Photon Detection System Performance in the DUNE 35 ton prototype - - PDF document

Prepared for submission to JINST

Photon Detection System Performance in the DUNE 35 ton prototype LAr-TPC detector

• F. Irst,a,b,1 S. Econd,c T. Hirda,2 and Fourthc,2 on behalf of the DUNE collaboration

aOne University,

some-street, Country

bAnother University,

different-address, Country

cA National Laboratory,

some-location, Country

E-mail: first@one.univ Abstract: The 35 ton (35t) prototype for the Deep Underground Neutrino Experiment (DUNE) far detector is a single phase liquid argon time projection chamber (LAr-TPC) integrated detector. The 35t took cosmic data for a six week run from the start of February to the middle of March 2016. The 35t had two drift volumes on either side of its anode plane assembly (APA) and utilized wire planes with wrapped wires and a photon detection system (PDS) read out by silicon photomultipliers (SiPMs). The PDS of the 35t demonstrated time resolution less than 100 ns, within the requirements

• f the DUNE far detector.

1Corresponding author. 2Also at Some University.

Contents

1 Introduction 1 2 The 35-ton Prototype 2 3 Photon Detector Operations 6 4 Photon Detector Performance 7 5 Conclusion 9

1 Introduction

Introduce argon scintillation and the role of the photon detectors in the intro. The Deep Underground Neutrino Experiment (DUNE) is an international, dual-site experiment that will study neutrino physics and search for proton decay. A beam of neutrinos with 2.5 GeV mean energy will be produced by the Long Baseline Neutrino Facility (LBNF) at Fermilab National Accelerator Laboratory (FNAL) and aimed 1300 km through the earth at a 40 kiloton liquid argon time projection chamber (LAr-TPC) far detector located in the Homestake Mine in SD. The beam produced at Fermilab will consist of νµ and will be measured with a near detector 574 m from the target; the far detector (FD) will measure νµ disappearance and νe appearance. [1] The DUNE FD will be the largest LAr-TPC ever constructed and will present multiple engineering and data- processing challenges. The cryostat, electronics, and field cage will need to be scaled up by a factor

• f 285. Cold digital electronics will be required to minimize the number and lengths of readout
• cables. In order to prototype and test the necessary technologies and solutions, the DUNE 35 ton

prototype detector was constructed and run at FNAL. The DUNE 35 ton prototype (35t) is a prototype single phase LAr-TPC integrated detector which tested DUNE far detector design and components. [2] Phase 1 of the 35t was a test of the membrane cryostat only from Dec. 20, 2013 to Feb. 15, 2014 and achieved the LAr purity required for LAr-TPC running, a 3 ms drift electron lifetime (equivalent to 100 ppt of O2). Phase 2 of the 35t tested new LAr-TPC features in a fully integrated system including both TPC and PDS to characterize the technology’s performance with cosmic ray observations. The 35t’s Phase 2 primary data-taking period was from Feb. 1, 2016 to March 12, 2016. Because the DUNE far detector’s TPC will have drift time on the order of milliseconds, it will be necessary to use another method to precisely measure the time of interaction T0 and match the interactions in the far detector with the neutrino beam spill timing.Not really. We need to be clear that this is only a requirement for non-beam physics. Make the focus NDK and SN physics. Maybe the way to structure this is an intro paragraph about what the PDs will do, then say we will use the 35ton to measure time resolution and coarse attenuation length. – 1 –

