The not so simple globular cluster ω Cen. I. Spatial distribution of the multiple stellar populations
A. Calamida, G. Strampelli, A. Rest, G. Bono, I. Ferraro, A. Saha, G. Iannicola, D. Scolnic, D. James, C. Smith, A. Zenteno
aa r X i v : . [ a s t r o - ph . GA ] F e b Draft version November 6, 2018
Preprint typeset using L A TEX style emulateapj v. 6/22/04
THE NOT SO SIMPLE GLOBULAR CLUSTER ω Cen. I. SPATIAL DISTRIBUTION OF THE MULTIPLESTELLAR POPULATIONS. A. Calamida , , G. Strampelli , A. Rest , G. Bono , , I. Ferraro , A. Saha , G. Iannicola , D. Scolnic , D.James , , C. Smith , A. Zenteno Draft version November 6, 2018
ABSTRACTWe present a multi-band photometric catalog of ≈ ≈ × ◦ across ω Cen. Photometry is based on images collected with the Dark Energy Camera onthe 4m Blanco telescope and the Advanced Camera for Surveys on the Hubble Space Telescope. Theunprecedented photometric accuracy and field coverage allowed us for the first time to investigate thespatial distribution of ω Cen multiple populations from the core to the tidal radius, confirming its verycomplex structure. We found that the frequency of blue main-sequence stars is increasing compared tored main-sequence stars starting from a distance of ≈
25’ from the cluster center. Blue main-sequencestars also show a clumpy spatial distribution, with an excess in the North-East quadrant of the clusterpointing towards the direction of the Galactic center. Stars belonging to the reddest and faintest red-giant branch also show a more extended spatial distribution in the outskirts of ω Cen, a region neverexplored before. Both these stellar sub-populations, according to spectroscopic measurements, aremore metal-rich compared to the cluster main stellar population. These findings, once confirmed,make ω Cen the only stellar system currently known where metal-rich stars have a more extendedspatial distribution compared to metal-poor stars. Kinematic and chemical abundance measurementsare now needed for stars in the external regions of ω Cen to better characterize the properties of thesesub-populations.
Subject headings: globular clusters: general — globular clusters: Omega Centauri
1. INTRODUCTION
The peculiar Galactic Globular Cluster (GGC) ω Cen (NGC 5139) has been subject to substantial obser-vational efforts covering the whole wavelength spectrumfrom the ultraviolet to the near-infrared. This gigan-tic star cluster, the most massive known in our Galaxy, M = 2 . × M ⊙ (van de Ven et al. 2006) has (at least)three separate stellar populations with a large undis-puted spread in metallicity (Norris & Da Costa 1995;Norris et al. 1996; Suntzeff & Kraft 1996; Kayser et al.2006; Villanova et al. 2007; Calamida et al. 2009; John-son & Pilachowski 2010). Table 1 summarizes the basicparameters of ω Cen.It has been suggested that ω Cen stellar populationsnot only show different chemical abundances but also Based on observations made with the Dark Energy Camera(DECam) on the 4m Blanco telescope (NOAO) under programs2014A-0327, 2015A-0151, 2016A-0189, PIs: A. Calamida, A. Rest,and on observations made with the NASA/ESA Hubble Space Tele-scope, obtained by the Space Telescope Science Institute. STScIis operated by the Association of Universities for Research in As-tronomy, Inc., under NASA contract NAS 5-26555. National Optical Astronomy Observatory - AURA, 950 NCherry Ave, Tucson, AZ, 85719, USA; [email protected] INAF - Osservatorio Astronomico di Roma - Via Frascati 33,00040, Monteporzio Catone, Rome, Italy Space Telescope Science Institute - AURA, 3700 San MartinDr., Baltimore, MD 21218, USA Dipartimento di Fisica, Universit´a di Roma Tor Vergata, Viadella Ricerca Scientifica 1, 000133, Roma, Italy The University of Chicago,The Kavli Institute for Cosmolog-ical Physics, William Eckhardt Research Center - Suite 499, 5640South Ellis Avenue, Chicago, IL 60637, USA Cerro Tololo Inter-American Observatory, Casilla 603, La Ser-ena, Chile Astronomy Department, University of Washington, Box351580, Seattle, WA 98195, USA have different kinematical properties. In particular, Nor-ris et al. (1997) matching the spectroscopic abundancesof ≈
500 red-giant (RG) stars with the radial velocities byMayor et al. (1997), observed that the metal-rich (MR)stars of ω Cen do not share the rotational velocity (V ≈ − ) of the metal-poor (MP) component. The mostMR stars also seem to have a smaller velocity dispersioncompared to the MP stars. However, these results werequestioned by Pancino et al. (2007) and Sollima et al.(2009), who found that the most MR stellar componentof ω Cen does not present any significant radial velocityoffset with respect to the bulk of stars. Moreover, thevelocity dispersion profile appears to decrease monoton-ically from σ v ≈ . σ v ≈ . ≤ r ≤ ω Cen formed in an indepen-dent stellar system later accreted by the cluster.Pancino et al. (2000, 2003), by analyzing the spatialdistribution of a sample of cluster RG stars, concludedthat the three main stellar populations of ω Cen (MP,MI, and MR) have different distributions: MP stars aredistributed along the direction of the cluster major axis(E–W), while the MI and MR along the N–S axis. Thisresult was confirmed by Hilker & Richtler (2000), based Calamida et al.on their Str¨omgren photometric metallicities for a sam-ple of ω Cen RGs. In particular, they found that themore MR stars seem to be more concentrated within aradius of 10 ′ from the cluster center. Sollima et al.(2005) also found that MR stars are more centrally con-centrated based on photometry of RG stars for a field ofview (FoV) of ≈ × ◦ across the cluster.Another peculiar property of ω Cen is the splitting ofthe main-sequence (MS). Hubble Space Telescope (HST)photometry revealed for the first time that ω Cen MSbifurcates into two main components, the so called blue-MS (bMS) and red-MS (Anderson 2002; Bedin et al.2004). Spectroscopic follow-up by Piotto et al. (2005)showed that bMs stars are more metal-rich than rMSstars, while their color being bluer. These authors thensuggested that bMS stars constitute a helium-enhancedsub-population in the cluster. The split of ω Cen MSwas also found by Sollima et al. (2007b) based on VeryLarge Telescope (VLT) photometry. They showed thatthe two sequences are still well-separated at ≈ ′ fromthe cluster center and that the bMS is more centrallyconcentrated compared to the rMS.The formation history and composition of ω Cen forma complex puzzle that is being slowly pieced togetherby investigations based on the latest generation of tele-scopes. However, all previous photometric studies werebased on catalogs covering a field of view of no morethan ≈ × ′ across ω Cen. We now push forward theongoing investigations by combining precise multi-bandDECam photometry, covering ≈ × ◦ across the cluster,and HST data for the cluster center to characterize theproperties of ω Cen multiple stellar populations from thecore to the tidal radius.The structure of the current paper is as follows. In § ω Cen. In § § § § § ω Cen stellar populations and we summarizeand discuss the results in §
2. OBSERVATIONS AND DATA REDUCTION
