Mon.Not.R.Astron.Soc.000,000–000(0000) Printed3September2009 (MNLATEXstylefilev2.2) Simulating the effect of AGN feedback on the metal enrichment of galaxy clusters 9 D. Fabjan1,2,3, S. Borgani1,2,3, L. Tornatore1,2,3, A. Saro1,2,3, G. Murante4 & K. Dolag5 0 0 2 1DipartimentodiAstronomiadell’Universita` diTrieste,viaTiepolo11,I-34131Trieste,Italy(fabjan,borgani,tornatore,[email protected]) 2INAF–IstitutoNazionalediAstrofisica,viaTiepolo11,I-34131Trieste,Italy p 3INFN–IstitutoNazionalediFisicaNucleare,Trieste,Italy e 4INAF–IstitutoNazionalediAstrofisica–OsservatorioAstronomicodiTorino,Str.Osservatorio25,I-10025,PinoTorinese,Torino,Italy([email protected]) S 5Max-Planck-Institutfu¨rAstrophysik,Karl-SchwarzschildStrasse1,GarchingbeiMu¨nchen,Germany([email protected]) 3 ] O 3September2009 C . h ABSTRACT p We presentastudyoftheeffectofAGNfeedbackonmetalenrichmentandthermalproper- - o tiesoftheintraclustermedium(ICM)inhydrodynamicalsimulationsofgalaxyclusters.The r simulationsare performedusing a version of the TreePM-SPH GADGET-2code, which also t s followschemo-dynamicalevolutionby accountingfor metalenrichmentcontributedbydif- a ferentstellarpopulations.Wecarryoutcosmologicalsimulationsforasetofgalaxyclusters, [ coveringthemassrangeM200 ≃(0.1−2.2)×1015h−1M⊙.Besidesrunsnotincludingany 1 efficientformofenergyfeedback,wecarryoutsimulationsincludingthreedifferentfeedback v schemes:(i)kineticfeedbackintheformofgalacticwindstriggeredbysupernovaexplosions; 4 (ii)AGNfeedbackfromgasaccretionontosuper-massiveblackholes(BHs);(iii)AGNfeed- 6 backinwhicha ’radiomode’isincludedwithanefficientthermalcouplingoftheextracted 6 energy,wheneverBHsenterinaquiescentaccretionphase.Besidesinvestigatingtheresulting 0 thermalpropertiesoftheICM,weanalyseindetailtheeffectthatthesefeedbackmodelshave . 9 ontheICMmetal-enrichment.WefindthatAGNfeedbackhasthedesiredeffectofquenching 0 starformationinthebrightestclustergalaxiesatz <4andprovidescorrecttemperaturepro- 9 filesinthecentralregionsofgalaxygroups.However,itseffectisnotyetsufficienttocreate 0 “coolcores”inmassiveclusters,whilegeneratinganexcessofentropyincentralregionsof : v galaxygroups.Asforthepatternofmetaldistribution,AGN feedbackcreatesa widespread i enrichmentintheoutskirtsofclusters,thankstoitsefficiencyindisplacingenrichedgasfrom X galactichalostothe inter-galacticmedium.ThisturnsintoprofilesofIronabundance,ZFe, r whichareinbetteragreementwithobservationalresults,andintoamorepristineenrichment a oftheICMaroundandbeyondtheclustervirialregions.Followingthepatternoftherelative abundancesofSiliconandIron,weconcludethatasignificantfractionoftheICMenrichment is contributed in simulations by a diffuse population of intra-cluster stars. Our simulations alsopredictthatprofilesoftheZSi/ZFeabundanceratiodonotincreaseatincreasingradii,at leastoutto0.5Rvir.Ourresultsclearlyshowthatdifferentsourcesofenergyfeedbackleave distinct imprints in the enrichment pattern of the ICM. They further demonstrate that such imprints are more evident when looking at external regions, approaching the cluster virial boundaries. Key words: Cosmology: Theory – Methods: Numerical – X–Rays: Galaxies: Clusters – Galaxies:Abundances–Galaxies:IntergalacticMedium 1 INTRODUCTION gions of relaxed clusters show little evidence of gas cooler than about a third of the virial temperature (e.g., Petersonetal. 2001; High quality data from the current generation of X–ray satel- Bo¨hringeretal.2002;Sandersonetal.2006);temperatureprofiles lites(XMM-Newton, Chandra andSuzaku) have now established have negative gradients outside core regions, a trend that ex- anumberofobservationalfactsconcerningthethermo-dynamical tendsout tothelargestradii coveredsofar byobservations (e.g., andchemo-dynamicalpropertiesoftheintra-clustermedium(ICM) DeGrandi&Molendi 2002; Vikhlininetal. 2005; Zhangetal. for statistically representative sets of galaxy clusters: core re- 2006; Baldietal. 2007; Prattetal. 2007; Leccardi&Molendi 2 Fabjanet al. 2008b); gas entropy is higher than expected from simple self- print of the conversion of mechanical energy associated to AGN similarscalingpropertiesoftheICM,notonlyincoreregions,but jets into thermal energy (and possibly in a non–thermal content alsoouttoR5001(e.g.,Sunetal.2009,andreferencestherein);ra- ofrelativisticparticles)throughshocks(e.g.,Mazzottaetal.2004; dialprofilesoftheIronabundanceshownegativegradients,more Fabianetal.2005;McNamaraetal.2006;Sanders&Fabian2007; pronounced for relaxed “cool core” clusters, with central values seeMcNamara&Nulsen2007 forareview).Although analytical of ZFeapproaching solar abundance and with a global enrich- argumentsconvincinglyshowthattheenergyradiatedfromgasac- mentatalevelofabout1/3–1/2ZFe,⊙(e.g.,DeGrandietal.2004; cretionontoacentralsuper-massiveblackhole(BH)isenoughto Vikhlininetal. 2005; dePlaaetal. 2006; Snowdenetal. 2008; suppress gas cooling, it is all but clear how this energy is ther- Leccardi&Molendi 2008a; see Mushotzky 2004; Werneretal. malisedanddistributedinthesurrounding medium. Alikelysce- 2008forrecentreviews). nario is that bubbles of high entropy gas are created at the ter- Galaxy clusters arise from the collapse of density perturba- minationofjets.Buoyancy ofthesebubblesintheICMthendis- tions over a scale of ∼ 10h−1 comoving Mpc. As such, the tributesthermalenergyoverlargerscales(e.g.,DallaVecchiaetal. aboveobservationalpropertiesoftheICMcomefromanon-trivial 2004; Cattaneoetal. 2007). Crucial in this process is the stabil- interplay between the underlying cosmological scenario, which ityofthebubblesagainstgasdynamicalinstabilitieswhichwould shapes thelarge-scalestructureof theUniverse, andanumber of tendtodestroythemquiterapidly.Indeed,detailednumericalsim- astrophysical processes (e.g., star formation, energy and chemi- ulations have demonstrated that gas circulation associated to jets cal feedback from supernovae and AGN) taking place on much (e.g., Ommaetal. 