MASARYKOVA UNIVERZITA přírodovědecká fakulta Centrum pro výzkum toxických látek v prostředí Vývoj metód pasívneho vzorkovania znečisťujúcich látok vo vodnom prostredí Habilitační práce Branislav Vrana Brno 2015 Poďakovanie Túto prácu venujem mojej žene Janke ako poďakovanie za jej lásku, s ktorou ma po celý náš spoločný život sprevádza. Ďakujem mojim rodičom za ich všestrannú podporu, mojim synom Andrejovi a Martinovi za trpezlivosť so mnou a za každodennú radosť, ktorú mi robia. Ďakujem mojim bývalým aj súčasným kolegom za plodnú prácu na spoločných projektoch. 2 1 Obsah 2 Zoznam skratiek a symbolov..............................................................................................5 3 Introduction........................................................................................................................7 4 Úvod...................................................................................................................................9 5 Koncept pasívneho vzorkovania......................................................................................10 5.1 Rozdeľovacie pasívne vzorkovače............................................................................11 5.2 Adsorpčné pasívne vzorkovače.................................................................................12 6 Koncept chemickej aktivity a rovnovážnej distribúcie látky v prostredí.........................14 7 Teória, modelovanie a kalibrácia pasívnych vzorkovačov..............................................16 7.1 Základné koncepty a modely pre rozdeľovacie pasívne vzorkovače........................18 7.2 Zovšeobecnený model pasívneho vzorkovača..........................................................23 7.3 Platnosť podmienok modelu......................................................................................23 7.4 Odpor k prestupu látky vo vodnej difúznej vrstve (WBL)........................................25 7.5 Odpor k prestupu látky v membráne.........................................................................27 7.5.1 Difúzny koeficient látky v membráne Dm..........................................................29 7.5.2 Rozdeľovači koeficient látky v systéme membrána-voda Kmv/ (alebo K^).......30 7.6 Kalibrácia pasívnych vzorkovačov............................................................................31 7.6.1 Statický expozičný dizajn...................................................................................31 7.6.2 Statický obnovovací dizajn................................................................................33 7.6.3 Prietokový dizajn................................................................................................33 7.6.4 Pasívne dávkovanie............................................................................................35 7.6.5 In situ kalibrácia.................................................................................................36 8 Zabezpečenie a kontrola kvality a štandardizácia............................................................39 8.1 Medzilaboratórne testy..............................................................................................39 8.2 Normalizácia pasívneho vzorkovania........................................................................41 9 Využitie pasívneho vzorkovania v regulačnom monitorovaní.........................................41 9.1 Rámcová smernica o vode.........................................................................................41 3 9.2 Európska stratégia boja proti znečisťovaniu vôd chemickými látkami.....................42 9.3 Hodnotenie stavu znečistenia povrchových vôd prioritnými látkami.......................43 9.4 Požiadavky na analytické metódy vo vzťahu k hodnoteniu povrchových vôd podľa Rámcovej Smernice o vode..................................................................................................45 9.5 Použiteľnosť pasívneho vzorkovania na monitorovanie prioritných látok podľa RSV 45 9.5.1 Hodnotenie súladu s ENK pre matricu voda......................................................48 9.5.2 Hodnotenie súladu s ENK pre matricu bio ta......................................................51 9.5.3 Úloha pasívneho vzorkovania vo viacstupňovom procese hodnotenia stavu vôd 54 10 Závery...............................................................................................................................56 11 Zoznam literatúry.............................................................................................................59 12 Zoznam publikovaných prác k téme habilitačnej práce...................................................74 12.1 Pôvodný vedecký článok v časopise......................................................................74 12.2 Kapitoly v odbornej knihe......................................................................................77 12.3 Ďalšie práce............................................................................................................77 4 2 Zoznam skratiek a symbolov A plocha vzorkovača, cez ktorú difunduje analyt APV Adsorpčný pasívny vzorkovač BAF bioakumulačný faktor BCF biokoncentračný faktor BMF biomagnifikačný faktor Cfree koncentrácia voľne rozpustenej látky ChA chemická aktivita látky D difúzny koeficient ENK environmentálna norma kvality ji difúzny tok látky i-tou fázou h koeficient prestupu látky i-tou fázou ke eliminačná rýchlostná konštanta prvého poriadku ^mw rozdeľovači koeficient membrána-voda Kow rozdeľovači koeficient oktanol-voda ^sw rozdeľovači koeficient vzorkovač-voda,; polymér-voda ^s vysoľovacia konštanta Setchenowovej rovnice LDPE polyetylén s nízkou hustotou PA polyakrylát PAH polycyklické aromatické uhľovodíky POM polyoxymetylén PCB polychlorované bifenyly PDMS polydimetylsiloxán PRC performančné referenčné látky (performance reference compounds) Q prietok vody v systéme; objem vody vymenený za jednotku času R univerzálna plynová konštanta 5 Re Reynoldsovo číslo RPV rozdeľovači pasívny vzorkovač Rs vzorkovacia rýchlosť látky RSV Rámcová smernica o vode 2000/60/ES RSV Rámcová smernica o vode 2000/60/ES S c Schmidtovo číslo SPE extrakcia na tuhej fáze SPME mikroextrakcia na tuhej fáze Sh Sherwoodovo číslo Sw rozpustnost' látky vo vode TMF faktor trofickej magnifikácie TWA time-weighted average; časovo vážený priemer koncentrácie Vs objem sorpčnej fázy vzorkovača WBL medzná difúzna vrstva vody (water boundary layer) 6 3 Introduction The quality of the environment is recognised as a high priority across the world, and measures towards its improvement have a positive effect on the quality of human life. Anthoropgenic pollutants in the aquatic environment may have a negative effect not only on the ecosystems, but, ultimately, also on the human health. In areas where water bodies cross national boundaries, there is a need to establish international monitoring networks that enable to obtain comparable, representative data on pollutant concentrations and trends. In order to succeed, it is necessary to obtain reliable information that is comparable between laboratories, is representative of environmental quality and underpins risk assessments and decisions on remedial actions. Much emphasis has been placed on the analytical chemical aspects of measuring pollutant levels in discrete samples but less attention has been paid to the underpinning sampling procedures despite the very much larger uncertainties associated with this crucial phase of the monitoring process. Passive samplers present an innovative monitoring tool for the time-integrated measurement of bioavailable contaminants in the aquatic environment. Passive sampling is based on the deployment in-situ, or use in the laboratory, of non-mechanical devices of simple construction capable of accumulating contaminants dissolved in water or sediment pore water. Such accumulation occurs via diffusion, typically over periods of days to weeks. This technology has great potential because of the simplicity of the principles underlying its function, and structure. In contrast to active samplers, passive samplers have no moving parts, they do not require a power source for their operation, and are relatively inexpensive. In addition, these devices can be deployed in almost any environmental condition, thus making them ideal for pollutant monitoring even in remote areas with minimal infrastructure. The presented habilitation thesis gives a brief introduction to passive sampling of pollutants in the aquatic environment. The development of selected methods is illustrated on my own scientific publications, or the reader is referred to available reviews. The thesis discusses functional principles of passive samplers and problems associated with the effects of environmental variables (temperature, water turbulence and sampler fouling) on their performance. Further, it gives a reference to the established or expected/potential performance of passive samplers for monitoring of the most discussed groups of aquatic pollutants and availability of calibration data that enable to obtain quantitative monitoring data. The document also explains the applicability of the passive sampling concept in the assessment of exposure of aquatic organisms in the process of risk assessment of pollutants. 7 Finally, the thesis discusses the state of the art and future research needs in development of method validation, quality assurance/quality control schemes and standardisation of the passive sampling technology. The thesis refers to studies that demonstrate the performance of passive samplers alongside conventional sampling schemes, and inter-laboratory studies that demonstrate reproducibility of data produced by different designs of passive samplers. These issues present the prerequisite for the future use of the passive sampling technology in scientific research as well as in regulatory monitoring. 8 4 Úvod Kvalita životného prostredia je prioritou v mnohých krajinách sveta, a opatrenia v tejto oblasti majú pozitivny vplyv na zlepšenie kvality ľudského života. Antropogénne znečisťujúce látky vo vodnom prostredí môžu mať negativny vplyv nielen na ekosystémy, ale v konečnom dôsledku i na zdravie človeka. V oblastiach, kde vodné toky prekračujú hranice štátov, je potrebné etablovat' medzinárodné monitorovacie siete, ktoré umožnia získať reprezentatívne a navzájom porovnateľné údaje o koncentráciách a trendoch znečisťujúcich látok. Pre úspešnosť týchto aktivít je potrebné získavať údaje, ktoré sú porovnateľné medzi laboratóriami, reprezentujú stav životného prostredia a sú vhodné na hodnotenie rizík i na rozhodovanie o opatreniach na zabránenie ďalšiemu znečisteniu. Hoci sa veľa dôrazu kladie na chemickú analýzu znečisťujúcich látok v diskrétnych vzorkách životného prostredia, menej pozornosti sa venuje aspektom vzorkovania, hoci táto kľúčová fáza procesu monitorovania je zvyčajne spojená s najväčšou neistotou. Pasívne vzorkovanie je inovatívny monitorovací nástroj na časovo integračné meranie biodostupných kontaminantov v životnom prostredí. Pasívne vzorkovanie je založené na in situ alebo ex situ použití nemechanických vzorkovačov jednoduchej konštrukcie, ktoré akumulujú rozpustené kontaminanty zvody alebo z pórovej vody v sedimentoch. Akumulácia látok do vzorkovačov prebieha samovoľne (difúziou) počas niekoľkých dní alebo týždňov expozície. Táto technológia má veľký potenciál využitia, hlavne vďaka jednoduchosti princípov, na ktorých je založená ich funkcia a konštrukcia. Na rozdiel od aktívnych vzorkovačov pasívne vzorkovače nemajú žiadne mechanické časti, väčšinou nevyžadujú na svoju prevádzku vonkajší zdroj energie, a navyše sú pomerne lacné. Tieto zariadenia môžu byť použité v takmer ľubovoľných podmienkach prostredia, čo umožňuje monitorovanie znečisťujúcich látok aj v odľahlých oblastiach bez infrastruktury. Predložená habilitačná práca podáva stručný úvod do problematiky pasívneho vzorkovania znečisťujúcich látok vo vodnom prostredí. Vývoj a využitie vybraných metód sú ilustruované na vlastných vedeckých publikáciách autora, alebo je čitateľ odkázaný na dostupné prehľadové štúdie. Práca diskutuje funkčné princípy pasívneho vzorkovania a problémy spojené s vplyvmi premenlivých podmienok vzorkovaného prostredia (napr. teplota, turbulencia vody a znečistenie vzorkovačov) na ich funkciu. Ďalej odkazuje na etablovaný alebo očakávaný potenciál pasívnych vzorkovačov na monitorovanie najdiskutovanejších skupín znečisťujúcich látok a tiež na kalibračné údaje, ktoré umožňujú získať kvantitatívne údaje 9 z monitorovania. Dokument tiež vysvetľuje použiteľnosť konceptu pasívneho vzorkovania v monitorovaní expozície vodných organizmov, potrebnom v procese hodnotenia rizík znečisťujúcich látok. Práca posktuje prehľad o súčasnom stave a potrebe ďalšieho výskumu v oblasti validácie metód, zabezpečovania a kontroly kvality a štandardizácie technológie pasívneho vzorkovania. Práca sa pritom odvoláva na štúdie, ktoré porovnávajú pasívne vzorkovače s konvenčnými metódami odberu vzoriek, a tiež na medzilaboratórne štúdie, ktoré demonštrujú reprodukovateľnosť dát získaných rôznymi typmi pasívnych vzorkovačov. Tieto témy tvoria predpoklad pre budúce využitie technológie pasívneho vzorkovania vo výskume i v regulačnom monitorovaní životného prostredia. 5 Koncept pasívneho vzorkovania Pasívne vzorkovanie je založené na použití in situ zariadenia, ktoré akumuluje kontaminanty z vody, alebo z iného média životného prostredia. Prestup kontaminantu z prostredia do vzorkovača je samovoľný difúzny proces, ktorý je hnaný rozdielom chemických aktivít monitorovanej látky medzi vzorkovaným médiom a sorpčnou fázou vzorkovača (Obrázok 3). Akumulácia látky vo vzorkovači prebieha až do ustálenia termodynamickej rovnováhy (resp. ustáleného stavu v dynamických systémoch, akými sú napr. rieky) medzi vzorkovačom a vodou, alebo až kým sa proces vzorkovania nepreruší. Doba expozície vzorkovačov je zvyčajne niekoľko dní až týždňov. Akumulované kontaminanty sa následne extrahujú a v extrakte sa stanovia ich koncentrácie. Ak sú vzorkovače kalibrované, je možné z množstva látky vo vzorkovači vypočítať koncentráciu látky rozpustenej vo vzorkovanom médiu. Pasívne vzorkovanie je často integračné, t.j. získaná vzorka reprezentuje koncentráciu látky vo vzorkovanom médiu za určité časové obdobie. Veľmi dôležitým aspektom pasívneho vzorkovania je možnosť vyjadriť množstvo látky vo vzorkovači v rovnováhe so vzorkovaným médiom formou chemickej aktivity (Mayer et al., 2003, viď kapitola 6.), ktorá je mierou hnacej sily pre samovoľný prestup látky medzi rôznymi zložkami životného prostredia. Vďaka vysokej sorpčnej kapacite a integračnému charakteru pasívnych vzorkovačov je možné monitorovať látky, ktoré sa nachádzajú rozpustené vo vode v extrémne nízkych koncentráciách (rádovo až pg/L). Konvenčné metódy vzorkovania vody, založené na bodových odberoch, neumožňujú stanovenie takýchto nízkych koncentrácií, hoci napr. environmentálne normy kvality, určené Rámcovou smernicou o vode(EU, 2013, 2008, 2000) vyžadujú monitorovať niektoré znečisťujúce látky vo vode metódami, ktoré majú medzu stanovenia na úrovni ng/L i nižšie. 10 V odbornej literatúre je dostupných niekoľko prehľadových prác, ktoré opisujú dizajn, kalibračné postupy, pracovné charakteristiky a príklady aplikácie rôznych pasívnych vzorkovčov na monitorovanie znečisťujúcich látok vo vodnom prostredí (Esteve-Turrillas et al., 2007; Kot-Wasik et al., 2007; Lobpreis et al., 2009; Lohmann et al., 2012; Lydy et al., 2014; Mills et al., 2007; Namiešnik et al., 2005; Ouyang and Pawliszyn, 2007; Sôderstrôm et al., 2009; Stuer-Lauridsen, 2005; Branislav Vrana et al., 2005; Vrana et al., 2010). (Booij, 2009)) v správe pre ICES Marine Chemistry Working Group sumarizoval potenciál využitia rôznych pasívnych vzorkovačov na monitorovanie látok regulovaných v Rámcovej Smernici o vode (EU, 2000) a v iných smerniciach a dohovoroch. (Vrana et al., 2010) vypracovali pre asociáciu laboratórií NORMAN pozičný dokument, ktorý uvádza prehľad použiteľnosti pasívneho vzorkovania pre monitorovanie emergentných (dosiaľ neregulovaných) znečisťujúcich látok vo vodnom prostredí. Ďalší aktuálny pozičný dokument asociácie NORMAN o pasívnom vzorkovaní bol aktuálne publikovaný medzinárodnou skupinou expertov, ktorú som v rokoch 2009-2014 koordinoval, na základe diskusií na špecializovanom workshope, ktorý sa konal v novembri 2014 v Lyone (Miége et al., 2015). Dokument identifikuje konkrétne aktivity, ktoré sú potrebné, aby pasívne vzorkovanie mohlo byť v budúcnosti využívané v rutinnom monitoringu vodného prostredia za účelom hodnotenia rizík a manažmentu kontaminantov. Užitočným zdrojom informácií o princípoch a aplikáciách pasívneho vzorkovania vo vodnom prostredí je i špecializovaná monografia, venovaná jednej z najznámejších vzorkovacích techník, tzv. semipermeabilným membránam (SPMD) (Huckins et al., 2006), a tiež prehľadová monografia pasívnych vzorkovacích technikách pre monitorovanie životného prostredia (Greenwood et al., 2007). V ďalšom texte je uvedené všeobecné rozdelenie metód pasívneho vzorkovania a tiež princípy ich funkcie. 5.1 Rozdeľovacie pasívne vzorkovače Rozdeľovacie pasívne vzorkovače (RPV) sú konštruované z hydrofóbnych polymérnych materiálov s vysokou permeabilitou pre nepolárné zlúčeniny. RPV absorbujú nepolárné látky z vody, pretože v porovnaní s vodou je rozpustnost' látok vo vzorkovači je oveľa vyššia ako vo vode. Hydrofóbne látky s nízkou rozpustnosťou vo vode sa dobre akumulujú v RPV, zatiaľ čo hydrofilné látky sa koncentrujú v oveľa menšej miere. Po dostatočne dlhej expozícii koncentrácia látky v RPV dosiahne dynamickú rovnováhu s koncentráciou vo vzorkovanom prostredí, napr. vo vode. Z rovnovážej koncentrácie látky v RPV je možné vypočítať koncetráciu vo vode pomocou rozdeľovačieho koeficienta vzorkovač-voda (Ksw). Táto koncentrácia vyjadruje koncentráciu voľne rozpustenej látky (Cfree), ktorá ale nie je totožná 11 s celkovou koncentráciou látky vo vzorke vody. Celková koncentrácia nepolárných látok vo vode závisí i od koncentrácie látky viazanej na rozpustené koloidy alebo dispergované častice organickej hmoty vo vode. Voľne rozpustná koncentrácia Cfree je priamo úmerná chemickej aktivite látky vo vode, a preto je vhodným parametrom, ktorý opisuje proces akumulácie chemických látok do vodných organizmov a tiež ich distribúciu medzi rôznymi zložkami životného prostredia. Pre použitie RPV sa predpokladá termodynamická rovnováha látky medzi vzorkovačom a vodou, ale pri praktickom použití vzorkovačov vo vode sa zvyčajne dosiahne rovnováha len pre látky s log Ksw < 5.5. Pre hydrofóbnejšie látky je prestup látky príliš pomalý (alebo sorpčná kapacita vzorkovača je príliš veľká) na rýchle dosiahnutie rovnováhy za typickú dobu expozície (2-8 týždňov). V takýchto prípadoch sa odhad Cfree opiera o meranie objemu vody, z ktorého vzorkovač in situ extrahuje sledovanú látku počas expozície. Extrahovaný objem vody (alebo vzorkovacia rýchlosť, ak je objem vyjadrený za jednotku času) sa dá odvodiť z rýchlosti uvoľňovania vybraných značených látok pridaných do vzorkovača pred expozíciou. V princípe ide o stanovenie rýchlosti uvoľňovania týchto látok, ktorá je kontrolovaná difúziou. Rýchlostná konštanta eliminácie prvého poriadku, meraná in situ je pre určitú látku rovnaká, ako je jej rýchlostná konštanta akumulácie, a preto môže byť použitá na výpočet Cfree i v situáciách, keď vzorkovač nie je v rovnováhe s okolitým prostredím. Boli vyvinuté modely a metódy pre odhad vzorkovacích rýchlostí látok (Booij and Smedes, 2010; Tatsiana P Rusina et al., 2010; Branislav Vrana et al., 2006) ako aj pre meranie rozdeľovacích koeficientov ^sw (Difilippo and Eganhouse, 2010; Hale et al., 2010; Smedes et al., 2009), čo umožňuje výpočet Cfree z koncentrácie látky vo vzorkovači. 5.2 Adsorpčné pasívne vzorkovače Adsorpčné pasívne vzorkovače(APV), obsahujú adsorbenty, ktoré sa tiež bežne používajú pri extrakcii na tuhej fáze (SPE) pri stanovení hydrofilných látok vo vode. V APV sa používa tenká vrstva takéhoto materiálu a od vodnej fázy ju zvyčajne oddeľuje filter alebo vhodná polopriepustná membrána. Podobne ako v prípade RPV látky difundujú cez hraničnú vrstvu vody a cez membránu alebo filter, ale akumulácia v sorpčnom materiáli prebieha adsorpciou na povrchu častíc sorbentu, a nie rozpúšťaním v objeme sorbentu. Adsorpcia hydrofilných látok je možná, pretože tieto látky sa môžu viazať rôznymi interakciami medzi povrchom sorbentu a funkčnými skupinami sledovaných látok, napr. van der Waals interakciami, 71-71 interakciami, vodíkovými väzbami 12 alebo elektrostatickými interakciami (Bäuerlein et al., 2012). Po dlhšej dobe expozície sa vzorkovacia rýchlosť v APV môže postupne znižovať nielen s dosiahnutím rovnovážneho stavu, ale môže byť obmedzená aj saturáciou adsorpčných miest sorbentu. Sorpcia iných ako sledovaných látok, napr. rušivých látok, prirodzene sa vyskytujúcich látok v prostredí, môže prispievať k zahlteniu sorpčných miest, alebo ku vzájomne kompetitívnej sorpcii sledovaných a interferujúcich látok. Aby sa týmto javom predišlo, alebo aby sa znížil účinok týchto javov na funkčnosť vzorkovača, sú APV zvyčajne exponované kratšie ako RPV. Krátka expozícia (niekoľko dní) je postačujúca, lebo polárne látky sa vo vodnom prostredí vyskytujú spravidla v rádovo vyšších koncentráciách ako hydrofóbne látky. Hoci pre APV bolo publikovaných veľa kalibračných štúdií, variabilita publikovaných vzorkovacích rýchlostí je vysoká (Harman et al., 2012, 2011) a proces akumulácie látok do APV dosiaľ nieje detailne pochopený, takže i použitie laboratórnych kalibračných dát v terénnych aplikáciách je doposiaľ iba empirické. Skupina látok Proces akumulácie Hnacia sila akumulácie Špeciácia Výber sorpčnej fázy Rozdeľovači pasívny vzorkovač Hydrofóbne látky Difúzia Absorpcia Rozdeľovači koeficient Vzorkovač/voda log Ksyi ~ log K0w nezávislý od koncentrácie Spravidla iba jedna forma Jediná polymérna fáza pre vzorkovanie širokého spektra Adsorpčný pasívny vzorkovač Polárne látky Kovy Difúzia ^ Adsorpcia Adsorpčný distribučný Koeficient AT. sw' Kc1/n Adsorpčné izotermy - závislé od koncentrácie Veľa látok disociuje niekoľko pKa - vzorkovanie viacerých foriem Výber z viacerých sorpčných fáz pre optimálne vzorkovanie Obrázok 1 Rozdiely v pasívnom vzorkovaní rozdeľovacími a adsorpčnými vzorkovačmi. 13 Rozdeľovači pasívny vzorkovač Adsorpčný pasívny vzorkovač Skupina látok Prestup látky Hydrofóbne látky Ovládaný membránou log Km Ovládaný WBL log K0IA >3 Polárne látky Ovládaný WBL Pre väčšinu látok Kinetika akumulácie Absorpčná a desorpčná kinetika sú izotropné ^é(sorpcia)- ^e(desorpcia) Ďesorpcia často nieje izotropná s adsorpciou Sorpcia na viacero typov sorpčných miest In situ kalibrácia Použitie performančných referenčných látok (PRC) Použitie PRC nie je univerzálne Potreba ďalšieho výskumu Obrázok 2 Rozdiely v mechanizme prestupu látok do rozdeľovacích a adsorpčných pasívnych vzorkovačov. Napriek týmto nedostatkom poskytujú APV cenné výsledky v skríningu polárnych znečisťujúcich látok vo vodách, pretože často umožňujú detegovať stopové množstvá látok v situáciách, kde klasické metódy vzorkovania, založené na nízkofrekvenčných bodových odberoch, zlyhávajú. Rozdiely medzi oboma typmi vzorkovačov sú ilustrované na obrázkoch 1 a 2. 6 Koncept chemickej aktivity a rovnovážnej distribúcie látky v prostredí Výsledkom pasívneho vzorkovania je odhad voľne rozpustenej koncentrácie (Cfree), ktorá sa považuje za najvhodnejší parameter expozície vodných živočíchov (Cornelissen et al., 2008). Dôvodom nie je, že by všetky látky, ktoré sa akumulujú v biote, pochádzali z vodného roztoku voľne rozpustených látok, ale fakt, že Cfree je priamo úmerná chemickej aktivite (CM) látky v sledovanej zložke životného prostredia, a dá sa vyjadriť ako pomer medzi koncentráciou a kapacitou pre akumuláciu látky v sledovanom médiu (CfreJSw), danom rozpustnosťou látky (vo stave podchladenej kvapaliny) (Sw)(Reichenberg and Mayer, 2006), teda ako podiel medzi koncentráciou látky v sledovanej fáze a kapacitou tejto fázy. Ak je 14 známa chemická aktivita ChA látky vo vode, látka má podľa teórie rovnovážneho rozdelenia tú istú chemickú aktivitu aj vo všetkých okolitých matriciach (zložkách životného prostredia), ktoré sa nachádzajú v rovnováhe s vodou. Qh/\ = Cfree _ Csed _ Cbiota _ Qjpjd q\q\\qž = ^Pasívny vzorto/a č S u u u u W sed biota lipid pasívny vzorková č kde Cx a Ux predstavujú koncentrácie a sorpčné kapacity matrice x (napr. sediment, vodný živočích, tukové tkanivo vodného živočícha, častice plaveniny a pod.). V systéme, ktorý je v stave termodynamickej rovnováhy, chemická aktivita je rovnaká vo všetkých zložkách prostredia. Naopak, rozdiel v chemickej aktivite medzi zložkami/matricami vo vodnom prostredí je hnacou silou pre pasívny/difúzny transport látok medzi nimi (Di Toro et al., 1991). Chemická aktivita látky v prostredí je teda vhodným parametrom na hodnotenie kvality životného prostredia. Táto teória je aplikovateľná napr. i pre akumuláciu látok zvody do vodných organizmov, dokonca i v prípade, že organizmy prijímajú tieto látky potravou, ak táto potrava je v termodynamickej rovnováhe s vodnou fázou (látka má v tejto fáze rovnakú ChA). V procese trávenia potravy jej akumulačná kapacita (U) klesá, čo spôsobuje zvýšenie ChA zložky oproti okolitému prostrediu. To urýchli ustálenie dynamickej rovnováhy, pri ktorej rýchlosť disipácie látky zvodného živočícha do vody je rovnaká ako opačný proces akumulácie. Nerovnovážny pomer chemickej aktivity látky medzi predátorom a jeho korisťou v zmysle tejto teórie vysvetľuje jav biomagnifikácie látok. 15 Difúzna dráha 5 i \_ = oo a \ ^ . á aktiv Difúzia y K *>• — _ ________t = tiy2_ _ Chemick \\ t = 0 Vodné prostredie Koeficienty prestupu látky: 10 10 Medzná vrstva vody (WBL) Membrána r) Ts Sorbent alebo rozpúšťadlo (Sorpčná fáza) S. Obrázok 3 Funkčný princíp pasívneho vzorkovača, ktorý ukazuje koncentračný profil látky počas difúzie a akumulácie z vodného prostredia (alebo iného vzorkovaného média, ľavá strana obrázka) do sorbentu (sorpčná fáza) cez permeabilnú (pórovitú alebo nepórovitú) membránu v čase t. Vysoká afinita látky k sorpčnej fáze je hnacou silou difúzie molekúl sledovanej látky do vzorkovača, až kým nedôjde k vyrovnaniu chemickej aktivity látky v oboch médiách, t.j. k ustáleniu termodynamickej rovnováhy. 7 Teória, modelovanie a kalibrácia pasívnych vzorkovačov Akumulácia kontaminantov do pasívneho vzorkovača je viacstupňový transportný proces. Na ilustráciu základných krokov tohoto procesu uvádzam najprv príklad akumulácie kontaminantu do vzorkovača, ktorý pozostáva z centrálnej sorpčnej fázy, obklopenej membránou. Ďalej sa predpokladá, že na povrchu vzorkovača sa nachádza vrstva biofilmu (biofouling), a že vzorkovač je umiestnený v ochrannej klietke (Obrázok 4). Analyt rozpustený vo vode najprv vstupuje konvekciou z okolitých vôd do ochrannej klietky, kde pohyb vody môže byť pomalší ako v prúde vody v okolí klietky. V blízkosti vrstvy biofilmu sa transport molekúl analytu konvekciou stále viac znižuje, až napokon celý prestup látky prebieha molekulárnou difúziou cez medznú vrstvu vody (water boundary layer, WBL). Po difúzii cez membránu sa analyt sorbuje do sorpčnej fázy. Tento všeobecný náčrt má rôzne modifikácie a dá sa aplikovať na rôzne typy vzorkovačov. Napr. niektoré vzorkovače nepoužívajú ochrannú klietku, vrstva biofilmu nemusí byť prítomná a membrána môže 16 zároveň plniť úlohu sorpčnej fázy (napr. v rôznych zariadeniach využívajúcich mikroextrakciu na tuhej fáze (SPME) (Ouyang et al., 2007), pásky z polyetylénu s nízkou hustotou (LDPE) (Estoppey et al., 2015), alebo z polydimetylsiloxánu (PDMS) (Smedes and Booij, 2012)), alebo vzorkovače môžu byť vybavené ďalšími fázami, ktoré sú umiestnené medzi membránou a centrálnou fázou (napr. membrane-enclosed sorptive coating (MESCO) (Vrana et al., 2001) alebo Chemcatcher (Kingston et al., 2000). 5, Centrál- mem- biofilm na fáza brána / 1 5m,efr voda Obrázok 4 Schematické znázornenie koncentračných profilov v dvojfázovom rozdeľovačom pasívnom vzorkovači, na ktorom sa nachádza vonkajšia vrstv biofilmu. Vzorkovač je vizualizovaný ako pravá strana symetrického vzorkovača, alebo ako celkový prierez vzorkovačom, ktorý obsahuje nepriepustnú stenu naľavo od centrálnej fázy. Čiarkované línie indikujú, ako je možné odhadnúť účinnú hrúbku jednotlivých vrstiev vzorkovača. Prevzaté z (K Booij et al., 2007). Za posledných dvadsať rokov bolo vyvinutých a je používaných niekoľko modelov, ktoré umožňujú lepšie pochopiť kinetiku prestupu kontaminantu do pasívneho vzorkovača. Tieto modely sú potrebné, aby bolo možné porozumieť, ako súvisí množstvo látky sorbované do vzorkovača s j ej koncentráciou vo vonkajšom prostredí (vo vode), ako aj pre navrhovanie a vyhodnocovanie kalibračných experimentov. Modely sa líšia v počte uvažovaných fáz, ako aj v zjednodušujúcich predpokladoch, ktoré sa berú do úvahy, ako je napr. existencia pseudo-ustáleného stavu (steady-state), prítomnosť alebo absencia lineárnych koncentračných gradientov pozdĺž priečneho profilu fáz, a ďalej spôsob, akým sa modeluje prestup látky 17 medznou vrstvou vody (WBL), a či je koncentrácia látky počas expozície vzorkovača konštantná. V nasledujúcom texte sú predstavené základné koncepty a modely používané v literatúre o pasívnom vzorkovaní. Ďalej je diskutovaný prestup látok cez rôzne fázy, z ktorých vzorkovače pozostávajú. Napokon sú diskutované dôsledky týcho modelov pre dizajn a evaluáciu kalibračných experimentov. 7.1 Základné koncepty a modely pre rozdeľovacie pasívne vzorkovače Koeficienty prestupu látky (k[) sa často používajú na prepojenie toku látky (ji) s koncentračným rozdielom AQ látky medzi okrajovými bodmi tejto fázy j, = h (2) Rovnica 2 vyjadruje predstavu, že tok látky (/',) je priamo úmerný hnacej sile AQ. Koeficient prestupu látky sa dá interpretovať ako vodivostný člen, s rozmerom rýchlosti (napr. m s"1). Tento postup bol použitý na modelovanie akumulácie látok do niekoľkých typov pasívnych vzorkovačov (Chen and Pawliszyn, 2004; Flynn and Yalkowsky, 1972; Huckins et al., 2006, 1993; Tcaciuc et al., 2015; B Vrana et al., 2005; Vrana et al., 2001; Wennrich et al., 2003). Diferenciálna rovnica, ktorá opisuje akumuláciu látky do vzorkovača, sa dá zapísať: dCs_ Ak0 f dt Va s c s (3) kde Cs a Cw sú objemové koncentrácie kontaminantu vo vzorkovači a v povrchovej vode, Vs je objem vzorkovača, A je plocha vzorkovača, cez ktorú difundujú do vzorkovača molekuly analytu a^sw je rozdeľovači koeficient látky v systéme vzorkovač-voda. Celkový koeficient prestupu látky do vzorkovača k0je daný: 111 1 U U U K U K K ' Ao 'V ^b^bw ftmAnw kde kw, kb, km sú koeficienty prestupu látky cez WBL, biofilm a membránu a ^bw a ^mw sú rozdeľovacie koeficienty látky v systéme biofilm-voda a membrána-voda. Rovnica (4) vyjadruje, že celkový odpor k prestupu látky (l/£0) je rovný súčtu odporov k prestupu látky v jednotlivých fázach vzorkovača. Ak uvážime, že koeficient prestupu látky je daný podielom difúzneho koeficienta a účinnej hrúbky difúznej vrstvy (ô), rovnica (4) sa dá napísať aj 18 -l = ^+^L + ^_ (5) k o Dví Dm Kmvi Db Kbvi (Bartkow et al., 2005) počítali aj s odporom k prestupu látky, spôsobeným ochrannou klietkou, ktorá je okolo vzorkovača a pridali v rovnici do súčtu člen A/Qv, kde Qv je prietok vody cez klietku a A je plocha vzorkovača. Tento čiastkový odpor sa však väčšinou môže zanedbať, okrem niektorých extrémnych dizajnov klietok, ktoré obmedzujú prietok vody vzorkovacím zariadením. Po krátkej dobe expozície vzorkovača je koncentrácia sledovanej látky vo vzorkovači oveľa nižšia ako je jej rovnovážna koncentrácia, t.j. Cs « KSWCW a rovnica sa zjednoduší Ak dCs*-fCwdt (6) a po integrácii v čase dostávame / Q ~ Cwdt = CwTWAt (7) kde Cw,TWAJe časovo vážený priemer (TWA) koncentrácie vo vodnej fáze. Na označenie prvej fázy vzorkovacieho procesu používajú tri pojmy. Keď je Cw konštantná v čase, koncentrácia akumulovaných kontaminantov lineárne narastá v čase. Tento časový úsek vzorkovania sa preto nazýva lineárna fáza akumulácie. Pre scenáre, keď vodné koncentrácie kolíšu v čase, koncentrácia vo vzorkovači je priamoúmerná TWA koncentrácii a vzorkovanie sa nazýva časovo integračné. Napokon, pretože rýchlosť zmeny koncentrácie vo vzorkovači je priamo úmerná koncentrácii vo vode, táto raná fáza vzorkovania sa nazýva kinetickým vzorkovaním. Zaujímavým aspektom rovnice (7) je, že produkt koA je ekvivalentný zdanlivému objemu vody, z ktorého vzorkovač vyextrahuje analyt za dobu expozície t. Na tento produkt (k0xÄ) nahliadame ako na vzorkovaciu rýchlosť (Rs): Rs=k0A (8) Pretože Rs reprezentuje objem vody extrahovaný za jednotku času, vytvára konceptuálne prepojenie medzi tradičnými vsádzkovými extrakčnými metódami a metódami založenými na pasívnom vzorkovaní. Rovnica ((8) vyjadruje, že vzorkovacia rýchlosť je priamo úmerná ploche vzorkovača. Preto porovnanie vzorkovacích rýchlostí medzi rôznymi dizajnmi vzorkovačov poskytuje relevantné výsledky iba v prípade, že sa berú do úvahy i rozdiely v ploche A. 19 o 1.2 1.0 - 0.8 - 0.6 - 0.4 0.2 - 0.0 Čas Obrázok 5 Efektívny objem vody ektrahovaný vzorkovačom (Ns/Cw) ako funkcia času. Pre dlhé expozičné doby je extrahovaný objem obmedzený sorpčnou kapacitou vzorkovača (ÄTswxVs) a pre krátke expozičné časy súčinom vzorkovacej rýchlosti a doby expozície. Približné modely, ktoré platia pre lineárnu časť akumulácie (krátka doba expozície) a rovnovážne vzorkovanie (dlhá doba expozície) sú znázornené čiarkovanými čiarami. Upravené podľa (K Booij et al., 2007). Pre veľmi dlhé expozičné časy a pri konštantnej hodnote Cw sa koncentrácia vo vzorkovači nemení v čase a riešením rovnice (3) je: 0 (9) čo je vyjadrením, že koncentrácia látky vo vzorkovači dosahuje rovnovážnu hodnotu (Cs = KswxCw). Príslušný vzorkovací režim sa nazýva rovnovážne vzorkovanie. Všeobecné riešenie rovnice rovnice (3) pre konštantnú koncentráciu Cwje dané (Vrana et al., 2001): Cs = Ksw Cw [1 -exp(-/ce0] + C0 exp(-/ce f) (10) kde Co je koncentrácia vo vzorkovači v čase ŕ = 0 a eliminačná rýchlostná konštanta (ke) je daná: 20 Čas [dni] Obrázok 6. Príklad izokinetickej výmeny látky medzi rozdeľovačmi pasívnym vzorkovačom a vodou. Graf ukazuje kinetiku akumulácie fluoranténu do pasívneho vzorkovača (Gerstel Twister, 2x0,5 cm) z vody s konštantnou koncentráciou Cw (čierne body) a kintetiku disipácie perdeuterovaného fluoranténu, ktorý bol pred experimentom pridaný do vzorkovača, a ktorého koncentrácia vo vode je počas experimentu udržiavaná pod medzou detekcie (Cw = 0). Plné čiary predstavujú fit experimentálnych dát modelom podľa rovnice (10). Akumulácia i disipácia látky je charakterizovaná tou istou hodnotou eliminačnej rýchlostnej konštanty ke, čo je princíp in situ kalibrácie - stanovenia vzorkovacích rýchlostí priamo v teréne. (Vrana, nepublikované). Rovnica (10) ukazuje, že akumulácia z prostredia a eliminácia počiatočného množstva látky vo vzorkovači (stanovuje sa analýzou tzv. fabrikačných blankov) sú aditívne. Odčítanie týchto koncentrácií môže byť problematické, keď pôvodná koncentrácia je vyššia alebo rovná rovnovážnej koncentrácii. Vtákom prípade koncentrácia v exponovanom vzorkovači môže byť menšia ako v kontrolných neexponovaných vzorkách (fabrikačné blanky) a odčítanie koncentrácie v kontrole by malo za následok negatívnu vypočítanú hodnotu v exponovanom vzorkovači. Preto sa v prípade rovnovážneho vzorkovania neodporúča odčítanie hodnoty fabrikačného blanku od výsledku merania v exponovanom vzorkovači. Rovnica (10) tiež ukazuje, že v rozdeľovači ch pasívnych vzorkovačoch je akumulácia i eliminácia jednej a tej istej látky charakterizovaná rovnakou hodnotou ke (Obrázok 6). Tento poznatok tvorí základ 21 odhadu in situ vzorkovacích rýchlostí z rýchlostí disipácie tzv. performančných referenčných látok (PRCs) (Huckins et al., 2002) Keď je počiatočná koncentrácii vo vzorkovači rovná nule, rovnica (10) sa dá integrovať: £S - ^sw Q/v 1 - exp v KswVsJ (12) a pre krátke časy expozície je možné ju zjednodušiť na lineárnu rovnicu: CwRst y. (13) Pre disipáciu látok, ktoré sa nenachádzajú v prostredí (Cw = 0), ale sú pridávané do vzorkovača pred expozíciou (napr. PRC), rovnica (10) sa dá zjednodušiť: Cs = C0 exp(-/ce 0 (14) Koncentrácie vo vode sa dajú vypočítať z množstva látky sorbovaného vo vzorkovači (Ns), in situ vzorkovacej rýchlosti látky Rs a jej rozdeľovačieho koeficienta v systéme vzorkovač-voda^sw, použitím preusporiadanej rovnice (12): 1 - exp Fist (15) Pre rovnovážne vzorkovače je člen v hranatých zátvorkách rovný 1 a vodné koncentrácie sa vypočítajú pomocou rovnice: (16) Pre kinetické vzorkovače, ktoré pracujú v lineárnom akumulačnom móde je člen v hranatej zátvorke približne rovný (Rsť)/(KSWVS), a koncentrácia vo vode sa dá vypočítať: (17) Menovatele v rovniciach (15(11) sa dajú interpretovať ako zdanlivý objem vody, z ktorého vzorkovač odstráni analyt počas expozície (Obrázok 5). V prípade rovnovážneho vzorkovania je tento objem obmedzený sorpčnou kapacitou vzorkovača (KSwxVs). Pri 22 kinetickom vzorkovaní je zdanlivý exktrahovaný objem vody obmedzený vzorkovacou rýchlosťou a expozičným časom (Rsxt). 7.2 Zovšeobecnený model pasívneho vzorkovača Diskusia v predchádzajúcej časti sa dá rozšíriť na iné pasívne vzorkovače, ktoré obsahujú ľubovoľný počet sub-fáz (bariér), za predpokladu, že sorpčná rovnováha je ustálená na rozhraniach medzi nimi, a že sú ustálené (steady-state) toky látky vo vnútri jednotlivých bariér medzi vodou a sorpčnou fázou (t.j. rozdiel medzi tokom látky dovnútra a von z každej čiastkovej bariéry je relatívne malý). Rovnica (5) sa dá zovšeobecniť (Vrana et al., 2001): -1=1 -5- (18) k o Di Kiw kde súčet platí pre všetky fázy i, z ktorých vzorkovač pozostáva. Vývoj množstva analytu akumulovanom v sorpčnej fáze vzorkovača (t.j. v tej časti vzorkovača, ktorá sa extrahuje a následne analyzuje) je daný rovnicou (12), kde celková hodnota Ksw je vyjadrená zovšeobecneným vzorcom =lYf~ (19) a objem vzorkovača Vs je rovný súčtu objemov všetkých sub-fáz, ktoré sa analyzujú. V literatúre venovanej SPME sa používa podobný empirický model, ktorý opisuje výmenu látky medzi vzorkovačom a vodou (Chen and Pawliszyn, 2003; H. J. Vaes et al., 1996): *£-=k,C„-k2Cs (20) Tento model je matematicky ekvivalentný s rovnicou (2), kde £2 = (A k0)/(Ksw Vs) and k\ = KsWk2. 7.3 Platnosť podmienok modelu Pre vyššie opísané modely sa predpokladá, že v membráne a v centrálnej fáze existujú lineárne koncentračné gradienty, že na rozhraní medzi fázami sa okamžite ustaľuje termodynamická rovnováha, a že molekulárna difúzia je hlavným transportným mechanizmom látok v membráne, nezávisle od času a koncentrácie. 23 V počiatočnej fáze expozície vzorkovača analyty musia penetrovať cez membránu, aby sa dostali do centrálnej sorpčnej fázy, čo spôsobuje tzv. lag fázu. Teoretický model toku látky cez plochú dosku s konštantnou koncentráciou látky na oboch stranách predpovedá dobu oneskorenia (lag time) (Crank, 1975): S2 ŕ = —=- (21) Difúzne koeficienty organických látok v polyméroch sú veľmi rozličné. Hodnoty difúzneho koeficienta polycyklických aromatických uhľovodíkov a polychlorovaných bifenylov pre polymér PDMS sa pohybujú v rozmedzí od 10"11 do 10~10 m2 s"1 (T. Rusina et al., 2010), pre polymér LDPE v rozmedzí hodnôt od 10"14 do 10"12 m2 s"1 (T. Rusina et al., 2010). Hodnoty difúzneho koeficienta benzénu v poly(metylmetakryláte) sú rádovo 10"16 m2 s"1 a v poly(vinylalkohole) 10"19 m2 s"1 (George and Thomas, 2001). Rovnica (21) predikuje pre membránu s hrúbkou 100 um dobu zdržania analytu vo vrstve PDMS 17-167 sekúnd, 30 minút až 46 hodín vo vrstve LDPE, cca. niekoľko mesiacov vo vrstve poly(metylmetakrylát)u a niekoľko storočí vo vrstve poly(vinylalkohol)u. Je zrejmé, že v prípade, ak akumulácia látok do vzorkovača je kontrolovaná difúziou cez WBL, distribúcia analytu v materiáli, z ktorého sú zhotovené membrány vzorkovača, neovplyvňuje vzorkovacie rýchlosti. Ak uvažujeme difúzny koeficient látky vo vode cca. 5x 10"10 m2 s"1 a účinnú hrúbku hraničnej vrstvy vody 30 až 300 um, pre akumuláciu kontrolovanú difúziou vo WBL sú očakávané doby zdržania látky vo vrstve medzi 0.3 a 30 s. V prípade, že membrána sa pri spracovaní vzorky vyhadzuje a analyzuje sa iba centrálna fáza (napr. vo vzorkovači POCIS), musí sa počítať s dobou zdržania analytu počas difúzie membránou aj v prípade, že rýchlosť určujúcim krokom je WBL. Lineárne koncentračné gradienty nemôžu existovať v membráne, ktorá akumuluje analyty, pretože v takom prípade tok látky do membrány musí byť väčší ako tok látky von z membrány na opačnej strane. Podľa tej istej argumentácie nemôže existovať lineárny gradient ani v centrálnej sorpčnej fáze vzorkovača. Koncentračný gradient v strede sorpčnej fázy (napr. pre SPMD, MESCO s PDMS tyčkou) alebo v blízkosti nepriepustnej steny (napr. Chemcatcher alebo SPME) by mal byť nulový (ináč by vznikala diskontinuita toku látky). Koncentračný gradient na vonkajšej strane centrálnej fázy by mal byť nenulový (inak by centrálna fáza nič neakumulovala). V prípade, že akumulácia látky je kontrolovaná WBL, existencia nelineárnych gradientov v membráne alebo v centrálnej fáze nespôsobuje 24 neplatnosť modelu, ale v prípade, že je akumulácia kontrolovaná difúziou v membráne, je potrebné tento jav zobrať do úvahy. Nelinearita koncentračných gradientov sa dá hodnotiť použití tzv. účinnej hrúbky vrstvy (<5t,eff), ako je zobrazené na Obrázok 4. (Louch et al., 1992) ukázali, že účinná hrúbka membrány sa od skutočnej hrúbky líši len menej ako 20% pre expozičné časy, ktoré sú vyššie ako doba zdržania látky v membráne. Predpoklad, že na rozhraní fáz je okamžite ustálená termodynamická rovnováha, je pravdepodobne splnená pre nízke hodnoty rýchlostí prestupu látky, aké sú typické pre metódy pasívneho vzorkovania, hlavne pre pryžové polyméry, ktoré majú krátku dobu relaxácie (George and Thomas, 2001). Hoci difúzne koeficienty látok v polyméroch závisia na koncentrácii difundujúcej látky, bolo ukázané, že táto závislosť je slabá (George and Thomas, 2001) a môže sa zanedbať, lebo pri pasívnom vzorkovaní sú nízke koncentrácie. 7.4 Odpor k prestupu látky vo vodnej difúznej vrstve (WBL) Exaktné modely prestupu látky cez WBL sú k dispozícii len pre niektoré jednoduché usporiadania toku vody, ako napr. pre tok v potrubí a paralelný tok pozdĺž absorbujúcej plochej dosky (Bird et al., 2007; Kader and Yaglom, 1972; Schlichting et al., 2000). V tesnej blízkosti platne sa moment vodného toku postupne od okraja znižuje vplyvom povrchového trenia. Keď sa voda pohybuje pozdĺž platne, táto spomalená vrstva atenuuje moment vodných vrstiev, ktoré sa nachádzajú vo väčšej vzdialenosti od povrchu, čo spôsobuje vznik viskóznej vrstvy, ktorej hrúbka postupne narastá so rastúcou vzdialenosťou od okraja platne v smere toku vody. Analogicky, analyty sa odstraňujú z vrstvy, ktorej hrúbka narastá v smere toku, čo spôsobuje vznik tzv. medznej koncentračnej vrstvy (WBL). S narastajúcou hrúbkou tejto vrstvy je významnejší prestup látky turbuletnou difúziou, pretože turbulentně difúzne koeficienty narastajú s rastúcou vzdialenosťou od povrchu (Kader and Yaglom, 1972; Son and Hanratty, 1967). Vo veľkej vzdialenosti od okrajovej hrany sa ustáli koncentračný profil, ktorý už nezávisí na vzdialenosti pozdĺž platne. Rovnice pre extrémne situácie - krátka platňa s rastúcimi koncentračnými hraničnými vrstvami a dlhá platňa (koncentračné medzné vrstvy sú nezávislé od vzdialenosti od hrany) boli odvodené pre laminárne toky (Opdyke et al., 1987). Koeficienty prestupu látky pre krátku platňu (spriemerované pre celý povrch) sú odvodené v (Opdyke et al., 1987). Vo všeobecnosti je však takmer nemožné odvodiť rovnicu, ktorá by umožnila presne odhadnúť koeficient prestupu látky vo WBL pre komplexnejšie geometrie vzorkovača umiestnené v prirodzenom toku, ktorého rýchlosť a turbulencia sa 25 menia v čase a priestore. (K Booij et al., 2007) však uvádzajú niekoľko zovšeobecnení, ktoré je možné uplatniť pri opise prestupu látky cez WBL do vzorkovača: 1. Počet premenných v modelových experimentoch prestupu látky cez WBL sa dá zmenšiť korelováním bezrozměrných kritérií používaných v chemickom inžinierstve (Sherwoodovo číslo (Sh), Reynoldsovo číslo (Re) a Schmidtovo číslo (Sc)), charakteristických pre zvolenú geometriu vzorkovača. 2. Pre širokú škálu takýchto empirických korelácií chemicko-inžinierska literatúra uvádza, že koeficient prestupu látky medznou vrstvou vody kw je priamoúmerný 2/3 molekulovému difúznemu koeficientu vo vode D podľa vzťahu kw ~ D (Bird et al., 2007; Worch, 1993). To v dôsledku značí, že hrúbka účinnej medznej vrstvy klesá -1/3 s hodnotou rastúceho difúzneho koeficienta podľa ä^-D' . 3. Účinná vodná medzná difúzna vrstva WBL, hoci je užitočná ako model pre vizualizáciu, kam až zasahuje koncentračný gradient sledovanej látky do vodného toku, by nemala byť dezinterpretovaná ako hrúbka fyzicky nereálnych objektov ako je napr. stagnantný film alebo nepremiešavaná hraničná vrstva vody. 4. Pre danú geometriu vzorkovača a prúdenie by mali byť hodnoty kw pre malé vzorkovače vyššie ako pre veľké vzorkovače. 5. kw narastá s rýchlosťou toku pre danú geometriu pasívneho vzorkovača (Obrázok 70brázok 7 Vplyv hydrodynamiky na vzorkovacie rýchlosti (Rs) látok do vzorkovača Chemcatcher s LDPE membránou. Experiment bol uskutočnený pri troch rýchlostiach prúdenia vody, dosiahnutými v laboratórnych podmienkach rôznymi frekvenciami otáčania karuselu s vzorkovačmi v prietokovom systéme. Prevzaté z (B Vrana et al., 2006).), ale predikcia jeho absolútnej hodnoty modelovaním je veľmi obtiažna. Navyše, porovnanie odhadutých a experimentálnych vzorkovacích rýchlostí je komplikované tým, že rýchlosti prúdenia vody v okolí vzorkovača sú väčšinou odhadované/vypočítané a nie fyzicky merané. 26 Obrázok 7 Vplyv hydrodynamiky na vzorkovacie rýchlosti (Rs) látok do vzorkovača Chemcatcher s LDPE membránou. Experiment bol uskutočnený pri troch rýchlostiach prúdenia vody, dosiahnutými v laboratórnych podmienkach rôznymi frekvenciami otáčania karuselu s vzorkovačmi v prietokovom systéme. Prevzaté z (B Vrana et al., 2006). 7.5 Odpor k prestupu látky v membráne V pasívnych vzorkovačoch sa používajú dva typy polymérnych membrán. Často používanými neporóznymi membránami sú najmä LDPE (Huckins et al., 1993, 1990; Kingston et al., 2000; B Vrana et al., 2005; Wennrich et al., 2003), PDMS (Tatsiana P Rusina et al., 2010; Smedes and Booij, 2012; van Pinxteren et al., 2010), polyakrylát (Leslie et al., 2002; Paschke and Popp, 2003) a iné nepolárné polyméry. Mikroporózne membrány môžu byť zhotovené z regenerovaného acetátu celulózy (CA) (Sabaliunas and Sôdergren, 1996; Sôdergren, 1990; Vrana et al., 2001), polyétersulfónu (Alvarez et al., 2007, 2004), polysulfónu (Kingston et al., 2000), polyakrylamidového (Zhang and Davison, 1995) či agarózového (Chen et al., 2012) hydrogélu. V niektorých aplikáciách membrána je zároveň aj primárnou sorpčnou fázou vzorkovača, napr. PDMS v Gerstel-Twister (Assoumani et al., 2015), pre LDPE pásky (Adams et al., 2007a), SPME vlákna (Ouyang et al., 2005), či pláty PDMS (Smedes and Booij, 2012). V iných aplikáciách je membrána určená na separáciu sorpčnej fázy od vody, napr. vo vzorkovači Chemcatcher (Greenwood et al., 2007), MESCO (Vrana et al., 2001), SPMD (Huckins et al., 1993), a tiež na zníženie difúzneho toku látky do sorpčnej fázy. Konduktivita membrány pre prestup látky je daná rovnicou: 27 I is _ Dm Knw Km A mw — ~ (22) m kde ť^je hrubka membrány (rovnica (5)). Hodnoty Dm i ^mw sú specifické pre každú látku. Úloha Kmw v rovnici (22) je zrejmá, ak uvážime, že látky s vysokou hodnotou rozdeľovačieho koeficienta membrána-voda majú i vysoké koncentrácie v membráne v blízkosti rozhrania membrána-voda, ak predpokladáme okamžité ustálenie sorpčných rovnováh na fázových rozhraniach. Dôsledkom toho je zvýšený koncentračný gradient naprieč membránou v porovnaní s látkami, ktoré majú nízke hodnoty Kmw. Strmší koncentračný gradient spôsobuje vyšší tok látky cez membránu. Naopak, výber membrány, voči ktorej majú analyty nízku afinitu (napr. hydrofilné membrány pri vzorkovaní hydrofóbnych látok) spôsobuje zvýšený odpor voči prestupu látky, čo vedie k zníženiu vzorkovacích rýchlostí. Je dokumentovaných niekoľko prípadov takéhoto efektu. Vzorkovacie rýchlosti chlórovaných pesticídov vo vzorkovačoch, ktoré obsahovali LDPE membránu, boli až stonásobne vyššie ako v prípade, keď bolo použité organické rozpúšťadlo naplnené do membrány z acetátu celulózy (Sabaliunas and Sôdergren, 1996). Náhrada hydrofilnej membrány vo vzorkovačoch MESCO a Chemcatcher polyetylénom viedla k výraznému zvýšeniu vzorkovacích rýchlostí (B Vrana et al., 2005; Wennrich et al., 2003) a vzorkovacie rýchlosti polárnych látok do pasívneho vzorkovača POCIS boli oveľa vyššie v prípade použitia polárnej polyétersulfónovej membrány, ako v prípade, keď sa použili nepolárné polyetylénové alebo nylonové membrány (Alvarez et al., 2007). Výber materiálu membrány má vplyv nielen na vzorkovaciu rýchlosť, ale aj na citlivosť vzorkovača na zmeny prúdenia vody. Keď sa zníži odpor membrány, rýchlosť vzorkovania je kontrolovaná medznou vrstvou vody (WBL), ktorá je silne závislá od hydrodynamických podmienok na rozhraní membrána-voda. Z toho vyplýva, že pokusy znížiť citlivosť pasívneho vzorkovania na prúdenie vody pridaním membrány, ktorá má nízke hodnoty rozdeľovacieho koeficienta pre sledované látky, spôsobia automaticky zníženie vzorkovacích rýchlostí. Naopak, pridanie membrán s vysokými hodnotami ^mw zvýši vzorkovacie rýchlosti, ale aj ich citlivosť na zmeny rýchlosti prúdenia vody (B Vrana et al., 2005). Zníženie vzorkovacej rýchlosti nemusí vždy znamenať problém. Závisí to od viacerých faktorov, napr. od koncentrácie látky vo vode, expozičnej doby a citlivosti analytického prístroja. Záverom predchádzajúcej diskusie je, že nie je možné vyvinúť pasívny vzorkovač, ktorý by mal dostatočne vysoké vzorkovacie rýchlosti vo všetkých prostrediach. 28 V prípade akumulácie kontrolovanej membránou sa predpokladá, že smernica závislosti log Rs voči log Kmw je približne jednotková, pretože Rs ~ k0 ~ DmxKmw. V praxi sa dosahujú mierne nižšie smernice, pretože hodnota Dm mierne klesá s rastúcou veľkosťou molekuly (Booij et al., 2003; H. J. Vaes et al., 1996; Verbruggen et al., 2000). Akumulácia kontrolovaná membránou sa dá identifikovať, ak smernica závislosti log ke od log Kmw je približne 0, alebo mierne nižšia, pretože podľa rovnice (11) platí ke ~ ^mw_1. Tieto podmienky sa typicky pozorujú pre látky s hodnotami log Kmw values < 3.5 pre SPME vlákna s polyakrylátovou fázou (H. J. Vaes et al., 1996; Verbruggen et al., 2000) a pre látky z log Kow hodnotami <4.5 pre SPMD vzorkovače (Vrana and Schúúrmann, 2002) (Obrázok 8). Je potrebné pripomenúť, že hranica medzi akumuláciou kontrolovanou WBL a membránou nezávisí iba od vlastností analytov, ale aj od hydrodynamických podmienok na rozhraní membrána-voda (5). Preto v stagnantnej vode môže byť kritická hodnota ^mw rozhrania medzi WBL a membránovou kontrolou posunutá smerom k nižším, v turbulentnej vode zase k vyšším hodnotám. 7.5.1 Difúzny koeficient látky v membráne Dm Odhad vzorkovacích rýchlostí pre transport kontrolovaný difúziou v membráne je možný na základe nameraných hodnôt difúznych koeficientov Dm sledovaných látok v materiáli, z ktorého sú membrány zhotovené. Difúzne koeficienty Dm je možné pomerne ľahko stanoviť pomocou metódy navrstvených filmov (Sjôberg et al., 1996), ktorá spočíva v jednorozmernej difúzii (kolmo na povrch filmu) látky cez na seba navrstvené filmy polyméru. Po vhodnom čase sa jednotlivé vrstvy analyzujú na obsah látky a zo získaného koncentračného profilu sa vypočíta difúzny koeficient z parciálneho riešenia druhého Fickovho zákona (Crank, 1975). (T. Rusina et al., 2010) použila túto metódu na stanovenie Dm pre polychlorované bifenyly a polyaromatické uhľovodíky v LDPE a PDMS. Odhadnuté hodnoty £)mboli 2-2.5 rádu nižšie v 2 -1 LDPE ako v polyméroch na báze PDMS. Log D hodnoty (m s" ) pre PCB sú v rozsahu -10.1 do -10.9 v PDMS a -11.9 do -13.7 v LDPE. Difúzne koeficienty v polyoxymetyléne, ktorý sa tiež používa v konštrukcii niektorých pasívnych vzorkovačov (Hawthorne et al., 2011), boli 2 -1 publikované iba pre fenantrén a pyrén, a ich log D hodnoty (m s" ) sa pohybujú okolo -14 (Ahn et al., 2005). Vo všeobecnosti hodnoty Dm klesajú s rastúcou mólovou hmotnosťou látky a tiež s rastúcim povrchom molekuly (T. Rusina et al., 2010), čo umožňuje extrapolovat' hodnoty difúznych koeficientov aj pre ďalšie látky. 29 7.5.2 Rozdeľovači koeficient látky v systéme membrána-voda Kmw (alebo Ksw) Kmw (alebo ^sw) hodnoty niektorých, hlavne perzistentných organických látok sa dajú nájsť v literatúre pre polyméry na báze PDMS (DiFilippo and Eganhouse, 2010; Mayer et al., 2000; Paschke and Popp, 2003; Smedes et al., 2009; H. J. Vaes et al., 1996; Yates et al., 2007), LDPE (Adams et al., 2007b; Fernandez et al., 2009; Hale et al., 2010; Lohmann, 2012; Smedes et al., 2009) a polyoxymetylén (POM) (Endo et al., 2011). Podľa termodynamických zákonov rozdelenie organickej látky z vody do organickej fázy (alebo do polyméru) narastá s poklesom teploty a nárastom salinity (Schwarzenbach et al., 1993). V oboch prípadoch sa znižuje rozpustnost' organickej látky vo vode, čoho dôsledkom je nárast hydrofóbnosti organickej látky, a tým aj afinita k hydrofóbnemu materiálu polyméru. Hodnoty ^sw je možné upraviť podľa lokálnych podmienok experimentu, použitím Vaď t Hoffovej rovnice na korekciu vplyvu teploty (Lohmann, 2012; Schwarzenbach et al., 1993): kde ^sw (T) a ^sw (298) sú hodnoty rozdeľovacieho koeficienta pri termodynamickej teplote T a pri 298 K, AHSW je entalpia distribúcie medzi polymér a vodu (kJ/mol) a R je univerzálna plynová konštanta (8.3143 J/mol/K). Podobne možno korigovať vplyv salinity (iónovej sily) na ^sw použitím empirickej Setchenowovej rovnice (Perron et al., 2013), ktorá vyjadruje závislosť rozpustnosti látky vo vode Cwso1 od molárnej iónovej sily roztoku [sol] a takzvanej vysoľovacej konštanty Ks: Iónová sila neovplyvňuje rozpustnost' analytov v hydrofóbnych polyméroch, preto hodnoty Ksw narastajú nepriamo úmerne s klesajúcou rozpustnosťou látky vo vode (Adams et al., 2007b). Pre polyméry na báze silikónovej gumy (PDMS) boli publikované i tieto závislosti Ksw hodnôt od teploty a salinity (Jonker et al., 2015). Pre ďalšie polyméry je potrebné uskutočniť ďalšie merania. (23) »so/ _ 'w ~ (24) 30 hydrofóbnosť Prestup látky do vzorkovača je kontrolovaný difúziou^ v membráne 80% -t—1 q 60% H— ^ 40% 20% 0% Karbamazepín ■ variabilita v rámci laboratória ■ medzilaboratórna variabilita Obrázok 11 Variabilita výsledkov medzilaboratórneho porovnania pasívnych vzorkovačov na rôznych úrovniach analytického postupu: príklad pre látku karbamazepín. STD - roztok štandardu; NPS - pasívny vzorkovač poskytnutý organizátorom; PPS - pasívne vzorkovače účastníkov. (N) - množstvo; (Cw) - koncentrácia vo vode. (Vrana a kol., nepublikované.) 40 Štúdia priniesla niekoľko na prvý pohľad prekvapivých zistení. Vo väčšine prípadov bola pozorovaná medzilaboratórna variabilita asi päťkrát vyššia ako vnútrolaboratórna precíznosť (Obrázok 11). Podobné výsledky merania, získané jednotlivými laboratóriami pre rôzne typy vzorkovačov, a tiež nízka variabilita výsledkov v rámci jednotlivých laboratórií naznačujú, že proces vzorkovania prispieva k celkovej variabilite merania menej ako následná laboratórna analýza. Zúčastnené laboratóriá mali problém s presným stanovením množstva sledovaných látok vo vzorkovači, ako aj s výpočtom koncentrácie látky vo vode z množstva látky sorbovaného vo vzorkovači. Výsledky meraní kompozitných vzoriek vody sa nachádzali v intervale hodnôt pasívneho vzorkovania. V budúcnosti bude potrebné výrazne zlepšiť presnosť pasívneho vzorkovania, najmä pre APV vzorkovače. Celkový záver tejto medzilaboratórnej štúdie je, že proces pasívneho vzorkovania prebieha podľa očakávania s dobrou reprodukovateľnosťou, ale laboratóriá, ktoré vzorky analyzovali, mali vo všeobecnosti problém s laboratórnou analýzou a interpretáciou dát. Závery štúdie boli zaslané na publikáciu v časopise TrAC (Vrana et al., n.d.), stav august 2015). 8.2 Normalizácia pasívneho vzorkovania V posledných rokoch bol urobený značný pokrok v normalizácii vzorkovacích metod. Jedným z výstupov projektu STAMPS financovaného európskou úniou v rámci 5. rámcového programu bol vývoj britskej národnej normy o pasívnom vzorkovaní (BSI, Publicly Available Specification: Determination of priority pollutants in surface water using passive sampling (PAS-61), May 2006., n.d.). Tento dokument sa stal podkladom pre prípravu medzinárodnej normy ISO 5667-23:2011 (ISO, 2011), ktorá predstavuje praktickú príručku pre pasívne vzorkovanie znečisťujúcich látok v povrchových vodách. 9 Využitie pasívneho vzorkovania v regulačnom monitorovaní 9.1 Rámcová smernica o vode Prijatím Rámcovej smernice o vode (RSV) 2000/60/ES (EU, 2000), ktorá nadobudla účinnosť v decembri 2000, sa mení pohľad na ochranu vodných zdrojov. Orientuje sa na vytváranie podmienok pre trvalo udržateľné využívanie vodných zdrojov. Kladie sa dôraz na zachovanie hydroekologických potrieb krajiny. Tento meniaci sa vzťah človeka k vode vyžaduje zo strany štátnych orgánov a inštitúcií zavedenie nových prístupov v chápaní a zabezpečovaní jej ochrany, ktoré vychádzajú z požiadavky zabezpečenia potrebného množstva vody 41 v zodpovedajúcej kvalite pre hospodárske využitie, za podmienky zachovania prírodných funkcií tokov a prírodného ekosystému a krajiny. V Rámcovej smernici o vode boli formulované nasledujúce hlavné ciele: rozšíriť ochranu vôd na všetky vody - tak povrchové ako aj podzemné, dosiahnuť „dobrý stav" všetkých vôd do roku 2015, špecifikovaný v smernici ako environmentálny cieľ, aplikovať reálny integrovaný manažment ľudských aktivít na báze riečnych povodí, uplatňovať kombinovaný prístup pri ochrane vôd, t. j. súbežnú aplikáciu limitných hodnôt emisií a environentálnych noriem kvality (ENK) životného prostredia, vrátane vylúčenia prísunu rizikových prioritných látok do vodného prostredia a znižovaniu obsahu prioritných látok vo vodnom prostredí., dosiahnuť aplikáciu cien za užívanie vôd, zodpovedajúcich „správnym cenám", stimulujúcich trvalo udržateľný rozvoj, dosiahnuť zapojenie celej spoločnosti do implementácie RSV, vypracovať a prijať efektívnu legislatívu. 9.2 Európska stratégia boja proti znečisťovaniu vôd chemickými látkami Znečistenie povrchových vôd chemickými látkami môže narúšať vodné ekosystémy a spôsobovať úbytok biotopov a zníženie biodiverzity. Znečisťujúce látky sa môžu hromadiť v potravnom reťazci a škodiť dravcom, ktoré konzumujú kontaminované ryby. Ľudia sú vystavení znečisťujúcim látkam konzumáciou rýb, pitnej vody a prípadne aj rekreačnými aktivitami. Znečisťujúce látky sa môžu nachádzať v prostredí mnoho rokov potom, ako boli zakázané. Niektoré sa môžu transportovať na veľké vzdialenosti a možno ich nájsť i v odľahlých oblastiach. Znečisťujúce látky môžu prenikať do životného prostredia z rôznych zdrojov, napríklad z poľnohospodárstva, priemyslu, spaľovaním, ako produkty alebo ako neúmyselne vypúšťané vedľajšie produkty. Mohli byť vypúšťané v minulosti, alebo sa aj naďalej uvoľňujú z výrobkov používaných v každodennom živote. Stratégia boja proti znečisťovaniu vôd chemickými látkami je vytýčená v článku 16 Rámcovej smernice o vode 2000/60/ES (RSV) (EU, 2000). Ako prvý krok tejto stratégie bol prijatý zoznam prioritných látok (EU, 2001), ktorý identifikoval 33 látok alebo skupín látok prioritného záujmu v povrchových vodách v celej Európskej únii kvôli ich rozšírenému používaniu a ich vysokým koncentráciám v riekach, jazerá, brakických a pobrežných vodách. 42 Tento zoznam je revidovaný každé štyri roky a podľa potreby aktualizovaný. Aktuálny zoznam zahŕňa hlavne organické zlúčeniny vrátane rôznych pesticídov, niektoré polycyklické aromatické uhľovodíky (PAU), benzén, halogenované rozpúšťadlá, spomaľovače horenia, zmäkčovadlá, povrchovo aktívne látky, antivegetatívne prípravky a aj niektoré ťažké kovy. 9.3 Hodnotenie stavu znečistenia povrchových vôd prioritnými látkami Európska komisia prijala Smernicu 2008/105/ES o environmentálnych normách kvality v oblasti vodnej politiky (EU, 2008). Táto smernica stanovuje limity na koncentrácie v povrchových vodách pre 41 nebezpečných chemických látok vrátane 33 prioritných látok a 8 ďalších znečisťujúcich látok, ktoré predstavujú významné riziko pre zdravie zvierat a rastlín vo vodnom prostredí a pre ľudské zdravie. Má za cieľ zabezpečiť vysokú úroveň ochrany proti rizikám pochádzajúcim z týchto 41 látok a stanovuje pre ne environmentálne normy kvality (ENK) na európskej úrovni. Okrem toho členské štáty EU ustanovujú ENK pre ďalšie syntetické a nesyntetické špecifické znečisťujúce látky relevantné pre jednotlivé povodia, ktoré môžu mať škodlivý účinok na biologickú kvalitu, a ktoré sú vypúšťané do povrchových vôd vo významných množstvách. Podľa RSV dodržiavanie ENKs pre prioritné látky je súčasťou hodnotenia chemického stavu útvarov povrchových vôd. Dodržiavanie ENK pre špecifické znečisťujúce látky je súčasťou hodnotenia ekologického stavu. Pre dodržiavanie predpisov na hodnotenie stavu vôd boli prijaté ENK pre vnútrozemské povrchové vody (rieky a jazerá) a ďalšie povrchové vôd (prechodné, pobrežné a teritoriálne vody). Boli stanovené dva druhy ENK: ročná priemerná koncentrácia (RP-ENK) pre ochranu proti dlhodobým a chronickým účinkom, a maximálne prípustná koncentrácia (NPK-ENK), aby sa predišlo nezvratným vážnym dôsledkom pre ekosystémy v dôsledku akútnej krátkodobej expozície. Vzhľadom na nedostatočný rozsah spoľahlivých informácií o koncentráciách prioritných látok v živých organizmoch a v sedimentoch na úrovni Spoločenstva, ako aj na skutočnosť, že informácie o povrchových vodách poskytujú dostatočný základ pre zabezpečenie komplexnej ochrany a účinnej kontroly znečistenia, hodnoty ENK boli v tomto štádiu pre väčšinu látok odvodené pre povrchové vody. V prípade troch prioritných látok (ortuť, hexachlórbenzén a hexachlórbutadién) boli ENK odvodené pre koncentrácie v organizmoch (biote). S výnimkou kadmia, olova, ortuti a niklu ENK sú vyjadrené ako celková koncentrácia stanovená vo vzorke vody. V prípade kovov ENK odkazujú na koncentráciu rozpustených látok, t.j. koncentráciu v kvapalnej fáze vzorky vody získanej filtráciou cez 0.45 um filter. 43 Sediment a vodné organizmy (biota) sú tiež dôležitými matricami pre monitorovanie niektorých prioritných látok a iných znečisťujúcich látok, ktoré majú tendenciu hromadiť sa v nich, s cieľom posúdiť dlhodobé vplyvy ľudskej činnosti a časové trendy. Cieľom monitorovania je zabezpečiť, aby sa existujúce úrovne kontaminácie v živých organizmoch a v sedimentoch nezvyšovali. V tejto súvislosti je relevantné monitorovať v sedimente a biote látky s akumulačným potenciálom ako sú polybrómované difenylétery (PBDE), C10-C13 chlóralkány, bis(2-etylhexyl)ftalát, hexachlórbenzén, hexachlórbutadién, hexachlór-cyklohexán, pentachlórbenzén, polycyklické aromatické uhľovodíky, tributylciničitý katión a kovy kadmium, olovo a ortuť. V roku 2013 bola prijatá Smernica Európskeho parlamentu a Rady 2013/39/EU, ktorou sa menia smernice 2000/60/EC a 2008/105/EC, pokiaľ ide o prioritné látky v oblasti vodnej politiky (EU, 2013). Obsahom tejto smernice je rozšírenie zoznamu prioritných látok o 12 nových látok a aktualizácia hodnôt ENK pre prioritné láty v povrchových vodách. Berúc do úvahy tendenciu niektorých látok bioakumulovať sa, pre 8 prioritných látok boli zavedené ENK hodnoty ako maximálne prípustné koncentrácie v biote (v mäkkýšoch alebo v rybách). Na základe smernice musia členské štáty postupne zaviesť program na monitorovanie koncentrácie týchto látok látok v živých organizmoch alebo vo vode, a hodnotiť, či stav povrchových vôd vyhovuje novo zavedeným ENK. ENK pre matricu „biota" sú odvodené ako koncentrácie v rybách, s výnimkou pre polycyklické aromatické uhľovodíy, kde sa uvádza odkaz na ryby, kôrovce a mäkkýše (v súlade s právnymi predpismi o bezpečnosti potravín). Členské štáty EU sa môžu rozhodnúť, používať pri hodnotení stavu povrchových vôd ENK v inej matrici, ako je špecifikovaná v smernici 2013/39/EU, prípade pre iné druhy živočíchov, ako sú uvedené v smernici. V prípadoch, kde je ENK nastavená pre živé organizmy, je dovolené, aby príslušné normy bolo možno odvodiť ako ekvivalentné koncentrácie vo vodnom stĺpci (pomocou biokoncentračného faktora (BCF)/biomagnifikačného faktora (BMF ) alebo bioakumulačného faktora (BAF)). Jednotlivé členské štáty EU sa môžu rozhodnúť, v ktorej matrici budú monitorovať prioritné látky za účelom hodnotenia stavu vôd, ale musia pritom zvážiť rad praktických a etických otázok, ako je napríklad nutnosť merať extrémne nízke koncentrácie látok vo vodách, alebo potreba odlovu veľkého množstva rýb na účel monitorovania. 44 9.4 Požiadavky na analytické metódy vo vzťahu k hodnoteniu povrchových vôd podľa Rámcovej Smernice o vode Pri kontrole kvality a stavu vôd je nevyhnutné, aby boli zabezpečené vyhovujúce analytické nástroje, umožňujúce stanovovať hladiny znečisťujúcich látok sledovaných na medzinárodnej úrovni (napr. monitoring hraničných tokov), ako aj na vnútroštátnej úrovni (potreby dané špecifickými zdrojmi znečistenia). Všetky analytické metódy, ktoré sa použijú na účely programov chemického monitorovania stavu vôd, musia spĺňať určité minimálne pracovné kritériá vrátane pravidiel neistoty meraní a limitov kvantifikácie metód (EU, 2009). Všetky metódy analýzy vrátane laboratórnych, terénnych a on-line testov používaných na účely programov sledovania chemických látok, uskutočňovaných v súlade s RSV, majú byť validované a dokumentované v súlade s normou EN ISO/IEC-17025 alebo inými zodpovedajúcimi normami uznanými na medzinárodnej úrovni. Všetky používané analytické metódy stanovenia sa musia opierať o hodnotu neistoty merania 50 % alebo nižšiu (k = 2) odhadnutú na koncentračnej úrovni príslušnej ENK a limit kvantifikácie rovný alebo nižší ako 30 % príslušnej ENK. Ak v prípade niektorého parametra neexistuje príslušná ENK, alebo ak neexistuje analytická metóda spĺňajúca vyššie uvedené minimálne pracovné kritériá, príslušná smernica vyžaduje, aby sa monitorovanie uskutočňovalo s použitím najlepších dostupných techník, ktoré nespôsobujú prílišné zvyšovanie nákladov. Laboratóriá musia preukázať svoju spôsobilosť na analyzovanie príslušných látok účasťou na programoch testovania odbornosti, ktoré zahŕňajú analytické metódy na úrovni koncentrácií, ktoré sú reprezentatívne pre programy monitorovania chemických látok uskutočňované podľa RSV a analýzou dostupných referenčných materiálov, ktoré reprezentujú odoberané vzorky obsahujúce primerané koncentrácie vzhľadom na príslušné ENK (EU, 2009). 9.5 Použiteľnosť pasívneho vzorkovania na monitorovanie prioritných látok podľa RSV Aktualizovaná smernica o environmentálnych normách kvality odporúča členským štátom aktívne postupovať pri implementácii inovatívnych monitorovacích nástrojov na hodnotenie koncentrácií a trendov prioritných látok v povrchových vodách: „Nové metódy monitorovania, ako napríklad pasívne odbery vzoriek a iné nástroje, sa z hľadiska budúceho uplatňovania javia ako sľubné a mali by sa preto ďalej rozvíjať" (EU, 2013). 45 Podobne, ako je akumulácia hydrofóbnych/lipofilných látok do tkanív vodných živočíchov hnaná lepšou rozpustnost'ou týchto látok v lipidoch ako vo vode, je aj prestup týchto látok z vody do pasívneho vzorkovača založený na lepšej rozpustnosti organických látok v materiáli, z ktorého sú vzorkovače zhotovené. Tieto vlastnosti pasívnych vzorkovačov určujú ich potenciálne využitie v regulačnom monitoringu, najmä pre hydrofóbne látky. Potenciál pasívneho vzorkovania na podporu monitorovania znečisťujúcich látok pri implementácii RSV bola prvýkrát diskutovaný na ad hoc expertnom stretnutí, organizovanom v roku 2009 asociáciiou NORMAN ("NORMAN Expert Group Meeting: Passive Sampling of Emerging Pollutants: state of the art and perspectives 27 May 2009 - Prague, The Czech Republic," 2009) a v pozičnom dokumente, ktorý bol spracovaný na základe tejto diskusie (Vrana et al., 2010). Ďalšími iniciatívami, kde bola riešená problematika využitia pasívnych vzorkovačov v regulačnom monitoringu bol „Utrechtský seminár" ("Include passive sampling in WFD-monitoring? Passive Sampling Workshop, Utrecht, The Netherlands 9-10 November 2011," 2011), workshop organizovaný SETAC o metódach pasívneho vzorkovania v sedimentoc h (Parkerton et al., 2012), ICES workshop o pasívnom vzorkovaní a pasívnom dávkovaní (International Council for the Exploration of the Sea, 2013), workshop organizovaný RECETOXom a asociáciou NORMAN ("Linking Environmental Quality Standards and Passive Sampling," 2013) a napokon workshop organizovaný asociáciou NORMAN v spolupráci s francúzskym referenčným laboratóriom pre oblasť vôd AQUAREF (Miégeetal., 2015, 2014). Pasívne vzorkovanie je považované za monitorovací nástroj - rovnovážnu (alebo nerovnovážnu) extrakčnú techniku, ktorá umožňuje stanoviť koncentrácie voľne rozpustených prioritných látok vo vode. Alternatívne sa na vzorkovač môže nahliadať ako na referenčnú matricu (zložku životného prostredia), ktorá je homogénna a má dobre definované vlastnosti, ktoré sú málo ovplyvnené okolitým prostredím. Výsledky meraní látok pasívnym vzorkovaním sa môžu prepočítať (konvertovať) v súlade s teóriou rovnovážnej distribúcie na ekvivalentné koncentrácie látky v iných zložkách životného prostredia. Najčastejšie ide o prepočet koncentrácie látky vo vzorkovači na koncentráciu látky voľne rozpustenej vo vode, v princípe je ale možné urobiť i prepočet na rovnovážnu koncentráciu látky v lipide vodných živočíchov (Jahnke et al., 2008). 46 Tabulka 1. Výhody (+) a nevýhody (-) priameho a pasívneho odberu vzoriek vody a možné riešenia Priamy odber vzoriek vody Pasívne vzorkovanie Analytická porovnateľnosť výsledkov meraní + tradičný prístup s dlhodobou históriou vývoja metód, dostupnosť validovaných metód a interkalibračných štúdií +/-tréning laboratórií a ďalšia kalibrácia vzorkovačov umožní výrazne zlepšiť vzájomnú porovnateľnosť meraní Vzájomná porovnateľnosť výsledkov meraní z rôznych vodných útvarov vzorky vody z rôznych útvarov, v rôznych režimoch toku a v rôznych obdobiach roka majú odlišné zloženie matrice; celková koncentrácia nedostatočne reflektuje riziko, ktoré znečisťujúce látky predstavujú pre vodné živočíchy +/-pasívny vzorkovač pozostáva z materiálu (matrice) s dobre definovaným zložením v rôznych podmienkach prostredia; výsledky meraní sú navzájom priamo porovnateľné a umožňujú identifikáciu priestorových a časových trendov znečisťujúcich látok vo vodách Meranie veľmi nízkych koncentrácií bežne používaný odber malého objemu vody (niekoľko litrov na účel analýzy) nieje vhodný na meranie ultrastopových koncentrácií látok vo vodách + pasívna akumulácia znečisťujúcich látok do vzorkovača z veľkého objemu vody (až niekoľko tisíc litrov) umožňuje dosiahnuť extrémne nízke medze stanovenia látok vo vode Reprezentativnost' vzoriek výsledok merania z bodového odberu vody reprezentuje iba koncentráciu za veľmi krátky časový úsek + pasívne vzorkovanie poskytuje integratívnu vzorku, ktorá je menej citlivá na krátkodobé zmeny vo vzorkovanom vodnom útvare Z koncentrácie látky v pasívnom vzorkovači je možné odhadnúť voľne rozpustenú koncentráciu látky rozpustenej látky, ktorá predstavuje hnaciu silu pre biokoncentráciu látok 47 do tkanív vodných živočíchov. Pasívne vzorkovače teda umožňujú stanoviť koncentráciu, v ktorej sú exponované vodné živočíchy na najnižších trofických úrovniach. Výsledky z rovnovážneho pasívneho vzorkovania je možné kovertovať na koncentrácie v lipide organizmu, ktorý je v rovnováhe s prostredím, v ktorom žije. Tento prístup je podobný, ako keď sa koncentrácie organických látok v sedimente normalizujú na obsah organického uhlíka a následne konvertujú na koncentrácie v iných environmentálnych matriciach. Na rozdiel od sedimentov v prípade pasívnych vzorkovačov nie je potrebné brať do úvahy variabilnú povahu organického uhlíka, pretože polymérne sorbenty používané na konštrukciu pasívnych vzorkovačov majú dobre definované a konštantné vlastnosti. Aplikácia pasívnych vzorkovačov môže pomôcť zefektívniť monitorovanie a následné hodnotenie a kvality vody, znížiť náklady spojené s monitorovaním, najmä pre látky s extrémne nízkymi koncentráciami vo vodnej fáze a pre látky, ktorých koncentrácie kolíšu v čase. Integratívne pasívne vzorkovanie má oproti bodovým odberom vzoriek výhodu, pretože poskytuje priemernú koncentráciu analytu vo vzorkovanej matrici za dlhšie časové obdobie. Nižšie sú uvedené niektoré požiadavky na monitorovanie vôd a porovnanie pre priamy a pasívny spôsob odberu vzoriek. 9.5.1 Hodnotenie súladu s ENK pre matricu voda Na hodnotenie chemického stavu vodného útvaru podľa Rámcovej smernice o vode sú určené ENK (EU, 2008)(EU, 2013). Tie sa zvyčajne vyjadrujú ako ako AA-ENK (ročný priemer) a MAC-NEK (maximálna prípustná koncentrácia). Ten je zvyčajne vyjadrená ako vysoko percentil, napr 90%(Hanke et al., 2009). Chemické monitorovanie diskrétnych environmentálnych vzoriek väčšinou spĺňa legislatívne požiadavky na analytické metódy (EU, 2009), ale existujú situácie, kedy pasívne vzorkovanie môže byť veľmi užitočné a doplniť chýbajúce informácie. Je zrejmé, že keď medza stanovenia vo vzorkách vody odohraných klasickým spôsobom je vyššia ako 30% príslušnej ENK (a preto nie je možné hodnotenie stavu vodného útvaru v súlade s vyššie uvedenými legislatívnymi normami) použitie metódy pasívneho vzorkovania, ktorá má vyhovujúcu medzu stanoveni, predstavuje logickú alternatívu. V prípade, že medza stanovenia metódy pasívneho vzorkovania + príslušná neistota neprekračuje príslušnú hodnotu ENK, je možné vykonať hodnotenie chemického stavu útvaru vôd pre danú látku i v prípade, že rozšírená neistota merania pri koncentrácii rovnej ENK 48 nespĺňa súčasné právu požiadavky, t.j. aby neistota stanovenia bola nižšia ako 50% (£=2)(EU, 2009). Ďalším prípadom, keď môže byť výhodné použiť pasívne vzorkovanie, je ak koncentrácie kolíšu v čase, čo nastáva najmä u prípravkov na ochranu rastlín v malých vodných útvaroch. Vtákom prípade je relevantné kontrolovať dodržanie MAC-ENK. Pasívne vzorkovanie síce poskytuje informáciu iba o časovo váženom priemere koncentrácie počas expozície (alebo počas polčasu ustálenia rovnováhy, ak je polčas kratší ako doba expozície), ale integratívny charakter vzorkovania umožňuje „zbadať" krátkodobý pulzný nárast koncentrácie. Faktom je, že intenzita ani trvanie takéhoto pulzného nárastu koncentrácie sa nedá odvodiť z jedného odberu pasívneho vzorkovača. Napriek tomu pasívne vzorkovanie umožňuje výrazne znížiť pravdepodobnosť, že takáto udalosť ostane nepovšimnutá, ako to často býva v prípade použitia konvenčného bodového odberu vzoriek s mesačnou frekvenciou odberu. Na odberových profiloch, kde pasívne vzorkovače namerajú najvyššie priemerné koncentrácie, je následne možné naplánovať intenzívnejšie vzorkovanie (napr. pomocou automatického vzorkovača), ktoré potvrdí alebo vyvráti prekročenie príslušnej ENK. Organické znečisťujúce látky vo vodách sa môžu vyskytovať, v závislosti od charakteru ich vypúšťania, s variabilnými alebo relatívne konštantnými koncentráciami. Na základe fyzikálnochemických vlastností sa môžu látky rozdeliť na hydrofilné (log Kow < 4) ahydrofóbne (log^ow >4), a pre tieto dve skupiny je nutné použiť dva rôzne typy pasívnych vzorkovačov. Tabulka 2 poskytuje prehľad použiteľnosti pasívnych vzorkovačov pri monitorovaní látok vo vodách. V prípadoch, keď koncentrácie látok v prostredí kolíšu iba málo, pasívne vzorkovanie je preferovanou technikou odberu, a ich hlavná výhoda je v možnosti dosiahnuť veľmi nízke hodnoty medze stanovenia. Prípadné výkyvy koncentrácií sú vo vzorke integrované, preto získaná vzorka dobre reprezentuje priemerné zloženie vody vo vodnom útvare za dlhšie časové obdobie. Ďalšiu komplikáciu v hodnotení stavu poďla RSV predstvuje fakt, že podľa smerníc, ktoré v súčasnosti platia (EU, 2008)(EU, 2013), hodnotenie súladu s ENK sa má vykonávať porovnaním „celkovej" koncentrácie látky vo vodnom stĺpci, zatiaľ čo pasívne vzorkovanie poskytuje iba informáciu o rozpustnej koncentrácii Cfree. V prípade, že koncentrácia celkového organického uhlíka (TOC) vo vodnom stĺpci neprekročí hodnotu 10 mg/l, vyskytujú sa látky, ktorých log Kov/ < 5 vo vode prevažne v rozpustenej forme (teda nenaviazané na častice) (Obrázok 12). Pre takéto látky je meranie pomocou pasívneho vzorkovania možné použiť na priame porovnanie s hodnotou ENK. 49 120% Obrázok 12 Odhad podielu rozpustenej látky (Cfree) vo vode v závislosti od jej hydrofóbnosti (log Kow) pre tri rôzne koncentrácie celkového organického uhlíka [TOC] vo vodnom stĺpci. Na odhad bol použitý vzťah ÄT0C = 0.41xäTow (Karickhoff, 1981) a model Cfree/Ctotal = 1/(1+[TOC]xäToc). Iná situácia vzniká pre hydrofóbnejšie látky, pre ktoré sa voľne rozpustený podiel vo vode prudko znižuje s rastúcou hodnotou log Kow. Vtákom prípade je riešením odvodiť z legislatívne ukotvenej ENK pre celkovú koncentráciu „odvodenú" hodnotu normy kvality, ktorá poskytuje rovnakú úroveň ochrany vodných živočíchov ako pôvodná ENK, ale vzťahuje sa na rozpustenú koncentráciu látky vo vode ENKVOda,c free (Whitehouse and Paya-Perez, 2011). Hodnoty Cfree, získané pasívnym vzorkovaním, je v takomto prípade možné porovnať s odvodenou hodnotou ENKVOda,cfree i pre hydrofóbnejšie látky. Tabulka 2. Použiteľnosť pasívneho vzorkovania na hodnotenie stavu znečistenia vôd log Kow Konštantná koncentrácia Fluktuujúca koncentrácia <4 Výhody pasívneho vzorkovania oproti štandardnému postupu vzorkovania vôd bodovými odbermi sú obmedzené Adsorpčné pasívne vzorkovanie má rolu skríningového nástroja, ale môže byť preferovanou metódou v prípade, že neistota vzorkovania je menšia ako variabilita koncentrácie vo vode >4 Rozdeľovacie pasívne vzorkovače umožňujú stanovenie voľne rozpustenej koncentrácie hydrofóbnych látok vo vode, hlavne v prípadoch extrémne nízkych hodnôt environmentálnych noriem kvality Rozdeľovacie pasívne vzorkovače poskytujú informáciu o priemernej hodnote koncentrácie, ale neumožňujú stanoviť maximálnu koncentráciu látky počas krátkodobého výkyvu koncentrácie 50 9.5.2 Hodnotenie súladu s ENK pre matricu biota Smernica 2013/39/EU (EU, 2013) umožňuje pre látky s bioakumulačným potenciálom použiť na hodnotenie chemického stavu vôd koncentrácie namerané v tkanivách vodných živočíchov. Pre skupinu 8 látok určuje i príslušné ENK pre biotu na úrovni Spoločenstva. Výhodou použitia bioty (napr. rýb) pri monitorovaní chemických látok je, že mnohé z týchto látok sa vo vode vyskytujú len vo veľmi nízkych koncentráciách, ale v dôsledku bioakumulácie sú ich koncentrácie dobre merateľné v tkanivách vodných živočíchov použitím dostupných analytických metód. Ďalšou výhodou tohto prístupu je, že meranie koncentrácií v tkanive živočíchov umožňuje priamo hodnotiť ich expozíciu, ak tieto látky nie sú aktívne metabolizované. Použitie organizmov na monitorovanie chemických látok však prináša niekolko problémov: neistota spôsobená variabilitou vzorkovaných druhov, veľkosti, veku, pohlavia, fyziologického stavu a trofickej úrovne organizmov, môže zaniesť do procesu hodnotenia stavu vôd významné skreslenie. Ďalším problémom je výsledná variabilita, ktorá komplikuje hodnotenie časových alebo priestorových trendov sledovaných látok, čo môže obmedziť možnosť porovnať výsledky meraní pre rovnakú látku medzi regiónmi • druhy rýb alebo iných vodných živočíchov, potrebné pre monitorovanie chemických látok, nemusia byť k dispozícii na všetkých odberových miestach monitorovanie bioty je ekonomicky (a prakticky) uskutočniteľné iba s nižšou frekvenciou odberov, ako je frekvencia odberov vzoriek vody • je potrebné deštrukčné vzorkovanie (nutnosť zabitia živočíchov odohraných na účel monitoringu), ktoré v prípade intenzívneho odlovu rýb za účelom chemického monitorovania môže dokonca ohroziť miestne populácie rýb Hoci je dostupná technická príručka Európskej komisie na monitorovanie chemických látok v biote v povrchových vodách (Deutsch et al., 2014), dáta získané týmto spôsobom budú veľmi pravdepodobne zaťažené značnou variabilitou, a v dôsledku toho i hodnotenie vôd bude zaťažené zvýšenou neistotou. To bude komplikovať následné rozhodnutia vodohospodárov pri nastavení opatrení na zlepšenie kvality vôd. Potenciálnym riešením problémov spojených s chemickým monitorovaním v živých organizmoch je aplikovať abiotické metódy monitorovania, napr. pomocou pasívnych vzorkovačov, ktoré poskytnú "biomimetické" meranie znečisťujúcich látok, t.j. budú 51 simulovať proces biokoncentrácie znečisťujúcich látok z vody do vodných organizmov s nízkou inherentnou variabilitou. Rozdeľovacie koeficienty vzorkovač-voda (Ksw) sú v prvom priblížení rovné hodnote rozdeľovačieho koeficienta v systéme oktanol-voda (Kow). Vzhľadom na to, že parameter Kow sa používa ako surogát lipidov, ktorý sa používa na opis biokoncentrácie látok do organizmov, existuje i vzťah medzi akumuláciou látok do pasívnych vzorkovačov a do organizmov. Za predpokladu, že kvantitatívne vzťahy sú dostatočne charakterizované (Rovnica (37), výsledky meraní z RPV umožňujú predikovať koncentrácie znečisťujúcich látok vbiote a môžu sa potenciálne použiť ako náhrada monitorovania chemických látok pomocou bioty: K CPS CJipid ENK^ ENKJipid Sjipid Clipid Cbiota ENKlipid ENKbiota kde Ä"s,iiPid je rozdeľovači koeficient látky medzi vzorkovačom a lipidom, CuPidje koncentrácia látky v biote, vztiahnutá na koncentráciu v lipide, /uPid je podiel lipidu v tkanive organizmu a ENKiipid je hodnota ENKbi0ta vyjadrená na základe koncentrácie v lipide. Návrh, aby sa konverzia výsledkov meraní pasívnym vzorkovaním prepočítala pomocou vyššie uvedenej rovnice na ekvivalentnú koncentráciu v modelovom lipide (napr. trioleíne), a aby aj príslušná hodnota ENK bola vyjadrená ako koncentrácia v lipide (Jahnke et al., 2008), žiaľ nenachádza v súčasnosti podporu u expertov, ktorí sú zodpovední za prípravu technickej dokumentácie na podporu implementácie RSV. Jedným z dôvodov je, že tento prístup vnáša do procesu hodnotenia stavu vôd ďalšiu neistotu, a tiež preto, že ciele ochrany takouto ENK by boli ťažko komunikovateľné vo vzťahu k verejnosti. Výhodami takéhoto prístupu je, že ENKiipid má vzťah ku koncentrácii látok vo vodných organizmoch, a tiež, že pre veľmi hydrofóbne látky hodnota ENK nepredstavuje extrémne nízku koncentráciu, ako je to často v prípade ENKVOda- V súčasnosti je pomerne ťažké vysvetliť neodborníkovi v tejto oblasti a presvedčiť verejnosť, že koncentrácie látok vo vode, ktoré sa pohybujú rádovo v pikogramoch na liter, môžu spôsobovať poškodenie ekosystému a sú škodlivé. Vzhľadom nato, že biomimetická extrakcia je založená na jednoduchom fyzikálnom procese rozďeľovacej rovnováhy, nemôže dokonale opísať proces bioakumulácie, ktorý zahŕňa i akumuláciu v potravnom reťazci, a tiež metabolizmus. Preto pasívne vzorkovanie simuluje iba proces biokoncentrácie v organizmoch, ktoré sledované látky nemetabolizujú (Verbruggen et al., 2000, 1999). Vo všeobecnosti nie je ani možné priamo porovnávať koncentráciu 52 získanú pasívnym vzorkovaním s hodnotou ENKbiota- Hoci pre niektoré druhy živočíchov sa našla veľmi dobrá korelácia medzi koncentráciami látok v pasívnych vzorkovačoch a v biote (Smedes, 2007), pre iné druhy bola táto korelácia slabá (Ashton et al., 2012). V princípe je možné nepriame hodnotenie súladu s ENK pre matricu biota pomocou pasívneho vzorkovania. RPV poskytujú spoľahlivé meranie (so známou neistotou) voľne rozpustenej koncentrácie Cfree väčšiny látok, ktoré majú tendenciu akumulovať sa v biote. Vychádzajúc z vyššie uvedenej tézy, že Cfree je naj relevantnej ší parameter expozície vodných organizmov účinkom znečisťujúcich látok, koncentrácie látok vo vzorkovači i v biote sú priamo úmerné hodnote Cfree (Obrázok 13), a tento vzťah je možné použiť ako spoločný menovateľ pri hodnotení súladu s ENK. koncentrácia v biote rovnovážna koncentrácia v pasívnom vzorkovači Obrázok 13 Distribúcia organických látok medzi vodou (voľne rozpustná koncentrácia), vodnou biotou a rozdeľovačmi pasívnym vzorkovačom. Zatiaľ čo vzťah koncentrácie Cps ku Cfree je pomerne jednoduchý a dá sa charakterizovať známou neistotou (Lohmann et al., 2012), v prípade bioty je vzťah medzi Cbiota a Cfree oveľa komplexnejší a zahŕňa nielen biokoncentráciu (charakterizovanú BCF), ale aj akumuláciu látky potravou v trofickom reťazci, charakterizovanú bioakumuláciou (BAF) a biomagnifikáciou (BMF). Aby bola možná kontrola súladu s ENKbiota, založenom na monitorovaní Cfree pomocou pasívneho vzorkovania, je potrebné najprv odvodiť potrebné kritérium hodnotenia súladu; t.j. prepočítať hodnotu ENKbiota na ENKVOda,cfree, ktorá poskytne ekvivalentnú ochranu vodných 53 organizmov pred negatívnymi účinkami sledovaných látok. Problém tkvie v tom, že sa pritom musia použiť hodnoty BAF, BMF a TMF, ktoré sú zaťažené veľkou variabilitou a môžu potenciálne vnášať veľkú neistotu do takejto konverzie (Moermond and Verbruggen, 2013). Konverzia je zmysluplná len v prípade, ak je možné variabilitu udržať v rozumných medziach. Je to možné urobiť niekoľkými prístupmi: • aplikáciou konzervatívnych (maximálnych) hodnoty BAF aplikáciou BAF hodnôt, ktoré úzko súvisia s lokálnym ekosystémom v monitorovanom vodnom útvare, aby bola zabezpečená dostatočná ochrana lokálnych vodných živočíchov • starostlivým výberom druhov monitorovaných organizmov z rôznych trofických úrovní, vo vzťahu k cieľom ochrany (t.j. ochrana ľudského zdravia, vodného vtáctva, vodných cicavcov), receptorov, ktoré sú ohrozené, ako aj expozičných ciest Tento spôsob hodnotenia chemického stavu, i spôsob založený na monitorovaní bioty sú oba založené na porovnaní nameraných koncentrácií s ENKbiota, ale použitie pasívneho vzorkovania poskytuje výhodu oproti monitorovaniu bioty, pretože umožňuje vyhnúť sa neistotám, ktoré do procesu hodnotenia vnáša vzorkovanie bioty. Pri zvažovaní, ktorý spôsob je vhodnejší, treba brať do úvahy i neistotu prepočtu hodnoty ENKbiota na ENKVOda,cfree, ako aj neistotu vzorkovania pomocou pasívnych vzorkovačov. Pokiaľ sa v budúcnosti podarí vymedziť dobre definované kritérium ENKVOda,cfree, pasívne vzorkovače môžu zohrať významnú úlohu v regulačnom monitorovaní znečisťujúcich látok, ako súčasť tzv. viacstupňového procesu hodnotenia stavu vôd. 9.5.3 Úloha pasívneho vzorkovania vo viacstupňovom procese hodnotenia stavu vôd Pasívne vzorkovače je možné použiť v prvom stupni tzv. viacstupňového procesu (tiered approach) hodnotenia stavu chemického znečistenia vôd a sedimentov (Deutsch et al., 2014). Viacstupňový postup nastavenia monitorovacích programov sa používa, pretože monitorovanie prioritných látok vo vodných organizmoch (v prípade hodnotenia chemického stavu na základe ENK pre matricu biota) si vyžaduje veľké nasadenie vzorkovacích, logistických a analytických kapacít. Preto je potrebné monitorovanie sústrediť na oblasti a vodné útvary, kde je zvýšené riziko prekročenia príslušných ENK. V takýchto oblastiach môže nastať situácia, že požadované druhy živočíchov nie sú k dispozícii, alebo sú 54 k dispozícii v nedostatočnom počte, veľkosti alebo vekovom rozmedzí. Existuje reálne riziko, že práve v oblastiach, kde sú prekročené ENK, sa biota vôbec nemusí vyskytovať. Viacstupňový skríningový prístup umožňuje v niekoľkých krokoch identifikovať problematické oblasti alebo hlavné zdroje rizík pre vodné živočíchy. Týmto postupom sa najprv rôzne geografické oblasti zoradia podľa informácie z dostupných monitorovacích dát alebo z modelovania, a následne sa identifikujú a prioritizujú oblasti/vodné útvary, kde sa očakávajú najvyššie koncentrácie znečisťujúcich látok. V prioritizovaných oblastiach sa následne uskutoční monitorovanie koncentrácií znečisťujúcich látok vo vodných živočíchoch. Prvostupňové hodnotenie môže byť založené na meraniach vo vode, plavenine, dnových sedimentoch, ale najmä meranie pomocou pasívneho vzorkovania môže výrazne spresniť prvotné hodnotenie chemického znečistenia. V prvom stupni sa monitorovanie látok uskutoční iba pomocou pasívnych vzorkovačov a aplikuje sa konzervatívne (najhorší možný scenár znečistenia) hodnotiace kritérium (ENK). Na miestach, kde pasívne vzorkovanie jasne ukazuje na dodržanie príslušných noriem kvality, ďalší intenzívny monitoring nebude potrebný. Len vo vodných útvaroch, kde pasívne vzorkovanie ukazuje prekročenie ENK, je potrebné v druhom kroku uskutočniť detailnejší monitoring, napr. pomocou bioty. Takýto postup by umožnil znížiť závislosť monitoringu na analýze tkanív vodných živočíchov, ale stále by zachoval využitie ENKbi0ta ako regulačnej normy pre hodnotenie stavu vôd, na základe ktorého sa uskutočňujú vodohospodárske opatrenia. 55 10 Závery Všeobecným úvodom do problematiky pasívneho vzorkovania, ktorý je v prílohách doplnený súborom mojich vlastných publikácií a prác, ktoré s témou habilitačnej práce úzko súvisia, som sa pokúsil ukázať, že pasívne vzorkovanie je sľubnou metódou, ktorá má veľký potenciál vo výskume osudu znečisťujúcich látok v životnom prostredí. Naznačil som aj možnosti praktického využitia pasívneho vzorkovania v regulačnom monitorovaní znečisťujúcich látok v povrchových vodách a smery budúceho výskumu, ktoré povedú k splneniu tohoto cieľa. V posledných rokoch bol dosiahnutý výrazný pokrok v porozumení faktorov, ktoré ovládajú akumuláciu kontaminantov do rozdeľovacích pasívnych vzorkovačov. Rovnovážne pasívne vzorkovanie umožňuje priamo porovnať úroveň znečistenia rôznych zložiek životného prostredia chemickými látkami, a preto je táto metóda veľmi vhodná pre štúdium distribúcie a transportu chemických látok v životnom prostredí. Pri vzorkovaní hydrofóbnych látok však ustálenie rovnováhy medzi vzorkovačovm a vodou trvá často veľmi dlho, preto sa na meranie koncentrácií týchto látok vo vode používajú kinetické parametre, ktoré charakterizujú rýchlosť prestupu látky z vody do vzorkovača. Prestup látky cez medznú vrstvu vody je vo všeobecnosti limitujúcim krokom pre akumuláciu hydrofóbnych látok do vzorkovača. Dôsledkom toho je závislosť vzorkovacích rýchlostí Rs od hydrodynamických podmienok na mieste expozície. Žiaľ, vzorkovacie rýchlosti nie je možné presne odhadnúť z lokálnych hydrodynamických podmienok (t.j. z rýchlosti prúdenia a z intenzity turbulencie toku), takže je potrebné použiť in situ kalibračné techniky, ktoré sú založené na použití performančných referenčných látok (PRC). Difúzia cez membránu je limitujúcim krokom akumulácie pre látky s nízkymi hodnotami permeability. Permeabilita je produktom a difúzneho a rozdeľovacieho koeficienta látky v membráne (DxKsw). Vzorkovacie rýchlosti týchto látok závisia iba na teplote a vzorkovacie rýchlosti získané v laboratóriu sa dajú priamo aplikovať v teréne. Vplyv závislosti vzorkovacích rýchlostí od rýchlosti prúdenia vody je možné eliminovať pridaním ďalších transportných bariér do vzorkovača, a tiež použitím polárnych membrán. Dôsledkom takéhoto snaženia je ale zvyčajne dramatický pokles vzorkovacích rýchlostí, čo napokon spôsobuje problémy s detegovateľnosťou sledovaných látok v pasívnom vzorkovači. Odhad vzorkovacích rýchlostí in situ je možný z rýchlosti disipácie PRC látok zo vzorkovača. Tento prístup je obmedzený malým intervalom hydrofóbnosti látok, pre ktoré sú disipačné rýchlostné konštanty stanovitelné. Preto je nutné použiť modely, ktoré extrapolujú 56 vzorkovacie rýchlosti, založené na PRC, pre hydrofóbnejšie látky. Pre spoľahlivý odhad týchto parametrov je potrebné poznať presné experimentálne hodnoty rozdeľovači ch koeficientov látok ^sw, a tiež hodnoty difúznych koeficientov týchto látok v polymérnych materiáloch, z ktorých sú vzorkovače zhotovené. Oveľa menej je známe o procesoch, ktoré ovládajú akumuláciu hydrofilných látok do adsorpčných pasívnych vzorkovačov (APV). Modely, ktoré boli odvodené pre rozdeľovacie pasívne vzorkovače (RPV), sú užitočné aj pre pochopenie funkcie vzorkovačov polárnych látok, aleje potrebné spomenúť niektoré významné rozdiely medzi vzorkovačmi APV a RPV. V literatúre je pomerne málo publikovaných hodnôt sorpčných distribučných koeficientov hydrofilných látok pre sorbenty, ktoré sa používajú v APV (Bäuerlein et al., 2012). Vzhľadom na komplexnosť interakcií medzi sorpčnou fázou a analytom je potrebné vyvíjať nové modely, ktoré by boli schopné odhadnúť tieto parametre zo štruktúry molekúl sledovaných látok i zo štruktúry sorpčných miest adsorbentov. Sorpcia hydrofilných látok do membrán a sorpčnej fázy zahŕňa sorpciu na povrchy, a teda sorpčné izotermy sú vo všeobecnosti nelineárne. Tento jav spôsobuje anizotropnú výmenu látok medzi sorbentom a vodou a tiež kompetíciu o sorpčné miesta na povrchu sorbentu. Vzorkovacie rýchlosti pre vzorkovače hydrofilných látok sú vo všeobecnosti nižšie ako pre hydrofóbne látky, čo má za dôsledok vyššie medze detekcie. Pasívne vzorkovanie má potenciál využitia v regulačnom monitorovaní, pretože umožňuje meranie extrémne nízkych (ale z hľadiska rizík pre životné prostredie a človeka veľmi relevantných!) koncentrácií znečisťujúcich látok vo vodách, poskytuje reprezentatívny obraz o kontaminácii a reflektuje expozíciu vodných organizmov. Rozdeľovacie pasívne vzorkovače už v súčasnosti umožňujú meranie kontaminantov v prostredí s neistotou, ktorá spĺňa požiadavky kladené na metódy, ktoré sa v Európskej únii môžu používať na účel regulačného monitorovania chemických látok vo vodách. Adsorpčné pasívne vzorkovače sú zatiaľ využiteľné hlavne ako nástroj pre skríning znečistenia a pre identifikáciu vodných útvarov so zvýšeným rizikom prekročenia environmentálnych noriem kvality. Dalši výskum a vývoj pasívnych vzorkovačov pre monitorovanie chemického znečistenia vodného prostredia by mal mať hlavný cieľ zabezpečiť presnosť meraní získaných touto metódou. Čiastkovými úlohami tohoto výskumu bude a) pochopenie a kvantitatívny opis funkcie adsorpčných pasívnych vzorkovačov, b) vývoj avalidácia robustných metód stanovenia znečisťujúcich látok v extraktoch pasívnych vzorkovačov, c) stanovenie kalibračných parametrov pasívnych vzorkovačov a vývoj modelov, ktoré umožnia 57 nevychýlený odhad koncentrácie sledovaných látok vo vode alebo v inej relevantnej zložke životného prostredia. 58 11 Zoznam literatúry Adams, R.G., Lohmann, R., Fernandez, L.A., MacFarlane, J.K., Gschwend, P.M., 2007a. Polyethylene devices: Passive samplers for measuring dissolved hydrophobic organic compounds in aquatic environments. Environ. Sci. Technol. 41, 1317-1323. Adams, R.G., Lohmann, R., Fernandez, L.A., MacFarlane, J.K., Gschwend, P.M., 2007b. Polyethylene devices: Passive samplers for measuring dissolved hydrophobic organic compounds in aquatic environments. Environ. Sci. Technol. 41, 1317-1323. Ahn, S., Werner, D., Karapanagioti, H.K., McGlothlin, D.R., Zare, R.N., Luthy, R.G., 2005. Phenanthrene and pyrene sorption and intraparticle diffusion in polyoxymethylene, coke, and activated carbon. Environ. Sci. Technol. 39, 6516-26. Allan, I.J., Booij, K., Paschke, A., Vrana, B., Mills, G. a, Greenwood, R., 2009. Field performance of seven passive sampling devices for monitoring of hydrophobic substances. Environ. Sci. Technol. 43, 5383-90. Allan, I.J., Booij, K., Paschke, A., Vrana, B., Mills, G. a, Greenwood, R., 2010. Short-term exposure testing of six different passive samplers for the monitoring of hydrophobic contaminants in water. J. Environ. Monit. 12, 696-703. Alvarez, D.A., Huckins, J.N., Petty, J.D., Jones-Lepp, T., Stuer-Lauridsen, F., Getting, D.T., Goddard, J.P., Gravell, A., 2007. Chapter 8 Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS), in: R. Greenwood, G.M. and B.V. (Ed.), Comprehensive Analytical Chemistry: Passive Sampling Techniques in Environmental Monitoring. Elsevier, pp. 171-197. Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L., Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development of a Passive, in Situ, Integrative Sampler for Hydrophilic Organic Contaminants in Aquatic Environments. Environ. Toxicol. Chem. 23,1640-1648. Ashton, D., Whitehouse, P., Gravell, A., Donlon, B., Balaam, J., Lyons, B., Walker, P., Fraser, D., 2012. Implementing WFD biota standards - a possible role for passive 59 samplers? Poster presentation, in: SETAC 6th World Congress/SETAC Europe 22nd Annual Meeting. SETAC, Berlin. Assoumani, A., Coquery, M., Liger, L., Mazzella, N., Margoum, C, 2015. Field application of passive SBSE for the monitoring of pesticides in surface waters. Environ. Sei. Pollut. Res. 22, 3997-4008. Banerjee, S., Sugatt, R.H., O'Grady, D.P., 1984. A simple method for determining bioconcentration parameters of hydrophobic compounds. Environ. Sei. Technol. 18, 79-81. Bartkow, M.E., Booij, K., Kennedy, K.E., Müller, J.F., Hawker, D.W., 2005. Passive air sampling theory for semivolatile organic compounds. Chemosphere 60, 170-6. Bäuerlein, P.S., Mansell, J.E., Ter Laak, T.L., De Voogt, P., 2012. Sorption behavior of charged and neutral polar organic compounds on solid phase extraction materials: Which functional group governs sorption? Environ. Sei. Technol. 46, 954-961. Bird, R.B., Stewart, W.E., Lightfoot, E.N., 2007. Transport Phenomena. John Wiley & Sons. Booij, K., 2009. Performance of passive samplers for monitoring priority substances. [WWW Document]. URLhttp://www.ices.dk/reports/MHC/2009/MCWG09.pdf Booij, K., Hofmans, H.E., Fischer, C. V, van Weerlee, E.M., 2003. Temperature-Dependent Uptake Rates of Nonpolar Organic Compounds by Semipermeable Membrane Devices and Low-Density Polyethylene Membranes. Environ. Sei. Technol. 37, 361-366. Booij, K., Sleiderink, H.M., Smedes, F., 1998. Calibrating the Uptake Kinetics of Semipermeable Membrane Devices using Exposure Standards. Environ. Toxicol. Chem. 17,1236-1245. Booij, K., Smedes, F., 2010. An improved method for estimating in situ sampling rates of nonpolar passive samplers. Environ. Sei. Technol. 44, 6789-94. Booij, K., Smedes, F., van Weerlee, E.M., 2002. Spiking of performance reference compounds in low density polyethylene and silicone passive water samplers. Chemosphere 46, 1157-61. 60 Booij, K., Vraná, B., Huckins, J., 2007. Theory, modelling and calibration of passive samplers used in water monitoring. ... Environ. Monit. 48, 141-169. Booij, K., Vraná, B., Huckins, J.N., 2007. Theory, modelling and calibration of passive samplers used in water monitoring, in: Greenwood, R., Mills, G., Vraná, B. (Eds.), Comprehensive Analytical Chemistry 48. Passive Sampling Techniques in Environmental Monitoring. Elsevier, Amsterdam, pp. 141-169. BSI, Publicly Available Specification: Determination of priority pollutants in surface water using passive sampling (PAS-61), May 2006., n.d. Chen, C.-E., Zhang, H., Jones, K.C., 2012. A novel passive water sampler for in situ sampling of antibiotics. J. Environ. Monit. 14, 1523-1530. Chen, Y., Pawliszyn, J., 2003. Time-weighted average passive sampling with a solid-phase microextraction device. Anal. Chem. 75, 2004-10. Chen, Y., Pawliszyn, J., 2004. Kinetics and the on-Site Application of Standards in a Solid-Phase Microextration Fiber. Anal. Chem. 76, 5807-5815. Cornelissen, G., Pettersen, A., Broman, D., Mayer, P., Breedveld, G.D., 2008. Field testing of equilibrium passive samplers to determine freely dissolved native polycyclic aromatic hydrocarbon concentrations. Environ. Toxicol. Chem. 27, 499. Crank, J., 1975. The Mathematics of Diffusion, 2nd editio. ed. Oxford University Press, Oxford. Deutsch, K., Leroy, D., Belpaire, C, Haan, K. den, Vraná, B., Clayton, H., Hanke, G., Ricci, M., Held, A., Gawlik, B., Babut, M., Perceval, O., Lepom, P., Heiss, C, Koschorreck, J., O'Toole, S., Valsecchi, S., Polesello, S., Carere, M., Hulscher, D. ten, Verbruggen, E., Dulio, V., Green, N., Vinas, L., Bellas, J., Lilja, K., Bignert, A., Whitehouse, P., Sumner, K., Law, R., Brant, J., Leverett, D., Merrington, G., 2014. Guidance document No. 32 on Biota Monitoring (The implementation of EQSbiota) under the Water Framework Directive. Common Implementation Strategy for the Water Framework Directive (2000/60/EC)., European Commission. European Commission, Brussels. 61 Di Toro, D.M., Zarba, CS., Hansen, D.J., Berry, W.J., Swartz, R.C., Cowan, C.E., Pavlou, S.P., Allen, H.E., Thomas, N. a., Paquin, P.R., 1991. Technical basis for establishing sediment quality criteria for nonionic organic chemicals using equilibrium partitioning. Environ. Toxicol. Chem. 10, 1541-1583. Difilippo, E.L., Eganhouse, R.P., 2010. Assessment of PDMS-water partition coefficients: Implications for passive environmental sampling of hydrophobic organic compounds. Environ. Sei. Technol. 44, 6917-6925. DiFilippo, E.L., Eganhouse, R.P., 2010. Assessment of PDMS-water partition coefficients: implications for passive environmental sampling of hydrophobic organic compounds. Environ. Sei. Technol. 44, 6917-25. Endo, S., Hale, S.E., Goss, K.-U., Arp, H.P.H., 2011. Equilibrium partition coefficients of diverse polar and nonpolar organic compounds to polyoxymethylene (POM) passive sampling devices. Environ. Sei. Technol. 45, 10124-32. Esteve-Turrillas, F.A., Pastor, A., Yusä, V., de la Guardia, M., 2007. Using semi-permeable membrane devices as passive samplers. TrAC - Trends Anal. Chem. 26, 703-712. Estoppey, N., Omlin, J., Schopfer, A., Esseiva, P., Vermeirssen, E.L.M., Delemont, O., De Alencastro, L.F., 2015. Low density polyethylene (LDPE) passive samplers for the investigation of polychlorinated biphenyl (PCB) point sources in rivers. Chemosphere 118,268-76. EU, 2000. EU, 2000. Directive 2000/60/EC of the European parliament and of the council of 23 October 2000 establishing a framework for community action in the field of water policy. Off. J. Eur. Union L327, 1-72. EU, 2001. Decision No 2455/2001/EC of the European Parliament and of the Council of 20 November 2001 establishing the list of priority substances in the field of water policy and amending Directive 2000/60/EC. Off. J. Eur. Communities L331, 1-5. EU, 2008. Directive 2008/105/EC of the the European parliament and of the council of 16 December 2008 on environmental quality standards in the field of water policy. Off. J. Eur. Union L348, 84-96. 62 EU, 2009. Commission Directive 2009/90/EC of 31 July 2009 laying down, pursuant to Directive 2000/60/EC of the European Parliament and of the Council, technical specifications for chemical analysis and monitoring of water status. Off. J. Eur. Union L 201, 36-38. EU, 2013. Directive 2013/39/EU of the European Parlament and of the Council of 12 August 2013 amending Directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy. Off. J. Eur. Union L226, 1-17. Fernandez, L. a, MacFarlane, J.K., Tcaciuc, A.P., Gschwend, P.M., 2009. Measurement of freely dissolved PAH concentrations in sediment beds using passive sampling with low-density polyethylene strips. Environ. Sci. Technol. 43, 1430-6. Flynn, G.L., Yalkowsky, S.H., 1972. Correlation and prediction of mass transport across membranes I: Influence of alkyl chain length on flux-determining properties of barrier and diffusant. J. Pharm. Sci. 61, 838-852. George, S.C., Thomas, S., 2001. Transport phenomena through polymeric systems. Prog. Polym. Sci. 26, 985-1017. Gonzalez, C, Greenwood, R., Quevauviller, P. (Eds.), 2009. Rapid Chemical and Biological Techniques for Water Monitoring - Water Quality Measurements Series. Wiley, New York. Greenwood, R., Mills, G.A., Vrana, B., 2006. Improving environmental monitoring. A report on the 2nd International Passive Sampling Workshop and Symposium, Bratislava, Slovakia, 3-6 May 2006, and associated satellite workshops. TrAC - Trends Anal. Chem. 25,751-754. Greenwood, R., Mills, G.A., Vrana, B., Allan, I., Aguilar-Martinez, R., Morrison, G., 2007. Monitoring of priority pollutants in water using chemcatcher passive sampling devices, in: Greenwood, R., Mills, G., Vrana, B. (Eds.), Comprehensive Analytical Chemistry Vol. 48. Elsevier, Amsterdam, pp. 199-229. 63 Gustafson, K.E., Dickhut, R.M., 1997. Distribution of polycyclic aromatic hydrocarbons in southern Chesapeake Bay surface water: Evaluation of three methods for determining freely dissolved water concentrations. Environ. Toxicol. Chem. 16, 452-461. Gustafson, K.E., Dickhut, R.M., 1997. Distribution of polycyclic aromatic hydrocarbons in Southern Chesapeake. Environ. Toxicol. Chem. 16, 452-461. Hale, S.E., Martin, T.J., Goss, K.-U., Arp, H.P.H., Werner, D., 2010. Partitioning of organochlorine pesticides from water to polyethylene passive samplers. Environ. Pollut. 158,2511-2517. Hanke, G., Lepom, P., Quevaviller, P. (Eds.), 2009. COMMON IMPLEMENTATION STRATEGY FOR THE WATER FRAMEWORK DIRECTIVE (2000/60/EC) Technical Report - 2009 - 025 Guidance Document No. 19 GUIDANCE ON SURFACE WATER CHEMICAL MONITORING UNDER THE WATER FRAMEWORK DIRECTIVE. Office for Official Publications of the European Communities, Luxembourg. Harman, C, Allan, I.J., Bäuerlein, P.S., 2011. The challenge of exposure correction for polar passive samplers—the PRC and the POCIS. Environ. Sci. Technol. 45, 9120-1. Harman, C, Allan, I.J., Vermeirssen, E.L.M., 2012. Calibration and use of the polar organic chemical integrative sampler-a critical review. Environ. Toxicol. Chem. 31, 2724-2738. Hawthorne, S.B., Jonker, M.T.O., van der Heijden, S. a, Grabanski, C.B., Azzolina, N. a, Miller, D.J., 2011. Measuring picogram per liter concentrations of freely dissolved parent and alkyl PAHs (PAH-34), using passive sampling with polyoxymethylene. Anal. Chem. 83,6754-61. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environ. Sci. Technol. 27, 2489-2496. Huckins, J.N., Petty, J.D., Booij, K., 2006. Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices. Springer, New York, USA. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F. V, Booij, K., Alvarez, D.A., Cranor, W.L., Clark, R.C., Mogensen, B.B., 2002. Development of the permeability/performance 64 reference compound (PRC) approach for in situ calibration of semipermeable membrane devices (SPMDs). Environ. Sci. Technol. 36, 85-91. Huckins, J.N., Petty, J.D., Orazio, C.E., Lebo, J.A., Clark, R.C., Gibson, V.L., Gala, W.R., Echols, K.R., 1999. Determination of uptake kinetics (sampling rates) by lipid-containing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water. Environ. Sci. Technol. 33, 3918-3923. Huckins, J.N., Tubergen, M.W., Manuweera, G.K., 1990. Semipermeable membrane devices containing model lipid: A new approach to monitoring the bioavailability of lipophilic contaminants and estimating their bioconcentration potential. Chemosphere 20, 533-552. Include passive sampling in WFD-monitoring? Passive Sampling Workshop, Utrecht, The Netherlands 9-10 November 2011 [WWW Document], 2011. URL http://www.passivesampling.net/utrechtworkshop/ International Council for the Exploration of the Sea, 2013. Report of the Working Group on Marine Sediments in Relation to Pollution (WGMS). Lowestoft, UK. ISO, 2011. Water quality - sampling - part 23: Guidance on passive sampling in surface waters ISO 5667-23:2011. Jahnke, A., McLachlan, M.S., Mayer, P., 2008. Equilibrium sampling: partitioning of organochlorine compounds from lipids into polydimethylsiloxane. Chemosphere 73, 1575-81. Jonker, M.T.O., van der Heijden, S.A., Kotte, M., Smedes, F., 2015. Quantifying the Effects of Temperature and Salinity on Partitioning of Hydrophobic Organic Chemicals to Silicone Rubber Passive Samplers. Environ. Sci. Technol. Kader, B.., Yaglom, A.., 1972. Heat and mass transfer laws for fully turbulent wall flows. Int. J. Heat Mass Transf. 15, 2329-2351. Karickhoff, S.W., 1981. Semi-empirical estimation of sorption of hydrophobic pollutants on natural sediments and soils. Chemosphere 10, 833-846. 65 Kingston, J.K., Greenwood, R., Mills, G.A., Morrison, G.M., Persson, B.L., 2000. Development of a novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments. J. Environ. Monit. 2, 487-495. Kot-Wasik, A., Zabiegala, B., Urbanowicz, M., Dominiak, E., Wasik, A., Namiešnik, J., 2007. Advances in passive sampling in environmental studies. Anal. Chim. Acta 602, 141-163. Leslie, H.A., Oosthoek, A.J.P., Busser, F.J.M., Kraak, M.H.S., Hermens, J.L.M., 2002. Biomimetic solid-phase microextraction to predict body residues and. Environ. Toxicol. Chem. 21, 229-234. Linking Environmental Quality Standards and Passive Sampling [WWW Document], 2013. URL www.eqsandps.passivesampling.net Lobpreis, T., Vraná, B., Dercová, K., 2009. Innovative approach to monitoring organic contaminants in aqueous environment using passive sampling devices . Inov. přístupy k Monit. Org. Kontam. vo Vodn. prostředí použitím pasivného vzorkovania 103, 548-558. Lohmann, R., 2012. Critical review of low-density polyethylene's partitioning and diffusion coefficients for trace organic contaminants and implications for its use as a passive sampler. Environ. Sci. Technol. 46, 606-618. Lohmann, R., Booij, K., Smedes, F., Vraná, B., 2012. Use of passive sampling devices for monitoring and compliance checking of POP concentrations in water. Environ. Sci. Pollut. Res. 19, 1885-1895. Louch, D., Motlagh, S., Pawliszyn, J., 1992. Dynamics of organic compound extraction from water using liquid-coated fused silica fibers. Anal. Chem. 64, 1187-1199. Lydy, M.J., Landrum, P.F., Oen, A.M., Allinson, M., Smedes, F., Harwood, A.D., Li, H., Maruya, K.A., Liu, J., 2014. Passive sampling methods for contaminated sediments: state of the science for organic contaminants. Integr. Environ. Assess. Manag. 10, 167-78. Mayer, P., Tolls, J., Hermens, L., Mackay, D., 2003. Equilibrium Sampling Devices. Environ. Sci. Technol. 37, 184A-191A. 66 Mayer, P., Vaes, W.H.J., Hermens, J.L.M., 2000. Absorption of hydrophobic compounds into the poly(dimethylsiloxane). Anal. Chem. 72, 459-464. Mazzella, N., Lissalde, S., Moreira, S., Delmas, F., Mazellier, P., Huckins, J.N., 2010. Evaluation of the use of performance reference compounds in an oasis-HLB adsorbent based passive sampler for improving water concentration estimates of polar herbicides in freshwater. Environ. Sei. Technol. 44, 1713-1719. Miěge, C, Mazzella, N., Allan, I, Dulio, V., Smedes, F., Tixier, C, Vermeirssen, E., Brant, J., O'Toole, S., Budzinski, H., Ghestem, J.-P., Staub, P.-F., Lardy-Fontan, S., Gonzalez, J.-L., Coquery, M., Vrana, B., 2015. Position paper on passive sampling techniques for the monitoring of contaminants in the aquatic environment - Achievements to date and perspectives. Trends Environ. Anal. Chem. Miěge, C, Mazzella, N., Coquery, M., Vrana, B., Tixier, C, Dulio, V., 2014. Workshop on Passive Sampling techniques for monitoring of contaminants in the aquatic environment. Achievements to date and future perspectives. Lyon. Miěge, C, Schiavone, S., Dabrin, A., Coquery, M., Mazzella, N., Berho, C, Ghestem, J.-P., Togola, A., Gonzalez, C, Gonzalez, J.-L., Lalere, B., Lardy-Fontan, S., Lepot, B., Munaron, D., Tixier, C, 2012. An in situ intercomparison exercise on passive samplers for monitoring metals, polycyclic aromatic hydrocarbons and pesticides in surface waters. TrAC - Trends Anal. Chem. 36, 128-143. Mills, G.A., Vrana, B., Allan, I., Alvarez, D.A., Huckins, J.N., Greenwood, R., 2007. Trends in Monitoring Pharmaceuticals and Personal-Care Products in the Aquatic Environment by Use of Passive Sampling Devices. Anal. Bioanal. Chem. 387, 1153-1157. Moermond, C.T.A., Verbruggen, E.M.J., 2013. An evaluation of bioaccumulation data for hexachlorobenzene to derive water quality standards according to the EU-WFD methodology. Integr. Environ. Assess. Manag. 9, 87-97. Namiešnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M., Wasik, A., 2005. Passive sampling and/or extraction techniques in environmental analysis: a review. Anal. Bioanal. Chem. 381,279-301. 67 NORMAN Expert Group Meeting: Passive Sampling of Emerging Pollutants: state of the art and perspectives 27 May 2009 - Prague, The Czech Republic [WWW Document], 2009. URL http://www.norman-network.net/?q=node/120 (accessed 8.1.14). Opdyke, B.N., Gust, G., Ledwell, J.R., 1987. Mass transfer from smooth alabaster surfaces in turbulent flows. Geophys. Res. Lett. 14, 1131-1134. Ouyang, G., Chen, Y., Pawliszyn, J., 2005. Time-Weighted Average Water Sampling with a Solid-Phase Microextraction Device. Anal. Chem. 77, 7319-7325. Ouyang, G., Chen, Y., Pawliszyn, J., 2006. Flow-through system for the generation of standard aqueous solution of polycyclic aromatic hydrocarbons. J. Chromatogr. A 1105, 176-179. Ouyang, G., Pawliszyn, J., 2006. SPME in environmental analysis. Anal. Bioanal. Chem. 386, 1059-1073. Ouyang, G., Pawliszyn, J., 2007. Configurations and calibration methods for passive sampling techniques. J. Chromatogr. A 1168, 226-235. Ouyang, G., Zhao, W., Alaee, M., Pawliszyn, J., 2007. Time-weighted average water sampling with a diffusion-based solid-phase microextraction device. J. Chromatogr. A 1138,42-46. Parkerton, T., Maruya, K., Lydy, M., Landrum, P., Peijnenburg, W., Mayer, P., Escher, B., Ghosh, U., Kane-Driscoll, S., Greenberg, M., Chapman, P., 2012. Guidance on passive sampling methods to improve management of contaminated sediments. Summary of a SET AC Technical Workshop. Pensacola FL (USA. Paschke, a, Popp, P., 2003. Solid-phase microextraction fibre-water distribution constants of more hydrophobic organic compounds and their correlations with octanol-water partition coefficients. J. Chromatogr. A 999, 35-42. Perron, M.M., Burgess, R.M., Suuberg, E.M., Cantwell, M.G., Pennell, K.G., 2013. Performance of passive samplers for monitoring estuarine water column concentrations: 1. Contaminants of concern. Environ. Toxicol. Chem. 32, 2182-9. 68 Prest, H., Petty, J.D., Huckins, J.N., 1998. Validity of using Smeipermeable membrane devices for determining aqueous. Environ. Toxicol. Chem. 17, 535-536. Reichenberg, F., Mayer, P., 2006. Two complementary sides of bioavailability: accessibility and chemical activity of organic contaminants in sediments and soils. Environ. Toxicol. Chem. 25, 1239-45. Richardson, B.J., Lam, P.K.S., Zheng, G.J., McCellan, K.E., De Luca-Abbott, S.B., 2002. Biofouling Confounds the Uptake of Trace Organic Contaminants by Semi-permeable Membrane Devices. Mar. Pollut. Bull. 44, 1372-1379. Rusina, T., Smedes, F., Klanová, J., 2010. Diffusion coefficients of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in polydimethylsiloxane and low-density polyethylene polymers. J. Appl. Polym. Sci. 116, 1803-1810. Rusina, T., Smedes, F., Klanová, J., Booij, K., Holoubek, I., 2007. Polymer Selection for Passive Sampling: a Comparison of Critical Properties. Chemosphere 68, 1344-1351. Rusina, T.P., Smedes, F., Kobližková, M., Klanová, J., 2010. Calibration of silicone rubber passive samplers: experimental and modeled relations between sampling rate and compound properties. Environ. Sci. Technol. 44, 362-7. Rusina, T.P., Smedes, F., Kobližková, M., Klanová, J., 2010. Calibration of silicone rubber passive samplers: Experimental and modeled relations between sampling rate and compound properties. Environ. Sci. Technol. 44, 362-367. Sabaliunas, D., Sódergren, A., 1996. Uptake of organochlorien pesticides by solvent-filled cellulose and. Ecotoxicol. Environ. Saf. 35, 150-155. Sara O'Brien, D., Chiswell, B., Mueller, J.F., 2009. A novel method for the in situ calibration of flow effects on a phosphate passive sampler. J. Environ. Monit. 11, 212-219. Schlichting, H., Gersten, K., Gersten, K., 2000. Boundary-Layer Theory. Springer Science & Business Media. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 1993. Environmental Organic Chemistry. Wiley-Interscience, New York, USA. 69 Shaw, M., Eaglesham, G., Mueller, J.F., 2009. Uptake and release of polar compounds in SDB-RPS EmporeGäó disks; implications for their use as passive samplers 75, 1-7. Sjôberg, FL, Bergman, R., Sundelôf, L.-O., 1996. A New Method for Diffusion Measurement in Polymeric Films Based on a Stacked Sheet Concept. Pharm. Res. 13, 1871-1874. Smedes, F., 2007. Monitoring of chlorinated biphenyls and polycyclic aromatic hydrocarbons by passive sampling in concert with deployed mussels, in: Greenwood, R., Mills, G., Vrana, B. (Eds.), Comprehensive Analytical Chemistry 48; Passive Sampling Techniques in Environmental Monitoring. Elsevier, Amsterdam, pp. 407-448. Smedes, F., Booij, K., 2012. Guidelines for passive sampling of hydrophobic contaminants in water using silicone rubber samplers [WWW Document]. ICES Tech. Mar. E Environ. Sci. URL http://info.ices.dk/pubs/times/times52/120621 TPMES 52 Final.pdf Smedes, F., Geertsma, R.W., Van Der Zande, T., Booij, K., 2009. Polymer-water partition coefficients of hydrophobic compounds for passive sampling: Application of cosolvent models for validation. Environ. Sci. Technol. 43, 7047-7054. Sôdergren, A., 1990. Monitoring of persistent, lipophilic pollutants in water and sediment by. Ecotoxicol. Environ. Saf. 19, 143-149. Sôderstrôm, H., Lindberg, R.H., Fick, J., 2009. Strategies for monitoring the emerging polar organic contaminants in water with emphasis on integrative passive sampling. J. Chromatogr. A 1216, 623-630. Son, J.S., Hanratty, T.J., 1967. Limiting relation for the eddy diffusivity close to a wall. AIChE J. 13, 689-696. Stuer-Lauridsen, F., 2005. Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environ. Pollut. 136, 503-524. Tcaciuc, A.P., Apeli, J.N., Gschwend, P.M., 2015. Modeling the transport of organic chemicals between polyethylene passive samplers and water in finite and infinite bath conditions. Environ. Toxicol. Chem. 70 Vaes, H.J., Hamwijk, C, Ramos, E.U., Verhaar, H.J.M., Hermens, J.L.M., 1996. Partitioning of organic chemicals to polyacrylate-coated solid phase. Anal. Chem. 68, 4458-4462. Vaes, W.H.J., Urrestarazu Ramos, E., Verhaar, H.J.M., Seinen, W., Hermens, J.L.M., 1996. Measurement of the Free Concentration Using Solid-Phase Microextraction: Binding to Protein. Anal. Chem. 68, 4463-4467. Van Pinxteren, M., Paschke, A., Popp, P., 2010. Silicone rod and silicone tube sorptive extraction. J. Chromatogr. A 1217, 2589-2598. Verbruggen, E.M.J., Vaes, W.H.J., Parkerton, T.F., Hermens, J.L.M., 2000. Polyacrylate-coated SPME fibers as a tool to simulate body residues and. Environ. Sci. Technol. 34, 324-331. Verbruggen, E.M.J., Van Loon, W.M.G.M., Tonkes, M., Van Duijn, P., Seinen, W., Hermens, J.L.M., 1999. Biomimetic extraction as a tool to identify chemicals with high. Environ. Sci. Technol. 33, 801-806. Vrana, B., Allan, I.J., Greenwood, R., Mills, G.A., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G., 2005. Passive sampling techniques for monitoring pollutants in water. TrAC Trends Anal. Chem. 24, 845-868. Vrana, B., Mills, G., Greenwood, R., Knutsson, J., Svensson, K., Morrison, G., 2005. Performance optimisation of a passive sampler for monitoring hydrophobic organic pollutants in water. J. Environ. Monit. 7, 612-620. Vrana, B., Mills, G.A., Dominiak, E., Greenwood, R., 2006. Calibration of the Chemcatcher passive sampler for the monitoring of priority organic pollutants in water. Environ. Pollut. 142, 333-343. Vrana, B., Mills, G.A., Kotterman, M., Leonards, P., 2006. Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water. Vrana, B., Popp, P., Paschke, A., Schuurmann, G., 2001. Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water. Anal. Chem. 73, 5191-5200. 71 Vrana, B., Schüürmann, G., 2002. Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics. Environ. Sei. Technol. 36, 290-296. Vrana, B., Smedes, F., Prokeš, R., Loos, R., Mazzella, N., Miege, C, Budzinski, H., Vermeirssen, E., Ocelka, T., Gravell, A., Kaserzon, S., n.d. An interlaboratory study on passive sampling of emerging water pollutants. TrAC - Trends Anal. Chem. Vrana, B., Smedes, F., Prokeš, R., Loos, R., Mazzella, N., Miege, C, Budzinski, H., Vermeirssen, E., Ocelka, T., Gravell, A., Mueller, J.F., 2015a. NORMAN Interlaboratory study (ILS) on passive sampling of emerging pollutants: A Chemical Monitoring On Site (CM Onsite) organised by NORMAN Association and JRC in support to the Water Framework Directive implementation. Vrana, B., Smedes, F., Rusina, T., Okonski, K., Allan, L, Grung, M., Hilscherova, K., Novák, J., Tarábek, P., Slobodník, J., 2015b. Passive sampling: chemical analysis and toxicological profiling, in: Liška, L, Wagner, F., Sengl, M., Deutsch, K., Slobodník, J. (Eds.), Joint Danube Survey 3. ICPDR - International Commission for the Protection of the Danube River, Vienna, pp. 304-315. Vrana, B., Vermeirssen, E.L.M., Allan, I., Kohoutek, J., Kennedy, K., Mills, G., Greenwood, R., 2010. Passive sampling of emerging compounds in the environment: state of the art and perspectives [WWW Document]. URL http://www.norman-network.net/sites/default/files/files/Events/2009/2009May27-Prague-PassiveSampling/norman_position_paper_pas_sampling.pdf Weast, R., 1983. Handbook of Chemistry and Physics. CRC Press, Boca Raton, FL. Wennrich, L., Vrana, B., Popp, P., Lorenz, W., 2003. Development of an integrative passive sampler for the monitoring of organic water pollutants. J. Environ. Monit. 5, 813-822. Whitehouse, P., Paya-Perez, A., 2011. Common Implementation Strategyfor the Water Framework Directive (2000/60/EC). Guidance Document No. 27 Technical Guidance For Deriving Environmental Quality Standards., Technical, ed. European Communities, Luxembourg. 72 Worch, E., 1993. Eine neue Gleichung zur berechnung von Diffusionskoeffizienten gelöster Stoffe. Vom Wasser 81, 289-297. Xu, Y., Wang, Z., Ke, R., Khan, S.U., 2005. Accumulation of Organochlorine Pesticides from Water Using Triolein Embedded Cellulose Acetate Membranes. Environ. Sei. Technol. 39,1152-1157. Xu, Y.P., Wang, Z.J., Ke, R.H., Khan, S.U., 2005. Accumulation of Organochlorine Pesticides From Water Using Triolein Embedded Cellulose Acetate Membranes. Environ. Sei. Technol. 39, 1152-1157. Yates, K., Davies, I., Webster, L., Pollard, P., Lawton, L., Moffat, C, 2007. Passive sampling: Partition coefficients for a silicone rubber reference phase. J. Environ. Monit. 9, 1116— 1121. Zhang, H., Davison, W., 1995. Performance-Characteristics of Diffusion Gradients in Thin-Films for the in-Situ Measurement of Trace-Metals in Aqueous- Solution. Anal. Chem. 67,3391-3400. 73 12 Zoznam publikovaných prác k téme habilitačnej práce Habilitačná práca obsahuje v prílohe niektoré vybrané publikácie, ktoré sa priamo vzťahujú k téme práce, t.j. k vývoju metód pasívneho vzorkovania znečisťujúcich látok vo vodnom prostredí. Sú uvedené publikácie, v ktorých som bol prvým autorom, korešpondujúcim autorom alebo spoluautorom. Tieto práce boli publikované od roku 2001 a sú uvedené v chronologickom poradí. Ďalšie rozdelenie prác je na pôvodné vedecké články v časopisoch, kapitoly v odborných knihách a ďalšie práce. 12.1 Pôvodný vedecký článok v časopise 1. Vrana B., Paschke A., Popp P., and Schúúrmann G., Use of semipermeable membrane devices (SPMDs): Determination of bioavailable, organic, waterborne contaminants in the industrial region of Bitterfeld, Saxony-Anhalt, Germany, Environ. Sci. Pollut. Res., 2001, 8, 27-34. 2. Vrana B., Paschke A., and Popp P., Polyaromatic hydrocarbon concentrations and patterns in sediments and surface water of the Mansfeld region, Saxony-Anhalt, Germany, /. Environ. Monit., 2001, 3, 602-609. 3. Vrana B., Popp P., Paschke A., and Schúúrmann G., Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water, Anal. Chem., 2001, 73, 5191-5200. 4. Vrana B. and Schúúrmann G., Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics., Environ. Sci. Technol, 2002, 36, 290-296. 5. Wennrich L., Vrana B., Popp P., and Lorenz W., Development of an integrative passive sampler for the monitoring of organic water pollutants, /. Environ. Monit., 2003, 5, 813-822. 6. Vrana B., Greenwood R., Mills G., Knutsson J., Svensson K., and Morrison G., Performance optimisation of a passive sampler for monitoring hydrophobic organic pollutants in water, /. Environ. Monit., 2005, 7, 612-620. 7. Vrana B., Allan I. J., Greenwood R., Mills G. A., Dominiak E., Svensson K., Knutsson J., and Morrison G., Passive sampling techniques for monitoring pollutants in water, TrAC - Trends Anal. Chem., 2005, 24, 845-868. 74 8. Vrana B., Paschke H., Paschke A., Popp P., and Schuurmann G., Performance of semipermeable membrane devices for sampling of organic contaminants in groundwater, /. Environ. Monit., 2005, 7, 500-508. 9. Vrana B., Paschke A., and Popp P., Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water, Environ. Pollut., 2006, 144, 296-307. 10. Vrana B., Mills G. A., Dominiak E., and Greenwood R., Calibration of the Chemcatcher Passive Sampler for the Monitoring of Priority Organic Pollutants in Water, Environ. Pollut., 2006, 142, 333-343. 11. Vrana B., Mills G. A., Kotterman M., Leonards P., Booij K., and Greenwood R., Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water, Environ. Pollut., 2007, 145, 895-904. 12. Schäfer R. B., Paschke A., Vrana B., Mueller R., and Liess M., Performance of the Chemcatcher passive sampler when used to monitor 10 polar and semi-polar pesticides in 16 Central European streams, and comparison with two other sampling methods., Water Res., 2008, 42, 2707-17. 13. Lobpreis T., Vrana B., Dominiak E., Dercová K., Mills G. A., and Greenwood R., Effect of housing geometry on the performance of ChemcatcherTM passive sampler for the monitoring of hydrophobic organic pollutants in water, Environ. Pollut., 2008, 153, 706-710. 14. Greenwood R., Mills G. A., and Vrana B., Potential applications of passive sampling for monitoring non-polar industrial pollutants in the aqueous environment in support of REACH, /. Chromatogr. A, 2009, 1216, 631-639. 15. Allan I. J., Booij K., Paschke A., Vrana B., Mills G. A., and Greenwood R., Field performance of seven passive sampling devices for monitoring of hydrophobic substances, Environ. Sci. Technol, 2009, 43, 5383-5390. 16. Lobpreis T., Vrana B., and Dercová K., Innovative approach to monitoring organic contaminants in aqueous environment using passive sampling devices , Inovativne prístupy k monitorovaniu organických kontaminantov vo vodnom prostredí použitím pasívneho vzorkovania, Chemické Listy 2009, 103, 548-558. 17. Allan I. J., Booij K., Paschke A., Vrana B., Mills G. a, and Greenwood R., Short-term exposure testing of six different passive samplers for the monitoring of hydrophobic contaminants in water., /. Environ. Monit., 2010, 12, 696-703. 18. Prokeš R., Vrana B., Klánová J., and Kupec J., Calibration of three passive samplers of hydrophobic organic compounds in water: Assessment of critical issues in experimental design data interpretation and field application, Fresenius Environ. Bull, 2010, 19, 2812-2822. 19. Tapie N., Devier M. H., Soulier C, Creusot N., Le Menách K., Ait-Aissa S., Vrana B., and Budzinski H., Passive samplers for chemical substance monitoring and associated toxicity assessment in water, Water Sci. Technol, 2011, 63, 2418. 20. Prokeš R., Vrana B., and Klánová J., Levels and distribution of dissolved hydrophobic organic contaminants in the Morava river in Zlín district, Czech Republic as derived from their accumulation in silicone rubber passive samplers., Environ. Pollut., 2012, 166, 157-66. 21. Jarošová B., Bláha L., Vrana B., Randák T., Grabic R., Giesy J. P., and Hilscherová K., Changes in concentrations of hydrophilic organic contaminants and of endocrine-disrupting potential downstream of small communities located adjacent to headwaters, Environ. Int., 2012, 45, 22-31. 22. Lohmann R., Booij K., Smedes F., and Vrana B., Use of passive sampling devices for monitoring and compliance checking of POP concentrations in water, Environ. Sci. Pollut. Res., 2012, 19, 1885-1895. 23. Jalová V., Jarošová B., Bláha L., Giesy J. P., Ocelka T., Grabic R., Jurčíková J., Vrana B., and Hilscherová K., Estrogen-, androgen- and aryl hydrocarbon receptor mediated activities in passive and composite samples from municipal waste and surface waters, Environ. Int., 2013, 59, 372-383. 24. Vrana B., Klučárová V., Benická E., Abou-Mrad N., Amdany R., Horáková S., Draxler A., Humer F., and Gans O., Passive sampling: An effective method for monitoring seasonal and spatial variability of dissolved hydrophobic organic contaminants and metals in the Danube river, Environ. Pollut., 2014, 184, 101-112. 25. Mills G. A., Gravell A., Vrana B., Harman C, Budzinski H., Mazzella N., and Ocelka T., Measurement of environmental pollutants using passive sampling devices - an updated commentary on the current state of the art., Environ. Sci. Process. Impacts, 2014, 369-373. 26. Amdany R., Chimuka L., Cukrowska E., Kukučka P., Kohoutek J., and Vrana B., Investigating the temporal trends in PAH, PCB and OCP concentrations in Hartbeespoort Dam, South Africa, using semipermeable membrane devices (SPMDs), Water SA, 2014, 40, 425-436. 76 27. Amdany R., Chimuka L., Cukrowska E., Kukučka P., Kohoutek J., Tôlgyessy P., and Vrana B., Assessment of bioavailable fraction of POPS in surface water bodies in Johannesburg City, South Africa, using passive samplers: An initial assessment, Environ. Monit. Assess., 2014, 186, 5639-5653. 28. Miěge, C, Mazzella, N., Allan, L, Dulio, V., Smedes, F., Tixier, C, Vermeirssen, E., Brant, J., O'Toole, S., Budzinski, H., Ghestem, J.-P., Staub, P.-F., Lardy-Fontan, S., Gonzalez, J.-L., Coquery, M., Vrana, B., 2015. Position paper on passive sampling techniques for the monitoring of contaminants in the aquatic environment -Achievements to date and perspectives. Trends Environ. Anal. Chem. 2015, in press. 12.2 Kapitoly v odbornej knihe 29. Booij K, Vrana B., Huckins J. N: Chapter 7 Theory, modelling and calibration of passive samplers used in water monitoring. In: Comprehensive Analytical Chemistry, R. Greenwood, G. Mills, B. Vrana (eds.). Elsevier, Amsterdam, Volume 48, 2007, Pages 141-169. 30. Greenwood R., Mills G.A., Vrana B., Allan I., Aguilar-Martinez R., Morrison G.: Chapter 9 Monitoring of priority pollutants in water using Chemcatcher passive sampling devices. In: Comprehensive Analytical Chemistry, R. Greenwood, G. Mills, B. Vrana (eds.). Elsevier, Amsterdam, Volume 48, 2007, Pages 199-229. 31. Paschke A., Vrana B., Popp P., Wennrich L. PaschkeH., Schuurmann G.: Chapter 10 Membrane-enclosed sorptive coating for the monitoring of organic compounds in water. In: Comprehensive Analytical Chemistry, R. Greenwood, G. Mills, B. Vrana (eds.). Elsevier, Amsterdam, Volume 48, 2007, Pages 231-249. 32. Vrana B., Smedes F., Rusina T., Okonski K., Allan I., Grung M., Hilscherova K., Novák J., Tarábek P., And Slobodník J., Passive sampling: chemical analysis and toxicological profiling, in Joint Danube Survey 3, I. Liška, F. Wagner, M. Sengl, K. Deutsch, and J. Slobodník, Eds. Vienna: ICPDR - International Commission for the Protection of the Danube River, 2015, 304-315. 12.3 Ďalšie práce 33. Vrana B., Vermeirssen E. L. M., Allan I., Kohoutek J., Kennedy K., Mills G., and Greenwood R., Passive sampling of emerging compounds in the environment: state of the art and perspectives, 2010. [Online]. Available: http://www.norman- 77 network.net/sites/default/files/files/Events/2009/2009May27-Prague-PassiveSampling/norman_position_paper_pas_sampling.pdf. 78 Príloha 1 Vrana B., Paschke A., Popp P., and Schuurmann G., Use of semipermeable membrane devices (SPMDs): Determination of bioavailable, organic, waterborne contaminants in the industrial region of Bitterfeld, Saxony-Anhalt, Germany, Environ. Sci. Pollut. Res., 2001, 8, 27-34. Research Articles Semipermeable Membrane Devices Research Articles Use of Semipermeable Membrane Devices (SPMDs) Determination of Bioavailable, Organic, Waterborne Contaminants in the Industrial Region of Bitterfeld, Saxony-Anhalt, Germany Branislav Vrana1, Albrecht Paschke1, Peter Popp2, and Gerrit Schüürmann1 1 Dept. of Chemical Ecotoxicology, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany 2 Dept. of Analytical Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany Corresponding address: Branislav Vrana; e-mail: bv@uoe.ufz.de DPI: http://dx.doi.orq/10.1065/espr2000.08.033 Abstract. Triolein-containing semipermeable membrane devices (SPMDs) were employed as passive samplers to provide data on the bioavailable fraction of organic, waterborne, organochlorine pesticides (OCPs), polychlorinated biphenyls (PCBs) and poly-nuclear aromatic hydrocarbons (PAHs) in streams flowing through a highly polluted industrial area of Bitterfeld in Saxony-Anhalt, Germany. The contamination of the region with organic pollutants originates in wastewater effluents from the chemical industry, from over one-hundred years of lignite exploitation, and from chemical waste dumps. The main objective was to characterise time-integrated levels of dissolved contaminants, to use them for identification of spatial trends of contamination, and their relationship to potential pollution sources. SPMDs were deployed for 43 days in the summer of 1998 at four sampling sites. The total concentration of pollutants at sampling sites was found to range from a low of 0.8 ug/SPMD to 25 ug/ SPMD for PAHs, and from 0.4 ug/SPMD to 22 ug/SPMD for OCPs, respectively. None of the selected PCB congeners was present at quantifiable levels at any sampling site. A point source of water pollution with OCPs and PAHs was identified in the river system considering the total contaminant concentrations and the distribution of individual compounds accumulated by SPMDs at different sampling sites. SPMD-data was also used to estimate average ambient water concentrations of the contaminants at each field site and compared with concentrations measured in bulk water extracts. The truly dissolved or bioavailable portion of contaminants at different sampling sites ranged from 4% to 86% for the PAHs, and from 8% to 18% for the OCPs included in the estimation. The fraction of individual compounds found in the freely dissolved form can be attributed to the range of their hydrophobicity. In comparison with liquid/liquid extraction of water samples, the SPMD method is more suitable for an assessment of the background concentrations of hydrophobic organic contaminants because of substantially lower method quantification limits. Moreover, contaminant residues sequestered by the SPMDs represent an estimation of the dissolved or readily bioavailable concentration of hydrophobic contaminants in water, which is not provided by most analytical approaches. Keywords: Bioavailability; monitoring; organochlorine pesticides; passive sampling; persistent organic pollutants; priority pollutants; polyaromatic hydrocarbons; polychlorinated biphenyls; semipermeable membrane devices (SPMDs); water contamination Introduction Qualitative and quantitative assessment of pollution of aquatic ecosystems by hydrophobic organic contaminants is a continuing challenge to environmental scientists. The fate and transport of these compounds depends on their physi-cochemical phase distribution. In aquatic systems, it is important to identify the freely dissolved concentration of a compound. The amount of substance freely dissolved in water also yields an approximate characterisation of the bioavailable fraction [1]. Concentrations of truly dissolved or bioavailable contaminants cannot be determined by most water sampling methods. Instead, total quantities of analytes are measured, including those molecules that are not readily bioavailable because they are bound to dissolved colloids present in water. Moreover, grab water samples provide information only about contaminant concentrations at the moment of sampling and may fail to account for episodic contamination events. Because of the low aqueous solubility of hydrophobic contaminants, it is often impossible to excise sufficiently large water samples to achieve instrumental detection limits. For these reasons, integrative sampling devices are needed which sequester truly dissolved contaminants over a longer time period and provide information about the time-averaged water concentration of contaminants. Huckins et al. [2,3] described the development of a semipermeable membrane device (SPMD) for passive and integrative in situ monitoring of waterborne contaminants. The SPMD sampler consists of layflat, polyethylene tubing containing a thin film of triolein, a high molecular-weight neutral lipid. The polyethylene used in SPMDs is commonly referred to as nonporous, even though transient cavities with diameters approaching about 1 nm are formed by random thermal motions of the polymer chains [3]. The thermally mediated transport corridors of the polyethylene exclude larger molecules, as well as those that are adsorbed on sediments or humic acids. Only truly dissolved (but generally nonionized) contaminants are sequestered. The process mimics the transfer of organic contaminants through biomembranes. The utility of the SPMD has been shown for monitoring aqueous residues of polychlorinated biphenyls, various organochlorine pesticides, polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans and polycyclic aromatic compounds. ESPR - Environ Sei & Pollut Res 8 (1) 27 - 34 (2001) © ecomed publishers, D-86899 Landsberg, Germany and Ft. Worth/TX, USA • Tokyo, Japan • Mumbai, India • Seoul, Korea 27 Semipermeable Membrane Devices Research Articles Results are reported here from a study where SPMDs were used to obtain information on spatial trends in bioavailable contaminants in a long-term polluted river system of the Mulde River, a major branch of the Elbe River flowing through a highly polluted industrial area of Bitterfeld in Saxony-Anhalt, Germany (Fig. 1). The contamination of the region with organic pollutants originates in wastewater effluents from the chemical industry, from over one-hundred years of lignite exploitation, and from chemical waste dumps. During the production of organochlorine insecticides like DDT (dichlorodiphenyl-trichloroethane) and lindane (y-hex-achlorocyclohexane, y-HCH) in the past decades, the toxic by-product HCH isomers (84% of the produced quantity of HCH isomers) was obtained and deposited on dumps. Most of these dumps located in former lignite pits are not sealed and waste HCH and DDT is washed out by the drainage water before entering the groundwater [4]. About 76 000 tons of HCH isomers, 3000 tons of DDT and its metabolites (DDX), and 13 000 tons of distillation residues containing chlorinated benzenes were deposited during the time period from 1960 to 1982 in the abandoned Antonie lignite pit, located in the centre of the industrial zone near Bitterfeld. The main stream in this area, called the Spittel-wasser, served as a wastewater channel for several decades. As a consequence of flood events, the soils of the wetland area along the Spittelwasser also became highly polluted with organic contaminants when the contaminated water covered the wetlands [5,6]. Fig. 1: Map of the streams in the area of Bitterfeld in Saxony-Anhalt, Germany. Hollow circles indicate SPMD deployment sites.The chemical plants in Bitterfeld, and the spring of the Schachtgraben Creek are located 6 km upstream from sampling site I. Arrows show the direction of flow in the streams The aim of the present study was to characterise spatial variations in the concentrations and relative proportions of the persistent organic pollutants at the various sites, and gradients in river water contamination were identified. In addition, average ambient water concentrations of the organic contaminants were estimated at each field site and compared with concentrations determined in bulk water samples. 1 Experimental Section 1.1 Materials and chemicals The solvents acetone, dichloromethane, hexane, isopropa-nol, and toluene in LiChrosolv quality were obtained from Merck. Acetonitrile (HPLC, Ultra Gradient Grade) and HPLC water were purchased from Baker. Hexachloroben-zene (HCB), HCH and DDX reference materials were obtained from Supelco, PAH standard materials from Supelco and PCB reference materials from Promochem. 1.2 Sampling devices SPMDs with standard configuration, designed by Huckins et al. [2] at the USGS in Columbia, MO, USA, consisting of a thin film of 1 mL of triolein (95% pure) sealed in a low-density, polyethylene, layflat tube (2.54x91.4 cm, 75-90 urn wall thickness), were purchased from Origo Hb, Tavelsjo, Sweden. They were stored in original, gas-tight, metal paint cans until just before field deployment. 1.3 SPMD deployment The SPMDs were placed at 4 sites in the streams of the Schachtgraben (Site I), the Spittelwasser (Sites II and III; above and below its confluence with the Schachtgraben, respectively), and the Mulde River (Site IV; below the tributary of the Spittelwasser), flowing through a highly polluted industrial area of Bitterfeld in Saxony-Anhalt, Germany (Fig. 1). The centre of the industrial area of Bitterfeld, including chemical plants and waste dumps, is located 6 km upstream from sampling Site I. The Schachtgraben Creek collects drainage water from this area. The SPMDs were deployed for 43 days during summer 1998 (29th July to 9th September). During the exposure, the water temperature varied from 17.5 to 21.5°C at all sampling sites. On the day of deployment and retrieval of the samples, water temperatures were equal (±1°C) at each sampling site. Two SPMDs were deployed at each of the four sampling sites. At each deployment site, SPMDs were removed from the metal can and placed into a stainless steel 1 cm mesh frame (20x100 cm) which protected the SPMDs from both sides. SPMDs were secured by fastening their ends to the frame using flat, stainless steel belts which were screwed to opposite ends of the frame. The frame was deployed in the horizontal orientation above the bottom of the stream and held by tent pegs. The depth below the water surface at which SPMDs were deployed at sampling sites ranged from 40 to 70 cm. On day 43, the SPMDs were removed from their frames and immediately sealed in individual amber glass jars. 28 ESPR - Environ Sei & Pollut Res 8 (1) 2001 Research Articles Semipermeable Membrane Devices The jars were transported to the laboratory in a cooler (on ice and in darkness) and were kept in a freezer at -20°C until processing. Exposed SPMDs were analysed for selected PAHs, PCBs and OCPs. 1.4 Sediment and soil samples A top soil sample from the site located in the floodplain of the Spittelwasser stream near sampling site III and a sediment sample from sampling site III were taken for PAH, PCB and OCP determination. Samples were extracted with toluene using accelerated solvent extraction technique and analysed for PAH, PCB and OCP content as described by Popp et al. (1997) [7]. 1.5 Water samples Water samples were taken from each sampling site at the end of the exposure period. Water samples were sealed in amber glass jars at 4°C until processing. One litre from each water sample was extracted three times with 30 mL toluene. Combined toluene extracts were filtered through anhydrous Na2S04, concentrated by rotary evaporation and by nitrogen blow-down, redissolved in 500 pL acetonitrile and analysed for PCBs, PAHs and OCPs. 1.6 Sample processing and residue enrichment Biofouling removal. The devices were subjected to an exterior cleanup for biofouling removal. SPMDs were shaken with 50 mL hexane for 20-30 s, rinsed with running deionized water, and scrubbed with soft toothbrushes. After that, they were submerged in 1 M HC1 for 30 s, then rinsed with running deionized water, rinsed with acetone and then with isopropanol. SPMDs were then allowed to dry in a glass column in a stream of high-purity nitrogen. Dialytic recovery of analytes. SPMDs were dialysed three times with 250 mL hexane per SPMD at 18°C for 24 hours. The dialysate volume was reduced to approximately 10 mL by rotary evaporation, and further reduced in volume with streams of high-purity nitrogen to dryness. The residue was redissolved in 5 mL dichloromethane and cleaned up by size exclusion chromatography (SEC). Size exclusion chromatography. The concentrated dialysate was cleaned up using a high performance SEC column (22.5 mm IDx250 mm, 10 pm particles) Lichrogel® PS 20 (Merck, Germany). The mobile phase for the SEC was dichloromethane (5 mL/min). The collected fraction containing the compounds of interest extended from 85-195 mL. The eluate from SEC was concentrated to approximately 10 mL by rotary evaporation. The volume of concentrated eluate was adjusted to 10 mL. Nonane (100 pL) was added as a keeper. From the eluate, 1 mL was taken and solvent exchanged to 1 mL acetonitrile for HPLC analysis of priority pollutant PAHs. In the remaining 9 mL of eluate, dichloromethane was evaporated using high-purity nitrogen. The residue was redissolved in hexane to 1 mL final volume, and was used for GC analysis of PCBs and OCPs. 1.7 Analysis The qualitative analysis of the PCBs and OCPs of interest was made by a mass spectrometric detector after separation of the contaminants by GC (HP 5890) using a capillary column (30 mx0.25 mm ID) with a non-polar stationary phase HP5MS (thickness 0.25 pm). Temperature conditions: Injector 250°C, column 80°C (6 min)- 6°C/min- 250°C (8.67 min). For quantitation of the PCBs and OCPs, sample extracts in hexane were analysed by GC (HP 5890) using an electron capture detector (300°C) and a capillary column (25 mx0.32 mm ID) with a non-polar stationary phase Ultra 2 (thickness 0.17 pm). Temperature conditions: Injector 300°C, column 80°C (6 min)- 6°C/min- 250°C (8.67 min). Quantitation of the residues was accomplished using a ten-point, external standard curve. The extract in acetonitrile was analysed for EPA priority pollutant PAHs by HPLC (HP 1050) using a programmable fluorescence detector and C18 Vydac201TP54 (250x4.6 mm ID) column. The mobile phase was acetonitrile/water pumped at 1 mL/min at 23°C with gradient elution. The composition gradient started with 60% water and 40% acetonitrile (3 min), then the acetonitrile content was increased to 100% in 24 min with a linear gradient. The contents were held constant for 13 min until the end of the analysis. Quantification of the PAH residues was accomplished using a 7-point, external standard curve. Acenaphthylene was not included in the analytical procedure because the substance shows no fluorescence. 1.8 Quality control Because SPMDs have a propensity to sequester vapour-phase contaminants [8], additional two trip blank SPMDs were exposed to air at each site while the water sampling SPMDs were being deployed and collected. Trip blanks were processed exactly as deployed samples and were used to define contamination of the SPMDs during transportation and handling. In addition, fresh SPMDs were taken through the entire dialytic and cleanup procedure (procedural blanks). Samples containing contaminant residues exceeding the procedural blank values were considered positive for contaminants. Spiked SPMDs were also analysed by fortifying fresh membranes and then processing them as a sample. The PCBs and OCPs were spiked at 500 ng per SPMD and PAHs at 80 ng per SPMD for each single component. Recovery rate values of the fortified PAHs from SPMDs were good and reproducible, with the exception of naphthalene. Average percent recoveries of the remaining PAHs varied from 56% for indeno[l,2,3] pyrene to 137% for phenanthrene, and the relative standard deviation of three spiked samples did not exceed 11 % for any compound. Average percent recoveries for the organochlorine compounds varied from 68% for (3-HCH to 125% for PCB 153, and the relative standard deviation of three spiked samples did not exceed 11 % for any compound. Method quantification limits (MQL) for PAHs in SPMDs ranged from 4 ng/SPMD for anthracene to 50 ng/SPMD for chrysene. In bulk water samples, the MQL ranged from 0.2 ng/L for anthracene to 2.5 ng/L for chrysene. MQL for PCBs and OCPs in SPMDs ranged from 0.5 to 2 ng/SPMD. ESPR - Environ Sei & Pollut Res 8 (1) 2001 29 Semipermeable Membrane Devices Research Articles In water samples, MQL for PCBs and OCPs ranged from 0.25 to 1 ng/L. 2 Results and Discussion 2.1 Occurrence of contaminants Concentrations of compounds of interest found in the SPMDs exposed for 43 days at different sampling sites are presented in Table 1. The concentrations of compounds are adjusted according to their recoveries from fortified SPMDs. The trip-blank SPMDs were devoid of quantifiable residues of OCPs, PCBs and PAHs, except for naphthalene and phen-anthrene; the phenanthrene concentration in the trip blank was much lower (more than 3 times) than its concentration in the SPMDs at sampling sites. Because of high trip-blank values, high differences in results from duplicate samples, and bad recovery from SPMDs, naphthalene was ignored in the further discussion. Duplicate SPMDs sequestered similar amounts of PAHs and OCPs of interest. In general, the relative percent differences between two SPMDs deployed at the same sampling site did not exceed 29% for PAHs, except for acenaphthene at sampling site III (44%). Relative percent differences for OCPs were not greater than 24%, except for p,p'-DDD at sampling site III (40%). Prest and Jacobson [9] have shown that the ratio of contaminant concentrations sequestered by two SPMDs at two sampling sites under similar conditions is equal to the ratio of time-averaged aqueous concentrations at the two sampling sites. When using SPMDs with a standard configuration, designed by Huckins et al. [2], mainly temperature and biofouling can effect the sampling rate. We assumed that the hydrodynamic conditions did not effect the uptake kinetics dramatically, although it has been reported that aqueous diffusion boundary layer at the membrane surface controls contaminant uptake for compounds with log Table 1: Mean concentrations of PAHs and OCPs found in SPMDs (ngSPMD, n=2) (a) and in sediment and top soil samples (ng/g: dry weight based), at sampling sites in the Bitterfeld region impling site Trip blank Site I Site II Site III Site IV Sediment at site III Top soil at site III \Hs Naphthalene (b) 86 71 1 117 378 205 729 1795 Acenaphthene <10 514 50 168 31 38 25 Fluorene <50 603 110 246 90 218 132 Phenanthrene 59 1 163 290 387 227 2461 1863 Anthracene <4 775 26 224 16 377 259 Fluoranthene <30 7420 365 2293 155 1002 1 168 Pyrene 7 8420 304 2214 92 2106 1768 Benzo[a]anthracene <20 1 198 41 410 <20 231 231 Chrysene <50 2918 124 1041 <50 714 1035 Benzo[b]flouranthene <20 747 36 335 <20 162 NQ (c) Benzo[k]fluoranthene <5 343 16 155 <5 114 115 Benzo[a]pyrene <5 470 14 204 <5 207 122 Dibenz[a,h]anthracene <15 21 <15 <15 <15 39 40 Benzo[g,h,i]perylene <15 165 <15 83 <15 212 151 ndeno[1,2,3]pyrene <25 153 <25 106 <25 210 188 E PAHs 152 25621 1493 8243 815 8820 8892 OCPs a-HCH 2 4597 21 3283 90 266 5330 P-HCH 1 944 34 1073 42 NQ 5329 y-HCH 1 1835 1 1 1551 43 NQ 274 S-HCH 2 2935 8 2330 32 NQ 660 HCB 3 3955 42 1482 66 402 3276 4,4'-DDD 1 3877 23 1397 106 NQ 842 4,4'-DDE 1 1286 22 318 29 NQ 14 4,4'-DDT 1 2934 15 545 39 NQ NQ EOCPs 12 22362 176 1 1979 446 668 15725 E PCBs (d) <6 <6 <6 <6 <6 <6 <6 Concentrations are recovery-rate-corrected I Concentrations of naphtalene are not recovery-rate-corrected NQ - not quantifiable (presence of interfering peaks) I PCBs quantified include congeners IUPAC No. 28, 52, 101, 138, 153 and 180 30 ESPR - Environ Sei & Pollut Res 8 (1) 2001 Research Articles Semipermeable Membrane Devices Kow>4.4 at water flow velocities of 30 cm/s and lower [10]. To correct the sampling rate values for the effects of biofouling and the flow velocity, Huckins et al. [3] suggested the use of a permeability reference compound (PRC). PRC is a non-interfering compound with moderate SPMD fugac-ity added to SPMD lipid prior to exposure. However, the data on this application in practice is still very limited. In this study, water temperatures were comparable (±1°C) at each sampling site. When assuming a similar biofouling at each sampling site, uptake rates were expected to be the same, and the ratio of contaminant concentrations found in SPMD samplers at two sampling sites was considered to be equal to the ratio of average water concentrations at these sites, a finding which must be taken into consideration for further discussion. 2.1.1 PAHs: Absolute concentrations in SPMDs The sites can be ranked from lowest to highest concentrations of total PAHs as follows: Mulde (site IV), Spittelwasser above its confluence with the Schachtgraben (site II), Spittelwasser below its confluence with the Schachtgraben (site III), and the Schachtgraben (site I). Concentrations of individual PAHs at the sampling sites generally followed the same pattern as the totals (Table 1). The PAHs found at the highest level at all four sampling sites (fluoranthene, pyrene, and phenanthrene) are ubiquitous contaminants found in the runoff from urban and industrialised areas, and their origin can be ascribed to combustion and industrial activity [11]. The highest total levels of PAHs sequestered by SPMDs in this study were comparable on the order of magnitude of the amounts found in SPMDs in other studies conducted under similar conditions [10,12,13]. Fluoranthene (the PAH found at the highest level in most samples) concentrations ranged from a low of 155 ng/sample at site IV to a high of 7.4 pg/sample at site I. The highest total concentration of PAHs, found in SPMDs from the Schachtgraben Creek (site I), is not surprising considering that the Schachtgraben drains away the water from the centre of the industrial area in Bitterfeld. Elevated concentrations of PAHs could also be found in the Spittelwasser River above its confluence with the Schachtgraben (site II). However, the total PAH concentration at the sampling site II was more than one order of magnitude lower than the concentration at the sampling site I. The PAHs found at sampling site II are likely to originate from air deposition and the passage of contaminated groundwater. Below the confluence of the Schachtgraben, the PAH concentration in the Spittelwasser River (site III) rises more than 10 times in comparison with that at site II. It is likely that this extreme concentration rise originates from the contribution of the Schachtgraben water. At the sampling site in the Mulde River (site IV) about 4 km downstream from sampling site III, near the confluence of the Mulde and the Spittelwasser, the concentration of several PAHs (acenaph-thene, fluorene, phenanthrene, anthracene, fluoranthene, and pyrene) was still measurable. It is likely that the water quality in the Mulde River at site IV is negatively influenced by the nearby tributary of the Spittelwasser. 2.1.2 PAH patterns in SPMDs To examine the relative concentrations of PAHs detected in SPMD samples among sites, the concentrations of individual PAHs were normalised by proportioning to fluoranthene's concentration for each site (Fig. 2). No major differences in relative concentrations were determined between site I and site III. In comparison with these sites, samples from sites II and site IV contain relatively higher normalised concentrations for compounds such as acenaphthene, fluorene, and phenanthrene. Among other PAHs, these compounds demonstrate a relatively good aqueous solubility and a low hydro-phobicity. These properties allow for transport to longer distances from the pollution source in the dissolved phase. Therefore, we conclude that sampling sites II and IV are more distant from a point source of pollution than sites I and III. This hypothesis is supported by a second observation that sites II and IV contain lower or negligible normalised concentrations for more hydrophobic compounds (log Kow>5.5) in comparison with sites I and III. Benzo(g, h, i)perylene Dibenz(a,h)anthracene Benzo(a)pyrene Benzo(k)fluoranthene Benzo(b)flouranthene Chrysene Benzo(a)anthracene Pyrene Fluoranthene Anthracene Phenanthrene Fluorene Acenaphthene EZ3 Site IV I I Site I I Site II H Site I relative concentration (%) Fig. 2: Patterns of polyaromatic hydrocarbons found in SPMDs exposed for 43 days at four deployment sites presented as fluoranthene normalized concentrations On the basis of total OCP residues, the sites can be ranked from lowest to highest as follows: Site II, site IV, site III, and site I. The sequestered amounts of OCPs (HCHs, HCB and DDX) were up to several orders of magnitude higher than observed in other studies [14-16]. None of the selected PCBs was present at quantifiable levels at any sampling site. Below the confluence of the Schachtgraben, the concentration of total OCPs of interest in the Spittelwasser stream (site III) rises almost two orders of magnitude. The OCP concentration in the Mulde River (site IV) was low; nevertheless, it was almost three times higher than that found at site II. HCB (the organochlorine compound found at the highest level in all samples) concentrations ranged from 42 ng/sample at site II to 4.0 pg/sample at site I. As concluded for PAHs, the most probably contamination source of the water in the Spittelwasser is the Schachtgraben tributary. The spatial trends in the HCH (summed-up) concentrations were similar as determined for total OCPs, ranging from 74 ng/sample at site II to 10.3 pg/sample at site I. ESPR - Environ Sei & Pollut Res 8 (1) 2001 31 Semipermeable Membrane Devices Research Articles The DDX components found in the samples were 4,4'-DDD (from 23 ng/sample to 3.9 pg/sample), 4,4'-DDE (from 22 ng/ sample to 1.3 pg/sample), and 4,4'-DDT (from 15 ng/sample to 2.9 pg/sample), with the spatial trends similar to total OCPs. 2.1.3 OCPs in SPMDs On the basis of the comparison of total contaminant concentrations sequestered by SPMDs at different sampling sites, but also after considering the distribution of individual compounds sampled by SPMDs at the sampling sites, we conclude that there is a point source of water pollution upstream to sampling site I in the Schachtgraben Creek. The concentration of PAHs and OCPs at sampling site III might be additionally elevated due to contaminant mobilization from historically contaminated sediments and soils of the wetland area along the Spittelwasser, which are likely to cause additional diffuse entries (Table 1). A simple, quantitative, inter-site comparison between water and sediment or soil from a specific site cannot be made since the sampled contaminants in the water column were sampled using SPMDs during a relatively short exposure period, whereas the contaminants in sediment were deposited during long time periods (months to years). Moreover, processes having control over the accumulation of contaminants from water to sediment, soil and SPMDs are of a different character. However, the PAH patterns of the three matrices sampled at this site (i.e. SPMD, sediment and top soil) have similar profiles, which suggests that the sources of the contamination might be the same between the past and present. 2.2 Water concentration estimation Using models previously developed [3] and applied [15-18], the time-averaged, bioavailable, waterborne concentrations of OCPs and PAHs at the sampling sites were estimated from concentrations in the SPMDs exposed in this study. The details of the model development are available [3,19] and will not be presented here. Equation 1 was used to calculate the dissolved (i.e. readily available), waterborne concentrations of compounds. c. clyl RJ (3) C V q _ y SPMD y SPMD (1) RJ As applied here, Cw is the concentration of the analyte in water, CSPMD is the concentration of the analyte in the SPMD (lipid+membrane), t is the exposure time in days, VSPMD is the volume of the SPMD (lipid+membrane), and Rsc is the SPMD sampling rate which is given by Rrr. — R„F, (2) where Ft is 1 - the fractional reduction in uptake flux or sampling rate due to fouling impedance. Sampling rates for the PAHs at 18°C have been reported [19], and were utilised for water concentration estimation in this study. Sampling rate (Rs) data for the OCPs at 26°C reported by Huckins et al. [20] is based on the analyte concentrations determined only in lipid. To utilise this data for an estimation of the water concentration of OCPs, the concentration of analytes in the lipid phase C, was calculated from the total analyte mass in a SPMD dialysate (Md: averaged value from two parallel SPMDs) as described by the relationship [3]. C^Md/{Ml+KmlMm) (4) where M, and Mm are the weights of the lipid and the membrane. For a first approximation, a substance-unspecific, membrane/lipid, partition coefficient (KmI) of 0.1 was used as shown by Huckins et al. [3]. This results in an analyte distribution of 73% in the lipid and 27% in the membrane [18] for the linear uptake phase. An average fouling resistance of 20% (Fi = 0.8) was employed for biofouled SPMDs to correct for reduction in SPMD uptake [16,18] SPMD sampling rates for PAHs and OCPs are given in Table 2. The estimation of time-averaged water concentration using equations (1) and (3) has one limitation, it is applicable only for highly hydrophobic substances with log Kow values higher than 4. Gale [21] showed that the linear uptake model cannot be applied for compounds with lower hydrophobicity, because SPMDs do not sample these chemicals integratively during exposure periods longer than several weeks. Therefore, the estimation was not applied for acenaphthene, flu-orene, and HCH isomers. The estimated ambient concentrations of selected contaminants are presented in Table 2. Note that the water concentration estimated with the SPMDs is an average concentration over a 43 d interval, not the maximum concentration during that interval. When method quantification limits in SPMDs for contaminants selected for an estimation of aqueous concentrations were substituted into the linear model equations (1) and (3), the resulting MQL values for PAHs in water were lowered to between 0.04 ng/L for anthracene and 0.39 ng/L for chrysene. For OCPs, the MQL in water were even lower, ranging between 1 pg/L for HCB and 10 pg/L for 4,4'-DDT The SPMD MQL values are substantially lower than the MQL determined with the water extraction method, on average by a factor of 5 for PAHs and by a factor of 150 for OCPs, respectively. The MQL of the direct water analysis could be lowered by extracting larger volumes of water, although quality control and physical difficulties are often encountered when collecting and extracting large water volumes needed for the quantitation of trace organic contaminants. Therefore, in comparison with direct water analysis, the SPMD method is more suitable for the assessment of background concentrations of hydrophobic organic contaminants. The estimated water concentrations (Cw) are comparable to or lower than concentrations measured in bulk water extracts (Ch). The discrepancy between C^and Cb can be explained by the fact that Cw represents only the truly dissolved (readily bioavailable) fraction of contaminants in 32 ESPR - Environ Sei & Pollut Res 8 (1) 2001 Research Articles Semipermeable Membrane Devices Table 2: Estimates of truly dissolved aqueous concentrations from SPMDs Cwat sampling sites, and concentrations measured in bulk water samples Cb. Concentrations reported are in ng/L. Published sampling rates (Rs) for individual PAHs [19] and forOCPs [20] were used to estimate the Cw values Rs Site I Site II Site III Site IV Compound (L/d) Cw cb Cw cb Cw cb Cw cb PAHs Phenanthrene 3.8 8.9 85.8 2.2 29.3 3.0 52.6 1.7 57.1 Anthracene 2.9 7.8 8.2 0.3 3.7 2.2 7.1 0.2 5.0 Fluoranthene 3.6 59.9 38.8 2.9 6.0 18.5 31.1 1.3 12.7 Pyrene 4.5 54.4 50.8 2.0 9.1 14.3 30.8 0.6 12.7 Benzo[a]anthracene 3.2 10.9 3.9 0.4 <1.0 3.7 2.6 NR 2.1 Chrysene 3.7 22.9 NQ 1.0 NQ 8.2 NQ NR <2.5 Benzo[b]flouranthene 2.8 7.8 3.4 0.4 <1.0 3.5 3.3 NR <1.0 Benzo[k]fluoranthene 2.9 3.4 3.5 0.2 0.4 1.6 1.4 NR 0.9 Benzo[a]pyrene 3.2 4.3 8.4 0.1 1.8 1.9 4.3 NR 3.6 Dibenz[a,h]anthracene 2.0 0.3 1.2 NR <0.8 NR <0.8 NR <0.8 Benzo[g,h,i]perylene 1.8 2.7 7.9 NR 2.2 1.3 4.1 NR 4.1 ndeno[1,2,3]pyrene 3.0 1.5 4.3 NR <1.3 1.0 1.8 NR <1.3 Ľ PAHs 184.7 216.1 9.4 53.4 59.2 139.8 3.7 98.2 OCPs HCB 8.2 10.2 NQ 0.1 NQ 3.8 NQ 0.2 NQ 4,4'-DDD 5.0 16.5 86.0 0.1 <0.5 5.9 75.0 0.5 5.0 4,4'-DDE 7.6 3.7 7.0 0.1 <0.4 0.9 5.0 0.1 <0.4 4,4'-DDT 4.4 14.1 221.0 0.1 <1 2.6 80.0 0.2 <1 Ľ OCPs 44.6 314.0 0.3 NQ 13.3 160.0 0.9 5.0 NQ - not quantifiable (presence of interfering peaks) NR - no residue was found in SPMD sampler water, whereas Cb includes both contaminants dissolved and bound to the dissolved organic matter. The ratio CJCb shows that the bioavailability of hydrophobic compounds is strongly affected by the dissolved organic carbon (DOC) content of the water, which ranged from 4.5 to 11.2 mg/L in water samples taken at the four sampling sites. The truly dissolved or bioavailable portion of contaminants at different sampling sites ranges from 4% to 86% for the PAHs, and from 8% to 18% for the OCPs included in the estimation. For individual compounds, differences in the percentages found in the freely dissolved form can be attributed to the range of their hydrophobicity. For PAHs with moderate hydrophobici-ty, the portion bound to DOC seems to be low or even negligible (with the exception of phenanthrene), whereas the PAHs with high log Kow values, and the DDX will be partitioned to DOC to a greater extent. 3 Conclusions This study is the first field application of SPMDs reported in Germany and it confirms that this sampling method is convenient for the continuous monitoring of hydrophobic organic contaminants in a watercourse. The SPMD technique has several important advantages over conventional episodic analytic measurements. The devices can be used in the field with only little technical support, which is particularly useful for monitoring programmes in remote regions. Due to the principally great flexibility in selecting sampling sites, this technique ena- bles a pseudo-continuous monitoring of hydrophobic pollutants along potentially relevant emission and transfer pathways, thus allowing one to unravel pollution sources in cases of both episodic contamination events as well as long-term, low-dose contamination. The integrative approach allows one to reduce the costs of monitoring campaigns by a substantial reduction of the required sampling frequency. An estimate of average water concentrations over an extended period of time can be obtained in contrast to conventional direct water sampling and liquid/liquid extraction unless a water sample is continually collected, composited and analysed over the time period. Moreover, the SPMD MQL values for estimation of the water concentration have been shown to be up to two orders of magnitude lower than MQL determined with the conventional water extraction method. However, more substance-specific, sampling rate data is needed to allow for the estimation of average ambient water concentrations of a broader range of contaminants. Also, more research is necessary to examine the effect of different milieu conditions on the sampling function (pH, salinity, hydrodynamics, temperature, dissolved organic matter), which is important for the purpose of deriving reliable quantitative results under in situ conditions. To make the technique more suitable for routine monitoring, a less expensive and less time consuming sample processing would be required. A simplified sample extraction and cleanup with reduced solvent consumption would also reduce the risk of sample contamination during handling in the laboratory. ESPR - Environ Sei & Pollut Res 8 (1) 2001 33 Semipermeable Membrane Devices Research Articles Acknowledgments. The authors would like to thank Elke Büttner, Petra Keil, Coretta Bauer, and Petra Fiedler for sample preparation and instrumental measurements. [12] Bennett, E.R.; Metcalfe, T.L.; Metcalfe CD. (1996): Semipermeable membrane devices (SPMDs) for monitoring organic contaminants in the Otonabee River, Ontario. Chemosphere 33, 363-375 [13] Moring, J.B.; Rose, D.R. (1997): Occurrence and concentrations of polycyclic aromatic hydrocarbons in semipermeable membrane devices and clams in three urban streams of the Dallas-Fort Worth Metropolitan Area, Texas. Chemosphere 34, 551-566 [14] Herve, S.; Prest, H. E; Heinonen, P.; HyotylAinen, T.; Koistinen, J.; Paasivirta, J. (1995): Lipid-Filled Semipermeable Membrane Devices and Mussels as Samplers of Organo-chlorine Compounds in Lake Water. ESPR - Environ. Sci. & Pollut. Res. 2, 24-30 [15] Ellis, G. S.; Huckins, J. N.; Rostad, C. E.; Schmitt, C. J.; Petty, J. D.; MacCarthy, P. (1995): Evaluation of Lipid-Con-taining Semipermeable Membrane Devices for Monitoring Organochlorine Contaminants in the Upper Mississippi River. Environ. Toxicol. Chem. 14, 1875-1884 [16] Petty, J.D.; Poulton, B.C.; Charbonneau, C.S.; Huckins, J.N.; Jones, S.B.; Cameron, J.T.; Prest, H.E (1998): Determination of bioavailable contaminats in the lower Missouri River following the flood of 1993. Environ. Sci. Technol. 32, 837-842 [17] Lebo, J.A.; Zajicek, J.L.; Huckins, J.N.; Peterman, P.H. (1992): Use of semipermeable membrane devices for in situ monitoring of polycyclic aromatic hydrocarbons in aquatic environments. Chemosphere 25, 697-718 [18] Petty, J.D.; Huckins, J.N.; Orazio, C.E.; Lebo, J.A.; Poulton, B.C.; Gale, R.W; Charbonneau, C.S.; Kaiser, E.M. (1995): Determination of waterborne bioavailable pesticide residues in the lower Missouri river. Environ. Sci. Technol. 29, 2561-2566 [19] Huckins, J.N.; Petty, J.D.; Orazio, C.E.; Lebo, J.A.; Clark, R.C.; Gibson, V.L.; Gala, W.R.; Echols, K.R. (1999): Determination of uptake kinetics (sampling rates) by lipid-contain-ing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water. Environ. Sci. Technol. 33, 3918-3923 [20] Huckins, J.N.; Petty, J.D.; Lebo, J.A.; Orazio, C.E.; Prest,.H.E; Tillitt, D.E.; Ellis, G.S.; Johnson, B.T.; Manuweera, G.K. (1996): Semipermeable membrane devices (SPMDs) for the concentration and assessment of bioavailable organic contaminants in aquatic environments. In: Techniques in Aquatic Toxicology. (Ed: Ostrander,GK) CRC Press (Lewis Publishers), Boca Raton. Florida, USA, 625-655 [21] Gale, R.W. (1998): Three-compartment model for contaminant accumulation by semipermeable membrane devices. Environ. Sci. Technol. 32, 2292-2300 Received: March 20th, 2000 Accepted: April 17th, 2000 Online-First: August 21 st, 2000 References [I] Schüürmann, G. (1997): Thermodynamische Modelle für die Bioverfügbarkeit und Bioakkumulation organischer Chemikalien. Z. Umweltchem. Ökotox. 9, 345-352 [2] Huckins, J.N.; Tubergen, M.W; Manuweera, G.K. (1990): Semipermeable membrane devices containing model lipid: a new approach to monitoring the bioavailability of lipophilic contaminants and estimating their bioconcentration potential. Chemosphere 20, 533-552 [3] Huckins, J.N.; Manuweera, G.K.; Petty, J.D.; Mackay, D.; Lebo, J.A. (1993): Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environ. Sci. Technol. 27, 2489-2496 [4] Krapp, L.; Körner, H.; Köhler, H.J. (1993): Umweltgeologie beim Pilotprojekt Bitterfeld-Wolfen. Geowissenschaften 1, 1-9 [5] Lauer, M.; Heymann, T.; Schneider, C. (1992): Grundlagen und erste Ergebnisse zur modellhaften Untersuchung einer durch industrielle Abwässer kontaminierten Flussaue mit dem Ziel einer ökologisch verträglichen Sanierung - Muldeaue zwischen Bitterfeld und Dessau. Schadstoffe Umwelt 10, 163-170 [6] Neumeister, H.; Ruske, R. (1995): Immissionsgeprägte Böden in der Industrieregion Bitterfeld. Mittlgn. Dtsch. Bodenk. Ges. 77, 339-372 [7] Popp, P.; Keil, P.; Moder, M.; Paschke, A.; Thuss, U. (1997): Application of accelerated solvent extraction followed by gas chromatography, high-performance liquid chromatography and gas chromatography-mass spectrometry for the determination of polycyclic aromatic hydrocarbons, chlorinated pesticides and polychlorinated dibenzo-p-dioxins and dibenzofurans in solid wastes. J. Chromatog. A 774, 203-211 [8] Petty, J.D.; Huckins, J.N.; Zajicek, J.L. (1993): Application of semipermeable membrane devices (SPMDs) as passive air samplers. Chemosphere 27, 1609-1624 [9] Prest, H.E; Jacobson, L.A. (1997): Passive water sampling for polynuclear aromatic hydrocarbons using lipid-containing semipermeable membrane devices (SPMDs): application to contaminant residence times. Chemosphere 35, 3047-3063 [10] Boon, K; Sleiderink, HM; Smedes, F (1998): Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ. Toxicol. Chem. 17, 1236-1245. [II] Lebo, J.A.; Zajicek, J.L.; Orazio, C.E.; Petty, J.D.; Huckins, J.N.; Douglas, E.H. (1996): Use of the semipermeable membrane device (SPMD) to sample polycyclic aromatic hydrocarbon pollution in a lotic system. Polycyclic Aromatic Com- 34 ESPR - Environ Sci & Pollut Res 8 (1) 2001 Príloha 2 Vrana B., Paschke A., and Popp P., Polyaromatic hydrocarbon concentrations and patterns in sediments and surface water of the Mansfeld region, Saxony-Anhalt, Germany, J. Environ. Monit., 2001, 3, 602-609. Polyaromatic hydrocarbon concentrations and patterns in sediments and surface water of the Mansfeld region, Saxony-Anhalt, Germanyf Branislav Vrana,*a Albrecht Paschke" and Peter Popp& "Department of Chemical Ecotoxicology, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany. Tel: + 49 341 235 26 18; Fax: +49 341 235 2401; E-mail: bv@uoe.ufz.de bDepartment of Analytical Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany Received 30th May 2001, Accepted 12th September 2001 First published as an Advance Article on the web 29th October 2001 The composition and spatial distribution of polyaromatic hydrocarbons (PAHs) and their relation to potential pollution sources were investigated in the Bose Sieben Creek, Saxony-Anhalt, Germany, using two techniques: semipermeable membrane devices (SPMDs) and sediment analysis. SPMD is an integrative device that passively samples hydrophobic chemicals of low to moderate molecular weight (< 600 Da) in water. SPMDs were placed in water for 34 days at three sites where sediments were also sampled. Fifteen PAHs were determined in SPMDs and in sediment samples to evaluate the concentration levels and specific PAH patterns. Time-weighted average aqueous PAH concentrations were estimated from the PAH amount accumulated in SPMDs during the deployment period using previously reported sampling rates. Sediment-water partition coefficients were used to estimate PAH concentrations in pore water from sediments. Calculated pore water concentrations were, on average, almost three orders of magnitude higher than those calculated from SPMDs. Thus, in addition to contamination from other sources, the water concentration at the sampling sites might be elevated due to contaminant mobilization from historically contaminated sediments. Relative PAH patterns from SPMDs and sediment were compared using principal component analysis, and were correlated with the PAH patterns from different potential contamination sources, including Theisen sludge, one of the by-products of the smelting process for copper production in the region in the past, which is likely to be the main contamination source of PAHs. Moreover, three origin indices (concentration ratios of PAH isomer pairs) were used to evaluate the suitability of these compounds as tracers to distinguish between the contamination arising from different sources. The evaluation of contaminant patterns permits the conclusion that the PAHs are of pyrolytic, industrial origin, possibly including contamination by Theisen sludge, and rules out a petrogenic source for the hydrocarbons. Introduction Polycyclic aromatic hydrocarbons (PAHs) form a widespread class of environmental chemical pollutants. They arise from the incomplete combustion of recent and fossil organic matter in flames, engines and industrial processes (pyrolytic origin), from emissions of non-combustion-derived matter (petrogenic origin) and from the post-depositional transformation of biogenic precursors (diagenetic origin).1'2 PAHs enter surface waters mainly via atmospheric fallout, urban run-off, municipal effluents and oil spillage or leakage. After entering the aquatic environment, their behaviour and fate depend on their physicochemical properties.3'4 Volatilization, dissolution, adsorption onto suspended solids and subsequent sedimentation, biotic and abiotic degradation, uptake by aquatic organisms and accumulation are all major processes to which PAHs in water are subjected. Due to their low aqueous solubilities and hydrophobic nature (log A"ow = 3-8), the concentrations of dissolved PAHs in water are very low. Otherwise, PAHs associate easily with particulate matter and are finally deposited in the sediment.3'5 In the Mansfeld region of Saxony-Anhalt, Germany, one of the potential emission sources of PAHs originates from the traditional mining and processing of Kupferschiefer, a ■(■Electronic Supplementary Information available. See http://www.rsc. org/suppdata/em/b 1/b 104707h/ 602 /. Environ. Monk., 2001, 3, 602-609 metalliferous Permian black marine shale. For more than 800 years, this bituminous shale was mined for copper and smelted in a coke-fired blast furnace to produce copper. One of the by-products of the smelting process, originating from the washing procedure of the flue dust from the furnace, is the 'Theisen sludge'. By 1990, when mining and copper production were stopped, a total of approximately 220000 tons of this slurry had been deposited at several sites. These deposits were neither sealed from the ground nor covered, and were therefore a major source of contamination to surface water, groundwater and soil. In addition to the large quantity of heavy metals and metalloids, the Theisen sludge material contains substantial amounts of PAHs, dioxins and furans.6 Although the majority of the hydrocarbons involve a mixture of polyaromatic hydrocarbons with a high boiling point, additional alkylated homo-logues and partially hydrogenated aromatics have also been detected. The extremely fine-grained nature of the material facilitates emissions into the environment, in particular into surface water and groundwater adjacent to the present deposit. Therefore, Theisen sludge can cause major environmental problems. No reliable information is available on the aqueous concentrations of PAHs in the streams of the region of Mansfeld. However, a few measurements have detected elevated PAH levels in the region. In addition, studies have been conducted to assess the leaching behaviour of Theisen sludge, which indicated the elevated remobilization potential of the PAHs from this material.7 DOI: 10.1039/bl04707h This journal is © The Royal Society of Chemistry 2001 The main objective of this study was to assess the level of contamination by PAHs in the stream of the Böse Sieben due to industrial historical activities in the study area and, in particular, to identify the PAH composition, spatial distribution and potential pollution sources. The small river collects surface water from the Mansfeld region. Surficial sediment samples and passive samplers (semipermeable membrane devices, SPMDs) from three sampling locations were analysed for PAHs to obtain information on the concentrations of PAHs in the sediment and aqueous phase. SPMDs are innovative devices suitable for the passive and integrative in situ monitoring of dissolved water-borne contaminants.8'9 The SPMD sampler consists of lay-flat polyethylene tubing containing a thin film of triolein, a high molecular weight neutral lipid. The utility of SPMDs has been shown in monitoring aqueous residues of polychlorinated biphenyls (PCBs),10 various organochlorine pesticides,11 polychlorinated dibenzo-furans and dibenzo-/?-dioxins12 and polycyclic aromatic compounds.13 Experimental section Materials and chemicals The 15 PAHs analysed in the samples are listed in Table 1. The solvents acetone, dichloromethane, hexane, isopropanol and toluene (LiChrosolv quality) were obtained from Merck (Darmstadt, Germany). Acetonitrile (HPLC, Ultra Gradient Grade) and HPLC water were purchased from Baker (Deventer, The Netherlands). A standard solution of a PAH mixture in methanol was purchased from Supelco (Deisenhofen, Germany). Sampling devices SPMDs with a standard configuration (2.54 cm x 91.4 cm; 75-90 |im membrane thickness; total mass, 4.3 g each), designed by Huckins et al.8 at the USGS in Columbia, MO, USA, were assembled from low density polyethylene lay-flat tubing containing a thin film of 95% pure triolein (1 mL). They were purchased from Exposmeter, Tavelsjö, Sweden, and were stored in original, gas-tight, metal paint cans until just before field deployment. SPMD deployment The SPMDs were placed at three sites in the stream of the Böse Sieben [near Wimmelburg (Site I), the Böse Sieben near Unterrißdorf (Site II) and the Süßer See lake (Site III), near the mouth of the Bose Sieben], in the Mansfeld region in Saxony-Anhalt, Germany (Fig. 1). The river basin of the Bose Sieben comprises almost the whole area affected by copper mining and processing. The SPMDs were deployed for 34 days during the autumn of 1998 (18th September to 22nd October). During exposure, the average water temperature varied from 12 to 14 °C at all sampling sites. Two SPMDs were deployed at each of the three sampling sites. At each deployment site, SPMDs were removed from the metal can and placed into a stainless steel 1 cm mesh frame (20 cm x 100 cm) which protected the SPMDs from both sides. SPMDs were secured by fastening their ends to the frame using flat, stainless steel belts screwed to opposite ends of the frame. The frame was deployed in a horizontal orientation above the bottom of the stream and held by tent pegs. The depth below the water surface at which SPMDs were deployed at the sampling sites ranged from 15 to 90 cm. On day 34, the SPMDs were removed from their frames and immediately sealed in individual amber glass jars. The jars were transported to the laboratory in a cooler (on ice and in darkness) and were kept in a freezer at — 20 °C until processing. Exposed SPMDs were analysed for selected PAHs. Sediment samples Sediment samples from the sampling sites were collected using an Ekman-Birge grab (Hydro-Bios Apparatebau GmbH, Kiel, Germany), which penetrated about 15 cm and collected about 0.1 m2 of surface sediments. Samples were air dried and the coarse material (>2.5mm) was removed. Duplicate sediment samples were extracted with toluene using the optimized accelerated solvent extraction technique, as described by Popp et al.1* Solvent exchange to acetonitrile was performed prior to HPLC analysis of PAHs. The organic carbon content of the sediment sample was measured with a total organic carbon (TOC) analyser (Leco) using sample combustion at 1000 °C. SPMD processing SPMD processing has been described previously.15 Briefly, the devices were subjected to exterior clean-up for biofouling removal. SPMDs were then dialysed three times with 250 mL of hexane per SPMD at 18 °C for 24 h and cleaned up by size exclusion chromatography (SEC). The SEC fraction containing the PAHs was collected, and solvent exchange from dichloromethane to acetonitrile was performed prior to HPLC analysis of PAHs. Table 1 Mean concentrations of PAHs in SPMDs (ng per SPMD; n — 2; mass of SPMD, 4.28 g) and in sediment samples (ng g dry weight based) at sampling sites in the Mansfeld region Sediment, Sediment, Sediment, SPMD", SPMD, SPMD, SPMD, PAH No. Site I Site II Site III trip blank Site I Site II Site III Naphthalene* <5 28 205 168 292 235 286 Acenaphthene 1 11 23 59 <10 24 36 51 Fluorene 2 52 33 216 <50 122 160 217 Phenanthrene 3 1024 546 2208 20 513 762 710 Anthracene 4 196 81 371 <4 52 62 91 Fluoranthene 5 1670 801 2393 <30 715 821 1048 Pyrene 6 1272 659 1624 8 933 956 1073 Benzo[a]anthracene 7 702 338 723 <20 105 89 97 Chrysene 8 682 667 798 <50 170 189 168 Benzo[é]fluoranthene 9 679 407 569 <20 74 57 80 Benzo[£]fluoranthene 10 274 210 215 <5 31 26 31 Benzo[a]pyrene 11 546 441 425 <5 31 26 33 Dibenz[a,/!]anthracene 12 103 79 78 <15 <15 15 <15 Benzo[g,/!,i]perylene 13 389 359 283 <15 14 31 25 Indeno[l ,2,3]pyrene 14 329 417 206 <25 <25 27 <25 Sum of PAHs 7929 5089 10373 197 3077 3491 3910 "Concentrations found in SPMDs are recovery rate corrected. ^Concentrations of naphthalene are not recovery rate corrected. /. Environ. Monit., 2001, 3, 602-609 603 km 0 12 3 Fig. 1 Map of the sampling sites (filled circles) in the stream Bose Sieben in the Mansfeld region in Saxony-Anhalt, Germany. The black areas on the left of the map represent the heaps and ponds from copper mining and processing. Quality control Because SPMDs have a propensity to sequester vapour phase contaminants,16 additional two trip blank SPMDs were exposed to air at each site while the water sampling SPMDs were being deployed and collected. Trip blanks were processed exactly as deployed samples and were used to define the contamination of the SPMDs during transportation and handling. In addition, fresh SPMDs were taken through the entire dialysis and clean-up procedure (procedural blanks). Spiked SPMDs were also analysed by fortifying fresh membranes and then processing them as samples. PAHs were spiked at 80 ng per SPMD for each individual component. Recovery rate values of the fortified PAHs from SPMDs were good and reproducible, with the exception of naphthalene. The average percentage recoveries of the remaining PAHs varied from 56% for indeno[l,2,3]pyrene to 117% for phenanthrene, and the relative standard deviation of three spiked samples did not exceed 11% for any compound. Analysis PAHs in sediment and SPMD extracts were analysed by HPLC (HP 1050) using a C18 Vydac201TP54 (250 mm x 4.6 mm id) column and a programmable fluorescence detector (FD). The mobile phase was acetonitrile-water pumped at 1 mL min-1 at 23 °C with gradient elution. The gradient elution started with 60% water and 40% acetonitrile (3 min); then, the acetonitrile content was increased with a linear gradient to 100% in 24 min, and was maintained in this condition for 13 min. Quantification of the PAH residues was accomplished using a seven-point, external standard curve. The standard curves were linear, with correlation coefficients for the investigated PAHs ranging between 0.996 and 0.999. No internal standards were employed for quantification using HPLC with FD; nevertheless, quantification using an external standard only is also permitted in the standard EPA 610 method. Method quantification limits (MQLs) for PAHs in sample extracts calculated from procedural blanks were determined as the average concentration plus ten times the standard deviation of the procedure blanks, ranging from 4ngmL_1 for anthracene to 50 ng mL-1 for chrysene. Acenaphthylene was not included in the analytical procedure because the compound shows no fluorescence. Distribution of PAHs using principal component analysis Principal component analysis (PCA) was used to compare the PAH distribution in different matrices (sediment, SPMD and aqueous phase) at the sampling sites. This was performed on data standardized to the total PAH concentration of each sample, focusing on relative patterns. The PCA represents the patterns by arranging the PAHs (variables) and sites along axes (principal components), which are assumed to represent basic factors or relationships. The first principal component (PCI) describes the maximum amount of variation of the data. Subsequent calculated principal components (PC2, PC3, etc.) describe the remaining variation of the data in decreasing order of importance. Each PC is orthogonal (uncorrelated) to the previous one. The results are depicted in plots of samples and variables, with the variables represented by lines running from the origin of the plot to the position of the variable loadings. The lines point in the direction of increasing variable relative concentrations, and the length of the line represents the extent of the increase. Data were modelled by PCA with KyPlot software.17 Results and discussion Absolute concentrations of PAHs in sediments and SPMDs The concentrations of the compounds of interest found in the sediments and the SPMDs exposed for 34 days at different sampling sites are presented in Table 1. Quantifiable amounts of all PAHs were found in sediments from all sampling sites. On average, the total PAH concentrations in the sediments were approximately ten times those found in the SPMDs on a ug g_1 basis. On the basis of total PAH residues, the sites can be ranked from lowest to highest as follows: Site II < Site I < Site III. The PAH concentrations in sediments ranged from 5.1 ug g_1 at Site II to 10.4 ug g_1 at Site III. The average relative percentage difference between duplicate sediment 604 /. Environ. Monit., 2001, 3, 602-609 samples from the same sampling site was 10%, and never exceeded 40%. The pollution levels can be assigned as elevated, although below the maximum admissible concentration of 20ugg_1 set by the Directive on the Disposal of Dredged Materials of Saxony-Anhalt.18 Apart from dibenz(a,/z)anthracene and indeno(l,2,3)pyrene at sampling sites I and III, all PAHs were also quantified in the SPMD samples. The concentrations of compounds in SPMDs (Table 1) were adjusted according to their recoveries from fortified SPMDs. The trip blank SPMDs were devoid of quantifiable residues of PAHs, except for naphthalene, phen-anthrene and pyrene. Only naphthalene was ignored in further discussions because of high trip blank values, large differences in results from duplicate samples and poor recovery from SPMDs; the phenanthrene and pyrene concentrations in the trip blanks were much lower (more than 20 and 100 times, respectively) than their concentrations in the SPMDs at the sampling sites. The average relative percentage difference between two SPMDs deployed at the same sampling site was 23%, and never exceeded 40%. The sites can be ranked from the lowest to the highest concentrations of total PAHs in SPMD samples as follows: Site I < Site II < Site III. The PAH concentrations increase downstream towards the mouth of Böse Sieben to the Süßer See lake. The concentrations of individual PAHs at the sampling sites generally followed the same pattern as the totals (Table 1). The highest total levels of PAHs sequestered by SPMDs in this study were of a comparable order of magnitude to the amounts found in SPMDs in other studies conducted under similar conditions.19-21 The PAH concentrations ranged from a low of 3.1 ug per SPMD at Site I to a high of 3.9 ug per SPMD at Site III. SPMD and sediment samples provide complementary information. PAH concentrations in sediments reflect long periods of time, because sediments are sinks for hydrophobic contaminants, while SPMDs only integrate water concentrations during the sampling period. Moreover, PAHs present in sediments are bound to particles, whereas SPMDs sample only PAHs truly dissolved in the water column. Water concentration estimation Sediment-based PAH concentrations in water (pore water) (CWs) were estimated from the concentrations found in sediment (Cs) using the equilibrium partitioning approach discussed by Di Toro et al.22 CWs = Cs/(focpKoc) (1) where foc is the fraction of sediment organic carbon, p is the sediment bulk density and Ä"oc is the sediment organic carbon-water partition coefficient. Ä"oc was calculated using Karickoff s approximation,23 i.e. Koc = 0.41 x Ä"ow. The octanol-water partition coefficient values (Ä"ow) of the PAHs utilized for the calculation are given in Table 2.foc measured in sediments was 3.92% at Site I, 1.56% at Site II and 4.86% at Site III. The substitution of Karickoff s equation by alternative correlations recently proposed to estimate Ä"oc from Kow25,26 yields PAH concentrations in pore water comparable (of the same order of magnitude) to the estimation results given in Table 2. The time-averaged water-borne concentrations of PAHs at the sampling sites can also be estimated from concentrations in exposed SPMDs. The details of the model development are available elsewhere,9'27'28 and are not presented here. In general, ambient water concentrations can be calculated using Cwm = CM/KM[l - exp(-ket)] (2) As applied here, CWM is the concentration of the analyte in water derived from SPMD (estimate of average value over the exposure period), CM is the concentration of the analyte in the SPMD (lipid + membrane), t is the exposure time in days, ke is the exchange rate constant for both overall uptake and elimination and KM is the equilibrium partition coefficient between SPMD and water. The selection of the most appropriate approach to estimate aqueous concentrations from concentrations in exposed SPMDs depends on whether the overall uptake is linear, curvilinear or equilibrium is attained between the SPMD and the aqueous phase during exposure.11 The time an analyte remains in the linear uptake phase (first-order uptake half-time, tn2) can be estimated from the reported equilibrium partitioning coefficient (KM) and actual sampling rate (RSc) for a specific average temperature value at each sampling site using tm = (\n2)KMVM/Rsc (3) where VM is the volume of the SPMD (lipid + membrane) and Rsc is the SPMD sampling rate given by Rsc = RsFi (4) where i7; is 1 — the fractional reduction in uptake flux or sampling rate Rs determined under defined conditions due to fouling impedance. Rsc is related to ke by Rsc = keKMVM (5) The chemical uptake into SPMD remains linear and integrative in the initial period of the exposure until the concentration factor (ratio Cm/CWm) in the SPMD reaches approximately half-saturation (ket < In 2) and eqn. (2) can be reduced to Cwm = (CmKmVCRscO (6) Among environmental variables, mainly temperature and biofouling can affect the sampling rate. We assumed that the hydrodynamic conditions at the sampling sites did not affect the uptake kinetics dramatically, although it has been reported that the aqueous diffusion boundary layer at the membrane surface affects contaminant uptake for compounds with log Kow > 4.5.15'29 In general, elevated sampling rates are expected in turbulent environments, and the application of laboratory-derived sampling rates may cause overestimation of the aqueous concentrations. Sampling rates for the PAHs at different temperatures have been reported,27 and were utilized for water concentration estimation in this study. The sampling Table 2 Estimates of dissolved pore water concentrations from sediments, Cws, at the sampling sites. Concentrations reported are in ng L_1. Recommended octanol-water partition coefficients for individual PAHs24 were used to estimate the CWs values using eqn. (1) PAH log Site I, Cws Site II, Cws Site III, Cws Acenaphthene 4.0 68.5 920.6 758.8 Fluorene 4.2 523.3 833.4 1752.8 Phenanthrene 4.5 5164.5 6910.7 8979.8 Anthracene 4.6 785.2 814.4 1198.5 Fluoranthene 5.1 2115.6 2546.6 2444.6 Pyrene 5.1 1611.4 2095.2 1659.0 Benzo[a]anthracene 5.9 140.9 170.3 117.1 Chrysene 5.7 217.0 532.7 204.8 Benzo[6]fluoranthene 5.8 171.6 258.2 116.0 Benzo[£]fluoranthene 6.0 43.7 84.1 27.7 Benzo[a]pyrene 6.2 54.9 111.4 34.5 Dibenz[a,/!]anthracene 6.8 2.9 5.6 1.8 Benzo[g,/!,z]perylene 6.9 7.8 18.1 4.6 Indeno[l,2,3]pyrene 6.8 8.3 26.5 4.2 X PAHs 10915.8 15327.6 17304.0 /. Environ. Monit., 2001, 3, 602-609 605 Table 3 Estimates of dissolved aqueous concentrations from SPMDs, CWM, at sampling sites. Concentrations reported are in ngL~'. Published sampling rates (Rs) and equilibrium SPMD-water partition coefficients (KM) for individual PAHs27 were utilized to estimate the CWM values using eqn. (2) (curvilinear model) and eqn. (6) (linear model), respectively PAH MW log KM 10°C/L d~' 18°C/Ld~' Site I, CWM Site II, CWM Site III, CWM Model used Acenaphthene 154.2 4.05 2.7 2.3 0.7 1.0 1.4 Curvilinear Fluorene 166.2 4.21 3.0 1.7 2.7 3.7 5.3 Curvilinear Phenanthrene 178.2 4.47 3.8 3.6 3.4 5.1 4.8 Linear Anthracene 178.2 4.67 2.9 3.6 0.4 0.5 0.7 Linear Fluoranthene 202.3 4.68 3.6 4.5 4.6 5.2 6.4 Linear Pyrene 202.3 4.79 4.5 5.2 4.9 5.0 5.5 Linear Benzo[a]anthracene 228.3 5.32 3.2 3.2 0.8 0.7 0.7 Linear Chrysene 228.3 5.32 3.7 4.8 1.1 1.1 1.0 Linear Benzo[Z)]fluoranthene 252.3 5.55 2.8 3.0 0.6 0.5 0.7 Linear Benzo[£]fluoranthene 252.3 5.44 2.9 3.9 0.2 0.2 0.2 Linear Benzo[a]pyrene 252.3 5.11 3.2 3.7 0.2 0.2 0.2 Linear Dibenz[a,/!]anthracene 278.4 4.83 3.0 3.8 NR" 0.1 NR Linear Benzo[g,/!,;]perylene 276.3 4.51 2.0 3.0 0.2 0.3 0.2 Linear Indeno[l,2,3]pyrene 267.0 4.04 1.8 1.9 NR 0.4 NR Linear X PAHs 22.6 27.9 30.6 "NR, no residue found in SPMD sampler. rates at individual sampling sites were interpolated from these values for an average temperature at each sampling site. An average fouling resistance of 20% (i7; = 0.8) was employed for biofouled SPMDs to correct for reduction in SPMD uptake.30 Laboratory-derived SPMD sampling rates Rs for PAHs utilized for calculation are given in Table 3. Except for acenaphthene and fluorene, the estimated first-order uptake half-time was longer than the exposure period, and the linear model [eqn. (6)] was used to calculate the dissolved water-borne concentrations of the compounds. For acenaphthene and fluorene, the curvilinear model [eqn. (2)] was used. It should be noted that the water concentration estimated with the SPMDs is an average concentration over a 34 day interval, not the maximum concentration during that interval. The estimated ambient concentrations of the selected contaminants by the two techniques are presented in Tables 2 and 3. Water-sediment equilibrium issues Absolute water PAH concentrations, estimated from sediment concentrations, were on average almost three orders of magnitude higher than those calculated from SPMDs, although the SPMDs were exposed close to the bottom sediment. To assess the net flux of PAHs between water and sediment at the sampling sites, fugacity quotients (ratio of the fugacity in the sediment fs to the fugacity in the water /w) can be calculated using the SPMD and the sediment concentration data. It can be shown that the fugacity quotient can be calculated using the ratio of the aqueous concentrations in equilibrium with individual compartments fs/f* Cws/Ci ws'^wm (7) The fugacity quotient can be cautiously interpreted as an indication of sediment-water equilibrium status. A ratio of unity indicates equilibrium, a ratio of less than unity indicates net movement from water to sediment and a ratio of more than unity indicates net movement from sediment to water. Fig. 2 shows the sediment/water fugacity quotients calculated using the approach outlined above. For PAHs in this study, movement is predicted to be from sediment to water, i.e. the sediment has a tendency to release these compounds. The low actual aqueous concentrations of PAHs in surface water can promote the dissolution of sediment-bound PAH residues. Thus, in addition to contamination from other sources, the water concentration at sampling sites might be elevated due to contaminant mobilization from historically contaminated sediments. Of the PAHs, anthracene and phenanthrene show the highest remobilization potential. The fugacity quotients have the highest values at Site II. PAH patterns In order to determine the nature of PAH pollution, we compared the PAH patterns of sediment, SPMD and the dissolved phase at different sampling sites. For this purpose, the relative concentrations of PAHs in samples were analysed by PCA. From an inspection of the PCA pattern analysis for the matrices, the score plot [PCI vs. PC2, Fig. 3(a)] shows the separation of the samples along the principal components. As can be seen in the loading plot [Fig. 3(b)], the compounds (PAHs) are separated on the principal component plane (PCI x PC2) according to their molecular weight or lipophi-licity. Four main groups can be distinguished. The first group represents the di-aromatics, acenaphthene and fluorene. The second group contains the tri-aromatics, phenanthrene and anthracene. These two groups represent the most water-soluble PAHs. The third group comprises the tetra-aromatics, fluoranthene and pyrene. Finally, the fourth group consists of the remaining tetra-, penta- and hexa-aromatics, i.e. the least water-soluble and most hydrophobic of the compounds studied. lndeno(1,2,3)pyrene Benzo(g,/i,/)perylene Dibenz( a, h)anthracene Benzo(a)pyrene Benzo(*r)fluoranthene Ben20(b)fluoranthene Chrysene Benzo(a)anthracene Pyrene Fluoranthene Anthracene Phenanthrene Fluorene Acenaphthene Fig. 2 Sediment/water fugacity quotients of the PAHs at the sampling sites, calculated as described in the text. The broken lines show the quotient range where the sediment is predicted to be close to equilibrium with the aqueous phase. (Fugacity quotients could not be calculated for Site I and Site III for indeno(l,2,3)pyrene and dibenz(a,/!)anthracene due to concentrations below the limit of detection in SPMDs at these sites.) 606 /. Environ. Monit., 2001, 3, 602-609 0.2 0.1 -0.3 SPMD and / mmm water from I ^A ' SPMD V • (a) in \ water from M7 sediment /wsi\ / • I sl sediment * ^ and source \ s" °' - materials V__^ i i V* * / 'áRM1579 L / U I I -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 PC1 [66%] 0.5 -0.5 Tetra-aromatics ( * ) Di-aromatics (b) I2 \ If \ Tetra-, penta and , 5.1). Along PCI, sediment, SPMD samples and water patterns estimated from SPMDs have negative scores, whereas water patterns estimated from sediments have positive scores. The second most important trend (PC2) explains 22% of the variance in the data. Samples with a positive score on the PC2 axis are characterized by higher relative concentrations of the di- and tetra-aromatics, whereas samples with negative scores have higher relative concentrations of penta- and hexa-aromatics. Along this principal component, the sediment samples have negative scores, whereas the SPMD samples and water patterns derived from SPMDs have positive scores. This is because higher molecular PAHs with high A"ow values are more likely to be associated with dissolved, particulate and sediment organic carbon than dissolved in the water, and thus available to be sampled by SPMDs. Sediment scores from different sampling sites have a much greater spread within their cluster than the PAH profiles within the SPMD matrix. The sediment sample from Site II contains higher relative concentrations of very hydrophobic PAHs (log Á"ow > 5.5) in comparison with sites I and III, which could indicate a contribution from an additional source of PAH pollution at this site. Sediment from Site III contains higher relative concentrations of fluorene, phenanthrene and anthracene and lower relative concentrations of more hydrophobic compounds (log A"ow > 5.5) in comparison with sites I and II. The estimated water patterns for SPMD are very similar to the SPMD patterns, whereas the calculated water patterns from sediment vary considerably from the sediment PAH patterns. Water patterns calculated from SPMD and sediment do not cluster together on the PCA plot. The pore water at equilibrium with sediment contains higher relative concentrations of di- and tri-aromatics (especially anthracene and phenanthrene, respectively) than the water column. In surface water, these compounds can be reduced due to their elevated volatility and by different degradation processes (i.e. bio-degradation and/or photodegradation). Sources of PAH contamination No dramatic rise in absolute PAH concentration was observed for any of the sampling sites, for any matrix, which indicates that the PAHs are likely to originate from diffusive sources rather than from a small number of discrete point sources. To identify a similarity in contamination pattern with some of the possible pollution sources (Table 4), principal component scores were calculated for potential source materials, including the Theisen sludge and its aqueous leachate, the Kupferschiefer and the coal tar SRM 1579, using eigenvectors (scoring coefficients) of the data covariance matrix obtained in the PCA [Fig. 3(a)]. Along PCI, Kupferschiefer has a very positive score, which clearly sets it apart from the remaining data. Theisen sludge patterns and the coal tar pattern are not separated well from each other on the PCI axis and their scores are slightly more positive in comparison with SPMD and sediment samples, respectively. Along PC2, Theisen sludge and coal tar have negative scores. This demonstrates a similarity in the composition of this sediment to that of these two materials. Another method to determine the PAH source is to calculate specific PAH/PAH ratios.33-35 These ratios can be compared with the fingerprints of PAHs from pyrolytic or petrogenic origin to identify the most likely contamination source material. The usual index of combustion and/or anthropogenic input is an increase in the proportion of the thermodynamically less stable parent PAH isomers relative to the stable isomers (e.g. anthracene relative to phenanthrene , fluoranthene relative to pyrene, benzofa]anthracene relative to chrysene, etc.). One difficulty in identifying PAH origins is the possible coexistence of many contamination sources, and the transformation processes that PAHs can undergo in matrices from diverse environmental compartments. The good correlation observed between PAH pairs with similar physicochemical properties indicates their similar behaviour irrespective of the sampling sites and matrices. On the principal component plane (PCI x PC2) [Fig. 3(b)], good correlation was observed, especially between relative concentrations of phenanthrene and anthracene (r = 0.94), fluoranthene and pyrene (r = 0.87) and benzo[a]anthracene and chrysene (r = 0.84). The ratios of phenanthrene/anthracene (Phe/Ant), fluoranthene/pyrene (Flt/Py) and chrysene/benzo-[a]anthracene (Chry/BaA) were examined as origin indices. The Phe/Ant ratio can be seen to be very high in petrogenic pollution by PAHs (i.e. Phe/Ant > 10), but lower in pyrolytic /. Environ. Monit., 2001, 3, 602-609 607 Table 4 PAH contents of some potential pollution source materials (d.w., dry weight) PAH Theisen sludge°/|ig g_1 d.w. Theisen sludge aqueous leachate'Vug L_1 Coal tar SRM 15797ugg~' d.w. Kupferschiefer/ ugg~' d.w. Acenaphthene 1.5 1.3 NRrf NR Fluorene 23.6 4.8 140.0 0.05 Phenanthrene 209.1 14.6 462.0 4.40 Anthracene 38.8 2.5 101.0 0.01 Fluoranthene 76.1 3.3 322.0 0.14 Pyrene 120.2 5.9 235.0 0.46 Benzo[a]anthracene 38.7 1.1 98.6 0.07 Chrysene 75.5 2.6 71.7 0.73 Benzo[6]fluoranthene 26.7 1.8 66.0 0.22 Benzo[£]fluoranthene 9.8 0.4 43.0 NR Benzo[a]pyrene 23.9 1.5 95.8 NR Dibenz[a,/!]anthracene 1.2 NR NR NR Benzo[g,/!,!]perylene 14.5 0.4 53.7 0.37 Indeno[l,2,3]pyrene 9.0 0.7 60.2 NR X PAHs 668.6 40.9 1749.0 6.47 "Taken from Popp et al.14 *Taken from Paschke et al.7 Taken from ref. 32. ''NR, no residue found. contamination cases. In the case of pyrogenic pollution, the Flt/Py ratio ought to be > 1, and the Chry/BaA ratio ought to be <1.35-36 The origin indices of the sediment and SPMD samples and those derived from the estimated dissolved aqueous phase composition at the sampling sites were compared with origin indices of potential pollution source materials, including the Theisen sludge, the Kupferschiefer and the reference material coal tar SRM 157 9 32 (Table 5). Coal tar SRM 1579 is characterized by origin indices typical of a material generated by pyrolytic processes, as indicated by the low Phe/Ant ratio of 4.57, the elevated Flt/Py ratio (1.37) and the low value of Chry/BaA (0.73). On the other hand, the criteria for a petrogenic PAH origin are confirmed very clearly when inspecting the origin indices calculated for the black shale Kupferschiefer. The characteristic extremely high Phe/Ant ratio of 344, low Flt/Py ratio (0.31) and high Chry/BaA ratio (10.37) allow for the clear differentiation of this petrogenic PAH source from other sources. Theisen sludge exhibits characteristic indices distinct from those of petrogenic origin. The Phe/Ant ratio is 10 in Theisen sludge samples. In contrast to typical material of pyrolytic Table 5 Selected PAH ratios (origin indices) for SPMDs, sediment samples, estimated water compositions and potential source samples in the Mansfeld region Sample Phe/Ant Flt/Py Chry/: Sediment at Site I 5.22 1.31 0.97 Sediment at Site II 6.74 1.22 1.97 Sediment at Site III 5.95 1.47 1.10 SPMD at Site I 9.91 0.77 1.62 SPMD at Site II 12.36 0.86 2.13 SPMD at Site III 7.79 0.98 1.74 Water from SPMD 8.13 0.94 1.30 at Site I Water from SPMD 10.49 1.04 1.66 at Site II Water from SPMD 6.84 1.17 1.31 at Site III Pore water from sediment 6.58 1.31 1.54 at Site I Pore water from sediment 8.49 1.22 3.13 at Site II Pore water from sediment 7.49 1.47 1.75 at Site III SRM 1579 (coal tar) 4.57 1.37 0.73 Theisen sludge 10.44 0.95 2.94 Kupferschiefer 344.05 0.31 10.37 "NR, no residue found. origin, Theisen sludge is characterized by Flt/Py values < 1 and Chry/BaA values > 1. The sediment, SPMD samples and calculated water patterns at the sampling sites are characterized by Phe/Ant values < 10. One exception is the SPMD sample at Site II and the water pattern derived from this sample (Phe/Ant = 12.36 and 10.49, respectively). However, this ratio might be additionally elevated because of selective photo-oxidation of anthracene during transport in the dissolved phase.37 This hypothesis is supported by the fact that the Phe/Ant values in the SPMD samples, which reflect the composition of the dissolved PAHs in water, are higher than those in sediment samples. The results of the Flt/Py ratios examined are ambiguous. Flt/Py ratios > 1 are observed in sediment samples and pore water patterns derived from sediment data, whereas ratios between 0.77 and 1.17 characterize the SPMD samples and water patterns from SPMD data. Except for the sediment sample at site I, Chry/BaA ratios > 1 were observed in all samples. Conclusions SPMD and sediment samples provide complementary information. The use of sediments to predict water concentrations and patterns may not be representative of the concentrations and patterns in the upper levels of the water column. PAH concentrations and patterns in sediment are changed by weathering and ageing, and reflect longer periods of time because sediments are sinks for hydrophobic contaminants, while SPMDs integrate water concentrations only during the sampling period. Moreover, PAHs present in sediment are bound to particles, whereas SPMD samples only PAHs truly dissolved in the water column. The comparison of data obtained by PAH analysis in sediment samples and SPMDs allows the specific distribution of PAHs to be determined in individual environmental compartments and the mobilization potential of these compounds to be assessed. Moreover, the evaluation of contaminant patterns in sediment and SPMD samples permits the assessment of the possible pyrolytic, industrial origin of the PAHs in the region. Although it was not possible to clearly identify one definite contamination source in the region, the results indicate that Theisen sludge cannot be ruled out as a possible source of PAH pollution. However, a conclusive statement about the origin of pollution will entail additional sampling with a higher density of sampling sites. In addition, studies currently being conducted on the assessment of the leaching behaviour of the Theisen sludge will produce more information on the potential contribution of this industrial waste to the pollution situation in the region of Mansfeld. 