/. Phys. Chem. A 2003, 107, 3305-3315 3305 Photohydration of Benzophenone in Aqueous Acicr Markus Ramseier, Paul Senn, and Jakob Wirz* Departement Chemie der Universität Basel, Klingelbergstrasse 80, CH-4056 Basel, Switzerland Received: August 1, 2002; In Final Form: November 13, 2002 Why is the triplet state of aromatic ketones quenched by protons? The long-known but unexplained quenching process was investigated in detail for benzophenone (1). Adiabatic protonation of triplet benzophenone, 31, encounters a state symmetry-imposed barrier, because the electronic structure of 31 is n,7T*, while that of its conjugate acid, 31H+, is Jt,jt*. Hence, the rate of protonation of 31, ka+ = 6.8 x 10s M-1 s-1, is well below the diffusion-controlled limit. The short-lived transient intermediate formed by protonation of 31 in 0.1 —1 M aqueous HCIO4 (Amax = 320 and 500 nm, t = 50 ns) is not 31H+, as was assumed in previous studies. The latter (Amax = 385 nm) is detectable only in acidified acetonitrile or in highly concentrated aqueous acid (>5 M HCIO4), where water activity is low. In moderately concentrated aqueous acids, adiabatic protonation of 31 is the rate-limiting step preceding rapid adiabatic hydration of a phenyl ring, 31H+ + H2O —* 31-H20, ko = 1.5 x 109 s-1. These findings lead to a revised value for the acidity constant of protonated 31, pATa(3lH+) = —0.4 ±0.1. Acetophenone (2) and several derivatives of 1 and 2 undergo a similar reaction sequence in aqueous acid. The acid-catalyzed photohydration of parent 1 and 2 is reversible. In meta-fluorinated derivatives, the reaction results in a clean and efficient formation of the corresponding phenols, a novel aromatic photosubstitution reaction. This indicates that hydration of 31H+ occurs predominantly at the meta position. A long-lived transient (Amax = 315 nm, r = 5.4 s) left after the decay of 31-H20 is attributed to a small amount of ortho-h^O that regenerates 1 more slowly. Introduction Since Hammond and co-workers1 elucidated the mechanism of the photoreduction of benzophenone (1) in 1962, the photophysical and photochemical properties of this molecule are known to any photochemist. Equally well-established is the fact that phenols become stronger acids upon electronic excitation,2 while aromatic ketones become stronger bases. Thirty years ago, Ledger and Porter3 reported that the phosphorescence of 1 in aqueous solution is quenched by protons with a rate constant of 6 x 10s M-1 s-1. They noted that "the quenching action of the proton is not understood". Several studies have since dealt with the photoprocesses of 1 in aqueous solution.4-13 Wyatt and co-workers4-5 investigated the initial absorbances and lifetimes of the triplet intermediates formed by flash photolysis of 1 over a wide range of acid concentrations and determined a p^"a of 1.5 ± 0.1 for the adiabatic dissociation constant of protonated triplet 1, 31H+. They found that the lifetime of 31 drops below 100 ns around pH « 0, whereas that of its conjugate acid 31H+ rises to 50 /is in degassed solutions of still higher acidities, and they concluded, "A possible explanation is that water is an effective quencher of [31H+]." Despite a number of follow-up studies,6-8 a satisfactory explanation for this puzzling phenomenon has not come forth. We report an investigation of 1 and of acetophenone (2) in aqueous acid by picosecond pump—probe spectroscopy, nanosecond laser flash photolysis (LFP), conventional flash photolysis, and photoacoustic calorimetry. We conclude that acid-catalyzed photohydration is responsible for the triplet quenching process observed in moderately acidic aqueous solutions. * To whom correspondence should be addressed. Tel. +41 61 267 38 42. Fax: +41 61 267 38 55. E-mail: J.Wirz@unibas.ch. f Part of the special issue "George S. Hammond & Michael Kasha Festschrift". Photohydration of the parent ketones 1 and 2 is reversible. However, in m-fluoro derivatives of 1 and 2, it results in efficient formation of the corresponding m-hydroxy ketones. Experimental Section The instruments used for picosecond pump—probe spectroscopy and for nanosecond LFP have been described previously.14 Briefly, pump—probe experiments were carried out using a 248 nm excitation pulse and a continuum probe pulse (300—700 nm), both with subpicosecond lifetimes. The delay line spanned a time range of 1.8 ns. Acetonitrile (2.0 M, about 10 vol %) was added as a cosolvent to achieve the required absorbance by 1 (A = 1 at the excitation wavelength of 248 nm, path length 2 mm) in neutral aqueous solutions. Addition of acetonitrile was not required to increase the solubility of 1 in aqueous acid. Nanosecond laser pulses at 248 or 351 nm were obtained from a Lambda Physik Compex 205 excimer laser (pulse width ca. 15 ns, pulse energy 100—600 mj). Pulses at 308 nm were taken from a Lambda Physik EMG 101 MSC excimer laser (pulse width ca. 20 ns, pulse energy 50—100 mj). Kinetic (1P28 photomultiplier) and spectrographic detection (MCP-gated diode array) of the transient absorption were available. Most kinetic measurements were done at ambient temperature (22 ± 2 °C). Perchloric acid solutions were made up by weighing the appropriate amounts of 70% (by weight) ultrapure HCIO4 and dilution with doubly distilled water. Fluorescence spectra were recorded on a Spex Fluorolog 111 instrument equipped with an R928 photomultiplier. Excitation spectra were corrected with a built-in Rhodamine-6G quantum counter. Fluorescence lifetimes were measured by excitation with subpicosecond pulses from a Ti: Sapphire laser (Clark MXR CPA-2001; frequency-doubled, 388 nm; frequency-doubled NOPA output, 270 nm). The emission spectra were recorded 10.1021/jp026663f CCC: $25.00 © 2003 American Chemical Society Published on Web 01/09/2003 3306 7. Phys. Chem. A, Vol. 107, No. 18, 2003 Ramseier et al. 220 300 400 500 600 ; ^h2o i i i r 1 1 | 1 1 1 1 | 1 1 1 1 | iml / / .■ > •' i • i ■ \ : i '■' > \ ■ i \ ' V / \ 1 12mhc10. •i 1 \ ■ \ 1 < ■ ' \\ / A/ N II i \ ■ - v / A J \ V" / v ■............,.......,.............. ■"r................................