Calibration of LHCf calorimeters for photon measurement by CERN SPS test beam

Calibration of LHCf calorimeters for photon measurement by CERN SPS test beam

Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136 Contents lists available at SciVerse ScienceDirect Nuclear Instruments and ...

536KB Sizes 5 Downloads 43 Views

Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

Contents lists available at SciVerse ScienceDirect

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

Calibration of LHCf calorimeters for photon measurement by CERN SPS test beam T. Mase a,n, O. Adriani b,c, L. Bonechi b, M. Bongi b, G. Castellini b,c, R. D’Alessandro b,c, K. Fukui a,d, M. Haguenauer f, Y. Itow a,e, K. Kasahara g, D. Macina h, K. Masuda a, H. Menjo b,e, G. Mitsuka a, M. Mizuishi g, Y. Muraki i, M. Nakai g, P. Papini b, A.-L. Perrot h, S. Ricciarini b, T. Sako a,e, Y. Shimizu g, T. Sumi a, K. Taki a, T. Tamura j, S. Torii g, A. Tricomi k, W.C. Turner l, A. Viciani b, H. Watanabe a, K. Yoshida m a

Solar-Terrestrial Environment Laboratory, Nagoya University, Nagoya, Japan INFN Section of Florence, Italy c University of Florence, Italy d Centro Siciliano di Fisica Nucleare e Struttura della Materia, Catania, Italy e Kobayashi-Maskawa Institute for the Origin of Particles and the Universe, Nagoya University, Nagoya, Japan f Ecole-Polytechnique, Palaiseau, France g RISE, Waseda University, Japan h CERN, Switzerland i Konan University, Japan j Kanagawa University, Japan k INFN Section of Catania and University of Catania, Italy l LBNL, Berkeley, California, USA m Shibaura Institute of Technology, Japan b

a r t i c l e i n f o

abstract

Article history: Received 17 December 2011 Accepted 22 December 2011 Available online 30 December 2011

Energy resolution and linearity of the LHCf calorimeters for electromagnetic showers were measured at the SPS H4 beam line in 2007 using electron beams of 50–200 GeV and muon beams of 150 GeV. The absolute energy scale was determined in these data. The results that were obtained (o 5% energy resolution) are well understood by using Monte Carlo simulations and are good enough for the requirements of the LHCf experiment. & 2011 Elsevier B.V. All rights reserved.

Keywords: LHC High-energy cosmic-rays Sampling calorimeter

1. Introduction The LHC forward experiment (LHCf) is one of the LHC experiments. LHCf is specifically designed for measurements of the very forward (pseudo-rapidity; Z 4 8:4) production spectra of gammarays, neutral pions and neutrons. These measurements are motivated by the desire to calibrate the hadron interaction models used to simulate cosmic-ray air showers. The LHCf data taking at pffiffi s ¼ 0:9 and 7 TeV proton–proton collisions has finished in the middle of July 2010 when the luminosity of LHC exceeded 1030 cm  2 s  1. Details of the experiment and detectors are found elsewhere [1–4]. In addition we have already verified the basic idea of the energy measurements using a prototype detector [5].

n

Corresponding author. Tel.: þ81 52 789 4320; fax: þ81 52 789 4313. E-mail address: [email protected] (T. Mase).

0168-9002/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2011.12.096

In this paper, we summarize the performance of the final LHCf detectors as they were installed on LHC. The beam test was performed at the CERN SPS North Area in 2007 before the installation in the LHC tunnel in 2008. We concentrate on the results for the energy measurement of electromagnetic showers.

2. The LHCf detectors The LHCf detectors are two independent shower calorimeters, named Arm1 and Arm2. Each detector consists of a pair of small sampling and imaging calorimeters, which we call a ‘tower’ hereafter, made of 16 layers of plastic scintillators (Eljen Technology EJ-260; 3 mm thickness) interleaved with tungsten converters (7 mm for the first 11 layers and 14 mm for the rest). The longitudinal structure of the towers is shown in Fig. 1. The longitudinal size is about 230 mm or 44 X0 ð1:55lÞ in units of radiation length (hadron interaction length). The transverse sizes

