Draftversion February5,2008 PreprinttypesetusingLATEXstyleemulateapjv.05/04/06 THE MILLENNIUM GALAXY CATALOGUE: THE LUMINOSITY FUNCTIONS OF BULGES AND DISCS AND THEIR IMPLIED STELLAR MASS DENSITIES Simon P. Driver and Paul D. Allen SUPA1,SchoolofPhysicsandAstronomy,UniversityofStAndrews,NorthHaugh,StAndrews,Fife,KY169SS,UK Jochen Liske EuropeanSouthernObservatory,Karl-Schwarzschild-Str.2,85748GarchingbeiMu¨nchen,Germany Alister W. Graham 7 CentreforAstrophysicsandSupercomputing, SwinburneUniversityofTechnology, Victoria3122, Australia 0 Draft versionFebruary 5, 2008 0 2 ABSTRACT n We derive the luminosity functions of elliptical galaxies, galaxy bulges, galaxy pseudo-bulges and a galaxydiscsfromourstructuralcatalogueof10,095galaxies. Inadditionwecomputetheirassociated J luminosity and stellar mass densities. We show that spheroidal systems (elliptical galaxies and the 5 bulgesofdiscgalaxies)exhibitastrongcolorbimodalityindicatingtwodistincttypesofspheroidwhich 2 are separated by a core color of (u−r) ∼ 2 mag. We argue that the similarity of the red elliptical and the red bulge luminosity functions supports our previous arguments that they share a common 1 originand surprisinglyfind that the same followsfor the blue ellipticals and blue bulges,the latter of v whichwereferto aspseudo-bulges. Intermsofthe stellarmassbudgetwefind that58±6percentis 8 currently in the form of discs, 39±6 per cent in the form of redspheroids (13±4 per cent ellipticals, 2 26±4percentbulges)andtheremainderisintheformofbluespheroidalsystems(∼1.5percentblue 7 ellipticals and∼1.5 per centpseudo-bulges). Interms ofgalaxyformationwe arguethatour dataon 1 0 galaxycomponents stronglysupports the notionof a two-stageformationprocess (spheroidfirst, disc 7 later) but with the additional complexity of secular evolution occurring in quiescent discs giving rise 0 to two distinct bulge types: genuine ’classical’bulges and pseudo-bulges. We therefore advocate that / therearethreesignificantstructuresunderpinninggalaxyevolution: classicalspheroids(old); pseudo- h bulges(young)anddiscs(intermediate). Theluminousnearbygalaxypopulationisamixtureofthese p threestructuraltypes. Thenatureofthe blue ellipticalgalaxiesremainsunclearbut onepossibilityis - o that these constitute recently collapsed structures supporting the notion of mass-dependent spheroid r formation with redshift. t s Subject headings: galaxies: spiral- galaxies: structure - galaxies: photometry - galaxies: fundamental a parameters - ISM: dust, extinction : v i X 1. INTRODUCTION more importantly, when the population was segregated r Inarecentpaper(Driver et al. 2006)wedemonstrated byHubbletypewefoundthatwhiletheearly-typegalax- a ies (E/S0s, i.e., bulge dominated) lay almost exclusively that galaxy bimodality is not just evident in color in the red-concentrated peak, and the late-type spirals (see Strateva et al. 2001; Baldry et al. 2004) but also in (Sd/Irr, i.e., disc dominated) in the blue-diffuse peak, the joint color-structure plane (see also Ball et al. 2006; the mid-type spirals (Sabc, i.e., bulge plus disc systems) Park et al. 2006; Choi et al. 2006; Conselice 2006). In straddledbothpeakswithnoobvioussignofbimodality. that work we used SDSS photometry and single S´ersic We inferred from this that galaxy bimodality arises be- (1963; Graham & Driver 2005) profile fits to investigate cause of the two component nature of galaxies and that the distribution of 10,095 relatively nearby luminous (M <−17mag)galaxiesinthecolor-S´ersicindexplane. spheroidal