Current and future applications of nanoclusters
Günter Schmid,* Monika Bäumle, Marcus Geerkens, Ingo Heim, Christoph Osemann and
Thomas Sawitowski
Institut für Anorganische Chemie der Universität Essen, Universitätsstrasse 5-7, D-45117 Essen,
Germany
Received 9th November 1998
This article deals with some recent developments in metal
and in semiconducting nanocluster science. Our studies on
the properties, mainly of metal nanoclusters, with respect to
future and also to current applications are reviewed,
including a series of unpublished results. The general
properties of metal clusters of one up to a few nanometers
G¨unter Schmid was born in Villingen, Germany, in 1937 and
studied chemistry at the University of Munich. He received his
Diploma in 1962 and his Doctor's Degree in Inorganic
Chemistry in 1965, both at the University of Munich. In 1966 he
moved to the University of Marburg, Germany, and finished his
Habilitation in 1969. In 1971 he received a professorship at the
University of Marburg and then he moved to the University of
Essen, Germany, where he became the director of the Institute
for Inorganic Chemistry (1977). His main research interests
include the synthesis and investigation of large transition metal
clusters and colloids, the generation of three-, two- and one-
dimensional arrangements of quantum dots, and heterogeneous
catalysis.
Thomas Sawitowski was born in Essen, Germany, in 1965. He
worked for 12 years as an industrial chemist in the department
for research and development of Th. Goldschmidt AG, Essen.
After studying chemistry at the University of Essen he left
industry and obtained his diploma in 1996. Currently he is
finishing his doctoral thesis under the supervision of Professor
G¨unter Schmid, dealing with the preparation and properties of
alumina nanocomposites.
Christoph Osemann was born in Haltern, Germany, in 1967.
After an apprenticeship and working for three years as a
laboratory assistant he left the catalytic research group of H¨uls
AG, Marl, Germany, and studied chemistry at the University of
Essen. He obtained his diploma in 1997 and is currently
preparing a doctoral thesis in the field of bimetallic clusters
under the supervision of Professor G¨unter Schmid.
Ingo Heim was born in K¨oln, Germany, in 1968. He studied
chemistry at the University of K¨oln and received his diploma in
1997. At present he is preparing a doctoral thesis in the field of
semiconductor nanoparticles under the supervision of Pro-
fessor G¨unter Schmid.
Marcus Geerkens was born in Wuppertal, Germany, in 1970.
He studied chemistry at the University of Essen and received his
Diploma in 1998. At present he is preparing a doctoral thesis in
the field of semiconductor nanoparticles under the supervision
of Professor G¨unter Schmid.
Monika B¨aumle was born in Bad S¨ackingen, Germany, in 1968.
After an apprenticeship as Diet-assistant in Bad Hersfeld,
Germany, she studied chemistry at the University of Essen and
received her diploma in 1996. At present she is preparing a
doctoral thesis in the field of metal cluster synthesis and
application under the supervision of Professor G¨unter
Schmid.
G¨unter Schmid Thomas Sawitowski
Christoph Osemann Ingo Heim Marcus Geerkens Monika B¨aumle
Chem. Soc. Rev., 1999, 28, 179­185 179
are discussed on the basis of numerous physical investiga-
tions in the course of the last few years. Quantum size effects
open the door to novel future technologies. The success of
future applications of nanoclusters will strongly depend on
the availability of three-, two- or one-dimensionally organ-
ized materials. Our own very recent results are promising,
but also indicate that much more work will be necessary.
Much more realistic is the use of metal nanoclusters in
heterogeneous catalysis. Finally, novel developments in
generating semiconducting nanomaterials in transparent,
nanoporous alumina membranes are discussed. CdS and
GaN can easily be prepared inside the pores to give
photoluminescent foils of unlimited size.
1 Introduction
The term `nanocluster' is used to name particles of any kind of
matter, the size of which is greater than that of typical
molecules, but which is too small to exhibit characteristic bulk
properties. The special nature of such nanoclusters, whether
consisting of atoms or composed of building blocks, is to be
traced back to a quantum confinement of electrons leading to a
change of the relevant properties compared to the bulk. Even
common materials such as water or carbon change their
behaviour if they become small enough: the stability of
buildings at temperatures below 0 °C is guaranteed by a
decisive decrease of the freezing point of water in the nanopores
of cement, and the extraordinary behaviour of slices of graphite,
too small to exist as a usual elementary modification, is ending
up in fullerenes and nanotubes, having initiated a completely
novel branch in chemistry and in material science.
