Dev. Chem.Eng. Mineral Process., 6(1R),pp.55-84, 1998. Recent Developments in Photocatalysis R.F. Howe Dept. of Physical Chemistry,University of New South Wales, Sydney, New South Wales 2052, AUSTRALIA This article reviews some recent developments in the area of photocatalysis. These include the development of metho& for preparing nanoctystallme (Q-sized) Ti02, the doping and promotion of titania to improve its photoreactivity, recent studies gf reaction mechanisms, the introduction of other oxide photocatalysts, and recent advances in the design of reactorsfor photocatalysis. It is concluded that practical implementation of photocatalysis, particularly in environmental applications, is now close to realitv. Introduction The subject of photocatalysis has been addressed in the literature for more than 25 years. Strictly speaking, heterogeneous photocatalysis involves absorbtion of light by a solid photocatalyst which then modifies the state of activation of at least one reactant adsorbed on the catalyst surface, thereby producing a new reaction path with a lower activation barrier than that for the thermal reaction proceeding in the absence of light [l]. This process does not involve direct absorption of light by reactant molecules, but resembles the homogeneous process of photo-sensitization in which a reactant molecule is electronically excited through interaction with a sensitizer which strongly absorbs the incident light. A distinction can also be made between the photocatalytic reaction, where the reaction in question does not proceed at all in the absence of light, and a photo-assisted catalytic reaction in which 55 R.F. Howe exposure to light enhances the rate of catalytic reaction which does proceed albeit slowly, in the dark. A further development is the closely related topic of photoelectrochemistry, where the combination of light and an electric field is used to drive a desired process. In all of the above, the central aspect is the use of light energy to promote or enhance an effective chemical reaction in the presence of a solid surface. This paper will outline some of the more recent developments in this general field. Other more comprehensive reviews, particularly of Ti02 photocatalysis, are available [2-91. Photocatalysis with Ti@: Some General Aspects Ti02 has received by far the most attention as a semiconducting oxide photocatalyst. The 1972 discovery of photocatalytic splitting of water over Ti02 electrodes by Fujishima and Honda [lo] prompted a vast research effort investigating the photoreactivity of titania. Titania is readily available, robust, and has a band-gap in the near ultraviolet region of the spectrum (3.2 eV or 387 nm for anatase). The anatase form of Ti02 is regarded as more photoactive than the mile form [l I], although that does depend on the precise preparation conditions of the titania and the particular reaction involved . The mechanism of photocatalysis over Ti02 is understood in broad outline, although as discussed further below mechanistic details of prospects for optimizing photocatalytic performance are still being hotly pursued. The initiating act in photocatalysis over a semiconductor oxide such as Ti02 is absorbtion of a photon with energy equal to or greater than the band gap of the semiconductor, producing holes and electrons. Figure 1, from reference , shows this process schematically. Once the holes and electrons are produced (corresponding to an excited state in the photocatalyst), there are several de-excitation events which may occur. The holes and electrons may recombine, either in the interior of the semiconductor or after migration to the surface. Various hole or electron trap sites (e.g. bulk impurity or dopant ions, surface defect states, etc.) may mediate the recombination process (not shown in Figure 1). Migration of holes andor electrons to the surface followed by interfacial electron transfer from/to adsorbed donor or acceptor 56 molecules respectively are the events leading to reaction. Recent developments in photocatalysis Figure 1. Schematic of phot-citation and de-excitation processes in an oxide semiconductor (Reproduced with permission@om reference ). The overall efficiency (quantum yield) of a photocatalytic process is then determined by the competition between interfacial electron transfer and ho1e:electron recombination. Interfacial electron transfer is in turn controlled by the band energy positions of the semiconductor relative to the redox potentials of the adsorbates: the acceptor species must lie below the bottom of the conduction band and the donor species above the top of the valence band. There is now a vast body of literature describing the use of Ti02 as a photocatalyst, particularly in recent times for environmental applications [ 6 ] . Table 1 provides a summary of different reaction types investigated. A much more comprehensive list is provided by Hofiinann et al. . The photodecompositionof water first reported by Fujushima and Honda [ 101 employed a Ti02 electrode in a photoelectrochemicalceli; separation of holes and 57 R.F. Howe Table 1. Reactions Photocatalyzed over Ti02. Reaction type 1. Decomposition [Go2+ 'So2+ 2 '60180 H20 + H2+f02 2. Photooxidation of Inorganic Species co+fo2+ CO2 N2+ + 022NO +NZ+ 3H20 CN-taq3 +: 0 2 + OCN-Iaq] 2NH3 + 0 2 2HX[aq] + 02 + X2 + H2O 3. Photo-reduction of Inorganic Species + 2NH3 + 3 O2 C02 + 2H20 + CH30H + O2 N2 + 3H20 ~~ 4. Photooxidation of Organic Species in the Gas Phase + acetone trichloroethylene+ 0 2 + C02, HCI,CO, Clz 4chlorophenol+ 0 2 + oxidation products all<ane+02 5. Photooxidation of Organic Contaminants in the Aqueous Phase chlorinated aromatics chlorinated aliphatics chlorinated olefins nitro compounds CFCs and HCFCs 6. Photodestruction of Malignant Cells Generation of H202 and OH radicals in vitro and in vivo 58 Reference Recent developments in photocatalysis electrons was achieved through an external circuit allowing electrons to flow to a platinum electrode where they reduce water to hydrogen, while the holes at the Ti02 surface oxidize water to oxygen. Water decomposition does not occur on irradiation of Ti02 alone; although the reaction is energetically favourable it is inhibited by the large overpotentials for evolution of H2 and 0 2 . Water decomposition can be achieved over Ti02 containing added metal and metal oxide components. Figure 2 shows for example a schematic of a Ti02 particle containing added Pt and RuO2. This functions as a short-circuited micro photoelectrochemicalcell in which the Pt is the cathode and Ru02 the anode; the role of the added components is to effectively lower the overpotential for evolution of H2 at the Pt surface and 0 2 at the RuO2 surface respectively . An alternative approach has been to provide a sacrificial species to remove either holes or electrons, allowing the other to react with water. For example, sustained production of H2 has been observed when aqueous suspensions of Ti02 are irradiated in the presence of methanol . Methanol scavenges positive holes from the structure of the Ti02 (being in the process oxidized to COz), allowing electrons to reduce water to H2. >O'"0 2 Figure 2. Schematic of water dissociation over a Ti02 photocatalyst contairiing added Pt and Ru02 (after reference [I 61). 