Role of Nanoparticles in Photocatalysis - Springer Link

Journal of Nanoparticle Research 1: 439–458, 1999. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.

Review paper

Role of nanoparticles in photocatalysis D. Beydoun1 , R. Amal1,∗ , G. Low2 and S. McEvoy3 Centre for Particle and Catalyst Technologies, School of Chemical Engineering and Industrial Chemistry, The University of New South Wales, Sydney, N.S.W. 2052 Australia; 2 Environment Protection Authority, Analytical Chemistry, Lidcombe, N.S.W. 2141 Australia; 3 CSIRO Division of Energy Technology, Lucas Heights Science and Technology Centre, Lucas Heights, N.S.W. 2234 Australia; ∗ Author for correspondence (Tel.: 2 9385 4361; Fax: 2 9385 5966; E-mail: [email protected])

1

Received 22 February 1999; accepted in revised form 17 June 1999

Key words: nanoparticles, quantum size, photocatalysis, dopants, sensitization, nanocrystalline films

Abstract The aim of this review paper is to give an overview of the development and implications of nanotechnology in photocatalysis. The topics covered include a detailed look at the unique properties of nanoparticles and their relation to photocatalytic properties. Current applications of and research into the use of nanoparticles as photocatalysts has also been reviewed. Also covered is the utilization of nanoparticles in doped, coupled, capped, sensitized and organic– inorganic nanocomposite semiconductor systems, with an effort to enhance photocatalytic and/or optical properties of commonly used semiconductor materials. The use of nanocrystalline thin films in electrochemically assisted photocatalytic processes has been included. Finally, the use of nanoparticles has made a significant contribution in providing definitive mechanistic information regarding the photocatalytic process. Background on photocatalysis The photocatalytic process Organic chemicals which may be found as pollutants in wastewater effluents from industrial or domestic sources, must be removed or destroyed before discharge to the environment. Such pollutants may also be found in ground and surface waters which also require treatment to achieve acceptable drinking water quality (Lindner et al., 1995). The increased public concern with these environmental pollutants has prompted the need to develop novel treatment methods (Zeltner et al., 1996) with photocatalysis gaining a lot of attention in the field of pollutant degradation. Much of the natural purification of aqueous system lagoons, ponds, streams, rivers and lakes is caused by sunlight initiating the breakdown of organic molecules into simpler molecules and ultimately to carbon dioxide

and other mineral products. There are various natural sensitizers that accelerate the process. The utilization of ‘colloidal semiconductors’ and the introduction of catalysts to promote specific redox processes on semiconductor surfaces were developed in 1976 (Kalyanasundaram, 1983). Since then, laboratory studies have confirmed that naturally occurring semiconductors could enhance this solar driven purification process (Matthews, 1993). The photocatalytic detoxification of wastewater is a process that combines heterogeneous catalysis with solar technologies (Zhang et al., 1994a). Semiconductor photocatalysis, with a primary focus on TiO2 , has been applied to a variety of problems of environmental interest in addition to water and air purification. The application of illuminated semiconductors for degrading undesirable organics dissolved in air or water is well documented and has been successful for a wide variety of compounds (Hoffmann et al., 1995).

440 Organic compounds such as alcohols, carboxylic acids, amines, herbicides and aldehydes, have been photocatalytically destroyed in laboratory and field studies. The photocatalytic process can mineralize the hazardous organic chemicals to carbon dioxide, water and simple mineral acids (Ahmed et al., 1984). Many processes have been proposed over the years and are currently used to remove organic toxins from wastewaters. Current treatment methods for these contaminants, such as adsorption by activated carbon and air stripping, merely concentrate the chemicals present, by transferring them to the adsorbent or air, but they do not convert them into non-toxic wastes. Thus, one of the major advantages of the photocatalytic process over existing technologies is that there is no further requirement for secondary disposal methods. Another advantage of this process is that when compared to other advanced oxidation technologies, especially those using oxidants such as hydrogen peroxide and ozone, expensive oxidizing chemicals are not required as ambient oxygen is the oxidant (Matthews et al., 1993). Photocatalysts are also self-regenerated and can be reused or recycled. Finally, the solar photocatalytic process can also be applied to destroy nuisance odours, taste and odour compounds, and naturally occurring organic matter, which contains the precursors to trihalomethanes formed during the chlorine disinfection step in drinking water treatment (Zhang et al., 1994b). During the photocatalytic process, the illumination of a semiconductor photocatalyst with ultraviolet (UV) radiation activates the catalyst, establishing a redox environment in the aqueous solution (Zhang et al., 1994a). Semiconductors act as sensitizers for lightinduced redox processes due to their electronic structure, which is characterized by a filled valence band and an empty conduction band (Hoffmann et al., 1995). The energy difference between the valence and conduction bands is called the band gap. The semiconductor photocatalyst absorbs impinging photons with energies equal to or higher than its band gap or threshold energy. Each photon of the required energy (i.e. wavelength) that hits an electron in the occupied valence band of the semiconductor atom, can elevate that electron to the unoccupied conduction band leading to excited state conduction band electrons and positive valence band holes (Schiavello, 1989). The fate of these charge carriers may take different paths (refer to Figure 1). Firstly, they can get trapped, either in shallow traps (ST) or in deep traps

Figure 1. Processes involved in semiconductor particles upon bandgap excitation. CB: Conduction Band, VB: Valence Band, ST: Surface Traps, DT: Deep Traps, rec: recombination.

(DT). Secondly, they can recombine, non-radiatively or radiatively, dissipating the input energy as heat. Finally, they can react with electron donors or acceptors adsorbed on the surface of the photocatalyst (Hoffmann et al., 1995). In fact, it was recently shown that any photoredox chemistry occurring at the particle surface, emanates from trapped electrons and trapped holes rather than from free valence band holes, and conduction band electrons (Serpone et al., 1996). The competition between charge-carrier recombination and charge-carrier trapping followed by the competition between recombination of trapped carriers and interfacial charge transfer are what determine the overall quantum efficiency for interfacial charge transfer (Hoffmann et al., 1995). Also of great importance are the band positions or flat band potentials of the semiconductor material. These indicate the thermodynamic limitations for the photoreactions that can take place (Hagfeldt et al., 1995). Photocatalysts An ideal photocatalyst should be stable, inexpensive, non-toxic and, of course, highly photoactive. Another primary criteria for the degradation of organic compounds is that the redox potential of the H2 O/• OH couple (OH− → • OH + e− ; E0 = − 2.8 V) lies within the bandgap of the semiconductor (Hoffmann et al., 1995). Several semiconductors have bandgap energies sufficient for catalysing a wide range of chemical reactions. These include TiO2 , WO3 , SrTiO3 , α-Fe2 O3 , ZnO and ZnS. TiO2 , the semiconductor most thoroughly investigated in the literature, seems to be the most promising for photocatalytic destruction of organic pollutants (Howe, 1998). This semiconductor provides the best

441 compromise between catalytic performance and stability in aqueous media (Aruna et al., 1996). The anatase phase of titanium dioxide is the material with the highest photocatalytic detoxification (Bahnemann et al., 1993). Binary metal sulphide semiconductors such as CdS, CdSe or PbS are regarded as insufficiently stable for catalysis, at least in aqueous media as they readily undergo photoanodic corrosion (Howe, 1998; Fischer, 1989). These materials are also known to be toxic. The iron oxides are not suitable semiconductors as they readily undergo photocathodic corrosion (Hoffmann et al., 1995). The band gap for ZnO (3.2 eV) is equal to that of anatase. ZnO however, is also unstable in water with Zn(OH)2 being formed on the particle surface. This results in catalyst deactivation (Howe, 1998).

Hence, as the size of the semiconductor particle is reduced below a critical diameter, the spatial confinement of the charge carriers within a potential well (Serpone et al., 1996), like a ‘particle in a box’ (Hagfeldt et al., 1995), causes them to behave quantum mechanically (Hoffmann et al., 1995). In solid state terminology this means that the bands split into discrete electronic states (quantized levels) in the valence and conduction bands (Weller et al., 1995) and the nanoparticle behaves more and more like a giant atom (Henglein, 1997). Nanosized semiconductor particles which exhibit size-dependent optical and electronic properties are called quantized particles (Q-particles) or quantum dots (Kamat, 1995).

Nanocrystalline photocatalysts

Quantum-size (Q-size) effects occur when the size of the semiconductor particles become smaller than the Bohr radius of the first excitation state (Hagfeldt et al., 1995). This has been stated by other authors as when the particle size of the colloidal particle becomes comparable to the DeBroglie wavelength of the charge carriers (Henglein et al., 1987; Weller et al., 1995). There seems to be discrepancy between the reported critical size below which quantization effects are observed. Many values have been quoted in the literature for the same semiconductor nanoparticles. There are also discrepancies between the estimated critical diameter and the actual diameters at which quantization effects are observed. These predictions depend critically on the effective masses of the charge carriers (Serpone et al., 1996). Kormann et al. estimated the excitation radii for titania particles to be between 7.5 and 19 Å. The exciton radii were computed by using various literature values and calculated by the equation put forward by Brus (Kormann et al., 1988). Gratzel estimated an excitation radius of ∼ 3 Å (Gratzel et al., 1989). Kormann et al. reported quantization effects with 20–40 Å titania particles. Q-size effects were observed during particle growth and at the final stage of synthesis of these transparent colloids. The synthesized anatase TiO2 particles resulted in a bandgap shift of 0.15 eV relative to bulk anatase TiO2 while the synthesized rutile TiO2 , with d ∼ 2.5 nm, were blue shifted by 0.1 eV. Anpo et al. (1987) reported excitation confinement with 55–2000 Å rutile particles and up to 530 Å for

Nanocrystalline photocatalysts are ultrasmall semiconductor particles which are a few nanometres in size. During the past decade, the photochemistry of nano semiconductor particles has been one of the fastest growing research areas in physical chemistry (Henglein, 1997). The interest in these small semiconductor particles originates from their unique photophysical and photocatalytic properties (Bahnemann, 1993). Several review articles have been published concerning the photophysical properties of nanocrystalline semiconductors (Henglein, 1988; 1989; Wang et al., 1991; Bahnemann 1993; Zeltner et al., 1996; Levy, 1997). Such studies have demonstrated that some properties of nanocrystalline semiconductor particles are in fact different from those of bulk materials. Nanosized particles, with diameters ranging between 1 and 10 nm, possess properties which fall into the region of transition between the molecular and the bulk phases (Bahnemann, 1993). In the bulk material, the electron excited by light absorption finds a high density of states in the conduction band, where it can exist with different kinetic energies (Henglein, 1997). In the case of nanoparticles however, the particle size is the same as or smaller than the size of the first excited state. Thus, the electron and hole generated upon illumination cannot fit into such a particle unless they assume a state of higher kinetic energy (Weller et al., 1995).

