J. Chem. SOC., Faraday Trans. 1, 1982, 78, 3331-3340 Ruthenium Dioxide: A Redox Catalyst for the Generation of Hydrogen from Water BY PATRICK KELLER AND ALEC MORADPOUR* Laboratoire de Physique des Solides, Universite de Paris-Sud, 91 405 Orsay, France AND EDMOND AMOUYAL Laboratoire des Processus Photophysiques et Photochimiques, Universite de Paris-Sud, 91 405 Orsay, France Received 19th February, 1982 Ruthenium dioxide, already known to be a catalyst for the oxidation of water to oxygen, has been shown to mediate effectively the generation of hydrogen from water in Ru(bipy)g+, methyl viologen and EDTA model systems, with efficiencies even higher (pH > 5) than those of previously investigated platinum catalysts. The main factor limiting the evolution of hydrogen is found to be the destruction of the organic electron-transfer relay, and this side-reaction is attributed to H&,,, species formed on the particles of catalyst. Ru0,-coated electrodes are widely used as so-called dimensionally stable anodes in the large-scale production of chlorine.In addition to this wide-spread industrial application, which is due to the high corrosion resistance and low overvoltage for chlorine evolution, other electrocatalytic properties as well as its basic electrochemical features have been widely studied1 during the last decade. Thus an exceptionally low overvoltage was also observed for oxygen evolution with this material;, however, the electrocatalytic activity decreased with time and the long-term stability of this type of electrode when generating oxygen was not sati~factory.~ In fact, in this case electrocatalysis has been attributed to a metastable substoichiometric RuO,(x < 2) oxidation state, and the fall-off of the oxygen- generation efficiency with time has been assigned to the possible formation of the more stable (less active) stoichiometric R U O , ~ ~ and/or to the oxidation to RuO,, which subsequently passes into solution causing degradation of these electrode^.^ On the other hand, RuO, exhibits hydrogen overvoltages similar to those on Pt electrodes,6 although hydrogen discharge produces a marked modification of the RuO, electrode and persistent H, evolution causes the RuO, layer to collapse, presumably by its reduction to metallic R U .~ Heterogeneous catalysts prepared from electrode materials, such as platinum or graphite, have long been known to promote various redox reactions.' The use of platinum as a catalyst for the reduction of water to hydrogen by vanadium(I1) was observed in 19028 and the oxidation of water to oxygen by cerium(rv) (also catalysed by platinum) has been studied more recently in a series of redox reaction^.^ These redox processes (catalysed by PtO,, Ir0,lo or RuO,ll) have recently been rediscovered, but the catalytic nature of the RuO, powders used in these experiments involving Ce4+ has been questioned.12 The renewed interest attracted by such redox catalysts has been stimulated by studies of the development of water-splitting processes to store solar energy.l 9 Thus, 333 13332 RuO, AS A REDOX CATALYST light-induced hydrogen formation catalysed by colloidal Pt149 l5 and oxygen generation promoted by Ru0,16 have been proposed in ' sacrificial ' model systems, where H, and 0, are produced at the expense of irreversible consumption of externally added compounds. Curiously, the catalytic activity of RuO,, which could have been used to promote both hydrogen and oxygen formation in these experiments because of its very low overvoltage for the generation of both species, was generally restricted to studies of oxygen production.l1?l6 Moreover, the necessity of the presence of Pt and RuO,, combined on semiconductor particles, was emphasized in recent studies of cyclic (non-sacrificial) water-splitting to hydrogen and oxygen by visible light.17 We have recently reported18 that RuO, is an effective redox catalyst for the generation of hydrogen with the Ru(bipy)i+/methyl viologen (MV2+)/EDTA sacrificial system.We now report