Data-driven molecular design rules for electrolytes in CO₂ electroreduction

CO₂ electroreduction offers a promising route to convert CO₂ into valuable chemicals and fuels. CO₂ electroreduction research has predominantly focused on catalyst innovation, yet the electrolyte, equally critical for governing CO₂ solubility and ion transport, remains comparatively underexplored. Here, we introduce a multiscale informatics framework integrating density functional theory, conductor-like screening model for realistic solvents (COSMO-RS) thermodynamics, molecular dynamics (MD), and machine learning to accelerate electrolyte discovery. The framework focuses on bulk-phase electrolyte properties that govern solvent stability, CO₂ solubility, and transport, and thus electrolyte performance in CO₂ electroreduction. Hierarchical screening of 1.3 million organic molecules identified six high-performing candidates, validated through MD simulations, and benchmarked against twenty experimentally tested electrolytes. All data, descriptors, and pretrained models are consolidated in the open-access COSMIC (CO₂ solvent materials informatics collection) database. Our analysis reveals how molecular polarity, protic character, and electronic structure govern CO₂ solubility and viscosity, providing transferable molecular design rules for next-generation CO₂ electroreduction electrolytes.