A comprehensive chemical abundance study of the outer halo globular cluster M 75
aa r X i v : . [ a s t r o - ph . GA ] A p r Astronomy&Astrophysicsmanuscript no. niki2˙v2 c (cid:13)
ESO 2018October 30, 2018
A comprehensive chemical abundance study of the outer haloglobular cluster M 75
N. Kacharov ⋆ , A. Koch , and A. McWilliam Landessternwarte, Zentrum f¨ur Astronomie der Universit¨at Heidelberg, K¨onigstuhl 12, D-69117 Heidelberg, Germany Carnegie Observatories, 813 Santa Barbara St., Pasadena, CA 91101, USAReceived 03.03.2013 / Accepted 11.04.2013
ABSTRACT
Context.
M 75 is a relatively young Globular Cluster (GC) found at 15 kpc from the Galactic centre at the transition region betweenthe inner and outer Milky Way halos.
Aims.
Our aims are to perform a comprehensive abundance study of a variety of chemical elements in this GC such as to investigateits chemical enrichment history in terms of early star formation, and to search for any multiple populations.
Methods.
We have obtained high resolution spectroscopy with the MIKE instrument at the Magellan telescope for 16 red giant stars.Their membership within the GC is confirmed from radial velocity measurements. Our chemical abundance analysis is performed viaequivalent width measurements and spectral synthesis, assuming local thermodynamic equilibrium (LTE).
Results.
We present the first comprehensive abundance study of M 75 to date. The cluster is metal-rich ([Fe / H] = − . ± .
02 dex,[ α / Fe] = + . ± .
02 dex), and shows a marginal spread in [Fe / H] of 0 .
07 dex, typical of most GCs of similar luminosity. A moderatelyextended O-Na anticorrelation is clearly visible, likely showing three generations of stars, formed on a short timescale. Additionallythe two most Na-rich stars are also Ba-enhanced by 0 . . M ∼ − M ⊙ )Asymptotic Giant Branch (AGB) stars. The overall n-capture element pattern is compatible with predominant r-process enrichment,which is rarely the case in GCs of such a high metallicity. Key words.
Stars: abundances – Globular clusters: general – Globular clusters: individual: M 75 – Galaxy: halo –
1. Introduction
With ages of 10 −
14 Gyr, globular clusters (GCs) are theoldest stellar systems in the Galaxy and therefore reflect itsearliest evolutionary stages. Although the Galactic halo GCsystem appears homogeneous (e.g. Cohen & Mel´endez 2005;Koch et al. 2009) and well compatible with the stellar halo inmany regards (e.g. Geisler et al. 2007), numerous propertiesshow broad di ff erences between individual clusters and are atodds with halo field stars. These characteristics comprise largespreads and anti-correlations of the light- elements involvedin p-capture reactions (C, N, O, F, Na, Mg, and Al, Osborn1971; Cottrell 1981; Kraft 1994; Gratton et al. 2001, 2004;Carretta et al. 2009b,c). Nowadays, these variations amongst thelight elements are considered as an evidence for the existence ofat least two generations of stars, present in all GCs studied todate (e.g. Gratton et al. 2012a, and references therein), as alsooften prominently seen in their colour-magnitude diagrams (e.g.Piotto et al. 2012). The multiple populations in GCs are tightlylinked to the “second-parameter e ff ect”, which needs to explaindiscordant horizontal branch (HB) morphologies at any givenmetallicity. Suggested solutions to this problem include a broadage range in the GCs (Searle & Zinn 1978) or variations in theirhelium content (e.g. D’Antona et al. 2002), whilst mass loss, α -abundances, rotation, deep mixing, binary interactions, core con- Send o ff print requests to : N. Kacharov, [email protected] ⋆ Member of the International Max Planck Research School forAstronomy and Cosmic Physics at the University of Heidelberg,IMPRS-HD, Germany. centration, or planetary systems cannot be ruled out as possiblesecond parameters (see Catelan 2009, for a detailed review).Currently, there are several theories trying to explain the for-mation of at least two stellar populations in GCs. The best candi-dates that pollute the interstellar medium (ISM) with p-captureelements whilst producing only little or no α - and Fe-peak-elements are massive ( ∼ − M ⊙ ) AGB stars (D’Ercole et al.2008, 2010) or fast rotating massive ( M > M ⊙ ) stars (FRMSDecressin et al. 2007). Both mechanisms work on very di ff er-ent timescales: The winds of FRMS enrich the ISM with p-capture products in ∼ × yrs, slightly before the explo-sions of the bulk of SNe II take place. On the other hand, thelong-lived AGB stars take a few 10 yrs before they enrich theISM with these elements (Gratton et al. 2012a). A recent studyby Valcarce & Catelan (2011) attempts to combine both mecha-nisms. A problem in all theories is the small fraction of primor-dial first generation (FG) stars ( ∼ ∼ ffi cient to form the numerousSG by a factor of ∼
10 (D’Ercole et al. 2008). For both mecha-nisms to work, one has to either invoke a top-heavy IMF for theFG or to assume that the GCs were much more massive and theylost a large fraction of their initial mass. De Mink et al. (2009)made an attempt to solve this problem by suggesting massive bi-naries as the main polluters. In fact, the similarity in chemistryand ages of the Milky Way halo field stars with the FG starsin GCs and the discovery of a few halo stars with modified CNand CH abundances, suggest that the bulk of the halo stars wereindeed formed in GCs (Gratton et al. 2012a; Martell & Grebel2010) and that the GCs originated in the centres of much larger
1. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 and later disrupted stellar systems (e.g. Carretta et al. 2010a;B¨oker 2008; Kravtsov & Gnedin 2005).Coupled with the lack of a metallicity gradient in the outerhalo GC system, the outer halo clusters’ second parameter prob-lem had prompted the first suggestion by Searle & Zinn (1978)that those GCs could have been donated by accreted dwarfgalaxy-like systems. The most distant Milky Way GCs presenta number of properties, which suggest a di ff erent origin than theinner halo GCs (Rodgers & Paltoglou 1984; Zinn 1993, 1996;Mar´ın-Franch et al. 2009). These comprise younger ages, dif-ferent kinematics, and possibly di ff erent chemical composition.Therefore, studies of GCs at larger Galactic distances are crucialfor the understanding of how the Galactic halo formed.In this paper we present the first ever chemical elementabundances derived from high-resolution spectra for the GCM 75. This cluster is located at a galactocentric distance of15 kpc, which tenants the transition region between the innerand outer Milky Way halo (Zinn 1993; Carollo et al. 2007). Itsyounger age ( ∼
10 Gyr; Catelan et al. 2002) and high metal-licity ([Fe / H] = − .
16 dex, this work) are compatible with theproperties of the outer halo GC system and suggest a possibleextragalactic origin. On the other hand, M 75 is amongst themost concentrated GCs ( c = log( r t / r c ) = .
