A pair potential analysis of dilute solutions of hydrogen in transition metals, alloys and compounds: Unrelaxed lattice model

Abstract

Hydrogen-metal potentials have been derived which are applicable to the prediction of partial molar energies of solution of hydrogen ( E ̄ H ∞ ) at infinite dilution in hosts having the A1 or A2 crystal structures as basis structures, e.g. L1 2 and B2, respectively. These potentials are based on fixed reference lattices, i.e. unrelaxed positions of host atoms at ideal site coordinates. The dependence of E ̄ H ∞ the lattice parameter in Pd-based alloy hosts is quantitatively accounted for if H occupies sites surrounded by Pd atoms only. Similarly, the E ̄ H ∞ values observed for certain A2-type (b.c.c.) alloy hosts correspond to those predicted for the most stable sites, which are the tetrahedral sites surrounded by four Ti atoms in 50 Nb-50 Ti and 50 Ti-50 Mo and by four Mo atoms in 70 Nb-30 Mo. Good agreement is also obtained between observed and predicted values of E ̄ H ∞ in the B2-type hosts FeTi, CoTi and NiTi. Contrary to previous experience, the short-range hydrogen-metal potential is not transferable between the Al and A2 structures, although the short-range potential in A2 hosts and the long-range potential in Al hosts yield the same predicted lattice parameter and cohesive energy for the corresponding metallic hydrogen. The prediction is made that FeRh (B2) should be similar to Fe Ti (B2) as a host for hydrogen, i.e. it should have the same E ̄ H ∞ value. Also, it is predicted that CoHf (B2) should attract hydrogen much more strongly than either component, i.e. a predicted E ̄ H ∞ of −80 000 cal (g atom) −1 compared with −60 000 cal (g atom H) −1 observed for Hf and −44 000 cal (g atom H) −1 observed for Co. The model provides an explanation for the Lundin effect and for exceptions to this effect. It also explains why ∂ E ̄ H ∂n values are negative for metallic hosts and suggests why these values are positive for certain compound hosts.