Researchers at the University of Colorado Boulder's Renewable and Sustainable Energy Institute, led by Alex Zunger, working with the University of Sao Paulo, have shown that two simple structural measurements can predict how stable a halide perovskite alloy will be and what band gap it will have, without running a full quantum simulation for every composition.
Alloying is central to tuning perovskite light absorption and stability, but predicting which combinations of halogens, alkali metals and other elements will work has typically required slow, computationally expensive density-functional theory (DFT) calculations run one composition at a time. The team instead looked for structural descriptors that could stand in for those calculations. Running systematic DFT calculations across alloy families with substitutions at the halogen (X-site), alkali (A-site) and group-IV (B-site) positions, they found that the excess in cation-anion-cation (B-X-B') bond angles - a measure of how much the perovskite's octahedra are distorted - closely tracks the alloy's mixing enthalpy, an indicator of thermodynamic stability. Separately, the average cation-anion (B-X) bond length emerged as a reliable predictor of how the band gap shifts across the alloy series.
The correlations held consistently as the team swapped halogens, alkali metals and other elements across a wide range of perovskite alloy compositions, comparing the simple geometric measurements against the full DFT calculations they were meant to approximate. Because both descriptors can be read directly from a structure's geometry rather than computed from scratch for each composition, the approach offers a much faster way to screen candidate alloys for the two properties that matter most for solar applications - durability and efficient light-to-electricity conversion - before committing to expensive full simulations.
