Latest results from the ARGO-YBJ experiment
aa r X i v : . [ a s t r o - ph . H E ] M a r Latest results from the ARGO-YBJ experiment
Di Sciascio Giuseppe on behalf of the ARGO-YBJ Collaboration
INFN - Sezione Roma Tor Vergata, Roma, ItalyE-mail: [email protected]
Abstract.
The ARGO-YBJ experiment has been in stable data taking for 5 years at theYangBaJing Cosmic Ray Observatory (Tibet, P.R. China, 4300 m a.s.l., 606 g/cm ). With aduty-cycle greater than 86% the detector collected about 5 × events in a wide energy range,from few hundreds GeV up to about 10 PeV. A number of open problems in cosmic ray physicshas been faced exploiting different analyses. In this paper we summarize the latest results incosmic ray physics and in gamma-ray astronomy.
1. Introduction
Aiming to face the open problems in Galactic cosmic ray (hereafter CR) physics through acombined study of photon- and charged particle-induced extensive air showers (hereafter EAS)with the same detector, ARGO-YBJ has been in stable data taking for more than 5 years at theYangBaJing Cosmic Ray Observatory (Tibet, P.R. China, 4300m a.s.l., 606 g/cm , 90.5 ◦ East,30.1 ◦ North). The detector acted simultaneously as a wide aperture ( ∼ sr), continuosly-operated γ -ray telescope at sub-TeV – TeV photon energies and as a high resolution CR detector in thebroad energy range between few TeV and 10 PeV.The combination of the high elevation of the site, the full coverage of the central carpet andthe high granularity of the readout provides very important advantages in various aspects ofCR physics. In fact, in addition to the decrease of the threshold energy down to few hundredsGeV, the location of the esperiment above 4000 m a.s.l. ensures that shower fluctuations aresmall (we are working in the shower maximum region) and that all nuclei produce showers withnearly the same electromagnetic size. The low energy threshold is crucial for the overpositionwith direct measurements carried out by balloon/satellite-born detectors allowing one to cross-calibrate the different energy scales. The nearly independence of the size on the mass of theprimary determinates the size – energy relation to be better defined, allowing the study of theelemental composition around the knee in a very efficient and reliable way.In this paper the latest results obtained by ARGO-YBJ in CR physics and in gamma-rayastronomy will be briefly summarized.
2. The ARGO-YBJ experiment
ARGO-YBJ is a full coverage air shower detector constituted by a central carpet ∼ ×
78 m ,made of a single layer of resistive plate chambers (RPCs) with ∼
93% of active area, enclosed bya guard ring partially instrumented ( ∼ ∼ ×
110 m . The apparatus has a modularstructure, the basic data acquisition element being a cluster (5.7 × ), made of 12 RPCs(2.85 × each). Each chamber is read by 80 external strips of 6.75 × (the spatialixels), logically organized in 10 independent pads of 55.6 × which represent the timepixels of the detector [1]. The readout of 18,360 pads and 146,880 strips is the experimentaloutput of the detector. In addition, in order to extend the dynamical range up to PeV energies,each chamber is equipped with two large size pads (139 ×
123 cm ) to collect the total chargedeveloped by the particles hitting the detector [2]. The RPCs are operated in streamer modeby using a gas mixture (Ar 15%, Isobutane 10%, TetraFluoroEthane 75%) for high altitudeoperation [3]. The high voltage settled at 7.2 kV ensures an overall efficiency of about 96% [4].The central carpet contains 130 clusters and the full detector is composed of 153 clusters for atotal active surface of ∼ . The total instrumented area is ∼ . For each eventthe location and timing of every detected particle is recorded, allowing the reconstruction of thelateral distribution and the arrival direction. The trigger is based on a time correlation amongthe pad signals depending on their relative distance. In this way, all the shower events givinga number of fired pads N pad ≥ N trig in the central carpet in a time window of 420 ns generatethe trigger. The whole system has been in stable data taking from November 2007 to January2013, with the trigger condition N trig = 20 and a duty cycle ≥ ∼ − , keeping good linearity up to a core density of about 15 particlesm − . This high granularity allows a complete and detailed three-dimensional reconstruction ofthe front of air showers at an energy threshold of a few hundreds GeV. Showers induced by highenergy primaries ( >
100 TeV) are also imaged by the charge readout of the large size pads whichallows to study the structure of the shower core region up to particle densities of ∼ /m [2].Details on the analysis procedure (e.g., reconstruction algorithms, data selection, backgroundevaluation, systematic errors) are discussed in [5, 6, 7]. The performance of the detector(angular resolution, pointing accuracy, energy scale calibration) and the operation stability arecontinuously monitored by observing the Moon shadow, i.e., the deficit of CRs detected in itsdirection [6, 8]. The measured angular resolution is better than 0.5 ◦ for CR-induced showerswith energy E > ∼ ◦ . According toMonteCarlo (MC) simulations the angular resolution for γ -induced events results smaller by30% - 40%. The absolute rigidity scale uncertainty of ARGO-YBJ is estimated at 10% level inthe range 1 - 30 TeV/Z [6, 8].