The critical size and changes in properties due to nanosize

442 anatase particles. Kavan et al. (1993) reported blue shifts with 25–100 Å particles. Weller et al. (1995) has stated that size quantization effects for semiconductor particles typically occur in a diameter range between 2 and 50 nm. In a recent review about photocatalysis, it was stated that the critical diameter for TiO2 is 10 nm (Howe, 1998). The same value was quoted by Martin et al., 1995. Recently, it has been proposed by Serpone et al. that what had been taken in the past as observed size quantization effects on 2 nm TiO2 particles, may in fact not be a Q-size effect (Serpone et al., 1996). Their deductions were made from observations with spectroscopic blue shifts in the effective bandgap of TiO2 with particle sizes of 21, 133 and 267 Å. The study argued that some of the bandgap shifts might simply reflect observations of direct electron transition in an otherwise indirect bandgap semiconductor such as TiO2 . When examining the absorption and photoluminescence of these particles, the authors found that an attempt to describe the absorption coefficients by the indirect band gap semiconductor relationship failed to give the expected linear correlation. However, a linear correlation was obtained with the relationship applicable to direct band gap semiconductors. Their conclusions were that the particle sizes examined seemed to affect charge carriers relaxation dynamics, however, these did not appear to be related to excitation confinement (Serpone et al., 1996). When considering semiconductors other than TiO2 , many reports have observed quantum size effects with many different particles. For example CdS and PbS a few nanometres in size have been found to exhibit quantum mechanical effects (Gallardo et al., 1989). Quantization effects have also been demonstrated with nanoclusters of hexagonal MoS2 and several of its isomorphous Mo and W chalcogenides (with cluster size down to ∼ 2 nm) (Wilcoxon et al., 1997). Henglein also observed size quantization effects with 1–10 nm CdS particles (Henglein et al., 1987). The authors recognized that quantization effects become evident when the particle size becomes comparable to the DeBroglie wavelength of the charge carriers. Quantization effects for an electron in an evacuated box were estimated to become significant at box dimensions of ∼ 0.1 nm (using the ‘particle in a box’ model). However, in the colloidal particles the effects were seen at much larger particle size. They found the reason to lie in the fact that the effective mass of a charge carrier, which moves in the periodic array of the constituents

of the crystal lattice, is generally much lower than the mass of an electron in free space. This results in a larger DeBroglie wavelength. The smaller the effective mass of the charge carriers, the more pronounced are the optical-size effects (Henglein et al., 1987). Size quantization in semiconductor particles leads to drastic changes in numerous important properties of the material. Firstly, size quantization affects the electronic properties of the semiconductor particle, with the ultrasmall crystallites composed of a few molecular units maintaining their discrete HOMOs (Highest Occupied Molecular Orbitals) and LUMOs (Lowest Unoccupied Molecular Orbitals) (Serpone et al., 1996). This in turn affects charge-carrier dynamics (Mimic et al., 1990; Smith et al., 1997). For example, for aqueous gold and silver colloids, spatial-size confinement was found to cause substantially slower electronic relaxation due to reduction of non-equilibrium electron transport and weaker electron–phonon coupling (Smith et al., 1997). Secondly, in the Q-size regime, the chemical and physical properties, which are related to electronic properties, strongly depend on the size of the nanoparticles (Weller et al., 1995). The electronic properties relate to the solid state physics aspects of the particles, that is, band structures and band gap, as well as the electrochemical positioning of the bandgap edge potential with respect to the Nernst potential of the solution (Hagfeldt et al., 1995). The band gap of the semiconductor becomes larger with decreasing particle size, and is indicated by an absorption shift to shorter wavelengths (Henglein, 1997). The band edges shift to yield larger redox potentials (Howe, 1998). The levels of the valence band are moderately shifted to lower energies, while those of the conduction band are strongly shifted to higher energies (Henglein, 1987). For a size-quantized particle, the increase in bandgap energy, 1EBG , varies with the radius of the particle, Rpart , the reduced effective mass of the excitation, µ, and the dielectric constant of the semiconductor, ε (Serpone et al., 1996). The reduced effective mass of the excitation, µ, is related to the effective masses of the charge carriers (me , mh ) by the following expres−1 sion: µ−1 = m−1 e + mh . From Eq. (1) it can be seen that 1EBG depends critically on the effective masses of the electrons and holes: 2 − 1.8e2 /εRpart . 1EBG = h2 /8µRpart

(1)

As a result of the change in electronic properties, quantization effects also have a pronounced effect

443 on spectral properties of semiconductors (Bahnemann et al., 1984). For instance, with decreasing particle size, the energy of the optical transition increases and the colour of the fluorescence light is blue shifted, that is, it is shifted to a shorter wavelength (Henglein, 1987). The Q-particles can have a different colour depending on the particle size. For example Cadmium sulphide normally exists as a yellow material, but becomes colourless when the particle is smaller than 22 Å. Cadmium phosphide, which is normally a black material, can be made in various colours depending on the particle size (Henglein, 1997a). Bahnemann et al. used a molecular orbital description to explain the optical properties of semiconductor particles. During particle growth of TiO2 particles from molecular to bulk sizes, molecular orbitals of an increasing number of molecules overlap. The energy gap between the HOMO and LUMO decreases. These changes result in a red shift (longer wavelength) of the accompanying optical transition (Bahnemann et al., 1987).

Activity of nano-photocatalysts One of the main advantages of the application of Q-sized particles is the increase in the bandgap energy with decreasing particle size. As was mentioned earlier, as the size of a semiconductor particle falls below the critical radius, the charge carriers begin to behave quantum mechanically and the charge confinement leads to a series of discrete electronic states. As a result there is an increase in the effective band gap and a shift of the band edges. Thus by varying the size of the semiconductor particles, it is possible to enhance the redox potential of the valence-band holes and the conductionband electrons (Hoffmann et al., 1992b). However, the solvent reorganization free energy for charge transfer to a substrate, remains unchanged (Martin et al., 1995). The increasing driving force and the unchanged solvent reorganizational free energy are expected to lead to an increase in the rate constants for charge transfer at the surface (Hoffmann et al., 1995). Thus, the use of size-quantized semiconductor particles may result in increased photoactivity for systems in which the rate-limiting step is interfacial charge transfer (Martin et al., 1995). Hence nanosized semiconductor particles can possess enhanced photoredox chemistry, with reduction reactions, which might not otherwise proceed in bulk materials, being able to occur readily using sufficiently

small particles (Serpone et al., 1996; Nedeljkovic et al., 1986). Another factor which could be advantageous is the fact that the fraction of atoms that are located at the surface of a nanoparticle is very large (Henglein, 1997). Q-sized particles also have high surface area to volume ratios, which further enhances their catalytic activity (Hoffmann et al., 1992b). One disadvantage of nanosized particles is the need for light with a shorter wavelength for photocatalyst activation. Thus a smaller percentage of a polychromatic light source will be useful for photocatalysis. The first account of light driven redox reactions in nanocrystalline systems was published in 1981. Since then, nanocrystalline semiconductor systems which have been well studied include oxides, sulphides and selenides, such as TiO2 , ZnO, WO3 , V2 O5 , Ag2 O, ZnS, CdS, PbS, Cu2 C, MoS2 , and CdSe (Hagfeldt, 1995). While some studies have found that nanoparticles have enhanced photoactivities over their bulk-phase counterparts, (Hoffmann et al., 1994; Anpo et al., 1987; Nedeljkovic et al., 1986; Nosaka et al., 1990), others have found the contrary to be true (Lepore et al., 1993; Hoffmann et al., 1992a; Martin et al., 1995). In cases where the activity of the nanoparticles was found to be lower than that of their bulk counterparts, it is thought that the positive effects of increased photo-potentials were offset by unfavourable surface speciation and surface defect sites on the size-quantized semiconductor particles, due to the preparation method (Hoffmann et al., 1995). Wang et al. (1997) demonstrated that there exists an optimal particle size in nanocrystalline TiO2 systems for maximum photocatalytic efficiency. In their experiments which involved the decomposition of chloroform, they observed an improvement in activity when the particle size was decreased from 21 to 11 nm, but the activity decreased when the size was reduced further to 6 nm. They concluded that for this particular reaction the optimum particle size was about 10 nm. Martin et al. (1995) saw the variation in the photoefficiencies of different forms of TiO2 as being fundamentally related to their charge-carrier dynamics. The authors studied the charge-carrier dynamics of QTiO2 and Degussa P25 (made up of 20–35 nm primary particles, the size of the aggregates is 50–200 nm) by time-resolved microwave conductivity. They found that the resultant interfacial electron transfer appeared to be faster for P25 than for Q-TiO2 . The slower

444 electron-transfer rates observed for Q-TiO2 were consistent with the lower steady-state quantum yields. This was supported by Zhang et al. (1998) who gave the following explanation for these observations. They suggested that in large TiO2 particles, volume recombination of the charge carriers is the dominant process, and can be reduced by a decrease in particle size. This decrease also leads to an increase in the surface area which can be translated as an increase in the available surface active sites. Thus a decrease in particle size should also result in higher photonic efficiencies due to an increase in the interfacial charge-carrier transfer rates. However, as the particle size is lowered below a certain limit, surface recombination processes become dominant since firstly, most of the electrons and holes are generated close to the surface, and secondly since surface recombination is faster than interfacial chargecarrier transfer processes. This is the reason why there exists an optimum particle size for maximum photocatalytic efficiency. The first accurately determined quantum yields of polycrystalline TiO2 were provided by Lepore et al. (1993). The authors compared the photochemical and photophysical behaviour of polycrystalline, optically transparent TiO2 suspensions (with a 62 nm average effective radius, prepared by successive centrifugation of P25 slurries) with quantum-sized (∼ 2.4 nm) TiO2 solutions, as well as P25 slurries. The quantum yields obtained with the centrifuged suspensions were similar to those of P25, but were higher than those obtained with the quantum-sized TiO2 . This difference was explained in terms of a longer lifetime of the charge carriers in the centrifuged samples and the P25 slurries, compared to the quantumsized (∼ 2.4 nm) TiO2 solutions. Also, from their findings, the authors suggested the lack of surfacetrapped electrons in P25 samples as a possible factor responsible for their higher photoactivity (Lepore et al., 1993). The first results demonstrating the enhancement of photoredox produced by size-quantization effects, were reported by Nedeljkovic et al. (1986). The nanocrystalline semiconductors studied were HgSe, PbSe, and CdSe colloids. For diameters less than 50 Å, the optical absorption edge of HgSe and PbSe was blue shifted by several volts. The enhancement effects of the photoredox chemistry of Q-size semiconductor particles were demonstrated with the evolution of H2 with aqueous colloids of HgSe and PbSe (d < 50 Å). The nanocrystalline CdSe particles (d < 50 Å) were

