Conduction in glassy and liquid semiconductors

Abstract

It is frequently stated that amorphous semiconductors have, in contrast to crystals, a value of the conductivity which is relatively insensitive to composition. This is explained by assuming that each atom in a glass has as many nearest neighbours as the number of bonds it can from (Ge, 4; As, 3; Te, 2), so that there are no free electrons available to carry a current. The validity of this concept will be examined; it is not true for some amorphous films (Mg-Bi) which are not strongly bonded. Also some glasses, when heated above the softening point, seem to change their coordination numbers and become metallic.

The theoretical models necessary to describe these results are outlined. In liquid metals and most amorphous metal films, the Ziman theory should be applicable, giving a conductivity equal to Se²L/12π³ħ, where S is the Fermi surface area and L the mean free path. When this is about 3000 Ω^–1 cm^–1, L is comparable with the distance between atoms and it cannot be smaller. For materials such as liquid Te for which the conductivity is lower, a “pseudogap” affects the conductivity. The lowest possible metallic conductivity is about 200 Ω^–1 cm^–1. For materials (liquids or non-crystalline solids) with lower conductivity, the current is due either to electrons excited to the “mobility shoulder” or to hopping conduction of the kind familiar in impurity conduction. A real gap (as contrasted with a pseudogap) must exist in transparent materials, and can be understood in terms of the tight-binding approximation.