Charge transfer by surface states in the photoelectrolysis of water using a semiconductor electrode

Abstract

THE photoelectrolysis of water using n-type semiconductors such as TiO₂, (refs 1–4) SrTiO₃, (refs 5–7) and so on is of current interest and importance for the utilisation of solar energy to produce hydrogen gas for fuel. In steady state conditions under illumination and in the absence of H₂ and O₂ in the electrolyte, the rate of H₂ evolution on a cathode is proportional to that of O₂ evolution on a semiconductor anode. The O₂ evolution process is considered to proceed by electron tunnelling from electrolyte levels to holes in the semiconductor through the Helmholtz layer, which must take place between states of equal energy. If only isoenergetic electron transfer between holes and occupied electrolyte states overlapping in energy the valence band occurs, the hole currents for O₂ evolution at n-TiO₂ would be expected to be very low, because the equilibrium redox level of the H₂O/O₂ couple is considerably above (at 1.8eV for pH = 4.7 (ref. 2)) the top of the valence band. Experimentally, however, the current of holes at the TiO₂-electrolyte interface is found to be quite high and the efficiency in charge transfer of holes at the interface is almost 100%⁸. To account for such a high efficiency, a contribution from some surface states has been proposed^8,9. A possible importance of surface states in electrochemistry at TiO₂ electrodes has been described by Frank and Bard¹⁰. Also, a recent electroluminescence measurement made for n-TiO₂ indicates that occupied surface states exist near the midgap at the TiO₂–electrolyte interface¹¹. At present, however, it is not understood which kind or character of surface states can improve the charge transfer efficiency for holes at the interface. We present here a theoretical treatment of electron tunnelling between the semiconductor and electrolyte by surface states. We apply a theory of surface states (more exactly interface states) in tunnel MOS (metal–oxide–semiconductor) devices of Freeman and Dahlke¹² to the semiconductor–electrolyte interface. A quite high efficiency for holes at the interface can be obtained when the electron population of the surface states is controlled by the electrolyte and the surface states are almost completely occupied.