Interfacial electrochemistry of colloidal ruthenium dioxide and catalysis of the photochemical generation of hydrogen from water

The formation of hydrogen from water using solar energy is a very attractive research topic, because of the potential use of hydrogen as an alternative, clean fuel. It has been shown by many workers in the field that photochemical hydrogen generation can be achieved in an aqueous system, containing a sensitizer (a light absorbing solute), an electron relay, and a dispersed catalyst. The electron relay transfers electrons from the light-excited sensitizer to the surface of the catalyst, where subsequent reduction of H +takes place. In an ideal photochemical system for solar energy conversion, water itself would ultimately provide the necessary electrons for hydrogen formation, under simultaneous oxygen evolution. However, complete ("cyclic") photodissociation of water involves a number of complications, like the recombination of intermediate photoproducts. To separately study the formation of hydrogen, these additional problems can be bypassed by adding an electron donor, which decomposes after having reduced the oxidized sensitizer. Such simplified systems are known as "sacrificial".The present thesis is concerned with the generation of hydrogen in such a sacrificial photochemical system. The main purpose has been to gain insight Into the processes that take place at the catalyst/solution interface. Because of its wide application in photochemical model systems for hydrogen production, methylviologen (MV 2+) was chosen as the electron relay. Via its reduced form MV +., electrons are transferred from the sensitizer to the catalyst. Colloidal ruthenium dioxide (RuO 2 ) was used as the catalyst compound. It has the advantage over the more commonly used Pt catalysts, that it does not catalyze the undesired, irreversible hydrogenation of MV 2+.The heterogeneous processes in a hydrogen photoproduction system cannot be investigated without taking into account the reactions in solution too. Therefore, ruthenium trisbipyridyl (Ru(bipy)32+) and EDTA were chosen as photosensitizer and sacrificial electron donor, respectively: most of the (light-induced) homogeneous reactions that take place in the Ru(bipy)32+/MV 2+/EDTA/colloidal catalyst system have been studied extensively by different groups of researchers. In our experiments, the standard reaction mixture (58 ml) for photogeneration of hydrogen contained 2 x 10 -4M Ru(bipy)32+, 5 x 10 -4M MV 2+, 0.02 M EDTA, and 0.05 M acetate buffer (pH 4.6).Colloidal RuO 2 was prepared by thermal decomposition of RuCl 3 at ca. 400 °C. The material obtained is crystalline and only slightly contaminated with residual Cl, which is mainly present at the surface of the particles. The BET surface area is 20-30 m 2/g. Dispersions of RuO 2 are colloid-chemically very unstable, even in the presence of polymers or surfactants. They manifest the same electric double layer characteristics as many other oxide dispersions. The point of zero charge (p.z.c.) in indifferent electrolyte (KNO 3 ) is positioned at pH 5.7-5.8.Experiments with RuO 2 film electrodes, prepared from the same colloidal material and sintered at 700 °C, revealed that the hydrogen evolution reaction is chemically reversible. Hydrogen evolution at moderate overpotentials does not modify the RuO 2 . In the presence of 0.05 M acetate buffer (pH 4.6), the mass transport limited current density for H +reduction is high since it is related to the buffer capacity and not to the actual proton activity. In the potential range studied, the hydrogen evolution reaction can be described by the Butler-Volmer equation, with a transfer coefficient αof about 0.33, and an exchange current density i o of ca. 0.09 mA/cm 2geometrical surface area. The true exchange current density is smaller by a factor depending on the surface roughness of the film electrodes.Adsorption, of MV 2+at the RuO 2 /solution interface is mainly a result of attractive coulombic interactions (above the p.z.c. of RuO 2 ), but it has been shown that there are also more specific interactions. However, the specific adsorption is weak and not noticeable at high concentrations of back-ground electrolyte and pH values below the p.z.c. of RuO 2 . No indications were found that MV 2+adsorbs at the catalyst surface under operational conditions of hydrogen evolution. Under these conditions, the sensitizer Ru(bipy)32+does not adsorb either. On the other hand, the electron donor EDTA strongly adsorbs on RuO 2 from a 0.05 M acetate buffer solution of pH 4.6. However, this seems not to affect the electron transfer between methylviologen and RuO 2 film electrodes, a process which takes place with a transfer coefficient αof ca. 0.35 and a standard heterogeneous rate constant k oof ca. 1.4 x 10 -5m/s (referred to the geometrical surface area).The colloidal RuO 2 turned out to be a good catalyst for photoproduction of hydrogen, in spite of the strong tendency of the particles to form aggregates. During the hydrogen evolution process, it does not loose its catalytic properties. It was confirmed that RuO 2 does not catalyze the hydrogenation of methylviologen. A disadvantage of RuO 2 is that it absorbs light throughout the entire visible region.Upon illumination of the reaction dispersion and after a certain induction time, hydrogen production takes place at a constant rate (steady state). After several hours, the production rate gradually decreases to zero. The maximum attainable amount of H 2 is determined by the initial amount of electron donor: each EDTA species can regenerate three oxidized sensitizer ions. However, in most experiments the total H 2 yield was less due to gradual destruction of methylviologen in the bulk solution.The steady state ratio [MV +.]