Wannier-driven search for novel BCS superconductors

A computational search for novel Bardeen-Cooper-Schrieffer superconductors is challenging due to the high cost of calculating the electron-phonon couplings that determine conventional superconductivity. The EPW (Electron-Phonon using Wannier functions) code can perform precise interpolations for the needed electronic and phonon dispersions at a much reduced cost, but a large-scale screening would still require manual construction of the Wannier functions. Here, we combine recent developments in automated Wannierization with the AiiDA computational infrastructure to enable a high-throughput search for novel superconductors, starting from the Materials Cloud 3-dimensional structure database (MC3D). We first compute the Allen-Dynes critical temperature for ~6,000 non-magnetic metals, using a direct coarse sampling of the electron-phonon couplings. For the most promising materials, we run a subsequent EPW workflow to obtain a high-quality dataset for over 300 dynamically stable structures, including spectral functions and Migdal-Eliashberg critical temperatures, from which we identify several novel candidates that can be promising for further experimental studies.