Unsupervised segmentation and clustering workflow for efficient processing of 4D-STEM and 5D-STEM data

Microscopy and Microanalysis (2026)

 

Scientists use a technique called 4D-STEM to map the structure of materials at the nanometer scale by collecting diffraction patterns at many points across a sample. While this provides a huge amount of information about crystal orientation, structure, and strain, the resulting data sets can be extremely large and difficult to analyze, especially when studying materials as they change over time.

In this paper, researchers developed a new data-analysis method that automatically groups together regions of a material that have similar crystal structures. By averaging the diffraction data within each region, the method improves data quality while dramatically reducing the amount of data that needs to be processed. The team demonstrated the approach by tracking the growth of gold nanoparticles in liquid. This technique makes it faster and easier to analyze large microscopy datasets and could help researchers study structural changes in a wide range of materials, including for quantum applications.

The paper authors are Serin Lee, Stephanie Ribet, Arthur McCray, Andrew Barnum, Jennifer Dionne and Colin Ophus, who conducted their work at Stanford University and Berkeley Lab.
#QuantumScience #QuantumMaterials