Photometric data discussed in this investigation be-long to two different sets from both space and groundbased telescopes. A set of 57 ugri images centered on ω Cen was collected over 3 nights, 2014 February 24,2015 June 22 and 2016 March 4, with the Dark EnergyCamera (DECam) on the 4m Blanco Telescope (CTIO,NOAO). DECam is a wide-field imager with 62 CCDsand covers a 3 square degree sky FoV with a pixel scaleof 0.263 ′′ . Exposure times for our observations rangedfrom 120 to 600s for the u -band and from 7 to 250s forthe other filters. Weather conditions were very good forall nights with image seeing ranging from 0.8 ′′ to 1.6 ′′ forthe u -band and from 0.7 ′′ to 1.2 ′′ for the other filters.Standard stars from the Sloan Digital Sky Survey (SDSS)Stripe 82 were observed in all filters and at different airmasses for the night of February 2014. The accuracy ofthe derived zero points (ZPs) ranges between 2% for the r and i filters to 4 −
5% for the g and u filters. Formore details on the photometric calibration see Table 3and Section §
3. Table 2 lists the log of the DECam ob-servations while Fig. 1 shows the footprint of DECamphotometric catalog (black dots) for ω Cen. Note thatstars are missing at the top and bottom of the footprintsince 2 out of the 62 DECam ccds, N7 and S7, are notoperational.Photometry in the F W , F W , F N filters wascollected with the Advanced Camera for Surveys (ACS)on board the Hubble Space Telescope (HST) for a regioncovering a FoV of ≈ ′ × . ′ centered on ω Cen. Formore details about these observations see Castellani et al.(2007, hereafter CS07).DECam images were pre-reduced by using a pipelinedeveloped by one of us, Photpipe . Photpipe is a robustpipeline used by several time-domain surveys (e.g., Su-perMACHO, ESSENCE, Pan-STARRS1; see Rest et al.2005, 2014), designed to perform single-epoch image pro-cessing including image calibration. We used Photpipeto perform bias subtraction, flat-fielding, cross-talk cor-rection, geometrical distortion correction and the imageastrometric calibration.Photometry was performed with a variant of DoPHOT(Schechter et al. 1993), with a pipeline developed by oneof us for reducing DECam images along the lines de-scribed in Saha et al. (2010). DoPHOT uses an analyti-cal function as a model point-spread function (PSF) fordescribing different object types. Aperture magnitudesare determined for bright isolated stars across each of the60 operational chips of the camera. The difference be-tween the PSF and aperture magnitudes is mapped overthe field of view, accounting for chip to chip offsets, andthis provides the correction factors for DoPHOT PSFmagnitudes (see Saha et al. 2010, for details). The samereduction pipeline performs a few quality selections onthe catalog: i) stars that lie too close to the chip bor-ders, i.e. less than 20 pixels away, are excluded; ii) ifa star has a close neighbor with significant comparativebrightness is excluded; iii) stars with photometric errorsmore than 3 sigma the average photometric error for allobjects in the corresponding 0.5 magnitude bin are ex-cluded.
3. PHOTOMETRIC CALIBRATION
During the night of 2014 February 24, a Stripe82field was observed in all the four filters, namelySDSS1048p0000. We retrieved the photometry for thestars included in this field from the Sloan Digital SkySurvey (SDSS) and from the Pan-STARRS1 (PS1) cat-alogs. To transform the photometry into the DECamnatural system we followed the approach of Scolnic et al.(2015). We derived the transformations from PS1 andSDSS to the DECam system based on the photometryof 205 out of the 379 standard stars of the Next Gen-eration Stellar Library (NGSL ). These standard starsspan a color range − . . g − i . . . g .
12 mag (for more detailson how these stars were selected see Scolnic et al. 2015).Fig. 2 shows the comparison of the SDSS and DECam r -band magnitudes of the selected standard stars versus https://confluence.stsci.edu/display/photpipearmin/Photpipe https://archive.stsci.edu/prepds/stisngsl/ he not so simple globular cluster ω Cen 3
Fig. 1.—
Field of view covered by DECam photometric cata-log (black dots) and by the ACS catalog (red) across the globularcluster ω Cen. The orientation is labeled in the figure. their SDSS g − i color. To convert the magnitudes fromthe SDSS to the DECam system, and later from PS1 toDECam, we used an iterative process. We first selecteda sample of stars by estimating the mean magnitude dif-ference for each 0.15 mag color bin and kept only starswith a difference ≤ σ . The figure shows the selected(blue dots) and the excluded (red) standard stars in the r SDSS − r DECam vs g SDSS − i SDSS plane. Another 1 σ selection was applied by using the fitted spline, and anew fit was performed. The final fitting spline is shownin the figure as a green solid line. This spline was thenused to transform the SDSS and PS1 photometry for thefield SDSS1048p0000 into the DECam natural system.Two different set of zero points (ZPs) were estimated byusing the two photometries and they are listed in Table 3together with their root mean square values (note thatthe ZP for the u filter was derived exclusively by usingthe SDSS photometry). The ZPs derived by using theSDSS and the PS1 photometry agree at the 2% level.The instrumental magnitudes of the ω Cen catalog arecorrected for aperture by selecting a few bright isolatedstars for each DECam ccd. The aperture correction isestimated accounting for chip to chip offsets and mappingthe entire camera FoV. The best seeing image is selectedas a reference for each filter, and the photometry of theother images is rescaled to this one, after bringing all theexposures to 1 second.We then calibrated the mean instrumental magnitudesby following the equation: M i = m i − K i × A i + ZP i (1)where M i and m i are the calibrated and instrumen-tal magnitudes, respectively, K i are the extinction co- −1.0 −0.5 0.0 0.5 1.0 1.5 2.0 2.5 g SDSS - i
SDSS −0.050.000.050.100.15 r S D SS - r D E C a m Selected N=192Excluded N= 10Spline fit
Fig. 2.—
Selected stars from the Next Generation Stellar Libraryin the
SDSS r − DECam r vs SDSS g − i color plane. Sigma clippingselected (blue dots) and excluded (red) stars are shown. The splinefit is over-plotted as a green solid line. efficients for the different filters, A i the air masses ofthe reference observations and ZP i the zero points. Weused the extinction coefficients obtained by the DECamLegacy Survey (DECaLS) collaboration (Li et al. 2016)for the g, r, i filters, namely K g = 0.18, K r = 0.0875 and K i = 0.065. For the u filter we used the value K u = 0.40obtained from DECam multiple u − band observations ofthe Galactic bulge. The air masses of the reference obser-vations are A u =1.19, A g =1.15, A r =1.19 and A i =1.19.As ZPs we used the average of the ZPs derived by usingthe PS1 and SDSS photometry, namely ZP u = − ZP g = − ZP r = − ZP i = − gri filters, while is ≈
5% for the u filter.The calibrated photometric catalog for ω Cen includes686 ,
746 stars with a measurement in the r and i filters,598 ,
429 with a measurement in the g filter, and 398 , u filter. Fig. 3 shows the i, u − i (left panel), i, g − i (middle) and i, r − i (right)color-magnitude diagrams (CMDs) for all stars observedin the FoV towards ω Cen. The Signal to Noise ratio(S/N) is ≥
20 down to u ≈
23 mag, g ≈
23 mag, r ≈ i ≈ FoV around ω Cenof Marconi et al. 2014 and their Figure 11).The ACS photometry was kept in the VEGA systemand we applied the camera charge transfer efficiency cor-rection and the available ZPs for the F W , F W , F N filters following the prescriptions by Sirianniet al. (2005). For more details on the photometric cali- Calamida et al.bration of this catalog see CS07.
