Gamma-ray Blazars and the Cosmic Background Radiation

Yoshiyuki Inoue (JAXA International Top Young Fellow)

PACIFIC 2015, Tahiti, 2015-09-15

10-10

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-6 10-4 10-2 100 102 104 106 108 1010 1012

E2 dN/

dE [e

rg cm

-2 s-1

sr-1

]

Photon Energy [eV]

Galaxies (Inoue et al. ’13)Pop-III Stars (Inoue et al. ’13)

AGNs (All)Radio-quiet AGNs (Inoue et al. ’08)

Blazars (Inoue and Totani ’09)Radio Galaxies (Inoue ’11)

Cosmic Background RadiationCMB

Galaxies

AGNs?

YI ’14

Cosmic Infrared Background

Cosmic Optical & Infrared Background (COB & CIB)

0.1

1

10

100

0.1 1 10 100 1000 10000h[µm]

iIi [

nW/m

2 /sr]

Madau & Pozzetti ’00 (HST)Elbaz et al. ’02 (ISO)

Papovich et al. ’04 (Spitzer)Fazio et al. ’04 (Spitzer)Xu et al. ’05 (GALEX)Dole et al. ’06 (Spitzer)

Frayer et al. ’06 (Spitzer)Gardner et al. ’00 (HST)

Berta et al. ’11 (Hershel/PEP)Wright & Reese ’00 (DIRBE)

Wright ’04 (DIRBE)Levenson et al. ’07 (DIRBE)

Levenson & Wright ’08 (DIRBE)Bernstein ’07 (HST)

Matsuoka et al. ’11 (Pioneer)Matsumoto et al. ’11 (IRTS)Matsuura et al. ’11 (AKARI)

CambrØsy et al. ’01 (DIRBE)Dwek & Arendt ’98 (DIRBE)

Gorijian et al ’00 (DIRBE)Finkbeiner et al ’00 (DIRBE)

Hauser et al ’98 (DIRBE)Lagache et al ’00 (DIRBE)

Edelstein et al ’00 (Voyger)Brown et al ’00 (HST/STIS)

Albert et al ’08 (MAGIC)(X, zc)=(1.0, 0.0)

(X, zc)=(50.0, 10.0)(X, zc)=(100.0, 10.0)

Stars Dust

YI+‘13a

Zodiacal Light

• Scattered solar emission by interplanetary dust (NIR)

• Interplanetary dust distribute around the plane of the ecliptic

• Brightest foreground emission for the COB/CIB measurement

黄道光

• 惑星間ダストにより散乱された太陽光　（近赤外線）

• 黄道光は空間分布を持つ→惑星間ダストは、黄道面に集中的に分布

• 散乱により偏光する

• 最も明るい前景放射（~60%をしめる）→大きな統計誤差の要因　　　　　　　→黄道光の理解が必要

!

黄道面

SunEarth

SunEarth

惑星間ダスト

黄道面

dust

the plane of the ecliptic

Zodiacal light

© Arai

Foregrounds for COB & CIB

• Foreground: Zodiacal light, Diffuse galactic light, Star light.

The Astrophysical Journal, 737:2 (19pp), 2011 August 10 Matsuura et al.

0.0001

0.001

0.01

0.1

1

10

100

0001001011wavelength [µm]

Sur

face

bri

ghtn

ess

I ν [MJy

/sr]

ZL

BMCEZ

SLDGL

ISD

Figure 1. Summary of the results of previous CIB measurements in the infrared and submillimeter wavelength ranges with from the following space missions:COBE/DIRBE (diamonds; Hauser et al. 1998) as upper limits, COBE/FIRAS (shaded region in the submillimeter range; Fixsen et al. 1998), IRAS and ISO (thinline with downward arrows at 60 and 90 µm and cross in 150–180 µm, respectively; Juvela et al. 2009), IRTS (shaded region in the near infrared; Matsumoto et al.2005), and HST (open triangles in the near infrared; Thompson et al. 2007). For comparison, we also show various foreground emission components of the dark sky:zodiacal light (ZL), zodiacal emission (ZE), starlight (SL; K > 9 mag), diffuse Galactic light (DGL), the Galactic cirrus (ISD), and the CMB. The integrated fluxfrom the galaxy counts, obtained by deep surveys from the ground in the near-infrared and submillimeter ranges and via space telescopes in the mid- and far-infrared(ISO, Spitzer, and AKARI) (Totani et al. 2001; Kawara et al. 1998; Puget et al. 1999; Matsuhara et al. 2000; Kawara et al. 2004; Dole et al. 2004; Frayer et al. 2006a;Matsuura et al. 2007; Shirahata et al. 2008), is indicated by the filled triangles connected with thin lines. Stacking the 24-µm galaxies for the Spitzer/MIPS map(open triangles) and the BLAST map (open circles) results in good agreement with the predicted CIB level from a galaxy evolution model by Lagache et al. (2004)(dotted line).

scattered by interstellar dust), Galactic cirrus emission (ther-mal emission from interstellar dust), and the cosmic microwavebackground (CMB). Note that at mid-infrared wavelengths itis currently impossible to detect the CIB because the zodiacalemission is too bright. In the near infrared and far infrared, theforeground emission is relatively weak, and careful modelingand subtraction of the foreground enable one to extract the CIBfrom the measured sky brightness.

As seen in Figure 1, the CIB spectrum at wavelengthslonger than 200 µm has been well constrained with the FIRASinstrument on COBE (Puget et al. 1996; Fixsen et al. 1998).However, results of photometric measurements at wavelengthsshorter than 200 µm with the DIRBE instrument on COBE aredivergent in the mean levels of the CIB brightness, mainlydue to the strong and uncertain foreground contamination ofzodiacal emission, which dominates the brightness of the entiresky, even though the Galactic foreground may be sufficientlyweak in low cirrus regions (Hauser et al. 1998; Lagache et al.2000; Finkbeiner et al. 2000). Although the CIB brightnesswas recently estimated using ISOPHOT data, independent ofCOBE data, the 90-µm data gave only an upper limit, andthe measurement accuracy of the CIB brightness in the 150–180-µm range was in fact worse than that of COBE (Juvelaet al. 2009). Figure 1 clearly shows a wavelength desert ofthe CIB measurement at shorter far-infrared wavelengths, i.e.,<200 µm. Hence, new measurements of the mean level of theCIB are required in this region.

In the last decade, many observational efforts were made toresolve the CIB into individual galaxies via far-infrared deepsurveys with infrared space telescopes such as ISO, Spitzer, andAKARI (Kawara et al. 1998; Puget et al. 1999; Matsuhara et al.2000; Kawara et al. 2004; Dole et al. 2004; Frayer et al. 2006a;Matsuura et al. 2007; Shirahata et al. 2008), and consequently

the origin of the CIB became clear. As shown in Figure 1,however, the detected galaxies account for only a small fraction(∼10%) of the measured CIB brightness in the far infrared.Frayer et al. (2006b) claimed that they resolved more thanhalf the model CIB at 70 µm into point sources in a singledeep survey toward the GOODS-N field. In the mid-infrared, alower limit of the CIB at 24 µm was derived from the integratednumber counts, and this is thought to account for ∼70% of themodel CIB at 24 µm (Papovich et al. 2004). Dole et al. (2006)obtained lower limits for the CIB at 70 and 160 µm via a stackinganalysis of detected sources at 24 µm, finding that the mid-infrared sources contribute ∼80% of the CIB in the far infrared,as shown in Figure 1 by the dotted line. In the submillimeterrange, a similar stacking analysis of 24-µm galaxies against thedeep surveys at 250, 350, and 500 µm by the Balloon-borneLarge-Aperture Submillimeter Telescope (BLAST) experimentrevealed that the 24-µm sources produce almost the entire CIBin the submillimeter range measured with FIRAS (Devlin et al.2009; Marsden et al. 2009). Although these studies, using theSpitzer 24-µm surveys, provided strong constraints on the meanCIB level, the current limit of direct measurement of the CIB asdiffuse emission in the far-infrared range is still high enough toallow the existence of new populations.

Measuring the spatial fluctuations (anisotropy) of the CIBis a powerful method of investigating the unresolved galaxypopulation below the detection limit and yields little con-tamination from the foreground. The CIB fluctuation analysiswas pioneered for the COBE/DIRBE data (Kashlinsky et al.1996a; 1996b). The angular power spectrum of the CIB fluc-tuations is an important observation to trace the distribution ofstar-forming galaxies with respect to the clustering bias in dark-matter halos. The fluctuation measurement is especially effec-tive at longer wavelengths, where direct measurement of the

2

Matsuura+’11

Gamma-ray Attenuation by The Cosmic Optical & Infrared Background

© MPI

Typical Spectra of Blazars• Non-thermal emission

from radio to gamma-ray

• Two peaks

• Synchrotron

• Inverse Compton

• Luminous blazars (Flat Spectrum Radio Quasars: FSRQs) tend to have lower peak energies (Fossati+’98, Kubo+’98)

41

42

43

44

45

46

47

48

49

-5 0 5 10

Log

10 i

Li(e

rg s

-1)

Log10 Ea (eV)Log10 (Energy [eV])

Log 1

0 (νL

ν [er

g/s]

)

FSRQ

BL Lac

Fossati+’98, Kubo+’98, YI & Totani ’09

Constraints from Gamma rays

• Fermi derived the COB opacity using the combined spectra of blazars (see also Gong &

Cooray ’13, Dominguez +’13).

• H.E.S.S. derived the COB/CIB intensity using the combined spectra of blazars.

• Assume 1) no pile-up in the TeV band & 2) extrapolation from unattenuated spectra.

sources above the critical energy (30). This inturn depends on a precise description of thegamma-ray spectra by our source parametriza-tion. To verify that this is the case and to ex-clude the possibility that the detected absorptionfeature is intrinsic to the gamma-ray sources (17),we performed the analysis in three independentredshift intervals (z < 0.2, 0.2 ≤ z < 0.5, and 0.5 ≤z < 1.6). The deviations from the intrinsic spectrain the three redshift intervals are displayed in Fig.2. In the local universe (z < 0.2), EBL absorptionis negligible in most of the Fermi-LAT energy

band (Ecrit ≥ 120 GeV). The lowest redshift in-terval therefore reveals directly the intrinsic spec-tra of the sources and shows that our spectralparametrization is accurate (18). The absorptionfeature is clearly visible above the critical energyin the higher redshift bins. Its amplitude and mod-ulation in energy evolve with redshift as expectedfor EBL absorption. In principle, the observedattenuation could be due to a spectral cutoff thatis intrinsic to the gamma-ray sources. The absenceof a cutoff in the spectra of sources with z < 0.2would require that the properties of BLLacs change

with redshift or luminosity. It remains an issue ofdebatewhether such evolution exists (31–34). How-ever, in case itwere present, the intrinsic cutoffwouldbe expected to evolve differently with redshift thanwe observe. To illustrate this effect, we fitted theblazar sample assuming that all the sources have anexponential cutoff at an energy E0. From sourceto source, the observed cutoff energy changes be-cause of the source redshift and because we as-sumed that blazars as a population are distributedin a sequence such as that proposed in (31–34).E0 was fitted to the data globally like b above. As

