The ProtoDUNE Photon Detection System: technology validation and performance

DUNE is a long-baseline accelerator experiment currently in construction at Fermilab and SURF (South Dakota). The science objectives of DUNE include the search for CP violation in the leptonic sector and the identification of the neutrino mass hierarchy, along with the observation of supernova neutrino bursts and proton decay. The Far Detector consists of four modules located deep underground, three of those instrumented with Liquid Argon TPCs and equipped with the DUNE Photon Detection System (PDS). The PDS is based on a novel light trapping technology that greatly enhances the DUNE physics reach, improving vertex identification, energy resolution and providing the trigger for non-beam events. Following a first run of data taking (from 2018 to 2020), the PDS of the two prototypes of the Far Detector located at CERN Neutrino Platform, ProtoDUNE-HD and ProtoDUNE-VD, is currently being reinstalled in order to implement the final design of the first (Horizontal Drift) and second (Vertical Drift) module. This paper presents the latest results of the PDS from test facilities and the status of the installation in ProtoDUNE in view of its second run. The most important achievements of the Vertical Drift PDS are reported, with emphasis on the new SiPM configuration and cold electronics, the custom WaveLength Shifting bars, and the latest generation of the dichroic filter designed for ProtoDUNE-VD.


Introduction
The Deep Underground Neutrino Experiment (DUNE) is a long-baseline experiment under construction in the United States.The Far Detector (FD) will be located in South Dakota, in the Sanford Underground Research Facility (SURF), 1300 km away from the accelerators in Fermilab, Illinois, which will provide a high-intensity neutrino   and antineutrino   source.The FD will be divided into 4 modules, three of them instrumented with Liquid Argon Time Projection Chambers (LArTPCs).For the fourth one, the DUNE collaboration is currently evaluating proposals.Each LArTPC will be housed inside a cryostat containing 17.5 kton of LAr, 10 kton fiducial mass [1].
The scientific goals of DUNE include the neutrino mass ordering, measurements of the CP phase, and the search for its violation in the neutrino sector.Physics with natural   sources is also in the program, such as the detection of the   flux from a core-collapse supernova within our galaxy and the search for proton decay.
The technology for the first two modules of the Far Detector (FD1 and FD2) will be validated with a dedicated campaign at CERN, where two prototypes (ProtoDUNE) of the detector are currently under renovation.

The Far Detector technology
A charged particle in LAr produces two signals proportional to the energy deposit: drifted electrons, that allow for precise imaging, and scintillation photons in the VUV region of the EM spectrum ( = 128 nm).The Far Detector is hence equipped with two independent readout systems, the charge readout system in the TPC anode and the Photon Detection System (PDS).The VUV scintillation light of the LAr is abundant (about 25k photons/MeV in the DUNE LArTPC field at 500 V/cm).The light signal collected by the PDS, combined with the charge read by the TPC, improves the calorimetric resolution of the system, especially at low energy.The fast component of the scintillation light has a  = 7 ns, that provides the trigger for non-beam events, such as Supernova Neutrino Bursts and proton decay.Moreover, the spatial resolution provided by the PDS improves the vertex reconstruction by one order of magnitude w.r.t. the TPC (from ∼ 1 cm to ∼ 1 mm).
The LAr VUV scintillation light needs to be downshifted to the visible range of the electromagnetic spectrum to be detected by commercially available photosensors.This task is performed by the -1 -X-ARAPUCA, a reflective box equipped with an entrance window, two photon-downshifting stages, one dichroic filter, and one light guide coupled to SiPMs (figure 1).The first downshifting stage occurs when the incoming VUV photons are absorbed and re-emitted by a layer of paraterphenyl (pTP) deposited on the substrate of the dichroic filters, from  = 128 nm to  = 350 nm.The second downshift, from  = 350 nm to  = 430 nm, is performed by the WaveLength Shifting (WLS) bar located inside the reflective box that also acts as a light guide towards the SiPMs.The light has now the proper wavelength to be reflected by the dichroic filters, which have a cut-off at 400 nm, and to be detected by the SiPMs [2,3].The lower limit to Photon Detection Efficiency (PDE) of the PDS is set at ∼ 2%, and is imposed by the combined physics requirements for the detection of   from Supernova Neutrino Bursts and of the proton decay.

