energy spectrum measured by the telescope array experiment

Energy Spectrum Measured by the Telescope Array Experiment in 1015.6 eV to 1020.3 eV Range

Toshihiro Fujii for the Telescope Array CollaborationKICP, University of ChicagoICRR, University of [email protected] 26th, 2008 1Image Credit: Brett Abernethy

mailto:[email protected]

mailto:[email protected]

2

The Energy Spectrum of Cosmic Rays above 1017.2 eV Measured by theFluorescence Detectors of the Telescope Array Experiment in Seven Years

R.U. Abbasia, M. Abeb, T. Abu-Zayyada, M. Allena, R. Azumac, E. Barcikowskia, J.W. Belza, D.R. Bergmana,S.A. Blakea, R. Cadya, M.J. Chaed, B.G. Cheone, J. Chibaf, M. Chikawag, W.R. Choh, T. Fujiii,1,∗, M. Fukushimai,j,

T. Gotok, W. Hanlona, Y. Hayashik, N. Hayashidal, K. Hibinol, K. Hondam, D. Ikedai, N. Inoueb, T. Ishiim,R. Ishimoric, H. Iton, D. Ivanova, C.C.H. Juia, K. Kadotao, F. Kakimotoc, O. Kalashevp, K. Kasaharaq, H. Kawair,S. Kawakamik, S. Kawanab, K. Kawatai, E. Kidoi, H.B. Kime, J.H. Kima, J.H. Kims, S. Kitamurac, Y. Kitamurac,

V. Kuzminp,2, Y.J. Kwonh, J. Lana, S.I. Limd, J.P. Lundquista, K. Machidam, K. Martensj, T. Matsudau,T. Matsuyamak, J.N. Matthewsa, M. Minaminok, Y. Mukaim, I. Myersa, K. Nagasawab, S. Nagatakin, T. Nakamuraw,

T. Nonakai, A. Nozatog, S. Ogiok, J. Ogurac, M. Ohnishii, H. Ohokai, K. Okii, T. Okudax, M. Onoy, A. Oshimaz,S. Ozawaq, I.H. Parkaa, M.S. Pshirkovp,ab, D.C. Rodrigueza, G. Rubtsovp, D. Ryus, H. Sagawai, N. Sakuraik,

L.M. Scottac, P.D. Shaha, F. Shibatam, T. Shibatai, H. Shimodairai, B.K. Shine, H.S. Shini, J.D. Smitha, P. Sokolskya,R.W. Springera, B.T. Stokesa, S.R. Strattona,ac, T.A. Stromana, T. Suzawab, M. Takamuraf, M. Takedai, R. Takeishii,

A. Taketaad, M. Takitai, Y. Tamedal, H. Tanakak, K. Tanakaae, M. Tanakau, S.B. Thomasa, G.B. Thomsona,P. Tinyakovaf,p, I. Tkachevp, H. Tokunoc, T. Tomidaag, S. Troitskyp, Y. Tsunesadak, K. Tsutsumic, Y. Uchihoriah,

S. Udol, F. Urbanaf, G. Vasiloffa, T. Wonga, R. Yamanek, H. Yamaokau, K. Yamazakiad, J. Yangd, K. Yashirof,Y. Yonedak, S. Yoshidar, H. Yoshiiai, R. Zollingera, Z. Zundela

aHigh Energy Astrophysics Institute and Department of Physics and Astronomy, University of Utah, Salt Lake City, Utah, USAbThe Graduate School of Science and Engineering, Saitama University, Saitama, Saitama, Japan

cGraduate School of Science and Engineering, Tokyo Institute of Technology, Meguro, Tokyo, JapandDepartment of Physics and Institute for the Early Universe, Ewha Womans University, Seodaaemun-gu, Seoul, Korea

eDepartment of Physics and The Research Institute of Natural Science, Hanyang University, Seongdong-gu, Seoul, KoreafDepartment of Physics, Tokyo University of Science, Noda, Chiba, JapangDepartment of Physics, Kinki University, Higashi Osaka, Osaka, JapanhDepartment of Physics, Yonsei University, Seodaemun-gu, Seoul, Korea

iInstitute for Cosmic Ray Research, University of Tokyo, Kashiwa, Chiba, JapanjKavli Institute for the Physics and Mathematics of the Universe (WPI), Todai Institutes for Advanced Study, the University of Tokyo, Kashiwa,

Chiba, JapankGraduate School of Science, Osaka City University, Osaka, Osaka, Japan

lFaculty of Engineering, Kanagawa University, Yokohama, Kanagawa, JapanmInterdisciplinary Graduate School of Medicine and Engineering, University of Yamanashi, Kofu, Yamanashi, Japan

nAstrophysical Big Bang Laboratory, RIKEN, Wako, Saitama, JapanoDepartment of Physics, Tokyo City University, Setagaya-ku, Tokyo, Japan

pInstitute for Nuclear Research of the Russian Academy of Sciences, Moscow, RussiaqAdvanced Research Institute for Science and Engineering, Waseda University, Shinjuku-ku, Tokyo, Japan

rDepartment of Physics, Chiba University, Chiba, Chiba, JapansDepartment of Physics, School of Natural Sciences, Ulsan National Institute of Science and Technology, UNIST-gil, Ulsan, Korea

tHigh Energy Astrophysics Institute and Department of Physics and Astronomy, University of Utah, Salt Lake City, Utah, USAuInstitute of Particle and Nuclear Studies, KEK, Tsukuba, Ibaraki, Japan

vHigh Energy Astrophysics Institute and Department of Physics and Astronomy, University of Utah, Salt Lake City, Utah, USAwFaculty of Science, Kochi University, Kochi, Kochi, Japan

xDepartment of Physical Sciences, Ritsumeikan University, Kusatsu, Shiga, JapanyDepartment of Physics, Kyushu University, Fukuoka, Fukuoka, Japan

zEngineering Science Laboratory, Chubu University, Kasugai, Aichi, JapanaaDepartment of Physics, Sungkyunkwan University, Jang-an-gu, Suwon, Korea

abSternberg Astronomical Institute, Moscow M.V. Lomonosov State University, Moscow, RussiaacDepartment of Physics and Astronomy, Rutgers University – The State University of New Jersey, Piscataway, New Jersey, USA

adEarthquake Research Institute, University of Tokyo, Bunkyo-ku, Tokyo, JapanaeGraduate School of Information Sciences, Hiroshima City University, Hiroshima, Hiroshima, Japan

afService de Physique Theorique, Universite Libre de Bruxelles, Brussels, BelgiumagDepartment of Computer Science and Engineering, Shinshu University, Nagano, Nagano, Japan

ahNational Institute of Radiological Science, Chiba, Chiba, JapanaiDepartment of Physics, Ehime University, Matsuyama, Ehime, Japan

Abstract

The Telescope Array (TA) experiment is the largest detector to observe ultra-high-energy cosmic rays in the northernhemisphere. The fluorescence detectors at southern two stations of TA are newly constructed and have now completedseven years of steady operation. One advantage of monocular analysis of the fluorescence detectors is a lower energythreshold for cosmic rays than that of other techniques like stereoscopic observations or coincidences with the sur-face detector array, allowing the measurement of an energy spectrum covering three orders of magnitude in energy.Analyzing data collected during those seven years, we report the energy spectrum of cosmic rays covering a broadrange of energies above 1017.2 eV measured by the fluorescence detectors and a comparison with previously publishedresults.

