A substrate doping variation study of the pinning states at metal/GaAs(110) interfaces

Abstract

Schottky barrier formation at metal/n-GaAs(110) interfaces has been studied with photoemission spectroscopy by varying the substrate doping from 4×1016 to 5×1018 cm−3. The results challenge certain assumptions that have been made in past barrier formation models. For the clustered systems investigated—In, Ga, and Ag on n-GaAs— the issue of lateral inhomogeneity of the interface state density is considered. Assuming that the interface states exist only beneath the clusters, lateral variation of the surface potential is expected. An increase of the substrate doping reduces the lateral depletion and should thereby affect this variation. However the core level line shapes exhibit no evidence of an inhomogeneous surface potential for any doping or coverage, even when the depletion length is in the neighborhood of the average cluster spacing. Consequently scenarios are considered in which pinning states are more uniformly distributed than the overlayer material itself. In modeling low-coverage potential barrier formation, one might suppose that the net interface charge for a given low coverage is independent of the substrate doping. Under this assumption the total band bending for a given coverage is expected to be inversely proportional to the dopant concentration. For the systems examined in this study—In, Ga, Ag, Sn, and Sb on n-GaAs—an inverse relationship between band bending and doping is observed, but the dependence is much weaker than a proportionality. In the context of a defect model of the interface charge states, this weakened dependence may be explained in terms of a defect state whose nature depends on the local Fermi level position, as has been suggested by Walukiewicz [J. Vac. Sci. Technol. B 5, 1062 (1986)]. The effect of metallic screening from the overlayer on the doping dependence of band bending is also considered.