Performance of VUV-sensitive MPPC for liquid argon scintillation light

Nuclear Instruments and Methods in Physics Research A 833 (2016) 239–244

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima

Performance of VUV-sensitive MPPC for liquid argon scintillation light T. Igarashi, M. Tanaka, T. Washimi n, K. Yorita Waseda University, Tokyo, Japan

art ic l e i nf o

a b s t r a c t

Article history: Received 25 March 2016 Received in revised form 16 June 2016 Accepted 4 July 2016 Available online 5 July 2016

A new multi-pixel photon counter (MPPC) sensitive to vacuum ultra-violet (VUV) light (wavelength λ < 150 nm ) has recently been developed and produced by Hamamatsu Photonics K.K. In this study, the basic properties of the new MPPC are measured at the cryogenic facility of the Waseda University using liquid nitrogen. The temperature dependence of the breakdown voltage, capacitance, and dark count rate of the MPPCs are also evaluated. Using an 241Am α-ray source, the absolute photon detection efficiency (PDE) of the liquid argon (LAr) scintillation light (λ ¼128 nm) for the latest MPPC model is estimated to be 13%. Based on these basic measurements a possible application of the new MPPC to LAr detectors in dark matter search is suggested. & 2016 Elsevier B.V. All rights reserved.

Keywords: MPPC SiPM Liquid argon VUV light

scintillation light;

As part of the ANKOK project [1], a direct dark matter search experiment using double phase Ar detector, we have been developing a new procedure to detect the LAr scintillation light using multi-pixel photon counters (MPPCs) [2], one of SiPMs [3,4]. Conventional MPPCs have high efficiency peaking at 400–500 nm and in general no sensitivity for VUV light (here, defined as wavelength below 150 nm). Meanwhile, an improved MPPC has been developed for the liquid xenon (LXe) scintillation light (wavelength of 175 nm) and is already being practically applied to the MEG experiment [5]. The sensitivity has been further extended to below 150 nm for a new MPPC currently under development at Hamamatsu Photonics K.K. [6] and thus this kind of new MPPC may lead to the development of a new type of liquid or gaseous argon scintillation detector in the near future. In this paper, the basic properties of the improved MPPC, its detection performance for the LAr scintillation light, and its physics applications are reported and discussed.

argon;

2. The VUV-sensitive MPPC

small signals and powerful background rejection.

The new MPPCs, designated “3X3MM-50UM VUV2”, “3X3MM50UM VUV3” and “3X3MM-100UM VUV3” [7], were developed by Hamamatsu Photonics K.K. for the direct detection of VUV light below 150 nm and were tested at the Waseda University campus under a cryogenic environment [8]. The VUV2 MPPC was developed in April 2014, and its basic parameters are similar to those of the commercially available MPPC, S12572-33-050C [9]. Meanwhile, the VUV3 MPPC, produced in April 2015, is a cross-talk suppressed model similar to the S13360-3050CS model [10]. The chip size is 3 mm  3 mm, and the labels of -50UM and -100UM denote pixel pitch of 50 μm and 100 μm , respectively (Fig. 1). The

1. Introduction Liquid argon (LAr) is known to be an excellent target material for various particle physics experiments, such as neutrino physics, nucleon decay, and the direct search for weakly interacting massive particles (WIMP dark matter). For the WIMP search, the ionization and scintillation signals of LAr and the scintillation pulse shape discrimination of LAr detectors provide a strong rejection power between the electron-recoil (main background) and the nuclear-recoil (WIMP signal) events. Detection of the LAr scintillation light using photosensors is difficult, particularly for the purpose of particle physics experiment, because it requires the photosensor to

 detect 128 nm vacuum ultra-violet (VUV) photons in the LAr  operate under an LAr temperature of −186 °C ;  operate at a relatively low voltage to avoid discharge into pure  demonstrate single-photoelectron counting to detect extremely Because of these technical requirements, the conventional procedure for detecting the LAr scintillation light is to convert the 128 nm light to visible light (420 nm) using tetraphenyl butadiene (TPB) as a wavelength-shifter and then detect these converted photons using cryogenic photomultiplier tubes (PMTs). n

Corresponding author. E-mail addresses: [email protected] (T. Washimi), [email protected] (K. Yorita). http://dx.doi.org/10.1016/j.nima.2016.07.008 0168-9002/& 2016 Elsevier B.V. All rights reserved.

