Knee @KEK, 2008425

Cosmic Ray Energy SpectrumM.Nagano, A.A.Watson (2000)

Cosmic Ray Energy SpectrumSommers (ICRC2001)

Experimentsiteg/cm2ehCAKENOJapan (35.5N, 138.5E)9301 GeVBLANCAUtah(40.2N,112.8W)870CASA-MIAUtah (40.2N,112.8W)870800 MeVDICE860800 MeVEAS-TopItaly (42.5N,13.6E)8201 GeVHEGRALa Palma (28.8N,17.9W)790KASCADE (electrons/muons)Germany (49.N, 8.E)1022230 MeVKASCADE (hadrons/muons)1022230 MeV50 GeVKASCADE (neural network)1022230 MeVMSU1020Mt. Norikura Japan735TibetTibet (30.1N,90.5E)606Tunka-13680Yakutsk (low energy)1020

All particle spectrumKnee around 3-5 PeVICRC2003 M. Takita

All particle energy spectrum

ICRC2007 Y. Tsunesada (BASJE)Energy dependence of< ln A>

Research purposeThus, measurements of the primary cosmic rays around the "knee" are very important and its composition is a fundamental input for understanding the particle acceleration mechanism that pushes cosmic rays to very high energies. According to the Fermiacceleration with supernovablast waves, the accelerationlimit EmaxZ * 100 TeV.The position of "knee" must be dependent on electric charge Z

KASCADE e/mHadron

Energy Spectrum of Single Elements

Kascade data BUT

KASCADE : Astroparticle phys. 24 (2005) 1-25

KASCADE : Astroparticle phys. 24 (2005) 1-25

TIBETYangbajing , Tibet, China 9053E, 3011N, 4,300 m a.s.l. (606g/cm2)

BD&ECAir Shower arrayPhys. Lett. B. 632(2006)58Tibet-II Air Shower array

Tibet-I to Tibet-II/HDNumber of detector I : 45 II : 185 HD: 109Mode Energy I : 10 TeV II : 10 TeV HD: 3 TeVArea I : 7 ,650 m2 II : 37,000 m2 HD: 5,200 m2

Characteristics of the Tibet Hybrid Experiment High altitude (4300m a.s.l. 606 g/cm2). Energy determination is made under minimum chemical- composition dependence around the knee.

Observe core structure by burst detectors (BD) & emulsion chambers (EC) Select air showers of light-component origin by high energy core detection. (A2/3) Young showers are mostly of proton and helium origins. Air shower axis is known with r < 1m. Ne and s are determined precisely.

Smaller interaction-model dependence for forward region than backward.

2nd particle density2nd particle timing Cosmic ray energyCosmic ray directionAir Shower Detection(ns)~10 TeV

NeNKG~31016eV

Constant fitting-0.0034o 0.011oSystematic pointing error < 0.01oAbsolute EnergyScale error 4.4% +- 7.9%stat +- 8%sysEnergy dependence ofDisplacementsCaused by Geomagnetic field

Verification Absolute energy scale Pointing errorCosmic Ray Energy Calib. by the Moon Shadow by Tibet-III

EC and BD Total EC area : 80 m2

EC and BDA structure of each EC used here is a multilayered sandwich of lead plate and photosensitive x-ray films, photosensitive layers are put every 2 (r.l.) (1 r.l.=0.5cm) of lead in EC. Total thickness of lead plates is 14 r.l.2) g family is mostly cascade products induced by high energy p0 decay g- rays which are generated in the nuclear interactions at various depths. 3) It is worthwhile to note that the major behavior of hadronic interactions as well as the primary composition are fairly well reflected on the structure of the family observed with EC.

-M.C.Simulation-Hadronic int.modelCORSIKA ( Ver. 6.030 ) QGSJET01 SIBYLL2.1 Primary composition modelHD (Heavy Dominant)PD (Proton Dominant)

The experimental conditions for detecting g family (Eg >= 4TeV, Ng>=4, SEg >=20 TeV) events with EC are adequately taken into account. For example, our EC has a roof, namely, the roof simulation and EC simulation are also treated.

