Characteristic structural knowledge for morphological identification and classification in meso-scale simulations using principal component analysis

Abstract

Meso-scale simulations have been widely used to probe aggregation caused by structural formation in macromolecular systems. However, the limitations of the long-length scale, resulting from its simulation box, cause difficulties in terms of morphological identification and insufficient classification. In this study, structural knowledge derived from meso-scale simulations based on parameters from atomistic simulations were analyzed in dissipative particle dynamic (DPD) simulations of PS-b-PI diblock copolymers. The radial distribution function and its Fourier-space counterpart or structure factor were proposed using principal component analysis (PCA) as key characteristics for morphological identification and classification. Disorder, discrete clusters, hexagonally packed cylinders, connected clusters, defected lamellae, lamellae and connected cylinders were effectively grouped.