Study of Crystal Defects Using Dark-Field X-ray Microscopy Combined with High-Energy Diffraction Microscopy and Computer Vision

Functional materials display magnetic, electronic and thermal properties that are intimately linked to their crystal structure. These crystalline structures are inherently imperfect, and their defects generate variations in local strain and orientation which impacts their macroscopic properties. The characterization of these defects, especially on the 'mesoscale', is a necessary step towards understanding their effect on material behavior. Here, we report on the use of dark-field x-ray microscopy (DFXM) and high-energy diffraction microscopy (HEDM) techniques to collect rich (~10 TB) experimental data to characterize the mosaic and strain state of a highly twinned and intergrown crystal structure. The measured sample was a single crystal of the known, highly faulted cathode material, NaMnO2. The wealth of information stored within this data is only accessible by leveraging computational resources and machine learning algorithms. An optical flow algorithm is being employed for motion correction, with plans to implement additional AI methods for defect identification and classification. Additionally, the correlation of DFXM and HEDM data is a first-of-its-kind study at the APS, defining a novel methodology for collecting crystal structure information across multiple length scales.