608 /. Environ. Monit., 2001, 3, 602-609 Acknowledgements The authors wish to thank Elke Büttner, Coretta Bauer and Doris Sonntag for sample preparation and instrumental measurements. References 1 J. M. Neff, Polycyclic Aromatic Hydrocarbons in the Aquatic Environment, Sources, Fates, and Biological Effects, Applied Science, London, 1979, p. 262. 2 Microbial Transformation and Degradation of Toxic Organic Chemicals, ed. L. Y. Young and C. E. Cerniglia, Wiley, New York, 1995. 3 J. L. Schnoor, Environmental Monitoring — Fate and Transport of Pollutants in Water, Air and Soil, Wiley, New York, 1996. 4 C. T. Chiou, L. J. Peters and V. H. Freed, Science, 1979, 206, 831. 5 C. T. Chiou, S. E. McGrody and D. E. Kile, Environ. Sei. Technol., 1998, 32, 264. 6 H. Weiss, M. Morency, K. Freyer, J. Bourne, D. Fontaine, B. Ghaleb, R. Mineau, M. Moder, P. Morgenstern, P. Popp, M. Preda, H.-C. Treutler and R. Wennrich, Sei. Total Environ., 1997, 203, 65. 7 A. Paschke, K. Freyer, H. C. Treutler, R. Wennrich, P. Popp, M. Moder, H. Weiß and G Schüürmann, in Theisenschlamm Report, ed. H. Weiß and B. Daus, UFZ Centre for Environmental Research, Leipzig, 2001, in press. 8 J. N. Huckins, M. W. Tubergen and G. K. Manuweera, Chemo-sphere, 1990, 20, 533. 9 J. N. Huckins, G. K. Manuweera, J. D. Petty, D. Mackay and J. A. Lebo, Environ. Sei. Technol, 1993, 27, 2489. 10 H. F. Prest, Chemosphere, 1992, 25, 1811. 11 J. D. Petty, J.N. Huckins, C. E. Orazio, J. A. Lebo, B.C. Poulton, R. W. Gale, C. S. Charbonneau and E. M. Kaiser, Environ. Sei. Technol, 1995, 29, 2561. 12 J. A. Lebo, R. W. Gale, D. E. Tillitt, J. N. Huckins, J. C. Meadows, C. E. Orazio and D. J. Schroeder, Environ. Sei. Technol, 1995, 29, 2886. 13 J. A. Lebo, J. L. Zajicek, J. N. Huckins, J. D. Petty and P. H. Peterman, Chemosphere, 1992, 25, 697. 14 P. Popp, P. Keil, M. Moder, A. Paschke and U. Thuss, /. Chromatogr. A, 1997, 774, 203. 15 B. Vrana and G. Schüürmann, Environ. Sei. Technol, submitted for publication. 16 J. D. Petty, J. N. Huckins and J. L. Zajicek, Chemosphere, 1993, 27, 1609. 17 K. Yoshioka, KyPlot 2000, ver. 2.0 beta 12, http://www.qualest. co.jp/Download/KyPlot/kyplot_e.htm 18 Richtlinie für die Entsorgung von Baggergut im Land Sachsen-Anhalt, ed. Ministerium für Raumordnung und Umwelt des Landes Sachsen-Anhalt, LSA Nr. 18/99, Magdeburg, 1999, pp. 583-589. 19 J. A. Lebo, J. L. Zajicek, C. E. Orazio, J. D. Petty, J. N. Huckins and E. H. Douglas, Polycycl. Arom. Comp., 1996, 8, 53. 20 J. B. Moring and D. R. Rose, Chemosphere, 1997, 34, 551. 21 B. Vrana, A. Paschke, P. Popp and G. Schüürmann, Environ. Sei. Pollut. Res., 2001, 8, 27. 22 D. M. Di Toro, C. S. Zarba, D. J. Hansen, W. J. Berry, R. C. Swartz, C. E. Cowan, S. P. Pavlou, H. E. Allen, N. A. Thomas and P. R. Paquin, Environ. Toxicol. Chem., 1991, 10, 1541. 23 S. W. Karickoff, Chemosphere, 1981, 10, 833. 24 D. Mackay, W. Y. Shiu and K. C. Ma, Illustrated Handbook of Physical-Chemical Properties and Environmental Fate of Organic Chemicals, Lewis Publishers, Boca Raton, FL, 1992, vol. II. 25 A. Sabljic, H. Güsten, H. Verhaar and J. Hermens, Chemosphere, 1995, 31, 4489. 26 J. R. Baker, J. R. Mihelcic and E. Shea, Chemosphere, 2000, 41, 813. 27 J. N. Huckins, J. D. Petty, C. E. Orazio, J. A. Lebo, R. C. Clark, V. L. Gibson, W. R. Gala and K. R. Echols, Environ. Sei. Technol., 1999, 33, 3918. 28 J. N. Huckins, J. D. Petty, H. F. Prest, R. C. Clark, D. A. Alvarez, C. E. Orazio, J. A. Lebo, W. L. Cranor and B. T. Johnson, A Guide for the Use of Semipermeable Membrane Devices (SPMDs) as Samplers of Waterborne Hydrophobic Organic Contaminants, API 4690, American Petroleum Institute (API), Washington, DC, 2000. 29 K. Booij, H. M. Sleiderink and F. Smedes, Environ. Toxicol. Chem., 1998, 17, 1236. 30 J. D. Petty, B. C. Poulton, C. S. Charbonneau, J. N. Huckins, S. B. Jones, J. T. Cameron and H. F. Prest, Environ. Sei. Technol, 1998, 32, 837. 31 D. Mackay, Environ. Sei. Technol, 1979, 13, 1218. 32 Reference Materials for Organic Analytes in Industrial, Foodstuff and Environmental Analysis, Promochem GmbH, Wesel, Germany, 1998. 33 M. Blumer and W. W. Youngblood, Science, 1975, 188, 53. 34 S. Sporstol, N. Gjos, R. G. Lichtenthaler, K. O. Gustavsen, K. Urdal, F. Oreld and J. Skei, Environ. Sei. Technol, 1983, 17, 282. 35 H. H. Socio, P. H. Garrigues and M. Ewald, Mar. Pollut. Bull, 2000, 40, 387. 36 K. T. Benlahcen, A. Chaoui, H. Budzinski, J. Bellocq and P. H. Garrigues, Mar. Pollut. Bull, 1997, 34, 298. 37 A. L. Irwan, A. A. Razak, F. Ni, M. F. Gin and E. R. Christensen, Water Air Soil Pollut., 1998, 101, 417. /. Environ. Monit., 2001, 3, 602-609 609 Príloha 3 Vrana B., Popp P., Paschke A., and Schúúrmann G., Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water, Anal. Chem., 2001, 73, 5191-5200. Anal. Chem. 2001, 73, 5191-5200 Membrane-Enclosed Sorptive Coating. An Integrative Passive Sampler for Monitoring Organic Contaminants in Water Branislav Vrana,* Peter Popp,41 Albrecht Paschke, and Gerrit SchUUrmann Department of Chemical Ecotoxicology, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany An integrative sampler that consists of a bar coated with poly(dimethylsiloxane) (PDMS) enclosed in a dialysis membrane bag has been developed combining the advantages of the passive sampling approach with solventiess preconcentration of organic solutes from aqueous matrixes and subsequent desorption of the sequestered analytes on-line with a capillary GC/MS system. The performance of the sampler was tested for integrative sampling of hydrophobic persistent organic pollutants including y-hexachlorocyclohexane, hexachlorobenzene, 2,2'-bis(4-chlorophenyl)-l, 1 '-dichloroethylene, polycyclic aromatic hydrocarbons, and polychlorinated biphenyls in the laboratory in a continuous-flow system. Linear uptake of all test analytes during exposure periods up to one week has been observed, and concentration proportionality of response of the sampler has been demonstrated. Over the range of controlled laboratory conditions, the magnitude of sampling rate values varied from 47 to 7 0 0 fiL, h 1 per sampler. The uptake rate of chemicals was dependent on their molecular mass, as well as on the partition coefficient between the PDMS and water. A decrease in sampling rates with decreasing water temperature was observed. The sampling device has the potential to detect low aqueous concentrations (ngto pgL_1) of test substances. Qualitative and quantitative assessment of pollution of ecosystems by persistent organic pollutants (POPs) is a continuing challenge to environmental scientists. In aquatic systems, it is important to obtain information on the time-weighted average (TWA) concentrations of pollutants, which is a fundamental part of an ecological risk assessment process for chemical stressors. Moreover, the quantification of freely dissolved concentrations of pollutants in water is needed for approximate characterization of the bioavailable fraction. Concentrations of truly dissolved contaminants cannot be determined by most water sampling methods. Instead, total quantities of analytes are measured, including those molecules that are not readily bioavailable because they are bound to * Corresponding author: (e-mail) bv@uoe.ufz.de; (phone) ++49 341 235 26 18; (fax) ++49 341 235 2401. * Department of Analytical Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany. 10.1021/ac010630z CCC: $20.00 © 2001 American Chemical Society Published on Web 10/04/2001 dissolved colloids present in water. Grab water samples provide information only about contaminant concentration at the moment of sampling and may fail to account for episodic contamination events. Because of the low aqueous solubility of many contaminants, it is often impossible to excise sufficiently large water samples to achieve instrumental detection limits. For these reasons, an integrative approach is needed, which would provide information about truly dissolved TWA contaminant concentrations over a long time period. Passive sampling devices allow convenient measurement of an average concentration over a long time period, on the order of several weeks. In contrast to active sampling, they require no mechanical devices to collect sample or a series of samples; this makes the method inexpensive, suitable to use at remote sites, and perhaps less prone to vandalism. The successful use of passive monitors in the industrial hygiene field for monitoring exposure of workers to chemicals in the air has contributed to the application of the same principle to dissolved organic contaminants in aquatic environments.12 Despite numerous shortcomings of the earlier developed devices, their use in field studies34 demonstrated that the in situ passive sampling approach had considerable potential. Most passive sampling devices typically consist of a receiving phase, with a high affinity for organic pollutants, separated from the aquatic environment by a diffusion-limiting membrane.5-8 They can be calibrated in the laboratory so that TWA concentrations of organic pollutants can be determined in field studies.9 Sbdergren2 developed a sampler design consisting of a dialysis membrane filled with organic solvents. The disadvantage of this design was the successive loss of the organic solvent from the device by diffusion through the membrane during exposure. (1) DiGiano, F. A.; Elliot, D.; Leith, D. Environ. Sei. Technol. 1988,22, 1365-1367. (2) Sbdergren, A. Environ. Sei. Technol. 1987, 21, 855-859. (3) Sbdergren, A. Ecotoxicol. Environ. Saf. 19 9 0, 19, 143-149. (4) Litten, S.; Mead, B.; Hassett, J. Environ. Toxicol. Chem. 1993, 12, 639-647. (5) Brown, R. H. J Environ. Monit. 2000, 2, 1-9. (6) Zabiegala, B.; Kot, A.; Namiesnik, J. Chem. Anal. 2 0 0 0, 45, 645-657. (7) Kot, A.; Zabiegala, B.; Namiesnik, J. Trends Anal. Chem. 2 0 0 0, 19, 446-459. (8) Namiesnik, J.; Gorecki, T. LC-GC Eur. 2000, (Sept), 678. (9) Huckins, J. N.; Petty, J. D.; Orazio, C. E.; Lebo, J. A.; Clark, R. C; Gibson, V.L.; Gala, W.R.; Echols, K. R. Environ. Sei. Technol. 1999,33, 3918-3923. Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5191 Huckins et al.10,11 described the development of a semipermeable membrane device (SPMD) for passive and integrative in situ monitoring of waterborne contaminants. The SPMD sampler consists of lay-flat polyethylene tubing containing a thin film of triolein, a high molecular weight neutral lipid. The utility of the SPMD has been shown for monitoring aqueous residues of polychlorinated biphenyls (PCBs),12 various organochlorine pesticides,13 polychlorinated dibenzofurans and dibenzo-p-dioxins,14 and polycyclic aromatic compounds (PAHs).15 The application of the device is limited to nonionized contaminants. Zhang et al.16 described a direct solid-phase micro extraction (SPME) of complex aqueous samples with hollow fiber membrane protection. In this approach, the fiber of an SPME device was placed inside a cellulose hollow membrane, which allows target analytes to diflu.se through while excluding high molecular weight interfering compounds. This arrangement can be used for determination of truly dissolved contaminants in aqueous samples; however, it is not suitable for passive sampling over a long time period. Recently, Alvarez et al.17 and Kingston et al.18 described development of passive samplers that enable to widen the application to a broader range of contaminants including low-hydrophobic substances (log K0„ < 4) such as atrazine,17,18 diazinon,17 17a-ethynylestradiol,17 or diuron.18 These samplers consist of a hydrophilic membrane material enveloping immobilized solid-phase materials as an alternative to a liquid receiving phase. The common disadvantage of the above-mentioned passive sampling techniques is a laborious recovery of analytes from samplers after exposure by solvent extraction or dialysis19 and a need for additional cleanup of the samples before gas chromatographic analysis.15,20,21 To make the passive sampling technology more suitable for routine monitoring, low-cost and less time-consuming sample processing is required. Sample processing with reduced solvent consumption would also minimize the risk of sample contamination during handling in the laboratory and enable to improve the quality control measures. Recently, a novel solventless and simple technique for pre-concentration of organic solutes from aqueous matrixes, the stir bar sorptive extraction (SBSE), was developed by Baltussen et al.22 In SBSE, a stir bar coated with poly(dimethylsiloxane) (10) Huckins, J. N.; Tubergen, M. W.; Manuweera, G. K. Chemosphere 1990, 20, 533-552. (11) Huckins, J. N.; Manuweera, G. K.; Petty, J. D.; Mackay, D.; Lebo, J. A Environ. Sci. Technol 1993,27, 2489-2496. (12) Prest, H. F. Jarman, W. M.; Burns, S. A.; Weismuller, T.; Martin, M.; Huckins, J. N. Chemosphere 1992, 25, 1811-1823. (13) Petty, J. D.; Huckins, J. N.; Orazio, C. E.; Lebo, J. A; Poulton, B. C; Gale, R W.; Charbonneau, C. S.; Kaiser, E. M. Environ. Sci. Technol. 1995, 29, 2561-2566. (14) Lebo, J. A; Gale, R. W.; Tillitt, D. E.; Huckins, J. N.; Meadows, J. C; Orazio, C. E.; Schroeder, D. J. Environ. Sci. Technol. 1995, 29, 2886-2892. (15) Lebo, J. A.; Zajicek, J. L; Huckins, J. N.; Petty, J. D.; Peterman, P. H. Chemosphere 1992, 25, 697-718. (16) Zhang, Z.; Pbrschmann, J; Pawliszyn, J. Anal. Commun. 1996, 33, 219— 221. (17) Alvarez, D. A.; Huckins, J. N.; Petty, J. D. SETAC-USA, Philadelphia, 1999; Poster. (18) Kingston, J. K; Greenwood, R.; Mills, G. A; Morrison, G. M.; Persson, B. L. J Environ. Monit. 2 0 0 0, 2, 487 -495. (19) Huckins, J. N.; Tubergen, M. W.; Lebo, J. A; Gale, W.; Schwartz, T. R. J. Assoc. Off. Anal. Chem. 1990, 73, 290-293. (20) Gustavson, K. E.; Harkin J. M. J Chromatog., A 2 0 0 0 , 883, 143-149. (21) Petty, J. D.; Jones, S. B.; Huckins, J. N.; Cranor, W. L; Parris, J. T.; McTague, T. B.; Boyle, T. P. Chemosphere 2000, 41, 311-321. Figure 1. Schematic diagram of the MESCO passive sampling device. A Gerstel-Twister bar used for SBSE (component 1) is enclosed in a dialysis membrane bag made from regenerated cellulose (component 2). The dialysis membrane bag is filled with 3 mL of bidistilled water (component 3) and sealed at each end with Spectra Por enclosures (component 4). (PDMS) is placed in the sample and stirred for a predetermined time. The stir bar is then thermally desorbed on-line with a capillary GC/ MS system. The use of PDMS as a receiving organic phase in extraction and thermodesorptionhas several advantages over other sorbents including inertness, negligible permanent sorption and reactions of analytes on it, and good blanks in GC analyses.23 Absorptive partitioning is the predominant mechanism of extraction of analytes into PDMS.24 The applicability of SBSE was demonstrated for the analysis of volatile and semivolatile micro pollutants from aqueous samples.22 In this work, we describe an adaptation of this novel technique to integrative passive sampling for hydrophobic persistent organic pollutants in aqueous environment THEORY Previously, models have been developed describing the uptake kinetics of organic contaminants in water by passive samplers constructed as a solvent-filled dialysis membrane25 or triolein-filled polyethylene membrane.11 The passive sampling device described in this study consists of a hydrophobic solid receiving phase enclosed in a water-filled hydrophilic semipermeable membrane (Figure 1). The passive sampler lowered in aqueous solution can be divided into several compartments including the bulk aqueous phase with constant solute concentration, the stagnant aqueous boundary layer, possible biofilm layer, the membrane, the inner aqueous phase, and (22) Baltussen, E.; Sandra, P.; David, F.; Cramers, C. J Microcolumn Sep. 1999, 11, 737-747. (23) Baltussen, E.; David, F.; Sandra, P.; Janssen, H.-G.; Cramers, C. A. J Chromatog., A 1998, 805, 237-247. (24) Baltussen, E.; Sandra, P.; David, F.; Janssen, H.-G.; Cramers, C. Anal. Chem. 1999, 71, 5213-5216. (25) Johnson, G. D. Environ. Sci. Technol. 1991,25, 1897-1903. 5192 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 the receiving organic phase. Under steady-state conditions, the flux of the solute is assumed to be constant and equal in each of the individual compartments. The overall mass transfer from the bulk aqueous phase to the receiving organic phase includes several diffusion and interfacial transport steps across all barriers. The resistances of each barrier to the mass transfer of analytes are assumed to be additive and independent,26 and the interfacial resistances are assumed to be negligible compared with diffusional resistances.27 Also, negligible accumulation of analyte in the diffusion-limiting membrane is assumed. Then, the rate of transport can be described by the overall mass-transfer coefficient km (m s-1), relating the net diffusive steady-state flux of the solute J(kg s-1) to its concentrations in the bulk aqueous phase Cw (kg m~3) and the receiving organic phase Cs (kg m~3) j= dMg/ at = vs dCg/ at = k0VAa( cw - cy ksw) ( d where Ms (kg) is the mass of analyte in the receiving organic phase, A is the membrane surface area (m2), a is the pore area of the membrane as fraction of total membrane area (membrane porosity), Ksw is the receiving organic phase/ water partitioning coefficient, and t (s) equals time. Equation 1 can be integrated : (Ms - M0)/Rst (6) Ca(t) Can + (CwKa Can) exp (2) where Cso is the concentration of analyte in the organic phase at t = 0. In the initial uptake phase, when the exponential term is very small (^1) or Cs/ Cw^-^sw, chemical uptake is linear or integrative. Thus, in the linear region, eq 2 can be reduced Cs(t) =CS0 + CwkJAoJVs)t For practical application, eq 3 can be rewritten Ms(t) =M0 + CwRst (3) (4) where Mo (kg) is the amount of analyte in the organic phase at s = 0. Rs (m3 s-1) is the sampling rate of the system (5) When fitting the eq 4 to experimental data, a negative intercept can be interpreted as a lag phase between initial deployment and penetration of analyte through the diffusion-limiting membrane. Sampling rate can be determined experimentally under fixed conditions at constant analyte concentration. Under environmental conditions, when the water concentration is changing during the exposure, the term Cw represents a TWA concentration during the deployment period. The TWA aqueous concentration can be then estimated from the amount of analyte accumulated in the sampler during the exposure (26) Scheuplein, R. J. J Theor. Biol 1968, 18, 72-89. (27) Flynn, G. L.; Yalkowsky, S. H. J. Pharm. ScL 1972, 61, 838-852. The chemical uptake into passive sampler remains linear and integrative approximately until the passive sampler concentration factor (ratio Cs(t)/ Cw) reaches half-saturation.9 When calibration data, i.e., Rs and Ksw, are available, the following equation can be used to estimate maximal exposure time in which the passive sampling system accumulates integratively under field conditions m2tfsWVg/i?g (7) where the term £50 is the first-order half-time of the uptake curve. Under these conditions the concentration of a chemical in the organic phase is directly proportional to the product of the concentration in the surrounding aqueous medium and the exposure time. EXPERIMENTAL SECTION Materials and Chemicals. Test chemicals (Table 1) included several groups of persistent organic pollutants: y-hexachlorocy-clohexane (lindane, y-HCH), hexachlorobenzene (HCB), 2,2'-bis-(4-chlorophenyl)-l,l'-dichloroethylene (DDE), PAHs, and PCBs. y-HCH reference material was obtained from Riedel-de Haen. HCB, DDE, and PAH reference materials were obtained from Dr. Ehrenstorfer. PCB reference material and test chemicals in high purity (>99<%s y-HCH, HCB, DDE, PAHs, and PCBs) were purchased from Promochem. Dialysis membrane Spectra/ Por 6 (molecular weight cutoff 1000) was obtained from Spectrum Laboratories. The Gerstel-Twister stir bar for sorptive extraction was obtained from Gerstel. Lichrolut (R) (diameter of particles 40—63 fim) was purchased from Merck. The solvents methanol and hexane were used in LiChrosolv quality from Merck. Sampler Design. The passive sampling device, referred to as the membrane-enclosed sorptive coating sampler (MESCO), consists in the actual investigation of a Twister bar used for SBSE (component 1, Figure 1) enclosed in a dialysis membrane bag made from regenerated cellulose (Spectra/ Por 6, molecular weight cutoff 1000, 18-mm flat width, 30-mm length; component 2, Figure 1). Twister is a stir bar (15 mm length) consisting of a magnetic core sealed inside a glass coated with 22 mg of PDMS. The PDMS sorptive layer (receiving phase) is 500 fim thick and its volume is 24 fiL. Prior to first use, the stir bar was placed into a vial containing 1 mL of a 1:1 mixture of methylene chloride and methanol, and the mixture treated for 5 min with sonication. Then the solvent mixture was rejected and the procedure repeated three times. The stir bar was dried in a desiccator at room temperature. Prior to each use, the stir bar was conditioned by heating for 180 min at 280 °C with a nitrogen stream of ~100 mL min-1. The dialysis membrane bag with Twister inside is filled with 3 mL of bidistilled water (component 3, Figure 1) and sealed at each end with 35-mm Spectra Por enclosures (component 4, Figure 1). The stir bar was allowed to freely move inside the membrane. As a relationship is likely to exist between the surface area and the rate of uptake, the area of the membrane was held constant at 1100 mm2. To enable a simultaneous exposure of a series of samplers, they were connected to a string, which was then exposed to organic analytes in a continuous-flow system. Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5193 Table 1. Selected Physicochemical Properties of Test Analytes at 25 °C no. compound MW log K„wb AGsiv/Y'CkJmol-1) AGs(o)' (kJmor1: 1 HCB 284.8 5.5 0.005 -6.9 -38.2 2 y-HCH 290.8 3.7 7.3 -9.5 -30.6 3 p,p'-DDE 318.0 5.7 0.04 -9.4 -41.8 4 PCB28 257.5 5.6 0.16 -8.7 -40.6 5 PCB52 292.0 6.1 0.03 -10.3 -45.0 6 PCB101 326.4 6.8 0.01 -10.1 -48.8 7 PCB138 360.9 7.6 0.0015 -11.0 -54.3 8 PCB153 360.9 7.8 0.001 -10.0 -54.4 9 PCB180 395.3 8.3 0.0003 -10.0 -57.3 10 acenaphthylene 152.2 4.0 16.1 -13.9 -36.7 11 acenaphthene 154.2 4.0 3.8 -8.6 -31.4 12 fluorene 166.2 4.2 1.9 -9.3 -33.2 13 anthracene 178.2 4.6 0.045 -10.9 -37.1 14 phenanthrene 178.2 4.5 1.10 -11.6 -37.2 15 fluoranthene 202.3 5.1 0.26 -17.1 -46.2 16 pyrene 202.3 5.1 0.132 -16.5 -45.6 17 be nzo [a] anthracene 228.3 5.9 0.011 -15.0 -48.6 18 chrysene 228.3 5.7 0.0019 -15.5 -48.0 19 be nzo [ b] fiuo ranthene 252.3 5.8 0.0015 -19.9 -53.0 20 be nzo [k] fluoranthene 252.3 6.0 0.0008 -19.8 -54.0 21 be nzo [a] pyrene 252.3 6.2 0.0038 -19.7 -55.0 22 indeno[ 1,2,3-ctflpyrene 276.3 6.8 0.0005 -24.5 -63.2 23 benzo[g/ři]perylene 276.3 6.9 0.0003 -24.5 -63.8 a Molecular weight. h Octanol—water partition coefficient.41'42 c Aqueous solubility.41,42 d Calculated free energy of aqueous solvation. e Calculated free energy of solvation in octanol. Laboratory Exposures. Batch Exposures. Twister bars designed for later use in flow-through exposures were individualized (by attributing a number to each bar) and the extraction efficiency and repeatability was tested in abatoh system, at first Conditioned Twister bars (without the membrane) were separately lowered to 20 mL of aqueous solution in a 25-mL closed amber glass vessel containing test solution of analytes. The test solution was prepared by spiking double-distilled water with a test substance mixture dissolved in methanol to give nominal concentration of individual analytes of 125 ng Lr1. The flask content was stirred at 1000 min-1 for 60 min at room temperature. After this, the Twister bars were removed from the sample, washed with a small amount of bidistilled water, and dried with a paper cloth. The accumulated analyte content was analyzed by GC/MS as described below. Detection of outliers was performed using the Mahalanobis distance technique (p = 0.05).28 The normal distribution of the errors for individual analytes in the sample set was tested by the Kolmogorov—Smirnov test (p = 0.05). Flow-Through Exposures. MESCO samplers were exposed to test chemicals at nominal concentration of 20 and 50 ng L-1 in a flow-through exposure system. Exposures were conducted at 14 and 19 °C. The experimental conditions of individual exposures are given in Table 2. The experimental setup of the flow-through exposure system has been described.29 Exposures were conducted at a linear flow velocity of 0.6 cm s-1. The exposures lasted between 4 and 7 days, during which the samplers were sampled at time intervals and their contents analyzed to determine accumulated concentrations of test chemicals as described below. Water samples from the exposure column (5 L) were taken at each time when samplers were sampled and analyte concentration in water was determined. (28) Egan, W. J.; Morgan, S. L. Anal. Chem. 1998, 70, 2372-2379. (29) Vraná, B.; Schüürmann, G., submitted to Environ. Sei. Technol Table 2. Summary of Passive Sampler Flow-Through Exposure Experimental Conditions nominal no. of expt concn temp exposure MESCO s no. (ng L-1) (°C) period (h) sampled 1 20° 19 0-166 16 2 20 14 0-165 12 3 50 19 0-96 6 a The nominal concentration of PCB180 was 40 ng L Sampler Processing. Following exposure, MESCOs were dismantled, Twister bars were washed with bidistilled water, dried with a paper cloth, checked visually for possible damage of the sorptive layer, and analyzed for accumulated analyte content (test substances only) by thermodesorption-GC/MS. Processing of Water Samples. The residues in the water samples were extracted using solid-phase extraction (SPE) using Lichrolut (R) sorbent The quantification of acenaphthylene, acenaphthene, fluorene, anthracene, phenanthrene, and HCB in water was carried out using SPME technique (Supelco 65-fim poly-(dimethylsiloxane)—divinylbenzene (PDMS—DVB) solid-phase microextraction fiber assembly) in combination with a gas chromatographic system. The detailed description of the procedures is given in the Supporting Information section. Instrumental Analysis. The quantitation and qualitative control of the compounds accumulated during exposures in Twister bars was performed by thermodesorption-GC/ MS. For thermodesorption, the Twister bar was positioned in the middle of the heated zone of a desorption tube (178-mm-length, 6-mm-o.d, 4-mm-i.d. glass tube) in a thermal desorption device. Thermodesorption-GC/ MS was performed on an Agilent Technologies (Palo Alto, CA) system equipped with aGerstel (Mulheim/ Ruhr, Germany) thermodesorption device TDS A. A cold injection 5194 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 system from Gerstel (CIS-4) with an empty liner was used for cryofocusing of the analytes prior to the transfer onto the analytical column. The cold injection system was cooled with liquid nitrogen to —150 °C during thermal desorption. The following conditions were chosen for the thermodesorption of the compounds from the stir bars: desorption temperature, 250 °C; helium flow rate, 100 mL min-1; desorption time, 10 min. The transfer line both from the thermodesorption device to the CIS and from the GC to the MSD ion source was set to 250 °C. After stir bar desorption, the CIS was heated to 250 °C with a rate of 12 °C s-1, the injector was used in the splitless mode with a splitless time of 1.5 min. A HP-5 MS column (30-m length, 0.25-^m i.d., 0.25-^m film thickness) was used with the following temperature program: 50 °C, 3 min isothermal, 15 °C min-1 to 160 °C, then at 3 °C min-1 to the final temperature of 280 °C, and held for 9 min. Helium was used as carrier gas at a linear velocity of 39 cm s-1. The single ion monitoring (SIM) mode applying one or two characteristic ions per compound was chosen for the detection. For the external calibration, a small bunch of glass wool was positioned to an empty desorption tube. The desorption tube was then connected to a cool injector of a GC and flushed with 20 mL min-1 nitrogen. The desorption tube with glass wool was then spiked with 2 fih of a calibration standard solution and flushed for 1 min by the nitrogen stream to allow the solvent (hexane) to evaporate. The desorption tube was then transferred to the thermodesorption device (TDS A) and processed by thermode-sorption-GC/ MS. Quantification of the residues sorbed on Twister bars was accomplished using a five-point external standard curve. Data Processing. The experimentally determined time courses of the amounts of individual test substances on the Twister sampler were fitted by linear regression analysis using eq 4. The adjustable parameters were the intercept (Mo) and the slope (Cw-Rs) of the linear uptake curve Ms = fit). Quality of the fit was characterized by the standard deviations of the optimized parameters, as well as the correlation coefficient adjusted for the degrees of freedom (r2 adjusted), the fit standard deviation (SD), and the Fisher test criterion on the accuracy of the model. The sampling rates of the device Rs for individual test compounds were calculated by dividing the slope of the linear uptake curve by the mean aqueous analyte concentration during the exposure. The required variances of Rs values were calculated from the coefficients of variation of the uptake slope parameters and of the concentrations in the aqueous phase which were obtained according to the law of error propagation. The free energies of solvation of the test substances in water AGs(w) were calculated using the quantum chemical continuum-solvation model SM2.30 For previous applications to calculate Henry's law constant from AGs(w), the reader is referred to the literature.31-33 The free energies of solvation of the test substances in octanol AGs(o), were calculated as follows. Under standard thermodynamic conditions, the equilibrium partitioning of a compound between the air phase (a) and the octanol phase (o) in terms of molar concentrations ca and c„ is governed by the solvation free energy AGs(o) (30) Cramer, C. J.; Truhlar, D. G. J. Comput.-AidedMol. Des. 1992, 6, 629-666. (31) Schuurmann, G. Environ. Toxicol Chem. 1995, 14, 2067-2076. (32) Schuurmann, G. J Comput. Chem. 2000, 21, 17-34. (33) Dearden, J. C; Schuurmann, G., submitted to Environ. Toxicol. Chem. AGs(o) = -BT In - = 2.3RT log(8) For the application of eq 8, the air—water partition coefficient Km is derived from the calculated free energy of aqueous solvation, log Km = AGs(w)/ 2.3RT (9) The multilinear regression analyses were performed with Origin 5.0 (Microcal Software, USA). RESULTS AND DISCUSSION Passive Sampler Performance. Batch Exposures. Normal distribution of the errors for individual analytes in the Twister samples was confirmed. The coefficient of variation of individual substances extracted from the solution by the 16 Twister bars incubated under the same conditions ranged from 6%(PCB 28) to 19%(PCB 180). Twisters checked for repeatability were used for construction of MESCO samplers exposed in flow-through studies. Flow-Through Exposures. The performance of the MESCO sampler was tested in continuous-flow exposures to constant concentrations of test chemicals. Concentrations of the analytes in water (Cw) and the amounts accumulated in the MESCO sampler (Cs) were two parameters measured regularly during the continuous-flow exposures. During exposure, the water concentration was held constant, which was confirmed by analyses of water samples. Characteristic uptake curves are shown in Figure 2. For all test substances, the uptake was linear in all exposure studies during the whole exposure period and without any sign of a leveling-offin the uptake curve. Satisfactory fits of kinetic eq 4 to the data from exposure were obtained for all test compounds. Correlation coefficient (r2 adjusted) values of the regression (model versus experimental) ranged from 0.74 to 0.97 with the exception of HCB in experiment 2 (r2 adjusted, 0.66). Coefficients of variation of the calculated slope did not exceed 29% in any case. A lag phase between approximately 0 and 46 h was observed for the test substances in experiment 1. In experiments 2 and 3, no significant (p > 0.05 in all tests) lag phases were detected for the test substances, except for PCB180 in experiment 2, for which a lag phase of 44 h was observed. The higher uncertainty in estimation of intercept values in these experiments results from lack of data in the initial uptake period (first sampling point available after 69 h). The average aqueous concentrations of individual analytes measured during exposures ranged from 50%to 130%of the nominal concentration. The maximum fluctuations of aqueous concentrations during exposure did not exceed 40% of the mean concentration for individual compounds. Concentration Proportionality of Response. The results of flow-through exposures indicate that the passive sampler responded proportionally to the range of test analyte concentrations (20—50 ng L-1, nominal). The independence of sampling rates Rs from aqueous solute concentrations was confirmed using an unpaired t-test (p = 0.05) for y-HCH, DDE, PCBs, and hydrophobic PAHs (log Kow > 5) (Figure 3). Time To Reach Steady State. The maximum exposure time in which the passive sampling system collects integratively Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5195 1.0-r 0.9- 0.8- CD 0.7- fwisl 0.6- CL 0.5- O) 0.4- 0.3- 0.2- 0.1- 0.0- a Acenaphthene ■ Fluoranthene • Benzo[3,ft,/]perylene Time (h) PCB28 a PCB101 • PCB180 0 20 40 60 80 100 120 140 160 180 Time (h) Figure 2. Uptake of selected PAHs and PCBs by the Twister-based MESCO sampler. The data used represent the 19 °C flow-through exposure (20 ng L~1). The lines are predicted concentrations in the sampler obtained by linear regression using eq 4. 1000 -i 900 - 800 700 P 600 I - 500 400 300 200 100 0 Nominal water concentration! T(ngL"1) B20 □ 50 <5? 800 700 -600 500 400 300 -200 100 0 Nominal water concentration (ng L'1) W20 D50 Kl Hi Figure 3. Relationship between aqueous concentration and MESCO sampling rate (Rs). The data used represent flow-through exposures at 19 °C. The independence of sampling rates from aqueous concentration was confirmed for the shown compounds using an unpaired Mest (p = 0.05). under field conditions or time to reach 50%of the Ksw value was estimated using eq 7 and the Rs values from the flow-through exposure study conducted at 19 °C and 20 ng L-1 nominal concentration (experiment 1). Because of physical difficulties in determination of the Ksw values in batch experiments (depletive extraction of test substances by the Twister from 20 mL of a 100 ng L-1 aqueous solution), the apparent distribution constants K? (PDMS), obtained with glass fibers coated with 100-^m PDMS for the analyte's partitioning between PDMS coating and aqueous sample was used as a substitute for Ksw in the estimation.34-36 The results of the first-order halftime *bo calculation are reported in Table 3. It is calculated that, for y-HCH and acenaphthylene, a passive sampler may sample integratively less than one week. For the rest of the PAHs taken into the calculation, the passive sampler may remain in the linear uptake phase more than one week; for the HCB, DDE, and PCBs, the £50 may be several months. The linear uptake of all test analytes in all exposure studies during the whole exposure period indicates that this condition of integrative sampling is fulfilled for at least one-week exposures. The *bo estimation indicates the possibility to use sampling rate (34) Paschke, A., unpublished work, Leipzig, 2001. (35) Doong, R.; Chang, S. Anal. Chem. 2 0 0 0, 72, 3647-3652. (36) Valor, I.; Perez, M.; Cortada, C; Apraiz, D.; Molto, J. C; Font, G. J Sep. Sci. 2001,24, 39-48. data obtained under laboratory conditions for estimation of TWA concentrations of analytes from the contaminant amounts accumulated in MESCOs during environmental exposures of several weeks. In general, deviations from the linear uptake in prolonged exposures are expected for compounds with log Km < 4.0, with the assumption that Ksw correlates well with 2fow within the hydrophobicity range. For a more accurate estimate of £50 values, direct measuring of Ksw in a Twister—water batch or flow-through system is necessary. Sampling Rate. The sampling rates Rs obtained in flow-through exposure studies conducted at 19 and 14 °C and 20 ng L-1 nominal concentration (experiments 1 and 2, respectively) are shown in Table 4. Over the range of controlled laboratory conditions, the magnitude of Rs values differed by 15-fold (i.e., from 47 to 700 fih h-1). This range of sampling rates is narrow relative to the broad Kovl range of almost 5 orders of magnitude (log Km ranged from 3.7 to 8.27). Using the average sampling rates for each chemical, a single MESCO deployed in water over 20 days would clear a total of 60—300 mL of water of the individual chemicals. This is a low volume when compared with clearance volumes of other common passive samplers, such as the triolein-filled SPMDs with standard configuration,9 which would clear 20—160 L of water in 20 days. Despite the fact that the extraction efficiency of MESCO is 3 orders of magnitude lower than that of SPMD, the method 5196 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 Table 3. Estimation of the Maximal Exposure Time (50 in Which MESCO Samples Integratively under Field Conditions at 1 9 °Ca compound log i«PDMS) «50 (d) HCB 4.36 119 y-HCH 3.2C 3 p,p'-DDE 5.26 344 PCB28 i.lb 69 PCB52 5.06 190 PCB101 5.36 655 PCB138 5.46 734 PCB153 5.46 910 PCB180 5.26 1020 acenaphthylene 3.40d 4 acenaphthene 3.63d 11 fluorene 3.71d 9 anthracene 3.98d 14 phenanthrene 3.96d 20 fluoranthene 4.71d 92 pyrene 4.86d 99 b e nz 0 [ a ] anthrac e ne 5.26d 211 chrysene 5.69d 530 b e nz o [ 6] fluo ranthe ne 5.17d 227 b e nz o [ & ] fluo ranthe ne 5.33d 299 benzo [a] pyrene 5.39d 439 indenof l,2,3-cd\ pyrene 4.28d 45 benzo [g/ři]perylene 4.43d 78 a Sampling rates taken for calculation were determined at nominal test substance concentrations of 20 ng L_1 at 19 °C. b The 100-//m PDMS fibers were exposed in 500 mL of stirred standard solution over a time sufficient to reach equilibrium distribution between the aqueous solution and fiber coating.34 c Data from ref 36. d Data from ref 35. Table 4. Summary of Passive Sampler Sampling Rates Rs Derived from Flow-Through Exposures at Different Temperatures at Nominal Analyte Concentration of 20 ng L1 T= 19 °C T = 14 °C Rs cv Rs C compound (fiL h-1) (% (fiL h-1 HCB 114 7 47 50 y-HCH 336 41 188 47 p,p-DDE 305 7 142 28 PCB28 337 49 497 57 PCB52 275 32 397 40 PCB101 226 13 266 28 PCB138 227 6 271 29 PCB153 188 7 229 30 PCB180 110 8 113 33 acenaphthylene 484 7 700 16 acenaphthene 280 8 238 14 fluorene 391 7 485 16 anthracene 462 15 543 21 phenanthrene 321 10 255 17 fluoranthene 389 11 217 31 pyrene 509 15 270 30 benzo[a]anthracene 597 4 212 33 chrysene 641 8 215 32 b e nz o [ 6] fluo ranthe ne 453 5 234 26 b e nz o [ & ] fluo ranthe ne 495 8 214 28 benzo [a] pyrene 388 7 301 18 inde no [1,2,3-cíí] pyrene 294 5 a benzo [g/ři] perylene 239 9 a a Indeno[l,2,3-cd]pyrene and benzo [g/ii]perylene were not determined during the experiment conducted at 14 °C. sensitivity of these two techniques is comparable. This is because the total amount of analyte sequestered by MESCO during deployment can be transferred to the GC system, whereas only a small portion of the SPMD extract is usually injected to the GC (to prevent introduction of large amounts of interfering contamination to the chromatographic system). The advantage of low clearance volume (i.e., Rst) of MESCO during exposure in comparison with other types of passive samplers (e.g., SPMDs) is the nondepletive extraction, which enables use of flow-through exposure calibration data also for TWA concentration estimation at sampling sites with very low exchange volumes of water in the vicinity of the sampler during an exposure (e.g., in wells with very low groundwater flux).37 The comparability of experimentally derived MESCO calibration data to actual values during field sampling generally depends on the similarity of laboratory and field exposure conditions. Besides temperature and biofouling, mainly flow velocity/ turbulence may affect the uptake kinetics. An increase in uptake rate can occur with increasing water flow velocity or turbulence as reported for passive sampling devices fitted with polyethylene membranes.18 29 38 On the other hand, Kingston et al.18 observed only minor effects of turbulence on the accumulation kinetics in a passive sampler fitted with a hydrophilic polysulfone membrane. Nevertheless, examination of potential rate-limiting barriers to analyte uptake by MESCOs is necessary. It is assumed that the (37) Gustavson, K. E.; Harkin, J. M. Environ. Sci. Technol 2 0 0 0 , 34, 4445-4451. (38) Booij, K.; Sleiderink, H. M.; Smedes, F. Environ. Toxicol. Chem. 1998, 17, 1236-1245. overall resistance (1/ km), to the uptake of a chemical is given by the sum of particular barrier resistances 1 „ di dM dw ds — = £-=-+-+- (10) ^ov i Kiw^i DMKMW Dw DgKsw where (5; is the particular barrier thickness, D; is the diffusion coefficient in the barrier, and 2f;w is the partition coefficient between the tth phase and water (designed as subscripts for the water (W), dialytic membrane (M), and receiving organic phase (S). The overall mass-transfer coefficient is expected to be affected mainly by the diffusion of solutes in individual phases (water, membrane pores, and the PDMS, respectively) and by their partitioning into the PDMS, since no accumulation of hydrophobic analytes is expected in the hydrophilic dialytic membrane (i.e., Xmw ~ 1)- As can be seen from eq 10, a resistance decrease in receiving organic phase is expected with increasing Ksw value for substances having a similar diffusion coefficient in the organic phase Ds- To obtain more information on the processes involved in the contaminant uptake, clearance (elimination) rate constants (ke) from the sampler into water are required for the test chemicals. In this study, we were able to make an estimation from the sampling rate and the PDMS/water partition coefficient (Kf (PDMS)) value only: e ^sw^s~^f(PDMS)ys Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5197 log Kf(PDMS) Figure 4. Logarithm of the clearance rate constant log ke (h~1) estimated using eq 11 versus the logarithm of PDMS/water partition coefficient log Ki(PDMS). The combination of eqs 10 and 11 allows recognition of the dominant barriers to mass transfer. When the diffusive transport is limited by the resistance in the PDMS and the resistance in water and dialytic membrane is negligible (i.e., if (5m/Dm-^mw + f5w/ flw 5.7) PAHs, too, with the exception of benzo[£]fluoranthene. For the rest of the tested substances, the dependence is less clear. For more insight into the connection between the sampling rate and the delay time, more detailed kinetic studies conducted in the initial uptake phase are needed. Effect of Temperature. The relationship between sampling rates of the test analytes and temperature was compared at two temperatures (14 and 19 °C, Table 4). The ratios derived by dividing analyte Rs values determined at 19 °C by those determined at 14 °C ranged from 0.7 to 3.0. No significant differences (unpaired í-test; p = 0.05) in sampling rates were observed between 14 and 19 °C treatments for PCBs and for y-HCH. Among PAHs, a significant decrease in sampling rates with decreasing temperature was observed for benzo[a]anthracene, chrysene, benzo[Ď]fluoranthene, and benzo[£]fluoranthene. Also, the sampling rate of HCB and p,p-DDE decreased significantly with decreasing temperature. The effector temperature on the sampling rate is not easy to model because of the complexity of the system. Both thermodynamic and kinetic parameters affecting the sampling rate are temperature dependent. For practical purpose, it is therefore necessary to determine the effects of temperature in the laboratory for each analyte of interest and to measure the environmental temperature during field deployment. Method Sensitivity and Selectivity. MESCO has the potential to detect low TWA water concentrations (ng to pg/L) for two reasons: (1) A substantial enrichment factor is built into MESCO sampling, because dissolved aqueous concentrations are concentrated up to the factor Ksw into a Twister. (2) The entire analyte amount on the Twister is introduced to the GC and directed to the detector. To estimate minimum quantifiable TWA aqueous concentrations, limits of quantitation in MESCO samplers Ms(loq) were substituted into eq 6. The calculated concentration quantiation limits depend on the sampling rate Rs, and the method sensitivity increases with increasing exposure period of the samplers. When taking a sampler exposure of 20 days for the calculation, estimated quantitation limits range from 4 pg L-1 for PCB28 to 140 pg L-1 for benzo[g/ii]perylene, respectively. Actual quantitation limits can be affected, e.g., by interfering substances or bleeding from the PDMS coating during thermodesorption. The MESCO sampling approach aims at measuring trace concentrations in water that will always contain interfering substances. The selectivity of the MESCO extraction technique is enhanced in two ways: (1) The dissolved molecules become separated from colloids during their diffusion across the dialysis membrane. (2) Hydrophobic target analytes are selectively extracted from the inner aqueous solution by the PDMS sorbent coating. CONCLUSIONS The MESCO sampling system combines the passive sampling approach with solventless preconcentration of organic solutes from aqueous matrixes and subsequent desorption of the sequestered analytes on-line with a capillary GC/ MS system. This combination presents a low-cost and robust alternative to the currently used passive sampling techniques. Moreover, the hydrophilic cellulose dialysis membrane is permeable for both nonpolar and polar organic species, whereas other passive sampling devices such as SPMDs allow for accumulation of nonpolar substances only. The user of MESCO can easily check the repeatability of the stir bars used for the preparation of the samplers. The Twister stir bar can be reused after each field deployment when no degradation or damage of the membrane occurs during exposure. The samplers are miniature and do not require use of large deployment devices in the field, which enables a nonconspicuous deployment Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5199 at sampling sites during monitoring campaigns. Instead of PDMS-coated stir bars, glass fibers coated with PDMS esc. may be used for construction of passive samplers. The advantage of SPME fibers is that accumulated analytes can be analyzed using conventional gas chromatographs without the need of a ther-modesorption unit and a cold injection system. However, the volume, and thus also the accumulation capacity of the stir bars, is between 1 and 2 orders of magnitude higher than that of SPME fibers, which makes a sampler with SPME fiber less sensitive. The performance of the MESCO sampler for integrative sampling of hydrophobic persistent organic pollutants has been demonstrated. The issues, which have to be addressed for further validation of MESCO, include testing (1) the stability of the dialysis membrane during in situ deployment and prevention of its possible degradation, (2) the effect of water turbulence on the uptake kinetics of analytes, (3) the effect of biofouling on the uptake (40) Popp, P.; Bauer, C; Wennrich, L. Anal Chim. Acta 2001, 436, 1-9. (41) Mackay, D.; Shiu, W. Y., Eds. Illustrated handbook of physical-chemical properties of environmental fate of organic chemicals; Lewis Publishers: Boca Raton, FL, 1992; Vol. 1. (42) Mackay, D.; Shiu, W. Y.; Ma, K. C, Eds. Illustrated handbook of physical-chemical properties of environmental fate of organic chemicals; Lewis Publishers: Boca Raton, FL, 1992; Vol. 2. kinetics, (4) the uptake capacity of Twister bars for individual analytes and determination of iisw, (5) the dissipation kinetics of individual analytes from MESCO at varying conditions, and (6) the applicability of the sampler for monitoring polar analytes. As an alternative to thermodesorption, reextraction of analytes from Twister bars by small volumes of organic solvents could be used.40 The extracts could be then subjected to analysis by HPLC or examined by bioassays. ACKNOWLEDGMENT The authors thank Uwe Schröter, Petra Keil, Heidrun Paschke, and Petra Fiedler for sample preparation and instrumental measurements. SUPPORTING INFORMATION AVAILABLE Description of the flow-through exposure system, description of processing and analysis of water samples, and results of the fits of kinetic eq 4 to the data from flow-through exposure experiments. This material is available free of charge via the Internet at http:/ / pubs.acs.org. Received for review June 6, 2001. Accepted July 30, 2001. AC010630Z 5200 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 Príloha 4 Vrana B. and Schuurmann G., Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics., Environ. Sci. Technol., 2002, 36, 290-296. Environ. Sci. Technol. 2002, 36, 290-296 Calibrating the Uptake Kinetics of Semipermeable Membrane Devices in Water: Impact of Hydrodynamics B RAN I SLAV VRAN A* AND GERRIT SCHÜÜRMANN Department of Chemical Ecotoxicology, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany The use of lipid-containing semipermeable membrane devices (SPM Ds) is becoming commonplace, butthe potential effects of environmental variables affecting the accumulation of contaminants into SPMDs had not been characterized sufficiently, yet. To characterize the effect of hydrodynamic conditions on the contaminant uptake kinetics, accumulation of pentachlorobenzene, hexachlorobenzene, and hexachlo-rocyclohexane isomers from water into SPMD was studied at various water flow rates. The accumulation kinetics of hydrophobic compounds (log Km > 4) are governed by the aqueous boundary layer in linear flow velocity range from 0.06 to 0.28 cm s~1 and sensitive to slight changes in flow rate. The effect of flow velocity on the exchange kinetics increases with increasing hydrophobicity. Under faster, but still laminar flow conditions (0.28-1.14 cm sH), the sensitivity to changes in flow decreases to a nonsignificant level for the substances under consideration. The results of this study confirm that the use of the laboratory-derived calibration data for estimation of analyte concentrations in the ambient environment is limited unless flow-sensitive performance reference compounds are used. Introduction Passive monitors are rapidly gaining wide acceptance for assessing integrated, or time-weighted, concentrations of organic chemicals in aquatic systems. One category ofpassive sampler, the lipid-containing semipermeable membrane devices (SPMD), introduced by Huckins et al. (7)has received a great deal of attention. The SPMD sampler consists of lay-flat polyethylene tubing containing a thin film of triolein, a high molecular weight neutral lipid. The polyethylene used in SPMD is commonly referred to as nonporous,even though transient cavities with diameters approaching ~10 A are formed by random thermal motions of polymer chains (2). The thermally mediated transport corridors of the polyethylene exclude larger molecules, as well as those that are adsorbed on sediments or humic acids. Only truly dissolved (but generally nonionized) contaminants are sequestered. The process mimics the transfer of organic contaminants through biomembranes. The utility of the SPMD has been shown for monitoring aqueous residues of polychlorinated biphenyls, various organochlorine pesticides, polychlorinated dibenzodioxins, polychlorinated dibenzofurans, and poly-cyclic aromatic compounds. * Corresponding author phone: ++49 341 235 26 18; fax: ++49 341 235 2401; e-mailbv@uoe.ufz.de. Current research results show that the SPMD can be used to estimate time-weighted average (TWA) concentrations of organic contaminants in aquatic environments. The theory and several mathematical models at different levels of complexity required to compute TWA ambient concentrations of analytes from SPMD concentrations have been described earlier (2—7). The uptake rates of contaminants into SPMD are affected by several factors including the sampler design, molecular properties, and environmental conditions. The environmental factors include temperature, biofouling impedance, and water velocity/turbulence. For correct estimation of ambient chemical concentrations from the field SPMD data, and for development of adequate calibration methods, it is necessary to sufficiently characterize the potential effects of environmental variables, in particular the impact of hydrodynamics on the uptake kinetics. Booij et al. (6) studied the effects of changes in flow turbulence on the exchange kinetics of organochlorine compounds with a wide range of K0„ values (4 < log K0„ < 8) in diluted sediment suspensions. He showed that the average exchange rate of chemicals between SPMD and water decreased by a factor of 4 under conditions of low turbulence. Huckins et al. (8) found a 1.5-fold increase in exchange rates with a 50-fold increase in velocity (range of 0.004—0.2 cm s~'). Most calibration studies have been conducted under low flow conditions; therefore, there is a need for characterization of the sensitivity of the calibration procedure to slight changes in laminar flow rate. To characterize the effect of hydrodynamic conditions on the contaminant uptake from water into SPMD, we examined the effect of various low linear flow velocities (flow rates) on the uptake kinetics of several organochlorine compounds with K0„ values ranging from 3.8 to 5.5. Modeling. To describe the uptake of contaminants from water to SPMD exactly, it would be necessary to use the Fick's second law for each compartment of the system, that is, for lipid, the SPMD membrane and water near the surface of the device, respectively (9). The inhomogeneity of each phase, which manifests itselfby the presence ofthe diffusion layers, and the different solvations of the substances in different phases ought to be also taken into account. The resulting description would be most probably too complicated for a direct comparison with our experimental data. Its simplification can be based on the plausible assumption ofquickdiffusion within the bulks ofthe compartments with regard to the duration of experiments, which can be justified using the solution of Fick's second law for the one-dimensionaldiffusion into a plane sheet of isotropic medium (9). This greatly simplifies the description ofthe transport, and the second-order partial differential equations based on Fick's second law can be replaced by a set of linear differential equations ofthe first order. The least complex approach for modeling the uptake of chemicals from water to SPMD given by Huckins et al. (5) is based on the description of the SPMD as a single compartment. The SPMD membrane is expressed as a lipid equivalent volume, and the SPMD sampler can be treated as a single compartment K*, (D where K values are partition coefficients among SPMD components [designated as subscripts for the whole SPMD, SPMD lipid (L), and the SPMD membrane (m) and water (w)] and V is the volume of a phase designated by subscripts. 290 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 2, 2002 10.1021 /es0100625 CCC: $22.00 © 2002 Am erican Chem ical Society Published on Web 12/08/2001 When assuming a constant concentration in water, the concentration for the whole device uptake is given by Qpmd — CwKSPMD(l exp[ kj]) (2) Here, C values are analyte concentrations and ke is the elimination rate constant, which is also called the overall exchange coefficient. The chemical uptake into SPMD remains linear and integrative in the initial period of the exposure until the concentration factor (CF, ratio Cspmd/Cw) in the SPMD reaches about half-saturation (ket < In 2) (7) and then eq 2 can be reduced to CsPMD ''w^SPMD^c' ' C,„RJ/V« (3) where Rs is the apparent SPMD samplingrate. The elimination rate constant ke can be broken down to several fundamental parameters 1 (4) where kw is the mass transfer coefficient in the aqueous boundary layer and &spmd is the apparent mass transfer coefficient for transport in the SPMD sampler from the surface of the SPMD to the location of chemical storage in the SPMD. The term in parentheses is the overall resistance to the uptake of a chemical. Analogously to the theory for diffusion through two films in series (JO), overall resistance is given by the sum of independent and additive resistances to mass transfer for the stagnant aqueous film at the surface of the SPMD (l//fc„) and for the SPMD (1/£Spmd£spmd). These resistances can be expressed as SPMD-specific residence times tw and Tspmd, which combine with ifspMD to model the elimination rate constant ke (or overall residence time x): 1 ^spmd ^"sPMD A 1 = r (5) The group VsrMo/Akw is thus the aqueous boundary layer residence time tw, and VspmdM^spmd is the SPMD residence time Tspmd- In extreme cases, the uptake rate for the whole device is controlled either by the SPMD (or, more specifically, by the polymeric membrane of the SPMD) or by the aqueous boundary layer, depending on the analyte properties and exposure conditions. Examination of eqs 4 and 5 indicates that a high partition coefficient between SPMD and water C^spmd) effectively reduces the resistance to mass transfer in the SPMD. Gale (7) predicted an SPMD control for the accumulation of large molecules with low polymer diffusivity or for accumulation at lower temperatures. On the other hand, aqueous film diffusion may control the accumulation of highly polymer-diffusive molecules and highly hydrophobic substances with low resistance to mass transfer in SPMD due to a high ifspMD value. Similar uptake rate constants (i.e., fce x -Kspmd) of chemicals with widely different partition coefficients are an indicator of the limitation of mass transport to and from the SPMD by the resistance in the aqueous phase, whereas increasing uptake rate constants with increasing ^spmd show that the membrane resistance likely governs the mass transport. Huckins et al. (2, 5) have suggested adding performance reference compounds (PRC) to SPMD lipid prior to deployment. PRCs are analytically noninterfering compounds with a low to moderate hydrophobicity (up to log K0„ of ~5.0), that can be used for in situ calibration of the exchange rates. This approach is based on the assumption that uptake rates of chemicals can be derived from measurements of loss rates of PRCs (2, 6). Uptake and release are considered to be isotropic processes. The release of a PRC from the SPMD, when the concentration ofthis compound in the environment is negligibly low, can be described by a first-order-decay kinetic equation Qpmd — Co exP( kj) (6) where Co is the concentration of PRC in SPMD at the beginning of exposure. Experimental Section Materials and Chemicals. The solvents acetone, dichlo-romethane, hexane, and 2-propanol of LiChrosolv quality were obtained from Merck. Hexachlorobenzene (HCBz), pentachlorobenzene (PeCBz),hexachlorocyclohexane (HCH), [13C6]-a-HCH, [13C6]PeCBz, and [13C6]HCBz standards were obtained from Supelco. [2Hi0]Anthracene (Dio-ANT) (98% pure) was obtained from DeuChem, Leipzig, Germany. SPMDs with standard configuration, designed by Huckins et al. at the U.S. Geological Survey in Columbia, MO, consisting of a thin film of 1 mLof triolein (95% pure) sealed in a low-density polyethylene lay-flat tube (2.54 x 91.4 cm, 75 — 90 fim wall thickness), were purchased from ExposMeter AB, Tavelsjo, Sweden. They were stored in originalgastight metal paint cans until just before exposure. Laboratory Exposures. Static Exposures. Batch exposures were conducted in amber glass flasks containing 1 L of double-distilled water and one SPMD each. Immediately before exposure, SPMD lipid was spiked with 1 fig of each HCH isomer or with PeCBz, HCBz, and D10-ANT. For spiking a small cut was made at one end of the SPMD, and 50 fiLof hexane solution of test chemicals in hexane (0.02 fig fiL~x) was injected into the membrane using an HPLC syringe (volume = 100 fiL). The punctured SPMD was heat-sealed again, and the spiked solution was homogenized with the triolein by squeezing the SPMD content several times from one end to the other using latex gloves. For the HCH exposure study two replicate SPMDs were sampled on days 0,1,5, 14, and 22 of the exposure. For PeCBz, HCBz, and D10-ANT, SPMDs were sampled on days 0, 35, and 47 of the exposure. Water samples (1L) were taken at each SPMD samp ling time. The triolein and the polyethylene membrane of each SPMD were analyzed separately as described below. Flow-through Exposures. SPMDs were exposed to test chemicals at a nominal concentration of 50 ng L_1 and to control water in a flow-through exposure system. Exposures were conducted at 19 °C. The experimental setup consisted of a 1 m high glass column with either 15 or 7.5 cm inner diameter with a sieve-like perforated bottom (openings of 0.5 cm diameter). The column was covered with dark foil to prevent photodegradation of analytes during exposure. The exposure water was pumped from the bottom to the top of the column. Test chemicals were dissolved in methanol, and the appropriate amounts of stock solution were delivered into exposure water in a 1 Lchamber positioned at the bottom of the column using a peristaltic pump (Minipuls 3, Gilson). The water in the chamber was mixed using a magnetic stirrer with the turning speed of 600 min-1. The methanol concentration in the exposure water was held constant at 0.01% (v/v) in all exposure studies. This concentration was not expected to significantly affect the partitioning of test chemicals between SPMD and water. Tap water was fed to the chamber using a membrane pump (Prominent) at 36— 180 Lh~'. This setup enabled the flow rate in the exposure column to be varied. Exposures were conducted at flow rates of 0.06, 0.18, and 1.14 cm s_1, respectively. Before exposure, SPMDs were spiked with 25 fiLof a hexane stock solution of Dio-ANT using an HPLC syringe (volume = 25 fiL) to give a final concentration of 10 fig per SPMD. SPMDs were fixed in the column in a vertical position using Teflon rings at the top VOL. 36, NO. 2, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY ■ 291 TABLE 1. Summary of SPMD Flow-Through Exposure Experimental Conditions expt 0 1 2 3 nominal concn of chemicals (ng L-1) 120 h) for calculation ofpartition coefficients oftest substancesbetween TABLE 2. Summary of SPMD Partition Coefficients Derived from Static Exposures log Kui log KLv. CV CV log compound log Ko„a (lit.)' (exptl) (%) KmL (%) KsPMDC a-HCH 3.80 3.82 25 0.052 11 3.31 ß-HCH 3.80 4.03 13 0.020 24 3.26 y-HCH 3.70 3.66 16 0.057 13 3.22 (5-HCH 4.10 3.91 14 0.015 15 3.56 PeCBz 5.20 5.27 5.38d 0.095c 41 4.73" HCBz 5.50 5.50 5.57d 0.122c 30 4.99" D-io-ANT 4.54 4.58d 0.167c 32 4.04' a Preferred or selected values from refs 76and 17. 6 Values from ref 18. 0 n- 4. "Method detection limits were taken as a substitute for the equilibrium aqueous concentration. " Literature value of KLw was taken for calculation. 'Value interpolated from linear dependence of log KSPMd on log Kow for the test substances. SPMD compartments and water. The Kl„ values estimated for HCH isomers from the static exposures are comparable with their octanol—water partition coefficients, which is in a good agreement with the study of Chiou (18), who observed almost equality of the partition coefficients for selected nondissociating organochlorine compounds with logifow < 5. For PeCBz, HCBz, and Dio-ANTthe concentration in water did not exceed the limits of detection (2, 2, and 6 ng L_1, respectively). Therefore, only Km^ values were obtained for these substances from static exposures, and the values directly measured by Chiou (18) were taken for calculation of ifspMD values of PeCBz and HCBz. ifspuro values were calculated using eq 1. The ifspuro value for Dio-ANT was interpolated from the linear dependence of log ifspuro on log K0„ for the test substances. An estimate ofifspMD values was also performed using the method detection limits in the aqueous phase as a substitute for the equilibrium aqueous concentration. ifspMD values of 4.7,4.9, and 3.9 were obtained for PeCBz, HCBz, and Dio-ANT, respectively. The difference in logifspMD values obtained by the two different approaches was not greater than 0.1 log unit. The mean values of determined partition coefficients are summarized in Table 2 together with related values from the literature. Flow-Through Exposures. The effect ofthe aqueous film resistance to mass transfer can be investigated when exposure studies are conducted under various hydrodynamic conditions. The exposures were conducted at flow rates for which a laminar character of the flow in the major part of the exposure column was observed. This was checked by observing the dissolution of KMnC>4 grains in water flowing through the exposure column. Concentrations ofthe analytes in water (Cw) and in the SPMD (Cspmd) were two parameters measured regularly during the continuous flow exposures. During the exposure the water concentration was held constant, which was confirmed by analyses of water samples. All SPMDs exposed to control water without addition of analytes, and also SPMD blank samples, contained less than method detection limits (MDL) of test substances. MDL values ranged from 8 ng/SPMD for a-HCH to 23 ng/SPMD for Dio-ANT. Average water concentrations oftest substances in exposure water ranged from 27 to 62 ng L_1. The independence of SPMD concentration factors (CFs) relative to aqueous solute concentrations was demonstrated previously (1—3, 5). Therefore, CFs were used to express the data. The variance of calculated concentration factor values up to 16% was estimated from the coefficients of variation (CV) of the concentration in the SPMD (12%) and ofthe concentrations in the aqueous phase (11%), according to the law of error propagation. Characteristic uptake curves are shown in Figure 1. The uptake was modeled using linear and nonlinear (Levenberg—Marquardt algorithm)regression analysis. Curve 292 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 2, 2002 8 60000 0 100 200 300 400 500 600 700 800 Time (h) FIGURE 1. Uptake of a-hexachlorocyclohexane (top) and hexachlo-robenzene (bottom) by SPM D exposed under conditions of different linear flow velocities: 0.06 cm s 1 (circles, dotted lines), 0.28 cm s_1 (triangles, dashed lines), and 1.14 cm s_1 (squares, solid lines). The flow-through exposures were conducted at 19 °C at nominal chemical concentrations of 50 ng L 1. The lines correspond to eqs 2 and 3 with the optimized values of the parameters given in Table 3. fitting was performed with Origin 5.0 (Microcal Software). The experimentally determined time courses of the CFs of individual HCHs in the SPMD were fitted by nonlinear regression analysis using eq 2 with ifspuro and ke as adjustable parameters. In the case when the CF of the compound did not reach half of the ifspuro value (for PeCBz and HCBz), eq 3 with ke as the only adjustable parameter was used for linear regression analysis of the uptake curves. The ifspuro values needed for these calculations were taken from Table 2. Note that estimated values of ifspMD were used for PeCBz, HCBz, and Dio-ANT, respectively. A satisfactory fit of kinetic eq 2 to the experimental data was obtained forallHCH isomers in allexposure experiments. Relatively accurate values of the parameters ifspMD (CVs not exceeding 14% of the estimate) and ke (CV < 39%, except of one case for /3-HCH, 49%) were obtained. The higher uncertainty in the estimation of ke values results from lack of data in the initial linear uptake period. A variation between 9 and 40% was observed in ifspMD values for individual HCHs among experiments conducted at different flow rates. A difference of up to a maximum of 0.5 log unit was observed between log ifspMD estimates from flow-through exposure data and values from static exposures for individual HCHs. With regard to PeCBz and HCBz, a good fit of eq 3 to the experimental data was obtained, too. CVs of the calculated ke did not exceed 5%. The release of Dio-ANT was modeled using eq 6 with Co and ke as adjustable parameters. Figure 2 shows the release kinetics of this compound under different flow conditions. Estimates of Co ranged between 89 and 109%ofthe calculated value. CVs of the ke for this compound varied between 12 and 17%. Effect of Flow on the Exchange Kinetics. To examine the effect of flow velocity on the mass transfer, best fit values of ke obtained for individual compounds under various flow Q_ 03 Time (h) FIGURE 2. Release of [2H10]anthracene from SPMDs exposed at different linear flow velocities: 0.06 cm s 1 (circles, dotted lines), 0.28cm s_1 (triangles, dashed lines), and 1.14cm s_1 (squares, solid lines). The flow-through exposures were conducted at 19 °C. The li nes correspond to eq 7 w ith the opti mized val ues of the parameters given in Table 3. conditions (0.06,0.28, and 1.14 cm s_1) were compared using an unpaired t test (p = 0.05). For HCHs, no significant difference was observed between ke values determined at different flow rates. The only exceptions were the ke values for /3-HCH and <5-HCH at flow rates of 0.28 and 1.14 cm s"1. However, in these cases the difference could not be attributed to the change of the aqueous film thickness, because the kc values determined at higher flow rate (1.14 cm s_1) were in both cases smaller than at the lower one (0.28 cm s_1). With regard to more hydrophobic chemicals, a significant difference was observed between ke values determined at the lowest flow rate (0.06 cm s_1) and at both higher flow velocities. At a flow rate of 0.28 cm s_1, ke for PeCBz is higher than that at 0.06 cm s_1 by a factor of 3 and for HCBz by a factor of 9. The further increase in flow rate from 0.28 to 1.14 cm s_1 did not cause any further significant increase in kc values. The release kinetics of Dio-ANT was used as an independent measure of the exchange kinetics between SPMD and water. It was significantly affected by the flow, too. A 5-fold increase in flow rate from 0.06 to 0.28 cm s_1 results in a 3-fold increase in ke. Only a slight, but insignificant, decrease in ke value was observed with the further 4-fold increase in flow velocity to 1.14 cm s_1. No significant effect of flow conditions on the uptake of moderately hydrophobic HCH, and, on the other hand, a strong effect increasing with the K0„ correspond well with the theory of diffusion through two films in series (10), which assumes a switch in the overallmass transferto the aqueous phase control for very hydrophobic compounds (eq 5). Mechanism of Accumulation. To obtain a more detailed insight into the mechanism ofthe accumulation process, we tried to characterize the contribution of aqueous and polymer film resistance to the overallmass transfer. For this purpose, ke values were fitted to eq 5. ifspMD values needed for the analysis were taken from Table 2. Direct linear regression of the data is not desirable because IIke and ifspMD vary over 2 orders ofmagnitude; thus, the regression is weighted heavily in favor of the larger values. Therefore, we preferred to perform the regression on the logarithmic quantities, assuming a log-normal distribution of the errors. In addition, the assumption was made that &spmd values are the same for all compounds and for all three exposure experiments and that k„ values are the same for all compounds within each single experiment, respectively. This assumption could be made because test compounds are nonpolar and they have VOL. 36, NO. 2, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY ■ 293 TABLE 3. Summary of SPMD Exchange Coefficients and Partition Coefficients Derived from Flow-through Exposures linear flow velocity 0.06 cms1 0.28 cms1 1.14 cm sH ke x 103 CV CV ke x 103 CV CV ke x 103 CV CV compound (h-1) (%) log /(spMD (%) (h-1) (%) log /(spMD (%) (h-1) (%) log Kstmn (%) a-HCH 10.56 28 3.29 6 10.98 33 3.37 13 8.67 18 3.32 5 ,3-HCH 18.09 49 2.91 9 9.99 18 2.87 8 4.65 32 3.17 14 y-HCH 14.07 39 3.17 8 13.07 20 3.22 7 7.07 21 3.33 7 r3-HCH 7.94 35 3.01 10 9.66 19 3.09 8 4.71 24 3.34 10 PeCBz 0.57 5 _a 1.77 5 _a 1.40 3 _a HCBz 0.13 4 —a 1.25 5 —a 1.40 4 —a D-io-ANT 1.87* 17 5.66* 15 5.75* 12 a Not used as adjustable parameter because equilibrium was not approached during the exposure. 6 Rvalue determined from the dissipation rate. TABLE 4. Values of Mass Transfer Coefficients for the SPMD (^spiud) and the Aqueous Boundary Layer (/(„) Obtained as Optimized Parameters of the Nonlinear Regression Analysis of the Elimination Rate Constant (ke) as Dependent on the SPMD-Water Partition Coefficient (Kspmd) Using Equation 7a flow velocity (cm s~1) 0.06 0.28 1.14 log *spmd (m s~1) -9.44 ± 0.09 -9.44 ± 0.09 -9.44 ± 0.09 tspmd (h) 79 79 79 log *„ (m s~1) (s) -6.03 ±0.12 112 -5.44 ±0.15 29 -5.57 ±0.14 38 (L dayH 3.6 14.0 10.4 a Statistical indices of the fit are the number of data points n = 21. the correlation coefficient r = 0.94, and the standard deviation of the fit s = 0.70. The aqueous boundary layer residence time rw and the SPMD residence time rspmd were calculated as VSpMD/<4ft„ and VSpmd/ Akspuo, respectively. The apparent sampling rate fis of compounds accumulated under aqueous boundary layer control was calculated as KA. similar molecular weights and sizes. Therefore, diffusion coefficients of the test substances and mass transfer coefficients determined at constant conditions in particular matrices (water or SPMD) are expected to be approximately the same. In this way, the nonlinear regression was simultaneously performed with all ke values obtained in all experiments using the rearranged eq 5 as a fitting function r = log(A/VSPMD)-log(£z, x 10A + 10 a) (7) where X = log ifspuro is the independent variable; z, are indicator variables taking the value zi = 1 for experimental data from the ith experiment, for the rest of the data, zi = 0; and Y = log ke is the dependent variable. Adjustable parameters are the mass transfer coefficient in the aqueous film for the ifh experiment A, = log £w„ and in the SPMD B = log fcspMD, respectively. The fit results are summarized in Table 4 and shown in Figure 3. Note that also ke values for Dio-ANT obtained from the kinetics of dissipation were included into the analysis. Membrane Resistance. According to the two-resistance theory, the less hydrophobic compounds (log ifspuro < 3.6); that is, HCHs seem to be accumulated under sampler (membrane and lipid) control. The resistance to mass transfer in the SPMD can be viewed as two particular resistances acting in series, one for transport in the polymeric membrane and the other for transport in the receiving lipid phase. It can be shown that these resistances are additive (8) ^SPMD-^SPMD FIGURE 3. Dependence of the elimination rate constant ke on the SPMD water partition coefficient Ksmo at different linear flow velocities: 0.06cm s_1 (circles, dotted lines), 0.28 cm s_1 (triangles, dashed lines), and 1.14 cm s_1 (squares, solid lines). The lines correspond to eq 7 with the values of adjustable parameters log frspMD and log frw given in Table 4. where constant diffusion coefficients Dm and Di,in the phase films of thickness 8m and <5Lare assumed for the membrane and the lipid, respectively. The last additive term in eq 8 can be neglected, because the resistance to diffusion in lipid is small in comparison with the resistance to diffusion in the membrane and in the aqueous boundary layer, respectively. From the estimated &spmd value of 3.7 x 10~10 m s_1 the corresponding polyethylene film diffusion coefficient Dm for the group of test substances was calculated using eq 8 after the introduction of Km„ = Km^K^„, taking 5m = 82.5 [im and neglecting the resistance to diffusion in lipid. The value of Dm, ranging from 6 x 10~H to 3 x 10~10 cm2 s~\ corresponds well with the value of 3 x 10~" cm2 s~' estimated for phenathrene by Huckins et al. (2) and the value of the order of 10~10 estimated for a series of chlorinated hydrocarbons by Booij et al. (6), respectively. In general, the diffusion coefficient in polymer is a substance-specific quantity, which is controlled by physicochemical properties of diffusant molecules. For nonpolar molecules steric effects could control diffusion in the polyethylene membrane. Therefore, &spmd derived in this study has to be considered as a rough estimate valid only for a group of compounds with properties similar to the compounds tested (small nonpolar molecules). The SPMD residence time Tspmd of 79 h is calculated, which indicates that SPMD sampling exceeding 2 days (i.e., In 2tspmd) cannot be considered as integrative for substances accumulated under membrane control. For exposures exceeding about four halftimes of 294 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 2, 2002 E. v (cm s") FIGURE 4. Dependence of the mass transfer coefficient In the aqueous boundary layer frw as a function of linear flow velocity v In the flow-through exposure system. the uptake curve [i.e., 4(ln 2tspmd)], aqueous concentrations can be estimated using the equilibrium partitioning approach, that is, Cw = Cspmd/^spmd- The actual concentration of a compound accumulated in SPMD will reflect concentration changes in the aqueous phase during exposure with a time delay of Tspmd- Aqueous Boundary Layer Resistance. For more hydrophobic compounds (log /?spmd >3.6), the transport kinetics are governed by the aqueous boundary layer. This is indicated by the decrease in ke values with increasing K0„ and also by the fact that kw values are a function of the flow rate. Aqueous film theory (13, 14) hypothesizes a liquid boundary layer of thickness <5W, which is postulated to be completely stagnant and nonconvected, so that a solute molecule crosses it by pure diffusion only. At steady state, the aqueous phase mass transfer coefficient is given by (9) where Dw is the diffusion coefficient in the aqueous phase. The film theory predicts an increase in k„ at faster flow rates because <5W decreases. The quantitative computations of the value of diffusional flux in a laminar fluid (13, 19) show that the mass transfer coefficient k„ should be a function of the characteristic fluid velocity v, in accordance with law v", for a great variety of geometrical shapes of streamlined bodies and for different types of surface. When the flow is across the surface of a plate in a fluid under forced convection, the exponent n is equal to 0.5. The dependence of k„ observed in our study seems to follow this trend (Figure 4). The mass transfer coefficient increases initially with the flow rate, but later the increase becomes less expressive (insignificant in our case). Unfortunately, even for the simple experimental setup used in our study, the hydrodynamics of the system is fairly complicated, that is, SPMDs affect the current profile, entrance effects occur at the bottom ofthe exposure column, etc. Therefore, a quantitative comparison ofthe experimentally determined dependence kw = f(v) with theoretical computations is precluded and a direct comparison ofthe data with other calibration studies (3, 6) is impossible, too. The same type of dependence, that is, k„(v) = Av03, is expected to be valid also for uptake data based on SPMD placement at right angles to a very slow flow (3, 5), provided that the streaming was laminar. Almost equal values of aqueous diffusion coefficients D„ were estimated for the tested group of compounds ranging from 6 x 10~6to 7 x 10~6cm2 s_1 (15). From eq 9, the estimated boundary aqueous film thickness decreases approximately from 780 to 110/im with increasing flow rate. The magnitude of the boundary layer thickness corresponds with that of ~400 fim estimated by Gale (7) from uptake data obtained in a relatively quiescent dilutor system (i.e., flow < 1 cm s~'). In environmental systems the effective thickness of the aqueous boundary layer can vary from ~10 fim (extremely turbulent/high flow conditions) to > 1000 fim (deep stratified lakes of deep seas) (20). In practice, the variation of flow at the surface of in situ exposed SPMDs can be reduced by the use of appropriate SPMD deployment devices. The advantage ofthe aqueous boundary layer control in comparison with the membrane layer control is that the transport kinetics are of low selectivity for compounds with similar molar mass and K0„ value. Diffusion coefficients in water ofthe magnitude of 10~5 cm2 s_1 are observed for the most compounds with molar masses up to 500 g mol-1. Therefore, the exchange rate parameters are likely to be similar for compounds of similar size and hydrophobicity. On the other hand, calibration studies and field exposures must manage the effect of flow, because the hydrodynamic regime can strongly affect the resistance ofthe aqueous boundary layer to mass transfer. The results of this study indicate that the accumulation kinetics ofhydrophobic compounds (logifow > 4.5) is sensitive to slight changes in flow approximately up to the flow rate of0.28 cm s_1.Underfaster,but still laminar, streaming (0.28 cm s~' < v < 1.14 cm s~'), the sensitivity ofthe mass transfer to changes in flow decreases to a nonsignificant level for the substances under consideration. Management of the Effect of Hydrodynamics on the Exchange Kinetics. This study confirms that for accurate estimation of aqueous contaminant concentrations from the amounts accumulated by SPMDs it is absolutely necessary to manage the effect of flow regime on the exchange kinetics. Achievement of a strictly controlled flow is rather complicated in laboratory experiments and almost impossible to reproduce in the field without expensive equipment. It is more realistic to conduct the calibrations under conditions of a low sensitivity to small changes in flow and to construct appropriate field deployment devices, buffering the flow efficiently so that a good correspondence of the exchange kinetics of contaminants in situ with the calibration data is obtained. The results of this study indicate that there might be an optimal regime under laminar flow conditions. In the case of a turbulent exterior flow, the theory leads to a proportionality ofthe limiting diffusional flux to the 0.8—0.9 power of velocity (19). Thus, an additional increase of k„ with increasing flow rate is expected when the flow regime switches from laminar to turbulent. Note that the change of fcw causes also a shift ofthe actual point of switch (i.e.,analyte ^spmd value) from aqueous layer control to membrane control (20). For the necessary laboratory—field comparison of the exchange kinetics, PRCs should be used. The desired attribute of a PRC is the high sensitivity of £e~PRc to changes in flow. The release kinetics of Dio-ANT, a PRC compound used in this study, follows the changes in flow rate with a quite satisfactory sensitivity (note that anthracene is photosensitive and should be used with caution in the field studies). Compounds with moderate hydrophobicity (log K0„ < 4) are disqualified as flow regime sensors because they are accumulated under membrane control, and their exchange kinetics is insensitive to changes in water flow regime. More hydrophobic PRCs (log K0„ > 5) might produce more significant differences in release kinetics under varying hydrodynamic conditions. However, very long exposure times (months or so) would be needed to achieve a significant decrease in SPMD concentrations of such substances. Nevertheless, the dissipation rates of a flow-sensitive PRC (i.e., £e~PRc) from environmentally exposed SPMD can be compared to the £e~cALderived forthe same compound during a laboratory calibration study to determine the effect of VOL. 36, NO. 2, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY ■ 295 exposure conditions on sampling. This approach has been shown by Huckins et al. (20). For compounds that are accumulated under aqueous layer control (i.e., if l/£w ^ l/&spMD/^spMD),the apparent samplingrate can be calculated as Rs = k„A. The values of apparent sampling rates Rs calculated for experiments conducted at different flow rates are given in Table 4. If the condition of equality of the temperature at the sampling site and in the laboratory calibration study is fulfilled, it can be shown that ^e-PRC Rs-fm\A = RS-CALZ (10) "e-CAL The Rs-fieid value can be introduced to eq 3 to calculate the TWAaqueous concentration of the analyte. Finally, the results of this study confirm that the use of the laboratory-derived calibration data for the estimation ofanalyte concentrations in the ambient environment is limited unless flow-sensitive performance reference compounds are used. Supporting Information Available Estimation ofthe time to reach steady-state fluxin individual SPMD compartments and the equations for the rate of transfer in steady state, determination of recovery rates of test chemicals from different matrices, extraction procedure of test substances from water, and an example ofthe use of PRCs to adjust the sampling rates. This material is available free of charge via the Internet at http://pubs.acs.org. Literature Cited (1) Huckins, J.N.;Tubergen,M. W.;Manuweera,G. K. Chemosphere 1990, 20, 533-552. (2) Huckins, J. N.;Manuweera, G. K.;Petty, J. D.;Mackay, D.;Lebo, J. A Environ. Sci. Technol. 1993, 27, 2489-2496. (3) Huckins, J. N.; Petty, J. D.; Lebo, J. A; Orazio, C. E.; Prest, H. F.; Tillitt, D. E.; Ellis, G. S.; Johnson, B. T.; Manuweera, G. K. In Techniques in Aquatic Toxicology; Ostrander, G. K., Ed.; CRC Press (Lewis Publishers): Boca Raton, FL, 1996; pp 625—655. (4) Meadows,J.C.;Echols,K.R.;Huckins,J.N.;Borsuk,F.A;Carline, R. F.; Tillitt, D. E. Environ. Sci. Technol. 1998, 32, 1847-1852. (5) Huckins, J. N.;Petty, J. D.; Orazio, C. E.; Lebo, J. A; Clark, R. C; Gibson, V. L.; Gala, W. R.; Echols, K. R. Environ. Sci. Technol. 1999, 33, 3918-3923. (6) Booij, K.; Sleiderink, H. M.; Smedes, F. Environ. Toxicol. Chem. 1998, 17, 1236-1245. (7) Gale, R. W. Environ. Sci. Technol. 1998, 32, 2292-2300. (8) Huckins, J. N.; Petty, J. D.; Prest, H. F.; Orazio, C. E.; Gale, R. W. In Abstracts, 18th Annual Meeting, SETAC, San Francisco, CA Nov 16—20,1997; Society of Environmental Toxicology and Chemistry: Pensicola, FL, 1997; p 206. (9) Crank, J. The Mathematics of Diffusion, 2nd ed.; Oxford University Press: New York, 1975; Chapter 4. (10) Mackay, D.; Paterson, S. Environ. Sci. Technol. 1981, 75, 1006-1014. (11) Mackay, D.; Hughes, A I. Environ. Sci. Technol. 1984,18, 439-444. (12) Vrana, B.; Paschke A; Popp, P.; Schüürmann, G. Environ. Sci. Pollut. Res. 2001, 8, 27-34. (13) Cussler, E. L. In Diffusion: Mass Transfer in Fluid Systems: Cambridge University Press: Cambridge, U.K., 1984. (14) Jeannot, M. A; Cantwell, F. F. Anal. Chem. 1997,69,235-239. (15) Lyman, W. J. In Handbook of Chemical Property Estimation Methods, Environmental Behavior of Organic Compounds', Lyman, W. J., Riehl, W. F., Rosenblatt, D. H., Eds.;McGraw-Hill: New York, 1982. (16) Mackay, D.; Shiu, W. Y; Ma, K. C. In Illustrated Handbook of Physical-Chem ical Properties of Environm enmtal Fate of Organic Chemicals; Lewis Publishers: Chelsea, MI, 1992; Vol. I. (17) Mackay, D.; Shiu, W. Y; Ma, K. C. In Illustrated Handbook of Physical-Chem ical Properties of Environm enmtal Fate of Organic Chemicals; Lewis Publishers: Chelsea, MI, 1997; Vol. V. (18) Chiou, C. T. Environ. Sci. Technol. 1985, 19, 57-62. (19) Levich, V. G. In Physicochemical Hydrodynamics; Prentice-Hall: London, U.K., 1962; pp 301-307. (20) Huckins, J. N.; Petty, J. D.; Prest, H. F.; Clark, R. C; Alvarez, D. A; Orazio, C. E.; Lebo, J. A; Cranor, W. L.; Johnson, B. T. In A Guide for the Use of Sem iperm eable Mem bran e Devices (SPMDs j as Sam piers of Waterborne Hydrophobic Organic Con tarn in ants. Report for the American Petroleum /rcsn'fHfefAP/^APIPublication 4690; API: Washington, DC, 2000. Received for review March 2, 2001. Revised manuscript received August 29, 2001. Accepted October 19, 2001. ESO 100625 296 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 2, 2002 Príloha 5 Wennrich L., Vrana B., Popp P., and Lorenz W., Development of an integrative passive sampler for the monitoring of organic water pollutants, J. Environ. Monit., 2003, 5, 813-822. Development of an integrative passive sampler for the monitoring of organic water pollutants Luise Wennrich,*0 Branislav Vrana,^ Peter Poppc and Wilhelm Lorenz^ "Leibniz-Institute of Surface Modification, Permoserstrasse 15, D-04318 Leipzig, Germany. E-mail: luise.wennrich@iom-leipzig.de; Fax: +49 341 235 2584; Tel: +49 341 235 3184 bDepartment of Chemical Ecotoxicology, UFZ-Centre for Environmental Research Leipzig-Halle, Permoserstrasse 15, D-04318 Leipzig, Germany cDepartment of Analytical Chemistry, UFZ-Centre for Environmental Research Leipzig-Halle, Permoserstrasse 15, D-04318 Leipzig, Germany d Mar tin-Luther-University Halle-Wittenberg, Institute of Analytical and Environmental Chemistry, Kurt-Mothes-Strasse 2, D-06120 Halle (Saale), Germany Received 27th March 2003, Accepted 10th July 2003 First published as an Advance Article on the web 1st August 2003 The development of convenient and competitive devices and methods for monitoring of organic pollutants in the aquatic environment is of increasing interest. An integrative passive sampling system has been developed which consists of a solid poly(dimethylsiloxane) (PDMS) material (tube or rod), acting as hydrophobic organic receiving phase, enclosed in a water-filled or an air-filled low-density polyethylene (LDPE) membrane tubing. These samplers enable the direct analysis of the pollutants accumulated during exposure in the receiving phase by thermodesorption-GC/MS, avoiding expensive sample preparation and cleanups. The capabilities of these sampling devices were studied for the sampling of 20 persistent organic pollutants (chlorobenzenes, hexachlorocyclohexanes, /?,/?'-DDE, PAHs, and PCBs) in laboratory exposure experiments. For the three sampler designs investigated the uptake of all target analytes was integrative over exposure periods up to 9 days (except PCB 101). The determined sampling rates range from 4 to 1340 ul h_1 for the water-filled samplers and from 20 to 6360 u.1 h_1 for the air-filled ones, respectively. The sampling rate of the analytes is dependent on their molecular weight, partition between water and sampler media (PDMS, polyethylene, water, air) and also of the sampler design. The passive samplers enable the estimation of time-weighted average (TWA) concentration of water pollutants in the lower ng 1_1 to pg 1_1 range. Introduction The monitoring of environmental pollutants in the ground and surface waters is of fundamental importance for both the protection of these ecosystems and the quality of human life. In particular the determination of persistent organic pollutants (POPs) is of ecotoxicological relevance due to their high toxic potential, their persistence and their tendency to bioaccumu-late. As is known, these pollutants can be present in the aquatic environment both freely dissolved and particle-bound. For ecological risk assessment the bioavailable fraction is of substantial interest. This corresponds with the freely dissolved fraction. Using conventional sampling techniques (grab sampling) only the total content of the pollutants is obtained. Furthermore, the conventional sampling and analysis of grab water samples only provide information about pollutant burden at the moment of sampling. Passive sampling techniques can overcome the problems mentioned above. These techniques allow the convenient determination of the time-weighted average concentration of the freely dissolved fraction of pollutants over several weeks or even months. Compared to conventional sampling the number of the samples and thus the expense of sampling and subsequent analysis can be reduced significantly. In addition, due to the accumulation of the pollutants over the whole sampling period, passive sampling allows the detection of even low analyte concentrations. Furthermore, the sampling devices are usually simple in design, small, inexpensive and require no power supply. This makes the technique inexpensive and suitable for use at remote sites. However, for the determination DOI: 10.1039/b303497f of TWA concentrations of organic pollutants in field studies samplers must be calibrated in laboratory experiments. Today the passive sampling technique represents an attractive alternative to the conventional snap-shot sampling for water monitoring of semivolatile POPs. In the last few years various passive sampling devices were designed for monitoring of pollutants in the aquatic systems. These sampling devices are usually so-called membrane samplers. Such membrane samplers typically consist of a receiving medium with a high affinity for the organic contaminants enclosed by a diffusion-limiting semipermeable membrane.1^4 Semipermeable membrane devices (SPMDs), introduced by Huckins and co-workers,5-9 attained the greatest importance and widespread application. Due to both their high membrane surface area and their relatively large volume of receiving medium SPMDs proved to be most effective in their capacity to accumulate lipophilic contaminants. The SPMD sampler consists of layflat low-density polyethylene tubing enclosing a thin film of triolein. The main disadvantage of the SPMD technique is the complex sample preparation procedure required to recover the accumulated pollutants from the collecting phase (triolein). This is achieved by dialysis using considerable amounts of organic solvents, followed by concentration of the extracts and an expensive cleanup before the chromatographic analysis.9'10 In the last few years several attempts have been made to develop passive sampling devices, which avoid the drawbacks mentioned above and also make passive sampling technology more attractive for routine monitoring programmes. Such passive samplers contain solid materials (granular adsorbents /. Environ. Monk., 2003, 5, 813-822 813 This journal is © The Royal Society of Chemistry 2003 and compact polymeric sorbents, like PDMS, respectively) instead of the liquid organic receiving phase. That allows the thermodesorption of the accumulated pollutants without additional sample preparation. Hardy et al.11-13 created a passive sampler consisting of a glass tube, sealed at one side with a silicone-polycarbonate membrane. Depending on the target analytes this sampler can be filled with various granular materials, such as activated charcoal, Tenax-TA, XAD-7, Chromosorb 103 and Porapak Q. After exposure, the granular receiving phase can be desorbed either with a suitable solvent or thermally. This sampler was successfully applied for the enrichment of more volatile organic compounds, like monocyclic aromatic compounds13 and phenols,12 whereas the less volatile compounds were not enriched effectively. At the end of the nineties Gratwohl and Martin14'15 patented a so-called ceramic dosimeter for the integrative sampling of organic compounds in ground water. This sampler consists of a porous ceramic tube which was filled with different grained adsorbents, e.g. the ion exchange resin Amberlite IRA-743 and Tenax. The porous ceramic tube enables only the dissolved analytes to pass the membrane. This sampler was applied for the monitoring of several PAHs in ground water. Concerning the subsequent thermodesorption of the analytes from Tenax difficulties appear due to the unexpected water permeability of the ceramic membranes. In a recently published paper, Vrana et al.16 described the application of a solid sorbent on the basis of PDMS as receiving phase in a membrane sampler. This so-called MESCO (membrane-enclosed sorptive coating) sampler consists of a stir bar coated with a thin PDMS layer (Gerstel Twister, a commercially available device used for stir bar sorptive extraction, SBSE17) enclosed in a water-filled dialysis membrane bag from regenerated cellulose. After exposure of the sampler, the PDMS coated stir bar is taken from the enveloping membrane and can be directly analysed by thermodesorption-GC/MS. Thus, laborious and time-consuming sample preparation can be avoided. PDMS is recommended as a receiving phase in extraction and thermodesorption as it has a number of benefits compared with other sorbents.18 The predominant mechanism of analyte extraction into the polymer PDMS phase is absorptive partitioning, which means that displacement effects of the analytes which are characteristic for adsorbents play no role. Although the MESCO sampler is a miniaturised version, this passive sampling approach enables lower quantification limits for hydrophobic POPs in the pg 1_1 level. The application of regenerated cellulose as a porous hydrophilic membrane material enables the widening of the applicability to a broader polarity range of pollutants including low-hydrophobic substances (log Kow < 4). Unfortunately, this material has relatively low chemical and thermal stability and is subject to microbial degradation,3 which potentially leads to the damage of the sampler in the field. The aim of the work presented here was to develop and to test a membrane sampler combining the advantages of the MESCO sampler with those of more stable membranes, such as low-density polyethylene. LDPE membranes were successfully applied in SPMDs. These membranes are hydrophobic, resistant to solvents and biodegradation and they can be heat-sealed. Furthermore, the commercially available stir bars as receiving phase should be substituted by less expensive PDMS materials with a significantly enhanced volume to increase the maximum exposure time of the passive sampler in the field. Theory Previously, models have been developed describing the uptake kinetics of organic contaminants in water by passive samplers constructed as solvent filled dialysis membranes,19 triolein filled polyethylene membranes20 or membrane enclosed sorptive coatings16 and can analogously be adapted for the description of the function of samplers designed in this study. These consist of a hydrophobic solid receiving phase (PDMS) enclosed in water-filled or air-filled semipermeable membrane made of nonporous LDPE. The mass transfer of an analyte in a sampler includes several diffusion and interfacial transport steps across all barriers, i.e. the stagnant aqueous boundary layer, possible biofilm layer, the membrane, the inner fluid (aqueous or gas) phase, and the receiving organic phase as rate control step is not assumed a priori. It can be shown that in the initial uptake phase, chemical uptake is linear or time-integrative. Under these conditions the concentration of a chemical in the organic phase is directly proportional to the product of the concentration in the surrounding aqueous medium Cw [kg m-3] and the exposure time t [s]. For practical application, uptake can be described by eqn. (1) Ms(t) = M0 + CwRst (1) where Ms [kg] is the amount of analyte accumulated in the receiving phase and M0 [kg] the initial amount of analyte in the sampler. Rs [m3 s_1] is the sampling rate of the system: Rs = kov A (2) where kov [m s_1] is the overall mass transfer coefficient and A [m2] is the membrane surface area. Sampling rate can be determined experimentally under fixed conditions at constant analyte concentration. Under environmental conditions, when the water concentration changes during the exposure, the term Cw represents a TWA concentration during the deployment period. As described by Huckins et al.,21 the uptake of an analyte into the passive sampler is linear and integrative approximately until the concentration factor of the sampler (ratio Cs(0/Cw) reaches half-saturation. If sampling rates Rs and organic receiving phase/water partitioning coefficients A"sw are available, the maximum exposure time in which the sampling device works integrative under field conditions can be estimated using eqn. (3): t50 ~ \n2KswVs/Rs (3) where t50 is the first-order half-time of the uptake curve and Vs the volume of the receiving phase. Experimental Chemicals and materials The test substances (Table 1) include several groups of semivolatile persistent organic pollutants: hexachlorocyclo-hexanes (HCHs), chlorinated benzenes (CBs), 2,2'-bis(4-chloropheny)-l,l'-dichloroethylene (p/-DDE), PAHs, and PCBs. HCH, chlorobenzene, PCB and PAH reference standards were obtained from Promochem (Wesel, Germany). The solvents w-hexane, methanol and dichloromethane (for organic trace analysis) were purchased from Merck (Darmstadt, Germany). HPLC-grade water was supplied by Baker (Deven-ter, The Netherlands). Layflat LDPE membrane tubing (layflat, 30 mm; wall thickness, 80 urn) was achieved from Polymer-Synthese-Werk GmbH (Rheinberg, Germany). Silicone tubing (3.0 mm x 3.6 mm) was obtained from Reichelt (Heidelberg, Germany). Silicone rod material (2.0 mm id) was purchased from Goodfellow (Bad Nauheim, Germany). Stir 814 /. Environ. Monit., 2003, 5, 813-822 Table 1 Selected physicochemical properties of the test analytes Compound Abbreviation No. MW° log W at 25 °C log KMWC at 25 °C d AAW at 25 °C £>A7cm2 s_1 at 20 °C V/cm2 s 1 at 20 °C 1,2,4,5-Tetrachlorobenzene TeCB 1 215.9 4.5 4.0 4.9 X io-2 0.06 6.2 x 10~6 Pentachlorobenzene PeCB 2 250.3 5.2 4.6 3.4 X io-2 0.057 5.8 x 10~6 Hexachlorobenzene HCB 3 284.8 5.5 4.8 5.3 X io-2 0.0543 5.5 x 10~6 ot-HCH ot-HCH 4 290.8 3.7 3.2 5.0 X 10~4 0.05 6.2 x 10~6 ß-HCH ß-HCH 5 290.8 3.8 3.3 1.8 X io-5 0.05 6.2 x 10~6 y-HCH y-HCH 6 290.8 3.7 3.2 2.1 X 10~4 0.05 6.2 x 10~6 5-HCH 5-HCH 7 290.8 4.1 3.6 1.8 X io-5 0.05 6.2 x 10~6 PCB 28 PCB 28 8 257.5 5.6 4.9 8.2 X io-3 0.0542 5.1 x 10~6 PCB 52 PCB 52 9 292.0 6.1 5.2 8.2 X io-3 0.054 4.9 x 10~6 PCB 101 PCB 101 10 326.4 6.8 5.6 1.4 X io-2 0.054 4.7 x 10~6 p,p'-DDE p,p'-DDE 11 318.0 5.7 5.0 1.7 X io-3 0.05 5.0 x 10~6 Acenaphthylene Ace 12 152.2 4.0 3.5 3.4 X io-3 0.063 6.5 x 10~6 Acenaphthene Acenaph 13 154.2 4.0 3.5 4.9 X io-3 0.063 6.3 x 10~6 Fluorene Flu 14 166.2 4.2 3.7 3.2 X 10~3 0.06 6.0 x 10~6 Phenanthrene Phe 15 178.2 4.5 4.0 1.3 X io-3 0.058 5.8 x 10~6 Anthracene Ant 16 178.2 4.6 4.4 1.6 X io-3 0.058 5.9 x 10~6 Fluoranthene FLU 17 202.3 5.1 4.5 4.2 X 10~4 0.055 5.5 x 10~6 Pyrene Pyr 18 202.3 5.1 4.5 3.7 X 10~4 0.055 5.6 x 10~6 Benzo[a]anthracene BaA 19 228.3 5.9 5.1 2.3 X 10~4 0.052 5.1 x 10~6 Chrysene CHR 20 228.3 5.7 5.0 2.6 X 10~5 0.052 5.1 x 10~6 "Molecular weight. *Octanol-water partition coefficient.22'23 cMembrane-water partition coefficient estimated from eqn. (4).28 ^Henry's Law constant.22'23 ^Diffusion coefficient in air.25 -^Diffusion coefficient in water.24 g8T — length of the diffusion path in the transfer medium — 0.3 cm. bars for SBSE (PDMS coating: 0.5 mm thickness, 10 mm length) were obtained from Gerstel (Mulheim/Ruhr, Germany). Physicochemical properties of substances Henry's Law constants KAW at 25 °C of substances under investigation were taken from the literature.22'23 Almost equal values of aqueous diffusion coefficients Dw were estimated for the tested group of compounds ranging from 5 x IO-6 to 7 x IO-6 cm2 s-1.24 Diffusion coefficients of the test analytes in air DA at 20 °C range from 0.05 to 0.06 cm2 s"1.25 An approximated value of 10~10 cm2 s_1 was used as diffusion coefficient of the analytes in the LDPE membrane DM.26,27 The membrane/water partition coefficients KMW were estimated from a predictive equation derived by Hofmans:28 log KMW = -0.0956(log Kow)2 + 1.7643 log i^ow - 1.98 (4) Preparation and test of the sampler components The materials provided for receiving phases in the passive sampling devices (silicone tubes and rods) were obtained from the manufacturers as endless materials. In order to obtain reproducible results the tubes and rods were carefully cut with a scalpel in pieces of each 40 mm length and then weighed. Outliers in the weight (CV > 1%) were discarded. In order to clean and condition the silicone tubing and rods, in each case ten of these were placed into a vial (50 ml) containing 50 ml of w-hexane and horizontally shaken for 2 h (tubing) or 4 h (rods). The materials were dried in a desiccator under vacuum and then thermally conditioned for 3 h at 250 °C in a nitrogen flow of about 50 ml min-1. For cleaning and conditioning of the stir bars these were placed separately into small vials filled with 2 ml of a mixture of dichloromethane and methanol (1 : 1) for 1 h. Then they were dried in a desiccator and subsequently heated at 250 °C for 90 min in a nitrogen flow. For cleaning of the layflat LDPE tubing, 3 pieces of this material with a length of each 1 m were put into a glass vessel (500 ml) containing 500 ml of w-hexane and shaken for 24 h. Then the solvent was rejected and the procedure was repeated once. The wet tubing was dried in a desiccator. To investigate the applicability of some PDMS materials as organic receiving phases in the sampling devices these were tested within the complete extraction and thermodesorption procedures. For this purpose, the conditioned receiving phases were separately shaken in each 50 ml water spiked with the test analytes (100 ng 1_1 of each compound). This solution was prepared by spiking a water sample with a mixture of test analytes dissolved in methanol. The vials containing the tubes and the rods were horizontally shaken for 2 h at 200 motions per min. The stir bars were stirred at 1000 rpm in Erlenmeyer flasks for 2 h. After extraction, the receiving phases were taken from the water sample, rinsed with a small volume of water, and dabbed dry with a lint-free tissue. It should be noted that the small water droplets inside the tubes should be carefully removed. The accumulated analytes were determined using thermodesorption-GC/MS as described later. The completeness of the desorption of the enriched analytes (carry over) was revised by a second desorption under equal conditions. Membrane samplers The membrane samplers used in this study (Fig. 1) consist of a layflat LDPE membrane tubing (length, 50 mm) enclosing a Fig. 1 Schematic diagram of the passive sampling device described here. The receiving phase (component 1, silicone tubes or rods) is enclosed in low-density polyethylene membrane tubing (component 3) filled with the transfer medium (component 2, water or air) and heat-sealed at each end. /. Environ. Monit., 2003, 5, 813-822 815 silicone tube (length, 40 mm; referred to as tube sampler) or a silicone rod (length, 40 mm; referred to as rod sampler). The layflat LDPE tubing with the receiving phase inside was water-filled (about 8 ml) or air-filled and heat-sealed at both ends. For both samplers the volume of the receiving phase (about 125 u.1) and the effective membrane surface area (30 cm2) were equal. In order to enable a simultaneous exposure of a set of samplers, they were connected to a string. Laboratory exposure experiments A set of passive samplers were exposed to contaminated water with a nominal analyte concentration of each 50 ng 1_1 in a flow-through exposure system. This system consisted of an exposure chamber, an 1 m high glass column (inner diameter 7.5 cm) with a perforated bottom. To prevent photodegrada-tion of the analytes during exposure the column was covered with dark foil. In a mixing chamber (1 1) positioned at the bottom of the exposure column tap water (60 1 h_1) and the appropriate amount of the test analytes dissolved in methanol (400 Ltg 1_1) delivered by a peristaltic pump (Gilson, USA) were carefully mixed using a magnetic stirrer. The resulting methanol concentration in the exposure water did not exceed 0.01% (v/v). Tap water was fed to the mixing chamber by a membrane pump (Prominent, Germany). The spiked water flowed from bottom to top through the exposure chamber. Using a heating-cooling system the water temperature in the exposure chamber was held constant at the predetermined temperature. The passive sampler string was fixed in the exposure column in a vertical position. The exposure experiments were performed at 14 and 8 °C, respectively, and at a linear flow velocity of the water of 0.38 cm s_1 (see Table 2). The samplers were removed one by one after predetermined exposure times. (The maximum exposure times varied between 176 and 236 h.) Then the receiving phases were immediately taken out of the enveloping LDPE tubing and carefully dried. The loaded receiving phases were stored in closed small glass vials at —18 °C in a freezer until thermodesorption-GC/MS analysis. Investigations concerning the loss of analytes during storage of the loaded receiving phases under these conditions resulted in the conclusion that these could be neglected. In order to determine the concentration of the analytes under investigation in the water during exposure, samples were taken from the exposure column at each time when samplers were removed and analysed as described below. Processing of the water samples The extraction of the water samples taken from the exposure column was performed using SBSE. The procedure was as follows: 50 ml of the water sample was filled into an Erlenmeyer flask (50 ml), the stir bar was lowered in the flask and then the sample was stirred at 1000 rpm for 2 h. After this the stir bar was taken out, washed with water and dried. For external calibration, spiked water samples containing 10, 30, 50, 70, and 100 ng l"1 of each analyte were prepared using a mixture of test analytes dissolved in methanol and extracted as described above. It should be noted that the content of methanol in the calibration solutions should be held constant (<1%). Thermodesorption-GC/MS analysis The pollutants accumulated during the exposure experiments in the receiving phases of the passive samplers and in the stir bars were analysed using thermodesorption-GC/MS. The solid receiving material was placed into an empty glass desorption tube. Thermodesorption-GC/MS was performed on an Agilent Technologies system 6890/5973 (Palo Alto, CA, USA) equipped with a Gerstel thermodesorption device with auto-sampler. For cryofocusing of the analytes prior to the transfer into the capillary column a Gerstel cold injection system (CIS 4) with an empty liner was used. During thermal desorption the CIS 4 was cooled with liquid nitrogen to a temperature of — 150 °C. For the desorption of the analytes from the receiving phases and the stir bars the following conditions were chosen: desorption temperature, 250 °C; helium flow rate, 100 ml min-1 and desorption time, 10 min. The transfer lines both from the thermodesorption device to the CIS 4 and from the GC to the MS ion source were set to 250 °C. After desorption of the receiving phase and cryofocusing of the analytes, the CIS 4 was heated to 250 °C at a rate of 12 °C s_1, whereas the system was used in the splitless mode with a splitless time of 1.5 min. An HP-5 MS capillary column (30 m, 0.25 mm id, 0.25 urn film thickness) was employed with the following temperature program: 50 °C, 3 min isothermal, 15 °C min"1 to 160 °C, then at 3 °C min-1 to the final temperature of 280 °C, and held for 8 min. Helium was used as carrier gas at a linear velocity of 39 cm s_1. The single ion monitoring (SIM) mode applying one or two characteristic ions per analyte was chosen for the detection. For external calibration of the accumulated pollutants in the receiving phases, a plug of silanised glass wool (length, about 4 cm) which was positioned in the heated zone of a desorption tube was spiked with the calibration solution (2 u.1). The desorption tube was flushed for 1 min with a nitrogen flow of 30 ml min-1 to allow the main part of the solvent (methanol) to evaporate and then thermally desorbed. In order to control analyte losses during the evaporation of methanol at external calibration, the flush time was varied in the range of 30 to 120 s. This investigations resulted in no significant decrease of the peak areas with increased flush time. Quantification of the analytes sorbed in the receiving phase was performed using a six-point calibration. Results and discussion Assessment of PDMS materials In a preliminary study the applicability of some commercially available PDMS materials—silicone tubes and silicone rods— as organic receiving phase in the passive sampling devices were investigated to achieve information about the extraction efficiency, the repeatability, completeness of the thermodesorption process (carry over), and the handling of the materials. For this purpose, each eight pieces of the receiving phases were object of the complete extraction and thermodesorption procedures (see the Experimental section—Preparation and test of the sampler components). Additionally, stir bars were included in the experiments, because they should serve on the one hand for comparison and they were employed for the analysis of the water samples on the other hand. The results of Table 2 Conditions of the flow-through exposure experiments Nominal Exposure Experiment no. Sampler design used concentration/ng 1_1 Temperatur e/°C Flow velocity/cm s 1 period/h la Water-filled tube sampler 50 14 0.38 176 lb Air-filled tube sampler 50 14 0.38 224 lc Water-filled rod sampler 50 14 0.38 224 2 Water-filled tube sampler 50 8 0.38 236 816 /. Environ. Monit., 2003, 5, 813-822 Table 3 Mean peak areas (n — 8), coefficients of variation (CV in %) and carry over (%) of different receiving phase materials obtained from extraction and thermodesorption-GC/MS analysis Compound Stir bars Tubes Rods Peak area x 10~3 CV (%) Carry over (%) Rel. peak area" (%) CV (%) Carry over (%) Rel. peak area" (%) CV (%) Carry over (%) 1,2,4,5-Tetrachlorobenzene 947 5.2 0.32 1.03 9.1 4.09 0.65 8.8 5.27 Pentachlorobenzene 965 6.0 0.15 0.65 6.8 1.36 0.68 6.5 4.37 Hexachlorobenzene 1004 6.9 0.12 0.65 15.5 nd 0.73 3.8 3.62 ot-HCH 391 7.6 0.05 0.72 6.1 nd 0.85 5.2 3.03 ß-HCH 57 9.6 3.00 1.33 4.5 2.25 1.71 19.7 4.28 y-HCH 305 8.7 0.11 0.64 6.3 0.40 0.84 6.8 3.52 5-HCH 117 8.5 0.86 0.85 7.8 2.64 1.10 20.6 0.70 PCB 28 1565 9.3 0.18 0.46 5.9 0.81 0.75 5.0 4.21 PCB 52 898 10.3 nd 0.43 5.2 0.28 0.74 6.9 4.73 PCB 101 510 11.1 0.09 0.36 11.1 0.33 0.67 16.5 6.07 p,p'-DDE 376 10.9 nd 0.36 13.2 0.09 0.65 20.5 5.78 Acenaphthylene 1272 5.5 0.26 1.00 6.9 1.31 0.82 7.4 3.41 Acenaphthene 1496 6.5 0.54 0.84 6.6 1.46 0.61 14.9 4.02 Fluorene 1287 7.6 0.50 0.72 7.3 1.36 0.84 4.6 3.72 Phenanthrene 2058 9.3 1.19 0.63 7.3 0.98 0.80 5.8 3.94 Anthracene 1386 9.8 0.14 0.54 5.9 0.42 p.i. Fluoranthene 1673 11.5 0.12 0.41 3.6 0.46 0.65 19.9 4.82 Pyrene 1643 11.2 0.12 0.41 6.7 0.44 0.55 20.9 5.17 Benzo[a]anthracene 281 12.8 nd 0.62 18.9 0.18 0.85 28.6 3.53 Chrysene 364 8.8 nd 0.52 14.0 0.25 0.89 25.5 3.97 "Related to the mean peak areas of the stir bars (n — 8). *nd — not detectable. cpi — peak interference. these investigations are summarised in Table 3. The carry over provides information about the completeness of the thermodesorption process. For this purpose, the already thermally desorbed materials were desorbed again and the peak areas of the first and second desorption were compared, setting the areas of the first desorption to 100%. It should be noted that the volume of the PDMS phase of the stir bars (24 u.1) and the other materials (125 u.1) differ considerably. The extraction yields (relative peak areas) of the analytes investigated using tubes and rods were in the range of 0.37 to 1.33 compared to stir bars. From the three receiving phases, the best repeatability was found for the stir bars. The variation coefficients of the peak areas of the individual analytes extracted from the spiked solution by the 8 stir bars ranged from 5 to 13%. Comparing the tubes and the rods, the first ones showed a better repeatability with variation coefficients from 4 to 19% (tubes) and from 4 to 29% (rods), respectively. The values for the carry over of the stir bars indicate that the thermodesorption of the most compounds under the given conditions was nearly quantitative (<1% except P-HCH and phenanthrene). The values for the tubes were slightly higher (in most cases lower than 1.5%). In contrast, the carry over of the rods was significantly increased (between 3.0 and 6.1%). The reasons for this finding we assume in the significantly larger thickness of the PDMS layer of the rods (2 mm id) compared to the other materials (tubes, 0.3 mm; stir bars, 0.5 mm). Thus, an increased time is needed for the quantitative diffusion of the analyte molecules dissolved in the PDMS phase to the surface area of the rods. For this reason the silicone tubes were favoured as receiving phase material in the passive sampler devices. However, it was found that the handling of the rods is more convenient, especially by the preparation of the air bubble-free water-filled samplers and in consideration of drying (removing of the small water droplets on the inner surface area of the tubes). Therefore, in one exposure experiment rod samplers were included, too. Calibration experiments The capabilities of the passive sampling devices described here for the long-term water monitoring of the target analytes were investigated by performing exposure experiments in a flow-through exposure apparatus. In Table 2 the experimental conditions are summarised. Over the exposure periods the analyte concentrations in the water as well as the temperature and the flow rate of the water were held constant. As described above, during the exposure experiments water samples and passive samplers were taken from the exposure column at time intervals to determine the analyte concentrations in the water (Cw) and the amounts accumulated in the samplers (Ms). The mean concentration of the individual analytes in the water samples within the exposures (Cw) were in the range of 78 to 131% of the nominal concentration (except p,p'-DDE in experiment 1). The calculated coefficients of variation of the average analyte concentrations were at maximum 11%. The experimentally determined time courses of the accumulated amounts of individual analytes on the receiving phases (Ms) were fitted by the linear regression analysis. According to eqn. (1) the adjustable parameters are the slope (Cw^s) and the intercept (M0) of the linear uptake curve. The quality of the fit was characterised by the standard deviations of the optimised parameters, as well as the correlation coefficient (K) and the fit standard deviation (SD). Typical uptake curves are shown in Fig. 2. Using tube samplers the uptake of all compounds was linear over the whole exposure time in all experiments. The correlation coefficients of the regression were in the range of 12 250 Exposure time (h) Fig. 2 Uptake of selected analytes by the air-filled tube sampler obtained from a flow-through exposure at 14 °C (nominal analyte concentration, 50 ng For abbreviations see Table 1. /. Environ. Monit., 2003, 5, 813-822 817 0.867 to 0.988. The variation coefficients of the calculated slope did not exceed 20%. Using rod samplers (experiment lc) the uptake of all analytes was linear except PCB 101. The correlation coefficients of the regression ranged from 0.698 to 0.990. The variation coefficients of the slope were in maximum 34%. The uptake curves of the analytes show partly negative intercepts. From the theory16 negative intercepts can be explained by the presence of a lag phase. This can be interpreted as the time needed for the analyte to pass the LDPE membrane. The duration of the lag phase or the so-called delay time is affected by the diffusivity of analyte and thickness of individual barriers (membrane and diffusion layers of fluid media). Moreover, steady-state flow of analyte from water to the receiving phase is not established immediately. However, the time to reach steady-state flux in the sampler can be estimated by the magnitude of the variable l2/Dt, where / is the film thickness, D is the diffusion coefficient and t is time.29 If the variable is less than unity, a steady-state flux is assumed. Using the thickness of the polyethylene membrane of 100 um and a typical diffusion coefficient of small non-polar molecules in LDPE membranes of 10~10 cm2 s_1, steady-state should be achieved after one or two days in the polyethylene membrane. This corresponds well with the lag phase observed in our experiments. In most cases the calculated lag phases were in the range between 5 and 30 h, however, for the PCBs lag phases up to 48 h were found. In aqueous and air boundary layers, steady-state should be established after few minutes only. To use the sampler for the monitoring purposes, analytes should approach steady-state in the individual compartments quickly with regard to the duration of experiments, i.e. duration of the transition phase should not be much longer than 10% of the exposure period. Sampling rates. The sampling rates Rs of the three types of passive samplers obtained in the exposure experiments 1 and 2 are given in Table 4. According to eqn. (1) the Rs values were calculated by dividing the slope of the linear uptake curve by the mean analyte concentration Cw in the water during exposure. The variances of the Rs values were calculated from both the coefficients of variation of the slope and of the analyte concentration in the aqueous phase, according to the law of error propagation. Over the range of the controlled exposure conditions, the Rs values of the analytes under investigation covered a range of 2 to 3 orders of magnitude. For example, for the water-filled tube sampler at 14 °C the Rs values were in the range of 5 to 1340 ul h_1. Comparing the sampling rates of the PAHs it can be seen that the values decrease with increasing molecular weight (size), increasing hydrophobicity (log A"0w ranged from 4.0 to 5.9) and decreasing water solubility of the compounds. A similar behaviour, significantly decreased sampling rates with increasing chlorination grade, was found for the chlorinated benzenes and the PCBs. Originally, higher chlorinated PCBs (PCB 138, PCB 153 and PCB 180) and EPA PAHs with high molecular weights (benzo[6]fluoranthene, benzo[/r]fluoranfhene, benzo[a]pyrene, indeno[l,2,3-cii]pyrene, dibenz[a/z]anthracene and benzo[ghi]-perylene) were to have been included in the exposure studies, too. However, in previous investigations it was found that these compounds were accumulated in the receiving phase only in very small amounts (near the detection limits). Therefore an accurate determination of the sampling rates was precluded. The sampling rates of the samplers described here are low (0.12 to 32 ml per day for the individual analytes using the water-filled tube sampler) compared with those of other sampling devices, such as standard SPMDs21 (1 to 8 1 per day). That means, the sampling efficiency of the SPMDs is about 3 orders of magnitude higher. Nevertheless, the sensitivity of the two methods should be approximately the same, because in the case of the samplers described here, the total amount of the analyte accumulated in the receiving phase is transferred to the GC/MS. In contrast, only a small portion of the obtained SPMD extract (usually 1-2 ul) is injected. Comparing the sampling rates given in Table 4 with those of the MESCO sampler16 it can be seen that the Rs values are in the same order of magnitude (in the ul h_1 range), as expected, but the MESCO sampling rates are more uniform. Additionally, for the PCBs and PAHs with high molecular weights Rs values could be determined, too. That means that measurable amounts of these analytes were accumulated in the PDMS material during the exposure. The main difference between these both sampling devices is in the membrane material employed. The membrane of the MESCO sampler consists of porous hydrophilic polymeric material (molecular weight Table 4 Sampling rates Rs of the 3 passive sampler designs derived from flow-through exposures at different temperatures (nominal analyte concentration 50 ng~') Compound Water-filled tube samplers Air-filled tube samplers Water-filled rod samplers T = 8 °C T = 14 °C T = 14 °C T = 14 °C CV (%) Rs/yá h_1 CV (%) Rs/\ú rT1 CV (%) Rs/vi ir1 CV (%) 1,2,4,5-Tetrachlorobenzene 737 9 647 9 6355 9 480 11 Pentachlorobenzene 201 11 192 11 4314 11 214 14 Hexachlorobenzene 21 13 56 22 904 14 87 18 oe-HCH 229 11 185 13 136 8 283 9 P-HCH 31 16 69 11 34 17 69 14 y-HCH 120 10 141 11 72 8 195 10 5-HCH 44 11 96 10 24 9 138 11 PCB 28 96 10 57 15 921 13 64 20 PCB 52 52 12 41 18 621 13 33 35 PCB 101 5 13 a 104 15 4 80 p,p'-DDE 4 14 5 20 53 14 5 21 Acenaphthylene 988 9 730 9 1398 8 507 7 Acenaphthene 897 9 671 9 2226 7 481 7 Fluorene 907 9 1342 11 1876 6 753 8 Phenanthrene 541 10 269 11 929 8 259 11 Anthracene 515 11 265 14 988 12 125 12 Fluoranthene 69 10 56 9 122 10 37 14 Pyrene 42 11 34 10 99 13 30 15 Benzo[a]anthracene 13 15 10 19 31 16 8 14 Chrysene 9 14 9 19 20 13 6 27 "PCB 101 could not be determined in this experiment. 818 /. Environ. Monit., 2003, 5, 813-822 cutoff 1000). Thus, the analytes pass the membrane by diffusion through the water-filled pores. In contrast, the membrane of the samplers used in this study consists of nonporous polyethylene. The organic analytes can pass such a nonporous polymeric membrane only by dissolving in the polymeric phase and subsequent diffusion through the membrane layer. (LDPE membranes can be passed only by truly dissolved organic molecules with cross-sectional diameters up to about 1 nm.20) Thus, the diffusion coefficients of the individual organic substances in the polymer DM and the membrane/water partition coefficients KMW are of crucial importance for the sampling efficiency. Influence of the transfer medium on the sampling rates. In order to investigate the influence of the medium, which is contained in the sampling device together with the receiving phase, water-filled and air-filled tube samplers were exposed together under the same conditions. The determined sampling rates and variances are listed in Table 4. Comparing the Rs values in the columns 4 and 6 it can be seen that the values of most of the analytes for the air-filled sampler are significantly higher as for the water-filled ones with exception of the four HCH isomers. Thus, for the chlorobenzenes and the PCBs the Rs values are increased 10- to 20-fold and for the PAHs up to 4-fold, respectively. The comparability of experimentally derived sampler uptake rates to actual values during environmental sampling generally depends on the similarity of laboratory and site exposure conditions. When sampler calibration and field conditions are dissimilar, the magnitude of the differences in lab and field uptake rates for an analyte depends on the source of analyte rate control. Thus, examination of potential rate-limiting barriers is important. The overall mass transfer coefficient is expected to be affected by the diffusion of solutes in individual phases (water, membrane, the inner transfer medium [air or water], and the PDMS, respectively) and by their partitioning into the PDMS and the LDPE membrane, since accumulation of hydrophobic analytes is expected also in the hydrophobic membrane. From the theory,30'31 it is assumed that the overall resistance (l/kov), to the uptake of a chemical is given by the sum of particular barrier resistances to mass transfer [eqn. (5)]: 1 Si (5) where St is the particular barrier thickness, Dt is the diffusion coefficient in the barrier and Kiw is the partition coefficient between the j-th phase and water. For water-filled tube sampler, the overall resistance (l/kovWS) is then given by eqn. (6): 1 (6) The subscripts B, M, W and S represent the boundary aqueous layer at the surface of the sampler [B], the membrane [M], the transfer aqueous layer inside the sampler [W], and the receiving organic phase [S]. The resistance to mass transfer in the air-filled tube sampler can be described analogously by eqn. (7): 1 DmKm (7) where the subscript A denotes the air layer between the receiving phase and the membrane, and KAW is the dimension-less Henry's Law constant. It is likely that the differences in the sampling rates determined under the same exposure conditions for two sampler designs differing from each other only in the composition of the filling medium (water or air) are caused by differences in the partial resistance to mass transfer in this medium. These particular resistances are described by the corresponding terms c5wADw and SA/DAKAW in eqn. (6) and (7), respectively. The diffusion paths of analyte molecules through the inner transfer medium are approximately the same for both sampler designs (i.e. c5w x c5A). Practically, the exact distance between the membrane and the PDMS rod or tube cannot be measured because the PDMS rod or tube was not in a fixed position inside the membrane. This distance may vary between 1 and 5 mm and an approximate average value of <5T = 3 mm was taken for calculations of particular resistances of the inner medium to mass transfer of an analyte. Note that for both sampler designs, the mass transfer by convection in the inner transfer medium is assumed to be negligible. Thus, the differences in sampling rates for an analyte may originate in unequal transfer medium permeability for the two sampler designs. To examine the effect of the inner transport medium on the mass transfer in the sampler, the sampling rate ratio for two sampler designs (i^sAs/^sws) determined for the same analyte under equal exposure conditions can be expressed using a combination of eqn. (2), (6) and (7): Rsas Rsws DaKawJ (8) where A = SB/DW + SM/DMKMW + SS/DSKSW and ST is the length of the diffusion path in the inner transfer medium. The sampling rate ratio is then modulated by the value of analyte's diffusion coefficient in water, the diffusion coefficient in air and the Henry's Law constant, respectively. We assume that the diffusion in membrane and/or the inner transfer medium are dominant diffusion limiting steps. The aqueous boundary layer at the surface of the sampler and in the PDMS layer present only a small part of the total diffusion path. Therefore, the term A in eqn. (8) can be rewritten A x SM/DMKMW In order to prove the applicability of eqn. (8) for prediction of the RSAS/RSWS ratio from the physicochemical properties of analytes, a correlation of estimated and measured ratio was performed using linear regression analysis [eqn. (9)]: (^SAs/^SWs)c -1.153 + 2.023(iWiW> e: (9) n = 19; SD = 8.68; r = 0.85; P < 0.0001 The fit yields a good correlation (see also Fig. 3). However, the calculated ratio overestimates the experimental value on average by a factor of 2. The systematic error is likely introduced into the calculation by using an imprecise value of the distance between membrane and PDMS phase c5T. A simulation of the effect of varying <5T on the estimated RSas/ Rsws ratio showed that a 5-fold increase in <5T from 1 to 5 mm results in a variation in the average slope of the linear dependence of calculated to measured RSAS/RSWS from 0.9 to 2.8. Despite this imprecision, experimental and estimated data correlate well for the whole range of simulated c5T. Thus, it appears that the observed differences in experimental Rs values for two sampler designs can be explained based on physico-chemical properties of analytes and theoretical considerations to mass transfer in samplers. The calculation of particular resistances shown in eqn. (6) and (7) allows also recognizing the dominant barriers to mass transfer. Any step or layer with more than 50% of the total resistance is considered rate limiting. The comparison relative contribution of individual barriers to the total resistance for each compound shows that the uptake rate control depends not only on the sampler construction, but also on the analyte properties (Table 5). The estimation of the rate limiting barrier will be verified by experiments in the future. Comparison of the tube and rod sampler. A comparison of the water-filled tube and rod samplers (Table 4, column 4 and /. Environ. Monit., 2003, 5, 813-822 819 PAHs, acenaphthylene, acenaphthene, phenanthrene, and anthracene, show a significant increase of the Rs values with decreasing temperature. The Rs values of the other PAHs (except fluorene) determined at the two exposure temperatures have no significant differences. The prediction of the temperature effect on the sampling rates is difficult because of the complexity of the system. Both thermodynamic and kinetic parameters affecting the sampling rate are temperature dependent. Based on widely applied relationships such as the Wilke-Chang equation and the Heyduk and Laude equation32 analyte diffusion coefficients in water are expected to be directly proportionally to temperature. On the other hand, the phenomenon of reduced or nearly constant solute permeability with increasing temperature has been observed in nonporous polymers such as LDPE.33 Typically, increased temperature should enhance mass transfer in all media and the uptake of target analytes should exhibit Arrhenius dependences. However, in membrane systems, non-ideal solute-polymer interactions may affect activation energy required for molecular diffusion, increasing complexity of the temperature-i^s relationship. Also, partition coefficients Kiw may decline enough with increasing temperature to offset increases in diffusion coefficients.34 Maximum exposure time t50. Maximum exposure time in which the passive sampling device accumulates a pollutant integrative under field conditions can be estimated according to eqn. (3) and the sampling rates Rs from the exposure experiments. As described in an earlier paper,16 the determination of distribution constants A"sw for the analyte partitioning between PDMS coating and aqueous sample in batch experiments causes difficulties. Therefore, the apparent distribution constants Af(PDMS), obtained from SPME experiments with PDMS coated fibers (100 um) was used as a substitute for the A"sw values in the estimation.16 The results of the t50 estimation for the water-filled tube sampler are given in Table 6. From the calculation results that for acenaphthylene, acenaphthene and fluorene the passive sampler may accumulate integrative about 2 to 3 weeks. Maximum exposure times from 3 to 10 weeks were estimated for HCHs. For the other PAHs investigated, HCB, DDE and PCBs, the t50 values may Table 5 Estimation of the main barrier to mass transfer in water-filled and air-filled passive sampler designs according to eqn. (6) or (7) Water-filled sampler Air-filled sampler Compound Membrane (%) Water (%) Rate limiting barrier Membrane (%) Air (%) Rate limiting barrier 1,2,4,5-Tetrachlorobenzene 14 86 W 99 1 M Pentachlorobenzene 4 96 W 93 7 M Hexachlorobenzene 2 98 W 92 8 M ot-HCH 49 51 W+M 79 21 M P-HCH 43 57 W+M 10 90 A y-HCH 49 51 W+M 62 38 M 5-HCH 27 73 W 5 95 A PCB 28 2 98 w 60 40 M PCB 52 1 99 w 41 59 M + A PCB 101 0 100 w 34 66 A p,p'-DDE 1 99 w 20 80 A Acenaphthylene 33 67 w 94 6 M Acenaphthene 32 68 w 96 4 M Fluorene 23 77 w 90 10 M Phenanthrene 13 87 w 66 34 M Anthracene 11 89 w 66 34 M Fluoranthene 4 96 w 15 85 A Pyrene 4 96 w 14 86 A Benzo[a]anthracene 1 99 w 3 97 A Chrysene 1 99 w 0 100 A °W = Water. *M = Membrane. CA = Air. 0 5 10 15 20 25 visas' ' sws'measured Fig. 3 Calculated versus experimental sampling rate ratio (RSAS/RSWS) for two sampler designs differing from each other only in the composition of the filling medium (air or water). The experimental ratio was determined for the two designs under the same exposure conditions in a flow-through experiment at 14 °C. The theoretical ratio Rs was calculated using eqn. (8). The line represents the linear regression given in eqn. (9). 8) shows that the sampling rates are similar. For the tube sampler the Rs values of PAHs are generally higher than those for the rod sampler. The variances of the sampling rates show increased values especially for the three PCBs for the rod sampler (PCB 28, 20%; PCB 52, 35% and PCB 101, 80%). A similar behaviour could be observed for chrysene (variation coefficient, 27%). The reason for this finding we assume in the relatively large thickness of the PDMS layer of the rods (2 mm id) and the associated deferred and incomplete desorption of these analytes. Because of the lower variances using the tube sampler this one has been favoured. Effect of the temperature. In order to study the influence of the temperature on the sampling rates, the water-filled tube samplers were exposed at two different temperatures (14 and 8 °C; see Table 4). A significant decrease (/-test) of the sampling rates with decreasing temperature was observed for hexachlorobenzene, P-HCH and 8-HCH. In contrast, the more volatile 820 /. Environ. Monit., 2003, 5, 813-822 Table 6 Estimated maximum exposure times /50 of the analytes using water-filled tube samplers in the field at 14 °C Compound log -Kf(pdms) /50/day Hexachlorobenzene 4.3° 1283 ot-HCH 3.2* 29 ß-HCH 2.7* 24 y-HCH 3.2* 40 5-HCH 3.3* 74 PCB 28 4.7° 3158 PCB 52 5.0° 8761 p,p'-DDE 5.2° 108749 Acenaphthylene 3.40c 12 Acenaphthene 3.63c 23 Fluorene 3.71c 14 Phenanthrene 3.96c 121 Anthracene 3.98c 129 Fluoranthene 4.71c 3287 Pyrene 4.86c 7625 Benzo[a]anthracene 5.26c 63769 Chrysene 5.69c 188469 "Data from reference 16. *Data from reference 36. cData from refer- ence 35. be several months and more. The results of the t50 calculation indicate that the passive sampler under investigation enables the estimation of TWA concentrations of pollutants from the amounts accumulated during field exposures of several weeks. As described above, the change of the inner transfer medium (from water to air) used in the samplers results in significantly increased sampling rates for most of the analytes investigated (except the HCH isomers) and thus, according to eqn. (3), in decreasing t50 values. It could be estimated, that the air-filled tube sampler may integrative sample the low molecular weight PAHs (acenaphthylene and acenaphthene) only up to one week. However, in the calibration experiment linear uptake were found to be up to nine days for these compounds. Sensitivity. The calculated sampling rates were used to estimate the potential of the sampling devices under study to detect low TWA concentrations of the target analytes. Based on eqn. (1), the minimum quantifiable TWA concentration of the analytes in ambient water were estimated, whereas the Ms values were replaced by the limits of quantification Ms(loq). According to the correlation mentioned above, the sensitivity of the entire analytical method depends on the sampling rate Rs and the exposure time of the sampler. That means, presuming the integrative uptake of the analyte from the sampler over the entire exposure period, the sensitivity improves with increasing exposure time. Assuming an exposure of 10 days, limits of quantification in the range of 3 pg 1_1 (fluorene) to 2.4 ng 1_1 (p,/?'-DDE) could be estimated for the water-filled tube sampler (at 14 °C). The use of the air-filled tube sampler enables a significant improvement in sensitivity for most of the target analytes except the HCH isomers. Therefore this sampler design is recommended if very low concentrations of pollutants are expected in the field. These results demonstrate that the sampling devices described here enable the detection of the target analytes in the lower ng 1_1 to pg 1_1 range. Conclusions Based on a previously described sampler (MESCO)16 a new passive sampler was designed which has on one hand the advantages of the earlier one, and overcomes its weakness (the low chemical and thermal stability as well as biodegradability of the dialysis membrane from regenerated cellulose) on the other hand. The membrane was substituted by a stable, in the SPMD technique successfully applied, LDPE membrane. Moreover, the stir bars used as receiving phase were substituted by less expensive PDMS materials (tubes and rods), which enabled additionally a significant increase of the PDMS volume and thus the accumulation capacity. The study of the PDMS materials regarding reproducibility and completeness of enrichment and thermodesorption yielded in comparable good results of tubes and stir bars. The investigation of the capability of three versions of the sampler (water-filled tube and rod sampler as well as air-filled tube sampler) resulted in the new samplers enabling the effective accumulation of the POPs under study and thus the estimation of low TWA concentrations of these water pollutants. The first comparison of samplers which differ only in the filling medium (water and air, respectively) was done, to our knowledge. This resulted in a significant increase of the sampling rates of most of the analytes and thus in enhanced sensitivities for the air-filled sampler. This finding could be confirmed by calculation of the sampling rates based on physico-chemical parameters. The new samplers are stable in field exposure (as tested in on-site experiments) and enable longer exposure times compared with the MESCO sampler because of their enlarged accumulation capacity. However, there is a lack in efficient sampling of analytes with larger molecular size, such as PCBs and PAHs with high molecular weights because of the application of the non-porous LDPE membranes. Acknowledgements This work was kindly funded by the Ministry of Education and the Arts of Saxony-Anhalt (Germany). References 1 R. H. Brown, /. Environ. Monit, 2000, 2, 1. 2 B. Zabiegala, A. Kot and J. Namiesnik, Chem. Anal. [Warsaw], 2000, 45, 645. 3 A. Kot, B. Zabiegala and J. Namiesnik, /. Trends Anal. Chem., 2000, 19, 446. 4 T. Gorecki and J. Namiesnik, /. Trends Anal. Chem., 2002, 21, 276. 5 J. N. Huckins, M. W. Tubergen and G. K. Manuweera, Chemo-sphere, 1990, 20, 533. 6 J. D. Petty, J. N. Huckins, C. E. Orazio, J. A. Lebo, B. C. Poulton, R. W. Gale, C. S. Charbonneau and E. M. Kaiser, Environ. Sei. Technol, 1995, 29, 2561. 7 H. F. Prest, W. M. Jarman, S. A. Burns, T. Weismüller, M. Martin and J. N. Huckins, Chemosphere, 1992, 25, 1811. 8 J. A. Lebo, R. W. Gale, D. E. Tillit, J. N. Huckins, J. C. Meadows, C. E. Orazio andD. J. Schroeder, Environ. Sei. Technol, 1995, 29, 2886. 9 J. A. Lebo, J. L. Zajicek, J. N. Huckins, J. D. Petty and P. H. Peterman, Chemosphere, 1992, 25, 697. 10 J. D. Petty, S. B. Jones, J. N. Huckins, W. L. Cranor, J. T. Parris, T. B. McTague and T. B. Boyle, Chemosphere, 2000, 41, 311. 11 R. D. Blanchard and J. K. Hardy, Anal. Chem., 1985, 57, 2349. 12 G.-Z. Zhang and J. K. Hardy, /. Environ. Sei. Health, 1989, A24, 279. 13 H. L. Lee and J. K. Hardy, Int. J. Environ. Anal. Chem., 1998, 72, 83. 14 P. Grathwohl, Patent No. 19824 082.1, Deutsches Patentamt München, 1998. 15 H. Martin, Entwicklung von Passivsammlern zum zeitlich integrierenden Depositions- und Grundwassermonitoring: Adsorberkar-tuschen und Keramikdosimeter, PhD Thesis, University of Tübingen, Germany, 2000. 16 B. Vrana, P. Popp, A. Paschke and G. Schüürmann, Anal.Chem., 2001, 73, 5191. 17 E. Baltussen, P. Sandra, F. David and C. Cramers, /. Microcolumn Sep., 1999, 11, 737. 18 E. Baltussen, C. A. Cramers and P. J. F. Sandra, Anal. Bioanal. Chem., 2002, 373, 3. 19 G. D. Johnson, Environ. Sei. Technol., 1991, 255, 1897. 20 J. N. Huckins, G. K. Manuweera, J. D. Petty, D. Mackay and J. A. Lebo, Environ. Sei. Technol., 1993, 27, 2489. 21 J. N. Huckins, J. D. Petty, C. E. Orazio, J. A. Lebo, R. C. Clark, /. Environ. Monit., 2003, 5, 813-822 821 V. L. Gibson, W. R. Gala and K. R. Echols, Environ. Sei. Technol., 1999, 33, 3918. 22 Illustrated handbook of physical-chemical properties of environmental fate of organic chemicals, eds. D. Mackay and W. Y. Shiu, Lewis Publishers, Boca Raton, FL, 1992, Vol. 1. 23 Illustrated handbook of physical-chemical properties of environmental fate of organic chemicals, eds. D. Mackay, W. Y. Shiu and K.C. Ma, Lewis Publishers, Boca Raton, FL, 1992, Vol. 2. 24 W. J. Lyman, in Handbook of Chemical Property Estimation Methods, Environmental Behavior of Organic Compounds, eds. W. J. Lyman, W. F. Riehl and D. H. Rosenblatt, McGraw-Hill Book Company, New York, 1982. 25 G. A. Lugg, Anal. Chem., 1968, 40, 1072. 26 B. Vrana and G. Schüürmann, Environ. Sei. Technol., 2002, 36, 290. 27 K. Booij, H. M. Sleiderink and F. Smedes, Toxicol. Chem., 1998, 17, 1236. 28 H. E. Hofmans, Numerical Modeling of the Exchange Kinetics of Semipermeable Membrane Devices, Master Thesis, University of Utrecht, Netherlands Institute for Sea Research, Den Burg, The Netherlands, 1998. 29 J. Crank, The Mathematics of Diffusion, 2nd edn., Oxford University Press, New York, 1975, Chapter 4. 30 R. J. Scheuplein, /. Theor. Biol, 1968, 18, 72. 31 G. L. Flynn and S. H. Yalkowsky, /. Pharm. Sei., 1972, 61, 838. 32 W. A. Tucker and L. H. Nelken, Diffusion coefficients in air and water, in Handbook of Chemical Property Estimation Methods, Environmental Behavior of Organic Compounds, eds. W. J. Lyman, W. F. Riehl and D. H. Rosenblatt, McGraw-Hill Book Company, New York, 1982, pp. (2-l)-(2-52). 33 S. T. Hwang and K. Kammermyer, Membranes in Separations. Robert E. Krieger Publishing Company, Malabar, FL, 1984. 34 J. N. Huckins, J. D. Petty, H. F. Prest, R. C. Clark, D. A. Alvarez, C. E. Orazio, J. A. Lebo, W. L. Cranor and B. T. Johnson, A guide for the use of semipermeable membrane devices (SPMDs) as samplers, API 4690, Columbia, 2000. 35 R. Doong and S. Chang, Anal. Chem., 2000, 72, 3647. 36 I. Valor, M. Perez, C. Cortada, D. Apraiz, C. J. Molto and G. Font, /. Sep. Sei, 2001, 24, 39. 822 /. Environ. Monit., 2003, 5, 813-822 Príloha 6 Vrana B., Greenwood R., Mills G., Knutsson J., Svensson K., and Morrison G., Performance optimisation of a passive sampler for monitoring hydrophobic organic pollutants in water, J. Environ. Monit., 2005, 7, 612-620. Performance optimisation of a passive sampler for monitoring hydrophobic organic pollutants in water Branislav Vrana,*a Graham Mills,6 Richard Greenwood," Jesper Knutsson,c Katarina Svenssonc and Gregory Morrisonc a School of Biological Sciences, University of Portsmouth, King Henry Building, King Henry I Street, Portsmouth, POl 2DY, UK. E-mail: bran.vrana@port.ac.uk; Fax: ++44/23 9284 2070; Tel: ++44/ 23 9284 2024 b School of Pharmacy and Biomedical Sciences, University of Portsmouth, White Swan Road, Portsmouth, POl 2DT, UK c Water Environment Transport, Chalmers University of Technology, SE-412 96 Goteborg, Sweden Received 20th December 2004, Accepted 11th April 2005 First published as an Advance Article on the web 29th April 2005 The performance of an integrative passive sampler that consists of a C18 Empore disk sorbent receiving phase fitted with low density polyethylene membrane was optimised for the measurement of time-weighted average concentrations of hydrophobic micropollutants in water. A substantial improvement of sampling characteristics including the rate of sampling and the sampling precision was achieved by decreasing the internal sampler resistance to mass transfer of hydrophobic organic chemicals. This was achieved by adding a small volume of n-octanol, a solvent with high permeability (solubility x diffusivity) for target analytes, to the interstial space between the receiving sorbent phase and the polyethylene diffusion-limiting membrane. Introduction There is an increasing requirement for monitoring the water quality across Europe, with particular emphasis on the contaminants in the list of priority pollutants in the Water Framework Directive (WFD) and in various water conventions, e.g. the Convention for the Protection of the Marine Environment of the North-East Atlantic (OSPAR). Among priority pollutants, persistent organic pollutants (POPs), such as organo-chlorine pesticides, polychlorinated biphenyls, and polycyclic aromatic hydrocarbons (PAHs) are of great importance. Due to their low aqueous solubilities and hydrophobic nature, the concentrations of POPs dissolved in water are very low, usually < 1 ppb. POPs associate easily with particulate matter and are finally deposited in the sediment. The fraction of the chemical truly dissolved in water is very small. Nevertheless, because organisms often bioconcentrate these low levels of contaminants in water to relatively high levels in their tissues, determination of the dissolved portion of environmental pollutants is critical for assessing the potential for detrimental biological impacts. The only monitoring method legally accepted for this purpose is spot sampling. This is both expensive and labour intensive and measures only instantaneous concentrations, which may not be representative of long-term average pollutant concentrations. There are a number of methods that attempt to overcome these problems, e.g. on-line continuous monitoring, biomonitoring or passive sampling.1 Among these methods passive sampling technology has the potential to become a reliable, robust, and cost-effective tool, that could be used in monitoring programmes across Europe.2'3 Integrative passive sampling involves the measurement of the concentration of an analyte as a weighted function of the g time of sampling.2 Ideally, the sampling device acts as an 5 infinite sink for contaminants of interest and the uptake of H analytes is time proportional. The use of passive sampling ° methods to monitor POPs has increased greatly over the past ° decade. Much has been published on the use of semipermeable 5 membrane devices (SPMD) for evaluating ultra-low concents trations of hydrophobic contaminants.4-7 Although SPMDs are widely used and very sensitive for assessment of waterborne POPs, laborious and time-consuming separation of lipid matrix components from target analytes using solvent dialysis and size exclusion chromatography is required.8 We developed previously a novel passive sampling system for the measurement of time-weighted average (TWA) concentrations of micropollutants in aquatic environments.9'10 The system is based on the diffusion of target compounds through a membrane and the subsequent accumulation of these pollutants in a bound, solid receiving phase. Accumulation rates and selectivity are regulated by the choice of both the diffusion-limiting membrane and solid-phase receiving material. One of the prototypes was designed for the sampling of non-polar organic compounds with log octanol/water partition coefficient (log Kovl) values greater than 3.9 This system used a 47 mm C18 Empore® disk as the receiving phase and a low-density polyethylene (LDPE) diffusion-limiting membrane. The LDPE is a nonporous material with no fixed pores, only transient cavities with a typical size of 1 nm. This solute size limitation excludes large molecules as well as those that are adsorbed on colloids or humic acids. Only truly dissolved and non-ionised contaminants diffuse through the LDPE membrane and can be sequestered by the sampler. The C18 Empore® disk has a very high affinity and capacity for the sampled non-polar organic pollutants. For a good sampler performance, a sufficiently high sampling rate is essential, i.e. the rate at which the sampler accumulates chemicals from water, usually expressed as volume of water cleared of a chemical per unit of time (e.g. L d_1). High sampling rates are needed, especially for non-polar chemicals due to their low concentrations in the water column. The sampling rate depends on the physicochemical properties of the analyte, the environmental conditions and the sampler design.2 An optimal sampler design can be achieved in two ways: by maximising the surface area of the sampler; i.e. the area through which the analytes are accumulated in the receiving phase, and by minimising the resistance of mass 6 1 2 J. Environ. Mo nit., 2005, 7, 612-620 This journal is © The Royal Society of Chemistry 2005 transfer across the various phases for the analytes being measured. The Empore® disk used for the receiving phase9 in our sampler design consists of octadecyl (CI8) bonded silica stationary phase particles, immobilised by polytetraffuoroethyl-ene (PTFE) fibrils. The disk presents a porous medium with a total porosity (inter-particle and intra-particle) of 0.52.11 When the LDPE diffusion-limiting membrane is placed over the C18 disk a thin layer of air, or in some cases water, remains trapped between the inner surface of the membrane and the adsorbing surface of the C18 silica particles embedded inside the PTFE disk. This layer of air or water both fills the pores in the Empore disk and forms a thin macroscopic film that fills the gap between the surfaces of receiving phase disk and diffusion-limiting membrane. The analytes taken up by this design of sampler by diffusion across the surface ofthe LDPE membrane are hydrophobic; air and water are media with very low permeability (solubility x diffusivity) for most of these chemicals. This layer trapped inside the sampler acts as (or represents) a significant additional barrier to mass transfer and potentially reduces the sampling rate ofthe analytes of interest. The aim of this study was to improve the performance ofthe passive sampler by minimizing its internal resistance to obtain higher sampling rates that are required for the measurement of low concentrations of non-polar organic pollutants. The effect of various gap-filling fluids (i.e. air, water and n-octanol) on the performance of the passive sampler was evaluated for a number of PAHs. These are non-polar compounds with a range of physicochemical properties (Table 1) and thereby provide a convenient test set of compounds. Theory The mass transfer of a given chemical through a passive sampling device includes several diffusion and interfacial mass transport steps across the different barriers that maybe present i.e. the stagnant aqueous boundary layer, possible bio-film layer, the diffusion-limiting membrane, the inner fluid (aqueous or gaseous) phase, and the receiving phase (Fig. 1). In the initial exposure phase, analyte uptake is expected to be linear or time-integrative after steady state flux of chemicals into the sampler has been achieved.4'12 Under these conditions the amount of a chemical in the receiving phase is directly proportional to the product of the concentration in the surrounding water (Cw) and the exposure time (t). For practical purposes, uptake in the linear phase can be described by: Passive sampling device Ms(t) = M0 + CwRst (1) chemical in water where Ms is the amount of analyte accumulated in the receiving phase, M0 is the initial amount of analyte in the receiving Fig. 1 Profile of the passive sampling device showing the barriers to analyte mass transfer into the sampler. phase, and Rs is the sampling rate of the system: Rs = kmA (2) where kov [m s~'] is the overall mass transfer coefficient and A [m2] is the surface area of the membrane. The uptake of an analyte is linear and integrative approximately until the concentration factor of the sampler (ratio Ms(7)/Cw) reaches half-saturation. The sampling rate of an individual chemical can be determined experimentally under fixed conditions at constant analyte concentration. Under environmental conditions, when the water concentration changes during the exposure, the term Cw represents a TWA concentration during the deployment period. Materials and methods Physicochemical properties of substances Henry's Law constants (KAW) at 25 °C, of substances under investigation were taken from literature.13'14 Nearly equal values of aqueous diffusion coefficients (Dw) were estimated for the test compounds, ranging from 5 x 10~6 to 6 x 10~6 cm2 s~V5 Diffusion coefficients of the test analytes in air, DA, at 20 °C ranged from 0.05 to 0.06 cm2 s-1.16 An approximated value of 10~10 cm2 s_1 was used as diffusion coefficient (DM) of the analytes in the LDPE membrane.17 Diffusion coefficients of test analytes in n-octanol (D0) of 7 x 10~7 cm2 s_1 were calculated from the assumption that the Da values are lower Table 1 Selected physicochemical properties of test analytes at 20 °C Sc Da" £>w xl06/ DQ x 10 Compound g moP1 log C girT3 cm2 s_1 cm2 s_1 cm2 s_1 Acenaphthene 154.2 4.0 3.8 4.9 x io-3 0.063 6.3 7.5 Fluorene 166.2 4.2 1.9 3.2 x io-3 0.06 6.0 7.2 Anthracene 178.2 4.6 0.045 1.3 x io-3 0.058 5.9 6.9 Phenanthrene 178.2 4.5 1.10 1.6 x io-3 0.058 5.9 7.0 Fluoranthene 202.3 5.1 0.26 4.2 x io-4 0.055 5.5 6.6 Pyrene 202.3 5.1 0.132 3.7 x io-4 0.055 5.6 6.6 Benzo[a]anthracene 228.3 5.9 0.011 2.3 x io-4 0.052 5.1 6.1 Chrysene 228.3 5.7 0.0019 2.6 x io-4 0.052 5.1 6.0 Benzo[/j]fiuoranthene 252.3 5.8 0.0015 3.4 x io-5 0.05 5.0 6.0 Benzo[fc]fiuoranthene 252.3 6.0 0.0008 3.4 x io-5 0.05 5.0 6.0 Benzo[a]pyrene 252.3 6.2 0.0038 4.6 x io-5 0.05 5.0 6.0 " Molecular weight (Mw). b n-Octanol/water partition coefficient Kovl. c Aqueous solubility S. d Dimensionless Henry's Law constant. e Da diffusion coefficient in air. f Dw diffusion coefficient in water. s Do diffusion coefficient in n-octanol. J. Environ. Mo nit., 2005, 7, 612-620 6 1 3 than those in water because of the higher viscosity of 1-octanol (7.49 cP at 25 °C) compared with water (0.89 cP at 25 °C).18 The LDPE membrane/water partition coefficients (KMW) were estimated from Hofman's predictive equation.19 The values of physicochemical properties are summarised in Table 1. Materials and chemicals C18 Empore® disks (47 mm diameter) were purchased from Varian Inc., Walton-on-Thames, UK. LDPE membrane material (40 urn thickness) was obtained from Fisher Scientific, Loughborough, UK. The solvents (HPLC grade quality or equivalent), acetone; n-hexane; 2,2,4-trimethyl pentane; ethyl acetate; n-octanol; n-nonane; methanol and water were obtained from Fisher Scientific. Certified pure (purity >98% in all cases) reference standards of the test compounds; surrogates, and internal standards were obtained from Qmx Laboratories, Saffron Walden, UK. Certified external calibration solutions of target analyte mixtures at a concentration of 10 iig mL-1 in cyclohexane were obtained from Qmx Laboratories. Sampler design The patented design of the passive sampler has been described previously.9'20 Briefly, the sampling device consisted of a PTFE body containing a C18 Empore disk as a receiving phase. A 40 urn thick LDPE disk (47 mm diameter) of diffusion-limiting membrane was placed on the top of the receiving phase. The PTFE body supported both the receiving phase and the diffusion-limiting membrane and sealed them in place. Three variants of sampler design were tested in this study. They differed only in the composition of the medium, filling the space between the receiving phase and the LDPE membrane: air (variant 1), water (variant 2) or n-octanol (variant 3). The effect of the filling medium on performance (sampling rate and sampling precision) of the sampler was evaluated. Preparation of the samplers C18 Empore® disks were conditioned by soaking them in methanol for 20 min until translucent and then stored in methanol until required. The Empore® disks were prepared in a 47 mm diameter disk vacuum manifold platform (Varian Inc.). Methanol (10 mL) was slowly passed through the disk, followed by 20 mL ultrapure water. 500 mL of 5 |ig L_1 aqueous solution of D12-benzo(a)anthracene (internal standard) was filtered through the disk. The subsequent treatment of the disks differed for the three sampler variants: Variant 1 A vacuum was applied for 30 min to ensure that the disk was completely dry at the end of the procedure. Variant 2 The filtration procedure was stopped immediately before the last portion of the 500 mL aqueous internal standard solution passed the disk. It was assured that the disk remained saturated with water after this procedure and the disk did not dry out during any of the preparation steps. To prevent the disk from drying between conditioning and exposure in the flow-through test system, the devices were loaded immediately before deployment. Variant 3 A vacuum was applied for 30 min to ensure that the disc was completely dry. The Empore® disk was then put on the sampler PTFE support disk. 1 mL solution of n-octanol in acetone (45% v/v) was applied. The acetone was allowed to evaporate from the disk for 10 min in the fume hood. The resulting volume of applied n-octanol was 450 uL. The final preparation step was the same for all sampler variants. The LDPE membrane was put on the top of the Empore® disk. Prior to sampler assembly, the LDPE membranes were pre-cleaned by soaking for 24 h in n-hexane and dried. Any air bubbles were smoothed away from between the two layers by gently pressing the top surface of the membrane using a clean paper tissue. The PTFE supporting disk was placed into the sampler body and fixed in place to form a watertight seal between the membrane and the top section of the sampler. Exposure experiments In each experiment up to 16 passive samplers were exposed in a constant concentration flow-through exposure system. A nominal concentration of 100 ng L_1 for each test analyte was maintained throughout the experiment. The configuration of the flow-through exposure system has been described.9 Briefly, it consisted of a 20 L glass tank with an overflow to waste. Water was fed to the exposure tank using a peristaltic pump at 2 L h_1. Test chemicals were dissolved in methanol (30 ug L_1) and the appropriate amounts of stock solution (100 uL min-1) were delivered into the exposure tank using a small peristaltic pump. The water in the chamber was mixed using an overhead stirrer. The resulting methanol concentration in the exposure water did not exceed 0.5% (v/v). To allow for the sorption of chemicals to exposed surfaces (e.g. glass walls of the tank), the system was allowed to equilibrate with the test solution for 48 h before samplers were deployed. The passive samplers were placed at the bottom of the exposure tank. Exposures were conducted at 11 °C. The exposures lasted 14 days, during which duplicate samplers were sampled at set time intervals and analysed (see below) to determine the concentrations of accumulated test chemicals. Duplicate samples (500 mL each) of water, sampled from the outlet of the exposure tank, were also taken each time the samplers were removed, and the concentration of test analyte in the water was determined. The experimental conditions of individual exposures are given in Table 2. Extraction of analytes from the passive samplers After exposure, the sampler was carefully disassembled and the LDPE membrane removed and rinsed with 5 mL acetone. Compounds were extracted from the Empore® disks in an ultrasonic bath (5 min) using acetone (5 mL) followed by 5 min in 50 : 50 (v/v) ethyl acetate: 2,2,4-trimethylpentane (5 mL). The disks were removed, the solvent extracts combined with the LDPE membrane rinsate and filtered through a drying cartridge containing 1 g of sodium sulfate (Varian Inc.). In the case where no «-octanol was used in the sampler construction, 100 uL of n-nonane was added to the extract to act as a solvent keeper. The solvent extract was gradually Table 2 Summary of flow-through exposure conditions" used for the different designs of passive sampler Sampler Permeation Exposure Number of variant medium6 period/h samplers 1 Air 0-336 16 2 Water 0-288 11 3 n-Octanol 0-284 15 " The nominal concentration of analytes in water was 100 ng LT1 at 11 °C. b The medium filling the gap between the Empore® receiving disk and the LDPE diffusion-limiting membrane. 6 1 4 J. Environ. Mo nit., 2005, 7, 612-620 reduced in volume under nitrogen to approximately 100 |iL, 800 |iL of «-hexane was added and transferred to a 2 mL vial for analysis. 100 |iL of 10 ng uL-1 solution of D10-anthracene in «-hexane was added as an internal standard. The final volume was adjusted to 1 mL with «-hexane. When n-octanol was used in sampler conditioning, the extract was gently reduced under nitrogen. Approximately 450 |iL of extract in n-octanol remained after this preparation step (n-octanol has a very low volatility). The reduced extract was transferred to 2 mL vials prior to analysis. 50 uL of the 10 ng uL-1 internal standard solution of D10-anthracene in n-octanol was added. The final volume was adjusted to 500 uL with n-octanol. In all cases, the percentage recovery of the test compounds from the C18 Empore® disks was between 95 and 100%. Extraction of analytes from water The test analytes in water samples taken from the outlet of the flow-through exposure system were extracted using solid-phase extraction (SPE) on Bondelut C18 LO SPE cartridges (3 mL/ 200 mg sorbent; Varian Inc.). The sorbent was first activated by the passage of 2 mL methanol followed by 10 mL doubly-distilled water through the bed. The water sample (500 mL) was passed through the cartridge at 30 mL min-1 using low-pressure. After the entire water sample has passed through the cartridge, the sorbent was dried by aspirating air through the bed. Extracted analytes were eluted with 1 mL «-hexane. 50 uL of internal standard (10 ng uL-1 D10-anthracene in «-hexane) was added prior to analysis. Water was spiked at multiple concentrations to estimate the recoveries of the test compounds using the SPE procedure. A procedural blank and four recovery solutions (20-100 ng L_1) were extracted concurrently with the water samples. The recovery standards were analysed alongside spiking standards and a mean percentage recovery for the four spiking concentrations was calculated. SPE recoveries of the test compounds were between 80 and 95%. Extraction of analytes from the PTFE sampler body To check the potential analyte adsorption to the PTFE material of the sampler, PAHs accumulated in a sampler body were extracted after 284 h of exposure in experiment 3. The sampler body was extracted in an ultrasonic bath (15 min) using acetone (200 mL). The extraction step was repeated twice. The extracts were combined and 100 uL of «-nonane was added. The extract was gradually reduced in volume under nitrogen to approximately 100 uL, 800 uL of «-hexane was added and transferred to a 2 mL GC vial for analysis. 100 uL of 10 ng uL-1 solution of D10-anthracene in «-hexane was added as an internal standard. The final volume was adjusted to 1 mL with «-hexane. Instrumental analysis The concentrations of all target analytes accumulated in samplers during the exposure studies were quantified using GC/MS. Analysis was performed with a 6890A series GC equipped with a mass-selective detector 5973 (Agilent Technologies, Bracknell, UK). GC/MS analysis of n-hexane sampler extracts (variants 1 and 2) and exposure tank water extracts The sampler extract (1 uL) was injected into the GC/MS system. Injections were carried out in splitless mode at an inlet temperature of 250 °C. The injector was coupled to a 30 m x 0.25 mm id, 0.25 urn film HP-5 MS capillary column (Varian Inc.). Helium was used as carrier gas at a column flow rate of 2 mL min The GC oven temperature programme was 60 °C (2 min) and then increased at 30 °C min~1 to 150 °C and then at 6 °C min-1 to 280 °C (5 min). Quantification of the test analytes was accomplished using a 7-point external calibration curve. All external standards were prepared in « -hexane. GC/MS analysis of n-octanol sampler extracts (variant 3) The sampler extract (1 uL) was injected into the GC/MS system. Injections were carried out in pulsed splitless mode at an inlet temperature of 250 °C. The injector was coupled to a methyl deactivated fused silica retention gap (2.5 m x 0.25 mm id) The other end ofthe retention gap was connected to a 30 m x 0.25 mm id, 0.25 urn film HP-5 MS capillary column (Varian Inc.). The pulse pressure was 50 psi for 2 min. Helium was used as carrier gas at a column flow rate of 2 mL min-1. The GC oven temperature programme was 120 °C (2 min) and then increased at 6 °C min-1 to 300 °C (5 min). Quantification ofthe test analytes was accomplished using a 7-point external calibration curve. All external standards were prepared in n-octanol. MS parameters The MS parameters for both GC methods were: interface temperature 280 °C, ion source temperature 250 °C, electron impact (EI) ionization mode at 70 eV. Analysis was performed by selected ion monitoring (SIM) applying two or three characteristic ions for each compound in both detection and quantification. Data processing The experimental time course accumulation rates of individual test substances on the Empore® disks were fitted by linear regression analysis using eqn. (1). The adjustable parameters were the intercept (M0) and the slope (Cw x Rs) of the uptake curve Ms = f(t). Quality of the fit was characterized by the standard deviations of the optimised parameters, as well as the correlation coefficient adjusted for the degrees of freedom (r2 adjusted), the fit standard deviation, and the Fisher test criterion on the accuracy of the model. The sampling rates Rs for individual test compounds were calculated by dividing the slope of the linear uptake curve by the mean aqueous analyte concentration during the exposure period. The required variances of Rs values were calculated from the coefficients of variation (relative standard deviations) of the uptake slope parameters and the concentrations in the aqueous phase, which were obtained according to the law of error propagation. Results and discussion Flow-through exposures The performance of the three sampler design variants was tested by exposure to constant concentrations of test chemicals in a continuous flow-exposure tank. Concentrations of the analytes in water (Cw) and the amounts accumulated in the receiving phase (Ms) were two parameters measured regularly during the continuous flow-exposures. During exposure the concentrations of the test compounds in water were held constant, and this was confirmed by regular analysis of water samples. Characteristic analyte uptake curves for the sampler variant 3 are shown in Fig. 2. Variant 1 (air-filled sampler) Satisfactory fits of the exposure data using eqn. (1) were obtained for all compounds (P < 0.05) excepting benzo[6]-fluoranthene, benzo[fc]fluoranthene and benzo[a]pyrene. For J. Environ. Mo nit., 2005, 7, 612-620 6 1 5 1600' 1200 0) T3 O) 1000 o 800 E 600 E =3 O 400 O < 200- A O Acenaphthene A Phenanthrene x ▲ Fluoranthene • Chrysene X A ▲ Á A.---" A'' 0 X ______________.1 X""' •■ •—-•— 100 150 Time [h] 200 250 300 Fig. 2 Uptake of selected PAHs and by the passive sampler variant 3, where the pores in the receiving phase were filled with n-octanol. The data used represent the 11 °C flow-through exposure at 100 ng L_1 nominal water concentration of each analyte. The lines are predicted concentrations in the sampler obtained by linear regression using eqn. (1). these three compounds, no significant uptake was observed. Correlation coefficient (r2 adjusted) values of the regression (model versus experimental) of the satisfactory fits ranged from 0.66 (acenaphthene) to 0.76 (fluoranthene). Coefficients of variation of the calculated sampling rate did not exceed 43% in any case, excepting acenaphthene (57%). Variant 2 (water-filled sampler) Satisfactory fits of the exposure data using eqn. (1) were obtained only for phenanthrene, anthracene, fluoranthene and pyrene (P < 0.05). For the rest of the PAH compounds, deviation of the data from linear uptake was observed. No significant uptake was observed for acenaphthene, fluorene, benzo[6]fluoranthene, benzo[fc]fluoranthene and benzo[a] pyrene. Correlation coefficient (r2 adjusted) values of the regression (model versus experimental) of the satisfactory fits ranged from 0.36 (anthracene) to 0.65 (fluoranthene). Coefficients of variation of the calculated sampling rate did not exceed 44% for any of these compounds. Variant 3 (n-octanol-filled sampler) Satisfactory fits of the exposure data using eqn. (1) were obtained for all test compounds. For all analytes the uptake was linear (P < 0.05) during the whole exposure period, without any sign of a levelling-off (reaching equilibrium). Correlation coefficient (r2 adjusted) values of the regression (model versus experimental) ranged from 0.72 (benzo[a]pyrene) to 0.96 (acenaphthene). Coefficients of variation of the calculated sampling rate did not exceed 32% in any case. The maximum fluctuations of water concentrations during exposures did not exceed 30% of the mean concentration for individual compounds. Performance comparison of the three sampler variants The sampling rates determined for the three sampler variants are shown in Fig. 3. The highest sampling rates (Rs up to 0.19 L d~') were determined for the sampling device filled with n-octanol (variant 3). The sampling rates obtained with the water-filled sampler (variant 2) were the lowest, except for 0 300 0 250 " 0 200 =1 0 150 a. o 100 0 050 0000 □ water □ air ■ octanol T T T \ \ f 1 Tl M h 1 ix» . 1 1 J J a . . //// / * y.x^ Compound Fig. 3 Sampling rates Rs of polycyclic aromatic hydrocarbons determined for the three passive sampler designs. The passive samplers consist of a receiving phase (47 mm C18 Empore® disk) fitted with a 40 |rm thick low-density polyethylene membrane. The pores in the receiving phase were filled with different media: water, air or n-octanol. Ra data were derived from 14-day flow-through exposures at 11 °C at the nominal analyte concentration of 100 ng L_1. benzo[a]anthracene and chrysene. For these two compounds, the uptake rates were comparable with the other two sampler variants. The sampler performance can be ranked from the slowest to the fastest as follows: water-filled sampler (variant 2) < air-filled sampler (variant 1) < n-octanol-filled sampler (variant 3). In general, sampling rates were lower for very hydrophobic PAHs with high molecular weight. The uptake of larger ringed PAHs (benzo[6]fluoranthene, benzo[fc]fluor-anthene and benzo[a]-pyrene) into the air and water-filled variant was so slow, that sampling rates could not be measured. A hydrophobicity profile (with an optimum at log Kow x 4.5) of sampling rates was observed for the n-octanol-filled device (variant 3). A similar hydrophobicity profile was described by Huckins et al. for lipid-filled SPMDs.21 The best precision of the calculated sampling rates was obtained for the n-octanol-filled sampler. The worst precision, accompanied by non-linear sampling behaviour, was observed for the water-filled sampler. Theoretical examination of the mass transfer To explain the observed differences in performance of the three sampler variants, a theoretical examination of the mass transfer processes involved was undertaken. The individual mass transfer layers through which a chemical must diffuse to reach the receiving phase are shown in Fig. 1. The overall mass transfer coefficient is affected by the diffusion of solutes in the individual layers (i.e. aqueous boundary layer, diffusion-limiting membrane, the inner transfer medium [air, water or n-octanol] and the receiving phase) and by their partitioning into the LDPE membrane and receiving phase; since accumulation of hydrophobic analytes is expected also in the membrane material.21 Moreover, accumulation is expected also in the n-octanol layer in sampler variant 3. From theory22'23 it is assumed that the overall mass transfer resistance (l/fcov), to the uptake of a chemical is given by the sum of particular barrier resistances to mass transfer: 1 (3) where 5t is the particular barrier thickness, Dt is the diffusion coefficient in the particular barrier and Kiw is the partition coefficient between the j'-th phase and water. The overall resistance (l/fcov) is then given by: 1 Dw Dm-Kmw DaKtw DsKsw (4) The subscripts B, M, W and S represent the boundary aqueous layer at the surface of the sampler [B], the membrane [M], the 6 1 6 J. Environ. Mo nit., 2005, 7, 612-620 inner transfer layer (air, water or «-octanol) [T] and the receiving phase [S]. The group DMKMW is the commonly used membrane permeability coefficient. In case where the medium that fills the space between the receiving phase and the membrane is air, water or «-octanol, the partition coefficient KTW in eqn. (4) stands for the dimensionless Henry's Law constant KAW, the unity (the number 1) or the octanol/water partition coefficient Kovl, respectively. Eqns. (3) and (4) show that an individual resistance increases with the increasing thickness of the barrier and with decreases in the diffusion and partition coefficients, respectively. A layer with more than 50% of the total resistance is considered rate-limiting.25 It is likely that the differences in the sampling rates for similar (but differing from each other in the composition of the fluid between the membrane and the receiving phase) sampler designs are caused by differences in the partial resistance to mass transfer of the type of interstitial fluid. The diffusion path of analyte molecules through the inner transfer medium is approximately the same for all sampler designs, because all three sampler designs have the same geometry. For all sampler designs, the mass transfer by convection in the inner transfer medium is assumed to be negligible. Thus, the difference in sampling rates is likely to originate in unequal inner transfer medium permeability of the individual sampler designs. To examine the effect of the inner transfer medium of the sampler on the mass transfer, an estimate of the magnitude of the resistance of this layer was made. This resistance (l/kT) was calculated as: 1 (5) The estimated values of (l/fcT) are shown in Fig. 4 together with the estimated resistance l/kM of the LDPE membrane, which was calculated as: 1 (6) The calculation was made using the available physicochemical properties of the test analytes including diffusion and partition coefficients (Table 1). The thickness of the transfer medium <5T was set to be approximately 1 mm (depth of pores in an Empore disk) and the thickness of the LDPE membrane <5M was 40 |im. A theoretical examination shows that the resistance to mass transfer of the water or air layer is expected to be several orders □ LDPE □ air □ water ■ octanol 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01 1.E+00 ///yv// Fig. Compound 4 Estimate of various individual barrier resistances (LDPE membrane and inner transfer media air, water and n-octanol) to mass transfer inside a passive sampler. The calculations were made using eqns. (5) and (6). of magnitude higher than that of the «-octanol layer. Despite the fact that the diffusion coefficients of the analytes in water and air are much higher than those in «-octanol, the permeability of these media is low due to very low solubility or vapour pressure of the PAHs in them. Moreover, the resistance of water and air is expected to increase with increasing the molecular weight of the analytes, whereas the resistance in «-octanol will decrease. This reduced internal resistance to uptake becomes significant when sampling (and successfully detecting) large ringed PAHs, which are present (in the dissolved form) only at trace levels in the aquatic environment. Furthermore, the resistance to mass transfer of the water layer is expected to be comparable to or even higher than the resistance of the LDPE membrane. Thus, in sampler variant 2, the water filling the pores of the Empore® disk is likely to be the rate-limiting barrier to the uptake of chemicals. Although the resistance of the air layer is expected to be lower than that of LDPE membrane for smaller PAHs (up to three aromatic rings), it is expected to dramatically increase for larger PAHs. The inner air layer is then expected to take control over the sampling kinetics in the variant 1 of the sampler. From theory,23 the individual resistances to mass transfer are additive. Fig. 5 shows sampling rates estimated (using eqn. (3)) theoretically for a combined mass transfer through the LDPE membrane and the inner transfer medium. The theoretical uptake scenario for the air (variant 1) and water-filled (variant 2) sampler designs estimate the maximum achievable values of sampling rates of 1.4 and 0.3 L d_1, respectively. In contrast, the predicted sampling rates for the «-octanol-filled sampler (variant 3) are much higher (up to 30 L d_1). Significant differences in favour of the «-octanol variant (3) are predicted especially for the larger ringed PAHs. Experimental sampling rates vs. theoretical predictions All experimentally determined sampling rates were lower than 26% of the theoretically calculated values. To compare the theoretically estimated kinetic parameters with the experimentally measured values, it is necessary to consider the simplifications made in the above calculations. Firstly, only the mass transfer of chemicals in physical sampler components (membranes and well-defined layers of fluid) was discussed. However, the first and very important step in uptake is the diffusion through the stagnant aqueous boundary layer at the surface of the membrane. This presents an additional barrier for the uptake of chemicals in the 30.0 25.0 20,0 ■15 C 10 0 5,0 1/ jo/ II 1 5 0.5 0 c / i «N ^ {J& Compound Fig. 5 Estimate of sampling rates (iy of the target analytes calculated for mass transfer through the LDPE membrane and the inner transfer medium layer of air, water or n-octanol using eqn. (4). The resistance of aqueous boundary layer at the outer surface of the sampler was not considered in the calculation. J. Environ. Monit., 2005, 7, 612-620 6 1 7 8874 sampler. The thickness of the aqueous boundary layer varies with exposure conditions; increased water turbulence during the sampling period reduces the thickness of the boundary layer and thus its resistance to mass transfer. For hydrophobic chemicals (with high values of KMW, KTW or Ksw) the aqueous boundary layer may become the rate-limiting barrier of the whole sampling process.24 Our ongoing calibration experiments with sampler variant 3 at varying exposure conditions have demonstrated that the sampling rates of PAHs can be increased by more than one order of magnitude by increasing the water turbulence around the face of the sampler.25 This confirms that the uptake of these compounds is indeed governed by diffusion across the aqueous boundary layer. The mass transfer conditions at the boundary layer for a non-streamlined body with a complicated geometry (e.g. the sampling device evaluated in this study) are difficult to model and for practical purposes are non-predictable.17 Secondly, limitations due to molecular size exclusion of the LDPE membrane were not taken into account. In the LDPE membrane resistance calculation, a uniform membrane diffusion coefficient (DM = 10~10 cm2 s_1) for the analytes was used. In reality, the DM value is related to molecular size and it is expected to significantly decrease with increasing molecular weight.19 The complex interactions of non-porous polymers with the diffusing analytes have so far prevented the establishment of a precise relationship between molecular size and the polymer diffusion coefficients.26 Thirdly, the theoretically derived kinetic parameters represent a situation at 25 °C, for which literature data of physico-chemical properties were available. A direct comparison with experimental data obtained at 11 °C would introduce a certain systematic error. However, this is expected to be only of minor importance, because the diffusion and partition coefficients are not expected to change dramatically (orders of magnitude or so) within the chosen temperature range (11-25 °C). Finally, the resistance to mass transfer of the receiving phase was assumed to be negligible in comparison to the other sampler layers. Hence, the theoretically predicted very high (up to 30 L d_1) sampling rates for high molecular weight PAHs are not realistic. The aqueous boundary layer, in combination with the low membrane permeability and the high resistance to mass transfer in the internal water or air may result in unfavourably low sampling rates for these and similar hydrophobic compounds. This assumption was confirmed by the observed very low or non-measurable uptake rate of large ringed PAHs. A steep decline in sampling rates for very hydrophobic compounds was observed also for lipid filled SPMDs; Huckins et al.26 discussed possible reasons for this phenomenon, which include: (a) a higher order rise in resistance to mass transfer across the aqueous boundary layer for large hydrophobic analytes; (b) a sharp reduction in solubility and permeability in the LDPE for analytes with large molecules; and (c) uncontrollable partitioning of very hydrophobic substances into the colloidal phase in water. Although measured sampling rates are expected to be lower than those estimated from theory, the calculations of the combined resistances of the LDPE membrane and the inner transfer medium represent an estimate of the best (highest accumulation rate) possible sampling performance. This could potentially be achieved for small molecules in an extremely turbulent environment where the thickness of the aqueous boundary layer is close to zero.23 The factors outlined above, that could not be estimated with the required precision, preclude a direct quantitative comparison of the model predictions with the experimental data. However, the theoretical considerations satisfactorily explain the observed differences in performance of the three sampler variants investigated. Robustness of the calibration data The factors affecting a chemical's sampling rate include sampler design, molecular properties and environmental conditions. This study was performed to find the optimum sampler design. The effects of environmental conditions such as temperature and water turbulence were not addressed, but will be reported separately.25 The fact that all experimentally determined sampling rates were lower than the theoretically derived values (i.e. which did not consider the resistance of the aqueous boundary layer) indicates that all sampling devices used in this study accumulated the target analytes under aqueous boundary layer control. Moreover, Fig. 4 illustrates that even a thin (1 mm) layer of water presents a more efficient barrier to mass transfer than the LDPE membrane. As a consequence, it is likely that the performance of the passive sampler will be susceptible to changes in water flow and turbulence during deployment. When the uptake of analytes is controlled by the diffusion in the aqueous boundary layer, the presence of the LDPE membrane in the sampler does not affect the sampling kinetics. Nevertheless, its role is crucial for the selectivity of sampling. LDPE allows only small (smaller than the size exclusion limit of 1 nm) and truly dissolved molecules to be sorbed by the receiving phase; contaminants bound to humic acids and colloids are excluded. In order to sufficiently validate the performance of the passive sampler, calibration at various flow and turbulence conditions as well as exposure temperatures was performed and will be reported separately. In addition, the in situ calibration concept of performance reference compounds can be applied to support the laboratory calibration data.27 This will also be reported along with investigations into the effects of biofouling of the membrane surface during field deployments. Practical aspects of the n-octanol usage in sampler construction When selecting a suitable solvent for application as an internal transfer medium in the sampler the following factors were taken into account. The most important factor is a good permeability (i.e. the diffusion coefficient x solubility) for the hydrophobic organic pollutants. Another factor is a low volatility and diffusivity to prevent solvent loss due to evaporation and diffusion from the device during prolonged (i.e. weeks to months) field deployments. Furthermore a viscous solvent helps to prevent unwanted spillage and leakage from the sampler during preparation, exposure and disassembly.28'29 The use of n-octanol (boiling point 195 °C) as a solvent which remains in the sample after its processing has consequences for the subsequent GC/MS analyses. Enriched residue extracts in n-octanol can, in principle, be directly injected into a hot split/splitless inlet of a GC. To ensure that the chromatographic peaks do not tail, an initial column temperature of at least 120 °C is necessary. Hence, only compounds with boiling points above 250 °C (e.g. PAHs with 3 or more rings) can be determined reproducibly by this analytical method. Furthermore, other practical details have to be considered when analysing hydrophobic compounds contained in n-octanol. These include the use a viscosity delay in the automatic injection cycle, use of pulsed splitless injection and the installation of a retention gap before the chromatographic column. Mixed solvents should generally not be used for samples analysed by GC and external standards dissolved in n-octanol are required for quantification of the accumulated residues. There are, however, several advantages to the use of n-octanol. Firstly, n-octanol does not represent a severe matrix interference. The extract from the sampler can be analysed by GC without the need of special cleanup procedures, unlike that from the widely used lipid-filled SPMDs. For the latter devices, laborious and time-consuming separation of lipid matrix com- 6 1 8 J. Environ. Mo nit., 2005, 7, 612-620 ponents from target analytes using solvent dialysis and size exclusion chromatography is required.8 Secondly, n-octanol acts as an efficient solvent keeper preventing the loss of target analytes due to evaporation during the sample preparation steps and subsequent storage prior to the chemical analysis. Thirdly, n-octanol is a well-characterised substance. Physico-chemical parameters such as partition coefficients between n-octanol and other media (e.g., water, air) are available in the literature for a large number of environmental pollutants. The availability of physicochemical property data facilitates the modeling of analyte uptake by the passive sampler. Comparison with other passive sampling devices The performance of the passive sampler optimised in this study can be compared with those of other types of sampling devices designed to collect hydrophobic organic pollutants. Calibration data for PAHs are available in the literature for SPMDs21 and the membrane enclosed sorptive coating (MESCO) sampling devices.30 These devices differ in their design geometry and their use of construction materials. However, for all of these, the sampling rates are directly proportional to the sampler surface area. Consequently, the highest sampling rates will be achieved with passive samplers having a very large surface area, such as the standard size SPMDs (450 cm2 in comparison to 17.5 cm2 for the sampler used in this study). Nevertheless, the sampling performance of these devices can be compared on the surface specific basis, i.e. when their sampling rates are expressed as volume of water cleared for a chemical, per unit time and unit surface area, or L d_1 cm~2. Furthermore, it is necessary to take into account that the reported sampling rates are likely to be affected by environmental variables (temperature, water turbulence, biofouling) and vary depending on the exposure conditions used to collect the laboratory derived calibration data. Although the most calibration studies reported in the literature were performed in flow-through systems, they were not conducted under the same exposure conditions or using water of a comparable quality (e.g., the presence of variable levels of dissolved organic carbon [DOC]). Taking into account these limitations, an approximate comparison of sampling rates can be made. Fig. 6 shows that the surface specific sampling rates of the three passive sampling devices are very similar for PAHs compounds with 3 and 4 aromatic rings, ranging from 5 to 13 mL d_1 cm~2. This indicates that the uptake of these compounds by the three different samplers is governed overall by a similar mass transfer process; most likely diffusion across the aqueous boundary layer. The uptake of PAHs with 5 and more aromatic rings in the molecule by our sampling device is up to one order of magnitude slower than that of the SPMD or MESCO. As has previously been mentioned, ongoing calibrations have demonstrated that the sampling of these chemicals can be accelerated (by an order of magnitude or more) by increasing the water turbulence around the face of the sampler.25 Thus, it is likely that the uptake of these compounds is also governed by diffusion across the aqueous boundary layer. However, there are a number of other possible explanations for the observed decrease in sampling rates for the very hydrophobic (5-ringed) PAHs. A decrease in the concentration of these compounds in the exposure tank over the time course of the experiments can be excluded as we have shown that the flow-through system provided a constant level of analyte. Once equilibrated, the potential losses of these compounds to the surfaces of the glass tank and the PTFE sampler bodies were compensated for by the continuous replenishment of analyte solution. Loss of analytes due to adsorption to the PTFE parts in close proximity to the active sampling area of the sampler (i.e. near the LDPE membrane) is one possibility. We have inves- 0.014 0.012 0.01 0.008 0.006 0.004 0.002 0 I □ SPMD □ MESCO ■ Sampler, variant 3 [J -gr ,a <& r& * "s & Ö1 ^XJ « 6.31 In our calibration system, DOC would be present in the water as a result of microbial activity during the 12-day exposure period. When analysing (by SPE) the water samples from the exposure tank, the concentration of analytes represents both the truly dissolved and colloidally bound fraction. However, the passive sampler sequesters only the truly dissolved fraction. This phenomena can lead to an underestimation of the calculated sampling rates for highly hydrophobic compounds. We speculate that this is the likely cause for the observed decrease in sampling rates for 5-ring PAHs compared to calibration studies with SPMD and MESCO samplers. The influence of DOC is a common problem for the laboratory calibration for all designs of passive sampling devices and further research is necessary to address this issue.32'33 Conclusions The performance of a passive sampler for integrative (TWA) sampling of hydrophobic organic pollutants has been optimised. Substantial improvements in sampling characteristics including the magnitude of sampling rate and the sampling precision were achieved by applying a small volume of J. Environ. Mo nit., 2005, 7, 612-620 6 1 9 n-octanol, to the space between the receiving sorbent phase and the LDPE diffusion-limiting membrane. The resulting device is simple to construct and deploy, and the procedures used for the analysis of compounds retained in the receiving phase are compatible with existing instrumental methods used by environmental laboratories that measure non-polar organic compounds in water. However, in situations where pollutant concentrations are very low (sub ppt levels) where high sampling rates would be required to sequester a sufficient amount of chemical for analysis, the SPMD would still remain the passive sampling method of choice. The issues that will be addressed in the further validation of the passive sampler include testing (1) the effect of environmental variables, i.e. water temperature and turbulence, on the uptake kinetics of analytes; (2) the dissipation kinetics of individual analytes or performance reference compounds from the sampler at varying conditions as an independent measure of the exchange kinetics between the sampler and water; (3) the uptake capacity of passive sampler for individual analytes; (4) adsorption of compounds by the part of the PTFE body in close proximity to the active sampling surface and (5) the effect of DOC and biofouling on sampler performance. Acknowledgements We acknowledge the financial support of the European Commission (Contract EVK1-CT-2002-00119; http://www.port. ac.uk/research/stamps/) for this work. References 1 C. J. Koester, S. L. Simonich and B. K. Esser, Anal. Chem., 2003, 75, 2813. 2 A. Kot, B. Zabiegala and J. Namiesnik, TrAC, Trends Anal. Chem., 2000, 19, 446. 3 J. Namiesnik and T. Gorecki, LC-GC Europe, 2000, 9, 678. 4 J. N. Huckins, G. K. Manuweera, J. D. Petty, D. Mackay and J. A. Lebo, Environ. Sci. Technol., 1993, 27, 2489. 5 J. D. Petty, C. E. Orazio, J. N. Huckins, R. W. Gale, J. A. Lebo, J. C. Meadows, K. R. Echols and W. L. Cranor, /. Chromatogr., A, 2000, 879, 83. 6 F. Verweij, K. Booij, K. Satumalay, N. van der Molen and R. van der Oost, Chemosphere, 2004, 54, 1675. 7 J. D. Petty, J. N. Huckins, D. A. Alvarez, W. G. Brumbaugh, W. L. Cranor, R. W. Gale, A. C. Rastall, T. L. Jones-Lepp, T. J. Leiker, C. E. Rostad and E. T. Furlong, Chemosphere, 2004, 54, 695. 8 J. N. Huckins, M. W. Tubergen, J. A. Lebo, W. Gale and T. R. Schwartz, /. Assoc. Off. Anal. Chem., 1990, 73, 290. 9 J. K. Kingston, R. Greenwood, G. A. Mills, G. M. Morrison and L. B. Persson, /. Environ. Monit., 2000, 2, 487. 10 L. B. Persson, G. M. Morrison, J. U. Friemann, J. Kingston, G. Mills and R. Greenwood, /. Environ. Monit., 2001, 3, 639. 11 W. P. N. Fernando, M. L. Larrivee and C. F. Poole, Anal. Chem., 1993, 65, 588. 12 G. D. Johnson, Environ. Sci. Technol, 1991, 255, 1897. 13 Illustrated Handbook of Physical-Chemical Properties of Environmental Fate of Organic Chemicals, ed. D. Mackay, W. Y. Shiu and K. C. Ma, Lewis Publishers, Boca Raton, FL, 1992, vol. 2. 14 Illustrated Handbook of Physical-chemical Properties of Environmental Fate of Organic Chemicals, ed. D. Mackay and W. Y. Shiu, Lewis Publishers, Boca Raton, FL, 1992, Vol. 1. 15 W. J. Lyman, in Handbook of Chemical Property Estimation Methods, Environmental Behavior of Organic Compounds, ed. W. J. Lyman, W. F. Riehl and D. H. Rosenblatt, McGraw-Hill Book Company, New York, 1982. 16 G. A. Lugg, Anal. Chem., 1968, 40, 1072. 17 B. Vrana and G. Schuurmann, Environ. Sci. Technol., 2002, 36, 290. 18 T. Seki, J. Mochida, M. Okamoto, O. Hosoya, K. Juni and K. Morimoto, Chem. Pharm. Bull, 2003, 51, 734. 19 H. E. Hofmans, Masters Thesis, University of Utrecht, Netherlands Institute for Sea Research, Den Burg, The Netherlands, 1998. 20 J. Kingston, R. Greenwood, G. Mills, G. M. Morrison and L. B. Bjoerklund Persson, GB Patent, 2353860, 2004. 21 J. N. Huckins, J. D. Petty, C. E. Orazio, J. A. Lebo, R. C. Clark, V. L. Gibson, W. R. Gala and K. R. Echols, Environ. Sci. Technol, 1999, 33, 3918. 22 R. J. Scheuplein, /. Theor. Biol, 1968, 18, 72. 23 G. L. Flynn and S. H. Yalkowsky, /. Pharm. Sei., 1972, 61, 838. 24 K. Booij, H. M. Sleiderink and F. Smedes, Environ. Toxicol-Chem., 1998, 17, 1236. 25 B. Vrana, R. Greenwood and G. Mills, unpublished work. 26 J. N. Huckins, J. D. Petty, H. F. Prest, R. C. Clark, D. A. Alvarez, C. E. Orazio, J. A. Lebo, W. L. Cranor and B. T. Johnson, Guide for the Use of Semipermeable Membrane Devices (SPMDs) as Samplers of Waterborne Hydrophobic Organic Contaminants, Report for the American Petroleum Institute (API), API Publication 4690, API, Washington, DC, 2000. 27 J. N. Huckins, J. D. Petty, J. A. Lebo, F. V. Almeida, K. Booij, D. A. Alvarez, W. L. Cranor, R. C. Clark and B. B. Mogensen, Environ. Sci. Technol, 2002, 36, 85. 28 S. Muller, M. Moder, S. Schräder and P. Popp, /. Chromatogr., A, 2003, 985, 99. 29 K. E. Rasmussen, S. Pedersen-Bjergaard, M. Krogh, H. Grefslie Ugland and T. Gronhaug, /. Chromatogr., A, 2000, 873, 3. 30 B. Vrana, P. Popp, A. Paschke and G. Schuurmann, Anal. Chem., 2001, 73, 5191. 31 L. P. Burkhard, Environ. Sci. Technol, 2000, 34, 4663. 32 R. Luellen and D. Shea, Environ. Sci. Technol, 2002, 36, 1791. 33 C. Miege, C. Ravelet, J. P. Croue and J. Garric, Anal. Chim. Acta, 2005, 536, 259. 620 J. Environ. Monit., 2005, 7, 612-620 Príloha 7 Vrana B., Allan I. J., Greenwood R., Mills G. A., Dominiak E., Svensson K., Knutsson J., and Morrison G., Passive sampling techniques for monitoring pollutants in water, TrAC - Trends Anal. Chem., 2005, 24, 845-868. Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends Passive sampling techniques for monitoring pollutants in water Branislav Vrana, Graham A. Mills, Ian J. Allan, Ewa Dominiak, Katarina Svensson, Jesper Knutsson, Gregory Morrison, Richard Greenwood We review the state of the art in using passive sampling technology for environmental monitoring of waterborne organic and inorganic pollutants. We discuss strategies for sampler design, calibration, in situ sampling and quality-control issues, and advantages and challenges associated with passive sampling in aqueous environments. We then review typical applications of passive samplers in assessing the aquatic environment. © 2005 Elsevier Ltd. All rights reserved. Keywords: Environmental monitoring; Environmental pollutants; Passive sampling; Sample preparation; Water analysis Abbreviations: ASV, Anodic stripping voltammetry; BTEX, Benzene, toluene, ethyl benzene and xylene; DET, Diffusion equilibrium in thin films; DGT, Diffusive gradient in thin films; GC, Gas chromatography; HPLC, High-performance liquid chromatography; LDPE, Low-density polyethylene; MESCO, Membrane-enclosed sorptive coating; nd-SPME; Negligible depletion solid-phase microextraction; NOM, Natural organic matter; OSPAR, The Convention for the Protection of the Marine Environment of the North-East Atlantic; PAH, Polycyclic aromatic hydrocarbon; PCB, Polychlorinated biphenyl; PCDD, Polychlorinated dibenzo[p]dioxin; PCDF, Polychlorinated dibenzo[p]furan; PDB, Polyethylene diffusion bag; PDBS, Passive diffusion bag sampler; PIMS, Passive integrative mercury sampler; PLM, Permeation liquid membrane; POCIS, Polar organic chemical integrative sampler; PRC, Performance reference compound; QA, Quality assurance; QC, Quality control; SBSE, Stir-bar sorptive extraction; SLM, Supported liquid membrane; SLMD, Stabilised liquid-membrane device; SPATT, Solid-phase adsorption toxin tracking; SPMD, Semi-permeable membrane device; SPME, Solid-phase microextraction; SVOC, Semi-volatile organic compound; TLC, Thin-layer chromatography; TRIMPS, Trimethylpentane-containing passive sampler; TWA, Time-weighted average; VOC, Volatile organic compound 1. Introduction Branislav Vrana*, Ian J. Allan, Richard Greenwood School of Biological Sciences, University of Portsmouth, King Henry Building, King Henry I Street, Portsmouth P01 2DY, United Kingdom It is necessary to monitor pollutants in the aquatic environment to satisfy the requirements of legislative frameworks and directives, as many of these compounds can pose a threat to both human health and ecosystems. A number of toxic compounds have been designated priority pollutants [e.g., those on lists of the US Environmental Protection Agency (EPA) and the Water Framework Directive of the European Union (EU)] and their measurement is necessary to ensure that water-quality standards are maintained. Sampling and analysis of such a broad range of organic (e.g., chlorophenols, organo- chlorine pesticides, polyaromatic hydrocarbons, polychlorinated biphenyls) and inorganic (e.g., heavy metals and some of their organo-metallic species) compounds represents an ongoing challenge to the environmental chemist. Graham A. Mills School of Pharmacy and Biomedical Sciences, University of Portsmouth, White Swan Road, Portsmouth P01 2DT, United Kingdom Ewa Dominiak Department of Analytical Chemistry, Chemical Faculty, Gdansk University of Technology, 80 952 Gdansk, G. Narutowicza 11/12, Poland Katarina Svensson, Jesper Knutsson, Gregory Morrison Water Environment Transport, Chalmers University of Technology, SE-412 96 Goteborg, Sweden "Corresponding author. Tel.: +44 23 9284 2024; Fax: +44 23 9284 2070; E-mail: bran.vrana@port.ac.uk 0165-9936/$ - see front matter © 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.trac.200S.06.006 845 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Most aquatic monitoring programmes rely on collecting discrete grab, spot or bottle samples of water at a given time. Often, where pollutants are present at only trace levels, large volumes of water need to be collected. The subsequent laboratory analysis of the sample provides only a snapshot of the levels of pollutants at the time of sampling. However, there are drawbacks to this approach in environments where contaminant concentrations vary over time, and episodic pollution events can be missed. One solution to this problem is to increase the frequency of sampling or to install automatic sampling systems that can take numerous water samples over a given time period. This is costly and in many cases impractical, since a secure site and significant pre-treatment of water are required. Such systems are rarely used in widespread monitoring campaigns. Spot sampling yields different apparent concentrations of pollutants depending on the pre-treatment applied (e.g., filtering) and does not provide information on the truly dissolved, bioavailable fraction of the contaminants. Another approach that yields information on biologically relevant concentrations of pollutants uses biota. A number of test species can be used, depending on the water body being investigated. These organisms can be deployed for extended periods of time, during which they passively bioaccumulate pollutants in the surrounding water. Analysis of the tissues or lipid extracts of the test organism(s) can give an indication of the equilibrium level of waterborne contamination. A number of factors can influence the results - metabolism, depuration rates, excretion, stress, viability and condition of test organism. Furthermore, extraction of analytes from the tissue of animals prior to instrumental analysis is complex. Estimates of pollutant concentrations in water can also be made by measuring concentrations in benthic sediments and then using equilibrium distribution coefficients to derive levels of dissolved analytes. This approach is limited by the assumption of equilibrium between the sediments and the water column, and the potential effects of organic carbon quality differences among sediments or the formation of non-extractable, sediment-bound residues that are not accounted for in current equilibrium-partition models. In the last two decades, alternatives have been sought to overcome some of these difficulties. Of these, passive sampling methods have shown much promise as tools for measuring aqueous concentrations of a wide range of priority pollutants. Passive samplers avoid many of the problems outlined above, since they collect the target analyte in situ and without affecting the bulk solution. Depending on sampler design, the mass of pollutant accumulated by a sampler should reflect either the concentration with which the device is at equilibrium or the time-averaged concentration to which the sampler was exposed. Such devices have been available for monitoring air quality since the early 1970s. These diffusion-based dosimeters have been employed extensively by industry to measure toxic chemicals in workplace air. a U) si £ 3 60 50 S 40 30 20 10 first passive sampler for organic micropollutants in water dialysis with a receiving resin for inorganic micropollutants first publication on SPMD first publicationon DET first publicationon SLM first publicationon DGT detection of compounds in water at pg/L use of passive samplers incombination with bioassays 9 - passive diffusion bag for sampling VOC 10 - in situ calibration using PRC 11 -first publicationon POCIS 12 - first publicationon Chemcatcher 13-first publicationon MESCO □ inorganic □ organic op óp op Op Op Op opopopopopopopcPc?' c?1 c?1 í ííí íí ííííí 1 2 3 4 5 6 78 9 10 11 12 13 Figure 1. Milestones in the development of passive sampling techniques. The figure shows the annual number of peer-reviewed publications on the development and the application of passive samplers for monitoring of organic contaminants in water. 