Skip to content

Reactive molecular dynamics simulations of sodium silicate glasses — toward an improved understanding of the structure

Summary

Yu, Wang, Wang, Bauchy, and Sant use ReaxFF reactive molecular dynamics to study sodium silicate glass structures and compare the resulting predictions to both a classical silicate force field and experimental structural references (Int. J. Appl. Glass Sci., DOI 10.1111/ijag.12248). The central question is pragmatic for the glass community: when modeling multicomponent silicates, does a bond-order reactive potential materially improve short- and medium-range order—including modifier Na⁺ effects on the Si–O network—relative to a cheaper classical model, without paying ab initio costs? The paper frames ReaxFF as an intermediate tier: more flexible than fixed-charge silicate models for chemically responsive environments, yet still accessible for large amorphous cells.

Methods

MD application (atomistic dynamics)

Yu et al. build glasses of ~3000 atoms (alkali-rich sodium silicate (NS)) in 3D periodic cubic cells using LAMMPS (explicitly named in the Methodology section of pdf_path). Glass formation (classical stage): atoms are placed randomly (avoiding overlaps), melted at 4000 K for 100 ps, then linearly quenched from 4000 K → 300 K at 1 K/ps under NPT at zero average pressure; the cell is further held at 300 K for 1 ns NPT to relax the quenched glass. Thermostat / barostat: the article documents NPT at 0 pressure through these classical stages but does not name a specific thermostat/barostat algorithm in the extracted Methodology text—confirm coupling details in pdf_path for exact integrator settings. This entire melt–quench pathway uses the classical Teter empirical silicate potential for efficiency.

ReaxFF refinement stage: the classical-quenched structure is relaxed for 1 ns with ReaxFF in NPT at zero pressure before structural analysis; the authors do not perform a full high-temperature ReaxFF quench, arguing that Na–O dynamics at melt temperatures would force an extremely small timestep and make a full ReaxFF quench prohibitively expensive. Timestep (classical melt/quench): N/A — the article discusses timestep demands qualitatively for ReaxFF at high T but does not print an explicit Δt value for the Teter stages in the PDF text extracted here—copy integrator settings from pdf_path if you need exact numbers.

Electric field: N/A — not used. Replica / enhanced sampling: N/A — not used. The introduction’s ~10× (ReaxFF vs classical MD) and ~10⁶× (ab initio vs classical) remarks are authored scaling statements, not timings from this study’s hardware.

Force-field training

N/A — not a new ReaxFF parameterization paper; the study applies ReaxFF and contrasts it to a classical model rather than reporting a fresh QM optimization campaign for NS glasses.

Static QM / DFT

N/A — not a DFT production study; QM appears as motivation and accuracy context for when reactive vs classical vs ab initio methods are appropriate.

Findings

  • Structural accuracy: The authors report that ReaxFF improves short- and medium-range structure relative to the tested classical potential, especially in capturing modifier effects on the Si–O network where bond-order responsiveness matters.
  • Na⁺ / network coupling: Because bond orders adapt to local environments, ReaxFF better tracks Na⁺-related structural signatures (e.g., Si–O bond elongation / connectivity patterns) than the fixed-charge classical description in their comparison.
  • Scope honesty (authored): The paper emphasizes static structure validation here and notes that dynamics, mechanics, and long-time aging need separate validation beyond this structural benchmark.
  • Workflow implication (authored discussion): For flexible depolymerized glasses, the authors discuss combining classical quench with subsequent ReaxFF refinement as a pragmatic route; silica-like rigidity makes that shortcut problematic (as they argue referencing prior silica work).

Relevance to group

Glass and oxide-network chemistry overlaps with geochemical and materials themes in ReaxFF applications.

Citations and evidence anchors