Measurement of the cosmic ray antiproton/proton flux ratio at TeV energies with the ARGO-YBJ detector

0.32 MB·English

by  The ARGO-YBJ Collaboration

Measurement of the cosmic ray antiproton/proton flux ratio at TeV energies with the
ARGO-YBJ detector.
B. Bartoli,1,2 P. Bernardini,3,4 X.J. Bi,5 C. Bleve,3,4 I. Bolognino,6,7 P. Branchini,8 A. Budano,8 A.K. Calabrese
Melcarne,9 P. Camarri,10,11 Z. Cao,5 R. Cardarelli,11 S. Catalanotti,1,2 C. Cattaneo,7 S.Z. Chen,5 T.L. Chen,12
Y. Chen,5 P. Creti,4 S.W. Cui,13 B.Z. Dai,14 G. D’Al´ı Staiti,15,16 Danzengluobu,12 I. De Mitri,3,4 B. D’Ettorre
Piazzoli,1,2 T. Di Girolamo,1,2 X.H. Ding,12 G. Di Sciascio,11,∗ C.F. Feng,17 Zhaoyang Feng,5 Zhenyong
Feng,18 E. Giroletti,6,7 Q.B. Gou,5 Y.Q. Guo,5 H.H. He,5 Haibing Hu,12 Hongbo Hu,5 Q. Huang,18
M. Iacovacci,1,2 R. Iuppa,10,11,† I. James,8,19 H.Y. Jia,18 Labaciren,12 H.J. Li,12 J.Y. Li,17 X.X. Li,5
G. Liguori,6,7 C. Liu,5 C.Q. Liu,14 J. Liu,14 M.Y. Liu,12 H. Lu,5 X.H. Ma,5 G. Mancarella,3,4 S.M. Mari,8,19
2 G. Marsella,4,20 D. Martello,3,4 S. Mastroianni,2 P. Montini,8,19 C.C. Ning,12 A. Pagliaro,16,21 M. Panareo,4,20
1 B. Panico,10,11 L. Perrone,4,20 P. Pistilli,8,19 X.B. Qu,17 F. Ruggieri,8 P. Salvini,7 R. Santonico,10,11 P.R. Shen,5
0
X.D. Sheng,5 F. Shi,5 C. Stanescu,8 A. Surdo,4 Y.H. Tan,5 P. Vallania,22,23 S. Vernetto,22,23 C. Vigorito,23,24
2
B. Wang,5 H. Wang,5 C.Y. Wu,5 H.R. Wu,5 B. Xu,18 L. Xue,17 Y.X. Yan,14 Q.Y. Yang,14 X.C. Yang,14
n
Z.G. Yao,5 A.F. Yuan,12 M. Zha,5 H.M. Zhang,5 Jilong Zhang,5 Jianli Zhang,5 L. Zhang,14 P. Zhang,14
a
J X.Y. Zhang,17 Y. Zhang,5 Zhaxiciren,12 Zhaxisangzhu,12 X.X. Zhou,18 F.R. Zhu,18 Q.Q. Zhu,5 and G. Zizzi9

8 (ARGO-YBJ Collaboration)
1 1Dipartimento di Fisica dell’Universit`a di Napoli “Federico II”,

Complesso Universitario di Monte Sant’Angelo, via Cinthia, 80126 Napoli, Italy.
]
E 2Istituto Nazionale di Fisica Nucleare, Sezione di Napoli,
H Complesso Universitario di Monte Sant’Angelo, via Cinthia, 80126 Napoli, Italy.
3Dipartimento di Fisica dell’Universit`a del Salento, via per Arnesano, 73100 Lecce, Italy.
h. 4Istituto Nazionale di Fisica Nucleare, Sezione di Lecce, via per Arnesano, 73100 Lecce, Italy.
p 5Key Laboratory of Particle Astrophysics, Institute of High Energy Physics,
- Chinese Academy of Sciences, P.O. Box 918, 100049 Beijing, P.R. China.
o
6Dipartimento di Fisica Nucleare e Teorica dell’Universit`a di Pavia, via Bassi 6, 27100 Pavia, Italy.
r
t 7Istituto Nazionale di Fisica Nucleare, Sezione di Pavia, via Bassi 6, 27100 Pavia, Italy.
s 8Istituto Nazionale di Fisica Nucleare, Sezione di Roma Tre, via della Vasca Navale 84, 00146 Roma, Italy.
a
[ 9Istituto Nazionale di Fisica Nucleare - CNAF, Viale Berti-Pichat 6/2, 40127 Bologna, Italy.
10Dipartimento di Fisica dell’Universit`a di Roma “Tor Vergata”,
1
via della Ricerca Scientifica 1, 00133 Roma, Italy.
v 11Istituto Nazionale di Fisica Nucleare, Sezione di Roma Tor Vergata,
8
via della Ricerca Scientifica 1, 00133 Roma, Italy.
4
12Tibet University, 850000 Lhasa, Xizang, P.R. China.
8
13Hebei Normal University, Shijiazhuang 050016, Hebei, P.R. China.
3
14Yunnan University, 2 North Cuihu Rd., 650091 Kunming, Yunnan, P.R. China.
.
1 15Universita` degli Studi di Palermo, Dipartimento di Fisica e Tecnologie Relative,
0
Viale delle Scienze, Edificio 18, 90128 Palermo, Italy.
2 16Istituto Nazionale di Fisica Nucleare, Sezione di Catania, Viale A. Doria 6, 95125 Catania, Italy.
1 17Shandong University, 250100 Jinan, Shandong, P.R. China.
:
v 18Southwest Jiaotong University, 610031 Chengdu, Sichuan, P.R. China.
i 19Dipartimento di Fisica dell’Universit`a “Roma Tre”, via della Vasca Navale 84, 00146 Roma, Italy.
X
20Dipartimento di Ingegneria dell’Innovazione, Universita` del Salento, 73100 Lecce, Italy.
r 21IstitutodiAstrofisicaSpazialeeFisicaCosmicadell’IstitutoNazionalediAstrofisica, viaLaMalfa153,90146Palermo,Italy.
a
22Istituto di Fisica dello Spazio Interplanetario dell’Istituto Nazionale di Astrofisica, corso Fiume 4, 10133 Torino, Italy.
23Istituto Nazionale di Fisica Nucleare, Sezione di Torino, via P. Giuria 1, 10125 Torino, Italy.
24Dipartimento di Fisica Generale dell’Universit`a di Torino, via P. Giuria 1, 10125 Torino, Italy.