1Deutsches Elektronen Synchrotron DESY, D-15738 Zeuthen,Germany. 2W. W. Hansen Experimental Physics Laboratory,Kavli Institute for Particle Astrophysics and Cosmology, De-partment of Physics and SLAC National Accelerator Laboratory,StanfordUniversity, Stanford, CA 94305, USA. 3Space SciencesLaboratory, 7 Gauss Way, University of California, Berkeley, CA94720–7450, USA. 4Max-Planck Institut für ExtraterrestrischePhysik, 85748 Garching, Germany. 5Università di Pisa andIstituto Nazionale di Fisica Nucleare, Sezione di Pisa I-56127Pisa, Italy. 6Laboratoire AIM, CEA-IRFU/CNRS/Université ParisDiderot, Service d’Astrophysique, CEA Saclay, 91191 Gif surYvette, France. 7Istituto Nazionale di Fisica Nucleare, Sezione diTrieste, I-34127 Trieste, Italy. 8Dipartimento di Fisica, Universitàdi Trieste, I-34127 Trieste, Italy. 9Istituto Nazionale di FisicaNucleare, Sezione di Padova, I-35131 Padova, Italy. 10Dipar-timento di Fisica e Astronomia “G. Galilei,” Università diPadova, I-35131 Padova, Italy. 11Istituto Nazionale di FisicaNucleare, Sezione di Pisa, I-56127 Pisa, Italy. 12Santa CruzInstitute for Particle Physics, Department of Physics andDepartment of Astronomy and Astrophysics, University ofCalifornia at Santa Cruz, Santa Cruz, CA 95064, USA.13Dipartimento di Fisica “M. Merlin” dell’Università e delPolitecnico di Bari, I-70126 Bari, Italy. 14Istituto Nazionale diFisica Nucleare, Sezione di Bari, 70126 Bari, Italy. 15LaboratoireLeprince-Ringuet, École Polytechnique, CNRS/IN2P3, Palaiseau,France. 16Institut de Ciències de l’Espai (IEEE-CSIC), CampusUAB, 08193 Barcelona, Spain. 17INAF-Istituto di AstrofisicaSpaziale e Fisica Cosmica, I-20133 Milano, Italy. 18AgenziaSpaziale Italiana (ASI) Science Data Center, I-00044 Frascati(Roma), Italy. 19Istituto Nazionale di Fisica Nucleare, Sezionedi Perugia, I-06123 Perugia, Italy. 20Dipartimento di Fisica,Università degli Studi di Perugia, I-06123 Perugia, Italy. 21Cen-ter for Earth Observing and Space Research, College of Science,George Mason University, Fairfax, VA 22030, USA. 22National Re-

search Council, National Academy of Sciences, Washington,DC 20001. 23INFN and Dipartimento di Fisica e Astronomia“G. Galilei,” Università di Padova, I-35131 Padova, Italy.24ASI Science Data Center, I-00044 Frascati (Roma), Italy.25Laboratoire Univers et Particules de Montpellier, UniversitéMontpellier 2, CNRS/IN2P3, Montpellier, France. 26Departmentof Physics, Stockholm University, AlbaNova, SE-106 91 Stock-holm, Sweden. 27The Oskar Klein Centre for CosmoparticlePhysics, AlbaNova, SE-106 91 Stockholm, Sweden. 28RoyalSwedish Academy of Sciences Research Fellow, SE-106 91Stockholm, Sweden. 29IASF Palermo, 90146 Palermo, Italy.30INAF-Istituto di Astrofisica Spaziale e Fisica Cosmica, I-00133Roma, Italy. 31Space Science Division, Naval Research Labora-tory, Washington, DC 20375–5352, USA. 32Department ofPhysical Sciences, Hiroshima University, Higashi-Hiroshima,Hiroshima 739-8526, Japan. 33NASA Goddard Space FlightCenter, Greenbelt, MD 20771, USA. 34INAF Istituto di Radio-astronomia, 40129Bologna, Italy. 35Department of Astronomy,Graduate School of Science, Kyoto University, Sakyo-ku, Kyoto606-8502, Japan. 36Department of Physics, Royal Institute ofTechnology (KTH), AlbaNova, SE-106 91 Stockholm, Sweden.37Research Institute for Science and Engineering, WasedaUniversity, 3-4-1, Okubo, Shinjuku, Tokyo 169-8555, Japan.38CNRS, IRAP, F-31028 Toulouse cedex 4, France. 39GAHEC,Université de Toulouse, UPS-OMP, IRAP, Toulouse, France.40Department of Astronomy, Stockholm University, SE-106 91Stockholm, Sweden. 41Istituto Nazionale di Fisica Nucleare,Sezione di Torino, I-10125 Torino, Italy. 42Department of Physicsand Department of Astronomy, University of Maryland, CollegePark, MD 20742, USA. 43Hiroshima Astrophysical Science Cen-ter, Hiroshima University, Higashi-Hiroshima, Hiroshima 739-8526, Japan. 44Istituto Nazionale di Fisica Nucleare, Sezione diRoma “Tor Vergata,” I-00133 Roma, Italy 45INTEGRAL ScienceData Centre, CH-1290 Versoix, Switzerland. 46Institute of Space

and Astronautical Science, JAXA, 3-1-1 Yoshinodai, Chuo-ku,Sagamihara, Kanagawa 252-5210, Japan. 47Department ofPhysics andAstronomy,University ofDenver, Denver, CO80208,USA 48Max-Planck-Institut für Physik, D-80805 München,Germany. 49Department of Physics and Center for Space Sci-ences and Technology, University of Maryland BaltimoreCounty, Baltimore, MD 21250, USA. 50Center for Researchand Exploration in Space Science and Technology (CRESST)and NASA Goddard Space Flight Center, Greenbelt, MD 20771,USA. 51Harvard-Smithsonian Center for Astrophysics, Cam-bridge, MA 02138, USA. 52Institut für Astro- und Teilchenphysikand Institut für Theoretische Physik, Leopold-Franzens-UniversitätInnsbruck, A-6020 Innsbruck, Austria. 53Department of Physics,California Polytechnic State University, San Luis Obispo, CA93401, USA. 54Department of Physics, University of Washing-ton, Seattle, WA 98195–1560, USA. 55Max-Planck-Institut fürKernphysik, D-69029 Heidelberg, Germany. 56Space Sci-ences Division, NASA Ames Research Center, Moffett Field,CA 94035–1000, USA. 57NYCB Real-Time Computing Inc.,Lattingtown, NY 11560–1025, USA. 58Astronomical Observatory,JagiellonianUniversity, 30-244Kraków, Poland. 59Department ofChemistry and Physics, Purdue University Calumet, Hammond,IN 46323–2094, USA. 60Institució Catalana de Recerca i EstudisAvançats (ICREA), Barcelona, Spain. 61Consorzio Interuniversitarioper la Fisica Spaziale (CIFS), I-10133 Torino, Italy. 62Dipartimentodi Fisica, Università di Roma “Tor Vergata,” I-00133 Roma, Italy.63Department of Physics, Center for Cosmology and Astro-ParticlePhysics, Ohio State University, Columbus, OH 43210, USA.

*To whom correspondence should be addressed. E-mail:majello@slac.stanford.edu (M.A.); buehler@stanford.edu(R.B.); anita.reimer@uibk.ac.at (A.R.)†Present address: Naval Research Laboratory, Washington, DC20375, USA.

Fig. 1. Measurement, at the 68 and 95% confi-dence levels (including systematic uncertaintiesadded in quadrature), of the opacity tgg from thebest fits to the Fermi data compared with predic-tions of EBL models. The plot shows the measure-ment at z ≈ 1, which is the average redshift of themost constraining redshift interval (i.e., 0.5 ≤ z <1.6). The Fermi-LAT measurement was derived com-bining the limits on the best-fit EBL models. Thedownward arrow represents the 95% upper limit onthe opacity at z = 1.05 derived in (13). For clarity,this figure shows only a selection of the models wetested; the full list is reported in table S1. The EBLmodels of (49), which are not defined for E ≥ 250/(1 + z) GeV and thus could not be used, are reportedhere for completeness.

Energy [GeV]

210

γγτ

-110

1

10

LAT best fit -- 1 sigmaLAT best fit -- 2 sigmaFranceschini et al. 2008Finke et al. 2010 -- model CStecker et al. 2012 -- High OpacityStecker et al. 2012 -- Low OpacityKneiske et al. 2004 -- highUVKneiske et al. 2004 -- best fitKneiske & Dole 2010Dominguez et al. 2011Gilmore et al. 2012 -- fiducialAbdo et al. 2010

1.0≈z

www.sciencemag.org SCIENCE VOL 338 30 NOVEMBER 2012 1191

REPORTS

on

April

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fro

m

H.E.S.S. collaboration: The EBL imprint on H.E.S.S. spectra

m ]µ [ λ1 10

]-1

sr

-2 [

nW

mλ

Fλ

10

210H.E.S.S. low energy

H.E.S.S. full datasetH.E.S.S. high energy

H.E.S.S. contour )

low | full | high(sys + stat

Direct measurements

ME12 exclusion region

Galaxy counts

Fig. 5. Flux density of the extragalactic background light versus wave-length. The 1σ (statistical) contours derived for several energy rangesare described in the top-right legend. The systematic uncertainty isadded quadratically to the statistical one to derive the H.E.S.S. con-tour. Lower limits based on galaxy counts and direct measurements arerespectively shown with empty upward and filled downward pointingtriangles (extracted from Gilmore et al. 2012). The region excluded byMeyer et al. (2012) with VHE spectra is represented by the dashed area.

Table 6. Measured normalization of the EBL optical depth, correspond-ing to the 1σ (statistical) contours shown in Fig. 5.

τmeasured/τFR08 λmin−λmax λFλ(λmin)−λFλ(λmax)µm nW m−2 sr−1

1.27+0.18−0.15 1.2−5.5 14.8+2.1

−1.7−4.0+0.6−0.5

1.34+0.19−0.17 0.30−5.5 3.1 ± 0.4−4.2+0.6

−0.5

1.05+0.32−0.28 1.2−17 12.2+3.7

−3.3−3.2+1.0−0.8

Notes. The second column indicates the wavelength range where thismeasurement is valid and the third column the corresponding flux den-sities. The first line corresponds to the full data set. The second andthird lines indicate the value derived with smaller data sets focussed onspecific energy ranges. The systematic uncertainty on the measurementslisted in the first column is 0.25.

The detailed study of the dependence of the systematic uncer-tainties on the wavelength, based e.g. on deviations from theEBL template model, is beyond the scope of this paper but thecomparison of various modellings in a complementary redshiftband and wavelength range by The Fermi-LAT Collaboration(Ackermann et al. 2012) supports our choice of template.

The contours lie in between the constraints derived withgalaxy counts and the direct measurements. A good agreementwith the VHE upper limit derived by Meyer et al. (2012) is alsofound over the wavelength range covered, with a maximum dis-crepancy between 1 and 2 µm smaller than the 1σ level. Theanalysis performed enables a measurement of the COB peak fluxdensity of λFλ = 15.0+2.1

−1.8 ± 2.8sys nW m−2 sr−1 at 1.4 µm, wherethe peak value and uncertainties are derived by scaling up theFR08 EBL flux density by a factor α0. This value is compatiblewith the previous constraints on the EBL flux density derivedwith H.E.S.S. data by Aharonian et al. (2006c).

5. Summary and conclusion

The spectra of the brightest blazars detected by H.E.S.S. were in-vestigated for an EBL absorption signature. Assuming intrinsicspectral smoothness, the intrinsic spectral curvature was care-fully disentangled from the EBL absorption effect. The EBLimprint is detected at an 8.8σ level, which constitutes the firstmeasurement of the EBL optical depth using VHE γ-rays. TheEBL flux density has been evaluated over almost two decadesof wavelengths with a peak amplitude at 1.4 µm of λFλ =15 ± 2sys ± 3sys nW m−2 sr−1, in between direct measurementsand lower limits derived with galaxy counts.

