Skip to content

Thermal conductivity of vitreous silica from molecular dynamics simulations: The effects of force field, heat flux and system size

Summary

Thermal conductivity of vitreous silica is computed with nonequilibrium molecular dynamics using the direct method under imposed heat flux. Three empirical models—BKS, Teter, and ReaxFF—are compared for the same glass preparation philosophy so that differences reflect potential form rather than distinct amorphization recipes. The study evaluates statistical uncertainty from thermal noise and finite-size effects by varying heat-flux magnitude and cross-sectional area. The abstract reports that all three potentials yield realistic amorphous structures, but ReaxFF agrees best with experimental thermal conductivity near 300 K, attributed to improved low-frequency vibrational content relative to BKS and Teter. Increasing heat flux and cross-section lowers the estimated standard deviation of the conductivity estimate from fluctuations, improving practical convergence.

Methods

Molecular dynamics (nonequilibrium thermal transport). Molecular dynamics simulations apply the direct method (nonequilibrium MD / NEMD) to vitreous silica supercells, imposing a steady heat flux and reading thermal conductivity κ from the induced temperature gradient once a steady profile appears. The study benchmarks three empirical potentials—BKS, Teter, and ReaxFF—on identically prepared amorphous structures so differences trace to the force field rather than distinct quench recipes. The abstract documents sweeps of heat-flux magnitude and cross-sectional area to trade thermal noise in κ estimates against finite-size effects and mild nonlinearity at large imposed flux. Engine naming (e.g., LAMMPS), exact atom counts, PBC details, timestep (fs), NVE/NVT staging, equilibration/production duration (ps/ns), and thermostat implementations for the hot/cold slabs are specified in J. Chem. Phys. 146, 054504 and should be taken from the PDF at pdf_path—the local extract is headline-only. NPT barostat-based pressure control is not the focus of the quoted abstract (conductivity work typically uses constant-volume hot/cold regions); confirm any stress control from the article text. Electric fields and metadynamics/umbrella enhanced sampling are not indicated for this κ study.

Force-field fitting. N/A — literature BKS, Teter, and ReaxFF SiO₂ parameter sets are compared as published; this paper does not refit them.

Static QM / DFT. N/A — κ extraction is classical MD; no on-the-fly DFT trajectories are reported for the conductivity workflow.

Review scope. N/A — primary research article benchmarking empirical silica models.

Findings

Outcomes and mechanisms. All three potentials reproduce realistic amorphous silica structures in the authors’ preparation, but ReaxFF tracks experimental κ near 300 K (~1.3–1.4 W m⁻¹ K⁻¹) better than BKS and Teter in this head-to-head study; the discussion attributes the gain chiefly to improved low-frequency vibrational content relative to rigid-ion models that historically overpredict κ for BKS-class fits.

Comparisons. Versus prior literature values cited in the introduction, BKS tends to overestimate κ; here ReaxFF aligns more closely with the experimental window under the same NEMD diagnostics.

Sensitivity / design levers. Increasing imposed heat flux and cross-sectional area lowers the standard deviation of κ caused by thermal fluctuations, improving statistical precision but moving the protocol farther into nonlinear-response territory if pushed too far.

Limitations / outlook. Classical models omit quantum corrections to heat capacity and phonon lifetimes in glasses; even when κ matches, elastic constants from ReaxFF may still deviate from experiment elsewhere—authors caution readers accordingly in the broader discussion.

Corpus honesty. Indexed text is abstract-level only; reproduce numerics, Green–Kubo comparisons (if any), and figure citations from papers/ReaxFF_others/2017-JCP-TianYe-thermal-conductivity-SiO2-ReaxFF-BKS-Teter.pdf.

Limitations

Classical simulations omit quantum corrections to heat capacity and phonon lifetimes in glasses; elastic constants from ReaxFF may still deviate from experiment even when κ matches well.

Relevance to group

Side-by-side ReaxFF vs rigid-ion silica thermal transport benchmarking relevant to laser-damage and fused-silica modeling contexts.

Citations and evidence anchors