Skip to content

ReaxFF reactive molecular dynamics on silicon pentaerythritol tetranitrate crystal validates the mechanism for the colossal sensitivity

Summary

Silicon pentaerythritol tetranitrate (Si-PETN) is known experimentally for extreme sensitivity, motivating mechanistic hypotheses beyond those typical of PETN alone. This PCCP article extends ReaxFF for Si–C–H–O–N energetic chemistry using QM training data, then applies LAMMPS reactive MD to contrast thermal decomposition of solid Si-PETN versus PETN and interpret Si-PETN’s exceptional sensitivity. The narrative ties bulk simulations to a previously proposed unimolecular carbon–silyl nitro-ester rearrangement that creates Si–O linkages not available to PETN in the same way, arguing that exothermic Si–O formation can couple to early NO\(_2\) release under conditions where PETN follows more conventional pathways. The contrast illustrates how heteroatom chemistry reroutes decomposition within the same nitrate-ester motif class—supporting the paper’s conclusion that colossal sensitivity can follow a low-barrier unimolecular route rather than dominant multi-body initiation.

Methods

Force-field training (ReaxFF, Si–C–H–O–N). The authors extend ReaxFF for silicon-containing nitrate-ester chemistry so bulk simulations use the same reactive manifold as prior single-molecule QM work on Si-PETN. The Electronic supplementary information bundled with papers/ReaxFF_others/TingTingZou_Goddard_PETN_SI_PCCP2014.pdf documents QM versus ReaxFF comparisons for Si–C, Si–O, and Si–N bond dissociations (Fig. S1), bending energies for Si–O–C, O–Si–C, C–Si–C, and Si–O–N angles (Fig. S2), bond-order cutoffs for fragment analysis (Table S1), and the parameter file ESI-ffield.txt.

MD application (ReaxFF-RMD in LAMMPS). Solid Si-PETN is modeled in a \(2\times2\times3\) supercell (696 atoms) after lattice optimization with three-dimensional periodic boundary conditions (PBC) on all axes. Equilibration uses 0.2 fs timesteps with Nose–Hoover thermostat (100 fs damping) and Parrinello–Rahman barostat (1000 fs damping) to relax lattice parameters at 5 K and 300 K (as stated in the PCCP/SI text), targeting atmospheric pressure during those NPT-style lattice-relaxation segments. Thermal-decomposition studies heat each crystal rapidly in NVT from 300 K to one of 1000, 1200, 1400, 1600, 1800, or 2000 K (example cited: 300 K → 1600 K in 0.1 ps), then continue NVT sampling at the target temperature for at least 20 ps before analyzing chemistry. To probe how exothermic reactions feed back on temperature and kinetics, additional microcanonical (NVE) RMD runs are started at 1200, 1400, 1600, and 1800 K and run for at least 100 ps with 0.1 fs timesteps. PETN crystals are simulated under parallel protocols for direct comparison. Pressure control: N/A — the high-temperature NVT/NVE decomposition stages use fixed-volume cells without hydrostatic barostat control; mechanical pressure/stress evolves only as a consequence of shear-free heating and chemistry, whereas the earlier lattice equilibration employs a barostat targeting 1 atm as quoted above.

Static QM / DFT-only: N/A — this publication’s new electronic-structure training data are referenced through the ReaxFF fitting and literature QM on Si-PETN; the PCCP/SI text should be consulted for any explicit DFT functional/basis details tied to those prior QM benchmarks.

Electric field / replica sampling / shear-shock: N/A — not used in the thermal RMD protocol described above.

Findings

At lower simulated temperatures, Si-PETN decomposition begins with the Liu carbon–silyl nitro-ester rearrangement that forms Si–O linkages—steps the authors state are absent for PETN under the same low-temperature regimes. As chemistry progresses, they argue that exothermic Si–O formation promotes earlier NO\(_2\) release from N–O/C-type cleavage channels that do not appear for PETN at comparable conditions. At higher temperatures PETN reacts by conventional NO\(_2\) loss and HONO elimination, yet Si-PETN remains markedly more reactive in their comparison. The bulk RMD results are presented as validating prior single-molecule QM arguments that Si-PETN’s extreme sensitivity follows a low-barrier unimolecular pathway rather than requiring dominant multi-body collision initiation. The PETN–Si-PETN contrast highlights how replacing the central C with Si reroutes early NO\(_x\) chemistry while preserving the overall nitrate-ester motif class. Reactive force fields inherit QM-training approximations; finite cells, rapid heating ramps, and nanosecond-scale sampling influence quantitative branching and timing, so barrier heights and product distributions should be read as model-dependent relative trends unless cross-checked against additional experiments or higher-level theory.

Limitations

Reactive FF energetics for energetic materials require ongoing validation; simulation heating rates and finite-size effects can influence branching ratios.

Relevance to group

Energetic materials ReaxFF reference from the Goddard/Zybin line, adjacent to EM initiation and shock chemistry topics in the corpus.

Citations and evidence anchors