Estimating warhead design properties from multiple resonances in physical cryptographic warhead verification

Abstract

Further progress towards complete nuclear disarmament hinges upon the development of technologies that can reliably verify the authenticity of nuclear weapons without revealing classified information about their design. One recently developed approach makes use of isotope-specific nuclear resonance fluorescence (NRF) measurements to verify a candidate warhead's isotopic and geometric characteristics. To conceal sensitive design information, the NRF signal from the warhead is not directly measured, but is first convolved with that of an encrypting foil containing relevant warhead isotopes in quantities unknown to inspectors. The measured NRF spectrum from the candidate warhead is then compared with that of a trusted template warhead to confirm or deny the authenticity of the candidate warhead. This work examines the information security of this protocol by evaluating the accuracy with which an inspector can estimate warhead properties from multiple NRF peaks. A mathematical model of the measured NRF spectrum is developed, and measurement spectra are simulated for a collection of simple, representative warheads. A least squares estimator for the areal density of each resonant isotope in the warhead is calculated. It is shown that for a single measurement with a 2.6 MeV bremsstrahlung beam, the areal densities of the fissile isotopes U-235 and Pu-239 are virtually always estimated with less than 100% error. By adjusting the beam endpoint energy to 2.16 MeV, the error on the plutonium estimators is shown to increase substantially. The effect of isotopic enrichment on the quality of inference is demonstrated by varying the grade of plutonium in the warhead from 6% Pu-240 to 18% Pu-240 by mass.