scieee AI-readable full text Open interactive document viewer

Enhancing Low-Energy Neutrino Sensitivity in Super-Kamiokande with the Wide-band Intelligent Trigger

Nascimento Machado, Lucas

Abstract

Parallel talk presented at the XXI International Workshop on Neutrino Telescopes - Padova 29 September - 3 October 2025 (https://agenda.infn.it/event/44606/) On behalf of the Super-Kamiokande Collaboration

Full text

Enhancing Low-Energy Neutrino Sensitivity in Super-Kamiokande with the Wide-band Intelligent Trigger September 30th, 2025 aSchool ofPhysics & Astronomy, University ofGlasgow +Correspondence: [email protected] L. N. Machado a,+, on behalf of the Super-Kamiokande Collaboration 1 Physics in the low-energy regime (< 4 MeV): •Access to the full solar neutrino spectrum (pp, 7Be, CNO) → Upturn in the electron neutrino survival probability expected at lower energies. •Search for astrophysical neutrinos, not yet observed. •Signals from reactor / geo-neutrinos also sit in the fewMeV window. The Super-Kamiokande Experiment The Super-Kamiokande (SK) Neutrino Detector is a50 kton water Cherenkov located in the Kamioka mine in Japan, overburden with 1000 m of rock. In operation since April 1996. 41 m 39 m 22.5 kton fiducial volume 2 > 11000 PMTs Super-Kamiokande is unique in covering neutrinos from MeV solar energies to TeV cosmic rays, while probing rare processes like nucleon decay. Accessing those energies promises new constraints on solar models, neutrino spectra, supernova physics, and star formation histories. Pushing traditional triggers hit threshold down to reach < 4 MeV causes huge fake rates from dark noise + radioactivity → DAQ/bandwidth blow-ups. Online reconstruction is the key discriminator •Reconstruct and preform vertex reconstruction of recorded hits in real time → reject random/noise hits before they look like an event. •Keep efficiency while suppressing fake triggers at lower energies. •Use parallel computing to keep up in real time The Wide-band Intelligent Trigger (WIT) has been developed to simultaneously trigger and reconstruct very low energy electrons (above 2.49 kinetic MeV) with an efficiency close to 100%. 3 The Wideband Intelligent Trigger (1) WIT is a computing farm running in parallel to other SuperK triggers. •17 machines processing data in parallel + one organizer. •~850 hyper-threaded cores dedicated to real-time data processing. •Each core handles blocks of 23 ms data files sequentially, applying a set of criteria to select goodquality events. •Floating threshold of 11 hits above dark noise rate. 4 The Wideband Intelligent Trigger (2) WIT also hosts: •Pre-supernova alarm: SK-only and SK+KamLAND •SN burst trigger •SN-triggered raw data saving system •Gravitational-wave follow up trigger (in development) Organizes reconstructed events that arrive timeunordered. Data files grouped into segments lasting about 1.5 minutes each. Using calibration data from Ni/Cf source at different positions in the SK detector (Z= +12, 0, -12). Two different methods compared: •Time of Flight (ToF) method: ≥ 8 hits above background in 20 ns with ToF consistent with source → reconstructed vertex. •Standard (STD) WIT chain: ≥ 11 hits above background in 230 ns over background → usual reconstruction (+ standard cuts). 5 WIT – Trigger Efficiency 𝜀 = 𝑆𝑆𝑇𝐷 −λ1𝐵𝑆𝑇𝐷 𝑆𝑇𝑜𝐹 −λ2𝐵𝑇𝑜𝐹 WIT trigger efficiency: Stable performance across source positions (Z = +12, 0, −12 m). WIT delivers high efficiency at few-MeV energies Find muon-induced hadronic showers by tagging neutron clouds (2.2 MeV γ from n-H capture) using WIT data Using space–time correlations + likelihood cuts; validate with a new FLUKA-based spallation simulation. Crucial to reduce backgrounds for solar neutrino analysis. Muon → Shower → n-captures → WIT tag → Likelihood cut. •Reconstruct neutron clouds from many 2.2 MeV γ-tags after a muon. •Cut on transverse (lt) and longitudinal (lLONG) distances to the cloud → isolate spallation decays from true low-E events. •Preselection + likelihood cuts (~90% spallation cut at high signal efficiency). 