3. Cosmic Ray Anisotropy
The CR arrival direction distribution and its anisotropy has been a long-standing problem eversince the 1930s. The study of the anisotropy is complementary to the study of their energyspectrum and elemental composition to understand CR origin and propagation. It is also apowerful tool to probe the structure of the magnetic fields through which CRs travel.The anisotropy in the CR arrival direction distribution hs been observed by differentexperiments with increasing sensitivity and details at different angular scales. Currentexperimental results show that the main features of the anisotropy are uniform in the energyrange (10 - 10 eV), both with respect to amplitude (10 − - 10 − ) and phase ((0 - 4) hr).The existence of two distinct broad regions, one showing an excess of CRs (called “tail-in”),distributed around 40 ◦ to 90 ◦ in R.A., the other a deficit (the “loss cone”), distributed around150 ◦ to 240 ◦ in R.A., has been clearly observed (for a review see, for example, [9]). In thelast years different experiments reported evidence of the existence of a medium angular scaleanisotropy in the both hemispheres [9].In Fig. 1 the ARGO-YBJ sky map of medium angular scale (order of 10 ◦ ) anisotropy ingalactic coordinates as obtained with 4.5 years data is shown (for details see [10, 11]). Themap center points towards the galactic Anti-Center. The zenith angle cut ( θ ≤ ◦ ) selectsthe declination region δ ∼ -20 ◦ ÷ ◦ . In this zenith angle bin the angular resolution is nearly ◦ ◦ -15 -12 -9 -6 -3 0 3 6 9 12 15 Figure 1.
ARGO-YBJ sky-map in galacticcoordinates. The color scale gives thestatistical significance of the observation instandard deviations. The map center pointstowards the galactic Anti-Center.
Energy (GeV) . ( G e V ) - . s . s r ) ( m . E × F l u x WFCT02-ARGO P+HeARGO P+HeCREAM P+HeCREAM PCREAM He KASCADE P+He QGSJETKASCADE P+He SIBYLLTibet-Phase II P+He QGSJETTibet-Phase II P+He SIBYLLTibet-2000 P QGSJET+HDTibet-2000 P QGSJET+PD
Figure 2.
Light component (p+He) en-ergy spectrum of primary CRs measured byARGO-YBJ compared with other experimen-tal results. The results obtained by theARGO-YBJ/WFCTA system are shown bythe filled red squares [23].constant and the energy of selected events is in the TeV range. According to the simulation,the median energy of the isotropic CR proton flux is E p ≈ ≈ ∼ ◦ implies multipoles of order 18). So far, no theory of CRs in the Galaxy exists which isable to explain the origin of these different anisotropies leaving the standard model of CRs andthat of the local Galactic Magnetic Field unchanged at the same time. The anisotropy problemis the most serious challenge to the standard model of the origin of galactic CRs from diffusiveshock acceleration [13].
