Environment-driven evolution of galaxies in the z = 0.209 cluster Abell 209

17 Pages·2016·3.02 MB·English

A&A585,A160(2016) Astronomy
DOI:10.1051/0004-6361/201527399 &
(cid:2)c ESO2016 Astrophysics
CLASH-VLT: Environment-driven evolution of galaxies
(cid:2)
in the z = 0.209 cluster Abell 209
M.Annunziatella1,2,A.Mercurio3,A.Biviano2,M.Girardi1,2,M.Nonino2,I.Balestra2,P.Rosati4,
G.BartoschCaminha4,M.Brescia3,R.Gobat5,C.Grillo6,M.Lombardi7,B.Sartoris1,2,8,G.DeLucia2,R.Demarco9,
B.Frye10,A.Fritz11,J.Moustakas12,M.Scodeggio11,U.Kuchner13,C.Maier13,andB.Ziegler13
1 DipartimentodiFisica,Univ.degliStudidiTrieste,viaTiepolo11,34143Trieste,Italy
e-mail:[email protected]
2 INAF−OsservatorioAstronomicodiTrieste,viaG.B.Tiepolo11,34131Trieste,Italy
3 INAF−OsservatorioAstronomicodiCapodimonte,viaMoiariello16,80131Napoli,Italy
4 DipartimentodiFisicaeScienzedellaTerra,Univ.degliStudidiFerrara,viaSaragat1,44122Ferrara,Italy
5 KoreaInstituteforAdvancedStudy,KIAS,85Hoegiro,130-722Dongdaemun-guSeoul,RepublicofKorea
6 DarkCosmologyCentre,NielsBohrInstitute,UniversityofCopenhagen,JulianeMariesVej30,2100Copenhagen,Denmark
7 DipartimentodiFisica,UniversitàdegliStudidiMilano,viaCeloria16,20133Milan,Italy
8 INFN,SezionediTrieste,viaValerio2,34127Trieste,Italy
9 DepartmentofAstronomy,UniversidaddeConcepción,Casilla160-C,Concepción,Chile
10 StewardObservatory/DepartmentofAstronomy,UniversityofArizona,933NCherryAve,Tucson,AZ,USA
11 INAF−IASF-Milano,viaBassini15,20133Milano,Italy
12 SienaCollege/DepartmentofAstronomy,515LoudonRoad,Loudonville,NY,USA
13 UniversityofVienna,DepartmentofAstrophysics,Türkenschanzstr.17,1180Wien,Austria
Received18September2015/Accepted14October2015
ABSTRACT
Context.Theanalysisofgalaxyproperties,suchasstellarmasses,colors,sizesandmorphologies,andtherelationsamongthemand
theenvironment,inwhichthegalaxiesreside,canbeusedtoinvestigatethephysicalprocessesdrivinggalaxyevolution.
Aims.WeconductathoroughstudyoftheclusterA209withanewlargespectro-photometricdatasettoinvestigatepossibleenviron-
mentaleffectsongalaxypropertiesthatcanprovideinformationongalaxyevolutioninclusterhostileenvironments.
Methods.We use the dataset obtained as part of the CLASH-VLTspectroscopic survey, supplemented with Subaru/SuprimeCam
high-quality imaging in BVRIz-bands,which yields 1916 cluster members (50% of them spectroscopically confirmed) down to a
stellarmassM(cid:2) = 108.6M(cid:3).Wedeterminethestellarmassfunctionofthesegalaxiesindifferentregionsofthecluster,byseparating
thesampleintostar-formingandpassiveclustermembers.Wethendeterminetheintra-clusterlightanditsproperties.Wealsoderive
theorbitsoflow-(M(cid:2) ≤ 1010.0M(cid:3))andhigh-mass(M(cid:2) > 1010.0M(cid:3))passivegalaxiesandstudytheeffectoftheenvironmentonthe
mass-sizerelationofearly-typegalaxies,selectedaccordingtotheirSérsicindex;theeffectsarestudiedseparatelyforthegalaxiesin
eachmassrange.Finally,wecomparetheclusterstellarmassdensityprofilewiththenumberdensityandtotal-massdensityprofiles.
Results.The stellar mass function of the star-forming cluster galaxies does not depend on the environment. The slope found for
passivegalaxiesbecomesflatterinthedensestclusterregion,whichimpliesthatthelow-masscomponent startstodominatewhen
movingawayfromtheclustercenter. Thecolor distributionoftheintra-clusterlightisconsistent withthecolor ofpassivecluster
members.Theanalysisofthedynamicalorbitsofpassivegalaxiesshowsthatlow-massgalaxieshavetangentialorbits,avoidingsmall
pericentersaroundtheBCG.Themass-sizerelationoflow-masspassiveearly-typegalaxiesisflatterthanthatofhigh-massgalaxies,
and itsslope isconsistent with the slope of the relation of field star-forming galaxies. Low-mass galaxies are also more compact
withinthescaleradiusof0.65Mpc.Theratiobetweenthestellarandnumberdensityprofilesshowsamasssegregationeffectinthe
clustercenter.Thecomparativeanalysisofthestellarandtotaldensityprofilesindicatesthatthiseffectisduetodynamicalfriction.
Conclusions.Our results are consistent with a scenario in which the “environmental quenching” of low-mass galaxies is due to
mechanisms such as harassment out to r , starvation, and ram-pressure stripping at smaller radii. This scenario is supported by
200
the analysis of the mass function, of the dynamical orbits and of the mass-size relation of passive early-type galaxies in different
clusterregions.Moreover,ouranalysessupporttheideathattheintra-clusterlightisformedthroughthetidaldisruptionofsubgiant
(M(cid:2) ∼ 109.5−10.0M(cid:3))galaxies.Infact,ourresultssuggestthatlow-massgalaxiesaredestroyedbytidalinteractions,andthatthose
thatavoidsmallpericentersaroundtheBCGareinfluencedbytidalinteractionsthatreducetheirsizes.Wesuggestdynamicalfriction
astheprocessresponsiblefortheobservedmasssegregation.
Key words.galaxies:luminosityfunction,massfunction–galaxies:clusters:general –galaxies:clusters:individual: Abell209–
galaxies:stellarcontent–galaxies:structure–galaxies:evolution
1. Introduction environmental(i.e.nurture)processesplayakeyroleintheevo-
lution of galaxies (Boselli & Gavazzi 2014) especially at the
Galaxies in clusters experience many different processes,
low-massend.Thesemechanismscanbedividedintotwomain
that affect their properties and evolution. Among them the
classes:gravitationalinteractions,whichincludeallsortsoftidal
(cid:2) Based in large part on data collected at the ESO VLT (prog. interactions, e.g. galaxy-galaxy mergers, dynamical friction,
and harassment and hydrodynamicinteractions, which include
ID186.A-0798),attheNASAHST,andattheNASJSubarutelescope.