240

T. Igarashi et al. / Nuclear Instruments and Methods in Physics Research A 833 (2016) 239–244

Fig. 1. The microphotograph of 3X3MM-50UM VUV3 [7].

MPPC

Type no.

V2-50UM-(1) 3X3MM50UM VUV2 V2-50UM-(2) 3X3MM50UM VUV2 V3-50UM 3X3MM50UM VUV3 V3-100UM 3X3MM100UM VUV3

Serial no. Bias voltage (V)

Gain

Dark counts (kHz)

A0010

66.65

1.25 × 106

572

A0011

66.77

1.25 × 106

701

A0011

54.87

2.00 × 106 674

A0003

53.78

5.50 × 106

553

number of pixel is 3600 for -50UM and 900 for -100UM. The basic parameters of these MPPCs as measured by Hamamatsu Photonics K.K. are summarized in Table 1. These measurements were all performed at room temperature ( 25 °C ).

3. Basic properties of VUV-sensitive MPPCs In addition to the room-temperature properties presented in Table 1, we have extensively measured the performance of three of the MPPCs (V2-50UM-(1), V3-50UM, and V3-100UM) at cryogenic temperature. The image at the top of Fig. 2 shows a schematic diagram of the test setup built at Waseda University. The MPPC is mounted inside a vacuum chamber filled with 1 bar of pure gas nitrogen and refrigerated by immersing the chamber in a open liquid nitrogen bath. Two thermometers (Pt100) located close to the MPPC are used to monitor the temperature around the MPPC, and the difference in the readout values of the two thermometers ( ±5 °C ) is considered as the systematic uncertainty of the MPPC temperature. A LED with a wavelength of 415 nm is located outside of the chamber at room temperature, and its emitted photons are injected into the MPPC surface through an optical fiber and feed-through. The average light yield of the LED signal is adjusted to approximately one photon per pulse at the surface of MPPC. The LED pulse time width is several nanoseconds, which is much shorter than the MPPC time response (40 ns for -50UM). The LED light yield per pulse is monitored by a PMT (Hamamatsu H1161), and its stability was found to be within 0.5% throughout the

3X3MM-50UM VUV3

12000

Events / 0.2 pC

Table 1 List of the new VUV-sensitive MPPCs and their basic properties measured by Hamamatsu Photonics K.K. at room temperature of 25 °C .

10000

Q0, N 0

8000 6000

Q 1, N 1

4000

Q2, N 2

2000 0

−2

0

2

4

6

8

Signal Charge (pC)

Fig. 2. Schematic diagram of the setup for measurements of the basic properties of MPPCs at cryogenic temperature (top) and the signal charge distribution of an MPPC (3X3MM-50UM VUV3) measured with 415 nm LED light at −190 °C (bottom).

measurement period. A MPPC driver kit (Hamamatsu C12332) is used to supply the bias voltage and signal amplification (gain Gamp = 10.9 ± 0.1). The signal is digitized by CAEN flash ADC (FADC) waveform digitizer (V1724, 100 Ms/s), and the digitized data are stored on a personal computer. The FADC waveform is analyzed to extract the signal charge by integrating the waveform in the time range of [  20 ns, 120 ns] (for -50UM) and [  20 ns, 500 ns] (for -100UM) from the LED pulse timing, because the pulse width changes in accordance with the pixel capacitance (proportional to the area of a pixel). The plot at the bottom of Fig. 2 shows an example of the signal charge distribution in units of pC for 50,000 data events, obtained at −190 °C with an MPPC (V3-50UM) bias voltage of 44.0 V. Each peak in the plot corresponds to zero, one, or two photoelectrons. The signal charge distribution is simultaneously fitted by three Gaussian distributions to estimate each mean charge (Q0, Q1, and Q2) and the number of events (N0, N1, and N2). The straight average of the charge distribution ( Q LED) is used to calculate the absolute value of photon detection efficiency (PDE) described later in Section 4. The MPPC gain (G) is obtained from the difference between the peak charge (Q G ≡ Q 1 − Q 0 = Q 2 − Q 1) by using the relation Q G = e × G × Gamp , where e is the elementary charge and Gamp is the signal amplification gain. As is well known, the gain of the MPPC is described as

