Test Beam Study of SiPM-on-Tile Configurations
A. Belloni, Y.M. Chen, A. Dyshkant, T. K. Edberg, S. Eno, V. Zutshi, J. Freeman, M. Krohn, Y. Lai, D. Lincoln, S. Los, J. Mans, G. Reichenbach, L. Uplegger, S. A. Uzunyan
VVersion 1.1, February 18, 2021
Prepared for submission to JINST
Test Beam Study of SiPM-on-Tile Configurations
A. Belloni 𝑏 Y.M. Chen 𝑏 A. Dyshkant 𝑎 T. K. Edberg 𝑏 S. Eno 𝑏 J. Freeman 𝑐 M. Krohn 𝑑, Y. Lai 𝑏 D. Lincoln 𝑐 S. Los 𝑐 J. Mans 𝑑 G. Reichenbach 𝑑 L. Uplegger 𝑐 S. A. Uzunyan 𝑎 V. Zutshi 𝑎 𝑎 Northern Illinois University 𝑏 University of Maryland 𝑐 Fermilab National Accelerator Laboratory 𝑑 University of Minnesota
Abstract: Light yield and spatial uniformity for a large variety of configurations of scintillator tileswas studied. The light from each scintillator was collected by a Silicon Photomultiplier (SiPM)directly viewing the produced scintillation light (SiPM-on-tile technique). The varied parametersincluded tile transverse size, tile thickness, tile wrapping material, scintillator composition, andSiPM model. These studies were performed using 120 GeV protons at the Fermilab Test BeamFacility. External tracking allowed the position of each proton penetrating a tile to be measured. Theresults were compared to a GEANT4 simulation of each configuration of scinitillator, wrapping,and SiPM.Keywords: Calorimeters | scintillators, scintillation and light emission processes | Detectormodelling and simulations I | Photon detectors for UV, visible and IR photons (solid-state)ArXiv ePrint: 1234.56789 Corresponding author. a r X i v : . [ phy s i c s . i n s - d e t ] F e b ontents The High Luminosity phase of the Large Hadron Collider (HL-LHC) [1] is scheduled to beginat CERN in 2027, with a designed instantaneous luminosity of 5 × cm − s − . In order tooperate in this environment, a new high granularity calorimeter (HGCAL) [2] will be installed inthe endcap regions of the CMS detector. In regions of the calorimeter where the fluence is highest,the design uses silicon sensors as the active material. In lower fluence regions, the design usesplastic scintillator tiles, with the scintillation light readout by silicon photomultipliers (SiPMs).To achieve the required performance, uniform light collection across the face of individualscintillator tiles is important. The SiPM-on-tile technology, where the SiPM is located in a dimple– 1 –achined into the tile surface, has been studied by the Calorimeter for Linear Collider Experimentcollaboration (CALICE) [3–5]. These studies reported that for 97% of a 3 × tile area theresponse is within 10% of the average response. Similar studies are needed for the tile designsthat will be used in the HGCAL. Some adjustment of the HGCAL SiPM-on-tile design may berequired to cover the full range of tiles in the calorimeter. CMS HGCAL will use scintillator tilesof approximately square shape varying in size from roughly 2 . × . . × . .We describe the apparatus and analysis used to study the responses of different scintillator tilegeometries and materials using the Fermilab Test Beam Facility (FTBF) [6] at the Fermi NationalAccelerator Laboratory. In this report, we present measurements for various geometries of EljenTechnology EJ-200 [7] and SCSN-81 [8] tiles and their comparison with GEANT4 [9] basedsimulations. The Fermilab Test Beam Facility [6] provides a primary beam containing 120 GeV protons bunchedat 53 MHz. The beam is delivered as a slow spill with a 4.2 second duration once per minute andintensity of approximately 5 × protons per spill. The beam spot size is roughly Gaussian, 4 cmin diameter with standard deviation in x and y of about 1 . Proton beam direction
Scintillator trigger counter 1 and 2Silicon tracking station 2 X Y
Silicon tracking station 1 X Y
Dark box
Scintillator tileSIPM
Testbeam Setup
Figure 1 . Test beam layout, not to scale. A dark box on a computer-controlled x-y stage contained theSiPM-on-tile sample. Upstream and downstream doublets of silicon strip layers provided particle tracking.Two downstream scintillation counters provided the trigger.
