Artificial intelligence deciphers the “clouds” of the detector to accelerate materials research



Structural dynamics (2022). DOI: 10.1063/4.0000161″ width=”700″ height=”374″/>

(a) An example of an XPFS detector image on 90×90 pixelated detectors. (b) Corresponding image of the photon map produced by the simulator for the detector image, plotted as photon distribution per spot. Credit: Structural dynamics (2022). DOI: 10.1063/4.0000161

X-rays can be used as a super-fast, atomic-resolution camera, and if researchers fire a pair of X-ray pulses moments apart, they get atomic-resolution snapshots of a system at two times. in time. Comparing these snapshots shows how material fluctuates in a tiny fraction of a second, which could help scientists design future generations of superfast computers, communications and other technologies.

However, resolving the information in these x-ray snapshots is difficult and time-consuming. Joshua Turner, a senior scientist at the Department of Energy’s SLAC National Accelerator Center and Stanford University, and ten other researchers turned to artificial intelligence to automate the process. . Their machine learning method, published on October 17 in Structural dynamicsaccelerates this X-ray probing technique, and extends it to previously inaccessible materials.

“The most exciting thing for me is that now we can access a different range of measurements that we couldn’t before,” Turner said.

Blob manipulation

When studying materials using this two-pulse technique, X-rays scatter across a material and are typically detected one photon at a time. A detector measures these scattered photons, which are used to produce a speckle pattern, a speckled image that represents the precise configuration of the sample at a given instant. The researchers compare the speckle patterns of each pair of pulses to calculate the fluctuations in the sample.

“However, each photon creates a burst of electrical charge on the detector,” Turner said. “If there are too many photons, these charge clouds coalesce to create an unrecognizable blob.” This cloud of noise means researchers have to collect tons of scattering data to get a clear understanding of the speckle pattern.

“You need a lot of data to understand what’s going on in the system,” said Sathya Chitturi, who holds a Ph.D. student at Stanford University who led this work. He is advised by Turner and co-author Mike Dunne, director of the Linac Coherent Light Source (LCLS) X-ray laser at SLAC.

With conventional methods, all the data first had to be collected and then analyzed using models that estimate how photons cluster at the detector, a time-consuming process to understand speckle patterns.

The machine learning method, on the other hand, uses the raw image from the scattered photon detector to directly extract the fluctuation information. This new method is ten times faster on its own and 100 times faster when combined with improved hardware, allowing data analysis closer to real time.

Part of the success of the new method is due to the efforts of co-author Nicolas Burdet, a research associate at SLAC, who developed a simulator that produced data with which to train the machine learning model. Through this training, the algorithm was able to learn how charge clouds coalesce and sort out the number of photons hitting the detector per drop and per pulse pair. The model proved to be accurate even in very blobby conditions.

See beyond the clouds

The model can extract information for a range of materials that have been difficult to study because X-rays scatter them too weakly to detect, such as high-temperature superconductors or quantum spin liquids. Chitturi said the new method could also be applied to other non-quantum materials, including colloids, alloys and glasses.

Turner said the research should help with the LCLS-II upgrade, which will allow researchers to collect up to a million images, or a few terabytes of data, per second, up from about a hundred images. per second for LCLS.

“We at SLAC are excited about this upgrade, but we’re also concerned about whether we can handle this amount of data,” Turner said. In a related article, the team found that their new technique should be fast enough to process all that data. “This new algorithm will really help.”

The increased speed offered by artificial intelligence also promises to alter the experimental process itself. Instead of making decisions after data collection and analysis, researchers will be able to analyze data and make changes during data collection, which could save time and money during the experiment. It will also allow researchers to spot surprises and redirect their experiments in real time to investigate unexpected phenomena.

“This method can allow you to further explore the materials science you are interested in and maximize scientific impact by allowing you to make decisions at different points in your experiment regarding changes in experimental variables such as temperature, field magnetic and material composition,” Chitturi said. said.

The study is part of a larger collaboration between SLAC, Northeastern University and Howard University to use machine learning to advance materials and chemistry research.

More information:
Sathya R. Chitturi et al, A Machine Learning Photon Detection Algorithm for the Coherent Analysis of Ultrafast X-ray Fluctuations, Structural dynamics (2022). DOI: 10.1063/4.0000161

Hongwei Chen et al, Testing the Data Framework for an AI Algorithm in Preparation for High Data Rate X-ray Facilities, arXiv (2022). DOI: 10.48550/arxiv.2210.10137

Journal information:

Provided by SLAC National Accelerator Laboratory

Quote: Artificial Intelligence Deciphers Detector ‘Clouds’ to Accelerate Materials Research (November 7, 2022) Retrieved November 7, 2022 from deciphers-detector-clouds.html

This document is subject to copyright. Except for fair use for purposes of private study or research, no part may be reproduced without written permission. The content is provided for information only.

Source link


Comments are closed.