High-Throughput Prediction of Stress-Strain Curve of Thermoplastic Elastomer Model Block Copolymers with Various Chain Structure by Combining Hierarchical Simulation and Deep Learning

Thermoplastic elastomer (TPE) is a typical industrial product, in which the microphase separation of block copolymer is utilized, and the stress-strain (S-S) curve is a key issue to design such products. The polymer chain and phase separated structure affect the S-S curve. However, it is not simple to find the relation between the structures and S-S curve. To tackle the problem, we applied hierarchical simulation and deep learning technique. We studied various type of AB type block copolymers, where A blocks and B blocks form glassy and rubbery domain respectively. S-S curves of wide variety of volume fractions and structures were investigated by hierarchical simulation of consistent field theory and coarse-grained molecular dynamics (CGMD). Furthermore, we applied 3D-convolutional neural network (3D-CNN) to make regression between the structures and S-S curves obtained by CGMD simulation. The predicted S-S curves of untrained structures using trained 3D-CNN showed good agreement with simulation, and high-throughput prediction could be realized comparing to the computationally intensive simulation.