The HST/ACS Coma Cluster Survey IV. Intergalactic Globular Clusters and the Massive Globular Cluster System at the Core of the Coma Galaxy Cluster

0.99 MB·English

by  Eric W. Peng

Accepted for publicationin The Astrophysical Journal
PreprinttypesetusingLATEXstyleemulateapjv.08/13/06
THE HST/ACS COMA CLUSTER SURVEY IV. INTERGALACTIC GLOBULAR CLUSTERS AND THE
MASSIVE GLOBULAR CLUSTER SYSTEM AT THE CORE OF THE COMA GALAXY CLUSTER 1
Eric W. Peng2,3,4, Henry C. Ferguson4, Paul Goudfrooij4, Derek Hammer5, John R. Lucey6, Ronald O.
Marzke7, Thomas H. Puzia8,9, David Carter10, Marc Balcells11, Terry Bridges12, Kristin Chiboucas13, Carlos
del Burgo14, Alister W. Graham15, Rafael Guzma´n16, Michael J. Hudson17, Ana Matkovic´18, David Merritt19,
Bryan W. Miller20, Mustapha Mouhcine10, Steven Phillipps21, Ray Sharples6, Russell J. Smith6, Brent Tully12,
and Gijs Verdoes Kleijn22
Accepted for publication inThe Astrophysical Journal
1
1
ABSTRACT
0
2 Intraclusterstellarpopulationsareanaturalresultoftidalinteractionsingalaxyclusters. Measuring
these populations is difficult, but important for understanding the assembly of the most massive
n
galaxies. The Coma cluster of galaxiesis one of the nearesttruly massive galaxyclusters, and is host
a
J to a correspondingly large system of globular clusters (GCs). We use imaging from the HST/ACS
ComaCluster Surveytopresentthe firstdefinitivedetectionofalargepopulationofintraclusterGCs
6
(IGCs)thatfillstheComaclustercoreandisnotassociatedwithindividualgalaxies. TheGCsurface

density profile around the central massive elliptical galaxy, NGC 4874,is dominated at large radii by
]
A a population of IGCs that extend to the limit of our data (R<520 kpc). We estimate that there are
G 47000±1600 (random) +−45000000 (systematic) IGCs out to this radius, and that they make up ∼70% of
the central GC system, making this the largest GC system in the nearby Universe. Even including
.
h the GC systems of other cluster galaxies,the IGCs still make up 30–45%of the GCs in the cluster
p core. Observational limits from previous studies of the intracluste∼r light (ICL) suggest that the IGC
- population has a high specific frequency. If the IGC population has a specific frequency similar to
o
high-S dwarf galaxies, then the ICL has a mean surface brightness of µ 27 mag arcsec−2 and
r N V
st a total stellar mass of roughly 1012M⊙ within the cluster core. The ICL m≈akes up approximately
a halfofthe stellarluminosityandone-thirdofthe stellarmassofthe central(NGC4874+ICL)system.
[ The color distribution of the IGC population is bimodal, with blue, metal-poor GCs outnumbering

red, metal-richGCs by a ratio of 4:1. The inner GCs associatedwith NGC 4874 also have a bimodal
2
distribution in color, but with a redder metal-poor population. The fraction of red IGCs (20%), and
v the red color of those GCs, implies that IGCs can originate from the halos of relatively massive, L∗
0
galaxies, and not solely from the disruption of dwarf galaxies.
0
0 Subject headings: galaxies: elliptical and lenticular, cD — galaxies: clusters: individual: Coma —
1 galaxies: halos — galaxies: evolution — galaxies: star clusters: general – globular
. clusters: general
1
0
1
1. INTRODUCTION
1
: 1BasedonobservationswiththeNASA/ESAHubbleSpaceTele- 1.1. Intracluster Stellar Populations and Hierarchical
v scope obtained at the Space Telescope Science Institute, which is
i operatedbytheAssociationofUniversitiesforResearchinAstron- Galaxy Formation
X omy,Inc.,underNASAcontract NAS5-26555.
Massive elliptical galaxies at the centers of galaxy
2Department ofAstronomy, PekingUniversity, Beijing100871,
r China;[email protected] clusters—often brightest cluster galaxies (BCGs) and
a
3Kavli Institute for Astronomy and Astrophysics, Peking Uni- sometimescDgalaxies—generallyhavelittleongoingstar
versity,Beijing100871, China
4SpaceTelescopeScienceInstitute,3700SanMartinDrive,Bal-
timore,MD21228, USA 13Institute for Astronomy, University of Hawai’i, 2680 Wood-
5DepartmentofPhysicsandAstronomy,JohnsHopkinsUniver- lawnDrive,Honolulu,HI96822, USA
sity3400N.CharlesSt.,Baltimore,MD21228,USA 14School of Cosmic Physics, Dublin Institute for Advanced
6Department of Physics, University of Durham, South Road, Studies,31FitzwilliamPlace,Dublin2,Ireland
DurhamDH13LE,UK 15CentreforAstrophysicsandSupercomputing,SwinburneUni-
7Dept. ofPhysics&Astronomy,SanFranciscoStateUniversity, versityofTechnology, Hawthorn,Victoria3122, Australia
1600HollowayAvenue, SanFrancisco,CA94132, USA 16Department of Astronomy, University of Florida, P.O. Box
8National Research Council of Canada, Herzberg Institute of 112055,Gainesville,FL32611, USA
Astrophysics, 5071 West Saanich Road, Victoria, BC V9E 2E7, 17Physics and Astronomy, University of Waterloo, 200 Univer-
Canada sityAvenueWest,Waterloo, Ontario,CanadaN2L3G1,Canada
9DepartmentodeAstronom´ıayAstrof´ısica,PontificiaUniversi- 18Department of Astronomy and Astrophysics, Pennsylvania
dadCat´olicadeChile,Av.Vicun˜aMackenna4860,7820436Macul, StateUniversity,UniversityPark,PA16802, USA
Santiago, Chile 19CenterforComputationalRelativityandGravitationandDe-
10AstrophysicsResearchInstitute,LiverpoolJohnMooresUni- partmentofPhysics,RochesterInstituteofTechnology,Rochester,
versity, Twelve Quays House, Egerton Wharf, Birkenhead CH41 NY14623,USA
1LD,UK 20GeminiObservatory, Casilla603,LaSerena,Chile
11InstitutodeAstrofisicadeCanarias,38200LaLaguna,Tener- 21AstrophysicsGroup,H.H.WillsPhysicsLaboratory,Univer-
ife,Spain sityofBristol,TyndallAvenue,BristolBS81TL,UK
12Department of Physics,EngineeringPhysics andAstronomy, 22Kapteyn Astronomical Institute, University of Groningen,
Queen’sUniversity,Kingston,ONK7L3N6,Canada P.O.Box800,9700AVGroningen, TheNetherlands
2 Peng et al.
formation with only minor evolution since z 1, and &Mendes deOliveira2005). Thesestudiestendtoshow
∼
onlyashallowrelationshipbetweentheirstellarmassand that the richer and more massive the group or cluster,
their host cluster mass (e.g., Lin & Mohr 2004; Whiley the larger the fraction of intergalactic light.
