Radionuclide Identification by an EJ309 Organic Scintillator-Based Pedestrian Radiation Portal Monitor Using a Least Squares Algorithm

摘要

EJ309 organic liquid scintillators are powerful tools for detecting both gamma rays and neutrons because of their ability to perform pulse shape discrimination. This feature is highly desirable for radiation portal monitors (RPMs) that screen occupants for special nuclear material, or other smuggled radioactive material; however, using such low-Z detection media to perform radionuclide identification is difficult because they do not exhibit discernable photo-peaks in the detected pulse height distributions (PHDs). Consequently, a least squares algorithm was developed that compares measured PHDs to a database of detector responses from previously measured nuclides to identify which is present. The method was applied for a pedestrian RPM prototype developed at the University of Michigan consisting of eight 7.62-cm diameter by 7.62-cm length EJ309 organic liquid scintillation detectors. The detectors are operated at two different gain settings to allow measurement of a wide range of gamma-ray energies, which allows the system to identify lower energy sources, such as Am-241, as well as higher energy sources, such as Co-60. The algorithm was fine-tuned for the system to reliably identify a variety of radioisotopes traveling at 1.2 m/s, 100 cm in front of the RPM using 3 s measurements. Under these conditions, large variances in the PHD bins are produced when measuring radionuclides with activities on the order of several microcuries. Accordingly, bin widths are chosen in a manner that maximizes the number of counts per bin while retaining the distinguishing features of spectrum. The identification technique further overcomes the challenge of poor counting statistics by applying a weighting function to the bins that emphasizes the differences between the detector responses in the database. Noise suppression is implemented to avoid amplification of the statistical fluctuations in background by the weighting function. Additional techniques are used to direct the algorithm to operate only on the response of the detectors with either high or low gain applied to prevent misidentification. The capabilities of the RPM, with its implementation of the algorithm, will be demonstrated through participation in the SCINITILLA Benchmark campaign at the European Commission Joint Research Centre in Ispra, Italy.