Pulse shape classification in liquid scintillators using the fuzzy c-means algorithm

A new approach to pulse shape classification for neutron detectors of type BC501A has been investigated. The method is based on the fuzzy c-means (FCM) algorithm, which allows finding clusters of similar shapes in a set of digitized detector pulses. The aim of the method is to provide principal pulse shapes, which further can be used to apply a pulse shape based particle identification. Since the method is not adapted to the case of liquid scintillator signals it is of general use and can also be applied to signals of other detector types. A detailed study of the quality of the FCM method for the search of principal pulse shapes in BC501A liquid scintillators is presented using a 500 Msample/s 12 bit digitizer and a 252Cf neutron source. A comparison to principal pulse shapes extracted using time-of-flight information proves the applicability of the method. Finally, an example of a pulse shape discrimination method based on the extracted principal pulse shapes is presented and compared to the well-known integration method.