Skip to content

Comparative molecular dynamics study of fcc-Al hydrogen embrittlement

Evidence and attribution

Authority of statements

Prose summarizes the published Corrosion Science article identified by doi and pdf_path. Duplicate proof ingests: 2015verners-corrosion-sc-comparative-molecular-2, 2015verners-corrosion-sc-comparative-molecular-3.

Summary

Reactive MD (ReaxFF via LAMMPS) probes hydrogen embrittlement in fcc Al nanoslabs with oxidized surfaces and notched Al/Al₂O₃ interfaces, comparing H distributions tied to initial vacancies vs random bulk H. Simulations connect H diffusion, plasticity, dislocation activity, and void/ decohesion motifs to environmentally assisted failure scenarios. The corrosion framing emphasizes hydrogen uptake scenarios that can accompany oxide-covered Al in aqueous or humid service environments, translated here into atomistic loading tests.

Methods

LAMMPS ReaxFF (papers/Verners_AlH_CorrosionScience_2015.pdf, §2) simulates a 3D periodic oxidized fcc Al nanoslab (~15.23 nm × 0.977 nm in-plane along [1̄10] and [11̄2], ~12.34 nm metal thickness) with (111) faces covered by ~1.271 nm amorphous Al₂O₃, a notch on one oxide face, and defined Al/Al₂O₃ stacking. 162 H atoms are placed either on 162 Al vacancies (paired) or randomly in bulk Al without vacancies—both strongly supersaturated cases. Plane-stress-like tensile cycling applies 0.25% strain along [1̄10] over 0.5 ps with 2.5 ps relaxations at 5 × 10⁻⁶ fs⁻¹ nominal rate, 0.2 fs timestep, 573 K, NPT with the strained dimension fixed and lateral stresses relaxed to 0 Pa via Nosé–Hoover thermostat (100 fs damping) and barostat (10,000 fs lateral pressure damping, three-chain volume coupling as referenced). No electric field or enhanced sampling is reported.

Force-field training: The study merges published Al/O, Al/H, O/H, and Al/O/H ReaxFF subsets (Table 1) and extends Al/H with VASP GGA-PBE PAW data on 32-atom fcc Al supercells (350 eV cutoff, 9×9×9 k-mesh, Methfessel–Paxton σ = 0.2) for interstitial H, vacancy–H, and diffusion-related energies—targeted augmentation, not a full new global parametrization. Bulk Al elastic and stacking-fault metrics are checked against DFT and experiment in the validation table.

Static QM / DFT: The VASP PBE bulk Al/H reference set above supports the ReaxFF extension; there is no separate large-scale non-ReaxFF MD stage.

Findings

Simulations show H-dependent decohesion, void growth, and H-enhanced dislocation emission / slip depending on initial H placement and defects (papers/Verners_AlH_CorrosionScience_2015.pdf). The abstract reports vacancy-associated H lowers effective H diffusivity versus random bulk H. ReaxFF elastic and fault metrics are compared to DFT and experiment in the validation table. Under identical 573 K cycling, vacancy-paired vs random bulk H shifts the balance between plastic and decohesion-dominated failure. The Discussion stresses nanoscale slab limits and ReaxFF chemistry bounds for macroscopic hydrogen embrittlement inference. For numbers and timelines prefer this version-of-record PDF; duplicate corpus paths are 2015verners-corrosion-sc-comparative-molecular-2 and 2015verners-corrosion-sc-comparative-molecular-3.

Limitations

  • Nanoscale slabs omit macroscopic crack plastic zones and complex microstructures.
  • ReaxFF oxidation/hydride chemistry remains parameterization-bound.
  • Strain-rate and loading mode choices in nanoscale MD may not map one-to-one onto compact-tension or fatigue experiments.

Relevance to group

Penn State ReaxFF application to Al hydrogen embrittlement with industrial corrosion framing.

Citations and evidence anchors

  • DOI: 10.1016/j.corsci.2015.05.008 — papers/Verners_AlH_CorrosionScience_2015.pdf.