Keywords: cosmic rays; ultra-high energy; fluorescence detector; energy spectrum; ankle; GZK cutoff

Preprint submitted to Astroparticle Physics October 23, 2015


























Abstract





























Abstract




1. Introduction1

The energy spectrum of cosmic rays has been mea-2

sured at energies from 108 eV to beyond 1020 eV in the3

century elapsed since the discovery of cosmic rays using4

a balloon measurement [1]. In the energy range above5

1011 eV, the cosmic-ray flux is not affected by solar ac-6

tivity; the spectrum follows the power-law structure of7

E−γ. The spectral index, γ, is ∼ 2.7 up to 1015.5 eV,8

where it changes to γ ∼ 3.0. The spectrum softens9

slightly to γ ∼ 3.3 around 1017 eV, and hardens again to10

γ ∼ 2.7 just below 1019 eV. These three spectral-index11

discontinuities are called, in order of increasing energy,12

the “knee,” the “second knee,” and the “ankle.” Cosmic13

rays above 1018 eV are called ultra-high-energy cos-14

mic rays (UHECRs). If assumed UHECRs are proton-15

dominated, UHE protons above 1019.7 eV expects to in-16

teract with cosmic microwave background radiation via17

pion production, and the mean free path of UHE pro-18

tons will be significantly reduced. As a result, the en-19

ergy spectrum above 1019.7 eV will be suppressed—the20

so-called GZK cutoff [2, 3].21

Since UHECRs are the most energetic particles in22

the universe, their origins are ostensibly related to ex-23

tremely energetic astronomical phenomena or other ex-24

otic phenomena, such as the decay or annihilation of25

super-heavy relic particles created in an early phase of26

the development of the universe. A transition of cos-27

mic ray origins from galactic to extragalactic sources28

might be happened around 1017 eV. Therefore, precise29

measurements of the energy spectrum covering a broad30

energy range and its structure are of utmost impor-31

tance to understand the origin and propagation of cos-32

mic rays. However, at the highest energies (∼ 1020 eV),33

the cosmic-ray flux becomes quite low, only one parti-34

cle per century per km2 area. A huge effective area and35

a large exposure are essential to measure cosmic rays36

with such energies.37

2. Telescope Array experiment38

The Telescope Array (TA) experiment is the largest39

cosmic-ray detector in the northern hemisphere [4]. TA40

consists of 507 surface detectors (SDs) deployed on a41

square grid with 1.2-km spacing, covering an effective42

area of about 700 km2 [5]; the SD array is overlooked by43

38 fluorescence detectors (FDs) at three locations [6].44

∗Corresponding authorEmail address: [email protected] (T. Fujii)

1Now at University of Chicago, USA2Deceased

One FD station, called “Middle Drum” (MD) and lo-45

cated northwest of the SD array, consists of 14 FDs pre-46

viously used in the High Resolution Fly’s Eye (HiRes)47

experiment [7]. Two other stations at the array’s south-48

east and southwest, respectively called “Black Rock49

Mesa” (BRM) and “Long Ridge” (LR), each consist of50

12 newly designed FDs [6].51

Because the duty cycle of the SDs is approximately52

100%, they have the greatest exposure, and hence the53

strongest statistics, of any TA analysis. The SD mea-54

surement of the cosmic-ray energy spectrum above55

1018.2 eV, including a suppression above 1019.7 eV con-56

sistent with the GZK cutoff, has been previously re-57

ported [8]. The energy spectrum observed by the MD58

station and a comparison with the HiRes experiment59

were reported after the first three years of operation [9].60

The energy spectrum above 1018 eV analyzing data col-61

lected during three-and-a-half years of operation by the62

FDs was previously reported [10]. In this paper, we re-63

port an update on the energy spectrum with double the64

statistics from seven years of observation, and extend65

the range of energies down to 1017.2 eV.66

The field of view of the BRM and LR stations’ FDs67

is approximately 18 in azimuth and 15.5 in elevation.68

At each station, 12 FDs are arranged in two layers of69

6 telescopes each. The upper-layer telescopes are ori-70

ented to view the sky at elevations from 3 to 18.5, and71

the lower layer observes higher elevations, from 17.572

to 33, for a combined elevation coverage of 3 ∼ 33.73

The directions of the telescopes are fanned out covering74

a combined 108 in azimuth at each station, with BRM75

looking generally west and northwest, and LR looking76

east and northeast. The BRM and LR FDs have spher-77

ical mirrors with a diameter of 3.3 m, which are com-78

posed of 18 hexagonal segments. The camera design79

of these FD consists of 256 hexagonal photo-multiplier80

tubes (PMTs) arranged in a 16×16 honeycomb array, in-81

stalled at the focal plane of the spherical mirror. The an-82

ode voltage of each PMT is digitized with a 12-bit, 40-83

MHz flash analog-to-digital converter (FADC). Sets of84

four adjacent time bins are summed to obtain an equiv-85

alent sampling rate of 10 MHz with a 14-bit dynamic86

range. The trigger electronics can select a track pat-87

tern of triggered PMTs in real time to reduce accidental88

noise [11].89

The absolute gain of a few standard PMTs (2 or90

3 installed in each camera) is measured by CRAYS91

(Calibration using RAYleigh Scattering) in the labora-92

tory [12]. The gain of the other PMTs in the same cam-93

era can be monitored relative to the standard PMTs by94

comparing the flash intensities of Xe lamps which are95

installed in the center of each mirror [13, 14]. In or-96

2Japan, USA, Korea, Russia, Belgium

3

Largest cosmic ray detector in the Northern hemisphere ~ 700 km2 at Utah, USAFluorescence detector + Surface detector array

PMT

16×16PMTs

Fluorescence Detector at BRM and LR stationsSpherical segment mirror (6.8 m2) + 256 Photomultiplier tube(PMTs)/camera, 12 newly designed telescopes

Surface Detector Array507 Scintillator, 1.2 km spacing

Fluorescence detector at MD stationRefurbished from HiRes experiment,Spherical mirror 5.2 m2,256 PMTs/camera,14 telescopes

Telescope Array Experiment (TA)

Telescope Array Experiment (TA)

4

Surface detector array (SD) Fluorescence detector (FD)

7 Years Steadily Operation

5

Surface detector array (SD) Fluorescence detector (FD)~100% duty operation Clear moonless night

~10% duty operation

Mar’08 Jun’15 Jun’15Nov’07

... to Observe Extensive Air Shower (EAS) induced by Ultra-High Energy Cosmic Ray (UHECR)

6Image credit: ASPERA_Novapix_L.Bret

Observed UHECR Event

7

Surface detector array (SD) Fluorescence detector (FD)Observe lateral density distribution

Charge density at 800 m, S800 as energy indicator

800 m

Azimuth angle [degree]40 50 60 70 80

Elev

atio

n an

gle

[deg

ree]

0

5

10

15

20

25

30

35

0 0 2 10 5 40 7 90 10 160 12 250 15 360 17 490 20 640 22 810 25 1000 28 1210 30 1440 33 1690 35 1960 38 2250 40 2560 43 2890 45 3240 48 3610

s) #p.e.µTime(

)2Slant Depth (g/cm500 600 700 800 900

Num

ber o

f Pho

to-E

lect

rons

0

50

100

150

200

250

300

350 /ndf = 235.0/329 (0.7)2χDataFluorescenceDirect CherenkovRayleigh Scatterd C.Mie Scattered C.