Particles of metals and semiconductors in a size-regime
where the wavelength of the electrons is of the same order as the
particle size itself are of extraordinary interest because they
behave electronically as zero-dimensional quantum dots. That
means that the laws of classical physics, valuable for bulk
materials, have to be substituted by quantum mechanical rules.
The transition from a bulk to a nanosized material is best
explained by the sketch in Fig. 1 where the electronic situation
in three different particles of metal atoms is illustrated. The
situation in (a) is that of a quasi-delocalized electronic state of
overlapping valence (VB) and conductivity bands (CB) as is the
case in a metallic bulk situation. In a semiconductor there exists
a gap between both bands, the size of which determines the
energy of electron transitions to achieve conductivity. On the
way from (a) to (b) in the case of a metal the overlap of VB and
CB becomes continuously smaller, finally ending up as a band
gap, too. Fig. 1(b) illustrates the extreme situation where the
particle diameter d corresponds with l/2 (l = de Broglie
wavelength) in the ground state. This energy state can be
compared with the s orbital of a giant metal atom (n = 0) which
can be occupied by two electrons. The first excited state (n = 1)
then corresponds to an atomic p orbital, etc. In (c), bonding
(MOb) and antibonding (MOab) molecular orbitals characterize
the localized bonds between a few atoms in a molecular
cluster.
This article will deal exclusively with particles of type (b),
either consisting of metal atoms or semiconducting materials.
The difference between both is only of quantitative, and not of
principal nature. In practice, quantum size effects are to be
expected even if the ultimate situation (b) is not reached, but
already in larger particles where d corresponds with some
multiples of the electronic wavelength. The investigations of a
series of nanoclusters in the last decade showed that it is
reasonable to expect quantum confinement in metal particles
between 1 and 10 nm, whereas nanoclusters of semiconductors
show quantum size behavior at larger sizes due to the different
conditions in the bulk.
It can be predicted that the very special situation in (b) or in
particles of similar size, if `metallic' or `semiconducting',
should enable manifold applications on very different fields of
science and technology. Some of them which have been the
object of our own interest will be described in the following.
Many others, of course, are under investigation by other groups.
Before discussing possible practical aspects of nanoclusters, a
first chapter will deal with the detection of quantum size effects
in nanoclusters. Metal clusters are exclusively considered in
connection with this, because semiconductor nanoclusters have
not been so intensively studied by ourselves on the one hand and
on the other hand because very good reviews have been
published.1
2 Quantum size effects in metal nanoclusters
To study quantum size behaviour of nanoclusters there is one
condition to be fulfilled: the particles must be separated from
each other to avoid coalescence and to keep the individual
nature of the particles. The availability of numerous ligand
protected metal clusters indeed enabled the systematic in-
vestigation of the properties of nanoclusters. Synthetic aspects,
however, will not be considered in this article. They are
described in detail in previous papers.2­6
The most frequently investigated cluster type is
Au55(PPh3)12Cl6 and some of its derivatives. This so-called
two-shell cluster (a central atom is embedded by two closely
packed shells consisting of 12 and 42 atoms, respectively) is of
special interest as its size of 1.4 nm (without ligands) seems to
represent the borderline between the situations (a) and (c) in Fig.
1.
The generation of hot electrons in gold clusters of different
size by femtosecond laser pulses allowed the observation of
their relaxation behavior.7 The relaxation behaviour of excited
electrons is determined by two opposite effects: the electron­
phonon coupling on the one hand, decreasing with decreasing
cluster size and, on the other hand, the collision rate on the
cluster surface, increasing with decreasing cluster size. For
small particles the latter process is dominating. So, on changing
from ca. 15 nm gold particles to the 1.4 nm Au55 cluster the
electronic relaxation increases characteristically, whereas a
Au13 ( ~ 0.7 nm) cluster shows a considerable decrease of the
relaxation compared with the 1.4 nm cluster. This can only
mean than on changing from Au55 to Au13, there is an expressed
transition from a nanosized metal to a molecule.