studied for CO2 reduction (to formic acid) since they have greater stability against photocorrosion compared to HgSe and PbSe. Large particle-sized CdSe colloids did not yield formic acid under the same experimental conditions. Anpo et al. (1987) synthesized extremely small TiO2 particles and investigated their use for the photocatalytic hydrogenation reactions of CH3 CCH with H2 O. They found the activity to increase as the diameter of the TiO2 particles decreased, especially below 100 Å. Absorption and photoluminescence spectra of these particles had exhibited blue shifts below 100 Å, suggesting a size-quantization effect. The authors suggested that the dependence of the yields on the particle size arose from the differences in the chemical reactivity not the physical properties, such as the surface area, of these catalysts. The authors recognized the presence of quantization effects in the systems they were studying, although they did not provide an explanation as to why these effects led to increased activity. The photocatalytic activity of Q-TiO2 particles was also demonstrated by Kormann et al. (1988). Henglein et al. (1989) studied the photochemistry of AgI nanoparticles. The absorption threshold of the prepared colloids was blue shifted, which is indicative of quantization effects. These AgI particles were found not to be photoactive with respect to the formation of silver or iodide. However, it was found that silver metal was deposited on the surface of the AgI colloids when the solution was de-aerated and sodium sulphite added. The fluorescence of these particles, which occurs close to the absorption threshold, was found to be quenched by the small amounts of deposited silver, believed to be acting as a charge scavenger. The scavenging process did not result in an increase in the reaction yield, with the silver deposit acting as a mediator for the recombination of charge carriers. Hoffmann et al. (1992a) showed that polymerization of methyl methacrylate readily occurred when using Q-sized ZnO semiconductors as photointiators. The quantum yields decreased as particle size decreased, due to either increased surface defects or enhanced rates of competing electron–hole recombination. Under the same experimental conditions, no polymerization occurred with bulk-size particles as photoinitiators. The same group also studied the efficiency of photoinitiation of polymerization of several vinylic monomers using both bulk and quantum-sized CdS, ZnO and TiO2 . The authors found that the Q-sized semiconductors demonstrated significantly higher

445 yields than their bulk counterparts. They attributed this to the greater total reactive surface area of the Q-particles. Also, when comparing the efficiencies of the three photocatalysts tested, they found that there was a correlation between the semiconductor photoinitiation efficiency and the reduction potential of its conduction band electrons. A comparison of the quantum yields for polymerization revealed that the rates increased in the following order: TiO2 , ZnO, CdS. This order paralleled an increase in the conduction-band potential (−0.1, −0.2, −0.4 V versus SCE, respectively) (Hoffmann et al., 1992b) (The potential of the SCE relative to the SHE is 0.2444 V at 25◦ C). Another area where photocatalysis is being applied is in the photocatalytic reductions of CO2 . Kanemoto et al. (1992) suggested the importance of the surface structures of the nanocrystallites in CO2 photoreductions. They deduced this from surface modification studies, which involved the addition of Zn2+ to a system of ZnS nanocrystallites. This addition resulted in product switching from formate to CO without loss in efficiency. In 1997, the same group carried out an investigation of a CdS photocatalytic system for CO2 reduction, which examined the role of the surface structures of the CdS nanocrystallites. They observed a marked improvement of photocatalytic activity through the formation of sulphur vacancies. They proposed that the CO formation process occurred via adsorption of CO2 to a Cd atom in the vicinity of a sulphur vacancy (Fujiwara et al., 1997). The application of nanoparticles in photocatalysts became a reality with the introduction of the now commercially available HombiKat UV 100 from Sachtleben Chemie GmbH. This novel photocatalyst consists of 5 nm TiO2 anatase particles forming micronsized aggregate structures. The activity of this novel material was investigated using dichloroacetic acid (DCA) as the model organic compound, by Lindner et al. (1995). The high photocatalytic activity exhibited by this new photocatalyst, compared to another commercially available catalyst, was explained by its very high surface area (> 250 m2 /g). The authors found, however, that higher concentrations of Hombikat UV 100 (5g/L) were required to reach its optimal performance as compared with 0.5 g/L of the other photocatalyst. They attributed this to the slightly blue shifted spectral response of this novel photocatalyst caused by its very small primary particles, although no explanation was provided regarding this conclusion.

Sato et al. (1996a) studied the photochemical reduction of nitrate to ammonia with and without methanol using layered hydrous titanate/cadmium sulphide (H2 Ti4 O9 /CdS) nanocomposites. The addition of methanol was found to be useful for the promotion of this reduction. Doping with Pt particles in the interlayer was also found to greatly increase the catalytic activity of the nanocomposites. This group has also looked at the photocatalytic properties of layered hydrous titanium oxide/CdS–ZnS nanocomposites incorporating CdS–ZnS into the interlayer. The hydrogen production activity, which was observed for these nanocomposites following irradiation with visible light, was found to be higher than that of unsupported CdS–ZnS (Sato et al., 1996b). Another area where considerable interest has been shown is the application of photocatalysis to silver recovery, for example from waste photographic effluents. Recently Sahyun et al. (1997) studied the photocatalytic deposition of silver from a silver salt ethanol solution onto TiO2 nanoparticles prepared with a chemisorbed surface alkoxide. From picosecondresolved transient absorption spectroscopy, the authors deduced that the rate of formation of the colloid silver deposit is determined by the reduction of Ag(I) by surface-trapped photoelectron states (Ti(III)). Size-quantized CdS (∼ 20 Å) nanocrystals have been used as photocatalysts for nitrate reduction (Korgel et al., 1997). Due to the effects of quantum confinement on electron and hole redox potentials, the nitrate reduction rates were found to depend strongly on the apparent particle size, with the fastest reduction rates being observed with the smallest nanocrystals. In the absence of an electron donor other than water, rapid photocorrosion was observed; therefore, formate was used as the sacrificial electron donor. Ogura et al. (1997) carried out tests using series of hexatitanates, M2 Ti6 O13 (M denotes Na, K, or Rb). The authors compared the photocatalytic activity of the hexatitanates supported on RuO2 , with the efficiency for the production of photoexcited charges. The test reaction was water decomposition. The hexatitanates showed high photocatalytic activities and high photoexcited radical formation (followed with ESR measurements). The results indicated a close relationship between the photocatalytic activity and the efficiency of electrons and holes production. The use of iron oxides as photocatalysts has been given particular attention due to their absorptivity in the visible region, abundance, and low cost. Faust et al.

446 (1989) examined the suitability α-Fe2 O3 as a photocatalyst by studying the kinetics and mechanisms of the photocatalytic oxidation of sulfur dioxide in aqueous colloidal suspensions of 3–25 nm α-Fe2 O3 . The results showed that the small hematite crystals possessed photocatalytic activity for the oxidation of sulfite (S(IV)), which readily depleted when the colloidal solutions containing 1 mM S(IV), and 0.1 mM of the nano α-Fe2 O3 particles were illuminated with light, of λ = 320 nm, in the presence of air. Kormann et al. (1989) also examined the suitability of α-Fe2 O3 as photocatalysts using transparent α-Fe2 O3 3–20 nm in size. They also compared the photocatalytic activity of hematite to the activities of colloids and suspensions of ZnO and TiO2 . While ZnO and TiO2 were found to be quite active photo-oxidation catalysts in the formation of hydrogen peroxide and in the degradation of chlorinated hydrocarbon molecules, only negligible photocatalytic activity was found for α-Fe2 O3 . Bahnemann (1993) has also carried out several tests to compare the efficiencies of several semiconductors as photocatalysts. This work involved the comparison of ultrasmall zinc oxide, titanium dioxide, hematite and titanium/iron mixed oxide particles. Q-size effects were observed during particle growth and at the final stages of synthesis of all of the oxides studied. While zinc oxide, titanium dioxide and titanium/iron mixed oxide particles exhibited considerable photocatalytic activity, hematite particles were only found to oxidize S(IV), with the quantum yields (φ) obtained being < 0.3. An explanation involving surface-bound molecules and free radical intermediates was provided to account for these differences in reactivity. In 1998, Cherepy et al. carried out ultrafast kinetic studies of photoexcited charge carriers dynamics in γ - and α-Fe2 O3 semiconductor nanoparticles using femtosecond laser spectroscopy. The aim of this work was to explain the observed low photoactivities of these iron oxide semiconductors (Cherepy et al., 1998). The results showed that the visible/near UV light induced charge separation of the photoexcited electrons in Fe2 O3 nanoparticles was very short-lived and the decay was primarily non-radiative. A power dependence, which reveals the relaxation mechanism for the early time decay, was also found to be very different from that in other semiconductor nanoparticles such as, CdS or TiO2 . The authors saw these results as providing some insight into the poor efficiency of Fe2 O3 as a photocatalyst. They also proposed that the identification of materials, which have an intrinsically slow decay

of photoexcited electrons, would help in the search of high-efficiency photocatalysts and photoelectrodes. Enhancement of photoactivity of nano-photocatalysts Under bandgap excitation, metal oxide semiconductor particles behave as short-circuited electrodes, with both oxidation and reduction processes occurring on their surfaces (Kamat, 1995). Thus two critical processes determine the overall quantum efficiency for interfacial charge transfer. These are the competition between the recombination and the trapping of the charge carriers, followed by the competition between the recombination of the trapped carriers and the interfacial charge transfer (Hoffmann et al., 1995). At least in colloidal sols of TiO2 , it was found that the charge carriers undergo 90% recombination ∼ 1 ns after excitation (Serpone et al., 1996). The authors concluded that the quantum yields of any surface photoredox reaction could not be greater than ∼ 10%. In order to enhance the photoactivity of bulk and colloidal TiO2 particles, interfacial charge-transfer reactions need to be enhanced. Improved charge separation and inhibition of charge carrier recombination is essential in improving the overall quantum efficiency for interfacial charge transfer (Bedja et al., 1995). This can be achieved by modifying the properties of the particles by selective surface treatment (Hoffmann et al., 1995). Several approaches have been taken to achieve this. These have included surface modification of the semiconductor particles with redox couples or noble metals (Choi et al., 1994, Aruna et al., 1996). Bahnemann et al. (1984) have shown that the efficiency of charge transfer at the semiconductor–electrolyte can also be improved by simultaneous scavenging of holes and electrons by surface adsorbed redox species. Another approach has involved the coupling of two semiconductor particles with different electronic energy levels (Bedja et al., 1995). Metal ion dopants Depositing or incorporating metal ion dopants into the titanium dioxide particles can influence the performance of these photocatalysts. This affects the dynamics of electron:hole recombination and interfacial charge transfer. The largest enhancement of photoactivity through doping was found in nanosized particles, in which the dopant ions are located within

