Challenges in Molecular Dynamics of Amorphous ZIFs Using Reactive Force Fields
Summary¶
The authors replicate literature melt-quench workflows that use the Yang et al. ReaxFF parameterization for zeolitic imidazolate frameworks (ZIFs) in LAMMPS and compare the resulting amorphous models to ab initio references and experiments. They conclude that outputs are highly sensitive to quench rate, thermostat/barostat choices, and supercell size, and that standard ReaxFF-quenched glasses can deviate strongly from AIMD and experiment, especially in local coordination statistics. The paper is explicitly cautionary for anyone porting ZIF melt-quench recipes across software versions or box sizes without revalidation.
Methods¶
1 — MD application (atomistic dynamics). The study uses LAMMPS with the Yang et al. ReaxFF for ZIF-4 (see J. Phys. Chem. C Computational Methods for the full parameter citation). Timestep 0.25 fs. System: a (2×2×2) ZIF-4 supercell of \(\sim 2176\) atoms with 3D PBC, following Yang et al. Melt, quench, and analysis stages are run mainly under isotropic NPT with a Nosé–Hoover thermostat and barostat; by default the paper fixes \(\sim 100\,\text{fs}\) and \(\sim 1000\,\text{fs}\) damping for temperature and pressure coupling (values retested in sensitivity analysis). NVT trajectories of \(\sim 1\,\text{ns}\) on ReaxFF-quenched glasses provide RDFs, angular distributions, coordination statistics, and mass density (sampling every 125 fs for RDFs in the published protocol). The authors replicate the published melt–quench schedule, then sweep quench rate, thermostat/barostat details, and (2×2×2) vs other supercell choices; representative input files are archived in the authors’ citable data repository (URL in the PDF). N/A — static external electric field in the production protocol; N/A — umbrella or metadynamics (standard melt–quench only). ReaxFF QEq and nonbonded cutoffs follow the Yang ZIF ReaxFF + LAMMPS conventions detailed in the article.
2 — Force-field training. N/A — the work applies the published Yang ZIF ReaxFF; it does not refit the parameter set.
3 — Static QM / DFT and AIMD (reference). The paper reports shorter PBE+TS-class VASP ab initio trajectories (\(\sim 60\,\text{ps}\) NVT segments) on melt–quenched ZIF-4 glasses for direct comparison to ReaxFF; PBE/PAW/k-mesh and related settings are in J. Phys. Chem. C.
4 — Experiments. N/A as a primary new experimental campaign; structural comparisons use ab initio data and literature Raman/SAXS-style information as referenced in the discussion.
Findings¶
Outcomes and mechanisms. Reproducing published Yang-style melt–quench procedures still leaves RDFs, coordination (including Zn–N and imidazolate connectivities), and angular statistics sensitive to small thermostat/barostat and quench choices. ReaxFF-quenched ZIF-4 glasses can differ markedly from the PBE+TS ab initio melts in the same study and from known experimental signposts for amorphous ZIFs, including unphysical local coordination that propagates to medium-range order, density, and derived properties.
Comparisons. Direct ReaxFF vs VASP-AIMD on ZIF-4 glasses; RDF peak positions and amplitudes shift with protocol enough to make blind transfer of RDFs between publications unsafe without revalidation. Literature experimental trends enter as context, not as new in-house synthesis data in this paper.
Sensitivity and design levers. Quench rate, thermostat/barostat damping choices, and box size are the main sensitivity axes demonstrated in the article.
Corpus / KB honesty. Cite numerical RDF/density values from the VOR PDF; this wiki emphasizes reproducibility warnings from the abstract and Discussion.
Limitations¶
AIMD length and size limits constrain reference data; force-field error may be entangled with protocol choices. Users should treat melt-quench “failure” here as a workflow sensitivity statement: modest thermostat tweaks or longer liquid annealing can shift glass densities enough to change conclusions about whether ReaxFF agrees with AIMD, so reproduction requires archiving full input decks rather than headline quench rates alone. Anyone reusing the Yang ZIF parameter set should therefore treat RDF/coordination checks as mandatory regression tests when LAMMPS builds or supercell sizes change.
Relevance to group¶
Critical assessment of ReaxFF melt-quench for MOF glasses—directly relevant to users of ZIF ReaxFF workflows. Anyone porting Yang-parameter melt recipes into new LAMMPS builds should treat this paper as a mandatory regression check on coordination statistics before publishing new amorphous structures.
Citations and evidence anchors¶
Reader notes (navigation)¶
- Critical ZIF / MOF glass melt-quench methodology; compare with porous-material applications in theme-reactive-md-corpus.