846 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends Later, the principles of passive dosimetry were applied in monitoring in aqueous environments. Milestones in the development of passive sampling devices for monitoring of water pollutants are shown in Fig. 1. This article reviews the state of the art of different passive sampling methods that have been developed to measure both organic and inorganic pollutants in water and highlights their range of applicability. Their potential for use in monitoring programmes is considered alongside other issues, such as quality control and detection limits. We discuss recent developments to extend their use (e.g., extracts from the devices being incorporated into bioassay-based ecotoxicology tests), challenges and limitations of the technology. 2. Principles Passive sampling can be defined in its broadest sense as any sampling technique based on free flow of analyte molecules from the sampled medium to a receiving phase in a sampling device, as a result of a difference between the chemical potentials of the analyte in the two media. The net flow of analyte molecules from one medium to the other continues until equilibrium is established in the system, or until the sampling period is stopped [1]. Sampling proceeds without the need for any energy sources other than this chemical potential difference. Analytes are trapped or retained in a suitable medium within the passive sampler, known as a reference or receiving phase. This can be a solvent, chemical reagent or a porous adsorbent. The receiving phase is exposed to the water phase, but without the aim of quantitatively extracting the dissolved contaminants. Pollutant adsorption or absorption from water into most passive sampling systems generally follows the pattern shown in Fig. 2. The exchange kinetics between a passive sampler and water phase can be described by a first-order, one-compartment mathematical model: Cs(0 = Cw^(l-e-n (1) where Cs(t) is the concentration of the analyte in the sampler at exposure time t, Cw is the analyte concentration in the aqueous environment, and /3 [21]. The design of the SPMD was first published in 1990. Since then, nearly 200 studies have been reported, and this is the most mature technique for sampling organic pollutants [22]. Several reviews and one monograph have been published on this technology [18,23-25]. http://www.elsevier.com/locate/trac 849 Table 1. Overview of passive sam pling devices for organic contaminants Sampler Full name Construction Analytes Sampling purpose Typical deployment Advantages Drawbacks Sample preparation for chemical analysis Ref. Ceramic dosimeter Ceramic dosimeter and taximeter Ceramic tube (5 x 1 cm) filled with a solid-phase sorbent material, closed with PTFE lids PAHs, BTEX, chlorinated hydrocarbons Integrative sampling in ground-water Up to 1 year No need for extensive laboratory calibrations. Robust design, suitable for long-term monitoring. Sorbent material of the "Toximeter" variant can be tested in contact bioassays Low sensitivity Solvent extraction or thermal desorption [32] Chemcatcher Universal passive sampler using Empore disk A housing made of inert plastic (e.g., PTFE), containing a disk of solid receiving phase bound in a porous polymer, and a disk of diffusion-modulating membrane. Polar and non-polar organics Integrative 14 days -1 month Selectivity of the sampler can be adjusted using appropriate combination of membrane and Empore disks. Calibration data available for many chemicals Solvent extraction [27] Dosimeter Activated carbon receiving phase in a perforated acrylic housing BTEX and atrazine Integrative Up to 2 months Solvent extraction [72] Ecoscope A sampler based on solvent-filled dialysis membranes and chelating sorbent discs A plastic housing containing a chelating sorbent disc for sampling metals and dialysis membrane filled with solvents Non-polar organics Qualitative screening [73] Gaiasafe Paper or fabric strips impregnated with a solution of binding agent Metals, anions, organic compounds Screening 2 days -2 months Solvent extraction [74] Gore-Sorber Various sorbent materials filled in a carrier hose made of Gore-Tex BTEX, MTBE, PAHs, VOCs, SVOCs Equilibrium 14 days Thermal desorption [75] LDPE and Low-density Hydrophobic Inte jrative 1 month Simple construction, Smaller Solvent [76] silicone polyethylene or silicone organic inexpensive, simple sampling extraction strips strips compounds sample processing, and capacity than calibration data available SPMDs for many analyte classes MESCO Membrane- PDMS-coated stir bar PAHs, PCBs, Inte jrative 2 weeks Miniaturised sampler, non- Low membrane Thermal [31] enclosed used in SBSE or a PDMS organochlorine depletive matrix stability of the desorption sorptive coating rod enclosed in a pesticides extraction, solventless sampler variant membrane made of sample processing, and with cellulose regenerated cellulose or both non-polar and polar dialysis low-density analytes are accumulated membrane polyethylene in the sampler equipped with a cellulose membrane nd-SPME Negligible A fibre coated with a Hydrophobic Equ librium Hours Negligible depletion Low sensitivity Thermal [30] depletion-solid liquid (polymer), a solid chemicals, extraction, a cheap, desorption in phase (sorbent), or a including PAHs, disposable device GC inlet microextraction combination of both PCBs, petroleum hydrocarbons, organochlorine pesticides, aniline, phenols Passive Sampler Silicone polycarbonate Chlorobenzenes, Inte jrative Up to 1 day Solvent [77] according to Lee permeation membrane nitrobenzenes extraction and Hardy and an adsorbent and nitrotoluenes receiving phase PDB Passive diffusion Dialysis membrane or a Polar organic Equ librium 2 weeks Relatively inexpensive, Not suitable for Conventional [35] bag samplers low-density compounds, sam 3ling in and sample recovery is sampling semi- analysis of the polyethylene bag filled VOCs, metals, grou ndwater rapid volatile organic receiving water with distilled water trace elements compounds phase PISCES Passive in situ Hexane in a PCB Inte jrative 2 weeks Volume [78] concentration- polyethylene membrane reduction of the extraction receiving phase sampler POCIS Polar organic Solid sorbent receiving Herbicides and Inte jrative Up to 2 High sensitivity; capacity Solvent [26] chemical phase material enclosed pharmaceuticals months of the sampler can be extraction integrative in a polyethersulphone with log Kqw < 3 adjusted using appropriate sampler membrane sorbent materials, membrane has low susceptibility to biofouling, and calibration data available for many chemicals (continued on next page) Table 1 (continued) Sampler Full name Construction Analytes Sampling purpose Typical deployment Advantages Drawbacks Sample preparation for chemical analysis Ref. Porous Sampler according to De Jonge and Rothenberg A water permeable porous sampler that acts as a semi-infinitive adsorptive sink Wide range of contaminants Flux-proportional sampling in soil and groundwater 1 month Tracers integrated in the sampler store information of water volume that passed the sampler during deployment Solvent extraction [79] Stainless steel housing Sampler according to Kot-Wasik A stainless steel housing, containing organic solvent in a chamber separated from water by a membrane Phenols, acid herbicides, triazines Integrative 1 month A sample of the receiving phase solvent can be taken without affecting the integrity of the sampler Low-sensitivity, receiving phase solvent may diffuse out ofthe sampler during field deployment Analysis of a sub-sample of solvent is taken and analysed without further clean-up steps [80] Solvent-filled dialysis membranes Non-polar solvent immiscible with water filled in a cellulose dialysis membrane Hydrophobic organic compounds Integrative 1 month Not prone to biofouling Low sensitivity for very hydrophobic compounds, and solvent diffuses out of the sampler during deployment Volume reduction of the receiving phase [81] SPATT Solid-phase adsorption toxin tracking Porous synthetic resin filled polyester fabric sachets Polar Phytotoxins Integrative 1 week Solvent extraction [82] SPMD Semi-permeable membrane devices Flat tube of LDPE filled with triolein Hydrophobic semi-volatile organic compounds Intej ;rative 1 month Widely used method, commercially available, well-established standard operation procedures, and calibration data available for many analyte classes, and high sensitivity Complicated sample cleanup, susceptible to biofouling Dialysis in organic solvents, size exclusion chromatography [21] TLC plate Thin-layer chromatography plate Orga no-phosphates Screening 1 month Good sensitivity because of a large surface area Solvent extraction [83] TRIMPS Trimethyl-pentane-containing passive sampler 2,2,3-Trimethylpentane filled in a low density polyethylene membrane Pesticides Intej ;rative 1 month Simple sample clean-up and analysis Receiving phase solvent diffuses out of the sampler during field deployment Direct analysis of the receiving phase solvent [84,85] TWA-SPME Solid-phase microextraction applied for determination of TWA concentrations A fibre coated with a liquid (polymer), a solid (sorbent), or a combination of both BTEX Intej ;rative A few minutes No need for extensive laboratory calibrations, and sampling rates can be estimated using empirical mass-transfer models Short-term sampling only, and fibre susceptible to damage or fouling in the field Thermal desorption in GC inlet [86] Table 2. Overview of passive sam pling devices for inorganic contaminants Sampler Full name Construction Analytes Sampling purpose Typical Advantages Drawbacks Sample Ref. deployment preparation for period chemical analysis Chemcatcher Comprises an Cd, Cu, Ni, Pb In situ sampling, 14 days Selectivity of the sampler Acid extraction [53] immobilized chelating and Zn integrative, -1 month can be adjusted using acceptor resin on a PTFE speciation appropriate combination base and a cellulose of membrane and acetate membrane filter Empore disks, and acting as a thin diffusion calibration data layer available for many chemicals DGT Diffusion Two layers of 55 metallic Integrative, 1 week Versatile, well Complicated Acid extraction [87] gradients in th n acrylamide gel mounted elements speciation, documented preparation of films in a holder device, one including the screening, device containing an acceptor common heavy mimicking phase, the other acting metals, biological uptake as a thin diffusion layer phosphorous, sulphide and "Tc PIMS Passive LDPE lay-flat tubing Neutral Hg Pre-concentration, Weeks- Membrane Further Direct analysis [52] integrative species screening months characteristics may be development of the receiving mercury altered for control of necessary for phase sampler sampling rates aquatic conditions PLM Permeation Microporous Cu, Pb Bioavailable metal Hours Selectivity of the sampler Complicated Solvent [88] liquid hydrophobic support species can be adjusted using preparation of extraction membrane separating test solution appropriate combination device from receiving solution of carrier media and receiving phase SLM Supported liqu d A strip solution with Doubly charged Integrative field Days Versatile, selectivity of Direct analysis, [89] membrane strong complexing agent cations sampling, pre- the sampler can be can be coupled is separated from the test concentration of adjusted on-line for real- solution by a macro- trace elements, time monitoring porous hydrophobic mimicking membrane biological membranes SLMD Stabilized liqu d LDPE lay-flat tubing Divalent metal Pre-concentration, Days-weeks Early Acid extraction [47] membrane containing an acidic ions in situ sampling, development device solution with high determination of stage affinity for the target labile metal ions in elements grab samples Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends 3.1.2. Polar organic chemical integrative sampler. The POCIS is used to monitor hydrophilic contaminants, such as pesticides, prescription and over-the-counter drugs, steroids, hormones, antibiotics and personal-care products [26]. Such compounds are entering water and ecosystems on a global scale and some have been linked with chronic toxicities. POCIS samples from the dissolved phase and thereby enables the biologically available fraction to be estimated. This sampler permits determination of TWA concentration in water over extended periods (several weeks). The POCIS comprises a solid receiving phase material (sorbent) sandwiched between two microporous poly-ethersulphone diffusion-limiting membranes. The type of sorbent used can be changed to target specifically certain compounds or chemical classes. Two configurations are commonly used: • a 'generic' configuration contains a mixture of three solid-phase sorbents (Isolute ENV+ polystyrene divinyl-benzene and Ambersorb 1500 carbon dispersed on S-X3 Biobeads); it is used to monitor most pesticides, natural and synthetic hormones, many wastewater-related chemicals, and other water-soluble organic chemicals and • the 'pharmaceutical' configuration contains a single (Oasis HLB) solid-phase sorbent and is designed for drug residues [26]. 3.1.3. Chemcatcher (organic version). This system uses a diffusion-limiting membrane and a bound, solid-phase receiving phase. Accumulation rates and selectivity are regulated by the choice of both the diffusion-limiting membrane and the solid-phase receiving material; both are supported and sealed in place by an inert plastic housing. For a range of priority pollutant classes, a number of designs are available with different combinations of receiving phase and diffusion-limiting membrane [27]. One design is used for the sampling of non-polar organic compounds with logKGw values greater than 4 [27]. This uses a 47-mm Ci8 Empore disk as receiving phase and an LDPE diffusion-limiting membrane. The Cis Empore disk has a high affinity and capacity for non-polar organic pollutants. Another design used for the sampling more polar organic contaminants combines a 47-mm Ci8 Empore disk as the receiving phase with a polyethersulphone diffusion-limiting membrane [27]. Other devices are being developed for a range of emerging pollutants, including alkylphenols, antiinflammatory drugs and other pharmaceuticals, poly-brominated flame retardants, steroids, sulphonamides and metals (e.g., mercury, tin and their organometallic species) [28]. 3.1.4. Negligible depletion-solid-phase microextrac-tion. Solid-phase microextraction (SPME) was developed by Pawliszyn et al. [29] as a simple extraction method with several advantages over liquid-liquid extraction and solid-phase extraction. The use of organic solvents is diminished and the SPME technique is simple, precise, and it may be automated easily, and the apparatus is inexpensive. The extraction medium is a thin layer of a polymer coating on an optical silica fibre, with a typical volume of 10-150 nL. Extraction equilibrium may generally be reached in 30 min. The mass of analyte on the fibre can be measured by either gas chromatography (GC) or high-performance liquid chromatography (HPLC). While most applications of SPME aim at the highest possible extraction efficiency, negligible depletion SPME (nd-SPME) represents a specific application to measure free concentrations based on negligible analyte extraction from the sampled matrix. In addition to the advantages of SPME, existing equilibria within the sample remain undisturbed during nd-SPME. The disadvantage of nd-SPME is the small amount of analyte that is available for analysis (typically only a few percent of the total amount in the sample), and this may lead to quantification problems. A review of nd-SPME has been published by Heringa and Hermens [30]. 3.2.5. Membrane-enclosed sorptive coating. This adaptation of the SPME technique to enable integrative passive sampling of hydrophobic organic pollutants has been reported. The device, referred to as the MESCO (membrane-enclosed sorptive coating), comprises a Gerstel Twister stir bar used for stir-bar sorptive extraction (SBSE) or a silicone polymer rod enclosed in a membrane made of regenerated cellulose. The receiving phases may be surrounded by air or water within the bag [31]. The miniature MESCO sampling system combines sampling with solventless pre-concentration. The sampler enables direct analysis of the accumulated contaminants by thermodesorption coupled on-line to GC, thereby avoiding time-consuming sample preparation and clean-up. Despite the small surface area and volume of the sampler, its sensitivity is comparable with other passive sampling systems, since the entire amount of analyte contained in the receiving phase is introduced into GC and subsequently detected. 3.1.6. Ceramic dosimeter. The ceramic dosimeter [32] uses a ceramic tube as the diffusion-limiting barrier to enclose a receiving phase comprising solid sorbent beads. Recently, the utility of the ceramic dosimeter as a robust groundwater-sampling device was demonstrated for benzene, toluene, ethyl benzenes, xylenes (BTEX) and naphthalenes, using Dowex Optipore L-493 as the receiving phase [33]. In up to 90 days of sampling in a contaminated aquifer, the ceramic dosimeter showed an excellent performance, as judged by comparing TWA contaminant concentrations derived from dosimeters with average aqueous concentrations determined by http://www.elsevier.com/locate/trac 855 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 frequent conventional spot-sampling methods. Based on the same principle, researchers proposed using Amberlite IRA-74 3 as a solid receiving phase for the measurement of PAHs [32,34]. 3.1.7. Polyethylene diffusion bags. There is potential for loss of volatiles during the collection of VOCs from groundwater. Polyethylene diffusion bag (PDB) samplers help to eliminate this problem [35,36]. The sampler comprises a membrane sealed in the form of a long cylindrical bag, filled with deionised water. The bag is made of LDPE and acts as a semi-permeable membrane allowing the passage of most chlorinated VOCs. VOCs in groundwater diffuse across the membrane into the de-ionised water in the bag until equilibrium is reached. Typically, PDBs take about 2 weeks to equilibrate in an aquifer [37]. Once this equilibration has occurred, sample recovery takes place. 3.2. Passive samplers for inorganic pollutants 3.2.1. Dialysis in situ. Equilibrium dialysis is a simple, size-based separation method applicable to the study of trace-metal speciation [38]. Sampling with a dialysis cell is based on a diffusive flux of species able to pass through the cell membrane towards a small volume of water as the acceptor solution, until equilibrium is reached. Metals associated with colloids and humic acid complexes, which are larger than the pores of the membrane, are excluded [39]. 3.2.2. Dialysis with receiving resins. An alternative configuration to the above is to add a receiving phase (e.g., a chelating resin) with a high affinity for the species being measured in the cell. Under these conditions, the diffusion rate is theoretically directly proportional to the metal concentration in the water being sampled [40]. If a suitable chelating resin is selected, the bioavailable metal species can be separated. In this case, diffusion across the dialysis membrane may simulate metal-transport processes across biological barriers. The use of the chelating resin, Chelex 100, showed a measurable, reproducible uptake of the soluble fraction of Cd, Pb and Zn at low ambient water concentrations [41]. Coefficients of variations were lower than for mussels, making this resin a promising acceptor phase for the measurement of dissolved metal species in sea-water. These devices have also been deployed in storm-water run-off and variations in the uptake rates of metals could be correlated to hydrological/hydrochemical parameters, such as rainfall volume and pH [42]. 3.2.3. Liquid membrane devices. Supported liquid membranes (SLMs) pre-concentrate trace elements from water and have been developed to mimic uptake across biological membranes. This system comprises an organic solvent with a complexing agent that is selective for the target element and is immobilised on a thin macro-porous hydrophobic membrane (either as a flat sheet or as a hollow fibre with a small lumen) [43,44]. One side of the membrane is exposed to the aqueous environment, while the other is in contact with a strip solution containing a complexing agent with a higher affinity towards the metals being separated than the one immobilised in the membrane. A proton, an anion or a metal-ion counter gradient drives the transport across the device. The device can be tailored to separate specific metal species by a careful selection of complexing agents or by altering the lipophilicity of the diffusion membrane [45,46]. SLM devices have been used to measure Cd, Co, Cu, Ni, Pb and Zn in natural waters. Effects of turbulence, pH and concentration variations on the performance of SLM devices have been reported [47]. The permeation liquid membrane (PLM) device is the result of further development of the SLM. This technique is based on carrier-mediated transport of metals across a hydrophobic membrane. The microporous support is impregnated with a hydrophobic organic solvent and placed between the sample and a receiving solution [48]. The transport of Cu and Pb complexes through a PLM with a neutral macrocyclic carrier has been described [49]. 3.2.4. Diffusive gradient in thin films. The diffusive gradient in thin-films (DGT) device is a development of a similar sampler - the diffusion equilibrium in thin-films (DET) device - initially suggested by Davison and co-workers in 1991 [50]. The first reported use of the improved DGT device was in 1994 for measuring Zn in sea-water. The DGT device comprises a gel-layer incorporating a binding agent (which acts as a solute sink) and a hydrated acrylamide diffusion gel separating it from the water column. This creates a diffusion layer of well-defined thickness. The initial design of the DGT utilised an ion-exchange resin as the receiving phase. Later, Zhang and co-workers [51] demonstrated the applicability of the technique to determination of trace metals (Cd, Cu, Fe and Mn) in sea-water. With a chelating resin embedded in the gel layer, metals could be quantified as low as 4 pmol/L after deployment for 1 h. The subsequent refinement of the design and the extended range of inorganic pollutants that may be sampled indicate the versatility and the widespread use of the DGT device. In principle, it is possible to sample any labile species for which a suitable binding agent can be embedded into the receiving phase gel. 3.2.5. Passive integrative mercury sampler. Attempts have been made to use the passive integrative mercury sampler (PIMS), originally designed for air sampling, to sample neutral Hg species in water [52]. The device comprises lay-flat LDPE tubing containing a reagent mixture of nitric acid and gold stock solution. 856 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends Experiments were performed in simulated freshwater and sea-water environments. The uptake rates remained linear for 2 weeks and preliminary results indicate that sampling of neutral Hg species from water is feasible. Sampling in freshwater was more effective than in sea-water, likely to be because a larger fraction of the total Hg in sea-water was present as charged chloro-anion complexes that could not readily permeate through the membrane. 3.2.6. Chemcatcher (inorganic version). An alternative configuration of the Chemcatcher (see Section 3.1.3) has been developed for the separation of metals. The device comprises a commercially available 47 mm diameter chelating extraction disk as receiving phase and a cellulose acetate diffusion-limiting membrane [53]. The sampler has been used to monitor Cd, Cu, Ni, Pb and Zn, in various aquatic environments, such as a storm-water pond, where the uptake of metals was compared with flow-weighted bottle samples. Results indicated a good correlation with the electro-available Cu fraction but were somewhat less clear for Zn [53]. The diffusion-limiting membrane can be treated with a low surface-energy coating (e.g., polyfluorinated sul-phonic acid polymer (Nafion)) to reduce biofouling on the surface of the membrane. The diffusion characteristics of the membrane, the influences of water turbulence and the radius of metal ions monitored have been investigated [54]. 4. Applications of samplers The first publications on the use of passive samplers to monitor aquatic contaminants were in 1980s (Fig. 1) and these devices have since received widespread recognition as effective tools in environmental research. Passive sampling technology is widely applicable in monitoring studies and the results obtained can be interpreted at different levels of complexity. Passive samplers have been employed in field studies aimed at: (a) screening for the presence and absence of pollutants; (b) investigating temporal trends in levels of water-borne contaminants; (c) monitoring spatial contaminant distribution and tracing point and diffusive pollution sources; (d) speciation of contaminants; (e) assessing pollutant fate and distribution between environmental compartments; (f) measuring TWA concentrations of waterborne pollutants; (g) comparing contaminant patterns in biota and passive samplers - biomimetic sampling to estimate organism exposure; and, (h) assessing toxicity of bioavailable pollutants in extracts from the receiving phase of passive samplers. Tables 3 and 4 illustrate the different field applications. These tables are not intended to be comprehensive, but rather to give the reader an overview of the variety of applications. A detailed review of the organic contaminant classes and aqueous matrices that can be sampled by passive samplers was recently published by Stuer-Lauridsen [55]. 4.1. Use in chemical monitoring There are several advantages in using passive samplers for monitoring pollutants in water including: (a) non-mechanical or passive operation; (b) ability to sample large volumes of water and (c) reduced effort required for deployment and sample processing compared to other commonly used methods. Currently available passive sampling devices are applicable to monitoring chemicals with a broad range of physicochemical properties (Fig. 3) and the detection limits obtained or the lowest measured concentrations (Fig. 4) suggest that passive samplers may find application in monitoring programmes. Stuer-Lauridsen [55] indicated that passive sampling devices can be used to monitor more than 75% of the organic micropollutants listed in water-quality criteria of the EU and US, the EU Water Framework Directive and the recommendations of The Convention for the Protection of the Marine Environment of the North-East Atlantic (OSPAR). 4.2. Contaminant speciation Speciation of environmental contaminants includes not only physicochemical speciation of the forms in which analytes are present in the sampled matrix (e.g., freely dissolved, colloidal and particle-bound forms), but also chemical speciation (e.g., the valency state of metals in the sampled water). Trace metals are present in water in various forms (hydrated ions, and inorganic and organic complexes) together with species associated with heterogeneous colloidal dispersions. The particulate phase also contains elements in a range of chemical associations, from weak adsorption to binding in the mineral matrix. These species coexist, although they may not necessarily be in thermodynamic equilibrium. The difficulty in differentiating the various forms arises from the low levels present in natural waters. The fractionation of species is recognised as an essential step in assessing bioavailability and toxicity in water. A problem is that solution equilibria may change after sample collection through adsorption or desorption of analytes to particulate and colloidal surfaces. A representative http://www.elsevier.com/locate/trac 857 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Table 3. Examples of field applications of passive sampling devices for monitoring orj *anic contaminants Application Sampler Environment Analytes Short description Ref. Screening of n-Hexane- Lake water Organochlorine Detection of contaminants in [90] contaminant for filled compounds passive samplers and presence or dialysis mussels absence membranes POCIS Wastewater Polar wastewater- Screening of contaminants [19] effluents related contaminants and pharmaceuticals SPMD River Hydrophobic Screening of contaminants [91,92] organic contaminants Speciation of SPMD Seawater PAHs Distribution of particulate, [57] contaminants dissolved, and colloidal PAHs in the water column nd-SPME River water PCBs, Determination of freely [58] chlorobenzenes dissolved contaminant fraction in presence of humic acids SPMD River PAHs Relationship between freely [59] dissolved contaminant levels and the quality of dissolved organic matter Monitoring of SPMD Seawater Organochlorine Temporal trend in sea-water [93] temporal compounds pollution by outflow of pollution trends contaminated freshwater following a flood episode SPMD Seawater PCBs and Time evolution in air, sea- [94] hexachloro-benzene water, and at the sea-air boundary layer Monitoring of SPMDs River PCBs Identification and [95] spatial contribution of point and distribution and diffusive sources to the total tracing pollution contaminant flux sources SPMD River PCDDs, PCDFs and Spatial distribution of [96] PCBs contaminants in a river basin SPMD River and sea- PAHs Spatial distribution of [97] water contaminants SPMD Surface water UV filter compounds A regional mass-balance [98] study PISCES Surface water PCBs Tracing a point source of [99] and effluent pollution wastewater SPMD Discharge Alkylphenol Spatial distribution of [100] from ethoxylates contaminants and their wastewater- degradation products in the treatment aqueous phase and their plants distribution between sediment and water column SPMD River PBDEs Assessment of spatial [101, contaminant levels and 102] contaminant-pattern profiles and their relation to contaminant sources SPMD Seawater PAHs Spatial levels and patterns of [103] contaminated bioavailable contaminant by discharged fraction oilfield- produced water 858 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends Table 3 (continued) Application Sampler Environment Analytes Short description Ref. SPMD Seawater Organotin Spatial levels and patterns of [104] compounds contaminants sampled by passive samplers and mussels compared to those with water samples Assessment of SPMD Irrigation PAHs Measuring the residence [105] contaminant water canal times (or persistence) of fate and analytes in the dissolved distribution phase water between SPMD Discharge PCBs, Comparison of contaminant [106] environmental from an chlorophenols, levels in SPMD, mussel and compartments industrial chlorobenzenes sediment source to sea- water SPMD Freshwater, Triclosan Fate of a bactericide in the [107,108] wastewater- aquatic environment treatment plants Low-density Seawater PCBs, PAHs and Distribution of dissolved [109] polyethy- hexach loro-benzene contaminants between lene strips sediment, pore-water and overlying water column SPMD River PCBs, PAHs, Comparison of dissolved [110-113] PCDDs, PCDFs and contaminant levels and substituted benzenes patterns estimated using sediment, fish and SPMD SPMD River Petroleum Pre-concentration of sub-part [114] hydrocarbons per billion levels for studying source, transport, and bioremediation using carbon- and hydrogen- isotope analysis Measurement of SPMDs River PCDDs, PCDFs Comparison of levels and [115,116] time-weighted congener profiles of average extremely hydrophobic aqueous compounds in SPMDs and concentrations water Ceramic Groundwater PAHs Comparison of passive [34] dosimeter samplers with spot sampling SPMD Groundwater PAHs Comparison of passive [70] samplers with spot sampling POCIS Effluent of Polar Assessment of prescription [117] wastewater- pharmaceuticals and illicit drugs in treated treatment sewage effluents plants Chem- Harbour Antifouling agents Comparison of passive [27] catcher samplers with spot sampling Estimate of SPMD Harbour Organochlorine Comparison of contaminant [118] organism pesticides levels and patterns in exposure mussels and SPMDs SPMD Seawater PAHs Assessment of contaminant [119] accumulation in mussels, fish and SPMDs exposed to dispersed crude oil SPMD Laboratory PCBs and Comparison of uptake [120,121] exposure in Organochlorine kinetics in SPMDs and fish groundwater pesticides spiked with contaminant (continued on next page) http://www.elsevier.com/locate/trac 859 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Table 3 (continued) Application Sampler Environment Analytes Short description Ref. SPMD Seawater PAHs Assessment of chemical [122,123] exposure in a side-by-side deployment of SPMD and bivalves TRIMPS River polluted Endosulfan Correlation of contaminant [124] by field run-off levels in passive samplers by pesticides with population densities of macroinvertebrates SPMD Wastewater- Synthetic musks Comparison contaminant [125] treatment levels and patterns in fish, plant mussels and SPMDs SPATT Seawater Algal toxins Assessment of shellfish [82] contamination by toxins using samplers and mussels deployed side by side Biomimetic Equilibrium Effluents and A complex mixture Estimate of total body [126,127] extraction for sampling surface water of hydrophobic residues in biota after toxicity using chemicals exposure to complex assessment of Empore disk chemical mixtures aqueous (sampling is contaminants not performed in situ) SPME A methodical A complex mixture Estimate of total body [65] study of hydrophobic residues in biota after chemicals exposure to complex chemical mixtures SPMD Effluents of Organochlorine Instrumental analysis and [128] wastewater- pesticides, PCBs, bioindicator tests to treatment PAHs determine toxic potential of plant bioavailable contaminants SPMD River A complex mixture Bioassay-directed [129] of hydrophobic fractionation to identify chemicals bioavailable and toxic chemicals SPMD Urban stream PAHs Assessment of toxic potency [130] of compounds collected by SPMDs using an in vitro bioassay value is particularly difficult to identify through conventional sampling procedures in environments where concentrations fluctuate [56]. 4.2.1. Organic contaminants. Passive samplers can be applied to characterise the distribution of organic contaminants between particulate, dissolved and colloidal phases in the water column [57-59]. The selectivity of devices may be adjusted to sample a desired fraction of an analyte present by choosing membrane materials with desired properties (e.g., pore size and charge on the surface). Most passive samplers collect only the truly dissolved fraction of chemicals, since: (a) the truly dissolved molecules become separated from colloids and particles during their diffusion across the membrane that separates water from the receiving phase [21]; and, (b) only dissolved molecules are sorbed by the receiving phase [30]. 4.2.2. Inorganic contaminants. Passive samplers have been used to gain understanding of the species of metals in the aquatic environment. Speciation of metals with the DGT device relies on two effects: the relative difference in diffusion coefficients; and, the relative difference in affinity to the binding agent between the species to be characterized. It is possible to differentiate between inorganic labile species and organic labile species by employing a systematic variation of diffusion gel pore sizes, resulting in a size-discriminating uptake in a similar fashion to voltammetry. However, diffusion coefficients of the model species have to be determined individually to make accurate measurements of the concentration of the labile species [60]. 860 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends Table 4. Examples of field applications of passive sampling devices for monitoring inorganic contaminants Application Sampler Environment Analytes Short description Ref. In situ metal SLM Natural Cd, Cu and The transport mechanisms [45,46,131] speciation waters Pb through supported liquid membrane devices for metal-ion separation and pre-concentration were studied and optimised SLMD Natural waters Cd, Co, Cu, Ni, Pb, Zn Effects of environmental conditions on the sampling of metals were investigated [47] Chem- Natural Cd, Cu, Ni, Integrative metal sampling was [53,54] catcher waters Pb, Zn compared with spot sampling and attempts made to reduce biofouling DGT and Natural Cr Simultaneous application of [132] DET waters DGT and DET to determine Cr(lll) and Cr(lll)/Cr(VI) fractions in resin layer and diffusive equilibrium layer, respectively DGT Lake water Cu, Fe, Mn and Zn Study of DGT performance in fivedifferent lakes (pH 4.7-7.5) and comparison between dialysis and predictions of a speciation model [133] DGT Natural freshwater Cu and Zn Comparison of DGT, competitive ligand exchange and voltammetric measurements, as well as examining the agreement of the results with predictions made by several speciation models [134] DGT Synthetic freshwater Cd Examination of DGT lability of Cd in solutions containing various synthetic (nitrilo-triacetic acid (NTA) and diglycolic acid) and natural (extracted fulvic acid) ligands. Diffusion gel of reduced pore size used to estimate portion of Cd complexed by fulvic acid [135] DGT Natural water Ni and Zn In situ determination of Zn and Ni speciation between humic and fulvic acid complexes through the use of diffusive gel layers with different pore sizes. Comparison with ASV results and predictions of speciation model [136] Mimics DGT Ion-poor Cu Comparison of Cu binding to [137] bioavailability water trout gills and results for ion-selective electrode and DGT measurements. Examination of the influence of NOM on Cu bioavailability DGT Freshwater Cu Investigation of the performance of DGT in the evaluation of toxic fraction of Cu to Daphnia magna, using synthetic ligands (EDTA, NTA, glycine and humic acids) [138] {continued on next page) http://www.elsevier.com/locate/trac 861 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Table 4 (continued) Application Sampler Environment Analytes Short description Ref. DGT Seawater Cd, Cu, Pb and Zn Parallel use of DGT devices and transplanted mussels to assess metal levels in marine environment [139] DGT Freshwater Al Comparison of the relevance of DGT performance to the observed bioavailability of Al with trout (Salmo trutta L.) compared with a pyrocatechol violet fractionation procedure [140] PLM Natural waters Cu, Pb Transport of metal complexes through the permeation liquid membrane depends on the lipophilicity of the complexes [88,141] Determination DGT Freshwater 134Cs Use of ammonium [142] of radionuclides molybdophosphate binding agent to collect and determine 134Cs in laboratory tests and applied to a natural freshwater lake Determination DGT Freshwater Al, Ba, Co, A novel sediment trap device [143] of metal Cu, Fe, Mn was used together with a remobilization and Ni DGT device to determine the metal remobilization from settling particles in a well-mixed lake 4.3. Quantification of concentrations in water Passive sampling methods can be used to calculate the concentrations of compounds in the aqueous phase, using the principles described in Section 2. Fig. 5 illustrates the way in which integrative passive samplers can provide representative information on TWA contaminant concentrations over a long period of time with a sampling frequency lower than in spot sampling. However, it is important to recognise that, in most cases, the aqueous concentration estimated using passive samplers reflects only the truly dissolved contaminant fraction and is not necessarily equal to the concentration measured in spot samples, particularly in very hydrophobic compounds in the presence of elevated levels of dissolved organic matter. Nevertheless, the comparison is possible, if all species and fractions of contaminants present in the sampled matrix are characterised (see Section 1). In many aquatic systems, contaminant concentrations are not constant, but fluctuate or occur in the form of unpredictable pulses. Concentrations reflected by integrative passive samplers are TWAs over the exposure period, but more research is needed to quantitate the uptake in passive samplers in scenarios involving pulsed and discontinuous exposure. Such research will provide sufficient evidence of realistic concentration estimates using passive samplers and convince the regulators of the application of passive samplers in monitoring programmes. 4.4. Estimate of organism exposure Sijm et al. [61] reviewed biomimetic passive sampling methods to study the bioavailability of chemicals in soil or sediment. Biomimetic equilibrium sampling approaches using SPME [29] and Empore disks can mimic partitioning of contaminants between the pore water and the organism. Both approaches assume that the freely dissolved contaminant concentrations will represent bioavailability. However, for substances that may be biotransformed in the organism, the methods will overestimate the concentration in the organism. For organisms that have several routes of uptake (in addition to via the water phase), the biomimetic method will underestimate the concentration in the organisms. Biomimetic sampling devices have been applied to sense dissolved sediment pore-water concentrations 862 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends LDPE and silicone strips - SPMD - PISCES - nd-SPME - MESCO - TRIMPS - Ecoscope - CHEMCATCHER - Solvent-filled dialysis membranes - TLC plate - Ceramic dosimeter - Dosimeter according to Kot-Wasik . - PDB - Sampler according to Lee and Hardy - TWA-SPME - POCIS - Dosimeter according DiGiano et al. - 10 log Kow Figure 3. Typical hyclrophobicity range of organic compounds sampled by selected passive sampling devices (characterised by the value of octanol/water partition coefficient, log /Cow)- SPMD LDPE and silicone strips TRIMPS CHEMCATCHER MESCO Solvent-filled dialysis membranes ■ POCIS Dosimeter according to Kot-Wasik TLC plate PDB Sampler according to Lee and Hardy Ceramic dosimeter ■ 10 12 14 16 - log c [g/L] Figure 4. Typical detectable concentrations of organic compounds by selected passive sampling devices. of contaminants [62,63] and to estimate the bioaccu-mulation potential in effluents and surface waters [64,65]. 4.5. Bioassays The pre-concentrated extracts obtained from the elution of receiving phases of passive samplers (particularly those used to measure organic pollutants) can subsequently be combined with a variety of bioassay procedures to assess both the level and the biological effects of water contaminants [66]. In some in vitro bioassays used to assess the health of an ecosystem, problems can occur due to the difficulty of obtaining suitable water samples for testing. For example, most hydrophobic organic contaminants are present in aquatic environ- ment only at trace levels (i.e., <1 fig/V). The extraction of several litres of water would be required to yield sufficient amounts of analyte for subsequent bioassay. The use of "bio-mimetically" separated extracts from passive samplers can overcome this problem [67]. It has been shown that the baseline toxicity of chemicals can be predicted (based on total body residue estimates) from the concentration of contaminants separated by passive samplers [68]. 5. Quality control The level of quality control (QC) applied to passive sampling varies with project goals and analytical procedures http://www.elsevier.com/locate/trac 863 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 60.0 50.0 o o -7 40.0 30.0 20.0 10.0 0.0 10 15 Time [d] 20 25 30 Figure 5. Comparison of a 28-clay TWA concentration of simazine obtained using passive sampling (dashed line; the Chemcatcher integrative sampler variant for polar organic chemicals) with the concentrations determined in filtered spot samples (circles) from the Meuse River, The Netherlands, in Spring 2004 [144]. involved. The application of appropriate QC procedures and parameters is a mandatory consideration in both sampler deployment and subsequent analysis. QC samples should address issues of purity of materials used to construct a device, and potential contamination during transport, deployment, retrieval and subsequent storage. QC protocols are also required for analyte recovery and further processing (enrichment and fractionation operations). Control charts are recommended for monitoring analyte recoveries throughout a project. The QC samples relevant to passive sampler studies include fabrication blanks, process blanks, reagent blanks, field blanks and sampler spikes. DeVita and Crunkilton [69] examined the QC issues associated with using SPMDs for monitoring PAHs in water. Their results showed that QC measures applied to SPMDs met or surpassed conventional guidelines (EPA method 610 for PAHs in water) for precision and accuracy. However, assessing the accuracy and the trueness of determinations made by passive samplers may prove difficult, as the results may not be directly comparable with total concentrations found in spot samples or by other sampling techniques. This is because only very few methods, other than passive samplers, can truly measure dissolved contaminant fractions. When environmental conditions at an exposure site differ from laboratory calibration conditions or calibration data are not available, samplers spiked with PRCs serve as a special type of QC sample. These provide information about in situ uptake kinetics [16,17]. QC samples involved in using passive sampling devices are shown in Fig. 6. Stuer-Lauridsen [55] discussed the quality assurance (QA) that would be required for passive samplers to be accepted in water-quality-monitoring programmes. 6. Future trends There are several major trends in the future development of passive sampling technology. The first is towards miniaturisation of devices. Small devices offer the advantages of inexpensive transportation to and from the sampling site, the requirement for small deployment devices and a low consumption of solvents and reagents during their subsequent processing. Moreover, miniaturised devices allow application in situations with limited space and volume of water (e.g., in groundwater boreholes [70]). Miniaturisation goes hand in hand with the trend to develop solventless sample-preparation techniques. Passive samplers based on in situ analyte pre-concentration using SPME or similar techniques allow sample processing (following exposure) using thermal desorption GC [31] or solvent microextraction followed by HPLC [71]. However, the practical application of SPME-based techniques in in situ passive sampling of aqueous trace contaminants will require their robustness and sensitivity to be further enhanced. The second trend is the development of passive sampling technology to monitor a wider range of chemicals. Recently, attention has been focused on compounds with medium-to-high polarity (e.g., polar pesticides and drugs [26]). Precise calibration of passive sampling devices for monitoring trace metals is essential for quantifying the various metal species and complexes found in water. 864 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Trends P repar ation Storage Transport Deployment Exposure Retrieval Storage Transport Processing Analysis Fabrication blanks Trip blanks > Field blanks J Performance reference compounds > Field blanks Trip blanks > Reagent blanks, process blanks Recovery spikes Figure 6. Quality control samples involved in passive sampling. This requires knowledge of the uptake kinetics of different metal moieties. Configuration of specific devices for monitoring well-defined fractions of metals will increase their potential as regulatory tools. A further challenge is to improve robustness by reducing or controlling the impacts of environmental conditions and biofouling on the sampler performance. Internal and external PRCs are being tested for improving the accuracy of TWA concentrations of contaminants. Another trend is the coupling of chemical and biological analysis of samples collected using passive samplers, with detection and identification of toxico-logically relevant compounds. The marriage of passive samplers and bio-marker and bio-indicator tests offers many avenues of investigation to provide information concerning the relative toxicological significance of waterborne contaminants. Finally, the development of efficient QA, QC and method-validation schemes for passive sampling techniques is essential to gain broader acceptance for the technology in regulatory programmes. Acknowledgements We acknowledge the financial support of the European Commission (Contract EVK1-CT-2002-00119; http:// www.port.ac.uk/research/stamps/) for this work. We thank Michiel Kotterman and Pim Leonards from The Netherlands Institute for Fisheries (RIVO), IJumiden, The Netherlands, for their permission to publish data in Fig. 5. References [1] T. Gorecki, J. Namiesmk, Trends Anal. Chem. 21 (2002) 276. [2] P. Mayer, J. Tolls, J. Hermens, D. Mackay, Environ. Sci. Technol. 37 (2003) 184A. [3] P.T. Harte, Ground Water Monit. Remediat. 22 (2002) 45. [4] http://diffusionsampler.itrcweb.org/. [5] J. Pawliszyn, Anal. Chem. 75 (2003) 2543. [6] B. Zabiegala, A. Kot, J. Namiesmk, Chem. Anal. 45 (2000) 645. [7] R.W. Gale, Environ. Sci. Technol. 32 (1998) 2292. [8] E.L. Cussler, in: E.L. Cussler, A. Varma (Editors), Diffusion: Mass Transfer in Fluid Systems, Cambridge University Press, Cambridge, UK, 1984. [9] J.N. Huckins, J.D. Petty, C.E. Orazio, J.A. Lebo, R.C. Clark, V.L. Gibson, W.R. Gala, K.R. Echols, Environ. Sci. Technol. 33 (1999) 3918. [10] D.R. Luellen, D. Shea, Environ. Sci. Technol. 36 (2002) 1791. [11] M.P. Harper, W. Davison, H. Zhang, W. Tych, Geochim. Cosmochim. Acta 62 (1998) 2 75 7. [12] C. Murdock, M. Kelly, L.Y. Chang, W. Davison, H. Zhang, Environ. Sci. Technol. 35 (2001) 4530. http://www.elsevier.com/locate/trac 865 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 [13] B. Vrana, G. Schüürmann, Environ. Sei. Technol. 36 (2002) 290. [14] B.J. Richardson, P.K.S. Lam, G.J. Zheng, K.E. McCellan, S.B. De Luca-Abbott, Marine Pollut. Bull. 44 (2002) 1372. [15] G.L. Flynn, S.H. Yalkowsky, J. Pharm. Sei. 61 (1972) 838. [16] K. Booij, H.M. Sleiderink, F. Smedes, Environ. Toxicol. Chem. 17 (1998) 1236. [17] J.N. Huckins, J.D. Petty, J.A. Lebo, F.V. Almeida, K. Booij, D. A. Alvarez, W.L. Cranor, R.C. Clark, B.B. Mogensen, Environ. Sei. Technol. 36 (2002) 85. [18] J.N. Huckins, J.D. Petty, K. Booij, Monitors of Organic Contaminants in the Environment: Semipermeable Membrane Devices, Springer, Berlin, Germany (in press). [19] D.A. Alvarez, P.E. Stackelberg, J.D. Petty, J.N. Huckins, E. T. Furlong, S.D. Zaugg, M.T. Meyer, Chemosphere (in press). [20] J. Namiesnik, B. Zabiegala, A. Kot-Wasik, M. Partyka, A. Wasik, Anal. Bioanal. Chem. 381 (2005) 2 79. [21] J.N. Huckins, G.K. Manuweera, J.D. Petty, D. Mackay, J.A. Lebo, Environ. Sei. Technol. 27 (1993) 2489. [22] http://wwwaux.cerc.cr.usgs.gov/spmd/SPMD_references.htm. [23] J.N. Huckins, J.D. Petty, J.A. Lebo, C.E. Orazio, H.F. Prest, D.E. Tillitt, G.S. Ellis, B.T. Johnson, G.K. Manuweera, in: G.K. Ostrander (Editor), Techniques in Aquatic Toxicology, CRC Press (Lewis Publishers), Boca Raton, FL, 1996, p. 625. [24] Y. Lu, Z. Wang, J. Huckins, Aquat. Toxicol. 60 (2002) 139. [25] J.D. Petty, C.E. Orazio, J.N. Huckins, R.W. Gale, J.A. Lebo, J.C. Meadows, K.R. Echols, W.L.J. Cranor, J. Chromatogr., A 879 (2000) 83. [26] D.A. Alvarez, J.D. Petty, J.N. Huckins, T.L. Jones-Lepp, D.T. Getting, J.P. Goddard, S.E. Manahan, Environ. Toxicol. Chem. 23 (2004) 1640. [2 7] J.K. Kingston, R. Greenwood, G.A. Mills, G.M. Morrison, B.L. Persson, J. Environ. Monit. 2 (2000) 487. [28] www.port.ac.uk/research/stamps/. [29] J. Pawliszyn, Solid-Phase Microextraction: Theory and Practice, Wiley, NY, USA, 1997. [30] M.B. Heringa, J.L.M. Hermens, Trends Anal. Chem. 22 (2003) 575. [31] B. Vrana, P. Popp, A. Paschke, G. Schüürmann, Anal. Chem. 73 (2001) 5191. [32] H. Martin, M. Piepenbrink, P. Grathwohl, J. Process Anal. Chem. 6 (2001) 68. [33] H. Martin, B.M. Patterson, G.B. Davis, Environ. Sei. Technol. 3 7 (2003) 1360. [34] S. Bopp, H. Weiss, K. Schirmer, J. Chromatogr., A 1072 (2005) 137. [35] D.A. Vroblesky, W.T. Hyde, Ground Water Monit. Remediat. 17 (1997) 177. [36] http://www.diffusionsampler.org. [3 7] C.E. Divine, J.E. McCray, Environ. Sei. Technol. 38 (2004) 1849. [38] R.E. Truitt, J.H Weber, Environ. Sei. Technol. 15 (1981) 1204. [39] J. Buffle, Trends Anal. Chem. 1 (1981) 90. [40] P. Benes, Water Res. 14 (1980) 511. [41] R.S.S. Wu, T.C. Lau, Marine Pollut. Bull. 32 (1996) 391. [42] G.M.P. Morrison, G.E. Batley, T.M. Florence, Chem. Brit. 25 (1989) 791. [43] J.A. Jonsson, L. Mathiasson, Trends Anal. Chem. 11 (1992) 106. [44] N. Parthasarathy, M. Pelletier, J. Buffle, Anal. Chim. Acta 350 (1997) 183. [45] N. Parthasarathy, J. Buffle, Anal. Chim. Acta 254 (1991) 1. [46] N. Parthasarathy, J. Buffle, Anal. Chim. Acta 284 (1994) 649. [47] W.G. Brumbaugh, J.D. Petty, J.N. Huckins, S.E. Manahan, Water, Air, Soil Pollut. 133 (2002) 109. [48] V.l. Slaveykova, N. Parthasarathy, J. Buffle, K.J. Wilkinson, Sei. Total Environ. 328 (2004) 55. [49] N. Parthasarathy, M. Pelletier, J. Buffle, J. Chromatogr., A 1025 (2004) 33. [50] W. Davison, H. Zhang, Nature (London) 367 (1994) 546. [51] H. Zhang, W. Davison, Anal. Chem. 67 (1995) 3391. [52] W.G. Brumbaugh, J.D. Petty, T.W. May, J.N. Huckins, Chemos. Global Change Sei. 2 (2000) 1. [53] L.B. Persson, G.M. Morrison, J.U. Friemann, J. Kingston, G. Mills, R. Greenwood, J. Environ. Monit. 3 (2001) 639. [54] L.B. Blom, G.M. Morrison, J. Kingston, G.A. Mills, R. Greenwood, T.J.R. Pettersson, S. Rauch, J. Environ. Monit. 4 (2002) 258. [55] F. Stuer-Lauridsen, Environ. Pollut. 136 (2005) 503. [56] G.M. Morrison, D.M. Revitt, J.B. Ellis, G. Svensson, P. Balmer, Water Res. 22 (1988) 1417. [5 7] J. Axelman, K. Naes, C. Naf, D. Broman, Environ. Toxicol. Chem. 18 (1999) 2454. [58] E.U. Ramos, S.N. Meijer, W.H.J. Vaes, H.J.M. Verhaar, J.L.M. Hermens, Environ. Sei. Technol. 32 (1998) 3430. [59] C. Miege, S. Durand, J. Garric, C. Gourlay, D. Wang, J.M. Mouchel, M.H. Tusseau-Vuillemin, Polycycl. Aromat. Compd. 24 (2004) 805. [60] H. Zhang, W. Davison, Anal. Chem. 72 (2000) 4447. [61] D. Sijm, R. Kraaij, A. Belfroid, Environ. Pollut. 108 (2000) 113. [62] P. Mayer, W.H.J. Vaes, F. Wijnker, K.C.H.M. Legierse, R.H. Kraaij, J. Tolls, J. Hermens, Environ. Sei. Technol. 34 (2000) 5177. [63] R. Kraaij, P. Mayer, F.J.M. Busser, M. van het Bolscher, W. Seinen, J. Tolls, Environ. Sei. Technol. 3 7 (2003) 268. [64] P.G.-J. DeMaagd, Environ. Toxicol. Chem. 19 (2000) 25. [65] E.M.J. Verbruggen, W.H.J. Vaes, T.F. Parkerton, J.L.M. Hermens, Environ. Sei. Technol. 34 (2000) 324. [66] D. Sabaliunas, J. Lazutka, I. Sabaliuniene, A. Sodergren, Environ. Toxicol. Chem. 17 (1998) 1815. [67] B.T. Johnson, J.N. Huckins, J.D. Petty, R.C. Clark, Environ. Toxicol. 15 (2000) 248. [68] H.A. Leslie, J.L.M. Hermens, M.H.S. Kraak, Environ. Toxicol. Chem. 23 (2004) 2017. [69] W.M. De Vita, R.L. Crunkilton, Environmental Toxicology and Risk Assessment, vol. 7, American Society of Testing and Materials, STP 1333 (1998) 237. [70] K.E. Gustavson, J.M. Harkin, Environ. Sei. Technol. 34 (2000) 4445. [71] P. Popp, C. Bauer, M. Moder, A. Paschke, J. Chromatogr., A 897 (2000) 153. [72] F.A. DiGiano, D. Elliot, D. Leith, Environ. Sei. Technol. 22 (1988) 1365. [73] www.alcontrol.se. [74] www.gaiasafe.de. [75] H. Sorge, P. Göttzelmami, M. Nallinger, Terra Tech. 4 (1994) 26. [76] K. Booij, F. Smedes, E.M. van Weerlee, Chemosphere 46 (2002) 1157. [77] H.L. Lee, J.K. Hardy, Int. J. Environ. Anal. Chem. 72 (1998) 83. [78] S. Litten, B. Mead, J. Hassett, Environ. Toxicol. Chem. 12 (1993) 639. [79] H. De Jonge, G. Rothenberg, Environ. Sei. Technol. 39 (2005) 274. [80] A. Kot-Wasik, Chem. Anal. 49 (2004) 691. [81] A. Sodergren, Environ. Sei. Technol. 21 (1987) 855. [82] L. Mackenzie, V. Beuzenberg, P. Holland, P. McNabb, A. Selwood, Toxicon 44 (2004) 901. [83] C.J. Leblanc, W.M. Stallard, P.G. Green, E.D. Schroeder, Environ. Sei. Technol. 3 7 (2003) 3966. [84] J.M. Zabik, L.S. Aston, J.N. Seibber, Environ. Toxicol. Chem. 11 (1992) 765. [85] A.W. Leonard, R.V. Hyne, F. Pablo, Environ. Toxicol. Chem. 21 (2002) 2591. 866 http://www.elsevier.com/locate/trac Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 [86] K. Sukola, J. Koziel, F. Augusto, J. Pawliszyn, Anal. Chem. 73 (2001) 13. [87] W. Davison, H. Zhang, Nature (London) 367 (1994) 546. [88] V.l. Slaveykova, N. Parthasarathy, J. Buffle, K.J. Wilkinson, Sei. Total Environ. 328 (2004) 55. [89] 1J.A. Jonsson, L. Mathiasson, Trends Anal. Chem. 11 (1992) 106. [90] S. Herve, H.F. Prest, P. Heinonen, T. Hyotylainen, J. Koistinen, J. Paasivirta, Environ. Sei. Pollut. Res. 2 (1995) 24. [91] L.L.P. Stee, P.E.G. Leonards, W.M.G.M. van Loon, A.J. Hendriks, J.L. Maas, J. Struijs, U.A.Th. Brinkman, Water Res. 36 (2002) 4455. [92] L.R. Zimmerman, E.M. Thurman, K.C. Bastian, Sei. Total Environ. 248 (2000) 169. [93] P.-A. Bergqvist, B. Strandberg, R. Ekelund, C. Rappe, A. Granmo, Environ. Sei. Technol. 32 (1998) 3887. [94] K. Booi), B.L. van Drooge, Chemosphere 44 (2001) 91. [95] J.F. McCarthy, G.R. Southworth, K.D. Ham, J.A. Palmer, Environ. Toxicol. Chem. 19 (2000) 352. [96] K. McCarthy, R.W. Gale, Hydrol. Process. 15 (2001) 1271. [97] M. Shaw, LR. Tibbetts, J.F. Muller, Chemosphere 56 (2004) 237. [98] T. Poiger, H.R. Buser, M.E. Balmer, P.A. Bergqvist, M.D. Muller, Chemosphere 55 (2004) 951. [99] S. Litten, B. Fowler, D. Luszniak, Chemosphere 46 (2002) 145 7. [100] E.R. Bennett, CD. Metcalfe, Environ. Toxicol. Chem. 19 (2000) 784. [101] M.G. Ikonomou, S. Rayne, M. Discher, M.P. Fernandez, W. Cretney, Chemosphere 46 (2002) 649. [102] S. Rayne, M.G. Ikonomou, Environ. Toxicol. Chem. 21 (2002) 2292. [103] T.I.R. Utvik, S. Johnsen, Environ. Sei. Technol. 33 (1999) 1963. [104] N. Folsvik, E.M. Brevik, J.A.J. Berge, Environ. Monit. 4 (2002) 280. [105] H.F. Prest, L.A. Jacobson, M. Wilson, Chemosphere 35 (1997) 3047. [106] A. Granmo, R. Ekelund, M. Bergren, E. Brorstrom-Lunden, P.A. Bergqvist, Environ. Sei. Technol. 34 (2000) 3323. [107] A. Lindstrom, I.J. Buerge, T. Poiger, P.-A. Bergqvist, M.D. Müller, H.-R. Buser, Environ. Sei. Technol. 36 (2002) 2322. [108] D. Sabaliunas, S.F. Webb, A. Hauk, M. Jacob, W.S. Eckhoff, Water Res. 37 (2003) 3145. [109] K. Booi), J.R. Hoedemaker, J.F. Bakker, Environ. Sei. Technol. 37 (2003) 4213. [110] K.R. Echols, R.W. Gale, T.R. Schwartz, J.N. Huckins, L.L. Williams, J.C. Meadows, D. Morse, J.D. Petty, C.E. Orazio, D.E. Tillitt, Environ. Sei. Technol. 34 (2000) 4095. [Ill] Z. Wang, Y. Wang, M. Ma, Y. Lu, J. Huckins, Environ. Toxicol. Chem. 21 (2002) 2378. [112] F. Verwei), K. Booi), K. Satumalay, N. van der Molen, R. van der Oost, Chemosphere 54 (2004) 1675. [113] R.W. Gale, J.N. Huckins, J.D. Petty, P.H. Peterman, L.L. Williams, D. Morse, T.R. Schwartz, D.E. Tillitt, Environ. Sei. Technol. 31 (1997) 178. [114] Y. Wang, Y.S. Huang, J.N. Huckins, J.D. Petty, Environ. Sei. Technol. 38 (2004) 3689. [115] A.L. Rantalainen, M.G. Ikonomou, LH. Rogers, Chemosphere 37 (1998) 119. [116] J.A. Lebo, R.W. Gale, D.E. Tillitt, J.N. Huckins, J.C. Meadows, C.E. Orazio, D.J. Schroeder, Environ. Sei. Technol. 29 (1995) 2886. [117] T.L. Jones-Lepp, D.A. Alvarez, J.D. Petty, J.N. Huckins, Arch. Environ. Contam. Toxicol. 47 (2004) 427. [118] CS. Hofelt, D. Shea, Environ. Sei. Technol. 31 (1997) 154. Trends [119] T. Baussant, S. Sanni, G. Jonsson, A. Skadsheim, J.F. Borseth, Environ. Toxicol. Chem. 20 (2001) 1175. [120] J.C. Meadows, K.R. Echols, J.N. Huckins, F.A. Borsuk, R.F. Carline, D.E. Tillitt, Environ. Sci. Technol. 32 (1998) 1847. [121] Y. Lu, Z. Wang, Water Res. 37 (2003) 2419. [122] J.N. Huckins, H.F. Prest, J.D. Petty, J.A. Lebo, M.M. Hodgins, R.C. Clark, D.A. Alvarez, W.R. Gala, A. Steen, R. Gale, C.G. Ingersoll, Environ. Toxicol. Chem. 23 (2004) 1617. [123] B.J. Richardson, G.J. Zheng, E.S.C. Tse, S.B. De Luca-Abbott, S.Y.M. Siu, P.K.S. Lam, Environ. Pollut. 122 (2003) 223. [124] A.W. Leonard, R.V. Hyne, R.P. Lim, F. Pablo, P.J. Van Den Brink, Environ. Toxicol. Chem. 19 (2000) 1540. [125] R. Gatermann, S. Biselli, H. Hiihnerfuss, G.G. Rimkus, M. Hecker, L. Karbe, Arch. Environ. Contam. Toxicol. 42 (2002) 437. [126] W.M.G.M. Van Loon, M.E. Verwoerd, F.G. Wijnker, C.J. Van Leeuwen, P. Van Duyn, C. Van DeGuchte, J.L.M. Hermens, Environ. Toxicol. Chem. 16 (1997) 1358. [127] E.M.J. Verbruggen, W.M.G.M. Van Loon, M. Tonkes, P. Van Duijn, W. Seinen, J.L.M. Hermens, Environ. Sci. Technol. 33 (1999) 801. [128] J.D. Petty, S.B. Jones, J.N. Huckins, W.L. Cranor, J.T. Parris, T.B. McTague, TP. Boyle, Chemosphere 41 (2000) 311. [129] D. Sabaliunas, J. Ellington, I. Sabaliuniene, Ecotoxicol. Environ. Safety 44 (1999) 160. [130] D.L. Villeneuve, R.L. Crunkilton, W.M. DeVita, Environ. Toxicol. Chem. 16 (1997) 977. [131] N. Parthasarathy, M. Pelletier, J. Buffle, Anal. Chim. Acta 350 (1997) 183. [132] H. Ernstberger, H. Zhang, W. Davison, Anal. Bioanal. Chem. 373 (2002) 873. [133] J. Gimpel, H. Zhang, W. Davison, A.C. Edwards, Environ. Sci. Technol. 37 (2003) 138. [134] S. Meylan, N. Odzak, R. Behra, L. Sigg, Anal. Chim. Acta 510 (2004) 91. [135] E.R. Unsworth, H. Zhang, W. Davison, Environ. Sci. Technol. 39 (2005) 624. [136] H. Zhang, Environ. Sci. Technol. 38 (2004) 1421. [137] CD. Luider, J. Crusius, R.C. Playle, P.J. Curtis, Environ. Sci. Technol. 38 (2004) 2865. [138] M. Tusseau-Vuillemin, R. Gilbin, E. Bakkaus, J. Garric, Environ. Toxicol. Chem. 23 (2004) 2154. [139] J.A. Webb, M.J. Keough, Marine Pollut. Bull. 44 (2002) 222. [140] O. Royset, B.O. Rosseland, T. Kristensen, F. Kroglund, O.A. Garmo, E. Steinnes, Environ. Sci. Technol. 39 (2005) 1167. [141] N. Parthasarathy, M. Pelletier, J. Buffle, J. Chromatogr., A 1025 (2004) 33. [142] C. Murdock, M. Kelly, L. Chang, W. Davison, H. Zhang, Environ. Sci. Technol. 35 (2001) 4530. [143] J. Hamilton-Taylor, E.J. Smith, W. Davison, H. Zhang, Limnol. Oceanogr. 44 (1999) 1772. [144] P. Leonards, M. Kotterman, Personal communication, 2005. Branislav Vrana is a research associate at the School of Biological Sciences, University of Portsmouth, UK. His research is focused on developing passive sampling devices for monitoring organic environmental pollutants. Graham Mills is a Reader in Environmental Chemistry at the University of Portsmouth. His research interests are the use of chromatographic and spectroscopic techniques for the analysis of biological fluids and environmental pollutants. http://www.elsevier.com/locate/trac 867 Trends Trends in Analytical Chemistry, Vol. 24, No. 10, 2005 Ian Allan is a research associate at the School of Biological Sciences, University of Portsmouth. His research is focused on contaminated soils and emerging tools for monitoring water quality. Ewa Dominiak is a PhD student at the Department of Analytical Chemistry, Gdansk University of Technology, Poland. Her research is focused on passive sampling of organic environmental pollutants. Katarina Svensson is a PhD student at the Department of Civil and Environmental Engineering, Chalmers University of Technology, Göteborg, Sweden. Her research is focused on development of passive sampling devices for monitoring inorganic environmental pollutants. Jesper Knutsson is a research engineer at Water Environment and Technology, Chalmers University of Technology. His field of expertise is trace-metal analysis. Gregory Morrison is Professor in sustainable aquatic systems at the Department of Water Environment Technology, Chalmers University of Technology. Richard Greenwood is Head of School of Biological Sciences at the University of Portsmouth, and coordinator of the European Union 5th Framework-funded project, STAMPS, on passive sampling in aquatic systems. 868 http://www.elsevier.com/locate/trac Príloha 8 Vrana B., Paschke A., and Popp P., Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water, Environ. Pollut, 2006,144, 296-307. Performance of semipermeable membrane devices for sampling of organic contaminants in groundwater! Branislav Vrana,*a Heidrun Paschke,6 Albrecht Paschke,c Peter Popp6 and Gerrit Schüürmannc a University of Portsmouth, School of Biological Sciences, King Henry Building, King Henry I Street, Portsmouth, Hampshire, UK POl 2DY. E-mail: bran.vrana@port.ac.uk; Fax: +44 23 92 84 2070; Tel: +44 23 92 84 2024 b Department of Analytical Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany c Department of Chemical Ecotoxicology, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany Received 29th July 2004, Accepted 3rd March 2005 First published as an Advance Article on the web 22nd March 2005 Lipid-filled semipermeable membrane devices (SPMDs) are receiving increasing attention as passive, in situ samplers for the assessment of environmental pollutant exposure. Although SPMDs have been successfully used in a variety of field studies in surface waters, only a few studies have addressed their characteristics as groundwater samplers. In this study, the performance of the SPMDs for monitoring organic contaminants in groundwater was evaluated in a pilot field application in an area severely contaminated by chemical waste, especially by chlorinated hydrocarbons. The spatial distribution of hydrophobic groundwater contaminants was assessed using a combination of passive sampling with SPMDs and non-target semiquantitative GC-MS analysis. More than 100 contaminants were identified and semiquantitatively determined in SPMD samples. Along the 6 field sites under investigation, a large concentration gradient was observed, which confirms a very limited mobility of hydrophobic substances in dissolved form in the aquifer. The in situ extraction potential of the SPMD is limited by groundwater flow, when the exchange volume of well water during an exposure is lower than the SPMD clearance volume for the analytes. This study demonstrates that SPMDs present a useful tool for sampling and analyzing of groundwater polluted with complex mixtures of hydrophobic chemicals and provides guidance for further development of passive sampling technology for groundwater. Introduction The monitoring of temporal and spatial trends in concentration levels of groundwater pollutants is essential for ecological risk assessment for chemical stressors as well as for surveillance of the success of remediation measures. This may be impossible without extensive repetitive sampling, when the conventional approach of spot sampling is used. This approach is to sample a quantity of a groundwater and determine the quantity of contaminant or analyte present, and then calculate the total concentration. It is both expensive and labour intensive, and measures only instantaneous concentrations, which may not be representative of long-term average pollutant concentrations. Moreover, measurements of organic chemicals in groundwater are highly prone to bias stemming from the choice of sampling techniques. Many hydrophobic chemicals tend to be particle-or colloid-associated, and not truly dissolved in groundwater.1 Thus, if the sampling method tends to increase the amount of suspended solids in groundwater samples (by re-suspending sediments in the well or by remobilizing particles from the aquifer), reported levels of organic chemicals may be erroneously high. In addition, groundwater sampling with pumps leads to a change in the hydraulic flow field, potentially causing a dilution of the contaminants. Furthermore, the use of con-u ventional sampling methods is affected by sorption of the S nonpolar analytes to the bailers, bags, filters, and tubing used.2 ■í -Q oi m o - ° t Electronic supplementary information (ESI) available: Tables with ^ detailed information on the contaminants identified in the SPMD Q extracts. See http://www.rsc.org/suppdata/em/b4/b411645c/ To overcome these limitations, improved sampling and analytical methods are needed, suitable for the characterization of contaminants in groundwater. Various sampling techniques have been developed to avoid aggravated disturbance of the groundwater wells and the surrounding aquifer during sampling.3 Passive samplers present a novel, non-invasive technology suitable for long-term monitoring of organic pollutants in groundwater.4 Passive sampling involves the deployment of a device, which uses a diffusion gradient to collect pollutants over a period of days to weeks, followed by extraction and analysis of pollutants in a laboratory to provide a measure of concentrations of pollutants to which the sampler was exposed. Two main regimes can be distinguished in passive sampler operation, these are integrative and equilibrium sampling. In the case of equilibrium sampling, the exposure time is sufficiently long to permit the establishment of thermodynamic equilibrium between the water and the reference phase. Equilibrium groundwater sampling devices called passive diffusion bag (PDB) samplers have been widely applied in groundwater monitoring.5 This sampler is suitable for sampling of volatile organic compounds, but its application for sampling of semi-volatile organic compounds is restricted. With integrative sampling, it is assumed that the rate of mass transfer to the reference phase is linearly proportional to the difference between the chemical activity of the contaminant in the water phase and in the reference phase. Unlike spot sampling, kinetic or integrative sampling methods also sequester contaminants from episodic events, can be used in situations of variable concentrations, and permit measurement of time-weighted average (TWA) concentrations over extended 5 00 J. Environ. Mo nit., 2005, 7, 500-508 This journal is © The Royal Society of Chemistry 2005 time periods. A comprehensive review with a full listing of available passive sampling techniques has recently been published by Namiesnik et al.6 Along with other passive sampling techniques, semipermeable membrane devices (SPMDs) present a convenient sampling and preconcentration method for instrumental methods of chemical analysis, as well as for bioassays.7 In the SPMD extraction, hydrophobic chemicals are sampled more efficiently than less hydrophobic chemicals, simulating the way xenobiotics are accumulated from the aqueous phase by biota.8'9 Moreover, the SPMD extraction allows restriction of sampling to the truly dissolved fractions in the aqueous phase of the water samples, while most of the sampling techniques include the fraction of the chemicals associated with suspended particles or colloids. SPMD is especially suitable for sampling of semivolatile organic compounds. In this study, we wanted to test the performance of the lipid-filled SPMDs for monitoring of organic contaminants in groundwater in a pilot field application, by evaluating a procedure that combines an innovative groundwater sampling technique with subsequent chemical screening and semiquantitative analysis of accumulated contaminants. The aim was to evaluate the potential of SPMDs to become a competitive tool for monitoring spatial and temporal distribution of organic groundwater pollutants in an area severely contaminated by chemical production residues, especially chlorinated hydrocarbons. Moreover, the study was conducted to provide an informative basis about the character of pollution with semivolatile organic compounds in order to target future method validation at the most relevant identified contaminants. The study was performed in the Bitterfeld region in Saxony-Anhalt, Germany, as part of the interdisciplinary joint research program called SAFIRA (SAnierungsForschung In Regional kontaminierten Aquiferen = Remediation Research in Regionally Contaminated Aquifers; abbreviation from German).10 In this program, suitable and innovative in situ remediation procedures have been developed and tested. The region was heavily polluted by mining, chemical industry and the uncontrolled deposition of chemical wastes over nearly 100 years. Groundwater in the area is still severely contaminated by chemical waste. A conservative estimate puts the volume of contaminated groundwater in the Bitterfeld region at some 200 million cubic metres. Serious ecological impact is to be expected when the groundwater contaminant plume reaches the zone of interaction with the nearby biosphere reservation area of the Mulde river floodplain. Although the main groundwater contaminants in the area are thought to be water soluble and volatile, the contribution of hydrophobic semivolatile contaminants is not known. Whilst hydrophobic compounds are present in groundwater in low concentrations (|ig L~'), they potentially can be accumulated by biota, and all the compounds in combination may cause severe biological effects. Experimental Materials and chemicals The solvents acetone, dichloromethane, hexane and isopropa-nol in LiChrosolv quality were obtained from Merck (Germany). Dimethylsulfoxide was obtained from Fluka (Germany). Perdeuterated polyaromatic hydrocarbons (D-PAHs) were obtained as pure neat compounds (purity >99%) from Promochem (Germany). Organic pollutant standards of 33 organic contaminants for determination of relative molar response factors and calculation of GC retention indices were prepared from neat compounds of high purity (>99%). These were purchased from Dr Ehrenstorfer (Germany), Pro- mochem (Germany), Riedel de Haen (Germany), Sigma Aldrich (Germany) and from Merck (Germany). Physicochemical properties of contaminants The octanol-water partition coefficient (Kovl) and the normal boiling point of substances identified in SPMD extracts were estimated using an incremental method.11'12 In cases where the substitution pattern of atoms in the identified molecular structure was ambiguous, Kow values were estimated for substances with all possible isomeric structures, and average values were calculated and utilised. Sampling devices SPMDs with standard configuration (2.54 x 91.4 cm, 75-90 |im membrane thickness, total mass 4.3 g each), assembled from low-density polyethylene lay-flat tubing and containing a thin film of 95% pure triolein (1 mL), were purchased from Exposmeter, Tavelsjo, Sweden. SPMDs were stored in original, gas-tight, metal paint cans until just before field deployment. Before groundwater exposure, SPMDs used for later quantification of accumulated compounds using GC-MS were spiked with performance reference compounds (PRCs), a mixture of D-PAHs including 2H10-biphenyl, 2H10-fiuorene, 2H10-phenan-threne (D-PHE), 2H10-anthracene, 2H10-fiuoranthene, 2H10-pyrene and 2H12-benz(a)anthracene. D-PAHs were spiked in 100 |iL of a hexane stock solution using an HPLC syringe (volume 100 |iL) to give a final concentration of 10 |ig of individual compound per SPMD. SPMDs without D-PAH addition were used for toxicity screening using bioassays. Sampling sites The study was performed in the Bitterfeld region in Saxony-Anhalt, Germany. The place of the SAFIRA project is located in an area of Bitterfeld free of previous mining activities, where a quaternary (Wechselion Mulde river gravel) and a tertiary (Bitterfeld mica sand) aquifer are separated by a lignite seam of 5-9 m thickness. In order to study the aquifer and the groundwater quality, almost 40 boreholes were installed and expanded to form groundwater monitoring wells in the past. Aquifer material analysis has shown that volatile chlorinated aliphatic hydrocarbons and chloroaromatics are present at elevated concentrations in the aquifers but are distributed differently above and below the coal horizon. In the upper aquifer, monochlorobenzene is the most important groundwater contaminant (20-30 mg L_1), whereas chloroethenes dominate in the aquifer below the lignite seam.13 The locations of the sampling wells used in this study are shown in Fig. 1 and their selected geophysical parameters are given in Table 1. The area near the well GWM 19/91 has been directly polluted by seepage of spilled chemicals in the past and represents the source zone of the contamination plume. In consideration of the main groundwater flow direction and known geological conditions, transport of organic pollutants in groundwater is expected toward the east and south from the source zone. Sampling was performed at the depth of the quaternary aquifer (19-32 m in the subsurface), except for the borehole SafBit 2/96, where groundwater from a greater depth (45 m) was also sampled. SPMD sampling The SPMDs were lowered into 5 groundwater wells in the study area for 20 days during spring 2000. SPMDs were mounted in perforated stainless steel deployment cages (5 x 5 x 80 cm long). Two SPMDs were mounted inside the deployment cage to form two open loops bent in the middle. Each loop was stretched between stainless steel pins at opposite ends of the deployment cage. Two sampling containers were J. Environ. Mo nit., 2005, 7, 500-508 50 1 4521000 rw 452i50o 4522cm) 4522500 Fig. 1 Map of the sampling area in Bitterfeld, Saxony-Anhalt, Germany. Circles indicate groundwater wells, where SPMDs were deployed. The area near to the well GWM 19/91 has been directly polluted by seepage of spilled chemicals in the past and represents the source zone of the contamination plume. deployed in each of the sampling wells. The first deployment cage contained two SPMD samples spiked with PRCs, which were used later for instrumental chemical analysis. Two SPMDs without PRCs were mounted in the second container and examined by bioassays after exposure. The results of the bioassay responses to the SPMD extracts will be reported separately. On day 20, SPMDs were removed from the deployment devices and immediately sealed in individual contaminant-free metal cans. The cans were transported on the same day, approximately within 6 hours, to the processing laboratory on ice and in darkness and were kept in a freezer at —20 °C until processing. SPMD processing The SPMD processing was described previously.14 Briefly, the devices were subjected to exterior cleanup. In contrast to our previous field studies with SPMDs in surface water (e.g. ref. 14), no fouling was observed at the surface of membranes after 20 days of exposure. This can be explained by a low activity of microorganisms in groundwater compared to that in surface water. SPMDs were then dialysed twice with 250 mL hexane per SPMD at 18 °C for 24 hours. The dialysate was concentrated to approximately 10 mL by rotary evaporation. Nonane (100 u,L) was added as a keeper and the volume was reduced using high-purity nitrogen. The residue was redissolved in dichloromethane. The concentrated extract was cleaned by size exclusion chromatography (SEC) using a high performance Table 1 SPMD sampling wells and selected geophysical parameters Sampling Sampling Flow well depth/m Aquifer direction SEC column (22.5 mm i.d. x 250 mm, 10 |im particles, Lichrogel PS 20 (Merck, Germany)) with dichloromethane (5 mL min-1) as mobile phase. The SEC fraction containing the contaminants (85-100 mL) was collected. This step results in the elimination of nearly all lipid materials and polyethylene oligomers. Solvent exchange from dichloromethane to 1 mL hexane was performed prior to examination by instrumental analysis. Instrumental analysis GC-MS non-target analysis (without standards15) was used for the identification and semiquantitative analysis of the contaminants in the SPMD extracts. The extracts were injected via an autosampler (1 |iL, splitless) into a GC (HP 5890) and separation of the contaminants was performed using a capillary column (30 m x 0.25 mm i.d.) with a nonpolar stationary phase HP5-MS (thickness 0.25 |im). The temperature conditions were as follows: injector 250 °C, column 50 °C (5 min)-5 °C rnin_1-280 °C (10 min), detector transfer line 280 °C. Ultra high purity helium was used as carrier gas. Detection was performed using a mass spectrometric detector operating in electron impact ionisation mode at 70 eV. The detector temperature conditions were: ion source temperature 230 °C and quadrupole temperature 150 °C. The detector was operated in the full-scan mode in the m/z range from 30 to 450. Quantification of D-PHE and other PRCs was accomplished in selected-ion monitoring mode (MS-SIM) using a six-point external standard curve under the same chromatographic conditions. Identification of contaminants The identification of substances in the SPMD extracts was performed by comparing the mass spectra obtained from the total ion chromatograms with the NIST 98 mass spectral library. The criterion for identifying a substance was the quality of match with the mass spectrum entry in the spectral database. A spectrum match quality value higher than 80% was considered sufficient for preliminary substance identification (see Electronic Supplementary Informationf, protocols from GC-MS analyses). It was verified by peak purity evaluation that each integrated peak resulted from only a single component, without co-elution of a major interference. The identity of the substances in samples from different sampling wells was confirmed also by the consistency of the retention times. To achieve a higher degree of certainty for correct component identification, retention time information was also incorporated into the identification. For this purpose, the Lee retention index (LRI) system was used.16'17 This system employs the polycyclic aromatic hydrocarbons naphthalene, phe-nanthrene and chrysene as the retention time markers. D-PAHs were used in our case since their LRIs do not differ Groundwater Hydraulic Hydraulic TOC/mg temperature/°C gradient (%o) conductivity/m s_1 L_1 .5 .0 SafBit 30/98 19.5- -20.5 Quaternary West - -> east 13.8 0.2-0.8 5.50E-04 (partially to the south) SafBit 31/98 19.0- -20.0 Quaternary West - -> east 13.7 0.2-0.8 2.20E-04 (partially to the south) SafBit 2/96 (tert.) 44.5- -46.0 Tertiary West - -> east 16.1 1.0-3.0 8.00E-06 SafBit 2/96 (quat.) 30.5- -32.0 Quaternary — 16.1 1.0-3.0 — SafBit 16/97 20.0- -21.0 Quaternary West - -> east 15.2 1.0-3.0 2.60E-05 GWM 19/91 24.5- -25.5 Quaternary — 16.5 — — " — Indicates no information about the parameter. 5 02 J. Environ. Mo nit., 2005, 7, 500-508 substantially from those of native substances. These compounds are assigned LRI values of 200 (n = 2), 300 (n = 3), and 400 (n = 4), respectively. LRIs of unknown compounds were calculated by linear interpolation: LRI = [100 x (RTunknown + 100(«) RT„)/(RT„+l - RT„)] (1) where -Rrunknown is the retention time of the unknown compound; RTn and RTn+l are the retention times of the markers that elute before and after the unknown. To prove the applicability of the Lee retention index system at the gas chromatographic conditions used in this study, LRI values were also calculated for 32 compounds chosen to reflect the contaminant spectrum identified in SPMD samples. The standard LRI values were compared with their published LRI values17 or their normal boiling point, if literature LRI data were not available (Table 2). Semiquantitative analysis of contaminants Determination of molar concentrations (expressed as umol of substance per SPMD sample) of the identified components was performed as follows. The total ion current (MS-TIC) technique was used for quantification. Concentrations of individual components were calculated using the approach shown by van Loon et a/.18 For this purpose, relative molar response factors (RMRFs) compared to D-PHE were determined for 33 com- pounds chosen to reflect the contaminant spectrum identified in SPMD samples (Table 2). The RMRF is defined in eqn. (2). Here, k\ and ^d-phe are the molar response factors of compound i and D-PHE, respectively. These values correspond to the slopes of a linear dependence S = ktC of the TIC area (S) on the molar amount injected (C). The kt values were calculated from ten-point calibration curves (injected concentration range 1-50 u.g mL_1). RMRF = ki/ku-i (2) An average RMRF value determined for the test set of 33 compounds was used for quantification of contaminants in SPMD extracts. The molar concentrations of individual components in SPMD extracts, Ct, were calculated as Ci 1 Cd-p RMRFa -5,- (3) where Cd.phe is the molar amount of D-PHE quantified separately in each sample using MS-SIM technique, Sd-phe and St are TIC areas of D-PHE and of the selected component in the full scan chromatogram of the same SPMD sample. Quality control Fresh SPMDs were taken through the entire dialytic and cleanup procedure (procedural blanks for instrumental analysis). Table 2 Molecular weight, boiling point, relative molar resp onse factors (RMRF) as compared to 2Hi0 phenanthrene and Lee retention indices of 33 organic standards as determined by GC-MS; extraction recovery of selectee organic standards from procedural spikes is given in the last column No. Compound MW° BP6 RMRFC LRť Recovery rate6 (%) 1 2-Bromotoluene 171.0 182 0.28 171 2 3-Bromotoluene 171.0 184 0.26 169 3 4-Bromotoluene 171.0 184 0.32 170 4 1,3-Dichlorobenzene 147.0 173 0.28 162 10 5 1,4-Dichlorobenzene 147.0 173 0.35 163 10 6 1,2-Dichlorobenzene 147.0 181 0.29 168 12 7 3-Nitroanisole 153.1 258 0.24 228 8 Bromobenzene 157.0 155 0.28 150 5 9 l-Chloro-3-nitrobenzene 157.6 236 0.25 210 10 1 -Chloro-4-nitrobenzene 157.6 242 0.27 212 11 1 -Chloro-2-nitrobenzene 157.6 246 0.25 213 12 1 -Bromo-2-chlorobenzene 191.4 204 0.40 188 13 2,5-Dichloroaniline 162.0 251 0.36 229 14 2-Chloronaphthalene 162.6 256 0.94 237 15 2-Chlorotoluene 126.58 159 0.34 151 16 3-Chlorotoluene 126.5 161 0.30 150 17 4-Chlorotoluene 126.5 162 0.34 151 18 1,3,5-Trichlorobenzene 181.4 208 0.52 190 70 19 1,2,4-Trichlorobenzene 181.4 214 0.51 199 55 20 1,2,3-Trichlorobenzene 181.4 219 0.49 207 50 21 frans-Azobenzene 182.2 293 0.41 278 62 22 2,4-Dichlorotoluene 161.0 200 0.43 187 23 l-Bromo-4-chlorobenzene 191.5 196 0.45 188 24 l,2-Dichloro-3-nitrobenzene 192.0 258 0.37 241 51 25 Azoxybenzene 198.2 — 0.34 — 26 1,2,3,5-Tetrachlorobenzene 215.9 246 0.56 228 83 27 1,2,4,5-Tetrachlorobenzene 215.9 246 0.78 229 80 28 1,2,3,4-Tetrachlorobenzene 215.9 254 0.61 237 82 29 Pentachlorobenzene 250.3 277 1.04 262 84 30 l,l,2,3,4,4-Hexachloro-l,3-butadiene 260.8 210 0.84 — 31 Hexachlorobenzene 284.8 332 1.06 291 100 32 a-Hexachlorocyclohexane 290.8 288 0.72 289 82 33 y-Hexachlorocyclohexane Average RMRF Standard deviation Relative standard deviation (%) 290.8 323 0.70 0.47 0.23 50% 298 75 " Molecular weight/g mol b Normal boiling point/°C. traction recovery from procedural spikes. Relative molar response factor; see Experimental section. d Lee retention index. e Ex- J. Environ. Mo nit., 2005, 7, 500-508 503 In addition, trip blanks were used to define contamination of the SPMDs during transportation and handling as described by Petty et al}9 Spiked SPMDs were also analysed by fortifying fresh membranes and then processing them as samples. The PRCs were spiked at 500 ng per SPMD for each single component. Procedural spikes were also analysed by fortifying fresh membranes with selected analytes and then processing them as samples (Table 2). The standards were spiked at 500 ng per SPMD for each single component. Results and discussion Identification of contaminants in SPMD samples Fig. 2 illustrates the results of the analysis of SPMD extracts from two of the sampling locations, and of an associated SPMD control. Up to 167 components were identified in the SPMD samples by mass spectrum library search and the identity of 123 substances characterized by a mass spectrum could be confirmed using LRI. For confirmation, calculated LRI were compared with published LRI data17 or normal boiling points of the substances identified by the mass spectrum library search as has been shown by Eckel.16 First, the applicability of the LRI system was confirmed for 32 standards chosen to reflect the spectrum of groundwater contaminants. The LRI values of the standards calculated from GC retention times using eqn. (1) correlated well with the normal boiling points of the respective substances (Fig. 3). The identity of a substance preliminarily characterized by a mass spectrum was confirmed only when the absolute difference between the boiling point (in °C) and the corresponding LRI was lower than 50 (Fig. 3). Identified contaminants included aliphatics and cycloaliphatics, chloroaliphatics, chlorinated and brominated benzenes, toluenes and xylenes, alkylated benzenes and naphthalenes, alkyl- and arylsulfides, sulfur containing heterocyclic aromatics, methylated and chlorinated aromatic amines, hexachlorocyclohexanes (HCHs), alkylphenols and nitrobenzenes. Matrix impurities identified in extracts from fresh procedural SPMD blanks (up to 20 compounds) included alkanes (C9-C16), bis(2-ethylhexyl)phthalate (DEHP), decahydrodimethyl naphthalene, methyl oleate and fatty acids (CI8). During groundwater exposures, some of the impurities dissipated from 500 -.--- / A 400- O 100--f- 100 200 300 400 500 Lee retention index Fig. 3 A plot of the Lee retention index (LRI) of preliminarily identified compounds (by a mass spectrum library search) in SPMD extracts (full circles) and of 32 standards chosen to reflect the spectrum of groundwater contaminants (triangles) versus their normal boiling points or LRI values published by Rostad et al. (ref. 17; hollow circles). Identity of a substance was confirmed only when the absolute difference between the boiling point and the corresponding LRI was lower than 50 (within the band given by the dotted lines). SPMDs. On average, only 5% of the initial amount of DEHP, 3% of the methyl oleate and less than 10% of the fatty acids were found in SPMD extracts after groundwater exposure. This finding indicates the need for specially adapted negative controls for bioassays, because these substances might cause additional inhibitory effects when SPMD extracts are subject to toxicity screening. Quantification of contaminants The applicability of the GC-MS method for total molar determinations strongly depends on the variation of the molar ill jí... ; r I ,~ i 11 I ii 1 1 , i Saf Bit 31/98 L.....i i 5.00 10.00 15.00 20.00 25.00 30.00 40.00 45.00 Time-> Fig. 2 GC-MS chromatograms of 2 SPMD sample extracts and a control (trip blank) SPMD. Samplers were deployed for 20 days in groundwater monitoring wells in Bitterfeld, Saxony-Anhalt, Germany. Peaks denoted by (D) are PRCs. Matrix impurities identified in procedural blanks are denoted by (I). Some of the matrix impurities dissipate from SPMDs during groundwater exposure. 5 04 J. Environ. Mo nit., 2005, 7, 500-508 response factors of organic contaminants.18 Since MS signals are not absolute, RMRFs were determined using D-PHE as internal standard. The average molar response factor relative to D-PHE of the test set of 33 substances was 0.47. The overall variation of the RMRFs was 50% for the test set (Table 2). Although this is a significant variation, it is not very large, when compared to the overall variability of the method (9% to 55%, see below). Information on molar concentrations, which are accurate within a factor of 2, is still highly relevant for environmental risk assessment purposes.18 Recovery rate values of the fortified PRCs from SPMDs were good and reproducible. Average percentage recoveries of PRCs varied from 50% to 100% and the relative standard deviation of three spiked samples did not exceed 10% for any PRC used. The analysis of procedural spikes (Table 2, last column) showed that elevated component volatility causes low recovery of accumulated analytes from SPMDs due to their partial loss during sample transport and processing. There is also a trend of decreasing precision of the entire sampling, cleanup and analytical procedure with decreasing boiling point of the analyte. Acceptable recoveries (> 50%) were determined for analytes with normal boiling point higher than 200 °C and thus only concentrations of identified semivolatile compounds with a normal boiling point higher than 200 °C were reported in this study. For quantitative recovery of more volatile compounds from SPMDs, a specific sample treatment would be required, e.g. application of purge and trap techniques. The results of the analysis of SPMD extracts after 20 days of groundwater exposure are presented in Fig. 4 and are reported in detail in the Supplementary Information.! On the basis of total semivolatile contaminant residues, the wells can be ranked from lowest to highest as follows: SafBit 2/96 (quat.), SafBit 31/98, SafBit 2/96 (tert.), SafBit 16/97, SafBit 30/98, and GWM 19/91. The observed extreme concentration gradient of contamination is a clear indicator of a low mobility of hydrophobic semivolatile contaminants in the aquifer over a short distance of less than 760 m between the two most distant sampling wells. The semiquantitative contaminant concentrations in SPMD extracts (given as a sum of all quantifiable semivolatile substances) ranged from 0.4 nmol per SPMD from well SafBit 31/ 98 to 20 umol per SPMD from well GWM 19/91. The average I = SPMD matrix impuriti 1E-4 1E-5- D 2(2) I - identified contaminants 3(6) 1(1) 1 D Blank 1 SafBit 1 SafBit 1 SafBit 1 SafBit 1 SafBit 1 GWM 30/98 31/98 2/96 2/96 16/97 19/91 (tert.) (quat.) Sampling groundwater well Fig. 4 Molar amounts of semivolatile compounds (boiling point > 200 °C) identified in extracts from SPMDs deployed for 20 days in groundwater monitoring wells in Bitterfeld. Floating bars show the concentration range determined in two samplers exposed side by side. The numbers above the bars denote the number of components quantified in the extracts followed by the total number of components identified (in brackets). Detailed information is listed in the Supplementary Information, t relative percentage difference between total concentrations of contaminants in duplicate SPMD samples from the same sampling site was in the range 14% (SafBit 31/98) to 61% (SafBit 30/98). The cummulative uncertainty uc (%) of the method employed for semivolatile organic chemicals sampling using SPMDs and their semiquantitative analysis can be estimated from the uncertainties of each sampling or analytical step (groundwater sampling, extraction, cleanup and gas chromatography), c(%) = {u (4) where us represents the uncertainty of sampling, ue represents the uncertainty of the extraction and cleanup procedure and «a represents the uncertainty of the analytical procedure. The uncertainty of sampling us can be deduced from the average relative percent difference of two parallelly deployed sampling devices, which varied between 14% and 61%. The uncertainty of the extraction and cleanup step can be estimated as the average relative standard deviation of extraction recovery of spiked samples. This was lower than 30% for the tested analytes with a boiling point higher than 200 °C (ue x 30%). Finally, the uncertainty of the analytical procedure was estimated by calculating the overall variation of the RMRFs for the test set of chemicals (ua « 50%). Thus, the overall uncertainty of the method employed in this study is expected to vary between 60% and 80%. As a result, the method gives semiquantitative information about the concentration levels with a precision within one order of magnitude. Although this is a relatively low precision, it is sufficient for a preliminary characterization of the pollution situation at the sampling sites. Further, substance-specific method validation for major identified (and environmentally relevant) analytes will enable a substantial improvement of the method precision. SPMDs have been developed as kinetic passive samplers, which integratively accumulate contaminants over a prolonged time period (days or weeks). Using known kinetic parameters, it is possible to calculate time-weighted average concentrations of the contaminants in the sampled medium from the amounts accumulated in the SPMD and the exposure time.7 There is sufficient evidence that the exchange kinetics of most organic analytes between SPMD and water can be described by first-order kinetics.20 Moreover, the kinetics are isotropic, i.e. both uptake and loss of an analyte are governed by the same mass transfer law. However, the sampling kinetics are affected by many factors including the physicochemical properties of the sampled analytes as well as the environmental conditions. To estimate the in situ sampling kinetics of SPMDs in groundwater in this study, the performance reference compound (PRC) approach was used. This approach was developed by Huckins et al.20 to enable estimation of exchange kinetics of contaminants between SPMDs and the sampled medium. PRCs are analytically non-interfering organic compounds that are added to the sampling device prior to exposure. The release of a PRC from the SPMD, when the concentration of this compound in groundwater is negligibly low (i.e. Cw = 0), can be described by a first-order-decay equation CsPMD(i) — CspMD(0)exP(~ ket) (5) Here, CSPMD is the PRC concentration in the SPMD, ke is the first-order exchange rate constant, which is also called the overall exchange coefficient and t stands for time. Assuming that isotropic exchange kinetics can be applied and that SPMD-water partition coefficients are known, measurement of PRC first order elimination rate constants ke during SPMD exposure permits estimation of the sampling rate, i.e. the volume of water that the SPMD has the potential to clear per day. J. Environ. Mo nit., 2005, 7, 500-508 505 Huckins et al?x showed that the daily SPMD clearance volume or sampling rate Rs is related to ke by 0.10' ■ keK< e^spmd V spmd (6) where ^SPMD is the SPMD-water partition coefficient and Fspmd is the SPMD volume. Measuring the sampling kinetics by taking samples in a time series appeared to be impractical because of a number of physical limitations. A maximum of two cages can be deployed at the depth of the screen, where groundwater is exchanged between the borehole and aquifer. Further, the manipulation of cages during the sampling would disturb the conditions of groundwater in the well. Moreover, the decreasing number of samplers in the well during the sampling period would likely cause temporal changes in analyte water concentrations, thus making the modelling of the sampling kinetics more complicated. Therefore, knowing the type of exchange kinetics from the literature, it was sufficient to measure PRC levels in SPMDs only at the beginning and the end of the field exposure. D-PAHs were used as PRCs in this study. To calculate the apparent first order exchange rate constant ke, eqn. (5) was solved to permit a two-point derivation of ke (assuming first-order kinetics): ln[C, spmd(0)/CsPMD(i) (7) Blanks spiked with D-PAHs were used to determine CSpmd(o)-A significant decrease of concentration in SPMD extracts during exposure was determined for PRCs with log KOVJ < 4.5. On the other hand, no significant decreases in the concentrations of 2H10-fluoranthene, 2H10-pyrene, and 2H12-benz (a)anthracene were observed after exposure in any of the SPMD samples. The calculated values of ke for PRCs determined for different sampling wells are shown in Fig. 5. Partitioning of chemicals between SPMD and groundwater The PRC approach can be applied to estimate appropriate exposure times needed to achieve equilibrium between SPMD and water for a specific group of compounds. The time required to reach 90% of the equilibrium concentration for uptake of contaminants or to offload 90% of the PRC from an SPMD can be considered as an approximation of equilibration time teq.22 This can be calculated using eqn. (7). PRCs having an elimination rate constant ke of 0.115 d~1 or higher are expected to achieve partitioning equilibrium between SPMD and groundwater within 20 days of exposure. The maximum log Kow value allowed for a substance in groundwater to achieve partitioning equilibrium within this time period was calculated by interpolation from the linear dependence ke = /(log Kovl) for each well as a value corresponding to ke = 0.115 d_1. Obtained threshold log Kovl values are given in Table 3. Compounds having log Kow equal to or lower than the threshold value achieve equilibrium partitioning between groundwater and SPMD within 20 days of exposure and their groundwater concentration can be estimated using the equilibrium partitioning model Cw — C< spmd/^spmd (8) For the remaining compounds, ambient groundwater concentrations can be estimated using a kinetic model described by Huckins et al?x Cw — Cspmd/^spmdO — exp[—fceř]) (9) SPMD extracts from wells GWM 19/91 and 2/96 (tert.) contained large amounts (50% and 60% on a molar basis) of chemicals with log Kovl higher than the calculated threshold log For these sites, additional accumulation of chemi- 0.08 0.06 ■ -& 0.04 - 0.03- 0.02- 0.01 - 0.00- ■ □ GWM 19/91 ■ SafBit 16/97 - A SafBit 2/96 (quat.) ▲ SafBit 2/96 (tert.) O SafBit 31/98 - • SafBit 30/98 I 1 I 1 ■ I ■ I ■ 4.1 4.2 4.3 4.4 4.5 4.6 log K Fig. 5 Dependence of the apparent first-order elimination rate constant (ke) of performance reference compounds (2H10-biphenyl, 2H10-fluorene, 2H10-phenanthrene, 2H10-anthracene) in sampling wells on the octanol-water partition coefficients Kovt. The lines correspond to linear regression analysis of the data (see Table 3). No significant decreases in the concentrations of 2Hi0-fluoranthene, 2Hi0-pyrene and 2Hi2-benz(a)anthracene were observed after exposure in any of the SPMD samples. cals from groundwater to SPMDs is expected during prolonged exposure periods. For the remaining sampling sites, no significant additional increase in total molar concentrations is expected after sampling longer than 20 days and the equilibrium partitioning model can be applied for estimating their concentration in groundwater from the levels accumulated in SPMDs. SPMD sampling kinetics limitation due to groundwater flow As can be seen in Fig. 5, differences in PRC release kinetics between different sampling wells are evident. Environmental variables including water velocity/turbulence, fouling and temperature may affect the exchange kinetics of SPMDs.23'24 We suppose that these factors only marginally contributed to the observed differences in contaminant uptake kinetics. It has been shown25 that for chemicals with log Kovl < 4.0 uptake rates are probably controlled by the diffusion through the polyethylene membrane rather than through the aqueous boundary layer at the SPMD surface, and therefore are not subject to effects caused by fluctuating hydrodynamic conditions. For these compounds there is little reason to believe that the observed variability in the data is due to exposure conditions other than the depletion of ambient environmental levels of measured analytes. Thus, it is more likely that the SPMD uptake is limited by groundwater flow.4 SPMDs are very efficient extractors with typical sampling rates of several litres of water per day.21 The model (eqn. (5)) describing the uptake of analytes assumes a constant concentration in the water surrounding the sampler and is relevant to the surface water situation when the water surrounding the membrane is exchanged quickly. This model may be inappropriate for use in groundwater in wells with low flow and low volume of the filtered zone (the span where samplers are deployed during the sampling). Permeability in a fine-grained aquifer can be very low, which may result in the depletion of target solutes at the membrane surface due to depletive sampling. As a consequence of the decreasing water concentration surrounding the SPMD due to depletion by the sampler, the 5 06 J. Environ. Monit., 2005, 7, 500-508 Table 3 Summary of the linear regression analysis of the dependence of the apparent elimination rate constant (ke) of performance reference compounds as dependent on the octanol-water partition coefficient using ke = A + B x log Kovl and estimated maximum log Kovl threshold value for a substance in groundwater to achieve equilibrium partitioning during 20 days of SPMD exposure in groundwater (also shown in Fig. 5). The analysis was performed with 4 PRC compounds (2Hi0-biphenyl, 2Hi0-fiuorene, 2Hi0-phenanthrene and 2Hi0-anthracene; N = 4) Sampling well A B Correlation coefficient log Kovl threshold GWM 19/91 0.48 -0.11 -0.99 3.5 SafBit 16/97 0.38 -0.08 -0.99 3.2 SafBit 2/96 (quat.) 0.67 -0.15 -0.99 3.8 SafBit 2/96 (tert.) 0.21 -0.05 -0.98 2.1 SafBit 31/98 0.76 -0.17 -0.99 3.9 SafBit 30/98 0.63 -0.14 -0.99 3.8 SPMD can accumulate analytes at a rate (contaminant amount per day) lower than that expected from the laboratory-derived sampling rate value Rs (determined at constant aqueous concentrations). To verify this, the groundwater flux in sampling wells Q was compared with the actual daily clearance volumes of SPMDs deployed in groundwater wells, represented by the in situ sampling rate Rs. These were calculated from the release kinetics of PRCs in each sampling well. Groundwater flux in sampling wells was estimated using Darcy's Law:4 Q = KiA (10) where K is the hydraulic conductivity, i is the hydraulic gradient and A is the cross-sectional area, assumed to be the SPMD deployment device length (ca. 1 m) times twice the diameter of the well.4 The permeability and gradient parameters were available from previous site assessments and were not obtained specifically for this study (Table 1). Calculated groundwater flow ranged from approximately 1 L d_1 in the well SafBit 16/97 to 5.3 L d"1 in the well SafBit 30/98. Due to a lack of information about the geophysical parameters, the calculation could not be performed for wells GWM 19/91 and SafBit 2/96 (quat). Rs values between 0.9 and 5.2 L d_1 were measured by Huckins et al?x (for PAHs with log Kovl < 5.3 and a standard SafBit 16/97 2/96 (tert.) 31/98 30/98 0 1 2 3 4 5 6 Q (L/d) Fig. 6 Dependence of estimated sampling rates of polycyclic aromatic hydrocarbons (Rs) on the groundwater flow (daily turnover volume; Q) in the sampling wells. Rs values were calculated with eqn. (6) using ke values from the PRC dissipation rate and log XSPMD values published by Huckins et at21 The solid line represents Q = Rs- SPMD at constant aqueous concentration under quiescent flow conditions and a temperature similar to that in groundwater at sampling sites). These published sampling rates are comparable to or smaller than turnover rates actually encountered in groundwater wells examined in this study. For the actual in situ Rs calculation, ke values from the dissipation rate of PRCs during SPMD exposure and -Kspmd values of PAHs published by Huckins et al?x were utilized and Rs values were calculated using eqn. (6). Fig. 6 confirms that the estimated in situ Rs values are affected by groundwater flow when the daily turnover volume in monitoring wells was lower than approximately 3 L d_1. Under low flow conditions, when the water in wells is refreshed slowly, the aqueous concentration of contaminants will not remain constant during the whole SPMD deployment period as the SPMD removes contaminants from surrounding water. Thus, the in situ extraction potential of the SPMD is limited by groundwater flow, when the exchange volume of well water during an exposure is lower than the SPMD clearance volume for the analytes. This limitation has to be taken into account when applying SPMDs as integrative passive samplers in groundwater. To avoid the complications caused by the possibility of a depletive SPMD extraction, the use of smaller SPMDs (with corresponding lower clearance volumes) is recommended for groundwater sampling. However, to apply the linear uptake model (non-depletive extraction), it must be assured that the calculated SPMD sampling rate is much lower than the daily groundwater turnover volume in the well. Alternatively, passive samplers with very low clearance volumes can be used, such as ceramic dosimeters.26 Recently a promising sampler design called Membrane Enclosed Sorptive Coating (MESCO) has been developed.27 This integrative passive sampler is based on non-depletive extraction. Generally, MESCO clearance volumes are lower than 1 ml h_1. Despite the low sampling rate, the sensitivity of this device is comparable to that of the SPMD, because the total amount of analyte sequestered by the MESCO during deployment can be transferred to the GC system, whereas only a small portion of the SPMD extract is usually injected into the GC (to prevent introduction of large amounts of interfering contaminants to the chromatographic system). Conclusions This study demonstrates performance of a procedure combining groundwater passive sampling using semipermeable membrane devices with chemical analysis of accumulated contaminants. Sampling the groundwater present in the screen zone using SPMDs provides the greatest chance of obtaining samples without increased turbidity and with minimal alteration of the groundwater chemistry caused by sampling. The SPMD method eliminates the need to dispose of potentially highly contaminated wastewaters produced by purge-type sampling methods. Cross contamination of samples is reduced by the use of SPMDs because the sampling equipment does not come into contact with water from multiple wells, unlike J. Environ. Mo nit., 2005, 7, 500-508 507 sampling conducted with non-dedicated pumps, tubing, or bailers. SPMDs have a minimal effect on water circulation within a well and thus preserve any stratification of water, whereas purging can induce vertical mixing of the water. Thus, SPMDs have the potential to provide representative concentrations of aqueous contaminants as they exist in the undisturbed subsurface. Although the methods employed in this study need further validation, our investigation provided a valuable informative basis about the character of pollution with semivolatile organic compounds in order to target future method calibration at the most relevant identified contaminants. The methodology demonstrated in this study is applicable for semivolatile organic groundwater contaminants. For accurate monitoring of a broad spectrum of contaminants, including volatile organic chemicals, a modification of the sample treatment procedure would be required, e.g. by application of purge and trap techniques. Alternatively, several passive samplers with a complementary selectivity, e.g. SPMD and PDB, can be deployed for screening/monitoring in multicomponent pollution situations. Use of sampling devices with low clearance volume is recommended to prevent limitation of the extraction potential by groundwater flow. Acknowledgements The authors would like to thank Ahmad Al-hallak, Petra Keil and Uwe Schröter for their assistance in field sampling and instrumental measurements, Ralf-Uwe Ebert for calculation of the physicochemical properties of groundwater contaminants, and Bernd Feist for his technical advice. References 1 H. Liu and G. Amy, Environ. Sei. Technol., 1993, 27, 1553. 2 L. V. Parker, Ground Water Monit. Rev., 1994, 14, 130. 3 J. Sevee, C. A. White and D. J. Maher, Ground Water Monit. Rev., 2000, 20, 87. 4 K. E. Gustavson and J. M. Harkin, Environ. Sei. Technol., 2000, 34, 4445. 5 D. A. Vroblesky and T. R. Campbell, Adv. Environ. Res., 2001, 5, 1. 6 J. Namiesnik, B. Zabiegala, A. Kot-Wasik, M. Partyka and A. Wasik, Anal. Bioanal. Chem., 2005, 381, 279. 7 J. N. Huckins, J. D. Petty, J. A. Lebo, C. E. Orazio, H. F. Prest, D. E. Tillitt, G. S. Ellis, B. T. Johnson and G. K. Manuweera, in Techniques in Aquatic Toxicology, ed. G. K. Ostrander, CRC Press, Lewis Publishers, Boca Raton, FL, 1996, pp. 625-655. 8 E. M. J. Verbruggen, W. H. J. Vaes, T. F. Parkerton and J. L. M. Hermens, Environ. Sei. Technol., 2000, 34, 324. 9 J. N. Huckins, G. K. Manuweera, J. D. Petty, D. Mackay and J. A. Lebo, Environ. Sei. Technol, 1993, 27, 2489. 10 H. Weiss, B. Daus, P. Fritz, F.-D. Kopinke, P. Popp and L. Wünsche, in Groundwater Quality: Remediation and Protection, ed. M. Herbert and K. Kovar, IAHS Publication No. 250, Wall-ingford, UK, 1998, pp. 443-450, also http://www.ufz.de/ index.php?en=1580. 11 W. M. Meylan, KOWWIN 1.66, Syracuse Research Corporation, Syracuse, NY, USA, 2000. 12 W. M. Meylan, MPBPWIN v 1.40., Syracuse Research Corporation, Syracuse, NY, USA, 1999. 13 J. Dermietzel and G. Christoph, Water Air Soil Pollut., 2001, 125, 157. 14 B. Vrana, A. Paschke, P. Popp and G. Schüürmann, Environ. Sei. Pollut. Res., 2001, 8, 27. 15 A. Kolbe, S. Geiss and J. W. Einax, Acta Hydrochim. Hydrobiol., 1999, 27, 58. 16 W. P. Eckel, Am. Lab., 2000, 32, 17. 17 C. E. Rostad and W. E. Pereira, HRC & CC J. High Resolut. Chromatogr. Chromatogr. Commun., 1986, 9, 328. 18 M. G. M. van Loon, F. G. Wijnker, M. E. Verwoerd and J. L. M. Hermens, Anal. Chem., 1996, 68, 2916. 19 J. D. Petty, J. N. Huckins and J. L. Zajicek, Chemosphere, 1993, 27, 1609. 20 J. N. Huckins, J. D. Petty, J. A. Lebo, F. V. Almeida, K. Booij, D. A. Alvarez, W. L. Cranor, R. C. Clark and B. B. Mogensen, Environ. Sei. Technol., 2002, 36, 85. 21 J. N. Huckins, J. D. Petty, C. E. Orazio, J. A. Lebo, R. C. Clark, V. L. Gibson, W. R. Gala and K. R. Echols, Environ. Sei. Technol, 1999, 33, 3918. 22 R. W. Gale, Environ. Sei. Technol, 1998, 32, 2292. 23 J. N. Huckins, J. D. Petty, H. F. Prest, R. C. Clark, D. A. Alvarez, C. E. Orazio, J. A. Lebo, W. L. Cranor and B. T. Johnson, in A guide for the use of semipermeable membrane devices (SPMDs) as samplers of waterborne hydrophobic organic contaminants, report for the American Petroleum Institute (API), API Publication 4690, API, Washington, DC, 2000. 24 B. Vrana and G. Schüürmann, Environ. Sei. Technol, 2002, 36, 290. 25 K. Booij, H. M. Sleiderink and F. Smedes, Environ. Toxicol. Chem., 1998, 17, 1236. 26 H. Martin, B. M. Patterson, G. B. Davis and P. Grathwohl, Environ. Sei. Technol, 2003, 37, 1360. 27 B. Vrana, P. Popp, A. Paschke and G. Schüürmann, Anal. Chem., 2001, 73, 5191. 508 J. Environ. Monit., 2005, 7, 500-508 Príloha 9 Vrana B., Paschke A., and Popp P., Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water, Environ. Pollut, 2006,144, 296-307. ^JSĚXSĚI- Availableonlineatwww.sciencedirect.com ^ri^nrpDirprť environmental #5Sl pollution ■ ' '___k_-' ELSEVIER Environmental Pollution 144 (2006) 296-307 www.elsevier.com/locate/envpol Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water Branislav Vrana a'*, Albrecht Paschke b, Peter Poppc School of Biological Sciences, University of Portsmouth, King Henry Building, King Henry I Street, Portsmouth, Hampshire POl 2DY, UK Department of Ecological Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany c Department of Analytical Chemistry, UFZ Centre for Environmental Research, Permoserstrasse 15, D-04318 Leipzig, Germany Received 1 July 2005; accepted 14 November 2005 A robust calibration method of a passive sampling device for monitoring of persistent organic pollutants in water is described. Abstract Membrane-enclosed sorptive coating (MESCO) is a miniaturised monitoring device that enables integrative passive sampling of persistent, hydrophobic organic pollutants in water. The system combines the passive sampling with solventless preconcentration of organic pollutants from water and subsequent desorption of analytes on-line into a chromatographic system. Exchange kinetics of chemicals between water and MESCO was studied at different flow rates of water, in order to characterize the effect of variable environmental conditions on the sampler performance, and to identify a method for in situ correction of the laboratory-derived calibration data. It was found that the desorption of chemicals from MESCO into water is isotropic to the absorption of the analytes onto the sampler under the same exposure conditions. This allows for the in situ calibration of the uptake of pollutants using elimination kinetics of performance reference compounds and more accurate estimates of target analyte concentrations. A field study was conducted to test the sampler performance alongside spot sampling. A good agreement of contaminant patterns and water concentrations was obtained by the two sampling techniques. © 2006 Elsevier Ltd. All rights reserved. Keywords: Organic pollutants; Passive sampling; Semipermeable membrane devices; Water monitoring 1. Introduction Monitoring of pollution of ecosystems by persistent organic pollutants (POPs) is an ongoing challenge for the analytical chemist. For qualitative and quantitative assessment of pollution, a large number of samples must be taken from a given location over the entire monitoring period, when spot/grab sampling is applied as the method of choice. This approach is time-consuming, laborious and can be very costly. Grab * Corresponding author. Tel.: +44 23 92 84 2024; fax: +44 23 92 84 2070. E-mail address: bran.vrana@port.ac.uk (B. Vrana). samples provide information only about the situation in the moment of sampling and may fail to account for episodic contamination events. Solutions for such situation are methods of passive sampling and/or extraction of analytes, which involve measurement of any analyte as a weighted average over the sampling time. The concentration of analyte is integrated over the whole exposure time, making such a method less bias-prone to fluctuations of pollutant concentrations. Long-term overview of pollutant levels at the sampling site is obtained in this way. Passive monitors are rapidly gaining wide acceptance for assessing time-weighted average (TWA), concentrations in aquatic systems. The current state-of-the art 0269-7491/$ - see front matter © 2006 Elsevier Ltd. All rights reserved, doi: 10.1016/j .envpol.2005.11.046 B. Vrana et al. I Environmental Pollution 144 (2006) 296—307 297 of passive sampling and extraction methods for long-term monitoring of environmental pollutants has recently been published by Namiesnik et al. (2005). The common disadvantage of most passive sampling techniques is a laborious recovery of analytes from samplers using solvent extraction. To make the passive sampling technology more suitable for routine monitoring, low-cost and less time-consuming sample processing methods are required. Sample processing with reduced solvent consumption would also minimize the risk of sample contamination during handling in the laboratory and enable to improve quality control measures. Recently, a solventless and simple technique for preconcen-tration of organic solutes from aqueous matrixes, the stir bar sorptive extraction (SBSE), was developed by Baltussen et al. (1998, 1999). The applicability of this extraction technique has been demonstrated for determination of polycyclic aromatic hydrocarbons (PAHs) and organochlorine pesticides (OCPs) in water (Popp et al., 2001, 2003; Leon et al., 2003). SBSE is suitable also for analysing real environmental samples including drinking water (Garcia-Falcon et al., 2004a), run-off water (Garcia-Falcon et al., 2004b) and precipitation water (Niehus et al., 2002). The method is very sensitive, with detection limits well below 10 ng LT1 level. Garcia-Falcon et al. (2004b) have shown that, in contrast to other extraction techniques, SBSE is suitable for determination of freely dissolved fraction of PAHs in environmental water samples. The determination of freely dissolved fraction of contaminants in water is important especially for assessment of organism exposure and bioavailability. Absorptive partitioning is the predominant extraction mechanism of analytes into poly(dimethylsiloxane) (PDMS), the sorptive material used in SBSE. Although SBSE was originally developed as method for batch extraction of water samples, we recently described an adaptation of SBSE for long-term continuous passive sampling of persistent organic pollutants in water (Vrana et al., 2001b). This so-called MESCO (membrane-enclosed sorptive coating) sampler consists of a stir bar coated with a thin PDMS layer [Gerstel Twister, a commercially available device used for SBSE (Baltussen et al., 1998)] enclosed in a water-filled dialysis membrane bag from regenerated cellulose. After exposure of the sampler, the PDMS coated stir bar is taken from the enveloping membrane and can be directly analysed by thermo-desorption—GC—MS. The performance of the MESCO sampler had been demonstrated for integrative sampling of hydrophobic persistent organic pollutants including y-hexachlorocyclohexane (y-HCH), hexachlorobenzene (HCB), 2,2'-bis(4-chlorophenyl)-l,l'-dichloroethylene (DDE), PAHs, andpolychlorinatedbiphe-nyls (PCBs) (Vrana et al., 2001b). In general, the sampler performance depends on the sampler design, physicochemical properties of the sampled analyte and the environmental conditions. To be able to apply laboratory-derived calibration data for calculation of TWA water concentrations in the field, it is necessary to consider (or determine) the effect of environmental variables, including temperature, hydrodynamics and biofouling, on the sampler performance. Because it appears impractical to conduct calibration studies for all exposure scenarios (e.g. for many combinations of temperature and water turbulence), a novel in situ calibration approach was developed by Huckins et al. (2002) for lipid-filled semipermeable membrane devices (SPMDs), passive samplers with working principle similar to MESCO. This involves the use of performance reference compounds (PRC), which are analytically non-interfering organic compounds with moderate to low affinity to the passive sampler that are added to the receiving phase (in our case to the PDMS) of the sampler prior to membrane enclosure. This approach is based on theory and experimental evidence that PRC dissipation rate constants are related to the uptake rates of target compounds. In order to test the robustness of MESCO performance against variable environmental conditions, exchange kinetics of PAHs, OCPs and PCBs between water and MESCO were studied under condition of varying flow rate. The PRC approach was tested to identify a method for in situ correction of the laboratory-derived calibration data. Also, a field study was conducted to test the sampler performance alongside spot sampling. 2. Theory The mass transfer of an analyte in a sampler includes several diffusion and interfacial transport steps across all barriers, i.e. the stagnant aqueous boundary layer, possible biofilm layer, the membrane, the inner aqueous phase, and the receiving organic phase. It has been shown that, the amount of the chemical accumulated in the MESCO sampler from water with constant chemical concentration can be described by (Vrana et al., 2001b): Ms (ř) = Mo + (Cw^swVs - Mo) 1 — exp kmAa (1) where Ms is the mass of analyte in the receiving phase (PDMS), Mo is the amount of analyte in the sampler at the start of the exposure, Cw represents the water concentration during the deployment period, Ksw is the receiving phase/water distribution coefficient, Vs is the volume of the receiving phase, kov is the overall mass transfer coefficient, A is the membrane surface area, a is the pore area of the membrane as fraction of total membrane area (membrane porosity), and t equals time. The coefficient in the exponential function is referred to as the overall exchange rate constant ke. k, =■ kovAa (2) In the initial uptake phase, when the exponential term is very small (<<1) or Ms/ysCw << Ksw, chemical uptake is linear or integrative. Thus, in the linear region Eq. (1) can be reduced to Ms (?) = M0 + Cy/kmAat (3) For practical application, the Eq. (3) can be rewritten Ms(t)=M0 + CwRst (4) B. Vrana et al. I Environmental Pollution 144 (2006) 296—307 298 where Rs is the sampling rate of the system. Rs = kmAa = keKSy,Vs. (5) Adding PRCs to the receiving phase prior to exposure of the passive sampler has been suggested as a means to calibrate the exchange rates in situ (Booij et al., 1998; Huckins et al., 2002). When PRCs are used that are not present in water (Cw = 0), Eq. (1) reduces to Ms(t) = M0exp(-ket) (6) which is a one-parameter equation, because the amount of PRC added to the MESCO sampler (M0) is known. 3. Experimental 3.1. Materials and chemicals Test chemicals (Table 1) included several groups of persistent organic pollutants: y-hexachlorocyclohexane (y-HCH), hexachlorobenzene (HCB), 2,2'-bis(4-chlorophenyl)-l,l'-dichloroethylene (DDE), PAHs and PCBs. y-HCH reference material was obtained from Riedel-de Haen. HCB, DDE and PAH reference materials were obtained from Dr. Ehrenstorfer. PCB reference material and test chemicals in high purity (>99%; y-HCH, HCB, DDE, PAHs and PCBs) were purchased from Promochem. Perdeuterated polycyclic aromatic Table 1 Selected physico-chemical properties of test analytes at 25 °C No. Compound MW1 log logiff" (gnioP1) ^ow (gnC3) (PDMS) 1 HCB 284.8 5.5 0.005 4.4e 2 Y-HCH 290.8 3.7 7.3 2.6e 3 p,p'-DDE 318.0 5.7 0.04 5.3e 4 PCB 28 257.5 5.6 0.16 4.8e 5 PCB 52 292.0 6.1 0.03 5.1e 6 PCB 101 326.4 6.8 0.01 5.5e 7 PCB 138 360.9 7.6 0.0015 5.7e 8 PCB 153 360.9 7.8 0.001 5.7e 9 PCB 180 395.3 8.3 0.0003 5.6e 10 Acenaphthylene 152.2 4.0 16.1 3.40f 11 Acenaphthene 154.2 4.0 3.8 3.63f 12 Fluorene 166.2 4.2 1.9 3.71f 13 Anthracene 178.2 4.6 0.045 3.98f 14 Phenanthrene 178.2 4.5 1.10 3.96f 15 Fluoranthene 202.3 5.1 0.26 4.71f 16 Pyrene 202.3 5.1 0.132 4.86f 17 Benzo[a]anthracene 228.3 5.9 0.011 5.26f 18 Chrysene 228.3 5.7 0.0019 5.69f 19 Benzo [b] fluoranthene 252.3 5.8 0.0015 5.17f 20 Benzo [k] fluoranthene 252.3 6.0 0.0008 5.33f 21 Benzo[a]pyrene 252.3 6.2 0.0038 5.39f 22 Indeno[ 1,2,3-crf]pyrene 276.3 6.8 0.0005 4.28f 23 Benzo [g, h, i] pery lene 276.3 6.9 0.0003 4.43f a Molecular weight (MW). b Octanol—water partition coefficient Kow (Mackay et al., 1992). c Aqueous solubility S (Mackay et al., 1992). d PDMS/water distribution coefficient. e Data from Paschke and Popp (2003). f Data from Doong and Chang (2000). hydrocarbons (D-PAHs) were obtained from Promochem. Phys-icochemical properties of test analytes are given in Table 1. Dialysis membrane Spectra/Por 6 (molecular weight cutoff 1000 Da) was obtained from Spectrum Laboratories. Twister™ stir bar for sorptive extraction was obtained from Gerstel. Lichrolut (R) (diameter of particles 40—63 (im) was purchased from Merck. The solvents methanol and hexane were used in LiChrosolv quality from Merck. 3.2. Sampler design The passive sampling device, referred to as the Membrane-Enclosed Sorptive Coating sampler (MESCO) has been described previously (Vrana et al., 2001b). It consists of a GERSTEL-Twister™ bar used for SBSE enclosed in a dialysis membrane bag made from regenerated cellulose (Spectra/ Por 6, molecular weight cutoff 1000 Da, 18 mm flat width, 30 mm length; component). Twister is a stir bar (15 mm length) consisting of a magnetic core sealed inside a glass coated with 22 mg PDMS. The PDMS sorptive layer (receiving phase) is 500 (im thick and its volume is 24 (iL. 3.3. Sampler preparation Prior to use, the stir bar was placed into a vial containing 1 mL of a 1:1 mixture of methylene chloride and methanol, and treated for 5 min with sonication. Then the solvent mixture was rejected and the procedure repeated three times. The stir bar was dried in a desiccator at room temperature. Prior to each use, the stir bar was conditioned by heating for 180 min at 280 °C with a nitrogen stream of about 100 mLmfrr1. Perdeuterated PAHs were utilised as PRCs. For loading the Twister stir bars with (PRCS), 20 mL of aqueous solution of containing 2H10-biphenyl (D10-BIP), 2H10-fluorene (D10-FLU), 2H10-phenanthrene (D10-PHE), 2H10-anthracene (D10-ANT), 2Hi0-fluoranthene (D10-FLT), 2H10-pyrene (D10-PYR) and 2H12-benz(a)anthracene (D12-BaA) was pipetted to a 25 mL amber glass vial with a flat base with a screw cap. The solution was prepared by spiking bi-distilled water with a PRC-mixture dissolved in methanol to give nominal concentration of individual analytes of 1 |igL~\ The vial was then placed on a magnetic stirrer. The pre-cleaned Twister stir bar was placed in the vial with the PRC solution and stirred at 1000 min-1 for 30 min at room temperature. In order to accelerate the procedure, up to six Twister bars were loaded in parallel. Following the loading with PRCs, Twisters were washed with bi-distilled water, dried with a soft paper tissue and stored closed in an amber glass GC vial in the freezer until use. For sampler assembly, the Twister was placed inside the dialysis membrane bag. The bag was filled with 3 mL of bi-distilled water and sealed at each end with 35 mm Spectra Por enclosures. As a direct relationship exists between the surface area and the rate of uptake, the area of the membrane was held constant at 1100 mm2. To enable a simultaneous exposure of a series of samplers, they were connected to a string, which B. Vrana et al. I Environmental Pollution 144 (2006) 296—307 299 was then exposed to organic analytes in a continuous-flow system. 3.4. Flow-through exposures MESCO samplers were exposed to test chemicals at a nominal concentration of 20ngL~1 in a flow-through exposure system. Exposures were conducted at 19 °C. The experimental conditions of individual exposures are given in Table 2. The experimental setup of the flow-through exposure system has been described (Vrana et al., 2001b; Vrana and Schuiirmann, 2002). Briefly, exposure water was pumped from the bottom to the top of a 1 m high glass column with either 7.5 or 15 cm inner diameter. Test chemicals were dissolved in methanol and the appropriate amounts of stock solution were delivered into exposure water in a 1 L chamber positioned at the bottom of the column using a peristaltic pump. The water in the chamber was mixed using a magnetic stirrer. The methanol concentration in the exposure water was held constant at 0.01% (v/v). This setup enabled to vary the flow rate in the exposure column. The string of MESCO samplers was fixed in the column in a vertical position between the top and the bottom of the exposure column. Exposures were conducted at linear flow velocities of 8, 35 and 68 cmmin-1. The exposures lasted up to 10 days, during which the samplers were sampled at time intervals and their contents analysed to determine accumulated concentrations of test chemicals as described below. Duplicate water samples from the exposure column (1 L) were taken at each time when samplers were sampled and analyte concentration in water was determined. 3.5. Field performance test To assess the performance of MESCO for monitoring POPs in the field, samplers were deployed in a river. The sampler data were compared with spot sampling. The sampling site was located in the stream Spittelwasser flowing through a highly polluted industrial area of Bitterfeld in Saxony-Anhalt, Germany (Vrana et al., 2001a). The MESCOs were deployed for 20 days during summer 2000 (15th June—4th July). During the exposure, the water temperature at the sampling site varied from 18.9 to 20.5 °C. Four samplers were deployed at the sampling site. On the day of deployment, MESCOs were freshly prepared in laboratory and transported to the field in amber glass jars filled with bi-distilled water to Table 2 Summary of passive sampler flow-through exposure experimental conditions Experiment Flow velocity Exposure Number of no. (cmmin-1) period (h) MESCOs sampled 1 35 0-163 16 2 8 0-233 15 3 68 0-168 12 Exposures were conducted at 19 °C and 20 ngL 1 nominal analyte concentrations in water. prevent drying of the dialysis membrane during transport. At the sampling site, MESCOs were removed from the jars and placed into a protective deployment device made of a stainless steel conduit of 5 cm inner diameter with perforated surface (5 mm openings). The deployment device protected MESCOs from abrasion and protected the sequestered pollutants from light. The depth below the water surface at which devices were deployed was 20 cm. On day 20, MESCOs were removed from the deployment device, checked visually for mechanical damage and immediately sealed in individual amber glass jars filled with bi-distilled water. The jars were transported to the laboratory in a portable icebox (on ice and in darkness). Additional trip blank sampler was exposed to air while MESCOs were being deployed and collected. Trip blank was processed exactly as deployed samples and was used to define contamination of the MESCOs during transportation and handling. Two 2 L water samples were taken from the sampling site at the beginning and the end of the exposure period, extracted and analysed for contaminant content using solid-phase extraction technique. 3.6. Sampler processing Following exposure, MESCOs were dismantled, Twister bars were washed with bi-distilled water, dried with a paper cloth, checked visually for possible damage of the sorptive layer, and analysed for accumulated target analyte and PRC content by thermodesorption—GC—MS. 3.7. Processing of water samples The residues in the water samples from the calibration apparatus and river water samples were extracted using solid-phase extraction (SPE) using Lichrolut (R) sorbent or SPME technique as described earlier (Vrana et al., 2001b). 3.8. Instrumental analysis The quantitation of the compounds accumulated during exposures in Twister bars was performed by thermodesorption— GC—MS under conditions described previously (Vrana et al., 2001b). Briefly, thermodesorption—GC—MS was performed on an Agilent Technologies (Palo Alto, CA, USA) system equipped with a Gerstel (Mulheim/Ruhr, Germany) thermodesorption device TDS A. A cold injection system from Gerstel (CIS-4) with an empty liner was used for cryofocusing the analytes prior to the transfer onto the analytical column. The single ion monitoring (SIM) mode of the mass selective detector applying one or two characteristic ions per compound was chosen for the detection. For the external calibration, a small bunch of glass wool was positioned to an empty desorption tube. The desorption tube was then connected to the cool injector of a GC and flushed with 20mLmin~1 of nitrogen. The desorption tube with glass wool was then spiked with 2 uL of a calibration standard solution and flushed for 1 min by nitrogen stream to allow the solvent (hexane) to evaporate. The desorption 300 B. Vrana et at. I Environmental Pollution 144 (2006) 296—307 tube was then transferred to the thermodesorption device (TDS A) and processed by thermodesorption—GC—MS. Quantification of the residues sorbed on Twister bars was accomplished using a five-point external standard curve. Method quantification limit for the analytes under investigation ranged from 0.01 to 0.2 ng/Twister. 3.9. Data processing The experimentally determined time courses of the amounts of individual test substances on the MESCO sampler were fitted by linear regression analysis using Eq. (4). The adjustable parameters were the intercept (M0) and the slope (Cw x Rs) of the linear uptake curve Ms =f(t). Quality of the fit was characterized by the standard deviations of the optimized parameters, as well as the correlation coefficient adjusted for the degrees of freedom (r2 adjusted), the fit standard deviation, and the Fisher test criterion on the accuracy of the model. The sampling rates of the device Rs for individual test compounds were calculated by dividing the slope of the linear uptake curve by the mean aqueous analyte concentration during the exposure. The required variances of Rs values were calculated from the coefficients of variation of the uptake slope parameters and of the concentrations in the aqueous phase, according to the law of error propagation. The release of PRC from the MESCO sampler was fitted by non-linear regression analysis using Eq. (6) with M0 and ke as adjustable parameters. Quality of the fit was characterized by the standard deviations of the optimized parameters, as well as the correlation coefficient adjusted for the degrees of freedom (r2 adjusted), the fit standard deviation, and the Fisher test criterion on the accuracy of the model. Acenaphthene ■ Phenanthrene o Chrysene l—1—I—1—I—1—r 100 120 140 160 180 Time (h) Fig. 1. Uptake of selected PAHs by the TWISTER-based MESCO sampler. The data represent the 19 °C flow-through exposure at linear flow velocity of 35 cmmin-1 and nominal water concentration of analytes 20 ngL~'. The lines are predicted concentrations in the sampler obtained by linear regression using Eq. (4). during exposure did not exceed 30% of the average concentration for individual compounds. 4.1.2. Sampling rate The sampling rates Rs obtained in flow-through exposure experiments conducted at 20 ng L_1 nominal water concentration and 19 °C and various linear flow velocities are shown in 4. Results and discussion 4.1. Flow-through exposures 4.1.1. Uptake kinetics The performance of the sampler was tested by exposure to constant concentrations of test chemicals in a continuous-flow exposure tank at three various linear flow velocities. Concentrations of the analytes in water (Cw) and the amounts accumulated in the receiving phase (Ms) were two parameters measured regularly during the continuous-flow exposures. During exposure the water concentration was held constant, which was confirmed by analyses of water samples. Characteristic analyte uptake curves are shown in Figs. 1 and 2. Satisfactory linear regression fits of the Eq. (4) to the uptake data of analytes from water to MESCO were obtained for all test compounds in all experiments. Correlation coefficient (r) values of the regression ranged from 0.79 to 0.99 except of PCB138, fluoranthene and pyrene in experiment 2, for which r values ranged from 0.59 to 0.69. Coefficients of variation (CV) of the calculated slopes of uptake curves ranged from 5 to 25% with a few exceptions of PCB138, fluoranthene and pyrene in experiment 2, for which CV ranged from 33 to 40%. The maximum fluctuations of aqueous concentrations 180 Time (h) Fig. 2. Uptake of selected PCBs by the TWISTER-based MESCO sampler. The data represent the 19 °C flow-through exposure at linear flow velocity of 35 cmmin-1 and nominal water concentration of analytes 20 ngL~'. The lines are predicted concentrations in the sampler obtained by linear regression using Eq. (4). B. Vrana et al. I Environmental Pollution 144 (2006) 296—307 301 Table 3. Over the range of controlled laboratory conditions, the magnitude of Rs values differed by 10-fold (i.e. from 100 to 983 uLh-1). This range of sampling rates is narrow relative to the broad Kow range of nearly 5 orders of magnitude. This is in a good agreement with our earlier observations (Vrana et al., 2001b). 4.1.3. Release kinetics The release of PRCs from the MESCO sampler to water was fitted by non-linear regression analysis using Eq. (6) with M0 and ke as adjustable parameters. Fig. 3 shows the release kinetics of Dio-BIP under various flow conditions. Satisfactory fits of the first order decay Eq. (6) to the elimination data were obtained for D10-BIP and D10-FLU, with correlation coefficient (r2 adjusted for degrees of freedom) values of the regression (model versus experimental) between 0.79 and 0.93. Coefficients of variation of the ke for these compounds varied between 7 and 20%. For the remaining PRCs, satisfactory fit of the first order kinetic decay Eq. (6) to the PRC elimination data was obtained only for D10-PHE in experiments 1 and 2, and for D10-ANT in experiment 1, with r2 (adjusted) values between 0.72 and 0.73. Coefficients of variation of the ke for these compounds varied between 22 and 33%. The release of the remaining PRCs from the MESCO was too slow to statistically evaluate the release kinetics. The results of the first order decay fits were poor and estimates of ke values for D10-FLT, D10-PYR and D12-BaA were statistically not significantly different from zero (p = 0.95). 4.1.4. Verification of isotropic exchange kinetics: absorption versus desorption Assuming that the uptake rate of target analytes Rs and the exchange rate constant ke of its labeled analogue (PRC) are measured under the same conditions and that the distribution coefficient Ksw is measured at the same temperature, comparison of the Rs derived using the PRC elimination (Eq. (5)) to the directly measured Rs of the target analytes can be viewed as a check of the isotropic exchange kinetics. For this purpose, we simultaneously measured the sampling rate of native fluorene and phenanthrene, and the exchange rate kePRC of their deuterated analogues in three exposure experiments. These exchange coefficients were determined at 19 °C and at various water flow conditions. The Ksw values used for the estimation of PRC-derived Rs of fluorene and phenanthrene were approximated by PDMS/water distribution coefficients taken from the literature (Doong and Chang, 2000). These are listed in Table 1. Directly measured Rs and PRC-derived SS.PRC for MESCOs in experiment 1 were as follows: fluorene, Rs = 680 uL h_1 and ^s-prc = 341 [iLh-1; phenanthrene, Rs = 734 uLh-1 and ^sprc = 390 (iLh-1. For the experiment 2, sampling rate values for the same compounds were as follows: fluorene, Rs = 403 uL h 1 and Rs prc = 271 uL h 1; phenanthrene, Rs = 451 uL h 1 and RS-prc = 394 uL h_1. Finally, for the experiment 3, sampling rate values for the fluorene were as follows: Rs = 675 uL h 1 and Rs -prc = 246 uLh \ Because of the bad quality of the fit of the PRC elimination data, the comparison in this experiment for phenanthrene was precluded. Table 3 Summary of passive sampler sampling rates Rs derived from flow-through exposures at different flow velocities at nominal analyte concentration of 20 ng LT Flow velocity (cmmin-1) 8 35 35 68 Rs OrLh-1) C.V. (%) Rs (uLh-1) C.V. (%) Rsa0rLh-') C.V. (%) Rs (uLh-1) c.v. (%) Compound HCB 347 29 218 22 114 7 278 li Y-HCH 183 33 287 22 336 41 347 19 p,p-DDE 287 35 334 26 305 7 179 36 PCB28 545 30 568 23 305 49 546 54 PCB52 349 29 409 21 337 32 471 57 PCB101 365 35 443 21 275 13 232 43 PCB138 404 31 311 23 188 6 160 34 PCB153 359 33 309 22 227 7 165 10 PCB180 287 36 238 21 110 8 172 24 Acenaphthylene 214 26 624 20 484 7 607 31 Acenaphthene 348 26 529 20 280 8 676 31 Fluorene 403 26 680 22 391 7 675 31 Phenanthrene 451 27 734 22 462 15 604 31 Anthracene 521 26 N.D. 321 10 983 31 Fluoranthene 322 30 720 21 389 11 280 32 Pyrene 332 32 371 30 509 15 242 31 Benzo[a]anthracene 307 47 N.D. 597 4 318 31 Chrysene 101 45 640 21 641 8 226 31 Benzo[6]fluoranthene 198 36 N.D. 453 5 293 32 Benzo[&]fluoranthene 188 37 N.D. 495 8 280 32 Benzo[a]pyrene 439 36 N.D. 388 7 478 32 Indeno[l,2,3-cd]pyrene 304 26 N.D. 294 5 261 34 Dibenzo[a,/i]anthracene 153 26 N.D. 9 158 39 Benzo[g,/i,;']perylene 260 26 N.D. 239 268 33 N.D. — not determined. a Data from Vrana et al. (2001b). 302 B. Vrana et al. I Environmental Pollution 144 (2006) 296—307 1.1 ■ 1.0 ■ 0.9 - 0.8 - 0.7 - 0.6 - 0.5 • ■ experiment 1 o experiment 2 a experiment 3 1 40 -1- 80 Time (h) I 120 —I— 160 Fig. 3. Release of Hi0-biphenyl from MESCOs (expressed as the mass fraction M/M0 of the initial amount M0 remaining in the sampler) exposed at different linear flow velocities: 8cmmin~' (experiment 2), 35cmmin~' (experiment 1), and 68 cmmin-1 (experiment 3). The flow-through exposures were conducted at 19 °C. By dividing Rs values by Rs-prc values, the bias or error encompassing the difference between predicted and measured values of Rs can be estimated. Application of this approach to our sampling rate data gave the following Rs/Rsprc ratios: flu-orene, ranging from 1.5 to 2.7; phenanthrene, ranging from 1.1 to 1.9. The predicted values of Rs-prc were lower than measured values. Overall, the mean Rs/Rsprc bias ratio was 2.0, and the coefficient of variation was 34%. This is a good agreement between sampling rate values calculated from uptake and elimination kinetic data, when taking into account three separate sources of error accumulated in the Rs/Rsprc ratio, originating in the measurement of the uptake (Rs) and elimination (ke) kinetics, as well as the distribution coefficient KSw Huckins et al. (2002) have demonstrated a similar accuracy and variance (i.e. 2-fold difference and 35% variance) in comparison of measured and PRC-predicted sampling rates for lipid-filled semipermeable membrane devices. The aforementioned experiments prove the isotropy of the uptake (absorption) and the elimination (desorption) of two analytes (fluorene and phenanthrene) onto and from a MESCO sampler. It is likely that isotropic exchange kinetics is valid also for a broad range of compounds, including the rest of analytes under investigation in this study. Using a similar approach, Chen and Pawliszyn have demonstrated the isotropy of the exchange kinetics between PDMS and water for BTEX aromatic compounds (Chen and Pawliszyn, 2004). The test is practicable only for compounds with moderate/ low affinity to the receiving PDMS, for which elimination kinetics can be measured in a reasonable time period (two weeks or so). The experiment demonstrates the isotropic exchange kinetics. By knowing the behavior of either the absorption or desorption, the opposite one will also be understood. 4.1.5. Time limit for integrative sampling Both uptake and elimination of a particular compound are characterized by the same exchange rate constant ke, according to the Eq. (1). This fact can be used to determine the maximum exposure time for integrative sampling with MESCO. The chemical uptake into passive sampler remains linear and integrative approximately until concentration factor reaches half saturation: (7) where t50 is the time required to accumulate 50% of the equilibrium concentration. Under these conditions, linear model (Eq. (4)) can be used to calculate TWA concentration of the analyte in water. The maximum exposure time t50 can be estimated, if both partition coefficient KSw and the sampling rate Rs are known. t5Q ~ In 2 - R* (8) However, the KSw values are not always available and the sampling rate in the field may differ from the value determined under laboratory conditions. Because the isotropic exchange kinetics applies, the first order halftime t50 for uptake and is mathematically identical to ti/2 for elimination, i.e. time required to lose 50% of the initial residue concentration in an exposure scenario, when the analyte is initially applied to the receiving phase (M0 0) and not present in water (Cw = 0). Thus, t50 of an analyte can be approximated by the elimination halftime f1/2 of a PRC with similar physico-chemical properties. f1/2 can be calculated using Eq. (6) and Ms(t1/2)=M0f2: t50 ~ h/2 — ' In 2 (9) The results of the first order halftime calculation for PRCs used in this study are reported in Table 4. It is calculated that, for a compound with physicochemical properties similar to D10-BIP or D10-FLU, MESCO would sample integratively for more than 10 days under conditions similar to flow-through exposures in this study. According to Eq. (8), t50 increases with increasing sampler capacity (ATSW) and with decreasing sampling rate (Rs). It has been shown that the range of sampling rates is relatively narrow over a broad hydrophobicity range. Thus, it is expected that the main factor determining the t5o is the magnitude of the partition coefficient Ksw- For practical purpose, the apparent distribution constants Kf (PDMS), obtained with glass fibres coated with 100 um-PDMS for analyte's partitioning between PDMS coating and aqueous sample can be used as substitute for Ksw (Doong and Chang, 2000; Valor et al., 2001; Paschke and Popp, 2003). With a few exceptions of y-HCH, acenaphthylene and acenaphthene, KSw values are higher than for that of fluorene. This implicates that, for most of the analytes under investigation and exposure conditions similar to the test exposures described in this study, B. Vrana et al. I Environmental Pollution 144 (2006) 296—307 303 Table 4 Summary of exchange coefficients derived from flow-through exposures Flow velocity (cmmin-1) 8 35 68 ks x 103 (h~') CV (%) tm (d) ke x 103 (h-1) CV (%) hn (d) ks x 103 (h-1) CV (%) hn (d) Compound Dio-biphenyl 2.7 11 11 3.0 13 10 2.6 8 11 Di0-fluorene 2.2 14 13 2.8 15 10 2.0 20 14 D i o-phenanthrene 1.8 33 16 1.8 23 16 Di0-anthracene 1.8 23 16 integrative uptake period is expected to be longer than f50 values indicated by D10-BIP and D10-FLU. 4.1.6. Robustness of sampler performance The comparability of experimentally derived MESCO calibration data to actual values during field sampling generally depends on the similarity of laboratory and field exposure conditions and the robustness of the sampler performance against fluctuations in environmental conditions. Besides temperature and biofouling, flow velocity/turbulence may affect the uptake kinetics. The uptake kinetics is sensitive to changes in flow velocity/turbulence when the dominant barrier to mass transfer of analytes is in the laminary aqueous boundary layer at the surface of the sampler. Such effect has been observed for passive sampling devices fitted with non-porous membranes made of low-density polyethylene (Booij et al., 1998; Vrana and Schuurmann, 2002), but also for samplers fitted with macro-porous polyethersulphone membranes (Kingston et al., 2000; Alvarez et al., 2004). The effect of flow velocity on the mass transfer of analytes to the MESCO samplers in the calibration experiments can be examined in three ways: (a) by examining the potential rate-limiting barriers to mass transfer of an analyte in MESCO; (b) by testing whether the varying flow conditions significantly affected the uptake of target analytes or (c) the elimination of PRCs. 4.1.7. Examination of the mass transfer in the sampler Accumulation of target analytes in MESCO requires their movement out of the bulk sample medium, across multiple layers of barriers, and into the sampler matrix. It is assumed that the overall resistance (l/km), to the uptake of a chemical in steady state is given by sum of particular barrier resistances kov ^—-'K^Di DytKypN Z)w DsKSy/ where <5, is the particular barrier thickness, £>, is the diffusion coefficient in the barrier and Kivi is the partition coefficient between the ;th phase and water (designed as subscripts for the water [W], dialytic membrane [M] and the receiving organic phase [S]). The overall mass transfer coefficient is affected mainly by the diffusion of solutes in individual phases (water, membrane pores and the PDMS, respectively) and by their partitioning into the PDMS, since no accumulation of hydrophobic analytes is expected in the hydrophilic dialytic membrane (i.e. KMW ~ 1). As can be seen from Eq. (10), a resistance decrease in receiving phase (PDMS) is expected with increasing Ksw value for substances having similar diffusion coefficient in this material Ds. When the diffusive transport is limited by the resistance in the PDMS and then the resistance in water and dialytic membrane being negligible (i.e. if <5m/£>m^mw + <5W/£»W << 8S/DSKSW), the exchange rate constant ke should be independent of ATSW and the sampling rate Rs should increase with increasing Ksw. On the other hand, if the transport is limited by the resistance in the water or dialytic membrane (i.e. if 8M/DMKMW + <5W/ £*w > > <5s/£>s^swX the exchange rate constant ke should be inversely proportional to the equilibrium partition coefficient Ksw and sampling rate should be independent of Ksw (Eqs. (2) and (10)). Inspection of exchange rate constants determined from the PRC release kinetics shows a decrease of ke with increasing log Kf in all experiments (Table 4). Further, there is no increasing trend of the sampling rate Rs versus log Kf. for neither of the contaminant classes under investigations (Figs. 4 and 5). On the contrary, the sampling rates of PCBs decrease with increasing log Kf, likely due to decreasing diffusivity in the rate-limiting barrier with the increasing molecular size/volume. The examination of potential rate-limiting barriers to analyte uptake by MESCO indicates, that the slowest kinetic step in the mass transfer is the diffusion in one of the aqueous barriers (i.e. in the pores in the cellulose membrane, in the water filling the sampler, or in the aqueous boundary layer, respectively) rather than the diffusion in the receiving phase (PDMS). Since the aqueous boundary layer presents only a small part of the total diffusion path, it is likely that the dominant rate-limiting barrier to mass transfer is in the cellulose membrane. Moreover, the net flux of non-polar molecules across the cellulose membrane is limited by the small area of the membrane pores and by small permeability (i.e. diffusivity x solubility) of the cellulose material for non-polar compounds. Further experiments were conducted in order to confirm the robustness of the MESCO calibration data against the fluctuation in hydrodynamic conditions. 4.1.8. Effect of flow hydrodynamics on the analyte uptake A one-way ANOVA test was also performed to check whether there was any significant difference between the Rs of individual compounds obtained in experiments conducted under conditions of varying water flow velocity. The ANOVA test was performed on sampling rates that included also previously published data. The extra data included in the test were 304 B. Vrana et dl. I Environmental Pollution 144 (2006) 296—307 1400 - 1200 - 1000 - 800 - 600 - 400 - 200 6.0 logK, Fig. 4. Mean values of sampling rates Rs of PAHs from all calibration exposures conducted at 19 °C as dependent on PDMS/water distribution coefficient Kf. obtained using the same experimental conditions as in this study; the experimental flow rate was 35cmmin_1 (Vrana et al., 2001b). Thus, these data represent a repeated experiment 1. With exception of PCB28, PCB 52, benzo[a]pyrene, indeno[l,2,3-c 5). This was achieved by adding a small volume of «-octanol, a solvent with high permeability (solubility x diffusivity) for target analytes, to the interstitial space between the receiving sorbent phase and the polyethylene diffusion-limiting membrane. The aim of this study was to characterise the effect of temperature and hydrodynamics on kinetic and thermodynamic parameters characterising the exchange of analytes between the sampler and water in order to calibrate the passive sampler for the measurement of TWA concentrations of non-polar organic pollutants. 2. Theory A number of authors have presented models describing the uptake kinetics of organic contaminants in water by passive sampling devices constructed from a receiving phase and a diffusion-limiting membrane (Johnson, 1991; Huckins et al., 1993; Gale, 1998). A comprehensive overview of theory and modeling of organic contaminant exchange between SPMDs and water has also recently been published by Huckins et al. (in press). The principles of analyte uptake described for SPMDs are also applicable to the sampler described in this study. The mass transfer of an analyte from water to the sampler includes diffusion, interfacial transport steps across several barriers (compartments), including the stagnant aqueous boundary layer, possible biofilm layer, the diffusion-limiting membrane, and finally the receiving phase, which is in this case an w-octanol-saturated Clg Empore® disk. Assuming a rapid establishment of steady-state conditions, the flux of an analyte is constant and equal in each of the individual compartments. This also assumed that sorption equilibrium exists at all compartment interfaces. The resistances of each barrier to the mass transfer of analytes are then additive and independent (Scheuplein, 1968; Flynn and Yalkowsky, 1972). Applying the assumptions given above, it can be shown that the amount of the chemical accumulated from water in the receiving phase of the sampler with constant analyte concentration can be described by the following equation: mD(7) =mD(0) + (CwATdwVd -»«d(0)) x i-expf-^JU (1) where mD is the mass of analyte in the receiving phase, mD(0) is the analyte mass in the receiving phase at the start of exposure, Cw represents the water concentration during the deployment period, KDW is the receiving phase—water distribution coefficient, VD is the volume of the receiving phase, k0 is the overall mass transfer coefficient, A is the membrane surface area, and t equals time. The overall mass transfer coefficient k0 is affected by the diffusion of analytes in the individual layers (i.e. aqueous boundary layer, diffusion-limiting membrane and the receiving phase) and by their partitioning into the LDPE membrane and receiving phase; since accumulation of hydrophobic analytes is expected also in the membrane material (Huckins et al., 1999). From theory (Scheuplein, 1968; Flynn and Yalkowsky, 1972), the overall mass transfer resistance to the uptake of a chemical is given by the sum of particular barrier resistances to mass transfer. Optimisation of the sampler design has been performed previously with the aim to minimise the internal resistance of the sampler to mass transfer of hydrophobic analytes (Vrana et al., 2005b). Thus, the contribution of the receiving phase to the overall resistance should be negligible. The coefficient in the exponential function is referred to as the overall exchange rate constant ke. B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 335 In the initial uptake phase, when the exponential term is very small (^Cl), chemical uptake is linear or integrative. Thus, in the linear region Eq. (1) can be reduced: mB(t) = mD(0)+CwU( (3) For practical applications, Eq. (3) can be rewritten: mv{t) = wD(0) + CwRst (4) where Rs is the sampling rate of the system, representing the equivalent extracted water volume per unit of time. Rs = k0A = ^DWVD (5) Adding chemical standards called performance reference compounds (PRCs) to the receiving phase prior to exposure of the passive sampler has been suggested as a means to calibrate the exchange rates in situ (Booij et al., 1998; Huckins et al., 2002). The use of PRCs can be based on the evidence, that analyte uptake and offload kinetics are governed by the same mass transfer law, and obey first order isotropic exchange kinetics. When PRCs are used that are not present in water (Cw = 0) and isotropic exchange kinetics applies, Eq. (1) reduces to: mD(t) = mD(0) &xp(-ket) (6) which is a one-parameter equation, since the amount of PRC added to the sampler (mD(0)) is always known. and the diffusion-limiting membrane. The PTFE body parts (components 1 and 4, Fig. 1) supported both the receiving phase (component 2, Fig. 1) and the diffusion-limiting membrane (component 3, Fig. 1) and sealed them in place. The sampler was sealed by means of a screw cap (component 5, Fig. 1) for storage prior to use. The original design used by Kingston et al. (2000) contained a protective mesh that prevented mechanical damage to the surface of the membrane. Preliminary field studies showed some disadvantages (adsorption of analytes, fouling); therefore, the mesh was not used in this calibration study. 3.4. Preparation of the sampler Cig Empore® disks were conditioned by soaking in methanol for 20 min until translucent and then stored in methanol until required. The Empore disks were prepared in a 47-mm diameter disk vacuum manifold platform (Varian Inc.). Perdeuterated polycyclic aromatic hydrocarbons were utilised as PRCs. For loading the Empore® disks with PRCs, 10 mL methanol was slowly passed through the disk, followed by 20 mL ultrapure distilled water. Aqueous solution (500 mL) of PRCs, containing 5 |ig L~1 of each of the following chemicals: Z)i0-biphenyl, Z)i0-acenaphthene, Z)i0-phenanthrene, Z)i0-pyrene and Z)i2-benzo[a]anthracene was filtered through the disk. A vacuum was applied for 30 min to ensure that the disc was completely dry. The extraction efficiency of the loading procedure for individual PRCs was between 50 and 100%, with the maximum coefficient of variation of 9%. The Empore® disk was then put on the sampler PTFE support disk (component 4, Fig. 1). One millilitre solution of n-octanol in acetone (45% v/v) was applied. The acetone was allowed to evaporate from the disk for 10 min in the fume cupboard. The resulting volume of n-octanol was 450 uL. The LDPE membrane (pre-cleaned by soaking for 24 h in n-hexane and dried) was put on the top of the Empore disk. Any air bubbles were smoothed away from between the two layers by gently pressing the top surface of the membrane using a clean paper tissue. The PTFE supporting disk was placed in the sampler body and fixed in place to form a watertight seal between the membrane and the top section of the sampler. 3. Materials and methods 3.5. Volume of the receiving phase and the membrane 3.1. Physicochemical properties of substances Values of physicochemical properties, including octanol/water partition coefficients (log Kow), aqueous solubilities (S) and aqueous diffusion coefficients (Z)w) are summarised in Table IS in the supplementary information (Mackay and Shiu, 1992; Mackay et al., 1992). Values of aqueous Z)w were estimated using Hayduk and Laude equation (Lyman et al., 1982). 3.2. Materials and chemicals Ci8 Empore® disks (47 mm diameter) were purchased from Varian Inc., Walton-on-Thames, UK. LDPE membrane material (40 urn thick) was obtained from Fisher Scientific, Loughborough, UK. The solvents (HPLC grade quality or equivalent), acetone, ethyl acetate, methanol, n-hexane, n-octanol, n-nonane, 2,2,4-trimethyl pentane, and water were obtained from Fisher Scientific. Certified pure (purity >98% in all cases) reference standards of the test compounds, surrogates, and internal standards were obtained from Qmx Laboratories, Saffron Walden, UK. Certified external calibration solutions of target analyte mixtures at a concentration of lOugmL-1 in cyclohexane were obtained from Qmx Laboratories. 3.3. Sampler design To calculate the distribution coefficients of compounds among the sampler compartments it is necessary to know the volumes of media of the receiving phase and membrane, i.e. the combined volume of Ci8 material and the 70 mm 50 mm 0 r i .-■2 " 4 The patented design of the passive sampler has been described previously (Kingston et al., 2000; Vrana et al., 2005b). Briefly, the sampling device consists of a PTFE body containing a Ci8 Empore® disk as a receiving phase. A 40-um thick LDPE disk (47 mm diameter) of diffusion-limiting membrane is placed on the top of the receiving phase. A small volume (450 uL) of n-octanol, a solvent with high permeability (solubility x diffusivity) for target analytes, is added to the interstitial space between the receiving sorbent phase Fig. 1. Schematic diagram of the Chemcatcher passive sampling device. 336 B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 n-octanol, and the LDPE membrane material. The receiving phase was not homogenous but consisted of a solid sorbent and a liquid (n-octanol) in a porous PTFE matrix. According to the manufacturer's documentation accompanying the Empore® disks, they consist of 10% (w/w) of PTFE fibres with 90% (w/w) of silica particles, chemically bonded octadecyl (Ci8) groups. The organic carbon content of this silica—C[8 material is 17% (w/w) (Verhaar et af, 1995), so 1 g of the silica—Ci8 material contains 0.20 g of Ci8. Assuming the density of the bonded C[8 is equal to that of octadecane (0.78 gmL~'), 1 g of the disk contains 0.25 mL of the Ci8 material. The 47 mm disk weighs 572 mg, so the volume of C[8 in the whole disk is 144 uL (Green and Abraham, 2000). The thickness of the disk is 0.5 mm. Four hundred and fifty microlitres of n-octanol was added to the disk before sampler assembly. The resulting total combined volume of the receiving phase VD is 600 uL. The 47-mm diameter LDPE membrane disk used for construction of the sampler weighs 55 mg. The thickness (<5m) of the LDPE membrane disk is 35 urn. The density of LDPE is 0.91 g cnC3; the resulting volume of the membrane disk is 60.4 uL. 3.6. Exposure experiments In each experiment up to 14 passive samplers were exposed in a constant concentration flow-through exposure system. This system was devised to allow calibration of the sampling devices to be made under controlled conditions of temperature, water turbulence, and analyte concentration. It was operated in a temperature-controlled dark room. The system consisted of a 20 L glass tank with an overflow to waste. The water and the solution of test analytes dissolved in methanol were pumped into the exposure tank separately at known and controlled rates. Water was fed to the exposure tank using a peristaltic pump at 2Lh~', allowing a complete renewal of water in the tank every 10 h. Test chemicals were dissolved in methanol (30 ugL~') and the appropriate amounts of stock solution (100 uLmin-1) were delivered into exposure tank using a second peristaltic pump. A nominal concentration of 100 ngL~' for each analyte was maintained throughout the experiment. The resulting methanol concentration in the exposure water did not exceed 0.5% (v/v). Prior to each exposure, the apparatus was operated for a minimum of 48 h without samplers to allow for stabilization of the water concentration of analytes. To ensure uniform hydrodynamic conditions in the vicinity of all samplers, 14 samplers were placed on two horizontal turntables (seven samplers on each turntable) at two levels (Fig. 2). The turntables were vertically interconnected by a shaft, which was driven by an overhead stirrer. All parts of the turntable in contact with water were made of PTFE to prevent excessive sorption of chemicals. The carousel device was placed in the glass tank. The carousel device was rotated at a selected stirring speed using an overhead stirrer. The exposures lasted 14 days, during which duplicate samplers were removed at set time intervals and analysed (see below) to determine the concentrations of accumulated test chemicals. Every time a sampler was removed for analysis it was replaced by an empty (without a disk and membrane) sampler body. This was necessary to keep constant hydrodynamic conditions within the calibration system. No carousel device was used in experiments, where conditions were set as "no stirring". Samplers were placed at the bottom of the exposure tank. To prevent the forming of concentration gradients in the calibration tank during the exposure, water in the tank was slowly stirred using a stainless steel propeller stirrer (diameter 60 mm) at 30 rpm. Following exposure, the devices were removed and dismantled, and the receiving phase of the exposed system was extracted to determine the mass of each analyte and PRC present in the sampler. In addition, a minimum of three samplers were analysed prior to exposure to determine the initial levels of PRCs and analytes in blank samplers. Duplicate samples (500 mL each) of water from the outlet of exposure tank were also taken at each time the samplers were removed, and the concentration of test analyte in the water determined (Vrana et af, 2005b). 3.7. Experimental design The calibrations were set up to measure the uptake of target analytes at different combinations of temperatures and hydrodynamic conditions in a full factorial design. The calibration data were gathered in order to determine ■ glass tank ■ carousel device sampler J Fig. 2. Exposure tank and a carousel device used in flow-through calibration of passive sampling devices. the sampling parameters and to observe how they are affected by environmental conditions. Each factor (temperature, stirring speed) was tested at three levels, resulting in the total number of nine experiments. The experimental conditions of individual exposures are given in Table 1. 3.8. Extraction of analytes from passive samplers and from water After exposure the sampler was carefully disassembled and the compounds were extracted from the Empore® disk using a two-step extraction procedure with organic solvents, described by Vrana et al. (2005b). The test analytes in water samples taken from the outlet of flow-through exposure system were extracted using solid-phase extraction (SPE) on Bondelut C[8 LO SPE cartridges (3 mL/200 mg sorbent; Varian Inc.). The extraction procedure has been described by Vrana et al. (2005b). 3.9. Instrumental analysis The concentrations of all target analytes in water and sampler extracts were quantified using GC/MS as described by Vrana et al. (2005b). Analysis was performed with a 6890A series GC equipped with a mass-selective detector 5973 (Agilent Technologies, Bracknell, UK). 3.10. Data processing The experimental time course accumulation rates of individual test substances on the Empore® disks were fitted by linear regression analysis using B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 337 Table 1 Summary of sampler flow-through exposure experiments Experiment no. 1 2 3 4 5 6 7 8 9 Temperature (°C) 6 11 18 Exposure period (h) 0-336 0-336 0-336 0-284 0-264 0-336 0-336 0-360 0-360 Rotation speed (min~ ) 0 40 70 0 40 70 0 40 70 Linear sampler velocity (cm s~')a 0 40 70 0 40 70 0 40 70 No. of samplers analysed 16 16 16 15 14 12 17 18 18 a Linear velocity vs was calculated as 2tzrf, where r is the radius between the centre of the calibration carousel and the centre of the sampler and/is the rotation speed. Eq. (4). The adjustable parameters were the intercept (mD(0)) and the slope (Cw x Rs) of the uptake curve mD =flt). Quality of the fit was characterised by the standard deviations of the optimised parameters, as well as the correlation coefficient adjusted for the degrees of freedom (r2 adjusted), the fit standard deviation, and the Fisher test criterion on the accuracy of the model. The sampling rates Rs for individual test compounds were calculated by dividing the slope of the linear uptake curve by the mean aqueous analyte concentration during the exposure period. The required variances of Rs values were calculated from the coefficients of variation (relative standard deviations) of the uptake slope parameters and the concentrations in the aqueous phase, which were obtained according to the law of error propagation. The release of PRCs from the sampler was fitted by non-linear regression analysis using Eq. (6) with mD(0) and ks as adjustable parameters. Quality of the fit was characterised by the standard deviations of the optimised parameters, as well as the correlation coefficient adjusted for the degrees of freedom (r2 adjusted), the fit standard deviation, and the Fisher test criterion on the accuracy of the model. 4. Results and discussion 4.1. Flow-through exposures 4.3. PRC offload kinetics The offload of PRCs from the Empore® disks was fitted by non-linear regression analysis using Eq. (6) with mD(0) and ke as adjustable parameters. Characteristic PRC offload curves are shown in Fig. 4 and the results are listed in Tables 5S— 7S in the supplementary information. Satisfactory fits of the first order decay, Eq. (6), to the offload data were obtained for £>10-biphenyl, £>10-acenapthene, £>10-fluorene and £>10-phenanthrene. The release of £>10-pyrene and £>12-benzo[a] anthracene from the sampler was too slow to be able to evaluate the kinetics statistically. For these PRCs, the results of the first order decay fits were poor and estimates of ke values for £>10-pyrene and £>12-benzo [a] anthracene were statistically not significantly different from 0 (P > 0.05). 4.4. Verification of isotropic exchange kinetics: absorption versus desorption The performance of the sampler was tested by exposure to constant concentrations of test chemicals in a continuous flow-exposure tank. Concentrations of the analytes in water (Cw) and the amounts accumulated in the receiving disk (mD) were two parameters measured regularly during the continuous flow-exposures. During exposure the water concentration was held constant, and this was confirmed by analyses of water samples. Characteristic analyte uptake curves for the sampler are shown in Fig. 3. Satisfactory linear regression fits of the Eq. (4) to the uptake data of analytes from water to the sampler discs were obtained for all test compounds in all experiments. 4.2. Sampling rate The sampling rates Rs obtained in flow-through exposure experiments conducted at 100 ngL-1 nominal water concentration and various linear flow velocities and temperatures are shown in Tables 2S—4S in the supplementary information. Over the range of controlled laboratory conditions, the magnitude of Rs values spanned over two orders of magnitude (i.e. from 0.008 for benzo[a]anthracene at 18 °C and a stirring speed of 0— 1.380 L d~1 for fluoranthene at 18 °C and a stirring speed of 40 rpm). This range of sampling rates is narrow relative to the broad Kow range of nearly five orders of magnitude. When the uptake rate of a target analyte Rs and the exchange rate constant ke of its deuterated analogue (PRC) 300 Fig. 3. Typical uptake curves of the analytes in the sampler. Data are presented from the flow-through exposure conducted at 11 °C and the carousel rotation speed 40 min~ (experiment 5). The drawn lines show the linear fits of the data using Eq. (4). 338 B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 1.2 1.0 0.6 0.4 0.2 0.0 • D10-Acenaphthene O D10-Fluorene ▼ D10-Phenanthrene V D10-Pyrene ■ D10-Biphenyl 0 50 100 150 t[h] 200 250 300 Fig. 4. Typical offload curves of PRCs from the sampler. Data are presented from the flow-through exposure conducted at 11 °C and the carousel rotation speed 40 min~ (experiment 5). The drawn lines show the best fits of the data using Eq. (6). are measured under the same conditions, the correlation between uptake and offload kinetic parameters can be viewed as a preliminary check of the isotropic exchange kinetics. Fig. 5 demonstrates that, for a broad range of environmental conditions (temperatures and water flow rates), there is a very good correlation between uptake and offload kinetic parameters of analytes and their deuterated analogues. A good correlation has been found not only for uptake of analytes and offload of their labelled analogue PRCs, but for a broad variety of analyte/PRC combinations (Table 8S, supplementary information). This indicates that the mass transfer 1.0 0.6 0.4 0.2 0.0 • Acenaphthene O Fluorene T Phenanthrene 0.00 0.10 0.12 Fig. 5. Correlation between sampling rates Rs of three polycyclic aromatic hydrocarbons and offload rate constants ke of their perdeuterated analogues (PRCs). The data represent nine flow-through exposures performed at various combinations of temperature and water turbulence. of many analytes and PRCs is governed by the same law and the isotropy of the uptake (absorption) onto and the offload (desorption) from the sampler. The test is practicable only for compounds with moderate/low affinity for the receiving phase, and for which significant offload can be measured within the time period of the experiment. A fall demonstration of the isotropic exchange kinetics would require a direct comparison of the exchange rate constants ke of a particular compound obtained from both offload and uptake curves. During the 2 weeks of sampler exposure, the uptake curves of the analytes under investigation remained in the linear uptake phase. Thus, the calculation of ke from the fit of an exponential function to the uptake data was precluded. A prolonged sampler exposure would enable to measure the whole uptake curve. However, such experiments were not performed in this study because of practical difficulties such as a progressive deterioration of water quality due to increasing microbial activity in the exposure tank during exposures longer than several weeks. Moreover, 14 days is the typical time scale for deployment of the devices in the field. 4.5. Receiving phase—water distribution coefficients The conventional approach to measuring the distribution coefficient between the receiving phase of the sampler and water is to perform a static exposure of the sampler in water and to measure concentration of the target analyte in water and in the receiving phase after equilibration. This approach is complicated for hydrophobic compounds, where difficulties might occur with the measurement of very low equilibrium concentrations in the water phase. Moreover, a time series of measurement needs to be performed to assure that the partitioning equilibrium has been reached. In this work, a kinetic approach to the measurement of the distribution coefficients was adopted. In the flow-through exposures, kinetic parameters for several compounds and their perdeuterated analogues (PRCs) were determined at a broad range of exposure conditions. These parameters included the sampling rates Rs for absorption and the desorption rate constants ke. Assuming the isotropy of the exchange kinetics of chemicals under investigation, and the validity of the model used to describe the kinetics, the value of the apparent receiving phase—water distribution coefficient can be calculated as a ratio of the absorption and desorption transport parameters for a particular compound: Rs keVD (7) There are only minimum differences in physicochemical properties of a compound and its deuterated analogue (PRC). Thus, it was assumed that the actual differences in their kinetic parameters were smaller than the experimental error associated with their determination. There were four compounds, for which the absorption and the desorption rate parameters of the corresponding PRC were measured in each experiment. These were acenaphthene, fluorene, phenanthrene and pyrene. B. Vrana et at. I Environmental Pollution 142 (2006) 333—343 339 The volume of the receiving phase VD is estimated to be lnKv 600 uL. The ATDW value was calculated using Eq. (7) and the required variance was calculated from the coefficients of variation of the uptake and elimination rate parameters. These were obtained according to the law of error propagation. Up to nine values of ATDW for each compound were calculated from the data available from individual exposure experiments (Table 9S, supplementary information). Among the exposure conditions that were varied in the experiments, only temperature is expected to affect the magnitude of ATDW. Thus, up to three independent measurements of ATDW were obtained for each of the three exposure temperatures. The temperature effect on ATDW is shown in Fig. 6. Parameters of the temperature dependence were estimated using the Van't log ATD Hoff plot for the temperature range from 6 to 18 °C in the form: A/T-B (8) where A and B are parameters of the linear dependence characterising the enthalpy and entropy components of the free energy, respectively, and T is the absolute temperature (K). The elevated variance of some of the calculated ATDW values precludes the closer investigation of the temperature effect on the distribution coefficients. Nevertheless, the experimental evidence indicates that ATDW values are not significantly affected by temperature in the range from 6 to 18 °C. This enables all log ATDW data to be described by a linear empirical function of log Kow (Fig. 7): 1.382log/Tow - 1-77 (R = 0, s = 0.13, n = 31) (9) 20000 ■ 18000 ■ 16000 ■ 14000 ■ 12000 ■ 10000 ■ 8000 ■ 6000 ■ 4000 2000 ■ 0 ■ Acenaphthene n-1-1-1-1-1- 8 10 12 14 16 18 20 25000 20000 - 15000 - 10000 - 5000 - Fluorene £ J -1-1-1-1-1-1- 6 8 10 12 14 16 18 20 T[°C] T[°C] 80000 ■ 70000 ■ 60000 ■ 50000 ■ 40000 ■ 30000 ■ 20000 ■ 10000 ■ 0 ■ -r- Phenanthrene 6e+5 n-1-1-1-1-1- 8 10 12 14 16 18 20 T [°C] T [°C] Fig. 6. Temperature dependence of apparent distribution coefficients between the sampler receiving phase (n-octanol-saturated Ci8-Empore® disk) and water iC"Dw. 340 B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 Fig. 7. The apparent receiving phase—water distribution coefficient log /("dw as a function of log Kow. Huckins et al. (in press) have shown that for SPMDs, the log Kow versus log SPMD/water partition coefficient plot for compounds with logATOw>5.0 deviated from linearity. This phenomenon has also been observed for plots of log biocon-centration factor versus log Kow (Connell, 1990). It was not possible to show in our study whether a deviation from linearity occurs for very hydrophobic compounds. 4.6. Time limit for integrative sampling The chemical uptake into the passive sampler remains linear and integrative approximately until concentration factor reaches half saturation: mD(t50)/VD/Cw = mD(°°)/2 = A:DW/2 (10) where t50 is the time required to accumulate 50% of the equilibrium concentration. Under these conditions, a linear model (Eq. (4)) can be used to calculate the TWA concentration of the analyte in water. The maximum exposure time t50 can be estimated, if both partition coefficient KDW and the sampling rate Rs are known: '50 = ln2JřDWyD//?s (11) According to Eq. (11), t50 increases with increasing KDW and with decreasing Rs. It has been shown that the range of sampling rates is relatively narrow over a broad hydrophobic-ity range. Thus, the main factor determining the t50 is the magnitude of the apparent distribution coefficient ATDW. However, the t50 estimate using this approach is not very precise because the sampling rates in the field differ from those determined under laboratory conditions. If the isotropic exchange kinetics apply, the first order half-time t50 for uptake is mathematically identical to ?1/2 for offload, i.e. the time required to lose 50% of the initial residue concentration in an exposure scenario, when the analyte is initially applied to the receiving phase (mD(0) 0) and is not present in the water (Cw = 0). Thus, t50 of an analyte can be approximated by the offload halftime ?1/2 of a PRC with similar physicochemical properties. f1/2 can be calculated using Eq. (12) and mD(t 1/2) = mD(0)/2: '1/2 : ln2/ke (12) In general, shorter halftimes are predicted at elevated temperatures and under turbulent hydrodynamic conditions, when the exchange kinetics is faster. It is calculated that, for compounds with hydrophobicity similar to £>10-biphenyl or £>10-fiuorene (log ATow ~ 4), the sampler would sample integratively during a time period between 1 and 10 weeks, depending on the temperature and turbulence level. For more hydrophobic compounds, this time period can be much longer. For example, the halftime of more than three months is calculated for compounds with log Kow > 5. 4.7. Sampling rates: effect of analyte properties The sampling rate is strongly affected by the physicochemical properties of the compounds. Among the non-polar priority pollutants under investigation, the highest sampling rates were observed for small, moderately hydrophobic compounds: anthracene, phenanthrene, fluoranthene and pyrene. The maximum sampling rates were measured for compounds with log Kow of 4.5. The lowest sampling rates were measured for indeno[l,2,3-Cii]pyrene, dibenzo[a,/z]anthracene and ben-zo[g,/z,;]perylene; large and extremely hydrophobic compounds. The typical dependence of sampling rates on hydrophobicity is illustrated in Fig. 8. Fig. 8. Effect of temperature and log Kow on analyte sampling rate values at 70 rpm rotation speed to illustrate the response surface. B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 341 4.8. Effect of temperature The relationship between sampling rates of the test analytes and temperature can be compared at three temperatures (6, 11 and 18 °C). In general, the sampling rate increases with the increasing exposure temperature. The typical dependence of sampling rate on temperature is shown in Fig. 9. We demonstrated that for the four polycyclic aromatic hydrocarbons with logATow range from 4.0 to 5.1, the apparent receiving phase-water distribution coefficient KDW was not significantly affected by temperature within the range from 6 to 18 °C. Thus, the temperature is expected to affect mainly the magnitude of the kinetic component of the sampling rate (ke; Eq. (7)). Typically, increased temperature should enhance mass transfer in all media. The temperature dependence of the sampling rate Rs can then be described by the Arrhenius-type equation: A£ ln/?s=lnA--- (13) RT y ' The activation energies range between 20 and 208 kJ moP . The average of all A£a values was 93 kJ moP1 with a standard deviation of 56 kJ mol-1. This would correspond to an increase in sampling/offload rate by a factor 5.2 over the temperature range from 6 to 18 °C. For a comparison, Huckins et al. (in press) calculated from the literature data available for SPMDs an average activation energy of 37 kJ mol-1. Thus, the effect of temperature on the Chemcatcher uptake kinetics appears to be more significant than that on SPMD sampling rates. The activation energies calculated for uptake of acenaph-thene, fluorene and phenanthrene were in line with the activation energies calculated for offload of £>10-acenaphthene, £>io-fluorene and £>io-phenanthrene. This is in agreement with isotropic exchange kinetics as well as with the assumptions that the temperature affects mainly the magnitude of the kinetic component of the sampling rate (ke). Note that the calculation of A£a was not performed for £>10-pyrene because of very low magnitude and a poor precision of the ke values. 4.9. Effect of hydrodynamics where R is the universal gas constant (kJ moP1 K-1), A is the pre-exponential factor expressing the maximum sampling rate at infinite temperature, T is the absolute temperature (K) and A£a is the activation energy (kJmol-1). Values of A£a were obtained by plotting the natural logarithm of Rs against the reciprocal value of absolute temperature (l/T). The intercept gives the value of In A. The activation energy A£a can be calculated by multiplying the slope of the regression line (&EJR) by R. An analogous equation was used for description of the temperature dependence of the offload rate constant ke. The calculation of the activation energy A£a using Eq. (13) was performed on three sets of calibration data, obtained at three levels of water turbulence. Because of a very low magnitude of sampling rates in stagnant water, evident temperature dependence was observed only for data obtained under conditions of turbulent water flow (40 and 70 min-1). The sampling rates obtained for individual compounds under various flow conditions were compared. With exception of the moderately hydrophobic lindane (log Kow = 3.7), a significant increase of sampling rate with increasing flow velocity was observed for all compounds under investigation. This corresponds well with the theory of diffusion through two films in series (Scheuplein, 1968; Flynn and Yalkowsky, 1972), which predicts a switch in the overall mass transfer to the aqueous phase control for hydrophobic compounds. A similar effect of hydrodynamics has been observed and explained also for SPMDs (Vrana and Schtitirmann, 2002). 4.10. Method sensitivity Minimum quantifiable TWA water concentrations were estimated by substituting the limits of quantification in the sampler extracts mD(LOQ) into Eq. (6). The calculated method Fig. 9. Effect of hydrodynamics on the analyte sampling rate values. Data are presented from the flow-through exposure conducted at 11 °C and various carousel rotation speeds (0, 40 and 70 min~'; experiments 4, 5 and 6, respectively; see Table 1). The compounds are sorted according to their increasing hydrophobicity. 342 B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 Table 2 Sensitivity of the passive sampling device Compound MLDa (ngL-1) MLQb (ngL-1) Acenaphthene 0.5-2.6 1.5-8.8 Fluorene 0.1-0.9 0.4-3.1 Phenanthrene 0.05-0.6 0.2-2.2 Anthracene 0.1-0.9 0.2-3.1 Fluoranthene 0.03-0.7 0.1-2.5 Pyrene 0.1-2.4 0.2-8.0 Benzo[a]antracene 0.4-25.2 1.3-83.3 Chrysene 0.2-7.7 0.7-25.2 Benzo[6]fluoranthene 0.6-20.6 2.1-68.2 Benzo[&] fluoranthene 1.8-21.1 6.1-69.9 Benzo[a]pyrene 1.3-18.1 4.3-59.7 Indeno[l,2,3-ai]pyrene 10.1 33.4 Dibenzo [a, h] an tracené 2.0-8.5 6.7-27.8 Benzo[g,/i,;']perylene 5.4-14.1 17.9-46.9 Pentachlorobenzene 0.1-0.9 0.4-2.9 Hexachlorobenzene 0.05-1.6 0.2-5.3 Lindane 2.2-12.0 7.3-40.1 Endosulfan I 0.6-9.1 2.0-30.5 Dieldrin 0.2-5.3 0.8-17.7 a MLD — method limit of detection, expressing the minimum TWA water concentration detectable by the sampler; the range of MLD was calculated for a typical 14 days sampler exposure and typical limits of detection for a GC/MS method using a splitless injection of 1 uL of sampler extract (0.5—6 ng/sampler). b MLQ — method limit of quantification, expressing the minimum time-weighted average (TWA) water concentration quantifiable by the sampler; the range of MLQ was calculated for a typical 14 days sampler exposure and typical limits of quantification for a GC/MS method using a splitless injection of 1 uL of sampler extract (1.7—20 ng/sampler). limits of quantification depend on the sampling rate Rs, and the method sensitivity increases with increasing sampler exposure period. Moreover, improved sensitivity can be achieved at elevated temperatures and turbulent hydrodynamic conditions. The calculated range of quantification limits for a typical 14-day sampler deployment is shown in Table 2. 5. Conclusions The study provided a calibration database necessary for reliable integrative sampling of hydrophobic micropollutants, including polyaromatic hydrocarbons and organochlorine pesticides, in water. It characterised the effect of two main environmental variables, temperature and water turbulence, on the sampler performance. The implication of the experiment demonstrating the apparent isotropic exchange kinetics is that, by knowing the behaviour of either the absorption or desorption kinetics, the opposite one will also be understood. This finding can be used practically for in situ recalibration of the sampler, where it is difficult to measure the level of environmental variables (especially turbulence and biofouling), but it is possible to determine the offload kinetics of PRCs. Sampling rates can be calculated from the known offload rate constants ke of PRCs and their correlations with the sampling rates Rs. This study contributes to the growing pool of evidence indicating that the PRC concept is widely applicable for the determination of in situ sampling kinetics, required for more accurate measurement of TWA concentrations using integrative passive samplers. The successful application of the PRC approach with other designs of water samplers including SPMDs (Booij et al., 1998; Huckins et al., 2002), silicone strips (Booij et al., 2002) and with membrane-enclosed sorp-tive coating samplers that use polydimethylsiloxane as a receiving phase (Vrana et al., 2001; Vrana et al., unpublished data) has been demonstrated. In addition this concept has been recently applied to passive air samplers, e.g. tristearin-based samplers (Mtiller et al., 2000), SPMDs (Soderstrom and Bergqvist, 2004) and polyurethane foam samplers (Bartkow et al., 2004). Recently, Chen and Pawliszyn (2004) demonstrated the applicability of PRCs for rapid field sampling/sample preparation using solid-phase microextraction (SPME). Nevertheless, more research is required to incorporate the PRC concept into sampler configurations with very strong analyte retention in the receiving phase, such as the polar organic chemical integrative sampler (POCIS; Alvarez et al., 2004), polar design of the Chemcatcher (Kingston et al., 2000) or samplers characterised by anisotropic analyte exchange kinetics (Persson et al., 2001). Our future work will focus on demonstrating the practical application of the laboratory calibration data, obtained in this study, for the measurement of TWA water concentration of priority pollutants in the field. Empirical and mechanistic models relating the calibration data to physicochemical properties of the sampled compounds will enable to apply the calibration data for measurement of a broader range of pollutants. More research is necessary to provide on understanding the effect of biofouling on the sampler performance. Acknowledgment We acknowledge the financial support of the European Commission (Contract EVK1-CT-2002-00119; http://www. port.ac.uk/research/stamps/) for this work. Appendix A. Supplementary information The supplementary information contains tables of selected physicochemical properties of test analytes; analyte sampling rates and PRC offload rate constants; apparent distribution coefficients between the sampler receiving phase and water and correlation coefficients between the sampling rates of analytes and offload rate constants of PRCs. Supplementary information for this manuscript can be downloaded at doi:10.1016/ j.envpol.2005.10.033. References Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L., Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environmental Toxicology and Chemistry 23, 1640—1648. B. Vrana et al. I Environmental Pollution 142 (2006) 333—343 343 Bartkow, M., Jones, K., Kennedy, K., Holling, N., Hawker, D., Müller, J., 2004. Evaluation of performance reference compounds in PUF passive air samplers at different wind speeds. Organohalogen Compounds 66, 139-144. Booij, K., Sleiderink, H.M., Smedes, E, 1998. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environmental Toxicology and Chemistry 17, 1236—1245. Booij, K., Smedes, E, van Weerlee, E.M., 2002. Spiking of performance reference compounds in low density polyethylene and silicone passive samplers. Chemosphere 46, 1157—1161. Chen, Y., Pawliszyn, J., 2004. Kinetics and the on-site application of standards in a solid-phase microextraction. Analytical Chemistry 76, 5807—5815. Connell, D.W., 1990. Bioaccumulation of Xenobiotic Compounds. CRC Press, Boca Raton, FL, USA. Flynn, G.L., Yalkowsky, S.H., 1972. Correlation and prediction of mass transport across membranes. I. Influence of alkyl chain length on flux-determining properties of barrier and diffusant. Journal of Pharmaceutical Sciences 61, 838-852. Gale, R.W., 1998. Three-compartment model for contaminant accumulation by semipermeable membrane devices. Environmental Science and Technology 32, 2292-2300. Green, C.E., Abraham, M.H., 2000. Investigation into the effects of temperature and stirring rate on the solid-phase extraction of diuron from water using a Cis extraction disk. Journal of Chromatography A 885, 41—49. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environmental Science and Technology 27, 2489-2496. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Cranor, W.L., Clark, R.C., Mogensen, B.B., 2002. Development of the permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environmental Science and Technology 36, 85-91. Huckins, J.N., Petty, J.D., Booij K. Monitors of Organic Contaminants in the Environment: Semipermeable Membrane Devices. Springer Verlag, in press. Huckins, J.N., Petty, J.D., Orazio, C.E., Lebo, J.A., Clark, R.C., Gibson, V.L., Gala, W.R., Echols, K.R., 1999. Determination of uptake kinetics (sampling rates) by lipid-containing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water. Environmental Science and Technology 33, 3918—3923. Johnson, G.D., 1991. Hexane-filled dialysis bags for monitoring organic contaminants in water. Environmental Science and Technology 255, 1897-1903. Kingston, J.K., Greenwood, R., Mills, G.A., Morrison, G.M., Persson, B.L., 2000. Development of novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments. Journal of Environmental Monitoring 2, 487—495. Koester, C.J., Simonich, S.L., Esser, B.K., 2003. Environmental analysis. Analytical Chemistry 75, 2813-2829. Lyman, W.J., Reehl, W.F, Rosenblatt, D.H., 1982. Handbook of Chemical Property Estimation Methods, Environmental Behavior of Organic Compounds. McGraw-Hill Book Company, New York. Mackay, D., Shiu, W.Y, 1992. In: Illustrated Handbook of Physical-Chemical Properties of Environmental Fate of Organic Chemicals, vol. 1. Lewis Publishers, MI, USA. Mackay, D., Shiu, W.Y, Ma, K.C., 1992. In: Illustrated Handbook of Physical—Chemical Properties of Environmental Fate of Organic Chemicals, vol. 2. Lewis Publishers, MI, USA. Müller, JE., Hawker, D.W., Conell, D.W., Kômp, P., McLachlan, M.S., 2000. Passive sampling of atmospheric SOCs using tristearin-coated fibreglass. Atmospheric Environment 34, 3525—3534. Namiesnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M., Wasik, A., 2005. Passive sampling and/or extraction techniques in environmental analysis: a review. Analytical and Bioanalytical Chemistry 381, 279—301. Persson, L.B., Morrison, G.M., Friemann, J.U., Kingston, J., Mills, G., Greenwood, R., 2001. Diffusional behaviour of metals in a passive sampling system for monitoring aquatic pollution. Journal of Environmental Monitoring 3, 639-645. Scheuplein, R.J., 1968. On the application of rate theory to complex multibar-rier flow co-ordinates: membrane permeability. Journal of Theoretical Biology 18, 72-89. Söderström, H.S., Bergqvist, P.A., 2004. Passive air sampling using semipermeable membrane devices at different wind-speeds in situ calibrated by performance reference compounds. Environmental Science and Technology 38, 4828-4834. Stuer-Lauridsen, F, 2005. Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environmental Pollution 136, 503-524. Verhaar, H.J.M., Busser, F.J.M., Hermens, J.L.M., 1995. Surrogate parameter for the baseline toxicity content of contaminated water — simulating the bioconcentration of mixtures of pollutants and counting molecules. Environmental Science and Technology 29, 726—734. Vrana, B., Mills, G.A., Allan, I.J., Dominiak, E., Svensson, K., Knutsson, J., Morrisson, G., Greenwood, R., 2005a. Passive sampling techniques for monitoring of pollutants in water. IrAC Trends in Analytical Chemistry 24, 845-868. Vrana, B., Mills, G., Greenwood, R., Knutsson, J., Svensson, K., Morrison, G., 2005b. Performance optimization of a passive sampler for monitoring hydrophobic organic pollutants in water. Journal of Environmental Monitoring 7, 612-620. Vrana, B., Popp, P., Paschke, A., Schüürmann, G., 2001. Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water. Analytical Chemistry 73, 5191—5200. Vrana, B., Popp, P., Paschke, A., unpublished results. Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water. Vrana, B., Schüürmann, G., 2002. Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics. Environmental Science and Technology 36, 290—296. Príloha 11 Vrana B., Mills G. A., Kotterman M., Leonards P., Booij K., and Greenwood R., Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water, Environ. Pollut., 2007,145, 895-904. Available online atwww.sciencedirect.com ^J|j^ *%' ScienceDirect ELSEVIER Environmental Pollution 145 (2007) 895-904 ENVIRONMENTAL POLLUTION www.elsevier.com/locate/envpol Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water Branislav Vrana a'*, Graham A. Mills b, Michiel Kottermanc, Pirn Leonardsc, Kees Booij d, Richard Greenwood a School of Biological Sciences, University of Portsmouth, King Henry Building, King Henry I Street, Portsmouth P01 2DY, United Kingdom b School of Pharmacy and Biomedical Sciences, University of Portsmouth, St Michael's Building, White Swan Road, Portsmouth P01 2DT, United Kingdom c Netherlands Institute for Fisheries Research, P.O. Box 68, Haringkade 1, 1970 AB IJmuiden, The Netherlands á Royal Netherlands Institute for Sea Research, P.O. Box 59, 1790 AB Texel, The Netherlands Received 16 December 2005; received in revised form 28 March 2006; accepted 12 April 2006 The exchange kinetics of hydrophobic organic pollutants between passive sampler and water were modelled to enable the measurement of time weighted average concentrations of pollutants. The applicability of the model was tested in a field study. Abstract Passive sampling of dissolved pollutants in water has been gaining acceptance for environmental monitoring. Previously, an integrative passive sampler consisting of a Cjg Empore® disk receiving phase saturated with «-octanol and fitted with low density polyethylene membrane, was developed and calibrated for the measurement of time weighted average (TWA) concentrations of hydrophobic pollutants in water. In this study, the exchange kinetics were modelled to obtain a better understanding of the mechanism of the accumulation process and to enable the measurement of TWA concentrations of hydrophobic pollutants in the field. An empirical relationship that enables the calculation of in situ sampling rates of chemicals using performance reference compounds was derived and its application was demonstrated in a field study in which TWA aqueous concentrations estimated from sampler data for target analytes were compared with TWA concentrations obtained from spot samples of water collected regularly during the sampler deployment period. © 2006 Elsevier Ltd. All rights reserved. Keywords: Chemcatcher; Hydrophobic organic pollutants; Passive sampling; Water monitoring 1. Introduction Passive sampling of organic pollutants in water has been gaining acceptance for environmental monitoring. A range of passive sampling devices has been developed for monitoring organic pollutants in water. These include the lipid-filled * Corresponding author. Present address: Water Research Institute, Nabr. arm. gen. L. Svobodu 7, 81249 Bratislava, Slovakia. Tel.: +421259343401. E-mail address: branovrana@googlemail.com (B. Vrana). 0269-7491/$ - see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2006.04.030 semi-permeable membrane device (SPMD; Huckins et al., 1993) and the membrane enclosed sorptive coating (MESCO; Vrana et al., 2001) for non-polar compounds and the polar organic chemical integrative sampler (POCIS; Alvarez et al., 2004) for polar compounds. The design and field performance of a wide range of passive samplers suitable for monitoring organic pollutants have recently been reviewed (Namiesnik et al., 2005; Stuer-Lauridsen, 2005; Vrana et al., 2005a). We previously developed a passive sampling device (Chemcatcher) for the measurement of time weighted average (TWA) concentrations of pollutants in aquatic environments 896 B. Vrana et al. I Environmental Pollution 145 (2007) 895—904 (Kingston et al., 2000; Vrana et al., 2005b). The sampler is based on the diffusion of target compounds through a membrane and the subsequent accumulation of these pollutants in a sorbent-receiving phase. Accumulation rates and selectivity are regulated by the choice of both the membrane and a receiving phase material. One of the prototypes was designed for the sampling of non-polar organic compounds with log octanol/ water partition coefficient (log Kow) values greater than three (Kingston et al., 2000). This system used a 47 mm Clg Em-pore® disk as the receiving phase and a low density polyethylene (LDPE) membrane. The Clg Empore® disk has a high affinity and capacity for the sampled pollutants. Despite the wide application of passive samplers, calibration data that relate absorbed amounts of chemicals to their aqueous concentrations are rare. As a result, field measurements using passive samplers are primarily reported in terms of absorbed amounts of chemicals, and only occasionally are the absorbed amounts translated into actual aqueous concentrations. To enable measurement of TWA water concentrations of non-polar organic pollutants, we calibrated the Chemcatcher sampler in a flow-through tank under controlled conditions. The calibration experiments were designed to characterize the effect of physico-chemical properties (compound hydrophobicity), temperature and hydrodynamics on kinetic and thermodynamic parameters characterising the exchange of analytes between the sampler and water. The calibration data have been reported recently (Vrana et al., 2006). In this study, the exchange kinetics of analytes between the sampler and water was modelled to obtain further insight into the mechanism of the accumulation process and to enable the measurement of TWA concentrations of non-polar priority pollutants in the field. An empirical relationship that enables the calculation of in situ sampling rates of non-polar chemicals using performance reference compounds (PRCs) was derived. Its application was demonstrated in a field study in which sampler data were compared with spot samples of water, collected regularly during the sampler deployment. 