(392f) Designing New Printable Thermoset Shape Memory Polymers Via Molecular Simulation and Machine Learning

Printable thermoset shape memory polymers (TSMPs) offer the opportunity to substantially expand the range of applications available to thermoset polymers. A TSMP, after a thermomechanical cycle (i.e., programming), can recover its original shape by heating. The act of programming reduces conformational entropy in the network and increases enthalpy through various interatomic interactions, thus supplying the thermodynamic driving force for shape recovery. The classic approach for developing new polymers is one of trial-and-error, but this brute-force strategy is costly when searching in a large compositional space. By fingerprinting TSMPs at multiple length scales and investigating a significant initial dataset using molecular simulation, we have identified salient features in epoxy/amine-based thermoset shape memory polymers that produce a large shape memory effect, identified storage mechanisms for entropy and energy in TSMPs, and proposed new molecules that we predict will result in improved recovery stress and recovery ratio.