Machine Learning Guided the Discovery of Superionic Delafossite AgFeO₂

The fundamental principle of solid-state electrolytes (SSEs) is predicated on the presence of superionic conductors (SICs), which are distinguished by the diffusion of liquid ions within the solid crystal lattice. These characteristics demonstrate considerable potential for achieving safe, high energy density, and reversible electrochemical energy storage in batteries. In this work, we employ molecular dynamics simulations based on machine learning force fields (MLFF) to elucidate the diffusion behavior of Ag⁺ through the Fe-O solid lattice in delafossite AgFeO₂ above 800 K. The analysis of atomic trajectories, mean square displacements (MSD), and radial distribution functions (RDF) indicates that Ag⁺ ions migrate primarily through the concerted mechanism within the structure. Madelung energy analysis revealed that the interaction between Fe and O is more intense than that between Ag and O. It is notable that the introduction of Ag vacancies resulted in a reduction of the superionic transition temperature from 800 to 600 K, thus demonstrating the impact of structural defects on ionic behavior. The present study opens up avenues for targeted materials in solid-state electrolytes and provides deeper insights into delafossite.