The low energy threshold achieved with the upgrade of theH.E.S.S. array, H.E.S.S. II, will enable the observation of theunabsorbed population of γ-rays and improve the constraintson the intrinsic spectra and thus on the absorption feature. Thetrough between the COB and the CIB will be characterized bythe Cherenkov Telescope Array (CTA, Actis et al. 2011) whichwill probe energies above 50 TeV. Finally, the increasing sizeof the sample of blazars detected at very high energies will im-prove the constraints on the redshift dependence of the EBL andestablish a firm observational probe of the thermal history of theUniverse.

Acknowledgements. The support of the Namibian authorities and of theUniversity of Namibia in facilitating the construction and operation of H.E.S.S.is gratefully acknowledged, as is the support by the German Ministry forEducation and Research (BMBF), the Max Planck Society, the German ResearchFoundation (DFG), the French Ministry for Research, the CNRS-IN2P3 and theAstroparticle Interdisciplinary Programme of the CNRS, the UK Science andTechnology Facilities Council (STFC), the IPNP of the Charles University, theCzech Science Foundation, the Polish Ministry of Science and Higher Education,the South African Department of Science and Technology and National ResearchFoundation, and by the University of Namibia. We appreciate the excellentwork of the technical support staff in Berlin, Durham, Hamburg, Heidelberg,Palaiseau, Paris, Saclay, and in Namibia in the construction and operation of theequipment.

ReferencesAbdo, A. A., Ackermann, M., Ajello, M., et al. (Fermi-LAT collaboration) 2010,

ApJ, 723, 1082Abramowski, A., Acero, F., Aharonian, F., et al. (H.E.S.S collaboration) 2010a,

A&A, 516, A56Abramowski, A., Acero, F., Aharonian, F., et al. (H.E.S.S collaboration) 2010b,

A&A, 520, A83Abramowski, A., Acero, F., Aharonian, F., et al. (H.E.S.S collaboration) 2011,

A&A, 533, A110Abramowski, A., Acero, F., Aharonian, F., et al. (H.E.S.S collaboration) 2012,

A&A, 539, A149Acero, F., Aharonian, F., Akhperjanian, A. G., et al. (H.E.S.S collaboration)

2010, A&A, 511, A52Ackermann, M., Ajello, M., Allafort, A., et al. (Fermi-LAT collaboratoin) 2012,

Science, 338, 1190Actis, M., Agnetta, G., Aharonian, F., et al. 2011, Exp. Astron., 32, 193Aharonian, F. A. 2000, New Astron., 5, 377Aharonian, F. A., Akhperjanian, A. G., Barrio, J. A., et al. 1999, A&A, 349, 11Aharonian, F., Akhperjanian, A., Beilicke, M., et al. 2003, A&A, 403, 523Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. (H.E.S.S collaboration)

2005a, A&A, 437, 95Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. (H.E.S.S collaboration)

2005b, A&A, 436, L17Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. (H.E.S.S collaboration)

2005c, A&A, 430, 865Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. (H.E.S.S collabo-

ration) 2005d, A&A, 442, 895Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. (H.E.S.S collabo-

ration) 2006a, A&A, 457, 899Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. (H.E.S.S collabo-

ration) 2006b, A&A, 455, 461Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. (H.E.S.S collabo-

ration) 2006c, Nature, 440, 1018

A4, page 7 of 11

Ackermann +’12Abramowski +’13

z~1.0

Two VHE (>100 GeV) Gamma Rays from PKS 0426-380 at z=1.1

• 2 VHE photons at flaring states, but not an exact correspondence to the peak of each flare.

• Spectral hardening from ~30 GeV.

– 21 –

10-11

10-10

10-9

E2 dN/d

E [e

rg c

m-2

s-1]

0.1 1 10 100

Energy [GeV]

Observed EBL corrected (Franceschini et al. 2008) EBL corrected (Inoue et al. 2013) Broken PL model

Fig. 4.— SED of PKS 0426−380 derived from the Fermi -LAT data accumulated during the

most energetic flaring state spanning MET 280000000 (17:46:38 UT on 2009 November 15)

to 302000000 (08:53:18 UT on 2010 July 28; see also the black horizontal line in Figure 2).

The observed spectrum is denoted by black squares and the highest energy bin is a 95%

confidence level upper limit. The spectra corrected for the EBL-related attenuation, using

the EBL models of Franceschini et al. (2008) and Inoue et al. (2013), are represented by

red circles and blue triangles, respectively. A broken power-law model which maximizes the

likelihood for the 0.1–300 GeV Fermi -LAT data is also indicated with black dashed line.

Tanaka, YI, +’13

– 19 –

100

Photo

n E

ner

gy [

GeV

]

50

8

6

4

2

0

0.1

-30

0 G

eV F

lux

[10

-7 p

hoto

ns

cm-2

s-1

]

5614255777554125504754682

MJD

200

Fig. 2.— top panel: Arrival times (MJD) and energies of the seven most energetic E >

50GeV ULTRACLEAN events detected from the vicinity of the direction of PKS 0426−380

(see Figure 1). bottom panel: Fermi -LAT weekly binned light curve of PKS 0426−380

(0.1−300GeV energy range). Black triangles denote the 95% CL flux upper limits (when TS

< 10). The two vertical dashed lines denote the VHE detection times (MJD 55209.11920977

and 56314.39746904; cf., Table 1). Also shown by a black horizontal line is the accumulation

period for which the Fermi -LAT spectrum was constructed (see Figure 4).

Is VHE Spectral Hardening Universal?

• It “seems” TeV blazars show spectral hardening.

The Astrophysical Journal Letters, 751:L11 (4pp), 2012 May 20 Essey & Kusenko

0

0.5

1

1.5

2

2.5

3

0 0.2 0.4 0.6 0.8 1 1.2 1.4

∆Γ

zFigure 1. Spectral change, ∆Γ = ΓTeV − ΓGeV, for TeV detected blazarsobserved by Fermi. Data points from the Fermi Second catalog (The Fermi-LAT Collaboration 2011) were separated into three sets: nearby sources (redinverted triangles), intermediate sources (green triangles), and distant sources(blue diamonds). The lines are the best fits to Equation (10) with D = 17.46(dashed line) and (Γp − Γs ) = 0.995 (solid line).(A color version of this figure is available in the online journal.)

effect would increase ∆Γ because the variation implies some ad-ditional softening due to moving past the Compton peak, whichis not supported by the data. TeV spectra, if they are secondarygamma rays produced along the line of sight, do not depend sig-nificantly on the gamma-ray or proton spectra of their sources(Essey & Kusenko 2010; Essey et al. 2010, 2011b; Murase et al.2012; Razzaque et al. 2012). The dependence on the EBL model(Finke et al. 2010; Franceschini et al. 2008; Stecker et al. 2006;Gilmore et al. 2009; Orr et al. 2011) is very weak (Essey et al.2011b). Thus, the spectral variation does not affect our con-clusion that the behavior in Figure 1 is consistent with a newcomponent taking over and dominating the signal for z ! 0.15.For the same reason, our best-fit line in Figure 1 does not dependon the choice of the EBL model.

Line-of-sight interactions of cosmic rays can account forthe hard spectra of distant blazars because, in this case, theobserved multi-TeV gamma rays are produced in interactionsof cosmic rays with the background photons relatively closeto Earth (Essey & Kusenko 2010; Essey et al. 2010, 2011b;Murase et al. 2012). For this reason, the distance to the sourceis much less important than in the case of primary sources.One, therefore, expects the spectra of secondary gamma rays toexhibit a slower change with redshift.

2. SOFTENING OF A TWO-COMPONENT SPECTRUM

We would like to generalize the Stecker & Scully (2006, 2010)scaling law to include the additional component at high redshift.The fluxes of primary gamma rays produced at the source andof secondary gamma rays produced in line-of-sight interactionsof protons scale with distance d as follows (Essey et al. 2011b):

Fprimary, γ (d) ∝ 1d2

e−d/λγ (2)

Fsecondary, γ (d) ∝ λγ

d2

!1 − e−d/λγ

"(3)

∼#

1/d, for d ≪ λγ ,

1/d2, for d ≫ λγ .(4)

Obviously, for a sufficiently distant source, secondary gammarays must dominate because they do not suffer from exponentialsuppression as in Equation (2). The predicted spectrum of γ -raysturns out to be similar for all the distant AGNs. Essey & Kusenko(2010) and Essey et al. (2010, 2011b) have calculated the spectrafor redshifts of 3C279, 1ES 1101-232, 3C66A, 1ES0229+200,and several other blazars, all of which yield a remarkably good(one-parameter) fit to the data (Essey & Kusenko 2010; Esseyet al. 2010, 2011b).

Based on our numerical results using a Monte Carlo propa-gation code described by Essey & Kusenko (2010) and Esseyet al. (2010, 2011b), we find that the spectra have a weak redshiftdependence and, in the TeV energy range, for 0.2 " z " 0.6, itcan be approximated by the following simple relation:

ΓTeV ≃ Γp + αz, (5)

where Γp is a constant and α ≈ 1.Let us now consider a flux of TeV gamma rays which is the

sum of two components that have the above-mentioned scalingwith distance:

FTeV = F11d2

exp(−d/λγ ) E−(Γs+DH0d)

+ F21d2

(1 − e−d/λγ )E−(Γp+αH0d) (6)

= 1d2

$e−d/λγ

%F1E

−(Γs+DH0d) − F2E−(Γp+αH0d)&

+ F2 E−(Γp+αH0d)' . (7)

While the overall 1/d2 factor does not affect the spectralindex, the exponential suppression of the first term in squaredbrackets in Equation (7) guarantees a sharp change from theStecker & Scully (2006, 2010) scaling law to a flatter scalinglaw which shows only a weak redshift dependence. The changeoccurs when the distance d is of the order of λγ , i.e., at a distancefrom the source where EBL optical depth approaches 1. Basedon our numerical calculations, and in agreement with Stecker& Scully (2006), the corresponding redshift is z ≈ H0d ≈ 0.1.Taking into account that F1 ≫ F2, one can write an approximatescaling law as

z2 FTeV ∝ e−z/0.1 F1 E−(Γs+Dz) + F2E−(Γp+αz). (8)

At lower energies, in the GeV energy range, the flux isexpected to show very little attenuation for z " 0.5 and to followthe simple relation

z2 FGeV ∝ F1 E−Γs . (9)

Thus, we expect that ∆Γ = ΓTeV − ΓGeV should exhibit thefollowing behavior:

∆Γ ≃#Dz for z " 0.1,

(Γp − Γs) + αz, for z ! 0.1.(10)

For practical reasons, it is easier and more instructive tocompare the spectral slopes given by Equation (10) with thedata rather than to fit the fluxes in Equation (9).

To select distant sources that are likely to be powerfulsources of cosmic rays (see Table 1), we applied two selectioncriteria: we selected gamma-ray emitters which (1) have been

2

The Astrophysical Journal, 768:197 (17pp), 2013 May 10 Inoue et al.

