ebook img

K+a galaxies in the zCOSMOS Survey: Physical properties of systems in their post-starburst phase PDF

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

Preview K+a galaxies in the zCOSMOS Survey: Physical properties of systems in their post-starburst phase

Astronomy&Astrophysicsmanuscriptno.0000 (cid:13)c ESO2010 January21,2010 ⋆ K+a galaxies in the zCOSMOS Survey: Physical properties of systems in their post-starburst phase D.Vergani1,2,G.Zamorani2,S.Lilly3,F.Lamareille4,C.Halliday5,M.Scodeggio6,C.Vignali2,P.Ciliegi2, M.Bolzonella2,M.Bondi7,K.Kovacˇ3,C.Knobel3,E.Zucca1,K.Caputi3,L.Pozzetti1,S.Bardelli1,M.Mignoli1, A. Iovino8,C.M.Carollo2,T.Contini4,J.-P.Kneib9,O. LeFe`vre9,V.Mainieri10,A.Renzini11,A.Bongiorno12, G.Coppa2,O.Cucciati9,S.delaTorre8,L. deRavel9,P.Franzetti6,B.Garilli6,P.Kampczyk2,J.-F.LeBorgne4,V.Le Brun9,C.Maier2,R. Pello4,Y. Peng2,E. PerezMontero4,E. Ricciardelli11,J.D.Silverman2,M. Tanaka10, 0 L. Tasca6,L. Tresse9,U.Abbas13,D. Bottini6,A. Cappi1,P. Cassata9,A. Cimatti2,L. Guzzo8,A.M.Koekemoer14, 1 A. Leauthaud15,D. Maccagni6,C. Marinoni16,H.J. McCracken17,P. Memeo6,B. Meneux12,18,P. Oesch2, 0 2 C. Porciani19,R. Scaramella20,P.Capak21,D.Sanders22,N.Scoville21,andY.Taniguchi23 n (Affiliationscanbefoundafterthereferences) a J Received30June2009;accepted9September2009 1 2 ABSTRACT ] O Aims. Theidentitiesofthemainprocessestriggeringandquenchingstar-formationingalaxiesremainunclear.Akeystageinevolution,however, appearstoberepresentedbypost-starburstgalaxies.Toinvestigatetheprevalenceofthesegalaxiesandtheirimpactongalaxyevolutioningeneral, C weinitiatedamultiwavelengthstudyofgalaxieswithk+aspectralfeaturesinthewell-studiedCOSMOSfield. . Methods.Weexamineamass-selectedsampleofk+agalaxiesintheCOSMOSfieldatz=0.48−1.2usingthespectroscopiczCOSMOSsample. h Toclassifygalaxiesintheirk+aphase,weuseaspectroscopiccriterion,basedontheamplitudeoftheHδabsorptionlineandtheabsenceof p the[OII]emissionline.Wedevelopouranalysisforawell-definedsamplebyimposingstringentconfidencelevelsonthespectroscopicredshifts - o andspectralmeasurements.Wecompareourresultsfortwomass-selectedsamplesofstar-formingandquiescentgalaxiesselectedusingapurely r spectralclassificationschemefromthe10,000zCOSMOScatalogue(i.e.,basedonmeasurementsof4000ÅbreakandEW[OII]). t s Results.In our mass-limited sample, k+a galaxies occupy the brightest tail of the luminosity distribution. They are as massive as quiescent a galaxiesandpopulatethegreenvalleyinthecolourversusluminosity(orstellarmass)distribution.Asmallpercentage(<8%)ofthesegalaxies [ haveradioand/orX-raycounterparts(implyinganupperlimittotheSFRof∼8M yr−1).Overtheentireredshiftrangeexplored,theclassofpost- ⊙ starburstgalaxiesismorphologicallyaheterogeneouspopulationwithasimilarincidenceofbulge-dominatedanddiskygalaxies.Thisdistribution 2 doesnotvarywiththestrengthoftheHδabsorptionlinebutinsteadwithstellarmassinawayreminiscentofthewell-knownmass-morphology v relation.TheresultsabouttheincidenceofasymmetriesandtheconcentrationofthelightdistributionderivedfromHST/ACSimagesimplythat 8 thisgalaxypopulationpossiblyrepresentsanintermediatestageofgalaxyevolution.Althoughk+agalaxiesarealsofoundinunderdenseregions, 6 they appear to reside typically in a similarly rich environment as quiescent galaxies on a physical scale of ∼ 2−8 Mpc, and in groups they 9 showamorphologicalearly-to-latetyperatiosimilartothequiescentgalaxyclass.Withthecurrentdataset,wedonotfindevidenceofstatistical 1 significantevolutionineitherthenumber/massdensityofk+agalaxiesatintermediateredshiftwithrespecttothelocalvalues,orthespectral . 9 properties,althoughmoresolidresultsonthisandotheraspectswillbeobtainedfollowingthecompletionofthesurvey. 0 Conclusions.Severalmechanismsrelatedandunrelatedtotheenvironmentareatworkinquenchingstar-formationactivityingalaxiesonshort 9 timescales(<1Gyr).Thosegalaxies,whichareaffectedbyasuddenquenchingoftheirstar-formationactivity,mayincreasethestellarmassof 0 thered-sequencebyuptoanon-negligiblelevelof∼10%. : v Keywords.Galaxies:formation-galaxies:evolution-galaxies:fundamentalparameters-galaxies:massfunction-cosmology:observations i X r a 1. Introduction Brinchmann&Ellis 2000), observational evidence exists that manygalaxieshaveexperiencedmorphologicaltransformations Numerous studies have demonstrated that a bimodal popula- at various redshifts. Brinchmann&Ellis (2000) reported that tion of galaxies exists up to high redshift (e.g., Franzettietal. a decline has occurred in the mass density of irregular galax- 2007; Mignolietal. 2009). The first population consists of ies between z ≃ 1 and today, in contrast to an increase for galaxies that are actively star-forming, gas-rich, rotationally- regular galaxies. Further support of this redistribution between supported, and morphologically disk-dominated. The second galaxy types has been provided by study of the stellar mass population represents typically quiescent galaxies that are function by many different surveys, e.g., K20 (Fontanaetal. gas-poor, pressure-supported, and have spheroidal morpholo- 2004), COMBO-17 (Borchetal. 2006), DEEP2 (Bundyetal. gies. As noted by many studies (e.g., Boweretal. 1998; 2006), SWIRE-VVDS-CFHTLS (Arnoutsetal. 2007), SDSS (Martinetal. 2007), and VVDS (Verganietal. 2008). It is im- Send offprint requests to: D.Vergani, e-mail: portant to assess the role of environment in these transforma- [email protected] tionsofgalaxytype.Forinstance,ithasbeenshownthatgalaxies ⋆ Based on data obtained with the European Southern Observatory in denseregionshavemodifieddistributionsof HI,the primary VeryLargeTelescope,Paranal,Chile,program175.A-0839. 2 D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield source of fuel for star-formation, and truncated star-formation vational constraints, the majority of studies cover small areas activity (e.g., Giovanelli&Haynes 1985; Cayatteetal. 1990; and might probe specific types of environments (large-scale Koopmann&Kenney 2004). However, several other mecha- structures in pencil surveys, or studies in clusters). In addition nisms may also contribute (see below). Thus, important ques- to the observational limitations, finding post-starburst galaxies tionsingalaxyformationandevolutionarewhetherstar-forming is also challenging because of the short duration of the post- galaxies are the precursors of the quiescent galaxy population starburst phase (1–1.5 Gyr) (see Couch&Sharples (1987) and observedatthepresentday,orwhethertherearealternativepaths Bargeretal.