X-Arapuca long term test

The photon detection system of the DUNE experiment is based on the X-ARAPUCA light trap. The basic elements of the X-ARAPUCA are the dichroic filters coated with wavelength shifter (para-Therphenyl), a waveshifting plate and an array of SiPMs which detects the trapped photons. A small scale prototype of the X-ARAPUCA has been installed in liquid argon in a dedicated facility at INFN-Napoli and exposed to alpha particles from a source. In order to test the stability of the overall device response the X-ARAPUCA was kept for 10 days in continuously purified liquid argon. The performed tests allowed for a preliminary estimation of the X-ARAPUCA absolute photon detection efficiency.


Introduction
Next generation neutrino experiments will investigate new physics beyond the Standard Model, addressing the measurement of the CP violating phase in the leptonic sector.A significant contribution is expected to the completion of the understanding of the standard neutrino oscillation picture by measuring the mixing parameters and the neutrino mass hierarchy.The Deep Underground Neutrino Experiment (DUNE) [1] on the Fermilab Long-Baseline Neutrino Facility (LBNF) represents one of the most relevant experiments in this field.LBNF provides an high intensity, broad band neutrino beam, peaked at 2.5 GeV.The neutrino beam flux, monitored by a near detector located at Fermilab, travels through the Earth crust for 1300 km and is finally detected by a far detector installed at the Sanford Underground Research Facility in South Dakota.The far detector is constituted by at least two 17.5 kton liquid argon (LAr) Time Projection Chambers (TPC).The huge target mass will enable a rich scientific program including among others searches for proton decay and detection of the neutrino flux from a core-collapse supernova within our galaxy.
LAr is known to be an excellent scintillator emitting 41 photons per keV of deposited energy by minimum ionizing particles.Scintillation photons are emitted through the de-excitation of Argon dimers (Ar * 2 ) singlet (S) and triplet (T) states with characteristic times of 6÷10 ns and about 1400÷1600 ns, respectively [2].The scintillation photons are emitted in Vacuum Ultra Violet (VUV) with a wavelength centered in a narrow band of ∼10 nm around 128 nm.The primary goal of the photon detection system of the DUNE far detector must meet several requirements: convert VUV light to visible through the use of wavelength shifting compounds in order to make it detectable by standard (cryogenic) photo-sensitive devices; provide large coverage at reasonable cost due to the huge detector dimensions; reach the required detection efficiency (>1%) to meet the supernovae DUNE scientific program.The device proposed for the light detection system of the DUNE far detector is the X-ARAPUCA [5].
It is constituted by a light collector coupled to an array of silicon photo-multipliers (SiPMs) which detect the collected photons.In this work the detection principle of a sample of two X-ARAPUCA devices has been probed in a continuous data-taking lasting about 10 days with the use of a LAr condenser and of a purification system.

The X-ARAPUCA
The X-ARAPUCA (XA) light trap is an evolution of the first ARAPUCA design [6].It consists of a box cavity with highly reflective internal walls.The entrance window of the box is made by a short-pass dichroic filter which has the properties of being highly transparent to photons with wavelength below a given cut-off (400 nm), while being highly reflective to photons with wavelength above the same cut-off.The dichroic filter is coated on the external side of the entrance window with para-Terphenyl (pTP), a wavelength shifter converting photons from 128 nm to 350 nm [7].Because its wavelength is below the dichroic cut-off, such photons can cross the dichroic filter.Inside the box immersed in LAr a second wavelength shifting step is performed by a WLS slab (EJ-286PS model manufactured by Eljen Technology) that shifts 350 nm photons to 430 nm.The light re-emitted from the WLS slab can be trapped by total internal reflections or escape, to be then reflected back by the dichroic filter.In both cases the photons eventually reach the SiPM photosensor located on the edge of the WLS slab.The light trap sequence is summarized in fig. 1.The pTP and EJ-286PS emission spectra are shown in fig. 1, where the separation with respect to the dichroic cut-off is clearly visible.The XA concept has been realised in several shapes and sizes.The XA model used in this work features two dichroic windows sizing 200×75 mm 2 overall, the same format employed in SBND experiment [8] (fig.2).The WLS slab is coupled to four photosensor boards, each containing four SiPMs (Hamamatsu model S13360-6050VE) ganged in parallel.A six windows XA is the basic photon detection unit for the DUNE far detector first module [9].

