ebook img

Thermal and non-thermal traces of AGN feedback: results from cosmological AMR simulations PDF

23 Pages·2012·4.69 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Thermal and non-thermal traces of AGN feedback: results from cosmological AMR simulations

MNRAS428,2366–2388(2013) doi:10.1093/mnras/sts213 Thermal and non-thermal traces of AGN feedback: results from cosmological AMR simulations F. Vazza,1,2,3‹ M. Bru¨ggen1,2 and C. Gheller4 1JacobsUniversityBremen,CampusRing1,D-28759Bremen,Germany 2HamburgerSternwarte,Gojenbergsweg112,D-21029Hamburg,Germany 3INAF/IstitutodiRadioastronomia,viaGobetti101,I-40129Bologna,Italy 4SwissNationalSupercomputingCentre,ViaCantonale,CH-6928Lugano,Switzerland D o w Accepted2012October12.Received2012October12;inoriginalform2012July8 n lo a d e d ABSTRACT fro m Weinvestigatetheobservableeffectsoffeedbackfromactivegalacticnuclei(AGN)onnon- h thermal components of the intracluster medium (ICM). We have modelled feedback from ttp s AGNincosmologicalsimulationswiththeadaptivemeshrefinementcodeENZO,investigating ://a c threetypesoffeedbackthataresometimescalledquasar,jetandradiomode.Usingasmall ad e set of galaxy clusters simulated at high resolution, we model the injection and evolution of m ic cosmicrays,aswellastheireffectsonthethermalplasma.Bycomparingboththeprofilesof .o u thermalgastoobservedprofilesfromtheACCEPTsampleandthesecondaryγ-rayemission p.c o to the available upper limits from Fermi, we discuss how the combined analysis of these m /m twoobservablescanconstraintheenergeticsandmechanismsoffeedbackmodelsinclusters. n ThosemodesofAGNfeedbackthatprovideagoodmatchtoX-rayobservationsyieldaγ-ray ras /a luminosity resulting from secondary cosmic rays that is about 10 times below the available rtic upperlimitsfromFermi.Moreover,weinvestigatetheinjectionofturbulentmotionsintothe le -a ICMfromAGN,andthedetectabilityofthesemotionsviatheanalysisoflinebroadeningof b s theFeXXIIIline.Inthenearfuture,deeperobservations/upperlimitsofnon-thermalemissions trac from galaxy clusters will yield stringent constraints on the energetics and modes of AGN t/4 2 feedback,evenatearlycosmicepochs. 8/3 /2 3 Key words: methods: numerical–galaxies: clusters: general–intergalactic medium–large- 6 6 scalestructureofUniverse. /1 0 7 0 9 8 5 b y centralsupermassiveblackhole(BH),tothe∼10–100kpccooling g 1 INTRODUCTION radius.Verysimilarproblemsareencounteredinellipticalgalaxies ues Radiativecoolingofgasingalaxyclustersissoefficientthatmost (e.g.Sarazin&White1987;Ciotti&Ostriker1997;Brighenti& t o n ofthehotgasphaseintheircoreoughttoberemovedonatime- Mathews2000;Ciotti&Ostriker2012).Forclusters,animportant 0 4 scalesmallerthanthelifetimeofthesystem,producinganinward issue is whether most of the energy input from AGN (or galac- M a motion of the cooling gas, a ‘cooling flow’ (e.g. Fabian, Nulsen ticwinds)totheICMhas occurred muchbeforetheassemblyof y 2 & Canizares 1984). However, dramatic cooling flows are not ob- the clusters via ‘pre-heating’ (e.g. Bialek, Evrard & Mohr 2001; 0 1 served in real clusters, and additional non-gravitational heating Brighenti & Mathews 2001; Tozzi & Norman 2001; McCarthy 9 mechanismsarelikelytokeepgasonahigheradiabat(e.g.Kaiser etal.2004),oratalowredshiftwithinalreadyformedclusters(e.g. 1991;Ponman,Cannon&Navarro1999;Lloyd-Davies,Ponman& Binney & Tabor 1995; Churazov et al. 2001; Bru¨ggen & Kaiser Cannon2000).Itislargelyagreedthatactivegalacticnuclei(AGN) 2002). The first mechanism requires a lower energy budget, with are a viable source of energy available for the self-regulation of totalenergies≤1062erg(McCarthyetal.2008),whilethesecond galaxyclusters.Observationsshowthattheenergyassociatedwith possibilityrequiresenergiesinexcessof∼1063–1064erg(Mathews AGN in clusters is in most cases sufficient to balance radiative &Guo2011b).However,thereisampleevidenceforstrongAGN lossesintheintraclustermedium(ICM;e.g.McNamara&Nulsen outflows in cool-core clusters (e.g. McNamara & Nulsen 2007, 2007,andreferencestherein).However,itislessclearhowtheen- 2012). ergyisreleasedfromthecompact((cid:2)kpc)regionsurroundingthe Theoretical work suggests that the real evolution of heating in clusters might be a combination of both (Brighenti & Mathews 2006;McCarthyetal.2008;Vazza2011;Duboisetal.2012;Short, (cid:3) Thomas&Young2012). E-mail:[email protected] (cid:5)C 2012TheAuthors PublishedbyOxfordUniversityPressonbehalfoftheRoyalAstronomicalSociety Thermalandnon-thermaltracesofAGNfeedback 2367 Quasar-induced outflows at high redshift, possibly following sionfromgalaxyclusterscanofferacomplementarywayoftesting mergers of gas-rich galaxies, have been observed (e.g. Nesvadba andfalsifyingfeedbackmodels.Onceaccelerated,CRhadronscan et al. 2006; Bhattacharya, Di Matteo & Kosowsky 2008; Dunn accumulateingalaxyclusters(Berezinsky,Blasi&Ptuskin1997) etal.2010).Atlowerredshift,themechanicalworkdonebyX-ray andproduceanon-thermalcomponentthatcouldbedetectedbyγ- cavities on the surrounding ICM may represent another viable rayobservations(e.g.Ackermannetal.2010).Secondaryparticles mechanismforheatingandmixingtheICM(e.g.Davidetal.2001; arecontinuouslyinjectedintotheICMviaproton–protoncollisions, McNamara & Nulsen 2007, 2012). Additional mechanisms can possiblyleadingtodetectablesynchrotronradiation(e.g.Blasi& modify the energy requirements of AGN feedback, by providing Colafrancesco1999;Dolag&Enßlin2000).Thecombinedanalysis complementary heating/mixing on various scales. The heating ofradioobservationsandγ-upperlimits,however,presentlysug- from Alfve´n waves in cosmic rays (CRs) enriched bubbles, and geststhatmostoftheobservedlarge-scaleradioemissioninclusters heatingfromCoulomblossesofCRsandthesurroundingthermal cannotbeduetosecondaryelectrons,basedontheoreticalestimates ICM, is an additional interesting topic of research (Loewenstein, oftherequiredtotalenergyinCRs(Brunettietal.2007;Brunetti Zweibel&Begelman1991;Colafrancesco,Dar&DeRu´jula2004; 2009;Donnertetal.2010b),andtotherequiredvaluesofthemag- D o Guo&Oh2008;Sijackietal.2008;Fujita&Ohira2011;Mathews neticfieldintheICM,inconflictwithobservations(Brunettietal. w n &Guo2011a). 