The test beam configuration is shown in Fig 1. The scintillator sample and the readout SiPMwere placed in a dark box mounted on a computer-controlled x-y stage that provided the ability– 2 –o move the tile transversely to the beam. The readout SiPM was connected to a 2 GHz invertingamplifier [10]. Bias voltage for the SiPM was provided by a Keithley 6410 source meter. Theanalog signal from the SiPM was sent to a DRS4 1 – 5 GS/s waveform digitizer [11]. The firmwareof the DRS4 was modified to provide an output "BUSY" signal [12].Silicon strip tracking stations were placed on the upstream and downstream sides of the darkbox containing the scintillator sample to measure the beam particle position. Two orthogonallayers of strips (tracker planes) in each tracking station recorded three-bit amplitude signals for eachtriggered event. The strips were 60 𝜇 m × . . × .
84 cm .We used collections of strips with amplitudes above a threshold to find clusters of adjoiningstrips. Amplitude-weighted coordinates of strips were assigned to these clusters to build particletracks. Assuming parallel beam propagation, we treated (X, Y) pairs of clusters in the upstreamand downstream stations with the minimum transverse distance between them as the most probabletrack trace. We used the (X, Y) position of the most probable track in the upstream tracker stationas the beam particle’s measured position (in events with no hits in the downstream tracker station,the first (X, Y) pair of clusters in the upstream station was used).The coincidence of two scintillator counters provided the trigger for the system. The coin-cidence was fed into a NIM+ module [13] which is based on the CAPTAN+ board designed atFermilab and a Kintex7 FPGA. This module was configured to accept triggers that occurred in theabsence of BUSY signals from the DRS4 and the silicon readout. The FTBF Data Acquisition System [14] is called OTSDAQ, "Off The Shelf Data Acquisition." Itwas designed at Fermilab and is based on the XDAQ libraries developed at CERN for the CMSexperiment. OTSDAQ was used to configure the DRS4 board and the silicon strip tracker stations,to provide triggers for both systems, and to perform online data quality monitoring and run control.The digitized waveforms from the DRS4 and the tracker information data streams were collected bythe independent DAQ computers. Offline, these streams were combined into events using matchingtriggers. The operating acquisition rate was about 1K events per beam spill.
This study investigates prototypes based on the SiPM-on-Tile concept [4]. Scintillators of varioussizes were machined with a "dimple" as shown in Fig 2. The dimple provided a physical space forthe SiPM and improved uniformity of response. In general, the SiPM will detect more light froma particle passing close to it. To reduce this geometry-induced hot spot, a dimple was machinedto reduce the amount of scintillator and hence the amount of light generated near the SiPM. TheSiPMs were located at the center of the dimple, with the active face located 0 .
55 mm into the dimplerelative to the tile face. A reflective wrapper surrounded the tile to reflect escaping photons backinto the tile, increasing the light yield. The wrapper covered 100% of the tile area except for a smallhole at the dimple to accommodate the SiPM. – 3 – .1 Description of tile geometries
Tiles were prepared from two scintillator materials, 3 . . . . . . × . . × . . Two SiPMs from Hamamatsu Photonics were tested: S13360-1350PE [15] and S14160-1315PS [16].These both have an active area 1 . × . . The S13360-1350 has 50 micron pixels, while theS14160-13115 has 15 micron pixels. The gain of the S13360 is consequently about 10 times larger.To ensure the pulse height of the two SiPMs were approximately the same, a second 10 × Figure 2 . Left: Cross section of "SiPM-on-Tile" SCSN-81 Design. Right: Cross section of EJ-200 tile.The dimples are centered on the tile face.
