Nuclear Warhead Verification: Virtual Gamma-Ray Spectrometry for Template-Matching

Abstract

The verification of nuclear warheads for treaty compliance is technically complex and requires a highly confident verification of nuclear component identity without revealing classified information about that component. Over the past 50-60 years, verification and information barrier initiatives have been undertaken in attempts to address these requirements. Current nuclear arms-control agreements limit the number of deployed strategic weapons, but of the approximately 16,000 warheads in existence today, only about 3,000 are regularly monitored and verified based on treaty obligations. With-out the ability to track countries’ capabilities and stockpiles accurately, concerns arise about countries modifying their nuclear warheads by removing fissile material to use for other purposes or to give or sell to other parties. Thus, future arms-control agreements are expected to be more complex, necessitating the identification and authentication of unique nuclear components. The ability to use an inspection system with an information barrier for verification is essential to the establishment and implementation of such agreements. This thesis focuses on the template-matching approach to verification that requires the analysis and algorithmic comparison of gamma ray spectra. Studies were conducted to create accurate nuclear source and detector models for virtual gamma ray spectra analysis. The studies included several diversion scenarios to ascertain the detectability of geometric and material diversions of nuclear components. The results indicate that while certain nuclear component/material diversions are undetectable by any means of gamma ray spectra analysis, regardless of the verification technique, the template-matching approach is a viable method for detecting some of the most likely diversions. In particular, the removal or replacement of weapons-grade plutonium with reactor-grade plutonium is detectable via gamma ray spectra comparison, although only for diversions of specific amounts. While a 50/50 reactor-grade/weapons-grade material diversion is detectable as compared with a 100 percent weapons-grade plutonium ball, neither a geometric diversion of 50 or even 75 percent of source material nor a 25/75 reactor-grade/weapons-grade diversion is statistically detectable. This suggests that there is a specific point at which a diversion of either type is detectable. Future studies should seek to identify these diversion detection points in order to better establish the efficacy and limitations of the template-matching approach to nuclear source verification.