Test of pulse shape analysis using single Compton scattering events

TL;DR

This paper evaluates pulse shape analysis techniques using single Compton scattering events in germanium detectors, employing neural networks to distinguish between single-site and multi-site events with about 80% efficiency.

Contribution

The study introduces a neural network-based method to effectively identify single- and multi-site events in germanium detectors using pulse shape analysis, validated with experimental data.

Findings

Achieved ~80% identification efficiency for single- and multi-site events.

Successfully used coincident detection to tag single Compton scattering events.

Demonstrated the effectiveness of neural networks in event classification.

Abstract

Compton scattering is one of the dominant interaction processes in germanium for photons with an energy of around two MeV. If a photon scatters only once inside a germanium detector, the resulting event contains only one electron which normally deposits its energy within a mm range. Such events are similar to Ge-76 neutrinoless double beta-decay events with just two electrons in the final state. Other photon interactions like pair production or multiple scattering can result in events composed of separated energy deposits. One method to identify the multiple energy deposits is the use of timing information contained in the electrical response of a detector or a segment of a detector. The procedures developed to separate single- and multiple-site events are tested with specially selected event samples provided by an 18-fold segmented prototype germanium detector for Phase II of the GERmanium Detector Array, GERDA. The single Compton scattering, i.e. single-site, events are tagged by coincidently detecting the scattered photon with a second detector positioned at a defined angle. A neural network is trained to separate such events from events which come from multi-site dominated samples. Identification efficiencies of ~80% are achieved for both single- and multi-site events.