Experimental setup and cryogenic facility
The two windows XA under test was installed in a 13 l cryostat (fig.3) connected to an argon gas liquefaction and purification system.The experimental setup allows for testing the photosensor performance in detecting events generated in pure liquid argon by a radioactive source.The device is hanging from the top flange and an 241 Am alpha source is mounted frontally in a peek holder fixed to a motion feedthrough.In this way the source can be translated between the two centers of the XA windows.The distance between the source and the dichroic surface is 4 cm.An optical fiber for single photoelectron calibration is inserted in the cryostat through an optical feedthrough externally connected to an Hamamatsu laser head PLP C8898.The cryostat is filled by liquefying gaseous Ar 6.0 (1 ppm impurities in total) from pressurised bottles.The argon liquefaction process is performed through a condenser made by two "brazed plate" and one "tube and shell" heat exchangers.The heat exchange is performed at the expenses of liquid nitrogen.After LAr filling the evaporated gaseous argon is recirculated by a gas pump through a SAES MonoTorr model PS4-MT50-R rare gas hot purifier [10].The cryostat is equipped with six PT100 level meter sensors and a pressure transducer.For this test, the four SiPMs boards from the XA were biased and read-out through the APSAIA board [11] developed within the SBND experiment.The board embodies 8 channels with input connectors.Both power supplies and amplifiers have remote control via RS232 serial port.The four XA output channels read-out by APSAIA are then sent to CAEN V1725B digitizers (250 MS/s, 14 bit).

Measurement results in Liquid Argon
With SiPM bias set at 4V over-voltage, we studied the single photon response of the different channels in laser-triggered runs to calibrate the average charge of the first photo-electron peak.This calibration was then used to study the XA response to events generated in argon by the 241 Am alpha source: a rate of about 100 Hz was found in self-trigger mode data taking.Data presented in this work refer to the case of alpha source positioned at the center of one of the two XA windows (the one located deeper in LAr).Fig. 4 shows an example of normalized and fitted waveform from scintillation light due to alpha particles.The slow decay component is in good agreement with expected value of 1.4 s.Fig. 5 shows the reconstructed charge, in number of photo-electrons (PE), for all four channels.For each channel, the  peak is fitted with a gaussian distribution.

Light yield and stability of the system
The system was kept in LAr with recirculation and purification system active for 10 consecutive days.Several self-trigger runs were acquired daily to monitor the stability of the light yield, measured as the total average number of PEs detected by the four channels.After a slight increase in the first 20 hours of data taking, due to impurities removal, the light yield was stable during the full period within ±1% (fig.6).The final light yield was found N < > =1595±10 on average.

X-ARAPUCA efficiency measurement
A preliminary efficiency of the XA photodetector can be known if the number of photons impinging on the dichroic window surface are estimated.The number of photons produced in LAr by excitation from the alpha particles is evaluated assuming a photon yield of Y  = 51000 ± 1000 photons/MeV and a quenching factor   = 0.71 ± 0.02 for alpha particles [2][3][4].With an alpha quasi-monochromatic energy of E  =5.48 MeV, the total amount of emitted photons is given by: The geometrical acceptance for the isotropically emitted VUV photons was evaluated with a Geant4 simulation in which the XA complete geometry was implemented together with the 241 Am source size and position.We found a geometrical efficiency   =21.1%, leading to a number of VUV photons impinging the XA surface given by: N    = N  ×   = 41300 ± 1400 The number of detected PEs obtained by summing all four channels must be corrected for the SiPMs secondary pulses induced by cross-talk and afterpulses.To this purpose we used the value 1.55 ± 0.05, expressed in number of avalanches generated per detected photon, reported in [12] for our SiPMs.The final measured efficiency is given by where the error is dominated by systematic uncertainties on the single photo-electron calibration.The obtained value obtained is in good agreement with other performed measurements [13].

Figure 2 .
Figure 2. X-ARAPUCA device used in the test.

Figure 4 .
Figure 4. Average waveform from alpha source scintillation

Figure 5 .
Figure 5. Alpha source spectra in PE for all four X-ARAPUCA channels

Figure 6 .
Figure 6.Total average PE from four X-ARAPUCA channels Vs time for 10 days