Figure 1. External muon counter positions on 35 ton

2 The 35-ton Prototype

The 35t was located on the Fermilab grounds in building PC4. This section will cover the experi- mental design. The primary focus will be on the photon detection system subsection. The 35t cryostat design is described in greater detail in (35 ton design paper). We should put in a few sentences of description here along with the reference. The Time Projection Chamber (TPC) consisted of two drift volumes inside the LAr-filled cryostat between the APAs and the CPAs on either side of them. High energy particles passing through the LAr ionized electrons along their path; these electrons drifted to the APAs by electric fields maintained between the CPAs and APAs. When reaching the APAs, the drift electrons induced signals on the wires wrapped around the induction planes and were then collected by the vertical wires on the collection planes. The induction and collection plane signals were read out by cold electronics and used to reconstruct 3D paths of the particles in the drift volumes. The TPC design and operation is described in greater detail in (35 ton design paper). The 35t used plastic scintillation counters placed around the outside of the cryostat structure to detect cosmic muons passing through the 35t. The positions of the external muon counters are depicted in Figure 1. Signals of coincidences of hits on multiple external muon counters in single event windows indicating throughgoing muons were used to trigger detector readout via the Penn Trigger Board (PTB) during triggered running mode of the system. The photon detection system (PDS) of the 35t consisted of the photon detectors (PDs) installed – 2 –

inside the cryostat, the SiPM Signal Processors (SSPs) used to read out and process the signals from the PD, and the calibration system that produced and diffused UV light inside the cryostat to test the capabilities of the photon detectors. The 35t contains eight separate photon detectors installed on the APAs. The photon detectors consisted of waveguides that led scintillation light to silicon photomultipliers (SiPMs) at their ends. SiPM outputs were read out by SSPs produced by Argonne National Laboratory. Figure 3 shows an example photon detector design. The eight photon detectors installed in the APAs tested several different waveguide technologies. Photon detector positions in the 35t APAs are shown in Figure 2. Photon detectors 0, 4, and 6 were developed by Colorado State University (CSU) and used a design of an array of wavelength-shifting fibers placed behind a radiator coated with TPB. The radiator converts scintillation light incident on it to visible blue light, of which approximately half is caught by the fibers and converted again to green light for transport to the top end of the detector, where they were read out by 8 SiPMs on each PD. Photon detectors 1, 3, and 7 were developed by Indiana University (IU) and used a waveguide consisting of four acrylic bars with 12 SiPMs and readout channels each (3 per bar). Each light guide was coated via dipping with a TPB solution dissolved in DCM at 0.6% by weight. The light guides convert scintillation light entering them into photons of visible blue light; total internal reflection catches a portion of the converted photons and conveys them to the 3 SensL C-series SiPMs on the readout end. Photon detector 2 was developed by Louisiana State University and used an acrylic plank coated with TPB with an embedded bundle of three wavelength-shifting fibers. Scintillation light that enters the plank is converted to blue light, which is captured either directly or after reflections by the fiber bundle which shifts the light to green photons and guided to both ends of the plank, where the fiber bundle is read out on both ends by one SensL B-series SiPM on each end. Photon detector 5 was developed by Lawrence Berkeley National Laboratory (LBNL) and used a waveguide four bar design with 12 SiPMs and readout channels each (3 per bar). [3] Each photon detector’s SiPMs were read out by one of seven SiPM Signal Processors (SSPs) designed and constructed by Argonne National Laboratory. The SSPs received the waveform output from the SiPMs as analog voltages, passed them through a fully-differential voltage amplifier, and digitized the waveforms with a 14-bit, 150 MSPS analog-to-digital converter (ADC). A Xilinx Artix-7 Field-Programmable Gate Array (FPGA) processed the digitized data from each channel with a leading edge discriminator to detect events. SSP readout can be configured in multiple ways including using external triggers for reading

• ut events or self-triggering on the measured waveforms when the amplitude exceeds a threshold

set for each channel. Both the externally triggered mode and self-triggering mode were employed during the 35t’s data run. In externally triggered mode, waveforms with the maximum allowed 2048 samples were saved (a length of about 15.5 µs) when triggers from the muon counters were

• received. In self-triggered mode, shorter waveforms with 700 samples (about 5 µs) were saved in
• rder to not overwhelm the 35t’s DAQ.

