Artificial Photosynthesis: Solar Splitting of Water to ... - Allen J. Bard

Acc. Chem. Res. 1995,28, 141-145 141Artificial Photosynthesis: Solar Splitting of Water toHydrogen and OxygenALLEN J. BARD* AND MARYE ANNE Fox*Department of Chemistry and Biochemistry, University of Texas at Austin, Austin, Texas 78712Received November 16, 1994Water SplittingThe maintenance of life on earth, our food, oxygen,and fossil fuels depend upon the conversion of solarenergy into chemical energy by biological photosynthesiscarried out by green plants and photosyntheticbacteria. In this process sunlight and available abundantraw materials (water, carbon dioxide) are convertedto oxygen and the reduced organic species thatserve as food and fuel. A long-standing challenge hasbeen the development of a practical artificial photosyntheticsystem that can roughly mimic the biologicalone, not by duplicating the self-organization andreproduction of the biological system nor the aestheticbeauty of trees and plants, but rather by being ableto use sunlight to drive a thermodynamically uphillreaction of an abundant materials to produce a fuel.In this Account we focus on “water splitting”, thephotodriven conversion of liquid water to gaseoushydrogen and oxygen:Beyond the intellectual challenge of designing andfabricating such a system, there are several practicalimplications. H2 could serve directly as a fuel, e.g.,for transportation or for the production of electricityin fuel cells, without producing pollutants or greenhousegases upon combustion. For some purposes,however, it might be useful to use the HZ as a reactantto produce a different fuel, such as one that is liquidat the usual temperatures and pressures. Thus, weseek as a “Holy Grail” a renewable energy sourcedriven by solar energy that produces a clean andstorable fuel.Let us define this Holy Grail more specifically. Wewant an efficient and long-lived system for splittingwater to Ha and 0 2 with light in the terrestrial(AM1.5) solar spectrum at an intensity of one sun. Fora practical system, an energy efficiency of at least 10%appears to be necessary. This means that the HZ and0 2 produced in the system have a fuel value of at least10% of the solar energy incident on the system. Inthe southern United States, the instantaneous maximumintensity is of the order of 1 kW/m2 and theaverage 24-h intensity throughout a year is about 250W/m2. Thus, the system should produce H2 at a rateof about 0.7 g/s or 7.8 L(STP)/s per m2 of collector atmaximum solar intensity. Long-lived implies that theAllen J. Bard holds the Hackerman-Welch Regents Chair in Chemistry at TheUniversity of Texas and works on the application of electrochemical methods tochemical problems.Marye Anne Fox currently occupies the M. June and J. Virgil Waggoner RegentsChair in Chemistry at the University of Texas and is director of the Center for FastKinetics Research. Her principal research interests are in physical organic chemistry0001-4842/95/0128-0141$09.00/0sensitizers and catalysts, as well as the materials ofconstruction, will not be consumed or degraded underirradiation for at least 10 years. The solar spectrumat sea level extends from the near infrared throughthe visible to the near ultraviolet with photon energiesup to 3.0 eV. This region is not absorbed by wateritself, so photochemical reactions are only possible inthe presence of some recyclable absorbing sensitizer.Finally, for practical applications, the cost of H2produced by the system (on an energy equivalentbasis) should be competitive with that of fossil fuels.Although we have defined our Holy Grail in termsof the water-splitting reaction, other chemical solarenergy conversions are possible and have been investigated.For example, there are semiconductor liquidjunction systems that, when irradiated with visiblelight, carry out the reactions 2HBr - H2 + Br2 and2HI - HZ + 12. Indeed, the “brine splitting” or“photochloralkali“ reaction,2H,O + 2C1- hv 20H- + C1, + H, (2)would probably be more useful than water splitting,but has so far not been achieved without applying anadditional external potential.Many investigations in this field have involvedsacrificial donors, which are reduced materials thatare oxidized more easily than water, for example,ethylenediamine tetraacetic acid or triethanolamine.The use of such compounds usually greatly improvesthe efficiency of the solar process, but clearly is not ofinterest in practical systems, especially if the sacrificialdonors