2009;Bonafedeetal.2010;Donnertetal.2010a;Bonafedeetal. lo a Thermalconductioncansomewhatreducetheenergybudgetthat 2011;Jeltema&Profumo2011). d e cBeenrttrsaclhAinGgeNr &havMeetiokspirnov1i9d8e6;inBorerdgemratno&steDmavcoidol1in9g88fl;oNwasra(ey.agn. cluTstheerrecfeonrter,esthemdayeteinctfioornmoruslaacbkoouft nthoen-ethneerrmgyalbeumdgisestioonffnroonm- d fro m & Medvedev 2001). Magnetorotational instabilities and heat-flux thermalparticles,andofthehistoryandmodalityofpreviousheat- h driven instabilities in the weak and anisotropic cluster magnetic ing episodes in the ICM. In addition, CR particles may have an ttp s fieldhavebeenproposedtoreducethecoolingofgas(e.g.Quataert importantdynamicaleffectontheICM(e.g.Brunetti&Lazarian ://a c 2008;McCourtetal.2011).Finally,alsomajormergershavebeen 2011b;Ruszkowski&Oh2011),andtheycanalsoaffecttheevo- a d suggestedasaviablemechanismtoreducethecoolingcatastrophe, lution of X-ray cavities powered by AGN jets (e.g. Mathews & e m eveniftherealefficiencyofthismechanismiscontroversial(Poole Brighenti2007;Guo&Oh2008;Sijackietal.2008;Mathews& ic .o etal.2006;Burnsetal.2008). Guo2011a). u p Cosmologicalsimulationswithradiativecooling,starformation Inthisworkwestudytheobservablenon-thermalfeaturesrelated .c o andgalacticwindsareunabletoreproducetheobservedprofilesof to AGN feedback models in a cosmological framework. A few m /m gastemperature,metallicityandentropyintheICM(e.g.Kravtsov single-objectsimulationshavebeenusedtoinvestigatetheroleof n &Borgani2012,forarecentreview).Themostpowerfulfeedback CRfeedbackinstoppingcoolingflows(Guo&Oh2008;Fujita& ra s frreocmentApGaNst.hSaesvbeeraelngsrtouudpiesdswucicthescsofusmllyoliomgpicleamlseinmteudlataiotnresaitnmtehnet OofhiCraR2s0a1t1,sh2o0c1k2s).tHriogwgeerveedr,bthyeAacGceNle,raastiowne(lalnadsrea-tacmceerlgereartiaonnd) /article ofthermalAGNfeedbackincosmologicalGADGETsimulations(e.g. accretionshocks,hasbeenneglectedinpreviousworks. -ab DallaVecchia etal.2004;Sijacki&Springel2006;Sijackietal. To our knowledge, the present study is the first in which stra 2007; Fabjan et al. 2010; McCarthy et al. 2010). In the afore- such detailed CR physics (e.g. particle acceleration, reduced c mentionedpapers,thegrowthofBHsatthecentresofgalaxiesis thermalization at the sub-shock, pressure feedback of CRs and t/42 8 followedusingsinkparticles,andtheenergyreleasefromeachBH effective adiabatic index of the baryon gas) as well as variety of /3 followsfromtheBondi–Hoyleaccretionrate(‘quasar’mode).The AGN feedback models (quasar, jet and bubble modes) have been /23 6 energyemittedbytheAGNisreleasedbyheatingupsmoothedpar- appliedtocosmologicalsimulations. 6 /1 ticle hydrodynamics (SPH)particles surroundingthe surrounding 0 7 ICM. Similar methods have been implemented in adaptive mesh 0 9 2 NUMERICAL METHODS 8 refinement(AMR)gridsimulationsbyCattaneo&Teyssier(2007) 5 andTeyssieretal.(2011)intheRAMSEScode.Thecreationof‘bub- WehaveproducedcosmologicalclustersimulationswiththeAMR by g bles’inflatedbyAGNduringtheir‘radiomode’wassimulatedby codeENZO.Onthebasisofthepublic1.5versionofENZOwehave ue DallaVecchiaetal.(2004)andSijacki&Springel(2006)inGADGET. implementedourmethodstomodeltheevolutionandfeedbackof st o Inthiscase,theenergyreleasedbytheBHisdepositedwithinpairs CRparticlesinjectedatshockwaves(Vazzaetal.2012a),aswell n 0 ofbubblesintheformofthermalenergyorCRenergy,andexerts asour(simplified)implementationsofenergyreleasefromAGN. 4 mechanicalworkPdVonthesurroundingICMwhilebuoyantlyris- ENZO is a grid and AMR code using the piecewise parabolic Ma ingintheclusteratmosphere.Othermodelsofmechanicalfeedback method(PPM)tosolvetheequationsofhydrodynamics,originally y 2 0 fromAGNhavealsobeenimplemented,byassigninga‘wind’drift writtenbyBryanetal.(1995)anddevelopedbytheLaboratoryfor 1 9 velocitytogasparticlessurroundingtheBH,withvelocitiesinthe ComputationalAstrophysicsattheUniversityofCaliforniainSan rangeof∼103–104kms−1(DallaVecchia&Schaye2008;Fabjan Diego(Normanetal.2007;Collinsetal.2010).1 etal.2010).Duboisetal.(2010,2011)implementedaschemeto ThedetaileddescriptionofourmodulesforCRphysicsandAGN followbipolarkineticoutflows(‘jet’mode)fromsimulatedAGN feedbackarepresentedinSection2.1.Inallrunsinthispaper,we inRAMSES,monitoringthegrowthofBHsusingthesameset-upof adoptedradiativecoolingforafullyionizedH–Heplasmawitha Teyssieretal.(2011).Intheframeworkofgalaxyformationstud- constantmetallicityofZ=0.3Z(cid:6),andacoolingfunctionwitha ies,run-timemodelsofradiativeandkineticfeedbackfromAGN cut-offatT=104K(Sarazin&White1987),asinthepublicversion ingalaxysimulationshavebeendevelopedbyKimetal.(2011)in ofENZO,whilethere-ionizationbackgroundduetotheultraviolet ENZOsimulations,andbyOppenheimer&Dave´ (2006)andGabor (UV)radiationfromearlystarsandAGNismodelledbykeepinga etal.