Major Merging: The Way to Make a Massive, Passive Galaxy
Arjen van der Wel, Hans-Walter Rix, Bradford P. Holden, Eric F. Bell, Aday R. Robaina
aa r X i v : . [ a s t r o - ph . C O ] N ov The Astrophysical Journal Letters, Accepted
Preprint typeset using L A TEX style emulateapj v. 08/22/09
MAJOR MERGING: THE WAY TO MAKE A MASSIVE, PASSIVE GALAXY
Arjen van der Wel , Hans-Walter Rix , Bradford P. Holden , Eric F. Bell & Aday R. Robaina The Astrophysical Journal Letters, Accepted
ABSTRACTWe analyze the projected axial ratio distribution, p ( b/a ), of galaxies that were spectroscopicallyselected from the Sloan Digital Sky Survey (DR6) to have low star formation rates. For these quiescentgalaxies we find a rather abrupt change in p ( b/a ) at a stellar mass of ∼ M ⊙ : at higher massesthere are hardly any galaxies with b/a < .
6, implying that essentially none of them have disk-likeintrinsic shapes and must be spheroidal. This transition mass is ∼ − ∼ M ⊙ , quiescent galaxiesshow a large range in axial ratios, implying a mix of bulge- and disk-dominated galaxies. Our resultstrongly suggests that major merging is the most important, and perhaps only relevant, evolutionarychannel to produce massive ( > M ⊙ ), quiescent galaxies, as it inevitably results in spheroids. Subject headings: galaxies: elliptical and lenticular, cD—galaxies: formation—galaxies: fundamentalparameters—galaxies: statistics—galaxies: structure INTRODUCTION
Even galaxies with little star formation activity con-tinue to evolve, as evidenced by the substantial in-crease of their cosmic stellar mass density over thepast 7 billion years (Bell et al. 2004; Faber et al. 2007;Brown et al. 2007). This must be related to the decreas-ing star formation activity over the same period (e.g.,Le Floc’h et al. 2005), and the production of such qui-escent galaxies through the truncation of star formation(e.g., Faber et al. 2007; Bell et al. 2007); the color scatteramong quiescent galaxies and its evolution are in preciseagreement with such a scenario (Ruhland et al. 2009).There are, however, quiescent galaxies at all redshifts z . . Max-Planck Institute for Astronomy, K¨onigstuhl 17, D-69117,Heidelberg, Germany University of California Observatories/Lick Observatory, Uni-versity of California, Santa Cruz, California, 95064, USA Department of Astronomy, University of Michigan, 500 ChurchStreet, Ann Arbor, Michigan, 48109, USA bly of galaxies. Merging activity among the massivegalaxy population is observed (e.g., van Dokkum et al.1999; van Dokkum 2005; Bell et al. 2006a,b; Lin et al.2008), and has been shown to produce a color-magnitude relation that is in agreement with observa-tions (Skelton et al. 2009). However, its cosmological rel-evance has always been difficult to determine, given theuncertainties in converting observed merger fractions tomerger rates and the associated growth in mass. An in-dependent and indirect indication that massive galaxiesundergo continuous evolution is provided by the recentresult that high-redshift quiescent galaxies are substan-tially smaller than local galaxies with the same mass (see,van der Wel et al. 2008, and references therein). Thisstrongly suggests that mergers are important (see, e.g.,van der Wel et al. 2009), and that the assembly of mas-sive galaxies is continuing up until the present day. An-other indirect, yet powerful, constraint is provided bythe evolution in the clustering and halo occupation dis-tribution of red galaxies (White et al. 2007; Conroy et al.2007; Brown et al. 2008): the evolution in the clusteringstrength of red galaxies is slower than expected in theabsence of merging.In this Letter we address the question whether majormerging is the dominant mechanism for the productionof very massive, quiescent galaxies. The argument thatwe invoke is simply that major merging generally leads torounder galaxies. An analysis of the shape distributionof quiescent galaxies can therefore constrain the impor-tance of merging. Since merging among galaxies withmass ratios of . p ( b/a ), of a large number of galax-ies, selected from the Sloan Digital Sky Survey (SDSS). Axial Ratios of Massive Galaxies Fig. 1.—
Left: axial ratio distribution, p ( b/a ), as a function of stellar mass for spectroscopically selected quiescent galaxies from theSDSS at 0 . < z < .
