Skip to content

Illuminating Invisible Grain Boundaries in Coalesced Single-Orientation WS2 Monolayer Films

Summary

Achieving wafer-scale single-crystal TMD films requires understanding coalescence defects that may be invisible in bright-field imaging yet limit electronic homogeneity. Hickey et al. combine electron microscopy and ReaxFF molecular dynamics to connect MOCVD growth conditions to grain-boundary microstructure in coalesced monolayer WS₂ films on 2 in. c-plane sapphire substrates. Dark-field TEM reveals defect arrays in films that selected-area electron diffraction and imaging show are nearly single-orientation yet imperfectly stitched; atomic-resolution STEM classifies translational grain boundaries. Films tilt out-of-plane when released from the substrate. The authors use statistical facet analysis over >1300 facets to relate nanoscale building blocks to multimicrometer morphology, bridging sub-Ångstrom structure with wafer-scale texture.

Methods

Experiments (CVD, microscopy, morphology)

Monolayer 2H-WS₂ is grown by MOCVD on 2 in. c-plane sapphire as described in Nano Lett. transferred films are imaged on TEM grids. SAED and bright-field TEM show millimeter-scale single-orientation regions; dark-field TEM (DF-TEM) reveals sub-micron faceted contrast not obvious in bright-field. ADF-STEM classifies local GB structures; the article reports >1300 facet measurements to connect nm-scale building blocks to μm-scale texture when Wulff-type arguments are invoked.

MD application (ReaxFF)

Hickey et al. use ReaxFF molecular dynamics—implemented in a LAMMPS-class workflow consistent with the cited ReaxFF literature and figure captions (see pdf_path/SI for the precise parameter set and input decks). Representative panels show ReaxFF MD structures equilibrated at 300 K alongside STEM-derived local models for type A vs type B translational boundaries and their transition segment. Boundary / periodicity: PBC slab supercells for the freestanding GB models as in the SI—N/A — full lateral cell vectors if not copied into this note. N/A — full supercell atom counts, 0.25 fs-class timestep if used, NVT thermostat name, and ps–ns run lengths: take from Supporting Information rather than this note. N/A — NPT barostat, external electric field, and shock/MSST in the main GB workflow. N/A — umbrella or metadynamics sampling for the primary GB case unless the SI states otherwise.

Force-field training in this work

N/A — this article applies an established 2D TMD/ReaxFF description (with references in the Nano Lett. text) to support GB models; the paper is not primarily a de novo CMA-ES/ParReax training report for a new parameter set.

Findings

  • Multiscale structure: Statistical facet analysis and TEM/STEM together show that nearly single-crystal SAED can coexist with widespread re-entrant faceting and invisible (in bright-field) defect arrays exposed by DF contrast.
  • GB types: The work distinguishes at least two translational-mismatch GB classes in coalesced films (including type A and B motifs in the figures) and a transition zone with mixed 3|W and 4|W-like W-site arrangements at the boundary core.
  • Kinetics link: The authors connect faster growth conditions to the observed defect catalog (see main text for the growth-rate comparison and figure references).
  • Simulations / experiment link: ReaxFF simulations (300 K equilibration in the figure captions) reproduce the out-of-plane tilt and AB-stacking motifs discussed for free-standing type B regions, supporting a chemically reasonable atomistic picture versus the imaging-derived models.
  • Comparisons / limitations: Where edge energetics or Wulff arguments appear, those benchmarks are in the paper; ab initio agreement is not the headline—## Limitations and the version-of-record PDF carry caveats for ReaxFF on S–W edges and for TEM projection effects. Corpus honesty: do not reuse lattice constants or angle labels without checking VOR pdf_path and supplementary figures.

Limitations

MD accesses local segments and short times relative to CVD reactor-scale processes; STEM contrasts may miss certain defect types. ReaxFF accuracy for WS₂ edges depends on the parameterization cited in the paper.

Confidence rationale: high—primary Nano Lett. article with abstract-level alignment to extract.

Citations and evidence anchors

DOI: 10.1021/acs.nanolett.1c01517. Nano Lett. 2021, 21, 6487–6495. ACS provides supporting information downloads alongside the article page; consult there for extra micrographs or methods not duplicated in this wiki.

Reader notes (navigation)

Coalescence microscopy pairs naturally with kinetic Wulff discussions on growth pages; keep cross-links explicit so retrieval can hop from TEM evidence to ReaxFF edge models.