CORRELATION-FUNCTIONS IN SURFACE-DIFFUSION - THE MULTIPLE-JUMP REGIME

Although the theories (transition state theory and jump diffusion) usu ally employed to describe surface diffusion cannot give information ab out the motion of adsorbed particles inside the potential wells, inter esting results were obtained recently by MD simulations showing enhanc ed oscillations in the mean-square displacement before the linear beha viour in time is finally reached; at the same time evidence of long co rrelated jumps was found. In this paper the single-particle diffusion on surfaces is studied in the framework of the continuous Brownian mod el (Klein-Kramers equation). The Klein-Kramers dynamics is first analy zed by qualitative considerations about the dissipation integral, obta ining necessary and sufficient conditions on typical time scales in or der to get different migration mechanisms. In particular, at high barr iers, conditions are found for multiple jumps to be inhibited or to ta ke place with small and considerable probability respectively. Then st arting from the dynamic structure factor, the relevant correlation fun ctions (velocity self-correlation spectrum and mean-square displacemen t) are evaluated, together with the jump probabilities, at the same po tential barrier and friction of the MD calculations. At these values o f the parameters diffusion proceeds, as expected, by a considerable fr action of multiple correlated jumps and many oscillations are found in the mean-square displacement in good agreement with MD results.