(Dated: January 19, 2012)
Cosmic ray antiprotons provide an important probe to study the cosmic ray propagation in the
interstellar space and to investigate the existence of dark matter. Acting the Earth-Moon system
as a magnetic spectrometer, paths of primary antiprotons are deflected in the opposite sense with
respecttothoseoftheprotonsintheirwaytotheEarth. Thiseffectallows, inprinciple,thesearch
for antiparticles in the direction opposite to the observed deficit of cosmic rays due to the Moon
(theso-called ‘Moon shadow’).
The ARGO-YBJ experiment, located at the Yangbajing Cosmic Ray Laboratory (Tibet, P.R.
2
China, 4300 m a.s.l., 606 g/cm ), is particularly effective in measuring the cosmic ray antimatter
contentviatheobservationofthecosmicraysshadowingeffectdueto: (1)good angularresolution,
pointing accuracy and long-term stability; (2) low energy threshold; (3) real sensitivity to the
geomagnetic field.
2
Based onallthedatarecordedduringtheperiod from July2006 throughNovember2009andon
a full Monte Carlo simulation, we searched for the existence of the shadow cast by antiprotons in
the TeV energy region. No evidence of the existence of antiprotons is found in this energy region.
Upper limits to the p¯/p flux ratio are set to 5% at a median energy of 1.4 TeV and 6% at 5 TeV
with a confidencelevel of 90%. In the TeV energy range these limits are thelowest available.
PACSnumbers: 14.20.Dh;13.85.Tp;96.50.S-;96.50.sd;95.85.Ry
I. INTRODUCTION Theestimatesofthesecondarypfluxaresufferingfrom
uncertaintiesonmodelsandparametersofparticleprop-
agation in the Galaxy, CR spectrum and composition,
A. Cosmic ray antiproton production
details of the nuclear crosssections for p production, an-
nihilation andscattering and, finally, on the heliospheric
Very high energy cosmic ray (VHE CR) antiprotons
modulation [5, 6]. On the contrary, there is a general
are an essential diagnostic tool to approach the solution
agreementinthecalculationofthesecondaryhigh-energy
ofseveralimportantquestionsofcosmology,astrophysics
p flux falloff: around 50 GeV the intensity decreases by
andparticlephysics,besidesstudyingfundamentalprop-
about 3 orders of magnitude below the maximum.
erties of the CR sources and propagation medium. The
Recent measurements of the antiproton flux up to
enigma of the matter/antimatter asymmetry in the lo-
about180GeVbythePAMELAsatellite[7,8]areconsis-
cal Universe, that of the existence of antimatter regions,
tent with the conventional CR model, in which antipro-
the search for signatures of physics beyond the standard
tons are secondary particles yielded by the spallation of
model of particles and fields, as well as the determina-
CR nuclei over the interstellar medium. Nevertheless,
tion of the essential features of CR propagation in the
given the current uncertainties on propagation parame-
insterstellar medium, these are only a few research top-
ters, exotic models of primary p production cannot be
icsthatwouldgreatlybenefitfromthe detectionofVHE
ruled out [8, 9]. As an example, recent calculations sug-
antiprotons (see for example [1, 2]).
gestthat the overallPAMELAp ande+ data[8,10]and
First of all, the observation of p abundance in the CR Fermi e+ +e− data [11] can be reproduced taking into
flux is a key to understand CR propagation. In fact, an-
accounta heavy dark matter particle (M ≥10 TeV) that
tiprotons are produced by standard nuclear interactions annihilatesintoW+W− orhh[12]. Thisscenarioimplies
of CR nuclei with the interstellar medium, the informa-
thatthep/pratio,consistentwiththebackgroundofsec-
tion coming from these spallation processes being com-
ondaryproductionuptoabout50GeV,increasesrapidly
plementary to that achievable fromsecondarynuclei like
reachingthe 10−2 level atabout 2 TeV. CR antiprotons,
Li, Be and B or secondariesof iron. It shouldbe noticed
as well as positrons, are therefore considered as prime
that antiprotons mostly trace the propagationhistory of
targets for indirect detection of dark matter [13–15].
protonspp→ p ppp, unliketheotherspallationproducts,
But it has been also suggested that the PAMELA
which may come from heavier nuclei. Also secondary p
positrondata may be a natural consequence of the stan-
represent a background flux that must be carefully de-
dardscenario for the originof galactic CRs, if secondary
termined to take out a primary p component due to any
e+ (and e−) production takes place in the same region
hypothetical exotic signal.
where CRs are being accelerated [16]. Since this is a
The observedamountof antiprotons in CRs is still far
hadronic mechanism, an associated rise of the p/p ra-
from being figured out. Detailed calculations show that
tio is predicted at energies ≥ 1 TeV [17]. Therefore, the
there is no model capable of accurately describing alto-
high-energyrangeoftheantiprotonspectrummayreveal
getherB/Candsub-Fe/Feratios,spectraofp,He, p,e+,
important constraints on the physics of the CR acceler-
e− anddiffuseγ-rays. Infact, conventionalmodelswith-
ation sites.
outreaccelerationfailinreproducingbothB/Cratioand
Antiprotons can be produced from primordial black
pfluxatthesametime[3]. Diffusivereaccelerationmod-
holes evaporation [18, 19] or in antigalaxies [20–23]. In
elsnaturallyreproducesecondary/primarynucleiratioin
particular, it seems possible that mechanisms exist that
CRs but produce too few antiprotons [4]. The introduc-
couldproduce the formationofseparatedantimatter do-
tion of a break in the diffusion coefficient [3] would lead
mainsduringthecosmicevolution,thusallowingthevis-
to consistent results, but it is not theoretically justified
ibleUniversetobegloballymatter-antimattersymmetric
so far, still resulting as an ad hoc assumption [4]. Some
[24]. In addition, the possibility exists to have antimat-
modelstakingintoaccountGalacticconvectivewindand
ter confined into condensed bodies like antistars in our
stochastic reaccelerationmay reproduce both antiproton
Galaxy [25, 26].
flux and secondary/primaryratio [6].
