The Co-evolution of Galaxies, Black Holes, and AGN in a Hierarchical Universe

ASPConferenceSeries,Vol.**VOLUME**,**YEAROFPUBLICATION****NAMESOFEDITORS**

TheCo-evolutionofGalaxies,BlackHoles,andAGNinaHierarchicalUniverse

RachelS.Somerville

Max-PlanckInstituteforAstronomy

Abstract.TheobservationallinkbetweenSupermassiveBlackHoles(SMBH)andgalaxiesatlowredshiftseemstobeverytight,andstatisticallytheglobalevolutionofstarformationactivityandBHaccretionactivityalsoseemtotraceeachotherclosely.However,pinningdowntheco-evolutionofgalaxiesandBHonanobject-by-objectbasisremainselusive.Ipresentresultsfromnewmodelsforthejointevolutionofgalaxies,SMBH,andAGN,whichmaybeabletohelpresolvesomeoftheobservationalpuzzles.Auniqueaspectofthesemodelsisourtreatmentofself-regulatedBHgrowthbasedonhydrodynamicsimulationsofgalaxy-galaxymergers.Althoughthesemodelsdoquitewellatreproducingtheobservedevolutionofgalaxies,theydonotreproducetheobservedhistoryofBHaccretion,predictingtoomuchearlyaccretionandnotenoughatlatetimes.Isuggesttwopossibleresolutionstothisproblem.

1.Introduction

Thereisstrongevidencethatthepresent-daypropertiesofSupermassiveBlackHoles(SMBH)andtheirhostgalaxiesaretightlylinked—forexampleweobserveatightrelationshipbetweenBHmassandthemassofthehostspheroidinnearbygalaxies(e.g.H¨aring&Rix2004).However,wedonotyetknowwhetherthisBH-galaxymassrelationshipevolvedovercosmictime,orhowgalaxiesandSMBHarrivedontoit.Perhapsmostimportantly,wedonotyetunderstandthephysicaloriginofthisrelationship.

Mergershaveoftenbeenproposedasamechanismthatcandrivegasintothenucleiofgalaxiesandfuelbothstarburstsandblackholeaccretion.Hopkinsetal.(2005)usedalargesuiteofhydrodynamicsimulationsofgalaxymergerstocharacterizetheepisodeofblackholegrowthandtheAGNaccretion“lightcurves”intheseevents.Hopkinsetal.(2006)thenusedthismodeltotestwhetherobservedgalaxymergerrates,thebuild-upofred/spheroidalgalaxies,andAGNluminosityfunctionsareconsistentwiththepictureinwhichmergersbothtriggerAGNactivityandcausemorphologicalandcolortransformation.Theyconcludedthatthereisexcellentstatisticalagreementbetweenallofthesequantities.

However,establishingadirectlinkbetweenmergersandAGNactivityhasprovencontroversial.LowluminosityAGN,whicharemorecommonandcanbestudiedinmoredetailatrelativelylowredshift(z<1),donotshowstrongevidenceforbeingpredominantlyinmorphologicallydisturbedhostsorclosepairs(e.g.Lietal.2006;Pierceetal.2007).Higherluminosityquasarsseemtobemoreoften(thoughnotalways)associatedwithdisturbedhostsortohaveclosecompanions(Bahcalletal.1997;Bennertetal.2008;Letaweetal.2008),1

2R.S.Somerville

butthesampleswithhighresolutionimagingaresmall.ItmaybethecasethatAGNaremosteasilyidentifiedintheverylatestagesofmergers,whenthetwonucleihavecompletelycoalescedandmostsignsofmorphologicaldisturbancehavebecomeinvisible.Especiallyathighredshift,atthetypicallyavailableimagesensitivity,theseverylatestagemergerremnantsmightwellappeartobe“normal”spheroidal-typegalaxies.

Ofcourseitisalsopossible,evenlikely,thattherearemultiplemodesofBHgrowth.ThereareamplesuggestionsintheliteraturethatAGNactivitycouldalsobefedbybarsorbystochasticaccretionofcoldgasingalacticnuclei.Hopkins&Hernquist(2006)presentedamodelforfuelingofBHinisolateddisksbystochasticaccretion,andarguedthatwhilethisfuelingmodemaybeimportantinlowerluminosityobjectsandatlatetimes,mostofthegrowthoftoday’smassiveBHoccurredinthemerger-drivenmodeathighredshift.