10-13

10-12

10-11

10-10

10-9

z = 0.165H 2356-309

E2dN

/dE

[T

eV/c

m2/s

]

z = 0.179RX J0648.7+1516

z = 0.1821ES 1218+304

z = 0.1861ES 1101-232

10-13

10-12

10-11

10-10

10-9

z = 0.1881ES 0347-121

z = 0.19RBS 0413

z = 0.2121ES 1011+496

z = 0.2871ES 0414+009

10-13

10-12

10-11

10-10

10-9

0.1 1 10

Energy [TeV]

z = 0.31S5 0716+714

0.1 1 10

Energy [TeV]

z = 0.4324C+21.35

0.1 1 10

Energy [TeV]

z = 0.4443C 66A

0.1 1 10

Energy [TeV]

z = 0.53623C 279

Figure 14. Same as Figure 13, but for sources at z > 0.15. References for the data are H 2356-309 (Aharonian et al. 2006b), RX J0648.7+1516 (Aliu et al. 2011),1ES 1218+304 (Acciari et al. 2009b), 1ES 1101−232 (Aharonian et al. 2007c), 1ES 0347−121 (Aharonian et al. 2007a), RBS 0413 (Aliu et al. 2012), 1ES 1011+496(Albert et al. 2007b), 1ES 0414+009 (Abramowski et al. 2012a), S5 0716+714 (Anderhub et al. 2009), 4C+21.35 (Aleksic et al. 2011a), 3C 66A (Aleksic et al. 2011b),and 3C 279 (Albert et al. 2008).(A color version of this figure is available in the online journal.)

harder spectra above several hundred GeV (see also Finke et al.2010).

To explain such intrinsically hard spectra, some authors haverecently suggested secondary cascade components generatedby very high energy cosmic-rays or gamma-rays, which mayalso offer a probe of intergalactic magnetic fields (e.g., Essey& Kusenko 2010; Essey et al. 2011; Essey & Kusenko 2012;Murase et al. 2012; Aharonian et al. 2013). Others haveproposed effects of time-dependence, stochastic acceleration,or multiple emission components (Lefa et al. 2011a, 2011b).Future CTA observations of these objects with high energy andtime resolution will elucidate such issues.

The signature of EBL absorption has not been seen in thespectrum of the extragalactic gamma-ray background (EGB)above 100 GeV (Ackermann 2011), even though it is naturallyexpected if its origin is cosmological (Inoue 2011a; Inoue &Ioka 2012). By considering the effects of cascade emission,Inoue & Ioka (2012) have recently shown that if the EGB at<100 GeV (Abdo et al. 2010b) is entirely composed of knowntypes of sources whose spectra are well constrained by existingobservations, then the measured EGB at >100 GeV would beinconsistent with this hypothesis, even for a low EBL such asproposed here. Further detailed spectral studies of extragalacticgamma-ray sources are required to resolve this issue.

6. CONCLUSIONS

We have developed models for the EBL over the redshiftrange z = 10 to z = 0 on the basis of a semi-analytical

model of hierarchical galaxy formation, into which Pop-III starswere incorporated in a simplified fashion. Our baseline model isconsistent with a wide variety of observational data for galaxiesbelow z ∼ 6 (Nagashima & Yoshii 2004; Kobayashi et al. 2007,2010), and is also capable of reionizing the universe by z < 8.However, in order to account for the Thomson scattering opticaldepth measured by WMAP, the ionizing photon emissivity isrequired to be 50–100 times higher at z > 10. This is in linewith recent observations of galaxy candidates at z ∼ 8, as longas the contribution from faint galaxies below the sensitivity ofcurrent telescopes is not large (e.g., Bouwens et al. 2012). The“missing” ionizing photons may possibly be supplied by Pop-IIIstars forming predominantly at these epochs in sufficiently smallgalaxies.

The EBL intensity at z = 0 in our model is generally not farabove the lower limits derived from galaxy counts. Our model isalso in good agreement with the data from Pioneer (Matsuokaet al. 2011) directly measured from outside the zodiacal region.The Pop-III contribution to the NIR EBL is !0.03 nW m−2 sr−1,less than 0.5% of the total in this band, even at the maximumlevel compatible with WMAP measurements. The putative NIREBL excess (Matsumoto et al. 2005), which also conflicts withthe upper limits from gamma-ray observations (Aharonian et al.2006a), may have a zodiacal origin rather than Pop-III stars.

Up to z ∼ 3–5, the γ γ opacity in our model is comparableto that in the majority of previously published models (Kneiskeet al. 2004; Franceschini et al. 2008; Finke et al. 2010; Gilmoreet al. 2012b) below Eγ ∼ 400/(1 + z) GeV, while it is a factorof ∼2 lower above this energy. The universe is predicted to belargely transparent below 20 GeV even at z > 4.

14

YI+‘13aEssey & Kusenko ’12

z

What is the origin of the hardening?

• Secondary gamma rays from cosmic rays along line of sight (Essey & Kusenko ’10, Essey

+’10, Essey+’11, Murase+’12,Takami+’13).

• Observed GeV-TeV photon index dependence on redshift will be different from simple CIB attenuation (Essey & Kusenko ‘12).

• Stochastic acceleration (Stawarz & Petrosian ’08, Lefa+’11).

• Lepto-hadronic emission (Cerutti+’14).

The Astrophysical Journal, 740:64 (9pp), 2011 October 20 Lefa, Rieger, & Aharonian

-12.5

-12

-11.5

-11

-10.5

-10

-9.5

14 16 18 20 22 24 26 28

log

vFv

[erg

.sec

.cm

2 ]

log v [Hz]

1ES 0229+200 (z=0.1396)

F v ∝

v1/

3

F v ∝

v1/

3

H.E.S.S. low level EBL

H.E.S.S. high level EBL

SWIFT

-12.5

-12

-11.5

-11

-10.5

-10

-9.5

14 16 18 20 22 24 26 28

v [e

rg.s

ec.c

m2 ]

1ES 0229+200 (z=0.1396)

F v ∝

v1/

3

F v ∝

v1/

3

E2Exp[-(E/Ec)3]

E2Exp[-E/Ec]

Figure 10. Hard spectrum blazar 1ES 0229+200 at z = 0.139 with SED modeled within an SSC approach using Maxwellian-type electron distributions. All parametersused are the same as in Figure 3. Data points shown in the figure are from Zacharopoulou et al. (2011), where the intrinsic (de-absorbed) source spectrum has beenderived based on the EBL model of Franceschini et al. (2008) with (1) EBL level as in their original paper (“low-level EBL”) and (2) (maximum) EBL level scaled upby a factor of 1.6 (“high-level EBL”).(A color version of this figure is available in the online journal.)

the high low-energy cutoffs needed in leptonic synchrotron-Compton models for the hard spectrum sources.

Although our main purpose here is not to fit data, Figure 10shows that a Maxwellian-type electron distribution could alsoprovide a satisfactory explanation for the hard TeV componentin 1ES 0229+200.

Our results illustrate that even within a leptonic synchrotron-Compton approach relatively hard intrinsic TeV source spectramay be encountered under a variety of conditions. While thismay be reassuring, the possibility of having such hard sourcespectra within “standard models” unfortunately constrains thepotential of extracting limits on the EBL density based on γ -rayobservations of blazars, one of the hot topics currently discussedin the context of next generation VHE instruments.

We thank S. Kelner and S. Wagner for helpful discussions.

REFERENCES

Aharonian, F., Akhperjanian, A. G., Barres de Almeida, U., et al. 2007a, A&A,475, L9

Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2007b, A&A,470, 475

Aharonian, F. A., Atoyan, A. M., & Nahapetian, A. 1986, A&A, 162, L1Aharonian, F. A., Khangulyan, D., & Costamante, L. 2008, MNRAS, 387, 1206Aharonian, F. A., Timokhin, A. N., & Plyasheshnikov, A. V. 2002, A&A, 384,

834Aharonian, F., et al. 2006, Nature, 440, 1081Atoyan, A. M., & Aharonian, F. A. 1999, MNRAS, 302, 253Begelman, M. C., Blandford, R. D., & Rees, M. J. 1984, Rev. Mod. Phys., 56,

255Bloom, S. D., & Marscher, A. P. 1996, ApJ, 461, 657Blumenthal, G. R., & Gould, R. J. 1970, Rev. Mod. Phys., 42, 237Bottcher, M., Dermer, C. D., & Finke, J. D. 2008, ApJ, 679, L9De Angelis, A., Mansutti, O., Persic, M., & Roncadelli, M. 2009, MNRAS, 394,

L21Derishev, E. V. 2007, Ap&SS, 309, 157

Derishev, E. V., Aharonian, F. A., Kocharovsky, V. V., & Kocharovsky, Vl. V.2003, Phys. Rev. D, 68, 043003

Dermer, C. D., & Schlickeiser, R. 1993, ApJ, 416, 458Essey, W., Kalashev, O., Kusenko, A., & Beacom, J. F. 2011, ApJ, 731, 51EFranceschini, A., Rodigliero, G., & Vaccari, M. 2008, A&A, 487, 837Fritz, K. D. 1989, A&A, 214, 14Georganopoulos, M., Kirk, J. G., & Mastichiadis, A. 2001, ApJ, 561, 111Giebels, B., Dubus, G., & Khelifi, B. 2007, A&A, 462, 29Gould, R. J., & Schreder, G. P. 1967, Phys. Rev., 155, 1408Henri, G., & Pelletier, G. 1991, ApJ, 383, L7Kardashev, N. S. 1962, SvA, 6, 317Katarzynski, K., Ghisellini, G., Tavecchio, F., Gracia, J., & Maraschi, L. 2006,

MNRAS, 368, L52Kifune, T. 1999, ApJ, 518, L21Krawczynski, H., Coppi, C. S., & Aharonian, F. 2002, MNRAS, 336, 721Longair, M. S. 1981, High Energy Astrophysics (Cambridge: Cambridge

University Press)Maraschi, L., Ghisellini, G., & Celotti, A. 1992, ApJ, 397, L5Medvedev, M. V. 2006, ApJ, 637, 869Neronov, A., Semikoz, D., & Taylor, A. M. 2011, A&A, submitted

(arXiv:1104.2801)Reville, B., & Kirk, J. G. 2010, ApJ, 724, 1283Rieger, F. M., Bosch-Ramon, V., & Duffy, P. 2007, Ap&SS, 309, 119Rybicki, G. B., & Lightman, A. P. 1979, Radiative Processes in Astrophysics

(New York: Wiley)Sauge, L., & Henri, G. 2006, A&A, 454, L1Schlickeiser, R. 1985, A&A, 143, 431Sikora, M., Begelman, M. C., & Rees, M. J. 1994, ApJ, 421, 153Sikora, M., Blazejowski, M., Moderski, R., & Madejski, G. M. 2002, ApJ, 577,

78Stawarz, L., & Petrosian, V. 2008, ApJ, 681, 1725Stecker, F. W., Baring, M. G., & Summerlin, E. J. 2007, ApJ, 667, L29Stecker, F. W., & Glashow, S. L. 2001, Astropart. Phys., 16, 97Stecker, F. W., & Scully, S. T. 2008, A&A, 478, L1Tavecchio, F., Ghisellini, G., Ghirlanda, G., Costamante, L., & Franceschini, A.

2009, MNRAS, 399, L59Tavecchio, F., Maraschi, L., & Ghisellini, G. 1998, ApJ, 509, 608Urry, C. M., & Padovani, P. 1995, PASP, 107, 803Zacharopoulou, O., Khangulyan, D., Aharonian, F., & Costamante, L.

2011, ApJ, 738, 157Zirakashvili, V. N., & Aharonian, F. 2007, A&A, 465, 695

9

1ES 0229+200 (z=0.1396) Stochastic Acc.

Lefa+’11

seen from sources beyond z ¼ 1 [8]. We will, therefore, considertwo different redshifts consistent with the lower bound.