4. Measurement of the Cosmic Ray Energy spectrum
There is a general consensus that Galactic CRs up to the all-particle knee originate in SNRsaccelerated by the first order Fermi mechanism in shock waves. The theoretical modelling of thismechanism can reproduce, in principle, the measured spectra and composition of CRs. RecentlyAGILE and Fermi observed GeV photons from two young SNRs (W44 and IC443) showing thetypical spectrum feature around 1 GeV (the so called “ π bump”, due to the decay of π ) relatedto hadronic interactions [14, 15]. This important measurement however does not demonstratehe capability of SNRs to accelerate CRs up to the knee and above.In the standard picture, mainly based on the results of the KASCADE esperiment, the kneeis attributed to the steepening of the p and He spectra [16]. However, a number of experimentsreported evidence that the bending of the light component (p+He) is well below the PeV andthe knee of the all-particle spectrum is due to heavier nuclei (see, for example, [17, 18, 19].A large number of theoretical papers discussed the highest energies achievable in SNRs and thepossibility that protons can be accelerated up to PeVs (for a recent review see [20] and referencestherein). Therefore, the determination of the proton knee, as well as the measurement of theevolution of the heavier component across the knee, are the key components for understandingorigin and acceleration mechanisms of Galactic CRs.A measurement of the CR primary energy spectrum (all-particle and light component) in theenergy range few TeV – 10 PeV is under way with the ARGO-YBJ experiment. To cover thiswide energy range different ’eyes’ have been used: • ’digital readout’ , based on the strip multiplicity, i.e. the picture of the EAS provided by thestrip/pad system, in the few TeV – 200 TeV energy range [21]; • ’analog readout’ , based on the particle density in the shower core region, in the 100 TeV –10 PeV range [22]; • ’hybrid measurement’ , carried out by ARGO-YBJ and a wide field of view Cherenkovtelescope, in the 100 TeV - PeV region [23].In the following we summarize results obtained below 100 TeV with the digital readout anddescribe the hybrid measurement up to the PeV range. Preliminary results obtained with theARGO-YBJ analog data are described in [22]. As described in [21], requiring quasi-vertical showers ( θ < ◦ ) and applying a selection criterionbased on the particle density, a sample of events mainly induced by p and He nuclei, with showercore inside a fiducial area (with radius ∼
28 m), has been selected. The contamination by heaviernuclei is found negligible. An unfolding technique based on the Bayesian approach has beenapplied to the strip multiplicity distribution in order to obtain the differential energy spectrumof the light component.The spectrum measured by ARGO-YBJ is compared with other experimental results in Fig. 2(blue inverted triangles). The ARGO-YBJ data agree remarkably well with the values obtainedby adding up the p and He fluxes measured by CREAM both concerning the total intensities andthe spectral index [24]. The value of the spectral index of the power-law fit to the ARGO-YBJdata is -2.61 ± × ◦ × ◦ with a pixel size of approximately 1 ◦ [26].From December 2010 to February 2012, in a total exposure time of 728,000 seconds, theARGO-YBJ/WFCTA system collected and reconstructed 8218 events above 100 TeV accordingto the following selection criteria: (1) reconstructed shower core position located well inside theARGO-YBJ central carpet, excluding an outer region 2 m large; (2) more than 1000 fired padson the central carpet; (3) more than 6 fired pixels in the PMT matrix; (4) a space angle betweenthe incident direction of the shower and the telescope main axis less than 6 ◦ . This selectionguarantees that the Cherenkov images are fully contained in the FOV, an angular resolutionbetter than 0.2 ◦ and a shower core position resolution less than 2 m. p0 1 2 3 L p -2-1012 ProtonHeliumCNOMgAlSiIron
Figure 3.
Scatter plot of the parameters p C and p L for showers induced by differentnuclei. The primary masses have beensimulated in the same fraction, assuming a -2.7 spectral index in the energy range 10 TeV– 10 PeV. (E/TeV) Log ) - s r - s - m . ( G e V Ω d A d t d E dd N × . E ARGO-YBJ G4ARGO-YBJ G1ARGO-YBJ Bayes-G4ARGO-YBJ Bayes-G1ARGO-YBJ WFCTA (p + He)ARGO-YBJ strip (p + He)Horandel (p + He) 2003 1 PeV × knee at Z Horandel (p+He) 2003 Gaisser et al. 2013 (p + He) Preliminary
Figure 4.