detailed studies of the light-induced, Ru0,-catalysed formation of H, and examine the unavoidable limitations of this catalysis encountered in the present sacrificial photosystem and which may be anticipated for other photocatalytic processes. EXPERIMENTAL MATERIALS The sources and methods of preparation and purification of Ru(bipy);+, methyl viologen (MV2+), EDTA and platinum hydrosols have already been described.15 The ruthenium dioxide catalyst was used either as the commercially available powder (Alfa Ventron, soluble form) or as a more finely dispersed sample obtained by loading solid supports [Linde molecular sieve type LZ-Y52 (Alfa Ventron, catalyst A) or titanium dioxide (Alfa Ventron, catalyst B)] as described elsewhere.16 The mixed catalyst (RuO,/IrO,) loaded on the above mentioned zeolite was also prepared by this method.lG Aqueous solutions were prepared from distilled water and the pH was adjusted to the desired value using acetate (pH 4 and 5 ) , phosphate (pH 6, 7 and 8) or borate (pH 9) buffers.METHODS PHOTOCATALYTIC H, PRODUCTION The photochemical procedures and apparatus have been described in detail.'*, l5 30 cm3 solutions containing Ru(bipy)i+ (2 x lop4 mol dm-3), MV2+ (5 x lop4 mol dmp3) and EDTA (0.2 mol dm-3) were first thoroughly purged by argon and then the required amount of catalyst powder was added. The continuously stirred mixtures were irradiated with a 250 W halogen slide-projector lamp.The production rates and the total amounts of hydrogen were measured, after the evolved gases had been bubbled through a 50% potassium hydroxide solution, with a 6cm3 scale gas volumeter. The amount of MV2+ present during irradiation (or the electrochemical experiments) was measured by high-pressure liquid chromatography (h.p.1.c.) analysis. H, FORMATION MEDIATED B Y THE ELECTROCHEMICALLY REDUCED MV2+ A specially designed two-compartment electrochemical gas-tight cell was used. Aqueous solutions containing EDTA (0.12 rnol dmP3) and MV2+ (5 x rnol dmP3), buffered to pH 6, 8 or 9 were introduced respectively into the anodic (50 cm3, graphite auxiliary electrode) and the cathodic (60 cm3, Hg pool working electrode and saturated calomel electrode) compartments. The two solutions, linked by an agar/KCl bridge, were purged with argon before the addition of the catalyst.Th,: required potential to generate MV'+ (-0.7 V us. SCE) was then applied (Tacussel PRT 100- IX), the cathodic compartment being connected to the above mentioned gas volumeter. The rates of formation of hydrogen (if produced) and yields, as well as the amounts of MV2+ present in these electrolytic runs, were measured as a function of time.P . KELLER, A. MORADPOUR A N D E. AMOUYAL 3333 CATALYTIC HYDROGENATIONS The possibility of the hydrogenation of MV2+ catalysed by RuO, (1 atm* H,) was also examined with aqueous mixtures (pH 5, 6 and 8) containing MV2+ and the catalyst, as previously effected with platinum ~ata1ysts.l~ RESULTS AND DISCUSSION Ru0,-c A TA L Y SED H, PHOTO PRODUCTION : pH E F F E c TS Visible-light irradiation of outgassed solutions of Ru(bipy)i+, MV2+ and EDTA results in appreciable evolution of hydrogen when RuO, is added to the aqueous mixtures; the catalytic efficiency of this oxide was first examined as a function of pH (table 1).TABLE 1 .--HETEROGENEOUS CATALYSIS OF H, PHOTOPRODUCTION FROM IRRADIATED AQUEOUS SOLUTIONS OF Ru(bipy)i+ + MV2+ + EDTAU H, formationd - MV2+ amount of rate yield (turnover entry pH catalyst/pmolbp /cm3 h-l /mmol number)e 1 2 3 4 5 6 7 8 9 10 1 1 12 4 5 6 7 8 9 5 5 5 5 5 6 I :30 I:30 I:30 I:30 I:30 I:30 I:60 I :90 I : 120 I : 180 I1 : 0.7 I1 :0.4 1.5 2.6 2.6 1.2 0.2 0.0 2.7 2.6 2.6 2.3 4.5 0.2 0.63 1.03 0.96 0.41 0.13 0.00 1.12 1.12 1.12 0.80 1.35 0.04 42 69 64 27 9 0 74 74 74 54 90 3 a See Experimental section for concentrations; I = RuO,, Alfa Ventron, soluble form; I1 = colloidal Pt, catalyst A, ref.