Skip to content

Molecular Dynamics Derived Gas-Surface Models for Use in Direct Simulation Monte Carlo

Summary

Molecular nitrogen–surface collisions on atomistically generated graphene and quartz sheets supply accommodation coefficients and post-collision velocity distributions for comparison with Maxwell and velocity-sampling gas-surface models in DSMC. The motivation is aerothermodynamics and rarefied gas applications where wall models must bridge atomistic scattering statistics and continuum or DSMC solvers; the paper’s graphene preparation uses ReaxFF relaxation to obtain realistic roughness and chemistry before extracting scattering kernels.

Methods

This contribution is experimental / computational fluid–surface coupling in spirit: classical molecular dynamics generates nitrogen–surface collision statistics that feed gas–surface interaction models in DSMC (papers/Accommodation_paper_Dec15submitB.pdf). The authors do not report a new force-field fit—they apply existing empirical models, using ReaxFF relaxation to prepare corrugated graphene and Lennard-Jones-style non-reactive dynamics for N\(_2\) impingement as stated in the manuscript.

Surface preparation (graphene). Expanded graphene supercells (up to 512 carbon atoms) sit in a nominally ~17.08 × 19.72 × 25.00 Å periodic box. ReaxFF relaxation is followed by NPT annealing at 1 atm with a Berendsen barostat (100 fs damping) and Berendsen thermostat (10 fs coupling) while the temperature is stepped 300 K → 1000 K → 300 K in 100 K segments of 2 ps each (40 ps total).

Surface preparation (quartz). Random SiO\(_2\) placement is minimized for 125 ps at 0.25 fs per step, then the slab is annealed with combined NVT/NPT segments from 300 K toward 4000 K in 100 K increments of 2 ps, yielding a 532-atom quartz slab (~17.32 × 19.98 × 22.53 Å).

N\(_2\) impingement trajectory bank. Non-reactive van der Waals trajectories integrate 36,000 steps at 0.25 fs (9.0 ps wall time per trajectory) with <0.05% energy drift as reported. About 2000 trajectories per incidence condition are accumulated on graphene at 1503 and 1100 m s⁻¹ with incidence angles 0°–80°, and on quartz at 750, 500, and 250 m s⁻¹ with angles 10°–40°. The checked-in PDF excerpt does not name the MD program explicitly—confirm the software string in the full PDF when auditing reproducibility.

Not emphasized in the collision protocol: replica exchange or other enhanced sampling; applied electric fields; production-stage reactive bond formation during the N\(_2\) impact ensemble (ReaxFF is used for graphene preparation, not as the integrator for the accommodation statistics bank).

Findings

Post-collision trajectories partition into single-bounce, multiple-bounce-with-escape, and multiple-bounce-without-escape classes whose probabilities depend strongly on incidence angle, speed, and whether the prepared surface is smooth versus intentionally rough—a simple mechanism-level picture of how momentum is thermalized at the wall. Energy accommodation coefficients track the kinematic variables and surface topology. A velocity-sampling gas–surface interaction model implemented in DSMC reproduces Maxwell-model predictions when fed MD statistics from rough (diffuse) surfaces under the conditions tested, i.e. close agreement with Maxwell baselines versus the velocity-resolved kernel built from MD. Sensitivity to incidence speed and angle is strong: the rate at which trajectories escape after multiple encounters shifts the accommodation distribution enough that angle-averaged coefficients alone can mislead DSMC boundary models. However, the manuscript’s MD bank is non-reactive N\(_2\) scattering after ReaxFF surface prep, so catalytic chemistry and electronic stopping are outside scope. Limitations include dependence on the specific temperature annealing used to roughen graphene and on finite slab thickness; future work would extend the same MD→DSMC workflow to oxidized or contaminated skins once validated against molecular beams. Corpus honesty: quantitative tables beyond this summary live in the registered PDF papers/Accommodation_paper_Dec15submitB.pdf.

Limitations

The manuscript excerpt in the corpus may omit some figure details; DOI is not populated in front matter here—anchor claims to the checked-in PDF and any updated normalized bibliography record when available. Non-reactive N\(_2\)–surface treatment limits chemistry beyond physisorption; oxidized or contaminated spacecraft materials may require extended models.

Relevance to group

Gas–surface MD with ReaxFF-prepared carbon surfaces connects to broader hypersonic and materials simulation interests; use this page when tracing how atomistic trajectories feed DSMC boundary models rather than as a standalone ReaxFF reactivity study.

Citations and evidence anchors

  • Manuscript PDF: papers/Accommodation_paper_Dec15submitB.pdf (DOI not populated in front matter for this ingest).