2. Observations and data reduction
Our spectroscopic observations of 16 giant stars in M 75 weretaken using the Magellan Inamori Kyocera Echelle (MIKE)spectrograph at the 6.5-m Magellan2 / Clay Telescope at LasCampanas Observatory, Chile. The instrument consists of twoarms sensible in the red and blue parts of the visible spectrum,which cover an entire wavelength range of 3340 Å to 9150 Å.Our data were collected over one night in April and four nightsin July 2011. By using a slit width of 0 . ′′ and 2 × ∼ ′′ on average. We reached a relatively highS / N of ∼
70 per pixel around 6500 Å on the red CCD and ∼
40 per pixel around 4500 Å on the blue CCD. The observ-ing log is presented in Table 1. The targets were selected fromthe catalogue of Kravtsov et al. (2007), choosing stars with ahigh membership probability, i.e. those within the tidal radiusof the cluster, yet avoiding the crowded central regions. Thiswas aided by visual inspection of archival FORS preimaging(Program ID 69.B-0305, P.I. E. Tolstoy). A colour-magnitudediagram (CMD) of M 75, highlighting our spectroscopic sam-ple, is presented in Figure 1. We also overplotted an isochroneof age 10 Gyr and metallicity Z = .
003 from the Padova li-brary (Girardi et al. 2010; Marigo et al. 2008), which best rep-resents the photometric data. This adopts an extinction value
Table 1.
Observing Log.
Star ID V Date Exp. time[mag] [s]239 15.04 Jul. 25 2011 3 × × + × × × × × × × + + × × + × × Notes. (1)
Based on the catalogue of Kravtsov et al. (2007). A V = .
49 mag ( E ( B − V ) = .
147 mag) from Schlegel et al.(1998), and a distance modulus of ( m − M ) = . f xcor toolin IRAF. From this, we found a mean heliocentric velocity of − . ± . − (standard deviation 8 . − ), confirm-ing the cluster membership of all stars. This is in an excellentagreement with the mean systematic radial velocity of M 75 of − . − , σ = . − (Harris 1996, 2010 version).Whilst this dispersion may seem large for a GC, it has to bekept in mind that M 75 is a very massive and concentrated sys-tem, so that this value is fully compatible with the large ve-locity dispersions found in other comparably luminous systems(Pryor & Meylan 1993).
3. Abundance analysis
We derived chemical element abundances through an equivalentwidth (EW) analysis, complemented by spectral synthesis us-ing the stellar abundance code MOOG (Sneden 1973). We usedan absolute abundance analysis method, which closely followsthe procedures described in Koch et al. (2009); Koch & Cˆot´e(2010). The line list was assembled from various sources(Koch & Cˆot´e 2010, and references therein), and complementedwith atomic data from the Kurucz data base . Additionaltransitions for some heavier elements were adopted from
2. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 V (B-V) Fig. 1.
CMD from Kravtsov et al. (2007). The stars from oursample are indicated by larger red symbols. An isochrone for[Fe / H] ∼ − . splot task in IRAF. We restricted our mea-surements to lines having reduced EWs (log(EW /λ )) less than − . ffi culties in plac-ing the continuum due to the strong blending owing to the rela-tively high metallicity of M75. We used the deblending option inIRAF’s task splot to account for blended lines where necessary.For the elements Rb, Zr, Gd, Dy, Er, Hf, and Th, we usedspectral synthesis instead. Accurate EW measurements were notpossible because of strong blending, too weak lines, or too lowS / N ratio.We applied corrections for Hyperfine Structure (HFS) split-ting for the odd-Z elements V, Mn, Co, Cu, Rb, La, and Eu usingthe blends driver of MOOG and atomic data for the splittingfrom McWilliam et al. (1995, 2012, in prep.). HFS correctionsfor Sc and Ba were small compared to the 1 σ measurement un-certainty and we ignored them. HFS corrections for the lighterodd-Z elements are generally negligible.Finally, the derived abundances were placed on the solarscale of Asplund et al. (2009). The full linelist and the measuredEWs are available in the online version of A&A. We present thefirst rows and columns of this table to guide the eye (Table 2). We interpolated the new grid of Kurucz plane-parallel,one-dimensional models without convective overshoot. spec T pho t based temperature Fig. 2.
A comparison between the spectroscopic and photometrictemperature estimates. The average temperature calculated from( V − J ) , ( V − H ) , and ( V − K ) colours is indicated by redcircles and the ( V − I ) calibration is indicated by blue squares.The dashed line is unity.These include the α -enhanced opacity distribution functions(AODFNEW; Castelli & Kurucz 2003) .As an initial guess, we calculated e ff ective temperaturesof our targets based on the ( V − I ) colours from the photo-metric catalogue of Kravtsov et al. (2007). This was comple-mented with photometry from 2MASS (Cutri et al. 2003) to ob-tain temperature-estimates based on the ( V − J ) , ( V − H ) , and( V − K ) colour indices. We used the temperature-colour calibra-tions of Ram´ırez & Mel´endez (2005). Additionally, we obtainedspectroscopic temperatures by measuring the EWs of a largenumber, typically about 60, Fe I lines and establishing excitationequilibrium. This is achieved by changing the temperature un-til there is no correlation between the derived abundances fromdi ff erent Fe I lines and their excitation potential. As a result,the mean temperature from all three 2MASS based indicators islower than the temperature based on the ( V − I ) colour alone andthe spectroscopic estimates by 200 K on average (Figure 2). Asimilar trend was also noted by Fabbian et al. (2009). One possi-ble explanation is the larger pixel size of the 2MASS detectors,which can lead to an undersampling in the crowded GC fieldcompared to better sampled optical images. This way, additionalflux contributions per pixel would yield overestimated infraredmagnitudes and thus lower e ff ective temperatures. We use thespectroscopic temperatures in the following analysis. The meandi ff erence between the temperatures from the ( V − I ) coloursand the spectroscopic ones is only 2 K with a 1 σ -scatter of 60 K,which we adopted as the temperature error for our targets.We derived physical gravities from the canonical Equation 1,using the dereddened V magnitudes with a bolometric correc-tion interpolated from the Kurucz grid, the spectroscopic temper-atures, and adopting the known distance to the cluster (19 kpc).We adopted a mass µ = .
