Skip to content

Benchmark of ReaxFF force field for subcritical and supercritical water

Summary

Reactive water under subcritical and supercritical conditions underpins geochemistry, combustion, and hydrothermal materials processing. This benchmark paper evaluates ReaxFF MD (LAMMPS reax/c) against experiments for H₂O spanning ambient through supercritical P–T windows. Properties examined include density, dipole moment, static dielectric constant ε(0), structure factors, hydrogen-bond statistics, self-diffusivity, and proton-transfer behavior. The study concludes ReaxFF matches many SC water observables quantitatively, while ε(0) is only qualitatively captured—an explicit caveat for electrostatic applications. Because water is the universal solvent in countless ReaxFF oxide and electrolyte simulations, this paper is frequently cited as the baseline sanity check before mixing H₂O with minerals, electrodes, or organic fragments in large reactive cells. Operators should still record which O/H parameter file they combine with mineral databases, because cross-term training may shift effective pH-like behavior even when pure water looks reasonable.

Methods

MD application: Simulations use LAMMPS (reax/c) with 309 H₂O molecules in a ~2.1 nm cubic supercell under 3D periodic boundary conditions (PBC). After energy minimization and Gaussian velocity initialization, trajectories integrate with Verlet and 0.2 fs timestep under NPT Nosé–Hoover thermostat and barostat control along prescribed temperature–pressure paths. ~500 ps aggregate sampling (~200 ps production after <10 ps equilibration) with 0.2 ps trajectory stride is reported, plus additional proton-transfer runs with finer output. State space: 400–900 K on 25–100 MPa isobars plus 293 K, 0.1 MPa baseline. N/A — external electric field or umbrella sampling protocols.

Force-field provenance (not a refit study): This J. Chem. Phys. contribution benchmarks an existing dissociative, QEq/EEM-polarizable ReaxFF water description against experiment; the QM training data and parameter optimization that produced the parent O/H set are cited from prior ReaxFF water publications rather than being regenerated here (N/A — new GA-based refit in this manuscript). Reference data in this paper are experimental equations of state, diffusivities, structure factors, and related observables tabulated for comparison.

Findings

Outcomes & mechanisms: ReaxFF tracks experimental density, hydrogen-bond statistics, diffusivity, and proton-transfer trends for subcritical and supercritical water across the state grid reported in J. Chem. Phys. 148, 234503 (2018).

Comparisons: agreement with experiment is described as quantitatively strong for many observables, while the static dielectric constant \(\varepsilon(0)\) is only qualitatively reproduced—an explicit limitation for electrostatic-sensitive workflows.

Sensitivity & design levers: performance depends on the chosen P–T path (temperature and pressure along isobars); proton-transfer statistics can be sensitive to output cadence and timestep choices, so port rates to other engines only after matching the article/SI settings.

Limitations / outlook (as authored): finite ~309-molecule cells affect fluctuations; global O/H parameters may behave differently when merged with oxide or organic cross-terms in other databases.

Corpus honesty: numeric tables and any SI-only sensitivity checks should be read from the PDF (pdf_path) rather than inferred from this wiki summary alone.

Limitations

Global O/H parameters are not a dedicated water model; 309-molecule finite size affects fluctuations; ε(0) mismatch may impact electrolyte simulations requiring accurate screening. Proton-transfer statistics are sensitive to timestep and thermostat coupling—compare SI-level settings before porting rates into other codes. JCP benchmarks should be cited whenever new oxide/water cross-terms are merged with this O/H core to document compatibility assumptions. Keep pressure units consistent when comparing isobars across tables.

Relevance to group

Foundational van Duin-group validation paper for high-T high-P aqueous ReaxFF chemistry cited whenever supercritical water appears in geochemical, hydrothermal, or combustion training discussions. Treat ε(0) caveat as a hard limit for electrolyte screening workflows.

Citations and evidence anchors

  • DOI: 10.1063/1.5031489; papers/Manzano_supercriticalH2O_JCP_2018.pdf; extract normalized/extracts/2018hegoi-manzano-the-journal-benchmark-reaxff_p1-2.txt.