T. Igarashi et al. / Nuclear Instruments and Methods in Physics Research A 833 (2016) 239–244

241

×10

1.5

3X3MM-50UM VUV2 3X3MM-100UM VUV3

15

MPPC Gain

Relative PDE (=1@Vov=3V)

3X3MM-50UM VUV3

10

5

0

0

2

4

6

MPPC Over-Voltage (V)

8

Cross-talk & Afterpulse Probability

6

20

3X3MM-50UM VUV2 3X3MM-50UM VUV3 3X3MM-100UM VUV3

1

0.5

0

0

2

4

6

8

1

3X3MM-50UM VUV2 3X3MM-50UM VUV3 3X3MM-100UM VUV3

0.8 0.6 0.4 0.2 0

0

2

MPPC Over-Voltage (V)

4

6

8

MPPC Over-Voltage (V)

Fig. 3. MPPC gain (left), relative PDE (middle), and cross-talk and afterpulse probability (right) as a function of the MPPC over-voltage at −190 °C measured with 415 nm LED light.

where C is the capacitance of one pixel, Vbd is the breakdown voltage, and Vov is the over-voltage [2]. The gain changes linearly with the over-voltage, even at −190 °C , as shown in the left plot in Fig. 3. Note that for this measurement, the breakdown voltage is obtained as 41.3 V from the Vbias –G plot by linear extrapolation to the zero-gain point; this voltage is already subtracted by Vbias in Fig. 3. We perform the same procedure for V2-50UM-(1) and V3100UM, and these results are shown in Fig. 3. The number of photons injected from the LED to the MPPC is assumed to follow a Poisson distribution with average μin . Thus the number of events with zero photoelectron is

N0 = Nall × e−μ = Nall × e−μin × PDE ,

(3)

The cross-talk and afterpulse probability is evaluated from a wider time window (integral range of [  20 ns, 600 ns]) to collect the afterpulses. As shown in the right plot in Fig. 3, the cross-talk and afterpulse probability X increases with the imposed over-voltage, and is lower for VUV3 than VUV2 at the same over-voltage. To evaluate the temperature dependence of the new MPPC, over a wider temperature range, we measure the gain and dark count rate at 8 different temperature points ( −190, −160, −130, −100, −80, −50, −20, and 20 °C ). The top plot in Fig. 4 shows the gain as a function of bias voltage at each temperature for V250UM-(1). The gain slope, which is proportional to the MPPC capacitance (about 85 fF), is independent of temperature while the breakdown voltage decreases with temperature (approximately 50 mV/K). The bottom plot in Fig. 4 shows the dark rate of V250UM-(1) counted with a threshold equivalent to one photoelectron without the LED light signals. The rate decreases significantly with decreasing temperature, becoming less than 1 Hz at cryogenic temperatures (below −150 °C ), where the upper bound is

3X3MM-50UM VUV2

2.0

°

-160 C ° -100 C

°

-190 C 1.5

°

-20 C

°

-50 C

1.0 °

-130 C

0.5

0.0

(2)

where Nall is the total number of events and PDE is the photon detection efficiency. In this setup, the absolute value of PDE cannot be extracted because the original μin is unknown. However, its dependence on Vov and the temperature can be estimated. The middle plot in Fig. 3 shows the relative PDE as a function of Vov , normalized to unity at Vov = 3.0 V . Because of the effects of the cross-talk and afterpulse, which may occur in the time window, the number of events observed in one photoelectron is decreased relative to the Poisson expectation by the cross-talk and afterpulse probability (X), defined as follows:

N1 = Nall × μe−μ × (1 − X ).

