PreprinttypesetusingLATEXstyleemulateapjv.2/19/04 BLAZARSASULTRA-HIGH-ENERGYCOSMIC-RAYSOURCES:IMPLICATIONSFORTEVGAMMA-RAY OBSERVATIONS KOHTAMURASE1,CHARLESD.DERMER2,HAJIMETAKAMI3,ANDGIULIAMIGLIORI4 ABSTRACT ThespectraofBLLacobjectsandFanaroff-RileyIradiogalaxiesarecommonlyexplainedbytheone-zone leptonicsynchrotronself-Compton(SSC)model. Spectralmodelingofcorrelatedmultiwavelengthdatagives 2 thecomovingmagneticfieldstrength,thebulkoutflowLorentzfactorandtheemissionregionsize. Assuming 1 thevalidityofthe SSC model,the Hillasconditionshowsthatonlyin rarecasescan suchsourcesaccelerate 0 protonstomuchabove1019 eV,so&1020eVultra-high-energycosmicrays(UHECRs)arelikelytobeheavy 2 ions if powered by this type of radio-loud active galactic nuclei (AGN). Survival of nuclei is shown to be n possibleinTeVBLLacsandmisalignedcounterpartswithweakphotohadronicemissions. Anothersignature a ofhadronicproductionisintergalacticUHECR-inducedcascadeemission,whichisanalternativeexplanation J oftheTeVspectraofsomeextremenon-variableblazarssuchas1ES0229+200or1ES1101-232.Westudythis 1 kindofcascadesignal,takingintoaccounteffectsofthestructuredextragalacticmagneticfieldsinwhichthe 3 sourcesshouldbeembedded.Wedemonstratetheimportanceofcosmic-raydeflectionsontheγ-rayflux,and show that requiredabsolute cosmic-rayluminositiesare largerthan the average UHECR luminosityinferred ] E from UHECR observationsand can even be comparable to the Eddingtonluminosity of supermassive black holes. Future TeV γ-ray observations using the Cherenkov Telescope Array and the High Altitude Water H Cherenkov detector array can test for UHECR acceleration by observing >25 TeV photons from relatively . h low-redshiftsourcessuchas1ES0229+200,and&TeVphotonsfrommoredistantradio-loudAGN. p Subjectheadings:galaxies:active—gammarays:galaxies—cosmicrays - o r t 1. INTRODUCTION Radio-loud AGN detected at TeV energies consist s mainly of high-synchrotron-peaked BL Lac objects, in- a Activegalacticnuclei(AGN)withextendedradiojetspow- cluding the ultra-variable TeV blazars Mrk 421 (z=0.031; [ eredbysuper-massiveblackholesareamongthemostlumi- Fossatietal. 2008), Mrk 501 (z=0.033; Albertetal. 2007b) nousobjectsin the low-redshiftuniverse. Since 2004, when 2 and PKS 2155-304 (z=0.116; Aharonianetal. 2007a), the present generation of imaging atomspheric Cherenkov v and the apparently non-variable TeV blazars 1ES telescopes began to operate, the number of extragalactic 6 0229+200 (Aharonianetal. 2007b) and 1ES 1101- 7 sourcesdetected at &0.1 TeV (very-highenergy; VHE) en- 5 ergies has grown rapidly, and is nearly fifty.5 The Fermi 232 (Aharonianetal. 2007c). Extragalactic VHE γ-ray galaxies include several Fanaroff-Riley (FR) class I radio 5 Gamma-raySpaceTelescope,nowinitsfourthmissionyear, galaxies (Cen A, M87, NGC 1275; Aharonianetal. 2009a; . is providing a wealth of new discoveries on γ-ray galax- 7 Aharonianetal. 2006; Aleksic´etal. 2011c, respectively). ies. In the high-confidence clean sample of active galac- 0 Cen A (Abdoetal. 2010d), M87 (Abdoetal. 2009c), tic nuclei associations in the First Fermi LAT AGN catalog 1 and NGC 1275 (Abdoetal. 2009b; Kataokaetal. 2010; 1 (1LAC;Abdoetal. 2010a),morethan600γ-rayblazars,di- Brown&Adams2011) have also been detected at GeV : vided about equally into BL Lac objects and flat spectrum v radio quasars (FSRQs), were reported. New classes of GeV energies. This list also includes the head-tail radio i galaxy IC 310 (Aleksic´etal. 2010; Neronovetal. 2010), X γ-ray galaxies, e.g., radio-loud narrow line Seyfert galax- intermediate-synchrotron-peaked objects like ies(Abdoetal. 2009a)andstar-forminggalaxiespoweredby r 3C 66A (Acciarietal. 2009a; Aliuetal. 2009; a supernovae rather than black holes (Abdoetal. 2010b), fol- Abdoetal.2011a) and BL Lac (Albertetal. 2007a; lowing closely the VHE detections of the starburst galaxies Abdoetal. 2011b), and the GeV luminous, high-redshift NGC253(Aceroetal. 2009)andM82(Acciarietal. 2009c), FSRQs 3C 279 (z=0.538; Aleksic´etal. 2011a), PKS 1510- arenowfirmlyestablished.Moreover,thespectralenergydis- 089 (z=0.361; Wagneretal. 2010), and 4C +21.35 (PKS tributions(SEDs)ofradiogalaxiesdetectedatGeVandVHE, being misalignedby large (&10◦) anglesto the jet axis and 1222+216,z=0.432;Aleksic´ etal. 2011b). The γ-ray data from weak-lined BL Lac objects and thoughttobetheparentpopulationofblazarsingeometrical FR-I radio galaxies are generally well fit with the unificationscenarios(Urry&Padovani1995), arehelpingto standard nonthermal electronic synchrotron self-Compton revealtheblazarjetgeometry.About10suchsourcesarenow (SSC) relativistic jet model(e.g., Mastichiadis&Kirk1997; detectedwithFermi(Abdoetal. 2010c). Tavecchioetal. 1998; Katoetal. 2006), but the use of 1 Department of Physics, Center for Cosmology and Astro-Particle archivaldataforhighlyvariableblazarsgavelargeparameter Physics,TheOhioStateUniversity,Columbus,OH43210,USA uncertaintiesinthepast. Withtherecentsimultaneousmulti- 2 SpaceScience Division, NavalResearch Laboratory, Washington, DC wavelengthdatasetsformanysources,accurateparameteres- 20375,USA timationcanbemade,eitherfromsimplescalingresultsinthe 3MaxPlanckInstituteforPhysics,FöhringerRing6,80805Munich,Ger- Thomsonregime,orfromdetailedspectralcalculationstaking many 4 Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cam- intoaccounttheKlein-Nishina(KN)effectthatisrelevantfor bridge,MA02138,USA high,UV/X-raysynchrotron-peakedblazars. 5Seetevcat.uchicago.edu/andwww.mpp.mpg.de/ rwagner/sources/ Thepaperisorganizedasfollows. InSection2,weassem- ∼ 2 Muraseetal. ble the derived parameter values obtained in various analy- 2.1. Two-ComponentSpectraofBlazarsandRadioGalaxies sesoftypicalBLLacobjectsandFR-Iradiogalaxies. From In the framework of one-zone leptonic synchrotron/SSC thesenumbers,weobtainmaximumenergiesofcosmicrays, models,thedouble-humpedSEDofblazarsandradiogalaxies and show that protons can be accelerated to &10 EeV en- (misalignedcounterparts),plottedaslog(νF )vs. log(ν),can ergiesonly in a few radio galaxiesand flares of BL Lac ob- ν be well described by a low-energy synchrotron component jects. Thenweexploretheassociatedhadronicsignaturesex- and a high-energy inverse-Compton (IC) curve. Each com- pected from TeV blazars in the case where jets of BL Lac ponent is characterized by the peak synchrotron flux νFs at objectsandFR-Iradiogalaxiesareacceleratorsofultra-high- ν peaksynchrotronphotonenergyε ,andthepeakICfluxνFC energy cosmic rays (UHECRs, with energies above the an- s ν at peak IC energy ε , respectively. The spectrum of radio- kle of 1018.5 eV), assuming the validity of the leptonic C loudAGNisoftenhighlyvariable,andrisesanddecayswith ≈ SSCparametersinferredfromrapidlyvariableAGN.Wedis- variability time t . Here, we define the variability time as var cussobservablesignalsproducedin thesourceandcalculate the shortest timescale in which the flux shows a significant thosegeneratedoutsidethesourcefrombothγraysandUHE- factor-of-twochange, and assume co-spatiality of the highly CRsescapingfromthejetacceleratorandpassingthroughthe variableemissionsinthedifferentenergybands,asindicated Mpc-scale regions of cosmic structure, magnetized clus- bycorrelatedvariability(Ackermannetal. 