The Hubble Space Telescope UV Legacy Survey of Galactic Globular Clusters. II. The seven stellar populations of NGC7089 (M2)
A. P. Milone, A. F. Marino, G. Piotto, L. R. Bedin, J. Anderson, A. Renzini, I. R. King, A. Bellini, T. M. Brown, S. Cassisi, F. D'Antona, H. Jerjen, D. Nardiello, M. Salaris, R. P. van der Marel, E. Vesperini, D. Yong, A. Aparicio, A. Sarajedini, M. Zoccali
aa r X i v : . [ a s t r o - ph . S R ] N ov Mon. Not. R. Astron. Soc. , 1– ?? (2013) Printed 10 July 2018 (MN L A TEX style file v2.2)
The
Hubble Space Telescope
UV Legacy Survey of Galactic GlobularClusters. II. The seven stellar populations of NGC 7089 (M 2) ⋆ A. P. Milone , A. F. Marino , G. Piotto , , L. R. Bedin , J. Anderson , A. Renzini ,I. R. King , A. Bellini , T. M. Brown , S. Cassisi , F. D’Antona , H. Jerjen , D. Nardiello , ,M. Salaris , R. P. van der Marel , E. Vesperini , D. Yong , A. Aparicio , ,A. Sarajedini , M. Zoccali , Research School of Astronomy & Astrophysics, Australian National University, Mt Stromlo Observatory, via Cotter Rd, Weston, ACT 2611, Australia Dipartimento di Fisica e Astronomia “Galileo Galilei”, Univ. di Padova, Vicolo dell’Osservatorio 3, Padova, IT-35122 Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5, Padova, IT-35122 Space Telescope Science Institute, 3800 San Martin Drive, Baltimore, MD 21218, USA Department of Astronomy, University of Washington, Box 351580, Seattle, WA 98195-1580 Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Teramo, Via Mentore Maggini s.n.c., I-64100 Teramo, Italy Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Roma, Via Frascati 33, I-00040 Monteporzio Catone, Roma, Italy Astrophysics Research Institute, Liverpool John Moores University, Liverpool Science Park, IC2 Building, 146 Brownlow Hill, Liverpool L3 5RF, UK Department of Astronomy, Indiana University, Bloomington, IN 47405, USA Instituto de Astrof`ısica de Canarias, E-38200 La Laguna, Tenerife, Canary Islands, Spain Department of Astrophysics, University of La Laguna, E-38200 La Laguna, Tenerife, Canary Islands, Spain Department of Astronomy, University of Florida, 211 Bryant Space Science Center, Gainesville, FL 32611, USA Universidad Cat`olica de Chile, Departamento de Astronom`ıa y Astrof`ısica, Casilla 306, Santiago 22, Chile Millennium Institute of Astrophysics, Av Vicuna Mackenna 4860, Macul, Santiago, Chile
Draft Version Nov, 7, 2014
ABSTRACT
We present high-precision multi-band photometry for the globular cluster (GC) M 2. We com-bine the analysis of the photometric data obtained from the
Hubble Space Telescope
UVLegacy Survey of Galactic GCs GO-13297, with chemical abundances by Yong et al. (2014),and compare the photometry with models in order to analyze the multiple stellar sequenceswe identified in the color-magnitude diagram (CMD). We find three main stellar components,composed of metal-poor, metal-intermediate, and metal-rich stars (hereafter referred to aspopulation A, B, and C, respectively). The components A and B include stars with di ff er-ent s -process element abundances. They host six sub-populations with di ff erent light-elementabundances, and exhibit an internal variation in helium up to ∆ Y ∼ ff erent metallicities,M 2 contains a third stellar component, C, which shows neither evidence for sub-populationsnor an internal spread in light-elements. Population C does not exhibit the typical photomet-ric signatures that are associated with abundance variations of light elements produced byhydrogen burning at hot temperatures. We compare M 2 with other GCs with intrinsic heavy-element variations and conclude that M 2 resembles M 22, but it includes an additional stellarcomponent that makes it more similar to the central region of the Sagittarius galaxy, whichhosts a GC (M54) and the nucleus of the Sagittarius galaxy itself. Key words: globular clusters: individual: NGC 7089 (M 2) — stars: Population II
The
Hubble Space Telescope ( HST ) “Legacy Survey of Galac-tic Globular Clusters: Shedding UV Light on Their Populationsand Formation” is designed to image 47 Galactic Globular Clus-ters (GCs) through the filters F275W, F336W and F438W of theultraviolet and visual channel (UVIS) of the Wide Field Cam-era 3 (WFC3) on board of
HST (GO-13297, PI. G. Piotto, see Pi- otto et al. 2014 — paper I hereafter — for details). GO-13297 willcomplement the F606W and F814W database from the AdvancedCamera for Survey (ACS) GC Treasury program (GO-10775, PI.A. Sarajedini, see Sarajedini et al. 2007) to provide homogeneousand accurate photometry of GCs in five bands, from ∼
275 to ∼ c (cid:13) A. P. Milone et al. shown a high sensitivity to abundance variations including light-element and helium (see Milone et al. 2012a, 2013 and Paper I fordetails).In the present paper we focus on NGC 7089 (M 2), which isone of the few GCs that exhibits a split sub giant branch (SGB)in the visual color-magnitude diagram (CMD, see the m F606W vs. m F606W − m F814W
CMD in Piotto et al. 2012). The U vs. U − V CMD of M 2 shows a poorly-populated red-giant branch (RGB) onthe redward side of the main RGB (Grundahl et al. 1999; Lardoet al. 2012). Spectroscopy has revealed that stars in the two RGBshave di ff erent abundances in terms of their overall metallicity(Yong et al. 2014, hereafter Y14) and in terms of their s -processelements, with the red-RGB being also s -rich (Lardo et al. 2013;Y14). More specifically, Y14 have shown that the metallicity distri-bution of M 2 stars has three peaks, around [Fe / H] ∼ − − − ff erence in iron abundance ( ∆ [Fe / H] & The other presently knownmembers of this short list are: ω Centauri, M 22, Terzan 5, M 54,and NGC 5824 (e.g. Norris & Da Costa 1995; Johnson & Pila-chowski 2010; Marino et al. 2009, 2011a,b; Da Costa et al. 2009,2014; Ferraro et al. 2009; Carretta et al. 2010a,b). NGC 1851 is anintriguing candidate as it exhibits two main stellar populations withdistinct s -element abundance and a di ff erence in [Fe / H] of ∼ m F275W vs. m F275W − m F814W
CMD of M 2is discussed in Sect. 3, while in Sect. 4 and Sect. 5 we identify thestellar populations along the RGB and the MS, respectively, and in-fer their abundance of iron and light elements. We estimate the ageand the helium content of multiple stellar populations in Sect. 6 andcompare M 2 with other anomalous GCs in Sect. 7. Summary andconclusions are given in Sect. 8.
