You are on page 1of 25

Surface Technology, 20 (1983) 357 - 381

357

A REVIEW OF PHOTOELECTROCHEMICAL METHODS FOR THE UTILIZATION OF SOLAR ENERGY L. PERALDO BICELLI
Department o f Applied Physical Chemistry o f the Milan Polytechnic, Research Centre on Electrode Processes, Consiglio Nazionale delle Ricerche, Piazza Leonardo da Vinci 32, 20133 Milan (Italy)

(Received September 20, 1982) Summary This review is concerned with photoelectrochemical methods which can directly transform solar energy into electrical energy, chemical products and synthetic fuels through appropriate devices using semiconductor electrodes. Systematic research in this field began in the 1970s, but it has only recently become very widespread and at present there are numerous research projects being conducted.

1. Introduction The Sun clearly represents a major source of energy if we consider the enormous a m o u n t of radiant energy that it transmits to the Earth's surface each day. In fact, the average flux of energy from the Sun is 1395 W m -2 (per unit of a surface that is perpendicularly exposed to the flux itself) measured outside the atmosphere at a distance half-way between the Earth and the Sun. Therefore, the total radiation intercepted by the Earth's surface is of the order of 1.73 1017 W [1]. Approximately 30% of this energy flux is directly reflected and dispersed into space; another 47% is absorbed by the atmosphere, the Earth's surface and the oceans, while the remaining 23% is consumed through evaporation, convection, precipitation and run-off, which characterize the hydrological cycle. Only a small fraction (about 0.03%) is stored in the form of chemical energy during photosynthesis (the process which provides nourishment for life itself and which was the source for all the fossil fuels that our present civilization is rapidly consuming [1 ] ). Solar energy can of course be transformed into different forms of energy that are useful to humanity, and there are various methods for doing this.

2. Classification of photoelectrochemical cells


A photoelectrochemical cell is defined as a cell in which the irradiation of an electrode in contact with an appropriate electrolyte produces a change 0376-4583/83/$3.00 Elsevier Sequoia/Printed in The Netherlands

358 in the electrode potential with respect to a reference electrode (under opencircuit conditions) or produces a change in the current flowing in the galvanic cell containing the electrode (under closed-circuit conditions). A possible m e t h o d of classifying photoelectrochemical cells takes into account the value and the sign of the free-enthalpy variation of the total chain reaction which takes place in the cell itself. 2.1. Photovoltaic electrochemical cells An example of this type of cell is the cell n-GaAsh K2Se, K2Se2, KOHLC which has been studied at the Bell Telephone Laboratories in the U.S.A. The electrolyte contains a single dissolved redox couple; thus, after illumination, oxidation of the reduced species takes place at the semiconductor photoanode, and exactly the opposite reaction occurs at the cathode, so that the composition of the electrolyte remains unchanged (at least in theory). Consequently, there is zero variation in the free enthalpy of the total process and solar energy is transformed into electrical energy in the same way as in a photovoltaic device with a solid-solid junction. Gerischer [2], who first studied the potentialities of these cells, also called them regenerative cells, precisely because of their ability to operate indefinitely. Since sunlight is a periodic and stochastic source of energy, any practical applications obviously require that these cells be integrated with a system for storing the electrical energy that is produced. The storage system may be independent, but special devices have been suggested with in situ storage, i.e. containing both the photoelectrochemical and the storage cells [3 - 6]. 2.2. Photoelectrosynthetic cells The electrolyte contains two different dissolved redox couples; thus, after illumination, the reaction that takes place at the anode is different from that at the cathode and, consequently, the solar energy causes a particular chemical process to occur, with modifications in the composition of the electrolyte. In this type of device the products formed, which are of interest because of their high energy, can be utilized to generate electrical energy in a separate cell during periods of darkness or to generate electrical energy even in the same cell which then functions as a fuel cell. These products can also be used for other purposes when needed. Since in these cells the variation in the free enthalpy of the total process is not zero, two subdivisions can be distinguished, which depend on whether this variation is positive or negative. 2.2.1. Photoelectrolytic cells: AG > 0 In this type of cell the chain reaction does not occur spontaneously and the chemical energy needed for the process is furnished by light energy. With devices of this type, we obviously want to transform widely available reagents into highly useful products, such as fuels. The most important reaction is the photodecomposition of water with the formation of hydrogen and oxygen, e.g. in the cell containing n-TiO21H +, H2 and 02, H2OIPt, which

359 was devised by Fujishima and Honda [ 7 ]. It is precisely the photoelectrolysis process that gave these cells their name.
2.2.2. Photocatalytic cells: AG ~ 0

In this type of cell the total chain reaction process is thermodynamically possible but it is kinetically hindered. Therefore the light energy is utilized, at least in part, to provide the activation energy needed during the reaction. An example is the reaction for the synthesis of ammonia from nitrogen and hydrogen that can be produced in a photoelectrochemical cell. Because of their mode of action, these devices are known as photocatalytic cells.

3. Theory Although it is probably very difficult to find electrode-electrolyte systems which are totally insensitive to absorbed light, the resulting photo electrochemical effects are often very limited. For example, with metallic electrodes which have a clean surface the photoelectrochemical effects are of the order of millivolts or nanoamp~res with quantum yields that are very low. Photoelectrochemical research on semiconductors, particularly in Germany [2, 8], has shown that photoelectrochemical effects are much more prominent in semiconductor electrodes that are either bare or covered with sensitizing dyes. Consequently, most of the devices under consideration at present operate with a semiconductor-electrolyte junction.

3.1. The semiconductor-electrolyte junction It is well known that at the instant at which a semiconductor electrode is dipped into a solution containing a redox couple there is an exchange of electrons between the two phases until the equilibrium conditions are reached. Indeed electrons are transferred by tunnelling from the phase with the smaller work function to that with the higher work function, i.e. from the phase with the higher Fermi level to that with the lower Fermi level. For example, let us now consider an n-type semiconductor which has a higher Fermi level than that of the only redox couple which is present in the solution (Fig. l(a)). After the equilibrium conditions have been reached, electrons are transferred from the semiconducting electrode to the electrolyte. However, owing to the relatively small concentration of the majority carriers in the semiconductor, the transfer process changes (decreases) the electron concentration with respect to the bulk value in a small region close to the interface, which is called the space charge region. In fact, a depletion layer of the majority carriers in the proximity of the junction is formed with a simultaneous bending-down of the semiconductor bands and band edges, whereas the Fermi level remains flat throughout the whole material (Fig. l(b)).

360
Energy

Depletion

EF

'Eee

'Eep

Eep

Eg

Ec EF

--~-

Eredox

E;--I- --

Er~ox

e-

E,,
SE~CONOUCTO
(a)

Eg',

ElY
SEM~O~UCIOR~[~ LECTROLYT[_
(b)

E.
SEMICONOUCTOR~ELECTROLYTE~COUNTERELECTROOE
(c)

Fig. 1. n-type semiconductor-redox electrolyte junction (Ev, valence band edge; Ec, conduction band edge; Eg, band gap; EF, Eredox and EFM , Fermi levels of the semiconductor, the redox electrolyte and the metal counterelectrode respectively; EBp , flat-band potential; EF* , Fermi level of the illuminated semiconductor at open-circuit conditions; Vp, photovoltage at open-circuit conditions; e, electron charge): (a) before equilibrium is reached; (b) in equilibrium in the dark; (c) under illumination in a regenerative photoelectrochemical cell at open-circuit conditions. (From ref. 9.)

