Skip to content

Effect of chemical structure on thermo-mechanical properties of epoxy polymers: Comparison of accelerated ReaxFF simulations and experiments

Evidence and attribution

Authority of statements

Prose sections below (Summary, Methods, Findings, etc.) are curated summaries of the publication identified by doi, title, and pdf_path in the front matter above. They are not new primary claims by this wiki.

For definitive numerical values, reaction schemes, and interpretations, use the peer-reviewed article (and optional records under normalized/papers/ when present)—not this page alone.

Summary

The study applies an accelerated ReaxFF workflow—supplying energy comparable to reaction barriers so that slow epoxy crosslinking events occur on MD-accessible timescales—to build atomistic networks from bisphenol-A epoxide cured with three classes of amines (aromatic, cycloaliphatic, aliphatic). Simulated thermo-mechanical properties are compared with experiments, highlighting how curing-agent chemistry shapes local heterogeneity (notably for cyclic amines) and strain-rate sensitivity (stronger for aliphatic systems). Adri C. T. van Duin is a corresponding author with Penn State engineering coauthors. The workflow is explicitly aimed at industrial epoxide–amine networks where cure chemistry controls composite performance. Thermoset design often trades Tg, toughness, and process window; the paper argues those macro knobs have traceable atomistic signatures in crosslink density and local strand stiffness.

Methods

Reactive cross-linking and acceleration

Accelerated ReaxFF supplies restraint energy \(E_\mathrm{rest}\) on selected atom pairs (functional form eq. 2 in the article) so that successful C–N epoxy–amine cross-links occur on MD-accessible time scales; restraint parameters (\(F_1,F_2,R_{12}\) for O–C, O–H, C–N pairs) and 10 000 steps of 0.25 fs per acceleration window are given in the Polymer Methods. The article states that ADF is used together with the accelerated ReaxFF cross-linking workflow (see their §2 and Fig. 2). Stoichiometry follows commercial cure ratios: bisA/DETDA and bisA/IPDA at 2:1 epoxide:amine, bisA/T403 at 3:1; systems are grown by doubling molecule counts as in the authors’ prior protocol. Polymerization stages use accelerated NVT runs at temperatures tied to the final cure stage of each formulation, with Berendsen thermostat (100 fs damping).

Annealing, \(T_g\), CTE, and modulus (LAMMPS)

After cross-linking, annealing ramps 300 K → 600 K in 100 K steps with 12.5 ps NVT or NPT segments (0.25 fs timestep); Berendsen thermostat (100 fs) and Berendsen barostat (500 fs damping) are used for the NPT portions. Density vs temperature for \(T_g\) uses NPT sweeps (300 000 steps of 0.25 fs per 20 °C interval from 25 °C to 325 °C). CTE segments repeat NPT at 300–330 K (300 000 steps, 0.25 fs). Tensile tests use LAMMPS stress/atom on relaxed networks (ReaxFF), with multiple strain rates as in the article and Poisson-relaxed lateral faces (ν = 0.5 where cited). All relaxed epoxy cells are treated as bulk systems with three-dimensional periodic boundary conditions in LAMMPS (PBC on the packed networks).

Experiments

Cure schedules, MDSC glass transitions, density, and modulus follow Table 1–2 in the paper (mix ratios and temperature windows for bisA/DETDA, bisA/IPDA, bisA/T403).

  • Replica / umbrella / metadynamics: N/A — not used (acceleration is restraint-based as above).

  • External electric field: N/A — not used.

Findings

  • Good correlation between simulation and experiment for the thermo-mechanical trends compared in the study.
  • Cyclic curing agents yield heterogeneous local structure that annealing can partly homogenize; aliphatic networks show more strain-rate dependence than aromatic-cured analogs in the simulations reported, linking cure-agent ring strain to mechanical dispersion.

  • The authors relate modulus shifts to free-volume-like indicators in the simulated cells, helping interpret why β-trends differ across amine families.

  • Corpus honesty: Tabulated moduli, \(T_g\), and CTE comparisons are authoritative in papers/Vashisth_Polymer_2018.pdf; this note is a navigation summary.

Limitations

  • Accelerated dynamics changes pathway competition relative to room-temperature processing; quantitative cure kinetics remain illustrative.
  • Force-field scope limits transfer to other epoxide/amine chemistries without re-parameterization.
  • Virtual mechanical tests on nanoscale cells omit fiber reinforcement, voids, and interphase regions present in composite coupons, so moduli comparisons should be read as chemistry-controlled trends within the simulated network class rather than drop-in predictions for full laminate stacks.

Relevance to group

Demonstrates group ReaxFF on thermoset cure with side-by-side validation, relevant to polymer composites and industrial processing questions.

Citations and evidence anchors

  • DOI: https://doi.org/10.1016/j.polymer.2018.11.005 (papers/Vashisth_Polymer_2018.pdf).

Reader notes (navigation)