1.1. Theory The theory of steady-state mass transfer of an analyte from water to the Chemcatcher passive sampler has been described (Vrana et al., 2006). The amount of the chemical accumulated from water in the receiving phase of the sampler with constant analyte concentration can be described by the equation: The overall mass transfer coefficient k0 is affected by the diffusion of analytes in the individual layers; i.e. aqueous boundary layer and LDPE membrane as well as by their partitioning into the membrane and receiving phase. The contribution of the receiving phase to the overall resistance is considered to be negligible (Vrana et al., 2005b). From theory (Scheuplein, 1968; Flynn and Yalkowsky, 1972), the overall mass transfer resistance to the uptake of a chemical is given by the sum of particular barrier resistances to mass transfer. The overall resistance (l/k0) is then given by: 1 _ 1 1 _ <5W ka hfi IcmKmw Ow DmKmw (2) where kw and kM are mass transfer coefficients in the aqueous boundary layer and the membrane, respectively. Eq. (2) shows that resistance to mass transfer increases with the increasing thickness of the barrier <5 and decreases in the diffusion and partition coefficients D and K, respectively. The coefficient in the exponential function (Eq. (1)) is referred to as the overall exchange rate constant ke. koA (3) ^dwVd In the initial uptake phase, pollutant uptake is linear or integrative. For practical applications, Eq. (1) can be reduced and rewritten: mD = mD0 + CwRst (4) where Rs [m3 s *] is the sampling rate of the device, and represents the equivalent water volume sampled per unit of time. Rs=k0A = keK^Vv (5) Adding chemical standards called PRCs to the receiving phase of the passive sampler prior to exposure has been suggested as a means to calibrate the exchange rates in situ (Booij et al., 1998; Huckins et al., 2002a). The use of PRCs can be justified providing that analyte uptake and offload kinetics are governed by the same mass transfer law, and obey first order exchange kinetics. When PRCs are used that are not present in water (Cw = 0) and isotropic exchange kinetics applies, Eq. (1) reduces to: niT, = mm exp(-ket) 2. Materials and methods (6) 1 — exp (1) where mD [kg] is the mass of analyte in the receiving phase, mD0 [kg] is the analyte mass in the receiving phase at the start of exposure, Cw [kg m~3] represents the water concentration during the deployment period, KDW is the receiving phase/water distribution coefficient, VD [m3] is the volume of the receiving phase, k0 [m s_1] is the overall mass transfer coefficient, A [m2] is the membrane surface area, and t [s] equals time. 2.1. Physicochemical properties of chemicals Preferred or selected values of physicochemical properties, including octa-nol/water partition coefficients (log ifow) and aqueous solubilities (S) were taken from Mackay et al. (1992a,b) and have been summarized previously (Vrana et al., 2006). Values of Z)w were estimated using Hayduk and Laudie equation (Tucker and Nelken, 1982). The distribution coefficient between the receiving phase of the sampler and water log iC"DW can be described by a linear empirical function of logifow (Vrana et al., 2006): log if D 1.382 logifow-1.77 (Ä = 0.97, s = 0.13, n = 31) (7) B. Vrana et al. I Environmental Pollution 145 (2007) 895—904 897 Booij et al. (2003) showed that the LDPE-membrane partition coefficient log ^mw can be described as a linear function of log K0w, with a temperature-dependent intercept. The intercept was reported to be close to zero for a temperature of approximately 10 °C and we used the predictive equation in form: logif„w = 0.972 log Kow (8) 2.2. Passive sampler design The passive sampler construction and preparation have been described previously (Vrana et al., 2005b, 2006). Briefly, the sampler consists of a PTFE body containing aCu Empore disk (47 mm diameter) as a receiving phase. A 40 urn thick LDPE membrane (47 mm diameter) is placed on the top of the receiving phase. 450 uL of n-octanol is added to the interstitial space between the receiving sorbent phase and the membrane. The PTFE body supports both the receiving phase and the LDPE membrane and seals them in place. The calculated total volume of the receiving phase VD is 600 uL (Vrana et al., 2006). 2.3. Calibration data The calibration data were obtained in experiments designed to measure the uptake of target analytes and offloading of PRC at different combinations of temperature and hydrodynamic conditions in a full factorial design. The calibration data were gathered in order to determine the sampling parameters for uptake of target analytes (sampling rate; Rs) and for the offload of PRCs (overall exchange rate constants; ks) and to observe how they are affected by environmental conditions. Briefly, in each experiment up to 14 passive samplers were exposed for up to 14 days in a constant concentration flow-through exposure system, under controlled conditions of temperature, water turbulence, and analyte concentration. Each factor (temperature, stirring speed) was tested at three levels, resulting in the total number of nine experiments. The experiments have been described in detail and the calibration data reported (Vrana et al., 2006). 2.4. Field performance test To assess the performance of the Chemcatcher for monitoring the target analytes in the field, samplers were deployed in a river. The sampler data were compared with data obtained using spot sampling of water. The sampling site was located in Eijsden at the location of a water quality monitoring station (Rijksinstituut voor Zoetwaterbeheer Integraal Afvalwaterbehandeling — RIZA), near the entry of the River Meuse into The Netherlands. Three replicate Chemcatcher samplers were deployed for 14 days from 24th May to 6th June 2004. During the exposure, the water temperature at the sampling site varied from 18 to 21 °C. On the day of deployment, samplers were transported to the field in a portable coolbox. At the sampling site, transport lids were removed from the samplers and samplers were placed into a protective deployment cage made of a stainless steel perforated sheet with 5 mm square holes. The dimensions of the cage were 350 mm in length, 245 mm in width, 240 mm in height. Samplers were hung in the cage about 20 mm from the top with the membranes facing downwards. The cage was deployed at depth of approximately 1 m below the surface, and was secured to a barge using a rope. To prevent the cage from floating in the current, weights were attached under the cage to a rope 1 m long. On day 14, three replicate samplers were removed from the deployment cage, checked visually for mechanical damage and the extent of biofouling, photographed and sealed with their transportation lids. The samplers were transported to the analytical laboratory in a coolbox. An additional field control sampler was exposed to air while samplers were being deployed and collected. The field control was processed as the deployed samplers and was used to measure contamination during transportation and handling. Three sampler fabrication controls were also analysed to determine contamination arising from the manufacturing process, sampler components, laboratory storage, processing and analytical procedures, but also to determine the initial concentration of PRCs in the samplers before exposure. The procedures of extraction and instrumental analysis by GC/MS of passive sampler extracts in n-octanol have been described (Vrana et al., 2005b). Analyte detection was performed using an MS detector with selected ion monitoring (SIM) of two or three characteristic ions for each compound in both detection and quantification. Detection limits of the method were calculated using the regression line of the chromatographic peak area against as the analyte amount in four external calibration standards in n-octanol with lowest concentrations. Detection limit corresponded to the analyte amount for which the peak area is equal to 3 times the standard deviation of the calibration curve intercept. River water samples were taken by means of specially designed apparatus at the Eijsden monitoring station. This was a continuously running stainless steel tap, fed by a pump. The intake water pump was located at about the same level as the cages with samplers. Six 1-L water samples were taken at regular intervals during the exposure period. Spot samples were filtered through a glass fibre filter (Whatman, 0.7 urn pore size), extracted using three aliquots (100 mL) of dichloromethane. Extracts were reduced in volume using a gentle stream of nitrogen and dried by filtering through sodium sulphate. Quality control samples were also prepared by fortifying pure water with target analytes (added in 0.5 mL acetone solution) and processed as samples. The spiking concentration was 100 ngL~' for each single component. Average percent recoveries of analytes from water ranged between 29% for pentachlor-obenzene and 101% for benzo[6]fluoranthene, respectively. The concentrations of analytes determined in water extracts were corrected using procedural recovery rates. The final volume was adjusted to 200 uL and samples were analysed by GC/MS for contaminant content. 3. Results and discussion 3.1. Mechanism of analyte uptake The mass transfer of a given chemical in a passive sampling device includes several diffusion and interracial mass transport steps across the different barriers that may be present (Vrana et al., 2005a). To obtain a more detailed insight into the mechanism of the accumulation process, the contribution of aqueous and polymer film resistance to the overall mass transfer was characterised. The combination of Eqs. (2) and (3) enables the modelling of the contribution of the resistances of the aqueous boundary layer and the polyethylene membrane to the exchange rate constant ke (or its reciprocal value l/ke, which is the overall sampler residence time t): ke A \&w kMKMwJ where kw and kM are mass transfer coefficients in the aqueous boundary layer and the polyethylene membrane, respectively. For the purpose of the fit, ke values of individual compounds in each experiment were calculated first using Eqs. (5) and (7). Eq. (9) was fitted to the data by nonlinear parameter estimation, adopting a log normal distribution of errors, and by estimating the mass transfer coefficients in the form log k rather than k, to speed up convergence. Details of the parameter estimation are given in supplementary information (Appendix A). According to the two-resistance film theory, moderately hydrophobic compounds should be accumulated under membrane control. With the exception of lindane, a significant increase of sampling rate with increasing flow velocity was observed for all compounds under investigation. Therefore, a test was performed to determine whether the contribution B. Vrana et at. I Environmental Pollution 145 (2007) 895—904 898 of the membrane to the overall resistance to mass transfer as estimated using Eq. (9) was significant. For this purpose the data were fitted to both the model including the membrane contribution and to a simple model in which this was neglected: 1^,1 ke A kw The significance of the contribution of the membrane resistance to explaining the variation in overall resistance to mass transfer was then tested using the extra sum of squares principle as described in Booij et al. (1998). Inclusion of the extra parameter (kM) did not yield a statistically significant reduction in variance and the simpler model was accepted. Thus, with the range of compounds used in our calibration studies it was not possible to calculate the contribution of membrane resistance to mass transfer. The fit results are summarized in Table 1 and shown in Fig. 1. For hydrophobic compounds under investigation (log Kow > 3.5), the transport kinetics are governed by the aqueous boundary layer. This hypothesis is supported by the decrease in ke values with increasing Kow and also by the fact that kw is a function of flow rate/turbulence. Film theory (Cussler, 1984; Jeannot and Cantwell, 1997) hypothesizes a liquid boundary layer of thickness <5W, which is postulated to be completely stagnant and non-convected, so that a solute molecule crosses it by only pure diffusion. At steady state, the aqueous phase mass transfer coefficient can be expressed as: *w=^ (11) <5w The film theory predicts an increase in kw at faster flow rates that decrease <5W. Table 1 Values of mass transfer coefficients for the aqueous boundary layer (£w) obtained as optimized parameters of the nonlinear regression analysis of the overall exchange rate constant log ks as dependent on the octanol—water partition coefficient log K0w using Eq. (10). Regression analysis was performed on data obtained at temperatures 6 °C (Fit 1), 11 °C (Fit 2) and 18 °C (Fit 3), respectively Fit Experiment Temperature Stirring log kw" Rsb no. no. [°C] speed [ms-1] [Ld-1] [rpm] 1 1 6 0 -6.60 ±0.11 0.04 1 2 6 40 -6.73 ±0.11 0.03 1 3 6 70 -6.28 ±0.11 0.08 2 4 11 0 -6.39 ±0.10 0.06 2 5 11 40 -6.00 ±0.10 0.15 2 6 11 70 -5.86 ±0.10 0.21 3 7 18 0 -6.49 ±0.19 0.05 3 8 18 40 -5.81 ±0.12 0.23 3 9 18 70 -5.62 ±0.14 0.36 a Statistical indices of the fits are the number of data points n\ = 45, n2 = 51, n3 = 38; the correlation coefficients rx = 0.95, r2 = 0.96 and r3 = 0.94; and the standard deviations of the fits Si = 0.18, sz = 0.18, S3 = 0.26. The apparent sampling rate Rs of compounds accumulated under aqueous boundary layer was calculated &WA. The calculations of diffusional flux in a fluid show that the mass transfer coefficient kw should be a function of the fluid velocity u, in accordance with law u" for a great variety of geometrical shapes of streamlined bodies and for different types of surface (Levich, 1962; Cussler, 1984). With the exception of the data measured at 6 °C (stirring rates 0 and 40 rpm), the kw increases when the flow increases in agreement with the film theory (Fig. 2). Unfortunately, the hydrodynamics in the experimental setup used in our study are complicated, since sampler bodies are non-streamlined and affect the current profiles and turbulence in the system. Further, the sampler contains sharp edges that are expected to dramatically change the local hydro-dynamic conditions. Therefore, a quantitative prediction of the dependence kw =f(u) using semi-empirical mass transfer calculations used in chemical engineering is difficult. For compounds that are accumulated under aqueous boundary layer control, the apparent sampling rates can be calculated as Rs = kwA (Table 1). Our model does not predict large differences in sampling rates for compounds accumulated under aqueous boundary layer control. In reality, sampling rate decreases with increasing log Kow. One of the factors causing the decrease is that the diffusion coefficient decreases with the increasing molecular size. Huckins et al. (2006) reviewed the literature on mass transfer in fluids and found that the typical empirical relations between the mass transfer coefficient and the diffusion coefficient are of the form kw ~ D^. Based on this relation, the Chemcatcher sampling rates at log AT0w = 7 are expected to be about 80% of the sampling rates at log Kow = 4. We observed that the decrease of sampling rate with increasing hydrophobicity was much sharper. Huckins et al. (2002b) observed a very similar trend for lipid-filled SPMDs and suggested several possible hypotheses to explain the drop in sampling rates for very hydrophobic compounds: (a) the sharp reduction in compound solubility in the polyethylene membrane; (b) potential formation of molecular dimers in the aqueous phase; and (c) the potential over-estimation of dissolved aqueous concentrations due to sorption to dissolved organic carbon (DOC). Hypothesis (a) can be rejected, because the observed sampling rates for very hydrophobic compounds were flow-dependent, indicating aqueous boundary layer control over the mass transfer. The most likely explanation is the underestimation of laboratory-derived sampling rates due to analyte sorption to DOC. Unfortunately, the actual DOC concentration was not measured continuously in the calibration studies. Thus, apparently low sampling rates of very hydrophobic compounds can potentially be caused by artefacts in the measurement of the water concentrations of hydrophobic chemicals. Sampling of DOC-bound residues by solid phase extraction (method to analyse water from the exposure tank in the calibration study) cannot be ruled out. However, there seems to be no alternative sampling equipment that would be suitable for accurate routine measurements of dissolved concentrations at a reasonable cost. Further research is necessary to investigate the impact of DOC on calibration data for all passive sampler designs. In order to estimate the thickness of the aqueous boundary layer (<5W) at the surface membrane of sampler, aqueous B. Vrana et al. I Environmental Pollution 145 (2007) 895—904 899 diffusion coefficients Dw of the test compounds at different temperatures were calculated using Hayduk and Laudie equation (Tucker and Nelken, 1982). The median Dw values (3.09, 3.67 and 4.96 x 10~6 cm2 s_1 at 6, 11 and 18 °C, respectively) were then used to calculate <5W from Eq. (11). In environmental situations the effective thickness of the aqueous boundary layer can vary from about 10 um (extremely turbulent/high flow conditions) to more than 1000 um (deep stratified lakes of deep seas) (Gale, 1998). The estimated boundary aqueous film thickness in this study decreases from more than 1000 um to less than 200 um with increasing flow turbulence. These thicknesses are higher than expected based on the highly turbulent flow conditions in the calibration tank. However, the calculated <5W reflects the local hydrodynamic conditions in the immediate vicinity of the membrane, located 20 40 vs [cm s"1] Fig. 2. The average mass transfer coefficient in the aqueous boundary layer kw as dependent on the linear velocity us of the sampler in the flow-through exposure system at various temperatures. Linear velocity us was calculated as 2tzrf, where r is the radius between the centre of the calibration carousel and the centre of the sampler and / is the rotation frequency. inside a 20 mm deep depression in the sampler body. Thus, the sampler design seems to effectively reduce the convective transport of analytes to the sampler membrane. A modification of the Chemcatcher design that reduces the depth of the cavity in the sampler body would be likely to increase the sampling rates in flowing water, thus improving the sampler's sensitivity, but this would also increase the variation of sampling rates caused by fluctuations in hydrodynamic conditions. 3.2. Empirical uptake rate model The mechanistic model explains the differences in sampling rates among compounds and among exposure conditions and discriminates among compounds accumulated under membrane control and aqueous boundary layer control. For a practical application of the calibration data for interpretation of results obtained with the Chemcatcher passive sampler in field studies, it is more convenient to derive, and easier to use the empirical equation for in situ estimation of sampling rates. Huckins et al. (2002b, 2006) showed that for SPMDs differences in exposure conditions cause sampling rates to be shifted by a constant factor for all compounds. In this study, a similar observation was made. The Chemcatcher log Rs vs. log Kow plots have very similar shapes, but show a varying offset for the different experimental conditions. Therefore, a nonlinear regression was performed for all log-transformed sampling rates Rs from the nine calibration experiments using a third order polynomial function of log Kqw: YiJ=Pi + aXJ + bX] + cX] (12) where Yq is the log Rs of compound j in experiment i, and Xj is the log AT0w value of compound j. The parameters a, b, and c characterize the shape of the hydrophobicity profile of the sampling rates, common for all calibration experiments, and the parameters Pi—Pg represent the offsets for the individual 900 B. Vrana et al. I Environmental Pollution 145 (2007) 895—904 experiments caused by varying environmental conditions. Sampling rates could be described by: log Rs=Pi + 22.755 log/row-4.061 log^ow + 0.2318 log^ow (R = 0.92, s = 0.22, n = 134) (13) Estimates of the optimised parameters a,b,c, and intercepts Pi are summarized in Table 2. A plot of (log /?s—Pj) as a function of log Kow is shown in Fig. 3. The standard deviation of the fit (0.22 log units) corresponds to an uncertainty factor of approximately 1.7, which is relatively good considering the large differences in exposure conditions tested (temperatures between 6 and 18 °C and a wide range of water turbulence) and the corresponding 172-fold difference in sampling rates. Information on concentrations that are accurate within a factor of two, is still highly relevant for environmental risk assessment purposes. Note that the empirical equation is applicable only for interpolation for compounds with log Kow values within the range 3.7—6.8. 3.3. Application of the empirical model to estimate in situ TWA concentrations To assess the applicability of the data obtained with Chem-catcher sampler for measuring TWA water concentrations of non-polar priority pollutants in the field, samplers were deployed for 14 days at a sampling site located at Eijsden in the River Meuse in The Netherlands. The sampler data were compared with spot water samples collected regularly during the sampler deployment period. The amounts of analytes accumulated in the Chemcatcher during field deployment are shown in Table 3. The fabrication blanks and field blanks contained quantifiable levels of phenanthrene, pyrene and benzo [a] anthracene. Quantifiable levels of eight PAHs (acenaphthene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benzo [a] anthracene and Table 2 Values of optimized parameters of the nonlinear regression analysis of the sampling rate logRs [Ld~'] as dependent on the octanol—water partition coefficient log Kow using Eq. (13) Experiment no. Temperature [°C] Stirring speed [rpm] Parameter Parameter value8 All experiments a 22.7549 ± 2.4590 All experiments b -4.0611 ±0.4762 All experiments c 0.2318 ± 0.0302 1 6 0 Pi -42.28 ±4.16 2 6 40 Pi -42.39 ±4.16 3 6 70 P3 -41.92±4.16 4 11 0 P4 -42.03 ±4.16 5 11 40 P5 -41.64 ±4.16 6 11 70 Ps -41.35 ±4.16 7 18 0 Pi -42.36 ±4.17 8 18 40 Ps -41.32±4.16 9 18 70 P9 -41.28 ±4.16 42.0 41.5 - 41.0 - 40.5 - 40.0 39.5 39.0 Fig. 3. Sampling rate as a function of compound hydrophobicity (expressed by log Kow)- The curve represents the best fit of the experimental data with the empirical model (Eq. (13)). The effect of environmental variables temperature and water turbulence is expressed by the offset of the curve on the y-axis (P,) and is eliminated in this projection. chrysene) were found in samplers exposed in the field for 14 days. The levels of acenaphthene, anthracene and benzo [a] anthracene found in field samplers after a 14 day exposure were not significantly different from the fabrication blank values (f-test, a = 0.05). The maximum coefficient of variation (or relative percent difference, where only duplicate samples were available) of samples with concentrations significantly elevated above the blank levels did not exceed 50%. The following algorithm was applied to calculate TWA water concentrations from the amounts of analytes accumulated in samplers during field deployment. 3.3.1. Estimation of the in situ exchange kinetics First, the in situ overall exchange rate constants ke were calculated from the rearranged Eq. (6). k, = rn(mD0/mD) (14) a Statistical indices of the fit are the number of data points n = 134, the correlation coefficient R = 0.92, and the standard deviation of the fit s = 0.22. Using the mean PRC concentrations (from triplicate samples) in the fabrication blanks mD0, the mean concentrations in field samplers mD, and an exposure time of 14 days. To estimate a statistically significant ke value, it is necessary to ensure that the PRC concentration found in field exposed samplers is significantly decreased in comparison with the concentration found in fabrication blanks. One of the options to test this is shown in Appendix A. The test showed that there was a significant offload of all PRC excepting D10-pyrene; the D10-pyrene data was not used in later calculations of sampling rates. The ke values obtained during the field exposure are shown in Table 4. 3.3.2. Calculation of sampling rates of PRCs Sampling rates Rs of the PRCs were calculated using Eqs. (5) and (7). The calculated Rs values are shown in Table 4. 3. Vrana et at. I Environmental Pollution 145 (2007) 895—904 901 Table 3 Mean amount of analytes found in passive samplers (ng per sampler) in fabrication blanks, in the field blank and at the sampling site in the River Meuse after 14 days of exposure Compound Fabrication blank CV (n = 3)a Field blank (n= 1) 14 days exposure CV (n = 3) Acenaphthene <2.4 n.d.b n.d. Fluorene <0.9 n.d. 21 29% Phenanthrene 16 7% n.d. 59 26% Anthracene <0.6 n.d. n.d. Fluoranthene <0.5 n.d. 34 31% Pyrene 7 7% n.d. 38 27% Benzo[a]anthracene 24 7% n.d. n.d. Chrysene <1.5 n.d. 4 16% Benzo[6]fluoranthene <2.9 n.d. n.d. Benzo [k] fluoranthene <2.9 n.d. n.d. Benzo[a]pyrene <2.3 n.d. n.d. Indeno[ 1,2,3-c,d]pyrene <5.4 n.d. n.d. Dibenz[a,/i]anthracene <1.5 n.d. n.d. Benzo [g, h, i] pery lene <3.2 n.d. n.d. Pentachlorobenzene <0.8 n.d. n.d. Hexachlorobenzene <0.5 n.d. n.d. Lindane <1.5 n.d. n.d. Dieldrin <1 n.d. n.d. a Number of replicates. b n.d. = Not detected. 3.3.3. Calculation of sampling rates of analytes The PRC-derived sampling rates were fitted to Eq. (13), using the exposure specific effect Pt as the only adjustable parameter (Table 4). The sampling rates of individual compounds were then estimated from Eq. (13) with the optimized value of parameter Pt. This approach is illustrated in Fig. 4. 3.3.4. Applicability of the linear uptake model The chemical uptake into the sampler remains linear and integrative in the initial period of the exposure until the sorbed amount approaches half of its equilibrium value (ha = hi 2/£e). Our previous study demonstrated that the offload halftime f1/2 of a PRC also characterizes the halftime of uptake saturation of an analogue compound (Vrana et al., 2006). In general, f1/2 increases with increasing affinity to the receiving phase for the compound; in case of the non-polar Chemcatcher sampler it increases with increasing hydropho-bicity of the analyte. The linear uptake model for calculation of TWA water concentrations can be applied only for compounds where the deployment period does not exceed the ty 2 value. The ty2 values of PRCs used in the field study are reported in Table 4. The deployment period did not exceed the ti/2 for any of the compounds. This indicates that the linear uptake model can be applied for all compounds with log Kow > 4. 3.3.5. Calculation of TWA water concentrations from passive sampler data TWA concentrations of target analytes at the sampling site in the River Meuse were estimated from concentrations in the exposed passive samplers using the rearranged Eq. (4): Cw — my - mm R*t (15) where Cw represents the TWA water concentration during the deployment period, mD is the analyte mass found in the sampler after field exposure, mDf is the average mass of analyte found in the field blank, Rs is the estimate of the in situ sampling rate derived as described above, and t equals exposure time. The TWA concentration was calculated as arithmetic average of the three estimates calculated from analyte amounts found in replicate samplers. The uncertainty level of this estimate was expressed as the standard error of the mean. TWA Table 4 Summary of in situ PRC exchange kinetic parameters and distribution coefficients obtained from the 14-day field exposure in the River Meuse. Optimized parameter P, of the empirical model (Eq. (13)) characterizes exposure specific conditions and can be used for calculating substance specific sampling rates Rs PRCs log Kow log ZfDW % PRC K [ 6 at this DOC level. This is in agreement with our observation of a twofold difference between TWA concentrations determined using spot sampling and passive samplers for pyrene and chrysene (log Kow = 5.1 and 5.7, respectively) at the average DOC level at the sampling site during deployment of 5.0 mgLT1. Another reason for the differences observed between spot sampling and passive sampling may be that higher or lower than average concentrations may have been present during the sampling interval between the individual spot samples. Any change in ambient concentration during these intervals is undetected. Despite the inherent difficulties, the current study demonstrates that a passive sampling technique delivers reasonable estimates of TWA concentrations for PAHs with log Kow < 6 when compared with the estimates based on repeat spot sampling. 4. Conclusions This study demonstrated that a calculation of TWA concentrations of waterborne hydrophobic pollutants was possible using the laboratory-derived passive sampler calibration data. Application of the mechanistic uptake model to the calibration data enabled the interpretation of differences in sampling rates among the test compounds and under varying exposure conditions. The model also permitted the classification of the compounds according to the mechanism of uptake, determined on the basis of its physicochemical properties of the compounds. Compounds with log Kow > 3.5 are accumulated in the Chem-catcher sampler under aqueous boundary layer control. Thus, their uptake kinetics is sensitive to both changes in temperature and water turbulence. Moreover, kinetic performance characteristics of the Chemcatcher sampler are likely to change with modifications to the geometry of the sampler body. An empirical relationship was derived that enabled the laboratory-derived sampling rates to be corrected for variations in the in situ exposure conditions. The correction is made using the information on in situ exchange kinetics of PRCs. This study contributes to the growing pool of evidence that supports the validity of using PRCs for the determination of in situ sampling kinetics. This method increases the accuracy of estimates of TWA concentrations obtained using integrative passive samplers. The successful application of the PRC approach with other designs of passive sampler including SPMDs (Booij et al., 1998; Huckins et al., 2002a) and silicone strips (Booij, unpublished data) has been demonstrated. The empirical equation is applicable for calculation of sampling rates of compounds with log KoW values in the range from 3.7 to 6.8. This approach is of value for pollutants for which no calibration data exist. Correlation of the calibration data with molecular descriptors other than octanol/water partition coefficients may bring additional information on the uptake process (Abraham, 1993; Abraham et al., 2004). The validity of this approach has been demonstrated by obtaining reasonable estimates of TWA concentrations for a range of PAHs in a field study. However, there are still a number of problems to be investigated when comparing data obtained from passive sampling with those obtained from spot samples. To improve the reliability of the data obtained with the two sampling methods, there is a need for equipment that can provide information on the concentrations of truly dissolved contaminants present in water at a reasonable cost. This would enable a more precise validation of the passive sampling technology. Acknowledgment We acknowledge the financial support of the European Commission (Contract EVK1-CT-2002-00119; www.port.ac. uk/stamps) for this work. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.envpol.2006. 04.030. References Abraham, M.H., 1993. Scales of solute hydrogen-bonding: their construction and application to physicochemical and biochemical processes. Chemistry Society Reviews 22, 73—83. Abraham, M.H., Ibrahim, A., Zissimos, A.M., 2004. Determination of sets of solute descriptors from chromatographic measurements. Journal of Chromatography A 1037, 29-47. Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L., Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environmental Toxicology and Chemistry 23, 1640—1648. Booij, K., Hofmans, H.E., Fischer, C.V., van Weerlee, E.M., 2003. Temperature-dependent uptake rates of non-polar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environmental Science and Technology 37, 361—366. Booij, K., Sleiderink, H.M., Smedes, R, 1998. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environmental Toxicology and Chemistry 17, 1236—1245. Burkhard, L.P., 2000. Estimating dissolved organic carbon partition coefficients for nonionic organic chemicals. Environmental Science and Technology 34, 4663 -4667. Cussler, E.L., 1984. Diffusion: Mass transfer in Fluid Systems. Cambridge University Press, Cambridge. Flynn, G.L., Yalkowsky, S.H., 1972. Correlation and prediction of mass transport across membranes. I. Influence of alkyl chain length on flux-determining properties of barrier and diffusant. Journal of Pharmaceutical Sciences 61, 838-852. Gale, R.W., 1998. Three-compartment model for contaminant accumulation by semipermeable membrane devices. Environmental Science and Technology 32, 2292-2300. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environmental Science and Technology 27, 2489-2496. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Cranor, W.L., Clark, R.C., Mogensen, B.B., 2002a. Development of the 904 B. Vrana et al. I Environmental Pollution 145 (2007) 895—904 permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environmental Science and Technology 36, 85—91. Huckins, J.N., Petty, J.D., Prest, H.F., Clark, R.C., Alvarez, D.A., Orazio, C.E., Lebo, J.A., Cranor, W.L., Johnson, B.T., 2002b. A guide for the use of semipermeable membrane devices (SPMDs) as samplers of waterborne hydrophobic organic contaminants. Report for the American Petroleum Institute (API). API Publication Number 4690. API, Washington, DC. Huckins, J.N., Petty, J.D., Booij, K., 2006. Monitors of Organic Contaminants in the Environment: Semipermeable Membrane Devices. Springer Verlag, New York (Chapter 3). Jeannot, M.A., Cantwell, F.F., 1997. Mass transfer characteristics of solvent extraction into a single drop at the tip of a syringe needle. Analytical Chemistry 69, 235-239. Kingston, J.K., Greenwood, R., Mills, G.A., Morrison, G.M., Persson, B.L., 2000. Development of novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments. Journal of Environmental Monitoring 2, 487—495. Levich, V.G., 1962. Physicochemical Hydrodynamics. Prentice-Hall, London. Mackay, D., Shiu, W.Y., Ma, K.C., 1992a. Illustrated Handbook of Physical-Chemical Properties of Environmental Fate of Organic Chemicals, vol. 1. Lewis Publishers, MI, USA. Mackay, D., Shiu, W.Y., Ma, K.C., 1992b. Illustrated Handbook of Physical-Chemical Properties of Environmental Fate of Organic Chemicals, vol. 2. Lewis Publishers, MI, USA. Namiesnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M., Wasik, A., 2005. Passive sampling and/or extraction techniques in environmental analysis: a review. Analytical and Bioanalytical Chemistry 381, 279—301. Scheuplein, R.J., 1968. On the application of rate theory to complex multibar-rier flow co-ordinates: membrane permeability. Journal of Theoretical Biology 18, 72-89. Stuer-Lauridsen, F, 2005. Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environmental Pollution 136, 503-524. Tucker, W.A., Nelken, L.H., 1982. Diffusion coefficients in air and water. In: Lyman, W.J., Reehl, W.F, Rosenblatt, D.H. (Eds.), Handbook of Chemical Property Estimation Methods. McGraw-Hill Book Company, New York (Chapter 17). Vrana, B., Mills, G.A., Allan, I.J., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G., Greenwood, R., 2005a. Passive sampling techniques for monitoring of pollutants in water. Trends in Analytical Chemistry 24, 845—868. Vrana, B., Mills, G., Greenwood, R., Knutsson, J., Svensson, K., Morrison, G, 2005b. Performance optimization of a passive sampler for monitoring hydrophobic organic pollutants in water. Journal of Environmental Monitoring 7, 612-620. Vrana, B., Mills, G.A., Dominiak, E., Greenwood, R., 2006. Calibration of the Chemcatcher passive sampler for the monitoring of priority organic pollutants in water. Environmental Pollution 142, 333—343. Vrana, B., Popp, P., Paschke, A., Schüürmann, G., 2001. Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water. Analytical Chemistry 73, 5191—5200. Príloha 12 Schäfer R. B., Paschke A., Vrana B., Mueller R., and Liess M., Performance of the Chemcatcher passive sampler when used to monitor 10 polar and semi-polar pesticides in 16 Central European streams, and comparison with two other sampling methods., Water Res., 2008, 42, 2707-17. WATER RESEARCH 42 (2008) 2707-2717 ELSEVIER Available at www.sciencedirect.com *%* ScienceDirect journal homepage: www.elsevier.com/locate/watres Performance of the Chemcatcher® passive sampler when used to monitor 10 polar and semi-polar pesticides in 16 Central European streams, and comparison with two other sampling methods Ralf Bernhard Schafera'h'*, Albrecht Paschkec, Branislav Vranad, Ralf Mueller6, Matthias Liessa department of System Ecotoxicology, UFZ-Helmholtz Centre for Enuironmental Research, Permoserstrasse 15, 04318 Leipzig, Germany institute for Ecology and Enuironmental Chemistry, Uniuersity Lüneburg, ScharnhorststraJSe 1, 21335 Lüneburg, Germany cDepartment of Ecological Chemistry, UFZ-Helmholtz Centre for Enuironmental Research, Permoserstrasse 15, 04318 Leipzig, Germany dNational Water Reference Laboratory for Slouafeia, Water Research Institute, Nabr. arm. gen. L. Suobodu 7, 81249 Bratislaua, Slouafeia department ET - Laboratory for Enuironmental Analytics, EWE AG, BürgerparfestraJSe 11, 49661 Cloppenburg, Germany article info Article history: Received 15 November 2007 Received in revised form 17 January 2008 Accepted 22 January 2008 Available online 2 February 2008 Keywords: Pesticides Monitoring Pollution Passive sampling Chemcatcher® abstract We investigated the performance of the Chemcatcher®, an aquatic passive sampling device consisting of a sampler body and an Empore® disk as receiving phase, when used to monitor acetochlor, alachlor, carbofuran, chlorfenvinphos, a-endosulfan, fenpropidin, linuron, oxadiazon, pirimicarb and tebuconazole in 16 Central European streams. The Chemcatcher®, equipped with an SDB-XC Empore® disk, detected seven of the aforementioned pesticides with a total of 54 detections. The time-weighted average (TWA) concentrations reached up to 1 |rg/L for acetochlor and alachlor. Toxic units derived from these concentrations explained reasonably well the observed ecological effects of pesticide stress, measured with the SPEAR index. In a follow-up analysis, we compared the Chemcatcher® performance with those of two other sampling systems. The results obtained with the Chemcatcher® closely matched those of the event-driven water sampler. By contrast, the TWA concentrations were not significantly correlated with concentrations on suspended particles. We conclude that the Chemcatcher® is suitable for the monitoring of polar organic toxicants and presents an alternative to conventional spot sampling in the monitoring of episodically occurring pollutants. © 2008 Elsevier Ltd. All rights reserved. 1. Introduction The monitoring of pesticide concentrations in surface waters is an inevitable step for the environmental risk assessment of pesticides. For these compounds, field runoff represents a relevant input path into streams in agricultural areas (Liess et al., 1999; Neumann et al., 2002). Runoff events occur discontinuously in association with heavy precipitation, and runoff-related pesticide exposure may have adverse effects on invertebrate communities (Leonard et al., 2000; Liess and von der Ohe, 2005). Since most pesticide concentrations during runoff events decrease to * Corresponding author at: Department of System Ecotoxicology, UFZ-Helmholtz Centre for Environmental Research, Permoserstrasse 15, 04318 Leipzig, Germany. Tel.: +49 3412351496; fax: +49 341235 2401. E-mail address: ralf.schaefer@ufz.de (R.B. Schäfer). 0043-1354/$- see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2008.01.023 2708 WATER RESEARCH 42 (2008) 2707- 2717 background levels within hours to a few days, routine water monitoring which mainly relies on spot (bottle) sampling at fixed intervals is likely to miss a great proportion of relevant events (Richards and Baker, 1993; Leu et al., 2004). Hence, environmental monitoring techniques are needed that allow for detection of runoff-related peak exposure and that are labour- and cost-efficient at the same time. Continuous water sampling represents an alternative to spot sampling. Throughout the last decade, passive sampling devices using various receiving phases have been employed successfully for continuous monitoring of various pollutants in surface waters (Stuer-Lauridsen, 2005; Vrana et al., 2005). The Chemcatcher® passive sampler with polar receiving phase and the polar organic chemical integrative sampler (POCIS) performed well in the monitoring of polar organic contaminants (Escher et al., 2006; Alvarez et al., 2007). Nevertheless, there is a paucity of studies addressing the monitoring of short-term pollution events with passive samplers (Greenwood et al., 2007). Furthermore, to our knowledge, only one study demonstrated a relationship between pesticide concentrations determined by passive samplers and effects on aquatic communities (Leonard et al., 2000). The establishment of such a relationship is hampered by the fact that time-weighted average (TWA) concentrations are obtained from passive sampling devices, whereas peak concentrations are required to assess potential acute ecotoxicological effects. In this study we present results of a field study at 16 sampling sites using the Chemcatcher® passive sampler to detect the polar and semi-polar pesticides acetochlor, alachlor, carbo-furan, chlorfenvinphos, a-endosulfan, fenpropidin, linuron, oxadiazon, pirimicarb and tebuconazole. The compounds were chosen on the basis of their ecotoxicological relevance in the sampling region (Schäfer et al., 2007a). In addition, we examine the extent to which the TWA concentrations can be related to a community-based biotic index—the Species At Risk (SPEAR)-index—designed to detect effects of pesticides on benthic invertebrates (Liess and von der Ohe, 2005). Since several sampling systems have been proposed to assess runoff-related pesticide exposure, there is also a need to compare the performance of different sampling systems. Therefore, another objective of this study was to compare the performance of the Chemcatcher® with the performances of two other sampling systems: an event-driven water sampler (EDS) and a suspended-particle sampler (SPS). (Technical drawings of all sampling methods can be found in the supplementary data.) Both methods have been proposed and used to catch runoff events in previous studies (Liess et al., 1996, 1999; Schulz et al., 2001; Liess and von der Ohe, 2005) and were deployed at the same sampling sites as the passive samplers in this study (Schäfer et al., 2007a,b). Comparison of the Chemcatcher® to these sampling methods comprised the following criteria: (1) number of pesticides detected and (2) the total number of detections above the limit of quantitation. Since sampling methods should deliver results that are relevant to assess effects on biota, we included as criteria also (3) the ability to explain variation in the SPEAR index. 2. Materials and methods 2.1. Study area Brittany, located in northwestern France, was chosen as the sampling region since (1) agriculture is the predominant land-use type there with 23.5% of the area (27,510 km2) being used for corn (19.2%), vegetable (2.6%), oil-seed (1.2%) and potato (0.5%) production and (2) in Western Europe pesticide usage is the highest globally in terms of expenditures per area (Oerke and Dehne, 2004). A total of 16 sampling sites in small agricultural streams (max. width: 5 m, max. depth: 0.8 m) were selected on the basis that they were expected to exhibit a gradient in pesticide contamination (Schäfer et al., 2007a). 2.2. Preparation, deployment and extraction of the passive sampler The Chemcatcher® passive sampling device (University Portsmouth, UK; commercially available at Alcontrol AB, Linkoping, Sweden) was employed for continuous water monitoring as described by Kingston et al. (2000). The Chemcatcher® consists of a polytetrafluorethylene (PTFE) sampler body and, for the purpose of this study, was equipped with SDB-XC Empore® disks (3M, Neuss, Germany) as the receiving phase (47 mm diameter; 15.9 cm2 surface area) containing polystyrene-divinylbenzene (PS-DVB) as sorbent. Before use, the SDB-XC Empore® disk was conditioned with 10mL acetone (HPLC-grade), 10 mL 2-propanol (analytical grade) and 10 mL methanol (HPLC-grade) obtained from Merck (Darmstadt, Germany). The conditioned disks were placed in the Chemcatcher® body, which was subsequently filled with purified water, closed and stored in zip-lock bags at 4°C until exposure (<48h). To obtain a rapid response to concentration changes, no diffusion-limiting membrane was used. Procedural blanks were stored non-exposed throughout the whole study period. The Chemcatcher® devices were deployed at the 16 sampling sites on 9-11 May for 10-13 days (Fig. 1), prior to a period with expected heavy precipitation according to the local weather forecast (www.meteofrance.com). The samplers were fixed to steel bars approximately 15 cm below the water surface. The open side of the Chemcatcher® was sealed with a copper mesh (mesh size 5 mm) to prevent mechanical damage and suppress biofouling (Vrana et al., 2005). It was directed towards the stream bottom. Four sites were equipped in duplicate and one in triplicate to assess the variability of the pesticide uptake. A field blank was exposed to the air during deployment and retrieval of samplers to account for potential airborne pollution. After exposure, the passive samplers were filled with stream water from the respective site, closed and stored in zip-lock bags at 4 °C in the dark. In the laboratory, the SDB-XC Empore® disks were carefully taken off the PTFE body, dried under vacuum using a vacuum manifold for about 15 min and subsequently eluted twice with 10 mL acetonitrile/methanol. The eluate was gently evaporated to dryness under nitrogen at 30 °C in a 200 mL evaporation vial using a TurboVap 2 concentration workstation (Zymark, Hopkington, USA) and WATER RESEARCH 42 (2008) 2707- 2717 2709 24/05 26/05 28/05 Runoff-Even! Fig. 1 - Sampling scheme for the three monitoring methods in 16 French streams. "Runoff-event" indicates a heavy precipitation event (> 10 mm/day). Table 1 - Physicochemical and analytical data for 10 measured pesticides Compound Type3 Class3 Log Lo| LOQ CC LOQ EDS LOQ SPS LOQ calc. LC50 Y ° ^ow Koc (ng/L)c'd (^kg)c'f (ug/L)a'S Acetochlor H Chloroacetamide 2.39 2.32 5.1 25 12.5 0.26 9000 Alachlor H Chloroacetamide 3.52 2.28 5.4 25 12.5 0.24 10,000 a-Endosulfan I Organochlorine 3.83 4.13 3.6 25 12.5 16.86 75 Carbofuran I Carbamate 2.32 1.75 10.4 25 12.5 0.07 38.6 Chlorfenvinphos I Organic phosphorous acid 3.10 2.47 5.2 25 12.5 0.37 0.3 Fenpropidin F Piperidine 2.90a 3.20' 4.1 25 12.5 1.98 500 Linuronh H Urea derivative 3.20 2.70 4.3 25 12.5 0.63 120 Oxadiazon H Oxadiazole 4.80 3.51 3.5 25 12.5 4.04 2400 Pirimicarb I Carbamate 1.70 1.90 4.5 25 12.5 0.10 17 Tebuconazole F Triazole 3.70a 3.50' 6.1 25 12.9 3.95 4200 a Taken from Tomlin (2003), I = insecticide, H = herbicide, F = fungicide. b Taken from Sabljic et al. (1995). c LOQ = limit of quantification for a sample obtained with the respective method. d CC = Chemcatcher®; computed for 14-day exposure. e For extraction of 10 g of suspended particles. 1 Sample LOQ for suspended particles that would correspond to the level of the EDS LOQ assuming equilibrium partitioning, computed according to LOQcaic = LOQ EDS ■ K0c - foe wnere jocis the mass fraction of organic carbon (assuming/0c = 5%). g LC50 for Daphnia magna. h Quantificated as 3,4-dichloroaniline. 1 Estimated with Chemprop 4.1 (http://www.ufz.de/index.php?en=6738). ' 25 and 100 for some samples with high matrix interference. redissolved with 200 uL acetonitrile. Prior to analysis, 5 uL triphenyl phosphate (TPP) was added as internal standard (IS). 2.3. Chemical analysis The selected compounds (Table 1) were quantified using an Agilent 6890N (Agilent Technologies Germany, Boeblingen, Germany) gas chromatograph (GC) equipped with a MPS2 autosampler, a CAS4 inlet (both from Gerstel, Miihlheim a.d. Ruhr, Germany) and an Agilent 5973 mass selective detector (MSD). The limit of quantification (LOQ) of the GC-MSD was 125pg/uL for all compounds. The sample LOQs differed between the sampling methods and between compounds for the Chemcatcher (Table 1). Typical total ion chromato-grams are given in Fig. 2. 2.4. Calculation of passive sampler TWA concentrations From the field-exposed passive samplers, the accumulated mass of each compound per sampler is obtained. To calculate TWA concentrations, a substance-specific sampling rate Rs, expressed in equivalent volume of sampled water per day, is required. For the compounds of this study, the sampling rates were previously determined in a laboratory flow-through experiment and found to range from 0.1 to 0.5L/day (Gunold et al., 2007). In addition, this calibration study showed that the Chemcatcher® remained in the linear integrative uptake regime for up to 14 days. Using the sampling rates of this study, the TWA concentrations for the sites in our study were calculated according to where Cw is the TWA concentration of the respective analyte in the water phase in the dimension mass/volume and ms is the accumulated mass after exposure time t. The procedural blank and the field blank yielded zero background contamination and had therefore not to be considered in Eq. (1). The calculated TWA concentrations should be regarded as approximation only, because between-site variation in water 2710 WATER RESEARCH 42 (2008) 2707- 2717 500000 450000 : ■-.00000 350000 300000 250000 200000 150000 100000 50000 45 ? ■ ' ' I 1 ■ ' ' 1 ' 1 ' 1 I ■ ' ' ' I ■ ' ■ ■ I ' 1 ' 1 I ' ■ ' ' I '' ■ ■ I 1 ■ ■ ■ I ■ ' ■ ' I ■ ' ■' 1 ' ■ ' ■ I ' ■ ' ■ I ■ ' ■ ' I ■ ■ ■ ■ I ' ' ■ ■ I 1 ' ■ ' I ' 1 ' ' I 1 ■ 1 ' I 1 ' ' ■ I ' ■ ' ' I ' ■ ■ ' I ' ' —I—r-400 5-00 6-00 7-00 8-00 900 1000 11-00 12-00 13-00 14.00 150C 1600 1 7-00 16 00 1900 20 00 21.00 22-00 23.00 24.00 25.00 26.00 27.00 2B.0O 29.00 Time—3 500000 450000 400000 350000 \ Š 300000 S 5 250OOO ■ 3 200000 150000 I 100000 50000 57 8 it 13 910 11 VI 14 400 5 00 6.00 7.00 8.00 9 00 10.00 11.00 12.00 1300 14 00 1500 16.00 17.00 18.00 19 00 2060 21.00 22.00 23.00 24.00 25.00 26.00 27.00 25 00 29.00 4000000 3500000 3000000 S 2500000 1 2000000 1500000 1000000 500000 • < i ' " ■ ■ i ' ■ ■ ■ i ■ ■ i ' i ■ ' ■ ■ i ■ ■ ■ ' i ' ■ ' ' i ' ■ ■ ■ i ■ ■ ■ ■ i ■ ■ ' ■ i ' ' ' < i ■ ■ ■ ■ i ■ ■ ■ ■ i ■ ■ ■ ■ i ■ ■ ' ■ i ' ■ ■ ' i ■ ■ ■ ■ * < ' • ' i ■ ■ ■ 1 i ■ • ■ ■ i * * ■ i ■ ■ ■ ■ i ' ■ ' ' i ' ■ 4 DO 5.00 6 00 7.00 6 00 9 00 1OO0 11-00 12-00 13 00 1400 15.0C 16 00 17 00 18.00 19 00 2000 21,00 22-00 23.00 24 00 25 00 26.00 27.00 26.00 2900 Fig. 2 - Typical total-ion chromatograms for (a) the event-driven water sampler (EDS), (b) the Chemcatcher®, and (c) the suspended-particle sampler (SPS). The samples were spiked with lug/L (SPS 100jig/kg) of pesticide standards. Deuterated internal standards were only used for comparison of the EDS and Chemcatcher®. Please note the different scaling of the y-axis for the SPS chromatogram. Analytes: 1: carbofuran, 2: pirimicarb D6, 3: pirimicarb, 4: acetochlor Dll, 5: acetochlor, 6: alachlor D13, 7: alachlor, 8: fenpropidin, 9: chlorfenvinphos D10, 10: chlorfenvinphos, 11: a-endosulfan D4, 12: a-endosulfan, 13: oxadiazon, 14: tebuconazol. temperature and biofouling were not taken into account, as the performance reference compound (PRC) concept (Huckins et al., 2002) was not applicable (Gunold et al., 2007). 2.5. Linking exposure to the SPEAR index We examined the extent to which the TWA concentrations determined with the Chemcatcher® can explain variation in the SPEAR index. Briefly, the SPEAR index predicts the effects of organic toxicants on the invertebrate community of a site, based upon traits of benthic invertebrates such as voltinism, migration potential, emergence time and physiological sensitivity (Liess and von der Ohe, 2005). Practically, these traits are used to classify the observed macroinvertebrate community of each sampling site into taxa potentially sensitive or tolerant towards organic toxicants. Subsequently, the SPEAR index value for a respective site is derived by computing the relative abundance of sensitive species in a community. WATER RESEARCH 42 (2008) 2707- 2717 2711 Details on the sampling of the benthic invertebrates and on the computation of the SPEAR index are given in Schäfer et al. (2007a). To assess and standardize the toxicity of the measured TWA concentrations, a log-transformed maximum toxic unit (TU) was computed using the 48-h acute median lethal concentration (LC50) for Daphnia magna (Table 1) as described by Schäfer et al. (2007a). A TU value of -5 was assigned to a site if no pesticide was found, corresponding to unpolluted sites in a previous study (Liess and von der Ohe, 2005). 2.6. Description of the EDS The EDS was designed to catch peak concentrations during pesticide runoff. The sampling system set into the streams consisted of a 1-L glass bottle fixed to a steel bar and was mounted approximately 5 cm above normal water level (Liess et al., 2001; Schulz et al., 2001). After a heavy rain event (>10mm precipitation/24 h) the filled sample bottles were retrieved and water samples were solid-phase-extracted using 6mL Chromabond HR-P columns containing 500 mg of PS-DVB, purchased from Macherey-Nagel (Düren, Germany), according to the method described in Schäfer et al. (2007a). The eluates were treated as described for the Chemcatcher®. The EDS monitoring results reported here refer to a single heavy-rain event (> 10 mm/day) during the study period that occurred between 12 and 13 May (Fig. 1). The TUs of this method were taken from Schäfer et al. (2007a). 2.7. Description of the SPS The SPS was designed to sample suspended particles and consisted of a 3-L sedimentation vessel that was buried in the streambed. Suspended particles that entered therein could settle down (Liess et al., 1996). The sampled suspended material was collected at 2-week intervals, freeze-dried and passed through a 2-mm sieve to remove needles, sticks and leaf parts. Approximately 10 g (dry weight) of the sample was extracted using an accelerated solvent extraction (ASE 200 system from Dionex, Idstein, Germany; extraction parameters: two 6-min cycles with ethyl acetate-acetone (2:1) at 110 °C and 11 MPa) with subsequent size exclusion chromatography (SEC) cleanup (Biobeads S-X3 cleanup column from Antec GmbH, Sindelsdorf, Germany) as described by Schäfer et al. (2007b). Due to matrix interferences the collected fraction in SEC was not evaporated further than to 1000 uL and, subsequently, 50 uL TPP was added as IS. To obtain comparable data sets, we used the results of the sampling period between 6 and 23-26 May for this method (Fig. 1). A maximum sediment TU was computed from the suspended particle concentrations as described in Schäfer et al. (2007b). Log-transformed sediment TUs are referred to as STU. 2.8. Data analysis Pearson's correlation coefficient r was calculated to indicate the similarity of two sampling methods followed by a t-test to detect significant correlations. Observations that were below LOQ for a compound at a certain site and for all sampling methods were excluded from analysis. In case an observation below LOQ corresponded to a measurement above LOQ in another sampling method, the observation below LOQ was replaced by half the LOQ. This substitution by a constant proved to be most reliable for small data sets in a comparative study (Clarke, 1998). Linear models were constituted (1) to analyse if the linear regression for two sampling methods differed significantly between sites or compounds which were included as covariate factors, and (2) to examine the explanatory power of TU (STU for SPS) for variation in the SPEAR index. Due to the low number of replicates (2 and 3) we calculated the relative range (RR) as dispersion measure for the TWA concentrations: RR(%) = (maX(X)Zmin(X)), (2) X where X are the observations for the respective compound at a certain site and X is the mean of X. The RR is a more conservative estimate of the sample dispersion compared to the relative standard deviation (RSD). All statistical computations and graphics were created with the open-source software package R (www.r-project.org) using version 2.6 (for Mac OS X, 10.4.10). 3. Results 3.1. Pesticide monitoring with the Chemcatcher® passive sampler At the 16 sites, seven of the 10 target pesticides were found with the Chemcatcher® passive sampler (Table 2); those not detected were chlorfenvinphos, a-endosulfan and fenpropi-din. Both chloroacetamide herbicides—acetochlor and ala-chlor—were detected most frequently above the LOQ and had the highest TWA concentrations, reaching up to 1 ug/L. Tebuconazole and pirimicarb were found only occasionally and had the lowest TWA concentrations. The TWA concentrations exhibited high variation at three of the five sampling sites with up to 150% in terms of RR (Table 2). The other sites showed medium (<50% RR) and low (<30% RR) variation for the majority of the compounds. The TUs for the sites ranged from -2.4, corresponding to 1/250 the LC50 of D. magna, to -5 (Table 2). The TU values explained reasonably well the variation in the SPEAR index (r2 = 0.5, p<0.01, n = 16) (Table 3), indicating effects of pesticides on the abundance of sensitive invertebrate taxa. 3.2. Comparison of the three sampling methods concerning pesticide monitoring All pesticides of the monitoring program were found in the water samples of the EDS and this sampling method yielded also a slightly higher number of total detections compared to the Chemcatcher® (Table 3). Nevertheless, the pesticide concentrations found by the two water sampling methods were significantly correlated (r = 0.79, p<0.01, n = 75). The concentrations determined with the EDS were in general a factor of 4-5 higher than the Chemcatchers' TWA concentrations (Fig. 3). The linear regression model, encompassing EDS' Table 2 - Time-weighted average concentrations in ng/L (+ relative rangea where replicates available) of pesticides determined with the Chemcatcher1 passive sampler as well as TUs and STUs for the three sampling methods'3 Site Acetochlor Alachlor Carbofuran Linuron Oxadiazon Pirimicarb Tebuconazole TU CCC TU EDSc'd STU SPSc'e 1 1158 184 124 54 10 bq bq -2.5 -0.4 0.7 2 14 7 21 bq bq bq bq -3.3 -2.2 -5.0 3 18 198 bq 37 bq bq bq -3.5 -2.7 2.5 4 196 40 36 9 7 bq bq -3.0 -2.5 -2.2 5 219 96 127 48 8 bq 6 -2.5 -2.0 1.1 6 60 (±148%) 12 (±99%) bq 16 (±94%) 4 (±72%) 5 (±86%) bq -2.6 -2.5 -5.0 7 37 132 92 57 bq 8 bq -3.5 -2.1 -5.0 8 454 (±102%) 681 (±99%) 159 (±27%) 41 (±116%) 9 (±103%) bq bq -2.4 -0.8 0.9 9 486 (±29%) 1233 (±14%) 52 (±22%) 22 (±25%) bq bq 15 (±10%) -2.9 -2.6 -2.0 10 388 (±55%) 182 (±44%) 20 (±13%) 66 (±48%) 26 (±95%) 6 (±26%) 11 (±33%) -3.3 -2.8 -4.1 11 20 14 bq bq bq 12 bq -3.2 -2.6 -5.0 12 bq bq bq bq bq bq bq -5.0 -5.0 -5.0 13 16 (±120%) 24 (±139%) bq bq bq bq bq -5.0 -4.7 1.0 14 bq bq bq bq bq bq bq -5.0 -5.0 -2.7 15 bq bq bq bq bq bq bq -5.0 -5.0 -5.0 16 bq bq bq bq bq bq bq -5.0 -5.0 1.2 a n = 2, except site 8 (n = 3). Calculated using Eq. (2). b bq = below limit of quantification; chlorfenvinphos, a-endosulfan and fenpropidine are not displayed because all observations were below limit of quantification. c Calculated with LC50 values taken from Tomlin (2003), see Table 1. d Calculated from data given in Schäfer et al. (2007a). e Calculated from data given in Schäfer et al. (2007b). WATER RESEARCH 42 (2008) 2707- 2717 2713 Table 3 - Comparison of the three sampling systems in 16 French sites Sampling Number of different pesticides method detected Total detections above the LOQ Explanatory power for the SPEAR index3 Chemcatcher® 7 EDS 10 SPS 5 54 66 22b r2 = 0.50 (p<0.01) r2 = 0.38 (p = 0.01) r2 = 0.01 (p>0.05) a Linear regression with the respective TUs/STUs as b Significantly lower than the total detections by the explanatory variable and SPEAR as response variable. other methods in multiple comparison tests (x2-test with Bonferroni correction, p<0.05). 20 50 100 200 500 1000 2000 EDS peak concentration [ng/L] Fig. 3 - Relationship between the Chemcatcher® TWA concentrations and the EDS peak concentrations in 16 agricultural streams, on a double logarithmic scale. Observations that were below LOQ for both sampling methods were excluded from analysis. Model parameters: r2 = 0.4, p<0.01, n = 75. Model parameters for non log-transformed concentration: r2 = 0.62, p<0.01, n = 75. concentrations as explanatory variable and the Chemcatch-ers' concentrations as response variable, was not significantly different between sites or compounds (analysis of variance of the models with and without the covariate factors, F-test, p>0.05). For the log-transformed pesticide concentrations inclusion of the covariate compounds in the linear model did increase the amount of explained variance significantly (analysis of variance, F-test, p<0.01). However, separate linear regression models for each compound yielded only two significant relationships (t-test, p<0.05) (Fig. 4). In the suspended particles sampled with the SPS, only 5 of the 10 pesticides were observed; any of the compounds alachlor, carbofuran, linuron, oxadiazon and pirimicarb was found. The total number of pesticide detections (22) in the particulate phase was significantly reduced (/2-test with Bonferroni correction, p<0.05) compared to both water phase methods (Table 3). No significant correlations were observed between water concentrations derived from the EDS and the Chemcatcher® on the one hand and the suspended particle concentrations monitored with the SPS on the other hand (r = 0.05 and 0.08, p>0.05, n = 76 and 72, respectively). 3.3. Comparison of the three sampling methods concerning effects assessment The STUs calculated on the basis of suspended particle concentrations were higher than the TUs based on water concentrations, with a maximum STU value of 2.5 corresponding to 321 times the LC50 for D. magna. For water concentrations, the TUs peaked at -0.42, equivalent to 1/2.5 the LC50 value for D. magna (Table 2). The TUs of the two water sampling methods were very similar, indicated by an r of 0.94 (p<0.01, n = 16). The SPEAR index was reasonably well explained by the TUs of the EDS and the Chemcatcher®, whereas no significant linear relationship was observed between STUs and SPEAR (Table 3). 4. Discussion 4.1. Using the Chemcatcher® for the monitoring of polar and semi-polar pesticides The Chemcatcher® passive sampler equipped with a SDB-XC Empore® disk detected all compounds included in the monitoring program except fenpropidin, chlorfenvinphos and a-endosulfan, although these compounds were found in samples obtained by the other sampling methods. In general, the Chemcatcher® should be suitable for detecting these substances, as they showed above average uptake rates in the samplers' receiving phase in a calibration study (Gunold et al., 2007). The non-detections with the Chemcatcher® are not likely to result from too low concentrations because in the EDS samples the concentrations of fenpropidin, chlorfenvinphos and a-endosulfan were not lower than those of the other monitored compounds. An explanation for the non-detection with the Chemcatcher® is that the period of exposure to these pesticides was shorter than in the case of the other compounds detected, resulting in a TWA concentration below LOQ. Since we have no temporal resolution of the water concentrations over the course of the runoff event, this issue remains unresolved. The levels of the TWA concentrations observed with the Chemcatcher® are in good agreement with another field study on 7 sites in southern England using the POCIS passive sampler, where concentrations up to 1 ug/L were reported for 2714 WATER RESEARCH 42 (2008) 2707- 2717 120 100 80 60 40 20 0 Linuron Oxadiazon 120 - 120 - 100 - r= 0.84; p = 0.04 100 - 80 - 80 - 0 ° 60 - 60 - 8 40 - 40 - 0 0 20 - 20 - 0 - 0 - 0 20 40 60 80 100120 Aceto chlor Hinmicarb 120 ■ 100 ■ 80 ■ 60 ■ 40 ■ 20 ■ 0 ■ lebuconazole 120 - 100 - 80 - 60 - 40 - 20 - o 0 - alpha.bndosultan ° 0 20 40 60 80 100120 Alachlor 0 20 40 60 80 100120 Carboturan 0 20 40 60 80 100120 0 20 40 60 80 100120 r = 0.35; p = 0.36 120 100 80 60 40 20 0 Chlortenvinphos 120 - 100 - 80 - 60 - 40 - 20 - o is 0 - henpropidin coro oa 0 500 1000 1500 2000 0 500 1000 1500 2000 0 100200300400500600 EDS peak concentration [ng/L] 0 20 40 60 80 100120 0 20 40 60 80 100120 Fig. 4 - Relationship between the Chemcatcher® TWA concentrations and the EDS peak concentrations in 16 agricultural streams, for single compounds. Observations that were below LOQ for both sampling methods were excluded from analysis. Dashed lines indicate LOQ, r = Pearson's correlation coefficient. Regression lines are shown for > 3 observations above LOQ for both methods. Diuron (Alvarez et al., 2004). Concerning variation in TWA concentrations for replicate deployments of passive samplers, some studies reported similar findings (Stuer-Laurid-sen, 2005; Alvarez et al., 2007), while another study with the Chemcatcher® found lower variability (RSD<20%, n = 2), though the exposure time was 3-fold reduced compared to our study (Escher et al., 2006). Variation in the rate of uptake into the receiving phase may result from differences in biofouling and environmental conditions such as temperature or current velocity. Since environmental conditions are nearly identical within a single sampling point, we suggest that the variation in our study resulted from the high biofouling that was observed on the samplers after deployment (Greenwood et al., 2007). Therefore, new techniques are needed for polar passive samplers that help to reduce variability during field exposure, such as the PRC approach for non-polar compounds (Alvarez et al., 2007). The derived TUs could reasonably well explain variation in the SPEAR index (Table 3). This suggests that variation in the composition of the invertebrate community could partly be attributed to pesticide stress and hence that the relative abundance of taxa classified as sensitive according to the SPEAR approach is reduced due to pesticides. A link between TWA concentrations and ecological effects was also found in two other studies (Leonard et al., 2000; Escher et al., 2006). Firstly, runoff-related endosulfan concentrations in passive samplers deployed in the Namoi river in Australia could be linked to the decline in invertebrate population densities (Leonard et al., 2000). Moreover, the Chemcatcher® was successfully employed to monitor herbicides and assess phytotoxicity in a small-scale field study in Australia (Escher et al., 2006). However, caution should be taken when relating TWA concentrations to effects on biota because no distinction can be made between a low-level chronic contamination and a short-term peak contamination on the basis of TWA concentrations. In a situation in which both chronic contamination and peak contamination are present, no link may be found between TWA concentrations and ecological effects. Furthermore, the relationship between TWA concentrations and biotic metrics will most likely not hold in situations in which more than one peak event occurs during the exposure time. Nevertheless, passive samplers with a polar receiving phase may constitute a labour- and cost-efficient tool for field monitoring of polar organic toxicants when the exposure characteristics are known and episodic events are rare. 4.2. Comparison of the Chemcatcher® with the EDS The Chemcatcher® passive sampler had a slightly lower number of total detections than the EDS (Table 3), but the concentrations were closely related (r = 0.79, p<0.01, n = 75). Since the EDS sampled only one precipitation-driven runoff event (Fig. 1), the similarity of the TWA and EDS concentrations suggests that this event was the most relevant source of the pesticides sampled with the Chemcatcher®. Thus, our findings emphasize the relevance of field runoff as input path for pesticides in aquatic ecosystems and hence are in accordance with the results of previous studies in streams of Germany (Liess et al., 1999; Neumann et al., 2002). On average, the TWA concentrations were 4- to 5-fold lower than the EDS concentrations (Fig. 3). The concentrations determined with the EDS were assumed to represent peak concentrations during runoff (Liess et al., 2001; Schulz et al., 2001). Assuming that concentrations following runoff events drop to below 10% of the peak water concentration within 1-4 days (Richards and Baker, 1993; Leu et al., 2004), one would expect the TWA water concentrations to be in the range of jTj-jj of the EDS concentrations, based on an average exposure time of 12 days (Eq. (1)). Furthermore, this should be dependent on physicochemical properties of investigated pesticides and thus lead to significant differences between compounds. Indeed, we observed a significant difference in the relationship between TWA and peak concentrations for different compounds, though only for log-transformed concentrations. Furthermore, the slopes of the regression lines were different in separate linear regressions for the various WATER RESEARCH 42 (2008) 2707- 2717 2715 compounds (Fig. 4). Nevertheless, we are aware that more extensive data are needed to prove these differences between compounds. 4.3. Comparison of the Chemcatcher® with the suspended-particles sampler Only five pesticides were detected on the suspended particles sampled with the SPS, and the total number of detections was significantly lower compared to the Chemcatcher® (Table 3). This may be explained by the polarity of the study compounds in view of the fact that the pesticides not detected had a log Kow <3.1 except for oxadiazon (Table 1). Moreover, the smaller number of observations related to the SPS samples may be partly due to the LOQ because it was a factor of 3-180 higher than the corresponding LOQs of the water samplers except for a-endosulfan, when assuming equilibrium partitioning between water and particulate phase (see LOQ calc, Table 1). The LOQ for the SPS could only be improved by stronger preconcentration of the eluate or by extracting an increased mass of suspended particles. Besides the fact that the amount of sample material from SPS was rather limited, both possibilities were hampered by the high magnitude of matrix coextraction masking the analyte peaks (Fig. 2). Thus, a more efficient SEC or solid phase extraction cleanup method for polar pesticides would be needed to achieve a lower LOQ (Dabrowska et al., 2003; Schäfer et al., 2007b). Consequently, the particle-associated pesticide concentrations exhibited no significant correlation with the TWA concentrations or the EDS peak concentrations which refer to the dissolved water phase. This low similarity was also expressed by the proportion of cases in which pesticides were found on suspended particles but not in samples collected by either the Chemcatcher® or the EDS. Similarly, no clear relationship between particle-associated contaminants and water concentrations was found in a 1-year monitoring study of 30 organic pesticides in six rivers in the UK (Long et al., 1998). Furthermore, high variability of the pesticide distribution between particulate and water phase was observed in tributaries of the Mississippi river (Pereira and Rostad, 1990) and in a field experiment on the release of six organic pesticides from a heavy clay soil during precipitation events (Brown et al., 1995). The contaminant distribution between particulate and water phase is influenced by environmental conditions, physicochemical properties and site-specific conditions that may explain the observed variation: (1) size of suspended particles, (2) composition and structure of organic matter in the particles (Zhou et al., 1995), (3) runoff-water flow rate (Gouy et al., 1999) and (4) lag time between pesticide application and runoff event. This variation in the pesticide partitioning between particulate and dissolved phase (Brown et al., 1995; Long et al., 1998) along with the high LOQ can explain why the results of the sampling with the SPS and the Chemcatcher® were very different. Although the SPS samples indicated much higher pesticide stress in terms of STU compared to the TUs derived from the TWA and peak concentrations, no significant relationship could be established to the SPEAR index. By contrast, other studies demonstrated significant linear relationships between STUs derived from bed sediments and the benthic community tolerance metrics (Wildhaber and Schmitt, 1998) or macroinvertebrate community composition (Friberg et al., 2003). The differing results of our study most likely result from monitoring suspended particle concentrations instead of bed-sediment concentrations. Suspended particles in field runoff usually have much higher contaminant concentrations than bed sediments and are rarely in equilibrium with the water phase, rendering questionable the application of the STU approach (Liess et al., 1996; Long et al., 1998). In the present study, results from passive sampling and event-driven water sampling were more informative when used to explain variation in the invertebrate community. We propose that water concentrations are more likely to explain effects of episodic events with polar toxicants, whereas the effects of chronic exposure to hydrophobic compounds may be predicted from analysis of the sediment phase. However, this should be tested in future studies, and passive samplers in different configurations can be useful tools for such studies. 5. Conclusions • The Chemcatcher® can be employed for continuous water sampling of polar organic toxicants for up to 14 days. • The Chemcatcher® configured with a SDB-XC Empore® and without diffusion-limiting membrane represents a promising method for the monitoring of short-term exposure that conventional spot water sampling is likely to miss. • Given the increasing attention that is paid to polar substances, a method similar to the performance reference compound concept is needed to account for variation in the passive sampling of polar compounds. • Exposure assessment with the Chemcatcher® passive sampler yields results similar to water sampling but differs from suspended-particles sampling. • In large-scale studies with frequently recurring pollution events, the Chemcatcher® is more labour- and cost-efficient than event-driven water sampling. Acknowledgements The authors are grateful to Graham Mills, Richard Greenwood and Jochen Mueller for support with the Chemcatcher® study. We would like to thank Laurent Lagadic, Thierry Caquet, Marc Roucaute and all the other people who contributed facilities to the field study. We thank Miro Vrana for providing the technical drawing of the Chemcatcher prototype. Special thanks to Bettina Egert and Henning Freitag for the chemical analyses. Peter von der Ohe and two anonymous reviewers provided valuable suggestions that improved the manuscript. The first author received funding through a scholarship of the "Studienstiftung des deutschen Volkes e.V." (Bonn, Germany). A.P., B.V and R.B.S. are also grateful to the British Council & 2716 WATER RESEARCH 42 (2008) 2707- 2717 German Academic Exchange Service (ARC project no. 1239) for financial support. Appendix A. Supplementary materials Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2008.01.023. references Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L., Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environ. Toxicol. Chem. 23 (7), 1640-1648. Alvarez, D.A., Huckins, J.N., Petty, I.D., Jones-Lepp, T., Stuer-Lauridsen, E, Getting, D.T., Goddard, J.P., Gravell, A., 2007. Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS). In: Greenwood, R., Mills, G.A., Vrana, B. (Eds.), Comprehensive Analytical Chemistry 48: Passive Sampling Techniques in Environmental Monitoring. Elsevier, Amsterdam, pp. 171-197. Brown, CD., Hodgkinson, R.A., Rose, D.A., Syers, J.K., Wilcockson, SJ., 1995. Movement of pesticides to surface waters from a heavy clay soil. Pestic. Sei. 43 (2), 131-140. Clarke, J.U., 1998. Evaluation of censored data methods to allow statistical comparisons among very small samples with below detection limit observations. Environ. Sei. Technol. 32 (1), 177-183. Dabrowska, H., Dabrowski, L., Biziuk, M., Gaca, ]., Namiesnik, ]., 2003. Solid-phase extraction clean-up of soil and sediment extracts for the determination of various types of pollutants in a single run. J. Chromatogr. A 1003 (1-2), 29-42. Escher, B.I., Quayle, P., Muller, R., Schreiber, U., Mueller, J.F., 2006. Passive sampling of herbicides combined with effect analysis in algae using a novel high-throughput phytotoxicity assay (Maxi-Imaging-PAM). J. Environ. Monitor. 8 (4), 456-464. Friberg, N., Lindstrom, M., Kronvang, B., Larsen, S.E., 2003. Macroinvertebrate/sediment relationships along a pesticide gradient in Danish streams. Hydrobiologia 494 (1-3), 103-110. Gouy, V., Dur, J.C., Calvet, R., Belamie, R., Chaplain, V, 1999. Influence of adsorption-desorption phenomena on pesticide run-off from soil using simulated rainfall. Pestic.Sci. 55 (2), 175-182. Greenwood, R., Mills, G.A., Vrana, B., Allan, I J., Aguilar-Martinez, R., Morrison, G., 2007. Monitoring of priority pollutants in water using Chemcatcher passive sampling devices. In: Greenwood, R., Mills, G.A., Vrana, B. (Eds.), Comprehensive Analytical Chemistry 48: Passive Sampling Techniques in Environmental Monitoring. Elsevier, Amsterdam, pp. 199-229. Gunold, R., Schäfer, R.B., Paschke, A., Schüürmann, G., Liess, M., 2007. Calibration of the Chemcatcher passive sampler for monitoring selected polar and semi-polar pesticides in surface water. Environ. Pollut., in press, doi:10.1016/j.envpol.2007. 10.037. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Clark, R.C., Mogensen, B.B., 2002. Development of the permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environ. Sei. Technol. 36 (1), 85-91. Kingston, J.K., Greenwood, R., Mills, G.A., Morrison, G.M., Persson, L.B., 2000. Development of a novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments. ]. Environ. Monitor. 2 (5), 487^195. Leonard, A.W., Hyne, R.V., Lim, R.P., Pablo, E, Van den Brink, P.J., 2000. Riverine endosulfan concentrations in the Namoi River, Australia: link to cotton field runoff and macroinvertebrate population densities. Environ. Toxicol. Chem. 19 (6), 1540-1551. Leu, C, Singer, H., Stamm, C, Muller, S.R., Schwarzenbach, R.P., 2004. Simultaneous assessment of sources, processes, and factors influencing herbicide losses to surface waters in a small agricultural catchment. Environ. Sei. Technol. 38 (14), 3827-3834. Liess, M., von der Ohe, P.C., 2005. Analyzing effects of pesticides on invertebrate communities in streams. Environ. Toxicol. Chem. 24 (4), 954-965. Liess, M., Schulz, R., Neumann, M., 1996. A method for monitoring pesticides bound to suspended particles in small streams. Chemosphere 32 (10), 1963-1969. Liess, M., Schulz, R., Liess, M.H.-D., Rother, B., Kreuzig, R., 1999. Determination of insecticide contamination in agricultural headwater streams. Water Res. 33 (1), 239-247. Liess, M., Schulz, R., Berenzen, N., Nanko-Drees, ]., Wogram, ]., 2001. Pesticide Contamination and Macroinvertebrate Communities in Running Waters in Agricultural Areas. UBA Texte 65, Umweltbundesamt, Berlin, 227pp. Long, J.L.A., House, W.A., Parker, A., Rae, J.E., 1998. Micro-organic compounds associated with sediments in the Humber rivers. Sei. Total Environ. 210 (1-6), 229-253. Neumann, M., Schulz, R., Schäfer, K., Müller, W, Mannheller, W, Liess, M., 2002. The significance of entry routes as point and non-point sources of pesticides in small streams. Water Res. 36 (4), 835-842. Oerke, E.C., Dehne, HW, 2004. Safeguarding production-losses in major crops and the role of crop protection. Crop Prot. 23 (4), 275-285. Pereira, W.E., Rostad, C.E., 1990. Occurrence, distributions, and transport of herbicides and their degradation products in the lower Mississippi river and its tributaries. Environ. Sei. Technol. 24 (9), 1400-1406. Richards, R.P., Baker, D.B., 1993. Pesticide concentration patterns in agricultural drainage networks in the Lake Erie basin. Environ. Toxicol. Chem. 12 (1), 13-26. Sabljic, A., Güsten, H., Verhaar, H., Hermens, J., 1995. QSAR modeling of soil sorption-improvements and systematics of Log Koc vs. Log Kow correlations. Chemosphere 31 (11-12), 4489-4514. Schäfer, R.B., Caquet, T., Siimes, K., Mueller, R., Lagadic, L., Liess, M., 2007a. Effects of pesticides on community structure and ecosystem functions in agricultural streams of three biogeo-graphical regions in Europe. Sei. Total Environ. 382 (2-3), 272-285. Schäfer, R.B., Mueller, R., Brack, W, Wenzel, K.-D., Streck, G., Liess, M., 2007b. Determination of 10 particle-associated multiclass polar and semi-polar pesticides from small streams using accelerated solvent extraction. Chemosphere 70 (11), 1952-1960. Schulz, R., Peall, S.K.C., Dabrowski, J.M., Reinecke, A.J., 2001. Current-use insecticides, phosphates and suspended solids in the Lourens River, Western Cape, during the first rainfall event of the wet season. Water SA 27 (1), 65-70. Stuer-Lauridsen, E, 2005. Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environ. Pollut. 136 (3), 503-524. Tomlin, C.D.S., 2003. The Pesticide Manual, A World Compendium. BCPC Publications, Hampshire, UK. Vrana, B., Allan, IJ., Greenwood, R., Mills, G.A., Dominiak, E., Svensson, K., Knutsson, ]., Morrison, G., 2005. Passive sampling WATER RESEARCH 42 (2008) 2707- 2717 2717 techniques for monitoring pollutants in water. Trends Anal. Chem. 24 (10), 845-868. Wildhaber, M.L., Schmitt, C J., 1998. Indices of benthic community tolerance in contaminated Great Lakes sediments: relations with sediment contaminant concentrations, sediment toxi- city, and the sediment quality triad. Environ. Monitor. Assess. 49 (1), 23-49. Zhou, J.L., Rowland, S., Mantoura, R.F.C., 1995. Partition of synthetic pyrethroid insecticides between dissolved and particulate phases. Water Res. 29 (4), 1023-1031. Príloha 13 Lobpreis T., Vrana B., Dominiak E., Dercová K., Mills G. A., and Greenwood R., Effect of housing geometry on the performance of ChemcatcherTM passive sampler for the monitoring of hydrophobic organic pollutants in water, Environ. Pollut., 2008,153, 706-710. ELSEVIER Available online at www.sciencedirect.com *l0 ScienceDirect Environmental Pollution 153 (2008) 706-710 Short communication ENVIRONMENTAL POLLUTION www.elsevier.com/locate/envpol Effect of housing geometry on the performance of Chemcatcher™ passive sampler for the monitoring of hydrophobic organic pollutants in water Tomáš Lobpreis a, Branislav Vrana b'*, Ewa Dominiak0, Katarína Dercováa, Graham A. Mills d, Richard Greenwoode a Faculty of Chemical and Food Technology, Slovak University of Technology, Radlinského 9, 81237 Bratislava, Slovakia Water Research Institute, Nabr. arm. gen. L. Svobodu 7, 81249 Bratislava, Slovakia c Department of Analytical Chemistry, Chemical Faculty, Gdansk University of Technology, 80 952 Gdansk, G. Narutowicza 11/12, Poland d School of Pharmacy and Biomedical Sciences, University of Portsmouth, St Michael's Building, White Swan Road, Portsmouth P01 2DT, United Kingdom " School of Biological Sciences, University of Portsmouth, King Henry Building, King Henry I Street, Portsmouth POl 2DY, United Kingdom Received 5 June 2007; received in revised form 30 August 2007; accepted 5 September 2007 The effect of passive sampler geometry on accumulation kinetics of organic pollutants from water was evaluated. Abstract Passive sampling of pollutants in water has been gaining acceptance for environmental monitoring. Previously, an integrative passive sampler (the Chemcatcher™) was developed and calibrated for the measurement of time weighted average concentrations of hydrophobic pollutants in water. Effects of physicochemical properties and environmental variables (water temperature and turbulence) on kinetic and thermodynamic parameters characterising the exchange of analytes between the sampler and water have been published. In this study, the effect of modification in sampler housing geometry on these calibration parameters was studied. The results obtained for polycyclic aromatic hydrocarbons show that reducing the depth of the cavity in the sampler body geometry increased the exchange kinetics by approximately twofold, whilst having no effect on the correlation between the uptake and offload kinetics of analytes. The use of performance reference compounds thus avoids the need for extensive re-calibration when the sampler body geometry is modified. © 2007 Elsevier Ltd. All rights reserved. Keywords: Chemcatcher™; Passive sampling; Water monitoring; Hydrophobic organic pollutants; Polycyclic aromatic hydrocarbons 1. Introduction Passive sampling devices are gaining acceptance as tools that can be used in monitoring programmes to measure concentrations of pollutants dissolved in water (Vrana et al., 2005a). One of these, the Chemcatcher™ passive sampler, was developed to measure time weighted average (TWA) concentrations of a range pollutants (including non-polar organic, polar organic and metals) in aquatic environments (Kingston et al., 2000). The sampler is based on the diffusion of * Corresponding author. Tel.: +421 259343466. E-mail address: branovrana@gmail.com (B. Vrana). compounds through a membrane and their subsequent accumulation in a sorbent receiving phase. The prototype designed to sample non-polar organic compounds (log octanol/water partition coefficient (logAT0w) greater than four) has a Clg Empore® disk saturated with «-octanol as the receiving phase and this is overlaid with a low density polyethylene (LDPE) membrane (Vrana et al., 2005b). The sampler has been calibrated for the measurement of TWA concentrations of hydrophobic pollutants in water (Vrana et al., 2006). In the calibration experiments the effect of physicochemical properties (e.g. compound hydrophobicity), water temperature and hydrodynamics on kinetic and thermodynamic parameters characterising the exchange of analytes between the sampler 0269-7491/$ - see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2007.09.011 T. Lobpreis et dl. I Environmental Pollution 153 (2008) 706—710 707 and water were investigated. It was found that the rate of uptake of test analytes from water to the sampler receiving phase is related to the rate at which they offload to the water. This enables the use of off-loading rates of performance reference compounds (PRCs) preloaded on to the receiving phase to be used to adjust uptake rates for the effects of temperature and hydrodynamic conditions in the field. The calibration procedures and data have been reported (Vrana et al., 2006, 2007). The rate of diffusion from the bulk water to the receiving phase is proportional to the surface area over which diffusion takes place and inversely proportional to the diffusion path length. Therefore, the physical dimensions of the sampler body will significantly affect the sampling rate for different analytes. The body of the Chemcatcher™ was optimised in terms of both materials of construction and geometry. PTFE was selected for the sampler body as it has a low sorption capacity for most environmental pollutants (Kingston et al., 2000; Vrana et al., 2005b, 2006, 2007). The housing was constructed to fit a 47 mm Empore® disk receiving phase, having an active sampling area of 17.5 cm2. Uptake kinetics of many hydrophobic analytes have been shown to be controlled by diffusion in the aqueous boundary layer at the surface of the LDPE membrane (Vrana et al., 2006). The resistance to mass transfer of the boundary layer depends on hydrodynamic conditions in the vicinity of the membrane, and these are significantly affected by the sampler geometry. The membrane and receiving phase of the first generation Chemcatcher™ (old design) were located inside a 20 mm deep depression in the front of the sampler body. This well effectively buffers the effect of fluctuating flow on sampler performance. It effectively reduces convective transport of analytes to the sampler membrane, thus reducing sampling rates (i.e. the rate at which the sampler accumulates chemicals). The depth of cavity in the Chemcatcher™ body (new design) was reduced to 7 mm (Fig. 1) in order to increase sampling rates; this is particularly important for hydrophobic Fig. 1. Views of the old (left) and the new (right) designs of the Chemcatcher1 sampler body. chemicals that are present in only low dissolved concentrations in the aquatic environment. The aim of this study was to compare the performance of the old and new designs in monitoring hydrophobic organic pollutants and to determine whether calibration data obtained with the old design could be used for the new design. The uptake kinetics of polycyclic aromatic hydrocarbons (PAHs) to and release kinetics of PRCs from the new design were measured in a flow-through system under conditions identical to those used by Vrana et al. (2006) with the old design. 2. Theory Mass transfer of an analyte from water to the Chemcatcher™ sampler has been described (Vrana et al., 2006), and accumulation of a chemical in the receiving phase of the sampler from water can be described by: /wD(f) = mD0 + (CwA:dwVd - mm)[l - exp(-ket)] (1) where mD [kg] is the mass of analyte in the receiving phase, mD0 [kg] is the analyte mass in the receiving phase at the start of exposure, Cw [kg m~3] is the concentration in the water during the deployment period, KDW is the receiving phase/ water distribution coefficient, UD [m3] is the volume of the receiving phase, ke [s_1] is the exchange rate constant and t [s] equals time. The initial uptake phase is approximately linear or integrative. Here the amount of a chemical in the receiving phase is directly proportional to the product of the concentration in the surrounding water (Cw) and the exposure time (t). Eq. (1) can be rewritten as: mB(t)= mm + Cw/?s t (2) where Rs is the substance specific sampling rate (Lday-1), which can be determined experimentally. When PRCs are used and exchange kinetics are isotropic, Eq. (1) reduces to a single parameter equation: mB(t) = mDOexp( - ket) (3) where the amount of PRC added to the sampler (mD0) is known. Mass transfer is affected by the diffusion of analytes in the individual layers (i.e. aqueous boundary layer, diffusion limiting membrane and the receiving phase) and by their partitioning into the LDPE membrane and receiving phase. Compounds with log Kow > 4 are accumulated in the Chemcatcher™ under aqueous boundary layer control (Vrana et al., 2006), and their uptake kinetics is therefore sensitive to changes in the boundary layer thickness, and this depends on hydrodynamic conditions at the sampling surface. For compounds with log Kow > 4, the kinetic performance characteristics of the Chemcatcher™ are likely to be highly dependent on the geometry of the sampler body. The new design effectively decreases the thickness of the boundary layer and 708 T. Lobpreis et at. I Environmental Pollution 153 (2008) 706—710 hence results in faster mass transfer of such compounds that are accumulated under boundary layer control. 3. Materials and methods 3.1. Materials and chemicals Ci8 Empore® disks (47 mm diameter) were from Varian Inc., Walton-on-Thames, UK. LDPE membrane (40 um thick) was from Fisher Scientific, Loughborough, UK. The solvents (HPLC grade), acetone, ethyl acetate, methanol, n-hexane, n-octanol, n-nonane, 2,2,4-trimethylpentane and water were from Fisher Scientific. Certified (purity > 98%) reference standards of the test PAHs, internal standards, PRCs [perdeuterated PAHs: 2Hi0-biphenyl (D10-BIP), 2H10-fluorene (D10-FLU), 2H10-phenanthrene (D10-PHE), 2H10-acenaphthene (Di0-ACE), 2Hi0-pyrene (Di0-PYR) and 2Hi2-benz(a)anthra-cene (Di2-BaA)], and, certified external calibration solutions (10 ugmL~' in cyclohexane) were from Qmx Laboratories, Saffron Walden, UK. accumulated in the disk with that in the measurement of aqueous concentration in the calibration system. The latter represents the main source of uncertainty in Rs. 4.2. Offload of PRCs The offload rate of PRCs from the Clg Empore® disks was fitted by non-linear regression analysis using Eq. (3) with mDo and ke as adjustable parameters. Characteristic PRC offload curves are shown in Fig. 2. Satisfactory first order decay fits were obtained for D10-BIP, D10-ACE, D10-FLU and D10-PHE, but the rate of release of D10-PYR and D12-BaA from the disk was too slow to be measured reliably and the ke values were not significantly different from 0 (P > 0.05). 3.2. Passive sampler construction The passive sampler preparation has been described (Vrana et al., 2006): the sampler body houses a Ci8 Empore® disk receiving phase overlaid with a 40 |im thick LDPE membrane (47 mm diameter). n-Octanol (450 uL) is added to the interstitial space between the receiving phase and membrane. The new design sampler (Fig. 1) consists of three components (two body parts and a lid for storage and transport), which are clipped together. This makes the sampler cheaper and assembly and disassembly faster than with the old design, where screw threads were used. The new sampler is designed as a disposable device for a single field deployment, thus eliminating the need for cleaning and accompanying quality control measures required for trace analysis. The new design is made of moulded polycarbonate and can be recycled. 3.3. Calibration experiment The exposure conditions were identical with those used for the calibration of the old design (Vrana et al., 2006). Twelve passive samplers (new design) were exposed for up to 7 days in a flow-through exposure system with a constant analyte concentration (nominally set to 100 ngL~'), under controlled temperature (18 °C), water turbulence (carousel rotation speed 40 rpm). Samplers were removed from the exposure tank at regular time intervals, and PAHs and PRCs extracted from the receiving phases. PAHs from water samples taken regularly during the calibration, instrumental conditions and data processing were performed according to Vrana et al. (2006). 4. Results and discussion 4.3. Effect of sampler geometry on the analyte uptake In line with theoretical expectations, higher (up to a factor of 2.5) sampling rates (Rs) of PAHs were obtained with the new design. This applied to all of the test compounds except fluoranthene and pyrene, for which no significant difference in Rs between designs was observed. The latter can be attributed to the errors associated with the determination of Rs being greater than the effect investigated. 4.4. Effect of sampler geometry on the PRC elimination The effect of body design on the elimination rate of the PRCs under constant exposure conditions was measured as the first order elimination rate constant (ke) that is independent of concentration. The uncertainty associated with ke is much lower than that associated with the estimation of Rs since the former is based solely on the measurement of the amount of analyte remaining in the Empore® disks, and unlike the measurement of the latter does not involve the measurement of the concentration in the calibration water. For the PRCs (D10-BIP, D10-FLU, D10-ACE and D10-PHE) with measurable The effects of body geometry on compound specific sampling rate (Rs) of target PAHs and the offload rates (overall exchange rate constants; ke) of PRCs were determined by comparing the calibration data with those obtained for the old design. 4.1. Uptake of analytes Concentrations of the test PAHs in water (Cw) in the test tank remained constant over the exposure period. Satisfactory linear regression fits of Eq. (2) for the uptake analytes from water to the sampler receiving phase disks were obtained for all test compounds. The sampling rates (Rs) ranged from 0.23 Lday-1 for benzo(k)fluoranthene to 1.14 Lday-1 for pyrene. The error (expressed as standard deviation) of Rs combines the error in the measurement of the chemical 2.50 2.00 1.50 1.00 0.50 0.00 j □ old design ■ new design J" J" ^ ,e.N r 0.05) was observed between Rvalues for the two sampler body designs. The off-loading of PRCs is 1.5 (D10-PHE) to 2.3 (D10-BIP) times faster with the new sampler design as shown in PRC offload curves (Fig. 3) and calculated values of ke (Fig. 4). 0.0200 Fig. 4. Comparison of exchange rate constant ks values for old and new Chemcatcher™ sampler designs. Calibration experiments were conducted at water temperature of 18 °C and a carousel rotation speed of 40 rpm. 4.5. Calibration data for the new Chemcatcher™ design Exchange kinetics of hydrophobic organic pollutants (log AT0w > 4) between sampler and water were faster with the new than with the old design of the non-polar Chemcatcher™, and this is likely to be due to the shallower cavity in the new design reducing the thickness of the aqueous boundary layer at the diffusion limiting membrane surface. In order to avoid a lot of extra work it is important to determine whether the extensive calibration data set for the old design can be used with the new design. Previously Vrana et al. (2006) demonstrated a strong correlation between uptake and offload kinetic parameters for non-polar analytes and their deuterated analogues over a wide range of temperatures and flow rates: (4) The correlations between Rs and ke for phenanthrene and fluorene are shown in Fig. 5. The data are based on nine flow-through experiments performed with the old design under various exposure conditions (Vrana et al., 2006) together with calibration data obtained with the new design under one set of exposure conditions in this study. The linear regression lines for the old sampler can be extrapolated to fit the observed 710 T. Lobpreis et dl. I Environmental Pollution 153 (2008) 706—710 K [d"1l Fig. 5. Correlation between sampling rates Rs of fluorene and phenanthrene and offload rate constants ks of their perdeuterated analogues (PRCs). The data represents nine flow-through experiments at various exposure conditions, performed with the old design of sampler body (Vrana et al., 2006) and one exposure experiment with the new sampler body. The lines demonstrate that the good correlation derived for the old design can be extrapolated for the new design. data for the new sampler design. Thus, Rs values for new sampler design can be accurately extrapolated from the Rs =f(ke) curve obtained using the old design. 5. Conclusions This study shows that PRCs can compensate uptake rates of non-polar pollutants for changes in local hydrodynamic conditions at the surface of the diffusion membrane caused by modification of the sampler body geometry. Since these compounds (log KQw > 4) are accumulated under aqueous boundary layer control, sampling rates are increased, and hence sensitivity is improved, by reducing the thickness of the stagnant layer associated with the cavity of the sampler. The sampler accumulates chemicals under aqueous boundary layer control, and the thickness of this will fluctuate with water turbulence. In the new design the thickness of the boundary layer will be smaller and hence both the sensitivity and the effect of turbulence will be proportionately greater. A balance has to be made between sensitivity and reducing the impact of turbulence on sampling rates. It is unlikely that a flow-insensitive passive sampler can be developed that has sufficiently high sampling rates for use in all environments (Booij et al., 2007). Nevertheless, where PRCs are used the need for extensive, time-consuming re-calibration is avoided when sampler body geometry is altered. This result has general consequences for all samplers (such as semipermeable membrane devices [SPMD] and membrane enclosed sorptive coating [MESCO]) used for non-polar organic pollutants where sampling rates are under boundary layer control (Huckins et al., 1993; Vrana et al., 2001). However, extension of this approach to samplers for polar organic compounds has proved problematic (Alvarez et al., 2007). Acknowledgement We acknowledge the financial support of the European Commission (Contract EVKl-CT-2002-00119; www.port.ac. uk/stamps) and Finance South East (SEEDA), UK for this work. We thank Arne Holmberg (Alcontrol, Sweden) and Miro Vrana for providing the technical drawing of the Chemcatcher™ prototype (Fig. 1). References Alvarez, D.A., Huckins, J.N., Petty, J.D., Jones-Lepp, T., Stuer-Lauridsen, F, Getting, D.T., Goddard, J.P., Gravell, A., 2007. Tool for monitoring hydro-philic contaminants in water: Polar Organic Chemical Integrative Sampler (POCIS). In: Greenwood, R., Mills, G, Vrana, B. (Eds.), Passive Sampling Techniques in Environmental Monitoring. Comprehensive Analytical Chemistry, vol. 48. Elsevier, Amsterdam, pp. 171—197. Booij, K., Vrana, B., Huckins, J.N., 2007. Theory, modelling and calibration of passive samplers used in water monitoring. In: Greenwood, R., Mills, G, Vrana, B. (Eds.), Passive Sampling Techniques in Environmental Monitoring. Comprehensive Analytical Chemistry, vol. 48. Elsevier, Amsterdam, pp. 141-169. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environmental Science and Technology 27, 2489-2496. Kingston, J.K., Greenwood, R., Mills, G.A., Morrison, G.M., Persson, L.B., 2000. Development of a novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments. Journal of Environmental Monitoring 2, 487—495. Vrana, B., Popp, P., Paschke, A., Schuiirmann, G, 2001. Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water. Analytical Chemistry 73, 5191—5200. Vrana, B., Allan, I.J., Greenwood, R., Mills, G.A., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G, 2005a. Passive sampling techniques for monitoring pollutants in water. TrAC-Trends in Analytical Chemistry 24, 845-868. Vrana, B., Mills, G, Greenwood, R., Knutsson, J., Svensson, K., Morrison, G, 2005b. Performance optimisation of a passive sampler for monitoring hydrophobic organic pollutants in water. Journal of Environmental Monitoring 7, 612-620. Vrana, B., Mills, G.A., Dominiak, E., Greenwood, R., 2006. Calibration of the Chemcatcher passive sampler for the monitoring of priority organic pollutants in water. Environmental Pollution 142, 333—343. Vrana, B., Mills, G.A., Kotterman, M., Leonards, P., Booij, K., Greenwood, R., 2007. Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water. Environmental Pollution 145, 895—904. Príloha 14 Greenwood R., Mills G. A., and Vrana B., Potential applications of passive sampling for monitoring non-polar industrial pollutants in the aqueous environment in support of REACH, J. Chromatogr. A, 2009,1216, 631-639. Journal of Chromatography A, 1216 (2009) 631-639 ELSEVIER Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Review Potential applications of passive sampling for monitoring non-polar industrial pollutants in the aqueous environment in support of REACH Richard Greenwood3*, Graham A. Mills b, Branislav Vranac a School of Biological Sciences, University of Portsmouth, King Henry I Street, Portsmouth P01 2DY, UK b School of Pharmacy and Biomedical Sciences, University of Portsmouth, White Swan Road, Portsmouth P01 2DT, UK c National Water Reference Laboratory for Slovakia, Water Research Institute, Nabr. arm. gen. L. Svobodu 5, 812 49 Bratislava, Slovakia article info Article history: Available online 1 October 2008 Keywords: REACH Passive samplers Non-polar organics Aquatic environment Monitoring Biomonitoring abstract Possible roles of passive sampling within the context of the European REACH legislation are discussed. Passive samplers can provide information on environmental concentrations, fate and behaviour of substances of concern. They can potentially replace biota in the assessment of bioavailability, having advantages including lower cost and variability, and greater repeatability and acceptability on ethical grounds. Where remedial actions (e.g., product withdrawal, replacement or redesign) may be required, wrong decisions are potentially very costly. Against this background it may be possible to develop strategies based on passive sampling that will protect the environment from potential damage whilst minimising operational costs. © 2008 Elsevier B.V. All rights reserved. Contents 1. Introduction......................................................................................................................................... 631 2. Current monitoring practice........................................................................................................................ 632 3. Passive samplers.................................................................................................................................... 632 3.1. Calibration................................................................................................................................... 634 4. Applications of passive sampling in monitoring industrial chemicals............................................................................. 635 4.1. Measurement of time-weighted average concentrations of substances in water.......................................................... 635 4.2. Use of passive samplers to assess the potential for bioaccumulation...................................................................... 637 5. Conclusions......................................................................................................................................... 638 References.......................................................................................................................................... 638 1. Introduction The European Union (EU) has introduced new legislation (enacted in June 2007) that aims to manage all anthropogenic chemicals (manufactured in Europe or imported) that are used in significant quantities in order to protect human health and the environment. The legislation is called Registration, Evaluation, Authorisation and Restriction of Chemical Substances (REACH-Regulation (EC) No. 1907/2006) and replaces more than 40 existing European Directives and Regulations. Details are given in Corrigendum to Regulation (EC) No. 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Corresponding author. Tel.: +44 23 9284 2065; fax: +44 23 9284 2070. E-mail address: richard.greenwood@port.ac.uk (R. Greenwood). Registration, Evaluation, Authorisation and Restriction of Chemicals [1]. A European Chemicals Agency has been established in Helsinki (Finland) to deal with the routine management of this legislation. REACH has removed the distinction between existing and new substances that was present in the old legislation, and has thus increased dramatically the number of chemicals that require registration. A further marked change is that the burden of proof that chemicals placed on the market are safe has been shifted from the regulatory authorities to the applicant for registration. There is a great incentive to register compounds since unregistered substances cannot be manufactured or placed on the European market, and it is expected that in the region of 180,000 substances will be pre-registered during 2008. Whilst the legislation will be phased in by tonnage to spread the burden, the process will continue, and by 2018 it will be necessary to register all compounds produced in quantities of 1 tonne or more per year. Some substances 0021-9673/$ - see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.09.091 f* f* ľ"\\# R. Greenwood et at /J. Chromatogr. A 1216 (2009) 631-639 632 (e.g., pharmaceuticals) that are covered by other legislation are exempt from REACH. A tighter time scale and higher priority apply to substances (e.g., carcinogens, persistent and bioaccumulative substances, and potential endocrine disruptors) that are recognised as hazardous. Registration is compulsory for compounds that are released into the environment during use. This legislation will require risk assessments of this large range of substances based on predicted environmental exposure. In some cases, especially for potentially hazardous chemicals, and those released into the environment during use, environmental monitoring will be required to provide the evidence to support registration. This may involve measurements in a range of compartments including water, sediment, suspended solids, and air for volatile substances. In order to measure environmental fate it will be necessary to use robust and representative monitoring data. The cost of obtaining this information is potentially high, and any methods that can reduce this will be helpful to a wide range of industries. A variety of approaches to monitoring has been developed to replace the current regulatory practice of intermittent grab sampling combined with classical laboratory analysis in order to reduce costs and increase representativeness and reliability of the data obtained [2 ]. This paper will not attempt a comprehensive review of passive sampling since in recent years there have been several exhaustive reviews of the available literature [3-7]. Instead this paper will identify areas where passive sampling could provide reliable information to support applications for registration under the REACH legislation in a cost effective manner, and focus on these after a brief overview of passive sampling technology. 2. Current monitoring practice Current regulatory monitoring practice has been used for many years, and has become accepted for legislative purposes because of the significant developments and improvements in laboratory-based analytical chemical methods over the last two decades. These have reduced levels of detection for a wide range of analytes, and provide increased confidence in measurements made. This has been helped by the introduction of quality control protocols and associated quality assurance procedures that are underpinned by the provision of good quality reference materials and inter-laboratory trials. Until recently relatively little attention has been paid to the sampling step that precedes the laboratory analysis [8]. Sampling is particularly problematic for many of the industrial chemicals of concern, many of which are classified as priority pollutants in the aquatic environment. This is particularly marked for compounds which are present at only trace levels, and for highly lipophillic compounds of which only a small fraction is truly dissolved, and most is bound to either dissolved or suspended organic matter. For many years sampling of the aquatic environment has relied on the collection of spot (bottle or grab) samples that are transported to a laboratory for qualitative and/or quantitative analysis. Although this apparently simple procedure is commonly used to underpin legislation, there are problems associated with it, and significant errors can arise, particularly where pollutants are present at only low levels. For some analytes it is necessary to take steps (e.g., the addition of a preservative such as a biocide, or for metals an acid) to ensure the integrity of the sample during transport and storage [9]. Even so the sample can become modified by processes such as adsorption to the walls of the sample container, volatilisation, and either chemical and/or microbial degradation. A further drawback of spot sampling is that it provides information on water quality only at the instant that the sample is taken. This may not be representative of average water quality, especially where concentrations of pollutants fluctuate in time due to factors such as run-off associated with seasonal application of pesticides, sporadic industrial discharges and rain events [ 10]. In order to overcome this latter problem, methods such as automated sampling equipment that collect samples at regular intervals to give a more representative sample of the water body over periods of time from hours to days have been used (e.g., composite sampling devices and on-line analytical systems). However, in these systems there is a large potential for contamination from, and adsorption to components such as sampling tubes, valves and pumps. For compounds present at only trace levels these losses can represent a significant proportion of the chemical originally present and this can introduce large uncertainties where either spot or automated sampling methods are used; and these will reduce confidence in any subsequent modelling or risk assessment procedures. The sampling stage is even more problematic for other environmental compartments such as sediment, suspended material, and sludge, despite recent improvements in the extraction methods available for these difficult matrices. Interpretation of the biological relevance of the levels of pollutants (particularly non-polar compounds) measured by current sampling, sample preparation, and analytical procedures is difficult. This is particularly important for substances identified as potentially persistent and bioaccumulative. In an attempt to obtain more toxicologically relevant information, living organisms, typically caged fish or caged sessile species such as bivalve molluscs, have been used as monitors. Organisms are deployed over extended periods and changes in the levels of pollutants of interest are measured in body tissues at the beginning and end of the trial. This approach can give an estimate of the average environmental concentrations of pollutants over the deployment period (up to several months). This bioaccumulation gives a qualitative indication of levels of pollutants and can be used in a comparative way between sites and between times to measure spatial and temporal variation, respectively [11 ]. However, this method has some limitations. It is not possible to expose organisms in harsh environments such as in some industrial and domestic discharges where concentrations of pollutants exceed toxic levels. A further difficulty is that the test species, even when taken from apparently uncontaminated sites, may contain measurable levels of some pollutants before deployment and it is necessary to depurate before use, and take large, representative control samples at the start of the monitoring campaign. The analysis of tissue samples from biota is expensive and time consuming because of the complex sample preparation step that is necessary. It is not possible in many cases to assume that because a particular chemical is not bioaccumulated it is not present in the water column. Some pollutants are eliminated by the test animals, and this can occur at rates ranging from negligible to matching or exceeding the uptake rates. Passive samplers have been developed to overcome some of the shortcomings of both spot sampling and biomonitoring procedures. Some forms of these devices have been designed to mimic the uptake of pollutants by living organisms, and as such may be particularly useful in preparing REACH registration applications for some classes of potentially bioaccumulative or toxic compounds. 3. Passive samplers Passive samplers have been used in environmental monitoring since the beginning of the 1970s. The early designs were used to measure concentrations of gaseous pollutants in air [12], and this technology is now widely used in monitoring ambient air quality and workplace exposures to potentially harmful compounds such as volatile organic solvents. These air samplers are now commercially available, and standards and official methods (e.g., ASTM, EPA, R. Greenwood et ai /J. Chromatogr. A 1216 (2009) 631-639 4 Diffusion Path -* Bulk phase of aquatic environment Water boundary Diffusion Receiving phase layer membrane Fig. 1. Schematic diagram showing the components of passive samplers that in kinetic mode maintain a low concentration at the surface of the receiving phase so that the rate of diffusion of substances across the water boundary layer and/or diffusion-limiting membrane is proportional to the concentration in the bulk water phase. NIOSH, CEN and ISO protocols) have been developed for use with these devices. A number of global networks of passive samplers has been established to map the movement of persistent anthropogenic organic pollutants across the world. More recently passive samplers have been developed for monitoring concentrations of pollutants in water, soils and sediments. However, this technology has not gained similar acceptance within the water regulatory context. Several designs of device are available either as experimental prototypes or as commercial products [4]. The same principles of operation apply to all passive sampler devices, both for use in air and water. Uptake of a chemical from the environment is by passive diffusion. The samplers comprise a receiving phase that accumulates contaminants, and has a very high affinity for them so that the concentration at its surface is maintained close to zero, and a diffusion-limiting layer that separates the receiving phase from the bulk water environment (Fig. I). Hence the mass of a contaminant accumulated is determined by its concentration in the water, the length of exposure, and the sampling rate (Rs) of the sampler. The latter is determined by a number of factors including the area of sampler available for diffusion, the properties of the diffusion-limiting layer (e.g., thickness and resistivity), and the properties (e.g., size and polarity) of the chemical. Rs can be interpreted as the apparent volume of water cleared of pollutant per unit of time. For kinetic samplers, operating in the linear uptake mode, far from the thermodynamic equilibrium between sampler and water, Rs is independent of the concentration of the pollutant in the water. For long exposure times that exceed the linear uptake phase the extracted volume is constrained by the sorption capacity of the passive sampler. When thermodynamic equilibrium between sampler and water is approached, sampling is no longer integrative, and the accumulated amount of analyte no longer reflects the time-weighted average concentration. Most passive samplers measure only concentrations of freely dissolved analytes and not the total amount of analyte present in the water column. Fractions that are bound to suspended particulate matter or to dissolve organic carbon (DOC) are not measured due to either their exclusion by the diffusion-limiting layer, or poor uptake by the receiving phase. In all passive samplers the mass accumulated is used to determine the external concentration, but depending on sampler design and mode of operation this can reflect either the equilibrium concentration or the time-weighted average (TWA) concentration over the deployment period (days to months). Where environmental concentrations fluctuate in time then the kinetic samplers are used, 633 but in more constant or slowly changing conditions the equilibrium samplers are deployed. Since the samplers accumulate substances over a prolonged period the analytes are effectively preconcen-trated, and this can bring them above the level of detection of the analytical method. It would be necessary to collect and extract large volumes of water in order to achieve a comparable sensitivity with spot sampling. Passive samplers have been developed for monitoring environmental pollutants from a range of chemical classes including metals, polar organics, non-polar organics, organo-metallics, and volatile organics [7]. Samplers have been used in both equilibrium and kinetic modes for some of these classes. Equilibrium samplers have been mostly used to measure concentrations of pollutants in ground water and in sediment pore water [13,14]. A number of designs is available, and one has been used to monitor volatile organic compounds in ground water [15]. A much wider range of kinetic samplers (Fig. 2) is available, and these have been used for all chemical classes of pollutant [16]. For metals two main designs of kinetic samplers are available for the measurement of TWA concentrations of the labile fraction of metals. In the diffusive gradients in thin films (DGT) sampler a thin hydrogel layer forms the diffusion-limiting membrane, and this overlays a chelating agent receiving phase. The Chemcatcher® (metals version) uses a similar receiving phase (in this case in the form of a commercially available Empore™ disk) and the diffusion-limiting layer is provided by a cellulose acetate membrane [ 17 ]. The DGT has an established record in the monitoring of metals such as cadmium, chromium, copper, lead, and zinc in a wide range of aquatic environments [18,19]. Information on the relative concentrations of labile and bound species of metal present in the water can be obtained by simultaneous deployment of DGTs with hydrogels of different porosities. This is important since some species of metals are far more toxic than others. The Chemcatcher® (polar organic version) and polar organic integrative sampler (POCIS) are designed to monitor concentrations of polar (logKow<4) organic pollutants [20,21]. In both samplers the diffusion-limiting membrane is a polyethersulphone sheet with water-filled micropores, and the receiving phases comprise a range of adsorbent materials, either bound in an Empore™ disk, or in a free particulate form. These have been used for measuring the TWA concentrations of a range of polar herbicides, pharmaceuticals, and personal care products, and are described more fully in another paper in this issue [22]. A much wider range of passive samplers is available for monitoring non-polar organic substances (34) that are dissolved in water at only trace levels (low ng L_1 to pg L_1). Large volumes of water would need to be processed in order to measure these compounds using bottle samples linked to conventional analytical methods. The applications have been extended to cover new and emerging compounds of concern including organo-metals used in wood preservation and in antifouling preparations (e.g., tributyl tin) [52,53]; polycyclic musk xylene, musk ketone used in domestic products [54], and polychlorinated naphthalenes [55]. Passive sampling has the potential to contribute the REACH registration process in a number of ways. A decision on the approval of a registration of a substance under the REACH regulations will 635 be based on a number of factors including volume of use, predicted environmental concentration (PEC), toxicological properties, exposure of aquatic organisms and bioavailability. For existing and new compounds initial assessments will be based, where possible, on a modelling approach. More work will be needed to evaluate substances that are lipophillic and stable, and that will potentially be released into the environment, either directly or indirectly. Compounds falling into this category will include some substances that are components of personal care and household products that will enter the environment via the waste water system. In order to estimate any environmental risk associated with individual substances, it is necessary to estimate the movement of compounds of interest to the aquatic environment. This must take into account the various possible sources, and in particular inputs from waste water effluent. For existing substances of concern it may be necessary to monitor concentrations of compounds in domestic effluents and surface waters in order to obtain data to estimate the degree of environmental risk. This is not straightforward where inputs into sewage treatment plants (STPs) and effluents from them fluctuate widely over a diurnal period, and vary between seasons. In addition, they can be markedly affected by sporadic weather events. Planning a monitoring programme is further complicated by a lack of spatial homogeneity following a discharge of effluent to a river. It is not uncommon for a plume of effluent to remain close to one bank of a river for many kilometres, and mixing is rarely instantaneous [56]. This situation can be even more complicated in tidal waters. There is a need for mapping the distribution of effluent in mixing/dilution zones in order to obtain a representative picture of dispersion and dilution of the substances of interest. Whilst monitoring in perceived 'hotspots' can provide a worst case scenario, it may be very misleading, and this will not provide representative information on average and/or maximum values of environmental concentrations. Such maximum environmental concentrations (MECs) may skew modelling and lead to unrealistic risk assessments. Appropriate sampling frequency, sampling period and pattern are prerequisites for representative sampling to be achieved. Castiglioni et al. [57] found a difference of a factor of two between maximum day time and minimum night time influent loads in an STP. Such a variation may introduce bias when using time-proportional sampling methods, in this case with an estimated underestimation of influent load of 5-15%. In order to obtain representative information that will give the necessary level of confidence in a risk assessment it would be necessary to use a high frequency of spot sampling, or flow-weighted composite sampling. This would be very expensive, particularly where there was marked local spatial variation. Costs could be reduced by using passive samplers deployed over a period of weeks at a range of sites to provide TWA concentrations of the substances of interest. An example of where the utility of passive sampling has been demonstrated is provided by the monitoring of polybrominated diphenyl ethers (PBDE) that are used as flame retardants in a wide range of goods and products for use in the home. These compounds are extremely hydrophobic (logKow from 4 up to as high as 10), and are present in surface waters at sub-ppb levels; but are of concern because they are very persistent and have been shown to bioaccumulate, and have been included in regulatory monitoring programmes in water and sediments. One particularly interesting monitoring campaign for the substances that illustrates the potential utility of passive sampling in this context was that of Booij et al. in the Scheldt estuary and along the North Sea coast of the Netherlands [58]. Using SPMDs this group was able to measure a series of PBDE congeners present at very low concentrations (0.1-5pgL_1). However, this is because of the large factor of pre-concentration exhibited by these devices. There are problems when dealing with the extremely hydrophobic congeners in this series 636 R. Greenwood et al. /]. Chromatogr. A 1216 (2009) 631-639 Table 1 Time-weighted average (TWA) water concentrations for a 14-day period estimated from the levels in the Chemcatcher® and those measured usingfiltered spot samples at the sampling site in the River Meuse. Compound log/Cow TWA concentration (ng L_1) Passive sampler3 Filtered spot samples'1 Fluorene 4.2 7.5 (±1.2) 1.6 (±0.1) Phenanthrene 4.5 10.2 (±2.0) 8.4 (±3.5) Pyrene 5.1 9.6 (±1.8) 22.9 (±10.8) Fluoranthene 5.1 10.5 (±1.8) 11.7 (±4.1) Chrysene 5.7 3.7 (±0.3) 8.7 (±0.8) a The TWA concentration was calculated as arithmetic average of the three estimates calculated from analyte amounts found in replicate samplers. The uncertainty level of this estimate was expressed as the standard error of the mean (in parentheses). b The arithmetic average of the six measurements of spot samples at regular intervals during 14 days of sampler exposure was taken as the best estimate of the TWA concentration. The uncertainty level on this estimate was expressed as the standard error of the mean (in parentheses). since as discussed above calibration procedures can give underestimates of uptake rates because of the tendency of these compounds to associate with DOC, suspended solids, and components of the calibration rig. Another application that demonstrates the utility of passive sampling is provided by the field trials carried out by Vrana et al. in a stream (the Spittelwasser) that flows through a highly polluted industrial area (Bitterfeld in Saxony-Anhalt, Germany) using the MESCO device [30]. They measured TWA concentrations of FAHs, PCBs, and some cyclodienes using 20-day exposures of the samplers alongside grab samples taken at the beginning and end of the deployment period. Some compounds (hexachloroben-zene, acenapthenene, fluorene, benzo[a]anthracene and chrysene) were measured quantitatively in the MESCO samplers but were not recovered from grab samples. Others (~y-hexachlorohexane, anthracene, phenanthrene, fluoranthene and pyrene) were measured in both samplers and grab samples, and the estimates of concentrations differed by up to a factor of two between the two methods. These differences could have been due to fluctuations in the concentrations in the period between grab sampling events. The variability between duplicate MESCO samplers was small (relative percentage difference in the range 6-15%). A field trial in the River Meuse (at Eijsden in the Netherlands) using the Chemcatcher® passive sampler to monitor PAHs and using the PRC approach demonstrated the utility of the samplers to provide measures of TWA concentrations where there are fluctuations in concentration with time, and enabled a comparison of spot and passive sampling for compounds covering a limited range of polarity [59]. In this trial six equally spaced spot samples were taken over a 14-day deployment period, and the average concentrations of five PAHs (chrysene, fluoranthene, fluorene, phenanthrene and pyrene) that were quantifiable in both spot samples and passive samplers were calculated. The estimates of TWA concentrations based on passive sampling and spot sampling are presented in Table 1. For fluoranthene and phenanthrene there was a reasonable agreement (<20% difference) between the estimates of concentration derived from spot and passive sampling. However, for chrysene and pyrene the passive sample-based estimates were markedly less than the estimates based on spot sampling. The concentration of fluorene estimated by passive sampling was higher than that measured in spot samples. There was a marked increase (approximately a factor of 10) in the concentration of phenanthrene measured in spot samples during the first week of the trial. However, the TWA concentrations estimated by the two sampling methods were very similar. This indicates that most of the phenanthrene in the filtered fraction was in the freely dissolved form, and demonstrates the Time [d] Fig. 3. Concentrations of phenanthrene in six spot samples (filtered at 0.45 p,m) taken at regular intervals during a 14-day deployment of three replicate Chemcatcher® samplers in the River Meuse (Netherlands). The water temperature varied between 18 and 21 °C, and the samplers were maintained at a depth of 1 m. The solid line represents the time-weighted average (TWA) based on the spot samples, and the dashed line the TWA estimated from the passive samplers. potential for passive sampling to provide TWA concentrations even where the concentration in the water fluctuates in time (Fig. 3). It is, however, difficult to compare estimates of concentrations of substances obtained by spot and passive sampling. The fractions of contaminants measured by the two methods can be different, especially in a regulatory context where for organics unfiltered spot samples are used for analysis. Any of the substance bound to particulate material or DOC will not be available for uptake by passive sampling. Even, as in this comparative study, where spot samples were filtered through 0.45 u,m filters prior to analysis, there can still be differences between the data from the two methods since some of the compound can be bound to DOC. The concentration of the truly dissolved fraction of hydrophobic analytes in water will depend on the level and quality of the DOC present, and this may change with time. It may change rapidly in, for example, a weather event such as heavy rainfall followed by run-off from surrounding land. The impact of the concentration of DOC on the freely dissolved fraction of organic compounds available for uptake by passive samplers will increase with increasing hydrophobicity of the pollutant. A further factor that can lead to apparent differences between the concentrations estimated by the two methods is the occurrence of fluctuations in concentration not detected by the frequency of spot sampling used. It will be particularly difficult to compare the values from the two methods where spot sampling is infrequent, for instance where spot samples are taken, for convenience, only at the beginning and end of the deployment period. This needs to be kept in mind when interpreting data from the monthly samples used in regulatory compliance monitoring under the WFD. The applications outlined above illustrate the potential utility of passive sampling for providing robust and representative information on concentrations of freely dissolved (biologically significant) fractions of non-polar chemicals in various divisions of the aquatic environment. This could be particularly useful within the context of the registration of existing substances of concern within the remit of the REACH legislation. These methods can provide representative data to underpin studies of the environmental fate and distribution of substances in waste, surface and ground waters, even where inputs fluctuate diurnally, seasonally or sporadically. In most cases the use of passive sampling would be less expensive than using high frequency spot sampling to obtain this information. R. Greenwood et at /]. Chromatogr. A 1216 (2009) 631-639 4.2. Use of passive samplers to assess the potential for bioaccumulation It is important to assess the potential for bioaccumulation for substances identified as potentially persistent and bioaccumula-tive. Passive samplers could provide an alternative to the use of biota in this context. The information from passive sampling could more representative of general bioavailability particularly where the substance of interest is metabolised by the test species. Since there is marked inter-specific variation in metabolic capability, the information from a single test species (usually selected on the basis of availability and ease of deployment) may not be representative of other organisms within the ecosystem under consideration. The potential of passive samplers for providing sound information to underpin registration packages under the REACH legislation is illustrated in a number of recent studies in which passive samplers were deployed alongside biomonitoring organisms. A detailed and rigorous study of the bioavailable fractions of PAHs, PCBs, and organo-chlorine pesticides at fresh water sites in the Amsterdam (Netherlands) area was carried out by Verweij etal. [60]. This group examined levels of compounds from these families of non-polar pollutants in the muscle tissue of caged carp, SPMDs, and sediments. The sampling sites covered a wide range from lakes that were thought to be relatively uncontaminated to sites that receive significant inputs from domestic effluent, dredging activities, and major organic chemical plants (including a coal tar refinery plant). SPMDs (loaded with PRCs) were deployed alongside caged carp at each site, and sediment samples were taken during the exposure period. Since for highly hydrophobic substances the concentrations in the water are generally very low, and typically below the level of detection of standard analytical methods, the concentrations were estimated on the basis of their SPMD-water partition coefficients. For some of the groups of pollutants investigated in this study the control levels in unexposed fish were relatively high, and the variation between individuals was large. This reduced their utility as indicators of exposure, though tissue levels of pollutants were elevated in fish exposed at heavily polluted sites compared with those in animals deployed at cleaner sites. In contrast the variation between individual SPMDs at each site was small ( □_ O cN ■a S 6 < Ll_ QQ E cd cd cn CD — O cd c cd CD 0_ □ 0_ □ 0_ □ O C" O CO i- ™ o ■n CM a> CJ UjCjiO ? oD0 a. -o u CM LU ^ QQ ^ O O LU (J CO 00 ^ o>- CJ : cj ■o T= & lu o <- X a QQ a- -£(J -?§1 _f P- LU _j QQ CO i- < ■Q "a ? S To Uj Q. — Q am suj "9 W -O ^ DC S= r-.

% cd 00 cd C cm °a r £ <-> -5 T °- §"£ o cd > H- O -p <-> ili ~ (J < o „ >■ c TJ c n Si » B..9-S in 5 £i i ^QQ^ 2 cdJ CJ w dc 0 cd Q_ CO 5 3 C cd O O C c cd t CD CJ Cfi 4—1 CO ^ CO i- P a.-B M- ■£ S O C CO £s 8 S2-S o cd S. J= < Dt « cd 3 CD ~ O j> CD cd t3 cn to "S > ^- co !- 6 cd " C3 -° > 2! §u » ° oi Q.11 C C cd cd c (- CD CD 4- tn c O O o s ™£ = .c LL 5 o cd cd dc 5: co Q_ „ « aJ "D ■ c o 0> g ^ N CD Tl ~ CD g N .. ™ - < CD CD . . cd E jd mop m '—" Q. 0} O cn cd ^ .E CD CD > " n CD CD cn CD 3 F 0 E °- £ E t o CD C3 c CD o cd a) ^ o) Q-a cd n: ■o CL 5384 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009 o ^ water; elution with n-pentane) and analyzed by GC-MS for PAHs. An electron capture detector was used for the detection and quantification of PCBs, hexachlorobenzene and p,jf-DDE. MESCO I (m) was prepared by inserting a precleaned silicone rod (I cm long; 2 mm diameter, Goodfellow Ltd., UK) into a dialysis membrane bag (18 mm flat width and 30 mm long) made from regenerated cellulose (Spectra/Por 6, molecular weight cutoff 600 Da) filled with Milli-Q water (10). Diffusion-limiting envelopes of MESCO II were composed of air-filled nonporous LDPE membrane (purchased from Polymer-Synthesewerk, Rheinberg, Germany) containing a 1.5 cm long silicone rod (from Goodfellow GmbH, Bad Nauheim, Germany) of 2 mm diameter as receiving phase spiked with PRCs (11). The bare silicone rods used were 8 cm long and 2 mm diameter. Following sampler retrieval, silicone rods were removed from the MESCO membranes and stored in glass vials at —20 °C until analysis. Bare silicone rods were quickly washed under tap water and dried with tissue paper before storage at —20 °C. The combined processing and analysis of silicone rods (1.5 cm long from the MESCOs and 1 cm pieces cut from silicone rods) consisted of a thermal desorption step followed by GC-MS analysis. A thermo-desorption unit (TDU) from Gerstel (Mülheim a.R., Germany) was placed on top of an Agilent 6890 GC (Agilent Technologies, Palo Alto, CA) equipped with a cold injection system CIS-4 (Gerstel) and a mass spectrometric detector [MSD) 5973N (Agilent). Full details of the analysis can be found elsewhere (11). In all cases, quality assurance procedures such as the use of internal standards for the extraction and analytical steps and the assessment of analyte recoveries were conducted. Sampler Deployment and Retrieval. All prepared samplers were stored at —20 °C and the temperature maintained below 0—4 °C during transport to and from the field site. Preparation and trip control samplers were prepared and transported in a similar way to exposed samplers and opened to the air during deployment and retrieval procedures. During deployment, controls were stored in closed containers at -20 °C. Samplers were mounted onto stainless steel cages, and moorings kept them 1 m below the surface of the water. In most cases, triplicate passive sampling devices of each type were exposed for a period of 7 days, two consecutive 14 day periods (14 days (1) and 14 days (2), respectively), and an overlapping 28 day exposure (further details on replication in SI). In addition, silicone strips, were deployed for four consecutive 7 day exposures. The 7 day sampling period for Chemcatcher, SPMDs and LDPE membranes was not undertaken in cages and samplers were therefore exposed to higher water turbulences. Results and Discussion Adequacy of the Performance Reference Compound Approach. The measurement of PRC dissipation provides information on contaminant exchange kinetics between water and the sampler and allows the estimation of Rs values in situ (6). Analytes for which the concentration in the sampler approaches equilibrium with the concentration in the water are characterized by significant or even complete elimination of PRC with similar log Kqw- However, negligible or little PRC dissipation is indicative of rates in the linear phase of uptake. The threshold between these two regimes is generally found for PRCs with log Kqw of 4.5—5 for exposure periods of several weeks (13,14). In addition, using multiple PRCs with a range of log Kqw makes it possible to establish when kinetics of uptake into the sampler are membrane- or boundary layer-controlled. The overall resistance to mass transfer (II ko) into the samplers can be expressed as the sum of the water (<5W/ Dw) and membrane-side (<5m/XmwDm) resistances: kn (1) with Kmw the membrane—water partition coefficient, <5W and <5M the boundary and membrane layer thicknesses (m), and Dw and Dm (m2 s-1) analyte diffusion coefficients in water and the membrane, respectively. Amounts of analytes absorbed by the samplers follow a first-order approach to equilibrium: N-- exp(-fcefj] (2) where AT is the amount of analyte absorbed (ng), ifSw the sampler—water partition coefficient (L L-1), Vthe volume of the sampler (L), fce the exchange rate constant (h-1), t the exposure time (h), and Ctwa is in ng L-1. PRC dissipation also follows first-order kinetics: (3) where AT0,prc and ArPRC are PRC masses in the samplers prior to and following exposure, respectively and where k. is given by k0A (4) where fc0 is the overall mass transfer coefficient (see eq 1), A the surface area of the sampler (m2), Vthe volume of the sampler (L) and Rs the analyte uptake rates (L d-1). PRC elimination rates, k.} were calculated for the various exposures and samplers and their statistical significance tested using a procedure described previously (15). Overall, it was possible to use most PRC data; however, data were not used when release was either close to 100% or insignificant, or when amounts remaining in trip controls were significantly lower than in fabrication controls (see SI for further details). Since configurations of the devices differ widely (Table 1) and k. is proportional to A/V(eq 4), elimination rates were normalized to this ratio. The relationship between k-V/A values for 14 and 28 day exposures and log Kqw is presented in Figure 1A. The spread of the data across the range of samplers is less than one log unit and the apparent plateau for PRCs with log Kqw < 5 is indicative of membrane-controlled mass transfer (13). The overlap of Chemcatcher and SPMD (both using LDPE membrane material) data and generally higher k-V/A values for silicone strips and MESCO II for PRCs with log Kqw < 5 reflects higher diffusion coefficients in the silicone material compared with LDPE (16). Overall mass transfer coefficients (fc0) determined as the product of keVIA and Ksw (eq 4), were plotted as a function of log^w (Figure IB). ifswfornondeuteratedPRC analogues were used (see following section for a detailed list of references). The transition between membrane-controlled mass transfer, where ko increases with increasing PRC hydrophobicity, to boundary layer-control becomes more apparent with the bell-shaped relationship between log k0 and log Kqw (Figure IB). Under boundary layer-controlled mass transfer, Rs is expected to decrease with increasing hydrophobicity. Here, a decrease can be observed for silicone strips (phenanthrene-d10 and fluoranthene-d10) and LDPE membranes (fluoranthene-d10 and chrysene-d12). The transition between membrane- and water-side-control of mass transfer appears to occur for compounds with log Kqw between 4.5 and 5.0 (Figure IB) and confirms previously observed cutoff points (13,14). One would expect similar fc0 values for fluoranthene-d10 under boundary layer-controlled exchange for LDPE membranes and silicone strips since both types of samplers have a similar configuration VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY ■ 5385 o ^ + Chemcatcher A LDPE Membrane V MESCO II X Silicone strip 0 SPMD FIGURE 1. (A) First-order performance reference compound elimination rates, ke, normalized to the sampler surface to volume ratio {AIV) for five different passive samplers. (B) Mass transfer coefficients, k0, as the product of keV/A and sampler-water partition coefficients, /fSw- Data are for 14 (1st and 2nd successive exposures) and 28 day exposures. Lines are intended as a guide to the eye only. and were disposed randomly in the same cages during exposure. Accounting for an uncertainty of log KSw of around 0.3 log units, differences observed here are not likely to be significant (16). PRC mass transfer coefficients were regressed (using Minitab version 14) against log Kow for those under membrane layer-controlled kinetics and molecular weight (MW) for those under boundary layer-limited exchange (Table 2). The observation of similar slopes for log ko versus log Kow regressions for Chemcatcher and SPMDs is not unexpected since both samplers use an LDPE membrane. The steeper slopes observed for both samplers for the 7 day exposure under higher water turbulences indicate that resistance to mass transfer in the boundary layer is of a similar order of magnitude to that in the membrane for analytes with log Kow near 4.5. According to eq 1, ko is influenced by both Ksw and the analyte diffusion coefficient Dm for the membrane material when mass transfer is membrane-controlled. With slopes of log i(^w-log Kow relationships close to unity, observed log fcb-log Kow slopes of 0.7—0.95 as shown in Table 2 are plausible. Slopes for silicone strips, however, are significantly lower. It is likely that resistance to mass transfer in the boundary layer is not negligible and contributes to the overall resistance to mass transfer of PRCs with higher log Kow used in these regressions. This is important since with an accurate knowledge of KSw and Dm values, estimates of fcw (Dw/<5W) may be obtained from PRCs under "membrane-controlled uptake". PRC-based information on boundary layer-controlled uptake is available only for LDPE membranes and silicone strips (Table 2). A linear regression of log fc0 on log MW gave values in the range —3 to —8.9 which are over an order of magnitude higher than the slope of —0.35 predicted if the reduction in mass transfer coefficients was solely the result of a decrease in analyte diffusion coefficients in water with increasing molecular weight (14). These slopes are, however, similar to those from log fcb-log MW regressions obtained in sediment slurries (17). This sharp decrease in k0 values has been observed previously during sampler calibration experiments (13,15) and attributed to (i) the transfer in the membrane of the Chemcatcher or SPMD becoming rate-limiting again owing to an increasing difficulty for larger molecules to diffuse in the LDPE, or (ii) contaminant sorption to DOC that reduces the fraction available to the samplers and results in the underestimation of Rs values of large molecular weight PAHs for example. Since the present data is based on PRC elimination rather than analyte uptake, log fco-log MW relationships suggest that a significant reduction in analyte diffusion coefficients in the membrane materials tested here is possible and contributes the strong decrease in uptake rates for compounds with log Kow > 5. Calculation of TWA Concentrations. Concentrations of dissolved contaminants in the Meuse river water were calculated using the following equation (combination of equations 2 and 4): N eXP("^) (5) Further details of the calculation of CWa are available in the SI. Literature values for Ksw for each sampler are needed and exposure-specific flshave to be determined. Ksw values for the Chemcatcher, LDPE membranes, MESCO I (m), MESCO II, silicone strips, silicone rods and SPMDs were obtained from refs 15, 14, 18—21 (using the experimental design as described in ref 22), and 13, respectively. Rs for the PRCs were calculated from Rs = fce,pRc Ksw V. Since PRC-based Rs are for a limited log Kow range, models relating Rs to analyte properties were used to estimate Rs for compounds outside the PRC range. A full description of the calculation of Rs values is provided in the SI. Briefly, sampling rates of the PRCs were fitted to the empirical log fls-log Kow relationships reported for the Chemcatcher (15) and SPMDs (13). For all other samplers, these relationships are not available and sampler-specific methods were used. For silicone strips, the PRC-based linear relationship between fce,pRc and Ksw'1 was used to extrapolate the Rs value for analytes with log Kow < 4.6. For those above this threshold, boundary layer-controlled uptake was assumed and fls-pRc for fluoranthene-d10 was used to extrapolate uptake rates for the remaining compounds according to Rs ~ (Vm)~0 39 where Vm is the analyte molar volume at boiling point (13). For LDPE membranes, the empirical Kow-Rs model developed by Booij and co-workers (14) based on SPMD/LDPE membrane experimental calibration data was used to estimate Rs for all analytes (see SI). Offloading of fluoranthene-d10 and chrysene-d12 and literature data were used to estimate two empirical parameters 5W and Bm representative of mass transfer in the boundary and membrane layers, respectively. The product of the mass transfer coefficient obtained and the surface area of the sampler is Rs. For MESCO I (m), no PRC data was available. Instead mean values of laboratory-based Rs were used. These were corrected according to eq 4 to account for the use of a different receiving phase (with different Vand KSw) {10, 23). For MESCO II, the overall mass transfer coefficients were calculated from the sum of theoretical mass transfer coefficients for the various layers of the sampler as previously undertaken (19). Waterside mass transfer was adjusted using available PRC data. Finally, analyte Rs for silicone rods were also estimated from semiempirical mass transfer coefficients calculated 5386 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009 TABLE 2. Slopes of Linear Regressions of Log k0 on Log /fow and Log k0 on Log MW for Each of the Samplers and Exposure Period of 7, 14, and 28 Days membrane-controlled uptake ((Alog k0)/{A\og flow)). (SE)" boundary layer-controlled uptake ((Alog k0)/{A\og MW)), (SE)a exposure (days) 7 14(1) 14(2) 28 7 14(1) 14(2) 28 Chemcatcher 0.95" (0.07) 0.91(0.08) 0.79(0.06) 0.85(0.70) SPMD 0.93" (0.29) 0.71(0.77) 0.78(0.05) 0.78(0.77) LDPE -8.5(2.6) -5.6(2.0) -8.9(7.8) -4.8(0.5) silicone strip -° 0.18(0.04) 0.39(0.05) 0.47(0.03) -° -8.8(7.6) -8.5(2.4) -3.1(7.2) 3 SE = standard error of the slope. b Deployment outside the cage resulting in higher mass transfer for PRC with log K0w ~ 5. ° Insufficient replication available. for the membrane and boundary layer according to eq 1. A 10 fim boundary layer thickness based on ke for fluoranthene-d10 was adopted. Interestingly, this value is similar to that obtained for MESCO II. In order to evaluate the performance of the various samplers, we compared (i) masses of analytes absorbed [normalized to the respective sampler surface areas) for all analytes that were in the linear phase of uptake, (ii) calculated Ctwa, and (iii) the precision of these Ctwa estimates. To compare surface area-normalized amounts of analytes, we first calculated the average amounts for each analyte and each sampler for the 7 day exposure. This was repeated for the 14 and 28 day exposures. These values were then divided by the corresponding values obtained for the LDPE membrane samplers. LDPE membrane samplers were selected based on the fact that the largest number of analytes was detected with this sampler. The size of data sets used to create the box-plots (Figure 2A) is indicative both of the number of analytes in the linear phase of uptake for the various samplers and of the relative method quantification limits (MQLs) ofthe various methods. These showthatMQLs generally increase in the order LDPE membrane ~ silicone strip ~ SPMD < silicone rod ~ MESCO II < MESCO I (m) ~ Chemcatcher. Generally samplers with large surface areas such as LDPE membranes, silicone strips and SPMDs enabled the quantification of all target compounds. The very similar mean analyte masses accumulated in silicone strips and in LDPE membranes result from the analysis being conducted in the same laboratory and the samplers having almost identical sizes and similar mounting in deployment cages. The uncertainty in the normalized mean ratio for SPMDs combines that associated with the analysis being conducted in a different laboratory with those due to differences in turbulences around the samplers resulting from their larger dimensions. Similar factors influence the data obtained for the other samplers. Some variability can be observed for these samplers though the significantly smaller size of data sets is likely to affect these results. The particularly small data set for MESCO I (m) is the result of membrane rupture in exposures of over 14 days. To compare CTwa values, we first calculated the geometric mean of CTwa for each compound and each exposure taken over all seven samplers. Ratios of individual Ctwa estimates over the geometric mean were then calculated. Dissolved contaminant concentrations varied over 3 orders of magnitude with low molecular weight PAHs at the ng L-1 level down to PCBs found at concentrations of tens of pg L-1. Marked differences in Ctwa generated by the various samplers can be observed in Figure 2B. Ctwa estimated by LDPE membranes, MESCO II and SPMDs are closest to respective mean concentrations. Concentrations measured by the Chemcatcher appear generally higher than mean concentrations. This could be explained by a reduction in uptake rates (as shown by PRC elimination rates) with increasing exposure time. Data obtained with MESCO I (m) and the silicone rods consistently under predict mean concentrations and appear much lower than those generated by the Chemcatcher, LDPE membrane or silicone strips. This could be the result of possible bias induced by the method used to calculate TWA concentrations from analyte masses accumulated or uncertainty in the PRC data for the silicone rods. Since the uptake of many of the analytes detected and quantified by these two samplers had reached a significant degree of equilibrium, most of the variability in Ctwa may be linked to the variability of KSw values (16). An uncertainty (or bias) of 0.3 log units is not impossible and would result in error equivalent to a factor of 2 when calculating Ctwa for analytes close to equilibrium. It should be noted here that Figure 2B reflects the variability among samplers and among laboratories. Finally, to compare the precision of Ctwa values, Ctwa for each analyte and each sampler were log-transformed before calculating standard deviations. The antilog of these standard deviations can be interpreted as an uncertainty factor and provides a comparison of the overall precision of the different passive sampling methods used here (Figure 2C). The observed variability for all samplers was in the range 1.2—1.5. The smallest variability is generally exhibited by the Chemcatcher and LDPE membranes. The precision of analytical measurements decreases with decreasing analyte concentration. Concentrations of these analytes in passive sampler extracts are closest to analytical LODs where analytical precision is worst. In contrast with the less hydrophobic PAHs [close to equilibrium), the calculation of Ctwa for analytes in the linear phase of uptake relies significantly more on PRC elimination rates. Therefore, the precision of Ctwa for these compounds cumulates errors from more sources since it includes differences in the physical preparation of the samplers, in masses accumulated by the samplers, in the PRC elimination rates and finally in the extraction and analytical measurements (generally close to analytical LODs) conducted in two different laboratories. Interestingly, here the spread of the LDPE membrane data is much lower than for silicone strips and SPMDs. Effect of Sampler Exposure Time. Sampler exposure time has an impact not only on PRC dissipation but also on masses of analyte accumulated. Longer deployments generally result in the accumulation of higher masses of contaminants that are in the linear phase of uptake (i.e., far from equilibrium) facilitating their analytical measurement while bringing sampler concentrations of less hydrophobic ones closer to equilibrium with the water phase. However, membrane fouling by biofilm-forming microorganisms or accumulation of suspended matter on the sampler surface may affect the exchange of analytes and PRCs between water and passive sampler when samplers are exposed long enough for these phenomena to occur. Exposures of 14 and 28 days resulted in significant biofouling comprising a large proportion of sediment particles. This may be due to a combination of relatively small openings on the cages used for deployment and the "zigzag" mounting of samplers within the cages that facilitates sediment particles settlement inside the cage. VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY ■ 5387 o ^ oj t3 3.3 2.5' C 0 1 2.0 ■g "D C ro 55 o - 1.5 1.0 V, (C) 36 17 27 38 I 87 / / FIGURE 2. (A) Box-plots of sampler surface area-normalized amounts absorbed for analytes under boundary layer controlled uptake (N/A). These were normalized with respect to those for LDPE membrane samplers ((N/A)LDPE membrane) and calculated for each analyte and each exposure. (B) Ratios of time-weighted average concentrations (Ctwa) measured by the different samplers to the geometric mean concentration (from all sampler replicates) for each analyte and exposure time. (C) Box-plot of standard deviations of log-transformed sampler-specific Ctwa calculated for each analyte and exposure. Values on the box-plots represent the sample size on which the box-plot is based. Dots are 5/95 percentiles. Significantly less fouling of the samplers was observed for the 7 day exposure outside the cages. Our approach here was to compare masses of analytes accumulated and this was possible for the 28 day or consecutive 14 day deployments since sampler-specific exposure conditions were identical. Figure 4 shows the ratio of the mass of contaminant accumulated over 28 days to the sum of masses accumulated over the two successive 14 day exposures. In the case of compounds in the linear phase of uptake (generally with log K0w > 5) during these 28 days, a 1.2 1.0 S 0.8 i 0.6 0.4 0.2 0.0 Chemcatcher LDPE membrane MESCO II Silicone rod Silicone strip SPMD A O <> los Kow FIGURE 3. Ratio of analyte masses accumulated over 28 days to the sum of masses accumulated during the two successive 14 day exposures. Reference lines at y = 1 and 0.5 indicate ideally compounds for which uptake is linear over the 28 days and those that have reached equilibrium, respectively. ratio of one would be expected. However, for those close to reaching equilibrium, a ratio of 0.5 should be obtained if the dissolved analyte concentration in the water phase did not change noticeably during the field test. For analytes with log Kqw > 5, ratios are in the range 0.5—1.0 with most values in the range 0.6—0.95 (Figure 3). Since PRC data demonstrated that most of these compounds were in the linear phase of uptake during the 28 days, ratios of one should be observed. The lower values seen here may result from increasing fouling of the samplers over time during exposure. When considering analytes that have neared equilibrium, most ratios are well below 0.5. While this could be explained by radical changes in dissolved analyte concentrations during exposures, contaminant masses accumulated during four successive 7-day exposures of silicone strips did not demonstrate such changes in concentration (data not shown). Masses of contaminants with log Kqw < 5.2 accumulated in all samplers appear to decrease with increasing exposure time and increasing membrane fouling, possibly as a result of degradation of the less hydrophobic PAHs. However, (photo-) degradation of analytes sorbed onto the receiving phase of the samplers is unlikely. Concerns can be raised when estimating Ctwa of the more mobile and degradable compounds when heavy membrane fouling is observed during long passive sampler deployment. Additional work is required to understand such a process and to estimate its overall importance in the determination of Ctwa- For the more hydrophobic contaminants, generally linear uptake was observed and exposure time/heavy fouling induced only minor changes in estimates of Ctwa- While minimal effects of biofouling have previously been observed (24), changes in uptake rates during exposure can be compensated since biofouling is expected to affect PRC release in a similar manner to analyte uptake (13). Here, only the PRC elimination data for the Chemcatcher showed a reduction in uptake rates when increasing sampler exposure from 14 to 28 days. Regulatory Use of Passive Sampling Data. Clear objectives and readily available methods with adequate limits of detection, precision and accuracy are required for regulatory monitoring. Water quality monitoring of hydrophobic organic contaminants as defined in the European WFD is based on the comparison of samples with "whole water" EQS. Since passive sampling measures the truly dissolved fraction of contaminants in water data generated by this method cannot be compared directly 5388 ■ ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 14, 2009 with currently set WFD EQS, even though the fraction sampled is more toxicologically relevant. Nevertheless, comparisons with "whole water" EQS values are possible after a further data manipulation to account for sorption to DOC and suspended particulate matter data (see SI). DOC-water (KDOc) and OC-water (Koc), partition coefficients (25, 26) may be used to calculate "whole water" concentrations from passive sampler-based Ctwa- Despite the high uncertainty of Koc and KDOc, the use of conservative values will result in an overestimation of "whole water" concentrations. If these are still well below EQS, compliance maybe demonstrated. Passive sampling-based whole water concentrations were compared with those obtained using bottle sampling collected during the field trial and with monthly institutional monitoring data for the period 2002-2005 (Table S5). Additionally, "whole water" concentrations were estimated from data obtained from monitoring of suspended particulate matter and of the fraction of organic carbon for the same 2002—2005 period. Bottle sampling was characterized by many measurements below limits of detection (LODs) that varied by a factor of 2—7. When comparing concentrations measured by bottle sampling with EQS values (Table S5), it is important to take account of limits of quantification, particularly for larger molecular weight PAHs (e.g., forbenzo[gfti]perylene) since for a method to be considered fit-for-purpose these values should not exceed one-third of the EQS. Mean whole water concentrations of benzo[gfti]perylene and inde-no[l,2,3-cd]pyrene estimated from passive sampling are very close to proposed WFD annual average EQS. Most mean concentrations estimated from suspended particulate matter monitoring for 2002—2005 were variable and close to or above EQS (2). Passive samplers generally provide data that is less variable than that from "whole water" sampling since the latter may be strongly influenced by levels of suspended particulate matter. This lower variability is an attractive characteristic in the monitoring of water quality and the detection of temporal trends in concentrations. The present study showed that the Ctwa estimated by the different samplers varied by a factor of 2 on average while short-term within-sampler variability was a factor of 1.3. Efforts should focus on quantifying the long-term within-sampler variability and understanding and reducing the variability between different types of samplers. LODs of passive samplers with large surface area are likely to be well below typical concentrations encountered across Europe for analytes with log Kow < 7.5, and this enables their use for monitoring tasks such as comparison with EQS or the monitoring of trends (4,27). For other samplers such as Chemcatcher and MESCO, screening for larger molecular weights PAHs can be undertaken with "field" LODs in a similar range to EQS levels. Investigative monitoring tasks or monitoring at sensitive sites or where elevated concentrations are expected (e.g., sewage/storm-water effluents) are therefore most appropriate applications for these devices. Acknowledgments We thank Nel Frijns and the RIZA monitoring team at Eijsden (The Netherlands), Uwe Schröter in Leipzig for the analysis of MESCO/ silicone rod samplers and guidance with the size-exclusion chromatography, and Ronald van Bommel for the extraction and analysis of silicone strips and LDPE membranes at NIOZ. We acknowledge financial support from the European Union's Sixth Framework Programme (Contract SSPI-CT-2003-502492; http://www.swift-wfd.com). Views presented here are those of the authors alone. Supporting Information Available Additional details on water quality, sampler replication data, lists of chemicals analyzed and detected by the various samplers, and details of the calculation of time-weighted average concentrations.This material is available free of charge via the Internet at http://pubs.acs.org. Literature Cited (1) Warren, N; Allan, I. J.; Carter, J. E.; House, W. A.: Parker, A. Pesticides and other micro-organic contaminants in freshwater sedimentary environments—A review. Appl. Geochem. 2003,18 (2), 159-194. (2) Lepom, P.: Brown, B.: Hanke, G.: Loos, R.: Quevauviller, P.; Wollgast, J. Needs for reliable analytical methods for monitoring chemical pollutants in surface water under the European Water Framework Directive./. Chromatogr., A2009, 1216, 302-315. (3) Vrana, B.: Mills, G. A.: Allan, I. J.; Dominiak, E.; Svensson, K.; Knutsson, J.; Morrison, G.: Greenwood, R. Passive sampling techniques for monitoring pollutants in water. TrAC, Trends Anal. Chem. 2005, 24 (10), 845-868. (4) Allan, I. J.; Vrana, B.: Greenwood, R.: Mills, G. A.: Roig, B.; Gonzalez, C. A "toolbox" for biological and chemical monitoring requirements for the European Union's Water Framework Directive. Talanta 2006, 69 (2), 302-322. (5) Kot-Wasik, A.: Zabiegala, B.: Urbanowicz, M.: Dominiak, E.; Wasik, A.: Namiesnik, J. Advances in passive sampling in environmental studies. Anal. Chim. Acta 2007, 602 (2), 141-163. (6) Booij, K.; Sleiderink, H. M.: Smedes, F. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ. Toxicol. Chem. 1998, 17 (7), 1236-1245. (7) Huckins, J. N; Petty, J. D.: Lebo, J. A.: Almeida, F. V.: Booij, K.; Alvarez, D. A.: Clark, R. C.; Mogensen, B. B. Development of the permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environ. Sci. Technol. 2002, 36 (1), 85-91. (8) Booij, K.; Smedes, F.: van Weerlee, E. M. Spiking of performance reference compounds in low density polyethylene and silicone passive water samplers. Chemosphere 2002, 46 (8), 1157-1161. (9) Vrana, B.: Mills, G. A.: Dominiak, E.: Greenwood, R. Calibration of the Chemcatcher passive sampler for the monitoring of priority organic pollutants in water. Environ. Pollut. 2006, 142 (2), 333-343. (10) Vrana, B.: Popp, P.: Paschke, A.: Schuurmann, G. Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water. Anal. Chem. 2001, 73 (21), 5191-5200. (11) Paschke,A.: Schwab, K.; BrummerJ.: Schuurmann, G.: Paschke, H.: Popp, P. Rapid semi-continuous calibration and field test of membrane-enclosed silicone collector as passive water sampler. /. Chromatogr., A 2006, 1124 (1-2), 187-195. (12) Vrana, B.: Schuurmann, G. Calibrating the uptake kinetics of semipermeable membrane devices in water: Impact of hydrodynamics. Environ. Sci. Technol. 2002, 36 (2), 290-296. (13) Huckins, J. N; Petty, J. D.: Booij, K., Monitors of organic chemicals in the environment: Semipermeable membrane devices. Springer: New York, 2006. (14) Booij, K.; Hofmans, H. E.; Fischer, C. V.: Van Weerlee, E. M. Temperature-dependent uptake rates of nonpolar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environ. Sci. Technol 2003, 37 (2), 361-366. (15) Vrana, B.: Mills, G. A.: Kotterman, M.; Leonards, P.: Booij, K.; Greenwood, R. Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water. Environ. Pollut. 2007, 145 (3), 895-904. (16) Rusina, T. P.: Smedes, F.: Klanová, J.; Booij, K.: Holoubek, I. Polymer selection for passive sampling: A comparison of critical properties. Chemosphere 2007, 68 (7), 1344-1351. (17) Booij, K.: Hoedemaker, J. R.: Bakker, J. F. DissolvedPCBs, PAHs, and HCB in pore waters and overlying waters of contaminated harbor sediments. Environ. Sci. Technol. 2003, 37 (18), 4213-4220. (18) Paschke, A.: Popp, R. Solid-phase microextraction fibre-water distribution constants of more hydrophobic organic compounds and their correlations with octanol-water partition coefficients. /. Chromatogr., A 2003, 999 (1-2), 35-42. (19) Wennrich, L.: Vrana, B.: Popp, P.: Lorenz, W. Development of an integrative passive sampler for the monitoring of organic water pollutants. /. Environ. Monit. 2003, 5 (5), 813-822. VOL. 43, NO. 14, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY ■ 5389 (20) Yates, K.; Davies, I.; Webster, L.; Pollard, P.; Lawton, L.; Moffat, C. Passive sampling: partition coefficients for a silicone rubber reference phase. /. Environ. Monit. 2007, 9 (10), 1116-1121. (21) Paschke, A.; Hanke, K.; Schuiirmann, G. in preparation. (22) Paschke, A.; Brummer, J.; Schuurmann, G. Silicone rod extraction of pharmaceuticals from water. Anal. Bioanal. Chem. 2007, 387 (4), 1417-1421. (23) Vrana, B.; Paschke, A.; Popp, P. Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water. Environ. Pollut. 2006, 144 (1), 296-307. (24) Booij, K.; van Bommel, R.; Mets, A.; Dekker, R. Little effect of excessive biofouling on the uptake of organic contaminants by semipermeable membrane devices. Chemosphere2006, 65(11), 2485-2492. (25) Karickhoff, S. W. Semiempirical estimation of sorption of hydrophobic pollutants on natural sediments and soils. Chemosphere 1981, 10 (8), 833-846. (26) Burkhard, L. P. Estimating dissolved organic carbon partition coefficients for nonionic organic chemicals. Environ. Sei. Technol. 2000, 34 (22), 4663-4668. (27) Allan, I. J.; Mills, G. A.; Vrana, B.; Knutsson, J.; Holmberg, A.; Guigues, N.; Laschi, S.; Fouillac, A. M.; Greenwood, R. Strategic monitoring for the European Water Framework Directive. TrAC, Trends Anal. Chem. 2006, 25 (7), 704-715. ES900608W 5390 ■ ENVIRONMENTAt SCIENCE & TECHNOtOGY / VOt. 43, NO. 14, 2009 Príloha 16 Lobpreis T., Vrana B., and Dercová K., Innovative approach to monitoring organic contaminants in aqueous environment using passive sampling devices, Inovatívne prístupy k monitorovaniu organických kontaminantov vo vodnom prostredí použitím pasívneho vzorkovania, Chemické Listy 2009, 103, 548-558. Chem. Listy 103, 548-558 (2009) Referát INOVATIVNE PRÍSTUPY K MONITOROVANIU ORGANICKÝCH KONTAMINANTOV VO VODNOM PROSTREDÍ POUŽITÍM PASÍVNEHO VZORKOVANIA TOMÁŠ LOBPREISa, BRANISLAV VRANAb a KATARÍNA DERCOVÁa "Oddelenie biochemickej technologie, Ústav biotechnologie a potravinárstva, Fakulta chemickej a potravinárskej technologie, Slovenská technická univerzita, Radlinského 9, 812 37 Bratislava, bNárodné referenčné laboratórium pre oblasť vôd na Slovensku, Výskumný ústav vodného hospodárstva, Nábrežie arm. gen. L. Svobodu 5, 812 49 Bratislava, Slovensko tomas. lobpreis@gmail. com Došlo 17.9.08, prepracované 8.12.08, prijaté 23.12.08. Kľúčové slová: pasívne vzorkovanie, organické kontami-nanty, monitorovanie životného prostredia, biomonitoring Obsah 1. Úvod 2. Monitorovanie kontaminantov 2.1. Bodové odbery vzoriek 2.2. Biomonitoring 2.3. Pasívne vzorkovanie 2.4. Porovnanie biomonitoringu a pasívneho vzorkovania 2.5. Faktory ovplyvňujúce pasívne vzorkovanie 3. Prehľad typov pasívnych vzorkovačov 3.1. Chemcatcher 3.2. Semipermeabilné membránové zariadenie 3.3. Polárne organické chemické integračné vzorkovače 3.4. Vzorkovače s membránou uzavretým sorpčným potahom 3.5. Keramický dozimeter 3.6. Pasívny difúzny vak 3.7. Mikroextrakcia na tuhú fázu 4. Záver 1. Úvod Problematika monitorovania kontaminantov vo vodnom prostredí je aktuálna najmä v prípade stopových organických látok. Mnohé z nich patria medzi ťažko degrado-vateľné zlúčeniny, pričom u mnohých z nich dochádza k bioakumulácii v organických tkanivách. Do životného prostredia bolo uvoľnené antropogénnou činnosťou veľké množstvo chemických látok s rôznymi fyzikálno-chemic- kými vlastnosťami, preto aj celkový vplyv týchto látok na ekosystém nie je jednoduché popísať. Tieto polutanty zahŕňajú pesticídy, organické rozpúšťadlá, priemyselné chemikálie, liečivá, látky z priemyselného a domáceho odpadu a degradačné produkty týchto látok. Osud kontaminan-tu v životnom prostredí je často neznámy, mnohé z nich prechádzajú dokonca procesom čistenia odpadových vôd bez zmeny. Problematikou perzistencie organických kontaminantov pri úprave pitnej vody sa zaoberal napr. Stac-kelberg a spol.1 Vzorkovanie2 patrí medzi najdôležitejšie kroky každého analytického postupu, pretože chyby, ktoré vzniknú pri odbere vzoriek už nie je možné neskôr odstrániť3 a tým výrazne vplývajú na celkovú nepresnosť merania4. Málokedy je možné odobratú vzorku analyzovať priamo, vo väčšine prípadov je nevyhnutné, aby finálnej analýze predchádzali rôzne úpravy. Zahŕňajú extrakciu analytu z vodného prostredia, aby sa odstránili prípadné interferencie a zakoncentrovanie vzorky z dôvodu zvýšenia citlivosti metódy. Tieto postupy, najmä analýza stopových koncentrácií kontaminantov, sú časovo veľmi náročné a nezriedka predstavujú 70-90 % z celkového času analýzy5. Preto trendy v tejto oblasti smerujú k zjednodušeniu analýzy, napríklad spojením vzorkovania a zakoncentrovania do jedného kroku, alebo k zníženiu objemov použitých rozpúšťadiel, čo je efektívnejšie z ekonomického, ale aj ekologického hľadiska. Pod pojmom pasívne vzorkovanie rozumieme techniku, ktorá je založená na voľnom prestupe analyzovanej látky zvodného prostredia do prijímajúcej fázy pasívneho vzorkovača ako výsledok rozdielov chemického potenciálu analytu medzi oboma fázami. Prestup látky sa riadi kineti-kou 1. Fickovho difúzneho zákona a prebieha až do vytvorenia termodynamickej rovnováhy v systéme. Predstavuje rýchlu, efektívnu a jednoduchú metódu na monitorovanie širokého spektra organických aj anorganických kontaminantov v prostredí. Jej výhoda spočíva v znížení nákladov, znížení objemov rozpúšťadiel, vo vysokej citlivosti a v poskytnutí informácie o časovo váženej koncentrácii (time-weighted average; TWA). Keďže ide o in situ metódu, nedochádza v porovnaní s bodovými odbermi k zmenám zloženia vzorky (napr. pH, teplota, obsah kyslíka) počas transportu6. Pasívne vzorkovače je možné okrem vodného prostredia použiť aj na analýzu kontaminantov vo vzduchu a pôde. Po prvýkrát boli patentované a použité v roku 1927 na sledovanie koncentrácie oxidu uhoľnatého vo vzduchu7. Odvtedy došlo k výraznému vývoju a rozšíreniu oblasti využitia, čo dokumentuje aj množstvo publikácií a literárnych prehľadov uverejnených na tému pasívneho vzorkovania. Najvýznamnejšie z rešerší sú uvedené v tabuľke I. 548 Chem. Listy 103, 548-558 (2009) Referát Tabuľka I Zoznam prehľadových prác k problematike pasívneho vzorkovania Rok Meno autora Predmet prehľadovej práce 1981 Fowler8 Teória a základy pasívneho vzorkovania vo vzduchu vrátane vplyvu teploty, tlaku a odozvy vzorkovača 2000 Kot a spol.9 Dlhodobé monitorovanie vo vodnom prostredí 2000 Lu a spol.10 Teória a aplikácia semipermeabilných membrán (SPMD) 2002 Gorecki a spol.5 Použitie pasívnych vzorkovačov v pôde, vo vzduchu a vo vode a biomonitorovanie 2003 Mayer a spol.11 Vzorkovanie v rovnovážnej oblasti 2005 Stuer-Laudrisen12 Monitorovanie organických mikropolutantov 2005 Namiesnik a spol.13 Pasívne vzorkovanie s dôrazom na mikroextrakciu na tuhej fáze (SPME) 2005 Vrana a spol.14 Metódy pasívneho vzorkovania vo vodnom prostredí 2007 Mills a spol.15 Monitorovanie farmaceutických látok 2007 Vrana a spol.16 Pasívne vzorkovanie na monitorovanie znečistenia životného prostredia 2008 Seethapathy a spol.17 Techniky pasívneho vzorkovania v environmentálnej analýze vo vodách, pôdach a vzduchu 2. Monitorovanie kontaminantov 2.1. Bodové odbery Najčastejšie používaným spôsobom vzorkovania je bodový odber vzoriek, pri ktorom sa vzorka odoberá v určitom okamihu a na konkrétnom mieste, pričom jeho voľba by mala reprezentovať vzorkovanú oblasť ako celok18. Problematická zostáva interpretácia bodových odberov, keď sú údaje získané zo vzoriek v jednom mieste a v jednom okamihu používané na charakterizáciu stavu celej lokality. Pri dodržiavaní európskej rámcovej smernice o vodách (WFD) je nevyhnutné zabezpečiť porovnateľ-nosť jednotlivých dát nameraných rôznymi členskými štátmi19. Na dosiahnutie tohto cieľa budú potrebné nové analytické metódy a prístupy20, výber vhodných certifikovaných referenčných materiálov a interlaboratórne experimenty21. Konvenčný postup pri monitorovaní znečistených alebo odpadových vôd pozostáva z odberu väčšieho množstva vody, zakoncentrovania vzorky v laboratóriu rôznymi extrakčnými technikami, vyčistenia vzorky od potenciálne interferujúcich příměsí a následnej inštrumentálnej analýzy. Monitorovanie vo vode komplikuje viacero problémov, najmä veľmi nízka hladina kontaminantov a jej premenlivosť. Pre dosiahnutie požadovanej medze detekcie je často nutné spracovať veľký objem vzorky, čo je v prípade ultra-stopových koncentrácií veľmi zložité. Veľkým problémom je predovšetkým stanovenie kontaminantov rozpustených vo vodnej fáze, teda biologicky dostupných22. Rovnako komplikované je stanovenie toxických účinkov polutantov v nízkych koncentráciách pomocou biotestov. Akútne testy toxicity spravidla nezaznamenajú odozvu a chronické testy sú omnoho náročnejšie a drahšie, aj keď umožňujú sledovať dlhodobé účinky aj nízkych koncentrácií. Takýto systém síce dokáže simulovať vplyv dlhodobej expozície polutantov na organizmus, ale opäť ide o vzorku odobranú v jednom okamihu23. Nevýhody bodových odberov možno zhrnúť nasledovne: Analýzy vzoriek získaných z bodových odberov reprezentujú iba zloženie vzorky v momente odberu a nemusia zachytiť náhodnú kontamináciu v inom čase. - Nastávajú problémy pri kontrole kvality pri manipulácii s veľkými objemami vody potrebnými na analýzu stopových koncentrácií. - Bežnými analytickými postupmi sa nedá stanoviť koncentrácia skutočne rozpustených a biodostupných polutantov. Toxikologické dáta a chemické kritériá kvality vody sú často založené iba na koncentrácii rozpustených látok a nie na celkovom množstve polutantov vo vodnom prostredí. - Konvenčné postupy bývajú často neúspešné pri stanovení ultrastopových množstiev bioakumulujúcich kontaminantov. Obmedzenia bodových odberov sa dajú čiastočne eliminovať použitím opakovaných odberov, ktoré sú však fyzicky, logisticky a ekonomicky náročné, predovšetkým pri monitorovaní vzdialenejších oblastí. Bez dostatočne opakovaných odberov nie je možné vyjadriť časovo priemernú koncentráciu sledovaných látok24. Na prekonanie nedostatkov monitorovania pomocou bodových odberov bolo vyvinutých viacero metód, ako napr. biomonitoring, on-line monitoring, in situ kontinuálny odber vzoriek, alebo pasívne vzorkovanie25. 549 Chem. Listy 103, 548-558 (2009) Referát 2.2. Biomonitoring Meranie koncentrácie polutantov v tkanivách živých vodných organizmov, najmä rýb, je obvyklou metodou používanou na monitorovanie úrovne kontaminácie vôd polutantmi. Táto metóda je založená na jave bioakumulá-cie, t.j. zakoncentrovani hydrofóbnych látok (napr. polychlorované bifenyly (PCB), polycyklické aromatické uhľovodíky (P AH) a organochlórované pesticídy (OCP)) v tukových tkanivách organizmov. Proces aktívnej a pasívnej akumulácie umožňuje merať koncentrácie skúmaného analytu, ktoré mnohonásobne prevyšujú jeho koncentráciu v prostredí a ktoré by nebolo možné detegovať konvenčnými analytickými postupmi. Využitie živých organizmov pri monitorovaní znečistenia životného prostredia eliminuje niektoré nedostatky bodových odberov. Takto získané dáta reprezentujú odozvu na skutočne bio-dostupnú frakciu kontaminantu a je možné priamo sledovať toxický vplyv na organizmus. Biomonitoring patrí medzi integrálne techniky, čo predstavuje výhodu oproti bodovým odberom, keďže dochádza ku kontinuálnej akumulácii analytu26. Takto získané údaje poskytujú informáciu o časovo spriemernenej koncentrácii polutantu. Využitie živých organizmov ako vzorkovačov má viacero výhod: odzrkadľujú skutočný vplyv stavu životného prostredia na živočíchy, vo väčšine prípadov sa využívajú natívne druhy, čiže odpadá potreba ich transportu na skúmanú lokalitu a aj ekonomický aspekt je nezanedbateľný27. Mali by však spĺňať isté kritériá28: a) nemalo by dochádzať k ich migrácii, b) musia byť rozšírené v celej sledovanej lokalite, c) pri dlhodobom monitoringu by mala byť zabezpečená stabilná populácia ad) na rôznych miestach by mala platiť rovnaká korelácia medzi koncentráciou v prostredí a tkanive. Všeobecne môžu byť živé organizmy použité v procese monitorovania životného prostredia dvoma spôsobmi: ako biomonitory a ako bioindikátory. Patria medzi ne organizmy alebo spoločenstvá organizmov, ktoré poskytujú informácie o kvalitatívnych alebo kvantitatívnych zmenách polutantov v životnom prostredí, resp. pri bioindikátoroch sa sledujú morfologické a histologické zmeny, často na bunkovej úrovni, ako aj metabolicko-biochemické procesy. Biomonitorovanie je možné použiť aj pre ťažké kovy29. Použitie živých organizmov na in situ monitorovanie kontaminantov v životnom prostredí je sprevádzané s viacerými problémami a obmedzeniami. Výsledná miera expozície je viazaná na druh, zdravotný stav a pohlavie organizmu, ako aj na konkrétnu oblasť, povahu vody a teplotu. Z dôvodu charakteristickej schopnosti metaboli-zácie určitého typu látok u niektorých zo sledovaných kontaminantov nemusí dochádzať k bioakumulácii, navyše je táto schopnosť závislá od veku a pohlavia organizmu. Významným obmedzením je migrácia druhov v závislosti na teplote, množstve a druhu potravy, nezohľadňuje sa ani výška hladiny toku. V konečnom dôsledku je obtiažne s istotou vzťahovať ryby k danému miestu odberu. Práve vyššie spomenuté podmienky môžu predstavovať nevýhody použitia živých organizmov pri monitorovaní znečiste- nia. Problematická zostáva aj interpretácia výsledkov, ako aj ich porovnanie z rôznych odberových oblastí. Rovnako ich nie je možné použiť v prostredí s vysokou koncentráciou kontaminantov (napr. čistiarne odpadových vôd). Živé organizmy je možné použiť aj ako biologické systémy skorého varovania, ktoré využívajú toxikologickú odozvu organizmu na prítomnosť kontaminantu v prostredí30. Ako indikátorové organizmy slúžia najčastejšie rôzne druhy rýb, larvy komárov, dafnie31, mikroorganizmy32, ustrice a iné mäkkýše. Systémy skorého varovania sa najčastejšie používajú pri monitorovaní pitnej vody a jej rozvodov33 a pri čistiarňach odpadových vôd. Polutanty sa v organizme zakoncentrujú z rozpustenej fázy vo vode (biokoncentrácia), ako aj príjmom týchto látok z ich potravinového reťazca (bioakumulácia). Živé organizmy teda nemôžu spoľahlivo plniť úlohu pre identifikáciu zdroja kontaminácie z dôvodu nedostatočnej proporcionality medzi koncentráciou v tukovom tkanive a vo vodnom prostredí. V prípade spracovania vzoriek živých organizmov (napr. rýb) nie je určená jednotná metóda spracovania vzoriek. Na analýzu sa používa tzv. jedlý podiel, ten však nie je jednoznačne definovaný. Tým je znížená porovnateľnosť dát, pretože niektoré sú generované len z čistej svalovej hmoty rýb a iné aj zo zhomogenizova-nej svaloviny akože. Namerané výsledky sa síce nemusia výrazne odlišovať, neposkytujú však opakovateľné údaje. Biomonitoringu sa bližšie venoval vo svojich prácach napr. Mora aspol.34 Pre porovnanie, postup odberu a spracovania vzoriek s použitím pasívnych vzorkovačov je jednoznačne daný a spracovaný do podrobného návodu. Biomonitorovanie bolo použité napríklad v rámcovom programe EU (WorkPackage 3, WP3) pri monitorovaní kvality vôd Stredozemného mora35, kde boli ako živé organizmy použité mušle36 a ryby37. 2.3. Pasívne vzorkovanie Pasívne vzorkovače fungujú vo vodnom prostredí na integratívnom princípe, t.j. počas expozície dochádza ku kontinuálnej extrakcii sledovaných látok zvody (akumuluje sa len rozpustný, ľahko bioprístupný podiel) bez toho, aby sa dosiahla termodynamická rovnováha medzi organickou fázou vo vzorkovači a vodou. Rýchlosť akumulácie látky do vzorkovača je priamo úmerná jej vodnej koncentrácii. Po ukončení expozície je možné z akumulovaného množstva analytu pomocou vzorkovacej rýchlosti odhadnúť hodnotu časovo váženého priemeru vodnej koncentrácie (TWA, time-weighted average con-centration) aj periodicky sa opakujúceho znečistenia počas expozície. Meraniu TWA koncentrácie v prostredí sa bližšie venoval napr. Zhao a spol.38 Týmto spôsobom je možné znížiť nutný interval vzorkovania a tým dosiahnuť aj zníženie nákladov. Metódy pasívneho vzorkovania ďalej umožňujú eliminovať viaceré nevýhody bodových odberov, ako napríklad zachytenie výkyvov koncentrácie kontaminantov vo vode, zjednodušeniu analytických postupov, alebo detekciu aj ultrastopových množstiev látok, 550 Chem. Listy 103, 548-558 (2009) Referát ktoré bežnými postupmi nie je možné detegovať. Výhodou je aj monitorovanie biodostupnej frakcie, ktorá je relevantná pre predikciu osudu látok v životnom prostredí, keďže metóda pasívneho vzorkovania je analogická s javom bio-koncentrácie kontaminantu z vody do živých organizmov. Základnou prednosťou tejto metódy je model expozície, ktorý je možné opísať fyzikálno-chemickými parametrami39. Nakoľko akumulačné schopnosti tohto modelu sú podobné ako u vodných živočíchov (s výnimkou vplyvu faktorov charakteristických pre živý organizmus: nezávislosť na druhu, pohlaví, bez metabolizácie, akumulácia nie je prahová pre prežitie organizmu), označuje sa model často ako „virtual fish"40. Technika pasívneho vzorkovania umožňuje prepočet koncentrácie konta-minantov na základe kalibračných dát a tým aj stanovenie koncentrácie sledovaných polutantov vo vode. Kontaminanty sú pri pasívnom vzorkovaní zachytené a viazané do vhodného média obsiahnutého vo vzorkovači, ktoré označujeme ako prijímajúca fáza. Môže ňou byť rozpúšťadlo, chemické činidlo alebo porózny adsorbent. Prijímajúca fáza je vystavená expozícii vo vodnom prostredí, ale nedochádza ku kvantitatívnej extrakcii, ako je tomu pri vsádzkovej extrakcii. Koncentrácia kontaminantu vo vodnom prostredí sa nemení vplyvom extrakčného procesu41. Princípom extrakcie je prestup analyzovanej látky zvodného prostredia cez fázové rozhrania do prijímajúcej fázy. Limitujúcou vrstvou prestupu by mala byť membrána, ale často, najmä pri nízkych hodnotách kon-vekcie, sa limitnou stáva laminárna difúzna vrstva vody na povrchu membrány. Adsorpcia alebo absorpcia kontami -nantov zvodného prostredia sa u väčšiny vzorkovačov riadi podľa modelu, ktorý odvodil Huckins a spol42. Kinetika akumulácie je riadená difúziou a dá sa opísať nasledovným vzťahom Cs(t) = Cs(0) + (Cw.KĽ -Cs(0)). 1 - exp kde Cs(t) je koncentrácia analytu v vzorkovači v čase t; Cs (0) koncentrácia analytu vo vzorkovači v čase 0; Cw koncentrácia analytu vo vodnom prostredí; KDW rovnovážny rozdeľovači koeficient systému prijímajúca fáza/voda; k0 celkový koeficient prestupu látky; KMw rovnovážny rozdeľovači koeficient systému membrána/voda; A plocha povrchu vzorkovača; t čas; VD objem prijímajúcej fázy. Priebeh akumulácie kontaminantu z prostredia do prijímajúcej fázy sa dá rozdeliť na dva režimy - kinetický (lineárny) a rovnovážny. Pri rovnovážnom vzorkovaní je doba expozície vzorkovača v prostredí dostatočne dlhá na to, aby došlo k ustáleniu termodynamickej rovnováhy medzi koncentráciou látky vo vode a v prijímajúcej fáze. Rovnovážnemu vzorkovaniu sa venovali viacerí autori11,43. Pri vzorkovaní v kinetickej oblasti akumulácie je tok látky priamo úmerný rozdielu chemickej aktivity vo vodnej a v prijímajúcej fáze. V počiatočnej fáze expozície vzorkovača v prostredí je hodnota desorpcie analytu z prijímajúcej fázy zanedbateľná a vzorkovač pracuje v lineárnej (tj. integrálnej) oblasti. Popri nesporných výhodách spojených s pasívnym vzorkovaním je potrebné spomenúť aj limitácie týchto postupov. Nie sú vhodné ako systémy skorého varovania, ale predovšetkým na dlhodobejší monitoring. Technológia je stále vo vývoji, doteraz neboli schválené referenčné štandardizované sústavy a chýba aj zakotvenie v legislatíve. Monitorovanie pasívnymi vzorkovačmi vyžaduje rozsiahly systém kalibračných dát a problematickým zostáva aj porovnanie s výsledkami získanými konvenčným spôsobom. Špecifickým problémom je aj znečistenie povrchu vzorkovačov mikroflórou (tzv. bioznečistenie), ktoré pri dlhodobejšom monitorovaní vytvára dodatočnú bariéru voči prestupu látok cez fázové rozhrania. V neposlednom rade je to aj riziko odcudzenia počas expozície. 2.4. Porovnanie biomonitoringu a pasívneho vzorkovania Koncentrácie perzistentných polutantov extrahovaných z pasívnych vzorkovačov sú proporcionálne ku koncentráciám týchto látok rozpustených vo vode. Naproti tomu polutanty v živých organizmoch sú viac ako na vodnej koncentrácii závislé na polčase vylučovania a dochádza aj ku skresleniu profilu kontaminácie vplyvom metabolizmu. Niektoré kontaminanty (napr. PAH) vzhľadom k ich rýchlej premene v organizme často nie sú v rybách detegovateľné. Táto skutočnosť platí všeobecne pre látky s nízkym rozdeľovacím koeficientom systému n-oktanol/voda (log Kow). Pri porovnávaní výsledkov merania z expozície pomocou pasívnych vzorkovačov a živých organizmov môže dôjsť ku zhode, čo platí najmä pri interpretácii látok, ktoré sa z tela organizmu prakticky nevylučujú a bioakumulujú sa. Štúdie v rôznych krajinách na rôzne kontaminanty (napr. PAH44, OCP a PCB45, ťažké kovy ahydrofóbne organické kontaminanty46) naznačujú, že použitie pasívnych vzorkovačov na monitorovanie stopových koncentrácií je efektívnejšie ako u živých organizmov a vďaka štandardizovaným postupom poskytujú porovnateľné údaje z rôznych lokalít. Výborná korelácia medzi metódami pasívneho vzorkovania a bodovými odbermi bola dosiahnutá aj pre endokrinné disruptory47 a vybrané aromatické uhľovodíky48. Výhodou pasívnych vzorkovačov je aj možnosť ich použitia vo vysokokonta-minovanom prostredí, napr. v čistiarňach odpadových vôd, kde by biomonitorovacie organizmy neboli schopné prežiť. 2.5. Faktory ovplyvňujúce pasívne vzorkovanie Pri pasívnom vzorkovaní je potrebné zohľadňovať mechanizmus výmeny sledovaného analytu medzi vodnou a prijímajúcou fázou. Na kompenzovanie environmentálnych vplyvov pri pasívnom vzorkovaní bolo vyvinutých viacero metód. Jednou z nich je použitie vnútorných štandardov, tzv. PRCs (Performance Reference Compounds)49. Ako vnútorný štandard slúžia štruktúrne, najčastejšie deu-terované analógy skúmaného kontaminantu. Sú to analy- 551 Chem. Listy 103, 548-558 (2009) Referát ticky neinterferujúce látky, ktoré sa v prostredí prirodzene nevyskytujú. Vnútorné štandardy sa pridávajú do prijímajúcej fázy ešte pred expozíciou v presne známej koncentrácii a táto metóda je založená na skúmaní vyplavovania štandardu z pasívneho vzorkovača. Mechanizmus akumulácie môže byť ovplyvnený viacerými faktormi. Rýchlosť prestupu látky je limitovaná difúziou cez semipermeabilnú membránu, alebo vodnou laminárnou difúznou vrstvou, ktorá vzniká na rozhraní membrána -voda. Laminárna difúzna vrstva predstavuje nepremiešava-nú vodnú vrstvu v tesnej blízkosti membrány, hrúbka tejto vrstvy, a následne aj odpor voči prestupu analytu, je silne závislá od turbulenci! v okolí pasívneho vzorkovača. Pri posudzovaní difúzie limitujúcej vrstvy je potrebné zohľadniť typ a vlastnosti membrány, vlastnosti prostredia počas vzorkovania, ako aj vlastnosti monitorovaného analytu. Vo všeobecnosti platí, že vzorkovacia rýchlosť Rs, ako aj akumulované množstvo monitorovaného analytu vo vzorkovači, sa so stúpajúcou turbulenciou výrazne zvyšuje, ak difúziu limituje laminárna difúzna vrstva vody na povrchu vzorkovača. Štúdie skúmajúce vplyv hydrodynamických podmienok na prestup látky dokázali, že redukovanie turbulenci! v okolí SPMD malo v prípade organochlórova-ných zlúčenín (log Kow 4-8) za následok až 4-násobné spomalenie prestupu látky, prípadne 1,5-násobné zvýšenie pri stúpajúcej rýchlosti prúdenia kvapaliny (v rozpätí 0,004-0,2 m s-1, cit.50). Niektoré typy vzorkovačov, ako napríklad vlákna SPME, sú však hydrodynamickými podmienkami ovplyvňované v menšej miere51. Vplyvu hydrodynamických podmienok na prestup látky cez fázové rozhrania sa venovali Vrana a Schúúrmann52. Ďalším významným faktorom, ktorý ovplyvňuje prestup látky do prijímajúcej fázy, je teplota. Vo všeobecnosti platí, že so zvyšujúcou sa teplotou rastie aj prestup látky do prijímajúcej fázy. Pre závislosť teploty od Rs platí rovnica Arrheniovho typu. Napríklad, pre vzorkovač typu Chemcatcher, vyvinutý pre vzorkovanie hydrofóbnych látok, zmena teploty zo 6 na 18 °C spôsobí až vyše 5-násobné zvýšenie vzorkovacej rýchlosti53. Charakteristickým problémom pri použití pasívnych vzorkovačov vo vodnom prostredí je tvorba biofilmu. Nechránený povrch vzorkovača ponoreného do vody je vystavený riziku kolonizovania baktériami a rôznou mikroflórou a faunou prirodzene sa vyskytujúcich vo vodnom prostredí. To môže viesť k vytvoreniu biofilmu na povrchu membrány, ktorý svojimi vlastnosťami znižuje celkový koeficient prestupu látky, najmä zvýšením hrúbky difúznej vrstvy a blokovaním pórov semipermeabilnej membrány. Hrúbka vrstvy biofilmu môže byť rozdielna aj medzi jednotlivými vzorkami vzorkovačov pochádzajúcich z toho istého experimentu. Zloženie biofilmu je závislé od mikrobiologického zloženia a vlastností vodného systému. Kolonizujúce mikroorganizmy môžu dokonca poškodiť povrch membrány, ak je vyrobená zbiodegrado-vateľného materiálu. Huckins a spol.54 zistili vo viacerých prípadoch 20-70% zníženie akumulácie polycyklických aromatických uhľovodíkov pri použití pasívnych vzorkovačov SPMD znečistených biofilmom. Model použitý na opis prestupu látky cez biofilm naznačuje, že v ideálnom prípade sa správa ako imobilizovaná vodná vrstva, s odporom voči prestupu látky nezávislým od rozdeľova-cieho koeficientu biofilm-voda. To znamená, že látky di-fundujú cez biofilm takmer rovnakou rýchlosťou, bez ohľadu na ich hydrofóbnosť. Toto bolo potvrdené aj ďalšou štúdiou55 pri monitorovaní PAH, kde bol zistený 50% pokles akumulácie v porovnaní s neznečistenými vzorkami, avšak vhodnou voľbou vnútorných štandardov bolo možné tento problém eliminovať. 3. Prehľad typov pasívnych vzorkovačov 3.1. Chemcatcher Pasívny vzorkovač Chemcatcher pracuje vo vodnom prostredí na integratívnom princípe, t.j. počas expozície dochádza ku kontinuálnej extrakcii sledovaných látok z vody bez toho, aby sa dosiahla termodynamická rovnováha medzi organickou fázou vo vzorkovači a vodou. Chemcatcher je zložený z viacerých častí. Podporný disk slúži na umiestnenie membrány a prijímajúcej fázy, predná a zadná časť pomocou vodotěsného závitu tento disk upevnia. Pri transporte sa používa aj uzáver, ktorý zabraňuje mechanickému poškodeniu povrchu membrány. Voliteľne sa môže použiť aj ochranná mriežka vyrobená z nehrdzavejúcej ocele, bronzu alebo medi. Celkový priemer predstavuje 70 mm a efektívna vzorkovacia plocha priemer 45 mm. Schránka vzorkovača je vyrobená z polytetrafluoroetylénu (PTFE; Teflon), ktorý je vhodný najmä kvôli veľmi nízkej schopnosti adsorbovat' sledované analyty na svojom povrchu. Pri expozícii v prostredí sa vzorkovače umiestňujú horizontálne membránou smerujúcou nadol, aby sa eliminovala akumulácia sedimentujúcich častíc na povrchu disku. Základom je difúzno-limitná membrána a viazaná tuhá prijímajúca fáza. Ich vhodnou kombináciou sa dá Chemcatcher použiť na monitorovanie širokého spektra látok, ako napr. vysoko nepolárných látok (DDT, DDE)56, organociničitých zlúčenín (MBT, TBT)57, polárnych a semipolárnych pesticídov58 alebo farmaceutík59. Na vzorkovanie nepolárných organických kontaminantov s hodnotou log Kow väčšou ako 4 sa používa polyetylénová (LDPE; low density polyethylene) membrána a ako prijímajúca fáza slúži Ci8 Empore™ disk. Táto fáza je založená na tuhom sorbente imobilizovanom do polymér-nej matice (90 % sorbent : 10 % PTFE hm.) vo forme disku a prekonáva viaceré problémy spojené s používaním kvapalných prijímajúcich fáz, ako napríklad vyplavenie do vodného prostredia. Takýto systém je aj odolnejší voči poškodeniu a naviac je možné vhodnou voľbou komerčne dostupných diskov zvýšiť spektrum analyzovaných látok, alebo naopak, selektívne zvoliť fázu zachytávajúcu úzku skupinu kontaminantov. Pre polárnejšie látky sa používa polyétersulfónová membrána (PES) a taktiež Empore™ disk, ktorý sa vyznačuje dostatočnou afinitou aj kapacitou pre väčšinu relevantných kontaminantov. Medzi ďalšie 552 Chem. Listy 103, 548-558 (2009) Referát používané mikroporózne limitno-difúzne membrány patria membrány zo sklenených vlákien, polykarbonátu, teflonu, polyvinylidéndifluoridu (PVDF), acetátu celulózy (CA), polysulfónu (PS) a regenerovanej celulózy. Membrána slúži ako semipermeabilná bariéra redukujúca prestup látky medzi vodným prostredím a prijímajúcou fázou a takisto zabraňuje prestupu molekúl väčších ako veľkosť pórov membrány, ako napr. znečiťujúce anorganické častice, makromolekuly alebo mikroorganizmy16. Pre polárnej-šie látky (log Kow < 3) sa používa fáza vyrobená zo sulfó-novaného polystyréndivinylbenzénu (SDB-RPS, alebo SDB-XC Empore™ disk)60. Ak sa zvolí ako prijímajúca fáza chelatačný disk, je možné pomocou vzorkovačov Chemcatcher monitorovať vo vodnom prostredí aj obsah toxických a ťažkých kovov (Cu, Cd, Co, Mn, Ba, Ca, Sr, Zn, Al, Cr, Sn, Pb, Fe, Ni, Mg)61. Pasívne vzorkovače Chemcatcher prešli od ich prvého použitia v praxi62 vývojom, čo vyústilo do prípravy vzorkovačov Chemcatcher II. generácie. Sú vyrobené z lisovaného plastu (polykarbonát), skladajú sa z troch častí a membrána s prijímajúcou fázou sa upevňuje jednoduchým „zacvaknutím". Cieľom vývoja bolo zefektívnenie činnosti vzorkovača, zníženie hmotnosti, zjednodušenie manipulácie a v neposlednom rade zredukovanie nákladov potrebných na výrobu. Táto optimalizácia, najmä zníženie profilu z 30 na 7 mm, viedla k zlepšenej kinetike vzorkovania a zníženiu vnútorného odporu vzorkovača voči prestupu hydrofóbnych organických látok s log Kow väčším ako 5 (cit.63). Toto bolo docielené pridaním malého množstva «-oktanolu do priestoru medzi prijímajúcu fázu a polyetylénovú membránu. «-oktanol je rozpúšťadlo s vysokou afinitou k sledovaným látkam64. 3.2. Semipermeabilné membránové zariadenie Ďalším typom pasívnych vzorkovačov sú semipermeabilné membrány (SPMDs, Semipermeable Membrane Devices) plnené trioleínom, syntetickým rybím tukom. Vzorkovací systém bol vyvinutý Huckinsom a spol.65 a jeho usporiadanie sa ustálilo v štandardne používanej konfigurácii. Vzorkovač SPMD sa skladá zpoloprie-pustnej membrány (hrúbky 75-95 um) rozmerov 94 x 2,5 cm s pórmi špecifického rozmeru do l-10~9m, čo je základné priblíženie k veľkosti molekúl, ktoré môžu difundovat' cez biomembrány. Vnútri membrány je uzavretý syntetický lipid triolein (l,2,3-tri-[c/í-9-octadecenoyl] glycerol). Pri expozícii dochádza k akumulácii lipofilných kontaminantov do prostredia trioleínu. Kapacita SPMD je daná jej rovnovážnym rozdeľova-cím koeficientom KTW (systém trioleín/voda) a objemom trioleínu. Schopnosť vzorkovania je určená veľkosťou tohto parametra. Uspokojivo sa dajú vzorkovať látky s hodnotou log Kow < 6,5, avšak kvôli hydrofóbnej membráne nie je tento systém vhodný na monitorovanie látok s log Kow < 3. Monitorovanie pomocou SPMD bolo overené pre rôzne analyty, ako napr. polychlorované dibenzo-dioxíny afurány (PCDD, PCDF)66, pesticídy (DDT, DDE, DDD)67, polycyklické aromatické uhľovodíky (PAH)68, organochlórované pesticídy alebo polychlorované bifenyly (PCB)70. Výhodou SPMD je možnosť expozície priamo v sedimentoch ako kontaktný priamy test a tieto výsledky sa môžu použiť pre odhad rizika pre bentické organizmy. Rovnako ako vzorkovače Chemcatcher, aj vzorkovače SPMD vyžadujú súbor kalibračných dát na elimináciu charakteristických environmentálnych podmienok71. Systém potom na základe zistených kinetických parametrov umožňuje prepočet koncentrácie kontaminantov a tým stanovenie výslednej koncentrácie sledovaných polutantov vo vode. Po expozícii v prostredí sú membrány extrahované «-hexánom alebo dichlórmetánom a následne je dialy-zát analyzovaný chemicky alebo toxikologický72. Chemická analýza prebieha metódami kvapalinovej chromatogra-fie (HPLC), plynovej chromatografie, alebo plynovej chro-matografie s hmotnostnou spektrometriou (GC/MS). Použitie SPMD dobre simuluje proces difúzie cez biomembrány (napr. epitel rybích žiabrov). Difúzia cez biomembrány je považovaná za rozhodujúcu pri biokon-centrácii polutantov. SPMD naplnené syntetickým rybím tukom dokážu simulovať proces biokoncentrácie v živom organizme. Toto bolo potvrdené aj štúdiou73, v ktorej boli porovnávané koncentrácie PCB, PAH aOCP vo vode, rybách a sedimentoch. SPMD sú vyrábané zo syntetických materiálov, ktoré zaisťujú väčšiu jednotnosť a reproduko-vateľnosť ako živé organizmy. Vďaka svojej vysokej citlivosti zachytia širokú škálu chemikálií aj v stopových koncentráciách, vrátane takých, ktoré sú organizmami metabo-lizované. Môžu byť exponované nielen vo vodnom prostredí, ale aj v sedimentoch74. SPMD patria v súčasnosti medzi najpoužívanejšie typy pasívnych vzorkovačov a detailne sa im venoval vo svojej práci napr. Esteve-Turrillas75. 3.3. Polárne organické chemické integračné vzorkovače Pasívne vzorkovače POCIS (Polar Organic Chemical Integrative Sampler) sú v princípe podobné vzorkovačom SPMD, používajú sa však na monitorovanie hydrofilných kontaminantov, ako sú napr. pesticídy, liečivá76, steroidné hormóny, herbicídy77 alebo antibiotiká. Tieto zlúčeniny sa dostávajú do vodného prostredia celosvetovo a u mnohých z nich bol pozorovaný efekt chronickej toxicity. Vzorkovač POCIS pozostáva z prijímajúcej fázy (sorbentu), ktorá je z oboch strán obklopená hydrofilnou mikroporóznou polyétersulfónovou membránou (pre vzorkovanie polárnych organických látok). Tá je upevnená medzi dvoma podpornými kruhmi. Zloženie prijímajúcej fázy (sorbentu) sa volí na základe charakteru látok, ktoré chceme monitorovať. Efektívna vzorkovacia plocha štandardne používaného systému POCIS predstavuje 41 cm2 na jeden vzorkovač o približnom priemere 10 cm (cit.78). Vzorkovanie prebieha iba z rozpustenej fázy, a teda umožňuje zistenie skutočne biologicky dostupného podielu. Tieto vzorkovače tiež pracujú na princípe integratívneho vzorkovania, sú v prostredí exponované počas viacerých 553 Chem. Listy 103, 548-558 (2009) Referát týždňov a poskytujú informáciu o časovo priemernej koncentrácii kontaminantu vo vodnom prostredí.79 3.4. Vzorkovače s membránou uzavretým sorpčným potahom Pasívne vzorkovače MESCO (Membrane-Enclosed Sorptive Coating) sa používajú na monitorovanie hydro-fóbnych organických polutantov. Medzi hlavné výhody patrí: malý rozmer, extrakcia a prekoncentrácia zvodného prostredia bez použitia rozpúšťadiel, bezstratová separácia prijímajúcej fázy a jednoduchá analýza v termodesorpčnej jednotke. Základ vzorkovača tvorí malá, približne 1,5 cm dlhá tyčinka pokrytá tenkou vrstvou polydimetylsiloxánu (PDMS), ktorá je umiestnená v dialyzačnom vrecku vyrobenom z regenerovanej celulózy, prípadne LDPE. Použitie PDMS ako prijímajúcej fázy je vhodné kvôli afinite kpolutantom, inertným vlastnostiam ako aj stabilite pri termodesorpcii80. Princípom vzorkovania je selektívny prestup látky z vodného prostredia a následná absorpcia na prijímajúcu fázu. Dialyzačná membrána je naplnená destilovanou vodou a uzatvorená na oboch koncoch. Pri expozícii sa používajú viaceré vzorkovače zoradené za sebou. Pri spracovaní vzoriek MESCO sa prijímajúca fáza vyberie z membrány, opláchne destilovanou vodou, vysuší a následne sa analyzuje obsah naakumulovaných látok pomocou termodesorpčnej GC/MS81. Táto metóda je kvôli nízkym stratám vhodná práve na stanovenie stopových koncentrácií vo vodnom prostredí. Podobne ako väčšina pasívnych vzorkovačov, aj vzorkovače MESCO prešli optimalizáciou. Prvý prototyp sa skladal z LDPE membrány uzavretej na oboch koncoch s vloženým SPME vláknom pokrytým PDMS, ktoré je vhodné pri monitorovaní nepolárných látok z vodného prostredia82. V ďalšej generácii, označovanej ako MESCO I (cit.83), bolo ako prijímajúca fáza použité miešadlo Twister™ (Gerstel, Múlheim/Ruhr, Germany) pokryté vrstvou PDMS, ktoré je používané na bezrozpúšťadlovú mikroex-trakciu. Zakladá sa na rovnakom princípe ako vlákno SPME, má však vyššiu extrakčnú kapacitu. Membrána je vyrobená z regenerovanej celulózy. Posledný typ, MESCO II, spája výhody vysokej kapacity prijímajúcej fázy druhej generácie so stabilitou LDPE membrány pôvodného typu vzorkovača84. Navyše bol pomerne drahý a krehký Twister™ nahradený silikonovou tyčinkou. Napriek malým rozmerom povrchu prijímajúcej fázy a objemu vzorkovača, citlivosť MESCO je porovnateľná s inými druhmi pasívnych vzorkovačov, pretože celé množstvo analytu obsiahnutého v PDMS sa prenesie do GC, kde je následne analyzované. Komplexnejšie je problematika vzorkovačov MESCO diskutovaná v literatúre16, ako aj príklady aplikácie v prostredí85. 3.5. Keramický dozimeter Keramický dozimeter patrí medzi pasívne vzorkovače, ktoré sú obzvlášť vhodné na dlhodobé monitorovanie kontaminantov vo vodnom prostredí a najčastejšie sa používajú na sledovanie kvality podzemných vôd. Skladá sa z keramickej tuby, ktorá predstavuje limitno-difúznu membránu a tuhého sorbentu vo forme guličiek (napr. Dowex Optipore L-493) ako prijímajúcej fázy. Tvarom pripomína rúrku s priemerom 15 mm, hrúbkou stien 1,5 mm o dĺžke najčastejšie 5-10 cm a pórmi 5-100 nm. Časovo vážené spriemernené koncentrácie namerané v podzemných vodách pomocou keramických dozimetrov veľmi dobre kore-lujú s hodnotami získanými často opakovanými bodovými odbermi86. Na monitorovanie stopových koncentrácií nepolárných látok, ako napr. PAH, sa používa ako prijímajúca fáza Amberlite IRA-743. Ide o iónovýmennú živicu na polystyrénovej báze s dostatočnou kapacitou viazať konta-minanty a dobrou zmáčavosťou. Výhodou tohto typu vzorkovača je možnosť dlhodobého použitia (90 dní) aj bez predchádzajúcej časovo náročnej kalibrácie87. Bolo dokázané88, že uvedená prijímajúca fáza (Amberlite IRA-743) je schopná udržať naakumulovaný kontaminant aj po premiestnení keramického dozimetra do deionizovanej vody na dobu 100 dní prakticky bez strát. Nedávno bol predstavený nový typ pasívneho vzorkovača, ktorý kombinuje jednoduchosť keramického dozimetra a možnosť biostanovenia. Bol označený ako „keramický toximeter"89 a obsahuje špeciálnu prijímajúcu fázu, Biosilon. Je upravená tak, aby bolo po naakumulovaní kontaminantu umožnené priľnúť bunkám stavovcov k jej povrchu a vyvolať biologickú odozvu. Na biostanove-nie sa využíva indukcia 7-etoxyresorufín-O-deetylázy v prítomnosti PAH. Táto metóda však nebola doteraz dostatočne preskúmaná a vyžaduje ďalšiu optimalizáciu. Monitorovaniu kontaminantov pomocou keramických dozimetrov sa bližšie venuje kapitola v knihe Passive Sam-pling Techniques in Environmental Monitoring16. 3.6. Pasívny difúzny vak PDB (Passive Diffusion Bag) patrí medzi pasívne vzorkovače pracujúce v rovnovážnej oblasti, pričom rovnováha je dosiahnutá do 24 h v prípade vzduchom plnených vzorkovačov (PVD, Passive Vapour Diffusion) a do 48 h v prípade vodou plnených vzorkovačov90. Približný rozmer štandardného PDB je 61 cm na dĺžku o priemere 32 mm. Primárne sa používa na získavanie informácií o koncentrácii nepolárných prchavých zlúčeninách (VOC) v podzemných vodách. Práve pri monitorovaní prchavých zlúčenín treba postupovať veľmi opatrne, keďže už aj pri samotnom odbere dochádza k stratám. PDB sú taktiež vhodné na dlhodobý monitoring nálezísk podzemnej vody a sledovanie prítomnosti predovšetkým trichlóreténu (TCE), benzénu, toluénu, etylbenzénu a xylénu (BTEX). Základ zariadenia tvorí semipermeabilná membrána, ktorá obsahuje deionizovanú vodu (prípadne vzduch). Ak je daný pasívny vzorkovač v kontakte so vzorkovacím médiom, nastáva difúzia kontaminantov cez semipermea-bilnú membránu do deionizovanej vody. Po ukončení expozície sa voda spolu s polutantmi, ktorá sa dostala cez 554 Chem. Listy 103, 548-558 (2009) Referát polopriepustnú membránu do vzorkovača, vypustí do nádoby na neskoršiu analýzu. PDB sa týmto úkonom regeneruje a vzorkovač je pripravený na ďalšiu expozíciu. Vhodnosť aplikácie PDB pri monitorovaní kontaminantov v podzemných vodách a porovnanie s klasickými technikami bola potvrdená experimentálne91. Semipermeabilná membrána slúži zároveň aj ako limitno-difúzna bariéra voči prestupu látky zvodného prostredia. Vhodnou voľbou materiálu je možné dosiahnuť selektivitu vzhľadom na monitorovanú skupinu kontaminantov a dokonca je možné sledovať prítomnosť anorganických zložiek92. Medzi najčastejšie používané patria dialyzačné membrány z regenerovanej celulózy a LDPE (Low Density Polyethylene) membrány. 