-tttv 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 Figure 1. Absorption spectra of 1 (ca. 4 x 10~5 M) in neutral (—) and 12 M HCIO4 (---) aqueous solution (left-hand scale). Fluorescence emission (— • —, excitation wavelength 335 nm) and excitation (•••, emission monitored at 522 nm) spectra of 1 in 12 M aqueous HCIO4 (right-hand scale). with a streak camera (Hamamatsu C5680) operated at 50 ps time resolution. The setup used for photoacoustic calorimetry has been described in detail.15 The design of the front-face irradiation flow cell and the algorithm used for the analysis of the sound waves recorded by the microphone were those of Caldwell. Melton, and co-workers.16 Solutions of 2-hydroxybenzophenone were used as a reference. The photoacoustic waveforms of the sample (E wave) were analyzed by convoluting the reference signal (J wave) with a dual or triple exponential heat deposition function and least-squares fitting of the resulting waveforms to the observed E wave. Five independent waveforms from separate experiments were analyzed for a given set of conditions. Goodness-of-fit was judged by comparing the residuals with the minimal error expected to arise from the eight bit resolution of the transient digitizer. Photohydrolysis of 3,4-Difluoroacetophenone. A solution of 3,4-difluoroacetophenone (44 mg) in 1 M aqueous HCIO4 (300 mL with 25% acetonitrile) was irradiated at 254 nm (low-pressure mercury arc) until the product absorption at 298 nm reached a maximum. The mixture was extracted with tert-butylmethyl ether, dried with MgSOzt, and evaporated, leaving a yellow oil that was chromatographed on silica using petroleum ether and methyl acetate (4:1) as an eluent. Besides 8 mg of the starting material, 16 mg of 4-fluoro-3-hydroxyacetophenone was isolated; mp 113-115 °C. !H NMR (CDCI3, TMS): <5 = 2.58 (s, 3H, Me), 5.74 (broad s, 1H, OH), 7.14 (dd, 1H, H5, 7 = 8.2 and 8.7 Hz), 7.51 (m, 1H, H6), 7.64 ppm (dd, 1H, H2, 7 = 8.4 and 2.1 Hz). 13C NMR (CDCI3, TMS): <5 = 154.22 (d, C4, 7 = 247 Hz), 143.84 (d, C3, 7 = 14.7 Hz), 134.33 (d, Q, 7=0.8 Hz), 121.75 (d, C6, 7=7.6 Hz), 117.65 (d, C2, 7=3.3 Hz), 115.73 (d, C5, 7= 19.0 Hz), 196.76 (s, C7), 26.53 ppm (s, Cg). MS (M+): mle 154. UV (/lniK/nm [logte/M-1 crrr1)]): in 0.1 M aqueous HC104: 253 [3.89], 298 [3.39]; in 0.1 M NaOH: 236 [4.34], 264 sh [3.7], 336 [3.43]. Quantum Yields of Photosubstitution. Fluorinated derivatives of 1 and 2 were irradiated at 254 nm with a medium pressure mercury arc (Hanau St 41, interference band pass filter) in degassed, 0.1 M aqueous HCIO4. The reaction progress was monitored by UV spectrophotometry. Quantum yields were determined using azobenzene in methanol as an actinometer.17 Results Absorption and Fluorescence Emission. Figure 1 displays the absorption spectra of 1 in neutral water and in 12 M aqueous HCIO4. The latter is attributed to oxygen-protonated benzophe- none, 1H+.4 Also shown are the fluorescence excitation and emission spectra of 1 in 12 M aqueous HCIO4 solution. The fluorescence of 1 in neutral aqueous solution was below the detection limit of the conventional fluorescence spectrometer used. Weak emission of 1H+ (2max = 520 nm) was observed in solutions with acid concentrations exceeding 1 M. Ground state protonation of 1 is negligible in 5.8 M HCIO4 (H0 = -2.7),18 and the excitation spectrum of the fluorescence by 1H+ in 5.8 M HCIO4 matched the absorption spectrum of neutral 1. Thus, adiabatic protonation of 1 occurs with low efficiency in the excited singlet state. The fluorescence lifetimes of 1H+ in 5.8 M (excitation at 270 nm) and 12 M air-saturated aqueous HCIO4 solution (excitation at 388 nm) were equal within experimental error, t = 700 ± 30 ps. Pump—Probe Spectroscopy. Transient absorption spectra obtained with 1 in neutral aqueous solution are shown in Figure 2. The spectra taken during the initial period (0—5 ps delay,19 not shown) were distorted by two photon absorption and exhibited spectral broadening arising from vibrational relaxation. The first spectrum shown (6 ps delay) is attributed to the lowest excited singlet state of 1 (2max = 335 and ca. 575 nm)20 as well as some broad absorption arising from solvated electrons (2max = 720 nm). Solvated electrons are also formed in the absence of a solute due to multiphoton absorption of the laser pulse by water. The initial changes in the transient spectra are attributed20 to intersystem crossing (ISC) from the lowest excited singlet state, '1, to the lowest triplet state, 31. Factor analysis21 of the spectra observed in the region of 500—550 nm required two significant components only, and global fitting to a single exponential rate law gave a singlet lifetime of 6.4 ± 0.2 ps. The spectrum of 31 Umax = 330 and 525 nm) persisted up to the maximum delay of 1.8 ns. Pump—probe spectra obtained with 1 in 1 M aqueous HCIO4 are shown in Figure 3. The spectral changes that are observed during the first 30 ps are quite similar to those in neutral water (Figure 2). However, two additional processes are indicated by the subsequent spectral changes. Factor analysis using the wavelength range of 450—550 nm indicated that four components are required for an adequate description of the spectral matrix and that a triexponential function is needed to fit the kinetics. A global least-squares fit of a triexponential function is shown in Figure 4. The spectral changes and the rate constant associated with the first process, k\ = (1.35 ± 0.09) x 1011 s_1, are the same as in neutral water (Figure 2) and are attributed to ISC of 1. The decay of the broad absorption arising from solvated electrons (2max = 720 nm) due to trapping by H+ provides the second rate constant, k = (1.7 ± 0.2) x 1010 s_1. A bimolecular rate constant k = 2.3 x 1010 M_1 s_1 is commonly quoted for this process,22 but that value pertains to low ionic strengths. The rate constant reported for the capture of solvated electrons by protons in 1 M aqueous HCIO4 is k = 1.1 x 1010 s-1.23 The