Development of a ReaxFF reactive force field for ammonium nitrate and application to shock compression and thermal decomposition
Evidence and attribution¶
Authority of statements
Prose sections below (Summary, Methods, Findings, etc.) are curated summaries of the publication identified by doi, title, and pdf_path in the front matter above. They are not new primary claims by this wiki.
For definitive numerical values, reaction schemes, and interpretations, use the peer-reviewed article (and optional records under normalized/papers/ when present)—not this page alone.
Summary¶
Shan, van Duin, and Thompson report a ReaxFF reparameterization for ammonium nitrate (AN) starting from an existing nitramine/TATB description, training against electronic-structure data for barriers, heats of formation, and crystal properties of AN phases. Applications shown in the abstract include isothermal compression of phase-IV AN (claimed agreement with DFT/experiment within about 10% over the studied compression range), unreacted principal Hugoniot states (noted as stiffer than experiment by about 17% in the abstract statement), and thermal decomposition simulations up to 2500 K with pathways said to align with experimental findings. Simulations are executed in LAMMPS using the group's ReaxFF workflow.
AN is a foundational energetic oxidizer whose phases and shock response are sensitive to hydrogen bonding and ion ordering; the parameterization aims to unify solid-state mechanics with reactive chemistry.
Methods¶
Force-field training (ReaxFF for ammonium nitrate). The authors extend an existing nitramine/TATB ReaxFF description and reoptimize bonded and nonbonded terms so NH\(_4\)NO\(_3\) matches electronic-structure targets for dissociation barriers, heats of formation, and crystal properties of AN phases (abstract). Optimization is carried out with the standalone ReaxFF code; production MD uses LAMMPS (Introduction). Program choices for QM reference data, basis/cutoff conventions, and the full training-set listing are in Section 2 of papers/Shan_Thompson_AN_JPCA_2014.pdf and the Supporting Information—not transcribed here. For a sibling reading copy with overlapping DOI text and more protocol detail extracted in-repo, see [[2014shan-venue-research]].
MD application (EOS, Hugoniot, thermal decomposition). The fitted field is applied to phase-IV solid AN for isothermal P–V curves, unreacted principal Hugoniot states, and thermal decomposition heated to 2500 K (abstract), using LAMMPS with ReaxFF as stated in the article. Supercell sizes, PBC vectors, ensemble staging (NVT/NPT as appropriate), timestep, equilibration/production segment durations (ps/ns in the article), and thermostat/barostat parameters are given in Section 2 and the SI rather than in the short in-repo extract used here—N/A to transcribe numerically on this page. Electric fields, shock pistons, and enhanced sampling are N/A for the validation cases summarized from the abstract.
Static QM. DFT supplies reference energies and equations of state for parametrization and static benchmarks; the headline applications in the abstract are ReaxFF MD, not production AIMD trajectories (N/A for standalone AIMD application block).
Findings¶
For phase IV, the isothermal P–V curve from the model agrees with DFT and experiment within about 10% over the compression range quoted in the abstract. The unreacted principal Hugoniot is nevertheless roughly 17% stiffer than experiment in the abstract’s summary—consistent with the introduction’s discussion of scatter across literature US–uP parametrizations for AN. Thermal decomposition trajectories to 2500 K are described as agreeing with experimental pathways at a qualitative level (article body for species-resolved detail). The abstract’s message is explicitly validation-sensitive: good isothermal agreement does not guarantee satisfactory Hugoniot response for this parameterization.
Limitations¶
- Hugoniot-level agreement is explicitly imperfect in the abstract; users should treat shock observables as validation-sensitive outputs.
- Full training-set tables, additional phases, and species-resolved timelines are in the J. Phys. Chem. A article + SI rather than a short extract-backed note.
- Extract is early pages only; quantitative plots and full barrier sets require the PDF body.
Relevance to group¶
Adri C. T. van Duin co-authorship ties this to ReaxFF development for energetic nitrogen/oxygen chemistry used across combustion and initiation modeling threads.