Zn-Assisted Mg Ion Transport in Spinel Oxide Cathodes: Insights From Neural Network Simulations

We investigated Mg insertion and ion transport in defect spinel Mg_xZnMnO₃ (0 ≤ x ≤ 1) using a neural network potential combined with molecular dynamics and genetic algorithm-based structure exploration. The diffusion coefficients showed a strong composition dependence: both stoichiometric spinel (x = 0.25) and rock-salt (x = 1.0) exhibited extremely low conductivity (<10^-10 cm² s^-1), whereas enhanced diffusivity appeared in the cation-deficient (x < 0.25) and cation-excess (0.25 < x < 1.0) regions, reaching 3.97 × 10^-7 cm² s^-1 at x = 0.50. Zn²⁺ consistently diffused faster than Mg²⁺, and both species migrated along the same 8a-16c-8a channels, suggesting possible concerted hopping interactions. Thermodynamic analyses revealed that the vacancy-driven spinel region (x < 0.25) is stabilized as a solid solution, whereas in 0.25 < x < 0.84 a biphasic spinel-rock-salt coexistence is favored. The precipitation of stoichiometric phases (x = 0.25, 1.0) may hinder Mg transport and reduce capacity, but the low energy-above-hull values suggest solid-solution pathways are kinetically accessible. These findings indicate that Zn plays a dual role by stabilizing tetrahedral sites and enhancing Mg mobility, providing design guidelines to achieve both high voltage and improved diffusivity for magnesium battery cathodes.