447 1–2 nm of the surface (Choi et al., 1994). Also, the high surface areas characteristic of nanoparticles (100– 500 m2 /g) appear to enhance the deposition process and the resulting activity of the catalyst (Zeltner et al., 1996). The work of Choi et al. involved a systematic study of the effects of 21 different metal ion dopants on nanocrystalline TiO2 . Chloroform oxidation and carbon tetrachloride reductions were used as photoreactivity tests. Their results showed that some doped Q-TiO2 particles had much greater photoactivity than their undoped counterparts. Doping with Fe(III), Mo(V), Ru(III), Os(III), Re(V), V(IV) and Rh(III) at the 0.5 at% level in the TiO2 matrix, significantly improved the photoreactivity for both oxidation and reduction. Choi used laser flash photolysis and time-resolved microwave conductivity measurements to correlate the effects of metal ion dopants to the lifetime of the photoexcited electron. In the Fe(III), V(IV), Mo(V) and Ru(III) doped samples, the lifetime of the generated electron was found to have increased to 50 ms compared to < 200 µs with the undoped Q-TiO2 . There was also an apparent linear correlation between the oxidation quantum yield (of chloroform) and the reduction quantum yield (of carbon tetrachloride) regardless of the nature of the dopant. This ion doping of large bandgap semiconductor colloids might not always be effective in lengthening the lifetime of the generated charge carriers. Recently, Smith et al. (1998) showed that in Ru(III) doped TiO2 colloids, the electronic decay was as fast as or even faster than in undoped TiO2 . The difference between the studies carried out by Smith et al. and those carried out by Choi et al., was the higher dopant level of Ru (III) of 3 at% used by Smith et al., compared with the 0.5 at% dopant level used by Choi et al. There could be several reasons for the variations in the reported effects of the dopant ions. One reason for these variations is the location and co-ordination of the dopant ions. These depend critically on the methods of sample preparation and pre-treatment as well as the concentration of the dopant ions. The dopant ions may be adsorbed on the surface, they may be incorporated into the interior of the particle on firing, or they may form separate oxide phases (Howe, 1998). The dopant ions can function as both hole and electron traps or they can mediate interfacial charge transfer (Choi et al., 1994). Once incorporated into the interior of the TiO2 , the dopant ions may occupy either

lattice (substitutional) or interstitial sites. Their ability to function as trap sites and/or to mediate interfacial charge transfer will depend on these factors (Howe, 1998). When incorporated in the interior of the particles, the d-electronic configuration of the dopant and its energy level within the TiO2 lattice also seem to significantly influence the photoactivity (Martin et al., 1995). Finally, the site where the electron gets trapped greatly influences the redox chemistry of the doped semiconductor. A dopant ion might act as an electron trap, and this might in fact lead to a lengthening in the lifetime of the generated charge carriers, resulting in an enhancement in photoactivity. However if an electron is trapped in a deep trapping site, it will have a longer lifetime, but it may also have a lower redox potential. This might result in a decrease in the photoreactivity (Hoffmann et al., 1995). The work carried out by Zhang et al. (1998) shed a new light on the role of dopant ions and their effect on photoactivity. Firstly, these authors provided further support for the existence of an optimum dopant concentration. Their main finding however was that this optimum concentration is particle-size dependent and decreases with an increase in size. The system they studied was Fe3+ doped TiO2 for the photocatalytic degradation of CHCl3 . They observed that for 6 nm particles, the optimum Fe concentration was 0.2 at%, while for 11 nm particles, the optimum concentration was 0.05 at%. They provided the following explanation for their observations. Their first explanation was with regards the existence of an optimal Fe3+ dopant concentration. Fe3+ ions serve as shallow trapping sites for the charge carriers and increase the photocatalytic efficiency by separating the arrival time of e− and h+ at the surface. If Fe3+ can act as a trap for both e− and h+ , at high dopant concentration, the possibility of charge trapping is high, and as such, the charge carriers may recombine through quantum tunneling. If Fe3+ acts as a h+ trap only, the recombination of the charge carriers is not of great concern at low dopant concentrations. At high concentrations however, a h+ may be trapped more than once as it tries to make its way to the surface. This hole which had been ‘held back’, might then recombine with an electron which is generated by a subsequent photon before it can reach the surface (i.e. increased incidence of volume recombination). Thus there exists an optimum Fe3+ concentration whether the Fe3+ acts as an e− and h+ trap or as a h+ trap only.

448 With regards to the optimum concentration depending on the particle size, Zhang et al. suggested that when the particle size becomes larger, the average path length of a charge carrier to the surface is longer. Thus, for a constant dopant concentration, the longer the path length which the charge carrier needs to travel, the higher the probability of meeting a dopant ion, and hence the greater the chance of multiple trappings. This multiple trapping leads to increased volume recombination. A lowering in the dopant concentration reduces the chance of multiple trappings for a larger particle. Thus the optimal Fe3+ dopant concentration should decrease with increasing TiO2 particle size. Dual semiconductor systems Another approach taken to modify the surface of semiconductor colloids, with an aim to improve charge separation and minimize or inhibit charge-carrier recombination, has been to dope with a second semiconductor. Excitation of these dual semiconductors results in an electron injection into the lower lying conduction band of the second semiconductor. In the composite nanoparticles, no electric field is necessary, as the charge separation is achieved by the tunneling of electrons (Henglein, 1997). Recent studies of these interparticle electron transfer occurs within 500 fs–2 ps (Evans et al., 1994). Henglein presented the first example in the literature of composite particles when he found that when small amounts of Cd2+ were added to ZnS, the ZnS bandgap fluorescence was quenched (Henglein, 1984). Since then there have been many papers published regarding the optical properties of mixed systems. Some of the systems studied include ZnS–CdS (Ueno et al., 1985), CdS–Ag2 S (Henglein et al., 1987), CdS–Ag2 S (Spanhel et al., 1987a), mixed crystals of Znx Cd1−x S, CdS–ZnS (Youn et al., 1988), AgI–Ag2 S (Henglein et al., 1989), ZnS–CdSe (Kortan et al., 1990) and CdS– PbS systems (Zhou et al., 1993). Reber et al. (1986) reported that modifying the surfaces of platinized CdS powders with silver ions could activate these photocatalysts, which are otherwise inactive with respect to H2 formation. The activation was attributed to the formation of a heterojunction between CdS and Ag2 S. It was also observed that the fluorescence of CdS was quenched by Ag+ ions, and that the spectral response for H2 formation was extended to wavelengths up to ∼ 600 nm where CdS itself does not absorb. Recently, emphasis has been placed on the development of coupled and capped semiconductor systems

and their application in photocatalysis. Many papers have been published regarding coupled semiconductors systems. These include CdS–TiO2 , CdS–ZnO (Spanhel et al., 1987b), CdS–Ag2 S (Spanhel et al., 1987), ZnO–ZnS (Rabani, 1989), ZnO–ZnSe (Kamat et al., 1992), AgI–Ag2 S (Henglein et al., 1989) and CdS–HgS (Haesselbarth et al., 1992). The charge separation mechanism in both capped semiconductor systems and coupled semiconductor systems involves the photogenerated electrons in one semiconductor being injected into the lower lying conduction band of the second semiconductor. However, the interfacial charge transfer is significantly different (Bedja et al., 1995). The charge-transfer processes involved in capped and coupled semiconductor systems are shown in Figures 2 and 3 respectively. In a coupled semiconductor system the two particles are in contact with each other and both holes and electrons are accessible on the surface for selective oxidation and reduction processes. Capped semiconductors on the other hand have a core and shell geometry. The electron gets injected into the energy levels of the core semiconductor (provided it has a conduction band potential which is lower than that of the shell). The electron hence gets trapped within the core particle, and is not readily accessible for the reduction reaction. In 1995 Bedja et al. synthesized TiO2 -capped SnO2 (SnO2 @TiO2 ) and TiO2 -capped SiO2 (SiO2 @TiO2 ) nanocrystallites. The photocatalytic properties of the capped semiconductor systems were tested for the oxidation of I− and SCN− . The SnO2 @TiO2 colloids were 80–100 Å in diameter and exhibited improved photocatalytic efficiencies compared to the uncapped TiO2 colloid. The quantum efficiency for I− oxidation was improved by a factor of 2–3 upon capping the SnO2 colloids with TiO2 . The improved charge separation in the SnO2 @TiO2 system was confirmed by laser flash photolysis experiments. The photocatalytic properties of SiO2 @TiO2 colloids were found to be similar to those of uncapped TiO2 colloids (Bedja et al., 1995). Similar work is being carried out by Amal et al., with this group being involved in the development of two novel photocatalysts. The first photocatalyst is α-Fe2 O3 @TiO2 , with an aim to expand the catalyst’s photoresponse into the visible region of the solar spectrum (Penpolcharoen et al., 1998). The second is Fe3 O4 @TiO2 , with an aim to enhance the catalyst’s separation properties from the treated water by imparting magnetic properties onto the photocatalyst (Beydoun et al., 1998).

449

Figure 2. Charge transfer in a coupled semiconductor system.

Figure 3. Charge transfer in a capped semiconductor system.

The development of these heterogeneous semiconductor systems is very promising and has the potential to contribute significantly to the area of photocatalysis. By changing certain parameters, such as the thickness of the shell or the particle radius of the core, this approach may allow the tailoring of important properties, such as photocatalytic, optical, and magnetic properties, of the photocatalyst. It may also be important in addressing problems such as photodissolution, which might otherwise occur in an ‘unprotected’ photocatalyst such as iron oxide (Hoffmann et al., 1995). In capped systems, if the thickness of the shell deposited is sufficiently thick, relative to the core radius, the individual identity of the two semiconductors can be maintained (Bedja et al., 1995). In these systems, only the holes are accessible at the surface of the photocatalyst. This allows for selective interfacial charge transfer, and improves the efficiencies of the oxidation reactions. However, one disadvantage of this approach is the fact that the reducing charge carriers (the electrons) are not being utilized during the

photocatalyic reaction, since they are being accumulated within the core semiconductor. Therefore these photocatalysts cannot be used for photoreduction reactions or reactions in which superoxide radicals (formed by the reduction of oxygen molecules) play an important role. Three-layered colloidal particles are another development in the field of surface-modified semiconductor nanoparticles. These consist of a quantum-sized semiconductor particle as the core, covered by several layers of another semiconductor material, onto which several layers of the core material are then deposited, and act as the outermost shell. These particles are called quantum dots or wells (Weller et al. (1995). The first example described in the literature was the system CdS–HgS– CdS (Mews et al., 1994). Another recent development in the application of nanoparticles in photocatalysis has been the emergence of organic–inorganic nanostructured composites. Interactions between organic and inorganic molecules are being used to generate a range of materials for catalytic technologies. Published work by Braun et al. (1996) describes the synthesis of stable semiconductororganic superlattices based on cadmium sulphide and cadmium selenide. By incorporating organic molecules in an inorganic lattice the authors anticipate that the electronic properties of this type of materials can be tailored. Therefore, these novel organic–inorganic nanostructured composites may be suitable for photocatalytic applications. Tenne et al. (1996) have also been working on the preparation of inorganic compounds, namely WSe2 and PtS2 , with a crystal structure similar to graphite. These compounds can be used to construct fullerene-like structures and nanotubes with potential applications in photocatalysis and nanoelectronics.

