The Snapshot Hubble U-Band Cluster Survey (SHUCS). I. Survey Description and First Application to the Mixed Star Cluster Population of NGC 4041
I. S. Konstantopoulos, L. J. Smith, A. Adamo, E. Silva-Villa, J. S. Gallagher, N. Bastian, J. Ryon, M. S. Westmoquette, E. Zackrisson, S. S. Larsen, D. R. Weisz, J. C. Charlton
aa r X i v : . [ a s t r o - ph . C O ] F e b The Snapshot
Hubble
U-Band Cluster Survey (SHUCS). I.Survey Description and First Application to the Mixed StarCluster Population of NGC 4041 ∗ . I. S. Konstantopoulos , , L. J. Smith , A. Adamo , E. Silva-Villa , J. S. Gallagher ,N. Bastian , , J. E. Ryon , M. S. Westmoquette , E. Zackrisson , S. S. Larsen ,D. R. Weisz , J. C. Charlton ABSTRACT
We present the Snapshot
Hubble
U-band Cluster Survey (SHUCS), a project aimed at char-acterizing the star cluster populations of ten nearby galaxies ( d <
23 Mpc, half within ≈
12 Mpc)through new F336W (U band equivalent) imaging from Wide Field Camera 3, and archival
BVI -equivalent data with the
Hubble Space Telescope . Completing the
UBVI baseline reduces theage-extinction degeneracy of optical colours, thus enabling the measurement of reliable ages andmasses for the thousands of clusters covered by our survey. The sample consists chiefly of face-onspiral galaxies at low inclination, in various degrees of isolation (isolated, in group, merging), andincludes two AGN hosts. This first paper outlines the survey itself, the observational datasets,the analysis methods, and presents a proof-of-concept study of the large-scale properties and starcluster population of NGC 4041, a massive SAbc galaxy at a distance of ≈
23 Mpc, and partof a small grouping of six giant members. We resolve two structural components with distinctstellar populations, a morphology more akin to merging and interacting systems. We also findstrong evidence of a truncated, Schechter-type mass function, and a similarly segmented lumi-nosity function. These results indicate that binning must erase much of the substructure presentin the mass and luminosity functions, and might account for the conflicting reports on the in-trinsic shape of these functions in the literature. We also note a tidal feature in the outskirtsof the galaxy in
GALEX
UV imaging, and follow it up with a comprehensive multi-wavelengthstudy of NGC 4041 and its parent group. We deduce a minor merger as a likely cause of itssegmented structure and the observed pattern of a radially decreasing star formation rate. Wepropose that combining the study of star cluster populations with broad-band metrics is not onlyadvantageous, but often easily achievable thorough archival datasets.
Subject headings: surveys: SHUCS — galaxies: individual (NGC 4041) — galaxies: star clusters: general— galaxies: interactions — galaxies: star formation — galaxies: groups: individual: LGG 266 Australian Astronomical Observatory, PO Box 915,North Ryde NSW 1670, Australia; [email protected]. Department of Astronomy & Astrophysics, The Penn-sylvania State University, University Park, PA 16802, USA. Space Telescope Science Institute and European SpaceAgency, 3700 San Martin Drive, Baltimore, MD 21218,USA. Max-Planck-Institut for Astronomy, K¨onigstuhl 17, D-69117 Heidelberg, Germany. D´epartement de Physique, de G´enie Physique etd’Optique, and Centre de Recherche en Astrophysique du Qu´ebec (CRAQ), Universit´e Laval, Qu´ebec, Canada. Department of Astronomy, University of Wisconsin-Madison, 5534 Sterling, 475 North Charter Street, MadisonWI 53706, USA. Excellence Cluster Universe, Boltzmann-Strasse 2,85748 Garching bei M¨unchen, Germany. Astrophysics Research Institute, Liverpool JohnMoores University, Egerton Wharf, Birkenhead, CH411LD, UK European Southern Observatory, Karl-Schwarzschild-Strasse 2, 85748 Garching bei M¨unchen, Germany. Department of Astronomy, Stockholm University, Os- . Introduction The launch of the
Hubble Space Telescope ( HST ) over two decades ago started a revolu-tion in the study of extragalactic star clusters.The discovery of large numbers of young com-pact clusters in star-forming galaxies led to thesuggestion that they could be the present-dayanalogues of globular clusters (see reviews ofWhitmore 2003; Larsen 2004a). The questionof whether these young clusters can survive for aHubble time has still not been settled but theirlongevity appears to be critically dependent on en-vironmental conditions within their host galaxies(de Grijs & Parmentier 2007; Bastian et al. 2011).The installation of Wide Field Camera 3 (WFC3)on
HST has vastly upgraded the imaging capa-bilities of the telescope shortward of 4000 ˚A. It isnow much easier to measure the age and mass dis-tributions of large populations of star clusters ingalaxies, and address fundamental questions suchas their long term survival chances. To obtainages and extinctions for clusters younger than ∼ UBVI baseline (Anders et al. 2004a).Prior to WFC3,
HST
U band imaging of suf-ficient depth and spatial coverage was feasiblefor only a few regions of nearby galaxies (e. g.Smith et al. 2007; Anders et al. 2004b), and onlydistant systems in their entirety (e. g. ¨Ostlin et al.2003; Adamo et al. 2010a).All local, late-type giant galaxies host popu-lations of young and intermediate-age star clus-ters, often with masses and densities that ri-val globular clusters. It has been proposed thatthe vast majority of stars are formed in clustersbut that most of these clusters ( ∼ car Klein Centre, AlbaNova, Stockholm SE-106 91, Sweden. Department of Astrophysics/IMAPP, Radboud Uni-versity Nijmegen, P.O. Box 9010, 6500 GL Nijmegen, TheNetherlands. Department of Astronomy, Box 351580, University ofWashington, Seattle, WA 98195, USA. ∗ Based on observations made with the NASA/ESAHubble Space Telescope, obtained at the Space TelescopeScience Institute, which is operated by the Association ofUniversities for Research in Astronomy, Inc., under NASAcontract NAS 5-26555. These observations are associatedwith program ters that survive this period of “infant mortal-ity” can disrupt through stellar evolution, two-body relaxation and the tidal field of the hostgalaxy on ∼ Gyr timescales (Bastian & Gieles2008). There has been much debate in theliterature over whether infant mortality exists,whether the later phases are mass-dependentor not (e. g. Lamers et al. 2005; Fall et al. 2005;Whitmore et al. 2007; Chandar et al. 2010; Bastian et al.2011), and even if most stars do indeed form inclusters (Bressert et al. 2010). By conducting asurvey of local star-forming galaxies, it will bepossible to obtain large samples of clusters cov-ering a wide range of masses and ages, and thushelp answer several open questions.In addition, the local environment has recentlybeen suggested as a major contributor to clus-ter disruption (e. g. Elmegreen & Hunter 2010;Kruijssen et al. 2011). The relationship betweenthe formation of clusters and the properties of thehost galaxy is, in fact, far from clear on both globaland local scales, and a survey would help to inves-tigate these relationships. So far, detailed studiesof large numbers of clusters in small samples ofgalaxies (e. g. Meurer et al. 1995; Goddard et al.2010; Adamo et al. 2011; Silva-Villa & Larsen2011) or small numbers of clusters in large sam-ples of galaxies (Larsen 2004b; Bastian 2008;Mullan et al. 2011) have suggested that environ-ments with a higher star-forming density form ahigher fraction of their stars in clusters. This con-nection is also relevant to the link between galaxyinteractions and increased star formation. Sincestar clusters can trace bursts of star formation,they can then provide a viable chronometer forthe interaction history of a galaxy.Naturally, understanding the physics that gov-ern these clusters is essential to utilizing themas tracers of star formation. After two decadesof
HST -driven research, several cluster parame-ters are considered as standard, such as their dis-tributions of luminosity and size. Perhaps mostnotably, many studies have investigated the starcluster mass distribution, with a common findingthat it can be represented by a power law of indexnear − HST imaging (
UBVI and H α ).The imprints of various detectors are visible here, arising from uneven spatial coverage across the opticalbaseline. The new F336W imaging envelops the archival WFPC2 pointings, while the ACS image leaves atrace of its chip gap. The face of NGC 4041 exhibits a composite structure. Pink H α bubbles trace thedistribution of young clusters along spiral arms that wind tighter in the inner regions than the outer galaxy.A bright central component seems to define this two-step structure, while the central peak is offset from thegeometrical center, the center of the outermost isophote, by ≈ ′′ .flects the mass function of Giant Molecular Clouds(e. g. Solomon et al. 1987), out of which star clus-ters form. The slope also appears to be a func-tion of brightness, that is to say, the brighter thesubsample, the steeper the extracted slope (Gieles2010). In addition, preliminary indications sug-gest that the characteristic ‘Schechter mass’, M ∗ ,where the truncation occurs, depends on envi-ronment (Larsen 2009; Gieles 2009; Bastian et al.2012). Again, a large survey of clusters in a varietyof host galaxies will be able to settle this issue. Another parameter of interest is the star clustersize distribution and whether this is related to en-vironment and/or age. Observations indicate thateffective radii are constrained to a range of 0.5 –10 pc despite the large dynamic range in clustermass (Portegies Zwart et al. 2010). The observedradius distribution is well described by a log-normal distribution with a peak at 3–4 pc for bothyoung clusters (Barmby et al. 2006; Bastian et al.2011) and old globular clusters (e. g. Jord´an et al.2005), with the exception of the “faint-fuzzy”3lusters discovered by Larsen & Brodie (2000).There are indications that the cluster core ra-dius increases with age (e. g. Mackey & Gilmore2003a,b; Scheepmaker et al. 2007; Bastian et al.2008; Hurley & Mackey 2010). A survey of clustersizes in different environments within galaxies willpermit studies of the cluster size distribution as afunction of age and environment.From the above, it is clear that an extensivesurvey of a large sample of star clusters in a va-riety of environments, aimed at measuring theirage, mass, and size distributions, will addressmany fundamental questions regarding their prop-erties, survival rates, formation histories, and re-lated environmental dependencies. In this paper,we describe such an endeavor: the Snapshot Hub-ble U-band cluster survey (SHUCS) combines newWFC3 F336W (U band-equivalent) imaging witharchival HST BVI -equivalent imaging, to mea-sure the properties of large samples of young clus-ters in nearby (mostly d .
