Reduction of coincident photomultiplier noise relevant to astroparticle physics experiments

https://doi.org/10.1016/j.nima.2005.01.319Get rights and content

Abstract

In low background and low threshold particle astrophysics experiments using observation of Cherenkov or scintillation light it is common to use pairs or arrays of photomultipliers operated in coincidence. In such circumstances, for instance in dark matter and neutrino experiments, unexpected PMT noise events have been observed, probably arising from generation of light from one PMT being detected by one or many other PMTs. We describe here experimental investigation of such coincident noise events and development of new techniques to remove them using novel pulse shape discrimination procedures. When applied to data from a low background NaI detector with facing PMTs the new procedures are found to improve noise rejection by a factor of 20 over conventional techniques, with significantly reduced loss of signal events.

Introduction

Photomultiplier noise can significantly limit the sensitivity of low background and low threshold astroparticle physics experiments searching for rare events such as neutrinos or WIMP dark matter using Cherenkov or scintillator technology. The drive for low threshold, which determines, for instance, the minimum energy of neutrino that can be detected in an astrophysical neutrino detector, requires detection of the lowest possible number of photons in the target medium. This translates to design geometries that use many photomultipliers to view the medium, operated in coincidence. Such designs provide the necessary good light collection but furthermore, because of the coincidence operation, provide an effective means of suppressing certain types of well known Photomultiplier Tube (PMT) noise, in particular single-photoelectron (SPE) noise and after-pulses. For these types of noise there is no causal relation between noise generated in a given tube and that in any other tube.

However, operation of these designs in which PMTs are effectively light coupled together, makes them vulnerable to other forms of PMT noise, in particular to effects often ascribed to “dynode glow” [1], [2]. This is believed to be due to light emitted from the dynode stack under electron bombardment being scattered and reflected back to the photocathode. This can produce noise in the host tube but additionally the photons produced can in principle also be observed by all light-coupled tubes. Such a coincident event is not suppressed by the coincidence operation and may mimic a signal event. Significant phenomena of this sort have been observed in the SNO detector, where they have been termed “flashers”. These are seen at 1000day-1 above an 18 hit threshold comprising relatively prompt flashes of light attributed to discharges from the dynodes [3]. In dark matter detectors based on NaI, liquid Xe and other scintillators, where high sensitivity and low background is needed, a likely related class of events sometimes termed “step” events are seen [4], [5]. The drive for lower thresholds and lower backgrounds means that such unusual and rare events will become an increasingly significant source of systematic error in these and other planned PMT-based low background detectors.

In this paper we describe investigation of such coincident PMT noise events and the development of new techniques to identify and suppress these events. The work was undertaken using apparatus at the Boulby Underground Laboratory comprising a low background shielding castle within which can be mounted pairs of face-to-face low background photomultipliers with and without scintillator in various configurations (see Section 2). The apparatus used is based on that of the modules of the Sodium Iodide (NaI) Advanced Detector (NAIAD) experiment [6]. The NAIAD experiment is a direct search for Weakly Interacting Massive Particle (WIMP) dark matter, comprising an array of 8 shielded low background NaI(Tl) crystals viewed by pairs of photomultipliers [7]. WIMP search experiments use various techniques to distinguish between nuclear recoil events, expected from WIMP interactions, and background electron recoil events. In the case of NaI(Tl) crystals, such as used in NAIAD, and several other experiments, this discrimination is achieved using pulse shape analysis to identify nuclear recoils [6], [7], [8], [9], [10], [11]. This is possible because in NaI(Tl), for instance, the scintillator time constant for nuclear recoil events is shorter (typically 170–200 ns) than for electron recoils (typically 240–300 ns).

Dark matter limits are set using this technique by determining statistically the maximum number of recoil events that may be present in a dataset. However, as with many such low background experiments it is crucial to have sensitivity at the lowest possible energy, to probe the maximum range of initial particle energy or mass. For this reason the coincidence threshold is usually set low, for instance at approximately 1 keV in NAIAD, allowing a sensitivity to WIMP masses in principle below 100 GeV. In such experiments the rate of PMT noise events is greatly reduced by using coincidence counting which demands that each PMT should see a minimum signal for an event to be accepted. However, any coincident PMT noise remains and strategies to reduce this (rather than just the SPE or non-causal noise) then become of major importance.

