Publication date: 28th August 2024
While III-V semiconductors have achieved the highest photo-electrochemical solar-to-hydrogen conversion efficiencies, they are remarkably unstable during operation in a harsh electrolyte. The first half of this talk will focus on the degradation mechanism of III-V cells and surface modification strategies aimed at protecting them from photocorrosion. We applied noble metal catalysts, oxide coatings by atomic layer deposition, and MoS2 in an effort to protect the GaInP2 surface that was in contact with acidic electrolyte. We also grew epitaxial capping layers from III-V alloys that should be more intrinsically stable than GaInP2. The ability of the various modifications to protect semiconductor surfaces was evaluated by operating each photoelectrode at short circuit for extended periods of time.
In the second part of this talk, I’ll discuss potential pitfalls to consider when characterizing semiconductor materials for durability. Measurements of photoelectrode durability in three-electrode (half-cell) configurations are typically not representative of the results obtained for nominally the same electrodes/materials when tested in a two-electrode (full-cell) configuration. While full-cell measurements are the best proxy to predict performance in a deployed photo-electrochemical water-splitting system, there are few materials that can drive both the water reduction and oxidation half-reactions. Component-level or half-cell testing is the only option for materials unable to perform unassisted water splitting. During this talk the results of multiple durability tests, some with and some without a reference electrode, will be used to evaluate key differences between the two types of testing in an effort to elucidate what leads to the incongruity between half- and full-cell measurements. It is anticipated that with this understanding, the photoelectrochemical water-splitting community can begin to move towards an accepted protocol for long-term durability testing that will have more predictive relevance to realistic device configurations.