Impedance measurements on Au55(PPh3)12Cl6 clusters in a
densely pressed pelleted form gave valuable information on the
intrinsic conduction behaviour of single clusters and also on that
between the perfectly packed grains.8­10 Following these results
Fig. 1 Illustration of the transition of a bulk metal via a nanocluster to a
molecule. The metallic band structure in (a) turns to a discrete electronic
energy level in (b) where the particle diameter corresponds with the de
Broglie wavelength. In (c) are shown bonding and antibonding molecular
orbitals, occupied by electrons localized in bonds.
180 Chem. Soc. Rev., 1999, 28, 179­185
the individual cluster behavior can be understood as being
caused by a doubly occupied electronic ground state, corre-
sponding with the situation shown in Fig. 1(b). Using the
formula for a three-dimensional electron gas, d = Hp
(2mEF)21/2 (d = lateral dimension of the quantum box, m =
mass of the electron, EF = Fermi energy) d ( = l/2) can be
determined as 1.4 nm, in perfect agreement with the calculated
data for a 55 atom cluster. Conductivity in the pellet is reached
by exciting the electrons to the first excited state and so halving
l. In Au55(PPh3)12Cl6 there is a special geometric situation
insofar as the thickness of the ligand shell is about 0.7 nm, i.e.
half of the diameter of the cluster core itself. Two ligand shells
of 1.4 nm in total separating two Au55 clusters from each other
are of the appropriate dimensions to enable perfect tunneling of
electrons through densely packed clusters. Indeed, the experi-
ments allow us to `metallize' the cluster material or to reduce
conductivity to a minimum. Fig. 2 elucidates these processes in
a simplified manner. As will be shown, thicker ligand shells or
spacer molecules enable tunneling processes too, but under
different energetic conditions.
Quantum size effects in nanoclusters have also been shown
by studying the susceptibility of variously sized Pd particles at
low temperatures.11 Odd and even numbers of electrons in Pd
clusters of identical size are expected to be present in a 50:50
ratio, depending on marginal deviations in geometry. Clusters
with a single electron couple more intensively the smaller they
are. Indeed, 2.2 nm Pd clusters show the most intensive
maximum of susceptibility compared with those of 3.0, 3.6 and
15 nm.
Using the same clusters, thermodynamic properties have
been shown to follow quantum size behaviour with respect to
the electronic specific heat for the first time. Again 2.2 nm Pd
clusters show the most significant deviations from bulk
behaviour at very low temperatures.11
These and several other experiments clearly indicate that
quantum behaviour of metal nanoclusters is observable, and is
most strongly expressed between 1 and 2 nanometers. Referring
to possible applications based on the quantum confinement of
electrons, particles in that size region should be of most
interest.
There are also several important results concerning the
electronic situation in single clusters, all indicating that they
follow quantum size behaviour when the current­voltage
characteristics are studied. Instead of a linear relationship,
typical for the Ohm behaviour of bulk metals, so-called
Coulomb blockades are observed, indicating single electron
transitions (SETs) between a tip and the cluster.12­14 The
temperature dependence of quantum size effects, based on the
relation Eel = e2/2C >> ET = kT has also been demon-
strated.12,13 This relation means that for initiating SET proc-
esses, the electrostatic energy Eel must be large compared with
the thermal energy ET. As the capacity C is smaller the smaller
the particle is, it becomes clear that SETs can only be observed
at relevant working temperatures if C is very small. From
various experiments we know C to be of the magnitude of 10219
Farad for Au55 clusters. In other words, the use of ca. 1.5 nm
metal clusters in principle enables switching with single
electrons around room temperature. This would not be possible
with semiconducting materials which lose their semiconducting
properties long before being miniaturized to 1.5 nm. This
knowledge prompted us and numerous other groups to look for
appropriate techniques to organize clusters in a three- (3D),
two- (2D) or one- (1D) dimensional manner. The use of
nanoclusters in future nanoelectronics is unambiguously linked
to the availability of well-ordered cluster arrangements. Indeed,
some considerable progress has been made during the last five
years. However, it is still difficult to make 3D, 2D or even 1D
arrangements routinely and in larger amounts. Therefore, the
following chapter deals with some very recent results in this
field as a contribution to the development of suitable conditions
for the realization of future nanoelectronic devices.
