Dead time correction of Feynman sampled rates suitable for neutron coincidence counting

Abstract

Neutron coincidence counting (NCC) based on shift-register logic (SRL) is extensively used by international nuclear safeguards inspectorates to verify special nuclear material inventories, particularly Pu. An alternative autocorrelation pulse counting analysis technique is founded on Feynman sampling and makes use of the well-known Feynman-Y statistic. Although the latter approach has been widely adopted by the reactor noise and sub-criticality communities it has not been used by the nonproliferation community. Both techniques may be interpreted using the same theoretical model and so should extract the same fundamental information about the detected neutron pulse train, namely the first and second reduced factorial multiplets that are determined by the joint properties of the measurement item and the surrounding external neutron detector. In the nuclear safeguards domain, a semi-empirical dead time treatment is conventionally applied to the trigger (singles) and doubles rates and, as already noted, the Feynman-Y method is not typically applied. In this article, we revisit the dead time treatment of Robba et al. (1983) for the Feynman sampling method. We show so that a dead time corrected multiplet analysis, familiar to safeguards practitioners, can be extracted. This is a first and necessary step to enabling a comparison of the relative performance of these two autocorrelation techniques on a like basis for deeply subcritical measurement items, which is the domain of concern to international nuclear safeguards inspectorates. From the analytical derivations it emerges that the Robba formalism, with minor corrections, and the commonly used NCC shift-register logic dead time correction formalism yield numerically equivalent corrections for practical purposes over the operational range of safeguards interest.