Correlation between self-diffusion in Si and the migration mechanisms of vacancies and self-interstitials: An atomistic study

The migration of point defects in silicon and the corresponding atomic mobility are investigated by comprehensive classical molecular dynamics simulations using the Stillinger-Weber potential and the Tersoff potential. In contrast to most of the previous studies both the point defect diffusivity and the self-diffusion coefficient per defect are calculated separately so that the diffusion-correlation factor can be determined. Simulations with both the Stillinger-Weber and the Tersoff potential show that vacancy migration is characterized by the transformation of the tetrahedral vacancy to the split vacancy and vice versa and the diffusion-correlation factor is about 0.5. This value was also derived by the statistical diffusion theory under the assumption of the same migration mechanism. The mechanisms of self-interstitial migration are more complex. The detailed study, including a visual analysis and investigations with the nudged elastic band method, reveals a variety of transformations between different self-interstitial configurations. Molecular dynamics simulations using the Stillinger-Weber potential show, that the self-interstitial migration is dominated by a dumbbell mechanism, whereas in the case of the Tersoff potential the interstitialcy mechanism prevails. The corresponding values of the correlation factor are different, namely 0.59 and 0.69 for the dumbbell and the interstitialcy mechanism, respectively. The latter value is nearly equal to that obtained by the statistical theory which assumes the interstitialcy mechanism. Recent analysis of experimental results demonstrated, that in the framework of state-of-the-art diffusion and reaction models the best interpretation of point defect data can be given by assuming. The comparison with the present atomistic study leads to the conclusion that the self-interstitial migration in Si should be governed by a dumbbell mechanism.