Quasiparticle band structures of CuCl, CuBr, AgCl, and AgBr: The extreme case

Abstract

We present a systematic study of the quasiparticle band structures of transition-metal halides CuCl, CuBr, AgCl, and AgBr. We show that GW calculations for cuprous halides are significantly more challenging computationally than ZnO, a much-discussed extreme case. The local-density approximation (LDA) within density functional theory severely underestimates the band gaps of CuCl and CuBr due to the inaccurate treatment of the semicore $d$ electrons. As a result, many-body perturbation calculations within the ${\mathrm{G}}^{0}{\mathrm{W}}^{0}$ approach fail to give accurate quasiparticle properties starting from the LDA mean-field solution. The $\mathrm{LDA}+U$ method (with the screened Coulomb and exchange parameters calculated using a constrained random-phase-approximation approach), on the other hand, provides a much better starting point for subsequent ${\mathrm{G}}^{0}{\mathrm{W}}^{0}$ calculations. When properly converged, the ${\mathrm{G}}^{0}{\mathrm{W}}^{0}/\mathrm{LDA}+U$ approach is able to reproduce the experimental minimum band gaps of all four compounds to within 0.1 eV. These results, however, can be achieved only by applying extremely high cutoff parameters, which would be very difficult without using our recently developed accelerated GW approach. Our work demonstrates the applicability and accuracy of the ${\mathrm{G}}^{0}{\mathrm{W}}^{0}/\mathrm{LDA}+U$ method in predicting the quasiparticle band structure of these materials and other systems involving localized semicore states.