78 M ⊙ for all stars, which is consistentwith the masses from the reference isochrones. Adopting lowermasses for the possible AGB stars in our sample would lead to asmall change in gravity, which would have a negligible e ff ect onthe derived abundances (See Section 3.3).log g = log( µ/µ ⊙ ) + T / T ⊙ ) − . M ⊙ − M ) + log g ⊙ (1)
3. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75
Table 2.
Line list and equivalent widths. The full table is available in the electronic version of the journal.
Element λ χ log gf EWs [mÅ] for each star[Å] [eV]
In the above equation, M = M V − BC denotes the absolute bolo-metric magnitude of the stars. We did not adjust the gravitiesto enforce ionisation equilibrium. As a result, the abundancesfrom the Fe II lines are higher by 0 .
18 dex ( σ = .
10 dex)on average, compared to the neutral species (Figure 3). Wenote that the di ff erences are larger for cooler stars, which mightbe due to the use of plane-parallel models instead of spheri-cal or 3D ones (Bergemann et al. 2012), departure from LTE(Heiter & Eriksson 2006; Bergemann et al. 2012; Ruchti et al.2013), or unknown blends. Additionally, the use of iron lineswith a broad range of excitation potentials (from 1 to 5 eV) couldalso cause some discrepancy, as noted by Worley et al. (2010)and Worley & Cottrell (2010). The discrepancy in Fe I vs. FeII is too large to be explained by systematic errors. Shifting thegravities by 0 . ∆ E ( B − V ) = .
35 mag or an increase of thetemperature scale by 100 K (200 K for the coolest stars) will alsorestore ionisation equilibrium. We deemed both possibilities un-likely, given the large, required changes compared to the smalluncertainties in the parameters and our overall excellent excita-tion equilibrium. Likewise, the Ti abundances from the ionisedspecies are larger than the ones based on Ti I lines by 0 .
07 dex( σ = .
15 dex).Microturbulent velocities ( ξ ) were determined by removingany trend in the plot of abundances versus EW of the Fe I lines.The derived values for M 75 are in the order of 2 . − . Atypical error of this method is ∼ . − : variations withinthis range still allow reasonably flat slopes in the EW plot.Since we did not have a prior knowledge of the metallici-ties of the individual stars, we started with atmosphere modelsfor [M / H] = − . −1.4 −1.2 −1.0 −0.8 −0.6[XI/H]−1.4−1.2−1.0−0.8−0.6−0.4 [ X II/ H ] FeTi
Fig. 3.
A comparison between the abundance results from neu-tral and ionised species. Fe abundances are shown with bluecircles and Ti abundances with red squares. The dashed line isunity.(Catelan et al. 2002), and updated it iteratively based on the FeI estimates from the previous step. Note that all the parameterswere iterated upon convergence. The derived stellar parametersfor all 16 red giants are summarized in Table 3.
We calculated the random error of the abundances as σ EW / √ N ,where N is the number of lines used for those elements where we
4. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75
Table 3.
Stellar parameters.
Star ID T ff T ( V − I )e ff T spece ff log g ξ [Fe I / H] [Fe II / H][K] [K] [K] log [cm s − ] [km s − ] [dex] [dex]239 4021 4208 4300 1.07 2.1 − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . − . ± . Notes. (1)
Possible AGB stars. measured more than one line and σ EW is the standard deviation.For those elements, for which we used the EW of a single line,we adopted a typical random error based on the mean abundancespread of all stars in our sample. For the abundances derived viasynthesis, we adopted random errors based on the minimum andmaximum abundance values that still yielded acceptable fits tothe observed spectra.To investigate the systematic abundance errors caused by theuncertainties of the stellar atmosphere parameters, we calculateda grid of new model atmospheres, varying the e ff ective temper-ature by ±
60 K, the surface gravity by ± . ± . − and the metallicity by ± . ff erence in their parameters.We also included calculations for Solar [ α / Fe] ratios using theKurucz ODFNEW models (column labeled ’ODF’ in Table 4).Then we recomputed the abundances of all elements with themodified atmosphere models. The results are summarized inTable 4 in terms of di ff erence to the default values. The col-umn labeled total lists the errors of all parameters combined inquadrature, including a 0 . α / Fe]. The lat-ter corresponds to 1 / ff erence between theAODFNEW and ODFNEW Kurucz models. We note, however,that these are upper limits due to the covariance of the atmo-spheric parameters (e.g. McWilliam et al. 1995).The change in temperature has a larger e ff ect on the specieswith a lower excitation potential (e.g. K I, Ti I, V I, Cr I), whilstthe change in gravity a ff ects mostly the ionised species. Forwarm GK giants the dominant Fe II species are more sensitiveto gravity and the Fe I species to changes in temperature. But wenote that the stars studied here, especially the coolest stars of oursample, have so low temperatures that they are in the transitionfrom Fe being dominated by the ionised species to Fe dominatedby the neutral species. That is why Fe II is so sensitive to vari-ations of the e ff ective temperature in this case. We note that ourprior ignorance of the metallicity of the model atmospheres hasonly a negligible e ff ect on the derived abundances. The overall,typical, systematic uncertainties are of the order of 0 .