×106

(1)

MPPC Gain

C (Vbias − Vbd ) CV = ov , e e

55

°

-80 C

°

20 C

60

65

MPPC Bias Voltage (V)

Dark Rate: > 1 P.E. (Hz)

G=

3X3MM-50UM VUV2

106 105 104 103 102 10 1 −1

10

10−2 −200

−150

−100

−50

0

50

o

Temperature ( C) Fig. 4. 3X3MM-50UM VUV2 (A0010) gain as a function of over-voltage at various operation temperatures (top). The solid squares in the plot correspond to the values measured by Hamamatsu Photonics K.K. at 25 °C . The bottom plot is the dark count rate with a threshold of one photoelectron at different temperatures.

limited by experimental conditions (e.g., small light leak).

4. Measurement of liquid argon scintillation light Fig. 5 shows a schematic diagram of the setup for measuring LAr scintillation light. The details of the setup for obtaining highpurity LAr are described elsewhere [1]. Each of the three MPPCs

242

T. Igarashi et al. / Nuclear Instruments and Methods in Physics Research A 833 (2016) 239–244

Table 2 Mean charge and energy resolution of Fig. 5 (top), and all parameters for deriving the PDE. The errors are only statistical uncertainty. See the text about the systematic uncertainty. Parameters

V2-50UM-(2)

Bias voltage ( Vbias ) Mean charge ( Q α ) Energy resolution ( σ /Q α ) Gain ( G )

56.20 V 44.85 V 45.27 V 592.6 ± 1.0 pC 345.0 ± 0.6 pC 2528.4 ± 3.4 pC 9.1 ± 0.2% 12.6 ± 0.2% 11.6 ± 0.2%

Cross-talk and afterpulse ( Npix ) # of pixel hit ( Nuncorr ) # of photoelectrons ( Ncorr ) # of pixel

160.5 96.7 3600 Counting loss correction ( k ) 1.01 PDE 7.3%

1cm

214

Fig. 5. Experimental apparatus for the liquid argon test setup at Waseda University.

(V2-50UM-(2), V3-50UM and V3-100UM) is immersed into the LAr, and the 241Am α-ray source (40 Bq) is placed 1 cm from each MPPC. The MPPC power supply and signal readout for the setup are the same as in Fig. 2. The top plot in Fig. 6 shows the signal charge distributions of the three MPPCs at the same over-voltage, 3 V, within [  20 ns, 10 μs] from the trigger timing. A clear peak corresponding to 5.5 MeV α-rays from 241Am is observed, and its signal rate is consistent with the expected value. The distribution is fitted by a single Gaussian function to obtain the mean charge ( Q α ) and the

300 3X3MM-50UM VUV2

250

3X3MM-50UM VUV3

Events / 2 pC

200

100

Npix =

50 0

1000

2000

3000

4000

Signal Charge (pC) 3X3MM-50UM VUV3

Average Pulse Height (mV)

1.8 × 106 1.1

7.5 × 106 1.1

117.9 108.1 3600 1.01 7.8%

180.5 170.4 900 1.06 12.9%

1 Q 1 . = α × Npix QG Npix

(4)

In this equation, Npix is the correction factor defined as the total number of MPPC pixel hits including the real one-photoelectron equivalent signal and the hits due to cross-talk and afterpulse:

150

102 10

Q LED 1 × . QG μ

(5)

By definition, Npix is always greater than 1 because of the effect of the cross-talk and afterpulse, and its over-voltage dependence is shown in the top plot in Fig. 7. This plot is used for PDE correction. For example, in the case of V2-50UM-(2), the number of photoelectrons at Vov ¼ 2.3 V with and without this correction are calculated to be Nreal = 96.7 ± 1.2 and Nuncorr = 160.5 ± 2.0, respectively. The expected number of LAr scintillation photons at the surface of the MPPC ( Nα ) is estimated using the following equation:

Nα = Eα/Wα × AMPPC = 5.5 MeV/(27.5 eV/photon) × 0.7 %= 1400 photons,

1 10−1 10−2

V3-100UM

energy resolution ( σ /Q α ). The results are summarized in Table 2. The energy resolution is improved by cross-talk suppression from VUV2 to VUV3 and the change of the pixel pitch from 50 μm to 100 μm . The bottom plot in Fig. 6 shows the average waveform for the signal events within 2σ around the peak of the charge distribution. The scintillation light of LAr is known to have two components with different decay times, fast ( O (10 ns)) and slow ( O (1 μs)) [11], and the waveform obtained by this measurement also has two components with a slow decay time of approximately one microsecond. Q α can be translated into the number of photoelectrons by dividing it by the gain (G). Because the charge distribution includes the effects of cross-talk and afterpulse, the following correction is necessary to evaluate the number of real photoelectrons in the signal:

