Finding “Trigger Sites” of Reactions Among Heterogeneous Materials from X-ray Microscopic Big Data Using Persistent Homology

Material properties are typically determined by specific features such as the heterogeneity (or “trigger sites”) of phases and chemical states. Trigger sites have been investigated in systems such as structural materials and batteries by using multiscale X-ray microscopy (XRM). However, identifying trigger sites manually or via computers has been challenging because the data involved are large and multidimensional. In this study, we developed a new approach to determine trigger sites on the basis of shapes of heterogeneity in XRM data by using persistent homology (PH) analysis. Our results revealed the following two aspects: (1) the trigger sites for the heterogeneous reduction of iron ore sinters, the complex of Ca–Fe–O oxides, and pores; and (2) the crack initiation sites at the nanoscale in carbon-fiber-reinforced plastics (CFRPs) under load. We could “non-empirically” identify trigger sites from the big data obtained by XRM. This approach can be applied to identify trigger sites in various materials.