4. Abundance results
In Table 6 we summarize the abundance results for M 75, relativeto Fe I for all neutral species and to Fe II for all ionised species.We also list the mean random error, ǫ rand , and the mean sys-tematic error, ǫ sys , on the abundance ratios [X / Fe]. The columnslabeled σ obs and σ obs contain the observed spreads of the abun-dances within the cluster. As noted above, the discrepancy be-tween Fe I and Fe II values are largest for the coolest stars in oursample, leading to a larger spread in the Fe II abundance. Forthis reason, we show the spreads calculated by using all stars( σ obs ) and by excluding the three coolest ones ( σ obs ). The lasttwo columns, σ and σ , show the cluster’s intrinsic spreadsfor all stars and without the three coolest stars, respectively, ob-tained by correcting for the measurement uncertainty as: σ = σ obs − ǫ rand . (2)Note that Equation 2 gives an over-estimate of the intrinsic dis-persion, because the true systematic uncertainties were not re-moved. Figure 4 shows the interquartile ranges (IQR) and themedian values of the abundances we derived. The only siginif-icant intrinsic spreads were found for the light elements O, Na,Al, and the s-process element Ba. We also note the presenceof one K-deficient star, which, however, does not present anyanomalous O, Mg, Na, or Al abundances. The scatters of allother elements are compatible with the observational errors.A table containing all chemical element abundance ratioswith associated random errors for all individual stars is availablein the electronic version of the journal. A part of it is presentedin Table 5 to guide the eye. The column labeled N shows thenumber of lines used to derive the particular element abundanceand ǫ rand shows the random error. With this study we derived the first measurement of the Fe abun-dance of M 75 based on high-resolution spectroscopy as [FeI / H] = − . ± .
03 dex (random) ± .
08 dex (systematic) witha marginal 1 σ spread of 0 .
07 dex. Using Fe II lines, we ob-tained a higher value of [Fe II / H] = − . ± .
03 dex (random)
5. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75
Table 4.
Systematic abundance errors.
Ion ∆ T ef f ∆ log g ∆ ξ ∆ [M / H] ODF total +
60 K −
60 K + − + − + / s − − + + − − + + − − − + + − − + + − − + − + − − + + − − + − − + − + + − − + − − + + + − + − − − + − − + + − − + + + − + − − − + + − − + − − + − + + + − − + + − − + + − − + − + − − + + − − − + + − − + + − − + − + − − + + − − + − − − + + − − + + − − + + − + + − − + + − − + − + − − + + − − + + − − + + − − − + − − − + + + − + − − + + − − + + − − + + − − + − + − − + + − − + + − − + + − − + − + − − + + − − + − + − − + + − − + − + − − + + − − − + + − − + + − − + − + − − + − − − + + − − + + − − + − − + − + − + + − − − + + + − − − − + − + + − − + + − − + − + − − + − + − + − − − + − + − + + − − + + − − + − − + − + − − + + − − + + − − + − + + − + + + − + − + − + + − + − − + + − + − + − − + + − + − + − − + + − − + − + − − + + − − − + + − − + + − − − + + − − + + − − + − + − − + + − − + − + − − + + − − + − − + + − − + − + − − + + − − − + − − + + − − + − + − − + + − − − + + − − + + − − ± .
16 dex (systematic) with a 1 σ spread of 0 .
13 dex. Thereis no trend of the [Fe I / H] or [Fe II / H] values with tempera-ture, except for the three coolest stars, which have considerablyhigher [Fe II / H]. Excluding these three coolest stars, the meandiscrepancy between the Fe I and Fe II values becomes [Fe I / FeII] = − . ± .
02 dex, which is still significant but the larger scatter of Fe II abundances is reduced to the same value as theFe I scatter ( σ = .
07 dex).Both values are in an excellent agreement with the metal-licity derived by Catelan et al. (2002), based on UBV photom-etry of [Fe / H] = − . ± .
17 dex and [Fe / H] = − . ± .
21 dex on the metallicity scales of Carretta & Gratton (1997)and Zinn & West (1984), respectively.
6. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75
Table 5.
Derived abundance ratios for the individual stars of the GC. The full table is available in the online version of the journal.
Star ID [FeI / H] ǫ rand N [FeII / H] ǫ rand N [O / H] ǫ rand N [Na / H] ǫ rand N [Mg / H] ǫ rand N ...239 − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − − Fig. 4.
Boxplot of the derived abundances in M 75 relative toiron. Fe I and Fe II abundances are relative to the Fe I clustermean. Neutral species are relative to Fe I and ionised speciesare relative to Fe II. The boxes designate the median values andIQR. The error bars show the minimum and the maximum value.Outliers are shown with circles. An outlier is defined if it devi-ates by more than 1 . Table 6.
Average abundance ratios, average random and system-atic errors, observational, and intrinsic spreads. See text for de-tails.
Element [X / Fe] ǫ rand ǫ sys σ obs σ obs σ σ Fe I / H − / H − − − − − − The scatter of 0 .
07 dex may seem large compared to the bulkof GCs ( σ . .
05 dex; Carretta et al. 2009a) but it is fully con-sistent with the higher luminosity of M 75. With an absolutemagnitude of M V = − .