(1 5 ) ; determined within lo%, ref. (1 5 ) ; defined as the ratio of the total amount of H, to the initial amount of MV2+. As irradiation proceeded, hydrogen was first produced at a constant rate, but then gas formation decreased and spontaneously stopped at every investigated pH. The rates and total yields, measured in the usual way, varied markedly as a function of pH (table 1, entries 1-7), maximum values being observed for pH 5 and 6. The irradiated solutions were also analysed by h.p.1.c. as a function of the irradiation time with respect to the amounts of MV2+ present in the mixtures.These analyses allowed us t o assign the termination of gas production to the almost total disappearance of the organic relay; the rates of this destruction were also observed to be dependent on the pH values (fig. 1). Several aspects of the Ru0,-catalysed formation of hydrogen are reminiscent of * 1 atm = 101 325 Pa.3334 RuO, AS A REDOX CATALYST 100 75 h E A 50 > z 25 0 L I 1 200 LOO 600 time/min FIG. 1 .-Analysis of MVZ+ during hydrogen photoproduction by Ru(bipy)$+ + MVz+ + EDTA + RuO,, using h.p. 1.c. Irradiations correspond to the experiments reported in table 1 : curve A (entry 2, pH 5), curve B (entry 4, pH 7). features typical of the results obtained previously with a colloidal-platinum catalyst :15 (i) the hydrogen-formation rates and the dependence of the overall yields on the pH and (ii) the limitation of gas production by the destructipn of the MV2+ organic relay.However, significant differences distinguish the two processes. With the platinum catalyst, an MV2+-hydrogenation reaction limits the catalytic efficiency of hydrogen photogeneration. This side-reaction (see also fig. 2) accounts for the observed variations in the hydrogen formation rates (quantum yieldslS) and the overall yields as a function of the amount of Pt catalyst added: both factors increased and then sharply decreased as the amount of Pt was increased.l49 1 5 9 l9 On the other hand, when the amount of catalyst added corresponded to that required to obtain the optimum efficiency of the photosystem drastic differences were observed, this time as a function of the pH of the irradiated solution, in the levels of hydrogen production.Thus, the yields at pH 6 were lower, by more than one order of magnitude, than those obtained at pH 5 (compare entries 1 1 and 12 in table 1); these profound differences were explained by the increasing efficiency of Pt to hydrogenate MV2+ as the pH value was raised.15 Turning now to RuO,, the possible hydrogenation of the MV2+ relay was also envisaged in the present photoproduction of hydrogen. In fact the use of an Ru0,-basedP . KELLER, A. MORADPOUR A N D E. AMOUYAL 3335 t o o 75 h E + pl > 2 50 25 C g B I I 50 100 time/min FIG. 2.-Analysis of the stability of MV2+ in the presence of Pt and RuO, under hydrogen, using h.p. 1.c. Curves A and B, Pt catalyst, pH 5 and 7, re~pectively;'~ curve C , RuO, catalyst, pH 5, 6 and 8 (see text for further details).catalyst in heterogeneous hydrogenation is not uncommon.2o This possibility was first examined (as with the Pt catalyst)15 in the absence of any extraneously formed photochemical intermediates, by the simple combination of MV2+ and RuO, under hydrogen (fig. 2). Under these conditions, the stability of MV2+ in H,-purged solutions in the presence of RuO, was higher in comparison with colloidal Pt. Whereas the latter promoted large decreases in the initial amounts of MV2+, with strongly pH-dependent hydrogenation rates (fig. 2, curves A and B), the former was inactive catalytically for hydrogenation over the photochemical timescale. The hydrogenation of MV2+ was only noticeable after extremely long induction periods, (as sometimes observed with Ru0,-based catalysts before ruthenium metal, effectively active in the hydrogenation, is p-oduced in situ20)* 8 h at pH 5 and > 24 h at pH 6 and 8 (fig. 2, curve C ) .Moreover, in Pt-catalysed processes the electron-exchange reaction between hydrogen and MV2+ [equilibrium (l)] MVZ++iH, MV+'+H* catalyst is readily identified (i.e. at pH 7) by the intense blue colour of MV'+. When