Decoding Composition-Structure-Performance Relations in Sodium Layered Oxide Positive Electrode Materials
arXiv:2610.30591
Abstract
Layered sodium transition-metal (TM) oxides, NaTMO, are among the most promising sodium-ion positive electrode materials, yet their Na/vacancy orderings, layer stacking sequences, and cooperative Jahn--Teller distortions are known only for a few single-TM compounds. We compile the Na/vacancy orderings reported in earlier studies into a single library of 41 unique orderings and apply this uniformly to five single-TM and all ten NaTMTMO oxides (TM = Ti, Mn, Fe, Co, Ni), screening more than 120,000 structures with machine-learning interatomic potentials prior to 1,025 accurate first-principles calculations. The phase diagrams explain the observed O3/P3/O1 sequences and voltages, and the O3-to-P3 competition hinges on ordering energies of a few meV per atom that scalar descriptors, such as the cationic potential and the interlayer-slab distance, miss. Our extensive analysis reveals that the redox couples form a nearly partner-independent ladder, Ti Mn Co Ni Fe, explaining pristine charge transfer and the oxidation sequence. Van Vleck analysis of every TMO octahedron reveals that TM mixing suppresses the cooperative Jahn--Teller distortions of Mn and Ni through charge transfer to Jahn--Teller-inactive states or dilution of Jahn--Teller-active centers. We formulate an energy-density-versus-strain map that yields a design rule: a high-voltage couple without Jahn--Teller strain, which Mn/Ni charge transfer realizes in NaMnNiO, the most practical composition. We release the ordering-resolved dataset as a training set for subsequent training of accurate machine-learning interatomic potentials.