A High-Throughput Wannierization Workflow for Predicting Functional Properties of Materials.

The discovery of functional materials for energy harvesting and electronic applications require accurate predictions of their optical and transport properties, which can be calculated from maximally localized Wannier functions. In this work, we develop a high-throughput Wannierization workflow that automates the construction of high-quality Wannier functions and the systematic evaluation of various functional properties. Starting from first-principles calculations, our workflow automatically selects the projection orbitals and the disentanglement and frozen energy windows, and validates the Wannier functions based on maximal spreads and band structure comparisons. The validated Wannier Hamiltonians are then used to compute tensorial optical and transport properties, including optical conductivity, shift current, electrical and thermal conductivity, Seebeck coefficient, and thermoelectric figure of merit zT. We tested this workflow on 15,000 materials, and around 90% of them can be Wannierized with the mean absolute error of the band energy difference within a few meV. We also generated a dataset of around 7,300 materials, representing the largest-to-date collection of tensorial optical and transport properties derived from the Wannier function method. This workflow provides a scalable pathway for constructing high-quality property databases and supports data-driven discovery of anisotropic optoelectronic and thermoelectric materials.