Skip to content

Effect of Fe–O ReaxFF on Liquid Iron Oxide Properties Derived from Reactive Molecular Dynamics

Summary

Liquid Fe–O is modeled at 2000 K over oxidation degrees \(Z_\mathrm{O} = \mathrm{O}/(\mathrm{O}+\mathrm{Fe})\) with 0 < \(Z_\mathrm{O}\) < 0.6, motivated by iron powder combustion as an energy carrier. This benchmark-style reactive MD study compares several ReaxFF parameterizations for the same thermodynamic state and observables, against experiment and thermodynamic data where available. The authors evaluate how choice of Fe–O ReaxFF affects minimum-energy paths, structure, (im)miscibility of liquid phases, transport coefficients, and mass and thermal accommodation coefficients, comparing to experiments and equilibrium calculations when possible, with the overarching goal of identifying which macroscopic observables are most sensitive to parameter-line differences.

Methods

Simulation matrix (B)

Liquid Fe–O at 2000 K, scanning oxygen fraction \(Z_\mathrm{O}=\mathrm{O}/(\mathrm{O}+\mathrm{Fe})\) with 0 < \(Z_\mathrm{O}\) < 0.6.

Observables

Minimum-energy paths, structure, liquid miscibility, transport coefficients (diffusivity/viscosity proxies), mass/thermal accommodation coefficients for gas–liquid models.

Parameter-set comparison (A)

Multiple literature Fe–O ReaxFF sets run side-by-side vs experiment and equilibrium references; diagnostics separate structural vs transport sensitivity.

The J. Phys. Chem. A framing ties the benchmark to iron powder combustion as a metal-fuel concept: liquid oxide microphysics controls evaporation, condensation, and gas–surface accommodation in spray/particle flames, so force-field differences that shift miscibility or transport can propagate to macroscopic burning models even when short-range pair correlations look similar.

1 — MD application (atomistic dynamics)

Engine / code: LAMMPS with each compared Fe–O ReaxFF parameter line (as listed in the article). System & composition: Bulk liquid Fe–O at 2000 K, 0 < \(Z_\mathrm{O}\) < 0.6 in O/(O+Fe). Boundaries / periodicity: 3D PBC melt cells. NVT production runs at 2000 K use fs-scale timesteps and multi-ns durations as tabulated; N/A on this page for every NVT thermostat constant and ns clock (see J. Phys. Chem. A Methods and SI). Pressure, electric field, shear, shock, enhanced sampling: N/A in this short summary for the liquid Fe–O benchmark (see PDF if a subsection adds one of these).

2 — Force-field training

N/A in the sense of a new fit—this paper compares existing Fe–O ReaxFF lines; training histories differ by line and are in the original parameterization papers.

3 — Static QM

N/A for on-the-fly ab initio MD; the article references thermodynamic and structural comparators and MEP-class analyses as described in the Methods.

Findings

Sensitivity conclusion

Different Fe–O ReaxFF lines predict markedly different liquid properties and accommodation/transport numbers at identical T/composition.

Implication

Authors argue an improved Fe–O parametrization is needed for iron combustion modeling, and identify which observables diverge most across parameterizations.

Across 2000 K liquid Fe–O compositions, transport and accommodation coefficients compared with literature and equilibrium surrogates show strong sensitivity to which ReaxFF line is chosen—this is a central limitation of empirical reactive MD for metal spray flames when oxidation reaction networks are not identically encoded between parameter sets. Future work would tighten the Fe–O database against high-T experiments; quote PDF tables in pdf_path for values not repeated here.

Limitations

ReaxFF accuracy is parameter-set- and training-data-dependent; high-temperature liquid iron oxide properties are sparsely characterized experimentally at atmospheric pressure, which constrains validation. When multiple Fe–O parameter lines disagree, the authors’ comparative framing should be read as identifying sensitivity hotspots—transport and accommodation numbers may shift non-linearly if short-range oxygen packing is misrepresented—even when qualitative liquid miscibility trends look similar across force fields. The 2000 K scan over \(Z_\mathrm{O}\) is a deliberate computational benchmark design: it separates structural sensitivity from transport/accommodation sensitivity when comparing published Fe–O lines side by side.

Relevance to group

Adri C. T. van Duin co-authors; connects ReaxFF combustion chemistry to metal fuels and group expertise in reactive MD parameter assessment.

Citations and evidence anchors

Reader notes (navigation)