Molecular dynamics simulations of the oxidation of aluminum nanoparticles using the ReaxFF reactive force field
Summary¶
Hong and van Duin report ReaxFF molecular dynamics of aluminium nanoparticle (ANP) oxidation at 300, 500, and 900 K for two initial oxygen densities (0.13 and 0.26 g cm\(^{-3}\)). The article argues that O\(_2\) adsorption and dissociation create localized hot regions and void-like disrupted metal/oxide regions at the surface; a bond-restraint barrier scan reported in the paper finds that voids can lower an illustrative oxygen diffusion barrier by up to ~92% and change the step from endothermic to exothermic in that model setup, after which oxide growth proceeds by accelerated oxygen transport. The authors further relate oxide thickness and density to the combined effects of temperature and oxygen density, and state qualitative agreement with selected experimental literature on aluminium oxidation kinetics.
Methods¶
Force-field lineage (Al/O). Simulations use the ReaxFF reactive force field for aluminium/oxygen chemistry as summarized in the article (energy decomposition and QEq-style charge treatment as referenced there), including a validation trajectory set on a bare Al (431) slab with oxygen before the ANP cases discussed in the Results section.
MD application — engine, cells, boundaries. ReaxFF MD is carried out with the 29 Jan 2014 LAMMPS build cited in the paper. Two primary setups are reported: (i) a 504-atom Al (431) slab in a 1.40 × 1.28 × 10.0 nm cell with 150 O\(_2\) molecules for the low-pressure validation trend study (additional high-density validation case described in the article); and (ii) an 864-atom amorphous Al cluster (~2.8 nm diameter) in a 5.0 × 5.0 × 5.0 nm box with 300 or 600 O\(_2\) molecules, giving the 0.13 and 0.26 g cm\(^{-3}\) initial oxygen densities used for the ANP oxidation sweeps. Three-dimensional periodic boundary conditions are used as stated for these gas–particle cells.
Ensemble, thermostat, timestep, duration. For the canonical ANP oxidation production runs, the authors use the NVT ensemble (constant volume) with a Nosé–Hoover thermostat (temperature damping parameter 100 fs) and an integration timestep of 0.2 fs, for 5,000,000 steps (1.0 ns total) at each reported temperature/density combination. A separate NVE demonstration at low oxygen density (0.01 g cm\(^{-3}\)) is described to examine energy-conserving heating from oxidation exothermicity without thermostatting.
Barostat / pressure. N/A — the primary ANP runs are constant-volume NVT; the article’s NVE subsection is used to discuss microcanonical behavior at low oxygen density rather than constant-pressure control.
Electric field, replica sampling. N/A — not used as a driving force in the protocol described for these oxidation trajectories.
Findings¶
Mechanism and microstructure. The authors connect highly exothermic early oxygen consumption to short-lived hot surface regions, void formation near the outer surface, and rapid initial oxide thickening, then analyze how temperature and oxygen density modulate oxide thickness/density and oxidation-state partitioning along the trajectory.
Comparisons. Bare-slab simulations at elevated oxygen density are used to recover qualitative trends of temperature-dependent limiting oxide growth relative to an cited XPS study, while acknowledging time-scale and pressure gaps between simulation and experiment.
Sensitivity / design levers. Reported sweeps vary system temperature (300–900 K for the ANP cases in Section 3.3) and initial oxygen density (0.13 vs 0.26 g cm\(^{-3}\)), and include the NVE low-density control to argue that void-assisted pathways are not an artefact of the thermostat alone.
Limitations (as framed in the article). The manuscript discusses high oxygen densities relative to ambient STP, nanosecond horizons, and the role of the thermostat as a stand-in for environmental heat transfer for small nanoparticles—see the Discussion for the authors’ caveats.
Limitations¶
- Nanoscale MD still uses high effective pressures/temperature ramps compared to many experiments; quantitative burning rates require careful upscaling analysis.
- ASAP PDF may differ slightly in pagination from the final issue version.
- Particle size polydispersity, native oxide thickness distributions, and alloying additives in technical Al powders can shift oxidation kinetics beyond the idealized ANP cores modeled here.
Relevance to group¶
Adri C. T. van Duin co-authorship; strengthens the Al/Al\(_2\)O\(_3\) ReaxFF storyline used across combustion, propellants, and corrosion-related modeling.