Development of a Mg/O ReaxFF Potential to describe the Passivation Processes in Magnesium-Ion Batteries
Summary¶
A Mg/O ReaxFF parameter set is trained against DFT data to capture bulk Mg and MgO, low-index Mg surfaces, adsorption, and diffusion, then applied to O₂ attack on Mg anode facets and related oxidation pathways relevant to magnesium-ion battery anode passivation by oxygen impurities. The framing emphasizes that trace O₂ exposure can dominate early-stage interface evolution even when electrolyte chemistry is nominally “dry,” so an accurate Mg/O reactive model is a prerequisite for realistic anode simulations. The introductory discussion in ChemSusChem stresses Mg-ion battery deployment barriers around passivating interphases at Mg anodes and positions O₂ impurities as a chemically simple yet practically important oxidation channel to model before tackling full electrolyte complexity.
Methods¶
ReaxFF training against DFT (A)¶
- Mg metal: hcp, fcc, bcc, A15, sc equations of state, cohesive energies, bulk moduli, and low-index surface energies.
- Mg–O: Bulk MgO, adsorption, and surface scenarios; diffusion barriers and other targets summarized in ChemSusChem + SI.
- QM level: DFT settings described in the paper feed the weighted optimization.
Validation simulations (B / GCMC + MD)¶
- GCMC + MD: Oxidation of a Mg nanoparticle + high-T anneal (~2000 K) to form rocksalt MgO.
- Facet MD: O\(_2\) on Mg(0001), Mg(10̄10)A, Mg(10̄11)—track dissociative adsorption, exothermic heating, oxide thickness (up to ~25 Å on the most reactive facet), and grain-boundary microstructure.
Analysis emphasis¶
Compare facets by oxide thickness, local temperature rise, and defect content—not terrace adsorption energy alone.
MD application (integrated)¶
Engine / code: LAMMPS and ReaxFF. Systems: O\(_2\) on Mg nanoparticle and (0001), (10̄10)A, (10̄11) slabs; GCMC + MD plus ~2000 K high-T anneal in ChemSusChem (see full paper for high-T segments). PBC for slabs; 3D nanoparticle cell. N/A — full atom counts, exact timestep, ps/ns segment lengths, thermostat damping, NVT setpoints, NPT usage, QEq and long-range settings, O\(_2\) dosing pressures in SI—not re-listed on this page to avoid inventing numbers. N/A — applied electric field; N/A — umbrella/REX in the main O\(_2\) line described.
Findings¶
Exothermic oxidation¶
O\(_2\) dissociates on contact, releasing enough energy to heat surfaces to thousands of kelvin in trajectories, accelerating MgO formation.
Facet ranking¶
Mg(10̄10)A is the most reactive facet examined, supporting the thickest transient MgO film.
Training-set quality¶
The field matches key DFT metrics with documented limitations (e.g. bulk modulus trends) in the article.
Multiscale implication¶
Localized hot spots and grain boundaries in nascent oxides arise from exothermic O incorporation—motivating heat dissipation models beyond isothermal MD.
The abstract-level narrative also highlights that O₂ dissociates on first contact with modeled Mg facets, heating surfaces to several thousand kelvin and driving rapid rocksalt MgO formation—consistent with the thick transient films and facet-dependent reactivity summarized above.
Limitations¶
ReaxFF remains semi-empirical; quantitative oxidation kinetics and oxide electronic structure require validation against experiment and higher-level theory for specific electrolyte environments.
Relevance to group¶
Direct ReaxFF parameterization collaboration (van Duin) on Mg/O chemistry for metal-anode passivation, adjacent to broader battery interface simulation themes.