6 Spallation Reduction Main target: 16N (from (n,p) interactions on 16O): 7.3s half-life, several meters from muon track. Impact on SK-IV solar analysis: New cuts using neutron-cloud info reduce deadtime by up to ~55%. +12.6% solar signal efficiency for 2790 × 22.5 kton·day exposure (equivalent to one extra year). This framework opens lower-energy windows and strengthens solar and other low energy analyses, especially with SK-Gd. New Methods and Simulations for Cosmogenic Induced Spallation Removal in Super-Kamiokande-IV - Phys. Rev. D 110, 032003 (2024) Lower the energy threshold of the SK-IV solar analysis, using boosted decision trees (BDT) in WIT data, aiming to observe upturn in the electron neutrino survival upturn (vacuum > MSW). •WIT data available for 618 days of SK-IV; •Many advanced machine learning methods compared, with BDT offering high background rejection; •Analysis of the solar neutrino flux at Ekin < 3.49 MeV. Observed/Expected = Background levels remain too high to have a meaningful effect on the study of the solar upturn. 7 Solar Analysis with WIT 0.11−0.36 +0.37, for Ekin 2.49–2.99 MeV 0.425−0.064 +0.148, for Ekin 2.99–3.49 MeV -> 2.66σ 8 Super-Kamiokande with Gadolinium 40 tons Gd2(SO4)3*8H2O 0.03% Gd Nuclear Inst. and Methods in Physics Research, A 1027 (2022) 166248 July/August 2020 13 tons Gd2(SO4)3*8H2O 0.01% Gd June/July 2022 Nuclear Inst. and Methods in Physics Research, A 1065 (2024) 169480 SK-VI: SK-VII/ SK-VIII: ~8 MeV ΔT0.03% Gd ~60 µs Enhance sensitivity to low energy electron anti-neutrinos! ΔT ~200 µs 2.2 MeV Neutron source: Am/Be Scans for IBD candidates in 10 second windows. •If >10 candidates are found, a SN alarm is issued. •Raw data buffer: last ~5 min continuously kept and saved on alarm. WIT can handle all SN-related data and issue SN alert in ≤ 22 s, even for a nearby SN (LED burst tests). Future improvements: •Deliver in ~40 s: event count, energy spectrum, and SN direction; •Machine learning-based methods for SN direction estimation. 9 WIT online SN alarm Distance [kpc] EGADS: Evaluating Gadolinium’s Action on Detector Systems Test facility for SK-Gd, standalone detector. HEIMDALL: High Efficiency IBD Monitoring Detector and Automated caLL can issue an alert even if SuperK is on Test/Calibration run or down! https://www-sk.icrr.u-tokyo.ac.jp/egadsSNalarm/ For automated emails: [email protected] Compact binary mergers are prime candidates for neutrino–gravitational wave coincidence searches: •Neutron Star–Black Hole (NSBH) mergers: may produce both high-energy and thermal neutrinos, depending on remnant and ejecta. •Binary Neutron Star (BNS) mergers: emit low-energy neutrinos (MeV) within milliseconds of the GW signal. Models expect that BNS and NSBH mergers emit thermal MeV neutrinos within < 1 s of merger, with peak luminosity > 10⁵³ erg/s, in the 5 to 30 MeV range. Higher luminosity: ҧ𝜈𝑒. 16 Gravitational Wave Follow-up in WIT Look for coincidences in SuperK in both lowand highenergy samples. Currently implementing real-time system in WIT: Summary Spallation reduction •Aimed to suppress long-lived isotopes (e.g., ¹⁶N). •Reconstruct neutron clouds from many 2.2 MeV γ-tags after a muon •Cuts save deadtime and keep signal efficiency. Solar analysis •Higher low-E efficiency (sub-4 MeV window) + BDTs on WIT data. •Aiming to observe upturn in the electron neutrino survival upturn. •2.66σ evidence for a solar signal in WIT. Supernova online monitor •Searches IBD candidates in 10 second windows; alarm if >10 found. •Improvements on-going: deliver in 40 seconds: event count, spectrum, improved direction fitter using machine-learning. Pre-supernova (pre-SN) •Sensitivity to few-MeV electron antineutrino from late burning (Si) hours–days before collapse. •Potential early-warning channel for nearby massive stars. Reactor neutrinos •Low-E capability + n-tagging improves reactor neutrino search. •Based on preSN selection. •ON/OFF and full data analyses conducted. •Currently validating and paper in preparation. GW follow-up •Rapid, low-threshold pipeline supports gravitational-wave alerts: prompt low-E neutrino searches. 17 The Wideband Intelligent Trigger (WIT): is a software trigger with online reconstruction running in parallel to standard triggers. The WIT system is a computing farm with 20 machines (19 processing data in parallel + one organizer) and ~900 hyper-threaded cores. WIT pushes traditional triggers hit threshold down to reach < 4 MeV causes huge fake rates from dark noise + radioactivity.