130

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

of the towers are 20 mm  20 mm and 40 mm  40 mm in Arm1, and 25 mm  25 mm and 32 mm  32 mm in Arm2. Four X–Y layers of position sensitive detectors, scintillating fiber (SciFi) belts in Arm1 [6] and microstrip silicon sensors in Arm2 [7], are inserted in order to provide incident positions of the showers. The calorimeters are designed to have energy and position resolutions better than 5% and 0.2 mm, respectively, for electromagnetic showers with energy 4 100 GeV. Because of the small aperture of the towers, the frequency of multi-particle hits in a single tower is reduced to a reasonable level. The double tower structure allows us to detect gamma-ray pairs from the decay of p0 s with a single photon induced shower in each tower. By reconstructing the invariant mass of gamma-ray pairs, we can identify the p0 s and hence measure their energy spectrum.

3. Beam test The fundamental properties of the LHCf detectors (conversion factors from measured charge to deposited energy, energy scale and linearity, and energy resolution) were investigated using the

0

50

100 150 Longitudinal size (mm)

200

Scintillator Tungsten Silicon

25, 32 mm

Detector#2 Vertical size

Scintillator Tungsten SciFi

20, 40 mm

Vertical size

Detector#1

CERN SPS North Area H4 beam line from 24 August to 11 September 2007. Beams of 50–200 GeV electrons, 150–350 GeV hadrons and 150 GeV muons were used in the test. The results were compared with a MC simulation that is used to predict the performance for the LHC energy range. 3.1. Set-up The set-up of the beam test is illustrated in Fig. 2. A microstrip silicon tracker was placed in front of the LHCf detector to measure the position of incident particles with high precision. This tracking system has been used for the ADAMO magnetic spectrometer [8] and is made of five identical detecting modules each of which is composed of two double-sided microstrip silicon sensors. Using this tracking system the impact points of particles on the detector are reconstructed with a resolution better than 15 mm. The trigger signal for the data acquisition system was generated by a coincidence of two small plastic scintillators (20 mm  20 mm and 40 mm  40 mm) placed at the exit of the beam pipe. These trigger scintillators were fixed relative to the beam line but the LHCf calorimeters and the silicon tracker were mounted on a movable stage that could be scanned through the beam. Data was taken under two sets of supplied high voltage to PMTs at low (450 V) and high (600 V) gains. The high (low) gain mode was identical (close) to the pffiffi condition used in the operation at s ¼ 0:9ð7Þ TeV proton–proton collisions. Throughout the paper, unless specified we present only the results from the high gain operation mode. The difference is essential only in the discussion on the energy resolution in Section 3.3.2. The electronics for data acquisition was essentially identical to that used in LHC except for the delay cables of the analog signal of the calorimeters. For the beam tests described here, we used RG58 cables with a total delay time of 300 ns while for the LHC operation we used 200 m long low attenuation cables (C-50-3-1, 850 ns delay). Additional tests to determine the different attenuation of these cables were also carried out during the beam tests at SPS. 3.2. Analysis

0

50

100 150 Longitudinal size (mm)

200

Fig. 1. Longitudinal structure of the LHCf calorimeters. Top (Bottom) panel shows the structure of the Arm1 (2) detector that uses SciFi (Silicon strip) as the position sensor.

3.2.1. Monte Carlo simulation and incident position For comparison with the beam test results, we used the MC simulation package COSMOS 7.49/EPICS 8.81 [9]. From the MC simulation, we can obtain deposited energy in each layer of calorimeter. When we compared the distribution of measured charge (ADC value, hereafter) from the beam test data and the deposited energy expected from the MC simulation, we always took account actually measured pedestal fluctuation in the MC data. We determined the track of each incident particle using the

Fig. 2. Set-up of the beam test at SPS. Signals for the data acquisitions were generated by the two fixed plastic scintillators (right in the figure) at the exit of the beam pipe. The LHCf detectors (left) and the ADAMO tracker (middle) were placed on a movable table to scan the calorimeters through the beams.