HD model1014eV1015eV1016eVProton22.611.08.1He19.211.48.4Iron22.239.151.7Other35.638.231.7

PD model1014eV1015eV1016eVProton39.038.137.5He20.419.419.1Iron9.49.910.2Other30.431.733.0

HD modelPD modelPrimary composition model

Model Dependence of g-family (Generation+Selection) Efficiency in EC QGSJET SIBYLLSIBYLL/QGSJET~1.3SIBYLL/QGSJET1.3SIBYLLQGSJETSIBYLLQGSJET

Model Dependence of Air Shower Size Accompanied by g-family

Procedures to ObtainPrimary Proton Spectrum( g-family selection criteria : Emin=4TeV, Ng=4, sumE >=20TeV, Ne >=2x105 )

AS+ECfamily matching event ANN Proton identification(Correlations)(Eg,Ng,< R >,,sec(), Ne )

Int. modelsQGSJETExpt.(80m2)(1996-1999)(699days)SIBYLLExpt.(80m2)(1996-1999)(699days)PrimaryHDPDHDPDTotal sampling primary2x1081x1082x1081x108

Number of g-family5252730317768019655177Selected by ANN(T

Event Matching between EC+BD+ASAS+ECfamily matching event ANN Proton identification(Correlations)(Eg,Ng,< R >,,sec(), Ne )

AS&family matching bytime coincidence, Nburst>105 and test177 ev selected192 + 14 ev expected

Fractions of P, He, M, Fe components (MC) making air showers accompanied by -families

Selection of proton-induced events by Artificial Neural Network (ANN)

sumE Total energy ECNg number of ganma family EC< R > ( mean lateral spread (< R > (H) / EC) mean energy flow spread ECsec() ( Zenith angle of gamma family ECNe Shower size of the tagged air showers AS

Selection of proton-induced events with ANN

Parameters for training( sumE, Ng, < R >, , sec(), Ne )

Target value for protons=0 others=1Define threshold value TthSelection efficiency of protonevents as a function of Tth Efficiency~75%Tth=0.4Purity~85%Target Value (T)

Comparison of Target Value Distribution. between DATA and MC

Back check: Selection of proton-induced events by ANN

Air shower size spectrum of p-like events vs MC (for proton like events (ANN out-put

Primary energy estimation ( for proton like events )( 1.0 < sec(theta)

Back check: Conversion factor for p-like EV ( by QGSJET + HDANN out-put

Energy resolution

Primary proton spectrum

Preliminary(KASCADE data: astro-ph/0312295)AllProtonKASCADE (P)Present Results(a) ( by QGSJET model) (b) ( by SIBYLL model )

Primary helium spectrum(a) (by QGSJET model)(b) (by SIBYLL model)

Primary ratio TibetKASCADE(a) (by QGSJET model)(b) ( by SIBYLL modelAll (P+He)All

Tibet IIIAS array + Burst Detector 733 ScintillatorsBurst hut80 m2 coverage by 100 burst detectors.

Phase II hybrid experimentScintillator 50cm x 160cm x 2cm.viewed with 4 PhotoDiodes.Measure size and position of the burst (e.g., e.m. cascade)Electromagnetic component over GeV is responsible for burst size.Scint. was calibrated by accelerator beam.

Proton+Helium spectrumPhase IPhase IPhase II

Proton+Helium spectrumPhase IPhase II

Tibet AS(~8.3m2) +MD(384ch, ~104m2)Tibet AS + MDgTibet AS+YAC(1~5m2)YACKnee p, He, Fe100TeVg

Summary ( 1 ) All particle E spectrum -> KASCADE ~= Tibet

( 2 ) Composition KASCADE: small sstat, but large ssyst(2~5) x100% Rigidity scenario not confirmed All particle knee bend by light elements Tibet: Large sstat(~10%), but small ssyst (~30% for p) The knee of all particle spectrum is composed of nuclei heavier than P + He

=== Research purpose == According to the Fermi acceleration with supernova blast waves, the acceleration limit EmaxZ * 100 TeV.The position of "knee" must be dependent on electric charge Z Thus, measurements of the primary cosmic rays around the "knee" are very important and its composition is a fundamental input for understanding the particle acceleration mechanism that pushes cosmic rays to very high energies.

The characteristics of the Tibet Hybrid Experiment can be summarized as follows.The first point is that the composition dependence of the energy determination around the knee can be minimized, because we can observe near shower maximum at Tibet altitude irrespective of the primary species.The second is that we can observe core structure by the burst detector which enables us to select air showers of light-component origin due to their penetrating feature. Young showers starting from deep first interaction point are purely proton and helium origin as shown later. Another merit of the core detector is that it helps to determine air-shower size and age precisely because the shower axis is known within an accuracy of 1m. The third point is the smaller interaction-model dependence for forward region than backward in the center of momentum system where target effect is important for backward region. The model difference on target effect is still large at present.