4. A CLEAN SAMPLE OF CLUSTER STARS
To separate field and cluster stars we adopted a sim-ilar approach as suggested by Di Cecco et al. (2015).To take advantage of the multi-band optical photometryavailable for globular clusters they estimated the clusterridge lines using different CMDs based on the same mag-nitude ( r ) and different colors. To improve the precisionof the cluster ridge lines the candidate cluster stars wereselected according to their radial distance and to theirphotometric errors. Once the multiple ridge lines wereestimated they generated a multi-dimensional CMD andthe candidate cluster stars were selected using a vari-able σ -clipping over the entire magnitude range. Thisapproach is seen to be quite robust, since they were ableto separate candidate field and cluster stars for M 71, ametal-rich globular projected onto the Galactic bulge.However, the quoted method can be hardly adopted ina globular cluster like ω Cen, due to the presence of well-defined multiple sequences mainly caused by a differencein metal content (Calamida et al. 2009; Johnson & Pila-chowski 2010). Therefore, the separation was performedusing a new improved approach. We estimated the ridgelines of the different sub-populations identified along thecluster red-giant branch (RGB), the main sequence turn-off (MSTO) and the main sequence (MS). The horizon-tal branch (HB) stars were not included since they aretypically bluer than field stars. These ridge lines wereestimated neglecting the stars located in the innermostcluster regions ( r ≤ ′ ) and applying several cuts in radialdistance and in photometric accuracy. Note that to fullyexploit the current photometric catalog we only selectedstars with accurate measurements in all the four ugri bands. Once the ridge lines (seven) have been estimatedwe performed a linear interpolation among them and gen-erated a continuos multi-dimensional surface. Finally, weused two different statistical parameters to separate fieldand cluster stars:1) we estimated the cumulative standard deviationamong the position of individual stars and the referencesurface;2) we associated a figure of merit to the distance inmagnitude and colors among the individual stars andthe reference surface.We have performed a number of test and trials tosharpen the selection criteria to separate cluster andfield stars. The approach was conservative, in the sensethat we preferred to possibly lose some of the candidatecluster stars instead of including possible candidate fieldstars. A glance at Fig. 4 shows the advantages of thecurrent approach. The left panel displays the color-color-magnitude diagram, u − r vs g − i vs r , for candidate field(gray dots) and cluster (multi-color) stars seen from thefront, while the right panel shows the same plot but seenfrom the back. Note that the candidate field stars dis-play a smooth distribution both in magnitude and incolor. In passing we note that the current approach canbe applied to separate field and cluster stars thanks tothe opportunity to use the u filter, since this band al-lows a better sensitivity to both effective temperatureand metallicity. Fig. 5 shows the candidate field stars inthe i, u − r CMD. No clear cluster sequence is present inthe field star sample in the entire magnitude range down
TABLE 1Positional, photometric and structuralparameters of the Galactic GlobularCluster ω Cen
Parameter Ref. a α (J2000) 201.694625 1 δ (J2000) -47.48330 1 M V (mag) a -10.3 3 r c (arcmin) b r t (arcmin) c e d σ V (km s − ) e ± . f . ± .
02 6( m − M ) (mag) g . ± . ± .
03 1 a References: 1) Braga et al. (2016); 2) Harris(1996); 3) Trager, King & Djorgovski (1995); 4)Geyer, Nelles & Hopp (1983); 5) Merrit, Meylan& Mayor (1997); 6) Calamida et al. (2005) a TotalVisual magnitude. b Core radius. c Tidal radius. d Eccentricity. e Stellar central velocity, disper-sion. f Reddening. g True distance modulus. to i ≈
23 mag. It is interesting to note the sequenceof disk white dwarfs at -0.5 ≤ u − r ≤ i ≥
18 mag that were excluded from the cluster sample afterapplying our selection method.
5. THE CLUSTER COLOR-MAGNITUDE DIAGRAMS
Following the procedure described in section § ugri filters. The u -bandphotometry is limiting the depth of the catalog, but itwas essential in allowing the cluster and field star separa-tion. Fig. 6 shows the same CMDs of Fig. 3 but for clus-ter members only. No selection in photometric accuracyis applied. All the cluster sequences are well-defined, in-cluding the extreme horizontal branch (EHB) at u − i ≈ -1, g − i ≈ -0.7, r − i ≈ -0.3 and 18 < i <
20 mag, andthe white dwarf (WD) cooling sequence at u − i ≈ -1, g − i ≈ -0.7, r − i ≈ -0.3 and i <
21 mag. The CMDsreach i ≈ S/N > . Thesemodels are in the Sloan photometric system (Fukugitaet al. 1996) and were transformed to the DECam systemby using the empirical transformations derived in section §
3. Extinction coefficients in the ugri filters were esti-mated by using the Cardelli et al. (1989) reddening lawand DECam filter transmission functions. We obtained A u = 1.70 × A V , A g = 1.18 × A V , A r = 0.84 × A V , and A i = 0.63 × A V . We used an absolute distance modulusof µ = 13.71 ± ± E ( B − V ) = 0.11 ± t = 12 Gyr, and different metallicities, namely Z =0.004, Y = 0.251, and Z = 0.0006, Y = 0.246. Thesevalues, -1.84 ≤ [ F e/H ] ≤ -1.01, approximately bracketthe bulk of ω Cen metallicity dispersion (Calamida et al.2009). The agreement between theory and observations http://albione.oa-teramo.inaf.it he not so simple globular cluster ω Cen 5
TABLE 2Log of the observations collected with DECam on the 4m Blanco telescope for ω Cen (CTIO, NOAO, proposal IDs:2014A-0327, 2015A-0151, 2016A-0189, PIs: A. Calamida, A. Rest).
Name Exposure time Filter RA DEC Seeing(s) (hh:mm:ss.s) (dd:mm:ss.s) (arcsec)February 24, 2014omegacen.u.ut140224.052814.fits 120 u 13:26:47.288 -47:28:45.894 1.2omegacen.u.ut140224.053340.fits 120 u 13:27:00.889 -47:33:26.593 1.3omegacen.u.ut140224.053907.fits 120 u 13:26:33.338 -47:31:08.396 1.2omegacen.u.ut140224.054435.fits 120 u 13:26:13.607 -47:34:28.294 1.6June 22, 2015omegacen.g.ut150622.035733.fits 250 g 13:26:47.047 -47:28:45.995 1.2omegacen.g.ut150622.040213.fits 250 g 13:26:27.319 -47:28:46.196 1.2omegacen.g.ut150622.040649.fits 250 g 13:26:27.337 -47:32:06.295 1.2omegacen.g.ut150622.041130.fits 250 g 13:26:47.058 -47:32:06.194 1.1March 4, 2016omegacen.r.ut160304.072010.fits 80 r 13:26:48.138 -47:27:41.994 0.8omegacen.r.ut160304.072202.fits 80 r 13:26:40.258 -47:27:41.695 0.8omegacen.r.ut160304.072350.fits 80 r 13:26:40.229 -47:26:21.494 0.8omegacen.r.ut160304.072538.fits 80 r 13:26:48.178 -47:26:21.793 0.8
This table is available in its entirety in a machine-readable form in the online journal.