final process is hard to discern by visual inspection of Figure 3. Proper analysis reveals that the band shapes and maxima gradually change from those of 31 (2max = 330 and 525 nm) to those of a species with 2max = 320 and 500 nm, which is attributed to the triplet state of a hydrate of 1, 31-H2O.20 The process is not complete within 2 ns after excitation (cf. Figure 4), and the associated rate constant is not well-defined in this time window. To avoid the perturbing influence arising from the decay of the solvated electrons, global analysis of four series of measurements in 1 M HCIO4 was restricted to the time window of 0.2—1.8 ns. In this way, an average value of £3 = Photohydration of Benzophenone in Aqueous Acid /. Phys. Chem. A, Vol. 107, No. 18, 2003 3307 300 Figure 2. Pump—probe transient absorption spectra of 1 (5 x 10 4 M) in neutral water detected with delays of 6—100 ps19 relative to the excitation pulses (248 nm, 0.5 ps half-width, 1 mJ per pulse). 300 Figure 3. Pump—probe transient absorption spectra of 1 (8 x 10 4 M) in 1 M aqueous HCIO4 detected with delays of 5 — 1826 ps19 relative to the excitation pulses (248 nm, 0.5 ps half-width, 1 mJ per pulse). (5 ± 1) x 10s s 1 was obtained for the rate constant of the slow process. Substantially different spectral changes were observed by pump—probe spectroscopy of 1 in 4 M HCIO4 (not shown). The initial changes with a rate constant of about k\ = 2 x 1011 s_1 are again attributed to ISC of !1 to 31. The kinetics of this process was somewhat distorted by the nearly simultaneous reaction of the solvated electrons with protons, for which a rate constant of about 5 x 10n s_1 is expected in this medium. Following these processes, a new, broad band with 2max = 385 nm is formed. At the same time, the absorption band at 525 nm, due to 31, shifts to 500 nm and is reduced in intensity. The rate constant of this process, determined from a biexponential fit to the series of spectra in the range of 360—430 nm, where interference by the absorption of solvated electrons is negligible, is k.2 « 3 x 1010 s_1. The species formed here is assigned to the triplet state of carbonyl-protonated benzophenone, 31H+.20 At delay times > 175 ps, the 385 nm absorption band due to 31H+ decays almost completely, leaving a similar end spectrum as in Figure 3. The rate constant of the last reaction is £3 « 1.2 x 109 s~\ 4.0 r/ps Figure 4. Loading coefficients c (•) of the first eigenvalue determined by factor analysis of the spectral matrix (450—550 nm) that is shown in Figure 3. The solid line shows the result of a global nonlinear least-squares fit of a triexponential function. The corresponding fit parameters (first-order rate constants) are given in the text. The inset is an expanded view of the first 180 ps. Pump—probe spectra determined with 5.8 M HCIO4 solutions of 1 are shown in Figure 5. Again, ISC of !1 to 31 with the 3308 /. Phys. Chem. A, Vol. 107, No. 18, 2003 Ramseier et al. 300 Figure 5. Pump—probe transient absorption spectra of 1 (7 x 10 4 M) in 5.8 M aqueous HCIO4 detected with delays of 5—100 ps19 relative to the excitation pulses (248 nm, 0.5 ps half-width, 1 mJ per pulse). TABLE 1: Observed First-Order Rate Constants ki—k5 in Air-Saturated Aqueous HCIO4 (Room Temperature, 22 ± 2 °C)" [H+] (M) pH or H0b yfc, (1011 s-'c) k2 (lO^s-'c) h (108 s" -V) kA (107 s-'c) fe (s-'c) 0 neutral 1.61 ± 0.05 (2) (7.3 ± 0.1) x 10-3(8) 0.001 3.0 (1.16 ±0.02) > c 10~2 (7) 0.005 2.3 (2.71 ±0.06) > c 10~2 (3) 0.154 0.008 2.1 (3.65 ± 0.02) > c 10-2 (5) 0.01 2.0 (4.57 ± 0.04) > c 10-2 (8) 0.02 1.7 (9 ± 1) x 10~2 (2) 0.03 1.5 (1.13 ±0.05) > c 10"1 (3) 0.05 1.3 (1.28 ±0.04) > c 10-' (7) 0.10 0.97 (2.2 ± 0.3) x 10"1(4) 0.156 ±0.003 (4) 0.20 0.65 2.2 ±0.1 (4) 0.50 0.18 2.0 ±0.1 (6) 1.0 -0.25 1.2 ± 0.2 (2) 5 ± 1 (4) 1.9 ±0.1 (6) 0.360 ± 0.002 (3) 1.5 -0.56 9 ± 2 (2) 2.0 -0.82 1.6 ± 0.2 (3) 1.0 ±0.1 (3) 11 ±2(3) 0.399 ± 0.003 (4) 2.5 -1.1 1.4 ±0.3(1) 12 ± 1(1) 3.0 -1.3 1.7 ± 0.2 (3) 1.3 ±0.2 (3) 10.9 ± 0.7 (3) 2.2 ±0.1 (6) 0.53 ± 0.01 (4) 4.0 -1.7 1.8 ± 0.9 (2) 3.0 ± 0.2 (2) 12 ± 2 (2) 2.4 ±0.1 (6) 1.42 ±0.01 (4) 5.0 -2.22 2.88 ± 0.08 (5) " Solutions used for pump—probe measurements contained 10% acetonitrile. b Values of the Ho acidity function (ref 18) are given for [H+] > 0.1 M. c Number of measurements given in brackets. associated spectral changes (2max = 335 and 575 nm shifting to 330 and 525 nm) is the first process, k\ « 8 x 1010 s_1. Almost as fast, hi « 5 x 1010 s_1, is the formation of a broad absorption band with 2max = 385 nm, which is attributed to adiabatic protonation of 31 at the carbonyl oxygen yielding 31H+. Further small changes, which cannot be assigned, occurred at longer delays. Nanosecond Flash Photolysis (LFP). The transient absorption spectrum of triplet benzophenone (31) obtained by 248 nm excitation of 1 (4 x lO-5 M) in neutral water (2max = 330 and 525 nm, not shown) was in excellent agreement with that obtained at the maximum delay (1.8 ns) in the pump—probe experiment (Figure 2). The decay of 31 was dominated by oxygen quenching and obeyed the first-order rate law accurately. An average rate constant k$ = (7.3 ± 0.1) x 105 s~] (air-saturated solutions) was obtained from eight measurements at different wavelengths in the range of 315—600 nm. The decay rate of 31 increased linearly with increasing acid concentration (fc,, Table 1). Transient spectra obtained by LFP of 1 (4 x lO-5 M) in 1.0 M HCIO4 are shown in Figure 6, together with two of the traces obtained by pump—probe spectroscopy. The transient spectrum obtained immediately after the nanosecond laser pulse (marked "10 ns") is attributed to the triplet state of the hydration product, 31-H20.20 Comparison with the pump—probe spectra recorded with delays of 176 and 1826 ps in the same medium indicates that the conversion of 31 to 31-H20 is not complete within 1826 ps (cf. Figures 3 and 4). Kinetic measurements by nanosecond LFP at various wavelengths from 315 to 520 nm gave an average rate constant of fct = (1.9 ± 0.1) x 107 s~! for the decay of 31-H20. The