structures (i.e., ellipticals and bulges) will lie B exclusivelyinthered-compactpeakanddiscsintheblue- The red peak is comprised of highly concentrated, high- diffuse peak. As classical bulges lie within thin rotating S´ersic index systems while the blue peak contains more discsystemsthisarguesforearlyspheroidformation(via diffuse, low-S´ersic index systems. This is important as rapid merging or collapse) followed by a more quiescent any movement from the blue peak to the red peak will phase in which the extended disc can form. To explore requiremodifyingtheorbits,angularmomentumanden- thisfurtherwehaveperformedbulge-discdecomposition ergy of the entire stellar population. A simple inert pro- of all 10,095 galaxies in the Millennium Galaxy Cata- cess (e.g., exhaustion of the gas supply, stripping, etc.) logue (see Allen et al. 2006) by fitting two-component could not achieve this, although a violent major merger S´ersic-bulge plus exponential-disc models using GIM2D event could (e.g., Barnes & Hernquist 1992). Perhaps (Simard et al. 2002). In this Letter we report the lumi- Electronicaddress: [email protected] nosity functions derived for various component samples Electronicaddress: [email protected] (e.g.ellipticals,bulges,discs)andtabulatetheassociated El1ecStrcoonttiicshadUdnreivsse:[email protected](SuU.aPuA) luminosityandstellarmassdensitiesforeachcomponent 2 Driver et al. class. Gyr spectrum). This spectrum can be represented by ThroughoutthispaperweassumeaΛCDMcosmology a fourth order polynomial valid over the redshift range with Ω = 0.3, Ω = 0.7, and adopt h = H /(100 km 0<z <0.18 only: m Λ 0 s−1 Mpc−1) for ease of comparison with other results. k(z)=3.86z+12.13z2−50.14z3. (1) 2. MGCCOMPONENTLUMINOSITYFUNCTIONS Wenotethatifwefollowasimilarprocedureforthediscs The Millennium Galaxy Catalogue (MGC) is a deep andbluebulgestheimpliedcharacteristicturnoverlumi- (µ =26 mag arcsec−2), wide area (37.5 deg2), B-band nosities (shown in Table 1) are systematically reduced lim imagingandredshiftsurveycoveringa0.6degwidestrip by ∼0.1 mag and the implied stellar masses reduced by along the equatorialsky from10hto 14h50′. The MGC ∼14 per cent. contains10,095galaxiesdowntoB =20mag,ofwhich To model evolution we assume pure luminosity evolu- mgc 9,696haveredshiftinformation. Fulldetailsoftheimag- tion of the following form: ing survey can be found in Liske et al. (2003), with the L =L(1+z)−β, (2) spectroscopic follow-up described by Driver et al. (2005; z=0 hereafter D05). In Allen et al. (2006) the bulge-disc de- where β is set to 0.75 for the global luminosity function compositions were reported for all 10,095B <20 mag (see Driver et al. 2005), 1 for blue components (discs, mgc galaxies,and a final structural catalogue produced, con- blue bulges,blue ellipticals), and0.5 forredcomponents sistingofbulge-only(S´ersicprofiles),disc-only(S´ersicor (red bulges and red ellipticals). These values are based exponential profiles) and bulge plus disc systems (S´ersic on the recent results reported in Zucca et al. (2006) plus exponential profiles). for red and blue systems (their type 1 and 4 respec- Luminosity functions are computed using the stan- tively). We do not model number evolution as the red- dard step-wise maximum likelihood (SWML) estima- shiftrangeissmallandourmergerrateestimates,based tor originally described by Efstathiou, Ellis & Peter- ondynamicallyclosepairswithinthe MGC,arelow(see son (1988), with samples divided into bins of absolute De Propris