Supporting information (part III): ReaxFF training-set weights and geometries (trainset.in style)
Abstract
Machine-readable ReaxFF training blocks in a trainset.in-style layout for Mo–S–H clusters and reactions used in the MoS₂ edge chemistry parametrization described in the parent J. Phys. Chem. Lett. article (DOI 10.1021/acs.jpclett.6b02902).
Summary¶
This corpus PDF is Supporting Information III for the MoS₂ ReaxFF work by Ostadhossein et al. It does not read as a standalone article; it supplies weighted training data for least-squares ReaxFF optimization in a format that lists CHARGE, GEOMETRY, and ENERGY sections. The extract shows atomic charges for small species (for example H₂S-like fragments), extensive bond and angle targets with per-observable weights for Mo–S–H stoichiometries such as MoS₆, MoS₄H₂, MoS₃H₂, MoS₄H₄, Mo₂S₆, and related fragments, and linear-combination reaction energies tying relative stabilities of hydrogenated sulfide and edge-like species together. The file’s role in the knowledge base is to preserve reproducibility of which observables and weights entered the fit, alongside human-readable tables in sibling SI files and the narrative validation on 2017ostadhossein-venue-research.
Methods¶
Force-field training / fitting. trainset.in-style blocks list CHARGE, GEOMETRY, and ENERGY targets with per-observable weights for Mo–S–H clusters and reaction-energy ladders used in the least-squares ReaxFF optimization described on [[2017ostadhossein-venue-research]].
MD application (atomistic dynamics). N/A — machine-readable training listings only.
Static QM / DFT. QM reference levels and DFT benchmarks for the same observables are given on SI II and in the parent letter.
Review / non-simulation framing. Supporting Information Part III (non-primary); use [[2017ostadhossein-venue-research]] for scientific narrative and DOI grounding.
Findings¶
Outcomes. The trainset shows a dense set of geometry constraints (many bonds and angles per cluster) with weights often near 0.02 for looser motifs and up to ~2.5 for angles used to stiffen key Mo-centered scaffolding, plus reaction-energy ladders tying dehydrogenation and sulfur rearrangement among MoS\(_x\)H\(_y\) fragments.
Comparisons. These weighted observables are what the optimizer matches to QM references documented on SI II and the parent letter—not to experiment directly in this raw listing.
Sensitivity / design levers. The relative weights reveal which charge, angle, and reaction-energy classes dominated the 2017 objective function for this parametrization.
Limitations / outlook. Edge catalysis, growth kinetics, and agreement with experiment are reported only on [[2017ostadhossein-venue-research]]; this PDF is a fitting artifact archive.
Corpus honesty. SI-only; no DOI in front matter—link [[2017ostadhossein-venue-research]] for bibliography.
Limitations¶
This entry is SI-only (see docs/corpus/NON_PRIMARY_ARTICLE_PAPER_SLUGS.md policy for SI packages). It lacks narrative Methods and Results, and no DOI is registered in front matter for the SI file itself. Operators must not cite this PDF as if it were peer-reviewed prose. Numerical weights are fitting artifacts; downstream users need the final ReaxFF and QM benchmarks from the parent work.
Confidence rationale: med because the page is complete for its corpus role but the canonical scientific story lives on the primary article page.
Reader notes (navigation)¶
- Parent letter and full context: 2017ostadhossein-venue-research
- Related SI parts: 2017ostadhossein-venue-microsoft-word-2
- Theme hub: theme-2d-epitaxy-growth
- reaxff-family
- Benchmarks: Phase 0 warmup questions