2011). ∼ ters and filaments, and the larger 100 Mpc-scale voids of Here we consider such a one-zone model. The emis- ∼ intergalactic space. For some apparently non-variable TeV sions are produced by electrons in a spherical blob moving blazars, where the one-zone synchrotron/SSC model typi- relativistically in a jetted geometry: the high-energy emis- cally requires extreme parameters for fits, the cascade radi- sion is produced via Compton scattering off the local syn- ationcanbeacrucialcomponentofthehigh-energyradiation chrotron photons that are generated by the electrons in the spectrum,dependingonthestrengthoftheintergalacticmag- blob. The synchrotron (IC) luminosities at peak energy are netic field (IGMF) in voids, as has recently been proposed Ls(C) 4πd2(ε Fs(C))andtheComptondominanceparam- to explain the γ-ray spectra of some extreme blazars such γ ≈ L s(C) ε eterisdefinedasA LC/Ls (ε FC)/(ε Fs). Theelectron as1ES0229+200(Essey&Kusenko2010;Esseyetal. 2010; C≡ γ γ ≈ C ε s ε distributionistypicallyassumedtoconsistofpower-lawseg- Esseyetal. 2011). We focus on such cascade emissions in ments. AslongasboththeνFs andνFC energyfluxesorigi- the VHE range in Section 3. Using numerical calculations, ν ν natefromelectronswiththesamemeancomoving-frameen- wealsodemonstratetheimportanceofstructuredextragalac- ergyγ′m c2,themagneticenergydensitycanberewrittenin ticmagneticfields(EGMFs)inclustersandfilamentsforfu- b e ture γ-ray detectability by the Cherenkov Telescope Array termsofACasB′2/8π∼Lsγ/(4πR′2δ4cAC). ThentheDoppler (CTA)andtheHighAltitudeWater Cherenkov(HAWC)de- factorisgiveninthisrelativisticsphericalblobformalismby tector array. Implications of this study and a summary are theexpression(Ghisellinietal. 1996) given in Section 4. Throughoutthis work, the cosmological parametersaretakenasH0=71kms- 1Mpc- 1,Ωm=0.3,and δ 31/2(Lsγ)1/4(εC/mec2)1/2 , (1) ΩΛ=0.7. ∼ 23/4c3/4t1/2A1/4B1/2(ε /m c2) var C Q s e andthecomovingmagneticfieldis 2. VHEBLAZARSANDUHECRS 211/4c3/4t1/2A1/4B1/2(ε /m c2)3 Of the wide variety of source classes that could po- B′ (1+z) var C Q s e tentially accelerate UHECRs, including, for example, ∼ 33/2(Ls)1/4(ε /m c2)3/2 γ C e GRBs (Waxman1995; Vietri1995; Muraseetal. 2008a), 4B (ε /m c2) fast rotating magnetars (Arons2003), structure forma- (1+z) Q s e , (2) tion shocks in galaxy clusters (Normanetal. 1995; ∼ 3δ(ε /ε ) C s Kangetal. 1996; Inoueetal. 2007), and quasar rem- providedtheComptonscatteringtakesplaceintheThomson nants (Boldt&Ghosh1999; Levinson2000), radio-loud AGN with jets seem privileged in that the most pronounced regime, which applies when δ & δT =2√3 εsεc/m2ec4(1+ excess in arrival directions of UHECRs is positionally z).7 Here the critical magnetic field is dpefined as BQ ≡ centered in the vicinity of the FR-I radio galaxy Cen- m2c3/e¯h 4.4 1013 G (e.g., Brainerd&Petrosian1987). taurus A (Abrahametal. 2008; Abrahametal. 2009). 6 Thee abov≃e equ×ations are derived by using the common FR-I radio galaxies, including their aligned counterparts relations ε /m c2 δ(B′/B )γ′2/(1+z), and ε /m c2 (eBmLissLivaictyoobfjects)1,0r4a5d–i1a0te46 aervgolMumpec-- 3anydr- 1timine-anvoenrathgeerd- (4/3)γb′2εss(i.ee., th≈e typicalQflubid-frame Lorentzcfacteor o≈f ≈ electrons radiating near the peak synchrotron and SSC fre- mal γ rays (Dermer&Razzaque2010), and FR-I radio quenciesisγ′ (√3/2) ε /ε intheThomsonlimit)(e.g., galaxies are found within the 100 Mpc Greisen- b≈ c s Zatsepin-Kuzmin radius. If co≈mparable power goes Sikoraetal. 2009). p For high-peaked BL Lac objects, the Compton scattering into the acceleration of UHECRs, then BL Lac ob- oftenoccursintheKNregime,wheremoredetailedmodeling jects and FR-I radio galaxies have more than sufficient is required. Nevertheless, equations(1) and (2) demonstrate emissivity to power the & 10 EeV UHECRs, which re- quire 1044 erg Mpc- 3 yr- 1 (Waxman&Bahcall1999; that source parameters such as δ and B′ can be determined ∼ fromthedouble-humpedSED.8 Berezinskyetal. 2006;Murase&Takami2009). 6SeeTakami&Sato2009fordiscussionsonissuesofUHECRanisotropy 7 The inequality is derived from 4γb′εs(1+z)/δ . mec2 and εC ≈ fnourclperio.toNnost,eaanlsdoLthemepoointeen&tialWcaoxnmtriabnut2io0n09f;roAmbrtehuebetacakl.gro2u0n1d1Cfoernthaeuarvuys (4/83)γOb′t2hεesr. parameters such as the acceleration efficiency η (ta′cc = superclusterpointedoutbyGhisellinietal.(2008). ηγ′mec2/(eB′c))dependondetails oftheelectron distribution. Forexam- BLLacObjectsasUHECRSources 3 Table1 givesmeasuredandinferredpropertiesforblazars there is the large scatter in parameters due to uncertainties. and radio galaxieswith goodmultiwavelengthcoveragethat However,evenwiththeallowedspreadintheparameterval- are well described by a synchrotron/SSC model, where de- ues(seeTable2),thisconclusionseemsrobust. rivedvaluesofmagneticfieldsandDopplerandLorentzfac- A similar conclusionis also reachedwhen consideringlu- tors based on detailed synchrotron and SSC modeling are minosity requirements for BL Lac objects and FR-I radio taken from the literature (rather than using equations 1 and galaxies (Dermer&Razzaque2010; Ghisellinietal. 2010). 