3 On the way to organized clusters
Metal and semiconductor clusters promise to substitute tradi-
tional materials in the micrometer size regime in the future. This
substitution would not only be linked with the miniaturization
of devices, but would especially be a big jump into a world of
novel technologies. First of all, there is the above mentioned
chance to work with single electrons at room temperature, a
condition for the development of new computer and laser
generations. Data storage capacities of orders of magnitude
better than at present are to be expected and even three-
dimensional neuronal networks, impossible to realize with
traditional materials, become possible. However, there are
numerous possible applications to be realized in shorter periods
of time, e.g. in the field of electroluminescence, non-linear
optics, surface-enhanced Raman spectroscopy, sensors or
catalysis. For some of these applications organized clusters are
not necessary, e.g. in catalysis or in sensoric. However, just for
the most attractive fields of nanoelectronics ordering is an
unambiguous condition.
As spheric particles in the size range of 1­10 nm do not tend
to form larger crystals, this usual route to optimized 3D
arrangements is in principle not possible, although some
crystalline particles on the micrometer scale have been found,
even consisting of relatively large nanoparticles of gold.15 Self-
assembled multilayers of nanosized particles have also been
observed, however, only with submicron dimensions.16­18
Our own activities in this field are based on the use of
bifunctional spacer molecules, linking clusters three-dimen-
sionally. The use of simple linear molecules such as H2N­
C6H4­(CH2)2­C6H4­NH2 and similar compounds indeed gave
three-dimensional cluster arrangements, but not with ordered
structures. We were not yet able to see larger cluster
arrangements by high resolution tunneling electron microscopy
(HRTEM).19 However, these 3D cluster materials gave another
important result, namely a direct relation between cluster
spacing and the activation energy needed to start electronic
tunneling via the spacers from one cluster to the next.14,20
Well-ordered cluster assemblies can be seen in Fig. 3, where
multiple layers of ca. 17 nm bimetallic particles can be seen,
consisting of a gold core, covered by a 3­4 atom layer thick
palladium shell. The very uniformly sized nanoclusters are
stabilized by a ligand shell of disodium 4,7-diphenyl-1,10-phe-
nanthrolinedisulfonate. This sulfonated phenanthroline deriva-
tive obviously favours ordered particle arrangements owing to
strong ionic interactions between the clusters. This example
demonstrates that appropriate interactions between nanoclusters
Fig. 2 Illustration of the electronic situation in a row of nanoclusters,
separated by barriers (ligand shells). In the ground state electronic tunneling
between the clusters is in principle not possible. In the excited state
tunneling is enabled by halving the electronic wavelength.
Chem. Soc. Rev., 1999, 28, 179­185 181
may lead to 3D arrangements which could be used for a
regulated communication between quantum dots. However, it is
to be stated that we are still at the beginning of a development
which will need much more effort to get routinely 3D
assemblies of nanoclusters of acceptable size.
Considering 2D nanocluster arrangements, there is some
progress to be registered during the last few years. Important
contributions by Shiffrin,16 Whetten,15 Andres,18 Möller,21,22
and others show that two-dimensionally organized nanoclusters
on suitable supports are possible. But, as already mentioned for
3D assemblies, routine work is still not possible. It may be that
the use of micelles in block copolymers, as used by Möller et al.,
is one of the most promising future techniques, as the
dimensions of the 2D array can be varied over a wide range and
the ordering of the particles is almost perfect. Our own efforts in
this field were focused on self-assembly processes between
functionalized clusters and modified surfaces.14,19,23 However,
as it turned out, strong chemical bondings between clusters and
surfaces do not usually give ordered 2D layers, but instead give
randomly oriented densely packed particles. It seems that good
ordering is only obtained if there is no or only weak cluster­
surface interactions, including the disadvantage that the struc-
ture can easily be destroyed by touching it, e.g. with the tip of
an AFM or STM.