In low background experiments using optically coupled PMTs such as this, some suppression of coincident noise can be achieved by the following two techniques: (a) additional pulse shape analysis of the events, and (b) asymmetry cuts. The former technique is based on the observation that the distribution of time constants of low-energy PMT coincident noise events can be fitted to an exponential function. This has a faster characteristic mean decay time than that of most scintillator events (though slower than the normal PMT response time). This distribution can be fitted and then subtracted from the data, even though the distribution overlaps with the distribution of time constants for scintillation events. Fig. 1 shows a typical low-energy time constant distribution for NaI including this noise feature. The asymmetry cut technique is based mainly on comparing the characteristics of coincident events, such as amplitude, absolute start time of the pulse and pulse shape [6], [7], [12]. The assumption used is that coincident noise seen in the tubes is due to light emitted from one tube being seen by one or more other tubes, as suggested occurs in SNO and dark matter experiments [3], [5]. In this case, unlike the situation for genuine signal events, the characteristics are likely to be asymmetric, for instance the pulse in one PMT being of much larger amplitude than the other, or with faster rise time.

Although these techniques are useful they both have clear drawbacks. The coincident noise subtraction technique, although powerful, assumes that the exponential fit of short time constant events may be extrapolated to higher time constants, through the data events. This assumption introduces systematic errors into the results. Such a subtraction also necessarily entails the possible rejection of genuine events, requiring the introduction of an efficiency factor. The asymmetry cuts cannot remove coincident noise events which give pulses in the tubes of similar characteristic. The asymmetry technique also requires operation in coincident mode, which may not always be possible or desirable. For dark matter searches any PMT noise events misidentified as nuclear recoil events reduce the effectiveness of pulse shape discrimination analysis.

Based on the drawbacks above there is a need to develop new techniques to reject coincident noise events with fit time constants comparable to signal events. We describe here new work in this area. Firstly, the apparatus used to acquire and characterise samples of low background coincident PMT noise data is outlined. This is followed by details of how the data were analysed and compared with data taken with similar configurations but with scintillator present. We then discuss the development of analysis techniques designed specifically to identify and reject coincident PMT noise. Finally, we will show comparisons between the effectiveness of the new procedures and more traditional techniques.

Section snippets

Experimental setup and data

All PMT noise measurements were performed at the Boulby Underground Laboratory at 1070 m depth, using the apparatus shown in Fig. 2. The set up comprises a shielding castle, incorporating 15 cm of low background lead, inside of which is a further 10 cm layer of copper shielding. The castle was supplied with an automatic source dropper to allow remote calibration of the scintillator using γ-ray sources (57Co and 60Co). A copper support structure allows 5 in. PMTs to be mounted face-to-face in a

Investigation and improved reduction of noise

Typical asymmetry cuts used to reject noise from NAIAD data, reject events where the fit time constants of pulses from each PMT differ by 100 ns or more, or where the start times of the pulses differ by 100 ns or more, or where the energy associated with the pulses differs by 40% of the combined energy or more [5], [12]. Cuts based on these parameters represent the current best procedures but, as discussed above, do not adequately deal with the coincident noise phenomena. To improve the situation

Conclusions

An experiment to acquire a sample of pure PMT coincident noise data was performed using a low background shielding array underground at the Boulby facility. These data were analysed and parameters developed to distinguish and thereby cut this noise from real data. The newly developed cuts have been shown to be more effective at removing noise from experimental data and more efficient in preserving scintillation events compared to the asymmetry cuts currently used. In the typical scintillator

Acknowledgements

The authors would like to thank the members of the UK Dark Matter Collaboration for their valuable assistance and advice. We are grateful to the Particle Physics and Astronomy Research Council for financial support and to Cleveland Potash Limited for their assistance. M. Robinson would also like to thank Hilger Crystals for their support of his Ph.D. work.

References (12)

  • J.J. Quenby

    Phys. Lett. B

    (1995)
  • N.J.C. Spooner

    Phys. Lett. B

    (2000)
  • V.A. Kudryavtsev

    Astroparticle Phys.

    (2002)
  • R. Lüscher

    Nucl. Phys. B (Proc. Suppl.)

    (2001)
  • H. Park

    Nucl. Instr. and Meth. A

    (2002)
  • H.J. Kim

    Nucl. Instr. and Meth. A

    (2002)
There are more references available in the full text version of this article.

Cited by (6)

  • Search of SIMP particles with a ground based detector

    2019, Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment
  • The status of KIMS-NaI experiment

    2016, Nuclear and Particle Physics Proceedings
  • Study of scintillation in natural and synthetic quartz and methacrylate

    2014, Optical Materials
    Citation Excerpt :

    The results of the fit are summarized in Table 5. The rise time constant is of the order of the one expected for the PMT model used while the decay time constants are in the typical range for inorganic scintillators; for example, for oxides with Si time constants reported in [11] range from some tens to some hundreds ns. The decay time constants found are fully compatible with those derived when fitting only the decay of the pulse to two exponential components.

View full text