57 mag (Harris 1996), it is amongstthe most luminous and hence most massive GCs in the Milky
7. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75
Way (only 18 of the ∼
150 MW GCs are brighter). Carretta et al.(2009a) reported a dependence on the scatter in [Fe / H] with var-ious cluster parameters. In their homogenous sample of 19 GCsthey showed that the 1 σ dispersion correlates with the clusterluminosity (a proxy for the present mass), the maximum e ff ec-tive temperature reached on the HB, and anticorrelates with thelevel of α -enrichment. All these correlations point to a better ca-pability of more massive clusters to self-enrich with the ejecta ofmassive stars. The deeper gravitational potential helps retainingthe massive stars ejecta and the hotter HB stars indicate largerHe-content and thus, more processing by previous generation ofstars. Our sample of 16 stars, however, does not allow us to makeany firm conclusions on the link between the observed Fe-scatterin M 75 and its self-enrichment. The production of α -elements such as O, Ne, Mg, Si, S, Ca, andTi is mainly associated with the eruptions of SNe II. The dif-ferent timescales of the occurrence of SNe II and SN Ia makethe [ α / Fe] abundance ratios a powerful tool for diagnosing thechemical evolution and star formation history (SFH) of any stel-lar population (Tinsley 1979). In the Milky Way, the relativelymetal poor halo stars form a plateau of enhanced [ α/ Fe] ∼ + . / H] & − . α / Fe] ratio is associatedwith rapid star formation episodes that ceased before the long-lived SNe Ia, the main source of iron, began to enrich the lo-cal environment through their ejecta. Dwarf galaxies, on theother hand, have slower star formation rates so that low val-ues of [ α/ Fe] are observed already at low metallicities (e.g.,Shetrone et al. 2001, 2003; Tolstoy et al. 2009). Although dif-ferent α -elements are produced on similar timescales, they showelement-to-element variations due to di ff erent production mech-anisms, either through hydrostatic He-burning in the cores ofmassive stars (e.g., O and Mg), or during the SNe explosionsthemselves (e.g., Si, Ca, and Ti).In M 75, we derived O-abundances by measuring the EWsand by spectral synthesis of the 6300 Å and 6364 Å lines, whichare free of telluric contamination owing to the fortunate radialvelocity of this GC. The 6364 Å line is, however, situated in thewing of a broad Ca autoionisation feature, so in the synthesis ofthis line we adopted the derived Ca-abundance. Additionally, weadopted Solar C- and N-abundances but we confirmed the resultsfrom Koch et al. (2009) that demonstrated that molecular (CNO)equilibrium does not a ff ect the derived O-abundances at thesemetallicities and levels of O-enhancements. We also confirmedthat there are not any extreme variations in the strength of theCH G-band around 4320 Å.Mg-abundances were derived from the line at 5711 Å, Sifrom the 5684 Å, 5949 Å, and 6155 Å lines, Ca-abundances werebased on 11 features between 5250 Å and 6750 Å, and Ti wasmeasured from various Ti I and Ti II absorption lines. We foundthat all α -elements are enhanced with respect to the Sun to a dif-ferent extent. The average [ α / Fe] ratio is 0 . ± .
02 dex, basedon the Mg, Si, and Ca abundances. This is consistent with thecanonical value for the old stellar population of the Milky Way(halo field stars and the majority of its GCs; Pritzl et al. 2005).Oxygen is the only α -element, which shows significant varia-tions in its abundance amongst the stars in our sample. Thesevariations are discussed in the following section in terms of mul-tiple populations. Elements like Na, Al, and K are produced through proton-capture reactions at high temperatures during the H-burningin the cores of massive stars. The above are the elements re-sponsible for creating the unique chemical pattern of GCs (e.g.Denisenkov & Denisenkova 1989; Langer et al. 1993), also seethe review from Gratton et al. (2004). Large variations in theabundances of Na and Al have been so far detected in all GCsstudied to date (Gratton et al. 2012a). Nowadays, it is largely ac-cepted that these variations are due to the presence of at least twostellar populations in every GC characterised by slightly di ff er-ent ages and abundance patterns. Whilst both populations showthe same content of Fe-peak elements, the later formed stars arecharacterised by enhanced N, Na, and Al, along with depletedC, O, and possibly Mg. M 75 is not an exception in this respect.We measured Na-abundances from the three lines at 5682 . .
2, and 6160 . , , ff ects in the derivedNa abundances from some commonly used lines. According tothis study, however, in the regime of our stars (bright cool gi-ants) the NLTE corrections are expected to be small (in the order0 . − . ff erent correlations between the light el-ements. The O-Na anticorrelation is clearly visible. We dividedour sample into Primordial (P) population, characterised with O-rich and Na-poor stars, and Intermediate (I) population, charac-terised by O-poor and Na-rich stars, following the empirical sep-aration introduced by Carretta et al. (2009c). The four stars with[Na / Fe] < . / . .
8. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 − [O/Fe] − − − − − − − − − − − [Al/Fe][O/Fe] [Al/Fe] [ N a / F e ][ A l / F e ] [ N a / F e ][ M g / F e ] Fig. 5.
Correlations between the light elements; upper left: Na-Oanticorrelation; upper right: (no) Al-Mg anticorrelation; bottomleft: O-Al anticorrelation; bottom right: Al-Na correlation.of an Extreme (E) population in M 75, characterised by ex-tremely low O, high Na, and lower Mg abundances. Although, arough calculation shows that the Al enhancement of + . .
06 dex, thus there is no need to be a strong slope in the[Mg / Fe] vs. [Al / Fe] diagram. The large Al variation is, however,strongly correlated with Na and anticorrelated with O (Figure 5).Neither of these elements shows a trend with e ff ective temper-ature and we can consider these correlations genuine propertiesof the cluster. Potassium is not correlated with any of those el-ements and its marginal variations could be due to significanttemperature-dependent NLTE e ff ects in the strong 7699 Å line(Zhang et al. 2006). Thus, the [K / Fe] ratios show a slight trendof decreasing abundance with decreasing temperature.
All the odd-Z iron-peak elements V, Mn, Co, and Cu su ff erfrom significant HFS corrections. These corrections vary from0 . . ff erent elements, with the largest ef-fects for Cu. As a result, Mn and Cu are depleted with re-spect to iron by 0 . ± .
06 dex and 0 . ± .
13 dex, re-spectively. Such values are not unusual and are observed in anumber of stellar systems (Cayrel et al. 2004; McWilliam et al.2003; McWilliam & Smecker-Hane 2005) and a number ofGalactic GCs (e.g. Koch et al. 2009; Koch & Cˆot´e 2010). Cobaltis slightly overabundant with respect to iron with a value of[Co / Fe] = . ± .
06 dex. All even-Z elements (Cr, Ni, Zn)plus Sc and V trace the dominant iron production in the long-lived SNe Ia in that the [X / Fe] ratios are compatible with thesolar values ([X / Fe] ∼ . We derived chemical element abundance ratios for a large va-riety of n-capture elements mainly based on EW measurements,but we employed spectral synthesis for those elements, for whichonly few weak or highly blended lines were available. We usedthe Kurucz atomic database to obtain a blending list for our syn-thesis. The lighter n-capture elements Rb, Y, and Zr, usually as- sociated with the weak s-process, have [X / Fe] ratios close to theSolar values (Table 6). We note, however, that the associated ran-dom errors on these ratios are large, owing to di ffi culties in mea-suring them. For instance, Rb and Zr abundances are derivedthrough a spectral synthesis of the 7800 Å and 5112 Å lines,respectively, which accounts for the severe blending of theselines. Yttrium abundances are derived based on EWs of threelines found in the blue region of the red arm of MIKE, whichis generally characterised by a lower S / N ratio. We also did notaccount for HFS e ff ects associated with this odd-Z element dueto lack of HFS data, but the corrections are expected to be small(Prochaska et al. 2000).Barium is the only n-capture element, which presents in-trinsic variations significantly exceeding the random errors. Inparticular, there are two Ba-rich stars with [Ba / Fe] = + . ± .
