Radiographic Testing
Plain-Language Commands Drive Autonomous X-ray Imaging
Title image: Tao Zhou of Argonne’s Center for Nanoscale Materials loads a sample into the 26-ID hard X-ray nanoprobe beamline before the SYNAPS-I demonstration. Credit: Jason Creps/Argonne National Laboratory
Science has increasingly used AI as a kind of microscope—sorting data, analyzing images, and revealing hidden patterns. Now, agentic AI can help scientists decide what to do next with simple, natural-language requests.
At the US Department of Energy’s (DOE) Argonne National Laboratory, researchers recently demonstrated this capability as part of the Synergistic Neutron and Photon Science–Intelligence (SYNAPS-I) project. SYNAPS-I is led by Lawrence Berkeley National Laboratory and combines data from neutron, X-ray, and microscopy experiments across national labs. It is part of DOE’s Genesis Mission, a national AI initiative.
In the project’s first phase, scientists built a real-time AI engine that reconstructs ptychographic images of microelectronics as fast as the data is generated. That turns hours or days of analysis into seconds. Ptychography collects many overlapping diffraction patterns, which are the distinctive ways X-rays scatter after interacting with a material. It then uses computation to rebuild sharp, high-resolution images.
“Real-time reconstruction changes the way experiments are performed, enabling researchers to adjust measurements on the fly instead of discovering key features they missed after the experiment has ended,” said Aileen Luo, an Argonne assistant computational scientist and lead developer of the SYNAPS-I model for ptychography.
The second phase adds an agentic AI layer that shifts the emphasis from data processing to experiment control.
“Instead of writing code or entering technical commands, researchers can talk to the AI, which then handles data collection, processing, and analysis automatically,” said Ming Du, an Argonne computational scientist and a member of the SYNAPS-I team. “By automating these tasks, researchers have more time to focus on the science and on operating the beamline itself.”
The system also automates region-of-interest searches, which means finding a small feature in a potentially large sample. Features vary so much from sample to sample that this task is nearly impossible to express in code.
“Agentic AI solves this by interpreting the images reconstructed in real time by PtychoFM, our ptychographic model, leveraging the vision capability of a multimodal large language model,” Du said. “It works step by step: collecting the image, examining the image, deciding where to move next, refining the search, and ultimately finding the target region within an image.”
The AI acts on its own for low-risk actions and asks for human approval when risk or uncertainty is higher. Once it finds the right spot, it switches to a high-resolution scan. The agent also uses Meta’s Segment Anything Model 3 (SAM3) to isolate and label features within each image.
The team tested the platform at the 26-ID hard X-ray nanoprobe beamline, which is shared by Argonne’s Advanced Photon Source and Center for Nanoscale Materials (CNM). They asked the agent to locate the interface between two regions in an integrated circuit device. Starting deep inside the device, the agent scanned, analyzed, segmented, and navigated through images until it found the correct region.
“By combining fast ptychographic reconstruction, agentic AI control, and SAM3-powered image segmentation, the demonstration showed that SYNAPS-I can analyze semiconductor devices with a level of speed and consistency that traditional workflows cannot match,” said Tao Zhou, a scientist with the CNM and member of the SYNAPS-I team.
“Microelectronics was simply the example used to showcase the technology. The real achievement is developing a next‑generation, AI‑enhanced microscope that can benefit many scientific fields,” said Mathew Cherukara, an Argonne computational scientist and leader of the Computing and AI group and the Argonne SYNAPS-I team.
“Combined, these SYNAPS-I advances lay the groundwork for self-driving microscopes, closed-loop experimentation, and automated defect-detection tools that promise to accelerate discovery across fields ranging from semiconductor manufacturing to advanced materials and national user facilities,” Cherukara said.
