Predicting Reactivity and Passivation of Solid-State Battery Interfaces.

Area of Science:

• Materials Science
• Computational Chemistry
• Electrochemistry

Background:

• Predicting solid-state electrolyte and lithium metal anode interface reactivity is crucial for solid-state battery development.
• Traditional methods like density functional theory (DFT) energetics and thermodynamic convex hull calculations have limitations in capturing kinetic and dynamic interfacial behaviors.
• Ab initio molecular dynamics (AIMD) simulations provide detailed insights but are computationally expensive and time-consuming for large-scale screening.

Purpose of the Study:

• To develop a computationally inexpensive, data-driven model for predicting the reactivity of interfaces between solid-state electrolytes and lithium metal anodes.
• To leverage machine learning on AIMD simulation data to capture both kinetic and thermodynamic factors governing interfacial stability.
• To accelerate the discovery and screening of novel, stable solid-state electrolyte materials for lithium-ion batteries.

Main Methods:

• Training machine learning models on atomistic structure information and AIMD simulation data of solid electrolyte-Li metal interfaces.
• Utilizing the trained models to predict interfacial reactivity for thousands of candidate materials rapidly.
• Comparing model predictions with traditional thermodynamic approaches to identify discrepancies and novel stable materials.

Main Results:

• Identification of over 300 new chemically stable solid electrolytes and over 780 passivating solid electrolytes predicted to be thermodynamically unfavored.
• Demonstration that purely thermodynamic approaches may mislabel many potential solid-state electrolyte candidates as unstable.
• Highlighting two borate materials (LiB13C2 and LiB12PC) predicted by the model and confirmed by AIMD to be highly conductive and chemically stable with Li.
• Suggesting the pool of promising, Li-stable solid-state electrolyte materials is significantly larger than previously estimated.

Conclusions:

• The developed data-driven model offers a computationally efficient and accurate method for predicting solid-state electrolyte-Li metal interfacial reactivity.
• This approach significantly accelerates materials discovery and screening, overcoming limitations of traditional computational methods.
• The findings expand the landscape of potential solid-state electrolyte materials, paving the way for next-generation solid-state batteries.