TABLE 3Zero points derived for Stripe 82 field SDSS1048p0000 by using SDSS and PS1 photometry.
System u g r i rms u rms g rms r rms i PS1 . . . -5.466 ± ± ± ± ± ± ± TABLE 4Values of the peaks and Full-Width half maximum (FW)for the three Gaussians that fit the three sub-populationsalong ω Cen MS: P for the red MS, P for the MS-a andP for the blue MS. See text for more details. ∆ Mag P FW P FW P FW ≤ r < ≤ i < ≤ i <
19 0.00 0.09 0.04 0.21 . . . . . .19 ≤ i < ≤ i <
20 0.00 0.12 0.03 0.30 -0.10 0.0520 ≤ i < ≤ i <
21 0.01 0. 21 0.20 0.09 -0.18 0.1610’ ≤ r < ≤ i < ≤ i <
19 0.00 0.05 0.02 0.13 . . . . . .19 ≤ i < ≤ i <
20 0.00 0.09 0.01 0.20 -0.09 0.0620 ≤ i < ≤ i <
21 0.00 0.09 -0.02 0.27 -0.15 0.0715’ ≤ r < ≤ i < ≤ i <
19 0.00 0.07 0.06 0.16 . . . . . .19 ≤ i < ≤ i <
20 0.00 0.05 0.03 0.09 -0.08 0.0920 ≤ i < ≤ i <
21 0.00 0.09 0.06 0.16 -0.13 0.14 is quite good in the entire magnitude range in all thethree CMDs. The HB in the i, u − i CMD is slightlybluer than the ZAHBs. This effect might be due to thecalibration uncertainties, ≈ u -band bolometric correction of the models.
6. ASTROMETRY AND COORDINATE SYSTEM
The astrometric calibration of ω Cen DECam catalogto the equatorial system J2000 was performed by usingPhotpipe and the Two Micron All Sky Survey (Cutriet al. 2003) catalog of stars as a reference. The finalaccuracy is better than 0.03” in both right ascension anddeclination.The astrometry of the ACS catalog was performed bymatching the photometry with stars from the catalogof van Leeuwen et al. (2000) with proper motions andmembership probabilities (for more details see CS07).We matched the ACS and DECam photometric cat-alogs for cluster members by using a 0.5” searchingradius. The matched catalog includes 1,722,810 starscovering a FoV of 2.3 × ◦ across ω Cen and includ-ing photometry in seven photometric bands, namely F W, F
W, F
N, u, g, r, i (see Fig. 1). To ourknowledge, this is the largest multi-band data set evercollected for a Galactic globular cluster after our ACS-WFI catalog published in CS07.The equatorial coordinates α and δ in degrees wereconverted to cartesian coordinates by following theprescriptions of van de Ven et al. (2006) with the clustercenter at α = 201.694625 ◦ and δ = -47.48330 ◦ (Bragaet al. 2016). Setting x in the direction of West and y inthe direction of North: x = − r · cosδ · sin ∆ αy = r · ( sinδ · cosδ − cosδ · sinδ · cos ∆ α )where ∆ α = α − α and r = 10 , /π to have x and y in arcminutes. Calamida et al. Fig. 3.—
DECam ugri color-magnitude diagrams towards ω Cen. Error bars are marked.
We then projected the cartesian coordinates x and y with the x and y axes aligned with the observed majorand minor axes of ω Cen, respectively. To accomplishthis we rotated the coordinates by the position angle ofthe cluster, defined as the angle between the major axisand the North direction measured counterclockwise byusing a value of 100 ◦ (van de Ven et al. 2006). Thecombined ACS-DECam photometric catalog with coor-dinates aligned on ω Cen major and minor axis will allowus to investigate the behavior of the cluster different sub-populations as a function of distance from the center.
7. THE MAIN-SEQUENCE SPLIT
Based on HST photometry Anderson (2002) before,and later Bedin et al. (2004), revealed that ω Cen MS isbifurcating into two main components, the so called blue-MS (bMS) and the red-MS (rMS). The color differencebetween the two sequences changes with magnitude, andthey are clearly separated in the magnitude interval 20.5 < V <
22. The HST observations included a central fieldand a field located at ≈ ′ from the cluster center. Aspectroscopic follow-up by Piotto et al. (2005) found thatbMS stars are more metal-rich than rMS stars counter to expectations. These authors proposed that bMS starsconstitute a helium-enhanced sub-population in the clus-ter to explain the observed anomaly. ω Cen MS split wasalso found by Sollima et al. (2007b) based on VLT pho-tometry. They showed that the two sequences are stillwell-separated at ≈ ′ from the cluster center and thatbMS stars are more centrally concentrated compared torMS stars. The ratio of bMS and rMS stars decreasesfrom a value of ≈ ≈ ′ down to 0.15 for distances larger than19 ′ . Bellini et al. (2010), by using deep HST observa-tions collected in different filters from the ultraviolet tothe red, showed the presence of a third MS, named MS-a,that better separates in the F W, F W − F W CMD, and seems to be connected to ω Cen faintest sub-giant branch (SGB), the so called SGB-a (Ferraro et al.2004), and the reddest and most metal-rich RGB, theso-called RGB-a Pancino et al. (2000), and named ω ′ (HST and VLT data). DECam photometryand the ability to remove the field component by usingcolor-color-magnitude diagrams opens up the possibilityhe not so simple globular cluster ω Cen 7to investigate the spatial distribution of the two MSs un-til ω Cen tidal radius.The left panel of Fig. 7 shows a zoom of DECam i, g − i CMD in the magnitude interval 18.0 ≤ i ≤ ≤ i ≤ ω Cen MS split is observedwith a 4m-class ground-based telescope. The g − i colordistance between the two sequences is changing withmagnitude and reaches a maximum of ≈ -0.18 mag at i ≈
21 mag. Unfortunately, our catalog does not havethe sufficient photometric accuracy to allow us to in-vestigate the behavior of the MS at fainter magnitudes.The ridge line of the rMS is over-plotted on the i, g − i CMD of Fig. 7 as a red solid line. The ridge line colorat the corresponding magnitude was subtracted to eachstar observed color to straighten the MS. The result ofthis process is shown in the right panel of the figure. Wethen estimated the distance of each star from the clustercenter by using the coordinates aligned with the majorand minor axes and divided the stars in three concentricannuli from 5 to 66 ′ , including approximately the samenumber of stars per radial annulus. Stars were then di-vided in six 0.5 magnitude bins from i = 18 down to i = 21. Fig. 8 shows the star g − i observed color minusthe ridge line color histograms for the six magnitude in-tervals for the annulus in the distance interval 10 < r < ′ . The panels show that the color distributions areasymmetric, being skewed towards the red in the entiremagnitude range, and they separate in two main peaksstarting at i ≈
19 mag. The skewness is probably dueto the presence of the third MS (MS-a), that the accu-racy of the photometry and the g − i color sensitivitydoes not allow us to separate it from the rMS, and tothe presence of blends and unresolved binaries. We fit-ted the six histograms with three Gaussians reproducingthe rMS ( P ), the MS-a ( P ) and the bMS ( P ), respec-tively. The three Gaussian functions used in the fit andtheir sum are shown in the figure as red (rMS), green(MS-a), blue (bMS) and black solid lines. The peaks andthe Full-Width Half Maximum values of the Gaussiansare indicated in the plots and listed in Table 4.DECam photometry clearly shows that ω Cen MS splitis present at all distances from the cluster center until thetidal radius. The g − i color separation between the rMS( P ) and the bMS ( P ) is the same, within the uncer-tainties, for the three different annuli, going from 0 .