etal.2008). Inthestandardhierarchicalparadigm,how-
1.2. Intergalactic Globular Clusters
ever,themostmassivehalosshouldbethelasttoassem-
ble, and so these galaxies have traditionally presented Another important clue to the formation of massive
problems for formation models. ellipticals is that those residing at the centers of galaxy
Recent simulations suggest that this paradox can be clusters often host extremely large populations of glob-
resolved in a picture where the stars that end up in the ular clusters (GCs). The star forming events that form
most massive galaxies form early, energy feedback from globularclusterswillmostlyformstarsthatendupinthe
supernovae and active galactic nuclei subsequently sup- field,so itis naturalthatthe numberofGCs ina galaxy
press star formation, and the assembly of these galaxies shouldroughlyscalewiththatgalaxy’sstellarluminosity
through dry mergers continues right up to the present ormass. However,theratioofGCs tostarlight—usually
day(DeLucia&Blaizot2007;althoughseeBildfelletal. characterizedasthespecificfrequency,S (Harris&van
N
2008 for evidence that feedback is not 100% efficient). den Bergh 1981)—has long been known to vary across
Thus, even if there is little star formation at late times, galaxy mass and morphology, with giant central ellipti-
these galaxiesare still expected to further assemble stel- calgalaxiesharboringthelargestGCsystemsandhaving
lar mass at z <2. This predicted increase in the masses someofthe highestspecificfrequencies. Thisabundance
ofBCGs overtime, however,may be in conflict with ob- ofGCsingalaxyclustersappearsexplainableifthenum-
servations that show little mass evolution of BCGs from ber of GCs scales with either total baryonic mass at the
z 1.5 to the present (Collins et al. 2009). cluster center, including hot gas (McLaughlin 1999), or
∼
It is expected that the dry merging or tidal stripping the total dynamical mass of the cluster (Blakeslee et al.
of satellites should not only contribute stars to the cen- 1997). Blakeslee (1997, 1999) observed that the number
tral galaxy itself, but also to an intracluster component of GCs in galaxy clusters was directly related to cluster
that has previously been associated with the extended mass, but the relatively constant BCG luminosity thus
stellar envelopes of cD galaxies (Matthews, Morgan & led to high S . It is possible that the high specific fre-
N
Schmidt 1964), and is sometimes labeled as a “diffuse quencies are because measurements of the galaxy lumi-
stellarcomponent”(Monacoetal.2006)orsimply“intra- nosity typically do not include a substantial ICL com-
clusterlight”(ICL).Thiscomponentcanmakeupalarge ponent, and that the high S in central cluster galaxies
N
fraction of the total luminosity at the center of galaxy would be more normal if the ICL was included. Galaxy
clusters (Oemler 1976), and if added to the stellar mass specificfrequencyalsovarieswithgalaxystellarmass(or
of central cluster galaxies, might naturally explain cur- luminosity)inawaythatisconsistentwiththeexpected
rent contradictions between simulation and observation. variationingalaxystellarmassfraction(ormass-to-light
Purcell, Bullock, & Zentner (2007)simulated the forma- ratio) (Peng et al. 2008; Spitler et al. 2008).
tion of the ICL from the shredding of satellite galaxies, ThisconnectionbetweenGCsandtotalmasshasinter-
finding that in massive clusters, the ICL can dominate esting implications, particularly in massive galaxy clus-
the total stellar mass of the combined ICL+BCG sys- ters where the predicted build up of stellar mass in cen-
tem,whichisconsistentwithobservationsoflowredshift tral galaxies should be paralleled by the build up of a
clusters (Gonzalez, Zabludoff, & Zaritsky 2005; Seigar, large GC system. If much of the stellar mass in galaxy
Graham, & Jerjen 2007). clusters resides in the low surface brightness ICL then
In fact, there is increasing observational evidence there should also be a corresponding population of intr-
that a significant fraction (10–40%) of the total stellar acluster GCs (IGCs) that are not gravitationally bound
light in a galaxy cluster is intergalactic. Starting with to individual galaxies, but directly to the cluster itself.
Zwicky(1951),manydetectionsoflowsurfacebrightness Moreover,thedetectionofpointsourceIGCsinthenear-
starlight in galaxy clusters—both in cD envelopes and estclustersis a mucheasierobservationalendeavorthan
in the regions between galaxies—support the existence measuring the faint ICL, giving us a window onto the
of substantial intracluster stellar populations (Welch & nature of the diffuse stellar content.
Sastry1971;Usonetal.1991;Vilchez-Gomezetal.1994; There are other reasons to expect substantial popu-
Gregg & West 1998; Trentham & Mobasher 1998; Feld- lations of IGCs. West et al. (1993) proposed that GC
meier et al. 2002,2004a;Lin & Mohr 2004;Adami et al. formation may be biased toward the largest mass over-
2005; Zibetti et al. 2005; Mihos et al. 2005; Gonzalez densities,i.e.galaxyclusters. Westetal.(1995)alsopro-
et al. 2005; Seigar et al. 2007; Gonzalez, Zaritsky, & posed that populations of IGCs were responsible for the
Zabludoff 2007; Krick & Bernstein 2007). In nearby highS seenincDgalaxies. Morerecently,spectroscopy
N
clusters, there have also been direct detections of in- of ultra-compact dwarfs (UCDs) and massive GCs (also
tergalactic red giant branch stars (Ferguson, Tanvir, & dubbeddwarf-globulartransitionobjects; Ha¸seganetal.
von Hippel 1998), asymptotic giant branch stars (Dur- 2005)haveuncoveredapopulationofcompactstellarsys-
rell et al. 2002) planetary nebulae (Theuns & Warren temsingalaxyclustersresemblingthemostmassiveGCs
1997; M´endez et al. 1997; Feldmeier, Ciardullo, & Ja- or dE nuclei stripped of their host galaxies (Drinkwater
coby 1998; Feldmeier et al. 2004b; Okamura et al. 2002; et al. 2003; Hilker et al. 2007; Mieske et al. 2008; Gregg
Arnaboldi et al. 2004; Gerhard et al. 2007; Arnaboldi et al. 2009; Madrid et al. 2010; Chiboucas et al. 2010).
et al. 2007, Castro-Rodrigu´ez et al. 2009; Doherty et al. Theseobjects,whilegenerallymoremassivethantypical
2009), novae (Neill, Shara, & Oegerle 2005), and super- GCs and consequently may have different origins, might
novae (Gal-Yam et al. 2003). Detections of intergalactic betheso-called“tipoftheiceberg”foralargepopulation
lighthavealso beenmade incompactgroups(Da Rocha of free-floating, normal globular clusters.