same initial transient spectra and decay rate constants were observed by LFP of aqueous solutions containing from 0.20 to 4.0 M HCIO4 (Table 1). The temperature dependence of the rate constant fct was measured with 1 M HCIO4 solutions (T = 274-335 K, 22 measurements). An Arrhenius plot of these data was linear and gave the activation parameters A = (1.6 ± 0.4) x 108 s"1 and £a = 1.13 ± 0.10 kcal mol-1. The transient spectrum generated by LFP of 1 in 5.8 M HCIO4 is substantially different (Figure 7). The initial spectrum agrees well with that observed at maximum delay by pump—probe spectroscopy with the same solvent (Figure 5). It also agrees closely with the initial transient spectrum recorded by LFP of 1 in a 0.5 M solution of HCIO4 (70%) in acetonitrile (not shown) Photohydration of Benzophenone in Aqueous Acid /. Phys. Chem. A, Vol. 107, No. 18, 2003 3309 Alnm 300 400 500 600 J_■ ■ 1_I_I_I_I_I_I_I-1_I-1-L 3.0 2.5 2.0 M^m-1 Figure 6. Transient absorption spectra obtained with 1 in 1 M HCIO4. The two upper traces (left-hand absorbance scale) were obtained by pump-probe spectroscopy, and the two lower ones (right-hand absorbance scale) were obtained by LFP. Delay times relative to the excitation flash at 248 nm are shown. 300 400 500 600 3.0 2.5 2.0 Figure 7. LFP of 1 (4.9 x 10~5 M) in 5.8 M aqueous HCIO4. Delay times relative to the excitation pulse at 248 nm (15 ns, 150 mJ) are shown. and is attributed to protonated triplet benzophenone, 31H+. It decays with a first-order rate constant of (8.7 ± 0.4) x 106 s_1. Similar results were obtained with 6.0 M HCIO4. In this medium, the transient decays with a rate constant of (5.6 ± 0.1) x 106 s"1. LFP of 1 in 12 M HCIO4, where it is protonated in the ground state (Figure 1), gave weak, broad transient absorption with maxima at about 390 and 500 nm, which is attributed to 31H+. In this medium, bleaching of the ground state absorption of 1H+ hides the 320 nm band of 31H+ (Figures 5 and 7). Transient 31H+ decays with a first-order rate constant of (1.6 ± 0.1) x 105 s"1 in air-saturated, 12 M HCIO4. Conventional (ms) Flash Photolysis. Excitation of 1 in aqueous acid with a discharge flash lamp (up to 1000 J electrical energy, 30 /is half-width) produced a long-lived transient absorbing below 350 nm, 2max = 315 nm, which is attributed to l-FLO.20 The same absorption was left after the decay of the transient 31-H20 in the spectra determined by nanosecond LFP (Figure 6, 500 ns delay). A dilution experiment ([1] = 1-20 x 10~6 M) in 0.1 M HCIO4 proved that the amount of 1-H20 was linearly related to the amount of light absorbed, i.e., that the yield of l-FLO did not depend on the concentration of 1. The lifetime of 1-H20 was t = 5.4 ± 1.0 s in 0.1 M aqueous HCIO4. Formation of this transient requires both water and acid. No such transient was formed in solutions of 1 in acetonitrile that was acidified with HC1 gas or concentrated HCIO4. The yield of l-FLO increased with acid concentration in dilute, air-saturated aqueous HCIO4 (0.001-0.10 M). The inverse of the initial absorbance Ao at 310 nm was linearly related to the inverse of [H+]. Linear regression (eight data points) gave an intercept of 30.0 ±0.1 and a slope of (5.86 ± 0.03) x 10~2 M. Upon further increase of the acid concentration, Ao reached a maximum value at about 1 M HCIO4 and decreased 3310 /. Phys. Chem. A, Vol. 107, No. 18, 2003 Ramseier et al. TABLE 2: Results of B3LYP/6-31G* Calculations compd E (hartree) ZPE (hartree) H (298 Kf (hartree) S° (298 K) (cal KT1 mor1) AH° (298 K) (kcal mol-1) o-l-HjO o-3l-H20 m-l-H20 m-3l-H20 p-i-n2o P- 31-H20 -653.007318 -652.953977 -652.971934 -652.952737 -652.999717 -652.941476 0.219528 0.216425 0.217936 0.215558 0.218711 0.214545 " Thermal enthalpy correction including zero-point energy. 0.232584 0.229886 0.231323 0.229338 0.232280 0.228889 108.7 112.8 110.3 114.5 111.7 119.9 = 0 31.8 21.4 32.2 4.6 39.0 at still higher concentrations. The transient was no longer detectable at acid concentrations of 6 M or higher. The temperature dependence of the decay of I/H2O in 0.1 M HC104 was determined in the range of 25-57 °C (10 data points). Activation parameters _Ea = 18.5 ± 0.2 kcal mol-1 and A = (5.5 ± 1.4) x 1012 s-1 were obtained from an Arrhenius plot. The decay rate constant of 1-H20 in 0.1 M aqueous HCIO4 at —5.9 °C was (3.2 ± 0.4) x 10-3 s-1 (three measurements). Product I/H2O could be extracted to pentane after irradiation of 1 at —6 °C in aqueous HCIO4. Its decay rate constant in pentane, measured by the absorbance decrease with time at 310 nm, was quite similar to that in aqueous solution, k = (3.1 ± 1.6) x 10-3 s-1 (three measurements, —5.9 °C). These experiments show that I/H2O is a neutral compound and that its decay rate is hardly solvent-dependent. However, the decay rate of I/H2O in water increased with acid concentrations exceeding 1 M (Table 1). Photoacoustic Calorimetry. Solutions of 1 and HCIO4 (4 M) in a mixture of acetonitrile (30% by volume) and water were excited at 248 nm. Acetonitrile was added to increase the thermal expansion coefficient of the solvent. Optical LFP of 1 in 4 M aqueous HCIO4 indicated that the intermediate 31-H20 was formed within the duration of the laser pulse and decayed with a lifetime of 42 ± 1 ns. Consistently, the photoacoustic waves indicated that a time-resolved heat deposition process occurred with a similar lifetime. The best fit was obtained with a biexponential function for the heat release. The rate constant for the fast processes, which occur rapidly on the time scale of the duration of the laser pulse, was arbitrarily fixed to 1 ns. Three parameters were adjusted, the two amplitudes of the two heat deposition processes and the rate constant for the time-resolved process. Of the total energy (248 nm => 115.2 kcal per mole of photons absorbed), 57.5 ± 4.3% was released within 1 ns and 34 ± 6% with a lifetime of 45 ± 7 ns. The goodness-of-fit was excellent and was not improved by using three exponentials in the kinetic trial function. Hence, the decay of 31-H20 