ArticlepublishedbyEDPSciences A160,page1of17
A&A585,A160(2016)
interactionsbetweenthegalaxy’sinterstellarmedium(ISM)and Table1.MainpropertiesoftheclusterAbell209.
thehotintergalacticmedium,i.e.rampressure(forareviewsee
Biviano2008).Theseprocessesareeffectiveindifferentregions BCG(α,δ) 01h31m52s.54,−13◦36(cid:7)40(cid:7).(cid:7)4
of the cluster (e.g. Treu et al. 2003; Boselli et al. 2006).Their MeanredshJi2f0t00 0.2090±0.0004
maineffectistohaltthestarformationofagalaxy,changingit Velocitydispersion[kms−1] 1320+64
fromstar-forming(SF)toquiescent.Otherpropertiesthatcould Virialradiusr [Mpc] 2.13±−06.705
200
beinfluencedbytheseprocessesare:stellarmass,size,andmor-
phologicaltype.Thestudyofthedistributionoftheseproperties Notes.Theradiusr200correspondstotheradiusofaspherewhosemass
andthecorrelationexistingamongthemandwiththeirenviron- overdensityis200timesthecriticaldensityattheclusterredshift.The
mentgivesinformationabouttherelativeimportanceofthepro- valueoftheredshiftistakenfromMercurioetal.(2003a).Thevelocity
cessesthataffecttheirevolution.Onepossiblewaytounderstand dispersioniscalculatedonourdataset(Sartorisetal.,inprep.),while
thevirialradiusistakenfromMertenetal.(2015).
the influence of these processes on the evolutionof galaxiesin
clustersistosearchforeffectsoflocalenvironmentonthedistri-
butionofgalaxystellar masses(M(cid:2)), i.e.thestellar massfunc- Anotherwaytoconstraintheeffectofdifferentenvironmen-
tion(SMF),andcomparetheSMFofgalaxiesinclusterstothat tal processeson the evolutionof galaxiesis to analyzethe dis-
inthefield. tribution of galaxy sizes and the relation with the stellar mass.
To disentangle the effects of the environment on the SMF It is knownthat differenttypes of galaxies, namelyearly-types
from the effects related to the morphology-density relation, (ETGs)andlate-typesshowdifferentdependenciesbetweensize
i.e., the fact that the fraction of different morphological types andstellar mass (Shenet al. 2003).In particular,the mass-size
changeswithlocaldensity(Dressler1980;Postmanetal.2005), relation of ETGs is steeper than the value found for late-type
oneshouldstudytheSMFseparatelyfordifferenttypesofgalax- galaxies. How the environment may change the size distribu-
ies.Sofar,thedeterminationoftheSMFhasbeenmainlybased tion of the galaxies is, instead, still a matter of debate. In the
ondatainthefield(Ilbertetal.2013),whiletheevolutionwith local Universe, Poggianti et al. (2013) claim that at fixed stel-
redshift (z) and the local environment of the SMF of different lar mass, cluster ETGs are slightly smaller than field galaxies.
galaxytypesinclustershasnotyetbeenstudiedingreatdetail. On the other hand, Huertas-Companyet al. (2013) did not de-
Onthe otherhand,the studyoftheluminosityfunction(LF)is tect any relation between galaxy sizes and environment.Many
betterknown.Byseparatingthecontributionofearly-andlate- ofthephysicalprocessesthatcanaffectthepropertiesofcluster
type galaxies using color indices, Popesso et al. (2006) show galaxiesarelikelytoberelatedtotheirorbits.Galaxiesonradial
thatther-bandLFoflocallate-typegalaxiescanbefitted with orbits,characterizedbysmallpericenters,canexperiencedenser
asingleSchechterfunction,whiletheLFofearly-typegalaxies environmentsthan galaxiesonmore tangentialorbits. Thiscan
requires a double Schechter function. They also show that the leadtotheirmorerapiddestruction(Taylor&Babul2004).Ithas
faint-endslopeoftheLFofearly-typegalaxiesexhibitsasignif- beensuggestedbySolanesetal.(2001)thatamongVirgocluster
icant and continuousvariationwith the environment,flattening galaxies,thosewithevidenceofgasstrippingareonmoreradial
in the central region of the cluster. They interpretthese results orbitsthantheothers.
as an effect of galaxy transformationfrom star-formingto qui- In this paper, we investigate the SMF of passive and
escent systems through harassment in the periphery and dwarf SF galaxies in the cluster Abell209 (hereafter A209) at
tidaldisruptioninthecore. z= 0.209 as a function of the local environment and down to
Vulcani et al. (2012, 2013) have found that global envi- amasslimitof108.6M(cid:3).Wealsoanalyzethedistributionofthe
ronment does not affect the SMF of different galaxy types. sizesofpassiveearly-typegalaxiesindifferentmassrangesand
Furthermore, they did not detect any difference between the in different cluster regions. Finally we determine the orbits for
SMFsofthetotalpopulationofgalaxieswithinandoutsidethe clusterpassivegalaxiesintwomassbins.
virialradiusofgalaxyclusters. Theseanalysesarebasedonanunprecedentedlargefraction
However, it is also importantto note that the effects of the (>50%)of∼1000spectroscopicallyconfirmedclustermembers
environmentcouldbeevidentonlybelowagivenmasslimitfor collectedattheESO-VLT,aspartoftheESOLargeProgramme
agivenredshift.Forexample,atz∼1vanderBurgetal.(2013) 186.A-0798 (hereafter CLASH-VLT, P.I. Piero Rosati, Rosati
findadifferenceinthe shapeoftheSMFofclustersandofthe et al. 2014). This is a spectroscopic follow-up of a subset of
fieldbelow1010M(cid:3). 14southernclustersfromthe “ClusterLensingAndSupernova
In Annunziatella et al. (2014), we studied the effect of en- surveywithHubble”(CLASH;seePostmanetal.2012).
vironmentontheSMFofSFandpassivegalaxiesinthegalaxy The main propertiesof A209 are listed in Table1. We per-
clusterMACS1206.2-0847(hereafterM1206)atthemeanred- formtheanalysisoftheSMFfortworeasons:tocheckwhether
shiftofz= 0.44,byuseofalargefraction(∼1/3)ofspectroscop- theresultsfoundforthedynamicallyrelaxedclusterM1206also
ically confirmedcluster members,downto M(cid:2)= 109.5M(cid:3). We holdforaclusterwithevidenceofsubstructures(Mercurioetal.
havefoundthattheSMF ofbothgalaxytypesdoesnotchange 2003b)and,tostudypossibleredshiftevolutioneffectscompar-
acrossthevirialradius1ofthatcluster.However,wediddetecta ingtheresultsobtainedinthetwoclusters,whichareseparated
differenceontheSMFofpassivegalaxiesintheinnermostand by ∼2 Gyr in cosmic time. For A209, we also study the intra-
densest region of the cluster in the form of a lack of galaxies cluster light(ICL) componentin order to get some hints about
with M(cid:2)< 1010.5M(cid:3). Our hypothesis is that these galaxies in itsprogenitorsandtounderstandhowitisrelatedtotheevolution
the very center of the cluster have been destroyed by tidal in- ofthegalaxiesinthecluster.Someoftheenvironmentaldepen-
teractionswiththeclusterpotentialandhaveenrichedtheintra- dentmechanismthatshapetheSMFofclustermembergalaxies,
clusterlight. infact,canalso beresponsiblefortheformationofthisdiffuse
component.The ICL was first discoveredby Zwicky(1951)in
1 Thevirialradiusr istheradiusofaspherewithmassoverdensityΔ theComacluster.Itoriginatesfromtidalinteractionsamongthe
200
timesthecriticaldensityattheclusterredshift.Unlessotherwisestated, galaxies or with the cluster potential during the cluster assem-
weadoptΔ=200throughoutthispaper. bly.Itisstillamatterofdebatewhichgalaxiescontributemore
A160,page2of17
M.Annunziatellaetal.:A209stellarmassfunction
totheICL.Fromatheoreticalpointofviewoneoftheproposed
leading scenarios is that this componentcomes from tidal dis-
ruptionofintermediatemassgalaxies(1010−11M(cid:3),Continietal.
