ReaxFF reactive force field for molecular dynamics simulations of liquid Cu and Zr metals
Evidence and attribution¶
Authority of statements
Training grids and validation metrics follow J. Chem. Phys. DOI 10.1063/1.5112794 and normalized/extracts/2019h-s-huang-j-chem-phys-reaxff-reactive_p1-2.txt.
Summary¶
Liquid metals underpin casting, fusion materials, and alloy processing, yet large-scale reactive MD needs accurate yet cheap interatomic models. This paper develops a ReaxFF description for Cu and Zr single-element liquids and crystals, fitting to VASP PAW-PBE equations of state across multiple crystal prototypes (fcc, bcc, hcp, a15, sc, dia) and low-index surface energies. Validation spans melting (liquid–crystal–liquid sandwich simulations), density, pair correlation functions, self-diffusion, and Voronoi local order in liquid and supercooled regimes, compared to experiment and EAM references. The work is explicitly positioned as expanding ReaxFF metallic capabilities for elemental liquids where EAM may be accurate for structures yet less flexible when bond-order-like effects matter in multicomponent extensions. For corpus navigation, keep this paper linked near alloy solidification and liquid-metal interface topics even though the present fit is pure Cu and Zr.
Methods¶
DFT: PBE PAW, 500–650 eV cutoffs, 0.2 Å⁻¹ k-spacing, tight SCF/force tolerances, 15 Å slab vacuum. ReaxFF: bond + overcoordination + vdW metal formulation with bond-order coupling; least-squares fit to DFT databases. MD: AMS ReaxFF integrator 0.25 fs, 125 ps windows; EAM baselines in LAMMPS (1 fs, 1 ns) using literature Cu and Zr EAM files. Melting: ~4116-atom Cu and ~4374-atom Zr sandwiches, NPT 0 pressure, interface velocity tracked.
MD validation environments. Melting and liquid benchmarks use three-dimensional periodic supercells with ~4116 Cu or ~4374 Zr atoms in sandwich geometries, NPT at 0 pressure with a Parrinello–Rahman-class barostat as described in JCP (pdf_path). AMS/ReaxFF trajectories use 0.25 fs timestep and 125 ps windows; LAMMPS EAM comparisons use 1 fs and 1 ns production. Thermostat: Nose–Hoover (or equivalent) details accompany those engines in the article. Temperature: thermal ramps and liquid/crystal hold points in K follow the schedules tabulated in the article for each element. External electric field: N/A. Enhanced sampling: N/A for the quoted melting and diffusion benchmarks.
Findings¶
ReaxFF reproduces liquid g®, diffusion, and interface behavior for Cu and Zr over the temperature ranges tested. Melting temperatures align more closely with experiment than the selected EAM potentials in the authors’ comparison, supporting use of ReaxFF for liquid-metal scenarios where bond-order flexibility matters. Supercooled liquid metrics and Voronoi local order further stress-test the parameterization beyond melting alone.
Limitations¶
Binary Cu–Zr or impurity chemistry requires additional training data beyond this elemental set. Liquid metal chemistry involving oxidation is not the focus of this benchmark. EAM comparisons depend on which EAM files are chosen—authors cite specific parameterizations that may not represent the latest experimental consensus for pure Cu/Zr melting. Archive the parameter file identifiers used in your LAMMPS scripts when reproducing their tables.
Relevance to group¶
A. C. T. van Duin co-authorship adds a liquid-metal ReaxFF benchmark complementary to oxide and organic lines in the corpus and supports method comparisons against EAM baselines in LAMMPS workflows.
Citations and evidence anchors¶
- DOI
10.1063/1.5112794;papers/Huang_JCP_2019_CuZr_liquid.pdf; extractnormalized/extracts/2019h-s-huang-j-chem-phys-reaxff-reactive_p1-2.txt.