Machine Learning for Analyzing Phase-Separated Structures of ABA Triblock Copolymer Blends in Elongation

An industrial thermoplastic material can form various metastable and complex phase-separated structures, depending on the molding process, resulting in various mechanical properties during deformation. Therefore, understanding the relationship between the morphology and the stress under deformation is important in material design. This study proposed a machine-learning-based morphology identification method and accordingly revealed the structural factor that affects the tensile stress of a neat ABA triblock copolymer as well as ABA-based blends with various compositional symmetricities, A block lengths, and B block lengths by combining this method with coarse-grained simulations as well as principal component analysis. By using this technique, researchers can visually and numerically analyze the morphological changes that occur with increasing strain.