Optimization of the n/gamma separation algorithm for a digital neutron spectrometer

An analysis of various n/gamma separation algorithms for a digital neutron spectrometer based on a stilbene crystal is performed. The same set of digital signals is used to test the following n/gamma separation algorithms: the algorithm based on the correlation analysis; the algorithms based on the direct comparison of the pulse fast component and the pulse tail; SDCC (Simplified Digital Charge Collection) algorithm; PGA (Pulse Gradient Analysis) and FGA (Frequency Gradient Analysis) algorithms. It is shown that the algorithm based on the correlation analysis makes it possible to carry out the most effective n/gamma separation for the entire pulse areas range. The shape of the distributions of the separation parameter for electrons and protons is analyzed. The correctness of these distributions approximation by the Gaussian curve is justified. An algorithm for the selection of the recoil protons resulting from the neutron interaction with the scintillator is proposed. It is based on the dynamic change of the separation parameter threshold depending on the pulse area. The distributions of the separation parameter for electrons and recoil protons strongly overlap for small pulse areas. In this case, the threshold for the separation parameter is chosen on the basis of the fixed probability of the event false identification. In our work, for example, the contribution of false events was fixed at 1%. With the increase of the pulse area, the distributions for electrons and protons begin to separate well. In this case it is convenient to set the threshold on the separation parameter at the point of their intersection. The use of this algorithm makes it possible to control the fraction of the falsely identified events for the entire range of the pulse areas. This is especially important for low-energy recoil protons, for which the probability of false event identification reaches its maximum. This algorithm was applied to the measurements performed by the time-of-flight method with a Cf-252 source. The suppression degree of the prompt gamma-ray peak was 1100 at the neutron detection threshold of 275 keV. The contribution of the falsely identified events for the different parts of the time-of-flight spectrum using the proposed event selection algorithm was studied. The largest contribution to the false pulses was found for the neutron distribution edges - for the neutrons with the energies above 8 MeV and below 0.5 MeV, and it did not exceed 0.23% in the entire range of the pulse areas.