In order to investigate multiple stellar populations in M 2 we haveused images taken with ACS / WFC and WFC3 / UVIS on board the
HST . The ACS / WFC dataset consists of 5 × s long exposuresin F606W and F814W plus one 20 s short exposure taken througheach of the same filters. These images were taken on April, 16,2006 as part of the ACS HST
Treasury Survey of GCs (GO 10775,PI. A. Sarajedini). The astro-photometric catalog of stars in theACS / WFC field used in this paper has been published by Ander-son et al. (2008).The WFC3 / UVIS dataset includes 6 × ∼ s exposures inF275W, 6 × ∼ s in F336W, and 3 × ∼ s in F438W. These im-ages were taken on August, 14, August, 29, and October, 18, 2013 Small star-to-star iron variations, at the level of . as part of GO-13297. These data were corrected for charge trans-fer deficiencies by using an algorithm that was developed specif-ically for UVIS and is based on the method and the softwareof Anderson & Bedin (2010). The reduction has been performedwith img2xym WFC3, software developed by Bellini et al. (2010)mostly based on img2xym WFI (Anderson et al. 2006). Photome-try and astrometry have been independently carried out for each ex-posure by using a set of spatially-variable empirical PSFs. We havecorrected the stellar positions for geometrical-distortion by usingthe solution from Bellini et al. (2009, 2011) and have calibratedthe photometry into the VEGA-mag flight system as in Bedin etal. (2005). Stellar proper motions have been obtained as in Ander-son & King (2003) by comparing the average stellar positions mea-sured from GO-10775 and from GO-13297 data. Our data cover atemporal baseline of 7.5 yr. Since the cluster has been centered inboth the ACS / WFC and WFC3 / UVIS fields of view, five-band pho-tometry and proper motions are available for stars in a ∼ × ff erent stellar populations is typically small, the study of multiplepopulations along the CMD of any GC requires very accurate pho-tometry. Because of this, we have selected the best-measured starsin our sample, by following the same selection criteria as describedin Milone et al. (2009), which are based on several diagnostics, in-cluding the amount of scattered light from neighboring stars, PSF-fit residuals, and rms scatter in position measurements. Photometryhas been corrected for di ff erential reddening by using the methodexplained in detail in Milone et al. (2012a). Briefly, we identifiedfor each star the 55 nearest well-measured main-sequence (MS)stars and determined the color distance from the MS ridge linealong the reddening direction. The median color distance of the55 neighbors has been assumed as the best estimate of di ff eren-tial reddening for each target star, and has been applied as a cor-rection to its color. The reddening in the direction of NGC 7089is E(B − V) ∼ ∆ E(B − V) ∼ − F814W CMD
The left panel of Fig. 1 shows the complete m F275W vs. m F275W − m F814W
CMD of M 2 members (black dots), and field stars (graycrosses), selected on the basis of their proper motions. The vector-point diagram (VPD) of proper motions in WFC3 / UVIS pixel unitsper year is plotted in the inset and includes all the stars plotted inthe CMD. Since we have used M 2 members as reference stars tocalculate proper motions, the bulk of stars around the origin of theVPD is mostly made of clusters members while field stars havelarger proper motions. We have thus drawn a red circle to separatethe probable cluster members from field stars. In this paper we an-alyze cluster members only. To identify them, we have chosen aradius of 0.03 pixel / yr that corresponds to five times the averageproper-motion dispersion along the X and Y direction for the bulkof stars around the origin of the VPD. There are 47 candidate-fieldstars with proper motions larger than 0.03 pixel / yr. This samplelikely includes also cluster stars with large proper motions indeed,from the Galactic model by Girardi et al. (2005) we expect less than20 field stars in a 2.7 × c (cid:13) , 1– ?? timodal horizontal branch (HB). The most surprising feature is apoorly-populated sequence, which runs on the red side of the ma-jority of the stars in M 2 and can be followed continuously fromthe lower part of the MS to the SGB, up to the RGB tip. The wideseparation from the bulk of M 2 stars is an unusual feature, as mul-tiple MSs and RGBs in ‘normal’ GCs usually merge around theSGB. Furthermore, in GCs, the color separation between multipleMSs typically increases when moving from the MS turn o ff towardsfainter magnitudes, in contrast with what is observed for the tinyreddest MS in M 2.An additional, sparsely-populated, SGB is clearly visible, be-tween the main SGB and the faintest SGB. The three SGBs aremore clearly seen in the upper-right panel of Fig. 1, which is azoom of the left-panel CMD around the SGB.In order to explore the origin of the multiple sequences ob-served on the m F275W vs. m F275W − m F814W
CMD we took advan-tage of the high-resolution spectroscopic data recently publishedby Y14. This study shows that red giants in M 2 exhibit a mul-timodal abundance distribution for iron and for those neutron-capture elements that are associated with s -processes in solar sys-tem material. Specifically there is a large metallicity variation, withthree groups of metal poor ([Fe / H] = ∼ − / H] = ∼ − / H] = ∼ − ∼ / Fe] vs. [Fe / H] fromY14, with black, red, and aqua symbols representing their metal-poor, metal-intermediate and metal-rich stars, respectively. Theten stars for which
HST
F275W and F814W photometry is avail-able are superimposed on the m F275W vs. m F275W − m F814W
