COLVARS-Driven Reactive Molecular Dynamics of Vitrimer Polymerization and Thermomechanical Property Analysis

Vitrimers are the new class of covalent adaptable networks (CANs) capable of topological rearrangement through dynamic bond exchange, enabling recyclability, self-healing and stress relaxation while maintaining network integrity. The ring opening of epoxides by carboxylic acids is a fundamental epoxy–acid esterification reaction underlying vitrimer network formation, yet its atomistic mechanism remains challenging to simulate as experimental time scales for crosslinking reactions in polymers range from minutes to hours—far beyond the reach of atomistic molecular dynamics (MD) simulations. In this work, we employ biased MD using the Collective Variables (COLVARS) module in LAMMPS combined with a reactive ReaxFF potential to promote the ring-opening reaction between Diglycidyl Ether of Bisphenol A (DGEBA) and oxalic acid. Three coupled collective variables were designed to represent epoxide C–O bond cleavage, C(epoxide)–O(acid) bond formation, and proton transfer between acid and epoxide oxygens. This external bias promotes nucleophilic addition of acid-derivedcarboxylate oxygen to the epoxide carbon, resulting in ring opening, followed by proton transfer to yield the β-hydroxy ester functionality characteristic of vitrimer networks. This method allows simulating cross-linking at realistic, low, temperatures which helps to mimic chemical reactions and avoids unwanted high-temperature side reactions. Since not all accelerated events lead to the desired product, careful selection of initial configurations, molecular orientations, and suitable external bias parameters increases the probability of meaningful simulation results. Furthermore, the stability of the resulting vitrimer was evaluated in the absence of any external bias, and its thermomechanical properties were subsequently analyzed. Results show that the biased MD approach successfully captures the atomistic steps of the esterification reaction and the resulting vitrimer network can be used for accurate prediction of thermomechanical properties.