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

Particle Flux at 20mm tower

Particle Flux at 40mm tower 60

40

131

40

40

35

30 40

20

30

10

20

Y axis (mm)

Y axis (mm)

50 30

30

25

20

20 15

10

10 10

0

0

5

0

0

10 20 30 X axis (mm)

0

40

0

Particle Flux at 25mm tower

40

Particle Flux at 32mm tower 100

90

40

10 20 30 X axis (mm)

40

80

60 50

20

40

10

30

Y axis (mm)

Y axis (mm)

80 70

30

30 60

20 40

10

20

0

10

20

0

0

0

10 20 30 X axis (mm)

40

0

0

10 20 30 X axis (mm)

40

Fig. 3. Flux distributions of electrons. Top left for the Arm1 20 mm calorimeter and top right for the Arm1 40 mm calorimeter. Bottom left for the Arm2 25 mm calorimeter and bottom right for the Arm2 32 mm calorimeter. The energies of the incident particles were 180 GeV electrons for Arm1 and 150 GeV electrons for Arm2. The solid lines indicate the edge of each calorimeter. Events falling within 2 mm of the edge of the calorimeters were removed from analysis. When comparing the MC and beam test data, the MC was sampled according to these distributions.

silicon tracker that gives the precise impact point on the calorimeters. The events in which the track was not well reconstructed or multi-tracks were identified were removed from the analysis. The typical number of events used in the analysis is some thousands (minimum 1600) at each energy. In the MC results presented throughout this paper, the beam incident position was obtained from the MC true value. Meanwhile, the beam incident position in the data analysis is calculated from the silicon tracker. It is of no matter because the position resolution of ADAMO tracker is quite good (15 mm). The detail and performance of ADAMO tracker is found in Ref. [8]. Fig. 3 shows the flux distributions of electron beam with energies 180 GeV and 150 GeV respectively for Arm1 and Arm2, under the high gain operation. Due to the non-uniform beam profile and scan coverage, the distributions were highly nonuniform. The MC simulation was carried out for uniform beam profile, and then sampled according to the beam profile at SPS when compared with the data. The impact of this non-uniformity with respect to the uniform beam profile is discussed in Section 3.3.2. The light yield of each scintillator was measured before assembly using a 90Sr beta-ray source. The measured non-uniformities of the light yield (about 10%) were taken into account in this MC simulation. Events falling within 2 mm of the edge of the calorimeters were removed from analysis.

3.2.2. Conversion factors from ADC to deposited energy CAEN ADC V965 was used to record the deposited energy at each scintillator layer. This ADC module have two simultaneous ranges, 0–900 pC and 0–100 pC for high and low range mode

respectively, and 12 bit resolution with 15 bit dynamics. As a gain calibration of each layer, the conversion factors from ADC value to deposited energy were determined channel by channel by comparing the data of 180 GeV (Arm1) and 150 GeV (Arm2) electron beams to the MC simulations. For the first scintillation layer and beyond the 11th layer, the matching was performed using muon data since electron showers do not produce sufficiently large signals in such layers. The two histograms obtained from the beam test data (ADC value) and from MC data (deposited energy) of each layer were compared to determine the conversion factor. The MC data was scaled by horizontal and the scaled histogram was compared with the beam test data until obtaining the minimum w2 . The bins which exceed 200 events per bin were used for w2 test to reduce the effect of the tail events. For calculating the w2 of muon data, each bin added the systematic error quadrature. The reduced w2 is nearly 1 for all histograms and the typical DOF is 5–15 and 13 for electron and muon, respectively. The factor which gives the minimum w2 was taken as the conversion factor of the layer. The conversion factors calculated by electron and muon data have been determined with about 1% and 5% accuracy, respectively. The best matched histograms are shown in Figs. 4 and 5 for the electron and muon data, respectively. In these figures, the deposited energy is scaled to the number of particles with a coefficient 1 particle¼0.453 MeV that corresponds to the most probable deposited energy by a 150 GeV muon passing through a 3 mm thick plastic scintillator. For analysis of muon beam, the tail are slightly worse. However the first and the backward layers have little effect on the energy determination of electrons and photons. It will be improved for future hadron measurements.

