On the problem of a suitable definition of the cluster in embedded-cluster treatments of defects in crystals

Abstract
The perturbed-cluster embedding scheme [J. Chem. Phys. 92, 7448 (1990)] adopts a self-consistent procedure where, at each cycle, a molecular cluster solution is calculated, and subsequently corrected for realizing its coupling to the crystalline environment. In the present work, it is shown that a proper design of the cluster is of critical importance for the scheme to work satisfactorily. Indications are given for defining clusters that are at the same time "self-satisfied," and optimally fitted to the environment, so that correction terms are minimized. The procedure is tested in the case of defects in crystalline silicon.