Experiments were performed on samples with various initial B/P ratios, which had been exposed to neutron fluences, followed by long-term room temperature annealing. The donor removal rate of P was measured, and the acceptor removal rate of B was evaluated. Upon extending a previous model, it was proposed that a slow return of the damage-induced B interstitial back to its original substitutional site was responsible for an acceptor increase that was observed during long-term annealing. This model, which involved consideration only of the behavior of known shallow donors and acceptors in the starting material, permitted the present data to be fitted without any need to invoke the existence of unidentified deep levels.

R.Wunstorf, W.M.Bugg, J.Walter, F.W.Garber, D.Larson: Nuclear Instruments and Methods in Physics Research A, 1996, 377[2-3], 228-33