deposits 39 ± 7 kcal mor1. In a second series of experiments, more dilute acid (0.17 M HCIO4, 30% acetonitrile) was used. Optical LFP of this mixture gave a lifetime of 32 ± 5 ns for 31 and 52 ± 5 ns for 31-H20. On the basis of purely statistical criteria, the fits to the acoustic waves were equally acceptable with a bi- or triexponential function. The fit parameters obtained with a triexponential function were chosen, because that was required by the analysis of the LFP data. The first exponential, formation of 31, was fixed to occur with t = 1 ns and to release 46.2 kcal mol-1 (115.2 kcal mol-1 — Ej(l)). In the delayed heat evolution, 16.5 ± 6.4% of the absorbed energy was deposited with t = 25 ± 10 ns and 30 ± 7% with t = 55 ± 15 ns. Thus, proton-catalyzed adiabatic hydration of 31 yielding 31-H20 affords 19 ± 7 kcal mol-1, and the decay of 31-H20 releases another 34.5 ± 8.0 kcal mol-1. The average of the two measurements (4 and 0.17 M HCIO4) for this process is 37 ± 6 kcal mol-1. No corrections were made for reaction volumes. Density Functional Calculations. The energies of three possible structures of hydrate I/H2O in the ground state and in the lowest triplet state were calculated using the B3LYP density functional with the 6-31G* basis set, as provided in the Gaussian 9824 program package. All structures were fully optimized, and frequency calculations were done. The energies given in Table 2 are those of the most stable conformers (OH bonds) and E/Z isomers shown below. o-l«H20 m-l»H20 p-l«H20 From the last column of Table 2, it is seen that the ortho isomer is the most stable hydrate in the ground state. In the triplet state, the energies of the ortho and the meta isomers are similar, that of the para isomer is higher. The calculated triplet excitation energies (0 K) of the three isomers are £T(ortho) = 31.5, £T(meta) = 10.6, and £T(para) = 33.6 kcal mol-1. The triplet energy of 1 was calculated as Ej = 61.5 kcal mol-1 by the same method. The enthalpy of dehydration (298 K) of o-l-H20 to 1 + H20 is calculated as Arff = -23.8 kcal mol-1. Energy Transfer to Biacetyl. The triplet energy of proto-nated triplet benzophenone, _ET(31H+) = 61.8 kcal mol-1,4 is known from phosphorescence measurements in strongly acidic media. With £T(biacetyl) = 56.4 kcal mol-1,25 it should be possible to observe energy transfer to biacetyl from both 31 and 31H+. The triplet—triplet absorption maxima of biacetyl are at 315 (e « 6000 M-1 cm-1) and 800 nm.26 LFP of 1 (3 x 10-5 M, excitation at 248 nm) in neutral, aerated water with added biacetyl (0—0.01M) showed accelerated first-order decay of 31 at 520 nm and biexponential decay at 325 nm. The fast component observed at 325 nm was attributed to the accelerated decay of 31, and the slow one was attributed to the decay of triplet biacetyl formed by energy transfer. A resolved growth followed by a decay of absorbance was seen at 800 nm, where only triplet biacetyl absorbs. The observed rate of energy transfer increased linearly with increasing biacetyl concentration. The best traces were obtained by monitoring at 520 nm (seven data points, 0—0.05 M biacetyl). A least-squares fit gave ket = (1.0 ± 0.1) x 109 M-1 s-1 for the bimolecular rate constant of energy transfer from 31 to biacetyl. In previous work,5-8 the transient observed by LFP of 1 in 1 M aqueous HCIO4 has been attributed to 31H+. Addition of up to 0.05 M biacetyl failed to accelerate the decay of the transient absorbance at 500 nm, which supports our proposal that this species is not 31H+ but 31-H20. The triplet energies of the three isomers of 31-H20 (Table 2) were calculated to be well below that of biacetyl. To demonstrate energy transfer from 31H+ to biacetyl, the experiment was repeated using acidified acetonitrile (1 M Photohydration of Benzophenone in Aqueous Acid /. Phys. Chem. A, Vol. 107, No. 18, 2003 3311 TABLE 3: Proton Quenching Rate Constants, Transient Absorption Spectra, and Lifetimes for Various Derivatives of 1 and 2 &obs,H+ Amx(3l/2-H20) r(3l/2-H20) Amx(l/2-H20) 77(l/2-H20)'1 compd (lO'M-'s-1) (nm) (ns) (nm) (s) benzophenone (1) 38 ± 1 320, 500 50 315 5.4 ± 1.0 3-methylbenzophenone 9.0 ±0.1 320,510 60 320 22.7 ± 0.8 4-methylbenzophenone 62 ±2 320,510 50 315 8.6 ± 0.5 3,4-dimethylbenzophenone 6.8 ±0.2 <340, 515 50 320 8.0 ± 1.1 3,5-dimethylbenzophenone 7.2 ±0.2 320,510 60 320 22.6 ± 0.9 4-aminobenzophenone 60 ±2 315,505 80 315 8.3 ± 0.8 4,4'-difluorobenzophenone 1.3 ±0.2 320, 500 50 305 0.15 ± 0.02 acetophenone (2) 22 ± 1 280 0.9 ± 0.1 2-chloroacetophenone 52 ± 1 285 3.3 ± 0.5 2-fluoroacetophenone 93 ±2 280 0.05 ± 0.01 4-fluoroacetophenone 1.0 ±0.1 280 0.06 ± 0.01 3-trifluoromethylacetophenone 0.88 ± 0.03 285 0.55 ± 0.03 1 Aqueous 0.1 M HCIO4; derivatives of 2 required addition of 10% acetonitrile, and derivatives of 1 required 20% acetonitrile to achieve sufficient solubility. HCIO4) as a medium, in which hydration of 31H+ does not interfere. Biacetyl was unstable in this medium, and its absorption band at 415 nm decayed with a half-life of 7 min. Thus, biacetyl was added immediately prior to LFP of 1 (1.5 x 1CT5 M). The lifetime of 31H+ (440 ns in air-saturated solution, measured at 315 nm) was reduced to 57 ns by the addition of 5 x 1CT3 M biacetyl, i.e., the rate constant for energy transfer from 31H+ to biacetyl amounts to ket « 3.0 x 109 M_1 s_1 in this solvent. Attempts to Detect 1-H20 by ^ NMR. Several attempts were made to detect the long-lived product LH2O by !H NMR. These experiments were hampered by the low yield of intermediate LH2O27 and by the fact that LH2O was, itself, quite sensitive to irradiation and could not be accumulated to a high extent even at reduced temperature. The best results were obtained by irradiation of 1 (6 x 10~3 M) in acidic D20/CD3OD 1:1 (0.3 M DCIO4) with 50 248-nm laser pulses at -30 °C in a quartz NMR tube. At this temperature, the lifetime of LH2O, monitored by its absorbance at 310 nm, was about 2 h. Following irradiation, the solution was quickly transferred to a cooled (—30 °C) 500 MHz NMR