(1996)).Becausetheyrepresentasmallfractionof in galaxyformation,and, if so, whatare the primaryprocesses thetotalgalaxypopulation,thesystematiceffectscausedbythe involvedandhowaretheycontrolled? aforementioned problems require particular attention when in- Themainlimitingfactorinansweringthisquestionisthatwe terpretingtheresultsinageneralframework. still have notrue understandingof whattriggersstar formation In this paper, we present for the first time a mass- ingalaxiesandwhythisactivityissuppressedinsomeofthem. selected sample of post-starburst galaxies at intermediate red- There is a class of galaxiesthat might shed light on this issue. shift in a wide variety of different environments. This mass- Theyarecalledk+aorE+Agalaxies(dependingontheauthor), selected sample and its parent galaxies are extracted from orsimply galaxiesin the post-starburstphase.Inthis work,we the largest multiwavelength survey existing up to now, the adoptthespectralterminologyk+ainsteadofthemorphological Cosmic Evolution Survey, or COSMOS (Scovilleetal. 2007) E+Anomenclature. imagedwithHST/ACS(Koekemoeretal.2007).Weprovidean Post-starburst galaxies, discovered by Dressler and Gunn overviewofthemainphysicalpropertiesofthesegalaxiesusing (1983), exhibit peculiar spectral characteristics, i.e., strong data of unprecedented quality and completeness in this study. Balmerabsorptionlines,indicativeofanintensestar-formation We presentafirstattempttoquantifytheircosmicevolution.A epochinthepastbillionyears,andanabsenceofemissionlines, moredetailedstudyofthissubjectanditsrelationtothegeneral asignatureofalackofongoingstar-formation.Onthebasisof pictureofgalaxyevolutionoverthepast8Gyrwillbepresented these spectral signatures, they have been assumed to represent oncompletionofthesurvey. galaxies in a transitional stage between being a star-forming, Thepresentworkisorganizedasfollows:thedataandsam- late-typegalaxyandapassive,early-typesystem. ple selection are presented in Sect.2; results are described in Variousmechanismshavebeenproposedtoexplainthechar- Sect.3, and a summary is given in Sect. 4. Throughout this acteristic spectral features of post-starburst galaxies. Cluster- work,weassumeastandardcosmologicalmodelwithΩM =0.3, relatedmechanisms,suchasram-pressuregasstripping,harass- ΩΛ = 0.7,and H0 = 70kms−1Mpc−1.Magnitudesaregivenin ment, or strangulation (Gunn&Gott 1972; Larsonetal. 1980; theABsystem. Balogh&Morris 2000) are efficient in suppressing the star- formation activity of a galaxy. Events such as galaxy mergers 2. Sampleselection and interactions (Toomre&Toomre 1972; Barnes&Hernquist 1992), commonly found in both the field and in clusters (al- 2.1.Thespectroscopicsurvey:zCOSMOS though more efficient in the field), have been proposed. The large fraction of morphological irregularities (of similar phys- ThezCOSMOSproject(Lillyetal.2007)isaredshiftsurveyof ical timescale as a post-starburst phase) and galaxy compan- galaxiesintheCOSMOSfieldconsistingoftwoobservingpro- ions in the vicinity of these galaxies, support the hypothesis grams.The “deep part”focuseson high-zgalaxies(1.4 < z < thateitheradynamicalinteractionoramergerwithexchangeof 3.0) using a combined flux-limited (BAB < 25.25) and colour material may have suppressed star formation. Strong AGN/SN selection(theseobservationsarenotusedinthepresentstudy). feedback might also contribute to star-formation quenching The “bright part” of zCOSMOS project aims to acquire spec- (Springeletal. 2005; Hopkinsetal. 2007), depending on the traofabout20,000galaxies(IAB < 22.5)witharatheruniform stellarmass(Kavirajetal.2007). sampling rate (about 60–70%) across the 1.7 deg2 COSMOS This galaxy population was first discovered, and has been field, a superb spectroscopic redshift success rate (> 97% at primarilystudiedsince,insideclusters(Dressler&Gunn1983; 0.5 < z < 0.8), and high velocity accuracy (∼ 100kms−1). Couch&Sharples1987;Tranetal.2003;Poggiantietal.1999, The brightpartis observedwith the red R ∼ 600VIMOSMR 2009), even if post-starburst galaxies have also been found grism covering the spectral range 5500–9500Å (2.55Å disper- in the field. Numerous studies have been undertaken to study sion), which provides an ideal configuration for detecting key the properties of field post-starburst galaxies at low red- spectralfeaturesingalaxiesouttoz∼1.2.Thispaperdealswith shift(Liu&Kennicutt1995;Zabludoffetal.1996;Changetal. the first “10k” galaxydata set completed to date for the bright 2001; Nortonetal. 2001; Goto&SDSScollaboration 2003; partofthezCOSMOSsurvey,whichwerefertohereafterasthe Quinteroetal. 2004; Yangetal. 2004), and at z > 0.3 zCOSMOSsurvey.Alldetailsofdataacquisition,reduction,red- (Hammeretal. 1997; Dressleretal. 1999; Poggiantietal. shiftmeasurements,andtheirqualitycanbefoundinLillyetal. 1999; Baloghetal. 1999; Tranetal. 2004; Yangetal. 2008; (2007)andLillyetal.(2009),whichalsodescribesotherimpor- Poggiantietal. 2009). These studies commonlyconcludedthat tantaspectsofthissurvey. post-starburstgalaxiesareaheterogeneousparentpopulation. Post-starburst galaxies appear to be rare at any epoch and 2.2.Thespectralmeasurementsandsampleselection discrepantresults abouttheir nature are found in the literature. The identifiable sources of discordance are numerous. Besides Wemeasurethespectralfeaturesinour10ksamplewithanau- differencesinselectioncriteriaandspectralresolution,themain tomatedsoftwarepackage(platefit VIMOS).Hereweprovidea source of discrepancy may be the different signal-to-noise ra- briefoutlineofthefittingprocedure.WerefertoLamareilleetal. tio achieved at different redshifts. This implies that we should (2006) for full details, and both Tremontietal. (2004) and also investigate the dependency of the underlying properties Brinchmannetal. (2004) for further details of the original on the host galaxies (e.g., the sampling of only the most mas- pipelinedevelopedforthehigherresolutionSDSSspectra.The sive or brightest systems). Furthermore, because of the obser- best-fit stellar continuum derived from a grid of stellar pop- D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield 3 In this paper, we use measurements of the spectral in- dex Hδ as defined by Worthey&Ottaviani (1997) and A Baloghetal. (1999), the equivalent width of the [OII]λ3727 doublet (EW[OII]), the equivalent width of the [OIII]λ5007 (EW[OIII]), and the amplitude of the 4000 Å break (D 4000) n (Baloghetal. 1999). A careful treatment of spectral fitting is mandatory because both emission and absorption lines are present in the Balmer lines. The index Hδ and the equivalent A width of the lines are defined such that positive valuesrefer to absorptionfeatures. AuthorsusuallyadoptHδAandEW[OII]featuresintheircri- teriaforselectingpost-starburstgalaxies,and/orHα depending on the spectral coverage, although some exceptions exist. For example, Tranetal. (2004) used a combination of Hδ and Hγ indices,in additionto [OII],Yanetal. (2006) addedanempiri- callimit to EW(Hβ) in constrainingEW[OII],and someothers used, in addition to a threshold on emission lines, the relative abundance of A-type to K-type stars, i.e., the ratio A/K (e.g., Quinteroetal.2004).Inthispaper,weadoptaconventionaldef- initionusingboththe[OII]emissionlineandHδA,tobeableto comparewiththemajorityofotherworks,althoughanycompar- isonisdifficultandonlypossiblequalitatively.Inafuturepaper, basedonthe20kgalaxysample,wewillexaminetheeffectsof different selection criteria on a larger sample of Hδ-enhanced