CMD.The three metallicity groups populate di ff erent RGBs: the bright,the middle, and the faint RGB correspond to metal-poor, metal-intermediate, and metal-rich populations of Y14, respectively. A number of recent studies (see Paper I and references therein) haveshown that the RGB, SGB, and the MS of GCs can often be sep-arated into distinct sequences of stars, and that appropriate com-binations of F275W, F336W, F438W, and F814W magnitudes arepowerful tools for identifying these multiple populations.The e ffi ciency of these filters in separating di ff erent stellarpopulations is closely connected to the chemical properties of GCsub-populations. The fact that the F275W filter includes the OHmolecular band, F336W the NH band, and F438W the CH and CNbands make them very sensitive to the e ff ect of molecular bandsin the stellar atmosphere, hence on the degree of CNO process-ing of the various sub-populations. First-generation stars are en-hanced in carbon and oxygen, have low nitrogen content, and arerelatively faint in F275W and F438W, and bright in F336W. Con-versely, second-generation stars, which are carbon / oxygen poorand nitrogen rich, are relatively bright in F275W and F438Wand faint in F336W. The result is that first-generation stars havebluer m F336W − m F438W colors than second-generation stars at thesame luminosity (Marino et al. 2008; Bellini et al. 2010; Sbordoneet al. 2011; Milone et al. 2012b), while, for the same stars, the m F275W − m F336W color order is reversed. As a consequence, the C F275W , F336W , F438W = ( m F275W − m F336W ) − ( m F336W − m F438W ) pseudo-color defined by Milone et al. (2013) is an extremely powerful andvaluable tool to maximize the separation among the various sub-populations. Figure 2 shows the m F438W vs. m F336W − m F438W
CMD and the m F814W vs. C F275W , F336W , F438W diagram of proper-motion-selectedM 2 members. We have marked stars that, on the basis of their po-sition in the left-panel CMD, are likely HB and asymptotic-giantbranch (AGB) stars with green dots and red crosses, while bluestragglers (BSSs), have been selected from the right-panel diagramand have been represented with blue circles. A visual inspectionof the latter diagram reveals that the RGB of M 2, which shows atmost a small color dispersion using m F336W − m F438W color, spans awide range in C F275W , F336W , F438W , with distinct RGBs. AGB stars arealso distributed over a wide interval of C F275W , F336W , F438W , in closeanalogy with what is observed for the RGB, suggesting that theAGB of M 2 also hosts multiple stellar populations.Multiple populations are also characterized by di ff erent he-lium content. Typically, second-generation stars are enhanced inhelium and are hotter than first-generation stars at the same lumi-nosity. The m F275W − m F814W color is quite sensitive to the oxygenabundance of the stellar populations through the OH molecule. Inaddition, the wide color baseline provided by F275W and F814W isvery sensitive to the e ff ective temperatures of stars, thus providinga valuable tool to identify stellar populations with di ff erent heliumcontent or metallicity (see Milone et al. 2012b for details).To further investigate stellar populations along the RGB ofM 2 we combined information from three distinct diagrams: (i)m F814W vs. m F275W − m F814W , (ii) m F814W vs. C F275W , F336W , F438W , (iii)m F814W vs. m F336W − m F438W . We show these three CMDs in the pan-els a , b , and c of Fig. 3, respectively. In the rest of this Section,we have excluded AGB and HB stars, and restricted our analysisto RGB stars with 12.1 < m F814W < , b , and c of Fig. 3 and mark the blue and the red envelope ofthe main RGB. The fiducials are then used to rectify the RGB insuch a way that the blue and the red fiducials translate into verticallines with abscissa equal to − , b , and c of Fig. 3 are named ∆ NF275W , F814W , ∆ NF336W , F438W ,and ∆ NCF275W , F336W , F438W , respectively, and have been calculated foreach star as: ∆ NX = [( X − X blue fiducial ) / ( X red fiducial − X blue fiducial )]-1 where X = ( m F275W − m F814W ), ( m F336W − m F438W ), or C F275W , F336W , F438W .Panel d of Fig. 3 shows ∆ NCF275W , F336W , F438W as a function of ∆ NF275W , F814W . In this section we exploit this diagram to identify thethree main stellar components of M 2 (A, B, and C) and the twosub-populations of the component B (namely B I , and B II ) along theRGB. Most of the stars, including those of the most metal-poormetallicity peak detected by Y14, are distributed in the left regionof the diagram and follow a well-defined pattern. In this paper wewill refer to these stars as population A and represent them withblack symbols. Noticeably, a small fraction of RGB stars in M 2exhibit large values of ∆ NF275W , F814W and ∆ NCF275W , F336W , F438W and areseparated from RGB-A stars by the dashed line that we have drawnby hand. We have arbitrarily divided stars on the right of this lineinto two samples: i) a group of stars with ∆ NF275W , F814W > .
5, whichwe have represented with aqua symbols and named population C,and ii) another group with intermediate ∆ NF275W , F814W values, markedwith red triangles and named population B. These colors and sym-bols are used consistently in the other panels of this figure.Because of their colors, stars on RGB-C correspond to the red-dest RGB discussed in Sect. 3 and include the only metal-rich spec-troscopic targets for which F275W, F336W, F438W, and F814Wphotometry is available.RGB-B has bluer m F275W − m F814W and m F336W − m F438W col-ors with respect than the RGB-C, but it is redder than the RGB-A. c (cid:13) , 1– ?? A. P. Milone et al.
Figure 1.
Left panel : m F275W vs. m F275W − m F814W
CMD of stars in the WFC3 / UVIS field of view centered on M 2. Cluster members and field stars, selectedon the basis of their proper motions, are represented with black dots, and gray crosses, respectively. The vector-point diagram of proper motions is plottedin the inset and the stars within the red circle are considered as cluster members. A zoom-in of the same CMD around the SGB is plotted in the upper-rightpanel.
Lower-right panel : [Y / Fe] vs. [Fe / H] for RGB stars from Y14. Metal-poor, metal-intermediate, and metal-rich stars are represented as black circles, redtriangles, and aqua stars, respectively. In the left panel we used the same symbols to highlight the Y14 stars we could cross-identify in our photometric catalog.