Figure 1: UHECR event observed by the fluorescence detector. The top figure shows the PMT

pointing directions and the brightness of signal (point size) and timing (point color). The

bottom figure shows a sum of selected PMT waveforms as a function of slant depth (black

plot), compared with the reconstructed result by the inverse Monte Carlo reconstruction

(histograms). The inverse Monte Carlo method can reproduce the obtained signal at the

camera.

20


Ele

va

tio

n a

ng

le [

deg

ree]

0

5

10

15

20

25

30

35

0 0 2 10 5 40 7 90 10 160 12 250 15 360 17 490 20 640 22 810 25 1000 28 1210 30 1440 33 1690 35 1960 38 2250 40 2560 43 2890 45 3240 48 3610

s) #p.e.µTime(

)2Slant Depth (g/cm500 600 700 800 900

Nu

mb

er o

f P

hoto

-Ele

ctro

ns

0

50

100

150

200

250

300


Figure 1: UHECR event observed by the fluorescence detector. The top figure shows the PMT

pointing directions and the brightness of signal (point size) and timing (point color). The

bottom figure shows a sum of selected PMT waveforms as a function of slant depth (black

plot), compared with the reconstructed result by the inverse Monte Carlo reconstruction

(histograms). The inverse Monte Carlo method can reproduce the obtained signal at the

camera.

20

Observe longitudinal development

Reconstruct calorimetric energy

Energy Estimation by TA SDA look up table made from Monte Carlo simulation

Event energy ETBL = function of S800 and zenith angle, sec(θ)

8

SD Energy 1/2

•  A look-up table made from the Monte-Carlo •  Event energy (ETBL) = function of reconstructed S800 and sec(θ) •  Energy reconstruction "# interpolation between S800 vs sec(θ) contours of

constant values of ETBL

•  The overall energy scale locked to the fluorescence detector

Y =

log 1

0 [S

800

(VE

M m

-2)]

X = Secant of zenith angle

Z =

Col

or =

log 1

0(E

/eV

)

ETBL = f[S800,sec(θ)]

TA Spectrum Summary Dmitri Ivanov

→East [1200m] 1 2 3 4 5 6 7 8 9 10 11 12

→N

orth

[120

0m]

12

13

14

15

16

17

18

19

20

21

22

7

7.5

8

8.5

9

S]µTime, [4 2008/06/25 19:45:52.588640 UTC

Array E

dge

- axis [m]uDistance along the -4000 -2000 0 2000

S]µ

Tim

e [

24

26

28

30

32

34

36

38

40

Lateral distance [m]310

]2

Sig

nal [

VEM

/ m

-110

1

10

210

Figure 2: A typical high energy event seen by the TA SD. Left: each circle represents a counter, positionedat the center of the circle, the area of the circle is logarithmically proportional to the counter pulse height,and the counter time is denoted by the color. The arrow represents the projection of the shower axis ontothe ground, denoted by u, and it is bisected by the perpendicular line at the location of the shower core.Middle: counter time versus distance from the shower core along the u direction, which is the shower axisprojected on the ground. Points with error bars are counter times, solid curve is the time expected by the fitfor counters lying on the u axis, dashed and dotted lines are the fit expectation times for the counters thatare correspondingly 1.5 and 2.0 km off the u axis. Right: Lateral distribution profile fit to the AGASA LDF.Vertical axis is the signal density in Vertical Equivalent Muon (VEM) per square meter units and horizontalaxis is the lateral distance from the shower core. 1 VEM is 2.05 MeV for the TA SD scintillator.

/ dof2χ0 1 2 3 4 5 6 7 8 9 100

100

200

300

400

500

600

700

800

900

θsec 1 1.1 1.2 1.3 1.4 1.5

) ]-2

[ S8

00 /

(VEM

m10

log

0

0.5

1

1.5

2

2.5

3

18

18.5

19

19.5

20

20.5(E/eV)

10log

(E/eV)10

TA SD, log18 18.5 19 19.5 20

(E/e

V)10

TA H

ybrid

, lo

g

18

18.5

19

19.5

20

Figure 3: Right: TA SD data and MC comparison of the lateral distribution fit c2 per degree of freedom.Points represent the data and solid line is the MC. Middle: Energy as function of reconstructed S800 andsec(q) made from the CORSIKA MC. Z-axis described by color represents the true (MC generated) valuesof energy. Right: TA SD reconstructed energies normalized by 1/1.27 and compared to the TA Hybrid resultsof BR, LR, and MD simultaneously. Superimposed 45o line shows no significant non-linearities.

4. TALE FD Bridge

The TALE bridge spectrum uses data collected in 2013/09/06 to 2014/01/09 period. Figure 4shows the resolution and exposure of the TALE bridge spectrum analysis using dotted lines. Theanalysis uses geometry reconstructed in monocular mode and both fluorescence and Cerenkovcomponents of light produced by particles of the shower. Details of the TALE bridge analysis aredescribed in [13].

4

ETBL is rescaled by the FD reconstructed energy to estimate final energy of SD, ESD,final

ESD,final = ETBL/1.27,

<20% resolution above 1019 eV

2008/05/11-2013/05/04

Parameter Proton Ironp1 3.55 ± 0.06 3.49 ± 0.15p2 17.12 ± 0.01 17.22 ± 0.01p3 0.68 ± 0.03 0.63 ± 0.05p4 17.56 ± 0.07 17.40 ± 0.13p5 0.29 ± 0.03 0.40 ± 0.01log10 Eb 18.27 ± 0.09 17.87 ± 0.03

Table 2: The fit parameters for aperture assuming proton and ironprimaries.

where252

γ =1 − exp

(− (log10 Eb − p2

)/p3

)

1 − exp(− (

log10 Eb − p4)/p5

) (5)253

and Eb is the energy (in eV) at the break. The best-fit254

values are described in Table 2.255

The aperture assuming the HiRes/MIA proton frac-256

tion, AΩ f , was estimated by the following formula:257

AΩ f = AΩP [R + f · (1 − R)

], (6)258

where f is the proton fraction and R ≡ AΩFe/AΩP is259

the ratio of the iron and proton best-fit apertures. The260

dependence of the aperture on primary species is most261

evident in the low-energy region, but becomes negligi-262

ble at high energies.263

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

sr]

2A

pert

ure

[km

-110

1

10

210

310

Proton

Iron

HiRes/MIA

Figure 4: The combined aperture of the BRM and LR FD stationsin monocular mode evaluated by MC simulations for primary com-positions of pure protons (red dotted line), pure iron (blue dashedline), and the proton fraction measured by the HiRes/MIA experiment(black solid line).