spectrometer, and the spectra were accumulated. Apart from the strong signals due to 1, very weak signals were detected, which disappeared irreversibly on warming. The spectrum is given as Supporting Information. Isotope Exchange. Solutions of 1 in acidic D2O (0.25 M D2SO4) were irradiated with 5000 248-nm pulses from an excimer laser. No deuterium incorporation was detected by gas chromatography/mass spectrometry (GC/MS) and !H NMR analysis of the irradiated sample. The dose was sufficient to excite each solute molecule about 1000 times. LFP of 2 and Derivatives of 1 and 2. A brief study of 2 and several derivatives of 1 and 2 by LFP indicated that the triplet state of these compounds is also quenched by protons. The observed bimolecular quenching rate constants fc0bs,h+ are collected in Table 3 and interpreted in the discussion (see eq 7). Acid quenching of 32 did not produce a transient similar to 31-H20, only a weak, long-lived transient, 2max « 280 nm, t = 0.9 s, presumably due to the ortho-hydrate 2-H20. The lifetime of the triplet hydrate formed from 2,32-H20, appears to be less than 20 ns. m-Fluoroacetophenone in 1 M HCIO4 did produce a strong transient absorption, 2max ^ 350 nm, t = 50 ns. Permanent absorbance was left in the range of 300—350 nm, ^■max = 308 nm, after the decay of this transient (cf. next section). As observed with LH2O, the yield of 2-H20 increased with acid concentration in dilute aqueous HCIO4. Linear regression of 1/A0 (280 nm) vs 1/[H+] ([H+] = 0.001-0.10 M) gave an intercept of 4.40 ± 0.05 and a slope of (1.69 ± 0.05) x 10~2M. SCHEME 1: Photohydrolysis of 3,4-Difluoroacetophenone in Aqueous Acid hv H+/H20 ■ HF Preparative Work and Quantum Yields of Photohydrolysis. Irradiation of 3-fluoroacetophenone in aqueous acid produced a new absorption band at 305 nm. Addition of excess base shifted the absorption maximum of the photoproduct to 343 nm. The shapes and maxima of these spectra were nearly identical to those of authentic 3-hydroxyacetophenone: 2max = 308 nm in aqueous acid, 348 nm in base. This indicated essentially clean photosubstitution of the fluoride. Similarly, irradiation of 3,3'-difluorobenzophenone produced a new band at 310 nm (370 nm in base). The photoproduct formed by irradiation of 3,4-difluoroacetophenone (Scheme 1) in aqueous acid was identified as 4-fluoro-3-hydroxyacetophenone (Experimental Section). Quantum yields of photohydrolysis were measured spectro-photometrically at low conversions in degassed 0.1 M aqueous HCIO4, assuming clean conversion to the corresponding phenols. They were high for derivatives carrying a meta-fluoro substitu-ent:

< 0.01. Discussion pH Rate Profiles and Assignment of Transient Intermediates. Five different transient intermediates are formed by excitation of 1 in aqueous acid. They were identified by their absorption spectra and assigned to the intermediates !1, 31, 31H+, 31-H20, and 1-H20 as shown in Scheme 2. The observed transient kinetics gave five first-order rate constants, k\—k5, Table 1. A reliable determination of five rate constants 3312 /. Phys. Chem. A, Vol. 107, No. 18, 2003 Ramseier et al. SCHEME 2: Postulated Reaction Sequence for the Photohydration of 1 in Aqueous Acid" 1 1«H20 31«H20 " The structure shown for 31is one of several possible isomers (cf. Scheme 3 below). TABLE 4: Absorption Maxima (Relative Intensities) of the Five Transient Intermediates Formed by Flash Photolysis of 1 in Aqueous Acid assignment Amax (nm) (relative intensity) spectrum 31 31H+ 31-H20 1-H20 335 (1), 575 (ca. 0.3) 330 (1), 525 (ca. 0.9) 320 (1), 385 (ca. 0.8), 500 (ca. 0.7) 320(1), 500 (ca. 0.2) 315 Figures 2, 3, and 5 (initial spectra) Figure 2 (last spectrum) Figure 7 (10 ns delay) Figure 6 (10 ns delay) broad Gaussian band (not shown) from transient absorption measurements is rarely possible. The present analysis was feasible, because the observed processes have quite different lifetimes spanning a time scale of 12 orders of magnitude and because the rates of these processes smoothly followed the pH dependencies expected from Scheme 2. Identification of the first two intermediates, !1 and 31, is straightforward. The initial absorption bands recorded immediately after excitation with a subpicosecond pulse at 248 nm (Figures 2, 3, 5; Table 4) are attributed to Si—Sx transitions of the excited singlet state of 1, !1. The visible absorption band of !1 (2max = 575 nm in acetonitrile) has been observed previously.28-29 The first absorption changes are due to ISC of '1 yielding the well-known triplet—triplet absorption of ben-zophenone, 31, 2max = 330 and 525 nm.4-26 The lifetime of !1 in aqueous solution is 6.5 ps. The ISC quantum yield of 1 is very close to unity,25 so that the observed rate constant of this process, kx = (1.55 ± 0.09) x 101 may be equated to the rate constant of ISC to the triplet state, kisc- In aqueous acid (>1 M), a small amount of protonation occurs in competition to ISC of l\. The adiabatic formation of !1H+ was detected by fluorescence spectroscopy (Figure 1). The lifetime of l\ in acetonitrile has been determined previously: h = 9 ± 228 and 9.6 ± 0.9 ps.29 Singlet-triplet ISC to the n,jr* triplet state of 1 appears to be accelerated by H-bonding. In remarkable contrast, H-bonding solvents reduce the rate of ISC to the jt,jt* triplet state of p-hydroxyac-etophenone.30 The similarity between the jt,jt* absorption spectra of 31 and benzophenone ketyl radical has been noted long ago.31 It was attributed to the fact that both species have the same Tt system bearing a single unpaired electron. For the same reason, the exited singlet state !1 also has a similar absorption spectrum. Assignment of the next three transients, the protonated triplet of 1, 31H+, its hydration product, 31-H20, and the long-lived ground state intermediate l-F^O, requires more detailed argumentation. We begin with an analysis of the kinetic system defined by Scheme 2. The proposed mechanism comprising five reactive intermediates and the end product 1 gives a set of six coupled differential