There are emulsion chambers and burst detectors in the emulsion chamber room as shown in Fig. . the total EC area is 80 m2.

There are the structure of the Emulsion Chambers and Burst detectors as shown in Fig. .1) A structure of each EC used here is a multilayered sandwich of lead plate and photosensitive x-ray films, photosensitive layers are put every 2 (r.l.) (1 r.l.=0.5cm) of lead in EC. Total thickness of lead plates is 14 r.l.2) g family is mostly cascade products induced by high energy p0 decay g- rays which are generated in the nuclear interactions at various depths. 3) It is worthwhile to note that the major behavior of hadronic interactions as well as the primary composition are fairly well reflected on the structure of the family observed with EC.

Hadronic int.modelCORSIKA ( Ver. 6.030 ) QGSJET01 SIBYLL2.1 Primary composition modelHD(Heavy Dominant)PD(Proton Dominant)The experimental conditions for detecting g family ( Emin = 4TeV, Ng>=4, SE>=20 TeV) events with EC are adequately taken into account. For example, our EC has a roof, namely, the roof simulation and EC simulation are also treated.

Comparison of the family production efficiency for primary protons between QGSJET and SIBYLL models as shown in figures. The figures show that the proton efficiency by SIBYLL model is roughly 30% higher than QGSJET model in the energy region related to this analysis.

How to obtain proton spectrum? Event matching between a gamma family event in EC and the accompanied air shower event is done via BD information,because EC has no time information. As discussed above, Tibet ASr experiment can detect a r family event accompained by an air-shower in the knee region, Using a neural network, we can further enrich the proton-induced events by the following parameters, which discriminate protons from other nuclei,for example lateral spread and air shower size ....and from Monte Calro simulation, we can estimate the primary proton energy, thus, we can obtain the primary proton spectrum.

How to obtain proton spectrum? Event matching between a gamma family event in EC and the accompanied air shower event is done via BD information,because EC has no time information. As discussed above, Tibet ASr experiment can detect a r family event accompained by an air-shower in the knee region, Using a neural network, we can further enrich the proton-induced events by the following parameters, which discriminate protons from other nuclei,for example lateral spread and air shower size ....and from Monte Calro simulation, we can estimate the primary proton energy, thus, we can obtain the primary proton spectrum.

Using ANN, we can select proton-induced events by the following parameters. ###### CHECK ####

#The target value is 0 for protons 1 for other nuclei, that is, when the target value is smaller than 0.4, the event is assigned as a proton-like event and when the target value larger than 0.4, the event is assigned as a heavy-like event.

Define threshold value Tth", we can get selection efficiency of proton events as a function of T. thus, we can use it to caculate the primary proton spectrum.

On the basis of all the discussions above, we estimated the primary proton spectrum as shown in Fig. by QGSJET, the red circle is our results by QGSJET + HD model, and the blue circle is our results by QGSJET + PD model. and there are all particle spectrum by the diffrent experiment, and there is the Gaisser line fit with lower energy region. and there is the results of Kascade by QGSJET model.and We also estimated the primary (All-(p+He))/All component as shown in Fig. by QGSJET, there are our results are compared with the other experiment, there is the results of Kascade by QGSJET model.This slide illustrates the detector configuration. There are 733 scintillators with 7.5m spacing and 100 burst detectors of the coverage area 80 m^2 located near the center of air-shower array.Burst detector used in the second phase hybrid experiment consists of lead plate of 7 r.l. , iron base of 1cm thick and scintillator of 2cm thick. The area of scintillator is 50cm x 160cm and 4 photodiodes are attached at every corner. The scintillator was calibrated using accelerator of 1 GeV electron beam. Knowing the attenuation length of the scintillator, one can determine the size and the center position of the cascade showers under the lead. Electromagnetic component over GeV is responsible for burst size. Therefore, we have much lower detection threshold than emulsion chamber at the sacrifice of the spatial resolution without using X-ray films.This is the result of the P+He spectrum obtained by an analysis using QGSJET+HD model. Present result agrees well with phase I experiment of Tibet-BD and Tibet-EC. The statistics of new data is higher than old data by one order. The plots of other works are the sum of the proton spectrum and helium spectrum calculated from their original papers.

We repeated the all procedure of the analysis using SIBYLL interaction model. Again the agreement between new data and old data is within statistical errors.