The expected optical responses of the various tiles were simulated using the GEANT4 [9] toolkit.The results are compared with data in Section 7.Fig. 3 shows the GEANT4 rendering of the simulation geometry. It is separated into threeparts: the tile with its dimple, the reflective wrapping, and the SiPM attached to a backplate. Thetile is completely wrapped in a reflective coating except for a hole centered on the dimple. Formost simulations, the SiPM back was flush with the tile surface. A circular backplate outside of thewrapping was attached to the back of the surface-mount SiPM and was simulated as either printedcircuit board white silk screen (WSS) or black tape with reflectivities at a wavelength of 425 nm of0.68 and 0.05, respectively.The simulated scintillator material was set to be the base material for SCSN-81 (polystyrene)and EJ-200 (polyvinyltolulene). Its optical properties were taken from the datasheet of the com-– 4 – igure 3 . GEANT4-rendering of simulated tile geometry as seen by the beam (left) and side view (right).The tile is a rectangle composed of plastic scintillator with a spherical dimple. The square at the center ofthe dimple is the rendering of the simulated SiPM. The sizes of this geometry follows Fig. 2 (right). mercial product EJ-200 [7] except for the absorption length. The effective light attenuation lengthat 425 nm was set to 3 . . ® and 3M Vikuiti ™ Enhanced Specular Reflectorfilm (ESR) [17] with effective reflectivities set to 0.79 and 0.985 at 425 nm, respectively. A thin airgap between the tile and the wrapping was included.A 120 GeV proton beam was generated, oriented perpendicular to the scintillator face. Thescintillating efficiency was 10 photons / keV for EJ-200 [7] and 8 . / keV for SCSN-81 [8].A photon entering into the SiPM sensitive area was counted if it passed an acceptance-rejection cuton the SiPM photon detection efficiency curve. The average number of detected photons for eachgeometry was used as a metric for the simulated light yield.For each geometry, the simulated proton’s impact positions were uniformly distributed acrossthe tile face. We studied the effect on the light yield of a uniform beam and a Gaussian beam, witha width similar to the test beam, and found the difference to be less than 2%.When simulations were compared with the data in Section 7, an additional normalizationconstant was used to scale down the predicted light yield from simulation. It was 1 .
15 and 1 .
01 forEJ-200 and SCSN-81, respectively. These quantities took into account the difference between the– 5 –ost probable value (MPV) of light yield from data and the mean light yield from simulation andthe estimations of the effective reflectivities that were used for the wrappings.
The DRS4 digitizes the SiPM analog voltage output using a 14 bit ADC with a configurablesampling frequency. For this experiment, we operated the DRS4 at 1 GS/s. For each trigger theSiPM waveform (as shown for example in Figure 4a) was recorded with 1024 samples. The DRS4trigger delay was set to provide the pre-signal region of the waveform for the pedestal evaluation.We used integrals of waveform pulses to measure the light output of the scintillator tiles. Thewaveform pulse was defined as a contiguous set of samples near a local maximum with an ADCamplitude above a threshold of 10 mV. Pulse voltage integrals 𝑉 𝐼 were calculated as the sums ofwaveform amplitudes in a region of 60 samples, starting with the sample with a signal above 0.25 ofthose at a pulse maximum. To obtain the conversion factor between the integrated voltage and thenumber of the photoelectrons (PEs) produced in the SiPM, the following procedure was applied: • we collected 10-20K waveforms for a scintillator tile with a known low light output perincident proton in order to see PE peaks; • for each waveform, we calculated the signal integrated voltage 𝑉 𝐼 under the pulse withmaximum amplitude; • the pedestal integrated voltage was calculated using a sample region of the signal pulse widthin the pre-signal part of the waveform; • after subtraction of the pedestal, signals were histogrammed to obtain the PE spectrum; • a fit (the sum of Gaussian curves) of the first six peaks of the spectra was used to calculatethe conversion factor as the mean distance between peaks in the fit region.Figure 4a shows a DRS4 waveform collected from a S14160-1315 SiPM, when a 120 GeVproton passes through a Tyvek ® -wrapped SCSN-81 scintillator tile. The difference of integrals inregions [562, 622] (signal) and [482, 542] (pedestal) is taken for the signal histogram in this event.Figure 4b shows a fitted spectrum with individual PE peaks corresponding to a calibrationfactor of 0 . ± . 