Skip to content

Searching for correlations between vibrational spectral features and structural parameters of silicate glass network

Experimental IR/Raman on sodium silicate glasses is compared to atomistic models using three potentials—Teter (fixed partial charges), MGFF (diffuse charges), and ReaxFF—to test long-used spectral assignment rules against simulated network structure.

Summary

Sodium silicate glasses with compositions [Na₂O]ₓ[Al₂O₃]₂[SiO₂]₉₋ₓ for x = 7, 12, 17, 22 mol% are synthesized and measured (IR/Raman), with fused silica and crystalline quartz as references. Parallel MD structures at the same compositions are built with Teter, MGFF, and ReaxFF; simulated IR trends vs composition agree best with experiment for ReaxFF. Bond-length and angle distributions, Qₙ speciation, and ring statistics from ReaxFF-MD are then compared to three widely used interpretation schemes (IR peak position vs Si–O–Si angle; Raman stretch deconvolution vs Qₙ; low-frequency Raman assignments vs ring sizes). The authors conclude that none of these traditional assignments is consistent with the ReaxFF-generated glass structures and suggest revised interpretations for future testing.

Methods

  • Experiments: IR and Raman on the glass series above plus silica substrate and quartz (details of acquisition in the article and Supporting Information).
  • Simulations: Classical MD in LAMMPS with periodic boundary (PBC) supercells of sodium aluminosilicate glasses at the same compositions (x = 7, 12, 17, 22 mol% in the [Na₂O]ₓ[Al₂O₃]₂[SiO₂]₉₋ₓ notation; O(10⁴–10⁵) atoms-scale cells in typical glass R&D—exact atom count in the JACE PDF) and unwrapped coordinates for analysis. Teter and MGFF use 1.0 fs time step; ReaxFF uses 0.25 fs to resolve charge transfer. Structures are equilibrated including ~300 K-type conditioning before vibrational/IR workflow; a thermostat (details in the VOR input) brings the box to the stated target temperature (K) before spectra are computed (duration in ps / ns as in the source). Molecular dynamics (reactive and nonreactive in turn) serves as the sampling engine for vibrational postprocessing (not NEB or rare-event CVHD in this JACE excerpt**).
  • Force fields: ReaxFF uses the authors’ Si/O/H/Na parameter set with Al terms merged from Pitman et al. (as cited in the paper); MGFF and Teter implementations follow prior references in the article.
  • Spectra: IR spectra computed from MD trajectories using the protocol described in the article (and prior work cited there). 1 — other MD / sampling slots. E-field, umbrella, metadynamics, replica exchange: N/A in the stated JACE workflow (standard NVT-type equilibration and vibrational analysis in the VOR at 300 K where cited). Barostat / NPT / hydrostatic pressure (GPa, bar): N/A or spelled in VOR if NPT stages exist; this wiki defers to pdf_path.

2 — ReaxFF reparameterization in this study. The ReaxFF set merges Al-containing terms from Pitman et al. (as cited) into the Si/O/H/Na line—a parameter merging / extension, not a de novo FF paper in full; 2 — full QM-only training set as a standalone JCP-style fit (block 2 in the AGENTS sense): N/A (uses prior Pitman + group ReaxFF assembly). 3 — static DFT (only) as the result: N/A—primary results are classical + ReaxFF MD + spectroscopy comparison to experiment (IR/Raman on glasses**).

Findings

  • Among the three potentials, ReaxFF best tracks the composition-dependent trends in the simulated IR spectra relative to measurements.
  • Standard correlations/deconvolutions linking (a) Si–O stretch IR peak position to Si–O–Si angle, (b) Raman stretch bands to Qₙ fractions, and © ~420–600 cm⁻¹ Raman features to small-ring bending do not match the bond-parameter and ring statistics from ReaxFF-MD networks.
  • The authors argue that widespread textbook-style assignments for silicate glasses should be revisited when interpreted against atomistic glass models.

Corpus honesty and limits. The JACE version-of-record is papers/Liu_JAmCerSoc_2020_vibrations_glass.pdf; for query-form or proof PDFs use 2020liu-venue-paper only as provenance, not as data-source for peak positions (see that sibling page). ReaxFF vibrational physics is classical; quantum phonon effects are incomplete (Limitations in the source**).

Limitations

Classical and ReaxFF models omit full quantum phonon physics; conclusions about “correctness” of empirical spectral rules are conditional on the ReaxFF glass structures being representative for this purpose.

Relevance to group

Co-authors van Duin (Mechanical Engineering, Penn State) and Kim/Liu/Hahn (Chemical Engineering / MRI, Penn State); joint with Corning and UNT collaborators.

Reader notes (navigation)