RuO, is added to aqueous solutions of MV2+ under hydrogen, the formation of the radical cation is no longer detected. These facts indicate the inability of this oxide to * Ruthenium metal as a catalytic species for the formation of hydrogen in the present photosystem may, however, be excluded because of its great ability to catalyse the hydrogenation of MV2+ with an efficiency even higher than that of colloidal Pt.3336 RUO, AS A REDOX CATALYST chemisorb hydrogen dissociatively21 (to hydrogenate MV2+ to undesired side-products) within the photochemical timescale, although the evolution of hydrogen from photochemically reduced MV'+ [through the reverse step of reaction (l)] is mediated efficiently by this catalyst.The striking difference in the observed effects of pH on the rates of hydrogen photoproduction catalysed by RuO, (table 1, entries 1-5) as compared with the Pt-mediated process (table 1, entries 1 1 and 12), is at first surprising if only the overpotential for H, evolution is considered (similar values have been determined for both catalysts6).This simply reflects the large reaction-rate differences for the catalysis by Pt and RuO, of the two competing processes, hydrogen formation [reaction (l)] and undesired hydrogen consumption. Considering the respective Nernst potentials associated with MV2+ oxidation (Emed) and water reduction (EH,) half-reactions [reactions (2) and (3)], MV2+ + e- g MV' + (2) H+ +e- $H2 (3) E H , = - (RT/nF) pH the catalysed formation of H, through the MV2+-mediated redox equilibrium [the sum of reactions (2) and (3)] requires, on a simple thermodynamic basis, the amount of MV'+ to exceed the equilibrium value at a given pH. Steady-state gas production then results from an excess of MV'+ generated upon illumination: at pH 8 the calculated I value of MV2+/MV'+ was 0.286 (MV'+ = 77%) and H, was effectively observed with RuO, as a catalyst (table 1, entry 5), whereas Pt failed to produce any detectable amount of H, under these conditions.However, an analysis of the amount of MV2+ in the irradiated aqueous mixture of Ru(bipy)t+ + EDTA + PtO,* at pH 8 showed a fast decrease in MV2+ (fig. 3, curve A) as compared with the rates of the uncatalysed (presumably radical-initiated) MV2+ destruction processL5 (fig. 3, curve B) or with the Ru0,-catalysed hydrogen production at this pH (fig. 3, curve C). Thus the Pt-based catalysts even prevented the production of H, in high-pH media, due to their high efficiency to hydrogenate the MV2+ relay. In this'case the H, produced was used exclusively in situ (at pH 8) to reduce the unsaturated organic compound.Note that a quantitative comparison of H, formation rates, corresponding to rather different H, yields, as a function of pH may be rather meaningless for the photosystems under study. Consequently, due to the interference of the side-reactions inherent in the MV2+/Pt couple (which depending on conditions may even be the main reaction, vide supra results at pH 8) these rate values may not bz appropriate as supporting data to test the otherwise interesting electrochemical predictive model of colloidal catalysis recently 23 RU0,/MV2+ COUPLE: THE LIMITING FACTOR I N H, GENERATION The hydrogenation of MV2+ has been ruled out for Ru0,-catalysed photoprocess by the independent investigation of the simple RuO, + MV2+ mixtures under hydrogen (fig. 2, curve C).Nevertheless the H, evolution obtained for visible-light irradiation of the Ru(bipy);+ + MV2+ + EDTA + RuO, solution stopped spontaneously, and the total hydrogen yields (table 1) were thus limited, as with Pt at every investigated pH, by the disappearance of the organic relay. The occurrence of other side-reactions, involving for example the EDTA radical-type oxidation by-products,15 seemed * PtO, was used here instead of colloidal Pt, which was unstable at this pH.P. 100 75 A 5 0 > E: 25 KELLER, A. MORADPOUR AND E. AMOUYAL 3337 I I-, 200 L 00 600 time/min FIG. 3.