12 Mpc) star-forminggalaxies. We present a full description of the sur-vey in terms of sample definition and data re-duction (Section 2); the detection and photom-etry of star clusters, and the derivation of age,mass, and extinction (Section 3). In the secondpart of this paper, we present a proof-of-conceptstudy for NGC 4041, a bright ( g = 12 .
2. The Survey: Target Selection and DataReduction
To define our galaxy sample for WFC3-F336Wimaging, we selected galaxies within 25 Mpc thathave
BVI images available in the
HST archive.We restrict the sample to any galaxy imaged withthe Advanced Camera for Surveys (ACS) or theWide Field and Planetary Camera 2 (WFPC2),but imposed no constraints on image depth. Wechose this distance limit and instrument set to en-sure that individual star clusters are resolved andthat deep WFC3 F336W imaging can be obtainedwith a 30 min exposure time. The depth is quan-tified as reaching m F336W ≈
26 mag at acceptable error levels, as will be demonstrated in Section 3.1.We excluded dwarf irregular galaxies as they typi-cally host very few clusters (e. g. Seth et al. 2004),but do not impose a strict lower mass limit on thedataset.The resulting sample of 22 galaxies was drawnfrom the combined target lists of the 11 Mpc H α Ultraviolet Galaxy Survey (11HUGS; Kennicutt et al.2008); the Local Volume Legacy survey (LVL;Dale et al. 2009); the ACS Nearby Galaxy SurveyTreasury (ANGST; Dalcanton et al. 2009); the
HST H α Snapshot survey (HHsnap; Ho 2003);and the Larsen (2004b) catalog. The last catalogon that list contained size measurements, photom-etry and information on morphology for clustersin galaxies with WFPC2 imaging in various filters(typically B, V and I equivalents) and ground-based UBVRI imaging. Like the present work,the Larsen study was also aimed at studying starclusters and their immediate environment.Our Cycle 18 Snapshot program “
HST
U bandSurvey of Star Clusters in Nearby Star-FormingGalaxies” (PI Smith, ID 12229) was awarded 22snapshot orbits, one for each of 22 proposed tar-gets. Eleven galaxies were observed, as listed inTable 1, as per the nominal completion rate of50% for SNAP programs (one observation failed).About half of observed galaxies are within ≈
12 Mpc, and four systems lie at 18 −
23 Mpc. Allexposures were taken with the F336W filter andthe duration was 1800 s. A three-point dither linepattern was chosen to cover the chip gap and toaid in the removal of hot pixels, cosmic rays andother artifacts. Either the UVIS-FIX or UVIS2-FIX aperture was used depending on the spatialextent of the galaxy compared to the 162 ′′ × ′′ UVIS field of view. The precise pointings werechosen to give maximum overlap with the archival
BVI observations.Throughout this series we will use the Johnsonfilter notation with the specific
HST filter sub-scripted, but will not at any point convert betweenthe two systems. For example, F336W will bedenoted as U , while F555W and F606W, bothroughly corresponding to Johnson V band, will bewritten as V and V . The U through I base will be referred to as UBVI .4 able 1WFC3/UVIS Target List. Name RA DEC Morphology Distance log( M ∗ ) Ref. Other HST data(h m s) ( ◦ ′ ′′ ) (Mpc) (M ⊙ )NGC 247 00 47 10.49 −
20 46 09.00 SAB(s)d 3.6 9.06 [1] F110W, F160W (N)NGC 672 01 47 54.06 +27 25 55.80 SB(s)cd 8.1 9.43 [2] F658N (A)NGC 891 02 22 32.90 +42 20 45.80 SA(s)b? 10.2 10.84 [1] F250W (A), F656N (W2), F160W (N)NGC 925 02 27 05.14 +33 34 54.50 SAB(s)d 9.3 9.65 [3] F160W (N)NGC 1003 02 39 16.40 +40 52 20.40 SA(s)cd 11.1 9.44 [3] − IC 356 04 07 46.47 +69 48 45.20 SA(s)ab pec 11.2 11.32 [2] F658N (A)NGC 2146 06 18 37.71 +78 21 25.30 SB(s)ab pec 17.2 11.04 [4] F658N (A), F160W (N)NGC 2997 09 45 38.70 −
31 11 25.00 SAB(rs)c 12.2 10.61 [2] F220W, F330W (A)NGC 3756 11 36 47.97 +54 17 37.25 SAB(rs)bc 21.7 10.38 [5] F658N (A)NGC 4041 12 02 12.17 +62 08 14.20 SA(rs)bc? 22.7 10.55 [2] F658N (A)NGC 6217 16 32 39.22 +78 11 53.60 (R)SB(rs)bc 18.3 10.46 [2] F658N (A)
Note.—
Coordinates (J2000) correspond to the UVIS aperture positions. Morphologies are taken from de Vaucouleurs et al. (1991,in the
NED homogenized notation), while distances are drawn from: [1] Willick et al. (1997); [2] Tully (1988); [3] Tully et al. (2009);[4] Kennicutt et al. (2011); [5] Springob et al. (2009). Stellar masses are derived through the ‘best’ K S photometry from 2MASS(Skrutskie et al. 2006). The final column (‘Other HST data’) lists non-proprietary imaging datasets from ACS (A), WFPC2 (W2),and NICMOS (N) in the MAST archive using filters other than
BVI -equivalents. NGC 6217 was scheduled and observed withUVIS/F336W, but the observation failed.
Our chosen proof-of-concept object was NGC 4041,a face-on SAbc galaxy near the upper end of ourdistance limit at 25 Mpc. This choice allows fora demonstration of the methodology of sourceselection and processing for approximately halfthe SHUCS galaxies, those at distances beyond ≈
15 Mpc. The processing and analysis of nearbysystems will be outlined in a future paper. Inaddition, it offers an interesting study of envi-ronmental effects on star cluster formation andevolution in a structurally segmented system.We discuss the structure and environment ofthis galaxy in Section 4 and show a compos-ite
HST image in Figure 1. The
HST observa-tions of NGC 4041 are shown in Table 2. Thearchival B , V , I , and H α (F658N) datawere obtained by two separate programs (9042and 9788). All data were retrieved from the Mikulski Archive for Space Telescopes (MAST)and each dataset was combined, corrected forgeometric distortion, and drizzled to the nativepixel scale using multidrizzle (Koekemoer 2005;Fruchter & Sosey 2009). For each WFPC2 image,two undithered exposures were obtained. Thesewere drizzled separately for each CCD and the fi-nal pixel scale was 0 . ′′
05 (PC) and 0 . ′′
10 (WF2, 3,4). The ACS F658N data consisted of a pair of images and these were drizzled to a pixel scaleof 0 . ′′
05. Only a single F814W ACS image wastaken; cosmic rays were removed using the la-cosmic task (van Dokkum 2001) and the resul-tant image drizzled. For the WFC3/UVIS data,three dithered images were obtained and thesewere drizzled to a pixel scale of 0 . ′′
3. Selection of Star Cluster Candidatesand Data Products
Star clusters are only marginally resolved at thedistance to NGC 4041, as the typical diameter of5 able 2Archival and new observations of NGC 4041
Instrument Date Filter Exposure Program ID P.I. Name(s)WFC3/UVIS 2011 Jan 30 F336W 1800 12229 SmithWFPC2 2001 Jul 04 F450W 320 9042 SmarttWFPC2 2001 Jul 04 F606W 320 9042 SmarttWFPC2 2001 Jul 04 F814W 320 9042 SmarttACS/WFC 2004 May 29 F814W 120 9788 HoACS/WFC 2004 May 29 F658N 700 9788 Ho7 pc subtends an angle of 0 . ′′ HST cameras. A considerable drawback is hence thepotential inclusion of stellar associations, whichare not discriminated by automatic source de-tection algorithms (see Silva-Villa & Larsen 2010;Bastian et al. 2011). Toward that end we employthe concentration index (CI), defined as the dif-ference in brightness between two apertures: onecomparable to the size of the point-spread func-tion (PSF), and another representing the typicalstar cluster (see Whitmore et al. 2007). This tech-nique helps to place clusters between stars, whichhave very small CI, and associations, at large CI.We will develop the use of this method in and Sec-tion 3.2.With the above in mind, we developed a selec-tion and photometry pipeline, which is run entirelyin IRAF † and consists of the following steps:1. Source selection in U . We run DAOfind to select sources brighter than 7 σ abovebackground, measured on various parts ofthe image. We do not restrict ‘roundness’(axial ratio) or ‘sharpness’ (size comparedto the stellar full width at half-maximum),in order to include elliptical and marginallyresolved clusters (the majority at this dis-tance, as noted above).2. Coordinate transformation . Our datasetconsists of images taken for different projects † IRAF is distributed by the National Optical Astron-omy Observatories, which are operated by the Associationof Universities for Research in Astronomy, Inc., under co-operative agreement with the National Science Foundation. at various times and with different cameras,which often leads to overlapping, but notidentical pointings. Our U observationswere designed for maximum overlap with thearchival imaging, and hence serve as the ref-erence frame for the World Coordinate Sys-tem (WCS). The coordinate lists from Step 1are converted from the U frame to those ofeach instrument/pointing used in the study.This step uses the tmatch algorithm withtypically some 20 reference sources (stars orcompact clusters) that span the entire imageas much as possible. The process is refineduntil the root-mean-squared errors for x andy positions do not exceed 0.1 px, a scaletested to eliminate source confusion. Whenregistering WFPC2 data, the four CCDs aremapped individually, and the number of ref-erence stars is usually ten or lower.3. Multi-band photometry . Photometricapertures of radius 0 . ′′
12 are placed at thecoordinates defined in Step 2, with the back-ground measured locally in annuli of 0 . ′′ . ′′
3, depending on the pixel scale of eachcamera and the corresponding PSF. Thesevalues were derived after testing the typi-cal growth curves of modeled stellar PSFsgenerated with
Tinytim (Krist et al. 2011),coupled with the
STScI Focus Model Util-ity ‡ to account for breathing at the timeof observation. We note that these val-ues will be revised for each galaxy depend-ing on its distance. We do not allow the DAOphot task to re-center sources as the ‡ . ′′
040 px − ), ACS/WFC(0 . ′′
050 px − ), WFPC2/PC (0 . ′′
045 px − ),WFPC2/WF (0 . ′′
100 px − ), the radius of theapplied photometric aperture corresponds toimage sizes of 3.0, 2.4, 2.7, and 1.2 px respec-tively. In the test case of NGC 4041, thesetranslate to a physical radius of 13.4 pc, cf. the typical effective cluster radius of ∼ . DAOphot recipe thattakes into account the gain and readout noiseof each detector. For cases where imagingfrom more than one
HST camera is availablein a given filter, we photometer all imagesand choose the measurement with the lowestassociated error. We calibrate the photome-try in the Vegamag system.The photometric process is completed by themeasurement of a CI for each source as thedifference in U brightness between aper-tures of one and three pixels. Such a tech-nique has been demonstrated in the past toproduce viable samples of star clusters atcomparable distances (e. g. The Antennæ,Whitmore et al. 2007). More informationcan be found in Section 3.2.4. Photometric corrections . At this stagewe correct for the size of the aperture.This has to be tailored to each galaxy in-dividually, as the correction is dependenton distance and whether the sources areresolved or not. In addition, we do notalways have enough bright, isolated clus-ters to derive growth curves. The correc-tions thus combine our own empirical de-ductions with the encircled energy curvesgiven for each instrument. For the test caseof NGC 4041, isolated sources were veryrare, so corrections were derived throughthe growth curves of model PSFs, gener-ated for each detector-filter combinationwith
Tinytim . Finally, we add to ourphotometric errors a factor of 0.05 mag inquadrature, to account for uncertaintiesin the photometric zeropoints (for detailssee Adamo et al. 2010a), and correct forforeground extinction (Schlegel et al. 1998).