Nreal = Nuncorr ×

3X3MM-100UM VUV3

0

1.8 × 106 2.6

V3-50UM

−2

0

2

4

6

8

10

Time (µs) Fig. 6. Signal charge distributions for three different MPPCs (top) and average waveform (bottom) for one MPPC (V3-50UM) for argon scintillation light by 241Am α-rays (5.5 MeV) at Vov ∼ 3 V .

(6)

where Eα is the energy of the 241Am α-rays, Wα is the LAr scintillation photon emission yield for α particle [11], and AMPPC is the acceptance calculated by the solid angle from the 241Am source to the MPPC. Finally, the photon detection efficiency is determined as Nreal/Nα . If two or more photons enter the same pixel within time constant of the MPPC, the total number of the fired pixel does not linearly correspond to the number of incident photons and needs to be corrected. In order to estimate the correction factor k = PDE/(Nreal/Nα ), we performed the Monte Carlo simulation by taking into account the LAr scintillation waveform. The same procedure is performed at each over-voltage studied. The bottom

Number of Pixel Hits per 1 P.E.

T. Igarashi et al. / Nuclear Instruments and Methods in Physics Research A 833 (2016) 239–244

4

triggered simultaneously or non-simultaneously (k ¼1). The relative uncertainty is approximately 1% for the 50 μm pitch (3600 pixel) MPPCs and approximately 10% for the 100 μm pitch (900 pixel) MPPC. Moreover, for the V2-50UM model, different MPPCs are used to observe the LAr scintillation (V2-50UM-(2)) and estimate the gain and cross-talk (V2-50UM-(1)). This difference comprises another major source of the systematic uncertainty. The relative uncertainty in the PDE is assigned as approximately 30% for the VUV2 MPPC and 10–20% for the VUV3 MPPCs, as shown by the dotted lines in the plot in Fig. 7.

3X3MM-50UM VUV2 3X3MM-50UM VUV3 3X3MM-100UM VUV3

3

2

1

0

5. Summary

0

2

4

6

8

Photon Detection Efficiency (%)

MPPC Over Voltage (V) 15

10

5 3X3MM-50UM VUV2 3X3MM-50UM VUV3 3X3MM-100UM VUV3

0

0

243

1

2

3

4

5

MPPC Over-Voltage (V) Fig. 7. Average number of total pixel hits per one photoelectron equivalent signal, Npix (top) and corrected PDE (bottom) as a function of MPPC over-voltage for 128 nm LAr scintillation light.

plot in Fig. 7 shows the PDEs obtained for the three MPPCs as a function of the over-voltage. The PDE is approximately 8% for the 50 μm pixel MPPCs and approximately 13% for the 100 μm pixel MPPC at Vov = 3 V . The PDE contains the fill factor, the sensitive area efficiency, such as PDE = QE × ε × Ptrigger , where QE is the quantum efficiency of the active area, ε is the fill factor, and Ptrigger is the probability that an incoming photon triggers a Geiger-avalanche multiplication [4]. The fill factor is 40% for -50UM and 62% for a for -100UM [7], so the difference of the PDE in our measurement is about consistent to the difference of the fill factor. The basic properties of the MPPC shown in Fig. 2 (gain and cross-talk) are measured as a function of the over-voltage at the liquid nitrogen temperature, while the α-ray data are obtained inside the liquid argon. Because the breakdown voltage depends fairly strongly on temperature (approximately 50 mV/K), the breakdown voltage at the liquid argon temperature needs to be determined independently. The precision of the breakdown voltage determination ( +−0.17 0.21V ) is found to be one of the major sources of the systematic uncertainty in the PDE measurement. The systematic uncertainty due to the correction factor k for the counting loss is conservatively calculated by assuming that all photons were