450 Sensitization of TiO2 Despite titanium dioxide being the most promising photocatalyst, it can only be activated by light of wavelength 390 nm or lower. The UV region (below 390 nm) constitutes only ∼ 4% of the energy available within the solar spectrum. Thus considerable research is being carried out to extend this photocatalyst’s response into the visible part of the solar spectrum. The sensitization of TiO2 with a second component to enhance activity and shift the wavelength of irradiation into the visible region is a goal that has been eagerly pursued. Again, several approaches have been taken. The first involves the sensitization of TiO2 with organic and organometallic dyes, these however are seen to be are less likely to succeed in photocatalysis due to their instability (Howe, 1998). Another approach has been to utilize narrowbandgap semiconductors to photosensitize TiO2 . The sensitizer has a high cross section for the absorption of visible light and injects electrons into the lower conduction band of the wide-bandgap semiconductor. The separated charge carriers can be used to initiate the chemical reactions. This principle has been used to construct photoactive layers, in which charge separation is achieved with an efficiency of 80% (Henglein, 1997a). Nanocrystalline, narrow-bandgap semiconductors can be used as sensitizers. These semiconductors can be tailored to suit specific purposes since changing their size can shift their electronic bands (Q-size effects). This concept is seen as more attractive alternative for TiO2 sensitization since they are more stable than organic dye (Howe, 1998). Sensitizing TiO2 with Q-sized narrow-bandgap semiconductors PbS and CdS was carried out by Vogel et al. (Vogel et al., 1994). These systems however, demonstrated a loss in efficiency, with photocorrosion of the narrow-bandgap semiconductors in aqueous media, under illumination with 460 nm light, also being a problem. Howe suggested the possibility of using such semiconductor sensitized nanocrystalline TiO2 for gas-phase photocatalysis (Howe, 1998). A different approach was taken by Bahnemann et al. in 1993 in order to synthesize particles which utilize parts of the visible solar spectrum when acting as photocatalysts. This approach involved the preparation of a mixed Ti(IV)/Fe(III) oxide catalyst. Not only did this photocatalyst have increased activity for the destruction of DCA (dichloroacetic acid), it also showed a photoresponse to 450 nm light. The prepared particles contained between 0.05 and 50 at% Fe(III) in their

matrix, and had particle diameters between 4 and 6 nm. The photoresponse shifted to longer wavelengths with an increasing content of ferric ions. However, this novel photocatalyst also suffered from the problem of a partial reductive dissolution upon illumination. This cathodic corrosion competes with the electron transfer to oxidants such as oxygen found in the system. In order to alleviate the problem of photodissolution, the authors suggested the introduction of H2 O2 , as an electron acceptor, into the system. H2 O2 being a better electron acceptor than O2 can be used to compete more efficiently with the photocatalyst’s dissolution (Bahnemann et al., 1993). Nanocrystalline films Another area of research that links the nanotechnology with heterogeneous photochemistry has been the development of semiconductor nanocrystalline films. Nanocrystalline semiconductor films constitute a network, in which electronic conduction can take place. The films are highly porous, and the pores between the particles are filled with a conducting medium such as electrolyte or semiconductor, forming junctions of extremely large area contact (Henglein, 1997). The thin films exhibit interesting photocatalytic and photoelectrochemical properties that are inherited from the native colloids (Kamat, 1995). Chemical vapour deposition or molecular beam epitaxy has been the preferred technique for depositing thin semiconductor films. The nanosized particles from which the films are made are in electronic contact allowing for electric charge percolation through such films. This charge transport is highly efficient, with the quantum yield being practically unity (Hagfeldt et al., 1995). One of the major advantages of nanocrystalline semiconductors is their high porosity. This facilitates surface modification with sensitizers, redox couples, and other semiconductors. Using nanocrystalline semiconductor films also allows the manipulation of the photocatalytic process by electrochemical methods (Kamat, 1995). Dye sensitization. Nanostructured films made up of wide bandgap semiconductors, such as TiO2 , respond only in the UV region. Sensitizing with dyes that absorb strongly in the visible spectrum, has been one method to extend the photoresponse of such films. In a porous film consisting of nanometre-sized TiO2 particles, the effective surface area can be enhanced 1000-fold, thus making light absorption efficient even with only a dye monolayer adsorbed on each particle (Hagfeldt et al.,

451 1995). Once photoexcited, the dye molecules inject electrons into the semiconductor’s conduction band. Recent developments in nanocrystalline electrochemical photovoltaic devices has meant that vast research efforts have been invested into optimising the properties of such films. O’Regan and Gratzel identified the improvement of light harvesting as the key to improving the efficiency of the photoelectrochemical cell. On a smooth surface, a monomolecular layer of sensitizer absorbs less than 1% of incident monochromatic light (O’Regan et al., 1991). O’Regan and Gratzel’s efforts led to the development and commercialization of a device capable of harvesting 46% of the incident solar energy flux. It was based on a 10 µm thick, optically transparent semiconductor film consisting of nanocrystalline TiO2 particles, coated with a monolayer of a Ru-based sensitizer dye. The absorption onset was shifted to 750 nm, with the light harvesting efficiency reading almost 100% in the whole visible region below 550 nm. Efforts have continued to try and improve the quantum yield of current photoelectrochemical devices. Recently, Salafsky et al. (1998) found that in the presence of an electron donor to the dye cation (I− /I3− redox couple), the injected charge is prevented from directly recombining with the cation and thus has a longer lifetime.

External bias/photoelectrochemical devices. Illuminated semiconductor particles behave as shortcircuited electrodes, with the interfacial charge transfer competing with the charge recombination process. The high rate of recombination between the charge carriers is a major problem in the photocatalytic degradation of organics and is a limiting factor in controlling the photocatalytic efficiency. During the past few years, considerable effort has been made in the preparation of nanoparticle films and their application in photoelectrochemical devices. In an electrochemically assisted photocatalytic process, the thin nanocrystalline semiconductor film is coated on a conducting glass surface, with the generated electrons being driven via an external circuit to a counter electrode by applying a positive bias. This leads to better charge separation and the problem of charge recombination is easily overcome (Kamat, 1995). In most of the photocatalytic reactions oxygen is essential for scavenging electrons from the irradiated semiconductor particle (Vinodgopal et al., 1996). Thus, an advantage of an electrochemically assisted

photocatalytic process is that oxygen is no longer required as an electron scavenger. This makes it possible to carry out the photocatalytic reaction under anaerobic conditions (unless O2 plays a role in the reaction mechanism of organic degradation (Schwitzgebel et al., 1995)). Another advantage is that the anodic and cathodic processes are separate. Photoelectrochemical devices can thus allow the isolation of the various reactions occurring in photocatalytic systems and provide a means to carry out selective oxidation and reduction in two separate compartments (Kamat, 1995, Viodgopal et al., 1996). Several people have studied the preparation of photoelectrochemically active thin films, using a number of different semiconductor materials. Films have been prepared from colloidal suspensions of ZnO (Spanhel et al., 1991), TiO2 (O’Regan et al., 1991), WO3 (Bedja et al., 1993; 1994a) and SnO2 (Bedja et al., 1994b). In 1997, Fu et al. studied the structure and electrochemically assisted photocatalysis of nanosize TiO2 /Pt/glass thin films prepared by the sol–gel process. Their results showed that a higher photocatalytic activity was obtained when an anodic bias is applied, with the activity becoming stronger with increasing anodic bias. The interlayer Pt and intensity of light source were also shown to affect the photocatalytic activities of the films (Fu et al., 1997). The effectiveness of the electrochemically assisted photocatalytic process have also been tested for the degradation of using 4-chlorophenol and Acid Orange 7 in aqueous solutions by Viodgopal et al. (Viodgopal et al., 1995; 1996). The semiconductor films used were SnO2 , TiO2 and a SnO2 –TiO2 nanoclusters. The authors found that combining photocatalysis and electrochemistry leads to improved photocatalytic degradation rates (even in the absence of oxygen). The highest rates were observed with the SnO2 /TiO2 coupled semiconductor films, with a ten-fold enhancement in the degradation rate being observed at an applied bias potential of 0.83 V versus SCE. The role of the coupled semiconductor was to further improve the charge separation. The development of multicomponent nanocrystalline semiconductor films is recognised as being of extreme importance to research in photoelectrochemistry (Kamat, 1995). These systems can serve two purposes. The first purpose is to allow the extension of the photoresponse of a wide-bandgap semiconductor by coupling it with a narrow-bandgap semiconductor. The second purpose is to inhibit (or slow down) the charge-carrier recombination, with enhanced charge

452 separation upon the injection of electrons from the conduction band of the narrow-bandgap semiconductor into the lower conduction band of the wide-bandgap semiconductor (Bedja et al., 1995). Coupled semiconductor systems in which increased photocurrent generation has been demonstrated include TiO2 –CdS (Vogel et al., 1994), ZnO–CdS (Hotchandani et al., 1992), TiO2 –CdSe (Liu et al., 1993) and TiO2 –SnO2 (Bedja et al., 1995). Hinogami et al. (1997) studied the modification of semiconductor surfaces with ultrafine metal particles for the efficient photoelectrochemical reduction of carbon dioxide. A p-type silicon was modified with small metal (Cu, Au and Ag) particles. The holes generated enter into the interior of the particle, while the photogenerated electrons come out to the surface. These electrons then transfer into the metal particles which reduce CO2 with the aid of their catalytic activity. The electrode has both high activity and very low surface recombination rate (Hinogami et al., 1997).

Using nanoparticles to study the mechanisms of photocatalytic reactions Interest in the detailed mechanism of illuminated semiconductor reactions arises from their intensive use in the photodegradation of organic pollutants (Rabani et al., 1998). Mechanistic studies of photocatalysed reactions are of paramount importance both for a fundamental understanding of the nature of heterogeneous photocatalysis and for optimizing its efficiency (Lepore et al., 1993). The electronic states of molecules play an important role in photochemistry. The population and lifetime of the charge carriers and the time scale of the interfacial charge-transfer process determining the overall efficiency of heterogeneous photocatalysis (Colombo et al., 1996). Thus, for a clear understanding of the redox processes across semiconductor– electrolyte interfaces, it is highly desirable to obtain detailed information on the dynamics of the photoinduced electron transfer by fast kinetic spectroscopy (Kalyanasundaram, 1983). Time-resolved optical spectroscopy and quantum yield measurements are usually the most useful sources of information in mechanistic photochemistry. Optical methods are difficult to employ when studying heterogeneous photocatalytic slurries. Due to the light scattering in these opaque systems, it is not easy

to determine accurately the irradiation light intensity absorbed by the particles. This makes it difficult to obtain accurate measurements of both the absorption spectra and photochemical quantum yields (Lepore et al., 1993). Another problem is the complexity of such a small-scale photoelectrochemical system. This has meant that details of the underlying reaction mechanisms of photocatalysis have not been well understood (Bahnemann et al., 1997). The use of nanoparticles has lead the way in the approaches taken to provide more definitive mechanistic information, particularly for TiO2 photocatalysts. Nanosize semiconductor particles have diameters smaller than the wavelength of the incident light. Their dispersions are optically transparent with the dimensions of the particles being small enough to render scattering of light negligibly small. Thus nanosize semiconductor particles have the added advantage of providing transparent solutions for detailed mechanistic studies of photoredox processes by laser photolysis techniques. Due to the quantum effects, the luminescence of colloidal semiconductors is strongly influenced by particle size. Typically, a broad emission is observed which blue shifts with decreasing radius of the particle. Luminescence analysis provides an important tool to study the dynamics of charge-carrier recombination in such colloidal semiconductors (Hagfeldt et al., 1995). Thus, with nanosize semiconductors, it is possible to directly monitor, by flash photolysis, the charge injection into or from the semiconductor to reagents present in solution (Kalyanasundaram, 1983). This makes it possible to obtain absorption spectra of the material and, therefore, the kinetics of processes taking place can be recorded, if they are reflected in a change of the absorption (Bockelmann et al., 1996). This enables elementary steps involved in the catalytic process to be conveniently analysed (Hagfeldt, 1995). Many mechanistic and spectroscopic studies have used optically transparent solutions of colloidal TiO2 . These investigations have led to a detailed understanding of the nature and reactivity of charge carriers in these colloidal particles (Lepore et al., 1993). Laser flash photolysis technique has been used to quantify the kinetics of various mechanistic steps (Howe, 1998) and has allowed the identification of factors that govern the dynamics of interfacial electron and hole transfer reactions (Hagfeldt et al., 1995). Early studies in 1985 used pico and nanosecond transient absorption measurements to monitor the fate of