Intheoriesinwhichmatterandantimatterarepresent
in equal amounts in spatially separated domains of sur-
vivable size it is expected that the p/p ratio should
∗Electronicaddress: [email protected] increase with energy in the framework of the energy-
†Electronicaddress: [email protected] dependent confinement model for CRs in the Galaxy. In
3
asimple”leakybox”modeltheenergyspectrumissolely B. The Moon shadowing effect
determined by the balance between generation at the
source and escape from the Galaxy. If the source spec- Since the Moon has an angular radius of about 0.26◦,
trum is proportional to E−α, the equilibrium spectrum
it must cast a shadow in the nearly-isotropic CR flux
of CRs inside the source region (i.e. inside the Galaxy) (the so-called shadow of the Moon). As first suggested
would be ∝E−(α+δ), due to the energy-dependent leak-
by Clark in 1957 [29], the shadowing of CRs from the
age of the source. Indeed, according to recent measure-
direction of the Moon is useful in measuring the angular
ments of B/C ratio in the primary CRs up to about 50
resolution of an air shower array directly, without need-
GeV/nucleon, the residence time of CRs in the Galaxy
ing Monte Carlo (MC) simulations. In fact, the shape
can be described by a power law in energy or rigidity
of the shadow provides a measurement of the detector
∝ R−δ, where δ ∼0.6 [5]. As a result, the energy spec-
point spread function, and its position allows the check
trum of the CRs leaked out from the antigalaxy has a
of possible pointing biases.
spectrum∝E−α. Ifweassumethatthereexistsageneral
Inaddition,duetothegeomagneticfield(GMF),posi-
accelerationmechanismforgeneratingCRswhichactsin
tivelychargedparticlesaredeflectedbyanangledepend-
bothgalacticandextragalacticsourcestogiveanuniver-
ing on the primary CR energy [30]. This effect produces
salsourcespectrumE−α,theextragalacticCRspectrum
a displacement of the shadow towards the West with re-
should reflect it. Thus, if antiprotons are assumed to be
spect to the Moon position and smears the shape in the
both primary and extragalactic, we should observe the
East-West direction, especially at low energies. The ob-
sourcespectrumofantigalaxies∝E−α andtheexpected
servationof the displacement of the Moon providesa di-
p/p ratio should increase with energy: p/p ∝ Eδ [21?
rect calibration of the relation between shower size and
,22]. Asaconsequence,theantiprotonfractioncouldin-
primary energy [30].
creaseuptoabout1%around500GeVorevento50%in
The same shadowing effect can be seen in the direc-
themulti-TeVenergyrangewithimportantobservational
tion of the Sun. Nevertheless, the interpretation of the
implications, being, at these energies, the backgroundof
shadow phenomenology is more complex. In fact, the
secondary antiprotons well below this prediction.
displacement of the shadow from the apparent position
of the Sun could be explained by the joint effects of
At high energies (≥ 100 GeV) the main observable re- the GMF and of the Solar and Interplanetary Magnetic
lated to the residence time of CRs in the Galaxy is the Fields (SMF andIMF,respectively)whose configuration
large-scale anisotropy in their arrival direction that is considerablychangeswiththephasesofthesolaractivity
known to be strictly related to the diffusion coefficient. cycle [31, 32].
Measurementsgivetheamplitudeofthefirstangularhar-
Linsley [33] and Lloyd-Evans [34] in 1985, following
monicofanisotropyattheorder10−3intheenergyrange a Watson’s suggestion, independently explored the pos-
1011 to 1014 eV where the most reliable data are avail-
sibility to use the Moon or Sun shadows as mass spec-
able [27, 28]. The data on CR anisotropy are consistent, trometers in order to measure the charge composition of
within a factor of about 3, with a diffusion coefficient
CR spectrum. In particular Linsley first discussed the
increasing with energy ∝ E0.3, as predicted in models idea to measure the CR antiprotons abundance exploit-
including stochastic reacceleration by Kolmogorov-type
ingtheseparationoftheprotonandantiprotonshadows.
hydromagneticturbolence(2ndorderFermiacceleration) In1990Urbanet al. [35]carriedoutdetailedcalculation
[5]. The measurement of the p/p ratio at high energies
of this effect proposing this method as a way to search
may be useful to constrain models for p production and for antimatter in primary CR at the TeV energies.
for the confinement of CRs, even if it is not straightfor-
The GMF should deflect the antimatter component in
wardtoinferthepropagationparameters,asthediffusion
the CRs in opposite direction with respect to the mat-
index δ, since they are partially degenerate with source
ter component. Therefore, if protons are deflected by
parameters [17].
the GMF towards East, antiprotons are deflected to-
wardsWest. Ifthe energyis lowenoughandthe angular
The first antiproton upper limits in the high energy resolution is adequate we can distinguish, in principle,
regionhavebeenobtainedbyStephensin1985exploiting two shadows, one shifted towards West due to the pro-
the observed charge ratio of muons at sea level. The tons and the other shifted towards East due to the an-
presence of p dilutes the chargeratio µ+/µ−. The limits tiprotons. At high energy (≥ 10 TeV) the magnetic de-
thusderivedforthep/pratioare7%,17%,10%and14% flection is too small compared to the angular resolution
respectively for the energy intervals 0.1 - 0.2 TeV, 1.0 - and the shadows cannot be disentangled. At low energy
1.5 TeV, 10 - 15 TeV and >30 TeV [23]. (≈100 GeV) the well deflected shadows are washed out
by the poor angular resolution, thus limiting the sensi-
In addition, deeper measurements of the p/p ratio at tivity. Therefore, there is an optimal energy window for
high energies have been performed exploiting the Earth- the measurement of the antiproton abundance.
Moon system as a magnetic spectrometer able to disen- In1991,theCYGNUScollaboration[36]firstobserved
tangle, in principle, the deflection of protons from that the CR shadowing effect measuring a deficit of 4.9 stan-
of antiprotons in the geomagnetic field. dard deviations (s.d.) in the CR background by super-
4
posing the Moon and Sun data at an energy of about 50 data recordedduring the period from July 2006 through
TeV.InthesameyearalsotheEAS-TOPexperimentob- November 2009.
servedtheshadowingeffectduetotheMoonandtheSun Thepaperisorganizedasfollows. IntheSectionIIthe
onthe 100TeVCRs flux with asignificanceofabout2.7 ARGO-YBJ detector is described. The description of a
s.d. [37]. In the following years this effect has been con- detailed MC simulation of the Earth-Moon spectrome-
firmed by other EAS-arrays (CASA-MIA[38], HEGRA ter system developed in order to evaluate the deficit of
[39], GRAPES [40]). eventsandtocalibratethedetectorisbrieflysketchedout
Thefirstobservationsofashadowingeffecthadtowait inSectionIII.InSectionIVthe data analysisis outlined
for the results of the CYGNUS and EAS-TOP experi- and the detector performance summarized. The p¯/p ra-
mentsin1991. Therearemainlytworeasonsforthislong tio calculation method is also described in Section IV.C.
delay, first the poor angular resolution of EAS-arrays in Finally,theresultsofthedataanalysisarepresentedand
comparison with the angular radius of the Moon or Sun discussed in Section V. A summary of the obtained re-
(∼0.26◦). Indeed,onlyatthebeginningofthe90sthean- sults is given in Section VI.
gular resolutions of EAS-arrays reached the 1 deg level.