3.2. Operating principles o f photoelectrochemical cells


In a s e m i c o n d u c t o r - e l e c t r o l y t e junction illuminated with radiation whose energy is equal to or greater than the band gap energy of the semiconductor, the absorption of a p h o t o n allows an electron t o be trangferred from the valence band to the c o n d u c t i o n band, i.e. an e l e c t r o n - h o l e pair is created. These charge carriers must be prevented from recombining so t hat t h e y can be utilized in order to pr oduce electrochemical reactions at the electrode surface with ions present in the electrolyte. At this p o in t we see how i m p o r t a n t the depletion layer at the semicond u c t o r surface is. In fact, a potential barrier exists in this layer which tends to separate the p h o t o p r o d u c e d e l e c t r o n - h o l e pairs in t hat the majority carriers (the electrons) t end to move downhill towards the bulk of the semic o n d u c t o r and the m i nor i t y carriers (the holes) t end to move uphill towards the s e m i c o n d u c t o r - e l e c t r o l y t e interface. In order for the m i nor i t y carriers t he n to be able to react with the ions present in the solution, a further condition must be satisfied. In fact, the Fermi level (which represents the Nernst potential) of the r e d o x couple must be situated within the forbidden band of the s e m i c o n d u c t o r at flat-band conditions. For n-type semiconductors the Fermi level of the redox couple must be below that of the s em i c onduc t or at flat-band conditions, whereas for p-type semiconductors it must be above it. Therefore, in an n-type semiconductor, which is part of a regenerative photoelectrochemical cell, the holes tend to move towards the electrolyte to p r o d u ce an anodic oxidation reaction, while the electrons move towards the

361

bulk o f the s e m i c o n d u c t o r and, by way of the external circuit, reach the c o u n t e r e l e c t r o d e to p r o d u c e a cathodic reduction reaction (Fig. l ( c ) ) . In a p-type s e m i c o n d u c t o r the opposite occurs. Let us now consider an n-type s e m i c o n d u c t o r in a solution which contains two different r edox couples, e.g. H+IH2 and O21H20 , as shown in Fig. 2(a). The first r e dox couple is the less noble and t h e r e f o r e it has the higher energy level. To understand this poi nt better, the absolute and the electrochemical energy scales are compared in Fig. 3. The t w o scales differ mainly in the energy level with zero value. In fact the absolute scale refers to the energy of an electron at rest in the vacuum at infinite distance, and the electrochemical scale refers to the normal hydrogen electrode (the H+IH2
Energy

1 e

"

h S M D D C O ELE~IROLYTE S M O D CO E CNUT R I E E N U I LELEtTROtVIE=:OBNTERELECIII~ESEMICONDUCI~ ELECI~LYIE :OUNIERELE'111~ R

E,

Ev I

(a)

(b)

(c)

Fig. 2. n-type semiconductor-redox electrolyte junction (Ev, valence band edge; Ec, conduction band edge; Eg, band gap; H+IH2, O21H20 and EFM , Fermi levels of the redox couples H+IH2 and O2]H20 in the electrolyte and of the metal counterelectrode respectively; EBp , flat-band potential; EF* , Fermi level of the illuminated semiconductor at open-circuit conditions; Vp, photovoltage at open-circuit conditions; Ep, external polarization; ~Ta and tic, anodic and cathodic overvoltages respectively; e, electron charge): (a) before equilibrium is reached; (b) in equilibrium in the dark; (c) under illumination in a photoelectrolysis cell at open-circuit conditions. (From ref. 9.)
0,0 0,5 -4, S -4,0 -3,0 -Z,0 -1,0 0,0 .1,0 *Z,0 .3,0 Electrochemical scale (Vott) H+IH 2

Vacuum
-

-I,5 -2,5 -3,5 -4,5 -5,5


-E.5

-7,5 Energy scale


(eV)

Fig. 3. Comparison between the absolute and the electrochemical energy scales. (From ref. 9.)

362 redox couple in standard conditions). It should be observed that, contrary to the usual electrochemical conventions, the positive direction of the electrochemical scale is now downwards. Let us return again to the semiconductor-redox electrolyte junction. When the equilibrium conditions are reached, a depletion layer is formed and the Fermi levels in the various phases are equal (Fig. 2(b)). Since there are two possible redox levels in the solution, the Fermi level of the electrolyte can be located anywhere between them, depending on the relative concentrations of hydrogen and oxygen in the solution. Thus, for example, if hydrogen is bubbled through the solution, the Fermi level would be located at the H+I H2 redox level; if oxygen is bubbled through the solution, it would be located at the O21H20 redox level. In the figure the Fermi level is arbitrarily shown to be not far from the O21H20 level. Figure 2 also shows a photoelectrolysis cell under illumination. At the anode (e.g. n-TiO2), holes are consumed according to the oxidation reaction H20+2h + hp 1 ~ - - O 2 + 2H 2

as electrons reach the cathode (e.g. platinum) through the external circuit, the hydrogen reduction reaction takes place: 2H + + 2ehv H20 - ~H2

Water is finally decomposed into hydrogen and oxygen: 1 ~ H2 + ~ O2

As previously stated, in order for the process to occur on n-type semiconductors, the Fermi level of both redox couples must be situated within the band gap, below the flat-band potential. This is not the case for the H+IH2 redox couple in Fig. 2(c). Therefore the counterelectrode has to be cathodically polarized by application of an external bias. This point will be referred to later. However, when all theoretical conditions are satisfied, the redox couple with the higher redox potential (lower energy level) is oxidized at the anode and the redox couple with the lower redox potential is reduced at the cathode in a photoelectrolysis cell, whereas the reverse occurs in a photocatalytic cell. In this way, solar energy is converted into chemical energy. In a photoelectrochemical cell working under short-circuit conditions, the Fermi levels (which also represent the electrochemical potential of the electrons) of the semiconductor, the electrolyte and the counterelectrode are equal for the three different phases, and the circulating photocurrent is the highest possible. On the contrary, under open-circuit conditions the Fermi level of the illuminated semiconductor is different from that of the semiconductor in the dark (and therefore different from that of the electrolyte), it is higher in n-type semiconductors and lower in p-type semiconductors. This shift in the Fermi level towards its original position is equal to the

363

photovoltage and it reaches its m a x i m u m value for very high illumination intensities of the semiconductor when flat-band conditions are reached. By external closure of the circuit on a suitable load, it is possible to obtain the entire current-voltage characteristic of the cell (Fig. 4). Thus, photoelectrochemical cells could be designed for optimal conversion of solar energy into chemical energy (short-circuit working conditions with maximum circulating current) or for optimal conversion into electricity (with the circuit closed on an appropriate external load in order to have the maximum power efficiency).

A
40

<I

E u 3; E

>,

'E

o16

02 04 cet[ voltage, V

06

Fig. 4. Current d e n s i t y - v o l t a g e curves of the n-MoSe212 M KI, 0.01 M I2[Pt regenerative p h o t o e l e c t r o c h e m i c a l cell with a single-crystal electrode with a very s m o o t h surface w h e n illuminated with various light intensities: curve A, 160 mW cm -2 (power efficiency of 7.0%, fill factor of 0.63); curve B, 100 mW cm -2 (power efficiency of 7.3%, fill factor of 0.58); curve C, 40 mW cm -2 (power efficiency of 7.6%, fill factor of 0.58). ( F r o m ref. 10.)

The actual power conversion efficiency depends on the p h o t o c u r r e n t voltage characteristic of the half-cell. If this is rectangular, the half-cell will deliver the nominal power; if it is linear, implying an internal ohmic resistance, the m a x i m u m deliverable power will be one-quarter of the nominal power. The ratio of the delivered power to the nominal power is called the fill factor. This factor, together with the power conversion efficiency, the open-circuit photovoltage and the short-circuit photocurrent, is a very important parameter in determining the behaviour of the cell.

3.3. Photocorrosion and photostability Because of the effect of illumination, electrochemical reactions occur on semiconductor electrodes and non-equilibrium hole-electron pairs are

364 produced. Hence the desired photoelectrochemical reactions may be accompanied by other unwanted reactions, e.g. photodecomposition of the material. In fact, the major problem in making practical photoelectrochemical solar cells is the prevention of corrosion of the semiconductor electrode. Photoproduced minority carriers move towards the semiconductor-electrolyte interface and break the bonds between the surface atoms and the electrode substrates. Then the surface atoms can be incorporated into a new phase or become solvated. These processes which involve the decomposition of the material would be thermodynamically impossible only if their Fermi level were located outside the forbidden band gap, but this never seems to be the case in any of the materials available at present. Therefore, there will generally be competition between the various possible processes and, from a t h e r m o d y n a m i c point of view, the processes with a higher free enthalpy will be favoured. Nevertheless, kinetic impediments intervene against unwanted reactions, thus allowing us even to use electrode materials that would normally be unstable. An a t t e m p t is usually made to favour such kinetic impediments by utilizing numerous factors, particularly (1) the choice of the redox couple and the electrolyte, (2) the selection of the pH of the electrolytic solution, as well as its general composition (concentration of the redox couple, addition of inert salts etc.) and (3) the operative conditions (stirring, current density etc.). Thus, in order to suppress corrosion, the kinetics of the redox couple introduced into the solution must be fast enough to capture the carriers as they arrive at the surface. Therefore, competition between carrier capture by a surface atom of the host crystal (corrosion) and carrier capture by the redox species in solution assumes an essential role in the functioning of the device. 4. Examples We shall now briefly review the problems and the information available at present for the various types of cells under consideration, i.e. photovoltaic and photoelectrosynthetic cells, the latter even in the form of semiconductor powder suspensions, and cells with organic semiconductors. A brief mention will also be made of systems which lie outside the scope of this review but which are of interest in that they allow us to complete the picture of solidliquid systems that can transform solar energy into other forms of energy. These systems are photogalvanic devices in which the incident radiation is absorbed by particular molecules present in the solution and photochemical systems in which the solid phase does not function as an electrode, but rather as a catalyst.
4.1. P h o t o v o l t a i c electrochemical cells Regenerative cells are at present under extensive study because of interest in their potential for practical applications and because of the advantages that they present with respect to cells with a solid-solid junction. Their two