Anothersignofco-evolutionisthattheglobalhistoriesofstarformationandBHaccretionseemtotraceeachotherremarkablycloselyoverthewholerangeoflookbacktimesoverwhichthesequantitieshaverobustobservationalestimates.Theglobalstarformationratedensityisalmostexactlyafactorof2000timestheBHaccretionratedensity,fromredshiftz∼5tothepresent.Moreover,evenwhendividedbymassoraccretionrate,“matched”populationsofgalaxiesandblackholesatz<1alsotraceoneanother’sactivity(Zhengetal.inprep).However,weknowthatatleastatz<1,thebulkofthestarformationactivityisassociatedwithisolateddisks,notmergersorspheroids(Belletal.2005).Therefore,itisalsostrangethatthesetwokindsofactivitytraceeachother,butdonotseemtobeoccuringinthesametypesofobjects(Zhengetal.inprep).

Athigherredshift,theremaybealargermismatchbetweenbetweenthetwokindsofactivity.Faucher-Giguereetal.(2008)obtainedconstraintsontheglobalstarformationdensityat22.ThismayindicatethatthereisatimedelaybetweenSFactivityandQSOactivity,whichisnaturallyexplainedinthemergerpicture(anygalaxywithcoldgascanformstars,butgalaxieshaveto“wait”foramajormergerbeforesignificantaccretionontotheBHcanoccur).

Onewaytotestthispicture,inwhichmergersareresponsiblefortriggeringbothAGNactivityandmorphologicalandspectrophotometrictransformationofgalaxies,istobuildcosmologicalmodelsthattreatthegrowthofgalaxies,blackholes,andAGNself-consistently.Inaddition,manyastronomersnowbelievethattheenergyreleasedbyaccretingblackholesmayplayacrucialroleinregulatinggalaxyformation.Herewedescribeonesuchsemi-analyticmodelforthejointformationofgalaxies,blackholes,andAGN,andpresentsomepredictionsfromthesemodels.2.

UnifiedModelsforGalaxies,BlackHoles,andAGN

Thefoundationsofourmodel,whichpertaintothegrowthofstructureinthedarkmattercomponent,andtheformationofgalaxies,aredescribedin

Co-evolutionofGalaxies,BH,andAGN3

Somerville&Primack(1999),Somervilleetal.(2001),andsubsequentworks.Briefly,themodelsareimplementedwithindarkmatterhalo“mergertrees”,andincludeapproximatetreatmentsforatomiccoolingofgas,photoionization,theformationofangularmomentumsupporteddisks(includingestimatesoftheirsizesandcircularvelocities),starformationaccordingtoanempirical“Kennicutt”Law,supernovafeedback,andchemicalevolution.Wehavere-centlyimplementednewmachinerywithinthisframeworktotreatblackholegrowthandtheassociatedAGNfeedback.OurnewmodelsarefullydescribedinSomervilleetal.(2008,S08).Herewegiveaverybriefsynopsisofthemostimportantnewingredients.Themodelspresentedherearefora“Concordance”ΛCDMcosmologywithΩm=0.3,ΩΛ=0.7,H0=70km/s/Mpc,σ8=0.9,ns=1.

