Application of the Recent Theory of Anderson Localization to Doped Semiconductors

Abstract

The purpose of this paper is to discuss the electrical conduction in highly doped semiconductors from the point of veiw of the recently developed theory of Anderson localization and other related theories, for instance, those of the effects of electron-electron interaction in disordered systems. Properties of doped semiconductors have been studied for many years [1], and now it is known that in the regions of metallic conduction their fundamental features can be interpreted largely on the basis of electron gas model. The effect of the scattering of electrons by impurities are considered, if necessary, in simple ways, for instance, in Born approximation. The effects of the higher order corrections are believed to be only qualitatively important. However, there are some phenomena which have not been given a clear explanation for a long time. For example, the mechanism of negative magnetoresistance was clarified only recently [2–4] although it was discovered more than 25 years ago [5] and has been known to be one of the general and fundamental properties of doped semiconductors [6]. It is known that the localized spin gives rise to negative magnetoresistance [7], and in some cases the data were analysed on the basis of the localized spin mechanism but no definite conclusion was obtained. Some years ago, KAWAGUCHI et al. [8] observed negative magnetoresistance in a quasi-two-dimensional system (Cs covered Si surface) and found that its magnitude is dependent only on the component of the magnetic field perpendicular to the surface.