2). In Table 2, we also show parametersderivedwith equa- InGhisellinietal. (2010),physicalparameterswereobtained tions(1)and(2)(onlywhenδ&δ issatisfied). Forblazars, viaspectralmodelingusingtheirone-zoneleptonicmodelof T weassumethatΓ δ,whereasvaluesoftheangleofthejet- all blazars with known redshift detected by the Fermi satel- ≈ tedemissionwithrespecttotheobserverinferredfromobser- lite during its first 3-month survey. The inferred magnetic cvδattvioarn/s(1a+rez)coisnusisdeedr.ed for radio galaxies. Also, R′ ≈ctv′ar = alunmdianlomsoitsytoafllBoLftLhaecmosbajteicsftsyiLsty.pi2cally10L4B7δ∼2 e1r0g46sδ- 112,ewrghes-re1 TheSED modelinghasmuchimprovedthankstothecon- B × 1 δ = δ/10. On the other hand, the required magnetic lu- 1 stantlyincreasingmulti-wavelengthcoverage,whichalso al- minosity for UHECR acceleration to 1020Emax eV is L lows simultaneous multi-band observations. Nevertheless, A,20 B ≈ there is still some degree of scatter among parameter sets 2×1047Γ21(EAm,a2x0)2Z- 2 erg s- 1 where Γ1 = Γ/10. Hence, it obtained by different groups, partially related to unavoid- also suggests difficulties in acceleration of &1020.5 eV pro- able parameter-degeneracy with respect to the observables. tonsintypicalBLLacobjects,thoughthesimpleSSCmodel In some cases, there is even significant scatter between de- cannotbesosimplyappliedtolower-peakedBLLacobjects rivedparametersforthesamestate ofaspecificsource(Cen wherevaluesofB′ andδ arenotwell-definedduepossiblyto A)orbetweendifferentstatesofthesamesource(Mrk501). externalComptonscatteringcomponents. In particular, the whole SED of Cen A up to the TeV-band If BL Lac objects and FR-I radio galaxies, as has cannot be fitted with a unique parameter set. Abdo et al. often been considered (e.g., Tinyakov&Tkachev2001; (2010d)show differentmodels and parameter values for the Berezinskyetal. 2002; Dermeretal. 2009), accelerate the samedata,whichreflectthelimitationsoffittingsingleepoch, UHECRs, then an ultra-high-energy (UHE) proton origin singlecomponentSEDsforderivationsofsourceparameters. of the highest-energy cosmic rays is disfavored from spec- However,ourconclusionsarenotaffectedbyuncertaintiesof tral modeling if the standard synchrotron/SSCmodelis cor- thepoorlyconstrainedparameters,aswediscussinmorede- rect. Heavier nuclei can, however, be accelerated up to tailbelow. ultra-high energies. The composition of UHECRs is an open question, with both proton and heavy-ion dominated 2.2. ImplicationsforUHECRAcceleration compositions having been claimed to be compatible with TheHillascondition(Hillas1984)limitsthemaximumac- HiRes(Abbasietal. 2010)andthePierreAugerObservatory celeratedenergyofionswithchargeZto (PAO)(Abrahametal. 2010)data,respectively.Asseenhere, the standard model for γ-ray emission from BL Lac objects EAmax≈ZeB′ΓR′, (3) and FR-I radio galaxiessuggests a transition from protonto in order that the particle Larmor radius is smaller than the heavy-iondominatedcompositionat (1018- 1019)eV. characteristic size scale R′ . ct′ = cδt /(1+z). The in- We have assumed that γ rays fro∼m TeV blazars and ra- var var equality is replaced by an equality in our estimates. Equa- dio galaxies are from leptonic Compton scattering in order tion (3) can be rewritten using the Thomson-limit relations thatsynchrotrontheorycanbeusedtoderivethevariouspa- givenabovewhenδ Γ,inwhichcaseonegets rameters. Thus the hadronic γ-ray flux must be consider- ≈ ablysmallerforaconsistentinterpretation.Sufficientlyhigh- 4(Ls)1/4t1/2B1/2ε energyprotonsand nucleiinteractwith synchrotronphotons EAmax∼Ze 3γc3/4Av1a/r4ε1Q/2 s. (4) in the jet via the photomeson process, with photopion pro- C C duction efficiency f for cosmic-ray protons estimated to pγ NoticethatEmaxisdefinedinthecosmicrestframeatUHECR be(e.g.,Murase&Beacom2010a) A production. InTable2,themaximumprotonenergyisestimatedusing f 2.3 10- 4 2.5 Ls t- 1 δ- 4 1keV Ep α- 1, equation(3)andparametersgiveninTable1,whichinturnare pγ ≃ × (cid:18)1+α(cid:19) γ,46var,4 1 (cid:18) ε (cid:19)(cid:18)Eb(cid:19) s p based on the results of synchrotron/SSCmodelfits forthese (5) sourcesfoundindedicatedmodelingpapers(ratherthanusing whereEb 1.6 1016 eV(ε /1keV)- 1δ2(1+z)- 2 isthetyp- equations1and2). Here,notethatotherlossesanddetailsof ical enerpg≃y of a×proton thats interacts w1ith a photon with theaccelerationprocesscouldlimitthemaximumparticleen- ε (where the proton energy is here defined in the observer s ergyfurther. ForthecasesconsideredinTable1, itisbarely frame).Also,αisthephotonindexatenergiesbeloworabove possible to accelerate protons up to 1020 eV, whereas Fe ε .Forα 1.5,whichistypicalofBLLacobjectsatE >Eb, nucleicouldeasily each reach &1020∼eV providedthatthey thsephoto≈mesonproductionefficiencyat 1019 eVbepcomeps cansurvivephotodisintegration. Alternately,&1020 eV pro- oforder f 6 10- 3,whichsuggestst∼hatthephotomeson pγ tonaccelerationcouldoccurtransientlyduringrareburstsor process is in∼effic×ient for this kind of blazar (thoughit could flares (Dermeretal. 2009; Murase&Takami2009), though bemoreefficientforlow-peakedBLLacobjectsandFSRQs). accordingtoTable2itmightstillbedifficultevenforbright TheefficiencycanalsobehigherifΓislower,providedthatΓ flares from Mrk 501 and PKS 2155-304. As noticed above, is consistentwith synchrotron/SSCmodelfits and minimum Lorentzfactorestimatesinferredfrom,e.g.,γγ opacityargu- ple,forthesimplepower-lawinjectionofacceleratedparticles,largevalues ments. ofηareoftensuggested(Inoue&Takahara1996),whichmaynotbethecase forinjectionwithmultiplepower-lawsegments. Roughly, half of the pions produced by photome- 4 TABLE1 MEASUREDANDINFERREDPROPERTIESOFVHEBLAZARSANDRADIOGALAXIES ID Source z Epoch t[vsa]r δ(a) Γ(/a[)d/eθgo(ab],sb) γb′(a) [mεe(sac)2] [νesrFgνsc(am)[-120-s-101]] R′(a[c)[m10]15] B[G′(a]) [mεeC(ac)2] ν[eCrFgνCc(ma)[-120s--110]] Ref. 1 CenA(core) 0.00183 2009 ≤1.0×105 1.0-3.9 2.0-7.0/15-30 (0.8- 400)×103 (0.8- 4000)×10-7 0.09-4.5 3.0- 11.0 0.02-6.2 0.17-(8.3×105) 0.025-8.5 1 2 M87 0.00436 2009 1.7×105 3.9 2.3/10 4×103 1.6×10-7 0.06 14.0 0.055 18.6 0.068 2 3 NGC1275 0.0179 Oct.2010d 8.6×104 2.3 1.8/25 960 2.4×10-3 0.9 2×103 0.05 2.9×103 0.3 3 4 NGC6251 0.024 – – 2.4 2.4/25 2×104 6.5×10-7 0.012 120 0.037 7.3 0.047 4 5 Mrk421 0.03 19March2001 1.0×103 80 80 9.3×104 0.005 7.4 3.0 0.048 8.1×104 7.0 5 6 Mrk501(h.(c),1997) 0.0337 16April1997 7×103 14-20 14-20 (7- 300)×104 0.3-0.5 8.0-8.5 1.0- 5.0 0.15-0.8 (1.4- 2.6)×106 2.9-3.4 6,7,8 7 Mrk501(l.(c),1997) 0.0337 7April1997 – 15 15 6×105 0.002 0.63 5.0 0.8 4.4×105 0.4 6 8 Mrk501(l.(c),2007) 0.0337 2007 – 25 25 1×105 0.002 0.63 1.0 0.31 4.4×105 0.4 9 9 Mrk501(l.(c),2009) 0.0337 2009 3.5×105 12-25 12-25 (6- 90)×104 0.002 0.55-0.63 1.0- 130 0.015-0.34 (1.3- 4.4)×105 0.3-0.4 7,10,11 10 1ES1959+650(h.(c)) 0.047 Sept2001-May2002 (2.2- 7.2)×104 18-20 14-20 4- 5×104 (0.07- 8)×10-3 1.0-3 5.8-9 0.04-0.9 8×105-6 0.2-2 12,13 11 1ES1959+650(l.(c)) 0.047 23-25May2006 8.64×104 18 18 5.7×104 0.003 2.6 7.3 0.25-0.4 1.2×105 0.22 14,15 12 PKS2200+420/BLLac 0.069 – – 15 15 900.0 5.3×10-7 0.76 2.0 1.4 1.6 0.4 14 13 PKS2005-489 0.071 – – 22 22 1.3×104 4.7×10-5 1.5 8.0 0.7 3.6×103 0.07 14 14 WComae 0.102 7-8June2008 5400 20-25 20-25 (1.5- 20)×104 8.0×10-5 0.4 3.0 0.24-0.3 8.1×103 0.15 14,16 15 PKS2155-304 0.116 28-30July2006 300 110 110 4.3×104 4×10-4 2.13 0.86 0.1 9.7×105 20.0 5 (a):parametervaluefromtheSEDmodelinginliterature(seereferences);(b):forblazarsourcesδ Γandθj 1/Γ;(c):high(h.)andlow(l.)state;References: 1-Abdoetal.