Deconvolving Plastic Scintillator Gamma-Ray Spectra Using Particle Swarm Optimization

Abstract

Plastic scintillators made from Polyvinyl Toluene 1(PVT) doped with fluorescent dye are used extensively in homeland security, scrap metal inspections, and other applications that require a large-area, cost-effective gamma detector. Unfortunately, gamma ray detection with PVT results in only Compton-edge events which contribute to a broad continuum spectrum having no discernible features. Other methods of obtaining information from low resolution PVT spectra have been described in publications but at the present time there is no practical method suitable for routine use. We have developed a method based on Particle Swarm Optimization (PSO) which analyzes raw PVT spectra and provides a histogram of contribution(s) vs. incident monoenergetic gamma energy(s). This is accomplished by summing multiple single-energy calculated PVT gamma responses into a ‘spectrum’ until a ‘best fit’ to the original raw data spectrum is obtained. The input set of response functions are calculated using MCNP5 and cover an appropriate energy range; we use a set of 255 response functions with 11 keV spacing between them (this corresponds to the ADC gain used in our PVT detectors). This set provides a range of 11 keV to 2805 keV for incident monoenergetic gamma energies which is suitable for most applications. Usually, we find that only one or two gamma response functions contribute to a ‘peak’ in the calculated histogram. Traditional radionuclide identification methods can be applied once the contribution gamma energies have been identified. As an added benefit: locating natural background gamma rays: 238 keV, 609 keV, 1460 keV, and 2614 keV in the deconvolved spectrum can be used to gain-stabilize the PVT detector.