5. Data Analysis264

We analyze data collected by the BRM and LR FD265

stations from 1 January 2008 to 1 December 2014, cor-266

responding to nearly seven years of observation. The267

total live time (subtracting the dead time for data acqui-268

sition) is 6960 hours at BRM and 5850 hours at LR. A269

cloud cut ensures that we only analyze data collected270

under weather conditions that can be accurately mod-271

eled in our MC simulation. This cut is applied by in-272

terpreting the visually recorded code at the MD FD sta-273

tion because it has the most coverage in this period, and274

we confirmed its consisntecy with the method described275

in Sec. 2. After the cloud cut, the live time is 4100276

hours at BRM and 3470 hours at LR, so that 41% of277

our data period was excluded by the cloud cut. The live278

time of simultaneous BRM and LR observation is 2870279

hours. Analyzing data using the monocular analysis280

under the same quality cuts, 28269 shower candidates281

above 1017.2 eV are obtained as shown in Figure 5. The282

number of events passing each selection in sequence is283

summarized in Table 1.284

log (E (eV))17 17.5 18 18.5 19 19.5 20 20.5

Num

ber

of E

vent

s

1

10

210

310Data (Jan/2008-Dec/2014)

Figure 5: Energy distribution of reconstructed showers from sevenyears of data.

5.1. Data/MC Comparison285

To further ensure the reliability of our analysis, the286

distributions of several parameters obtained from recon-287

struction of the observed data are compared with the288

predictions estimated from MC simulations using the289

QGSJetII-03 model. The MC simulations are weighted290

according to the energy spectrum measured by the sur-291

face detectors [7]. The comparisons of a variety of292

parameters are shown in Figs. 6–8 within three en-293

ergy ranges: 1017.2−18.0 eV, 1018.0−19.0 eV, and above294

1019.0 eV. In each figure, the black plot indicates the295

distribution observed at the BRM or LR station, and the296

red or blue histograms indicate the expected distribu-297

tions estimated from MC simulations at the same sta-298

tion for respective compositions of pure protons or pure299

iron. The distributions of each MC parameter is normal-300

ized to the number of data observations. These plots are301

in reasonable agreements between data and MC simula-302

tions in all energy ranges.303

5

Exposure from Monte Carlo Zenith Angle LDF χ2/DOF Pulse Height

6300 km2 sr yr

TA SD 2008/05/11-2015/05/11

!  Detailed Monte Carlo used for exposure calculation in all measurements of TA


Elev

atio

n an

gle

[deg

ree]

0

5

10

15

20

25

30

35

)2Slant Depth (g/cm400 500 600 700 800 900 1000

Num

ber o

f Pho

to-E

lect

rons

0

50

100

150

200

250

300


Figure 1: UHECR event observed by the fluorescence detector. Thetop figure shows the PMT pointing directions and the brightness ofsignal (point size) and timing (point color). The bottom figure showsa sum of selected PMT waveforms as a function of slant depth (blackplot), compared with the reconstructed result by the inverse MonteCarlo reconstruction (histograms). The inverse Monte Carlo methodcan reproduce the obtained signal at the camera.

with a 1.0 km scale height, corresponding to a verti-150

cal aerosol optical depth (VAOD) of 0.034. The ab-151

solute fluorescence yield is obtained from a measure-152

ment by Kakimoto et al. [20], with a wavelength spec-153

trum adopted from the result of the FLASH collabora-154

tion [21]. The emission of fluorescence photons is pro-155

portional to the longitudinal projection of atmospheric156

energy deposit, with a lateral distribution described by157

the NKG function [22, 23].158

3.2. Air Shower Model159

We use the CORSIKA software [24] to simulate160

the development of UHECRs using proton and iron161

primary particles, each according to five types of162

hadronic-interaction model: QGSJet01C, QGSJetII-03,163

QGSJetII-04, Sibyll 2.1 and Epos-LHC. CORSIKA also164

allows us to estimate the fraction of a primary’s en-165

ergy that is not deposited and does not contribute to the166

calorimetric energy. This missing energy for proton or167

iron primaries is parameterized by168

EEcal= a1 + a2 log10

Ecal

eV+ a3

(log10

Ecal

eV

)2(2)169

where E is the primary energy and Ecal is calorimetric170

energy. Using the QGSJetII-03 model, the fit parame-171

ters (a1, a2, a3) are (3.083, -0.1947, 0.004695) for pro-172

tons, and (4.051, -0.2757, 0.006390) for iron. The ob-173

tained missing energy is 8% – 13% for protons depend-174

ing on energy; the difference between proton and iron175

is at most ∼ 6% above 1017.2 eV. Our missing energy176

correction combines the proton and iron results, assum-177

ing an energy-dependent proton fraction given by the178

HiRes/MIA experiment [25] as shown in Figure 2. The179

uncertainty of the proton fraction was evaluated as 20%.180

The HiRes/MIA result indicates ∼50% proton and iron181

primaries at 1017 eV, increasing to ∼90% for energies182

above 1018 eV.183

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

Prot

on fr

actio

n

0

0.2

0.4

0.6

0.8

1

HiRes/MIA

20% uncert.

Figure 2: Proton fraction reported by the HiRes/MIA experiment andits uncertainty [25]. The uncertainty of the proton fraction is indicatedby the dashed line. If the fraction is larger than 1, purely proton isassumed.

3.3. Quality Cuts184

The geometries of some showers, e.g., those that185

are too short or faint are difficult to reconstruct accu-186

rately. Thus, we apply quality cuts to select only well-187

reconstructed events in our analysis: the number of hit188

PMTs is larger than 10; the track length is larger than189

10; the time extent is larger than 2 µs; and the depth190

of EAS maximum, Xmax, is within the station’s field of191

view, falling between the first and the last visible depths192

(Xstart and Xend, respectively). To avoid Cherenkov-193

light contamination, we require an angle on the shower-194

detector plane less than 120, and a minimum viewing195

angle greater than 20. Since our Monte Carlo (MC)196

simulations are generated with zenith angles up to 65197

and core locations in a circle of 35 km radius from the198

CLF, the last two cuts are required to avoid contamina-199

tion of the data with events from beyond the generated200

range, due to the monocular geometry resolution. A full201

set of the quality cuts are summarized in Table 1.202

3

Aperture, Exposure CalculationDetailed Monte Carlo used for aperture calculation in all measurement of TA. Exposure = Aperture × live-time.FD aperture needs to assume mass composition.Use the proton fraction measured by the HiRes/MIA experiment with 20% uncertainty [Astrophys. J. 622 (2005) 910].