(2010d)(seeFigure5andTable2in thepaperforthedifferentmodels),2-Abdoetal. (2009c),3-Abdoetal. (2009b)(seealsoBrown≈&Adams2≈011),4-Migliorietal. (2011),5-Finkeetal. (2008),6-Pianetal. (1998),7-Acciarietal. (2011),8- Katarzynskietal. (2001),9-Albertetal. (2007b),10-Anderhubetal. (2009),11-Abdoetal. (2011c),12-Tagliaferrietal. (2003),13-Krawczynskietal. (2004),14-Tavecchioetal. (2010),15-Tagliaferrietal. (2008),16-Acciarietal.(2009b). M u r a s e e t a l. BLLacObjectsasUHECRSources 5 TABLE2 INFERREDPROPERTIESOFVHEBLAZARSANDRADIOGALAXIES ID Source [MdpLc] [erLgsγc[m10-245s]-1] [erLgCγc[m10-245s]-1] AC γb′(e) δT(b) δ(c) R′([dc)[m10]15] B[′G(a]) EAmax([f)e/VZ][1019] 1 CenA(core) 3.7 (0.15- 7.3)×10-4 (0.04- 14)×10-4 0.3-1.9 890-2.1×104 9.9-(6×10-4) 0.12-3.7 3.0-12 0.02-9.1 0.01-4 2 M87 16.7 2.0×10-4 2.3×10-4 1.1 9.3×103 0.006 2.7 20 0.021 0.05 3 NGC1275 75.3 0.06 0.02 0.35 960 0.005 – – – 5 4 NGC6251 104 2×10-3 6.6×10-3 3.3 2.9×103 0.007 – – – 0.3 5 Mrk421 130.0 1.5 1.4 0.95 3.4×103 74 – – – 0.3 6 Mrk501(h.(g),1997) 146.0 2.0-2.2 0.7-0.9 0.36-0.41 (1.4-2.5)×103 (3.0-3.5)×103 – – – 0.1-2 7 Mrk501(l.(g),1997) 146.0 0.2-0.4 0.1-0.2 0.44-0.63 (0.08-1.3)×104 100-1700 – – – 2 8 Mrk501(l.(g),2007) 146.0 0.2 0.1 0.63 1.3×104 100 – – – 0.2 9 Mrk501(l.(g),2009) 146.0 0.1-0.2 0.08-0.1 0.55-0.63 (0.7- 1.3)×104 58-100 – – – 0.2-0.7 10 1ES1959+650(h.(g)) 206 0.5-1.5 0.1-1.1 0.2-0.8 (2.7- 9.5)×104 27-910 – – – 0.1-3 11 1ES1959+650(l.(g)) 206 1.3 0.1 0.08 6600 66 – – – 1-2 12 PKS2200+420/BLLac 307.0 0.8 0.45 0.53 2.8×103 0.006 – – – 1 13 PKS2005-489 316.0 1.8 0.07 0.04 7.6×103 1.5 – – – 4 14 WComae 464.0 1.0 0.38 0.38 8.7×103 3.1 7.2 3.0-3.7 2.1-2.6 0.4-0.7 15 PKS2155-304 533.0 7.2 68 9.4 1.3×104 24 – – – 0.3 γ(ab′):≈ob√ta3i/n2edpfεrCom/εse;q(ufa):tioobnta(i2n)e;d(bf)r:omδTeq=u2at√io3np(3ε)Cuεssi/nmg2eBc′4,(Γ1,+azn)d;R(c′):reopbotratiendedinfTroamblee1q;u(agt)i:ohnig(1h)(;h(.d)):ancdallcouwla(tle.)dsatastseu.mingR′≈ctv′ar =cδtvar/(1+z); (e): son production are charged, and each neutrino car- nosity is L & LC = A Ls. In the proton synchrotron UHECR γ C γ ries 1/4 of the pion energy, so the total (isotropic- blazar model (Aharonian2000; Mücke&Protheroe2001; ∼ equivalent) neutrino luminosity at given E 0.05E Mückeetal. 2003),whichistypicallyviableforhigh-peaked ν p ≈ is estimated to be ELν (3/8)f (E )ELCR 3.8 BLLacobjects,protonsynchrotronradiationisemittedupto i1n0g42fepγrg.s-11 lfipkγe,-B2(LELLCEaRc/Eo1b0∼j4e5cetsr,gws-h1ep)rγefofrppγa =soE1u0rc- e2≃fspaγt,-is2fy×is- FeenneeermrrggiyieasicscεdeMsleet≈rear4tmiδoi1nn/es(dc1eb+nyza)rtihToeesV.syT(nhincehtprhohetorltoiomnmicteotshooalnitnptghr)oeidmnucaetxfifiiomcniueenmf-t used. Then, one finds that the neutrino flux from an indi- ficiencyfor protonsis stronglydependenton δ, but the con- vidual source is typically too low to be detected with Ice- dition L Ls suggests that the synchrotron energy-lossis Cube. Onecanalsosee thatthecumulativebackgroundflux B ≫ γ dominantatultra-highenergieswheretheprotonsynchrotron from high-peaked BL Lac objects is low. The UHECR en- radiation is typically prominent at TeV energies (though ergyinputinthelocaluniverseis 5 1043 ergMpc- 3 yr- 1 ∼ the photohadroniccascade componentmay become relevant at 1019 eV (Murase&Takami200∼9),×so that assuming that atlowerenergies). Thestrongmagneticfieldalsosuppresses suchBLLacobjectsandFR-IgalaxiesarethemainUHECR electronicSSC emission because fewerelectronsare needed sources,theexpectedcumulativemuonneutrinobackground togeneratethesamesynchrotronflux. fluxisestimatedtobe(Murase&Beacom2010a) E2Φ 10- 10GeVcm- 2s- 1sr- 1 fpγ(20Eν) E2- p f , 2.3. SurvivalofNucleiintheSource ν ν ∼ (cid:18) 10- 2 (cid:19) ν,17.7 z If the standard synchrotron/SSC scenario holds for TeV (6) blazars and their misaligned counterparts, then the protons where p is the cosmic-ray spectral index and f is a pre- z canhardlyreach1020 eV, asshownin Table2 (seeEmax ob- factorcomingfromtheredshiftevolutionofthesources.Low A tained by detailed modeling in the literature). For BL Lac photomeson productionefficiencies also follow if the UHE- objects and FR-I galaxies to be the steady sources of UHE- CRs are heavy nuclei, whose losses are dominated by pho- CRs, therefore,UHECRs wouldprimarilybeheaviernuclei. todisintegration(seebelow). Moreluminousblazars,includ- In such a scenario, one has to examine whether ions can inglow-peakedBL Lacobjects,mayhoweverleadtohigher survive photodisintegration losses (cf. Muraseetal. 2008a; photomeson production efficiencies, so that the cumulative Wangetal. 2008;Pe’eretal. 2009). Thephotodisintegration neutrino background could be dominated by this class of opacity is estimated similarly to the photomesonproduction AGN(Mückeetal. 2003). efficiency. Approximatingthe photodisintegrationcrosssec- Our evaluation is based on the standard synchrotron/SSC tion by the giant dipole resonance (GDR) cross section as model for BL Lac objects and FR-I radio galaxies. One σ σ δ(ε- ε¯ )∆ε¯ , for a sufficiently soft photon couldabandonthestandardsynchrotron/SSCmodelandcon- Aγ ∼ GDR GDR GDR sider a highly magnetized, 10- 100 G jet model, which spectrum(withα&1),weget(Murase&Beacom2010a,see ∼ alsoMuraseetal. 2008aforthenon-GDReffect) is needed in hadronic blazar models to accelerate protons to &1020 eV (Aharonian2000). Correspondingly, the min- t′ 2σ ∆ε¯ Ls E α- 1 i1m04u7mergmsa-g1neΓt21ic(Elpmu,2amx0)i2n,owsihtyichisiselsatrimgeartethdantothbeetyLpBica≈l s2yn×- τAγ ≈ tAv′aγr ≃ 1+GDαR ε¯GGDDRR4πδ4tvγarc2εs(cid:18)EAAb(cid:19) , (7) chrotronluminosityofBLLacobjects,Ls 1046 ergs- 1. In the hadronic models, γ-ray emission is aγtt∼ributed to proton wheretA′γ isthephotodisintegrationinteractiontime,σGDR≈ synchrotron radiation and/or proton-induced cascade emis- 1.45 10- 27A cm2, ε¯GDR 42.65A- 0.21 MeV (for A > 4), sion, which leads to the requirementthat the UHECR lumi- ∆ε¯ × 8 MeV, and Eb ≈0.5δ2(m c2ε¯ /εs)(1+z)- 2 (in GDR ∼ A ≈ A GDR 6 Muraseetal. theobserverframe).Thenwenumericallyfind flow.Therefore,thehighlyvariableVHEradiationfromthese 2.5 ε - 1 E α- 1 BL Lac objects is likely to be either leptonic synchrotronor τ (E ) 0.16 Ls t- 1 δ- 4 s A , SSC,orprotonsynchrotronradiationproducedinthejet(even Aγ A ≃ (cid:18)1+α(cid:19) γ,46var,4 1 (cid:16)1keV(cid:17) (cid:18)EAb(cid:19) thoughanotherslowlyvariablecomponentmaybeproduced (8) bythesecondaryemission). where EAb ≃4.8×1016 eV (A/56)0.79(εs/1keV)- 1δ12(1+z)- 2 This conclusion does not however hold for a fraction of istheenergyofanucleusthattypicallyinteractswithapho- blazars and radio galaxies from which prominent variabil- tonwithε . Hence,heavynucleiwithE (Z/26)1020.5 eV ity has not been seen. An interesting source is the ex- s A ∼ (given in the observer frame) undergo some photodis- treme TeV blazar 1ES 0229+200, which has a hard VHE integration reactions unless δ is high enough. The component extending to > 10 TeV, but has not been re- nucleus survival condition τ (E ) . 