Second, due to the high energy threshold (≈100 TeV)
of the experiments the statistical significance of the ob- II. THE ARGO-YBJ EXPERIMENT
servations was small and the position of the shadow not
affectedbythe GMF.Therefore,the deficitofthe count- A. The detector
ing rate was observed as a function of the angular dis-
tance from the Moon position, without any information
The ARGO-YBJ experiment, located at the YangBa-
on the East-West asymmetry and consequently, without
Jing Cosmic Ray Laboratory (Tibet, P.R. China, 4300
any possibility to study the CR antimatter content.
m a.s.l., 606 g/cm2), is constituted by a central carpet
In1993,theTibetASγ experimentmeasuredboththe ∼74× 78 m2, made of a single layer of Resistive Plate
Moon and Sun shadows with an energy threshold low Chambers (RPCs) with ∼93% of active area, enclosed
enough(about10TeV)toallowa2-dimensionalstudyof by a guard ring partially instrumented (∼20%) up to
the effect. In particular, they observed for the first time ∼100×110 m2. The RPC is a gaseous detector work-
a westward displacement of the Moon shadow (0.16◦ at
ing with uniform electric field generated by two parallel
the 7.1 σ level)fromits actualposition. With this result electrode plates ofhigh bulk resistivity (1011Ω cm). The
the first upper limit to the p/p ratio with this technique
intense field of 3.6 kV/mm at 0.6 atm pressure provides
wassetatabout30%[41]. Afterwards,the collaboration
verygoodtime resolution(1.8ns)andthehighelectrode
set an upper limit to the p/p at 10 TeV at about 10%
resistivity limits the area interestedby the electrical dis-
[42]. chargeto few mm2. The apparatushasa modular struc-
The CR shadowing effect has been observed also in ture, the basic data acquisition element being a cluster
the highenergymuondistributionwithundergroundde- (5.7×7.6 m2), made of 12 RPCs (2.85×1.23 m2 each).
tectors (SOUDAN-2 [43], MACRO [44, 45], L3+C [46], Each chamber is read by 80 external strips of 6.75×61.8
BUST [47], MINOS [48]). Recently, the ICECUBE ex- cm2 (the spatial pixel), logically organized in 10 inde-
periment observed the Moon shadow in the Southern pendentpadsof55.6×61.8cm2 whichrepresentthe time
hemisphere with a statistical significance of more than pixel of the detector [52]. The read-out of 18360 pads
10 s.d. [49]. Upper limits to the p/p ratio have been set and146880strips arethe experimentaloutput ofthe de-
by the MACRO (48% at 68% c.l. at about 20 TeV) [45] tector. The RPCs are operated in streamer mode by
and L3+C (11% at 90% c.l. around 1 TeV) [46] collabo- usingagasmixture(Ar15%,Isobutane10%,TetraFluo-
rations. roEthane75%)forhighaltitudeoperation[53]. Thehigh
HighsensitivityobservationsoftheMoonshadowhave voltage settled at 7.2 kV ensures an overall efficiency of
beenrecentlyreportedinthemulti-TeVenergyregionby about96%[54]. The centralcarpetcontains130clusters
an upgraded version of the Tibet ASγ array [50] and by (hereafter ARGO-130)andthe full detector is composed
the MILAGRO collaboration [51]. While the Tibet ASγ of 153 clusters for a total active surface of ∼6700 m2.
experiment set a limit to the p/p ratio to 7% at 90% c.l. The total instrumented area is ∼11000 m2.
at about 3 TeV, so far the MILAGRO collaboration did Asimple,yetpowerful,electroniclogichasbeenimple-
not publish any result on the antimatter searchwith the mented to build an inclusive trigger. This logic is based
Moon shadow. on a time correlationbetween the pad signals depending
TheARGO-YBJexperimentisparticularlyeffectivein on their relative distance. In this way, all the shower
measuringtheCRantimattercontentviatheobservation events giving a number of fired pads N ≥ N in
pad trig
of the CRs shadowing effect due to: (1) good angular the central carpet in a time window of 420 ns generate
resolution,pointingaccuracyandlong-termstability;(2) the trigger. This trigger can work with high efficiency
low energy threshold; (3) real sensitivity to the GMF down to N = 20, keeping negligible the rate of ran-
trig
due to the absence of any systematic shift in the East- domcoincidences. The timing calibrations ofthe pads is
Westdirection. Inthispaperwereportthemeasurement performed according to the method reported in [55, 56].
of the p/p ratio in the TeV energy region with all the The whole system, in smooth data taking since July
5
2006 with ARGO-130, is in stable data taking with the field of view, extensive primaries are generated with ar-
full apparatus of 153 clusters since November 2007 with rival direction sampled within the Moon disc. For each
the trigger condition N = 20 and a duty cycle ≥85%. chemical species (p, He, CNO group, Mg-Si group and
trig
The trigger rate is ∼3.5 kHz with a dead time of 4%. Fe),thenumberofprimariestobegeneratediscomputed
Once the coincidence of the secondary particles has on the basis of the effective exposure time, according to
been recorded, the main parameters of the detected the energy spectrum resulting from a global fit of the
shower are reconstructed following the procedure de- main experimental data [59].
scribed in [30]. In short, the reconstruction is split into Once the number of CRs expected to be hampered
the following steps. Firstly, the shower core position is by the Moon has been calculated, the charge sign of ev-
derived with the Maximum Likelihood method from the ery primary is inverted and it is propagated back to the
lateraldensity distributionofthe secondaryparticles. In Moon, the magnetic field bending its trajectory. The
the second step, given the core position, the shower axis propagation stops anyway at the Moon distance, giving
is reconstructed by means of an iterative un-weighted agoodapproximationofthedeflectionundergonebythe
planar fit able to reject the time values belonging to the CR before reaching the atmosphere. In fact, if we firstly
non-gaussian tails of the arrival time distribution. Fi- consider a positively charged CR arriving to the Earth
nally,aconicalcorrectionisappliedto the survivinghits atmosphereandthenweinvertitschargeanditsmomen-
in order to improve the angular resolution. Unlike the tum and threw it back to the space, the two trajectories
information on the plane surface, the conical correction do not overlap because of the numerical approximation.
is obtained via a weighted fit which lowers the contribu- Nonetheless, as long as the primary energy is above sev-
tion from delayed secondary particles, not belonging to eral tenth of GeV, both trajectories give similar devia-
the shower front. tions, the difference ranging from 15% at 50 GeV down
The analysisreportedinthis paper refersto eventsse- to3%above1 TeV. Furtherdetailsandresultsfromthis
lected according to the following criteria: (1) more than simulation can be found in [30, 57].