365 basic advantages are (1) the easy attainment of the junction that allows for the separation of the photogenerated charges (by simply immersing the semic o n d u c t o r in the electrolyte) and (2) the relatively small sacrifice in conversion efficiency when single crystals are replaced by polycrystals or polycrystalline films. However, there are also some negative factors which must be taken into consideration, including (1) the problems related to the sealing of the cell because of the presence of a liquid and (2) the above-mentioned possibility of photocorrosion, especially in aqueous electrolytes. For this reason the various materials have also been studied in nonaqueous electrolytes (e.g. as reported in ref. 11), in room temperature molten salt electrolytes (e.g. as discussed in ref. 12) and recently also in solid polymer electrolytes [13]. However, the results in terms of conversion efficiency still do not seem comparable with those obtained in aqueous electrolytes. Until now, only limited attention has been given to the possibility of using protective coatings which might broaden the available number of stable semiconductor-electrolyte systems. However, Noufi et al. [14] succeeded in electrodepositing a conductive stable polymer film on a semiconductor electrode to provide protection against photodissolution-degradation. Polypyrrole was deposited anodically onto various semiconductor materials (e.g. cadmium chalcogenides, GaAs and silicon) from non-aqueous solutions containing pyrrole. Typical performance characteristics attainable with such film-coated photoanodes are comparable with those obtained with bare electrodes. Another recent approach to circumventing photocorrosion utilizes stable conducting electrodeposited metal oxide films (e.g. nickel oxide) to act as barriers to ion-solvent transport while permitting efficient electron exchange with the electrolyte [15].

4.1.1. Class I (e.g. CdSe) and class H (e.g. M o S e s) materials With regard to problems of photostability, it should be kept in mind that the first materials used (n-CdS, n-CdSe and n-GaAs) were particularly subject to photodecomposition, in that their valence band is due to anion orbitals and their conduction band is due to cation orbitals. Thus, the photogeneration of the electron-hole pair necessarily involves a weakening of the bond. Considerable improvement was obtained in the stability of these materials (particularly in cadmium chalcogenides) by using adequate redox couples (particularly chalcogenideJpolychalcogenide couples), especially those with the anion in c o m m o n with the semiconductor electrode [3, 16]. Additional improvements have also been observed by adapting the electrode surface. For example, various workers have studied the effect of depositing metals or metal ions onto the electrode surface and have observed an improvement in stability when less than a monolayer is deposited (e.g. gold on GaAs [17] and zinc on CdSe [18]). Moreover, organic reductants have been used as kinetic inhibitors of photocorrosion (e.g. thiocresol and pmercaptobenzoic acid for CdS [19] ).

366 In addition to ionic compounds characterized by phototransitions (p band of the anion -> s band of the cation), Tributsch and coworkers [20] have suggested some dichalcogenides of transition metals (sulphides and selenides of m o l y b d e n u m and tungsten) as materials for photoelectrochemical cells. These materials possess typical layered structures, and both their valence and conduction bands are due to d orbitals of the metallic atom; thus phototransitions occur between bands of the same type and do not cause a weakening of the metal-chalcogen bond. Good photostability and efficiency are predicted for these metals, and this has already been experimentally proved for single-crystal electrodes having active surfaces parallel to the plane of the layers (the van der Waals plane) in aqueous solution containing the I21 I- couple. Conversely, when other planes are also exposed to the electrolyte (e.g. in stepped surfaces (Fig. 5)), the photoelectrochemical properties decrease considerably and the material is subject to photocorrosion. In order to avoid these problems, Parkinson and coworkers [21] and Razzini et al. [22] have suggested that the material should be treated with solutions containing particular molecules capable of blocking the dangling bonds of the transition metal atoms which point towards the electrolyte, corresponding to the edges of the steps. Since they are not shielded by the p orbitals of the chalcogen atom, these orbitals seem to function as recombination centres and seem to be easily subject to corrosive attack [23]. In fact, charge collection scanning microscopy has recently identified steps on the surface of layered materials, growth irregularities and bulk dislocations as recombination sites [24]. After the treatment, electrodes both illuminated and in the dark improve in performance. This improvement, however, does not last very long (a few hours) if pyridine derivatives which seem to semi-intercalate between the layers [21] are used. Conversely, the improvement lasts longer (hundreds of hours) if agents that are capable of forming coordination compounds with transition metals, such as ethylenediaminetetraacetic acid [22], are used. Blocking the recombination centres associated with surface imperfections is only one approach that can be used. An alternative is the dark electrochemically initiated polymerization of an impervious coating at these sites. In fact, recent work indicates that the polymerization reaction occurs only at growth irregularities in the otherwise smooth van der Waals surface, providing for the specific passivation of the recombination sites [25]. From observations of the electrode surface in situ, i.e. directly under the microscope while it is working, differences have been identified even in the smooth regions, e.g. "active" areas, where the performance is greater, and "inactive" areas [26]. The active areas have particular photocatalytic properties. Simple illumination of the electrode under open-circuit conditions causes photo-oxidation of the I- ions contained in the electrolyte to iodine; this reaction does not occur on the inactive areas. Consequently, it is believed that the surfaces of transition metal dichalcogenides may not be homogeneous, probably because of different doping levels [27] or even because of the presence of impurities [10], and it is t h o u g h t that these

367 factors influence the performance and stability of the material. Control of the orientation of the exposed surface does not always seem to be sufficient to obtain active electrodes because even smooth van der Waals' surfaces contain a large number of local imperfections.

4.1.2. Polycrystalline and thin film electrodes Despite various difficulties, it has been possible to prepare polycrystalline and thin film electrodes from materials belonging to class I; although these electrodes are inferior to those of monocrystalline materials, they nevertheless offer some interesting characteristics. Traditional methods which have recently been used are electrodeposition [28], chemical deposition [29], deposition by simultaneous vacuum evaporation of metal and chalcogen on suitable substrates [30] and sintering at high pressures [31]. One of the new methods of fabricating semiconductor layers consists of painting a slurry of the semiconductor on a substrate and is particularly successful for cadmium and zinc chalcogenides [32]. Electrodes based on sprayed thin films of CdS, CdSe, CdSx Sel- x and Bi2Se 3 have also been prepared; these films adhere well to both smooth and porous substrates, and their photoelectrochemical properties can be modified by annealing and surface treatments [33 ]. In fact, the films that are thus obtained are often subjected to particular treatments in order to improve their performance. Much of the recent success in photoelectrochemistry can be attributed precisely to the discovery of heat and chemical treatments or etching procedures which decrease the extent of recombination reactions at the semiconductor electrode. For example, thin film photoelectrodes of CdSe show a substantial increase in solar energy conversion efficiency after annealing [34], photoelectrochemical etching [35] and photoactivation [36]. This last procedure consists of illumination of the semiconductor electrode which is immersed in an appropriate salt solution under closed-circuit conditions. As can be concluded from the above considerations, greater difficulties are encountered in the preparation of polycrystalline films of layered materials. Nevertheless, the problem does n o t seem to be insurmountable, given the recent success of preliminary attempts to obtain functional polycrystalline electrodes. Thin film and polycrystalline photoactive electrodes of MoSe2 have been prepared by various techniques, such as powder pressing, flux annealing of pasted films and addition of organic polymer binders. The action of the polymer matrix is twofold: orientation of the crystallites and passivation of the recombination centres at the surface steps in contact with the electrolyte [37]. Moreover by h o t pressing WSe 2 powder with 2 wt.% Se under vacuum, Parkinson and coworkers [38] have recently prepared polycrystalline n-type WSe2 photoelectrodes, which show light efficiencies of better than 2% in regenerative photoelectrochemical cells.

