Probing Interfacial Degradation in Solid-State Batteries Using Machine Learning Force Fields

Abstract

Listen

Translate article

Solid-state batteries (SSBs) are next-generation energy storage technologies with improved safety and potentially higher energy densities compared to conventional Li-ion batteries, which is enabled by using fast ion-conducting solid electrolytes (SEs). However, practical applications of SSBs are hindered by the electro-chemo-mechanical instabilities at interfaces within the SEs (grain boundaries (GBs)) and between SEs and the electrodes, which deteriorates Li transport and the chemical and mechanical integrity of the cell. To mitigate, fundamental understanding of the intrinsic physico-chemical properties at interfaces is required. To this end, we directly probe atomic structures of the interfaces and GBs by performing large-scale atomistic simulations, enabled by validated machine-learning force fields (MLFFs) and resolve the structure-property relationship at complex interfaces that governs Li-ion transport and stability of SSBs.In this talk, we will discuss the characteristics of garnet Li7La3Zr2O12 (LLZO) SE/LiCoO2 (LCO) cathode interfaces as well as the internal interfaces within the LLZO (GBs). It is observed from our simulations that the propensities for interfacial degradation strongly depend on the surface chemistry of LLZO and LCO. Doping LLZO is found to have a segregation effect at the LLZL GBs, here we will discuss its implication towards secondary phase formation and Li transport kinetics. At last, we will address the impact of interfaces on micro-crack propagation behavior in the LLZO-LCO system. In summary, our results reveal how atomic details of the dynamically evolving interfaces dictate the performance of SSBs, and provide guidance for processing and interface design to achieve desired performance .This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract number DEAC52-07NA27344. Authors acknowledge funding support from the Vehicle Technologies Office, Office of Energy Efficiency and Renewable Energy, U.S. Department of Energy and computational resource support from the Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program. This research used resources of the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC02-06CH11357.