Calibration of the photon detector is important for quantifying phenomena such as the energy range of interest, the scintillation light’s fast and slow components, and the propagation of photons including reflections and scattering. The 35t’s calibration system was designed to be capable of examining the above phenomena as well evaluating multiple photon detectors’ relative efficiencies – 3 –

Figure 2. Photon detector positions in 35 ton

– 4 –

Figure 3. Example photon detector panel design, showing waveguide bar and SiPMs of the IU or LBNL designs.

and monitoring the stability and response of the entire PDS as a function of time. The 35t’s calibration system was UV-light based, using a set of UV LEDs in the VUV wavelength range as light sources. Light pulses were generated by a 1U rack mounted Light Calibration Module (LCM) which was based on a repurposed SSP unit and operated outside of the liquid argon cryostat in a NIM crate alongside the 35t’s SSPs. The UV LEDs inside the LCM were coupled to quartz fiber-

• ptic cable to transmit light via a feedthrough into the cryostat and detector volume at particular

locations on the cathode plane assembly (CPA) and distribute the light uniformly across the eight photon detector planks mounted in the 35t’s APAs. Five diffusers were mounted on the CPA with

• ne placed in the center and four mounted near the corners of the CPA, as illustrated in Figure 4.

Figure 4. Diffusers emit UV light (top left figure) mounted at five CPA locations shown via arrows (right figure). Quartz fiber transports the UV light from the LCM to diffusers (lower left figure).

The calibration system was tested by running the LCM to produce flashes of UV light in the detector volume while triggering the SSPs from the LCM in order to record the response to the light. Special calibration runs were taken with this setup, varying the pulse width and pulse amplitude while flashing only the central diffuser and also flashing the other diffusers in isolation with constant pulse width and amplitude to test the response against each other. Figure 5 shows the response of

• ne channel to flashes from the central diffuser with different pulse widths.

– 5 –

s) µ Time ( 1 2 3 4 5 Amplitude (ADC) 1400 1600 1800 2000 2200 2400 2600

40 ns 30 ns 20 ns 10 ns 6.67 ns 3.33 ns

Figure 5. Waveforms from photon detector channel 63 taken during calibration runs with LED voltage at 30 V and varying pulse widths in time from 3.33 ns to 40 ns.

3 Photon Detector Operations

In this section we describe the trigger rate studies performed in order to investigate possible origins

• f the noise in the TPC as well as in the PDS. One of the main motivations for this noise studies

was the high singles rate observed by µBooNE. We separate the noise as coming from two main sources: from possible radiological contamination, that could have been seen by µBooNE, and from

• electronics. For the former case we performed runs with a special configuration with photon detector

data only, externally triggered by muon counters and self-trigger with low threshold, meaning 2.3 photo-electrons for the well behaved channels. A run was performed for approximately 5 minutes every few hours taking 5.5 µs waveforms. This was a fairly stable configuration that allowed monitoring light sources and light yield as the detector conditions changed, as the purity increase with time, for example. The later case study was conducted looking for correlations between the RCE and SSP trigger rate RMS. The trigger rate per run is calculated taking the trigger rate average

• f the channels, ignoring channels with zero counts in that run.

– 6 –

Figure 6. Scatter plots for SSP and RCE RMS for different run periods. We observe a discretization of the RCE RMS as the run number increases, which also means as the time increases. Can we get just Run17000s as a figure to itself?

4 Photon Detector Performance

The DUNE far detector will use its photon detectors to measure T0 by reconstructing optical hits

• ccurring in conjunction with signals from the TPC.Waveforms of about 15.5 µs were saved when

the photon detectors were externally triggered by muon counters; an example is shown in 7. Optical hits are reconstructed by finding peaks on the waveforms. The optical hit is an object containing parameters of a pulse found on one of the SSP waveforms. In order to find optical hits the waveform is first processed by the pedestal-finder algorithm, which uses several of its first or last ticks to calculate the pedestal. Then the pedestal is passed to the

