Oxidation of Tungsten at Room Temperature Irradiated by Oxygen Plasma
Summary¶
A W–H–O ReaxFF parametrization is developed with DFT references from ADF (molecular scans) and VASP (PBE PAW, 520 eV, DFT-D3, Γ-centered 10×10×1 k-meshes, 20 Å vacuum for slabs), including ZBL close-range W–W and W–H pairs and W–O training against bulk, surface, oxide EOS, and oligomer energetics. LAMMPS ReaxFF simulations prepare BCC W(001) slabs (28×28×84 Å, 4374 W), equilibrate to 300 K, then cumulatively deposit 4500 O atoms at 1, 10, or 30 eV (below physical W sputtering by O at ~40 eV as cited), with 1 fs timestep, cascade evolution, and 0.5 ps Langevin cooling to 300 K between impacts. Additional runs probe O reflection and chemical sputtering from pre-oxidized surfaces.
Methods¶
- FF training: Single-parameter search minimizing QM vs ReaxFF error; EEM charges vs Mulliken PBE; W BCC lattice, defect, surface, H interstitial/adsorption data; WO\(_x\) EOS, surface energies, oxide molecule ΔH ladder (Section 2.1).
- Surface preparation: Steep descent minimization (10\(^{-8}\) force tolerance); Langevin heating to 300 K; NVE relaxation; cleave to 2D periodic slab; repeat; 0.5 fs timestep with charge equilibration each step.
- O irradiation: 4500 sequential O impacts per scenario; impact area ~529 Å\(^2\) with 5 Å buffer; interval 3–4 ps between impacts depending on energy; 3000–4000 steps per impact; post-impact Langevin 300 K cooling 0.5 ps; reported fluxes ~2.1–2.83×10\(^{32}\) O m\(^{-2}\) s\(^{-1}\).
- Analysis: Adlayer thicknesses, O depth profiles, n\(_O\)/n\(_W\) ratios approaching WO\(_3\)-like stoichiometry in upper bins; separate 1000-impact studies on low vs high O coverage slabs (A-deps / B-deps).
1 — MD (O bombardment / cascade + cooling). Engine: LAMMPS ReaxFF; 1 fs timestep in O-impact stages; EEM + 0.5 fs during equilibration as stated. System: BCC W(001), 4374 W in ~28×28×84 Å; O cumulative 4500 impacts, 1 / 10 / 30 eV; NVE during impact, 0.5 ps Langevin to 300 K between events; 2D PBC slab. Barostat, NPT, E-field, umbrella, MTD: N/A in the summarized protocol. 2 — W–H–O ReaxFF training in the first FF bullet; 3 — Static QM (ADF + VASP) in the summary lede and DFT bullet list.
Findings¶
- Cumulative low-energy O impacts build mixed W–O adlayers 12–23 Å thick (depending on 1–30 eV scenarios) atop W(001) before saturation, after which additional O predominantly reflects or ejects adlayer species (Figure 4 trends).
- Higher impact energy drives O deeper before saturation (O extends up to ~5–13 Å into the metal depending on case), altering saturation fluence vs 1 eV impacts.
- n\(_O\)/n\(_W\) profiles show ≥3:1 O:W in upper adlayer bins, with B–A–B interfacial regions nearer ~1.7:1 as described—linking atomistic profiles to WO\(_x\)-like oxidation fronts.
- Follow-on single-impact studies on oxidized slabs quantify O reflection vs sputtering channels that matter for plasma-facing W under oxygen impurity flux. Sensitivity to impact energy and O coverage is the main parametric axis in the text above; J. Phys. Chem. C ASAP PDF (issue/pages per DOI) is authoritative.
Limitations¶
Classical ReaxFF cannot capture electronic sputtering or full fusion D/T inventories; 300 K substrate cooling between ps-scale impacts is a stand-in for experimental flux/time separation.
Relevance to group¶
van Duin/Shin collaboration on fusion-relevant W oxidation with ReaxFF plasma-like O bombardment.