-H.p. 1.c. analysis of the amount of MV2+ present in the photochemical experiments at pH 8. Irradiated solution corresponds to the composition reported in table 1. Curve A, PtO,; curve B, without catalyst; curve C, RuO, (3 x lops mol).reasonable. Therefore hydrogen formation, mediated by electrochemically reduced MV'+ and catalysed by RuO,, was investigated in the absence of any extraneous photochemical electron-transfer process. The stability of MV2+ was examined, by h.p.1.c. analysis, in a simple two-component RuO,/MV'+ system. Hydrogen formation was in fact observed when the radical cation was continuously generated at pH 6 and 8, and the corresponding amounts of MV2+ were found to decrease with time (fig. 4, curves A and B). At pH 9 however, the organic compound was found to be very stable (fig. 4, curve C) and hydrogen formation was no longer detected in this medium. Thus whenever H, was generated, a side-reaction inherent to the simple RuO,/MV'+ couple took place and destroyed the organic relay.The possibility of Ru0,-catalysed hydrogenation of MV'+ was first considered. However, this hypothesis (which was in contradiction with the inability of RuO, to activate hydrogen21) was rejected, as MV'+ was also found to be stable (pH 9) when H,-purged solutions (instead of argon) were used in the electrochemical cell. The 'reactivity' of H,, produced by the Ru0,-catalysed process [reaction (4)] MV'+ + H+ + MV2+ + BH, (4) RuO, on MV'+ was thus completely different from that of the added hydrogen gas. I on FAR 783338 100 75 E + N > E 50 25 RuO, AS A REDOX CATALYST \ 200 400 600 time/min FIG. 4.-Analysis of the stability of MV2+ in the Ru0,-catalysed formation of hydrogen mediated by electrochemically reduced MV'+, using h.p.1.c.Curves A and B, pH 6 and 8, respectively; curve C, pH 9 (see Experimental section for the details). These results clearly show that Hiads) species, spilled over the catalyst surface by the electron-transfer relay,24 react efficiently with the organic salt. With electrodes made of RuO, these adsorbed species are assumed to lead to hydrogen through an ion-atom recombination step2' [reactions (5) and (6)] H++e-,H;,d,, ( 5 ) (6) and are also envisaged to be the catalytically active intermediates in heterogeneous hydrogenations. RuO, did indeed catalyse effectively the ' self-destruction ' of MV' + by these Hiads) species, although it was not, strictly speaking, a hydrogenation catalyst under the present conditions. Hiads) + H,O+ + e- * H2 + H,O MODIFIED RUO,-BASED CATALYSTS The maximum efficiencies in the photoproduction of hydrogen obtained in the Ru0,-catalysed system were closely comparable to the values obtained in the colloidal-Pt-catalysed photosystem (cf. entries 7 and 11 in table l), but the amount of RuO, macrodispersed powder required in these irradiations (entry 7 = 60 pmol) was roughly two orders of magnitude higher than the corresponding amount of the more finely dispersed Pt hydrosols (entry 11 = 0.7 pmol).In order to increase the catalyst efficiency (turnover number) solid-support deposited oxides were investigated. Very efficient zeolite-supported metal oxides have been proposed for the photoinduced generation of oxygen from water,16 and now several oxides have been prepared by a similar procedure and examined as catalysts in hydrogen photogeneration.TheP. KELLER, A. MORADPOUR AND E. AMOUYAL 3339 hydrogen-formation rates and yields promoted by these supported catalysts (table 2) are close to those previously measured using unsupported RuO, powders (entry 1, table 2), but these values were obtained with significantly lower amounts of added oxides (cf. entries 1 and 3, table 2). Moreover, the most efficient sample for hydrogen production (RuO, + 11-0, mixed oxides, entry 4) was also found to be the best catalyst for oxygen formation in the Ru(bipy):+ + Co(NH,),Cl* Cl, sacrificial mode1,16 although this is probably not related to any similarity between the catalytic sites (not yet precisely defined) involved in hydrogen and oxygen formation. TABLE 2.