2014). This scenario has also been supported by some obser-
vationalworks(Annunziatellaetal. 2014;Presotto et al. 2015;
DeMaio et al. 2015; Montes & Trujillo 2014). However, some
authors (e.g. Conroy et al. 2007) suggest that the central ICL,
associatedwiththeBCG,couldformfromdestructionofdwarf
galaxies. The joint analysis of the ICL and the SMF of cluster
memberscanshedsomelightonthistopic.
The paper is organized as follows. In Sect. 2 we describe
the data sample and explain how we assign the membershipto
thegalaxiesinthefieldofview.InSect.3wederivethestellar
massesof galaxiescluster members.InSect. 4 we estimate the
SMFforclustermembersandexamineitsdependenceongalaxy
typeandenvironment.InSect.5,wegiveabriefdescriptionof
themethodusedtoderivethemapoftheintra-clusterlightand
study some properties of the ICL in A209. In Sect. 6, we per-
formtheanalysisoftheorbitsofpassivegalaxies.InSect.7we
analyzethemass-sizerelationofpassivegalaxiesintworegions
of the cluster. In Sect. 8 we show the stellar mass density and
number density profiles compared with the total mass density
profile obtained from an independentlensing analysis (Merten
etal.2015).InSects.9and10wediscussourresultsanddraw Fig.1.HistogramoftheR -bandmagnitudeofthe51166objectsinthe
C
ourconclusions. photometric(blackline)sampleandofthe2622sourceswithreliable
Throughout this paper, we use H = 70kms−1Mpc−1, redshifts(redline).
0
ΩM = 0.3, and ΩΛ = 0.7. At the cluster redshift the scale is
206kpcarcmin−1.Allthemagnitudesusedinthisworkreferto
usingthe software SExtractor(Bertin & Arnouts1996)in con-
theABsystem.
junction with PSFEx (Bertin 2011; Annunziatella et al. 2013).
By using a method based on neural networks (Brescia et al.
2013), we obtain the photometric redshift z for all galaxies
2. Datasample phot
forwhichwehavethemagnitudesinalltheobservedbands.In
Atotalof11LR-BluemaskswereacquiredwithVLT/VIMOS, this paper, we consider only galaxies down to the limit of the
as part of the CLASH-VLT ESO large programme, using four observedspectroscopicsample:RC = 24.0mag(seeFig.1).
separate pointings, each with a different quadrant centered on Todetermineifgalaxieswithz canbeclassifiedasmem-
phot
the cluster core. The LR-Blue masks cover the spectral range bers, we investigate the diagram shown in Fig. 2. We use the
370–670 nm with a resolving power R = 180, for a total ex- objectswithspectroscopicredshifttofindthebestselectioncri-
posure time of 10.5 h. Spectra are classified according to the teria to identifycluster membergalaxies.Then,we applya cut
reliability of the redshift (see Biviano et al. 2013)as: “secure” inthephotometricredshift.Weselectasclustermembersthose
(QF = 3; 1946 objects) with a probability of >99.99%,“based galaxies with 0.175 < z < 0.272 (see horizontal lines in
phot
on emission-lines” (QF = 9; 110 objects) with a probabilityof Fig. 2). With these cuts in z we correctly classify 92% of
phot
∼92%,and“likely”(QF=2;492objects)withaprobabilityof thespectroscopicmembers,minimizingthenumberofinterlop-
∼75%. ers. The sample of photometrically selected members contains
Membership for objects with spectroscopic redshift is 1805 galaxies down to the magnitude limit of RC= 24.0 mag.
assigned according to what is known as the “peak+gap” Theselectionofmembersbasedonphotometricredshiftsisnot
(P+G) method (Fadda et al. 1996), already applied to M1206 asreliableasthatbasedonspectroscopicredshifts.Wetakethis
(Bivianoetal.2013;Girardietal.2015).Thismethodidentifies intoaccountasexplainedinSect.2.1.
apeakintheredshiftdistributionusinga1Dadaptivekernelal-
gorithm,thenrejectsgalaxiesfromoverlappingradialbinsthat
2.1.Completenessandmembershipcorrections
havelineofsightvelocitiesverydifferentfromthemeanvelocity
ofthecluster. Oursampleneedstobecorrectedfortwodifferentfactors.One
Thefinalspectroscopicdatasetconsistsof2548reliablered- is the incompleteness of the sample with photometric redshift
shifts,observedaspartofCLASH-VLT,inadditionto110taken withrespecttothephotometricsample,sincetoobtainz ,we
phot
fromMercurioetal.(2003a)andMercurioetal.(2008).Intotal, requirethatthegalaxiesbedetectedinallfiveSUBARUbands.
1116areconfirmedmembers. Theothertakesintoaccountpossiblediscrepanciesbetweenz
phot
PhotometricdatawereobtainedwiththeSUBARU/Suprime- andz inthephotometricselectionofmembers,whichwecall
spec
Cam from SMOKA (Baba et al. 2002) covering a 30(cid:7) × 30(cid:7) membershipcorrection.Inordertoaccountforthesecorrections,
fieldofview,withtotalexposuretimesof2400s,1800s,2400s, weapplythesamemethodasinAnnunziatellaetal.(2014).
1320s and 4800s in the B, V, R , I and z-band, respectively. The photometriccompleteness C is defined as the ratio be-
c p
Thetypicalseeinginthefinalsky-subtractedimagesvariesfrom tween the number of galaxies with photometric redshift and
0.486arcsecintheR-bandupto0.77intheB-bandwithapixel the total number of galaxies in the R -band. The complete-
C
scaleof0.2arcsec.Detailsonthedatareductioncanbefoundin nessC hasbeenevaluatedforthewholesampleandseparately
Nonino et al. (2009). The photometric catalogs were extracted for red and blue galaxies, where we use the threshold value
A160,page3of17
A&A585,A160(2016)
1.0 2.0
0.8 1.5
r
o
ct
a
f
n
0.6 o1.0
cti
e
hot orr
zp C
0.5
0.4
0.0
0.2 18.5 19.5 20.5 21.5 22.5 23.5 0.25 1.25 2.25 3.25
RC(mag) R(Mpc)
Fig.3.Correctionfactor(f )formembershipselectionasafunctionof
M
R magnitude(leftpanel)andclustercentricdistance(rightpanel),for
C
0.0 red(dots)andblue(diamonds)galaxies.
0.0 0.2 0.4 0.6 0.8 1.0
z
spec
Fig.2.Photometricvs.spectroscopicredshiftsforgalaxiesintheclus- 3. Measurementsofstellarmass
terfieldandinthemagnituderange18 ≤R ≤ 24mag.Blackdotsare
C
spectroscopicallyconfirmedmembers.Thetwodashedlinesrepresent We use the spectral energy distribution (SED) fitting method
thecriteriainphotometricredshiftschosentoclassifytheclustermem- to derive the stellar masses of the cluster members galaxies.
bers.Galaxieswithinthisrangeofphotometricredshiftarecoloredin We make use of the code MAGPHYS (da Cunha et al. 2008).
red.Galaxieswithz inthisrangeandwithoutspectroscopicinforma- For each galaxy, MAGPHYS finds the template at the redshift
phot
tionareclassifiedasmembers. Bluecrossesaregalaxiesoutsideboth of the galaxy that best reproduces the observed galaxy fluxes.
thespectroscopicandphotometricselections. The templates are assembled from stellar population synthesis
models (Bruzual & Charlot 2003 or Bruzual & Charlot 2007)
withaChabrier(2003)stellarinitialmassfunctionandametal-
licity value in the range 0.02–2 Z(cid:3). We adopt the Bruzual &
Charlot (2003) library of models. The two libraries differ in
the treatment of thermally pulsating asymptotic giant branch
B−R = 1.5 (calculated on the basis of aperture magnitudes stellar phase which is not relevant at the mean redshift of the
C
within5(cid:7)(cid:7))todiscriminatebetweenthetwosamples.Thisvalue cluster A209 (Zibetti et al. 2013). In order to obtain the SED,
forthecolorischosentomatchthelimitinspecificstarforma- MAGPHYS parametrizes the star formation history of a galaxy
tionrate()usedtodistinguishbetweenPassiveandSFgalaxies withacontinuum,SFR∝e−γt,andsuperimposedrandombursts.