et al. 2005 and De Propris et al. 2007). magnitude. The MGC spectroscopic sample has a nom- inal Kron magnitude limit of B = 20 mag. How- 2.2. Luminosity functions mgc ever, here we will use the GIM2D total magnitudes Fig.1showstheluminositydistributionsandSchechter (derived by integrating the light profiles to infinity) so function fits for our full galaxy sample (upper left), el- that there is no longer a single limit that applies to lipticals only (upper right; i.e. objects with B/T = 1 the sample as a whole. To accommodate for this, each after logical filtering, see Fig. 13 of Allen et al. 2006), galaxy now has a unique magnitude limit, defined by: discs (lower left) and bulges (lower right). Note that Blim = 20+Bmgc(S´ersic)−Bmgc(Kron). Following D05, the Schechterfunction inthe upper left differs fromthat we restrict our sample to galaxies in the redshift range shown in D05 because the magnitudes are now based 0.013 < z < 0.18 and within carefully defined size and onS´ersic profiles integratedto infinity rather than Kron surface brightness boundaries (see D05 and Liske et al. magnitudes. This difference is significant, resulting in a 2006 for full details). brighter M∗ value by about 0.1 mag but a comparable faint-end slope, α. Analysis of independent repeat ob- 2.1. k+e corrections servations of ∼700 galaxies suggests that our decompo- In D05 individual k-corrections were derived for each sitions are valid to good accuracy (∆M = ±0.1 mag bulge galaxy by comparing the total galaxy broad-band colors and ∆M =±0.15 mag, see Allen et al. 2006)for com- disc (uBgriz)tothe 27spectraltemplatesgiveninPoggianti ponentss with luminosities with M <−17 mag. Below B (1998)andidentifying the best fitting spectrum. Having thislimitourdecompositionsbecomeincreasinglylessre- now separated the MGC galaxies into bulges and discs liable andthese dataareshownwithopen symbols. The (see Allen et al. 2006) the globalk-correctionis notnec- moststrikingresultfromFig.1istherapidlyrisingfaint- essarily valid. However, approximately 50 per cent of end slope for the elliptical population. This was noted our sample are best fit by one component profiles (i.e., previously in Driver et al. (2006) and was shown to be bulge-only or disc-only galaxies) and for these systems duetocontaminationoftheclassicalellipticalsampleby we adopt the k-correctionsas previously derivedin D05. lowluminosity blue spheroids(see alsoEllis et al. 2005). For the remaining, two-component systems we consider In Fig. 2 we show the color-structureplane defined by our componentcolorstoo coarseto be used to derivero- SDSS core (u− r) PSF color versus component S´ersic bust individual k-corrections (as our decompositions are index for the ellipticals (upper left) and galaxy bulges done in a single filter only). For the case of blue bulges (upper right). The bimodality of the ellipticals is strik- and blue discs the global k-correction is likely to be ap- ing,withablueandredpopulationbeingapparent. The propriate for both (assumming that the blue bulge has blue sample defines what we label blue ellipticals which formed from the disc). In the case of discs surrounding were identified in Ellis et al. (2005) and quantified in classical red bulges we note that for low-B/T systems Driver et al. (2006). A cut at (u−r) = 2 mag provides the k-correction is likely to be appropriate for the discs a clear division. The lower panels of Fig. 2 show the lu- but not the bulges. In the case of high-B/T systems