are more expensive than the H2 produced.It might be possible to use reduced waste materialsin this role, but is it unlikely that this approach willbe practical in large-scale fuel production.Photoelectrochemical approaches may be useful,however, as a means of water treatment, destroyingorganic wastes and removing metals. There are alsoa number of chemical schemes for converting solarenergy to electrical energy, e.g., in liquid junctionphotovoltaic cells. Indeed, devices with single-crystalsemiconductors have been constructed which showsolar efficiencies of above 10%. It remains to be seenwhether such chemical photovoltaic systems will bepractically competitive with solid state ones based, forexample, on single-crystal or amorphous Si. Althoughmany of these alternative chemical solar energysystems are interesting, we focus here on the watersplittingreaction, because it effectively represents thescientific challenges typical of all such efforts.History and ProgressA. Efficiency. The free energy change for reaction1 is AGO = 237.2 kJ/mol or 2.46 eV/molecule of H2O.0 1995 American Chemical Society

142 Acc. Chem. Res., Vol. 28, No. 3, 1995Since two electrons are involved in the reaction aswritten (n = 2), this corresponds to 1.23 eV/e, whichis also the standard emf for the reaction. The photonsin the solar spectrum provide sufficient energy to drivethis reaction, but the efficiency of the reaction dependsupon how the reaction is carried out. It is possible tocause water splitting thermally with light via concentratorsand a solar furnace by heating water to 1500-2500 K.I However, the efficiency of this process istypically below 2%, and the cost of the capital equipmentand material stability problems suggest that thisapproach to solar water splitting is not a promisingone.Since water itself does not absorb appreciable radiationwithin the solar spectrum, one or more lightabsorbingspecies (photoconverters or sensitizers)must be used to transduce the radiant energy tochemical (or electrical) energy in the form of electronlhole pairs, i.e., to the oxidizing and reducing potentialneeded to drive the reaction. The maximum efficiencyfor photochemical solar converters has been consideredin a number of papers2 and depends upon the bandgap (or threshold energy), E,, of the photoconverter.Radiation of energy below E, is not absorbed whilethat above E, is partly lost as heat by internalconversion or intraband thermalization processes.Additional thermodynamic losses occur because theexcited state concentration is only a fraction of thatof the ground state and because some excited statesare lost through radiative decay.2 When these factorsare taken into account, the threshold photon energyand the maximum efficiency can be calculated. For asingle photoconverter system, wavelengths below 770nm (or energies above 1.6 eV) are required to yield amaximum efficiency of about 30%. Lower photonenergies and higher efficiencies are attainable if oneemploys two photoconverters. Thus for a system withtwo photoconverters with two different, optimized E,values, one finds a maximum solar efficiency of 41%.2These calculations show that, in principle, the desiredefficiency for water splitting is attainable, even witha system involving a single photoconverter.B. Semiconductor Solid State Photovoltaic-Based Systems. A number of different approachesare possible with semiconductors as the photoconverter.The most direct, brute force, approach employsa solid state photovoltaic solar cell to generate electricitythat is then passed into a commercial-typewater electrolyzer (Figure 1A). The maximum theoreticalefficiency for a Si photovoltaic cell is 33%, andthe efficiencies of the best laboratory cells have beenreported to be about 24%. Commercial single-crystalSi solar cells generally have efficiencies in the 12-16% range. The electrolysis of water at a reasonablerate in a practical cell requires applied voltagessignificantly larger than the theoretical value (1.23 Vat 25 "C), and electrolysis energy efficiencies of about60% are typical. Thus, the efficiency of the combinedsolar/electrolyzer system using commercially availablecomponents is close to the desired 10% defined forsolar hydrogen generation. Moreover, the componentsare rugged and should be long-lived. The problemwith such a system is its cost. Solar photovoltaicscannot currently produce electricity at competitive(1) Etievant, C. Solar Energy Muter. 1991, 24, 413.