Accurate band gaps from exchange potentials designed from a cuspless hydrogen density-based exchange hole model

The explicit forms of exchange-correlation (XC) potentials, which are not functional derivatives of any XC energy functional, are reasonably efficient in predicting the band gap of materials. The most successful example in this genre is the MBJ [F. Tran et al., Phys. Rev. Lett., 2009 102, 226401] exchange potential, which is based on the asymptotically correct Becke-Roussel (BR) exchange potential. We employ the cuspless hydrogen density and corresponding exchange hole to construct a BR like potential. The modified BR potential is again utilized in the framework of MBJ for band gap calculations. Also, we employ a Laplacian free model of the exchange hole in the framework of MBJ. These methods are analyzed by calculating band gaps of various test sets containing narrow, intermediate, and wide bandgap materials. Besides, we apply these potentials to eighteen ternary semiconductors with a chalcopyrite structure, exciting materials for photovoltaic applications. By comparing them with MBJ, we find that the band gaps obtained using the new potentials are not uniformly larger values than the MBJ potential in all cases. But, in many instances where MBJ overestimates the gap, the new potentials' band gaps are comparatively smaller and closer to the experimental ones. We also show that these potentials can correctly predict the band structure of three-dimensional topological insulators.

Automated summary

The study focuses on modifying exchange-correlation potentials to improve the prediction accuracy of material band gaps, indicating that the new potentials can better fit experimental values in some cases, particularly showing good performance in predicting the band structures of three-dimensional topological insulators.