59 R.F.Howe Photooxidation of inorganic species over Ti02 has been undertaken both in the gas phase and in aqueous solution. For example, Bickley et al.  detected NO by thermal desorption after irradiation of an N2, 0 2 mixture in the presence of Ti02 at room temperature, while Fitzmaurice et al.  observed 12- and I2 formation on irradiation of colloidal sols of Ti02 in the presence of iodide ions. Photoreduction of nitrogen over an iron-doped Ti02 catalyst was reported by Schrauzer et al.  to yield ammonia and hydrazine. Photoreduction of C02 has attracted some attention. In this case optimum conversion of C02 to hydrocarbon products is achieved by adding a metal component to the T i 0 3 for example Pd-Ti02 shows very high selectivity for C b formation . Presumably, the Pd functions in the same way as the metal component in water decomposition photocatalysis, as a cathode. A thin-film photoelectrocatalytic cell for reduction of carbon dioxide has recently been described by Ichikawa and Doi . This combines a Ti02 film on a conducting substrate functioning as the anode with a metal (e.g. Pt or Cu) cathode, separated by a Nafion polymer layer. Water is oxidized to 0 2 at the photoanode, while electrons flowing through the external circuit to the metal cathode reduce carbon dioxide. Addition of a small bias voltage enhanced the C02 conversion, and in a very interesting observation operating the bias voltage in a pulsed mode was found to inhibit catalyst deactivation. Photooxidation of hydrocarbons in the gas phase over Ti02 catalysts was the subject of intensive study in the 1970s by the Teichner group [l]. Products obtained were primarily ketones and aldehydres; it was pointed out that the low temperatures at which photocatalysis is conducted avoided the gas phase free radical reactions leading to complete combustion in the thermal catalytic oxidation. More recently, the possibility of using photocatalysis to destroy organic compounds present as environmental impurities in air has been realised. Miller and Fox  have indicated that from an economic viewpoint photocatalytic treatment of contaminated air is commercially viable for high quantum efficiency reactants such as trichloroethylene (TCE). Anderson et al.  have described, for example, complete conversion of TCE to C02, HCl and H20 in 60 Recent developments in photocatalysis a single pass through a reactor containing Ti02 photocatalyst at TCE levels of 450 ppm. Other hydrocarbons, particularly aromatics, are much less reactive, and photo-oxidation compares unfavourably with incineration or carbon absorption separation as means of removal. 011;s et al.  have recently reported that photocatalyzed destruction rates of low quantum efficiency contaminant compounds in air can be dramatically promoted by adding a high quantum efficiency promoter such as TCE, perchloroethylene or trichloropropene. Addition of TCE was found, for example, to produce high quantum efficiency conversion at contaminant levels of 50 mg m-3 for several aromatics and volatile oxygenates. The authors propose that adsorbed chlorine radicals play a key role in the enhancements observed, although they note that not all hydrocarbon contaminants can be influenced in this way, and the reaction mechanisms remain uncertaiti. The comprehensive review by Hoffmann et al. on environmental applications of semiconductor photocatalysis lists more than 400 references to published work on photooxidation of organic compounds in water over titania photocatalysts. The catalysts used include solid powders, colloidal sols, and thin films. The potential of photocatalysis for wastewater treatment is clearly established, but to date no commercially available process uses W irradiation of Ti02 photocatalysts for remediation of contaminated waters. However, a commercial unit for analysis of total organic carbon in water based on photochemical conversion of organics to C02 over Ti02 has been developed from the work of Mathews et al. . In reviewing the existing literature on both gas-phase and liquid-phase photocatalysis over Ti02, there are several reasons which can be identified as to why photocatalytic processes have not so far become commercially viable, and why the ultimate objective of utilizing sunlight to drive chemical reactions has not been attained. First, intrinsic activity of many photocatalysts reported in the literature is low, and variable. A commercial anatase (Degussa P25, which is actually a rutileanatase mixture) is frequently used as a standard catalyst, but has a relatively low surface area (approx. 55 m2g-l). Tanaka et al. have reported wide variations in 61 R.F. Howe photocatalytic activity over different crystalline forms of Ti02 having different surface areas and calcined at different temperatures. More recently, sol-gel methods are being used to produce higher surface area forms of titania (with in principle at least higher reactivity); the properties of nanocrystalline Ti02 are discussed further below. Enhancement of activity by doping the Ti02 photocatalyst with transition metal ions has been widely attempted, but with variable success. Another motivation for modifying the Ti02 with dopants is to shift the absorbtion spectrum from the near ultra-violet into the visible, in order to match more closely the solar spectrum. These issues are considered below. Although the general features of photocatalytic reaction mechanisms are understood, the detailed events following creation of the hole and the electron and in particular the interfacial electron and hole transfers to adsorbed species are not well characterized. As in other types of catalysis, an improved understanding of reaction mechanisms is essential if catalyst performance is to be improved. The complexity of the solid-liquid interface, particularly where the solid surface is not well defined, contributes to the mechanistic uncertainty. Recent approaches to this problem include the application of surface science techniques to well defined single crystal Ti02 surfaces , and the use of time-resolved spectroscopic methods to probe the kinetics of fundamental events . These and other developments in mechanistic studies are discussed further below. Another crucially important aspect of photocatalysis is reactor design. Early studies of photocatalysis were not particularly concerned with reactor efficiency, and used typically a fixed bed of Ti02 powder or an aqueous slurry. The absorbtion coefficient for UV radiation of Ti02 is however extremely high, such that 99% of light absorbtion occurs within a 5 pm depth of Ti02 powder . The typical particle sizes of 30 to 100 nm for commercial anatase powder mean that light scattering from fixed beds or powder slurries will act to reduce photochemical efficiency. Some of the recent activities in reactor design intended to overcome these problems are described below. 