132

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

Layer 2

600

400 200

0

500

500 400 300

500 400 300

200

200

100

100

0

1000 1500 2000 2500 particles

counts/bin

600

Layer 4

700

600 counts/bin

counts/bin

Layer 3

700

800

0

800

800

1000

0

500

0

1000 1500 2000 2500 particles

1200

0

500

1000 1500 2000 2500 particles

1200 1000 Layer 5

1000

Layer 6

Layer 7

1000

600 400

500

0

1000 1500 2000 2500 particles

Layer 8

1200

counts/bin

1000 counts/bin

400

800 600 400 200 0

500

1000 1500 2000 2500 particles

2400 2200 2000 1800 1600 1400 1200 1000 800 600 400 200 0

800 600 400 200

0

500

0

1000 1500 2000 2500 particles

0

500

1000 1500 2000 2500 particles

4000 3500

Layer 9

Layer 10

3000 counts/bin

0

1400

0

600

200

200 0

counts/bin

counts/bin

counts/bin

800 800

2500 2000 1500 1000 500

0

500

1000 1500 2000 2500 particles

0

0

500

1000 1500 2000 2500 particles

Fig. 4. The distributions of the scaled ADC value and number of particles (from MC results) in the 180 GeV electron initiated showers in case of the Arm1 20 mm calorimeter. The black dots show the data and the solid histograms indicate the MC results.

3.3. Analysis results 3.3.1. Energy determination The sum of the energies deposited within the 2nd to 13th scintillation layers is used as an estimator of the primary energy. The beam test data was converted from ADC value to deposited energy using the factor determined in Section 3.2.2. The nonuniformity of light yield of each scintillator was corrected according to the incident position. Some fraction of shower particles leak out from the edge of the calorimeters. The function of leak out is dependent on the impact position but independent from the incident energy. The two-dimensional leakage correction function was determined using the MC simulations and used for this correction. In the summation the energies deposited in layers

beyond the 11th layer are doubled because the spatial sampling of the deeper layers is half that of the first 11 layers. The distributions of total deposited energies measured in the calorimeters were compared with the MC simulations for different primary electron energies in Fig. 6. The tails in the low energy side of data are caused by the contamination of low energy particles. The data and the MC result are in good agreement in all energies except the low energy tails. The peak values of Fig. 6 were plotted as function of the beam energy in the top panels of Fig. 7. Linear fits to the beam test data and residuals from the fits are also shown. We found a deviation from linearity of less than 2% in the beam test data (filled symbols) and also found good agreement between the data and MC simulations (open symbols) except at 50 GeV.

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

133

1400 Layer 1

counts/bin

1200 1000 800 600 400 200 0

0

1

2

3

4

5

6

7

8

particles 1400

1200 Layer 11

1200

Layer 12

Layer 13

1000

1000 800 600 400 200 0

0

1

2

3 4 5 particles

6

7

600

600

400

400

200

200

0

8

800

800

counts/bin

counts/bin

counts/bin

1000

0

1

2

3 4 5 particles

6

7

0

8

0

1

2

3 4 5 particles

6

7

8

1200

1000 Layer 14

Layer 15

1000

Layer 16

1000

800 600 400 200 0

counts/bin

counts/bin

counts/bin

800 600 400

1

2

3 4 5 particles

6

7

8

0

600 400

200

0

800

200 0

1

2

3 4 5 particles

6

7

8

0

0

1

2

3 4 5 particles

6

7

8

Fig. 5. The distributions of the scaled ADC value and number of particles (from MC results) in the 150 GeV muon initiated showers in case of the Arm1 20 mm calorimeter. The indications are same as Fig. 4.

3.3.2. Energy resolution The width taken by the Gaussian fit to the deposited energy distribution in Fig. 7 is used as a definition of the energy resolution. The results for the energy resolution are shown in Figs. 8 and 9 for the high and low gain data, respectively. The energy resolution entirely depends on the shower fluctuation and noise conditions. Taking account of the actual noise conditions (ADC pedestal fluctuations) in the MC simulation, we obtained very good agreement between the experiment and MC simulations at both of the high and low gain mode. Our target of better than 5% energy resolution at above 100 GeV is achieved. Until now, we discussed the energy resolution for the particles non-uniformly entered in the calorimeters as shown in Fig. 3. Fig. 10 shows the energy resolution for uniformly injected particles calculated only by the MC simulation. The results of the uniform and non-

uniform have essentially no difference. Thanks to the correction of position dependency and the event selection described in Section 3.2.1, the calorimeters give almost uniform response to the different particle incident position.

