Materials & Energyarticle2026-10-02

Nucleation and Coalescence of Wustite-like Motifs in Oxygen-Deficient Magnetite: A Deep-Potential Molecular Dynamics Study

0 citations

Abstract

Abstract Oxygen loss in magnetite drives the cation redistribution and local structural reconstruction that control the rate of iron oxide reduction, but this coupling has resisted direct simulation: the reconstruction develops far beyond DFT-accessible scales, and classical potentials cannot describe mixed-valence Fe–O with the required fidelity. We developed a DFT-validated Fe–O Deep Potential trained on bulk, defective, and surface configurations and applied it in large-scale molecular dynamics with an imposed stochastic oxygen-removal protocol. The potential reproduces DFT energy and force trends together with vacancy and surface energetics. DFT single-point checks on oxygen-deficient configurations and equation-of-state calculations for Fe3O4 and FeO further delineate the transferability of the potential, with larger deviations emerging in the most strongly oxygen-deficient states. As oxygen is removed, four-coordinated O environments decrease while six-coordinated wüstite-like motifs emerge, grow, and coalesce into extended local domains. Deep-Potential nudged elastic band (DP-NEB) calculations, benchmarked against selected DFT single-point energies, show that the coupled O/Fe-vacancy configuration gives the lowest Fe-migration barrier among the selected local rearrangement pathways. Time-windowed mean-squared displacements yield apparent Fe and O diffusivities that evolve in the nonstationary lattice. These simulations establish a defect-mediated solid-state response toward wüstite-like local order under imposed oxygen-deficient conditions, rather than the formation of a complete equilibrium Fe1–xO phase. The protocol isolates oxide-side reconstruction and does not explicitly treat hydrogen chemistry.

// Source

View paper (DOI)OpenAlexChemistry of MaterialsPublished 2026-10-02

Authors: Zeng Liang, Kejiang Li, Jianliang Zhang, Alberto N. Conejo

Institutions: University of Science and Technology Beijing