Atomistic insights into Cu chemical mechanical polishing mechanism in aqueous hydrogen peroxide and glycine: ReaxFF reactive molecular dynamics simulations
Scope
ReaxFF MD of a Cu / aqueous H₂O–H₂O₂–glycine / hydroxylated SiO₂ abrasive stack models CMP-like shear under high normal load, emphasizing slurry chemistry and Cu removal pathways.
Summary¶
Wen et al. simulate chemical mechanical polishing (CMP)-relevant chemistry at a copper surface contacted by a hydroxylated silica abrasive and an aqueous mixture of water, hydrogen peroxide, and glycine. The goal is atomistic clarity on how oxidation, proton transfer, and mechanochemical shear combine to enable material removal in Cu CMP, where industrial slurries are far more complex but share the same oxidant + complexing agent + abrasive motifs. The ReaxFF model spans Cu, Si/O/H abrasive chemistry, and organic glycine fragments, with development context and DFT checks (including NEB references for key water steps) summarized in the article. Adri C. T. van Duin and Diana M. van Duin appear among authors, tying the publication to the group’s tribochemistry and oxide–metal interface expertise.
Methods¶
ReaxFF model lineage (A)¶
ReaxFF spans Cu, Si/O/H abrasive chemistry, and organic glycine/oxidant chemistry with DFT/NEB support cited in JPCC for key steps (see article).
Mechanochemical CMP protocol (B)¶
System (Fig. 1): Cu(111) slab (4369 Cu), hydroxylated SiO₂ abrasive (1296 Si, 2708 O, 232 H), slurry 200 H₂O, 10 H₂O₂, 5 glycine in ~6.34 × 5.90 × 4.50 nm; periodic xy; bottom Cu layers frozen, silica tool rigid/movable. Load to ~6.0 GPa normal pressure on abrasive, 25 ps equilibration, then shear 40 m/s (0.4 Å/ps) for 300 ps along x. LAMMPS, NVT, 300 K, Nose–Hoover (100 fs thermostat damping), Δt = 0.25 fs; Ovito visualization.
Static QM (C)¶
DFT/NEB references support specific barrier claims—not a standalone static study.
Barostat / hydrostatic pressure: N/A — production shear uses NVT at 300 K with ~6.0 GPa normal load on the abrasive (mechanical stress in the interfacial stack) rather than NPT fluid pressure control. Electric field: N/A — not used. Replica / enhanced sampling: N/A — not reported for this CMP shear study.
Findings¶
Mechanisms (removal pathways)¶
H₂O chemisorbs; H₂O₂ dissociates toward Cu–OH-rich surfaces. Silanol groups mediate proton-transfer coupling abrasive to slurry. Under shear, Cu removal proceeds via glycine-, OH-, and Cu–O–Si bridge channels (§3.2+)—oxidation + ligand-assisted lift-off vs pure abrasion.
Glycine carboxylate/amine interactions with Cu oxides compete with peroxide-driven oxidation, so removal is not a single mechanical scrape but a coupled ligand/oxide/abrasive sequence consistent with industrial CMP slurry design heuristics.
Limitations / outlook¶
Single-asperity, high load/speed idealization; industrial slurry complexity (pH, inhibitors, particles) not fully captured. Sensitivity in the modeled stack is to normal load (~6.0 GPa), shear speed (40 m/s), and reagent counts (H₂O/H₂O₂/glycine) as in Fig. 1; broader process maps in the PDF add nuance. Corpus honesty: use the JPCC PDF for table-level repro detail.
Limitations¶
A single asperity and high load/speed are idealizations; industrial slurries include particles, inhibitors, and pH buffers not modeled here.
Relevance to group¶
Detailed ReaxFF CMP case study with explicit mechanochemical pathways (van Duin / consulting-affiliation context in the publication).