Detection of Illicit HEU (235U) Production in a Fuel Cycle Enrichment Plant

The purpose of this effort is to demonstrate the ability to detect illicit HEU production in a declared, nominal LEU production plant. Illicit HEU production in a declared LEU production facility is a pernicious problem in international safeguards because the physical differences between LEU and HEU production are slight; indeed, both can use the same enrichment cascade. Previous measurements have measured the Uranium enrichment locally, meaning at a particular gas centrifuge or a particular header pipe. These were typically gamma-ray measurements that were collimated to view a particular point. The difficulty with this method is that modern gas centrifuge plants can be quite large, and the number of individual measurements required using this approach becomes prohibitive. By contrast, the method we employ uses an array, or a grid, of isotropic neutron detectors arranged on the ceiling of the enrichment cascade. The method builds on the Distributed Source Term Analysis Method (DSTA). The DSTA method made total neutron measurements using isotropically sensitive neutron detectors, on a grid of points around a facility to be measured. Then, an MCNP model was constructed of the entire facility and the grid of measurements. The quantity of nuclear material was adjusted as part of the input of the MCNP model, iteratively, until the MCNP output matched the actual measurements. The result was the nuclear material distribution. We have extended the DSTA approach to the detection of illicit HEU production based on two observations: First, the neutron detector response to the presence of nuclear material is linear. That is, the detector response is a linear, weighted sum of the contributions from various volume elements. There is no parametric (i.e. nonlinear) variation. Second, the total neutron emission is proportional to the enrichment for a gas centrifuge plant which uses UF6 gas. The alpha-n emission from the 234U alpha decay on the fluorine dominates the neutron rate. Because the system is linear, a simple matrix relationship exists between the distribution of neutron emission and detector response. This matrix can be inverted to provide a nearly instantaneous measure of neutron emission throughout the cascade. The significance is that an array of simple, inexpensive, identical, and somewhat remote detectors can provide a direct, automated, and near real time indication of illicit HEU production over the entire cascade hall.