1 gives δ & ported to be variable in observations taken over a period Aγ A 17(Z/26)0.1(A/56)- 0.079(Ls )1/5t- 1/5(ε /1keV)- 0.1(1+z)- 1/5 of 3- 4 yr (Aharonianetal. 2007b; Perkinsetal. 2010). If γ,46 var,4 s the apparent absence of the variability comes from observa- (forα 1.5),butsignificantphotodisintegrationlossiseasily ∼ tional limitations and if fast variability is present, the emis- avoidedforreasonablylargebulkoutflowDopplerfactors. sion should be produced in/near the blazar region. If it is Recalling from equation (5) that the photomeson produc- the case that there is no rapid variability, the observedcom- tion efficiency has the same dependence on δ, we can con- ponentmaycomefromanextendedjet(Böttcheretal. 2008, clude that when heavy nuclei survive photodisintegration, see also Section 4 for further discussion on the γ-ray emis- the photomeson production efficiency is so low that the sion region). In addition, as we see in this paper, a corresponding neutrino and γ-ray fluxes are not easily de- slowly variable componentcan be γ-ray-inducedintergalac- tected(Murase&Beacom2010a,butseealsoMurase&Bea- tic cascade emission. Furthermore, if it is non-variable, com2010b). proton-induced intergalactic cascade emission (i.e., inter- 3. EXTREMETEVBLAZARSANDINTERGALACTICCASCADES galactic cascades caused by UHE γ rays and pairs gen- Theγγ opacityargumentallowsustoplaceconstraintson erated via the photomeson production with the CMB and δ in BL Lac objects observed at TeV energies, and requires EBL) can be responsible for the observed emission. These δ&60forPKS2155-304(Begelmanetal. 2008)forthema- cascade emissions may confuse interpretation of the mini- jorJuly/August2006TeVflares(Aharonianetal. 2007a),and mumbulkLorentzfactorfromγγ opacityargumentsandthe δ &100 to be furthermore consistent with synchrotron/SSC levelof the EBL (Essey&Kusenko2010; Esseyetal. 2010; model fitting for different models of the extragalactic back- Esseyetal. 2011). The intergalactic cascade components groundlight(EBL;Finkeetal. 2008).Anotherimportantfact could be present not only in extreme blazars, but also on isthatVHEphotonscaninteractwiththecosmicphotonback- longertimescalesinotherblazarsandradiogalaxies. Forex- grounds. VHE γ rays produce electron-positron pairs via ample,aslowlyvariable( month)emissionatGeVenergies ∼ γγ pair creation, and the resulting high-energy pairs make wasobservedfromMrk501(Abdoetal. 2011c),whichcould high-energy photons via Compton scattering. Hence, the arisefromtheintergalacticcascadeinducedbyvariableTeV cascaded γ rays, which are often called pair echoes (e.g., sourcephotons(Neronovetal. 2011). Plaga1995; Muraseetal. 2008b) and/or pair haloes (e.g., Here we focus on the intergalactic cascade scenarios in Aharonianetal. 1994; Neronov&Semikoz2007), are ex- order to explain hard VHE spectra of extreme TeV blazars pected at GeV-TeV energies. In particular, γ rays with en- whosevariabilityisapparentlyabsent. We calculatethecas- ergies below 100 TeV are likely to leave structured re- cadeemissionbysolvingtheBoltzmannequations,whereγγ gions of the u∼niverse, inducing the cascade in the void re- pair creation, IC scattering, synchrotronradiation, and adia- gion(Muraseetal. 2008b). batic energy loss are taken into account (Lee1998). As for Many γ-ray blazars show variability, and often dis- protonpropagation,wedirectlysolvetheequationofmotion play spectacular flares. Some of them are ultra-variable, of protons one by one, with photomeson production simu- as seen in multi-TeV flaring episodes from PKS 2155- lated by SOPHIA (Mückeetal. 2000) and the Bethe-Heitler 304 (Aharonianetal. 2007a; Aharonianetal. 2009b), Mrk process included to treat interactions with the ambient pho- 501 (Albertetal. 2007b) and Mrk 421 (Galanteetal. 2011). ton field (Chodorowskietal. 1992). Then the electromag- Such rapidly varying γ-ray emission should be produced in neticcascadeiscalculatedseparately.FortheEBLmodel,we theblazarregion. ThisisbecausetheIGMFwillintroducea employ the low-IR and best-fit models (Kneiskeetal. 2004; significanttimespreadinthecascaderadiation,whichseems Kneiske&Dole2010, seeFinkeetal. 2011fordetaileddis- incompatible with rapidly varying emissions. For example, cussions on the EBL). We focus on the possibility that the considerprimary10TeVγ rays,sotheCompton-upscattered cascade interpretation is a viable explanation of VHE γ-ray cosmic microwave background (CMB) photons have E spectra of extreme TeV blazars. Here, the IGMF in voids γ o(4f/B3)γ′2λε1C/M2B&≃1808- 18G- eV10-γ17′27.G BMapsecd1/2on(Dtohleagloewtaelr. 2li0m1i≈1ts; hthaesctoascbaedeweraadkiaetnioonugaht T(BeVIGVeλn1ceo/rh2g.ies1i0s- 1n5otGsuMppprce1s/s2e)dthbayt IGV coh the IGMF if it is to make the measured flux in the VHE Dermeretal. 2011; Takahashietal. 2012) obtained for 1ES range. On the other hand, the void IGMF cannot be be- 0229+200 (where the term λ is the coherence length of the magnetic field, and this rceolhation is understood to apply low 10- 18 G Mpc1/2 due to constraints from Fermi (e.g., ∼ whenλ issmallerthanthecoolinglengthforGeVproduc- Dolagetal. 2011; Dermeretal. 2011; Tavecchioetal. 2011; coh tion),thetimescaleofthe 0.1TeVpairechoisestimatedto Tayloretal. 2011;Ahlers&Salvado2011). be(Muraseetal. 2008b;D∼ermeretal. 2011), An important point is that cosmic magnetic fields are al- ∆tIGV≃1.4yrEγ- 2,11B2IGV,- 17(λγγ/100Mpc)(1+z)- 1, (9) mweoasktcIeGrtMaiFnlsyininvhooimdso,gBeneoλu1s/.2Whe1r0e-a9sGonMempca1y/2e,xtpheecsttvruercy- whereλγγ isthemeanfreepathfortheγγ paircreation,and turedregionoftheuniveIrGsVeicsohlik≪elytobesignificantlymag- the void IGMF B is defined in the frame of the Hubble IGV BLLacObjectsasUHECRSources 7 netized. Clusters ofgalaxiesare knownto haveB 0.1- EG -7 ∼ 1 µG (e.g., Vallée2004), and recent simulations have sug- z=0.05 gestedthatfilamentshaveB 1- 10nG(Ryuetal. 