07 to0 .
15 mag for 10 ≤ r < ′ , according to the magnitudeinterval.7.1. The ratio of blue and red main-sequence stars
To characterize the spatial distribution of ω Cen MSstars we computed the ratio of bMS and rMS stars, r ( bM S/rM S ) = N ( bM S ) /N ( rM S ), as a function of theradial distance. To select the sample of bMS and rMSstars we first produced a 3D CMD for stars in the mag-nitude interval 19.25 < i < g − i , magnitude i , and luminosity function. To overcome subtle problemsin constraining the position of the MS peaks caused bythe binning of the data, we associated to each star aGaussian kernel with a sigma equal to its intrinsic errorin the g − i color measurement. The green surface wascomputed by summing all the individual Gaussians over the entire color and magnitude range. A glance at thesurface discloses two distinct backbones tracing the bMSand the rMS (blue and red solid lines, respectively). Tofurther improve the identification of bMS and rMS stars,the blue and the red lines display the peaks of the twosequences, while the black solid line marks the valley be-tween the two different sub-populations, i.e. the relativeminimum between the two relative maxima.The 3D CMD plotted in Fig. 9 clearly shows that theseparation between bMS and rMS stars is far from be-ing trivial, since the difference in color is magnitude de-pendent. Moreover, the MSs associated to the two sub-populations display, at fixed magnitude, different broad-enings in color. To overcome thorny problems in the se-lection criteria adopted to identify bMS and rMS stars,we adopted an incremental approach. Firstly, we onlyselected stars that are located within ± < i < g − i color from the backbone of ∆ = 0.04mag. The adopted minimum color bin was driven bythe typical color uncertainty in the selected magnituderange. We then repeated the same selection but increas-ing the distance in color from the backbone. Note thatthe bMS and the rMS samples never overlap, since theinner boundary is traced by the valley plotted in Fig. 9.To improve the statistics of the two samples, the rangein color was increased up to 0.3 mag. We performed anumber of tests and this color bin is a good compromisebetween the width in color of the bMS plus the rMS andthe contamination of field stars. It is clear that with awider bin in color, we mainly select stars that are locatedalong the slopes either of the bMS or of the rMS back-bone. As a whole we ended up with 28 bins in g − i colorranging from 0.02 to 0.3 mag.To validate the criteria adopted for the selection ofbMS and rMS stars Fig. 10 shows the quoted selection inthe i, g − i CMD. The left panel shows the selection basedon a g − i color range ∆ = 0.04 mag. Note that for thisselection the blue and the red samples are approachingthe valley in the bright regime, i < g − i color range.Fig. 11 shows the 3D plot of the ratio between bMSand rMS stars as a function of the radial distance andof the g − i color range used in the selection for the en-tire sample of stars included in the 19.25 < i < r ( bM S/rM S ) = 0 . ± . < r < ′ , in which the ra- Calamida et al.tio decreases by almost a factor of two from the half-mass radius (5 ′ ), and then it starts to steadily increase.Note that all previous investigations concerning the ra-dial trend of bMS and rMS stars reached a maximumdistance of ≈ ′ from the cluster center.2) The population ratios display two relative maximain approaching the cluster center and for radial distancesof the order of 45–50 ′ . These findings are independentof the g − i color bin used to select the sample of bMSand rMS stars and indeed the radial trends are quitesimilar when moving from the narrower to the widerbin. Moreover, the current finding is also independentof the approach used to select candidate cluster and fieldstars. The population ratios are similar in the left panelof Fig. 11, where the ratio is the entire sample of stars,and in the middle panel, where it is based on only can-didate cluster members.3) Data plotted in the middle panel of Fig. 11 furthersupport the evidence that the maximum in the popu-lation ratio is attained in the outskirts of ω Cen ( r ≈ ′ ), where bMS stars are overwhelming rMS stars, theratio being of the order of 1.2 (see Table 5). The pop-ulation ratio increase in the innermost cluster regionsonly produces a relative maximum. This trend becomes,for statistical reasons, more clear when moving from thenarrower to the wider color bins.4) The radial trends of the population ratio display asteady decrease when moving from the maximum ( r ≈ ′ ) to the truncation radius ( r ≈ ′ ) of the cluster.This decrease is affected by statistics, and indeed, thenarrower color bins are slightly noisier when comparedto the wider ones.The population ratios based on only candidate fieldstars plotted in the right panel of Fig. 11 show a trendthat at glance might appear counterintuitive, since theydisplay a well defined maximum in approaching the in-nermost cluster regions, with r ( bM S/rM S ) = 4 . ± .
15. The expected trend would have been a relativelyflat distribution, as observed for distances larger than ≈ ′ . However, this is a consequence of the fact that weare plotting the ratio between bMS and rMS stars andnot the star counts. The ratios attain similar values, butthe number of bMS and rMS stars among candidate clus-ter and field stars are significantly different. In passingwe also note that the ripples showed by the cumulativepopulation ratios are a consequence of small fluctuationsin the number of blue and red candidate field stars.To further quantify the difference in star counts be-tween cluster and field candidate stars the left panel ofFig. 12 shows the ratio of blue candidate field and clusterstars. The radial trends show that star counts of bluefield stars are at least two order of magnitude smallerthan those of blue cluster stars. This means that themaximum in the bMS and rMS star ratio in the rightpanel of Fig. 11 is caused by a non-perfect separation be-tween candidate field and cluster stars. However, the starcounts of blue field stars are at most a few hundredthsof the candidate blue cluster members. This means thatthey do not affect the current findings concerning theradial trend of the population ratio. The same outcomeapplies to the minimum and to the maximum of the pop-ulation ratios. The candidate cluster bMS stars outnum-ber the candidate blue field stars up to radial distancesof the order of 35 ′ . In these regions the two samples at- tain, within the errors, star counts log ( N field /N MS ) ≈ ≈ ′ . The decreasein the population ratio at larger distances requires in-dependent confirmation possibly based on photometriccatalogs selected using proper motion.To further support our result we plotted in Fig. 13 the i, g − i CMDs of candidate cluster (left panels) and field(right) stars for two external radial annuli, 30 ≤ r < ≤ r < ′ , respectively. The figure shows thatour method to separate cluster and field stars is veryeffective until large distances, r ≥ ′ , from the clustercenter. However, as discussed before and illustrated inthe previous figures, some ω Cen stars might be miss-classified as field stars in the more internal regions ofthe cluster. A residual contamination of field stars inthe cluster MS samples might also be present at largedistances from the center (see bottom left panel).7.2.