2004; Brighenti&Mathews 2006; Heinzetal. smaller scales. Cosmological hydrodynamical simulations repre- 2006),gasviscosity,magneticfields(e.g.,Ruszkowskietal.2007) sent the modern instrument withwhichsuch complexities can be andinjectionofcosmicrays(e.g.,Ruszkowskietal.2008)areall described as the result of hierarchical assembly of cosmic struc- expectedtocooperateindeterminingtheevolutionofbuoyantbub- tures(e.g.,Borgani&Kravtsov2009,forarecentreview).Indeed, bles.Although highly instructive, allthesesimulationshavebeen simulationsofgalaxyclustersreachnowadaysahighenoughres- performedforisolatedhalosand,assuch,theydescribeneitherthe olution, whileincluding a realisticdescription of theabove men- cosmologicalgrowthandmergingofblackholes,northehierarchi- tioned astrophysical processes, for them to provide a coherent calassemblyofgalaxyclusters. interpretative framework for X-ray observations. Quite remark- Springeletal. (2005) and DiMatteoetal. (2005) have pre- ably, simulation predictions for the thermodynamical properties sentedamodelthatfollowsinacosmologicalsimulationtheevolu- of the ICM are in good agreement with observations, at least tionoftheBHpopulationandtheeffectofenergyfeedbackresult- outside the core regions: simulated profiles of gas density and ingfromgasaccretionontoBHs.Inthismodel,BHsaretreatedas temperature match the observed ones at cluster-centric distances sinkparticleswhichaccretefromthesurroundinggasaccordingto ∼>0.1R500 (e.g.,Lokenetal.2002;Borganietal.2004;Kayetal. aBondiaccretionrate(e.g.,Bondi1952),withanupperlimitpro- 2004; Roncarellietal. 2006; Prattetal. 2007; Nagaietal. 2007; videdbytheEddingtonrate(seealsoBooth&Schaye2009).This Crostonetal.2008);theobservedentropylevelatR500iswellre- modelwasusedbyBhattacharyaetal.(2008),whoruncosmolog- produced by simulations including radiative cooling and star for- ical simulations to study the effect of AGN feedback on galaxy mation (e.g., Nagaietal. 2007; Borgani&Viel 2009). The situa- groups.Theyfoundthatthisfeedbackiseffectiveinreducingstar tion is quite different within cluster cores, where simulations in- formationincentralregionsandtodisplacegastowardsouterre- cluding only stellar feedback generally fail at producing realistic gions.However,theresultingentropylevelwithinclustercoresis coolcores.Thetwoclearestmanifestationsofthisfailurearerep- higher thanobserved. Initsoriginalversion, theenergy extracted resented by the behaviour of temperature and entropy profiles at fromBHaccretionislocallydistributedtothegasaroundtheBHs small radii, ∼<0.1R500. Observations of cool core clusters show accordingtoaSPH-kernelweightingscheme.Thismodelhasbeen thattemperatureprofilessmoothlydeclinetowardthecentre,reach- modifiedbySijacki&Springel(2006)whoincludedthepossibil- ingtemperaturesofabout1/3–1/2ofthemaximumvalue,whilethe itytoinflatehigh-entropybubblesintheICMwheneveraccretion entropylevelatthesmallestsampledscalesisgenerallyverylow ontothecentralBHentersinaquiescent“radiomode”.Theunder- (e.g.,Sunetal.2009;Sandersonetal.2009a).Onthecontrary,ra- lying idea of injecting bubbles with this prescription was to pro- diativesimulationsincludingavarietyofmodelsofstellarfeedback vide a more realistic description of the effect of jet termination predictsteepnegativetemperaturegradientsdowntotheinnermost ontheICM,althoughtheeffectofthejetitselfwasnotincluded. resolvedradiiandcentralentropylevelsmuchhigherthanobserved Puchweinetal. (2008) simulated a set of clusters and groups to (e.g., Valdarnini 2003; Tornatoreetal. 2003; Borganietal. 2004; showthatthisfeedbackschemeisabletoreproducetheobserved Nagaietal.2007;cf.alsoKayetal.2007).Thisfailureofsimula- slope of the relation between X–ray luminosity and temperature. tions is generally interpreted as due to overcooling, which takes Sijackietal.(2007)showedinasimulationofasinglepoorgalaxy place in simulated clusters even when including an efficient su- clusterthattheinjectionof bubblesisquiteeffectiveinsuppress- pernova(SN)feedback,andcausesanexcessofstarformationin ingstarformationincentralregions.However,alsointhiscasethe thesimulatedbrightestclustergalaxies(BCGs;Romeoetal.2005; temperatureprofileinthecoreregionsdoesnotmatchtheobserved Saroetal.2006). slope,whilemorerealistictemperatureprofilescanbeproducedif Thegenerallyacceptedsolutiontotheseshortcomingsofsim- bubbleinjectionisassociatedtotheinjectionofanon–thermalpop- ulationsisrepresentedbyAGNfeedback.Indeed,thepresenceof ulationofrelativisticparticles(Pfrommeretal.2007;Sijackietal. cavitiesintheICMattheclustercentreisconsideredasthefinger- 2008).Althoughtheseauthorspresentedresultsconcerningtheef- fectofAGNontheICMthermodynamics,nodetailedanalysishas been so far carried out to study the interplay between AGN and 1 HereandinthefollowingwewillindicatewithR∆ theradiusencom- chemicalenrichment,byincludingadetailedchemo-dynamicalde- passingameandensityequalto∆ρcr,beingρcr =3H2/8πGthecritical scriptionoftheICM. cosmicdensity.Inasimilarway,wewillindicatewithM∆themasscon- Indeed,theamountanddistributionofmetalsintheICMpro- tainedwithinR∆. videanimportantdiagnostictoreconstructthepasthistoryofstar AGNfeedbackand metalenrichmentofclusters 3 formationandtheroleofgas-dynamicalandfeedbackprocessesin Table1.Characteristicsofthesimulatedclusters.Column1:clustername. displacingmetalsfromstarformingregions(e.g.,Mushotzky2004; Column 2: mass contained within R200 (units of 1014h−1M⊙). Col- Borganietal.2008;Schindler&Diaferio2008,forreviews).Cos- umn 3: value of R200 (units of h−1Mpc). Column 4: value of the mological chemo-dynamical simulations of galaxy clusters gen- spectroscopic-liketemperaturewithinR500,T500.Columns5and6:mass erally show that the predicted profiles of Iron abundance are ofthecentralBHhostedintheBCGfortheAGN1andAGN2runs(unitsof steeperthanobserved(e.g.,Valdarnini2003;Tornatoreetal.2004; 1010h−1M⊙),respectively.Fortheg51cluster,thetwoadditionalvalues Romeoetal. 2005; Tornatoreetal. 2007; Dave´etal. 2008), with reportedforthemassofthecentralBHrefertotheAGN2(0.8)andAGN2W runs(seetext)andareindicatedwiththe†and‡symbol,respectively. anexcessofenrichmentinthecoreregions.Thisisgenerallyinter- pretedasduetothesameexcessofrecentstarformationinsimu- Cluster M200 R200 T500 MBH latedBCGs.However, Fabjanetal.