Simulation of Template Spectra for Scintillator Based Radionuclide Identification Devices Using GEANT4

The performance of radioisotope identification devices (RID) for homeland security applications is strongly influenced by nuclide identification algorithms. The identiFINDERtrade exploits template matching - a technique correlating measured gamma ray energy spectra with a library of reference spectra (templates) stored in the device. Of course the template quality impacts the identification performance. Templates can be measured, but such a purely experimental approach is cumbersome and might be expensive, particularly for nuclides which are not always accessible. Therefore Monte Carlo (MC) simulations have been utilized for template generation. The GEANT4 toolkit provides the benefits of an open source program with easy access to appropriate material and radioactive decay data, considering all relevant physical processes. However, dedicated computing techniques had to be applied in order to obtain the necessary (statistical) accuracy at a justifiable computing time. The implemented models have also been used to study the influence of detector size and material, source mantling or matrix, and detector-source geometry on the measured gamma ray spectrum.