| We investigate the electronic structure and optical properties of semiconductor interfaces and defects by a many-body perturbation theory, determining the self-energy of quasiparticles in the GW approximation (GWA) [W. G. Aulbur, L. Jonsson and J. W. Wilkins, Solid State Physics 54, 1 (1999)]. We describe two specific implementations, based on different basis sets: a plane wave method, and a Projector Augmented Wave (PAW) method. We also present a new parallelization procedure that we have implemented for computationally demanding defect calculations.; Using the plane wave based method, we obtain conduction and valence band offsets for hexagonal and cubic III-V semiconductor interfaces. For the III-N interfaces, we present the first quasiparticle calculation of the band offsets, including important effects of interface strain, polarization fields, and semicore states. We report significant trends, such as the dependence of the valence and conduction band offsets on the in-plane lattice constant, as well as on the choice of pseudopotential used. We find that large macroscopic polarization fields exist in the bulk regions of the polar (0001) superlattice, and a charge density decomposition in monopole and dipole terms is necessary in order to determine the electrostatic interface dipole. For the experimentally relevant AlN-GaN (0001) wurtzite interface, we find type-I offsets ranging from VBO = 1.3 eV and CBO = 1.5 eV for the in-plane lattice constant of a 6H-SiC (0001) substrate to VBO = 0.8 eV and CBO = 1.8 eV for the in-plane lattice constant of a GaN (0001) substrate. This sensitivity may explain the range of experimental results for systems whose in-plane lattice constants could not be directly measured.; Using the parallel version of the code, we determine the levels inside the band gap of silicon introduced by planar {lcub}311{rcub} interstitial defects. We find that the level introduced by the planar /II/ defect has a quasiparticle energy of Ev + 0.56 eV, in excellent agreement with measurements obtained by deep-level transient spectroscopy (DLTS). This excited-state calculation confirms previous total-energy results [J. Kim, F. Kirchhoff, J. W. Wilkins, and F. S. Khan, Phys. Rev. Lett. 84, 503 (2000)] suggesting that /II/ defects act as building blocks of larger Si planar {lcub}311{rcub} defects. |
