Destructive and Nondestructive Analyses of Plutonium Separated from a Low-Burnup Low-Enriched Uranium for Nuclear Forensics

Abstract

The ever-changing global geopolitical landscape greatly influences how nuclear threats grow and evolve. To counteract these threats, equal growth must be present in the research and development (R&D) sector to counter the use of nuclear weapons or nuclear explosive devices. The work presented here examines tools and techniques which have been used for nuclear forensics (NF), international nuclear safeguards, and nuclear nonproliferation. These tools include neutronics codes, radiochemistry, mass spectrometry, radiation spectroscopy, and machine learning, to name a few. The objective of the dissertation research was to validate NF techniques developed at Texas A&M University that can attribute “interdicted” Pu by determining the reactor-type that produced it, fuel burnup (a measure of irradiation time in a reactor), and time since irradiation completed (TSI). The research conducted for this dissertation aimed to add an additional experimental point (coordinates: reactor-type, fuel burnup, and TSI) in the database of isotopic signature ratios estimated as a function of reactor-types, fuel burnup, and TSI by performing Monte Carlo radiation transport code (MCNP) simulations. Previous research had validated two other points in the database for depleted U irradiation in a fast neutron spectrum and natural U irradiated in a thermal neutron spectrum. Specifically, this research investigated the capability of MCNP to accurately predict technical forensic signatures of Low-Burned Low-Enriched Uranium (LEU) fuel material irradiated in a thermal neutron spectrum. Forensic signatures, which included intra-elemental ratios of Pu and fission products, were to be fed to statistical based technical NF methods. Furthermore, MCNP calculations were to be validated with the experimental data obtained from neutron irradiation of LEUO2, chemical separation of Pu, and by characterizing the LEUO2 material used for neutron irradiation. The objective encompasses four parts. First, an examination of the efficacies of the neutronics codes, specifically the Monte Carlo (MC) neutronics codes, used for predicting the nuclide concentrations in neutron irradiated nuclear fuel by simulating the nuclear fuel depletion in nuclear reactors. These codes were evaluated against experimental values in benchmark and validation studies. On their own, MC codes do not simulate nuclear fuel depletion, rather radiation/particle transport. Therefore, MC codes can be coupled to nuclear fuel depletion solvers, such as CINDER90 or ORIGEN2 based on Markov Chain and Matrix Inversion methods, respectively, to solve the Bateman equation of isotope generation and depletion. Additionally, various nuclear data libraries can be used. Combination of these two parts can affect the accuracy of depletion calculation results. From our literature review of the various validation and benchmark studies on nuclear fuel material irradiation, we found that the nuclides 133Cs, 135Cs, 137Cs, 148Nd, 239Pu, 240Pu, and 241Pu production in irradiated nuclear fuel were accurately predicted by MC codes. However, nuclides 125Sb, 242Cm, 243Cm, 244Cm, 245Cm, and 246Cm were poorly predicted. Second, perform destructive and nondestructive assays (DA and NDA) to measure NF signatures of irradiated nuclear fuel, specifically in low enriched uranium dioxide (LEUO2) irradiated in a thermal neutron spectrum and compare the measured results with that obtained from the Monte Carlo radiation transport code (MCNP) simulations. The LEUO2 samples of a few milligrams were irradiated to a burnup of ~1GWd/MTU, cooled, then dissolved in nitric acid to conduct gamma-ray spectrometry and mass spectrometry. Several nuclides were measured and compared with the MCNP predictions. The nuclides measured were 91Y, 95Zr, 95Nb, 103Ru, 133Cs, 134Cs, 135Cs, 137Cs, 136Ba, 138Ba, 140Ba, 140La, 141Ce, 144C, 149Sm, 150Sm, 152Sm, 153Eu, 154Eu, 239Pu, 240Pu, 241Pu, and 242Pu. Most of the nuclide concentrations were measured to be within 15% of the MCNP predictions. These nuclides were used to estimate fuel burnup and TSI. Third, measure total Q-value of separated Pu from the alpha decays of its isotopes using total DES. 239Pu/240Pu ratios were measured in a cryogenic microcalorimeter at Los Alamos National Laboratory (LANL) through a technique called total Decay Energy Spectroscopy (DES). The DES technique takes advantage of the physical phenomenon at the transition edge of superconductors. The total energy of individual alpha radiation decays can be measured with great accuracy. Separated Pu, which was produced in the High Flux Isotope Research (HFIR) reactor was measured and the 239Pu/240Pu ratio was found to be within two standard deviations confidence interval. Fourth, validation of NF evaluation methodologies using experimental measurements conducted through post irradiation examination (PIE) of LEU irradiated in a thermal neutron spectrum. Experimental results from second part discussed above were fed to two statistical methodologies which can predict the reactor-type the material came from, quantify fuel burnup, and calculate TSI. The two nuclear methodologies (Maximum Likelihood and Machine Learning) were developed at Texas A&M University. Both techniques were able to accurately classify the experimental data and quantify fuel burnup within one standard deviation confidence interval. TSI was not as accurately predicted due to inaccurate Cs measurements.