The dependence of the secondary spectrum on the redshift ofthe source [4] can be easily understood. The electromagneticshowers induced by the interactions continue to soften the spec-trum until the gamma rays produced have an energy such thatthe optical depth s K 1. The optical depth grows with distance,and it also grows with energy. Thus the low-energy peak of the sec-ondary spectrum occurs at a lower energy for more distant sources,leading to a softer spectral shape. For two different redshifts, thespectral shapes at high energies are similar, as long as the condi-tion s" 1 holds true for both redshifts.

We will show that some combinations of redshift and EBL give agood fit to the VHE data. In particular, we compare with the data ourtheoretical predictions for secondary gamma rays assuming that (i)the source is at z ¼ 0:6 and EBL spectrum based on Ref. [27]; (ii) thesource is at z ¼ 1:0 and z ¼ 1:3 for a lower level of EBL based on Ref.[25]. Each of these possibilities produces a substantially better fit tothe data than a primary gamma ray spectrum obtained in the ab-sence of cosmic rays. This is remarkable because the shape of thehigh-energy spectrum is quite robust and independent of any mod-el parameters. This successful fit to the data provides additionalcredence to the idea that distant gamma-ray sources are dominatedby a hard, secondary spectrum. However, the spectral shape and fitto experimental data alone cannot distinguish between the differ-ent models of EBL and different redshifts. One can, however, useflaring and variability data to infer further information on the tran-sition from primary to secondary spectrum and to gain informationabout both the EBL and the redshift of the source.

While the shape of the spectrum is robust, its normalization de-pends on the source luminosity in protons Lp, which is not known.In what follows, we obtain the best fit for isotropic equivalentsource power in protons Lp;iso ¼ ð1$ 10Þ & 1049 erg/s. The trueintrinsic luminosity can be smaller by several orders of magnitude,depending on the jet opening angle h. In particular, for h ¼ 3', therequired intrinsic luminosity is comparable to the Eddington lumi-nosity of a black hole with a mass M ( ð1$ 8Þ & 108 M). Of course,only a fraction of the jet energy is transferred to high-energy par-ticles. For a non-spherical accretion with a jet, the Eddington lumi-nosity is not necessarily a limiting factor, and there is growingevidence of super-Eddington luminosities in relativistic outflowsin GRBs and in very powerful blazars [33]. The required luminosityis consistent with general principles of acceleration, as long asmost of the accretion energy is converted efficiently to the kineticenergy of the jet, rather than to thermal radiation of the accretionflow [8].

Some assumptions must be made about the strengths of inter-galactic magnetic fields (IGMFs) along the line of sight. If IGMFsare greater than ( 3& 10$14 G, they can cause sufficient deflectionsof protons to diminish a contribution of secondary gamma rays. Forvery small magnetic fields, below 10$17 G, the low-energy part ofthe spectrum (which is not as robust as the TeV part) may exceedthe observations of Fermi [12]. Therefore, we choose to fit the datafor IGMFs of the order of 10$15 G, in the middle of the allowedrange inferred from observations [12].

Blazars are known to be variable with the variability correlatedacross a wide range of energies [29]. Secondary gamma rays, on theother hand, would have this variability washed out [7], and no var-iability should be seen in the secondary component on the time-scale of the VERITAS observations. Thus it is reasonable toconsider the possibility that the observation at lower energieswas due to a flaring state, which should typically have a timescaleof the order of days to months [29]. The flaring is further supportedby Swift data for the VERITAS observation period [30,32]. One canexpect further observations to detect a lower flux state (unless lowenergy data again points to a flaring state), which should be clearly

evident at energies where s* 1. However, the secondary compo-nent should not correlate with this lower energy state and wouldremain roughly constant.

3. Results

In Figs. 1–3 we illustrate the effect of flaring and show the ex-pected spectra for both a high state and a low state for two differ-ent redshift-EBL combinations. The data are available fromVERITAS [30,32] and MAGIC [31]. These data are not contempora-neous, and there can be differences in systematic and statistical er-rors between the two sets of data. In Figs. 1–3 we show the datafrom VERITAS, which were accumulated over a longer period ofobservation and have smaller error bars. Our best-fit curves thatagree with VERITAS data fit the MAGIC data as well.

In each case the flux of the primary component is decreased bya factor 4 (which is reasonable for illustrative purposes and is con-sistent with observations [30,32,31]), while the secondary compo-nent remains constant. Although the flaring spectral data revealedlittle difference between the two scenarios, a comparison of thehigh and low states shows some marked differences. Firstly, thespectral shape of the higher redshift source’s low state is signifi-cantly softer than the lower redshift source. Secondly, the ratio

0.1 1 10 100 1000 1041 10 10

5 10 10

1 10 9

5 10 9

1 10 8

5 10 8

1 10 7

E GeV

E2d

Nd

EG

eVcm

2s

1

Fig. 1. Predicted spectrum of PKS 1424 + 240 (solid black line), assuming z ¼ 0:6,the lowest redshift consistent with the lower bound [23], for EBL model of Ref. [24].The blue dot-dashed line is primary spectrum (spectral index c ¼ 1:65); red dashedline is secondary gamma rays for the mean IGMF of B ¼ 10$15 G with a correlationlength of 1 Mpc; black solid line is the combined spectrum of primary andsecondary gamma rays. Also shown are the VERITAS and Fermi data points. Thepurple line represents the total spectrum with primary signal suppressed by afactor of 4 (low state). (For interpretation of the references to colour in this figurecaption, the reader is referred to the web version of this article.)

0.1 1 10 100 1000 1041 10 10

5 10 10

1 10 9

5 10 9

1 10 8

5 10 8

1 10 7

E GeV

E2d

Nd

EG

eVcm

2s

1

Fig. 2. Same as Fig. 1, but assuming redshift z ¼ 1:0 and using EBL model of Ref.[25].

W. Essey, A. Kusenko / Astroparticle Physics 57–58 (2014) 30–32 31

PKS 1424+240 (z=0.6) Secondary Gamma Rays

Essey & Kusenko ‘14

TeV blazar sample• Select 36 blazars with z

from the default TeVcat catalog.

• Low-state data are available for 31/36.

• 3FGL SED data.

• CIB correction by YI+’13.

• Systematic jet parameter study w/ MWL data is also on-going.

10-1310-1210-1110-1010-910-810-7

z = 0.1301ES 1215+303

E2dN

/dE

[T

eV

/cm

2/s

]

z = 0.1381ES 0806-524

z = 0.141ES 0229+200

10-1310-1210-1110-1010-910-810-7

z = 0.1426391RXS J101015.9 - 311909

z = 0.165H 2356-309

z = 0.179RX J0648.7+1516

10-1310-1210-1110-1010-910-810-7

10-4 10-3 10-2 10-1 100 101

z = 0.1821ES 1218+304

Energy [TeV]10-4 10-3 10-2 10-1 100 101

z = 0.1861ES 1101-232

Energy [TeV]10-4 10-3 10-2 10-1 100 101

z = 0.1881ES 0347-121

Energy [TeV]YI & Tanaka in prep.

0

1

2

3

4

5

0 0.1 0.2 0.3 0.4 0.5

6K

TeV-

GeV

Redshift z

VHE BlazarsEssey and Kusenko ’12 (z>0.1)

Inoue et al. ’13

YI+ in prep.

GeV-TeV index dependence on redshift

• No clear correlation (see also Sanchez+’13).

• But, 1ES 0229+200 seems to be peculiar.

• additional components at TeV band are not significantly seen via F-test. No sources with P(F)<0.05.

• Future test by CTA is necessary (e.g. Takami+’13, YI+’14b).

1ES 0229+200

The Astrophysical Journal Letters, 771:L32 (5pp), 2013 July 10 Takami, Murase, & Dermer

10-15

10-14

10-13

10-12

10-11

10-10

1010 1011 1012 1013 1014

E2 F

E [e

rg c

m-2

s-1

]

E [eV]

H.E.S.S. I

CTA

γ-induced (low IR)γ-induced (best fit)

CR-induced (low IR)CR-induced (best fit)

Becherini et al. (2012)

Figure 1. SEDs calculated for gamma-ray-induced (red) and UHECR-induced(blue) cascade scenarios for KUV 00311−1938 (z = 0.61) using low IR (thick)and best fit (thin) EBL models deduced by Kneiske et al. (2004) with the analyzedLAT data (green) with a H.E.S.S. preliminary spectrum (magenta; Becheriniet al. 2012). We take s = 1.76. The isotropic equivalent energy of input gammarays for the gamma-ray-induced cascade Liso

γ and of UHECR source protons fora UHECR-induced cascade Liso

p are 3.5×1046 erg s−1 and 1.1×1047 erg s−1, re-spectively. The differential sensitivity curve for a 50 hr observation with H.E.S.S.I (http://www.mpi-hd.mpg.de/hfm/HESS/pages/home/proposals/; dashed line),and the 50 hr sensitivity goal of the Cherenkov Telescope Array (CTA; Actiset al. 2011; dotted line) are also plotted. The flux lower than the sensitivitycurve can be achieved under a relaxed criterion of wider energy-bins and lowersignificance required to estimate flux in each bin.(A color version of this figure is available in the online journal.)

reproduced by both gamma-ray- and UHECR-induced cascadescenarios between 10 and 100 GeV. The UHECR-induced cas-cade predicts larger flux above 200 GeV and harder spectrumthan the gamma-ray-induced scenario above ∼1 TeV. Prelimi-nary H.E.S.S. data support the hadronic interpretation. Note thatthe redshift of this object is uncertain (see Section 5).

We confirmed that the SEDs of the other more distant sourcesin the list, excepting sources with steep spectra, namely PKS0426−380 and PKS 2142−75, are reproduced by both gamma-ray-induced and UHECR-induced cascade scenarios for thequoted redshifts. More distant sources allow the possibilityto distinguish the two scenarios clearly by the difference inpredicted spectral fluxes above ∼1 TeV. Due to their largedistances, a sharper cutoff of the gamma-ray-induced spectracompared to the UHECR-induced spectra is predicted at thecharacteristic EBL absorption energy Ec (Murase et al. 2012b),and a plateau of emission extending to >10 TeV is predicted inthe hadronic scenario.

In general, differential sensitivity is defined more conserva-tively than integral sensitivity for IACTs. Conventionally, thedifferential sensitivity requires a 5σ signal for a 50 hr obser-vation in each of four equal-width logarithmic bins per decade,whereas the integral sensitivity is defined as a 5σ excess ofgamma rays above a given threshold energy for a 50 hr obser-vation (e.g., Aleksic et al. 2012). Thus, integral flux is moresensitive to the scenario distinction.

Figure 2 shows the integral flux corresponding to the pre-dictions in Figure 1. Here, we can obviously recognize thatthe UHECR-induced scenario can be distinguished from thegamma-ray-induced scenario by the Cherenkov Telescope Ar-ray (CTA). This source is detectable at the 5σ level up to ∼3 TeVfor the low-IR model and ∼1 TeV for the best-fit model in theUHECR-induced scenario, while it should only be detected upto ∼500 GeV in the gamma-ray-induced scenario. Detection of

10-15

10-14

10-13

10-12

10-11

10-10

1010 1011 1012 1013 1014

F(>

E)

[cm

-2 s

-1]

E [eV]

H.E.S.S. I

CTA

γ-induced (low IR)γ-induced (best fit)

CR-induced (low IR)CR-induced (best fit)

0

1

2

54500 55000 55500 56000 56500F(>

10 G

eV)

[10-9

cm

-2 s

-1]

Modified Julian Day

Figure 2. Integral flux corresponding to the SEDs in Figure 1 (KUV00311−1938) with the H.E.S.S. I integral sensitivity (presented by Y. Becheriniin Rencontres de Moriond 2009; http://moriond.in2p3.fr/J09/) and the integralsensitivity goal of CTA for a 50 hr observation (Actis et al. 2011). The insetshows a >10 GeV light curve with 16 equal time bins, each lasting 90.3 days.The light curve is consistent with a constant flux hypothesis with χ2

r = 0.95which is calculated only from finite flux points.(A color version of this figure is available in the online journal.)