At the distance to NGC 4041, the bulk of itscluster population is nearly point-like. There-fore, we follow the traditional method of artifi-cial star counts to estimate completeness. In fu-ture works focussed on nearby systems, however,we will use artificial clusters instead, to overcomethe uncertainties introduced by stochastic vari-ations of the colors of low-mass clusters (intro-duced by under-sampled stellar mass functions;see Fouesneau & Lan¸con 2010; Popescu & Hanson2010; Silva-Villa & Larsen 2011).Fig. 2.— Completeness tests tracing the fractionof artificial stars recovered at a range of bright-ness. The dotted vertical lines indicate 50% and90% completeness fractions for the inner disk andouter galaxy, performed separately to account forthe different background levels. The top axis dis-plays the range of absolute magnitudes covered.We estimate 50% and 90% completeness levels at(25.5, 23.6) mag and (26.3, 24.6) mag for the innerdisk and outer galaxy respectively.We created sets of artificial stars in a 15 ×
15 ob-ject grid, using the
MKSynth algorithm of Larsen(1999). We used
TinyTim models of the WFC3PSF as described in Step (3) of the pipeline (Sec-tion 3). The generated objects were assigned mag-nitudes in the range [20, 30] mag, in steps of onemagnitude. Since completeness varies with the lo-cal background (e. g. Scheepmaker et al. 2007), as7ell as crowding and confusion, we use two fieldsfor this process: one covering most of the innerdisk, and another, equally-sized field that coversthe outer spiral structure. Results are presented inFig. 2, where we show the 50% and 90% recoveryfractions to occur at (25.5, 23.6) mag and (26.3,24.6) mag for the inner disk and outer galaxy re-spectively.Fig. 3.— Brightness-error plots across the
UBVI filter-set. The separation into inner and outergalaxy (blue dots, red triangles) accentuates theissues faced with photometric errors in the innerdisk, an effect of crowding. This is taken into ac-count in all analysis. Using this plot we tentativelyestablish V as the limiting band for this study,as it shows the most irregular brightness-error dis-tribution. Regular statistical measures, such asthe standard deviation of the error distribution,provide no distinction between the BVI filters.Another photometric property that will impactour star cluster analysis is the limiting filter. Weestimate this through the brightness-error plot ofFigure 3. Regular statistical measures, such asthe standard deviation of the error distribution,provide no distinction between the
BVI filters. Wetherefore look for the most irregular distributionand adopt V as the limiting band. In Step 3 of the photometry pipeline (Section 3)we note the measurement of the CI, as the dif-ference in magnitude between apertures of ra-dius 1 and 3 px. We employ this metric in or-der to distinguish between star clusters and stel-lar associations, large, short-lived structures (seePortegies Zwart et al. 2010; Bastian et al. 2011,2012) that can contaminate samples and affect theanalysis of star cluster populations.In order to employ a CI cut in our source fil-tering process, we test the theoretical expectationfrom modeled clusters of various ages and sizes.First we use
MKSynth (see Section 3.1) to convolvean Elson et al. (1987) surface brightness profilewith a PSF derived from point sources in our im-ages. Since we could not identify suitable starson the F336W image of NGC 4041, we employedSHUCS images of NGC 891 and NGC 2146. In ad-dition, while various images are taken at differentfocus settings, the PSF does not change to a levelthat will affect our analysis – the effect is smallerthan our typical photometric error of 0.1 mag.We assign a range of sizes to these model clusters,incorporate them on the F336W image and mea-sure their CI in the same way as for the detectedsources. The top panel of Figure 4 shows a plot ofCI versus assigned radius. The first datapoint, at R eff = 0 pc, denotes individual stars, which reg-ister values of ≈ . . c = r tidal r core = 30,while the age, mass and effective (half-light) radiusassume the following ranges:log( M/ M ⊙ ) = [3 . , . , . , . , . τ /yr ) = [6 . , . , . , . R eff = [1 , , , , , , , , ,
20] pc8he range of CI assigned to clusters is shown asa hashed area in the top panel of Figure 4, whilein the center and bottom panels, the CI area isbracketed by dotted lines to aid illustration. TheCIs of stochastically modeled clusters are slightlyhigher (by ≈ . U brightness in Figure 5. Ap-plying the above CI cut and the 90% completenesslimits of Section 3.1, we restrict cluster candidatesto a very specific part of photometric parameterspace. Yggdrasil models:derivation of age, mass and extinction
The age, mass, and extinction of all candi-date star clusters are derived through fits to theirspectral energy distribution (SED) with
Yggdrasil models (see Zackrisson et al. 2011, for a detaileddescription). Such a model set is ideal for ourwork, as it combines single stellar population(SSP) synthesis models with nebular spectra, bothemission lines and continuum, and is tuned tothe redshift of each galaxy. Analyses using
Yg-gdrasil models have in the past been successfulin describing the cluster populations of intenselystar forming galaxies (e. g. Adamo et al. 2010b).For the SHUCS sample, we use Starburst99SSPs (Leitherer et al. 1999; V´azquez & Leitherer2005), compiled using a Kroupa (2001) stellarinitial mass function throughout the mass range0 . −
100 M ⊙ ; Padova stellar evolutionary tracks;and three different metallicities: 0.4, 1, and 2.5times solar. We note no spatial trends in terms ofmetallicity, as shown in Figure 6. While this resultmight seem to oppose the well-established metal-licity gradients in local spirals (e. g. in the LocalGroup; Cioni 2009), the photometric derivation ofmetallicity is subject to large uncertainties. Nofirm conclusions can be drawn about the metal-licity distribution. To reproduce the integratedfluxes of the earlier stages of cluster evolution, i. e. luminous H ii regions surrounding the recentlyformed stars, the models include a self-consistenttreatment of the nebular emission performed withthe Cloudy photoionization code (Ferland et al.1998). The metallicities of the gas and the starsare assumed to be the same. Using the gas proper- Fig. 4.— Concentration Index tests on artificialclusters. The top panel shows a CI versus radiusplot for clusters created as Elson et al. (1987) lightprofiles of index γ = 3 .
0, convolved with an empir-ical PSF. The center and bottom panels show ar-tificial clusters created as aggregates of individualstars, arranged according to a King profile (moreappropriate for older clusters). Cluster ages arelog( τ / yr) = 7 ,
8, while different symbols representa range of masses. As they age, clusters displayless regular behavior, as evidenced by the errorbars in the log( τ / yr) = 8 model set. A consistentasymptotic behavior is noted through both meth-ods. R eff &
10 pc are rarely observed, so we set theCI at this radius as the upper limit. An R eff = 0represents individual stars, and the correspondingCI ≃ . bottom panelshows all sources that register UBVI photometryas green points, while the open and closed graycircles apply a cut according to photometric error.The top panel shows the space chosen to representhigh-quality cluster candidates, the ones used forall further analysis, against a backdrop of the graycircles of the lower panel. Vertical lines denote the90% completeness limits for the inner and outergalaxy.ties of typical local H ii regions (Kewley & Dopita2002) we assume a filling factor f = 0 .
01, a hydro-gen density of n (H) = 100 cm − , and a coveringfactor, i. e. fraction of ionizing photons absorbedby the gas, of c = 0 .
5. Different values of c , n (H), and f only impact the SED fits for the veryyoungest clusters. We therefore never attempt toresolve the 1 − χ fittingalgorithm, as detailed in Adamo et al. (2010a). Inbrief, the program considers sources detected infour or more filters and applies the Cardelli et al.(1989) attenuation law at each age step to con-strain the visual extinction in the line of sight.The final age, mass, and extinction are assignedto a cluster given the set of parameters that min-imizes the χ . Uncertainties are carried throughthe fitting process. The program estimates thebest reduced χ ν,best ( χ divided by the degrees of freedom), and saves, as the range of acceptable val-ues, all the solutions with χ ≤ χ ν,best + 3 .