(2011)inGADGETsimulations. gastemperaturefloor(∼2×104K)intheredshiftrange4≤z≤7 Todate,however,littleattentionhasbeenpaidtotheamountof (asinVazzaetal.2010a). non-thermalenergydepositedintheICMbythevariousfeedback mechanisms.Thisisanimportantaspect,sincenon-thermalemis- 1http://lca.ucsd.edu 2368 F.Vazza,M.Bru¨ggenandC.Gheller Wedidnotincludestarformationandfeedbackthroughwinds orsupernovaeintheseruns,therebyreducingthecomplexityand memory usage of the code. Hydrodynamical simulations suggest thatwhilesupernovaeareimportanttoreproducetheobservedmetal distribution of the ICM, they do not have a significant impact on the thermal history of the ICM on large scales (e.g. Short et al. 2012). For the simulations presented here, we assumed a ‘concor- dance’ (cid:4) cold dark matter ((cid:4)CDM) cosmology with (cid:5) = 1.0, 0 (cid:5)B=0.0441, (cid:5)DM=0.2139, (cid:5)(cid:4) = 0.742, Hubble parameter h = 0.72 and a normalization for the primordial density power spectrumσ =0.8. 8 D o w n lo 2.1 Cosmicrayphysics a d e d TfeheedbbaacskicomfeCtRhosdisntooumrEoNdZeOltrhuensinajreecteioxnp,laaidnvedecitniodneatanidlipnreVsaszuzrae Figure1. AccelerationefficiencyofCRsasafunctionofMachnumberfor from ecMte,laelwr.a(ht2iic0oh1n2iesaf)fig.ciWvieeennacsbys,yuηmd(iMeff)tuh,satihvtaeCtRsohnsolaycrkedieanpcjceeencldteesrdaotanitotsnhheo(ecM.kgsa.cwBhietnlhlua1mn9b7aec8r-;, td=oif0rf.ee0dr6e,,nt=htep0rd.e1i-f2ef,ex0ries.1nti8tn,lgi=nre0ast.i2so4hsoaownfdtPh=cer0/a.Pc3cg.eilnertahteiopnree-ffishcoiecnkcryegfoiornP.cFr/rPomg=bl0a.c0k, https://a c Blandford & Ostriker 1978; Drury & Voelk 1981; Kang & Jones a d 1990;Ellison,Baring&Jones1995;Malkov&O’CDrury2001; 2.1.1 Shockre-acceleration em Kang & Jones 2007; Caprioli 2012). New CR energy is injected ic Wemodelatrun-timetheshockre-accelerationofCRs,byshocks .o inthesystembymultiplyingtheenergyfluxthrougheachshocked u cellbythetimestepandthecellsurface(ateachAMRlevel): rinujnencitniognosv.eTrhaismceadniubmeaplarretaicduylaernlryicrheeledvoafntCiRnethneercgaysebyopfrsehvoiockuss p.com ρ v3 (cid:9)t causedbyAGNfeedback,whereatlateredshifttheICMisalready /m Ecr=η(M)· u2s · (cid:9)xll, (1) teonrtihceherdeswulittshiCnRKsa(nPgcr&/PJgo∼nes10(2−020to7)1t0h−e1partesze≤nc0e.5o)f.CARccsoirnditnhge nras/a whereρ isthepre-shockdensity,v istheshockvelocity,(cid:9)t and pre-shock region mimics an increased injection efficiency of CR rtic (cid:9)resxpleacretivtuheelyt.imToeesntespuraendentehregyspcaotinaslesrrevsaotliuotniotnheatththeermAaMleRnelergvlyel-iln, eunsienrggyadinifftehreenptoasnta-slyhtoicckal.fTuhnicstiodnynfoamrηic(Me)f,fedcetpecnandebnetotnretahteedratbiyo le-abs the post-shock region is reduced proportionally at run-time. The Pcr/Pginthepre-shock.Inourcase,wecalculateη(M)viaalinear trac 1dthy)eengato+mtai(clγsecforf−fectht1ie)veemcrp]i,rxewtsuhsrueerreoefγogf=atsh5ea/nt3wd,oγC,cRrP=sefwf4=i/t3hP,inegge+iasctPhhcecreg=lalsρfeo[nl(leγorwg−ys cainnatsdeerPpthcore/laPetfigfio=nci0ebn.e3ctwy(beηoe(tnMhtt)ahikeseiendxeftrnroetmimcaeKlatconasgthe&esooJnofenEeocsfr/2VE0a0gz7z=)a.Fe0tiga[i.ln.12sw0h1ho2iwcahs] t/428/3/23 densityandecr istheCR-energydensity.Thedynamicalfeedback the acceleration efficiency, based on the interpolation of Kang & 66 of CR pressure is treated in the Riemann solver by updating the Jones(2007),asafunctionofthepressureratiobetweenCRsand /10 gas matter fluxes in the in 1D sweeps along the coordinate axes thermalgasinthepre-shockregion.Thepost-shockthermalization 70 and using the effective gamma factor (γeff = (γPPg++γPcrPcr)), for the isthenreducedatrun-timeaccordingly,asinVazzaetal.(2012a). 985 computationofthelocalsoundspeedincells. g cr BasedonourtestsinVazzaetal.(2012a),therun-timemodelling by 4/3AesvienryVwazhzearee,twalh.i(c2h0c1o2rare),spwoenudssetothteherefllaattitveissttpicosvsailbuleeomfoγmcren=- othfeshfioncakldreis-atrcicbeulteiorantioofnCoRfCeRnserdgoyesthnaottishaovneaavderraamgeatiinccirmeapsaecdtboyn gues tfuomrγscpre=ctr4u/m3]o.fFCixRinsg,tthhreouvgalhuγecorf=γqcr/w3i[twhiinththfe(ps)im∝upl−atqeadnvdoqlu=m4e ∼r(aSt1ei0octipboeenrtw3c.ee2ne)nt,aiCnnsRdidstheaetnrhedefotvhrieerritamhlearliamdgpiausasc.itnTohaflirwse-aiasycsbceeticlneayruastieendstihCdeeRpcsrleousnsstutehrrees t on 04 M isanunavoidableassumptionofthetwo-fluidmodeladoptedhere. a However,oncetheCR-energydensityisspecified,theCRpressure finalbudgetofCRenergyoftheICMissmall. y 2 0 dependsonlyweaklyonthespectralshapeoff(p)andonthecut-off 1 9 ofthespectrum(e.g.Jones&Kang1993;Jubelgasetal.2008). 2.1.2 HadronicandCoulomblosses To model the injection of CRs at each shocked cell, we devel- opedarun-timeshockfinderbasedonpressurejumps,similarto CRscanloseenergyviabinaryinteractionswiththermalparticles Ryuetal.(2003).Ateachtimestep,weflagcellswithanegative of the ICM. This channel of energy exchange between thermal 3Ddivergence,∇·v<0,andwithconcordantlocalgradientsof and relativistic particles in the ICM is important for the high gas temperatureandentropy,∇S·∇T>0(e.g.Ryuetal.2003).The density]ρ/(μm )>10−2cm−3]ofcoolcores.Relativisticprotons p local Mach number is computed by inverting standard Rankine– transferenergytothethermalgasviaCoulombcollisionswiththe Hugoniot jump conditions for gas pressure. This method can run ionizedgas.Theycanalsointeracthadronicallywiththeambient in either fixed grid resolution or AMR mode in ENZO runs, and thermalgasandproducemainlyπ+,π−andπ0,providedthattheir compares very well with the shock-finding methods we develop kineticenergyexceedsthethresholdof282MeVforthereaction. elsewhere (Vazza, Brunetti & Gheller 2009b). Compared to our The neutral pions decay after a mean lifetime of ≈9 × 10−17s firstpaperonthissubject(Vazzaetal.2012a),weaddthetreatment intoγ-rays.ToestimatethetotalenergytransferratebetweenCRs ofseveraladditionalphysicalprocessesthatarelistedbelow. and thermal gas in both mechanisms, we need to determine the Thermalandnon-thermaltracesofAGNfeedback 2369 CR-energyspectrum.Sincethisinformationisnotreadilyavailable andgroups(e.g.Sijackietal.2007;Teyssieretal.2011;Martizzi inthetwo-fluidmodel,wemustassumeanapproximatesteady-state