One-dimensional cluster arrangements (cluster wires) are of
remarkable interest for several reasons: they can be used for the
study of electron transitions in one direction, probably much
easier to understand than those in 3D probes, and as semi-
conducting nanopaths for several technical applications. 1D
cluster arrangements need a template to be formed. We chose
nanoporous alumina which has numerous advantages compared
to other porous materials: it is easy to prepare, the pores in the
transparent oxide sheets are all running parallel through the
membrane, perpendicular to the surface and, most importantly,
there is a broad variability of the pore width between ca. 5 and
200 nm.24­26 If special conditions during the preparation are
considered, even hexagonally perfectly ordered pores can be
reached. Fig. 4 shows an atomic force microscopic (AFM)
image of a surface of such an ordered pore system. Filling the
pores with clusters of an appropriate size leads to assemblies of
cluster wires, isolated from each other by the alumina material,
but open for being contacted from one or from both sides to
study electronic behavior. Such measurements have not yet
been performed, but promising success has been achieved in
making cluster wires.27
Fig. 5 shows a TEM image of a cluster wire containing
membrane, sectioned along the pores. The clusters consist of 1.4
nm Au55 cores, enlarged to 4.2 nm by a ligand shell of thiol-
functionalized silsequioxanes.28 The difficulty is to fill the
pores without interruptions. This is obviously not possible if the
pores are longer than ten nanometers. Such thin alumina films
Fig. 3 A transmission electron microscopic (TEM) image of some ordered
layers of shell-structured gold­palladium particles.
Fig. 4 Atomic force microscopic (AFM) image of the surface of a
nanoporous alumina membrane. The pore diameter is ~ 40 nm. The sample
has been ion-beam milled previously.
Fig. 5 TEM image of a single 1.4 nm Au cluster wire in a nanoporous
alumina membrane (a), (b) is a schematic representation of the helical
situation in the 7 nm pore. Reproduced with permission from Chem. Eur. J.,
1997, 3, 1951. Copyright 1997 Wiley-VCH.
182 Chem. Soc. Rev., 1999, 28, 179­185
are obtained by ion beam milling processes. Electronic
measurements are just at the beginning.
4 Catalysis
The use of ligand stabilized transition metal clusters on the one
hand and of unprotected, bare nanoparticles on the other hand as
homogenous and heterogeneous catalysts, respectively, is
extensively described in the literature.29 In particular, supported
metal particles are traditionally applied in industrial catalysis
for many purposes. The catalytic behavior of bare particles on
supports has been studied as a function of size and shape in a
huge number of papers in the course of the last decades, whereas
ligand protected clusters are much less investigated as im-
mobilized catalysts. The main interest was focused on the
function of the ligands. Indeed, they can increase or reduce
activity, however, it has also been shown that the influence of
ligands with respect to selectivity may also be of some interest.
We have been able to show that ligand stabilized Pd clusters in
the size range of 3­4 nm show very good activities and
selectivities on various supports when they are used for the
semihydrogenation of hex-2-yne to cis-hex-2-ene.30,31 The
ligands consisted of variously substituted phenanthrolines. It
could be observed that, depending on the kind of substituent,
e.g. alkyl groups of various lengths, the activities changed
considerably, whereas the selectivity was in any case close to
100%.
Here, we report for the first time on the use of very small Pd
nanoparticles (1.5 nm) in a supported form with and without
ligands to semihydrogenate hex-2-yne. This comparison of
protected and bare clusters for the same catalytic reaction is
important since there is, to our knowledge, no information on
comparable reactions. This type of cluster is formed when
palladium acetate and 1,10-phenanthroline in a 1.3 :10 molar
ratio are dissolved in 3-methylbutanol and heated to 60 °C for 7
days. The clusters, which can be isolated from solution by
centrifugation as a black powder, can be redissolved in a 1:1
water­pyridine mixture. Immobilization occurs from solution
by the addition of supports such as active carbon, TiO2, Al2O3
or various zeolites. Catalysis was performed with 1 wt.% Pd on
support. The ligands were removed by heating the immobilized
clusters to 100­130 °C for 4­5 hours in high vacuum.
Transmission electron microscopy was used to prove that the
particle size was the same before and after heating. The
hydrogenation reactions were carried out in ethanol at room
temperature and with hydrogen gas at 1 atm. Figs. 6(a) and 6(b)
show the results.
