Issues in first-principles calculations for defects in semiconductors and oxides

Recent advances in density-functional theory (DFT) calculations of defect electronic properties in semiconductors and insulators are discussed. In particular, two issues are addressed: the band-gap underestimation of standard density-functional methods with its harmful consequences for the positioning of defect-related levels in the band-gap region, and the slow convergence of calculated defect properties when the periodic supercell approach is used. Systematic remedies for both of these deficiencies are now available, and are being implemented in the context of popular DFT codes. This should help in improving the parameter-free accuracy and thus the predictive power of the methods to enable unambiguous explanation of defect-related experimental observations. These include not only the various fingerprint spectroscopies for defects but also their thermochemistry and dynamics, i.e. the temperature-dependent concentration and diffusivities of defects under various doping conditions and in different stoichiometries.