Galaxy-galaxyMergersWeassumethatgalaxymergersmayinduceaburstofstarformationanddestroyanypre-existingdiskcomponent,dependingonthemassratioofthesmallertothelargergalaxy,µ.Weparameterizethestrengthandtimescaleoftheseburstsaccordingtotheresultsofalargesuiteofhy-drodynamicsimulationsofgalaxymergers(Robertsonetal.2006b;Coxetal.2008),asdescribedinS08.Mergerscanalsoheatandthicken,orevendestroy,apre-existingdiskcomponent,drivinggalaxiestowardsmorphologicallyearlierHubbletypes.Weassumethatthefractionofthepre-existingstarsthatistrans-ferredfromadisktoaspheroidalcomponentisastronglyincreasingfunctionofthemergermassratioµ,suchthatminormergerswithµ<0.2havelittleeffect,andmajormergerswithµ>0.25leavebehindaspheroid-dominatedremnant.BrightModeAGNandAGN-drivenwindsInourmodel,mergersalsotriggertheaccretionofgasontosupermassiveblackholesingalacticnuclei.Eachtop-levelhaloinourmergertreesisseededwithablackholeofmassMseed≃100M⊙.Followingamerger,theblackholeisallowedtogrowattheEddingtonrateuntiltheBHreachesacriticalmass,wheretheradiativeenergybeingemittedbytheAGNbecomessufficienttohaltfurtheraccretion.Thisself-regulatedtreatmentofblackholegrowthisbasedonhydrodynamicsimulationsincludingBHgrowthandfeedback(Springeletal.2005b;DiMatteoetal.2005;Hopkinsetal.2007a),andisdescribedinmoredetailinS08.Energyradiatedbyblackholesduringthis“bright”,quasar-likemodecanalsodrivegalactic-scalewinds,clearingcoldgasfromthepost-mergerremnants(Springeletal.2005a).Ourmodelformomentum-drivenAGNwindsisdescribedinS08.

RadioModeFeedbackInadditiontotherapidgrowthofBHinthemerger-fueled,radiativelyefficient“brightmode”describedabove,weassumethatBHalsoexperiencealow-Eddington-ratio,radiativelyinefficientmodeofgrowthassociatedwithefficientproductionofradiojetsthatcanheatgasinaquasi-hydrostatichothalo.TheaccretionrateinthisphaseismodelledassumingBondiaccretionusingtheisothermalcoolingflowsolutionofNulsen&Fabian(2000).WethenassumethattheenergythateffectivelycouplestoandheatsthehotgasisgivenbyLheat=κheatηm˙radioc2,wherem˙radioistheaccretionrateontotheBH,η=0.1istheassumedconversionefficiencyofrest-massintoenergy,andκheatisafreeparameteroforderunity.

4R.S.Somerville

Figure1.Thepredictedrelationshipbetweenbulgemassandblackholemass(greyshadingindicatestheconditionalprobabilityP(mbh|mbulge);lightbluesolidanddashedlineshowsthemedianand16thand84thpercentiles)comparedwiththeobservedrelationfromH¨aring&Rix(2004,greenlines).SymbolsshowthemeasurementsforindividualgalaxiesfromH¨aring&Rix(2004).

3.TheGlobalEvolutionofGalaxiesandBlackHoles

Wenormalizethefreeparametersinourmodelstoreproduceakeysetofob-servationalquantitiesatz∼0,suchasthestellarmassfunction,thecoldgasfractionindiskgalaxies,andthestellarmetallicity(S08).WeshowourmodelpredictionsandcomparewithobservationsforabroadrangeoflowredshiftgalaxypropertiesinS08.

TherelationshipbetweengalaxymassandBHmassisclearlyakeyresultthatourmodelshouldreproduce.Inourmodel,thisrelationshipissetbythedepthofthepotentialwellofthegalaxyatthetimewhentheBHforms,whichinturnisdeterminedbythegasfractionoftheprogenitorgalaxiesofthelastmerger.Moregas-richprogenitorssuffermoredissipationwhentheymerge,andproducemorecompactremnantswithdeeperpotentialwells.Adeeperpotentialwellrequiresmoreenergy,andthereforeamoremassiveBHinordertohaltfurtheraccretionandgrowth.WeseefromFig.1thatourfiducialmodelreproducestheobservedslopeandscatteroftheMBH–Msph(blackholemassvs.spheroidmass)relationship.OurmodelalsopredictsthattheMBH–Msphrelationshouldevolvewithtime.Becausemoregas-richmergerprogenitorsproduceremnantswithlargerblackholes,andthegalaxiesinourmodelsweresignificantlymoregasrichinthepast,galaxieshavelargerBHfortheirspheroidmassthantheydoatthepresentday.Thisleadstoarelativelymildamountofevolution,ofafactoroflessthantwosincez∼1andafactorofaboutfoursincez∼3(Hopkinsetal.2007a).