minositydistributionsandSchechterfunctionfitsforthe we note that Peletier & Bacells (1996) report that such blue and red samples. We see that red ellipticals and discs are typically redder. We therefore consider it ap- blue ellipticals follow markedly different trends and it is propriate to continue to adopt the global k-correction indeed the blue ellipticals which are responsible for the for both our discs and blue bulges. For the red bulges, apparentupturn in the total elliptical galaxy luminosity however,weadoptthespectraltemplatemostfrequently functionatveryfaintabsolutemagnitudes. Wenotethat adopted by our single componentred ellipticals (a Sa 15 whenthebulgesaredividedinthesamemanner,theblue The MGC: Luminosity functions of bulges and discs 3 bulge luminosity function is very similar to that of the centofthestellarmassliesintheredpeakdominatedby ellipticals, possibly indicating some common origin. early-typegalaxies. Aswehavenowseparatedtheearly- Itistemptingtoassociatethebluebulgeswithpseudo- type galaxies into bulges and discs one expects that the bulges (see Kormendy& Kennicutt 2004),whichare be- spheroid (elliptical+bulge) stellar mass density should lieved to arise from inner disc instabilities giving rise to be lower than the early-type stellar mass density, which a ’swelling’ of the disc in the central region. As many of of course includes the associated disc components of the our blue bulges have M > −17 mag, where our bulge- lenticular galaxies. Examining the color-structure plane B disc decompositions become unreliable, we cannot un- (Fig. 2 upper panels) for ellipticals and bulges we see ambiguously confirm this population as pseudo-bulges that the red populations (of each type) lie in the same but for the moment adopt this nomenclature for ease location. This is indicative of a shared origin for red of discussion. The blue ellipticals remain somewhat in- ellipticals and classical red bulges. When combined the triguing and appear to define a new class of object as red spheroids account for 39±6 per cent of the stellar previously noted by Ellis et al. (2005) and Driver et al. mass density. Hence the bulk of the stellar mass (96 (2006; see also Ilbert et al. 2006 who identify them as a per cent) exists in the two classical structures originally rapidly fading population). We are currently exploring defined by de Vaucouleurs (1959). these systems further (Ellis et al. 2007) and for the mo- The remaining 3 per cent lie in the form of blue ellip- mentsimplyflagthemasinteresting. Fromtheirdistinct ticals (∼1.5 per cent) and blue bulges (∼ 1.5 per cent). luminosityfunctionitiscleartheyarepredominantlylow Theselattertwopopulationscanthereforebeconsidered luminosity systems and could potentially represent the minor from a cosmological perspective. Furthermore, as local counterparts to the luminous blue compact galax- the blue bulges are likely to represent either difficulties ies studied by Guzman et al. (1997) and Phillips et al. in the decomposition (e.g., bars) or pseudo-bulges (disc (1997). swelling), their stellar mass could arguably be added, Fig. 3 shows the final component luminosity functions in either case, to that of the galaxy discs (i.e., ∼ 61 with the red ellipticals and red bulges combined into a per cent). The nature of the blue ellipticals (previously singleredspheroidgroupandtheblueellipticalsandblue dubbed blue spheroids) remains uncertain but they ap- bulgesgroupedtogetherintoasinglebluespheroidclass. pear to constitute a very small fraction of the stellar