equations. As protons are in large excess relative to the amount of the transient intermediates formed, [H+] may be assumed to be constant during the transient decays, and the products fcH+[H+] and fcy+[H+] can be treated as first-order rate constants. All reactions are then first order, which results in a rate law for the time-dependent absorbances A consisting of five observable exponentials, eq 1. A(l,f) = A0(l,o i=l,5 (1) The relations between the observable rate constants k\—k5 and the microscopic rate constants shown in Scheme 2 were determined by standard methods of linear algebra,32 eqs 2—6. ct + Va1 4§_ a- 2), kobsii+, is lower than the rate constant for protonation of 31 by the factor/ = ko/(ka + k-u+) = 0.45 ± 0.10, which represents the fraction of 31H+ that is quenched by water addition in dilute acid. Least-squares fitting of eq 7 gave kobsM+ = (3.8 ± 0.1) x 108 M_1 s"1, in reasonable agreement with the less accurate value of (6.4 ± 1.3) x 108M- that Porter and Ledger3 had determined by phosphorescence quenching. The quantum yield for the formation of 31-H20 in dilute acid is then given by hydr = &obs,H+ 3314 /. Phys. Chem. A, Vol. 107, No. 18, 2003 Ramseier et al. SCHEME 3: Tentative Assignment of Intermediates 31-H20 and 1-H20 3r , 3r + H20,-H+ 5nr fast OH OH 3m-l'H20, major 3o-l*H20, minor -H20 £-h20 -H,0 [H+l/7(/tobs,h+[H+l/'+ k'l&c). Hence, the inverse yield of 1-H20 should be linearly related to the inverse of [H+] in dilute acid, eq 8. The initial absorbance, Ao (315 nm), by the long-lived transient 1-H20 accurately obeyed eq 8 for acid concentrations in the range of 0.001-0.10 M. 1/A0 « a + ak'lscJkobsiI+ [H+] (8) The ratio of slope and intercept of eq 8 (determined by linear regression; see Results, Conventional Flash Photolysis), k'lsc/kohsM+ = (1.95 ± 0.01) x 10~3 M, is in excellent agreement with the ratio of the individual rate constants k'isc (Table 5, footnote a) and fc0bs,h+ (Table 3) determined independently, k'lsc/kobsM+ = (1.92 ± 0.06) x 10~3 M. The quantitative agreement between these two results confirms that transient 1-H20 results from proton quenching of 31. Similarly, k'lsc/kobSili+ = (3.84 ± 0.12) x 10~3 M was obtained from the ratio of slope to intercept of the corresponding plot for the initial absorbance Ao (280 nm) by the transient 2-H20 and fcJsc/fc0bs,h+ = (3.4 ± 0.3) x 10~3 from the individual rate constants. This result corroborates our assumption that the same mechanism (Schemes 2 and 3) applies for 2, although 32-H20 escaped detection by nanosecond LFP. Acidity of 1H+ in the Electronic Ground State and in the Lowest Excited Singlet and Triplet States. The ratio of the rate constants for adiabatic dissociation of 31H+, k-a+, and protonation of 31, ka+, provides the acidity constant of 1 in the excited triplet state: ^a(3lH+) = fc-H+/fcH+ = 2.65 ± 0.66, pK„ (31H+) = -0.4 ±0.1. The rate constants k-a+ and ka+ (Table 5) were determined from the kinetic data obtained in strongly acidic solutions, Ho = —0.25 to —1.7. The resulting acidity constant is, strictly, a concentration quotient determined over a substantial range of (high) ionic strengths. However, dissociation of 31H+ to 31 and H+ is a charge shift reaction, which should not depend much on ionic strength. Both k-a+ and ka+ are substantially slower than expected35 for proton transfer between two oxygen atoms, i.e., "normal" acids and bases. This unusual finding may be attributed to the fact that the electronic wave functions of 31 (3n,jr*) and of 31H+ (%,jr*) have different symmetry (assuming planar structures), so that adiabatic proton transfer to 31 encounters a symmetry-imposed barrier. Depending on the acidity function used, values for the acidity constant of protonated benzophenone ranging from pK^ (1H+) = —6.2 to —4.7 have been reported. We adopt the more recent estimates based on the Ho acidity function pK^ (1H+) = —4.745 and —4.71.36 The acidity constant of 1H+ in the lowest excited singlet state may then be the estimated by the Förster cycle, pKz (11H+) s pK^ (1H+) + Avo-o/(2.3RT). Using p^a (1H+) = -4.7,36 vo-o (!1) = 26 040 crrr1,25 and v0-o (!1H+) = 24 250 crrr1 estimated from Figure 1, we obtain pK^ (11H+) s -1. On the basis of the older value pK^ (1H+) = —5.7 and somewhat different estimates of the excitation energies vo-o, Ireland and Wyatt4 estimated p^a (11H+) s -3.6. We detected adiabatic protonation of !1 by fluorescence emission from !1H+ following excitation of 1 in solutions with acid concentrations exceeding 1 M. This indicates that the latter estimate is low. Comments on Previous Studies. Favaro and Bufalini6 have studied the pH dependence of energy transfer from 31 to biacetyl by phosphorescence. By steady state analysis of the phosphorescence intensities, they obtained a lifetime of 2.3 x 10~4 s for 31 in degassed, neutral, aqueous solution (extrapolated to infinite dilution of 1) and a lifetime of ca. 40 ns for the triplet species formed in acidic solution (pH < l).37 That species was attributed to an "excited complex" different from protonated benzophenone in the triplet state, 31H+. This assignment comes remarkably close to that made in the present study. It was based on the previous38 observation that the phosphorescence emission at 77 K from frozen aqueous solutions of pH 3 to H0 -4 was different from that of frozen 70% HC104 or 96% H2SO4, where protonation of 1 to 31H+ occurs in the ground state. However, Rayner and co-workers7 subsequently showed that the differences of the phosphorescence spectra noted by Favaro and Bufalini cannot be taken as evidence for a benzophenone—hydroxonium complex, because similar changes were found by addition of sodium acetate. The shift was attributed to an effect of the additive (acid or salt) on the frozen solvent structure. Both Wyatt and co-workers4-5 and Shizuka and co-workers8-11 have assumed that the transient species observed after protonation of 31 in dilute aqueous acid is the conjugate acid 31H+. The analysis of their data should be revised accordingly. In particular, the acidity