Machine-Learning-Based Study of Ionic Diffusion and Lattice Dynamics in K₂Se₂Te and Related K-Based Superionic Materials

Although K₂Se₃ and K₂Te₃ were discovered in the last century, most previous studies have focused on the battery performance of K₂Te₃. In contrast, the single-crystal compound K₂Se₂Te was only accidentally discovered in 2021, and its physical properties remain largely unexplored both theoretically and experimentally. While extensive studies have been devoted to phonon transport in Ag-based and Cu-based superionic materials and more recently in certain K-based systems—the intrinsic relationship between ionic transport and thermal properties remains elusive. In this work, we employ molecular dynamics simulations based on machine-learning interatomic potentials to investigate ionic diffusion and lattice dynamics in K₂Se₂Te and related compounds with DFT-level accuracy. Our analysis reveals that the liquid-like motion of potassium atoms significantly enhances the coherent contribution to the thermal conductivity. This study provides new insight into the complex atomic dynamics governing transport phenomena in superionic conductors.