Intergalactic Globular Clusters in the Coma Cluster 3
In fact, a number of extragalacticGC studies over the (Kavelaars et al. 2000; Harris et al. 2000; Harris et al.
past few years have strongly suggested the presence of 2009)all point to large GC systems aroundthe cluster’s
IGCsinnearbygalaxyclusters. IntheVirgoandFornax giant elliptical galaxies, particularly around the central
Clusters, serendipitous discoveries of GCs in intergalac- massivegalaxy,NGC4874(Harrisetal.2009). Although
tic regions using HST imaging (Williams et al. 2007), manyphotometricstudiessupporttheexistenceofanin-
ground-based imaging (Bassino et al. 2003) and spec- tracluster stellar light component in Coma (e.g., Zwicky
troscopy (Bergond et al. 2007) point to the existence of 1951; de Vaucouleurs & de Vaucouleurs 1970; Welch &
IGC populations. However, it is often unclear whether Sastry 1971; Kormendy & Bahcall 1974; Matilla 1977;
these GCs are truly intergalactic, or are part of the ex- Melnick, White & Hoessel 1977; Thuan & Kormendy
tended halos of cluster galaxies (c.f. Schuberth et al. 1977; Gregg & West 1998; Calca´neo-Rolda´n et al. 2000;
2008). A recent study by Lee, Park & Hwang (2010), Adami et al. 2005), and even velocities for intracluster
however, used data from the Sloan Digital Sky Survey PNehavebeenmeasured(Gerhardetal.2007;Arnaboldi
(SDSS) and found statistically significant detections of etal.2007),evidencefororagainsttheexistenceofIGCs
GC candidates throughout the Virgo cluster. In more ismuchmoremuddled. AsearchforIGCsinComausing
distant galaxy clusters, candidate IGC populations have ground-based data by Mar´ın-Franch et al. (2002, 2003)
been identified as point source excesses in HST imaging did not find a surface brightness fluctuation signal that
(Jorda´n et al. 2003; West et al. 2011). would have hinted at the presence of IGCs, mainly be-
It is possible that some IGCs and intracluster stars causeoftheshallowlimitingmagnitudeoftheirphotom-
formed in situ, i.e. in cold, intergalactic gas that never etry.
accreted onto or was stripped from galaxies. It is also At a distance of 100 Mpc, the value we adopt for this
possible that IGCs formed very early and at high effi- paper (m M = 35, Carter et al. 2008), 1′ on the sky
−
ciencies in dwarf-sized subhalos (e.g., Moore et al. 2006; subtends 29 kpc in the cluster, and the mean of the GC
Peng et al. 2008), and whose host galaxies were subse- luminosityfunction(GCLF)ingiantellipticalsisI =
Vega
quentlytidallydestroyedbyinteractionswiththecluster 26.44 mag. This regime of projected areal coverage and
potential. Another possibility is that IGCs were formed GC apparent brightness makes it reasonable to conduct
in larger galaxies and were stripped through tidal inter- a contiguous survey of the Coma cluster core for GCs,
actions with the cluster potential or with other galaxies unbiasedbythelocationsofindividualgalaxies,usingthe
(see, e.g., simulations of Yahagi & Bekki 2005; Bekki & HST Advanced Camera for Surveys (ACS) Wide Field
Yahagi 2006). Channel (WFC).
The formation of the IGC population is obviously TheHST/ACSComaClusterSurveyisaTreasurySur-
linkedto that of the ICL, althoughthe observedproper- vey originally approved for 164 orbits. One of the main
ties of the two populations may be different. For exam- components of this survey was a contiguous ACS/WFC
ple, the detectability of the ICL is highly dependent on mosaic of the core of the Coma cluster, making it the
its surface brightness, whereas IGCs are detectable even idealdatasettoinvestigatetheexistenceofintergalactic
in isolation. Simulations by Rudick et al. (2009) show GCs. Hints of this population have already appeared in
that the ICL is supplied by tidal streams that originally studies of UCDs by Madrid et al. (2010) and Chiboucas
have relatively high surface brightness but then disperse et al. (2010). This paper presents the first compilation
tobecomefainterandhardertodetect. ICLstudiesusing and description of IGCs in the Coma cluster.
surfacephotometryarethusmoresensitivetorecentdis- 2. OBSERVATIONSANDDATA
ruptions,whereastheIGCpopulationisalesstemporally
2.1. Imaging Data
biased tracer of the full intracluster stellar population.
ThestudyofextragalacticGCsystemshasbeentrans- The data used in this study are from the HST/ACS
formed by the high spatial resolution imaging of the ComaClusterSurvey. Thesurveyobservationsanddata
HubbleSpace Telescope(HST),withobservationsofhun- reduction are described in detail by Carter et al. (2008),
dreds of GC systems now in the archives and published the catalog generation for the public data release is de-
literature (e.g., Jord´an et al. 2009). HST’s deep sen- scribed by Hammer et al. (2010), and an in depth anal-
sitivity to compact or unresolved sources, and its abil- ysis of galaxystructuralparameters and completeness is
ity to distinguish background galaxies from likely GCs, presented in Balcells et al. (2010). We summarize the
makesitanidealtoolforextragalacticGCstudies. How- relevant information here.
ever, the relatively narrow field of view of HST’s cam- The Coma Cluster Survey,as originallydesigned, con-
eras and the close proximity of the galaxies being stud- sistedofa largecentralACS mosaicof the Comacluster
ied (D . 100 Mpc) means that most HST studies have core, and 40 targeted observations in the outer regions
focusedonGCsystemsdirectlyassociatedwithgalaxies. of the cluster. The centralmosaic was designedto be 42
Observations of wider fields are usually conducted with contiguous ACS/WFC pointings in a 7 6 tiling config-
ground-basedtelescopes (e.g. McLaughlin 1999; Bassino uration, and covering an 21′ 18′ area×. Each pointing
etal.2006;Rhodeetal.2007),thatgainareaatthe cost is observed in two filters, F4×75W (g) and F814W (I),
of spatial resolution. with exposure times of 2560s and 1400s, respectively.
Unfortunately, the failure of the ACS/WFC (January,
1.3. The Coma Cluster of Galaxies
2007)meantthatonly28%ofthesurveywascompleted:
TheComaClusterofgalaxies(Abell1656)isoneofthe 19 pointings in or around the central mosaic, and 6 in
nearest rich, dense clusters, and is a fundamental target the outer regions. Within the core, the central galaxy
forextragalacticstudies. StudiesofGCsystemsinComa NGC 4874 was observed, but the other giant elliptical,
from the ground using surface brightness fluctuations NGC 4889, was not imaged before the ACS failure. De-
(Blakeslee et al. 1997; Blakeslee 1999) and using HST spite the shortfall, the current observations still provide
4 Peng et al.
Fig. 1.—(a)TheACSF814Wimageofthe central pointing(Visit19) containingNGC 4874and manyother brightellipticalgalaxies.
Northis left, and East isdown. Theimage is202′′ (98 kpc) ona side. (b) Thesame pointing but after our iterative galaxy subtraction.
Residuals at galaxy centers are still visible at this contrast level, but the overall large scale gradients in the background light have been
removed.
the largest set of deep, high resolution imaging avail- except in the close vicinity of cluster galaxies. The one
able for this important galaxy cluster. Recent studies of exception is Visit 19, which contains NGC 4874 and
compact galaxies in Coma (Price et al. 2009) and spec- many other ellipticals in the cluster core (Figure 1a).
troscopyofComaclustermembers(Smithetal.2009)are This pointing is nearly entirely dominated by the light
partofaconcertedefforttostudygalaxyevolutioninthe from one galaxy or another; it is also the one with
Coma cluster built around this HST Treasury survey. the highest concentration of GCs. To address this, we
The ACS data reduction was performed using a dedi- implemented an iterative galaxy subtraction algorithm
cated Pyraf/STSDAS pipeline that registered and com- thatproducedafullybackgroundsubtractedimage(Fig-
bined images while performing cosmic-ray rejection. ure 1b).