3.7. Mikroextrakcia na tuhú fázu Pri mikroextrakcii na tuhú fázu (SPME, Solid phase microextraction)93 ide o jednoduchú extrakčnú metódu. V tomto prípade je extrakčným médiom tenké křemenné vlákno potiahnuté tenkou vrstvou polyméru, často PDMS (polydimetylsiloxán). Extrakčnú rovnováhu možno dosiahnuť, v závislosti od fyzikálno-chemických vlastností látok, už v priebehu tridsiatich minút a množstvo analytu, ktoré je naviazané na vlákne sa analyzuje plynovou alebo vyso-koúčinnou kvapalinovou chromatografiou. Napriek krátkemu času potrebnému na dosiahnutie rovnováhy, je možné zaradiť metódy SPME medzi pasívne vzorkovanie, najmä kvôli rovnakému princípu voľného prestupu látky z prostredia do prijímajúcej fázy. Táto metóda umožňuje monitorovanie hydrofóbnych chemikálií, vrátane P AH, PCB, chlórovaných pesticídov a fenolov. SPME je možné použiť aj na monitorovanie pôdy. Priame porovnanie s koncentráciou analytu (PAH) v dážďovkách dokázalo, že pri technike SPME skutočne dochádza k akumulácii voľne dostupnej frakcie94. Výhoda SPME spočíva v rýchlom dosiahnutí rovnováhy, relatívne jednoduchej analýze, nízkych nákladoch, dobrej korelácii so živými organizmami a nenáročnej manipulácii. Vlákna SPME pokryté vrstvou polyakry-látu je tiež možné použiť na simuláciu javu bioakumulácie a odhad akútnej toxicity na živý organizmus95. Zatiaľčo mnoho aplikácií SPME sa usiluje o najvyššiu možnú efektívnosť extrakcie, nd-SPME (negligible depletion SPME - mikroextrakcia na tuhú fázu so zanedbatelným večerpaniem) predstavuje špecifickú aplikáciu na merania voľnej koncentrácie testovanej vzorky, pričom sa extrahuje len nepatrné množstvo analytu, čo môže predstavovať problém pri celkovej kvantifikácii. Ide o novú metódu, ktorá umožňuje stanovovať voľne dostupnú frakciu kontaminantu, ako aj zisťovanie rozdeľovacích koeficientov. To sa dá využiť pri skríningu a identifikácii bioa-kumulujúcich zlúčenín v prostredí96. Technika nd-SPME je detailnejšie opísaná v literatúre97. Metóda SPME sa často používa aj pri sledovaní úrovne znečistenia sedimentov, pričom ich toxické vlastnosti sú priamo vztiahnuteľné na množstvo biodostupnéhu podielu kontaminantu obsiahnutého v pórovej vode98. Pri monitorovaní v sedimentoch sa používa termín matrix- SPME (matricová SPME) , pretože pri extrakcii z prostredia sa na dosiahnutí rovnováhy zúčastňuje matrica sedimentu ako celok a výsledky je možné použiť na výpočet fugacitných koeficientov11. Sklenené vlákno je pokryté 15 um hrubou vrstvou PDMS, umiestni sa do prostredia až po dosiahnutie rovnováhy (podľa charakteru analytu 1 až 30 dní) a následne analyzuje plynovou chromatografiou. 4. Záver Technológia pasívneho vzorkovania, uvedené možnosti použitia a jej implementácia do praxe poukazujú na značný potenciál tejto inovatívnej metódy pri monitorovaní kvality životného prostredia. Napriek tomu, že prešla dlhoročným vývojom, metóda pasívneho vzorkovania sa stále rozvíja a zdokonaľuje. Výskum sa v tejto oblasti sústreďuje na elimináciu možných environmentálnych faktorov ovplyvňujúcich kinetické parametre vzorkovačov (teplota, hydrodynamické podmienky, bioznečistenie), hľadanie materiálov vhodných pre vzorkovanie nových skupín látok a spojenie chemickej a toxikologickej odozvy na prítomnosť kontaminantu v prostredí. Medzi významné výhody v porovnaní s konvenčnými prístupmi vzorkovania patrí jednoduchosť, nízka ekonomická náročnosť, možnosť použitia bez vonkajšieho zdroja energie, poskytovanie informácie o časovo váženom priemere koncentrácie a detekcia aj ultrastopových hladín kontaminantu. Samotným cieľom pasívneho vzorkovania nie je nahradiť klasické bodové odbery, ale poskytnúť dodatočné informácie ku stavu znečistenia životného prostredia. LITERATÚRA 1. Stackelberg P. E., Furlong E. T., Meyer M. T., Zaugg S. D., Henderson A. K., Reissman D. B.: Sci. Total Environ. 329, 99 (2004). 2. Stoeppler M. (ed.): Sampling and Sample Preparation: Practical Guide for Analytical Chemists. Springer Verlag, Berlin 1997. 3. Batley G. E.: Mar. Pollut. Bull. 39, 31 (1999). 4. Lyn J. A., Ramsey M. H., Fussell R. J., Wood R.: Analyst 128, 1391 (2003). 5. Gorecki T., Namiesnik J.: Trends Anal. Chem. 21, 276 (2002). 6. Paschke A.: Trends Anal. Chem. 22, 78 (2003). 7. Gordon C. S, Lowe, J. I.: US Patent 1.644 014, 1927. 8. Fowler W. K.: Int. Lab. 14, 80 (1982). 9. Kot A., Zabiegala B., Namiesnik J.: Trends Anal. Chem. 19, 446 (2000). 10. Lu Y., Wang Z., Huckins J.: Aquat. Toxicol. 60, 139 (2002). 11. Mayer P., Tolls J., Hermens J. L. M., Mackay D.: Environ. Sci. Technol. 37, 185 (2003). 12. Stuer-LauridsenF.: Environ. Pollut. 136, 503 (2005). 13. Namiesnik J., Zabiegala B., Kot-Wasik A., Partyka M., Wasik A.: Anal. Bioanal. Chem. 381, 279 (2005). 14. Vrana B., Allan I. J., Greenwood R., Mills G. A., Do- 555 Chem. Listy 103, 548-558 (2009) Referát miniak E., Svensson K., Knutsson J., Morrison G.: Trends Anal. Chem. 24, 845 (2005). 15. Mills G. A., Vrana B., Allan I. J., Alvarez D. A., Huc-kins J. N., Greenwood R.: Anal. Bioanal. Chem. 387, 1153 (2007). 16. Greenwood R., Mills G. A., Vrana B. (ed.): Passive Sampling Techniques in Environmental Monitoring. Comprehensive Analytical Chemistry Series. Elsevier, Amsterdam 2007. 17. Seethapathy S., Gorecki T., Li X.: J. Chromatogr., A 1184, 234 (2008). 18. Dixon W., Smyth G. K., Chiswell B.: Water Res. 33, 971(1999). 19. Bloch H.: Houille Blanche 1, 60 (2003). 20. Roig B., Valat C, Allan I. J., Greenwood R., Berho C, Guigues N., Mills G. A., Ulitzur N.: Trends Anal. Chem. 26, 21A (2007). 21. Coquery M., Morin A., Bécue A., Lepot B.: Trends Anal. Chem. 24, 117 (2005). 22. Kočí V., Ocelka T., Kochánková L.: Vodní Hospodářství 11, 331 (2001). 23. http://wwwaux.cerc.cr.usgs.gov/SPMD/SPMD-Tech_Tutorial.htm, stiahnuté 5. september 2008. 24. Alvarez D. A., Stackelberg P. E., Petty J. D., Huckins J. N., Furlong E. T., Zaugg S. D., Meyer M. T.: Che-mosphere 61, 610 (2005). 25. Koester C. J., Esser B. K., Simonich S. L.: Anal. Chem. 75, 2813 (2003). 26. Namiesnik J., Zabiegala B., Kot-Wasik A., Partyka M., Wasik A.: Anal. Bioanal. Chem. 381, 279 (2005). 27. Allan I. J., Vrana B., Greenwood R., Mills G. A., Roig B., Gonzalez C: Talanta 69, 302 (2006). 28. Längsten W. J., Spence S. K., v knihe: Metal Specia-tion and Bioavailability in Aquatic Systems (Tessier A., Turner D. R., ed.), str. 407. J. Wiley, Chichester 1995. 29. Antonelli M. L, Ercole P., Campanella, L.: Talanta 45, 1039 (1998). 30. Baldwin I. G., Kramer K. J. M. (ed.): Biological Early Warning Systems (BEWS). CRC Press, Boca Raton 1994. 31. Gerhardt A., De Bisthoven L. J., Soares A. M. V.: Environ. Sei. Technol. 39, 4150 (2005). 32. Küster E., Dorusch F., Vogt C, Weiss H., Altenburger R.: Biosens. Bioelectron. 19, 1711 (2004). 33. Gruber D. S., Diamond L. M. (ed.): Automated Biomonitoring: Living Sensors as Environmental Monitors. Ellis Horwood, Chichester 1988. 34. Mora M. A., Wainwright S. E.: Rev. Environ. Contain. Toxicol. 158, 1 (1996). 35. Beliaeff B., Bocquené G.: Mar. Environ. Res. 58, 239 (2004). 36. Zorita L, Apraiz L, Ortiz-Zarragoitia M., Orbea A., Cancio L, Soto M., Marigómez L, Cajaraville M. P.: Environ. Pollut. 148, 236 (2007). 37. Zorita L, Ortiz-Zarragoitia M., Apraiz L, Cancio L, Orbea A., Soto M., Marigómez L, Cajaraville M. P.: Environ. Pollut. 153, 157(2008). 38. Zhao W., Ouyang G., Alaee M., Pawliszyn J.: J. Chromatogr., A 1124, 112 (2006). 39. Vrana B., Mills G. A., Kotterman M., Leonards P., Booij K., Greenwood R.: Environ. Pollut. 145, 895 (2007). 40. http://wwwaux.cerc.cr.usgs.gov/SPMD/, stiahnuté 5. september 2008. 41. Vrana B., Allan I. J., Greenwood R., Mills G. A., Do-miniak E., Svensson K., Knutsson J., Morrison G.: Trends Anal. Chem. 24, 845 (2005). 42. Huckins J. N., Manuweera G. K., Petty J. D., Mackay D. , Lebo J. A.: Environ. Sei. Technol. 27, 2489 (1993). 43. Ouyang G., Zhao W., Bragg L., Qin Z., Alaee M., Pawliszyn J.: Environ. Sei. Technol. 41, 4026 (2007). 44. Kukkonen J. V. K., Landrum P. F.: Aquat. Toxicol. 42, 229 (1998). 45. Petty J. D., Jones S. B., Huckins J. N., Cranor W. L., Parris J. T., McTague T. B., Boyle T. P.: Chemos-phere 41, 311 (2000). 46. Barber L. B., Keefe S. H., Antweiler R. C, Taylor H. E. , Wass R. D.: Environ. Sei. Technol. 40, 603 (2006). 47. Zhang Z., Hibberd A., Zhou J. L.: Anal. Chim. Acta 607, 37 (2008). 48. Ouyang G., Zhao W., Alaee M., Pawliszyn J.: J. Chromatogr., A 1138, 42 (2007). 49. Booij K., Sleiderink H. M., Smedes F.: Environ. Toxicol. Chem. 17, 1236 (1998). 50. Huckins J. N., Petty J. D., Prest H. F., Orazio C. E., Gale R. W.: 18th Annual SET AC Meeting, San Francisco, CA, Nov 16-20, 1997, Book of Abstracts (bez editora), str. 206. 51. Ouyang G., Chen Y., Pawliszyn J.: Anal. Chem. 77,7319 (2005). 52. Vrana B., Schüürmann G.: Environ. Sei. Technol. 36, 290 (2002). 53. Vrana B., Mills G. A., Dominiak E., Greenwood R.: Environ. Pollut. 142,333 (2006). 54. Huckins J. N., Petty J. D., Booij K. (ed.): Monitors of Organic Contaminants in the Environment: Semipermeable Membrane Devices. Springer, Berlin 2006. 55. Richardson B. J., Lam P. K. S., Zheng G. J., McClel-lanK. E., De Luca-Abbott S. B.: Mar. Pollut. Bull. 44, 1372 (2002). 56. de la Cal A., Kuster M., de Alda M. L., Eljarrat E., Barceló D.: Talanta 7<5, 327 (2008). 57. Aguilar-Martínez R., Palacios-Corvillo M. A., Greenwood R., Mills G. A., Vrana B., Gómez-Gómez M. M.: Anal. Chim. Acta 618, 157 (2008). 58. Schäfer R. B., Paschke A., Vrana B., Müller R., Liess M.: Water Res. 42, 2707 (2008). 59. Vermeirssen E. L. M., Asmin J., Escher B. L, Kwon J. H., Steimen L, Höhender J.: J. Environ. Monitor. 10, 119 (2008). 60. Gunold R., Schäfer R. B., Paschke A., Schüürmann G., Liess M.: Environ. Pollut. 755, 52 (2008). 61. Allan I. J., Knutsson J., Guigues N., Mills G. A., Fou-illac A. M., Greenwood R.: J. Environ. Monit. 9, 672 556 Chem. Listy 103, 548-558 (2009) Referát (2007) . 62. Kingston J. K, Greenwood R., Mills G. A., Morrison G. M., Persson L. B.: J. Environ. Monit. 2, 487 (2000) . 63. Lobpreis T., Vrana B., Dominiak E., Dercova K, Mills G. A., Greenwood R.: Environ. Pollut. 153, 706 (2008) . 64. Vrana B., Mills G., Greenwood R., Knutsson J., Sven-sson K., Morrison G.: J. Environ. Monit. 7, 612 (2005) . 65. Huckins J. N., Tubergen M. W., Manuweera G. K.: Chemosphere 20, 533 (1990). 66. Charlestra L., Courtemanch D. L., Amirbahman A., Patterson H.: Chemosphere 72, 1171 (2008). 67. Ellis S. G., Booij K, Kaputa M.: Chemosphere 72, 1112 (2008). 68. Ke R., Li J., Qiao M., Xu Y., Wang Z.: Arch. Environ. Con. Tox. 53,313 (2007). 69. Liao L. B., Xiao X. M.: Chemosphere 64, 1592 (2006) . 70. Booij K, Van Drooge B. L.: Chemosphere 44, 91 (2001) . 71. Huckins J. N., Petty J. D., Orazio C. E., Lebo J. A., Clark R. C, Gibson V. L., Gala W. R., Echols K. R.: Environ. Sei. Technol. 33, 3918 (1999). 72. Chec E., Podgörska B., Wegrzyn G.: Environ. Monit. Assess. 140, 83 (2008). 73. Verweij F., Booij K, Satumalay K., Van Der Molen N., VanDerOostR.: Chemosphere 54, 1675 (2004). 74. Vrana B., Paschke A., Popp P.: J. Environ. Monit. 3, 602 (2001). 75. Esteve-Turrillas F. A., Yusa V., Pastor A., de la Guar-dia M.: Talanta 74, 443 (2008). 76. Togola A., Budzinski H: Anal. Chem. 79, 6734 (2007) . 77. Mazzella N., Dubernet J. F., Delmas F.: J. Chroma-togr., A 1154, 42 (2007). 78. Alvarez D. A., Petty J. D., Huckins J. N., Jones-Lepp T. L., Getting D. T., Goddard J. P., Manahan S. E.: Environ. Toxicol. Chem. 23, 1640 (2004). 79. MacLeod S. L., McClure E. L., Wong C. S.: Environ. Toxicol. Chem. 26, 2517 (2007). 80. Baltussen E., Cramers C. A., Sandra P. J. F.: Anal. Bioanal. Chem. 373, 3 (2002). 81. Montero L., Popp P., Paschke A., Pawliszyn J.: J. Chromatogr., A 1025, 17 (2004). 82. Paschke A., Popp P.: J. Chromatogr., A 999, 35 (2003). 83. Vrana B., Popp P., Paschke A., Schüürmann G.: Anal. Chem. 73, 5191 (2001). 84. Wennrich L., Vrana B., Popp P., Lorenz W.: J. Environ. Monit. 5, 813 (2003). 85. Paschke A., Schwab K, Brummer J., Schüürmann G., Paschke H, Popp P.: J. Chromatogr., A 1124, 187 (2006). 86. Martin H, Patterson B. M., Davis G. B., Grathwohl P.: Environ. Sei. Technol. 37, 1360 (2003). 87. Bopp S., Weiß H, Schirmer K.: J. Chromatogr., A 7 072, 137 (2005). 88. Martin H, Piepenbrink M., Grathwohl P.: Proc. Soc. Anal. Chem. 6,68 (2001). 89. Bopp S. K, Mclachlan M. S., Schirmer K.: Environ. Sci. Technol. 41, 6868 (2007). 90. Vroblesky D. A., Campbell T. R.: Adv. Environ. Res. 5, 1 (2001). 91. Sorel D., Longino B. L., Warner S. D., Hamilton L. A.: Proceedings of the Third International Conference on Remediation of Chlorinated and Recalcitrant Compounds, Monterey, California, May 20-23, 2002, Book of Abstracts (Gavaskar A. R., Chen A. S. C, ed.), str. 2547. 92. Ehlke T. A., Imbrigiotta T. E., Dale J. M.: Ground Water Monit. R. 24, 53 (2004). 93. Eisert R., Levsen K: J. Chromatogr., A 733, 143 (1996). 94. Jonker M. T. O., Van Der Heijden S. A., Kreitinger J. P., Hawthorne S. B.: Environ. Sci. Technol. 41, 7472 (2007). 95. Verbruggen E. M. J., Vaes W. H. J., Parkerton T. F., Hermens J. L. M.: Environ. Sci. Technol. 34, 324 (2000). 96. Hermens J. L. M., Freidig A. P., Ramos E. U., Vaes W. H. J., Van Loon W. M. G. M., Verbruggen E. M. J., VerhaarH. J. M.: ACS Symp. Ser. 773, 64 (2001). 97. Heringa M. B., Hermens J. L. M.: Trends Anal. Chem. 22, 575 (2003). 98. Bondarenko S., Spurlock F., Jianying G. A. N.: Environ. Toxicol. Chem. 26, 2587 (2007). 99. Mayer P., Vaes W. H. J., Wijnker F., Legierse K. C. H. M., Kraaij R., Tolls J., Hermens J. L. M.: Environ. Sci. Technol. 34, 5177 (2000). T. Lobpreis", B. Vrana\ and K. Dercova" ("Department of Biochemical Technology, Faculty of Chemical and Food Technology, Slovak University of Technology, Bratislava, b Water Research Institute, Slovak National Water Reference Laboratory, Bratislava, Slovak Republic): Innovative Approach to Monitoring Organic Contaminants in Aqueous Environment Using Passive Sampling Devices The aim of this review is to introduce new methods of monitoring organic contaminants in aqueous environment. Passive sampling devices are able to overcome many of the limitations associated with conventional spot sampling of waters. They work in the integrative mode allowing the estimation of time-weighted average concentrations of contaminants in water, soil, sediments or air. Unlike most monitoring methods, passive samplers measure the dissolved, i.e. bioavailable fraction of water pollutants. In addition, they are able to effectively concentrate the pollutants that are present in trace amounts. The passive sampling devices should not replace conventional sampling; they provide additional information on the environment pollution at a reasonable cost. 557 Chem. Listy 103, 548-558 (2009) Referát 61. ZJAZD CHEMIKOV 7.-11. september 2009 Vysoké Tatry, Tatranské Matliare Vážení priatelia, v mene organizačného a programového výboru, sponzorov a čestného predsedníctva je nám potešením Vás pozvať na náš ďalší spoločný zjazd chemikov a to opäť do Vysokých Tatier. Centrom zjazdu bude opäť ho-telový komplex Hutník situovaný v Tatranských Mat-liaroch. Určite ste si všimli, že postupne budu-jeme tradíciu našich tatranských zjaz-dov. Popri rôznych pozvaných prednášateľoch (PP) sa môžete tešiť na výber (po dvoch nositeľoch Nobelovej ceny) zaujímavého plenárneho prednášateľa. Novinkou bude tématický večer venovaný 80 rokom SChS a Kurz aplikácií kvantovej chémie. Organizačný výbor Dušan Velič - predseda Monika Aranyosiová - výkonný tajomník Miroslav Michalka - technická podpora Zuzana Hloušková - hospodár Milan Drábik - vedecký tajomník Pavel Drašar - vedecký tajomník Programový výbor Prof. Ing. Dr. Jozef Tomko, DrSc. (SChS) Doc. Ing Viktor Milata, CSc. (SChS) Ing. Miloš Revús (SSPCH, BA) RNDr. Dalma Gyepesová, CSc. (SChS) Doc. RNDr. Marta Sališová, CSc. (SChS) Prof. Ing. Vlasta Brezová, DrSc. (SChS) Ing. Mária Omastová, PhD. (SChS) Ing. Marián Janek, PhD. (SChS) RNDr. Jozef Tatiersky, PhD. (SChS) Mgr. Katarína Javorová>5ChS) ^ Prof Ing. Ján Labuda, DrSc. (STU, BA) Ing. Michal Korenko, PhD. (SAV, BA) Prof Ing. StanislavBiskupič,DrSc.(SWBA) Prof RNDr. Jozef Cársky, CBc. rtjK^A) Ing. Milan Vrška,CSc. (fTIfríjfl) Prof RNDr. Dušan Kaniansky, DrSc. (UK, BA) Doc. RNDr. Jozef Kuruc, PhD. (UK, BA) Prof Ing. Milan Remko, DrSc. (UK, BA) Prof Ing. Lubor Fišera, DrSc. (STU, BA) Doc. Ing. Dušan Berek, Drsl. (SAWBÄÍ Doc. Ing. Stefan Schmidt, Php. (STU\BA) Ing. Ján Hirsch, DrSc. (3AV, BA)\ Prof Ing. Peter Šimon, DrSij. (STU, BÄl Prof Ing. Vasil Koprda, DrSc. (STU, BA) Doc.Ing. Ján Reguli, PhDSXU, TrV^ Doc. RNDr. JánBenko, CSc. (bííJJA) C Doc. RNDr. Martin Pútala, PhD. (UK"B^1 4 Doc. RNDr. Taťána Gondová, CSc. (UPJŠ, KB*. Doc. RNDr. Mária Řeháková, CSc. (UPJŠ, KE) Doc. RNDr. Renáta Oriňáková, CSc. (TU, KE) Prof RNDr. Naděžda Števulová, PhD. (TU, KE) RNDr. Slávka Hamuľáková, PhD. (UPJŠ, KE) Doc. RNDr. Mária Ganajová, CSc. (UPJŠ, KE) Doc. RNDr. Magdaléna Bálintová, PhD. (TU, KE) Ing. Elena Kulichová (Nováky) Sekcie: 1. Analytická a fyzikálna chémia 2. Anorganická a materiálová chémia 3. Organická chémia a polyméry 4. Vyučovanie a história chémie 5. Životné prostredie a biotechnológia 6. CHEMPROGRESS Konferenčný poplatok: účastník, člen* 300 € študent, doktorand, člen* 200 € dôchodca, člen* 250 € príplatok za nečlena 100 € príplatok za jednolôžkovú izbu 150 € sprevádzajúca osoba 250 € * ASChFS, AČChS, SChS Poplatok zahŕňa: konferenčné materiály, ubytovanie v dvojposteľovej izbe s plnou penziou (od večere 7. 9. po obed 11. 9.), uvítací večierok, vínny a pivný večer, přestávkové občer-stvenie, slávnostný večierok, plaváreň, miestny poplatok, poistenie nákladov na zásah Horskej záchrannej služby. Termíny: Registrácia do 1. júna 2009 ^tetba dol. júla 2009 ^BJrt?a1*L do 1. júla 2009 Registrácia po\ (TÉ009 pri zaplnenej ubytovacej kapacite, bude navýšená o 100 € na zabezpečenie náhradného ubytovania. Formy prezentácie: ťoster (8O0 milí sirka* f 000 mm dĺžka) Súťaže formou komentovaných posterov študenti, doktorandi (ceny: 150, 100, 50 €) n i m ijf n rnn 200,100€) Prednáška Formát MS Powerpoint pozvaná prednáška 40 min. + 10 min. diskusia prednáška 20 min. + 5 min. diskusia Panelová diskusia ako záver zjazdu Abstrakt v časopise ChemZi 5/9 2009 Publikácia v nasledujúcich číslach ChemZi Srŕíakt: Slovenská chemická spoločnosť, Radlinského 9/1111, 812 37 Bratislava, fax: +421/2/52495205 e-mail: zjazd.chemikov@gmail.com web: http://www.schems.sk/61zjazd 558 Príloha 17 Allan I. J., Booij K., Paschke A., Vrana B., Mills G. a, and Greenwood R., Short-term exposure testing of six different passive samplers for the monitoring of hydrophobic contaminants in water., J. Environ. Monit., 2010,12, 696-703. PAPER www.rsc.org/jem | Journal of Environmental Monitoring Short-term exposure testing of six different passive samplers for the monitoring of hydrophobic contaminants in water| Ian J. Allan,*" Kees Booij,* Albrecht Paschke,c Branislav Vrana/ Graham A. Millse and Richard Greenwood^ Received 12th October 2009, Accepted 18th December 2009 First published as an Advance Article on the web 14th January 2010 DOI: 10.1039/b921326k Passive sampling devices are increasingly relied upon for monitoring non-polar organic contaminants in water. While many types of devices are available they have seldom been evaluated alongside each other. We tested six passive sampling devices namely: Chemcatcher, two modified versions of the membrane enclosed sorptive coating (MESCO I (m) and MESCO II), silicone rod and strip and semipermeable membrane device (SPMD). Samplers spiked with a range of performance reference compounds (PRCs) were exposed (5 days) in a continuous flow-through tank using Meuse river water fortified with fluctuating concentrations (20-700 ng Lr1) of polycyclic aromatic hydrocarbons, polychlorinated biphenyls, hexachlorobenzene and p,p'-DT>E. Dissipation rates of PRCs appeared to provide reliable information on exchange kinetics even under these short-term exposure conditions. They accounted for differences between masses of contaminants accumulated by replicate samplers, indicating that the variability between replicates was in part due to differences in water turbulences and hence boundary layer thickness. In this system, resistances in the membrane and boundary layers are likely to be in the same order of magnitude for PRCs. Sampler performance was evaluated by comparing masses accumulated in the devices only for analytes for which uptake was linear (integrative) and limited by transport across the boundary layer. Consistent data were obtained across the range of samplers despite their different configurations, and the analysis being conducted in three separate laboratories. The pattern in analyte masses accumulated by Chemcatcher and MESCO II data could be explained by the extraction and analysis being conducted only on the receiving phase of the samplers and a significant impact of the lag-phase prior to obtaining a steady flux of contaminants across the polyethylene membranes. "Norwegian Institute for Water Research, Gaustadalleen 21, NO-0349 Oslo, Norway. E-mail: ian.allan@niva.no; Fax: +47 22185200; Tel: +47 22185100 b Royal Netherlands Institute for Sea Research, PO Box 59,1790 AB Texel, The Netherlands 'Department of Ecological Chemistry, UFZ-Helmholtz Centre for Environmental Research, Permoserstrasse 15, 04318 Leipzig, Germany dNational Water Reference Laboratory for Slovakia, Water Research Institute, Nabr. arm. gen. L. Svobodu 5, 81249 Bratislava, Slovakia 'School of Pharmacy and Biomedical Sciences, University of Portsmouth, White Swan Road, Portsmouth, UK POl 2DT fSchool of Biological Sciences, University of Portsmouth, King Henry I Street, Portsmouth, UK POl 2DY f Electronic supplementary information (ESI) available: Lists of analytes used in this study and of performance reference compounds spiked into the different passive samplers. See DOf: 10.1039/b921326k Introduction Passive sampling is a monitoring technique that is increasingly being used to measure dissolved concentrations of hydrophobic contaminants in water.1-3 Accumulation of contaminants relies on diffusion of chemicals present in the water phase into the device as a result of a difference in chemical activity of the contaminant in water and that in the sampler. Devices may be simple polymeric membranes such as low density polyethylene (LDPE) membrane or silicone strips,4 or more complex designs such as the Chemcatcher5 composed of a LDPE membrane superimposed onto a receiving phase («-octanol-loaded C18 Empore® disk), SPMD6 with triolein-filled LDPE tubing or the MESCO samplers with a silicone rod or Gerstel Twister bar Environmental impact Passive sampling devices have the potential to respond to many of the challenges set by legislative texts such as the EU Water Framework Directive regarding the monitoring of aquatic micropollutants. Advantages include measurement of the freely dissolved concentration of contaminants in water and continuous monitoring over periods of time. Since different types of devices exist, the equivalence of the data they provide must be demonstrated and their working principle understood before they can be widely applied within monitoring programmes. This paper is one of few studies comparing the performances of number of passive samplers for organic compounds and demonstrating the consistency of the data they provide. Useful insights in the working principles of the some of the samplers are also given. 696 I J. Environ. Monit., 2010, 12, 696-703 This journal is © The Royal Society of Chemistry 2010 receiving phase enclosed in various types of membrane.7-9 The mass transfer of contaminants into these samplers is dependent on both the characteristics of the device and the physico-chemical properties of the analytes being sequestered.1 Mass transfer of pollutants into the samplers is a multi-step phenomenon involving (i) transport of the chemical across the water boundary layer present at the surface of the sampler, (ii) transfer across the membrane and/or receiving phase layer of the sampler and in some cases (iii) transport across the biofilm layer that develops as a result of extended exposure in natural waters.1 The overall resistance to mass transfer (l/k0) into a sampler can be described as the sum of resistances in the water (<5W/Z>W), membrane (ôM/KMWDM) and biofilm (ôB/KBWDB): 1 _ dw ôM | <3B with k0 the overall mass transfer coefficient (m s-1), KMW (L L-1) the membrane-water (or sampler-water when using single-phase polymeric samplers) partition coefficient, KBW the biofilm-water partition coefficient, <5W, <5M and <5B the boundary, membrane and biofilm layer thicknesses (m), and Dw, DM and DB (m2 s~') analyte diffusion coefficients in water, membrane and biofilm layers, respectively. It has been shown that, depending on exposure conditions (water temperature and turbulence), resistance to mass transfer is generally influenced more by membrane-side processes for analytes with relatively low log Kow and more by transport across the boundary layer for those with high log KoW values.10,11 The threshold partition coefficient separating these two mass transfer processes is typically in the region of log 3.5—5.0.10-12 Integrative monitoring can be achieved if the exposure duration is kept well below the time required for the concentration in the sampler to reach equilibrium with the dissolved concentration in water.1 The amount of analyte absorbed by passive samplers may be represented by a first-order kinetic approach to equilibrium: m = Ksw KCW[1 - exp(-fcer)] (2) where m is the amount of analyte absorbed (ng), Ksw the sampler-water partition coefficient (L Lr1), K the volume of the sampler (L), ke the exchange rate constant (d~'), t the exposure time (d) and Cw the analyte concentration (ng L-1). ke is given by: Ksw V KSw V where A is the surface area of the sampler (cm2), kQ the overall mass transfer coefficient (m s-1) and Rs the analyte uptake rate (L d~'). Rs is the analyte specific equivalent volume of water cleared by the sampler per unit of time, and is needed to calculate analyte concentrations in the water column. First-order dissipation kinetics of PRCs, non-naturally occurring chemicals spiked into the sampler prior to deployment, can be used to estimate ke, the exchange rate constant (eqn (4)): rapRc = wo.prc exp(-kj) (4) where oto.prc and mPRC are masses of PRC in the sampler prior to and following exposure, respectively. When PRC dissipation is complete (or close to), then equilibrium between the concentration of analytes with similar log Kow in the sampler and that in water may be assumed. However, little (or negligible) PRC dissipation indicates that the uptake is in the linear phase. The boundary between these two regimes is generally found for PRCs with log KoW of 4.0-5.0 for typical exposures of up to several weeks. Many sources of error contribute to the uncertainties in the measurement of Cw obtained using integrative passive samplers. These arise from the reproducibility in sampler fabrication, PRC spiking, the accuracy of Ksw values (eqn (2)), the accuracy of PRC measurements and potential effects of biofouling and DOC.13,14 For highly hydrophobic compounds whose uptake is generally under boundary layer-control and for which data on dissipation of PRC analogues are not commonly available, further uncertainty arises from the extrapolation of their uptake rates from data for less hydrophobic PRCs.6,11,15 The application of a range of different passive sampling devices with distinctly different working principles, and inter-laboratory variation in sampler processing and analysis are likely to increase the overall uncertainty in these measurements.10 Further variability may also result from the quantification of analyte masses in extracts from passive samplers close to the analytical limits of detection. The present study assessed the performance of six different passive samplers for the measurement of hydrophobic organic contaminant concentrations in water. Samplers were exposed over a short period (five days) in a tank with a through flow of natural Meuse river water fortified with a range of PAHs, PCBs and pesticides. The concentrations used were made to fluctuate over the exposure period, reaching high levels at the peaks, in order to simulate a pollution event. Contaminant concentrations higher than those commonly found in surface waters were used in order to extend the number of analytes that were accumulated by all of the types of samplers tested. The evaluation of the performances of the different sampler designs compared dissipation rates of PRCs from the various samplers, and masses of analytes absorbed by the samplers normalised to sampler surface area. As this comparative study involved a number of laboratories, it provided a realistic evaluation of the performance of the samplers since in normal practice the overall uncertainty of measurements will include inter-laboratory variability. Materials and methods Experimental procedure for sampler exposure The study was undertaken in Eijsden (The Netherlands) for a period of 5 days in April 2005. This experiment was designed to expose simultaneously six different types of passive sampling device to Meuse river water fortified with various PAHs, PCBs, hexachlorobenzene and p,p'-DDE (standards purchased from Qmx Ltd, UK). A full list of compounds spiked into the river water for this test and/or analysed for can be found in the ESIf (Table SI-1). The system consisted of a custom-made stainless steel tank designed to hold 200 L of river water and to house a 27 cm diameter Teflon® carousel composed of five platforms for exposure of the various passive sampling devices (Fig. 1). The carousel was operated using an electrical overhead stirrer with a rotation speed set to 30 rpm for the entire duration of the experiment. This system allowed uniform contaminant This journal is © The Royal Society of Chemistry 2010 J. Environ. Monit., 2010, 12, 696-703 | 697 Overhead st rrer River water Fig. 1 Diagram of the test tank used for the exposure of the various types of passive sampling devices. concentrations in water throughout the container for the duration of the experiment. Samplers were exposed using a flow-through system and fresh Meuse river water was pumped into the tank at the rate of 10 L fr1 to ensure that the overall removal of test analytes from the water phase by the samplers was negligible. The spiking solution was prepared by dissolving PAHs, PCBs, hexachlorobenzene and p,p'-DDE into methanol. A Watson-Marlow peristaltic pump and pre-cleaned silicone tubing were used to deliver the methanolic solution to the test tank (0.7 mL hr'). The volumetric flow of methanol was kept low so that the concentration of methanol in water was negligible ( -4.4 -4.8 4 Chemcatcher 0 MESCO II □ Silicone rod A Silicone strip 0 SPMD A A A A Fig. 3 First-order PRC dissipation rates normalised to sampler volume to surface area ratio (keVIA) for five samplers (A) and overall mass transfer coefficients, k0, for PRCs (B). Note: no PRCs were used with MESCO I (m) samplers. samplers for hydrophobic organic compounds, and the robustness of the approach since all of the devices are functioning in a similar manner and yielding comparable results. However, the range of polarity of pollutants for which the various samplers can be used in an integrative manner in practice varies, and is determined by the time taken to approach equilibrium in relation to the required length of field deployment. Half-time to equilibrium (r50) was calculated to estimate thresholds between linear and non-linear uptake phases. t50 values for analytes with log 7f0w < 4.5-4.6 were under 7 days for most samplers (based on PRC data). Half-time to equilibrium increased in the order silicone rod < SPMD < silicone strip < chemcatcher < MESCO II. In this range, t50s as low as 1-4 days were found for SPMDs and silicone rods. Those for the Chemcatcher were in the range 2 to 8 days for analytes with log Kow below 5. Based on limited PRC data, t50s for MESCO II were between 7 and 22 days. Therefore, masses of the least non-polar analytes absorbed by these samplers may not be representative of the whole five days of exposure. This has been exemplified previously in simulations of deviations between true and measured CTwa in relation to the timing of the occurrence of peaks under conditions of fluctuating concentrations of the This journal is © The Royal Society of Chemistry 2010 J. Environ. Monit, 2010, 12, 696-703 | 699 analyte of interest and the t50 for that particular compound.20 Linear uptake is, however, expected for all samplers and analytes with log ,K0w > 5. The t50 for MESCO I (m) exposure is likely to be >10 days for most analytes studied here.7 For MESCO II, the t50 calculated here for low molecular weight PAHs is in agreement with the one week estimate by a previous study.9 Increases in overall mass transfer with increasing PRC hydrophobicity can be seen for analytes with log ,K0w between 3.9 and 4.5 in the Chemcatcher, silicone rod, and SPMD. This positive slope indicates that resistance to mass transfer in the membrane for analytes with log ,K0w < 4.8 cannot be neglected. However, it does not appear as steep as expected under fully membrane-controlled exchange for SPMDs11 and Chemcatcher.15 PRC dissipation data for silicone strip samplers clearly show a decrease in overall mass transfer coefficient with increasing PRC hydrophobicity and this is indicative of boundary layer-controlled contaminant exchange between sampler and water for analytes with log ,K0w > 4.5. Analyte masses absorbed A number of different criteria can be used for the comparison of the performance of different types of passive sampling devices. These can include the measurement of analyte masses absorbed normalised to sampler volume or area depending on whether analyte concentrations in the sampler have reached equilibrium or if uptake remains in the kinetic sampling stage.6,10,11 In the present system, measured total water concentrations (including bound and dissolved material) are likely to be higher than dissolved concentrations driving the uptake by the samplers. It is possible to calculate and compare TWA concentrations of pollutants for all samplers. However, uncertainty in the estimates obtained with the several types of sampler may be introduced by the different methods of calculation (e.g. use of empirical log Rs-\og ,K0w models6), uncertainties in the PRC dissipation rates and the values used for parameters such as sampler-water partition coefficients (Ksw). Hence it is challenging to make direct comparisons between estimates of dissolved concentrations of pollutants based on passive sampling and measurements based on analysis of spot water samples. Despite the short exposure time of five days, the spiking of the matrix with relatively high concentrations of contaminants resulted in the detection of a higher number of compounds with higher limits of detection than were found in a previous study in the field. The full list of compounds analysed for in extracts from each type of passive samplers was published previously.10 In the latter the same range of devices was deployed but for much longer times (up to 28 days).10 It is possible to assess the performance of the different sampler designs by comparing masses of analytes accumulated by the samplers, assuming linear uptake and boundary layer-control. In this case, masses are normalised to the surface area (A) of the sampler (eqn (3)). In order to aid interpretation of the data, normalised masses for each analyte from each sampler were further divided by the mean of analyte masses absorbed into SPMD samplers. These were plotted on a log scale against log .Kow f°r eacri sampler (Fig. 4). A ratio of 1 implies a negligible effect of the differences in sampler configuration, sampler placement in relation to water turbulences in the tank and variability in the analysis in different laboratories. In this case based on PRC data, normalised masses for analytes with log Kow > 4.6 would be expected to be similar and independent of the material used for the preparation of the sampler. MESCO I (m), silicone rod and strip samplers (and SPMD) appeared to be generally consistent for analytes with log .Kow > 4.5 with values reasonably close to one within the observed variability. Normalised masses as a proportion of masses absorbed into SPMDs are generally within a factor of 2. Based on the masses accumulated, the transition to boundary layer-controlled uptake does appear to occur at a log .Kow ~ 4.6. Below this threshold, ratios are far from the reference value of one. This indicates that for these analytes uptake is affected by the volume of the sampler and the sensitivity of the sampler to fluctuations in concentration. The within-sampler variability for MESCO I (m), MESCO II, silicone rod and silicone strip samplers is smaller than that observed for the SPMD and Chemcatcher. However, for both of the latter, a good relationship was observed between the relative masses of analyte absorbed in the different replicate samplers (with log .Kow > 4.6) and respective PRC dissipation data. This also indicates that despite PRCs being under "membrane-controlled" exchange kinetics, they appear to be representative of boundary layer resistance. In this system resistances in the membrane and boundary layers are likely to be in the same order of magnitude. This is further confirmed by comparing the relative standard deviations for PAH masses accumulated by triplicate SPMDs and the resulting Cw calculated using d10-phenanthrene Rs values for respective replicates. As shown in Fig. 5, the within-sampler variability reduces when PRC-based uptake rates are taken into account. While a similar picture can be seen for the Chemcatcher data, it is further complicated by the lag-phase effect. Replicate MESCO I (m), MESCO II and silicone rod samplers were exposed in a very similar fashion (since samplers are physically linked in a strip or rod) resulting in very low variability between replicate samplers. Although silicone strips are not joined together, the three replicates were deployed in a similar way, and this contributes to the low variability between replicates. Further, silicone strips and rods are made of a single polymer/membrane material and the entire sampler is generally extracted following exposure. Although the SPMD has a more complicated design, the whole sampler (including the LDPE membrane) is commonly extracted for analysis. However, for Chemcatcher and MESCO samplers, only the receiving phase, and not the LDPE layer, is extracted, and both Chemcatcher and MESCO II exhibit a decrease in the ratio of normalised masses with increasing analyte log ,K0w (Fig. 4). This is particularly marked for MESCO II where ratios close to or below 0.1 (a factor of ten or more below the expected value of 1) are found for the largest PAH and PCB compounds. It is likely that this behaviour results from the analysis of only analytes accumulated in the receiving phase, and a lag-phase in the accumulation of analytes in the receiving phase. This lag-phase can be considered as the time required for a steady state flux to be established across barriers of the sampler.6 Ratios for largest PAHs and PCBs are lower for MESCO II than those found for the Chemcatcher and this may be due to the additional air layer present in MESCO II. Wennrich et al.9 estimated MESCO II lag times in the range of 5-30 h and values as high as 48 h for PCBs. 700 | J. Environ. Monit, 2010, 12, 696-703 This journal is © The Royal Society of Chemistry 2010 10 Chemcatcher O □ 0.1 10 < co < 0.1 -e— -^-A-*Z£- □ □ MESCO II 0 Q Q 8r Q 10 □ CD O Silicone rod 8 i--------- 0.1 0.1 LogK( ow LogKow Fig. 4 Masses, m, of PAHs, PCBs, /),//-DDE and HCB absorbed in the six different samplers normalised to the surface area of the samplers (^sampler^sampler) and divided by the mean of triplicates measurements of normalised masses absorbed by SPMDs (»3spmeA4spmd). On the plot of MESCO II data, white and grey symbols are for PAHs and PCBs, respectively. Different symbols represent replicates. Estimates of membrane-based lag time6 (tL = 8M2/6DM), using published DM values,21,22 ranged between <1 h and 26 h for MESCO II (not accounting for the air layer), and between 20 min and 4 h for Chemcatcher. The lower values for the latter are due to the thinner LDPE membrane used in this sampler.5 Interestingly, MESCO II data for PAHs and PCBs appeared to follow two distinct trends (Fig. 4). Rather than plotting ratios as a function of log Kow, these are shown in Fig. 6 as a function of analyte diffusion coefficient in LDPE polymer.21 Improved agreement between ratios of PAHs and PCBs can be observed. This confirms that processes occurring in the LDPE membrane of MESCO II affect the overall uptake into this sampler under the present conditions. Implications for the use of passive sampling devices This study generally confirms and furthers our understanding of the principles of contaminant uptake into six types of passive sampling devices for monitoring non-polar organic contaminants. The ability to select test compounds and to use relatively This journal is © The Royal Society of Chemistry 2010 J. Environ. Monit., 2010, 12, 696-703 | 701 ■p ro d) Log Kow Fig. 5 Relative standard deviations (%) of triplicates measurements of PAH masses absorbed by SPMDs (•) and resulting triplicate Cw calculated using respective d10-phenanthrene Rs (O) for analytes whose uptake is expected to be linear and under boundary layer-control. -12.5 -13.0 14.0 Fig. 6 Masses of PAHs, PCBs, p,p'-DDE and HCB absorbed in MESCO II normalised to the surface area of the samplers (»3sampler/ ^sampler) ana divided by the mean of triplicates measurements of normalised masses absorbed by SPMDs (»3spmeA4spmd) plotted against log diffusion coefficient in the LDPE membrane.21 White and grey symbols are for PAHs and PCBs, respectively. membrane- and boundary layer-controlled uptake is in agreement with masses of contaminants accumulated under boundary layer-controlled uptake. The variability exhibited by certain sampler replicates could be explained by their positioning in the tank and differences in water turbulences around the replicates and this was demonstrated by the PRC dissipation rates. This work underlines the utility of the PRC approach since in natural waters conditions at the sampler surface can vary over short distances and in time. This in situ calibration method increases the robustness of applications of passive sampling in monitoring water quality. Devices tested in this trial had a range of properties, and some are better suited for use for compounds of lower log Kow and at higher concentrations (e.g. Chemcatcher), whilst others (e.g. SPMD, silicone strips and rods) are well suited for monitoring very non-polar compounds of which the dissolved concentration will be low. The latter samplers have high uptake rates and equilibrium is approached over short exposure times for compounds with log Kow < 4.6, when sampling ceases to be integrative, and information on TWA concentration is lost. Another significant property that needs to be considered is the occurrence of a lag-phase for Chemcatcher and MESCO II where only the receiving phase is extracted. This reduces their utility for very short deployment times. The comparison of sampler performance for analytes with log Kow < 4.6 is more complex since under present conditions, sampling was not truly integrative for all analytes and all samplers and contaminant concentrations in water varied significantly during the exposure. Uncertainties associated with samplers with very different configurations, possibly exhibiting lag phases, or different receiving phase volumes and those with sampler-water partition coefficients, Ksw values needed for the comparison, render such comparison futile here. Together with the use of reference sites, one way forward for the intercomparison of passive sampling technologies is the use of laboratory or pilot scale tank tests using either ultra pure, distilled or natural waters. Such trials have proved to be useful for conducting comparisons of the performance of passive sampling devices in measuring dissolved concentrations of metals23 and, in the present study, non-polar organics. This can help reduce uncertainties in the measurement of the concentration of pollutants present and improve control over environmental variables such as temperature and turbulence. high contaminant concentrations in the exposure tank allowed the evaluation to be effected across a wider range of compounds (and properties) than was possible in a previous comparative field study.10 This was particularly important for samplers (e.g., Chemcatcher) with relatively low uptake rates, and allowed a more thorough comparison of the devices. In addition, with higher contaminant masses accumulated in the samplers (further away from limits of detection), the inter-laboratory variability in the analysis is likely to be lower. These masses were used to compare the performance of the various samplers, where the overall variability included that due to analysis being conducted in three different laboratories. PRC dissipation data are consistent across the range of samplers and the transition between Acknowledgements We thank Nel Frijns and the RIZA monitoring team at Eijsden (The Netherlands), Uwe Schröter in Leipzig for the analysis of MESCO/silicone rod samplers and guidance with the size-exclusion chromatography and Ronald van Bommel for the extraction and analysis of silicone strips at NIOZ. We are grateful to Nathalie Guigues and the team from the Bureau de Recherche Geologique et miniere (BRGM) who contributed to the experimental set-up and running the test. Clive Thompson and ALcontrol Laboratories (Rotherham, UK) are acknowledged for their contribution to the analysis of water samples in this experiment. We acknowledge financial support from the European Union's Sixth Framework Programme (Contract 702 I J. Environ. Monit., 2010, 12, 696-703 This journal is © The Royal Society of Chemistry 2010 SSPI-CT-2003-502492; http://www.swift-wfd.com). Views presented here are those of the authors alone. Notes and references 1 B. Vrana, G. A. Mills, f. J. Allan, E. Dominiak, K. Svensson, J. Knutsson, G. Morrison and R. Greenwood, TrAC, Trends Anal. Chem. (Pers. Ed), 2005, 24, 845-868. 2 f. J. Allan, G. A. Mills, B. Vrana, J. Knutsson, A. Holmberg, N. Guigues, S. Laschi, A. M. Fouillac and R. Greenwood, TrAC, Trends Anal. Chem. (Pers. Ed), 2006, 25, 704-715. 3 f. J. Allan, B. Vrana, R. Greenwood, G. A. Mills, B. Roig and C. Gonzalez, Talanta, 2006, 69, 302-322. 4 K. Booij, F. Smedes and E. M. van Weerlee, Chemosphere, 2002, 46, 1157-1161. 5 B. Vrana, G. Mills, R. Greenwood, J. Knutsson, K. Svensson and G. Morrison, /. Environ. Monit., 2005, 7, 612-620. 6 J. N. Huckins, J. D. Petty and K. Booij, Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices, Springer, New York, 2006. 7 B. Vrana, A. PaschkeandP. Popp, Environ. Pollut., 2006,144,296-307. 8 B. Vrana, P. Popp, A. Paschke and G. Schuurmann, Anal. Chem., 2001, 73, 5191-5200. 9 L. Wennrich, B. Vrana, P. Popp and W. Lorenz, /. Environ. Monit., 2003, 5, 813-822. 10 1. J. Allan, K. Booij, A. Paschke, B. Vrana, G. A. Mills and R. Greenwood, Environ. Sei. Technol, 2009, 43, 5383-5390. 11 K. Booij, H. E. Hofmans, C. V. Fischer and E. M. Van Weerlee, Environ. Sei. Technol, 2003, 37, 361-366. 12 K. Booij, H. M. Sleiderink and F. Smedes, Environ. Toxicol. Chem., 1998, 17, 1236-1245. 13 K. Booij, R. van Bommel, A. Mets and R. Dekker, Chemosphere, 2006, 65, 2485-2492. 14 P. Mayer, M. M. Fernqvist, P. S. Christensen, U. Karlson and S. Trapp, Environ. Sei. Technol, 2007, 41, 6148-6155. 15 B. Vrana, G. A. Mills, M. Kotterman, P. Leonards, K. Booij and R. Greenwood, Environ. Pollut., 2007, 145, 895-904. 16 A. Paschke, K. Schwab, J. Brummer, G. Schuurmann, H. Paschke and P. Popp, /. Chromatogr., A, 2006, 1124, 187-195. 17 B. Vrana, G. A. Mills, E. Dominiak and R. Greenwood, Environ. Pollut., 2006, 142, 333-343. 18 B. Vrana and G. Schuurmann, Environ. Sei. Technol, 2002, 36, 290-296. 19 A. Paschke and R. Popp, /. Chromatogr., A, 2003, 999, 35^12. 20 C. Gourlay-France, C. Lorgeoux and M. H. Tusseau-Vuillemin, Chemosphere, 2008, 73, 1194-1200. 21 T. P. Rusina, PhD thesis, Masaryk University, 2009. 22 T. P. Rusina, F. Smedes, J. Klanová, K. Booij and 1. Holoubek, Chemosphere, 2007, 68, 1344-1351. 23 1. J. Allan, J. Knutsson, N. Guigues, G. A. Mills, A. M. Fouillac and R. Greenwood, /. Environ. Monit., 2007, 9, 672-681. This journal is © The Royal Society of Chemistry 2010 J. Environ. Monit, 2010, 12, 696-703 | 703 Príloha 18 Prokeš R., Vraná B., Klánová J., and Kupec J., Calibration of three passive samplers of hydrophobic organic compounds in water: Assessment of critical issues in experimental design data interpretation and field application, Fresenius Environ. Bull., 2010,19, 2812-2822. CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul CALIBRATION OF THREE PASSIVE SAMPLERS OF HYDROPHOBIC ORGANIC COMPOUNDS IN WATER: ASSESSMENT OF CRITICAL ISSUES IN EXPERIMENTAL DESIGN, DATA INTERPRETATION AND FIELD APPLICATION Roman Prokeš1, Branislav Vraná1*, Jana Klánová1 and Josef Kupec2 'Masaryk University, Faculty of Sciences, Research Centre for Toxic Compounds in the Environment RECETOX, Kamenice 126/3, 625 00 Brno, Czech Republic. 2 The T. G. Masaryk Water Research Institute, Mojmírovo náměstí 16, 612 00 Brno, Czech Republic. ABSTRACT Three types of passive sampling devices (two samplers based on a silicone polymer sheet and semipermeable membrane devices) were calibrated for the measurement of time-weighted average (TWA) concentrations of hydrophobic micropollutants, including polycyclic aromatic hydrocarbons and organochlorine pesticides, in water. During a 28 day exposure to constant analyte concentrations, linear uptake was observed into silicone rubber sheets for compounds with log ^Tow > 4.5. With exception of compounds with log K0Vi < 4 in SPMDs, uptake into all passive samplers was controlled by the water boundary layer (WBL). Thus, sampling rates are expected to vary widely depending on hydrodynamic conditions during field exposure. Sampling rates of highly hydrophobic polycyclic aromatic hydrocarbons (log K0Vi > 6) in all calibrations were significantly underestimated in comparison with the theoretical model that is based on diffusion of analytes in WBL. The difference could be explained by the effect of sorption to colloids present in water in the calibration tank. Since an independent measurement of analyte exchange kinetics using performance reference compounds was not performed, sampling rates in the field were calculated for anthracene using its concentrations in collected spot samples. Field sampling rates for the rest of compounds with log K0Vi > 4.5 were estimated using laboratory-derived calibration data adjusted for field exposure conditions. Application of this approach is demonstrated in a field study in which TWA aqueous concentrations from sampler data for target analytes correlated well with concentrations obtained from spot samples of water collected during the sampler deployment. KEYWORDS: calibration, passive sampling, persistent organic pollutants, semipermeable membrane devices, silicone rubber INTRODUCTION Passive sampling techniques are widely applied to assess exposure and contamination in water, air and soils. These techniques allow determination of the time-weighted average concentration of freely dissolved pollutant fraction over extended periods of time. Passive samplers are cheaper than conventional methods, their manipulation is simple, no power is needed and they can be used in harsh conditions. Diffusion of organic pollutants from sampled media to the sampler is driven by a difference in the chemical potentials. One of the most common applications of the passive sampling devices is the estimation of time weighted average (TWA) concentrations of pollutants for environmental risk assessment. The concentration found in a passive sampler can be used for estimation of TWA water concentration in field situations providing accurate calibration data is available. The accumulation of chemicals by passive samplers is characterized by an initial linear uptake stage followed by curvilinear and equilibrium partitioning stages. The exchange process between a passive sampler and water is described as follows [1]: N = m,KmCy,Q.-e-k-t) (1) where N is the mass of a target compound in the sampler at time /, ke is the exchange rate coefficient, Ksw is the sampler/water partition coefficient, m& is the mass of the sampler and Cw is the concentration of a target analyte in water. In the initial uptake phase, when the exponential term is very small (« 1), chemical uptake is linear or integrative. Then, in the linear region Eq. 1 can be reduced: N = RsCwt (2) where Rs is the sampling rate of the system, representing the equivalent extracted water volume per unit of time. A model for estimation of Rs that combines the resis- 2812 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin UJ tance to transport in both the water phase and the sampler is often applied in the passive sampling literature [2-4]: R. = kA i kK (3) sw j where k0 is the overall mass transfer coefficient, A is a surface area of the sampler, and ks are the mass transfer coefficients through the water boundary layer (WBL) and the sampler, respectively. The sampling rate R& is dependent on a variety of factors including water flow velocity, water temperature and biofouling [2, 5-7]. Correction for this variability can be achieved by estimation of R& from the dissipation rates of performance reference compounds (PRCs) spiked into the passive sampler prior to exposure. The dissipation rate is equal to the rate of the uptake process, and it is equally affected by variability in environmental factors [1,8]. Laboratory-derived calibration data are necessary to establish the relationship between Rs and K0Vi in order to apply R& estimated for compounds with moderate hydro- phobicity (e.g. PRCs) to compounds in a higher K0VI range. In this study, the relationship between Rs and log K0Vi was investigated for three different passive samplers (two silicone polymer sheets and semipermeable membrane devices, SPMDs) for a range of polychlorinated biphenyls [9], or-ganochlorine pesticides [10] and polycyclic aromatic hydrocarbons [6] in a flow-through system. Application of calibration data in a field exposure without the use of PRCs is demonstrated. MATERIALS AND METHODS_ Materials and chemicals Organic solvents dichloromethane, methanol, «-hexane, cyclohexane and chloroform were obtained from Lab-Scan, Ireland and Sigma-Aldrich, Czech Republic. Standards of 16 polyaromatic hydrocarbons (PAHs), 6 polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs) and internal standards (^-terphenyl, PCB 121) were obtained from Sigma-Aldrich, Czech Republic. Physicochemical properties of test analytes are given in Table 1. TABLE 1 - Selected physicochemical properties of test analytes at 25°C. Compound MW log A™" logiTswc logiW Rs (g mol1) Altesil LDPE Altesil (L ď1) SPMD Naphthalene 128.2 3.37 3.03 2.81 Eq.8 n.±h Acenaphthylene 154.2 3.92 3.02 3.16 Eq.8 n.d.h Acenaphthene 152.2 4.00 3.25 3.62 Eq.8 Eq.8 Fluore ne 166.2 4.18 3.79 3.77 0.98 2.19 Anthracene 178.2 4.54 4.21 4.33 1.53 15.59 Phenanthrene 178.2 4.57 4.11 4.22 4.83 3.07 Fluoranthene 202.3 5.18 4.62 4.93 1.22 10.76 Pyrene 202.3 5.22 4.67 5.10 1.66 7.29 Benzo(a)anthracene 252.3 5.90 5.32 5.73 0.48 5.83 Chrysene 228.3 5.86 5.25 5.78 1.14 7.01 B enzo (b)fluoranthene 228.3 5.91 5.74 6.66 0.82 5.36 B enzo (k)fluoranthene 252.3 5.90 5.74 6.66 0.84 4.70 Benzo(a)pyrene 252.3 6.04 5.70 6.75 0.56 2.55 Indeno(l,2,3-cd)pyrene 276.6 6.50 6.06 7.40 0.35 2.84 Dibenzo(a,h)anthracene 278.4 6.75 6.24 7.32 0.12 2.35 Benzo(g,h,i)perylene 276.3 6.50 6.02 7.27 0.33 2.45 PCB28 257.6 5.67 5.53 5.40 1.79 6.62 PCB52 292.0 5.84 5.80 5.55 1.89 6.17 PCB 101 326.4 6.38 6.28 6.18 1.66 4.80 PCB138 326.4 6.74 6.77 6.82 1.75 4.94 PCB153 360.9 6.83 6.72 6.81 1.37 3.34 PCB 180 360.9 6.92 6.98 7.24 1.42 3.89 p,p' -DDE 318.0 5.70 5.67" - 1.92 5.20 p,p-DDD 320.1 5.50 5.48" - 1.52 7.71 p,p'-DDT 354.5 6.19 6.14" - 0.53f 3.04 a-HCH 290.8 3.70 2.60e - 0.87 1.018 P-HCH 290.8 3.80 2.60e - Eq.8 0.338 y-HCH 290.8 3.80 2.60e - 0.84 1.878 5-HCH 290.8 4.10 2.60e - Eq.8 0.548 'Molecular weight (MW) bOctanol-water partition coefficient 18]. "Silicone rubber-water and LDPE-water partition coefficients [19] dData interpolated from the log Km- log K„ correlation "Data from Paschke and Popp [201 8 Partitioning equilibrium has been likely achieved during 28-days exposure. V>t determined 2813 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul Sampler preparation SPMDs were prepared according to the procedure described by Huckins et al. [11]. The layflat low density polyethylene (LDPE) tubing was purchased from Polymer Institute Brno, Czech Republic. The LDPE tubing (width 3 cm, thickness 80 urn) was cut into 83 cm pieces and thermo-sealed at one end with a heat sealer (ETA 0762, Czech Republic). To remove monomers and other impurities the LDPE was extracted in «-hexane for 48 h with solvent exchange after 24 h. The tubing was filled with 1ml of triolein (>97% purity, Sigma-Aldrich, Czech Republic) configured as a thin film and thermo-sealed at the other end. SPMDs had surface area of « 460 cm2. Silicone rubber (SR) sheets from two producers Rubena (Rubena, Czech Republic) and Altesil (Altec, Great Britain) were used in this study. The wall thickness of SR Rubena and SR Altesil was 0.1 and 0.5 mm, respectively. Two types of SR were prepared using the procedure described by Rus-ina et al. [4, 12]. SR sheets were cut into pieces of size 25 x 9.3 cm with surface area of «460 cm2. Two cleaning steps were applied to remove oligomers, other impurities and talcum powder from the surface. At first, SR was shaken in ethyl acetate for 1 d, and then Soxhlet-extracted in methanol for 12 h, wiped with a paper tissue and air dried in a fume-hood overnight. Calibration experiments In each experiment up to 9 passive samplers of each type were exposed in a constant concentration flow-through exposure system. This system was devised to allow calibration of the sampling devices to be made under controlled conditions of temperature (22°C), water turbulence, and analyte concentration (Fig. 1). It was operated in a temperature-controlled dark room. The system consisted of two glass tanks with an overflow to waste. Sterilised (using an UV lamp) and degassed (using helium) tap water and the solution of test analytes dissolved in methanol were mixed using a magnetic stirrer in a mixing tank at known and controlled rates and pumped into the exposure tank. Uncon-taminated tap water was pumped into the second, control tank. Water was fed to the exposure tank using a peristaltic pump at a constant flow of 3-5 L h"1, allowing a complete renewal of water in the tank every 4-6 h. Test chemicals were dissolved in methanol (667 ug L"1) and the appropriate amounts of stock solution (1.5 - 2 uL min"1) were delivered into exposure tank using a second peristaltic pump. A nominal concentration of 15-35 ng L"1 for each analyte was maintained throughout the experiments. The resulting methanol concentration in the exposure water did not exceed 0.0001 % (v/v). The effective flow velocity in both exposure tanks was 4 cm s"1. Prior to each exposure, the apparatus was operated for a minimum of 4-6 h without samplers to allow for stabilization of the water concentration of analytes. To ensure uniform hydrodynamic conditions in the vicinity of samplers, samplers were placed on stainless steel holders inside exposure tank. Two different orientations of sampler holders were used for exposure of SR and SPMDs, respectively. The control tank contained three samplers and the exposure tank up to 9 samplers. The exposures of SR lasted up to 28 days, during which triplicate samplers were removed at set time intervals and analysed to determine the concentrations of accumulated test chemicals. Following exposure, the devices were removed and the samplers were extracted to determine the mass of each analyte present in the sampler. In the first experiment with SPMDs, five samplers were exposed for 28d. The average concentration of pollutants in the exposure tank was 33.5 ng/1 and the flow rate was 3.05 1/h. In the second experiment, two types of SR (Altesil and Rubena) were exposed in the same exposure tank. Nine sheets of each type were installed and triplicate sheets from each type were collected after 7, 20 and 28 d, respectively. The average concentration of pollutants was 15 ng/1 and the flow rate was 5 1/h. V 10 11 w FIGURE 1 - Exposure apparatus used in flow-through calibration of passive sampling devices: 1-tap water inlet, 2-vertical tube, 3-helium cylinder, 4-UV tube for water disinfection, 5, 6, 8-peristalic pumps, 7-solution of analytes in methanol, 9-mixing tank, 10- tank for exposure of samplers to uncontaminated water, 11- tank for exposure of samplers to contaminated water. Extraction and analysis Following exposure, SPMDs were rinsed with tap water and distilled water and then wiped with a paper tissue. For a better analyte recovery, the cleaned SPMDs were perforated with scissors and SPMDs (membrane with triolein) placed into a vial and extracted by dichloromethane for 3 d according to Lebo et al. [13]. The extract was reduced in volume to about 2 ml using gentle stream of high purity nitrogen. To remove polyethylene waxes and triolein the extract was cleaned by gel permeation chromatography (GPC) using BioBeads S-X3 200-400 mesh according to Luellen and Shea [14]. The flow rate of chloroform was 0.6 ml/min and sample collected between 25 and 42 min. The cleanup for analysis of PAHs was performed on a silica gel column, a sulphuric acid modified silica gel column was used for PCB/OCP analysis in samples [2]. Extracts were reduced under a stream of nitrogen. Ter-phenyl and PCB 121 were used as internal standards for 2814 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul analysis of PAHs and PCBs, respectively. Samples were analysed using GC/MS. The surface of SRs was cleaned before extraction in the same way as SPMDs. SRs were Soxhlet-extracted for 12 h in methanol. The extracts were reduced in volume to 15 ml using Kuderna Danish concentrator. The final evaporation was provided with a gentle stream of high purity nitrogen to about 2 ml. The samples were cleaned up using silica gel column, reduced and internal standards were added. Samples were analysed by GC/MS [4, 15]. The extracts were analyzed by GC/MS on an Agilent 6890 equipped with a DB-5MS column (60 m x 0.25 mm i.d., film thickness 0.25 urn, carried gas He). The sample was injected in splitless mode. The temperature program was 80°C (hold 4 min) increase at 15°C/min to 180°C (hold 15 min), increase at 5°C/minto 310°C (hold 20 min). The MS parameters for both GC methods were: interface temperature 280°C, ion source temperature 250°C, electron impact (EI) ionization mode at 70 eV. Analysis was performed by selected ion monitoring (SIM) applying two or three characteristic ions for each compound in both detection and quantification. Field study The field study was performed at the sampling site Spytihnev (WGS84: 49°08'08,7" N, 17°30'11,9" E, altitude 188 m) in the Morava river (south-eastern part of the Czech Republic). This area is an industrial and agricultural region with 10 cities and 72 villages [16]. SRs were transported to the field in a cool box. SRs were placed in a stainless steel holder and exposed for 28 days. After period of sampling SRs were packed in two layers of aluminium foil, put in a polyethylene zip-lock bag and transported to the laboratory. Until analysis samples were stored in a freezer at -18°C. RESULTS AND DISCUSSION_ Data analysis The experimentally determined time courses of the amounts of individual test substances in the Altesil sampler were fitted by linear regression analysis using modified Eq. 2 in form r V TT/ CF(t) Rst (4) Where CF(t) is the concentration factor obtained by dividing the accumulated amount of analyte after defined time period / (0, 7, 20 and 28 days, respectively) by the average concentration in the exposure tank for a given time period (0 to /). CF presents equivalent volume of water extracted by sampler for a given period of time. Characteristic analyte uptake curves for the sampler are shown in Figures 2 and 3. 15 20 Time [days] FIGURE 2 - Uptake of selected PAHs in the Altesil sampler in a flow-through exposure at nominal water concentration of analytes 15 ng L"1. The drawn lines show the linear fits of the data using Eq. 4. Time [days] FIGURE 3 - Uptake of selected PCBs and organochlorinated pesticides in the Altesil sampler in a flow-through exposure at nominal water concentration of analytes 15 ng L"1. The drawn lines show the linear fits of the data using Eq. 4. Because an independent measurement of analyte exchange kinetics using PRC was not performed, an alternative check was performed that the uptake of analytes was linear and integrative during the whole exposure period. Uptake remains essentially linear until 50% of equilibrium concentration is reached. The time it takes to reach the equilibrium concentration (ti/2) is related to the elimination rate constant and can be estimated: 2815 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul tV2=\n2lke*\n2KswmsIRs (5) where Ksw is the sampler/water equilibrium partitioning coefficient. Ksw values published recently by Smedes et al. [17] were used. For HCH isomers, Ksw values for bulk silicone material were not available and log Ksw =2.6, obtained using solid phase microextraction fibre coated with polydimethylsiloxane was used [20]. The estimate shows that, with exception of HCHs (excepting 5-HCH), naphthalene and acenaphthylene (ty2 = 5, 10 and 18 days, respectively) with low values of partition coefficient (log KSVi < 3.7) compounds should accumulate into samplers in linear uptake mode during the whole exposure period of 28 days. For SPMDs, uptake kinetics were not measured. Thus, sampling rates were estimated using a single point calculation from amounts accumulated after 28 days of exposure. Linear uptake regime during this exposure time was also checked for individual analytes using Eq. 5 and Ksw data available from literature [21]. Calculated sampling rates are shown in Table 1. Comparison of two silicone rubber samplers In the experiment with SR, identical exposure conditions were applied by the use of samplers with exactly the same surface area and geometry. They were also positioned in the exposure chamber in the same position. When uptake is linear (integrative) and WBL controlled and samplers are exposed in the flow through system under the same hy-drodynamic conditions, masses of analytes absorbed by the samplers should be the same. Moreover, because both samplers are made of silicone rubber with similar properties (diffusion and partition coefficients of analytes), absorbed masses of analytes in the linear uptake phase are expected to be very similar even for compounds accumulated under membrane control (less hydrophobic compounds). The only difference in absorbed mass is expected for compounds that reach partition equilibrium during exposure (some of those with lowest Ksw values), for which the ratio of accumulated amounts in both samplers should be: No K 0m i 2co sw2 si where N, x is the amount in a sampler accumulated at equilibrium. When equality of partition coefficients KSVi in both polymers is assumed, msX= 14 g (Altesil) and ms2 = 12 g (Rubena), the expected ratio of Ni/N2 at equilibrium is 1.17. Considering the average variation in the experimental data of cca 10%, it seems unlikely that a significant difference in accumulated masses could be observed for compounds that reached partitioning equilibrium. In agreement with theoretical considerations, the 28 day exposure resulted in the measurement of similar masses of all analytes for both SR polymers (Figs. 4 and 5). In comparison with Rubena, Altesil contained slightly higher amounts of PAHs and lower concentrations of PCBs and organochlorine pesticides. The same pattern was obtained for data from 7 and 20 day exposures. Because these differences were small, it cannot be unambiguously judged whether the differences originate in differences of SR material properties or in the bias of methods used for analyte quantification. In linear uptake phase, Rubena shows a comparable performance with Altesil and calibration data obtained for Altesil can be applied for both polymers. Po- 2816 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul # «M" «M- ^ ^ ^ ^ ^ ■ / / / FIGURE 5 - Mean amounts of PCBs and organochlorine pesticides accumulated in two different SR samplers after 28 days of exposure at 15 ng L"1. tential differences in partition coefficients require further investigation. However, practical experience with Rubena shows that the polymer contains some components that complicate sample extraction, cleanup and instrumental analysis. Rubena is a type of industrial silicone that is surface treated with coating of chalk and we recommend Altesil as a better choice. Comparison of Altesil and SPMD As stated earlier, when a) uptake is linear (integrative) and WBL controlled and b) samplers with the same surface area are exposed in the flow through system under the same hydrodynamic conditions, masses of analytes absorbed by the samplers should be the same and independent of the sampler material. When comparing data obtained from SPMD and Altesil exposures, it is necessary to consider that the condition b) was not fulfilled. Although the samplers had the same surface area, their geometry was different. SPMDs are stripes 83x3 cm and Altesil sheets were rectangles 25x9 cm. Although the linear flow velocity in the calibration apparatus was maintained the same (4 cm s"1), the difference of sampler orientation in the exposure chamber can cause significant differences in local hydrodynamic conditions. These result in differences in the thickness of the WBL at the surface of the sampler and consequently in the sampling rates of compounds accumulated under WBL control. Indeed, sampling rates of analytes accumulated in integrative regime in SPMDs were on average 5.8 times higher than those obtained for SR. Prolonged linear uptake is favoured in samplers with high accumulation capacity, given as a product msxKsw. High capacity of the Altesil sampler used in this study is achieved by the use of higher amount of sorptive material (14 g) in comparison with other samplers, e.g. the standard SPMD (5 g). Theoretical model of analyte uptake To obtain information on the processes that affect the sampling rates obtained in calibration studies, data were compared with the theoretical model for estimation of R& that combines the resistance to transport in both the water phase and the sampler (Eq. 3). Membrane control The value of mass transfer coefficient in sampler material ks (Eq. 3) can be calculated from the diffusion coefficient in the sampler material (Z)s) and the half-thickness of the sampler 5S: (7) The corresponding sampling rate of compounds accumulated fully under membrane control with negligible resistance to mass transfer in WBL (l/kw « l/(ksKsw) is given: fi> - lr TT A - ^sK-swA (8) Diffusion coefficients of analytes of interest in SR were reported recently [12]. At laboratory temperature, for analytes of interest they range from 10"9 to 10"11 m2 s"1 and they decrease with molecular volume. Diffusion coefficients in LDPE are 2 to 4 orders of magnitude lower than those in SR, ranging from 10"12 to 10"15 m2 s"1 at 25°C [12]. Partition coefficients between various polymers including silicone rubber, LDPE and water were published, too [19]. Application of these values, e>s = 2.5 and 0.08 mm for Altesil and SPMD, respectively, yields the estimate of sampling rates that can be potentially achieved if the resis- 2817 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul tance of the WBL was negligible and compounds were accumulated completely under sampler control. For Altesil, estimated maximum achievable sampling rates are more than three orders of magnitude higher than those determined in our experiment. For SPMDs, calculation yields sampling rates for compounds with log K0Vi > 4 that are more than two orders of magnitude higher than those determined in our experiment. This indicates that all compounds with log ^Tow > 4 were accumulated under WBL and the resistance to mass transfer in the sampler material can be neglected. Less hydrophobic compounds (e.g. HCH isomers, and some less hydrophobic PAHs) may be accumulated under membrane control in SPMD [7]. Water boundary layer control We demonstrated that most compounds under investigation were accumulated under WBL control. When neglecting the resistance to mass transfer in the sampler, sampling rates that are controlled by the WBL can be modeled as Rs=kwA (9) where &w is the mass transfer coefficient in the WBL. In general, increases when flow rates and turbulence intensities increase. The typical relation between and the diffusion coefficient can then be summarized as [22] k xD2'3 (10) ww v ' Because sampling rates are commonly given as a function of log ^Tow, Booij et al. [2] expressed log Dw for PCBs, PAHs and chlorobenzenes as a function of log K0Vi, and obtained Rs=ABwK-or (ID where 5W is a constant for a given exposure, but may vary among exposures according to differences in hydro-dynamic conditions and sampler geometry. ABV has the same units as Rs and the value equals the hypothetical sampling rate for K0Vi = 1. The equation predicts the sampling rates to weakly decrease with increasing log ^Tow in the high log ^Tow range. Depending on models used for estimation of diffusion coefficients, the dependence may vary frames ~ ^ow"°02 - ^ow"°06 [4]. Rusina et al. [4] confirmed this theory; in a calibration of silicone strips for PAHs and PCBs that experimental Rs was proportional to KOVi'om. In contrast to the theoretical model, steeper decrease of sampling rates of SR with increasing hydrophobicity was observed in our study. In the range where uptake is WBL controlled (log ^Tow > 4.5), our data show that Rs ~ KOVi'0'13, however, with a low correlation coefficient (R2 = 0.09). Much better correlations where obtained when Rs were related to K0Vi for individual classes of compounds. DependencesRs~ AT0W"°62 (R2 = 0.90) andRs~Km°M (R2 = 0.56) were obtained for PAHs and organochlorine pesticides, respectively. Similar dependencies have been shown in experimental data obtained in other calibration studies, e.g. Rs ~ Km°26and ~ AT0W"°85 for SPMD [23] and Chemcatcher [24], respectively. Analytes adsorbed on colloids or particles are not directly available for sampling with passive samplers. A possible reason for a large drop in measured sampling rates of very hydrophobic compounds may be the overes-timation of dissolved aqueous concentrations due to sorption of analytes to dissolved organic carbon (DOC). Burk-hard [25] reviewed contaminant sorption by dissolved organic matter. Using several hundreds of DOC-water partition coefficients (^doc) reported in these studies, he found that DOC-water partition coefficients for naturally occurring DOC (humic and fulvic acids, sediment pore water, soil pore water, groundwater, and surface water) was best described by log^oc=log^-l.ll (12) The 95% confidence interval amounted to 1.3 log units, which corresponds to a scatter in the KDOc values by a factor of 20. Adopting Eq. 11 for the sampling rate of truly dissolved analytes, and the Burkhard relationship for sorption to DOC, the apparent sampling rate (i?s,apP) is given by R Rs _ ABwK;r _ AByKtr (13) s'app l + [DOC]KDOC 1 + [DOC]Kdoc 1 + [DOC]QKow where Q is dependent on DOC quality (Q = 10_111 ~ 0.078 for DOC of average quality; see Eq. 11). In order to check if Eq. 13 sufficiently describes the experimental sampling rates, this model was fitted to the calibration data for compounds accumulated under WBL (with log ^Tow > 4.5) and assuming a log normal distribution of errors. logRs = logABlr -0.044 log Zor - £ z,log(l + [DOC]Q,Kow) (14) (=i where log ^Tow is the independent variable; z; are indicator variables taking the value z; = 1 for experimental data for the z'-th group of compounds (/'= 1 for PAHs and z'=2 for organochlorine compounds, respectively), for the rest of the data, z;= 0; and log Rs is the dependent variable, log A8W and log Q, [DOC] are adjustable parameters. Sampling rates of DDT were excluded from the calculation, because they seemed underestimated due to a measurement error of water concentration that was taken for calculation. Results of the fit are shown in Figure 6 and in Table 2. Inclusion of a DOC-sorption term in the model significantly improved the log Rs - log ^Tow fit for the calibration data. Unfortunately, DOC concentrations in water from the calibration apparatus were not measured. Thus, the DOC quality could be only roughly estimated, assuming a hypothetical concentration of DOC of 0.5 mg L"1. Values of log Q fall within the 95% confidence range of log Q values (-2.4 to +0.2) reported by Burkhard [25]. Although the evidence is indirect, sorption to DOC probably caused an underestimation of the sampling rates of highly hydrophobic compounds in this study. The applied model indicates that different decrease of apparent sampling rate with 2818 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul TABLE 2 - Nonlinear regression analysis results of logfls to log ^o„ using Eq. 14." Sampler n r s logABwCLď1) log öpah [DOC] log Qpcb [DOC] log ÖpV (cm3 g1) log öpcb°(cm3 g"1) Altesil 21 0.94 0.13 0.54 ±0.04 -5.74 ±0.09 -7.55 ±0.33 -0.4 -2.3 SPMD 21 0.86 0.13 1.16 ±0.05 -6.24±0.12 -6.94± 0.16 -0.5 -1.2 "nonlinear regression was performed using SigmaPlot for Windows Version 11.0 "assuming [DOC] = 0.5 mg L"1 increasing hydrophobicity for both groups of compounds may be explained by stronger adsorption of PAHs to colloidal matter in the experimental system than was the case for PCBs and organochlorine pesticides included in the datasets. An underestimation of the sampling rate by a factor of 2 occurs when Q [DOC] K0Vi = 1. Inspection of the data shows that sampling rates of compounds with log ^Tow values greater than 5.7 for PAHs and 7.55 for organochlo-rinated pesticides and PCBs, respectively, may have been underestimated. Experimental data for PCBs seem to be less affected by sorption phenomena than data for PAHs. For SPMD data, where only a single point in time calculation of R& was performed, similar results were obtained (Table 2). '°3 Kow FIGURE 6 - Dependence of the sampling rate log Rs in Altesil on the octanol/water partition coefficient log Km for PAHs (full dots), PCBs and OCPs (empty dots). The lines correspond to Eq. 14 with the values of optimized parameters given in Table 2. The dashed line shows the theoretical model of diffusion in WBL [2], Eq. 11. The the difference of log ABV in the two experiments is 0.62 which means that sampling rates determined in the experiment with SPMDs were on average four times higher than those in experiment with SR. This corresponds well with the mean value of sampling rate ratio of 5.8 for compounds accumulated in integrative regime. Data corrected for the effect of adsorption do not contradict the validity of the theoretical model (k^ « Dw2/3). Methods for independently measuring the extent of sorption to DOC should be included in future calibration experiments. Alternatively, methods of calibration, based on distribution of analytes between dosing and acceptor sheets that do not require measurement of analytes in the water phase should be applied [4]. Application of calibration data in field situations Evaluation of calibration data obtained with the two passive samplers indicates WBL control over accumulation of all analytes with log ^Tow > 4. Because of WBL control over the mass transfer, sampling rates are expected to vary widely depending on hydrodynamic conditions during exposure. Without the availability of information on in situ exchange kinetics from performance reference compounds (PRC), the error in estimate of sampling rates in a real situation can reach several orders of magnitude, depending on the difference of the hydrodynamic conditions between laboratory and the field. Without the availability of data from PRC elimination, an alternative method for minimizing the error in in situ Rs estimate is essential. In situ sampling rates can be estimated using Eq. 11. The value of Bw is variable in field conditions and it depends on local hydrodynamic conditions, sampler geometry and temperature. Thus, it has to be determined for each field exposure. For this purpose, it is necessary to determine the sampling rate for at least one compound under investigation. Preferably, it should be a) a compound that can be found both in the passive sampler and in the water phase at quantifiable concentration; b) a compound that is accumulated under WBL control and remains in linear uptake phase during the whole sampler exposure; c) the compound should be present in the sampled water predominantly in the dissolved phase. These conditions are fulfilled for moderately hydrophobic compounds with log ^Tow «4.5, e.g. phenanthrene or anthracene. For such compound, in situ Rs can be estimated using rearranged Eq. 2: N C t (15) where Cw is the mean value of anthracene concentration in spot samples of water taken during sampler exposure. Concentration of anthracene should not fluctuate widely during exposure, otherwise the calculation may be biased. A check has to be performed using Eq. 5 that the compound does not equilibrate during exposure. In the next step, value of log ABW is calculated by substituting the calculated Rs value of anthracene into Eq. 11. Sampling rates of compounds with log ^Tow > 4.5 are then extrapolated using Eq. 11 with adjusted exposure specific log ABW value. This approach was tested on data from a field study performed in the River Morava at the sampling site 2819 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul Spytihněv in July 2007. Analyte concentrations obtained using passive sampling with Altesil (C^a) and spot samples taken before and after sampler exposure (Cb) are shown in Table 3. For anthracene, site specific value of log A8W of 0.74 was calculated. This means that sampling rates at the sampling site were less than factor two higher than in the calibration study with Altesil. Nevertheless, this similarity in exposure conditions is likely only a coincidence. The mean calculated ratio of CTwA/Cb of 2.8 is acceptable, considering possible fluctuation of water concentrations at the sampling site during exposure, which is not reflected in data from spot samples. A good correlation was obtained between concentrations obtained from Altesil SR and those obtained from spot sampling, assuming log normal data distribution. logQ = -0.263+ 0.868^0^ (16) N= 21, R = 0.861, s = 0.40 Passive sampling data slightly overestimate data obtained using spot sampling. From theory, elevated difference between spot and passive sampling is expected for very hydrophobic compounds (log K0Vi > 6) that adsorb on colloids and particles present in the water phase. Our limited dataset does not show any trend of difference increasing with hydrophobicity. For compounds that likely achieved partition equilibrium during exposure, concentration in water was calculated as C^N/K^/m^ However, this concentration does not does not represent a TWA value. -2-10 1 2 '°9 CTWA FIGURE 7 - Correlation between the TWA analyte concentrations determined using Altesil sampers (log CTWa) and those determined in two spot samples taken before and following passive sampler exposure (log Cb) at the sampling site Spytihnev. Dashed lines present 95% confidence and prediction intervals, respectively. TABLE 3 - Mean concentrations of PAHs, PCBs and OCPs found in Altesil (ng per sampler; n=2), calculated sampling rates Rs, estimates of TWAconcentrations from Altesil CTWa, and concentrations measured in bulk water samples Cb collected at the beginning and the end of a 28 day sampler exposure at the sampling site Spytihnev in July 2007. Compound N[ng] Sampling mode Rs [L d"1] Ctwa [ng L"1] Cti [ngL1] Cti [ngL1] Naphthalene 17 eq.a n.e.° 1.2" 4.7 7.8 Acenaphthylene 9 eq. n.e. 0.6" 0.3 0.1 Acenaphthene 29 eq. n.e. 1.2" 2.1 1.2 Fluorene 66 eq. n.e. 0.8" 2.3 1.2 Phenanthrene 260 Linear 3.5 2.6 7.2 3.1 Anthracene 385 Linear 3.5 3.8 5.6 2.0 Fluoranthene 1236 Linear 3.3 13.0 11.4 3.4 Pyrene 1416 Linear 3.3 14.9 8.5 2.0 Benz(a)anthracene 165 Linear 3.1 1.9 0.6 ad. Chrysene 206 Linear 3.1 2.3 1.7 0.3 Benzo(b)fluoranthene 60 Linear 3.0 0.7 0.3 ad. Benzo(k)fluoranthene 36 Linear 3.1 0.4 0.2 ad. Benzo(a)pyrene 97 Linear 3.0 1.1 0.1 ad. Indeno(l,2,3cd)pyrene 5 Linear 2.9 0.1 ad. ad. Dibenz(a,h)anthracene n.d. Linear 2.8 n.d. ad. ad. Benzo(g,h,i)perylene 6 Linear 2.9 0.1 ad. ad. PCB28 16 Linear 3.1 0.2 ad. ad. PCB 52 7 Linear 3.1 0.1 0.1 ad. PCB 101 4 Linear 2.9 0.1 ad. ad. PCB 118 1 Linear 2.8 n.d. ad. ad. PCB 153 7 Linear 2.8 0.1 0.1 0.2 PCB 138 4 Linear 2.8 0.1 ad. ad. PCB 180 n.d. Linear 2.6 n.d. ad. ad. p,p'-DDE 36 Linear 3.1 0.4 0.1 0.1 p,p'-DDD 15 Linear 3.2 0.2 0.1 0.1 p,p'-DDT n.d. Linear 3.0 n.d. ad. ad. a-HCH 32 eq. n.e. 5.7" 0.3 0.7 P-HCH n.d.c eq. n.e. n.d. ad. ad. 7-HCH 19 eq. n.e. 3.4" 0.6 0.5 5-HCH n.d. eq. n.e. n.d. ad. ad. aeq. - partitioning equilibrium between sampler and water has likely been achieved bn.e. - not estimated; cn.d. - not detected; estimated using equilibrium partitioning model 2820 CO © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin UJ CONCLUSIONS The data confirms that for compounds accumulated under WBL control differences in water flow velocities can cause sampling rates to vary several orders of magnitude. Because of the complexity of the hydrodynamics involved, there is little hope that sampling rates can be expressed as a simple function of ambient flow rates [22]. Therefore, estimation of in situ sampling rates by measuring the dissipation rates of performance reference compounds should be mandatory. In their absence, alternative method must be applied that allows reliable estimate of in situ sampling rate of at least one compound under investigation. An option is the measurement of this compound using spot regular spot sampling during sampler exposure. Sampling rates for compounds that cannot be easily detected by spot sampling because of their very low concentrations can then be estimated from laboratory-derived relationships between Rs and log K0VI or other properties (diffusion coefficients, molecular mass etc.). ACKNOWLEDGEMENTS This research was supported by the Czech Ministry of Education (project MSM 0021622412) and by the EU Operational Programme "Research and Development for Innovations", the CETOCOEN project (no. CZ. 1.05/2.1.00/01.0001). REFERENCES [1] Booij, K., Sleiderink, H. M. and Smedes, F. (1998) Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environmental Toxicology and Chemistry 17, 1236-1245. [2] Booij, K., Hofmans, H. E., Fischer, C. V. and van Weerlee, E. M. (2003) Temperature-dependent uptake rates of nonpo-lar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environmental Science & Technology 37, 361-366. [3] Vraná, B., Mills, G.A., Kotterman, M., Leonards, P., Booij, K. and Greenwood, R. (2007) Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water. Environmental Pollution 145, 895-904. [4] Rusina, T., Smedes, F., Kobližková and M., Klanová, J. (2010) Calibration of silicone rubber passive samplers: experimental and modeled relations between sampling rate and compound properties. Environmental Science & Technology 44, 362-367. [5] Rantalainen, A.-L., Cretney, W. J. and Ikonomou, M. G. (2000) Uptake rates of semipermeable membrane devices (SPMDs)forPCDDs, PCDFs. Chemosphere 40, 147-158. [6] Huckins, J. N., Petty, J. D., Orazio, C. E., Lebo, J. A., Clark, R. C, Gibson, V. L., Gala, W. R. and Echols, K. R. (1999) Determination of uptake kinetics (sampling rates) by lipid-containing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water. Environmental Science & Technology 33, 3918-3923. [7] Vraná, B. and Schüürmann, G. (2002). Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics. Environmental Science & Technology 36, 290-296. [8] Huckins, J. N., Petty, J. D., Lebo, J. A., Almeida, F. V, Booij, K, Alvarez, D. A., Cranor, W. L., Clark, R. C. and Mogensen, B. B. (2002) Development of the permeability/performance reference compound (PRC) approach for in situ calibration of semipermeable membrane devices (SPMDs). Environmental Science & Technology 36, 85-91. [9] Heinisch, E., Kettrup, A., Bergheim, W., and Wenzel, S. (2006) Persistent chlorinated hydrocarbons (PCHC), source-oriented monitoring in aquatic media 5. Polychlorinated biphenyls (PCBs). Fresenius Environmental Bulletin 15, 1344-1362. [10] Heinisch, E., Kettrup, A., Bergheim, W., Martens, D., and Wenzel, S. (2005) Persistent chlorinated hydrocarbons (PCHC), source-oriented monitoring in aquatic media 2. The insecticide DDT, constituents, metabolites. Fresenius Environmental Bulletin 14,69-85. [11] Huckins, J. N., Manuweera, G. K, Petty, J. D., Mackay, D. and Lebo, J. A. (1993) Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environmental Science & Technology 27, 2489-2496. [12] Rusina, T.P., Smedes, F., Klanová, J., Booij, K. and Holoubek, I. (2007) Polymer selection for passive sampling: a comparison of critical properties. Chemosphere 68, 1344-1351. [13] Lebo, J. A., Huckins, J. N., Petty, J. D., Ho, K. T. and Stern, E. A. (2000) Selective removal of organic contaminants from sediments: a methodology. Chemosphere 40, 811-819. [14] Luellen, D.R. and Shea, D. (2002) Calibration and field verification of semipermeable membrane devices for measuring polycyclic aromatic hydrocarbons in water. Environmental Science & Technology, 36 1791-1797. [15] Luszezanec A. Silicone rubber as a reference phase to study sediment-water interactions, National Institute for Coastal and Marine, The Netherlands, Werkdocument RIKZ/IT-2001.647x, (2001) [16] Hilscherová, K, Dušek, L., Kubik, V., Čupr, P., Hofman, J., Klánová J. and Holoubek, I. (2007) Redistribution of organic pollutants in river sediments and alluvial soils related to ma-jorfloods. Journal of Soils Sediments 7, 167-177. [17] Illustrated Handbook of Physical-Chemical Properties of Environmental Fate of Organic Chemicals, vol. 1. ed. D. Mac-kay, W.Y. Shiu, K.C.Ma, Lewis Publishers, MI, USA,1992. [18] Illustrated Handbook of Physical-Chemical Properties of Environmental Fate of Organic Chemicals, vol. 2. ed. D. Mac-kay, W.Y. Shiu, K.C.Ma, Lewis Publishers, MI, USA,1992. [19] Smedes, F., Rinze, W., Geertsma, A., Van der Zande, T. and Booij, K. (2009) Polymer-water partition coefficients of hydrophobic compounds for passive sampling: application of cosolvent models for validation. Environmental Science & Technology 43, 7047-7054. [20] Paschke, A. and Popp, P. (2003) Solid-phase microextraction fibre-water distribution constants of more hydrophobic organic compounds and their correlations with octanol-water partition coefficients. Journal of Chromatography A 999, 35-42. [21] Monitoring of Organic Chemicals in the Environment: Semipermeable Membrane Devices, ed. J.N. Huckins, J.D. Petty and K. Booij, Springer, New York, 2006. 2821 OQ © by PSP Volume 19-No 12. 2010 Fresenius Environmental Bulletin Ul [22] Booij K., Vrana B. and Huckins J.N. Theory, modeling an calibration of passive samplers used in water monitoring. In Monitoring of Organic Chemicals in the Environment: Semipermeable Membrane Devices., ed. J.N. Huckins, J.D. Petty and K. Booij, Springer, New York, 2006, pp. 141-169. [23] Huckins, J. N, Petty, J. D., Orazio, C. E., Lebo, J. A., Clark, R. C, Gibson, V. L., Gala, W. R. and Echols, K. R. (1999) Determination of uptake kinetics (sampling rates) by lipid-containing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water. Environmental Science & Technology 33, 3918-3923. [24] Vrana, B., Mills, G. A., Kotterman, M., Leonards, P., Booij, K. and Greenwood, R. (2007) Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water. Environmental Pollution 145, 895-904. [25] Burkhard, L. P. (2000) Estimating dissolved organic carbon partition coefficients for nonionic organic chemicals. Environmental Science & Technology 34, 4663-4667. Received: March 23, 2010 Accepted: June 01, 2010 CORRESPONDING AUTHOR Branislav Vrana Masaryk University Faculty of Sciences Research Centre for Toxic Compounds in the Environment RECETOX Kamenice 126/3 625 00 Brno CZECH REPUBLIC Phone: +420 608 96 6271 Fax: + 420 549 49 2840 E-mail: vrana@recetox.muni.cz FEB/ Vol 19/ No 12/ 2010 - pages 2812 - 2822 2822 Príloha 19 Tapie N., Devier M. H., Soulier C, Creusot N., Le Menach K., ATt-ATssa S., Vraná B., and Budzinski H., Passive samplers for chemical substance monitoring and associated toxicity assessment in water, Water Sci. Technol., 2011, 63, 2418. ) IWA Publishing 2011 water Science & Technology | 63.10 | 2011 Passive samplers for chemical substance monitoring and associated toxicity assessment in water N. Tapie, M. H. Devier, C. Soulier, N. Creusot, K. Le Menách, S. Ait-Aissa, B. Vrana and H. Budzinski ABSTRACT The European legislation, and in particular the water Framework Directive requires the development of cost efficient monitoring tools that can provide the required information for the assessment of water contamination. Passive sampling methods represent one of the novel tools that have a potential to be used in various regulatory monitoring programmes aimed at assessing the levels of chemical pollutants. These methods are particularly interesting for sampling polar organic pollutants in water because they provide representative information of the water quality over extended time periods (days to weeks) in environments with fluctuating contaminant concentrations. This is achieved by integrative sampling of pollutants over the whole sampler deployment period. These tools can be coupled to toxicity testing using bioassays that give information on toxic and ecotoxic hazards associated to substances that are present, these substances being identified or not. in this study the polar organic chemical integrative sampler (POCIS) was used in surface water to evaluate the water contamination by polar organic compounds and their potential toxicity. Keywords | Biotests, hormones, passive sampling, pesticides, pharmaceuticals, POCIS N. Tapie M. H. Devier c. Soulier K. Le Menach H. Budzinski (corresponding author) LPTC/ISM - UMR 5255 CNRS University Bordeaux 1, 33405 Talence, France E-mail: h.budzinski@ism.u-bordeauxl.fr N. Creusot s. Ait-Aissa NERIS, Unite Ecotoxicologie in vitro et in vivo - BP2 60550 Verneuil en Halatte, France B. vrana Research Centre for Toxic Compounds in the Environment (RECETOX) Faculty of Science, Masaryk University Kamenice 126/3 625 00 Brno Czech Republic INTRODUCTION The restoration of good ecological and chemical status of all water bodies in Europe by 2015 as requested in the Water Framework Directive (WFD, Directive 2000/6O/EC) is an important stake. One of the goals of this directive is to improve the water quality by reversing, when necessary, the degradation trend of underground and surface waters by gradually reducing the discharges of substances that have been classified as priority pollutants. Discharges should even be stopped for substances classified as hazardous priority compounds. To achieve this goal, the WFD requires the development of cost efficient monitoring tools that can provide the required information for the assessment of water contamination. Checking water quality compliance with regulatory provisions is usually based on the chemical analysis of spot (bottle) samples of water taken at a defined frequency. This approach suffers from several drawbacks. Spot samples doi: 10.2166/wst.2011.129 provide concentrations of pollutants only at the moment of sampling. Thus, in water bodies characterized by marked temporal and spatial variability there is an increased risk of a false classification of the chemical status. Further, the laboratory methods commonly used for the analysis of spot samples of water are often not sensitive enough to fulfil the required minimum performance criteria associated with the current environmental quality standards for pollutants (Commission Directive 2009/9O/EC). A promising alternative for monitoring pollutants in aquatic systems is based on the use passive sampling techniques. In comparison to spot sampling techniques, passive samplers provide a more representative picture of the water quality. This is achieved by the integrative sampling of contaminants during sampler deployment periods up to several weeks. Passive samplers can be used alongside spot sampling in order to corroborate or contradict the data obtained. This 2419 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 approach can provide additional 'weight-of-evidence' in water bodies where concentrations of contaminants are expected to fluctuate widely with time. The measurement of time-weighted average (TWA) concentrations over periods of weeks to months using passive sampling seems to be a promising approach. A range of passive samplers has been developed for monitoring organic pollutants in water. Their different designs and field performance have been reviewed (Namies-nik et al. 2005; Stuer-Lauridsen 2005; Vrana et al. 2005). Among available passive sampling techniques polar organic chemical integrative sampler (POCIS) has shown a potential to be used in various monitoring programmes aimed at assessing the levels of polar organic compounds in the aquatic environment. In addition to instrumental analysis of pollutants in sampler extracts, these can be subjected to toxicity testing using bioassays that give information on toxic and ecotoxic risks associated with the sampled substances (substances being identified or not (Alvarez et al. 2007)). In this study field trials were carried out to assess the performance of the POCIS alongside spot sampling for monitoring a wide range of polar organic pollutants in surface water. Moreover, toxicity of the extracts from the field exposed samplers was evaluated to identify potential environmental hazards from compounds accumulated in the samplers during exposure, by using in vitro bioassays that detect endocrine-like and dioxin-like compounds. MATERIALS AND METHODS Materials POCIS samplers (version for sampling pharmaceuticals) were provided by Exposmeter AB (Tavelsjo, Sweden). Acetonitrile, dichloromethane, isooctane and methanol (HPLC reagent grade, Scharlau) were purchased from ICS (Instrument Consommable Service, Belin Belief, France). Glass solid phase extraction (SPE) cartridges of 6 mL with PTFE frits (20 um porosity) and Oasis HLB bulk sorbent (60 um) were purchased from Supelco (Saint Quentin-Falla-vier, France) and Waters (Guyancourt, France), respectively. Pharmaceuticals and hormones were provided by Sigma Aldrich (Saint Quentin Fallavier, France), polycyclic aromatic hydrocarbons (PAH) by LGC Standard (Molsheim, France), Alkylphenols and pesticides by Cluzeau (Sainte Foy La Grande, France). The studied compounds were: Pharmaceuticals (PHARM): amitryptiline, aspirin, caffeine, carbamazepine, diazepam, doxepin, gemfibrozil, ibuprofen, imipramine, ketoprofen, naproxen, nordiazepam, paracetamol, theophylline. Polycyclic aromatic hydrocarbons (PAHs): acenaphthene, acenaphthylene, anthracene, benzo(a)anthracene, benzo(a) pyrene, benzo(b)fluoranthene, benzo(j)fluoranthene, benzo(k) fluoranthene, benzo(e)pyrene, benzo(g,h,i)perylene, chrysene, dibenzo(a)anthracene, fluoranthene, fluorene, indeno(l,2, 3-c,d)pyrene, naphthalene, perylene, phenanthrene, pyrene. Hormones (HORM): 17a-ethynylestradiol, 17ß-estradiol, levonorgestrel, mestranol, norethindrone, estrone, progesterone. Pesticides (PEST): alachlore, atrazine, desethyl atrazine (DEA), desisopropyl atrazine (DIA), bifenthrin, chlorfenvin-phos, chlortoluron, methyl chlorpyrifos, chlorpyrifos, chlorsulfuron, cyanazine, cyfluthrine, Cypermethrin, cyroma-zine, l-(3,4dichlorophenyl)-3 methyl-urea (DCPMU), l-(3, 4dichlorophenyl)-urea (DCPU), l-(2,4dichlorophenyl)-urea, dichlorodiphenyltrichloroethane (DDT), dichlorodiphenyl-dichloroethylene (DDE), dichlorodiphenyldichloroethane (DDD), deltamethrin, diazinon, dichlorvos, dimethachlor, dimethoate, diuron, esfenvalerate, ethroprophos, fenithro-thion, fenvalerate, irgarol, isoproturon, lindane, linuron, malathion, metazachlor, metoxuron, nicosulfuron, permethrin, phosmet, prometryn, propachlor, propazine, pymetrozine, simazine, s-metolachlor, temephos, terbutryn, terbuthyla-zine, methyl tolclofos. Phenols (AKP): 4-nonylphenol (NP), 4-ter-octylphenol (OP), nonylphenol ethoxyacetic acid (NP1EC), 4-nonylphe-nol monoethoxylate (NP1EO), 4-nonylphenol diethoxylate (NP2EO), bisphenol A. Field experiments POCIS (pharmaceutical version) were exposed in Nerac in the surface water in the Bai'se River (Garonne basin, south west of France) (Figure 1). Two triplicates of POCIS (3 for chemical analysis and 3 in vitro bioassays, respectively) were deployed in May 2007 over a period of one month. 2420 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 Figure 1 | Sampling site in Nerac in the Baise River (Garonne River Basin, South west of France). During exposure POCIS were placed in a perforated canister made of high quality stainless steel to protect them from mechanical damage. Field control POCIS were used to follow an eventual contamination during transport and manipulation with samplers during deployment and retrieval. They were taken to the sampling site and exposed to the air during the immersion and the withdrawal of POCIS. Control POCIS were processed simultaneously and equally to the exposed samplers. Spot water samples were also collected at the beginning, in the middle and at the end of the one month exposure period, to compare with the data obtained with passive sampling for several groups of compounds (PAHs, pharmaceuticals, alkylphenols, hormones, organophosphate pesticides, organochlorine pesticides, pyrethroid pesticides, triazines, phenylurea herbicides). Chemical analysis POCIS: After exposure, each POCIS was rinsed with ultra pure water to remove particles and biofilms present on the outer surface of the membranes. Control POCIS were processed using the same procedure. The metal disks were disassembled and the membranes were detached from the disks. The sorbent was carefully transferred into an empty glass solid phase extraction (SPE) tube by rinsing it with ultrapure water. The sorbent was dried by applying vacuum for 1 h. Analytes were eluted by 30 mL of dichloromethane/methanol mixture (50:50 v/v). The extract was concentrated first by using a rapidvap vaccum evaporation system (25 min), then by a gentle stream of nitrogen and finally dissolved in 150 uL of a solvent suitable for injection to an analytical instrument. Water: Spot samples of water were collected during POCIS exposure. Water samples were collected to 4 L amber glass bottles. Before use, the bottles were detergent washed, acid rinsed and heated at 450 °C for 6 h. Immediately after collection, the samples were filtered through a glass fibre filter (GF/F 0.7 um pore size). The analytes were measured in the filtrate. Pharmaceuticals, alkylphenols, and phenylurea herbicides were analysed by LC/MS/MS. PAHs, hormones and remaining pesticides (triazines, organophosphate pesticides, pyrethroid pesticides) were analysed by GC/MS. The analytical procedures were adapted from Togola & Budzinski 2421 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 (2008) for the pharmaceuticals, from Labadie & Budzinski (2005) for the hormones, from Budzinski et al. (2000) for the PAHs, from Alder et al. (2006) for the pesticides analysis and from Cailleaud et al. (2007) for the alkylphenols. Procedural blanks were also regularly performed during the sample extraction process and all the results presented are corrected by taking the blank levels into account. The performance of analytical methods was checked by the extraction of a spiked sample in each series of analyses. Analysed compounds were quantified using internal standard calibration. The response factors of the various compounds were measured by injecting a mixture of standard reference solutions. Extraction recoveries POCIS sampler contains 200 mg of Oasis HLB sorbent enclosed between two polyethersulfone (PES) membranes. The membranes which confine the sorbent are compressed between two metal disks (5.4 cm ID). The total exchanging surface area of the membranes is about 46 cm2. The ratio surface area to mass of sorbent is 230 cm2g_1. To determine the extraction recovery of analytes from the Oasis HLB sorbent empty glass solid phase extraction tubes with PTFE frits were packed with 200 mg of Oasis HLB sorbent and triplicate solide phase extraction cartridges were placed on a Visiprep vacuum manifold (Supelco). Each cartridge was spiked with a mixed standard solution of pharmaceuticals, pesticides, alkylphenols, hormones and PAHs by adding 50 uL of a standard solution in ethyl acetate to the sorbent. The analytes were eluted from the sorbent by 30 mL of dichloromethane/methanol mixture (50:50 v/v), which is quite a large volume that has been optimized to maximize extraction recoveries. The extract was concentrated first by using a rapidvap vaccum evaporation system (25 min) and then by using a gentle stream of nitrogen. These stages of evaporation require about 45 min. The losses of compounds have been tested during the development of the extraction method; no significant losses were observed (less than 5%). After the evaporation steps, the extract was dissolved into a solvent suitable for instrumental analysis. Morover the validity of the methods analysis was confirmed by the extraction of a spiked sample for each series of analyses. Table 1 | Extraction recoveries of some compounds from the Oasis HLB sorbent (n = 3) Recovery (%) (n = 3) Compound Mean rsd PAH Acenaphthene 79 ±7 Acenaphthylene 92 ±12 Anthracene 99 ±4 B enz 0 (a) anthrac ene 86 ±11 Benzo(a)pyrene 99 ±3 Benzo(b,j,k)fluoranthene 88 ±8 Benzo(e)pyrene 101 ±5 Benzo(g,h,i)perylene 104 ±2 Chrysene 100 ±8 Dibenzo(a,c)anthracene 104 ±4 Fluoranthene 105 ±6 Fluorene 84 ±6 Indeno(l,2,3-c,d)pyrene 109 ±8 Naphthalene 108 ±13 Perylene 83 ±5 Phenanthrene 99 ±10 Pyrene 102 ±7 PHARM Amitryptiline 97 ±3 Aspirin 66 ±19 Caffeine 77 ±34 Carbamazepine 71 ±5 Diazepam 105 ±2 Doxepin 58 ±12 Gemfibrozil 97 ±7 Ibuprofen 71 ±9 Imipramine 82 ±11 Ketoprofen 64 ±12 Naproxen 61 ±17 Nordiazepam 99 ±1 Paracetamol 63 ±3 Theophylline 81 ±16 AKP NP 96 ±5 OP 75 ±2 NP1EC 78 ±3 NP1EO 97 ±6 NP2EO 85 ±9 BPA 92 ±9 HORM 17a-Ethynylestradiol 103 ±15 17ß-Estradiol 99 ±17 (continued) 2422 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 Table 1 I continued Table 1 I continued Recovery (%) (n = 3) Recovery (%) (n = 3) Compound Mean rsd Compound Mean rsd Levonorgestrel 94 ±16 Metoxuron 102 ±1 Mestranol 92 ±19 Nicosulfuron 131 ±3 Norethindrone 93 ±20 Permethrin 74 ±21 Estrone 108 ±18 Phosmet 95 ±25 Progesterone 98 ±15 Prometryn 112 ±9 PEST Alachlore 104 ±4 Propachlor 108 ±15 Atrazine 87 ±5 Propazine 97 ±4 Desethyl atrazine 92 ±3 Pymetrozine 76 ±37 Desisopropyl atrazine 106 ±5 Simazine 93 ±4 Bifenthrin 70 ±27 s-Metolachlor 123 ±5 Chlorfenvinphos 133 ±19 Temephos 83 ±22 Chlorotoluron 100 ±1 Terbutryn 96 ±9 Methyl chlorpyrifos 94 ±2 Terbuthylazine 134 ±11 Chlorpyrifos 100 ±1 Methyl tolclofos 97 ±1 Chlors ulfuron 128 ±1 Cyanazine 70 ±36 Cyfluthrine 72 ±22 Coupling of passive sampling with in vitro bioassays Cypermethrin 73 ±20 Cyromazine 85 ±4 In addition to the chemicals analysis, POCIS were also used DCPMU 71 ±1 for toxicity testing using bioassays. After exposure, the sorbent DCPU 73 ±2 was transferred into glass solid phase extraction tube for 124-Dichlorodiphenylurea 72 ±2 extraction. The organic compounds were eluted in 3 fractions: DDT+DDE+DDD 53 ±9 the first fraction (Fl) with 10 ml of dichloromethane, the Deltamethrine 61 ±23 second fraction (F2) with 10 ml of dichloromethane/metha- Diazinon 117 ±5 nol mixture (50:50 v/v) and the final fraction (F3) with 10 Dichlorvos 74 ±30 ml of methanol. Each fraction was analysed for all selected Dimethachlor 148 ±7 compounds. Toxicity tests were performed on each fraction. Dimethoate 104 ±14 The estrogenic, (anti-)androg enic and dioxin-like activities Diuron 94 ±1 of the extracts were assessed by using three in vitro bioassays Esfenvalerate 71 ±24 based on MELN (MCF-7 cells stably transformed with the fire- Ethroprophos 151 ±23 fly luciferase gene under the control of endogenous estrogen Fenithrothion 115 ±14 receptor; Balaguer et al. 2001), MDA-kb2 (MDA-MB-453 Fenvalerate 80 ±18 cells stably transformed with the firefly luciferase gene Irgarol 113 ±10 driven by a promoter regulated by endogenous androgen Isoproturon 107 ±1 receptor; Wilson et al. 2002^ and PLHC-1 (fish hepatoma Lindane 59 ±6 derived cells; Louiz et al. 2008) cell lines, respectively. Linuron 70 ±1 Description of cell lines and protocols for routine cell culture Malathion 108 ±2 and environmental sample assessment has been reported in Metazachlor 132 ±11 details previously (Louiz et al. 2008; Creusot et al. 2010). In (continued) brief, cells were seeded in 96-wells plates and left to grow up 2423 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 to confluence before being exposed to carrier solvent (negative control) and serial dilutions of reference ligand (positive control) and POCIS extracts (test sample). In the MELN and MDA-kb2 assays, cells were exposed for 16 h and processed for luciferase activity assay. In the PLHC-1 assay, cells were exposed for 4 h (PAH-like activity) and 24 h (dioxin-like activity) and then were processed for 7-ethoxyr-esorufin-O-deethylase (EROD) activity assessment in intact cells. Toxic-equivalent quantities relative to reference compounds in samples were determined by comparing modelled dose-response curves of samples and reference compounds, as previously described (Louiz et al. 2008). RESULTS AND DISCUSSION Extraction of POCIS and recoveries To determine the extraction recovery of analytes from the Oasis HLB sorbent, elution of analytes from spiked sorbent was performed. The percentage recoveries (n = 3) were higher than 70% for most of the compounds (Table 1). Only 9 compounds showed low extraction recovery. These were bifenthrin (70 ± 27%), cyanazine (70 ± 36%), aspirin (66 ± 19%), ketoprofen (64 ± 12%), paracetamol (63 ± 3%), naproxen (61 ± 17%), deltamethrin (61 ± 23%), lindane (59 ± 6) and DDT + DDE + DDD (53 ± 9). The applied extraction protocol is efficient for 91 compounds belonging to 5 classes including PAHs, pharmaceuticals, alkylphenols, hormones and pesticides. For each set of analysed samples, a spiked sample was processed in order to monitor the performance of the extraction protocol. Field exposures POCIS samplers were deployed in May 2007 in the Bai'se River (South West of France) during a period of one month. After exposure, sorbent was extracted to determine the mass of compounds (Ms) accumulated in POCIS. Only compounds found in POCIS are presented (Table 2). Assuming linear uptake of all contaminants in the sampler during field exposure, TWA concentrations of studied compounds in water were calculated from the amount of analytes accumulated in POCIS (Ms) using laboratory-derived Table 2 I Mass of analyte accumulated in the sorbent after an exposure time (Ms) and sampling rates (Rs) used for the calculation of the TWA concentration of sampled analytes (only compounds detected in field exposed samplers are shown) M, (ng) References Atrazine desisopropyl (DIA) 6 0.06 Mazzella et al. (2007) Atrazine desethyl (DEA) 18 0.12 Mazzella et al. (2007) Simazine Nd 0.31 Budzinski et al. (2009) Atrazine 2 0.33 Budzinski et al. (2009) Terbuthylazine 2 0.25 Mazzella et al. (2007) Promethryn 2 0.36 Personal data Terbuthryn 1 0.34 Personal data Lindane 2 0.09 Alvarez et al. (2007) I DDT 1 0.02 Alvarez et al. (2007) Diuron 23 0.25 Mazzella et al. (2007) Isoproturon 6 0.22 Mazzella et al. (2007) Metoxuron 35 0.20 Mazzella et al. (2007) Linuron 12 0.24 Mazzella et al. (2007) Chlorsulfuron 5 0.11 Alvarez et al. (2007) Nicosulfuron 119 0.04 Mazzella et al. (2007) Cafeine 13 0.08 Togola & Budzinski (2007) Carbamazepine 2 0.40 Togola & Budzinski (2007) Aspirine 15 0.01 Togola & Budzinski (2007) Paracetamol 18 0.02 Togola & Budzinski (2007) Gemfibrozil 3 0.05 Togola & Budzinski (2007) Diclofenac 2 0.17 Togola & Budzinski (2007) Nonylphenol 26 0.02 Budzinski et al. (2009) NP1EC 9 0.28 Personal data 17-Oestradiol (E2) 2 0.04 Zhang et al. (2008) Testosterone (T) 2 - No Rs data availabe sampling rates Rs: where, Cw is the TWA concentration in water over the sampler deployment period, Ms(t) is the mass of analyte 2424 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 □ Passive sampling ■ Spot sampling J t n ľ! ■ J ^ rl n „ fl ^ .-, 1 Figure 2 | Comparison of water concentrations determined by spot sampling and by passive sampling (n = 3). accumulated in the sorbent after an exposure time (t) and Rs is the sampling rate. Data (Rs and Ms) used for the calculation are shown in Table 2, unfortunately no Rs data are available for testosterone. The concentrations in water obtained by POCIS were compared to the water concentrations determined by spot sampling (Figure 2). Both sampling techniques are not fully comparable because spot sampling gives only a snapshot of contamination while POCIS provides an integrated concentration (the TWA value). Moreover, the filtering threshold used for water filtration is 0.7 um, while the pores of the membranes of POCIS are 0.1 um. Nevertheless, a good correlation between the concentration of water obtained by spot sampling and that obtained by passive sampling can be observed if the compounds are in sufficient concentration to be detected by passive sampling and if the concentration in water is constant over time. This was the case for DEA, lindane, diuron and NP (Figure 2). On the contrary several phenylurea pesticides, like nico-sulfuron, chlorotoluron, isoproturon, metoxuron, linuron, chlorsulfuron were detected by POCIS whereas they were not found in spot samples. This was also the case for pharmaceuticals gemfibrosil and diclofenac which were present in POCIS, but not detected in water samples. POCIS are able to accumulate quantifiable amounts compounds that were below the detection limit of the spot sampling method. Significant preconcentration of analytes from water together with integrative sampling that allows retention of episodic concentration peaks (normally not detected by low frequency spot sampling) enable POCIS to provide a better (more sensitive and more representative) information on the pollution of the sampled environment by polar organic compounds in comparison with spot sampling. Coupling of passive sampling with in vitro bioassays After exposures in the Bai'se River, some POCIS were fractionated and each fraction was tested for estrogenic, (anti) androgenic, PAH-like, and dioxin-like activities using in vitro bioassays. Estrogenic activity was detected in the Fl fraction and to a lesser extent in F2 (Table 3). The most abundant compounds in fraction Fl were alkylphenols and pesticides, including estrogenic ones like 4-tert-octylphenol, bisphenol A and DDT metabolites (Table 4). Unexpectedly, the F3 fraction, which contained trace levels of steroid hormones, was not estrogenic in the MELN bioassay. The two fractions (Fl and F2) exhibited PAH-like activity of 47.5 and 15.8 ng of BaP-EQ per POCIS, respectively. Accordingly, PAHs were mainly detected in Fl but not in F2. So the activity detected in F2 could be due either to a higher sensitivity of the bioassay 2425 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 Table 3 I Estrogenic, PAH-like, dioxin-like and androgenic activities detected in POCIS fraction using in vitro bioassays E2-EQ" BaP-EQ" TCDD-EQC DHT-EQ" POCIS fractions (ng/POCIS) (ng/POClS) (ng/POClS) (ng/POCIS) Fl 0.44 47.5 <0.6 <0.2 F2 0.06 15.8 <0.6 <0.2 F3 <0.02 <1.1 <0.6 <0.2 "E2-EQ: l7bestradiol-equivalents. °BaP-EQ: benzo(a)pyrene-equivalents. CTCDD-EQ: dioxin-equivalent. aDHT-EQ: dihydrotestosterone-equivalent. Table 4 I Quantity of studied compounds in POCIS Fraction (ng) Fraction 1 Fraction 2 Fraction 3 (ng) (ng) (ng) Pharmaceuticals 20 14 Nd PAHs 128 1 Nd Alkylphenols + BPA 36 12 3 Hormones nd nd 4 Triazine 31 nd Nd Pyrethroids nd nd Nd Organophosphate pesticides nd nd Nd Phenyl urea pesticide 212 5 7 or to compounds that were not targeted in this study, such as transformation product of PAHs. Finally, no dioxin-like or (anti)androgenic activity could be detected in these samples. CONCLUSIONS Field studies in which the results obtained with passive samplers are compared to those obtained with conventional sampling techniques increase the body of evidence that is available to underpin acceptance of the validity of passive sampling. The data sets obtained in this study show the effectiveness of the POCIS in integrative sampling of a broad range of organic chemicals in the surface water. Moreover, the potential of coupling chemical and toxicological characterization of water quality using passive samplers was demonstrated. However, the detection of biological activities that could not be explained by chemical analyses supports further investigation to identify biologically active substances sampled by POCIS. ACKNOWLEDGEMENTS The 'Agence de l'eau Adour Garone', the 'Region Aquitaine' and the 'Agence Nationale de la Recherche' (ANR) are acknowledged for their financial support, and by the French Ministry of Environment (grant P189-AP08 to INERIS). REFERENCES Alder, L., Greulich, K., Kempe, G. & Vieth, B. 2006 Residue analysis of 500 high priority pesticides: better by GC-MS or LC-MS/MS? Mass Spectrometry Reviews 25, 838-865. Alvarez, D. A., Huckins, J. N., Petty, J. P., Jones-Lepp, T., Stuer-Lauridsen, F. S., Getting, D. T., Goddard, J. P. & Gravel, A. 2007 Tool for monitoring hydrophilic contaminants in water. Polar organic chemical integrative sampler (POCIS). In: Passive Sampling Techniques in Environmental Monitoring (R. Greenwood, G. Mills & B. Vrana, eds). Wilson and Wilson Comprehensive Analytical Chemistry, Elsevier, Amsterdam, pp. 171-197. Balaguer, P., Boussioux, A. M., Demirpence, E. & Nicolas, J. C. 2001 Reporter cell lines are useful tools for monitoring biological activity of nuclear receptor ligands. Luminescence 16, 153-158. Budzinski, H., Letellier, M., Thompson, S., LeMenach, K. & Garrigues, P. 2000 Combined protocol for the analysis of polycyclic aromatic hydrocarbons (PAHs) and polychlorobiphenyls (PCBx) from sediments using focussed microwave assisted (FMW) extraction at atmospheric pressure. Fresenius' Journal of Analytical Chemistry 367 (2), 165-171. Budzinski, H., Soulier, C, Lardy, S., Capdeville, M. J., Tapie, N., Vrana, B., Miége, C. & Ai't-Aľssa, S. 2009 Xenobiotics in the Urban Water Cycle - XENOWAC 2009. Paphos, Cyprus. Cailleaud, K., Forget-Leray, J., Souissi, S., Lardy, S., Augagneur, S. & Budzinski, H. 2007 Seasonal variations of hydrophobic organic contaminant concentrations in the water-column of the Seine Estuary and their transfer to a planktonic species Eurytemora affinis (Calanoid, copepod). Part 2: Alkylphenols polyethoxylates. Chemosphere 70, 281-287. Commission Directive 2009 Commission Directive 2009/90/EC of 31 July 2009 laying down, pursuant to Directive 2000/60/EC of the European Parliament and of the Council, technical specifications for chemical analysis and monitoring of water status. Creusot, N, Kinani, S., Balaguer, P., Tapie, N., Maillot-Maréchal, E., Porcher, J. M., Budzinski, H. & Ai't-Aľssa, S. 2010 Evaluation of an hPXR reporter gene assay for the detection of aquatic emerging pollutants: screening of chemicals and application to water samples. Analytical and Bioanalytical Chemistry 396, 569-583. 2426 N. Tapie et al. | Passive samplers for Chemical substance monitoring and associated toxicity assessment in water water Science & Technology | 63.10 | 2011 Directive 2000 Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. Labadie, P. & Budzinski, H. 2005 Development of an analytical procedure for determination of selected estrogens and progestagens in water samples. Analytical and Bioanalytical Chemistry 381 (6), 1199-1205. Louiz, L, Kinani, S., Gouze, M. E., Ben-Attia, M., Menif, D., Bouchonnet, S., Porcher, J.-M., Ben-Hassine, O. K. & Ai't-Ai'ssa, S. 2008 Monitoring of dioxin-like. estrogenic and anti-androgenic activities in sediments of the Bizerta lagoon (Tunisia) by means of in vitro cell-based bioassays: contribution of low concentrations of polynuclear aromatic hydrocarbons (PAHs). Science of the Total Environment 402, 318-329. Mazzella, N., Dubernet, J.-F. & Delmas, F. 2007 Determination of kinetic and equilibrium regimes in the operation of polar organic chemical integrative samplers. Application to the passive sampling of the polar herbicides in aquatic environments. Journal of Chromatography A 1154, 42-51. Namiesnik, J., Zabiegala, B., Kot-Wasik, A., Partyka, M. & Wasik, A. 2005 Passive sampling and/or extraction techniques in environmental analysis: a review. Analytical and Bioanalytical Chemistry 381, 279-301. Stuer-Lauridsen, F. 2005 Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environmental Pollution 136, 503-524. Togola, A. & Budzinski, H. 2007 Development of polar organic integrative samplers for analysis of pharmaceuticals in aquatic systems. Analytical Chemistry 79, 6734-6741. Togola, A. & Budzinski, H. 2008 Multi-residue analysis of pharmaceutical compounds in aqueous samples. Journal of Chromatography A 177 (1), 150-158. Vrana, B., Mills, G. A., Allan, I. J., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G. & Greenwood, R. 2005 Passive sampling techniques for monitoring pollutants in water. Trends in Analytical Chemistry 24, 844-868. Wilson, V. S., Bobseine, K., Lambright, C. R. & Gray, L. E. 2002 A novel cell line. MDA-kb2 that stably expresses an androgen-and glucocorticoid-responsive reporter for the detection of hormone receptor agonists and antagonists. Toxicological Sciences 66, 69-81. Zhang, Z., Hibberd, A. & Zhou, J. L. 2008 Analysis of emerging contaminants in sewage effluent and river water: Comparison between spot and passive sampling. Analytica Chimica Acta 607, 37-44. Príloha 20 Prokeš R., Vraná B., and Klánová J., Levels and distribution of dissolved hydrophobic organic contaminants in the Morava river in Zlín district, Czech Republic as derived from their accumulation in silicone rubber passive samplers., Environ. Pollut., 2012,166,157-66. ENVIRONMENTAL POLLUTION Environmental Pollution 166 (2012) 157-166 Contents lists available at SciVerse ScienceDirect ELSEVIER Environmental Pollution journal homepage: www.elsevier.com/locate/envpol Levels and distribution of dissolved hydrophobic organic contaminants in the Morava river in Zlín district, Czech Republic as derived from their accumulation in silicone rubber passive samplers Roman Prokeš, Branislav Vraná*, Jana Klánová Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment RECETOX, Kamenice 126/3, 625 00 Brno, Czech Republic article info abstract Article history: Received 19 September 2011 Received in revised form 6 February 2012 Accepted 21 February 2012 Keywords: Dissolved concentration Hydrophobic organic compounds Monitoring Passive sampling Silicone rubber Water Dissolved waterborne polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs) were assessed over a period of one year at five sampling sites in a model industrial region in the Czech Republic using silicone rubber passive samplers. The spatial variability of POPs in the studied region in water was small and diffusive pollution sources predominate. Concentrations of the most volatile PAHs decreased with increasing water temperature in the whole region, which reflects the seasonality in atmospheric deposition. The dissolved concentrations of more hydrophobic PAHs, PCBs and OCPs in and downstream the industrial zone are related to desorption from suspended particles. Upstream the industrial area, a positive correlation of dissolved and particle-bound contamination was observed only for DDT metabolites and hexachlorobenzene. Calculated fugacities in water and bottom sediment indicated a fair degree of equilibrium between these compartments for OCPs and PCBs, whereas sediment represented a potential source of PAHs. © 2012 Elsevier Ltd. All rights reserved. 1. Introduction The freely dissolved concentration of persistent organic pollutants (POPs) in water is one of the important parameters for the assessment of their bioavailability and fate in the aquatic environment. It is generally assumed that particle and colloid-bound compounds cannot cross biological membranes, bioconcentrate and cause biological effects (Landrum et al., 1985). The freely dissolved concentration of POPs in the water column is directly proportional to their fugacity in the water phase (Mayer et al., 2003). Pollution monitoring based on direct water measurement of dissolved concentrations of POPs by bottle sampling is not reliable, as the individual spot samples of water collected at the sampling sites reflect only the pollution situation at the moment of sampling. Determination of extremely low (but toxicologically relevant) dissolved concentrations of hydrophobic compounds (levels below 1 ng L-1) is complicated since the loss of such trace amounts of analytes through volatilization, glassware adsorption and degradation during transport and sample processing steps (filtration and extraction). Moreover, measurement of truly dissolved concentration of these compounds in water cannot be easily * Corresponding author. E-mail address: vrana@recetox.muni.cz (B. Vrana). 0269-7491/S - see front matter © 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2012.02.022 achieved by conventional liquid/liquid or solid phase extraction techniques because of potential bias of these methods introduced by co-extraction of analytes bound to colloids present in water samples. Passive sampling techniques are widely applied to assess exposure and contamination in water, air and soils (Greenwood et al., 2007). Diffusion of organic pollutants from sampled media to the sampler is driven by the high affinity of analysed compounds to the sorbent material of the receiving phase in the sampler. The concentration found in a passive sampler can be used for calculation of time weighted average (TWA) water concentration over extended periods of time for environmental risk assessment, providing accurate calibration data is available. In this study, passive samplers made from polydimethylsiloxane (PDMS) sheets, better known as silicon rubber (SR), were deployed to characterize the spatial and temporal distribution of the hydrophobic organic pollutants including polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), organo-chlorinated pesticides (OCPs) and hexachlorobenzene (HCB) and their dynamics in the dissolved phase in the water column of the Morava river and its tributary Dřevnice in a model industrial area of Zlín in the Czech Republic (Fig. 1). Previous studies conducted in the Zlín region evaluated the risk related to POPs contamination of river sediments and alluvial soils (Hilscherova et al., 2007). A long-term monitoring showed different dynamics of PAHs and PCBs during 158 R. Prokeš et al. / Environmental Pollution 166 (2012) 157-166 0 2 4 6 km Popovice Fig. 1. Map of the sampling sites in the Zlin area, Czech Republic. floods when PAHs were redistributed from the sediments to alluvial soils while PCBs have been washed out of the study region. This paper presents particular results of a larger study aimed at characterization of contaminant distribution and dynamics in a fine temporal resolution between various aquatic compartments (surficial and suspended sediments, water) in the Zlín region. 2. Materials and methods 2.1. Materials and chemicals Organic solvents dichloromethane, methanol, n-hexane, cyclohexane and chloroform were obtained from Lab-Scan, Ireland and Sigma—Aldrich, Czech Republic. Standards of 16 polyaromatic hydrocarbons (PAHs), 6 polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs), HCB and internal standards (p-terphenyl, PCB 121) were obtained from Sigma—Aldrich, Czech Republic. Physico-chemical properties of analytes are given in Supplementary information. 2.2. Sampling sites A one year study was conducted from July 2007 to July 2008 at 5 sampling sites in the river Morava and its tributaries in the model impacted area close to the town of Zlín in south-eastern part of the Czech Republic (Fig. 1). It belongs to the basin of the river Danube and it includes a part of the major river Morava with its tributary Dfevnice. This area is an industrial and agricultural region with 10 cities and 72 villages, with the largest industrial city Zlín, it has high economic and cultural significance and is noted for its industrial and agricultural activities. Water and sediment in this area has been historically impacted by extensive industrial activities as well as agriculture. The five sites have been previously shown (based on analysis of contaminants in sediments and alluvial soils) to represent three regions within this area according to their location and contamination characteristics and the division into regions has been previously validated by cluster analysis (Hilscherova et al., 2007). Table 1 describes the sampling site locations. Actual water temperature and volume discharge data were obtained from Czech Hydrometeo-rological Institute. 2.3. Passive samplers Silicone rubber (SR) sheets from Altesil (Altec, Great Britain) were applied as passive samplers. The application was first time described by Smedes (2007). Rusina et al. (2007) characterized the polymer properties used for sampler construction. Among the materials tested, PDMS-based polymer Altesil showed the best overall performance including low release of oligomers, moderate swelling in solvents, fast diffusion coefficients of nonpolar compounds in the PDMS materials and high partition coefficients of hydrophobic compounds between the polymer and water. Rusina et al. (2010) calibrated the silicone rubber passive samplers and derived relations between the calibration parameters (sampling rates) and physicochemical properties of sampled compounds. Recently, Smedes et al. (2009) reported for a number of hydrophobic organic compounds (PAHs and PCBs) reliable partition coefficient data between silicone rubber polymer and water using a co-solvent method. The knowledge of sampling rates and polymer/water partition coefficients of hydrophobic organic compounds in combination with site specific exchange kinetics of contaminants between the sampler and water allow for the application of silicone rubber based passive samplers in quantitative measurement of these compounds dissolved in water. The wall thickness SR Altesil was 0.5 mm. SR were prepared using the procedure described by Rusina et al. (2007, 2010). SR sheets were cut into pieces of size 25 x 9.3 cm with surface area of =460 cm2. Two cleaning steps were applied to remove oligomers, other impurities and talcum powder from the surface. At first, SR was shaken in ethyl acetate for 1 d, and then Soxhlet-extracted in methanol for 12 h, wiped with a paper tissue and air dried in a fume-hood overnight. 2.4. Sampling campaign Before and after each passive sampler exposure, bottom sediment, suspended sediment and water samples were collected for analysis of target compounds. Temperature, pH, conductivity, oxygen content and water flow were also measured during each visit at the sampling site. SRs were transported to the sampling site in a cool box, wrapped in two layers of aluminium foil, and a polyethylene zip-lock bag. At the sampling site, two replicate samplers were placed in a stainless steel wire holder that was then suspended at depth of approximately 1 m below the water surface on a rope with a buoy, and secured to the shore using a rope. Weights were f* f* ľ"\\# R Prokeš et al. / Environmental Pollution 166 (2012) 157-166 159 Table 1 Description of sampling sites in the model study area. No. Sampling site Symbol Water body WGS84 WGS84 Altitude [m] Mean annual discharge3 [m3/s] 1. Bělov BE Morava river 49,21811 17,50136 185 44.74 2. Malenovice MA Dfevnice river 49,20772 17,55469 199 1.67 3. Spytihněv SP Morava river 49,13581 17,50211 182 49.92 4. Čerták — Morava CE Morava river 49,06719 17,43628 177 49.92 5. Čerták — branch CR Oxbow lake of the Morava river 49,06844 17,43561 181 a Calculated from volume discharge data available for the period July 2007—July 2008. b Volume discharge was not measured at this sampling site located in an oxbow lake. attached (using a rope) under the cage to prevent the cage from floating up in the current. Passive samplers were exposed to water column in 28-day deployment periods. Following exposure, two replicate SRs were removed from the holder, packed in two layers of aluminium foil, put in a polyethylene zip-lock bag and transported to the laboratory in a cooling box Passive samplers were replaced by fresh samplers. The monitoring continued for one year. The dates of deployment periods are given in Supplementary information (Table SI). Several samples were not retrieved due to bad accessibility, unsuitable deployment conditions (e.g. forming of a thick ice layer on water surface) and loss of sampler during field exposure. After exposure, samplers were stored in a freezer at -18 ° C until analysis. All samples were analysed for hydrophobic organic pollutants PAHs, PCBs, OCPs and HCB. 2.5. Sampling and analysis of water, suspended matter and surficial sediment Water samples were collected into amber glass bottles (2.5 L) with a screw cap. Samples were collected in the flow line 1 m below the water surface. Suspended matter was separated by filtration of the whole water sample through a glass fibre filter (Whatman, 2.2 |im pore size). Collection of surface sediments was performed as described in Hilscherova et al. (2007). Sample processing and analysis are described in Supplementary information. 2.6. Extraction and analysis of passive samplers Following exposure, SRs were rinsed with tap water and distilled water and then wiped with a paper tissue. SRs were Soxhlet-extracted for 12 h in methanol. The extracts were reduced in volume to 15 ml using Kuderna Danish concentrator. The final evaporation was provided with gentle stream of high purity nitrogen to about 2 ml. The samples were cleaned up using silica gel column (sulphuric acid modified silica gel was applied for organochlorines), reduced and internal standards were added. Terphenyl and PCB 121 were used as internal standards for PAHs and PCBs, respectively. Samples were analysed by GC/MS as described by Prokes et al. (2010). 2.7. Statistical analysis A total number of 70 samples of SR were analysed. The average value of analyte determination in two duplicate samples exposed under the same conditions was taken for analysis, thus the statistical analysis was performed on JV = 35 samples. Standard robust measures were applied for summary statistics of all examined parameters: estimate of median supplied with 10% and 90% quantiles. Non-parametric strategy was also applied for two samples (Median test) and multiple comparisons (Kruskal—Wallis test). Most concentration parameters revealed log-normal sample distribution (Kolmogorov Smirnov normality test) and therefore log transformation was applied prior to ANOVA analysis that requires normal distribution. Parametric strategy applied for two sample comparisons was also applied (Holm Sidak test). 3. Results and discussion 3.1. Calculation of dissolved water concentrations from passive sampler data The accumulation of chemicals by passive samplers is characterized by an initial linear uptake stage followed by curvilinear and equilibrium partitioning stages (Booij et al., 2007). In the initial uptake phase, chemical uptake is linear and thus integrative: N = tfsCwt (1) where N is the mass of a target compound in the sampler at time t, Rs is the sampling rate of the system and Cw is the concentration of a target analyte in water. Because of water boundary layer (WBL) control over the mass transfer to SR samplers for hydrophobic compounds, Rs is expected to vary depending on hydrodynamic conditions during exposure (Huckins et al., 1999; Rantalainen et al., 2000; Vrana and Schuurmann, 2002; Booij et al., 2003a). Estimation of the sampling rates in the absence of information on in situ exchange kinetics based on dissipation of performance reference compounds (PRCs) (Booij et al., 1998; Huckins et al., 2002) can result in significant errors. Their extent depends on the difference between the hydrodynamic conditions in the laboratory and the field, and on the application of the alternative method for minimizing the error in in situ Rs estimation. Such alternative approach was described in our previous work (Prokes et al., 2010) and has been applied in this study as follows. Booij et al. (2003a) expressed Rs for PCBs, PAHs and chloro-benzenes as a function of log Kow: Rs = ABw/C044 (2) where A is the sampler surface area and Bw is a constant for a given exposure, but may vary among exposures according to differences in hydrodynamic conditions and sampler geometry. Depending on models used for estimation of diffusion coefficients in water, the dependence for a water boundary layer controlled uptake may vary from Rs = K-°02 - K-°06 (Booij et al., 2003a). Rusina et al. (2010) confirmed this in a calibration of silicone strips for PAHs and PCBs. In situ sampling rates were estimated using Eq. (2). The value of Bw was determined for each field exposure. For this purpose, it was necessary to determine the absolute value of exposure specific Rs for at least one compound under investigation. The conditions for selection of such compound were: a) it can be found both in the passive sampler and in the water phase at quantifiable concentration; b) it is accumulated under WBL control and remains in linear uptake phase during the whole sampler exposure; c) in filtered water samples it is predominantly present in the dissolved phase. These conditions are most likely fulfilled for moderately hydrophobic compounds with log ^ow = 4—5, e.g. phenanthrene or anthracene. For such a compound, in situ Rs can be calculated using the rearranged Eq. (1): <-wr where Cw is the mean value of the compound concentration in spot samples of water taken during sampler exposure (two values; before and after exposure). It was assumed that concentration of anthracene did not widely fluctuate during exposure. A check was performed using Eq. (4) that the sampler does not equilibrate with water for the compound. In the next step, value of log ABW was calculated by substituting the calculated reference Rs value of anthracene into Eq. (2). Sampling rates of compounds with log Kow > 4.5 were then extrapolated using Eq. (2) with adjusted exposure specific log ABw value and the log Kow values of individual compounds. It has been verified that comparable Rs values are obtained, when the calculation of exposure specific log ABW is based on other reference compounds with similar physicochemical properties, such as phenanthrene or fluoranthene (Supplementary information, Table S3). R. Prokes et al. / Environmental Pollution 166 (2012) 157-166 160 3.2. Sampler equilibration time In the linear uptake mode, i.e. far from equilibrium, the aqueous concentration of analyte was calculated from the absorbed amount using rearranged Eq. (1) (Booij et al., 2003b). When the analyte has reached equilibrium, its aqueous concentration was calculated using the SR/water partition coefficient. An uptake remains essentially linear until 50% of equilibrium concentration is reached. The time interval it takes to reach the equilibrium concentration (tip) can be estimated (Booij et al.,2007): t1/2 = \n2 Kswms/Rs (4) where Ksw is the sampler/water partition coefficient, ms is the mass of the sampler. Ksw values published by Smedes et al. (2009) were used. For HCH isomers, Ksw values for bulk silicone material were not available and log Ksw = 2.6, obtained using solid phase micro-extraction fibre coated with polydimethylsiloxane was used (Paschke and Popp, 2003). A correction of Ksw for temperature was not performed (Booij et al., 2003a). HCHs, naphthalene, acenaphthylene, acenaphthene and fluorene, compounds with low values of partition coefficient (log Ksw < 3.7), equilibrate during 28 days in all sampler exposures and the equilibrium partitioning model was applied: msKsw For the remaining compounds, linear uptake model (Eq. (1)) was applied. This approach is illustrated in Table 2. Table 2 Mean concentrations of PAHs, PCBs and OCPs found in SR (ng per sampler: n = 2), calculated sampling rates Rs, estimates of concentrations from SR Cws, and concentrations measured in bulk water samples Q, collected at the beginning and the end of a 28 day sampler exposure at the site MA (period 14, Supplementary information, Table SI). Compound N[ng] Sampling Rs Cm 0,2 mode [Ld-1 IngL-1] IngL-1 1 IngL-1] Naphthalene 96 Equilib.3 n.e.b 6.72" 6.9 3.7 Acenaphthylene 31 Equilib. n.e. 0.59d 0.4 0.3 Acenaphthene 695 Equilib. n.e. 28.83d 4.5 3.6 Fluorene 814 Equilib. n.e. 10.10" 3.6 3.1 Phenanthrene 2648 Linear 14.0 6.77 7.9 7.1 Anthracene 293 Linear 13.9 0.75 0.7 0.8 Fluoranthene 3545 Linear 13.1 9.67 9.6 7.5 Pyrene 2078 Linear 13.0 5.69 7.5 6.5 Benz(a)anthracene 188 Linear 12.2 0.55 1.1 0.8 Chrysene 328 Linear 12.2 0.96 2.0 1.5 Benzo(b)fluoranthene 57 Linear 12.2 0.17 0.7 0.5 Benzo(k)fluoranthene 37 Linear 12.2 0.11 0.6 0.4 Benzo(a)pyrene 23 Linear 12.0 0.07 0.7 0.4 Indeno(l,2,3cd)pyrene 5 Linear 11.5 0.01 0.4 0.2 Dibenz(a,h)anthracene <1 Linear 11.2 <0.01 n.d.c n.d.c Benzo(g,h,i)perylene 6 Linear 11.5 0.02 0.6 0.4 PCB 28 29 Linear 12.5 0.08 0.1 0.1 PCB 52 6 Linear 12.2 0.02 <0.1 0.1 PCB 101 6 Linear 11.6 0.02 <0.1 <0.1 PCB 118 2 Linear 11.2 0.01 <0.1 <0.1 PCB 153 8 Linear 11.1 0.03 0.1 0.1 PCB 138 6 Linear 11.0 0.02 <0.1 <0.1 PCB 180 3 Linear 10.5 0.01 <0.1 <0.1 p,p'-DDE 53 Linear 12.4 0.15 0.2 0.1 p,p'-DDD 22 Linear 12.7 0.06 0.1 0.1 p,p'-DDT 6 Linear 11.8 0.02 0.1 0.1 a-HCH 24 Equilib. n.e. 5.21" 0.2 <0.1 y-HCH 18 Equilib. n.e. 6.37" 2.0 0.8 HCB 58 Linear 12.7 0.16 0.2 0.2 a Eq. — partitioning equilibrium between sampler and water has likely been achieved. b n.e. — not estimated. c n.d. — not detected. d Estimated using equilibrium partitioning model (Eq. (5)). 3.3. Relation between spot sample and passive sampler data With a few exceptions, a good correlation was obtained between water concentrations obtained from SR (Cws) and the mean water concentration value from filtered samples of water taken before and after sampler exposure sampling (Cb), assuming log-normal data distribution. Such comparison was only possible for compounds that were present at quantifiable concentrations in samples from both matrices. The mean value of the linear regression correlation coefficient from 30 exposures was 0.77. Results of this correlation for individual sampling sites and sampler deployment periods are available in Supplementary information (Table S4). Passive sampling provides concentrations that are lower than those obtained using from filtered spot samples of water. In general, the difference of both values increased with decreasing analyte concentrations; a typical example is shown in Fig. 2. Concentrations of compounds in samples decreased with their increasing hydrophobicity. An increased difference between spot and passive sampling is expected for more hydrophobic compounds (log Kow > 6) since passive samplers accumulate only dissolved chemicals, whereas even filtered water samples contain a significant fraction of compounds sorbed to colloids that can pass through the filter. This was confirmed, when the observed difference (expressed as log Cb - log Cws) was plotted against the compound hydrophobicity, as illustrated in Fig. 3. Concentrations found in filtered samples of spot water overestimated the truly dissolved concentrations of very hydrophobic compounds (log Kow > 6) by up to 2 orders of magnitude in most cases. This observation illustrates the usefulness of passive samplers for the measurement of truly dissolved concentrations of extremely hydrophobic compounds, which is not possible using conventional techniques, e.g. filtration of water samples followed by liquid—liquid extraction. -2-1012 Fig. 2. An example of relation between dissolved concentrations in water determined using SR samplers (log Cws) and those determined as a mean of two samples of water taken before and following passive sampler exposure (log Cb) at the site MA (period 14). Thin dashed lines present 95% confidence and prediction intervals, respectively. The thick dashed line indicates the equality of values. f* f* ľ"\\# R. Prokeš et at. / Environmental Pollution 166 (2012) 157-166 161 time for most studied contaminants with a few exceptions. A significantly elevated concentration of anthracene was observed at sites SP and CE in comparison with the site BE. Also, the sum of PAHs was significantly elevated at the site MA in comparison with all remaining sites. Finally, concentration of HCB was significantly lower at the site MA in comparison with the sites located in the main stream of the river Morava (BE, SP, CE). Low MW PAHs were found to be relatively more dynamic contamination components in the dissolved phase over the year than high MW PAHs. Also, the absolute concentrations in the dissolved phase were dominated by low MW PAHs, which can be explained by their better water solubility and lower adsorption to suspended solids. Highest median concentrations of all PAHs in water (excepting anthracene) were observed in river Dřevnice (site MA), which is likely due to the presence of local pollution sources as Dřevnice river collects pollution from the industrial agglomeration with chemical, plastic, rubber, shoe and machinery industry. The relatively small river with less than 5% average flow discharge of the river Morava at their confluence is more impacted by local pollution sources than the river Morava, where pollutants can be diluted more effectively. A similar pattern of PAHs was observed at all sampling sites, which indicates that the main pollution sources are similar (Fig. 4). Their spatial variability in water was relatively small, ranging between 7.8 and 25.7%, with the exception of anthracene, characterized by a higher variability of 47.7%. Elevated concentration and large fluctuation of anthracene in the area downstream the industrial zone of Zlín (site SP) is likely caused by specific contamination originating from a local point source, the anthraquinone producer DEZA Otrokovice, located between sites BE and SP on the river Morava. Regional distribution of recent sediment contamination in the Morava river also revealed a distinct anthracene peak just downstream of Otrokovice (Babek et al., 2008). Highest concentrations of PCBs, HCHs and DDT (but not HCB) were observed at the site MA. The contamination pattern by PCBs, OCPs and HCB was similar, which indicates that diffusive pollution sources dominate over local point sources (Fig. 5). The Table 3 Summarised dissolved concentrations of PAHs in water, derived from SR passive samplers, at the five sampling sites. Contaminants [ng L Median values (10%; 90% quantiles); calculated over the whole monitored period BE MA SP CE CR p valuec Components No. of measurements JV = 9 JV = 6 JV = 11 JV = 4 JV = 5 of variabilityd (Between sites differences) Naphthalene 15.87ab (4.91 28.20) 24.99ab (6.14; 35.20) 8.87ab (2.20; 27.40) 3.95ab (2.12; 25.80) 9.02ab (5.29; 12.84) 0.329 12.3% Acenaphthylene 0.69ab (0.28 3.92) 0.96ab (0.54; 3.41) 0.40ab (0.13; 4.66) 0.26ab (0.14; 3.18) 0.54ab (0.27; 1.20) 0.519 7.8% Acenaphthene 14.65ab (7.90 19.84) 20.56ab (8.70; 31.25) 10.61ab (1.40; 17.53) 8.54ab (1.93; 15.78) 9.25ab (4.53; 17.04) 0.317 n.a. Fluorene 8.60ab (4.71 10.05) 8.38ab (5.17; 14.66) 7.34ab (0.89; 12.05) 5.45ab (1.60; 9.43) 6.78ab (3.87; 10.70) 0.666 11.3% Phenanthrene 8.91ab (3.19 16.51) 12.52ab (7.75; 43.74) 4.78ab (2.56; 12.63) 6.25ab (2.40; 12.32) 3.84ab (2.08; 7.37) 0.072 21.3% Anthracene 0.90ab (0.29 1.33) 1.05ab (0.80; 4.28) 3.40ab (1.65; 8.70) 3.45ab (1.56; 5.55) 1.80ab (1.26; 3.19) 0.003 47.7% Fluoranthene 14.27ab (2.76 17.47) 18.36ab (10.37; 50.30) 10.97ab (5.08; 23.68) 8.74ab (6.44; 13.79) 7.74ab (4.84; 10.01) 0.226 18.5% Pyrene 7.58ab (1.37 11.00) 12.22ab (7.14; 30.74) 7.89ab (3.95; 14.93) 6.65ab (4.64; 9.99) 4.80ab (2.95; 6.22) 0.182 21.6% Benzo(a)anthracene 0.38ab (0.24 1.07) 1.24ab (0.71; 2.35) 0.88ab (0.35; 1.84) 0.66ab (0.35; 0.93) 0.53ab (0.37; 0.79) 0.187 20.0% Chrysene 0.83ab (0.30 1.82) 2.1 lab (1.13; 4.04) 1.43ab (0.55; 2.58) 1.07ab (0.62; 1.60) 0.97ab (0.58; 1.24) 0.215 18.9% Benzo(b)fluoranthene 0.18ab (0.08 0.39) 0.35ab (0.24; 0.65) 0.2 lab (0.11; 0.68) 0.26ab (0.15; 0.33) 0.26ab (0.17; 0.37) 0.455 14.4% Benzo(k)fluoranthene 0.10ab (0.03 0.19) 0.24ab (0.15; 0.39) 0.13ab (0.05; 0.41) 0.15ab (0.09; 0.30) 0.15ab (0.10; 0.22) 0.255 16.2% Benzo(a)pyrene 0.10ab (0.02 0.16) 0.15ab (0.10; 0.30) 0.25ab (0.05; 0.46) 0.10ab (0.05; 0.35) 0.09ab (0.06; 0.15) 0.099 25.0% Indeno( 1,2,3-cd)pyrene 0.03ab (0.01 0.06) 0.06ab (0.02; 0.10) 0.02ab (0.01; 0.10) 0.02ab (0.01; 0.02) 0.03ab (0.02; 0.04) 0.408 10.2% Dibenzo(a,h)anthracene 0.01ab (<0.01; 0.03) 0.02ab (0.01; 0.03) 0.01ab (<0.01; 0.02) <0.01ab (<0.01; 0.01) 0.01ab (<0.01; 0.01) 0.419 n.a. Benzo(g,h,i)perylene 0.03ab (0.01; 0.06) 0.06ab (0.03; 0.10) 0.02ab (0.01; 0.09) 0.03ab (0.02; 0.03) 0.03ab (0.02; 0.04) 0.352 9.5% £PAHs 76.22ab (27.82; 104.80) 93.19ab (77.90; 202.71) 72.15ab (40.81; 88.18) 47.77ab (25.18; 94.61) 48.37ab (42.84; 53.51) 0.012 25.7% n.a. — not analysed. ab Marks of statistical significance of multiple comparison tests between sampling sites. Values within one row marked by the same letter are not mutually significantly different (p > 0.05; multiple median test). c Overall p value of Kruskal—Wallis test comparing sampling sites. d Component of overall variability that belongs to the differences between sampling sites. This was calculated as ratios of relevant sum of squares (ANOVA model; based on log-transformed concentration data). Fig. 3. A typical relation of the observed difference between dissolved concentrations in water determined using SR samplers and those from spot samples of water (log Cb - log Cws) as dependent on compound hydrophobicity (log /Cow). The example is from the measurement at the site MA (period 14). The thin dotted line illustrates an empirical model that assumes sorption of compounds to colloids according to their hydrophobicity. The thick dashed line indicates the equality of values. 3.4. Space vs. time-related changes of dissolved contaminants Tables 3 and 4 show the overall summary of dissolved concentration of contaminants at the five sampling sites during one year monitoring campaign where the sources of variability are assessed. No significant differences could be observed among regions across 162 R. Prokeš et al. / Environmental Pollution 166 (2012) 157-166 Table 4 Summarised dissolved concentrations of PCBs, OCPs and HCB in water, derived from SR passive samplers, at the five sampling sites. Contaminants [ng L 1 ] Median values (10%; 90% quantiles); calculated over the whole monitored period BE MA SP CE CR p valuec Component of variability11 (Between sites differences) No. of measurements JV = 9 JV = 6 JV= 11 JV = 4 JV = 5 PCB 28 0.07 (0.01; 0.11) 0.17 (0.09; 0.51) 0.07 (0.03; 0.32) 0.21 (0.12; 0.27) 0.07 (0.04; 0.23) 0.130 24.6% PCB 52 0.02 (0.01; 0.05) 0.05 (0.02; 0.11) 0.02 (0.01; 0.08) 0.04 (0.03; 0.06) 0.03 (0.01; 0.05) 0.537 12.3% PCB 101 0.02 (0.01; 0.04) 0.03 (0.02; 0.09) 0.02 (0.01; 0.05) 0.03 (0.03; 0.04) 0.03 (0.02; 0.06) 0.239 15.2% PCB 118 0.01 (<0.01; 0.03) 0.01 (0.01; 0.06) 0.01 (<0.01; 0.02) 0.01 (0.01; 0.02) 0.01 (0.01; 0.02) 0.248 13.4% PCB 153 0.02 (0.01; 0.05) 0.05 (0.03; 0.12) 0.02 (0.01; 0.09) 0.05 (0.04; 0.05) 0.05 (0.03; 0.10) 0.112 20.8% PCB 138 0.01 (<0.01; 0.04) 0.03 (0.01; 0.10) 0.01 (0.01; 0.05) 0.03 (0.02; 0.04) 0.03 (0.02; 0.05) 0.140 17.5% PCB 180 0.01 (<0.01; 0.04) 0.02 (0.01; 0.07) 0.01 (<0.01; 0.02) 0.02 (0.02; 0.02) 0.02 (0.01; 0.05) 0.065 21.5% £PCBs 0.17 (0.04; 0.37) 0.44 (0.18; 0.98) 0.14 (0.09; 0.57) 0.38 (0.28; 0.49) 0.24 (0.14; 0.56) 0.203 18.9% a-HCH 3.57 (1.39; 5.99) 6.31 (1.56; 20.91) 4.09 (2.38; 12.58) 4.78 (2.14; 11.23) 2.51 (1.91; 5.12) 0.661 8.7% y-HCH 3.67 (0.18; 15.63) 11.59(5.31; 74.74) 2.05 (0.18; 7.07) 4.23 (1.29; 12.98) 2.93 (1.64; 5.00) 0.061 24.3% £HCHs 6.98 (3.36; 22.65) 17.16(11.62; 91.65) 6.14 (3.04; 18.82) 7.15 (5.27; 23.85) 4.67 (3.94; 10.12) 0.087 23.9% p,p'-DDE 0.08 (0.02; 0.11) 0.14 (0.06; 0.55) 0.06 (0.04; 0.35) 0.15 (0.13; 0.23) 0.08 (0.05; 0.11) 0.132 18.4% p,p'-DDD 0.07 (0.01; 0.12) 0.10 (0.04; 0.27) 0.03 (0.02; 0.19) 0.09 (0.08; 0.11) 0.06 (0.04; 0.09) 0.467 10.8% p,p'-DDT 0.01 (0.00; 0.05) 0.02 (0.01; 0.09) 0.01 (<0.01; 0.01) 0.01 (<0.01; 0.02) <0.01 (<0.01; 0.01) 0.083 14.6% J2 DOTS 0.15 (0.04; 0.29) 0.26 (0.11; 0.90) 0.10 (0.07; 0.57) 0.25 (0.25; 0.31) 0.15 (0.09; 0.21) 0.247 20.1% HCB 1.44a (0.63; 2.67) 0.21b(0.17; 0.75) 1.28a (0.39; 4.54) 1.69a(1.20; 2.10) 0.42ab (0.23; 0.72) 0.007 27.2% abMarks of statistical significance of multiple comparison tests between sampling sites. Values within one row marked by the same letter or unmarked are not mutually significantly different (p > 0.05; multiple median test). c Overall p value of Kruskal—Wallis test comparing sampling sites. d Component of overall variability that belongs to the differences between sampling sites. This was calculated as ratios of relevant sum of squares (ANOVA model; based on log-transformed concentration data). spatial component of variability in water was small, ranging between 8.7 and 27.2%. The compounds concentrations of which are susceptible of covarying in the environment were identified in this study on the basis of the correlation coefficient values. This statistical approach is based on the fact that each pollution source produces a characteristic compound pattern; so, the correlation factors between the concentrations of all the individual compounds can give an idea whether they all originate from the same source or not (Socio et al., 2000). At sites in and downstream the industrial zone of Zlín (MA, SP), concentrations of PAHs with 3—5 aromatic rings in water positively correlate with concentrations of PCBs, DDT congeners and HCB (Supplementary information; Tables S6 and S7). The industrial complex of Zlín presents an over 100-year old environmental burden with multiple contaminant sources and deserves a more detailed investigation in the future. Contaminant pattern observed upstream the industrial zone of Zlín (site BE) is different (Supplementary information; Tables S5), which can be explained by a different type of human activities (agriculture) prevailing in that area. 3.5. Effect of environmental variables To investigate effects of environmental variables, i.e. temperature, flow and suspended particulate matter (SPM) content on the concentration of pollutants in the water column, data on mean temperature, volume discharge, SPM content and concentration of analytes in water bound on SPM were correlated with concentrations of individual compounds in water, estimated from SR. The correlation was possible only at sites with the minimum of six measurements per year (BE, MA, SP). Correlations are shown in Tables 5—7. Water temperature and volume discharge were negatively correlated at all investigated sampling sites, with correlation coefficients ranging from -0.75 and -0.80, respectively. SPM did not significantly correlate with temperature or volume discharge at any of the three sites. 3.5.1. Effect of temperature A clear trend in concentration decrease with increasing water temperature was observed at sites BE, MA, SP for the two most volatile PAHs, naphthalene and acenaphthylene, respectively (Fig. 6). A weak trend of concentration decrease of fluorene and phenanthrene with increasing temperature was also observed at 100 <# <& <& <# „ 4.5) measured in the dissolved phase at sites MA and SP reflect the release of contaminants from polluted suspended particles in the water column. The particles likely appear in the water column as a result of emission, dry and wet atmospheric deposition and soil erosion from industrial zones. 3.6. Sediment/water fugacity ratios To assess the net flux of PAHs between water and sediment at the sampling sites, fugacity ratios (ratio of the fugacity in the sediment/s to the fugacity in the water/w) were calculated using the passive sampler — derived dissolved concentration (Cw) and the sediment concentration data (Cs) (Mackay, 1979; Di Toro et al., 1991). fs_ = Cs 1 /w Cw/ocpKoc The derivation of Eq. (6) has been shown previously (Vrana et al., 2001); /oc is the fraction of sediment organic carbon, p is the sediment bulk density (relative to water) and Koc is the sediment organic carbon—water partition coefficient. Koc was calculated using Karickoffs approximation (Karickoff, 1981), i.e. Koc ~ 0.41 x KoW-/oc measured in sediments ranged between 2.1% and 5.3%. The substitution of Karickoffs equation by alternative correlations proposed to estimate Koc from Kow (Sabljic et al., 1995; Baker et al., 2000) yields comparable PAH concentrations in pore water (of the same order of magnitude). The fugacity ratio can be cautiously interpreted as an indication of sediment—water equilibrium status. A ratio of unity indicates equilibrium, a ratio of less than unity indicates net flux from water to sediment and a ratio of more than unity indicates net flux from sediment to water. A fair degree of equilibrium (fslfw between 0.2 and 5) exists between the pore water and the overlying water for organochlorine pesticides and PCBs at sites BE, SP and MA. An example for the SP site is shown in Fig. 8. During the monitored period of one year, investigated sediments presented neither a contaminant sink nor a significant pollutant source for these compounds. By contrast, the sediment at all three locations are a significant potential source of ▼ □ □ o * o • Nap (SP) 0 Ace (SP) ▼ Nap (MA) a Ace (MA) ■ Nap (BE) □ Ace (BE) ▼ ■ □ o 0 5 10 15 20 25 Temperature [°C] Fig. 6. The dependence of naphthalene (Nap) and acenaphthylene (Ace) concentration on temperature at sampling sites BE, MA and SP, respectively. • Anthracene o Fluoranthene ▼ PCB 28 a PCB 52 o 0 00 0 0 • • o ° •2. 0 • 0 ••• • • ▼ T T T a a a ▼ ▼ ▼ a a aa a 1 10 1 100 SPM content (mg/L) Fig. 7. The dependence of anthracene, fluoranthene, PCB 28 and PCB 52 dissolved concentrations on average SPM content in water samples collected before and after sampler exposure at the site SP. R. Prokeš et al. / Environmental Pollution 166 (2012) 157-166 165 a 1000 1 it I I iy it u I Ü ra -a ±= q- Compound £ - Ü o o o • 1 i dj 1 I o O Ü ü a a I i q q. E q. 'a. alp E d co Compound Fig. 8. Sediment/water fugacity ratios of the PAHs (a), PCBs and OCPs (b) at the sampling site SP, calculated as described in the text. The ratio range between 0.2 and 5 is approximately where the sediment is predicted to be close to equilibrium with the aqueous phase. The boundaries of the box indicate the 25th and 75th percentile, a line within the box marks the median. Data points that lie outside the 10th and 90th percentiles (whiskers) are shown as symbols. PAHs, with the concentrations in the pore water being up to three orders of magnitude higher than in the water column. However, the fugacity in sediment may be overestimated by the presence of PAHs sorbed to soot particles in sediment. These were co-extracted during the analysis of PAHs in sediment, but they are not readily available for desorption to sediment pore water. Thus, the estimate of PAH flux between sediment and water column is not unambiguous without further experiments that will characterize the PAHs fraction that can be mobilized to water in a reasonable timescale. 4. Conclusions Passive samplers provide a reliable tool for an assessment of long-term pollution status within the water bodies. For compounds accumulated under WBL control, however, fluctuations in water flow velocities result in significant fluctuations of the sampling rates making it more difficult to derive truly dissolved concentrations of organic pollutants from their amounts accumulated in the sampling media. In situ assessment of the sampling rates based on dissipation rates of the performance reference compounds (PRCs) is therefore strongly recommended in the field studies. In this study, we demonstrated application of an alternative method that can reduce bias in estimation of the truly dissolved concentrations in the absence of PRCs. The passive samplers provide complementary information to the sediment samples. Sediment concentration patterns may not be representative for estimation of bioavailable concentrations in the upper levels of the water column as they provide a long-term contamination record which is further a subject to change due to weathering and ageing. On the contrary, the passive samplers integrate water concentrations only during the sampling period and reflect the actual pollution situation in a water body. The spatial variability of dissolved POPs in the studied region was relatively small, which indicates that diffusive pollution sources dominate over local point sources. The only exception was anthracene. Elevated and fluctuating concentrations, as well as a unique ratio of anthracene to phenanthrene concentration downstream the city of Zlín can be related to a specific industrial point source of this compound. Concentrations of the most volatile PAHs decreased with increasing water temperature in the whole region indicating that atmospheric emission from domestic heating sources and consequent dry and wet deposition represent an important pathway of PAHs to aquatic ecosystem of the region in the winter period. Distinct hydrophobic contaminant (log Kow > 4.5) distribution patterns between dissolved phase and SPM were observed upstream and downstream the industrial zone of Zlín. Dissolved phase concentrations of these compounds were positively correlated with their particle-bound concentrations downstream the industrial zone (sites MA, SP) indicating release of contaminants from polluted suspended particles deposited to the water bodies via dry and wet atmospheric deposition and soil erosion. Upstream (site BE), a positive correlation of dissolved water concentration with SPM content was observed only for hydrophobic organochlorine pesticides (DDT including metabolites and HCB) suggesting erosion of soil particles from agricultural areas, and re-suspension of sediment particles during the high water discharge events. Inspection of the sediment/water fugacity ratios revealed a fair degree of equilibrium between the pore water and the overlying water for organochlorine pesticides and PCBs. In contrast, the calculated concentrations of PAHs in pore water were up to three orders of magnitude higher than those in the water column indicating that sediments can act as a potential pollution source. However, the applied model did not account for the specific sediment carbon composition or quality that affects the availability of PAHs for desorption. Our study provided insight into the spatial and temporal variability of bioavailable concentrations of POPs in aquatic ecosystem and their potential sources. Similar field studies increase the body of information available for assessment of factors that affect distribution and fate of POPs in the natural environment. Moreover, they support regulators in assessing opportunities for using passive sampling for monitoring water quality within a legislative framework. Acknowledgements This research was supported by the EU Operational Programme "Research and Development for Innovations", the CETOCOEN project (no.CZ.1.05/2.1.00/01.0001). f* f* ľ"\\# 166 R. Prokeš et al. / Environmental Pollution 166 (2012) 157-166 Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.envpol.2012.02.022. References Babek, O., Hilscherova, IC, Nechyba, S., Zeman, J., Famera, M., Franců, J., Holoubek, I., Machat, J., Klanová, J., 2008. Contamination history of suspended river sediments accumulated in oxbow lakes over the last 25 years. Journal of Soils and Sediments 8,165-176. Baker, J.R., Mihelcic, J.R., Shea, E., 2000. Estimating Koc for persistent organic pollutants: limitations of correlations with Ji0w- Chemosphere 41, 813—817. Booij, IC, Hofmans, H.E., Fischer, C.V., Van Weerlee, E.M., 2003a. Temperature-dependent uptake rates of nonpolar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environmental Science and Technology 37, 361—366. Booij, IC, Hoedemaker, J.R., Bakker, J.F., 2003b. Dissolved PCBs, PAHs, and HCB in pore waters and overlying waters of contaminated harbor sediments. Environmental Science and Technology 37, 4213—4220. Booij, IC, Sleiderink, H.M., Smedes, F., 1998. Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environmental Toxicology and Chemistry 17,1236—1245. Booij, IC, Vraná, B., Huckins, J.N., 2007. Theory, modeling an calibration of passive samplers used in water monitoring, in: Greenwood, R., Mills, G., Vraná, B. (Eds.), Passive Sampling Techniques in Environmental Monitoring. Comprehensive Analytical Chemistry Series, vol. 48, Barcelo, D. (Series Ed.), Elsevier, Amsterdam, 141-169 pp. Di Toro, D.M., Zarba, CS., Hansen, D.J., Berry, W.J., Swartz, R.C, Cowan, C.E., Pavlou, S.P., Allen, H.E., Thomas, N.A., Paquin, PR., 1991. Technical basis for establishing sediment quality criteria for nonionic organic chemicals using equilibrium partitioning. Environmental Toxicology and Chemistry 10,1541—1583. Greenwood, R., Mills, G., Vraná, B., 2007. Passive sampling techniques in environmental monitoring. In: Barcelo, D. (Ed.), 2007. Comprehensive Analytical Chemistry Series, vol. 48. Elsevier, Amsterdam. Hilscherova, IC, Dušek, L, Kubik, V., Čupr, P., Hofman, J., Klanová, J., Holoubek, I., 2007. Redistribution of organic pollutants in river sediments and alluvial soils related to major floods. Journal of Soils and Sediments 7,167—177. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Cranor, W.L, Clark, R.C, Mogensen, B.B., 2002. Development of the permeability/performance reference compound (PRC) approach for in situ calibration of semipermeable membrane devices (SPMDs). Environmental Science and Technology 36, 85-91. Huckins, J.N., Petty, J.D., Orazio, C.E., Lebo, J.A., Clark, R.C, Gibson, V.L, Gala, W.R., Echols, ICR., 1999. Determination of uptake kinetics (sampling rates) by lipid-containing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water. Environmental Science and Technology 33, 3918-3923. Karickoff, S.W., 1981. Organic pollutant sorption in aquatic systems. Chemosphere 10, 833-846. Landrum, PF, Reinhold, M.D., Nihart, S.R., Eadie, B.J., 1985. Predicting the bioavailability of organic xenobiotics to Pontoporeia Hoyi in the presence of humic materials and natural dissolved organic matter. Environmental Toxicology and Chemistry 4, 459—467. Mackay, D., 1979. Finding fugacity feasible. Environmental Science and Technology 13,1218-1223. Mayer, P., Tolls, J., Hermens, L, Mackay, D., 2003. Equilibrium sampling devices. Environmental Science and Technology 37,184A—191A. Paschke, A., Popp, P., 2003. Solid-phase microextraction fibre-water distribution constants of more hydrophobic organic compounds and their correlations with octanol-water partition coefficients. Journal of Chromatography A 999, 35—42. Prokes, R., Vrana, B., Klanová, J., Kupec, J., 2010. Calibration of three passive samplers of hydrophobic organic compounds in water: assessment of critical issues in experimental design, data interpretation and field application. Fresenius Environmental Bulletin 19, 2812—2822. Rantalainen, A.-L, Cretney, W.J., Ikonomou, M.G., 2000. Uptake rates of semipermeable membrane devices (SPMDs) for PCDDs, PCDFs. Chemosphere 40, 147-158. Rusina, T., Smedes, F., Koblizkova, M., Klanová, J., 2010. Calibration of silicone rubber passive samplers: experimental and modeled relations between sampling rate and compound properties. Environmental Science and Technology 44, 362-367. Rusina, T.P, Smedes, F., Klanová, J., Booij, IC, Holoubek, I., 2007. Polymer selection for passive sampling: a comparison of critical properties. Chemosphere 68, 1344-1351. Sabljic, A., Güsten, H., Verhaar, H., Hermens, J., 1995. QSAR modelling of soil sorption. Improvements and systematics of log/Coc- Chemosphere 31, 4489—4514. Smedes, F, 2007. Monitoring by passive sampling in concert with deployed mussels, in: Greenwood, R., Mills, G., Vrana, B. (Eds.), Passive Sampling Techniques in Environmental Monitoring. Comprehensive Analytical Chemistry Series, vol. 48, Barcelo, D. (Series Ed.), Elsevier, Amsterdam, 407^153 pp. Smedes, F., Rinze, W., Geertsma, A., Van der Zande, T., Booij, IC, 2009. Polymer-water partition coefficients of hydrophobic compounds for passive sampling: application of cosolvent models for validation. Environmental Science and Technology 43, 7047-7054. Socio, H.H., Garrigues, PH., Ewald, M., 2000. Origin of Polycyclic Aromatic Hydrocarbons (PAHs) in Coastal Marine sediments: case studies in Cotonou (Benin) and Aquitaine (France) areas. Marine Pollution Bulletin 40, 387—396. Vrana, B., Paschke, A., Popp, P., 2001. Polyaromatic hydrocarbon concentrations and patterns in sediments and surface water of the Mansfeld region, Saxony-Anhalt, Germany. Journal of Environmental Monitoring 3, 602—609. Vrana, B., Schüürmann, G., 2002. Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics. Environmental Science and Technology 36, 290-296. Príloha 21 Jarošová B., Bláha L, Vraná B., RandákT., Grabic R., GiesyJ. P., and Hilscherová K., Changes in concentrations of hydrophilic organic contaminants and of endocrine-disrupting potential downstream of small communities located adjacent to headwaters, Environ. Int., 2012, 45, 22-31. Environment International 45 (2012) 22-31 ELSEVIER Contents lists available at SciVerse ScienceDirect Environment International journal homepage: www.elsevier.com/locate/envint ENVIRONMENT INTERNATIONA! '3 Changes in concentrations of hydrophilic organic contaminants and of endocrine-disrupting potential downstream of small communities located adjacent to headwaters B. Jarosova a, L. Blaha a, B. Vrana a, T. Randak b, R. Grabic b, J.P. Giesy c-d-e-{-z, K. Hilscherova a'* a Research Centre for Toxic Compounds in the Environment (RECETOX), Faculty of Science, Masaryk University, Kamenice 126/3, 62500, Brno, Czech Republic b University of South Bohemia in Ceske Budejovice, Faculty of Fisheries and Protection of Waters, South Bohemian Research Center ofAquaculture and Biodiversity of Hydrocenoses, Zatisi 728/11, 389 25 Vodnany, Czech Republic c Department of Biomedical Veterinary Sciences and Toxicology Centre, University of Saskatchewan, Saskatoon, Saskatchewan, Canada d Zoology Dept. and Center for Integrative Toxicology, Michigan State University, East Lansing, MI 48824, USA e Department of Biology and Chemistry, City University of Hong Kong, Hong Kong SAR, PR China ' Zoology Department, College of Science, King Saud University, P. 0. Box 2455, Riyadh 11451, Saudi Arabia g Environmental Science Program, Nanjing University, Nanjing, PR China ARTICLE INFO ABSTRACT Article history. Received 7 January 2012 Accepted 4 April 2012 Available online 8 May 2012 Keywords: Androgen Dioxin-like activity Estrogen In vitro assay POCIS Waste Water Treatment Plant Endocrine-disruptive potential and concentrations of polar organic contaminants were measured in seven headwaters flowing through relatively unpolluted areas of the Czech Republic. Towns with Wastewater Treatment Plant (WWTP) discharges were the first known sources of anthropogenic pollution in the areas. River water was sampled several kilometers upstream (US) and several tens of meters downstream (DS) of the WWTP discharges, by use of Pesticide and Pharmaceutical Polar Organic Integrative Samplers (POCIS-Pest, POCIS-Pharm). Extracts of passive samplers were tested by use of a battery of in vitro bioassays to determine overall non-specific cytotoxicity, endocrine-disruptive (ED) potential and dioxin-like toxicity. The extracts were also used for quantification of polar organics. There was little toxicity to cells caused by most extracts of POCIS. Estrogenicity was detected in all types of samples even though US locations are considered to be background. At US locations, concentrations of estrogen equivalents (EEq) ranged from less than the detection limits (LOD) to 0.5 ng EEq/POCIS. Downstream concentrations of EEqs ranged from less than LOD to 4.8 ng EEq/POCIS. Concentrations of EEq in POCIS extracts from all DS locations were 1 to 14 times greater than those at US locations. Concentrations of EEq measured in extracts of POCIS-Pest and POCIS-Pharm were in a good agreement. Neither antiestrogenic nor anti/androgenic activities were detected. Concentrations of 2,3,7,8-TCDD equivalents (TEqbio) were detected in both types of POCIS at concentrations ranging from less than the LOD to 0.39 ng TEqbio/POCIS. Nearly all extracts of POCIS-Pharm contained greater concentrations of TEqbio activity than extracts of POCIS-Pest. Concentrations of pesticides and pharmaceuticals in extracts of POCIS were generally small at all sampling sites, but levels of some pharmaceuticals were significantly greater in both types of POCIS from DS locations. Chemical analyses along with the results of bioassays documented impacts of small towns with WWTPs on headwaters. © 2012 Elsevier Ltd. All rights reserved. Abbreviations: AEq, androgenic equivalent; AhR, Aryl hydrocarbon receptor; DS, downstream; El, estrone; E2,17(3-estradiol; E3, Estriol; EC, effective concentration; ED, endocrine disruption; EDCs, endocrine disruptive compounds; EE2,17a-ethynylestradiol; EEq, estrogenic equivalent; HpOCs, hydrophilic organic compounds; Kow, octanol-water partition coeffident; LOD, limit of detection; LOQ, limit of quantification; NR, Neutral Red; PCBs, poly-chlorinated biphenyls; PCDDs, polychlorinated dibenzodioxins; PCDFs, polychlorinated di-benzofurans; PNEC, Predicted No Effects Concentration; POCIS, Polar Organic Chemical Integrative Sampler; POCIS-Pest, Polar Organic Chemical Integrative Sampler optimized for polar Pestiddes; POCIS-Pharm, Polar Organic Chemical Integrative Sampler optimized for most Pharmaceuticals; Rs, sampling rate (L/day); TEqbio, dioxin-like equivalent obtained in bioassay; TCDD, 2,3,7,8-tetrachlorodibenzo-p-dioxin; US, upstream; WWTP, Waste Water Treatment Plant * Corresponding author. E-mail address: hilscherova@recetox.muni.cz (K. Hilscherova). 0160-4120/$ - see front matter © 2012 Elsevier Ltd. All rights reserved, doi: 10.1016/j.envint.2012.04.001 1. Introduction Municipal and industrial waste waters can be sources of compounds that are able to cause acute toxicity as well as sublethal chronic abnormalities including disruption of hormonal balance in aquatic organisms (endocrine disruption, ED). Persistent and bioaccumulative organic chemicals have been conventionally monitored, but less persistent and less hydrophobic organic compounds are currently used as pesticides, prescription and non-prescription drugs and personal care products. Despite their lesser bioconcentration potential, relatively large fluxes of some of these compounds into aquatic systems might be acutely toxic and/or induce sublethal chronic abnormalities (Alvarez B. Jarošova et al. / Environment International 45 (2012) 22-31 23 et al., 2007). Furthermore, some of these chemicals (particularly pharmaceuticals) can be highly potent, such that even concentrations at or near analytical detection limits may have biological activity. Concentrations and/or ecotoxicological effects of hydrophilic organic compounds (HpOCs, contain one or more polar functional groups or a significant molecular dipole moment) have been reported in discharges of Waste Water Treatment Plants (WWTP) and/or downstream receiving waters (Aguayo et al., 2004; Bolong et al, 2009; Caliman and Gavrilescu, 2009). Downstream reaches of rivers have been shown to be polluted by compounds of both industrial and communal origin (Bolong et al., 2009), and therefore it is difficult to evaluate contributions and effects of pollutants released by individual towns. There are fewer sources of HpOC pollution in the headwaters and their potential impacts are not easy to assess, since there is limited information on concentrations of pollutants in the background areas. Although different groups of HpOCs can contribute to adverse effects, xenoestrogens and xenoandrogens have emerged as environmental issues due to their ability to mimic or otherwise adversely affect functions of natural reproductive hormones, which could result in impaired reproduction of aquatic organisms (Matthiessen and Johnson, 2007). Even though the efficiencies of conventional WWTPs with activated sludge systems to remove estrogenic and androgenic compounds seem to be relatively high (88-> 99% for estrogens and 96-> 99% for androgens (Korner et al., 2000; Leusch et al., 2010; Murk et al, 2002; Svenson and Allard, 2004), concentrations of these endocrine disruptive compounds (EDCs) in some effluents are sufficient to cause ED (Kirk et al, 2002). Since some EDCs can cause adverse effects at small concentrations (ng/L), it is difficult and expensive to detect them by instrumental analyses (Korner et al, 2000). Moreover, because they occur in mixtures, even if they can be quantified, it is difficult to predict the potential effects of these compounds (Leusch et al, 2005). Therefore, in vitro bioassays can serve as cheaper and more environmentally relevant alternative to screen for the combined effects of mixtures on specific biological endpoints (Kinnberg, 2003). The most frequently reported effect connected with EDs in surface waters is feminization of male fish downstream of WWTPs (Jobling and Tyler, 2003). Among estrogenic EDCs, the steroidal estrogens estrone (El), estradiol (E2), and synthetic estrogen analogue, ethinyl estradiol (EE2), are some of the most potent endocrine disrupters in sewage effluents, all having more than thousand times greater potency to cause ED, at least in fish, than most other xenobiotics (Young et al., 2004). Under environmental conditions, steroidal hormones have been identified to be primarily responsible for observed adverse estrogenic effects on fish downstream of WWTPs although other weakly estrogenic compounds, such as alkylphenols and bisphenol A, can contribute to the effects (Desbrow et al, 1998; Gross-Sorokin et al, 2006). Important is also the fact that effluents from WWTPs can contain antiandrogenic chemicals as well. Their presence has been suggested by previous studies as a potential complication in establishing the chemical causation of fish sexual disruption (Tyler and Jobling, 2008). Efforts to identify the contributing antiandrogens are now underway, using a targeted fractionation process combined with screening by recombinant yeast assay and high-quality analytical chemistry. It should also be mentioned that certain compounds may act as both estrogens and antiandrogens (e.g. Suzuki et al, 2005). There are two different approaches of sampling water, either active or passive. We chose to use passive integrative sampling, rather than traditional grab or composite sampling, for two reasons: i) passive sampling permits determination of time-weighted average concentrations of HpOCs in water, which is especially important when concentrations of HpOCs fluctuate over time because of changes in weather or variable diurnal patterns of consumption of products which are primary sources of HpOCs and, ii) the most potent EDCs usually occur at small concentrations (ng/L) and passive integrative samplers serve as an effective alternative to collecting and handling large volumes of water (Alvarez et al., 2007). One useful passive sampler for HpOCs is the Polar Organic Chemical Integrative Sampler (POCIS). Relatively good correlations have been observed between concentrations of estrogenic equivalent (EEq) determined in bioassays for POCIS and grab water samples (Arditsoglou and Voutsa, 2008; Vermeirssen et al., 2005). POCIS has been shown to sample a wide variety of polar as well as moderate hydrophobic organic compounds with log Kow of less than 4. Two types of adsorbents are considered standard for deployment of POCIS in the field. One of the two standard configurations, POCIS-Pest, preferentially concentrates waterborne HpOCs such as polar pesticides, natural and synthetic hormones, and other wastewater-related contaminants. The other, POCIS-Pharm, incorporates a sorbent optimal for sequestering polar pharmaceuticals (Alvarez et al, 2007). Both types of POCIS exhibited linear uptake of phenolic and steroid compounds during 28-day tests conducted in laboratory during which concentrations of analytes in water were held constant. The correlation coefficients of the linear regression with respect to time-scale were greater than 0.995 for POCIS-Pest and 0.985 for POCIS-Pharm, which suggests that uptake was time-integrative and the rate of uptake was not time-dependent during the exposure period. Moreover, rates of sampling (Rs) were not affected by changes in concentrations of tested compounds (Arditsoglou and Voutsa, 2008; Matthiessen and Johnson, 2007). In the present study, water quality in terms of HpOCs and EDCs was studied in several headwaters in the Czech Republic. A combination of instrumental analyses of individual chemicals and in vitro assays with extracts from POCIS-Pest and POCIS-Pharm was conducted to: i) determine background levels of anti/estrogenic, anti/androgenic and dioxin-like activities in headwater streams upstream of known sources of anthropogenic pollution, and ii) evaluate the impacts of small towns and their WWTP discharges on concentrations of mixtures of EDCs in rivers. 2. Methods 2.1. Collection of samples One POCIS-Pest and one POCIS-Pharm (Exposmeter AB, Sweden) sampler were deployed at each location. Study locations were upstream and downstream of seven municipal WWTPs, which were situated on small rivers and streams in relatively unpolluted areas of the Czech Republic (Fig. 1). Upstream (US) POCIS were placed from 2 to 5 km upstream of WWTPs in highland forest areas with minimal anthropogenic impact, while downstream (DS) sites were within 150 to 250 m of WWTP effluents. The towns studied, Králíky, Jilemnice, Cvikov, Tachov, Volary, Vimperk and Prachatice, are the upstream- Fig. 1. Location of the sampling sites on small rivers in the Czech Republic: 1 — River Tichá Orlice near town Králíky: 2 — Stream Roudnický potok (upstream) and Jizerka river (downstream) near town Jilemnice; 3 — Stream Boberský potok near town Cvikov; 4 — River Mže near town Tachov; 5 — River Volyňka near town Vimperk; 6 — Stream Volarský potok near town Volary; 7 — Stream Živný potok near town Prachatice. B. Jarosova etal. / Environment International 45 (2012) 22-31 24 most sources of anthropogenic pollution on the assessed rivers/ streams. These rivers/streams have natural or seminatural habitats flowing mostly through woodlands but there are agricultural fields or pastures in close proximity (0.2-3 km) to most of the towns. All WWTPs applied mechanical-biological treatment with activated sludge and Cvikov WWTP had an additional stabilizing pond (1.4 ha). All locations were sampled in June 2008, except for Pracha-tice, which was sampled in January 2008. Duration of deployment of samplers was 2 to 3 weeks. Duration of deployment should be within the linear uptake period for most HpOCs. Characteristics of WWTPs and river/stream conditions are summarized (Table 1). 2.2. Extraction ofPOCIS After collection of POCIS, all samples (entire POCIS) were stored at — 18°C until analysis. The exposed POCIS was disassembled; the sorbent was transferred to the glass gravity flow chromatographic column with glass wool plug and analytes were eluted by the appropriate solvent mixture. Methanol was used as the eluent for POCIS-Pharm and a mixture of dichlormethane: methanol: toluene (8:1:1) was used for POCIS-Pest. The eluate was then evaporated to a small volume, the solvent was changed to methanol and the sample volume was adjusted to 2 mL for chemical analyses. Hexane, dichloromethane, acetone, toluene (all in Suprasolv purity), water and methanol (Hyper-grade for LC/MS) were purchased from Merck (Darmstadt, Germany). The aliquots of extracts were further concentrated four-fold under a gentle stream of nitrogen to decrease the LOD for in vitro assays. The process blank samples were prepared following sample preparation procedure of both POCIS types and they were analyzed together with the other samples. 2.3. Bioassays Four individual bioassays were used to determine overall cytotoxicity, anti/estrogenicity, anti/androgenicity and dioxin-like potencies of extracts of POCIS-Pest and POCIS-Pharm samplers. The reporter gene assays employed mammalian cell lines MVLN and H4IIE-/uc and two types of recombinant Saccharomyces cerevisiae. MVLN are human breast carcinoma cells stably transfected with luciferase gene under the control of estrogen receptor, which were used for the assessment of cytotoxicity and anti/estrogenicity. Cytotoxicity of the samples was also investigated by recombinant strain of S. cerevisiae which expresses genes for enzyme luciferase under standard conditions (Leskinen et al., 2005). The potency of POCIS extracts to modulate androgen receptor-mediated responses was examined by use of recombinant S. cerevisiae that were modified to express human androgen receptor along with firefly luciferase under transcriptional control of androgen-responsive element (Michelini et al., 2005). H4IIE-/uc are rat hepato-carcinoma cells stably transfected with the luciferase gene under control of Aryl hydrocarbon receptor (AhR) and they were used for the assessment of dioxin-like activity (Sanderson et al., 1996). At least two independent experiments were conducted in each bioassay for each exposure variant. All dilutions of POCIS extracts or controls were tested at least in triplicate. Cytotoxicity of the samples can bias the results of the bioassays, therefore viability of cells was assessed several ways: Viability of MVLN cells was determined by use of the Neutral Red (NR) test where the NR dye is incorporated in the lysosomes of living cells and the uptake of NR is proportional to the number of viable cells. For cytotoxicity testing by NR-test, MVLN cells were seeded at a density of 25000 cells/well in 96-well microplate ViewPlates™ (Packard, Meriden, CT, USA) and incubated for 24 h at 37 °C under atmosphere enriched with 5% C02. During this period cells were grown in DMEM-F12 without phenol red (Sigma Aldrich, USA) containing 10% foetal calf serum previously treated with dextran-coated charcoal to reduce concentrations of natural steroids in the serum. After 24 h, cells were exposed to dilutions of extracts from POCIS and solvent control (methanol, 0.5% v/v). Cytotoxicity was determined after 24 h of exposure, when NR (Sigma-Aldrich, Czech Republic) was added to the exposure medium in microplates to make a final concentration of 0.5 mg/mL. Cells were then incubated for 1 h at 37 °C. Afterwards, the cells were washed twice with phosphate buffered saline and lysed in the presence of acetic acid-ethanol solution (25:25:0.5; ethanol:water:acetic acid) for 15 min on a shaker. Finally, NR uptake was determined spectrophotometrically (Power Wave, BioTek, USA) at 570 nm. Absorbance was related to the response of the solvent control and the percentage of cytotoxicity of each sample dilution (viability of the cells exposed to the sample dilution relative to viability of cells exposed to solvent control (considered as 100%)) was determined. For the other way of assessing the viability, the recombinant strain of S. cerevisiae which expresses genes for enzyme luciferase under standard conditions (Leskinen et al., 2005) was used. In the presence of cytotoxic substances in the medium, luminescent light, produced normally by interaction between luciferase and added substrate luciferin, is less. When reaching a linear phase of growth, yeast were seeded into 96-well culture ViewPlates™ (Packard, Meriden, CT, USA) and exposed to vehicle, dilutions of POCIS extracts or to medium alone. Yeast cells were incubated for 2.5 h at 30 °C and then the signal was detected after addition of D-luciferin substrate. Detected luminescence was used to express the percentage of cytotoxicity caused by each sample dilution, as determined by the viability of the cells exposed to sample dilution relative to viability of cells exposed to solvent control, which was assigned a value of 100%. Exposure for the determination of the anti/estrogenic potency of extracts in MVLN cells was conducted the same way as for the NR cytotoxicity evaluation described above with the following difference: cells were exposed to dilutions of POCIS extracts, calibration of the reference estrogen E2 (dilution series 10~12-0.5x 10~9 M E2, Sigma-Aldrich, Czech Republic) and solvent control (methanol, 0.5% v/v). After 24 h of exposure, the intensity of luminescence was measured Table 1 Description of sampling sites, river parameters and sampling dates and duration. Site no. Name of town Inhabitants no. Name of recipient Effluent %a River 0355 River flow Sampling duration Date of sampling13 river(stream) [m3/s] velocity [m/s] [day] 1 Kraliky 4800 Ticha Orlice 20% 0.07 0.23 16 26 May-11 June 2 Jilemnice 6000 Roudnicky potok 5% 0.02 0.08 (US) 16 26 May-11 June [USyjizerka (DS)C 0.02 (DS) 3 Cvikov 1900 Bobersky potok 10% 0.08 0.13 21 21 May-11 June 4 Tachov 13000 Mze 15% 0.40 0.17 22 21 May-12 June 5 Vimperk 7650 Volynka 4% 0.11 0.06 21 22 May-12 June 6 Volary 4000 Volarsky potok 5% 0.07 0.12 21 22 May-12 June 7 Prachatice 13000 Zivny potok 30% 0.15 0.17 23/16" 7/14"-30 January a Average contribution of WWTP effluent to the recipient. b All samples were taken in 2008. c US = upstream site, DS = downstream site. " US POCIS-Pest and both DS POCISes have been exposed for 23 days while US POCIS-Pharm for 16 days. B.Jarosova et al. I Environment International 45 (2012) 22-31 using Promega Steady Glo Kit (Promega, Mannheim, Germany). After subtraction of the response of the solvent control, luminescence in the estrogenicity assay was related to the maximal response of standard li-gand (E2max for estrogenicity) and converted to percentages of E2max. Maximal induction as well as the shape of the curve differed among samples, thus equal efficacy or parallelism of the dose-response curves could not be assumed (Villeneuve et al., 2000). To avoid any predictions beyond the measured responses with all samples and to estimate the estrogenic equivalents (EEq) in the samples (expressed in ng E2/ POC1S) the EEq2o estimate based on the 20% E2max response was reported, since most of the active samples did not reach the 50% E2max. EEq20 values were based on relating the amount of E2 causing 20% of the E2max response (EC2o) to the amount of sample causing the same response determined from regression analysis (equivalent of amount of E2 per amount of sample). The EC values were calculated by nonlinear logarithmic regression of dose-response curve of calibration standard and samples in Graph Pad Prism (GraphPad Software, San Diego, USA). The anti/estrogenicity was assessed by simultaneous exposure of the sample extract and 17p-estradiol (33 pM E2). Duration of sampling varied from 16 to 23 days at different locations. Based on the evidence from previous research that uptake of phenolic as well as steroidal estrogens is linear in terms of time and concentration up to at least 28 days (Alvarez et al., 2007; Arditsoglou and Voutsa, 2008), we present our results normalized to 20 days of deployment along with the primary data in Table 3. The normalization was performed to simplify the comparability of our results among different locations and also with other studies in discussion. The data are presented both these ways to demonstrate the possible influence of the somewhat different deployment periods of the samplers on the results and their interpretation. Concentrations of EEq in water were estimated by use of the sampling rate of E2 (0.09 L/day) previously determined by Matthiessen and Johnson (2007). It is important to stress, that these recalculated values represent approximate estimates of EEq concentrations in water and the values should not be considered as definite concentrations. This estimation will be further discussed in detail. 25 Concentrations of EEq in water were calculated (Eq. (1)). Cw = Cpoas/Rst (1) where: Cw is the estimated concentration of EEq in water (ng/L), Cmas are concentrations of EEq in extracts from POC1S (ng/POCIS; primary not normalized values), Rs is sampling rate (L/day) of E2 previously determined by Matthiessen and Johnson (2007) and t is the sampling period (days). As it was mentioned, anti/androgenity of POC1S extracts was determined by use of recombinant strain of S. cerevisiae. Plating and dosing were the same as for determination cytotoxicity of sample extracts in another strain of S. cerevisiae described above, but in this case, yeast cells were exposed not only to POC1S extracts and controls of pure medium and vehicle but also to dilutions of standard (testosterone in a range from 10~n to 10~6M, Sigma-Aldrich, Czech Republic). The H411E-/uc model was used for analysis of dioxin-like activity of the samples (Sanderson et al., 1996). Cells were seeded at a density of 15000 per well in 96-well microplate ViewPlates™ (Packard, Meriden, CT, USA) and incubated for 24 h under 5% C02 at 37 °C, in DMEM-F12 medium with phenol red (Sigma Aldrich, USA) containing 10% foetal calf serum. After 24 h, cells were exposed to the reference compound 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, with a dilution series of 10~12-0.5x 10~9 M, Ultra Scientific, USA), or dilutions of POC1S extracts and solvent control (methanol, 0.5% v/v). After 24 h of exposure, the intensity of luminescence was measured using Promega Steady Glo Kit (Promega, Mannheim, Germany). Results from the H411E-/uc in vitro assay were analyzed by the same approach as described for the determination of the EEq above. Presented TEqt,i0 are expressed in ng of TCDD per POC1S. TEqbio values were based on EC20 values because most samples did not reach greater EC responses. For each bioassay the limit of detection was determined as the lowest observable effect concentration of standard chemical divided by the greatest non-cytotoxic extract concentration expressed as POC1S equivalent. Table 2 List of pesticides and pharmaceuticals analyzed in extracts from both POCIS-Pest and POCIS-Pharm and list of perfluorinated organic compounds analyzed in extracts from POCIS-Pest. Pharmaceuticals Pesticides Perfluorinated organics Carbamazepine 2,4,5-T MCPA Perfluoro-1 -hexanesulfonate Cephalexin 2,4-D MCPPJVIECOPROP 2H-perfluoro-2-octenoic acid Ciprofloxacin Acetochlor Metalaxyl Perfluoro-1 -octanesulfonamide Diaveridine Alachlor Metamitron N-methylperfluoro-l-octanesulfonamide Diclofenac Atrazine Methabenzthiazuron Perfluorooctanoic acid Enrofloxacin Atrazine desethyl Methamidophos Perfluorooctane sulfonic acid Erythromycin Azoxystrobin Methidathion Perfluorononanoic acid Metronidazole Bentazone Metobromuron Norfloxacin Bromacil Metolachlor Ofloxacin Carbofuran Metoxuron Sulfachloropyridazine Cyanazine Metribuzin Sulfamethazine Desmetryn Monolinuron Sulfamethoxazole Diazinon Nicosulfuron Sulfamethoxypyridazine Dichlobenil Phorate Sulfapyridine Dichlorprop Phosalone Trimethoprim Dimethoate Phosphamidon Diuron Prometryn Fenarimol Propiconazole Fenhexamid Propyzamide Fipronil Pyridate Fluazifop-p-butyl Rimsulfuron Hexazinone Simazine Chlorbromuron Tebuconazole Chlorotoluron Terbuthylazine Imazethapyr Terbutryn Isoproturon Thifensulfuron-methyl Kresoxim-methyl Thiophanate-methyl Linuron Tri-allate 26 B. Jarosova etal. / Environment International 45 (2012) 22-31 2.4. LC/MS/MS analyses Chemicals such as natrium sulfate, silicagel, methanol etc. were purchased from Merck (Darmstadt, Germany). 13C labeled and native per-fluorinated compounds were purchased from Wellington Laboratories. 13C labeled Simazine, Sulfamethoxazol, 2.4D and Ciprofloxacin were purchased from Cambridge Isotope Laboratories. Native compounds were purchased from Dr. Ehrenstorfer, AccuStandards and Absolute Standards. All of the standards were purchased from Labicom ltd. (Olomouc, Czech Republic). A list of analyzed compounds is given in Table 2. A cocktail of internal standards was spiked into each POC1S extract (100 uL of the standard mixture in water was added to 100 uL of POC1S extract). Chemicals were identified and quantified by use of LC/MS/MS. Analyses were performed using three different LC/MS/ MS methods. Chemicals in POC1S extracts were quantified by use of internal standards. A subsample (20 uL for pesticide and 10 uL for pharmaceuticals) was injected onto an analytical column (Phenomenex C18 Aqua, 2mmx50mm, 5 pm particles). The HTS PAL (CTC) autosampler, Rheos2000 (Flux) quaternary pump and TSQ Quantum AccessTM (ThermoScientific, USA) triple quadrupole tandem mass spectrometer were used for analyses of polar pesticides, pharmaceuticals and per-fluorinated compounds. Two MS/MS transitions were monitored (where possible) for native analytes to confirm identity. An agreement of results obtained from both transitions better than 30% was accepted as a confirmed result. Isotope dilution and internal standard methods were used for the quantification of target compounds. Quantification limits (LOQs) of analytes were calculated the same way as concentration but peak area corresponding to instrument LOQ was used instead of peak area found in sample. Thus, LOQs are adjusted to internal standards. Most detected compounds have been shown to be in the linear uptake phase for at least 23 days (the maximal deployment period in our study) (Alvarez et al., 2007). Thus, we present concentrations of those compounds normalized to 20 days of deployment to enable more precise interpretation of our results across different locations and also better comparability with other studies in discussion. 2.5. Statistical analysis Due to violations of the assumptions of parametric statistical testing, differences between results of the two applied cytotoxicity detection systems as well as between potencies of POClS-Pest and POCIS-Pharm extracts to induce nonspecific cytotoxicity and act through specific modes of action were evaluated by nonparametric Wilcoxon Matched Pairs test. The same test was applied to assess differences between concentrations of pollutants detected in POClS-Pest and Pharm extracts. The nonparametric Spearman rank correlation was used to assess the similarity of the potential of POClS-Pest and Pharm extracts to act through specific modes of action. All statistical analyses were performed with Statistica for Windows® 9.0 (StatSoft, Tulsa, OK, USA), the tests were considered significant at p<0.05. 3. Results There was no response above detection limits observed for blanks in any of the bioassays. The limits of detection in blanks were 0.06 ng EEq/POCIS for estrogenity, 1.29 ng AEq/POCIS for androgenity and 0.03 ng TEqbio/POCIS for dioxin-like activity. 3.1. Cytotoxicity Most tested concentrations of POCIS extract equivalents (0.00125%-0.25% POCIS/mL) were not cytotoxic to yeast or to MVLN cells. At the greatest tested POCIS extract equivalent concentration 0.5% POCIS extract/mL samples from some locations caused cytotoxicity of as much as 50% (Fig. 2). For both types of POCIS the cytotoxic effects were comparable or greater at DS locations than at US locations with a single exception where the POCIS-Pharm extract at location 5 exhibited greater cytotoxicity at the US location (Fig. 2B). However, the greater cytotoxicity observed DS of WWTPs compared to US was statistically significant only for extracts of POCIS-Pest measured by yeast test In all other cases, including all extracts of POCIS-Pharm in both bioassays and POCIS-Pest in MVLN cells, the magnitude of differences in cytotoxicity was not statistically significant between US and DS. Although the yeast test was significantly more sensitive to cytotoxicity of POCIS-Pharm extracts (p = 0.009) than the MVLN test, the results of the two tests were comparable among POCIS extracts, with no significant difference between the results of the two tests with extracts of POCIS-Pest (p = 0.79). The yeast test was also significantly more sensitive to POCIS-Pharm extracts than POCIS-Pest extracts (p = 0.01), whereas there was no statistically significant difference between cytotoxicity of extracts of the two types of samplers in the MVLN test. 3.2. Anti/estrogenicity Estrogenicity was detected in extracts of both types of POCIS and differences were observed between US and DS locations. No extract showed significant antiestrogenic activity (data not shown). Although samples from DS locations were more estrogenic than those from US locations at all sites, some EEq was detected also in most US samples (Table 3). Because uptake of the more potent and also some less potent estrogens has previously been demonstrated to be time integrative for more than 25 days (e.g. Arditsoglou and Voutsa, 2008), here estrogenic potentials detected in extracts of POCIS are reported also as normalized to 20 days of POCIS deployment. However, differences between data obtained before and after the normalization to 20 days of POCIS deployment were negligible (Table 3). Concentrations of EEq greater than the LOD (0.1 to 0.6 ng/POCIS) were observed in four out of seven US locations in both types of POCIS. The variation among LOD is caused by slightly different cytotoxicity of extracts. Detected concentrations of EEq in US samples ranged from 0.3 to 0.5 ng/POCIS20 days in POCIS-Pest as well as in POCIS-Pharm extracts. Since there were no known anthropogenic impacts near US sites, the detected EEq concentrations can be considered as background. Estrogenic equivalents in extracts from DS samples were greater than the LOD at all sites with the single exception of the POCIS-Pest extract at site 2. Concentrations ranged from 0.7 to 4.0 ng/POCIS20 days for POCIS-Pest and from 0.5 to 4.2 ng/POCIS20 days for POCIS-Pharm extracts. The greatest concentrations of EEq were observed at DS locations at sites 3 and 7 (Table 3). At site 3 DS samples contained more than 10-fold greater concentration of EEq than the US sample in the case of POCIS-Pest and more than 14-fold greater concentration of EEq than the US POCIS-Pharm. At site 7 DS samples contained more than 7-fold greater concentrations of EEQ. than the US sample from POCIS-Pest and more than 5-fold greater concentration than the US sample from POCIS-Pharm, respectively. Estrogenic potential of water was estimated (Eq. (1)). For US localities sampled by both types of POCIS the calculated water EEq concentrations detected above LOD varied from 0.1 to 0.3 ng/L. Estimated estrogenic potential in water in DS locations sampled by POCIS-Pest ranged from less than 0.4 to 2.2 ng EEq/L and for those sampled by POCIS-Pharm from 0.3 to 2.3 ng EEq/L (Table 3). There were statistically significant correlations between estrogenic potentials of the pesticide and pharmaceutical POCIS extracts (Spearman rank 0.79, N = 7, LOD values were replaced by value of 1/2 LOD), despite the discrepancy at the DS location at site 6. At DS location at site 6, repeated evaluation of estrogenic potential confirmed the difference of estrogenicity in extract of POCIS-Pharm compared to POCIS-Pest The likeness of estrogenicity in extracts of POCIS-Pest and Pharm was also confirmed by nonparametric Wilcoxon Matched Pairs test, which indicated no significant difference between POCIS-Pest and Pharm (p = 0.81). 3.3. Anti/androgenicity There was no significant androgenic activity in any extract in the test with recombinant yeast assay (data not shown). Detection limit was 1.29 ng AEq/POCIS. None of the extracts has shown antiandrogenic activity (data not shown). 3.4. Dioxin-like activity Dioxin-like activity was detected in most extracts. At US locations sampled by POCIS-Pest, concentrations exceeded the detection limit of 0.03 ng TEqbio/POCIS in only two cases whereas extracts from the POCIS-Pharm sampler deployed at the same locations had detectable concentrations at six out of seven sites (Fig. 3). Concentrations of TEqbio at US locations ranged from less than the LOD to 0.08 and to 0.22 ng TEqbio/POCIS for extracts of POCIS-Pest and POCIS-Pharm, respectively. DS sites mostly showed greater concentrations of TEqbio in extracts from POCIS-Pharm than from POCIS-Pest. Extracts from DS POCIS-Pest contained concentrations of TEqbio that ranged from less than LOD of 0.08 to 0.26 ng TEqbio/POCIS and from 0.08 to 0.39 ng TEqbio/POCIS in extracts of POCIS-Pharm. When considering all samples together, significantly greater concentrations of TEqMo were observed in extracts of POCIS-Pharm than extracts of POCIS-Pest (Wilcoxon Matched Pairs test; P=0.0029). Nevertheless, similar patterns of greater concentrations of TEqbio at DS locations with similar orders of magnitudes were observed in extracts of both types of POCIS. At most sites, concentrations of TEqbio were greater DS of VWVTPs (Fig. 3). Concentrations TEqbio in extracts of DS POCIS-Pest at sites 4 and 7 were greater than those in extracts of POCIS-Pest from US, by 1.4- and 4.9-fold, respectively. Concentrations of TEqbio in extracts of POCIS-Pharm at sites 1,2 and 5 were approximately equivalent B.Jarosova et at. I Environment International 45 (2012) 22-31 27 B x o 4— o % 100 75 50 25 0 100 75 50 25 0 m POCIS Pest ■ POCIS Pharm n US DS US I DS US I D3 US I DS US I DS US DS US DS 2 3 4 5 6 7 Location number ta POCIS Pest ■ POCIS Pharm US DS [SM B US DS US DS US DS US DS US DS 2 3 4 s 6 7 US , DS 1 Location number Fig. 2. Cytotoxicity of extracts (concentration of 0.5% POCIS/mL) from upstream (US) and downstream (DS) measured by the yeast screen (A) and by Neutral Red test with MVLN cells (B). Error bars show standard deviations. For samples without any cytotoxic effect, no values are presented. for US and DS locations, whereas they were about 3 -fold greater at the DS location of sites 3 and 4 and at least about 5-fold greater at the DS location at sites 6 and 7. 