Molecular Dynamics Study of Alkylsilane Monolayers on Realistic Amorphous Silica Surfaces
Summary¶
Molecular dynamics simulations compare reactive (ReaxFF) and classical (nonreactive) models for n-alkylsilane monolayers on silica. The authors introduce a synthesis mimetic simulation (SMS) sequence in which amorphous silica is generated and exposed to H₂O₂ to build a hydroxide-rich surface analogous to piranha treatment of silicon wafers, then alkylsilanes are assembled. Results are compared to more idealized setups (crystalline silica, non-H₂O₂ amorphous surfaces, controlled roughness).
The central claim is tribological: laboratory-relevant oxidation roughens and hydroxylates silica at nanometer scales that classical monolayer studies often omit.
Methods¶
MD application (atomistic dynamics)¶
All trajectories use LAMMPS. ReaxFF (Fogarty et al. Si/O; Rahaman et al. C/O/H) builds amorphous silica substrates with and without synthesis mimetic simulation (SMS) exposure to H₂O₂, intended to mimic piranha-type oxidation and hydroxylation before n-alkylsilane assembly; OPLS-AA (Lorenz et al. silica; Jorgensen et al. alkanes) carries nonequilibrium shear because it is ~50× cheaper than ReaxFF at similar sizes, with the authors reporting close equilibrium monolayer agreement between ReaxFF and OPLS-AA so friction uses OPLS-AA on SMS-prepared surfaces. SMS is melt/quench silica followed by H₂O₂ processing, then monolayer deposition, compared to crystalline, defected-crystalline, and non-H₂O₂ amorphous surfaces.
Shear cells use C₁₀ alkylsilanes at 3.9 chains nm⁻² (~experimental 4.0), PBC in xy, and PPPM slab electrostatics with no long-range Coulomb across the nonperiodic z normal. Runs are NVT at 300 K with a Nose–Hoover thermostat (50 fs damping); 0.5 fs timestep for ReaxFF and rRESPA for OPLS-AA (0.3 / 0.6 / 1.2 fs for bonds/angles/LJ+Coulomb). 2 ns production samples equilibrium monolayer structure (ReaxFF); 5–10 ns production applies ±5 m s⁻¹ sliding velocities to mirrored stacks (faster than typical AFM but consistent with prior monolayer friction MD). The outer 50% of each silica slab is rigid; normal load is scanned by fixing inner surface separations at 4–5 spacings between 1.8 and 3.0 nm (N/A — stochastic NPT barostat). OPLS-AA LJ cutoff 10 Å. Applied electric fields and replica sampling: N/A — not used.
Force-field training¶
N/A — this work applies published ReaxFF/OPLS-AA parameterizations; it does not report a new QM training loop.
Static QM / DFT¶
N/A — central claims are from reactive/classical MD and nonequilibrium shear.
Findings¶
SMS produces hydroxyl-rich, rough amorphous silica that changes monolayer packing and tilt versus idealized crystalline or unprocessed amorphous substrates; nematic order and tilt distributions track RMS roughness in the article’s analysis. Friction correlates negatively with global orientational order under monolayer–monolayer shear: defects modestly raise friction, while substrate roughness (especially after SMS) lowers order and raises friction more strongly, with defects plus SMS roughness acting together. Omitting postsynthesis oxidation therefore underestimates friction and overestimates order versus laboratory-relevant surfaces, so ideal crystalline silica can misrank lubricant chemistries.
Limitations¶
Wear via C–C/C–Si bond scission is explicitly out of scope (OPLS-AA shear avoids reactive scission); extending to wear chemistry would require reactive shear models and longer timescales than reported here.
Relevance to group¶
Illustrates ReaxFF use for oxidative surface preparation and tribology on silica, with explicit comparison to classical modeling of the same interfaces.