Speaker
Description
In recent years, some materials from the elpasolite crystal family have been under development for either or both gamma ray and neutron detection. These include Cs2LiYCl6 (CLYC) and Cs2LiLaBr6 (CLLB). Since these crystals have different luminescence decay times under neutron and gamma irradiation. The pulse shape discrimination (PSD) is widely used to discriminate between neutron and gamma signals in nuclear detection. In our previous work, the PSD Figure-of-Merit (FOM) value was optimized to 2.5 for CLYC, 1.2 for CLLB by using PSD method. However, the discrimination between neutrons and gamma rays becomes more difficult in the low gamma-equivalent energy region. Therefore, we trained the neutron and gamma waveforms measured by CLYC and CLLB crystal coupled PMT under Am-Be source irradiation based on the Convolutional Neural Network (CNN) method. The PSD FOM value of the CNN method was better than 6.0 at the gamma-equivalent energy region of more than 300 keV. In addition, we constructed CNN model for complicated n-g discrimination under piled-up condition, the accuracy for the particle identification is over 97% for each class(g+g, g+n, n+g and n+n). The timing information carried by the PMT waveform is also important when reconstructing the particle trajectory and discrimination of particles. With the development of fast analog-to-digital converter, the whole waveform information could be available. Compared with the traditional Constant Fraction Discrimination (CFD) timing method, a new one based on the CNN model for the timing of a pair of Cherenkov-detection MCP-PMTs improves the coincidence time resolution (CTR) by 50%. These results show that the CNN model is not only suitable for identification of neutron gamma signals, but also for the the reconstruction of particle trajectories.
| Consider for long presentation | No |
|---|
