Revealing atomic-scale patterns of nature in living color — ScienceDaily

A machine learning method to use large volumes of X-ray data will speed up material discovery.

Color coding makes aerial charts much easier to understand. By color we can tell at a glance where there is a road, forest, desert, city, river or lake.

Working with several universities, the US Department of Energy’s (DOE) Argonne National Laboratory has developed a method to create color-coded graphs of large volumes of X-ray analysis data. This new tool uses computational data sorting to find clusters related to physical properties, such as atomic distortion in a crystal structure. This should greatly speed up future studies of atomic-scale structural changes induced by changing temperature.

“Our method uses machine learning to rapidly analyze massive amounts of X-ray diffraction data,” said Raymond Osborne, senior physicist in Argonne’s Materials Science Division. “What might have taken us months in the past now takes about a quarter of an hour, with much finer results.”

For more than a century, X-ray diffraction, or XRD, has been one of the most prolific of all scientific methods for materials analysis. It has provided key information on the 3D atomic structure of countless technologically important materials.

In recent decades, the amount of data obtained in XRD experiments has increased dramatically at large facilities such as the Advanced Photon Source (APS), a user facility of the DOE Office of Science at Argonne. However, analysis methods that can handle these huge data sets are sorely lacking.

The team calls their new method X-ray Temperature Clustering, or XTEC for short. It accelerates material discovery by rapidly clustering and color-coding large X-ray data sets to reveal previously hidden structural changes that occur when temperature is increased or decreased. A typical large data set would be 10,000 gigabytes, which equates to approximately 3 million songs for streaming music.

XTEC draws on the power of unsupervised machine learning using methods developed for this project at Cornell University. This machine learning does not depend on initial training and training with already well-explored data. Instead, it learns by finding patterns and clusters in large data sets without such training. These patterns are then represented by color coding.

“For example, XTEC can assign red to a cluster of data that is associated with a certain property that changes with temperature in a certain way,” Osborne said. ‚ÄúThen cluster two will be blue and associated with another property with a different temperature dependence, and so on. The colors indicate whether each cluster represents the equivalent of a road, forest or lake on an aerial map.’

As a test case, XTEC analyzed data from beamline 6-ID-D at APS taken from two crystalline materials that are superconducting at temperatures close to absolute zero. At this ultra-low temperature, these materials go into a superconducting state, offering no resistance to electric current. More important to this study, other unusual features appear at higher temperatures related to changes in the material’s structure.

Applying XTEC, the team extracted an unprecedented amount of information about changes in atomic structure at different temperatures. These include not only distortions in the ordered arrangement of atoms in the material, but also fluctuations that occur when such changes occur.

“Because of machine learning, we can see the behavior of materials that is not visible from conventional XRD,” Osborne said. “And our method is applicable to many big data problems, not only in superconductors, but also in batteries, solar cells, and any temperature-sensitive device.”

APS is undergoing a massive upgrade that will increase the brightness of its X-rays up to 500 times. Along with the upgrade will come a significant increase in the data collected at APS, and machine learning techniques will be essential to analyze this data in a timely manner.

In addition to Osborne, Argonne writers include Matthew Krogstad, Daniel Phelan, Puspa Upreti, Michael Norman and Stephan Rosenkrantz. The primary collaborating partner is Cornell University (Eun-Ah Kim, Jordan Venderley, Krishnanand Malalayya, Michael Matty, Geoff Pleiss, Varsha Kishore, and Kilian Weinberger) and the Cornell High Energy Synchrotron Source (Jacob Ruff). Other partners include the University of Tennessee (David Mandrus), the University of Maryland (Lech Poudel), and New York University (Andrew Gordon Wilson).

Argonne funding was provided by the DOE Office of Basic Energy Sciences and the National Science Foundation.

Leave a Comment