In the development of painting materials, controlling the strength properties of materials is crucial for ensuring durability. While conventional techniques can measure the strength properties of polymer films, elucidating the relationship between their internal structure and strength properties remains challenging. In this study, we applied mathematical science methods, specifically persistent homology analysis, to extract structural information from molecular dynamics simulations of tensile tests on polymer films. This approach enabled us to analyze structural changes in the polymer membranes during the tensile tests. Our analysis revealed that changes in the ring structures extracted through persistent homology analysis correlated closely with the stress variations observed in the stress-strain curve. Furthermore, through an inverse analysis of the persistence diagram, we successfully identified specific structures within the polymer films that predominantly influence the stress-strain curve behavior. Notably, our approach does not rely on prior training with similar data, as required in conventional machine learning methods. Instead, it automatically and efficiently elucidates the relationship between the strength and structure of materials without any prior knowledge. This technique is expected to drive significant progress in the design and development of new materials, including paints.
Papers
Journal: Journal of Chemical Theory and Computation
Title: Dynamic Correlation Analysis between Stress-Strain Curve and Polymer Film Structure Using Persistent Homology
Authors: Ryuhei Sato*, Shinya Kawakami, Hirotaka Ejima, Takahiro Ujii, Koichi Sato, Takanori Ichiki, Yasushi Shibuta