9

Aperture difference on FD

Proton and Iron model

Energy Spectrum from TA FD and SD

10

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×Fl

ux

-110

1

10

Telescope Array 2015 Preliminary

TA FD

TA SD

Ankle at logE=18.72±0.02,

Suppression at logE=19.78±0.05

Item Uncertainty

Fluorescence 11%

Atmosphere 11%

Calibration 10%

Reconstruction 9%

Total 21%


Ele

vatio

n an

gle

[deg

ree]

0

5

10

15

20

25

30

35

)2Slant Depth (g/cm400 500 600 700 800 900 1000

Num

ber o

f Pho

to-E

lect

rons

0

50

100

150

200

250

300


Figure 1: UHECR event observed by the fluorescence detector. Thetop figure shows the PMT pointing directions and the brightness ofsignal (point size) and timing (point color). The bottom figure showsa sum of selected PMT waveforms as a function of slant depth (blackplot), compared with the reconstructed result by the inverse MonteCarlo reconstruction (histograms). The inverse Monte Carlo methodcan reproduce the obtained signal at the camera.

with a 1.0 km scale height, corresponding to a verti-150

cal aerosol optical depth (VAOD) of 0.034. The ab-151

solute fluorescence yield is obtained from a measure-152

ment by Kakimoto et al. [20], with a wavelength spec-153

trum adopted from the result of the FLASH collabora-154

tion [21]. The emission of fluorescence photons is pro-155

portional to the longitudinal projection of atmospheric156

energy deposit, with a lateral distribution described by157

the NKG function [22, 23].158

3.2. Air Shower Model159

We use the CORSIKA software [24] to simulate160

the development of UHECRs using proton and iron161

primary particles, each according to five types of162

hadronic-interaction model: QGSJet01C, QGSJetII-03,163

QGSJetII-04, Sibyll 2.1 and Epos-LHC. CORSIKA also164

allows us to estimate the fraction of a primary’s en-165

ergy that is not deposited and does not contribute to the166

calorimetric energy. This missing energy for proton or167

iron primaries is parameterized by168

EEcal= a1 + a2 log10

Ecal

eV+ a3

(log10

Ecal

eV

)2(2)169

where E is the primary energy and Ecal is calorimetric170

energy. Using the QGSJetII-03 model, the fit parame-171

ters (a1, a2, a3) are (3.083, -0.1947, 0.004695) for pro-172

tons, and (4.051, -0.2757, 0.006390) for iron. The ob-173

tained missing energy is 8% – 13% for protons depend-174

ing on energy; the difference between proton and iron175

is at most ∼ 6% above 1017.2 eV. Our missing energy176

correction combines the proton and iron results, assum-177

ing an energy-dependent proton fraction given by the178

HiRes/MIA experiment [25] as shown in Figure 2. The179

uncertainty of the proton fraction was evaluated as 20%.180

The HiRes/MIA result indicates ∼50% proton and iron181

primaries at 1017 eV, increasing to ∼90% for energies182

above 1018 eV.183

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

Prot

on fr

actio

n

0

0.2

0.4

0.6

0.8

1

HiRes/MIA

20% uncert.

Figure 2: Proton fraction reported by the HiRes/MIA experiment andits uncertainty [25]. The uncertainty of the proton fraction is indicatedby the dashed line. If the fraction is larger than 1, purely proton isassumed.

3.3. Quality Cuts184

The geometries of some showers, e.g., those that185

are too short or faint are difficult to reconstruct accu-186

rately. Thus, we apply quality cuts to select only well-187

reconstructed events in our analysis: the number of hit188

PMTs is larger than 10; the track length is larger than189

10; the time extent is larger than 2 µs; and the depth190

of EAS maximum, Xmax, is within the station’s field of191

view, falling between the first and the last visible depths192

(Xstart and Xend, respectively). To avoid Cherenkov-193

light contamination, we require an angle on the shower-194

detector plane less than 120, and a minimum viewing195

angle greater than 20. Since our Monte Carlo (MC)196

simulations are generated with zenith angles up to 65197

and core locations in a circle of 35 km radius from the198

CLF, the last two cuts are required to avoid contamina-199

tion of the data with events from beyond the generated200

range, due to the monocular geometry resolution. A full201

set of the quality cuts are summarized in Table 1.202

3

Uncertainty attributed to Proton Fraction Assumption

11

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×Fl

ux

1

10

0.04±=-3.26 1γ

0.04±)=18.62 ankle

log(E

0.06±=-2.65 2γ

/ndf=18.6/19 (1.0)2χ

Figure 10: Energy spectrum observed by the BRM and LR fluores-cence detector stations.

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×Fl

ux 1

10

HiRes/MIA

Composition uncert.

Proton

Iron

Figure 11: Systematic uncertainty attributed to the assumed protonfraction. The red open-square is the flux estimated by an aperture as-suming primary protons, and the blue triangle assumes primary iron.The square bracket corresponds to the uncertainty caused by 20% pro-ton fraction uncertainty.

scale.408

7. Conclusions409

We have measured the cosmic-ray energy spectrum410

covering three orders of magnitude at energies above411

1017.2 eV using the monocular analysis of data taken412

during the first seven years of operation by the new flu-413

orescence detectors of the Telescope Array experiment.414

The obtained spectrum has an overall broken-power-law415

structure, with an obvious spectral-index break at an en-416

ergy of log10(Eankle/eV) = 18.62 ± 0.04, corresponding417

to the ankle. The structure is in good agreement with418

the spectra reported using the TA surface detectors and419

by HiRes-II.420

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×Fl

ux

-110

1

10

TA (this work)TA MDTA SD

HiRes-IHiRes-IIAuger ICRC2013

Figure 12: Energy spectrum compared with results reported byHiRes [8], Auger [27] and other detectors within TA [7, 28].

