Accurate and efficient band-gap predictions for metal halide perovskites at finite temperature

TL;DR

This paper presents a fast and accurate computational method for predicting finite-temperature band gaps in halide perovskites, combining hybrid functional estimates with corrections for spin-orbit coupling and thermal vibrations, suitable for large-scale screening.

Contribution

The authors introduce a novel, efficient scheme that accurately predicts finite-temperature band gaps using minimal hybrid functional calculations and vibrational corrections, improving upon existing methods.

Findings

Achieved a mean absolute error of 0.17 eV compared to experimental data.

The method requires only a single hybrid functional calculation plus inexpensive corrections.

Suitable for large systems and high-throughput material screening.

Abstract

We develop a computationally efficient scheme to accurately determine finite-temperature band gaps. We here focus on materials belonging to the class ABX3 (A = Rb, Cs; B = Ge, Sn, Pb; and X = F, Cl, Br, I), which includes halide perovskites. First, an initial estimate of the band gap is provided for the ideal crystalline structure through the use of a range-separated hybrid functional, in which the parameters are determined nonempirically from the electron density and the high-frequency dielectric constant. Next, we consider two kinds of band-gap corrections to account for spin-orbit coupling and thermal vibrations including zero-point motions. In particular, the latter effect is accounted for through the special displacement method, which consists in using a single distorted configuration obtained from the vibrational frequencies and eigenmodes, thereby avoiding lengthy molecular dynamics. The sequential consideration of both corrections systematically improves the band gaps, reaching a mean absolute error of 0.17 eV with respect to experimental values. The computational efficiency of our scheme stems from the fact that only a single calculation at the hybrid-functional level is required and that it is sufficient to evaluate the corrections at the semilocal level of theory. Our scheme is particularly convenient for large-size systems and for the screening of large databases of materials.