Breaking interdisciplinary barriers in solid-state battery research: BatteryAgent for multifaceted analysis

Abstract

Energy-dense All-Solid-State Batteries (ASSBs) require simultaneous optimization from atomic-scale material properties to cell-level manufacturing constraints—a challenge beyond conventional, single-domain approaches. Here, we present BatteryAgent, an end-to-end Large Language Model (LLM)-Agent framework that autonomously orchestrates interdisciplinary ASSB analysis through three key innovations: (1) a collaborative expert team architecture for integrated cross-disciplinary insights, eliminating traditional analytical silos; (2) modular computational tools enabling experts to dynamically acquire knowledge, perform numerical analysis, and visualize data on demand; and (3) adaptive coordination mechanisms ensuring consistency across scales from materials to battery cells. Demonstrated through comprehensive ASSB optimization, the framework quantitatively analyzes crucial material-level trade-offs: sulfide electrolytes exhibit superior ionic conductivity but stability limitations, oxides offer superior stability at the expense of higher density and resistivity, while halides provide balanced intermediate performance. Cell-level modeling further validates that ultrathin (<30 μm) sulfide electrolytes enable battery designs approaching ∼500 Wh kg⁻¹ configurations, a critical milestone for advanced energy applications. BatteryAgent thus accelerates ASSB development and provides a robust blueprint for AI-driven interdisciplinary research, broadly applicable to complex, multi-property optimization challenges across energy storage domains.