Influence of monoenergetic surface state occupation on Fermi level pinning of metal–GaAs interfaces

Abstract

Fermi level pinning at metal–GaAs interfaces is conventionally attributed to a high density of surface states. This article points out that not only the density and energy level of the surface states, but more importantly their electron occupancy determines the location and degree of pinning. For device applications, Fermi level pinning is most often based upon the Cowley–Sze model, which postulates a thin, electrically transparent interfacial layer, as would be expected for metalizations on GaAs which form arsenide compounds such as aluminum. However, this model is very difficult to apply to the more common surface spectroscopies (x‐ray and ultraviolet photoemission spectroscopies) because of its assumption of a uniform distribution of surface states with energy. This article presents a modified model which is based upon any multiplicity of monoenergetic donorlike or acceptorlike surface states and which therefore can be better correlated to the measured surface density of states. This model allows the full electronic structure of the surface states to be included, and it determines the degree and location of Fermi level pinning as a function of this defect structure alone. This eliminates the need for any empirically determined surface neutral energy, while still encompassing both the ideal Schottky and Bardeen limits of the Cowley–Sze model. An important result is that a high density of surface states is not by itself sufficient to cause pinning. The electrostatic restrictions of Gauss’ law require a surface charge on the interfacial layer in order to support any band bending within the semiconductor. This surface charge can arise only from sufficient occupancy of the surface states, e.g., the ionized surface acceptor states associated with arsenic vacancies or antisite defects in n‐type GaAs.