Fig.2showstheglobalstarformationratedensityofallgalaxiespredictedbyourmodels.Weshowboththeresultsofour“fiducial”model,aswellasamodelinwhichgalaxyformationhasbeensuppressedinlow-masshalos(wesimplydonotallowgastocoolinhaloswithmasslessthan1011M⊙).Wewill

Co-evolutionofGalaxies,BH,andAGN5

Figure2.Left:Starformationratedensityasafunctionofredshift.TheuppersetofthickerlinesshowsthetotalSFRinthemodels,andthelowersetofthinlinesshowstheSFRduetobursts.Symbolsandsolidlinesshowacompilationofobservationalresults,convertedtoaChabrierIMF(seeS08fordetails).Right:Theintegratedglobalstellarmassdensityasafunctionofredshift.Symbolsandthesolidlineshowobservationalestimates(seeS08).Inbothpanels,dashed(purple)curvesshowour“fiducial”ΛCDMmodel,anddot-dashed(red)curvesshowamodelwithreducedgalaxyformationinsmallmasshalos(seetext).

refertothelattermodelasthe“low”modelforbrevity.Theresultsofthelowmodelaresimilartothosefrommodelswithreducedsmallscalepower,e.g.withlowerσ8.Wehavekeptthesamevaluesforthefreeparametersinbothmodels.Bothmodelsfitthedatafairlywellfrom03)thanthelowmodel,becausemuchofthestarformationattheseredshiftsistakingplaceinlow-masshalos.

IntherightpanelofFig.2weshowthecomplementaryquantityρstar,theintegratedcosmicstellarmassdensity.Thefiducialmodelpredictsasignificantlyearlierassemblyofstarsingalaxiesthanobservationsofhighredshiftgalaxiesindicate.However,thelowmodelproducesverygoodagreementwiththestellarmassdensityasafunctionofredshift.Wenotethatthistensioninthemodelresultsisconnectedwithapossibleinconsistencybetweenthetwoobservationaldatasets(starformationratesandstellarmasses)thathasbeennotedrecentlyinseveralpapers(e.g.Hopkins&Beacom2006;Fardaletal.2007;Wilkinsetal.2008;Dav´e2008).OnepossibleresolutionofthistensioncanbeobtainedifthestellarIMFhaschangedwithtime,andwasmoretop-heavyathighredshift.4.

Evolutionof“BrightMode”AGN

Fig.3showstheintegratedbolometricluminositydensityof“brightmode”BHaccretioninourmodels,comparedwithobservationalestimatesfromquasarlu-minosityfunctions(Hopkinsetal.2007b).Weseethatinourfiducialmodel,QSOactivitypeaksattoohigharedshift,andfallsofftoorapidlyatlowred-

6R.S.Somerville

Figure3.Thebolometricluminositydensityof“brightmode”AGN(quasars)asafunctionofredshift.Symbolswitherrorbarsshowtheobser-vationalestimatesfromHopkinsetal.(2007b).Linesshowmodelpredictionsfrom:ourfiducialmodel(solid),our“low”model(short-dashed),amodelinwhichblackholesareejectedaftereverymerger(longdashed),andthemodelofCrotonetal.(2006).

Co-evolutionofGalaxies,BH,andAGN7

shift.The“low”modelpredictsariseandfallinQSOactivitythatisinsome-whatbetteragreementwiththedata,butstilldoesnothavequitetherightshape.Forcomparison,weplotthebrightmodeaccretionrateinthemodelsof(Crotonetal.2006),whicharesimilartoourfiducialmodelsinmanyrespects,buttreatBHgrowthandAGNfeedingdifferently.Weseethatthismodelalsofailstoreproducetheobservations:QSOactivitypeakstooearly,anddoesnotfalloffrapidlyenoughatlowredshift.