(Z)Archer, M. D.; Bolton, J. R. J. Phys. Chem. 1990, 94, 8028 andreferences therein.te'b%-JPhotovoltaic hv, cell ,/ElectrolysisAn-tw semiconductornlpBBard and FoxII 4 2/ \semiconductor dye layerCDFigure 1. Schematic diagrams of different types of semiconductor-basedsystems proposed for solar water splitting: (A)solid state photovoltaic cell driving a water electrolyzer; (B) cellwith immersed semiconductor p/n junction (or metallsemiconductorSchottky junction) as one electrode; (C) liquid junctionsemiconductor electrode cell; (D) cell with dye-sensitized semiconductorelectrode.prices, and hydrogen from water electrolyzers issignificantly more expensive than that produced chemicallyfrom coal or natural gas. Lower cost solar cellsare possible, e.g., through the use of polycrystallineor amorphous Si or other semiconductors (CdS, CdTe,CuInSed, and some improvements in water electrolyzerefficiency through better configurations andcatalysts and the use of higher temperatures arepossible. However, it probably will be difficult to bringthe overall cost down to levels that make such aconfiguration practical in the foreseeable future.An alternative system involves the semiconductorphotovoltaic cell immersed directly in the aqueoussystem (Figure 1B). At the least this eliminates thecosts and mechanical difficulties associated with separateconstruction and interconnection of solar andelectrochemical cells. In one such system, the electrodesare composed of single or multiple semiconductorp/n junctions that are irradiated while they arewithin the cell. This simpler apparatus is attainedat the cost of encapsulating and coating the semiconductorsto protect them from the liquid environmentand probably with a more limited choice of electrocatalystfor 0 2 or H2 evolution. Moreover, the opencircuitphotovoltage of a single Si p/n junction is only0.55 V, so at least three of these in series would beneeded to generate the necessary potential for watersplitting. For example, in a system developed at TexasInstruments3 (TI) p-Si/n-Si junctions were producedon small (0.2 mm diameter) Si spheres embedded inglass and backed by a conductive layer to form an(3) Kilby, J. S.; Lathrop, J. W.; Porter, W. A. US. Patent 4 021 323,1977; U.S. Patent 4 100 051, 1978; U.S. Patent 4 136 436, 1979. Seealso: Johnson, E. L. In Electrochemistry in Industry; Landau, U., Yeager,E., Kortan, D., Eds.; Plenum: New York, 1982; pp 299-306.

Artificial Photosynthesisarray. Each sphere behaved as a photovoltaic cell andproduced about 0.55 V. The use of two arrays,protected with noble metal catalysts (M), i.e., Wp-Si/n-Si and M/n-Sup-Si, connected in series and incontact with HBr, allowed H2 and Bra to be generatedwith about an 8% efficiency. Multiple TI photoarraycells to carry out water splitting and other reactionsrequiring higher potentials are possible4 at a considerablesacrifice in efficiency. Note that, in addition top/n semiconductor junctions, those between a metaland semiconductor (Schottky barriers) can be used toproduce a photopotential, e.g., in electrodes such asAdn-Gap, PtSUn-Si, and Pun-GaAs.C. Semiconductor Electrode (Liquid Junction)Systems. Of more interest to chemists aresystems in which the photopotential to drive thewater-splitting reaction is generated directly at thesemiconductorAiquid interface (Figure IC). In 1839Becquerel noted small photoeffects when metal electrodeswere irradiated in electrochemical cells.5 Ratherextensive research was carried out on various metalelectrodes, sometimes covered with oxide or otherfilms, and immersed in a variety of solutions, includingsome containing fluorescent dyes.6 The effects seenwere usually small, and given the state of electrochemistryand knowledge of the electronic propertiesof solids, the results were generally poorly understood.The discovery of the transistor and interest by chemistsand physicists in semiconductor materials, notablySi and Ge, led to more extensive electrochemical andphotoelectrochemical studies, usually with the goal ofcharacterizing the semicond~ctor.~,~The modern era of semiconductor electrodes andinterest in these in photoelectrochemical devices forenergy conversion, especially via the water-splittingreaction, can be traced to the work of Honda andFujishima on single-crystal Ti02 electrode^.