/[MV 2+] appeared to be always low, even in the absence of catalyst. This must be the result of a yet unspecified reaction which reconverts MV +.into MV 2+. Probably, a photogenerated intermediate species is involved in this process.In all the experiments with the hydrogen photoproduction system, the incident light intensity was a rate-determining factor. The steady state rate of hydrogen production depends also, but to a lower extent, on the sensitizer concentration. It has been shown in a simple way that the first step in the hydrogen evolution process, i.e. the excitation of Ru(bipy)32+, is first order in the light intensity and less than first order in the sensitizer concentration.The hydrogen production rate increases with EDTA concentration up to a plateau above ca. 0.02 M. At the plateau, the oxidized sensitizer is regenerated efficiently, preventing back-reaction with MV + .As a function of methylviologen concentration, the production rate exhibits a maximum around 2 x 10 -3M.At low quantities of RuO 2 (< 10 mg), the available catalytic surface area is rate-limiting. At higher catalyst amounts, the production rate is fairly constant; it decreases slightly with increasing RuO 2 amount due to the absorption of light by the RuO 2 particles.For any amount of RuO 2 , the stirring rate affects the rate of hydrogen evolution. Mass transfer of H +to the catalyst surface is not rate-limiting, as is also confirmed by the insensitivity of the production rate to the buffer concentration. This implies that the mass transfer of MV + .to the catalyst surface is a rate-determining factor.Most of the abovementioned experimental results can be satisfactorily simulated using a quantitative model, in which the homogeneous reactions are described by steady state kinetic equations and the heterogeneous processes as electrode reactions. The catalytic properties of RuO 2 can be understood and predicted by considering the RuO 2 aggregates as microelectrodes. Probably, the electrical conductivity of RuO 2 -on the level of a metallic conductoris essential for its catalytic performance.Hydrogen evolution at the catalyst surface takes place near the equilibrium potential of the H +/H 2 couple. At these potentials, reconversion of MV 2+into MV + .at the catalyst surface is negligible. The rate of the heterogeneous processes is determined by the rate of mass transfer of MV + .to the surface and, to a lower degree, by the rate of interfacial electron transfer. The mass transfer coefficient of methylviologen, under the standard stirring conditions, appeared to be in the order of 10 -5m/s.Mass transfer of methylviologen would undoubtedly be favoured by a better dispersion of the catalyst, since aggregation of the RuO 2 particles makes the surface less accessible. If the same or higher hydrogen production rates could be reached with lower catalyst amounts, the disadvantage of light absorption by the RuO 2 particles would become less important. Therefore, it seems worth trying again to stabilize dispersions of RuO 2 , for example by covalently linking polymers to the oxide surface.The simulations further indicate that, if the total surface area of the RuO 2 particles is assumed to be catalytically active, the kinetic parameters i o and k oare only ca. 10 times lower than the corresponding values found for the RuO 2 film electrodes per unit geometrical surface area. This is surprising, because the roughness factor of these electrodes was estimated to be in the order of several hundreds. This point deserves further attention. Aspects that could be investigated, are the influence of heat treatments on the reductive catalytic properties of RuO 2 and the comparison with kinetic parameters for single crystal RuO 2 electrodes.The presented model for the hydrogen production system does not account for the maximum in hydrogen production rate as a function of methylviologen concentration. The differences between model predictions and experimental results point to a progressive inhibition of the heterogeneous processes with increasing MV 2+concentration. This aspect will be the subject of further study, including investigation of the dependency of the electron transfer rate constant on the bulk concentration of methylviologen.The overall quantum yield of the hydrogen production in our standard system is low; even with an excess of catalyst, it is less than 4 %. Since reconversion of MV 2+into MV + .at the catalyst surface does not take place (each MV + .species that reaches the surface is used for hydrogen production), the low efficiency of the system results from the homogeneous proceases. Reconversion of MV + .into MV 2+in solution is competitive with the production of hydrogen and makes the system less efficient. The quantum yield is also limited by the low efficiency of the quenching of the excited sensitizer by methylviologen. At pH 4.6, less than 25 % of the quenching acts results in charge separation (according to our numerical simulations ca. 16 %). Furthermore, the gradual destruction of methylviologen under illumination of the reaction mixture, makes this compound unsuitable for use in any practical device for photogeneration of hydrogen.Combination of information regarding the homogeneous and interfacial aspects of the hydrogen production system leads to a picture that is at least semiquantitatively, and in many aspects quantitatively consistent. Extentions of this approach could be useful for the rational design of catalytic systems for solar energy conversion.