Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
  • Open Access

Detector signal characterization with a Bayesian network in XENONnT

E. Aprile et al. (XENON Collaboration)
Phys. Rev. D 108, 012016 – Published 26 July 2023
PDFHTMLExport Citation
Abstract
Authors
Article Text
  • INTRODUCTION
  • THE XENONnT EXPERIMENT
  • SIGNAL CLASSIFICATION
  • EVENT SELECTION USING THE NBC SCORES
  • CONCLUSION AND FUTURE WORK
  • ACKNOWLEDGMENTS
  • APPENDICES
    • References

    Abstract

    We developed a detector signal characterization model based on a Bayesian network trained on the waveform attributes generated by a dual-phase xenon time projection chamber. By performing inference on the model, we produced a quantitative metric of signal characterization and demonstrate that this metric can be used to determine whether a detector signal is sourced from a scintillation or an ionization process. We describe the method and its performance on electronic-recoil (ER) data taken during the first science run of the XENONnT dark matter experiment. We demonstrate the first use of a Bayesian network in a waveform-based analysis of detector signals. This method resulted in a 3% increase in ER event-selection efficiency with a simultaneously effective rejection of events outside of the region of interest. The findings of this analysis are consistent with the previous analysis from XENONnT, namely a background-only fit of the ER data.

    • Figure
    • Figure
    • Figure
    • Figure
    • Figure
    • Figure
    • Figure
    1 More
    • Received 12 April 2023
    • Accepted 23 June 2023

    DOI:https://doi.org/10.1103/PhysRevD.108.012016

    Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. Funded by SCOAP3.

    Published by the American Physical Society

    Physics Subject Headings (PhySH)

    1. Research Areas
    Particle dark matter
    1. Techniques
    Bayesian methodsDark matter detectors
    Particles & FieldsGravitation, Cosmology & Astrophysics

    Authors & Affiliations

    Click to Expand

    Article Text

    Click to Expand

    References

    Click to Expand
    Issue

    Vol. 108, Iss. 1 — 1 July 2023

    Reuse & Permissions
    Author publication services for translation and copyediting assistance advertisement

    Authorization Required

    Log In
    Other Options
    ×

    Images

    • Figure 1
      Figure 1

      Individual examples of an S1 signal (green) and S2 signal (purple) from simulated data used as input for training. The S1 signal example has a total area of 18 PE and duration of 2000 ns; the S2 signal area is 1766 PE and duration is 12000 ns. Top: waveforms, downsampled to 50 samples total, with the elapsed time of the signal in the secondary x axis, illustrating the different temporal profiles between a typical S1 and S2 signal. Bottom: quantiles, 50 total, with relative total observed area in the secondary x axis.

      Reuse & Permissions
    • Figure 2
      Figure 2

      Graph structure of a naive Bayes classifier. Class node (C, either S1 or S2) is assumed to directly influence the value of each attribute node (Ai). Each attribute node is conditionally independent from all other attribute nodes, given the class node.

      Reuse & Permissions
    • Figure 3
      Figure 3

      Kr83m calibration data corrected S2 area vs corrected S1 area (cS2, cS1). Corrections on area are calculated to normalize detector effects that vary across time and space; see [3] for details. The color scale indicates the S1 signal’s NBC score [Eq. (3)]. The rightmost population contains the merged (41.5-keV) S1 signals, and thus is shown to have an NBC score that is less canonically S1-like, owing to the true underlying physical process being a merging of two S1 signals. The 32.1-keV (middle) and 9.4-keV (left) populations have S1 NBC scores which are more canonically S1-like. The S2 signals for the 32.1- and 9.4 keV are merged into a single S2 signal; thus, all three populations of S1 signals shown have equivalent-sized S2 signals.

      Reuse & Permissions
    • Figure 4
      Figure 4

      Performance of the S1 NBC and S2 NBC selection criteria on Rn220 calibration data, within the fiducial volume. A bin colored black indicates the fraction of remaining events to be exactly 0. Adjacent panels show the fraction of remaining events per bin, projected along the cS1 and cS2 axes. Events outside of the ROI are effectively targeted by the S1 and S2 NBC selection criteria. Above the ER band, gaseous S2 signals are misreconstructed into events within the fiducial volume, becoming the primary source of background events. Following the S1 and S2 NBC selection criteria, selections to target multiple-scatter events, mis-paired S1 and S2 signals, and accidental coincidences are applied, which remove the remaining events outside of the ER ROI.

      Reuse & Permissions
    • Figure 5
      Figure 5

      Comparison of total event-selection efficiency between this work and the previous work reported from XENONnT SR0 [3]. The step increase at 10 keV in efficiency is due to the nuclear-recoil region of data being blinded below 10 keV during both analyses. On average, this method is 3% more efficient, with the greatest relative improvement in efficiency being in the 2–20-keV energy region.

      Reuse & Permissions
    • Figure 6
      Figure 6

      Top panel: SR0 ER data (black) selected using the Bayesian network-based approach with the best-fit background-only model B0 (red). The subdominant AC contribution is not visible. Bottom panel: Δevents/bin corresponds to the difference in observed data in events/(tonne-year· keV) between those selected using the Bayesian network-based approach and those selected using the previous method described in [3]. The increase of observed events particularly at energies below 20 keV is consistent with the increase in efficiency of this work relative to [3].

      Reuse & Permissions
    • Figure 7
      Figure 7

      Observed S1 signals from the 32.1-keV (left), 9.4-keV (middle), and merged 41.5-keV (right) decays of Kr83m during SR0 calibration. In particular, the 41.5-keV peak has an S1 NBC score of 0, indicating that it is neither a canonical S1 nor a canonical S2. In reality, this is a merged double S1. The NBC scores between the three populations allow for selection of the merged waveforms based solely on shape. The NBC selection criteria described in Sec. 4 would ensure that such double S1s were rejected in ER event selection for analysis. The color of each waveform corresponds to the NBC score as shown in Fig. 3.

      Reuse & Permissions
    • Figure 8
      Figure 8

      Observed S2 signals from Kr83m calibration. A canonical S2 signal (left) produced from Kr83m decay passes the NBC selection criteria. A high-energy interaction in the GXe region of the detector produced a merged, noncanonical S2-classified signal (middle), and this signal is vetoed by the NBC selection criteria. Finally, an ionization signal produced in gas (right) is also rejected by the NBC selection criteria. Robust rejection of such events is important both for proper ER event selection, and for calibration and efficiency calculations.

      Reuse & Permissions
    ×

    Sign up to receive regular email alerts from Physical Review D

    Reuse & Permissions

    It is not necessary to obtain permission to reuse this article or its components as it is available under the terms of the Creative Commons Attribution 4.0 International license. This license permits unrestricted use, distribution, and reproduction in any medium, provided attribution to the author(s) and the published article's title, journal citation, and DOI are maintained. Please note that some figures may have been included with permission from other third parties. It is your responsibility to obtain the proper permission from the rights holder directly for these figures.

    ×

    Log In

    Cancel
    ×

    Search

    Article Lookup

    Paste a citation or DOI
    Enter a citation
    ×