Nanoscale structure and dynamics of water on Pt and Cu surfaces from MD simulations
Summary¶
Classical molecular dynamics with the third-generation charge-optimized many-body (COMB3) Pt/O/H potential reproduces DFT-supported sub-monolayer water motifs on Pt(111), including √37 and √39 reconstructions with “575757” di-interstitial defects. For liquid films, the first wetting layer shows a buckled bilayer-like distribution from the balance of metal–water adsorption and H-bonding. Molecular kinetic spreading analysis shows water nanodroplets spread an order of magnitude slower on large Pt than on close-packed Cu due to interfacial molecular packing differences. The study is positioned as a fixed-bond alternative to reactive treatments when the scientific question is wetting morphology and kinetic spreading rather than dissociative surface chemistry.
Methods¶
A — Interatomic potential (classical, fixed reactive chemistry)¶
- COMB3 third-generation charge-optimized many-body potential for Pt/O/H; Cu-related wetting comparisons use the COMB treatment described in the article (not ReaxFF).
B — Molecular dynamics (LAMMPS)¶
- Engine: LAMMPS with
field.comb3.PtOHparameter file referenced in the paper. - Systems: Pt(111) sub-monolayer ice/water motifs (0.67 ML hexagonal, √37/√39 reconstructions, 575757 defects) and multilayer liquid films; nanodroplet spreading on large Pt vs Cu surfaces.
- System size & composition: Metal–water slab supercells containing hundreds to thousands of atoms (exact counts in Langmuir Methods).
- Boundaries / periodicity: In-plane periodic boundary conditions with finite slab thickness/vacuum as standard for surface MD.
- Ensemble: NVT canonical trajectories for wetting and spreading benchmarks (per paper_keywords).
- Timestep: Femtosecond integration timestep (exact fs in Methods).
- Duration: Equilibration and production run lengths in ps/ns tabulated in the article.
- Thermostat: Nose–Hoover or Langevin thermostat parameters as listed for each stage.
- Barostat: N/A — hydrostatic barostat not used for the cited constant-volume surface NVT cells unless SI documents NPT film thickness relaxation—verify in PDF.
- Temperature: K setpoints for sub-monolayer vs liquid-film studies.
- Pressure: N/A — isotropic pressure targets not applied for the NVT slab setups described at abstract level.
C — Pure DFT (validation literature)¶
- DFT references from the literature benchmark COMB3 sub-monolayer structures on Pt(111).
D — Analysis¶
- Molecular kinetic theory treatment of spreading rates tied to interfacial density profiles and first-layer buckling motifs discussed in Results.
Findings¶
- COMB3 captures experimentally reported low-coverage ice/water motifs on Pt(111), including √37/√39 networks and 575757 defects, consistent with DFT literature cited in the paper.
- The first liquid layer next to Pt(111) exhibits a characteristic two-layer buckling pattern tied to adsorption vs hydrogen-bond competition.
- Spreading rates from molecular kinetic theory differ by about one order of magnitude between Pt and Cu for the large-surface droplet setups described, traced to interfacial water density profiles.
- The discussion ties first-layer buckling and √37/√39 motifs to kinetic spreading: interfacial packing differences between Pt and Cu propagate into order-of-magnitude changes in predicted spreading rates for large droplets.
Limitations¶
Not ReaxFF: chemistry is fixed-bond COMB3; dissociative chemistry is outside the model’s scope. van Duin appears via the Penn State author list as collaboration context, not as ReaxFF development here. COMB3 Pt/O/H parameters should not be mixed ad hoc with ReaxFF water models in the same simulation cell without a defined multiscale handshake, because the electrostatic and bonded partitioning differ fundamentally. Spreading comparisons between Pt and Cu assume comparable droplet sizes and thermostat protocols; confirm boundary conditions when reusing published kinetic coefficients.
Relevance to group¶
Penn State Sinnott/Liang-adjacent classical potential work on metal–water interfaces, relevant to benchmarking and multiscale contexts around reactive methods.
Citations and evidence anchors¶
- DOI:
10.1021/acs.langmuir.8b02315.