It is clearly to be seen that the selectivity in the case of
phenanthroline protected clusters (a) is only ca. 80­90%,
whereas in the case of the bare clusters (b) it is 100%. From our
experience this result could not be expected. Whatever the
reason for that behavior may be, it becomes clear that catalytic
studies with clusters of definite size and environment are
valuable materials to work out principles.
5 Luminescence of semiconducting nanoclusters
The well-defined nanoporous alumina membranes, briefly
described in Chapter 3, can not only be used to fill in clusters
which have been prepared before, but they offer the unique
chance to synthesize nanoparticles inside tubes of defined
length and width. The advantage of that system is the absolute
transparency in the visible range so that the generation of
optically interesting materials in the membrane can directly lead
to useful materials.
Among the possible candidates to form photoluminescent
nanoclusters inside the pores, the compounds CdS and GaN
have been selected, owing to their large band gaps of 2.3832 and
3.4033 eV, respectively.
To generate crystalline CdS nanoclusters in the pores,
membranes which are open on both sides have been used, and
aqueous solutions of CdCl2, beginning with 5 wt.% up to
saturated ones have been filled in by vacuum induction,
followed by the reaction with gaseous H2S which diffuses
rapidly into the pores to precipitate CdS. GaN was produced by
the ligand precursor compound (C2H5)2GaNH2
34 which, after
its transfer into the pores, was decomposed at 350­550 °C in an
atmosphere of nitrogen, corresponding to eqn. (1). The gaseous
DT
(C2H5)2GaNH2 --? GaN + 2C2H4 + H2 (1)
products ethene and hydrogen leave the pores and GaN is
deposited. In contrast to the CdS nanoclusters, the size and
structure of the GaN particles has not yet been determined.
The luminescence of CdS monocrystals was observed
between 2.08 eV (595 nm) and 2.05 eV (605 nm).35 The
luminescence of the CdS in the pores was found between 1.94
eV (640 nm) and 1.70 eV (730 nm), depending on the conditions
to be discussed in the following. Fig. 7 shows the photo-
luminescence spectrum of a 25 nm pore-containing membrane,
filled and dried at room temperature. Besides the broad peak at
ca. 650 nm there is a high-energetic excitonic fluorescence at
2.84 eV (437 nm).
Excitonic fluorescence of CdS nanoclusters is normally only
observed after modification of the particle surface.1,36 The
modification consists of a substitution of free valences on the
surface by Cd ions blocking the S and SH radicals. These are
responsible for the non-radiative recombination by trapping
holes and preventing activation of fluorescence. Excess Cd ions
on the surface allow radiative deactivation of electrons from the
conductivity band and holes from the valence band to give the
high-energetic emission. The excitonic emission of CdS
nanoclusters in the alumina pores is obviously caused by the
Fig. 6 Semihydrogenation of hex-2-yne to cis-hex-2-ene by ligand protected
(a) and bare (b) 1.5 nm Pd clusters on active carbon in ethanol at room
temperature and 1 atm of hydrogen pressure.
Chem. Soc. Rev., 1999, 28, 179­185 183
interaction of the particle surface with the pore walls, consisting
of reactive Al­OH functional groups, and by excess Cd ions
blocking the hole-trapping by free reactive valences of the
clusters.
We also studied the effect of repeated CdS precipitation in the
pores and observed that the luminescence intensity increased
considerably by a second CdS formation in 20 nm pores, but
decreased with further CdS depositions. Fig. 8 shows this
behavior. As can also be seen, there is no significant effect on
the wavenumbers. We interpret the intensity maximum after the
second filling by the formation of additional active particles, the
decrease of intensity after three- or four-fold precipitations is to
be traced back to the progressive formation of bulk-like larger
and inactive particles. This agrees with the fact that membranes
with larger pores, e.g. 30 nm, show the maximum of
luminescence only after the fourth filling, followed by an
intensity decrease with further CdS formations.
The situation in the CdS-containing pores is best seen from
TEM images of ion-beam milled, very thin membranes. In Fig.
9(a), a larger cutout of an ion-beam milled membrane is shown.
The pores have been filled by CdS, precipitated from a
concentrated CdCl2 solution, leading to a relatively high
particle loading. Most of the pores at the distinct positions,
accidentically reached by the milling process, are completely or
partially filled, depending on the local situation. In Fig. 9(b) a
single pore is shown, filled with polycrystalline CdS matter.