10 dex and [Ba / Fe] = + . ± .
06 dex, whilst the mean [Ba / Fe]ratio is 0 . ± .
02 dex (Figure 8). Both stars are certainly notluminous and cool enough to have produced the s-process them-selves and the high Ba abundances are most likely due to enrich-ment from AGB stars in the cluster’s environment. The signifi-cance of these Ba-rich stars is further discussed in the Discussionsection. Our Ba abundances were derived mostly based on EWsof the easily accessible 5854 Å line, whilst the 6142 Å and6497 Å lines are saturated in most of our stars. For the warmerstars with higher gravity, where the latter lines are not saturated(reduced EW < − . ffi ciently strong lines,whilst the abundances for Gd, Dy, Er, Hf, and Th are derivedbased on spectral synthesis of one or several lines of thesespecies. We note that the associated errors of the latter elementsare large owing to the weakness of the lines and / or the low S / Nratios in their vicinity. One might also be interested if the Ba-enhanced stars also show variations in the other s-process el-ements La and Ce. These two stars show a marginal enhance-ment in their La abundance by ∼ σ above the cluster’s average[La / Fe] ratio and statistically insignificant Ce-enhancement byless than 1 σ above the cluster’s average [Ce / Fe] ratio. We ap-plied HFS corrections to the odd-Z elements La and Eu but therewere no available HFS data for Pr. We assumed that they arecomparable in magnitude to the HFS corrections for the othern-capture elements and thus very small.Besides Ba, Ce, and Hf, which show solar [X / Fe] ratios, allother heavy elements are enhanced with respect to the Sun; seeSection 5.1 for discussion.
5. Discussion
It is surprising that, at [Fe / H] = − .
16 dex, M 75 seems to beone of the rarer cases of GCs compatible with predominant r-process production of the n-capture elements (Figure 6, see alsoSneden et al. 2000; Yong et al. 2008; Koch et al. 2009, for M 15,M 5, and the distant GC Pal 3, respectively). In Figure 6 weshow the total r + s-process Solar curve and the pure r-processcurve from Burris et al. (2000) plotted over the mean n-captureelement abundances derived from our spectroscopic sample ofM 75. Both curves are normalised to the mean Ba-abundancefor an easier distinction. A χ test shows that the best fit to theproduction of the elements from Ba (Z =
56) to Th (Z =
90) isfound for a scaled Solar pure r-process enrichment plus an ad-mixture of 10% of the Solar s-process yields (Figure. 7; upper
9. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 panel). In this fit we included all stars in our sample but the twoBa-enhanced ones and we excluded the elements Rb, Y, Zr, andHf. Hafnium lies o ff the general pattern in all stars and we as-sumed that its abundances likely su ff er from a severe systematico ff set. The lighter n-capture elements Rb, Y, and Zr, on the otherhand, have very complicated production channels, which are notyet fully understood (Travaglio et al. 2004). For instance theyare associated with the weak s- and r-processes, which appear inmassive (M ∼ M ⊙ ) stars on similar timescales as the r-processproduction from SNe II (Raiteri et al. 1993). We conclude thatonly a small number of AGB stars have contributed to the en-richment of the primordial cloud from which M 75 formed.We note that there is not any di ff erence in the n-capture el-ement abundance pattern between the P- and I-generations inour sample with the exception of the two Ba-enhanced stars. Wefound the best χ -fit for both Ba-rich stars to be scaled solarr- plus an admixture of 50% and 100% of the scaled Solar s-process yields, respectively. The abundance pattern of the Ba-rich star χ tests to the abundances of all Ba-normal stars, averaged together and the two Ba-rich stars - sep-arately. The results show that the Ba-normal stars have againexperienced only 10% of the solar s-process and the two Ba-richones 40% and 70%, respectively.Since the s-process enhancement is usually associated withAGB stars, this can be seen as evidence that FRMS were themain polluters, which ejecta formed the intermediate stellar pop-ulation. One should note, however, that only the most massive( ∼ − M ⊙ ) AGB stars reach the necessary high ( > × K)temperatures at the bottom of the convective envelope to acti-vate the ON cycle and thus to reduce the O abundance. TheseAGB stars experience only a few dredge-up processes, whichcannot alter the s-process abundances (D’Ercole et al. 2008).Thus, the AGB scenario cannot be ruled out. In fact, it is evenmore favourable, considering the presence of 2 Ba-rich stars, dis-cussed in the previous section. In conclusion, the heavy elementswere not significantly modified during the self-enrichment pro-cesses in the early cluster evolution (with the noted exceptionof the few most Na-rich stars) but their pattern is genuine to thecloud, from which M 75 formed.
Despite the large formal errors of the measured Th abundance,the star-to-star variations of Th are small, which leads to a pre-cise, mean Th abundance for the cluster. We used the [Th / Eu]ratio to derive an approximate age estimate for M 75 based onthe radioactive decay of Th. The mean log ǫ (Th / Eu) is − . ± .
02 dex, or N Th / N Eu = .
28. Whilst Eu is a stable element, thehalf life period ( t / ) of the isotope Th is 1 . × yr. Usingthe universal law of radioactive decay, namely N ( t ) = N e − λ t ,where λ = ln(2) / t / , and the Solar System initial Th / Eu ra-tio, N Th / N Eu = .
46, (Cowan et al. 1999), we obtained an ageestimate of 10 Gyr. This method usually leads to a precisionof ± N Th / N Eu = .
25 for M 15, which leads to an age of 12 .
50 60 70 80 90
Atomic Number−1.0−0.50.00.51.01.52.0 l og (cid:2) solar r+spure r−process Rb YZr BaLaCePrNdSmEuGdDy Er Hf Th
Fig. 6.