Acknowledgments421

The Telescope Array experiment is supported422

by the Japan Society for the Promotion of Sci-423

ence through Grants-in-Aid for Scientific Research424

on Specially Promoted Research (21000002) “Ex-425

treme Phenomena in the Universe Explored by High-426

est Energy Cosmic Rays” and for Scientific Re-427

search (19104006), and the Inter-University Re-428

search Program of the Institute for Cosmic Ray Re-429

search; by the U.S. National Science Foundation430

awards PHY-0307098, PHY-0601915, PHY-0649681,431

PHY-0703893, PHY-0758342, PHY-0848320, PHY-432

1069280, PHY-1069286, PHY-1404495 and PHY-433

1404502; by the National Research Foundation of Ko-434

rea (2007-0093860, R32-10130, 2012R1A1A2008381,435

2013004883); by the Russian Academy of Sciences,436

RFBR grants 11-02-01528a and 13-02-01311a (INR),437

IISN project No. 4.4502.13; and Belgian Science Pol-438

icy under IUAP VII/37 (ULB). The foundations of Dr.439

Ezekiel R. and Edna Wattis Dumke, Willard L. Eccles,440

and George S. and Dolores Dore Eccles all helped with441

generous donations. The State of Utah supported the442

project through its Economic Development Board, and443

the University of Utah through the Office of the Vice444

President for Research. The experimental site became445

available through the cooperation of the Utah School446

and Institutional Trust Lands Administration (SITLA),447

U.S. Bureau of Land Management, and the U.S. Air448

Force. We also wish to thank the people and the of-449

ficials of Millard County, Utah for their steadfast and450

warm support. We gratefully acknowledge the contri-451

butions from the technical staffs of our home institu-452

tions. An allocation of computer time from the Center453

for High Performance Computing at the University of454

Utah is gratefully acknowledged.455

8

Parameter Proton Ironp1 3.55 ± 0.06 3.49 ± 0.15p2 17.12 ± 0.01 17.22 ± 0.01p3 0.68 ± 0.03 0.63 ± 0.05p4 17.56 ± 0.07 17.40 ± 0.13p5 0.29 ± 0.03 0.40 ± 0.01log10 Eb 18.27 ± 0.09 17.87 ± 0.03

Table 2: The fit parameters for aperture assuming proton and ironprimaries.

where252

γ =1 − exp

(− (log10 Eb − p2

)/p3

)

1 − exp(− (

log10 Eb − p4)/p5

) (5)253

and Eb is the energy (in eV) at the break. The best-fit254

values are described in Table 2.255

The aperture assuming the HiRes/MIA proton frac-256

tion, AΩ f , was estimated by the following formula:257

AΩ f = AΩP [R + f · (1 − R)

], (6)258

where f is the proton fraction and R ≡ AΩFe/AΩP is259

the ratio of the iron and proton best-fit apertures. The260

dependence of the aperture on primary species is most261

evident in the low-energy region, but becomes negligi-262

ble at high energies.263

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

sr]

2A

pert

ure

[km

-110

1

10

210

310

Proton

Iron

HiRes/MIA

Figure 4: The combined aperture of the BRM and LR FD stationsin monocular mode evaluated by MC simulations for primary com-positions of pure protons (red dotted line), pure iron (blue dashedline), and the proton fraction measured by the HiRes/MIA experiment(black solid line).

5. Data Analysis264

We analyze data collected by the BRM and LR FD265

stations from 1 January 2008 to 1 December 2014, cor-266

responding to nearly seven years of observation. The267

total live time (subtracting the dead time for data acqui-268

sition) is 6960 hours at BRM and 5850 hours at LR. A269

cloud cut ensures that we only analyze data collected270

under weather conditions that can be accurately mod-271

eled in our MC simulation. This cut is applied by in-272

terpreting the visually recorded code at the MD FD sta-273

tion because it has the most coverage in this period, and274

we confirmed its consisntecy with the method described275

in Sec. 2. After the cloud cut, the live time is 4100276

hours at BRM and 3470 hours at LR, so that 41% of277

our data period was excluded by the cloud cut. The live278

time of simultaneous BRM and LR observation is 2870279

hours. Analyzing data using the monocular analysis280

under the same quality cuts, 28269 shower candidates281

above 1017.2 eV are obtained as shown in Figure 5. The282

number of events passing each selection in sequence is283

summarized in Table 1.284

log (E (eV))17 17.5 18 18.5 19 19.5 20 20.5

Num

ber

of E

vent

s

1

10

210

310Data (Jan/2008-Dec/2014)

Figure 5: Energy distribution of reconstructed showers from sevenyears of data.

5.1. Data/MC Comparison285

To further ensure the reliability of our analysis, the286

distributions of several parameters obtained from recon-287

struction of the observed data are compared with the288

predictions estimated from MC simulations using the289

QGSJetII-03 model. The MC simulations are weighted290

according to the energy spectrum measured by the sur-291

face detectors [7]. The comparisons of a variety of292

parameters are shown in Figs. 6–8 within three en-293

ergy ranges: 1017.2−18.0 eV, 1018.0−19.0 eV, and above294

1019.0 eV. In each figure, the black plot indicates the295

distribution observed at the BRM or LR station, and the296

red or blue histograms indicate the expected distribu-297

tions estimated from MC simulations at the same sta-298

tion for respective compositions of pure protons or pure299

iron. The distributions of each MC parameter is normal-300

ized to the number of data observations. These plots are301

in reasonable agreements between data and MC simula-302

tions in all energy ranges.303

5

Change the proton fraction by the uncertainty of HiRes/MIA of ±20%, and recalculate aperture of FD.

Calculated the energy spectrum with those aperture.

Comparison with Other Measurements

12

(E (eV))10

log17 17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×Fl

ux

-110

1

10

TA FD

TA MD

TA SD

HiRes-IHiRes-IIAuger ICRC2013

Consistent with the HiRes result in a broad energy range

Consistent with TA MD result

8.5% difference with Auger result around ankle, however consistent within systematics uncertainty.

(E (eV))10

log17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×Fl

ux

-110

1

10Preliminary

TA Combined 2015

Auger ICRC 2013 +8.5%

Discrepancy on the SuppressionEven if we correct the energy difference, the suppression shows large discrepancy above 1019.3 eV.

Possible reasons of discrepancy: fluorescence yield, atmospheric model, missing energy correction, detector: scintillator or water-tank, Northern/Southern hemisphere.

TA×4 : fourfold statistics at the suppression

13

187187cm

124cm LeftModule RightModule

Practical implementation

Two modules in one box per station, readout by one PMT, area ~4 m2

12

10 cm10cm

1cm

4.2. THE SCINTILLATOR DETECTOR 69

Figure 4.12: 3D view of the SSD module with the support bars. The bars are connected to the tankusing lifting lugs present in the tank structure.

4.2.7 Calibration and control system

The SSD calibration is based on the signal of a minimum ionizing particle going through thedetector, a MIP. Since this is a thin detector, the MIP will not necessarily be well separatedfrom the low energy background but, being installed on top of the WCD, a cross triggercan be used to remove all of the background. About 40% of the calibration triggers of theWCD produce a MIP in the SSD. The statistics of calibration events recorded in a minute, thenormal WCD calibration period, are therefore enough to obtain a precise measurement of theMIP. Figure 4.13 shows the MIP calibration histogram from a 2 m2 test module, obtained inone minute of acquisition. The MIP is clearly defined, and will allow an absolute calibrationof the SSD to better than 5%.