2008; -8 z=0.1 aDraesleatrgael.r2t0h0an8,laenvdelsseeexaplseoc,EteeG.dg∼.f,oDrotnhneeIrGteMtaFl.i2n0v0o9i)d,sw,haincdh -1s]) -9 zzz===000...357 2 galaxiesincludingAGNarelikelylocatedinthesestructured -m -10 c regions of the universe. The mean free path of .100 TeV V and&3EeVγraysislargerthan Mpc(e.g.,Dermer2007), Ge -11 sothatonemayexpectthattheca∼scadeemissioninducedby F [E-12 VHE/UHEprimaryγraysisprimarilydevelopedinthevoids. 2 E Ontheotherhand,ionsmustpropagateintheclustersand/or g( -13 o filaments, so that they are deflected (and delayed) by their l -14 magneticfields. Indeed,asdemonstratedbyanumberofau- thors, the structuredEGMFs play a crucialrole on propaga- -15 tion of UHECRs (e.g., Takamietal. 2006; Dasetal. 2008), 2 2.5 3 3.5 4 4.5 5 5.5 6 and this is even more so the case for lower-energy cosmic log(E [GeV]) rays. FIG. 1.—Spectra of VHE γ-ray-induced cascade emission for various sourceredshifts. Weassumethetotalγ-rayluminosityofLγ=1045ergs- 1 3.1. CascadesbyPrimaryVHE/UHEGammaRays withβ=2/3andEmax=100TeV.Thelow-IREBLmodelofKneiskeetal. (2004)isusedhere. SEDs of high-peaked BL Lac objects are generally well reproduced by the standard one-zone electronic syn- chhavroetrtohne/ShSaCrdemstoVdeHl.E γA-mraoyngspethcetrma,atextre1m- e10TeTVeVblaeznaerrs- -8 bb ==52//43 ww.. EEmmaaxx==110022 TTeeVV gies,asindicatedbydeabsorptionofthem∼easuredγ-rayspec- b =2/3 w. Emax=101.5 TeV trum based on conventional EBL models (discussed below) -1s]) -9 b =5/4 w. Emax=101.5 TeV and supported by non-detections of GeV γ rays by Fermi. -2m Also, in some cases (RGB J0152+017, 1ES 0229+200 and c V 1ES0548-322),theoptical/UVdatashowarathersteepspec- e -10 G trumwhichisthoughttobetheemissionfromthehostgalax- [E ies(Tavecchioetal. 2011).Althoughthesynchrotroncompo- F 2 nentof extreme TeV blazarsseem unremarkableat the opti- E g( -11 cal/UVband,inthesecases,comparisonbetweenoptical/UV o l and X-ray data requires a strong roll-off of the nonthermal spectrumbelowtheX-rayband,suggestingthatFs ν1/3for ν∝ -12 1ES0229+200(Tavecchioetal. 2011). 2 2.5 3 3.5 4 4.5 5 5.5 6 Itispossibletoexplainsuchhardγ-rayspectrabytheSSC log(E [GeV]) model, but extreme parametersseem necessary comparedto FIG. 2.—Spectra of VHE γ-ray-induced cascade emission for various casesoftypical,variablehigh-peakedBL Lacobjects. Itof- intrinsicphotonspectra.Thesourceredshiftissettoz=0.14. tensuggestsaverynarrow-rangeenergydistributionofelec- trons, and unusually large values of δ 102- 103 may be ∼ necessary to avoid the Klein-Nishina suppression. For 1ES 0229+200,PKS0548-322and1ES0347-121,extremevalues for recollimation shocks (e.g., Bromberg&Levinson2009; oftheelectronminimumLorentzfactorofγe,m 104- 105are Nalewajko&Sikora2009), or acceleration at knots and ∼ required from spectral modeling, where the hard SSC spec- hotspots,notingthatvariationsonmuchlongertimescalescan trum,FE E- 2/3(intheThomsonregime)canbeexpectedin beexpectedinthesemodels. ∝ theVHErange(Tavecchioetal. 2011). When VHE γ rays are emitted from a source, they in- Thereare severalalternate blazarmodelsthatpredictvery duceanelectromagneticcascadein intergalacticspace. This hard 30- 100 TeV γ-ray emission. In the hadronic cascadeunavoidablyaccompaniesspectralproductionofex- ∼ model, the proton synchrotron process leads to multi- treme TeV blazars as long as the IGMF in voids is weak TeV emission if the outflow is ultra-relativistic, Γ 102- enough. To demonstrate this, we show in Figures 1 and 2 103. Then, further hardening may be caused b∼y inter- theVHE γ-ray-inducedcascadeemission forsourcesatvar- nal absorption due to some soft photon field outside the ious redshifts. In Figure 1, primary source photons with blob (Zacharopoulouetal. 2011). Another possibility to F E- β with β =2/3 and Emax =100 TeV (in the source E ∝ make hard TeV emission is electromagnetic radiation pro- rest frame) are assumed. One sees that the observed cutoff ducedbynonthermalelectronsinthevacuumgapoftheblack duetotheEBLbecomeslowerformoredistantsourcessince holemagnetosphere(e.g.,Levinson2000). theγγpair-creationopacityincreases.InFigure2,adifferent Böttcher et al. (2008) suggested that hard VHE emission photon index (β =5/4) and/or a different maximum energy originates from CMB photons Compton-upscattered by rel- (Emax=101.5TeV)areassumedforcomparison,whichcauses ativistic electrons that are accelerated in the extended jet. slightdifferencesinspectra.Asindicatedbyequation(9),the In this model, if the electron spectrum is hard, p 1.5, intergalacticcascadeemissioninducedbyprimaryγrayswill ∼ the resulting number spectrum of the IC emission has FE beslowlyvariableoralmoststeady. E- (1+p)/2 E- 5/4, which is compatible with the observe∝d Anotherwaytohaveγ-rayinducedemissioninvolvesUHE ∼ VHE γ-rayspectrum. The same processcould be important γ rays produced in blazar jets or radio galaxies. Such a 8 Muraseetal. matedtobe(Takami&Murase2011) -7 z=0.05 -8 z=0.1 √2λ l λ 1/2 l 1/2 2-1 s]) -9 zzz===000...357 θCR≈ 3rLcoh ≃8◦ZEA- ,119BEG,- 8(cid:18)0.1cMohpc(cid:19) (cid:18)Mpc(cid:19)(10). -m -10 Therefore,thedeflectionbythestructuredEGMFsisnotneg- c V ligibleforcosmicrayswithenergies.1019 eV,sincethede- e -11 G flection angle is larger than the typical jet opening angle of F [E-12 θj 0.1 6◦. Thecorrespondingtimespreadduetoastruc- 2E tur∼edEG∼MFaroundthesource(thatiscomparabletothetime g( -13 delay)isexpectedtobe o l -14 ∆t 1 l CR θ2 -15 1+z ≈4 CRc 2 2.5 3 3.5 4 4.5 5 5.5 6 log(E [GeV]) ≃2×104yrZ2EA- ,219B2EG,- 8(cid:18)0.1λcMohpc(cid:19)(cid:18)Mlpc(cid:19)2(1,1) FIG. 3.— Spectra of UHE γ-ray-induced cascade emission for various sourceredshifts.WeassumeLγ=1045ergs- 1at1018.75- 1019.25eV. whichisunavoidableaslongascosmicrayspassthroughthe structured region around the source, and it implies that the resulting cascade emission is essentially regarded as steady emission. Noticethatthetotaltimespread∆T couldgen- CR erally be longer than ∆t due to additionaltime spread by CR interveningstructuredEGMFsandthevoidIGMF. case is shown in Figure 3 assuming much higher injected In order to model the structured EGMFs, we have as- photon energies than before. Here we assume an injection sumed a simplified two-zone model with structured EGMF spectrum centered at 10 EeV spanning one decade. The andIGMFinvoids(seeTakami&Murase2011,fordetails). photomeson production by UHE protons, which can be ex- We model a cluster of galaxies by a sphere with the radius pectedinhadronicmodels,leadstoUHEphotonswithenergy of 3 Mpc, and B (r)=B (1+r/r )- 0.7, with B =1 µG and EG 0 c 0 Eγ ≈0.1Ep ≃1019 eV Ep,20. In the synchrotron source in rc =378 kpc. The magnetic field direction is assumed to be whichUHEprotonsareaccelerated,onemayexpectthatthe turbulent with the Kolmogorov spectrum and the maximum synchrotronself-absorptioncutoffcurtailsthenumberoflow length of λ = 100 kpc. In addition to the EBL, the in- max energyphotonsimpedingUHEphotonescapefromtheemis- frared background in the cluster is considered as the super- sionregion(Murase2009). Acaveatofthismodelinourcase position of the SEDs of 100 giant elliptical galaxies calcu- isthatgenerationofUHEγ raysinthesourcerequiresaccel- latedbyGRASIL(Silvaetal. 