etal.2012).3 spectrumfortheCR-energydistribution.Wefixedaspectralindex Inthefollowing,wewillrefertothecellsexceedingthisdensity ofα=2.5fortheparticleenergy2andcomputedthetotalCoulomb thresholdandpoweringenergyfeedbackas‘AGNcells’. andhadroniclossratesasafunctionoftheICMdensityandofE WehaveimplementedthreemodesofAGNfeedback:a‘quasar’ cr foreachcell,asinGuo&Oh(2008): mode (i.e. thermal output of energy from AGN, Section 2.2.1), a n E ‘jet’mode(i.e.kineticenergyoutputfrombipolarjetsaroundAGN, (cid:12)coll=−ζccme−3ergccmr−3 ergs−1cm−3, (2) Section2.2.2)anda‘radio’mode(i.e.creationofbuoyantbubbles inpressureequilibriumwiththeICM,Section2.2.3). wheren ≈nistheelectronnumberdensity,andζ =7.51×10−16 Once the feedback mode is specified, the only parameters that e c isthecoefficientforallcollisionalenergylossterms.Inhadronic mustbesetare(a)theinitialredshiftforthestartofAGNfeedback collisions, only ∼1/6 of the inelastic energy goes into secondary (z )and(b)theenergyreleaseofeachsingleAGNevent,E . AGN AGN electrons(Colafrancescoetal.2004;Guo&Oh2008).Theenergy- Inthecaseof‘quasar’and‘jet’modesthisdirectlymeasuresthe D dominating region of CR electrons (γ ∼ 102) will heat the ICM energyweprovideforeachburstofeitherthermalorkineticenergy, ow n through Coulomb interactions, plasma oscillations and excitation whileforthe‘bubble’feedbackthisrepresentstheestimatedtotal lo a ofAlfve´nwaves(e.g.Guo&Oh2008).Therefore,wecanassume energyreleasedtotheICMbythecreationofbubbleswithinternal d e tthhaetrmthaeliszeatsieocnoannddartyhueslehcetraotntsheloIsCeMm.oSsitmoiflatrhteoireqenueartigoynt(h2r)o,utghhe p&reBssruu¨gregePnbb2a0n0d8)v.olumeVbb,EAGN≈3PbbVbb/2(e.g.Scannapieco d from heatingrateoftheICMthroughCoulombandhadroniccollisions Evenifjetsandradiobubblesareassociatedwiththesametype h canbecomputedas ofAGNfeedback(e.g.McNamara&Nulsen2007,2012),inour ttp s n E studytheyareconsideredasalternative scenarios.Thisallows us ://a (cid:12) =ξ e cr ergs−1cm−3, (3) to distinguish the CR effects of buoyancy and impulsive kinetic ca heat ccm−3ergscm−3 d feedbackinaclearerway. e m whereξ =2.63×10−16(Guo&Oh2008).Inoursimulationswith Inapreliminarysetoftests(seeAppendixA)weexploredvarious ic c .o radiativecoolingandAGNfeedback,therateofenergylossdueto recipesfortheimplementationoffeedbackmodesinonereference u p thesecollisionsisextremelysmall,typically∼10−3to10−4ofE cluster,beforevaryingtheefficienciesinthewholesetofclusters. .c cr o ore duringthetimestepateachAMRlevel.Thisallowsustouse Here,however,wewilldiscussonlythe‘fiducial’subsetofpa- m g /m asimplefirst-orderintegrationtocomputetheenergylossesofCRs rametersforwhicheachimplementationoffeedbackmodesshowed n (andthecorrespondinggasheatingrate)atrun-time. thebestperformance.Whiletherangeofspatialresolutionachieved ras hadInroonuicrapnrdevCioouuslowmobrklo(sVseasz.zIanegteanle.ra2l0,1m2oa)d,ewlliengditdhinsoptroincecslusdaet ifneaotuurrersuansssoiscipartoedbawbliythneoatcshuAffiGciNenmttoodestu(ed.yg.stpheecimficorspmhoallol-gsycaolef /article run-timedecreasestheCRenergybyafactorof∼10withinclus- jetsorrisingbubbles),ourresolutionandphysicalset-uparesuitable -a tercores,whileyieldingidenticalresultsfortheremainingcluster for"studyingthelarge-scalefeaturesofCRsintheICM. bstra volumecomparedtorunsthatneglectlosses. c t/4 2.2.1 ThermalfeedbackfromAGN 2 8 /3 2.2 ModelsoffeedbackfromAGN AsignificantfractionoftheenergyemittedfromAGNcanthermally /2 3 coupletothesurroundinggas.Onecandefineefficienciessuchthat 6 InVazza(2011),weimplementedasimplemodelofAGNfeedback theenergyaddedintime(cid:9)tis 6/1 inENZO,viainjectionofthermalenergyattheoppositesidesofthe 07 coolingregionofgalaxyclusters.Here,weexploremorecomplex (cid:9)EAGN,g=(cid:15)r(cid:15)fM˙BHc2(cid:9)t, (4) 098 recipesofenergyfeedbackbetweenthecoldgasandthesurrounding where (cid:15) ∼ 0.1 is the bolometric radiative efficiency for a 5 b ICM,allowingalsoforadirectinputofCRenergyfromAGN. SchwarzsrchildBH(Shakura&Sunyaev1973),M˙ ,and(cid:15) isthe y g Ateachtimestep,theidentificationofasuitablelocationofthe BH f ue couplingefficiencywiththethermalgas,whichisusuallyassumed s cgeanstoravlersduepnesrimtya.sFsiirvset,BwHeflisagbacseeldlsohnostthinegsaimgpalsedmenesaistyureexcoefeldoicnagl tMo bevienrstuhseσranregleatoiofn(cid:15)(fe≈.g.0D.0i5M–a0t.t1e5oeintaolr.d2e0r05to;Bfiotoththe&obSscehrvayede t on 0 agiventhreshold,n≥nBH,andthenweselectasactive‘AGNcells’ 20B0H9;Yangetval.2012).ThisthermalcouplingbetweentheAGN 4 M onlythemaximawithincubicregionsofsize≈1Mpch−1.Basedon a moredetailedmodellingofBHgrowthusingsinkparticles(Sijacki andthesurroundingICMisusuallycalled‘quasar’feedback,which y 2 maybeamechanismforpre-heatingoftheICM(McCarthyetal. 0 etal.2007;Teyssieretal.2011;Martizzietal.2012),wetunedthe 19 2004, 2008; Lapi, Cavaliere & Menci 2005; Sijacki, Springel & thresholdvaluetoexciteAGNfeedbackton ≈10−2cm−3.The min Haehnelt2009;Lapi,Fusco-Femiano&Cavaliere2010).Inaddi- use of this threshold is motivated by the fact that in our fiducial tion, quasar-induced outflows at high redshift, possibly following set-up(peakresolutionof25kpch−1percell)themassenclosedin mergers of gas-rich galaxies, have been observed in many cases acellwithn=10−1to10−2cm−3is≈1010to1011M(cid:6).Thisisthe (e.g. Nesvadba et al. 2006; Bhattacharya et al. 2008; Dunn et al. typicalgasmasssurroundingBHsofM ∼108 to5×109M(cid:6), BH which are commonly hosted inside the masses of galaxy clusters 3Thechoiceofrelyingonlyonthegasdensityasaproxymaytriggeralso feedbackfromcoldfilamentsconnectinggalaxies,whereAGNfeedbackof 2ThischoiceofαcorrespondstoaMachnumberofM=3attheparticle thetypeconsideredhereisunlikely.However,oursimulationshavesufficient injection.Onaverage,thisrepresentsthetypicalinjectionspectraofparticles highresolutiononlyingalaxyclusters.Thisseemstoexcludeanyspurious acceleratedatpowerfulmergershocksincosmologicalsimulations,which releaseoffeedbackenergyfromcoldfilaments,asindeedwefind.Tofully are dominant sources of thermalization and CR injection within clusters avoidthispossibility,onewouldhavetoresorttomorecomplexmodels (e.g.Vazzaetal.2011b,andreferencestherein). (e.g.sinkparticles). 