Lawrence Berkeley National Laboratory has launched SYNAPS-I, a groundbreaking effort to harness the power of artificial intelligence to revolutionize scientific discovery. This multi-laboratory effort is part of the Department of Energy’s new Genesis mission, which will transform petabytes of data from advanced light and neutron scattering facilities into actionable knowledge across fields such as energy, semiconductors and medicine. “Our national research facilities already lead the world in scientific discovery, and SYNAPS-I fundamentally accelerates the path from experiment to insight by incorporating AI directly into analytical workflows,” said Alex Hexemer, Berkeley Lab’s Advanced Light Sources (ALS) senior scientist and director of SYNAPS-I. By uniting seven DOE facilities and integrating large-scale machine learning models, SYNAPS-I promises a new era of increased data output and accelerated insights, especially with upgrades like the ALS-U project.
SYNAPS-I Initiative Accelerates Scientific Discovery with AI
The SYNAPS-I (SYnergistic Neutron and Photon Science – Intelligence) initiative seeks to revolutionize scientific analysis by directly integrating artificial intelligence into the workflow of the nation’s leading light and neutron scattering facilities. SYNAPS-I is one of three AI modeling teams led or co-led by Berkeley Lab, leveraging existing expertise in high-performance computing and large-scale dataset management. A central focus is building machine learning pipelines that dramatically speed up the analysis of data generated by techniques such as X-ray microscopy and neutron scattering, which reveal important details about a material’s composition and structure. Currently, tasks such as image segmentation (identifying individual particles within a material) are often painstakingly performed by hand. “Currently, there are segmentation AI models available for images of everyday objects, but they don’t work well for scientific data. We’re building SYNAPS-I to fill that gap.”
This effort will integrate analysis across seven DOE basic energy science user facilities, including ALS, which is undergoing an upgrade (ALS-U) that is expected to yield even larger data outputs. By sharing resources and expertise with partners such as Argonne, Brookhaven, and SLAC, SYNAPS-I aims to build AI capabilities beyond what any single facility can accomplish on its own. “By pooling expertise and data across our facilities, we can build AI capabilities that benefit all users and accelerate scientific discovery in ways that no single facility can achieve alone,” added Heksemer.
Ptychography and image segmentation in X-ray/neutron science
Advances in X-ray microscopy and neutron scattering have led to increasingly complex datasets, requiring new approaches to data analysis and interpretation. SYNAPS-I aims to fill this gap. Scientists use these techniques to carefully study phase changes and molecular structures within active materials, which are essential to developing improved batteries and other advanced technologies. The primary method, ptychography, uses a lensless computational X-ray microscope to analyze samples at the atomic level. Advanced Light Sources (ALS) researchers used this technique more than a decade ago to achieve imaging of 5-nanometer structures in lithium iron phosphate. This breakthrough reveals insights into defect formation during chemical changes.
By building a machine learning pipeline, the SYNAPS-I team aims to significantly accelerate knowledge extraction from both X-ray microscopy and neutron scattering. This pipeline powers existing algorithms for automated ptychography and, importantly, image segmentation, the process of identifying features within complex X-ray and neutron data. Dimitrios Argyriou said, “SYNAPS-I is the first step into an exciting new era of science in modern facilities.”
ALS-U upgrade enables high-resolution data collection
Recently completed and ongoing upgrades of the National Science Facilities, with Lawrence Berkeley National Laboratory’s Advanced Light Source Upgrade (ALS-U) at the center stage, are poised to usher in a new era in materials science. This enhancement, combined with the SYNAPS-I initiative, is expected to dramatically accelerate the translation of complex image data into actionable scientific insights. SYNAPS-I, a multi-laboratory effort, is part of the Department of Energy’s Genesis mission and is designed to advance artificial intelligence and discovery across multiple disciplines. The main focus is streamlining traditionally time-consuming tasks such as ptychography (a lensless X-ray microscopy technique) and image segmentation.
The researchers previously used ptychography at ALS to image a 5-nanometer structure in lithium iron phosphate, a breakthrough that revealed the details of defect formation. Currently, manually identifying individual particles within a material during image segmentation is a laborious task. SYNAPS-I aims to replace this with an automated tool that can characterize particles in real time in an X-ray or neutron beamline. “Automated segmentation in advanced microscopy remains an important challenge in science,” said Dimitrios Argyriou, ALS-U project interim director. “With ALS, especially after the ALS-U upgrade, we will gain an unprecedented perspective on the inner workings of nature and technology.”
By pooling expertise and data across facilities, we can build AI capabilities that benefit all users and accelerate scientific discovery in ways that no single facility can achieve alone.
Alex Hexemer, Berkeley Lab Senior Scientist and SYNAPS-I Principal Investigator
Multi-lab collaboration powers innovative AI models
Department of Energy facility. Harnessing artificial intelligence to accelerate scientific advances in fields from energy to medicine. This multi-laboratory effort is a key component of the broader Genesis mission, which aims to use unique DOE data and facilities to build and deploy self-improving AI models. SYNAPS-I unites researchers from Berkeley Lab, Argonne, Brookhaven, SLAC, and Oak Ridge, demonstrating a commitment to sharing resources and expertise. Recent upgrades such as the ALS-U project are generating unprecedented amounts of data, creating great opportunities for AI-driven discovery.