Finally we studied the temperature dependence of the
luminescence. It was found that tempering of the filled
Fig. 7 Photoluminescence spectrum at 300 K of a CdS filled nanoporous
alumina membrane with 25 nm pores.
Fig. 8 Dependence of the photoluminescence intensity of CdS containing
alumina membranes on the number of CdS precipitations.
Fig. 9 Transmission electron microscopic images of CdS filled alumina
membranes. (a) shows the surface of an ion-beam milled, very thin
membrane. Most of the pores show the presence of CdS; (b) shows a
magnified pore filled with crystalline CdS.
184 Chem. Soc. Rev., 1999, 28, 179­185
membranes leads to a continuous decrease of luminescence,
completely ending at ca. 350 °C. The disappearance of the
orange luminescence by thermal treatment is known from
literature and is caused by the phase transition from the cubic to
the hexagonal modification.37
The GaN loaded membranes showed a luminescence signal at
3.62 eV (341 nm). Compared with literature data of cubic or
hexagonal GaN, this is a short-wave shift of 41 and 21 nm,
respectively.38 A relationship between the photoluminescence
intensity, the time of tempering and the number of GaN
depositions can not yet be confirmed. However, as mentioned
above, these results are still preliminary and should just show
that photoluminescent foils of unlimited size can be produced in
a very simple way, if appropriate reactions can be performed in
the pores or if suitable precursors are available. In this
connection it should be mentioned that luminescent silicon,
probably in the form of a siloxene, has also been generated in
the pores by us recently.39
6 Outlook
Nanoclusters of metals or semiconductors can more and more
be considered as the building blocks of future modern
technologies. This is due to the size dependant electronic
properties of these particles. Nanoclusters of transition metals
become semiconductors if small enough. If the more techno-
logical problems such as organization and addressing of these
quantum dots can be solved, there is an almost unlimited field
of applications to be foreseen. The properties of nanosized
semiconductors have long been known to depend very sensi-
tively on the particle size. There are numerous and important
applications which become possible considering these facts.
The use of transparent alumina membranes (and of course other
comparable materials) gives an additional chance to prepare and
to apply semiconducting nanoclusters in new fields in the near
future.
7 Acknowledgements
We gratefully acknowledge financial support by the Deutsche
Forschungsgemeinschaft and the Fonds der Chemischen In-
dustrie. We also wish to thank the BASF AG, Ludwigshafen,
for their support of the catalytic work.
8 References
1 H. Weller, Angew. Chem., Int. Ed. Engl., 1993, 32, 41.
2 G. Schmid, Inorg. Synth., 1990, 7, 214.
3 G. Schmid, B. Morun and J.-O. Malm, Angew. Chem., Int. Ed. Engl.,
1989, 28, 778.
4 G. Schmid, M. Harms, J.-O. Malm, J.-O. Bovin, J. v. Ruitenbeck, H. W.
Zandbergen and Wen T. Fu, J. Am. Chem. Soc., 1993, 115, 2046.
5 G. Schmid, S. Emde, V. Maihack, W. Meyer-Zaika and St. Peschel,
J. Mol. Catal. A, 1996, 107, 95.
6 G. Schmid, R. Pugin, J.-O. Malm and J.-O. Bovin, Eur. J. Inorg. Chem.,
1998, 6, 813.
7 B. A. Smith, J. Z. Zhang, U. Giebel and G. Schmid, Chem. Phys. Lett.,
1997, 270, 139.
8 U. Simon, G. Schmid and G. Schön, Angew. Chem., Int. Ed. Engl., 1993,
32, 250.
9 G. Schön and U. Simon, Colloid Polym. Sci., 1995, 273, 101.
10 G. Schön and U. Simon, Colloid Polym. Sci., 1995, 273, 202.
11 Y. Volokitin, J. Sinzig, L. J. de Jongh, G. Schmid and I. I. Moiseev,
Nature, 1996, 384, 621.
12 A. Bezryadin, C. Dekker and G. Schmid, Appl. Phys. Lett., 1997, 71,
1273.