2370 F.Vazza,M.Bru¨ggenandC.Gheller 2010).SimilartoMcCarthyetal.(2010)andTeyssieretal.(2011), in our cosmological runs. For this reason, we created already whenwedetectcellswithn≥n ,wereleasethethermalenergy, formedevacuatedbubblesaroundcellshostinganAGN,similarto min E ,addingtothetotalandinternalgasenergyinsidecellsatthe Scannapieco&Bru¨ggen(2008).Thebubblesarecreatedin(approx- AGN highestavailableAMRlevel. imate)pressureequilibriumwiththesurroundingICM,bydecreas- Thisimplementationofquasarfeedbackisonlyanapproximation ingthegasdensityinsidethecellsbyρ(cid:12)=δ ρ,andcorrespondingly bb ofthetruephysicalprocessesatplay,i.e.thelaunchingofstrong increasingthethermalenergyinordertoconservetheoriginalgas windsduetotheradiationpressureofphotonsfromtheaccretion pressure. During run-time, this modification is performed over a disc.Thisisunavoidable,giventhatourbestresolutionisordersof single time step of the simulation, at the highest available AMR magnitudelargerthanthetheoreticalaccretiondiscregion,andalso level. This obviously leads to a loss of gas mass in these cluster giventhedifficultyofmodellingtheradiativetransferofphotons runs. The simulations show that on average this loss amounts to fromtheaccretionregion. ∼5–10percentofthegasmassbytheendofthesimulation. Ourchoiceofaminimumgasdensityn =10−2cm−3 selects Theinitialunderdensityofthebubblesissetbyenergyconser- min thetypicalenvironmentofmassiveBHswithinclustersandgroups. vationinsidethebubbles,afterthatEAGN isaddedtothegasand D o Furthermore,weassumeanenergyoutputsimilartomosttheoretical CRenergywithinthesamevolume.FortherangeofEAGNandnmin w n models,withoutactuallymeasuringtheaccretionpowerofBHsat adopted inthiswork,theunderdensityinsidethebubblesistypi- lo a run-time.Thisisdifferentfromsimulationswherethemassgrowth cally0.1–0.01.Thebubblesaregeneratedasapairofcubicblocks d e orefcBonHsstriuscmtioondeolfletdheusBinHgmsinatktepraartciccrleetsi,ownhriactheeantarubnle-stiamnea.cTchuirsaties iosfa23vecreyllsc,ruadtethaepdpirsotaxnimceatoifontwaondceilslsexfrpoemctethdetoclulesatedrtcoeenxtrpee.dTiehnist d fro m notpossibleinourcase.Ourapproachisonlyafirststeptoinclude numericallymixingofthethermalenergy. h theeffectsofAGNfeedbackinourversionofthecode,anditallows Inprinciple,wecanalsoallowforthepresenceofCRsinjectedby ttp s usafastandefficientstudyofAGNmodes. theAGN,parametrizedbyφcr=Ecr,AGN/EAGNofthetotalinjected ://a c energy,similartoSijackietal.(2008).Inthiscasetheenergyofthe a d 2.2.2 KineticfeedbackfromAGN bubblesisconservedalsobymodifyingtheeffectiveadiabaticindex em ofthemixtureofgas+CRs(asinVazzaetal.2012a).However,inthe ic TheinnermostICMcanbeaffectedbytheinjectionofkineticenergy mainbodyofthispaperweonlyrefertopurelythermalbubbles.We .ou p throughbipolarjetsoriginatingfromtheAGN(e.g.Binney&Tabor presentafewresultswithgas+CRsinflatedbubblesinAppendixA .c o 1995;Ciotti&Ostriker2012).Thisenergycanbethermalizedby (theresults,however,arenotsignificantlydifferent). m impactingontheICMafterashortdistancefromtheclustercentre /m n (10–100kpc;Pope2009).Gasparietal.(2011a,b)andDuboisetal. ra s (2010,2011)recentlysimulatedthemechanicalcouplingbetween 3 RESULTS /a purelykineticjetsfromtheAGNandthesurroundingICMinthe rtic The results discussed in the main part of this paper have been le coolingregion. -a producedconsideringthefollowingphysicalmechanisms:(a)pure b Inthismodeleachofthetwojetsisinitializedasapureinputof s kineticenergydensityE =1/2ρv2 ,withvelocity,v ,pointing cooling; (b) cooling and thermal feedback by AGN from z ≤ 4, tra radiallyoutwardsfromtAhGeNclustercenjettre.Evenifinourjevtersionof withafixedenergyreleasepereventofEAGN =1059erg(oralso ct/4 1060erginthecaseofE1andE5A);(c)coolingandkineticenergy 2 ENZOthelaunchingdirectionofthejetsissetbyarandomselection feedbackbyAGNfromz≤4,withafixedenergyreleaseperevent 8/3 ofcoordinateaxes,inthisworkwekeepthejetaxisfixed.Inthis ofE =1059erg;(d)coolingandenergyfeedbackfrombuoyant /23 way, velocity effects related to the direction of jet launching can AGN 6 easilybedetected(Section3.4).Everytimeajetisgenerated,we bubblesinjectedbyAGNfromz≤4,withanapproximateenergy 6/1 modifythegasvelocity,thetotalandinternalenergyatthehighest preesrimveunltatoefdEfoAGuNrc∼lus1t0e5r9seinrgt.hWeimthastsheraanbgoeve10i1m4p≤lemMe≤nta1t0io1n5sM,w(cid:6)e. 0709 available AMR level in a pair of cells on opposite sides of the 8 gasdensitypeak.Thewidthandtheinitialextensionofthejetare Thelistofthemostimportantparametersofall‘fiducial’models 5 b set by the maximum resolution, which is ∼10 times larger than investigatedinthearticleisgiveninTable1.Thefirstmodel(pure y g thebestresolutionavailablein‘single-object’runs(Gasparietal. cooling) obviously produces strong cooling flows in all systems. ues 2011a).Atourresolution,theequivalentopeningangleofeachjetis In this work, this model is therefore regarded only as a standard t o ∼30◦–40◦withrespecttotheclustercentre.Thisisanunavoidable referencetoassesstheroleofeachfeedbackmodel. n 0 drawback of the fact that in such cosmological runs achieving a Inordertoachieve thelargest possibledynamical rangeinside 4 M thevolumewhereeachclusterforms,thefourclusterswereresim- a