25 strips N should be fired on the ARGO-130 car- After accounting for the arrival direction correction
strip
pet; (2) the zenith angle of the shower arrival direction ought to the magnetic bending effect, the air showers
should be less than 50◦; (3) the reconstructed core posi- development in the atmosphere has been generated with
tionshouldbeinsideanarea150×150m2centeredonthe the CORSIKA v. 6.500 code [60]. The electromagnetic
detector. After these selectionsthe number ofeventsan- interactionsaredescribedbytheEGS4packagewhilethe
alyzedisabout2.5×1011(about109insidea10◦×10◦an- hadronic interactions above 80 GeV are reproduced by
gular region centered on the Moon position). According theQGSJET-II.03andtheSYBILLmodels. Thelowen-
tosimulations,themedianenergyoftheselectedprotons ergy hadronic interactions are described by the FLUKA
is E ≈1.8 TeV (mode energy ≈0.7 TeV). package. CR spectra have been simulated in the energy
50
range from 10 GeV to 1 PeV following the relative nor-
malization given in [59]. About 108 showers have been
III. MONTE CARLO SIMULATION sampled in the zenith angle interval 0-60 degrees. The
secondaryparticleshavebeenpropagateddowntoacut-
off energies of 1 MeV (electromagnetic component) and
The deficit of events in the Moon direction, the ab-
100MeV(muonsandhadrons). Theexperimentalcondi-
solute energy calibration and the angular resolution, as
tions (trigger logic,time resolution, electronic noises, re-
well as the systematic pointing biases, have been stud-
lationbetweenstripandpadmultiplicity,etc.) havebeen
ied by comparing the observed Moon shadow charac-
taken into account via a GEANT4-based code [61]. The
teristics (East-West and North-South displacements and
corepositionshavebeenrandomlysampledinanenergy-
shape) with the expectations from a detailed MC simu-
dependent area large up to 2·103 × 2·103 m2, centered
lation of the CR propagationin the Earth-Moon system
on the detector. Simulated events have been generated
[30, 57, 58].
in the same format used for the experimental data and
With this simulation we also estimated the expected
analyzed with the same reconstruction code.
antiprotons flux in the opposite CR Moon shadow side,
as described in Section IV.C.
FromtheMCsimulationstrategyviewpoint,theMoon
shadow has been treated like an extensive excess signal, IV. DATA ANALYSIS
insteadofalackintheisotropicCRsflux. Inotherwords,
oursimulationdealswiththeMoonasifitwasthesource For the analysis of the shadowing effect, the signal is
of the CRs which it intercepts in reality. collected within a 10◦×10◦ sky region centered on the
The simulation has been realized on the basis of the Moon position. We used celestial coordinates (right as-
real data acquisition time. The Moon position has been cension and declination, R.A. and DEC. hereafter) to
computed at fixed times, starting from July 2006 up to buildtheevent andbackground skymaps,with0.1◦×0.1◦
November 2009. Such instants are distant 30 seconds binsize. Finally,afterasmoothingprocedure,thesignif-
eachother. Foreachtime, aftercheckingthe dataacqui- icance map, used to estimate the statistical significance
sition was effectively running and the Moon was in the of the observation, is obtained.
6
N 4 5
event number1990050 · 103 ) 231 -05
85 (cid:176)(m 0
d-
80-4 S - 3 - 2 - 1d-dm0((cid:176)) 1 2 ( a 3 )N 4 4 W3 2 a(1a- 0m)c-o1sd-(cid:176)(2) - 3 - 4E dS --21 --1150
-3
· 103 -20
684
-4
-4 -3 -2 -1 0 1 2 3 4
682
W (a -a )cosd ((cid:176) ) E
m
680
er 678 FIG. 2: Significance map of the Moon region observed with
mb all events detected by ARGO-YBJ. The event multiplicity is
ent nu 676 2a5re≤RN.Ast.riαp a<nd40DaEnCd.zδencietnhtearnegdleonθ t<he50M◦.ooTnhpeocsoitoirodnin(aαtmes,
ev 674 δm). Thecolorscalegivesthestatisticalsignificanceinterms
of standard deviations.
672
670
Fig. 2. It contains all events belonging to the lowest
668-10 -8 -6 -4 -2 0 2 4 6 8 10 multiplicitybininvestigated(25≤N <40),collected
strip
W (a -a m)cosd ((cid:176)) E by ARGO-YBJ during the period July 2006 - November
(b) 2009(about 3200hours on-sourcein total). N is the
strip
number of fired strips on the central carpet ARGO-130.
FIG.1: Plot (a): showersfiringNstrip >100collected around The significance of the maximum is about 22 s.d.. The
the Moon position. The coordinates are R.A. α and DEC. δ
observedwestwarddisplacementof the Moon shadowby
centeredontheMoon position (αm,δm). Theplot (b)shows about 1.5◦ allows to appreciate the sensitivity of the
themap projections along theR.A. direction.
ARGO-YBJ experiment to the GMF. This means that
a potential antiproton signal is expected eastward within
1.5◦ from the actual Moon position (i.e., within 3◦ from
As it can be appreciated in Fig. 1, the Moon shadow
the observed Moon position). The median energy of se-
turns outto be alackinthe smoothCRsignal,observed
lectedeventsisE ≈750GeV(modeenergy≈550GeV)
byARGO-YBJevenwithoutsubtractingthebackground 50
for proton-induced showers. The corresponding angular
contribution nor smoothing the signal.
resolution is ∼1.6◦.
The large displacement of the shadow is only one el-
ement of this analysis, the other one being the angu-
A. Moon shadow analysis lar resolution which is not adequate in this multiplicity
range. Indeed, as can be seen from the Fig. 2, the mat-
Cosmic rays blocked by the Moon must be as much ter shadow is visible on the antimatter side with a sig-
as the background events lying within a region as large nificance ofabout10s.d., thus limiting the sensitivity to
as the Moon disc. A suitable background estimation is the antiproton abundance measurement. We note that
therefore a crucial point of the analysis. As it can be this is the first time that an EAS array is observing the
seen also from the projection along the R.A. from Fig. Moon shadow cast by sub-TeV primary CRs.
