Neutron experiments can take days to complete, requiring researchers to work long shifts to monitor progress and make necessary adjustments. But thanks to advances in artificial intelligence and machine learning, experiments can now be done remotely and in half the time.
The DEMAND single crystal diffractometer at the High Flux Isotope Reactor, or HFIR, is the latest neutron instrument at the Department of Energy’s Oak Ridge National Laboratory to be equipped with machine learning-assisted software. Developed by ORNL researchers in the Neutron Sciences and Computing and Computational Sciences Directorates, the new software, called Real-Time Image Analyzer, or ReTIA, analyzes images in real-time to detect structural and magnetic scattering signals with confidence. By comparing neighboring frames, ReTIA filters out unwanted background noise, making it easier to detect weak signals with significantly fewer measurements.
What’s more, ReTIA enables experiments on DEMAND to be fully automated from start to finish. With the push of a button, the software takes control of the instrument to measure samples only at angles where high quality data is likely to be found.
“This is a critical step toward realizing ‘smart’ beamlines,” said ORNL postdoctoral researcher Yiqing Hao. “ReTIA allows us to bypass unnecessary data collection that we would have been measured and discarded in the past. This allows us to do experiments 30 to 50% faster. The improvement also allows users from outside institutions to mail in samples and study them remotely with comparatively little human intervention by our team.”
Next up, the team plans to continue making improvements to DEMAND through 2024, as well as working to develop the technology for use on more neutron instruments at HFIR and its sister facility, the Spallation Neutron Source.
In addition to Hao, the project team members include Erxi Feng, Dan Lu, Leah Zimmer, Zachary Morgan, Bryan Chakoumakos, Guannan Zhang and Huibo Cao. Read more about the research published in the Journal of Applied Crystallography.
HFIR is a DOE Office of Science user facility.
UT-Battelle manages ORNL for the Department of Energy’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. The Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.