muchlargerresolutioniscomputationallyveryexpensive.However, y ulatedathighspatialandDMmassresolutionstartingfromparent 2 forthedynamicalfeedbackofjetsthemostimportantquantityis 0 theinjectedkineticenergy,whichissimilartosimulationsathigher simulationsatlowerresolutionandaddingnestedinitialconditions 19 resolution,oncethedifferentmassloadofthejetsandlaunching velocityareconsidered.Alsotheinitialvelocityofourjets(which Table1. Listofthephysicalmodelsadoptedinourruns.Column1:identi- depends on the typical density of AGN cells, through the total ficationname.Column2:cooling.Column3:energyperevent.Column4: kinetic energy released in the AGN burst) is ∼500–1000kms−1, detailsonfeedbackmode(withapproximatevaluesofheatingtemperature, jetvelocityandinitialbubbleunderdensity). about one order of magnitude lower than the typical velocity of jetsinsingle-objectsimulationsatmuchhigherresolution(Gaspari eta"l.2011a,b). ID Cooling EAGN(erg) Feedbackmode Cooling Yes 0 No Quasars Yes 1059 Thermal,TAGN∼5×107K 2.2.3 BubblefeedbackfromAGN Quasars2 Yes 1060 Thermal,TAGN=5×108K ThecreationofbuoyantbubblesintheICMinflatedbyjetsrequires Jets Yes 1059 Kinetic,vjet∼750kms−1 a spatial resolution (∼0.1–1 kpc) beyond what can be achieved Bubbles Yes 1059 Buoyant,δbbl∼0.05 Thermalandnon-thermaltracesofAGNfeedback 2371 Table 2. Main parameters of the clusters simulated in shocksandintenseturbulentmotionsintheICMouttotheclusters this work. Column 1: identification name. Column 2: to- outskirts.Shocksandturbulenceareanalysedinthenextsections tal(gas+DM)massatz=0.Column3:virialradius(Rv). (Sections3.2–3.4).Inaddition,ourtestwithENZOhasshownthat Column4:dynamicalstateatz=0.Theidentificationname thiscompositeAMRcriterionbettercapturesphysicalmixingmo- ofeachclusterischosentobeconsistentwithotherworks tionsintheICM,andreducestheamountofnumericalmixing,with ofours(Vazzaetal.2010a,2012a).Theclusterparameters importantconsequencesintheamountofcoldgaswithintheclus- arereferredtothesimplenon-radiativesimulationofeach tervolume(Vazza2011).However,thismethodiscomputationally object. muchmoreexpensivewithrespecttothegas/DMoverdensitycri- ID Mtot(×1014M(cid:6)) Rv(Mpc) Dynamicalstate teriaalone,andwecouldaffordtorunitonlyinthe‘fiducial’AGN models(Table1). E1 11.2 2.67 Post-merger InAppendixAweshowtheeffectsofdifferentimplementations E5A 8.2 2.39 Merging andenergiesbudgetinAGNfeedbackinthesmallestofthesesys- H5 2.4 1.70 Post-merger tems,usingthestandardAMRcriterionwhichworksongas/DM D H10 1.2 1.20 Relaxed o overdensity.Inthemainpaperwewillstudytheobservablether- w n mal and non-thermal features of the ‘fiducial’ implementation of lo a of increasing spatial and mass resolution (e.g. Abel et al. 1998). eachfeedbackmodeinourversionofENZO.Theparametersofthe de Tcewnotrleedveolnsothfenecsltuesdteinrictieanltcroesn.dTitihoensbowxeraetptlhaecefidrsint lceuvbeilchreagdiothnes sfemeadlblacclkusitsergisvaemnpilneTreasbilmeu2l.atedwiththedifferentrecipesofAGN d from sizeof≈95Mpch−1(withm ≈5.4×109M(cid:6)h−1andconstant Theevolutionofgastemperaturebetweenz=2and0incluster h spatialresolutionof(cid:9)1≈425dmkpch−1).Thesecondboxhadasize H5 (quasar mode) is shown in the panels of Fig. 2. Blast waves ttps of ≈47.5Mpch−1 (with mdm ∼ 6.7 × 108M(cid:6)h−1 and constant triggeredbyAGNfeedback(highlightedbyarrows)inthethermal ://a spatialresolutionof(cid:9) ≈212kpch−1).Foreveryclusterrun,we modedrivepowerfulshocksthroughtheICM,addingtothepattern ca 1 d identifiedcubicregionswiththesizeof∼6Rv(whereRvisthevirial ofmergerandaccretionshockwaves.Weobservethatingeneral em radiusofclustersatz=0,calculatedinlowerresolutionruns),and only two to three AGN cells are located inside forming clusters ic .o allowed for three additional levels of mesh refinement, achieving at a high redshift, while at a lower redshift only one AGN cell u p apeakspatialresolutionof(cid:9)≈25kpch−1 (inthefollowing,we is usually found inside the virial volume. The shocks triggered .c o willrefertothissubvolumeastothe‘AMRregion’).Fromz=30 byAGNfeedbackareefficientinremovingthecoldgasphaseat m (theinitialredshiftofthesimulation)toz=2,meshrefinementis z>1,whileproducingsignificantamountsofCRsintheforming /mn triggeredbygasorDMoverdensity.Fromz=2anadditionalre- cluster.Blastwavesalsocausetheexpulsionofasignificantfraction ras fiisnsewmietcnhtecdritoenr.ioTnhbisasseedcoonnd1ADMveRloccriittyerjiuomnpiss(dVeasizgznaeedttaol.c2a0p0t9uare) oetfatlh.e(2h0o1t0)g.as from the cluster volume, in line with McCarthy /article -a b s tra c t/4 2 8 /3 /2 3 6 6 /1 0 7 0 9 8 5 b y g u e s t o n 0 4 M a y 2 0 1 9 Figure2. Timesequenceofgastemperature[inunitsof(K)inthecolourcoding]forasliceof5×5comovingMpch−1andwidth25kpch−1forrunH5 with‘quasar’feedback.Thearrowspointtothelocationsofrecent‘quasar’eventsinthevolume.The‘coarse’resolutionofthefirstsnapshotsisduetothe factthatweturnonAMRbasedonvelocityjumpsatz=2. 