Monte Carlo Modeling of Correlations in Pulsed Neutron Data

Identifying special nuclear materials concealed within various objects or encased in shielding continues to be of interest in ensuring national security in the United States. One particular method of identification involves using a pulsed 8-10 MeV bremsstrahlung photon beam to interrogate an item of interest. If special nuclear materials are present, then the photons will induce photofissions and photoneutrons from the nuclear material. Helium-3 tubes surrounded by polyethylene and cadmium can then be used to detect fast neutrons from the object. Experimental data has been collected using this method at Idaho National Laboratory (INL) showing discrimination between fissile and other nuclear materials by measuring the auto- correlation in count rates in the detectors following the pulse [1], [2]. The pulsing of photons at a fixed repetition rate means that the conventional point model coincidence counting equations no longer apply to the measured quantities (‘singles’ and ‘doubles’ rates). A dedicated Monte Carlo program has been written that simulates the key processes of the neutron behavior, such as detection probability, thermalization, induced fission probability and deadtime. This has proven to be capable of accurately reproducing the experimental data from INL. The program has been used in conjunction with traditional shift register analysis to gain an understanding of doubles behavior in the non-steady state case and to show how the measured quantities are related to the key quantities of interest such as induced fission rate.