1998),usingfittingformulafor erationofUHEprotonsandmoderatelyefficientphotomeson thegasdistribution(Rordorfetal. 2004). Filamentsaremod- production. As noted before, the photomeson productionin eled by a cylinder with a radius of 2 Mpc (Ryuetal. 2008) the source may not be too efficient in high-peaked BL Lac and a height of 25 Mpc. The magnetic field is assumed to objects, which implies that the required UHECR luminosity beturbulent,whichisdescribedbytheKolmogorovspectrum has to be very large. As can be seen, there is some notable with B = 10 nG and λ =100 kpc, although these val- EG max differencesatlowredshiftsz.0.1duetothelongereffective uesare veryuncertain. Somenumericalsimulationsimplya energylosslengthofUHEγrays,butthereceivedspectraare large-scalecoherentcomponentof the magneticfield in fila- notstronglysensitive to theenergyatwhichthe photonsare ments(e.g.,Brüggenetal. 2005),whichmaydeflectcosmic- injectedforhigherredshiftsources. raytrajectoriesevenmoreeffectively. UHECRs are injected Note that the intergalactic cascade scenario makes a non- fromthecenterofthefilamenttowardadirectionperpendic- variable or slowly variable component, even when the γ- ular to the cylindrical axis in order to examine a relatively ray emission made in the jet contributes to a separate conservativecase. Throughoutthiswork,theIGMFinvoids highly variable component. Although there is no strong ev- isassumedtobeweakenoughtobelessimportantforcosmic- idence of time variability for several extreme TeV blazars, raydeflections. future sensitive observations by CTA (Actisetal. 2011), In Figure 4, we show our numerical results for the case HAWC (Sandovaletal. 2009), LHAASO (Cao2010), or Emax = 1019 eV expected in the standard synchrotron/SSC SCORE (Hampfetal. 2011) willbe crucialfor identifyinga p model of typical, variable BL Lac objects and FR-I galax- slowlyvariableγ-rayemissioncomponent. ies. One sees that the structured EGMFs play an important rolebysuppressingtheresultingγ-rayfluxbymorethanone 3.2. CascadesbyPrimaryUHECRs order of magnitude compared to the case without them. In Intheprevioussubsection,weconsideredcascadeemission Figure 5, we show the case of Emax = 1020 eV, which can p inducedbyprimaryγrays.VHEγraysat.100TeV,orUHE be achievedin the hadronicmodel. While the Bethe-Heitler γ rayswithenergies&3EeV,wheretheopacityoftheback- pair-creationprocessdominantlyprovidesanelectromagnetic groundradiationisnotso large,canleavestructuredregions componentin Figure 4, contributionof photomesonproduc- of the universe, whereas cosmic rays should feel structured tion is more importantin Figure 5. In the filamentcase, the EGMFs in clusters and filaments. The deflection of cosmic deflectionangleofUHECRsaround1020 eVisstilllessthan rays by the structured region with size l Mpc, magnetic thejetopeningangle,sothattheγ-rayfluxisdilutedbyonly ∼ fieldofB 10nG,andcoherencelengthofλ 0.1Mpc a small factor. On the other hand, in the cluster case, be- EG coh ∼ ∼ (whichmaybe typicaloffilaments; Ryuetal. 2008)isesti- cause UHECRs cannot be beamed, the γ-ray flux becomes BLLacObjectsasUHECRSources 9 101 no Structured EGMF Cluster -10 Filament Filament Cluster Jet 2-1 s])-12 utions 100 -eV cm-14 Contrib 10-1 [GE ve 2g(E F-16 Relati 10-2 Isotropic o l -18 10-3 1018 1019 1020 1021 1 2 3 4 5 6 7 log(E [GeV]) Ep [eV] FIG. 4.—EffectsofthestructuredEGMFontheγ-rayflux. Weassume FIG. 6.— Effects of the structured EGMFs on the deflection of UHE LUHECR =1045 erg s- 1, with Epmax =1019 eV and p=2. Here, as in the protons. Relative contributions represent how much the apparent cosmic- resultsoncascadeemissioninducedbyprimaryγrays,weusetheisotropic- luminosityatwhichcosmicraysenterthevoidregionisdilutedfromELCR E equivalentcosmic-rayluminosityatthesource(definedforUHECRsabove atthesource.Notethatatwo-sidedjetisconsideredthroughoutthiswork. 1018.5eV),whichisrelatedtotheabsolute(beaming-corrected) cosmic-ray jleutmoipneonsiintyg,LanUgHlEeCiRs,θj,ja=sL0U.1H.ETChRe≡so(u1r-cecorsedθsjh)-if1tLiUsHsEeCtRto,j.zH=e0r.e5t.heassumed -7 z=0.05 -8 z=0.1 z=0.3 no Structured EGMF -1s]) -9 zz==00..57 -10 Filament 2 Cluster -m -10 1]) V c 2- s -12 Ge -11 -cm F [E-12 V 2 e -14 E G g( -13 [E lo 2E F -16 -14 g( -15 o l 2 2.5 3 3.5 4 4.5 5 5.5 6 -18 log(E [GeV]) FIG. 7.—Spectra ofUHEproton-induced cascade emissionforvarious 1 2 3 4 5 6 7 sourceredshifts.WeassumeLUHECR=1045ergs- 1withEpmax=1019eVand log(E [GeV]) p=2.ThesourceisassumedtobelocatedinthefilamentwithBEG=10nG FIG.5.—SameasFigure4,butwithEpmax=1020eV. andλmax=0.1Mpc.Thelow-IREBLmodelishereassumed. with Figures 1 and 3). Importantly for distant sources, the proton-inducedcascadespectrumismuchharderthantheγ- almost isotropic and the corresponding flux is reduced ac- rayinducedspectrum,especiallyaboveTeVenergies. Future cording to the jet beaming factor (1- cosθ ) 1/200 for j VHEobservationsbyCTAandHAWCareimportanttoiden- ≃ θ =0.1. TheeffectsofthestructuredEGMFsareillustrated j tify the originof UHECRs throughdetectionof high-energy in Figure 6, where the relative contributions are calculated γ rays,aswenowdemonstratefor1ES0229+200inthenext fromtwo-dimensionalGaussianfits. Notethatifweexpress subsection. theisotropic-equivalentcosmic-rayluminositywherecosmic In this work, we are interested in cases where IC cascade rays leave the structured region as ELCR, then the relative E emission in voids is important in the VHE range, since it contributionsare(1- cosθj)(ELCER)/(ELCER,j). Inthefilament canexplainhardVHEspectraofextremeTeVblazarsassug- case, isotropization becomes significant at 1019 eV rather gestedbyEsseyetal.