Nitride semiconductor is an interesting photocathode material for water splitting, but the preceding research suffered from low efficiency due to the wide bandgap and poor crystalline quality of p-type nitride crystals. We have proposed a tandem nitride electrode structure, n^--GaN (low-doped n-GaN) /AlN/n-GaN contact/substrate. This structure can function as a photocathode without using p-type layers, and it can be extended to multi-junction structure consisted of InGaN layers for better utilization of visible photons. By optimizing the structure and the crystal growth method, it was demonstrated that the onset potential of photocurrent can slightly exceed the redox potential for oxygen evolution (E^o_O2/H2O). However, the onset potential was more negative than expected and the cathodic current at E^o_O2/H2O was too low. Two major issues were identified: (1) excessive overpotential at the photocathode, and (2) leak current through the photo-active top n^--GaN to the n-GaN bottom contact. The latter is the anodic (reverse) current from the bottom contact through the pinholes in top n^--GaN, which are not easy to be removed completely.

In order to solve the issue, we here aim at the selective introduction of co-catalyst on the surface of n^--GaN active layer. Pt was chosen as a well-known co-catalyst for H₂ evolution with low overpotential. We propose the wise use of Pt electrodeposition: selective introduction of Pt on the n^--GaN surface using photo-induced cathodic current. In this method, no Pt was deposited on the exposed part of n-GaN contact at the bottom of the pinholes. It is expected that only the cathodic current generated in the top n^--GaN active layer is enhanced by Pt co-catalyst. This is indeed a self-alignment process.

As compared with the electrodeposition in dark, photo-electrodeposition allowed us to obtain fine and dense Pt particles (10~100 nm in diameter and ~10¹⁰ cm^-2 in density) as confirmed by AFM observation. As a result, the onset potential for cathodic current shifted to positive direction and the cathodic current increased. The improvement brought about the Pt selective introduction is especially drastic on the tandem photocathode with large number of pinholes on the top layer. When this technology was applied to the tandem photocathode with the most successful crystal growth with much reduced number of pinholes, the onset potential well exceeded E^o_O2/H2O. Cathodic current at E^o_O2/H2O amounted to 81 μA/cm², which is a promising value for zero-bias water splitting.