(2008)showed thatanexcess AGN1 AGN2 of recent star formation has also the effect of efficiently locking g1.a 12.69 1.76 9.71 56.22 15.74 recently-produced metals into stars, thereby preventing a too fast g1.b 3.64 1.16 3.39 7.96 2.52 increaseofthemetallicityofthehotdiffusemedium. g1.c 1.36 0.84 2.03 4.63 1.32 All the above analyses are based on different implementa- g1.d 1.04 0.76 1.82 1.59 0.46 tionsofSNenergyfeedback, whileonlymuchlessdetailedanal- g1.e 0.62 0.64 1.44 0.84 0.30 yses of ICM metal enrichment have been so far presented by also including AGN feedback in cosmological simulations (e.g., g8.a 18.51 2.00 12.82 113.50 27.73 Sijackietal.2007;Molletal.2007).Forinstance, Roedigeretal. g8.b 0.65 0.65 1.41 2.58 0.88 (2007)showedfromsimulationsofisolatedhalosthatbuoyancyof g8.c 0.52 0.61 1.42 1.74 0.63 bubblescanactuallydisplacealargeamountofthecentralhighly g51 10.95 1.68 7.55 43.38 16.50 enriched ICM, thus leading toa radical change of the metallicity 6.18† profiles,oreventoadisruptionofthemetallicitygradients. 15.02‡ Theaimofthispaperistopresentadetailedanalysisofcos- g72.a 10.57 1.66 8.67 48.74 13.87 mological hydrodynamical simulations of galaxy clusters, which g72.b 1.48 0.86 2.18 3.79 1.34 havebeencarriedoutwiththeGADGET-2code(Springel2005),by combining the AGN feedback model described by Springeletal. g676 0.87 0.72 1.91 3.03 0.95 (2005)withtheSPHimplementationofchemo-dynamicspresented g914 0.88 0.72 2.03 2.89 1.13 byTornatoreetal.(2007).Besidesshowingresultsontheeffectof g1542 0.83 0.71 1.79 2.54 0.79 combining metallicity–dependent cooling and AGN feedback on g3344 0.92 0.74 2.03 3.15 1.29 g6212 0.89 0.73 2.02 2.57 0.99 theICMthermodynamics,wewillfocusourdiscussiononthedif- ferent effects that SNe and AGN feedback have on the chemical enrichmentoftheICM.Althoughwewillshortlydiscusstheeffect Table2.Resolutionofthedifferentruns.Column2-3:massoftheDMpar- ofdifferentfeedbacksourcesonthepatternofstarformationinthe ticlesandinitialmassofthegasparticles(unitsof108h−1M⊙).Column BCG, the results presented in this paper will mainly concern the 4:valueofthePlummer-equivalentgravitationalforcesofteningatz = 0 enrichmentofthehotdiffuseintra-clustergas.Wedefertoaforth- (unitsof h−1kpc).Column5:clusterssimulatedateachresolution. comingpaperadetailedanalysisoftheeffectofAGNfeedbackon MDM mgas ε Clusters the population of cluster galaxies. The scheme of the paper is as follows.WepresentinSection2thesimulatedclusters,andbriefly Lowres. 11.0 2.03 5.00 g1,g8,g51,g72,g676 describe the implementation of the chemical evolution model in Highres. 1.69 0.31 2.75 g676,g914,g1542,g3344, g6212 the GADGET-2codealongwiththeSNandAGNfeedbackmod- els.InSection3weshowourresultsonthethermalpropertiesof theICMandtheir comparison withobservational results.Section 18clusterswithM200∼>5×1013h−1M⊙,whosebasiccharacter- 4isdevoted tothepresentation of theresultsontheICMchemi- isticsarereportedinTable1. calenrichmentfromoursimulationsandtheircomparisonwiththe Eachregionwasre-simulatedusingthezoomedinitialcondi- mostrecentobservations.Wediscussourresultsanddrawourmain tion(ZIC)techniquebyTormenetal.(1997).Therelativemassof conclusionsinSection5. thegasanddarkmatter(DM)particleswithinthehigh–resolution regionofeachsimulationissetsothatΩ = 0.045fortheden- bar sityparametercontributedbybaryons.ForeachLagrangianregion, initialconditionshavebeengeneratedatabasicresolutionandat 2 THESIMULATIONS 6.5timeshighermassresolution.Atthishigherresolution,themass 2.1 Thesetofsimulatedclusters mofgtahseD=M2.a8nd×g1as07phar−ti1cMles⊙arreesmpeDctMive≃ly,1w.9it×h P1l0u8mhm−e1rMeq⊙uiavnad- Our set of simulated clusters is obtained from nine Lagrangian lentsofteninglengthforthecomputationofthegravitationalforce regions extracted from a simulation containing only dark matter set to ε = 2.75h−1kpc in physical units below z = 2 and (DM)withaboxsizeof479h−1 Mpc(Yoshidaetal.2001),per- to ε = 8.25h−1kpc in comoving units at higher redshifts. In formed for aflat Λcold darkmatter (CDM) cosmological model the lower resolution runs, these values are rescaled according to with relevant parameters Ωm = 0.3, h100 = 0.7, σ8 = 0.9. m−D1M/3.WedecidedtosimulatethenineLagrangianregionswith ThissetofsimulatedclustersisdescribedindetailbyDolagetal. massiveclustersatthelowerresolution,whileweusedthehigher (2008).Theregionsarecentredonninegalaxyclusters:fiveofthem resolutionforthefivelow-massclusters.Onlythelow-massg676 havemassesM200 ≃1.0×1014h−1M⊙,whiletheremainingfour clusterwassimulatedatbothresolutions.WeprovideinTable2a haveM200 ≃ (1.0−2.2)×1015h−1M⊙. Theregionsof three descriptionoftheparametersforthetwodifferentresolutions,also largestobjectsalsocontainotherclusters,sothatwehaveintotal listingtheclusterssimulatedateachresolution. 4 Fabjanet al. 2.2 Thesimulationcode centralBHofmass105h−1M⊙,providedthehalodoesnotcon- tainanyBHyet.Fortherunsatthelowerresolutionthevalueof OGuArDGsEimT-u2latcioodnes (wSperreingepler2fo0r0m5e).d AullsinsgimutlhaetionTsreeinPcMlu-dSePHa thehalomassassumedtoseedBHsisMth = 5×1010h−1M⊙, so that it is resolved with about 40 DM particles. At the higher metallicity-dependent radiative cooling (Sutherland&Dopita resolution the halo mass threshold for BH seeding decreases to 1g9ro9u3n)d, he(Hatainagrdtfr&omMaaduaunifo1r9m96t)imaen-ddeptheendeenfftecutlitvraevimoloetdeblacbky- Mth = 1010h−1M⊙, so astoresolve it withapproximately the same number of particles. We verified that using in the higher- Springel&Hernquist(2003)forthedescriptionofstarformation. resolutionrunsthesamevalueofM asinthelow-resolutionruns In this model, gas particles above a given density are treated as th causes theeffect of BHaccretion tobeshiftedtoward lower red- multiphase, so as to provide a sub–resolution description of the shift, since its onset has to await the formation of more massive inter-stellar medium. In the following, we assume the density halos, whileleavingthefinalpropertiesof galaxyclustersalmost thresholdfortheonsetofstarformationinmultiphasegasparticles unaffected. to be n = 0.1cm−3 in terms of number density of hydrogen H Onceseeded,eachBHcanthengrowbylocalgasaccretion, atoms.Oursimulationsalsoincludeadetailedmodelofchemical witharategivenby evolution by Tornatoreetal. (2007, T07 hereafter). We address the reader to T07 for a more detailed description of this model. M˙BH = min(cid:0)M˙B,M˙Edd(cid:1) (1) Metals are produced by SNe-II, SNe-Ia and intermediate and low-mass stars in the asymptotic giant branch (AGB hereafter). orbymergingwithotherBHs.HereM˙B istheaccretionrateesti- We assume SNe-II to arise from stars having mass above 8M⊙. matedwiththeBondi-Hoyle-Lyttletonformula(Hoyle&Lyttleton As for the SNe-Ia, we