5. Themaximum and minimum age, mass, and extinc-tion contained in this pool of likely solutions areassigned as errors. This is a typical method of es-timating errors when the χ statistic is employed,e. g. the photometric fits of Bik et al. (2003) andthe spectroscopic fits of Konstantopoulos et al.(2009). Lampton et al. (1976) provide a full rea-soning.Example SED fits are shown in Figure 7. Firstwe provide an illustration of the best fitting Yg-gdrasil model, with source photometry marked asfilled dots. The red squares mark the best-fitmodel flux on the y axis, along with the centralwavelength and bandpass of each employed
HST filter. The plots of the bottom row plot χ inthe age-extinction and age-mass spaces. Red con-tours mark the range of acceptable solutions tothe SED fit, defined as χ ν < χ ν,best + . ν . The toprow of figure sets features two cluster candidatesthat meet the selection criteria (detailed in nextsection), while the ones in the bottom row do not. The primary data products of each study area Photometry Table and an SED Table. Thesedata products will be made available to the com-munity in their entirety when the survey has beencompleted. The Photometry Table collects posi-tional and photometric measurements and appliesa photometric flag, f UBVI , according to the follow-ing scheme:1: Source detected and photometered in
UBVI ,11: . . . plus
UBVI errors less than 0.3 mag,111: . . . plus a CI in the range [1.1, 1.8] mag,1111: . . . plus U brighter than the 90% com-pleteness limit.All analysis is performed on clusters with f UBVI = 1111. The scheme is complemented by asecond flag, f UBVI + , which also considers the pho-tometric quality in bands other than UBVI (inthis case H α ), and is marked with a similar se-quence of the number 2 for clarity (2, 22, 222,2222). The effect of filtering the sample acrossthe UBVI baseline on photometric completenessis negligible, and will be discussed fully in thecontext of the luminosity functions of Section 5.2.10ig. 6.— Metallicity distribution ( left ) of all sources in the filtered catalogue ( f UBVI = 1111, see Section 3.4),divided according to galactocentric radius. We find sub- and super-solar metallicities to be preferred inNGC 4041. However, the uncertainties in photometrically derived metallicity are considerable, so no firmconclusions can be drawn from this histogram. The metallicity distribution also does not show any spatialtrends. This is shown on the right , where all clusters in the final sample are symbol coded according toderived metal content.This filtering process, as applies to NGC 4041,is demonstrated in Fig. 8. The top row showstwo segments of our F336W image where we marksources with f UBVI of 1 through 1111 as reddots and green, blue, and yellow circles. The ra-dius of 3 px matches the photometric aperture in U . The bottom row shows a UBVI color di-agram, color-coded in the same way, to demon-strate the effect of each filter on the selectionprocess, along with the corresponding brightness-error plot. From these images it is clear that thedistance to this galaxy gives rise to a crowded fieldof clusters and associations. Coupled with thestochastic variations in the light profile (presentat all distances), this severely cuts down the num-ber of clusters for which reliable size measurementscan be performed. At 22 . ≈ . HST imag-ing, which corresponds to just under 2 pc. Thatway, the smallest effective radius that can be sam-pled is ≈ R eff > α , I, J (e. g. NICMOS-F110W), H (e. g.NICMOS-F110W). The shorthand filter represen-tations are defined in the table header. Positionalinformation is also included, in the form of coor-dinates measured on the F336W image.The results of the SED fits are collected in anSED Table, featuring a minimum, best-fit, andmaximum value for age, mass, and extinction; thenumber of data points (bands) employed by thefit; and the reduced χ statistic of each fit.11ig. 7.— SED fits of four cluster candidates in NGC 4041 with Yggdrasil models, following the process ofAdamo et al. (2010a). The filled blue dots represent the observed integrated fluxes, with associated errorbars. The corresponding integrated model fluxes are labelled as red squares and the horizontal lines indicatethe bandpass of the given
HST filter. The quality of the fit is shown in the lower panels of each set, wherethe χ ν is plotted across the parameter spaces of age-extinction and age-mass. The red contour line indicatesthe range of acceptable fit solutions. Each plot is labelled with an identifier, and the derived age and massof the cluster candidate. The top row shows good fits, while the cluster candidates of the bottom row arefiltered out of the final catalog. Cluster 4427 is found in a crowded region and its light is contaminated byneighboring sources. As such it does not pass the CI test employed by the filtering process and receives f UBVI = 11. 12ig. 8.— The cluster selection process, demonstrated for the population of NGC 4041. The top panels showclose-ups of the F336W frame (inner, outer galaxy on the left and right). Each image measures at approxi-mately 1 . × . f UBVI values of 1, 11, 111, and 1111 as red dots and green, blue, and yellowcircles of radius 3 px, coinciding with the photometric aperture. The bottom panels follow this selection ona
UBVI color diagram, with associated single stellar population model tracks (Zackrisson et al. 2011). Themetallicities of the dashed (green), solid (black) and dashed-dotted (magenta) lines are 0 . , . , . × Z ⊙ . Onthe right we show a U magnitude-error plot. None of the criteria appear to introduce color-space biases.
4. The Structure and Environment ofNGC 4041
We begin the analysis of the SHUCS datasetwith the SA(rs)bc galaxy NGC 4041 (de Vaucouleurs et al.1991). We adopt a distance of 22.7 Mpc and a cor-responding distance modulus of 31.78 from Tully (1988). Given this distance modulus, we derivean absolute magnitude of g = − . HST color-compositeimage of the galaxy in
UBVI and H α . The13alaxy appears to have a partitioned structure,with tightly wound inner spiral arms giving way toa looser, outer network. The boundary is definedby a smooth, yellow bulge-like feature, suggestiveof an old underlying stellar population. The bulgeis overlaid with bright, blue star clusters and dustlanes, and it fades quite rapidly toward the out-skirts of the galaxy, where blue star clusters canbe found in large numbers. Finally, we observethe nuclear source to be split into two componentswith comparable fluxes, both offset from the centerof the outermost isophote by 0 . ′′ ′′ . The fol-lowing sections treat the galaxy and its surround-ings in more detail. Multi-component spirals have been observed inthe past, first the southern supergiant NGC 6902(Gallagher 1979), and then in the context of lowsurface brightness galaxies (Bosma & Freeman1993). Matthews & Gallagher (1997) observedsuch structures in a sub-sample of ‘extreme’late-type galaxies. They also find offset cen-tral peaks in such galaxies, as do Odewahn et al.(1996) in another catalog of late-types, andKarachentsev et al. (1993) in Magellanic spirals.Matthews & Gallagher discuss the possibility ofoffset central peaks representing the true dynam-ical centers of their hosts, i. e. centers of grav-ity offset from the geometric or isophotal center.Marconi et al. (2003) touch on this scenario bysuggesting that the dynamical center of NGC 4041might be decoupled from its geometric center.The composite structure of NGC 4041 is investi-gated further through plotting the surface bright-ness of the galaxy in azimuthally averaged bins,normalized to the central peak (despite the smallcentral offset, to simplify illustration). This isshown in Figure 9, where the central exponen-tial disk measures at a radius of ≈ ′′ (2.4 kpc)and we derive a radius of ≈ ′′ (9 kpc) for theouter galaxy from the optical isophotes. As notedabove, the spiral structure is not discontinuous,with some spiral arms stemming from the floc-culent inner structure and developing into looseouter arms. Across the disk the inter-arm dis-tance increases from ∼ . ≈ . ≈ . . ′′ . r / law,while an index of 1 equals an exponential profile.spiral galaxies, while the outer disk is better de-scribed by Sersic (1968) profiles of high index, typ-ical of early-type galaxies (e. g. Cˆot´e et al. 2006).The F336W profile shows a periodic fluctuation.While one might expect the population of youngstar clusters to contribute to this, Larsen (2004b)show the contrary: star clusters are never found tocontribute more than 10% of the overall U bandluminosity of a spiral galaxy (and up to 20% instarbursts). The complex structure of NGC 4041 couldbe the result of a past dynamical event, so wesearch the
NASA Extragalactic Database ( NED )for neighbors at accordant redshifts. We limitthe search to galaxies in a cone of 1 degreeabout NGC 4041 and in a velocity space of∆ v <
400 km s − . The search yields five neigh-bors, so the galaxy is by no means isolated. Detailsare given in Table 3, and a finding chart shown inFigure 10.The first five galaxies in this list comprise groupLGG 266 in the listing of Garcia (1993). Thesixth, MGC+11-15-013, also fits the LGG cata-log limiting criteria – a velocity difference of less14ig. 10.— The loose grouping of galaxies around NGC 4041. One is a distorted spiral, three appear irregular(albeit the high inclination hinders this classification), and one is an early-type galaxy with a faint innerdust lane (evident in regular contrast scaling, but not here). We mark the velocity of each galaxy on theSDSS mosaic image on the left, and show high-contrast zoom-ins on the right (not to scale). The foursmaller galaxies (top two rows) also exhibit considerable FUV fluxes on GALEX images, indicative of recentor ongoing star formation. Their calculated SFRs are listed in Table 3. The color of the inner disk ofNGC 4041 appears similar to that of the lenticular NGC 4036, while the outer disk shines bright in shorterwavelengths.than 600 km s − and a projected separation lessthan 0.52 Mpc – but was only discovered recently(Liske et al. 2003). Since we find no morphologicalclassification for this galaxy in the literature, weinspect its z band SDSS image to find a linear pro-file with no pronounced central peak. We thereforeclassify it as Irr. The close-up of Figure 10 showsa clumpy profile, but it is biased by bright starforming regions at the extremes of the disk thatare not detectable in the z band. The dominant galaxy of this group is NGC 4036, a lenticular. Itfeatures an inner equatorial dust lane and registersLINER emission (V´eron-Cetty & V´eron 2006). Upon inspecting
GALEX
UV imaging ofNGC 4041 we find a faint, three-pronged featureto the south of the UV disk. It extends between125 ′′ and 220 ′′ , or ∼ −
24 kpc. This ‘tidalfork’ is shown in the far-ultraviolet plus near-15 able 3The NGC 4036 Galaxy Group
ID RA DEC m g v R SFR log( M ∗ ) Morphology References(h m s) ( ◦ ′ ′′ ) (mag) (km s − ) (M ⊙ yr − ) (M ⊙ )NGC 4036 12 01 26.7 +61 53 45 11.97 1385 0.02 10.42 S0 [1] [1] [2]NGC 4041 12 02 12.2 +62 08 14 12.19 1234 0.63 10.58 SA(rs)bc? [1] [1] [3]IC 0758 12 04 11.9 +62 30 19 14.25 1275 0.19 9.26 SB(rs)cd? [4] [4] [3]UGC 7009 12 01 44.1 +62 19 33 14.33 1119 0.12 8.78 Im [4] [4] [4]UGC 7019 12 02 29.4 +62 25 02 15.05 1518 0.05 · · · Im [5] [6] [7]MGC+11-15-013 12 02 43.3 +62 29 52 16.07 1059 0.03 · · ·
Irr [4] [4] [4]
Note.—
Morphologies are taken from de Vaucouleurs et al. (1991), apart from MGC+11-15-013, which we classify inSection 4.2. Stellar masses are derived though ‘best’ 2MASS K S magnitudes (Skrutskie et al. 2006), except NGC 4041,for which we use an aperture of 72 . ′′ GALEX
FUV imaging (see Section 4.4). SDSS photometry is model photometry in the g band, where applicable.Positions, photometry, and radial velocities are drawn from the following sources, quoted in triplets in column 7: [1] SDSSData Release 6, Adelman-McCarthy et al. (2008); [2] Cappellari et al. (2011); [3] de Vaucouleurs et al. (1991); [4] SDSSData Release 2, Abazajian et al. (2004); [5] Cotton et al. (1999); [6] VATT B band photometry, Taylor et al. (2005);[7] Springob et al. (2005); The first five galaxies in this list make up group LGG 266 in the listing of Garcia (1993), andthe sixth also obeys the inclusion criteria. Given the individual distance moduli, we derive absolute g band magnitudesof − . − . − . − . − .