3.5. Chemical analyses Although most of the selected chemicals that were monitored were not detected in extracts at concentrations greater than the LOQ. (0.1 to 14 ng/POCIS), concentrations of several pharmaceuticals were greater at DS relative to US locations (Table 4). The greatest concentrations of pharmaceuticals were observed at the DS location of site 7. Pharmaceuticals found most frequently and also at the greatest concentrations were car-bamazepine and diclofenac. Concentrations of carbamazepine ranged from less than the detection limit (2-8 ng/POCIS) to 9 ng/POCIS2o days in extracts from US locations and from 13 to 339 ng/POCIS2o days in extracts from DS locations. The concentrations of diclofenac ranged from less than the LOQ. (2-8 ng/POCIS) to 31 ng/POCIS2o days in extracts from US locations and from 18 to 409 ng/POCIS20 days in extracts from DS locations. Concentrations in extracts of POCIS-Pest and POCIS-Pharm were comparable with a few exceptions, such as sulfapyridine at sites 3 and 4. Except pharmaceuticals presented in Table 4, a few other compounds — ofloxacin, norfloxacin, ciprofloxacin and erythromycin were detected above the detection limits (LOQ0.6-14 ng/POCIS), all detected concentrations were lower than 100 ng/POCIS20 days-Concentrations of most pesticides that were monitored were less than the LOQ (0.1-6.5 ng/POCIS). Most pesticides, which were quantifiable, were triazines, and their concentrations were generally small (<100 ng/POCIS20 days). Concentrations of all detected triazines, including atrazine, atrazine desethyl, hexazinone, simazine and terbuthylazine are summarized in Table 5. Besides triazines, acetochlor at a concentration of 1375 ng/POCIS20 days was detected in one isolated POCIS-Pest sample from US location of site 2. Beside the pharmaceuticals and pesticides, perfluorinated organic compounds (listed in Table 2) were also monitored in extracts of POCIS-Pest However, concentrations greater than the LOQ of 0.21-1.15 ng/POCIS were observed only in a few cases Table 3 Estrogenic activities in POCIS-Pest and POCIS-Pharm extracts measured by MVLN in vitro assay expressed as ng EEq/POCIS, normalized to sampling period of 20 days and recalculated (according to Eq. (1)) to approximate EEq water concentrations. Site no. US/DSa POCIS depl.b (day) EEq in POCIS extracts EEq in POCIS extracts normalized Estimated EEq in water derived (ng/POCIS) to 20 days of POCIS deployment from E2 Rsc and EEq of POCIS (ng/POCIS20days) extract (ng/L) POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm 1 US 16 0.2 ±0.01 <0.2 0.3 <0.3 0.1 <0.1 DS 1.0 ±0.1 0.7 ±0.2 1.3 0.9 0.7 0.5 2 US 16 <0.3 <0.3 <0.4 <0.4 <0.2 <0.2 DS <0.3 0.7 ±0.6 <0.4 0.8 <0.2 0.5 3 US 21 0.4 ±0.3 0.3 ±0.1 0.4 0.3 0.2 0.2 DS 4.2 ±1.5 4.3 ±0.4 4.0 4.1 2.2 2.3 4 US 22 0.5 ±0.2 0.3 ±0.1 0.5 0.3 0.3 0.1 DS 0.9 ±0.2 0.5 ±0.02 0.8 0.5 0.5 0.3 5 US 21 0.4 ±0.1 0.5 ±0.1 0.4 0.5 0.2 0.3 DS 0.9 ±0.6 1.0 ±0.04 0.9 1.0 0.5 0.5 6 US 21 <0.3 <0.3 <0.3 <0.3 <0.2 <0.2 DS 0.7 ±0.7 2.3 ±0.3 0.7 2.2 0.4 1.2 7 US 23/16" <0.6 <0.6 <0.5 <0.8 <0.3 <0.4 DS 4.5 ±1.3 4.8 ±1.0 3.9 4.2 2.2 2.3 a US = upstream site, DS = downstream site. b Duration of POCIS deployment. c Rs = sampling rate. " US POCIS-Pest and both DS POCISes have been exposed for 23 days while US POCIS-Pharm for 16 days. 28 ß. Jarošova etat I Environment International 45 (2012) 22-3! 0.5 W 0.4 O O t 0.3 S 0.2 O" i" 0.1 0.0 s POCISPest ■ POCIS Pharm Jiřin! J US | DS US | DS US | DS US | DS US | DS US | DS US | DS 3 4 5 Location number Fig. 3. Dioxin-like activity of upstream (US) and downstream (DS) POCIS-Pest and POCIS-Pharm extracts determined by H4IIE-/uc in vitro assay. White columns indicate TEqbfo concentrations less than our detection limit (0.03 ng/POCIS); error bars show standard deviations. and were less than 5 ng/POCIS with single exception of perfluorooctane sulfonic acid, which was detected at DS location 2 at concentration 36 ng/POCIS. 4. Discussion Most previous studies assessing ED contamination of rivers focused on the influence of urbanized areas and larger WWTPs (Kinnberg, 2003), but there is less information on the impact of smaller sources on headwaters where better quality of water would be expected. Our study brings important information on the background levels of ED and HpOCs compounds and the influence of smaller towns without major industrial activities on headwaters pollution. Seven small rivers or streams were sampled by use of POCIS-Pest and POCIS-Pharm passive samplers US and DS of the most upstream sources of anthropogenic pollution, which were small towns with WWTP discharges. Sampling rates for most compounds, which were investigated by use of POCIS in turbulent conditions, have been reported to range from 0.12 to 0.26 L/day (95%centile of published Rs; Alvarez et al., 2007; Arditsoglou and Voutsa, 2008; Harman et al., 2008; Macleod et al., 2007; Mazzella et al., 2007). This means that in 16 days, which is the minimal time of deployment of POCIS in the study, the results of which are reported here, the amount of the chemicals present in POCIS would be equivalent to 1.92-4.16 L of river water (0.12-0.26 L/ day x 16 days). Thus, the least concentration causing cytotoxic effect — 0.5% POCIS/mL, would represent 9.6- to 20.8-fold concentrated river water. Therefore our results suggest little overall cytotoxicity of river water and weak impact of WWTPs onto this unspecific toxicity. The results of the two systems used to detect cytotoxicity, yeast and mammalian cells, were similar with the exception of greater cytotoxicity of extracts of POCIS-Pharm in the yeast cells. This observation indicates greater sensitivity of the yeast model toward some chemicals that are more concentrated by POCIS-Pharm. Chemical analyses of POCIS-Pest and Pharm extracts did not reveal any significant differences in concentrations of monitored pollutants. However, it has been suggested that some pharmaceuticals have multiple functional groups, which have a tendency to strongly bind to the carbonaceous component of the triphasic adsorbent mixture contained in POCIS-Pest, which results in poor solvent extraction recoveries of some members of this class of compounds during sample processing (Alvarez et al., 2007). Our results demonstrating weak cytotoxicity correspond to another study of Alvarez et al. (2008), who used Microtox® assay to evaluate toxicity of POCIS from surface waters burdened by extensive agriculture. In that study, no extract from passive samplers (POCIS, SPMD) exposed for 29 to 65 days displayed acute toxicity. Although the study, the results of which are reported here, was conducted in relatively unpolluted areas, some estrogenic activity was detected even at US locations (Table 3). Authors of some other studies had referred to detect concentrations of EEq in reference rivers. Nadzialek et al. (2010), who used active sampling and MCF-7 assay, found EEq concentrations at both tested reference sites in Belgium to be 0.01 and 0.03 ng/L. These concentrations are comparable with those estimated in our study (<0.1-0.3 ng EEq/L) especially if we consider our recalculated results as the worst case scenario. In contrast, Sellin et al. (2009), who used POCIS-Pharm and chemical analyses of their Table 4 Results of the LC/MS/MS analyses — pharmaceuticals with greatest detected concentrations in extracts from POCIS-Pest and POCIS-Pharm (ng/POCIS20 days)- Results are normalized to sampling period of 20 days. Site us/ Sulfapyridine Sulfamethoxazole Trimethoprim Carbamazepine Diclofenac no. DS a POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm 1 US - - - - - - - - - - DS - - 74 16 13 9 44 28 60 49 2 US - - - - - - - - - - DS - 14 9 - - - 15 15 18 30 3 US - - 11 - - - 6 - 31 24 DS 90 25 27 - 10 8 95 36 133 57 4 US 9 3 - - - - 9 3 - - DS 100 13 59 8 28 10 61 13 100 23 5 US - - - - - - - - - - DS 12 16 - - 8 14 24 40 31 70 6 US - - - - - - - - - - 7 DS US DS 42 26 30 15 35 32 190 238 181 190 1 50 36 200 122 209 209 339 304 391 409 '-" less than LOQ (0.6-14 ng/POCIS). a US = upstream site, DS = downstream site. B. Jarošova et al. / Environment International 45 (2012) 22-31 29 Table 5 Results of the LC/MS/MS analyses - concentrations of triazines (ng/POCIS20 days), which were the most frequently detected pesticides at tested sites. Results are normalized to sampling period of 20 days. Site US/ Atrazine Atrazine desethyl Hexazinone Simazine Terbuthylazine no. DSa POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm POCIS Pest POCIS Pharm 1 US _ _ 5 7 _ 15 21 DS . . 8 6 - - - 2 3 2 US 8 12 18 19 1 - . 5 1375 1875 DS 4 7 5 5 4 3 1 1 475 713 3 US 7 7 8 3 32 19 5 4 2 1 DS 24 11 17 5 49 20 8 4 3 1 4 US 2 3 8 5 6 5 - - 2 2 DS 5 2 11 3 8 3 1 - 4 3 5 US 8 7 13 7 18 12 - - 2 2 DS 5 11 7 9 12 16 - 1 2 3 6 US - - - - 1 - - - 1 1 DS 21 31 25 22 20 18 - 1 6 6 7 US 2 2 16 13 9 9 1 2 - - DS 14 11 25 18 10 9 2 1 2 1 "-" less than LOQ (0.1-6.5 ng/POCIS). a US = upstream site, DS = downstream site. extracts to monitor estrogens in rivers of Nebraska, reported calculated EEq concentrations above detection limit (1 ng/POCIS7 days) in 2 out of 3 reference sites and the concentrations (1.9 and 1.5 ng/POCIS7 days) were at least one order of magnitude greater than those found in our study. Matthiessen and Johnson (2007) evaluated, among others, estrogenic potential of 6 British headwaters with only few sources of estrogenic contamination (isolated houses with septic tanks). They used POCIS, which was previously calibrated in a laboratory study and yeast estrogen screen assay to evaluate estrogenic potential of the POCIS extracts. Their EEq concentrations ranged from less than the LOD (0.08 ng/L) to 1.4 ng/L with a median of 0.3 ng/L (except of 1 site with extremely great EEq value), which are slightly greater but comparable results to ours. Greater estrogenic potential DS of WWTPs compared to US was detected at all sampled sites (Table 3). Comparable results were obtained by Vermeirssen et al. (2005), who monitored estrogens in POCIS Pest and Pharm extracts deployed US and DS of 5 municipal WWTPs in Switzerland. Four out of the five rivers were, according to earlier DS samples analyses, chosen as moderate to greatly estrogenic whereas one river as less estrogenic. The concentrations of EEq at the least burdened site were very similar to those obtained in our study (0.4 ng EEq/POCIS22 days in extracts of both types of POCIS placed US and 1.9-2.0 ng EEq/POCIS22 days in extract of POCIS-Pest and 1.7-1.9 ng EEq/POCIS22 days of POCIS-Pharm situated DS of the WWTP). In contrast, the river with the greatest estrogenic pollution contained more than 20 ng EEq/POCIS22 days in both POCIS extracts of US samples and comparable EEq concentrations in DS ones. Similar to our results most DS samples displayed increase of estrogenic activity compared to US ones. Greater concentrations of estrogens in all POCIS samplers deployed DS of municipal WWTPs of smaller towns compared to US sites were also found in Nebraska (Sellin et al., 2009). Those authors determined estrogenic equivalents analytically (based on known potential of steroidal estrogens to cause the effect) and the recalculated EEq concentrations were greater (up to 22.7 ng/POCIS7 days) than those detected by bioassays in our study. However, the greatest EEq concentrations were detected DS of WWTP with trickling filters technology which had been previously proved to be less effective in estrogens removal than activated sludge systems (Svenson et al., 2003) such as those in all WWTPs in our study. Concentrations of EEq in POCIS extracts were converted to approximate concentrations of EEq in water by use of sampling rate of E2 because: i) in numerous studies steroidal estrogens have been identified to be responsible for most (often more than 90%) of estrogenic activity detected by in vitro assays in municipal waste waters effluents (e.g. Korner et al., 2001; Routledge et al., 1998) ii) compared to El, Estriol (E3) and EE2, E2 has the least Rs (Arditsoglou and Voutsa, 2008), which enabled to estimate the worst case scenario (the greatest concentration) and iii) E2 is the standard reference compound used for EEq calculations. For estimating concentrations of EEq in water, Rs for E2 previously established for the same standardized POCIS configuration as used in our study was applied in calculation (0.09 L/day; Matthiessen and Johnson, 2007). From the rates of sampling for E2 given in the literature (Arditsoglou and Voutsa, 2008; Matthiessen and Johnson, 2007), the Rs calibrated at 10 °C was used because the temperature was similar to the conditions in the studied streams and rivers and the application of the lowest Rs value resulted in the worst case scenario estimate. Furthermore, application of the E2 sampling rate calibrated at 23.5 ± 0.5 °C by Arditsoglou and Voutsa (2008) would result in a range <0.1 to 1.8 ng/L EEq, which is similar to the currently presented results (Table 3). Rate of sampling can vary under different environmental conditions (e.g. diverse water flow rates, pH or temperature) but all the stations (with exception of location 7) were sampled at the same time eliminating thus at least partially variability. Moreover, the flow rates were always greater than 0.02 m/s and it has been demonstrated that under turbulent conditions sampling rates do not dramatically change as a function of flow velocity (Li et al., 2010). Another line of evidence, which supports the approach of EEq calculation applied in the study, is direct comparison of POCIS with grab samples as reported by Vermeirssen et al. (2005). Those authors measured estrogenic activity in both extracts of POCIS and grab samples and concentrations of EEq in extracts of POCIS were approximately 3-fold greater than the average concentrations of EEq in grab samples. These findings indicated the rate of sampling for estrogenic compounds is approximately 0.14 L/day. This experimentally established Rs is consistent with the results observed in this study where it was assumed that use of Rs for E2 could serve as an approximation to estimate concentrations of EEq in water and that these recalculated results represent a realistic estimate of the worst case scenario. Even though the most estrogenic extracts came from POCIS exposed DS of Prachatice town (site 7), which has the most inhabitants and the largest proportion of WWTP effluent in relation to the recipient river (Table 1), these two parameters did not correlate with the estrogenic potentials in POCIS extracts from other sites. Other forces, for example different primary sources of estrogens or different WWTP capacity or technology, probably influenced the EEq concentrations in DS samples. Estrogenic activity detected in extracts of POCIS-Pest or POCIS-Pharm was similar, this observation is consistent with previous field as well as calibration studies (Arditsoglou and Voutsa, 2008; Vermeirssen et al., 2005). 30 B. Jarosova etal. / Environment International 45 (2012) 22-31 Although dioxin-like compounds are usually investigated in less polar matrices such as SPMD or sediments, some recent studies (Dagnino et al., 2010; Reungoat et al., 2010) affirmed this activity also in water phase. In this study, dioxin-like activity was detected in both types of POC1S (0.05-0.39 ng TEqbio/POClS), even at several US locations. Sampling rates for known AhR active compounds and kinetic of their sampling has not been reported for POC1S yet. Therefore our results cannot be recalculated to water concentrations nor to unified number of days of their deployment. Dioxin-like activity has been traditionally connected with hydrophobic compounds such as polychlorinated dibenzodioxins (PCDDs), polychlorinated dibenzofurans (PCDFs) or polychlorinated biphenyls (PCBs). Since experimentally-determined values for log KqW range from 6.1 to 8.2 for PCDD and PCDF congeners (Chrostowski and Foster, 1996) and from 4.66 up to 7.44 for PCB congeners, respectively (Zhou et al., 2005), these compounds are not expected to be sampled by POC1S. Our results suggest that less hydrophobic compounds like PAHs, which are also known to bind to AhR, or some unknown compounds might represent non-negligible part of dioxin-like activities in aquatic environment and this issue desires further research. In this study concentrations of TEqbio in extracts of POClS-Pharm were approximately 2-fold greater than those in extracts of POCIS-Pest. Up to authors' knowledge, no other comparisons of concentrations of TEqbio in extracts of POClS-Pest and POClS-Pharm have been published. However, since the same sorbent mass and membrane were used for both types of POC1S, it seems that different affinity of dioxin-like compounds to the POClS-Pest vs. POClS-Pharm sorbent might be responsible for the observed difference. Another reason could be the efficiency of extraction methods. However, the most potent and traditionally studied dioxin-like pollutants are hydrophobic substances and POClS-Pest was extracted by less polar solvent than POClS-Pharm. Even though in vitro assays revealed some specific potencies of mixtures that might cause effects to the aquatic biota, chemical analyses of a wide range of compounds (Table 2) did not show significant contamination. The greatest effects were observed in estrogenic activity screening assay. However, steroidal estrogens, which have been shown to be responsible for most of the estrogen equivalents in waste waters (Desbrow et al., 1998), were not monitored in this study. Among detected chemicals, some triazines are known to be able to disturb endocrine system of organisms (Danzo, 1997; Vonier et al., 1996). In this study, triazines were detected at concentrations from less than 0.1 to 1875 ng/POCIS2o days (Table 5) and their previously published sampling rates varied from 0.12 to 0.26L/day (Alvarez et al., 2007; Mazzella et al., 2007). Estimated concentrations of triazines in water ranged from less than 0.02 ng/L to 781 ng/L, but these compounds are known to be effective at concentrations greater than mg/L (Danzo, 1997; Vonier et al., 1996) and thus their contribution to the responses detected by the in vitro systems can be considered negligible. Concentrations of all monitored chemicals were small compared to the results of other studies (Arditsoglou and Voutsa, 2008; Soderstrom et al., 2009), which was in good agreement with our intention to sample relatively unpolluted areas. Despite the small concentrations of studied contaminants there were obviously increased concentrations of pharmaceuticals in DS samples. This was not so remarkable in case of pesticides. The reason of greater differences of pharmaceuticals concentrations in US and DS extract than pesticides might be the fact that pharmaceuticals are used only in human quarters or farms whereas pesticides are used also in areas distant from towns. When considering the environmental significance of our results, some of the detected estrogenic equivalents concentrations had been reported to cause adverse effects. Authors of most studies, who observed estrogenic adverse effects on aquatic biota, reported EEq concentrations or corresponding concentrations of estrogens higher than those detected in our study (e.g. Sellin et al., 2009; Vermeirssen et al., 2005; Young et al., 2004). However, for example, Vethaak et al. (2005) found elevated levels of yolk protein vitellogenin in male bream (Abramis brama) in river with EEq levels determined by in vitro ER-CALUX assay as low as 0.17 ng/L In that study, steroidal hormones were identified as the main contributors to the EEq (Vethaak et al., 2005). To authors' knowledge, the only estrogen, for which LOEC concentrations lower than 0.5 ng/L in vivo has been reported, was EE2 (Young et al., 2004). For example, Zha et al. (2008) demonstrated that the reproduction of the F-l minnows was completely inhibited at EE2 concentration as low as 0.2 ng/L in a multigeneration study with Chinese rare minnows (Gobiocypris rams). In our study, the upstream locations (with estimated EEqs <0.1-0.3 ng/L) were chosen as background sites without any grasslands or human settlements near the catchments and therefore we do not expect steroidal estrogens, particularly the synthetic EE2, to be responsible for the detected EEq. Contrariwise, at downstream locations with estimated EEq <0.2-2.3 ng/L, where municipal waste water effluents were considered as the main sources of estrogens, the presence of highly potent steroidal estrogens would be expected. The relative potency of any estrogens to E2 can differ for in vitro and in vivo studies (e.g. Johnson and Sumpter, 2001). The greatest difference has been reported for EE2. In the in vitro assay that we used (MVLN) the estrogenic potency of EE2 relative to E2 is 1.25 whereas in in vivo studies concerning production of yolk protein vitellogenin or alteration of ovarian somatic index in fish it has been reported to be approximately 25-30 (Gutendorf and Westendorf, 2001; Young et al., 2004). This indicates that the overall estrogenic equivalents for in vivo situation might be even greater that those derived from in vitro tests. As far as the authors know, there are no studies available on potential in vivo adverse effects in similar locations as examined in our study. Therefore it is not possible to reliably estimate the environmental significance of detected EEq yet. The levels of vitellogenin in brown trout (Salmo trutta fario L.) from US and DS Prachatice (corresponding to our location 7) were investigated in September 2007 by researchers from Faculty of Fisheries and Protection of Waters, University of South Bohemia. There were significantly increased levels of vitellogenin in male brown trout captured downstream compared to the upstream site. The number of examined fish males was 6 at each US and DS location. The median plasma concentration were bellow detection limit of 10 ng/mL in male fish from upstream site and 3035 ug/mL in those from downstream site (Zlabek, personal communication). This corresponds with the results of our study, where the estrogenic activity was bellow detection limit in POC1S exposed upstream of Prachatice, while there were the greatest EEq among all sites in our study detected in POC1S from the Prachatice downstream site (2.3 ng/L). Thus, the increased EEq values from in vitro studies might indicate potential in vivo effects. Generally, the relevance of in vitro determined estrogenic equivalents for in vivo situation is a very important issue, which requires further research and which is also in focus of our further studies. 5. Conclusion The study brought new information about concentrations of polar organic contaminants and endocrine-disruptive potential in relatively unpolluted rivers and about the influence of smaller towns on this type of contamination in affected headwaters. There was an obvious impact on all sites despite the fact that the towns are equipped with municipal WWTPs with advanced activated sludge systems of treatment. Increased exposure potential of estrogenic and dioxin-like compounds (determined by in vitro assays) downstream of the towns were demonstrated. Some of the detected estrogenic equivalents concentrations had been reported to cause adverse effects. The study also demonstrated the suitability of passive sampling combined with chemical analyses and in vitro bioassays to reveal these impacts. B.Jarosova et al. I Environment International 45 (2012) 22-31 31 Acknowledgments This study has been supported by the projects of Ministry of Education CR. (ENVISCREEN no. 2B08036 and INCHEMBIOL MSM0021622412), by the project CETOCOEN (CZ.1.05/2.1.00/01.0001) from the European Regional Development Fund, CENAKVA (CZ.l.05/2.1.00/01.0024) and the project SP/2e7/229/07 (Ministry of Environment CR). The research was also supported by a Discovery Grant from the Natural Science and Engineering Research Council of Canada (project # 326415-07) and a grant from the Western Economic Diversification Canada (project # 6578 and 6807). The authors wish to acknowledge the support of an instrumentation grant from the Canada Foundation for Infrastructure. Prof. Giesy was supported by the Canada Research Chair program, an at large Chair Professorship at the Department of Biology and Chemistry and State Key Laboratory in Marine Pollution, City University of Hong Kong, The Einstein Professor Program of the Chinese Academy of Sciences and the Visiting Professor Program of King Saud University. References Aguayo S, Munoz MJ, de la Torre A, Roset J, de la Pena E, Carballo M. Identification of organic compounds and ecotoxicological assessment of sewage treatment plants (STP) effluents. Sei Total Environ 2004;328:69-81. Alvarez DA, Huckins JN, Petty JD, Jones-Lepp T, Stuer-Lauridsen F, Getting DT, et al. Chapter 8 Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS). In: Greenwood R, Mills G, Vrana B, editors. Passive sampling techniques in environmental monitoring, vol. 48. Comprehensive analytical chemistry; 2007. p. 171-97. Alvarez DA, Cranor WL, Perldns SD, Clark RC, Smith SB. Chemical and toxicologic assessment of organic contaminants in surface water using passive samplers. J Environ Qual 2008;37:1024-33. Arditsoglou A, Voutsa D. Passive sampling of selected endocrine disrupting compounds using polar organic chemical integrative samplers. Environ Pollut 2008:156: 316-24. Bolong N, Ismail AF, Salim MR Matsuura T. A review of the effects of emerging contaminants in wastewater and options for their removal. Desalination 2009;239:229-46. Caliman FA, Gavrilescu M. Pharmaceuticals, personal care products and endocrine disrupting agents in the environment — a review. CLEAN—Soil Air Water 2009;37: 277-303. Chrostowski PC, Foster SA A methodology for assessing congener-specific partitioning and plant uptake of dioxins and dioxin-like compounds. Chemosphere 1996;32: 2285-304. Dagnino S, Gomez E, Picot B, Cavailles V, Casellas C, Balaguer P, et al. Estrogenic and AhR activities in dissolved phase and suspended solids from wastewater treatment plants. Sei Total Environ 2010;408:2608-15. Danzo BJ. Environmental xenobiotics may disrupt normal endocrine function by interfering with the binding of physiological ligands to steroid receptors and binding proteins. Environ Health Perspect 1997;105:294-301. Desbrow C, Routledge EJ, Brighty GC, Sumpter JP, Waldock M. Identification of estrogenic chemicals in STW effluent. 1. Chemical fractionation and in vitro biological screening. Environ Sei Technol 1998;32:1549-58. Gross-Soroldn MY, Roast SD, Brighty GC Assessment of feminization of male fish in English rivers by the environment agency of England and Wales. Environ Health Perspect 2006;114:147-51. Gutendorf B, WestendorfJ. Comparison of an array of in vitro assays for the assessment of the estrogenic potential of natural and synthetic estrogens, phytoestrogens and xenoestrogens. Toxicology 2001;166:79-89. Harman C, Tollefsen K-E, Boyum O, Thomas K, Grung M. Uptake rates of alkylphenols, PAHs and carbazoles in semipermeable membrane devices (SPMDs) and polar organic chemical integrative samplers (POCIS). Chemosphere 2008;72:1510-6. Jobling S, Tyler CR. Endocrine disruption in wild freshwater fish. Pure Appl Chem 2003;75:2219-34. Johnson AC, Sumpter JP. Removal of endocrine-disrupting chemicals in activated sludge treatment works. Environ Sei Technol 2001 ;35:4697-703. Kinnberg K. Evaluation of in vitro assays for determination of estrogenic activity in the environment Copenhagen, Denmark: Danish Environmental Protection Agency; 2003. Kirk LA, Tyler CR Lye CM, Sumpter JP. Changes in estrogenic and androgenic activities at different stages of treatment in wastewater treatment works. Environ Toxicol Chem 2002;21:972-9. Korner W, Bolz U, Sussmuth W, Hiller G, Schuller W, Hanf V, et al. Input/output balance of estrogenic active compounds in a major municipal sewage plant in Germany. Chemosphere 2000;40:1131-42. Korner W, Spengler P, Bolz U, Schuller W, Hanf V, Metzger JW. Substances with estrogenic activity in effluents of sewage treatment plants in southwestern Germany. 2. Biological analysis. Environ Toxicol Chem 2001;20:2142-51. Leskinen P, Michelini E, Picard D, Karp M, Virta M. Bioluminescent yeast assays for detecting estrogenic and androgenic activity in different matrices. Chemosphere 2005;61:259-66. Leusch FDL, Chapman HF, Korner W, Gooneratne SR Tremblay 1A Efficacy of an advanced sewage treatment plant in southeast Queensland, Australia, to remove estrogenic chemicals. Environ Sei Technol 2005;39:5781-6. Leusch FDL, De Jager C, Levi Y, Lim R Puijker L, Sacher F, et al. Comparison of five in vitro bioassays to measure estrogenic activity in environmental waters. Environ Sei Technol 2010;44:3853-60. Li HX, Vermeirssen ELM, Helm PA, Metcalfe CD. Controlled field evaluation of water flow rate effects on sampling polar organic compounds using polar organic chemical integrative samplers. Environ Toxicol Chem 2010;29:2461-9. Macleod SL, McClure EL, Wong CS. Laboratory calibration and field deployment of the polar organic chemical integrative sampler for pharmaceuticals and personal care products in wastewater and surface water. Environ Toxicol Chem 2007;26: 2517-29. Matthiessen P, Johnson I. Implications of research on endocrine disruption for the environmental risk assessment, regulation and monitoring of chemicals in the European Union. Environ Pollut 2007;146:9-18. Mazzella N, DubernetJF, Delmas F. Determination of kinetic and equilibrium regimes in the operation of polar organic chemical integrative samplers: application to the passive sampling of the polar herbicides in aquatic environments. J Chromatogr A 2007;1154:42-51. Michelini E, Leskinen P, Virta M, Karp M, Roda A A new recombinant cell-based biolumi-nescent assay for sensitive androgen-like compound detection. Biosens Bioelectron 2005;20:2261-7. Murk AJ, Legier J, van Lipzig MMH, Meerman JHN, Belfroid AC, Spenkelink A, et al. Detection of estrogenic potency in wastewater and surface water with three in vitro bioassays. Environ Toxicol Chem 2002;21:16-23. Nadzialek S, Vanparys C, Van der Heiden E, Michaux C, Brose F, Scippo M-L, et al. Understanding the gap between the estrogenicity of an effluent and its real impact into the wild. Sei Total Environ 2010;408:812-21. ReungoatJ, Macova M, Escher BI, Carswell S, Mueller JF, Keller J. Removal of micropol-lutants and reduction of biological activity in a full scale reclamation plant using ozonation and activated carbon filtration. Water Res 2010;44:625-37. Routledge EJ, Sheahan D, Desbrow C, Brighty GC, Waldock M, Sumpter JP. Identification of estrogenic chemicals in STW effluent. 2. In vivo responses in trout and roach. Environ Sei Technol 1998;32:1559-65. Sanderson JT, Aarts J, Brouwer A, Froese KL, Denison MS, Giesy JP. Comparison of Ah receptor-mediated luciferase and ethoxyresorufin-O-deethylase induction in H4IIE cells: implications for their use as bioanalytical tools for the detection of polyhalogenated aromatic hydrocarbons. Toxicol Appl Pharmacol 1996:137: 316-25. Sellin MK, Snow DD, Akerly DL, Kolok AS. Estrogenic compounds downstream from three small cities in eastern Nebraska: occurrence and biological effect J Am Water Resour Assoc 2009;45:14-21. Soderstrom H, Lindberg RH, Fick J. Strategies for monitoring the emerging polar organic contaminants in water with emphasis on integrative passive sampling. J Chromatogr A 2009;1216:623-30. Suzuki T, Kitamura S, Khota R, Sugihara K, Fujimoto N, Ohta S. Estrogenic and antian-drogenic activities of 17 benzophenone derivatives used as UV stabilizers and sunscreens. Toxicol Appl Pharmacol 2005;203:9-17. Svenson A, Allard AS. Occurrence and some properties of the androgenic activity in municipal sewage effluents. J Environ Sei Health A Tox Hazard Subst Environ Eng 2004;39:693-701. Svenson A, Allard AS, Ek M. Removal of estrogenicity in Swedish municipal sewage treatment plants. Water Res 2003;37:4433-43. Tyler CR Jobling S. Roach, sex, and gender-bending chemicals: the feminization of wild fish in English rivers. Bioscience 2008;58:1051-9. Vermeirssen ELM, Körner O, Schonenberger R Suter MJF, Burkhardt-Holm P. Characterization of environmental estrogens in river water using a three pronged approach: active and passive water sampling and the analysis of accumulated estrogens in the bile of caged fish. Environ Sei Technol 2005;39:8191-8. Vethaak AD, Lahr J, Schrap SM, Belfroid AC, Rijs GBJ, Gerritsen A, et al. An integrated assessment of estrogenic contamination and biological effects in the aquatic environment of The Netherlands. Chemosphere 2005;59:511-24. Villeneuve DL, Blankenship AL, Giesy JP. Derivation and application of relative potency estimates based on in vitro bioassay results. Environ Toxicol Chem 2000;19: 2835-43. Vonier PM, Crain DA, McLachlan JA, Guillette LJ, Arnold SF. Interaction of environmental chemicals with the estrogen and progesterone receptors from the oviduct of the American alligator. Environ Health Perspect 1996;104:1318-22. Young WF, Whitehouse P, Johnson I, Sorokin N. Proposed predicted-no-effect-concentrations (PNECs) for natural and synthetic steroid oestrogens in surface waters. Technical Report P2-T04/1, Environment Agency, Bristol; 2004. Zha JM, Sun LW, Zhou YQ, Spear PA, Ma M, Wang ZJ. Assessment of 17 alpha-ethinylestradiol effects and underlying mechanisms in a continuous, multigeneration exposure of the Chinese rare minnow (Cobiocypris rams). Toxicol Appl Pharmacol 2008;226:298-308. Zhou W, Zhai Z, Wang Z, Wang L. Estimation of n-octanol/water partition coefficients (Kow) of all PCB congeners by density functional theory. J Mol Struct Thoechem 2005;755:137-45. Príloha 22 Lohmann R., Booij K., Smedes F., and Vrana B., Use of passive sampling devices for monitoring and compliance checking of POP concentrations in water, Environ. Sci. Pollut. Res., 2012, 19, 1885-1895. Environ Sei Pollut Res (2012) 19:1885-1895 DOI 10.1007/sll356-012-0748-9 POPs WORKSHOP, TEN YEARS AFTER THE SIGNATURE OF THE STOCKHOLM CONVENTION Use of passive sampling devices for monitoring and compliance checking of POP concentrations in water Rainer Lohmann • Kees Booij • Foppe Smedes • Branislav Vrana Received: 20 October 2011/ Accepted: 6 January 2012 © Springer-Verlag 2012 Abstract Background The state of the art of passive water sampling of (nonpolar) organic contaminants is presented. Its suitability for regulatory monitoring is discussed, with an emphasis on the information yielded by passive sampling devices (PSDs), their relevance and associated uncertainties. Almost all persistent organic pollutants (POPs) targeted by the Stockholm Convention are nonpolar or weakly polar, hydrophobic substances, making them ideal targets for sampling in water using PSDs. Widely used nonpolar PSDs include semipermeable membrane devices, low-density polyethylene and silicone rubber. Results and discussion The inter-laboratory variation of equilibrium partition constants between PSD and water is mostly 0.2-0.5 log units, depending on the exact matrix Responsible editor: Philippe Garrigues R. Lohmann (El) Graduate School of Oceanography, University of Rhode Island, 215 South Ferry Rd, Narragansett, 02882 Rl, USA e-mail: lohmann@gso.uri.edu K. Booij NIOZ Royal Netherlands Institute for Sea Research, P.O. Box 59, 1790 AB Texel, The Netherlands F. Smedes Deltares, Geo-environmental Research Laboratory, P.O. box 85467, 3508 AL Utrecht, The Netherlands F. Smedes • B. Vrana Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment RECETOX, Kamenice 126/3, 625 00 Brno, Czech Republic used. The sampling rate of PSDs is best determined by using performance reference compounds during field deployment. The major advantage of PSDs over alternative matrices applicable in trend monitoring (e.g. sediments or biota) is that the various sources of variance including analytical variance and natural environmental variance can be much better controlled, which in mm results in a reduction of the number of analysed samples required to obtain results with comparable statistical power. Conclusion Compliance checking with regulatory limits and analysis of temporal and spatial contaminant trends are two possible fields of application. In contrast to the established use of nonpolar PSDs, polar samplers are insufficiently understood, but research is in progress to develop PSDs for the quantitative assessment of polar waterborne contaminants. In summary, PSD-based monitoring is a mature technique for the measurement of aqueous concentrations of apolar POPs, with a well-defined accuracy and precision. Keywords Persistent organic pollutants • Passive sampler ■ Water • Monitoring • Compliance • Quality control ■ Sampler-water partition coefficient • Sampling rate 1 Introduction Measuring aqueous phase concentrations of persistent organic pollutants (POPs) is an important activity for assessing the effectiveness of international treaties on pollution prevention and reduction, such as the Stockholm Convention (SC) on POPs, which was adopted in 2001 and entered into force on 17 May 2004 after being ratified by the fiftieth country (UNEP 2001). It is a global treaty to protect human health and the environment from the adverse effects posed Springer 1886 Environ Sci Pollut Res (2012) 19:1885-1895 by POPs. Within the convention, POPs are defined as compounds that are persistent, prone to long-range transport, bioaccumulate and elicit adverse effects. The 12 (groups of) compounds that were included in the SC in 2001 were all hydrophobic: poly chlorinated biphenyls (PCBs), poly-chlorinated dibenzo-/?-dioxins and dibenzofurans and several organochlorine pesticides (namely aldrin, chlordane, DDT, dieldrin, endrin, heptachlor, hexachlorobenzene (HCB), mirex and toxaphene). Nine additional compounds were added to the Convention in 2009 (chlordecone, a-, |3-and y-hexachlorocyclohexanes (HCHs)), pentachloroben-zene, hexabromobiphenyl, tetra-, penta-, hexa- and hepta-bromodiphenylethers, perfluorooctane sulfonic acid and its salts (PFOS) and perfluorooctane sulfonyl fluoride. In 2011, endosulfan was added to Annex A of the Stockholm Convention. With the exception of PFOS, all other POPs are nonpolar or weakly polar, hydrophobic substances, making them ideal targets for sampling in water using nonpolar passive sampling devices (PSDs). Ten years after the adoption of the SC, an expert meeting was organised on 22-24 May 2011 in Brno, Czech Republic, by the Research Centre for Toxic Compounds in the Environment, which serves as Stockholm Convention Regional Centre for capacity building and transfer of technology in Central and Eastern European countries (Klanová et al. 2011). Among the ten priority areas identified by the participants, there was a need expressed to make better use of 'advanced and cost-effective sensors capable of providing quasi-real time concentrations at different latitudes'. This was felt to be particularly important to help with the Global Monitoring Plan (GMP), a key element to the Effectiveness Evaluation of the SC. Currently, the SC monitors human milk and air, with a call for adding water to the matrices being observed regularly (Lohmann and Muir 2010). Overall, there is a dearth of POPs concentrations measured in water, making it difficult to verify multimedia modelling approaches. Another priority area identified at the Brno meeting was to make POPs measurements integral to the Global Earth Observation System of Systems (GEOSS). The GMP of GEOSS aims to interlink existing information systems for environmental and health monitoring. Regulators recognise the potentially prohibitive cost of incorrect actions based on the use of unrepresentative data in risk assessment. Passive sampling technology is proving to be a reliable and robust tool that could be used in monitoring programmes on a regional and global scale (EU 2009; Lohmann and Muir 2010). Passive sampling enables the determination of concentrations of dissolved contaminants, which is a fundamental part of an ecological risk assessment for chemical stressors (e.g. Leslie et al. 2002; Mayer and Holmstrup 2008).These devices are now being considered as part of a new strategy for monitoring a range of priority and emerging pollutants. The Advisory Committee on the Marine Environment of the International Council for Exploration of the Sea (ICES) recommended that the ICES member countries should continue working on passive sampling techniques as a monitoring tool. They further suggested to the Oslo and Paris Commission for the protection of the marine environment of the North-East Atlantic (OSPAR) that the draft guidelines for integrated monitoring should be formulated in such a way that these techniques can be included (ICES 2005). Although regulatory monitoring of organic contaminants under the European Union's Water Framework Directive (WFD) heavily relies on batch sampling followed by chemical analysis of unfiltered water sampling, the Pan-European drafting group for the chemical monitoring of surface water, sediment and biota under the WFD has listed passive sampling as a complementary method that can help to 'corroborate or contradict spot sampling data', improve water quality assessment and reduce monitoring costs, particularly when concentrations are low and time variable (EU 2009, 2010). In the following, we present the state of the art of passive water sampling and discuss its suitability for regulatory monitoring, with an emphasis on the information provided by passive sampling results, their relevancy and associated uncertainties. 2 Role of passive sampling in water monitoring 2.1 Partitioning of POPs Passive sampling methods can measure the concentration of freely dissolved contaminants (Cw), which is directly related to the contaminants' chemical activity (aw) (Mayer et al. 2003): where 5W is the contaminant solubility in water (at the same temperature and salinity). The difference in chemical activity between the two compartments quantifies the potential for spontaneous uptake. This also indicates the bioavailability or pressure (fugacity) of contaminants on organisms (Reichenberg and Mayer 2006) and consequently represents the exposure level for organisms. In an equilibrium situation, Eq. 1 can be extended to all environmental compartments like air, biota, but also sub-compartments such as suspended particulate matter (SPM) and dissolved organic matter (DOM). In non-equilibrium situations, the difference in chemical activity is the driving force for transport of compounds towards an equilibrium situation. Nonpolar POPs are largely sorbed to particulate material, such that their freely dissolved concentrations are extremely Springer Environ Sci Pollut Res (2012) 19:1885-1895 1887 low. Large volumes of water often need to be sampled with the connected risk of contamination or losses through wall adsorption. Whole water sampling, as presently prescribed in the EU's Water Framework Directive (EU 2000) yields a very poor estimate of a compound's chemical activity, because of the large contribution of contaminant fractions that are bound to SPM and DOM. Similar problems occur with batch water sampling that is followed by filtration and extraction, because the SPM and DOM partly pass through the filter. Furthermore, the freely dissolved fraction may partly adsorb to the filter. Hermans et al. (1992) reported that concentrations of (operationally defined) 'dissolved' PCBs and HCB in coastal water samples were profoundly dependent on the SPM separation method used (filtration versus centrifugation), as well as on the extent of filter clogging. Risk assessment of such hydrophobic substances is therefore often done by estimating aqueous phase concentrations from concentrations in other environmental compartments, e.g. sediment, using equilibrium partition theory (Di Toro et al. 1991) and the appropriate partition coefficients. Kd K&PM BAF where K& is the sediment-water partition coefficient, A^SPM is the SPM-water partition coefficient, BAF is the bioaccumu-lation factor and Cse(j, CSPM and Q, are the concentrations in sediment, SPM and biota. A practical problem with the application of Eq. 2 is that Kd, KSPM and BAF depend on matrix properties (e.g. amount and quality of the organic carbon and lipids, physiological state of the organism). Nonpolar passive sampling devices (PSDs) absorb hydrophobic compounds from the aqueous phase and concentrate them to a level that can be easily analysed with standard equipment, thereby avoiding the procedural errors that result from the processing of large water volumes needed in batch water sampling. Uptake is driven by the difference in chemical activity between the PSD and the surrounding environment. The use of nonporous and non-polar polymeric membranes limits the uptake of particle bound compounds, because the transient cavity sizes that are formed by the random thermal motion of the polymer chains are of the order of 1 nm, which only allows the diffusion of single contaminant molecules (Huckins et al. 1993, 2006). The rate at which PSD-water equilibrium is attained depends on the contaminants, deployment conditions and the PSDs used. For highly hydrophobic compounds and for thick nonpolar PSDs, equilibrium attainment can take months to years. In this case, the samplers yield a time-weighted average Cw over the exposure period. By contrast, equilibrium can be attained within hours to days for compounds with low hydrophobicity and for thin PSDs. More details are given below. Various PSDs are available for the sampling of POPs in water. The semi-permeable membrane device (SPMD) was the first sampler that was used on an appreciable scale (Huckins et al. 1993, 2006). The typical configuration is a 90-cm low-density polyethylene (LDPE) lay flat tubing of 2.5 cm wide (~460 cm2 surface) with a wall thickness of ~70-95 urn, filled with 1 mL synthetic triolein. Single-phase polymeric sheets and films also have been used as a PSD: LDPE (Booij et al. 1998; Adams et al. 2007), silicone (Rusina et al. 2010b; Smedes et al. 2009) and polyoxy-methylene (Jonker and Koelmans 2001; Cornelissen et al. 2008). A special version of the Chemcatcher was designed for the sampling of nonpolar organic compounds using an octanol-soaked Cig Empore disk as an adsorptive phase behind an LDPE membrane (Vrana et al. 2005). The surface area of these samplers can (within limits) be tailored to the needs for a particular sampling programme, but typically ranges between 10 and 1,000 cm2. Chemical analysis of these samplers includes the conventional extraction and cleanup procedures, followed by injection of an aliquot of the final extract for instrumental analysis. By contrast, a group of other (much smaller) samplers are analysed without extraction and cleanup, by thermal desorption of analy-tes from the whole sampler, such as solid-phase micro extraction (Arthur and Pawliszyn 1990), stir bar sorptive extraction (Baltussen et al. 1999) and membrane-enclosed sorptive coating (Vrana et al. 2001). Although the principles of operation are similar to those of SPMDs and polymer strip samplers, these samplers appear to be less widely used in environmental monitoring, and our primary focus for this review will not be on this latter group. 2.2 PSDs versus biota Biomonitoring is a widely used method for assessing environmental POP levels, as exemplified by the 'Mussel Watch Programs' (Goldberg 1975; Monirith et al. 2003; Kimbrough et al. 2009). Some well-known difficulties with biomonitoring are inter-species variability for programmes that cover a wide geographical area, the interaction between environmental conditions and contaminant uptake kinetics of the organisms, mortality and uncertainties that are associated with high initial concentrations in the case of transplanted organisms. Booij et al. (2006) evaluated literature data on co-deployed SPMDs and biota, and concluded that SPMDs yield more reliable estimates of exposure concentrations. SPMDs were deemed more reliable due to a better understanding of their contaminant uptake kinetics and less certainty in knowing in situ BAF values of the organisms. A 4-year monitoring study in Dutch coastal waters with PSDs and co-deployed mussels (eight stations, sampled twice per year) revealed a strong correlation between concentrations in mussels and PSD-derived aqueous concentrations (Smedes 2007). Similarly, a laboratory study in Springer 1888 Environ Sci Pollut Res (2012) 19:1885-1895 which polychaetes and LDPE were exposed to contaminated sediments also showed a strong relationship between both sets of results (Friedman et al. 2009). 3 Theory of contaminant uptake by PSDs 3.1 ApolarPOPs The theory of contaminant uptake by nonpolar PSDs is well established (Vrana et al. 2001; Huckins et al. 2006; Booij et al. 2007). At the initial stage of the sampler deployment, the absorbed amounts (Ns) increase linearly with time (f) if the aqueous concentration (Cw) is constant CwRst (3) During the initial stage, the product Rst can be seen as the water volume that is extracted during the deployment (amount=concentrationx volume), which is why Rs is known as the water sampling rate. At very long exposure times, the sampler equilibrates with the water, and the ana-lyte amounts are given by Ns = CwKswms (4) where Ksvi is the sampler-water partition coefficient. The product Ksv/ ms represents the water volume that is extracted by a given PSD at equilibrium. Eqs. 3 and 4 are special cases of the general uptake equation A/* = CwKswms 1 — exp _Rs± KSVJm, (5) which is valid during the linear uptake stage (t—>0), the equilibrium stage (t—>oo), as well as for the transition stage in between. Eq. 5 is always exact, whereas Eqs. 3 and 4 are always approximate, though useful for back-of-the-envelope calculations. Equation 5 allows for estimating the effectively extracted water volume (Ve) at any deployment time, which helps to compare the results from passive sampling with those of batch sampling. Analytes differ widely in the rate at which sampler-water equilibrium is attained. The quotient RJ(Ksvims) in Eq. 5 is a first-order equilibration rate constant (ke), and the characteristic time scale for equilibrium attainment (req) is given by 1 Rs (6) Thus, compounds with low Ksw values quickly attain equilibrium (e.g. naphthalenes, HCHs). PSDs yield a time-integrated Cw for exposure times that are much shorter than and closely follow the (possibly variable) environmental ' eq; Cw for t» Teq (Hawker 2010). req may therefore be used to identify the time window for time-integrative sampling. Analyte uptake includes advective and diffusive transfer from bulk water, through a water region with reduced turbulence and flow (the water boundary layer, WBL), via a biofilm (if present), into the membrane. In some PSD configurations, an additional sorption phase is present behind the membrane (e.g. triolein in the case of SPMDs, and octanol-soaked Cig bonded silica in the case of the nonpolar Chemcatcher). Each of the above transfer steps may be rate limiting, but in most cases, the uptake rates are either controlled by the membrane or by the WBL. Membrane control generally occurs for compounds with low logrvow values and low diffusion coefficients in the membrane, and for PSDs that are exposed at high flows. Aqueous concentrations can be calculated from the absorbed amounts, using Eq. 5, when Ksw and Rs are known. The accuracy of these Cw estimates obviously depends on the accuracy of these parameters, and the choice for one PSD or another should be based on the availability of high-quality calibration parameters for the compounds of interest (see Section 4.6). In situ calibration of PSDs is necessary, because ^s depends on the exposure conditions, such as temperature, flow and biofouling. This in situ calibration is done by spiking the PSDs with performance reference compounds (PRCs) before exposure (Huckins et al. 2002). Suitable PRCs do not occur in the environment (e.g. isotopically labelled compounds), and have logivsw values in the range 3-7 to ensure that dissipation data cover the full loss range between 0% and 100%. Sampling rates can be determined by fitting the retained PRC fraction (f) as a function of Ksv/ by nonlinear least-squares estimation (Booij and Smedes 2010) f N R* t (7) Because Rs not only depends on the exposure conditions, but also is weakly compound-dependent, a suitable sampling rate model should be chosen that relates Rs to compound properties, such as molecular size (see Section 4.6). Using the above method typically allows Rs to be estimated with a precision of about 10%. Uncertainties in the Ksvi values of the PRCs may result in a bias of about 0.3 log units (Booij and Smedes 2010). 3.2 Polar POPs With the addition of PFOS, HCH isomers and endosulfan to the SC, consideration has been given to include water as a recommended matrix in the GMP for POPs. Global oceans and large lake waters represent a major sink for PFOS, HCHs and endosulfan and to a lesser extent for other POPs. Nonpolar PSDs such as LDPE, silicone rubber or SPMD are Springer Environ Sci Pollut Res (2012) 19:1885-1895 1889 well applicable for monitoring of HCHs and endosulfane. Due to their low log Kow values these substances attain sampler-water partition equilibrium within several days. Their integrative sampling can be extended by PSDs with a higher mass and/or a smaller surface area (Eq. 6). This also favours lower detection limits, because higher sampler mass implies a larger equivalent water volume that is extracted at equilibrium (Eq. 4). Passive sampling of PFOS presents a specific challenge due to its low affinity to hydrophobic polymer materials used in nonpolar PSDs. Although no quantitative studies aimed at quantification of PFOS and other fluorinated surfactants in water with PSDs have been reported, several studies reported identification of these compounds in adsorbent-based polar organic chemical integrative sampler extracts (Alvarez et al. 2007; Vrana et al. 2010). Performance of various PSDs for sampling PFOS in effluent from a wastewater treatment plant, including comparison with continuous water sampling, is recently being evaluated in an interlaboratory study organised by the NORMAN association (network of reference laboratories for monitoring emerging environmental pollutants, www.norman-network.net). The PSDs that are applied for these polar emerging organic compounds are based on analyte diffusion through microporous membranes and sorption to selective adsorbent materials. Accumulation of polar organic compounds by adsorbents is more complex than absorption of hydrophobic chemicals in nonporous polymers such as LDPE or PDMS. Adsorption distribution coefficients (unlike partition coefficients in sub-cooled liquid polymers) are obtained from sorption isotherms and are concentration-dependent, and competitive adsorption of non-target analytes cannot be ruled out. The PRC approach for in situ Rs estimation is complicated by strong sorption of most compounds to the adsorbents, and desorption kinetics are generally not isokinetic with the uptake. At present, the samplers for polar POPs are not sufficiently well understood to warrant their inclusion in regulatory monitoring, but intense research is being conducted to extend applicability of PSDs for quantitative assessment of polar waterborne contaminants (including PFOS). A position paper that reviews the state of the art has been presented by the NORMAN association (Vrana et al. 2010). 4 Quality assurance and quality control of passive sampling A number of quality assurance issues that are specific for the use of PSDs should be considered before starting a monitoring project with these samplers, to ensure that the measured aqueous concentrations are fit for purpose. Most importantly, the available Ksw values and sampling rate models should be sufficiently accurate. The effect of uncertainties in Ksvi and Rs on the final Cw estimate can be evaluated by applying the method of error propagation, using Eqs. 3 (kinetic sampling) or 4 (equilibrium sampling). For example, a bias of 0.3 log units in Ksvi results in a bias of 0.3 log units in the Cw estimate of compounds that attain equilibrium during the exposure. In addition, an initial estimate of the detection limits should be evaluated, based on the analysis of solvent blanks, fabrication control PSDs and field control PSDs (Petty et al. 2000; Huckins et al. 2006; ISO 2011). 4.1 Accuracy of Ksw The accuracy of Ksw values is often difficult to assess. For many compound-sampler combinations, only values from a single study are available, often with a very small reported error that is based on replicates in one experiment. Interlaboratory variability (ILV) data yield more realistic error estimates. For LDPE, this ILV amounts to 0.18 log units (RMS value for 18 PCBs and 9 PAHs, based on two to seven laboratories, temperature range 18-24°C), except for the PAHs with five or six aromatic rings, for which the ILV was 0.78 log units (two to four studies) (Muller et al. 2001; Booij et al. 2003b; Adams et al. 2007; Cornelissen et al. 2008; Smedes et al. 2009; Perron et al. 2009; Fernandez et al. 2009b; Hale et al. 2010). For SPMDs, the ILV was estimated from the (salinity-corrected) data compilation by Huckins et al. (2006) as 0.21 log units (RMS value for one PCB and seven PAHs, two to three studies). The ILV for PDMS was estimated from the data compilation by Difi-lippo and Eganhouse (2010) as 0.45 log units (RMS value based on SPME fibres, PDMS sheets and PDMS traps at 25°C for studies that passed the quality criteria defined by these authors; 11 PAHs, 9 PCBs, 5 chlorobenzenes, 4 BTEX compounds, 6 pesticides, but excluding the pyrethroids; two to nine studies). ILV did not increase with hydrophobicity in this data set. Because not all silicone rubbers are pure PDMS, but may contain fillers and functional groups, users of silicone rubbers should carefully identify the source of the polymer that they intend to use. Differences in Ksw values of PCBs and PAHs for silicone rubbers from different sources may differ by up to 0.55 log units (Smedes et al. 2009). For the nonpolar Chemcatcher Ksv/ data are only available from a single study (Vrana et al. 2006). The accuracy of Ksvi is only critical for compounds that reach (partial) PSD-water equilibrium, and is irrelevant for compounds that remain in the linear uptake stage (c.f., Eq. 3). However, the accuracy of Cw estimates for the latter group strongly depends on the accuracy of which in turn largely depends on the quality of the Ksw values of the PRCs (c.f., Eq. 7). Springer 1890 Environ Sci Pollut Res (2012) 19:1885-1895 4.2 Effect of temperature and salinity The effects of temperature and salinity on Ksv/ are small and well established. The temperature effect on KSVi can be estimated using the Van't Hoff equation log^w,2 = l0g^ " ^Mii (i " £) (8) where A//Sw is the water-sampler transfer enthalpy, Tis the absolute temperature and R is the gas constant. Adams et al. (2007) showed for two PAHs and one PCB that A//Sw for LDPE may be estimated from for T Tm:AHsw = -AHsol 1 > where AHfus is enthalpy of fusion, AHsol is the enthalpy of solution and Tm is the melting temperature of the target analyte. These authors report A//Sw values between -12 and -29 kJ mol l, which indicates a decrease in Ksw by a factor of 1.2 to 1.5 for a 10°C temperature increase. Similar A//Sw values can be estimated from the log^sw values reported by Booij et al. (-9 to -45 kJ mol :) (Booij et al. 2003b). Muijs and Jonker (2009) report water-PDMS transfer enthalpies of PAHs to decrease with molecular size from -16 kJ mol 1 for phenanthrene to -35 kJ mol 1 for indeno [l,2,3-crf]pyrene, i.e. similar to the A//Sw for LDPE. Lohmann (2012) suggested using a value of AHSW of -25 kJ mol 1 for nonpolar compounds. The Ksw of chlo-robenzenes, PCBs, PAHs and DDE for SPMDs appears to be largely independent of temperature (Booij et al. 2003b; Huckins et al. 200!), possibly with the exception of phenanthrene, for which AHSW values of +5 and -23 kJ mol1 have been found. The effect of dissolved salts on the KSVi of nonpolar compounds is small, and can be calculated from the Setschenow equation, log^sw = logZsw,0 + KJ (10) where / is the ionic strength (in moles per liter), Ks is the Setschenow constant (in liters per mole) and Ksw0 is the sampler-water partition coefficient at an ionic strength of 0. Xie et al. (1997) showed that Ks increases with increasing LeBas molar volume (^LeBas) of the analyte from 0.1 to 0.3 L mol1 when ^LeBas increases from 60 to 180 cm3 mol1 and levels off to a constant value of 0.3 Lmol1 for larger compounds. Jonker and Muijs (2010) reported ^s=0.35±0.02 L mol 1 for PAHs with three to six aromatic rings (^LeBas >197 cm3 moF1). Adopting an ionic strength of average seawater at 20°C of 0.713 mol L 1 (Millero and Sohn 1992; Millero and Huang 2009), and a Setschenow constant of 0.35 Lmol1, the salt effect on Ksw can be seen to be smaller than 0.25 log units in most environments. Springer The uncertainties in the temperature and ionic strength corrections on Ksvi are small. With the temperature correction, an error in AHSW of 10 kJ mol 1 results in a 15% error in Ksw over a 10°C temperature range (~ 0.06 log units). Similarly, if the Setschenow constant used is off by 0.1 Lmol l, then this results in an error in Ksvi of only 18% for average seawater. This leaves the Ksvi values determined in ultrapure water as the primary source of error: 0.18 to 0.45 log units, depending on the compound and the sampler. The primary challenge with nonpolar samplers is therefore to reduce the inter-laboratory variability in Ksvi determinations, rather than to test 'novel' PSDs equipped with yet another sampling matrix. 4.3 Estimating Ksvi When experimental values are not available, Ksvi has to be estimated from empirical correlations with compound properties, and PSD users have to check if the accuracy of these correlations is sufficient, prior to starting a sampling project. Traditionally, correlations with logffow have been widely used, but this method suffers from the fact that logKow values often have an uncertainty of about 0.2 log units or more, possibly with the exception of values that have been optimised for thermodynamical consistency (Beyer et al. 2002; Aberg et al. 2008; Ma et al. 2010; Schenker et al. 2005; Shen and Wania 2005; Li et al. 2003; Xiao et al. 2004) . Alternatively, Ksw values can be estimated using polyparameter linear free energy relationships (Sprunger et al. 2007; Arp et al. 2010; Difilippo and Eganhouse 2010; Endo et al. 2011). Again, the accuracy of Ksvi is only an issue for PRCs and for target analytes that attain equilibrium with the sampler. 4.4 Accuracy of the Rs model The next step is to determine if the sampling rate model to be used in Eqs. 3 and 5 is sufficiently reliable. It is well established that sampling rates are limited by diffusion into the sampler for compounds with low ^ow values and by transport through the WBL for analytes with high hydrophobicity (Vaes et al. 1996; Leslie et al. 2002;Booij et al. 2003b; 2007; Huckins et al. 2006). The modelling of WBL-controlled uptake is relatively straightforward. Hydrodynamic theory and evidence from the chemical engineering literature states that sampling rates are proportional to the aqueous diffusion coefficient to the power 2/3 (Levich 1962; Boudreau and Guinasso 1982; Booij et al. 2007; Knudsen et al. 1999; Stephens et al. 2005) , which only leaves the proportionality constant to be derived from PRC dissipation data. Experimental values of diffusion coefficients of organic contaminants are virtually absent, but the necessary estimates may be obtained from semi-empirical correlations with molar mass (M) or molar volume. Thus, depending on the diffusion coefficient model that is used, Environ Sci Pollut Res (2012) 19:1885-1895 1891 Rs is expected to be proportional to AT0 35(Booij et al. 2003b), AT047 (Rusina et al. 2010b), l^LeBas °39(Huckins et al. 2006). For example, the sampling rate of PCB180 is expected to be smaller than that of phenanthrene by a factor of 1.3, 1.4 and 1.2, respectively. Because molecular size and hydrophobicity are strongly correlated for nonpolar compounds, a weak decrease in Rs with increasing Kow can be expected. Thus, Leslie et al. (2002) found for SPME fibres that Rs/Vs (kľ in the terminology of these authors) is virtually independent of Kow in the range 310 ms-1). In most cases, the compounds that experience membrane-controlled uptake also attain equilibrium during the exposure period. As a rule of thumb, it can be assumed that the uptake is membrane controlled for compounds with log&ľow<4.5 for LDPE and SPMDs, and logAľow<3.5 for silicone samplers, except for some rare cases where very high sampling rates are observed. WBL control of the uptake can always be estimated afterwards from where Dm is the diffusion coefficient in the membrane (Rusina et al. 2007, 2010a; Adams et al. 2007; Hale et al. 2010), Rs is obtained from PRC dissipation data, A is the surface of the sampler, Kmvi is the membrane-water partition coefficient and 5m is the membrane thickness for biphasic samplers (SPMD, Chemcatcher) or half the sampler thickness for single-phase samplers of which both sides are exposed. The role of the membrane in the uptake can be neglected if Eq. 11 is satisfied for a PRC that is 50% dissipated. 4.5 Detection limits After assuring the quality of Ksw and the Rs model, the expected detection limits should be assessed from the amounts detected in fabrication controls and field controls. Fabrication controls yield information on the contaminant amounts that are taken up from the laboratory atmosphere during construction. The field controls yield information on analyte uptake during transport and deployment/retrieval operations. Initial approximations of the detection limits can be obtained from Eq. 5 by substituting the average amounts detected in the field control samples or the limit of detection of the analytical method if the compound was not detected. Required estimates of the sampling rate can of course only be obtained after the exposure, but an initial estimate ofRs/A of 2-5 LdmT2 day-1 usually is a realistic range to work with if the true value of Rs is not yet known. 4.6 Quality control A number of quality control measures are needed to certify the quality of a running PSD-based monitoring project. These include the analysis of solvent blanks, fabrication controls, field controls and matrix spikes. Comparison of solvent blanks, fabrication controls and field controls can help to identify possible sources of contamination, after which appropriate measures can be taken. Blank subtraction may be done for analytes that were in the linear uptake stage during the exposure, as inferred from PRC dissipation data. Blank subtraction should not be done for compounds that attain (partial) equilibrium, because the amounts detected in exposed PSDs can be lower than those in the field controls, due to dissipation of pre-deployment contamination during the exposure. A conservative sample rejection criterion is to set the minimum amount in exposed samplers to ten times the amounts detected in the field controls, and to review sampler construction and transport operations if this condition is not met. A further quality control parameter is the precision of the PRC-based sampling rates. The quantitation of a highly hydrophobic PRC that is insignificantly dissipated also yields useful information on the precision of the chemical analysis on a per sample basis. Springer 1892 Environ Sci Pollut Res (2012) 19:1885-1895 4.7 Standardisation Progress has been made on the normation of passive sampling methods for their use in monitoring water quality. In March 2011, an ISO standard has been published that specifies procedures for the determination of time-weighted average concentrations and equilibrium concentrations of dissolved organic, organometallic and inorganic substances, including metals, in surface water by passive sampling, followed by analysis (ISO 2011). Collaborative studies are required to obtain information about the robustness of the whole sampling process including instrumental analysis, sampler calibration and field sampling. Several collaborative exercises were performed recently with the aim to compare performance of various samplers. Allan et al. (2009) evaluated the performance of seven PSDs for the monitoring of PAHs, PCBs, HCB andp,p'-DDE through simultaneous field exposures of 7-28 days in the River Meuse. Despite the absence of the analytical comparability test of participating laboratories and different modes of calculation, relatively consistent Cw values were obtained for the different samplers and sources of observed variability were critically discussed. In 2010, the French national water reference laboratory AQUAREF (www.aquaref.fr) organised an inter-laboratory study targeting compounds relevant in chemical monitoring under the WFD, including PAHs, heavy metals and polar pesticides. Another study aimed at monitoring of emerging pollutants was organised by NORMAN. The latter two exercises were designed to cover individual aspects in the passive sampling process, including analytical comparability and, where it was possible, comparison with conventional sampling of water. The results of these studies will be available in 2012. More proficiency testing schemes are needed for the most frequently used PSD designs to evaluate the contribution of the analytical uncertainty component to total variability of the sampling process. Inter-laboratory studies that compare the performance of various available passive sampler designs at a reference site will allow a realistic evaluation of passive sampling variability for the tested compounds and give information whether a particular passive sampling method provides a satisfactory result within an agreed performance interval. Finally, campaigns where water samples are analysed in parallel with passive samplers are required to evaluate the comparability of these two methods. 5 Application of passive samplers in regulatory monitoring We are not aware of any cases yet where PSDs have already been accepted for compliance checking. So far, the use of PSDs in monitoring programmes has been limited to occasional studies. PSDs have been used by a number of governmental agencies in the USA (e.g. US Geological Survey, US Environmental Protection Agency, US National Park Service, US Fish and Wildlife Service, Virginia Department of Environmental Quality, Washington State Department of Ecology), the United Kingdom (UK Environment Agency) and the Czech Republic (Institute of Public Health) (ITRC 2006). The Dutch monitoring authorities have used PSDs for trend monitoring since 2001 at eight coastal stations (Smedes et al. 2007), and included several freshwater stations in 2008. Beside this in many other countries (e.g. Australia, Belgium, France, Germany, Ireland, Norway, Slovakia, Sweden and Switzerland), trials and/or repeated sampling using PSDs is occurring, although mainly on project basis. Within ICES/OSPAR, a trial survey was organised in 2006 in order to investigate the possibilities of passive sampling for OSPAR monitoring. In a mutual effort 13 laboratories sampled 30 stations all over Europe demonstrating the potential of PSDs for wide-scale monitoring (Smedes et al. 2007). For many POPs, regulatory limit values such as EU environmental quality standards (Lepom et al. 2009) refer to concentrations in water that are extremely low (low nanograms per liter) and traditionally established low-volume water sampling techniques very often fail to comply with minimum performance criteria in terms of limit of quantification and measurement uncertainty. Alternative, more sensitive sampling techniques, such as high-volume sampling devices are costly and hardly applicable in monitoring campaigns on a large scale. Moreover, discontinuous water sampling with a low sampling frequency may not provide information with required confidence and precision for compliance checking where concentrations of pollutants fluctuate in time (e.g. with seasonal variation in use of pesticides or sporadic industrial discharges). The unique performance characteristics of passive samplers that include their time integrative nature combined with extremely low limits of quantification for most POPs may represent the only practicable way to monitor these substances in the water column. Since reliable values of Ksw and Rs with associated uncertainty can be derived for most of priority pollutant POPs in nonpolar PSDs such as LDPE and silicone rubbers, fulfilment of legally binding minimum method performance criteria and QA/QC provisions for compliance checking can be demonstrated. 6 Summary and recommendations PSDs can effectively be used as a tool in regulatory monitoring as the obtained freely dissolved concentration is a strong indicator for exposure to aqueous organisms. PSDs are suitable matrices for trend monitoring of hydrophobic Springer Environ Sci Pollut Res (2012) 19:1885-1895 1893 POPs because they integrate concentration fluctuations in time in a specific water body and long-term comparisons can be made with lower sampling frequency at the required sensitivity and statistical power to detect temporal or spatial trends. The major advantage of PSDs over alternative matrices used for trend monitoring, e.g. sediments or biota, is that PSDs constitute a well-defined sample medium with known uptake capacity. In contrast to results based on sediment or biota, PSD data require no corrections for organic carbon, lipid content or species to compare data on a worldwide scale. Passive samplers can safely be sent around and deployment requires no specialists, making it possible to monitor POPs across the world. Furthermore, different sources of variance including analytical and environmental variance can be much better controlled, which in turn results in reduction of the required number of analysed samples to obtain results with comparable statistical power. Compliance checking with regulatory limits and analysis of temporal and spatial contaminant trends are two possible fields of application. The objection against passive sampling has been that PSDs are qualitative (or at best semiquantitative) tools for assessing water quality. In the present article, we argue that PSDs can now be regarded as fully quantitative tools with well-defined accuracy and precision that allow concentrations of dissolved organic contaminants to be compared against legal standards. This would require an adaptation of legal standards away from total concentrations towards dissolved concentrations that better reflect the compound's chemical activity and related exposure level in the environment. Meanwhile, the scientific community should take further steps towards improving the accuracy and precision of passive sampling technology, by means of inter-laboratory comparison studies and inter-calibration studies. The focus of these studies should be on PSD handling, chemical analysis and data processing, as well as on the development of strict protocols for the accurate determination of PSD-water partition coefficients. In addition, further research is needed for improving the accuracy of PSDs for polar organic compounds. Acknowledgements RL. acknowledges funding from EPA's Great Lakes Restoration Initiative Award GLAS # 00E00597-0 supporting passive sampler research at URL References Aberg A, MacLeod M, Wiberg K (2008) Physical-chemical property data for dibenzo-p-dioxin (DD), dibenzofiiran (DF), and chlorinated DD/ Fs: a critical review and recommended values. J Phys Chem Ref Data 37(4): 1997-2008. doi:l 0.1063/1.3005673 Adams RG, Lohmann R, Fernandez LA, Macfarlane JK, Gschwend PM (2007) Polyethylene devices: passive samplers for measuring dissolved hydrophobic organic compounds in aquatic environments. Environ Sci Technol 41(4):1317-1323 Allan IJ, Booij K, Paschke A, Vrana B, Mills GA, Greenwood R (2009) Field performance of seven passive sampling devices for monitoring of hydrophobic substances. Environ Sci Technol 43 (14):5383-5390. doi:10.1021/es900608w Alvarez DA, Huckins JN, Petty JD, Jones-Lepp T, Stuer-Lauridsen F, Getting DT, Goddard JP, Gravell A (2007) Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS). In: Greenwood R, Mills G, Vrana B (eds) Passive sampling techniques in environmental monitoring. Elsevier, Amsterdam, pp 171-197 Arp HPH, Endo S, Goss KU (2010) Comment on "Assessment of PDMS-water partition coefficients: Implications for passive environmental sampling of hydrophobic compounds". Environ Sci Technol 44(22):8787-8788 Arthur CL, Pawliszyn J (1990) Solid phase microextraction with thermal desorption using fused silica optical fibers. Anal Chem 62:2145-2148 Baltussen E, Sandra P, David F, Cramers C (1999) Stir bar sorptive extraction (SBSE), a novel extraction technique for aqueous samples: theory and principles. J Microcolumn Sep ll(10):737-747 Beyer A, Wania F, Gouin T, Mackay D, Matthies M (2002) Selecting internally consistent physicochemical properties of organic compounds. Environ Toxicol Chem 21(5):941-953 Booij K, Hoedemaker JR, Bakker JF (2003a) Dissolved PCBs, PAHs, and HCB in pore waters and overlying waters of contaminated harbor sediments. Environ Sci Technol 37(18):4213-4220 Booij K, Hofmans HE, Fischer CV, Van Weerlee EM (2003b) Temperature-dependent uptake rates of nonpolar organic compounds by semipermeable membrane devices and low-density polyethylene membranes. Environ Sci Technol 37 (2):361-366 Booij K, Sleiderink HM, Smedes F (1998) Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ Toxicol Chem 17(7):1236—1245 Booij K, Smedes F (2010) An improved method for estimating in situ sampling rates of nonpolar passive samplers. Environ Sci Technol 44(17):6789-6794. doi: 10.1021/es 101321 v Booij K, Smedes F, van Weerlee EM, Honkoop PJC (2006) Environmental monitoring of hydrophobic organic contaminants: the case of mussels versus semipermeable membrane devices. Environ Sci Technol 40(12):3893-3900. doi:10.1021/es052492r Booij K, Vrana B, Huckins JN (2007) Theory, modelling and calibration of passive samplers used in water monitoring. In: Greenwood R, Mills GA, Vrana B (eds) Passive sampling techniques in environmental monitoring. Elsevier, Amsterdam, pp 141-169 Boudreau BP, Guinasso NL (1982) The influence of a diffusive sublayer on accretion, dissolution, and diagenisis at the sea floor. In: Fanning KA, Manheim FT (eds) The dynamic environment of the ocean floor. Lexington Books, Toronto, pp 115-145 Cornelissen G, Pettersen A, Broman D, Mayer P, Breedveld GD (2008) Field testing of equilibrium passive samplers to determine freely dissolved native polycyclic aromatic hydrocarbon concentrations. Environ Toxicol Chem 27(3):499-508 Di Toro DM, Zarba CS, Hansen DJ, Berry WJ, Swartz RC, Cowen CE, Pavlou SP, Allen HE, Thomas NA, Paquin PR (1991) Technical basis for establishing sediment quality criteria for nonionic organic chemicals using equilibrium partitioning. Environ Toxicol Chem 10:1541-1583 Difilippo EL, Eganhouse RP (2010) Assessment of PDMS-water partition coefficients: Implications for passive environmental sampling of hydrophobic organic compounds. Environ Sci Technol 44 (18):6917-6925. doi:10.1021/esl01103x Endo S, Hale SE, Goss KU, Arp HP (2011) Equilibrium partition coefficients of diverse polar and nonpolar organic compounds to polyoxymethylene (POM) passive sampling devices. Environ Sci Technol 45(23):10124-10132 Springer 1894 Environ Sci Pollut Res (2012) 19:1885-1895 EU (2000) Directive 2000/60/ec of the European parliament and of the council of 23 October 2000 establishing a framework for community action in the field of water policy. Off J Eur Union L327:l-72 EU (2009) Guidance on surface water chemical monitoring under the water framework directive. Common implementation strategy for the water framework directive (2000/60/ec); guidance document no. 19. vol http://ec.europa.eu/environment/water/water-framework/ facts_figures/guidance_docs_en.htm. Office for Official Publications of the European Communities, Luxembourg EU (2010) Guidance on chemical monitoring of sediment and biota under the water framework directive; common implementation strategy for the water framework directive (2000/ 60/ec). Office for Official Publications of the European Communities, Luxembourg Fernandez LA, Harvey CF, Gschwend PM (2009a) Using performance reference compounds in polyethylene passive samplers to deduce sediment porewater concentrations for numerous target chemicals. Environ Sci Technol 43(23):8888-8894. doi: 10.102l/es901877a Fernandez LA, MacFarlane JK, Tcaciuc AP, Gschwend PM (2009b) Measurement of freely dissolved PAH concentrations in sediment beds using passive sampling with low-density polyethylene strips. Environ Sci Technol 43(5):1430-1436 Friedman C, Burgess RM, Perron MM, Cantwell MG, Ho KT, Lohmann R (2009) Comparing polychaete bioaccumulation and passive sampler uptake to assess the effects of sediment resuspension on PCB bioavailability. Environ Sci Technol 43:2865-2870 Goldberg ED (1975) The mussel watch: a first step in global marine monitoring. Mar Pollut Bull 6:111-114 Hale SE, Martin TJ, Goss KU, Arp HPH, Werner D (2010) Partitioning of organochlorine pesticides from water to polyethylene passive samplers. Environ Pollut 158(7):2511-2517. doi: 10.1016/j. envpol.2010.03.010 Hawker DW (2010) Modeling the response of passive samplers to varying ambient fluid concentrations of organic contaminants. Environ Toxicol Chem 29(3):591-596. doi:10.1002/etc.69 Hermans JH, Smedes F, Hofstraat JW, Cofino WP (1992) A method for estimation of chlorinated biphenyls in surface waters: influence of sampling method on analytical results. Environ Sci Technol 26:2028-2034 Huckins JN, Manuweera GK, Petty JD, Mackay D, Lebo JA (1993) Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environ Sci Technol 27:2488-2496 Huckins JN, Petty JD, Booij K (2006) Monitors of organic chemicals in the environment - semipermeable membrane devices. Springer, New York Huckins JN, Petty JD, Lebo JA, Almeida FV, Booij K, Alvarez DA, Clark RC, Mogensen BB (2002) Development of the permeability/ performance reference compound approach for in situ calibration of semipermeable membrane devices. Environ Sci Technol 36 (1):85-91 ICES (2005) Passive sampling techniques for contaminants. Report of the advisory committee on the marine environment (ACME), section 2.3.5. http://www.ices.dk/committe/acom/comwork/report/2005/ may/passive%20samplers.pdf. ICES, Copenhagen ISO (2011) Water quality - sampling - part 23: Guidance on passive sampling in surface waters. International Organization for Standardization, Geneva, ISO 5667-23:2011(E) ITRC (2006) Technology overview of passive sampler technologies. http://www.itrcweb.org/documents/dsp_4.pdf. Interstate Technology & Regulatory Council, Washington, D.C. Jonker MTO, Koelmans AA (2001) Polyoxymethylene solid phase extraction as a partitioning method for hydrophobic organic chemicals in sediment and soot. Environ Sci Technol 35 (18):3742-3748 Springer Jonker MTO, Muijs B (2010) Using solid phase micro extraction to determine salting-out (Setschenow) constants for hydrophobic organic chemicals. Chemosphere 80(3):223-227. doi:10.1016/j. chemosphere.2010.04.041 Kimbrough KL, Johnson WE, Lauenstein GG, Christensen JD, Apeti DA (2009) An assessment of polybrominated diphenyl ethers (PBDEs) in sediments and bivalves of the U.S. Coastal zone. NOAA Technical Memorandum NOS Silver Spring, MD Klanova J, Diamond M, Jones K, Lammel G, Lohmann R, Pirrone N, Scheringer M, Balducci C, Bidleman T, Blaha K, Blaha L, Booij K, Bouwman H, Breivik K, Eckhardt S, Fiedler H, Garrigues P, Harner T, Holoubek I, Hung H, MacLeod M, Magulova K, Mosca S, Pistocchi A, Simonich S, Smedes F, Stephanou E, Sweetman A, Sebkova K, Venier M, Vighi M, Vrana B, Wania F, Weber R, Weiss P (2011) Identifying the research and infrastructure needs for the global assessment of hazardous chemicals ten years after establishing the Stockholm Convention. Environ Sci Technol 45 (18)7617-7619. doi:10.1021/es202751f Knudsen JG, Hottel HC, Sarofim AF, Wankat PC, Knaebel KS (1999) Heat and mass transfer. In: Green DW, Maloney JO (eds) Perry's chemical engineer's handbook. 7th ed. McGraw-Hill, New York, pp 5-1 to 5-79 Lepom P, Brown B, Hanke G, Loos R, Quevauviller P, Wollgast J (2009) Needs for reliable analytical methods for monitoring chemical pollutants in surface water under the European water framework directive. J Chromatogr A 1216:302-315 Leslie HA, Ter Laak TL, Busser FJM, Kraak MHS, Hermens JLM (2002) Bioconcentration of organic chemicals: is a solid-phase microextraction fiber a good surrogate for biota? Environ Sci Technol 36:5399-5404 Levich VG (1962) Physicochemical hydrodynamics. Prentice Hall, Inc., Englewood Cliffs, NJ Li NQ, Wania F, Lei YD, Daly GL (2003) A comprehensive and critical compilation, evaluation, and selection of physical-chemical property data for selected polychlorinated biphenyls. J Phys Chem Ref Data 32(4):1545-1590 Lohmann R, Muir DCG (2010) Global aquatic passive sampling (AQUA-GAPS): using passive samplers to monitor POPs in the waters of the world. Environ Sci Technol 44(3):860-864 Lohmann R (2012) A critical review of low-density polyethylene's partitioning and diffusion coefficients for trace organic contaminants and implications for its use as a passive sampler. Environ Sci Technol 46:606-618 Ma YG, Lei YD, Xiao H, Wania F, Wang WH (2010) Critical review and recommended values for the physical-chemical property data of 15 polycyclic aromatic hydrocarbons at 25°C. J Chem Eng Data 55:819-825 Mayer P, Tolls J, Hermens L, Mackay D (2003) Equilibrium sampling devices. Environ Sci Technol 37(9):184A-191A Mayer P, Holmstrup M (2008) Passive dosing of soil invertebrates with polycyclic aromatic hydrocarbons: limited chemical activity explains toxicity cutoff. Environ Sci Technol 42(19)7516-7521 Millero FJ, Huang F (2009) The density of seawater as a function of salinity (5 to 70 g kg"1) and temperature (273.15 to 363.15 K). Ocean Sci 5:91-100 Millero FJ, Sohn ML (1992) Chemical oceanography. CRC Press, Boca Raton Monirith I, Ueno D, Takahashi S, Nakata H, Sudaryanto A, Subramanian A, Karuppiah S, Ismail A, Muchtar M, Zheng J, Richardson BJ, Prudente M, Hue ND, Tana TS, Tkalin AV, Tanabe S (2003) Asia-Pacific mussel watch: monitoring contamination of persistent organochlorine compounds in coastal waters of Asian countries. Mar Pollut Bull 46:281-300 Muijs B, Jonker MTO (2009) Temperature-dependent bioaccumulation of polycyclic aromatic hydrocarbons. Environ Sci Technol 43 (12):4517^1523. doi:10.1021/es803462y Environ Sci Pollut Res (2012) 19:1885-1895 1895 Muller JF, Manomanii K, Mortimer MR, McLachlan MS (2001) Partitioning of polycyclic aromatic hydrocarbons in the polyethylene/ water system. Fresenius J Anal Chem 371(6):816-822 Perron MM, Burgess RM, Ho KT, Pelletier MC, Friedman CL, Cantwell MG, Shine JP (2009) Development and evaluation of polyethylene reverse samplers for marine phase II whole sediment toxicity identification evaluations. Environ Toxicol Chem 28:749-758 Petty JD, Orazio CE, Huckins JN, Gale RW, Lebo JA, Meadows JC, Echols KR, Cranor WL (2000) Considerations involved with the use of semipermeable membrane devices for monitoring environmental contaminants. J Chromatogr A 879:83-95 Reichenberg F, Mayer P (2006) Two complementary sides of bioavailability: accessibility and chemical activity of organic contaminants in sediments and soils. Environ Toxicol Chem 25:1239-1245 Rusina TP, Smedes F, Klanová J (2010a) Diffusion coefficients of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in polydimethylsiloxane and low-density polylethylene polymers. J Appl Polymer Sci 116(3): 1803-1810. doi:10.1002/app.31704 Rusina TP, Smedes F, Klanová J, Booij K, Holoubek I (2007) Polymer selection for passive sampling: a comparison of critical properties. Chemosphere 68(7): 1344-1351. doi: 10.1016/j. chemosphere.2007.01.025 Rusina TP, Smedes F, Kobližková M, Klanová J (2010b) Calibration of silicone rubber passive samplers: experimental and modeled relations between sampling rate and compound properties. Environ Sci Technol 44(l):362-367. doi::10.1021/es900938r Schenker U, MacLeod M, Scheringer M, Hungerbuhler K (2005) Improving data quality for environmental fate models: a least-squares adjustment procedure for harmonizing physicochemical properties of organic compounds. Environ Sci Technol 39 (21):8434-8441 Shen L, Wania F (2005) Compilation, evaluation, and selection of physical-chemical property data for organochlorine pesticides. J Chem Eng Data 50:742-768 Smedes F (2007) Monitoring of chlorinated biphenyls and polycyclic aromatic hydrocarbons by passive sampling in concert with deployed mussels. In: Greenwood R, Mills G, Vraná B (eds) Comprehensive analytical chemistry, vol 48. Elsevier, pp 407-448 Smedes F, Geertsma RW, van der Zande T, Booij K (2009) Polymer-water partition coefficients of hydrophobic compounds for passive sampling: application of cosolvent models for validation. Environ Sci Technol 43(18)7047-7054. doi:10.1021/es9009376 Smedes F, Van der Zande T, Davies IM (2007) ICES passive sampling trial survey for water and sediment (PSTS) 2006- 2007. Part 3: preliminary interpretation of field data, http:// www.ices.dk/products/cmdocs/cm-2007/j/j0407.pdf. Sprunger L, Proctor A, Acree WE, Abraham MH (2007) Characterization of the sorption of gaseous and organic solutes onto polydi-methyl siloxane solid-phase microextraction surfaces using the Abraham model. J Chromatogr A 1175(2):162-173 Stephens BS, Kapernick A, Eaglesham G, Mueller J (2005) Aquatic passive sampling of herbicides on naked particle loaded membranes: accelerated measurement and empirical estimation of kinetic parameters. Environ Sci Technol 39(22):8891-8897 UNEP (2001) Final act of the plenipotentiaries on the Stockholm Convention on persistent organic pollutants. United Nations Environment Program Chemicals, Geneva, Switzerland Vaes WHJ, Hamwijk C, Ramos EU, Verhaar HJM, Hermens JLM (1996) Partitioning of organic chemicals to polyacrylate-coated solid phase microextraction fibers: kinetic behavior and quantitative structure-property relationships. Anal Chem 68(24):4458-4462 Vrana B, Mills G, Greenwood R, Knutsson J, Svensson K, Morrison G (2005) Performance optimisation of a passive sampler for monitoring hydrophobic organic pollutants in water. J Environ Monitor 7(6):612-620. doi: 10.1039/b419070J Vrana B, Mills GA, Dominiak E, Greenwood R (2006) Calibration of the Chemcatcher passive sampler for the monitoring of priority organic pollutants in water. Environ Pollut 142(2):333-343 Vrana B, Mills GA, Kotterman M, Leonards P, Booij K, Greenwood R (2007) Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water. Environ Pollut 145(3):895-904. doi: 10.1016/j.envpol.2006.04.030 Vrana B, Popp P, Paschke A, Schuurmann G (2001) Membrane-enclosed sorptive coating an integrative passive sampler for monitoring organic contaminants in water. Anal Chem 73(21):5191-5200. doi:10.1021/ac010630z Vrana B, Vermeirssen ELM, Allan IJ, Kohoutek J, Kennedy K, Mills GA, Greenwood R (2010) Passive sampling of environmental pollutants in the aquatic environment: State of the art and perspectives. Position paper, www.norman-network.net/public_docs/ slides_prague/norman_position_paper_pas_sampling.pdf. Xiao H, Li NQ, Wania F (2004) Compilation, evaluation, and selection of physical-chemical property data for alpha-, beta-, and gamma-hexachlorocyclohexane. J Chem Eng Data 49:173-185 Xie WH, Shiu WY, Mackay D (1997) A review of the effect of salts on the solubility of organic compounds in seawater. Mar Environ Res 44:429-444 Springer Príloha 23 Jalová V., Jarošová B., Bláha L, Giesy J. P., Ocelka T., Grabic R., Jurčíková J., Vraná B., and Hilscherová K., Estrogen-, androgen- and aryl hydrocarbon receptor mediated activities in passive and composite samples from municipal waste and surface waters, Environ. Int., 2013, 59, 372-383. Environment International 59 (2013) 372-383 ELSEVIER Contents lists available at SciVerse ScienceDirect Environment International journal homepage: www.elsevier.com/locate/envint ENVIRONMENT INTERNATIONAL Estrogen-, androgen- and aryl hydrocarbon receptor mediated activities in passive and composite samples from municipal waste and surface waters V. Jalová a, B. Jarošová a, L. Bláha a, J.P. Giesy b'c'd'e'f, T. Ocelka g R. Grabic h, J. Jurčíková g, B. Vraná a, K. Hilscherová a'* a Research Centre for Toxic Compounds in the Environment (RECETOX), Masaryk University, Kamenice 753/5, 625 00, Brno, Czech Republic b Dept. of Biomedical Veterinary Sciences and Toxicology Centre, University of Saskatchewan, Saskatoon, Saskatchewan, Canada c State Key Laboratory of Pollution Control and Resources Reuse, School of the Environment, Nanjing University, Nanjing, 210046, PR China d Department of Zoology, Center for Integrative Toxicology, Michigan State University, East Lansing, MI, USA e School of Biological Sciences, University of Hong Kong, Hong Kong, China ' Department of Biology and Chemistry, State Key Laboratory for Marine Pollution, City University of Hong Kong, Hong Kong, China g Institute of Public Health Ostrava, National Reference Laboratory for POPs, Ostrava, Czech Republic h University of South Bohemia in Ceske Budějovice, Faculty of Fisheries and Protection of Waters, South Bohemian Research Center of Aquaculture and Biodiversity of Hydrocenoses, Zatisi 728/11, Vodnany, 389 25 Czech Republic ARTICLE INFO ABSTRACT Article history: Received 1 January 2013 Accepted 30 June 2013 Available online xxxx Keywords: Estrogenic Androgenic Cytotoxicity Bioassay in vitro Passive sampling Dioxin-like Passive and composite sampling in combination with in vitro bioassays and identification and quantification of individual chemicals were applied to characterize pollution by compounds with several specific modes of action in urban area in the basin of two rivers, with 400,000 inhabitants and a variety of industrial activities. Two types of passive samplers, semipermeable membrane devices (SPMD) for hydrophobic contaminants and polar organic chemical integrative samplers (POCIS) for polar compounds such as pesticides and pharmaceuticals, were used to sample wastewater treatment plant (WWTP) influent and effluent as well as rivers upstream and downstream of the urban complex and the WWTP. Compounds with endocrine disruptive potency were detected in river water and WWTP influent and effluent. Year-round, monthly assessment of waste waters by bioassays documented estrogenic, androgenic and dioxin-like potency as well as cytotoxicity in influent waters of the WWTP and allowed characterization of seasonal variability of these biological potentials in waste waters. The WWTP effectively removed cytotoxic compounds, xenoestrogens and xenoandrogens. There was significant variability in treatment efficiency of dioxin-like potency. The study indicates that the WWTP, despite its up-to-date technology, can contribute endocrine disrupting compounds to the river. Riverine samples exhibited dioxin-like, antiestrogenic and antiandrogenic potencies. The study design enabled characterization of effects of the urban complex and the WWTP on the river. Concentrations of PAHs and contaminants and specific biological potencies sampled by POCIS decreased as a function of distance from the city. © 2013 Elsevier Ltd. All rights reserved. 1. Introduction There is increasing evidence that environmental contaminants have the potential to disrupt endocrine processes. This might result in adverse effects on reproduction, cause certain cancers, and other toxicities related to (sexual) differentiation, growth, and development (Giesy et al., 2000; Miles-Richardson et al., 1999; Sanderson and van den Berg, 2003; Snyder et al., 2000). A variety of pollutants that are found in surface and waste waters, such as organochlorine pesticides (OCPs), polychlorinated biphenyls (PCBs), polychlorinated dioxins and furans (PCDD/Fs), polycyclic aromatic hydrocarbons (PAHs), alkylphenols, synthetic steroids, pesticides, pharmaceuticals and personal care products (PPCPs), but also natural products such * Corresponding author. Tel.: +420 54949 3256; fax: +420 54949 2840. E-mail address: hilscherova@recetox.muni.cz (K. Hilscherovä). 0160-4120/$ - see front matter © 2013 Elsevier Ltd. All rights reserved. http://dxdoi.org/! 0.1016/j.envint.2013.06.024 as phytoestrogens, have been shown to elicit endocrine disruptive effects. Sources of endocrine disrupting compounds (EDCs) are associated with larger urbanized and industrial areas. However, influences of smaller local sources can also be significant, especially where dilution is minimal (Jarošova et al., 2012). EDCs are also released to aquatic environments from both municipal and various industrial waste waters (Garcia-Reyero et al., 2004). Relative contributions of EDCs to surface waters depend on efficacies of sewage treatment systems, which is dependent on both capacity and technology of the wastewater treatment plant (WWTP). Potential risks of adverse effects of effluents from WWTPs to aquatic environments are influenced by volume of effluent, discharge of the receiving river, weather conditions and probably other factors that affect dissipation through dilution and/or degradation (Sumpter, 1995). Wastewater treatment plants receive mixtures of molecules from domestic, agricultural, and/or industrial wastes and V. jalová et al. I Environment International 59 (2013) 372-383 373 thus waste waters can contain mixtures of many of the above listed pollutants and their degradation products (Alvarez et al., 2005). Despite intensive removal of xenobiotics by municipal WWTPs, which can range from 88 to >99% and 96 to >99% for xenoestrogens and xenoandrogens, respectively (Korner et al., 2000; Leusch et al., 2010; Murk et al., 2002; Svenson and Allard, 2004), they often do not remove all chemicals from the effluent. Moreover, during treatment some contaminants can be deconjugated to their more biologically active forms (Desbrow et al., 1998). Thus, most effluents still contain complex mixtures of molecules, including transformation products formed during treatment. Adverse effects on endocrine function and/or reproductive health associated with exposure to effluents from WWTPs, which can persist several kilometers from the point of effluent entry (Harries et al., 1996), have been demonstrated in wild fish populations (Jobling et al., 1998) or fishes caged downstream from WWTPs (Snyder et al., 2004). Several studies combining the use of chemical analyses and in vitro assays have revealed steroid estrogens as the most potent endocrine disruptors in WWTP effluents with thresholds for adverse effects of a few ng/L (Korner et al., 2000; Matsui et al., 2000; Nakada et al., 2004; Routledge et al., 1998; Snyder et al., 2000). However, other EDCs can be effective in various landuse conditions (Sole et al., 2000) and special consideration should be paid to mixtures of pollutants. Also, more information is needed to assess the potential contribution from other sources than just the WWTPs. Selection of an appropriate sampling approach is crucial to determining the presence of contaminants and assessment of their potential for effects on aquatic environment. Traditional grab samples represent the immediate situation, thus only those contaminants present at the time of sampling are characterized. Episodic events such as spills or stormwater runoff can be missed since contaminants can dissipate prior to the next sampling (Alvarez et al., 2005; Huckins et al., 1990,1993). A more representative way to sample, that represents an integrated estimate of the time-averaged exposure is composite samples collected over time. But, even this type of extensive sampling represents isolated conditions over relatively short durations. This sort of intensive sampling program is resource-intensive, requiring sampling staff and/or special equipment, which cannot be easily employed at many sites, especially at locations where equipment might be at risk to vandalism. An alternative protocol is passive sampling, which enables estimation of time-weighted concentrations of contaminants and sequesters residues from episodic events commonly not detected by use of intermitent grab sampling. Passive sampling requires minimal resources of both personnel and equipment. Passive samplers have no moving parts to fail and require no electricity to function. They can be placed out of sight to avoid vandalism. Passive sampling can be used in situations of variable water conditions and because they concentrate residues from water they can enable detection of ultra-trace, yet toxicologically relevant concentrations of contaminant mixtures over extended durations (Alvarez et al., 2004). Other advantages include relatively simple, single deployment as compared to collecting and processing multiple water samples, greater mass of chemical residues sequestered, and the ability to detect chemicals which dissipate quickly (Alvarez et al., 2005; Huckins et al., 1990). Passive sampling also eliminates the need for some tedious and time-consuming cleanup steps associated with other types of sample collection. Semipermeable membrane devices (SPMDs) have been developed as in situ, integrating passive samplers for monitoring of trace-level, waterborne hydrophobic contaminants (Huckins et al., 1993) and have been used for effective sampling of multiple classes of chemicals, including PAHs, PCBs, OCPs, PCDD/Fs, alkylated phenols, moderately polar organophosphate insecticides, pyrethroid insecticides, neutral organometallic compounds, and certain heterocyclic aromatic compounds (Petty et al., 2000a). Since SPMDs can mimic accumulation by aquatic organisms that can bioconcentrate trace amounts of organic contaminants, SPMDs measure not only the presence, but also the bioavailability and bioconcentration potential of organic contaminants (Huckins et al., 1990; Petty et al., 2000b). Polar Organic Chemical Integrative Samplers (POCIS) sequester waterborne hydrophilic contaminants, such as polar pesticides, pharmaceuticals, ingredients from personal care and consumer products, natural and synthetic hormones (Alvarez et al., 2004, 2005; Petty et al., 2004). Depending on the sorbent used, POCIS can be modified for sampling of general hydrophilic contaminants or pharmaceuticals (Alvarez et al., 2005). The aim of this study was characterization of the influence of the industrialized urban region of Brno, Czech Republic and its associated municipal WWTP on contamination of the Svratka and Svitava rivers by compounds with endocrine disruptive potency by joint use of bioassays, two types of passive samplers and identification and quantification of selected organic chemicals. One goal was to assess the year-round variability in endocrine disruptive potency of WWTP influent and effluent water and thus treatment efficiency for EDCs by collecting composite samples monthly. The second major goal was to determine the relative magnitude of contributions of the urban area and the WWTP on contamination of these two urban rivers by endocrine disruptive compounds that can modulate the arylhydrocarbon (AhR), estrogen (ER) and androgen (AR) receptors. A battery of in vitro bioassays was used to assess potencies of agonists of these three receptors. Two types of passive samplers, POCIS and SPMD, were used to collect integrated samples of hydrophobic and hydrophilic compounds and assess their potencies to interfere with the three receptors signalling. 2. Materials and methods 2.1. Sampling design Samples were collected from the region around Brno, the second largest metropolitan district of the Czech Republic in Central Europe. The metropolitan region of Brno with more than 400,000 inhabitants is spread through the basin formed by the Svratka and Svitava Rivers. The city has a central wastewater treatment plant and a variety of industrial activities. The municipal WWTP treats wastewater conveyed by a system of sanitary sewers from the city of Brno and increasingly also by a system of pumping stations from its surroundings. The WWTP was recently reconstructed and enhanced to a capacity of 513,000 population equivalent with permissible volume of discharged wastewater of 4222 L/s. Waste water is subjected to primary (mechanical) treatment followed by biological stage of activation with pre-denitrification and anaerobic phosphorus removal (system of circulatory activation with change of anaerobic, anoxic and aerated zones). Excess activated sludge is then anaerobically stabilized (Brněnské vodárny a kanalizace, 2010; Ministry of the Environment, 2010). The influent and effluent of the WWTP were sampled monthly from May 2007 until April 2008. In addition, SPMD and POCIS passive samplers were placed in the influent (site 5) and effluent (site 6) of the WWTP and at seven sites in the Svratka, Svitava and Bobrava Rivers at locations upstream and downstream of Brno and downstream of the WWTP effluent (Fig. 1). Passive samplers were deployed for 23 days and collected during October 2007. Sampling locations in the Svratka River were: Kninicky (site 1) upstream of the city of Brno (downstream of the dam of Brno reservoir) and a site downstream of Brno upstream of the confluence with the Svitava River (Svratka before confluence, site 2). Locations monitored in the Svitava River included Bílovice and Svitavou (site 3), a small town upstream of Brno, and another site downstream of Brno upstream of the confluence with the Svratka River (Svitava before confluence, site 4). Another sampling site was selected in the Bobrava River (site 9), which is a tributary affected mostly by agriculture that flows into the Svratka River downstream of the WWTP. Downstream of the WWTP and the confluence of the Bobrava and Svratka rivers samples were collected near a small town Rajhradice (site 7) and at Zidlochovice (site 8, approximately 20 km downstream from Brno). 374 V. Jalová et al. I Environment International 59 (2013) 372-383 Germany ■WWTP Legend • sampling site u J built up area Fig. 1. Map of the Czech Republic showing locations of sampling sites in the vicinity of Brno. Sampling sites: 1—Svratka River, Kninicky, 2—Svratka River before confluence, 3— Svitava River, Bílovice nad Svitavou, 4—Svitava River before confluence, 5—WWTP Modrice, influent, 6—WWTP Modrice, effluent, 7—Svratka River, Rajhradice, 8—Svratka River, Zidlochovice, 9—Bobrava River. 2.2. Passsive water sampling and preparation of extracts SPMD and POCIS disks were obtained from Exposmeter AB, Tavelsjo, Sweden. Prior to passive sampling, the sampling protocol was prepared with QA/QC. One POCIS was used for both chemical analysis and bioassay testing. Two SPMDs were used in duplicates for chemical analysis, one SPMD was used for toxicity assessment. SPMDs for chemical analysis contained performance reference compounds (PRC) used as onsite SPMDs calibration. Four deuterated PAHs ([2Hin]acenaphthene, [2Hi0]fluorene, [2Hi0]phenanthrene, and [2H12]chrysene) and four 13C12-labeled PCBs (PCB 3, 8, 37, and 54) were used as PRCs. Transport, field and laboratory blanks were used. A standard sampling arrangement was used as described in Grabic et al. (2010). It consists of a combination of POCIS and SPMDs mounted on commercially available stainless steel holders in protective deployment canisters made of perforated stainless steel plates. These samplers were suspended at 0.5-1 m depth of the water column in cryptic locations to minimize vandalism. After exposure for 23 days, samplers were recovered, cleaned and sealed in airtight, metal cans and placed on ice in a cooler for transport to the laboratory. Membranes were stored in sealed cans in a freezer at —18 °C until analysis. Before analysis SPMDs were cleaned and dialyzed with hexane in accordance with previously published methods (Ellis et al., 1995). Combined dialysates were adjusted to a volume of 10 mL. Chemical residues sampled by POCIS were recovered from the sorbent by organic solvent elution with a combination of methanol:toluene:dichloromethane (1:1:8, v/v/v). Volumes of all extracts were reduced by rotary evaporation and under a gentle stream of nitrogen, then solvent was exchanged to methanol (Alvarez et al., 2005). The final equivalent concentrations were 1 sampler/mL. A portion of each extract was transferred into DMSO for testing in bioassays. 2.3. Processing of waste water Samples of influent and effluent were collected from the municipal WWTP on the Svratka River, downstream of Brno, once a month for 12 months. Water was collected every 2 h and composited over a 24-h period. Samples of influent were prefiltered through glass wool and 47 mm diameter glass fiber filter with 2.7 um pores (Filap, Czech Republic) and both influent and effluent samples were filtered through glass fiber filters (1 um pores, Whatman, Sigma-Aldrich, Czech Republic) to prevent solid phase extraction (SPE) cartridges from clogging during later extraction. Filters were extracted and tested separately to ensure that no compounds with significant potency in any of the assays were removed by filtration. Organic compounds in filtrates were extracted within 24 h by SPE by use of Oasis HLB cartridges (Waters, Czech Republic). Cartridges were activated by methanol and equilibrated by water according to producer instructions. After samples had passed through cartridges, they were dried by air for 10-15 min and eluted by use of 15 mL methanol. Extracts were rotary evaporated to reduce the volume to approximately 2 mL and then evaporated in a gentle stream of nitrogen to final volumes of 1 mL. 2.4. Instrumental analyses Organic extracts of SPMD and POCIS samplers were analyzed for wide range of organic compounds. Samples were analyzed in accordance with standard EN ISO/IEC 17025. Detailed analytical procedures were described in Grabic et al. (2010). A set of internal standards was used in the analyses. These included carbon 13Ci2-labeled PCBs (3, 15, 31, 52, 118, 153, 180, 194, 206, 209), TCS, PFOC (perfluorooctanesulfonic acid [PFOS], perfluoro-nonanoic acid [PFNA], perfluoro-octanoid acid [PFOA]), and native standards purchased from Wellington Laboratories (Canada). 13C-labeled OCPs (7-HCH and DDE), PAH (13C2_6-labeled PAHs U.S. Environmental Protection Agency [U.S. EPA] 16 PAH cocktail), and polar compounds (simazine, 2,4-D, sulfamethoxazol, ciprofloxacin) were purchased from Cambridge Isotope Laboratories (USA). The native ones were purchased from Dr. Ehrenstorfer, AccuStandards, and Absolute Standards via Labicom (Czech Republic). All solvents, including hexane, dichloromethane, acetone, toluene (SupraSolv purity), water, and methanol (hypergrade for LC/MS) were of the highest quality from Merck (Germany). Organic extracts of SPMDs were characterized by quantifying 16 US EPA polycyclic aromatic hydrocarbons (PAH): acenaphthene, acenaph-thylene, anthracene, benzo[a]anthracene, benzo[a]pyrene, benzo[o] fluoranthene, benzo[ghi]perylene, benzo[k]fluoranthene, chrysene, dibenzo[a,h]anthracene, fluoranthene, fluorene, indeno(l,2,3-cd)pyrene, naphthalene, phenanthrene, and pyrene), polychlorinated biphenyls (PCBs): tri-, tetra-, penta-, hexa-, hepta-, octa-, nona-, and deca-congeners, organochlorine pesticides (OCPs): hexachlorbenzene, a-, (3-7-, 8-stereoisomers of hexachlorohexane (HCH), two congeners of dichlorodiphenyltrichloroethane (DDT) and its degradation products, V. jalová et al. / Environment International 59 (2013) 372-383 375 dichlorodiphenyldichloroethylene (DDE) and dichlorodiphenyldi-chloroethane (DDD), triclosan (TCS) and its environmental transformation product methyl triclosan (MeTCS) and polybrominated diphenyl ethers (PBDEs), expressed as the sum of congeners. POC1S extracts were analyzed for polar pesticides, pharmaceuticals and perfluorinated compounds (PFCs), expressed as the sum of per-fluoroorganic compounds (PFHxS, FHUEA, FOSA, N-MeFOSA, PFOA, PFOS, PFNA). A complete list of individual pesticides and pharmaceuticals analyzed in POC1S is attached in footnotes to Table 1. Gas chromatography/mass spectrometry (GC/MS) was used for identification and quantification of PAHs. PAHs with more rings that could not be analyzed by use of GC/MS were analyzed by use of high performance liquid chromatography with fluorescence detector (HPLC/FLD). Quantification of PCBs, OCPs, PBDEs, triclosan and its metabolite were performed by GC/MS-MS. Polar pesticides, pharmaceuticals and PFCs were identified and quantified by use of HPLC/MS-MS. Limits of detection for identified groups of chemicals were as follows: PAHs 3 ng/SPMD, MeTCS/TCS 3 ng/SPMD, OCPs 0.2 ng/SPMD, PCBs 0.1 ng/SPMD, polar pesticides: 0.5-5 ng/POCIS, antibiotics: 1-2 ng/POCIS, other pharmaceuticals 5 ng/POCIS. Analytical procedure involved evaluation of recoveries of internal standards. Recoveries were within following ranges: PAHs: 80-100 %, MeTCS/TCS: 60-100 %, OCPs, PCBs: 60-100 %, polar pesticides, pharmaceuticals: 55-80 %. Both trip and analytical blanks were analyzed. Laboratory blanks were subtracted. Trip blanks contributed 0-5 % of the total exposure, therefore no subtraction was performed. 2.5. In vitro bioassays Four transactivation reporter gene bioassays were used to assess receptor-mediated potencies of organic extracts of waters from the WWTP and passive samplers. All assays were conducted in 96 well microplates and included several dilutions of extracts in triplicate to provide a dose-response curve for each sample. All media and chemicals were purchased from Sigma-Aldrich (Czech Republic) unless otherwise specified. 2.5.1. AhR-mediated potency AhR-mediated (dioxin-like) potency was determined by use of the H411E-/uc bioassay, which is rat hepatoma cell line containing a lucif-erase reporter gene under control of dioxin-responsive enhancers (DRE) (Hilscherova et al., 2001; Sanderson et al., 1996; Villeneuve et al., 2002). H411E-/uc cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (BioTech, Czech Republic) supplemented with 10% fetal calf serum Mycoplex (PAA, Austria). The H411E-/uc cells were seeded in the culture medium at density of 15,000 cells/well and after 24 h exposed to samples, calibration reference or solvent control. Standard calibration was performed with 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD; Ultra Scientific, USA; dilution series 1-500 pM). After 24 h of exposure, intensity of luciferase luminescence corresponding to the receptor activation was measured by use of Promega Steady GloKit(Promega, USA). 2.5.2. ER-mediated potency Estrogen receptor mediated potency was evaluated by use of the MVLN bioassay, a human breast carcinoma cell line transfected with the luciferase gene under control of estrogen receptor activation (Demirpence et al., 1993; Freyberger and Schmuck, 2005; Hilscherova et al., 2002). MVLN cells were cultured in medium DMEM/F12 supplemented with 10% fetal calf serum Mycoplex (PAA, Austria). MVLN cells were seeded at density of 20,000 cells/well in DMEM/ F12 supplemented with 10% dialyzed fetal calf serum (PAA, Austria), which was additionally dextran/charcoal treated to further decrease background concentrations of hormones. Approximately 24 h after plating, cells were exposed to samples, calibration reference or solvent control in DMEM/F12. Standard calibration was performed with 17p-estradiol (E2; dilution series 1-500 pM). Effects of extracts on MVLN were assessed either singly or in combination with competing Table 1 The results of chemical analysis of passive samplers extracts. Ranges: the sum of detected compounds—the sum of detected compounds plus limit of detection for the nondetected compounds. POCIS Pesticides3 Sulfonamides'3 Other antibiotics1 Other pharmaceuticals'1 PFCs Sampling site ng/POCIS 1 376-464 157-172 12-68 231-239 6-9 2 285-388 104-128 2-52 253-261 3-6 3 382-491 824-838 54-105 904-911 33-36 4 463-603 721-733 32-81 808-814 38-41 5 279-394 924-938 290-317 1242-1249 12-15 6 2726-2836 10,087-10,104 1534-1551 18,550-18,559 272-274 7 474-599 992-1004 120-157 1344-1350 29-32 8 342-441 889-903 98-138 1147-1154 21-24 9 613-723 926-938 51-108 1003-1009 10-12 SPMD PAHs PCBs OCPs Triclosan MeTriclosan PBDEs Sampling site ng/L Pg/L Pg/L Pg/L Pg/L Pg/L 1 40.8 408-438 809-825 431 168 16-27 2 52.9 724-734 831-845 190 155 8.8-14 3 38.2 2155-2168 737-747 360 812 21-28.2 4 40.8 1370-1373 718-720 247 642 13.7-16.8 5 2160 825-861 831-839 32,817 84.2 162 6 31.6 1440-1446 1183-1194 8747 24,365 136-140 7 36.2 1252-1259 775-782 1115 3197 27.6-30.2 8 28.6 1548-1567 1040-1044 1680 3344 30.3-37.4 9 51.2 507-526 684-701 554 867 10.3-19.2 a Pesticides: clopyralid, bentazone, bromoxynil, 2,4-D, MCPA, dichlorprop, mecoprop (MCPP), 2,4,5-T, imazethapyr, thifensulfuron-methyl, methamidophos, nicosulfuron, rimsulfuron, metamitron, dimethoat, atrazin_desethyl, metoxuron, phosphamidon, cyanazin, metribuzin, simazin, bromacil, carbofuran, hexazinon, thiophanate-methyl, monolinuron, chlorotoluron, isoproturon, metobromuron, atrazin, desmetryn, dichlobenil, methabenzthiazuron, diuron, methidathion, ethofumesat, azoxystrobin, linuron, terbuthylazine, chlorbromuron, propyzamide, prometryn, metolachlor, fenhexamid, fenarimol, acetochlor, terbutryn, fipronil, kresoxim-methyl, tebuconazole, diazinon, propiconazole, phorate, phosalone, fluazifop-p-butyl, tri-allate, pyridate, alachlor, metalaxyl. b Sulfonamides: sulfapyridin, sulfamethazin, sulfamethoxypyridazin, sulfachloropyridazin, Sulfamethoxazol. c Other antibiotics: metronidazol, cefalexin, ofloxacin, norfloxacin, ciprofloxacin, enrofloxadn, erythromycin, trimetoprim. d Other pharmaceuticals: diaveridin, Carbamazepin, didofenac. 376 V. Jalová et al. / Environment International 59 (2013) 372-383 endogenous ligand (33 pM 17p-estradiol)—given concentration is near its EC50 value. Exposure duration and final measurement was the same as in the case of H411E-/uc bioassay described above. 2.5.3. AR-mediated potency (Anti)androgenicity of passive samplers extracts was assessed in a bioassay with MDA-kb2 cells, a human breast carcinoma cell line stably transfected with luciferase reporter gene under control of functional endogenous androgen receptor (AR) and glucocorticoid receptor (GR) (Wilson et al., 2002). MDA-kb2 cells were cultured in L-15 Leibovitz medium supplemented with 10% fetal calf serum Mycoplex (PAA, Austria). MDA-kb2 were seeded at density of 50,000 cells/well and exposed after 24 h to samples, calibration reference or solvent control in L-15 Leibovitz medium supplemented with 10% dextran/ charcoal treated dialyzed fetal calf serum. Standard calibration was performed with dihydrotestosterone (DHT; dilution series 1 pM-10 uM). In addition to androgenic effects, antiandrogenicity was assessed in combination with competing endogenous ligand (1 nM dihydrotestosterone). After 24 h of exposure, intensity of luciferase luminescence was measured with prepared luciferase reagent (Wilson et al., 2002). Organic extracts of influent and effluent waters were assessed in a bioluminescent yeast assay based on recombinant Saccharomyces cerevisiae cells modified to express human androgen receptor along with firefly luciferase under transcriptional control of androgen-responsive element to detect compounds affecting AR-mediated hormonal signalling. The assay with the androgen-responsive yeast model was performed according to Leskinen et al. (2005). Yeast cells were seeded in 96-well microplates and exposed to reference testosterone (T; dilution series 1 pM-10 uM), the sample alone or in combination with testosterone (10 nM) to determine antiandrogenic effect. Yeast cells were incubated for 2.5 h and then the signal was detected after addition of D-luciferin substrate. 2.5.4. Cytotoxicity Non-cytotoxic sample concentrations to be used in each bioassay with mammalian cell lines were determined by use of the neutral red uptake assay (Freyberger and Schmuck, 2005). Particular bioas-says with individual cell lines were processed as previously described. At the end of the exposure period, neutral red solution (0.5 mg/mL of media) was added and cells were incubated for 1 h at 37 °C. Medium was removed and cells washed with PBS and lysed with 1% acetic acid in 50% ethanol. Absorbance was measured in a microplate spectrophotometer at 570 nm. Yeast strain of recombinant S. cerevisiae constitutively expressing luciferase, which has shown greater sensitivity compared to the mammalian cells, was used for detailed cytotoxicity assessment (Leskinen et al., 2005; Michelini et al., 2005). Complete dose-responses relationships of cytotoxic effects for all samples were determined after 2.5 h exposure. The intensity of luciferase luminescence after addition of D-luciferin corresponded to the number of surviving cells (Leskinen et al., 2005). 2.6. Data analysis Sample responses expressed as relative luminescence units were converted to percentage of maximum response of the standard curves (% TCDDmax/E2max/DHTmax/Tmax). The response of the solvent control was substracted from both standard and sample responses prior to the conversion. EC values were calculated by nonlinear logarithmic regression of dose-response curves of calibration standards and samples (Graph Pad Prism, GraphPad® Software, San Diego, California, USA). Relative potencies expressed as TCDD equivalents (BIOTEQJ/E2 equivalents (EEQJ/androgen equivalents (AEQJ were calculated by relating the EC50 value of standard calibration with the concentration of the tested sample inducing the same response (Villeneuve et al., 2000). Due to cytotoxicity, it was not possible to obtain complete dose-response curves in testing of waste water samples in the yeast assay. Thus, their AEQ values were calculated as point estimates because maximum detected luminescence induction at noncytotoxic concentrations did not exceed 15%. Cytotoxicity, antiestrogenicity and antiandrogenicity corresponded to the decrease in detected luminescence/absorbance signal given by solvent control in case of cytotoxicity and specified amount of competing standard ligand for the other effects. 1C50 values for antiestrogenicity and antiandrogenicity or lC2o values in cases that the effects did not cause 50% response, were calculated from dose-response curves expressed in percentage of signal of competitive concentration of added natural ligand (33 pM E2, 1 nM DHT, 10 nM testosterone). For better clarity of the trends in graphs the values are expressed as an index of antiestrogenicity (AE) or antiandrogenicity (AA), which corresponds to reciprocal value of IC20 or 1C50. Similarly, the index of cytotoxicity was derived as the reciprocal value of 1C20 or 1C50 for the cytotoxic response. 2.7. Calculation of dissolved water concentrations from passive sampler data Concentrations of target analytes in water were calculated from the mass absorbed by the SPMD, the in situ sampling rate of the compounds and their sampler-water partition coefficients using the kinetic uptake model by Huckins et al. (2006). Sampling rates of target compounds were estimated from dissipation of performance reference compounds (PRCs) from SPMDs during exposure using nonlinear least squares method by Booij and Smedes (2010), considering the fraction of individual PRCs that remain in the SPMD after the exposure as a continuous function of their partition coefficients, with sampling rate as an adjustable parameter. The necessary sampler-water partition coefficients values were estimated from the respective octanol/water partition coefficients according to Huckins et al. (2006). For the purpose of comparison of toxic potencies of extracts from SPMDs from different sampling sites the measured toxic equivalent concentrations (TEQJ in extracts [ng/SPMD] were translated to water concentrations Cw-teq [ng/L or pg/L] at the individual sites. Since phys-icochemical properties of the compounds that exhibit bioassay response in the extracts are not known, linear uptake was assumed (Eq. (1)). C (1) w—TEQ — ^ j \l) Where: Rs is the sampling rate and t is the exposure time. The necessary Rs values were obtained using the PRC model described above. Since Rs is only a weak function of hydrophobicity, values of Rs with a medium molecular mass (MW = 300) were applied in all calculations. For POC1S data, no correction for the potential effect of environmental variables was performed and results were simply compared on the basis of toxic equivalent concentrations (TEQ) in sampler extracts [ng/POCIS]. It has been demostrated that water flow rate has a relatively minor influence on the accumulation of a number of pollutants including EDCs into POC1S (Li et al., 2010). Thus, it appears not necessary to adjust sampling rates for POC1S when they are deployed in areas where the water flows vary only slightly. 3. Results 3.1. Concentrations of individual residues Greatest concentrations of polar pesticides, pharmaceuticals and perfluoroorganic compounds in POC1S were detected at site 6 (WWTP effluent) (Table 1). Concentrations of contaminants found in V. jalová et at. I Environment International 59 (2013) 372-383 377 POCIS from WWTP influent (site 5) were less than in POC1S at WWTP effluent and comparable or greater than in those from the other sites. The explanations of greater detected levels of some contaminants and biological potencies in passive samplers from WWTP effluent are elaborated in detail in the Discussion section. Concentrations of some pharmaceuticals in POCIS from the sites upstream of Brno were slightly greater than downstream, but concentrations in the Svratka River were generally approximately 4-fold less than in the Svitava River. Similarly, concentrations of PFCs were approximately 6-fold greater in Svitava than in Svratka, while concentrations of pesticides were comparable in both rivers. Greater concentrations of pesticides were found at site 9 on the tributary of the Svratka River. Concentrations of pharmaceuticals were greater bellow the WWTP effluent. There was a slight decrease of concentrations of contaminants in POCIS as a function of distance from the city and WWTP. The greatest concentrations of most pollutants sampled by SPMD were observed in samples from the WWTP, with concentrations of PAHs and triclosan greatest in the influent (site 5), while concentrations of methyl triclosan were greatest in the effluent (site 6) (Table 1). Greater concentrations of PCBs and methyl triclosan were detected already upstream of Brno in the Svitava River (sites 3, 4). Concentrations of most pollutants did not increase much directly downstream of Brno on both rivers (sites 2, 4), except for PCBs in the Svratka River. Concentrations of PAHs were slightly lesser downstream of the WWTP (site 7) and further decreased at the longer distance from the city (site 8), while no such trend was observed for concentrations of PCBs and OCPs. Concentrations of PBDEs, triclosan and methyl triclosan were significantly greater downstream of the WWTP. I 1 =í o 15 %. 20 0 0 g o 111 ^ £ _> 10 Ik □ Influent B 2000 -i 0 in O 3- 1500 - >. o X 0 1000 - 0 %. 0 0 500 ■ X . as Č3 12- 'x 0 cytol 8- 0 ndex 4- n n n n 123456789 Sampling site Fig. 2. Cytotoxicity of samples extracts detected in the bioluminescent yeast assay: (A) influent and effluent water samples from the WWTP; (B) POCIS (Index of cytotoxicity expressed as reciprocal value of IC50, [sampler/mLp1); (C) SPMD (Index of cytotoxicity expressed as reciprocal value of IC2o, [L/mLp1); no column = no significant activity. insert). Concentrations of BIOTEQs were between 0.3 and 2 ng TCDD/ POCIS. Potency detected in the WWTP effluent (site 6) was 5-fold greater than that in the influent (site 5). All extracts of SPMD contained detectable AhR-mediated potency with the greatest response in the WWTP influent sample (site 5) and also in the Bobrava River which was affected by agriculture (site 9, Fig. 3B). Concentrations of BIOTEQ determined from SPMD ranged from 8.2 to 14.6 pg TCDD/L. 3.4. ER-mediated potency Potency of ER agonists was detected in water from the WWTP during all samplings throughout the year (Fig. 4). Values of 17p-estradiol (E2) equivalents (EEQJ varied from 5.4 to 124 ng E2/L in influent and from 0.1 to 5.1 ng E2/L in effluent. Efficiency of treatment to remove EEQ ranged from 80 to greater than 99 %. POCIS sample from the WWTP influent (site 5) had a concentration of EEQ of 7.3 ng E2/ 378 V. Jalová et al. / Environment International 59 (2013) 372-383 A 4 -i FCDD, 3 ■ OJ c 2 ■ O LU 1— 1 ■ O CD 0 ■ □ Influent .<£ .<$» o* .vq?> 140 - 120 - _l Ki 100 - LU E3J 80 - C 60 - O LU 40 - LU 20 - 0 - □ Influent ■ Effluent fin n c?1 ,č?> x$ ^ ^ ^ o* o* o* # Fig. 4. Estrogenic potency, expressed as estradiol equivalents (EEQJ of extracts of WWTP influent and effluent water, detected in MVLN assay; no column = no significant activity. greatest antiandrogenic potency in extracts of POC1S was observed at site 4 in the Svitava River, directly downstream of Brno (Fig. 6A). The antiandrogenic potency of the extract of the POC1S exposed to WWTP influent (site 5) was comparable with the potency observed in samples from most sites on the rivers. There was no antiandrogenic potency observed in POC1S exposed to WWTP effluent (site 6). There was generally no antiandrogenic potency in extracts of SPMD exposed upstream of the WWTP, while there was antiandrogenic potency in samples from the WWTP (sites 5, 6) and from sites downstream of the WWTP. The antiandrogenic potency of compounds sampled by SPMD was approximately 60% greater in WWTP influent than that in effluent (Fig. 6B). Sampling site Fig. 3. AhR-Mediated (Dioxin-like) potency of samples extracts detected in H4IIE-/uc assay expressed as BIOTEQ. equivalents: (A) influent and effluent water from the WWTP; (B) SPMD and POCIS. sampler. The concentration of EEQ in the extract of POCIS exposed to effluent (site 6) was less than 0.6 ng E2/sampler, which was the limit of detection. There were no EEQ detectable in POCIS from the rivers or in any SPMD samples. Influent and effluent water samples from the WWTP showed no significant antiestrogenic potency when tested in the presence of E2. Alternatively, antiestrogenic potency was detected in extracts of SPMD and POCIS from all sites. Data from SPMDs indicate greater antiestrogenicity in sites from river Svratka compared to Svitava already upstream of Brno. Greatest antiestrogenicity was observed in POCIS exposed to WWTP effluent while all samples from rivers and WWTP influent showed comparable potency (Fig. 5). 0J CS! o _ .5Í O OJ "O B 2400 -i 2000 - 1600 - 1200 - 800 - 400 - n 4 5 6 7 Sampling site 24 3.5. AR-mediated potency Significant androgenic potencies were found mostly at the greatest non-cytotoxic concentrations of influent water samples and concentrations of androgen equivalents (AEOJ ranged from <23 to 193 ng testosterone/! (Table 2). Concentrations of AEQdetermined for non-cytotoxic concentrations of effluent extracts were less than the limit of detection, which was 1-4 ng testosterone/L Efficiency of treatment to remove androgenic compounds was greater than 96-99%. POCIS from WWTP influent and effluent were the only other samples to exhibit detectable AEQ with concentrations of 32.6 and 6.9 ng DHT/sampler, respectively. No antiandrogenic potency was observed in non-cytotoxic concentrations of samples from influent or effluent water from the WWTP. Antiandrogenic potency in competition with the added endogenous ligand DHT was detected in most extracts of SPMD and POCIS. The & 20 - 5 0 I 1 I I I I I I I I I 123456789 Sampling site Fig. 5. Antiestrogenic potencies of samples extracts determined by use of the MVLN assay in the presence of 33 pM estradiol expressed as index of antiestrogenicity: (A) POCIS— reciprocal value of IC50 [sampler/mLp1), (B) SPMD—reciprocal value of IC20 [L/mLp1). V. jalová et at. I Environment International 59 (2013) 372-383 379 Table 2 Androgenic activity of influent and effluent water extracts from the WWTP detected in the yeast assay. (LOD ranged from 1.3 to 70 ng testosterone/L because of variable cytotoxicity of samples). Sampling date AEQ. (ng testosterone/L) Influent Effluent May 07 155 <3.7 June 07 97 <2.2 July 07 <70 <2.2 August 07 <70 <2.6 September 07 <23 <1.3 October 07 80 <1.3 November 07 193 <1.3 December 07 96 <1.3 January 08 107 <1.3 February 08 140 <1.3 March 08 47 <1.3 April 08 35 <1.3 4. Discussion Rivers can be contaminated by many chemicals, some of which have the potential to affect normal reproduction, development and behavior of wildlife species and potentially also human health. Some of these compounds can be released to rivers from large city agglomerations via WWTP and other point-discharge or diffuse sources (Cargouet et al., 2004; Jobling et al., 1998; Sabaliunas et al., 2000; Snyder et al., 2000). In recent years, WWTP have been studied as potential sources of endocrine disruptive compounds to the aquatic environment (Harries et al., 1996; Murk et al., 2002; Tan et al., A 1000 n 123456789 Sampling site Fig. 6. Antiandrogenic potency of samples extracts determined by use of the MDA-kb2 assay in the presence of 1 nM dihydrotestosteron (DHT), expressed as an index of anti-androgenicity (reciprocal value of IC50): (A) POCIS [sampler/mL]"1, (B) SPMD [L/mL]-1; no column = no significant activity. 2007). There are several studies that have investigated WWTPs by use of various approaches including passive sampling combined with instrumental analysis and/or bioassays (Tan et al., 2007; Vermeirssen et al., 2005). However, there has been less information on other possible sources. Moreover, the studies using bioassays were focused mainly on estrogenic potency and there is limited data on other specific biological potencies in mixtures extracted from surface or waste waters. In addition, mostly known endocrine disruptive compounds, such as estrogens, androgens, phthalates or alkylphenols are analyzed, but more data is needed for other pollutants, such as widely used compounds from the group of pharmaceuticals and personal care products. In this study potencies for ligands in mixtures to interact with specific receptors as well as concentrations of several classes of pollutants were measured in waste waters and surface waters of two rivers in an urban metropolitan area in Central Europe with a variety of industries and modern recently renovated WWTP with advanced treatment capacity and efficiency. The sampling design and a complex approach using passive sampling along with chemical analysis and bioassays enabled to characterize the distribution and sources of pollutants in the model part of river basin. Based on measured residues, water of the Svitava River upstream of Brno seems to be more polluted than the Svratka River. Specifically, concentrations of pharmaceuticals, PFCs, PCBs and methyl triclosan were lower in the Svratka River. Furthermore, greater potencies for cytotoxicity of the hydro-philic fraction were observed in the Svitava River upstream of Brno. These data point to some pollution sources on river Svitava upstream of Brno agglomeration. There was no obvious influence of the city itself or WWTP on the concentrations of PAHs and organohalogenated compounds except of somewhat increased PCBs in Svratka downstream of Brno. Thus, neither runoff from the metropolitan region of Brno nor the effluent of the WWTP contributed significantly to the pollution with these compounds. Alternatively, concentrations of pharmaceuticals, antibiotics, triclosan and PBDEs were not affected by the city, but increased downstream of the WWTP, despite its up-to-date treatment technology. The data from passive samples document highly efficient removal of hydrophilic antiandrogenic and about 60% removal of hydrophobic antiandrogenic pollutants during WW treatment. Despite this removal, the concentrations of hydrophobic antiandrogenic pollutants in the river increased downstream of the WWTP similarly to the cytotoxic potency. Concentrations of triclosan and methyl triclosan were increased by the WWTP. For polar pesticides there was no influence of the city itself or WWTP. Concentrations of most of the polar compounds sampled by POCIS and associated biological potencies went down at the last study site about 20 km downstream of the city. There was no such decrease in levels of hydrophobic pollutants sampled by SPMD and their biological potencies, except of PAHs. The decrease of PAHs concentrations downstream of WWTP was not due to particle adsorption and sedimentation after flow out from WWTPs, since there was no increase of PAHs levels in river sediments (data not shown). For all pollutants sampled by POCIS as well as some pollutants sampled by SPMD, the greatest concentrations were detected in WWTP effluent. Similarly, in the POCIS exposed to effluent there was also the greatest cytotoxicity, dioxin-like and antiestrogenic potency. All these concentrations and potencies were greater than for the WWTP influent. There are at least two explanations of the observed elevated concentrations and toxic potencies of compounds accumulated in passive samplers in the WWTP effluent in comparison to influent. Passive sampling methods measure the concentration of freely dissolved contaminants, which is directly related to the contaminants' chemical activity (Mayer et al., 2003). This also indicates the bioavailability or pressure (fugacity) of contaminants on organisms and consequently represents the exposure level for organisms. In the WWTP influent hydrophobic compounds are largely sorbed to the suspended particulate material so that their freely dissolved concentration is small (Lohmann et al., 2012). In the wastewater 380 V. Jalová et al. I Environment International 59 (2013) 372-383 treatment process the content of suspended material is efficiently reduced, which in turn results in a strong decrease of sorption capacity for hydrophobic compounds in WWTP effluent. However, some persistent compounds are not eliminated by the treatment process. As a result of the reduced uptake capacity of the particulate matter, free dissolved concentrations (chemical activity) in the effluent are higher than in the influent, which is in turn reflected in their levels found in passive samplers, especially in SPMDs. Differences in uptake might be affected by different passive sampler exposure conditions in WWTP influent and effluent, respectively. Among potential factors that affect uptake kinetics into passive samplers, hydrodynamics and fouling are the most important ones. The visual observation of channels in WWTP influent and effluent indicates a similar turbulent water flow character in both cases. Thus, influent/effluent differences in hydrodynamics can hardly explain the observed up to ten-fold increase in accumulated amounts of some compounds in passive samplers (e.g. compounds in POC1S; Table 1). We hypothesize that fouling of samplers is the more important factor that affects the uptake of both hydrophobic as well as hydrophilic compounds into passive samplers. The raw waste water is a very complex mixture which contains debris, mud, various particles and even dispersed emulsions of liquids that are non-miscible with water (such as fats). Fouling and layers of dirt can reduce uptake of compounds into passive samplers (Stuer-Lauridsen, 2005) and lead to lower sampling rates by a) physical blockage of active surface of samplers by debris; b) thickening the diffusion barriers; c) reduction of the driving force for sampler uptake by shifting the partitioning equilibria between sampler and the surrounding environment. Our study indicates that passive sampling (especially for POC1S samples) may not be a reliable method in raw sewage water and could lead to significant underestimation of actual concentrations of dissolved pollutants. This problem is really specific to the raw sewage water and does not concern passive samples from any other site. Most studies using in vitro assays include cytotoxicity tests, which determine the greatest possible sample concentration that is not cytotoxic for the cells to be used as the maximal tested concentration for the specific effects. In this study, dose-response curves and 1C50 of extracts on yeast cells were determined. The efficient decrease of cytotoxicity in SPMD and waste water after waste water treatment might be due to activated sludge processes as well as flocculation, which have been shown to have the greatest efficiency of removal of cytotoxic compounds (Ma et al., 2005). Cytotoxicity of waste waters did not correlate with estrogenic or androgenic potencies of these waste waters. This observation is consistent with the results reported by Vega-Lopez et al. (2007), who found no correlation between estrogenic disruption and toxicity determined in MCF-7 cells for samples of water from two Mexican lakes, which receive domestic and industrial wastewaters after secondary treatment. These results support the theory that estrogenic potency in waste waters is caused primarily by steroidal estrogens, which are potent at ng/L concentrations and therefore does not correlate with the overall cytotoxicity. Cytotoxicity of extracts of all POC1S in the yeast assay can be related to sesquestered pollutants, especially antibiotics and other pharmaceuticals determined by chemical analysis. There are few studies that have focused on effects of urban pollution on the overall toxicity of waters in municipal rivers. Toxicity determined by the Microtox assay was directly proportional to urban land cover in streams around six metropolitan areas in the USA (Bryant and Goodbred, 2009). Toxicity of river water sampled by SPMD in Microtox and Daphnia pulex test has been observed in the Neris River after flowing through the capital city of Lithuanina (Sabaliunas et al., 2000). This finding is consistent with the observation of greater toxicity of compounds sampled by SPMD from the Svitava River downstream of the metropolitan area compared to upstream of Brno observed in this study. Detected AhR-mediated potency in both SPMD and POC1S indicated contribution of both hydrophobic and polar compounds to the overall dioxin-like potential of samples. Similarly in river sediments, mass-balance calculations based on fractionation with subsequent quantification have suggested that PAHs can account for a considerable portion of the dioxin-like potency together with unidentified more polar AhR-active compounds (Hilscherova et al., 2001). Dioxin-like potency found in all extracts of SPMDs was probably linked with the presence of known hydrophobic AhR ligands, such as PAHs or PCBs. Although dioxin-like compounds are usually investigated in less polar matrices such as SPMD or sediments, some recent studies (Dagnino et al., 2010; Reungoat et al., 2010) confirmed AhR potency in water phase. Results of another study (Jarošova et al., 2012) reported dioxin-like potency of 0.05 to 0.39 ng B10TEQ/POC1S in headwaters with small local sources of pollution. In the current study, POC1S samples exhibited dioxin-like potency only at three sites, inside and downstream of the WWTP, which suggests that waste waters contain some hydrophilic dioxin-like compounds that are not completely removed during treatment. This result is in agreement with the dioxin-like potencies detected WWTPs influent and effluent waters. The data for waste water samples show dioxin-like potency specifically for the polar methanolic extracts and thus might not include influence of some hydrophobic pollutants. Efficiency of treatment by the WWTP determined from BlOTEQs of the waste water samples was not as great for chemicals with dioxin-like potency as in the case of elimination of cytotoxicity or hormone-like potencies. Efficiencies of treatment varied substantially throughout the year. Release of some particle-bound compounds during treatment and lesser efficiency of treatment related to greater persistence of some AhR-active compounds might have contributed to this difference. However, the absolute concentrations of B10TEQ were less than those observed in other studies eventhough only a limited number of papers report dioxin-like potency in the dissolved phase. For example, Dagnino et al. (2010) detected AhR potency (by the same method as we used) in influent and effluent of French municipal WWTPs with an activated sludge system supplemented with biofilter to be as great as 37 to 112 ng TCDD/L, and 2.8 to 11.6 ng TCDD/L, respectively. Efficiency of removal was approximately 90% and the authors concluded that removal of AhR potency in this type of WWTPs depends primarily on removal of suspended solids with which they are associated. Alternatively, Ma et al. (2005) did not find concentrations of B10TEQ that were greater than 14 pg TCDD/L in either influents or effluents from a pilot plant in a Beijing WWTP, China. The observation that xenoestrogens and xenoandrogens were detected in waste water and POC1S samples from the WWTP, but not in SPMDs, implies that polar compounds accounted for the estrogenic and androgenic potencies. Since feminization of fish downstream from WWTPs has been observed in rivers worldwide, estrogenic potential of different types of waters has been evaluated in multiple studies. Examples of estrogenic potencies detected by various in vitro assays documenting the comparability of our findings to the situation in other parts of the world are compiled in Table 3. Relatively great efficiency of removal of estrogenic potency in various WWTPs has been documented both by composite water sampling as well as POC1S sampling. The majority of municipal or domestic WWTPs have implemented at least physical and biological treatment techniques. Activated sludge processes, similar to those of WWTP investigated in this study, are the most widely used types of biological treatment processes worldwide. Most studies that have focused on WWTP of similar types to that studied here found the treatment efficiencies for estrogens ranging from >88 to >99% (Leusch et al., 2005; Murk et al., 2002), 90-95% (Korner et al., 2000; Murk et al., 2002) or greater than 95% (Tan et al., 2007), but other studies have reported lesser efficiencies (Cargouet et al., 2004). Efficiency of removal of estrogenic potency, as determined by the MVLN assay, in four mechanical-biological municipal or domestic WWTPs in Paris V.Jälovä et al. / Environment International 59 (2013) 372-383 381 Table 3 Examples of estrogenic activities in waste waters and surface waters as detected by various in vitro assays. Matrix EEQng/L Country In vitro assay3 Reference Wastewater influent 51-70 Germany E-Screen Korner et al. (2000) 17-23 Queensland, Australia E-Screen Leusch etal. (2005) 1.1-120 The Netherlands ER-CALUX, YES Murketal. (2002) 35-72 Japan YES Onda et al. (2002) 1-30 Sweden YES Svenson etal. (2003) 108-356 Queensland, Australia E-Screen Tan et al. (2007) 5.4-124 Czech Republic MVLN This study Wastewater effluent 6 Germany E-Screen Korner et al. (2000) <0.75 Queensland, Australia E-Screen Leusch etal. (2005) 0.03-16 The Netherlands ER-CALUX, YES Murketal. (2002) 4-25 Japan YES Onda et al. (2002) <0.1-15 Sweden YES Svenson etal. (2003) 0.6-6.2 Japan YES Nakada et al. (2004) 1.9-15 USA MVLN Snyder et al. (2001) 99%, but in most tested samples it was greater than 96%. In previous studies, concentrations of estrogen equivalents (EEQJ of river water upstream and downstream of several WWTPs, quantified by use of the yeast estrogen screen (YES), was significantly correlated with EEQ based on chemical analysis of steroidal estrogens for grab samples and POCIS (Vermeirssen et al., 2005). Also chemical and biological (E-Screen assay) analyses used to determine the concentrations of 15 endocrine disrupting compounds and estrogenicity in grab and passive samples from five municipal WWTPs showed good agreement (Tan et al., 2007). Alternatively, assessment of contamination of headwater streams from livestock farms documented that measured waterborne steroids accounted for some of the detected estrogenicity, but a considerable portion of estrogenicity could not be attributed to concentrations of identified estrogens (Matthiessen et al., 2006). Androgenic potency of waste water in bioassays was shown to decrease during progression through the WWTP (Michelini et al., 2005). Concentrations of AEQ and efficiencies of removal observed in our study are similar to those reported for three Swedish municipal WWTPs that used activated sludge systems, and had androgenic potencies in yeast androgen screen (YAS) in influents ranging from 30 to 75 AEQ ng/L (and 0.8-3 AEQ ng/L in effluents) with efficiencies of removal of 96-98% (Svenson and Allard, 2004). However, some studies detected androgenic potencies in waste water influents that were greater than those observed in our study (Kirk et al., 2002; Leusch et al., 2006). Androgenic potencies in effluents of some WWTPs were as great as hundreds of ng AEQ/L, but in other WWTPs effluents they were less than the limits of quantification (Blankvoort et al., 2005; Kirk et al., 2002; Leusch et al., 2006; Sousa et al., 2010). Efficiencies of removal of androgens ranged from 82 to more than 99% when activated sludge was included in treatment processes, but significantly less when only primary treatment or for example biological trickling filters were employed (Kirk et al., 2002; Leusch et al., 2006). This observation is consistent with efficiencies of removal determined in this study which were greater than 96% in all cases. Also results obtained with POCIS samples confirmed significant removal of compounds with estrogenic and androgenic potency. Our results document that the efficiency of removal of both estrogenic and androgenic potency of the Brno WWTP can be ranked among the most efficient clarification WWTPs that do not implement advanced treatment. However, the results reported here also show that the efficiency of treatment can vary especially for dioxin-like and cytotoxic compounds, and thus one timepoint sampling might not be sufficient for its determination. Results of this study provide unique information on the variability of cytotoxicity and specific potencies in waste waters during the whole year. Estrogenic potency seemed to be greater in the dryer summer season when there is less dilution than during winter when more precipitation results in greater runoff, but also greater dilution (Fig. 4). However, there was no clear trend for androgenic potencies. Lower temperatures in winter did not negatively influence removal of estrogenic potency by the WWTP, but it might have affected the breakdown of more persistent compounds causing the dioxin-like potency. The greatest cytotoxicity was observed during summer, which might be correlated with lesser dilution (Fig. 2), but with another peak in winter, when probably some other types of pollutants associated with more typical winter sources (such as combustion) might play more significant role. However, the dioxin-like potency did not vary as much as estrogenicity throughout the year, except for August when it was approximately 3-fold greater than during the rest of the year. This observation is probably due to less dilution in summer and possibly also some immediate pollution situation that can affect the samples collected during a single day. There is limited information on seasonal variability of specific potencies of contaminants in waste waters. A study conducted in the UK (Kirk et al., 2002) found that estrogenic and also androgenic potencies in influents and effluents were less in samples collected in months of rainy weather. The recombinant yeast assay was used to assess variability of estrogenic potencies in influent and effluent of Canadian municipal WWTP implementing an additional cleaning step of UV disinfection (Fernandez et al., 2008). Estrogenic potencies of composite samples of influent taken every week from September to December were not dependent on sampling season, while EEQ levels in final effluents were very high, exceeding 100 ng EEQ/L in September and ranging from about 50 to 80 ng EEQ/L from the end of October till the end of the campaign. Lower EEQ concentrations in effluent in autumn and winter compared to summer were seen also in our study, but the ranges of EEQ values were much lower than those reported by Fernandez et al. (2008). 382 V. Jalová et al. / Environment International 59 (2013) 372-383 Similar to the results of this study, small estrogenic potencies and/or concentrations of industrial estrogen mimics and natural estrogens were frequently detected in WWTP discharges, due to their incomplete removal by WWTPs (Table 3). However, even these concentrations have been shown to be effective in causing some biological effects. It has been demonstrated in a 7-year whole-lake experiment that long term exposure to estrogens (5-6 ng/L ethinyl estradiol) can affect sustainability of wild fish populations (Kidd et al., 2007). Moreover, a multigeneration study of Chinese rare minnows (Gobiocypris rams) demonstrated that reproduction of the Ft minnows was completely inhibited at the ethinyl estradiol concentration as low as 0.2 ng/L (Zha et al., 2008). These results suggest that even when efficiencies of removal of estrogen are as great as those observed in this study, risks to aquatic organisms can still occur due to the concentrations of estrogens that are constantly released from waste water effluents. The risk seems to be greatest in cases when the volume of effluent waters represents a greater proportion in relation to the receiving waters. Next to the estrogenic and androgenic potencies detected in POC1S and water from WWTP, there were also some antiestrogenic and antiandrogenic pollutants in passive samples from WWTP, which however were not detected in the influent and effluent water samples. This difference indicates that antiestrogenic and antiandrogenic potency is related probably to less polar compounds, which were not in sufficient concentrations included in the methanolic extract of waste water. Moreover, the antiestrogenic/antiandrogenic potencies in waste waters could be masked by relatively great cytotoxicity of the methanolic extracts. Furthermore, passive samples enable higher preconcentration of the compounds compared to the composite water samples and thus the antiestrogenic/antiandrogenic activity detected in passive samples might have been bellow the limit of detection for the water samples. The passive samples from rivers exhibited neither estrogenic nor androgenic potency, but rather antiestrogenic and antiandrogenic potential. The antiestrogenic potency was detected in extracts from passive samplers exposed upstream of the city. In the study by Garcia-Reyero et al. (2001) (anti)estrogenicity was detected by recombinant yeast assay in waste waters and all samples of river water. The lack of estrogenic potency in POC1S and SPMD from river water in the study reported here could be caused by the presence of sufficient concentrations of chemicals that have been shown to have antiestrogenic potency, including pesticides, such as linuron or atrazine (Orton et al., 2009). Antiandrogenic potency was detected at most sampling sites. Hydrophilic antiandrogenic compounds were found in POC1S at sampling sites upstream of the city, whereas antiandrogenic potency in SPMD associated with the more hydrophobic pollutants was detected namely in the WWTP and downstream of the WWTP. Multiple contaminants are known to be associated with antiandrogenic potency (Orton et al., 2009; Sohoni and Sumpter, 1998), including some pesticides, which were detected by chemical analysis (e.g. p,p'-DDE, diuron). 5. Conclusion This study revealed the presence of compounds with endocrine disruptive potency in both river water and WWTP influent and effluent. The results of year-round waste water assessment confirmed high treatment efficiency of the WWTP for cytotoxic compounds, xenoestrogens and xenoandrogens. There was significant seasonal variability of efficiency of treatment, especially of dioxin-like potencies. Despite its high efficiency WWTP had impact on the pollution with endocrine disruptive compounds. The approach employed enabled determination of contributions of the metropolitan urban area and the WWTP to contamination of the rivers. Concentrations of PAHs and most pollutants sampled by POC1S decreased as a function of distance downstream of the city. Passive sampling, along with in vitro bioassays and chemical analysis allowed determination of a broad spectrum of contaminants and specific biological potencies and revealed the pollution situation in this model region. More research should be performed in the future to better characterize passive sampler performance under complex exposure conditions in raw wastewaters. Acknowledgments This research was supported by CETOCOEN (CZ.1.05/2.1.00/ 01.0001) project granted by the European Union and administered by the Ministry of Education, Youth and Sports of the Czech Republic, and by the projects of the MSMT 2B06093 and ENV1SCREEN 2B08036. Prof. Giesy was supported by the program of 2012 "High Level Foreign Experts" (#GDW20123200120) funded by the State Administration of Foreign Experts Affairs, the P.R. China to Nanjing University and the Einstein Professor Program of the Chinese Academy of Sciences. He was also supported by the Canada Research Chair program, and an at large Chair Professorship at the Department of Biology and Chemistry and State Key Laboratory in Marine Pollution, City University of Hong Kong. References Alvarez DA Petty JD, Huckins JN,Jones-Lepp TL, Getting DT, Goddard JP, et al. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environ Toxicol Chem 2004;23:1640-8. Alvarez DA Stackelberg PE, Petty JD, Huckins JN, Furlong ET, Zaugg SD, et al. Comparison of a novel passive sampler to standard water-column sampling for organic contaminants associated with wastewater effluents entering a New Jersey stream. Chemosphere 2005;61:610-22. Blankvoort BMG, Rodenburg RJT, Murk AJ, Koeman JH, Schilt R Aarts J. Androgenic activity in surface water samples detected using the AR-LUX assay: indications for mixture effects. Environ Toxicol Pharmacol 2005;19:263-72. Booij K, Smedes F. An improved method for estimating in situ sampling rates of nonpo-lar passive samplers. Environ Sei Technol 2010;44:6789-94. Brněnské vodárny a kanalizace. Sewage water treatment at WWTP Brno (Odvádění a čištění odpadních vod/ČOV Brno—Modříce), http://www.bvk.cz/o-spolecnosti/ odvadeni-a-cisteni-odpadnich-vod/cov-brno-modrice/, 2010. [in Czech]. Bryant WL, Goodbred SL. The response of hydrophobic organics and potential toxicity in streams to urbanization of watersheds in six metropolitan areas of the United States. Environ Monit Assess 2009;157:419-47. Cargouet M, Perdiz D, Mouatassim-Souali A, Tamisier-Karolak S, Levi Y. Assessment of river contamination by estrogenic compounds in Paris area (France). Sei Total Environ 2004;324:55-66. Dagnino S, Gomez E, Picot B, Cavailles V, Casellas C, Balaguer P, et al. Estrogenic and AhR activities in dissolved phase and suspended solids from wastewater treatment plants. Sei Total Environ 2010;408:2608-15. Demirpence E, Duchesne MJ, Badia E, Gagne D, Pons M. MVLN cells—a bioluminescent MCF-7-derived cell-line to study the modulation of estrogenic activity. J Steroid Biochem Mol Biol 1993;46:355-64. Desbrow C, Routledge EJ, Brighty GC, Sumpter JP, Waldock M. Identification of estrogenic chemicals in STW effluent. 1. Chemical fractionation and in vitro biological screening. Environ Sei Technol 1998;32:1549-58. Ellis GS, Huckins JN, Rostad CE, Schmitt CJ, Petty JD, Maccarthy P. Evaluation of lipid-containing semipermeable-membrane devices for monitoring organochlo-rine contaminants in the Upper Mississippi River. Environ Toxicol Chem 1995;14: 1875-84. Fernandez MP, Buchanan ID, Ikonomou MG. Seasonal variability of the reduction in estrogenic activity at a municipal WWTP. Water Res 2008;42:3075-81. Freyberger A Schmuck G. Screening for estrogenicity and anti-estrogenicity: a critical evaluation of an MVLN cell-based transactivation assay. Toxicol Lett 2005;155: 1-13. Garcia-Reyero N, Grau E, Castillo M, De Aida MJL, Barcelo D, Pina B. Monitoring of endocrine disrupters in surface waters by the yeast recombinant assay. Environ Toxicol Chem 2001;20:1152-8. Garcia-Reyero N, Raldua D, Quiros L, Llaveria G, Cerda J, Barcelo D, et al. Use of vitellogenin mRNA as a biomarker for endocrine disruption in feral and cultured fish. Anal Bioanal Chem 2004;378:670-5. Giesy JP, Pierens SL, Snyder EM, Miles-Richardson S, Kramer VJ, Snyder SA et al. Effects of 4-nonylphenol on fecundity and biomarkers of estrogenicity in fathead minnows [Pimephales promelas). Environ Toxicol Chem 2000;19:1368-77. Grabic R Jurčíkova J, Tomsejova S, Ocelka T, Halířová J, Hypr D, et al. Passive sampling methods for monitoring endocrine disrupters in the Svratka and Svitava Rivers in the Czech Republic. Environ Toxicol Chem 2010;29:550-5. Harries JE, Sheahan DA, Jobling S, Matthiessen P, Neall P, Routledge EJ, et al. A survey of estrogenic activity in United Kingdom inland waters. Environ Toxicol Chem 1996;15:1993-2002. Hilscherova K, Kannan K, Kang YS, Holoubek I, Machala M, Masunaga S, et al. Characterization of dioxin-like activity of sediments from a Czech river basin. Environ Toxicol Chem 2001;20:2768-77. V. jalová et al. I Environment International 59 (2013) 372-383 383 Hilscherova K, Kannan K, Holoubek I, Giesy JP. Characterization of estrogenic activity of riverine sediments from the Czech Republic Arch Environ Contam Toxicol 2002;43: 175-85. Huckins JN, Tubergen MW, Manuweera GK. Semipermeable-membrane devices containing model lipid—a new approach to monitoring the bioavailability of lipophilic contaminants and estimating their bioconcentration potential. Chemosphere 1990;20:533-52. Huckins JN, Manuweera GK, Petty JD, Mackay D, Lebo JA Lipid-containing semipermeable-membrane devices for monitoring organic contaminants in water. Environ SciTechnol 1993;27:2489-96. Huckins JN, Booij K, Petty JD. Theory and modeling. In: Huckins JN, Booij K, Petty JD, editors. Monitors of organic chemicals in the environment. Semipermeable membrane devices. New York: Springer; 2006. p. 45-85. Jarosova B, Blaha L, Vrana B, Randak T, Grabic R, Giesy JP, et al. Changes in concentrations of hydrophilic organic contaminants and of endocrine-disrupting potential downstream of small communities located adjacent to headwaters. Environ Int 2012;45:22-31. Jobling S, Nolan M, Tyler CR Brighty G, Sumpter JP. Widespread sexual disruption in wild fish. Environ Sei Technol 1998;32:2498-506. Kidd KA, Blanchfield PJ, Mills KH, Palace VP, Evans RE, Lazorchak JM, et al. Collapse of a fish population after exposure to a synthetic estrogen. Proc Natl Acad Sei U S A 2007;104:8897-901. Kirk LA, Tyler CR Lye CM, Sumpter JP. Changes in estrogenic and androgenic activities at different stages of treatment in wastewater treatment works. Environ Toxicol Chem 2002;21:972-9. Korner W, Bolz U, Sussmuth W, Hiller G, Schuller W, Hanf V, et al. Input/output balance of estrogenic active compounds in a major municipal sewage plant in Germany. Chemosphere 2000;40:1131-42. Leskinen P, Michelini E, Picard D, Karp M, Virta M. Bioluminescent yeast assays for detecting estrogenic and androgenic activity in different matrices. Chemosphere 2005;61:259-66. Leusch FDL, Chapman HF, Korner W, Gooneratne SR, Tremblay 1A Efficacy of an advanced sewage treatment plant in southeast Queensland, Australia, to remove estrogenic chemicals. Environ Sei Technol 2005;39:5781-6. Leusch FDL, Chapman HF, van den Heuvel MR Tan BLL, Gooneratne SR, Tremblay 1A Bioassay-derived androgenic and estrogenic activity in municipal sewage in Australia and New Zealand. Ecotoxicol Environ Saf 2006;65:403-11. Leusch FDL, De Jager C, Levi Y, Lim R Puijker L, Sacher F, et al. Comparison of five in vitro bioassays to measure estrogenic activity in environmental waters. Environ Sei Technol 2010;44:3853-60. Li H, Vermeirssen EL, Helm PA, Metcalfe CD. Controlled field evaluation of water flow rate effects on sampling polar organic compounds using polar organic chemical integrative samplers. Environ Toxicol Chem 2010;29:2461-9. Lohmann R, Booij K, Smedes F, Vrana B. Use of passive sampling devices for monitoring and compliance checking of POP concentrations in water. Environ Sei Pollut Res 2012;19:1885-95. Ma M, Li J, Wang ZJ. Assessing the detoxication efficiencies of wastewater treatment processes using a battery of bioassays/biomarkers. Arch Environ Contam Toxicol 2005;49:480-7. Matsui S, Takigami H, Matsuda T, Taniguchi N, Adachi J, Kawami H, et al. Estrogen and estrogen mimics contamination in water and the role of sewage treatment. Water Sei Technol 2000;42:173-9. Matthiessen P, Arnold D, Johnson AC, Pepper TJ, Porringer TG, Pulman KGT. Contamination of headwater streams in the United Kingdom by oestrogenic hormones from livestock farms. Sei Total Environ 2006;367:616-30. Mayer P, Tolls J, Hermens L, Mackay D. Equilibrium sampling devices. Environ Sei Technol 2003;37:184A-91A Michelini E, Leskinen P, Virta M, Karp M, Roda A A new recombinant cell-based biolumi-nescent assay for sensitive androgen-like compound detection. Biosens Bioelectron 2005;20:2261-7. Miles-Richardson SR Kramer VJ, Fitzgerald SD, Render JA Yamini B, Barbee SJ, et al. Effects of waterbome exposure of 17 beta-estradiol on secondary sex characteristics and gonads of fathead minnows (Pimep/iafespromefas). Aquat Toxicol 1999;47:129-45. Ministry of the Environment of the Czech Republic, Czech Environmental Inspectorate. Report on the reconstruction and modernization of the VWVTP in Brno, (in Czech) http://www.cizp.cz/(blobdbr4uyzcvq454qlmucvb)/default.aspx?id=511&ido= 362&sh=-711127208, 2010. Murk AJ, Legier J, van Lipzig MMH, Meerman JHN, Belfroid AC, Spenkelink A, et al. Detection of estrogenic potency in wastewater and surface water with three in vitro bioassays. Environ Toxicol Chem 2002;21:16-23. Nadzialek S, Vanparys C, Van der Heiden E, Michaux C, Brase F, Scippo ML, et al. Understanding the gap between the estrogenicity of an effluent and its real impact into the wild. Sei Total Environ 2010;408:812-21. Nakada N, Nyunoya H, Nakamura M, Hara A Iguchi T, Takada H. Identification of estrogenic compounds in wastewater effluent Environ Toxicol Chem 2004;23:2807-15. Oh SM, Kim HR, Park HK, Choi K, Ryu J, Shin HS, et al. Identification of estrogen-like effects and biologically active compounds in river water using bioassays and chemical analysis. Sei Total Environ 2009;407:5787-94. Onda K, Yang SY, Miya A, Tanaka T. Evaluation of estrogen-like activity on sewage treatment processes using recombinant yeast. Water Sei Technol 2002;46:367-73. Orton F, Lutz I, Kloas W, Routledge EJ. Endocrine disrupting effects of herbicides and pentachlorophenol: In vitro and in vivo evidence. Environ Sci Technol 2009;43: 2144-50. Petty JD, Orazio CE, Huckins JN, Gale RW, Lebo JA, Meadows JC, et al. Considerations involved with the use of semipermeable membrane devices for monitoring environmental contaminants. J Chromatogr A 2000a;879:83-95. Petty JD, Jones SB, Huckins JN, Cranor WL, Parris JT, McTague TB, et al. An approach for assessment of water quality using semipermeable membrane devices (SPMDs) and bioindicator tests. Chemosphere 2000b;41:311-21. Petty JD, Huckins JN, Alvarez DA, Brumbaugh WG, Cranor WL, Gale RW, et al. A holistic passive integrative sampling approach for assessing the presence and potential impacts of waterborne environmental contaminants. Chemosphere 2004;54:695-705. Reungoat J, Macova M, Escher BI, Carswell S, Mueller JF, Keller J. Removal of micropollutants and reduction of biological activity in a full scale reclamation plant using ozonation and activated carbon filtration. Water Res 2010;44:625-37. Routledge EJ, Sheahan D, Desbrow C, Brighty GC, Waldock M, Sumpter JP. Identification of estrogenic chemicals in STW effluent 2. In vivo responses in trout and roach. Environ Sci Technol 1998;32:1559-65. Sabaliunas D, Lazutka JR Sabaliuniene I. Acute toxicity and genotoxicity of aquatic hydrophobic pollutants sampled with semipermeable membrane devices. Environ Pollut 2000;109:251-65. Sanderson T, van den Berg M. Interactions of xenobiotics with the steroid hormone biosynthesis pathway. Pure Appl Chem 2003;75:1957-71. Sanderson JT, Aarts J, Brouwer A, Froese KL, Denison MS, Giesy JP. Comparison of Ah receptor-mediated luciferase and ethoxyresorufin-O-deethylase induction in H4IIE cells: Implications for their use as bioanalytical tools for the detection of polyhalogenated aromatic hydrocarbons. Toxicol Appl Pharmacol 1996;137:316-25. Snyder SA, Snyder E, Villeneuve D, Kurunthachalam K, Villalobos A, Blankenship A Giesy J. Instrumental and bioanalytical measures of endocrine disrupters in water. Analysis of Environmental Endocrine Disruptors; 2000. p. 73-95. Snyder SA, Villeneuve DL, Snyder EM, Giesy JP. Identification and quantification of estrogen receptor agonists in wastewater effluents. Environ Sci Technol 2001 ;35: 3620-5. Snyder EM, Snyder SA Kelly KL, Gross TS, Villeneuve DL, Fitzgerald SD, et al. Reproductive responses of common carp (Cyprimis carpio) exposed in cages to influent of the Las Vegas Wash in Lake Mead, Nevada, from late winter to early spring. Environ Sci Technol 2004;38:6385-95. Sohoni P, Sumpter JP. Several environmental oestrogens are also anti-androgens. J Endocrinol 1998;158:327-39. Sole M, de Alda MJL, Castillo M, Porte C, Ladegaard-Pedersen K, Barcelo D. Estrogenicity determination in sewage treatment plants and surface waters from the Catalonian area (NE Spain). Environ Sci Technol 2000;34:5076-83. Sousa A Schonenberger R Jonkers N, Suter MJF, Tanabe S, Barroso CM. Chemical and biological characterization of estrogenicity in effluents from VWVTPs in Ria de Aveiro (NW Portugal). Arch Environ Contam Toxicol 2010;58:1-8. Stuer-Lauridsen F. Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environ Pollut 2005;136:503-24. Sumpter JP. Feminized responses in fish to environmental estrogens. Toxicol Lett 1995;82-3: 737-42. Svenson A, Allard AS. Occurrence and some properties of the androgenic activity in municipal sewage effluents. J Environ Sci Health A Tox Hazard Subst Environ Eng 2004;39:693-701. Svenson A Allard AS, Ek M. Removal of estrogenicity in Swedish municipal sewage treatment plants. Water Res 2003;37:4433-43. Tan BLL, Hawker DW, Muller JF, Leusch FDL Tremblay LA, Chapman HF. Comprehensive study of endocrine disrupting compounds using grab and passive sampling at selected wastewater treatment plants in South East Queensland, Australia. Environ Int 2007;33:654-69. Vega-Lopez A Ramon-Gallegos E, Galar-Martinez M, Jimenez-Orozco FA, Garcia-Latorre E, Dominguez-Lopez ML. Estrogenic, anti-estrogenic and cytotoxic effects elicited by water from the type localities of the endangered goodeid fish Cirardinichthys viviparus. Comp Biochem Physiol C Toxicol Pharmacol 2007:145:394^03. Vermeirssen ELM, Korner O, Schonenberger R Suter MJF, Burkhardt-Holm P. Characterization of environmental estrogens in river water using a three pronged approach: active and passive water sampling and the analysis of accumulated estrogens in the bile of caged fish. Environ Sci Technol 2005;39:8191-8. Villeneuve DL, Blankenship AL, Giesy JP. Derivation and application of relative potency estimates based on in vitro bioassay results. Environ Toxicol Chem 2000;19: 2835-43. Villeneuve DL, Khim JS, Kannan K, Giesy JP. Relative potencies of individual polycyclic aromatic hydrocarbons to induce dioxinlike and estrogenic responses in three cell lines. Environ Toxicol 2002;17:128-37. Wilson VS, Bobseine K, Lambright CR Gray LE. A novel cell line, MDA-kb2, that stably expresses an androgen- and glucocortkoid-responsive reporter for the detection of hormone receptor agonists and antagonists. Toxicol Sci 2002;66:69-81. Zha JM, Sun LW, Zhou YQ, Spear PA, Ma M, Wang ZJ. Assessment of 17 alpha-ethinylestradiol effects and underlying mechanisms in a continuous, multigeneration exposure of the Chinese rare minnow (Cobiocypris rams). Toxicol Appl Pharmacol 2008;226:298-308. Príloha 24 Vrana B., Klučárová V., Benická E., Abou-Mrad N., Amdany R., Horáková S., Draxler A., Humer F., and Gans O., Passive sampling: An effective method for monitoring seasonal and spatial variability of dissolved hydrophobic organic contaminants and metals in the Danube river, Environ. Pollut, 2014,184,101-112. ENVIRONMENTAL POLLUTION 1 Environmental Pollution 184 (2014) 101-112 i ELSEVIER Contents lists available at ScienceDirect Environmental Pollution journal homepage: www.elsevier.com/locate/envpol Passive sampling: An effective method for monitoring seasonal and flVrossMark spatial variability of dissolved hydrophobic organic contaminants and metals in the Danube river Branislav Vrana a'bi *, Veronika Klučárová c, Eva Benická c, Ninette Abou-Mrad b, Robert Amdany d, Soňa Horáková a, Astrid Draxler e, Franko Humer e, Oliver Gans e a Water Research Institute, Nabr. Arm. Gen. L. Svobodu 5, 812 49 Bratislava, Slovakia b Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment RECETOX, Kamenice 753/5, 625 00 Brno, Czech Republic c Slovak University of Technology, Faculty of Chemical and Food Technology, Radlinského 9, 812 37 Bratislava, Slovakia A University of Witwatersrand, Department of Chemistry, P/Bag 3, Johannesburg, 2050, South Africa e Umweltbundesamt Environment Agency Austria, Spittelauer Lände 5, 1090 Vienna, Austria article info abstract Article history: Received 11 March 2013 Received in revised form 11 July 2013 Accepted 23 August 2013 Keywords: Danube Free dissolved concentration Persistent organic pollutants Metals Passive sampling Application of passive samplers is demonstrated for assessment of temporal and spatial trends of dissolved polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and priority metals in the middle stretch of the Danube river. Free dissolved concentrations of PAHs, measured using SPMD samplers, ranged from 5 to 72 ng IT1. Dissolved PCBs in water were very low and they ranged from 5 to 16 pg IT1. Concentration of mercury, cadmium, lead and nickel, measured using DGT samplers, were relatively constant along the monitored Danube stretch and in the range <0.1, <1—20, 18—74, and 173 —544 ng IT1, respectively. Concentrations of PAHs decreased with increasing temperature, which reflects the seasonality in emissions to water. This has an implication for the design of future monitoring programs aimed at assessment of long term trends. For such analysis time series should be constructed of data from samples collected always in the same season of the year. © 2013 Elsevier Ltd. All rights reserved. 1. Introduction In December 2000 the European Union adopted the Water Framework Directive (WFD) to secure water resources for future generations (EU, 2000). In the implementation process of the WFD, all EU member states are required to perform trend monitoring on several pollutants priority substances in surface water that tend to accumulate in sediment and/or biota in surface water (EU, 2008). Long term measurements in water provide important information that can be used in evaluation of effects of accepted measures on lowering the emissions. Such a trend monitoring can be carried out in water, suspended particles and sediments as well as in biota. The decision, which matrix to survey is difficult especially for compounds present in water at very low concentrations, such as heavy metals and hydrophobic organic pollutants like polycyclic aromatic hydrocarbons (PAH) or polychlorinated biphenyls (PCBs). Among * Corresponding author. E-mail addresses: branovrana@googlemail.com, vrana@recetox.muni.cz (B. Vrana). 0269-7491/$ - see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.Org/10.1016/j.envpol.2013.08.018 other available monitoring methods passive sampling presents a promising approach because it provides sensitive and time integrative measurement of free dissolved concentrations of contaminants in water (Greenwood et al„ 2007). Diffusion of organic pollutants from sampled media to the sampler is driven by the high affinity of analysed compounds to the sorbent material in the sampler. The concentration found in a passive sampler can be used for calculation of time weighted average (TWA) water concentration over extended periods of time. The major advantage of passive samplers over alternative matrices used for trend monitoring, e.g. sediments or biota, is that passive samplers constitute a well-defined sampling medium with a known uptake capacity. In contrast to results based on sediment or biota, passive sampling data require no corrections for organic carbon, lipid content or species to compare data on a temporal or spatial scale. Free dissolved concentration is a measure of organism exposure in water and passive sampling allows measurement even for compounds that cannot be measured in biota because of their excretion or metabolism by organisms. Furthermore, different sources of variance including analytical and environmental variance can be much better controlled, which in turn results in reduction of the required 102 B. Vrana et al. / Environmental Pollution 184 (2014) 101-112 number of analysed samples to obtain results with comparable statistical power (Lohmann et al., 2012). Another advantage of use of passive samplers is the determination of free dissolved concentration in water, which is one of the important parameters for the assessment of pollutant bioavailability and fate in the aquatic environment. The freely dissolved concentration of contaminants in the water column is directly proportional to their fugacity in the water phase (Mayer et al., 2003). Pollution monitoring based on direct water measurement of dissolved concentrations of hydrophobic organic compounds by bottle sampling is not reliable, since the individual spot samples of water collected at the sampling sites reflect only the pollution situation at the moment of sampling. Measurement of truly dissolved concentration of these compounds in water cannot be easily achieved by conventional liquid/liquid or solid phase extraction techniques because of potential bias of these methods introduced by co-extraction of analytes bound to colloids present in water samples. In this study, passive samplers were applied to characterize the temporal and spatial variability of dissolved heavy metals, PAHs and PCBs in the Danube river between the cities of Vienna and Bratislava (Fig. 1). This paper presents particular results of a larger study aimed at comparison of the most promising available monitoring methods (bottom sediments, suspended particulate samplers and passive samplers) for those pollutants (PAHs, selected heavy metals) to give a technical recommendation on how to perform a trend monitoring in the aquatic environment (www. umweltbundesamt.at/umweltsituation/hestia_home). This comparison will provide the basis for a technical recommendation on how to implement the WFD as well as for a future national and regional cooperation in monitoring and consistent evaluation of the quality of the water body. 2. Materials and methods 2.1. Chemicals Organic solvents: acetone (Mikrochem, Slovakia), n-hexane SupraSolv (Merck, Germany), dichloromethane SupraSolv (Merck, Germany), hydrochloric acid 36%, p.a. (Merck, Germany), Triolein (Sigma Aldrich, Belgium), silicagel 60 (Merck, Germany). Gases for GC—MS/ECD equipment: nitrogen ECD and helium 6.0 (both Messer Tatragas, Slovakia). Etalons of 16 polycyclic aromatic hydrocarbons for calibration of equipment (PAH mix 9,100 pg mL-1 in cyclohexane), 6 polychlorinated biphenyls (10 pg mL-1 in cyclohexane), perdeuterated polycyclic aromatic hydrocarbons applied as performance reference compounds (Dio-acenaphthene, D10-fluorene, D10-phenanthrene, D12-chrysene, D12-benzo(e)pyrene), surrogates (D8-naphthalene, Dio-anthracene, Dio-pyrene, Di2-benzo(a)anthracene, D^-ben-zo(k)fluoranthene, Di2-benzo(a)pyrene Di2-benzo(g,h,i)perylene), PCB30 and PCB185 were purchased from Dr. Ehrenstorfer, Germany. Terphenyl and PCB 121, the internal standards for instrumental analysis by GC/MS were purchased by Sigma— Aldrich, Germany. Physicochemical properties of analytes are given in Supplementary Information. 2.2. Passive samplers 2.2.1 SPMDs The SPMDs consisting of an LDPE membrane filled with 1 mL of triolein (95% purity), in nominal dimensions 2.54 x 91.4 cm (exposure surface area 460 cm2), wall thickness of 75—90 |im were purchased from (Exposmeter, Sweden). Samplers contained 2 |ig/sampler of individual performance reference compounds (PRCs; Dio-Acenaphthene, D10-Fluorene, D10-Phenanthrene, D12-Chrysene, D12-Benzo(e)pyr-ene). Before use they were stored in gas tight metal containers at -20 °C The volume of sampler (triolein + membrane) is 4.95 mL 2.2.2. DGTs DGT (diffusive gradients in thin film samplers) samplers were purchased by DGT Research Ltd, Lancaster, UK. Two versions of the sampler were applied: one for sampling mercury ions and another version for sampling heavy metals nickel, cadmium and lead. The sampler is composed of a plastic body, which contains a pre-filter with a surface area A = 3.14 cm2, diffusive hydrogel (0.8 mm thick) and adsorptive resin-gel (0.16 mL volume) layers. 2.3. Sampling sites 2.3.1. Altenwdrth an der Donau Altenworth on the Danube represents the location upstream of the Vienna area. The actual sampling site was located at the bridge on the left bank Danube river side arm in Altenworth, approximately at the river kilometre (rkm) 1980, cca 55 km upstream Vienna agglomeration. This sampling site was not located directly in the main stream of the Danube, since the installation of the sampling equipment would have been logistically very difficult in the area of the adjacent Danube power plant. The sampled surface water is not affected by the backwater area of the Danube dam that is located downstream. The water level gradient at the bridge provides suitable conditions for operation of suspended sediment traps that were deployed simultaneously with passive samplers. The fast water current at the bridge enabled to achieve elevated sampling rates with SPMDs and thus to accumulate higher amounts of analytes. 2.3.2. Langenzersdorf The site Langenzersdorf is located at the weir 2 of the Marchfeld channel just upstream the main Vienna city agglomeration. The artificially constructed channel represents an important source of irrigation water for vegetable farmers of the Marchfeld area between the rivers March/Morava and the Danube. The site is located 1 km downstream the intake structure of Marchfeld channel from the left Fig. 1. Map of the sampling sites in the Danube river in Austria and Slovakia. Site symbols are given next to the site location names. B. Vrana et al. / Environmental Pollution 184 (2014) 101-112 103 bank of the Danube at the rkm 1938. It is assumed that the flow velocity of Danube is slightly affected by the backwater of the Freudenau dam that is located 17 km downstream, but this should have only a minimum effect on water quality. 2.3.3. Wolfsthal The sampling was performed at the Wolfsthal on-line monitoring station at Hainburg/Donau on the right Danube bank, approx. 15 km upstream Bratislava at the rkm 1879. The water from the Danube for passive sampler exposure was pumped into the monitoring station using submersible pumps operating at 1000—2000 L hr1 that were installed in the main stream of Danube. During the sampling campaign performed in 2010, passive sampling was simultaneous with other alternative sampling methods including continuous collection of water samples and suspended particulate matter. For the purpose of this sampling campaign non-filtered water from the Danube was evenly distributed to particular sampling devices that included automatic water sampler, suspended particulate matter sampler and passive samplers, respectively. Data comparing various sampling techniques will be reported separately. The station facility was adapted to perform passive sampling as described in the sampling campaign description below. 2.3.4. Cunovo The sampling site is located 15 km downstream the city of Bratislava at the rkm 1836 on the dam at the right bank in Čunovo. The Čunovo dam is a part of the Danube dam system Gabčíkovo and its basic function is to ensure the flow into the old Danube riverbed in the agreement between Slovakia and Hungary. The sampling was performed at the water intake object of the hydroelectric power plant in the Čunovo dam (http://www.gabcikovo.gov.sk/svdgn/stup_Cun.htm). In 2011, after nearly 20 years since the completion of the dam system, sediment dredging activities started in the reservoir Hrušov that is located just upstream the Čunovo dam. There are about 2 million cubic metres of sediment that must be removed in the coming years. These dredging activities can, potentially mobilize also pollutants that are assessed in the present study. During the sampling campaign the dredging project was performed at the right bank, and the flow of sampled water passed the area of dredging activities. 2.4. Sampling campaigns 2.4.1. 2010 In 2010 passive sampling was performed at a single monitoring site at the online monitoring station in Wolfsthal. Continuous sampling in 14-days passive sampler exposure periods started in July and ended in December 2010 with a single interruption from 14th September to 5th October for station maintenance. During each of the ten 14-day exposures three samplers of each type (SPMD, DGT) were deployed in parallel. For deployment samplers were mounted using a stainless steel wire holder inside aim high glass cylinder with 5 cm inner diameter. The flow of Danube river water through the cylinder was kept constant at 140 L hr1 for the entire 14-day exposure period. The dates of deployment periods are given in Table 1. In parallel with passive sampling, ten composite samples of whole water, representative for each of the 14-day sampler deployment period, were collected using an automatic water sampler installed in the monitoring station. Details are given in Supplementary Information. Concentrations of PAHs and heavy metals were determined in these composite samples. 2.4.2. 2011 The monitoring included 4 season sampling campaigns, in the months of February, April, July, and October 2011. The dates of deployment periods are given in Table 2. During each campaign 3 samplers in parallel were deployed at each site during 14 days cycles. For deployment at the online monitoring station Wolfsthal samplers were mounted in a flow-through column as described for the 2010 campaign. At the Table 2 Description of sampling sites in the study area. No. Sampling site Symbol Water River Longitude Latitude body kilometre 1. Altenwörth AL Danube 1980 15°51'54" 48°22'44" 2. Langenzersdorf L Danube 1938 16°21'22" 48°17'35" 3. Wolfsthal w Danube 1879 16°59'15" 48°09'52" 4. Čunovo c Danube 1836 17°13'29" 48°01'49" remaining three sites samplers were placed into protective cage made of perforated stainless steel plate, preventing their mechanical damage and deployed in the river approximately 1 m below the water level with the help of ropes, buoys and anchors. After 14 days of exposure the samplers were collected, inspected for mechanical damage and the biofilm formation, photographed, transported to the laboratory in the protection package in a portable cool box. The prevention of contact of SPMDs with plastic materials and other potential sources of contamination were ensured. An additional field control sampler was exposed to air while samplers were being deployed and collected. The field control was processed as the deployed samplers and was used to measure contamination during transportation and handling. Three sampler fabrication controls were also analysed to determine contamination arising from the manufacturing process, sampler components, laboratory storage, processing and analytical procedures, but also to determine the initial concentration of PRCs in the SPMD samplers before exposure (Huckins et al., 2002; Booij et al., 2007). Several samplers were not retrieved due to loss of samplers during field exposure, namely by vandalism at site Langen-zersdorf in April, and by sampler cage tear off at site Altenworth in July. The SPMD samplers and their extracts were stored at separate place from chemicals, in a freezer under the temperature -20 °C SPMD samplers were analyzed for hydrophobic organic pollutants PAHs and PCBs. DGT samplers were stored at 4 °C until processing and analysed for priority pollutant heavy metals nickel, cadmium, lead and mercury. 2.5. Sample extraction and analysis 2.5.1. SPMDs SPMD samplers were cleaned from debris and mud and analytes were extracted two times 24 h by dialysis to hexane. Dialysates were further cleansed by gel permeation chromatography and silica gel or sulphuric acid modified silica gel for PAH and PCB analysis, respectively. The analysis of PAHs was performed using 6890N GC (Agilent, USA) equipped with a 30 m x 0.25 mm x 0.25 |im HP5-MS column (Agilent, USA) coupled to 5972 MS operated in electron impact ionization mode. PCB analysis was performed using GC—MS/MS 6890N GC (Agilent, USA) equipped with a 60 m x 0.25 mm x 0.25 |im DB5-MS column (Agilent J&W, USA) coupled to Ouattro Micro GC MS MS (Waters, Micromass, UK) operated in EI+ ionization mode. Details of sample processing and instrumental analysis are given in Supplementary Information. 2.5.2. DGTs Heavy metals accumulated in the DGT sampler adsorption resin were extracted with 1 mL of 1 mol L_1 HN03 solution for 24 h. The determination of heavy metals nickel and lead in extracts was performed according to ISO 15586:2003, whereas cadmium was analysed according to DIN 38406/19. The analysis proceeded by atomic absorption spectrometry with graphite furnace (ET-AAS). Mercury analysis was performed by a microwave digestion with hno3 and h2o2 and an amalgam enrichment and reduction with sodium borohydride, followed by analysis of mercury by cold vapour atomic absorption spectrometry. Table 1 Description of the sampling campaign at the site Wolfsthal in the Danube in 2010. Exposure nr. Exposure period Exposure Water Mean discharge SPMD-sampling Start End (days) temperature (°C) (m3 s-1)11 ratefisOd-1) I 06.07.2010 20.07.2010 14 21 2176 16.4 II 20.07.2010 03.08.2010 14 20 2766 11.4 III 03.08.2010 17.08.2010 14 18 3371 6.0 IV 17.08.2010 31.08.2010 14 18 2417 6.6 V 31.08.2010 14.09.2010 14 15 2918 7.2 VI 05.10.2010 19.10.2010 14 12 1428 4.1 VII 19.10.2010 02.11.2010 14 10 1414 3.7 VIII 02.11.2010 16.11.2010 14 10 1364 3.4 IX 16.11.2010 30.11.2010 14 8 1469 3.7 X 30.11.2010 14.12.2010 14 4 1897 2.0 a Rs is the equivalent water volume extracted by SPMD per day for a compound with a medium molecular weight (Mw =178; phenanthrene). b Calculated from volume discharge data available for the monitoring station in Bratislava. 104 B. Vrana et al. / Environmental Pollution 184 (2014) 101-112 Table 3 Description of the sampling campaign in the Danube in 2011. Campaign Nr. and Exposure period Exposure SPMD-sampling Water Mean discharge sampling site Start End (days) rate Ks(L d"1)3 temperature (°C) (m3 s-1)11 I Al Altenworth 16.02. 02.03. 14 15.8 4 1496 I L Langenzersdorf 16.02. 02.03. 14 11.5 3 IW Wolfsthal 03.03. 17.03. 14 4.4 3 I C Cunovo 03.03. 17.03. 14 12.9 3 II Al Altenworth 14.04. 28.04. 14 16.3 12 1346 II L Langenzersdorf 14.04. 28.04. 14 NAC 13 II W Wolfsthal 14.04. 28.04. 14 4.14 13 II C Cunovo 14.04. 28.04. 14 18.6 13 III Al Altenworth 22.06. 7.07. 14 NAC 19 2063 III L Langenzersdorf 22.06. 07.07. 14 20.9 19 III Wolfsthal 22.06. 07.07. 14 12.0 19 III C Cunovo 22.06. 07.07. 14 14.8 19 IV Al Altenworth 13.10. 27.10. 14 26.2 12 2021 IV L Langenzersdorf 13.10. 27.10. 14 17.6 11 IV W Wolfsthal 13.10. 27.10. 14 3.4 11 IV C Cunovo 13.10. 27.10. 14 19.7 11 a Rs is the equivalent water volume extracted by SPMD per day for a compound with a medium molecular weight (Mw = 178; phenanthrene). b Calculated from volume discharge data available for the monitoring station in Bratislava. c NA-not available because of SPMD sampler loss. 2.6. Calculation of dissolved water concentrations from passive sampler data 2.6.1 SPMDs Dissolved water concentrations of target analytes were calculated from amounts accumulated in SPMDs as follows. Amounts of analytes absorbed by the samplers follow a first-order approach to equilibrium. Aqueous concentrations were calculated from the mass (JVs) absorbed by the SPMD, the in situ sampling rate of the compounds Rs and their sampler—water partition coefficients JCsw: C„ = ■ JVc Vc/ŕc, (1) where Vs is the volume of the SPMD (4.95 mL) and t is the sampler exposure time. PRC dissipation also follows first-order kinetics. Sampling rates Rs were estimated from dissipation of PRCs from SPMDs during exposure using nonlinear least squares method by Booij and Smedes (2010), considering the fraction/of individual PRCs (D10-acenaphthene, D10-fluorene, D10-phenanthrene and D10-chrysene) that remain in the SPMD after the 14-day exposure as a continuous function of their JCsw, with Rs as an adjustable parameter. / = exp KswVs (2) Here,/= JVpRc/No.PRci No,prc = initial amount of the PRC at t = 0, JVPRC _ amount of each PRC remaining after exposure, and t is exposure period (14 days). Assuming water boundary layer controlled uptake, Rs of individual target compounds in the higher hydrophobicity range was estimated by substituting Eq. (3) derived by Rusina et al. (2010) into Eq. (2). Kc = FAM- (3) Here M is the molecular weight of the analyte, A is the surface area of SPMD (460 cm2). The factor F represents the effects of environmental conditions (temperature, flow, biofouling). It was obtained as an optimized value of adjustable parameter using nonlinear least squares method for estimating sampling rates (Booij and Smedes, 2010) after substitution of Rs in Eq. (2) by Eq. (3). The necessary JCsw values were intrapolated from the empirical equation (Huckins et al., 2006) log/Csw = -0.1618(log/Cow)2 + 2.321 log/Cow - 2.61 (4) Booij et al. (2003a) observed that SPMD-water partition coefficients J(Sw did not significantly change with temperature in the range from 2 °C to 30 °C, thus, for our calculations partition coefficients were not corrected for effect of temperature. 2.6.2. DGTs Dissolved water concentrations Cdgt of metals were calculated from their masses accumulated in DGTs (AT) according to Warnken et al. (2007). JVAg DtA (5) where Ag is the thickness of the diffusion gel layer, t is exposure time, A is the sampler surface area and D is the temperature dependent diffusion coefficient of a metal ion. Applied values of D were taken from the DGT manufacturer (www. dgtresearch.com). It is assumed that mass transfer of metal species into DGT is controlled by diffusion in the gel layer. Thus, sampling by DGT should not be affected by the flow velocity/turbulence, as is the case for SPMDs. 2.7. Assessment of PAH patterns using principal component analysis Principal component analysis (PCA) was used to compare the PAH levels and patterns in the dissolved phase, which was monitored at four sampling sites during four seasons in the 2011 sampling campaign. PCA analysis was based on absolute analyte concentrations and data were modelled according to the procedure described by Vrana et al. (2001). 3. Results and discussion 3.1. Aspects of sampling with SPMDs The repeatability within three parallel determinations of PAH concentrations represented by mean relative standard deviation 10.0 a: 8.5 3.35E-03 3.40E-03 3.45E-03 3.50E-03 1/T [K"1] 3.55E-03 3.60E-03 3.65E-03 Fig. 2. Effect of water temperature on SPMD sampling rate of phenanthrene obtained during individual 14-day exposures at the site Wolfsthal in 2010 (black circles) and 2011 (white circles), respectively. The line represents linear regression of all SPMD sampling rates (expressed as natural logarithm; In Rs) vs. reciprocal value of absolute temperature (1 /T). The activation energy of mass transfer AE„ of 58 ± 10 kj mor1 was calculated from the slope of the line multiplied by gas constant R according to Eq. (6). B. Vrana et al. / Environmental Pollution 184 (2014) 101-112 105 was 24%, in that all the processes of analytical determination are included — sampling, extraction and determination by GC—MS. SPMD fabrication and field blanks contained concentrations of PAHs and PCBs that were below the instrumental limit of detection, with exception of naphthalene (up to 40 ng/SPMD). Blank subtraction for naphthalene in field exposed samplers was not done, because any naphthalene present in blanks dissipates from SPMDs during exposure to level which is at equilibrium with water (Lohmann et al., 2012). Solvent blanks processed concurrently with samplers did not contain quantifiable amounts of target analytes. In some exposed samples compounds as benzo[b]fluorantene, benzo [k]fluorantene, benzo[a]pyrene, indeno[l,2,3-cd]pyrene, dibenz [a,h]anthracene, benzo[ghi]perylene were present at concentrations below limit of quantification. Those compounds are ~i-1-1-1-1-1-1-r- III IV V VI VII VIII IX X Period I II III IV v VI VII VIM IX x Period „ 4 -#- BaA dissolved -O- Chr dissolved A BaA whole water Chr whole water c d) u c o Ü 3 - 1 - „ 4 - o) c c d) u c o Ü 3 - 1 - BbF dissolved BbF whole water II III IV V VI VII VIII IX X Period IV v VI VII VIII ix x Period Fig. 3. Temporal variability of free dissolved (using SPMD) water and whole water (using continuous water sampler) PAH concentrations, at the online monitoring station in Wolfsthal in July—December 2010. Samplers were continuously exposed in 14-day deployment periods. Deployment and retrieval dates are reported in Table 1. 106 B. Vrana et at. / Environmental Pollution 184 (2014) 101-112 hydrophobic and predominantly partitioned to suspended particles and colloids in water and only a small fraction is present in the dissolved phase. For the calculation of the mean concentration of those compounds in water according to Eq. (1). the mass in sampler Ns was substituted by instrumental limit of quantification (LOQJ. In such case the calculated water concentration of those compounds represents the highest possible concentration. Water concentrations estimated from LOQ for the above compounds were in range 0.01—0.29 ng IT1. The highest LOQ values were calculated for the sampling site in Wolfsthal, because of low Rs values obtained at this site due to slow water motion inside the exposure tube. Where method LOQ was applied in mean value estimates, data in Figs. (4— 5) are labelled with an asterisk. Detection limits in water can be significantly improved by a longer sampler exposure time or by exposure conditions, e.g. higher water turbulence. One option to increase sampling rates would be the use of samplers with a larger surface area, since the sampling rate is a product of mass transfer coefficient and sampler surface area. The 14-day exposure of samplers in this study was a result of a compromise to enable a direct comparison with other tested sampling methods (suspended particle traps and composite water samples). The PRC-derived sampling rates Rs for phenanthrene from all field exposures are shown in Tables 1 and 3. Rs values for other compounds were derived using Eq. (3), which estimates a slight decrease in Rs with increasing molecular mass. Phenanthrene Rs values ranged from 2.0 L d-1 at Wolfsthal in December 2010 to a maximum of 26.2 L d-1 at Altenwôrth in October 2011. In agreement with assumption of water boundary layer uptake different Rs values were obtained at different sites and during different seasons, which is related to differences in flow rates of water in the river and the position of sampler in the stream. Further relevant factors that affect mass transfer include the temperature and possibly the presence of biofouling and particle deposition on the surface of sampler. At the monitoring station Wolfsthal the sampler was placed in a glass tube, where the river water was pumped with lower flow velocity/turbulence than is in the river. This explains the generally lower sampling rates at this site. With exception of the sampling site Čunovo SPMD sampling rates increase with increasing temperature. The effect of temperature on passive sampling with SPMDs could be quantified at the site Wolfsthal (Fig. 2). Since the effect of flow velocity/turbulence on mass transfer into passive sampler at this site could be kept relatively constant, temperature was the only variable factor in exposures, when neglecting potential variable effects of biofouling and suspended particulate matter concentrations on mass transfer. At other sites, such evaluation was not possible because flow conditions could not be controlled for caged samplers. The effect of temperature on Rs can be quantified in terms of activation energies (AEa) for mass transfer, as modelled by the Arrhenius equation Rs = RSmexp(-^j (6) where RSoo is the sampling rate at the hypothetical upper limit where temperature is infinite, R is the gas constant and T is the absolute temperature. Values of AEa can be determined by plotting the natural logarithm of Rs (In Rs) vs. the reciprocal absolute temperature (1/r). The activation energy can then be calculated by multiplying the slope of the linear regression line with the gas constant. The calculated activation energy for phenanthrene in this study A£Q of 58 ± 10 kj mol-1 is in line with the average A£Q of 37 ± 21 kj mol-1 summarised for a broad range of studies by Huckins et al. (2006). This means that a temperature increase from 10 to 20 °C causes an increase in sampling rate by a factor about 2.3. 3.2. Temporal and spatial variability ofPAHs in the Danube river In 2010 temporal variability of PAH concentrations was investigated at a single sampling site in the Danube at Wolfsthal. Sum of concentrations of free dissolved PAHs determined from SPMDs deployed during the 2010 campaign were 5—39 ng L_1. The SPMD data (Fig. 3) show that the freely dissolved concentrations of individual PAHs in the water column increase during the winter months. This may reflect the higher PAH emissions from pollution sources, mainly from burning of fossil fuels, in winter. The atmospheric deposition is one of the important transport processes, by which PAHs enter the water phase. The higher activity of emission sources in winter in combination with climatic conditions such as temperature inversion that limits the vertical dispersion and less intensive atmospheric reactions create favourable conditions for PAH deposition to water phase. Moreover, Henry's law constant increases with increasing temperature and thus, higher equilibrium concentrations in water are expected at lower temperatures even when atmospheric concentration remains constant (Staudinger and Roberts, 2001). In addition to the general trend of concentration increase with decreasing water temperature, an increase of concentrations of Table 4 Correlation of free dissolved (Cfree) and whole water (Ctotal) concentrations of PAHs at the site Wolfsthal during the sampling campaign in 2010 with mean water temperature (T), suspended particulate matter content (SPM), and total organic carbon content (TOC). Compound Log/Cow ctoa\ Cfri Ore e T SPM TOC T SPM TOC Naphthalene NAP 3.37 a0.70 0.03 0.28 0.34 -0.41 0.19 0.30 Acenaphthylene ACE 4.00 a0.79 -0.31 0.28 0.15 -0.61 0.02 0.06 Acenaphthene ACY 3.92 0.23 0.54 0.31 0.57 -0.33 0.38 0.57 Fluorene FLU 4.18 0.37 0.27 0.28 0.57 -0.60 0.03 0.12 Phenanthrene PHE 4.57 -0.01 0.29 0.17 0.55 a-0.62 0.06 0.11 Anthracene ANT 4.54 0.00 0.52 0.48 a0.65 -0.61 0.14 0.16 Pyrene PYR 5.18 -0.11 0.31 0.31 a0.64 a-0.68 0.01 -0.12 Fluoranthene FLT 5.22 -0.15 0.19 0.10 0.42 -0.72 -0.09 -0.20 Chrysene CHR 5.86 -0.14 0.34 0.33 a0.67 a-0.69 -0.02 -0.14 Benzo[b]fluoranthene BbF 5.90 -0.29 0.43 0.42 a0.67 a-0.62 -0.01 -0.15 Benz[ a ]anthracene BAA 5.91 -0.19 0.33 0.27 0.58 -0.75 -0.17 -0.24 Benzo[a]pyrene BAP 6.04 -0.32 0.56 0.49 a0.75 a-0.81 -0.20 -0.23 Benzo[ghi]perylene BP 6.50 -0.53 a0.63 0.52 0.59 a-0.77 -0.13 -0.20 Indeno[ 1,2,3-cd ]pyrene IP 6.50 0.36 0.08 0.37 0.47 a-0.77 -0.13 -0.20 Dibenz[a,h]anthracene DahA 6.75 -0.53 0.41 0.31 0.48 a-0.77 -0.13 -0.20 a Significant Pearson product moment correlation coefficients (n = 10, p < 0.05; non-directional t-test) and higher than 0.62. B. Vrana et al. / Environmental Pollution 184 (2014) 101-112 107 Table 5 Dissolved concentrations (ng L"1) of sum of PAHs, Cd, Ni and Pb measured in urban impacted European rivers. River PAHs Cd Ni Pb Reference Danube 13-72 2-14 205- -544 18-74 This study Morava 25-203 Prokeš et al., 2012 Marne 7-19 Thévenot et al., 1998 Seine 9-70 11-67 Thévenot et al., 1998 Chiffoleau et al., 1999 15-50 8-111 338- -3760 Tusseau-Vuillemin et al., 2007 3.5-106 Bourgeault and Gourlay-Francé, 2013 Thames 800 Neal et al., 2000 Rhône 55 11 423 76 Miege et al., 2012 Bosna 20-480 Harman et al., 2013 1-24 218- -2981 8-1000 Vrana et al., authors unpublished data some lighter PAHs (naphthalene, acenaphthene, fluorene and phenanthrene) was observed during the third sampler exposure period (03.08. to 17.08. 2010). In August 2010, a local flood occurred at the Danube sampling profile in Wolfsthal and the elevated concentrations of dissolved compounds may be related to mobilization of these compounds during the event. At the Wolfsthal site free dissolved concentrations of PAHs obtained with passive sampling (Cfree) can be compared with whole water concentrations (Ctotai) determined in composite water samples representative of each of the 14-day sampler deployment periods (Fig. 3). The comparison reveals that Cfree in water decreases with increasing compound hydrophobicity (Supplementary information Fig. S5), which reflects the adsorption of hydrophobic compounds on particles or colloids. A significant positive correlation (Table 4) between Cfree and Ctotai was observed only for the two most hydrophilic compounds (naphthalene and acenaphthylene), which are predominantly present in water in the dissolved phase. While Cfree was negatively correlated with temperature for most compounds, such trend was not observed for Qotai- With exception of the most hydrophilic compounds (naphthalene and acenaphthylene), Ctotai of PAHs was positively correlated with total organic carbon (TOC) content in water, which confirms that hydrophobic compounds are associated with organic matter present on particles and in colloids in water. One hypothesis for the absence of correlation between Cfree and Ctotai is that a fraction of compounds adsorbed on suspended particulate matter is bound irreversibly and cannot partition into dissolved phase, however, such investigation was beyond the scope of this study and more research is needed to prove it. Since whole water concentration measurements were performed with a single composite sample during each sampling period, no data on precision of whole water sampling in one laboratory is available in this study. Collection and analysis of replicate samples would likely reveal whether absence of correlation with free dissolved concentration can be attributed to low precision of sampling and analysis. However, considering the very high sampling and processing effort needed to obtain a representative water sample for a 14 day period, such experiment is practically not feasible. In addition to samples collected during our study, information is available on concentrations of PAHs in spot samples of whole water (1 L) that were collected monthly in 2010 at the Wolfsthal monitoring station by Water Research Institute Bratislava for the purpose of chemical status assessment in the river Danube (Water ■a 4 4.0 Altenwörth □ February_2011 ■ April_2011 ■ October 2011 SS///, ■ŕ 8.0 7.0 5.0 Wolfsthal □ February 2011 □ April 2011 ■ July_2011 ■ October_2011 L í j n*, f li v. Langenzersdorf □ February_2011 ■ July_2011 ■ October_2011 iLk "luk >/'///Y V///// „ 5.0 Ml Čunovo m ň! raj -□ February_2011-~DApril_2011 □ July_2011 -■ Odober 2011 - V ■ j- ✓ ✓ s / / / Fig. 4. Temporal variability of free dissolved PAH concentrations, monitored using SPMD passive samplers at four sampling sites along the Danube river in 2011. Data points labelled with asterisks include individual measurements below limit of quantification. 