The low specific capacity of sodium-ion batteries (SIBs) limits their practical use in high-capacity energy storage devices. Recently, cumulative cationic and anionic redox reactions have been identified as promising approaches to achieving high capacity in SIBs. However, the excess oxidation of labile oxygen during anionic redox leads to structural degradation and voltage hysteresis in Na-rich cathode materials. In this work, we employ first-principles density functional theory (DFT) calculations to elucidate the contributions of cationic and anionic redox reactions in a prototype Na-rich cathode material (Na 2 RuO 3 ) across different voltage windows. Additionally, we utilized machine learning interatomic potentials (MLIPs), CHGNet and MACE-MP-0, to illustrate the phase transitions at varying degrees of deintercalation in Na 2 RuO 3 . To understand the redox chemistry of this material, we investigated the electronic structures, the O 2 binding energies, the bond covalency, and the local magnetic moments. Our study demonstrates that the strongly constrained and appropriately normed (SCAN) functional outperforms PBE and PBE + U methods across all voltage ranges within the operating window. Furthermore, our computed electrochemical potentials with the SCAN functional are in agreement with the available experimental data. Additionally, by incorporating a series of Hubbard U values (U = 2, 4, 5 eV), we highlight the importance and accuracy of suitable U parameters depending on the element of interest. Our results indicate that in Na 2 RuO 3 , cationic redox is primarily dominant despite it being a Na-rich material. Moreover, we demonstrate that CHGNet and MACE-MP-0 MLIPs can be effectively used to prescreen Na-rich cathode materials with reasonable accuracy for their electrochemical properties.
