Simulated migration and reaction of correlated point defects

Abstract

Abstract Monte Carlo techniques are applied to defect migration in the vicinity of a fixed reaction volume. Intended to simulate “stage ID ” annealing of correlated interstitials and vacancies in fcc metals, large reaction spheres of up to 720 atomic volume are employed, with mobile defects in the (100) split configuration symmetrically diffusing from maximum distances of more than twice the capture radius r 0 and for a maximum of 100 jumps. This discrete approach and continuum theory are judged to be equally valid. Random walk recovery probabilities are not a smooth function of initial distance r, but fQr large reaction volumes agree with continuum theory nearly as well for r ∼ r 0 as for r > 2r 0. Absolute agreement is improved as the “dumbbell” separation of the split defect is increased. Recovery due to an extended distribution of defects is obtained by weighting individual walks from distances < 3r 0: the resulting composite annealing curves disallow observation of structure and compare favorably with resistivity data. The number of symmetric random walk jumps N required to reach the maximum rate of correlated recovery-the “ID peak”-is found for a wide range of initial distributions, the best parameter estimates for Cu giving N = 50 ± 15. This value and resistivity data give N for the uncorrelated recovery peak in agreement with theory. One dimensional migration is excluded by these results.