RGB-B stars span a wide m F275W − m F814W and m F336W − m F438W color range and appear to cluster around two di ff erent regions inthe ∆ NCF275W , F336W , F438W vs. ∆ NF275W , F814W plane, indicating the exis-tence of at least two stellar populations within group B. We namethem populations B I and B II as illustrated in Fig. 3. The fourmetal-intermediate targets studied by Y14 for which photometryis available belong to the RGB-B group. Among them, two starsare highly enhanced in sodium ([Na / Fe] ∼ / Fe] ∼ I and B II regions in the ∆ NCF275W , F336W , F438W vs. ∆ NF275W , F814W diagram, respectively. This fact confirms that thesestar groups represent di ff erent stellar sub-populations that have dif-ferent sodium abundance.Lardo et al. (2012) have performed a CN- and CH-index studyof 38 red giants of M 2 and have identified two groups of CN-strong (CH-weak) and CN-weak (CH-strong) stars. Furthermore,they have determined the abundance of carbon and nitrogen in 35stars and found significant star-to-star variations of both elementsat all the luminosities that form an extended C-N anticorrelation. HST photometry is available for only one star ( ff erent groups of panel d ofFig. 3 we infer that in the central region of M 2 96.1 ± ± ± I and B II , contain approximately thesame number of stars (48 ±
12% and 52 ±
12% of RGB-B stars, re-spectively).The upper-left panel of Fig. 4 shows the Hess diagram in the ∆ NF336W , F438W vs. ∆ NF275W , F814W plane. Here RGB-A stars exhibit aneven more complex pattern, defining a kind of semi-circle.About half of the stars are clustered around( ∆ NF275W , F814W , ∆ NF336W , F438W ) ≃ ( − − ∆ NF336W , F438W vs. ∆ NF275W , F814W diagram and we have named them A I , A II , A III , and A IV , andcolored them in black, green, magenta, and cyan, respectively.The same colors are used consistently in the ∆ NCF275W , F336W , F438W vs. ∆ NF275W , F814W diagram plotted in the lower-right panel.In order to estimate the fraction of stars in each sub-population, we have introduced the polar reference frame shownin the upper-left panel of Fig. 4. To do this, we have arbitrar-ily translated the origin to the point indicated by the red circle c (cid:13) , 1– ?? Figure 2.
Left : m F438W vs. m F336W − m F438W
CMD (left panel);
Right : m F814W vs. C F275W , F336W , F438W diagram for M 2 stars. AGB, HB, and BSSs, selectedby eye, have been marked with red crosses, green dots, and blue circles, respectively. ( ∆ NF275W , F814W , ∆ NF336W , F438W ) = ( − − o counterclockwise. The histogram distribution of the polar an-gle, θ , in the upper-right panel clearly shows four main peaks. Wehave fitted this distribution with a least squares fit of the sum of fourGaussians and estimated the fraction of stars in each componentfrom the area of the four Gaussians. We infer that 8 ± ± ± ±
4% of stars belong to populations A I -A IV , respec-tively.The nature of these di ff erent stellar populations on RGB-Acan be clarified by again combining our photometric results, withthe chemical abundances from the spectroscopy in Y14. Thegreen, magenta, and cyan circles superimposed on the diagramof Fig. 4 refer to sodium-poor stars ([Na / Fe] = − ± / Fe] = ± / Fe] = ± II . The fact that populations A II and A III host theNa-poor and the Na-intermediate stars, respectively, suggests thatthese two stellar populations host stars with di ff erent light-elementabundances. The most Na-rich star of Y14 is not clearly associatedto any bump of stars, although it is possible that it might be relatedto population A IV , at least for the ∆ NCF275W , F336W , F438W value. Itsanomalous position in the ∆ NF275W , F814W and ∆ NF336W , F438W diagramsmay be intrinsic, but we cannot exclude the possibility that it isdue to photometric errors.
In section 3 we have shown that M 2 includes a poorly-populatedsequence of stars that is associated with population C (hereafterMS-C) and is clearly visible in the CMD of Fig. 1 on the redwardside of the main MS. We postpone the analysis of MS-C to the next section and analyze in this section the bulk of MS stars which aremostly associated with population A (MS-A). It is not possible toidentify population-B stars along the MS from our dataset.In order to identify possible multiple stellar populations alongMS-A of M 2, we followed the same approach adopted in Sect. 4for the RGB. The m F814W vs. m F275W − m F814W
Hess diagram forMS stars is shown in the left panel of Fig. 5. Below m F814W ∼ m F814W ∼ ∼ m F814W ∼
21. It is significantly larger than the m F275W − m F814W colorerror, which is between 0.03 and 0.08 mag in the same magnitudeinterval . The verticalized m F814W vs. ∆ NF275W , F814W is plotted in theinset for stars with 20.05 < m F814W < ∆ NF275W , F814W isskewed towards the blue, further confirming the presence of a bluetail of stars.The m F814W vs. C F275W , F336W , F438W
Hess diagram of MS starsin the upper-right panel of Fig. 5 furthermore reveals a red tailof stars, confirmed by the m F814W vs. ∆ NC F275W , F336W , F438W ver-ticalized diagram and the histogram of the ∆ NC F275W , F336W , F438W distribution shown in the inset. In the lower-left panel we plot ∆ NC F275W , F336W , F438W as a function of ∆ NF275W , F814W for MS-A starswith 20.05 < m F814W < The m F275W − m F814W error has been estimated as δ ( m F275W − m F814W ) = p δ ( m F275W ) + δ ( m F814W ) , where δ ( m F275W ) and δ ( m F814W ) are the r. m. s.of magnitude measurements from the six F275W and five F814W expo-sures, respectively divided by the square root of the number of images mi-nus one.c (cid:13) , 1– ?? A. P. Milone et al.
Figure 3.
