The metal-semiconductor transition in three-dimensional disordered systems-reanalysis of recent experiments for and against minimum metallic conductivity

Abstract

Recently a series of experiments has been performed to investigate the transition from metallic to activated conduction, the so called metal-semiconductor transition. Most of these studies favour a continuous transition of sigma (0K). The author analyses presumptions and uncertainties in the interpretation of such experiments; why the use of an ansatz for sigma (T) may be misleading is discussed. Studies of amorphous Si1-xCrx show that there is a continuous phase transition from metallic to activated conduction at finite temperatures, whereas the metal-insulator transition at zero temperature is discontinuous. A phenomenological model for the behaviour of the conductivity near the metal-semiconductor transition in Si1-xCrx is given. By reinterpreting data from the literature this model is shown to be applicable to other amorphous semiconductors alloyed with transition metals, as well as to compensated and uncompensated heavily doped crystalline semiconductors, including Si:P, and, moreover, to granular metals. The metal-semiconductor transition should have the same nature for all these substances. The logical structure of the phenomenological model together with the results of tunnelling studies imply the hypothesis that the transition is related to the formation of a soft Coulomb gap. The 'width' of this gap should tend to zero as the transition is approached.