Accelerated Screening of Halide Double Perovskites via Hybrid Density Functional Theory and Machine Learning for Thermoelectric Energy Conversion

A comprehensive first‐principles and machine learning study is conducted on 102 halide double perovskites to identify promising candidates for thermoelectric applications. The HSE06 hybrid functional within the Quantum ATK framework is used to accurately determine electronic structures, bandgaps, and total and partial densities of states. Boltzmann transport theory is applied to figure out important thermoelectric parameters, such as the Seebeck coefficient, electrical conductivity, and ZT values over a wide range of temperatures. Supervised machine learning models are trained on density functional theory (DFT)‐derived descriptors to speed up the discovery of new materials. These models demonstrate high predictive accuracy for thermoelectric performance across different chemical spaces. A detailed analysis of the electronic band structures and orbital contributions is carried out for Rb2GeI6, Rb2PbI6, Cs2SnBr6, and In2PtCl6, some of the best‐performing compounds. A wide range of behaviors is observed, including metallic, degenerate, and wide‐bandgap semiconducting, which correlate with distinct transport properties. This unified method shows how using accurate DFT, transport theory, and machine learning together can help find new materials with specific functions. This will lead to the development of next‐generation thermoelectric technologies based on environmentally friendly halide perovskites.