Mean neutron capture element abundances for all starsin M 75, normalized to Ba. The lines display the scaled solarpure r- and r + s-process contributions from Burris et al. (2000).The uncertainties represent the 1 σ scatter of the derived elementabundances in all stars.in the Th abundance can still lead to large errors in the age es-timate. For instance, a systematic error of 0 . ff erence in age. More accurate spectroscopic age-dating would be feasible once U-abundances can be measured(Frebel et al. 2007), but given the low S / N ratios in the relevantblue regions, this is an unlikely endeavour in the remote M 75.
The phenomenon of multiple populations in GCs often extendsto the appearance of the CMD. High-precision photometry hasrevealed multiple main sequences, subgiant branches, and RGBsin the CMDs of many clusters, which do not exhibit large vari-ations in metallicity (Piotto et al. 2007, 2012; Han et al. 2009).These e ff ects are mostly driven by CNO variations and di ff erentHe-content in stars from di ff erent generations (D’Antona et al.2002, 2005; Piotto et al. 2005). Moreover, the presence of mul-tiple populations has been proven to be one of the key parame-ters that shapes the HB, where the e ff ect is most pronounced,because stars of the same age but di ff erent He-content havedi ff erent initial masses and, thus, occupy di ff erent regions ofthe HB. Whilst the stars with primordial He-abundance (com-patible with the Big Bang nucleosynthesis) preferably populatethe red part of the HB, stars with enhanced He are responsiblefor the formation of extended blue tails (D’Antona et al. 2002;D’Antona & Caloi 2004). Strong correlations between the He-content and the abundances of p-capture elements with the e ff ec-tive temperature amongst stars from the HB have been recentlyfound in M 4 (Marino et al. 2011), NGC 2808 (Gratton et al.2011), and NGC 1851 (Gratton et al. 2012b). M 75 is one ofthe most curious cases in this respect. It has a very extended andpeculiar HB with a trimodal distribution (Catelan et al. 2002).Apart from the well separated BHB and RHB, its CMD showsa distinct third extension of a very blue, faint tail (Figure 1, seealso Figure 2 in Catelan et al. 2002). The extended blue tail ofM 75’s HB is at odds with our findings of only a moderate Na-Oanticorrelation on the RGB (Figure 5), which so far lacks an ex-treme (E) population. Such a population is often found in GCswith extended HBs (Carretta et al. 2009c) and characterised by
10. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 −0.50.00.51.01.52.02.5 l og (cid:0) s(cid:1)(cid:3)(cid:4)(cid:5)(cid:6) (cid:7)(cid:8)(cid:9)(cid:10)(cid:11) (cid:12) (cid:13) (cid:14)(cid:15)(cid:16) (cid:17) −process (cid:18)(cid:19)(cid:20)(cid:21)(cid:22)(cid:23) (cid:24)(cid:25)(cid:26)(cid:27)(cid:28) (cid:29)(cid:30)(cid:31) !"
6’ 7( 8) 9*A+,-./ 01235: −0.50.00.51.01.52.0 l og ; <=>?@B CDEFG HIJK L −process MNOPQR STUVW XYZ [\]^_‘
Rb YZr BaLaCePrNdSmEuGdDy Er Hf ThRb YZr BaLaCePrNdSmEuGdDy Er Hf Th
Fig. 7.
Upper panel: Mean neutron-capture element abundancepattern for all “Ba-normal” (s-process deficient) stars in M 75.The solid line represents the best fit model for a scaled solarr-process plus 10% of the scaled solar s-process yields. Thedashed line shows the scaled solar total r + s-process yields forcomparison. The error bars represent the 1 σ scatter of the de-rived element abundances of all s-process deficient stars; Bottompanel: Derived n-capture elements abundances for star + s-process yields.The dashed line shows the scaled solar pure r-process model forcomparison. The error bars represent the random errors for thisparticular star.extremely Na-rich and O-poor stars, and accompanied by largeHe variations. Having in mind, however, the tiny populated ex-tremely blue tail in M 75 and the limited number of our sample,it is possible that such E population has just been missed by ourselection criteria. But curiously, we do not detect a Mg-Al anti-correlation, neither, which is commonly found in those GCs witha more complex HB morphology. Furthermore, clues for He-,CNO-, or age-variations have not been detected in the CMD ofM 75 (in terms of multiple RGBs, subgiant branches, or mainsequences). We note however, that there is not any narrow-bandphotometry available for this GC, which might reveal multiplepopulations amongst the main sequence, the subgiant branch andthe RGB (e.g. Carretta et al. 2011a). Still, the distribution of Heand p-capture elements amongst the HB of M 75 remains anopen question and thus, high-resolution spectroscopic observa- tions amongst stars on the HBs are needed to ascertain if its pe-culiar morphology is mostly driven by the presence of multiplepopulations or if there are other parameters with major influ-ence. Possibly, the younger age and higher concentration, com-plemented with its remote location in the Milky Way halo holdimportant clues about its origin. A closer look to the Na-O and Al-O anticorrelations of M 75(Figure 5) shows three distinct populations rather than a contin-uous anticorrelation, as found in most GCs (the P-population,which consists of stars with Na- and O-abundances typical ofthe halo field; a second group of stars mildly enriched in Naand depleted in O; and a third group of the most Na-rich, O-depleted stars, which is separated by a clear gap from the sec-ond group). But the question whether the Na-O anticorrelationin GCs is actually continuous or rather discrete has recentlyraised attention because the findings of discrete main sequencesin some GCs (D’Antona et al. 2005; Piotto et al. 2007) and thediscrete distribution of the Na and O abundances amongst theHB (Marino et al. 2011) suggest a discrete chemical distributionof the di ff erent populations also amongst the RGB. Currently,more precise, high-resolution studies of large number of GCs’RGB stars are being carried out to clear out this question (e.g.Carretta et al. 2012). Besides the three populations seen in theNa-O plane, M 75 also hosts a number of Ba-rich stars, whichcould represent a fourth population.A possible explanation of the formation of four populationscould be found within the pristine gas dilution scenario sug-gested by D’Ercole et al. (2008, 2011). On a timescale of sev-eral tens of Myr after the P-generation has formed, the centreof the cluster still hosts only gas from the higher mass AGBstars’ ejecta. At this point, the Na-rich, O-depleted, Ba-normalpopulation is formed. In the next ∼
10 Myr, pristine gas fallsin the central region of the cluster and mixes with the gas en-riched by the AGB winds. Another population forms from thediluted gas, which is mildly Na-rich and O-depleted. After thediluted gas is fully processed, the lower mass AGB stars re-main the only source of gas in the cluster. Their ejecta could alsobe enhanced in s-process elements. The last Ba-rich populationforms, which is also strongly Na-rich and O-depleted and indis-tinct from the other stars formed from none-diluted gas (Figure8). Hydrodynamical and N-body simulations can test the viabil-ity of this idea and better constraint the time-scales and polluters’masses. Meanwhile, some improvements of the AGB models arealso needed, since we do not yet fully understand these very lateevolutionary stages in terms of mass-loss, convection, and nu-clear reaction cross-sections.
