A surface diffusion mechanism at high temperature

Abstract

A theoretical model is proposed to account for the experimentally determined temperature dependence of surface self-diffusion (Arrhenius plot) of fcc metals which is known to exhibit a strong upward curvature near the melting point. The model is based on an idea, put forward earlier, that complex defects will contribute to the surface diffusion flux at temperatures approaching the melting point. Two different surface diffusion states are postulated to exist for surface defects: localized and non-localized states. The localized state is characterized by all vibrational modes of adatom, dimer and trimer, and one translational degree of freedom along the reaction coordinate. The non-localized state includes free translation of the adatom, and free translation and rotation of dimer and trimer. This model leads to an increase of the apparent activation energy of surface diffusion and an increase of the pre-exponential factor with increasing temperature as demanded by the experimental results. A numerical evaluation of the theory is presented for the case of a (110) copper surface which results in a diffusion coefficient of 1.2 × 10 −3cm 2/sec at the melting point in good agreement with experiment. A further result of this theory is a large “jump distance” of defects in the non-localized diffusion state which increases with increasing complexity of the defect. Critical experiments on possible anisotropy effects are discussed. The present model is believed to be supplementary to the surface roughening effect.