Star counts across the body of the cluster
The spatial distribution of ω Cen MS stars appears tobe even more complicated than suggested by the radialgradient. The density map of the ratio between bMSand rMS stars is plotted in the left panel of Fig. 14.The population ratio shows a clumpy distribution, witha well-defined North/South asymmetry in the outermostcluster regions, being the bMS stars significantly moreabundant in the Northern quadrants. Thus suggestingthat the clumping in the radial gradient might be as-sociated to azimuthal variations across the body of thecluster. It is worth noting that the main over-density ofbMS stars is pointing towards the Galactic center (GC)(Dauphole et al. 1996; Leon et al. 2000, see the arrowsplotted in the left panel of Fig. 14). These findings seemto suggest a connection between the spatial distributionof bMS stars and ω Cen dynamical evolution. A morequantitative analysis of the difference between bMS andrMS stars in these cluster regions does require new kine-matic and spectroscopic data.The anonymous referee suggested us to investigatewhether possible extinction variations across the bodyof the cluster could affect the current population ratio.To verify that our result is not affected by foregroundreddening, we downloaded reddening values provided bySchlafly & Finkbeiner (2011) for the region covered bythe current photometric catalog. The reddening colordensity map of the observed field is shown in the rightpanel of Fig. 14. The reddening in the current FoV hasa minimum value of ≈ ≈ E ( B − V ) = 0.11 magand a total dispersion of σ E ( B − V ) = 0.02 mag. Thesevalues are in very good agreement with the dispersion σ E ( B − V ) . ω Cen. This low differential reddening could movestars from the blue to the red MS sample. However, the http://irsa.ipac.caltech.edu/applications/DUST/ he not so simple globular cluster ω Cen 9
Fig. 4.— ugri
DECam color-color-magnitude diagram of ω Cen members (different colors mark different selected cluster sequences) andfield stars (gray dots) seen from the front (top panel) and the back (bottom).
Fig. 5.—
DECam i, u − r color-magnitude diagram of candidatefield stars in the observed field of view. observed over-densities of bMS stars when moving to-wards the outermost cluster regions cannot be caused byan increase in the extinction. An increase in differentialreddening causes a decrease in the population ratio, sincetruly bMS stars are moved into the rMS sample. There-fore, the current population ratio can be considered as a lower limit to the real one.On the other hand, the presence of higher dust ex-tinction ( E ( B − V ) ≈ ω Cen proper motion.7.3.
Comparison with literature
Our result is in agreement with the findings of Belliniet al. (2009, hereinafter BE09), based on HST data, fordistances 5 . r . ′ . At larger distances and up to r ≈ ′ , i.e. the region of ω Cen sampled by Bellini et al., ourratio of bMS and rMS stars is significantly lower than theratio found by the quoted authors. At r ≈ ′ , for in-stance, Bellini et al. found a ratio of 0.36 ± ± σ smaller. Onthe other hand, our findings do not agree with the studyof Sollima et al. (2007b) for distances smaller than r ≈ ′ , while they agree very well at larger distances and upto r ≈ ′ , i.e., the cluster region sampled by VLT pho-tometry. BE09 claim that Sollima et al. ratio is lowercompared to their findings due to the wider color rangethey used to select rMS stars, which would include un-resolved binaries and members of the third MS, makingthe bMS and rMS population ratio smaller. The differ-ent approaches used to select bMS and rMS stars mightalso be the origin of the difference we find between ourratios and Sollima et al. ratios at small distances fromthe cluster center.As far as the difference between our ratios and BE09values for distances larger than 10 ′ , we have to take intoaccount several circumstantial evidence. Our sample ofrMS stars is contaminated by unresolved binaries, MS-astars, and marginally by blends in these more externalcluster regions. Photometric and spectroscopic analysisprovide a binary frequency for ω Cen of ≈
5% (Mayoret al. 1996; Sollima et al. 2007a), and MS-a stars, which0 Calamida et al.
Fig. 6.—
DECam ugri color-magnitude diagrams of ω Cen cluster members. Isochrones for the same age, t = 12 Gyr, and differentmetallicities are over-plotted (see labeled values). The respective zero age horizontal branch (ZAHB) tracks are also shown. Error bars aremarked. are the counterpart of RGB-a stars, are less than 5% ofcluster stars (Pancino et al. 2000, CS07). These factorswill cause an artificial increase in the star counts of therMS, i.e. a decrease in the population ratio we are deal-ing with. By accounting for these factors, our populationratio would agree with the findings of BE09. However,these factors cannot explain the global decreasing trendof the ratio of bMS and rMS stars observed with DECamdata and not found by BE09. The number of binaries isindeed expected to decrease at increasing distances fromthe cluster center, and the MS-a stars are supposed tobe more centrally concentrated compared to metal-poorstars (Pancino et al. 2003; Bellini et al. 2009). The num-ber of these objects and blends decreases at larger dis-tances from the cluster center, with the net effect of anincrease of the bMS and rMS ratio. However, DECamdata are clearly showing that this population ratio is de-creasing from ≈ ′ up to a distance of ≈ ′ , where itattains a broad minimum value of ≈ ≤ r ≤ ′ ) and attains a value r ( bM S/rM S ) ≈ ≤ r ≤ ′ , and show a decreasing population ratiofrom a value r ( bM S/rM S ) ≈ ≈ r ( bM S/rM S ) is increasing for distances larger than ≈ ′ reaching a peak value of ≈ ≈ ′ The above empirical evidence brings forward a veryinteresting implication. Spectroscopic measurements areavailable for 17 blue MS stars and they suggest thatthe bMS is more metal-rich than the rMS (Piotto et al.2005). Our photometric analysis shows that the radialtrend of this sub-population becomes more and more rel-evant when moving towards the outskirts of the cluster.This behavior is different from what it is observed insome nearby dwarf galaxies, where more metal-rich starsare centrally concentrated when compared with the moremetal-poor ones (Bono et al. 2010; Fabrizio et al. 2015).This radial trend is further supported by metallicity gra-dients based on spectroscopic measurements of nearbydwarf galaxies (Ho et al. 2015) and on photometric in-dices (Mart´ınez-V´azquez et al. 2016), suggesting eithera relatively flat metallicity distribution or a steady de-crease when moving from the innermost to the outer-he not so simple globular cluster ω Cen 11
Fig. 7.—
Left - Zoom of the ω Cen i, g − i color-magnitude diagram. The ridge line for the rMS is over-plotted as a red solid line. Errorbars are marked. Right - Observed star g − i color minus the ridge line color for the respective i magnitude. most galaxy regions. ω Cen seems to show an oppositetrend with a population of more metal-rich stars less con-centrated compared to the more metal-poor population.The reasons for the possible difference in the metallic-ity trend between ω Cen and nearby dwarf galaxies arenot clear and new spectroscopic measurements to inferkinematical and abundance properties of a larger sampleof bMS and rMS stars at different radial distances arerequired to better characterize their nature or nurture.
