Asaf Kay1 Daniel Grave1 David Ellis1 Kirtiman Malviya1 Hen Dotan1 Avner Rothschild1

1, Technion-Israel Institute of Technology, Haifa, , Israel

Iron oxide (α-Fe2O3, hematite) is a promising material for use as a photoanode in photoelectrochemical cells for solar water splitting due to its long term stability under operating conditions, cost, abundance, and visible light absorption capabilities. However, state of the art photoanodes still fall significantly short of the theoretical efficiency. This poor performance is generally attributed to short lifetime of photogenerated minority carriers, i.e. holes, primarily measured by optical pump-probe methods, and also to a reported small diffusion length, resulting in significant bulk recombination. Despite much research into hematite photoanodes, an accurate model describing the device physics and charge transport of hematite photoanodes is still lacking. The most widely used model to describe photoanode behavior is the Gartner model, which describes the photocurrent as a sum of all carriers photogenerated within the depletion region plus those generated in the bulk which are able to diffuse to the depletion region. It is widely accepted that hematite photoanodes display short depletion and diffusion lengths of only several nm, therefore most of the efforts to improve hematite performance have centered on nanostructured porous layers. In this talk, we show that, in seeming contradiction to one or more of the preceding assumptions (i.e., Gartner model, diffusion length, or depletion region width), holes generated at least 700 nm away from the surface in a thick Ti-doped hematite planar photoanode are able to reach the surface and contribute to the photocurrent. Furthermore, we show that photogeneration of holes closer to the surface does not necessarily result in higher probability of charge carrier extraction and that the wavelength dependence of the generation plays a significant role in the ability to extract charges.