368

4.2. Photoelectrosynthetic cells 4.2.1. Water photoelectrolysis Ever since Fujishima and Honda [7] published their first results about the possibility of decomposing water through photoelectrochemistry, their research has provoked intense interest owing to the importance of hydrogen as an energy vector and because of the difficulties t h a t are encountered in the thermal decomposition of water. Fujishima and Honda used a photoelectrochemical cell with an anode made up of a semiconductor electrode of n-TiO2 connected to a platinum black counterelectrode through an external load. The semiconductor anode was illuminated with UV radiation since the prohibited band of TiO2 is very high (about 3.3 eV). It was immediately recognized that there are considerable problems with the water photoelectrolysis reaction. First, the value of the free enthalpy of the process is very high, equal to 1.23 eV referred to the absolute energy scale. Furthermore, apart from the t h e r m o d y n a m i c contribution, the effect of electron transfer kinetics across the interface has to be accounted for, since rather high overvoltages can be predicted for the process of oxygen evolution (as well as for hydrogen evolution) on semiconductors [39]. Consequently, it is often necessary to apply an external bias in order to raise the Fermi level of the cathode (Fig. 2(c)), especially because a high hydrogen production requires to be operated in conditions of maximum circulating current, i.e. conditions of m a x i m u m overvoltage. In order for the process to occur without the need for an external bias, it is imperative t h a t semiconductors be used which have a band gap greater than about 2.2 eV and which are thus capable of utilizing only the most energetic part of the solar spectrum. It should be added that, in practice, there are very few semiconducting materials which contain the Fermi levels of the two redox couples (H+l H 2 and O21H20) under study within their band gap (below the flat-band potential if n type, and above it if p type) and which are also chemically stable both in the dark and under illumination. In practice the last requirement limits the fields of possible materials to oxide semiconductors. Nearly all oxide materials investigated so far have been n type and therefore are used as photoanodes. Fujishima and Honda have already described in their first paper [7] how the performance may be improved by choosing a second photoelectrode (a suitable p-type semiconductor) rather than platinum. Unfortunately, there are very few materials to choose from; one of these is p-GaP. However, p-GaP is not very stable in aqueous solutions and it gives a limited photoresponse. A photoelectrochemical cell with n-TiO2 as the photoanode and p-GaP as the photocathode is capable of decomposing water without the intervention of an external bias [40]. Much effort has recently been directed towards the use of semiconductor-based photoelectrochemical solar cells for splitting water into hydrogen and oxygen [41]. The principal aim of m a n y research projects is to find both n-type and p-type semiconductor materials t h a t are stable and capable of absorbing light in the visible region.

369 The research has resulted in n-type semiconductor materials which are stable but which have wide band gaps, or materials which have smaller band gaps but which are unstable under photoelectrolysis conditions [42]. The possibility has therefore been studied of photosensitizing stable semiconductors which have large band gaps by the addition of a dye to the electrolyte in order to extend their photoresponse into the visible region of the spectrum [43]. Unfortunately, it has been found that dyes contributed little to light absorption; quantum efficiencies are of the order of a few per cent at best, and power efficiencies are two orders of magnitude smaller than the quantum efficiencies. This is because only dye molecules or ions adsorbed on the electrode surface can inject electrons into the semiconductor on photoexcitation and thus become effective. Moreover, performance very often deteriorates after the sensitized electrode has been illuminated for many hours. This is the case, for example, with n-TiO2, n-SrTiO3 and n-SnO 2 electrodes sensitized with ruthenium-based dyes. Proposals are thus being made for the design of a dye molecule that will improve this situation [44].

4.2.2. S e m i c o n d u c t o r materials
As we have shown, oxides are extremely important as semiconductor anodes for water photoelectrolysis and for this reason various workers have examined a wide range of them. A recent analysis showed that in practice it is impossible to find oxides which do not contain partially filled d levels with both the small band gap and the large negative flat-band potential that are required for efficient operation in the unbiased photoelectrolysis of water. High efficiency operation with these oxides is only possible in some special examples. For oxides containing partially filled d levels the simultaneous requirements of stability, fiat-band potential and band gap impose severe limitations [45]. Another point is that several oxides which are used as anodes for water photodecomposition (e.g. TiO2, SrTiO3, WO 3 and SnO2) must be made conductive by the creation of oxygen defects. Oxygen deficiencies, however, are u n d o u b t e d l y responsible for the long-term instability of the electrodes studied to date. Therefore the chemical substitution of oxygen by fluorine has been proposed as an alternative m e t h o d for enhancing the conductivity of the electrodes [46]. n-TiO2 {e.g. as discussed in refs. 7 and 4 7 - 5 0 ) seems even t o d a y to represent one of the most important materials because it has been possible to extend its spectral response into the visible portion of the spectrum through sensitization with organic dyes [51] and by doping with some special impurities [50] or with transition metals [52]. Various methods, such as thermochemical oxidation, anodization etc., have been perfected for obtaining polycrystalline films of n-TiO2 which exhibit a high degree of photoelectrochemical activity comparable with that of monocrystalline anodes [53]. Also the m e t h o d of preparing polycrystalline films of n-TiO2 by sintering [54] and sputtering [55] is being studied.

370 Given the promising results obtained with electrodes of TiO2, proposals have also been made for the use of heterojunction photoanodes of TiO2 thin films on thermally grown ~-Fe203 [56] and for the use of mixtures with other oxides, e.g. oxides of ruthenium [57] and vanadium [58]. Other oxides taken into consideration are n-WO 3 [59] (also amorphous and for which a substantial increment is observed in the solar-chemical conversion efficiency during prolonged anodic polarization [ 4 8 , 6 0 ] ) , n-Fe203 [61], n-SnO2 [62], n-SrTiO3 [63] and n-KTaO3 and nKTa0.77Nb0.2303 [64]. Fe203 is a potentially interesting material because it is very easily produced, resists photodecomposition and possesses a flat-band potential close to the hydrogen potential. Its band gap of 2.2 eV allows, in theory, a solar energy conversion efficiency of up to 20%. Unfortunately, Fe203 always displays a low quantum yield when used as a photoanode. Even single-crystal samples show a quantum yield of only 20% at high anodic polarization [65]. Sputtered (amorphous) semiconducting films have also been obtained from this material. These behave as p type, whereas the single-crystal electrodes are always n type and are interesting because the observed photocurrent is constant with time and the surface of the electrode is not damaged [66]. In addition, sintered disks have been prepared of undoped c~-Fe203 and of c~-Fe203 doped with various oxides of group IVa (silicon, germanium, tin and lead) and group IVb (titanium and zirconium) elements to find out which of them shows the highest activity [67]. The results agree with the general aliovalent doping concept that the 4 + ion introduces an electron into the conduction band. Other materials which seem to be very interesting are the lanthanides of chromium, rhodium, vanadium and gold. They have been prepared by heating a paste of La203 with the individual oxides on a titanium base. The anodic photocurrents observed were higher by a factor of 2 - 6 than those obtained with n-TiO2 single-crystal electrodes [68]. As we have shown, there is a certain degree of variety in the alternative materials available for photoanodes, whereas the choice is more limited for photocathodes. In addition to the above-mentioned p-type GaP, studies have also been conducted on p-type InP, particularly for p h o t o p r o d u c t i o n of hydrogen and chlorine, and p-type silicon for water photoelectrolysis in NaOH and NaC1 electrolytes. In order to avoid dissolving the silicon surface, a voltage bias is applied between the p-type silicon cathode and the platinum anode [69] or the electrode surface is coated with a plasma-polymerized siloxane film [70]. Much more promising are the rare earth rhodates with a distorted perovskite structure; these have a band gap of 2.2 eV and can be doped to give either p-type or n-type electrodes. At the Laboratories of du Pont de Nemours and Co. it was found that, when a ceramic p-type LuRhO3 cathode and an n-type TiO2 anode were used, it was possible to photoelectrolyse

371 water in sunlight {without any externally applied voltage) and to generate power simultaneously [ 71]. Cells composed of p-GaP and n-Fe203 can also be taken as examples of systems with two semiconductor electrodes; these cells have recently been proposed for the decomposition of sea water using incident solar radiation. Both photoelectrode materials show enhancements of their photocurrents when catalysts {platinum on p-GaP and ruthenium dioxide on n-Fe203) are deposited onto their surfaces [72]. Finally, water photoelectrolysis cannot be achieved with layer-type transition metal dichalcogenides because of the fact that either a sulphate or a selenate appears as a photoreaction product with water. In addition, none of these compounds has an energy gap exceeding the thermodynamic threshold for water decomposition. However, hydriodic and hydrobromic acids seem to be hydrogen sources that are nearly as convenient as water and photoelectrochemically more practical than water; layered materials have been proposed for the photodecomposition of these acids. 4.2.3. Other photoelectrochemical processes In addition to water photoelectrolysis, other processes have been studied for producing hydrogen. One of the most important is the photoelectrosynthesis of hydrogen and halogen from aqueous solutions of hydrohalic acids. In these processes, in order to create a device having practical applications, it is imperative that the electrochemical system is integrated with another device (usually chemical) which permits the regeneration of the initial reactive species. One example of a photoelectrosynthetic cell is the p-InP]HClaq, KClaqJPt cell, which is capable of converting 12% of the incident radiation into hydrogen and chlorine. Hydrogen evolution occurs on microscopic islands of catalysts such as rhodium, ruthenium and platinum located on the semiconductor [73]. Another interesting example is the p-WSe21MV 2 I-InWSe 2 cell {where MV - m e t h y l viologen), which allows the p h o t o p r o d u c t i o n of hydrogen and iodine [74]. A completely different process is represented by the photoelectrochemical iodide-mediated oxidation of SO2 in strong acid solutions at illuminated metal dichalcogenide electrodes (n-MoS2, n-MoSe 2 or n-WS2). The cathode reaction is still hydrogen evolution, but the overall reaction is now 2H20
+

