Speaker
Description
This research work consists in the design, development, and experimental characterization of a γ-ray spectrometer based on large lanthanum bromide scintillator crystals (3” × 3”) coupled with SiPMs. In nuclear physics experiments where photon’s energy ranges from 100 keV to 30 MeV, GAMMA provides state-of-the-art energy resolution (<3% at 662 keV) with a compact, modular and robust structure.
The interaction position reconstruction in the crystal volume is a fundamental information to compensate the Relativistic Doppler effect which leads to an undesired energy shift of the measured photon’s energy. To accomplish this task, imaging capabilities have been implemented on GAMMA: different Artificial Intelligence algorithms such as Decision Trees and Neural Networks have been tested. Results about 1-D and 2-D position sensitivity will be shown explaining how an RMS-Error lower than 1.5 cm$_{rms}$ has been met on the 2-D reconstruction; furthermore, an estimate of the DOI in such a thick scintillator will be presented.
Finally, Artificial Intelligence algorithms have been synthesized in an FPGA to allow Real Time position sensitivity. Important results about latency and percentage of hardware resources used (DSP, LUT) will be discussed comparing Decision Tree and Neural Network solutions.