The performance requirements for the SSD come mainly from calibration requirements:in shower measurement mode, the dominant measurement errors are due to Poisson fluc-tuations of the number of particles detected, and the overall calibration constant determi-nation. Detector non-uniformity contributes a small error when compared to the Poissonerror, as long as non-uniformities are below 20%. While the FWHM of the WCD calibrationhistogram will be clearly smaller than that of the SSD (the calibration unit for the WCD, theVEM, is at about 100 pe), the fact that the SSD can be cross-triggered by the WCD meansthat the MIP is clearly visible against very little background. The width of the MIP distri-bution is mostly determined by Poisson statistics of the number of photoelectrons per MIP,the non-uniformity of the detector, and the intrinsic fluctuation of the response to a singleparticle, mainly due to different track lengths in the scintillator. The latter factor was deter-mined from simulations to be around 18%. The baseline design chosen for the SSD produces12 photoelectrons per MIP [146], which would degrade to 8 photoelectrons after 10 years ofoperation due to aging. This amounts to a 35% contribution to the MIP distribution width.

60 CHAPTER 4. THE SURFACE DETECTOR

Figure 4.1: 3D view of a water-Cherenkov detector with a scintillator unit on top.

The scintillator units have to be precisely calibrated with a technique similar to the cal-ibration procedure of the WCD (cf. section 4.2.7). The size of the detector and its intrinsicmeasurement accuracy should not be the dominant limitations for the measurement. Thedynamic range of the units has to be adequate to guarantee the physics goals of the pro-posed upgrade.

The detector will be assembled and tested in parallel in multiple assembly facilities toreduce the production time and, therefore, has to be easily transportable. The mechanicalrobustness of the scintillator units must be ensured. The units will be shipped after assem-bly, and validated at the Malargue facilities of the Pierre Auger Observatory before beingtransported to their final destination on top of a WCD in the Pampa. They will then haveto operate for 10 years in a hostile environment, with strong winds and daily temperaturevariations of up to 30C.

4.2.2 Detector design

The baseline design relies only on existing technology for which performance measurementshave been made. The Surface Scintillator Detectors (SSD) basic unit consists of two modulesof 2 m2 extruded plastic scintillator which are read out by wavelength-shifting (WLS)fibers coupled to a single photo-detector. Extruded scintillator bars read by wavelength-shifting fibers have already been employed in the MINOS detector [143]. The active part ofeach module is a scintillator plane made by 12 bars 1.6 m long of extruded polystyrene scin-tillator. Each bar is 1 cm thick and 10 cm wide. The scintillator chosen for the baseline designis produced by the extrusion line of the Fermi National Accelerator Laboratory (FNAL) [144].

The bars are co-extruded with a TiO2 outer layer for reflectivity and have four holes inwhich the wavelength-shifting fibers can be inserted. The fibers are positioned following thegrooves of the routers at both ends, in a “U” configuration that maximizes light yield andallows the use of a single photomultiplier (at the cost of a widening of the time responseof the detector by 5 ns, which has a totally negligible impact). The fibers are therefore read

Read-out of scintillators with WLS fibers

Simple and robust construction of detector module and mounting frame, double roof for thermal insulation

Both WCD and SSD to be connected to new 120 MHz electronics

R, Engel et al, ICRC2015R. Takeishi et al, ICRC 2015

Scintillator on Water-Tank (AugerPrime)

Comparison between the Surface Detectors of the Pierre Auger Observatory and the Telescope ArrayR. Takeishi

Figure 1: The two Auger SD stations deployed at the TA Central Laser Facility.

PMTs through optical windows. The signals are processed using front-end electronics having six10-bit Fast Analog to Digital Converters (FADCs) running at 40 MHz. A dynamic range of 15 bitsis realized using signals derived from the anode and from the last dynode (×32). The digitizedsignals are sent to a programmable logic device board used to make various triggering decisions.

The Auger North SD station is a one PMT water Cherenkov surface detector used in the PierreAuger Research and Development Array in Colorado, USA [11]. It is a cost-effective version ofthe Auger South SD station, with the same footprint, height and water volume. The Auger Northand South SD stations deployed at the TA CLF are shown in figure 1. The design of the electronicsfor the Auger North surface detector is based on the one used at the Auger South SD. In this casehowever, the digitization is performed with commercial 10-bit ADCs with 100 MHz sampling rate.The dynamic range is extended to 22 bits, using signals derived from the anode (×0.1, ×1 and×30) and from a deep (5th out of 8) dynode.

The TA SD station is composed of two layers of plastic scintillator with two PMTs, one foreach layer [12]. It has an area of 3 m2 and each layer has 1.2 cm thickness. The scintillators andPMTs are contained in a stainless steel box which is mounted under a 1.2 mm thick iron roof toprotect the detector from large temperature variations. Photons that are generated in the scintillatorare collected by wavelength shifting fibers and read out by PMTs. The signals from PMTs aredigitized by a commercial 12-bit FADC with a 50 MHz sampling rate on the CPU board.

3. Analysis and Results

In order to start collecting data immediately after its deployment, the Auger North SD stationwas configured to record data locally. This was done by installing a large capacity (512GB) flashdrive directly onto the local station controller. The second level trigger (T2) data, obtained fromthe standard Auger calibration procedure [1], were obtained and written on the drive at a rate ofabout 20 Hz. Only a very small fraction of those events arises from UHECR showers. A smallerdataset of atmospheric muons from the T1 trigger (100 Hz) was also collected to derive the VerticalEquivalent Muon (VEM) calibration from the single muon energy loss spectrum. In this analysis,the data from two observation periods are used; the first is Oct. 21, 2014 - Nov. 17, 2014 andthe second is Nov. 19, 2014 - Dec. 7, 2014. The flash drive was swapped between the twoperiods. The exchange requires a shutdown of the station to open the tank and access the local

3

Water-tank installed at TA site

Further Lower Energy = TALE (Telescope Array Low-energy Extension)

14

TALETA MD

TALE

TA MD

Enlarge field of view of FD in elevation to observe lower energy showers down to 1015.6 eV.

> 1017.4 eV Fluorescence dominated

< 1017.4 eV Cherenkov dominated

Resolution and Exposure as a Function of Energy

15

16

Energy Spectrum in 1015.6 eV to 1020.3 eVTALE + TA FD + TA SD

Preliminary

17

log(E(eV))=16.34±0.04 log(E(eV))=17.30±0.04

log(E(eV))=18.72±0.02

log(E(eV))=19.80±0.05

Combined Spectrum and Fitted Result

Preliminary

Comparison with other Measurements

18

(E (eV))10

log17.5 18 18.5 19 19.5 20 20.5

)-1 s

-1 sr

-2 m2

(eV

24/1

03

E×

Flux

-110

1

10Preliminary

TA Combined 2015

Auger ICRC 2013 +8.5%

Summary and Future Plans

19187187cm

124cm LeftModule RightModule

Practical implementation

Two modules in one box per station, readout by one PMT, area ~4 m2

12

10 cm10cm

1cm

4.2. THE SCINTILLATOR DETECTOR 69

Figure 4.12: 3D view of the SSD module with the support bars. The bars are connected to the tankusing lifting lugs present in the tank structure.