Light (p+He) component energyspectrum of primary CRs measured byARGO-YBJ with different analyses. Thesystematic uncertainty is shown by theshaded area and the statistical one by theerror bars. The parametrizations given in [28,29] are shown for comparison. A Horandel-like spectrum with a modified knee at Z × max recorded by a RPCin a given shower is a useful parameter to measure the particle density in the shower core region,i.e. within 3 m from the core position. For a given energy, in showers induced by heavy nucleiN max is smaller than in showers induced by light particles. Therefore, N max is a parameteruseful to select different primary masses. In addition, N max is proportional to E . rec , where E rec is the shower primary energy reconstructed using the Cherenkov telescope. We can define a newparameter p L = log ( N max ) − . · log ( E rec /T eV ) by removing the energy dependence [23].The Cherenkov footprint of a shower can be described by the well-known Hillas parameters[27], i.e. by the width and the length of the image. Older showers which develop higher in theatmosphere, such as iron-induced events, have Cherenkov images more stretched, i.e. narrowerand longer, with respect to younger events due to light particles which develop deeper. Therefore,the ratio between the length and the width (L/W) of the Cherenkov image is expected to beanother good estimator of the primary elemental composition.Elongated images can be produced, not only by different nuclei, but also by showers withthe core position far away from the telescope, or by energetic showers, due to the elongationof the cascade processes in the atmosphere. Simulations show that the ratio of L/W is nearlyproportional to the shower impact parameters R p , the distance between the telescope and thecore position, which must be accurately measured. An accurate determination of the showergeometry is crucial for the energy measurement. In fact, the number of photoelectrons collectedin the image recorded by the Cherenkov telescope N pe varies dramatically with the impactparameter R p , because of the rapid falling off of the lateral distribution of the Cherenkov light.Only an accurate measurement of the shower impact parameters R p , and a good reconstructionof the primary energy allow to disentangle different effects. A shower core position resolutionbetter than 2 m and an angular resolution better than 0.2 ◦ , due to the high-granularity ofthe ARGO-YBJ full coverage carpet, allow to reconstruct the shower primary energy with aresolution of 25%, by using the total number of photoelectrons N pe . The uncertainty in absoluteenergy scale is estimated about 10%.Therefore, in order to select the different masses we can define another new parameter p C = L/W − . · ( R p / m ) − . · log ( E rec /T eV ) by removing both the effects due to the showeristance and to the energy. The values of these parameters for showers induced by differentnuclei are shown in the Fig. 3. As can be seen from the figure, a suitable selection in the p L – p C space allows to pick out a light composition sample with high purity. In fact, by cuttingoff the concentrated heavy cluster in the lower-left region in the scatter plot, i.e. p L -0.91and p C ∼
170 m sr above 100 TeV and shrinks to ∼
50 m sr after theselection of the (p+He) component.The light component energy spectrum measured by the ARGO-YBJ/WFCTA hybrid systemis shown in the Fig. 2 by the filled red squares. A systematic uncertainty in the absolute flux of15% is shown by the shaded area. The error bars show the statistical errors only. The spectrumcan be described by a power law with a spectral index of -2.63 ± ± × − GeV − m − sr − s − .This result is consistent for what concern spectral index and absolute flux with themeasurements carried out by ARGO-YBJ below 200 TeV and by CREAM. The flux difference isabout 10% and can be explained with a difference in the experiments energy scale at level of 4%.The measurement of the west-ward displacement of the Moon shadow under the effect of thegeomagnetic field, as a function of the event multiplicity, allowed to calibrate the relation showersize - primary energy, thus calibrating the absolute energy scale of the detector at 10% levelbelow 20 TeV [6]. Above this energy the overposition with CREAM provides a solid anchorageto the absolute energy scale at few percent level.In order to extend the measurement of the ARGO-YBJ/WFCTA hybrid experiment to thePeV, the selection cuts in the p L – p C space have been modified as follows: events for whichp L -1.25 and p C ∼ to 10 eV are required to observe the knees of different nuclei and to investigate indetail the end of the spectrum of Galactic CRs and the contradictory results among differentexperiments.
5. The Fermi Cocoon in the Cygnus region
The Cygnus X region is one of the most luminous region of the Northern γ -ray sky and it isrich in potential CR accelerator sites, e.g. Wolf Rayet stars, OB associations and SNRs. TheFermi collaboration detected a complex extended source, in a position consistent with the sourceARGO J2031+4157, attributed to the emission of freshly accelerated CRs interacting with gasand radiation, filling a bubble (a “cocoon”) carved by stellar winds and multiple supernovaeshock waves [30].The significance map around ARGO J2031+4157 as observed by ARGO-YBJ is shown inthe left plot of Fig. 5. To improve the sensitivity to γ -induced events, an optimizated selectionbased on the shower core position is applied [31]. A smoothing is applied with an energy-dependent point-spread function evaluated for point γ -ray sources. For comparison, the knownTeV sources and the Cygnus Cocoon are marked in the figure. The sizes of markers indicate the alactic longitude (deg) G a l ac t i c l a t i t ud e ( d e g ) -6-4-20246 -4-20246 S i gn i f i ca n ce MGRO J2031+41Cygnus CocoonARGO J2031+4157 VER J2019+407TeV J2032+4130
Energy (eV) ) - s - F l u x ( e r g s c m E -12 -11 -10 -9 Fermi-LATARGO-YBJMilagro =150 TeV c Model:E =40 TeV c Model:E
Figure 5.