galaxies(e.g.,LeBorgneetal.2006)andthepossiblecontribu- tionfromobscuredAGNs(Yanetal.2006).Inthepresentpaper, ourpost-starburstcandidatecriteriaarethattheequivalentwidth ofHδ islargerthan+3Åandtheequivalentwidthoftheemis- A sion line [OII] is larger than −3Å. This is illustrated in the top panelofFig.1,wheretheequivalentwidthofHδ isplottedver- A susEW[OII]forthe10k-parentgalaxiesintheredshiftinterval 0.48 ≤ z ≤ 1.2(theredshiftbinexploredinthiswork,whichis imposedbyourinstrumentalset-up).Thisplotshowsthespec- troscopic criteria adopted to select the post-starburst candidate galaxies.AsisevidentinFig.1,theregioncorrespondingtoour k+acandidatesisnotcompletelyseparatefromthatofthe10k- parent population as it is in the SDSS data (cf. the left panel of Fig.2 by Yanetal. (2009)). This is mainly due to the errors associatedwiththespectralmeasurements.Toevaluatetheirim- pact on the classification scheme, we tested differentempirical thresholds and applied a confidence level to the spectral fea- tureswhendefiningoursamples.Thisapproachaimstoexclude (include) residual signatures of ongoing star formation and to Fig.1.(Top)DistributionofequivalentwidthsofHδvs.EW[OII] testthereliabilityofourspectralmeasurements.Whilethenum- for all galaxies in the 10k-parent sample in the redshift inter- bersofobjectschangedependingonthethresholdsadopted,of val 0.48 ≤ z ≤ 1.2. This plot shows the criteria adopted to course, we obtained an optimal compromise between the abil- select the post-starburst candidate galaxies, i.e., the region be- ity to completea robust,statistical analysis and reliable results tween HδA> +3Å and EW[OII]> −3Å marked with dashed by including only galaxies with EW[OII] larger than -3Å and lines. Isophotal contours are in steps of 10% with the faintest EW(Hδ)largerthan3Åatthe2σconfidencelevel. isophotelevelstartingatthe10%level.(Bottom)Templatespec- trum of a typical galaxy in a post-starburst phase obtained by To maintain strong control on occasional poor wavelength stacking 69 zCOSMOS post-starburst galaxies more massive calibration that can affect in particular the bluest part of our than log(M/M )> 10.0 (corrected by selection effects as ex- spectra, we visually inspected each spectrum of potential can- ⊙ plainedinSect.2.5). didates. For spectra with doubtful measurements of EW[OII], we imposed an additional limit on the equivalent width of the EW[OIII]line.Weusedasubsampleofhighsignal-to-noise10k spectra and high confidence spectral measurements in which ulation synthesis models (Bruzual & Charlot 2003, hereafter bothEW[OII]andEW[OIII]linesweredetectableinthespectra, BC03)wassubtractedfromtheobservedspectrum.Anyremain- to define this limit on the [OIII] lines. From the best-fit model ing residual was removed by fitting a low-order polynomialto solution, we inferred a threshold for the EW[OIII] of −6.9Å. thecontinuum-subtractedspectrum,andallemissionlineswere Furthermore,fromallofourcatalogues,weremovedbothstars then fitted simultaneously with a Gaussian profile. Finally, ab- and broad-line AGNs and adopted a high-quality flag scheme sorption features and spectral breaks were measured after sub- for the spectroscopic redshifts (> 95.5% confidencelevel) (for tractingemissionlinesfromtheoriginalspectrum. details see Lillyetal. (2007) and Lillyetal. (2009)). Although 4 D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield weinspectedeachspectrumindividually,ourconfidencelevelin theredshiftmeasurementsis100%forourpost-starburstsample. We also performed quality checks of our post-starburst candi- datespectratoverifytheimpactofresidualskylinesonspectral features.WeexaminedthereducedVIMOSspectrumfortheim- pactoffringingeffects,whichmaybesignificantatwavelengths above λ > 8,500Å because of the thinned, back-illuminated CCDsusedinVIMOS(Scodeggioetal.2005). The final sample presented in Sect.3 contains spectra that passedalloftheaforementionedtests. Usingtheconstraintsdescribedabove,weobtainasampleof 74 post-starburstgalaxiesin the flux-limited zCOSMOS bright sampleintheredshiftintervalz∈[0.48−1.2].Threeofthemare atredshiftshigherthanz>1.Foroneobject,thecomputationof a precise absolute magnitudeis impossible because of the lack of sufficientphotometriccoverage,and it is excludedfrom our statisticalanalysis(fordetailsseeinZuccaetal.(2009)). 2.3.Post-starburstspectraltemplate Giventhedifficultiesinselectingthisrarepopulationofgalaxies, weprovideatemplatespectrum1 tothecommunityofthepost- starburstgalaxypopulationobtainedbystacking69zCOSMOS post-starburstgalaxiesmoremassiveoflog(M/M )>10.0(cor- ⊙ rectedbyselectioneffectsasexplainedinSect.2.5).Adescrip- tion of the steps performed to obtain this template spectrum is Fig.2. Stellar massand B-bandabsolute magnitudeas a func- giveninthefollowingtext. tion of redshift of the zCOSMOS bright sample (points, gray The bottom panel of Fig.1 shows the template spectrum, symbols). The (violet) triangles show the location of post- which exhibits the distinctive spectral features of galaxies that starburstgalaxies.Thelightanddarkgrayareasshowtheselec- have recently experienced a starburst before its star-formation tioninstellarmassadoptedinsampleMCORandsampleMST, wasquenchedpriortoobservations.Thesefeaturescanbesum- respectively. The dashed (black) line in the upper panel illus- marizedasfollows: tratesthecompletenessinmass-to-lightratiofortheglobalpop- i) The strength of the [OII] nebular emission line, which is ulation(for comparison,it is plotted with a dotted,red line the emitted in HII regions around O and B stars, is almost absent completeness limit for the early-type population). The dashed (black) line in the bottom panel is plotted as reference to il- (EW[OII]<−2.5Å)inourtemplatespectrum.Giventhat[OII]is anindicatorofrecentstarformation(withlifetimeof107yr),this lustrate that our galaxies of sample MST are brighter than the evolving-luminositylimitatM <−20.1−z. propertyindicatesthatthestarformationduringtheepochofthe B observationhasstopped.TheHαemissionlineisanotherrobust indicatorofongoingstarformation,butcannotbedetectedwith ThecombineduseofthetwoSFRestimators(Hδand[OII]), our adopted R ∼ 600 VIMOS MR-grism observational setup, which are sensitive to stars of differentstellar masses and life- whichcoversthespectralrange5500Å<λ<9500Å. times,allowsustoobtaininthenextsectionaquiteprecisecos- ii) Another typical spectral feature of this galaxy populationis micclockoftheirrecent(<1Gyr)star-formationhistory,andto thestrongabsorptionfeaturesofBalmerlines.AlthoughHαand characterizetheirphysicalproperties. Hβ are often filled in by nebular emission that can make their interpretation difficult, Hδ can be used to estimate the stellar ages.Fromlate-typestarstoearlier-typestars, theHδstrength- 2.4.Stellarmasses ens.ThiscanbeachievedifA-starsdominatethemain-sequence Since we studied galaxies that are understood to have expe- contributionneartheepochoftheobservation. rienced a recent secondary burst of star formation, we used iii) As noted originally by Rose (1984), the ratio of the equiv- Charlot&Bruzualmodelswiththeinclusionofsecondarybursts alentwidthsoftheCaIIHλ3968+Hǫ toCaIIKλ3934Åspectral (Charlot&Bruzualinprep.,orCB07).Thesemodelshavebeen linesisconstantinstarslaterthanaboutF02,butincreasesdra- frequently used to model the complex star-formation history maticallyforearliertypestarsastheCaII linesweakenandHǫ of a galaxy more realistically than a purely exponentially de- strengthens.Thisspectralpropertyisalsoillustratedinourtem- clining law (Kauffmannetal. 