assume their progenitors to be binary 1939;Bondi&Hoyle1944;Bondi1952),whileM˙Edd istheEd- systems, whose total mass lies in the range (3–16)M⊙. Metals dington rate. The latter is inversely proportional to the radiative and energy are released by stars of different mass by properly efficiencyǫr,whichgivestheradiatedenergyinunitsoftheenergy accounting for mass–dependent lifetimes. In this work we as- associatedtotheaccretedmass: sume the lifetime function proposed by Padovani&Matteucci L ǫ = r . (2) (1993),whileweassumethestandardstellarinitialmassfunction r M˙ c2 BH (IMF) by Salpeter (1955). We adopt the metallicity–dependent FollowingSpringeletal. (2005), weuse ǫ = 0.1 as areference stellar yieldsby Woosley&Weaver (1995) for SNe-II, theyields r value, which istypical for a radiatively efficient accretion onto a by vandenHoek&Groenewegen (1997) for the AGB and by Schwartzschild BH (Shakura&Syunyaev 1973). The model then Thielemannetal.(2003)forSNe-Ia.Theversionofthecodeused assumes that a fraction ǫ of the radiated energy is thermally forthesimulationspresentedhereallowedustofollowH,He,C, f O,Mg,S,SiandFe.Onceproducedbyastarparticle,metalsare coupled to the surrounding gas, so that E˙feed = ǫrǫfM˙BHc2 is the rate of provided energy feedback. In standard AGN feed- thenspreadtothesurroundinggasparticlesbyusingtheB-spline backimplementationweuseǫ =0.05followingDiMatteoetal. kernelwithweightscomputedover64neighboursandtakentobe f (2005), who were able with this value to reproduce the observed proportional to the volume of each particle (see T07 for detailed M −σrelationbetweenbulgevelocitydispersionandmassof tests on the effect of choosing different weighting schemes to BH thehostedBH(e.g.,Magorrianetal.1998).Thischoicewasalso spreadmetals). found to be consistent with the value required in semi-analytical modelstoexplaintheevolution ofthenumber densityof quasars (Wyithe&Loeb2003). 2.3 Feedbackmodels GasswallowedbytheBHisimplementedinastochasticway, Inthesimulationspresentedinthispaperwemodeltwodifferent by assigning to each neighbour gas particle a probability of con- sources of energy feedback. The first one is the kinetic feedback tributingtotheaccretion,whichisproportionaltotheSPHkernel modelimplementedbySpringel&Hernquist(2003),inwhichen- weightcomputedattheparticleposition.DifferentlyfromSDH05, ergy released by SN-II triggers galactic winds, with mass up- weassumethateachselectedgasparticlecontributestotheaccre- load rate assumed to be proportional to the star formation rate, tionwith1/3ofitsmass,insteadof beingcompletelyswallowed. M˙ = ηM˙ .Therefore,fixingtheparameterη andthewindve- Inthisway,alargernumberofparticlescontributetotheaccretion, W ⋆ locity v amounts to fix the total energy carried by the winds. whichisthenfollowedinamorecontinuousway.Weremindthat W Inthefollowing,weassumeη =2forthemass-uploadparameter intheSDH05scheme,thisstochasticaccretionisusedonlytoin- andv =500kms−1forthewindvelocity.IfeachSN-IIreleases creasethedynamicmassoftheBHs,whiletheirmassenteringin W 1051ergs,theabovechoiceofparameterscorrespondstoassuming the computation of the accretion rateis followed in acontinuous thatSNe-IIpowergalacticoutflowswithnearlyunityefficiencyfor way,byusingtheanalyticexpressionforM˙ .Oncetheamount BH aSalpeterIMF(seeSpringel&Hernquist2003). of energy to be thermalised is computed for each BH at a given Furthermore, we include in our simulations the effect of time-step,onehastodistributethisenergytothesurroundinggas feedback energy released by gas accretion onto super-massive particles.Intheiroriginalformulation,SDH05distributedthisen- blackholes(BHs),followingtheschemeoriginallyintroducedby ergyusingtheSPHkernel. Springeletal. (2005, SDH05 hereafter; see also DiMatteoetal. Besides following thisstandard implementation of the AGN 2005,2008;Booth&Schaye2009).WerefertoSDH05foramore feedback, wealsofollow analternativeprescription, whichdiffer detailed description. In thismodel, BHsare represented by colli- fromtheoriginaloneintwoaspects. sionlesssinkparticlesofinitiallyverysmallmass,thatareallowed Firstly,followingSijackietal.(2007),weassumethatatran- tosubsequently grow viagasaccretionandthroughmergerswith sitionfroma“quasar”phaseto“radio”modeoftheBHfeedback otherBHsduringcloseencounters.Duringthegrowthofstructures, takes place whenever the accretion rate falls below a given limit weseedeverynewdarkmatterhaloaboveacertainmassthreshold (e.g.,Churazovetal.2005,andreferencestherein),corresponding M ,identifiedbyarun-timefriends-of-friendsalgorithm,witha toM˙ /M˙ < 10−2,whichimpliesanincrease of thefeed- th BH Edd AGNfeedbackand metalenrichmentofclusters 5 backefficiencytoǫ =0.2.AthighredshiftBHsarecharacterised (d) Modified version of the AGN feedback (AGN2 hereafter), f byhighaccretionratesandpowerveryluminousquasars,withonly withfeedback efficiencyincreasingfromǫ = 0.05toǫ = 0.2 f f asmallfractionoftheradiatedenergybeingthermallycoupledto whenM˙ /M˙ <10−2,anddistributionofenergyaroundthe BH Edd the surrounding gas. On the other hand, BHs hosted within very BHwithatop-hatkernel. massive halos at lower redshift are expected to accrete at a rate wellbelowtheEddingtonlimit,whiletheenergyismostlyreleased Inordertofurtherexploretheparameterspaceoftheconsid- in a kinetic form, eventually thermalised in the surrounding gas ered feedback models we also carried out one simulation of the throughshocks.Secondly,insteadofdistributingtheenergyusing g51clusterbasedontheAGN2scheme,butwiththefeedbackeffi- aSPHkernel,wedistributeitusingatop-hatkernel,havingradius ciencyincreasedtoǫ =0.8(AGN2(0.8)hereafter).Wealsonote f givenbytheSPHsmoothinglength.Inordertoavoidspreadingthe that our simulations include either winds triggered by SN explo- energywithinatoosmall sphere, weassume aminimum spread- sionsorAGNfeedback.Whilethischoiceisdonewiththepurpose ing length of 2h−1kpc. The rationale behind the choice of the ofseparatetheeffectsofthesetwofeedbacksources,weexpectin top-hatkernelistoprovideamoreuniformdistributionofenergy, realisticcases that both AGN and SN feedback should cooperate thus mimicking the effect of inflating bubbles in correspondence in determining the star formation history of galaxies. In order to of thetermination of the AGN jets. It iswell known that a num- verify the effect of combining the two feedback sources, we car- berofphysicalprocessesneedtobeadequatelyincludedforafully riedout one simulation of theAGN2 scheme for the g51 cluster, self-consistentdescriptionofbubbleinjectionandbuoyancy: gas- inwhichalsogalacticwindswithavelocityv =300kms−1are w dynamicaleffectsrelatedtojets,magneticfields,viscosity,thermal included(AGN2Whereafter). conduction,injectionofrelativisticparticles. Beforestartingthepresentationoftheresultsonthethermal Inparticular,anumberofstudiesbasedonsimulationsofiso- andenrichmentpropertiesoftheICM,webrieflycommentonthe lated halos (e.g., Ommaetal. 