Skip to content

Atomistic mechanisms of Si chemical mechanical polishing in aqueous H2O2: ReaxFF reactive molecular dynamics simulations

Summary

Chemical mechanical polishing (CMP) of silicon uses slurry chemistry and mechanical abrasion to planarize wafers; hydrogen peroxide is a common oxidizer in silica-based slurries. This Computational Materials Science article models CMP at the atomistic level with ReaxFF reactive molecular dynamics, placing a silica abrasive asperity in sliding contact with hydrogen-terminated Si(100) in explicit aqueous hydrogen peroxide. The study contrasts peroxide-driven chemistry with pure-water slurries to isolate how pre-oxidation and shear-induced bond rupture contribute to material removal and friction. van Duin and Lu co-authorship ties the paper to the group’s broader silicon–water and tribochemistry portfolio, including standalone water studies on multiple Si facets. CMP slurries are multicomponent industrial fluids; this model isolates oxidizer chemistry on a single facet to build mechanistic intuition before adding pH buffers or surfactants.

Methods

Molecular dynamics (reactive). The article employs ReaxFF reactive molecular dynamics to follow a silica abrasive sliding on hydrogen-terminated Si(100) in explicit water, comparing runs with and without aqueous H₂O₂ so oxidizer chemistry can be isolated while pressure and velocity remain the paired mechanical knobs (the Preston-style emphasis in the introduction). The indexed excerpt does not reproduce supercell dimensions, exact atom counts, or integration timestep values—those appear in the Comput. Mater. Sci. Methods of the PDF at pdf_path. Periodic supercell conventions, target temperature (K) and thermal coupling of the Si slab and aqueous film, NVT/NPT staging, thermostat damping, equilibration versus production duration (ps/ns), and barostat usage should be read from the same source; this page does not substitute for those tables.

Force-field fitting. N/A — the study applies a published Si/O/H ReaxFF parameterization suited to silicon oxidation and tribochemistry; it does not report a new parameter fit in this article.

Static QM / DFT. N/A — trajectories are reactive MD, not on-the-fly DFT dynamics (QM cost is cited as prohibitive for sliding wear in the introduction).

Review scope. N/A — primary application paper; sibling [[2017wen-venue-jp6b11310]] covers facet-resolved silicon–water adsorption.

Findings

Outcomes and mechanisms. H₂O₂ oxidizes the Si surface before severe asperity contact, promoting interfacial Si–O–Si bridges that link the abrasive to the wafer. Once contact occurs, shear strains Si–Si and Si–O bonds until they rupture, ejecting Si-containing fragments consistent with material removal. Reactive pathways therefore couple mechanical work directly into bond scission rather than treating removal as purely abrasive grooving.

Comparisons. Versus pure H₂O, the peroxide slurry yields a more oxidized interface, removes more Si atoms in the simulated window, and produces a higher friction force because stronger covalent interfacial contacts must be broken during sliding.

Sensitivity / design levers. The introduction stresses that industrial CMP depends jointly on contact pressure and sliding velocity (Preston-type scalings) and that H₂O₂ concentration feeds both roughness and removal trends in experiments cited there; the MD portion is constructed to mirror those paired mechanical and chemical controls.

Limitations / outlook. Idealized single-asperity geometry and simplified slurry chemistry omit particle polydispersity, buffers, and surfactants present in fabs.

Corpus honesty. Quantitative loads, timesteps, and thermostat labels are not in the short indexed excerpt—confirm all protocol numerics from papers/Wen_Polishing_CompMatSci_2017.pdf before reproducing trajectories.

Limitations

Single surface orientation and idealized asperity geometry simplify industrial CMP, where particle size distributions, pH buffers, and additives matter. ReaxFF barriers are empirical. Slurry pH shifts acid–base chemistry at the interface beyond the explicit compositions shown in the idealized model.

Relevance to group

Direct van Duin-group contribution on Si/SiO\(_x\) tribochemistry paired with other silicon–water entries.

Citations and evidence anchors