Skip to content

Development of the ReaxFF Reactive Force Field for Inherent Point Defects in the Si/Silica System

Summary

ReaxFF parameters for Si/O/H are re-fit (ReaxFFpresent) to capture point defects, O interstitial behavior in bulk Si, and Si/SiO2 interface oxidation pathways where earlier SiOH(2010) fits spuriously allowed O diffusion in Si at 300 K. Training follows a QM-heavy dataset; validation includes O diffusion in Si (800–2400 K), amorphous SiO2 density, and a-SiO2/Si annealing scenarios. The J. Phys. Chem. A contribution is explicitly aimed at semiconductor process modeling where defect-mediated oxidation must remain physically plausible across temperature.

Methods

  • Parameterization: ReaxFFpresent Si/O/H terms optimized against QM reference data for defect energies, migration paths, and interface-relevant configurations (see article §2).
  • Bulk Si – O migration: MD tracks bond-centered (BC) O hops in the (110) plane with an asymmetric saddle; reported O-interstitial barrier ~64.8 kcal/mol matching DFT/experiment trends in the abstract.
  • Temperature study: O diffusion activated above ~1400 K in their simulations; diffusion coefficient vs T compared with experimental literature (abstract).
  • Silica validation: a-SiO2 prepared at 2.21 g/cm³, close to 2.20 g/cm³ experimental density; a-SiO2/Si stack annealed below the computed Si melting point to probe interfacial O transport without the SiOH(2010) low-temperature leakage.

Training reaction classes, QM levels, and MD analysis scripts for O mean-square displacement versus T are documented in papers/Nayir_JPC_C_SiOx_2019_proof.pdf (confirm final layout against the published issue when citing tables).

MD application (proof Methods). Molecular dynamics is ReaxFF-driven on PBC supercells (example: 5488-atom two-phase Si melting cell 38.3 × 38.3 × 76.5 Å with a frozen half-slab and a molten zone for coexistence tests). Ensemble: NVT and NPT segments for melting/interface equilibration; NVT annealing and NPT relaxation blocks appear in the two-phase and a-SiO₂/Si annealing workflows described in the text. Timestep: 0.5 fs for the O-diffusivity runs quoted in the proof. Thermostat: Berendsen with 100 fs coupling for 1 ns NVT anneal to target temperature, then 3.5 ns continuation with 1000 fs coupling in the same paragraph of the proof. Barostat / pressure: NPT is used in equilibration and a-SiO₂ density stages as described (e.g., NPT relaxation of a-SiO₂ and NPT segments for phases competing to fill the box in melting tests). Engine / code: the proof text does not recover a clear third-party MD label from this PDF text layer; treat as ReaxFF MD in the computational setup of J. Phys. Chem. A 9b01481 and confirm software in the version-of-record if you need a package name. Duration: multi-ns windows for some O-diffusion statistics; 50 ps NVT melt shots and NPT a-SiO₂ relax blocks are quoted for specific figures in the Methods section. Temperature: 800–2400 K for O-diffusivity trends (abstract) plus ~2421 K / ~2425 K melting bracketing for Si in the two-phase test described in the proof. Electric field / enhanced sampling: N/A — not part of the quoted diagnostics.

Findings

  • Mechanism: O migrates in bulk Si via BC-site hopping along the expected crystallographic path (abstract).
  • Barriers and kinetics: 64.8 kcal/mol barrier for O-interstitial migration and temperature-dependent diffusivity consistent with experiments (abstract).
  • Fix vs SiOH(2010): ReaxFFpresent removes unphysical O diffusion through Si at 300 K that plagued SiOH(2010) while reproducing targeted oxidation behavior in the a-SiO2/Si tests (abstract).

The validation section emphasizes agreement on barrier scale and T-dependent diffusivity trends as primary acceptance criteria for the refit.

Limitations

Proof PDF; confirm final layout and any parameter table edits in the published J. Phys. Chem. A article. High-temperature MD may still miss charged defect states important in devices.

Wiki prose here is a navigation aid. Definitive numbers, protocol details, and figure-level claims should be taken from the peer-reviewed article at pdf_path (and any Supporting Information cited there), not from this page alone.

Relevance to group

Semiconductor oxidation and defect modeling with ReaxFF; central to Si/SiO2 process and reliability studies in the van Duin ReaxFF line.

Citations and evidence anchors