The dithered images were combined using Multidrizzle We subtracted the 10 brightest galaxies from Visit
(Koekemoer et al. 2003), which uses the Drizzle algo- 19. We started from the brightest (NGC 4874 itself)
rithm (Fruchter & Hook 2002). For this study, we used and worked to the faintest of the ten. In each case,
thedatadrizzledfortheACSComaSurveyDataRelease we first manually masked all the bright galaxies except
2 (DR2). However, except for Visits 3, 10, and 57, the for the one being subtracted. We then used the IRAF
F814W images on which the bulk of this paper is based ellipseandbmodeltasks(Jedrzejewski1987)to model
are identical to those in Data Release 1 (DR1). the isophotes of the object galaxy. Because the ellipse
fitting only occurs out to a finite radius, the resulting
model will have finite extent, and the subsequent sub-
2.2. Object Catalogs and Galaxy Subtraction
traction will leave a sharp discontinuity in the image.
Our images of the Coma cluster reveal a striking For convenience of object detection, we extended the
amountofdetail: cluster members acrossthe massspec- ellipse-generated models with a power-law fit to the
trum, globular clusters, background galaxies, and a few last five data points in the profile at fixed ellipticity.
foreground stars. Given the different spatial scales of This allowed for a smooth subtraction out to the bor-
these objects on the sky, it is important to generate cat- dersofthe ACSimage. After subtractingonegalaxy,we
alogs with detection parameters optimized for the sub- thenrepeatedthe processwiththe nextbrightestgalaxy
ject under study. For our purpose of studying the glob- on the subtracted image. After the last galaxy was sub-
ular clusters between galaxies, we used the well-tested tracted,weusedSourceExtractortocreateandsubtract
Source Extractor software package (Bertin & Arnouts a background map that removed large scale variations.
1996)with parameters optimized for point source detec- Thislaststepisimportantbecauseitallowsustorecover
tion, and which are effectively identical to those used by from any large scale over- or under-subtractions due to
Hammer et al. (2010) in the public data release1. Pho- mismatches between the power-law extensions and the
tometry was put onthe AB magnitude systemusing the true surface brightness profiles of the galaxy. A similar
zeropointsofSirannietal.(2005). Allmagnitudesinthis technique was used with success by Jord´an et al. (2004)
paper are AB unless otherwise specified. in ACS images of Virgo cluster galaxies, although that
For most visits, the area between galaxies is much was only for single galaxies.
larger than that occupied by galaxies and the catalogs After galaxy subtraction, we generated catalogs with
can be considered effectively complete to the same level Source Extractor, using variance maps that accounted
fortheextraPoissonnoiseexpectedfromthe subtracted
1 For details, visit the Coma Cluster Survey website at galaxylight. These catalogscontainobjects muchcloser
http://astronomy.swin.edu.au/coma/
Intergalactic Globular Clusters in the Coma Cluster 5
to the centers of galaxies, and to NGC 4874 in partic-
ular. Photometry was obtained by using 3 pixel radius 22 Visit 19 Visit 59
(0′.′15) circular apertures with aperture corrections and
zeropoints from Sirianni et al. (2005). Unless otherwise
23
specified,allmagnitudesinthispaperareontheABsys-
tem. These objects are included in the DR2 catalogs of
Hammer et al. (2010). 24
2.3. Completeness I
25
We use artificial star tests to quantify the spatially
varying detection efficiency across our images. This is
particularly important for the galaxy-subtracted image 26
containing NGC 4874, where the bright galaxy light af-
fects the depth of our observations.
27
We first use routines in DAOPHOT II (Stetson 1987)
toconstructanempiricalPSFusingbrightpointsources
−0.5 0.0 0.5 1.0 1.5 −0.5 0.0 0.5 1.0 1.5
in Visit 19. At the distance of the Coma Cluster, nearly C C
4−10 4−10
allglobularclustersareunresolvedwithHST,andcanbe
well-approximatedby point sources (the mean half-light Fig. 2.— I magnitude versus concentration index (C4−10 =
radius of GCs, rh 3 pc, is only 6% the full width m4pix − m10pix) for Visit 19, the central pointing containing
half maximum of t≈he point spread∼function). Because NGC 4874 (left) and Visit 59, the background pointing most dis-
tant from the cluster center which contains mostly background
detection is done only in the F814W band, we only add
galaxies(right). TheverticallocusofpointsaroundC4−10 =0.45
artificial stars to these images. contains point sources. Most of the point sources in Visit 19 are
When adding point sources into the images, we avoid likelytobeGCs. Thebackground galaxies aremostlyresolvedto
bemoreextended thantheGCsuntilI ∼27,wheresomeoverlap
objects in the image as well as artificial stars already
the stellar locus. The red outlines shows our selection region for
placed so as to avoid incompleteness due to confusion.
GCs.
We run the exact same detection pipeline on these im-
ages as we do to create our object catalog and record ing a variable width of the selection region with mag-
whether the objects were detected as a function of mag- nitude, the variations were not significant and we de-
nitudeandposition. Thenumberofartificialstarsadded cided in the end that simplicity wasbest, choosinga cut
and measured—7,000,000in Visit 19 alone, and approx- of 0.2 mag around the median concentration for point
imately4240arcmin−2 forthe othervisits—ensuresthat sou±rces ( C4−10 =0.45).
h i
we can derive a completeness curve for any position in For the purposes of this study, we wish to maximize
the survey, and for any GC selection criteria. the number of good GC candidates, while also balanc-
In a typical blank area in our images observed with ing the increasing number of background contaminants
the full exposure time, the 90% completeness level is at with magnitude. Because of the depth and high spa-
I 26.8 mag, and the 50% completeness level is at I tialresolutionof our data,we chose a fairly conservative
27≈.3 mag. At R 2′ from the center of NGC 487≈4, magnitude limit, including objects with I < 26.5 mag.
≈
however,these limits are 1.5 mag shallower. At this magnitude, our data is 97% complete in re-
gions free of galaxy light, so completeness corrections
2.4. Globular Cluster Candidate Selection
are only important toward the centers of galaxies. At
OneofthemainbenefitsofGCstudieswithHSTisthe the Coma Cluster distance (m M = 35), assuming
−
abilitytousemorphologyandresolutiontoseparateGCs an extinction A = 0.017 mag for NGC 4874 (Schlegel,
I
from their main contaminants, background galaxies. At Finkbeiner, & Davis 1998), this limit should include a
Coma distances, GCs are point sources when observed significant fraction of the GCs in a Gaussian GC lumi-
with HST, but the great majority of background galax- nosity function (GCLF) typical of giant ellipticals. We
ies are resolved. We use this ability to select against use the recently measured I-band GCLF measurement
backgroundcontaminantsandproducearelativelyclean for the Virgo cD galaxy M87 (Peng et al. 2009), which
sample of GC candidates. was performed with deep HST/ACS observations in the
We use a rough but effective concentration crite- sameF814Wfilter usedbythe ACSComaSurvey. Peng
rion to select GCs. Figure 2 shows the “magnitude- et al. (2009) quote a GCLF Gaussian mean and sigma
concentration”diagramfor objects, where we measure a of µ = 8.56 mag and σ = 1.37 mag. For AB
I,Vega
−
concentration index, C4−10 using the difference in mag- magnitudes,weadd0.436magto µI,Vega (Siriannietal.
nitude measured in a 4 pixel diameter aperture and a 2005). Assuming these values fora GaussianGCLF, our
10 pixel diameter aperture. Figure 2 shows that this GC catalog magnitude limit should include 39% of all
∼
index works well to distinguish point sources from ex- GCs, and 75% of the luminosity in GCs.
