Data-Driven Knowledge Discovery Reveals Quantitative Electrolyte Design Rules for Anode-Free Sodium Metal Batteries

Sodium-based batteries are promising for large-scale energy storage due to the abundance and low cost of sodium, yet their practical energy density remains constrained. Anode-free configurations enhance energy density, but they require electrolytes that enable highly reversible Na plating/stripping with exceptional Coulombic efficiency (CE). Here, by establishing an interpretable machine learning-driven knowledge discovery framework, quantifiable design rules leading to high CE were formulated within our data set. We reveal that reducing solvent oxygen content and increasing nonpolar surface area promote an anion-dominated solvation structure by improving the solvent steric hindrance effect, resulting in NaF-rich solid electrolyte interphase and suppressed inactive sodium formation. Guided by these insights, an optimal electrolyte based on asymmetric diethylene glycol butyl methyl ether with NaPF6 was developed. It achieved an average CE of 99.9% over 800 cycles and enabled an Ah-level pouch cell with an energy density exceeding 230 Wh kg-1. This work bridges data science with solvation and interphase chemistry, establishing quantitative principles for the development of high-performance electrolytes toward sustainable energy storage.