08. The gray scale represents, normalized to the total number of galaxies in narrow bins of stellar mass, the fractionof galaxies with axial ratio b/a . The upper boundaries below which, as a function of mass, 10%, 30%, 50%, 70%, and 90% of galaxies arelocated, are delineated by the red lines.
Right: fraction of galaxies with axial ratios b/a smaller than 0.8, 0.6, 0.4, and 0.2, as a functionof stellar mass. These figures clearly show that at M ∗ ≥ M ⊙ , the fraction of galaxies with small b/a decreases rapidly with mass.At lower masses, p ( b/a ) is approximately uniform in the range 0 . < b/a < .
9, implying a significant contribution of disks. At highermasses, axial ratios are approximately evenly distributed in the range 0 . < b/a < .
9, which shows that disks must be rare, and galaxiesintrinsically round.
Through a detailed analysis, they infer the intrinsicshape distribution and the effect of extinction. Bothdivide the sample into ’elliptical’ and ’spiral’ galaxies,and confirmed that luminous ’elliptical’ galaxies are,on average, rounder and tri-axial, compared to low-luminosity ’ellipticals’, which are more elongated andoblate (Davies et al. 1983; Franx et al. 1991), and dis-play disky isophotes (Jørgensen & Franx 1994). Thisphenomenon is not recent: Holden et al. (2009) showedthat this trend persists at least out to z ∼
1. Here wepresent a complementary, modified analysis, focusing on p ( b/a ) as a function of stellar mass for quiescent, i.e.,non-star-forming, galaxies. Because mass-to-light ratiosare well constrained by broad-band colors for quiescentgalaxies, stellar mass estimates are robust. This is es-sential for our purposes, as we are interested in the mostmassive objects, i.e., those that populate the exponen-tial tail of the mass function. Furthermore, as opposedto previous studies, we pre-select galaxies independentof their photometric properties. Our shape-independent,spectroscopic selection criteria circumvent the biases thatare potentially introduced by selecting galaxies by their’morphological’ properties, or some pre-defined surfacebrightness profile.With this sample, for which we have determined ax-ial ratios from our own fits to two-dimensional light dis-tributions, we address the following specific questions.Are high-mass, quiescent galaxies rounder than low-mass quiescent galaxies? If so, is there a mass limit atwhich p ( b/a ) distinctly changes, and above which disk-dominated are completely absent? Such evidence wouldimply that the only evolutionary path to such masses is a disk-destroying mechanism, i.e., major merging. THE SAMPLE
We select a sample of 17,480 quiescent galaxies fromData Release 6 of the SDSS (Adelman-McCarthy 2008).Our sample includes galaxies at redshifts 0 . < z < . α emission lines. The se-lection criteria are described and motivated in full byGraves et al. (2009); but as opposed to that work, we donot exclude galaxies with a low concentration index andgalaxies that are fit better by an exponential profile thanby a de Vaucouleurs (1948) profile, because this may ex-clude quiescent, yet disk-like galaxies, which are obvi-ously relevant for quantifying p ( b/a ) of quiescent galax-ies. As a consequence, our sample may include galax-ies with star formation in an extended disk outside theSDSS spectroscopic fiber. This effect, however, does notaffect our main conclusion that quiescent massive galax-ies with prominent disks are extremely rare (see Section3). Rather, such a bias works in the opposite directionin the sense that it would lead to the mistaken inclusionof galaxies with large disks.The exclusion of all galaxies with emission lines alsoexcludes quiescent galaxies with active galactic nuclei.Their number, however, is small, and make up a smallfraction of the population (e.g., Pasquali et al. 2009) thatis negligible for our purposes.The axial ratios were obtained as described byvan der Wel et al. (2008). Briefly, GALFIT (Peng et al.2002) is used to determine from the r -band the radii, ax-ial ratios, position angles, and total magnitudes, assum-ing a de Vaucouleurs (1948) surface brightness profile.We have verified that adopting surface brightness mod-an der Wel et al. 3els with a free S´ersic index does not lead to a significantlydifferent p ( b/a ).The stellar masses are derived with the simple conver-sion from color to mass-to-light ratio (Bell et al. 2003),but are normalized to correspond to the Kroupa (2001)stellar initial mass function. The assumed cosmology is(Ω M , Ω Λ , h ) = (0 . , . , . . < z < .