2004; Brighenti&Mathews 2006) resultsconcerningthemassofthecentralblackholesinthesimu- havepointedoutthatgascirculationgeneratedbyjetsprovidesan lationsincludingAGNfeedback,andthestarformationrate(SFR) important contribution for the stabilization of cooling flows (see historyofthebrightestclustergalaxies(BCGs). alsoHeinzetal.2006,foracosmologicalsimulationofclusterfor- Looking at Table 1, we note that our simulations predict mationincludingjets).Initscurrentimplementation,themodelof ratherlargemassesforthesuper-massiveBHshostedinthecentral BHfeedbackincludedinoursimulationsneglectsanykineticfeed- galaxies.Quiteinterestingly,theAGN2runsgenerallyproduceBH backassociatedtojets.Basedonananalyticalmodel,Pope(2009) masseswhicharesmaller,byafactor3–5,thanfortheAGN1runs. computed the typical scale of transition from kinetic to thermal Thisdemonstrates that including themoreefficient “radiomode” feedback regime for AGN in elliptical galaxies and clusters. As for the feedback and distributing the energy in a more uniform aresult,hefoundthattheeffectofmomentumcarriedbyjetscan wayhasasignificant effect inregulating gasaccretion. Although beneglectedonscales∼>20kpc,theexactvaluedependingonthe BCGs are known to host BHs which are more massive than ex- localconditionsofthegasandontheinjectionrateofkineticand pected for normal early type galaxies of comparable mass (e.g., thermal energy. In order to compare such a scale to that actually Laueretal.2007),theBHmassesfromour simulationsseemex- resolvedinoursimulations,weremindthereaderthatSPHhydro- ceedingly large, also for the AGN2 model. For instance, the BH dynamicsisnumericallyconvergedonscalesabout6timeslarger mass hosted by M87, within the relatively poor Virgo cluster, is thanthePlummer–equivalentsofteningscaleforgravitationalforce mBH ≃ (3 −4) ×109M⊙ (e.g., Raffertyetal. 2006). This is (e.g.,Borganietal.2002).Owingtothevaluesofthegravitational aboutafactor3–5smallerthanfoundinourAGN2runsforclus- softeningreportedinTable2,thescalesresolvedinoursimulations tersofcomparablemass,M200 ≃2×1014h−1M⊙.Wenotethat arenotintheregimewherekineticfeedbackisexpectedtodomi- increasingtheradio-modefeedbackefficiencyfromǫ =0.2to0.8 f nate,thusjustifyingtheadoptionofapurelythermalfeedback.As (AGN2(0.8)run)reducesthemassofthecentralblackholeinthe afurther test,wehave computed theradius of thetop–hat kernel g51clusterfrommBH ≃16.5×109M⊙to≃6.2×109M⊙(see within which energy is distributed around the central BHsin our Table1).Thissuggeststhatahighlyefficientthermalizationofthe simulatedclusters.Asafewexamples,wefoundatz = 0thisra- energyextractedfromtheBHisrequiredtoregulategasaccretion diustobe21kpcand23kpcfortheAGN2runsoftheg72andg676 totheobservedlevel.Quiteinterestingly,wealsonotethatadding clusters,respectively. Thisimpliesthat weareinfactdistributing the effect of galactic winds in the AGN2 scheme (AGN2W run) thermal energy over scales where kinetic feedback should not be onlyprovidesamarginalreductionofthefinalmassofthecentral dominant. BH.Therefore,althoughgalacticwindscanplayasignificantrole Inviewofthedifficultyofself–consistentlyincludingtheco- in regulating star formation within relatively small galaxies, they operative effect of all the physical processes we listedabove, we arenotefficientindecreasinggasdensityaroundthelargestBHs, preferheretofollowarathersimplifiedapproachandseetowhat soastosuppresstheiraccretionrate. extent[resultson]thefinalresultsofoursimulationsaresensitive As for the comparison with previous analyses, Sijackietal. tovariationsintheimplementationoftheBHfeedbackmodel. (2007)performed asimulationofthesameg676 cluster included Insummary,weperformedfourseriesofruns,corresponding in our simulation set at the lower resolution, for their feedback toasmanyprescriptionforenergyfeedback. schemebasedontheinjectionofAGNdrivenbubbles.Theyfound thatthefinalmassofthecentralBHismBH ≃6×109h−1M⊙. (a) Nofeedback(NFhereafter):neithergalacticwindsnorAGN This value is about 30 per cent lower than what we find for the feedbackisincluded. AGN2 runs of g676. In order to compare more closely with the (b) Galacticwinds(Whereafter)includedbyfollowingthemodel result by Sijackietal. (2007), we repeated the run of the g676 bySpringel&Hernquist(2003),withv =500kms−1andη=2 cluster by switching off the metallicity-dependence of the cool- w forthewindmassupload. ing function. As a result, BH accretion proceeds in a less effi- (c) Standard implementation of AGN feedback from BH accre- cientway,andtheresultingcentralBHmassinthiscasedropsto tion(AGN1hereafter),usingǫf =0.05forthefeedbackefficiency. ≃3.5×109h−1M⊙. 6 Fabjanet al. AsfortheSFRhistoryoftheBCG,thisisestimatedbyidenti- DMparticles,usingalinkinglengthof0.15intermsofthemean fyingfirstallthestarsbelongingtotheBCGatz=0.Thishasbeen inter-particleseparation. Withineach FOFgroup, weidentify the donebyrunningtheSKIDgroup-findingalgorithm(Stadel2001), DMparticlehavingtheminimumvalueofthegravitationalpoten- usingthesameproceduredescribedbySaroetal.(2006).Aftertag- tialandtakeitspositiontocorrespondtothecentreofthecluster. gingallthestarparticlesbelongingtotheBCG,wereconstructthe Allprofilesarethencomputedstartingfromthiscentre.Thesmall- SFRhistorybyusingtheinformationontheredshiftatwhicheach estradiusthatweusetocomputeprofilesencompassesaminimum starparticlehasbeenspawnedbyaparentgasparticle,according number of 100 SPH particles, a criterion that gives numerically tothestochasticalgorithmofstarformationimplementedintheef- converged results for profiles of gas density and temperature in fectivemodelbySpringel&Hernquist(2003).TheresultingSFR non–radiative simulations (e.g., Borganietal. 