Unveiling the intricacies of steel corrosion induced by chloride: Insights from reactive molecular dynamics simulation
Summary¶
Shen and colleagues develop Fe/Cl extensions for ReaxFF to study chloride-assisted corrosion of an Fe(100) surface in alkaline aqueous electrolyte using LAMMPS. Cluster DFT (B3LYP-D3(BJ), 6-311++G(2df,2p)) and periodic CASTEP (GGA-PW91) data train and validate bond dissociation, angle distortion, adsorption energies (top/bridge/hollow Cl on Fe(100)), and Mulliken charges. The NVT trajectory (300 K, Δt = 0.1 fs, 500 ps total) for a ~34.4 × 34.4 × 66.3 Å\(^3\) cell with 3168 Fe atoms and explicit water/Na\(^+\)/Cl\(^-\)/OH\(^-\) at pH 13.5 and 1 mol/L Cl\(^-** captures **oxide growth**, **Fe dissolution**, **chloride catalysis**, and **competitive OH\(^-\) vs Cl(^- adsorption.
Methods¶
1 — MD application (ReaxFF, §2.3). Engine / code: LAMMPS with ReaxFF molecular dynamics. System: Fe(100) slab in ~34.4 × 34.4 × 66.3 Å\(^3\) cell, 3168 Fe atoms, 1160 H\(_2\)O, 27 Na\(^+\), 20 Cl\(^-\), 7 OH\(^-\) (pH 13.5, 1 mol/L Cl\(^-\)). Boundaries: PBC in x,y; fixed z with reflective upper wall. Ensemble / T: NVT at 300 K with NVT thermostat settings as in §2.3 (damping/algorithm in VOR PDF if not repeated here). Timestep / duration: Δt = 0.1 fs; 500 ps total trajectory; snapshots every 1000 steps; OVITO for film analysis. Barostat / pressure control: N/A for this NVT slab setup. Electric field / external bias: N/A. Replica / metadynamics / umbrella: N/A in the reported protocol.
2 — Force-field training. Parent / scope: Fe/Cl extensions to ReaxFF for aqueous Fe corrosion chemistry. QM reference: B3LYP-D3(BJ)/6-311++G(2df,2p) cluster data; CASTEP GGA-PW91 periodic (2×2) Fe(100)+Cl slab (4 layers, 15 Å vacuum, spin-polarized, 340 eV cutoff, 4×4×1 k mesh) for E_ads Cl (top/bridge/hollow) and Mulliken data. Training set / optimization: Fe–Cl bond curves, angle distortions, adsorption energies; weighted ReaxFF error functional (§3.1); Tables 2–4 for parameters. Reference data: QM primary; RDF peaks compared to published Fe–Cl lengths in Findings.
3 — Static QM — used for parameterization and spot validation; not a separate static-only DFT application paper beyond the training set.
Findings¶
Corrosion sequence. Early dynamics show oxide initiation from OH/O interaction, then Cl\(^-\) accumulation that weakens Fe–Fe bonding and promotes Fe dissolution and vacancy formation, consistent with catalytic chloride roles emphasized in the discussion. Charge analysis tracks Fe oxidation, O reduction, and Cl\(^-\)-mediated electron transfer patterns over 5–500 ps.
Mechanistic detail (§3.2.2). The paper traces short-lived intermediates (e.g., Fe(OH)\(^+\) formation and Cl substitution steps) culminating in Fe(OH)\(_2\)-like products, with Cl\(^-\) argued to remain surface-catalytic rather than bulk-intercalated—supporting a catalytic picture over direct bulk chlorination.
RDF and morphology. Fe–O and Fe–Cl RDFs align with Fe–Cl ~2.35 Å vs a ~2.32 Å experimental mean cited in the text. Surface roughening and pitting progress through nucleation–growth–stabilization by 500 ps.
Authored model limits — short ns-scale sampling and simplified electrolyte (see ## Limitations).
Limitations¶
Short nanosecond-scale trajectories and finite electrolyte model do not capture long-time passivation or macroscopic transport in cement pores; ReaxFF accuracy is lower far from equilibrium bond lengths (noted in validation).
Relevance to group¶
Adri C. T. van Duin co-authors Fe/Cl ReaxFF development applied to infrastructure corrosion chemistry—useful reference for aqueous Fe oxidation parameterization.
Citations and evidence anchors¶
- Simulation parameters: §2.3; validation: §3.1; kinetics/mechanism: §3.2–3.2.4 (Constr. Build. Mater. 443, 137839 (2024)).