∼
tended sources. Here, we show the distribution of ob- This is likely an oversimplification, however, as both
jects in Visit 19 (the one containing NGC 4874), which the mean and width of the GCLF is known to vary with
has the largest number of GC candidates. We overplot galaxy mass (Jorda´n et al. 2006; 2007). If we assume
the objects from Visit 59, which is the most remote of a Gaussian GCLF typical of dwarf ellipticals in clusters
ourfields,andcontainsmostlybackgroundgalaxies. The (µ = 8.1 mag and σ = 1.1 mag, Miller & Lotz
I,Vega
−
red lines show our selection region where we exclude 2007), then our limit includes 22% of the total num-
∼
nearly all of the background galaxies. Although we ex- ber. This discrepancy is one of the main systematic un-
perimented with different cuts in this diagram, includ- certainties in our analysis. We emphasize, however,that
6 Peng et al.
Fig. 4.— Smoothed spatial distribution of GCs in the Coma
Cluster core (30.′8×23.′0, 900×670 kpc). Pixels are 20′′ on a
side, and color represents the surface density of GCs, corrected
for completeness (blue to red denotes low to high density). The
entire image has been smoothed by a Gaussian kernel with σ =
30′′. The dominant concentration of GCs is around NGC 4874,
Fig. 3.— Spatial distribution of ACS GC candidates shown
on a 1◦ ×1◦ Digitized Sky Survey image of the Coma Cluster andanextended structureof GCs appears toconnect NGC 4874,
4889,and4908. Somepeaksinthedistributionrepresentindividual
with North up and East to the left (1.75×1.75 Mpc at Coma
clustergalaxies.
distance). Atthetopleftoftheimageistheobservedportions of
the cluster core central mosaic. The largestconcentration of GCs
is around the central galaxy, NGC 4874. The other large galaxy, to NGC 4839(top) and NGC 4827,two giantearly-type
unobserved by ACS, is NGC 4889. At the bottom of the image galaxies. Thethreeotherfieldstothesoutharenotnear
are six fields in the outer regions of the cluster. The three outer
massive cluster members. We take these three southern
fields to the right (west) show higher numbers of GC candidates
because of their proximity to large galaxies. The eastern three fieldsasanupperlimitonthebackgroundcontamination
outerfields(bottomcenter)arenotnearlargegalaxiesandareused from foreground stars and compact distant galaxies. All
asbackgroundfields. ThedensityofGCcandidatesthroughoutthe
of these fields have fewer GC candidates than does any
entireclustercoreismuchhigherthaninthebackgroundregions,
field in the central mosaic.
implying a large population of intracluster GCs. The ACS field
sizesareroughly202′′onaside. Other than an obvious concentration around
NGC 4874 and NGC 4889 (the latter of which was
changing the assumed GCLF does not affect the signif- not observed with ACS), the GC distribution is rela-
icance of our result, just the inferred total number of tively uniform across most of the central mosaic and
GCs. Given that the depth of our data is not sufficient not spatially clustered; i.e., with the exception of the
to measure the GCLF parameters directly, we choose to two central ellipticals, the spatial structure of the GCs
assume the brighter GCLF, seen in giant ellipticals, as is not highly correlated with the positions of cluster
thiswillgiveusalowerestimateforthenumberofGCsin galaxies. This is partly a bias introduced by the failure
any given area. The numbers could be higher by 80% of Source Extractor to detect GCs that are immediately
∼
in regions where the GCLF for dwarf ellipticals is more inthevicinityofbrightgalaxies. However,GCdetection
representative. should not be a problem in the halos of the galaxies,
We also introduce a broadcolor cut of 0.6<(g I)< and except in a few cases we do not detect the kind of
−
1.5 that should include all old globular clusters. This small-scale substructure one would expect in the cluster
color range is based on the transformed g–z colors of GC distribution if all the GCs were tightly associated
GCs in the ACS Virgo Cluster Survey (Coˆt´e et al. 2004; with galaxies.
Peng et al. 2006), and mainly eliminates distant, com- Figure 4 shows more clearly the distribution of GCs
pact red galaxies. The ages of extragalactic GCs across inthe cluster core. To produce this figure, we divide the
all metallicities are primarily old (> 5 Gyr), especially coreregioninto20′′ 20′′“pixels”witheachrepresenting
those associated with massive early-type galaxies (e.g. thesurfacedensityo×fGCs,correctedforspatiallyvarying
Peng et al. 2004; Puzia et al. 2005; Beasley et al. 2008; completeness,andsmoothedwithaGaussiankernelwith
Woodley et al. 2010), so this color range should include σ = 30′′. The large concentration of GCs at the center-
all bona fide GCs. right is the GC system of NGC 4874. The GC system
of NGC 4889 is also evident, although the galaxy itself
3. SPATIALDISTRIBUTIONOFGCCANDIDATES wasnot observed. While Figure 3 showsthat the overall
Figure 3 plots the locations of GC candidates in our surfacedensityofGCsiswellabovethebackground,Fig-
ACSimagesonaDSSimageofthecluster. Whileitisnot ure 4 shows hints of large-scale substructure in the GC
surprisingthatthenumberofGCsishigharoundmassive spatial distribution. There appears to be an extended
ellipticals suchas NGC 4874,what is striking aboutthis structureofIGCsconnectingNGC4874toNGC4889to
figureisthatthenumberofGCsacrosstheentirecentral NGC 4908 and IC 4051, both of which lie just beyond
mosaic is high and is significantly elevated when com- the eastern edge of the mosaic.
pared to the numbers in the outer fields. Even the cor- These observations suggest the existence of a large in-
nerfieldsofthecentralmosaichavemanymoreGCs. Of tergalacticpopulationofglobularclusters. Inthefollow-
the six outer fields, three in the southwest (lower right) ing sections, we seek to verify and quantify their exis-
havevisiblyelevatedGCnumbersduetotheirproximity tence.
Intergalactic Globular Clusters in the Coma Cluster 7
4. BACKGROUNDESTIMATIONANDGALAXYMASKING ments of GC system radial profiles. The parameters of
this model are estimated based on scaling relations for
Contaminants to our sample of GCs consists of fore-
S and R from Peng et al. (2008, 2011). This model-
ground stars and faint, unresolved background galax- N e
ing is done for allComa galaxiesinthe Eisenhardtetal.
ies (the sum of which we generically refer to as “back-
(2007)catalog,whichiscompletetoM < 16magand
ground”). This background is important to quantify, as V
−
extends to M < 14 mag. This is described in greater
a smooth background can mimic a smooth IGC popula- V
−
detail in Appendix A.2.
tion. Ground-based IGC studies in Coma are typically
Althoughwehavetakengreatpainstomodelandsub-
plaguedbyhighbackgroundduetotheirinabilitytodis-
tract any residual GCs that may belong to Coma galax-
tinguish distant galaxies from point sources.
ies, we find that our final result is largely insensitive to
As a measure of our background, we choose the three
the assumed parameters. The detection of IGCs, as we
outer ACS fields—visits 45, 46, and 59—that are not
will show below, is highly significant and not dependent
near giant galaxies, and are shown at the bottom-center
on the details of the background or the modeling of GC
of Figure 3. For each of these fields, we select GC can-
systems. We estimate the systematic uncertainty due to
didates as described earlier, and also mask the regions
ourmodelingproceduretobe+4000GCs(AppendixA.3),
containing a few obvious Coma members using the pre- −5000
scriptions described below. The surface density of GC only 5–9% of the inferred IGC population.
candidates over these three fields is 2.8 0.3 arcmin−2,
or 28per ACSfield. As we willshow,±this is nearlyan 5. RESULTS
∼
order of magnitude lower than the density of GCs even 5.1. Radial Profile of GCs in the Coma Cluster
in the outer fields of the Coma core.
