Pulse shape recognition for CdZnTe semiconductor detector by using multi-shaping amplifiers method with neural network algorithm

A new pulse shape recognition method with multi-shaping amplifiers, combined with a neural network algorithm, has been developed, where four pulse heights are sampled from one signal pulse through four linear amplifiers with different shaping time constants. The four pulse heights are used as characteristic parameters to recognize the pulse shape with a neural network. This method has been applied to signal processing for a CdZnTe semiconductor detector to improve the deteriorated energy spectra caused by pulse height deficits due to the different mobilities of electrons and holes in the detector. The neural network recognizes the pulse shape patterns and provides the corrective magnification factors of the pulse heights. After the corrective procedure, the energy spectrum for ¹³⁷Cs gamma-rays is improved from 9.3 keV to 7.4 keV in the energy resolution (FWHM) of the 662 keV gamma rays photopeak. The photopeak becomes a considerably symmetrical shape without a low-energy tail. It has been verified that this method is simple and useful for pulse shape analyses, which can be used for many other applications