^ Indeed,water splitting in TiO2-based cells can be accomplished,but only with an additional electrical bias. Theproblem with Ti02 is that the conduction band is toolow (i.e,, at an insufficiently negative potential) togenerate hydrogen at a useful rate. Moreover, becausethe Ti02 band gap is large (3.0 eV for rutile), only asmall fraction of the solar light is absorbed and theefficiency of TiOz-based cells can never attain thespecified 10% level. Cells with Ti02 electrodes ofvarious types (e.g., single crystal, polycrystalline, thinfilm) have nevertheless been heavily investigated,largely because Ti02 is very stable and is a good modelfor understanding the semiconductorAiquid interface.The stability of the semiconductor in contact witha liquid and under irradiation is an important factor.To generate oxygen, rather energetic photogeneratedholes are required, and these tend to cause decompositionof the semiconductor. Thus, a key requirementin cells of this type, involving a single photojunction,is the discovery of a semiconductor with an appropriateband gap (< about 2.5 eV), with the conduction(4) White, J. R.; Fan, F.-R. F.; Bard, A. J. J. Electrochem. SOC. 1985,132, 544.(5) Becquerel, E. C. R. Hebd. Seances Acad. Sci. 1839, 9, 561.(6) Copeland, A. W.; Black, 0. D.; Garrett, A. B. Chem. Reu. 1942,31, 177.(7) Gerischer, H. In Advances in Electrochemistry and ElectrochemicalEngineering; Delahay, P., Ed.; Interscience Publishers: New York, 1961;Vol. 1, p 139.(8) Myamlin, V. A,; Pleskov, Yu. V. Electrochemistry of Semiconductors;Plenum Press: New York, 1967.(9) Fujishima, A,; Honda, K. Nature 1972, 238, 37.Ace. Chem. Res., Vol. 28, No. 3, 1995 143band sufficiently negative for hydrogen evolution andthe valence band sufficiently positive for oxygenevolution, so that it remains stable under irradiation.This single-junction semiconductor electrode has notyet been discovered. Indeed, it is only with materialswith band gaps even larger than that of TiO2, likeSrTiOs, that water splitting can be carried out withoutan additional electrical bias. The solar efficiency ofsuch cells is very small.D. Semiconductor Particle Systems. A considerablesimplification of the apparatus is possible if theelectrochemical cell can be replaced by simple dispersionsof semiconductor particles. In such dispersions,the semiconductor particles can be coated with islandsof metals that behave as catalytic sites, with eachparticle behaving as a microelectrochemical cell.lo Ti02has been a favorite material, although other compounds,such as CdS and ZnO, have also been studied.While a number of interesting photoreactions havebeen carried out, including the use of particles todestroy organics and to plate metals from wastewaterll and for synthetic purposes,12 reports on theuse of particulate systems for water splitting remaincontroversial. At best the solar efficiencies of processesreported to date have been very small (

144 Ace. Chem. Res., Vol. 28, No. 3, 1995 Bard and Foxlarge fraction of the incident light while still maintaininga high quantum efficiency. A photoelectrochemicalcell based on a ruthenium complex sensitizeradsorbed on Ti02 showed an efficiency above 7% indirect sunlight for conversion of light to electricity. Inthis example, iodine was generated at the dye-coatedelectrode and was reduced at the counter e1ectr0de.l~For water splitting, the oxidation reaction at the dye/liquid interface would have to generate oxygen, aconsiderably more difficult process. Indeed, a fundamentalproblem with dye-sensitized systems may bethe photochemical instability of the sensitizer, especiallywhen holes sufficiently energetic to liberateoxygen are produced during irradiation.F. Homogeneous and MicroheterogeneousSystems. An alternative to these electrochemicalmethods is to employ multicomponent solutions ordispersions as complex arrays for multistep watersplitting. Such a system would typically combine anefficient sensitizer, one or more relays or chargeaccumulation sites, and one or more gas evolutioncatalysts. Much effort has been devoted to understandingand optimizing the function of each constituentof these arrays, but their combination (even of thebest single components) has so far not produced thedesired synergism needed to attain the high solarconversion efficiency that is the goal of this quest.The most thoroughly studied systems employ microheterogeneousarrays, with either the oxidation(H2O - 0 2) or the reduction (HzO - H2) half-reactionsbeing the targeted goal. In such arrays, the excitedstate of the sensitizer is quenched either reductivelyor oxidatively, and most require a sacrificial electrondonor or acceptor. A major disadvantage of mostsystems studied