The next talk in the PhyLife seminar series will be given by Lau Blom Grondahl on 23rd of November at 11:00 am at Lykeion.
Title: Development and usage of coarse-grained computational models of polymer systems
Abstract: Various analytical approaches aim to describe the properties of polymer systems, but none captures sufficient detail to accurately describe their behavior at both a microscopic and macroscopic scale. For this reason, much modern polymer research relies heavily on molecular dynamics simulation to probe the microscopic behavior of these systems and how they are coupled to macroscopic phenomena. One successful molecular dynamics model is the Kremer-Grest model. In this model, polymers are essentially represented as hard beads connected by springs. While the Kremer-Grest model does not describe any specific polymer, it can be tuned to describe most commodity polymers by introducing stiffness into the bead-spring chains.
Despite the Kremer-Grest model already being coarse-grained compared to more atomistic models, it still requires many beads in a simulation to produce useful results. State-of-the-art simulations usually include about 5 million beads corresponding to a system with an approximate volume of 0.014 um^3 This means that simulations are slow, making them impractical for studying larger systems or the behavior of systems on longer timescales. In order to probe such systems, we introduce a mesoscopic model, where polymer network constraints are represented by nodes on a graph with edges of the graph representing the connectivity of the constraints. Dynamics of the nodes can then be simulated by considering the edges as entropic springs, which provide forces on the nodes.
Two types of nodes exist in this description: Crosslinks and entanglements. Crosslinks essentially represent chemical bonds between different polymers, while entanglements represent topological constraints from one polymer on another and vice versa. Using this model, we are able to reduce the number of dynamic elements in a Kremer-Grest simulation with 5 million beads to one with a few hundred thousand nodes, depending on the polymer stiffness. This reduction yields a significant computational speedup. Preliminary results of simulations only considering crosslinks show promising results when compared to those obtained from Kremer-Grest simulations where the network topology has been neglected. Implementing entanglement dynamics into the model, however, is difficult without making ad hoc assumptions. Once completed, this approach may enable studies of larger systems or long timescale phenomena, such as network relaxation and prediction of storage and loss modulus. The model could also be applied to the study of similar topology-dominated systems involving long polymer-like molecules.