(seeSect.3).TheparameterCisapproximatelyconstant,≥80%, The star formation timescale γ is distributed according to the
downtoR ≤ 24magindependentlyfrommagnitudeorgalaxy probabilitydensityfunction p(γ)=1 − tanh(8γ − 6),whichis
C
type.Therefore,weadoptaconstantmeanvalueC=0.81anda uniform between 0 and 0.6 Gyr−1 and drops exponentially to
correctionfactor,f =1/C=1.23. zero at 1 Gyr−1. The age of the galaxy is free to vary with an
C
upperlimitimposedbytheageoftheUniverseattheconsidered
Toobtainthemembershipcorrectionfactor,weevaluatethe
redshift.
completenessandthepurityofthesampleofgalaxiesclassified Inordertofindthebest-fitmodel,MAGPHYSusesaBayesian
as membersbased on their photometricredshift. We define the
approach. As outputs, it produces the probability distribution
completenessastheratiobetweenthenumberofgalaxiesiden-
function of all the parameters and their best-fit values. For the
tified as members(bothspectroscopicallyand photometrically)
parameters we are interested in we adopt the median value as
and the number of spectroscopic members CM=Npm∩zm/Nzm. themorerobustestimatewitha1σintervalgivenbythediffer-
The purity is defined as the ratio between the number of
ence between 16 and 84 percentilesof the probabilitydistribu-
photometric members that are also confirmed spectroscopi- tion.Themeanvalueforthe1σuncertaintyonthestellarmass
cally and the number of photometric members that have zspec, is∼0.07dex.
P=Npm∩zm/Npm∩z. The membership correction factor is de- In Annunziatellaet al. (2014)we checkedthat the value of
finedasf ≡P/C . InFig. 3we analyzethedependenceoff
M M M thestellarmassisindependentofthespecificalgorithmofSED
ongalaxytype,magnitude,anddistancefromtheclustercenter.
fitting(providedoneconsidersthesamestellarIMF).However,
Wefindthatthecorrectionfactordependsongalaxytype,while wefoundthatwhenonlyopticaldataisusedMAGPHYStendsto
itisindependentofbothmagnitudeandclustercentricdistance. overestimatethe mass of a galaxybecausethe publicMAGPHYS
Weusetwodifferentvaluesforredandbluegalaxiesintheentire
librariescontainanexcessofstar-formingdustymodelswithre-
magnitudeandradialrange,f =0.9and0.75,respectively.
M spect to passive ones (E. da Cunha, priv. comm.). This means
The finalcorrectionfactorwe appliedis givenbytheprod- that if the fluxes of a galaxy can be fitted by a dusty model or
uct f · f (∼1.125 and 0.937 for passive and SF galaxies, by a truly passive one, the median value is always biased to-
C M
respectively). wards the dusty model. In order to avoid this systematic error
A160,page4of17
M.Annunziatellaetal.:A209stellarmassfunction
2.2
12
2.0
11
1.8
1.6)
) g
(cid:3) 10 a
M m
/ 1.4(
∗ C
M R
g( 9 1.2−
o
L B
1.0
8
0.8
0.6
7
16 18 20 22 24 26 Fig.5. SMFsof SFand passive cluster member galaxies. The best-fit
RC(mag) SchechterfunctionsareshownwiththebluedashedlineforSFgalax-
iesandredtriple-dot-dashedlineforpassivegalaxies,whilethenum-
Fig.4. Relation between the stellar mass and the R magnitude. The
C ber counts(dividedbybinsize)arethebluesemicircleandredtrian-
points are color-coded according to the B−R color. Dashed vertical
C gle,respectively.Violetdiamondsarethecountsobtainedforallcluster
andhorizontallinesarethemagnitudeandmasslimits,respectively.
membersandthevioletsolidlinerepresentsthesumofthetwoSMFs.
Counts are evaluated in bins of 0.2 dex in M(cid:2). The 1σ errors on the
countshavebeenestimatedviathebootstrapresamplingprocedure.
we made use of a non-public library of templates heavily bi-
ased tofit oldstellar populationswith verylittle star formation
(provided by E. da Cunha). We proceed as follows. We per-
free in these fits, while the normalization is fixed by imposing
formedthe SED fitting for all the galaxiesin oursample using
thattheintegraloftheSMFinthechosenmassrangebeequalto
the standard libraries. Among its output parameters, MAGPHYS
thetotalnumberofgalaxiesoftheconsideredsamplecorrected
also provides an estimate of the specific star formation rate
fortheweightfactor.Inthesefitseachgalaxyhasaweightequal
(i.e., star formation rate per unit mass, sSFR ≡ SFR/M(cid:2)). We totheproductf ·f definedinSect.2.1.Wehavecheckedthat
examined the values obtained for the sSFR and checked that C M
thecontributionoftheerrorsonthestellarmassvalueissmaller
this parameter has a bimodal distribution, with a local mini-
thanthestatisticalerrorsassociatedwitheachoftheparameters
mum at sSFR = 10−10 yr−1 (see also Lara-López et al. 2010,
oftheSM.ThecorrectionfactorinFig.3,asshown,isindepen-
and references therein). We then classified as passive galaxies
dentfrommagnitudeorradialdistance.Therefore,theerrorson
those with values of sSFR< 10−10 yr−1. Star-forming galax-
this factor would influence only the normalizationof the SMF.
ies, instead,havesSFR≥ 10−10 yr−1 Atthispointwe fitted the
Theonlytwocasesinwhichwetookthenormalizationsintoac-
passive galaxies again using the non-public library of models.
countarewhenstudyingthetotalSMF(Sect.4.1)andwhenes-
The limiting magnitude of RC= 24 mag corresponds roughly timatingthecontributionofgalaxiesofdifferentmasstotheICL
to a completeness limit of 108.6M(cid:3) and to a surface bright- (Sect.5.2).Toestimatethecontributionoftheerrorsinthesetwo
ness of 24.7 mag/arcsec2 for passive galaxies,accordingto the cases, we repeatedour calculationsusingf ±Δf . The results
mass vs. R dispersion of Fig. 4 and to the measured galaxy M M
C we obtained are statistically consistent with those obtained us-
sizes (see Sect. 7). The mass limit for SF galaxies would be
ing f . Thus, for ouranalysis we do notconsiderthe errorson
lower. However, we chose the same value for passive galaxies M
the membership correction factor. The cumulative SMF for all
in order to be consistent when comparing the SMF of differ-
galaxiesinA209isgiveninFig.5(violetline).
entgalaxytypes.Withthismasslimit,ourfinalsampleconsists
We are interested in studying how the SMF depends on
of1916galaxies,morethanhalfofwhicharespectroscopically galaxytypeorenvironmentinordertoshedlightonthediffer-
confirmedmembers. entprocessesthatcanaffecttheevolutionofgalaxiesinclusters.