2003a; Brinchmannetal. 2004; plate spectrum, where the CaII Hλ3968+Hǫ is deeper than the Salimetal.2005). CaIIKλ3934Åabsorptionlines. Stellarmasseswerederivedbyfittingstellarpopulationsyn- iv)Wenotethelackofemissionfrom[OIII]andHβlinesinour thesismodelsfromthelibraryofCB07tothebroad-bandoptical templatespectrum,whichindicatesthatnarrow-lineSeyfertsdo (CFHT:u,i,Ks;Subaru:B,V,g,r,i,z;Capaketal.2007)and notrepresentanimportantcomponentofoursample,incontrast near-infrared(Spitzer/IRAC:3.6µm,4.5µm;Sandersetal.2007) tootherworks(e.g.,Yanetal.2009). photometryby minimizingthe chi-squaretest statistic for each galaxy.Theestimateofthestellarmassprovidedbythebest-fit 1 This template can be accessed from the web page modelspectralenergydistributionisthecurrentmasscontained http://www.bo.astro.it/∼daniela/template.fits. instarsineachgalaxyattheepochoftheobservationobtained D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield 5 byintegratingthestarformationhistoryoverthegalaxyageand complex scheme that introducesa redshift dependenceof SSR subtractingthecumulativemasslostduringstellarevolution. doesnotalterthefinalresultsappreciablyintheredshiftbinex- The measurement of stellar mass included some assump- ploredhere(seeBolzonellaetal.2009).InthesampleMCOR, tions.WeusedaChabrierinitialmassfunction(Chabrier2003) themasscompletenesslimit,whichdependsonbothredshiftand withaloweranduppercutoffat0.1and100M ,adustattenua- mass-to-lightratio, also takes accountof the colour-magnitude ⊙ tiondescribedbythemodelofCharlot&Fall(2000),andmetal- relation(Pozzettietal.2009;Bolzonellaetal.2009). licities between 0.1 and 2 Z⊙. The star-formation history was The second approach permits us to increase the number of modelledbytwocomponents.Thefirstcomponentwasacontin- objects(from18to69)inouranalysisandexploreawiderred- uousstar-formationmodelwith an exponentiallydeclininglaw shiftbin(z∈[0.48−1.2])thanthatofz∈[0.48−1.0]forsample oftheformSFR(t)∝exp(-t/τ),wheretimescaleτandagetare MST. However, we always compare the results obtained with intherangeτ/[Gyr]=[1,∞]andt/[Gyr]=[0.1,20].Thesecond thisMCORsamplewiththoseobtainedwiththecompleteMST componentwasrepresentedbyrandomburstssuperimposedon sampletovalidatefurtherourconclusions. the first component that had an equal probability of occurring Figure2showsthestellarmassofthezCOSMOSbrightsam- throughoutthelifeofthegalaxy. pleasafunctionofredshiftand,overplotted,thedataofthepost- Ofcoursetheresultsdependontheadoptedassumptionsand starburstgalaxiesinourmass-completesamples(withviolet,tri- modelsused.Forexample,theassumptionofa Chabrierinitial anglesymbols).Wenotethattheconservativelimitimposedby mass functionproducesa consistentdifference in the valuesof Mock catalogues agrees well with the completeness in the ob- thestellarmassatfixedagethatiswithinafactorof1.7,butwith servedmass-to-lightratiofortheglobalpopulation.Thislimitis a smalldispersion,of thosederivedwith the SalpeterIMFpre- similartothemoreconservativelimitobtainedfortheearly-type scription (Pozzettietal. 2007). The inherent uncertainty in the population (defined using the spectrophotometricclassification initial mass function adopted is not a global limitation for re- describedinZuccaetal.2009).Thestatisticalweightsandmass sults based on stellar masses given the constancy of this rigid completenessadoptedarethevaluesusedintheluminosityand shift over a wide range of star formation histories, although to mass function studies by Zuccaetal. (2009), Bolzonellaetal. limitthevariousdegeneraciesalargesetofgridsofstellarpop- (2009), and Pozzettietal. (2009). In the same figure, we also ulationsynthesismodelswereused.Furtherdetailsofthestellar plottheB-bandabsolutemagnitudeasafunctionofredshift.It massmeasurementscanbefoundinBolzonellaetal.(2009)and isimportanttonotethatoursampleMST,whichiscompletein Pozzettietal.(2007,2009). terms of stellar mass such that log(M/M )> 10.8, is brighter ⊙ thantheevolving-luminositylimitatM <−20.1−z(illustrated B withthedashed,blacklineoverplottedintheleftpanelofFig.2). 2.5.Post-starburstrepresentativesample Sinceouroriginalselectionisbasedonaflux-limitedsurvey,we 2.6.Comparisonsamples used subsamples of galaxies complete in terms of stellar mass (adopting two independent approaches) to avoid selection bi- We compiled two comparison catalogues of strongly star- ases. forming and quiescent galaxies. For these galaxies, we also In the first approach, we adopted a conservative limit to adopted consistently the same stellar mass completeness crite- the stellar mass of log(M/M )> 10.8 in the redshift bin z ∈ ⊙ ria,redshiftflags,andconfidencelevelsonthespectralmeasure- [0.48 − 1.0] (from now on, this stellar Mass STringent crite- mentsasusedindefiningtheclassofpost-starburstgalaxies. rion refers to sample MST). This stringent mass limit was de- Thestar-formingandquiescentgalaxyclasseswereselected fined using Mock catalogues from the Millennium simulation usingapurespectralclassification.Wecombinedmeasurements (DeLucia&Blaizot2007)ofthezCOSMOSbrightsample(see of the amplitude of the 4000 Å break, taken as an age esti- Meneuxetal. (2009) for an exhaustiveexplanationon the sub- mator of the underlying stellar population, and the equivalent ject).Weselected18post-starburstgalaxiesforthisconservative width of the [OII]λ3727 line, assumed to be an indicator of mass-completesample,orsampleMST. the star-formation activity. Thus, quiescent galaxies were de- To maximize the quality of the statistical analysis, we also fined to be galaxies with a zero, or negligible, star-formation constructed a second sample, or sample MCOR, including rate inferred using the EW[OII], i.e., EW[OII]> −3Å, and an galaxies more massive than log(M/M )> 10.0 in the redshift ⊙ binz∈[0.48−1.2],andaccountingforunder-representedgalax- oldunderlyingstellarpopulationdescribedbytheirlarge4000Å break (D 4000> 1.5) (e.g., Kauffmannetal. (2003b)). We de- iesathighzbycorrectingwithstatisticalweightsandaV /V n tot max formalism(Schmidt1968).Thetotalvolumesampledbythesur- fined a sample of strong EW[OII] emitters by constructing a veywithintheredshiftbinexploredisdenotedbyV .Thevol- catalogue of objects with EW[OII]< −15Å and small 4000 Å tot ume Vmax is the maximum volume in which each galaxy of a break (Dn4000≤ 1.5). From the 10k sample, we also compiled givenI magnitudeisstillobservable.Weappliedthestatistical the 10k-parentcatalogue of galaxies following the selection in AB weightsadoptedinthestudiesofluminosityandmassfunctions stellar massadoptedforsample MCOR and sample MST, and of the zCOSMOS sample (Zuccaetal. 