Abstract
For pt.III see ibid., vol.8, p.775-84 (1978). Calculations of the energy, Delta E, of hydrogen in the transition metals Pd, Rh and Ni previously reported were in error because of an inaccurate approximation used for the exchange and correlation energy contribution xi xc. Corrected results are presented and now disagree with the experimental results. The theory has therefore been critically reexamined and several refinements considered in detail. It is shown that the original theory correctly included the non-orthogonality of the basis functions. Further, the inclusions of second-neighbour sites of the impurity in the Vdd' matrix and gradient corrections to the exchange and correlation potential are shown to have little effect. The original perturbation treatment of the Vss' matrix has been tested by treating the g=0 orthogonal plane waves more rigorously through a Wannier-function representation. Significant changes occur at low energies in the electron density of states but the Delta E values are little affected. It is possible that the g not=0 Wannier functions could produce significant effects. A model calculation for a non-spherical impurity potential shows that a spherically symmetric screened Coulomb potential may be inadequate to describe hydrogen in transition metals.