Modeling of glycidoxypropyltrimethoxy silane compositions using molecular dynamics simulations
Summary¶
Reactive and non-reactive atomistic models are used to relate the fraction of silane condensation species (T0–T3) in a glycidoxypropyltrimethoxy silane (GPS) sizing network to density, structure, and mechanical response under deformation. The motivation is composite manufacturing: sizing layers on glass fibers influence matrix adhesion and damage tolerance before a full epoxy interphase is formed.
Reactive all-atom MD probes GPS-based glass-fiber sizing networks in which Si–O–Si linkages create T0–T3 silicon sites. Six T0:T1:T2:T3 compositions are built; GAFF cross-links the network, then ReaxFF captures bond breaking during mechanical tests. LAMMPS is the engine. Structure (density, RDF, void space) and mechanics (modulus, strength, strain, energy absorption, failure modes) are reported as functions of species composition.
Methods¶
MD application. LAMMPS drives all-atom MD in two stages. Stage 1 (network build, non-reactive): GAFF interactions support a single-step cross-linking algorithm (cited to prior publications) that condenses glycidoxypropyltrimethoxy silane (GPS) into Si–O–Si-linked films. Six T0:T1:T2:T3 compositions are built—100% T3, 100% T2, 50:50 T2:T3, 75:25 T2:T3, 10:10:40:40, and 25:25:25:25—each with three independent replicas to average conformational variability. Stage 2 (mechanical test, reactive): ReaxFF replaces GAFF during deformation so Si–O bond scission and damage nucleation are permitted; the manuscript contrasts GAFF vs ReaxFF functional form (continuous bond order, polarizable geometry-dependent charges).
Supercells are PBC GPS films with atom counts and cell vectors tabulated in Comput. Mater. Sci. §2. Reported observables include mass density, radial distribution functions, interstitial/void measures, stress–strain curves, elastic modulus, strength, failure strain, energy absorption, and failure morphology classifications.
Force-field training is N/A (GAFF + literature ReaxFF). Static QM, electric fields, and enhanced sampling are N/A for the main workflow.
Thermodynamic ensemble, timestep, thermostat/barostat, temperature, strain rate, and run lengths for cross-linking equilibration and ReaxFF mechanical loading appear in Comput. Mater. Sci. §2 and are not transcribed from the short indexed extract used here.
MD blueprint honesty. LAMMPS is named in the article for molecular dynamics on PBC GPS films. NVT/NPT/NVE choices, timestep, thermostat, barostat/pressure control during network curing vs ReaxFF deformation, and equilibration/production durations (ps/ns) are N/A on this page—retrieve from Comput. Mater. Sci. §2.
Findings¶
Outcomes: Mechanical response of the GPS sizing layer depends systematically on T-species composition; structure (density, RDF, void space) and mechanics (modulus, strength, strain, energy absorption) co-vary across the six models. Damage: ReaxFF deformation captures bond-breaking pathways and failure modes not accessible under GAFF alone. Statistics: three independent networks per composition are used to average variability. Comparisons / limitations: the paper discusses GAFF→ReaxFF handoff stresses and the omission of film former, epoxy, and full fiber/matrix interphase chemistry—see ## Limitations below and the manuscript discussion. ## Limitations
GAFF-to-ReaxFF handoff can introduce internal stress; the paper discusses equilibration implications. Film former, epoxy, and full interphase chemistry are not simulated here—focus is pre-interphase GPS network. Future work would need to embed these sizing networks adjacent to polymer matrices to capture interphase formation explicitly.
Relevance to group¶
Direct ReaxFF application from van Duin (RxFF Consulting) on composite interphase-relevant organosilicon chemistry, with explicit LAMMPS + ReaxFF workflow. The paper is a practical reference for how condensation speciation maps to measurable mechanical metrics in sizing layers.