4.2.7 Calibration and control system

The SSD calibration is based on the signal of a minimum ionizing particle going through thedetector, a MIP. Since this is a thin detector, the MIP will not necessarily be well separatedfrom the low energy background but, being installed on top of the WCD, a cross triggercan be used to remove all of the background. About 40% of the calibration triggers of theWCD produce a MIP in the SSD. The statistics of calibration events recorded in a minute, thenormal WCD calibration period, are therefore enough to obtain a precise measurement of theMIP. Figure 4.13 shows the MIP calibration histogram from a 2 m2 test module, obtained inone minute of acquisition. The MIP is clearly defined, and will allow an absolute calibrationof the SSD to better than 5%.

The performance requirements for the SSD come mainly from calibration requirements:in shower measurement mode, the dominant measurement errors are due to Poisson fluc-tuations of the number of particles detected, and the overall calibration constant determi-nation. Detector non-uniformity contributes a small error when compared to the Poissonerror, as long as non-uniformities are below 20%. While the FWHM of the WCD calibrationhistogram will be clearly smaller than that of the SSD (the calibration unit for the WCD, theVEM, is at about 100 pe), the fact that the SSD can be cross-triggered by the WCD meansthat the MIP is clearly visible against very little background. The width of the MIP distri-bution is mostly determined by Poisson statistics of the number of photoelectrons per MIP,the non-uniformity of the detector, and the intrinsic fluctuation of the response to a singleparticle, mainly due to different track lengths in the scintillator. The latter factor was deter-mined from simulations to be around 18%. The baseline design chosen for the SSD produces12 photoelectrons per MIP [146], which would degrade to 8 photoelectrons after 10 years ofoperation due to aging. This amounts to a 35% contribution to the MIP distribution width.

60 CHAPTER 4. THE SURFACE DETECTOR

Figure 4.1: 3D view of a water-Cherenkov detector with a scintillator unit on top.

The scintillator units have to be precisely calibrated with a technique similar to the cal-ibration procedure of the WCD (cf. section 4.2.7). The size of the detector and its intrinsicmeasurement accuracy should not be the dominant limitations for the measurement. Thedynamic range of the units has to be adequate to guarantee the physics goals of the pro-posed upgrade.

The detector will be assembled and tested in parallel in multiple assembly facilities toreduce the production time and, therefore, has to be easily transportable. The mechanicalrobustness of the scintillator units must be ensured. The units will be shipped after assem-bly, and validated at the Malargue facilities of the Pierre Auger Observatory before beingtransported to their final destination on top of a WCD in the Pampa. They will then haveto operate for 10 years in a hostile environment, with strong winds and daily temperaturevariations of up to 30C.

4.2.2 Detector design

The baseline design relies only on existing technology for which performance measurementshave been made. The Surface Scintillator Detectors (SSD) basic unit consists of two modulesof 2 m2 extruded plastic scintillator which are read out by wavelength-shifting (WLS)fibers coupled to a single photo-detector. Extruded scintillator bars read by wavelength-shifting fibers have already been employed in the MINOS detector [143]. The active part ofeach module is a scintillator plane made by 12 bars 1.6 m long of extruded polystyrene scin-tillator. Each bar is 1 cm thick and 10 cm wide. The scintillator chosen for the baseline designis produced by the extrusion line of the Fermi National Accelerator Laboratory (FNAL) [144].

The bars are co-extruded with a TiO2 outer layer for reflectivity and have four holes inwhich the wavelength-shifting fibers can be inserted. The fibers are positioned following thegrooves of the routers at both ends, in a “U” configuration that maximizes light yield andallows the use of a single photomultiplier (at the cost of a widening of the time responseof the detector by 5 ns, which has a totally negligible impact). The fibers are therefore read

Read-out of scintillators with WLS fibers

Simple and robust construction of detector module and mounting frame, double roof for thermal insulation

Both WCD and SSD to be connected to new 120 MHz electronics

Comparison between the Surface Detectors of the Pierre Auger Observatory and the Telescope ArrayR. Takeishi

Figure 1: The two Auger SD stations deployed at the TA Central Laser Facility.

PMTs through optical windows. The signals are processed using front-end electronics having six10-bit Fast Analog to Digital Converters (FADCs) running at 40 MHz. A dynamic range of 15 bitsis realized using signals derived from the anode and from the last dynode (×32). The digitizedsignals are sent to a programmable logic device board used to make various triggering decisions.

The Auger North SD station is a one PMT water Cherenkov surface detector used in the PierreAuger Research and Development Array in Colorado, USA [11]. It is a cost-effective version ofthe Auger South SD station, with the same footprint, height and water volume. The Auger Northand South SD stations deployed at the TA CLF are shown in figure 1. The design of the electronicsfor the Auger North surface detector is based on the one used at the Auger South SD. In this casehowever, the digitization is performed with commercial 10-bit ADCs with 100 MHz sampling rate.The dynamic range is extended to 22 bits, using signals derived from the anode (×0.1, ×1 and×30) and from a deep (5th out of 8) dynode.

The TA SD station is composed of two layers of plastic scintillator with two PMTs, one foreach layer [12]. It has an area of 3 m2 and each layer has 1.2 cm thickness. The scintillators andPMTs are contained in a stainless steel box which is mounted under a 1.2 mm thick iron roof toprotect the detector from large temperature variations. Photons that are generated in the scintillatorare collected by wavelength shifting fibers and read out by PMTs. The signals from PMTs aredigitized by a commercial 12-bit FADC with a 50 MHz sampling rate on the CPU board.

3. Analysis and Results

In order to start collecting data immediately after its deployment, the Auger North SD stationwas configured to record data locally. This was done by installing a large capacity (512GB) flashdrive directly onto the local station controller. The second level trigger (T2) data, obtained fromthe standard Auger calibration procedure [1], were obtained and written on the drive at a rate ofabout 20 Hz. Only a very small fraction of those events arises from UHECR showers. A smallerdataset of atmospheric muons from the T1 trigger (100 Hz) was also collected to derive the VerticalEquivalent Muon (VEM) calibration from the single muon energy loss spectrum. In this analysis,the data from two observation periods are used; the first is Oct. 21, 2014 - Nov. 17, 2014 andthe second is Nov. 19, 2014 - Dec. 7, 2014. The flash drive was swapped between the twoperiods. The exchange requires a shutdown of the station to open the tank and access the local

3

TA measured the energy spectrum over 4.7 orders of magnitude in 1015.6 eV to 1020.3 eV range.

4 features seen: low energy ankle at 1016.34 eV, 2nd knee at 1017.30 eV, ankle at 1018.72 eV, suppression at 1019.80 eV

Large discrepancy with Pierre Auger above 1019.3 eV, which cannot be resolved by rescaling energies of the experiments.

TA×4 will provide us fourfold statistics at the suppression.

Activities to understand the suppression discrepancy.

energy spectrum measured by the telescope array experiment

Documents