Skip to content

Extension of the ReaxFF combustion force field toward syngas combustion and initial oxidation kinetics

Summary

This Journal of Physical Chemistry A article retrains the widely used Chenoweth et al. CHO-2008 ReaxFF combustion parameter set into an updated CHO-2016 description, coauthored by Chowdhury Ashraf and Adri C. T. van Duin. The authors identify two limitations in CHO-2008 for their targets: inaccurate small-molecule oxidation chemistry relevant to syngas, especially conversion between CO and CO\(_2\), and an overly fast hydrogen abstraction by molecular oxygen from hydrocarbons, which depresses predicted oxidation initiation temperatures relative to expectations. The manuscript expands the DFT-based training set with additional reactions and transition-state geometries along syngas oxidation routes and oxidation initiation pathways, then reoptimizes parameters while aiming to preserve CHO-2008 quality for large hydrocarbon chemistry such as jet-fuel surrogates.

Methods

Force-field training (ReaxFF). Starting from CHO-2008 (Chenoweth et al.), the authors expand the DFT-backed training set with reaction energies and transition-state geometries along syngas oxidation routes and hydrocarbon oxidation-initiation pathways where CHO-2008 is known to fail—especially CO ↔ CO₂ interconversion for small-molecule oxidation and an overly facile O₂ hydrogen abstraction that suppresses predicted oxidation initiation temperatures. Reoptimization yields CHO-2016 while the manuscript states an explicit goal to preserve CHO-2008-level behavior for large-molecule / jet-surrogate chemistry. QM level, weighting, and optimizer details (least-squares / CMA-ES-style workflows) are documented in the main text and Supporting Information.

MD application (validation). High-temperature gas-phase NVT reactive MD exercises CHO-2016 on syngas, methane, JP-10, and n-butylbenzene for both oxidation and pyrolysis scenarios (fuel list and qualitative outcomes summarized in the abstract). Supercell compositions, initial stoichiometries, target temperatures, timestep, thermostat, total simulated times, and MD software are tabulated in the article + SI; this wiki page does not copy those run cards from the short front-matter extract.

Static QM underpins the fit and selected spot checks rather than serving as a standalone production method. Electric-field drives and enhanced sampling are N/A in the indexed framing of this parametrization paper.

FF-training blueprint honesty. Parent CHO-2008 ReaxFF; QM (DFT) training structures/energies; optimization language (least-squares / CMA-ES) and reference QM/experimental benchmarks are all in the article + SI—this page summarizes intent only.

MD validation blueprint honesty. High-temperature reactive molecular dynamics on gas-phase PBC supercells uses NVT in the abstract’s wording; LAMMPS is the usual engine for published CHO-2016 workflows—confirm in SI. Timestep, thermostat, equilibration/production durations (ps/ns), barostat/pressure if any NPT segments exist, and boundary conditions beyond PBC are N/A on this page—copy from the PDF/SI.

Findings

Syngas and methane oxidation simulations with CHO-2016 show substantially improved C\(_1\) chemistry relative to CHO-2008 and resolve the low-temperature oxidation initiation problem attributed to overly facile O\(_2\) abstraction in the older parametrization. For JP-10, Arrhenius parameters for decomposition obtained with CHO-2016 agree with experiment and with CHO-2008 simulation results within the scope claimed in the abstract. For n-butylbenzene, initiation mechanism pathways from CHO-2016 remain in good agreement with both experiment and CHO-2008 outcomes, supporting transferability across fuel classes in the authors’ tests. The article nonetheless acknowledges the vast size of combustion reaction networks and the need for case-by-case assessment beyond the validation suite.

The introduction also contrasts detailed kinetic models for small hydrocarbons with the practical need to simulate complex fuels and fuel mixtures at elevated pressures and in condensed phases where hand-built mechanisms become incomplete, motivating ReaxFF as a reaction-discovery tool that does not require prespecifying every elementary step, while acknowledging that quantum accuracy still limits affordable system sizes and timescales. The article cites enormous computational costs for illustrative ab initio nanoreactor studies as a contrast point for why ReaxFF remains attractive for statistically meaningful reactive sampling despite its empirical approximations.

Limitations

  • Combustion networks remain enormous; transferability must be assessed case-by-case beyond tested fuels.

Relevance to group

Landmark group publication extending the widely used CHO ReaxFF line.

Citations and evidence anchors

  • DOI: 10.1021/acs.jpca.6b12429 (papers/Chowdhury_CHO_2017_JPCA.pdf).
  • Text-aligned pointers: normalized/extracts/2017chowdhury-venue-jp6b12429_p1-2.txt

Reader notes (navigation)