Comparative Morphological Signatures of Strike Ordered Uranium Oxides for Nuclear Forensics

Area of Science:

• Nuclear Chemistry
• Materials Science
• Machine Learning Applications

Background:

• Nuclear forensics is crucial for detecting illicit radioactive materials.
• Characterizing uranium ore concentrates (UOCs) is essential for nuclear security.
• Understanding UOC synthesis pathways and their resulting properties is vital.

Purpose of the Study:

• To investigate how precipitation strike order influences UOC morphology.
• To differentiate synthetic routes and strike orders using machine learning.
• To establish morphological fingerprints for forensic attribution of UOCs.

Main Methods:

• Synthesized UOCs (U3O8) via four routes: AUC, ADU, UO4, and MDU.
• Analyzed SEM images using a ResNet34-based CNN for morphological feature extraction.
• Employed machine learning classification to distinguish between synthetic routes and strike orders.

Main Results:

• Most synthetic routes showed significant morphological differences based on strike order (p < 0.05).
• CNN analysis achieved high discrimination accuracy for AUC, ADU, and UO4 routes.
• The MDU route exhibited minimal morphological variation and lacked statistical significance (42-72% accuracy).

Conclusions:

• The influence of strike order on UOC morphology is synthesis-route dependent.
• Distinct morphological signatures were identified for AUC, ADU, and UO4, but not MDU.
• Strike order-dependent morphological fingerprints have implications for nuclear forensic attribution and process monitoring.