Abstract
I consider the possibility that the excited-state oscillator wave functions of dilute hydrogen in bcc metals overlap sufficiently with nearest-neighbor occupancy sites so as to produce hydrogenic energy bands, analogous to electronic energy bands in narrow-band semiconductors. The theory is motivated by the experiments of Magerl et al. as well as the earlier observation of ground-state tunnel splitting by Wipf et al., demonstrating quantum coherence in the motion of the hydrogen, despite the necessity of correlated motion by the surrounding metal atoms. Because of the latter complication, the relevant overlap integrals are not calculated from first principles. The band structures are given for the first (nondegenerate) and second (doubly degenerate) excitations ωI and ωII of the local oscillators, modulo a few irreducible overlap integrals, which are then determined by comparison with experiment. The fact that the experimental bandwidths for inelastic neutron scattering from dilute hydrogen in V, Nb, and Ta satisfy Γ(V)>Γ(Nb)>Γ(Ta) at room temperature (Rush, Magerl, and Rowe) finds a natural explanation in the theory. It is shown that the ωI and ωII bandwidths satisfy ΔEIIΔEI=(HIIHI)ϒ, where HI and HII are irreducible overlap integrals and ϒ is an (almost) universal constant for H in bcc metals, determined (essentially) by the geometry of the tetrahedrally coordinated hydrogen occupancy sites. On the basis of the band structure that I obtain, I estimate that ϒ34. Based upon physical reasoning, the relation (HIIHI)=(ωIIωI)2 is proposed. Given the (model-consistent) empirical result, ωIIωI212, this leads to the prediction ΔEIIΔEI32, to be compared with the neutron-measured ratios ΓIIΓI=1.3 and 2.0 for dilute hydrogen trapped at O and N impurities in Nb metal at T=4 and 10 K, respectively. The variation in