Supporting Information: Computational Synthesis of MoS2 Layers by RMD (Hong et al., Nano Letters 2017)
Non-primary corpus PDF
This pdf_path is the Supporting Information PDF for Hong et al., Nano Letters 2017. The version-of-record letter text and reactive MD results on MoO3 sulfurization are documented under 2017hong-venue-nl-2017-01727n (papers/ReaxFF_others/Hong_MoS2_acs.nanolett_2017.pdf). See the maintainer catalog NON_PRIMARY_ARTICLE_PAPER_SLUGS.md for how SI-first slugs are handled.
Summary¶
The SI package documents reoptimization of a Mo/O/S ReaxFF description aimed at gas-phase MoOxSy intermediates and reactions during sulfurization of MoO3 surfaces toward MoS2, using MoO3 surfaces and sulfur-containing gas species in later CVD-style reactive MD. Initial parameters are taken from prior Mo/O and Mo/S ReaxFF work; the authors assemble a quantum training set of small clusters (up to 14 atoms), alpha- and beta-MoO3, and 2H/1T MoS2 crystal references, then refit bond, off-diagonal, and angle terms (including selected Al-O-Mo angles) with a weighted least-squares objective. The SI also records merge of sulfur parameters from Islam et al. and of Al/O parameters for alumina substrate models, with explicit caveats on branch transferability when combining combustion-branch Mo/O/S fits with water-branch Al/O data.
Methods¶
A — Force-field training / fitting: Mo/O/S ReaxFF reoptimization via weighted least squares / single-parameter parabolic search on QM vs ReaxFF errors. Training includes charges, structures, bond dissociation, angles, cluster cohesive energies, MoO\(_3\) and MoS\(_2\) equations of state; weights emphasize small gas-phase MoO\(_x\)S\(_y\) clusters for early sulfurization. Merges: S from Islam et al.; Al/O for Al\(_2\)O\(_3\) substrates—SI warns limited transferability across branches when Al/O not co-refit to Mo/O/S.
B — Molecular dynamics / sampling: Supporting Information focuses on parameterization; large-scale RMD protocols for MoO\(_3\) sulfurization are documented on 2017hong-venue-nl-2017-01727n.
C — DFT / static QM: Clusters: Q-Chem, PBE + dispersion, 6-31G** on O/S, Hay–Watt ECP on Mo, spin scanning. Periodic: VASP PAW, MoS\(_2\) PBE 400 eV cutoff; MoO\(_3\) vdW-DF2 550 eV; optimizations/NEB to 1e-5 eV/atom, 5e-3 eV/Å forces (SI).
D — Review / non-simulation framing: SI methods supplement for Hong et al. Nano Lett.—see primary letter for scientific conclusions.
MD note. Large-scale RMD production protocols for MoO\(_3\) sulfurization are documented on 2017hong-venue-nl-2017-01727n and its PDF, not in this SI-first file. QM training clusters in this SI reach up to 14 atoms, alongside bulk MoO\(_3\) and MoS\(_2\) reference cells. Ensemble: NVT (Nosé–Hoover) for production RMD is stated on the primary letter page; this SI does not re-list those decks. Timestep / thermostat damping / duration / PBC / barostat / temperature / pressure / electric field / enhanced sampling for production RMD: N/A — copy remaining knobs from [[2017hong-venue-nl-2017-01727n]] rather than this parameterization supplement.
Findings¶
Mulliken charges from Q-Chem on small Mo/O/S clusters match the ReaxFF charge model closely enough that charge parameters are left unchanged during the bond and angle refit. Cohesive energies for listed clusters (e.g. MoO, MoS, MoO2, MoS2, MoO3, MoS3, MoO2S, MoOS2, and larger Mo2OxSy and Mo3OxSy stoichiometries) show typical errors of a few percent in Table S1, with intentionally higher accuracy for small clusters due to weighting. Bond dissociation and angle distortion data are reproduced at least qualitatively in Figures S3-S4 of the SI. The text stresses that only DFT data generated in this study enter the refit, so numerical agreement with ReaxFF results quoted in earlier references is not guaranteed after reoptimization. Merging Al/O and Mo/O/S lines risks poor bulk or surface descriptions for alumina unless validated, as noted in the SI narrative.
Limitations¶
The SI is a methods supplement: reaction pathway and RMD synthesis conclusions belong to the primary letter page. Combined ReaxFF branches may fail outside the validated Mo/O/S and substrate test cases described here.
Reader notes (navigation)¶
- 2017hong-venue-nl-2017-01727n (primary Nano Letters article PDF in corpus)
- reaxff-family