1, the background events are not uniformly distributed
around the Moon, because of the non-uniform exposure
of the map bins to the CR radiation. The background B. Detector performance
has been estimated with the equi-zenith angle method,
as described in detail in [30]. Theperformanceofthedetectoranditsoperationsta-
A significance map of the Moon region is shown in bility have been studied in detail in [30] exploiting the
7
CRMoonshadowingeffectwithalldatasinceJuly2006. 4000
Themeasuredangularresolutionisbetterthan0.5◦for 2000
CR-induced showers with energies E > 5 TeV, in good 0 40<Nstrip<100
agreement with MC expectations. The Point Spread nts -2000
ou -4000
Fpruonjcetcitoinonofnottheaffdeectteecdtorb,ysttuhdeieGdMinF,thise GNaourstshi-aSnouftohr deficit c --86000000
Nstrip ≥200, while for lower multiplicities is better de- -10000 Data
scribed for both MC and data with a linear combination -12000 MC
of two Gaussian functions. The second Gaussian con- -14000-4 -3 -2 -1 0 1 2
(a -a )cosd ((cid:176))
tributes for about 20%. m
4000
Thelong-termstabilityoftheARGO-YBJexperiment
2000
has been checkedby monitoring both the position of the N >100
0 strip
Moon shadow, separately along R.A. and DEC. projec-
nts -2000
tions,andthe amountofshadowdeficiteventsinthepe- ou -4000
rmioodntNhoavnedmfboerrev2e0n0t7s–wiNthovNember>2100100., fAosrsehaocwhnsiidneFreiga.l deficit c --86000000
strip
17 of ref. [30], the position of the Moon shadow turned -10000
outto be stable atalevelof0.1◦ andthe angularresolu- -12000
tion stable at a level of 10%, on a monthly basis. These -14000-4 -3 -2 (a --a1)cosd ((cid:176)) 0 1 2
m
results make us confident about the detector stability in
the long-term observation of the Northern sky. A sys- FIG. 3: Deficit counts measured around the Moon projected
tematic uncertainty of (0.19±0.02)◦ towards the North along the East-West axis for two different multiplicity bins
in the absolute pointing accuracy is observed. The most (black circles) compared to MC expectations (red squares).
important contribution to the systematics is likely due EventscontainedinanangularbandparalleltotheEast-West
to a residual effect not completely corrected by the time axis and centered on the observed Moon position, propor-
calibration procedure. Further studies are under way. tional to the multiplicity-dependent angular resolution, are
used (see text).
We haveestimatedthe primaryenergyofthe detected
showers by measuring the westward displacement as a
function of the shower multiplicity, thus calibrating the
relationbetweenshowersizeandCRenergy. Thesystem- tations: 40≤Nstrip <100 and Nstrip >100, in the upper
aticuncertaintyintheabsoluterigidityscaleisevaluated and lower panels, respectively. The vertical axis reports
to be less than 13% in the range from 1 to 30 TeV/Z, the events contained in an angular band parallel to the
mainly due to the statistical one [30]. East-Westaxisandcenteredonthe observedMoonposi-
tion. The widths of these bands are chosen on the basis
of the MC simulation so that the shadow deficit is max-
C. The CR p¯/p flux ratio calculation imized. They turn out to be proportional to the Nstrip-
dependentangularresolution. Thewidthsofthesebands
are ±2.80◦ and ±1.90◦, respectively.
AllchemicalspeciesofCR,eachwithitsownspectrum,
The data are in good agreement with the MC sim-
contribute to form the Moon shadow signal. Hence, the
ulation and the observed shadows are shifted westward
chance of unfolding all contributions relies on MC simu-
of (-0.75±0.05)◦ and (-0.30±0.05)◦, as expected. A de-
lations, as well as the searchfor antiprotons demands to
tailed analysis of this sortof projections as a function of
properly reproduce the Moon shadow signal. The shape
the shower size is given in [30].
oftheMoonshadowistightlyconnectedtotheprimaries
energy spectrum, which mostly determines the tails of The GMF shifts westward the dip of the signal from
the signal. At first, we assume the energy spectrum of positively charged primaries. Searching antiprotons
antiprotonsfollowsapowerlawdN/dE =k·E−γ,where meanslookingforexcessesintheeasternpartoftheR.A.
thespectralindexγ istakentobeaslargeasthatofpro- projection, i.e. trying and fitting the Moon shape ex-
tons. Aninvestigationofthe dependence onthe spectral pected from combining CR and antimatter to the shape
index will follow in the Section V. obtained from experimental data. Of course, whichever
In order to evaluate the CR p¯/p flux ratio, only the matter-antiprotons combination is obtained, the total
projection along the R.A. is important, as it has been amountof triggeredevents mustnotbe changed,so that
shownthatatYangbajingtheGMFhasanon-nulleffect the fitting procedure consists in transferring MC events
on the CR trajectories only along such direction [30]. from the CR to the antiprotons shadow and comparing
The R.A. projection results from an integration along the result with data.
the DEC. direction. Tomakesuchacomparison,wefirstlyadoptedthefol-
ThedeficitcountsobservedaroundtheMoonprojected lowingmethod. WeobtainedtwokindsofMoonshadow,
ontheR.A.axis(thatisontheEast-Westaxis)areshown cast by all CRs and protons, respectively. After project-
inFig. 3fortwomultiplicitybinscomparedtoMCexpec- ing them along the R.A. direction, we used a superposi-
8
tion of several Gaussian functions to describe the deficit the following multiplicity ranges: 40≤N <100 and
strip
event distribution in each shadow [50]. Four Gaussian N ≥100. Intheformerbinthestatisticalsignificance
strip
functions were found to be adequate for fitting both dis- ofMoonshadowobservationis 34 s.d., the measuredan-
tributions within 5◦ from the Moon disc center. Let us gular resolution is ∼1◦, the proton median energy is 1.4
name θ the angular distance from the Moon disc center TeVandthe numberofdeficiteventsabout183000. The
and f (θ) the Gaussian function superposition describ- shadow is shifted of about 1◦ westward. In the latter
m
ing the CR shadow. Let F (θ) be the proton shadow, multiplicity bin the significance is 55 s.d., the measured
p
obtained by imposing a given power law spectrum. The angular resolution ∼0.6◦, the proton median energy is 5
observed Moon shadow should be expressed by the fol- TeV and the number of deficit events about 46500. The
lowing function: shadow is shifted of about 0.4◦ westward.
There exists, however, no evidence indicating deficits
fMOON(θ)=(1−r)fm(θ)+rFp(θ) of CRs at the opposite positions around θ = 1◦ and 0.4◦
=(1−r)f (θ)+rF (−θ) in the eastward direction, corresponding to the particles
m p
with negative charge (anti-matter) such as p¯, H¯e, C¯,...
(0 ≤ r < 1) where the first term represents the deficit and F¯e, if any.
in CRs and the second term represents the deficit in an-
Therefore, we applied both methods described in the
tiprotons. This function must be fitted to the data to
previous section to evaluate upper limits to the CR p¯/p
obtain the best value of r.
flux ratio. The r parameter which best fits the expecta-
We alsoappliedasecondmethodtodetermine the an-
tions to the data turns out to be always negative, i.e. it
tiproton content in the cosmic radiation. Without intro-
assumesnon-physicalvaluesthroughoutthewholeenergy
ducing functions to parameterize the expectations, we
rangeinvestigated. WithadirectcomparisonoftheR.A.
directly comparedthe MC signalwith the data. We per-
projections, the r-values which maximize the likelihood
formed a Maximum Likelihoodfit using the p¯content as are: -0.076±0.040 and -0.144±0.085 for 40≤N <100
strip
a free parameter with the following procedure: and N ≥100, respectively. The corresponding upper
strip
limits with 90% confidence level (c.l.), according to the
1. the Moon shadow R.A. projection has been drawn
unified Feldman & Cousins approach [62], are 0.034 and
both for data and MC.
