Skip to content

Multistage reaction pathways in detonating high explosives

Evidence and attribution

Authority of statements

Prose below summarizes the publication identified by doi, title, and pdf_path in the front matter. For definitive numerical values and figures, use the peer-reviewed article.

Summary

Large-scale reactive MD of shocked RDX crystal—validated against shorter quantum MD—resolves a two-stage chemistry: rapid N₂ and H₂O production within ~10 ps followed by delayed CO evolution past nanoseconds. Metastable carbon- and oxygen-rich clusters with fractal shapes are argued to bottleneck further oxidation to small gases; distinct unimolecular versus intermolecular channels are identified for the fast N₂ and H₂O channels (abstract; extract). The Letter situates the problem in the broader detonation context: reaction-zone width and reaction time are tied to intermediate products, and sub-nanosecond RDX pathways under high P/T had been poorly mapped because few reactive MD studies spanned both ~100 nm length and ~100 ps time scales simultaneously—motivating the scalable ReaxFF implementation (introduction, extract).

Methods

Reactive model and validation ladder

  • Large-scale parallel ReaxFF MD (domain decomposition + message passing) targets shock-induced chemistry in crystalline RDX, with subset trajectories compared to shorter quantum molecular dynamics (QMD) benchmarks (abstract; introduction).

Initial crystal / supercell (pre-shock)

  • The Letter cites an 168 × 5 × 5 stack of RDX unit cells in a simulation box ~22.28 × 5.787 × 5.354 nm³ equilibrated near room temperature before shock loading (APL 105, 204103; extract).

ReaxFF formulation (qualitative)

  • ReaxFF uses bond orders, dynamic charges, and an electronegativity equalization-style scheme to follow bond making/breaking at cost below full QMD (extract/introduction).

Shock protocol details

  • Piston speed, thermostatting, and duration of the production shock runs appear in APL Methods—beyond the short extract on this wiki page.

1 — MD application (shocked crystalline RDX)

  • Engine / code: Large-scale parallel ReaxFF MD with domain decomposition + message passing as stated in the abstract/introduction (extract); N/A — explicit program name string not on indexed extract p1–3 (confirm in papers/ReaxFF_others/Li-RDXdetonation-APL14.pdf).
  • System size & composition: 168 × 5 × 5 stack of RDX unit cells in a box ~22.28 × 5.787 × 5.354 nm³ equilibrated near room temperature before shock loading (APL 105, 204103; extract).
  • Shock / shear: shock loading of the crystal after equilibration (abstract/introduction framing); N/A — piston speed, shock direction, and Hugoniot state variables not on indexed extract p1–3 (APL Methods).
  • Boundaries / periodicity: PBC implied by crystal supercell shock setup language in the Letter class; N/A — explicit PBC statement not copied here (PDF).
  • Ensemble: N/A — headline shock MD ensemble (NVT vs NVE vs MSST) not stated on indexed extract p1–3 (APL Methods).
  • Timestep / thermostat / barostat / production duration: N/A — not stated on indexed extract p1–3 (APL Methods).
  • Temperature / pressure: Room-temperature pre-shock equilibration is noted for the initial cell (extract); post-shock thermodynamic control—N/A — not on indexed extract p1–3.
  • Electric field: N/A — not indicated in the indexed extract opener.
  • Replica / enhanced sampling: N/A — not indicated in the indexed extract opener.

2 — Force-field training

N/A — new ReaxFF parameterization is not the focus of the indexed abstract/extract window; the Letter emphasizes applying ReaxFF chemistry with QMD validation subsets (abstract).

3 — Static QM validation ladder

  • Quantum MD (QMD): shorter QMD trajectories used as a validation reference against ReaxFF subsets (abstract); N/A — DFT functional/basis/cutoffs not on indexed extract p1–3.

Findings

Simulations report a two-stage mechanism: rapid N\(_2\) and H\(_2\)O formation on ~10 ps timescales, then delayed CO evolution on nanosecond timescales. Large metastable carbon- and oxygen-rich clusters with fractal morphology bottleneck further oxidation to small gas products, especially CO. Distinct unimolecular versus intermolecular routes dominate the fast N\(_2\) versus fast H\(_2\)O channels, respectively (abstract; extract pages 1–3). The authors highlight this as first evidence in their setup for large C/O-rich clusters as intermediates that inhibit CO production, yielding the staged picture of N\(_2\)/H\(_2\)O first and CO later (abstract paragraph in extract). The introduction also sketches the detonation-wave picture (von Neumann spike, reaction zone) to motivate why intermediate inventory sets effective reaction time in HE modeling (introduction, extract).

Limitations

Reactive MD fidelity depends on the underlying reactive force field; extreme shock conditions stress any classical model.

Citations and evidence anchors

  • DOI 10.1063/1.4902128 (extract citation line).
  • Abstract (extract page 1).