08 down to M ∗ ∼ × M ⊙ , set by thespectroscopic magnitude limit of the SDSS ( r = 17 . b/a , reported in the following section, is notin any way compromised by selection effects or measure-ment errors. RESULTS AND DISCUSSION
In Figure 1(a) we show p ( b/a ) of the 17,480 spectro-scopically selected, quiescent galaxies as a function ofstellar mass. p ( b/a ) is shown in gray scale, with the per-centiles of the cumulative b/a distribution shown as (red)lines. Figure 1(a) immediately demonstrates that for qui-escent galaxies, the projected axial ratio distribution is astrong function of stellar mass. In the narrow mass range8 × . M ∗ /M ⊙ . × there is a rapid decreasein the number of galaxies with small axial ratios. As fur-ther illustrated by Figure 1(b), above M ∗ ∼ × M ⊙ quiescent galaxies with b/a < . M ∗ & × M ⊙ all but exclude the existence, or the survival, of highlyflattened, disk-like stellar components. As highly flat-tened stellar systems are quite common at lower masses,in the possible realm of plausible progenitors of high-mass galaxies, this result implies the destruction of theflattened component in whatever process causes growthbeyond M ∗ ∼ × M ⊙ . Therefore, our result thatessentially all quiescent galaxies with masses larger than M ∗ ∼ × M ⊙ are round strongly suggests that forsuch galaxies major mergers are the dominant, perhapseven unique, formation channel. The destruction of astellar disk requires a major merger, i.e., a merger involv-ing progenitors with a relatively small mass ratio of atmost ∼
3, mergers with a larger mass ratio leaving stel-lar disks intact (see, e.g., Bekki 1998; Bournaud et al.2004). Moreover, most likely, the progenitors are notvery gas rich, as this would produce a disky remnant(e.g., Naab et al. 2006).It has been suggested that cold flows are responsiblefor the formation of massive, classical bulges at high red-shift (Dekel et al. 2009; Ceverino et al. 2009). In thisscenario, intensely star-forming ’knots’ merge, forminga massive bulge (see also Noguchi 1999). However, asubstantial fraction of the mass ( ∼ b/a > .
8) at the veryhighest masses (
M > × M ⊙ ) signifies that suchhigh mass galaxies are typically brightest group/clustergalaxies, which tend to be slightly more elongated than’normal’ massive elliptical galaxies (see, Bernardi et al.2008).As already noted above, at masses lower than M ∗ ∼ M ⊙ , quiescent galaxies display a large range in ax-ial ratios, which implies that star formation truncationmechanisms below 10 M ⊙ are often not associated withthe destruction of the disk. It remains to be testedwhether p ( b/a ) of low-mass quiescent galaxies is simi-lar to or different from p ( b/a ) of star-forming galaxiesin the same mass range. Such an analysis, which is non-trivial because of the effects of extinction and color gradi-ents, will constrain the degree to which mergers or bulgegrowth regulate star formation at these lower masses.Another open question concerns the number and prop-erties of massive, star-forming galaxies. Morphologicalstudies suggest that a large fraction (20% − M ∼ M ⊙ are late-typegalaxies (van der Wel 2008; Bamford et al. 2009), andtheir high masses of at least some of these objects are con-firmed by their rotational velocities (e.g., Courteau et al.2007). Yet, the degree to which such galaxies are disk-dominated and should be considered actively star form-ing remains to be determined. It would, therefore, bepremature to conclude that merging is the only way toproduce a massive galaxy in general, and therefore re-strict this proposition to the formation of massive, qui-escent galaxies.The picture sketched by the axial ratio distribution ofquiescent galaxies is in agreement with the strong cor-relation between structure and mass for galaxies in gen-eral (e.g., Kauffmann et al. 2003; van der Wel 2008) andearly-type galaxies in particular (e.g., Caon et al. 1993;Graham & Guzm´an 2003): high-mass galaxies are moreconcentrated and have higher S´ersic indices than low-mass galaxies. These trends are an indirect indication ofa decreasing importance of disks for galaxies with highermasses, although part of this trend is caused by the in-crease in S´ersic index with galaxy mass among spheroidalgalaxies. In our sample we see a similar trend: in themass range 10 < M ∗ /M ⊙ < × , 41% of the galax-ies have S´ersic indices n <
3, whereas at higher masses, M ∗ > × M ⊙ , only 3% have such low S´ersic indices.We postpone a full exploration of the joint behavior ofshape and structure as a function of galaxy mass until afuture paper, but it is encouraging that the apparent ab-sence of prominent disks in high-mass, quiescent galaxiesis reflected in both the S´ersic index and p ( b/a