Left plot: The significance map of the ARGO J2031+4157 region observed bythe ARGO-YBJ experiment. Right plot: Spectral energy distribution of the Cygnus Cocoon.Different markers stand for the spectra measured by different detectors. The dot-dashed lineshows the best fit to the Fermi-LAT and ARGO-YBJ data using a simple power-law function.The red solid (dashed) line is obtained by a hadronic model with a proton cutoff energy at 150(40) TeV. For references and details see [31].68% containment size of the extension (for details see [31]).The energy spectrum of the Cygnus Cocoon measured by Fermi-LAT, ARGO-YBJ andMilagro is shown in the right plot of Fig. 5. The flux determined by ARGO-YBJ appearsconsistent with the extrapolation of the Fermi-LAT spectrum suggesting that the emission ofARGO J2031+4157 can be identified as the counterpart of Cygnus Cocoon at TeV energies.The combined spectrum of Fermi-LAT and ARGO-YBJ can be described by a differentialpower law (dot-dashed line) dN/dE = (3 . ± . × − · ( E/ . T eV ) ( − . ± . photons cm − s − TeV − , suggesting the same origin for both GeV and TeV extended gamma-ray emission.Only statistical errors are shown, the systematic errors on the flux are estimated to be less than30%. The upper limits of Fermi-LAT and ARGO-YBJ indicate the presence of a slope changeor cutoff below ∼ ∼
10 TeV, respectively. At a distance of 1.4 kpc, the observedangular extension of about 2 ◦ corresponds to more than 50 pc, making the Cygnus Cocoon thelargest identified Galactic TeV source. The observation of this large TeV source, not removedin the Milagro analysis of the γ -ray diffuse flux from the Cygnus region, can explain the excessobserved by them in the region 65 ◦ < l < ◦ , | b | < ◦ .Such a large region can be related to different scenarios. As discussed in [30], the favoredscenario to explain the emission in the Cygnus Cocoon is the injection of CRs via accelerationfrom the collective action of multiple shocks from supernovae and the winds of massive stars,which form the Cygnus superbubble. Such superbubbles have been long advocated as CRfactories, therefore the Cygnus Cocoon could be the first evidence supporting such hypothesis.In order to test a possible hadronic origin of the γ -ray emission through the π decay, weconsidered inelastic collisions between accelerated protons and target gas. We assumed that theprimary protons follow a power law with an index similar to the gamma spectrum and withan exponential cutoff at 150 TeV, as suggested in [30] to describe CR acceleration by randomstellar winds in the Cygnus superbubble. This energy is the maximum proton cutoff allowed bythe ARGO-YBJ upper limit. The resulting spectrum is shown in the right plot of Fig. 5 by thesolid red line [31]. It is worth noting that the Milagro data are not described by this model andcan be reconciled only with a cutoff of about 40 TeV (dashed red line) non consistent with theRGO-YBJ results.
6. Conclusions
The ARGO-YBJ experiment has been in stable data taking for more than 5 years at theYangBaJing Cosmic Ray Observatory. With a duty-cycle greater than 86% the detectorcollected about 5 × events in a wide energy range, from few hundreds GeV up to about10 PeV, exploiting the performance of the RPCs to image the front of atmospheric showers withunprecedented resolution and detail.Since November 2007 to January 2013 ARGO-YBJ monitored the Northern sky at TeVphoton energies with a cumulative sensitivity ranging from 0.24 to ∼ References [1] Aielli G. et al. 2006
NIM
A562
Astropart. Phys. NIM
A443
NIM
A608
Astrophys. J. Lett.
L208.[6] Bartoli B. et al. 2011
Physics Rev. D Astrophys. J.
Physics Rev. D ”On the observation of the Cosmic Ray Anisotropy below 10 eV” , in ”Homageto the Discovery of Cosmic Rays, the Meson-Muons and Solar Cosmic Rays”, Chapter 9, pagg. 221-257(Nova Science Publishers, Inc., New York, 2013). Preprint: arXiv:1407.2144.[10] Iuppa R. and Di Sciascio G., 2013 Astrophys. J.
Physics Rev. D JCAP J. Phys. G R95.[14] Giuliani A. et al. 2011
Astrophys. J. Lett.
L30[15] Ackermann M. et al. 2013
Science
Astropart. Phys. Phys. lett. B
Astropart. Phys. Astropart. Phys. Astron. Astrophys. Rev. Physics Rev. D et al Frascati Physics Series Chinese Phys. C Astrophys. J.
Chinese Phys. C NIM
A629
Proc. 19th ICRC (La Jolla) Astropart. Phys. Frontiers of Physics Science
Astrophys. J.790