ProtoDUNE-HD, Far Detector 1
A prototype for FD1 was built and tested at the CERN Neutrino Platform in the first run of ProtoDUNE.It collected data from the SPS with electron and muon beams and from cosmic rays over two years, from 2018 to 2020 [5].The TPC was installed in a membrane cryostat with internal dimensions of 8.5 × 8.5 m 2 for the base and 7.9 m in height, and, as planned for the FD1, the electric field was horizontally oriented (figure 2).
Prototypes for the PDS bar-shaped photon detectors with 207.4 × 8.2 cm 2 optical area each-were embedded in the wire anode planes assemblies (APA), 10 for each of the six APA.Three different technologies were tested for the PDS and the 60 modules were subdivided as follows: 2 based on the ARAPUCA technology (a previous version of the X-ARAPUCA) [3], 29 dip-coated light guides [6,7], 29 double-shift light guides [8].For both the muon and electron beams, with momentum ranging from 2 to 7 GeV/c, the ARAPUCA efficiency was evaluated to be about 2%.The dip-coated light guides and the double-shift light guides efficiencies were, respectively, 0.1 and 0.2%.
-2 - After the first validation in this ProtoDUNE run, the X-ARAPUCA technology will be employed for both FD1 and FD2 of DUNE and tested in the ProtoDUNE PDS run 2. The new design of the X-ARAPUCA for the Far Detector 1 (FD1-XA) employs 48 ganged SiPMs that are read by a single channel and an active window of 46.2 × 10 cm 2 .The SiPMs are located on the long edges of the 48 × 9.3 × 0.38 cm 3 WLS bar, 24 for each side, mounted on eight PCBs hosting six 0.6 × 0.6 cm 2 SiPMs each.Six dichroic filters with dimensions 7.7 × 10 cm 2 close the optical window, whose dimensions for the active area are given by the combination of the filters and the WLS bar ones minus the dead areas introduced by the ribs that support the filters.All of the X-ARAPUCA components are held by a G10 (high-pressure fiberglass laminate) mechanical assembly (figure 3), internally lined with Vikuti ® , a highly reflective material.For the FD1-XA, the ratio between the WLS area and SiPM is 26.A single FD1 PDS module holds 4 FD1-XA, read by four independent channels; the ProtoDUNE FD1 has a total of 4 APA, 10 PDS modules each, for a total of 160 FD1-XA.The FD1-XA in ProtoDUNE have been equipped with WLS bars by either Glass To Power "G2P" (Italy) [10,11], Poly(methyl methacrylate) (PMMA) based, or Eljen "EJ-286" (U.S.A.) [12], polyvinyltoluene (PVT) based, and either HAMAMATSU HPK DUNE-75μm-High Quenching Resistance SiPMs [13] or -3 -FBK Triple Trench SiPMs [14], for a total of 4 different configurations.All configurations employ OPTO-Campinas (Brazil) dichroic filters, deposited on B270 from SCHOTT ® .For both SiPM types dedicated tests performed in Milano-Bicocca achieved a signal to noise ratio S/N>4 for 48 SiPMs ganged together, thus satisfying the DUNE requirements.Before installation, the PDE for the four FD1-XA configurations was assessed in Milano-Bicocca and CIEMAT.The most performing combination turned out to be the one with the Glass to Power WLS bar and the Hamamatsu SiPMs, with efficiency for the overall cell of about 2-2.5%.The other combinations reported an efficiency of 1.5-2% (table 1).Efforts to enhance the PDE are currently ongoing; increasing the light sealing of the WLS bar and breaking the optical path of photons undergoing total internal reflection inside of it have reported promising results.Prior to being installed in ProtoDUNE FD1, all the 160 FD1-XA were tested in LN 2 , without dichroic filters, to assess their performance in terms of S/N ratio and Dark Count Rate (DCR) in Milano-Bicocca, CIEMAT, and Colorado State University.Among the 160 FD1-XA, 47 were tested in Milano-Bicocca in a shallow laboratory.The analysis on this subset reported a slightly better S/N for the FD1-XA equipped with Hamamatsu SiPMs, regardless of the WLS bar (figure 4 and table 2), and a considerably better DCR (almost a factor 4) for the FD1-XA equipped with G2P WLS bar, regardless of the SiPMs (figure 5 and table 3).The DCR results were expected, as the PVT is a scintillating material.These overall trends were confirmed by CIEMAT.The second ProtoDUNE installed at the CERN Neutrino Platform, which will test the expected performance of the DUNE FD2, exploits the Vertical Drift technology [15].The main difference with the FD1 TPC is that in the FD2 the anodes are conductive planes instead of wires, so the PDS cannot be placed in the APA as in FD1.In FD2, the PDS is positioned outside the field cage ("membrane modules") and on the cathode.The light uniformity is improved via Xenon doping of the LAr [16].
The placement of the PDS modules on the cathode poses a challenge, as its SiPMs need to be biased and read in a 300 kV electric field.The solutions are the Power over Fiber (PoF) and the Signal over Fiber technologies [15].The PoF is based on a commercial system which includes a GaAs laser and an Optical Power Converter (OPC).The laser light travels through an optical fiber and is converted to volts (6 V) by the OPC at cold, subsequently powering the cold electronics.A DC-DC converter then brings the 6 V to the adequate SiPM bias, which differs for Hamamatsu and FBK SiPMs.With the Signal over Fiber, the analog signal is transmitted using IR laser light.