108 B. Vrana et at. / Environmental Pollution 184 (2014) 101-112 Research Institute, Bratislava, 2013). During the whole duration of sampling campaign, concentrations of all monitored priority pollutant PAHs were below their respective LOQs. The LOQs of individual compounds were relatively high (2—30 ng IT1; Supplementary data, Table SI). The data do not contradict our observations, however, no statements on temporal variability of pollution can be made on their basis. Although data from regulatory monitoring, obtained using low volume spot sampling, can be applied for checking compliance with environmental quality standards, they are not suitable for assessment of temporal and spatial variability of PAHs. The availability of water discharge data at the Wolfsthal monitoring station enabled to estimate fluxes (as a product of discharge and concentration) of free dissolved as well as total PAHs in the rived Danube. The estimated flux of dissolved PAHs (sum of 16 compounds) ranged from 0.9 kg d_1 in October to 6.6 kg d_1 in December 2010, respectively. Estimated total PAH flux in Danube ranged from 3.3 kg d_1 in October to 16.7 kg d_1 in August (period III), respectively. The maximum total flux coincides with the above mentioned elevated water flow event. The average contribution of free dissolved compounds to total flux was 31%. We stress that the ultimate aim of passive sampling is to obtain a measure of the level of pollution that gives a representative measure of the exposure of organisms and compare the contaminant levels in time and space, but not to assess mass balance of compounds in water bodies. In 2011 the samplers were deployed during four seasons at four sampling sites to characterize the temporal and spatial variability of priority metals, PAHs and PCBs in the water column of the Danube river between the cities of Vienna and Bratislava. Total concentration of PAHs determined from SPMDs in the campaign conducted at four sampling sites in 2011 were 13—72 ng IT1. A comparison with free dissolved concentrations measured with passive sampling in other urban impacted European rivers shows that the pollution of Danube by PAHs is 1.5—7 times lower than in the rivers for which data is compiled in Table 5. Temporal variability of PAH concentrations at the four sites is shown in Fig. 4. In agreement with observations from 2010 the highest PAH concentrations at all four sampling sites were observed in winter (February) and the lowest ones in summer Quly). respectively. A single exception to this general trend were elevated concentrations of fluoranthene, pyrene and chrysene that were observed at Čunovo in July 2011. This event may have been related to on site sediment dredging activities or from accidental release of PAHs from ships, but would require a more detailed investigation. Spatial variability of PAHs during different seasons along the monitored Danube stretch is shown in Fig. 5. To visualize spatial trends of free dissolved concentrations, data from sites downstream the Altenwôrth site (AL) were presented as percentual concentration increase or decrease against the levels measured at AL site. Visualisation was performed only for compounds where concentrations exceeded their respective LOQs. No systematic spatial trends in PAH concentrations could be observed along the monitoring stretch since different and often opposite trends were observed during different seasons. The spatial variability of PAH concentration was not dramatic and for most compounds the concentrations varied less than two-fold in both directions in comparison with those measured at the AL site. In February 2011 ^ ŕ ý ^ / ^ a April 2011 4 I 'i f ľ / # & <ŕ ŕ # ŕ ŕ j- 0# 4 <ŕ 4 October 2011 V5" DÍL 1 ŕ ŕ č <ŕ <ŕ ŕ <ŕ s <ŕ 97% pure) was purchased from Sigma-Aldrich Chemie GmbH, Steinheim, Germany. Performance reference compounds - Dl0-acenaphthene, Dl0-fluorene, Dl0-phenanthrene and Dl0-pyrene - were sourced from Dr Ehrenstorfer GmbH, Augsburg, Germany. Dg-Naphthalene, Dl0-anthracene, Dl2-fluoranthene, Dl2-benzo(a)anthracene, Dl2-benzo(k)fluoranthene, Dl2-benzo(g,h,i)pyrene, PCB 30, PCB 185 and d6-gamma HCH (Dr Ehrenstorfer GmbH, Augsburg, Germany) were used as recovery standards. PCB 121 and terphenyl (Dr Ehrenstorfer GmbH, Augsburg, Germany) were used as internal standards for PCB and PAH instrumental analysis, respectively. Pesticide residue analysis grade n-hexane, dichloromethane, trichloromethane and all other solvents Crocodile V River Magalies River Sample site Hartbeespoort Crocodile • Pretoria J L ^'Veľ^J_^ Hennops River r* \ Jukskei } ^Ziittb abwe,. Botswana y3"*"^0Zunique f Pretoria] 11 mil . \ f^^r • \£ \ f Johannes-QJJ V J* Durban ♦ Sample site Johannesburg Figure 1 Map showing the sampling site in the Hartbeespoort Dam, South Africa (all > 97% pure) used were purchased from Sigma-Aldrich (Prague, Czech Republic). Milli-Q water (18MH-cm) was obtained from the Millipore Simplicity 185 system (Millipore, Bedford, MA, USA). Sampling site The sampling site was located in the Hartbeespoort Dam, 25°45'09.97"S, 27°53'04.39"E, about 37 km west of Pretoria and on the Crocodile River in North West Province, South Africa (Fig. 1). The dam is a 20.7 km2 water reservoir sandwiched between the Magalies mountain range in the Highveld region of northern South Africa (Nyoni et al., 2011). The dam reservoir receives water from a catchment area of about 4 100 km2, via the Jukskei and Hennops rivers that flow into the Crocodile River (Harding et al., 2004). The five catchment basins of the dam are, from west to east: the Magalies/Skeerpoort, the Crocodile, the Juskei, the Hennops and the Swartspruit basin (Van Rei, 1987). The Crocodile River accounts for about 90% of the dam's water supply with rainwater being the major source in summer. This scenario dramatically changes during the dry season (winter) as 50% of the water received by the dam then comprises treated wastewater from urbanised areas upstream (Harding et al., 2004), which creates environmental challenges for the water body. Although the origins of the Crocodile River system can be traced to the north of the city of Johannesburg, extensive rural crop farming is still carried out within the dam's drainage area, using its water. Considerable urban development is also present along the shorelines of the basin, and a portion of the impounded water from the dam is utilised for domestic supply, both within the riparian community and in downstream urban centres (DWA, 2012). Monitoring of the water body for PAHs, PCBs and OCPs using SPMDs was done in each of the four seasons of the year: winter, spring, summer and autumn, as described in Table 1. Sampling procedure At the deployment site, the samplers, including the field controls, were unpacked from the metal cans and placed on clean aluminium foil. The samplers were then mounted onto the deployment devices (protected by a steel casing). Once ready, they were quickly immersed in the water at between 1 and 1.5 m depth below the water surface. The steel cages housing http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 TABLE 1 Deployment periods and some water quality parameters Sampling period Water temperature (°C) pH Dissolved oxygen (mg-£') Winter 19-04-2011 to 04-05-2011 11.6 8.0 5.01 Spring 19-08-2011 to 02-09-2011 14.2 8.2 5.23 Summer 18-11-2011 to 02-12-2011 25.5 9.0 6.04 Autumn 24-02-2012 to 09-03-2012 19.7 8.6 5.75 the samplers were tied using ropes and anchored firmly to buoys. Finally, the field blanks were placed in airtight tin cans and transported in portable ice chests ('cool boxes') to the laboratory, where they were stored at -20°C. SPMD sampler preparation and deployment SPMDs (dimensions: 2.5 x 91.4 cm, 460 cm2 external surface area, and a wall thickness of 70 u.m) were prepared from LDPE layflat tubing (Brentwood Plastics, MO, USA) and filled with 1 mf; of high purity triolein (l,2,3-tri(cis-9-octadecenoyl)glyc-erol) (99% pure) which had previously been spiked with performance reference compounds, namely, fluorene-d10, acenaph-thene-d10, anthracene-d10, phenanthrene-d10 and pyrene-d10, to yield a nominal concentration of 2 pig-g1 triolein. Prepared samplers were stored in airtight sealed metal cans under freezing temperatures (-20°C) awaiting deployment. SPMDs were deployed in the water body in triplicates for a 14-day period. On retrieval, samplers were placed in airtight metal containers and quickly transported to the laboratory where they were stored at -20°C until processing. SPMD processing After removing particulates and biofouling from the surface of affected SPMDs using a soft brush and tap water, they were briefly immersed in diluted (10%) hydrochloric acid to rid them of adsorbed carbonates acquired during field deployment. The samplers were again flushed with sufficient amounts of tap water, and dried using acetone and a soft paper tissue. Each sampler was transferred into a pre-cleaned, empty 250 mf glass bottle with a ground joint stopper and 100 mf of HPLC grade n-hexane added. Each sampler was then spiked with surrogate standard solutions, namely, naphthalene-dg) fluoranthene-dl2, benzo(a)anthracene-dl2, benzo(k)fluoranthene-dl2, benzo(g,h,i) pyrene-dl2, PCB 30 and 85, and d6-gamma HCH, and extraction done twice for 24 h in the dark at room temperature. The extracts were combined and reduced to about 10 mf using a rotary evaporator (Heidolph Laborata4000, Germany) at 40°C before concentrating further to about 0.5 mf\ Finally, the extracts were reconstituted in 1 mf of pesticide residue analysis-grade trichloromethane. Removal of lipids that diffused into the extract during dialysis was achieved using a gel permeation chromatography (GPC) system equipped with a high pressure pump (HPP5001) and a fraction collector (ECOM, Prague, Czech Republic). A gel 5 u.m 50 A, 7.5 x 300 mm, high performance size exclusion chromatography column (Agilent PL) was used to fractionate the extracts with chloroform as the mobile phase at a flow rate of 0.6 mf-minAnalytes were collected from 18 min 20 s to 41 min 40 s and reduced to the last drop using a gentle stream of nitrogen gas. Subsequently, the extract was reconstituted to 1 ml n-hexane. The GPC eluate was subjected to further cleanup by activated silica gel. One portion (20%) of the GPC extract targeting PAHs was cleaned using activated silica gel packed in a glass column and eluted with 10 ml of n-hexane followed by 20 ml of dichloro-methane. The remaining portion (80%) targeting PCBs and OCPs was cleaned with sulphuric acid-modified activated silica gel, prepared by mixing 33 ml of concentrated sulphuric acid (> 98%) with 50 g of freshly prepared activated silica gel. Thorough homogenisation of the mixture was ensured before column packing. Target analytes were eluted with 30 mf dichloromethane. After reduction to 1 ml using a gentle stream of nitrogen gas, terphenyl or PCB 121 internal standards were added to the sample, and ultimately analysed by GC-MS/MS for PAHs and PCBs/OCPs, respectively. Instrumentation The PAHs of interest were analysed using a 6890 GC system coupled with a 5971 mass selective detector (Agilent Technologies). Chromatographic separation of the components was done using a capillary column (30 m x 0.25 mm internal diameter, 0.25 \xm film thickness) HP-5MS and helium as the carrier gas flowing at 1.5 m£-min_1. Conditions of gas chromatography separation were as follows: injector temperature was set at 250°C, initial column temperature was set at 70°C and held for 0.5 min. This ramped at 25°C-min1 to 150°C. It was then ramped at 30°C-min1 to 200°C. This was further ramped at 8°C-min1 to 280°C and held for 20 min. Detection of the separated PAHs was achieved using a MS/MS system operated in selected ion monitoring mode with the electron impact ionisation set at 70eV. The temperatures of the ion source, transfer line and the quadrupole were held at 230°C, 280°C and 150°C, respectively. Quantitation of the residues was accomplished using a 7-point standard calibration curve in the concentration range of 0 to 1 000 ng-£_1. GC-MS/MS was used for indicator PCBs and OCPs analysis. 6890N GC (Agilent, USA) equipped with a 60 m x 0.25 mm x 0.25 um DB5-MS column (Agilent J&W, USA) coupled to Quattro MicroGC MS (Waters, Micromass, UK) operated in EI+ was used; at least 2 MRM transitions were recorded for each compound analysed. Injection was done in splitless mode at 280°C and 1 \xl sample loaded. Helium was used as carrier gas at the flow of 1.5 mf-minThe GC temperature programme was 80°C (1-min hold), then 15°C-min1 to 180°C, and finally 5°C-min1 to 300°C (5-min hold). Raw data were processed using TargetLynx software (Waters, Micromass, UK). Quality control Fabrication controls and field blanks were used to account for contamination of the SPMDs during device construction, and sampler deployment and retrieval from the site. Vapour-phase contamination during deployment of the SPMDs was factored in by the field blanks. These blanks were subjected to identical processing treatment as the deployed devices. http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 TABLE 2 Mean concentrations of PAHs, OCPs and DDTs in SPMDs (ng SPMD \ n = 3) Compound Season Winter Spring Summer Autumn PAHs Naphthalene 307 ± 11 149 ±11 171 ±4 286 ± 20 Acenaphthylene 158 ±5 76 ±6 132 ±2 180 ±7 Acenaphthene 37 ± 3 20 ±2 25 ± 1 38 ±3 Fluorene 76 ± 11 61 ± 7 49 ±4 106 ±7 Phenanthrene 164 ± 36 71 ± 9 75 ± 5 217 ± 13 Anthracene 875 ± 40 279 ± 5 56 ± 12 164 ±3 Fluoranthene 147 ±4 65 ±3 119 ± 11 97 ± 7 Pyrene 105 ±2 49 ±3 88 ±5 83 ±8 Benz [a] anthracene 31 ± 1 16 ± 1 32 ±2 33 ± 1 Chrysene 41 ± 1 22 ±5 45 ±6 44 ± 1 Benzo [b]fluoranthene 35 ±4 26 ±8 36 ± 1 38 ±2 Benzo [kjfluoranthene 36 ±2 ND 36 ±2 38 ±2 Benzo [a] pyrene 34 ±6 32 ±0 39 ± 7 43 ±5 Indeno [1,2,3 -cd]pyrene 37 ±0 19 + 1 38 ± 1 39 ± 1 Dibenz [a,h]anthracene ND ND ND ND Benzo [ghi]perylene 35 ± 1 20 ±2 41 ±3 45 ±2 ZPAHs 2 117 ±57 905 ±21 984 ±21 1450 ± 29 HCHs HCB28 1.1 ±0.2 0.9 ±0.1 0.7 ± 0.0 0.6 ±0.1 a-HCH 55.9 ±3.5 52.5 ± 5.1 32.3 ±3.6 64.8 ± 5.5 ß-HCH 154.3 ±7.8 153.4 ±0.4 70.9 ± 4.0 45.5 ± 2.9 Lindane 8.8 ± 0.9 8.1 ±0.1 5.9 ±0.8 9.8 ±0.9 Ô-HCH 3.2 ± 0.2 3.1 ±0.2 1.4 ±0.2 2.5 ±0.4 e-HCH 9.9 ± 0.6 9.8 ± 0.6 4.5 ±0.1 5.6 ±0.7 ZHCHs 233.2 ± 8.6 227.7 ± 5.2 106.5 ±5.4 128.7 ±6.3 DDTs o,p'-DDE 0.6 ±0.1 0.8 ±0.1 0.4 ± 0.0 0.3 ±0.0 p,p'-DDE 5.3 ±0.2 5.5 ±0.4 3.6 ±0.3 3.1 ±0.2 o,p'-DDD 10.8 ± 1.1 13.6 ±0.2 5.0 ±0.4 4.2 ±0.4 p,p'-DDD 31.1 ±0.6 43.5 ±0.8 15.7 ±3.1 19.9 ± 2.7 o,p'-DDT ND ND ND ND p.p'DDT 0.6 ±0.1 0.7 ±0.1 0.7 ±0.1 0.7 ±0.1 ZDDTs 48.5±1.3 64.1±0.9 25.3±3.1 28.2±2.7 PCBs PCB 28 3.5 ±0.50 3.1 ±0.30 1.5 ±0.10 2.1 ±0.17 PCB 52 1.1 ±0.06 1.0 ±0.09 0.6 ± 0.00 0.5 ±0.04 PCB 101 0.4 ± 0.06 0.4 ± 0.03 0.3 ± 0.07 0.4 ± 0.01 PCB 118 0.2 ± 0.02 0.2 ± 0.00 0.2 ± 0.01 0.2 ± 0.00 PCB 153 0.5 ±0.03 0.5 ±0.07 0.4 ± 0.02 0.7 ± 0.05 PCB 138 0.6 ± 0.05 0.3 ±0.00 0.5 ± 0.05 0.5 ± 0.05 PCB 180 0.3 ± 0.02 0.3 ±0.00 0.5 ± 0.02 0.4 ± 0.02 ZPCBs 6.5 ±0.51 5.8 ±0.32 3.9 ± 0.14 4.8 ±0.19 ND: not detected RESULTS AND DISCUSSION Occurrence of PAHs, PCBs and OCPs in the SPMDs The absolute contaminant concentrations sequestered by SPMDs deployed at Hartbeespoort Dam during the 14-day deployment period in each of the four seasons of the year are presented in Table 2. Characteristic ions (m/z values) used in the analysis of polycyclic aromatic hydrocarbons in single ion monitoring (SIM) mode by GC/MS and characteristic MRM transitions (m/z values of parent and daughter ions) used in the analysis of PCBs and OCPs are given in the Appendix (Tables Al and A2). Analyte concentrations were adjusted with respect to their recoveries obtained from recovery standards introduced prior to the dialytic process. The SPMD field blanks showed no quantifiable concentrations of the target http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 contaminants. Determination of recoveries for all samples was carried out by spiking them with surrogate standards prior to extraction. Good recoveries were recorded that ranged from 55% to 123% for PAHs, 73% to 94% for PCBs and 72% to 104% for OCPs. The relative standard deviations between co-deployed triplicate samplers did not exceed 22% for PAHs, 24% for PCBs and 17% for OCPs. Limits of detection (LOD) and quantification (LOQs) for the method were 0.1 and 0.4 ng-f \ respectively for PAHs. LODs for PCBs and OCPs were all less than 0.1 ng-f 1 whereas LOQs were 0.1 ng-f Estimation of dissolved water concentrations of analytes Dissolved water concentrations of target analytes were calculated from amounts accumulated in SPMDs as follows: Amounts of analytes absorbed by the samplers follow a first-order approach to equilibrium. Aqueous concentrations were calculated from the amounts (N) absorbed by the SPMD, the in-situ sampling rate of the compounds R and their sampler-water partition coefficients K : K*,VS[1 - exp( -Rj/'KwVs))] where: V"s is the volume of the SPMD and t is the sampler exposure time. PRC dissipation also follows first-order kinetics. Sampling rates Rs were estimated using the non-linear least -squares method of Booij and Smedes (2010), considering the fraction/of individual PRCs (D10-acenaphthene, D10-fluorene, D10-phenanthrene and D10-pyrene) that remained in the SPMD after the 14-day exposure as a continuous function of their K w, with Rs as an adjustable parameter. / = exp"]TlH (2) V sw * s J v ' where: f= NmJN„ ■> N„ „„^ is the initial amount of the PRC at t = 0 Np is amount of each PRC remaining after exposure t is exposure period (14 days). Assuming water boundary layer controlled uptake, Rs of individual target compounds in the higher hydrophobicity range was estimated by substituting Eq. (3), derived by Rusina et al. (2010), into Eq. (2). Rs = FAM "aA1 (3) where: M is the molecular weight of the analyte, A is the surface area of SPMD (460 cm2) and F is the regression coefficient that was optimised using the non-linear least squares method for estimating sampling rates. The necessary K values were interpolated from the empirical equation (Huckins et al., 2006). log K = -01618(log K )2 + 2.321 log K -2.61 (4) The calculated free dissolved water concentrations of the PAHs, PCBs and OCPs are presented in Table 3. Temporal trends of water-dissolved contaminants Equation (3), which estimates a slight decrease in Rs with increasing molecular mass, was used to calculate compound-specific R values for all of the compounds studied. Depending on the water flow velocities, different R values were obtained in s the various seasons, in agreement with the assumption of water boundary layer uptake. Mass transfer of analytes may also be affected by other factors such as temperature, biofouling and deposition of particulates on the surface of the SPMDs. Estimated water soluble concentrations generally followed the trend: PAHs > OCPs > PCBs. PAHs are ubiquitous organic pollutants characterised by many natural and anthropogenic sources, unlike OCPs and PCBs (industrial products). Since the dam receives over 90% of its water from the Crocodile River, which originates in Johannesburg city, it is possible that a good portion of the pollutants sampled could be of industrial origin. PCB concentrations are on average 2 to 3 orders of magnitude lower than those of PAHs and OCPs because most of these manufactured products have long been banned and their use stopped, in line with the Stockholm Convention, and whatever was captured by the samplers is attributable to their environmental persistence due to slow degradation. The sum total of water-borne concentrations of the compounds ranged from 30.2 to 60.8 ng-f1 (PAHs), 10.0 to 10.7 ng-f1 (OCPs) and 38 to 150 pg-f1 (PCBs). Generally, the seasonal trends for all of the compounds mirrored the amounts accumulated in the SPMDs. An observed predominance of smaller molecular weight PAHs was evident in all four seasons. This maybe attributed to their higher solubility in water due to lower hydrophobicity and, hence, transportation from the point sources was probably more efficient. PAHs A remarkable seasonal variability in the amounts of sequestered PAHs was shown by the deployed SPMDs. Estimated total analyte concentrations ranged from 30.0 ng-f 1 (in summer) to a high of 60.8 ng-f 1 (in winter). These concentrations are comparable to those reported by Wang et al. (2009) (13.8-97.2 ng-f ') at the Three Gorges River, China, and Vrana et al. (2014) (5-72 ng-f ') in the Danube River, Slovakia/Austria. The trend of total concentrations of PAHs dissolved in water was as follows: winter > spring > autumn > summer. Individual PAH concentrations obtained in the various seasons also generally followed the same trend as the totals (Fig. 2). Smaller molecular weight PAHs constituted the highest percentage of the sequestered compounds. The reported water-soluble concentrations of the heavy molecular weight PAHs in the current study were on average up to 2 orders of magnitude lower than the maximum contaminant limits (MCL) set by international regulatory bodies such as the United States Environmental Protection Agency (USEPA) (0.01 to 0.04 ug-f ')• The elevated concentrations recorded during winter may be attributed to a number of factors. During the winter months, very little precipitation is recorded (average of about 4-9 mm for the study area) and, since the dam depends on river water for replenishment, its volume drastically drops. This in turn increases the percentage of the dam's water originating from treated wastewater, which can exceed 50% of the total volume (Harding et al., 2004). These wastewater treatment plants are located in the industrialised areas north of Johannesburg. In addition, average temperatures substantially drop during winter (to an average air temperature of 4-7°C as measured in the study area) which in turn discourages analyte losses via volatilisation. Atmospheric deposition of PAHs represents an http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On line) = Water SA Vol. 40 No. 3 July 2014 TABLE 3 Estimated dissolved water concentrations, (ng-£_1) of PAHs, PCBs and OCPs from SPMDs deployed at Hartbeespoort Dam in different seasons of the year Compound Season Winter Spring Summer Autumn PAHs Naphthalene 43.153 43.206 23.980 38.499 Acenaphthylene 3.862 3.897 3.214 4.354 Acenaphthene 1.115 1.173 0.759 1.099 Fluorene 1.176 1.858 0.716 1.646 Phenanthrene 1.283 1.316 0.386 1.808 Anthracene 7.321 2.200 0.310 1.519 Fluoranthene 0.709 1.340 0.223 0.639 Pyrene 0.698 0.994 0.170 0.561 Benz [a] anthracene 0.169 0.287 0.051 0.231 Chrysene 0.226 0.422 0.072 0.302 Benzo [b]fluoranthene 0.221 0.511 0.060 0.279 Benzo [k]fluoranthene 0.222 ND 0.060 0.282 Benzo [a]pyrene 0.199 0.294 0.065 0.233 Indeno [1,2,3 -cd] pyrene 0.221 0.362 0.063 0.297 Dibenz [a,h] anthracene ND ND ND ND Benzo [ghi]perylene 0.211 0.365 0.070 0.343 ZPAHs 60.768 58.225 30.199 52.082 HCHs a-HCH 2.320 2.212 0.666 2.921 P-HCH 6.780 6.661 8.102 6.151 Lindane 0.445 0.389 0.189 0.542 6-HCH 0.054 0.062 0.024 0.087 E-HCH 0.442 0.386 0.187 0.260 ZHCHs 10.350 10.201 9.168 9.961 DDTs o,p'-DDE 0.004 0.006 0.010 0.007 p,p-DDE 0.033 0.052 0.088 0.062 o,p'-DDD 0.065 0.104 0.176 0.124 p,p'-DDD 0.203 0.323 0.547 0.384 p,p'-DDT ND ND ND ND p,p'-DDT 0.004 0.006 0.011 0.008 ZDDTs 0.309 0.491 0.832 0.585 PCBs PCB28 0.019 0.029 0.067 0.020 PCB 52 0.006 0.012 0.025 0.010 PCB101 0.003 0.005 0.014 0.007 PCB 118 0.001 0.001 0.003 ND PCB 138 0.003 0.006 0.013 0.005 PCB 153 0.004 0.005 0.018 0.004 PCB 180 0.002 0.004 0.012 0.003 ZPCBs 0.038 0.062 0.150 0.049 ND: not detected important pathway for PAHs into the aquatic ecosystem. The increased concentrations in water in winter may correspond with elevated atmospheric concentrations during the same period due to enhanced combustion of coal for heating. The summer months experience high rainfall coupled with high temperatures. Resuspension of sediment-immobilised PAHs was expected to increase PAH concentrations in the water phase. Inputs from runoff and rivers originating from polluted areas upstream were also thought to be potential PAH sources. Although these factors may have been at play, it seems dilution effects (larger water volumes) as well as losses through volatilisation may have tempered the expected increase in contaminant concentrations. The autumn season is characterised by less precipitation and dropping temperatures. These conditions may have led to lower contaminant losses via volatilisation coupled with increased concentration due to decreased bulk water volumes. The PCB concentrations obtained from the deployed http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 Naphthalene Winter Spring Summer Autumn Deployment period Benzo[a]pyrene Winter Spring Summer Autumn Deployment period Winter Spring Summer Autumn Deployment period Benzo[k]fluoranthene Winter Spring Summer Autumn Deployment period Figure 2 Temporal changes in water-dissolved concentrations of some individual PAHsatthe site SPMDs were generally lower, by about 2 to 3 orders of magnitude, than PAH and OCP concentrations. PCBs are organic contaminants of purely anthropogenic origin, in contrast to PAHs that have both natural and anthropogenic sources. In addition, most PCBs were banned some years back and the remnants captured by the samplers are as a result of the strong persistence of PCBs in the environment. Due to their strong hydrophobicity (shown by higher log K values of up to several orders of magnitude), PCBs tend to partition away from the water phase and preferentially adsorb strongly onto particulate matter, colloids and sediments in water. Moreover, their emissions are likely to be much lower than those of PAHs because, unlike the western industrialised countries, South Africa may not have utilised PCBs heavily during its economic growth in the 1980s or later when usage of PCBs was banned (Ogata et al., 2009). Estimated water concentrations of the sum total of PCBs are shown in Table 3. When ranked in increasing order, the water-dissolved analyte concentrations followed the trend: summer > spring > autumn > winter. Concentrations of the compounds ranged from a low of 0.038 ng-f1 in winter to a high of 0.150 ng-f1 in summer. These concentrations were comparable to those obtained by Vrana et al. (2014) in the Danube River (5 to 16 pg-f ') and Allan and Ranneklev (2011) in the Alna River, Norway (0.7 to 85 pg-f ')• Clearly, PCB levels in summer were significantly higher than those recorded in all of the other seasons. This observation may be explained as follows: In the summer rainfall region of South Africa, within which the study area lies, the summer period usually experiences heavy rainfall (90-125 mm). Most PCB congeners are highly hydrophobic compounds which preferentially adsorb strongly onto soil particles and sediments. Therefore, heavy rain events may disrupt these strong interactions thereby remobilising them into the water phase. This is partly supported by the fact that usage of these compounds has been banned for several years and therefore a majority of inputs could be coming from sediment samples. Surface runoff from urban centres (where these compounds are found in higher quantities) may also add to the pollutant load. A good portion of the water that eventually finds its way to the sample site can be traced to the industrial areas of Johannesburg (Fig. 1). The estimated freely dissolved water concentrations of OCPs are given in Table 3. Seasonal ranking from lowest to highest followed the trend: summer, autumn, spring, winter. The sequestered amounts of OCPs that comprised hexachloro-cyclohexanes (HCHs), and DDX (DDTs, DDDs and DDEs) were up to 2 orders of magnitude higher than PCBs but slightly less than those of PAHs. Among the analysed OCPs, HCHs contributed over 78% of the quantified amount and their free dissolved concentrations ranged from about 9.2 ng-f 1 in summer to 10.4 ng-f 1 in winter. Figure 3 presents the water dissolved concentrations of selected HCH isomers. Particularly high levels of |3-HCH were detected at the sampling site in all four seasons. This HCH isomer is characterised by a much lower vapour pressure, better solubility in water, and lower Henry's law constant than all of the other HCH isomers, which favour partitioning from air to water. Compared to the gamma- and alpha-HCHs, it is the most recalcitrant isomer (Stockholm Convention, 2007). In a global monitoring study of persistent organic pollutants (POPs) in coastal waters, Ogata et al. (2009) reported high concentrations of HCHs in samples from South Africa. This was in contrast to levels obtained in other parts of the world (such as USA, Asia and Europe) which were lower. They attributed this observation to the application of lindane in South Africa, which contains y-HCH as its main component. It is possible that technical-grade HCHs that also contain considerable amounts of |3-HCH (5-12%) were applied. Because of its relative volatility, this globally-banned pesticide can easily find its way into water systems via atmospheric deposition. In the soil-air interface, ratios of HCH isomers have been used to identify the historical pollution sources (Willett et al., 1998). |3-/(a+y)-HCH > 0.5 is an indicator of historical pollution while a ratio less than 0.5 indicates new introduction of HCHs. In the case of the current study, these ratios ranged from 1.8 in autumn to a high of 9.5 in summer. It is therefore proposed that in all four seasons, HCH input to the sampling site is predominantly historical in nature with minimal inputs from current application. Since the overall seasonal trends of HCHs generally mirrored those of PAHs (with the exception of values obtained in spring), we conclude that the same factors may have affected http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 Winter Spring Summer Autumn Deployment period JL Winter Spring Summer Autumn Deployment period Winter Spring Summer Autumn Deployment period Winter Spring Summer Autumn De ployme nt pe riod Figure 3 Water dissolved concentrations of individual HCHs as affected by seasonal change them. However, the decrease in OCP concentrations from winter to spring was insignificant. This observation may be attributed to their comparatively lower volatility. Thus, losses through volatilisation resulting from increased temperatures during spring may not have been a major factor. Interestingly, DDT and its metabolite residues showed seasonal patterns similar to those of PCBs, even though their water dissolved concentrations were generally higher. Estimated water concentrations of the DDT sum ranged from 0.31 ng-f1 in winter to 0.83 ng-f 1 in summer, respectively. Volatilisation is a major route through which DDT and its metabolites are released into the atmosphere and, once there, these chemicals are cycled back to surface water through dry and wet deposition (Stockholm Convention, 2007). Findings from this study suggest that wet deposition of DDT and its metabolites may be playing an important role in re-introducing them to the sample site, as higher water concentrations coincide with high precipitation (summer). Moreover, considering the relatively high log Kow values associated with DDX (5.8-6.79), remobilisation of the particle/sediment-bound fraction as a consequence of heavy rainfall may have been a possibility. Contributions from runoff originating from fields also cannot be ignored. Taken together, these factors may explain the seasonal trends of DDX. Source identification of PAHs in the Hartbeespoort Dam The principal sources of PAHs in the environment can be classified as either pyrogenic or petrogenic, with the pyrogenic inputs predominating in aquatic environments (Ekpo et al., 2012). Based on the SPMD-obtained PAH concentrations, identification of the probable sources was attempted. Reports by many authors on the apportionment of PAH sources in the environment using molecular ratios of certain PAHs are available in the literature (Baumard et al., 1998; Vrana et al., 2001; Zhang et al, 2004; Brandli et al, 2008). With respect to passive sampling, ratios of PAHs must be for compounds with near identical sampling rates to minimise bias arising from the mode of calculation of the rates for compounds with widely differing log K (Allan and Raneklev, 2011). Furthermore, the same authors observed that unless PAHs are directly emitted to surface water, dissolved phase concentrations may not necessarily be representative of sources of contamination. From among the several available approaches, ratios of fluoranthene/ (fluoranthene + pyrene) [Flt/(Flt + Pyr)] and anthracene/ (anthracene + phenanthrene) [Ant/(Ant + Phe)] calculated from waterborne concentrations were applied in the identification of the possible sources of PAHs in the site. Figure 4 shows the diagnostic ratios of PAH concentrations measured with SPMDs in Hartbeespoort Dam. An Fit/ (Fit + Pyr) ratio > 0.5 indicates a pyrogenic source, as does an Ant/(Ant + Phe) ratio > 0.5. Ratios of indeno[l,2,3-cd]pyr-ene/(indeno[l,2,3-cd]pyrene + benzo[g,h,i]perylene) greater than 0.5 point to fossil fuel combustion or pyrogenic sources (Brandli et al, 2008) for the PAHs in the Hartbeespoort Dam. Thus, with the exception of concentrations obtained from SPMDs deployed in winter, all PAHs pointed to a pyrogenic origin. The winter-derived data showed a mixture of both pyrogenic and petroleum combustion sources. The spike in the petrogenic PAH fraction during winter may be attributed to the increased proportion of treated wastewater originating from Johannesburg. As Harding et al. (2004) reported, during winter, precipitation is almost nil and, consequently, more than 50% of the reservoir's inlet water is composed of treated wastewater. A steep increase in the Flt/(Flt + Pyr) ratio was observed from winter to spring before decreasing during summer. A further drop in the ratio, albeit gently, occurred between summer and autumn. CONCLUSIONS SPMDs are potentially effective tools for monitoring hydrophobic contaminants in aqueous systems such as those present in the Hartbeespoort Dam, South Africa. In addition to detecting concentrations of PAHs, PCBs and OCPs in the toxicologically most relevant dissolved phase, SPMDs also captured their seasonal variation in the water body. Generally, total contaminant concentrations in the dam increased in the order: summer, spring, autumn, winter. Concentrations of the PAH and HCH isomers decreased with increasing water temperature, which likely reflects seasonality of atmospheric deposition. The dissolved concentrations of PCB and DDT isomers are most likely related to desorption from suspended particles. Diagnostic ratios of PAHs measured in SPMDs were used to identify the possible sources of PAHs in the water. These ratios indicated http://dx.doi.org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 P exogenic P exogenic 3.7 0.6 - 3.3 3.4 E 0-2 Pyrogenic 0.7 -0.6 -£ 0.4- SP SU * Vu wi Grass, wood arid coal combustion Petroleum combustion iH 0J> -3.1 -Q Petroleum ( ) 0.2 0.4 0.6 D.B 1 AntfAnt + Phe Pyrogenic SU ♦A.U SP WI - -♦- 3.4 3.6 AntfArit +■ Phe Grass, wood arid coal combustion Petroleum combustion Petroleum Figure 4 Diagnostic ratios of PAH concentrations measured with SPMDs in the Hartbeespoort Dam. WI: winter; SP: spring, SU: summer, AU: autumn, Fit: fluoranthene, Pyr:pyrene, Ant: anthracene, Phe: phenanthrene. that the PAH concentrations in the dam during spring, summer and autumn were mainly of pyrogenic origin while the winter levels comprised both pyrogenic and petrogenic sources. ACKNOWLEDGEMENTS The authors would like to thank Lenka Vaňková of RECETOX, Masaryk University, Czech Republic, for technical assistance and for financial support from the European Regional Development Fund, from the Ministry of Education of the Czech Republic (LM2011028 and L01214), the project "Employment of Best Young Scientists for International Cooperation Empowerment" (CZ. 1.07/2.3.00/30.0037) co-financed from European Social Fund and the state budget of the Czech Republic, as well as from the National Research Foundation and Water Research Commission of South Africa. REFERENCES ALLAN IJ and RANNEKLEV SB (2011) Occurrence of PAHs and PCBs in the Alna River, Oslo (Norway). /. Environ. Monit. 13 2420-2426. ATSDR (AGENCY FOR TOXIC SUBSTANCES AND DISEASE REGISTRY) (2009) Toxicological profile for polycyclic aromatic hydrocarbons (PAHs). Case studies in environmental medicine. United States Department of Health Services, Public Health Service, Atlanta, Georgia, U.S.A. BAUMARD P, BUDZINSKI H, GARRIGUES Q, BURGEOT T and BELLOCQ J (1998) Origin and bioavailability of PAHs in the Mediterranean Sea from mussel and sediment records. Estuar. Coast. Shelf Sci. 47 77-90. BOOIJ K and SMEDES F (2010) An improved method for estimating in situ sampling rates of nonpolar passive samplers. Environ. Sci. Technol. 44 6789-6794. BOOIJ K, SLEIDERINK HM and SMEDES F (1998) Calibrating the uptake kinetics of semipermeable membrane devices using exposure standards. Environ. Toxicol. Chem. 17 1236-1245. BOOIJ K, VRANA B and HUCKINS JN (2007) Theory, modelling and calibration of passive samplers used in water monitoring. In: Greenwood R, Mills G and Vrana B (ed.) Passive Sampling Techniques in Environmental Monitoring: Comprehensive Analytical Chemistry Series Vol. 48. Elsevier, Amsterdam. 141-169. BOUWMAN H (2003) POPs in South Africa: The handbook of environmental chemistry. Persistent Org. Pollut. 3 (0) 297-320. BOUWMAN H, COOPAN RM, REINECKE AJ and BECKER PJ (1990) Levels of DDT and metabolites in breast milk from KwaZulu mothers after DDT application for malaria control. Bull. World Health Organ. 68 761-768. BRANDLI RC, BUCHELI TD, AMMANN S, DESAULES A, KELLER A, BLUM F and STAHEL WA (2008) Lipid containing semipermeable membrane devices for monitoring organic contaminants in water./. Environ. Monit. 10 1278-1286. DAS SK, ROUTH J and ROYCHOUDHURY A (2008) Sources and historic changes in polycyclic aromatic hydrocarbon input in a shallow lake, Zeekoevlei, South Africa. Org. Geochem. 39 1109-1112. DEGGER N, WEPENER V, RICHARDSON BI and WU RSS (2011) Brown mussels {Perna perna) and semipermeable membrane devices (SPMDs) as indicators of organic pollutants in the South African marine environment. Mar. Pollut. Bull. 63 91-97. EKPO BO, FUBARA FB, EKPA OD and MARYNOWSKI HL (2012) Determination of hydrocarbon sources using n-alkane and PAH distribution indices in sediments from coastal areas of Bonny River in Niger delta, Nigeria. ARPN J. Earth Sei. 1 9-20. HARDING WR, THORNTON JA, STEYN G, PANUSKA J and MORRISON IR (2004) Hartebeespoort Dam Remediation Project (Phase 1) Volume 1 Action Plan. Department of Agriculture, Environment and Tourism of the North West Province Government (DACET, NWP), South Africa. HUCKINS JN, MANUWEERA GK, PETTY JD, MACKAY D and LEBO JA (1993) Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environ. Sci. Technol. 27 2489-2496. HUCKINS JN, PETTY JD and BOOIJ K (2006) Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices. Springer, New York. http://dx.doi.Org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 HUCKINS JN, PETTY JD, LEBO JA, ALMEIDA FV, BOOIJ K, ALVAREZ DA, CRANOR WL, CLARK RC and MOGENSEN BB (2002) Development of the permeability/performance reference compounds approach for in situ calibration of semipermeable membrane devices. Environ. Sei. Technol. 36 85-91. KOT A, ZABIEGALA B and NAMIESNIK J (2000) Passive sampling for long-term monitoring of organic pollutants in water. Trends Anal Chem. 19 (7) 446-459. LANDRUM PF, REINHOLD MD, NIHART SR and EADIE BJ (1985) Predicting the bioavailability of organic xenobiotics to Pontoporeia hoyi in the presence of humic and fluvic materials and natural dissolved organic matter. Environ. Toxicol. Chem. 4 459-467. LEVINSON J, CHANNSLUSZNY S, YASMAN Y, BUTALOV V and SCHETCHERI (2008) Detector for particulate PAHs in water. Anal Bioanal. Chem. 381 1884-1591. MAYER P, TOLLS J, HERMENS L and MACKAY D (2003) Equilibrium sampling devices. Environ. Sei. Technol 37 184A-191A. NIEUWOUDT C, PIETERS R, QUINN L, KYLIN H, BORGEN AR and BOUWMAN H (2011) Polycyclic aromatic hydrocarbons (PAHs) in soil sediment from industrial, residential, and agricultural areas in central South Africa: An initial assessment. Soil Sediment Contam. 20 188-204. NYONI H, CHIMUKA L, VRANA B and CUKROWSKA E (2011) Membrane assisted passive sampler for triazines compounds-characterisation of environmental conditions and field performance. Anal Chim. Acta 694 (1-2) 75-82. OGATA H, TAKADA K, MIZUKAWA H, HIRAI S, IWASA S, ENDO Y, MATO M, SAHA K, OKUDA A, NAKASHIMA M, MURAKAMI N, ZÜRCHER R, BOOYATUMANONDO MP, ZAKARIA LQ, DUNG C, MIQUEZ MM, GORDON S, SUZUKI C, MOORE HK, KARAPANAGIOTI T, MCCLURG S, WEERTS E, BURRES W, SMITH M, VAN VELKENBURG JS, LANG RC, LANG D, LAURSEN B, DANNER N, STEWARDSON RC and THOMPSON T (2009) International pellet watch: Global monitoring of persistent organic pollutants (POPs) in coastal waters. 1. Initial phase data on PCBs, DDTs and HCHs. Mar. Pollut. Bull. 58 1437-1446. QUINN L, PIETERS R, NIEUWOUDT C, BORGEN AR, KYLIN H and BOUWMAN H (2009) Distribution profiles of selected organic pollutants in soils and sediments of industrial, residential and agricultural areas of South Africa. /. Environ. Monit. 11 1647-1657. RUSINA T, SMEDES F, KOBLIZKOVA M and KLANOVA J (2010) Calibration of silicone rubber passive samplers: experimental and modelled relations between sampling rate and compound properties. Environ. Sei. Technol 44 362-367. SABALIUNAS D and SODERGREN A (1997) Use of semipermeable membrane devices to monitor pollutants in water and assess their effects: a laboratory test and field verification. Environ. Pollut. 96 (2) 195-205. STOCKHOLM CONVENTION (2007) Stockholm Convention on Persistent Organic Pollutants (POPs). URL: http://chm.pops.int (Accessed November 2013). TIEMANN U (2008) In vivo and in vitro effects of the organochlorine pesticides DDT, TCPM, methoxychlor, and lindane on the female reproductive tract of mammals: a review. Reprod. Toxicol 25 (3) 316-326. VAN REI WF (1987) The Hartbeespoort dam - a magnet to millions? In: Thornton JA and Walmsley RD (eds.) Hartbeespoort Dam -Quo Vadis. FRD Ecosystem Programmes Occasional Report Vol. 25. 83-93. VERWEIJ F, BOOIJ K, SATUMALAY K, VAN DER MOLEN N and VAN DER OOST R (2004) Assessment of bioavailable PAH, PCB and OCP concentrations in water, using semipermeable membrane devices (SPMDs), sediments and caged carp. Chemosphere 54 1675-1689. VRANA B, MILLS GA, ALLAN IJ, DOMINIAK E, SVENSSON K, MORRISON G and GREENWOOD R (2005) Passive sampling techniques for monitoring pollutants in water. Trends Anal Chem. 24 845-868. VRANA B, PASCHKE A, POPP P and SCHUURMANN G (2001) Use of semipermeable membrane devices (SPMDs): determination of bioavailable, organic, waterborne contaminants in the industrial regions of Bitterfield, Saxony-Anhalt, Germany. Environ. Sei. Pollut. Res. 8 27-34. VRANA B, KLUCAROVA V, BENICKA E, ABOU-MRAD N, AMDANY R, HORAKOVA S, DRAXTER A, HUMER F and GANS O (2014) Passive sampling: An effective method for monitoring seasonal and spatial variability of dissolved hydrophobic organic contaminants and metals in the Danube River. Environ. Pollut. 184 101-102. WANG J, BI Y, PFISER G, HENKELMANN B, ZHU K and SCHRAMM KW (2009) Determination of PAH, PCB and OCP in water from the Three Gorges Reservoir accumulated by semipermeable membrane devices (SPMD). Chemosphere 75 1119-1127. WILLET KL, ULRICH EM and HITES RA (1998) Differential toxicity and environmental fate of hexachlorocylohexane isomers. Environ. Sei. Technol 32 2197-2207. ZHANG ZL, HUANG J, YU G and HONG HS (2004) Occurrence of PAHs, PCBs and organochlorine pesticides in the Tonghui River of Beijing, China. Environ. Pollut. 130 249-261. http://dx.doi.Org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 APPENDIX TABLE AI Details of characteristic ions (m/z values) used in the analysis of polycyclic aromatic hydrocarbons in single ion monitoring (SIM) mode by GC/MS Compound Retention time (min) 'm/z 1 m/z 2 m/z 3 2Terphenyl 23.04 230 215 202 Naphthalene 8.37 128 129 126 Biphenyl 10.48 154 153 155 Acenaphthylene 11.49 152 153 150 Acenaphthene 11.91 154 153 155 Fluorene 13.3 166 167 164 Phenanthrene 16.42 178 179 176 Anthracene 16.61 178 179 176 Fluoranthene 21.13 202 203 200 Pyrene 22.09 202 203 200 Retene 23.46 219 234 205 Benzo [b]fluorene 23.9 216 215 217 Benzonaphthothiophene 26.39 234 235 232 Benzo [ghi]fluoranthene 26.58 226 227 224 Cyclopenta[cd]pyrene 27.45 226 227 224 Benzo [a] anthracene 27.49 228 229 226 Triphenylene 27.6 228 229 226 Chrysene 27.66 228 229 226 Benzo [b]fluoranthene 32.17 252 253 250 Benzo [j]fluoranthene 32.18 252 253 250 Benzo [k]fluoranthene 32.29 252 253 250 Benzo [e]pyrene 33.27 252 253 250 Bezno[a]pyrene 33.48 252 253 250 Perylene 33.8 252 253 250 Indeno [12 3cd] pyrene 38.39 276 277 274 Dibenzo [ah] anthracene 38.51 278 279 276 Dibenzo [ac] anthracene 38.52 278 279 276 Benzo [ghi]perylene 39.7 276 277 274 Anthanthrene 40.41 276 277 274 Coronene 50.13 300 301 298 3Dg-Naphthalene 8.37 136 137 134 3D10-Phenanthrene 16.33 188 189 184 3D12-Perylene 33.69 264 265 260 'The ion in the first column was used for quantification, the other two were used as qualifier ions to confirm compound identity instrumental internal standard ^Recovery internal standard http://dx.doi.Org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 TABLE A2 Details of characteristic MRM transitions (m/z values of parent and daughter ion are given) used in the analysis of PCBs and OCPs by GC/MS/MS Name Retention time (min) 'MRM transition (Quantifcation) MRM transition (Qualifier) 2PCB 121 20.24 325.9 > 255.9 327.9 > 255.9 3PCB 30 17.2 256>186 258>186 3PCB 185 27.28 393.8 > 323.9 395.8 > 325.9 PCB 28 17.13 256>186 258>186 PCB 52 18.15 289.9 > 220 291.9 > 220 PCB 101 21.05 325.9 > 255.9 327.9 > 255.9 PCB 118 23.27 325.9 > 255.9 327.9 > 255.9 PCB 153 23.95 359.8 > 289.9 361.8 > 289.9 PCB 138 24.91 359.8 > 289.9 361.8 > 289.9 PCB 180 27.22 393.8 > 323.9 395.8 > 325.9 PeCB 11.85 250 >215 252 >215 HCB 14.52 283.8 > 248.9 285.8 > 213.8 a-HCH 14.31 219> 183 181 >145 ß-HCH 15.1 219> 183 181 >145 y-HCH (Lindane) 15.29 219> 183 181 >145 6-HCH 16.19 219> 183 181 >145 o.p'-DDE 20.89 246 >176 318 > 248 p,p'-DDE 22.01 246 >176 318 > 248 o.p'-DDD 22.27 235 > 165 237>165 p,p'-DDD 23.52 235 > 165 237>165 o,p'-DDT 23.59 235 > 165 237>165 p.p'DDT 24.84 235 > 165 237>165 E-HCH 16.45 181 >145 219> 183 'The MRM transition in the first column was used for quantification, the other was used to confirm compound identity instrumental internal standard ^Recovery internal standard http://dx.doi.Org/10.4314/wsa.v40i3.5 Available on website http://www.wrc.org.za ISSN 0378-4738 (Print) = Water SA Vol. 40 No. 3 July 2014 ISSN 1816-7950 (On-line) = Water SA Vol. 40 No. 3 July 2014 Príloha 27 Amdany R., Chimuka L, Cukrowska E., Kukučka P., Kohoutek J., Tolgyessy P., and Vrana B., Assessment of bioavailable fraction of POPS in surface water bodies in Johannesburg City, South Africa, using passive samplers: An initial assessment, Environ. Monit. Assess., 2014, 186, 5639-5653. Environ Monit Assess (2014) 186:5639-5653 DOI 10.1007/sl0661-014-3809-3 Assessment of bioavailable fraction of POPS in surface water bodies in Johannesburg City, South Africa, using passive samplers: an initial assessment Robert Amdany • Luke Chimuka • Ewa Cukrowska • Petr Kukučka • Jiří Kohoutek • Peter Tolgyessy • Branislav Vraná Received: 2 November 2013 /Accepted: 6 May 2014 /Published online: 29 May 2014 © Springer International Publishing Switzerland 2014 Abstract In this study, the semipermeable membrane device (SPMD) passive samplers were used to determine freely dissolved concentrations of polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs) in selected water bodies situated in and around Johannesburg City, South Africa. The devices were deployed for 14 days at each sampling site in spring and summer of 2011. Time weighted average (TWA) concentrations of the water-borne contaminants were calculated from the amounts of analytes accumulated in the passive samplers. In the area of interest, concentrations of analytes in water ranged from 33.5 to 126.8 ng F1 for PAHs, from 20.9 to 120.9 pg F1 for PCBs and from 0.2 to 36.9 ng F1 for OCPs. Chlorinated pesticides were mainly composed of hexachlorocyclohexanes (HCHs) (0.15-36.9 ng F1) and dichlorodiphenyltrichloromethane (DDT) with its metabolites (0.03-0.55 ng F1). By applying diagnostic ratios of certain PAHs, identification of possible sources of the contaminants in the various sampling sites was R. Amdany • L. Chimuka (El) • E. Cukrowska Molecular Sciences Institute, School of Chemistry, University of the Witwatersrand, P/Bag 3, WITS, Johannesburg 2050, South Africa e-mail: luke.chimuka@wits.ac.za P. Kukucka • J. Kohoutek • B. Vrana Research Centre for Toxic Compounds in the Environment (RECETOX), Masaryk University, Kamenice 573/5, CZ-625 00 Brno, Czech Republic P. Tolgyessy Water Research Institute, Nabr. Arm. Gen. L. Svobodu 5, 812 49 Bratislava, Slovakia performed. These ratios were generally inclined towards pyrogenic sources of pollution by PAHs in all study sites except in the Centurion River (CR), Centurion Lake (CL) and Airport River (AUP) that indicated petrogenic origins. This study highlights further need to map up the temporal and spatial variations of these POPs using passive samplers. Keywords Free dissolved concentration • Passive sampling devices • Hydrophobic organic compounds • Monitoring • Passive sampling • SPMDs Introduction Water systems that have roots in urbanised areas are normally prone to severe contamination by an array of pollutants that include hydrophobic organic contaminants (HOCs). Pollution may be caused by current and/or previous industrial activities or both. Such water resources need to be secured for the benefit of current and future generations. Assessment of the pollution levels and distribution of the contaminants in water systems can be achieved by employing sound monitoring practices using a variety of available tools and techniques. Grab sampling has traditionally been applied in the determination of HOCs in water. However, successful monitoring is hampered by their existence at very low concentrations in water phase, in addition to frequent temporal changes. Increased sampling frequency, use of large sample volumes, installing automatic samplers and applying more sensitive analytical Springer 5640 Environ Monit Assess (2014) 186:5639-5653 techniques are possible solutions, but they come with cost implications. Passive sampling devices (PSDs) have shown promise as better alternatives since they permit unattended large volume- and time-integrated sampling, which compensate for fluctuating concentrations and also give lower detection limits (Harman et al. 2008; Vrana et al. 2014). Use of passive samplers is also advantageous because only the freely dissolved concentration of the analyte in water is sampled. This fraction of the contaminant is critical for the assessment of its bioavailability and fate in the aquatic environment and the risk associated with exposure of aquatic organisms to these contaminants. Among PSDs, the semipermeable membrane devices (SPMDs) have been successfully used as quantitative tools to assess concentrations of HOCs in the waters of various aquatic ecosystems (Huckins et al. 1993; Lu et al. 2002; Vrana et al. 2005, 2014). SPMDs passively accumulate lipophilic organic contaminants by mimicking biological membranes in its ability to allow selective diffusion of the compounds. Typically, organics with partition coefficients (log Kow) higher than 3 are suitable for extraction by this technique (Huckins et al. 1993; Vrana et al. 2005). In the current study, SPMDs were employed for the initial assessment of the bioavailable fractions of poly-cyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs) in the water columns of streams and rivers originating from Johannesburg City, South Africa. Further investigation of the pollution levels in water bodies that receive water from the urban streams and rivers was also undertaken. So far, very little studies have been reported that have investigated the presence of these POPs in water bodies in greater Johannesburg area, South Africa (Sibali et al. 2008; Sibiya et al. 2012, 2013a;). The reported studies have looked at the total concentrations and only for a few PAHs (Sibiya et al. 2012, 2013a) and organochlorine pesticides (Sibali et al. 2008). Materials and methods Chemicals and reagents The 16 US EPA PAH standards with purities>97 % pure were purchased from Sigma-Aldrich Chemie GmbH, (Steinheim, Germany). Performance reference compounds (PRCs): dio-acenaphthene, di0-fluorene, di0-phenanthrene and dio-pyrene, as well as recovery standards d8- naphthalene, dio-anthracene, di2-fiuoranthene, di2-benzo(a)anthracene, d12-benzo(k)fluoranthene, d12-benzo(g,h,i)pyrene, PCB 30, PCB 185 and d6-gamma hexachlorocyclohexane (HCH) were purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany). Internal standards (PCB 121 and terphenyl) for instrumental analysis were also purchased from Dr. Ehrenstorfer GmbH. Sulphuric acid (98 %) and hydrochloric acid (36 %) were purchased from Merck (Darmstadt, Germany). Triolein (97 %,) was purchased from Sigma Aldrich, (Ghent, Belgium), while silica gel 60 was from Merck (Darmstadt, Germany). High-purity (>99 %) ra-hexane, dichloromethane and trichloromethane were bought from Sigma-Aldrich (Prague, Czech Republic). Reagent water was drawn from a Milli-Q water system (Millipore, Bedford, MA, USA). SPMD sampler preparation and deployment Standard size SPMDs in the dimensions 2.54 x 91.4 cm, 460 cm2 external surface area were prepared from LDPE membranes (Brentwood plastics, MO, USA) and filled with 1 ml of high-purity triolein (97 % pure) to give a final total sampler volume of 4.95 ml. Initially, 100-cm long portions of the tubes were cut off from the roll before inserting them into pre-cleaned, dry glass bottles using a pair of blunt tweezers. Cleaning was done twice by soaking them overnight in hexane with the aim of removing organic contaminants. After air-drying, the membranes were heat-sealed at one end to form a loop. Each SPMD was spiked (a solution in n-hexane was spiked to SPMD using a GC syringe) with individual PRCs at a concentration of 2 ug sampler-1 (Vrana et al. 2014). The LDPE membrane was closed using a thermal seal (Impulse sealer ME-300 HI, Mercier Corporation). The devices were stored in airtight sealed metal cans at -20 °C awaiting deployment. Thereafter, the stainless steel housings containing the SPMDs were lowered about 40 cm below the surface of a river or dam bank. Its end was then tied to nearby branch of tree using a nylon string. At Hartbeespoort Dam (HD), the passive samplers were tied to the bottom of the floating bridges used for recreation purposes in the same way using a string. Samplers were deployed for 14 days; the more commonly used deployment time (Vrana et al. 2014). Sites for sampler deployment were based on our previous study in the same area (Sibiya et al. 2013b) and are described in Table 1. The sampling points are also shown in Fig. 1. Springer Environ Monit Assess (2014) 186:5639-5653 5641 Table 1 Description of sampling sites Sampling site Symbol Water body Longitude Latitude Ifafi (Hartbeespoort Dam) IFA Dam 25° 45'09.97" S 27° 53' 04.39" E Juskei River 1 JR 1 River 26° 01' 07.49" S 28° 05' 34.69" E Juskei River 2 JR2 River 26° 00'25.30" S 28° 04'45.74" E Eagles (Hartbeespoort Dam) EGL Dam 25° 44' 56.29" S 27° 50' 06.18" E Homestead Lake HSL Dam 26° 10'25.64" S 28° 17'04.71" E Airport River AUP River 26° 08' 29.74" S 28° 17'04.71" E Centurion Lake CL Dam 25° 51' 55.41" S 28° 12' 23.36" E Centurion River CR River 25° 51' 40.01" S 28° 11'22.93" E Sampling sites Jukskei Rivers as the densely populated Alexandra Township. Two sampling sites were chosen along the Jukskei River (JR 1 and JR 2) for the deployment of SPMDs. The Jukskei River (JR) is one of the river catchments in the Johannesburg metropolis that covers over 800 km2 (Campbell 1996). The source of the river can be traced to the Bruma Lake situated at the foot of the Witwatersrand area. The river eventually empties its water into the Hartbeespoort Dam after merging with the Crocodile River downstream. The Jukskei River (JR) meanders northwards through a number of residential areas such Centurion Lake and Centurion River The Centurion Lake (CL) and Centurion River (CR) form part of the Hennops, a relatively small perennial river, that originate from a marshy area situated a few kilometres east of Kempton Park, Johannesburg (Torien and Walmsley 1979). Further downstream, the river receives treated effluent from a number of wastewater Crocodile River Key I Sampling site • Towns Megalies River Fig. 1 Map of the sampling sites Springer 5642 Environ Monit Assess (2014) 186:5639-5653 treatment plants. Kempton Park area is home to several industries that release their effluent into the river via the treatment plants. The Centurion Lake is surrounded by vibrant business installations such as shopping malls, hotels, recreational facilities and garages. Hartbeespoort Dam Two sites were located in the Hartbeespoort Dam: Ifafi (JFA) and Eagles (EGL). The dam is located about 37 km west of Pretoria and along the Crocodile River in North West Province of South Africa. The water body is a 20 km2 water reservoir sandwiched between the Megaliesberg mountain range in the Highveld region of northern South Africa (Hely-Hutchinson and Schumann 1997; Nyoni et al. 2011). The reservoir receives water from an approximately 4,100 km2 area all the way from Johannesburg City via the Jukskei and Hennops Rivers that flow into the Crocodile River. This accounts for about 90 % of the dam's water inlet supply with rain water being the major source (Harding et al. 2004). The five catchment basins of the dam are from west to east, the Megalies/Skeerpoort, the Crocodile, the Jukskei, the Hennops and the Swartspruit basin (Van Rei 1987). Homestead Lake and Airport River Both Homestead Lake (HSL) and Airport River (AUP) are located about 28 and 35 km, respectively, east of Johannesburg City central business district and a few kilometres from the O.R. Tambo international airport. The origin of the AUP is a swampy area situated at the periphery of the airport. It flows downstream in an easterly direction, passing through a few residential areas before discharging its water into a man-made reservoir called HSL. This dam is surrounded by many residential developments. The HSL also receives water from a small stream originating from the southern part of the airport and about 5 km west of the dam. After exiting the dam, the river flows towards the east rand area of Johannesburg. Extraction of SPMDs Debris, particulates and biofouling were removed from the surfaces of retrieved SPMDs using a stream of tap water before briefly immersing in diluted (10 %) hydrochloric acid to rid them of adsorbed carbonates. Further washing of the samplers with deionised water and air-drying at room temperature followed before placement of each device in a 250-ml glass container with a ground joint glass stopper. One hundred milliters of ra-hexane was added into each container to fully immerse the SPMD and spiking with surrogate standards in hexane (0.5 Lig sampler-1 of each compound) done. Dialytic extraction of analytes was carried out over a 24-h period at room temperature and in the dark. After this period, dialy-sates were transferred into clean, labelled glass containers and fresh batches of 100 ml of ra-hex-ane added to the samplers and the process repeated. The extracts were combined and reduced to about 10 ml at 40 °C using a rotary evaporator (Heidolph Laborata 4000, Germany) and further concentrated to the last drop with a gentle stream of nitrogen gas and, thereafter, reconstituted in 1 ml of trichloromethane. Further processing by gel permeation chromatography (GPC) was done to remove triolein and sulphur contaminants prior to instrumental analysis. One thousand microliters of the extract was introduced into a GPC system equipped with a high-pressure pump (HPP5001) and a fraction collector (ECOM, Prague, Czech Republic) and fractionation achieved using a Gel 5 \xm 50 A, 7.5 x 300 mm, column (Agilent PL). Dichloromethane acted as the mobile phase flowing at 0.6 ml min-1. Target analytes were collected in the fractions that eluted as from 18.3 to 41.7 min. Prior to eluent volume reduction to near dryness using nitrogen gas, a solvent keeper (0.1 ml of ra-nonane) was added. The sample was finally reconstituted in 1 ml of ra-hexane and subjected to activated silica gel cleanup. For PAHs, each column was packed with 5 g of activated silica gel (prepared by drying at 120 °C for 8 h). Conditioning of the column was done by flushing it with 10 ml of ra-hexane and the analytes of interest eluted using 20 ml of dichloromethane after sample introduction. The eluate was evaporated to 10 ml at 40 °C and reduced further to 1 ml with nitrogen gas. Sulphuric acid-modified activated silica gel (mixture of 50 g freshly prepared activated silica gel and 33 ml of concentrated sulphuric acid, 98 %) was used to clean PCBs and OCPs. Subsequent elution of the analytes was done with 30 ml dichloromethane. After evaporation and further concentration to 1 ml, terphenyl or PCB 121 internal standards were added to the samples. GC-MS analysis of the compounds followed. Springer Environ Monit Assess (2014) 186:5639-5653 5643 Instrumentation Prior to analysis of all samples, calibration using standards with concentrations ranging from 0.01 to 0.200 |ig ml-1 was done. For PAHs, separation was achieved using a HP-5MS capillary column with the dimensions 30 mx0.25 mm internal diameter, 0.25 |im film thickness and helium as the carrier gas flowing at 1.9 ml min"1. Working conditions: splitless injection of 1 |il sample at 250 °C. Column temperature programme 70 °C (0.5 min hold) then ramped at 25 °C min"1 to 150 °C followed by 3 °C min"1 to 200 °C and finally increased at 8 °C min"1 to 280 °C and held for 20 min. Detection of separated analytes was done using a 5971 MS system (Agilent Technologies, USA) set at 320 °C and 70 eV electron impact ionisation. Selected ion monitoring mode was used in the measurements and two to three characteristic ions were chosen for detection and quantification of each compound. The ion source, the transfer line and the quadrupole temperatures were maintained at 230, 280 and 150 °C, respectively. Using external calibration methods, analyte concentrations in the samples were calculated based on the peak areas of the highest characteristic ion in the mass spectrum of the compounds. Recoveries of surrogate standards introduced into the sampler containers prior to dialytic extraction were used to correct such concentrations. Analysis of PCBs and OCPs in the samples was done on a 6,890 N GC (Agilent technologies, Santa Clara, USA) linked to a Quattro MicroGC MS (Waters, Micromass, UK) operated in EI+ mode was used. Chromatographic separation of target analytes was achieved using a 60 mx0.25 mmx 0.25 urn DB5-MS column (Agilent J&W, USA) with the column flow rate of carrier gas (helium) maintained at 1.5 ml min"1. The inlet was operated in the splitless mode at 280 °C and 1 |il sample loaded. For each compound analysed, a minimum of two MRM transitions were recorded. The column temperature was initially 80 °C and held for 1 min, then ramped at 15 °C min"1 to 180 °C and finally 5 °C min"1 to 300 °C (5 min). TargetLynx software (Waters, Micromass, UK) was applied in processing the raw data. Contaminations that may have occurred during sampler fabrication, deployment and/or retrieval were corrected using fabrication and field blanks. Water-dissolved concentrations of compounds Since amounts of analytes absorbed by the samplers follow a first-order approach to equilibrium, water-dissolved concentrations were determined from the quantities sequestered (Ns) by SPMDs, compound-specific in situ sampling rates (i?S) and their sampler-water partition coefficients (Ksw): w K^Vsll-expi-Rst/iK^Vs))] y ' where FS=SPMD volume and ^exposure time. The dissipation of PRCs also obeys first-order kinetics. The nonlinear least squares method (Booij and Smedes 2010) was adopted in the estimation ofi?s based on the fraction (J) of individual PRCs that remained in the SPMD after exposure as a continuous function of their _rvsw, with Rs as an adjustable parameter. where, /=A^prc/A^o,prc and A^0,prc=PRCs at t=0, 7VpRC=PRCs at t=l4 days. With the assumption that uptake is controlled by the aqueous boundary layer, Eq. (3) (Rusina et al. 2010) was substituted in Eq. (2) enabling the estimation of sampling rates of individual compounds in the higher hy-drophobicity range. Rs = FAM~°A7 (3) where M=molar mass of compound, A=SPMD surface area (460 cm2) and F regression coefficient that was optimised using nonlinear least squares method for estimating sampling rates. ^sw values were intrapolated from the empirical Eq (4) (Huckins et al. 2006). Log Kow values were obtained from various literatures (Vrana et al. 2014). LogKsw = -01618(logKovl,)2 + 2.3211og£OM,-2.61 (4) Results and discussion Occurrence of PAHs, PCBs and OCPs in the SPMDs Amounts of PAHs accumulated in SPMDs after the 14-day exposure period in the various sampling sites are shown in Table 2, while for PCBs and OCPs are presented in Table 3. Field blank SPMDs were devoid of Springer 5644 Environ Monit Assess (2014) 186:5639-5653 Table 2 Mean concentrations of PAHs in SPMDs (ng sampler ') sequestered from different sample sites (n=3) Compound Sampling site IFA JR 1 JR 2 EGL HSL AUP CL CR Naphthalene 268±41 247±56 262±16 283±61 389±88 307±26 338±71 148=±=22 Acenaphthylene 156±30 156±18 158±29 150±12 164±26 197±28 132±11 55±4 Acenaphthene 34±5 38±7 124±6 32±7 38±9 334±6 47±4 74±8 Fluorene 84±15 77±4 355±8 65±11 82±20 71 ±11 36±8 213=±=22 Phenanthrene 144±12 246±5 1,192±97 132±15 138±18 174±18 1,220±98 425=±=13 Anthracene 308±14 287±83 198±11 102±14 49±10 179±22 121±9 31=1=5 Fluoranthene 127±10 424±47 999±35 156±37 61±15 83±21 118±3 260=±=19 Pyrene 96±13 367±38 825±91 115±12 42±5 97±19 279±35 260±22 Benz[a]anthracene 32±8 62 ±7 119±6 32±2 29±0 31±2 189±14 18=t=4 Chrysene 44±9 133±18 203±28 46±8 34±4 42±5 11±2 31=t=4 B enzo [bjfluoranthene 38±3 57±4 86±4 34±1 39±6 36±2 17±4 9±0 B enzo [k] fluoranthene 25±5 62±3 86±17 38±4 37±2 71±5 5±1 5±1 Benzo[a]pyrene 30±4 37±13 55±4 79±12 48±6 ND 11±1 4±1 Indeno[ 1,2,3-cd]pyrene 38±3 41±2 48±10 38±5 37±3 38±0 4±1 8±0 Dibenz[a,h]anthracene ND ND ND ND ND ND ND ND B enzo [ghijperylene 40±4 45±3 56±5 36±2 36±5 37±2 18=1=3 ND EPAHs 1,464±60 2,279±121 4,766±147 1,335±79 1,223±98 1,394±57 2,543±128 1,541 ±46 ND not detected quantifiable amounts of PAHs, PCBs and OCPs. Prior to dialytic extraction of analytes in the samplers, recovery standards were introduced. Information obtained from their recoveries was then used to adjust the concentrations of the target compounds. Compounds of interest showed good recoveries that ranged from 55 to 115 % for PAHs, 76 to 103 % for PCBs and 69 to 111 % for OCPs. Comparable quantities of the analytes were obtained from triplicate SPMDs deployed in the same site. Relative percent differences between such replicates in all sites were not greater than 25 % for PAHs, except for anthracene at sampling site JR 1 (29 %) and benzo[k] fluoranthene at sample site AUP (34 %). For PCBs, these differences did not exceed 25 %. Save for a-HCH recorded in sampling site JR 2 (34 %), all OCPs exhibited a relative percent difference of less than 21 % between replicates. In this study, summed-up amounts of SPMD-sequestered analytes ranged from 1,223 to 4,766 ng sampler-1 for PAHs, 6.5 to 76.1 ng sampler-1 for PCBs and 4.5 to 921.6 ng sampler-1 for OCPs. The chlorinated pesticides were primarily composed of HCHs (3.0 to 870.0 ng sampler-1) and dichlorodiphenyltrichloromethane (DDT) and its metabolites (1.5 to 86.5 ng sampler-1). Springer The use of PRCs (D-PAHs) for other compounds other than PAHs is justifiable. Since hydrophobic compounds with log Kow>4 are accumulated in SPMD under water boundary layer control (WBL), sampling rate is determined by diffusion in water. Diffusion coefficients in water of PAHs, PCBs and OCPs are a weak function of molecular weight (Eq. 3) (Rusina et al. 2010). Since diffusion in water is assumed not to be affected by parameters other than molecular volume/weight, sampling rates of all compounds can be estimated using Eq. 3. Water-dissolved concentrations of the analytes Rs values for individual compounds were determined using Eq. 3. Table 4 presents the PRC-derived Rs values for fluorene resulting from SPMD field deployment at the various sampling sites. These values ranged from a low of 1.0 1 d-1 at HSL in October 2011 to a high of 26.1 1 d-1 at JR 2 in December 2011. Differences in Rs values at different sites may be attributed to variations in water flow velocities, in agreement with assumption of water boundary layer uptake. Although no flow velocities were measured, dams generally had much lower Environ Monit Assess (2014) 186:5639-5653 5645 Table 3 Mean concentrations of PCBs and OCPs in SPMDs (ng sampler ) recorded in the different sample sites (n=3) Sampling site Compound IFA JR 1 JR2 EGL HSL AUP PCBs PCB 28 PCB 52 PCB101 PCB 118 PCB 138 PCB 153 PCB 180 EPCBs OCPs HCHs a-HCH (3-HCH y-HCH 6-HCH E-HCH EHCHs DDX o,p'-DDE p,p'-DDE o,p'-DDD p,p'-DDD o,p'-DDT p,p'-DDT SDDX EOCPs 34.3±5.5 1.2±0.2 0.6±0.1 0.2±0.0 0.6±0.0 0.6±0.1 0.7±0.1 38.2±0.5 44.4±8 127.9±16 7.0±1.1 2.6±0.4 6.7±1.3 188.5±10 0.7±0.1 7.6±1.3 20.3±4.8 55.8±3.4 1.5±0.0 0.6±0.1 86.5±6.0 275±12 44.1±7.8 2.3±0.5 1.1±0.3 0.5±0.1 0.9±0.2 1.2±0.1 0.5±0.1 50.6±0.9 120.4±41 626.8±98 16.5±3.7 46.5±4.0 11.4±1.9 821.6±107 0.8±0.2 8.0±1.6 3.2±0.5 17.7±3.0 3.0±0.7 11.1±2.6 43.8±4.4 865 4± 107 68.1±12 2.4±0.4 l.liO.l 0.8±0.3 1.3±0.2 1.4±0.2 1.0±0.1 76.1±0.6 131.5±26 630.2±82 13.6±2.2 65.9±134 28.8±5.6 870.0±87 1.5±0.2 14.4±2.7 5.2±0.5 26.6±L4 1.8±0.4 2.1±0.2 51.6±3.1 921.6±88 32.0±6.3 0.8±0.1 0.3±0.0 0.2±0.0 0.5±0.1 04±0.1 04±0.0 34.6±0.6 43.2±7 114.1±20 5.9±1.2 2.3±0.0 7.5±1.3 173.0±21 0.6±0.1 5.5±1.0 28.6±5.5 47.2±8.1 1.8±0.3 0.6±0.0 84.3±9.8 257.3±23 5.9±1 0.3±0.0 0.1 ±0.0 ND 0.2 ND ND 6.5±0.1 0.8±0.2 0.8±0.1 1.4±0.2 ND ND 3.0±0.3 0.1 ±0.0 0.5±0.1 0.1 ±0.0 0.5±0.0 0.3±0.0 ND 1.5±0.1 4.5±0.3 3.2±0 2.4±0 2.0±0 ND 3.1±0 5.1±0 4.3±0 20.1±0 1.2±0 0.7±0 1.2±0 ND ND 3.1±0 0.1±0 2.1±0 0.6±0 1.5±0 0.5±0 0.6±0 5.4±0 8.5±0 ND not detected All summed numbers are in italic velocities compared to those samplers deployed on river banks. Other contributors to differences in sampling rates include temperature, biofilm infestation and deposition of particulates on the surface of sampler (Baxter 1990; Cailleaud et al. 2007; Brandli et al. 2008; Booij and Smedes 2010). South Africa has got four seasons with summer starting from mid-October to mid-February and is very hot characterised by afternoon thunderstorms. Autumn is from mid-February to mid-April with little rain and not very hot. Winter starts from May to July, while spring is from August to mid-October. Most of the sampling was done in spring (Table 4) where there is little or no rainfall and is beginning to get hot. :2 :\ '.0 '.1 '.1 '.0 :2 '.3 '.1 :2 14 1.0 1.4 1.1 1.2 i.l i.l 15 1.6 PAHs Estimation of free dissolved concentrations of the PAHs in water based on the amounts accumulated in deployed SPMDs are presented in Table 5. Since the sampling was not done at the same time and season, it is not easy to compare the results for spatial trends. Total analyte concentrations by site varied from 22.1 ng F1 at RC to 126.7 ng F1 at HSL. The high concentration of PAHs at this site could be linked to previous reported oil spill in the upstream of the Airport River. Airport River (AUP) flows into the Homestead Lake, and this suggests that it is acting as a recipient of PAHs. In the same way, the concentrations of PAHs in the Centurion Lake were Springer 5646 Environ Monit Assess (2014) 186:5639-5653 Table 4 Description of the sampling campaign at the sites Sampling site Season Exposure period Exposure (days) Water temperature (°C) SPMD-Sampling rate Rs (1 O Start End IFA Spring 2 September 2011 16 September 2011 14 19 19.2 JR 1 Summer 3 December 2011 17 December 2011 14 21 19.4 JR2 Summer 3 December 2011 17 December 2011 14 21 26.1 EGL Spring 2 September 2011 16 September 2011 14 19 18.3 HSL Spring 6 October 2011 20 October 2011 14 20 1.0 AUP Spring 06 October 2011 20 October 2011 14 20 20.0 CL Spring 12 August 2011 26 August 2011 14 18 10.7 CR Spring 12 August 2011 26 August 2011 14 18 18.8 much higher than those in the Centurion River. This again may suggest that the dam is acting as a recipient and perhaps a sink of PAHs. A study of PAHs in sediments in the same area found high concentration levels (Sibiya et al. 2013b). The concentration of PAHs in sediments at the Centurion Lake ranged from 61 to 1,690 Ltg kg-1 and 84 to 1,545 |ig kg-1 at the Homestead Lake (Sibiya et al. 2013b). The reported concentrations in Table 5 for PAHs are comparable to those reported by Karacik et al. (2013) (8.36-76.68 ng F1) and Wang and coworkers (2009) (19.14-97.17 ng F1). They were also comparable to those obtained by Vrana et al. (2014) in the Danube River (13-72 ng L-1). However, they were significantly higher than what Allan and Ranneklev (2011) (0.033-9.3 ng F1) obtained in the Alna River, Norway. Evidently, water phase PAH concentrations of individual compounds (Fig. 2) generally reflected the trend exhibited by the cumulative concentrations at any given sampling site. The smaller molecular weight compounds (< four rings) accounted for the highest percentage (77.6 % at HSL to 96.5 % at AUP) of total PAHs in the water phase. Their relatively higher water solubilities as indicated by lower log A"ow values enhance their availability and, hence, uptake by the samplers. On the other hand, strong hydrophobicity of larger molecular weight PAHs encourages increased sorption to larger particulates and colloids in the water column resulting in diminished availability. PCBs Freely dissolved PCB levels in the waters of the various sampling sites are presented in Table 5. Estimated water Springer phase concentrations were in the low picograms per liter range. Sum of seven indicator PCB congeners varied between 21 pg F1 at AUP and 121 pg F1 at HSL. These values were about three orders of magnitude lower than for PAHs and up to two orders lower than OCPs. Of the many PCB congeners known, seven of them (PCB 28, PCB 52, PCB 101, PCB 118, PCB 138, PCB 153 and PCB 180) were quantifiable in most of the sites, with the less-chlorinated PCBs predominating (up to 89 %). Figure 3 presents concentrations of some of the PCB congeners. Residue levels obtained were lower than those reported by Allan and Ranneklev (2011), Liu et al. (2013) and Cailleaud et al. (2007) but generally comparable to those obtained by Wang et al. (2009) (66-519 pg F1). PCBs enter the environment mainly through volatilisation from in-use and disposed equipment or as re-emissions from soils (Wang et al. 2007). Although these compounds were never produced in South Africa, PCB oils as well as equipment containing such oils were imported for use mainly for electricity generation and in manufacturing industries. However, as in many other countries, PCBs are currently outlawed in the country, and their presence in the environment is attributed to previous applications, since these compounds are persistent organic pollutants. Old electricity transformers contained PCBs, but these are now being phased out by Eskom, a South African electricity generation and supply company. Although sampling sites IFA and EGL are located in an area devoid of major industrial activities, it still recorded significant quantities (33.6 and 27.1 pg F1, respectively) of the contaminants. Apparently, most of Environ Monit Assess (2014) 186:5639-5653 5647 Table 5 Estimated dissolved water concentrations (ng 1 ') of PAHs, PCBs and OCPs at the various sampling sites Compound Sampling site IFA JR 1 JR 2 EGL HSL AUP CL CR PAHs Naphthalene 37.205 34.705 36.813 38.92 61.185 43.135 47.491 20.795 Acenaphthylene 3.832 3.795 3.828 4.084 13.307 4.777 3.279 1.338 Acenaphthene 1.037 1.154 3.764 0.941 3.194 1.017 1.439 2.247 Fluorene 1.349 1.157 5.280 0.905 6.422 1.063 0.595 3.200 Phenanthrene 1.198 1.699 7.487 0.900 10.465 1.171 11.773 2.942 Anthracene 2.838 1.656 1.306 0.581 3.743 0.918 1.200 0.224 Fluoranthene 0.755 1.938 3.757 0.483 4.753 0.363 0.909 1.194 Pyrene 0.579 1.694 3.152 0.429 3.265 0.427 2.164 1.207 Benz[a]anthracene 0.185 0.276 0.423 0.130 2.409 0.131 1.470 0.081 Chrysene 0.256 0.592 0.726 0.179 2.822 0.178 0.086 0.139 Benzo[b]fluoranthene 0.268 0.263 0.317 0.151 3.370 0.157 0.139 0.042 Benzo[k]fluoranthene 0.141 0.284 0.313 0.198 3.182 0.319 0.041 0.023 Benzo[a]pyrene 0.195 0.171 0.200 0.189 2.023 ND 0.089 0.019 Indeno [1,2,3 -cd]pyrene 0.236 0.196 0.182 0.172 3.361 0.174 0.034 0.039 Dibenz[a,h]anthracene ND ND ND ND ND ND ND ND Benzo[ghi]perylene 0.247 0.214 0.212 0.165 3.27 0.167 0.152 ND EPAHs 50.32 49.79 67.76 48.43 126.78 53.99 70.86 33.49 OCPs HCHs a-HCH 1.884 5.106 5.576 1.832 0.092 0.051 (3-HCH 5.737 28.103 28.254 5.116 0.094 0.031 y-HCH 0.371 0.874 0.721 0.313 0.170 0.064 6-HCH 0.046 0.790 1.095 0.039 ND ND e-HCH 0.301 0.511 1.291 0.336 ND ND EHCHs 8.339 35.385 36.937 7.635 0.356 0.146 DDX o,p'43DE 0.004 0.004 0.006 0.003 0.010 ND p,p'-DDE 0.048 0.040 0.055 0.026 0.048 0.010 o,p'43DD 0.129 0.016 0.020 0.133 0.010 0.003 p,p'-DDD 0.354 0.090 0.102 0.227 0.048 0.007 o,p'43DT 0.010 0.016 0.007 0.009 0.030 0.003 p,p'-DDT 0.004 0.059 0.008 0.003 ND 0.003 EDDTs 0.549 0.225 0.198 0.401 0.146 0.026 SOCPs 8.888 35.610 37.135 8.036 0.502 0.172 PCBs EPCBs 0.034 0.058 0.074 0.027 0.121 0.021 ND not detected All summed numbers are in italic the water at the sites is supplied through the Crocodile of the two rivers can be traced to the outskirts of River (Harding et al. 2004) whose major tributaries Johannesburg City—a probable source. This is especial-include the Jukskei and Hennops Rivers. The origins ly reinforced by the closeness in the estimated Springer 5648 Environ Monit Assess (2014) 186:5639-5653 contaminant concentrations at IFA (34 pg F1) and EGL (27 pg F1) sampling sites (both in the same water body). The lighter-molecular-weight PCB congeners are usually more prone to atmospheric transport (Ockenden et al. 2003) and volatilisation (Wang et al. 2007). However, presence of larger-molecular-weight PCBs such as the dioxin-like PCB 118 (Quinn et al. 2009) in the water body is likely a result of previous application in the immediate surrounding area. Typically heavier PCB molecules are known to deposit close to the main source, resulting in relatively increased levels in the areas, even decades after initial use, whereas their lighter counterparts can travel for long distances from their sources. Although the heavier congeners — 700 _1 be a. 500 ->-- c 500- o - 400- c o 300- u c o :oo u 100- 0 PCB 28 PCB 52 HSL AUP Springer Environ Monit Assess (2014) 186:5639-5653 5649 Fig. 4 Percent composition of HCH in water of the several sample sites I a-HCH I B-HCH V-HCH : 6-HCH E-HCH Bi c "5. E re AUP HSL JR2 J31 EGL 40% 60% Percentage (%) 100% were found at very low concentrations in the water phase, their high lipophilicity and biomagnification effects through the food web may be a cause of concern (Degger et al. 2011). OCPs Estimated ambient water concentrations of OCPs in the various sampling sites are shown in Table 5. Sum total OCP concentrations ranged from 0.172 ng F1 at AUP to 37.135 ng F1 at JR 2. The principal compositions of these compounds in all the sites were HCH and DDX (DDT, DDD and DDE). HCHs levels were higher than those of DDX with SHCHs ranging from 0.146 to 36.937 ng F1 and SDDX varied between 0.026 and 0.549 ng F1. Concentrations of individual isomers generally followed a similar trend as the totals. HCH concentrations from this study were in agreement with those reported by Luo et al. (2004) (5.7-23.3 ng F1) but higher than those obtained by Wang et al. (2009) (0.10-0.63 ng F1). Considering all the study sites, mean water-borne concentrations of individual HCH isomers increased in the order: 6-HCH (2.134 ng F1), e-HCH (2.250 ng F1), Y-HCH (2.446 ng F1), a-HCH (12.990 ng F1) and (3-HCH (68.998 ng F1), (Fig. 4), with a- and (3-HCHs accounting for over 90 % of the totals. In all cases, the |3-HCH predominated. This isomer is characterised by higher water solubility, lower volatility and stronger environmental stability to physical, chemical and biological degradation (Willet et al. 1998; Wang et al. 2009). Concentrations of a-HCH were also higher than for y-, 6- and £-isomers. Technical-grade HCH was widely used as an insecticide, and together with its accompanying isomers, it is still readily found in the environment (Wu et al. 1997). Since this type of HCH contains between 60 and 70 % a-HCH (Li and Macdonald 2005), it is expected that for every quantity of the pesticide used (and eventually ending up in the environment), a big percentage constitutes a-HCH. Moreover, its relative volatility aids in long-range transportation to regions afar. However, the lower concentrations (than a- or (3-isomers) ofy-HCH observed in all the sampling sites suggest no recent applications of the insecticide in the catchment areas of the water systems. As a signatory of the Stockholm convention, South Africa has phased out the production and use of these compounds. Sampling sites JR 1 and JR 2, both of which are found in the Jukskei River, recorded the highest water-dissolved HCH concentrations. However, a slight variation in contaminant levels between them was witnessed going downstream (from JR 1 to JR 2). Since the origin of the Jukskei River is very close to Johannesburg City, the high levels of the contaminants recorded may thus be related to previous agricultural activities. Most of the developed parts of Johannesburg long the Jukskei River were previous farms which were later sold and developed for industrial and residential properties. Further downstream (JR 2), the marginal increase in HCH levels is likely resulting from additional input from also previous agricultural activities. Site IFA recorded slightly higher water-dissolved SHCH concentrations than EGL, despite the two sites being in the same water body Springer 5650 Environ Monit Assess (2014) 186:5639-5653 Fig. 5 Percent composition of ■ o.P'-DDE ■ p.p'-DDE ■ ojP'-DDD ■ p,p'-DDD o,p'-DDT p,p'-DDT DDX in water of the various sample sites Percentage (%) (Hartbeespoort dam). The closer proximity of the Crocodile River's entry point into the dam to IFA than to EGL (Fig. 1) may explain this discrepancy. The summed-up concentrations of estimated free dissolved DDX in each sampling site are presented in Table 5. SDDX ranged from 0.026 ng F1 in AUP to 0.549 ng F1 in IFA. Total DDX concentrations were at most two orders of magnitude lower than HCH values. IFA and EGL recorded significantly higher contaminant concentrations (at least twofold) compared to all the other sites. DDX levels obtained in this study were comparable to those reported by Quemerals et al. (1994) and Wang et al. (2009) but lower by several orders of magnitude than those reported by Karacik et al. (2013) and Rajendran et al. (2005). Ultraviolet radiation as well as microbial activity can degrade DDT to its metabolites, DDD and DDE. These degradation products are representative of historic use of DDT (Wang et al. 2009). The percent contributions of DDT and its metabolites in each of the sampling sites are shown in Fig. 5. Evidently, most of the DDX existed as DDD and DDE, with the former constituting the highest percentage in the majority of sites (IFA=86.8 %; JR 1 = 47.4 %; JR2=62.0 %; EGL=88.3 %; HSL=38.6 %; AUP=41.7 %). This was more pronounced at IFA and EGL sampling sites. Two inferences can be made from these observations. Firstly, contamination of the sites is mainly due to past use of DDT and the contribution of current application appears limited. Secondly, reductive dechlorination mechanisms (Baxter 1990; Wedemeyer 1966) of DDT are more prevalent in the studied aquatic systems and, especially, at IFA and EGL. Sibali et al. (2008) is also reported to have looked at the level of organochlorine pesticides along the Juskei River in Johannesburg and including the Hartbeespoort Dam (HD). Soxhlet extraction was used for solid samples, Fig. 6 PAH cross plots for the ratios Ant/(Ant + Phe) vs Flt/(Flt + Pyr) Petrogenic Pyrogenic 0.65 i 0.5 $ 0.55 i £h 0.45 £ 0-4 0.35 0.3 ♦E.C LC HSL ♦ HD JR2 ♦ EGL JÖR.1 ♦ ♦ ♦ A.TJP ♦ 0.2 0.4 0.6 AJNT/ANT+PHE 0.8 Grass, wood and coal combustion Petroleum combustion Petroleum Springer Environ Monit Assess (2014) 186:5639-5653 5651 while water samples were extracted with liquid-liquid extraction; both techniques determines the total concentration. The concentrations of these pesticides were much higher in sediments, mostly in three-digit micrograms per kilogram levels while in water were mostly single- and double-digit micrograms per kilogram levels. High concentration in the sediment indicate accumulation from previous use. Re-desorption processes could be contributing to what is observed in water bodies. Possible sources of PAHs PAHs enter the environment through two major pathways: pyrogenic or petrogenic sources. Water-dissolved concentrations of the analytes have been used to predict their probable sources by utilising molecular ratios of certain PAHs (Yunker et al. 2002; Zhang et al. 2004; Brandli et al. 2008; Allan and Ranneklev 2011). Specifically, variations in the ratios of the thermody-namically less-stable PAHs are used as indices in apportioning such sources. For passive samplers, Allan and Ranneklev (2011) suggest that ratios of PAHs must be for compounds with near to identical sampling rates so as to minimise bias arising from the mode of calculation of the sampling rates for compounds with widely differing log Kow values. In the current study, source apportionment of PAHs in each sampling site was attempted using ratios of Anthracene/(Anthracene + Phenanthrene) [(Ant/(Ant + Phe))] against Fluoranthene/(Fluoranthene + Pyrene) [(Flt/(Flt + Pyr))]. As shown in Fig. 6, the majority of the sites sampled gave Flt/(Flt + Pyr) ratios that were greater than 0.5, indicating pyrogenic origins. This may have occurred through combustion of biomass and coal. PAHs at sampling sites AUP and CL were clearly inclined towards petroleum combustion sources. A small ratio (<0.1) of Ant/(Ant + Phe) indicates dominance of petrogenic sources. It is proposed that since AUP is in very close proximity to a busy international airport, aircraft, vehicular and other machinery exhausts may be finding their way into the waters of the sampling site. PAHs detected at sites CL and CR also seem to point to petrogenic origins probably due to the myriad of human activities taking place in the immediate surroundings of this busy urban setup. In general, the use of Ant/Ant + Phe ratio in differentiating petrogenic from pyrogenic sources is at times limited by photolytic degradation of anthracene which can result in lowered ratios (Kamens etal. 1988; Liuetal. 2009). Conclusions Passive sampling devices and SPMDs in particular are potentially viable tools in determining water-dissolved concentrations of HOCs such as PAHs, PCBs and OCPs in water systems. The free dissolved fraction of organic micropollutants is critical in terms of their bioavailability and ecotoxicological impacts on aquatic organisms. In this study, though still an initial assessment, three classes of HOCs, PAHs, PCBs and OCPs, were quantifiable in all the study sites. PAH levels were at least one and two orders of magnitude higher than OCPs and PCBs, respectively Using molecular ratios of certain PAHs, the identification of probable sources of the contaminants in the water phase was attempted. The findings indicated an inclination towards pyrogenic origins in all sample sites except at AUP and CL and CR, which indicated petroleum combustion sources. In general, our study has significant importance in providing basic preliminary data of POP pollution in water systems situated in and around Johannesburg City, South Africa. Acknowledgments The authors appreciate the technical assistance rendered by Lenka Vaňková of RECETOX, Masaryk University, Czech Republic, and for financial support from the Czech Ministry of Education of the Czech Republic (LM2011028 and L01214), National Research Foundation (NRF) and Water Research Commission (WRC) of South Africa References Allan, I. J., & Ranneklev, S. B. (2011). Occurrence of PAHs and PCBs in the Alna River, Oslo (Norway). Journal of Environmental Monitoring, 13, 2420-2426. Baxter, R. M. (1990). Reductive dechlorination of certain chlorinated organic compounds by reduced hematin compared with their behaviour in the environment. Chemosphere, 21, 451-458. Booij, K., & Smedes, F. (2010). An improved method for estimating in situ sampling rates of nonpolar passive samplers. Environmental Science and Technology, 44, 6789-6794. Brandli, R. C, Bucheli, T. D., Ammann, S., Desaules, A., Keller, A., Blum, F., et al. (2008). Lipid containing semipermeable membrane devices for monitoring organic contaminants in water. Journal ojEnvironmental Monitoring, 10, 1278-1286. Springer 5652 Environ Monit Assess (2014) 186:5639-5653 Cailleaud, K., Forget-Leray, J., & Souissi, S. (2007). Seasonal variations of hydrophobic organic contaminant concentrations in the water-column of the Seine Estuary and their transfer on a planktonic species Eurytemora affinis (Calanoida, copepoda). Parti: PCBs and PAHs. Chemosphere, 70, 270-280. Campbell, L.A. (1996). A study of storm water runoff from Alexandra Township in the Jukskei River. MSc dissertation, University of the Witwatersrand, Johannesburg, South Africa. Degger, N., Wepener, V., Richardson, B. I., & Wu, R. S. S. (2011). Brown mussels (Pernaperna) and semipermeable membrane devices (SPMDs) as indicators of organic pollutants in the South African marine environment. Marine Pollution Bulletin, 63, 91-97. Harding, W.R, Thornton, J.A., Steyn, G., Panuska, J. & Morrison, I.R. (2004). Department of Agriculture, Environment and Tourism of the North West Province Government, South Africa. Harman, C, Tollefsen, K.-E., Boyum, O., Thomas, K., & Grung, M. (2008). Uptake of alkylphenols, PAHs and carbazoles in semipermeable membrane devices (SPMDs) and polar organic chemical integrated sampler (POCIS). Chemosphere, 72, 1510-1516. Hely-Hutchinson, J. R., & Schumann, E. H. (1997). Seasonal and inter-annual variability in the physical properties of Hartbeespoort Dam. South African Journal of Science, 93, 283-290. Huckins, J. N., Manuweera, G. K., Petty, J. D., Mackay, D., & Lebo, J. A. (1993). Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environmental Science and Technology, 27, 2489-2496. Huckins, J. N., Petty, J. D., & Booij, K. (2006). Monitors of organic chemicals in the environment: semipermeable membrane devices. New York: Springer. Kamens, R. M, Guo, Z, Fulcher, J. N., & Bell, D. A. (1988). The influence of humidity, sunlight and temperature on the daytime decay of polyaromatic hydrocarbons on atmospheric soot particles. Environmental Science and Technology, 22, 103-108. Karacik, B., Okay, O. S., Henkelmann, B., Pfister, G, & Schramm, K. W. (2013). Water concentrations of PAH, PCB and OCP by using semipermeable membrane devices and sediments. Marine Pollution Bulletin, 70, 258-265. Li, Y. F., & Macdonald, R. W. (2005). Sources and pathways of selected organochlorine pesticides to the arctic and the effect of pathway divergence on HCH trends in biota: a review. Science of the Total Environment, 342, 87-106. Liu, A. X., Lang, Y. H., Xue, L. D, Liao, S. L., & Zhou, H. (2009). Probabilistic ecological risk assessment and source apportionment of polycyclic aromatic hydrocarbons in surface sediments from Yellow Sea. Bulletin of Environmental Contamination and Toxicology, 83, 681-687. Liu, Y, Beckingham, B., Ruegner, H., Li, Z., Ma, L., Schwientek, M., et al. (2013). Comparison of sedimentary PAHs in the rivers of Ammer (Germany) and Liangtan (China): differences between early- and newly-industrialised countries. Environmental Science and Technology, 47, 701-709. Lu, Y, Wang, Z., & Huckins, J. (2002). Review of the background and application of triolein-containing semipermeable membrane devices in aquatic environmental study. Aquatic Toxicology, 60, 139-153. Luo, X., Mai, B., Yang, Q., Fu, J., Sheng, G, & Wang, Z. (2004). Polycyclic aromatic hydrocarbons (PAHs) and Springer organochlorine pesticides in water columns from the Pearl River and the Macao harbor in the Pearl River Delta in South China. Marine Pollution Bulletin, 48, 1102-1115. Nyoni, H., Chimuka, L., Vrana, B., & Cukrowska, E. (2011). Membrane assisted passive sampler for triazines compounds-characterisation of environmental conditions and field performance. Analytica Chimica Acta, 694, 75-82. Ockenden, W. A., Breivik, K., Meijer, S. N., Steinnes, S. E., Sweetman, A. J., & Jones, K. C. (2003). The global recycling of persistent organic pollutants is strongly retarded by soils. Environmental Pollution, 121, 75-80. Quémerals, B., Lemieux, C, & Lum, K. R. (1994). Concentrations and sources of PCBs and organochlorine pesticides in the St. Lawrence River (Canada) and its tributaries. Chemosphere, 29, 591-610. Quinn, L., Pieters, R., Nieuwoudt, C, Borgen, A. R., Kylin, H., & Bouwman, H. (2009). Distribution profiles of selected organic pollutants in soils and sediments of industrial, residential and agricultural areas of South Africa. Journal of Environmental Monitoring, 11, 1647-1657. Rajendran, R. B., Imagawa, T, Tao, H., & Ramesh, R. (2005). Distribution of PCBs, HCHs and DDTs, and their ecotoxico-logical implications in Bay of Bengal, India. Environment International, 31, 503-512. Rusina, T, Smedes, F., Koblizkova, M., & Klanová, J. (2010). Calibration of silicone rubber passive samplers: experimental and modelled relations between sampling rate and compound properties. Environmental Science and Technology, 44,362-367. Šibali, L. L., Okwonkwo, J. O., & McCrindle, R. I. (2008). Determination of selected organochlorine pesticides (OCPs) compounds from the Jukskei River Catchment area in Gauteng, South Africa. Water SA, 34, 611-622. Sibiya, P., Cukrowska, E., Jônsson, J. Á., & Chimuka, L. (2012). Development and application of solid phase extraction (SPE) method for polycyclic aromatic hydrocarbons (PAHs) in water samples in Johannesburg area, South Africa. South African Journal of Chemistry, 65, 206-213. Sibiya, P., Cukrowska, E., Jônsson, J. Á., & Chimuka, L. (2013a). Hollow fiber liquid phase microextraction (HF-LPME) for the determination of polycyclic aromatic hydrocarbons (PAHs) in Johannesburg Jukskei River, South Africa. Chromatographia, 76, 427-436. Sibiya, P., Cukrowska, E., Tutu, H., & Chimuka, L. (2013b). Development and application of microwave assisted extraction (MAE) for the extraction of five polycyclic aromatic hydrocarbons in sediments samples in Johannesburg area, South Africa. Environmental Monitoring and Assessment, 185, 5537-5550. Torien, D. F., & Walmsley, R. D. (1979). The chemical composition of the upper Hennops River and its implications on the water quality of the Rietvlei Dam. Water SA, 35, 77-89. Van Rei, W.F. (1987). The Hartbeespoort dam—a magnet to millions? In: J.A. Thornton, R.D. Walmsley (Eds.), Hartbeespoort Dam—Quo Vadis, vol. 25, FRD Ecosysystem Programmes Occasional Report 83-93. Vrana, B., Mills, G. A., Allan, I. J., Dominiak, E., Svensson, K, Morrison, G, et al. (2005). Passive sampling techniques for monitoring pollutants in water. Trends in Analytical Chemistry, 24, 845-868. Vrana, B., Klucarova, V, Benicka, E., Abou-Mrad, N., Amdany, R., Horáková, S., et al. (2014). Passive sampling: an effective Environ Monit Assess (2014) 186:5639-5653 method for monitoring seasonal and spatial variability of dissolved hydrophobic organic contaminants and metals in the Danube River. Environmental Pollution, 184, 101-102. Wang, Z., Chen, J., Yang, P., Qiao, X., & Tian, F. (2007). Polycyclic aromatic hydrocarbons in Dalian soils: distribution and toxicity assessment. Journal of Environmental Monitoring, 9, 199-204. Wang, J., Bi, Y, Pfiser, G., Henkelmann, B., Zhu, K., & Schramm, K. W. (2009). Determination of PAH, PCB and OCP in water from the Three Gorges Reservoir accumulated by semipermeable membrane devices (SPMD). Chemosphere, 75, 1119-1127. Wedemeyer, G. (1966). Dechlorination of DDT by Aerobacter aerogenes. Science, 152, 641-647. Willet, K. L., Ulrich, E. M., & Hites, R. A. (1998). Differential toxicity and environmental fates of 5653 hexachlorocyclohexane isomers. Environmental Science and Technology, 32, 2197-2207. Wu, W. Z., Schramm, K. W., & Kettrup, A. (1997). Study of sorption, biodegradation and isomerisation of HCH in stimulated sediment/water system. Chemosphere, 35, 1887-1894. Yunker, M. B., Macdonald, R. W., Vingarzan, R., Mitchell, R. H., Goyette, D., & Sylvestre, S. (2002). PAHs in the Frazer River basin: a critical appraisal of PAH ratios as indicators of PAH source and composition. Journal of Organic Geochemistry, 33, 489-515. Zhang, Z. L., Huang, J., Yu, G, & Hong, H. S. (2004). Occurrence of PAHs, PCBs and organochlorine pesticides in the Tonghui River of Beijing, China. Environmental Pollution, 130, 249-261. Springer Príloha 28 Miěge, C, Mazzella, N., Allan, I., Dulio, V., Smedes, F., Tixier, C, Vermeirssen, E., Brant, J., O'Toole, S., Budzinski, H., Ghestem, J.-P., Staub, P.-F., Lardy-Fontan, S., Gonzalez, J.-L., Coquery, M., Vrana, B., 2015. Position paper on passive sampling techniques for the monitoring of contaminants in the aquatic environment - Achievements to date and perspectives. Trends Environ. Anal. Chem. 2015, in press. Trends in Environmental Analytical Chemistry 8 (2015) 20-26 ELSEVIER Contents lists available at ScienceDirect Trends in Environmental Analytical Chemistry journal homepage: www.elsevier.com/locate/teac TrEAC ĽJ Position paper on passive sampling techniques for the monitoring of contaminants in the aquatic environment - Achievements to date and perspectives Cecile Miěge3*, Nicolas Mazzellab, Ian Allanc, Valeria Duliod, Foppe Smedesef, Celine Tixierg, Etienne Vermeirssen h, Jan Brant1, Simon OTooleJ, Helene Budzinskik, Jean-Philippe GhestemPierre-Francois Staub m, Sophie Lardy-Fontan n, Jean-Louis Gonzalez0, Marina Coquery3, Branislav Vranae 'Irstea, UR MALY (Freshwater Systems Ecology and Pollution Research Unit), 5 rue de la Doua, CS70077, 69626 Villeurbanne Cedex, France bIrstea, UR REBX (Water Research Unit), 50 Avenue de Verdun, Gazinet, 33612 Cestas Cedex, France CNIVA, Section 312, Gaustadalleen 21, 0349 Oslo, Norway dINERIS - Direction Risques Chroniques, Rue Jacques Taffanel, Pare technologique ALATA, 60550 Verneuil-en-Halatte, France e Masaryk University, Faculty of Science, Research Centre for Toxic Compounds in the Environment RECETOX Kamenice 753/5, 625 00 Brno, Czech Republic fDeltares, P.O. Box 85467, 3508 AL Utrecht, The Netherlands glfremer, RBE-BE-LBCO, rue de l'lle ďYeu, B.P. 21105, 44311 Nantes Cedex 3, France h Swiss Centre for Applied Ecotoxicology Eawag-EPFL, 8600 Dabendorf, Switzerland 'Cefas, Pakefield Road, Lowestoft, Suffolk NR34 9DX, UK 'Environmental Protection Agency, Richview, Clonskeagh Road, Dublin 14, Ireland k Universitě' Bordeaux 1, EPOC - UMR 5805 CNRS, LPTC, Bat. A12, 351 crs de la Libe'ration, 33405 Talence, France 1BRGM, Laboratory Division, 3 Avenue Claude Guillemin, 45060 Orleans Cedex 02, France m ONEMA, 5 allée Felix Nadar, 94300 Vincennes, France "LNE, DMSI, 1 rue Gaston Boissier, 75724 Paris Cedex 15, France "Ifremer, RBE-BE-LBCM, B.P. 330, Zone Portuaire de Brégaillon, 83507 La Seyne/mer Cedex, France (D CrossMark ARTICLE INFO ABSTRACT Article history: Received 17 July 2015 Accepted 20 July 2015 Keywords: Passive sampling Water framework directive Monitoring programmes Priority substances Emerging substances Environmental quality standards This paper, based on the outcome of discussions at a NORMAN Network-supported workshop in Lyon (France) in November 2014 aims to provide a common position of passive sampling community experts regarding concrete actions required to foster the use of passive sampling techniques in support of contaminant risk assessment and management and for routine monitoring of contaminants in aquatic systems. The brief roadmap presented here focusses on the identification of robust passive sampling methodology, technology that requires further development or that has yet to be developed, our current knowledge of the evaluation of uncertainties when calculating a freely dissolved concentration, the relationship between data from PS and that obtained through biomonitoring. A tiered approach to identifying areas of potential environmental quality standard (EQS) exceedances is also shown. Finally, we propose a list of recommended actions to improve the acceptance of passive sampling by policymakers. These include the drafting of guidelines, quality assurance and control procedures, developing demonstration projects where biomonitoring and passive sampling are undertaken alongside, organising proficiency testing schemes and interlaboratory comparison and, finally, establishing passive sampler-based assessment criteria in relation to existing EQS. © 2015 Published by Elsevier B.V. 1. Introduction For two decades, several passive sampling devices have been developed for the monitoring of organic and inorganic * Corresponding author. E-mail address: cecile.miege@irstea.fr (C. Miěge). contaminants in aquatic environments. These passive samplers (PS) enable the improvement of limits of quantification (LOQJ by accumulation and concentration of contaminants over long-term exposure. Moreover, when they are used in the integrative phase of uptake (i.e. integrative samplers), time-weighted average (TWA) concentrations over the exposure period can be derived, leading to a better representativeness of measurements. http://dx.doi.org/10.1016/j.teac.2015.07.001 2214-1588/© 2015 Published by Elsevier B.V. C. Miěge et al./Trends in Environmental Analytical Chemistry 8 (2015) 20-26 21 Such passive sampling techniques have been recommended in the European Commission Guidance Document on surface water chemical monitoring [1], then in the Water Framework Directive (WFD) daughter Directive 2013/39/EU [2] as complementary methods to improve the level of confidence in water monitoring data in comparison with conventional spot sampling. PS are assumed to have a positive influence on the future design and output of monitoring programmes in the context of the WFD and the Marine Strategy Framework Directive (MSFD). However, some barriers still remain that prevent regulatory acceptance and actual implementation of these tools for routine monitoring of contaminants in aquatic systems. In order to endorse PS use in monitoring programmes, several actions have been conducted, including interlaboratory studies (ILS) to evaluate the performances of passive sampling methods with a focus on (i) hydrophobic substances in situ [3], (ii) hydrophobic substances in laboratory (ECLIPSE project, [4]), (iii) priority substances in situ (AQUAREF, www.aquaref.fr, [5]), and (iv) emerging substances in situ (NORMAN network, http://www. norman-network.net/?qHome, with the Joint Research Centre's Institute for Environment and Sustainability, JRC-IES, [6]). Moreover, a NORMAN Expert Group meeting on "Linking Environmental Quality Standards and Passive Sampling" was organised in July 2013 in Brno (CZ) to discuss the possible routes for the implementation of passive sampling in regulatory monitoring for checking of compliance with Environmental Quality Standards (EQS) for WFD priority and river basin-specific substances. And, in collaboration with the International Commission for the Protection of the Danube River (ICPDR) and within the framework of the Joint Danube Survey (JDS3) in 2013, the NORMAN network launched a study to develop and test a methodology for continuous screening of large rivers using passive sampling. The aim was to assess the applicability of a temporally and spatially integrative sampling approach as a water quality monitoring tool for various substances. The results of this study have been published recently [7]. In November 2014, a "Workshop on Passive Sampling techniques for monitoring of contaminants in the aquatic environment", was organised jointly by the NORMAN network and AQUAREF, at Irstea, Lyon, France. This workshop brought together experts involved in passive sampling activities carried out by the NORMAN network and beyond. They discussed the state of the art and defined the strategy and a roadmap of further actions to be fostered by NORMAN, for 2015 and beyond, to improve implementation of passive sampling techniques in environmental monitoring. The present paper is addressed to scientists and to water managers and decision-makers at river basin, national and European level. The aim of this paper is to provide a common position, as discussed at the workshop in Lyon, of the passive sampling community experts regarding concrete actions required to improve the use of passive sampling techniques in support of risk assessment and risk management and to point to ways of overcoming the remaining barriers to regulatory acceptance and actual implementation of these tools for routine monitoring. Particular attention is given to organic contaminants, for which various types of PS can be used according to their hydrophobicity (Sections 3.1 and 3.2). The discussion on PS for monitoring programmes in water and biota (Sections 3.3 and 3.4) also includes the case of metals, as sampled with the generally accepted PS: Diffusive Gradient in Thin Films (DGT) [8]. 2. Method The first day of the meeting focused on discussions between scientific experts on technical issues surrounding the features and performance of passive sampling techniques. Participation on the second day was also open up to stakeholders and embraced the applicability of PS in regulatory monitoring programmes in the aquatic environment (WFD - MSFD, OSPAR Convention, etc.). The workshop was organised in four sections which reflect the recurrent questions and challenges identified by decision-makers as regards the use of passive sampling techniques for environmental monitoring: 1. Which PS are suitable for monitoring hydrophobic organic compounds in water? Can we expect to obtain accurate time-weighted average (TWA) concentrations with these PS? 2. Which PS are suitable for monitoring hydrophilic organic compounds in water? Can we expect to obtain accurate time-weighted average (TWA) concentrations with these PS? 3. What is the role of passive and grab sampling approaches in monitoring programmes? Are data obtained by passive sampling comparable with those from grab sampling? 4. What role can passive sampling play in support to chemical monitoring in biota? The conclusions presented in this paper are organised following these 5 successive items. Parts 1 and 2 focus on organic contaminants, whereas parts 3 and 4 cover all contaminants, including metals. 3. Results and discussion 3.2. Which passive samplers are suitable for monitoring hydrophobic compounds in water? Various types of PS are available for hydrophobic compounds: the Semi-Permeable Membrane Device (SPMD, biphasic system), silicone rubber and Low Density PolyEthylene (LDPE) strips (monophasic systems) are the most commonly used [8]. It is not possible to recommend a single specific PS. Rather, PS calibration data should satisfy certain performance or quality standard criteria, and uptake and release processes should be in agreement with theory. Recommending a specific PS would also lead to a loss of information and prevent an improvement of existing techniques or new developments. SPMD is a biphasic PS (a polyethylene membrane filled with lipid), and can therefore generally be considered more complex than monophasic polymers concerning sample processing in the laboratory and modelling of contaminant uptake mechanisms. Given these constraints, it is expected that the use of monophasic samplers will be favoured over the use of SPMD. Nevertheless, the use of SPMD for more than 20 years has generated numerous laboratory and field data. Moreover, it is at present the only standardised and commercially available PS for hydrophobic compounds. Even so, for practical reasons, monophasic polymers (e.g. silicone rubber, LDPE) appear to be the most suitable PS for sampling of hydrophobic compounds. Monophasic polymers can be of different qualities and made of different materials; but at the moment, there are no standard commercial products available. It was therefore unanimously agreed that there is a need for commercial supplies of standard monophasic PS. Suitable polymers should meet the following criteria: • the uptake of the polymers must be based on absorption (not adsorption) and sampler/water partition coefficients for the compounds of interest should be sufficiently high in order to allow good performance in terms of substance accumulation; • the diffusion coefficients of target substances inside the polymer should be sufficiently high so water boundary control dominates 22 C. Miege et al./Trends in Environmental Analytical Chemistry 8 (2015) 20-26 the uptake process, even under severe turbulence conditions. This allows the uptake process to be calibrated from the release of Performance Reference Compounds (PRC, i.e. a sort of internal standards) that are dosed prior to deployment [9,10]. For each new monophasic polymer, sufficient diffusion should be confirmed and partition coefficients should be determined either independently or through cross-calibration against a polymer with already known partition coefficients. Such a polymer (e.g. silicone) could serve as a reference material for sampler cross-calibration. For accurate analysis of PS, there is also a need for certified reference materials (CRMs) of polymers used in passive sampling containing the most widely monitored and regulated compounds. Preparation of such CRMs could be the role of the EuropeanJRC for Reference Materials and Measurements (IRMM) and/or of the National Metrology Institutes (NMIs). The application of PS in waters requires knowledge of polymer-water partition coefficients (Kpw) and knowledge that diffusion coefficients (Dp) in the polymer are sufficiently high, both for substances of interest and for those used as PRC. When commercial PS products and CRMs become available, their routine use for monitoring compounds whose diffusion and partition coefficients (and their uncertainty) have been published will not require additional calibration experiments by end-users. The use of accurate Kpvj constants, PRC for measurement of in situ exchange kinetics, and the application of validated uptake models are sufficient for accurate measurements of contaminant concentrations in waters using PS. Thus, in order to support the use of PS, it is important to: • Develop harmonised guidelines, in particular for: o the measurement of polymer-water partition coefficients (Kpw); o the measurement of substance diffusion coefficients (Dp) in PS polymers; o the definition of criteria for an appropriate application of PRC; o the definition of suitable and validated models for calculation of water concentration from PS. • Perform interlaboratory studies to improve validation of PS for routine use. As to the latter, it is recommended that interlaboratory studies aimed at validation of PS for routine use should be designed as two-step exercises, in which Step 1 is the Proficiency Test (PT) for the analysis of the contaminants in the extracts of PS, and Step 2 is an interlaboratory study for intercomparison of PS field-deployment and analysis of contaminants in PS, including estimation of water concentration. Only skilled laboratories (i.e., those that succeeded in Step 1) should be allowed to participate in Step 2. For the choice of contaminants, the focus should be on hydrophobic WFD Priority Substances and other substances (including the new Priority Substances) for which robust analytical methods already exist (for analysis in PS exposed in the aquatic environment). With respect to the influence of temperature and salinity, Kpvj values used for calculation of freely dissolved concentrations are usually determined for T = 20 °C and salinity = 0%0. Workshop participants concluded that there is no need to correct Kpw for temperature nor salinity, since EQS values are not corrected for the effects of these parameters, when used for compliance monitoring (to be noted that there are specific EQS in marine waters). Moreover, the approach using Kpvj without correction provides more conservative water concentration estimates (higher concentrations are estimated in scenarios with low temperature and high salinity); such estimates are therefore more protective when referring to compliance with EQS (worst case scenario). 3.2. Which passive samplers are suitable for monitoring ofhydrophilic compounds in water? Can we expect to obtain accurate time weighted average (TWA) concentrations with passive sampling? Various types and configurations of PS exist today for hydrophilic compounds: the Polar Organic Chemical Integrative Sampler (POCIS) (e.g. with different membranes and sorbent phases), the Chemcatcher and the Empore disks are the most commonly used [8]. At present, it is not possible to recommend a preferred specific PS for sampling of hydrophilic compounds. It was acknowledged that at present the mechanisms of uptake and release of hydrophilic substances from water into these adsorption-based PS are not fully understood. The exchange of compounds between the PS and the aqueous phase can often be considered an anisotropic process. Consequently, it is generally not possible to use the release of PRC to calibrate the uptake rate and allow calculation of time weighted average (TWA) water concentrations for a wide range of compounds. Nonetheless, PRC should be used as surrogates to check that exposure conditions (e.g. temperature, salinity, water flow) are within the limits for which the laboratory derived the calibration data (quality controls). Currently, adsorption-based PS for hydrophilic compounds allow only semi-quantitative information to be obtained. This is because of the uncertainty in applying laboratory-based sampling rates to in situ field conditions. However, when confidence intervals of estimated TWA concentration are available, these PS data could be used for EQS compliance checking. One of the possible approaches to apply PS data for assessing compliance with a regulatory limit involves the calculation of the upper 90% confidence limit of the PS-derived TWA concentration. Accurate analyses and the use of an equivalent volume of water sampled by the PS smaller than the actual sampled volume to calculate water concentrations would ensure that estimated TWA concentrations are an overestimate of actual concentrations and a robust use of PS. The good status cannot be considered as achieved if the calculated upper TWA concentration limit exceeds the EQS. This is possible for substances for which linear uptake is confirmed for the period of exposure. Poulier et al. [11] recently proposed a method to determine confidence intervals for each TWA concentration estimate by POCIS, over a period of one year (Fig. 1). The means of maximum and minimum limits of these confidence intervals are defined as MAX and MIN, respectively. Thereafter, the MAX and MIN values are compared to the AA-EQS (annual average EQS) and good chemical status is considered to be achieved if MAX is lower than the AA-EQS (Fig. 1). Understanding the uptake mechanism of polar compounds into adsorption-based PS is the first and most important issue that needs to be resolved in order to reduce the currently observed uncertainty in passive sampling data. New solutions have to be found to simplify PS construction to an effective minimum. In this process, it is possible that some of the traditionally applied passive sampling designs will have to be abandoned (e.g. application of membranes in PS, which often cause undesired complications of the uptake mechanism). Even if PS tools for hydrophilic substances still need developments and adaptations, guidelines describing how to conduct PS calibrations are required. In particular, such guidance should define a common set of metadata and calibration conditions (temperature, water flow, type of the exposure system, type of water) to be reported together with the obtained sampler calibration parameters. All this information is required for the C. Miěge et al./Trends in Environmental Analytical Chemistry 8 (2015) 20-26 23 Step 1: Analysis of the POCIS exposed monthly in the water body Step2: C P0C(S are corrected with the appropriate factors to account for the lower and upper limits of the POCI. uncertainty. Step 3: Calculation of the average concentration MIN and MAX MIN Values belowLQ are replacedby 0 MAX Values belowLQ are replacedby the LQ Step4 Water diagnosis MIN > AA-EQS MIN < AA-EQS < MAX MAX < AA-EQS Good statusis not achieved Fig. 1. Proposed procedure to use POCIS data for surveillance monitoring. From Poulier et al. [ 11 ] Wo diagnosis possible Good statusis achieved assessment of the possible relationship between the observed variability in available calibration data and the exposure conditions used in calibrations [12]. In situations where the effect of environmental conditions on the PS performance (especially the sampling rate) in the field cannot be either determined or controlled, application of laboratory-derived calibration parameters will always introduce a systematic error into derived water TWA concentrations. When water concentrations are calculated from passive sampling data, expected variability of applied calibration parameters should be included in the calculation of the reported concentration. The value and uncertainty of applied sampling rates and the approach for calculation of uncertainty should also be reported. More generally, the reporting of passive sampling data requires improved practice, focusing particularly on the data and models used to estimate water concentrations from contaminant masses sorbed into the PS. In contrast with spot sampling, PS provides time-integrated concentrations of pollutants. If the uncertainty of water concentrations obtained from PS is lower than the variability of environmental concentrations, data obtained by PS represent the contamination situation in the water body as well as or better than the low frequency spot sampling (e.g. based on 4-12 sampling times per year) that is currently used in compliance monitoring for the WFD. Previous interlaboratory studies (including the AQUAREFILS [5] in 2010 and NORMAN ILS [6] in 2011) showed that accurate analysis of certain hydrophilic substances (pesticides, pharmaceuticals, steroid hormones, perfluorinated compounds) remains a challenge for a number of laboratories. Inaccurate analyses contributed significantly to the observed high variability of water concentrations derived from PS data which cannot be attributed to inadequacies of the PS process. It was therefore recommended to organise further intercomparison studies. As for hydrophobic compounds, in order to ensure validation of the different parts of the PS process, future intercomparison studies should be designed as two-step exercises, where Step 1 is the FT for analysis of contaminants in extracts of PS, and Step 2 is Interlaboratory comparisons for PS field-deployment and analysis of contaminants in PS. Only skilled laboratories (i.e., those that succeeded in Step 1) should be allowed to participate in Step 2. For the choice of contaminants, the focus should be on WFD Priority Substances and other hydrophilic substances (including new Priority Substances) for which robust analytical methods exist (in PS exposed in real water). Finally, workshop participants identified the need to develop PS for ionic and highly hydrophilic compounds (e.g. glyphosate). 3.3. Passive versus grab sampling approaches in monitoring programmes Passive sampling measures the dissolved phase concentration of a contaminant (and not the whole water concentration, as required by Directive 2013/39/EU [2]). As a result, passive sampling cannot be used today to assess compliance with EQS for all organic contaminants under the WFD, but only for moderately polar to polar organic compounds (with log KOVJ < 5) where the concentration in the water column is not dominated by the fraction adsorbed to colloids and particles in water. On the other hand, passive sampling is recommended in the European Commission Guidance Document on surface water chemical monitoring [ 1 ] and in the Directive 2013/ 39/EU [2] as a complementary method to improve the quality of the assessment and as a resource saving measure. In this regard, passive sampling could be used in conjunction with investigative monitoring as a risk-based screening tool to evaluate the presence or absence of chemical contaminants, to identify sources of pollution when the concentration levels (and therefore the required limits of detection) are extremely low or when the source of pollution is intermittent. Passive sampling can also be employed in trend monitoring both as a qualitative and a quantitative tool. PS offer added value compared to grab sampling when applied as an "early-warning tool" to detect increasing (or decreasing) trends. Exceedance of defined threshold values could be used to trigger further monitoring using conventional sampling techniques, e.g. grab sampling and/or biota monitoring. Some practical advantages of passive sampling can be highlighted: • low limits of detection and quantification can be achieved, especially with samplers for hydrophobic compounds; • in situ sample preconcentration is possible and the handling of large water volumes can be avoided (thereby allowing lower costs for transport and storage in comparison with conventional spot sampling, and easier sampling in remote locations); • thanks to higher stability of the sampled compounds, it is possible to allow prolonged sample storage; • analysis of samples can be delayed and, if needed, combined to composite samples; 24 C. Miége et at/Trends in Environmental Analytical Chemistry 8 (2015) 20-26 • unlike water samples, sorbents or extracts of PS are more suitable for long term storage in specimen banks. As to the quality of the information obtained from PS measurement results: • information obtained with PS is representative of an extended time period; this integrated information is more relevant to describe the status of a water body than the information which can be obtained with spot sampling; • only freely dissolved compounds are sampled: for hydrophobic compounds, PS provide a measure directly proportional to the chemical activity of the contaminant of interest in the medium being sampled; • PS allow a reduction in the effect of blank contamination, since the integrative character of sampling allows concentrations in exposed PS to be found that are significantly higher than levels found in blanks. There is still a need for pilot field studies to gain experience and demonstrate the usefulness and relevance of passive sampling strategies compared to grab sampling. Such demonstration studies should be designed to show the difference between conventional monitoring (i.e. 4-12 spot water samples/year, or integrative biota monitoring for hydrophobic compounds and metals) and a new, more relevant and practical concept using PS. The study should aim to demonstrate that a TWA concentration via PS is more representative and relevant - compared to conventional monitoring - for the characterisation of the chemical status of water bodies. In France, such a demonstration exercise is planned by AQUAREF for the next WFD monitoring cycle, in close connection with policy-makers, stakeholders and end-users (water agencies). This action could be extended to the European level through NORMAN network activity. In the Netherlands, local water authorities have been using PS for monitoring POPs in surface and coastal waters in parallel with monitoring in mussels [13] for more than a decade. In addition, demonstration studies applying passive sampling in parallel with biota monitoring and led by the Environment Agency in the UK are under way. Indeed, regulatory implementation of PS requires decisionmakers to be convinced of the need to globally change the current monitoring and compliance checking concept under the WFD. The relevance of the signal obtained by passive sampling (integrative sampling, relation of TWA concentrations with the environmental risk to aquatic organisms) should be stressed. Such a change in the monitoring concept recently took place in the anti-doping sector in sports where controls are now performed on hair (integrative information) rather than in urine (punctual information). It is acknowledged that there is much more experience of large scale PS application for marine water monitoring than for freshwater monitoring. It is therefore necessary to better share this experience between the two expert communities. For example, the three-level approach in place within OSPAR, which consists of drafting of guidance documents, organisation of proficiency tests (via QUASIMEME, http://www.quasimeme.org) and definition of water quality assessment criteria, could also be applied to continental waters [14]. In order to allow improved compilation and comparison of measurement data from PS, experts agreed that it is necessary to define a common and harmonised set of metadata that should accompany the measurement results to be reported in the literature and/or in databases. It is recommended that such a harmonised set of metadata should be included in the next update of the ISO 5667/23 standard [15]. A central European repository (database) would be useful to better share PS monitoring data. This database should gather information on the PS used, the conditions of deployment, the analytical method, the method to treat the results, the concentration in the PS and the estimated water TWA concentration. There is already a NORMAN template for collecting PS data (used for passive sampling data collection from the Joint Danube Survey 3 [7]). This template could be used by the PS community as the basis of a possible upgrade before final validation and adoption as a common data collection template. Finally, to facilitate communication and dissemination, there is a need to adopt harmonised terminology within the PS research area. Some knowledge gaps remain as regards the battery of passive sampling devices suitable for very hydrophilic and/or ionisable substances, for some priority substances (e.g. PFOS and mercury) for which biota EQS exist, and for substances with extremely low EQS in water (e.g. dichlorvos, dicofol and heptachlor) [2]. 3.4. Applicability of passive sampling in support of chemical monitoring in biota With the recent update of the EQS Directive 2013/39/EU [2], there is a demand for cost efficient monitoring tools that could support data obtained from chemical monitoring in biota. The newly introduced EQSbiota for hydrophobic compounds call for the use of analytical methods that meet the requirements of the QA/QC Directive (2009/90/EU) [16]. With these EQSbiota, protection of human health via consumption of fishery products, and protection of predators against secondary poisoning were also introduced as new protection goals. Hence, these EQSbiota bring new challenges in the design of monitoring programmes and data interpretation for compliance checking and assessment of trends (e.g., the need to normalise biota data based on lipid content, trophic magnification factor). According to the European Commission technical guidance for the implementation of EQSbiota [17], PS can be applied in a tiered approach to identify or rank areas of potential EQS exceedance (Fig. 2, [18]). In sucha tiered approach, trigger values (i.e. threshold concentrations, exceedance of which triggers the second tier, monitoring of biota) are needed. Experts discussed further possibilities of the application of PS, beyond the current recommendation of the European Commission, to support or replace chemical monitoring of hydrophobic compounds and mercury in biota. Despite the recommended normalisation of biota monitoring data prior to chemical status assessment, the establishment of temporal and spatial trends of bioaccumulating compounds is still expected to be complicated by the inherent variability of the sampled aquatic organisms. Even if "active biomonitoring" for biota (caged organisms) offers some practical solution for marine waters and more recently for continental waters [17,19], experts believe that the inherent variability of passive sampling data can be much better controlled, which presents the main advantage of the abiotic sampling approach. Experts agreed that passive sampling cannot predict actual concentrations of priority compounds in biota. Passive sampling data can predict the concentrations that would be determined in biota (lipid) if the organism were at thermodynamic equilibrium or steady state with the environment. However, deviations from equilibrium cannot be easily forecasted because of the complexity of uptake processes, trophic magnification, growth dilution, seasonal influences and the "home-range" of the species, which result in a large variability of accumulation of chemical contaminants in biota. As a result, bioconcentration factor (BCF) and bioaccumulation factor (BAF) data reported in the literature are extremely variable. The application of these BCF and BAF literature values to predict concentrations of contaminants in biota C. Miěge et al./Trends in Environmental Analytical Chemistry 8 (2015) 20-26 25 Tier 1: Use of validated passive sampler Tier 3: Refinement of understanding of risk to develop evidence for action Pass Confirmed Exceedance 1 Programmes of measures Tier 1: Validated PS "screen" where EQSbj0ta available ->Presence/absence. Calibration to validate "non-detection" = no risk to biota. Positive detection -> Biota screening Tier 2: Risk to predators/humans via food chain. Collect larger numbers of small organisms. Human health based EQS -» sample Fish/biota. < EQSbiota = STOP Tier 3: Refinement of risk and increasing confidence in assessment (increased sampling programme, geographical aspects etc.) Fig. 2. Proposed tiered approach to identify potential EQS exceedance using PS. From Whitehouse [18] from passive sampling derived aqueous concentrations thus lead to a large variability. In spite of those limitations, experts are of the opinion that PS reflect very well the contaminant levels to which biota are exposed in their natural environment. The same contaminant trends (in time and space) could be observed both in biota data and in passive sampling data (as demonstrated for example by the long-term observation of PS vs mussels performed in the Netherlands for marine waters [13]). Experts concluded that passive sampling is a suitable tool to determine spatial and temporal trends, with lower inherent data variability compared to chemical monitoring in biota. The expert view is that (except for secondary poisoning purposes) measuring contaminant levels in waters can be more appropriate for assessing aquatic biota exposure than measuring their concentration in the organisms. For example, some compounds that are actively metabolised would not be found in organisms (or only at low concentrations), although organisms were exposed to them (e.g. polycyclic aromatic hydrocarbons in fish). Nonetheless, it must be noted that recent studies showed that active biomonitoring in gammarids could provide useful data for metals exposure in freshwater systems [20]. If EQSbiota were set only to protect human health from exposure via consumption of fish, there would be no role for passive sampling in water monitoring. In this case it would be sufficient to assess that levels of contaminants in fish used for human consumption do not exceed the defined thresholds. However, since the definition of EQSbiota also embraces other protection goals, including protection of aquatic life, PS can still play a significant role in WFD monitoring. According to the WFD, it is possible to convert EQSbiota to equally protective EQS in water (EQSwater) and use such standards in regulatory monitoring. The uncertainty of PS concentrations of the most hydrophobic priority substances in water is sufficiently low to allow in principle for a comparison with EQSwater [3,21 ]. This is possible especially because limits of quantification that are achievable by passive sampling for those hydrophobic compounds are lower than the respective EQSs. From the uptake of hydrophobic pollutants by PS, the freely dissolved concentration is estimated, which represents the driving force for bioconcentration in organisms. PS thus enable the in situ determination of hydrophobic bioaccumulative organic compounds that organisms at the lowest trophic level are exposed to. The results from passive sampling can also be converted into lipid-based concentrations for an organism considered at equilibrium with the environment to which the sampler was exposed (using lipid-polymer partition coefficients). The advantage of expressing results on a lipid basis is, besides being more closely related to concentrations in biota, that it is an easier unit to communicate to regulators and the public, since it is difficult for a layman to understand that concentrations in the range of fg/L to pg/L in water can pose a hazard. Lipid-polymer partition coefficients will be needed for all substances of interest (i.e. those with existing EQSbiota); and for those for which values already exist, further validation may be required. A major recommendation resulting from this workshop is that, on the sites across Europe where biota monitoring is undertaken for WFD or OSPAR purposes, biota monitoring should be as far as possible complemented by PS exposures. This will help develop the much needed datasets to improve our understanding of bioaccumulation factors. Parallel exposures of PS with biota monitoring (ideally, including multiple trophic levels) at a number of sites in Europe (with different exposure levels) will enable assessment of the variability of BAFs used in the conversion of EQSbiota to EQSwater (BAF = Cbiota/CWater. Cwater is the freely dissolved concentration from PS, BAF could be established at different river basins). When such variability is known and 26 C. Miége et al./Trends in Environmental Analytical Chemistry 8 (2015) 20-26 acceptable, biota monitoring could be subsequently replaced by monitoring with PS for compliance checking. 4. Conclusions This paper summarises the outcome of discussions that were held during a NORMAN Network-workshop in Lyon (France) in November 2014. We aimed to provide commonly agreed recommendations to enable the future use of passive sampling for regulatory monitoring of contaminants in aquatic environments. We hope these steps will contribute to increase acceptance of passive sampling by policy-makers. A number of concrete actions required to advance the use of passive sampling techniques in support of contaminant risk assessment and management have been identified: • Monophasic polymers (e.g. silicone rubber or low density polyethylene) are recommended as the PS of choice for hydrophobic, non-ionised organic substances and the community unanimously agrees that there is a need for commercial supplies of monophasic passive samplers. • Currently, for hydrophilic organic substances, adsorption-based samplers (e.g. POCIS) provide semi-quantitative data only and further research is needed to either (a) reduce uncertainty of measurement of existing devices, or (b) develop a new sampler design with a simpler (and better controlled) contaminant uptake mechanism. Another viable route for application of these devices in regulatory monitoring, for EQS compliance checking of WFD Priority Substances, is to establish intervals of estimated TWA concentrations and to compare the maximum and minimum limits of these confidence intervals to the AA-EQS values. • For the future, the development of new PS for ionic and highly hydrophilic compounds is required. • Uncertainty associated with passive sampling-derived aqueous concentrations can be evaluated and taken into account when PS are used for trend and compliance monitoring. This is confirmed by experience from previous interlaboratory studies, which clearly showed that for certain groups of emerging compounds, inaccurate analysis, rather than the passive sampling technique, is still the main cause of the observed high variability of the results reported by the laboratories. Future intercomparison studies should be organised so that they include different steps in order to ensure validation of each critical part of the sampling and analytical process (i.e. analysis of the contaminants in the extract, PS-field deployment and analysis of the contaminants in the PS, including calculation of water concentration). • One major feature of passive sampling compared to grab sampling is that PS provide TWA concentration results. These integrated TWA measurement data provide more representative and relevant information for characterisation of the chemical status of water bodies than conventional monitoring (mean values of 4-12 spot samples) data. However, such a shift demands a radical change in the regulatory procedure with which water agencies and decision-makers are familiar. The launch of field studies where the two approaches, the conventional one and the PS approach, would be applied in parallel on a number of selected sites, is highly recommended in order to convince decision-makers that it is advantageous to make this shift. • PS reflect the contaminant levels to which biota have been exposed in their natural environment. • As regards chemical monitoring of hydrophobic priority substances in biota, PS can be applied in a tiered approach to identify or rank areas of potential risk of exceedance of EQSs before chemical monitoring in biota. Replacement of chemical monitoring in biota by PS can also be envisaged. The main advantage of such an alternative route is that PS can ensure a lower inherent variability of the concentration data compared to biota monitoring data. PS cannot predict actual concentrations of priority compounds in biota, but passive samplers reflect well the contaminant levels to which biota have been exposed in their natural environment. Since the definition of EQSbiota is not limited to protection of human health but also to the protection of aquatic life, and in consideration of the fact that the WFD allows EQSbiota to be converted in equally protective EQSwater, concentration data obtained with PS can be considered compatible with the protection objectives set by EQSbiota. • In consideration of all the above, steps to be undertaken to convince policy-makers to accept passive sampling in regulatory monitoring are: o Drafting of guidelines and clear Quality Assurance/Quality Control rules; o Running of demonstration projects/case studies with passive sampling undertaken alongside spot sampling and biota monitoring, in order to demonstrate their applicability for compliance monitoring purposes; o Organisation of proficiency testing (FT) schemes and interlaboratory exercises for passive sampling in water; o Development of assessment criteria in relation to EQSs. References [1] European Commission, Guidance Document no. 19 - Common implementation strategy for the Water Framework Directive (2000/60/EC). Guidance on surface water chemical monitoring under the Water Framework Directive. Technical Report, 2009 025,132 pp.. [2] Directive 2013/39/EU of the European Parliament and of the Council amending Directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy, Off. J. Eur. Union L 226 (2013) 1-17. [3] I.J. Allan, K. Booij, A. Paschke, B. Vrana, G.-A. Mills, R. Greenwood, Environ. Sei. Technol. 43 (2009) 5383. [4] RJacquet, C. Miege, F. Smedes, C. Tixier,J. Tronczynski, A. Togola, C. Berho, I. Valor, J. Llorca, B. Barillon, P. Marchand, M. Coquery, Chemosphere 98 (2014) 18. [5] C. Miege, N. Mazzella, S. Schiavone, A. Dabrin, C. Berho, J.-P. Ghestem, C. Gonzalez, J.-L. Gonzalez, B. Lalere, S. Lardy-Fontan, B. Lepot, D. Munaron, C. Tixier, A. Togola, M. Coquery, Trends Anal. Chem. 36 (2012) 128. [6] B. Vrana, F. Smedes, R Prokeš, R. Loos, N. Mazzella, C. Miege, H. Budzinski, E. Vermeirssen, T. Ocelka, A. Gravell, S. Kaserzon, Trends Anal. Chem. (2015) (submitted for publication). [7] B. Vrana, F. Smedes, T. Rusina, K. Okonski, I. Allan, M. Grung, K. Hilscherova, J. Novák P- Tarábek J. Slobodník, in: I. Liška, F. Wagner, M. Sengl, K. Deutsch, j. Slobodník (Eds.), Vienna: ICPDR - International Commission for the Protection of the Danube River, 2015, p. 304, http://www.danubesurvey.org/results. [8] B. Vrana, G.A. Mills, I.J. Allan, E. Dominiak K. Svensson, J. Knutsson, G. Morrison, R. Greenwood, Trends Anal. Chem. 24/10 (2005) 845. [9] TP.Rusina,F.Smedes.J.Klanová, K.Booij,I.Holoube,Chemosphere68(2007) 1344. [10] TP. Rusina, F. Smedes, M. Koblizkova, J. Klanová, Environ. Sei. Technol. 44 (2010) 362. [11] G. Poulier, S. Lissalde, A. Charriau, R Buzier, F. Delmas, K. Gery, A. Moreira, G. Guibaud, N. Mazzella, Sei. Total Environ. 497/498 (2014) 282. [12] N. Morin, C. Miege, J. Randon, M. Coquery, Trends Anal. Chem. 36 (2012) 144. [13] F. Smedes, in: R Greenwood, G.A. Mills, B. Vrana (Eds.), Passive Sampling Techniques in Environmental Monitoring, Elsevier, Amsterdam, 2007, p. 407. [14] OSPAR Commission, London Publication, no. 379, 2008. http://www.ospar.org/ documents/dbase/publications/p00379/p00379_cemp_assessment_manual.pdf. [15] ISO 5667-23, Water quality - Sampling - Part 23, 2011 23 pp.. [16] Directive 2009/90/EC of 31 July 2009 laying down, pursuant to Directive 2000/60/ EC of the European Parliament and of the Council, technical specifications for chemical analysis and monitoring of water status, Off. J. Eur. Union L 201 (2009) 36-38. [17] European Commission, Guidance document no. 25. Common implementation strategy of the Water Framework Directive (2000/60/EC).' Guidance on chemical monitoring of sediment and biota under the Water Framework Directive. Technical Report, 2015 74 pp.. [18] P. Whitehouse, Workshop on "Passive Sampling as a monitoring tool for emerging chemicals", Dublin, (Personal communication), 2014. [19] J.-P. Besse, O. Geffard, M. Coquery, Trends Anal. Chem. 36 (2012) 113. [20] j.-P. Besse, M. Coquery, C. Lopes, A. Chaumot, H. Budzinski, P. Labadie, 0. Geffard, Water Res. 47/2 (2013) 650. [21] R. Lohmann, K. Booij, F. Smedes, B. Vrana, Environ. Sei. Pollut. Res. 19 (2012) (1885). Príloha 29 Booij K, Vrana B., Huckins J. N: Chapter 7 Theory, modelling and calibration of passive samplers used in water monitoring. In: Comprehensive Analytical Chemistry, R. Greenwood, G. Mills, B. Vrana (eds.). Elsevier, Amsterdam, Volume 48, 2007, Pages 141-169. Chapter 7 Theory9 modelling and calibration of passive samplers used in water monitoring Kees Booij, Branislav Vrana and James N. Huckins 7.1 INTRODUCTION Contaminant uptake by passive sampling devices (PSDs) can be seen as a multi-stage transport process. To illustrate the basic steps involved, we will first discuss contaminant uptake by a PSD that consists of a central sorption phase, surrounded by a membrane. For this exercise, we assume that the sampler is biofouled, and is contained within a protective cage (Fig. 7.1). Coming from the surrounding waters, analytes first have to enter the protective cage, where the motion of water may be reduced relative to the water outside the cage. Close to the biofouling layer, convective transport of analyte molecules is reduced more and more, until all transport takes place by molecular diffusion within the water boundary layer (WBL). When ventilating organisms are present, diffusion may be amended with convective currents that are set up by the organisms. After diffusion through the membrane, analytes are finally sorbed by the central sorption phase. This general picture may differ from case to case. For example, protective cages and biofouling layers may be absent, the membrane may act as the final sorption phase (e.g. various types of solid-phase microextraction devices (SPMEs), and low-density polyethylene (LDPE) and polydimethylsiloxane (PDMS) strip samplers), or the sampler may be equipped with additional phases between the membrane and the central phase (e.g. membrane-enclosed sorptive coating (MESCO) and Chemcatcher samplers). A variety of models has been used over the past 15 years to better understand the kinetics of contaminant transfer to passive samplers. These models are essential for understanding how the amounts of absorbed contaminants relate to ambient concentrations, as well as for Comprehensive Analytical Chemistry 48 R. Greenwood, G. Mills and B. Vrana (Editors) Volume 48 ISSN: 0166-526X DOI: 10.1016/S0166-526X(06)48007-7 © 2007 Elsevier B.V. All rights reserved. 141 K. Booij, B. Vrana and J.N. Huckins water Fig. 7.1. Schematic representation of concentration profiles in a dual-phase PSD with exterior biofilm (i.e. the right half of a symmetrical sampler, or the whole cross section of a sampler with an impermeable boundary located to the left of the central phase). Dashed lines indicate how the effective thickness of the respective phases may be estimated (see Sections 7.5 and 7.6). the design and evaluation of calibration experiments. Models differ in the number of phases and simplifying assumptions that are taken into consideration, such as the existence of (pseudo-) steady-state conditions, the presence or absence of linear concentration gradients within the membrane phase, the way in which transport within the WBL is modelled and whether or not the aqueous concentration is constant during the sampler exposure. In the next sections, we will introduce the basic concepts and models used in the literature on passive samplers for the special case of trio-lein-containing semipermeable membrane devices (SPMDs). These can easily be extended to samplers with more or with less sorption phases. Then we will discuss the transport of chemicals through the various phases constituting PSDs. Finally, we will discuss the implications of these models for designing and evaluating calibration studies. 7.2 BASIC CONCEPTS AND MODELS FOR SPMDS Mass-transfer coefficients (kO are frequently used to link the flux (/j) through a phase (i) to the concentration difference AQ between the end points of that phase ji = k1ACi (7.1) 142 Theory, modelling and calibration Equation (7.1) is an expression of the notion that mass fluxes (J) are linearly proportional to a driving force (ACj). The mass-transfer coefficient can be interpreted as a conductivity term, with the dimension of a velocity (e.g. cm day-1). This approach has been followed to model contaminant uptake by a number of PSDs [1-7]. Huckins et al. [3] have applied this scheme for the case of contaminant uptake by triolein-filled SPMDs in the presence of biofouling, assuming that the fluxes at both sides of each interface are equal, and that local sorption equilibrium exists at the interfaces. In addition, these authors assumed that the ratios of space-averaged concentrations in the triolein and in the membrane phases are close to the triolein-membrane partition coefficient at all times. The latter assumption was confirmed for the case of SPMDs, by numerical integration of Fick's second law [8]. The differential equation that governs the uptake process can then be expressed as da Akn („ a dt Vs V Kt (7.2) lsw, where Cs and Cw are the volume-based contaminant concentrations in SPMD and in water respectively, Vs is the SPMD volume, A is the SPMD surface area, and Ksw is the SPMD-water partition coefficient. The Ksw equals the volume-averaged partition coefficient for the triolein phase CKlw) and the membrane (Kmw), as shown by Huckins et al. [9] ksw = m ;r ';tl (7.3) Vm-K'mw + Vl-^Lv vm + vL The overall mass-transfer coefficient kQ is given by ± = ± + ^ + ^- (7.4) where kw, k^, km are the mass-transfer coefficients for the WBL, the biofilm and the LDPE membrane, and Khw, Kmw are the biofilm-water and the membrane-water partition coefficients, respectively. Equation (7.4) is an expression of the fact that the total mass-transfer resistance (l/k0) equals the sum of the resistances posed by the respective phases. Acknowledging that a mass-transfer coefficient equals the ratio of a diffusion coefficient and an effective phase thickness (3), Eq. (7.4) can also be written as 1 _ <5W | <5m | <^b (75) k0 Dw DmKmw DhKhw 143 K. Booij, B. Vrana and J.N. Huckins Bartkow et al. [10] have accounted for the transport resistance posed by a protective cage that may surround SPMDs, by adding a term A/QY to the right-hand side of Eq. (7.4), where Qv is the volume rate of water flow to the protective cage and A the surface area of the SPMD. These authors concluded, however, that this resistance can be neglected, except for some rather extreme cage designs. For short exposure times, the concentration in the SPMD is much smaller than its equilibrium value (i.e. Cs KawCw), and Eq. (7.2) reduces to Ak0 dCs^—^Cwdt (7.6) which yields after integrating over time [3] JdCa~^J Cwdt = ^Cw,TWAt (7.7) where CW;twa is the time-weighted average (TWA) concentration in the water phase. Three names may be used to refer to the initial stage of the sampling process. When Cw is constant with time, the concentration of accumulated contaminants increases linearly with time. This stage of the uptake is therefore called the linear uptake stage. For scenarios where aqueous concentrations vary with time, the concentration in the SPMD is linearly proportional to the TWA concentration, and sampling is called time-integrative. Finally, because the rate of change of concentrations in the sampler is linearly proportional to the aqueous concentration, this initial sampling stage may be called kinetic sampling. An interesting aspect of Eq. (7.7) is that the product AkQt is equivalent to the apparent water volume extracted during the exposure time t. Hence, the product AkQ can be viewed as an apparent water sampling rate (Rs) Ra = kA (7.8) Because Rs represents the volume of water extracted per unit time, it forms a conceptual link between traditional batch water extraction methods and PSD-based methods. Equation (7.8) shows that water sampling rates are linearly proportional to the surface area of the sampler. For this reason, a comparison of sampling rates among different sampler designs only yields meaningful results when differences in surface area are taken into account. 144 Theory, modelling and calibration For very long exposure times and a constant Cw, the concentration in the SPMD does not change with time, and Eq. (7.2) reduces to XV i sw which merely is an expression that the concentration in the SPMD attains its equilibrium value (Cs = KSWCW). The corresponding sampling scenario is called equilibrium sampling. A general solution to Eq. (7.2) for constant Cw is given by [4] Ca = KawCw[l - exp(-ket)] + C0 exp(-ket) (7.10) where C0 is the concentration at t = 0, and the elimination rate constant (ke) is given by ke=inr=inr (7-n) 1 sw v s 1 sw v s Equation (7.10) shows that the uptake from the environment and the elimination of the initial amounts (found in the PSD fabrication controls) are additive. Subtraction of these levels can be problematic when the initial concentration is higher than, or about equal to, the equilibrium concentration. In that case, the concentrations in exposed samplers can be smaller than the concentrations observed in fabrication controls, and control subtraction would yield negative concentrations. Equation (7.10) also shows that the uptake and elimination process of a particular compound are characterised by the same ke value. This observation is the basis of estimating in situ sampling rates from the dissipation rates of performance reference compounds (PRCs) (Section 7.9.4) [11]. When the initial concentration equals zero, Eq. (7.10) takes the form of the more familiar release equation [2] Rtt ' (7.12) C — K C ^s — ^^-sw^w 1 — exp | which in the short time limit reduces to the linear uptake equation s For the dissipation of PRCs that do not occur in the environment (Cw = 0) and that are spiked into the sampler prior to exposure, Eq. (7.10) reduces to the more familiar release equation [2] Ca = C0exp(-ket) (7.14) 145 K. Booij, B. Vrana and J.N. Huckins Aqueous concentrations can be calculated from the amounts (Ns) absorbed by the PSD, the in situ sampling rate of the compounds and their sampler-water partition coefficients, using the rearranged Eq. (7.12) Ns °w = KSWVS[1 - exp(-Rst/KswVs)] (7-15) For equilibrium samplers, the term in square brackets equals 1 to good approximation, and aqueous concentrations are calculated from cw^7fV (7-16) For kinetic samplers, operating in the linear uptake mode, the term in square brackets is approximately equal to (Rst)/(KSWVS), and aqueous concentrations can be calculated from Cw^^ (7.17) The denominators in Eqs. (7.15)-(7.17) can be interpreted as the apparent water volume that is cleared of analyte during the exposure (Fig. 7.2). In the case of equilibrium sampling, this volume is limited by the sorption capacity of the sampler (KSWVS). For kinetic sampling, the apparently extracted water volume is limited by the sampling rate and the exposure time (Rst). 7.3 MODEL APPLICATION TO OTHER PASSIVE SAMPLERS The discussion in the previous section can easily be extended to other passive samplers that contain any number of sub-phases, provided that sorption equilibrium exists at the interfaces and that (pseudo-) steady-state conditions apply within the barriers between the water and the collection phase (i.e. the difference between the inward and outward fluxes for each intermediate phase should be relatively small). Equation (7.5) may then be generalised as [4] T=^mr (7-18) where the summation runs over all phases i. The evolution of the analyte amounts accumulated in the receiving phase (i.e. that part of the sampler that is actually extracted and analysed) is given by 146 Theory, modelling and calibration lime Fig. 7.2. Effectively extracted water volume as a function of time. For long exposure times the extracted volume is constrained by the sorption capacity of the PSD (KSWVS), and at short exposure times by the product of sampling rate and time (Rst). Eq. (7.12), where Ksw takes the general form Ksw = ^^ (7.19) and the sampler volume Vs equals the sum of the volumes of all the sub-phases that are analysed. In the SPME literature, a slightly different (empirical) model is used to describe the sampler-water exchange kinetics [7,12] d* "#Vi~w #v^s = &iCw - k2Cs (7.20) This model is mathematically equivalent to Eq. (7.2), with k2 = (Ak0)KKsyys) and kx = Kswk2. 7.4 VALIDITY OF THE MODEL ASSUMPTIONS For the models above, it was assumed that linear concentration gradients exist in the membrane and in the central phase; that equilibrium exists at the interfaces; that molecular diffusion is the predominant 147 K. Booij, B. Vrana and J.N. Huckins transport mechanism in the membrane with a diffusion coefficient that is independent of time and of concentration. In the initial stages of the exposure, analytes have to penetrate the membrane to get to the central phase. The resulting time lag has been experimentally confirmed to be about 10 h for the uptake of PCB 52 by SPMDs [2]. A theoretical model for the mass flux through a plane sheet with constant concentration on both sides of the sheet predicts a lag time of [13] (7.21) Diffusion coefficients differ widely among polymers. Values for benzene include SxlO^m^"1 in PDMS, 2xlCr12m2s-1 in LDPE, 2xl0~16m2s_1 in poly(methylacrylate) and lxl0~19m2s_1 in poly (vinylalcohol) [14,15]. For 100 |im thick membranes of these polymers Eq. (7.21) predicts lag times of 6 s, 14min, 4 months, and 4 centuries, respectively, and these values are expected to increase with molecular size. Evidently, for WBL-controlled uptake, the analyte distribution within the membrane does not affect the uptake rates. Adopting an aqueous diffusion coefficient of 5 x 10~10m2s_1 and an effective boundary layer thickness of 30-300 |im (Section 7.5), lag times of 0.3-30 s may be expected for WBL-controlled uptake. However, when the membrane is discarded, and only the central phase is analysed, the lag time for membrane passage has to be accounted for, even in the case of WBL-controlled uptake. Linear gradients in the membrane cannot exist when the membrane accumulates analytes, because in this case the flux into the membrane must be larger than the flux out of the membrane. By the same argument, linear gradients cannot exist in the central receiving phase either. The concentration gradient in the middle of the receiving phase (e.g. SPMD, MESCO equipped with a PDMS rod), or next to an impermeable wall (e.g. Chemcatcher, SPME) should be zero. (Otherwise, a discontinuity in the flux would occur.) Yet, the concentration gradient at the outer side of the central phase should differ from zero. (Otherwise, the central phase would not accumulate anything.) Again, for WBL-controlled uptake, the existence of non-linear gradients in the membrane or in the central phase does not invalidate the model, but for membrane-controlled uptake, this phenomenon may have to be accounted for. The non-linearity of concentration gradients can be assessed in terms of an effective phase thickness (<5i;eff) as shown in Fig. 7.1 for the membrane phase. Using an analytical radial diffusion model for uptake by SPME 148 Theory, modelling and calibration fibres, Louch et al. [16] showed that the effective membrane thickness deviates less than 20% from the actual membrane thickness for times that are larger than the lag time (Eq. (7.21)). Using numerical methods, Hofmans [8] obtained similar results for SPMDs. The assumption that instantaneous equilibrium exists at the interfaces is likely to be met for the small mass-transfer rates encountered in passive sampling methods, particularly for rubbery polymers, which are characterised by short relaxation times [14]. Although diffusion coefficients in polymers have been shown to depend on diffusant concentration, this dependence is reported to be weak [14], and can probably be neglected in passive sampling because of the relatively low analyte concentrations encountered. 7.5 WATER BOUNDARY LAYER RESISTANCE Exact models for mass transfer through the WBL exist only for some simple flow arrangements, such as the flow through ducts and pipes, and parallel flow along an absorbing flat plate [17-20]. Starting at the leading edge of the plate, the momentum of the water that is immediately adjacent to the plate is reduced due to surface friction. As the water moves along the plate, this retarded water layer in turn attenuates the momentum of the water layers at larger distance from the surface, which results in the development of a viscous sublayer, with a thickness that increases with distance downstream of the leading edge. Similarly, analytes are removed from a layer with a thickness that increases downstream, leading to the development of a concentration boundary layer. With increasing thickness of this layer, transport by eddy diffusion becomes increasingly important, since turbulent diffusion coefficients increase with increasing distance from the surface [19,21]. At large distances from the leading edge, a steady-state concentration profile is established that no longer depends on the distance along the plate. Equations for the short-plate limit (growing concentration boundary layers) and the long-plate limit (distance-independent concentration boundary layers) have been given by Opdyke et al. [22] for hydrodynamically smooth flows (i.e. flows along surfaces where the roughness elements are embedded in the viscous sublayer). The (surface averaged) mass-transfer coefficients for the short-plate limit are given by [22,23] (7.22) 149 K. Booij, B. Vrana and J.N. Huckins where v is the kinematic viscosity of the water, L is the length of the plate, and u* is the friction velocity, which is frequently used in the literature on hydrodynamics to parameterise the shear stress (t) u* = (7.23) where p is the density of water. In turbulent flows, u* can be interpreted as the characteristic eddy velocity relative to the main stream [24,25]. The friction velocity for an essentially laminar flow along a flat surface is related to the free-stream velocity (U) by [18] ul = 1.328í/V— (7.24) Equation (7.24) is arranged so as to stress that it is dimensionally consistent (i.e. u* has the same dimension as the main stream velocity U, and v/(UL) is dimensionless). The transition from laminar to turbulent flow takes place at values of UL/v > 4 x 106 when special precautions are taken to reduce the turbulence intensity of the main flow [17]. When no such precautions are taken, the transition to turbulent flow takes place at lower values, i.e. UL/v > 350,000 to 500,000, depending on the turbulence intensity of the main flow [17]. In the long-plate limit, the mass-transfer coefficients are given by [19,22] ÍD \2/3 K = 0MuA-^-\ (7.25) and u* (for fully developed turbulent flow) may be estimated from the free-stream velocity by [18] í4 = 0.074í/2(^-)1/5 (7.26) For more complex scenarios, such as mass transfer for cylinders and packed bed reactors, empirical correlations have been established of the form [18,26] Sh = BRem (7.27) ReSc1/3 where the (dimensionless) Sherwood (Sh), Reynolds (Re) and Schmidt (Sc) numbers are defined by Sh = ^ (7.28) 150 Theory, modelling and calibration (7.29) Sc = v (7.30) D w where d is a conveniently chosen characteristic length scale and u a characteristic velocity. The constant .B in Eq. (7.27) is of the order 1 and m k—0.5 (range—0.3 to —0.7). For the case of mass transfer to a cylinder with its main axis perpendicular to the flow, d equals the diameter of the cylinder, B = 0.6 and m = —0.487, which is valid for the range 100 < Re < 3500 and 1000 1 (7.37) w where Da is the diffusion coefficient in air and Kaw the dimensionless Henry's law constant. Calculated ratios (DaKaw/Dw) ranged from 520 for HCB to 0.1 for jS-HCH and <5-HCH. For compounds with Kaw values > 10~4, the effect of an unfavourable air-water partition coefficient is offset by a more favourable diffusion coefficient in air (~5x 10-6m2s-1) compared with that in water (—5 x 10~10m2s_1). Similar observations have been made for a Chemcatcher sampler, equipped with a central compartment of Ci8-coated silica and an LDPE membrane [5]. 157 K. Booij, B. Vrana and J.N. Huckins Replacing water by air as intermediate phase resulted in an increase in sampling rates up to a factor of 6. Decreasing sampling rates were observed for the 5-ring PAHs, which showed a decrease in sampling rate by a factor of 2-3 as a result of their very low Kaw values (<3 x 10~5). The use of 1-octanol as intermediate phase resulted in an approximately 20-fold increase in sampling rates compared with water as intermediate phase [5]. It should be noted again, however, that reducing the transfer resistances of the internal phases, enhances the relative importance of the mass-transfer resistance of the WBL (Eq. (7.5)), and hence the sensitivity of the sampler to changes in flow conditions. 7.9 CALIBRATION 7.9.1 Static exposure design In the experimentally convenient static exposure scenario, passive samplers are exposed in a single volume of contaminated water. This method has been used in the past for determining bioaccumulation factors and uptake rates of contaminants by fish and mussels. The evolution of aqueous concentrations in the exposure water is given by [58-60] Cw0U + KawVa C-w — V. exp w 1 + KawVa- Rj V w KawVa 1 + KawVa (7.38) V. w where Cw0 is the aqueous concentration at t = 0. The concentration in the sampler can be evaluated from the mass balance (VSCS = Vw[Cw0 - Cw]) Cw0Kaw{ 1 - exp Ca = KawV„ « KawVa (7.39) which reduces to Eq. (7.12) in the limit Vw^oo. With Eqs. (7.38) and (7.39) it is assumed that there are no competing sorption phases (equipment and particulate/dissolved organic matter) in the exposure system. In the short time limit, Eq. (7.38) may be approximated by (7.40) 158 Theory, modelling and calibration and the concentration in the sampler may be approximated by CwoRst i lRst 1 Rst \ (7.41) Vs { 2VW 2KSWVS When the concentration in the sampler is much lower than its equilibrium value (i.e. Rat KawVa), the third term between the parentheses in Eq. (7.41) may be neglected, and Eq. (7.41) reduces to where Cw;twa is the TWA concentration during the exposure. Static exposures have been used in the calibration of SPMDs and similar samplers [9,11,28,59,61] and also is the typical calibration scenario in SPME research [36,60,62]. Equilibration times obtained with static exposures are sometimes erroneously assumed to also apply to field exposures [59,61]. Equation (7.39) shows that the evolution of analyte concentrations in the samplers follows first-order kinetics, with a rate constant that is dependent on the water volume, among other factors. High rate constants can be found when the water volume is small compared with the sorption capacity of the sampler (Vw KSWVS). In this case, the rate constant is approximately equal to RJVW. However, the water volume in the field is essentially infinite (Vw ;» KawVa), and the rate constant for the attainment of equilibrium equals RJ(KSWVS) in that case. The intuitive explanation of short equilibration times that may be observed in static exposure designs is that both the accumulation in the sampler and the depletion of the water favour the attainment of equilibrium [63]. By contrast, depletion of the water phase in the field is insignificant. 7.9.2 Static renewal design In static renewal designs, the exposure water is refreshed batchwise [41,54]. This design may be used when static or continuous flow exposure designs are not an option. This may occur, for example, when a static exposure would result in an excessive depletion of the water phase, or when problems occur in maintaining stable aqueous concentrations during flow-through exposures. Aqueous concentrations should be measured at least at the beginning and at the end of each renewal period, in order to estimate their average. Uptake curves may be generated when it can be assumed that the amounts removed from the water are absorbed by the sampler (i.e. loss terms like evaporation Cs = (7.42) 159 K. Booij, B. Vrana and J.N. Huckins and wall sorption, as well as sorption on to dissolved/particulate matter can be neglected) and that the average aqueous concentrations do not vary greatly among renewals. Even then, the mathematical modelling of such data is not so easy, except for the case of kinetic sampling over the entire exposure period (Eq. (7.42)). 7.9.3 Continuous flow design Continuous flow designs aim at preventing depletion of the water phase during the exposure by ensuring a constant supply of freshly contaminated water to the exposure chamber. As with the static and static renewal designs, sorption to dissolved/particulate matter should be negligible in order to prevent overestimating Cw. However, sorption to the equipment used in the exposure system has no detrimental effect, provided that the equipment has equilibrated with the water. Stable aqueous concentrations can be maintained during the entire exposure if the flushing rate (Q: volume per unit time) of the exposure chamber is much larger than the total sampling rate of all samplers [30] Q » nRs (7.43) where Rs is the sampling rate per sampler and n the total number of samplers in the exposure system. For example, an exposure system that contains five passive samplers that have a sampling rate for a particular compound of 4 L day-1 would require a flushing rate at the beginning of the experiment, that is much higher than 20 L day-1. Such a set-up would therefore require a flushing rate of at least 100 L day-1 of water with dissolved organic carbon (DOC) levels that are low enough to ensure that contaminant sorption to DOC is insignificant. With the gradual removal of samplers during the experiment, the flushing rate may be reduced, provided that the hydrodynamic conditions in the exposure chamber can be kept constant, e.g. by additional stirring or by recirculation pumping. Because sampling rates are linearly proportional to the sampler surface, the use of smaller samplers may help to reduce the water demand. It should be realised in this case, however, that for WBL-controlled uptake the sampling rate may be a weak function of the sampler length (Eq. (7.22)). Mixing of stock solutions in methanol or acetone is the most widely used method for preparing contaminated water needed in the exposure experiments [2,6,32], but generator column techniques based on Ci8-coated silica [28,30] or permeation through a dialysis membrane [64] have also been used. 160 Theory, modelling and calibration When constant aqueous concentrations can be maintained during the entire experiment, sampling rates and sampler-water partition coefficients may be obtained by curve fitting of Eq. (7.12). In case the extent of equilibrium attained is insufficient to estimate Ksw, the linear uptake equation (Eq. (7.13)) should be used. Decision methods for selecting the correct model are discussed elsewhere [28,65]. Slightly more complicated models should be used when aqueous concentrations are not sufficiently constant during the exposure. Suppose that the aqueous concentrations can be described by a second-order polynomial in time Cw(t) = Co + dt + C2t2 (7.44) the solution to the differential equation (Eq. (7.2)) can be found as [66] Cs = (Co - ^ + ^Vl - expHM)] + (Ci - + C2t2 (7.45) where ke is given by Eq. (7.11). The solution for constant concentrations (Eq. (7.12)) and aqueous concentrations that vary linearly with time [30] can be seen to be special cases of Eq. (7.45). 7.9.4 In situ calibration The evaluation of dissipation rate constants of PRCs has been used as a method for calibrating the uptake rates of PSDs in situ [2,11,65,67-69]. When PRCs are selected that do not occur in the environment in significant amounts (e.g. 13C-labelled PCBs or perdeuterated PAHs), their dissipation rate constants can be estimated from the rearranged Eq. (7.14) k, = -^m (7.46) where C0 is the PRC concentration at t = 0. Consequently, the sampling rate of this PRC can be obtained from the rearranged Eq. (7.11) Rs = keKswVs (7.47) PRCs can be used only if their dissipation rate is large enough to quantify the difference in PRC concentration at the beginning and at the end of the exposure. Analytical precision is the controlling factor in this case. For compounds with large dissipation rates, detection limits may be an issue. As a result, PRC-derived sampling rates can be obtained only for compounds that span a 1.5 log units wide range in log Kow. In the case of SPMDs, this range spans log Kow values between 4.5 and 6, but for PSDs 161 K. Booij, B. Vrana and J.N. Huckins with smaller sorption capacities, these values may be shifted towards the higher Kow end. Extrapolation of PRC-based sampling rates to compounds with much lower log Kow values is not so critical, because these compounds will have attained a substantial, if not complete, degree of equilibrium, and Eq. (7.15) is quite insensitive to uncertainties in sampling rates for this group of compounds. However, uncertainty exists on the question of how PRC-based sampling rates should be extrapolated to the high log Kow range. Huckins et al. [11] defined the exposure adjustment factor (EAF) as the ratio of the (PRC-based) sampling rate in the field and the sampling rate of compounds with the same physicochemical properties obtained during laboratory calibration studies These authors showed that the EAF is only a weak function of log Kow, and that PRCs may be used to reduce the effect of exposure conditions on sampling rates from 3- to 10-fold to about 2-fold. The EAF approach has recently been generalised [3]. An alternative method of using PRC-based sampling rates to estimate Rs values of more highly hydrophobic compounds is based on the assumption that the conductivity of the WBL is proportional to D2J3 [3,30,69]. Since Dw is a weak function of molecular size, Rs can be estimated from [3] where Vprc and V are the LeBas molar volumes of PRC and analyte respectively. The PRC should be subject to WBL-controlled kinetics, in this case. Experimental sampling rates for WBL-controlled uptake decrease much stronger with molecular size than indicated by Eq. (7.49), but this decrease may well be caused by the overestimation of concentrations of dissolved analyte, due to sorption to DOC [3,5,30]. However, to date, experimental proof of this assumption is not available. 7.10 CONCLUSION AND OUTLOOK Considerable progress has been made in understanding the factors that control hydrophobic organic contaminant uptake by passive samplers. • Transfer through the water boundary layer generally is the rate-limiting step for the uptake of highly hydrophobic compounds. As a EAF = (7.48) (7.49) 162 Theory, modelling and calibration result, the sampling rates (Rs) for these compounds depend on the hydrodynamic conditions at the exposure site. Unfortunately, sampling rates for these compounds are difficult to estimate from the local flow velocities and turbulence intensities, and in situ calibration techniques based on the dissipation of performance reference compounds (PRCs) are necessary. • Diffusion through the membrane is the rate-limiting step for compounds with low membrane-water partition coefficients (Kmw). Sampling rates for these compounds are only dependent on temperature, and sampling rates obtained in the laboratory can be applied in the field. • Attempts have been made to eliminate the flow-dependency of sampling rates for highly hydrophobic compounds, by adding additional transport barriers in the sampler and by using more polar membranes. These attempts have been unsuccessful due to a dramatic drop in sampling rates, resulting in detectability problems. • The dissipation of PRCs allows for estimating sampling rates in situ. This technique is hampered by the limited range of log Kow values (4.53) density polyethylene (LDPE) Hydrophilic organic Cis Empore™ disk Microporous compounds (log Kow<3) polysulfone (PS) SDB-RPS Empore™ Microporous disk polyethersulfone (PES) Metals Chelating Empore Microporous cellulose disk acetate (CA) Mercury Chelating Empore Microporous disk polyethersulfone (PES) Organotin compounds Ci8 Empore™ disk Microporous cellulose acetate (CA) Empore extraction disks were selected as convenient receiving phases for use in the Chemcatcher samplers. They are available as standard 47-mm diameter sorbent particle loaded disks. The particles are held together within an inert matrix made of polytetrafluoro-ethylene (PTFE) (90% sorbent: 10% PTFE, by weight). The variety of sorbent materials used in the Empore disk technology enabled the selection of suitable receiving phases for all classes of pollutants under investigation, including polar and non-polar organic analytes, organo-metallic compounds and metals (Table 9.1). A further advantage is the availability of published extraction protocols for a number of analytes and a simple analyte elution with consistent recoveries. Moreover, procedures enabling the disks to be loaded (using procedures developed for solid-phase extraction) in a reproducible manner with internal standards or performance reference compounds (PRCs) by filtering an aqueous standard solution through the disk were developed [1]. 9.2.2 Diffusion membranes Two types of polymeric membranes have been tested for construction of Chemcatcher samplers; non-porous membranes including LDPE and 201 R. Greenwood et al. microporous membranes including glass fibre, nylon, polycarbonate, PTFE, polyvinylidenedifluoride (PVDF), cellulose acetate (CA), poly-sulfone (PS), polyethersulfone (PES) and regenerated cellulose. The membranes separate the sorption phase from the bulk water phase, and reduce the flux to the sorption phase. The membrane acts as a semipermeable barrier between the receiving phase and the aqueous environment. The dissolved analytes can pass through to the receiving phase, while particulates, microorganisms and macromolecules with a size greater than the exclusion limit cannot permeate. Without the protection of the membrane, there is a risk of deterioration of the receiving phase disks in the aqueous environment due to biofouling. The criteria for selecting an optimum membrane for sampling a specific group of analytes have been discussed in Chapter 7. The physical strengths, handling properties and chemical resistance of membrane materials were assessed during the initial evaluation. These tests were followed by accumulation studies of test analytes in prototype devices fitted with different membranes in a flow-through system. The latter studies were designed to determine the conductivity to mass transfer of membranes for a broad range of organic and organometallic pollutants and metal ions. Differences in conductivity of various membrane materials are shown in Fig. 9.1. In this first evaluation stage, optimum combinations of diffusion membrane/receiving phase systems were selected for a comprehensive evaluation, including calibration in the laboratory and testing in the field (Table 9.1). PS and PES membranes were selected for sampler devices designed to sample polar organic pollutants (log KqW<3) and mercury. These membranes have a high degree of physical strength and good antifoul-ing properties, due to their low surface energy that prevents adsorption of macromolecules to the surface. Polar molecules readily diffuse through the 0.2-Lim wide water-filled pores. In contrast, more hydrophobic compounds sorb to the polymer matrix of the membrane. Due to low diffusivity in the polymer matrix, conductivity of the membrane decreases dramatically with increasing hydrophobicity of sampled compounds. CA was selected as a material suitable for construction of Chemcatcher samplers for inorganic ions and organotin compounds, due to their optimum diffusion through the water-filled membrane pores, combined with negligible adsorption to the membrane material. The non-porous LDPE allows permeation of hydrophobic analytes (log Kow>3-4), due to the favourable combination of high membrane/ water partitioning coefficients and membrane diffusivities for those compounds (see Chapter 7). On the other hand, the membrane has a 202 Monitoring of priority pollutants in water Membrane type Fig. 9.1. The effect of diffusion membrane materials on the patterns of uptake of seven organic compounds. The exposure was performed at constant analyte concentration in water at 11°C in a flow-through tank. A 47-mm Cis Empore M disk was used as receiving phase in all cases. high resistance to mass transfer of more polar compounds and completely excludes the permeation of ions and molecules with effective diameter larger than 1 nm.This material was used in the Chemcatcher designed to sample non-polar organic pollutants. 9.2.3 Sampler body 9.2.3.1 Reusable sampler body prototype The principles of Fickian diffusion state that the flux of a substance to the receiving phase is proportional to the surface area over which diffusion takes place and is inversely proportional to the diffusion path length. Therefore, if passive sampling obeys Fickian diffusion, the physical dimensions of the sampler body significantly affect the sampling rate for analytes. During the development phase, the design of the Chemcatcher body was optimised in terms of both construction materials and sampler geometry. 203 R. Greenwood et al. In the evaluation stage, PTFE was selected as a construction material for the sampler body. Its advantage is a low sorption capacity for most environmental pollutants. Moreover, PTFE is denser than water and is not buoyant in the sampled environment, making it easy to deploy this prototype in the field by suspending it from a wire or a string. The system was constructed to fit a 47-mm Empore disk as the receiving phase, with the chosen diffusion membrane material being laid directly on its surface. Both were supported by means of a 50-mm rigid PTFE backing plate (Fig. 9.2). The active surface area of the Chemcatcher sampler is 17.5 cm2. To seal the sampler, a sleeve open at the back was screwed into place to hold the individual body sections together. In addition, a sealing plate allowed the system to be filled with water and sealed during storage and transport. Thus, the sampler body also acts as a container for storage and transport. The PTFE body could be reused several times, but only after a thorough cleaning involving a multi-step washing procedure. 4- 70 mm -». —_ 50 mm _ 0 r r 5 m Fig. 9.2. Schematic diagram of the prototype Chemcatcher device, used during the sampler development. The PTFE body parts (components 1 and 4) support the receiving phase (component 2) and the diffusion membrane (component 3) and sealed them in place. The sampler is sealed by means of a screw cap (component 5) for storage and transport. 204 Monitoring of priority pollutants in water In the early stages of development [2], a protective steel mesh was used to protect the surface of the membrane. However, the use of a mesh was later abandoned, because it proved to accumulate particulate matter in the field and also to provide shelter for colonising organisms that cause fouling or degradation of the membrane. 9.2.3.2 Disposable sampler body prototype In subsequent performance tests, the uptake kinetics of many analytes were shown to be controlled by diffusion in the aqueous boundary layer on the membrane surface. The resistance to mass transfer of the boundary layer depends on hydrodynamic conditions in the membrane vicinity. These are significantly affected by the construction geometry of the sampler body. The membrane and receiving phase of the first-generation Chemcatcher prototype were located inside a 20-mm deep depression in the sampler body. This sampler design effectively buffers the effect of fluctuating flow on the sampler performance. However, it also effectively reduces convective transport of analytes to the sampler membrane, causing reduced sampling rates (i.e. the rate at which the sampler accumulates chemicals). For an optimum sampler performance, high sampling rates are essential, especially for sampling non-polar chemicals, due to their extremely low concentrations in the water column. In order to increase sampling rates, the geometry of the body was further refined in the latest version of Chemcatcher body prototype by reducing the depth of the cavity to a minimum (Fig. 9.3). In comparison with the first-generation prototype, the second-generation sampler collects analytes with increased sampling rates. Tests showed that the sampling rate for non-polar compounds (log.Kow>3-4), which are accumulated under aqueous boundary layer control, was increased by a factor of 2. This provides improved sensitivity, but also increased variation of sampling rates in response to fluctuations in turbulence (water flow). In the latest design, the Chemcatcher body is made of mouldable plastic materials. The body consists of three components (two body parts and a lid for storage and transport), which are clipped together (Fig. 9.3). This makes the sampler assembly and disassembly faster than it was in the first-generation prototype, where assembly was made using screw threads. This sampler body prototype was designed as a disposable device for a single field deployment. This removes difficulties connected with cleaning procedures and accompanying quality control measures required for use in trace analysis. The plastic material can be recycled. 205 R. Greenwood et al. Fig. 9.3. Views of the disposable Chemcatcher sampler. Depending on the nature (temperature, turbulence, presence of suspended solids) of the environment to be sampled and on the target analyte properties, a sampler design can be selected to provide an optimum performance. 9.3 THEORY The general theory of passive sampling is described in Chapter 7, and this is applicable to the various Chemcatcher designs. In summary, mass transfer of a chemical into the sampler involves several diffusion and interfacial mass transport steps across the various barriers that may be present; i.e. the stagnant aqueous boundary layer, possibly a biofilm, the diffusion membrane, the inner fluid (liquid or gaseous) phase, and the receiving phase. In the initial stages of exposure, analyte uptake is expected to be linear or time-integrative after steady-state flux of chemicals into the sampler has been achieved [3,4]. Under these conditions, the amount of a chemical in the receiving phase is directly proportional to the product of the concentration in the surrounding water (Cw) and the exposure time (t). For practical purposes, uptake in 206 Monitoring of priority pollutants in water the linear phase can be described by mD(i) = m0 + CwRst (9.1) where mD is the amount of analyte accumulated in the receiving phase, m0 is the initial amount of analyte in the receiving phase, and R$ is the sampling rate of the system: where kOY (m s-1) is the overall mass transfer coefficient and A (m2) is the surface area of the membrane. The uptake of an analyte is linear and integrative approximately until the concentration factor of the sampler (mD(ř)/Cw) reaches half saturation. The sampling rate of an individual chemical can be determined experimentally under fixed conditions at constant analyte concentration. Under environmental conditions, when the water concentration changes during the exposure, the term Cw represents a time-weighted average (TWA) concentration during the deployment period. 9.4 CALIBRATION The sampling rate depends on the physicochemical properties of the analyte, the environmental conditions and the sampler design. To enable measurement of TWA water concentrations of a range of pollutants, the Chemcatcher sampler was calibrated in flow-through tank studies under controlled conditions of temperature and water turbulence. Concentrations of the analytes in water (Cw) and the amounts accumulated in the receiving disk (mD) were measured regularly during the exposure. In each experiment, passive samplers were exposed for up to 14 days in a constant concentration of analyte. Each factor (temperature and stirring speed (turbulence)) was tested at three levels. The calibration experiments were designed to characterise the effect of physicochemical properties, temperature and hydrodynamics on kinetic and thermodynamic parameters characterising the exchange of analytes between the sampler and water. So far, calibration data have been reported for the non-polar Chemcatcher [1,5] and calibration data for other Chemcatcher designs will be reported shortly [6,7]. 9.5 SAMPLING OF HYDROPHOBIC ORGANIC CONTAMINANTS Kingston et al. [2] designed one of the Chemcatcher prototypes for the sampling of non-polar organic compounds with log Kqw values greater Rs - kovA (9.2) 207 R. Greenwood et al. than 3. This system uses a 47-mm Ci8 Empore disk as the receiving phase and a 35-um thick LDPE diffusion membrane. The Ci8 Empore disk has a very high affinity and capacity for the sampled hydrophobic organic pollutants. LDPE is a non-porous material, even though transient cavities with diameters approaching about 1 nm are formed by random thermal motions of the polymer chains. The thermally mediated transport corridors of the polyethylene exclude large molecules, as well as those that are adsorbed on sediments or colloidal materials such as humic acids. Only truly dissolved and non-ionised contaminants are sequestered. Recently, the optimisation of this sampler design has been reported [8]. This involved the improvement of sampling characteristics including the enhanced sampling kinetics and precision by decreasing the internal sampler resistance to mass transfer of hydrophobic organic chemicals (\og KqW>5). This was achieved by adding a small volume of rc-octanol, a solvent with high permeability (solubility x diffusivity) for target analytes, to the interstitial space between the receiving sorbent phase and the polyethylene diffusion membrane. The use of rc-octanol as an interstitial phase resulted in an approximately 20-fold increase in sampling rates compared with those observed with water as the interstitial phase [8]. 9.5.1 Calibration data Calibration data for the non-polar Chemcatcher were obtained in laboratory experiments designed to measure the uptake of target analytes (sampling rate; R$) and offloading of PRCs (elimination rate constants; ke) at different combinations of temperature and hydrodynamic conditions in a full factorial design. The calibration data were gathered in order to determine the sampling parameters and to observe how they are affected by environmental conditions to enable a more precise measurement of TWA concentrations of non-polar priority pollutants in the field [1]. Over the range of controlled laboratory conditions (temperature and turbulence), the magnitude of R$ values of hydrophobic chemicals spanned over two orders of magnitude (i.e. from 0.008 L day-1 up to 1.380 L day-1). The sampling rate is strongly affected by the physico-chemical properties of the compounds. Among the non-polar priority pollutants, the highest sampling rates were observed for small, moderately hydrophobic compounds: anthracene, phenanthrene, fluoran-thene and pyrene. The lowest sampling rates were measured for 208 Monitoring of priority pollutants in water Fig. 9.4. Effect of water turbulence (expressed as rotation speed of a carousel device loaded with samplers) and log K0w on the sampling rates for a range of non-polar organic compounds in the Chemcatcher at 11°C. indeno[l,2,3-cd]pyrene, dibenz[a,h] anthracene and benzo[g,h,i]perylene; large and extremely hydrophobic compounds. The typical dependence of sampling rates on hydrophobicity is shown in Fig. 9.4. Sampling rates increase with the increasing temperature, and the temperature dependence of the sampling rate Rq can be described by an Arrhenius-type equation. The mean activation energy for all of the hydrophobic analytes under investigation was gSkJmol-1. This corresponds to an increase in sampling/offload rate of a factor of 5.2 over the temperature range 6-18°C. For comparison, Huckins et al. [9] calculated from the literature data available for semipermeable membrane devices (SPMDs) an average activation energy of 37 kJ mol-1. Thus, the effect of temperature on the Chemcatcher uptake kinetics appears to be more significant than that on SPMD sampling rates. With the exception of the moderately hydrophobic lindane (log Kqw = 3.7), a significant increase in sampling rate with increasing flow velocity was observed for all compounds under investigation (Fig. 9.4). This corresponds well with the theory of diffusion through two films in 209 R. Greenwood et al. series [10,11], which predicts a switch in the overall mass transfer to the aqueous boundary layer control for hydrophobic compounds. A similar effect of hydrodynamics has been observed and explained for SPMDs [12]. 9.5.2 Performance reference compound concept Figure 9.5 shows that for a range of environmental conditions (temperatures and water flow rates) there is a good correlation between uptake kinetics (sampling rate R$) of analytes and offload kinetic parameters (elimination rate constant ke) of their deuterated analogues (used as PRCs). This demonstrates isotropy of the uptake (absorption) onto and the offload (desorption) from the sampler for a range of hydrophobic analytes. Thus, the PRC concept can be applied to the measurement of in situ exchange kinetics in the field. 1.0 0.8 0.6 i dc 0.4 0.2 0.0 • Acenaphthene O Fluorene T Phenanthrene 0.00 0.02 0.04 0.06 0.08 0.10 0.12 K Id"1] Fig. 9.5. The correlation between the sampling rates Rs of three polycyclic aromatic hydrocarbons and the elimination rate constants ke of their per-deuterated analogues demonstrates the isotropic exchange kinetics for the non-polar Chemcatcher sampler variant. The data represent laboratory flow-through exposures performed at various combinations of water temperature and turbulence. Reproduced from Ref. [1] with permission from Elsevier. 210 Monitoring of priority pollutants in water 9.5.3 Non-polar Chemcatcher/water distribution coefficients Assuming isotropy of the exchange kinetics of the chemicals under investigation, and the validity of the model used to describe the kinetics, the value of the receiving phase water distribution coefficient iřDW can be calculated as the ratio of the absorption and desorption transport parameters for a particular compound (see also Chapter 7): k™ = t^t- (9-3) The experimental evidence indicates that iřDW values are not significantly affected by temperature in the range 6-18°C. This enables the derivation of an empirical equation to calculate the distribution coefficient .Kdw of a compound between the non-polar Chemcatcher sampler and water using the rc-octanol-water partition coefficient: logifDW = 1.382 logifow - 1-77 (r = 0.97, s = 0.13,,i = 31) (9'4) Huckins et al. [9] have shown that for SPMDs, the log Kow versus log SPMD/water partition coefficient plot for compounds with log Kow>5.0 deviated from linearity. This phenomenon has also been observed for plots of log bioconcentration factor versus log Kow [13]. It has not yet been demonstrated whether or not a deviation from linearity occurs for very hydrophobic compounds in the non-polar Chemcatcher. 9.5.4 Empirical uptake rate model It is convenient to derive an empirical equation for the in situ estimation of sampling rates for use in the interpretation of results obtained with the Chemcatcher passive sampler in field studies. Huckins et al. [9,14] showed that for SPMDs, differences in exposure conditions cause sampling rates to be shifted by a constant factor for all compounds. A similar observation was made for the non-polar Chemcatcher. Log R$ versus log Kow plots from all calibration studies for the Chemcatcher have very similar shapes, but show a varying offset for the different exposure conditions (combinations of water temperature and turbulence). A nonlinear regression analysis of log-transformed sampling rates R$ on a third-order polynomial function of log Kow from all calibration studies enabled the derivation of an empirical model that can be used to calculate the sampling rate as a function of hydrophobicity. 211 R. Greenwood et al. This relationship is applicable within the range of log Kow 3.7 to 6.8 and for a range of exposure conditions (temperatures between 6 and 18°C and water turbulence (stirring speeds from 0 to 70rpm)): logics =P +22.755 logKow- 4.061 log2if0w + 0.2318 log3if0w (r = 0.92,s = 0.22,rc = 134) (9.5) The relative ratios of sampling rates of any two compounds within the calibration range are constant for a broad range of exposure conditions. The knowledge of the parameter P is sufficient to characterise the effect of varying environmental conditions on the absolute magnitude of the sampling rates. The standard deviation of the fit (0.22 log units) corresponds to an uncertainty factor of approximately 1.7, which is relatively low considering the large differences in exposure conditions tested. Information on concentrations, that are accurate within a factor of 2, is still highly relevant for environmental risk assessment purposes. 9.5.5 Estimation of in situ TWA concentrations An algorithm has been derived to calculate TWA water concentrations from the amounts of analytes accumulated in non-polar Chemcatcher samplers during field deployment [5]. This involves the characterisation of in situ exchange kinetics, using PRCs. The PRC elimination rate constant ke is calculated using two points: amount of PRC in a sampler prior to and after a field exposure. Isotropic first-order exchange kinetics are assumed. Sampling rates R$ of PRCs are calculated using Eqs. (9.3) and (9.4). The PRC-derived sampling rates are then fitted to Eq. (9.5), using the exposure-specific effect P as the only adjustable parameter. The sampling rates of individual compounds are then estimated from Eq. (9.5) with the optimised value of parameter P. TWA concentrations of target analytes at the sampling site can be estimated from concentrations in the exposed samplers using the rearranged Eq. (9.1): CW = mB(t)- mm (9.6) Rst where Cw represents the TWA water concentration during the deployment period, mD(ř) is the analyte mass found in the sampler after field exposure, mDf is the average mass of analyte found in the field blank, Rs is the estimate of the in situ sampling rate derived as described above and t equals exposure time. 212 Monitoring of priority pollutants in water 9.6 SAMPLING OF HYDROPHILIC ORGANIC CONTAMINANTS 9.6.1 Integrative sampler Kingston et al. [2] designed a Chemcatcher prototype for integrative sampling of polar organic compounds with log Kqw values lower than 3 over long exposure times. This system uses a 47-mm Ci8 Empore disk as the receiving phase and a 100-Lim thick PES diffusion mem-brane. The Ci8 Empore disk, used as a receiving phase in this Chemcatcher prototype, has been shown to have a high affinity and capacity for many organic pollutants. The octadecyl functional groups bonded to the silica surface provide mainly non-polar interactions with hydrophobic molecules. However, a fraction of the silica material has non-substituted silanol groups with a high affinity for molecules with polar functional groups. These interactions involve mainly hydrogen bonding or dipole-dipole interactions. Thus, this sorbent disk exhibits can retain analytes with a broad range of physicochemical properties. As described earlier, PES is a porous membrane with a high permeability for polar organic chemicals. This material has also been used in other passive samplers, e.g. polar organic chemical integrative samplers (POCIS) [15] (also see Chapter 8). Retention of some polar compounds on Ci8 Empore disks is stronger than one would expect from their hydrophobicity. This high receiving phase affinity permits the sampling of pollutants over a prolonged period without reaching the saturation of the sorbent material. On the other hand, this high affinity complicates the selection of compounds with a medium sampler fugacity that could be used as PRCs, since offloading rates are extremely low and it is not possible to measure in situ analyte exchange kinetics. This is shown in Fig. 9.6. Linear uptake of atrazine (a compound with relatively low hydrophobicity: log Kow = 2.61) into the Chemcatcher was observed during a period of 14 days under a range of exposure conditions. No significant elimination of Z)5-atrazine, loaded onto the Empore disk prior to exposure, was observed over this period. This demonstrates an ideal performance of this variant of Chemcatcher as an integrative sampler for polar compounds. However, it is impossible to see whether the uptake kinetics of atrazine was correlated with the elimination kinetics of _D5-atrazine. Thus this compound cannot be used as a PRC in the time scale of a typical field exposure. Several other compounds, including Z)5-atrazine, .Dio-chlorpyrifos, Z)8-naphthalene, Di0-simazine and Di4-trifluralin, were tested and none was identified to be suitable as a potential PRC. 213 R. Greenwood et al. Q 700 600 500 400 300 200 100 0 o o o o o o o ° • ° • atrazine o D5-atrazine -1-1-1-1-1-1-1- 0 50 100 150 200 250 300 350 Time [h] Fig. 9.6. Uptake of atrazine in the Chemcatcher prototype fitted with Ci8 Empore™ disk and a polyethersulfone membrane in a flow-through laboratory exposure (14 days). No significant elimination of D5-atrazine, loaded onto the Empore™ disk prior to exposure was observed. Data are presented from an exposure conducted at 4°C in turbulent water (rotation speed 70rpm). The aqueous concentration of atrazine was held constant at 1 LigLr1, and the water-exchange rate in the flow-through system was 50 L day \ Calibration data for the polar variant of Chemcatcher were obtained in laboratory experiments in a similar experimental set up as described in Section 9.5.1. Experiments were designed to determine sampling rates R$ of a selected number of triazine and phenylurea herbicides for various combinations of temperature and hydrodynamic conditions. An example of sampling rates of the triazine herbicides is shown in Fig. 9.7. The sampling rates increase with increasing temperature, and the activation energy for the triazine herbicides under investigation (simazine and atrazine) was 130kJmol_1. This would correspond to an increase in R$ of nearly a factor 10 over the temperature range 6-18°C. Thus, the temperature dependence of sampling rate for devices fitted with PES membranes seems to be greater than for those fitted with LDPE membranes. On the other hand, the observed effect of hydrodynamic conditions on sampler performance was only moderate. 214 Monitoring of priority pollutants in water (A) (B) 0.30 0.25 ^ 0.20 i -i 0.15 0.10 0.05 0.00 iff ■ Simazine -Atrazine 10 15 Temperature [°C] 20 0.06 t 0.05 ■ 0.04- I ■o ■ _l 0.03 J Jf) EC 0.02 - 0.01 ■ 0.00 ■■ .Simazine • Atrazine 20 40 60 80 Stirring speed [rpm] Fig. 9.7. Effect of temperature (A: measured in turbulent water) and water turbulence (B: expressed as rotation speed of a carousel device loaded with samplers; measured at 11°C) on the sampling rate of the polar Chemcatcher for triazines. 9.6.2 Short pollution event detector Many pesticides, some of which are polar molecules, are released at high concentrations into streams and rivers in episodic events, such as field runoff after pesticide spraying, heavy rain and storm events, or during wastewater discharge. These events usually last only a few hours and in order for these compounds to be detected by passive samplers, a device with a short response time is required. However, the device fitted with a PES membrane, although ideal for long-term monitoring, has a lag phase of several hours that represents the time necessary for the analytes to diffuse through the membrane to reach the receiving phase. The lag phase of the device can be predicted using a theoretical model for the mass flux through a plane sheet with constant concentration on both sides of the sheet, as outlined in Chapter 7. Since the PES membrane is discarded before analysis (only the receiving phase is analysed), the lag time for passage through the membrane has to be taken into account. Shaw and Muller [16] suggested the use of a device fitted with only an Empore disk receiving phase (without a diffusion membrane) to reduce the response time and make the sampler more reactive to ac-cidental pollution events. The naked Empore disks deployed in stainless steel cages secured between two squares of wire mesh that 215 R. Greenwood et al. allowed the disks to be exposed on both surfaces. Later, Stephens et al. [17] used a device with a naked Empore disk fitted in the Chem-catcher PTFE body, and accumulation in such a device is shown in Fig. 9.1. Such samplers have a very short lag phase that represents only the time taken for the analyte to diffuse across the aqueous boundary layer. The analyte sampling rates are higher than in devices fitted with PES membranes as the resistance to mass transfer is lower in absence of the membrane. The disadvantage of such device is a fast equilibration of the sampling device with the water phase, which restricts to a few days the time over which the sampler operates in time-integrative mode. Moreover, because the main barrier to the mass transfer is the aqueous boundary layer, the sampling kinetics of such devices are sensitive to changing hydrodynamic conditions [18,19]. Potentially, problems may arise with sample clean-up due to fouling of the receiving phase during a direct contact with sampled water in the field. More work is required to minimise the uncertainty caused by sampling rate fluctuations with degree of water turbulence. Nevertheless, this approach is very useful for detecting and semi-quantitative evaluation of short pollution events. 9.7 SAMPLING OF METALS A Chemcatcher variant based on diffusion through a porous CA membrane to a receiving phase, where the analyte is removed by chelation in a chelating Empore disk has been developed for monitoring metals [19]. Uptake rates to the receiving phase were determined in both batch and flow-through laboratory exposures for different metal ions. Sampling rates were found to be diffusion controlled and inversely related to pH. The uptake rate can be used for calculating the diffusion coefficients for specific compounds under defined laboratory conditions [19]. In situ deployment of the passive sampler was demonstrated to provide metal concentrations, corresponding to the electrochemically available fraction of total metal [20]. Laboratory handling procedures were developed that enabled a direct analysis of the accumulated metals on the receiving membrane by laser ablation inductively coupled plasma mass spectrometry [20]. In a later study, a calibration database of Rs values for five metals for independently varied temperature and turbulence conditions was established in an experimental setup similar to that described in Section 9.5.1 [6]. Rs for cadmium, copper, nickel and zinc were within 216 Monitoring of priority pollutants in water the same order of magnitude (50-150 mL day-1) and showed similar variations with temperature and turbulence. Somewhat lower sampling rates (12-17 mL day-1) were measured for lead. Both changes in temperature and turbulence were shown to have a significant effect on sampling rates of metal ions [6]. 9.8 SAMPLING OF ORGANOMETALLIC COMPOUNDS Another version of Chemcatcher has been developed for the measurement of the TWA concentrations of organotin compounds (monobutyl-tin, dibutyltin, tributyltin and triphenyltin) in water. The receiving phase is a Ci8 Empore disk and the diffusion membrane is CA. The effects of environmental variables (pH, salinity and biofouling) that could influence accumulation in receiving phase have been evaluated in the laboratory. Linear uptake was observed for at least for 14 days of exposure at constant aqueous concentration of analytes. Compound-specific sampling rates varied between 0.063 and 0.038 L day-1 [7]. 