8. THE SPATIAL DISTRIBUTION OF RED-GIANTBRANCH STARS
The bMS has its counterpart in one of the multipleRGBs of ω Cen, possibly at a metallicity intermediatefor the cluster range (for more details on the correspon-dence between the triple MS and the multiple sub- andred-giant branches see Bellini et al. 2010). Unfortu-nately, it is not possible to clearly separate the differ-ent intermediate RGBs without using the informationon the star chemical composition. For the more cen-tral regions of the cluster, up to ≈ ′ , low- and high-resolution spectroscopy and photometric metallicities areavailable. For the outskirts of the cluster, no metallic-ity information is available so far. Therefore, we decidedto investigate the spatial distribution of the bluest and brightest SGB/RGB, the most MP sub-population in ω Cen according to spectroscopy, corresponding to therMS. We compared the properties of this sub-populationto the ones of the faintest and reddest SGB/RGB, themost MR cluster sub-population according to spectro-scopic measurements and corresponding to the reddestMS. The reddest MS, or MS-a, is difficult to separatefrom the rMS because it overlaps with its sequence ofunresolved binaries. However, MS-a has its continua-tion on ω Cen reddest RGB, the ω ω i, u − i or u, u − i CMDs and the F W, F W − F W or F W, F W − F W CMDs, where thetemperature sensitivity is larger (see Fig. 6). Therefore,we used the u, u − i CMD, and for the internal regionsof the cluster, r . ′ , the F W, F W − F W CMD to select a sample of stars along the ω ω u, u − i and2 Calamida et al. Fig. 8.— g − i color histograms of ω Cen straightened main-sequence for six i -band magnitude intervals for the distance annulus10 < r < ′ . The three Gaussian functions and their sum used tofit the histograms are indicated as red (rMS), green (MS-a), blue(bMS) and black (total) solid lines. The Peak (P) and Full-WidthHalf Maximum (FW) values for the Gaussians are shown. F W, F W − F W CMD, respectively, and se-lected stars 0.1 mag fainter and brighter than these ridgelines. Note that the aim of this analysis is to select asample of MP stars and of stars with a metallicity inter-mediate between the MP and the ω u, u − i (top panel) and ACS F W, F W − F W CMD (bottom) with the se- . . . . . . i [ M a g ] g - i [ M a g ] . . . Fig. 9.— i, g − i CMD of ω Cen main-sequence in themagnitude interval 19.25 < i < § lected sample of MP (blue dots), MI (green) and MR(red) stars. The three samples of cluster stars have asimilar completeness since both DECam and ACS pho-tometric catalogs are complete down to the turn-off level.The total sample of MP stars includes ≈ ≈ ≈ ≈ ′ , while the fraction of MR starsincreases for larger distances until ≈ ′ . The spatialdistribution of the MI sample is different from either theMP and the MR spatial distributions starting from a dis-tance of ≈ ′ from the cluster center. The frequency ofMI stars is lower compared to MP and MR stars, fromthis distance until the tidal radius. The inset of Fig. 16shows a zoom of the radial distributions from 10 to 40 ′ .It is clear how the number of MR stars increases com-pared to the MP ones starting at ≈ ′ and then startdecreasing again at ≈ ′ , while the number of MI stars isalways lower in this distance range. For distances largerthan ≈ ′ , statistics is preventing us to fully character-ize the behavior of these sub-populations. To verify thatour method to select the sample of MI red giants did notalter the analysis we performed the following test. Wehe not so simple globular cluster ω Cen 13 g-i [Mag] g-i [Mag] g-i [Mag] i [ M a g ] Fig. 10.— i, g − i CMDs of ω Cen main-sequence in the magnitude interval 19.25 < i < g − i color limits, ∆, indicated at the top of each panel. The cyan and greensolid lines mark the blue and red main-sequence ridge lines, respectively. See the detailed explanation in § Fig. 11.—
Ratio of bMS and rMS stars as a function of distance from the cluster center and g − i color bins used to select the stars,∆. The three panels show the same ratio for all the stars (left), for only candidate cluster members (middle) and for only candidate fieldmembers (right) in the magnitude interval 19.25 < i < selected MI stars by moving the RGB ridge line 0.1 and0.2 mag brighter and fainter on both the CMDs. Wethen compared the spatial distribution of the new sam-ple of MI stars with those of the MP and MR stars. Theresult is the same within uncertainties, with the MI stars being more centrally concentrated compared to the MPand the MR stars.Fig. 17 shows the cumulative radial distributions ofMP, MI and MR stars. This plot clearly shows thatMI stars are more centrally concentrated compared to4 Calamida et al. Fig. 12.—
Logarithm of the ratio of candidate field and cluster blue (left panel) and red (right) main-sequence stars in the magnitudeinterval 19.25 < i < g − i color bins used to select the stars, ∆. the MP and the MR stars, while the MP stars are moreconcentrated than the MR stars for distances 10 . r . ′ . Moreover, the ω ′ . This result confirms the findings of CS07,where an increase of ω ω ≈
3% at r ≈ ′ to ≈
7% for larger distances (formore details please see their Table 3).The current results also agree with previousfindings by Hilker & Richtler (2000) based onStr¨omgren photometric metallicities for a sample of ω Cen RGs. These authors showed that more metal-rich RGs are more concentrated compared to moremetal-poor stars within a radius of 10 ′ from the clustercenter. Sollima et al. (2005), based on photometry of asample of RGs for a field of view of ≈ × ◦ acrossthe cluster, found a similar result. All these previousstudies were based on photometric catalogs covering aradial distance until ≈ ′ . DECam data allowed usfor the first time to analyze the spatial distributionof ω Cen RGB sub-populations across a much largerportion of the cluster and to disclose the peculiarbehavior of ω N ( M R ) /N ( M P )and N ( M I ) /N ( M P ), in the four quadrants of ω Cen,i.e. NW, NE, SW and SE. For the MR to MP star ra-tio, values are in agreement within uncertainties in thethree NW, SW and NE quadrants, while a deficiency ofMP stars is present in the SE quadrant. The MI to MPstar ratio show a clear West-East asymmetry, due to thedecrease of MP stars in the Southern quadrant of thecluster, while MI stars are more numerous in the NW, NE, and SE quadrants of ω Cen. The same table liststhe number of MP, MI and MR stars in the four differ-ent regions. A clear asymmetry is present, with the MPbeing much more numerous in the two Northern quad-rants, while MR and MI stars have a deficiency in theSW quadrant of ω Cen.Jurcsik (1998) showed that more metal-rich stars in ω Cen ([
F e/H ] ≥ − .
25) are segregated in the South-ern part of the cluster, while the metal-poor ([
F e/H ] ≤− .
75) stars in the Northern. The centroid of these twogroups are ≈ ′ apart. The segregation of MR starsin the Southern half of ω Cen was confirmed by Hilker &Richtler (2000), while they did not find an equivalent seg-regation for the MP sub-population. Our data seem tosupport a different distribution of MP, MI and MR starsin ω Cen, and a clear excess of MP stars in the Northernhalf of the cluster and a deficiency of MI and MR stars inthe SW quadrant. However, we do not have homogenousabundance measurements for MP, MI and MR stars upto ω Cen tidal radius. Spectroscopic measurements forstars belonging to the different sub-populations in theoutskirts of the cluster are now needed to better charac-terize the spatial distribution of MP, MI and MR RGBstars in ω Cen.
9. SUMMARY AND CONCLUSIONS
We presented multi-band photometry of ω Cen for atotal FoV of ≈ × ◦ across the cluster. Images were col-lected with the wide field camera DECam and combinedwith ACS data for the crowded regions of the clustercore. The availability of the u -band photometry allowedus to use a new method based on color-color-magnitudediagrams to separate cluster and field stars. We ended upwith a final photometric catalog of ≈ F W, F
W, F
N, u, g, r, i . To our knowledge, thishe not so simple globular cluster ω Cen 15
Fig. 13.— i, g − i CMDs of candidate cluster (left panels) and field (right panels) stars for two different external radial bins.