A new multi-pixel photon counter (MPPC) with a sensitivity to VUV light (wavelength <150 nm ) has recently been developed by Hamamatsu Photonics K.K. We tested the new MPPC under cryogenic temperatures ( −190 °C ) and measured several of its basic properties. We then successfully detected the scintillation light of LAr (wavelength¼128 nm) with this MPPC. The photon-detection efficiency was measured to be approximately 8% for 50 μm pixel MPPCs and approximately 13% for a 100 μm pixel MPPC at Vov = 3 V. In the recent WIMP dark matter search experiments using liquefied noble gas (argon and xenon), it has been reported that the background events at the detector surface may be mis-reconstructed as events at the center of the detector. These mis-reconstructed events remain in the signal region and limit the instrument sensitivity. For the ANKOK experiment, we are attempting to improve the spatial reconstruction resolution by arranging the MPPC. For the double-phase argon detector the spatial resolution in terms of electron drift direction is determined very precisely (O(
Acknowledgments We first acknowledge the Solid State Division of Hamamatsu Photonics K.K. for providing us with the new VUV MPPC samples. We particularly appreciate useful discussions with Y. Hakamata, K. Sato, R. Yamada, and Y. Ohashi. We are grateful to W. Ootani for helpful comments and support. This work is a part of the outcome of research performed under the Waseda University Research Institute for Science and Engineering (Project numbers 13C09 and 14C12), supported by JSPS Grant-in-Aid for Challenging Exploratory Research Grant Number 25610060.

References [1] M. Tanaka, Status of r&d on double phase argon detector: the ANKOK project, in: CYGNUS 2013: 4th Workshop on Directional Detection of Dark Matter, vol. 469, 2013, p. 012012, http://dx.doi.org/10.1111/10.1088/1742-6596/469/1/

244

T. Igarashi et al. / Nuclear Instruments and Methods in Physics Research A 833 (2016) 239–244

012012. [2] S. Gomi, et al., Development and study of the multi pixel photon counter, Nucl. Instrum. Methods Phys. Res. Sect. A 581 (2007) 427–432, http://dx.doi.org/ 10.1016/j.nima.2007.08.020. [3] P. Buzhan, et al., Silicon photomultiplier and its possible applications, Nucl. Instrum. Methods Phys. Res. Sect. A 504 (2003) 48–52, http://dx.doi.org/ 10.1016/S0168-9002(03)00749-6. [4] D. Renker, Geiger-mode avalanche photodiodes, history, properties and problems, Nucl. Instrum. Methods Phys. Res. Sect. A 567 (2006) 48–56, http://dx. doi.org/10.1016/j.nima.2006.05.060. [5] W. Ootani, et al., Development of deep-UV sensitive MPPC for liquid xenon scintillation detector, Nucl. Instrum. Methods Phys. Res. Sect. A 787 (2014) 220–223, http://dx.doi.org/10.1016/j.nima.2014.12.007. [6] K. Sato, et al., The UV sensitivity improvement of MPPC, Nucl. Instrum.

[7] [8]

[9] [10] [11]

Methods Phys. Res. Sect. A 732 (2013) 427–430, http://dx.doi.org/10.1016/j. nima.2013.06.054. Private Communication with Hamamatsu Photonics K.K. T. Washimi, M. Tanaka, K. Yorita, Direct detection of liquid argon scintillation with MPPC, in: Light Detection in Noble Elements (LIDINE 2015), vol. 11, 2016, p. C02077, http://dx.doi.org/10.1088/1748-0221/11/02/C02077. URL 〈http://www.hamamatsu.com/jp/en/product/category/3100/4004/4113/ S12572-050C/index.html〉. URL 〈http://www.hamamatsu.com/jp/en/product/category/3100/4004/4113/ S13360-3050CS/index.html〉. T. Doke, K. Masuda, Present status of liquid rare gas scintillation detectors and their new application to gamma-ray calorimeters, Nucl. Instrum. Methods Phys. Res. Sect. A 420 (1999) 62–88, http://dx.doi.org/10.1016/S0168-9002(98) 00933-4.