We wanted to compare only the shape of the SMF of different
samples,sowerenormalizedtheSMFsbydividingthemforthe
4. Stellarmassfunction
totalnumberofobjectsineachsample(exceptinFig.5).Inorder
The SMF of cluster members is obtained by considering the toestablishiftwoSMFsarestatisticallydifferentwecompared
number of galaxies in each stellar mass bin weighted for the the shape parameters (αandM∗) without taking the normaliza-
completenessandthemembershipcorrectionfactors.Theerrors tionvaluesintoaccount.Tohighlightthedifferencebetweentwo
onthecountsareestimatedwiththebootstrapprocedure(Efron SMFs, we also performedthe Kolmogorov-Smirnov(hereafter,
&Tibshirani1986).TheSMFcanbeanalyticallydescribedbya K-S) test (e.g., Press et al. 1993). This is a non-parametrical
Schechter(1976)function test that determines if two samples are drawn from the same
(cid:2) (cid:3) (cid:2) (cid:3) parentpopulationbyevaluatingthemaximumdistancebetween
Φ(logM)=ln(10)Φ∗ M 1+αexp −M d(logM), (1) their two cumulativedistributions. We used a modified version
M∗ M∗ of the K-S test in order to include the weights on stellar mass
(Annunziatellaetal.2014).
whereαistheslopeofthelow-massend,M∗ istheexponential
cutoff, and Φ∗ is the normalization. We fit the Schechter func-
tion to the M(cid:2) distribution downthe mass limit 108.6M(cid:3) using 4.1.Differentgalaxytypes
the maximum likelihood technique (Malumuth & Kriss 1986). The separate SMFs of SF and passive cluster member galax-
Theuseofthismethodimpliesthatonlytwoparametersareleft ies are shown in Fig. 5. The best-fit values of αandM∗ with
A160,page5of17
A&A585,A160(2016)
Table2.Best-fitSchechterfunctionparameters. Table 3. Best-fit Schechter function parameters for different
environments.
Galaxytype α log(M∗/M(cid:3))
Passive −1.09±0.02 11.10±0.05 Galaxytype Environment α log(M∗/M(cid:3))
SF −1.63±0.03 11.02±0.30 SF R>r200 −1.58±0.08 10.95±0.30
All −1.17±0.02 11.15±0.05 SF R≤r200 −1.69±0.07 11.42±0.45
Passive R>r −1.09±0.04 11.09±0.09
200
Passive R≤r −1.09±0.02 11.10±0.05
200
Passive Region(a) −1.01±0.06 11.21±0.12
Passive Region(b) −1.09±0.04 11.09±0.08
Passive Region(c) −1.10±0.03 11.09±0.08
Passive Region(d) −1.10±0.03 11.03±0.07
Table4.ResultsoftheK-Stests.
Comparedsamples N1,N2 Prob.(%)
Typedependence
Passivevs.SFinthecluster 1580,336 <0.01
Passivevs.SFinRegion(b) 425, 34 <0.01
Passivevs.SFinRegion(c) 488, 75 <0.01
Passivevs.SFinRegion(d) 75,219 <0.01
Environmentdependence–SFgalaxies
Fig.6.Best-fitvaluesoftheSchechterparametersαandM∗ withtheir
1σlikelihoodcontours. SFwithinandoutsider200 161,175 >10
SFinRegions(b)and(c) 34, 75 >10
SFinRegions(c)and(d) 75,219 >10
their 1σ errors are listed in Table 2 and shown in Fig. 6. The
uncertaintiesareobtainedbymarginalizingoverthenormaliza- Environmentdependence–passivegalaxies
tion parameter. The SMFs of, separately, SF and passive clus-
ter members are adequately fitted by a single Schechter func- Passivewithinandoutsider200 438,408 >10
PassiveinRegions(a)and(b) 160,425 8
tion. In order to assess the goodness of the fits, we compared
PassiveinRegions(b)and(c) 425,488 >10
themassdistributionofgalaxiesinthe consideredsample with
PassiveinRegions(c)and(d) 488, 57 >10
thebest-fitSchechterfunctionusingtheK-Stest.Weperformed
thistestforalltheSMFsconsideredinthispaper,exceptforthe Notes.ThevaluesN1andN2arethenumberofgalaxiesineachofthe
SMFofallclustermembers,alwaysobtaining P-values>10%. comparedsubsamples.Theprobabilitiesinthelastcolumnsarereferred
Therefore,wecannotrejectthenullhypothesisthatthetwodis- tothenullhypothesisthatthestellarmassdistributionsofthetwosam-
tributions are statistically identical. The SMF of passive and plesaredrawnfromthesameparentpopulation. Avalue≤1%means
SF cluster members is given by the sum of the SMFs of sep- that the two distributions are statistically different. Samples with less
arately SF and passive galaxies. The dependence of the SMF that10objectsareexcludedfromthetest.
on galaxy type is statistically significant. This is evident from
the comparison of the shape of the two SMFs and also from
theresultsoftheK-Stest(seeTable4).Thedifferencebetween passivegalaxiesinthetwoclustersatdifferentredshiftsisqual-
themassdistributionofpassiveandSFgalaxiesholdsnotonly itatively in agreementwith the results shown in Rudnick et al.
when considering the whole cluster, but also in different clus- (2009).Intheirpaper,theauthorsfoundthattheluminosityfunc-
ter regions. From Fig. 5 we see that the passive SMF domi- tionsofpassivegalaxiesshowedastrongevolutionwithredshift
nates over the SF SMF at all masses down to the complete- in terms of an increase of faint galaxies relative to bright ones
ness limit in A209. In Annunziatella et al. (2014), we found towards lower redshifts. Another feature of the passive galaxy
that the two SMFs of SF and passive galaxies for the cluster SMFisthedipvisibleinthegalaxycountsat∼1010.10M(cid:3).The
M1206 at z ∼ 0.44 crossed at the value Mcross ∼ 1010.1M(cid:3), if presence of this dip was also found in the luminosity function
wifewceornessitdreicreteddththeewahnoalleyscilsuisntesridreegthioenv,iarniadlMregcrioossn∼of1th0e9.5cMlu(cid:3)s-, oatftrheed-mseaqguneintucedegainlaxthieesRof-btahnisdc∼lu9s.t8erm(aMg,erwcuhriciohectoarlr.e2sp0o0n3dbs)
ter.Thedifferencebetweenwhathappensinthetwoclusterscan (according to the fit in Eq. (3)) to our mass value. This dip is
bepartiallyexplainedbytheButcher−Oemlereffect(Butcher& probablys an indication that there are two populations of pas-
Oemler1978),accordingtowhichgalaxyclustersatintermedi- sivegalaxies,oneoflowermassesprobablyoriginatingfromthe
ateredshiftcontainalargerfractionofbluegalaxiesthanlower quenchingofSFgalaxies,andanotherofhighermasses.Tover-
redshift ones. On the other hand, there is a striking difference ifythiswefitadoubleSchectertoourdatainSect.4.3.
betweentheslopeoftheSMFofpassivegalaxiesthatwefindin
thisworkandthevaluewefoundforpassivegalaxiesinM1206
4.2.Effectsofglobalandlocalenvironments
(α = −0.38 ± 0.06).Wecheckedthatthisdifference
passiveM1206
in slope is not due to the differentcompleteness limits that we To investigate the environmental dependences of the SMF, we
reach in the two works. The difference between the SMF of considered SF and passive galaxies separately. In this way we
A160,page6of17
M.Annunziatellaetal.:A209stellarmassfunction
12.0
10
10.5
)
−2
min) 5 1.800 1.800 9.0 min
c
Y(arcBCG 0 9.000 4.500 67..05 ensity(ar
− 4.5 d
−5 el
Y 3.0 ern
K
−10 RTH 1.5
O
EASTN
15 10 5 0 -5 -10 -15
X − XBCG(arcmin)
Fig.9.Spatialdistributionofclustermembersdowntothestellarmass
limit. The points are colored according to the local number density.
