Skip to content

Reactive Force Field for Simulations of the Pyrolysis of Polysiloxanes into Silicon Oxycarbide Ceramics

Evidence and attribution

Authority of statements

Prose sections below are curated summaries of the J. Phys. Chem. C article identified by doi and pdf_path. Parameter counts and training-data sizes are quoted from the publication text; treat SI tables as authoritative for numerical FF entries.

Summary

The article develops a new ReaxFF parameterization for Si–C–O–H chemistry aimed at silicon oxycarbide (SiCO) ceramics and polymer-to-ceramic synthesis routes starting from cross-linked polysiloxane precursors. The scientific motivation emphasizes that SiCO combines attractive thermal and mechanical properties for applications from thermal protection to electronics and energy storage, but that atomistic modeling has been limited either by DFT cost or by insufficient empirical reactive models for the amorphous, hydrogen-bearing networks formed during pyrolysis.

The authors position their work against prior ReaxFF sets (notably Newsome et al. for Si–C–O contexts and later Soria et al. extensions), arguing those parameterizations can miss key SiCO features and produce inadequate structures for some glass and free-carbon morphologies. The abstract claims extensive validation against experimental and computational thermochemistry, ab initio molecular dynamics benchmarks at elevated temperature, and an application to simulated polymer pyrolysis producing amorphous SiCO with graphene-like carbon segregations embedded in an oxycarbide matrix, in qualitative agreement with experimental observations.

Methods

1 — MD application (polysiloxane pyrolysis). Molecular dynamics in LAMMPS-class RMD codes (as in the J. Phys. Chem. C article): large-scale reactive MD converts cross-linked polysiloxane precursors toward amorphous SiCO; 3D PBC supercells with 10⁴-scale atom counts in the reported pyrolysis runs ( confirm in text / SI). NVT thermostat-mediated ramped-T (temperature in K) schedules with timestep in sub-1 fs and long-ns-class cumulative durations as in Main+SI; N/A to restate every K-ramp and equilibration here. Barostat / NPT / hydrostatic pressure : ** N/A for T-ramped NVT stages unless a NPT block is in the SI. External E-field, shock, umbrella: N/A in the stated RMD path**.

2 — Force-field training (ReaxFF, Si–C–O–H, SiCO). Parent set builds from Newsome et al. Si–C–O data plus Srinivasan-class C–C updates. DFT library: >10,000 crystalline SiCO-type hypothetical cells and >1000 amorphous models from network construction and melt–quench AIMD, in a self-learning retraining loop. QM targets include thermochemistry and high-T AIMD-style comparisons; functionals / cutoffs in the DFT protocol are in the main / SI. Optimization follows standard ReaxFF least-squares / ParReaxFF-class workflows as stated.

3 — Static QM alone. The paper uses extensive DFT training; N/A as a standalone static only result section separate from the RMD application above.

Findings

Mechanisms and morphology

The new field passes thermochemistry/AIMD checks emphasized in the abstract more faithfully than earlier SiCO attempts, enabling long reactive runs at large sizes. Pyrolysis simulations produce amorphous SiCO with graphene-like carbon segregations in an oxycarbide matrix—qualitatively aligned with experiment (figures in Results).

Limitations / future validation

SiCO space is broad; H content, T, and O/C stoichiometry shift microstructures—case-by-case validation remains necessary.

Limitations

SiCO composition space is vast; hydrogen content, processing temperature, and oxygen stoichiometry can all shift microstructure. ReaxFF remains an empirical model: quantitative NMR-level predictions and long-range carbon ordering may require cross-validation against DFT or experiment for each target composition. Training emphasizes SiCOH chemistry relevant to polysiloxane routes; transfer to unrelated organosilicon chemistries needs explicit checks.

Relevance to group

UT Arlington / PSU collaboration combining Kroll group SiCO modeling with van Duin-line ReaxFF development—central to ceramic pyrolysis and nanocarbon-in-glass microstructures in the group’s materials portfolio.

Citations and evidence anchors

  • DOI 10.1021/acs.jpcc.9b03810; J. Phys. Chem. C 2019, 123, 16804–16812.
  • Excerpt alignment: normalized/extracts/2019ponomarev-j-phys-chem-reactive-force_p1-2.txt.