A molecular understanding of the solvation and dynamics of ions under static electric fields is crucial for modeling a wide range of natural and technological processes. Yet, traditional simulation methods suffer from a trade-off that has to be made between accuracy and statistical convergence. To bridge this gap, herein, we extend our recently introduced perturbed neural network potential molecular dynamics (PNNP MD) approach to investigate the solvation structures and ionic transport mechanisms of electrified alkali cationic solutions. We obtain ionic conductivities for Li+, Na+, and Cs+ from the field dependence of the ionic current density in good agreement with experiment. Surprisingly, the migration mechanism is found to be strikingly different for the three ions, despite their similar ionic conductivities. While Li+ conducts predominantly through vehicular migration of a stable 4-fold coordinated ion at all field strengths, Cs+ conducts strictly through a structural diffusion mechanism, where 9-12 transient first shell water coordination bonds are continuously broken and reformed. Notably, aqueous Na+ emerges as a "Goldilocks" ion: its ion-water interactions are strong enough to maintain distinct 5-6-fold coordination shells at zero field (unlike Cs+) yet labile enough to be strongly perturbed by electric fields (unlike Li+). As a consequence, we observe an electric-field-induced transition from vehicular to structural ionic transport for Na+ that is accompanied by a marked increase in ionic current density. Our results imply that the conductance mechanism of ions with moderate ion-solvent interactions can be effectively tuned by external electric fields.
Joll et al. (2026) studied this question.