Thisresultissurprising,becauseHopkinsetal.(2008b)foundgoodagree-mentwiththesesameobservations,usingthesamemodelforBHgrowththatwehaveimplementedhere.WecheckedthatthemergermassfunctionpredictedbyourmodelagreeswiththatintheHopkinsetal.models.Thegasfractionsofourmergerprogenitorsalsoagreewell.Eventuallywerealizedthatanim-portantdifferenceintheassumptionsmadebyHopkinsetal.(2008b)andthosecontainedinthemodelspresentedherewasthat,intheHopkinsetal.model,theblackholeswereallowedtogrowfromtheseedmasstothefinalmassineachmerger.Thiseffectivelyneglectedthepre-existingBHthatgrewinearliermergers.Putanotherway,inourmodels,byz∼1–2,thereisalreadyasig-nificantamountofmassinspheroidsinplace,eachcontainingamassiveblackhole.Ifthepre-existingBHisalreadyabovethecriticalmassdiscussedabove,thennoAGNactivitytakesplace.Thus,inourmodels,manylatemergersdonottriggeranyAGNactivity.

Asacheck,were-ranourmodel,assumingthattheblackholesareejectedfromthegalaxyattheendofeverymerger.Thebrightmodeaccretionhistorynowagreesmuchbetterwiththeobservationsatleastinshape,althoughthenormalizationisabithigh.Althoughthismodelmayseemartificial,itispossiblethatinsomecasesblackholesareactuallyejectedbythegravitationalrocketmechanism,whichmayimpartkicksofseveralhundredsto1000km/stomergingBH(Volonteri2007,andreferencestherein).However,thisejectionofmassiveBHcanclearlyhaveaproblematicimpactonthepredictedMBH–Msphrelation,asshowninFig.reffig:mbh.However,themodelshownhereisextremeinthatBHareejectedfromeverygalaxyaftereverymerger.ItispossiblethatamorephysicalimplementationofthegravitationalrocketmechanismcouldimprovetheBHaccretionhistorywithoutleavingbehindtoomanymassivespheroidswithnoBH,orwithverylowmassBH,whicharenotobserved.

Anotherpossibleresolutiontothisproblemcouldperhapsbeobtainedifspheroidsarenotformedasefficientlyinhighredshiftmajormergersaswehaveassumedhere.Becauseinourmodels,thepropertiesofthespheroidregulateBHformation,ifspheroidsformlater,thenBHaccretionwilloccurlateraswell.Thereareindicationsfromadetailedanalysisofthehydrodynamicmergersimu-lationsthatgas-richmergersmayoftenproducedisk-like,ratherthanspheroidal,remnants(Robertsonetal.2006a;Hopkinsetal.2008a).Becauseourdiskstendtobemoregasrichathighredshift,thiseffectwouldindeedshiftspheroidfor-mationtolatertimes.Wearecurrentlyworkingonimplementingthiseffectinourmodels.

85.

R.S.Somerville

SummaryandConclusions

Wehavepresentedsomefirstpredictionsfromnewmodelsforthejointevolutionofgalaxies,blackholes,andAGN.Ourmodelsdifferfromothersemi-analyticmodelsintheliteratureinthatwehaveimplementedself-regulatedBHgrowthbasedontheresultsofhighresolutionhydrodynamicsimulationsofgalaxymerg-ers.

Weshowedthatourfiducialmodeldoesafairlygoodjobofreproducingtheglobalstarformationhistory,butoverproducesthestellarmassinplaceathighredshift.A“low”modelwithcoolingsuppressedinlow-masshalos,oramodelwithreducedsmall-scalepower,producesgoodagreementwiththestellarmassdensityhistorybutdoesnotagreeaswellwiththestarformationhistory.Thisreflectsaninternaltensionbetweenthesetwodatasets,whichmaybeduetosystematicerrorsarisingfromatime-varyingstellarInitialMassFunction.

Wefoundthatourmodel,whichdoesfairlywellatreproducinggalaxyobservationsatbothlowandhighredshift,failstoreproducetheformoftheBHaccretionhistoryastracedbythebolometricquasarluminositydensity.Inourmodels,QSOactivitypeaksattoohigharedshift,andfallsofftoorapidlyatlowredshift.ThemodelsofCrotonetal.(2006)alsofailtoreproducethisbehavior.Wesuggestedtwopossibleresolutionstothisproblem:1)BlackHolesare(atleastsometimes)ejectedfromtheirhostgalaxyfollowingamergeror2)majormergersathighredshiftarenotasefficientatproducingspheroidalremnantsaswehaveassumed.

