New Methods for Understanding and Controlling the Self-Assembly of Reacting Systems Using Coarse-Grained Molecular Dynamics

This research aims at developing new computational methods to understand the molecular self-assembly of reacting systems whose complex structures depend on the thermodynamics of mixing, reaction kinetics, and diffusion kinetics. The specific reacting system examined in this study is epoxy, cured with linear chain thermoplastic tougheners whose complex microstructure is known from experiments to affect mechanical properties and to be sensitive to processing conditions. Mesoscale simulation techniques have helped to bridge the length and time scales needed to predict the microstructures of cured epoxies, but the prohibitive computational cost of simulating experimentally relevant system sizes has limited their impact. In this work, we develop an open-source plugin for the molecular dynamics code HOOMD-Blue that permits epoxy crosslinking simulations of millions of particles to be routinely performed on a single modern graphics card. Using these capabilities, we are able to use ensembles of epoxy processing pathways to obtain realistic bond kinetics and relaxation times that sensitively depend on stochastic bonding rates and a diffusive drag parameter respectively. This work also demonstrates the first implementation of fully customizable temperature-time curing profiles and the largest cross-linked structures obtained using molecular dynamics simulation. We evaluate coarse-grained models based on Dissipative Particle Dynamics (DPD) and compare with Lennard-Jones(LJ) models for their suitability to study glassy dynamics which is important for modeling epoxies or any other glassy material. We find that “hard” particle potentials such as the LJ potential are necessary to model glassy materials and characterize multiple methods for measuring the glass transition temperature (Tg) in simulations. We find that variations in temperature-time curing profiles result in significant differences in the final cured morphologies. Finally, we apply our general techniques to the specific DGEBA/DDS/PES system and validate our predicted glass transition temperatures against experiment