AI-Guided Design and Discovery of Silicon-Based Anode Materials for Lithium-Ion Batteries

David Garcia Valcarce, Eduardo Abenza, Roberto Gómez-Espinosa Martín

Keywords: Silicon anodes, Lithium-ion batteries, Graph neural networks, Generative models, Property prediction, Li-ion transport, Materials discovery, AI-driven optimization

Abstract:

The development of high-capacity anodes remains a central challenge for advancing
lithium-ion battery (LIB) technology. Silicon (Si) offers an exceptional theoretical
capacity but suffers from severe volume expansion, structural degradation, and
limited Li-ion mobility after cycling. Here, we present an AI-driven framework for
the analysis, generation, and optimization of silicon-based anode materials with
enhanced Li-ion transport and controlled volume variation during lithiation and
delithiation. Our approach integrates large-scale materials databases with state-
of-the-art machine learning to: (i) analyze Li-ion migration pathways in known
compounds, (ii) generate lithiation states from charged configurations, and (iii)
design novel Si-based materials with enlarged Li-ion migration channels while
tracking their volume changes across lithiation levels. By combining graph-based
generative models with efficient property predictors, we accelerate the discovery
of candidate anodes through database-guided screening. The results highlight
the potential of AI to identify next-generation silicon-based anode materials with
improved stability and performance, providing valuable guidance for experimental
development.

Submission Track: Paper Track (Full Paper)

Submission Category: AI-Guided Design

Institution Location: Madrid, Spain

AI4Mat Journal Track: Yes

AI4Mat RLSF: Yes

Submission Number: 95