7, and − . ultraviolet (FUV+NUV) image of Figure 11. Westack the two GALEX channels to boost the flux,and smooth by a three-pixel Gaussian kernel toremove noise. We present this image at high con-trast to accentuate low surface brightness features.The contours mark optical brightness, derived bystacking the g and r SDSS images, and smoothingby a ten-pixel Gaussian kernel.The feature is detectable in the UV image atthe 3-5 σ level, but not quite as bright in thestacked and smoothed optical image ( . σ de-tection). Assuming this is a tidal feature physi-cally associated with NGC 4041, we can attributeit to a past interaction. The low g + r flux denoteslow or no H α emission, and therefore no currentstar formation in the ‘fork’. The UV flux shouldthen mostly originate from activity on the order of100 Myr ago, rather than current star formationand O-stars. Since there are no published values for thestar formation rate of NGC 4041, we apply theKewley et al. (2002) methodology to obtain esti- mates from the
IRAS infrared fluxes at 60 µ m and100 µ m. In brief, we first employ the Helou et al.(1988) prescription to obtain the far-infrared flux, F FIR , from the
IRAS fluxes, which we then con-vert to a luminosity through L FIR = 4 π D × F FIR ,where D is the distance to the galaxy. We thenuse the Kewley et al. adaptation of the Kennicutt(1998) law to derive a FIR star formation rate(SFR) of 4.10 M ⊙ yr − for NGC 4041. Thiscompares well with galaxies of similar morpho-logical type and brightness in the SINGS survey(Kennicutt et al. 2003). The flux ratio between 60and 100 µ m of 0.44 is also in accord with SINGS.We also obtained an FUV SFR for NGC 4041through archival GALEX imaging, which we ex-pect to be lower, due to the absorption of UVphotons by dust. To estimate the flux fromthe target we fit it with the
ELLIPSE functionin IRAF and added up the counts correspond-ing to the two segments of the galaxy disk (as-suming the above radii of 80 ′′ , ′′ and no el-lipticity), while accounting for a flat backgroundlevel and a foreground extinction of E ( B − V ) =0 .
15 mag (following A FUV = 7 . × E ( B − V ) fromGil de Paz et al. 2007). Note that the low reso-16ig. 11.— FUV+NUV image from GALEX ,smoothed with a three-pixel Gaussian kernel, andpresented at high contrast to emphasize low sur-face brightness features. We reveal the presenceof material beyond even the extended UV disk ofNGC 4041, in the form of a three-pronged struc-ture. This ‘tidal fork’ might be part of a greaternetwork of debris, but it is the only feature de-tected at a level of 3-5 σ above the background.The contours show the isophotal structure of a g + r image from SDSS at the 1, 2, 3, 5, 10 σ level.Only the brightest, central prong is detectable inSDSS at the 1-2 σ level. The absence of H α emis-sion (transmitted in the r band filter) suggests anage on the order of 100 Myr. Concentric circlesmark the outer disk of NGC 4041, a faint, ex-tended UV disk, and the full radial extent of thefork. The angular distances correspond to 9, 14,and 24 kpc at the distance of 22.7 Mpc. The brightfeature in the southwest is a foreground F-typestar.lution of IRAS images does not allow for a sim-ilar treatment. We then convert fluxes to starformation rates of 0.18, 0.65 M ⊙ yr − throughthe FUV Kennicutt (1998) relation. Consider-ing the area of each segment, we derive SFRdensities of [0 . , .
08] M ⊙ yr − arcmin − , or[1 . , . × − M ⊙ yr − kpc − for the innerdisk and outer galaxy respectively. We apply theabove methodology to also derive the FUV SFRsof the group around NGC 4041, and list results inTable 3. The observation of a markedly higherSFR density in the central regions of a galaxy is common in post-interaction systems, where gasis funneled inwards (F¨orster Schreiber et al. 2003;Konstantopoulos et al. 2009) until it is the nucleusalone that is experiencing a starburst, or until itignites an active nucleus (Ellison et al. 2011).Stellar masses, M ∗ were derived by applyingthe Bell et al. (2003) prescription with a K S solarbrightness of 3.32 mag to catalogued 2MASS K S photometry (Skrutskie et al. 2006). The faintesttwo members of the group are below the detectionlimit, and the obtained values for the four brightergalaxies are listed in Table 3. While we generallyuse the 2MASS ‘standard’ radius, in the case ofNGC 4041 we use a radius of 72 ′′ in order to en-velop the entirety of the outer galaxy. This givesrise to a larger M ∗ for NGC 4041 than NGC 4036,despite the brighter tabulated g band magnitudeof the latter. Additionally, we derive a specificSFR (sSFR) of NGC 4041, through division withthe FIR SFR, of 10 . × − yr − , consistentwith the morphological type of NGC 4041 – cf. its near morphological counterparts HCG 7C andHCG 59A, that also register values ≃ − yr − (Konstantopoulos et al. 2010, 2012).Finally, we look for information on the gaseouscomponent of NGC 4041. The survey of Couto da Silva & de Freitas Pacheco(1989) offers a value of log( M H i / M ⊙ ) = 9 . α CO =4 . ⊙ (Solomon et al. 1987), to convert thelog( L mol ) = 8 .
48 K km s − pc to a gas massof log( M mol / M ⊙ ) = 9 .
14. The combined gasmass, M gas = 1 . M H i + M mol (the factor of0 . M H i accounts for helium, e. g. Hunter et al.1982), is therefore log( M gas / M ⊙ ) = 9 .
95. Dividedby the SFR, we get a gas depletion timescale of ∼ . Many of the traits examined in this sectionare consistent with a past dynamical event inNGC 4041: a segmented brightness profile; a dou-ble central peak, offset from the isophotal center;a UV-bright tidal feature; and a markedly highercentral SFR. Of particular interest is the offsetcentral peak, which potentially represents the dy-namically decoupled core of NGC 4041 (following17arconi et al. 2003). Such features herald pastmergers (e. g. Emsellem et al. 2011), not interac-tions.The group environment is thought to beconducive to dynamical events and acceleratedgalaxy evolution (e. g. Verdes-Montenegro et al.2001; Johnson et al. 2007; Gallagher et al. 2010;Konstantopoulos et al. 2010, 2012). Furthermore,while a dense environment can enhance star for-mation during interactions (Martig & Bournaud2008), the effects of minor events are often notvery pronounced, leaving but the faintest tracesof their passing (Konstantopoulos et al. 2010). Atthe same time, compact galaxy groups have beennominated as the sites of lenticular galaxy forma-tion (Wilman et al. 2009), as minor mergers buildup mass and gradually exhaust the gas reservoir ofthe group, while retaining the structure of individ-ual disks. The high-index S´ersic outer brightnessprofile for NGC 4041 (Figure 9) is more akin to alenticular than a spiral.The overall symmetry of the disk seems to ruleout a major merger. We therefore propose an ac-cretion event, perhaps that of a gas-rich dwarf, asthe origin of the present star formation activityand the two-component disk of NGC 4041.
5. The Bimodal Star Cluster Populationof NGC 4041
In the Introduction of Section 1 we argued thatthe understanding of a galaxy and its surround-ings can be enhanced by a study of the star clus-ter population. We now present such a study forNGC 4041. We split the sources according togalactocentric radius ( r gc ), in order to contrastthe sub-populations of the inner and outer regions,and derive analysis only from high-confidence de-tections, i. e. sources flagged in Section 3.4 as f UBVI = 1111.
The colors of star clusters are indicative of theage of the stellar population they represent, henceSSP models can be used to interpret the colors ofcluster candidates across NGC 4041. The color-magnitude diagram (CMD) of Figure 12 plots theevolution of SSP models of various masses (from10 to 10 M ⊙ ), and shows dissimilar distributionsfor clusters in the inner and outer galaxy (blue circles/red triangles). While this hints at differentmass functions, incompleteness will eliminate thelow-luminosity end of the inner disk distribution.Still, such dramatic contrasts between subpopu-lations within a single system are not normallyobserved outside merging and interacting galaxies(e. g. Fedotov et al. 2011). We remind that no spa-tial trends were discovered in the metallicity dis-tributions of clusters in the inner and outer galaxy(see Section 3.3), and note that the reddening dis-tribution shows no change between inner disk andouter galaxy.Fig. 12.— Color-magnitude diagrams of high-confidence cluster candidates with Yggdrasil mod-els of solar metallicity overplotted on isotropicaxes. Model tracks represent masses of 10 , 10 ,10 , and 10 M ⊙ from bottom to top, each trac-ing ages between 6 Myr and ∼
10 Gyr. Num-bers in filled circles mark each age dex of modelSSP evolution. There is a pronounced differencein the color distributions of clusters in the innerdisk (blue dots) and outer galaxy (red triangles),likely a combination of differing detection limitsand possibly also mass distributions. Such dispar-ities are often observed in interacting systems, butnot elsewhere. The crosshairs in the bottom leftrepresent the median error in each axis.We further investigate this disparity throughthe color-color plot of Figure 13. As in the CMDof Figure 12, the model track spans the full evo-18ution of an SSP, from 1 Myr to about 10 Gyr.The SSPs account for the transmission of nebularcontinuum and emission lines in the F606W filter.This extends the purely stellar SSP toward greenervalues and therefore cover sources that scatter to-ward the top left of the color distribution. Weshow color contours on the right panel to aid com-parison, where blue and red shades represent theinner/outer galaxy. To avoid over-representingoutliers we omit the lowest two contour levels,hence plotting only two-dimensional bins thatsample at least 15% of the full population. Sourcesin the inner disk appear to often diverge towardbluer values. We attribute this to crowding andaperture effects: on the vertical axis, the B flux is measured from WFPC2, at lower resolu-tion; and on the horizontal axis, we might expectsome contamination by the diffuse, irregularly dis-tributed H α emission in the inner disk. In addi-tion, the colors of low-mass clusters might pref-erentially diverge toward this region of colorspace(Fouesneau & Lan¸con 2010; Popescu & Hanson2010; Silva-Villa & Larsen 2011). A third possi-bility is a variety of filling factors in the gas, whichmight lead to variations of up to 0.5 mag in themodel tracks. Since we adopt a given filling factorand electron density, we might be underestimatingthe blue emission from these regions.The loci of the color contours (on the right), in-dicated by stars, are offset by ≈ . cf. the cluster populationsof the Antennæ (Whitmore & Schweizer 1995),NGC 3256 (Trancho et al. 2007a,b), and most no-tably NGC 7252, which exhibits an inner starforming disk and an outer halo-type distribu-tion (Miller et al. 1997; Schweizer & Seitzer 1998).Even so, such an orderly inner-versus-outer galaxydistinction has not been noted in past analyses. The Luminosity Function (LF) of star clus-ter populations is routinely observed to have asmooth, power-law shape, as noted in the Intro-duction. While this shape can be seen in thebinned LF of Figure 14 (left), the cumulative func-tions on the right panel reveal a striking substruc-ture of: (i) an incomplete part with an evolvinggradient; (ii) a smooth, power-law segment; and(iii) a truncation at the bright end. This indi-cates that substructure may have been habituallyignored in the literature by arranging clusters inbins of arbitrary width (see the cautionary note ofMa´ız Apell´aniz & ´Ubeda 2005).We measure the slope from the cumulative func-tion, in two ways. First we fit only the smoothpart of each function, but maintain the fit consis-tent between the four bands. Then we fit over alarge range, in order to emulate the effect of bin-ning. The results, listed in Table 4, show shallowerfunctions for the full range fit (consistent with theliterature but slightly steeper than the canonicalvalue of − U returns markedly shallower fits thanthe rest of the bands, an effect noted in the liter-ature review of Gieles (2010), where redder bandswere shown to have steeper distributions. We notethat the extinction does not change the slope ofthe LF (Larsen 2002). The Gieles analysis alsoshowed LF slopes to be a function of the mean lu-minosity. The mean brightness over which we fit(21 or 22 mag; see Table 4) translates to a lumi-nosity of log( L/ L ⊙ ) ≃
5, corresponding to litera-ture slopes of − .