𝑃𝐸 / 𝑉 𝐼 . We selected events which 1) have a matching trigger number for both DRS4 and tracker data, and2) passed quality selection criteria. We required: • a clean waveform - the falling edge of the signal pulse was required to reach the level of0.25 of the pulse’s maximum and the pre-signal region should be wider than the number ofsamples used for signal integration; – 6 – ample Number0 200 400 600 800 1000 S i P M S i gna l ( V ) - - - - - Integrated voltage0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 E v en t s / b i n (a) (b) Figure 4 . (a) A typical SiPM signal waveform (1 ns per sample) and (b) a photoelectron spectrum withmultiple Gaussians fit over the peaks. The signal and pedestal regions of the waveform are integrated inregions [562, 622] (signal, green) and [482, 542] (pedestal, red). • noise suppression - SiPM signal pulses were required to have the peak amplitude above tenmV; • clean tracker data - the upstream tracker station was required to have silicon strips fired inboth x and y planes but have not more than one two-strip cluster in each.These cuts were sufficient to ensure that a single particle passed through the tile and thatthe particle can be treated as a minimum-ionizing particle (MIP). MIP yield distributions for the3 × SCSN-81 tile wrapped in ESR before and after quality selections are shown in Figure 5a.
To determine the light yield of a scintillator tile, the MIP yield distribution is fit with the sumof a Gaussian and a Landau function, where the lower light yield region is modelled to followGaussian statistics, but the high light yield tail is modelled with a Landau. We report the lightyield as the most probable value (MPV) of the fit function. We observed that the MPV statisticaluncertainty in a typical single measurement (5-10K clean events) was smaller than the ±
3% opticalcoupling systematic uncertainty described below in Sec 8.1. Figure 5b shows the fit of the MIPyield distribution for clean events in the 3 × SCSN-81 tile test.
An attractive feature of the experimental setup is the ability for studies of the spatial uniformityacross the face of a tile of the collected signals. Fig. 6 shows the average MIP yield distribution forthe 3 × × . EJ-200 tile wrapped in ESR (reference tile) as a function of the proton impactposition. The observed pattern in the MIP yield, including a bright spot slightly below the dimple– 7 – umber of Photoelectrons0 10 20 30 40 50 60 E v en t s / b i n
10 Number of Photoelectrons20 40 60 80 100 120 140 E v en t s / b i n FitMean/Sigma = 23.89+-0.25/6.86+-0.16FitMPV/FWHM = 23.91/17.08 (a) (b)
Figure 5 . (a) MIP yield distributions for the 3 × SCSN-81 tile wrapped in ESR before (black) andafter quality selections (red), and (b) the fit of the PE yield distribution after quality selections. region, is consistent with a slightly misaligned SiPM as reported in [19]. Fig. 7 shows profiles ofMIP yield distributions in data and simulation normalized to the maximum value of the entire twodimensional distribution for this EJ-200 tile. The profiles are along 2 mm horizontal and verticalbands that either pass through the tile center or are offset by 8 mm from the tile center in x or y. Weobserved uniform (within statistical uncertainty) MIP yield in bands located 8 mm away from thedimple center and 7 mm away from the side edges of the reference tile.To validate the performance of our experimental equipment, we prepared a SCSN-81 tile with alarge dimple machined into it. This configuration was expected to exaggerate the effect of the dimpleon the uniformity of the tile response. Figure 8 shows the beam intensity profile and the averageMIP yield distribution as a function of the proton position in the upstream tracker plane for thistile. The depression of the MIP yield in the tile center is clearly evident and this corresponds to thedimple in the scintillator material. Fig. 9 shows profiles of MIP yield distributions, normalized