English

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

Instrumentation and Detectors 2026-03-31 v1 Artificial Intelligence High Energy Physics - Experiment

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^{14}C in LS, the photons induced by the β\beta decay of the 14^{14}C isotope inevitably contaminate the signal, degrading the energy resolution. In this work, we propose three models to tag 14^{14}C photon hits in e+e^+ events with 14^{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^{14}C and one e+e^+ with kinetic energy below 5 MeV, the models achieve 14^{14}C recall rates of 25%-48% while maintaining e+e^+ to 14^{14}C misidentification below 1%, leading to a large improvement in the resolution of total charge for events where e+e^+ and 14^{14}C photon hits strongly overlap in space and time.

Keywords

Cite

@article{arxiv.2603.27727,
  title  = {Suppression of $^{14}\mathrm{C}$ photon hits in large liquid scintillator detectors via spatiotemporal deep learning},
  author = {Junle Li and Zhaoxiang Wu and Guanda Gong and Zhaohan Li and Wuming Luo and Jiahui Wei and Wenxing Fang and Hehe Fan},
  journal= {arXiv preprint arXiv:2603.27727},
  year   = {2026}
}

Comments

14 pages, 11 figures