The justification for this is the similarity in the shapes mass budget, although we must note the near divergent and ranges of the luminosity distributions from Fig. 2. faint-end slopes. The primary conclusion then is that the stellar mass is mainly divided between two distinct 2.3. Luminosity densities and stellar-mass densities structures: blue 2D discs and red 3D spheroids. FinallywenotethattheresultspresentedinthisLetter Table1showsthe Schechterfunctionvalues forthe lu- are based on B-band data and therefore susceptible to minosityfunctionfitsshowninFigs.1,2&3,alongwith dust attenuation (see Shao et al. 2006). Bell & de Jong theirassociatedb -bandluminosity,j ,andstellarmass J bJ (2001)arguethattheeffectonanindividualgalaxy’sstel- densities, ρM. These are derived using the following ex- lar mass is less than one might expect (see their fig. 1) pressions: because the observed decrease in total luminosity is off- jbJ =φ∗10−0.4(MB∗−M⊙)Γ(α+2) (3) set by the increased stellar mass-to-light ratio inferred from the redenned colors, therefore yielding comparable and final stellar masses. We explore this in detail in Driver N et al. (2007) and note that while the masses are mod- ρM =X(φi/Ni)10(1.93(g−r)i−0.79)10−0.4(MB,i−M⊙). ified somewhat the final stellar mass breakdown is not i dramatically altered. (4) ThelatterexpressionisfirstshowninDriveretal.(2006) and is based on the color to mass-to-light ratios given by Bell & de Jong (2001) which assume a Salpeter-’lite’ The Millennium Galaxy Catalogue consists of imag- IMF. Note that the (g −r) colour for each galaxy was ing data from the Isaac Newton Telescope and spectro- obtainedbymatchingtheMGCtotheSloanDigitalSky scopic data from the Anglo Australian Telescope, the Survey first data release (Abazajian et al. 2003). ANU 2.3m, the ESO New Technology Telescope, the TelescopioNazionaleGalileo,andthe GeminiTelescope. 3. DISCUSSION The survey has been supported through grants from FromTable 1 we see that 58±6 per cent of the stellar the Particle Physics and Astronomy Research Coun- mass is in the form of galaxy discs, 13±4 per cent in cil (UK) and the Australian Research Council (AUS). redelliptical galaxiesand26±4 per centin classicalred The data and data products are publicly available from bulges. Previouslyit hasbeenreported(Belletal.2003; http://www.eso.org/∼jliske/mgc/ or on request from Baldry et al. 2004; Driver et al. 2006) that 54−60 per J. Liske or S.P. Driver. REFERENCES AbazajianK.,etal.,2003, AJ,126,2081 Ball N., Loveday J., Brunner R.J., 2006, MNRAS, submitted, AllenP.,Driver S.P.,Graham A.W.,Cameron E.,Liske J., Cross (astro-ph/0610171) N.J.G.,DeProprisR.,2006,MNRAS,371,2 BarnesJ.E.,HernquistL.,1992, Nature,360,715 Baldry I., Glazebrook K., Brinkmann J., Ivezic Z., Lupton R.H., BellE.,McIntoshD.H.,KatzN.,WeinbergM.D.,2003,ApJS,149, NicholR.C.,SzalayA.S.,2004,ApJ,600,681 289 BellE.,deJongR.S.,2001, ApJ,550,212 4 Driver et al. TABLE 1 Sample MB∗(−ma5gl)ogh α (10−2h3 Mpcφ−∗3(0.5mag)−1) (108hLj⊙bJMpc−3) (108hMρ⊙MMpc−3) N All −19.84±0.02 −1.15±0.01 1.72±0.05 2.65±0.13 6.2±0.3 7786 AllDiscs −19.44±0.04 −1.15±0.03 1.74±0.09 1.85±0.20 3.6±0.4 6024 Bulges −19.23±0.07 −1.00±0.08 0.64±0.05 0.50±0.11 1.7±0.4 1431 Ellipticals −19.36±0.12 −0.91±0.11 0.38±0.05 0.32±0.11 0.9±0.3 835 BlueEllipticals −19.89±0.50 −1.88±0.22 0.02±0.01 0.11±∞a 0.1±∞a 229 RedEllipticals −19.02±0.11 −0.26±0.13 0.36±0.02 0.21±0.03 0.8±0.2 606 BlueBulges −19.87±0.50 −2.08±0.21 0.02±0.01 0.92±∞a 0.1±∞a 249 RedBulges −19.11±0.07 −0.75±0.08 0.65±0.05 