Effect of interatomic potential on simulated grain-boundary and bulk diffusion: A molecular-dynamics study.

Abstract

Molecular-dynamics methods are used to model diffusion in a \ensuremath{\Sigma}=5 [100] Al tilt boundary and in bulk. The diffusion coefficient D and activation energy Q for atoms in the boundary and in bulk are calculated for several different Al empirical interatomic pair potentials. These include a Morse potential, spline potentials fitted to bulk experimental data (elastic constants, phonon spectra, etc.), and a pseudopotential. Reasonable agreement is obtained with experimental diffusion values for Al, although activation energies are low. There is also a wide variation in results from one potential to another because the atomic motion is sensitive to the shape of the primary well or minimum of the interatomic potential. Rescaling the data with rough estimates of the different bulk melting temperatures that each potential predicts reduces the discrepancy between potentials. This is shown to be true for virtually any pair potential by calculating diffusion rates for Morse potentials with changing potential-well depth, position, and width. The variations between potentials are also explained in a quantitative sense by a simple calculation of potential-energy barrier height for vacancy migration in a bulk model. A method is given for using a linear relation between barrier height and melting temperature to predict diffusion coefficients and general transport properties in grain boundaries and bulk for any pair potential in any fcc metal.