tothe maximum value of the intensity profile for this SCSN-81 tile. Points with error bars correspondto profiles of SiPM signals measured in vertical or horizontal bands of 2 mm width, spanning thetile center. Dashed lines show profiles in bands of 2 mm width that are 8 mm offset from the tilecenter in x or y.We show in Fig. 10 the average MIP yield distribution as a function of proton position for a5 . × . EJ-200 tile. For the same tile, we show the light yield for annular regions centeredon the dimple in Fig. 11. We observe an increase of the light yield directly in the region outside thedimple.To quantify a tile’s uniformity, we divided a tile into 2 × bins, determined the averagelight yield in each bin, and calculated the RMS / mean for all average light yield values in a giventile. The average light yield distributions are shown in Fig. 12 for the 3 × EJ-200 tile, the5 . × . EJ-200 tile, and the SCSN-81 tile with large dimple. To establish that the RMS / mean– 8 – roton Position (mm)0 5 10 15 20 25 30 35 40 P r o t on P o s i t i on ( mm ) L i gh t Y i e l d ( PE ) Figure 6 . The average MIP yield distribution of the EJ-200 reference tile as a function of proton positionin the upstream tracker. distribution provides useful insights on the uniformity of tiles, we performed a simplified simulationof a 3 × tile. The simulated light response across the tile was perfectly uniform, following aPoisson distribution with an average light yield of 30 PE. We simulated 10 events in each 2 × bin and compared the result to data by quantifying the non-uniformity as RMS / mean tile RMS / mean uniform −
1. Thenon-uniformity of the different sized EJ-200 tiles and the special SCSN-81 tile is listed in Table 1.As expected, the EJ-200 tiles show considerably lower non-uniformity than the special SCSN-81tile.
Table 1 . The non-uniformity,
RMS / mean tile RMS / mean uniform −
1, of the average light yield for the different sized EJ-200tiles and the SCSN-81 tile with the large dimple machined into it.
Tile Non-UniformityEJ-200 2 . × . . ± . . × . . ± . . × . . ± . . × . . ± . . × . . ± .
11– 9 – (mm) proton
X0 5 10 15 20 25 30 35 40 m a x L Y / L Y Data Center Y-profileSimulation Center Y-profile (mm) proton
Y0 5 10 15 20 25 30 35 40 m a x L Y / L Y Data Center X-profileSimulation Center X-profile (a) (b) (mm) proton
X0 5 10 15 20 25 30 35 40 m a x L Y / L Y Data Side Y-profileSimulation Side Y-profile (mm) proton
Y0 5 10 15 20 25 30 35 40 m a x L Y / L Y Data Top X-profileSimulation Top X-profile (c) (d)
Figure 7 . Normalized MIP yield distribution profiles in data and simulation for the EJ-200 reference tilealong a) 2 mm y-bands at the tile center, b) 2 mm x-bands at the tile center, c) 2 mm y-bands offset by 8 mmin the x direction from the tile center, and d) 2 mm x-bands offset by 8 mm in the y direction from the tilecenter.
The light yield was measured for different sized square tiles composed of 3 mm thick, EJ-200scintillator wrapped in ESR. Tile sizes ranged from 2 . × . . × . . The results arecompared to simulation and shown in Fig 13. A 𝜒 fit to data is performed with the function 𝑝 × ( Tile Area / ) 𝑝 , where 𝑝 and 𝑝 are parameters of the fit and fitted values of 𝑝 = . ± .
48 PE and 𝑝 = − . ± .
02 are obtained. The light yield is roughly inverselyproportional to the square root of the tile area.– 10 – roton Position (mm)0 5 10 15 20 25 30 35 40 P r o t on P o s i t i on ( mm ) N u m be r o f P r o t on s P r o t on P o s i t i on ( mm ) L i gh t Y i e l d ( PE ) (a) (b) Figure 8 . The a) beam intensity and b) average MIP yield distribution profile as a function of proton positionin the upstream tracker. The MIP yield in b) is that of a SCSN-81 tile with large dimple. The reduced lightyield at the dimple is clearly seen. (mm) proton
X0 10 20 30 40 50 m a x L Y / L Y Center Y-profileSide Y-profile (mm) proton
Y0 10 20 30 40 50 m a x L Y / L Y Center X-profileTop X-profile (a) (b)
Figure 9 . Normalized MIP yield distribution profiles for the tile with an increased dimple size shown inFigure 8 along a) 2 mm y-bands at the tile center and 8 mm to the right side of the dimple area and b) 2 mmx-bands at the tile center and 8 mm to the top side of the dimple area. Points with error bars correspond tocenter profiles, while dashed lines show the off-center profiles.