5. We attribute the consistentlysteeper slopes derived from the binned histogramsto sampling uncertainties, especially given the rel-atively small cluster population being fit. This dis-crepancy exemplifies the loss of information that19ig. 13.—
Left: U − B vs V − I colors of star cluster candidates in NGC 4041 on isotropicaxes. We overplot Yggdrasil models, following the conventions of Figure 12, and also include sub- andsuper-solar tracks in this case. Clusters in the inner disk and outer galaxy are marked as blue circles andred triangles respectively, while the crosses on the lower left denote the median photometric error of eachsubsample. The extinction vector marks one magnitude of attenuation in the V band (Cardelli et al. 1989).Sources in the inner disk often diverge from the model track toward bluer values (top left), a likely effect ofcrowding. This could alternatively be attributed to differing filling factors in the gas, in which the emissionoriginates (see Section 3.3).
Right:
Two-dimensional color-density plot for the inner disk (blue), and outergalaxy (red). We omit areas that sample less than 15% of the 2D-histogram peak. The two distributionsappear slightly offset, extending each other in opposite directions. The median colors (marked as stars) areseparated by ≈ . UBVI baseline. Thisimplicitly includes the
BVI bands in defining thecompleteness of our sample. In order to estimatethis effect, we plotted the LFs by omitting thecriteria pertaining to multi-band photometry andfound the shape of each function to be unchangedby this process. The measured slopes were con-sistent within the errors with those listed in Ta-ble 4. We also examined the effect of potentiallyage-dependent reddening, by repeating the aboveexercise after limiting the age and extinction of thesample being fit. In all cases, the derived slopeswere consistent with the fits of Table 4.20ig. 14.— Binned ( left ) and cumulative ( right ) luminosity functions (LF) of star cluster candidates inNGC 4041, arranged according to filter and r gc (inner disk, outer galaxy). The lines (bands on the right)denote the range over which we fit the slope of each function, and we note slopes next to each pair of LFs.The U LF of the inner disk is shallower than other functions, which are generally consistent within theerrors. A higher overall extinction would not explain this discrepancy. Slopes steeper than − Right : LF slopes,measured on the cumulative functions of the center panel. Dashed lines indicate the 90% completenesslimits. As expected by the literature review of Gieles (2010), the slope becomes steeper with increasingbrightness. We attribute this to the segmented structure of the function, as it gives rise to shallow slopes inthe incomplete part, values consistent with the literature in the power-law segment, and very steep slopesonce the upper truncation sets in. The increasing errors reflect the sampling (decreasing numbers).Table 4: Luminosity Function Fit Parameters and Outcomes.Filter Fit Range (in/out) Slope, unbinned (in/out) Slope, binned (in/out)(mag) ( − ) ( − )F336W (U) [20 . , .
0] [21 . , .
5] 2 . ± .
24 2 . ± .
15 2 . ± .
10 2 . ± . . , .
0] [21 . , .
5] 2 . ± .
57 2 . ± .
56 3 . ± .
12 2 . ± . . , .
0] [21 . , .
5] 3 . ± .
40 3 . ± .
48 3 . ± .
13 3 . ± . . , .
0] [21 . , .
5] 3 . ± .
37 2 . ± .
21 2 . ± .
11 2 . ± . . , .
6] [20 . , .
6] 2 . ± .
21 2 . ± .
23 1 . ± .
04 1 . ± . . , .
6] [21 . , .
6] 2 . ± .
38 2 . ± .
16 2 . ± .
06 2 . ± . . , .
6] [21 . , .
6] 2 . ± .
42 2 . ± .
25 1 . ± .
07 2 . ± . . , .
6] [21 . , .
6] 2 . ± .
43 2 . ± .
27 1 . ± .
07 2 . ± . Note.—
The top and bottom tiers show fits to the ‘smooth’ and ‘full’ ranges (see text). The limits of the full range fitsmeasure between 20 mag and the 90% completeness limits for the inner disk and outer galaxy (see Section 3.1). .3. Mass Function In order to quantify the cluster mass distribu-tions in the inner and outer disk, we plot the MassFunction (MF) in Figure 15 as a set of cumulativedistributions (counting from high-to-low mass).This plot will suffer from various incompletenesseffects at different star cluster ages. On the youngend we need to exclude short-lived, unbound stel-lar associations that masquerade as star clusters atlarge distances (Gieles & Portegies Zwart 2011).Toward older ages, the diagram is affected bythe evolutionary fading and reddening of SSPs.This decreases the mass-to-light ratio, thereforeincreasing the mass required to detect a star clus-ter with increasing age. These evolving detectionlimits are represented by the dotted and dashedlines (inner/outer galaxy) on Figure 15, and shapethis parameter space.We therefore restrict the MF to cluster candi-dates aged between [10, 100] Myr. We note anoffset of ≃ . α = − . − .
0. In brief, we follow the methodology ofBastian et al. (2012) and run a series of modelsof the MF with a sample size equal to the ob-served sub-populations, and assume a power-lawdistribution. The slope is derived after few trial-and-error iterations. The median of all models isshown as a solid line, and dashed and dotted linesindicate the range containing 50% and 90% of allmodeled outcomes. The observed functions areseen to diverge from the power-law models at thelow-mass end, ]bf and also the and high-mass endfor the outer galaxy. While incompleteness willaffect low-mass clusters, at the high-mass end weexpect to detect the vast majority of sources. Thisdivergence is therefore established as a physical ef-fect and the MF of the outer galaxy interpretedas a Schechter-type distribution with a trunca-tion at log( M ∗ / M ⊙ ) ≈ .
4. This follows on the Fig. 15.— Mass functions (MFs) of star clustersin the inner disk and outer regions of NGC 4041.The MFs cover only clusters aged 10 −
100 Myrand thus avoid fitting on young, unbound associa-tions, and incomplete samples at older ages. BothMFs show a composite structure in three parts: anincomplete low-mass section, followed by a power-law, which is truncated above log( M ∗ / M ⊙ ) ≈ . − . Combining mass with age we obtain the dia-gram of Figure 16. The relative shift in mass notedabove (Section 5.3) is obvious here as a verticalshift according to different detection limits, whilethe age distributions appear largely similar andnot unlike others studied in the past. The three in-ner disk outliers at high mass and young age repre-sent the double nuclear peak and another, nearbysource. Assuming these are nuclear star clusters,we do not expected them to follow normal scalinglaws (Seth et al. 2008; Scott & Graham 2012); asmixed stellar populations their mass-to-light ratioswill be unlike those of SSPs.From the first age dex, log( τ /yr ) ∈ [6 , Yggdrasil model across thegiven range of ages. This limit evolves accord-ing to the temporally increasing mass-to-light ra-tio of SSPs. The two sub-populations are offsetby approximately 0.5 dex in mass, which followson the relative shift of Figure 15. The age dis-tributions appear consistent and typical of theirlate-type host galaxy.We perform a maximum likelihood analysis ofthe age and mass distributions, shown in Fig-ure 17, given the cluster disruption formulationsof Lamers et al. (2005) and Gieles (2009). Thisway we simultaneously compute M ∗ and t , thedissolution time of a 10 M ⊙ cluster. We employa Schechter function and the modeled evolution ofan Yggdrasil
SSP in V , the limiting filter (seeSection 3.1, Figure 3). As with the LF analysis, weexclude sources younger than 10 Myr to avoid stel-lar associations, and take into account incomplete-ness at older ages (through the Yggdrasil model).This leaves the black dots of Figure 17, which are23t with models based on a range of variations of M ∗ and t .The fits favor a t of a few hundred Myr in theouter galaxy and a longer time in the inner disk,contrary to Lamers et al. (2005), who found en-vironment density to correlate inversely with t .However, the small number of data-points in theinner disk provides broad confidence contours, ex-pressed as shallow topography, therefore no de-ductions should be made from this fit. The M ∗ seems to change significantly with local environ-ment (inner/outer galaxy), in a manner consistentwith the M83 study from which this methodologyis derived: denser regions display higher M ∗ . A final test for a physical difference be-tween the inner/outer galaxy populations canbe drawn from the cluster age distribution ofFigure 18 (left panels, often referred to as a‘d N/ d t plot’), which counts the number of clus-ters surviving from every age-step of clusterformation (see Lamers et al. 2005; Fall et al.2005; Whitmore et al. 2007; Chandar et al. 2010;Bastian et al. 2011, and references therein). Whenplotting all sources, as we do on the top panel, thisdiagnostic is affected by evolutionary fading anddisplays a power-law decline over time. Given astable star formation history, setting a lower limitto the mass of a cluster sample will flatten thediagram to a certain age – the higher the masscut, the farther back in time this diagnostic canreach.In the bottom panel of Figure 18 we performthis exercise for a mass cut that ensures com-pleteness to log( τ / yr) = 8 . ≈
300 Myr). Theprecise mass of each cut (different for the innerand outer galaxy) is extracted from the age-massdiagram of Figure 16, as the intersection of thelog( τ / yr) = 8 . N/ d t plot. We note,however, that future SHUCS analyses focussed onnearby galaxies will not lack this diagnostic power. In previous Sections we have argued that bin-ning the typically small data-sets of extra-galacticstar cluster populations erases information con-tained in the individual data-points. In order toinvestigate this effect in the age distribution, weadd two more plots to the right panels of Fig-ure 18. There we modulate the age distribution bythe mass of each cluster, to obtain a mass outputplot. The full populations show a slight dissimi-larity in their outputs over time, which, however,disappears when only plotting the complete sam-ple of high-mass clusters. In all, the study of star clusters in NGC 4041is in tune with the derivations of the traditional,broad-band diagnostics of Section 4. The colordifference between inner and outer galaxy is mir-rored in the differing color, luminosity, mass, andage distributions of the corresponding star clus-ter populations. Unfortunately, the d N/ d t plot islimited by the number of clusters available abovethe 90% completeness limit. Therefore, the pastdynamical or accretion event deduced throughmulti-wavelength metrics above cannot be con-firmed through the cluster population. In follow-ing SHUCS investigations, however, we expect thesmaller distances to the galaxies targeted to offeran opportunity to explore their dynamical histo-ries via the age distribution.
