Off-the-Shelf ReaxFF Parameter Sets Underestimate O–O Bond Strengths in Organic Peroxides: A Static and Kinetic Benchmark for Thermal-Hazard Assessment
- 1SEIHWAN Inc.
Abstract
ReaxFF reactive molecular dynamics (MD) is increasingly used to study organic peroxide decomposition, although the parameter sets supplied with common MD codes were developed for other chemistries. We benchmarked the six C/H/O ReaxFF parameter sets distributed with LAMMPS against CBS-APNO O–O bond dissociation enthalpies (BDEs) of 13 peroxides and tested whether the static error carries over to the kinetics. All six sets underestimate the O–O BDE (mean signed errors of −11.9 to −30.5 kcal mol-1) and correlate only weakly with the references (r = 0.14–0.50). The newer CHO-2016 set (not distributed with LAMMPS) gives −19.5 kcal mol-1, whereas the same protocol yields 85–90 kcal mol-1 for the ethane C–C bond (experiment, about 90 kcal mol-1). The errors increase with alkyl and acyl substitution. For di-tert-butyl peroxide (DTBP), the sets predict 0.4–16.7 instead of 42.4 kcal mol-1, four of them stretch the O–O bond to 1.68–1.76 Å, and a fifth breaks it. Isothermal MD of DTBP at 575–1480 K with three sets gave activation energies of 45.0–76.5 kJ mol-1, consistent with each set’s static BDE (48.7–69.9 kJ mol-1) but 2.0–3.4 times lower than the experimental 153 kJ mol-1. At 121 °C, the extrapolated rate constants exceed manufacturer data by factors of 3.4 × 108 to 5.0 × 1012, and the predicted 10 h half-life temperatures (−147 to −59 °C) lie 180–270 °C below the measured 121 °C. Calculating the O–O BDE of the target peroxide is therefore a simple but necessary check before using a ReaxFF parameter set for hazard assessment.
Related
- Nickel-Catalyzed Carbonylative Synthesis of 1,6-Diketones from Cycloalkylsilyl Peroxides and Arylboronic Acids.
- URSA: Chemistry-Aware Benchmark for Utilitarian Retrosynthesis Assessment
- N -Acyl Hemiperaminals: From Bioactive Skeleton Editing of Artemisinin to Programmable Handles for α-Amido Functionalization and Bioactive Peroxides
- From Peroxides to Qubits: Using Glow-Stick Chemistry to Introduce Excited States, Conical Intersections, and Quantum Computing to Undergraduates
- Single-electron Fluorine Bonding. I. The SEFB-15 and CSFB-14 Benchmark Datasets, Reference Binding Energies and Intermolecular Bond Distances: Assessment of M06-2X density functional
BibTeX
@misc{kim2026offtheshelf,
title = {{Off-the-Shelf ReaxFF Parameter Sets Underestimate O–O Bond Strengths in Organic Peroxides: A Static and Kinetic Benchmark for Thermal-Hazard Assessment}},
author = {Seihwan Kim},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010540/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010540/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Off-the-Shelf ReaxFF Parameter Sets Underestimate O–O Bond Strengths in Organic Peroxides: A Static and Kinetic Benchmark for Thermal-Hazard Assessment
AU - Seihwan Kim
PY - 2026
DA - 2026/10/15
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010540/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010540/v1
AB - ReaxFF reactive molecular dynamics (MD) is increasingly used to study organic peroxide decomposition, although the parameter sets supplied with common MD codes were developed for other chemistries. We benchmarked the six C/H/O ReaxFF parameter sets distributed with LAMMPS against CBS-APNO O–O bond dissociation enthalpies (BDEs) of 13 peroxides and tested whether the static error carries over to the kinetics. All six sets underestimate the O–O BDE (mean signed errors of −11.9 to −30.5 kcal mol-1) and correlate only weakly with the references (r = 0.14–0.50). The newer CHO-2016 set (not distributed with LAMMPS) gives −19.5 kcal mol-1, whereas the same protocol yields 85–90 kcal mol-1 for the ethane C–C bond (experiment, about 90 kcal mol-1). The errors increase with alkyl and acyl substitution. For di-tert-butyl peroxide (DTBP), the sets predict 0.4–16.7 instead of 42.4 kcal mol-1, four of them stretch the O–O bond to 1.68–1.76 Å, and a fifth breaks it. Isothermal MD of DTBP at 575–1480 K with three sets gave activation energies of 45.0–76.5 kJ mol-1, consistent with each set’s static BDE (48.7–69.9 kJ mol-1) but 2.0–3.4 times lower than the experimental 153 kJ mol-1. At 121 °C, the extrapolated rate constants exceed manufacturer data by factors of 3.4 × 108 to 5.0 × 1012, and the predicted 10 h half-life temperatures (−147 to −59 °C) lie 180–270 °C below the measured 121 °C. Calculating the O–O BDE of the target peroxide is therefore a simple but necessary check before using a ReaxFF parameter set for hazard assessment.
ER -