2009; Bolzonellaetal. allothercriterialistedabove(fromnowonwhenweusetheter- 2009;Pozzettietal.2009).Theseweightsaretheproductofthe minology“allobjects”,werefertothe10k-parentcatalogue). inverseofthetargetsamplingrate(TSR)andtheinverseofthe OurselectedsamplesaredisplayedinFig.3,wheretheselec- spectroscopic sampling rate (SSR). The target sampling rate is tioncriteriausedtoclassifygalaxiesinourcomparativeanalysis the ratio of the objects spectroscopically observed to the total areshown.Thisfigureillustrateshowthethreeclassesofgalax- numberofobjectsintheparentphotometriccatalogue.Thespec- ies (highly star-forming, quiescent, and post-starburst galax- troscopic sampling rate is the ratio of the number of observed ies) populate the plane of the amplitude of 4000 Å break and objects with reliable redshifts to the total number of observed EW[OII].Smallpointsandisophotalcontoursin steps of15%, objects,calculatedinbinsofapparentmagnitude.Theapparent with the faintest isophote level starting at 15%, show the 10k- magnitude dependence takes into account the decline towards parentgalaxieslevel.UsingcriteriaadoptedforsampleMCOR, faintermagnitudes,inourabilitytomeasurearedshift.Amore we obtain a comparison sample of 295 quiescent galaxies and 6 D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield amount of dust. Galaxies with these characteristics should be dusty, starburst galaxies (and misclassified post-starburst ob- jects). This is a hypothesis that should be verified especially when dealing with galaxies understood to have experienced a recent, strong burst of star formation. OB stars responsible for the [OII] emission, as other younger stellar generations, are more strongly obscured by dust than older stellar populations. This age selective dust extinction enables the true strength of the Balmer absorption lines to be measured, but only a frac- tionoftheemissiontobedetectedinstrongerstarburstgalaxies (Poggiantietal. 2001, 2009). To ascertain whether our sample of post-starburst galaxies may be contaminated by dusty star- burst galaxies, we performeddifferent tests. First, we explored the dusty starburst scenario by computing both the extinction valuesandstar-formationrate by SED fitting. Secondly,taking advantageoftheavailablemultiwavelengthobservations,weex- ploredthe X-ray and 1.4GHz radiodata sets (Schinnereretal. 2007;Cappellutietal.2009;Brusaetal.2009). 1) Dust extinctionandSED fitting SFR. We foundthatpost- starburstgalaxieshaveingenerallowvaluesofextinctionasesti- matedbyfittingCB07modelSEDs(medianvalue<τ >=0.6). V ThemedianSFRis0.6M yr−1,andthe25%and75%quartiles ⊙ Fig.3. The amplitude of 4000 Å break is plotted as a func- ofthedistributionrangefrom0.01M yr−1 to4.2M yr−1.The ⊙ ⊙ tion of EW[OII] to show the selection criteria used to classify distributionsofbothτ andSFRsforpost-starburstgalaxiesdif- V theclassesofstar-forming(square,bluesymbols)andquiescent fer at a high confidence level from the same distributions for (circle,orangesymbols)galaxies.Triangle(violet-coded)sym- star-forminggalaxies(<τ >∼2and<SFR>∼21M ).While V ⊙ bols show the post-starburst galaxies. Small points (in black) post-starburst galaxies exhibit a similar cumulative τ distri- V show the 10k-parentcatalogue (isophotal contoursare in steps bution to quiescent galaxies, the latter galaxies have an even of 15% with the faintest isophote level starting at 15% level). lower rate of star-formation activity, as expected when using The4000Åbreakhistogramsofthe10k-parentsample(dashed, the used spectroscopic criteria described in Sect.2 to identify gray line) and of the post-starburst galaxies (solid, violet line) them (< τ >∼ 0.4 and < SFR >∼ 0.07M ). For only one V ⊙ are also shown along the x-axis. The verticallines indicate the post-starburst galaxy do we infer from SED fitting a high rate locationofthecutinDn4000(at1.5).Thehorizontallinesindi- ofstar-formationactivity(∼ 140M⊙yr−1)andahighextinction catethelocationofthetwoextremethresholdsat-3Åand-15Å (τ ∼2).Thisobjectisoneofthetwopost-starburstgalaxiesde- V inEW[OII].ArrowsareupperlimitstoEW[OII]values.Typical tectedatbothX-rayandradio1.4GHzfrequencies,evenifthe errors in the spectral measurements are plotted as errorbars at radiodetectionisonlyatthe3σconfidencelevel(seebelow). thebottomofthepanel. 2) 1.4 GHz radio frequency. We investigated the dusty star- burst hypothesis further by cross-correlating our sample with 174 star-forminggalaxies, while we have 178 quiescentgalax- the 1.4GHz VLA-COSMOS survey (Schinnereretal. 2007; iesand45star-forminggalaxiesinsampleMST.The10k-parent Bondietal.2008).IfitisnotproducedbyAGNemission,thera- galaxiesnumber1508followingthecriteriaofsampleMCOR, diofluxisindicativeofthecurrentstar-formationactivity(over and665forsampleMST. shorttimescalesof108 yr)unbiasedbydust(seeCondon1992, forareview). We found that one post-starburst galaxy has a counterpart 3. Results at 1.4 GHz radio frequency with a detection confidence level We present the main physical properties of the mass-selected of 7.5σ and five others have possible 1.4 GHz emission at post-starburstgalaxiesidentifiedinthezCOSMOSsurveyinat- ≃3σconfidencelevel.Fortheentiresampleof74post-starburst temptingtoobtainacoherentpictureoftheirorigininthecon- galaxies,we thereforehavea detectionrateof1.3±1.3%,if we text of galaxy formation.Our analysis refers to galaxies of the adopt the threshold of 4σ for assuming a radio detection to statisticallycorrected,mass-completesampleatz∈[0.48−1.2] be true (as adopted in Schinnereretal. (2007) and Bondietal. (sample MCOR) and the conservative mass-limited sample at (2008)).Thisdetectionrate increasesto 8.1±2.7%,if we adopt z∈[0.48−1.0](sampleMST). thelessconservativethresholdof3σforradiodetection. Toinvestigatethemeanradiopropertiesofthepost-starburst galaxies,weperformedastackinganalysisofthe1.4GHzradio 3.1.Robustnessofoursample:dustystarbursthypothesis? images. From the stacking analysis, we excluded the six post- Galaxies with spectral characteristics similar to those used to starburstgalaxieswithapossibleradiocounterpartplusthreead- define a post-starburstgalaxy are often interpreted to be a sys- ditionalsourcesthatexhibit,within2.5′′oftheiropticalposition, tem in which star formation has suddenly and almost com- eitherradio emissionmoresignificantthan3σthatis notasso- pletelystoppedwithinthepastfewGyrs.Thiscessationofstar- ciatedwiththepost-starburstgalaxy,oranegativeradiopeakof formationisassumedtoexplainthelackofemissionlinesinthe highersignificancethan3σ.Fortheremaining65post-starburst galaxyspectrum. galaxies,imagesof40×40pixels(correspondingto14′′×14′′) An alternative explanation, however, is that these galaxies wereextractedaroundtheopticalposition.Finally,allthe65im- do have emission, but that this is extinguished by a significant ageswerecombinedto obtaina medianimage with a 1σradio D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield 7 noiseof≈2µJy.Amarginaldetectionof≈ 3σconfidencelevel TheopticalspectraofthesethreesourceswiththehighestX-ray was found at a distance of 1.4′′ from the centre of the stacked countsin thehardband,mightindicatethattheyareAGNhost image with a radio flux of ≈7 µJy. To access the reliability of galaxies. this detection,we repeated the stacking analysisfor 5 different samples, each of them consisting of 75 random positions. The OfthesefiveX-raysources,twohaveaclearassociationwith criterionusedtoexcludefromthestacking,post-starburstgalax- sub-mJysourcesintheVLAcatalog(Bondietal.2008),oneof ies with a radio detection(ora negativeradio peak,see above) which has the highest SFR of all sources and dust extinction was applied similarly to the simulated samples. All 5 stacked inferredfromSEDfitting. images from the simulated sample had a 1σ radio noise in the range 1.4-1.8µJy. No radio detection was identified for any of To place constraints on the properties of the X-ray unde- thesimulatedstackedimages. tected post-starburst galaxies, X-ray stacking analysis was ap- Therefore, assuming the marginal detection in the stacked pliedtotheremaining49sources,extractingsourcecountsfrom imageto be realandassuming its fluxof 7 µJy to be the mean circles of 5′′ radius and background counts from a region at radiofluxofradio-quietpost-starburstgalaxies,wecanestimate least10′′ awayfromthesourceposition.X-raysourcespresent their mean star-formation rate, or more realistically its upper in the Chandra source catalog (Elvis et al. 2009) were care- limit since some objects may be AGNs. We obtained an upper fully masked. To limit the problems with the PSF at large off- limit of 8.2M yr−1 by adopting the calibration by Belletal. axisangles,weextractedsourcecountsusingonlyobservations ⊙ (2003). The radioluminosityL was calculatedby assum- wherethesourcepositionwaslocatedwithin8′ofthefieldcen- 1.4GHz ingamedianredshiftofz =0.7014. tre(seeMiyajietal.(2008)andKimetal.(2009)foradescrip- median We then explored the physical properties of the six (> tion of the tool adopted here). We did not find any significant 3σ)radio-detectedpost-starburstgalaxies.Radio-detectedpost- signal from the stacked post-starburst galaxies using Chandra; starburst galaxies are slightly more massive and brighter than the 3σ flux upper limit to the stacked 0.5–8 keV counts is non-radio-detected post-starburst galaxies (logM/M = ≈ 2×10−16 cgs,correspondingtoarest-frame0.5–8keVlumi- ⊙1.4GHz 11.22 ± 0.09 versus log M/M = 11.08 ± 0.05; nosityof≈4.9×1041ergs−1attheaverageredshift(hzi≈0.73) ⊙no−1.4GHz M =−22.64±0.37versusM =−22.44±0.14). ofthestackedsources.Thisupperlimittothe stackedsignalis B,1.4GHz B,no−1.4GHz Post-starburstgalaxieswithacounterpartat1.4GHzalsoshow consistentwithemissionfromnormalandstarburstgalaxies. slightlybluercolours((U-B) =1.10±0.1)thannon-radio- 1.4GHz detectedpost-starburstgalaxies((U-B) = 1.18±0.03). Dewanganetal.(2000)reportedadetectionofanactivenu- no−1.4GHz Although these differences are not statistically significant per cleusinonepost-starburstgalaxyselectedduringasearchforul- se, their properties and low detection rate suggest that radio- trasoftsourcesinaROSATcampaign,butnoradioemissionwas detectedpost-starburstgalaxiesmightrepresenttheinitialphase detected within a radiusof 15′′ in the NRAO/VLA and FIRST in the evolution of post-starburst galaxies, close to the peak of sky surveys. In previous studies, similar counter-checks led to thestarburst(thetimescaleofradioemissionisanorderofmag- a detection rate of the order of 1-15%. Miller&Owen (2002) nitudeshorterthanthatofthek+aphase). using the galaxies selected by Zabludoffetal. (1996), detected 2 in 15 post-starburstsystemsin radiocontinuum(correspond- 3) X-ray counterparts. To characterize the multiwavelength ing to a star-formation rate of about 2-6 M yr−1), and 1 in propertiesof post-starburstgalaxies, we also searchedfor their ⊙ 56objectsweredetectedbyBlakeetal. (2004) (at7 M yr−1). possible X-ray counterparts using the available data provided ⊙ by XMM-Newton for the entire 2 deg2 of the COSMOS field UsingChandraobservationsoftheExtendedGrowthStripfrom DEEP2,Georgakakisetal.(2008)foundanassociationbetween (Cappelluti et al. 2009; Brusa et al. 2009) and Chandra over the inner 0.9 deg2 (Elvisetal. 2009) at considerably deeper post-starburstgalaxiesat z ∼ 0.8 andAGN, supportinga mod- eratelevelsofAGNobscuration.Althoughtheseresultsaredif- X-ray flux levels. While only one post-starburst galaxy is de- tectedbyXMM-Newtonforamatchingradiusof6′′,fiveX-ray ficulttocomparebecauseofdifferentselectioncriteriaanddif- ferentmassandluminosityranges,theseresultshavebeenused sources(four“new”plustheXMM-Newtonone)aredetectedby Chandra(within4′′ because of the sharperPSF) at a detection toclaimthatatleastsome“so-calledpost-starburst”galaxiesare still actively forming stars, implying that a selection based of likelihood>10.8inatleastoneX-rayband(seeElvisetal.2009 the blue-part of the optical spectra is imprecise. Some authors and Puccettiet al. 2009for furtherdetails) amongthe 54 post- haveshownthatopticalspectratypicalofapost-starburstgalaxy starburstgalaxieslocatedinthefieldofChandramosaic.Visual inthebluepartoftheirspectrum,havesubstantialHαemission inspection of all of these sources supports their likely physi- (e.g.,Liu&Kennicutt1995).Wefullyagreewiththestatement calassociationwiththe post-starburstgalaxiesunderinvestiga- by e.g., Miller&Owen (2002) that the minimum requirement tion.Allofthefivepost-starburstgalaxiesdetectedbyChandra fordefiningasystemintheirk+a,orpost-starburstphaseisthe are in regions of moderate-to-high exposure time (>100 ks), coverageofbothblue-andred-partsoftheopticalspectrum,or wheremorethanoneChandrapointingisavailable(Elvisetal. complementingtheopticaldatawithotherwavelengthdatasets. 2009). However, the limited number of counts (≈ 9–40) in the 0.5–7keVbandpreventsusfromperforminganadequateX-ray spectral analysis. X-ray counts were converted into fluxes (us- ing a power-lawmodelwith photonindexΓ = 1.4, as adopted in the Chandra-COSMOScatalog; Elvis et al. 2009), and then into0.5–10keVrest-frameluminosities(giventhesourcespec- Basedonourcomplementaryradio-continuumanddeepX- troscopicredshiftsof0.67–1.10)of1.1–8.6×1042ergs−1. raydata,wefoundthatanegligiblefractionofourpost-starburst At leastforthethreesourceswiththe highestX-raycounts galaxies,selectedbyusingbluediagnosticsoftheopticalspec- inthehardband(above2keV),thisluminosityrangesuggestsa tra, have possible residual star-formation activity, and a small significantcontributionfromAGNemission.Theassumptionof percentageshowsacontributionfromAGNemission.Thus,the aphotonindexofΓ= 1.9,whichistypicalofunobscuredAGN mainconclusionofourpresentstudyisthatwehaveselecteda emission,woulddecreaseX-rayfluxesandluminositiesby25%. robustsampleofpost-starburstgalaxies. 