In agreement with experimental data on hydrogen diffusion in amorphous metals the density of sites on an energy scale could be described by a Gaussian distribution. Therefore, in Monte Carlo simulations of interstitial diffusion Gaussian distributions were used for the energies in the equilibrium and the saddle point configuration. In one case correlation between site energy and saddle point energy was taken into account. The computer simulations yield the interstitial concentration in the different sites, the long range diffusion coefficient and the average jump rates as a function of temperature, total concentration of interstitials for different widths of the Gaussian distributions and for a two-dimensional simple square and a three dimensional fcc lattice. The partial concentrations in different sites could be described by Fermi-Dirac statistics and the temperature and concentration dependence of the diffusion coefficient was in agreement with models assuming a constant saddle point energy. Thus activation energy of diffusion decreased and diffusivity increased with increasing concentration. In an Arrhenius plot of the diffusion coefficient negative curvature was observed for small concentrations and/or higher temperatures and linear behavior at high concentrations and/or lower temperatures which could be explained by a competition between entropy and energy gain.

Monte-Carlo Simulations of Interstitial Diffusion and Trapping-II. Amorphous Metals. R.Kirchheim, U.Stolz: Acta Metallurgica, 1987, 35[2], 281-91