Coarse-grained molecular dynamics (MD) simulations allow us to predict the mechanical responses of polymers, starting merely with a description of their molecular architectures. It is interesting to ask whether, given two competing molecular architectures, coarse-grained MD simulations can predict the differences that can be expected in their mechanical responses. We have studied two crosslinked polyimides PMR15 and HFPE52—both used in high- temperature applications—to assess whether the subtle differences in their uniaxial stress–strain responses, revealed by experiments, can be reproduced by carefully coarse-grained MD models. The coarse graining procedure for PMR15 is outlined in this work, while the coarse grain forcefields for HFPE52 are borrowed from an earlier one (Pandiyan et al 2015 Macromol. Theory Simul. 24 513–20). We show that the stress–strain responses of both these polyimides are qualitatively reproduced, and important insights into their deformation and failure mechanisms are obtained. More importantly, the differences in the molecular architecture between the polyimides carry over to the differences in the stress–strain responses in a manner that parallels the experimental results. A critical assessment of the successes and shortcomings of predicting mechanical responses through coarse-grained MD simulations has been made.

Titanium, due to its high strength to weight ratio and polyimides, due to their excellent thermal stability are being increasingly used in aerospace applications. We investigate the bonding between a (110) rutile substrate and a popular commercial polyimide, PMR-15, starting from the known atomic structure of the rutile substrate and the architecture of the polymer. First, the long PMR-15 molecule is divided into four fragments and an all-atom non-bonded forcefield governing the interaction between PMR-15 and a rutile substrate is developed. To this end, parameters of Buckingham potential for interaction between each atom in the fragments and the rutile surface are fitted, so as to ensure that the sum of non-bonded and electrostatic interaction energy between the substrate and a large number of configurations of each fragment, calculated by the quantum mechanical route and obtained from the fitted potential, is closely matched. Further, two coarse grained models of PMR-15 are developed—one for interaction between two coarse grained PMR-15 molecules and another for that between a coarse grained PMR-15 and the rutile substrate. Molecular dynamics simulations with the coarse grained models yields a traction separation law—a very useful tool for conducting continuum level finite element simulations of rutile-PMR-15 joints—governing the normal separation of a PMR-15 block from a flat rutile substrate. Moreover, detailed information about the affinity of various fragments to the substrate are also obtained. In fact, though the separation energy between rutile and PMR-15 turns out to be rather low, our analysis—with merely the molecular architecture of the polyimide as the starting point—provides a scheme for in-silico prediction of adhesion energies for new polyimide formulations.