Coordinates refer to the position of the BCG (black star). Red, light
Fig.7.SMFsofSF(left)andpassive(right)galaxiesinsideandoutside greenandbluelinesidentifygalaxiesbelongingtoRegions(a),(b)and
the virial radius of the cluster. The SMFs are normalized to the total (c)definedinthetext.PointsoutsidethecontoursbelongtoRegion(d).
numberofgalaxiesineachsample. Thebluestarrepresentsthepositionofthehighestdensitypeak.
differentregionsaccordingtotheirnumberdensity(intermsof
galaxiesperarcmin−2):
(a) Σ> 9.0;
(b) 4.5< Σ ≤ 9.0;
(c) 1.8< Σ ≤ 4.5;
(d) Σ ≤ 1.8.
It can be seen in Fig. 9 that regions identified by Σ are not
spherically symmetric. Another peculiar characteristic is that
Region(a)isnotcenteredontheBCGandthatRegions(b)and
(c) also incorporate a group in the N-NE direction. Mercurio
et al. (2003a) found that the center of the cluster was coinci-
dentwiththepositionoftheBCG.However,theyusedasample
ofbrighterspectroscopicallyconfirmedmembergalaxies.Their
magnitude limit (R < 22 mag) roughly corresponds to a mass
Fig.8.AsinFig.6:best-fitSchechterparametersM∗ andαwiththeir limit>109.5M(cid:3) (seeFig.4).Ifwelimitouranalysistothissam-
ple of galaxiesthe center is coincidentwith the position of the
1σcontourindifferentregionsofthecluster
BCG andtheregionathighdensityin theNWdirectiondisap-
pears;inotherwords,themorphologicalappearanceoftheclus-
terchangesaccordingtothelimitingmagnitudeofthesample.
areabletoexcludetheeffectsofthedependenceofgalaxypop- The group in the N-NE direction contains very few cluster
ulationfromenvironment(Dressler1980). spectroscopicmembers.Forthisreason,weremoveditfromour
First,weconsideredtheSMFsofSFandpassivegalaxiesin- analysis after checkingafter we checkedthat its presence does
sidethevirialradiusofthecluster(r200)andoutside(∼1.7r200). notaffecttheSMFofgalaxiesindifferentregions.
TheSMFsandtheirbest-fitSchechterparametersareshownin Figure10showstheSMFofpassivegalaxiesinRegions(a)
Figs. 7 and 8. The SMFs of SF galaxiesinside and outside the to(d),andtheirbest-fitparametersarereportedinFig.11andin
virialradiusarestatisticallyindistinguishable(seetheresultsof Table3.Figures10and11showthattheSMFofpassivegalaxies
theK-StestinTable4).Thesameresultholdsforpassivegalax- dependsonthelocaldensity.Theenvironmenthastwoeffectson
iesaswell. theSMFofpassivegalaxiesinthedensestregionofthecluster:
We define the “environment” in units of the local density thereisahighernumberofhigh-massgalaxiesthaninlessdense
(numberofgalaxiesperarcmin−2).Wedecidedtousethisdefini- regions(whichcorrespondstoahighervalueofM∗)butthereis
tionofenvironmentinsteadoftheradialdistancefromtheBCG alsoadropatthelowendoftheSMF.
(Annunziatellaetal.2014)sinceA209isanelongated,dynami- We applied the K-S test separately to the M(cid:2) distributions
callynotfullyrelaxedcluster(Fig.9)andtheregionsidentified ofSFandpassivegalaxiesinadjacentregions.Accordingtothe
bylocaldensityareelongatedwithrespecttothoseidentifiedby resultsoftheK-Stest,inRegions(a)to(d)theM(cid:2)distributions
clustercentricradius. of both SF and passive galaxies are not statistically different.
Thelocaldensityisobtainedwithakerneldensityfunction, However,this can be related to the fact that low-mass galaxies
wheretheprojecteddistributionofthegalaxiesissmoothedwith statisticallydominatethesamples.Ifwerepeatthetestrestrict-
a Gaussian kernel in an iterative way. Figure 9 shows the dis- ing it to the galaxies with M(cid:2)>109.5M(cid:3), we find a significant
tribution of galaxies in the cluster. The points are color-coded difference between the mass distribution of passive galaxies in
according to the value of the local density (Σ). We define four Regions(a)and(b).
A160,page7of17
A&A585,A160(2016)
Fig.10. SMFs of passive galaxies in different density regions. The Fig.12. SMFs of passive cluster member galaxies within and outside
SMFsare normalized tothetotal number of galaxies ineach sample. r200. The red solid and orange dashed lines are the best-fit double
Redcircles,orangetriangles,maroonsquares,andmagentadiamonds Schechterfunctions.Thepointsandtherelativeerrorbarshavethesame
arenumbercountsinRegion(a)to(d). meaningasinFig.7(rightpanel)
.
Fig.11.AsinFig.6:best-fitSchechterparametersM∗andαwiththeir
1σcontourindifferentdensityregionsofthecluster. Fig.13.SMFsofpassiveclustermembergalaxiesinRegions(a)to(d).
Thelinesarethebest-fitdoubleSchechterfunctions.Thepointsandthe
relativeerrorbarshavethesamemeaningasinFig.10.
4.3.DoubleSchecter
EvenifthesingleSchechterisstatisticallyanacceptablefittoall (Mortlocketal. 2015;Baldryetal. 2012).In Fig.12,we show
theSMFspresentedinthispaper,thedipexistinginthenumber the double Schechter fit to the SMF of passive galaxies within
countsofpassivegalaxiesinallenvironmentsandtheupturnin and outside the virialradiusof the cluster, while in Fig. 13 we
thenumbercountsatthelow-massendvisibleinthelowdensity showthedoubleSchechterfitinRegions(a)to(d).Thebest-fit
regionsseem to indicate that perhapsa doubleSchechterfunc- valuesoftheparametersarelistedinTable5.BothFig.13and
tion is more suitable to describe the SMF of this galaxy pop- Table 5 show thatin the densest regionof the cluster the slope
ulation. For this reason we fit a double Schechter function to ofthetwoSchechterfunctionsarestatisticallyindistinguishable,
theSMFofpassivegalaxiesinalltheenvironments.Thedouble meaningthatadoubleSchechterfunctiondoesnotprovideabet-
Schechterfunctionusedhasfourfreeparameters: terfitthanasingleone.Theevidenceoftwopopulationsofpas-
Φ(logM)=ln(10)Φ∗ sive galaxies becomes stronger when movingto outer and less
⎡(cid:7) (cid:8) (cid:7) (cid:8) ⎤ (cid:7) (cid:8) denseregionsofthe cluster.Thisisalso shownbytheincrease
×⎢⎢⎢⎢⎢⎣ M 1+α+f M 1+α2⎥⎥⎥⎥⎥⎦×exp − M d(logM). (2) ofthedifferencebetweenthetwoslopesoftheSchechterfunc-
M∗ 2 M∗ M∗ tions.ThesecondcomponentintheSMFofpassivegalaxiesis
hypothesizedinthemodelofPengetal.(2010b),asaresultof
Compared to the single Schechter function there are two addi- the environmental quenching of SF galaxies. The fact that we
tional free parameters: the slope of the second component α , observeasecondcomponentthatisstrongerintheouterregions
2
and the ratio between the normalizationsof the two functions, couldmeanthattheenvironmentalprocessesresponsibleforthe
f =Φ∗/Φ∗. We considerthe same M∗ forthe two functionsto quenchingaremosteffectiveintheexternalregionsoftheclus-
2 2
minimize the number of free parameters and to follow the ap- teror thatin thecenterthesemechanismsarestrongenoughto
proach already used for the study of the SMF of field galaxies completelydestroythispopulationofgalaxies.