9.9 FIELD APPLICATIONS 9.9.1 Pan-European field trials to compare the performances of the Chemcatcher and spot sampling in monitoring the quality of river water In 2004, field performance of the non-polar Chemcatcher was tested in a field trial in rivers in four European countries (the Czech Republic, Finland, The Netherlands and Norway). The sampler exposure was repeated twice at each of the four sampling sites, once in spring and once in autumn. The uptake of selected organic priority pollutants (PAHs and OCPs) in the Chemcatchers during deployment periods up to 28 days were compared with the contaminant levels found in extracts from filtered spot samples of water collected regularly over the exposure period. The resulting dataset provides a solid basis for the evaluation of the passive sampling method for hydrophobic chemicals with logi^ow from 3 to 7. The main objective of the study was to evaluate the ability of non-polar Chemcatcher samplers to estimate TWA concentrations of selected PAHs and OCPs under various exposure conditions (contaminant spectrum, temperature, water turbulence and fouling). For practical estimation of the chemical exchange kinetics between Chemcatcher and water, the PRC approach was successfully applied and validated. The coefficients of variation of the two-point estimate of 217 R. Greenwood et al. the PRC overall exchange rate constants ke ranged from 1% to 34% and the precision was sufficient to allow significant ke estimates for a number of PRCs in each of the individual field studies. The PRC offload data confirmed that the chemical exchange kinetics are site specific and depend significantly on exposure conditions, including temperature, turbulence and biofouling. The knowledge of PRC offload kinetics in combination with laboratory-derived Chemcatcher calibration data enabled estimation of in situ sampling rates for the whole range of target analytes that were expected to be found in the monitored rivers. The compound-specific sampling rates ranged from 0.003 to 0.424 L day-1. Maximum in situ sampling rates were measured for compounds with moderate hydrophobicity (log Kow 4-6). The method sensitivity decreased for very hydrophobic (log Kow>6) compounds. The examination of the site-specific exchange kinetics of PRCs indicated in eight field exposures for European rivers that the uptake remained linear for up to 28 days for compounds with log Kqw > 4.3 at all sampling sites. Heavy biofouling of the samplers was observed at all four sampling sites. This may be the reason for the deterioration of the exchange kinetics of the samplers with increasing time. Confocal laser microscopy was used to obtain semi-quantitative measure (film thickness and density) of the biofilm layer. Method detection limits of target analytes in sampler extracts ranged from 0.2 to 10 ng per sampler. Instrumental method detection limits can be translated into site-specific minimum detectable water concentrations of 0.1-138 ngL-1 on the basis of compound-specific in situ sampling rates over a 14- or 28-day exposure period. The lowest detection limits were achieved for compounds with a favourable combination of a low instrument detection limit and high sampling rate. This was the case for the OCPs including dieldrin, a-endosulfan, hexa-chlorobenzene, lindane and pentachlorobenzene, as well as for PAHs with less than five aromatic rings. Mean masses of PAHs found in Chemcatchers exposed in the field ranged between one and tens of ng per sampler. Compounds with two, three and four aromatic rings per molecule dominated the PAH spectrum. These are more water soluble than the heavier PAHs, and are thus likely to be present in water at higher concentrations. Moreover, the sampling performance characteristics of the Chemcatcher favour the uptake of compounds with moderate hydrophobicity. The concentrations of analytes found in Chemcatcher extracts were converted into the corresponding TWA aqueous concentrations, using the calculated in situ sampling rates. The estimated TWA concentrations of individual 218 Monitoring of priority pollutants in water truly dissolved PAHs at the sampling sites ranged between the detection limit and 60.3 ngL-1. The estimated TWA concentrations of individual truly dissolved OCPs ranged between the detection limit and 3.4 ngL-1. The TWA concentrations estimated from the passive sampler data were compared with concentrations of analytes determined from filtered water samples to assess the performance of Chemcatcher. When comparing the TWA concentrations calculated from spot samples and passive samplers, it is important to consider the differences in contaminant fractions in water that are measured using the two methods. TWA concentrations estimated using passive samplers reflect the truly dissolved concentrations and do not account for the pollutants bound to particles and colloids in water. Water samples filtered through 0.45-|im pore size filters still contain a contaminant fraction that is bound to dissolved organic material (DOM) present in water. The truly dissolved fraction of hydrophobic analytes in water will depend on the level and quality of DOM, which may fluctuate during the sampling period. Unfortunately, there is a lack of equipment that is suitable for routine measurements of dissolved concentrations at a reasonable cost. The comparison was limited to cases where a particular analyte was detected in both the spot samples and the passive samplers. With a few exceptions (namely hexachlorobenzene and lindane) a comparison with spot samples was possible for the pesticides and for the PAHs with a maximum of four aromatic rings per molecule. The difference in water concentrations calculated using both methods never exceeded one order of magnitude. 9.9.2 Monitoring pesticide runoff in Brittany, France In 2005, Schäfer et al. [21] used Chemcatcher fitted with naked SDB-XC Empore disks to investigate whether they can be applied to monitor field runoff of ecotoxicologically relevant pesticides in current use. The field study was performed in Brittany, in the North-western France, a region with intensive agriculture and pesticide usage. Between 1 and 3 samplers were deployed for 10-13 days at each of the 16 small streams. The target analytes were mainly polar or moderately polar pesticides with logifow values between 1.4 and 4.13. These belonged to multiple classes of pesticides: chloracetanilide herbicides (alachlor, acetochlor), the phenylurea herbicide linuron, the oxadiazolone herbicide ox-adiazon, carbamate insecticides (pirimicarb, carbofuran), the organo-phosphate insecticide chlorfenvinphos, the organochlorine insecticide 219 R. Greenwood et al. endosulfan, the pyperidine fungicide fenpropidin and the conazole fungicide tebuconazole. A significant accumulation of all compounds except fenpropidin, chlorfenvinphos and a-endosulfan was observed in the devices. These results indicate the potential utility of these samplers in providing semi-quantitative or qualitative data on compounds present in episodic events, and the utility of the SDB-XC Empore disks for sequestering polar compounds. This phase may be more useful than the Ci8 disks described for the polar variant of the Chemcatcher, and further work in this area is ongoing. 9.9.3 Field trial in the River Meuse in The Netherlands A field test of the wide range of passive sampling devices presently available was conducted at RIZA's monitoring station at Eijsden (NL) in April 2005 as part of the Screening method for Water data InFor-mation in support of the implementation of the Water Framework Directive (SWIFT-WFD) project. The aim of this trial was to evaluate the suitability of passive samplers for monitoring water quality to meet the requirements of the European Union's Water Framework Directive (WFD) legislation. The trial was designed to provide data on the robustness and utility of this technology in order to strengthen the case for its introduction into monitoring programmes. Passive samplers for metals, polar and non-polar organic pollutants were deployed for overlapping periods of 7, 14 and 21 or 28 days in the River Meuse. Chemcatchers with different configurations were tested alongside SPMD, membrane-enclosed sorptive coating (MESCO), POCIS and DGT. TWA concentrations obtained were compared with those obtained from conventional spot sampling and analysis by an accredited laboratory. In addition, since the field deployment was undertaken at RIZA's monitoring station, concentrations from continuous monitoring for organic contaminants and composite sampling for metals were available for further comparisons. It was therefore possible to evaluate information provided by the passive samplers alongside that from in situ, spot and composite sampling for the monitoring of metals in water. TWA zinc concentrations measured with Chemcatcher were calculated from the masses of zinc accumulated over exposures of 7, 14 and 21 days and available calibration data. These were compared with spot sampling and weekly composite sampling conducted to determine total and filtered (0.45 |im) fractions of zinc (Fig. 9.8). TWA concentrations measured with the Chemcatcher for 7-, 14- and 21-day exposures are generally in good 220 Monitoring of priority pollutants in water 10 15 20 Time (Days) 10 15 20 Time (Days) Fig. 9.8. Comparison of TWA zinc concentrations obtained for exposures of 7, 14 and 21 days of the Chemcatcher in the River Meuse with spot sampling (A) and composite sampling (B). Both sets of water samples were analysed without (•) and with nitration (o) to 0.45 |im. TABLE 9.2 Comparison of mean zinc concentrations measured with the Chemcatcher and spot sampling (with and without filtration) for exposure times of 7, 14 and 21 days Exposure TWA concentration ((ig L x) Spot sampling dissolved (days) concentration (\xg L"1) Mean Std dev. Mean Std dev. 7 8.0 1.8 10.5 1.3 14 6.9 1.7 10.9 1.4 21 7.9 2.4 13.9 6.2 agreement with those determined by spot and composite sampling. While Chemcatcher-measured zinc concentrations are similar to mean dissolved concentrations from spot sampling for 7- and 14-day exposures, the precision of the measurement appears lower (Table 9.2). Higher fluctuations in concentrations observed during the 21-day exposure resulted in a significant loss of precision for spot sampling, while lower precision for Chemcatcher may have resulted from environmental impacts such as biofouling. However, it remains difficult to judge the accuracy of each of these methods in determining the TWA labile fraction of zinc. Slight underestimation of time-integrated filtered concentration 221 R. Greenwood et al. of zinc by the Chemcatcher may be the result of the uncertainty or bias from the calibration data used or due to a fraction of filtered zinc not available for uptake by the Chemcatcher. The time-integrated nature of in situ sampling is likely to offer more representative information than that provided by infrequent spot samples, and should be useful in assessing long-term trends in contaminant levels. 9.9.4 Field trial in the estuary of the River Ribble in the United Kingdom A field trial was conducted as part of the SWIFT-WFD project in the United Kingdom Pilot River Basin, the Ribble catchment. Pressure points along the Ribble estuary were identified, and a risk assessment was then effected. A trial was then designed to be carried out in October 2005, and passive samplers were selected to monitor some of the contaminants that might be present as a result of past and present industrial activity, including boat building, shipping and oil drilling. These pollutants potentially included metals (e.g. cadmium and mercury), or-ganotin compounds (MBT, DBT and TBT) and PAHs. Chemcatchers for polar, non-polar organic pollutants, metals and organotins were deployed along with other sampling devices over a 5-week period. A number of sampling sites was selected along the estuary including Preston docks and a control site upstream of the tidal area. One aim of this trial was to demonstrate the value of these tools in comparison with standard monitoring methods used in the estuary. The estuary was an aggressive environment with high tidal flows, and episodic storm events carrying debris down the river. Some of the sampling devices were lost because of physical damage in which the moorings were dislodged and swept away. However, sufficient deployment rigs survived to allow the measurement of pollutants at four sites over the deployment period. An example that illustrates the utility of the samplers is provided by the measurements of TWA concentrations of cadmium along the estuary (Fig. 9.9). Masses of cadmium accumulated in the Chemcatchers were generally low. Concentrations upstream of the tidal area, in Preston docks and downstream of the dock appeared similar while the concentration in one sampler from the site in mid-estuary was significantly higher. Despite possible error in the estimation of uptake rates, R$ due to the uncertainty in the levels of turbulence at the different sites, the Chemcatcher samplers yielded more useful information than that provided by the routine spot sampling carried out over the period of the trial. 222 Monitoring of priority pollutants in water 0.016 0.014- Cadmium o 0.010- S 0.008- o g 0.006- ■I 0.004- I X 0.000 Mid-estuary DS docks Preston docks US Preston Site Fig. 9.9. TWA cadmium concentrations measured using the Chemcatcher passive sampler at various sites along the Ribble estuary. Data for the mid-estuary and DS docks sites are from a single sampler and data shown for Preston docks and upstream of Preston are the mean of two measurements (DS: downstream; US: upstream). The standard monitoring by the Environment Agency for England and Wales was conducted on two occasions during the trial. Cadmium concentrations were found below limits of detection (LOD: 0.04 LigL-1) for all sites monitored. This is in agreement with concentrations measured with Chemcatcher and emphasises the advantage of in situ time-integrative sampling over spot sampling in term of detection limits, since useful data that could be used in determining trends were obtained. This contrasts with the spot sampling where only categorical information (not detected) was obtained. 9.10 COMPARISON OF THE PERFORMANCE OF THE CHEMCATCHER WITH THAT OF OTHER SAMPLING DEVICES The performances of passive samplers can be compared for a range of classes of pollutants, and evaluated alongside other methodologies. For example, calibration data for hydrophobic organic pollutants are available in the literature for SPMDs [22] and the MESCO sampling devices [23,24]. These devices differ in their design geometry and the materials 223 R. Greenwood et al. used in their construction. However, the sampling rate is directly proportional to the sampler functional surface area. Consequently, the highest sampling rates will be achieved with passive samplers having the largest surface area, such as the standard size SPMDs (450 cm2 in comparison to 17.5 cm2 for the Chemcatcher). It is therefore necessary to compare the performances on a surface area specific basis, i.e. with sampling rates expressed as volume of water cleared for a chemical, per unit time and unit surface area (L day-1 cm-2). In making this comparison it is necessary to take into account reported variations in sampling rates with exposure conditions. Although the most calibration studies reported in the literature were performed in flow-through systems, they were not all conducted under identical conditions (temperature and turbulence). However, if these limitations are taken into account an approximate comparison of sampling rates can be made. The surface-specific sampling rates of three passive sampling devices (MESCO, SPMD and non-polar Chemcatcher) are similar for PAHs compounds with three and four aromatic rings, and range from 5 to 13 mL day-1 cm-2. This indicates that the uptake of these compounds by the three different samplers is governed overall by a similar mass transfer process; this is most likely to be diffusion across the aqueous boundary layer. A similar comparison can be made for the polar variant of Chemcatcher and the POCIS. The surface area of the standard configuration of POCIS is 41 cm2 (see Chapter 8), in comparison with 17.5 cm2 for the Chemcatcher. The two samplers are fitted with similar diffusion membrane materials, both are made of PES. The surface-specific sampling rates at room temperature for atrazine and simazine were approximately a factor 2 higher for the Chemcatcher than those reported by Alvarez (Table 8.4 in Chapter 8). This is a reasonable agreement, and the observed difference may be caused by differences in the calibration conditions for the two sets of samplers. While for the metal version of Chemcatcher, uptake is limited by diffusion in water across the boundary layer and the CA membrane, for the DGT it is restricted by metal diffusion across the hydrogel and only minor effects of the boundary layer are reported [25]. For both samplers, free ions and organic/inorganic metal complexes are able to dissociate within the time required to cross the diffusion layers will accumulate and therefore the TWA concentration will be representative of these fractions. A major difference between these devices is the procedure for the calculation of TWA concentrations. While laboratory-based calibration data are used to calculate TWA concentrations with 224 Monitoring of priority pollutants in water the Chemcatcher, concentrations for DGT are obtained using known metal diffusivities for the hydrogel layer measured in the laboratory. In order to evaluate the performance of the Chemcatcher and the DGT when responding to simulated peaks of metal concentrations, a 5-day tank experiment was conducted using Meuse river water. TWA concentrations were measured and compared with the equivalent concentrations from unfiltered, filtered (0.45 um) and ultra-filtered (5 kDa) spot samples. Figure 9.10 shows a comparison of TWA concentrations measured by the Chemcatcher and the DGT, relative to spot sampling concentrations. While for Cd and Ni, the Chemcatcher slightly underestimates TWA concentrations, the DGT is in better agreement with filtered fractions of these metals. Similar results are obtained for both samplers for Zn and closest agreement is with the filtered fraction. For Cu, both samplers underestimate the filtered concentration while clearly overestimating the ultra-filtered fraction. Generally, results appear in agreement with the speciation of these metals under those conditions. Overall, TWA concentrations obtained using the Chemcatcher appear to have a slight bias as most data points are below the 1:1 relationship. This may be related to the selection of laboratory c o s c o u < 1.4 - 1.2 - 1.0 D 0.8 0.6 ■ / / / □ / A / 4A relationship o Unfiltered 0 Filtered • Ultrafiltered 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Chemcatcher / TWA Concentration Fig. 9.10. Comparison of TWA Cd (O), Cu (A), Ni (V) and Zn (□) concentrations measured by Chemcatcher and DGT relative to TWA concentrations (unfiltered, filtered (0.45 urn) and ultra-filtered (5 kDa)) measured by spot sampling during a 5-day long tank experiment with spiked metals simulating fluctuating concentrations in natural Meuse river water. 225 R. Greenwood et al. calibration data for set levels of temperature and turbulence that differ slightly from conditions observed during the experiment. 9.11 FUTURE TRENDS The advantage of passive sampling over classical spot sampling is that it provides a measure of average conditions in a body of water over extended periods of time. This gives a more representative picture of water quality than a few instantaneous measurements of pollutant levels taken at intervals over a year. Monitoring programmes based on passive sampling will therefore provide better information on which to assess long-term trends in pollutant concentrations. For metal samplers, it is possible to obtain extra information on speciation that is pertinent to their bioavailability and potential toxicity [26,27] and hence underpin robust risk analysis. In order to facilitate recognition of the value of passive sampling, and its potential for underpinning legislation it is essential to demonstrate the validity of the method, and to develop standards for use in this field. One national standard (BSI PAS 61) [28] is available, and this covers the preparation, field deployment in surface waters and preparation for analysis of passive samplers. It is also important, however, to recognise the limitations of passive samplers, and to address some of the challenges laid down by these. One important challenge is the assessment of the impact of biofouling of the diffusion membrane on uptake rates. A further challenge is to develop sampler designs that can be used to detect and quantify peaks of concentrations during short but significant pollution events. This may be especially important for the measurement of, for example, intermittent industrial releases that may otherwise not be detected. Currently, it is difficult to assess whether an observed accumulation in a sampler is the result of a transient event or a lower but more constant concentration. In order to be able to interpret passive sampler data, particularly over the short-term deployments needed to detect peak episodic events, a better knowledge of observed lag phases between the appearance of a peak of contaminant concentration in water and its detection by a passive sampling device will be required to allow a clearer interpretation of passive sampling data. ACKNOWLEDGMENTS We acknowledge the financial support of the European Commission (Contracts EVK1-CT-2002-00119; http://www.port.ac.uk/stamps/ and 226 Monitoring of priority pollutants in water SSP1-CT-2003-502492; http://www.swift-wfd.com) for this work. We thank Arne Holmberg (Alcontrol, Sweden) and Miro Vrana for providing the technical drawing of the Chemcatcher prototype (Fig. 9.3). REFERENCES 1 B. Vrana, G.A. Mills, E. Dominiak and R. Greenwood, Calibration of the Chemcatcher passive sampler for the monitoring of priority organic pollutants in water, Environ. Pollut., 142 (2006) 333. 2 J.K. Kingston, R. Greenwood, G.A. Mills, G.M. Morrison and B.L. Pers-son, Development of novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments, J. Environ. Monit, 2 (2000) 487. 3 J.N. Huckins, G.K. Manuweera, J.D. Petty, D. Mackay and J.A. Lebo, Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water, Environ. Sci. Technol., 27 (1993) 2489. 4 G.D. Johnson, Hexane-filled dialysis bags for monitoring organic contaminants in water, Environ. Sci. Technol., 25 (1991) 1897. 5 B. Vrana, G.A. Mills, M. Kotterman, P. Leonards, K. Booij and R. Greenwood, Modelling and field application of the Chemcatcher passive sampler calibration data for the monitoring of hydrophobic organic pollutants in water, Environ. Pollut, 145 (2007) 895-904. 6 K. Runeberg, S. Rauch, B. Vrana, G.A. Mills, R. Greenwood, M. Kotterman, S. Herve, M. Weideborg, V. Kocourek and G.M. Morrison, Assessment of the performance of the Chemcatcher sampler for the monitoring of metals in water (2006, in preparation). 7 R. Aguilar, R. Greenwood, G.A. Mills, B. Vrana, M.A. Palacios and M. Gomez, Development of a passive sampling system for the time weighted average monitoring of mercury and organotin species in water (2006, in preparation). 8 B. Vrana, G. Mills, R. Greenwood, J. Knutsson, K. Svensson and G. Morrison, Performance optimisation of a passive sampler for the monitoring of hydrophobic organic pollutants in water, J. Environ. Monit., 7 (2005) 612. 9 J.N. Huckins, J.D. Petty and K. Booij, Monitors of Organic Contaminants in the Environment: Semipermeable Membrane Devices, Springer Verlag, New York, 2006. 10 R.J. Scheuplein, On the application of rate theory to complex multibarrier flow co-ordinates: membrane permeability, J. Theor. Biol., 18 (1968) 72. 11 G.L. Flynn and S.H. Yalkowsky, Correlation and prediction of mass transport across membranes. I. Influence of alkyl chain length on flux-determining properties of barrier and diffusant, J. Pharm. Sci., 61 (1972) 838. 227 R. Greenwood et al. 12 B. Vrana and G. Schüürmann, Calibrating the uptake kinetics of semipermeable membrane devices in water: impact of hydrodynamics, Environ. Sei. Technol., 36 (2002) 290. 13 D.W. Connell, Bioaccumulation of Xenobiotic Compounds, CRC Press, Boca Raton, FL, 1990. 14 J.N. Huckins, J.D. Petty, H.F. Prest, R.C. Clark, D.A. Alvarez, C.E. Orazio, JA. Lebo, W.L. Cranor and B.T. Johnson, A guide for the use of semipermeable membrane devices (SPMDs) as samplers of waterborne hydrophobic organic contaminants. Report for the American Petroleum Institute (API), API publication number 4690, API, Washington, DC, 2002. 15 D.A. Alvarez, J.D. Petty, J.N. Huckins, T.L. Jones-Lepp, D.T. Getting, J.P. Goddard and S.E. Manahan, Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments, Environ. Toxicol. Chem., 23 (2004) 1640. 16 M. Shaw and J.F. Müller, Preliminary evaluation of the occurrence of herbicides and PAHs in the Wet Tropics region of the Great Barrier Reef, Australia, using passive samplers, Mar. Pollut. Bull., 51 (2005) 876. 17 B.S. Stephens, A. Kapernick, G. Eaglesham and J. Müller, Aquatic passive sampling of herbicides on naked particle loaded membranes: accelerated measurement and empirical estimation of kinetic parameters, Environ. Sei. Technol., 39 (2005) 8891. 18 C.E. Green and M.H. Abraham, Investigation into the effects of temperature and stirring rate on the solid-phase extraction of diuron from water using a Cis extraction disk, J. Chromatogr. A, 885 (2000) 41. 19 L.B. Persson, G.M. Morrison, J.U. Friemann, J. Kingston, G. Mills and R. Greenwood, Diffusional behaviour of metals in a passive sampling system for monitoring aquatic pollution, J. Environ. Monit., 3 (2001) 639. 20 L.B. Blom, G.M. Morrison, J. Kingston, GA. Mills, R. Greenwood, T.J.R. Pettersson and S. Rauch, Performance of an in situ passive sampling system for metals in storm water, J. Environ. Monit., 4 (2002) 258. 21 R. Schäfer, R. Mueller, A. Paschke, B. Vrana, L. Lagadic and M. Liess, Comparison of three methods for determination of moderately polar pesticides in small streams in agricultural areas (2006, in preparation). 22 J.N. Huckins, J.D. Petty, C.E. Orazio, JA. Lebo, R.C. Clark, V.L. Gibson, W.R. Gala and K.R. Echols, Determination of uptake kinetics (sampling rates) by lipid-containing semipermeable membrane devices (SPMDs) for polycyclic aromatic hydrocarbons (PAHs) in water, Environ. Sei. Technol, 33 (1999) 3918. 23 B. Vrana, P. Popp, A. Paschke and G. Schüürmann, Membrane-enclosed sorptive coating. An integrative passive sampler for an integrative passive sampler for monitoring organic contaminants in water, Anal. Chem., 73 (2001) 5191. 228 Monitoring of priority pollutants in water 24 B. Vrana, A. Paschke and P. Popp, Calibration and field performance of membrane-enclosed sorptive coating for integrative passive sampling of persistent organic pollutants in water, Environ. Pollut., 25 (2006) 296. 25 K.W. Warnken, H. Zhang and W. Davison, Accuracy of the diffusive gradients in thin-films technique: diffusive boundary layer and effective sampling area considerations, Anal. Chem., 78 (2006) 3780. 26 I.J. Allan, B. Vrana, R. Greenwood, G.A. Mills, B. Roig and C. Gonzalez, A "toolbox" for biological and chemical monitoring requirements for the European Union's Water Framework Directive, Talanta, 69 (2006) 302. 27 S.C. Apte, G.E. Batley, K.C. Bowles, P.L. Brown, N. Creighton, L.T. Hales, RV. Hyne, M. Julli, S.I. Markich, F. Pablo, N.J. Rogers, J.L. Stauber and K. Wilde, A comparison of copper speciation measurements with the toxic responses of three sensitive freshwater organisms, Environ. Chem., 2 (2005) 320. 28 British Standards Institute (BSI), Publicly Available Specification: Determination of priority pollutants in surface water using passive sampling (PAS-61), May 2006. 229 Príloha 31 Paschke A., Vrana B., Popp P., Wennrich L. PaschkeH., Schuurmann G.: Chapter 10 Membrane-enclosed sorptive coating for the monitoring of organic compounds in water. In: Comprehensive Analytical Chemistry, R. Greenwood, G. Mills, B. Vrana (eds.). Elsevier, Amsterdam, Volume 48, 2007, Pages 231-249. Chapter 10 Membrane-enclosed sorptive coating for the monitoring of organic compounds in water Albrecht Paschke, Branislav Vrana, Peter Popp, Luise Wennrich, Heidrun Paschke and Gerrit Schúúrmann 10.1 INTRODUCTION Membrane-enclosed sorptive coating (MESCO) denotes the recently developed miniaturised passive sampling devices consisting of a membrane which encloses polydimethylsiloxane (PDMS) coatings or coarse silicone material (embedded in a fluid) as the collecting phase for organic compounds.1 The general advantages of the MESCO samplers are (i) the simple and loss-free separation of the collector phase; (ii) its processing without further clean-up steps by direct thermal desorption or solvent microextraction; (iii) the possibility to spike the collecting phase before deployment with so-called performance reference compounds (PRCs) and (iv) that, in addition to chemical target or non-target analysis, the collecting phase can also be subject to biological effect screening (after digestion using an appropriate solvent). In our work we took advantage of commercially available PDMS coatings or silicone materials as the collecting phase. PDMS is recommended as a receiving phase in extraction and thermodesorption as it has a number of benefits compared with other sorbents [1]. The predominant mechanism of analyte extraction into PDMS/silicone phase is absorptive partitioning which has the advantage that displacement effects of the analytes (competitive enrichment), characteristic for adsorbents, play no role. When neat silicone material is used as collecting phase instead of a sorptive coating, one can take the abbreviation MESCO also for membrane-enclosed silicone collector. Comprehensive Analytical Chemistry 48 R. Greenwood, G. Mills and B. Vrana (Editors) Volume 48 ISSN: 0166-526X DOI: 10.1016/S0166-526X(06)48010-7 © 2007 Elsevier B.V. All rights reserved. 231 A. Paschke et al. The chapter gives an overview of theoretical aspects and design of the different MESCO sampler formats for water monitoring2 and summarises our efforts to calibrate the samplers for several priority pollutants in laboratory studies and to test them under field conditions. 10.2 PASSIVE UPTAKE MODEL FOR MESCO SAMPLER It has been shown that the amount of the chemical accumulated in the MESCO sampler from water with constant chemical composition can be described by the following equation [2]: where nis is the mass of analyte in the receiving phase (PDMS), m0 is the amount of analyte in the sampler at the start of the exposure, Cw represents the water concentration during the deployment period, Ksw is the receiving phase/water distribution coefficient, V$ is the volume of the receiving phase, kOY is the overall mass transfer coefficient, A is the membrane surface area, a is the pore area of the membrane as fraction of total membrane area (membrane porosity) and t equals time, a will be set to 1 for non-porous membranes. The coefficient in the exponential function is referred to as the overall exchange rate constant ke: k0YAa Rs ke K V K V (10'2) where Rs is the sampling rate, expressing equivalent volume of water cleared of chemical per unit of time in the linear (integrative) uptake phase. Adding standards (i.e. PRCs) to the receiving phase prior to exposure of the passive sampler has been suggested as a means to calibrate the exchange rates in situ [3,4]. When PRCs are used that are not present in water (Cw = 0), Eq. (10.1) reduces to mg(ř) = moexp(—ket) (10.3) which is a one-parameter equation, because the amount of PRC added to the MESCO sampler (m0) is known. 2Some other MESCO variants designed for monitoring semi-volatile organic compounds in air are described in Chapter 5. ms(t) = mQ + (CWKSWVS - m0) 1 - exp t ) (10.1) 232 MESCO for monitoring in water 10.3 DESIGN OF THE DIFFERENT MESCO FORMATS 10.3.1 PDMS-coated fibre enclosed in an LDPE membrane As a precursor of the MESCO [5] we tested membrane bags (13 x 2.5 cm) of 100 Lim thick low-density polyethylene (LDPE) tubing (Polymer -Synthesewerk Rheinberg, Germany), heat-sealed at both ends, in combination with a 100 Lim PDMS-coated SPME fibre (Supelco, Deisenhofen, Germany) as collector phase (Vq = 0.68 liL) and 25 mL of a 40/60 iso-propanol/water mixture (v/v) as inner fluid. LDPE is the membrane material also used in construction of SPMDs [6] and the PDMS-coated fibre is a rational tool for solid-phase microextraction of analytes from aqueous samples, and provides high enrichment factors for more hydrophobic substances [7]. Figure 10.1 shows the design of this permeation sampler. The coil spring (of stainless steel) prevents the fibre coating from a direct contact with the membrane. A serious shortcoming of this sampler is that the polymer-coated quartz glass fibre tip is fragile and difficult to handle during removal from and re-inserting into the steel needle of the commercial SPME syringe device. 10.3.2 PDMS-coated stir bar enclosed in a dialysis membrane bag (MESCO I) This type, first described by Vrana et al. in 2001 [2,8], uses the PDMS-coated stir bar as collector phase. The stir bar is known under the rFA/T trademark Twister (Gerstel, Mulheim/Ruhr, Germany) and is commonly used for solvent-free microextraction using the same principle as ■ PDMS-coated SPME fibre LDPE membrane ■ Coil spring 1 Fluid filling Fig. 10.1. Construction of MESCO precursor [5]. 233 A. Paschke et al. Fig. 10.2. Diagram of MESCO I [2]. an SPME fibre, but with a larger extraction capacity. Figure 10.2 shows a diagram of the sampler. Specifically, we tested dialysis membrane bags made of regenerated cellulose (Spectra/Por 6) with molecular weight cut-off of 1 kDa (3 x 1.8 cm), sealed at each end with a 35 mm Spectra/Por closure, in combination with Twister bars of 15 mm length coated with a 500-um-thick layer of PDMS (Vs = 24uL) and 3mL bi-distilled water as membrane bag filling. Regenerated cellulose is a porous hydrophilic membrane material that enables widening the applicability to a broader polarity range of pollutants, including low-hydrophobic substances (\og Kqw<4). Unfortunately, this material has relatively low chemical and thermal stability and is subject to microbial degradation, which potentially leads to the damage of the sampler in natural surface waters during prolonged exposure of several weeks. 10.3.3 Silicone material enclosed in an LDPE membrane (MESCO II) This sampler type combines [8,9] the advantages of a high-capacity collector phase with that of a more stable membrane material, LDPE. These membranes are hydrophobic, resistant to solvents and biodegradation and they can be heat-sealed. Furthermore, the relatively expensive and fragile Twister bar is substituted by a cheap silicone material (pieces of a tube or rod) as collector phase. Figure 10.3 shows the schematic design of the sampler. Additional investigations have shown the usefulness of these materials for an effective pre-concentration of several classes of persistent organic pollutants from water samples and the applicability of thermodesorption-GC-MS analogously to the processing of Twister bars [9,10]. The significantly enhanced volume of the collector phase 234 MESCO for monitoring in water LDPE membrane Fluid filling Silicone rod/tube Fig. 10.3. Schematic design of MESCO II [8]. (> 100 liL) increases the maximum exposure time of the passive sampler in the field. A practical drawback of silicone tubes, when used as collecting phase in combination with water as fluid filling, is that remaining water droplets (inside the tube) can disrupt the GC-MS analysis. Since 2004 we have focused our work on improvement of the promising MESCO II format with silicone rods enclosed. Several thicknesses of LDPE membrane were tested as well as other membrane materials, such as the dense polypropylene bag, usually used for membrane-assisted solvent extraction of water samples in the laboratory [11]. Interestingly, it turns out in a preliminary laboratory study that this latter material is not useful for MESCO devices because it obviously prevents the transfer of substances to the inner receiving phase (silicone rod). 10.4 LABORATORY-DERIVED SAMPLING RATES OF THE VARIOUS MESCO FORMATS The performance of the PDMS-coated fibre in LDPE membrane bag (MESCO precursor) was tested by time-dependent exposure in a flow-through system [2] at 19°C (upstream flow: 36 Lh-1; nominal water concentration for each test substance: 50ngL_1; exposure times: up to 360 h). The sampling rates obtained are summarised in the second column of Table 10.1 (for details of SPME fibre desorption and gas chromatographic analysis see Ref. [7]). MESCO I samplers were also tested in this flow-through apparatus (under the same conditions as above; see Ref. [2] for experimental 235 A. Paschke et al. TABLE 10.1 Sampling rates CRS) of the two early MESCO formats in comparison with that of a standard SPMD for selected priority pollutants Substance Rs of SPMD Rs of MESCO Rs of MESCO I (mLh-1)a precursor (mLh x)b (mLh-1)b a-HCHc 108 0.0005 0.40 Hexachlorobenzene 58 0.0022 0.25 Anthracene 150 0.0014 0.22 Fluoranthene 188 0.0015 0.25 Pyrene 217 0.0012 0.27 Benz o [a] anthracene 133 0.0009 0.37 PCB 28c 350 0.0070 0.15 PCB 52 258 0.0088 0.15 PCB 101 258 0.0063 0.13 PCB 138 200 0.0046 0.09 PCB 153 133 0.0031 0.10 aAt 18°C for a-HCH, hexachlorobenzene and the polyaromatic hydrocarbons, at 12°C for PCBs; taken from Ref. [12]. bAt 19°C. "Substance abbreviations: HCH—hexachlorocyclohexane; PCB—polychlorinated biphenyl. details). The determined sampling rates are given in the last column of Table 10.1. Due to its much larger sampling capacity, the standard SPMD (of 450 cm2 surface area) has up to five orders of magnitude higher sampling rates than the MESCO formats tested. But one should bear in mind that the substances trapped in the PDMS coating (of an SPME fibre or a Twister bar) are, in contrast to that sampled using an SPMD, transferred quantitatively to the injector of the analytical instrument. This prevents, at similar sampling sensitivity, possible volumetric dilution errors but has on the other hand the disadvantage of having only "one shot" per sampler specimen that can be overcome only by multiple exposure of samplers (as a MESCO string). Further flow-through calibration experiments showed that the sampling rates in MESCO I were not significantly affected by the flow velocity, within the tested range of exposure conditions [2,13]. Different configurations of the MESCO II sampler were exposed to spiked water in a similar flow-through system at 14°C (upstream flow: 60 Lh-1; nominal concentration: 50ngL_1 for each test chemical; exposure times: up to 176/236 h). The membrane bags (5 cm x 3 cm) consisted of 100 |im thick LDPE tubing (Polymer-Synthesewerk Rheinberg, 236 MESCO for monitoring in water Germany). Four cm long pieces of silicone tube (with 3.6mm O.D., 3.0 mm I.D.; Reichelt, Heidelberg, Germany) or 4-cm-long pieces of silicone rod (2.0 mm O.D., Goodfellow, Bad Nauheim, Germany) were used as collector phase. The silicone material was embedded in 8mL water in one series of experiments or in air for another series (see Ref. [9] for further experimental details). The sampling rates calculated from the accumulated analyte mass are given in Table 10.2. Remarkably higher Rs values (in the same order of magnitude as those obtained for MESCO I) were obtained with air as fluid filling of the membrane bags. This can be explained by a detailed consideration of the mass-transfer resistances [9]. Tube and rod material yielded similar results but the variances in R$ were lower for the tube-containing sampler. Recently, we determined preliminary sampling rates for new MESCO II sampler formats in rapid semi-continuous batch extraction tests [14]. These consisted of lay-flat membrane strips, 15 x 3 cm of the 100 |im thick membrane or 8 x 4 cm of that with 50 |im wall thickness. The strips were segmented by heat-sealing into four or two uniform parts, respectively. Each segment (2 cm long) contained a 15 mm long piece of pre-conditioned SR "embedded" in air. Such an SR piece is equivalent to 47 |iL of receiving phase. These data are given in Table 10.2. There is a reasonably good agreement with R$ values obtained in the previous study. Additional flow-through experiments are in progress to investigate the influence of temperature and water flow on the sampling rates of these inexpensive MESCO variants and to test the applicability of the PRC concept for R$ adjustment to varying sampling conditions. 10.5 FIELD APPLICATION OF MESCO SAMPLERS 10.5.1 A case study with MESCO I for monitoring of persistent organic pollutants in surface water 10.5.1.1 Sampling site To assess the performance of MESCO for monitoring persistent organic pollutants (POPs) in the field, samplers were exposed in water at a site located in the river Weisse Elster at the locality Halle-Burgholz in Saxony-Anhalt, Germany, close to the confluence of the River Weisse Elster with the River Saale (ol^lCN; ll°59/47//E, estimated using Google Earth). Three MESCOs were deployed at the sampling site for 28 days during summer 2002 (24th July-21st August). The last two weeks of sampler exposure coincided with the major flood that occurred 237 TABLE 10.2 Sampling rates (Rs) of different MESCO II configurations (SR—silicone rod; ST—silicone tube) for selected priority pollutants determined in various laboratory experiments Substance Rs of Rs of Rs of Rs of Rs of SR+water in ST+water in ST+air in SR+air in SR+air in 100 urn LDPEa 100 urn LDPEa 100 urn LDPEa 100 urn LDPEb 50 urn LDPEb (mLr1) (mLr1) (mLh-1) (mLh-1) (mLh-1) a-HCHc 0.28 0.18 0.14 0.031d 0.039d 1,2,3,4-TCBC not det.e not det.e not det.e 1.47 0.61 Pentachlorobenzene 0.21 0.19 4.30 1.30 2.24 Hexachlorobenzene 0.09 0.06 0.90 0.65 0.87 Naphthalene not det.e not det.e not det.e 0.13d not det.e Acenaphthylene 0.51 0.73 1.40 0.07d not det.e Acenaphthene 0.48 0.67 2.23 0.35 not det.e Fluorene 0.75 1.34 1.88 0.49 not det.e Phenanthrene 0.26 0.27 0.93 0.63 0.72 Anthracene 0.13 0.26 0.99 0.40 0.83 Fluoranthene 0.04 0.06 0.12 0.33 0.26 Pyrene 0.03 0.03 0.10 0.26 0.23 PCB 28c 0.06 0.06 0.92 0.74 0.63 PCB 52 0.03 0.04 0.62 0.66 4.12f PCB 101 0.004 not det.e not det.e 0.39 not det.e PCB 138 not det.e not det.e not det.e 0.14 0.05 PCB 153 not det.e not det.e not det.e 0.15 0.05 aDetermined in a flow-through apparatus with a nominal analyte concentration of 50ngL_1 at 14°C [9]. bDetermined in serial batch extraction tests with a nominal analyte concentration of 25ngLT1 at room temperature [14]. cSubstance abbreviations: HCH—hexachlorocyclohexane; TCB—tetrachlorobenzene; PCB—polychlorinated biphenyl. distribution constant CKsw — CMEsco(eq.)/Cw(eq.)) calculated by assuming that CW(eq.) — 25ngL_1. eNot determined. fPotential outlier. MESCO for monitoring in water in the river basins of Elbe and Danube in Central Europe in August 2002. A local flood event was observed also at the Weisse Elster, accompanied with the rise in water level up to 2 m against the typical summer average. The samplers were retrieved after the flood wave retreated. During the exposure, the water temperature at the sampling site varied from 19 to 22°C. 10.5.1.2 Target pollutants The analytes included several groups of POPs: y-hexachlorocyclohexane (y-HCH), hexachlorobenzene (HCB), 2,2/-bis(4-chlorophenyl)-l,l/-dichloro-ethylene (DDE), selected polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs). The Ksw values, needed in further data evaluation, were approximated by PDMS/water distribution coefficients from the literature and were reported previously [13]. 10.5.1.3 Sampler preparation MESCO I preparation has been described in detail elsewhere [2,13]. Briefly, the cleaned and conditioned Twister stir bar was pre-loaded with six PRCs: 2H10-biphenyl (D10-BIP), 2H10-phenanthrene (D10-PHE), 2H10-anthracene (D10-ANT), 2H10-fluoranthene (D10-FLT), 2H10 -pyrene (D10-PYR) and 2H12-benzo[a]anthracene (D12-BaA). This was performed by stirring the Twister bar for 30min at 1000 min-1 at room temperature in 25 mL of solution containing l|igL_1 of each PRC. For sampler assembly, the Twister bar was placed inside a dialysis membrane bag. The bag was filled with 3 mL of bi-distilled water and sealed at each end with 35 mm Spectra Por closures. Four control samplers were prepared together with the three field-deployed samplers; these were stored in the laboratory at —20°C until analysis. Controls were processed exactly as deployed samplers and were used to define contamination during preparation and storage, and to determine nominal PRC concentrations in MESCO samplers. 10.5.1.4 Sampler deployment and retrieval On the day of deployment, MESCOs were freshly prepared in the laboratory and transported to the field in amber glass jars filled with bi-distilled water to prevent drying of the dialysis membrane during transport. At the sampling site, MESCOs were removed from the jars and placed into a protective deployment device designed for sampling using SPMDs. The deployment device was a canister made of perforated stainless steel sheet (5 mm openings), containing five racks designed for holding standard length SPMDs. One of these carriers was used to hold 239 A. Paschke et al. the MESCOs. Three SPMDs with standard configuration (2.54 x 91.4 cm, 75-90 Lim membrane thickness, total mass 4.3 g each) were exposed next to MESCOs, in the same deployment device. Before exposure, SPMDs were spiked with PRCs (10Lig/SPMD of each standard) as described earlier [15]. The deployment device protected MESCOs and SPMDs from abrasion and the sequestered pollutants from sunlight. The canister was held at depth of approximately 1 m below surface using a buoy, rope and anchor, and was secured to the shore with a rope. On day 28, samplers were removed from the deployment device and checked visually for mechanical damage. Although disintegration (mechanical or biological degradation) of the cellulose bags occurred during the exposure of MESCOs, the Twister bars were found to be intact, sticking by their magnets to the inner surface of the deployment canister. The Twister bars were dried using a soft paper tissue, transferred using clean forceps to GC vials (2mL), sealed and transported to the laboratory in a portable icebox (on ice and in darkness) and stored at —20°C till analysis. Field exposed samplers were analysed together with the control samplers. 10.5.1.5 Sampler processing and analysis The quantification of the compounds accumulated during field exposure on Twister bars of the MESCO samplers was carried out as described previously [2,13]. Briefly, analyses were performed on an Agilent Technologies (Palo Alto, CA, USA) GC 6890 with MSD 5973N system equipped with a Gerstel (Mulheim/Ruhr, Germany) thermodesorption system TDS A and a cold injection system CIS-4 from Gerstel with an empty liner that was used for cryofocusing the analytes prior to the transfer onto the analytical column. The single ion monitoring (SIM) mode of the mass selective detector applying one or two characteristic ions per compound was chosen for the detection. Quantification of target substances sorbed on Twister bars was accomplished using a five-point external standard curve. Method quantification limit for the analytes under investigation ranged from 1 to 5 pg per Twister. Details of SPMD processing were described earlier [15,16]. SPMD data evaluation was performed using the empirical uptake model derived by Huckins et al. [17]. 10.5.1.6 Accumulated amount of water pollutants Table 10.3 shows the mass of each target analyte accumulated in the MESCOs during a 28-day field deployment. Quantifiable amounts of all target analytes were found in field-exposed samplers. Control blanks 240 TABLE 10.3 Average mass of pollutants (in pg per Twister bar) determined in the control MESCOs (m0) and in the field-exposed MESCOs (ms; n = 3), and in situ aqueous concentrations of organic analytes estimated from MESCO (Cw) Compound m0 (pg) CVa (n = 4) (%) ms (pg) CV (n = 3) (%) ke (day"1) Cw (ngL^1) HCB 1 13 79 2 0.085 0.14 7-HCH 1.8 1695 27 0.130 182 p,p'-DDE <1 132 8 0.069 0.03 PCB 28 1 16 62 7 0.077 0.05 PCB 52 <1 43 6 0.072 0.02 PCB 101 <1 27 12 0.065 0.004 PCB 138 <1 33 5 0.062 0.003 PCB 153 <1 22 9 0.062 0.002 PCB 180 <1 8 10 0.064 0.001 Acenaphthylene 4 45 124 11 0.107 2.16 Acenaphthene 10 10 1172 3 0.102 12.2 Fluorene 18 9 1128 4 0.100 9.76 Anthracene 8 30 1494 9 0.094 7.03 Phenanthrene 62 30 3128 7 0.094 15.4 Fluoranthene 13 30 3135 8 0.079 2.86 Pyrene 13 15 3302 8 0.076 2.16 Benzo[a]anthracene 2 76 1185 3 0.069 0.32 Chrysene 4 42 967 2 0.063 0.10 Benzo [b] fluoranthene <5 450 1 0.071 0.15 Benzo[k]fiuoranthene <5 244 3 0.068 0.06 Benzo[a]pyrene <5 455 4 0.067 0.09 Indeno[l,2,3-cd]pyrene <5 121 7 0.087 0.29 Dibenzo [a,h] anthracene <5 46 10 0.084 0.03 Benzo[g,h,i]perylene <5 115 10 0.076 0.20 The samplers were exposed 28 days in August 2002 at a site in the river Weisse Elster in Saxony-Anhalt, Germany. aCV, coefficient of variation or relative standard deviation of multiple samples. A. Paschke et al. contained quantifiable amounts of lindane, PCB 28 and PAHs with up to four aromatic rings. Nevertheless, analyte levels found in field exposed samplers were in all cases significantly higher than those in control blanks. The variation of the masses recovered from three replicate field exposed devices ranged from 1% (benzo[b]fluoranthene) to 27% (lindane). This is an excellent precision despite the degradation of the protective cellulose membranes of the MESCOs during exposure. 10.5.1.7 In situ exchange kinetics from PRC offload Our previous investigations have shown that both uptake and elimination of a particular compound in MESCO I are characterised by the same exchange rate constant ke, according to Eq. (10.1) [13]. The use of PRCs allowed a two-point estimation of the first-order exchange rate constants ke. These were calculated from the rearranged Eq. (10.3) using mean values (from replicate samples) of the PRC amounts found in field exposed samplers (ms) and in the controls (m0) and exposure time of 28 days: ke = ln(m°/ms) (10.4) The calculated ke values ranged from 0.072 day"1 (Di0-PYR) to 0.126 day-1 (D10-BIP). Student's t-test (a = 0.05) was performed to ensure that changes in PRC residue concentrations were statistically significant, according to the law of error propagation. This was the case for all PRCs excepting Di2-BaA with no significant offload during exposure. The field-derived ke values were two to three times higher than those reported in a laboratory calibration study [13]. This indicates faster exchange kinetics at the sampling site than those observed under laboratory conditions. The temperature at the sampling site during the field study was similar to that in the calibration study. Although this investigation [13] indicated that the flow velocity had no significant effect on the exchange kinetics, this was tested only at low velocities. The flow around the cage with samplers in the field was much faster than the simulated flow in the calibration apparatus, and the increased water turbulence might have affected the analyte mass transfer between water body and samplers, despite the buffering effect of the protective cage. The elevated exchange kinetics can also be explained by degradation of cellulose membranes during the field exposure, resulting in a significant loss of resistance to analyte exchange between Twister and water. 242 MESCO for monitoring in water -0.80 -0.85 -0.90 -0.95 ^ -1.00 o -1.05 -1.10 -1.15 -1.20 2.5 3.0 3.5 4.0 4.5 5.0 log Ksw Fig. 10.4. Correlation between estimated in situ exchange rate constants ke and PDMS/water distribution constant Ksw. The sampling using MESCO I was performed in August 2002 in the river Weisse Elster near confluence with River Saale. The sub stance-specific ke values were estimated from the linear correlation between log ke and log Ksw (Fig. 10.4): log&e = -0.6187 - 0.1029 \ogKsw (n = 5,s = 0.05, r = 0.904) (10.5) The estimated in situ ke values of target analytes are shown in Table 10.3. 10.5.1.8 Sampling-mode considerations The knowledge of in situ ke values enables to estimate the percentage of sampler saturation with target analytes after 28 days of exposure, when a constant pollutant concentration in the river water is assumed. This can be calculated as (1—exp(-ke t)) x 100% and shows that the accumulated concentrations of target analytes approached 83-97% of partitioning equilibrium, determined by the magnitude of the PDMS/water distribution constant Ksw (Fig. 10.5). The sampler exposure seems to have exceeded the maximum time period allowing time-weighted average (TWA) sampling, which lasts approximately until the sampler approaches 243 A. Paschke et al. 100- cr CD 80- ■§ 60- 40- 20- Fig. 10.5. Percentage of target-analyte equilibration between passive samplers and water as dependent from their hydrophobicity (expressed as log K0w) after 28 days of field exposure in the river Weisse Elster. The dashed line indicates the maximum limit of saturation (50%) permitting a time-integrative sampling. half-saturation. The prolonged field exposure was due to the flood event that made it impossible to retrieve the samplers any earlier. A comparison of percentage of sampler saturation with target analytes in MESCO and a standard-size SPMD shows that after 28 days of exposure, partitioning equilibrium was reached in both samplers for compounds with \ogKow<4.5 (Fig. 10.5). Compounds with log-Kow>5.5 have not exceeded half-saturation in SPMDs. This indicates that SPMDs sampled those compounds in time-integrative mode. In contrast, all compounds have likely exceeded the half-saturation in MESCO samplers (Fig. 10.5). Thus, after 28 days, MESCO was in the curvilinear or equilibrium sampling phase. This is caused by the fact that MESCO has much lower absorption capacity than SPMD, due to its very small receiving-phase volume. The calculation of saturation halftime ty2 = \n2/ke shows that the MESCO remained in the linear or integrative uptake phase during the first two weeks of exposure for most of the analytes under investigation. This information is valuable for further method validation, indicating that field exposures using MESCO I in warm and turbulent water should not exceed two weeks, 244 MESCO for monitoring in water if the study is aimed the estimation of TWA concentrations. Two weeks seems to be also a compromise time period during which no degradation of the cellulose membrane is expected. 10.5.1.9 Estimate of ambient aqueous concentrations As a consequence of the different exchange kinetics between the field study and laboratory experiments, a direct application of laboratory-derived calibration data for calculation of ambient aqueous concentrations of target analytes was not appropriate in this particular case. Nevertheless, the calculation of aqueous concentrations was performed using Eq. (10.1), knowing the necessary substance-specific parameters ke and Ksw: Cw = irswys[i-exP(-M] (10'6) The estimated aqueous concentrations are shown in Table 10.3. They range from lpgL-1 (PCB 180) to more than 180 ngL-1 (lindane), demonstrating that MESCO allows for in situ measurement of very low contaminant levels. It is important to stress that the calculated aqueous concentrations are an estimate of the truly dissolved fraction present in water as shown by Garcia-Falcon et al. [18]. The sampling-mode considerations indicate that the calculated values in this particular study did not provide an accurate TWA concentration estimate, nevertheless, MESCO I has a great potential for time-integrative sampling, provided the deployment period is restricted to a shorter time. 10.5.1.10 Comparison of MESCO I with SPMD Figure 10.6 shows a comparison of aqueous concentrations of PAHs estimated from analyte amounts accumulated in MESCOs and SPMDs during a 28-day field deployment. Both methods provide information on a dissolved fraction of analytes, enabling a direct comparison of results obtained using the two approaches. Aqueous concentrations estimated using both methods showed similar patterns, with higher levels of less hydrophobic light PAHs (with four and less aromatic rings) and low concentrations of more hydrophobic, heavy (less water soluble PAHs with five and more aromatic rings). MESCO-derived aqueous concentrations of light PAHs were higher than those derived from SPMDs. The opposite trend was observed for heavy PAHs. There may be various sources of differences in absolute values calculated using the two methods. First, neither of the two methods provided accurate estimates of TWA concentrations for light PAHs, because both samplers nearly approached partitioning equilibrium. Thus, values 245 A. Paschke et al. 0 I" ,1I,1I,1I,1I,1I,1I,JI, J,-n,-n,-n,JI, <$> <$> <$> <$> <$> <$> <$> <$> O® ^ #vc? <^10 000. 29.2.4.2 Analysis of polar compounds Polar pesticides and pharmaceuticals were analysed by liquid chromatography (Waters Acquity) with MS detection (Waters Xevo TQ-S). Analytes were separated on reverse phase column (Waters Acquity UPLC BEH-C18) using gradient elution with methanol and water, both with 0.1% formic acid. Eluting analytes were ionized using electrospray in positive mode and detected in MRM mode. 29.2.4.3 Toxicological profiling For toxicological profiling, a battery of bioassays has been established. The same tests are employed for assessment of toxic potential of samples from high volume active sampling (Chapter 27). The set consists of eight assays provided by four laboratories (INERIS, RECETOX, RWTH, and University of Queensland (UQ)). The selected bioassays cover several important steps in the toxicity pathway including induction of xenobiotic metabolism, specific and reactive modes of toxic action, activation of adaptive stress response pathways. The diverse modes of action provide broad range of information on toxic potential. Specifically, there are assays for assessment of endocrine disruptive potential (anti-)estrogenicity (MELN) and (anti-)androgenicity (MDA-kb2), activation of receptors for xenobiotics (CAFLUX and HG5LN-hPXR), immune response (NF-KB-bla THP-1), mutagenicity and DNA damage -related apoptosis (Ames fluctuation assay and p53-bla HCT-116, resp.) and detection of response to oxidative stress (ARE-bla Hep G2). The model cell lines are exposed to dilution series of the ED and SR extracts to describe dose-response relationship of the effects. The potentials are quantified in comparison with negative control and positive control describing the effect of a model chemical with known toxic potency specific for each of the bioassay endpoints. Table 90: List of bioassays employed in the toxicological profiling of passive sampler extracts Laboratory Bioassay Endpoint INERIS MELN Binding to and activation of human estrogen receptor (ER)1 HG5LN-hPXR Binding to and activation of the human pregnane X receptor (PXR)2 RECETOX CAFLUX Binding to and activation of aryl hydrocarbon receptor (AhR)3 MDA-kb2 Binding to and activation or inhibition of activity of human androgen receptor (AR)4 RWTH Ames fluctuation assay Assessment of mutagenic activity in Salmonella typhimurium after metabolic activation of compounds with S9 liver fraction5 UQ p53-bla HCT-116 Assessment of p53-mediated apoptosis rate in response to DNA damage6 ARE-bla Hep G2 Induction of the Nrf-2-mediated oxidative stress pathway7 NF-KB-bla THP-1 Induction of inflammatory response8 1(Balaguer et al„ 1999), 2(Lemaire et al„ 2006), 3(Aarts et al„ 1998), "(Wilson et al„ 2002), ^Reifferscheid et al„ 2012), 6(Yeh et al„ 2014), 7(http://tools.lifetechnologies.com/content/sfs/manuals/cellsensor_AREblaHepG2_man.pdf, n.d.), 8("http://tools.lifetechnologies.com/content/sfs/manuals/CellSensor_NFkBbla_THP1_man.pdf," n.d.) CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 308 29.2.5 QA/QC The applied quality control measures included the analysis of procedural solvent blanks, fabrication controls, field controls and matrix spikes. 29.2.6 Data analysis Dissolved water concentrations of were calculated from analyte amounts accumulated in SR. and LDPE samplers, the in situ sampling rate (Rs) of the compounds and their sampler-water partition coefficients (Smedes et al., 2009) as described in Smedes and Booij (2012). Sampling rates were estimated from dissipation of PRCs from samplers during exposure using methods described by Booij and Smedes (2010). For ED samplers calibration data are not available so far. For compounds under investigation we assumed an integrative uptake with a constant sampling rate. Identification of pollutant gradients along the Danube was performed based on the amount of a compound sampled by the ED in individual stretches, normalised to an average sampler exposure time (1.6 days). 29.3 Results 29.3.1 Analysis of hydrophobic compounds- use of silicone rubber samplers SR samplers were deployed at 8 successive Danube stretches to characterise the spatial variability of hydrophobic compounds in the water column of the river. 29.3.1.1 Polychlorinated biphenyls and brominated diphenyl ethers Calculated dissolved PCB concentrations were in sub ng l"1 range (Figure 150). Sums of 6 indicator PCB congeners ranged from 158 to 369 pg l"1. Over the set of PCBs investigated there is a decrease in free dissolved concentration as hydrophobicity increases. The highest spatial variability is observed for the more water soluble congeners PCB28, 52 and 101. There was no clear spatial trend of PCB contamination along the river. Concentrations of freely dissolved PBDEs (referring to the sum of the concentrations of congener numbers 28, 47, 99, 100, 153 and 154) were below the limit of quantification of 3 pg l"1 with the exception of the stretch Passau to Bratislava, where the summed concentration of the 6 congeners was 12 pg l"1. Measurement of such low concentrations would require longer exposure times for integrative sampling, which was not available during the JDS3 cruise. A parallel 43 day sampling using a caged SR sampler statically deployed at a sampling site downstream Bratislava in the period August-October 2013 provided a concentration estimate of 2 pg l"1 for the sum of 6 PBDE congeners (Vrana, unpublished data). PCB 28 PCB 52 PCB 101 PCB 118 PCB 153 PCB 138 PCB 180 Figure 150: Free dissolved concentration of PCBs measured by SR samplers in 8 Danube stretches CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 309 29.3.1.2 Organochlorine compounds The free dissolved concentrations of OCs were in sub ng l"1 range (Figure 151).The highest concentration of pentachlorobenzene (PeCB) up to 96 pg l"1 was observed in the stretch between Budapest and Belgrade whereas the highest level of hexachlorobenzene (HCB) of 97 pg l"1 was measured in the lowest Danube stretch between Ruse and Tulcea. The spatial variability of PeCB concentration was higher than that of HCB. Among the hexachlorocyclohexane (HCH) congeners, only (3-HCH is reported because of low extraction recovery of the remaining isomers. There is an increasing trend of (3-HCH concentration along the river, ranging between 9 pg l"1 in the upper stretches and 259 pg l"1 in the river delta area, respectively. The same spatial trend can be observed also for the sum of total DDT (given as sum of 4 isomers according to the Directive 2008/105/EC) as well as for p,p-DDT. Concentrations of p,p-DDT (1-21 pg l"1) comprised only 2-7% of the total DDT, which indicates no current use of DDT in the Danube catchment. In the delta area concentration of DDT metabolites reach levels up to 864 pg l"1. 1000 900 800 700 (I/Bd) 600 c ro 500 c a u 400 o o 300 200 100 0 I Passau-Bratislava I Bratislava-Budapest I Budapest-Vukovar I Vukovar-Belgrade I Belgrade-Turnu-Severin iTurnu-Severin-Ruse I Ruse-Braila I Braila-Tulcea PeCB HCB b-HCH p,p'DDT(x 10) S DDT Figure 151: Free dissolved concentration of OCPs measured by SR samplers in 8 Danube stretches 29.3.1.3 Polycyclic aromatic hydrocarbons Summed concentrations (El6 US EPA PAHs) of free dissolved PAHs in the water column ranged between 10.6 ng l"1 in stretch 7 and 45.1 ng l"1 in stretch 4, respectively. Summed concentrations were largely composed of PAHs with up to 4 aromatic rings. As for PCBs there is a strong decrease of free dissolved concentration with increasing compound hydrophobicity (Figure 152). Concentrations of compounds with 6 aromatic rings were mostly below the limit of quantification (tens of pg l"1). Elevated PAH concentrations were observed in the stretches 4 and 5 (Budapest to Vukovar) and stretch 5 (Vukovar to Belgrade) with distinct pollutant patterns, which indicates different sources of PAHs along those river stretches. Concentrations of individual PAHs measured in stretch 2 (Passau to Bratislava) are within the concentration range that was measured in that stretch in spring till autumn 2011 using SPMD passive samplers (Vrana et al., 2014). This indicates that free dissolved PAH concentrations and their patterns in that Danube stretch in the summer period remained stable over a period of several years. A comparison with free dissolved concentrations measured using passive sampling in other European rivers (Vrana et al, 2014) shows that the concentrations of PAHs in the Danube is comparable to about 10 times lower. CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 310 10000 8000 li || 1 L I Passau-Bratislava I Bratislava-Budapest I Budapest-Vukovar ■ Vukovar-Belgrade I Belgrade-Turnu-Severin Turnu-Severin-Ruse Ruse-Braila Braila-Tulcea 1 Ü 11 " ■r ■S' & & >> I7> »> »> Figure 152: Free dissolved concentration of PAHs measured by SR samplers in 8 Danube stretches 29.3.1.4 Alkylphenols The highest concentrations of free dissolved 4-nonylphenol (4-NP; 9.2 ng l"1) and that of 4-tert-octylphenol (4-t-OP; 0.36 ng l"1) was observed in the stretch between Vukovar and Belgrade (Figure 153). Concentration of 4-t-OP was on average 50 times lower than that of 4-NP. 10.0 4-tert-octylphenol (xlO) 4-nonylphenol Figure 153: Free dissolved concentration of alkylphenols measured by SR samplers in 8 Danube stretches CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 311 29.3.2 Analysis of polar compounds - use of Empore disk samplers 29.3.2.1 Polar pesticides A suite of 40 polar pesticides was analysed in extracts from the ED samplers. Results of analysis of five WFD priority pollutant polar pesticides, namely alachlor, atrazine, diuron, isoproturon and simazine are shown in Figure 154. Alachlor and diuron were present at concentrations less than or close to limit of quantification, which roughly corresponds to concentrations less than 100 pg l"1 in water. Estimated concentrations of atrazine, simazine and isoproturon in water were in the order of units of ng l"1 with the maxima of these pesticides in the stretch from Ruse to Braila. The results indicate that concentrations of the priority polar pesticides were far below their EQS values. It has to be noted that the main period of pesticide application is April-July and therefore the JDS results are not representative for the application season of these compounds. 700 600 200 100 0 1 ■ Passau-Bratislava ■ Bratislava-Budapest ■ Budapest-Vukovar ■ Vukovar-Belgrade ■ Belgrade-Turnu-Severin 1. ■ Turnu-Severin-Ruse Ruse-Braila II ■ 1 1 ■ _ ■ Braila-Tulcea bJ hJ Alachlor Atrazine Diuron Isoproturon Simazine Figure 154: Spatial variability of WFD priority pollutant polar pesticides in the water column measured by ED samplers in 8 Danube stretches. Data is expressed as amount of compound taken up by an integrative sampler during an average sampler exposure (1.6 days) 29.3.2.2 Alkylphenols The longitudinal relative concentration profile of alkylphenols in the Danube, measured by ED samplers (Figure 155), was similar to that reported by SR samplers. The highest concentrations of both 4-t-OP and 4-NP, but also of bisphenol A was measured in the stretch from Vukovar to Belgrade. In ED samplers concentration of 4-t-OP was on average 40 times lower than that of 4-NP. CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 312 1400 Bisphenol A 4-t-OP 4-NP Figure 155: Spatial variability of alkylphenols in the water column measured by ED samplers in 8 Danube stretches. Data is expressed as amount of compound taken up by an integrative sampler during an average sampler exposure (1.6 days) 29.3.2.3 Pharmaceuticals Results of analysis of caffeine and two pharmaceuticals, carbamazepine and diclofenac in extracts from the ED samplers are shown in Figure 156. The trend of caffeine concentration in the water column along the river was similar to that of bisphenol A. Estimated caffeine concentration levels were up to several tens of ng 1-1 with the maximum observed concentration in the stretch from Vukovar to Belgrade. For comparison, analyses of caffeine in discrete spot samples taken collected the cruise and analysed by ELISA showed median concentration in Danube of 93 ng 1-1 (Chapter 26). Estimated concentrations of carbamazepine along the river were in units of ng 1-1 and less variable than that of caffeine. In agreement with the measurements made during JDS2 diclofenac was present at concentrations less than or close to limit of quantification, which can be explained by the biodegradability of this compound (Loos et al., 2008). 1200 1000 I Passau-Bratislava I Bratislava-Budapest I Budapest-Vukovar I Vukova r-Belgrade I Belgrade-Turnu-Severin ITurnu-Severin-Ruse Ruse-Braila I Braila-Tulcea caffeine carbamazepine diclofenac Figure 156: Spatial variability of caffeine and selected pharmaceuticals in the water column measured by ED samplers in 8 Danube stretches. Data is expressed as amount of compound taken up by an integrative sampler during an average sampler exposure (1.6 days) CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 313 29.3.3 Toxicological profiling Selected toxic/bioactive potentials (see Table 90) of extracts of SR and ED passive samples are currently under evaluation. Preliminary results indicate that SR extracts contain significant amounts of dioxin-like compounds assessed by CAFLUX bioassay (Figure 157). Estimated toxic equivalents (bioTEQ) of samples recalculated for the sampled volume are between 6-10 pg l"1. MELN bioassay has indicated estrogenic activity in SR samples. The specific estrogenic potential needs to be quantified yet. Available data from HG5LN-hPXR bioassay show that some SR extracts can significantly activate pregnane X receptor, but not the androgenic receptor. Negative results have been obtained in case of mutagenicity of SR extracts in Ames assay. Preliminary data indicate that at least some of the ED samples possess quantifiable estrogenic and PXR-related potential significantly higher than field blank samples. 12 Figure 157: Estimate of toxic equivalent of TCDD in the water column measured by SR samplers in eight Danube stretches determined in CAFLUX bioassay 29.4 Conclusions Despite the low or sub- ng l"1 concentrations of most organic pollutants present in the free dissolved phase, passive sampling enabled to clearly identify spatial gradients of a broad range of organic pollutants in the water column, including PCBs, OCs, PAHs, alkylphenols, selected polar pesticides and pharmaceuticals. In many cases, the integrative character of passive sampling allowed measurement of compounds down to pg l"1 levels where methods based on low volume spot sampling of water applied in the previous JDS2 survey failed to detect them (Sengl, 2008). Passive samplers in most cases confirmed similar spatial distribution of pollutants along the river, as was observed in JDS2. The highest levels of PAHs, alkylphenols and caffeine in passive samplers were observed in the Danube stretches between Budapest and Belgrade. In agreement with JDS2, the downstream profile of PCBs and HCB showed a low variability and did not suggest particular emission maxima (Umlauf et al, 2008). In accordance with the findings during the JDS1 and JDS2, the downstream profile of (3-HCH, DDT and its metabolites displays a sharp increase in the water column downstream Braila towards the Black Sea (Umlauf et al, 2008). The low percentage of p,p-DDT of the total DDT concentration indicates that there was no current use of DDT in the area. The levels of priority pollutant polar pesticides alachlor, atrazine, diuron, isoproturon and simazine were comparable with the levels found in water samples during JDS2 and well below their respective EQS values (Loos et al, 2008). Whereas data from spot sampling reflects the pollution at the individual JDS sampling sites at a single moment of time, passive samplers continuously sampled pollutants for several days, including river CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 314 stretches between individual JDS sampling sites. Thus, the information provided by spot sampling and passive sampling should be considered as complementary. Finally, the combination of passive samplers with bioassays presents a very promising approach for detection of various trace organic pollutants and toxic potentials along the river and for identification of areas of concern for further investigation. 29.5 References AARTS, J.M.M.J.G., JONAS, A., DIKKENBERG, VAN DEN L.C., BROUWER, A., 1998. CAFLUX, a simplified version of the CALUX assay for Ah receptor (ant)agonist, based on enhanced green fluorescent protein (EGFP) reporter gene expression. Organohalogen Compd. 37, 85-88. ALLAN, I.J., HARMAN, C, RANNEKLEV, S.B., THOMAS, K. V, GRUNG, M., 2013. Passive sampling for target and nontarget analyses of moderately polar and nonpolar substances in water. Environ. Toxicol. Chem. 32, 1718-26. BALAGUER, P., FRANCOIS, F., COMUNALE, F., FENET, H., BOUSSIOUX, A.-M., PONS, M., NICOLAS, J.-C, CASELLAS, C, 1999. Reporter cell lines to study the estrogenic effects of xenoestrogens. Sci. Total Environ. 233, 47-56. BOOIJ, K., SMEDES, F., 2010. An improved method for estimating in situ sampling rates of nonpolar passive samplers. Environ. Sci. Technol. 44, 6789-94. http://tools.lifetecMologies.com/contenťsfs/manuals/cellsensor_AREblaHepG2_man.pdf [WWW Document], n.d. http://tools.lifetecMologies.com/contenťsfe^ [WWW Document], n.d. LEMAIRE, G., MNIF, W., PASCUSSI, J.-M., PILLON, A., RABENOELINA, F., FENET, H., GOMEZ, E., CASELLAS, C, NICOLAS, J.-C, CAVAILLĚS, V., DUCHESNE, M.-J., BALAGUER, P., 2006. Identification of new human pregnane X receptor ligands among pesticides using a stable reporter cell system. Toxicol. Sci. 91, 501-9. LOOS, R., LOCORO, G., CONTINI, S., 2008. Polar water-soluble contaminants in the liquid water phase by SPE-LC-MS2, in: Liska, L, Wagner, F., Slobodnik, J. (Eds.), Joint Danube Survey 2 - Final Scientific Report. -International Commission for the Protection of the Danube River, Vienna, pp. 170-173. MAYER, P., TOLLS, J., HERMENS, L., MACKAY, D., 2003. Equilibrium Sampling Devices. Environ. Sci. Technol. 37, 184A-191A. REICHENBERG, F., MAYER, P., 2006. Two complementary sides of bioavailability: accessibility and chemical activity of organic contaminants in sediments and soils. Environ. Toxicol. Chem. 25, 1239-45. REIFFERSCHEID, G., MAES, H.M., ALLNER, B., BADUROVA, J., BELKIN, S., BLUHM, K., BRAUER, F., BRESSLING, J., DOMENEGHETTI, S., ELAD, T., FLUCKIGER-ISLER, S., GRUMMT, H.J., GURTLER, R., HECHT, A., HERINGA, M.B., HOLLERT, H., HUBER, S., KRAMER, M., MAGDEBURG, A., RATTE, H.T., SAUERBORN-KLOBUCAR, R., SOKOLOWSKI, A., SOLD AN, P., SMÍTAL, T., STALTER, D., VENIER, P., ZIEMANN, C, ZIPPERLE, J., BUCHINGER, S., 2012. International round-robin study on the Ames fluctuation test. Environ. Mol. Mutagen. 53, 185-97. SENGL, M., 2008. EU WFD organic priority substances in water, suspended particulate matter, sediments and biota and other organic pollutants, in: Liska, L, Wagner, F., Slobodnik, J. (Eds.), Joint Danube Survey 2 - Final Scientific Report. ICPDR-International Commission for the Protection of the Danube River, Vienna, pp. 132-146. SMEDES, F., BOOIJ, K, 2012. Guidelines for passive sampling of hydrophobic contaminants in water using silicone rubber samplers. ICES Tech. Mar. E Environ. Sci. 52. SMEDES, F., GEERTSMA, R.W., VAN DER ZANDE, T., BOOIJ, K, 2009. Polymer-water partition coefficients of hydrophobic compounds for passive sampling: Application of cosolvent models for validation. Environ. Sci. Technol. 43, 7047-7054. UMLAUF, G., CHRISTOPH, E., HUBER, T., MARIÁNI, G., MUELLER, A., SKEJO, H., WOLLGAST, J., 2008. Cross matrix inter-comparison of semi-volatile organic compounds in water, suspended particulate matter, sediments and biota, in: Liska, L, Wagner, F., Slobodnik, J. (Eds.), Joint Danube Survey 2 - Final Scientific Report. ICPDR-International Commission for the Protection of the Danube River, Vienna, pp. 174-191. VRANÁ, B., KLU ČÁROVÁ, V., BENICKÁ, E., ABOU-MRAD, N, AMD ANY, R., HORÁKOVÁ, S., DRAXLER, A., HUMER, F., GANS, O., 2014. Passive sampling: An effective method for monitoring seasonal and spatial variability of dissolved hydrophobic organic contaminants and metals in the Danube river. Environ. Pollut. 184, 101-112. CPDR / International Commission for the Protection of the Danube River / www.icpdr.org 29 Passive sampling: chemical analysis and toxicological profiling 315 Website of the European Ferrybox Community [WWW Document], 2014. URL http://www.ferrybox.org/ (accessed 6.18.14). WILSON, V.S., BOB SEINE, K., LAMBRIGHT, C.R., GRAY, L.E., 2002. A novel cell line, MDA-kb2, that stably expresses an androgen- and glucocorticoid-responsive reporter for the detection of hormone receptor agonists and antagonists. Toxicol. Sci. 66, 69-81. YEH, R.Y.L., FARRE, M.J., STALTER, D., TANG, J.Y.M., MOLENDIJK, I, ESCHER, B.I., 2014. Bioanalytical and chemical evaluation of disinfection by-products in swimming pool water. Water Res. 59C, 172-184. 29.6 Acknowledgments We acknowledge the NORMAN association www.norman-network.net, the SOLUTIONS Project supported by the European Union Seventh Framework Programme (FP7-ENV-2013-two-stage Collaborative project) under grant agreement 603437, and the RECETOX NETWORKING project supported by the EU Operational Programme "Education for Competitiveness" (CZ1.07/2.3.00/20.0053) for the financial support. This research has been co-funded from the European Social Fund and the state budget of the Czech Republic. Ian Allan and Merete Grung acknowledge NIVA funding through the RivScreen project (2013-2014), project O-13036. Authors thank to Petra Přibylová, Petr Kukučka, Simon Vojta, Ondřej Audy, Jiří Kohoutek, Jitka Bečanová, Marek Pernica and Zdeněk Simek from RECETOX, Masaryk University for the instrumental analysis of samples. CPDR / International Commission for the Protection of the Danube River / www.icpdr.org Príloha 33 Vrana B., Vermeirssen E. L. M., Allan I., Kohoutek J., Kennedy K., Mills G., and Greenwood R., Passive sampling of emerging compounds in the environment: state of the art and perspectives, 2010. [Online]. Available: http://www.norman-network.net/sites/default/files/files/ Events/2009/2009May27-Prague-PassiveSampling/norman_position_paper_pas_sampling.pdf. rma NORMAN Network of reference laboratories, research centres and related organisations for monitoring of emerging environmental substances Passive sampling of emerging pollutants in the aquatic environment: state of the art and perspectives Position Paper Branislav Vranaad*, Etienne L. M. Vermeirssenb Ian J. Allanc, Jiff Kohoutekd, Karen Kennedy6 Graham A. Mills' and Richard Greenwood9 aSlovak National Water Reference Laboratory, Water Research Institute, Nabr. arm. gen. L. Svobodu 5, 81249 Bratislava, Slovakia bEawag, Swiss Federal Institute of Aquatic Science and Technology, Uberlandstr. 133, 8600 Dubendorf, Switzerland Norwegian Institute for Water Research, Gaustadalleen 21, NO-0349, Oslo, Norway dResearch Centre for Toxic Compounds in the Environment (RECETOX), Masaryk University, Kamenice 126/3, 625 00 Brno, Czech Republic eThe University of Queensland, The National Research Centre for Environmental Toxicology (Entox), 39 Kessels Rd, Brisbane QLD, 4108, Australia 'School of Pharmacy and Biomedical Sciences, University of Portsmouth, White Swan Road, P01 2DT, UK 9School of Biological Sciences, University of Portsmouth, King Henry I Street, Portsmouth, P01 2DY, UK 'Contact person: Branislav Vrana T +421 2 59343466 F +004212 54418047 e-mail branovrana(5)googlemail.com This document has been written as a follow-up to the expert group meeting organised by the NORMAN association on 27th May 2009 in Prague. It reflects the position of the NORMAN association experts and invited speakers on the topic of passive sampling and its application in the monitoring of emerging pollutants in aquatic environment. NORMAN Association N° W604002510 Network of reference laboratories, research centres and related organisations for monitoring of emerging environmental substances http:/Avww. norman-network. net Table of contents I. Introduction............................................................................................................................3 II. Concept of passive sampling................................................................................................5 III. Applications in aquatic monitoring of emerging compounds...........................................9 III. 1. Algal toxins...................................................................................................................9 111.2. Antifouling compounds - organotins.........................................................................9 111.3. Brominated flame retardants...................................................................................10 111.4. Endocrine disrupting compounds............................................................................10 111.5. Fluorinated surfactants..............................................................................................11 111.6. Organosiloxanes.........................................................................................................11 111.7. Pharmaceuticals.........................................................................................................12 111.8. Polar pesticides..........................................................................................................13 111.9. Sunscreen and ultra-violet filters.............................................................................14 IV. Application in sediment monitoring.................................................................................14 V. Application in monitoring of contaminants in biota.........................................................16 VI. Application in ecotoxicity assessment...............................................................................18 VI. 1. Passive samplers as mimics for bioconcentration...................................................18 VL2. Which passive sampler suits which bioassay?.........................................................19 VI.3. The link between biological and chemical analysis.................................................20 VI.4. Identification of toxic compounds in passive samplers: effect-directed analysis.20 VI.5. How does the bioassay response in passive sampler extracts relate to sampler exposure conditions?...........................................................................................................21 VII. Quality assurance, quality control and normation.........................................................21 VIII. Application of passive samplers in regulatory monitoring...........................................22 IX. Future trends.....................................................................................................................24 Executive Summary Passive samplers represent an innovative monitoring tool for the time-integrated measurement of bioavailable contaminants in water and sediment. Passive sampling technology is proving to be a reliable, robust and cost-effective tool that could be used in monitoring programmes across Europe. These devices are now being considered as a part of an emerging strategy for monitoring a range of priority and emerging pollutants. Passive sampling is based on the deployment in-situ, or use in the laboratory, of non-mechanical devices of simple construction capable of accumulating contaminants dissolved in water or sediment pore water. Such accumulation occurs via diffusion, typically over periods of days to weeks. Contaminants accumulated in exposed samplers are subsequently extracted and their concentration levels measured, allowing the quantification of time-weighted average (TWA) concentrations in water or equilibrium pore water concentrations in sediment. These devices can be deployed in most aquatic conditions (fresh and saline) and associated water treatment facilities, thus making them ideal for monitoring across the entire water cycle and even in remote areas with minimal infrastructure. Passive sampling can also be employed in batch sediment extractions to provide estimates of contaminant concentrations in pore water or assessment of bioavailable concentrations of contaminants in sediment. In 2009, the NORMAN association organised a meeting of experts in the field of passive sampling. As a result of this meeting a position paper was produced, which reflects the view of the experts on the topic of passive sampling and its application in the monitoring of emerging pollutants in the aquatic environment and indicates future research and development needs in this area. The position paper discusses functional principles of passive samplers and problems associated with the effects of environmental variables (temperature, water turbulence and sampler fouling) on their performance. Further, it lists the established or expected/potential performance of passive samplers for monitoring of the most discussed groups of emerging substances (such as cyanobacterial toxins, antifouling agents, brominated flame retardants, endocrine disrupting compounds, fluorinated surfactants, organosiloxanes, pharmaceuticals, polar pesticides, sunscreen filters etc.) and availability of calibration data that enable estimation of TWA concentrations. The document also shows the applicability of the passive sampling concept in risk-oriented monitoring of emerging substances in sediments and in determination of the bioaccumulative exposure of organisms. The great potential of this technology in combination with toxicological assays to determine the biological relevance of mixtures of toxicants with specific modes of action, and present at low concentrations, is also demonstrated. If passive sampling is to become accepted and used in a regulatory context for monitoring water quality across Europe, then there is a need for the development of improved validation methods and setting-up of the appropriate quality control and quality assurance schemes for the technology. Successful demonstration of the performance of passive samplers alongside conventional sampling schemes, and inter-laboratory studies that demonstrate reproducibility of data produced by different designs of passive samplers, are urgently needed to facilitate the acceptance of passive sampling in routine regulatory monitoring programmes in the future. page 2 I. Introduction Improvements in analytical methods, primarily the introduction of more sensitive and specific mass spectrometry techniques, have increased awareness of the presence of emerging substances from many sources at trace levels (low ng L"1) in the aquatic environment [1]. These substances include industrial chemicals and products, consumer products such as pharmaceuticals (both prescription and non-prescription drugs) and personal-care products, pesticides, natural bioactive compounds such as cyanotoxins and hormones, and metabolites of all these chemicals. Previous research focused mainly on non-polar and mono-polar compounds such as PCBs (polychlorinated biphenyls), PAHs (polycyclic aromatic hydrocarbons), chlorinated solvents, or chlorinated pesticides such as DDT or lindane. More recently attention has turned to the modern polyfunctional and often ionisable pesticides, biocides, drugs and personal care products. Currently there is a lack of knowledge regarding the fate and effects of many chemicals released into the environment either as products or accidentally. Although most of these compounds are present in the environment at low concentrations, many of them raise considerable toxicological concerns, particularly when present as components of complex mixtures [2]. Exposure assessment in the aquatic environment is based primarily on analytical measurements of chemical compounds in samples from various environmental compartments - water, sediments, soils, air - as well as from organisms from different trophic levels within a food chain [2]. Understanding and quantification of processes which emerging compounds can undergo in the environment, such as adsorption and partitioning between solid and aqueous phases, formation of complexes in solution as well as abiotic and biological transformation, are also urgently required. Both effective sampling and analytical methods are therefore essential to obtain reliable data on the concentrations, speciation and fate of these compounds in the aquatic environment. While a lot of effort has been put into research and development of increasingly sensitive instrumental analytical methods for the measurement of emerging substances in various matrices in the aquatic environment, less interest has been paid to the development of suitable sampling techniques. Until recently, sampling methods for emerging substances were the same as those routinely used for monitoring priority pollutants in the aquatic environment. These are based on periodic collection of spot or grab bottle samples of water. The subsequent laboratory analysis of the sample provides a snapshot of the levels of pollutants at the time of sampling. There are, however, drawbacks to this approach in environments where contaminant concentrations vary over time, and where episodic pollution events such as spills or storm water runoff can easily be missed. This problem is particularly relevant to polar (hydrophilic) emerging substances. The residence times of these compounds in aquatic systems are generally lower than those of hydrophobic organic compounds. However, the presence of these more hydrophilic compounds in these systems (wastewater, surface water) may occur as a result of relatively episodic events (frequent, short duration and high concentration peaks). Thus, there is an urgent need for the development of suitable sampling and analytical methods capable of detecting and identifying contaminants in an integrative manner for an adequate assessment of the environmental risk posed by emerging substances. One solution to this problem is to increase the frequency of sampling or to install automatic sampling systems that can collect numerous water samples over a given period. For example, the pooling of samples collected hourly into a 24 h composite sample, or page 3 continuous on-line monitoring for specific sets of compounds can be used to provide representative data. These methods are both costly and in many cases impractical, since a secure site and additional infrastructure or personnel are required to protect, operate and maintain the mechanical automatic sampling devices. Over the last decade alternative methods for monitoring water quality have been sought to overcome some of the difficulties. A developing alternative strategy to these traditional sampling methods is to employ passive sampling devices that can be deployed over extended time periods (days to weeks) to provide time-weighted average (TWA) concentrations [3,4]. Passive sampling is a relatively easily applied sampling technique, based on the use of non-mechanical samplers of simple construction, often consisting of a single polymeric sorbing phase. In most cases these samplers do not require any external energy source to function. These devices can be deployed in most aquatic conditions (fresh and saline) and associated water treatment facilities, thus making them ideal for monitoring across the entire water cycle and even in remote areas with minimal infrastructure. Furthermore, these samplers assist with the sensitivity of subsequent analytical methods as they pre-concentrate and preserve chemicals sampled within these polymeric receiving phases. This enables improved sensitivity for a greater range of compounds and improved stability of chemicals within the sample without additional treatment (e.g. pH adjustment) unlike more traditional grab sampling techniques. In some cases, the use of passive samplers can also help to reduce or even eliminate the use of excessive volumes of toxic extraction solvents. Passive samplers have been used for environmental monitoring since the 1970s, when the first samplers for the assessment of ambient air quality and workplace exposures to potentially hazardous air pollutants were developed and applied. To date, a number of sampler designs are commercially available and there are now established standards and official methods (e.g. ASTM, EPA, NIOSH, CEN and ISO protocols) for the use of these devices, which form part of legal frameworks. More recently, worldwide monitoring networks have been set up using passive air samplers to monitor persistent organic pollutants on a global scale [5,6]. In contrast, the application of passive samplers in monitoring water quality is some way behind the situation for air, and the technologies available for monitoring soils and sediments are even further from recognition. Since the introduction of the semi-permeable membrane device (SPMD), designed at USGS by Huckins et al. [7] in the early 1990s, passive samplers have become widely used for monitoring persistent organic pollutants and other non-polar organic compounds in the aquatic environment. Nearly ten years later, the passive sampling technology suitable for sampling hydrophilic organic compounds including modern pesticides, pharmaceuticals and personal care products has been reported in the work of Alvarez (POCIS sampler) [8] and Kingston et al. (Chemcatcher concept) [9]. Since then, the number of publications on development, performance optimisation and field application of passive samplers for emerging substances has grown rapidly. A number of recent reviews have been published describing the design, calibration procedures, figures of merit and applications of the different devices for monitoring the aquatic environment [3,10,11,12]. Booij summarised in a report for the ICES Marine Chemistry Working Group the established or expected/potential performance of various passive samplers of compounds that are listed under WFD and other directives or conventions [13]. Recently, several review papers addressing passive sampling of emerging pollutants have been published [14,15]. In addition, a book describing the SPMD [16] and a page 4 general text describing many passive sampling techniques for environmental monitoring [17] are available. II. Concept of passive sampling Passive sampling is based on the deployment in-situ or use in the laboratory of devices capable of accumulating contaminants dissolved in water or sediment pore water. Such accumulation occurs via diffusion, typically over periods of days to weeks. Contaminants accumulated in exposed samplers are subsequently extracted and their concentration levels measured, allowing the quantification of TWA concentrations in water or equilibrium pore water concentrations in sediment. It enables temporally-representative sampling or sampling of the truly dissolved concentration of contaminants in water or aquatic sediments. Even for those chemicals that are present at extremely low concentrations in the dissolved phase and are primarily accumulated in biota via the dietary uptake, passive samplers generally extract sufficient amounts of residues for analysis. Passive sampling can also be employed in batch sediment extractions under laboratory conditions to provide estimates of contaminant concentrations in pore water or assessment of bioavailable fraction of contaminant in sediment [18,19]. Passive sampling is based on the diffusion of analyte molecules from the sampled environmental medium (water or sediment pore water) to a receiving phase in the sampling device. The diffusion occurs as a result of a difference between chemical potentials of the analyte in the two media (Figure 1). The net flow of analyte molecules from one medium to the other continues until equilibrium is established in the system, or until the sampling is stopped. The mass of chemical sorbed in the sampler following a given exposure period is initially proportional to the TWA concentration in the environmental medium to which the sampler was exposed (integrative samplers) and subsequently once equilibrium is achieved to the concentration in the environmental medium with which the device is at thermodynamic equilibrium (equilibrium samplers). The main advantage of kinetic or integrative sampling is that even contaminants from episodic events commonly not detected with spot sampling are collected by the sampler. This permits the measurement of time weighted average (TWA) contaminant concentrations over extended time periods using a single sample (extract from the passive sampler). This gives a more representative picture of contaminant levels than that obtained with the use of infrequent spot samples. To achieve equilibrium sampling, for a given sampler the sampling period needs to be sufficiently long to establish thermodynamic equilibrium between the water and the sorbent phase of the sampler. To achieve equilibrium within reasonable sampling periods samplers of relatively low capacity for the analytes of interest or with modified surface area to volume ratios may be required [20]. Application of the sampler-water distribution coefficient then enables the calculation of the analyte concentration in the sampled medium. Analytes are accumulated in a suitable sorbent material within the passive sampler, known as a receiving phase. This can be a solvent, chemical reagent, absorbent polymer or a porous adsorbent material. Whereas most samplers of hydrophobic compounds are based on diffusion and absorption in non-porous polymers, most samplers of polar organic compounds (i.e. majority of emerging compounds) and metals are based on diffusion through porous membranes and sorption to selective adsorbent materials. The difference in selection of materials applied in sampler construction results in different sorption phenomena that define the driving force of the sampling process (Figure 2). In general, accumulation of hydrophilic organic compounds to porous adsorbents is more complex than absorption and page 5 dissolution of hydrophobic chemicals in non-porous polymers (polyethylene or polydimethylsiloxane). This is because adsorption distribution coefficients (unlike partition coefficients in solvents and sub-cooled liquid polymers) described by sorption isotherms can be concentration-dependent. Competitive adsorption of analytes and possible interferences are also possible. The polar organic compounds are mainly retained by specific interactions with functional groups at the surface of the adsorbent. Although the use of adsorptive polymers with specific interactions is preferred in certain cases, the risk always exists of saturating the fixed number of superficial bonding sites when these polymers are applied to a complex sample matrix. Finally, many compounds may speciate into multiple forms depending on their pKa parameters and the pH of the sampled medium. Where a sorbent phase only accumulates a single form of a specific compound such as the neutral species, these phenomena will also influence the observed uptake. Sampling description is thereby complicated by the presence of several species with different diffusion and sorption properties that may dynamically change during the sampling process, depending on a milieu of properties of both the sampled medium, the receiving phase and of the individual compound. Recently, a novel absorptive equilibirum passive sampler for polar organic compounds has been reported by Magnér et al. [21]. This is based on a plastic material, polyethylene-co-vinyl acetate-co-carbon monoxide (PEVAC). This receiving phase operates as a homogenous, non-porous liquid in which the analytes are retained by dissolution rather than by specific interactions with the surface of the polymer. The PEVAC material showed enhanced sorption of several polar pesticides and pharmaceuticals compared to the silicone material. Identification of suitable absorbent polymer materials with high retention capacity of polar compounds presents a promising approach in future development of passive sampling technology and may replace currently used complex adsorption-based samplers for which data conversion into aqueous concentrations is often difficult. For devices that operate in the kinetic or integrative mode, the sampling rate is given by the product of the overall analyte mass transfer coefficient and the active surface area of the sampler (Rs = k0 A). Sampling rate may be interpreted as the volume of water cleared of analyte per unit of exposure time (e.g. ml_ h"1 or L day"1) by the device and is independent of the analyte concentration in the sampled medium. It can be affected and modulated by the analyte diffusion and partition properties in the media along the diffusional path, and is determined in laboratory calibration studies. Often the main barrier to mass transfer is the water boundary layer (WBL) located at the external surface of the sampler. In such a case the sampling rate is significantly affected by environmental variables such as water temperature, turbulence and biofouling. If laboratory calibration data is to be used for calculation of TWA concentrations, the effect of these variables has to be either controlled or quantified. For samplers used to measure concentrations of non-polar organic analytes, one method of overcoming some of the problems associated with the impact of fluctuating in situ environmental conditions (temperature and turbulence) on sampling rate is the use of performance reference compounds (PRCs) [22]. These are analytically non-interfering compounds (typically deuterium or 13C labelled analogues of the compounds to be measured) and are loaded onto the receiving phase of the sampler prior to deployment. These PRCs are eliminated from the receiving phase during the deployment period. Where the kinetics of uptake and elimination are isotropic, that is the rate constants for the elimination of the PRCs are affected by environmental variables in a manner similar to the uptake rates of pollutants, these elimination rate constants can be used to correct the sampling rates of pollutants in field page 6 deployments. There is also some evidence that the elimination rate constants of PRCs can be used to compensate for the impact of biofouling on uptake; however, more work is needed in this area [23,24,25]. Diffusional path Analyte net flow Water Water boundary (sampled medium) layer Diffusion Membrane (porous or nonporous) Receiving phase (acceptor so rbent) Figure! Functional principle of a passive sampling device, showing the concentration profile of a compound during diffusion and accumulation from bulk of the sampled medium to the sorbent (receiving phase) through a permeable (porous or non-porous) membrane. High affinity to the sorbent inside the sampler drives the diffusion of analyte molecules from the sampled medium into the sampler until the thermodynamic equilibrium is established, (adapted from Mills et al. [ 14]). The correction for the effect of environmental variables in samplers where the sequestration process depends on adsorption of the analyte presents one of the major challenges in the development of the technology. In many cases, uptake of analytes (polar organic compounds and metals) into these devices is WBL-controlled and thus sensitive to changes in flow turbulence. The PRC concept cannot, however, be generally used to correct calibration data for changes in field conditions because of the complex character of the desorption kinetics that may not be isotropic with the adsorption [26]. Mazzella et al. [27] and Budzinski et al. [28] have recently demonstrated isotropic exchange in certain exposure scenarios, but this concept still remains to be fully explored. In cases where PRC loss is not isotropic with uptake of target analytes, an alternative in situ calibration approach is to load PRCs into co-deployed sampling phases from which elimination is observed and which may subsequently be related to uptake. An in situ calibration technique, using PRC-loaded absorbent polydimethylsiloxane (PDMS) disks deployed alongside the Empore™ adsorbent disk samplers as a surrogate calibration phase, has been proposed by Shaw et al. [26] and shows promise for future applications. Alternatively a passive flow monitor based on dissolution gypsum has been developed which may predict the sampling rate in response to in situ flow conditions [29]. Differences in mass transfer in absorption- and adsorption-based samplers are illustrated in Figure 3. rma Compound class Uptake process Driving force Hydrophobic compounds Diffusion ^Absorption I Sampler/water partition coefficient Log KSyj " log Kc ow Independent of concentration Speciation Usually single form sampled Receiving phase selection Single receiving phase for a broad range of compounds Polar compounds Metals Diffusion ^ Adsorption I Adsorption distribution K5W- Kc"" Adsorption isotherms - dependent on concentration Many compounds dissociate multiple p/Ca - multiple species sampled A range of phases to optimise performance Figure 2. Differences in passive sampling in (left) absorption- and (right) adsorption- based samplers. The majority of emerging substances are polar or semi-hydrophobic. Thus, the use of adsorbent-based samplers presents the most suitable sampling approach for these compounds. Compound class Hydrophobic compounds Polar compounds Mass transfer Desorption kinetics In situ calibration Membrane control log Km < 3 WBL control log Kow > 3 Absorption and desorption kinetics are isotropic k - k e(iptake) e(release) Performance reference compounds WBL control for most analytes Desorption is often not isotropic with adsorption Sorption to multiple binding sites PRC use is not fully explored More research is needed Figure 3. Differences in mass transfer in (left) absorption- and (right) adsorption-based samplers r m a III. Applications in aquatic monitoring of emerging compounds A detailed description of sampler designs available for monitoring emerging polar organic compounds has recently been published by Sóderstróm et al. [15]. Applications of passive samplers for some important groups of emerging substances are discussed in the following section. Table 1 lists the most discussed emerging pollutants in the aquatic environment, the established or expected/potential performance of passive samplers of these compounds and availability of calibration data that enable calculation of TWA concentrations. 111.1. Algal toxins Algal toxins are a group of natural products which may occur in fresh, brackish and marine waters. However, possibly because of anthropogenic eutrophication and global climate changes, and subsequent blooms of potentially toxin-producing cyanobacteria, the incidence of contamination of water bodies with these compounds seems to have increased over recent years[30]. Algal toxins are structurally, functionally and phylogenetically diverse group of compounds with variable chemical and toxicological characteristics. These pollutants may cause serious health problems as documented by cases of human and animal intoxications as well as by the results of laboratory studies [30]. Based on the toxicity data, the World Health Organization (WHO) suggested the tolerable daily intake (TDI) value for microcystin-LR (a widespread hepatotoxin produced by cyanobacteria) is 0.04 ug kg"1 body weight, and corresponding safety guideline value 1.0 ug U1 is recommended for drinking waters. There are no obligatory guidelines for other cyanobacterial and algal toxins. However the presence of these compounds in water is highly undesirable and tools for proper monitoring are necessary. Owing to the quite high spatial and temporal variability of the occurrence and subsequent development of algal blooms, and hence potentially of co-occurring toxin production, passive samplers may prove to be a useful tool for monitoring of natural toxins. The first use of integrative passive sampling for algal toxins was described in the work of MacKenzie et al. They developed a passive sampler (SPATT bag) based on synthetic resin enclosed in porous sachets and used it for monitoring a group of marine toxins known as paralytic shellfish poisons [31]. The device was designed as an early warning of developing cyanobacterial blooms to protect consumers and prevent the harvesting of contaminated seafood products. This work was continued by other authors. Fux et al. evaluated various sorbents in the SPATT system [32]. Rundberget et al. redesigned the device and used it for monitoring of various natural toxins on the southern coast of Norway [33]. Shea et al. described the development of a monophasic device for monitoring of brevetoxins, highly toxic compounds produce during red tide events. Devices constructed of polydimethylsiloxane sheets were successfully used for integrative sampling [34]. Kohoutek et al. employed POCIS for the monitoring of microcystins in freshwater. The study was focused on evaluation of various configurations of the sampling device [35], and described calibration procedures and monitoring of the toxins under conditions of natural algal blooms. Concentrations of toxins obtained by passive sampling correlated well with the overall concentration of dissolved microcystins, demonstrating the suitability of passive sampling for the determination of TWA concentrations [35]. 111.2. Antifouling compounds - organotins Due to their bioaccumulation potential and toxicity, organo-metallic substances are considered as emerging pollutants of concern. In some cases organo-metallic compounds page 9 (e.g. some organic forms of tin) are more toxic than inorganic complexes or free forms of the parent metal. Passive sampling devices have been used to measure a number of organo-metallic species, including those of lead, mercury and tin. F0lsvik et al. [36,37] reported the use of SPMDs for monitoring organotin compounds using SPMDs. Both dibutyl- and tributyltin were accumulated by the devices, but no accumulation of monobutyltin was observed during several weeks of SPMD exposure in a Norwegian fjord. Using this method, it was possible to identify concentration gradients of organotin compounds at the sampling site. Later, a variant of the Chemcatcher® sampler was developed and calibrated for the measurement of the TWA concentration of organotin compounds. [38,39]. Using gas chromatography (GC) with either ICP-MS or flame photometric detection, favourable limits of quantification for the device (14-day deployment) for the different organotin compounds in water were in the range of 0.8-25 ng L"1, and once accumulated in the receiving phase the compounds were stable over prolonged periods [39]. 111 - 3- Brominated flame retardants Polybrominated diphenyl ethers (PBDEs) are widely used as flame retardants in products such as furniture, textiles, plastics, paints and electronic appliances. Due to their extreme hydrophobicity (log KqW values 4-10), these compounds are dissolved in the aqueous phase at extremely low (sub-ppb) concentrations. Nevertheless, because of their possible environmental risks due to their persistence and bioaccumulation, the inclusion of certain PBDE congeners in monitoring programmes is justified. Booij et al. [40] used SPMDs for sampling and in situ pre-concentration of PBDEs from water at several sampling stations in the Scheldt estuary and the North Sea along the Dutch coast. The application of integrative sampling enabled the back-calculation of extremely low concentrations (in range 0.1-5 pg L"1) of PBDE congeners in water from SPMD-accumulated amounts. Rayne and Ikonomou [41] employed SPMDs for sampling PBDEs in water in the Fraser River near Vancouver, Canada. The concentrations of PBDE found in SPMDs, their physicochemical properties, and their SPMD uptake parameters were used in an aquatic transport model to reconstruct the patterns of PBDE in pollution sources. The reconstructed patterns of accumulation in SPMDs closely approximated the composition of known technical mixtures of PBDEs. III.4. Endocrine disrupting compounds Over the last two decades the presence in the environment of endocrine disrupting compounds, such as those which mimic or block the action of endogenous hormones on steroid (oestrogen and androgen) receptors and subsequently alter the normal functioning of the endocrine system in wildlife and humans, has emerged as a major environmental issue [42,43]. Natural oestrogens (such as oestrone, E1, and 17-B oestradiol, E2) and synthetic oestrogens (e.g. 17-a-ethinyloestradiol, EE2, the active component of oral contraceptives) are very powerful endocrine disruptors. They derive mainly from excreta of humans and livestock [44]. Anthropogenic industrial chemicals such as nonylphenol (NP), bisphenol A (BPA) and phtalates are, however also known to influence the hormonal system of aquatic organisms. Wastewater treatment plants are important sources of pollution, since many endocrine disrupting compounds are not fully removed by the treatment processes. Several studies have demonstrated applicability of passive samplers for integrative sampling of these compounds during exposure periods up to several weeks [126,128,129,142]. For many compounds, calibration data that enable quantitative translation of amounts accumulated by the sampler into TWA concentrations are available (Table 1). page 10 III.5. Fluorinated surfactants Fluorinated surfactants (also referred to as poly- and perfluoroalkyl compounds, including perfluoroalkyl carboxylic acids, perfluoroalkyl sulfonates, fluorotelomeric acids, alcohols, etc.) have been used for decades to make stain repellents that are widely applied to fabrics, carpets and paper. They are still used in the manufacture of paints, adhesives, waxes, polishes, metal coatings, electronics and caulks. Due to concern over their persistence and global occurrence in humans and wildlife, two of these fluorinated surfactants, perfluorooctanoic acid (PFOA) and perfluorooctanesulfonate (PFOS) are within the family of compounds currently attracting the greatest attention as emerging pollutants.[45] It is difficult to identify the origin of pollution by fluorinated surfactants found in wastewater. Although no quantitative studies aimed at monitoring of these substances with passive sampling methods have been reported, Casey et al. [46] reported identification of these compounds in POCIS extracts at levels above associated controls. Recently, Gunther et al. described the application of a passive sampler based on active carbon adsorbent [47]. Further research in development of passive samplers suitable for monitoring of these compounds in water is needed. 111-6- Organosiloxanes Another important class of emerging pollutants is the organosiloxanes. These polymers comprise a backbone of alternating silicon-oxygen units with organic side chains attached to each silicon atom. Over the last 30 years organosiloxanes (silicones), both cyclic and linear forms, have been extensively used in a number of consumer products. These include for example anti-perspirants, and hair and skin care items. It has been estimated that in the USA adult women are exposed to up to 307 mg of organosiloxanes daily [48]. The most commonly used organosiloxane is decamethylcyclopentasiloxane (abbreviated to D5) although others such as octamethylcyclotetrasiloxane (D4) and their linear versions can be used in products [48]. These compounds have unusual physico-chemical properties combining high hydrophobicity (e.g. D5 has a log Kow of 6-8, depending on the literature reference used)) with a high Henry's Law constant and low water solubility [49]. Owing to these properties, most (c. 90%) of the organosiloxanes used in personal protection products are expected to be evaporated to the atmosphere during and after use, with the remainder being discharged into the wastewater. Several organosiloxanes are under assessment for classification as very persistent and very bioaccumulative in the environment. Hence there is an urgent need for monitoring levels of these compounds in different environmental compartments. Analytically, siloxanes are difficult to measure at trace levels as they are ubiquitous atmospheric environmental contaminants, they are contained in sample vial caps, septa, gas chromatographic columns and they give problems of cross-contamination by laboratory workers using personal care products containing these substances. The maintenance of good procedural blanks and rigorous quality assurance and quality control measures are needed to ensure confidence in any quantitative results. For these reasons reliable environmental monitoring data are sparse. Most analytical methods for both cyclic and linear siloxanes employ headspace gas chromatography/mass spectrometry techniques [49], although large volume direct injection methods using n-hexane have also proved to be useful [50]. Sparham et al. [49] have recently analysed D5 in the Rivers Great Ouse and Nene, UK (concentration range < 10-29 ng L"1) and in treated wastewater (concentration range 31-400 ng L"1). There are few other quantitative studies for D5 and the other organosiloxanes of environmental concern. page 11 Owing to the low concentrations of organosiloxanes found in the aquatic environment, the use of passive samplers in monitoring campaigns may offer the opportunity to pre-concentrate these compounds prior to instrumental analysis. To date, however, there is little experience of their use with this class of pollutants. Work in this area is being undertaken by researchers (Mills and Greenwood) at the University of Portsmouth, Portsmouth, UK. Preliminary findings show that pre-cleaned thin sheets of low density polyethylene (LDPE) membrane can be effectively used as passive samplers for D4 and D5. Work is currently being undertaken to identify PRCs that are suitable for use with the samplers and that are appropriate for the organosiloxanes of major environmental concern. Polydimethylsiloxane (PDMS) sheets cannot be used for this purpose because of background contamination with these smaller siloxane polymers. This makes it difficult to obtain good procedural blanks. Even with extensive washing it is still hard to remove all traces of D4 and D5 from these materials. Other polymers such as polyethylene terephthalate (PET), polyoxymethylene (POM), polytetrafluoroethylene (PTFE) and polycarbonate could potentially be used as either equilibrium or kinetic samplers for these compounds. Because the organosiloxanes are volatile, care must be taken during field deployments not to lose the sequestered analytes during retrieval and transport of samplers and in subsequent laboratory processing. Extensive QA and QC procedures must also be employed. Data from the Portsmouth group on the initial field use of the LDPE samplers for measuring this class of compounds are expected in 2011. III.7. Pharmaceuticals Concern over pharmaceutical residues (and personal care products) entering the aquatic environment has been growing since the mid-1990s. Both classes of compounds enter the environment largely as a result of human use, although some come from veterinary use. Several studies have reported the presence of a wide range of these chemicals at ng L"1 and sub ng L"1 concentrations in various water bodies. A complex mixture of chemicals is often present comprising the parent molecule, associated metabolites and environmental degradation products. Some of these substances may subsequently enter the food chain. The biological effects of pharmaceutical residues on aquatic organisms have been reviewed recently [51]. Effluent from wastewater treatment works is the most common source of pharmaceutical residues in streams and rivers. Some of these chemicals are resistant to treatment. Often the treatment process can break down conjugated drug metabolites to release the parent molecule back into the environment. A range of tertiary treatment processes (e.g. chlorination, ozonation and UV light) can be employed to reduce these levels, but these are expensive to operate continuously at the treatment plant. Pharmaceuticals have a wide range of physico-chemical properties and concentrations in the aquatic environment and this can make their measurement challenging. Many drugs are either weak acids or bases with pKa values in the range 4-10. The degree of ionisation will therefore differ in different water bodies that have pH values typically over the range 5.5-8.4 (i.e. from soft to hard fresh water and sea water). Likewise, these substances have a range of log Kow values, but most are considered polar compounds. In some cases the chirality of the drug molecule also needs to be considered. Most compounds of environmental concern can be analysed using LC/MS/MS instrumental methods after extraction and concentration. Typically a wide range of analytes can be separated and quantified at the trace level in a single analysis. page 12 There is a need to obtain reliable data on the fate of pharmaceuticals in the aquatic environment. These data can then be used to develop appropriate models and assist in the risk assessment process. As most discharges of these substances are sporadic and seasonal it is difficult to obtain such information using spot or grab sampling alone. Passive sampling therefore offers a number of opportunities in this area and this has been summarised by Mills et al. [14]. Recently, Soderstrom et al. [15] reviewed performance characteristics of samplers suitable for monitoring pharmaceuticals and other polar organic pollutants in the aquatic environment. Two types of passive sampler (polar version of the Chemcatcher and POCIS) have been used for measuring TWA concentrations of pharmaceuticals (and some personal care products). The devices use either an immobilised (Chemcatcher) or loose (POCIS) receiving phase. The Chemcatcher uses a 47 mm Empore™ disk, usually based on divinylbenzene copolymer chemistry, although ion-exchange (both anion and cation) receiving phases can be used for certain classes of analyte. The POCIS uses a commercially available solid-phase extraction adsorbent (typically c. 200 mg Oasis HLB) that is specially designed to sequester pharmaceuticals. The same diffusion-limiting membrane (polyethersulphone) is used in both devices. This membrane has a low surface energy and this can limit biofouling of its surface during field use. The uptake rates of the two devices for these more polar analytes are low (typically less than 1 L d"1) compared with the sampling of non-polar compounds by, for example, SPMDs. This can limit their usefulness in some applications, but - unlike non-polar compounds - polar compounds are usually present at higher concentrations, so that sampling rates below 1 L d"1 are not an obstacle. Although a number of laboratory and field studies have been carried out using the POCIS, there is an urgent need for reliable calibration data (Table 1). In many cases different calibration systems (e.g. flow through and static with renewal) [52] and different water turbulences and temperatures have been used and this increased the variation in the data obtained. Much of the field data reported is therefore either qualitative (presence or absence of a pollutant) or semi-quantitative (amount extracted from the receiving phase) rather than using uptake rates to calculate actual water concentrations (ng L"1). 111-3- Polar pesticides Use of pesticides can have unintended effects on the environment. Over 98% of sprayed insecticides and 95% of herbicides reach a destination other than their target species, including non-target species, air, water, bottom sediments, and food [53]. There are four major routes through which pesticides reach water, including: spray-drift outside of the intended application area, percolation, or leaching, through soil column, with water runoff or concomitant soil erosion, or through accidental or negligent releases [54]. There is an increased demand for environmental monitoring of pesticides because some of them are either already identified as priority substances under the Water Framework Directive (e.g. atrazine, simazine, diuron, isoproturon), or may become priority substances in the future or are relevant as river basin-specific pollutants in selected European regions [55]. An EU "Thematic Strategy on the Sustainable Use of Pesticides" calls for environmental monitoring to be done for other new pesticides in order to verify whether the concentrations in the aquatic environment are "safe" [56]. The first passive sampler reported for this chemical class was the POCIS [57,58]. Typically, for sampling of polar pesticides POCIS remains in the time-integrative mode for exposure periods of up to several weeks. This sampler has found application in integrative sampling of page 13 a wide range of polar pesticides and, for many of them, calibration data are available that enable quantitative translation of amounts accumulated by the sampler into TWA concentrations (Table 1). Polar pesticides are often released at high concentrations into streams and rivers in episodic events. These events usually last only a few hours and for these compounds to be detected by passive samplers, a device with a short response time is required. But passive sampling devices fitted with microporous membranes (e.g. polyethersulphone membrane in POCIS), although ideal for long-term monitoring [59], have a lag-phase of several hours which represents the time necessary for the analytes to diffuse through the membrane to reach the receiving phase [24]. In situations where detection of short pollution events in the monitored water body is required, a long lag-phase of the sampling device presents a potential disadvantage. Shaw and Mueller [60] suggested the use of a device fitted with an Empore™ disk bonded polymeric sorbent as receiving phase (without a diffusion limiting membrane) to reduce the response time and make the sampler more reactive to sudden pollution events [61]. The disadvantage of such devices is a fast equilibration of the sampling devices with the water phase, which restricts to a few days the time over which the samplers operate in time-integrative mode. Comparison of the performance of two different types of Empore™ disks as passive samplers showed that the styrene-divinylbenzene-reverse phase sulfonated (SDB-RPS) Empore™ disk had better performance as sorbent phase for very polar compounds compared to C18 [62]. III.9. Sunscreen and ultra-violet filters The analysis of sunscreens/organic ultra-violet (UV) filters in water has increased substantially in the last two years. Due to their use in a variety of personal care products, these compounds can enter the aquatic environment indirectly from showering, washing clothes, via wastewater treatment plants and also directly from recreational activities. In one of the first studies, Poiger et al. [63] detected four organic UV filters (80-950 ng SPMD"1) in SPMDs deployed at Lakes Zurich and Greifensee, Switzerland. SPMD-derived water concentrations were in the range of 1-10 ng U1 and corresponded well with those determined in spot samples of water. In a later study, Balmer et al. [64] investigated the occurrence of four important organic UV filter compounds in water, wastewater and fish from various Swiss lakes. Data from passive sampling using SPMDs supported the presence of these UV filters in lakes and rivers and suggested some potential for accumulation of these compounds in biota. Recently, Fent and Zenker et al. [65,66] demonstrated the applicability of the POCIS sampler for monitoring oestrogenic UV filters in surface water. They found that processing of POCIS samples with subsequent instrumental measurements was much less time consuming than processing of fish samples for environmental monitoring. Hydrophilic compounds like benzophenone-4 which do not accumulate in fish lipids could also easily be determined using the POCIS sampler. IV. Application in sediment monitoring Until recently sediment monitoring has relied on the determination of total or normalised contaminant concentrations. This approach, however, does not distinguish between freely dissolved and bound molecules and aims to assess the presence of chemicals rather than their activity and availability [67]. Since many laboratory and field studies have demonstrated that biological effects in benthic organisms are not generally related to the total concentration page 14 of contaminants in sediments, alternative and more representative measures of the bioavailable fraction of contaminants in sediments are required [68]. In addition, it has been shown that traditional empirical models tend to overestimate pore water concentrations. Application of passive sampling to sediment monitoring can be undertaken in situ with buried passive samplers or in batch experiments in the laboratory following grab sampling or coring (and sectioning). Passive samplers can be used to: • Determine freely dissolved contaminant concentrations in pore water; • Estimate sediment-pore water partition coefficients for contaminants of interest; • Measure contaminant desorption rates; • Estimate the fraction of contaminants available for desorption within a relatively short time scale or fraction effectively contributing to the partitioning with pore water and/or biota; • Measure surface water/pore water activity or fugacity ratios to estimate whether sediments act as a source or sink for contamination in the overlying water; • Measure the total contaminant amount in sediment that is available for release to the aqueous phase within a given time. The most commonly used passive sampling approach is based on the principle that the passive sampler is exposed to a sediment sample until a thermodynamic equilibrium between the two phases is established. According to partition theory, the concentration of a compound in the sampler is directly proportional (by the equilibrium partitioning coefficient between sampler and water) to the freely dissolved concentration of sampled compounds in pore water. Because this concentration is considered to be the driving force for the uptake of the contaminants by aquatic organisms, the bioavailability of a substance can be directly assessed using passive samplers. However, depending on sampler characteristics (e.g. surface area and volume), equilibrium may not be established for the most hydrophobic compounds during exposure and therefore performance reference compounds (such as used for surface water deployments) can be used to quantify sampler-pore water exchange kinetics and dissolved concentrations in such situations [67,69]. In all cases it is absolutely crucial to select an appropriate combination of sampler and sediment volumes in order to avoid significant depletion of the pore water phase. The true freely dissolved concentration of contaminant in pore water can be determined when the sampler's sorption capacity is kept well below that of the sediment sample to avoid depletion during the extraction [20,70,71]. When the sorption capacity of sampler to sediment is kept high, samplers can be used to measure the total contaminant amount in sediment that is available for release to the aqueous phase within a given time. This represents the fraction available to take part in partitioning with sediment organisms. The contaminants remaining in the sediment following such extraction can be considered effectively unavailable [72]. This fraction can also be estimated by repeated/successive extractions of the sediment with an adsorbent phase such as Tenax [73,74]. Such procedures also enable the quantification of contaminant desorption rates. The concentration difference between the in situ deployed samplers from the sediment and those from the overlying water give direct information on the fugacity difference between sediment and water, and on the direction of the contaminant diffusion at the sediment-water interface as well [20,71,75]. This enables identification of sites where remedial treatment of sediment may be appropriate. Other parameters, such as sedimentation rates and the spatial page 15 resolution of sediment sampling close to the sediment-water interface, are crucial for such measurements. For metals, the technique of diffusive gradients in thin films (DGT) provides an important contribution to understanding processes that metals undergo in sediments. DGT provide measurements in sediments that can be reported either as the mean flux of labile metal species to the device during the deployment time, or as the mean interfacial concentration in pore water. For a given device and deployment time, the interfacial concentration can be related directly to the effective concentration of labile metal [76]. This concentration represents the supply of metal to any sink, be it DGT or an organism that comes from both diffusion in solution and release from the solid phase. The primary use of DGT in sediments has been to investigate the distribution of solutes (metals) at high spatial resolution and to interpret the dynamics of the pollutant release from sediment [76]. Pore water concentration profiles with a fine resolution can be obtained by deploying DGT probes vertically in sediment and across the sediment-water interface. Modelling of metal accumulation in DGT with increasing exposure time can allow the estimation of sediment-water partition coefficients for metals of interest. It is crucial that the risk assessments of contaminants in sediment are as reliable as possible. It is widely accepted that it is the dissolved fraction of pollutants that is available for interaction with biological tissues and that can thereby cause bioaccumulation and/or biological effects. Several studies have shown that biota concentrations, calculated from partition coefficients based on classical equilibrium partition theory, are often orders of magnitude higher than the actual measured concentration in the sediment-dwelling organisms. But, using the freely dissolved concentration derived from passive samplers, the calculated concentrations in biota are in good agreement with the actual measured values [77].The methodology using passive sampling is leading to a much better understanding of how hydrophobic contaminants interact with sediment. This will allow a better estimation of (bio)availability, as can be validated through comparison with uptake by organisms. Data obtained with passive samplers can be used in risk calculations for sediment-bound contaminants with regard to any need for remedial measures for contaminated sediments and these studies would be an important input with regard to environmental quality standards for contaminants in water proposed in the EU Water Framework Directive. So far, the methodology of passive sampling in sediment has been tested and successfully validated in studies focused mainly on priority groups of contaminants that cause major environmental problems, such as polycyclic aromatic hydrocarbons or polychlorinated biphenyls. Nevertheless, this concept can also be successfully applied in risk-oriented monitoring of other groups of contaminants in sediments, including emerging substances. Further research is needed to develop novel solid phases with strong affinity to a broad range of compounds that may be found in sediments. These sampler materials should allow an easy extraction and analysis of accumulated substances [68]. V. Application in monitoring of contaminants in biota Knowledge of dissolved phase chemical concentrations is a critical part of understanding how aqueous exposure levels relate to the concentrations of residues measured in organisms in various trophic levels of aquatic ecosystems. The freely dissolved concentrations of pollutants represent the driving force for bioconcentration. Thus, passive samplers enable in situ determination of the bioaccumulative exposure of organisms at the lowest trophic level (filter feeders, e.g. mussels), in nearly all food chains, to hydrophobic organic compounds [78,79]. The estimation of bioaccumulation factors (BAFs) in certain species of concern (e.g. mussels) has also been demonstrated [79,80]. Moreover, since the contribution of dietary uptake for organic compounds with log Kows < 5.5 is generally very small, organism exposure assessment can be potentially extended to higher trophic levels for less hydrophobic compounds. Bayen et al. [81] recently reviewed kinetic studies of the uptake of neutral non-polar chemicals from the aqueous phase into organisms (fish, bivalve, crustacean, insect, worm, algae, and protozoan) and passive samplers. They demonstrated that passive samplers are biomimetic when diffusional partitioning processes mediate concentrations in organisms of concern (i.e., when residue accumulation in organism tissues follows equilibrium partitioning theory). Huckins et al. [78] discussed in detail accumulation into the SPMD sampler compared with that into biomonitoring organisms. The large number of variables, which potentially affects the accumulation of hydrophobic organic compounds in biota, suggests that it is unrealistic to expect any single passive sampler to be biomimetic of all biomonitoring organisms. Also, it is similarly unrealistic to expect that one or two species of biota mimic bioaccumulation in all organisms of concern. Variables affecting pollutant accumulation in passive samplers are limited to the sampler properties, physicochemical properties of the sampled chemical, exposure site conditions (e.g. temperature and turbulence, and exposure scenario factors such as the constancy of chemical concentrations during the exposure period). The ability to generate chemical-specific calibration data and then adjust these values to site-specific conditions (e.g. using PRCs) [22] means that analyte concentrations obtained using passive samplers are directly comparable across sampling sites. There are some fundamental similarities in the characteristics and processes affecting the accumulation in biota and passive samplers, especially for hydrophobic organic compounds. Diffusion of non-polar compounds through non-porous polymers used in passive sampler construction mimics the diffusion across bio-membranes. Also, partitioning between the polymers, organism lipids and the exposure water is similar and can be described by the equilibrium partitioning theory. Finally, the surface-to-volume ratio appears to be a critical parameter for the uptake rate of the more hydrophobic chemicals, both for samplers and organisms. Monitoring by passive samplers has some advantages over the use of biota. Passive samplers can be prepared to a standardised quality characterised by low initial concentration of contaminants, uniform composition, diffusion and sorption properties. In contrast, test organisms often contain background contamination levels and they are naturally variable in composition. As a result, variability of chemical analysis of biota or sediment is in most cases higher than that associated with analysis of passive samplers. Moreover, the simple polymeric matrix composition of passive samplers provides sample extracts that contain much less matrix interference in comparison with extracts from biota and sediment. Samplers can be applied in almost any environment with a broad range of water quality properties and even in very polluted sites where biomonitoring organisms may not survive. In contrast, biomonitoring organisms can be applied only within a certain geographical range and they do not tolerate extreme exposure conditions (e.g. temperature, pH, pollution, and salinity). The uptake process of pollutants in passive samplers is simple (by diffusion and sorption), whereas it is more complex in organisms since it includes bioconcentration, bioaccumulation page 17 and metabolism. The complexity of these processes is increased by behavioural, physiological and anatomical characteristics of biomonitoring organisms. The uptake capacity of polar organic compounds in biomonitoring organisms is in most cases low. Also, these compounds reach steady state within a short period of time, so that biological sampling of polar organic compounds has a very limited applicability [82]. In comparison with biomonitoring organisms, passive samplers demonstrate better retention of contaminants that are absorbed during peak exposure events. The amount of chemicals accumulated in passive samplers in most cases reflects the dissolved, readily bioavailable, concentration in sampled water, whereas the estimation of contaminant bioavailability from total amount found in an organism body may be difficult, owing to the presence of a non-incorporated portion of the pollutant in its intestines. For metals, the DGT technique measures directly the variables needed to assess water quality. Uptake of trace metals across living membranes is determined by free ion concentrations when membrane transport is slow and by the total concentration of labile species when membrane transport is fast. Deployment of twin DGT devices with different diffusive gel layers can provide an in situ measurement of both labile inorganic and total labile species. Free ion activities can be calculated from labile (free and/or kinetically-labile species in solution) inorganic concentrations. VI. Application in ecotoxicitv assessment Ecotoxicity assessments are an invaluable tool for the evaluation of water quality and in some countries ecotoxicity assessments are compulsory, for example, with direct toxicity assessments of effluents released to the environment [83]. One of the main advantages of ecotoxicity assessments is that they give an integrated picture of the total toxic burden of the complex mix of chemicals that are present in environmental samples. It is often the case that toxic substances cannot be identified and chemical monitoring methods cannot be targeted, but ecotoxicity assessments can still measure the effect of these unknowns in environmental samples. Such samples can be tested, either at the level of organisms (e.g. daphnids or fish embryos [83],[84]), at the level of cells (e.g. fish cell lines) [84] or at the sub-cellular level (e.g. specific binding of chemicals to receptors using reporter gene assays). An example of such a reporter assay comes from research on endocrine disruptors, where cells have been modified to express oestrogen receptors ([85],[86]). The binding of oestrogens - or oestrogen-like compounds - to the receptors leads to the production of an enzyme which in turn induces a colour change in the medium (or light emission) that can be quantified easily. Commonly, bioassays are applied to whole water samples, extracts of water samples or extracts of organism tissues. Applying the same bioassays to extracts of passive samplers is straightforward and an increasing number of studies have explored this. VI.1. Passive samplers as mimics for bioconcentration Combining bioassays with (grab) water samples has the same limitations (or advantages) as compared to combining chemical analyses with water samples. Grab samples give an accurate picture of the total concentration only at a certain point in time. Grab samples again provide data on toxic effects that relate only to the time of sampling. As an alternative, combining ecotoxicity assessments with monitoring of chemicals in biota, for example by analysing extracts of aquatic organisms, is certainly feasible, and produces more representative results than analysing grab samples, but has the same limitations associated page 18 with monitoring of contaminants in biota as discussed in the previous section (i.e. section V. ). Combining bioassays with passive sampling circumvents the limitations that are associated with grab samples and chemical monitoring in biota. Furthermore, a passive sampler mimics bioconcentration of freely dissolved chemicals over cell walls, membranes or a filter feeding apparatus or gills. Thus, testing passive sampler extracts in bioassays has a high relevance as this reflects exposure scenarios in the aquatic environment. VI.2. Which passive sampler suits which bioassay? Numerous biological assays have already been used successfully in combination with passive samplers. Many studies deal with quantification of environmental oestrogens with reporter gene assays in extracts from SPMDs ([87,88]), POCIS ([89],[90],[91],[92],[93],[94]) and Chemcatchers ([95]). An assay that covers compounds such as PAHs and dioxin-like compounds, the EROD assay, has been used with extracts from SPMDs ([87]) and in combination with the Toximeter ([96]). Several studies describe the use of Chemcatchers and POCIS to measure photosystem II (PS-II) inhibitors ([97],[98],[99],[100]). Microtox, a bacterial whole cell assay that is used to measure baseline toxicity, has also been used in combination with POCIS ([94],[100]), Chemcatcher ([98]) and SPMD ([101]) extracts. Muller et al. tested Chemcatchers extracts in the umuC assay, which is used to assess genotoxic effects in response to the presence of DNA-damaging chemicals within the sampled mixture. [98]. Mutagenicity has been assessed in extracts from SPMDs by Rastall et al. [87]. Shaw et al. used Chemcatchers in combination with two invertebrate bioassays, coral larval settlement and sea urchin larval development, in addition to bacterial luminescence and microalgal photosynthesis [102]. The above listing is certainly not complete but illustrates that the range of bioassays is very diverse, spans across organisational levels - from gene expression to whole organisms -and covers multiple modes of action. In addition, both relatively hydrophobic absorptive passive samplers and adsorptive samplers used to sample more polar chemicals have been used in combination with these multiple end-point bioassays. Although various combinations of passive sampler and bioassays have been explored, it is difficult to list fixed combinations for passive samplers and biotests. The reason for this is that the range of compounds that is targeted by bioassays is often very diverse and no single sampler can adequately target a set of chemicals with diverse physicochemical properties. This issue can be illustrated for an algal test that is used to quantify the effects of herbicides such as diuron and atrazine that inhibit PS-II. Log Kow values for PS-II inhibitors range from below 1 (e.g. metamitron) to 4 (dipropetryn). Metabolites of these compounds can also be active PS-II inhibitors and may further extend the log Kow range of possible PS-II inhibitors. Log Kow ranges for compound classes targeted by other bioassays can be even larger; e.g. PCBs with log Kow values up to 7 are oestrogenic whereas benzotriazole, with a log Kow of 1.4, is anti-oestrogenic. As passive samplers usually target a range of log Kow values spanning 2 to 3 orders of magnitude [87], it is clear that not all compounds that are active in a bioassay will be sampled in a similar, integrative fashion. Some toxic compounds may reach equilibrium well before others. Thus, even when the concentration ratios of various toxicants in the environment are constant, different integrative sampling windows of individual compounds will cause their concentration ratios in a passive sampler to vary over the deployment time of the sampler. In addition, different compounds with the same mode of action may have very different diffusion coefficients within a given sampler (or over a membrane that envelops the sampling phase), and thus behave differently in response to changing environmental conditions. page 19 Although no single passive sampler covers all compounds that act on a certain organism or have a certain mode of action, this does not negate the rationale of combining passive samplers with ecotoxicity assessments. The use of bioassays is a more holistic approach to assessing the risk associated with exposure, since the technique provides a functional integrative assessment of mixture toxicity for chemicals accumulated by passive samplers to levels sufficient to induce a biological response. So, by combining passive sampling with bioassays it is possible to avoid intensive chemical analyses. However, when using a specific bioassay in a sampling campaign, one has to attempt to identify the main possible toxicants that may be present at the sampling locations and select a sampler that best covers the log Kows of those toxicants. VI.3. The link between biological and chemical analysis It is common to express the effect of water samples in ecotoxicity tests as a dilution factor, i.e. at what dilution the sample still leads to a certain effect level in the bioassay [83]. The same approach can be used for a passive sampler and one can express the toxic effect in terms of a certain portion of a sampler extract [89]. An alternative approach was developed by Koči et al., a toxicity measure corrected for the volume sampled by a passive sampler (vtox [103]). Although these approaches are clearly informative, and one can classify more or less polluted sites and derive water quality criteria on this basis, it is difficult to compare chemical and biological analyses directly. Another system to evaluate effects in bioassays is the toxic equivalent (TEQ) concept. It was first established for effects caused by dioxins and PCBs on the arylhydrocarbon receptor [104]. Subsequently, the concept has been applied to oestrogenic activity, phytotoxicity and other types of toxicity. In essence the TEQ concept revolves around comparing the dose response curve induced by a sample to the dose response induced by a reference compound (see [105]). The biological response to the sample can then be expressed in terms of an amount or concentration of the reference compound. This approach can then be complemented by testing many individual compounds in the bioassay to establish their dose-response curves; from these one can derive their potencies relative to the reference. When a set of compounds has been quantified in an environmental sample by means of chemical analysis, concentrations of these compounds can be multiplied by the potencies of the compounds and added together (assuming concentration addition applies) [106]. The sum of the individual chemicals signifies the toxicity based on chemical analysis and the minimum expected response of the environmental sample in the biological test. This approach is well established and many legal TEQ limits are in place for dioxin-like compounds (e.g. the EU limit for fish = 4 pg WHO-PCDD/F-TEQ /g fresh weight) [107]. Being able to relate results from a bioassay directly to those obtained by chemical analyses has the main advantage that one can assess whether most of the toxicity has been accounted for by the chemical analyses, or whether major toxicants have been missed. In passive sampling, linking biological analyses to chemical analyses has been done in several studies ([90],[92],[93],[97],[99]). Attention has focused on oestrogens, PAHs and herbicides and recently also on baseline toxicity ([100]). VI.4. Identification of toxic compounds in passive samplers: effect-directed analysis Effect-directed analysis (EDA) is another area where ecotoxicity assessments can be used [108]. In EDA, an environmental sample is fractionated chromatographically and next, the page 20 various fractions are tested individually for toxic effects. Once toxicity has been detected in a fraction, this fraction can be analysed chemically to identify possible toxicants. This is a very powerful method for identifying major toxicants in a complex environmental sample, particularly when the bioassay data are expressed as TEQ to allow for direct comparisons between data from chemical and biological analyses. The EDA approach has been applied frequently in sediments [68,109]. As yet, only one example comes from passive sampling. Rastall et al. [110] fractionated SPMD extracts and tested these for activity in a reporter gene assay for oestrogen receptor agonists. They found that oestrogens sampled by SPMDs cover a wide log K0vv range, but individual oestrogens could not be identified. This area is one where much progress can be made. In a recent field study where POCIS were deployed for five weeks in treated sewage effluents, a toxic spill occurred at one of 21 sites. The toxic spill caused a fish kill in the receiving river, and the POCIS from this site recorded the highest baseline toxicity in a bacterial test [100]. Using chemical analyses of water samples taken directly following the fish kill, the toxicant(s) causing fish mortality could not be identified (A. Stockli, personal communication). Although EDA was not attempted with these POCIS, it clearly points to an effective use for passive samplers as monitors for such peak toxic events. VI.5. How does the bioassay response in passive sampler extracts relate to sampler exposure conditions? The rate at which a compound is sampled by a passive sampler depends on the properties of the compound, the properties of the sampler and the environmental conditions at the deployment site. For individual chemicals it is fairly straightforward to establish relationships between compound properties, environmental conditions and sampling rates [111]. In contrast, the response in bioassays is the sum of the effects caused by contributions from at best a few (for highly specific endpoints) to a large number of individual compounds. As the composition of the mixtures and the relative abundance of the toxicants can vary widely across sites, and over time, this poses certain limitations on how bioassay results can be interpreted with respect to varying environmental conditions. Interpretation can be even harder when a sampler includes a membrane. For example, it was shown that more polar compounds (log K0vv < 2) move more rapidly over a polyethersulphone membrane than less polar compounds (log K0w > 3) into the SDB sampler phase in the Chemcatcher [99]. For short sampling windows, less polar compounds may be under-represented in the mixture of toxicants which will skew results. Thus, when combining bioassays and passive sampling one has to appreciate the uncertainties caused by the fact that the suites of target chemicals cover a wide range of physicochemical properties. As a result, different mixtures of chemicals with the same mode of toxic action will respond differently to varying exposure conditions. VII. Quality assurance, quality control and normation If passive sampling is to become accepted and used in a regulatory context for monitoring water quality across Europe, then there is a need for the development of improved validation methods and setting up of the appropriate quality control and quality assurance schemes for the technology. This would involve a set of activities (e.g. development of standard certified reference materials, setting-up of round robin exercises and the publication of standard methods) as those have been established for the validation of analytical techniques for the measurement of various analytes of importance in different environmental matrices. There is also a need for associated accreditation schemes laboratories involved in passive sampler calibration measurements in the lab and those using passive samplers in the field. The implementation of the above is not straightforward. For laboratory calibrations of the samplers, there is a need for large volumes of reference materials to be available. For field trials it may be possible to use reference sites that are well characterised and stable in chemical composition. An attempt to compare various water monitoring methods that could potentially be used in support of the Water Framework Directive was undertaken as part of a European Union-funded project [112] and the results of this activity have been summarized [113]. A number of field trials were undertaken in different water bodies across Europe and the results from these multiple comparisons indicated the potential utility of this approach. But these activities are expensive to develop and organize and therefore regulators and other end-users need to be convinced of the value of these alternative monitoring techniques so that they can support the provision of EU funding to enable this important research in support of policy and associated legislation. Several interlaboratory field trials, where a range of passive sampling technologies will be evaluated at European riverine sites, are being set up in 2010. The first is being facilitated within the framework of AQUAREF (the organisation coordinating French laboratories involved in water monitoring) [114]. A call was made in early 2010 for the participation of research groups across Europe who are involved in either developing or using passive sampling technology. Several field sites were selected and include both surface water and a marine lagoon in France. This trial focuses on the monitoring of pesticides, PAHs and metals. The second exercise is being proposed by the NORMAN network, where the focus of this exercise will be on the application of passive sampling for monitoring pollutants of emerging concern. Further, an interlaboratory proficiency testing scheme aimed at the chemical analysis of a range of hydrophobic organic compounds and metals in two commercially available passive samplers has been launched recently in the Czech Republic. [115] The results of these exercises will be of value in demonstrating the future utility of the technology and will be helpful in the design of similar activities in the future. Progress has been made on the normation of passive sampling methods. One of the deliverables of the European Union-funded project STAMPS [116] was the development of a British Standards Institution Publicly Available Specification [117]. This specification provides guidance for end-users on the preparation, deployment and associated quality assurance requirements for the use of passive samplers in surface waters. The specification is currently under consideration for development of a CEN/ISO standard [118]. VIII. Application of passive samplers in regulatory monitoring "Emerging pollutants" can be defined as pollutants that are currently not included in routine monitoring programmes at the European level and which may be candidates for future regulation, depending on research on their (eco)toxicity, potential health effects and public perception and on monitoring data regarding their occurrence in the various environmental compartments. In many cases knowledge of their ambient and background levels in water, sediments and biota is still limited and even less is known of the long-term ecotoxicological effects of these emerging contaminants. At such an early stage, it is difficult if not impossible to derive appropriate environmental quality standards (EQS) for these chemicals without the use of significant safety factors. Therefore compliance testing against EQS values is not page 22 often undertaken for these substances. Most monitoring programmes that include emerging pollutants are in general screening studies [119,120] aimed at obtaining additional information on the occurrence of these compounds in various aquatic environmental matrices, where they are likely to accumulate. Passive sampling may be favoured over matrices such as sediments and biota for such screening. It draws advantage from a simple matrix composition that enables simplified sample extraction, cleanup and the subsequent instrumental analysis. Moreover, field exposure of passive samplers in various matrices such as surface waters, wastewaters and sediment can be standardised. In addition, the use of, for example absorption-based samplers for the screening of non-ionic hydrophobic substances in water and sediments results in limits of detection which are generally substantially lower than those that can be achieved through bottle sampling [121]. Another factor to be taken into account in screening studies is the possible (mostly unknown) temporal variability in the concentration of emerging pollutants in water. Continuous monitoring possible with passive samplers can help in reducing the uncertainty associated with sampling when concentrations vary in time. For example, variable concentrations may be observed for emerging contaminants that are emitted in the urban environment and that can ultimately be released from sources such as landfill and wastewater effluents. This is, however, also valid for compliance monitoring of more conventional pollutants for which EQS have been derived and are in use (e.g. for the EU WFD). Despite the measurement of a different fraction of contaminants in water, passive samplers can be used to support data collected by infrequent bottle sampling [122,123] or through monitoring in biota. This allows continuous monitoring in conditions where this would not be feasible and improves the representativeness of the sampling. The integrative nature of passive sampling combined with extremely low limits of detection for non-ionic hydrophobic organic contaminants may represent the only acceptable way to monitor some of these substances in surface waters. Since passive sampling is based on the measurement of dissolved phase pollutants, further comparison with EQS based on "whole water" concentration values may require additional information to account for the fraction of contaminants associated with other phases such as dissolved organic carbon and suspended particulate matter. In the long term, such a strategy requires the development of water body-specific knowledge of contaminant speciation and partitioning. The additional information on non-dissolved fractions of compounds can be obtained in parallel representative measurements of these compounds in suspended particular matter or bottom sediments. The sum of the representative (e.g. TWA) contaminant concentration in the dissolved phase (provided by passive samplers) and that bound to colloids and particles (provided by sampling of suspended particulate matter) will provide the measure of total concentration that can be applied in compliance checking with EQS. Moving towards an implementation of passive sampling for regulatory monitoring of emerging substances will require the identification of suitable material for accumulation of target compounds and an accurate characterisation and calibration of the devices. In this regulatory context, passive samplers may be applied to the monitoring of surface waters in both populated and remote areas and other aqueous matrices such as wastewaters and other effluents. Samplers can be deployed simultaneously in different media in order to detect gradients in chemical activity/concentration and understand fluxes of these emerging substances. page 23 IX. Future trends There are several future trends for the development of passive sampling techniques for emerging substances. Novel materials will need to be tested as selective receiving phases (e.g. ionic liquids, molecularly imprinted polymers, and immuno-adsorbents), together with membrane materials that permit the selective diffusion of chemicals. Novel synthetic absorbent polymer materials with high retention capacity of polar organic compounds may enable the replacement of currently used adsorption-based samplers for which data conversion into aqueous concentrations is often difficult. A major challenge in the future development of the technology is the calibration of devices to enable the quantification of the concentration of emerging substances present in water. In comparison with devices designed for sampling hydrophobic organic compounds, sampling of most emerging substances is more complex. In addition to the common factors (temperature, water turbulence and biofouling), other factors (e.g. salinity, DOC level, pH, and the presence of complex mixtures of contaminants) may significantly affect the performance of samplers of emerging substances and these need to be evaluated. There are several routes to reduce uncertainty associated with the passive sampler data. These include quantitative assessment, reduction or control of the known factors which impact on sampler performance. For samplers where analytes are accumulated in the receiving phase by absorption mechanisms, PRCs can be successfully employed for improving the accuracy of the measurement of TWA concentrations of contaminants in the field. However, further research is needed to understand accumulation kinetics in samplers fitted with adsorbent-type receiving phases. Mechanical control of constant water flow conditions around the receiving phase in the field enables sampling rates of WBL-controlled samplers that are unaffected by turbulence [124]. Such devices require an in situ use of rotors or pumps that force water motion around the sampling devices. Thus, they cannot be classified as true "passive samplers". However, miniaturised devices that require only a low energy supply (e.g. batteries or solar cells) for the operation of pumps can be deployed in the same way as passive samplers. Miniaturised devices present a further trend in technology development. Small samplers are usually less expensive to use because of the lower costs of materials needed for their preparation and the reduced equipment requirements for their deployment. Lower volumes of solvents and reagents are consumed during their subsequent processing. Small samplers also offer the advantage of easy transportation to and from the sampling site. As miniaturised devices should not deplete the bulk matrix, they can be used in situations where space, volume and the flow of water are limited; for example, in groundwater boreholes. The ability to predict kinetic and thermodynamic uptake parameters for passive samplers using quantitative structure property relationship (QSPR) models describing interactions of sampled compounds with materials used in the construction of devices is also important. This may help to reduce the amount of required laboratory-based calibration experiments. Development of biomimetic devices capable of simulating the accumulation of toxic chemicals in tissues of aquatic organisms will enable a reduction in the use of chemical monitoring in biota in routine monitoring programmes. It will also decrease the uncertainty page 24 associated with the data obtained, as this is based on highly variable samples of biological material. The combination of the deployment of passive samplers followed by the biological testing of sampler extracts with the aim of detecting and subsequently identifying toxicologically relevant compounds offers much potential. This approach can provide information concerning the relative toxicological significance of waterborne contaminants and hence help to improve risk assessments for different water bodies. Finally, further development of QA/QC, method validation schemes, and standards for the use of passive sampling devices is urgently needed. Successful demonstration of the performance of passive samplers alongside conventional sampling schemes as well as inter-laboratory studies that demonstrate reproducibility of data produced by different designs of passive samplers will help to facilitate the acceptance of passive sampling in routine regulatory monitoring programmes in the future. Table 1. List of most discussed emerging pollutants in the aquatic environment and the established or expected/potential performance of passive samplers of these compounds. Category / class Sub-class Individual substances Potential of non-polar samplers3 Potential of polar samplers" Stage of development0 Sampler calibration datad _ 17) + - P Category / class Sub-class Individual substances Medium chain PCAs (mPCAs, C14-17) Technical PCA products Potential of non-polar samplers3 + + Potential of polar samplers" Stage of development0 P P Sampler calibration datad Fragrances Fragrances Acetylcedrene Benzylacetate Benzylsalicylate Camphor g-Methylionone Hexylcinnamaldehyde Isoborneol Isobornylacetate Isoquinoline d-Limonene Methyldihydrojasmonate Methylsalicylate p-t-Bucinal Terpineol + + + + + + + + + + + + + + P P P P P P P P P P P d P P Nitro musks Muskketone Muskxylene Musk ambrette + + + - d d P Macrocyclic musks Polycyclic musks AHTN (Tonalide) Galaxolide OTNE AHDI (Phantolide) ADBI (Celestolide) ATM (Traseolide) + + + + + + d d d d d d Gasoline additives Dialkyl ethers Methyl-tert-butyl ether (MTBE) Indus-trial chemi-cals Industrial chemicals TCEP Triphenyl phosphine oxide Perfluoro-alkylated substances Perfluoroalkylat ed substances Perfluorooctane sulfonate (PFOS) Perfluorooctanoic acid (PFOA) + + P P Pers onal care prod ucts Sun-screen agents 4-Methylbenzylidene camphor + + d Potential Potential Stage of development0 Sampler calibration datad Category / class Sub-class Individual substances of non-polar of polar samplers3 samplers" Benzophenone - + d Benzophenone-3 - + d Butyl methoxydibenzoyl- P methane Ethylhexyl + + methoxycinnamate Eusolex Homosalate N,N-Diethyltoluamide - + d Octocrylene Oxybenzone Insect N,N-diethyl-m-toluamide + d (DEET) repellents Bayrepel Octamethylcyclotetrasilo + - P xane (D4) Decamethylcyclopentasil + - P oxane (D5) Dodecamethylcyclohexa + - P siloxane (D6) Carriers Hexamethyldisiloxane (HM or HMDS) Octamethyltrisiloxane (MDM) Decamethyltetrasiloxane (MD2M) Dodecamethylpentasilox ane (MD3M) + + + + - P P P P Methyl-paraben - + P Parabens Ethyl-paraben + P (hydroxybenzoi c acid esters) Propyl-paraben Isobutyl-paraben - + + P P Acetochlor - + d [26,131,132] Amitrole - + Bentazone - + d Bromofos-ethyl - + Carbazole - + Carbendazim - + d [99] <2 ■o Polar pesticides Carboxin - + o and their Glyphosate - + c/3 a> degradation Chloridazon _ + d Q_ products Clopyralid Chlorpropham - + + Chlorpyrifos - + d [130] Chlorotoluron - + d 2,4 D - + d [59] Dicamba - + P [59] Category / class Sub-class Individual substances Desethylterbutylazine Desmedipham Desmetryn Potential of non-polar samplers3 Potential of polar samplers" + + + Stage of development0 d Sampler calibration datad Diazinon + + d [99] Diclobenil - + d-Dichlorvos + + d [57] Dinoterb - + Endosulfan-sulfate + + d [133] Ethoprophos - + Ethofumesate - + d Fluroxypyr - + Heptenophos - + lodofenphos - + Imidacloprid - + MCPA - + d [59] MCPB - + P MCPP (Mecoprop) - + P [99] Metalaxyl - + d [27] Methomyl - + Metamitron - + d Mevinphos - + Phenmedipham - + Prometryn + + P Prometon - + d Secbumeton - + Terbutryn + + P [99] Terbutylazine - + d [134,99] Thiabendazyl - + d Triadimefon - + Other pesticides Cypermethrin Deltamethrin Permethrin + + + - d d d [135] Sulfonyl urea New pesticides Degradation products of pesticides Desisopropylatrazine Desethylatrazine - + + d d [27] [27,99] ■ <" o 1> Biocides Triclosan + + d [129,136] ml Methyltriclosan + + d [137] äceutical s Analgesic Acetaminophen (paracetamol) Codeine + + d P [129,138,139 ] Pharm* Hydrocodone - + Anorexic Fenfluramine - + P Category / class Sub-class Individual substances Potential of non-polar samplers3 Potential of polar samplers" Stage of development0 Sampler calibration datad Anthelmintic Ivermectin + P Amoxicillin - + P Ampicillin - + P Azithromycin - + d [128,140] Chloramphenicol - + P Chlortetracycline - + P Ciprofloxacin - + P Clarithromycin - + d [95,141] Cloxacillin - + P Danofloxacin - + P Dicloxacillin - + P Doxycycline (anhydrous) - + P Doxycycline (monohydrate) - + P Enoxacin - + P Enrofloxacin - + P Erythromycin - + d [141] Flumequine - + P Antibacterial Josamycin Lincomycin Methicillin Minocycline Norfloxacin Novobiocin Ofloxacin Oleandomycin Oxacillin Oxytetracycline Penicillin G Penicillin V - + + + + + + + + + + + + P P P P P P P P P d P P Roxithromycin - + d [141] Spiramycin - + P Sulfadiazine - + d Sulfamerazine - + d [128] Sulfamethazine - + d [141] Sulfamethoxazole - + d [99,129] Sulfapyridine - + d [129,138, 141] [95,129, 138,141] Anticonvulsant Carbamazepine - + d Primidone - + Antidepressant Tetracycline Tiamulin - + + d page 31 Category / class Sub-class Individual substances Citalopram Escitalopram Sertraline Fluoxetine Fluvoxamine Paroxetine Potential of non-polar samplers3 Potential of polar samplers" + + + + + + Stage of development0 d d d Sampler calibration datad [129] [129] [129,141, 140] [129] Antidiabetic Glyburide (glibenclamid; glybenzcyclamide) Metformin + + P Antiemetic Diphenhydramine - + d Antihistaminic Loratadine - + Antihypertensive Nadolol Verapamil + + Antiinflammatory Aceclofenac Acemetacin Acetylsalicylic acid (aspirin) Alclofenac Diclofenac Fenoprofen Fenoprofen calcium salt dihydrate Ibuprofen Indomethacin Ketoprofen Meclofenamic acid Mefenamic acid Naproxen Phenylbutazone Phenazone Propyphenazone Tolfenamic acid - + + + + + + + + + + + + + + + + + d d d d d d d [138] [99,138, 141] [141] [129,138] [138,141] [129,138, 141] Antimicrobial agent Clotrimazole - + Antineoplastic Cyclophosphamide Cyclophosphamide (anhydrous form) Daunorubicin Doxorubicin Epirubicin - + + + + + P Category / class Sub-class Individual substances Potential of non-polar samplers3 Potential of polar samplers" Stage of development0 Sampler calibration datad Fluorouracil Ifosfamide - + + P Antiulcerative Famotidine Lansoprazole Omeprazole Ranitidine - + + + + d P [141,140] Anxiolytic Alprazolam Bromazepam Diazepam Lorazepam Medazepam Meprobamate Nordiazepam Oxazepam Temazepam - + + + + + + + + + d d d P P P P P d [138] [138] [141] Acebutolol - + P Beta-Blockers Atenolol Betaxolol Bisoprolol Carazolol Metoprolol Oxprenolol Pindolol Propranolol Sotalol Timolol - + + + + + + + + + + d P P P P P P d P P [129,141] [129] [129,141] [129] Blood viscosity agents Pentoxifylline + Bronchodilators Albuterol Albuterol sulfate Clenbuterol Fenoterol Salbutamol Terbutaline + + + + + + d d d [138] [138] [138] Diuretic Caffeine Furosemide Hydrochlorothiazide - + + + d P d [128, 129,138] [141] Lipid regulators Bezafibrate Clofibric acid Etofibrate - + + + d [141] Category / class Sub-class Individual substances Fenofibrate Fenofibric acid Potential of non-polar samplers3 Potential of polar samplers" + + Stage of development0 Sampler calibration datad Gemfibrozil - + d [129,138,141 i Lovastatin - + j Mevastatin - + Pravastatin - + Simvastatin - + P Acecarbromal - + Allobarbital - + Amobarbital - + Sedatives, hypnotics Butalbital Hexobarbital Pentobarbital Aprobarbital Secobarbital sodium - + + + + + 17-alpha-Oestradiol 17-alpha-Ethinyloestradiol 17-beta-Oestradiol Beta-sitosterol - + + + + d d d d [25,128,129, 142] [25,128] [25,128,129, ,142] Cholesterol - + d Steroids and hormones Diethylstilbestrol Oestriol - + + P d [142] Oestrone - + d [25,128,129] Oestrone 3-sulphate - + P Prednisolone - + P Dexamethasone - + P Bethametasone - + P Mestranol - + d Amitryptiline - + d [138] Psychiatric drugs Doxepine Imapramine Nordiazepam Zolpidem - + + + + d d [138] [138] Diatrizoate - + X-ray contrast media lohexol lomeprol lopamidol lopromide - + + + + Trace 1 1 i\j ICIIO Trace metals and their compounds Tetramethyllead Tetraethyllead + + - Category / class Sub-class Individual substances 4-Methyl-1H-benzotriazole Potential of non-polar samplers3 Potential of polar samplers" + Stage of development0 P Sampler calibration datad Benzotriazoles 5-Methyl-1H-benzotriazole 5,6-Dimethyl-1-H-benzotriazole - + + d P Tolyltriazoles Tolyltriazole (TT) 4-/5-Tolyltriazole (TTri) Wood "eserva-tives Phenols para-Cresol + d Q. Cocaine - + P Codeine - + d [141] Dihydrocodeine - + P Drugs of abuse Heroin Hydrocodone Morphine - + + + P P P i_ CD r— Oxycodone - + P -1—» o Benzothiazole - + d Benzothiazoles (BT) 2-Mercapto-benzothiazole - + d Benzothiazole sulfonic acid - + P Nicotine Cotinine - + d [128] metabolite The following considerations apply. potential of non-polar samplers: (e.g. SPMD, LDPE, silicone, non-polar Chemcatcher) + = log Kow > 4; - = log Kow < 3 "potential of hydrophilic samplers (POCIS, the hydrophilic version of Chemcatcher, Empore™ disks and others) + = log Kow < 3 ; - = log Kow >4 cstage of development: d = performance has been demonstrated in the laboratory and/or in the field; p = performance is likely to be good, but experimental evidence is not available, selected references are given to publications containing sampler calibration data Acknowledgment Supported by the EU Operational Programme "Research and Development for Innovations", the CETOCOEN project (no.CZ. 1.05/2.1.00/01.0001 ) Entox is a partnership between Queensland Health and the University of Queensland. Passive sampling research at Entox is supported under an Australian Research Council linkage project (LP0883675). References 1. C. Zwiener and F.H. Frimmel (2004). LC-MS analysis in the aquatic environment and in water treatment - A critical review: Part II: Applications for emerging contaminants and related pollutants, microorganisms and humic acids. Analytical and Bioanalytical Chemistry 378: 862-874. 2. R.P. Schwarzenbach, B.I. Escher, K. Fenner, T.B. Hofstetter, C.A. Johnson, U. von Gunten, and B. Wehrli. The challenge of micropollutants in aquatic systems (2006). Science 313: 1072-1077. 3. B. Vrana, I.J. Allan, R. Greenwood, G.A. Mills, E. Dominiak, K. Svensson, J. Knutsson, and G. Morrison (2005). Passive sampling techniques for monitoring pollutants in water. TrAC Trends in Analytical Chemistry 24: 845-868. 4. Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, D. Barcelo (series ed.), ed. R. Greenwood, G. Mills and B. Vrana, Elsevier, Amsterdam, 2007, 453 p. 5. T. Harner, K. Pozo, T. Gouin, A.-M. Macdonald, H. Hung, J. Cainey, and A. Peters (2006) . Global pilot study for persistent organic pollutants (POPs) using PUF disk passive air samplers. Environmental Pollution 144: 445-452. 6. K. Pozo, T. Harner, F. Wania, D.C.G. Muir, K.C. Jones and A. Leonard (2006). Towards a global monitoring network for persistent organic pollutants in air: results from the GAPs study. Environmental Science and Technology 40: 4867-4873. 7. J.N. Huckins, G. K. Manuweera, J.D. Petty, D. Mackay, and J.A. Lebo (1993). Lipid-containing semipermeable membrane devices for monitoring organic contaminants in water. Environmental Science and Technology 27: 2489-2496. 8. D.A. Alvarez, PhD. thesis: Development of an integrative sampling device for hydrophilic organic contaminants in aquatic environments. University of Missouri-Columbia, Columbia, MO, USA, 1999. 9. J.K. Kingston, R. Greenwood, G.A. Mills, G.M. Morrison, and B.L. Persson (2000). Development of a novel passive sampling system for the time-averaged measurement of a range of organic pollutants in aquatic environments. Journal of Environmental Monitoring 2: 487-495. 10. F. Stuer-Lauridsen (2005). Review of passive accumulation devices for monitoring organic micropollutants in the aquatic environment. Environmental Pollution 136: 503-524. 11. G. Ouyang, J. Pawliszyn (2007). Configurations and calibration methods for passive sampling techniques. Journal of Chromatography A 1168: 226-235. 12. A. Kot-Wasik, B. Zabiegata, M. Urbanowicz, E. Dominiak, A. Wasik, and J. Namiešnik (2007): Advances in passive sampling in environmental studies. Analytica Chimica Acta 602: 141-163. 13. K. Booij. Performance of passive samplers for monitoring priority substances. Report for ICES Marine Chemistry Working Group, 2009. http://www.ices.dk/reports/MHC/2009/MCWG09.pdf 14. G.A. Mills, B. Vrana, I. Allan, D.A. Alvarez, J.N. Huckins, and R. Greenwood (2007) : Trends in monitoring pharmaceuticals and personal-care products in the page 36 aquatic environment by use of passive sampling devices. Analytical and Bioanalytical Chemistry 387: 1153-1157. 15. H. Söderström, R.H. Lindberg, and J. Fick (2009): Strategies for monitoring the emerging polar organic contaminants in water with emphasis on integrative passive sampling: Tools for the REACH Programme - analytical methods for the evaluation of industrial contaminants Journal of Chromatography A. 1216: 623-630. 16. J.N. Huckins, J.D. Petty, and K. Booij (2006) (eds.) Monitors of organic chemicals in the environment: semipermeable membrane devices. Springer, New York. 17. R. Greenwood, G. Mills, and B. Vrana, (2007) (eds.) Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, Barcelo, D. (series Editor), Elsevier, Amsterdam. 18. J. Harmsen (2007). Measuring bioavailability: From a scientific approach to standard methods. Journal of Environmental Quality 36: 1420-1428. 19. ISO 17402:2008 (TC 190 Soil quality) Soil quality - Requirements and guidance for the selection and application of methods for the assessment of bioavailability of contaminants in soil and soil materials. 20. P. Mayer, J. Tolls, J.L.M. Hermens, and D. Mackay (2003). Equilibrium Sampling Devices. Environmental Science and Technology 185A-191A. 21. J. A. Magnér, T. E. Alsberg and D. Broman (2009): Evaluation of poly(ethylene-co-vinyl acetate-co-carbon monoxide) and polydimethylsiloxane for equilibrium sampling of polar organic contaminants in water. Environmental Toxicology and Chemistry 28: 1874-1880. 22. J.N. Huckins, J.D. Petty, J.A. Lebo, F.V. Almeida, K. Booij, D.A. Alvarez, W. Cranor, L.R.C. Clark, and B.B. Mogensen (2002). Development of the permeability/performance reference compound (PRC) approach for in situ calibration of semipermeable membrane devices (SPMDs). Environmental Science & Technology 36: 85-91. 23. K. Booij, R. van Bommel, A. Mets, and R. Dekker (2006): Little effect of excessive biofouling on the uptake of organic contaminants by semipermeable membrane devices. Chemosphere 65: 2485-2492. 24. K. Booij, J.N. Huckins, and B. Vrana. Theory, modeling and calibration of passive sampling devices in water monitoring. In: R. Greenwood, G. Mills, B. Vrana, (2007) (eds.) Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, Barcelo, D. (series Editor), Elsevier, Amsterdam, p.141-164. 25. C. Harman, O. Boyum, K. V. Thomas, and M. Grung (2009). Small but different effect of fouling on the uptake rates of semipermeable membrane devices and polar organic chemical integrative samplers. Environmental Toxicology and Chemistry. 28:2324-2332. 26. M. Shaw, G. Eaglesham and J.F. Mueller (2009): Uptake and release of polar compounds in SDB-RPS Empore(TM) disks; implications for their use as passive samplers Chemosphere 75:1-7. 27. N. Mazzella, S. Lissalde, S. Moreira, F. Delmas, P. Mazellier and J.N. Huckins, (2010). Evaluation of the use of performance reference compounds in an Oasis- page 37 HLB adsorbent based passive sampler for improving water concentration estimates of polar herbicides in freshwater. Environmental Science and Technology 44: 1713-1719. 28. H. Budzinski, personal communication. 29. D.S. O'Brien, B. Chiswell, and J. F. Mueller, (2009). A novel method for the in situ calibration of flow effects on a phosphate passive sampler. Journal of Environmental Monitoring 11: 212-219. 30. I. R. Falconer, Cyanobacterial Toxins of Drinking Water Supplies: cylindrospermopsins and microcystins. Boca Raton, Florida, USA: CRC Press;(2006), 279pp. 31. L. MacKenzie, V. Beuzenberg, P. Holland, P. McNabb, and A. Selwood (2004). Solid phase adsorption toxin tracking (SPATT): a new monitoring tool that simulates the biotoxin contamination of filter feeding bivalves. Toxicon 44: 901-918. 32. E. Fux, C. Marcaillou, F. Mondeguer, R. Bire, and P. Hess (2008). Field and mesocosm trials on passive sampling for the study of adsorption and desorption behaviour of lipophilic toxins with a focus on OA and DTX1. Harmful Algae. 7:574-583. 33. T. Rundberget, E. Gustad, I. A. Samdal, M. Sandvik, and CO. Miles (2009). A convenient and cost-effective method for monitoring marine algal toxins with passive samplers. Toxicon.; 53:543-550. 34. D. Shea, P. Tester, J. Cohen, S. Kibler, and S. Varnam (2006). Accumulation of brevetoxins by passive sampling devices. African Journal of Marine Science. 28:379-381. 35. J. Kohoutek, P. Babica, L. Blaha and B. Marsalek (2008). A novel approach for monitoring of cyanobacterial toxins: development and evaluation of the passive sampler for microcystins. Analytical and Bioanalytical Chemistry 390:1167-1172. 36. N. Folsvik, E. M. Brevik, J. A. Berge (2000). Monitoring of organotin compounds in seawater using semipermeable membrane devices. Journal of Environmental Monitoring 2: 281-284. 37. N. Folsvik, E. M. Brevik, and J. A. Berge (2002). Organotin compounds in a Norwegian fjord. A comparison of concentration levels in semipermeable membrane devices (SPMDs), blue mussels (Mytilus edulis) and water samples. Journal of Environmental Monitoring 4: 280-283. 38. R. Aguilar-Martfnez, R. Greenwood, G.A. Mills, B. Vrana, M.A. Palacios-Corvillo, and M.M.Gomez-Gomez (2008). Assessment of Chemcatcher passive sampler for the monitoring of inorganic mercury and organotin compounds in water. International Journal of Environmental Analytical Chemistry 88: 75-90. 39. R. Aguilar-Martfnez, M.A. Palacios-Corvillo, R. Greenwood, G. A. Mills, B. Vrana, and M.M. Gomez-Gomez (2008). Calibration and use of the Chemcatcher® passive sampler for monitoring organotin compounds in water. Analytica Chimica Acta 618: 157-167. 40. K. Booij, B.N. Zegers, and J.P. Boon, (2002). Levels of some polybrominated diphenyl ether (PBDE) flame retardants along the Dutch coast as derived from page 38 their accumulation in SPMDs and blue mussels (Mytilus edulis). Chemosphere 46: 683-688. 41. S. Rayne, and M.G. Ikonomou (2002). Reconstructing source polybrominated diphenyl ether congener patterns from semipermeable membrane devices in the Fraser River, British Columbia, Canada: comparison to commercial mixtures. Environmental Toxicology and Chemistry 21: 2292-2300. 42. T. Colborn, F.S. vom Saal, and A.M. Soto (1993). Developmental effects of endocrine disrupting chemicals in wildlife and humans. Environmental Health Perspective 101:378-384. 43. IPCS & WHO 2002. Global assessment of the state-of-the-science of endocrine disruptors (eds: T. Damsra, S. Barlow, A. Bergman, R. Kavloc & G. Van Der Kraak). 44. R.J. Williams, A.C. Johnson, J.J.L. Smith and R. Kanda (2003): Steroid estrogens profiles along river stretches arising from sewage treatment works discharges. Environmental Science and Technology 37, 1744-1750. 45. S.D. Richardson, and T.A. Ternes (2005). Water analysis: Emerging contaminants and current issues. Analytical Chemistry 77: 3807-3838. 46. CR. Casey, L. Strattan, T. L. Jones-Lepp, and D. Alvarez (2004). EPA Science Forum 2004: Healthy Communities and Ecosystems, Washington D.C. 47. T. Günther, M. Strauss, J.B. Kopp, and R. Hartmann (2009): Identifizierung und Verminderung der PFT-Belastung im Klärschlamm der Kläranlagen Hann, Münden und Hedemünden. KA Korrespondenz Abwasser, Abfall, 56:690-695. 48.. Y. Horii and K. Kannan (2008). Survey of organosilicone compounds, including cyclic and linear siloxanes, in personal-care and household products. Archives of Enviromental Contamination and Toxicology 5: 701-710. 49. C. Sparham, R. Van Egmond, S. O'Connor, C. Hastie, M. Whelan, R. Kanda and O. Franklin (2008). Determination of decamethylcyclopentasiloxane in river water and final effluent by headspace gas chromatography/mass spectrometry. Journal of Chromatography A 1212: 124-129. 50. R. Dewil, L.Appels, J. Baeyens, A. Buczynska and L. Vaeck (2007). The analysis of volatile siloxanes in waste activated sludge. Talanta 74: 14-19. 51. K. Fent, A. A. Weston, and D. Caminada (2006). Ecotoxicology of human pharmaceuticals. Aquatic Toxicology 76:122-159. 52. B.S. Stephens and J.F. Müller. Techniques for quantitatively evaluating aquatic passive sampling devices. In: R. Greenwood, G. Mills, B. Vrana, (2007) (eds.) Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, Barcelo, D. (series Editor), Elsevier, Amsterdam, p. 329-346. 53. Miller GT (2004), Sustaining the Earth, 6th edition. Thompson Learning, Inc. Pacific Grove, California. Chapter 9, Pages 211-216. 54. http://en.wikipedia.org/wiki/Environmental_effects_of_pesticides 55. European Commission, 2001. Decision 2455/2001/EC of 20 November 2001 establishing a list of priority substances in the field of water policy. Off. J. Eur. Comm. L 331, 15.12.2001. page 39 56. European Commission 2006. A thematic strategy on the sustainable use of pesticides, COM(2006) 372, 12.7.2006. 57. D.A. Alvarez, J.D. Petty, J.N. Huckins, T.L. Jones-Lepp, D.T. Getting, J.P. Goddard, and S.E. Manahan (2004). Development of a passive in situ sampler for hydrophilic organic contaminants in aquatic environments. Environmental Toxicology and Chemistry 23:1640-1648. 58. D.A. Alvarez, J.N. Huckins, J.D. Petty, T. Jones-Lepp, F. Stuer-Lauhdsen, D.T. Getting, J.P. Goddard, and A. Gravell. Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS). In: In: Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, D. Barcelo (series ed.), ed. R. Greenwood, G. Mills and B. Vrana, Elsevier, Amsterdam, 2007, 171-197. 59. A.T.K. Tran, R.V. Hyne, and P. Doble, (2007). Calibration of a passive sampling device for time-integrated sampling of hydrophilic herbicides in aquatic environments. Environmental Toxicology and Chemistry 26 435-443. 60. M. Shaw, and J.F. Mueller, (2005). Preliminary evaluation of the occurrence of herbicides and PAHs in the Wet Tropics region of the Great Barrier Reef, Australia, using passive samplers, Marine Pollution Bulletin, 51: 876-881. 61. R.B. Schäfer, A. Paschke, B. Vrana, R. Mueller, and M. Liess, (2008). Performance of the Chemcatcher (R) passive sampler when used to monitor 10 polar and semi-polar pesticides in 16 central European streams, and comparison with two other sampling methods. Water Research 42: 2707-2717. 62. B.S. Stephens, A. Kapernick, G. Eaglesham, and J. Mueller, 2005. Aquatic passive sampling of herbicides on naked particle loaded membranes: accelerated measurement and empirical estimation of kinetic parameters. Environmental Science and Technology 39: 8891-8897. 63. T. Poiger, HR. Buser, M.E. Balmer, P.-A. Bergqvist, and M.D. Muller. (2004). Occurrence of UV filter compounds from sunscreens in surface waters: regional mass balance in two Swiss lakes. Chemosphere 55: 951-963. 64. M.E. Balmer, HR. Buser, M.D. Muller, and T. Poiger (2005). Occurrence of some organic UV filters in wastewater, in surface waters, and in fish from Swiss lakes. Environmental Science & Technology 39: 953-962. 65. A. Zenker, H. Schmutz, and K. Fent (2008). Simultaneous trace determination of nine organic UV-absorbing compounds (UV Filters) in environmental samples. Journal of Chromatography A. 1202: 64-74. 66. K. Fent, A. Zenker, and M. Rapp (2010). Widespread occurrence of estrogenic UV-filters in aquatic ecosystems in Switzerland. Environmental Pollution 158:1817-1824. 67. F. Smedes, C. Tixier, I. Davies, P. Roose, T. van der Zande, and J. Tronczynski, ICES Passive Sampling Trial Survey of for hydrophobic contaminants Water and Sediment; including laboratory intercalibration. http://www.passivesampling.net 68. W. Brack, N. Bandow, K. Schwab, T. Schulze, and G. Streck (2009). Bioavailability in Effect-Directed Analysis of Organic Toxicants in Sediments. Trac-Trends in Analytical Chemistry 28: 543-549. page 40 69. L.A. Fernandez, J.K. MacFarlane, A.P. Tcaciuc, and P.M. Gschwend (2009).. Measurement of freely dissolved PAH concentrations in sediment beds using passive sampling with low-density polyethylene strips. Environmental Science & Technology 43: 1430-1436. 70. P. Mayer, W.Vaes, F. Wijnker, K. Legierse, R. Kraaij, J. Tolls, and J. Hermens (2000). Sensing dissolved sediment porewater concentrations of persistent and bioaccumulative pollutants using disposable solid-phase microextraction fibers. Environmental Science & Technology 34: 5177-5183. 71. K. Booij, J.R. Hoedemaker, and J.F. Bakker (2003). Dissolved PCBs, PAHs, and HCB in pore waters and overlying waters of contaminated harbor sediments. Environmental Science and Technology 37: 4213-4220. 72. F. Smedes, and I. Davies. ICES Annual Report 2007. Theme Session J -Applications of passive sampling devices in environmental monitoring, assessment, research and testing. http://www.ices.dk/iceswork/asc/2007/ThemeSessions/synopses/SessionJ.pdf 73. T.E.M. ten Hulscher, J. Postma, P.J. den Besten, G.J. Stroomberg, A. Belfroid, J.W. Wegener, J. H. Faber, J.J.C. van der Pol, A. J. Hendriks, and P.CM. van Noort. (2003). Tenax Extraction mimics benthic and terrestrial bioavailability of organic compounds. Environmental Toxicology and Chemistry 22: 2258-2265. 74. A. de la Cal, E. Eljarrat, T. Grotenhuis, and D. Barcelo (2008). Tenax(R) extraction as a tool to evaluate the availability of polybrominated diphenyl ehters, DDT, and DDT metabolites in sediments. Environmental Toxicology and Chemistry 27: 1250-1256. 75. G. Cornelissen, A. Pettersen, D. Broman, P. Mayer, and G.D. Breedveld (2008). Field testing of equilibrium passive samplers to determine freely dissolved native polycyclic aromatic hydrocarbon concentrations. Environmental Toxicology and Chemistry 27: 499-508. 76. W. Davison, H. Zhang, and K.W. Warnken. Theory and applications of DGT measurements in soils and sediments. In: Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, D. Barcelo (series ed.), ed. R. Greenwood, G. Mills and B. Vrana, Elsevier, Amsterdam, 2007, pp. 353-378. 77. ICES. 2006. Report of the Working Group on Marine Sediments in Relation to Pollution (WGMS), 27-31 March 2006, ICES Headquarters, Copenhagen. ICES CM 2006/MHC:01. 44 pp. 78. J.N. Huckins, J.D. Petty, and K. Booij. Chapter 7 Comparison to biomonitoring organisms. In: Monitors of organic chemicals in the environment. Semipermeable membrane devices, ed. J. N. Huckins, J.D. Petty and K. Booij, Springer, New York, 2006, 139-167. 79. F. Smedes. Monitoring by passive sampling in concert with deployed mussels. In: Passive sampling techniques in environmental monitoring. Comprehensive Analytical Chemistry Series, D. Barcelo (series ed.), ed. R. Greenwood, G. Mills and B. Vrana, Elsevier, Amsterdam, 2007, pp. 407-453. 80. H.A. Leslie, A.J.P. Oosthoek, F. J. M. Busser, M.H.S. Kraak, and J.L.M. Hermens. (2002). Biomimetic solid-phase microextraction to predict body residues page 41 and toxicity of chemicals that act by narcosis. Environmental Toxicology and Chemistry 21:229-234. 81. S. Bayen, T.L. ter Laak, J. Buffle, and J.L.M. Hermens (2009). Dynamic exposure of organisms and passive samplers to hydrophobic chemicals. Environmental Science and Technology 43: 2206-2215. 82. J.N. Brown, N. Paxqus, L. Forlin, and D.G.J. Larsson, (2007) Variations in bioconcentration of human pharmaceuticals from sewage effluents into fish blood plasma. Environmental Toxicology and Pharmacology. 24:267-274. 83. S. Gartiser, C Hafner, S. Oeking, and A. Paschke A. (2009). Results of a "Whole Effluent Assessment" study from different industrial sectors in Germany according to OSPAR's WEA strategy. Journal of Environmental Monitoring 11:359-369. 84. B. Roig, I. Allan, G.A. Mills, N. Guigues, R. Greenwood, and C Gonzalez (2009). Existing and new methods for chemical and ecological status monitoring under the WFD. In C.Gonzalez, R. Greenwood and P. Quevauviller (Eds.) Rapid Chemical and Biological Techniques for Water Monitoring. Wiley, Chichester. 85. E.J. Routledge and J.P. Sumpter (1996). Estrogenic activity of surfactants and some of their degradation products assessed using a recombinant yeast screen. Environmental Toxicology and Chemistry 15: 241-248. 86. B. Roig, I. Bazin, S. Bayle, D. Habauzit, and J. Chopineau, (2009). Biomolecular recognition systems for water monitoring. In C.Gonzalez, R. Greenwood and P. Quevauviller (Eds.) Rapid Chemical and Biological Techniques for Water Monitoring. Wiley, Chichester. 87. A. Rastall, A. Neziri, Z. Vukovic, C Jung, S. Mijovic, H. Hollert, S. Nikcevic, and L. Erdinger (2004). The identification of readily bioavailable pollutants in lake shkodra/skadar using semipermeable membrane devices (SPMDs), bioassays and chemical analysis. Environmental Science and Pollution Research 11: 240-253. 88. J.A. Lebo, F.V. Almeida, W.L. Cranor, J.D. Petty, J.N. Huckins, A. Rastall, D.A. Alvarez, B.B. Mogensen, and B.T. Johnson. (2004). Purification of triolein for use in semipermeable membrane devices (SPMDs). Chemosphere 54: 1217-1224. 89. J.D. Petty, J.N. Huckins, D.A. Alvarez, W.G. Brumbaugh, W.L. Cranor, R.W. Gale, A.C. Rastall, T.L. Jones-Lepp, T.J. Leiker, CE. Rostad, and E.T. Furlong (2004). A holistic passive integrative sampling approach for assessing the presence and potential impacts of waterborne environmental contaminants. Chemosphere 54: 695-705. 90. E.L.M. Vermeirssen, O. Körner, R. Schönenberger, M.J.-F. Suter, and P. Burkhardt- Holm (2005). Characterization of environmental estrogens in river water using a three pronged approach: active and passive water sampling and the analysis of accumulated estrogens in the bile of caged fish. Environmental Science and Technology 39: 8191-8198. 91. E.L.M. Vermeirssen, M.J.-F. Suter, and P. Burkhardt-Holm 2006. Estrogenicity patterns in the Swiss midland river Lützelmurg in relation to treated domestic sewage effluent discharges and hydrology. Environmental Toxicology & Chemistry 25: 2413-2422. 92. P. Matthiessen, D. Arnold, A.C. Johnson, T.J. Pepper, T.G. Pottinger, and K.G.T. Pulman (2006). Contamination of headwater streams in the United Kingdom by page 42 oestrogenic hormones from livestock farms. Science of The Total Environment 367: 616-630. .93. C. Liscio, E. Magi, M. Di Carro, M.J.-F. Suter, E.L.M. Vermeirssen (2009). Combining passive samplers and biomonitors to evaluate endocrine disrupting compounds in a wastewater treatment plant by LC/MS/MS and bioassay analyses. Environmental Pollution 157: 2716-2721. 94. D.A. Alvarez, W.L. Cranor, S.D. Perkins, R.C. Clark, and S.B. Smith (2008). Chemical and toxicologic assessment of organic contaminants in surface water using passive samplers. Journal of Environmental Quality 37:1024-1033. 95. E.L.M. Vermeirssen, J. Asmin, B.I. Escher, J.-H. Kwon, I. Steimen, and J. Hollender (2008). The role of hydrodynamics, matrix and sampling duration in passive sampling of polar compounds with Empore™ SDB-RPS disks. Journal of Environmental Monitoring 10: 119 -128. 96. S.K. Bopp, M.S. McLachlan, and K. Schirmer (2007). Passive sampler for combined chemical and toxicological long-term monitoring of groundwater: the ceramic toximeter. Environmental Science and Technology 41: 6868-6876. 97. B.I. Escher, P. Quayle, R. Muller, U. Schreiber, and J.F. Mueller (2006). Passive sampling of herbicides combined with effect analysis in algae using a novel high-throughput phytotoxicity assay (Maxi-lmaging-PAM). Journal of Environmental Monitoring 8: 456-464. 98. R. Muller, J.Y.M. Tang, R. Thier, and J.F. Mueller (2007). Combining passive sampling and toxicity testing for evaluation of mixtures of polar organic chemicals in sewage treatment plant effluent. Journal of Environmental Monitoring 9: 105-110. 99. E.L.M. Vermeirssen, N. Bramaz, J. Hollender, H. Singer, and B.I. Escher (2009). Passive sampling combined with ecotoxicological and chemical analysis of pharmaceuticals and biocides - evaluation of three Chemcatcher™ configurations. Water Research 43: 903-914. 100. E.L.M. Vermeirssen, J. Hollender, N. Bramaz, J. Van der Voet, and B.I. Escher. Linking toxicity in algal and bacterial assays with chemical analysis in passive samplers deployed in 21 treated sewage effluents. Environmental Toxicology and Chemistry, accepted. 101. V. Koči, T. Ocelka, M. Mlejnek, and R. Grabic (2004). Efficiency assessment of wastewater treatment plant based on SPMD sampling. Central European Journal of Chemistry 2: 91-112. 102. M. Shaw, A. Negri, K. Fabhcius, and J.F. Mueller (2009). Predicting water toxicity: Pairing passive sampling with bioassays on the Great Barrier Reef. Aquatic Toxicology 95: 108-116. 103. V. Koči, M. Mlejnek, L. Kochánková (2003). Toxicological evaluation of exposed SPMD membranes. Central European Journal of Chemistry 1: 28-34. 104. M. Van den Berg, L. Birnbaum, A.T.C. Bosveld, B. Brunstrôm, P. Cook, M. Feeley, J.P. Giesy, A. Hanberg, R. Hasegawa, S.W. Kennedy, T. Kubiak, J.C. Larsen, F.X.R. van Leeuwen, A.K.D. Liem, C. Nolt, R.E. Peterson, L. Poellinger, S. Safe, D. Schrenk, D. Tillitt, M. Tysklind, M. Younes, F. Waern, and T. Zacharewski. page 43 (1998). Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environmental Health Perspectives 106: 775-792. 105. B.I. Escher, N. Bramaz, J.F. Mueller, P. Quayle, S. Rutishauser, E.L.M. Vermeirssen (2008). Toxic equivalent concentrations (TEQs) for baseline toxicity and specific modes of action as a tool to improve interpretation of ecotoxicity testing of environmental samples. Journal of Environmental Monitoring 10: 612-621. 106 A. Kortenkamp (2007). Ten years of mixing cocktails: a review of combination effects of endocrine-disrupting chemicals. Environmental Health Perspectives 115 (Suppl 1): 98-105. 107. Commission Regulation (EC) No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs. Official Journal of the European Union L 364/5-24. 108. W. Brack (2003). Effect-directed analysis: a promising tool for the identification of organic toxicants in complex mixtures? Analytical and Bioanalytical Chemistry 377: 397-407. 109. W. Brack, R. Altenburger, U. Ensenbach, M. Moder, H. Segner, and G. Schüürmann, (1999). Bioassay-directed identification of organic toxicants in river sediment in the industrial region of Bitterfeld (Germany) - A contribution to hazard assessment Archives of Environmental Contamination and Toxicology, 37: 164- 110. A.C. Rastall, D. Getting, J. Goddard, DR. Roberts, and L. Erdinger (2006). A biomimetic approach to the detection and identification of estrogen receptor agonists in surface waters using semipermeable membrane devices (SPMDs) and bioassay-directed chemical analysis. Environmental Science and Pollution Research 13: 256-267. 111. T. Rusina, F. Smedes, M. Kobližková, and J. Klanová (2010). Calibration of silicone rubber passive samplers: experimental and modeled relations between sampling rate and compound properties. Environmental Science & Technology 44: 362-367. 112. http://www.swift-wfd 113. C. Gonzalez, R. Greenwood, and P. Quevauviller (eds.). Rapid Chemical and Biological Techniques for Water Monitoring - Water Quality Measurements Series, New York, Wiley, 2009. 419pp. 114. http://www.aquaref.fr 115. http://www.cslab.cz/IPSIC2010_The%20invitation_April2010.pdf 116. http://www.port.ac.uk/research/stamps/projectdescription/ 117. BSI, Publicly Available Specification: Determination of priority pollutants in surface water using passive sampling (PAS-61), May 2006. 118. ISO/DIS 5667-23. Water quality - Sampling - Part 23: Determination of priority pollutants in surface water using passive sampling; draft under development. 119. R. Loos, B.M. Gawlik, G. Locoro, E. Rimaviciute, S. Contini, and G. Bidoglio (2008). EU wide monitoring survey of polar persistent pollutants in European river waters. JRC Scientific and Technical Report. Institute for Environment and Sustainability, Joint Research Centre. ISBN 978-92-79-10649-1. 174. page 44 120 N. A. Warner, A. Evenset, G, Christensen, G. W. Gabrielsen, K. Borgá, and H. Leknes (2010). Volatile siloxanes in the European arctic: assessment of sources and spatial distribution. Environmental Science & Technology, in press. 121. D.A. Alvarez, P.E. Stackelberg, J.D. Petty, J.N. Huckins, E.T. Furlong, S.D. Zaugg, and M.T.Meyer (2005). Comparison of a novel passive sampler to standard water-column sampling for organic contaminants associated with wastewater effluents entering a New Jersey stream., Chemosphere, 61: 610-622. 122. I.J. Allan, J. Knutsson, N. Guigues, G.A. Mills, A.M. Fouillac, and R. Greenwood (2008). Chemcatcher(R) and DGT passive sampling devices for regulatory monitoring of trace metals in surface water. Journal of Environmental Monitoring 10: 821-829. 123. I.J. Allan, J. Knutsson, N. Guigues, G.A. Mills, A.-M. Fouillac, and R. Greenwood, (2007). Evaluation of the Chemcatcher and DGT passive samplers for monitoring metals with highly fluctuating water concentrations. Journal of Environmental Monitoring 9: 672-681. 124. J. Llorca, Julio, C. Gutiqrrez, E. Capilla, R. Tortajada, L. Sanjubn, A. Fuentes, and I. Valor, (2009). Constantly stirred sorbent and continuous flow integrative sampler: New integrative samplers for the time weighted average water monitoring. Journal of Chromatography A. 1216:5783-5792. 125. J. Kohoutek, B. Maršálek, and L. Bláha (2010). Evaluation of the novel passive sampler for cyanobacterial toxins microcystins under various conditions including field sampling. Analytical and Bioanalytical Chemistry 397:823-828. 126. A. Arditsoglou, and D. Voutsa (2008). Passive Sampling of Selected Endocrine Disrupting Compounds Using Polar Organic Chemical Integrative Samplers. Environmental Pollution 156:316-324. 127. C. Harman, K.E. Tollefsen, O. Boyum, K. Thomas, and M. Grung (2008). Uptake rates of alkylphenols, PAHs and carbazoles in semipermeable membrane devices (SPMDs) and polar organic chemical integrative samplers (POCIS). Chemosphere. 72:1510-1516. 128. Z.L. Zhang, A., Hibberd, and J.L. Zhou (2008). Analysis of emerging contaminants in sewage effluent and river water: comparison between spot and passive sampling. Analytica Chimica Acta; 607:37-44. 129. H. Li, P.A. Helm, and CD. Metcalfe (2010). Sampling in the great lakes for pharmaceuticals, personal care products, and endocrine-disrupting substances using the passive polar organic chemical integrative sampler. Environmental Toxicology and Chemistry 29:751-762. 130. R.V. Hyne, F. Pablo, M. Aistrope, A.W. Leonard, and N. Ahmad (2004). Comparison of time-integrated pesticide concentrations determined from field-deployed passive samplers with daily river-water extractions. Environmental Toxicology and Chemistry 23: 2090-2098. 131. N. Mazzella, J.F. Dubernet, and F. Delmas (2007). Determination of kinetic and equilibrium regimes in the operation of polar organic chemical integrative samplers - application to the passive sampling of the polar herbicides in aquatic environments. Journal of Chromatography A 1154:42-51. page 45 132. R. Gunold, R.B. Schafer, A. Paschke, G. Schuurmann, and M. Liess (2008) Calibration of the Chemcatcher passive sampler for monitoring selected polar and semi-polar pesticides in surface water. Environmental Pollution 155: 52-60. 133. A.W. Leonard, R.V. Hyne, and F. Pablo (2002). Trimethylpentane-containing passive samplers for predicting time-integrated concentrations of pesticides in water: laboratory and field studies. Environmental Toxicology and Chemistry 21: 2591-2599. 134. N. Mazzella, T. Debenest, and F. Delmas (2008). Comparison between the polar organic chemical integrative sampler and the solid-phase extraction for estimating herbicide time-weighted average concentrations during a microcosm experiment. Chemosphere. 73:545-550. 135. F.A. Esteve-Turrillas, A. Pastor, and M. De La Guardia (2007). Behaviour of semipermeable membrane devices in neutral pesticide uptake from waters. Analytical and Bioanalytical Chemistry 387: 2153-2162. 136. D. Sabaliunas, S.F. Webb, A. Hauk, M. Jacob, and W.S. Eckhoff (2003). Environmental fate of Triclosan in the River Aire Basin, UK. Water Research 37: 3145-3154. 137. H. Singer, S. Müller, C. Tixier, and L. Pillonel (2002).Occurrence and environmental behavior of the bactericide triclosan and its methyl derivative in surface waters and in wastewater. Environmental Science & Technology 36: 4998-5004. 138. A. Togola, and H. Budzinski, (2007). Development of polar organic integrative samplers for analysis of pharmaceuticals in aquatic systems. Analytical Chemistry 79: 6734-6741. 139. S.L. Bartelt-Hunt, D.D. Snow, T. Damon, J. Shockley, and K. Hoagland (2009). The occurrence of illicit and therapeutic pharmaceuticals in wastewater effluent and surface waters in Nebraska. Environmental Pollution 157:786-791. 140. D.A. Alvarez, J.D.Petty, J.N. Huckins, T.L. Jones-Lepp, D.T. Getting, J.P. Goddard, and S.E. Manahan, (2004). Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environmental Toxicology and Chemistry 23: 1640-1648. 141 S.L. Macleod, E.L. McClure, and CS. Wong (2007). Laboratory calibration and field deployment of the polar organic chemical integrative sampler for pharmaceuticals and personal care products in wastewater and surface water. Environmental Toxicology and Chemistry 26:2517-2529. 142. B.L.L. Tan, D.W. Hawker, J.F. Muller, F.D.L. Leusch, L.A. Tremblay, and H.F. Chapman, (2007). Comprehensive study of endocrine disrupting compounds using grab and passive sampling at selected wastewater treatment plants in South East Queensland, Australia. Environment International 33: 654-669.