About this Abstract |
Meeting |
MS&T23: Materials Science & Technology
|
Symposium
|
Materials Informatics for Images and Multi-dimensional Datasets
|
Presentation Title |
Semi-automated Hierarchical Clustering Model for 4D-STEM Datasets |
Author(s) |
Chuqiao Shi, Nannan Mao, Yao Yang, Jing Kong, Yimo Han |
On-Site Speaker (Planned) |
Chuqiao Shi |
Abstract Scope |
Four-dimensional scanning transmission electron microscopy (4D-STEM) has made major strides in understanding materials structure. Compared with conventional STEM images with one-pixel intensity per scanning point, 4D-STEM are comprised of a moment-resolved diffraction pattern per scanning point, which provide extremely rich lattice information of materials. Despite its advancements, the processing of big 4D data remains difficult and requires priory knowledge of both 4D-STEM and the materials structure. Here, we developed a semi-automated model based on unsupervised hierarchical clustering. Our approach complements the existing 4D data processing approaches by providing a rapid and automatic initial analysis of the 4D data, as well as uncovering the unexpected but significant fine structures and deformation sin the materials, including WS2-WSe2 superlattice with minor ripples and 2D van der Waals ferroelectrics with complex domain structures. We also applied the approach to catalytic nanoparticles and improve our understanding of in the catalyst behavior in chemical reactions. |