Manybody GW calculation of the oxygen vacancy in ZnO

TL;DR

This study uses many-body GW calculations to analyze the oxygen vacancy in ZnO, revealing insights into defect state shifts, finite size corrections, and the robustness of GW results over different DFT functionals.

Contribution

It demonstrates the application of GW calculations to defect states in ZnO, highlighting the importance of self-interaction correction and finite size effects, and confirms the stability of GW results across different functionals.

Findings

GW shifts increase splitting between defect states

Charged defect energies need finite size corrections

GW results are consistent across DFT functionals

Abstract

Density functional theory (DFT) calculations of defect levels in semiconductors based on approximate functionals are subject to considerable uncertainties, in particular due to inaccurate band gap energies. Testing previous correction methods by many-body GW calculations for the O vacancy in ZnO, we find that: (i) The GW quasi-particle shifts of the VO defect states increase the spitting between occupied and unoccupied states due to self-interaction correction, and do not reflect the conduction versus valence band character. (ii) The GW quasi-particle energies of charged defect states require important corrections for supercell finite size effects. (iii) The GW results are robust with respect to the choice of the underlying DFT or hybrid-DFT functional, and the (2+/0) donor transition lies below mid-gap, close to our previous prediction employing rigid band edge shifts.