Assessment of the GLLB-SC potential for solid-state properties and attempts for improvement

TL;DR

This paper evaluates the GLLB-SC potential's effectiveness for predicting solid-state properties, compares it with other methods like mBJ, and explores potential improvements to enhance its accuracy for various materials.

Contribution

It provides a comprehensive comparison of GLLB-SC with other methods and investigates modifications to improve its predictive performance.

Findings

GLLB-SC performs similarly to hybrid functionals for band gaps.

GLLB-SC overestimates magnetic moments in metals.

Variants of GLLB-SC are tested for improved accuracy.

Abstract

Based on the work of Gritsenko et al. (GLLB) [Phys. Rev. A 51, 1944 (1995)], the method of Kuisma et al. [Phys. Rev. B 82, 115106 (2010)] to calculate the band gap in solids was shown to be much more accurate than the common local density approximation (LDA) and generalized gradient approximation (GGA). The main feature of the GLLB-SC potential (SC stands for solid and correlation) is to lead to a nonzero derivative discontinuity that can be conveniently calculated and then added to the Kohn-Sham band gap for a comparison with the experimental band gap. In this work, a thorough comparison of GLLB-SC with other methods, e.g., the modified Becke-Johnson (mBJ) potential [F. Tran and P. Blaha, Phys. Rev. Lett. 102, 226401 (2009)], for electronic, magnetic, and density-related properties is presented. It is shown that for the band gap, GLLB-SC does not perform as well as mBJ for systems with a small band gap and strongly correlated systems, but is on average of similar accuracy as hybrid functionals. The results on itinerant metals indicate that GLLB-SC overestimates significantly the magnetic moment (much more than mBJ does), but leads to excellent results for the electric field gradient, for which mBJ is in general not recommended. In the aim of improving the results, variants of the GLLB-SC potential are also tested.