constant is 2 orders of magnitude lower than previously estimated,5 pK^ (31H+) = —0.4 ±0.1. Conclusion Rapid water addition to the triplet state of protonated benzophenone (31H+) is responsible for the hitherto unexplained quenching of aromatic ketone triplets by protons in water. The rate constant for protonation of 31 is well below the diffusion-controlled limit, and this is attributed to a state symmetry-imposed barrier for adiabatic proton transfer to 31 (n,jr*) yielding 31H+ (jt,jt*). The acidity constant of 31H+ is revised to pK^ (31H+) = -0.4 ± 0.1. Up to [H+] < 1 M, protonation of 31 is rate-limiting for adiabatic hydration of 31H+, which yields predominantly the meta-hydrate 31-H20. The photohydration Photohydration of Benzophenone in Aqueous Acid /. Phys. Chem. A, Vol. 107, No. 18, 2003 3315 reaction is reversible for parent 1 and 2. Derivatives of 1 and 2 carrying fluoride as a leaving group in the meta position form the corresponding m-hydroxy derivatives cleanly and efficiently, constituting a hitherto unexplored, efficient aromatic photosub-stitution reaction. Acknowledgment. This work was supported by the Swiss National Science Foundation, Project Nos. 20-61746.00 and 2160-064525.01/1. We thank Sonja Mayer for skillful experimental support. Supporting Information Available: HNMR of benzophenone in irradiated solution. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Moore, W. M.; Hammond, G. S.; Foss, R. P. J. Am. Chem. Soc. 1961, 83, 2789-2794. (2) Tolbert, L. M.; Solntsev, K. M. Acc. Chem. Res. 2002, 35, 19-27. (3) Ledger, M. B.; Porter, G. J. Chem. Soc, Faraday Trans. 1 1972. 68, 539-553. (4) Ireland, J. F.; Wyatt, P. A. H. J. Chem. Soc, Faraday Trans. 1 1973, 69, 161-168. (5) Rayner, D. M.; Wyatt, P. A. H. J. Chem. Soc, Faraday Trans. 2 1974, 70, 945-954. (6) Favaro, G; Bufalini, G. J. Phys. Chem. 1976, 80, 800-804. (7) Rayner, D. M.; Tolg, P. K.; Szabo, A. G. J. Phys. Chem. 1978, 82. 86-89. (8) Shizuka, H.; Kimura, E. Can. J. Chem. 1984, 62, 2041-2046. (9) Bensasson, R. V.; Gramain, J.-C. J. Chem. Soc, Faraday I1980. 76, 1801-1810. (10) Lougnot, D. J.; Jacques, P.; Fouassier, J. P.; Casal, H. L. Kim-Thuan, N; Scaiano, J. C. Can. J. Chem. 1985, 63, 3001-3006. (11) Hoshi, M.; Shizuka, H. Bull. Chem. Soc Jpn. 1986, 59, 2711-2715. (12) Elisei, F.; Favaro, G.; Gorner, H. J. Photochem. Photobiol. A: Chem. 1991, 59, 243-253. (13) Canonica, S.; Hellrang, B.; Wirz, J. J. Phys. Chem. A 2000, 104. 1226-1232. (14) Hasler, E.; Hormann, A.; Persy, G; Platsch, H.; Wirz, J. J. Am. Chem. Soc 1993, 115, 5400-5409. (15) Senn, P.; Wirz, J. Res. Chem. Intermed. 1995, 21, 877-883. (16) Ni, T.; Caldwell, R. A.; Melton, L. A. J. Am. Chem. Soc. 1989. Ill, 457-464. (b) Melton, L. A.; Ni, T.; Lu, Q. Rev. Sci. Instrum. 1989. 60, 3217-3223. (c) Arnaut, L. G.; Caldwell, R. A.; Elbert, J. E.; Melton. L. A. Rev. Sci. Instrum. 1992, 63, 5381-5389. (17) Gauglitz, G; Hubig, S. J. Photochem. 1985, 30, 121-125. (18) Kresge, A. J.; Chen, H. J.; Capen, G. L.; Powell, M. F. Can. J. Chem. 1983, 61, 249-256. (19) The quoted delays relative to the subpicosecond excitation pulse refer to the blue end of the spectrum (300 nm). Red-shifted colors of the probe continuum reach the sample with smaller delays (chirping, ca. —1 ps delay per 80 nm shift), i.e., a quoted delay of 5 ps corresponds to a delay of 1.5 ps at 600 nm. Colors arriving simultaneously with the pump pulse are depleted by two photon absorption. (20) Arguments for the assignment of transient intermediates are given in the Discussion. (21) Gampp, H.; Maeder, M.; Meyer, C. J.; Zuberbühler, A. D. Talanta 1985, 32, 95-101. (22) Buxton, G. V.; Helman, W. P.; Ross, A. B. J. Phys. Chem. Ref. Data 1988, 17, 513-886. (23) Jonah, C. D.; Miller, J. R.; Matheson, M. S. J. Phys. Chem. 1977. 81, 931-934. (24) Frisch, M. J.; Tracks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G; Montgomery, J. A., Jr.; Stratmann, R. E.; Burant, J. C; Dapprich, S.; Millam, J. M.; Daniels, A. D; Kudin, K. N; Strain, M. C; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C; Adamo, C; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C; Challacombe, M.; Gill, P. M. W.; Johnson, B. G; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98; Gaussian, Inc.: Pittsburgh. PA, 1998. (25) Murov, S. L.; Carmichael, I.; Hug, G. L. Handbook of Photochemistry, 2nd ed.; Dekker: New York, 1993. (26) Carmichael, I.; Hug, G. L. J. Phys. Chem. Ref. Data 1986, 15. 1-250. (27) Assuming a molar absorption coefficient e of ca. 30 000 M~' cm-1, we estimate a quantum yield of <10% for the formation of 0-LH2O in aqueous acid. (28) Miyasaka, H; Morita, K.; Kamada, K.; Mataga, N. Bull. Chem. Soc. Jpn. 1990, 63, 3385-3397. (29) Tamai, N; Asahi, T.; Masuhara, H. Chem. Phys. Lett. 1992, 198. 413-418. (30) Pelliccioli, A.-P.; Klán, P.; Zabadal, M.; Wirz, J. J. Am. Chem. Soc 2001, 123, 7931-7932. (31) Godfrey, T. S.; Hilpern, J. W.; Porter, G. Chem. Phys. Lett. 1967. 490-492. (32) Elmer, F.-J. Differentialgleichungen in der Physik; Verlag Harri Deutsch: Frankfurt am Main, 1997. (b) Mauser, H; Gauglitz, G. Photo-kinetics, Theoretical Fundamentals and Applications; In Chemical Kinetics; Compton, R. G, Hancock, G., Eds.; Elsevier: Amsterdam, 1998; Vol. 36. (33) Shizuka, H. Acc. Chem. Res. 1985, 18, 141-147. (34) Boyd, D. R.; McMordie, R. A. S.; Sharma, N. D.; More O'Ferrall. R. A.; Kelly, S. C. J. Am. Chem. Soc. 1990, 112, 7822-7823. (b) Rao, S. N.; More O'Ferrall, R. A.; Kelly, S. C; Boyd, D. R.; Agarwal, R. J. Am. Chem. Soc. 1993, 115, 5458-5465. (c) Dey, J.; O'Donoghue, A. C; More O'Ferrall, R. A. J. Am. Chem. Soc. 2002, 124, 8561-8574. (35) Eigen, M. Angew. Chem. 1963, 75, 489-508. (36) Bagno, A.; Lucchini, V.; Scorrano, G. J. Phys. Chem. 1991, 95. 345-352. (37) The case for energy transfer to biacetyl from the short-lived species in the data of Favaro and Bufalini is weak. We were unable to observe energy transfer from 31H+ by transient absorption. (38) Favaro, G. Chem. Phys. Lett. 1975, 31, 87-90. (b) Favaro, G: Romani, A. Chem. Phys. Lett. 1991, 184, 596-598.