Intrinsic defect properties in GaN calculated byab initioand empirical potential methods

Density functional theory (DFT) has been used to investigate the formation, properties, and atomic configurations of vacancies, antisite defects and interstitials in GaN, and the DFT results are compared with those calculated by molecular dynamics simulations using two representative potentials. The DFT calculations reveal that the relaxation of vacancies is generally small, but the relaxation around antisite defects is large, especially for the Ga antisite that is not stable and converts to a ${\mathrm{N}}^{+}\ensuremath{-}\mathrm{N}⟨0001⟩$ split interstitial plus a Ga vacancy at the original site. The N interstitials, starting from all possible sites, eventually relax into a ${\mathrm{N}}^{+}\ensuremath{-}\mathrm{N}⟨11\overline{2}0⟩$ split interstitial. In the case of Ga interstitials, the most stable configuration is a Ga octahedral interstitial, but the energy difference among all the interstitials is small. The ${\mathrm{Ga}}^{+}\ensuremath{-}\mathrm{Ga}⟨11\overline{2}0⟩$ split interstitial can bridge the gap between nonbonded Ga atoms, thereby leading to a chain of four Ga atoms along the $⟨11\overline{2}0⟩$ direction in GaN. The formation energies of vacancies and antisite defects obtained using the Stillinger-Weber (SW) potential are in reasonable agreement with those obtained by DFT calculations, whereas the Tersoff-Brenner (TB) potential better describes the behavior of N interstitials. In the case of Ga interstitials, the most stable configuration predicted by the TB model is a ${\mathrm{Ga}}^{+}\ensuremath{-}\mathrm{N}⟨11\overline{2}0⟩$ split interstitial; while for the SW model the Ga tetrahedral configuration is more stable, which is in contrast to DFT results.