Zoom-in of the m F814W vs. m F275W − m F814W (panel a ), m F814W vs. m F336W − m F438W
CMD (panel b ), and vs. C F275W , F336W , F438W diagramsaround the RGB. The two fiducials used to verticalize the RGB are shown as thick red and blue lines (see text for details). The verticalized diagrams areplotted in panels a , b , and c . Large black dots, red triangles, and aqua stars respectively mark the metal-poor, metal-intermediate, and metal-rich starsobserved by Y14. Panel d shows the ∆ NCF275W , F336W , F438W vs. ∆ NF275W , F814W diagram for RGB stars. RGB-A, RGB-B, and RGB-C stars are colored black, red,and aqua, respectively. The mean error bar in shown the lower-left corner. The magenta and green large triangles mark the metal-intermediate / Na-rich starsand metal-intermediate / Na-poor stars, respectively. The star analyzed by Lardo et al. (2012) has been represented with a large asterisk. the RGB reveals that MS-A stars and the RGB-A share similari-ties. In both cases there is a tail of stars with small ∆ NF275W , F814W and ∆ NC F275W , F336W , F438W and second one with large ∆ NF275W , F814W and ∆ Nc F275W , F336W , F438W .To further investigate multiple populations along the MS-A,we selected by eye two groups of stars with extreme m F275W − m F814W and C F275W , F336W , F438W and highlighted them in the lower-right panel of Fig. 5 with cyan and green colors, respectively.To minimize the e ff ect of binaries or stars with large photomet-ric errors we have excluded stars with very large, and very small m F275W − m F814W and C F275W , F336W , F438W values. We will use the samecolor code in Fig. 6.Milone et al. (2013) have shown that RGB stars with both ex-treme m F275W − m F814W and C F275W , F336W , F438W values are the progenyof MS stars with correspondingly extreme m F275W − m F814W and C F275W , F336W , F438W . According to this scenario, the cyan and green stars selected in Fig. 5 correspond to the groups of stars that havebeen defined as A IV and A II , respectively, along the RGB (see lowerleft panel of Fig. 4). The majority of both MS-A and RGB-A starsare clustered at intermediate ∆ NF275W , F814W and ∆ NC F275W , F336W , F438W values and are the progenitors of group A
III in the RGB (Fig. 4).The present data set does not allow us to identify the progenitors ofRGB-A I along the MS.The extreme ∆ NF275W , F814W and ∆ NC F275W , F336W , F438W values ofthe A II and A IV stellar groups imply that they must have extremecontents of helium and light elements. In the next section, we willestimate the helium spread within MS-A. We remind the readerthat both the m F275W − m F814W color and the C F275W , F336W , F438W indexare sensitive to light-element abundances such as OH, CN, NH,CH molecules of di ff erent strength a ff ect the flux in the F275W,F336W, F438W filters. As expected, the three groups of RGB-Astars in M 2, which are clearly distinguishable in Fig. 4, are more c (cid:13) , 1– ?? Figure 4.
Upper-left panel : ∆ NF336W , F438W vs. ∆ NF275W , F814W
Hess diagram for stars in the RGB-A. The reference frame adopted to measure the polar angle, θ , for RGB-A stars is also represented. The long- and short-notched lines mark di ff erent angles, from θ = o to θ = o in steps of 90 o , and 45 o , respectively.See text for details. Upper-right panel : Histogram distribution of θ for RGB-A stars. The least-squares best-fit multi-Gaussian function is represented withgray line, while its four components are colored black, green, magenta, and cyan. Lower panels : ∆ NF336W , F438W vs. ∆ NF275W , F814W diagram for RGB-A stars (leftpanel). Stars observed by Y14 are represented with large symbols. We have defined four groups of stars (A I -A IV ), and colored them black, green, magenta,and cyan, respectively. The same stars are plotted with the same color codes in the ∆ NCF275W , F336W , F438W vs. ∆ NF275W , F814W diagram shown in the lower-rightpanel. The filled and open red triangles represent RGB-BI and RGB-BII stars, respectively. The mean error bar is plotted in the lower-left corner of each panel. confused along MS-A. This is both due to the photometric errorsand to a temperature di ff erence, as the MS stars are hotter than theRGB. As a consequence, light-element variations have a lower in-fluence on the photometric passbands used in our study, due to thediminished strengths of the OH, CH and CN molecular bands (seeSbordone et al. 2011; Cassisi et al. 2013). Y14 have determined abundances for 34 elements for 16 M 2 redgiants, thus providing an accurate chemical pattern of the multi-ple stellar populations of M 2. In the following, we will use theirmeasurements to constrain age and helium-abundance di ff erencesamong stars in M 2. The left panel of Fig. 6 shows the m F814W vs. m F606W − m F814W
CMD of M 2 from ACS / WFC photometry (An-derson et al. 2008). MS-C stars are highlighted with aqua diamondsymbols.In order to measure the age and helium content, we have usedthe photometry from the F606W and F814W bands, as it is not sig-nificantly a ff ected by the light-element abundance variations (Sbor-done et al. 2011; Milone et al. 2012b). A set of isochrones fromDotter et al. (2007) corresponding to di ff erent chemical composi-tion and ages has been superimposed on the CMD of Fig. 6. Wehave used a primordial-helium (Y = α -element content as in Y14 ([Fe / H] = − α / Fe] = A II .To do this, we compared the CMD with a grid of isochroneswith the same composition but di ff erent age, distance modulus, andreddening. The best fit corresponds to an apparent distance modu- c (cid:13) , 1– ?? A. P. Milone et al.
Figure 5.
Upper panels : m F814W vs. m F275W − m F814W (left), m F814W vs. C F275W , F336W , F438W (right) diagrams of MS stars. The insets show m F814W against ∆ NF275W , F814W and m F814W against ∆ NC F275W , F336W , F438W for stars with 20.05 < m F814W < ∆ NF275W , F814W and ∆ NC F275W , F336W , F438W have been obtained by subtracting to each star the color of the fiducial line drawn by hand to reproduce the CMD.
Lower-left panel : ∆ NC F275W , F336W , F438W vs. ∆ NF275W , F814W
Hess diagram. In the right panel we have colored green and cyan the stars we consider to be the progeny of group A II and A IV identified along the RGB. lus ( m − M ) V = .