Acknowledgments.IthanktheorganizersoftheSubarumeetingforputtingonanexcellentconferenceandforinvitingme.Ithankmycollaborators,P.Hop-kins,T.Cox,B.Robertson,Y.Li,andL.Hernquistforallowingmetopresentourunpublishedworkhere.IalsothankmycollaboratorsX.Zheng,E.Bell,H.-W.Rix,K.Jahnke,andF.Fontanotforinterestingdiscussions.Iamgrate-fultoD.Crotonforprovidinghismodelpredictionsinelectronicform,andforusefuldiscussions.

References

BahcallJ.N.,KirhakosS.,SaxeD.H.,SchneiderD.P.,1997,ApJ,479,2BellE.F.,etal.,2005,ApJ,625,23

BennertN.,CanalizoG.,JungwiertB.,StocktonA.,SchweizerF.,PengC.Y.,Lacy

M.,2008,ApJ,677,846

CoxT.J.,JonssonP.,SomervilleR.S.,PrimackJ.R.,DekelA.,2008,MNRAS,384,

386

CrotonD.J.,etal.,2006,MNRAS,365,11Dav´eR.,2008,MNRAS,152

DiMatteoT.,SpringelV.,HernquistL.,2005,Nat,433,604FardalM.A.,KatzN.,WeinbergD.H.,Dav´eR.,2007,MNRAS,379,985

Faucher-Giguere,C.-A.,Lidz,A.,Hernquist,L.,Zaldarriaga,M.,2008,ApJinpress,

astro-ph/0806.0372H¨aringN.,RixH.-W.,2004,ApJ,604,L

HopkinsA.M.,BeacomJ.F.,2006,ApJ,651,142HopkinsP.F.,HernquistL.,2006,ApJS,166,1

HopkinsP.F.,CoxT.J.,YoungerJ.D.,HernquistL.,2008a,ArXive-prints,806HopkinsP.F.,HernquistL.,CoxT.J.,KereˇsD.,2008b,ApJS,175,356

Co-evolutionofGalaxies,BH,andAGN9

HopkinsP.F.,HernquistL.,CoxT.J.,RobertsonB.,KrauseE.,2007a,ApJ,669,45HopkinsP.F.,HernquistL.,MartiniP.,CoxT.J.,RobertsonB.,DiMatteoT.,Springel

V.,2005,ApJ,625,L71

HopkinsP.F.,RichardsG.T.,HernquistL.,2007b,ApJ,6,731

HopkinsP.F.,SomervilleR.S.,HernquistL.,CoxT.J.,RobertsonB.,LiY.,2006,

ApJ,652,8

LetaweY.,MagainP.,LetaweG.,CourbinF.,Hutsem´ekersD.,2008,ApJ,679,967LiC.,KauffmannG.,WangL.,WhiteS.D.M.,HeckmanT.M.,JingY.P.,2006,

MNRAS,373,457

NulsenP.E.J.,FabianA.C.,2000,MNRAS,311,346PierceC.M.,etal.,2007,ApJ,660,L19

RobertsonB.,BullockJ.S.,CoxT.J.,DiMatteoT.,HernquistL.,SpringelV.,Yoshida

N.,2006a,ApJ,5,986

RobertsonB.,CoxT.J.,HernquistL.,FranxM.,HopkinsP.F.,MartiniP.,Springel

V.,2006b,ApJ,1,21

SomervilleR.S.,HopkinsP.,CoxT.J.,RobertsonB.,HernquistL.,2008,MNRAS,

accepted,astro-ph/0808.1227

SomervilleR.S.,PrimackJ.R.,1999,MNRAS,310,1087

SomervilleR.S.,PrimackJ.R.,FaberS.M.,2001,MNRAS,320,504SpringelV.,DiMatteoT.,HernquistL.,2005a,ApJ,620,L79—,2005b,MNRAS,361,776VolonteriM.,2007,ApJ,663,L5

WilkinsS.M.,TrenthamN.,HopkinsA.M.,2008,ArXive-prints,801