Tight-binding molecular dynamics study of vacancy-interstitial annihilation in silicon

We have studied the annihilation of a tetrahedral interstitial by a vacancy using molecular dynamics and a tight-binding model due to Goodwin-Skinner-Pettifor. At both 1473 and 1173 K, the annihilation process was dominated by the movement of the faster-diffusing vacancy. Once the vacancy moved to a nearest-neighbor position relative to the interstitial, annihilation was inevitable. As the initial distance between the T-interstitial and the vacancy increased, the number of pathways by which annihilation occurs increased accordingly. Fast and slow annihilation pathways were identified. The process was found to be highly stochastic in nature, requiring multiple simulations to provide adequate statistics. System size was not a factor in altering the results. The capture radius, within which annihilation is assured, was found to be between 5-6 Angstrom at 1473 K, although it is clear from the orientation dependence of the results that a spherically symmetric view of the capture radius is somewhat inappropriate. Limited investigation of annihilation between a (110) split interstitial and a vacancy confirmed the basic results of Tang el al. (Ref. 21) (using a Kwon tight-binding model) that the formation of a double-vacancy: double-interstitial configuration, dubbed a "bond-defect" by those authors, occurs in some but not all cases as part of the annihilation mechanism.