Electronic structure and identification of deep defects in GaP

We report a theoretical study of the Ga vacancy, the P antisite defect (i.e., a P atom occupying a normally Ga site), and carbon impurities at Ga sites. The analysis employs a modified version of the self-consistent Green's-function method that was previously used to describe deep centers in Si. In all three cases, the number, orbital content, and relative energy positions of the localized states can be understood in terms of simple models, as in the case of corresponding centers in Si. In particular, we find that, in all cases, the gap state consists primarily of atomiclike $p$ orbitals centered on the four nearest neighbors, revealing that deviations from $s{p}^3$-hybrid character are possible even without any relaxation of the surrounding lattice. In the case of the antisite defect, we find an $A_1$ bound state in the gap which is consistent with the EPR spectra that led to the identification of the defect. The calculated ionization energies are in agreement with optical data and more recent deep-level-transient-spectroscopy data. In the case of the Ga vacancy, we find a $T_2$ level in the gap above an $A_1$ resonance inside the valence bands, and in the case of carbon, we find a deep $A_1$ level in the gap. These results contradict key assumptions that led to the identification of the NRL-1 EPR signal as the $V^{2-}$ charge state of the Ga vacancy. More recent experiments redetermined some of the parameters of NRL-1 and reassigned the center to the neutral state of the Ga vacancy. Here, we discuss the interplay between experiment and theory that led to the new identification and describe existing difficulties that limit our ability to construct a detailed model which is consistent with all experiments and with theory.