(2010).Whenpairsaremainlysupplied thanat 1021eVfortheclustercase. ∼ viatheBethe-Heitlerprocess,thetimescaleofsecondarypho- In Fi∼gure 7, we show resulting γ-ray spectra for various tonsproducedbyaprotonbeamroughlybecomes rloedssshleifntgs.thOwGinpgc,toUHthEepBreotthoen-sHceointlteirnuperotocessuspwpliytheleencetrrogny-- ∆tIGV≃14yrEγ- 2,11B2IGV,- 17(λBH/Gpc)(1+z)- 1, (12) positronpai∼rsfora longerdistance thanthe photomesonen- which is more relevant than ∆T when the void IGMF CR ergyloss lengthof 100 Mpc. As a result, the dependence is so strong that ∆T < ∆t is satisfied. Here, CR IGV ∼ of the proton-inducedγ-ray fluxeson distance is much gen- λ is the Bethe-Heitler energy loss length. One BH tler than γ-ray-inducedfluxes. Indeed, one sees thatthe rel- should also keep in mind that the proton-induced GeV- ative importance of the proton-induced γ-ray flux to the γ- TeV synchrotron emission from the structured region it- ray induced flux increases with distance (compare Figure 7 self, where the EGMFs are stronger, should also be ex- 10 Muraseetal. pected (see Gabici&Aharonian2005; Koteraetal. 2009; 10-11 Koteraetal. 2011,andreferencestherein).ForaweakIGMF 1ES 0229+200 thatisofinterestinthiswork,itsrelativeimportanceissome- whatsmaller whenthevolumefillingfractionof themagne- tizedregionistakenintoaccount. We have demonstrated the likely importance of the struc- turedEGMFsforproton-inducedintergalacticcascadeemis- -1s) 10-12 2 sion. They are also importantfor UHE nuclei. Since nuclei -m with energy ZE have the same deflection angle as protons c p g with energyEp, ourresultsindicatethatFe nucleishouldbe er significantly isotropized for all observed UHECR energies. (ν F For UHE nuclei, the photodisintegration energy loss length ν HESS is 100 Mpc, for which the energy fraction carried by γ 10-13 VERITAS ray∼s and neutrinos is small as long as Emax is not too high. E20,low IR A Ontheotherhand,UHE nucleisupplyhigh-energypairsvia E19,low IR E19, best fit the Bethe-Heitler process, whose effective cross section is E14, low IR κBH,AσBH,A κBH,pσBH,p(Z2/A), which induces cascades in 5 hr CTA sensitivity ∼ thesamemannerasUHEprotons. Therefore,theintergalac- 50 hr tic cascade signal, which is generated outside the source, is 10-14 alsoimportantforsourcesofprimaryUHEnuclei.9 102 103 104 105 E (GeV) 3.3. ImplicationsforTeV-PeVObservations InawiderangeofEBLmodels,deabsorptionofmeasured FIG.8.—SpectralfitstoHESSandVERITASdataof1ES0229+200.Blue datapointsarefromHESS(Aharonianetal.2007b),andreddatapointsare TeVblazarspectraleadstohardexcessesat>TeVenergiesin, preliminaryVERITASdata(Perkinsetal.2010). Thecurveslabeled“E20, e.g.,1ES1101-232,1ES0229+200,and1ES0347-121(see, lowIR"and“E19,lowIR"arethecascadespectrainitiatedbytheE- 2 in- e.g.,Fig.8inFinkeetal. 2010). TheseunusualTeVspectral jection with Emax =1020 eV and 1019 eV protons, respectively, using the p emissioncomponentsareconventionallyexplainedby(either low-IREBLmodel(Kneiskeetal.2004), whereas thecurvelabeled “E19, leptonicorhadronic)emissionsatthe source,buttheycould bestfit"isthespectrumwithEpmax=1019eVforthebest-fitEBLmodel.The also be explained by intergalactic cascade emissions. Non- curvelabeled “E14, lowIR"isthespectrumresulting fromthecascadeof Emax=1014eVphotonswithβ=5/4producedatthesourceforthelow-IR simultaneousTeV excesses are also seen above the extrapo- EBLmodel. Doubledot-dashedanddottedcurvesgive,respectively,the5σ lationoftheGeVfluxinNGC1275(Abdoetal. 2009b)and differentialsensitivityfor5and50hrobservationswithCTA(configuration the core of Cen A (Abdoetal. 2010d), but because of their E;Actisetal.2011). proximity,theseexcessesareunlikelytobeUHECR-induced emissionsmadeinintergalacticspace. Figure8demonstratesthat1ES0229+200canbefitbyboth theγ-rayinducedcascadeandproton-inducedcascadeemis- featurebyfutureCherenkovdetectorssuchasCTAorHAWC sions.BecauseoftheuncertaintyinEBLmodels,itisnoteasy is possible, and the differential sensitivity goal of CTA is to distinguishbetweenthe two possibilitiesat 0.1- 1TeV shown(Actisetal. 2011). Notethatthisisadifferentialsen- energies. Athigherenergies,however,ourcalc∼ulationsshow sitivity curvewith the requirementof 5σ significance for 50 thatUHECR-inducedcascadeemissionbecomesharderthan hourobservationsper bin, with 4 binsper decade. This is a γ-ray-inducedcascadeemissionresultingfromattenuationof much more stringent requirementthan detection of a source hard γ-ray source photons, for a given EBL model. More with5σbasedonintegratedflux,whichcanbedividedinto3 importantly, the emission spectrum measured as a result of datapointswith 3σsignificanceeach.Giventhedifferential ≈ the injection of VHE/UHE photonsat the source is strongly CTAsensitivityfora50hrobservation,thespectralhardening suppressed above 10 TeV for a wide range of EBL mod- associatedwithhadroniccascadedevelopmentcanbeclearly els, whereasa cosm∼ic-ray-inducedcascadedisplaysasignif- detected. icantly harder spectrum above this energy, and detection of It is theoretically expected that cosmic-ray-induced and > 25 TeV γ rays from 1ES 0229+200 is only compatible γ-ray-induced cascade emissions are more easily discrim- if the γ rays are hadronic in origin. This is because UHE inated in higher redshift sources. For the γ-ray-induced protons (and UHE nuclei) can inject high-energy pairs over cascade, there should be a cutoff because of γγ pair cre- the Bethe-Heitler energy loss length (λ (A/Z2) Gpc at ation by the EBL, while spectra of the cosmic-ray-induced BH E A1019eV)thatistypicallylongerthan∼theeffectiveloss cascade emission are hardened by the continuous injection A ∼ throughtheBethe-Heitlerprocess. Hence,deepobservations lengthofVHE/UHEphotons. Forsteady,non-variableγ-ray at & TeV energies by CTA or HAWC for moderately high- sources,thisintergalacticcascadesignalinducedbyUHECRs redshiftblazarswillalsobeimportanttoresolvethisquestion, provides a crucial probe of UHECR sources. Its identifica- alongwithdetailedtheoreticalcalculationsforindividualTeV tion would demonstrate that a distant blazar is an UHECR blazars. sourcethroughelectromagneticchannels,whichprovidesan- Now that IceCube has been completed, it has started to otherimportantcluebesidesγ-rayvariability.Identifyingthis give important insights into the origin of UHECRs by itself 9 Ontheotherhand, theemissionofγ raysandneutrinos producedin- and with GeV and VHE γ-ray observations. But detection side the source of primary UHE nuclei is limited by the nuclear survival of neutrino signals produced outside the source seems diffi- condition, asshowninMurase&Beacom (2010a; 2010b). Giventhatthe cult for high-peaked BL Lac objects and FR-I galaxies, be- observed UHECRs are dominated by heavy nuclei, this limitation is also cause the point source flux sensitivity at >10 PeV is order applied to neutrinos produced outside the source, i.e., cosmogenic neutri- nos(Murase&Beacom2010a). of 10- 11TeVcm- 2s- 1 (Spiering2011;Abbasietal. 2011), ∼