A160,page8of17
M.Annunziatellaetal.:A209stellarmassfunction
Table5.Best-fitparametersofthedoubleSchechterfunctionofpassivegalaxiesinallenvironments.
Environment α log(M∗/M(cid:3)) f2 α2
Passive −0.61±0.25 10.96±0.08 0.76±0.25 −1.18±0.23
r ≤ r −0.93±0.27 11.08±0.07 0.76±0.33 −1.18±0.28
200
r > r −0.05±0.34 10.77±0.10 0.44±0.25 −1.21±0.10
200
Region(a) −1.01±0.14 11.21±0.12 0.65±0.41 −1.01±0.22
Region(b) −0.94±0.08 11.06±0.10 0.34±0.35 −1.24±0.26
Region(c) −0.41±0.35 10.93±0.10 0.81±0.32 −1.19±0.09
Region(d) −0.34±0.36 10.70±0.11 0.10±0.23 −1.48±0.29
)(cid:3) 11.3 strippedbythesegalaxiescouldcontributetotheICL.Tocheck
M
11.2 SingleSchechter ifthishypothesisholdsinthisclusteraswellweneedtoobtain
/
∗M 11.1 anICLmapandanestimateoftheICLcolorandmass.
g( 11.0 Intra-clusterlightisadiffusecomponentthatoriginatesfrom
o
l starsboundtotheclusterpotential;thesestarsaretidallystripped
−1.10
from the outer regions of galaxies that interact with the BCG
−1.05
α or other cluster members (Murante et al. 2007; Contini et al.
−1.00
−0.95 SingleSchechter 2014). Some properties of the ICL, such as the radial profile,
thecolor,andpresenceofsubstructurearestrictlylinkedtothe
) 11.4
(cid:3) accretionhistoryanddynamicalevolutionofgalaxyclusters.For
M 11.2
/ 11.0 example,thepresenceoftidal arcscangiveinformationonthe
∗
M 10.8 assemblyhistoryofthe cluster.If theICL is formedbeforethe
g( 10.6 DoubleSchechter cluster has undergonevirialization,the bulk of the ICL will be
o
l smoothwithfewfainttidalarcs.Accordingtotheoreticalmodels
−1.0
−0.8 thesefeaturesaretheresultofrecentinteractionamonggalaxies
α1 −−00..64 with the cluster potential (Rudick et al. 2009). The color can
−0.2 DoubleSchechter alsogivesomehintsonwhicharethemainprogenitorgalaxies
0.0
oftheICL.EveniftheICLcanshedlightonmanymechanisms
−1.7
occurringingalaxyclusters,isverydifficulttostudyitowingto
−1.5
2 −1.3 DoubleSchechter itsveryfaintsurfacebrightness,i.e.,∼1%oftheskybrightness.
α
−1.1 Furthermore, the detection of the ICL at high redshift is made
−0.9 even more difficult by the surface brightness dimming, which
scaleswithredshiftas(1 + z)4.
0.65
R ToobtaintheICLmapwefollowedthemethoddevelopedby
N0.55
GS0.45 Presottoet al. (2015).Inthe followingsections, we summarize
themainaspectsofthismethod.
0.35
0 2 4 6 8 10 12
Σ(numberofgalaxies/arcmin2)
5.1.ICLdetectionmethod
Fig.14.Fromtoptobottom:best-fitparametersofthesingleSchechter
There is no consensus on the method for the detection of the
(panels1and2)andofthedoubleSchechterfunctions(panels3−5)for
ICL.Ideally,oneshouldsubtractthelightcontributionfromall
passive galaxies and the GSNR (panel 6) as a function of the cluster
galaxiesintheclusterincludingtheBCG.However,itisnotal-
localnumberdensity.Regions(d)to(a)fromlefttoright.
wayspossibletoperfectlyfitthelightdistributionofeachgalaxy.
For this reason, in many works the galaxies are masked down
In Fig. 14, we show the parameters of the SMF of passive to a certainlimitin surfacebrightness.The approachwe use is
galaxiesasafunctionofthelocaldensityofthecluster.Wealso a hybridmethod which first fits a Sérsic profile to each galaxy
showtheratiobetweenthenumberofgiantandsubgiantgalax- andsubtractsthebest-fitmodelwheneverpossible,thenitmasks
ies(GSNR)asafunctionoftheenvironment.Wedividethegi- all the pixels with high residuals. We use this method to ana-
ant and subgiant galaxies using a mass threshold of 1010.0M(cid:3), lyze the RC image of the cluster. Since the region affected by
i.e. the mass value at which the SMFs of passive galaxies in the ICL is ∼500kpcaround the BCG, we limit our analysis to
Region(a)and(b)intersect.ThetrendintheGSNRshowsthat a squaredimage ∼1 Mpc in size centeredon the BCG. We use
in Region (a) the ratio between the number of giant and sub- an automated procedure that relies on the GALAPAGOS soft-
giantgalaxiesishigherthaninlessdenseregions.Thistrendis ware (Barden et al. 2012). The main goal of GALAPAGOS is
consistentwiththeonewefoundinM1206. to obtain a Sérsic fit for all the sources detected in the image
bySExtractor(Bertin& Arnouts1996).Thefits are performed
bytheGALFITcode(Pengetal.2010a).Detailedexplanations
5. Intra-clusterlight
of how GALAPAGOS works is given in Barden et al. (2012).
