Autonomous Characterization of Materials Across Scales 

The National Nuclear Security Administration's (NNSA’s) efforts to modernize the stockpile rely on the evaluation of alternative metallurgical processing techniques and alloy designs, yet today’s microstructural characterization is so labor-intensive that only a fraction of critical features are analyzed, limiting data for producing process–structure–property–performance maps. 

The Autonomous Characterization of Materials Across Scales project removes that burden with an AI that autonomously operates advanced microscopy tools—including legacy systems with no programming interface—to deliver comprehensive data 30–40× faster with 5× more measurements per sample, enabling higher-quality process–structure–property–performance relationships and accelerating the pace at which materials and manufacturing processes can be qualified for the stockpile.

Frontier AI for Accelerated Materials Qualification

Accelerating the process of robust detection of material and component defects in national security applications, by adapting multi-modal foundation models to distinguish major defects requiring material reprocessing from acceptable minor defects and by developing targeted image manipulations to boost detection accuracy, robustness, and data efficiency.

PNNL POC: Aaron Luttman