UC Berkeley

This dissertation examines the measurement of nuclear resonance fluorescence gamma-rays as a technique to non-destructively determine isotopic compositions of target materials that are of interest for nuclear security applications. The physical processes that can result in non-resonant background to nuclear resonance fluorescence measurements are described and investigated using a radiation transport computer code that relies on the Monte Carlo technique, MCNPX. The phenomenon of nuclear resonance fluorescence is discussed with consideration of the angular distributions of resonance emissions, the effects of nuclear recoil, and the influence of thermal motion.

Models describing two ways of measuring nuclear resonance fluorescence rates in materials are considered. First the measurement of back-scattered photons is considered. In this type of measurement, the portion of the interrogating photon beam that is scattered into large relative angles is measured. When the radioactivity of the target can be overcome by shielding or by use of intense photon sources, direct measurement of gamma-rays, emitted during nuclear resonance fluorescence can provide quantitative signatures that appear to be useful for applications such as forensic age-dating of large radiological sources. However, if the target radioactivity is too intense, as in the case for most spent nuclear fuel, a second measurement type, where indirect measurement of transmitted resonant-energy photons can also provide quantitative information. This method allows radiation detectors to be better-shielded from target radioactivity, but suffers from a slower accrual rate of statistical confidence. The models described herein indicate that very intense photon sources and large high-resolution detector arrays would be needed to measure ²³⁹Pu content in spent fuel to precisions desired by nuclear safeguards organizations. However, the rates at which statistics accrue are strongly proportional to the strengths of the resonances, and measurement of a plutonium isotope with stronger resonances may provide more practical measurement rates.

The model for predicting relative detection rates of nuclear resonance fluorescence gamma-rays in the transmission measurement was experimentally tested using the ²³⁸U in a mixture of depleted uranium and lead as a surrogate for ²³⁹Pu in spent fuel. The experiment indicated that the model was approximately correct, but that the process of notch refilling, which was excluded from the initial model, appears to be visible. Data files of the computer code, MCNPX, were modified to allow for nuclear resonance fluorescence to be simulated and a bug in the code was repaired to allow the code to more accurately simulate non-resonant elastic photon scattering. Simulations using this modified version of MCNPX have indicated that the magnitude of the notch refill process is comparable to that of the difference between the analytical model and the experimental data.