55, age 13.0 ± − V) = / WFCF606W and F814W band as in Bedin et al. (2005). We assumedthe same values of reddening and distance modulus for the otherisochrones of Fig. 6. The best-fit age was estimated as in Dotteret al. (2010) by determining the isochrone that best fit the CMD inthe region between the MS turn o ff and the SGB. The correspond-ing uncertainty was inferred from the intrinsic magnitude and colorspread of the MS turn o ff and the SGB stars and we considered as1 σ -uncertainty the range of age that envelope the bulk of these stars(see Dotter et al. 2010 for details).The aqua and the yellow isochrones both have age 12.0 Gyr,primordial helium, and the same iron abundance as measured byY14 ([Fe / H] = − ff erent α -element con-tent. The aqua isochrone corresponds to [ α / Fe] = A II and its uncertainty corresponds to 0.75 Gyr.As a consequence, populations A and C are consistent withbeing coeval within ∼ = m F814W ∼ m F606W − m F814W colors than the bulk of M 2 stars at the same lumi-nosity further supports our conclusion that the population C is notconsistent with being significantly helium enhanced.In the upper-right panel of Fig. 6 green and cyan color codesmark the two stellar groups A II and A IV of MS stars that we iden-tified in Fig. 5. The corresponding fiducial lines have been repre-sented with the same colors. It is clear that MS- A II stars are redderthan stars in the MS- A IV .In the lower-right panel, we have superimposed onto the fidu- c (cid:13) , 1–, 1–
55, age 13.0 ± − V) = / WFCF606W and F814W band as in Bedin et al. (2005). We assumedthe same values of reddening and distance modulus for the otherisochrones of Fig. 6. The best-fit age was estimated as in Dotteret al. (2010) by determining the isochrone that best fit the CMD inthe region between the MS turn o ff and the SGB. The correspond-ing uncertainty was inferred from the intrinsic magnitude and colorspread of the MS turn o ff and the SGB stars and we considered as1 σ -uncertainty the range of age that envelope the bulk of these stars(see Dotter et al. 2010 for details).The aqua and the yellow isochrones both have age 12.0 Gyr,primordial helium, and the same iron abundance as measured byY14 ([Fe / H] = − ff erent α -element con-tent. The aqua isochrone corresponds to [ α / Fe] = A II and its uncertainty corresponds to 0.75 Gyr.As a consequence, populations A and C are consistent withbeing coeval within ∼ = m F814W ∼ m F606W − m F814W colors than the bulk of M 2 stars at the same lumi-nosity further supports our conclusion that the population C is notconsistent with being significantly helium enhanced.In the upper-right panel of Fig. 6 green and cyan color codesmark the two stellar groups A II and A IV of MS stars that we iden-tified in Fig. 5. The corresponding fiducial lines have been repre-sented with the same colors. It is clear that MS- A II stars are redderthan stars in the MS- A IV .In the lower-right panel, we have superimposed onto the fidu- c (cid:13) , 1–, 1– ?? Figure 6.
Left : m F814W vs. m F606W − m F814W
CMD of M 2, with five superimposed isochrones with the values of [Fe / H], [ α / Fe], age, and Y listed in the figureinset. Aqua symbols highlight population C stars.
Right panels
Zoom-in of the same CMD shown in the left panel around the MS (upper panel). A II andA IV stars, as defined in Fig. 5, are plotted with green and cyan points, respectively. The green and the cyan line are the fiducials of A II and A IV MSs. In thelower-right panel we have superimposed the two metal-poor isochrones on the fiducials. cial lines defined in the upper-right panel the metal-poor isochronesfrom the left panel, using the same distance modulus and redden-ing. Population A II is well fitted by an isochrone with primordialhelium (Y = / H] = − α / Fe] = A IV is reproduced by an isochrone corre-sponding to a stellar population with the same age, and the samecontent of iron and α -elements as A IV , but with enhanced helium(Y = α / Fe] and [Na / Fe] used in this paper, and with the fractionof stars within each sub-population.
In order to better understand the properties of the seven sequencesphotometrically identified in M 2, in this Section we will combinethe available spectroscopic information with photometric results.In Fig 7 we summarize what we currently know about clustersthat show an intrinsic variation in iron content. Figure 7 shows theabundance of sodium, aluminum, and barium vs. iron content forfive GCs: M 2 (Y14), M 22 (Marino et al. 2011a), M 54 includingthe Sagittarius (Sgr) dwarf galaxy nucleus (Carretta et al. 2010a), ω Cen (Marino et al. 2011b) and Terzan 5 (Origlia et al. 2011). Wehave represented with red triangles stars of M 22 and M 2 that areenhanced in s -processes elements. Stars in the Sgr dwarf galaxy, inthe population C of M 2, and metal-rich stars of Terzan 5 are plottedwith aqua star symbols, while blue diamonds represent stars in themost-metal-rich RGB of ω Cen.Figure 7 shows that populations A and B of M 2 have di ff er-ent content of s -elements with iron-intermediate stars also being s - c (cid:13) , 1– ?? A. P. Milone et al.
Figure 7.
Left : [Na / Fe] vs. [Fe / H] for four GCs with intrinsic variations in metallicity: M 22 (Marino et al. 2011a); M 2 (Y14); ω Centauri (Marino et al. 2011b);M 54 and the Sagittarius dwarf galaxy (Carretta et al. 2010a).
Right : the three upper panels show [Ba / Fe] against [Fe / H] for M 22 (Marino et al. 2011a); M 2(Y14); ω Centauri (Marino et al. 2011b), while the two lower panels show [Al / Fe] vs. [Fe / H] for M 2 (Y14) and Terzan 5 (Origlia et al. 2011). The s -rich starsof M 22 and M 2 are represented with red triangles, while the blue diamonds mark stars in the RGB-a of ω Centauri. Population-C stars of M 2, stars in theSagittarius dwarf galaxy nucleus, and the metal-rich stars of Terzan 5 are plotted with aqua symbols.
POP Population [Fe / H] [Na / Fe] [ α / Fe] Y age Sect. / Fig. Sect. / Fig.ratio [dex] [dex] [dex] [dex] [Gyr] RGB MS A I ± − ± / A II ± − − ± / / A III ± − ± / A IV ± − ± / / B I ± − / B II ± − / C ± − − ± / / Table 1.