13 L. F. Chi, M. Hartig, T. Drechsler, Th. Schwaack, C. Seidel, H. Fuchs
and G. Schmid, Appl. Phys. A, 1998, A66, 187.
14 G. Schmid and L. F. Chi, Adv. Mater., 1998, 10, 515.
15 R. L. Whetten, J. T Khoury, M. M. Alvarez, S. Murthy, I. Vezmar, Z.
Wang, P. W. Stephens, C. L. Cleveland, W. D. Luedtke and U.
Landmann, Adv. Mater., 1996, 8, 428.
16 M. Brust, M. Walker, D. Bethell, D. J. Shiffrin and R. Whyman, J.
Chem. Soc., Chem. Commun., 1994, 801.
17 M. Brust, D. Bethell, D. J. Shiffrin and C. J. Kiely, Adv. Mater., 1995,
7, 795.
18 R, P. Andres, J. D. Bielefeld, J. I. Henderson, D. B. Janes, V. R.
Kolagunta, C. P. Kubiak, W. J. Mahoney and R. G. Osifchin, Science,
1996, 273, 1690.
19 G. Schmid and St. Peschel, New J. Chem., 1998, 7, 669.
20 U. Simon, R. Flesch, H. Wiggers, G. Schön and G. Schmid, J. Mater.
Chem., 1998, 8, 517.
21 J. P. Spatz, A. Roescher and M. Möller, Adv. Mater., 1996, 8, 337.
22 J. P. Spatz, S. Mößner and M. Möller, Chem. Eur. J., 1997, 3, 1552.
23 St. Peschel and G. Schmid, Angew. Chem., Int. Ed. Engl., 1995, 34,
1442.
24 J. P. O'Sullivan and G. C. Wood, Proc. R. Soc. London, Ser. A, 1970,
317, 511.
25 H. Masuda and F. Fukada, Science, 1995, 248, 1466.
26 C. Martin, Science, 1994, 266, 1961.
27 G. L. Hornyak, M. Kröll, R. Pugin, Th. Sawitowski, G. Schmid, J.- O.
Bovin, G. Karsson, H. Hofmeister and S. Hopfe, Chem. Eur. J., 1997, 3,
1951.
28 G. Schmid, R. Pugin, J.-O. Malm and J.-O. Bovin, Eur. J. Inorg. Chem.,
1998, 6, 813.
29 For a summarizing review containing ca. 400 references see L. N.
Lewis, Chem. Rev., 1993, 93, 2693
30 G. Schmid, S. Emde, V. Maihack, W. Meyer-Zaika and St. Peschel,
J. Mol. Catal. A, 1996, 107, 95.
31 G. Schmid, V. Maihack, F. Lantermann and St. Peschel, J. Chem. Soc.,
Dalton Trans., 1996, 589.
32 I. Broser, R. Broser and M. Rosenzweig, Landolt-Börnstein, Vol. III,
p. 166, Springer, Berlin, 1982.
33 R. D. Dupris, J. A. Edmond, S. Nakamura and F. A. Ponce, Gallium
Nitride and Related Materials, Vol. 395, p. 551, Mater. Res. Soc. Symp.
Proc., 1997.
34 J. Andrews and M. Littlejohn, J. Electrochem. Soc., 1975, 122, 1273.
35 N. Susa, H. Watanabe and M. Wada, Jpn. J. Appl. Phys., 1976, 15,
2365.
36 L. Spanhel, M. Haase, H. Weller and A. Henglein, J. Am. Chem. Soc.,
1987, 109, 5649.
37 H. Ariza-Caderon, R. Lozada-Morales, O. Zelaya-Angel, G. Mendoza-
Alvares and L. Banos, J. Vac. Sci. Technol. A, 1990, 14, 2480.
38 R. P. Dubois, J. A. Edmond, S. Nakamura and F. A. Ponce, Gallium
Nitride and Related Materials, Vol. 395, p. 551, Mater. Res. Soc. Symp.
Proc., 1997.
39 A. Heilmann, P. Jutzi, A. Klipp, U. Kreibig, R. Neuendorf, Th.
Sawitowski and G. Schmid, Adv. Mater., 1998, 10, 398.
Review 8/01153B
Chem. Soc. Rev., 1999, 28, 179­185 185