-5 - The PDS for the FD2 uses the same X-ARAPUCA technology as in FD1 but with different dimensions and SiPM coverage.The WLS plate, produced by Glass to Power in PMMA [10], is 60.7 × 60.7 cm 2 with a baseline thickness of 0.38 cm and its sides are covered by 160 SiPM (40 per side) either by Hamamatsu or FBK, 0.6 × 0.6 cm 2 .The SiPMs are passively ganged in groups of 5, and 8 of these are actively ganged at each of the two inputs of the front-end electronics.Each of the two independent front-end electronics circuits processes and connects 80 SiPMs signal for a total of two output channels per FD2-XA module.
The ratio between the WLS area and SiPM is 128, to be compared with the 26 of the FD1-XA.These differences require tailoring the mechanics and some parameters of the FD2-XA modules to maximize the light collection.A measurement of PMMA bars at room temperature and after being cooled down to LAr temperature showed that this material shrinks up to 1%, while following the same procedure with G10 pieces didn't report an appreciable shrinking.So, given the WLS plate dimensions, a gap of 3 mm between the WLS plate and the SiPMs is expected on each of the 4 sides.To minimize this gap, the SiPMs are mounted on flex circuits and backed by a spring-loaded mount; the springs are loaded at room temperature and they relax at LAr temperature, keeping the SiPMs in contact with the WLS plate.Some modules deployed in ProtoDUNE have dimple cuts (flat or cylindrical) machined at the edges of the WLS, of positions and widths that allow the SiPMs to fit inside.The cylindrical dimples' goal is to increase the light extraction from the WLS bar and onto the SiPMs, while the flat ones were used to test the gluing of the SiPMs to the WLS plate.
The WLS plate dimensions also require optimizing the Optical Path and fine-tuning it with the WLS absorption, acting on the dye concentration and plate thickness.For PMMA-based WLS plates, the critical angle   for light trapping at the surfaces is 56 • in LAr.For  >   the light is trapped and guided by total internal reflection to the SiPMs.The trapped photons hence undergo multiple reflections and the optical path inside the large-size WLS plate may reach a couple of meters, increasing the probability of re-absorption.For this reason, it's important to fine-tune the WLS absorption versus the optical path, acting on the chromophore (BBT, a thiophene based compound) concentration and the plate thickness.The latter is re-tuned to keep a probability > 90% to absorb the incoming 350 nm photons emitted by the pTP, the primary WLS.The simulations indicate that an optimum is reached with a 16-24 mg/kg dye concentration in the PMMA and a plate thickness of 0.55-0.6cm.The WLS plate thickness is constrained by the SiPM dimension, so it is limited to 0.6 cm.Six FD2-XA in ProtoDUNE have a WLS plate that is 0.55 cm thick, three of them with 40 mg/kg dye concentration and three with 25 mg/kg.Their efficiency will be compared with the other WLS plates installed in ProtoDUNE-FD2, which have a chromophore concentration of 80 mg/kg, the same as ProtoDUNE-FD1 bars.
For the FD2-XA different solutions for the dichroic filters were also investigated in order to maximize the active area and the performances in LAr.In ProtoDUNE-FD2 the PDS mounts either 10 × 20 cm 2 or 15 × 15 cm 2 dichroic filter.These larger filters were produced by ZAOT (Italy), which employs Borofloat 33 Optical Glass as substrate, and by PhotonExport (Spain), which uses Fused Silica.ZAOT filters, in particular, were optimized to operate in LAr for angles of incidence close to 45 • .With respect to the baseline filters by OPTO-Campinas, which were designed for the same angles on incidence but in air, the ZAOT filters have higher transmittance in the pTP emission range, higher reflectivity in the light guide WLS chromophore emission range, but narrower reflectivity range.Measurements for the filter performances were executed in water, whose refraction index is n = 1.33 and acts as a viable proxy for LAr (n = 1.24) (figure 6).
-6 - The impact on the PDE of the ZAOT filters technology was tested at Milano-Bicocca using a FD1-XA equipped with three ZAOT filters and three OPTO-Campinas ones; the PDE on side covered by the ZAOT filters was measured being +9% w.r.t. the side covered by the OPTO-Campinas filters; this increase was also predicted by a simulation [17] made with GEANT4 [18].
The PDE of the "Membrane" FD2-XA for various configurations is currently under test at INFN-Napoli (Italy) and CIEMAT (Spain).Preliminary results from INFN-Napoli reported a PDE of ∼ 2% for a FD2-XA equipped with 15 × 15 cm 2 ZAOT dichroic filters, SiPMs from FBK and a Glass to Power WLS plate with dimples, baseline (80 mg/kg) chromophore concentration [19].Preliminary measurements with an FD2 amplifier adopted for the modules located on the walls of the cryostat ("Membrane" modules) in Milano-Bicocca also reported achieving a S/N of about 8 with 80 SiPMs.Dedicated PDE and ganging measurements will also be performed on PoF modules.
In the ProtoDUNE-FD2, there are currently installed 8 "Membrane" FD2-XA and 8 cathode FD2-XA (powered with PoF); other six modules with optimized chromophore concentration will be deployed by the end of 2023.