Fraction of stars, and summary of the main properties of the seven stellar populations of M 2 used throughout this work. As discussed in the paperthe values of [Fe / H], [Na / Fe], and [ α / Fe] are derived from Y14. The last two columns indicate the section and the figure where the populations have beenidentified along the RGB and the MS. rich (Y14). In addition, Fig. 4 shows that RGB-A and RGB-B starshost stellar sub-populations with di ff erent light-element abundancein close analogy with what is observed in M 22. Y14 suggestedthat the populations A and B of M 2, and other anomalous GCs(M 22, NGC 1851, and ω Cen) have experienced a similar com-plex star-formation history. We refer the reader to papers by Marinoet al. (2009, 2011a,b), Carretta et al. (2010a), Da Costa & Marino(2011), Y14 and reference therein for a discussion of this topic.Y14 noticed that the RGB-C stars of M 2 exhibit chemical properties that are rarely observed in any GC. In this section wefocus on the population C of M 2 and compare it with the threeextreme cases of ω Cen, M 54, and Terzan 5. • ω Cen exhibits a multimodal iron distribution that spans awide range of metallicity, with [Fe / H] ranging from ∼ − ∼− c (cid:13) , 1– ?? is made of metal-poor and helium-normal stars, a blue MS, whichhosts metal-intermediate stars and is highly helium enhanced, byup to Y ∼ HST photom-etry to investigate the multiple MSs in ω Cen and showed thatMSa stars have redder m F275W − m F814W , m F336W − m F814W , and m F435W − m F814W colors than the red MS and the blue MS. Theyalso noticed that, when using m F606W − m F814W , m F625W − m F814W , or m F658N − m F814W colors, MSa becomes bluer than the red MS. SinceMSa stars are the progenitors of the RGBa (Pancino et al. 2002),they are significantly more metal-rich than the red MS. Thereforethe blue colors of MSa would imply that its stars are enriched inhelium, as suggested by Norris (2004).M 2 is similar to ω Cen in having stellar populations that arehighly enhanced in iron relative to the others, but there are twoimportant di ff erences: (1) in ω Cen, the stars of the most metal-rich population are strongly enhanced in sodium, aluminum and s -process elements (Norris & Da Costa 1995; Johnson & Pila-chowski 2010; Marino et al. 2011b, D’Antona et al. 2011), in con-trast with the population C of M 2, where the content of Na, Al,Y, and Zr are comparable with those of first-generation, normal-population stars (Y14, see also Fig. 6). (2) The population C ofM 2 is not consistent with a high helium abundance. We concludethat ω Cen and M 2 have certainly experienced a di ff erent chemical-enrichment history. • M 54 is another massive GC with star-to-star variations inmetallicity (Sarajedini & Layden 1995; Bellazzini et al. 2008; Car-retta et al. 2010a,b). It lies in the nuclear region of the Sgrdwarf galaxy, although it is not clear whether it formed in situor was pushed into the center by dynamical friction. Carretta etal. (2010a,b) have derived chemical abundances for Fe, Na, and α elements for 103 red giants in the Sgr nucleus (Sgr-N). They usedboth radial velocities and photometry to identify 76 M 54 mem-bers and 27 Sgr stars. Carretta and collaborators showed that M 54stars span a wide range in metallicity, with [Fe / H] ranging from ∼ − − σ ∼ / Fe] vs. [Fe / H] for RGB stars in M 54 (black circles)and Sgr (aqua stars) from Carretta et al. (2010b). Stars in M 54 ex-hibit a large star-to star variation in [Na / Fe], and, on average, high[Na / Fe]. By contrast, Sgr-N stars exhibit a low sodium abundance(see Fig. 7) .The stellar system including M 54 and the Sgr-N shares somesimilarities with M 2. M 54 seems to include populations similarto M 2’s A and B, while the Sgr-N star chemistry is similar to theabundance pattern of M 2 population C. Unlike the case of ω Cen,however, where most of the metal-rich stars are strongly enhancedin sodium, both the Sgr-N stars and population C of M 2 exhibitlow [Na / Fe]. In addition, population C of M 2 is depleted in [ α / Fe]by ∼ α -elements than M 54 stars. • Terzan 5 also exhibits a very peculiar chemical compositionwith two main groups of stars with [Fe / H] ∼ − / H] ∼ + α -element abundance with respect to the moremetal-poor population, thus suggesting that Type Ia SNe may haveplayed a role in the star-formation history of this cluster. Notice-ably, at odds with the populations A and B of M 2, there is no evi-dence for light-element variations among neither the metal-rich northe metal-poor stars of Terzan 5. To further investigate similarities between M 2 and Terzan 5, we show in the lower-right panels ofFig.7 [Al / Fe] vs. [Fe / H]. We note that all the metal-rich stars ofboth M 2 and Terzan 5 have almost-solar aluminum ([Al / Fe] ∼ We have exploited multi-wavelength photometry from the
HST
UVLegacy Survey GO-13297 to investigate the stellar populations ofthe GC M 2. We have identified three main components, which wenamed A, B, and C. Within these three main components, we iden-tified seven stellar sub-populations.The main component, A, which includes sub-populations A I ,A II , A III , and A IV , hosts the metal-poor stars identified by Y14. Itexhibits an intrinsic spread in helium, with Y ranging from pri-mordial values (Y ∼ ∼ ∼
96% of stars.Noticeably, the three stars with di ff erent sodium abundance iden-tified by Y14 are located on the di ff erent sequences, with helium-rich stars also having higher [Na / Fe]. Therefore the component Ais similar to the multiple stellar populations we have identified inthe majority of GCs. These multiple sequences host stars with thesame heavy element abundance, have almost-homogeneous contentof s -elements, and exhibit star-to-star variations in helium and lightelements as expected for material which has been gone through hy-drogen burning at high temperatures (CNO cycle).Component B has intermediate metallicity and includes ∼ I , B II , with a di ff erent light-element abundance. It could be the analogous of the s -rich andiron-rich stellar population of M 22, and shares similarities withthe iron-rich stellar populations of ω Cen and M 54.Component C includes ∼
1% of stars, is highly enhanced iniron ([Fe / H] ∼ − α / Fe] and [Al / Fe] ratiothan populations A and B. Its stars are s -poor and are not enhancedin helium. Such properties are not compatible with self-enrichmentdue to either AGB or fast rotating massive stars and are not consis-tent with the early-disc accretion scenario.The combination of the photometric and spectroscopic stud-ies of M 2 presented in this paper and in Y14 suggest that M 2is composed by at least two distinct entities. The main one, con-taining most of the stars in M 2, includes populations A and Band can be further separated into six distinct sub-populations. Theminor component (population C), which makes up ∼
1% of thecluster stars and apparently has not produced any secondary sub-populations, is populated by stars rich in metals and with s-processelements in nearly solar proportion. Due to the chemical proper-ties of population-C stars, one possibility is that population-C starsformed from material that made the first A sub-population but hav-ing been slightly contaminated by supernovae of either type.In any event, the component A + B and the component C ofM 2 must have experienced independent star-formation histories,and as an alternative M 2 may be final result of the merger of twostellar systems. The fact that M 2 shares many similarities with thestellar system composed by M 54 and the nucleus of the Sagittariusdwarf galaxy, makes it very tempting to speculate that it could bethe remnant of a much larger stellar system which merged with theMilky Way in the past. c (cid:13) , 1– ?? A. P. Milone et al.
