Supporting information: QM/experimental vs ReaxFF comparison for CHO-2016 combustion force field training
Summary¶
This Supporting Information PDF belongs to the CHO-2016 extension of the ReaxFF combustion parameterization focused on syngas chemistry and oxidation initiation pathways, with the peer-reviewed narrative appearing in the companion Journal of Physical Chemistry A article by Chowdhury Ashraf and Adri C. T. van Duin (see [[2017ashraf-venue-research]] and the version-of-record slug [[2017chowdhury-venue-jp6b12429]] when citing pagination from the issue PDF). The SI pages tabulate quantum-mechanical reference energies alongside ReaxFF (CHO-2016) reaction energies for a hierarchy of small-molecule reactions, including dehydrogenation steps and bond dissociations involving C=O, C=C, O–O, C–H, H–O, and C–O bonds. The tables are intentionally granular: they expose per-reaction residuals so that parameter changes during the CHO-2016 reoptimization can be audited against the underlying quantum targets rather than summarized only by global error metrics. The purpose is training-set documentation: auditors can verify which energetic targets anchored the reoptimization and where residual errors remain. Frontmatter author strings in the corpus may not match publisher ordering; treat the linked CHO-2016 article as authoritative for authorship and DOI.
Methods¶
1 — MD application. N/A — this PDF is Supporting Information of gas-phase QM versus CHO-2016 ReaxFF reaction energy tables; any condensed-phase or production MD validation protocols live in the companion J. Phys. Chem. A article ([[2017chowdhury-venue-jp6b12429]] and related corpus slugs).
2 — Force-field training (SI content). Tables list QM reference energies alongside CHO-2016 ReaxFF values for bond dissociations and dehydrogenation motifs spanning C=O, C=C, O–O, C–H, H–O, and C–O classes so residuals can be audited reaction-by-reaction. QM level (functional, basis, geometry conventions) for each training reaction is defined in the main article and referenced here rather than fully re-derived on this page.
3 — Static QM / DFT. The SI is primarily QM documentation for the fit; detailed Monte Carlo optimization weights, liquid/interface benchmarks, and global error metrics are N/A — not reproduced here—see the primary publication.
Production MD validation protocols (cells, ensembles, timesteps, run lengths) are N/A — not defined in this SI; see the companion J. Phys. Chem. A article ([[2017chowdhury-venue-jp6b12429]] / [[2017ashraf-venue-research]] per corpus linking). For completeness of molecular dynamics indexing on this SI: LAMMPS-style supercell PBC setups, NVT/NPT staging, timestep (fs), equilibration/production run lengths (ps/ns), thermostat and barostat/pressure control, temperature (K), electric field coupling, and umbrella/metadynamics/replica-exchange sampling are N/A — here because this file is QM vs ReaxFF tables, not trajectory protocols.
Findings¶
Training-set diagnostics. Side-by-side QM and CHO-2016 rows show where the reactive field tracks QM rankings and energies for small-molecule combustion motifs and where large residuals identify reactions that remained difficult to fit during reoptimization.
Scope limits. These gas-phase tables do not establish performance for liquids, soot, or interfacial systems; those benchmarks are N/A — not in this SI—see [[2017chowdhury-venue-jp6b12429]].
Confirm every number against pdf_path and the version-of-record companion article; this ingest may lack publisher DOI in frontmatter.
Limitations¶
SI-only files lack standalone discussion of simulation performance on condensed-phase or interfacial systems; confidence remains low for automated metadata because DOI and venue fields may be incomplete in this ingest stub.
Relevance to group¶
The document is a primary transparency artifact for group-led combustion ReaxFF development and should accompany any reproduction of CHO-2016 parameters in LAMMPS decks.
MAS / retrieval notes¶
Ingest metadata lacks a DOI in frontmatter—link to the main CHO-2016 J. Phys. Chem. A article for bibliographic completion. Automation should treat QM vs ReaxFF energy tables as training-set evidence, not as standalone experimental results. Confidence stays low until frontmatter aligns with publisher identifiers and the companion article slug is canonicalized in manifests.
Citations and evidence anchors¶
- Companion article: Ashraf & van Duin, J. Phys. Chem. A 2017, CHO-2016 extension (
papers/Ashraf_CHO_2017_JPCA_SI.pdf).