62 Recent developments in photocatalysis Nanocrystalline Ti02 The surface area of a catalyst can be dramatically increased by reducing the particle size into the nanometre range (e.g. 10 nm = 100 A). The photoreactivity of colloidal dispersions of Ti02 in this size range was first investigated by Gratzel et al. . The advantages of such colloidal dispersions were pointed out to be the optical transparency (minimal light scattering), rapid carrier diffusion to the interface, and high surface to volume ratio. Since that time the chemistry involved in producing colloidal sols has been investigated by many groups  and methods for producing large quantities of material developed . The reduction of particle size below 10 nm has other important consequences for photocatalytic reactivity. The spatial confinement of charge carriers within a nanometre size range produces so-called size quantization effects. The band gap of the semiconductor increases and the band edges shift to yield larger redox potentials, thus increasing rate constants for charge transfer at the surface. For systems in which interfacial charge transfer is the rate limiting step, size-quantized (Q-sized) Ti02 particles may show enhanced photoefficiency . Three methods have been used to study the photoreactivity of Q-sized Ti02 particles. Gratzel et al.  initially studied aqueous colloidal dispersions as prepared. Such dispersions, although optically transparent and convenient for spectroscopic investigation [47-49] are metastable, and do not allow convenient recovery of catalyst. Q-sized Ti02 powder can be recovered from colloidal sols by, for example, freeze drying . Alternatively, films of Q-sized particles can be formed by dip-coating or spin-coating suitable substrates with the sol solutions. Figure 3 shows for example a scanning electron micrograph of a Q-sized Ti02 film prepared by dip-coating a quartz slide with a colloidal Ti02 sol . The particle sizes evident in the micrograph are remarkably uniform, in the range 5-10 nm, and are unchanged when the film is fired at temperatures up to 800°C. Grazing incidence X-ray diffraction and EXAFS measurements have revealed however that the average crystalline domain size increases progressively on firing. The films as prepared comprise poorly crystalline anatase, as judged by X-ray diffraction, EXAFS and Raman spectroscopy. The crystallinity improves on firing but even 63 R.F. Howe after firing at 800°C the films remain predominantly anatase (Q-size anatase bulk powders convert to rutile at about 500°C). The dramatic effect of crystalline domain size on the band gap of Q-size anatase films is illustrated in Figure 4 . This shows the uv-visible transmission spectra of films after firing to various temperatures, compared with the diffuse reflectance spectrum of bulk anatase. The increased band gap associated with Q-size particles is clearly evident; the difference between the film and bulk anatase decreases as the firing temperature increases, but the band gap is still blue shifted even after firing at 800°C. Figure 3. SEM image of Q-size Ti02 film prepared by dip coatirig a quartz substrate with colloidal sol Cfrom reference  ). The films prepared by dip-coating illustrated in Figures 3 and 4 are typically 50 nm thick. Thicker films can be prepared by multiple coating, although the properties of such thicker films have not yet been investigated in detail. The films are extremely porous and may be expected to show high photoreactivity. Anderson 64 Recent developments in photocatalysis et al.  have reported the activity of similar films (prepared by spin-coating) for photodegradation of chlorophenols, and show that the optimum performance is obtained with films coated three times in succession. They point out however that the optimum film thickness will be dependent on reactor design. Likewise, Cui et al.  report that the photocatalytic activity of nanocrystalline Ti02 films for degradation of salicyclic acid does not increase with increasing film thickness beyond about 500 nm. Their data show in fact that there is only a small increase in activity beyond a film thickness of 200 nm. 200 400 300 Nanometers Figure 4. UV-visible spectra of I96 Fe doped Q-size Ti02 film after firing to (a) 125OC. (b)350°C,(c) SOOOC, ( d ) Bulk anatase. 65 R.F. Howe Kato et al.  have examined the performance of nanocrystalline Ti02 films prepared by dip-coating from colloidal sols for the photodegradation of acetic acid, as a function of film pretreatment. They find that films prepared to be nanocrystalline anatase show good activity for acetic acid photodegradation, whereas films pretrated (by rapid heating cycles) to contain a mixture of anatase and rutile are less active, and show an approximately linear correlation between activity and the fraction of anatase in the films. Much thicker nanocrystalline Ti02 films have been used extensively by Gratzel et al.  in photoelectrochemical and photovoltaic cells. Films of 3-10 Frn thickness are prepared by spreading a paste of nanocrystalline colloidal Ti02 particles on to a conducting glass support then sintering; the properties of the resulting films depend very much on the prior history of the colloidal sol (particularly particle size and particle size distribution) as well as the sintering conditions used. [7,55,56] The dynamics of light induced charge separation in thick nanocrystalline titania films have been studied by ORegan et a]. , following injection of electrons from the photo-excited state of a ruthenium complex into the conduction band of the titania. Two important findings from this work were that the colloidal particles constituting the film are in electronic contact, forming a three dimensional array, and that the films contain a high concentration of electron trap sites which exert a strong influence on hole : electron recombination processes. The nature of these electron trap sites was not determined, although earlier EPR studies of colloidal Ti02 sols  indicated that interstitial Ti4+ sites at grain boundaries may be important. Characterization of bulk nanocrystalline Ti02 powders prepared by sol-gel methods has been described by Terwillinger and Chiang . The initially amorphous freeze-dried