6. Summary
We have presented the observational and tech-nical setup of the Snapshot
Hubble
U-band Clus-ter Survey (SHUCS), comprising new F336W (U-band equivalent) observations of 10 galaxies withexisting
BVI coverage. We will use this new imag-ing to complete the
UBVI baseline, as requiredfor the accurate photometric age dating of starclusters. The survey is focussed primarily on starclusters in the first ∼ Gyr of their evolution, andis aimed at understanding the formation and earlyevolution of these objects, as well as their utilityas tracers of star formation history. In this paperwe have also demonstrated the high scientific yieldof combining star cluster statistics with broad-band, multi-wavelength metrics. We propose that,with large-scale information readily available from24ig. 17.— Disruption time for a 10 M ⊙ cluster ( t ) versus Schechter Mass ( M ∗ , where the truncation setsin) for the inner disk and outer regions of NGC 4041. This plot uses the age-mass diagram of Figure 16and completeness (based on an Yggdrasil
SSP model) to estimate the two quantities through a series ofmaximum-likelihood tests. Similarly to the MFs of Figure 15, only the black dots are used, in order to avoidincompleteness effects and short-lived stellar associations. We find values consistent with the literature inthe outer galaxy, while the atypically high M ∗ derived for the inner disk might be affected by crowding andincompleteness. This is evident from the shallow gradient of the contours, i. e. the large range of values thatfit the data well. GALEX , SDSS, 2MASS, IRAS, and other surveys,this two-pronged approach is not only advanta-geous, but imperative in achieving a well-roundedunderstanding of each cluster population at hand.This combined approach was demonstratedthrough the analysis of the large-scale propertiesand stellar populations of NGC 4041, a massivegalaxy at the upper tier of the SHUCS distancescale ( d &
15 Mpc). This system features a com- plex physical and dynamical structure, expressedas a discontinuous brightness profile perhaps rem-iniscent of M64 from a different vantage point. Wefind a strong distinction in the colors and masses ofstar clusters when applying a cut in galactocentricdistance where this break occurs, reminiscent ofthe color and age distributions in interacting andmerging systems. Most notably, we find the innerdisk to have been forming more massive clusters25ig. 18.— The cluster age distribution histograms (‘d N/ d t ’ plots) of the left panels count the numberof surviving clusters per time interval over the past Gyr. Isotropic axes show an x = − y slope in themagnitude-limited sample on the top, as expected from the literature. Applying a mass cut should flattenthe d N/ d t of a galaxy with a stable star formation history. In this case, the slope lessens, but is difficultto interpret given the small number of data-points. In this work we have argued that binning might eraseinformation, therefore we also present unbinned data in the cumulative age distribution on the ( right ). Thetwo sets of plots show consistent results.( M & M ⊙ ) in the past ≈
100 Myr, despitethe overall redder appearance of its subpopula-tion, which indicates an older mean age. Thisis in tune with our analysis of archival
GALEX
FUV imaging, which reveals a higher SFR inthe inner disk than the outer galaxy, similar topost-starburst systems. This could be linked tothe theoretical expectation of a higher star for-mation efficiency and a higher fraction of stars forming in clusters in regions of higher gas den-sity (Kruijssen 2012). We also discovered a tidalfeature, which, combined with information in theliterature, strongly favors an accretion event inthe recent past ( ∼
100 Myr) as the origin of thesegmented morphology of NGC 4041.Throughout the galaxy we found a truncationin the star cluster mass function, in accord withthe recent studies (Larsen 2009; Gieles 2009).26his truncation occurs at a higher mass in theinner disk, as expected by these recent results(Bastian et al. 2012). Furthermore, we found boththe mass and luminosity function to break downinto three segments when binning is abandonedin favor of cumulative distributions: one dom-inated by incompleteness, one following the fa-miliar power-law shape, and one encompassing atruncation at the upper end. Our results stronglyadvocate for the use of all available information,rather than binning, when characterizing the age,mass, and luminosity distributions of star clusterpopulations. In this work we have sought that re-sult through the use of cumulative distributions.The strength of the survey-at-large derivesfrom the availability of deep U observations atthe highest resolution available (currently HST -WFC3), enabling the precision age dating of hun-dreds of clusters per galaxy – potentially thou-sands in nearby galaxies. The full survey willconsist of the analysis of the cluster populationsof ten late-type galaxies of various morphologicaland spectroscopic types. In addition to the un-precedented statistical value of this analysis, weexpect to focus a few works on individual sys-tems, namely the ongoing merger in NGC 2146(see the pilot study of Adamo et al. 2012), granddesign spiral NGC 2997, and NGC 247, where wewill contrast the star formation history derivedfrom field stars to that derived from star clusters.Combined, the individual studies will help:1. Search for a characteristic value ( M ∗ REFERENCES
Abazajian, K., Adelman-McCarthy, J. K.,Ag¨ueros, M. A., et al. 2004, AJ, 128, 502, 502Adamo, A., ¨Ostlin, G., & Zackrisson, E. 2011,MNRAS, 417, 1904, 190427damo, A., ¨Ostlin, G., Zackrisson, E., et al. 2010a,MNRAS, 407, 870, 870Adamo, A., Zackrisson, E., ¨Ostlin, G., & Hayes,M. 2010b, ApJ, 725, 1620, 1620Adamo, A., Smith, L. J., Gallagher, J. S., et al.2012, MNRAS, 426, 1185, 1185Adelman-McCarthy, J. K., Ag¨ueros, M. A., Al-lam, S. S., et al. 2008, ApJS, 175, 297, 297Anders, P., Bissantz, N., Fritze-v. Alvensleben, U.,& de Grijs, R. 2004a, MNRAS, 347, 196, 196Anders, P., de Grijs, R., Fritze-v. Alvensleben, U.,& Bissantz, N. 2004b, MNRAS, 347, 17, 17Barmby, P., Kuntz, K. D., Huchra, J. P., &Brodie, J. P. 2006, AJ, 132, 883, 883Bastian, N. 2008, MNRAS, 390, 759, 759Bastian, N., & Gieles, M. 2008, in AstronomicalSociety of the Pacific Conference Series, Vol.388, Mass Loss from Stars and the Evolution ofStellar Clusters, ed. A. de Koter, L. J. Smith,& L. B. F. M. Waters, 353–+Bastian, N., Gieles, M., Goodwin, S. P., et al.2008, MNRAS, in press, arXiv:0806.1460,0806.1460Bastian, N., Adamo, A., Gieles, M., et al. 2011,MNRAS, L298+, L298+—. 2012, MNRAS, 419, 2606, 2606Bell, E. F., McIntosh, D. H., Katz, N., & Wein-berg, M. D. 2003, ApJS, 149, 289, 289Bik, A., Lamers, H. J. G. L. M., Bastian, N., Pana-gia, N., & Romaniello, M. 2003, A&A, 397, 473,473Bosma, A., & Freeman, K. C. 1993, AJ, 106, 1394,1394Bressert, E., Bastian, N., Gutermuth, R., et al.2010, MNRAS, 409, L54, L54Cappellari, M., Emsellem, E., Krajnovi´c, D., et al.2011, MNRAS, 413, 813, 813Cardelli, J. A., Clayton, G. C., & Mathis, J. S.1989, ApJ, 345, 245, 245 Chandar, R., Whitmore, B. C., Kim, H., et al.2010, ApJ, 719, 966, 966Cioni, M.