2002). As we shall historiesfortheBCGoftheg51clustersareshownintheleftpanel seeinthefollowing,profilescomputedwiththiscriterionwillex- ofFigure1. tendtosmallerradiiforthoserunswhichhaveahighergasdensity Asexpected,therunwithnoefficientfeedback(NF)produces atthecentre,whilestoppingatrelativelylargerradiifortherunsin- thehigheststarformationatallepochs:SFRhasapeakatz ≃ 4 cludingAGNfeedback,whicharecharacterisedbyalowercentral andthendropsasaconsequenceoftheexhaustionofgaswithshort gasdensity. cooling time. Despite this fast reduction of the SFR, its level at z = 0 is still rather large, ≃ 400h−1M⊙ yr−1. As for the run 3.1 Theluminosity-temperaturerelation withwinds,ithasareducedSFRsincetheverybeginning,owing to the efficiency of thisfeedback scheme in suppressing star for- Therelationbetween bolometricX–rayluminosity, L andICM X mationwithinsmallgalaxieswhichformalreadyathighredshift. temperature, T, provided one of the first evidences that non– Alsointhiscase,thepeakofstarformationtakesplaceatz ∼ 4, gravitationaleffectsdeterminethethermo-dynamicalpropertiesof butwithanamplitudewhichisabout40percentlowerthanforthe the ICM (e.g., Voit 2005, for a review). Its slope at the scale of NFrunsandwithamoregentledeclineafterwards.Asfortheruns clustersisobservedtobeL ∝Tαwithα≃2.5–3(e.g.,Horner X withAGNfeedback,theirSFRhistoryisquitesimilartothatofthe 2001;Prattetal.2009),andpossiblyevensteeperorwithalarger Wrundowntoz ≃ 5.Starformationisthensuddenlyquenched scatter at the scale of galaxy groups (e.g., Osmond&Ponman at z∼<4. We note in general that SFR for the AGN1 model lies 2004).Theseresultsareatvariancewithrespecttotheprediction, slightlybelowthatoftheAGN2model,asaconsequenceofboth α = 2,ofself-similarmodelsbasedonlyontheeffectofgravita- thedifferentwayofdistributingenergyand,atlowredshift,ofthe tionalgasaccretion(e.g.,Kaiser1986).Attemptstoreproducethe inclusionoftheradiomodeassumedinthequiescentBHaccretion observed L –T relationwithhydrodynamic simulations of clus- X phase. Suppression of the SF at relatively low redshift is exactly tershavebeenpursuedbyseveralgroups(seeBorganietal.2008, thewelcomeeffectofAGNfeedback.Atz =0,theresultingSFR forarecentreview).Simulationsofgalaxyclustersincludingtheef- isofabout70M⊙yr−1forbothmodels,avaluewhichiscloserto, fectofstarformationandSNfeedbackintheformofenergy–driven althoughstillhigherthanthetypicalvaluesofSFRobservedinthe galactic winds produce results which are close to observations BCGsofclustersofcomparablemass.(e.g.,Raffertyetal.2006). at the scale of clusters, while generally producing too luminous Quiteinterestingly,increasingthefeedbackefficiencytoǫf =0.8 galaxygroups(e.g.,Borganietal.2004).Althoughacloseragree- intheAGN2(0.8)rundoesnotsignificantlyaffecttheleveloflow- ment withobservations atthescaleofgroups canbeobtained by redshiftstarformation,whileitsuppressesstarformationbyonly using SN–triggered momentum–driven winds (Dave´etal. 2008), ∼ 10 per cent around the peak of efficiency. Therefore, while a thereisageneralconsensusthatstellarfeedbackcannotreproduce higher efficiency is indeed effective in suppressing gas accretion atthesametimeboththeobservedL –T relationandthelowstar X onto thecentral BH,a further reduction of star formation associ- formationrateobservedincentralclustergalaxies.Puchweinetal. atedtotheBCGshouldrequireadifferentwayofthermalizingthe (2008)presented resultsontheL –T relationfor simulationsof X radiatedenergy.AsfortheAGN2Wrun,itsSFRathighredshiftis galaxy clusters which included the bubble–driven AGN feedback lowerthanfortheAGN2run,duetotheeffectofwinds,whileno scheme introduced by Sijackietal. (2007). They concluded that, significantchangeisobservedatz∼<3. while simulations not including any efficient feedback (in fact, IntherightpanelofFig.1weplotthestellarmassfoundinthe quitesimilartoourNFruns)produceoverluminousobjects,their BCG at z = 0that isformed before agiven redshift. According mechanismforAGNfeedbackisefficientinsuppressingtheX–ray tothisdefinition,thisquantityistheintegraloftheSFRplottedin luminosityofclustersandgroupsattheobservedlevel. therightpanel,computedfromagivenredshiftz toinfinity.This We present here our results on the L –T relation, keeping X plot clearly shows the different effect that winds and AGN feed- inmindthat our simulationsdiffer fromthose by Puchweinetal. backhaveinmaking theBGCstellarpopulation older.IntheNF (2008)bothinthedetailsoftheimplementationoftheAGNfeed- and W runs, theredshift at which 50per cent of theBCG stellar backschemeandinthetreatmentofthemetallicitydependenceof masswasalreadyinplaceisz50 ≃ 2.Thisindicatesthatevenan thecoolingfunction.IntheleftpanelofFigure2weshowthere- efficientSNfeedbackisnotabletomakethestellarpopulationof sultsforourrunsbasedonSNgalacticwinds(Wruns)andforthe theBCGolder.Onthecontrary,theeffectofAGNfeedbacktakes runs not including any efficient feedback (NF runs). Filled sym- placemostlyatrelativelylowreshift.Asaconsequencetheageof bols refer to the main halo of each simulated Lagrangian region, theBGCstellarpopulationincreases,withz50 ≃3.0−3.6,almost while open circles are for the “satellites”. Although we find sev- independentofthedetailoftheAGNfeedbackscheme. eralsatelliteshavingatemperaturecomparabletothoseofthelow- massmainhalos(seealsoTable1),weremindthatthesesatellites aredescribedwithamassresolutionsixtimeslowerthanthatofthe low-massmainhalos.WenotethattheNFrunsprovideaL –T re- 3 THERMALPROPERTIESOFTHEICM X lationwhichisnotfarfromtheobservedone,especiallyatthescale Galaxy clusters are identified in each simulation box by running ofgroups.Thereasonforthiscloseragreement,withrespecttothe firstafriends-of-friends(FOF)algorithmoverthehighresolution resultbyPuchweinetal.