At-Scale Data-Driven Exploration of High-Voltage Cathode-Active Materials for Sodium Batteries

Abstract

Sodium-ion batteries (SIBs) share similar electrochemistry with Li but offer several advantages, including high abundance in nature and low cost, as well as suitability for fast charging due to a Na-ion mobility higher than that of Li. The development of high-voltage SIBs heavily relies on the discovery of novel, robust cathode-active materials (CAMs). All-inorganic materials represent the most mature and practical choice as CAMs for next-generation SIBs; however, their family spans a vast and chemically diverse space. In this work, we present a large-scale, chemically validated database of stable materials for SIB cathode discovery, curated from four major databases: Materials Project, AFLOW, OQMD, and GNoME. Generalizable and transferable descriptor-based machine learning (ML) models are developed based on a dataset of charged-only structures rather than charged/discharged pairs. Using a committee of the top four trained ML models, average voltage and specific capacity are predicted as target properties. Finally, a subset of top-ranked candidate CAMs is validated through explicit, high-throughput first-principles calculations of voltage profiles, phase stability, structural robustness upon sodiation/desodiation, and electronic properties. Together, this integrated data curation, ML ranking and predictions, and first-principles validation strategy establishes a scalable and transferable framework for accelerating the discovery of stable, high-voltage CAMs for SIBs and beyond.