Sensitivity of lattice defect calculations for fcc metals to empirically derived interatomic potentials

The relevant factors which must be considered for the empirical fitting of a two-body interatomic interaction to the elastic moduli and phonon dispersion relations are discussed. It is shown that for nickel and copper there is no necessity to use any more complex interaction than a near-neighbor central force. There is, however, no unique way to determine an appropriate potential from perfect lattice data alone, since the interaction must be known at distances which do not occur in the perfect lattice for the calculation of lattice defect properties. A set of three potentials for copper, all of which are equivalent for the perfect lattice, were used to investigate the sensitivity of vacancy and interstitial formation and migration energies to the details of the potential. Vacancy calculations were not very sensitive, the ratio E v/(E +E v) varying from about 0.4 to 0.5. The interstitial results showed an overall greater sensitivity, as the relative formation energy of different defect configurations, including which defect is the most stable, varied with the potential. In all cases, however, the migration energy remained small. Although the detailed results are dependent on variations in the interactions, the overall picture presented by the calculations is unchanged-vacancy migration energies equal to or slightly less than formation energies, large interstitial formation energies, and small interstitial migration energies. Independent of the details of the potential and in agreement with earlier calculations, this work indicates that radiation damage annealing due to free migration of interstitials occurs in the so-called Stage I region.