SO 2

) H 2 + H2SO 4

and is part of a hybrid cycle for splitting water into hydrogen and oxygen [75]. In fact, thermoelectrochemical hybrid cycles for hydrogen production have been studied for some time; in these cycles the electrochemical step is achieved in an electrolytic cell but, as we have shown, it can also be achieved in a photoelectrochemical cell. The sulphuric acid solution that is obtained is concentrated. The acid is catalytically decomposed into water, oxygen and SO2 at high temperatures ( 8 0 0 - 1 0 0 0 C), and the SO2 is recycled [76]. Obviously, this type of process (the so-called Westinghouse process) is of

372

practical interest only when it is connected to a nuclear power plant which can provide not only the electrical energy (although only a modest a m o u n t is needed since the electrochemical process occurs at low voltage) but also the thermal energy needed for inexpensive regeneration of the SO2. Because of the problems involved in obtaining hydrogen via water photoelectrolysis, other worthwhile reactions have been studied which could, in principle, be photoelectrochemically driven by illumination of semiconductor liquid-junction solar cells [77]. Rather than direct hydrogen photoproduction, it might be best to achieve direct photogeneration of a reducing agent that is capable of effecting water reduction or, in general, of producing hydrogen via a catalysed process at a reactor that is far from the solar device [78]. The advantage of these methods is that semiconductors can be used which have band gap energies that are well matched to the solar spectrum, and correspondingly high solar energy conversion efficiencies can thus be obtained. The drawback is that the chemical process may be quite complicated.

4.2. 4. Semiconductor powder suspensions


Particularly interesting reactions, of the type that occur in photoelectrosynthetic cells, have been obtained by utilizing powdered polycrystalline semiconductor materials suspended in water, which have been inappropriately called " p o w d e r photocatalysts". Nozik [79] was the first to use the term "photochemical diode" to indicate this particular configuration of photoelectrosynthetic cells. (It would perhaps be more precise to call it a "photoelectrochemical diode".) In theory, these are cells which have collapsed into monolithic particles containing no external wires. Simple immersion of these particles in an appropriate electrolyte followed by illumination can produce particular chemical reactions in the electrolyte. In fact, each quantum of light that is absorbed by the surface of the powder grain produces a hole-electron pair that leads to both anodic and cathodic processes on the same surface but in different regions, so that the overall reaction is electrically balanced. Nozik distinguishes between two types of photochemical diodes on the basis of materials that function as anodic and cathodic regions respectively; these two types are Schottky-type diodes, comparable with a metalsemiconductor junction (in this case the powder grain is covered with metallic islands), and p - n - t y p e diodes respectively. Powder photocatalysts present numerous advantages over the electrodes of photoelectrochemical cells. In fact, the instability of the electrodes becomes a less serious problem, and materials can be used which have very active surfaces and which can be easily prepared. The quantum and energy outputs, however, are lower (at least at present) because of the lower efficiency in the hole-electron separation, the lower electronic conductivity of the powders and the less efficient charge transfer. Numerous reactions have been studied; some of the most interesting are as follows: (1) the photoelectrolysis of water, in the liquid and vapour

373 state, on powdered semiconductors made of TiO2 [80 - 83] and metal-loaded 8rTiO3 [84]; (2) the p h o t o p r o d u c t i o n of halogens (chlorine, bromine and iodine) in oxygen-saturated aqueous solutions containing the respective halide ions and platinized n-TiO2 powder (the production rate of the halogen is greatly enhanced by using a platinized n-TiO2 powder instead of a pure n-TiO2 powder catalyst and by lowering the pH of the solution [85] ); (3) the photo-assisted production of ammonia [80, 86] and/or hydrazine [80] from nitrogen and water on various TiO2-Fe203 catalysts supported on A1203 [87] (this process is particularly interesting because of the proposed use of ammonia as a multipurpose energy vector; ammonia could also serve as a large-scale agricultural input and as a feedstock for local processing industries [88]); {4) the photo-assisted reduction of aqueous CO 2 into formic acid, formaldehyde, methanol and methane, and the simultaneous development of oxygen on various powder semiconductors, including TiO2, WO3 and CdS [89] ; (5) the photoreduction of CO2 into methane with SrTiO3-Pt catalysts [81]; (6) the photoreaction of carbon with water and the production of CO (or CO2) and hydrogen with TiO2-RuO2 catalysts [83]; (7) the same reaction with lignite to produce CO2 and hydrogen [90]. On the basis of these results, slurry electrodes based on semiconductor powder suspensions of TiO2, WO3 and CdS have recently been prepared. They behave in the same way as the single-crystal and polycrystalline semic o n d u c t o r materials at present being investigated and possess all the advantages of powder photocatalysts [91].

4.3. Organic semiconductor cells Numerous organic semiconductors, particularly many dyes, are characterized by a high sensitivity to visible light [92, 93] and they often show a good stability in electrolytic solutions [94]. Thus, it has long seemed natural to extend photoelectrochemical research to these c o m p o u n d s as well. These materials, usually supported on SnO2, gold and platinum [95], exhibit the same photoelectrochemical behaviour as inorganic semiconductors. Thus on n-type semiconductors (e.g. crystal violet), after illumination, oxidation processes occur and the photocurrent is anodic whereas, on p-type semiconductors (e.g. phthalocyanines), reduction processes take place and the photocurrent is cathodic. In order for these systems to function, it is imperative that either redox couples be present in the electrolyte or compounds which function as electron donors or acceptors be present, i.e. compounds which are capable of self-oxidation and self-reduction respectively. For example, ascorbic acid acts as an electron donor for n-type dyes, whereas benzoquinone acts as an electron acceptor for p-type dyes. The usually very modest photoelectrochemical effect, especially for the photocurrent, increases with the doping of the photoconducting layer and becomes more stable and reproducible [93]. A decrease in the thickness of the supported semiconducting film also involves an increase in the photocurrent.

374 Many workers have dealt with this field of research [96], particularly Meier et al. [97] who recently were also able to achieve the photodecomposition of water by applying a bias voltage of 0.5 - 0.7 V. Although the efficiency of photoelectrochemical cells with an organic semiconductor is at present inferior to that obtained with inorganic electrodes, higher values are predicted for the future, once the bending of the bands can be better controlled. Furthermore, the remarkable variety of available compounds provides this field with very promising potential. Photoelectrochemical systems which involve solid-liquid interfacial layers of chlorophylls have also recently been taken into consideration. Studies are being conducted of monolayers, mixed monolayers and multilayers of chlorophyll deposited onto semiconducting SnO2 and/or metal electrodes. The highest efficiency is obtained with a homogeneously mixed monolayer of chlorophyll-lecithin on SnO2 [98]. A sandwich-type photocell using two chlorophyll electrodes has also been proposed. In addition, chlorophyll electrodes are being used for photoenergy conversion modelled on photosynthesis [99].

5. Other systems A short account will now be given of systems which, strictly speaking, do not belong to the class of photoelectrochemical cells and are outside the aims of this review, but which are nevertheless significant.

5.1. Photogalvanic cells Photogalvanic cells deserve a separate discussion. Although some workers [79, 97] consider them to belong to the class of photoelectrochemical cells, these devices are fundamentally different. In photogalvanic cells, optical energy is converted into electrical energy. However, the photoreaction occurs homogeneously in a solution containing the appropriate molecules, often dyes, which are capable of passing from their equilibrium energy level to an excited energy level, thus becoming particularly active for charge transfer reactions to inert electrodes (e.g. platinum). Therefore the illuminated electrode is not a semiconductor which absorbs a quantum of light and produces a hole-electron pair, but rather it is a metallic electrode in contact with a light-absorbing solution containing a redox couple, usually inorganic [ 9 4 , 1 0 0 ] . As a result, the space charge region and the important function that it performs are completely lacking. According to some theoretical estimates of the m a x i m u m power obtainable from photogalvanic cells [101], a m a x i m u m power of 70 pW cm -2 has been calculated on the assumption of a quantum efficiency of unity for the photochemical process. After this value has been compared with that for semiconductor-electrolyte systems already experimentally achieved (greater than 10 mW cm-2), it must be concluded that photogalvanic systems are really not of practical interest for the direct conversion of solar energy.

