Machine Learning Waveform Fitting to Improve Energy Resolution in P-Type Point Contact Germanium Detectors

The Majorana Demonstrator is an array of High-Purity Germanium (HPGe) detectors searching for neutrinoless double-beta (0νββ) decay in ⁷⁶Ge. The Demonstrator combines low-noise electronics with the excellent intrinsic energy resolution of HPGe detectors to attain the best energy resolution of any 0νββ search, thus reducing the background rate in the 0νββ region-of-interest. Trapping of drifting charges in the bulk of the detectors can negatively impact the energy resolution and detector performance. The Majorana collaboration has developed a machine learning approach that can model and fit HPGe detector signals with sub-percent precision, and can be used to extract event information from a waveform. Once a number of parameters describing the detector and electronics response are determined, this waveform fitting technique can be used to infer the interaction locations for a particular event. Using the deposition origin, we can model the drift path of the charge-carriers. A charge trapping correction based on the carrier’s drift path length, which is in development, has the potential to further improve the Demonstrator's overall energy resolution.