Clustering Method to Process Signals from a CdZnTe Detector

概要

論文の詳細を見る
The poor mobility of holes in a compound semiconductor detector results in the imperfect collection of the primary charge deposited in the detector. Furthermore the fluctuation of the charge loss efficiency due to the change in the hole collection path length seriously degrades the energy resolution of the detector. Since the charge collection efficiency varies with the signal waveform, we can expect the improvement of the energy resolution through a proper waveform signal processing method. We developed a new digital signal processing technique, a clustering method which derives typical patterns containing the information on the real situation inside a detector from measured signals. The obtained typical patterns for the detector are then used for the pattern matching method. Measured signals are classified through analyzing the practical waveform variation due to the charge trapping, the electric field and the crystal defect etc. Signals with similar shape are placed into the same cluster. For each cluster we calculate an average waveform as a reference pattern. Using these reference patterns obtained from all the clusters, we can classify other measured signal waveforms from the same detector. Then signals are independently processed according to the classified category and form corresponding spectra. Finally these spectra are merged into one spectrum by multiplying normalization coefficients. The effectiveness of this method was verified with a CdZnTe detector of 2 mm thick and a 137Cs gamma-ray source. The obtained energy resolution was improved to about 8 keV (FWHM). Because the clustering method is only related to the measured waveforms, it can be applied to any type and size of detectors and compatible with any type of filtering methods.
社団法人日本原子力学会の論文
2001-07-25

Clustering Method to Process Signals from a CdZnTe Detector

スポンサーリンク

概要

著者

関連論文

スポンサーリンク