10-15

10-14

10-13

10-12

10-11

10-10

1010 1011 1012 1013 1014

F(>

E)

[cm

-2 s

-1]

E [eV]

MAGIC II

CTA

γ-induced (low IR)γ-induced (best fit)

CR-induced (low IR)CR-induced (best fit)

0

1

54500 55000 55500 56000 56500F(>

10 G

eV)

[10-9

cm

-2 s

-1]

Modified Julian Day

Figure 3. Same as Figure 2, but for PG 1246+586 (z = 0.847). Lisoγ =

7.5×1046 erg s−1 and Lisop = 2.0×1047 erg s−1. We take s = 1.94. The inset is

a light curve similar to Figure 2, with χ2r = 0.40 for a constant flux hypothesis.

(A color version of this figure is available in the online journal.)

this source above 1 TeV would be very strong evidence for ahadronic origin of the radiation.

We demonstrate this behavior for a more distant source, PG1246+586, in Figure 3. Despite its distance, this source canbe detected by CTA below ∼200 GeV for both scenarios. Itis possible to distinguish between the two scenarios becausethe difference in detecting photons for the two scenarios wouldbe larger than the range of uncertainties implied by the EBLmodels used, even with the flux of the characteristic hadronicplateau at high energies being below the CTA sensitivity. Thus,even gamma-ray sources with z ∼ 0.85 can be utilized todisentangle the two scenarios. Other sources detectable with50 hr observations with CTA in the source list are Ton 116,B3 1307+433, 4C +55.17, and PKS 1958−179. Note thatthe sensitivity of CTA North may be somewhat worse above∼10 TeV because no small-size telescopes are projected to be apart of the array.

3

Takami+’13

KUV 00311-1938 (z=0.61) Secondary Gamma Rays

Cosmic Gamma-ray Background

Cosmic Gamma-ray Background Spectrum at >0.1 GeV

• Fermi has resolved 30% of the CGB at ~1 GeV and more at higher energies.

• Updated LAT measurement of IGRB spectrum – Extended energy range: 200 MeV – 100 GeV x 100 MeV – 820 GeV

• Significant high-energy cutoff feature in IGRB spectrum – Consistent with simple source populations attenuated by EBL

• Roughly half of total EGB intensity above 100 GeV now resolved into individual LAT sources

34

CGB Spectrum

Ackerman+’15

Possible Origins of CGB at GeV

Markus Ackermann | 220th AAS meeting, Anchorage | 06/11/2012 | Page

The origin of the EGB in the LAT energy range.

4

Unresolved sources Diffuse processesBlazars

Dominant class of LAT extra-galactic sources. Many estima-tes in literature. EGB contribu-tion ranging from 20% - 100%

Non-blazar active galaxies27 sources resolved in 2FGL ~ 25% contribution of radio galaxies to EGB expected. (Inoue 2011)

Star-forming galaxiesSeveral galaxies outside the local group resolved by LAT. Significant contribution to EGB expected. (e.g. Pavlidou & Fields, 2002)

GRBsHigh-latitude pulsars

small contributions expected. (e.g. Dermer 2007, Siegal-Gaskins et al.

2010)

Intergalactic shockswidely varying predictions of EGB contribution ranging from 1% to 100% (e.g. Loeb & Waxman 2000, Gabici & Blasi 2003)

Dark matter annihilationPotential signal dependent on nature of DM, cross-section and structure of DM distribution (e.g. Ullio et al. 2002)

Interactions of UHE cosmic rays with the EBL

dependent on evolution of CR sources, predictions varying from 1% to 100 % (e.g. Kalashev et al. 2009)

Extremely large galactic electron halo (Keshet et al. 2004)

CR interaction in small solar system bodys (Moskalenko & Porter 2009) © M. Ackermann

Blazar contribution to CGB

• Padovani+’93; Stecker+’93; Salamon & Stecker ‘94; Chiang + ‘95; Stecker & Salamon ‘96; Chiang & Mukherjee ‘98; Mukherjee & Chiang ‘99; Muecke & Pohl ‘00; Narumoto & Totani ‘06; Giommi +’06; Dermer ‘07; Pavlidou & Venters ‘08; Kneiske & Mannheim ‘08; Bhattacharya +’09; YI & Totani ‘09; Abdo+’10; Stecker & Venters ‘10; Cavadini+’11, Abazajian+’11, Zeng+’12, Ajello+’12, Broderick+’12, Singal+’12, Harding & Abazajian ’12, Di Mauro+’14, Ajello+’14,Singal+’14, Ajello, YI, +’15,

• Blazars explain ~50% of CGB at 0.1-100 GeV.

Results

•  EGB total intensity of 1.1×10-5 ph cm-2 s-1 sr-1 •  Blazars contribute a grand-total of (5-7)×10-6 ph cm-2 s-1 sr-1

–  Resolved sources : ~4×10-6 ph cm-2 s-1 sr-1 –  Unresolved blazars: ~(2-3)×10-6 ph cm-2 s-1 sr-1 (in agreement with Abdo+10)

Preliminary

Ajello, YI+’15

The Astrophysical Journal, 751:108 (20pp), 2012 June 1 Ajello et al.

]-1 erg s48[10γ L

-310 -210 -110 1 10

]-1 )

48/1

0γ

(L-3

,z=0

) [M

pcγ

(LΦ

-1810

-1710

-1610

-1510

-1410

-1310

-1210

-1110

-1010

-910

-810

-710

LF of Fermi’s FSRQs (this work)

uncertainty on Fermi LFσ1

De-evolved Fermi LF

Narumoto & Totani (2006)

Inoue, Totani, & Mori (2010)

Stecker & Venters (2011)

Figure 5. Local (z = 0) LF of the Fermi FSRQs as derived from the best-fit LDDE model in Section 4.2 (solid line). The gray band represents the ±1 σ uncertaintycomputed as described in the text. The long- and short-dashed lines show the LF models based on the EGRET blazars derived by Narumoto & Totani (2006) and Inoueet al. (2010), respectively. The dash-dotted line shows the prediction from the model of Stecker & Venters (2011).(A color version of this figure is available in the online journal.)

]-1 erg s48[10γ L

-110 1 10

]-1 )

48/1

0γ

(L-3

,z=1

) [M

pcγ

(LΦ

-1410

-1310

-1210

-1110

-1010

-910

-810

LF of Fermi’s FSRQs (this work)

uncertainty on Fermi LFσ1

De-evolved Fermi LF

Narumoto & Totani (2006)

Inoue, Totani, & Mori (2010)

Stecker & Venters (2011)

Figure 6. LF of the Fermi FSRQs at redshift 1.0 as derived from the best-fit LDDE model in Section 4.2 (solid line). The gray band represents the ±1 σ uncertaintycomputed with the method described in the text. The long-dashed and short-dashed lines show the LF models based on the EGRET blazars derived, respectively, byNarumoto & Totani (2006) and Inoue et al. (2010). The dash-dotted line shows the prediction from the model of Stecker & Venters (2011).(A color version of this figure is available in the online journal.)

9

Gamma-ray Luminosity

Function

Ajello+’12

CGB Spectrum

Radio Galaxies

• Strong+’76; Padovani+’93; YI ’11; Di Mauro+’13; Zhou & Wang ’13

• Use gamma-ray and radio luminosity correlation.

• ~20% of CGB at 0.1-100 GeV.

The Astrophysical Journal, 733:66 (9pp), 2011 May 20 Inoue

10-5

10-4

10-3

10-2

10-1

10-2 10-1 100 101 102 103 104 105 106

E2 dN

/dE

[MeV

2 cm

-2 s

-1 M

eV-1

sr-1

]

Photon Energy [MeV]

IntrinsicAbsorbed

CascadeTotal

HEAO-ISwift-BAT

SMMCOMPTELFermi-LAT

Figure 3. EGRB spectrum from gamma-ray-loud radio galaxies in the unit of MeV2 cm−2 s−1 MeV−1 sr−1. Dashed, dotted, dot-dashed, and solid curves show theintrinsic spectrum (no absorption), and the absorbed, cascade, and total (absorbed+cascade) EGRB spectrum, respectively. The observed data of HEAO-1 (Gruberet al. 1999), Swift-BAT (Ajello et al. 2008), SMM (Watanabe et al. 1997), COMPTEL (Kappadath et al. 1996), and Fermi-LAT (Abdo et al. 2010e) are also shown bythe symbols indicated in the figure.(A color version of this figure is available in the online journal.)

jets (Urry & Padovani 1995). The fraction of radio galaxieswith viewing angle <θ is given as κ = (1 − cos θ ). In thisstudy, the fraction of gamma-ray-loud radio galaxies is derivedas κ = 0.081, as discussed in Section 3.3. Then, the expectedθ is !24◦. The viewing angle of NGC 1275, M 87, and CenA is derived as 25◦, 10◦, and 30◦ by SED fitting (Abdo et al.2009b, 2009c, 2010c), respectively. Therefore, our estimationis consistent with the observed results.

Here, beaming factor δ is defined as Γ−1(1−β cos θ )−1, whereΓ is the bulk Lorentz factor of the jet and β =

!1 − 1/Γ2.

If Γ ∼ 10, which is typical for blazars, δ becomes ∼1 withθ = 24◦. This value means no significant beaming effectbecause the observed luminosity is δ4 times brighter than that inthe jet rest frame. On the other hand, if 2 ! Γ ! 4, δ becomesgreater than 2 with θ = 24◦ (i.e., the beaming effect becomesimportant). Ghisellini et al. (2005) proposed the spine and layerjet emission model, in which the jet is composed of a slow jetlayer and a fast jet spine. The difference of Γ between blazarsand gamma-ray-loud radio galaxies would be interpreted usinga structured jet emission model.

We note that κ depends on αr , as in Section 3.2. By changingαr by 0.1 (i.e., to 0.7 or 0.9), κ and θ change by a factor of 1.4 and1.2, respectively. Thus, even if we change αr , the beaming effectis not effective if Γ ∼ 10 but with a lower Γ value, 2 ! Γ ! 4.

5.2. Uncertainty in the Spectral Modeling

As pointed out in Section 2, there are uncertainties in SEDmodeling because of small samples, such as the photon index (Γ)and the break photon energy (ϵbr). In the case of blazars, Stecker& Salamon (1996) and Pavlidou & Venters (2008) calculatedthe blazar EGRB spectrum including the distribution of thephoton index by assuming Gaussian distributions even with∼50 samples. We performed the Kolomogorov–Smirnov testto determine the goodness of fit of the Gaussian distributionto our sample, and to check whether the method of Stecker &Salamon (1996) and Pavlidou & Venters (2008) is applicable to

our sample. The chance probability is 12%. This means that theGaussian distribution does not agree with the data. To investigatethe distribution of the photon index, more samples would berequired.

