Unraveling\nSuperionic Conductivity in Na<sub>2</sub>B<sub>12</sub>H<sub>12</sub>: Molecular Dynamics Study of Phase Transition,\nAnion Reorientation, and Ionic Conductivity via Machine Learning Force\nField

Metal\ncloso-hydroborates, particularly Na<sub>2</sub>B<sub>12</sub>H<sub>12</sub>, have emerged as promising solid-state electrolytes\nfor sodium-ion batteries, owing to their high ionic conductivity at\nelevated temperatures. Despite their potential, the mechanisms underpinning\ntheir superionic conduction and phase transitions remain incompletely\nunderstood. In this study, we develop a machine learning force field\n(ML-FF) for Na<sub>2</sub>B<sub>12</sub>H<sub>12</sub>, enabling large-scale\nmolecular dynamics simulations that capture the intricacies of its\nphase transition, anion reorientation dynamics, and ionic conductivity.\nOur simulations reveal a martensitic transformation from a monoclinic\nto a body-centered cubic (bcc) phase at ∼675 K, resulting in\na dramatic enhancement in Na<sup>+</sup> conductivity and a significant\ndecrease in the activation energy for anion reorientation. Additionally,\ncooling simulations indicate an incomplete reverse phase transition\ndue to rapid cooling, highlighting the complexities of phase stability\nin Na<sub>2</sub>B<sub>12</sub>H<sub>12</sub>. These findings emphasize\nthe critical role of anion reorientation and the cooperative movement\nof cations and anions in facilitating superionic conduction. By leveraging\nthe accuracy and efficiency of ML-FFs, this study provides unprecedented\natomistic insights into the mechanisms driving high-temperature conductivity\nin Na<sub>2</sub>B<sub>12</sub>H<sub>12</sub>, offering pathways for\nthe optimization of closo-hydroborates as next-generation solid-state\nelectrolytes.