-R. L. 2009, A&A, 506, 1137, 1137Cˆot´e, P., Piatek, S., Ferrarese, L., et al. 2006,ApJS, 165, 57, 57Cotton, W. D., Condon, J. J., & Arbizzani, E.1999, ApJS, 125, 409, 409Couto da Silva, T. C., & de Freitas Pacheco, J. A.1989, Rev. Mexicana Astron. Astrofis., 17, 127,127Dalcanton, J. J., Williams, B. F., Seth, A. C.,et al. 2009, ApJS, 183, 67, 67Dale, D. A., Cohen, S. A., Johnson, L. C., et al.2009, ApJ, 703, 517, 517de Grijs, R., Fritze-v. Alvensleben, U., Anders, P.,et al. 2003, MNRAS, 342, 259, 259de Grijs, R., & Parmentier, G. 2007, Chinese J.Astron. Astrophys., 7, 155, 155de Vaucouleurs, G. 1948, Annalesd’Astrophysique, 11, 247, 247de Vaucouleurs, G., de Vaucouleurs, A., Corwin,Jr., H. G., et al. 1991, Third Reference Cata-logue of Bright GalaxiesElfhag, T., Booth, R. S., Hoeglund, B., Johansson,L. E. B., & Sandqvist, A. 1996, A&AS, 115,439, 439Ellison, S. L., Patton, D. R., Mendel, J. T., &Scudder, J. M. 2011, MNRAS, 418, 2043, 2043Elmegreen, B. G., & Hunter, D. A. 2010, ApJ,712, 604, 604Elson, R. A. W., Fall, S. M., & Freeman, K. C.1987, ApJ, 323, 54, 54Emsellem, E., Cappellari, M., Krajnovi´c, D., et al.2011, MNRAS, 414, 888, 888Fall, S. M., Chandar, R., & Whitmore, B. C. 2005,ApJL, 631, L133, L133Fedotov, K., Gallagher, S. C., Konstantopoulos,I. S., et al. 2011, AJ, 142, 42, 4228erland, G. J., Korista, K. T., Verner, D. A., et al.1998, PASP, 110, 761, 761F¨orster Schreiber, N. M., Genzel, R., Lutz, D., &Sternberg, A. 2003, ApJ, 599, 193, 193Fouesneau, M., & Lan¸con, A. 2010, ArXiv e-prints, arXiv:1003.2334Fruchter, A., & Sosey, M. 2009,Gallagher, J. S. 1979, AJ, 84, 1281, 1281Gallagher, S. C., Durrell, P. R., Elmegreen, D. M.,et al. 2010, The Astronomical Journal, 139, 545,545Garcia, A. M. 1993, A&AS, 100, 47, 47Gieles, M. 2009, MNRAS, 394, 2113, 2113—. 2010, ASPC, 423, 123, 123Gieles, M., & Portegies Zwart, S. F. 2011, MN-RAS, 410, L6, L6Gil de Paz, A., Boissier, S., Madore, B. F., et al.2007, ApJS, 173, 185, 185Goddard, Q. E., Bastian, N., & Kennicutt, R. C.2010, MNRAS, 405, 857, 857Helou, G., Khan, I. R., Malek, L., & Boehmer, L.1988, ApJS, 68, 151, 151Ho, L. 2003, in HST Proposal, 9788–+Hunter, D. A., Gallagher, J. S., & Rautenkranz,D. 1982, ApJS, 49, 53, 53Hurley, J. R., & Mackey, A. D. 2010, MNRAS,408, 2353, 2353Johnson, K. E., Hibbard, J. E., Gallagher, S. C.,et al. 2007, AJ, 134, 1522, 1522Jord´an, A., Cˆot´e, P., Blakeslee, J. P., et al. 2005,ApJ, 634, 1002, 1002Karachentsev, I. D., Karachentseva, V. E.,& Parnovskij, S. L. 1993, AstronomischeNachrichten, 314, 97, 97Kennicutt, R. C., Calzetti, D., Aniano, G., et al.2011, PASP, 123, 1347, 1347 Kennicutt, Jr., R. C. 1998, in ASPCS, Vol.142, The Stellar Initial Mass Function (38thHerstmonceux Conference), ed. G. Gilmore &D. Howell, 1–+Kennicutt, Jr., R. C., Lee, J. C., Funes, Jos´e G.,S. J., Sakai, S., & Akiyama, S. 2008, ApJS, 178,247, 247Kennicutt, Jr., R. C., Armus, L., Bendo, G., et al.2003, PASP, 115, 928, 928Kewley, L. J., & Dopita, M. A. 2002, ApJS, 142,35, 35Kewley, L. J., Geller, M. J., Jansen, R. A., &Dopita, M. A. 2002, AJ, 124, 3135, 3135Koekemoer, A. 2005, Space Telescope EuropeanCoordinating Facility Newsletter, 38, 16, 16Konstantopoulos, I. S., Bastian, N., Smith, L. J.,et al. 2009, ApJ, 701, 1015, 1015Konstantopoulos, I. S., Gallagher, S. C., Fedotov,K., et al. 2010, ApJ, 723, 197, 197—. 2012, ApJ, 745, 30, 30Krist, J. E., Hook, R. N., & Stoehr, F. 2011, inSociety of Photo-Optical Instrumentation En-gineers (SPIE) Conference Series, Vol. 8127,Society of Photo-Optical Instrumentation En-gineers (SPIE) Conference SeriesKroupa, P. 2001, in Astronomical Society of thePacific Conference Series, Vol. 228, Dynamics ofStar Clusters and the Milky Way, ed. S. Deiters,B. Fuchs, A. Just, R. Spurzem, & R. Wielen,187–+Kruijssen, J. M. D. 2012, MNRAS, 426, 3008, 3008Kruijssen, J. M. D., Inti Pelupessy, F., Lamers,H. J. G. L. M., et al. 2011, ArXiv e-prints,arXiv:1112.1065Lamers, H. J. G. L. M., Gieles, M., Bastian, N.,et al. 2005, A&A, 441, 117, 117Lampton, M., Margon, B., & Bowyer, S. 1976,ApJ, 208, 177, 177Larsen, S. S. 1999, A&AS, 139, 393, 393—. 2002, AJ, 124, 1393, 139329arsen, S. S. 2004a, in Space Telescope ScienceInstitute symposium series, Vol. 18, Planets toCosmology: Essential Science in the Final Yearsof the Hubble Space Telescope, ed. M. Livio, &S. Casertano, astro–ph/0408201—. 2004b, A&A, 416, 537, 537—. 2009, A&A, 494, 539, 539Larsen, S. S., & Brodie, J. P. 2000, AJ, 120, 2938,2938Leitherer, C., Schaerer, D., Goldader, J. D., et al.1999, ApJS, 123, 3, 3Liske, J., Lemon, D. J., Driver, S. P., Cross,N. J. G., & Couch, W. J. 2003, MNRAS, 344,307, 307Mackey, A. D., & Gilmore, G. F. 2003a, MNRAS,338, 120, 120—. 2003b, MNRAS, 338, 85, 85Ma´ız Apell´aniz, J., & ´Ubeda, L. 2005, ApJ, 629,873, 873Marconi, A., Axon, D. J., Capetti, A., et al. 2003,ApJ, 586, 868, 868Martig, M., & Bournaud, F. 2008, MNRAS, 385,L38, L38Matthews, L. D., & Gallagher, III, J. S. 1997, AJ,114, 1899, 1899Mayya, Y. D., Romano, R., Rodr´ıguez-Merino,L. H., et al. 2008, ApJ, 679, 404, 404Meurer, G. R., Heckman, T. M., Leitherer, C.,et al. 1995, AJ, 110, 2665, 2665Miller, B. W., Whitmore, B. C., Schweizer, F., &Fall, S. M. 1997, AJ, 114, 2381, 2381Mullan, B., Konstantopoulos, I. S., Kepley, A. A.,et al. 2011, ApJ, 731, 93, 93Noeske, K., Baggett, S., Bushouse, H., et al. 2012,WFC3 UVIS Charge Transfer Eciency October2009 to October 2011,Odewahn, S. C., Windhorst, R. A., Driver, S. P.,& Keel, W. C. 1996, ApJ, 472, L13, L13 ¨Ostlin, G., Zackrisson, E., Bergvall, N., &R¨onnback, J. 2003, A&A, 408, 887, 887Popescu, B., & Hanson, M. M. 2010, ApJ, 724,296, 296Portegies Zwart, S. F., McMillan, S. L. W., &Gieles, M. 2010, ARA&A, 48, 431, 431Scheepmaker, R. A., Haas, M. R., Gieles, M., et al.2007, A&A, 469, 925, 925Schlegel, D. J., Finkbeiner, D. P., & Davis, M.1998, ApJ, 500, 525, 525Schweizer, F., & Seitzer, P. 1998, AJ, 116, 2206,2206Scott, N., & Graham, A. W. 2012, ArXiv e-prints,arXiv:1205.5338Sersic, J. L. 1968,Seth, A., Ag¨ueros, M., Lee, D., & Basu-Zych, A.2008, ApJ, 678, 116, 116Seth, A., Olsen, K., Miller, B., Lotz, J., & Telford,R. 2004, AJ, 127, 798, 798Silva-Villa, E., & Larsen, S. S. 2010, A&A, 516,A10+, A10+—. 2011, A&A, 529, A25, A25Skrutskie, M. F., Cutri, R. M., Stiening, R., et al.2006, AJ, 131, 1163, 1163Smith, L. J., Bastian, N., Konstantopoulos, I. S.,et al. 2007, ApJL, 667, L145, L145Solomon, P. M., Rivolo, A. R., Barrett, J., &Yahil, A. 1987, ApJ, 319, 730, 730Springob, C. M., Haynes, M. P., Giovanelli, R., &Kent, B. R. 2005, ApJS, 160, 149, 149Springob, C. M., Masters, K. L., Haynes, M. P.,Giovanelli, R., & Marinoni, C. 2009, ApJS, 182,474, 474Taylor, V. A., Jansen, R. A., Windhorst, R. A.,Odewahn, S. C., & Hibbard, J. E. 2005, ApJ,630, 784, 784Trancho, G., Bastian, N., Miller, B. W., &Schweizer, F. 2007a, ApJ, 664, 284, 28430rancho, G., Bastian, N., Schweizer, F., & Miller,B. W. 2007b, ApJ, 658, 993, 993Tully, R. B. 1988,Tully, R. B., Rizzi, L., Shaya, E. J., et al. 2009,AJ, 138, 323, 323Ubeda, L., & Anderson, J. 2012, Study of theevolution of the ACS/WFC charge transfer ef-ficiency,van Dokkum, P. G. 2001, PASP, 113, 1420, 1420V´azquez, G. A., & Leitherer, C. 2005, ApJ, 621,695, 695Verdes-Montenegro, L., Yun, M. S., Williams,B. A., et al. 2001, A&A, 377, 812, 812V´eron-Cetty, M.-P., & V´eron, P. 2006, A&A, 455,773, 773Whitmore, B. C. 2003, in Space Telescope Sci-ence Institute symposium series, Vol. 14, ADecade of Hubble Space Telescope Science, ed.M. Livio, K. Noll, & M. Stiavelli, 153Whitmore, B. C., Chandar, R., & Fall, S. M. 2007,AJ, 133, 1067, 1067Whitmore, B. C., & Schweizer, F. 1995, AJ, 109,960, 960Willick, J. A., Courteau, S., Faber, S. M., et al.1997, ApJS, 109, 333, 333Wilman, D. J., Oemler, A., Mulchaey, J. S., et al.2009, ApJ, 692, 298, 298York, D. G., Adelman, J., Anderson, Jr., J. E.,et al. 2000, AJ, 120, 1579, 1579Zackrisson, E., Rydberg, C.-E., Schaerer, D.,¨Ostlin, G., & Tuli, M. 2011, ApJ, 740, 13, 13Zhang, Q., & Fall, S. M. 1999, ApJL, 527, L81,L81