Suppression of $^{14}\mathrm{C}$ photon hits in large liquid scintillator detectors via spatiotemporal deep learning

2026-03-29

AI Summary

This study addresses the degradation of energy resolution in liquid scintillator detectors caused by low-energy positron signals contaminated by $^{14}$C β-decay photons. To tackle this challenge, the authors propose a hit-level discrimination method based on spatiotemporal deep learning. The approach innovatively integrates a gated spatiotemporal graph neural network with a Transformer architecture that encodes both scalar and vector charge information, enabling high-precision separation of $^{14}$C photon events from genuine $e^+$ events even under strong spatiotemporal overlap. Experimental results demonstrate that the method achieves a positron misidentification rate below 1% while attaining a $^{14}$C recall rate of 25%–48%, substantially improving the energy resolution of the total event charge.

Abstract

Liquid scintillator detectors are widely used in neutrino experiments due to their low energy threshold and high energy resolution. Despite the tiny abundance of $^{14}$C in LS, the photons induced by the $β$ decay of the $^{14}$C isotope inevitably contaminate the signal, degrading the energy resolution. In this work, we propose three models to tag $^{14}$C photon hits in $e^+$ events with $^{14}$C pile-up, thereby suppressing its impact on the energy resolution at the hit level: a gated spatiotemporal graph neural network and two Transformer-based models with scalar and vector charge encoding. For a simulation dataset in which each event contains one $^{14}$C and one $e^+$ with kinetic energy below 5 MeV, the models achieve $^{14}$C recall rates of 25%-48% while maintaining $e^+$ to $^{14}$C misidentification below 1%, leading to a large improvement in the resolution of total charge for events where $e^+$ and $^{14}$C photon hits strongly overlap in space and time.

Problem

• liquid scintillator
• carbon-14
• energy resolution
• photon hits
• neutrino detection

Innovation

• spatiotemporal deep learning
• graph neural network
• Transformer-based model
• liquid scintillator detector
• ¹⁴C background suppression