We evaluate the uncertainties in SED models by using variousSEDs. Figure 4 shows the total EGRB spectrum (absorbed +cascade) from the gamma-ray-loud radio galaxies with variousphoton index and break energy parameters. The contributionto the unresolved Fermi EGRB photon flux above 100 MeVbecomes 25.4%, 25.4%, and 23.8% for Γ = 2.39, 2.11, and2.67, respectively. In the case of Γ = 2.11, the contribution tothe EGRB flux above 10 GeV becomes significant. For the MeVbackground below 10 MeV, the position of the break energyand the photon index is crucial to determine the contributionof the gamma-ray-loud radio galaxies. As shown in Figure 4,higher break energy and softer photon index result in a smallercontribution to the MeV background radiation. To enable furtherdiscussion on the SED modeling, the multiwavelength spectralanalysis of all GeV-observed gamma-ray-loud radio galaxies isrequired.

5.3. Flaring Activity

It is well known that blazars are variable sources in gammarays (see, e.g., Abdo et al. 2009a, 2010d). If gamma-ray-loudradio galaxies are the misaligned populations of blazars, theywill also be variable sources. Kataoka et al. (2010) have recentlyreported that NGC 1275 showed a factor of ∼2 variation inthe gamma-ray flux. For other gamma-ray-loud radio galaxies,such a significant variation has not been observed yet (Abdoet al. 2010b). Currently, therefore, it is not straightforward tomodel the variability of radio galaxies. In this paper, we usedthe time-averaged gamma-ray flux of gamma-ray-loud radiogalaxies in the Fermi catalog, which is the mean of the Fermi 1 yrobservation. More observational information (e.g., frequency)is required to model the gamma-ray variability of radio galaxies.Further long-term Fermi observation will be useful, and futureobservation by ground-based imaging atmospheric Cherenkov

6

YI ’11

The Astrophysical Journal, 733:66 (9pp), 2011 May 20 Inoue

spectrum is given by

dN/dϵ ∝!ϵ−(p+1)/2 ϵ ! ϵbr,ϵ−(p+2)/2 ϵ > ϵbr,

(1)

where ϵbr corresponds to the IC photon energy from electronswith γbr (Rybicki & Lightman 1979).

The SED fitting for NGC 1275 and M87 shows that the ICpeak energy in the rest frame is located at ∼5 MeV (Abdo et al.2009b, 2009c). In this study, we use the mean photon index, Γc,as Γ at 0.1–10 GeV and set a peak energy, ϵbr, in the photonspectrum at 5 MeV for all gamma-ray-loud radio galaxies as abaseline model. Then, we are able to define the average SEDshape of gamma-ray-loud radio galaxies for all luminositiesas dN/dϵ ∝ ϵ−2.39 at ϵ >5 MeV and dN/dϵ ∝ ϵ−1.89 atϵ ! 5 MeV by following Equation (1).

However, only three sources are currently studied with multi-wavelength observational data. We need to make further studiesof individual gamma-ray-loud radio galaxies to understand theirSED properties in wide luminosity ranges. We examine otherspectral models in Section 5.2.

3. GAMMA-RAY LUMINOSITY FUNCTION

3.1. Radio and Gamma-ray Luminosity Correlation

To estimate the contribution of gamma-ray-loud radio galax-ies to the EGRB, we need to construct a GLF. However, becauseof the small sample size, it is difficult to construct a GLF usingcurrent gamma-ray data alone. Here, the RLF of radio galax-ies has been extensively studied in previous works (see, e.g.,Dunlop & Peacock 1990; Willott et al. 2001). If there is a cor-relation between the radio and gamma-ray luminosities, we areable to convert the RLF to the GLF with that correlation. Inthe case of blazars, it has been suggested that there is a corre-lation between the radio and gamma-ray luminosities from theEGRET era (Padovani et al. 1993; Stecker et al. 1993; Salamon& Stecker 1994; Dondi & Ghisellini 1995; Zhang et al. 2001;Narumoto & Totani 2006), although it has also been discussedthat this correlation cannot be firmly established because of flux-limited samples (Muecke et al. 1997). Recently, using the Fermisamples, Ghirlanda et al. (2010, 2011) confirmed that there is acorrelation between the radio and gamma-ray luminosities.

To examine a luminosity correlation in gamma-ray-loud radiogalaxies, we first derive the radio and gamma-ray luminosityof gamma-ray-loud radio galaxies as follows. Gamma-rayluminosities between the energies ϵ1 and ϵ2 are calculated by

Lγ (ϵ1, ϵ2) = 4πdL(z)2 Sγ (ϵ1, ϵ2)(1 + z)2−Γ , (2)

where dL(z) is the luminosity distance at redshift, z, Γ is thephoton index, and S(ϵ1, ϵ2) is the observed energy flux betweenthe energies ϵ1 and ϵ2. The energy flux is given from the photonflux Fγ , which is in the unit of photons cm−2 s−1, above ϵ1 by

Sγ (ϵ1, ϵ2) = (Γ − 1)ϵ1

Γ − 2

"#ϵ2

ϵ1

$2−Γ− 1

%

Fγ , (Γ = 2) (3)

Sγ (ϵ1, ϵ2) = ϵ1 ln(ϵ2/ϵ1)Fγ , (Γ = 2). (4)

Radio luminosity is calculated in the same manner.

40

41

42

43

44

45

46

47

48

38 39 40 41 42 43 44 45

log 1

0(L

γ [er

g/s]

)

log10(L5 GHz [erg/s])

FRIFRII

Figure 1. Gamma-ray luminosity at 0.1–10 GeV vs. radio luminosity at 5 GHz.The square and triangle data represent FRI and FRII galaxies, respectively. Thesolid line is the fit to all sources.(A color version of this figure is available in the online journal.)

Figure 1 shows the 5 GHz and 0.1–10 GeV luminosity relationof Fermi gamma-ray-loud radio galaxies. Square and triangledata represent FRI and FRII radio galaxies, respectively. Thesolid line shows the fitting line to all the data. The function isgiven by

log10(Lγ ) = (−3.90±0.61) + (1.16±0.02) log10(L5 GHz), (5)

where errors show 1σ uncertainties. In the case of blazars, theslope of the correlation between Lγ (>100 MeV), luminosityabove 100 MeV, and radio luminosity at 20 GHz is 1.07 ± 0.05(Ghirlanda et al. 2011). The correlation slopes of gamma-ray-loud radio galaxies are similar to those of blazars. This indicatesthat the emission mechanism is similar in gamma-ray-loud radiogalaxies and blazars.

We need to examine whether the correlation between theradio and gamma-ray luminosities is true or not. In the flux-limited observations, the luminosities of samples are stronglycorrelated with redshifts. This might result in a spurious lu-minosity correlation. As in previous works on blazar samples(Padovani 1992; Zhang et al. 2001; Ghirlanda et al. 2011),we perform a partial correlation analysis to test the correla-tion between the radio and gamma-ray luminosities exclud-ing the redshift dependence (see the Appendix for details).First, we calculate the Spearman rank–order correlation co-efficients (see, e.g., Press et al. 1992). The correlation co-efficients are 0.993, 0.993, and 0.979 between log10 L5 GHzand log10 Lγ , between log10 L5 GHz and redshift, and betweenlog10 Lγ and redshift, respectively. Then, the partial correlationcoefficient becomes 0.866 with chance probability 1.65×10−6.Therefore, we conclude that there is a correlation between theradio and gamma-ray luminosities of gamma-ray-loud radiogalaxies.

3.2. Gamma-ray Luminosity Function

In this section, we derive the GLF of gamma-ray-loud radiogalaxies, ργ (Lγ , z). There is a correlation between the radioand gamma-ray luminosities as shown in Equation (5). Withthis correlation, we develop the GLF by using the RLF of radiogalaxies, ρr (Lr, z), with radio luminosity, Lr. The GLF is given

3

YI ’11

Star-forming Galaxies

• Soltan ’99; Pavlidou & Fields ’02; Thompson +’07; Bhattacharya & Sreekumar 2009; Fields et al. 2010; Makiya et al. 2011; Stecker & Venters 2011; Lien+’12, Ackermann+’12; Lacki+’12; Chakraborty & Fields ’13; Tamborra+’14

• Use gamma-ray and infrared luminosity correlation

• ~10-30% of CGB at 0.1-100 GeV.

The Astrophysical Journal, 755:164 (23pp), 2012 August 20 Ackermann et al.

E (GeV)-110 1 10 210

)-1

sr

-1 s

-2 d

N/d

E (G

eV c

m2

E

-810

-710

-610

= 2.2)ΓThompson et al. 2007 (Fields et al. 2010 (Density-evolution)Fields et al. 2010 (Luminosity-evolution)Makiya et al. 2011Stecker & Venters 2011 (IR)Stecker & Venters 2011 (Schecter)

Isotropic DiffuseStar-forming GalaxiesMilky Way Spectrum, 0<z<2.5

= 2.2, 0<z<2.5ΓPower Law

Figure 7. Estimated contribution of unresolved star-forming galaxies (bothquiescent and starburst) to the isotropic diffuse gamma-ray emission measuredby the Fermi-LAT (black points; Abdo et al. 2010f). The shaded regions indicatecombined statistical and systematic uncertainties in the contributions of therespective populations. Two different spectral models are used to estimate theGeV gamma-ray emission from star-forming galaxies: a power law with photonindex 2.2, and a spectral shape based on a numerical model of the global gamma-ray emission of the Milky Way (Strong et al. 2010). These two spectral modelsshould be viewed as bracketing the expected contribution since multiple star-forming galaxy types contribute, e.g., dwarfs, quiescent spirals, and starbursts.We consider only the contribution of star-forming galaxies in the redshift range0 < z < 2.5. The gamma-ray opacity of the universe is treated using theextragalactic background light model of Franceschini et al. (2008). Severalprevious estimates for the intensity of unresolved star-forming galaxies areshown for comparison. Thompson et al. (2007) treated starburst galaxies ascalorimeters of CR nuclei. The normalization of the plotted curve depends onthe assumed acceleration efficiency of SNRs (0.03 in this case). The estimatesof Fields et al. (2010) and by Makiya et al. (2011) incorporate results from thefirst year of LAT observations. Fields et al. (2010) considered the extreme casesof either pure luminosity evolution and pure density evolution of star-forminggalaxies. Two recent predictions from Stecker & Venters (2011) are plotted: oneassuming a scaling relation between IR-luminosity and gamma-ray luminosity,and one using a redshift-evolving Schechter model to relate galaxy gas mass tostellar mass.(A color version of this figure is available in the online journal.)

this component and to predict the cosmogenic ultra-high energyneutrino flux originating from charged pion decays of the ultra-high energy CR interactions (Ahlers et al. 2010; Berezinskyet al. 2011; Wang et al. 2011).

Galactic sources, such as a population of unresolved millisec-ond pulsars at high Galactic latitudes, could become confusedwith isotropic diffuse emission as argued by Faucher-Giguere& Loeb (2010). Part of the IGRB may also come from our SolarSystem as a result of CR interactions with debris of the OortCloud (Moskalenko & Porter 2009).

Finally, a portion of the IGRB may originate from “newphysics” processes involving, for instance, the annihilation ordecay of dark matter particles (Bergstrom et al. 2001; Ullio et al.2002; Taylor & Silk 2003).

