The metal-insulator transition in Si:P

Abstract

Recent experiments on the metal-insulator (MI) transition in Si:P are reviewed. This transition occurs as a function of phosphorus concentration N at N c ≈3.5·1018 cm−3. It has become a prototype MI transition bearing features both of an Anderson transition due to the statistical distribution of donor atoms and of a Mott transition due to electron-electron interactions. In the metallic regime above N c the temperature dependence of the electrical conductivity e at low temperatures is governed by localization and electron-interaction effects. Upon approaching N c , e goes to zero as(N−N c ) v with v ≈ 0.5 for uncompensated and v ≈ 1 for compensated Si:P. This difference may lie in the importance of electron-electron interactions in the former. In the vicinity of the MI transition both delocalized and localized electrons are simultaneously present as evidenced by a large number of different experiments such as magnetic susceptibility, specific heat and magnetic resonance. In particular, the magnetic field dependence of the specific heat allows a detailed analysis of the contributions of localized and itinerant electrons. A field-induced shift towards localization is clearly observed. Finally, the delocalized electrons exhibit a well-defined plasma edge in the far-infrared reflectivity.