so far is that they do not accomplishtrue water splitting; rather, they focus on eitherhydrogen generation in the presence of a sacrificialelectron donor or oxygen production in the presenceof a sacrificial electron acceptor. Even when optimized,a successful melding of these half-reactions intoa simple composite array remains elusive. Furthermore,the consumption of a sacrificial reagent inthe optimized half-reactions obviates the economicadvantages of simple water splitting and mars itsenvironmental attractiveness as a means for chemicalproduction of fuel by solar energy conversion. Includinga sacrificial reagent also often completely reversesthe net thermodynamics, converting an inherentlyuphill (photosynthetic) reaction such as water splittinginto a net energetically downhill (photocatalytic)transformation by coupling the endothermic steps withother highly exothermic ones.In the half-reaction involving the production ofhydrogen, the following coupled sequence is typicallyfollowed. The key photochemically-driven step is theoxidation of the excited sensitizer S, which is accomplishedby transferring an electron to a moleculeR that acts as a relay to a proton source, usually inthe presence of a catalyst Cat. The singly oxidizedsensitizer is then reconverted to its original neutralstate by interaction with a sacrificial electron donorD, which is consumed in a sequence of one or morechemical steps initiated by irreversible electron donation.The radical anion produced upon oxidation ofthe sensitizer acts as a relay reagent, transferring its(15) O'Regan, B.; Gratzel, M. Nature 1991,353, 737PSnPS*extra electron, usually in the presence of a catalyst,to water, thus producing hydrogen gas. The catalystthus functions both to accumulate electrons to providethe necessary electrochemical potential and the numberof electrons for the reduction of water and to serveas a gas evolution site.For efficient hydrogen evolution, several featuresare necessary: (1) the oxidative trapping must occurfaster than radiative or nonradiative relaxation of thesensitizer excited state; (2) each of the intermediatecharge relays must be stable in both the oxidized andreduced states so that the sensitizing dye can bemultiply recycled; (3) a robust catalyst for efficientevolution of hydrogen must be effective in trappingthe reduced relay; (4) the sacrificial donor must beinexpensive and readily available; and (5) each of thecomponents of the array must be chemically andoptically compatible. In operation, these requirementsare often structurally contradictory, and even when ahigh quantum efficiency (production of fuel per absorbedphoton) is observed, low solar efficiency (productionof fuel over the entire range of incidentphotons) is commonly encountered.Lehn and co-workers have provided an early exampleof one such ensemble, using a metal complexRu(bpy)~Cls, as sensitizer, triethanolamine as thesacrificial donor, and colloidal platinum generated insitu from KaPtC16 as the gas evolution cata1yst.l6Numerous other variants of these studies were undertakenin the subsequent decade,17 with sensitizersincluding metal complexes, metalated porphyrins andphthalocyanines, organic dyes, (e.g., xanthines orcyanines), or organic polymers [e.g., poly(viny1pyridinesll.In such studies, a range of relays (e.g.,viologens, phenanthroline complexes, multivalent cationslike V3+ or C$+, or redox enzymes like cytochromec3) and catalysts (e.g., metal alloys, enzymes, supportedmetal oxides, colloidal metals, etc.) were employed.In a typical optimized system [with Ru(bpy)3Cl~ assensitizer and methyl viologen as relay for hydrogenevolution on colloidal platinum or gold in the presenceof sacrificial ethylenediamine tetraacetic acid (EDTA)],hydrogen evolves efficiently, with catalytic turnover(about 100) limited by competing hydrogenation of themethyl viologen relay.'* Turnover efficiency could besubstantially improved by substituting a family ofnewly synthesized viologens that are resistant tocatalytic hydr~genation'~ and by replacing the hydrogenevolution catalyst with colloidal platinum. Moststudies with these artificial arrays, however, have(16, Lehn, J.-M.; Sauvage, J.-P. NOUL~. J. Chim. 1977, 1. 449.(17) Gratzel, M. Energy Resources through Photochemistry and Catalysis:Academic Press: New York, 1983.(181 Moradpour, A,: Amouyal, E.: Keller. P.; Kagan, H. NOUL.. J. Chim.1978, 2, 547.(19) See, for example: Crutchley. R. J.: Lever. A. B. P. J. Am. Chem.Soc. 1980, 102, 7128.