0.42±0.07 1.6±0.3 1182 Ellipticals+Bulges −19.34±0.07 −1.01±0.06 0.96±0.08 0.84±0.16 2.7±0.5 2266 Red(Ellipticals+Bulges) −19.16±0.07 −0.67±0.07 0.97±0.06 0.64±0.08 2.4±0.3 1788 Blue(Ellipticals+Bulges) −19.89±0.32 −1.97±0.14 0.04±0.02 0.20±∞a 0.3±∞a 478 a Theseluminosityfunctionsarepotentiallydivergentwithinthespecifiederrors,givingrisetoextremeluminosityandstellarmassdensities. As wedonotknowwhetherthesedistributionscontinuetodivergeweinfertheluminosityandstellarmassdensitiesrequiredforafullyself-consistent table,i.e.,thestellarmassofbluespheroidsisderivedbysubtractingtheredellipticals’stellarmassfromthetotalellipticals’stellarmass,etc. Choi Y., Park C., Vogeley M., 2006, ApJ, in press, astro- IlbertO.,etal.,2006,A&A,453,809 ph/0611607 KormendyJ.,KennicuttR.C.Jr.,2004,ARA&A,42,603 ConseliceC.J.,2006, MNRAS,submitted, astro-ph/0610016 LiskeJ.,Lemon D.,Driver S.P.,CrossN.J.G.,Couch W.J., 2003, DeProprisR.,LiskeJ.,DriverS.P.,AllenP.D.,CrossN.J.G.,2005, MNRAS,344,307 AJ,130,1516 LiskeJ.,DriverS.P.,AllenP.D.,CrossN.J.G.,DeProprisR.,2006, DePropris R.,Liske J., Conselice C.,Driver S.P.,Graham A.W., MNRAS,369,1547 AllenP.D.,2007,MNRAS,submitted Park C., Choi Y., Vogeley M., Gott II J.R., Blanton M.R., 2006, deVaucouleursG.,1959, HDP,53,275 ApJ,inpress,astro-ph/0611610 DriverS.P.,LiskeJ.,CrossN.J.G.,DeProprisR.,AllenP.D.,2005, PeletierR.F.,BacellsM.,AJ,111,2238 MNRAS,360,81 PhillipsA.C.,etal.,1997,ApJ,489,543 DriverS.P.,etal.,2006,MNRAS,368,414 PoggiantiB.,1997,A&AS,122,399 DriverS.P.,PopescuC.,TuffsR.J.,LiskeJ.,GrahamA.W.,Allen SersicJ.-L.,1963,BAAA,6,41 P.D.,DeProprisR.,2007,MNRAS,submitted Shao Z., Xiao Q., Shen S., Mo H.J., Xia X., Deng Z., 2006, ApJ, EfstathiouG.,EllisR.S.,PetersonB.A.,1988,MNRAS,232,431 submitted,astro-ph/0611714 EllisS.C.,DriverS.P.,AllenP.D.,LiskeJ.,Bland-HawthornJ.,De SimardL.,etal.,2002, ApJS,142,1 ProprisR.,2005, MNRAS,363,1257 StratevaI.,etal.,2001,AJ,122,1861 EllisS.C.,DriverS.P.,LiskeJ.,Graham A.W.,2007, MNRAS,in ZuccaE.,etal.,2006,A&A,455,879 preparation GrahamA.W.,DriverS.P.,2005,PASA,22,118 Guzman R., Jangren A., Koo D.C., Bershady M.C., Simard L., 1998, ApJ,495,13 The MGC: Luminosity functions of bulges and discs 5 Fig. 1.— B-band luminosity functions derived using our variable limitSWML method for the global sample (top left), discs (bottom left),ellipticals(topright),andbulges(bottom right). Ineachcasethederiveddata(squares),andfittedSchechter functions(solidlines) areshown. TheSchechter fitfortheglobal sampleusesalldata points but forthe component luminosityfunctions (discs, ellipticals,and bulges)weonlyusedatapointswithMB <−17mag(seeSection2.3). 6 Driver et al. Fig. 2.—Thetoprowshowsthe(u−r)−ndistribution(dots)forellipticals(left),andbulges(right)withMB <−17mag. Thecontours show the volume-corrected luminosity density for these objects. The natural division between the red and blue populations appears to occur at (u−r)=2mag. The bottom row shows the B-band luminosityfunctions forellipticals (left), andbulges (right), splitinto red, (u−r)>2mag,componentswheredatapointsareshownbysquares,andblue,(u−r)≤2mag,componentswheredatapointsareshown bytriangles. The MGC: Luminosity functions of bulges and discs 7 Fig. 3.— Derived data and Schechter fits for B-band luminosity functions for our three final structural components: discs (circles), redbulgesandellipticals(squares), andblue(pseudo-)bulges andblueellipticals(triangles). TheSchechter fitsonlyusedata pointswith MB <−17mag. ThethickerblacklineshowstheglobalB-bandluminosityfunction.