Proceedings of nanoGe Fall Meeting 2021 (NFM21)
DOI: https://doi.org/10.29363/nanoge.nfm.2021.235
Publication date: 23rd September 2021
The fundamental band gap is a decisive observable when characterizing energy converting materials. Being able to predict it reliably at affordable numerical cost is a prerequisite for the large-scale computational search for new materials. Density Functional Theory (DFT) is often the method of choice for calculating a material's electronic structure. The band gap, however, has been a notoriously difficult observable for DFT. Inexpensive exchange-correlation approximations such as LDA and GGAs systematically underestimate the gap because they lack a derivative discontinuity. Hybrid functionals often yield more realistic results, but at a sharply increased computational cost. In any case, many-body perturbation theory methods, such as the GW approach, have often been required for a reliable prediction of the gap, adding a further layer of methodological complexity and cost. We here show that band gaps can be predicted accurately within DFT, without any empirical parameters, from the TASK meta-generalized gradient approximation (meta-GGA) functional. The latter has been derived from first principles by fulfilling known constraints and by taking into account the derivative discontinuity. Unlike some other meta-GGAs, the TASK functional leads to well converging calculations and predicts band gaps with reasonable accuracy for a large range of systems at the same order of computational expense as a GGA calculation.