3.3.3. Cable attenuation The results of energy calibration and performance obtained in this beam test are directly applicable to the measurements performed at LHC except for the difference of the attenuation in the cables and in the electrical noise. The latter has been carefully measured at the LHCf site on the LHC and was taken account the final LHCf performance simulations. The difference in cable attenuation was calibrated during the beam test reported here. One of the long cables used at the LHC site was brought to the

134

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

1000

800 50GeV

100GeV

700

800

counts/bin

counts/bin

600 600 400

500 400 300 200

200

100 0

1500

2000 particles

0 3000

2500

3500

4000 particles

4500

900 1000

800

150GeV

counts/bin

800

counts/bin

180GeV

700

600 400

600 500 400 300 200

200

100 0

5000

6000 particles

1000

7000

0

6000

7000 particles

8000

200GeV

counts/bin

800 600 400 200 0 6000

7000

8000

9000

particles Fig. 6. The sum of deposited energy for each incident energy in case of the Arm1 20 mm calorimeter. The black points with error show the beam test results and the solid histograms show the MC results.

beam test at SPS and used to measure 100 GeV shower signals. The ADC distribution taken with the long cable was compared with one taken at nominal cable set-up at SPS. Applying a scaling factor of 0.89170.017 (95% CL) to the nominal set-up, we could explain the data taken with the long cable (see Fig. 11 ). This factor is used to reconstruct energy measured at LHC. An independent calibration of this cable attenuation factor was carried out at the LHC site. We sent a signal produced by a pulse generator (Techtronics AFG3252) from where the detectors are installed and recorded it in the counting room. For the comparison, the same measurement was carried out in the counting room using the RG58 cable used in the beam test. The source pulse shape was generated using a plastic scintillator and a PMT used in the LHCf detector excited by fast (0.3 ns) nitrogen laser pulse using a laser generator (USHO KEN-1020). The correction factor (signal amplitude of long LHC cable divided by signal

amplitude of RG58 cable) determined in this test is 0.89970.004 and consistent with the one obtained above.

4. Conclusions The calibration and performance tests for electromagnetic showers were carried out with the final LHCf detectors using electron and muon beams at SPS in 2007. The energy resolution for 4100 GeV electron beams is less than 5%. The absolute energy scale was determined in 100–200 GeV within 72%. This energy scale is used for data taking of LHC together with the result of cable attenuation measurement. The environmental differences between SPS and LHC were also calibrated. The results are good enough for the requirements and are well understood by MC simulations. The LHCf detectors successfully started first data

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

True energy vs sum of particles (Arm1)

10000

8000

8000 7000 sum of particles

sum of particles

6000 5000 4000

6000 5000 4000

3000

3000

2000

2000

1000

1000 50

0

100 150 200 true energy [GeV]

0

250

0

50

100 150 200 true energy [GeV]

250

0

50

100 150 200 true energy [GeV]

250

10 Deviation[%]

10 Deviation[%]

SPS2007 25mm cal. SPS2007 32mm cal. MC 25mm cal. MC 32mm cal.

9000

7000

0

True energy vs sum of particles (Arm2)

10000

SPS2007 20mm cal. SPS2007 40mm cal. MC 20mm cal. MC 40mm cal.

9000

135

5 0 -5 -10 0

50

100 150 200 true energy [GeV]

5 0 -5 -10

250

Fig. 7. Correlation between the primary electron energy and the sum of the deposited energy (dE) measured by the calorimeters. The squares and triangles show the results of the 20 mm or 25 mm and 40 mm or 32 mm calorimeters, respectively. The filled symbols show the beam test results while the open symbols show the results of the MC simulations. The dotted lines in the upper panels show linear fits to the beam test results and the residuals from the fits are shown in the bottom panels. In the plots of residuals, the triangles are shifted by 5 GeV to avoid overlap. Energy Resolution (arm1)

14

SPS2007 20mm cal. with 600V SPS2007 40mm cal. with 600V MC 20mm cal. MC 40mm cal.

12 10

600V (High Gain)

8 6

10

6 4

2

2 0

50

100

150

200

0

250

600V (High Gain)

8

4

0

SPS2007 25mm cal. with 600V SPS2007 32mm cal. with 600V MC 25mm cal. MC 32mm cal.

12 Resolution [%]

Resolution [%]

Energy Resolution (arm2)

14

0

50

100

energy [GeV]

150

200

250

energy [GeV]

Fig. 8. Energy resolutions for each tower and arm. The filled (open) plots show the beam test (MC) results. The squares are for the small tower of each arm (20 mm and 25 mm for Arm1 and Arm2, respectively) and triangle plots are for the large tower of each arm (40 mm and 32 mm for Arm1 and Arm2, respectively). These are results for high gain mode.