Artificial Photosynthesisfocused on forward- and back-electron-transfer ratesand on structural improvement of photophysicalproperties: for example, with a complex five-componentantenna system, Sasse and co-workers havereported a quantum efficiency of 93% for hydrogenevolution from an array containing g-anthracenecarboxvlatein acetate buffered to DH 5 in a solutionconiaining EDTA and methyl vioiogen.20Similarly, oxygen evolution could be induced byirradiation with redox catalysts that act as pools ofoxidative equivalent^.^^^^^^^^ In a recent variant, Calzaferrihas employed Ag+-loaded zeolites toward thisobjective, with self-sensitization by 02 being demonstratedto improve 02 evolution effi~iency.~~ In general,much lower efficiencies are observed for theevolution of oxygen than of hydrogen, mainly becauseof the difficulty in accumulating the necessary fouroxidative equivalents required to convert water tooxygen.Self-assembling arrays have also been constructedin media in which one or more of the components isconfined within a fured matrix, either a sol-gel or azeolite. In functionalized zeolites, for example, Malloukand co-workers have shown that a substantiallyextended charge separation lifetime of 37 ps can leadto a quantum efficiency for charge separation of about17% from flash photolysis meas~rements,~~ althougha quantum efficiency for hydrogen evolution of only0.05% has been reported by Dutta and co-workers fora similarly constructed array.25 Given that very longcharge carrier lifetimes (from excitation of pyrene toa viologen confined to a silicate matrix in the presenceof a mobile electron relay26) have been reported, onemight expect high efficiencies for hydrogen productionin analogous sol-gels if mass transport limitations canbe overcome.Another option to a self-assembling, multicomponent,microheterogeneous array is using a covalentsupramolecular assembly to accomplish long-termelectron-hole pair separation. Such arrays containingthree, four, or five absorbers, relays, or traps havebeen described by a number of groups as capable ofrelatively long distance (tens of angstroms) chargeseparation with high quantum efficiency and longlifetimes (83% for an electron-transfer pentad with acharge separation lifetime of 55 ps in chl~roform~~),(20) Johansen, 0.; Mau, A. W. H.; Sass, W. H. F. Chem. Phys. Lett.1983, 94, 107, 113.(21) Lehn, J.-M.; Sauvage, J.-P.; Ziessel, R. Nouu. J. Chim. 1979, 3,423.(22) Kalyanasundaram, K.; Gratzel, M. Angew. Chem., Int. Ed. Engl.1979, 18, 701.(23) Sulzberger, B.; Calzaferri, G. J. Photochem. 1982, 19, 321.(24) Krueger, J. S.; Mayer, J. E.; Mallouk, T. E. J. Am. Chem. SOC.1988, 110, 8232.(25) Borja, M.; Dutta, P. K. Nature 1993, 362, 43.(26) Slama-Schwork, A.; Ottolenghi, M.; Avnir, D. Nature 1992,355,240.(27) Gust, D.; Moore, T. A,; Moore, A. L.; Lee, S. J.; Bittersmann, E.;Luttrull, D. K.; Rehms, A. A,; DeGraziano, J. M.; Ma, X. C.; Gao, F.;Belford, R. E.; Trier, T. T. Science 1990, 248, 199.(28) van Damme, H. V. D.; Nijs, H.; Fripiat, J. J. J. Mol. Cutal. 1984,27, 123.(29) Katakis, D. F.; Nitsopoulou, C.; Konstantatos, J.; Vrachnou, E.;Falaras, P. J. Photochem. Photobiol. A (Chem.) 1992, 68, 375.(30) Smotkin, E.; Cervera-March, S.; Bard, A. J.; Campion, A,; Fox,M. A.; Mallouk, T. E.; Webber, S. E.; White, J. M. J. Phys. Chem. 1987,91, 6.