453 photo-produced holes and electrons. Electron trapping was reported to be faster than 30 ps, the time resolution of the laser system used. The trapping of holes was estimated to occur within < 250 ns. Electron:hole recombination was estimated to occur within 30 ns (Rothenberger et al., 1985). It is important to note that early flash photolysis studies suffered from the problem of a large number of electron–hole pairs being generated in each TiO2 per laser pulse (Bahnemann, 1997). In 1995, using nanosized TiO2 , Skinner et al. were able to employ experimental conditions that ensured that on the average less than one electron–hole pair was formed per semiconductor cluster during each light pulse. The authors carried out femto spectroscopic measurements, and under the well-defined experimental conditions, this research group found that electron-trapping time was approximately 180 ± 10 fs. They also found that the electron:hole recombination occurred on a picosecond time scale rather than nanosecond, with the mean lifetime for the electron–hole pair was found to be 30 ps not 30 ns as reported earlier (Skinner et al., 1995). Serpone et al. (1996) carried out a picosecond study on colloidal solutions TiO2 . They reported a rapid trapping of electrons and holes in less than 30 ps. The second-order kinetics measured in their experiments revealed two decay processes. A fast decay, which was assigned to recombination of charge carriers in shallow traps. The longer decay occurred on a much longer time scale, with the recombination occurring at times greater than 20 ns. This decay was attributed to deeply trapped charge-carrier recombination. Serpone et al. proposed that it is the longer decay which determines the photocatalytic activity. Colombo et al. (1996) examined the charge-carrier dynamics suspensions of Degussa P25 TiO2 , which they probed with femtosecond time-resolved diffuse reflectance spectroscopy. They also studied transient events in dry P25 TiO2 powders. They compared the electron trapping in both systems (wet and dry powder) to electron trapping in quantum sized TiO2 . In all three systems the electron trapping was very rapid and occurred in less than 200 fs. Colombo et al. also examined the interfacial holetransfer dynamics with P25/SCN− complexes being probed as a function of thiocyanate ion concentration. The presence of thiocyanate adsorbed to the surface of the P25 nanocrystalline particles dramatically influences the recombination kinetics. Both an increase in signal strength, which meant an increase

in the population of trapped charge carriers, and the much slower decay in the transients indicated this. This demonstrated the hole-scavenging ability of the SCN− . It also demonstrated that the interfacial charge transfer, of an electron from the SCN− to a hole on the photoexcited TiO2 , effectively competes with electron–hole recombination. It is known that it is the number of holes available for the interfacial reaction, which ultimately determines the efficiency of a photocatalytic process. Thus, this study presented a major finding which can have significant implications on photocatalysis. The authors showed that it was possible to adsorb species with electron-scavenging ability such that the hole-transfer reaction can successfully compete with the picosecond electron–hole recombination process. Nosaka et al. (1990) studied the electron transfer from a laser-excited CdS colloid particle to surfaceadsorbed viologens (organic reducing agents). The quantum yield of the photoinduced electron transfer in CdS and In2 S3 semiconductor particles to surfaceadsorbed violgen molecules, was measured as a function of the density of adsorbed photons in the particles. The authors found that at a lower photon density, the quantum yields became constant, with the smaller particles having a lower quantum yield. The authors presented a kinetic model, which explained the decrease of quantum yield with the decreasing diameter of the particle. In their model, the ratio of the rate constant for the electron–hole pair recombination (Kr) to the rate constant for photoinduced electron transfer (Ke) was found to be dependent on the size of the semiconductor particle. For CdS particles the value of Kr/Ke increases by a factor of about 20 with doubling of the radius. For In2 S3 the dependence of Kr/Ke on the radius was found to be very weak. Kr/Ke increases only by two times with doubling of the radius. From their kinetic model the authors proposed that when the electron-transfer rate (Ke) does not increase much with decreasing size, the value of the quantum yield decreases because the electron–hole recombination rate (Kr) probably increases largely. As was previously mentioned, Lepore et al. (1993) were the first to provide accurately determined quantum yields of polycrystalline TiO2 (with an average radius of 62 nm). The quantum yields obtained with these samples were similar to those of P25, but higher than those obtained using the quantum-sized (∼ 2.4 nm) TiO2 . A longer lifetime of the charge carriers in the polycrystalline samples and the P25 slur-

454 ries, compared to the quantum-sized TiO2 solutions, was found to be the reason behind this. Martin et al. (1995) studied the charge-carrier dynamics of Q-TiO2 and Degussa P25 using time-resolved microwave conductivity. They found that the resultant interfacial electron transfer was faster for P25 than for Q-TiO2 . Schwitzgebel et al. (1995) studied the role of oxygen and the generated electrons in photocatalysed oxidation of organics. The organics oxidized were n-octane, 3-octanol, 3-octanone, or n-octanoic acid, with the oxidation carried out using buoyant nanocrystalline n-TiO2 -coated glass microbubbles. The effect of dissolved Fe3+ ions on the reaction was also studied. Their results showed that both the photogenerated holes and electrons participate in the oxidation reaction. They also found that molecular oxygen has two roles. Firstly it accepts the electron and is reduced to a superoxide radical (O2 •− or HO2 • ). Secondly it combines with the organic radical, generated upon the reaction between the organic and the hole or OH radical reaction. This produces an organoperoxy radical (ROO• ). The superoxide radical, which is a relatively ineffective oxidizing agent, combines with the organoperoxy radicals to form an unstable tetraoxide that decomposes to give CO2 . The effect of the dissolved Fe3+ ions was to reduce the CO2 yields in the photocatalytic air oxidation of the four reactants. This was due to the fact that they competed for the photogenerated electrons and oxidized the superoxide to O2 . In reactions that do not involve the superoxide radical, for example the photocatalytic air oxidation of n-octanal, the oxidation reaction was not inhibited by Fe3+ . Bockelmann et al. (1996) performed detailed photophysical and photochemical investigations of colloidal Ti/Fe mixed oxide suspensions using a flash photolysis apparatus for time-resolved measurements. In their efforts to understand the process of photodissolution of this mixed semiconductor, they found that during photodissolution the photogenerated electrons are not mainly captured at Ti4+ centres but in fact reduced Fe3+ centres within the microcrystals forming Fe2+ . More recently, in 1997, Bahnemann et al. carried out time-resolved laser flash photolysis under well-defined experimental conditions, which resemble the steady-state irradiation conditions in photocatalysis. Bahnemann et al. used nanosized TiO2 particles, of mean diameter 24 Å, to investigate the details of

the photocatalytic oxidation of the model compounds DCA− and SCN− . The authors found that in the absence of hole scavengers, the electrons were trapped instantaneously (i.e. within the duration of the laser flash). The group suggested that at least two different types of hole traps need to be considered. Deeply trapped holes, which are long-lived and unreactive, and shallowly trapped holes, which exhibit a very high oxidation potential. The traps are interstitial atoms, impurities, dislocations, and defects. (Weller et al. (1995) suggested that traps in nanosized particles were found mainly on the surface as a result of the large surfaceto-bulk ratio). In the same study Bahnemann et al. also investigated the electron transfer kinetics to donors and acceptors. They found that the adsorption of the model compounds DCA and SCN on the TiO2 surface prior to the bandgap excitation, was a prerequisite for an efficient hole scavenging (Bahnemann et al., 1997). Smith et al. (1997) have also studied the ultra-fast dynamics of photoinduced electrons in several metal and semiconductor colloidal nanoparticle systems using femtosecond laser spectroscopy. For aqueous gold and silver colloids, they found that spatial size confinement caused slower electronic relaxation due to reduction of non-equilibrium electron transport and weaker electron–phonon coupling. The effect of surfactant on the electron dynamics in CdS colloids was also examined and found to be significant. The authors put forward that this finding may provide evidence to suggest that the electrons were dominantly trapped at the liquid–solid interface. Rabani et al. (1998) studied nanosecond kinetics of dry TiO2 using pulsed laser photolysis. They used layers of closely packed 5 nm titanium dioxide nanocrystallites. Pulsed laser photolysis was carried out in the presence and absence of added reactants. Their results showed that TiO2 layers immersed in liquids (acidic or alkaline water, CCl4 , CCl4 /CBr4 mixture, cyclohexane) have the same absorption versus time profiles as the dry layers. Murakata et al. (1998) studied the effect of the solvent in which the particles are dispersed on their photocatalytic properties using ESR spectra measurements. The dependence of activity upon kind of solvent was found to be due to the different concentrations of Ti3+ formed with the different solvents (these are responsible for the reductive catalytic activity of the TiO2 particles).

455 Conclusions Considering the topics covered in this review, it is evident that the emergence of nanotechnology is bound to have significant implications in the research field of photocatalysis. First, the nanostructure their remarkable properties will prove invaluable in the development of novel photocatalysts and the enhancement of existing ones. In areas such as doping, coupling, capping and sensitizing, benefits have already been seen with enhanced photocatalytic and optical properties. Secondly, the use of nanocrystalline thin films in electrochemically assisted photocatalytic processes is another area where nanoparticles will be very useful. Finally, the use of nanoparticles for providing mechanistic information will continue to make significant contributions in enhancing the understanding of photoinitiated processes in semiconductor materials.