The light yield was measured for tiles of different thicknesses wrapped in either ESR or Tyvek ® .This was done by stacking 3 . roton Position (mm)0 10 20 30 40 50 60 70 P r o t on P o s i t i on ( mm ) L i gh t Y i e l d ( PE ) Figure 10 . The average MIP yield distribution profile of the 5 . × . EJ-200 tile as a function ofproton position in the upstream tracker. grease in between to simulate a solid tile. Care was taken to avoid bubbles in the optical greasecoupling. The edges of the tiles were diamond fly-cut to polish them. Sets of three square tileswith dimensions 3 ×
3, 4 ×
4, and 5 × were prepared. From each set, one tile had a dimplemachined in it and the other two tiles had unmachined faces. The results are compared to simulationand shown in Fig 14. A 𝜒 fit is performed to the light yield measured in data of each tile thicknesswith the function 𝑝 × ( Tile Thickness / . ) 𝑝 , where 𝑝 and 𝑝 are parameters of the fit andthe fitted values are shown in Table 2. In all cases the increase in light yield was less than linear intile thickness.The light yield difference between tiles wrapped with ESR vs. Tyvek ® is shown in Table 3,where the ratio of MPV from ESR wrapped tiles to Tyvek ® wrapped tiles for each sample size islisted. In all cases, ESR wrapped tiles demonstrate much higher light yields than tiles wrapped inTyvek ® . The light yield as a function of hole size in the ESR reflector was measured for a 3 × EJ200 tilewrapped in ESR. One expects that as the hole diameter in the reflective foil decreases, the light yieldshould increase, since fewer photons escape through the gap between SiPM and wrapper. Likewise,it is expected that if the white silkscreen is changed to a non-reflective surface, the light yield shoulddecrease, since fewer photons reflect back into the tile through the gap. The light yield was studied– 12 – enter ring outer radius (mm)0 10 20 30 40 50 M PV ( PE ) Data MPV, R-profileSimulation MPV, R-profile
Entire tile MPV = 14.31 PE
Figure 11 . The MPV from concentric annular regions centered at the tile’s dimple (black points) for the5 . × . ±
3% systematic, which dominate the statistical errors in the fits.Simulated results are indicated by open triangles.
Table 2 . The fit values of 𝑝 and 𝑝 in the 𝑝 × ( Tile Thickness / . ) 𝑝 fit to the different sized andwrapped tiles. SCSN-81 tile area, mm Wrapping 𝑝 fit values 𝑝 fit values30 . × . . ± .
74 0 . ± . . × . ® . ± .
22 0 . ± . . × . . ± .
44 0 . ± . . × . ® . ± .
16 0 . ± . . × . . ± .
21 0 . ± . . × . ® . ± .
13 0 . ± . . . .
35 mm diameter. For these hole sizes, the white backplate was compared to data taken with thewhite backplate covered with black tape, which is a good approximation to a nonreflective surface.The results are compared to simulation and shown in Fig. 15. A 𝜒 fit to data is performed withthe function 𝑝 × ( Hole Diameter / ) 𝑝 where 𝑝 and 𝑝 are parameters of the fit and fittedvalues of 𝑝 = . ± .
31 PE and 𝑝 = − . ± .
06 for WSS and 𝑝 = . ± .
