Atomistic-scale analysis of carbon coating and its effect on the oxidation of aluminum nanoparticles by ReaxFF molecular dynamics simulations

Summary¶

An Al/C ReaxFF is developed against QM training sets and exercised in ReaxFF MD to study hydrocarbon-derived carbon coatings on aluminum nanoparticles (ANPs) and their influence on subsequent oxidation. Simulations report hydrogen transfer to Al sites, C–C bond-preserving binding modes for some precursors, and precursor-dependent carbon layer growth. Oxidation of coated particles is compared to bare ANPs, highlighting reduced reactivity at low temperature but high susceptibility once the coating is disrupted at elevated temperature, consistent with the experimental literature cited in the article.

Methods¶

This slug points at papers/Hong_AlCOx_JPCC_2016.pdf; narrative and locator notes for the same DOI also appear on 2016hong-venue-research (ACS proof bytes). The protocol text below matches that article.

Force-field training (ReaxFF for Al/C/H/O): QM training (periodic): VASP GGA-PBE with PAW, 400 eV cutoff; Al(111) slabs (6 layers, lower layers fixed) for hydrocarbon adsorption/decomposition; k-meshes including 8×8×8 (bulk Al\(_4\)C\(_3\)) and 5×5×1 (slab adsorbates); NEB (L-BFGS, 3 images) for barriers. QM clusters: Jaguar B3LYP/6-311G for nonperiodic Al/C/H/O geometries; bond/angle scans for Al–C dissociation and distortion. ReaxFF optimization: sequential refinement of Al–C, Al/H, and angle/off-diagonal terms to QM targets; ReaxFF-NEB cross-checks vs DFT-NEB on selected pathways.

MD application (coatings and oxidation): Engine: ADF for ReaxFF molecular dynamics (same protocol as 2016hong-venue-research). Ensemble / controls: NVT with Berendsen thermostat (100 fs damping); timestep 0.1 fs (authors cite high T up to ~3000 K). Coating cycles: 864-atom Al nanoparticle in 45×45×45 Å³ with 350 gas molecules per cycle; ANP at 300 K, hydrocarbon gas 2500 K for 15 ps, then cool to 300 K in 8.5 ps; repeated; ethylene, ethane, acetylene compared. Oxidation: coated ANP with 600 O\(_2\) in 60×60×60 Å³; 300 K vs 3000 K runs to 150 ps with elevated O\(_2\) density ~0.15 g/cm³ to accelerate chemistry in short windows. Barostat / hydrostatic pressure control: N/A — NVT oxidation and coating stages in the summarized protocol. Electric field / enhanced sampling: N/A — not used in the summarized protocol.

Static QM / DFT: covered under the VASP/Jaguar training blocks above (not a separate post-hoc DFT results section for the oxidation trajectories).

Findings¶

ReaxFF reproduces Al/C interaction energies from QM for the training comparisons presented, and qualitatively captures hydrocarbon binding and surface reaction sequences on bare ANPs (abstract).
Carbon layer growth depends strongly on which hydrocarbon precursors are used (abstract).
Coatings form with H migration to Al sites while often preserving C–C connectivity during deposition stages in the summarized trajectories (abstract).
Carbon-coated ANPs are less reactive at low temperature but become highly susceptible to oxidation when the coating is removed or disrupted at elevated temperature—trends described as consistent with experimental literature in the abstract.

Limitations¶

Combustion-relevant conditions span pressure, size polydispersity, and oxide polymorphism beyond any single MD study.
ReaxFF cannot capture electronically excited or plasma-driven chemistry without additional extensions.
If Hong_AlCOx_JPCC_2016.pdf pagination differs from the ACS proof on 2016hong-venue-research, prefer the journal version-of-record for locators.

Relevance to group¶

Core Hong + van Duin line on energetic Al nanoparticles and passivation engineering, tightly coupled to ReaxFF parameterization practice.

Citations and evidence anchors¶

Abstract in papers/Hong_AlCOx_JPCC_2016.pdf; DOI: 10.1021/acs.jpcc.6b00786.