Simple models for band gap estimations in halide perovskites

<jats:p>Halide perovskites (HP) are a well-known class of compounds with compositions based on the ABX3 stoichiometry. The diversity of HP compositions leads to an enviable range of optoelectronic applications, such as in solar cells, LEDs, lasers, photocatalysts, power electronics, and communications. To be used as such, band gap energy (𝐸𝑔) values emerge as a crucial property. The specialized community dedicated to control and predict 𝐸𝑔 from experimental and theoretical results. While data analysis and machine learning (ML) methods can be and have been used to understand composition-property relations, current literature lacks simple, generic, and computationally inexpensive tools to estimate BG values for generic HP compositions. Here, we propose two empirical models derived from existing literature data. Using above 400 entries, we started from elemental properties of the ABX3 components, namely, their ionic radii and B and X electronegativities, and converted them to their effective descriptors tolerance factor, octahedral factor, and ionic character. These descriptors were correlated with the 𝐸𝑔 from which the empirical models were derived using statistical and ML-derived regressions. Despite their simplicity, the models can estimate the band gap energies of HPs from easily determined information, with errors on the order of only 0.12 eV for 𝐸𝑔 in the typical range from 1 to 5 eV. Potentialities and limitations of these empirical models are also discussed.</jats:p>