Molecular-based approaches for understanding and tailoring structure-property-processing relations in materials, based on the fundamental principles of quantum and statistical mechanics, have gained ground in academic research and industrial practice. They have been greatly aided by an unprecedented growth in computer power, but also by new, efficient theoretical and computational methods and algorithms. The broad spectra of length and time scales governing structure and dynamics in real-life materials have demanded the advancement of multiscale modeling strategies, involving several levels of representation, to bridge atomic-level constitution and interactions with macroscopic properties.

In this talk, Professor Doros Theodorou will discuss three examples of molecular modelling of structure-property relations in polymeric materials: (a) prediction of linear and nonlinear rheological properties of high-molecular weight polymer melts, such as polyethylene and cis-1,4 polyisoprene, through hybrid particle-field mesoscopic simulations employing slip-springs to represent entanglements and parameterized on the basis of atomistic calculations (1-4); (b) tracking structural relaxation in polymer glasses, such as polystyrene, as a sequence of elementary transitions between basins on their energy hypersurface, with transition rate constants computed from atomistic infrequent-event analysis (5,6) ; (c) quantifying the conditions for local interfacial failure in epoxy/graphene or graphene oxide interfaces through stress-controlled nonequilibrium molecular dynamics simulations (7). In each example he will outline how scale-hopping algorithms can be devised, based on rigorous statistical mechanical principles, to meet the challenges of long time and length scales in a computationally tractable way. The computational results lead to property predictions that are validated by available experimental measurements and elucidate molecular-level processes that are critical to materials design.