2372 F.Vazza,M.Bru¨ggenandC.Gheller D o w n lo a d e d fro m h ttp s ://a c a d e m ic .o u p .c o m /m n ra s /a rtic le -a b s tra c t/4 2 8 /3 /2 3 6 6 /1 0 7 0 9 Figure3. Timesequenceofmass-weightedaveragegastemperatureforacomovingvolumeof∼(5Mpch−1)3intheformationregionofclusterH5.Weshow 85 theresultsofthethreeresimulationsemployingAGNfeedback,alwayswithanenergypereventofEAGN=1059erg.Horizontalarrowssuggestthelocation by ofinterestingepisodesofAGNfeedbackwithinthevolume. g u e s Fig.3showsacomparisonofprojectedgastemperatureacross ofclusterE5AinrunsemployingAGNfeedback.Jetsandquasars t o n theentireAMRregionofthesamecluster,atthreedifferentcosmic arenotactiveanymoreatz=0inthisrun,andnoclearfeatures 0 4 epochs.Whilethelarge-scalemorphologyoftheclusterissimilar relatedtotheiractivitycanbeseenintheinnermostclusterregion M a inallrunstowardstheendofthesimulation,eachfeedbackmode atthisredshift.Apairofbubbles,ontheotherhand,hasrecently y 2 locallyperturbstheICMinadifferentway,drivingpowerfulout- beeninjectedandcanbeseenasacoupleofunderdenseblobson 0 1 bursts (or buoyant bubbles) whenever strong cooling causes n > oppositesidesoftheclustercentre(shownbygreencirclesinthe 9 n . image).OtherimagesofactivejetsandbubblescanbeseeninFig. min Inthequasarmode,thefinallarge-scalemorphologyofthecluster A1ofAppendixA. isalsoaffectedbytheoverallactionofpowerfulthermalbursts,and Weusephasediagramsforthecellsinsidethevirialvolumein presents a large amount of ejected hot gas outside of the cluster order to characterize relevant differences. In Fig. 5 we show the volume. In the jet mode, prominent jets are launched along the example of phase diagrams for an early cosmic epoch before the horizontalaxis,drivingpowerfulshocks(M∼5–10)intothe∼106– cluster formation, z = 2 (top panels), and after the formation of 107Kcoolinggasaswellasgasmotionsalongtheiraxis.Thebubble thecluster,z=0.2(bottompanels)forthefourdifferentfeedback mode,ontheotherhand,hasbarelydetectableeffectsonthelarge modes.Themostevidentdifferencebetweentherunsistheabsence scale, and is globally similar to a simple pure cooling run (not ofthe‘coolingflow’phaseonthelowerrightpartofthediagram, shown). in all runs employing AGN feedback. In the bubble mode, the Fig. 4 presents a zoomed image of the gas density maps for a feedbackfrombuoyantbubblesonlyaffectsthehigh-densityregions sliceof2×2Mpc2anddepth25kpch−1forthefinalconfiguration aroundtheclustercore.Athighredshift,thephasediagramisvery Thermalandnon-thermaltracesofAGNfeedback 2373 D o w n lo a d e d fro Figure4. Gasdensity(toppanels)forasliceof2×2Mpch−2 andwidth25kpch−1 acrossresimulationsofclusterE5Aatz=0.Thecolourcodingis m [ρ/ρcr,b](whereρcr,bisthebaryoncriticaldensity).Thegreencirclesshowtheprojectedlocationoftwobubblespreviouslyinjectedintheclustercentre. http s ://a c a d e m ic .o u p .c o m /m n ra s /a rtic le -a b s tra c t/4 2 8 /3 /2 3 6 6 /1 0 7 0 9 8 5 b y g u e s t o n 0 4 M a y 2 0 1 9 Figure5. PhasediagramsofcellswithinthevirialvolumeoftheresimulationsofclusterE1atz=2(toppanels)andatz=0.2(bottompanels).Thecolour codingshowsthecell-wisepressureratiobetweenCRandgaswithincells;thethicklineshowstheaveragetrendwithineachphasediagram. similartothepurecoolingrun,withtheexceptionoftheselective phasehasalmostdisappearedinjetandquasarfeedback,duetothe removalofgaswithn >10−2cm−3cells,whichareheatedtohigh efficientandvolume-fillingnatureoftheheatingeventsdrivenby min temperatures(>5×107K)bytheinjectionofbubbles.Ontheother such powerful mechanisms. In the bubble run the phase diagram hand,inthequasarandjetmodestheactionoffeedbackdrastically is similar to the pure cooling run, with the exception of the n ∼ altersthephasediagramswithrespecttothepurecoolingrun.This 10−2cm−3regime,wherebubblesareinjectedalmostcontinuously isparticularlyevidentinthequasarmodeatz=2,wherea‘cloud’of andovercoolingisbalancedbymixing.Thisleavesasignificant high-temperaturecells(T>107Kandn∼10−4cm−3)isleftafter portionof‘cold’(<106K)anddensecellsinthesimulatedvolume, thepassageofanAGN-drivenblastwave.Atz=0.2,the‘cooling’ characterizedbylargevaluesofP /P . cr g 2374 F.Vazza,M.Bru¨ggenandC.Gheller if the power per event is 10 times larger. In Fig. 7 we show the 3.1 X-rayproperties total energy released by AGN in the formation region of cluster WecompareoursimulationswiththecollectionofChandracluster E1,intheredshiftrange0≤z≤1.Bytheendofthesimulation, observationsofCavagnoloetal.(2009),publiclyavailableviathe allfeedbackmodesusedatotalamountofenergyintherangeof ACCEPT catalogue (http://www.pa.msu.edu/astro/MC2/accept). ∼2–9×1061erg,correspondingto∼0.01–0.09ofthetotalthermal The catalogue consists of 241 clusters mostly located in the red- energyoftheclusteratz=0. shiftrange0.05≥z≥0.5andwithaveragecentraltemperaturesin Interestingly,bytheendofthesimulationthehigherpowerAGN therangeof0.5–10keV.Forourcomparisonweselectedasubsam- mode(EAGN=1060erg)usedabout∼30percentlessenergythan pleof∼60objectsfromtheACCEPTcatalogue,withredshiftsand thelowerpowerrun(EAGN=1059erg),whichis∼1/3ofthetotal averagecentraltemperaturescompatiblewithourclusterdataset: energy used in the kinetic feedback mode at a lower power. The z≤0.2and0.2≤T≤5keVinside0.2R .Forfurtherdetailson totalenergybudgetusedinthebubblemodeissimilartothequasar 200 theACCEPTcatalogueandsetofobservations,wereferthereader modeatthesamepowerperevent.Wenotethatinthissystemthe toCavagnoloetal.(2009). useofAGNfeedbackissignificantlyreducedforz>0.6,duetothe D o In Fig. 6, we show the radial profile of reduced gas entropy4 onsetoflarge-scalemergersandpowerfulshockheatingwithinthe w (K/TSL, where K = TSL/n2/3 and TSL is the spectroscopic-like cluster volume. However, episodes of AGN feedback are present nloa temperature; Rasia et al. 2005), gas number density, n, and even at later epochs. Overall, a smaller amount of energy from d e sTpheectsriomscuolaptiecd-lipkreofitelmespaerreatucroemfpoarrethdetodiftfheerepnrtofifeleedsboafckobmseordveesd. ApuGreNlyfethederbmacaklfiesendebcaecsksa,royrtboubbabllaensc,ecothmepcaarteadsttroopkhinicetciocofeliendgbwacitkh. d from clusters, divided by the colour coding into ‘cool-core’ systems if This is consistent with analytical results of Pope (2009, fig. 1), h theircentralentropyis<30keVcm2 (inblue),or‘non-cool-core’ whereitisshownthatahigherinjectionrateofkineticenergy,with ttps systemsotherwise(red). respecttothermalenergy,isnecessarytobalancethecoolingflow ://a As expected, none of the pure cooling runs can reproduce