Block chemistry for accurate modeling of epoxy resins

Abstract

Accurate molecular modelling of the physical and chemical behavior of highly cross-linked epoxy resins at the atomistic scale is important for the design of new property-optimized materials. However, a systematic approach to parametrizing and characterizing these systems in molecular dynamics is missing. We, therefore, present a unified scheme to derive atomic charges for amine- based epoxy resins, in agreement with the AMBER force field, based on defining reactive fragments – blocks – building the network. The approach is applicable to all stages of curing, from pure liquid, to gelation, to fully cured glass. We utilize this approach to study DGEBA/DDS epoxy systems, incorporating dynamic topology changes into atomistic Molecular Dynamics simulations of the curing reaction with 127,000 atoms. We study size effects in our simulations and predict the gel point utilizing rigorous percolation theory to accurately recover the experimental data. Furthermore, we observe excellent agreement between the estimated and the experimentally determined glass transition temperatures as a function of curing rate. Finally, we demonstrate the quality of our model by the prediction of the elastic modulus, based on uniaxial tensile tests. The presented scheme paves the way for a broadly consistent approach for modelling and characterizing all amine-based epoxy resins.

Block Chemistry for Accurate Modeling of Epoxy Resins

“…We finally determined the Young's modulus of our system by performing a tensile stress simulation. To make sure that the system was not trapped in a metastable equilibrium, which is known to affect the mechanical properties of MD polymeric networks, 24 we subjected each cured sample to a postcuring annealing procedure consisting of 10 ns at a temperature of 1000 K (see Zenodo 63 for input scripts), after which the temperature was quickly lowered to 300 K. Each system was then allowed to equilibrate in the NPT ensemble at 298 K using isotropic pressure coupling until its density reached steady state (see Table S5). To rule out the possibility of insufficient annealing time, we also extended the postcuring annealing period up to 20 ns but without observing significant changes in the mechanical response of the system.…”

“…To rule out the possibility of insufficient annealing time, we also extended the postcuring annealing period up to 20 ns but without observing significant changes in the mechanical response of the system. After completing equilibration at room temperature, each cross-linked sample underwent a uniaxial tensile test along the x, y, and z axes at an engineering strain rate of 2•10 −8 s −1 (input available at Zenodo 63 ). This strain rate was chosen as a compromise between the computational efficiency of sampling and straining the system as slow as possible.…”

“…All parameters can be found in the LAMMPS data files in our Zenodo repository. 63 After minimization of the initial geometry using a combination of steepest descent and conjugate gradient, each cell was heated for 20 ps in the NVT ensemble, raising the temperature to 503 K, and then equilibrated in the NPT ensemble for 2 ns, allowing the density to reach a steady state (see Figure S7 and Zenodo 63 for input scripts). A Nose− Hoover thermostat and isotropic barostat at 1 atm with a 100 fs relaxation time, periodic boundary conditions, and a 1 fs integration time step were employed throughout.…”

“…The LAMMPS scripts developed for the implementation of this protocol are available on Zenodo. 63 Curing cycles were performed using fix bond/react, 23 which executed its own inner loop consisting of three processes (Figure 2): reactive-group search, topology change, and structural relaxation. Specifically, bond/react effected accurate topology updates from the pre-to the postreacted configurations (Scheme 5) triggered by a distance cutoff between reacting atoms.…”

“…The AMBER force field library files of our blocks and all the LAMMPS scripts necessary to reproduce our simulations are publicly available on Zenodo. 63 With these block fragments, our modular charge scheme can be applied to any cured or uncured DGEBA-DDS network, including any polymerization state n of DGEBA. Notably, the block chemistry principle can also be easily adapted to other epoxy/amine systems at the minimum cost of two charge computations: one for the new hardener and one for the short OPO block linking the latter to the epoxy system.…”