04 PE and 𝑝 = − . ± .
06 for black tape are extracted. In accordance with our expectations, larger holeshad lower light yields, and the effect is stronger for the black backplate.– 13 –
Average light yield (PE)20 25 30 35 40 45 50 55 60 N u m be r o f b i n s Average light yield (PE)15 20 25 30 35 40 N u m be r o f b i n s (a) (b) Average light yield (PE)10 15 20 25 30 35 40 45 N u m be r o f b i n s (c) Figure 12 . The average light yield in 2 × × EJ-200 tile, b) the 5 . × . EJ-200 tile, c) and the SCSN-81 tile with large dimple.
A systematic uncertainty was calculated for the reproducibility of the optical coupling between tileand SiPM by comparing the MPV across nominally identical measurements of the same tile withthe same wrapping. A typical reproducibility of 3 × tiles was found to be ± Tile Area (cm M PV ( PE ) SimulationDataData Fit ) Tile Area (cm5 10 15 20 25 30 S i m . D a t a Figure 13 . The light yield reported as MPV in data and simulation as a function of tile area for a 3 mmthick, wrapped in ESR, EJ-200 scintillator tile. A systematic uncertainty due to reproduciblitiy of opticalcoupling is estimated to be 3 . M PV ( PE ) Data Fit ·
3 Data Fit ·
4 Data Fit · ·
3 Data ·
3 Simulation ·
4 Data ·
4 Simulation ·
5 Data · M PV ( PE ) Data Fit ·
3 Data Fit ·
4 Data Fit · ·
3 Data ·
3 Simulation ·
4 Data ·
4 Simulation ·
5 Data · Figure 14 . The light yield reported as MPV in data and simulation as a function of tile thickness for 3 × ×
4, and 5 × square SCSN-81 tiles, wrapped in ESR (left) or Tyvek ® (right). The different coloredcurves are the fits to data. We studied intentional misalignment of the SiPM relative to its nominal location in the x-y plane,and also in its depth into dimple (z direction). For a SiPM misaligned by 1 . . . able 3 . The ratio of MPV for ESR divided by Tyvek ® wrapped tiles. SCSN-81 tile dimensions, mm ESR/Tyvek ® light yield MPV ratio30 . × . × . . ± . . × . × . . ± . . × . × . . ± . . × . × . . ± . . × . × . . ± . . × . × . . ± . . × . × . . ± . . × . × . . ± . . × . × . . ± . Hole Diameter (mm) M PV ( PE ) Simulation WSSData WSSSimulation Black TapeData Black TapeData WSS FitData Black Tape Fit
Hole Diameter (mm)3.5 4 4.5 5 5.5 6 6.5 S i m . D a t a Figure 15 . The MPV of the light yields in data and simulation as a function of hole size are shown forSiPMs sitting on top of a white silkscreened backplate and a black tape backplate. (S13360 SiPM, EJ-200tile.) this is a significant effect, we believe it is not important for our studies due to the design of themechanics of our test stand that naturally provided consistent z placement. Additionally, the opticalcoupling reproducibility measurement, Sec 8.1, includes any residual effect of z misalignment. Ourtile-to-SiPM mounting was very reproducible for most tiles.
The temperature of the SiPM was measured using a PT10000 temperature sensor placed on thePCB close to the SiPM. Using this sensor, we observed a temperature variation between 26 and28 ◦ C for all runs. For our SiPMs and our choice of over-voltage, the expected change in the number– 16 –f detected photons due to temperature variation is less than 1%. The systematic uncertainty inoptical coupling covers this uncertainty.