References Ahmed S. & D.F. Ollis, 1984. Solar assisted catalytic decomposition of the chlorinated hydrocarbons trichloroethylene and trichloromethane. Solar Energy 32(5), 597–601. Anpo M., T. Shima, S. Kodama & Y. Kubokawa, 1987. Photocatalytic hydrogenation of CH3 CCH with H2 O on small particle TiO2 : Size quantization effects and reaction intermediates. The Journal of Physical Chemistry 91, 4305–4310. Aruna S.T. & K.C. Patil, 1996. Synthesis and properties of nanosized titania. Journal of Materials Synthesis and Processing 4(3), 175–179. Bahnemann D.W., 1993. Ultrasmall metal oxide particles: Preparation, photophysical characterisation, and photocatalytic properties. Israel Journal of Chemistry 33, 115–136. Bahnemann D.W., A. Henglein, J. Lillie & L. Spanhel, 1984. Flash photolysis observation of the absorption spectra of trapped positive holes and electrons in colloidal TiO2 . The Journal of Physical Chemistry 88, 709–711. Bahnemann D.W., C. Kormann, & M.R. Hoffmann, 1987. Preparation and characterisation of quantum size zinc oxide: A detailed spectroscopic study. The Journal of Physical Chemistry 91, 3789. Bahnemann D.W., D. Bockelemann, R. Goslich, M. Hilgendorff & D. Weichgrebe, 1993. Photocatalytic detoxification: Novel catalysis, mechanisms and solar applications. In: Ollis D.F. and Al-Ekabi H. eds. In: Photocatalytic purification and treatment of water and air. Elsevier Science Publishers. pp. 301–319. Bedja I. & P.V. Kamat, 1995. Capped semiconductor colloids. Synthesis and photoelectrochemical behaviour of TiO2 -capped

SnO2 nanocrystallites. The Journal of Physical Chemistry 99, 9182–9188. Bedja I., S. Hotchandani & P.V. Kamat, 1993. Photoelectrochemistry of quantized WO3 colloids. Electron storage, electrochromic, and photoelectrochromic effects. The Journal of Physical Chemistry 97, 11064–11070. Bedja I., S. Hotchandani, R. Carpentier, K. Vinodgopal & P.V. Kamat, 1994a. Electrochromic, and photoelectrochemical behaviour of thin WO3 films prepared from quantized colloid particles. Thin Solid Films 247(2), 195–200. Bedja I., S. Hotchandani & P.V. Kamat, 1994b. Preparation and photochemical characterisation of thin SnO2 nanocrystalline semiconductor films and their sensitization with bis(2,20 -bipyridine)(2,20 -bibyridine-4,40 -dicarboxylic acid)ruthenim(II) complex. The Journal of Physical Chemistry 98, 4133–4140. Beydoun D., R. Amal, G. Low & S. McEvoy, 1998. Novel photocatalyst: Titania coated magnetite. World Congress on Particle Technology 3, 6–9 July 1998, Brighton, UK, Paper no. 385. Bockelmann D., M. Lindner & D. Bahnemann, 1996. From nanosized particles to commercial products: The search for novel photocatalysts. In: Pelizzetti E. ed. Fine Particles Science and Technology, Kluwer Academic Publishers. pp. 657–689. Braun P.V., P. Osenar & S.-I. Stup, 1996. Semiconducting superlattices templated by molecular assemblies. Nature 380(6572), 325–328. Brus I.E., 1986. Electronic wave functions in semiconductor clusters: Experiment and theory. The Journal of Physical Chemistry 90, 2555–2560. Cherepy N.J., D. Liston, J.A. Lovejoy, H. Deng & J.Z. Zhang, 1998. Ultrafast studies of photoexcited electron dynamics in γ - and α-Fe2 O3 semiconductor nanoparticles. The Journal of Physical Chemistry B 102, 770–776. Choi W., A. Termin & M.R. Hoffmann, 1994. The role of metal ion dopants in quantum-sized TiO2 : Correlation between photoreactivity and charge carrier recombination dynamics. The Journal of Physical Chemistry 98, 13669–13679. Colombo D.P. Jr. & R.M. Bowman, 1996. Does interfacial charge transfer compete with charge carrier recombination? A femtosecond diffuse reflectance investigation of TiO2 nanoparticles. The Journal of Physical Chemistry 100(Nov 21), 18445– 18449. Evans J.E., K.W. Springer & J.H. Zhang, 1994. Femtosecond studies of interparticle electron transfer in a coupled CdS–TiO2 colloidal system. The Journal of Chemical Physics 101, 6222– 6225. Eychmueller A., A. Haesselbarth & H. Weller, 1992. Quantumsized HgS in contact with quantum-sized CdS colloids. Journal of Luminescence 53(1–6), 113–115. Faust B.C., M.R. Hoffmann & D.W. Bahnemann, 1989. Photocatalytic oxidation of sulfur dioxide in aqueous suspensions of α-Fe2 O3 . The Journal of Physical Chemistry 93, 6371–6381. Fischer Ch.-H., H. Weller, L. Katsikas & A. Henglein, 1988. Photochemistry of colloidal semiconductors. 30. HPLC investigation of small CdS particles. Langmuir 5, 429–432. Fischer Ch.-H., J. Lillie, H. Weller, L. Katsikas & A. Henglein, 1989. Photochemistry of colloidal semiconductors. 29. Fractionation of CdS sols of small particles by exclusion chromatography. Ber. Bunsenges. Phys. Chem 93, 61–64.

456 Fu X., X. Zhang, S. Song, X. Wang & M. Tao, 1997. Preparation of nano-size TiO2 /Pt/glass thin films by sol–gel process and their photoelectrocatalytic properties. Gongneng Cailiao Journal of Functional Materials 28(4, Aug 20), 411–414. Fujiwara H., H. Hosokawa, K. Murakoshi, Y. Wada, S. Yanagida, T. Okada & H. Kobayashi, 1997. Effect of structures on photocatalytic CO2 reduction using quantized CdS nanocrystallites. The Journal of Physical Chemistry B 101(Oct 9), 8270–8278. Gallardo S., M. Gutierrez, A. Henglein & E. Janata, 1989. Photochemistry and radiation chemistry of colloidal semiconductors. 34. Properties of Q-PbS. Berichte der Bunsen-GesellschaftPhysical Chemistry 93, 1080–1090. Gratzel M., 1989. Heterogeneous photochemical electron transfer. CRC Press, Boca Raton, FL. Hagfeldt A. & M. Gratzel, 1995. Light-induced redox reactions in nanocrystalline systems. Chemical Reviews 95(Jan/Feb), 49–68. Haque S.A., Y. Tachibana, D.R. Klug & J.R. Durrant, 1998. Charge recombination kinetics in dye-sensitized nanocrystalline titanium dioxide films under externally applied bias. The Journal of Physical Chemistry B 102, 1745–1749. Henglein A., 1987. Q-particles: Size quantization effects in colloidal semiconductors. Progress in Colloid Polymer Science 73, 1–4. Henglein A., 1988. Mechanism of reactions on colloidal microelectrodes and size quantization effects. Topics in Current Chemistry 143: 113–180. Henglein A., 1989. Small particle research: Physicochemical properties of extremely small colloidal metal and semiconductor particles. Chemical Reviews 89: 1861–1873. Henglein A., 1997. Nanoclusters of semiconductors and metals – colloidal nano-particles of semiconductors and metals: Electronic structure and process. Berichte der BunsenGesellschaft-Physical Chemistry 101(11): 1562–1572. Henglein A., M. Gutierrez, H. Weller, A. Fojtik & J. Jirkovsky, 1989. Photochemistry of colloid semiconductors. 30. Reactions and fluorescence of AgI and AgI–Ag2 S colloids. Ber. Bunsenges. Phys. Chem. 93: 593–599. Hinogami R., Y. Nakamura, S. Yae & Y. Nakato, 1997. Modification of semiconductor surface with ultrafine metal particles for efficient photoelectrochemical reduction of carbon dioxide. Applied Surface Science 121/122: 301–304. Hoffmann A.J., H. Yee, G. Mills & M.R. Hoffmann, 1992a. Photoinitiated polymerisation of methyl methacrylate using Q-sized ZnO colloids. The Journal of Physical Chemistry 96: 5540–5546. Hoffmann A.J., G. Mills, H. Yee & M.R. Hoffmann, 1992b. Q-sized CdS: Synthesis, characterisation, and efficiency of photoinitation of polymerisation of several vinylic monomers. The Journal of Physical Chemistry 96: 5546–5552. Hoffmann A.J, E.R. Carraway & M.R. Hoffmann, 1994. Photocatalytic production of H2 O2 and organic peroxides on quantumsized semiconductor colloids. Environmental Science and Technology 28: 776–785. Hoffmann M.R., S.T. Martin, W. Choi & D. Bahnemann, 1995. Environmental applications of semiconductor photocatalysis. Chemical Reviews 95: 69–96.

Hotchandani S. & P.V. Kamat, 1992. Charge transfer in coupled semiconductor systems. Photochemistry and photoelectrochemistry of the colloidal CdS–ZnO system. The Journal Physical Chemistry 96: 6834–6839. Howe R.F., 1998. Recent developments in photocatalysis. Dev. Chem. Eng. Mineral Process. 6(1): 55–84. Howe R.F. & M. Gratzel, 1985. EPR observation of trapped electrons in colloidal TiO2 . The Journal of Physical Chemistry 89: 4495. Kalyanasundaram K., 1983. Semiconductor particulate systems for photocatalysis and photosynthesis: An overview. In: Michael Gratzel ed. Energy Resources Through Photochemistry and Catalysis. Academic Press. pp. 217–260. Kamat P.V., 1995. Tailoring nanostructured thin films. Chemtech 22–28. Kamat P.V. & B. Patrick, 1992. Photophysics and photochemistry of quantized ZnO colloids. The journal of physical chemistry 96: 6829–6834. Kavan L., T. Stoto, M. Gratzel, D. Fitzmaurice & V. Shklover, 1993. Quantum size effects in nanocrystalline semiconducting TiO2 layers prepared by anodic oxidative hydrolysis of TiCl3 . The Journal of Physical Chemistry 97: 9493–9498. Korgel B.-A. & Monbouquette H.-G., 1997. Quantum confinement effects enable photocatalyzed nitrate reduction at neutral pH Using CdS nanocrystals. The Journal of Physical Chemistry B 101: 5010–5017. Kormann C., D.W. Bahnemann & M. Hoffmann, 1988. Preparation and characterisation of quantum-size titanium dioxide. The Journal of Physical Chemistry 92: 5196–5201. Kormann C., D.W. Bahnemann, M.R. Hoffmann, 1989. Environmental photochemistry: Is iron oxide (hematite) an active photocatalyst? A comparative study: α-Fe2 O3 , ZnO, TiO2 . Journal of Photochemistry and Photobiology A: Chemistry 48: 161–169. Kortan A.R., R. Hull & R.L. Opila, 1990. Nucleation and growth of CdSe on ZnS quantum crystallite seeds, and vice versa, in inverse micelle media. Journal of the American Chemical Society 112: 1327–1332. Lepore G.P., C.H. Langford, J. Vichova & A. Vlcek, 1993. Photochemistry and picosecond absorption spectra of aqueous suspensions of a polycrystalline titanium dioxide optically transparent in the visible spectrum. Journal of Photochemistry and Photobiology A: Chem 75: 67–75. Levy B., 1997. Photochemistry of nanostructured materials for energy applications. Journal of Electroceramics 1(3): 239–272. Lindner M., D. Bahnemann, B. Hirthe & W. Griebler, 1995. Solar water detoxification: Novel TiO2 powders as highly active photocatalysts. Solar Engineering-Vol.1, ASME: 399–408. Liu D. & P.V. Kamat, 1993. Photoelectrochemical behaviour of thin CdSe and coupled TiO2 /CdSe semiconductor films. The Journal of Physical Chemistry 97: 10769–10773. Martin S., H. Herrmann, W. Choi & M. Hoffmann, 1995. Photochemical destruction of chemical contaminants on quantum-sized semiconductor particles, Solar EngineeringVol.1, ASME: 409–413. Matthews R., 1993. Photocatalysis in water purification: Possibilities, problems and prospects. In: Ollis D.F. and Al-Ekabi H. eds. Photocatalytic Purification and Treatment of Water and Air. Elsevier Science Publishers. pp. 121–139.

