Machine‐Learning‐Assisted Elucidation of Ion Transport Mechanism in Heterogeneous Interphases Formed at the Electrolyte–Electrode Interfaces

Abstract

The structural and compositional heterogeneity at electrolyte/electrode interfaces critically determines the functionality, performance, and stability of modern Li‐ion batteries. However, atomic‐scale characterization remains challenging due to the high degree of disorder, spatial confinement, and dynamic evolution of these interfaces, leaving their role in Li‐ion transport poorly understood. LiF and are among the most prevalent interfacial components, forming either intentionally or through electrode contamination and electrolyte decomposition during cycling. Here, we employ large‐scale molecular dynamics simulations accelerated by machine‐learning interatomic potentials to investigate how local heterogeneity at LiF/ interfaces affects Li transport. Our results show that highly mixed Li–F– chemistries at the interface enable rapid Li conduction and are both energetically stable and dynamically robust. These findings suggest that incorporating an appropriate fraction of can disrupt homogeneously LiF‐covered interlayers, thereby enhancing Li‐ion transport kinetics while improving the mechanical robustness of the interlayer structure.