375 Nevertheless research is also very active in this field, and some recently proposed alternative systems seem to offer interesting results, e.g. photogalvanic cells capable of photodecomposing water and producing hydrogen [102] and hybrid systems composed of a photogalvanic half-cell and a photovoltaic half-cell, combining the responses of both these cells [ 103 ].
5.2. P h o t o c h e m i c a l s y s t e m s These systems are only of interest in connection with the production of chemical fuels. In particular, the photolysis of water into hydrogen and oxygen has long been the dream of many photochemists. It has recently been verified that this objective is attainable in laboratory experiments by utilizing the appropriate photosensitizers and catalysts, i.e. colloidal platinum to support the hydrogen evolution reaction and ruthenium and/or titanium oxides to support the oxygen evolution reaction. There is, however, a severe problem which limits the efficiency of photochemical processes in homogeneous solutions, i.e. the back reactions in the dark. Therefore research in this field, although intense, must still be considered to be at an initial stage [104].

6. Conclusions In this review we described the various types of photoelectrochemical cells with semiconductor electrodes, and we classified them according to the value of the free-enthalpy variation of the total reaction that takes place in the galvanic cell. We then concentrated on the basic principles by which these cells function. Finally, we examined in greater detail the so-called regenerative photoelectrochemical and the photoelectrosynthetic cells. The first type allows the transformation of solar energy into electrical energy; the second type allows the transformation of solar energy into chemical energy, i.e. into chemical fuels and in particular hydrogen. We have shown that there is extensive research being conducted in this field. For the most part the present studies are fundamental in character but oriented towards practical applications for the medium- to long-term future and are thus attracting the attention of numerous industrial groups. As a result, the number of patents being issued is constantly increasing; these relate not only to the electrode materials but also to the entire cell, since the prospects for their design and operation are continuously being reanalysed. Research in photoelectrochemistry has led to an improved understanding of the semiconductor-liquid interface, and many new materials have been investigated. It has also provided new insight into a variety of processes {electrochemical, catalytic, photolytic and photographic). The results obtained are very interesting from a scientific point of view; however, they still cannot be considered as conclusive from the point of view of practical applications.

376 Numerous and complex problems must still be faced in order to obtain practical devices. However, it must be kept in mind that, unlike systems with a solid-solid junction, those with a solid-liquid junction (which are thought to constitute a possible alternative) have only been proposed as recently as 1975 by Gerischer [2]. This has been due to our limited knowledge of the electrochemical behaviour of semiconductors; this knowledge, at a theoretical level, had to be integrated by extending the band theory to electrolytes containing a redox couple. As we have shown, even if we consider only the semiconducting electrode, there are considerable difficulties in obtaining materials which are stable to corrosive processes and whose performance offers interesting applications. Although it can be argued that acceptable solar-electrical conversion efficiencies have been reached with regenerative solar cells, particularly with cells based on single-crystal electrodes of selenides and tellurides, the lifetimes required for these devices (10 - 20 years) have yet to be achieved. From this point of view, however, different types of treatment have recently been proposed for electrode surfaces; these treatments are capable of passivating the active centres for recombination of the electron-hole pairs and of improving both the performance and the stability characteristics of the surface. If we want to produce practical applications for these cells, we must also obtain functional polycrystalline and thin film electrodes. The problem has already been successfully dealt with for some materials belonging to the first class, although their performance is still inferior to that of monocrystalline materials. The problem has yet to be resolved, however, for the dichalcogenides of the transition metals. Greater difficulties are encountered with photoelectrosynthetic cells, and in particular photoelectrolysis cells, which have received by far the most attention and are thus the most advanced and promising in their results. Also in this field the lack of stable electrode materials which are capable of matching the solar spectrum and of satisfying all energetic conditions constitutes a strong limitation. Therefore some workers [105] have already questioned whether it is really worthwhile to search for materials that can be used for large-area photoelectrolysis cells since the chemical hydrogen fuel has to be collected and transferred into some kind of storage medium. From this point of view, it seems more feasible to combine a large-area regenerative photoelectrochemical cell, which only produces electrical power, with a small electrolysis cell, where special electrodes with small hydrogen and oxygen overvoltages are used. Although these conclusions may still be open to discussion, there is no question that the devices at present available for the photoelectrolysis of water involve numerous problems which must be resolved if such devices are to offer prospects for the future. As for other photoelectrochemical processes, the most advanced relate to the p h o t o p r o d u c t i o n of halogens. Other processes, particularly those

377 involving hybrid cycles, are still unsatisfactory. Powder photocatalysts are interesting for both the versatility of the possible processes and the materials, but their efficiency must be improved before any practical application can possibly be foreseen for them. While the goal of an efficient and stable system for the direct solar production of fuels remains elusive, semiconductor-based systems still remain the most efficient chemical systems described so far for such reactions. In fact, photogalvanic and photochemical systems are not competitive, owing to their lower power output, and this situation seems destined to remain unchanged for several years. Finally, it may be worthwhile to consider the problem of alternative ways of producing hydrogen. There is no question that in the fossil fuel era the most economical m e t h o d of producing hydrogen is via chemical processes, i.e. steam reforming of methane or other hydrocarbons, or coal gasification. After the depletion of fossil fuels, the demand for hydrogen will increase, particularly as a readily transportable fuel. Of fundamental importance at that time will be the various methods of producing hydrogen by using energy from renewable resources. According to the most recent analyses, the electrochemical m e t h o d will be the most advantageous. In particular, electrolysis is considered to be the only advanced technology for the splitting of water through the use of either solar or nuclear energy. Thermochemical, thermoelectrochemical, thermophotoelectrochemical, photochemical and biochemical methods, however, are interesting from a scientific point of view, but they cannot compete with water electrolysis [106]. As for the utilization of solar energy through photoelectrochemical devices, it is once again suggested that a combined system consisting of a regenerative photocell and an electrolysis cell would be more competitive than a photoelectrolysis cell for direct water decomposition. As a conclusion to this review, we must consider that there is still extensive work to be carried out on the physical properties of the materials, their solid state chemistry and their photoelectrochemistry. In contrast, the ease of formation of junctions in photoelectrochemical solar cells permits the use of semiconductor preparations that would be unsuitable for homojunction or heterojunction devices. Thus, simple and inexpensive semiconductor preparation techniques m a y be possible in the near future, and this may more than outweigh the lower efficiencies suggested by theoretical analyses. Moreover, the possibility of adding storage within the regenerative cells represents an obvious advantage over more conventional solar cells. Although present results are still unsatisfactory, they do show that, with more efficient photoelectrodes and better storage elements, these devices could be very effective. Their versatility and modular nature are certainly worth further investigation.
References

1 M. K. Hubbard, Sci. A m . , 224 (1971) 61. 2 H. Gerischer, J. Electroanal. Chem. Interfacial Electrochem., 58 (1975) 263.