sols crystallize first to anatase, but nucleation to rutile occurs soon after. (300-4Oo0C)The addition of tin (1 mole %) was found to promote the crystallization of rutile, such that nanocrystalline mile with particle sizes less than 20 nm could be prepared. This approach offers the possibility of preparing well defined nanocrystalline rutile for comparitive studies with the 66 Recent developments in photocatalysis corresponding anatase material. It is also clear that the issue of phase transitions in nanocrystalline Ti02 is an important one; the role of a film substrate in stabilizing the anatase phase and inhibiting particle growth is one which has important consequences for photocatalysis. Another new approach to producing highly dispersed titania photocatalysts is the anchoring of Ti02 onto high surface area supports such as Vycor glass. Anpo et al.  describe the reaction of T i c 4 with hydroxyl groups on porous Vycor glass, followed by hydrolysis of the anchored compounds, to produce supported Ti02. EXAFS and photoluminescence measurements indicate that, at least at low titanium loadings, isolated Ti02 species are produced. The resulting photocatalysts showed high activity for oxidative decomposition of l-octanol (up to 1000 times higher activity per unit mass of Ti02 powders). The conventional description of photocatalysis in terms of semiconductor band structure is clearly inappropriate for such catalysts, and Anpo et al.  prefer to describe the activity in terms of a charge transfer complex : Ti"02-+ hv 4 [Ti3+@]* ...(1) Doped and Promoted Titania There have been many reports of influencing the activity of titania photocatalysts by incorporating transition metal dopants into the titania, but also widespread disagreement as to sign and magnitude of the effects. For example, Mu et al.  reported that doping with trivalent andor pentavalent metal ions was detrimental to the photocatalytic activity, whereas others  showed that activity was enhanced by doping with pentavalent ions. Fe3+ enhances the photoreduction of N2 , but has little influence on the photodegradation of phenol . Photoactivity of Ti02 doped with Mo and V is reported to be lower than that of the parent oxide 1631. There is disagreement however as to whether Cr3+ enhances or inhibits the photoreactivity of titania [60,64]. 67 R.F. Howe Doping with transition metal ions is also known to influence the dynamics of ho1e:electron recombination and interfacial charge transfer in Ti02. Flash photolysis experiments on aqueous colloidal sols showed that doping Ti02 with Fe3+ or V4+ dramatically augmented the lifetime of the hole electron pairs created by band gap irradiation . It was further shown by EPR spectroscopy that irradiation of such doped colloids causes the EPR signal of Fe3+ to fade and new Ti3+ signals to appear. However in the case of V4+, a new V4+ signal first appears and then fades as a Ti3+ signal develops . These observations were attributed to the ability of Fe3+ and V& to function as both electron and hole traps; hole trapping at the transition metal site allowed the accumulation of trapped electrons at Ti4+ sites, accompanied by partial dissolution of the colloids. More recently, Hoffman et al.  have undertaken a systematic study of nanocrystalline Ti02 doped with a range of different metal ions, using transient spectroscopy to measure charge carrier recombination dynamics and correlating these with photoreactivity for oxidation of CHC13 and reduction of CCl4. Figure 5 from reference  shows a periodic table summarizing the effects of different dopant ions on the photoreactivity of Ti02. Relative to Ti4+ (data for the undoped TiO2), the largest enhancement effects are seen with Fe3+, Mo5+, Ru3+, V4+ and V3+, while A13+ and appear to cause some inhibition of the photooxidation activity at least. The quantum yields in Figure 5 could be quantitatively correlated with the transient recombination dynamics. According to Hoffman et al. , dopant ions can function as both hole and electron traps, and as mediators of interfacial charge transfer. To be photo-active, a dopant should act as both an electron trap and a hole trap. Thus ions such as Fe3+, Mo5+, Ru3+, V4+ and V3+ are most effective, being capable of both oxidation (hole trapping) or reduction (electron trapping). Ions such as V5+ which can function only as electron traps do not appreciably inhibit hole-electron recombination, since the trapped electrons can readily combine with the still mobile holes. However, at high light intensity levels when the trap sites are saturated, they may then begin to function as recombination centres. 68 Recent developments in photocatalysis Dopant ions also mediate interfacial charge transfer. This was shown in reference  by the variation in photoactivity enhancement by different metal ions as a function of particle size of the Ti02. The largest enhancements were found in nano-sized particles, in which all dopant ions are located within 1-2 nm of the surface. As the particle size increased, the number of near-surface dopant ions decreased, and the enhancement effects lessened (undoped Ti02 showed no variation of activity with particle size over the same range). Hoffmann et al. suggested that enhancement of interfacial charge transfer is in fact the single most important factor in enhancement of photoreactivity of doped Ti02, at least for nanocrystalline materials. (0.08 0.141 4.08 I: I 1 051 0.66 Ni2+ 050 0.09 4.08 027 1.20 1.60 4.08 0.80 0.84 (0.08 Figure 5. Periodic table of the photocatalytic effects of various metal ion dopants in Ti02 (reproduced with permission from reference  ). The upper numbers are quantum yields for CHClj oxidation, and the lower for CCl4 reduction. 69 R.F. Howe Undoubtedly one reason for variations in dopant effects reported in the literature is differences in the location and coordination of the dopant ions, which depend critically on the methods of sample preparation and pretreatment. In preparing nanocrystalline Ti02 for example by sol-gel methods, dopant ions may be introduced during the initial hydrolysis step, during peptization of the gel produced by hydrolysis, or following dispersal of the sol. Dopant ions initially adsorbed on the surface of sol particles may be incorporated into the particles on firing, or may form separate metal oxide phases. Dopant ions incorporated into the interior of the Ti02 may occupy either lattice (substitutional)or interstitial sites. The ability of the dopant ions to function as trap sites and/or to mediate interfacial charge transfer will depend on all of these factors. Characterizationof the doped catalysts is thus crucial if their performance is to be understood. A detailed characterization of vanadium doped nanocrystalline Ti02 has been reported by Hoffmann et