-FORMATION OF HYDROGEN BY THE VISIBLE-LIGHT IRRADIATION OF AQUEOUS SOLUTIONS OF Ru(bipy)i+ + MV2+ + EDTAU PROMOTED BY VARIOUS HETEROGENEOUS METAL-OXIDE CATALYSTS catalyst H, formationC amount added rate yield entry typeb /pmol /cm3 h-l /mmol 1 RuO, (powder) 60 2.7 1.12 2 RuO,/zeolite 15 1.8 0 .9 2 3 RuO,/TiO, 7.5 2.4 0.67 (catalyst A) (catalyst B) zeolite (catalyst C) 4 RuO,+IrO,/ 11.4 3.0 1.12 a , See Experimental section for concentrations and for catalyst preparation method; see note (d) of table 1. CONCLUSIONS The efficiency of Ru0,-based catalysts for the generation of hydrogen from aqueous mixtures of Ru(bipy)i+ + MV2+ + EDTA irradiated by visible light was found to be at least comparable to (or even higher than, depending on the pH values) the efficiency of previously investigated colloidal-Pt catalysts.The higher efficiency of RuO, is related to an inability to chemisorb hydrogen dissociatively and to catalyse the undesired hydrogenation of the organic relay, which limits the formation of hydrogen promoted by Pt. Nevertheless, MV2+ was found to be unstable in this medium and its destruction is attributed to the side-reactions initiated by Hiads) species confined to the catalyst surface. More generally, these H a adsorbed intermediates, unavoidable in any catalytic hydrogen-formation process, may obviously contribute to other undesired short-circuit processes (e.g. if oxygen was produced at a neighbouring site). These reactions must be considered in the absence of any specific catalyst (such as Pt if H, and 0, are considered) for the recombination steps.The design of a new method of H, and 0, production using separate c~mpartments~~ is therefore highly attractive and will certainly be developed in the future. 108-23 340 RUO, AS A REDOX CATALYST S. Trasatti and W. E. O'Grady, in Advances in Electrochemistry and Electrochemical Engineering, ed. H. Gerischer and C. W. Tobias (John Wiley, New York, 1981), p. 177. ' L. D. Burke, 0. J. Murphy, F. F. O'Neill and S . Venkatesan, J . Chem. Soc., Faraday Trans. 1 , 1977, 73, 1659, and references cited therein. R. S. Yeo, J. Orehotsky, W. Wissher and S . Srinivasan, J. Electrochem. Soc., 1981, 128, 1900. S. Trasatti, J . Electroanal. Chem., 1980, 111, 125. L. D. Burke and J. F. Healy, J . Electroanal.Chem., 1981, 124, 327, and references cited therein. D. Galizzioli, F. Tantardini and S. Trasatti, J . Appl. Electrochem., 1974, 4, 57. M. Spiro, J . Chem. Soc., Faraday Trans. I , 1979, 75, 1507. A. Piccini and L. Marino, Z . Anorg. Allg. Chem., 1972, 32, 55. M. Spiro and A. B. Ravno, J . Chem. Soc., 1965, 78. lo J. Kiwi and M. Gratzel, Angew. Chem., Int. Ed. Engl., 1978, 17, 860. l 1 J. Kiwi and M. Gratzel, Angew. Chem., Int. Ed. Engl., 1979, 18, 624. l 2 A. Mills and M. L. Zeeman, J . Chem. SOC., Chem. Commun., 1981, 948. l 3 N. Sutin and C. Creutz, Pure Appl. Chem., 1980, 52, 2717. l 4 P. Keller and A. Moradpour, J . Am. Chem. Soc., 1980, 102, 7193 and references cited therein. l5 P. Keller, A. Moradpour, E. Amouyal and H. B. Kagan, Nouv. J . Chim., 1980, 4, 377. l6 J. M. Lehn, J. P. Sauvage and R. Ziessel, NOUD. J . Chim., 1980, 4, 355. l7 E. Borgarello, J. Kiwi, E. Pelizzetti, M. Visca and M. Gratzel, J . Am. Chem. Soc., 1981, 103, 6324. l8 E. Amouyal, P. Keller and A. Moradpour, J . Chem. Soc., Chem. Commun., 1980, 1019. l9 E. Amouyal, D. Grand, A. Moradpour and P. Keller, Nouu. J. Chim., 1982, 6, 241. * O See, for example, L. F. Fieser and M. Fieser, Reagents for Organic Synthesis (John Wiley, New York, 21 D. Galizzioli, F. Tantardini and S . Trasatti, J . Appl. Electrochem., 1975, 5, 203. 22 D. S. Miller, A. J. Bard, G . McLendon and J. Ferguson, J . Am. Chem. Soc., 1981, 103, 5336. 23 D. S. Miller and G. McLendon, J . Am. Chem. SOC., 1981, 103, 6791. 24 K. Kopple, D. Meyerstein and D. Meisel, J. Phys. Chem., 1980, 84, 870, and references cited therein 25 F. Chojnowski, P. Clechet, J. R. Martin, J. M. Herrmann and P. Pichat, Chem. Phys. Lett., 1981, 1967), p. 983. for similar processes. 84, 555. (PAPER 2/3 1 1)