378 3 J. Manasseh, G. Hodes and D. Cahen, Nature (London), 261 (1976) 403; J. Electrochem. Soc., 124 (1977) 532. 4 P. G. P. Ang and A. F. Sammells, Meet. o f the Electrochemical Society, Hollywood, FL, October 5- 10, 1980, Abstract 156; Faraday Discuss. Chem. Soc., 70 (1981) 207. 5 P. G. P. Ang and A. F. Sammells, J. Electrochem. Soc., 129 (1982) 233. 6 P. G. P. Ang, C. J. Liu, A. A. Rossignuolo, A. J. Tiller and A. F. Sammells, Meet. o f the Electrochemical Society, Denver, CO, October 11 - 16, 1981, Abstract 511. 7 A. Fujishima and K. Honda, Bull. Chem. Soc. Jpn., 44 (1971) 1148; Nature (London), 238 (1972) 37. 8 H. Gerischer, J. Electrochem. Soc., 113 (1966) 1174. H. Tributsch and H. Gerischer, Ber. Bunsenges. Phys. Chem., 73 (1969) 850. H. Gerischer, Surf. Sci., 13 (1969) 265. 9 L. Peraldo Bicelli and B. Scrosati, Chim. Ind. (Milan), 63 (1981) 172. 10 L. Peraldo Bicelli and G. Razzini, Surf. Technol., 16 (1982) 37. 11 S. N. Frank and A. J. Bard, J. Am. Chem. Soc., 97 (1975) 7427. M. Miyake, H. Yoneyama and H. Tamura, Electrochim. Acta, 22 (1977) 319. P. A. Kohl and A. J. Bard, J. A m . Chem. Soc., 99 (1977) 7531. K. Nakatani, S. Matsudaira and H. Tsubomura, J. Electrochem. Soc., 125 (1978) 406. P. A. Kohl and A. J. Bard, J. Electrochem. Soc., 126 (1979) 603. L. F. Schneemeyer, M. S. Wrighton, A. Stacy and M. J. Sienko, Appl. Phys. Lett., 36 (1980) 701. M. E. Langmuir, P. Hoenig and R. D. Rauh, J. Electrochem. Soc., 128 (1981) 2357. R. Noufi, D. Tench and L. F. Warren, J. Electrochem. Soc., 128 (1981) 2363. G. Nagasubramanian and A. J. Bard, J. Electrochem. Soc., 128 (1981) 1055. F. Di Quarto and A. J. Bard, J. Electroanal. Chem. Interfacial Electrochem., 127 (1981) 43. 12 P. Singh, K. Rajeshwar, J. Du Bow and R. Job, J. A m . Chem. Soc., 102 (1980) 4676. A. L. Edgecombe, J. Phillips and H. F. Gibbard, Meet. o f the Electrochemical Society, Hollywood, FL, October 5- 10, 1980, Abstract 294. P. Singh, R. Singh, K. Rajeshwar and J. Du Bow, J. Electrochem. Soc., 128 (1981) 1145. 13 T. Skotheim, Appl. Phys. Lett., 38 (1981) 712;J. Electrochem. Soc., 129 (1982) 894. 14 R. Noufi, D. Tench and L. F. Warren, J. Electrochem. Soc., 127 (1980) 2310; 128 (1981) 2596. 15 D. Tench and L. F. Warren, Meet. o f the Electrochemical Society, Montreal, May 9 14, 1982, Abstract 337. 16 A. B. Ellis, S. W. Kaiser and M. S. Wrighton, J. A m . Chem. Soc., 98 (1976) 1635. B. Miller and A. Heller, Nature (London), 262 (1976) 680. 17 K. W. Frese, Jr., M. J. Madou and S. R. Morrison, J. Electrochem. Soc., 128 (1981) 1939. 18 J. Reichman and M. A. Russak, J. Appl. Phys., 53 (1982) 708. 19 A. Kirsch-De Mesmacker, P. Josseaux and J. Nasielski, Meet. o f the Electrochemical Society, Denver, CO, October 11 -16, 1981, Abstract 570. 20 H. Tributsch, Z. Naturforsch., 32a (1977) 972. H. Tributsch and J. C. Bennet, J. Electroanal. Chem. Interfacial Electrochem., 81 (1977) 97. H. Tributsch, Ber. Bunsenges. Phys. Chem., 82 (1978) 169. H. Tributsch, J. Electrochem. Soc., 125 (1978) 1086. J. Gobrecht, H. Tributsch and H. Gerischer, J. Electrochem. Soc., 125 (1978) 2085. 21 B. A. Parkinson, T. E. Furtak, D. Canfield, K. Keung-Kam and G. Kline, Faraday Discuss. Chem. Soc., 70 (1981) 234. D. Canfield and B. A. Parkinson, J. Am. Chem. Soc., 103 (1981) 1279. 22 G. Razzini, L. Peraldo Bicelli, G. Pini and B. Scrosati, J. Electrochem. Soc., 128 (1981) 2134.

379 23 W. Kautek, J. Gobrecht and H. Gerischer, Ber. Bunsenges. Phys. Chem., 84 (1980) 1034. 24 H. J. Lewerenz, S. D. Francis, C. J. Doherty and H. J. Leamy, J. Electrochem. Soc., 129 {1982) 418. 25 H. S. White, H. D. Abruna and A. J. Bard, J. Electrochem. Soc., 129 (1982) 265. 26 G. Razzini, J. Power Sources, 7 (1982) 275. 27 S. Menezes, L. F. Schneemeyer and H. J. Lewerenz, Appl. Phys. Lett., 38 (1981) 949. 28 S. Chandra and R. K. Pandey, Phys. Status Solidi A, 59 {1980) 787. D. J. Miller and D. Haneman, Sol. Energy Mater., 4 (1981) 223. 29 D. R. Pratt, M. E. Langmuir, R. A. Boudreau and R. D. Rauh, J. Electrochem. Soc., 128 (1981) 1627. R. N. Bhattacharya and P. Pramanik, J. Electrochem. Soc., 129 (1982) 332. 30 M. A. Russak, J. Reichman, J. De Carlo and C. Creter, Rep. SERI-TR-8OO2-8-T1, 1980 (Solar Energy Research Institute); Energy Res. Abstracts, 6 (1981), no. 15465. 31 M. Robbins, Adv. Chem. Ser., 186 (1980) 151. 32 G. Hodes, D. Cahen, J. Manassen and M. David, J. Electrochem. Soc., 127 (1980) 2252. 33 C. J. Liu and J. H. Wang, J. Electrochem. Soc., 129 (1982) 719. 34 K . T . L . De Silva, D. J. Miller and D. Haneman, Sol. Energy Mater., 4 {1981) 233. 35 R. Tenne, Appl. Phys., 25 (1981) 13. N. Mueller and R. Tenne, Appl. Phys. Lett., 39 (1981) 283. 36 C. H. Liu, J. Olsen, D. R. Saunders and J. H. Wang, J. Electrochem. Soc., 128 (1981) 1224. 37 G. Djemal, M. Miiller, U. Lachish and D. Cahen, Sol. Energy Mater., 5 (1981) 403. 38 D. S. Ginley, R. M. Biefe|d, B. A. Parkinson and K. Keung-Kam, J. Electrochem. Soc., 129 (1982) 145. 39 H. Gerischer, Adv. Electrochem. Electrochem. Eng., 1 (1961) 197. 40 H. Yoneyama, H. Sakamoto and H. Tamura, Electrochim. Acta, 20 (1975) 341. A. J. Nozik, Appl. Phys. Lett., 29 (1976) 150. 41 A. J. Nozik, Annu. Rev. Phys. Chem., 29 (1978) 189. M. S. Wrighton, Acc. Chem. Res., 12 (1979) 303. M. Tomkiewicz and H. Fay, Appl. Phys., 18 (1979) 1. 42 A. J. Bard and N. W. Wrighton, J. Electrochem. Soc., 124 (1977) 1706. 43 H. Tsubomura, M. Matsumura, K. Nakatani and Y. Nomura, in A. Heller (ed.), Proc. Conf. on Semiconductor Liquid Junction Solar Cells, Vol. 77-3, Electrochemical Society, Princeton, NJ, 1977, p. 178. R. Memming, J. Electrochem. Soc., 125 (1978) 117. 44 M. P. Dare-Edwards, J. B. Goodenough, A. Hamnett, K. R. Seddon and R. D. Wright, Faraday Discuss. Chem. Soc., 70 (1981) 285. 45 D.E. Scaif~, Sol. Energy, 25 (1980) 41. 46 A. Wold and K. Dwight, Adv. Chem. Set., 186 (1980} 161. 47 A. Fujishima, K. Kohayakawa and K. Honda, Bull. Chem. Soc. Jpn., 48 (1975) 1041. 48 M. A. Butler, R. D. Nashby and R. K. Quin, Solid State Commun., 19 (1976) 1011. 49 D. Haneman, Surf. Sci., 86 (1979) 462. M. Ulmann and J. Augustynski, Meet. o f the Electrochemical Society, Denver, CO, October 11 - 16, 1981, Abstract 573. 50 A. K. Ghosh and H. P. Maruska, J. Electrochem. Soc., 124 (1977) 1516. 51 A. Fujishima, E. Hayashitani and K. Honda, Seisan Kenkyu, 23 (1971) 363; Chem. Abstracts, 76 (1972), no. 79315g. 52 Y. Matsumoto, J. Kurimoto, T. Shimizu and E. Sato, J. Electrochem. Soc., 128 (1981) 1040. V. M. Arutyunyan, A. G. Sarkisyan, Zh. R. Panosyan, V. M. Arakelyan, A. O. Arakelyan and G. E. Shakhnazaryan, Elektrokhimiya, 1 7 (1981) 1471. 53 A. Fujishima, K. Kohayakawa and K. Honda, J. Electrochem. Soc., 122 (1975) 1487. K. J. Hartig, J. Lichtscheidl and N. Getoff, Z. Naturforsch., 36a (1981) 51.