457 Mews A., A. Eychmuller, M. Giersig, D. Schooss & H. Weller, 1994. Preparation, characterization, and photophysics of the quantum dot quantum well system CdS/HgS/CdS. The Journal of Physical Chemistry 98: 934–941. Micic O.I., M. Meglic, D. Lawless, D.K. Sharma & N. Serpone, 1990. Semiconductor photophysics. 5. Charge carrier trapping in ultrasmall silver iodide particles and kinetics of formation of silver atom clusters. Langmuir 6: 487–492. Murakata T., R. Yamamoto, Y. Yoshida, M. Hinohara, T. Ogata & S. Sato, 1998. Preparation of ultrafine TiO2 particles dispersible in organic solvents and their photocatalytic properties. Journal of Chemical Engineering of Japan 31(1): 21–28. Nedeljkovic J.M., M.T. Nenodovic, O.I. Micic & A.J. Nozik, 1986. Enhanced photoredox chemistry in quantized semiconductor colloids. The Journal of Physical Chemistry 90: 12–13. Nosaka Y., N. Ohta & H. Miyama, 1990. Photochemical kinetics of ultrasmall semiconductor particles in solution: Effect of size on the quantum yield of electron transfer. The Journal of Physical Chemistry 94: 3752–3755. Ogura S., M. Kohno, K. Sato & Y. Inoue, 1997. Photocatalytic activity for water decomposition of RuO2 -combined M2 Ti6 O13 (M equals Na, K, Rb, Cs). Applied Surface Science 121– 122(Nov 2): 521–524. O’Regan B. & M. Gratzel, 1991. A low cost, high efficiency solar cell based on dye-sensitized colloidal TiO2 films. Nature 353(Oct 24): 737–740. Penpolcharoen M., A. Amal & V. Chen, 1998. Synthesis of titania coated hematite particles. Proceedings, CHEMECA 98, The 26th Australian Chemical Engineering Conference, 28–30 September 1998, Port Douglas, Queensland, Australia, Paper no. 222. Rabani J., 1989. Sandwich colloids of ZnO and ZnS in aqueous solutions. The Journal of Physical Chemistry 93: 7707–7713. Rabani J., K. Yamashita, K. Ushida, J. Stark & A. Kira, 1998. Fundamental reactions in illuminated titanium dioxide nanocrystallite layers studied by pulsed laser. The Journal of Physical Chemistry 102: 1689–1695. Reber J.F. & M. Rusek, 1986. Photochemical hydrogen production with platinized suspensions of cadmium sulfide and cadmium zinc sulfide modified by silver sulfide. The Journal of Physical Chemistry 90: 824–834. Rothenberger G, J. Moser, M. Gratzel, N. Serpone & D.K. Sharma, 1985. Charge carrier trapping and recombination dynamics in small semiconductor particles. Journal of the American Chemical Society 107: 8054–8059. Rothenberger G., D. Fitzmaurice & M. Gratzel, 1992. Spectroscopy of conduction band electrons in transparent metal oxide semiconductor films: Optical determination of the flatband potential and colloidal titanium dioxide films. The Journal of Physical Chemistry 96: 5983–5986. Sahyun M.R. & N. Serpone, 1997. Primary events in the photocatalytic deposition of silver on nanoparticulate TiO2 . Langmuir 13(Sep 17): 5082–5088. Salafsky J.S., W.H. Lubberhuizen, E.I van Faassen & R.E. Schropp, 1998. Charge dynamics following dye photoinjection into a TiO2 network. The Journal of Physical Chemistry B 102: 766–769. Sato T., K.I. Sato & Y. Fujishiro, 1996a. Photochemical reduction of nitrate to ammonia using layered hydrous titanate/cadmium

sulphide nanocomposites. Journal of Chemical Technology and Biotechnology 67: 345–349. Sato T., K. Masaki & K.-I. Sato, 1996b. Photocatalytic properties of layered hydrous titanium oxide/CdS–ZnS nanocomposites incorporating CdS–ZnS into the interlayer. Journal of Chemical Technology and Biotechnology 67: 339–344. Schiavello M. & A. Sclafani, 1989. Thermodynamic and kinetic aspects in photocatalysis. In: Serpone N. and Pelizzetti E. eds. Photocatalysis: Fundamentals and Applications. John Wiley & Sons. pp. 159–173. Schlichthorl G., S.Y. Huang, J. Sprague & A.J. Frank, 1997. Band edge movement and recombination kinetics in dye-sensitized nanocrystalline TiO2 solar cells: A study by intensity modulated photovoltage spectroscopy. The Journal of Physical Chemistry B 101: 8141–8155. Schwitzgebel J., J.G. Ekerdt, H. Gerischer & A. Heller, 1995. Role of the oxygen molecule and of the photogenerated electron in TiO2 photocatalyzed air oxidation reactions. The Journal of Physical Chemistry 99(15): 5633–5638. Serpone N., D. Lawless & E. Pelizzetti, 1996. Subnanosecond characteristics and photophysics of nanosized TiO2 particulates from Rpart = 10 A to 34 A: Meaning for heterogeneous Photocatalysis. In: Pelizzetti E. ed. Fine Particles Science and Technology. Kluwer Academic Publishers. pp. 657–673. Shiragami T., S. Fukami & Y. Wada, 1993. Semiconductor photocatalysis: Effect of light intensity on nanoscale CdS-catalyzed photolysis of organic substrates. The Journal of Physical Chemistry 97(Dec 9): 12882–12887. Skinner D.E., D.P. Colombo & J.J. Cavaleri, 1995. Femtosecond investigation of electron trapping in semiconductor nanoclusters. The Journal of Physical Chemistry 99: 7853–7856. Smith B.A., D.M. Waters, A.E. Faulhaber, M.A. Kreger, T.W. Roberti & J.Z. Zhang, 1997. Preparation and ultrafast optical characterization of metal and semiconductor colloidal nano-particles. Journal of Sol–Gel Science and Technology 9(2, Mar): 125–137. Spanhel L. & M.A. Anderson, 1991. Semiconductor clusters in the Sol–Gel process: Quantized aggregation, gelation and crystal growth in concentrated ZnO colloids. Journal of the American Chemical Society, 113: 2826–2833. Spanhel L., H. Weller, A. Fojtik & A. Henglein, 1987a. Photochemistry of semiconductor colloids. 17. Strong luminescing CdS and CdS–Ag2 S particles. Berichte der BunsenGesellschaft- Physical Chemistry 91: 88–94. Spanhel L., H. Weller & A. Henglein, 1987b. Photochemistry of semiconductor colloids. Electron injection from illuminated CdS into attached TiO2 and ZnO particles. Journal of the American Chemical Society 113: 2826–2833. Tenne R., 1996. Fullerene-like structures and nanotubes from inorganic compounds. Endeavour (Oxford, England) 20(3): 97–104. Ueno A., N. Kakuta, K.H. Park, M.F. Finalayson, A.J. Bard, A. Campion, M.A. Fox, S.E. Webber & J.M. White, 1985. Silica-supported ZnS–CdS mixed semiconductor catalysts for photogeneration of hydrogen. The Journal of Physical Chemistry 89: 3828–3833. Vinodgopal K. & P.V. Kamat, 1995. Electrochemically assisted photocatalysis using nanocrystalline semiconductor thin films. Solar Energy Materials and Solar Cells 38(1–4, Aug): 401–410.

458 Vinodgopal K & P.V. Kamat, 1996. Combine electrochemistry with photocatalysis. Chemtech 26(Apr): 18–22. Vogel, R., P. Hoyer & H. Weller, 1994. Quantum-sized PbS, Ag2 S, Sb2 S3 , and Bi2 S3 particles as sensitizers for various nanoporous wide-bandgap semiconductors. The Journal of Physical Chemistry 98: 3183–3188. Wang Y. & N. Herron, 1991. Nanometer-sized semiconductor clusters: Materials synthesis, quantum size effects, and photophysical properties. The Journal of Physical Chemistry 95: 525–532. Wang C.-C., Z. Zhang & J.Y. Ying, 1997. Photocatalytic decomposition of halogenated organics over nanocrystalline titania. Nanostructured Materials 9: 583–586. Weller H. & A. Eychmuller, 1995. Photochemistry and photoelectrochemistry of quantized matter: Properties of semiconductor nanoparticles in solution and thin-film electrodes. In: Douglas C. Neckers, David H. Volman and Gunther von Bunau, eds. Advances in Photochemistry. Volume 20. John Wiley and Sons. Inc. Wilcoxon, J.P., P.P. Newcomer & G.A. Samara, 1997. Synthesis and optical properties of MoS2 and isomorphous nanoclusters in the quantum confinement regime. Journal of Applied Physics 81(12): 7934–7944.

Youn H.C., S. Baral & J.H. Fendler, 1988. Dihexadecyl phosphate, vesicle-stabilized and in situ generated mixed CdS and ZnS semiconductor particles. Preparation and utilization for photosensitized charge separation and hydrogen generation. The Journal of Physical Chemistry 92: 6320–6327. Zeltner W.A. & M.A. Anderson, 1996. The use of nanoparticles in environmental applications. In: Pelizzetti E. ed. Fine Particles Science and Technology. Kluwer Academic Publishers. pp. 643–656. Zhang Y., J.C. Crittenden, D.W. Hand & D.L. Perram, 1994a. Fixed-bed photocatalysis for solar decontamination of water. Environmental Science Technology 28: 435–442. Zhang Y., J.C. Crittenden & D.W. Hand, 1994b. The solar photocatalytic decontamination of water. Chemistry and Industry 714–717. Zhang, Z., C.-C. Wang, R. Zakaria & J.Y. Ying, 1998. Role of particle size in nanocrystalline TiO2 -based photocatalysts. The Journal of Physical Chemistry B 102: 10871–10878. Zhou H.S., I. Honma, H. Komiyama & J.W. Haus, 1993. Coated semiconductor nanoparticles: The CdS/PbS system’s synthesis properties. The Journal of Physical Chemistry 97: 895.