AI-Enhanced X-Ray Tomography Tracks Heat Shield Ablation in Real Time

Researchers at the Department of Energy’s Lawrence Berkeley National Laboratory have achieved a breakthrough in aerospace engineering by capturing real-time, three-dimensional observations of spacecraft heat shield degradation during simulated atmospheric reentry. The study, published in the journal npj Materials Degradation, addresses a long-standing challenge in thermal protection system design: the inability to directly monitor the microscopic ablation process as it occurs. Traditional modeling of heat shield materials has relied heavily on pre- and post-test inspections, forcing engineers to extrapolate performance from static data. This limitation became evident following NASA’s Artemis I test flight, where actual heat shield performance diverged from computational predictions. To bridge this gap, a collaborative team from Berkeley Lab, the University of Illinois Urbana-Champaign, and NASA deployed in situ X-ray micro-computed tomography at the Advanced Light Source. The technique subjects commercial ablators to precisely controlled temperatures up to 1,652 degrees Fahrenheit, atmospheric pressure, and reentry-equivalent gas mixtures while capturing continuous volumetric scans. To overcome the trade-off between rapid data acquisition and microscopic detail, the researchers integrated AI-driven generative adversarial networks. By training the algorithm on high-resolution pre- and post-exposure snapshots, the system enhanced lower-resolution time-lapse scans into a complete, high-fidelity record of structural evolution. This approach allowed the team to track multiphase chemical decomposition and porosity changes in real time. The imaging revealed fundamentally different degradation pathways for two widely used ablative materials. SLA-561V, which incorporates cork as a structural filler, underwent complete thermal breakdown of the cork, leaving behind isolated, empty pockets. In contrast, SLA-220 lacks organic fillers and instead undergoes a thermomechanical transformation, forming a dense, branching network of interconnected silicone channels. These microstructural distinctions directly influence how heat and reentry gases permeate the material, dictating its overall thermal protection efficiency. According to project researchers, including NASA Johnson Space Center scientist Vishnu Oruganti and Advanced Light Source senior scientist Liz Clark, the dataset provides unprecedented empirical validation for predictive modeling. By directly observing internal structural changes during heating, engineers can now refine computational models with greater accuracy, reducing uncertainty in mission planning and thermal shielding design. The technology has already been applied to major NASA ablative systems, including those protecting the Perseverance rover during Mars entry and components slated for upcoming Artemis and deep-space exploration missions. This integration of synchrotron imaging and generative artificial intelligence establishes a new standard for materials characterization under extreme conditions. The resulting high-resolution degradation maps enable faster iteration cycles for thermal protection systems, directly enhancing the safety, reliability, and operational readiness of future crewed spacecraft navigating planetary atmospheres.