Nature of the Amorphous-Amorphous Interfaces in Solid-State Batteries Revealed Using Machine-Learned Interatomic Potentials

CH Wang and M Aykol and T Müller, CHEMISTRY OF MATERIALS, 35, 6346-6356 (2023).

DOI: 10.1021/acs.chemmater.3c00993

Non-crystalline solid materials have attracted growingattentionin energy storage for their desirable properties such as ionic conductivity,stability, and processability. However, compared to bulk crystallinematerials, fundamental understanding of these highly complex metastablesystems is hindered by the scale limitations of density functionaltheory (DFT) calculations and resolution limitations of experimentalmethods. To fill the knowledge gap and guide the rational design ofamorphous battery materials and interfaces, we present a moleculardynamics (MD) framework based on machine-learned interatomic potentialstrained on the fly to study the amorphous solid electrolyte Li3PS4 and its protective coating, amorphous Li3B11O18. The use of machine-learned potentialsallows us to simulate the materials at time and length scales thatare not accessible to DFT while maintaining a near-DFT level of accuracy.This approach allows us to calculate amorphization energies, amorphous-amorphousinterface energies, and the impact of the interface on lithium ionconductivity. This study demonstrates the promising role of activelylearned interatomic potentials in extending the application of abinitio modeling to more complex and realistic systems such as amorphousmaterials and interfaces.

Return to Publications page