The metal-nonmetal transition

An account is given of some of the mechanisms which can lead to a transition from a metallic to a nonmetallic state, when a parameter such as the interatomic distance or temperature is varied. The simplest of these is the band overlap or Wilson transition, which occurs when a conduction band overlaps a valence band; this is discussed in § 2 and for noncrystalline systems in § 15. These transitions can be described in the Hartree-Fock approximation. If the insulating property is due essentially to the repulsion between electrons (e2/r12), then the nonmetallic state is normally antiferromagnetic. The possibility of describing it by normal band theory with a spin-dependent potential is discussed in § 5. It is emphasized that antiferromagnetism can exist in the metallic state, and that the conditions for the appearance of conductivity and the disappearance of antiferromagnetism are not always the same. The nonmetallic behaviour, that is the existence of a Hubbard gap, normally persists above the Néel temperature (as in NiO), as does the gap in some metals, but not in chromium. Disordered systems, such as doped semiconductors, are discussed; here in the metallic state we suggest that the two Hubbard bands overlap, and that the metal-nonmetal transition can be described as an Anderson transition (§ 16). This model gives a simple explanation of the negative magnetoresistance. In some materials a transition occurs which does not involve magnetic moments or structural change, and for d bands, following Halperin and Rice, and Weger, we introduce the concept of an `orbital orientation wave' in degenerate d bands (§§ 5, 19.3, 19.4). A number of specific materials are discussed.