Skip to content

ReaxFFMgH: reactive force field for magnesium hydride systems

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

The work introduces ReaxFF\(_{\mathrm{MgH}}\), a ReaxFF parameterization for Mg, H, and MgH\(_x\) chemistry trained on QM data for clusters and equations of state of bulk Mg and MgH\(_2\) phases. The model reproduces lattice parameters, densities, EOS, bond/angle energetics, and reaction energies for small hydride species. MD demonstrations follow H\(_2\) absorption/desorption in nanoscale MgH\(_2\), emphasizing size-dependent thermochemistry: heat of formation becomes less negative for smaller nanoparticles (roughly −16 to −19 kcal/Mg between 0.6–2 nm), trending toward the bulk value near −20 kcal/Mg beyond ~2 nm—linking nanostructure to desorption behavior relevant to hydrogen storage kinetics. The introduction motivates nanoscaling as a lever to weaken Mg–H binding relative to bulk MgH\(_2\), while noting that ball-milled powders often remain in a size regime that still looks bulk-like in thermochemistry.

Methods

ReaxFF training (checklist A)

  • Lineage / form: ReaxFF\(_{\mathrm{MgH}}\) follows the same development strategy as prior ReaxFF families (hydrocarbons, Si/SiO\(_2\), Al/Al\(_2\)O\(_3\)). Total energy includes bond/order, over/undercoordination, valence, an additional \(E_{\mathrm{MgH}}\) term to distinguish distinct Mg–H energetics in different bonding environments, plus van der Waals and Coulomb with EEM-style charge equilibration (Sec. 2, J. Phys. Chem. A 109, 851–859; DOI 10.1021/jp0460184).
  • Training structures / observables: fit to QM equations of state for Mg (hcp, fcc, bcc, simple cubic, diamond) and multiple MgH\(_2\) phases (α, β, γ, CaF\(_2\)-type, diamond-type, plus ground-state rutile-type R-MgH\(_2\) as labeled in the paper); additional cluster data for bond dissociation, angle bending, and reaction energies; EEM parameters for Mg adjusted to match DFT Mulliken charges on small Mg–H clusters (H parameters inherited from ReaxFF\(_\mathrm{CH}\)) (Sec. 2–2.2).

QM reference (checklist C)

  • Clusters / molecules: Jaguar 4.01, B3LYP, 6-311G**++ for cluster energetics; Mulliken charges for EEM training sometimes computed with a smaller basis without diffuse functions (Sec. 2.2).
  • Condensed phases: CASTEP plane-wave DFT, GGA-PBE, ultrasoft pseudopotentials; k-mesh via Monkhorst–Pack with spacing 0.1 Å\(^{-1}\); kinetic-energy cutoff 380 eV (Sec. 2.2).

Optimization / fitting

  • Procedure: successive one-parameter optimization against the QM database; parameters tabulated in Tables 5–10 with supplementary material referenced (Sec. 2.2).

Molecular dynamics (checklist B)

  • System build: Mg and MgH\(_2\) nanoparticles ~0.6–2.2 nm cut as spherical clusters from hcp Mg and R-MgH\(_2\) supercells using Cerius2 “Crystal Builder” with a bond-order cutoff (Sec. 2.2).
  • Integration: NVT with Berendsen thermostat for equilibration and annealing; timestep 0.25 fs; charges updated every MD step (Sec. 2.2).
  • Electrostatics / ReaxFF: standard ReaxFF treatment of nonbonded vdW + Coulomb with dynamic charges (Sec. 2).

MD protocol details (nanoparticle demonstration): The application section reports molecular dynamics of H\(_2\) absorption/desorption in MgH\(_2\) nanoparticles using ReaxFF\(_{\mathrm{MgH}}\). N/A — specific MD package named in the short extract—confirm in papers/cheung-et-al-2005-reaxffmgh-reactive-force-field-for-magnesium-hydride-systems.pdf. System: spherical Mg and MgH\(_2\) nanoparticles ~0.6–2.2 nm cut from hcp Mg and R-MgH\(_2\) supercells (Cerius2 “Crystal Builder”, bond-order cutoff) as stated above. Boundaries / periodicity: N/A — full PBC vs vacuum boundary wording not recovered line-by-line from the excerpt; the nanoparticle demo is described as clusters carved from bulk cells—see Sec. 2.2 in the PDF. Ensemble: NVT with Berendsen thermostat for equilibration/annealing. Timestep: 0.25 fs. Duration / stages: ps/ns timing for full equilibration/production is not restated beyond “equilibration and annealing” in the excerpt—see PDF for schedule. Thermostat: Berendsen; N/A — coupling constant in the indexed text. Barostat: N/A — NPT not stated for this NVT nanoparticle block. Temperature: 552 K appears in the introduction as a cited bulk dehydrogenation condition (1 atm), not necessarily every production MD temperature—verify trajectory conditions in the PDF. Pressure / stress: N/A — hydrostatic pressure control for the excerpted NVT demo. Electric field: N/A. Replica / enhanced sampling: N/A.

Findings

  • FF quality (static tests): ReaxFF\(_{\mathrm{MgH}}\) reproduces the QM-derived cell parameters, densities, and EOS trends for the Mg and MgH\(_2\) phases included in training, and matches cluster bond/angle scans and reaction energetics within the figures/tables reported in Sec. 3 (J. Phys. Chem. A 109, 851–859).
  • Nanoparticle thermochemistry: heat of formation becomes less exothermic as particle size decreases—roughly −16 to −19 kcal/(mol Mg) for ~0.6–2.0 nm particles, approaching the bulk ~−20 kcal/(mol Mg) by ~2 nm and above—so ball-milled 20–100 nm particles remain near-bulk in this model (Sec. 3 / abstract).
  • Implication stated by the authors: size-dependent thermochemistry motivates exploring smaller nanoscale hydrides for H\(_2\) release conditions relative to bulk MgH\(_2\) (high dehydrogenation temperature at 1 atm cited in the introduction).
  • Limitations (authors / model): kinetics of H\(_2\) evolution includes barriers, catalysis, and restructuring beyond a single nanothermochemistry demonstration; FF transferability follows the usual ReaxFF bounds of the training set.

Limitations

  • Kinetics of H\(_2\) release involves kinetics barriers, surface reconstruction, and catalysis not fully captured in a single nanoparticle thermochemistry demo.
  • QM training coverage bounds transferability to alloys, defects, and extreme conditions.

Relevance to group

Core ReaxFF lineage paper with Adri C. T. van Duin on metal hydrides and hydrogen-storage motivated MD—directly adjacent to materials energy applications.

Citations and evidence anchors

  • Primary citation via PDF metadata; add journal DOI in front matter when confirmed from the published version if this file is a preprint capture.