Conclusions
The installation of ProtoDUNE-HD, which is a prototype of DUNE Far Detector 1, was completed in September 2022, with the modules of the Photon Detection System equipped with four different combinations of two WaveLength Shifting bars (manufactured by Glass to Power and Eljen) and two SiPMs (by Hamamatsu and FBK).The cooldown of the cryostat is planned for Spring 2024.The tests for the Photon Detection Efficiency of these four combinations were performed in two sites: Milano-Bicocca and CIEMAT, reporting results ranging from 1.5 to 2.5% depending on the adopted SiPM and WLS components.The best results were achieved with the Hamamatsu-Glass to Power combination and with the SiPMs biased at +3.0 V overvoltage, namely at SiPM PDE of 50%.The installation of ProtoDUNE-VD, a prototype of DUNE Far Detector 2, is currently in progress, along with tests for the PDE of its PDS in Napoli and CIEMAT.These tests reported a preliminary efficiency of ∼ 2% and techniques to further increase the PDE of the PDS of both Far Detector 1 and Far Detector 2 are currently under development.

Figure 4 .
Figure 4. S/N distributions of the 47 FD1-XA tested in Milano-Bicocca, grouped by type.

Table 1 .
Photon Detection Efficiencies for the four FD1-XA configurations installed in ProtoDUNE.

Table 2 .
Average S/N for the four FD1-XA configurations tested in Milano-Bicocca.FD1-XA type S/N mean S/N st.dev.