This project aims to combine ultrafast X-ray imaging and AI to see and understand rapid phase-change processes inside energy systems that are currently impossible to observe directly.
Many next-generation energy systems rely on phase-change processes that occur inside fast, opaque, and often cryogenic environments, limiting our ability to directly observe and optimize them. Gas–solid transitions such as carbon dioxide desublimation influence the efficiency and reliability of heat pumps, cryogenic heat exchangers, carbon capture and storage systems, and other critical infrastructure. However, current diagnostic methods are limited in their ability to capture how these processes initiate and evolve under realistic operating conditions. This project will develop a novel diagnostic platform that integrates ultrafast X-ray imaging with physics-guided artificial intelligence to reconstruct real-time, three-dimensional phase-change behavior. Radiographic images will capture transient structural changes, while AI-based reconstruction methods, incorporating material-specific attenuation properties and phase constraints, will infer physically consistent 3D morphologies from limited data. Initial experiments will use water frosting as a controlled test case, followed by carbon dioxide desublimation as a demonstration relevant to cryogenic separations and carbon management technologies. The resulting datasets and algorithms will provide new insights into nucleation pathways, interface motion, and morphological evolution, enabling improved design and optimization of advanced energy systems.