(2008)liesinthefact thattheseauthors AGNfeedbackand metalenrichmentofclusters 7 7000 AGN1 2.5e+13 NF AGN1 W NF 6000 AGN2 W AGN2W AGN2 AGN2(0.8) AGN2W 2e+13 AGN2(0.8) 5000 ]⊙ M s [ M/yrs]⊙ 4000 ar masse 1.5e+13 SFR [ 3000 ve Stell 1e+13 ati 2000 mul u C 5e+12 1000 0 0 0 2 4 6 8 10 0 2 4 6 8 10 redshift redshift Figure1.Leftpanel:thehistoryofstarformationrateofthebrightestclustergalaxy(BCG)intheg51clusterfortherunswithnofeedback(NF,blueshort- dashed),withgalacticwinds(W,greendashed),withstandardAGNfeedback(AGN1,reddot-dashed)andwithmodifiedAGNfeedback(AGN2,light-blue solid).AlsoshownaretheresultsfortheAGN2runwithhigherradio-modeBHfeedbackefficiency(AGN2(0.8),purplelong-dashed)andfortheAGN2run alsoincludinggalacticwinds(AGN2W,magentadotted).Rightpanel:thestellarmassfoundintheBCGatz=0thatwasformedbeforeagivenredshift. adoptedacoolingfunctioncomputedforzerometallicity.Includ- indeedabletoproducearealisticL –T relation,almostindepen- X ingthecontributionofmetallinestotheradiativelossesincreases dentofthedetailofhowtheenergyisthermalisedinthesurround- coolingefficiencyand,therefore,gasremovalfromthehotX–ray ingmedium. emittingphase.However,thepricetopayforthisreductionofX– rayluminosity isthat a far too largebaryon fraction isconverted intostarswithinclusters(seebelow).Quiteparadoxically,wealso 3.2 TheentropyoftheICM note than including an efficient feedback in the form of galactic Entropy level in central regions of clusters and groups isconsid- windsturnsintoanincreaseofX–rayluminosity.Thisisduetothe ered another fingerprint of the mechanisms which determine the fact that this feedback prevents a substantial amount of gas from thermodynamical history of the ICM. Early observational results cooling,withoutdisplacingitfromthecentralclusterregions,thus on the presence of entropy cores (e.g., Ponmanetal. 1999) have increasingtheamountofX-rayemittingICM. been more recently revised, in the light of higher quality data Inordertoassestheeffectofthedifferentresolutionusedfor from Chandra (e.g., Cavagnoloetal. 2009) and XMM–Newton largeand smallclusters, wecarried out runsof g676 at thesame (e.g.,Johnsonetal.2009)observations. Thesemorerecentanaly- lowerresolutionofthemassiveclusters.Theresultsareshownin sesshowthatentropylevelofclustersandgroupsatR500ishigher Fig. 2 with black diamonds connected with arrows to the corre- thanpredictedbynon–radiativesimulations,withentropyprofiles sponding higher resolution result. We note that resolution effects forrelaxedsystemscontinuouslydecreasingdowntothesmallest go in opposite directions for the NF and W runs. In fact, in the resolvedradii(seealsoSunetal.2009). absenceofwinds,therunawayofcoolingwithresolutionremoves Whileradiativesimulationsofgalaxyclustershavebeengen- fromthehotphasealargeramountofgas,thusdecreasingX–ray erally shown to reproduce observationalr results at R500 (e.g., luminosity. On the contrary, higher resolution allows a more ac- Nagaietal.2007;Dave´etal.2008),theygenerallypredicttoolow curatedescriptionofkineticfeedback,whichstartsheatinggasat entropylevelsatsmallerradii.Forinstance,Borgani&Viel(2009) higherredshift.Asaconsequence,radiativelossesarecompensated have shown that the entropy at R2500 can be reproduced by re- athigherresolutionforalargeramountofdiffusebaryons,which sortingtoafairlystrongpre-heatingatz = 4.However,thispre- remaininthehotphase,therebyincreasingtheX–rayluminosity. heatingmusttargetonlyrelativelyoverdenseregions,topreventthe As for the runs with AGN feedback, it is quite efficient in creationoftoolargevoidsinthestructureoftheLyman-αforestat decreasing X–ray luminosity at the scale of galaxy groups, thus z∼2(seealsoShangetal.2007). wellrecoveringtheobservedL –T relation.Thisconclusionholds In the following we use the standard definition of entropy, X forbothimplementationsoftheAGNfeedback,whichhaverather whichisusuallyadoptedinX–raystudiesofgalaxyclusters(e.g., smalldifferences. Therefore,althoughtheAGN2schemeismore Sunetal.2009): efficientinregulatingthegrowthofthecentralBHshostedwithin T∆ theBCGs,ithasamarginaleffectontheX–rayluminosity.These K∆ = n2/3 , (3) resultsareinlinewiththosefoundbyPuchweinetal.(2008),who e,∆ however usedtheinjectionofheatedbubblestodistributetheen- whereT∆andne,∆arethevaluesofgastemperatureandelectron ergyextractedfromtheBHaccretion.Thiswitnessesthatthefeed- numberdensitycomputedatR∆.Asforthetemperature,itiscom- backenergyassociatedtogasaccretionontosuper-massiveBHsis puted by following theprescription of spectroscopic liketemper- 8 Fabjanet al. 100 100 Horner 2001 Horner 2001 Arnaud&Evrard 99 Arnaud&Evrard 99 Osmond&Ponman 04 Osmond&Ponman 04 Pratt et al. 09 Pratt et al. 09 NF AGN-1 W AGN-2 s] 10 s] 10 g/ g/ er er 4 4 4 4 0 0 1 1 ) [0 1 ) [0 1 0 0 5 5 R R < < (ol (ol b b X, X, L 0.1 L 0.1 0.01 0.01 1 10 1 10 T (< R ) [keV] T (< R ) [keV] sl 500 sl 500 Figure2.RelationbetweenX–rayluminosityandtemperatureforsimulated(colouredsymbols)andobserved(greypointswitherrorbars)clustersandgroups. ObservationaldatapointsarefromArnaud&Evrard(1999)(greydiamonds)andPrattetal.(2009)(greysquares)forclusters,fromOsmond&Ponman(2004) (greycircles)forgroupsandfromHorner(2001)(greytriangles).DatafromsimulationswerecomputedinsideR500.Leftpanel:resultsforthenofeedback case(NF,bluetriangles)andforthecasewithgalacticwinds(W,greensquares).Rightpanel:resultsfortherunswithstandardAGNfeedback(AGN1,red circles)andwiththemodifiedAGNfeedbackscheme,wherealsoaradio-moderegimeisincluded(seetext,AGN2;cyancircles).Foreachseriesofruns, filledandopensymbolsrefertothemainhalowithineachresimulatedLagrangianregionandto“satellite”halosrespectively.Blackdiamondsrefertothe runsat1×resolution,witharrowspointingtoresultsat6×higherresolutionfortheg676cluster. atureintroducedbyMazzottaetal.(2004).Thisdefinitionoftem- withtheresultontheL –T relation,thisresultshowsthatarather X peraturehasbeenshowntoaccuratelyreproduce, withinfewper- tunedenergyinjectionisrequired,whichmustbeabletosuppress cents,theactualspectroscopictemperatureobtainedbyfittingspec- X–rayluminositybydecreasinggasdensity,whileatthesametime traofsimulatedclusterswithasingle–temperatureplasmamodel, reproducingthelowentropylevelmeasuredatsmallradii. withinthetypicalenergybandswheredetectorson-boardofpresent X–raysatellitesaretypicallysensitive. 3.3 Temperatureprofiles WeshowinFigure3thecomparisonbetweenoursimulations andobservationaldataongroups(Sunetal.2009)andonclusters Anumberofcomparisonsbetweenobservedandsimulatedtemper- (Vikhlininetal. 2009) for the relation between entropy and tem- atureprofilesofgalaxyclustershaveclearlydemomnstratedthata peratureatR500andR2500(upperandlowerpanels,respectively). remarkableagreementexistsatrelativelylargeradii,R∼>0.2R180, Inorder toreproduce theprocedure adopted bySunetal.