Metal oxide photoelectrodes for solar water splitting
Abstract
Efficient solar water splitting – using sunlight to produce hydrogen from water – has been an ambitious goal of the scientific community for over 40 years. At its heart this is a materials problem, with the photoelectrodes used in a photoelectrochemical cell having to satisfy all the constraints of a photovoltaic material (light absorption, charge transport) as well as being stable in water and having appropriately positioned band edges. Of the metal oxide systems studied for this purpose, we identified iron oxide (hematite, α-Fe2O3), tungsten trioxide (WO3) and an emerging (at the time) material, bismuth vanadate (BiVO4) as the most promising. In this dissertation we sought to understand and address the shortcomings of these materials, namely, carrier transport in BiVO4 and α-Fe2O3 and light absorption in WO3. We synthesized high quality single crystals of undoped and Mo and W-doped BiVO4 using the floating zone technique and carried out fundamental transport measurements. Electrons were shown to form small polarons and the Hall effect mobility was low, ~10-1 cm2 V-1 s-1 at 300 K. Critically, the mobility measured by the Hall effect may be vastly different from the drift mobility. Small-polaron hopping was found to be in the adiabatic regime and anisotropic conductivity was related to the structural arrangement of vanadium ions. Electrons are also thought to form small polarons in α-Fe2O3, but a thorough analysis had not been performed. We grew single crystals of Ti:α-Fe2O3 and characterized their electron transport to evaluate this model and probe the large anisotropy thought to occur between the basal planes. These revealed that the adiabatic small-polaron model was appropriate. Interestingly, electron transport in Ti:α-Fe2O3 was shown to be near-isotropic, contradicting the common view in the literature. Finally, we studied the effects of sulfur or iodine incorporation in WO3 with the aim to improve its visible light harvesting ability. Both of these impurities did increase visible light absorption, but performance was degraded in all cases except for very low concentrations of sulfur doping. These impurities likely form inter-gap defect bands which allow the absorption of longer wavelength light, but also degrade transport properties if present in large amounts.