TABLE 5Star counts and population ratio for bMS and rMS candidate ω Cen and field stars.
Distance N ( rMS ) N ( NbMS ) N ( rMS ) field N ( bMS ) field r ( bMS/rMS ) r ( bMs/rMs ) field (arcmin)7.5 21128 7672 69 50 0.36 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Fig. 14.—
Left - Density color map of the ratio of bMS and rMS stars as a function of star position in arcminutes. The green arrowsmark the direction of the Galactic center (GC), ω Cen proper motion (PM), and the direction perpendicular to the Galactic center (PGP).The color scale is shown at the bottom. Right - Reddening color density map for the observed region across ω Cen as a function of starposition in arcminutes. The direction of the GC, the cluster PM and the PGP are indicated with red arrows and the color scale is shownat the bottom. he not so simple globular cluster ω Cen 17
Fig. 15.—
Top - DECam u, u − i CMD of ω Cen cluster mem-bers. The selected metal-poor (MP) red-giant branch stars (bluedots), metal-intermediate (MI, green), and metal-rich (MR, red)are over-plotted. Bottom: ACS F W, F W − F W CMDof ω Cen cluster members. Selected MP, MI and MR stars aremarked.
TABLE 6Number of metal-poor (MP) and metal-rich (MR) sub- andred-giant branch stars in the different quadrants of ω Cen. N ( MR ) /N ( MP ) N ( MI ) /N ( MP ) N ( MP ) N ( MI ) N ( MR )NW 0.34 ± ± ±
51 1504 ±
39 829 ± ± ± ±
43 1056 ±
32 636 ± ± ± ±
44 1377 ±
37 761 ± ± ± ±
40 1318 ±
36 682 ± N
10 15 20 25 30 35 40r [arcmin]0.000.050.100.150.20 N Fig. 16.—
Histograms of the spatial distribution of the metal-poor (blue solid line), metal-intermediate (green dotted-dashed),and metal-rich (red dashed) sub- and red-giant branch stars. Theinset shows a zoom of the region included in 10 < r < ′ . is the largest multi-band data set ever collected for aGalactic globular cluster and covering the widest FoVafter our ACS-WFI catalog published in CS07.DECam precise photometry allowed us to observe thesplit along ω Cen MS and to show that it is present atall distances from the cluster center. The bMS is well-separated from the rMS in the magnitude range 19.0 ≤ i ≤ g − i color distance between the twosequences is changing with magnitude and reaches a max-imum of -0.18 mag at i ≈
21 mag. The color separationbetween the rMS and the bMS is the same, within theuncertainties, at all distances from ω Cen center, rangingfrom ≈ ≈ ω Cen sub-populations across the cluster until the nominal tidal ra-dius of 57 ′ . In particular, we were able to investigate thespatial distribution of the two cluster main sequences.We found that stars belonging to the bMS are more cen-trally concentrated compared to stars belonging to therMS up to a distance of ≈ ′ . The frequency of bMSstars is then steadily increasing up to ≈ ′ , with theratio of bMS and rMS stars being larger than 1. Theratio of bMS to rMS stars shows an asymmetric clumpydistribution across the cluster, with an excess of bMSstars in the Northern half. The over-density of bMS starsin the cluster North-East quadrant is pointing towardsthe Galactic center, suggesting a connection between thespatial distribution of bMS stars and ω Cen dynamicalevolution.8 Calamida et al.Unfortunately, our photometry does not allow us toidentify the continuation of the bMS on the sub- andred-giant branch phases. We then analyzed the spatialdistribution of a sample of MP and MI stars selectedalong the sub- and red-giant branch by using ridge linesin the u, u − i and F W, F W − F W CMDs.Moreover, stars belonging to the ω . r . ′ . Data clearly show that the ω ′ . This result confirms thefindings of CS07, where an increase of the number of ω ω Cen is present.Moreover, MP stars are more numerous in the Northernhalf of the cluster.Stellar populations with different metallicities and ageshow different spatial distributions with the more metal-rich sub-populations being more centrally concentratedin some nearby dwarf spheroidal galaxies such as Carina,Sculptor, Fornax (Monelli et al. 2003; del Pino et al.2013; Fabrizio et al. 2015; Ho et al. 2015; Mart´ınez-V´azquez et al. 2016). The same behavior is observedin a Galactic globular cluster presenting a significantspread in iron abundance, Terzan 5. The metal-richsub-population in this cluster is more centrally concen-trated compared to the metal-poor one (Ferraro et al.2016). The case of ω Cen seems to be different from bothknown Local Group dwarf spheroidal galaxies and clus-ters presenting stellar populations with different metal-licities. ω Cen MI selected RGs are indeed more cen-trally concentrated compared to the MP ones, but the ω • ω Cen hosts a metal-intermediate RGB sub-population that behaves as the metal-intermediateand metal-rich stellar populations in some nearbydwarf spheroidal galaxies and Terzan 5, beingmore centrally concentrated compared to the clus-ter metal-poor RGB sub-population. This stellarsub-population is peculiar compared to the typicalsecond generation of stars present in other glob-ular clusters, presenting not only a different ironabundance and its own chemical anti-correlations,according to spectroscopic analyses (Gratton et al.2011; Marino et al. 2011, 2012), but also a verydifferent spatial distribution; • ω Cen bMS stars are more centrally concentratedcompared to rMS stars up to a distance of ≈ ′ . The frequency of bMS stars, supposedly moremetal-rich than the rMS stars according to spectro-scopic measurements, steadily increases at largerdistances, outnumbering the rMS stars until ap-proximately the cluster tidal radius. Their spatial distribution is asymmetric and clumpy, with an ex-cess of bMS stars in the direction of the Galacticcenter; • ω Cen hosts a metal-rich sub-population makingup the cluster third MS, MS-a, and the reddestand faintest RGB, the ω ≈ ′ fromthe cluster center, and their frequency increases atlarger distances, showing a more extended spatialdistribution; • The behavior of the bMS and the ω ω Cen. These results, if confirmed, willmake ω Cen the only stellar system known to have moremetal-rich stars with a more extended spatial distribu-tion compared to more metal-poor stars.Further data are now needed to solve the ω Cen puzzle.The current photometry combined with abundance andradial velocity measurements for stars of the differentstellar sub-populations across the entire cluster will allowus to better understand the origin of ω Cen.Further homogeneous photometric data are alsoneeded to better characterize the behavior of the differ-ent stellar sub-populations until and beyond the nominaltidal radius of 57 ′ . We have an approved DECam pro-posal to observe an area around ω Cen beyond its nom-inal tidal radius. The new data will allow us to bettercharacterize the spatial distribution of ω Cen differentstellar sub-populations. In particular, we are interestedin confirming the more extended spatial distribution ofthe bMS and the ω × ◦ across thecluster and including the u -filter, we will be also able toconfirm the presence of stellar over-densities tracing tidaltails around ω Cen, previously found by Marconi et al.(2014), and to detect new stellar debris if present.We thank the anonymous referee for very helpful sug-gestions which led to an improved version of the paper.he not so simple globular cluster ω Cen 19
Fig. 17.—
Cumulative radial distributions of metal-poor (bluesolid line), metal-intermediate (green dashed-dotted), and metal-rich (red dashed) sub- and red-giant branch stars.0 Calamida et al.