NGC4874haspreviouslybeenobservedtohavealarge
Toverifythisbackgroundlevel,wecomparedourpoint
number of globular clusters (Blakeslee & Tonry 1995;
source counts in these fields to those in the COSMOS
Harriset al.2000,2009). Ithas also been shownto have
HST Treasury project (Scoville et al. 2007). The COS-
MOS survey imaged 1.8 deg2 at high Galactic latitude a GC systemwhose spatialprofile is shallowerand more
extended than those for other elliptical galaxies (Harris
withthe samecamera(ACS/WFC),filter(F814W),and
etal.2009). CouldtheGCsthatweseefillingthecluster
depthasthe ComaClusterSurvey. Thenumberofpoint
core simply be the extended GC system of NGC 4874?
sources in our three background fields is entirely consis-
The situation is complicated by the fact that the core of
tentwiththe numbersexpectedfromthe surfacedensity
thecluster containsnotonebuttwogiantellipticals, the
of stars in the COSMOS fields. Down to I <25 mag,
814
other being NGC 4889, as well as many other member
wedetect104 10pointsourcesinourthreebackground
± galaxies.
fieldsand98areexpectedusingthe averagesurfaceden-
In Figure 5, we show the radial distribution of GCs
sity from the COSMOS data. This independent check
in the cluster core, centered on NGC 4874. For each
gives us more confidence that our background value is
bin in radius, we sum up the number of observed GC
correct.
candidates in unmasked regions, subtract the expected
The fact that the global background is so low com-
contribution of GCs from other Coma galaxies (shown
pared to the detections in our Coma core fields gives us
as the dotted line in Figure 5), and subtract the global
confidence that we are indeed detecting GCs within the
backgroundlevel(dot-dashline),leavingwhatshouldbe
Coma cluster. The more difficult question is whether
the NGC 4874 and IGC population. We determine the
these objects are truly “intergalactic” or simply part of
meancompletenessofthesamplewithintheannulus,and
extended galactic systems. This debate is not one easily
extrapolate the total number of GCs assuming the M87
resolved by imaging data alone. We can, however, ad-
GCLF as described above. We then sum the total ob-
dress the contribution from galactic GC systems in two
served, unmasked area within the annulus to determine
ways. First, we aggressivelymask regionsaroundknown
the surface number density of GCs. The random errors
bright galaxies. Second, we can make certain assump-
in each bin are derived from the Poisson errors for the
tions about the numbers and spatial extent of the GC
number of candidate GCs as well as from the Poisson
systems of observed cluster members and compare sim-
error in the background, added in quadrature.
ulated GC distributions to the observations. We do this
This profile, tabulated in Table 1, represents our best
in order to test the hypothesis that the GCs observedin
estimate of the radial surface density distribution of the
the cluster core are an intergalactic population.
GC system surrounding NGC 4874, uncontaminated by
The details of these two methods are described in Ap-
the GC systems of other cluster members. Perhaps the
pendix A. Inshort, we generatemasks aroundallgalax-
mostinterestingfeatureofthis profileisamarkedinflec-
ies with luminosities down to M < 17 mag, both in
g
− tionatR 200kpc,beyondwhichtheGC surfacenum-
and around all of our fields. The detection algorithm
∼
berdensity decreasesmuchmoreslowlywithradius. We
that we use for GCs actually ends up masking GCs
interpretthisflatteningoftheprofileastheregionwhere
around fainter galaxies because our chosen background
a large and extended population of IGCs starts to dom-
estimation parameters cannot follow the steeply rising
inate the GCs directly associated with NGC 4874. The
surface brightness profiles at the centers of galaxies. For
significance of this detection is extremely high, as the
each galaxy, we apply a liberal, size-dependent mask to
background level is shown in Figure 5 by the horizontal
the surrounding regions. These masks should eliminate
dot-dashed line at the bottom, with the estimated error
90% of the “galactic” GCs from our catalogs. For the
∼ of the background denoted as the shaded gray region.
remainingouterGCs,wesubtractedamodelGCsystem
TheinferredIGC surfacedensityisafactor 7overthe
usinganassumedS´ersicn=2 profilefor the GC surface
∼
background. Another point of comparison is with the
density,areasonableassumptiongivenpreviousmeasure-
modeled surface density of remaining unmasked galactic
8 Peng et al.
TABLE 1
GC radialsurfacedensity profile
centered on NGC4874
hRi σ Area
1.00 (arcmin) (arcmin−2) (arcmin2)
0.036 3276±2321 0.00546
−2pc 00..005882 41083516±±21303716 00..0001741298
kC 0.115 2175±825 0.02834
NG0.10 0.162 2950±507 0.05617
0.229 2737±326 0.11152
0.322 1978±181 0.22114
0.454 1837±128 0.38152
0.639 1423±91 0.56911
0.900 1164±71 0.73770
0.01 1.268 828±52 0.96299
1.786 641±39 1.45523
2.516 318±15 4.62493
3.543 161±7 12.59246
1 10 100 1000 4.991 97±4 22.04740
R (kpc) 7.030 62±4 16.82656
10.066 44±3 19.62118
Fig. 5.— The radial distribution of GCs in the Coma Cluster 13.674 53±2 40.98172
core centered on NGC 4874. The surface density of GCs in each 17.886 38±3 14.17776
bin(blackpoints)iscalculatedaftermaskingaroundknowngalax-
ies, and statistical subtraction of GCs belonging to these cluster
Note. — These surface densi-
members. Theradialprofileexhibitsaflatinnercoreaswellasan
ties arecorrected for completeness,the
inflection and flattening at large radii. We interpret the flat dis-
full Gaussian GCLF, and include the
tributionat largeradii as evidence of alargepopulation of IGCs.
masking and subtracting of GCs be-
The dot-dashed line and gray band at bottom denote the surface
longing to other cluster members, as
densityofbackgroundobjects(plus1σerrors)determinedfromour
described in Section 5.1. The surface
outer ACS fields and subtracted from all radial bins. The back-
density of contaminants (also correct-
groundlevelisafactor∼7belowthatintheoutermostbins. The
ing for the GCLF) as marked by the
arrow at bottom right shows the mean distance from NGC 4874
dot-dashedlineinFigure5is7.2±1.2.
of the three background fields. The dotted line is the modeled
radial distribution of GCs belonging to cluster members that are
still visible after masking. These have been subtracted from the Although we cannot determine the shape of the IGC
GCradialprofile,althoughtheytooarewellbelowtheoveralllevel component’s density profile, one constraint is that it
by a factor of a few. The data are well fit by a S´ersicmodel plus must fall rapidly after the limits of our data. The mean
aconstantlevel(solidline). TheS´ersiccomponentaloneisshown
distance of the three fields we are using to measure the
asthedashedline.
background is shown as the vertical arrow at 1.3 Mpc.
