Skip to content

A ReaxFF Force Field for 2D-WS2 and Its Interaction with Sapphire

Scope

New ReaxFF parameters for W/S/H/Al/O describe monolayer and few-layer WS\(_2\) (phases, defects, edges, grain boundaries) and WS\(_2\)/sapphire interfaces, trained on quantum-mechanical data and checked against ADF-STEM experiments reported in the article.

Summary

The work introduces a ReaxFF parameter set for 2D WS\(_2\) and its interaction with sapphire, aimed at large-scale reactive molecular dynamics of synthesis-relevant structure and defects. The training set includes periodic and nonperiodic quantum data on semiconducting and metallic WS\(_2\), point and line defects (including ripplocations), edge excess energies at different sulfur coverages, and W(SH)\(_2\)S\(_2\) bond/angle energetics. For the metal–oxide interface, data cover corundum Al\(_2\)O\(_{3-x}\)S\(_x\), sulfur in fcc Al, and sulfur adsorption and diffusion on Al(100) and Al(111). Benchmark molecular dynamics and comparison to post-training DFT and annular dark-field scanning transmission electron microscopy (ADF-STEM) experiments are used to assess the model; fitting details, parameters, and extended methods are documented in the Supporting Information.

Methods

Force-field training (ReaxFF for W/S/H/Al/O). A ReaxFF set is fitted to DFT/QM reference data covering 2D WS\(_2\) (2H/1T, defects, edges, grain boundaries), W(SH)\(_2\)-class local hydration motifs, and sapphire-relevant Al–O–S and Al–S environments (corundum Al\(_2\)O\(_3\)-based training points, S in fcc Al, and S adsorption/diffusion on Al(100) and Al(111)), with tables in the J. Phys. Chem. C paper and SI. Optimization uses the group’s ReaxFF Monte Carlo / CMA-ES-style workflow; the parameter file is in the SI.

MD application (LAMMPS ReaxFF). LAMMPS with the new ReaxFF is used for slab/ribbon and WS\(_2\)/sapphire case studies (phase changes, defect transport, interface morphology). NVT thermostat at reported temperatures in K and ps–ns-scale trajectory lengths as tabulated in Section 2 and the Results of J. Phys. Chem. C (exact K and fs values in pdf_path, not re-keyed on this page); 3D PBC with slab buffer freezing as specified in Section 2. NPT stress tests only where listed in the VOR+SI. Barostat — N/A in strict NVT defect blocks; NPT only where the text lists it. Hydrostatic pressure in (bar): N/A as a separate user knob in NVT defect blocks; 1 bar-style NPT only in dedicated segments. E-field, metadynamics, shock — N/A in the benchmarks summarized in the abstract-level summary on this page.

Static QM (training only). DFT is the reference for fitting; not a separate PES publication beside the ReaxFF work.

Experiments. ADF-STEM of CVD WS\(_2\) on sapphire where cited for interface morphology.

Findings

Model behavior. The ReaxFF recovers the 2H↔1T displacive trend, S-vacancy migration barriers, line-defect energetics (including ripplocations), 1T domain nucleation on 2H basal/edge regions, and grain-boundary trends vs S chemical potential (including S- vs W-terminated edges). Coalesced WS\(_2\) on sapphire is used to discuss epitaxy-relevant shapes.

Comparisons. Post-hoc DFT checks and STEM images support selected barriers/structures; tabulated meV/Å values are in the PDF/SI, not re-keyed on this page.

Sensitivity and levers. T, S chemical potential / coverage, and substrate orientation shift predicted defect populations and morphology.

Limitations

Training and validation scope follow the systems and conditions in the article and Supporting Information; transfer to other substrates or chalcogen chemistries requires separate assessment.

Relevance to group

Develops and validates group-lineage ReaxFF for a major 2D TMD system and its oxide substrate, with joint experimental (STEM) validation.

Citations and evidence anchors

Reader notes (navigation)

  • Cross-link other 2D TMD or oxide-interface theme hubs in wiki/concepts/ when curated.