Efficient machine learning approach for optimizing the timing resolution of a high purity germanium detector

R.W. Gladen,V.A. Chirayath,A.J. Fairchild,M.T. Manry,A.R. Koymen,A.H. Weiss

DOI: https://doi.org/10.1016/j.nima.2020.164505

2020-11-01

Abstract:<p>We describe here an efficient machine-learning based approach for the optimization of parameters used for extracting the arrival time of waveforms, in particular those generated by the detection of 511 keV annihilation <span class="math"><math>γ</math></span>-rays by a 60 cm³ coaxial high purity germanium detector (HPGe). The method utilizes a type of artificial neural network (ANN) called a self-organizing map (SOM) to cluster the HPGe waveforms based on the shape of their rising edges. The optimal timing parameters for HPGe waveforms belonging to a particular cluster are found by minimizing the time difference between the HPGe signal and a signal produced by a BaF₂ scintillation detector. Upon applying these variable timing parameters to the HPGe signals, we achieved a <span class="math"><math>γ</math></span>-<span class="math"><math>γ</math></span> coincidence timing resolution of <span class="math"><math>∼</math></span>4.3 ns at the 511 keV photo peak (defined as 511 ± 50 keV) and a timing resolution of <span class="math"><math>∼</math></span>6.5 ns for the entire <span class="math"><math>γ</math></span>-ray spectrum—without rejecting any valid pulses. This timing resolution approaches the best obtained by analog nuclear electronics, without the corresponding complexities of analog optimization procedures. We further demonstrate the universality and efficacy of the machine learning approach by applying the method to the generation of secondary electron time-of-flight spectra following the implantation of energetic positrons on a sample.</p>