A new approach to studying the electronic energy band structure on solids has been developed and calculations are reported for lithium metal. This framework, the GI method, leads to one-electron orbitals which in general are singly occupied, have no orthogonality constraints, are no longer required to have the full symmetry of the system, and lead to a description which is valid at all internuclear distances. Yet they still retain an independent particle interpretation.
In the application to solids, similar considerations apply. It is found that, for the alkalis, the resulting one-electron conduction orbitals can be taken to be Bloch functions for a smaller symmetry group than the bcc symmetry of the lattice. Thus, the resulting Brillouin zone (BZ) is smaller than that in Hartree-Fock (HF), and gaps can occur at the Fermi surface where none were previously permitted. For lithium these gaps are found to be sufficiently small so that many of the expected properties are not significantly affected and the resulting Fermi surface is found to be quite spherical in good agreement with, for example, position annihilation results. However, for such properties as the high field transverse magneto resistance, the soft X-ray emission spectrum, the optical absorption spectrum, the thermoelectric power, and the Hall coefficient, striking alterations in the description are obtained which lead to an appealing explanation of many of the anomalous properties of the alkalis, and seem to be in at least qualitative agreement with the experimental observations. The Mott paradox is also resolved; the metal is found to change continuously from a conductor to an insulator as the system is dilated.