(2009), wheretheeffectofcoolingisrelativelyunimportant.Whilethisre- wecomputethespectroscopic–liketemperatureofsimulatedclus- sult, which holds almost independently of the physical processes tersbyexcludingthecoreregionswithin0.15R500.Asfortheruns includedinthesimulations(e.g.,Lokenetal.2002;Borganietal. withnoefficientfeedback(NF)wenotethattheyproduceentropy 2004;Kayetal.2007;Prattetal.2007;Nagaietal.2007),should levels, at both R500 and R2500, which are close to the observed be regarded as a success of cosmological simulations of galaxy ones. Thisresult can beexplained inthesameway asthat found clusters, the same simulations have much harder time to predict for the LX–T relation: overcooling, not balanced by an efficient realisticprofileswithincool-coreregions(e.g.,Borganietal.2008, feedbackmechanism,removesalargeamountofgasfromtheX– forarecentreview). Inthisregime,radiativesimulationssystem- rayemittingphase,whileleavinginthishotphaseonlyrelatively atically produce steep negative temperature profiles, at variance highentropygas,whichflowsinfromlargerradiiasaconsequence with observations, as a consequence of the lack of pressure sup- oflackofcentralpressuresupport. portcausedbyovercooling. Thisfurtherdemonstratesthatasuit- Including winds (W runs) has the effect of increasing the ablefeedbackmechanismsisrequiredtopressurisethegas,soas amountoflow-entropygas,whichisnowallowedtoremaininthe topreventovercoolingandturningthetemperaturegradientsfrom hotphasedespiteitsformallyshortcoolingtime,thankstothecon- negativetopositiveinthecoreregions.Whilefeedbackassociated tinuousheatingprovidedbywinds.Asaresult,entropydecreases toSNehasbeenprovednottobesuccessful,AGNfeedbackisgen- atbothradii,forthesamereasonforwhichX–rayluminosityin- erallyconsideredasalikelysolutionforsimulationstoproducere- creases. As for the runs with AGN feedback, its effect is almost alisticcoolcores.Basedonsimulationsofonerelativelylow–mass negligibleformassiveclusters.Ontheotherhand,AGNfeedback cluster,Sijackietal.(2008)foundthatAGNfeedbackcanprovide providesasignificantincreaseoftheentropylevelinpoorsystems, reasonabletemperatureprofilesonlyifapopulationofrelativistic aneffectwhichislargeratsmallercluster-centricradii.Theresult- particlesisinjectedalongwiththermalenergyininflatedbubbles. ingentropyishigherthanindicatedbyobservationaldata.Together WepresentinFigure4thecomparisonbetweensimulatedand AGNfeedbackand metalenrichmentofclusters 9 2] 2] m m c c V 1000 V 1000 e e k k [0 [0 0 0 5 5 K K 3 3 4/ 4/ z) z) E( E( NF AGN1 W AGN2 Sun et al. 2006 Sun et al. 2006 Vihlinin et al. 2009 Vihlinin et al. 2009 100 100 1 10 1 10 T (T ) (< R ) [keV] T (T ) (< R ) [keV] spec sl 500 spec sl 500 2] 2] m 1000 m 1000 c c V V e e k k [0 [0 0 0 5 5 2 2 K K 3 3 4/ 4/ z) z) E( E( 100 100 NF AGN1 W AGN2 Sun et al. 2006 Sun et al. 2006 Vihlinin et al. 2009 Vihlinin et al. 2009 1 10 1 10 T (T ) (< R ) [keV] T (T ) (< R ) [keV] spec sl 2500 spec sl 2500 Figure3.Relationbetweenentropyandtemperatureforoursimulatedclusters(colouredcirclesandsquares)andfortheobservationaldatapointsatR500 (upperpanels)andR2500 (lowerpanels)fromSunetal.(2009)(blackfilleddiamonds)andVikhlininetal.(2009)(blackopendiamonds).Leftandright panelsshowresultsfortheninecentralclustersfortherunswithout(NF:bluesquares;W:greencircles)andwithAGNfeedback(AGN1:redsquares;AGN2: cyancircles),respectively.Forafaircomparisonwithobservations,spectroscopic-liketemperaturesofthesimulatedclustersarecomputedbyexcludingthe regionswithin0.15R500. observed temperatureprofilesfor galaxyclusters withT∼>3 keV andtheAGN2feedbackschemespressurisetheICMinthecentral (leftpanel)andforpoorerclustersandgroupswithT∼<3keV(right regions, thus preventing adiabatic compression in inflowing gas. panel). Observational results are taken from Leccardi&Molendi From the one hand, this result confirms that a feedback scheme (2008b) and Sunetal. (2009) for rich and poor systems, respec- not relatedto star formation goes indeed inthe right direction of tively. As for rich clusters, none of the implemented feedback regulatingthethermalpropertiesoftheICMincoolcoreregions. schemeiscapabletoprevent thetemperaturespikeatsmallradii, Ontheotherhand, italsodemonstratesthattheschemesofAGN whileallmodelsprovideatemperatureprofilequitesimilartothe feedbackimplementedinoursimulationsdoanexcellentjobatthe observedoneatR∼>0.3R180.Thesituationisdifferentforgroups. scaleof galaxygroups, whiletheyarenot efficient enough at the Inthiscase,bothschemesofAGNfeedbackprovideresultswhich scaleofmassiveclusters. gointherightdirection.Whilegalacticwindsarenotabletosignif- icantlychangethesteepnegativetemperaturegradients,theAGN1 10 Fabjanet al. 2 2 AGN-1 NF W W AGN-2 AGN-1 1.8 1.8 NF AGN-2 Leccardi&Molendi 2008 Sun et al. (2009) 1.6 1.6 R)180 1.4 R)2500 1.4 < 1.2 < 1.2 T(mean 1 T(mean 1 T / sl 0.8 T / sl 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0.2 0.4 0.6 0.8 1 0 0.5 1 R/R R/R 180 500 Figure4.ComparisonbetweenthetemperatureprofilesforsimulatedandobservedclusterswithT∼>3keV(leftpanel)andforgroupswithT∼<3keV(right panel).Ineachpanel,differentlinescorrespondstotheaveragesimulatedprofilescomputedoverthefourmainmassiveclustersintheleftpanelandover thefivemainlow-massclustersintherightpanel,forthedifferentsetsofruns:nofeedback(NF,blueshortdashed),galacticwinds(W,greenlongdashed), standardAGNfeedback(AGN1,reddot-dashed),modifiedAGNfeedback(AGN2,cyansolid).Forreasonsofclarity,weshowwitherrorbarsther.m.s.scatter overtheensembleofsimulatedclustersonlyfortheAGN1runs.ObservationaldatapointsforclustersintheleftpanelaretakenfromLeccardi&Molendi (2008b),whilethoseforgroupsintherightpanelarefromSunetal.(2009). 0.18 0.18 0.16 0.16 0.14 0.14 0.12 0.12 ) 00 0.1 ) 00 0.1 5 5 R R < < (as 0.08 (as 0.08 g g f f 0.06 0.06 0.04 0.04 Vikhlinin et al. 2006 Vikhlinin et al. 2006 0.02 0.02 NF AGN-1 W AGN-2 cosmic baryon fraction cosmic baryon fraction 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 T (T ) [keV] T (T ) [keV] spec sl spec sl Figure5.ComparisonofthegasfractionwithinR500insimulations(colouredcircles)andobservationaldatafromChandradata(diamondswitherrorbars) analysedbyVikhlininetal.(2006).Leftpanel:resultsforthenofeedback(NF)runs(darkblue)andfortherunswithgalacticwinds(W,lightgreen).Right panel:resultsfortherunswiththestandardAGNfeedback(AGN1,darkred)andwithmodifiedAGNfeedback(AGN2,lightcyan).Thehorizontaldottedline marksthecosmicbaryonfractionassumedinthesimulations.
Description: