Surface states and Fermi level pinning at semiconductor/ electrolyte junctions

Abstract

The influence of surface modification procedures on the electronic interface properties of p-InP, p-Si, p- and n-type GaAs and p-CuInS 2 is investigated at the semiconductor redox electrolyte contact. Plots of photovoltage versus redox potential allow the determination of surface state densities using a simple model. The calculated surface-state densities D s range from 10 11 cm −2 eV −1 in an unpinned situation to about 2×10 14 cm −2 eV −1 for pronounced Fermi level pinning evidenced by independence of the photovoltage on changes of the redox potential. On InP and GaAs, deviations from the linear relationship of photovoltage and redox potential are observed. The deviations are attributed to an energetically non-uniform distribution of surface states. The model considerations are extended to include variations of D s with electron energy. In some cases, a dependence of the cell voltage in the dark on redox potential is observed which is attributed to electrochemical reactions which lead to a dynamic pinning of the Fermi level. The plot of photovoltage versus redox potential can give semi-quantitative information on semiconductor surface state densities for sufficiently high light intensity. The model is particularly simple for a uniform distribution of surface states but can be extended to varying density yielding an essentially similar mathematical form. The results presented here were all obtained in vanadium redox electrolyte and care should be taken if various redox couples are used to span a wider potential range. The specific surface interaction which might take place between the semiconductor and the respective solution can drastically alter the surface properties. Evidence for such a pronounced influence of the surface chemistry on surface electronic properties is given in this article. As the potential range scanned in this work was quite small, the data for D s can be considered to be relatively accurate. The model does so far not include, however, changes of the surface recombination velocity with changes of D s and cases where complete Fermi level pinning occurs. Investigations at lower light levels could provide such information. The function used in the determination of D s from the photovoltage measurements has hyperbolic character and is shown in Fig. 13. The steep change for slopes larger than about 0.9 also shows that an experimental uncertainty of ±0.05 in slope can result in a difference for D s of greater than an order of magnitude. It must therefore be concluded that the data obtained for c > 0.97 cannot be taken for an evaluation of D s. In fact, the uncertainties associated with the influence of a voltage-dependent surface recombination rate affecting V r would make the evaluation even more problematic in this range, although S r should be low for these surface state densities in the range of 5 × 10 11 cm −2 eV −1.