the forclustermasses>1015M(cid:6).Thisisbecauseinamassivecluster ca d observed profiles. In these runs, steep gas density profiles with thecriticalmomentuminjectionrateneededtoovercomethepull em centralvaluesof∼0.1–1cm−3 andreducedcentralentropybelow of gravity and the surrounding gas pressure in the cluster core is ic .o ∼10cm2 are found, at odds with observations and in agreement more difficult to reach than the critical injection rate of thermal u p with standard radiative runs (e.g. Li & Bryan 2012, for a recent energyrequiredtoovercomethecoolingflowviathermalfeedback. .c o review).OnlyclusterE1ismarginallywithintherangeallowedby Inaddition,theenergyreleasefromjetsclosetotheclustercoreis m /m observationsfordensityandentropy.However,consideringthatthis veryanisotropic,anditbecomesmoreisotropic(duetothedriving n systemhasjustgonethroughastrongmajormergerat(z∼0.25), ofshocks)onlyatadistanceof∼100–200kpcfromthecore. ras ibtycaitnsngoatsbedecnlasistsyifiaenddatsemacplearsastiucraelpcoroofil-lceo.rOencltuhseteor,thaesrsuhgangdes,taeldl obsOevrveeradlll,otwhe-tXem-rpayerpartuorpeerntioens-ocofoolu-rcoclruesctelursstseeresmsicnocnestihsteentytpwicitahl /article runs employing AGN feedback yield a much better comparison internaldropintemperatureofcool-coreclustersisnotobserved. -a b withobservations.Jetandbubblefeedbackmodesproduceroughly On the other hand, in the epoch in which the AGN feedback is stra similarprofilesforallclusters:aratherflatentropyprofilewithin triggeredinoursimulations,thethermalstructureoftheinnermost c clustercores(compatiblewiththehighentropyfloorofnon-cool- clusterregionsisnotcompatiblewiththeappearanceoftypicalcool t/42 8 coreclustersinCavagnoloetal.2009),ashallowdensityprofileand cores either, because the internal drop in temperature is typically /3 analmostisothermaltemperatureprofileinsidethecentral∼Mpc muchstrongerthanwhatisobserved.Alackofresolutioncompared /23 6 fromthecentre.Lessclearresultsarefoundinthecaseofquasar toobservationscouldberesponsibleforthat. 6 /1 feedback. While in clusters E1, H5 and H10 a reasonable match However,wefindnoevidenceofbimodalityinthedistribution 0 7 with observations is found with the ‘fiducial’ energy budget of of clusters entropy or gas density in our resimulations, at odds 09 E =1059ergperevent,thisisinsufficienttoquenchthecooling withobservations(Cavagnoloetal.2009;Prattetal.2010;Eckert 85 caAtGaNstrophe in cluster E5A, where a rather standard cooling flow et al. 2012). This might be connected to our coarse sampling of by g takesplacebytheendofthesimulation.Byincreasingtheavailable theparameterspaceofAGNfeedback,ormayinsteadcallforthe u e energypereventbyoneorderofmagnitude,EAGN =1060erg,the implementationofmorecomplexphysicsinoursimulations.How- st o coolingflowisstoppedalsoinclusterE5A(lowerrowofFig.6). ever,theinvestigatedsampleistoosmalltoaddressthisimportant n 0 Interestingly,thesameistrueforaresimulationofclusterE1using issueindetail,andwewillleavethisforfuturework. 4 M thesamehigherenergybudget.However,inthecaseoftheother We illustrate the effects of AGN feedback on the radial distri- a smallerclustersthehigherenergybudgetistoolargeandthethermal bution of baryon fraction by referring to the relevant example of y 2 structure of both is destroyed, leading to a gas-poor cluster (see cluster H5 at z = 0, shown in Fig. 8. As expected, the onset of 01 9 AppendixA). thecoolingflowcausestoostrongaconcentrationofbaryonsin- This may simply suggest that different cluster masses must be sidethecoolingradius,clearlyatoddswithobservation(e.g.Ettori characterizedbydifferenttypicalpowersperevent,mirroringthe etal.2009;Sun2012,andreferencestherein).Ontheotherhand, factthatalargerpowerpereventisneededtobalancethegravityand allfeedbackrunsproduceafinaldistributionofgasmasswhichis thepressureofthecoolinggaseousatmosphere(e.g.Pope2009). moreinagreementwithobservations,showingincreasingprofiles However,itisinterestingtonotethattheself-regulatingnatureof towardstheclusterouterregions(e.g.Ettorietal.2009;Sun2012). AGNfeedback(evenwiththissimpleimplementation)yieldsvery Approaching R200 the enclosed gas mass fraction is 90–100 per similarfinalprofilesinthemostmassiveobjectofoursample,even cent of the cosmological baryon fraction. However, the different implementations of feedback produce significantly different and time-dependentfeaturesintheprofile.Inparticular,themechanical 4Weusethe‘reduced’entropy,ratherthantheusualentropy,K=T/n2/3, action of jets changes the shape of the enclosed baryon fraction sincethisremovesthedependenceonthehostclustermass(Borganietal. in this low-mass system, due to a more recent feedback episode. 2008). Similar trends are found in the other systems, provided that the Thermalandnon-thermaltracesofAGNfeedback 2375 D o w n lo a d e d fro m h ttp s ://a c a d e m ic .o u p .c o m /m n ra s /a rtic le -a b s tra c t/4 2 8 /3 /2 3 6 6 /1 0 7 0 9 8 5 b y g u e s t o n 0 4 M a y 2 0 1 9 Figure6. Profilesofreducedgasentropy,gasnumberdensityandspectroscopic-liketemperatureforthesimulatedclusters(thicklineswithdifferentstyles) andforobservedclustersintheACCEPTcatalogue(Cavagnoloetal.2009,theredlinesarefornon-cool-coreclustersandtheblueareforcool-coreclusters). Eachrowshowstheprofilesofadifferentfeedbackmode.Thelastrowshowstheprofilesforthequasarmode,withEAGN=1060ergperevent(onlythetwo mostmassiveclustersareshown).

Description:
cosmic rays, as well as their effects on the thermal plasma. By comparing both the fusion (e.g. Hanasz & Lesch 2000; Jubelgas et al. 2008), should . Brown S., Emerick A., Rudnick L., Brunetti G., 2011, ApJ, 740, L28. Brüggen M.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.