ACKNOWLEDGMENTS
Support for Hubble Space Telescope proposal GO-13297 was providedby NASA through grants from STScI, which is operated by AURA, Inc.,under NASA contract NAS 5-26555. APM and HJ acknowledge sup-port by the Australian Research Council through Discovery Project grantDP120100475. MZ acknowledges support by Proyecto Fondecyt Regular1110393, by the BASAL Center for Astrophysics and Associated Tech-nologies PFB-06, and by Project IC120009 ‘Millennium Institute of As-trophysics (MAS)’ of Iniciativa Cient´ıfica Milenio by the Chilean Ministryof Economy, Development and Tourism.
REFERENCES
Anderson, J., & King, I. R. 2003, AJ, 126, 772Anderson, J., & King, I. R. 2006, Instrument Science Report ACS 2006-01, 34 pages, 1Anderson, J., Bedin, L. R., Piotto, G., Yadav, R. S., & Bellini, A. 2006,A&A, 454, 1029Anderson, J., Sarajedini, A., Bedin, L. R., et al. 2008, AJ, 135, 2055Anderson, J., & Bedin, L. R. 2010, PASP, 122, 1035Bedin, L. R., Piotto, G., Anderson, J., et al. 2004, ApJ, 605, L125Bedin, L. R., Cassisi, S., Castelli, F., et al. 2005, MNRAS, 357, 1038Bellini, A., & Bedin, L. R. 2009, PASP, 121, 1419Bellini, A., Bedin, L. R., Piotto, G., et al. 2010, AJ, 140, 631Bellini, A., Anderson, J., & Bedin, L. R. 2011, PASP, 123, 622Carretta, E., Bragaglia, A., Gratton, R. G., et al. 2010, ApJ, 714, L7Carretta, E., Bragaglia, A., Gratton, R. G., et al. 2010, A&A, 520, A95Carretta, E., Gratton, R. G., Lucatello, S., et al. 2010, ApJ, 722, L1Cassisi, S., Mucciarelli, A., Pietrinferni, A., Salaris, M., & Ferguson, J.2013, A&A, 554, A19Da Costa, G. S., Held, E. V., Saviane, I., & Gullieuszik, M. 2009, ApJ,705, 1481Da Costa, G. S., & Marino, A. F. 2011, PASA, 28, 28Da Costa, G. S., Held, E. V., & Saviane, I. 2014, MNRAS, 438, 3507D’Antona, F., & Caloi, V. 2004, ApJ, 611, 871D’Antona, F., D’Ercole, A., Marino, A. F., et al. 2011, ApJ, 736, 5Dotter, A., Sarajedini, A., Anderson, J., et al. 2010, ApJ, 708, 698Ferraro, F. R., Dalessandro, E., Mucciarelli, A., et al. 2009, Nature, 462,483Girardi, L., Groenewegen, M. A. T., Hatziminaoglou, E., & da Costa, L.2005, A&A, 436, 895Gratton, R. G., Villanova, S., Lucatello, S., et al. 2012, A&A, 544, AA12Grundahl, F. 1999, Spectrophotometric Dating of Stars and Galaxies, 192,223Harris, W. E. 1996, AJ, 112, 1487Johnson, C. I., & Pilachowski, C. A. 2010, ApJ, 722, 1373King, I. R., Bedin, L. R., Cassisi, S., et al. 2012, AJ, 144, 5Lardo, C., Pancino, E., Mucciarelli, A., & Milone, A. P. 2012, A&A, 548,A107Lardo, C., Pancino, E., Mucciarelli, A., et al. 2013, MNRAS, 433, 1941Marino, A. F., Villanova, S., Piotto, G., et al. 2008, A&A, 490, 625Marino, A. F., Milone, A. P., Piotto, G., et al. 2009, A&A, 505, 1099Marino, A. F., Milone, A. P., Piotto, G., et al. 2011, ApJ, 731, 64Marino, A. F., Sneden, C., Kraft, R. P., et al. 2011, A&A, 532, A8Marino, A. F., Milone, A. P., Yong, D., et al. 2014, MNRAS, 442, 3044Milone, A. P., Bedin, L. R., Piotto, G., & Anderson, J. 2009, A&A, 497,755Milone, A. P., Piotto, G., Bedin, L. R., et al. 2012a, A&A, 540, A16Milone, A. P., Piotto, G., Bedin, L. R., et al. 2012b, ApJ, 744, 58Milone, A. P., Marino, A. F., Piotto, G., et al. 2013, ApJ, 767, 120Norris, J. E., & Da Costa, G. S. 1995, ApJ, 447, 680Norris, J. E. 2004, ApJ, 612, L25Origlia, L., Rich, R. M., Ferraro, F. R., et al. 2011, ApJ, 726, L20Pancino, E., Pasquini, L., Hill, V., Ferraro, F. R., & Bellazzini, M. 2002,ApJ, 568, L101Pietrinferni, A., Cassisi, S., Salaris, M., & Castelli, F. 2004, ApJ, 612, 168 Pietrinferni, A., Cassisi, S., Salaris, M., & Castelli, F. 2006, ApJ, 642, 797Piotto, G., Villanova, S., Bedin, L. R., et al. 2005, ApJ, 621, 777Piotto, G., Milone, A. P., Anderson, J., et al. 2012, ApJ, 760, 39Piotto, G., Milone, A. P., Bedin, L. R., et al. 2014, arXiv:1410.4564Sarajedini, A., & Layden, A. C. 1995, AJ, 109, 1086Sarajedini, A., Bedin, L. R., Chaboyer, B., et al. 2007, AJ, 133, 1658Sbordone, L., Salaris, M., Weiss, A., & Cassisi, S. 2011, A&A, 534, A9Yong, D., & Grundahl, F. 2008, ApJ, 672, L29Yong, D., Mel´endez, J., Grundahl, F., et al. 2013, MNRAS, 434, 3542Yong, D., Roederer, I. U., Grundahl, F., et al. 2014, MNRAS, 441, 3396,Y14Villanova, S., Geisler, D., Gratton, R. G., & Cassisi, S. 2014, ApJ, 791,107 c (cid:13) , 1–, 1–