8 D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield Fig.4.Stellarmass,B-bandabsolutemagnitude,andstellarmass-to-lightratiohistogramsareshownforeachclassofgalaxies.The cumulativedistributionsofeachsamplearealsoplotted.Medianvaluesofpost-starburstgalaxiesand10k-parentsampleareplotted as solid and dashed lines, respectively. Blue and red arrows are median values of star-forming and quiescent galaxies. The bars assigned to each median value correspondto 1σ error computedwith resamplingtechnique.Galaxies classified as post-starburst (violet),quiescent(red),star-forming(blue),andthe10k-parent(black)samplefollowthedefinitiondescribedinSect.2.5. 3.2.Basicproperties:luminosities,stellarmasses,and acombinationofahighstellarmassandrelativelylowmass-to- colours light ratio. Among the most massive galaxies (criterion MST), there is a trend for post-starburst galaxies to be the most mas- The B-band absolute magnitude distribution plotted in Fig.4 sive objects, more massive than quiescentgalaxies (at 1σ) and shows that post-starburst galaxies of the conservative sam- star-forminggalaxies(at∼ 2σ).Sinceourpost-starburstgalax- ple MST populate the brightest tail of our 10k-parent galaxy iesare alreadyso massive,we mayconcludethattheirprogen- population. The post-starburst galaxies more massive than itorscannotbe commonstar-forminggalaxiesthathaveassem- log(M/M )> 10.8 are brighter by ∼ 0.7 − 0.8 mag than the ⊙ bled their stellar mass by means of a continuous, smooth star- galaxiesinbothourcomparisonclasses.Asimilaranalysisofthe formation history and that have stopped, for whatever reason, sampleMCORgalaxiesconfirmsthatthepost-starburstgalaxy forming stars. Indeed to justify the observed stellar mass as- sampleissystematicallybrighterby∼0.4−0.5magthantheme- sembly of a post-starburstgalaxywith a typicalstellar mass of dianbrightnessescomparedtomedianvaluesoftheothergalax- log(M/M )= 10.58 (see Fig.4), a specific-SFR (SFR/M) of ⊙ ies. Both these results are statistically significant at more than ∼1.2×10−9(orSFR∼20M yr−1)isrequiredforagalaxywith ⊙ the 3σlevel.One of the reasonsfor thisB-bandbrightnessex- astellarmassoflog(M/M )=10.23(where10.23isthemedian ⊙ cessisprobablyarecentburstofstarformationinpost-starburst stellarmassfoundforourstar-forminggalaxies).Thisspecific- galaxycandidates,ascommonlyconcludedinpaststudies. SFRvalueistypicalofanobjectwithahighstar-formationef- Post-starburst galaxiesalso populatethe massive tail of the ficiency.Furthermore,wecanalsoexcludethedrymerginghy- 10k-parentgalaxies(see Fig.4). Their mass distribution differs pothesis, i.e., the mergingof galaxyprogenitorsin the absence fromthatofthe10k-parentgalaxiesat∼ 91%confidencelevel ofgas,asthestrengthoftheHδabsorptionlinesimpliesarecent usingtheKolmogorov-Smirnovtest(Pressetal.1992),andour burstofnewstars,i.e.,thatfreshgasshouldhavebeenrecently post-starburst candidates are as massive as quiescent galaxies. available. Therefore, the possible progenitors might be the al- Our different samples selected using the MCOR criteria show readymassive,lessnumerousstar-forminggalaxiesateachcos- thetypicalincreasingtrendinstellarmassandmass-to-lightra- mictime,or/andthemostactivelystar-formingsourcesthatina tioexpectedonthebasisoftheirgalaxyclassificationfromstar- shortperiodofcosmictimecanassemblysufficientstellarmass: forminggalaxypopulationtoquiescentgalaxies.Giventhesim- in other words, the k +a’s progenitors might be galaxies with ilar magnitudesandstellar masses spannedbythe star-forming thehighestvaluesofspecificstar formationratesatanyepoch, andquiescentgalaxiesselectedusingtheMSTcriteria,thistrend assumingthatthe mass assemblyis a processsustained by star isnotvisibleinthecorrespondingpanelsofFig.4.Post-starburst galaxiesappeartohaveabrighterB-bandluminositybecauseof D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield 9 formationactivityattheobservedrate,i.e.,nomergereventsare invoked. Quiescent and star-forming galaxies populate the red se- quence and the blue cloud, respectively, as shown in Fig.5 for sampleMCOR.Thepeakofthedistributionforquiescentgalax- iesisat(U−B)∼1.3,andstar-forminggalaxiesaredistributed around a locus of (U − B) ∼ 0.8 in sample MCOR (colours areslightlyredderforstar-forminggalaxiesinsampleMSTby ∼ 0.2 mag, due to the higher stellar mass selection). The un- derpopulatedregionin-betweenthetwo distributions,i.e.,a re- gioncalledthegreenvalley,isthelocuswithinwhichourpost- starburstgalaxiesarelocated.Inthecolour-magnitudediagram, they are on average (U − B) ∼ 0.25 redder than star-forming galaxies,and∼ 0.1bluerthanquiescentgalaxies.Althoughthe median distribution of colours appears to imply that the green valleyispopulatedbypost-starburstgalaxies,theyspana wide rangeofcolours.ThispropertyisconfirmedinFig.3,wherethe 4000 Å break histograms of the 10k-parent sample and post- starburstgalaxiesareshownalongthex-axis(withasolid,vio- letlineanddashed,grayline,respectively).The4000Åbreakof post-starburstgalaxiesrangesbetweenquitesmallvalues(typi- calofstar-forminggalaxies)andquitehighones(characteristics of quiescent, passive galaxies), or 1.2 < D 4000 < 2, with a Fig.5. Distributionofthe(U-B) colourversusthestellar mass n peakatD 4000∼1.5. for post-starburst galaxies (triangle, violet colour-coded sym- n Based on this result, selecting post-starburst-candidate bols) and comparison samples (quiescent galaxies, circle, red- galaxies such as those that exhibit stronger Balmer absorption coded symbols, and star-forming galaxies, square, blue-coded symbols). Isophotal contours are in steps of 10% with the linesthanexpectedfortheir4000Åbreak,e.g.,inawaysimilar faintestisophotelevelstartingatthe10%level.Theverticalsolid tothecriterionadoptedbyWildetal.(2009),shouldunderesti- (dashed)lineindicatesourmass-completenessatlog(M/M )> mate their truefraction.Ourresultmightjustify the low detec- ⊙ 10.0(>10.8). tionofthisclassofgalaxiesintheVVDSfield.Wildetal.find only 5 post-starburst galaxies in the VVDS field, scaling from their quoted 18 Hδ strong galaxieswith nebular emission lines late-type divisions is independent of their Hδ strength shown to 5 properly-definedpost-starburstgalaxies(i.e., those objects in panel A of Fig.6. There is, instead, an increase with Hδ amongthe18detectedoneswithoutnebularemissionlines). in the fraction of post-starburst galaxies in irregular systems. Ourresultindicatesthatpost-starburstgalaxies,whichpop- In panel C, we show the relative fractions of the three dif- ulate the green valley for a very short time, provide a suitable ferentmorphologicaltypes(bulge-dominated,disky-dominated, galaxypopulationtooccupythegapbetweenblueandredgalax- and irregulars) that are post-starburst galaxies compared to ies in the colour-magnitude relation observed at any redshift the same fractions for the samples of quiescent and star- fromthelocalUniverseuptoz∼1.