Therefore, the GC surface density must fall to zero, or
GCs, shown as the dotted line, which also has already
at least the level of the dashed line, by this distance. A
been subtracted from our total GC profile. The overall
steep falloff like this favors a low n S´ersic profile for the
surfacedensityofGCsinthisclusterprofileiswellabove
IGCs(n=1–2),similartotheICLprofilesinSeigaretal.
the surface density of masked galactic GCs (by a factor
(2007) and X-ray gas in galaxy clusters (Demarco et al.
4–7), and thus the GCs we see are likely to be truly in-
2003), but lower (i.e., steeper in the outer regions) than
tergalactic. We have also found that our results do not
dark matter halo density profiles (Merritt et al. 2006).
change significantly if we only use data from the eastern
However, at these radii, it may not make as much sense
or western half of the Coma core.
tospeakofacircularlysymmetricGCradialprofile,and
A single S´ersic profile, normally a good fit to the sur-
it would be more useful to map in two dimensions the
face density profiles of GC systems, is not sufficient to
spatial distribution of GCs.
describe the data for the centralComa Cluster GCs. In-
stead,wefitamodelcombiningaS´ersicprofileandacon-
5.2. Total Numbers of GCs and Specific Frequency
stant. ItislikelythattheIGCshavearadiallydecreasing
density profile (although the simulations of Bekki & Ya- Weuseourradialspatialdensityprofiletoestimatethe
hagi(2006)suggestthattheycanalsohaveaflatdensity totalnumber ofGCs inthe cluster core,whichwe define
distributionwithin the centralfew hundredkiloparsecs), to be the extent of our data. Integrating this profile
but the data only allow us to measure their mean sur- for R < 520 kpc gives a remarkable 70000 1300 GCs,
±
face density. The solid line in Figure 5 traces the best withtheIGCcomponentdominatingtheGCpopulation
fit model, and the dashed line that follows it until large beyond 150 kpc. As listed in Table 2, the number of
radii is the best fit S´ersic component. The fitted surface GCsbelongingtoNGC4874’s“S´ersiccomponent”outto
density of IGCs is 0.055 0.002 kpc−2, which is a 19σ this radius is 23000,leaving a remaining 47000 1600
detection over the backg±round, 0.00845 0.001 kpc−2, (random)+400≈0 (systematic) to be IGCs. There a±re over
± −5000
assuming Poisson random errors. As we discuss in Sec- twice as many IGCs as there are GCs from the S´ersic
tion 4 and Appendix A.3, we also need to account for component, resulting in an IGC fraction of the entire
systematic uncertaintiesfromour modeling,but they do central GC system of 70%.
∼
notaffectthe mainconclusion,whichishighsignificance WithameasurementofNGC4874’sluminosity,wecan
of the detection. We list the best fit parameters in Ta- calculatean“intrinsic”specific frequency for the galaxy.
ble 2. Harris et al. (2009) use a luminosity of M = 23.46
V
−
Intergalactic Globular Clusters in the Coma Cluster 9
TABLE 2
18
Best parametersforS´ersic plusconstant (ΣIGC) model KPNO 4m
fittoComa central GCsystem
10.00 SDSS
TK77
20
Parameter Value Description
n 1.3±0.1 S´ersicindex
Re 62±2kpc S´ersiceffective radius
ΣΣeIGC 00..403575±±00..003042kkppcc−−22 GMCeasnuIrGfaCcesduernfasciteydaetnsRitey −2pc 1.00 222csec)
NNNGGIGCCC,,tSoetrsic 47000±72031006000000±±(r1)730+−00045000000 (s) T“IGSo´etCrasslicwG”iCtGhsiCnwsi5tw2h0iitnhk5ipn2c052k0pckpc N kGC 0.10 24µ (mag/arr
(adjusted to D = 100 Mpc), but surface brightness pro-
26
files from SDSS imaging (J. Lucey, private communica-
tion)andKPNO4-meterCCDimaging(R.Marzke,pri-
0.01
vatecommunication)showthegalaxytobesubstantially
28
brighter. Measurements of the total r-band luminosity
1 10 100
from mosaicked SDSS frames gives a value of r = 10.23 R (kpc)
(M = 24.77,assumingE(B V)=0.009fromSchlegel
r
etal.19−98),whichincludes a0−.32magextrapolationus- Fig. 6.—TheradialdistributionofGCscenteredonNGC4874
ingthebest-fitS´ersicprofileforthelightbeyondR=7′. (blackdots)comparedtothesurfacebrightnessprofileoffieldstar
light around NGC 4874. Three different sources are used for the
The mean color of the galaxy is g r 0.8 mag, which
surface brightness profiles: KPNO (solid), SDSS (dotted), Thuan
− ≈
produces MV = 24.47 using the Lupton (2005) trans- &Kormendy(1977,dashed). Theshadedandstripedregionsrep-
formation. − resent a change in sky determination of ±2% for the KPNO and
SDSSprofiles,respectively. TheGCradialsurfacenumberdensity
With this, we calculate an “intrinsic” specific fre-
profile and the field star surface brightness profile do not exhibit
quency for NGC 4874 of SN = 3.7 0.1 (the errors are similarshapesateithersmallorlargeradii. Thesurfacebrightness
±
purelyfromthetotalnumbersofGCsanddonotinclude measurementsatlargeradii,however,areentirelydependentonan
errors in the luminosity). 2 This value is very much in accuratemeasureoftheskybrightness.
line with the those of non-cD, giant early-type galaxies
in the Virgo and Fornax clusters. brightness profile of the “intracluster background light”
If we assume that all GCs (including IGCs) are part in the Coma cluster as determined photographically by
of the NGC 4874 system and that we are not missing Thuan & Kormendy (1977), transforming from to r
G
anyluminosityfromthe galaxy,thenthis wouldgivethe magnitudes using an offset of ( r)=0.37 mag, based
G−
specificfrequencywithin520kpc ahighervalueofSN = on a (B V)=0.7 mag, their published transformation
−
11.4 0.2,avaluesimilartothosemeasuredforsomecD from to Thuan-Gunn r (Thuan & Gunn 1976), and
± G
galaxies. Another interpretation,whichwe discuss later, thenanoffsettoSDSSr(Fukugita,Shimasaku,Ichikawa
is that the specific frequency of the system is the lower, 1995). This profile appears to match the SDSS photom-
morenormalvalue,butthattheIGCsaretracingalarge etry atR 100kpc but then continues with a shallower
∼
amountofintraclusterstarlightthatisunaccountedfor. slope.
It is clear that at large radii (R > 100 kpc), the de-
5.3. Comparison to NGC 4874 Surface Brightness termination of the sky is crucial to the measurement of
Profile the ICL. To illustrate this, we shade in the regions cor-
A relevant comparison for the GCs is to the surface responding to a change in sky determination of 2% of
±
brightness profile of the field star light of NGC 4874. In theskylevelaroundtheKPNOandSDSSmeasuredpro-
Figure 6, we plot the light profile in circular apertures files. Although there is no evidence for a “break” in the
aroundNGC 4874fromtwoindependent datasets(with measuredsurface brightness profiles akinto what we see
arbitrary normalization). As mentioned in the previous in the GCs, the surface brightness profile in the regions
section, the first is from measurements using SDSS r- where IGCs dominate GC counts is entirely dependent
bandimaging (J. Lucey, in preparation),and the second on the determination of the sky level to better than 1%
uses imaging from the Mosaic-I camera on the KPNO and thus is difficult to quantify.