al. . They show evidence for the presence of surface bound V02+, interstitial V4+, surface V2O5, solid solutions of V, Ti1-~02and lattice substituted V4+, depending on the exact conditions of sample preparation and treatment. In all cases, photoreactivity of the doped catalysts was reported to be lowered, implying that vanadium centres both at the surface and in the bulk act as ho1e:electron recombination centres. The complexity of these materials (and the resulting difficulty in interpreting photoreactivity data) has been confirmed in a later EPR and 5IV N M R study by Luca et al. . In contrast, recent studies of Fe3+ doped nanocrystalline Ti02 films prepared by dip coating from a colloid sol where Fe3+ was added during the peptization step have shown that Fe3+ (at levels of 1% or less) is incorporated completely into substitutional sites in the anatase lattice upon firing at elevated temperatures . Figure 6 shows Fourier transforms of Fe K-edge E M S data for 1% Fe3+ doped titania films after firing at different temperatures. The film dried at 150°C shows Fe3+ with a single oxygen coordination shell, corresponding to isolated Fe3+ adsorbed on the surface of the Ti02 particles. After firing at 800°C. on the other hand, the coordination environment is identical to that of Ti4+ in anatase: a distorted octahedral oxygen first shell, and a titanium second shell, proving that upon firing 70 Recent developments in photocatalysis Fe3+ migrates into substitutional lattice sites in the Ti02 (anatase) structure. Preliminary EXAFS analysis indicates that this does not happen in the corresponding films doped with Cr3+ or MoG. -'I 'T LF r........ .... . . j 5 . ... 4 t ......-<. . 1 Figure 6. Fourier transfonns of Fe K-edge EXAFS front 1% F2+ doped titania f i l m fired at (A) 150°C; (B) 800°C. Solid curves are experimental data and dotted curves simulations. The addition of a second transition metal (or non-metal) oxide to Ti02 at high concentrations must be described as promotion rather than doping. There have been 71 R.F. Howe a number of studies of such promoted Ti02 photocatalysts. For example, Do et al.  report that addition of WO3 to Ti02 greatly enhanced its photocatalytic activity for degradation of 1,4 dichlorobenzene. Their data show a factor of 2 increase in rate of degradation at an optimum W03 content of 3 mole %. The enhancement is attributed to enhanced electron transfer from Ti02 to the promoter. The flat-band potential of WO3 is more positive than that of TiO-2, thus photoelectrons will transfer to the W03 conduction band, and holes will accumulate in the Ti02 (and thus be available for oxidation of dichlorobenzene). A similar enhancement has been described on addition of Nb2O5 to Ti02 photocatalysts. Cui et al.  point out that the 2-3 mole % promoter found to be optimum corresponds to the theoretical monolayer capacity of P25 anatase, although no direct evidence for monolayer dispersion of the promoter has been presented. The increase in photocatalytic activity could be correlated with an increase in surface acidity of the catalysts as the Nb2O5 content increased, suggesting that surface acidity may play an important role in the photocatalytic reactivity. Surface acidity was also considered by Anderson et al.  as a possible explanation for the enhanced performance of TiOz/Si02 and Ti02/ZrO2 catalysts for the photooxidation of ethylene in the gas phase. They concluded however that the up to 3-fold enhancement observed must be due to the increased surface area of the promoted catalysts and the inhibition of the anatase to rutile phase transition. Weller et al.  have considered the sensitization of Ti02 (and other wide band-gap oxide semiconductors) photocatalysts by addition of quantum-sized narrow band-gap semiconductors. This concept follows from the use of organic dyes to sensitize Ti02 electrodes . The sensitizer has a high cross-section for absorption of visible light, and injects electrons into the conduction band of the underlying semiconductor. The advantages of using quantum sized semiconductors as sensitizers are that the band gap can be adjusted by varying the particle size and the materials are in principle more stable than organic dye sensitized Ti02. In practice, the measurements of Weller et al.  indicate that photocorrosion or other processes causing loss of efficiency are still a problem with the dual semiconductor systems. For example, the photocurrent quantum yield for a PbS 72 Recent developments in photocatalysis coated nanocrystalline Ti02 photoelectrode decreased from 65%to 25% after 4 days of illumination with 460 nm light. Photocorrosion in aqueous media is a characteristic probIem with narrow band-gap semiconductors such as PbS or CdS. The possibility of using such semiconductor sensitized nanocrystalline Ti02 for gasphase photocatalysis may however be attractive. Studies of Reaction Mechanisms As noted above, the mechanistic events occumng in photocatalysis following creation of holes and electrons in the oxide semiconductor remain a matter of speculation. There have however been several approaches taken recently to provide more definitive mechanistic information, particularly for Ti02 photocatalysts. The interfacial processes of electron and hole transfer from the semiconductor to adsorbed molecules are crucial steps in any photocatalytic mechanism. The surface structure of micro or nanocrystalline Ti02 is however difficult to study and define. There have been a number of surface science studies made of single crystals of Ti02 which attempt to relate chemistry observed on the well-defined single crystal surfaces to that occumng on photocatalysts. These surface science studies have been reviewed in detail by Yates et al. . Most authors have studied single crystals of rutile. Photoemission spectroscopy shows that the electronic structure of nearly perfect rutile surfaces is closely similar to that of the bulk , but the introduction of surface defects (Ti3+ sites) introduces additional states in the band-gap around 0.8 eV below the F e d level. One of the few studies on anatase single crystals reports similar electronic structure to rutile surfaces . The chemisorption properties of single crystal Ti02 surfaces are dominated by defect sites. Dissociation of water fills surface oxygen vacancies to produce hydroxyl groups , while coordinatively unsaturated Ti cations are highly reactive towards adsorbed organic molecules. To date, only a single surface science study of a photooxidation process on a well defined single crystal surface has been reported. Lu et al.  examined the photodegradation of methyl chloride on a rutile (1 10) surface, and showed that the active sites for this reaction are surface Ti3+ sites 73 R.F. Howe (oxygen vacancies). Interaction of oxygen