Charge corrections for supercell calculations of defects in semiconductors

We investigate the validity of the Makov-Payne correction if applied to supercell calculations for charged defects by comparing the results of a supercell calculations with those obtained using a Green's function approach. For several defects in 3C-SiC the observed energy differences between both calculations are not bridged by the corrections of Makov-Payne. Whereas a correction proportional to the defect charge squared is predicted, the observed differences between supercell and Green's function approaches depend almost linearly on the charge. Possible reasons for this discrepancy are discussed