4-meter telescope (R. Marzke, private communication). We can use these profiles to calculate the “local” spe-
Both profiles are in good agreement in the inner regions cific frequency of the outer GCs, although any calcula-
(R < 20 kpc), but start to diverge in the outer regions tion is highly uncertain due to sky subtraction for the
due to differences in the sky measurements. The differ- surfacephotometry. Nevertheless,wecantakethesesur-
ence between the two profiles is at the level of 2% of the face brightness profiles at face value to see if the cal-
sky. culated values are reasonable. Assuming that the sur-
Intheregionsbeyond100kpc,wealsoplotthesurface face brightness at R = 200 kpc is µr 26.2 mag
≈
(following the Thuan & Kormendy profile), and using
2 If we use the older, fainter value for the luminosity, then V r = 0.2 mag for old metal-poor stellar populations,
−
SN = 9.5±0.3, which is consistent with the value found in the then µV(200 kpc) 26.4 mag. Given the IGC sur-
HST/WFPC2 study of Harris et al. (2009). Their data only ex- face density at thes≈e radii (46 arcmin−2), we estimate
tended to R∼65 kpc, and thus were not able to detect the IGC
S (200 kpc) =5. If we assume the profile derived from
population. Thehighervalueisalsoconsistentwiththevalueesti- N
matedbyBlakesleeetal.(1997),whoalsousedafainterluminosity. SDSS data, however, then µV(200 kpc) 27.5 mag, re-
≈
10 Peng et al.
sulting in S (200 kpc) = 13. We emphasize that the
N
surfacephotometryisextremelyuncertainattheseradii,
NGC 4874/Coma
and the upper error bar on this number is essentially M87/Virgo (McL99)
unconstrained. The local values of S at these kinds 10.00 M87/Virgo (T06)
N M87/Virgo (ACSVCS)
of radii have previously been reported to be quite high
(Tamura et al. 2006 in M87 and Rhode & Zepf 2001 for
NGC 4486), but those measurements are equally uncer-
tain for similar reasons.
In the inner regions, there is a notable divergence be- −2c 1.00
p
tween the GCs and the galaxy light. The galaxy does kC
G
not show the prominent core within 10 kpc that the GC N
systemdoes,onlydisplayingaflatteningintheprofileat
a smaller radius. 0.10
5.4. Comparison to the M87 GC System
Perhaps the most relevant local comparison for the
NGC 4874/Coma Cluster GC system is that of M87 in
0.01
the Virgo Cluster. In Figure 7 we show the GC radial
1 10 100
surface density profiles of the two GC systems. The R (kpc)
outer M87 profile is taken from the data of McLaugh-
lin (1999) and Tamura et al. (2006), while the central Fig. 7.—TheradialdistributionofGCscenteredonNGC4874
(blackdots)comparedtothedistributionofGCsaroundtheVirgo
regions of the profile are from the ACS Virgo Cluster
cD galaxy M87. M87 data is from McLaughlin (1999, McL99,
Survey(ACSVCS) datashowninPengetal.(2008). We orange diamonds), Tamura et al. (2006, T06, red asterisks), and
notethatthephysicalresolutionoftheComaHSTdatais the ACS Virgo Cluster Survey (Peng et al. 2008; blue triangles)
very competitive with ground-based Virgo observations withaS´ersicfittothecombineddatasetoverplotted(dot-dashed).
(0′.′1resolutionatComadistanceisequivalentto0′.′6res- The Coma GCs have a much shallower and larger core, as well
as an inflection where IGCs begin to dominate. Even with the
olution at Virgo distance), but obviously cannot match largerT06dataset,thereisnotyetevidenceforaprofileinflection
the ACSVCS observations of M87. None of the Virgo around M87 like what we see in Coma, although the data do not
data sets go as far out in physical radius as our Coma gocomparablyfaroutinradii.
data,buttheystillprovideausefulcomparison. Thetwo however,onlyreachesaradiusof130kpc,andthuswould
GCsystemsprofilesaresimilarintherangeofintermedi- not be sensitive to the kind of IGC population we see in
ateradii(20–100kpc),butdifferencesappearinthevery Coma. In fact, if the Coma data had the same physical
inner and outer regions. Most noticeably, the Coma GC radial extent, it would have been very difficult to de-
systems displays a very pronounced core within 10 kpc, tect the IGC population. More deep, wide-field imaging
which does not appear to be present in the Virgo sys- of the area around M87, such as the Next Generation
tem except perhaps within 1 kpc. This deficit of GCs Virgo Survey3, will be necessary to detect or place more
at the center of NGC 4874 is also evident in the analy- stringent limits on a population of IGCs in Virgo.
sis of Harris et al. (2009). This core could be the result
of dynamical friction destroying GCs at the center of 5.5. GC Color Distributions
NGC 4874.
For old GCs, the broadband color is an indicator of
ThecoreintheGCprofile,andthedivergencefromthe
metallicity. The color distributions of extragalactic GC
M87 GC profile, is most evident within 10 kpc. Could
systems have been studied extensively with HST (e.g.,
this be due to unaccounted observational incomplete-
Larsen et al. 2001; Peng et al. 2006), and are often bi-
ness? Thefourinnermostradialbins(R<3.5kpc),have
modal in nature (Gebhardt & Kissler-Patig 1999), espe-
the largesterrorsandshallowestobservationsbecause of
ciallyinmassiveearly-typegalaxies. InFigure8,weplot
the bright galaxy light (completeness of the GCLF is
the color distribution of bright GCs (applying a magni-
25% in these bins). There are multiple reasons, how-
≈ tudelimitofg <25magforhigherS/N)intheunmasked
ever, why we believe these lower surface densities to be
regionsofthe ComaClustercore. The colordistribution
real. The radius at which the core becomes apparent,
of all GCs shows the typical bimodality seen in extra-
20′′,is largeforHST imaging. Evenexcludingtheinner
galacticGCsystems,displayingaprominentpeakofblue
10′′(4.8kpc),thedifferenceinslopebetweenthetwoGC
(metal-poor) GCs with (g I) 0.9, and a red (metal-
profilesis stillapparent. Also,the completenesstests we − ≈
rich) peak with (g I) 1.15.
apply also take into account incompleteness due to im- − ≈
We plot the color distributions divided by distance
perfect profilesubtraction, andthus is a true measureof
from the center of NGC 4874—those within 50 kpc
the completeness in these radialannuli. In orderto turn
(galactic GCs) and those outside of 130 kpc (predomi-
the M87 GC profile into the Coma GC profile through a
nantlyIGCs). WeusetheKaye’sMixtureModel(KMM;
systematic overestimationof the completeness, the com-
McLachlan&Basford1988;Ashman,Bird,&Zepf1994)
pleteness would have to be overestimated by nearly an
implementation of the expectation-maximization (EM)
orderofmagnitude. Lastly,thisdeficitofGCsrelativeto
method to fit twoGaussians with the same standardde-
the galaxy light profile was also independently found by
viation to the GC color distributions of each sample.
Harris et al. (2009) using HST/WFPC2 data (see their
Both the galactic and intergalactic GCs are much bet-
Figure 6).
In the outer regionsthe M87 profile follows the single- 3TheNextGenerationVirgoSurvey(NGVS)isaLargeProgram
S´ersic fit out to the limits of the data. The Virgo data, withtheCanada-France-HawaiiTelescope.