In Annunziatella et al. (2014) we found that the SMF of pas- Here,we brieflydescribeonlythestepsthatare crucialforour
sive galaxies in the innermost and densest region of the clus- analysis.
terM1206declinesatthelow-massend.Wesuggestedthatthis The detection of the sources is performedby SExtractor in
steeper drop could be explained by the tidal stripping of sub- two-stepruns:onecalled“cold”,whichistunedtoproperlyde-
giant galaxies (M(cid:2) < 1010.5M(cid:3)). In our scenario, the material tectanddeblendonlybrightsources,andtheothercalled“hot”,
A160,page9of17
A&A585,A160(2016)
whichisfocusedonthedetectionoffaintones.Thetwodetec- Table6.Best-fitparametersofthethreecomponentsoftheSérsicfitof
tions are then combined by rejecting the faint sources that are theBCG+ICL.
located inside the cold source as determined by the Kron el-
lipse (Kron 1980). The input parameters are chosen to detect Component R n mag
e tot
in the hotmode sourceswith S/N > 1σ . For each detected (h−1kpc) (ABmag)
sky 70
source, GALAPAGOS creates a stamp image large enough to 1(BCG) 0.86 0.5 20.07
include neighbor galaxies and centered on the source posi- 2(BCG) 44.28 3.07 16.15
tion. Then,GALFIT simultaneouslyfits all the sourcesin each 3(ICL) 120 1.51 16.57
stamp. The starting parameters for the fits are the SExtractor
estimate of source position (X_IMAGE, Y_IMAGE), the source
total magnitude (MAG_BEST), its effective radius (function of We used the magnitude value obtained from the fit to esti-
matetheICLmass.Weperformabi-weightedfitoftherelation
FLUX_RADIUS),anditspositionangle(THETA_IMAGE).The
between the stellar masses of red cluster member galaxies that
outputparametersfromthefitsandthoseobtainedbySExtractor
are in the subsample image analyzed with GALAPAGOS and
arethencombinedinafinalcatalog.ThenweusetheGALtoICL
theirtotalR magnitudeprovidedbyGALFIT:
procedure presented in Presotto et al. (2015). This procedure C
merges the stamps created by GALAPAGOS to form a global log(M/M(cid:3)) = (18.41±0.07)−(0.41±0.02)×RCtotmag. (3)
modelimage which is later subtracted from the originalimage
to obtain a residual image called BCG+ICL map. The sources Using this relation with the total RC magnitudes ob-
tained from the multicomponent Sérsic fit, we estimate
thathavehighresidualsareidentifiedbycomparingthedistribu-
tion of the residualswith the distributionofthe pixelvaluesof MICL = (2.9±0.7)× 1011M(cid:3) and MBCG = (6.2 ± 1.2) ×
askyregion.Thepixelswhosevaluesdeviateatdifferentσlev- 1011M(cid:3). The mass of the BCG obtained in this way is con-
els from the sky (1, 2, 3, 4, 5σ ) are flagged and associated sistent with the mass obtained when applying MAGPHYS, i.e.
tothecorrespondingsource.Thesskeypixelscanbemaskedorone 6.2±1.2 × 1011M(cid:3).
To check if the ICL can form from tidal stripping of sub-
thesourcescanbemanuallyre-fitthetothecorrectlevel.Before
the BCG+ICL map is created one can also decide to manually giants galaxies (i.e. galaxies with M(cid:2) <1010.0M(cid:3)), we have to
evaluatethemissingmassfromsubgiantgalaxiesinRegion(a).
addmasks(forexampleincaseofstellarspikes).
Themissingmassfromsubgiantgalaxiesisdeterminedwiththe
InFig.15,weshowtheR imageofA209(firstpanel),the
C
globalmodelimage(secondpanel)andtheBCG+ICLmapwith equation
pixelsat1σsky level(thirdpanel).Wecancheckifthesubtrac- (cid:12) 1010.0M(cid:3) (cid:12) 1010.0M(cid:3)
tionofthelightfromthesourceswasdoneinanefficientwayby ΔM ≡ MΦ∗(M)dM− MΦ (M)dM, (4)
sub a a
countingthe percentageof maskedpixels. In ourcase thisper- 108.6M(cid:3) 108.6M(cid:3)
centagevariesfrom7%to10%ifwemaskat5σskyor1σsky. whereΦa istheSMFofpassivegalaxiesinRegion(a)notnor-
malized, while Φ∗ is the SMF of passive galaxiesif we substi-
a
tute the value of the slope with the best-fit value obtained for
5.2.ICLproperties
the SMF in Region(b). The integralis donebetween the com-
We estimate the B − R color of the BCG+ICL map. To do pletenessmasslimitandthemassatwhichtheSMFsofpassive
C
this we take as a benchmarkthe global model in the RC-band. galaxiesinRegions(a)and(b)intersecteachother(∼1010.0M(cid:3)).
From this we obtain a model for the B-band by changing the The upper and lower error on ΔM are estimated by
sub
zeropointand fitting only the magnitudesof the objects. Then, repeating the evaluation of Eq. (4) and fixing the slope to
wgeetstuhbetrBa-cbtatnhdisBmCoGde+lIfCrLommtahpe.BA-tbathnidsipmoaingte,owfethdeegclruasdteerthtoe αof±ΔMδα. TihsefvoarlmuealloyfiΔnMasgubreiesm1e.n4t−+00w..75ith×M1011Mw(cid:3)i.thTinhe1vσa.luAe
sub ICL
BCG+ICL map in the R -band to the PSF of the B-band by similar agreement between ΔM and M was also found
C sub ICL
convolving it with a Gaussian function whose σ is the differ- in M1206. Breaking the integral in Eq. (4) we can estimate
encebetweenthetwoPSFsandfromthiswesubtracttheB-band the contribution to ΔM coming from galaxies in different
sub
BCG+ICLmap.TheB−R coloroftheBCG+ICLis∼2.1mag mass ranges: the main contribution is given by galaxies in the
C
at the position of the BCG while it drops to 1.2 mag at a dis- massrange109.5 < M(cid:2) <1010.0M(cid:3)(60%),galaxiesinthemass
tancefromtheBCG of∼110kpc.Inthisradialrange,thecolor range109.0 < M(cid:2) <109.5M(cid:3) contribute∼30%,while galaxies
oftheBCG+ICLremainsinthecolorrangeofthepassivecluster in the range 108.6 < M(cid:2) <109.0M(cid:3) only 10%. This suggests
members(seeFig.4). thattheICLcouldoriginatefromthestrippingofstarsfromsub-
ThesurfacebrightnessprofileofBCG+ICLdeviatesfroma giant galaxies in the mass range 109.0 < M(cid:2) <1010.0M(cid:3). On
singleSérsicprofile(Presottoetal.2015).Forthisreasonwefit the contrary,this result seems to rule outthat the disruptionof
theBCG+ICLlightprofilewithasumofmultiplecomponents. dwarf galaxies is the main channel for the formation of ICL,
Moustakasetal.(inprep.)modelthesurfacebrightnessprofileof which agrees with the predictions from semi-analytical mod-
theBCGsofeachCLASHcluster.Theyfindthatthelightprofile els (Contini et al. 2014). Our result is also consistent with re-
oftheBCGofA209,contrarytotheotherBCGsinthesample, centfindingsfromobservationalworks.Indetail,DeMaioetal.
deviates from a single Sérsic profile. They perform the fits in (2015),useasubsampleofCLASHclustersat0.44 ≤ z ≤ 0.57
all HST bandpasses and assume a universal value of nandR . todeterminethecolorgradientsandtotalluminosityoftheICL.
e
We chose to fit the BCG+ICL profile with three components: Their main conclusionis that the ICL originatesfrom the tidal
twocomponentsfortheBCGswiththeeffectiveradius(R )and destruction of 0.2 L* galaxies (which corresponds to galaxies
e
the Sérsic index fixed to the values found in Moustakas et al., of ∼1010.0 M(cid:3)). In a completely independent way, Longobardi
and another one for the ICL. The best-fit parameters for the etal.(2015)assessthatthePlanetaryNebulaeintheintra-cluster
three components are listed in Table 6. The position angles of mediumoftheVirgoclustercanbeassociatedwithagalaxypro-
thethreecomponentsareconsistentwithintheuncertanties.The genitorthatisfourtimesthemassoftheLargeMagellanicCloud
reducedχ2ofthefitis1.535. (∼109.5M(cid:3)).
A160,page10of17