High-throughput Materials-to-Device Discovery of Lead-free Double Perovskites Beyond Radiative-limit Metrics

Metal halide perovskite solar cells (PSCs) are leading candidates for next-generation photovoltaics, yet their instability and lead toxicity motivate environmentally benign alternatives. Lead-free double perovskites offer improved structural robustness but often exhibit weak near-band-edge absorption and modest efficiencies, making it crucial to identify compositions and phases that balance stability with photovoltaic performance. Here we assemble an initial enumerated phase-diverse library of 31,276 structure-composition entries spanning six symmetry-distinct cubic, tetragonal, and monoclinic configurations and establish a device-guided multiscale workflow for realistic performance evaluation. This framework combines symmetry-aware optical screening, finite-temperature structural-retention assessment, multiscale parameter transfer, 300 K thermal-displacement-averaged optical response, and self-consistent drift-diffusion device modeling. The workflow selects 23 candidates for comparative device-level evaluation under unified conditions. The device-level analysis shows that candidates favored by bandgap or Spectroscopic Limited Maximum Efficiency (SLME) do not necessarily retain their advantage as recombination losses increase. By varying the effective bimolecular recombination coefficient and the Shockley-Read-Hall carrier lifetime, we quantify PCE degradation and ranking evolution as a model-conditioned measure of device-level loss tolerance. The shortlisted candidates reach baseline predicted PCEs of up to 15.26%, while pronounced rank changes under increasing nonradiative recombination loss highlight the limitations of radiative-limit metrics alone. This work therefore recasts lead-free double-perovskite screening from identifying the highest radiative-limit efficiency to prioritizing materials that combine competitive device performance with robust loss tolerance and experimental relevance.