The description of
the band gap of halide perovskites at the level
of density functional theory (DFT) has been subject of several studies
but still presents significant problems and deviations from experimental
values. Various approaches have been proposed, including the use of
system-specific hybrid functionals with a variable amount of exact
exchange or the explicit inclusion of spin–orbit coupling (SOC)
effects. In this work, we present a pragmatic recipe to compute the
band gap of halide perovskites with a minimum average error. The recipe
is tested on a set of 36 halide perovskites of the type ABX3 [A = Cs, methyl-ammonium (MA), and formamidinium (FA); B = Ge, Sn,
and Pb; and X = Cl, Br, and I] for which experimental estimates of
the band gap have been reported in the literature. Upon assessment
of the accuracy of commonly used DFT functionals and the analysis
of their performances based on error and statistical analysis, we
suggest a strategy to compute band gaps in halide perovskites with
a single functional. This is based on the use of the hybrid HSE06
functional where SOC is included exclusively for Pb-containing compounds.
The results are rationalized in terms of the materials’ chemical
nature and are corroborated by the prediction of their expected efficiencies
in solar cells. The calculated efficiencies from band gaps obtained
with the proposed approach closely follow the experimental trend,
demonstrating the importance of adopting a reliable but material-independent
computational strategy to screen new halide perovskite materials for
solar energy conversion.