High-precision detection modeling reveals fundamental structural scales of amorphous materials

The characterization of amorphous materials has been a long-standing challenge in materials science due to their lack of long-range order, which makes it difficult to define structural metrics. Here, we describe a 2D amorphous system-based high-precision detection model(HPDM) that utilizes the concepts of “bag” and instances to achieve image-level classification to structure-level segmentation of amorphous systems without any additional dynamical information. We introduce an order parameter that describes the structural evolution in amorphous phase transitions and demonstrate the accuracy of HPDM in detecting amorphous microstructures. A fundamental structural scale for 2D amorphous systems can be obtained based on the optimal detection dimensions of HPDM for finite-scale systems, eliminating the need to calculate structural and dynamical information. Our findings and methodology hold significant scientific implications for understanding the nature of disordered structures in amorphous materials.