380 N. Giordano, V. Antonucci, S. Cavallaro, R. Lembo and J. C. J. Bart, Adv. Hydrogen Energy, 2 (1981) 621. Ya. L. Kogan and A. M. Vakulenko, Sol. Energy Mater., 3 (1980) 357. A. A. Soliman and H. J. J. Seguin, Sol. Energy Mater., 5 ( 1 9 8 1 ) 9 5 ; Can. J. Phys., 59 (1981) 1674. F. T. Liou, C. Y. Yang and S. N. Levine, J. Electrochem. Soc., 129 (1982) 342. H. Yoneyama, H. Sasaki and H. Tamura, Denki Kagaku Oyobi Kogyo Butsuri Kagaku, 49 (1981) 222; Chem. Abstracts, 95 (1981), no. 15044c. T. E. Phillips, K. Moorjani, J. C. Murphy and T. O. Poehler, J. Electrochem. Soc., 129 (1982) 1210. F. Di Quarto, A. Di Paola and C. Sunseri, Electrochim. Acta, 26 (1981) 1177. W. Gissler and R. Memming, J. Electrochem. Soc., 124 (1977) 1710. K. L. Hardec and A. J. Bard, J. Electrochem. Soc., 123 (1976) 1024; 124 (1977) 215. J. H. Kennedy and K. W. Frese, J. Electrochem. Soc., 125 (1978) 709; 125 (1978) 723. P. Iwansky, J. S. Curran, W. Gissler and R. Memming, Programme Prog. Rep., 1981 (Commission of the European Communities, Joint Research Centre, Ispra Establishment) (category 1.9, number 3907). F. Moilers and R. Memming, Bet. Bunsenges. Phys. Chem., 76 (1972) 469. M. S. Wrighton, A. B. Ellis, P. T. Wolczanski, D. L. Morse and H. B. Abrahamson, J. Am . Chem. Soc., 98 (1976) 44. M. S. Wrighton, A. B. Ellis, P. T. Wolczanski, D. L. Morse and D. S. Ginby, J. Am . Chem. Soc., 98 (1976) 2774. R. K. Quinn, R. D. Nashby and R. J. Baughman, Mater. Res. Bull., 11 (1976) 1011. A. M. Redon, J. Vigneron, R. Heindl, C. Sella, C. Martin and J. P. Dalbera, Sol. Cells, 3 (1981) 179. M. V. C. Sastri and G. Nagasubramanian, Adv. Hydrogen Energy, 2 (1981) 597. J. H. Kennedy, M. Anderman and R. Schinar, J. Electrochem. Soc., 128 (1981) 2371. V. Guruswamy, P. Keillor, G. L. Campbell and J. O'M. Bockris, Sol. Energy Mater., 4

54 55 56 57 58 59 60 61

62 63 64 65 66 67 68

(1980) 11.
69 70 71 72 73 74 75 76 V. Guruswamy, O. J. Murphy, V. Young, G. Hildrech and J. O'M. Bockris, Sol. Energy Mater., 6 (1981) 59. M. Noda, Adv. Hydrogen Energy, 2 (1981) 609. R. C. Hughes, D. S. Ginley and A. K. Hays, Meet. o f the Electrochemical Society, Denver, CO, October 11 -16, 1981, Abstract 558. H. S. Jarrett, A. W. Sleight, A. H, Kung and J. L. Gillson, Surf. Sci., 101 (1980) 205. H. Mettec, J. W. Otvos and M. Calvin, Sol. Energy Mater., 4 (1981) 443. A. Heller, R. G. Vadimsky, Phys. Rev. Lett., 46 (1981) 1153. A. Heller, W. A. Bonnet and B. Miller, Meet. o f the Electrochemical Society, Denver, CO, October 11 - 16, 1981, Abstract 556. F. F. Fan, H. S. White, B. L. Wheeler and A. J. Bard, J. Am. Chem. Soc., 102 (1980) 5142. G. S. Calabrese and M. S. Wrighton, J. A m . Chem. Soc., 103 (1981) 6273. R. J. Remick, A. F. Sammells and S. E. Foh, Meet. o f the Electrochemical Society, Hollywood, FL, October 5 - 10, 1980, Abstract 590. M. Butti, S. Fattori and D. O. Rodriguez, Thesis, Politecnico di Milano, 1981. A. J. Bard, Science, 207 (1980) 139. N. S. Lewis, M. G. Bradley, A. B. Bocarsly and M. S. Wrighton, J. Am. Chem. Soc., 101 (1979) 7721. F. F. Fan, B. Reichman and A. J. Bard, J. A m . Chem. Soc., 102 (1980) 1488. A. B. Bocarsly, D. C. Bookbinder, R. N. Dominey, N. S. Lewis and M. S. Wrighton, J. Am . Chem. Soc., 102 (1980) 3683. A. J. Nozik, Appl. Phys. Lett., 30 (1977) 567. N. Schrauzer and T. D. Guth, J. Am. Chem. Soc., 99 (1977) 7189. J.C. Hemminger, R. Cart and G. A. Somorjai, Chem. Phys. Lett., 57 (1978) 100.

77 78

79 80 81

381 82 V. F. Kiselev, A. S. Petrov and R. V. Prudnikov, Kinet. Katal., 19 (1978) 1593. S. Sato and J. M. White, Rep. A082768, 1980 (Texas University, Austin, TX). T. Sakata, T. Kawai, T. Koiso and M. Okuyama, Adv. Hydrogen Energy, 2 (1981) 773. F. T. Wagner and G. A. Somorjai, J. Am. Chem. Soc., to be published. 83 T. Kawai and T. Sakata, J. Chem. Soc., Chem. Commun., (1979) 1047. T. Kawai and T. Sakata, 7th Int. Congr. on Catalysis, Tokyo, Japan, June 1980, Preprint B38. 84 J. M. Lehn, J. P. Sauvage and R. Ziessel, Nouv. J. Chim., 4 (1980) 623. 85 B. Reichman and C. E. Byvik, J. Phys. Chem., 85 (1981) 2255. 86 H. Miyama, N. Fujii and Y. Nagae, Chem. Phys. Lett., 74 (1980) 523. 87 V. Augugliaro, A. Lauricella, L. Rizzuti, M. Schiavello and A. Sclafani, Adv. Hydrogen Energy, 2 (1981) 589. 88 L. Green, Jr., Adv. Hydrogen Energy, 2 (1981) 1265. 89 T. Inoue, A. Fujishima, S. Konishi and K. Honda, Nature (London), 277 (1979) 636. 90 S. Sato and J. M. White, Ind. Eng. Chem., Prod. Res. Dev., 19 (1980) 542. 91 W. W. Dunn, Y. Aikawa and A. J. Bard, J. Electrochem. Soc., 128 (1981) 222. 92 H. Meier, Die Photochemie der Organischen Farbstoffe, Springer, Berlin, 1963, p. 127. 93 H. Meier, Organic Semiconductors -- Dark- and Photoconductivity in Organic Solids, Verlag Chemie, Weinheim, 1974, p. 307; Top. Curr. Chem., 61 (1976) 85. 94 H. Meier, W. Albrecht, U. Tschirwitz and E. Zimmerhackl, Bet. Bunsenges. Phys. Chem., 77 (1973) 843. H. Meier, U. Tschirwitz, E. Zimmerhackl, W. Albrecht and G. Zeitler, J. Phys. Chem., 81 (1977) 712. 95 J. O'M. Bockris and L. Handley, Energy Convers., 18 (1978) 1. 96 C. H. Langford, B. R. Hollebone and D. Nadezhdin, Can. J. Chem., 59 (1981) 652. 97 H. Meier, W. Albrecht, U. Tschirwitz, N. Geheeb and E. Zimmerhackl, Chem.-Ing.Tech., 51 (1979)653. 98 T. Miyasaka and K. Honda, Surf. Sci., 101 ( 1 9 8 0 ) 5 4 1 ; A m . Chem. Soc. Syrup. Set., 146 (1981) 231. 99 F. Takahashi and T. Komori, Bull. Chem. Soc. Jpn., 54 (1981) 1305. M. Aizawa, J. Yoshitake and S. Suzuki, Mol. Cryst. Liq. Cryst., 70 (1981) 1407. 100 H. Meier, in K. Venkataraman (ed.), The Chemistry o f Synthetic Dyes, Vol. IV, Academic Press, New York, 1972, p. 475. W. D. Clark and J. A. Eckert, Sol. Energy, 17 (1975) 147. R. Gomer, Electrochim. Acta, 20 (19~ 5) 13. K. Shigehara and E. Tsuchida, J. Phys. Chem., 81 (1977) 1883. 101 W. J. Albery and M. D. Archer, J. Electroanal. Chem. Interfacial Electrochem., 86 (1978) 1. 102 T. Yamase, Photochem. Photobiol., 34 (1981) 111. 103 J. I. Tsipouridis and B. M. Gibbs, Proc. 3rd Int. Conf on Future Energy Concepts, in Inst. Electr. Eng. Conf. Publ. 192 (1981) 87. 104 B. V. Koryakin, T. S. Dzhabiev and A. E. Shilov, Dokl. Akad. Nauk S.S.S.R., 233 (1977) 359. K. Kalyanasundaram and M. Gr~'tzel, Angew. Chem., Int. Edn. Engl., 18 (1979) 701. N. Sutin and C. Creutz, Pure Appl. Chem., 52 (1980) 2717. E. Pelizzetti, M. Visca, E. Borgarello, E. Pramauro and A. Palmas, Chim. Ind. (Milan), 63 (1981) 805. 105 R. Memming, Electrochim. Acta, 25 (1980) 77. 106 S. Srinivasan, Meet. o f the Electrochemical Society, Montreal, May 9 - 14, 1982, Abstract 442.

You might also like