High-throughput study of electrical conductivity in ordered metals

Electrical conductivity is a key property for the design of new metallic conductors, but its accurate first-principles prediction is computationally expensive, which has limited its exploration in high-throughput materials screening. In the associated work, we combine machine-learning surrogate models with density-functional theory (DFT) and density-functional perturbation theory (DFPT) to screen approximately 2.8 million inorganic compounds from the Alexandria database for metallic electrical conductivity, using the electron–phonon coupling constant λ from Eliashberg theory as a proxy for scattering strength. From this pool, only the compounds predicted to exceed a high-conductivity threshold were carried forward for explicit electron–phonon coupling calculations and Boltzmann transport theory using the EPW code, solving the iterative linearized Boltzmann transport equation (BTE), narrowing the candidates down to 52 top-performing compounds. This dataset provides the DFT, DFPT, and EPW input and output files for these 52 candidates, together with 24 elemental benchmark metals and 3 spin-orbit-coupling reruns computed for validation.