−1.5.Althoughk+agalax- forminggalaxies.Asexpected,star-forminggalaxiesaremostly iesmayrepresentsomeofthegalaxiesinthegreenvalley,wedo disky-dominatedwith ∼ 20−30% having irregular morpholo- nothoweverproposethattheyconstitutetheentirepopulationof gies, while the majority of the quiescent galaxies are bulge- green-valleygalaxies. dominated. Morphologically, post-starburst galaxies are a het- erogeneouspopulation,althoughwenotethatthemostmassive 3.3.Morphologiesandstructuralparameters post-starburstgalaxies(sampleMST)arepredominantlybulge- dominatedsystems(seelowerpanelsofFig.6)andhaveasim- We investigated the morphological properties of post-starburst ilardistributionamongmorphologicaltypesasquiescentgalax- galaxiesusingthehighspatialresolutionHST/ACSF814Wim- ies.ItisinterestingtonotethatinthesampleMST,thereisno ages(Koekemoeretal.2007).Galaxymorphologiescontainthe post-starburstgalaxywithanirregularmorphology,butwealso imprintsofstellarpopulations,gascontent,anddynamicalstruc- notethelownumberstatisticsinthecurrentsampleMST. tures.Thus,ifaneventsuchasmerginghasoccurredinthere- In some studies, post-starburst galaxies are found to be centhistoryofagalaxy,itssignatureshouldbeobserved,given mostlybulge-dominatedand/orhavea morphologythatis con- thatthetimescaleofgravitationaldisturbancesinducedbymerg- sistentwithbeingearly-type(Goto&SDSScollaboration2003; ingissimilartothatofthepost-starburstphase. Quinteroetal. 2004; Tranetal. 2004; Blakeetal. 2004; Goto We used the ZEST morphological classification of 2005; Baloghetal. 2005; Poggiantietal. 2009), and in some Scarlataetal. (2007a) based on five non-parametric diag- othershaveahighincidenceofdisksandspiralarms(e.g.,Franx nosticstodefineearly,late,andirregularmorphologicalclasses 1993;Couchetal. 1994;Dressleretal. 1994;Caldwell&Rose of all studied galaxies in this work. We also explored in more 1997;Dressleretal.1999).Forexample,Quinteroetal.(2004) detail properties such as the asymmetry index, the degree of visually inspected a sample of 160 galaxies from the SDSS at irregularities,andtheconcentrationofthelight. 0.05<z<0.20,andclassified90%ofthemasbulge-dominated The upper panel of Fig.6 illustrates that a similar frac- systems. Yangetal. (2004) used the term “diverse morpholo- tion of post-starburstgalaxiesin the sample MCOR are bulge- gies”todescribefivehigh-resolutionHSTimagesofk+agalax- dominated (E) and disky-dominated (L) (see panel B). How iesselectedfromthesampleofZabludoffetal.(1996),although the galaxies are distributed between the so-called early- and we note that their selection did favour the bluest systems, i.e., 10 D.Verganietal.:Post-starburstgalaxiesinCOSMOSfield the ones with smallest D 4000 that have experienced most re- n centlyasignificantstarburstepisode.Theirdetailedstudyusing bulge-to-disktwo-dimensionalfitting decompositionof four of them identified one barredS0, two disky systems, and one un- classifiablegalaxy.Theirinferredcentrallightdistributionswere foundtohavea profilesimilarto,butmoreluminous,thanthat ofatypicalearly-typepower-law.Athigherredshifts,Tranetal. (2004)foundmorebulge-dominatedsystemsintheirk+agalaxy sample than in field galaxies. The sample was selected from a program designed to study galaxy evolution in clusters, and k+a galaxieswere selected using a conservativeconstrainton boththeHδandHγBalmerindicesandaccordingtotheabsence of[OII]emissionline.Astudybasedonvisualclassificationof ACS/HST imagesby Poggiantietal. (2009) identified a higher fractionof S0 and Sa types amongfield post-starburstgalaxies than Dressleretal. (1999) who instead found a higher fraction oflatertypesamongpost-starburstgalaxies. Basedonthemorphologicalresultsobtainedforthepresent post-starburst galaxy sample, we agree with the widely estab- lished result that post-starburst galaxies are morphologically a heterogeneous population. We also found an increasing frac- tion of early-type morphologies at increasing stellar mass in agreementwiththewellknownmass-morphologyrelation(e.g., Tascaetal. 2009). We mightfind in the future with larger data sets spanning a wider range of galaxy properties, that the flux limits and other selection effects might play a critical role in themorphologicalclassification,e.g.,favouringanearly-typeas- signmentforbrighter/massivesystemsasfoundinthiswork. Apart from being heterogeneousin their morphologies,the second common feature used to describe post-starburst galax- ies by most authors is the large fraction of asymmetries and clumpinesspropertiesassociatedwiththem(e.g.,finestructure, and tidal tails). We quantify the degree of asymmetry in the light distribution using the asymmetry coefficient computed as the difference between the image rotated through 180 degrees and the original frame following the usual prescription (e.g., Conseliceetal. 2000) (for details see Scarlataetal. (2007b)). Figure7 shows that for post-starburst galaxies in the sample MCOR, the median and range of the asymmetry index are in- termediatebetweenvaluesforquiescentgalaxies(andthe 10k- parentsample)andstar-forminggalaxies.Comparingtheasym- metrydistributionsofpost-starburstgalaxieswiththoseofstar- forming and quiescent galaxies, they both differ at higher than the99.9%KS-testconfidencelevel.Similarresultsareobtained forthelightconcentration,whichisdefinedtobetheratioofthe radiusincluding80%to thatincluding20%of the galaxylight andmeasuresthecentraldensityofthegalaxylightdistribution. In Fig.7, we show that the light concentration indices of post- starburst galaxies are typically in-between the values for qui- escent and star-forminggalaxies. These differencesin both the Fig.6.Morphologicalclassificationofpost-starburstgalaxiesfor morphological classification and the parameters of asymmetry sampleMCOR(top)andsampleMST(bottom).InpanelAwe andlightconcentrationofpost-starburstgalaxiesandthatofboth show the fraction of the three different morphologicaltypes in star-formingand quiescentgalaxies, suggest that post-starburst post-starburst galaxies (bulge-dominated (E), disky-dominated galaxiesrepresentanintermediatestageofgalaxyevolution. (L)andirregulars(I))asafunctionofthestrengthofHδ.Similar UsingtheZESTscheme,thedistortionofthelightdistribu- plot is shown in panel B for post-starburst galaxies (indepen- tion is quantified by the clumpiness index. This index is avail- dently by the Hδ strength). Panel C shows the fraction of the able forgalaxiesclassified as early-and late-types,i.e., notfor three different morphological types in quiescent galaxies (or- irregular galaxies, which by definition are very distorted and ange)comparedtothesampleofpost-starburst(violet)andstar- for which the index might be associated with only the max- forming galaxies (blue), respectively. The error bars represent imum degree of clumpiness. The flags for the clumpiness in- thePoissonianerrors. dexrangefromzero,indicativeofasmoothlightdistribution,to three,correspondingtoaclumpylightdistribution(seeTable2, Scarlataetal.(2007a)).Wefoundthatthemajorityofquiescent galaxies have a smooth light distribution (86% have flags of 0 or1),whereasstar-forminggalaxiestendtohaveaclassification

See more

The list of books you might like

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