Development of a ReaxFF Reactive Force Field for Pt/Cl Systems with Application to Platinum Metal Etching with Chlorine and Hydrogen Chloride Gases
Summary¶
A Pt/Cl/H ReaxFF parameter set is trained on VASP GGA-PBE data for Pt slabs, Cl and HCl adsorption on Pt(100) and Pt(111), PtCl\(_2\) crystal equations of state, and Cl diffusion barriers (CI-NEB), then coupled to existing Pt/O/H and Cl/O/H ReaxFF libraries. LAMMPS NVT simulations at 1500 K (Berendsen thermostat, 0.25 fs timestep) study Cl\(_2\) and HCl attack on Pt(100) and Pt(111) slabs and on a cuboctahedral Pt nanoparticle, comparing chlorination, etching, and relative etchant behavior to experiment.
Methods¶
- DFT training data: VASP PAW GGA-PBE; 450 eV cutoff; (14×14×14) k mesh for bulk Pt; Pt slabs 6 layers with relaxation of top 3 layers + adsorbates; dipole correction; adsorption energies from the article’s formula; CI-NEB (3 images) with 5.0 eV/Å\(^2\) spring and 0.05 eV/Å force criterion; k meshes per coverage in Table S1 (Supporting Information).
- ReaxFF optimization: Successive one-parameter search minimizing weighted QM vs ReaxFF errors for Pt–Cl bonded/off-diagonal/angle/dihedral terms combined with Pt/O/H and Cl/O/H sets as described in §3.1.
- MD: Slab box 110 × 110 × 256 Å; 9600 Pt per slab with 4500 Cl\(_2\) or 6000 HCl molecules initially; cuboctahedron ~586 Pt NP in 65–78 Å cubic boxes with 765–1020 etchant molecules; NVT, 1500 K, 100 fs Berendsen damping; 2 ns slabs / 1 ns NP; 0.25 fs timestep.
Analysis outputs tied to etching chemistry. Trajectories monitor surface chloride coverage, volatile Pt\(_x\)Cl\(_y\) evaporation events, and facet-resolved metal loss curves used to compare Cl\(_2\) versus HCl efficiency. RDF and coordination metrics (as plotted in the article) connect subsurface Cl penetration depths to passivation differences between (100) and (111) terraces.
1 — Application MD. LAMMPS ReaxFF; NVT 1500 K; 0.25 fs timestep; Berendsen thermostat ( 100 fs damping as stated in summary); 2 ns (slabs) / 1 ns (NP) production; ~9600 Pt slabs, cuboctahedral NPs, PBC boxes. NPT /barostat: N/A. E-field, umbrella, MTD: N/A in the summarized work. 2 — ReaxFF / Pt–Cl training (DFT, §3.1) in the first bullets. 3 — DFT used for training data, not a standalone PES survey paper.
Findings¶
- The fitted field reproduces key QM trends for bulk Pt, PtCl\(_2\) phases, surface Cl/H configurations, and diffusion barriers well enough for the subsequent MD survey (figures/tables in the article and SI).
- Pt(100) shows higher susceptibility to chlorination and Pt removal than Pt(111), which remains more passivated against Cl penetration, slowing formation of volatile Pt\(_x\)Cl\(_y\) species.
- HCl vs Cl\(_2\) etching ratios from simulation align satisfactorily with experimental references cited in the paper, supporting use of the model for atomistic Pt halogen etching scenarios. Facet and etchant comparisons and RDF-based subsurface Cl diffusion narratives are the main sensitivity levers discussed; T is fixed at 1500 K in the long runs summarized. J. Phys. Chem. A PDF is authoritative.
Limitations¶
1500 K aggressive etching conditions accelerate chemistry; transferability to industrial ALE-like conditions and alloy/contaminated Pt surfaces is not established in the main text. If this note and the PDF disagree on numerical detail, treat the peer-reviewed article as authoritative.
Relevance to group¶
van Duin-group ReaxFF parametrization for transition-metal halogen chemistry with Shin / Fichthorn collaboration.