Skip to content

Si/C/H ReaxFF Reactive Potential for Silicon Surfaces Grafted with Organic Molecules

Evidence and attribution

Authority of statements

Prose below summarizes the JPCC article identified by title and pdf_path.

Summary

A Si/C/H ReaxFF parameterization is developed for alkyl-grafted silicon surfaces, trained against DFT reaction energies and barriers for decomposition pathways on small Si clusters and extended surfaces. ReaxFF MD in ADF and LAMMPS follows thermal decomposition of alkyl monolayers at high temperature; Arrhenius analysis of MD rates is compared to DFT barriers and to transition-state theory expectations for pre-exponential trends with chain length. Coverage-dependent stability of monolayers on Si(111), Si(100) 2×1, and “half-flat” Si(100) is compared between ReaxFF and experiment/DFT. Applications include thermal stability screening of SAM coatings used in microfabrication and sensing stacks.

Methods

  • Parameterization: Combine prior Si/H and hydrocarbon ReaxFFs; optimize on a DFT training set of reaction energies for β-hydride elimination, alkene elimination, radical-mediated steps, and related elementary reactions on cluster models; error-function minimization between ReaxFF and QM energies (equation and weighting in the paper). Training emphasizes radical-mediated β-hydride sequences that dominate high-T decomposition channels.
  • DFT validation: Barrier and energy comparisons for elementary steps observed in MD; additional checks on coverage-dependent energetics vs prior DFT on ethyl-terminated Si(111).
  • Production reactive MD: ADF2017 and LAMMPS ReaxFF NVT runs on alkyl-grafted Si(111), Si(100) 2×1, and half-flat Si(100) slab supercells with PBC; 0.1 fs velocity Verlet; 100 fs temperature damping after 1 ps at 300 K; 800 ps at 1500–2200 K for high-T decomposition statistics. Thermostat name beyond damping: N/A — see JPCC Methods. Barostat / pressure: N/A — constant-volume NVT (no NPT). Electric field: N/A — not used. Enhanced sampling: N/A — umbrella / metadynamics / replica exchange not used (Arrhenius from direct MD).
  • Kinetics: Arrhenius plots from MD rates; comparison of activation energies and prefactors to DFT and TST expectations.

Findings

  • Mechanism: Dominant decomposition path is β-hydride abstraction by silyl radicals followed by alkene elimination to the gas phase, consistent with DFT barrier ordering.
  • Barriers & kinetics: MD activation energies match DFT well; prefactor vs chain length trends align with TST.
  • Coverage: ReaxFF reproduces experimental trends for alkyl coverage on several Si orientations as a function of chain length; DFT and ReaxFF both indicate decyl monolayers can be stable at high coverage (~0.8) at moderate temperatures.
  • Cross-code checks (ADF vs LAMMPS) are used to increase confidence that observed kinetics are not an artifact of a single integrator implementation.

Limitations

High-temperature MD accelerates chemistry; direct comparison to low-T experiments is indirect. Cluster/surface models may omit full device-scale defects. Ambient oxidation or aqueous hydrolysis pathways are not the focus of the thermal decomposition benchmarks presented.