with these sites produced a species activated by electron transfer from the conduction band of electrons generated by band-gap irradiation, and this activated oxygen species then reacted with methyl chloride. It should be emphasized that the circumstances under which the experiments of Lu et al. were undertaken (low pressure gas phase reaction) are quite different from those of normal photocatalysis. Nevertheless, the demonstration that catalyst mediated photoactivation of oxygen is the crucial step in photodegradation of methyl chloride will undoubtedly have bearing on mechanistic considerations in photocatalysis (and finds some support from EPR studies which failed to detect surface hydroxyl radicals commonly postulated as the reactive species [781 ). The photoelectrochemical properties of single crystals of anatase were recently reported by Gratzel et al. . This work confirmed that the main difference between anatase and rutile surfaces is the position of the conduction band edge; the flat band potential of anatase (101) surfaces is shifted by 0.2 eV relative to that of rutile (001). Interestingly, the photosensitized electron injection from an adsorbed ruthenium complex dye into the conduction band of the anatase single crystal had a quantum efficiency 3-fold less than that of the corresponding nanocrystalline Ti02 thin film electrode, suggesting that surface texture plays an important role. The other new development in studying mechanisms of photocatalytic reactions is the adoption of laser flash photolysis techniques to quantify the kinetics of various mechanistic steps. Early studies on aqueous colloidal sols by Rothenberger et al. [941 used pico- and nano-second transient absorbtion measurements to monitor the fate of photo-produced holes and electrons. Electron trapping was reported to occur within 30 ps, the time resolution of the system used, while the trapping time for holes was estimated to be <250 ns. Second order kinetics were observed for e1ectron:hole recombination, and a mean lifetime for the e1ectron:hole pair of 30 ns was estimated. Subsequent studies with femto-second time resolution on similar aqueous colloidal sols revealed that e1ectron:hole recombination occurred on a picosecond time scale rather than nano-second, and the electron trapping time was determined to be approx. 200 femto-seconds . 74 Recent developments in photocatalysis Bowman et al. [97, 981 have recently applied for the first time the technique of time-resolved diffuse reflectance spectroscopy to observe transient events in solid Ti02 powders.Comparison of electron trapping in quantum sized Ti02 colloids, wet and dry P-25 anatase powder revealed that in all 3 cases electron trapping occurred in less than 200 femto-seconds. The resulting decay of the trapped electron absorbtion signal due to recombination occurred within 50 ps, and was also independent of the nature of the sample. Addition of SCN- as a hole scavenger profoundly lengthened the lifetime of the trapped electrons, indicating that there is efficient competition of charge transfer of holes to the adsorbed SCN- with recombination. Kinetic data of this kind are essential to the development of detailed mechanistic models of photocatalysis. Other Oxide Photocatalysts Although Ti02 has received by far the most attention in the photocatalysis literature, other oxide semiconductors are attracting some interest (as noted above, non-oxide binary semiconductors such as CdS, CdSe or PbS are regarded as insufficiently stable for catalysis, at least in aqueous media). The band gap of ZnO (3.2 eV) is identical to that of anatase, and some early reports describe photoactivity of ZnO for CO oxidation, for example [go]. ZnO is however also unstable in water, leading to catalyst deactivation . Other candidate oxides include W 0 3 (2.8 eV) and Sr Ti03 (3.2 eV). SrTi0-j modified with NiO has been shown to decompose water into H2 and 0 2 catalytically , and evidence has been presented for the presence of NiO, Ni and Ni(OH)2 phases in the active photocatalysts . W03 and doped W03 catalysts have been shown  to be active for the photochemical reaction of methane with water to form methanol : C& + H20 + CH30H + H2 ...( 2) 75 R.F. Howe This reaction occurs homogeneously in the gas phase under irradiation with 185 nm light [ 8 5 ] , through a mechanism involving photolysis of water to produce OH radicals which abstract a hydrogen atom from CH4. Taylor et al.  found that adding a La promoted W 0 3 catalyst to this reaction promoted the rate of methanol production, as illustrated in Figure 7. Cu doping inhibited methanol formation, whereas doping with Pt or Cu + La had no effect on the homogeneous (no catalyst) reaction. Addition of hydrogen peroxide to the reaction enhanced 1O-fold the yield of methanol in the presence of the WOg catalysts, supporting the suggestion that OH radicals play an important role. There remain many unanswered questions about this reaction, the exact role of the catalyst, and why La doping is important, but it does represent an interesting new development. TiME (HOURS) Figure 7. Methanol production @om reaction of CH4 with H 2 0 over various catalysts under w illumination (reproduced with permission@om reference f84J). No catalyst (-El-); W O r L a (-); WOyPt (-0-); WOrLa-Cu (*); and woj-cu (+). Other non-conventional oxide photocatalysts have also been reported recently [86, 871. K4Nb6017 shows activity for the water decomposition reaction with a 76 Recent developments in photocatalysis noticeably high quantum efficiency. The band-gap of K4Nb6017 is 3.5 eV, which is also in the useful range for practical photocatalysis. Domen et al.  investigated the dependence of reaction rate on light intensity, and concluded that the reaction mechanism involves competition between hole and electron reaction with H20 and recombination. They suggest an upper limit for the quantum efficiency of 20%. Other ternary oxide systems based on perovskite structures are referred to in the patent literature . Reactor Design Oxide photocatalysis has been traditionally practiced with slurries of Ti02 dispersed in water (for aqueous reactions), or with a thin-layer fixed bed of Ti02 powder for gas phase reactions. For aqueous systems, the problem of catalyst recovery prevents the use of this method for large scale water treatment. For both liquid and gas phase reactions, powdered catalyst beds are inefficient from the aspect of irradiation. A number of groups have therefore begun investigating reactors in which thin coatings of the photocatalyst are employed, such that all Ti02 particles receive in principle uniform irradiation. The important issues then become preparation of the coating (and its adherence to the substrate) and design of the reactor to achieve optimum contact between reactants and the irradiated photocatalyst.  