(31)Fan, F.-R. F.; Keil, R. G.; Bard, A. J. J. Am. Chem. SOC. 1983,105, 220.(32) Nakato, Y.; Ueda, K.; Yano, H.; Tsubomura, H. J. Phys. Chem.1988,92, 2316.(33) Weller, H. Ber. Bunsen-Ges. Phys. Chem. 1991, 95, 1361.Ace. Chem. Res., Vol. 28, No. 3, 1995 145with the rate of forward electron transfer favored bya factor substantially greater than lo3 over the energydissipative back electron transfer.Despite the inherent difficulties encountered inattempts to optimize the oxidative and reductive halfreactionsin solution phase systems, it is clear thatimpressive progress has been made. Nonetheless,combinations of such half-reactions to a more complexarray capable of water splitting (with concomitanthydrogen and oxygen evolution) have proven to bedifficult. To our knowledge, no truly homogeneousnonsacrificial system for water splitting has yet beenreported. Even so, compartmentalized tandem oxidation-reductiondoes take place on several nonelectroactivespatially confined solids, although the catalyticturnover numbers remain small (turnover number ofabout 5 for a R~(bpy)3~+/clay/Eu~+ array29, the reactionsoscillatory, and the net quantum efficiencies low[4% upon excitation of tris[l-(4-methoxyphenyl)-2-phenyl-1,2-ethylenedithiolenicl tungsten in the visible29(400-500 nm)].Prospects for the FutureThese scientific difficulties continue to pose tantalizingchallenges to scientists bent on devising commerciallyviable means for solar energy conversion,especially as it relates to chemical fuel production andits contribution to a hydrogen-based economy. Althoughwe claim no special insight into the ultimatesolutions to these problems, we can speculate on thoseresearch directions that may seem most promising.Given the difficulties associated with homogeneousroutes employed so far, it is a fair bet that heterogeneousarrays will be needed for high-efficiency solarenergy conversion. The problem of accumulatingmultiple redox equivalents for the oxidation andreduction half-reactions also suggests that these systemswill require multifunctional components. Onepossibility is the use of multijunction devices, such asthe bipolar CdSe/CoS semiconductor photoelectrodepanel array on which unassisted water photodecompositioncan be stoichiometrically attained with anequivalent solar energy efficiency of about l%.30 It isalso likely that new semiconductor materials andmultielectron catalysts will be key components of suchheterogeneous arrays, perhaps with small metal clustersor films to protect the semiconductor surface, forexample, as noble metal silicides31 or microscopicallydiscontinuous metal over layer^.^^ Dramatically newfamilies of photosensitizers might also overcome someof the practical barriers still encountered in attemptsto utilize the whole solar spectrum: the use of nanoparticlesof one semiconductor on a highly porous filmof Ti0233 seems a very interesting possibility. In anycase, it seems clear that a creative breakthrough,probably focused within the boundaries defined byintriguing basic research conducted over the lastdecade, is needed to reach this “Grail”.MAF. gratefully acknowledges useful discussions on theseand related topics with Dr. Edmond Amouyal (Orsay,France).AR9400812