Studies of anisotropies in the IGRB intensity on small angularscales provide another approach to identify IGRB constituentsource populations (Siegal-Gaskins 2008). The fluctuation an-gular power contributed by unresolved star-forming galaxies isexpected to be small compared to other source classes becausestar-forming galaxies have the highest spatial density amongconfirmed extragalactic gamma-ray emitters, but are individ-ually faint (Ando & Pavlidou 2009). Unresolved star-forminggalaxies could in principle explain the entire IGRB intensitywithout exceeding the measured anisotropy (Ackermann et al.2012a). By contrast, the fractional contributions of unresolved

Redshift0 0.5 1 1.5 2 2.5

) (L

mµ8-

1000

lo

g L

8

9

10

11

12

13

Cum

ulat

ive

Con

trib

utio

n F

ract

ion

0<z<

2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

IR Survey Completeness Threshold

Redshift0 0.5 1 1.5 2 2.5C

umul

ativ

e C

ontr

ibut

ion

Fra

ctio

n 0<

z<2.

5

0

0.2

0.4

0.6

0.8

1

Figure 8. Relative contribution of star-forming galaxies to the isotropic diffusegamma-ray background according to their redshift and total IR luminosity(8–1000 µm) normalized to the total contribution in the redshift range 0 < z <2.5. Top panel: solid contours indicate regions of phase space which contributean increasing fraction of the total energy intensity (GeV cm−2 s−1 sr−1) from allstar-forming galaxies with redshifts 0 < z < 2.5 and 108 L⊙ < L8–1000 µm <

1013 L⊙. Contour levels are placed at 10% intervals. The largest contributioncomes from low-redshift Milky Way analogues (L8–1000 µm ∼ 1010 L⊙) andstarburst galaxies comparable to M82, NGC 253, and NGC 4945. The blackdashed curve indicates the IR luminosity above which the survey used to generatethe adopted IR luminosity function is believed to be complete (Rodighiero et al.2010). Bottom panel: cumulative contribution vs. redshift. As above, only theredshift range 0 < z < 2.5 is considered.(A color version of this figure is available in the online journal.)

blazars and millisecond pulsars to the IGRB intensity are con-strained to be less than ∼20% and ∼2%, respectively, due tolarger angular power expected for those source classes.

6. GALAXY DETECTION OUTLOOKFOR THE FERMI-LAT

The scaling relations obtained in Section 4.3 allow straight-forward predictions for the next star-forming galaxies whichcould be detected by the LAT. We use the relationship betweengamma-ray luminosity and total IR luminosity to select the mostpromising targets over a 10 year Fermi mission.

We begin by creating an IR flux-limited sample of galaxiesfrom the IRAS Revised Bright Galaxies Sample (Sanders et al.2003) by selecting all the galaxies with 60 µm flux densitygreater than 10 Jy (248 galaxies). Next, 0.1–100 GeV gamma-ray fluxes of the galaxies are estimated using the scalingrelation between gamma-ray luminosity and total IR luminosity.Intrinsic dispersion in the scaling relation is addressed bycreating a distribution of predicted gamma-ray fluxes for each

18

Ackermann+’12

The Astrophysical Journal, 755:164 (23pp), 2012 August 20 Ackermann et al.

)-1 (W Hz1.4 GHzL

1910 2010 2110 2210 2310 2410

)-1

(er

g s

0.1-

100

GeV

L

3710

3810

3910

4010

4110

4210

4310

)-1 yrSFR (M-210 -110 1 10 210 310

M33M31

SMC

NGC 6946

Milky Way

M82M83

Arp 220

NGC 4945

IC 342

NGC 253

NGC 1068NGC 2146

LMC

LAT Non-detected (Upper Limit)

LAT Non-detected with AGN (Upper Limit)

LAT DetectedLAT Detected with AGN

Best-fitFit UncertaintyDispersion

Calorimetric Limit erg)50 = 10η

SN(E

) (W Hz1.4 GHzL

1910 2010 2110 2210 2310 2410

) (

erg

s0.

1-10

0 G

eVL

3710

3810

3910

4010

4110

4210

4310

)-1 (W Hz1.4 GHzL

1910 2010 2110 2210 2310 2410

1.4

GH

z /

L0.

1-10

0 G

eVL

10

210

310

410

510

)-1 yrSFR (M-210 -110 1 10 210 310

M33

M31SMCNGC 6946

Milky Way

M82

M83

Arp 220

NGC 4945IC 342

NGC 253

NGC 1068

NGC 2146

LMC

LAT Non-detected (Upper Limit)

LAT Non-detected with AGN (Upper Limit)

LAT DetectedLAT Detected with AGN

Best-fitFit UncertaintyDispersion

Calorimetric Limit erg)50 = 10η

SN(E

) (W Hz1.4 GHzL

1910 2010 2110 2210 2310 2410

1.4

GH

z /

L0.

1-10

0 G

eVL

10

210

310

410

510

Figure 3. Top panel: gamma-ray luminosity (0.1–100 GeV) vs. RC luminosityat 1.4 GHz. Galaxies significantly detected by the LAT are indicated with filledsymbols whereas galaxies with gamma-ray flux upper limits (95% confidencelevel) are marked with open symbols. Galaxies hosting Swift-BAT AGNs areshown with square markers. RC luminosity uncertainties for the non-detectedgalaxies are omitted for clarity, but are typically less than 5% at a fixed distance.The upper abscissa indicates SFR estimated from the RC luminosity according toEquation (2) (Yun et al. 2001). The best-fit power-law relation obtained using theEM algorithm is shown by the red solid line along with the fit uncertainty (darkershaded region), and intrinsic dispersion around the fitted relation (lighter shadedregion). The dashed red line represents the expected gamma-ray luminosityin the calorimetric limit assuming an average CR luminosity per supernovaof ESN η = 1050 erg (see Section 5.1). Bottom panel: ratio of gamma-rayluminosity (0.1–100 GeV) to RC luminosity at 1.4 GHz.(A color version of this figure is available in the online journal.)

Although these three SFR estimators are intrinsically linked,each explores a different stage of stellar evolution and issubject to different astrophysical and observational systematicuncertainties.

Figures 3 and 4 compare the gamma-ray luminosities ofgalaxies in our sample to their differential luminosities at1.4 GHz, and total IR luminosities (8–1000 µm), respectively.

) (Lmµ8-1000 L

810 910 1010 1110 1210

)-1

(er

g s

0.1-

100

GeV

L

3710

3810

3910

4010

4110

4210

4310

)-1 yrSFR (M-210 -110 1 10 210 310

M33

M31

SMC

NGC 6946

Milky Way

M82

M83

Arp 220

NGC 4945

IC 342

NGC 253

NGC 1068

NGC 2146

LMC

LAT Non-detected (Upper Limit)

LAT Non-detected with AGN (Upper Limit)

LAT DetectedLAT Detected with AGN

Best-fitFit UncertaintyDispersion

Calorimetric Limit erg)50 = 10η

SN(E

) (Lmµ8-1000 L

810 910 1010 1110 1210

)-1

(er

g s

0.1-

100

GeV

L

3710

3810

3910

4010

4110

4210

4310

) (Lmµ8-1000 L

810 910 1010 1110 1210

mµ8-

1000

/

L0.

1-10

0 G

eVL

-510

-410

-310

-210

-110

)-1 yrSFR (M-210 -110 1 10 210 310

M33

M31SMC

NGC 6946

Milky Way

M82

M83

Arp 220

NGC 4945

IC 342

NGC 253 NGC 1068

NGC 2146

LMC

LAT Non-detected (Upper Limit)

LAT Non-detected with AGN (Upper Limit)

LAT DetectedLAT Detected with AGN

Best-fitFit UncertaintyDispersion

Calorimetric Limit erg)50 = 10η

SN(E

) (Lmµ8-1000 L

810 910 1010 1110 1210

mµ8-

1000

/

L0.

1-10

0 G

eVL

-510

-410

-310

-210

-110

Figure 4. Same as Figure 3, but showing gamma-ray luminosity (0.1–100 GeV)vs. total IR luminosity (8–1000 µm). IR luminosity uncertainties for the non-detected galaxies are omitted for clarity, but are typically ∼0.06 dex. Theupper abscissa indicates SFR estimated from the IR luminosity according toEquation (1) (Kennicutt 1998b).(A color version of this figure is available in the online journal.)

A second abscissa axis has been drawn on each figure toindicate the estimated SFR corresponding to either RC or totalIR luminosity using Equations (2) and (1). The upper panelsof Figures 3 and 4 directly compare luminosities betweenwavebands, whereas the lower panels compare luminosity ratios.Taken at face value, the two figures show a clear positivecorrelation between gamma-ray luminosity and SFR, as hasbeen reported previously in LAT data (see in this context Abdoet al. 2010b). However, sample selection effects, and galaxiesnot yet detected in gamma rays must be taken into account toproperly determine the significance of the apparent correlations.

We test the significances of multiwavelength correlationsusing the modified Kendall τ rank correlation test proposed byAkritas & Siebert (1996). This method is an example of “survival

9

Ackermann+’12

Sum of Components

•  Blazars, star-forming galaxies and radio galaxies can explain the intensity and the spectrum of the EGB

Preliminary

As usual: it does not include the systematic uncertainty on the EGB!

Components of the Cosmic Gamma-ray Background

• Blazars (Ajello+’15), Radio gals. (YI’11), & Star-forming galaxies (Ackermann+’12) make up almost 100% of CGB from 0.1-10

3 GeV.

• But,,,

• # of detected radio gals. and star-forming gals. is ~10.

• TeV spectra of blazars are not well established. Redshifts of ~50% of BL Lacs are not measured.

Ajello, YI +’15

Upper Bound on the Cosmic TeV Gamma-ray Background

• Cascade component from VHE CGB can not exceed the Fermi data

(Coppi & Aharonian ’97, YI & Ioka ’12, Murase+’12, Ackermann+’14).

• No or negative evolution is required -> low-luminosity BL Lacs show negative evolution (Ajello+’14).

Cascade

Absorbed

ULIntrinsic

YI & Ioka ’12

Cosmic TeV Gamma-ray Background

• The TeV blazar data give lower limit on to the cosmic gamma-ray background.

• Current limit at 0.3-10 TeV is

• Fermi has resolved more portion of the TeV sky than IACTs do?

• Need to remove ~3 orders higher electron background to detect the CGB with CTA.

10-8

10-7

10-6

10-5

10-4

10-3

103 104 105 106 107 108 109

E2 dN/

dE [M

eV2 /c

m2 /s/

sr/M

eV]

Photon Energy [MeV]

Total CGB (Fermi)Resolved CGB (Fermi)

VHE BlazarsUpper Limit (II12)

Mrk 421Mrk 501

10-7

10-6

10-5

10-4

10-3

10-2

10-1

100

101

103 104 105 106 107 108 109 1010

E2 dN

/dE

[MeV

2 /cm

2 /s/sr

/MeV

]

Photon Energy [MeV]

Total CGB (Fermi)Resolved CGB (Fermi)

VHE BlazarsUpper Limit (II12)

IceCube (i per flavour)AMS (e++e-)

3⇥ 10�5

✓E

100GeV

◆�1

[MeV/cm2/s/sr] < E2 dN

dE< 5⇥ 10�5

✓E

100GeV

◆�0.7

[MeV/cm2/s/sr]

YI & Tanaka in prep.

Summary• We have not understood blazar emission mechanism to constrain

the cosmic infrared background from gamma-ray observations

• New emission mechanisms: secondary gamma rays, stochastic acceleration, lepto-hadronic emission…

• At least, the GeV - TeV index distribution of blazars has no correlation.

• The cosmic GeV gamma-ray background is from blazars, radio galaxies, and star-forming galaxies

• Current GeV & TeV observations give strict constraints on the cosmic TeV gamma-ray background.