Anderson et al.  have described 3 different types of reactor design for the specific task of photodecomposing organic contaminants in water. These are shown in Figures 8-10. The annular photoreactor illustrated in Figure 8 houses a mercury vapour lamp which irradiates a dipcoated titania sol film; reaction solution suitably oxygenated is pumped through the annular space between the film and the outer cooling jacket. Problems encountered with this reactor include difficulties with temperature control, transmission of radiation through the catalyst film inducing homogeneous photochemistry in the annular space, and perhaps most seriously, the limited amount of catalyst that is in the reactor (the catalyst film thickness is typically less than 1 micron). 77 R.F. Howe Figure 9 shows an alternative flat plate design in which a glass slide coated with Ti02 is sandwiched between two halves of a teflon block. In this case sample heating is not a problem, and the light intensity is readily controlled with an external lamp (it is also possible to place an array of flat plate reactors under a single lamp.) This design still suffers however from the same process capacity limitation as the annular reactor. Inner tube I Catalyst / coating Cooling chamber/ Reaction chamber / 02/Air Cooling entry Figure 8. Annular plzotoreacrorfor water purification (reproduced with pernrission from reference 139j ). A third design, which allows more catalyst to be incorporated into the reactor without losing irradiation efficiency is shown in Figure 10. In this design, described 78 Recent developments in photocatalysis in detail in reference , light is coupled into the reactor via optical fibres coated with titania thin films. Coating of a bundle of e.g. 200 fibres with Ti02 gives a much higher loading of catalyst in the reactor, and has given conversions and apparent quantum efficiencies comparable with those of a slurry reactor . A practical difficulty referred to in  is the fragility of the Ti02 coated optical fibres; improving the durability of the nanocrystalline Ti02 coatings is an obvious objective. Nevertheless, the fibre optic reactor concept offers many attractive advantages, including the possibilities of use in remote locations. Figure 9. Flat plate reactur for water purification (reproduced with permission from reference  ). Catalyst recovery is not an issue in performing photocatalysis on gas phase reactions. The optical efficiencies of the flat plate and fibre optic designs would 79 R.F. Howe also make them attractive however for gas phase reactions. A flat plate reactor has been used by Obee  to remove formaldehyde and toluene from the gas phase. Nimlos et al.  used small diameter coated tubes in a reactor design otherwise not dissimilar to that in Figure 9 for oxidation of ethanol in the gas phase, while Anderson et al.  refer to use of a packed (coated) capillary tube reactor in their gas phase studies. , Optical fibers coated with anatase membrane Contaminated water in Figure 10. Opticalfibre reactor (reproduced wit11permissionfrom refereme  ). Conclusions and Outlook After 20 years of largely empirical and often contradictory studies, heterogeneous photocatalysis is now approaching a level of understanding where practical implementation on a large scale, particularly for environmental applications, can be contemplated. Some particular issues which still confront the field are : 80 Recent developments in photocatalysis *The apparent improvements in catalytic performance of Ti02 that can be achieved by doping with transition metal ions must be understood more clearly if this method is to be used in practice. The sensitization of Ti02 with a second component to enhance activity and shift the wavelength of irradiation into the visible region is a goal which should be pursued further. Organic and organometallic dyes of the type used to sensitize Ti02 electrodes are less likely to succeed in photocatalysis, but the concept of photosensitizing with narrow band-gap semiconductors is an attractive one which warrants investigation. *The use of more novel semiconductors (WO3, KqNb6017, perovskite mixed oxides) in practical reactors will require methods for preparing high surface area thin films of these materials to be developed. As in all other forms of catalysis, improved understanding of reaction mechanisms in photocatalysis is another important objective. Time resolved optical spectroscopy and EPR spectroscopy will play important roles. Optimization of reactor design. Further improvements in reactor design for both liquid phase and gas phase reactions may be anticipated. Environmental catalysis has become a major theme of the 1990s. Heterogeneous photocatalysis is now poised to play a pivotal role in environmental control, and may contribute also to new pathways to chemical products. Acknowledgements The author's current research on titania sol-gel films is supported by the Australian Research Council as part of a joint project with the Advanced Materials Group of ANSTO (Drs J. Bartlett and J. Woolfrey). References 1. 2. 3. 4. 5. Formenti, M. and Teichner, S.J. 1978. Catalysis Chemical Society Specialist Periodical Reports), 2, 87. Serpone, N., and Pelizetti, E. eds. 1989. "Photocatalysis - Fundamentals and Applications",Wiley, New York.. Fox,M.A., and Dulay, M.T. 1993. Chem. Rev., 93,341. Ollis, D.F., and Al-Ekabi, H. (eds.) 1993. "PhotocatalyticPurification and Treatment of Water and Air", Elsevier, New York. Schiavello, M.(ed.) 1988. "Photocatalysisand the Environment", Kluwer. 81 R.F. Howe 6. Hoffmann, M.R.; Martin, S.T.; Choi, W., and Bahneman, D.W. 1995. Chem. Rev., 95.69. 7. Hagfeldt, A., and Gratzel, M., 1995. Chem. Rev., 95,49. 8. Mathews, R.W. 1992. Pure and Applied Chem., 64,1285. 9. Linsebigler, A.; Lu, G., and Yates, J.T. 1995. Chem. Rev., 95,735. 10 Fujishima, A., and Honda, K. 1972. Nature, 37,2338. 1 1 . Rao. M.V.; Rajeshwar, K.; Vemerker, V.R., and Dubow, J. 1980. J. Phys. Chem., 84, 1987. 12. Tanaka, K. 1974. J. Phys.Chem., 78,555. 13. Courbon, H.;Formenti, M., and Pichat, P. 1977. J. PhysChem., 81,550. 14. Sato, S., and White, J.M. 1981. J. Phys.Chem., 85,592. 15. Kalyanasundaram, K.; Gratzel, M., and Pelizzetti, E. 1986. Coord. Chem. Rev., 69, 57. 16. Duonghong, D.; Borgarello, E., and Gratzel, M. 1981. J. Am. Chem. SOC.,103,4685. 17. Kawai, T., and Sakata, T. 1980. J. Chem. SOC.Chem. Commun., 694. 18. Thevenet, A,; Juillet, F., and Teichner, S.J. 1974. Japan J. Appl. Phys., Supp 2, 2, 529. 19. Bickley, R.I.; Jayanty, R.K.; Navio, J.A.; Real, C., and Macias, M. 1991. Surface Science 251,252, 1052. 20. Frank, S.N., and Bard, A.J. 1977. J. Phys. Chem., 81,1484. 21. Draper, R.B., and Fox, M.A. 1990. Langmuir, 6, 1396. 