Energy Resolution (arm1) 14

14

SPS2007 20mm cal. with 450V SPS2007 40mm cal. with 450V MC 20mm cal. MC 40mm cal.

10

450V (Low Gain)

8 6

10

6 4

2

2 0

50

100

150

energy [GeV]

200

250

450V (Low Gain)

8

4

0

SPS2007 25mm cal. with 450V SPS2007 32mm cal. with 450V MC 25mm cal. MC 32mm cal.

12 Resolution [%]

12 Resolution [%]

Energy Resolution (arm2)

0

0

50

100

150

200

250

energy [GeV]

Fig. 9. Energy resolutions for each tower and arm. The filled (open) plots show the beam test (MC) results. The circles are for the small tower of each arm (20 mm and 25 mm for Arm1 and Arm2, respectively) and stars are for the large tower of each arm (40 mm and 32 mm for Arm1 and Arm2, respectively). These are results for low gain mode.

136

T. Mase et al. / Nuclear Instruments and Methods in Physics Research A 671 (2012) 129–136

Energy resolution (arm1) 14

14

MC 20mm cal. with 600V MC 40mm cal. wuth 600V MC 20mm cal. with 450V MC 40mm cal. wuth 450V

10 8 6

MC 25mm cal. with 600V MC 32mm cal. with 600V MC 25mm cal. with 450V MC 32mm cal. with 450V

12 resolution [%]

12 resolution [%]

Energy resolution (arm2)

10 8 6

4

4

2

2

0

0 0

50

100

150

200

0

250

50

energy [GeV]

100

150

200

250

energy [GeV]

Fig. 10. Energy resolution for each tower and arm for the uniform incident. The definition of markers are same as Figs. 8 and 9 but only the MC results are plotted. These are results for high and low gain mode.

interaction models and improving understanding of the origin of high-energy cosmic-rays [10].

1200

1000

Acknowledgments The authors are grateful to the CERN SPS staff for supporting our experiment. This work is partly supported by Grant-in-Aid for Scientific research by MEXT of Japan and by the Grant-in-Aid for Nagoya University GCOE ‘‘QFPU’’ from MEXT. This work is also supported by Istituto Nazionale di Fisica Nucleare (INFN) in Italy. The receipt of Japan Society for the Promotion of Science (JSPS) Research Fellowship (HM, TM) is also acknowledged. A part of this work was performed using the computer resource provided by the Institute for the Cosmic-Ray Research (ICRR), University of Tokyo.

counts

800

600

400

200

0

200

400

600

800 ADC

1000

1200

1400

Fig. 11. The ADC distribution for 100 GeV electron induced shower. The dash line shows the distribution which was taken by the cable of SPS beam test. The solid line use the cable of LHC site. The shaded area indicate the distribution which applied the attenuation factor 0.891 for the dash line histogram.

References [1] [2] [3] [4] [5] [6]

taking at LHC at the end of 2009 and finished operation at pffiffi s ¼ 0:9 and 7 TeV proton–proton collision in the middle of July 2010. As a result of the successful calibration by the beams, LHCf already started providing fruitful data for calibrating hadron

[7] [8] [9] [10]

LHCf Technical Design Report, CERN-LHCC-2006-004. O. Adriani, et al., Journal of Instrumentation 3 (2008) S08006. R. D’Alessandro, et al., Acta Physica, Polonica B 38 (2007) 829. H. Menjo, et al., Astroparticle Physics 34 (2011) 513520. T. Sako, et al., Nuclear Instruments and Methods in Physics Research A 578 (2007) 146. M. Mizuishi, et al., Journal of Physical Society of Japan 78 (Suppl. A) (2009) 173. O. Adriani, et al., Journal of Instrumentation 5 (2010) P01012. L. Bonechi, et al., in: Proceedings of 29th International Cosmic Ray Conference, Pune, vol. 9, 2005, p. 283. K. Kasahara, EPICS web page, /http://cosmos.n.kanagawa-u.ac.jp/S. O. Adriani, et al., Physics Letters B 703 (2011) 128.