See the full schedule here.
Title: From Scattering to Diffraction: Unified Analysis of Non-Crystalline and Crystalline Phases Using Data Science Tools
Keywords: X-ray Scattering, X-ray diffraction, Automated Pipelines
Abstract: The large photon flux and broad X-ray energy spectrum of synchrotron sources enable a multitude of experimental techniques producing terabytes of data. High energy, X-ray scattering experiments are especially interesting due to the rapid collection of data using 2D areal detectors to provide structural information of materials. This data in the form of images or stacks of sequential images has information about the crystals within the sample, their orientation, distribution, size, and shape. When samples consist of several types of crystals, the volume fractions can be determined. In practice, scientists struggle to fully analyze every image in the data collections due to the manual nature of the analysis.
Our work focuses on automation of data analysis to enable scientists to focus on interpretation of the data. In this presentation, we will provide illustrative examples of the importance of ontology development (e.g. mds:Xray domain ontology) for synchrotron science and the use of FAIR principles in working with metadata.1 We will then describe the importance of this approach to automated workflows that integrate pre-processing, analysis, and visualization of the results (i.e. the FAIRshake python package).2 To aid in the use of artificial intelligence for machine learning of the data, we will present a kinematic-diffraction simulator (KDS) that is capable of providing labeled 2D diffraction data sets for use as training data.3 The use of the KDS to generate training data will be applied to the phase fraction analysis of Ti-6Al-4V during in-situ thermal annealing diffraction experiments at NSF’s Cornell High Energy Synchrotron Source (CHESS).4,5 We will also discuss our approach to handling liquid or amorphous scattering as we develop tools for handling non-crystalline phases in training data simulation. To illustrate this approach, the crystallization of Ti3Nb alloys will be presented where data was collected at DOE’s Advanced Photon Source (APS).
Related Papers
1. B. P. Rajamohan et al., “Materials Data Science Ontology(MDS-Onto): Unifying Domain Knowledge in Materials and Applied Data Science,” Sci. Data, 12 (2025) 628, doi: 10.1038/s41597-025-04938-5.
2. Finley Holt, “FAIRshake: A Framework for Automating High-Throughput XRD Data Processing and Machine Learning Integration” MS Thesis, Case Western Reserve University, Jan. 2025. http://rave.ohiolink.edu/etdc/view?acc_num=case1736871877028537
3. R. Medhi, et al., “2D-diffractogram analysis: Kinematic-diffraction simulator for neural-network training-data generation”, Comp. Mater. Sci. 252 (2025) 113777. doi: 10.1016/j.commatsci.2025.113777
4. W. Yue, et al., “Phase Identification in Synchrotron X-ray Diffraction Patterns of Ti-6Al-4V Using Computer Vision and Deep Learning”, Integr. Mater. Manuf. Innov. 13 (2024) 36-52.
5. W. Yue, et al., “Exploring 2D X-ray diffraction phase fraction analysis with convolutional neural networks: Insights from kinematic-diffraction simulations”, MRS Advances 9 (2024) 921-928. doi: 10.1557/s43580-024-00862-9