• ptical-hit-finder algorithm, along with several other parameters (the primary and the secondary

thresholds and the minimum hit width). The algorithm loops through all of the TDC ticks in the waveform and searches for a value that exceeds the primary threshold. Once found, the algorithm defines the region of the waveform around that TDC tick, starting from a TDC tick with the ADC value exceeding the secondary threshold and ending with a TDC tick with the ADC value falling below the secondary threshold as a potential hit. If the width of that region is greater than or equal to the minimum width specified, the optical hit is created. The parameters stored in the optical hit are the optical channel (the number of the SiPM that produced the waveform), pulse time (the time of the TDC tick with the maximum ADC value in the pulse), pulse width (the length of the pulse), pulse area (the total area of the pulse), pulse amplitude (the maximum height of the peak), and the number of photoelectrons (obtained by dividing the hit area by the single-PE peak area). In order to measure the time resolution of our PDS, we computed the time difference between the reconstructed optical hit times from the photon detectors and the trigger times delivered by the muon counters. The data sample used was all runs containing photon detector data externally triggered by the muon counters. To exclude the large noise present in the output, we reconstructed hits with a peak of about 4.5 – 7 –

Figure 7. Waveform from photon detector showing a strong optical hit

PE above the pedestal, corresponding to an ADC count of about 100. Looking at ∆t for just photon detector 3 in Figure 8, we observed a narrow peak of width < 100 ns, which is much smaller than the TPC drift time on the order of milliseconds. As a test of whether the photon detectors are seeing true light, we examined the attenuation loss

• f light in the 35t vs. distance. Using the 35t’s full data set of triggered events, we selected events

with muons that triggered pairs of counters on opposite sides of the 35t to find samples of muons passing through the 35t at different distances from the APAs. We expect that events with muons passing closer to the photon detectors will contain a greater number of observable photoelectrons than events with horizontal muons passing farther away from the APAs. We used seven pairs of muon counters, with each pair of counters facing each other directly

• pposite on the outer walls of the cryostat. Figure ?? shows the number of events present for each

pair of muon counters, total and requiring that at least one optical hit be found; Figure ?? shows the ratio of events with optical hits to total events. We require that there be no more than the two triggers from the muon counter pairs in each event and that the triggers fall within 10 time ticks of each other. The reconstructed optical hits are required to have amplitude above a threshold of 100 ADC counts and cannot be from one of the 10 excluded noisy optical channels. Figures ?? and ?? show that counter pairs closer to the APAs register a greater number of events. – 8 –

s) µ Time ( 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1 500 1000 1500 2000 2500 3000 3500 4000

Muon counter time - peak time of optical hits for PD3

peaktime_minus_trigtime_per_hit_zoomin_no1

Entries 9907 Mean 0.8797 RMS 0.01414 Underflow 219 Overflow 1419

Muon counter time - peak time of optical hits for PD3

Figure 8. Optical hit peak times minus muon counter trigger times for photon detector 3

For each event with triggers on a pair of muon counters, all reconstructed amplitudes are summed; if no optical hit is present, the summed amplitude is zero. We average over total events with triggers to adjust for the greater number of events found on the muon counter pairs closer to the APA. Figure 9 shows a downward trend in the average total amplitudes as we move from the closer muon counter pairs to the farther pairs, showing that fewer photoelectrons produced by muons farther from the APAs reach the photon detectors.

5 Conclusion Acknowledgments

This work was supported by grants to be filled in later. We would like to thank (insert list).

References

[1] R. Acciarri et al. (DUNE Collaboration), Long-Baseline Neutrino Facility (LBNF) and Deep Underground Neutrino Experiment (DUNE) Conceptual Design Report Volume 2: The Physics Program for DUNE at LBNF, arXiv:1512.06148. [2] R. Acciarri et al. (DUNE Collaboration), Long-Baseline Neutrino Facility (LBNF) and Deep Underground Neutrino Experiment (DUNE) Conceptual Design Report Volume 4: The DUNE Detectors at LBNF, arXiv:1601.02984 .

– 9 –

Center position of counter in x (cm)

50 100 150 200

Average optical hit amplitude per event

50 100 150

Average amplitude per event vs. counter distance Average amplitude per event vs. counter distance

Figure 9. Average Optical Hit Amplitude per Event vs. Counter Pair Positions [3] D. Whittington. (DUNE Collaboration), Photon Detection System Designs for the Deep Underground Neutrino Experiment, arXiv:1511.06345.

– 10 –