0.041, respectively.
2. the MC Moon shadow has been split into a “mat- Since the anti-shadow was assumed to be the mirror
ter” part plus an “antiproton” part, again so that image of the proton shadow, we assume for the antipro-
the total amount of triggered events remains un- tonsthesamemedianenergy. Thep¯/pratioisΦ(p¯)/Φ(p)
changed: = 1/f · Φ(p¯)/Φ(matter), therefore, being the assumed
p
proton fraction f = 73% for 40≤N <100 and f =
p strip p
ΦMC(mat)−→ΦMC(r;mat+p¯) 71% for N ≥100 [59], we are able to set the follow-
strip
=(1−r)Φ (mat)+Φ (p¯) ing upper limits at 90% c.l.: 0.05 for 40≤N <100
MC MC strip
and 0.06 for N ≥100. Notice that the two values are
strip
3. for each antiproton to matter ratio, the expected similar, in spite of the different multiplicity interval. It
Moon shadow R.A. projection Φ (r;mat+p¯) is is a consequence of the combination of the two opposite
MC
comparedwith the experimentalone via the calcu- effects of the angular resolution and of the geomagnetic
lation of the likelihood function: deviation.
In Fig. 4 the ARGO-YBJ results are compared with
B
all the available measurements. The energy bin is 34%
L(r)=XNiln[Ei(r)]−Ei(r)−ln(Ni!) around the median energy for each multiplicity interval.
i=1 The solid curves refer to a theoretical calculations for a
where N is the number of experimental events in- pure secondaryproductionofantiprotonsduring the CR
i
cluded within the i-th bin, while E (r) is the num- propagation in the Galaxy by Donato et al. [14]. The
i
ber of events expected within the same bin, which curves were obtained using the appropriate solar mod-
is calculated by adding the contribution expected ulation parameter for the PAMELA data taking period
fromMC(Φ (r;mat+p¯))tothemeasured back- [8]. The long-dashed line refer to a model of extragalac-
MC
ground. ticprimaryp¯production[20,23]. Therigidity-dependent
confinementofCRsintheGalaxyisassumed∝R−δ,be-
Both methods described above give results consistent ing R the rigidity, and δ = 0.6. We note, however, that
within 10%. this curve has been normalized by authors to the low
energy p¯/p measurements carried out in 1980s. Recent
measurements show that the sub-TeV p¯/p flux ratio is
V. RESULTS AND DISCUSSION about a factor 10 lower. The dotted line refers to a pos-
siblecontributionofantiprotonsfromtheannihilationof
The optimal energy windows for the measurement a heavy dark matter particle [12]. The short-dashedline
of the antiproton abundance in CRs is identified by shows the calculation by Blasi and Serpico [17] for sec-
9
1
Golden 79 BSP 81 ARGO-YBJ - 90% c.l.
Golden 84 Stephens 85
Bogomolov 87-90 LEAP 90
10-1 PBAR 90 MASS 91
IMAX 95 BESS 95-97
Caprice 97 CAPRICE 98
HEAT 01 L3+ 03
MACRO 03 Tibet-ASg 07
10-2 PAMELA 10
o
ti
ra10-3
p
/p
10-4
10-5
10-6
10-1 1 10 102 103 104 105
energy (GeV)
FIG. 4: The antiproton to proton flux ratio obtained with the ARGO-YBJ experiment compared with all the available mea-
surements. The solid curve refers to a theoretical calculations for a pure secondary production of antiprotons during the CR
propagation in the Galaxy by Donato et al. [14]. The long-dashed line refer to a model of extragalactic primary p¯production
[20, 23]. The rigidity-dependent confinement of CRs in the Galaxy is assumed to be ∝ R−δ, where δ = 0.6. The dotted line
refers to the contribution of antiprotons from the annihilation of a heavy dark matter particle [12]. The short-dashed line
shows the calculation by Blasi and Serpico [17] for secondary antiprotons including an additional p¯component produced and
accelerated at CR sources.
ondaryantiprotons including anadditional p¯component Manyunknownfactorscontributetoitsvalue,mostlyre-
produced and accelerated at CR sources. lated to the diffusion coefficient inside galaxy. To inves-
tigatethispoint,primaryantiprotonsareassigneddiffer-
Two sources of systematic errors have been investi-
entspectralindices,γ=2.0,2.2,2.4,2.6,2.8,3.0. Results
gated: the presence of electrons in the cosmic radiation
fortheinvestigatedmultiplicityintervalsaresummarized
andtheunknownantiprotonspectralindex. Asdiscussed
intableI.Thelimitsoftheantiproton/protonratiovaries
in[30]theabsoluterigidityscaleuncertaintiesassociated
of 20%-30% with the spectral index.
totheCRchemicalcompositionandtodifferenthadronic
modelsintheMCcalculationsareabout7%and12%re-
spectively.
TABLE I: Theeffect of different spectral indices
Like antiprotons, electrons are supposed to shift west-
ward, giving a Moon shadow opposite to that produced index 90%upperlimit(40-100) 90%upperlimit(>100)
2.0 3% 4%
bypositivelychargedCR.Nonetheless,theMoonshadow
2.2 4% 4%
provided by electrons is expected to be slightly different 2.4 4% 4%
2.6 5% 5%
from that of antiprotons. Firstly, for a given multiplic-
2.8 5% 7%
ity range the median energy of electron-induced show- 3.0 6% 7%
ers is 30%-40% less than that of CR showers, i.e. the
shadow dip is expected to be further displaced 30% -
40%. Then, the angular resolution is better for electron
primaries than for CRs (∼30%). Last, the electron flux
at TeV energies is less than 10−3 of CR flux [63]. As VI. CONCLUSIONS
a consequence, we estimate that the systematic uncer-
tainty due to misinterpreting electrons as antiprotons is
The ARGO-YBJ experiment is observing the Moon
below 10%.
shadow with high statistical significance at an energy
Since the spectral index of antiprotons is unknown, threshold of a few hundreds of GeV. Using all data col-
there is no reason to assume the proton spectral index. lected until November 2009, we set the upper limits on
10
the p¯/p flux ratio to 5% at an energy of 1.4 TeV and 6% Acknowledgments
at 5 TeV with a confidence level of 90%.
