Scientists have developed deep-learning models that can sift through the massive amounts of data generated by X-ray diffraction techniques.
Scientists from the say deep learning can supercharge a technique that is already the gold standard for characterizing new materials. In an , the interdisciplinary team describes models they developed to better leverage the massive amounts of data that X-ray diffraction experiments produce.
During X-ray diffraction experiments, bright lasers shine on a sample, producing diffracted images that contain important information about the material鈥檚 structure and properties. Project lead , an associate professor in the and a scientist at the , says conventional methods of analyzing these images can be contentious, time-consuming, and often ineffective.
鈥淭here is a lot of materials science and physics hidden in each one of these images and terabytes of data are being produced every day at facilities and labs worldwide,鈥� says Abdolrahim. 鈥淒eveloping a good model to analyze this data can really help expedite materials innovation, understand materials at extreme conditions, and develop materials for different technological applications.鈥�
The study, led by student Jerardo Salgado, holds particular promise for high-energy-density experiments like those conducted at LLE by researchers from the . By examining the precise moment when materials under extreme conditions change phases, scientists can discover ways to create new materials and learn about the formation of stars and planets.
Abdolrahim says the project, funded by the US Department of Energy鈥檚 National Nuclear Security Administration and the National Science Foundation, improves upon previous attempts to develop machine learning models for X-ray diffraction analysis that were trained and evaluated primarily with synthetic data. Abdolrahim, Associate Professor from the , and their students incorporated real-world data from experiments with inorganic materials to train their deep-learning models.
More X-ray diffraction analysis experimental data needs to be publicly available to help refine the models, according to Abdolrahim. She says the team is working on creating platforms for others to share data that can help train and evaluate the system, making it even more effective.