A setup to study the responses of different scintillator tile geometries and materials has been installedat the Fermilab Test Beam Facility. This was used to collect and analyze data in a January-February2020 test beam run. The light yield and uniformity was measured for various values of the tilesize, tile thickness, type of the reflective wrapper, and the hole diameter in the wrapper. ESR wasfound to be substantially more reflective than Tyvek ® , with a ratio of light yield of roughly 2 to 4times, depending on other tile parameters. The light yield of a tile was measured as approximatelyinversely proportional to the square root of the tile area. The light yield was observed to increasemuch less than linearly with thickness, with the exact amount depending on tile size and wrappingmaterial. Simulation was developed that agrees well with the light yield and uniformity measuredin test beam data across the different tiles. Acknowledgments
Work supported by the Fermi National Accelerator Laboratory, managed and operated by FermiResearch Alliance, LLC under Contract No. DE-AC02-07CH11359 with the U.S. Department ofEnergy. The U.S. Government retains and the publisher, by accepting the article for publication,acknowledges that the U.S. Government retains a non-exclusive, paid-up, irrevocable, world-widelicense to publish or reproduce the published form of this manuscript, or allow others to do so, forU.S. Government purposes. Work also supported by the US-DOE Office of Science (High EnergyPhysics) under Award Number DE-SC0011845 and DE-SC0010072. Additional support providedby the University of Maryland Physics Department.– 17 – eferences [1] Apollinari G. et al.,
High-Luminosity Large Hadron Collider (HL-LHC): Technical Design Report V.0.1 , CERN-2017-007-M (2017), doi:10.23731/CYRM-2017-004.[2] CMS Collaboration,
The Phase-2 Upgrade of the CMS Endcap Calorimeter , CERN-LHC-2017-023,CMS-TDR-019 (2019), https://cds.cern.ch/record/2293646.[3] Y. Liu et al.,
A Design of Scintillator Tiles Read Out by Surface-Mounted SiPMs for a Future HadronCalorimeter , in
Proceedings, 21st Symposium on Room-Temperature Semiconductor X-ray andGamma-ray Detectors (RTSD 2014) , Seattle, WA, USA, November 8-15, 2014, p. 7431118. 2016.arXiv:1512.05900. doi:10.1109/NSSMIC.2014.7431118.[4] G. Blazey et al.,
Directly Coupled Tiles as Elements of a Scintillator Calorimeterwith MPPCReadout , Nucl. Instrum. Meth. A (2009) 277,doi:10.1016/j.nima.2009.03.253.[5] N. Tsuji,
Study on Larger Scintillator Tile for AHCAL , in CALICE Collaboration Meeting, Universityof Tokyo, September 2017.[6] E. Niner,
Fermilab Test Beam Facility Status and Plans , 8th Beam Telescopes and TestbeamsWorkshop, Tbilisi, Georgia, Jan 27-31, 2020.[7] Eljen Technology, (January 2016),
General Purpose Plastic Scintillator EJ-200, EJ-204, EJ-208,EJ-212 , eljentechnology,https://eljentechnology.com/images/products/data_sheets/EJ-200_EJ-204_EJ-208_EJ-212.pdf.[8] Kurary, (June 2006),
Scintillation Materials
Geant4—a simulation toolkit , Nucl. Instrum. Meth. A , 506: 250 (2003).[10] S. Los and J. Freeman,
A Low Cost SIPM Evaluation and Control Prototyping System IncludingAccurate Bias Voltage Generation, Leakage Current Measurement, and Peltier Temperature Control
Fermilab-TM-2749 2020.[11] Paul Sherrer Insitut, (Septemember 2018),
DRS4 Evaluation Board User’s Manual
PREP Modernization , CDCVS,https://cdcvs.fnal.gov/redmine/projects/prepmodernization/wiki[14] K. Biery et al,
The Fermilab Test Beam Facility Data Acquisition System Based on 𝑜𝑡𝑠𝑑𝑎𝑞 , 2018,arXiv:1806.07240 [physics.ins-det].[15] Hamamatsu, (August 2016),
MPPC S13360 series
MPPC S14160-1310PS/-1315PS/-3010PS/-3015PS
Reflectivity Spectra for Commonly Used Reflectors , United States: N. p., 2012. Web.doi:10.1109/TNS.2012.2183385.[18] M. Tanabashi et al. (Particle Data Group),
Phys. Rev. D
98, 030001 (2018).[19] L.M.S. de Silva and F. Simon,
Effects of misalignment on response uniformity of SiPM-on-tiletechnology for highly granular calorimeters , JINST
15 P06030, 2020.15 P06030, 2020.