In the few-TeV energy range the ARGO-YBJ results
provide the strongest p¯/p limits obtained to date, useful
to constrain any primary antiproton production model
which foresees high fluxes at TeV energies. As dis- This work is supported in China by NSFC (No.
cussed in Section IV.A the main limiting factor in the 10120130794),theChineseMinistryofScienceandTech-
p¯/pratiomeasurementexploitingtheMoonshadowtech- nology,the Chinese AcademyofSciences,the KeyLabo-
nique is the angular resolution. The new generation of ratoryofParticleAstrophysics,CAS,andinItalybythe
EAS-arraysunderconstruction(HAWC,LHAASO)isex- Istituto Nazionale di Fisica Nucleare (INFN). We also
pectedtoimprovetheangularresolutionbyafactorof≈ acknowledge the essential supports of W. Y. Chen, G.
3 in the TeV energy range [64, 65]. Taking into account Yang, X. F. Yuan, C. Y. Zhao, R. Assiro, B. Biondo, S.
alsoan expectedincrease ofthe effective areaby at least Bricola, F. Budano, A. Corvaglia, B. DAquino, R. Es-
a factor of 3, we expect that in the next future the sen- posito, A. Innocente, A. Mangano, E. Pastori, C. Pinto,
sitivity to the p¯/p ratio could be lowered by a factor of E. Reali, F. Taurino, and A. Zerbini, in the installation,
10 in the TeV energy range. debugging, and maintenance of the detector.
[1] F.W.Stecker,in”Matter-Antimatter Asymmetry” ed.L. [28] M. Amenomori et al.,Science 314, 439 (2006).
Iconomidou-Fayard&J.TranThanhVan(ed.Hanoi:The [29] G.W. Clark, Phys.Rev.108, 450 (1957).
Gioi) 5 (2003) (preprint arXiv:hep-ph/0207323v1) [30] B. Bartoli et al., Phys.Rev.D84, 022003 (2011).
[2] M.DineandA.Kusenko,Rev.Mod.Phys.76,1(2003). [31] M. Amenomori et al.,Astrophys. J. 541, 1051 (2000).
[3] I.V.Moskalenko et al.,Astrophys.J. 565, 280 (2002). [32] G. Aielli et al.,Astrophys.J. 729, 113 (2011).
[4] I.V. Moskalenko et al., Proc. of the 28th International [33] J. Linsley, Proc. of the 19th International Cosmic Ray
Cosmic Ray Conference (ICRC 03), Tsukuba, Japan Conference (ICRC 85), La Jolla, USA (1985), Vol. 3, p.
(Universal Academy Press, Inc., Tokyo, Japan, 2003), 465.
Vol. 3, p.1921. [34] J. Lloyd-Evans, Proc. of the 19th International Cosmic
[5] A.W. Strong et al., Annu. Rev. Nucl. Part. Sci. 57, 285 Ray Conference (ICRC 85), La Jolla, USA (1985), Vol.
(2007). 2, p.173.
[6] F. Donato et al.,Astrophys.J. 563, 172 (2001). [35] M. Urban et al., Nucl. Phys. B (Proc. Suppl.) 14B, 223
[7] O.Adriani et al.,Phys. Rev.Lett. 102, 051101 (2009). (1990).
[8] O.Adriani et al.,Phys. Rev.Lett. 105, 121101 (2010). [36] D.E. Alexandreaset al.,Phys.Rev. D43, 1735 (1991).
[9] G. Kaneet al.,Phys.Lett. B681, 151 (2009). [37] M.Agliettaetal.,Proc.ofthe22thInternationalCosmic
[10] O.Adriani et al.,Nature 458, 607 (2009). Ray Conference (ICRC 91), Dublin, Ireland (1991), Vol.
[11] A.A.Abdoet al., Phys.Rev.Lett. 102, 181101 (2009). 2, p.708.
[12] M. Cirelli et al.,Nucl. Phys.B813, 1 (2009). [38] A. Borione et al.,Phys. Rev.D49, 1171 (1994).
[13] A.Bottino et al.,Phys. Rev.D58, 123503 (1998). [39] M. Merck et al.,Astrop. Phys.5, 379 (1996).
[14] F. Donato et al.,Phys. Rev.Lett. 102, 071301 (2009). [40] A. Oshima et al.,Astrop. Phys. 33, 97 (2010).
[15] C.Evolietal.,preprintarXiv:1108.0664v1[astro-ph.HE] [41] M. Chantell et al., Nature367, 25 (1994).
(2011). [42] M.Amenomorietal.,Proc.ofthe24thInternationalCos-
[16] P.Blasi, Phys. Rev.Lett. 103, 051104 (2009). micRayConference(ICRC95), Rome, Italy (University
[17] P. Blasi and P.D. Serpico, Phys. Rev. Lett. 103, 081103 of Rome, Rome, Italy,1995), Vol. 3, p.84.
(2009). [43] J.H. Cobb et al.,Phys. Rev.D61, 092002 (2000).
[18] S.A. Stephens, Proc. of the 17th International Cosmic [44] M. Ambrosio et al.,Phys. Rev.D 59, 012003 (1999).
Ray Conference (ICRC 81), Paris, France (1981), Vol. [45] M. Ambrosio et al.,Astrop. Phys. 20, 145 (2003).
4, p. 89. [46] P. Achard et al., Astrop.Phys. 23, 411 (2005).
[19] A.Barrau et al.,A&A 388, 676 (2002). [47] Y.M. Andreyevet al.,Cosmic Research 40, 559 (2002).
[20] S.A. Stephens and R.L. Golden, Space Sci. Rev. 46, 31 [48] P. Adamson et al.,Astrop. Phys. 34, 457 (2011).
(1987). [49] H. Stiebel et al., Proc. of the 32th International Cos-
[21] F.W. Stecker and A.W. Wolfendale, Nature 309, 37 mic Ray Conference (ICRC 11), Beijing, China (2011),
(1984). HE2.3 (2011) p. 1235.
[22] F.W.SteckerandA.W.Wolfendale,Proc.ofthe19thIn- [50] M. Amenomori et al.,Astrop. Phys. 28, 137 (2007).
ternationalCosmicRayConference(ICRC85),LaJolla, [51] X. Xu et al., Proc. of the 28th International Cosmic
USA (1985), Vol. 2, p.354. Ray Conference (ICRC 03), Tsukuba, Japan (Universal
[23] S.A.Stephens,A&A 149, 1 (1985). Academy Press, Inc., Tokyo, Japan, 2003), Vol. 6, p.
[24] M.Yu.Khlopov et al.,Phys.Rev. D62, 083505 (2000). 4065.
[25] A.DudarewiczandA.W.Wolfendale, MNRAS268, 609 [52] G.Aiellietal.,Nucl.Instrum.MethodsPhys.Res.,Sect.
(1994). A 562, 92 (2006).
[26] M.Yu.Khlopov, Gravitation & Cosmology 4, 69 (1998). [53] C.Baccietal.,Nucl.Instrum.MethodsPhys.Res.,Sect.
[27] G. Guillian et al.,Phys. Rev.D75, 062003 (2007). A 443, 342 (2000).