Modeling the electrical effects of radiation damage in semiconductor devices requires a detailed description of the properties of point defects generated during and subsequent to irradiation. Such modeling requires physical parameters, such as defect electronic levels, to describe carrier recombination. Density functional theory (DFT) is the method of choice for first-principles simulations of defects. However, DFT typically hugely underestimates the fundamental band gap in semiconductors, and the band gap is the energy scale of interest for defect levels. Moreover, boundary conditions in the supercell approximation used in DFT calculations of defects also can inject large errors and uncertainties. I describe a new, more rigorous methodology for supercell calculations, implemented in the SeqQuest DFT code, that incorporates a proper treatment of electrostatic boundary conditions, locates a fixed chemical potential for the net defect electron charge, includes the bulk dielectric response, and creates a robust computational model of isolated defects. Using this methodology, the computed DFT defect level spectrum for a wide variety of Si defects spans the experimental Si gap, i.e., exhibits no band gap problem, and the DFT results agree remarkably well with experiment for those values that are experimentally known. The new scheme adds rigor to computing defect properties, and has important implications for density functional theory development.
Armstrong 1109, Purdue University, West Lafayette