22. Schrauzer, G.N.; and Guth, T.D. 1977. J. Am. Chem. SOC.,99,7 189. 23. Inoue, I.; Fujishima, A.; Konishi, S., and Honda, K. 1979. Nature, 277,637. 24. Hirana, K.; Inoue, K., and Yatsu, T. 1992. J.Photochem.Photobio1.A : Chem., 64, 255. 25. Appo, M., and Chiba, K.J. 1992. J. Molec. Catal., 74,207. 26. Djeghri, N.; Formenti, M.; Juillet, F., and Teichner, S.J. 1974. Farad. Disc. Chem. SOC.,58, 185. 27. Phillips, L.A., and Raupp, G.B. 1992. J. Molec. Catal.. 77,297. 28. Dibble, L.A., and Raupp, G.B. 1992. Environ. Sci. Tech., 26.492. 29. d’Henneze1, O., and Ollis, D.F. 1996. Hightower, J.W.; Delgass, W.N.; Inglesia, E., and Bell, A.T. (eds) 11th International Congress on Catalysis, 101, 435. Elsevier Stud. Surface Sci. Catal. 30. Stafford. U.; Gray, K.A.; Kamat, P.V., and Varma, A. 1993. Chem. Phys. Lett., 205, 55. 31. Mao, Y.; Schoneich, C., and Asmus, K.D. 1991. J. Phys. Chem., 95,80. 32. Sabin, F.; Turk, T.. and Vogler, AJ. 1992. Photochem. Photobiol. A : Chem., 63,99. 33. Cai, R.; Hashimoto, K.; Itoh, K.; Kubota, Y., and Fujishima, A. 1991. Bull. Chem. SOC.Japan, 64, 1268. 34. Cai, R.; Kubota, Y.; Shuin. T.; Sakai, H.; Hashimoto, K., and Fujishima, A. 1992. Cancer Research, 52,2346. 35. Fitzmaurice, D.J.; Eschle, M.; Frei, H., and Moser, J. 1993. J. Phys. Chem., 97, 3806. 36. Ishitani, 0.; Inoue, C.;Suzuki, Y., and Ibusuki, T. 1993. J. Photochem. Photobiol. A : Chem., 72,269. 37. Ichikawa, S., and Doi, R. 1996. Catal. Today, 27, 271. 38. Miller, R., and Fox, R. 1993. In: Ollis, and Al-Ekabi, H. (eds.) Photocatalytic Purification and Treatment of Water and Air, p.183. Elsevier, New York 39. Zeltner, W.A.; Hill, C.G., and Anderson, M.A. May 1993. Chemtech., 21. 40. Mathews, R.W. 1990. U.S.Patent 5,244,811, Anal. Chim. Acta, 233, 171. 82 Recent developments in photocatalysis 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. Tanaka, K.; Hisanaga, T., and Rivers. A.P. 1993. In Ollis, D.F., and Al-Ekabi, H. (eds.) Photocatalytic Purification and Treatment of Water and hr,p.169, Elsevier , New York. Fonnenti, M.; Juillet, F.; Merideau, P., and Teichner, S.J. 1 Nov, 1971. Chemtech., 680. Gratzel, M. 1981. Acc. Chem. Res., 14, 376. Brinker, C.J., and Scherer, G.W. 1990. "Sol-Gel Science : The Physics and Chemistry of Sol-Gel Processing", Academic Press, New York. Bartlett, J.R., and Woolfrey, J.L. 1992. Mater. Res. Soc. Symp. Proc., 271, 309. Hoffman, A.J.; Yee, H.; Mills, G., and Hoffman, M.R. 1992. J. Phys. Chem., 96, 5540. Moser, J.; Gratzel M., and Gallay, R. 1987. Helv. Chim. Acta, 70, 1596. Howe, R.F., and Gratzel, M. 1985. J. Phys. Chem., 89,4495. Moser J., and Gratzel, M. 1993. Chem. Phys., 176, 493. Martin, S.; Momson, C.L., and Hoffmann, M.R. 1994. J. Phys. Chem., 98, 13695. Hanley, T.L. 1997. Ph.D. Thesis, University of New South Wales. Hanley, T.L.; Luca, V.; Bartlett, J.L.; Woolfrey, J., and Howe, R.F. manuscript in preparation. Cui, H.; Shen, H-S.; Gao, Y-M.; Dwight. K., and Wold. A. 1993. Mat. Res. Bull., 28,195 Kato, K.; Tsuzuki, A.; Taoda, H.; Toni, Y.; Ato, T., and Butsugan, Y. 1994. J. Mater. Sci., 29,591 1. ORegan, B., and Gratzel, M. 1 9 9 1 . Nature, 353,737. Nazeruddin, M.K.; Kay, A.; Rodicio, I.; Humphry-Baker, R.; Muller, E.; Liska, P.; Vlachopoulos, N., and Gratzel, M. 1993. J. Am. Chem. SOC.,115,6382. ORegan, B.; Moser, J.; Anderson, M., and Gratzel, M. 1990. J. Phys. Chem., 94, 8720. Terwillinger, C.D., and Chaing, Y.-M. 1993. Nanostructured Materials 2.37. Yamashita. H.; Ichihasi, Y.; Harada, M.; Stewart, G.; Fox, M.A., and Anpo, M. 1996. J. Catal.. 158.97. Mu, W.; Hemnann, J-M., and Pichat, P. 1989. Catal. Lett., 3.73. Karakitsou, K.E., and Verykios, X.E. 1993. J. Phys. Chem., 97, 1184. Palmisano, L.; Augugliaro, V.; Sclafani, A., and Shiavello, M. 1988. J. Phys. Chem., 92,6710. LUO,Z., and Gao, Q-H.Photochem. 1992. Photobiol. A : Chem., 63, 367. Borgarello, E.; Kiwi, J.; Gratzel, M.; Pellizetti, E.. and Visca, M. 1982. J. Am. Chem. SOC.,104,2996. Gratzel, M., and Howe, R.F. 1990 J. Phys. Chem., 94, 2566. Choi, W.; Termin, A., and Hoffmann, M.R. 1994. J. Phys. Chem., 98, 13669. Martin, S.T.; Momson, C.L., and Hoffmann, M.R. 1994. J. Phys. Chem., 98, 13695. Luca, V., Thomson, S., and Howe, R.F. 1997. J. Chem. SOC.Faraday Trans. 93,2195 Hanley, T., Luca V., Pickering, I., and Howe, R.F. to be published. Do. Y.R., Lee, W., Dwight, K., and Wold, A. 1994. J. Solid State Chem., 108, 198. Cui, H., Dwight, K., Soled, S., and Wold, A. 1995. J. Solid State Chem., 115, 187. Fu. X.; Clark, L.A.; Yang, Q., and Anderson, M.A. 1996. Environ. Sci. Techno]., 30, 647. Vogel, R., Hoyer, P., and Weller, H. 1994. J. Phys. Chem., 98, 3183. Henrich, V.E. 1985. Rep. Progr. Phys., 48, 1481. See, A.K., and Bartynski, R.A. 1992. J. Vac. Sci. Techno]., A 10,2591. Lu, G., Linsebigler, A., and Yates, J.T. 1994. J. Phys. Chem., 98, 11733. Lu, G.; Linesbigler, A., and Yates. J.T. 1995. J. Phys. Chem., 99,7626. 83 R.F. Howe 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. Howe, R.F., and Gratzel, M. 1987. J. Phys. Chem., 91, 3906. Kavan, L., Gratzel, M., Gilbert, S.E.; Klemenz, C., and Scheel, H.J. 1996. J. Am. Chem. SOC.,118,6716. Murphy, W.R.; Veerkamp, T.F., and Leland, T.L. 1976. J. Catal., 43,304. Carraway, E.R.; Hoffman, A.J., and Hoffmann, M.R. 1994. Environ. Sci. Technol., 28, 786. Domen, K. Naito, S. Soma, M. Onishi, T. and Tamaru, K. 1982. J. Phys. Chem., 86,3657. Domen, K.; Kudo, A.; Onishi, T.; Kosugi, N., and Kuroda, H. 1986. J. Phys. Chem., 90, 292. Taylor, C.E.; Noceti, R.; DEste, J.R., and Martello, D.V. 1996. In : Hightower, J.W. Delgass, W.N. Iglesia, E. and Bell, A.T. (eds), Proc.llth International Congress on Catalysis, Elsevier Stud. Surface Sci. Catal., 101,407. Ogura, K., and Kataoka, M. 1988. J. Mol.Catal., 43,371. Sayama, K.; Tanaka, A. Domen, K. Maruya, K., and Onishi, T. 1990. Catal. Lett., 4, 217. Tabata, S., Onishi, H., Yagasaki, E.; Ippommatsu, M., and Domen, K. 1994. Catal. Lett., 28,417. Tanaka A., and Domen, K. 1995. Jap.Patent 07 24329. Legrini, E., Oliveros, E., and Braun, A.M. 1993. Chem. Rev., 93,671. Peill N.J., and Hoffmann, M.R. 1995. Environ. Sci. Technol., 29, 2974. Peill, N.J., and Hoffmann, M.R. 1996. Environ. Sci. Technol., 30, 2806. Obee, T.N. 1996. Environ. Sci. Technol., 30, 3578. Nimlos, M.R.; Wolfrum, E.J.; Brewer, M.L.: Fennek J.A., and Bintner, G. 1996. Environ. Sci. Technol., 30,3102. Rothenberger, G.; Moser, J.; Gratzel, M.;Serpone, N., and Sharma, D.K. 1985. J. Am. Chem. SOC.,107,8054. Skinner, D.E.; Colombo, D.P.; Cavaleri, J.J., and Bowman, R. 1995. J.Phys. Chem., 99,7853. Serpone, N.; Lawless, D.; Khairutdinov, R., and Pelizzetti, E. 1995. J. Phys. Chem., 99, 16655. Colombo, D.P., and Bowman, R.M. 1996. J. Phys. Chem.. 100,11752. Colombo, D.P, and Bowman, R.M. 1996. J. Phys. Chem., 100,18445. Received: 84 3 'February 1997; Accepted after revision: 3 March 1997.