Now we can place some qualitative constraints on the averagemass of the polluters in M 75 provided that enrichment wasdominantly from AGB stars. Despite its high luminosity, whichis often considered as a reason for the presence of higher masspolluters, there are several factors prompting that lower massAGB stars also contributed to the formation of the intermediatepopulation in this GC:i) Whilst the maximum amount of Na produced in mostGCs is approximately the same, high mass AGB stars man-age to process more O through the ON cycle and, hence, theyare more depleted in this element. Therefore, the amount of
11. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 −0.2 0.0 0.2 0.4 0.6 0.8[Na/Fe]−0.4−0.20.00.20.40.60.8 [ B a / F e ] Fig. 8. [Ba / Fe] vs. [Na / Fe] ratios in M 75’s RGB stars. The twomost Na-rich stars are also Ba-enhanced by 0 . ± .
10 dex and0 . ± .
06 dex, respectively.O-depletion is a proxy for the polluters’ average mass in thesense that more extended Na-O anticorrelations imply highermass polluters (see Carretta et al. 2009c). For instance, the Na-O anticorrelation in M 75 is less extended than in other GCsof similar luminosity (e.g. NGC 1851, NGC 2808, M 5) andmore similar to the less massive M 4, prompting for lowermass polluters in M 75. Another example of a massive GCthat does not have an extended Na-O anticorrelation is 47 Tuc(M V = − .
42 mag). This GC, however hosts significantlymore O-depleted stars compared to M 75 and suggests predomi-nantly high-mass AGB polluters, in accordance with the conclu-sions drawn by Carretta et al. (2012). This is best represented inFigure 9, where we plot the Na-O anticorrelations of M 75 andthe three reference clusters – NGC 1851, M 4, and 47 Tuc. Wealso show simple dilution models that we computed as describedin Carretta et al. (2009c), tuning by eye the input parameters (themaximum and minimum Na- and O-abundance ratios) for all ofthem. Figure 9 clearly shows that the Na-O anticorrelation ofM 75 is more similar to the less massive GC M 4 compared tothe extended anticorrelation of NGC 1851 and “steeper” thanthe Na-O anticorrelation of 47 Tuc. Thus, we suggest that theaverage mass of the polluters di ff ers from cluster to cluster, re-gardless of its luminosity. M 75 and M 4 were likely enriched bythe ejecta of lower mass AGB stars, whilst 47 Tuc was mainlyenriched by more massive AGB stars. GCs with very extendedNa-O anticorrelations (e.g. M 5, M22, NGC 2808, NGC 1851)most likely experienced continuous star formation ( ∼
100 Myr)and were enriched by polluters of broader mass range.ii) Furthermore, the two stars, which present anomalouslyhigh Ba abundances are also the most Na-rich ones (Figure8). This is indicative of s-process enrichment from intermedi-ate mass AGB stars (M < − ⊙ ), that are able to alter thes-process pattern of the I-population stars (Gallino et al. 1998).Ba-rich stars are very rare in GCs but have also been found inNGC 1851, M 22, and ω Cen, which are all noted for their com-plex evolution, likely characterised by longer star formation pe-riod extending after these low mass stars manage to pollute thecluster environment with s-process enhanced ejecta.iii) The [Rb / Zr] ratio can also be a proxy for the polluters’mass because it is sensitive to the neutron density. The main neu- −0 a b c d [ N a / F e ] NGC 1851M 4M 75−0.6 −0.4 −0.2 0.0 0.2 0.4 0.6 0.8[O/Fe]−0.50.00.51.0 [ N a / F e ] M 7547 Tuc
Fig. 9.
A comparison of the extent of the Na-O anticorrelationsof several GCs relative to M 75; Top panel: the Na-O anticor-relations in M 75, NGC 1851 (Carretta et al. 2011b), and M 4(Carretta et al. 2009c); Bottom panel: the Na-O anticorrelationsin M 75 and 47 Tuc (Carretta et al. 2009c). Simple dilution mod-els are overimposed. The [O / Fe] abundances of M 75 are shiftedby the mean discrepancy of [Fe I / Fe II] = − .
14 dex to match theother studies.tron source in the He shell of low mass (1 < M < ⊙ ) AGBstars is the reaction C( α, n) O, whilst in more massive AGBstars, neutrons are mainly released by Ne( α, n) Mg and sincethe Ne source produces much higher neutron densities than the C neutron source, the [Rb / Zr] ratio can discriminate betweenthe two (Garc´ıa-Hern´andez et al. 2009). The mean [Rb / Zr] ratioof our sample of M 75 stars is = − . ± .
03 dex and there arenot any significant star-to-star variations in our sample of stars.Both Ba-rich stars have a [Rb / Zr] = . ± . ff erence with respect to the mean[Rb / Zr] ratio. We conclude that we cannot use the [Rb / Zr] ra-tio to discriminate between di ff erent AGB masses, owing to thelarge uncertainties of this ratio in individual stars. In order to investigate M 75’s origin, we present in Figure 10a comparison of the abundances in M 75 with the abundancesof Galactic disk and halo stars at di ff erent metallicities, and
12. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 with the average abundances of other GCs. A small represen-tative sample of individual dSph stars is also plotted for com-parison. The abundances of the Milky Way halo and disk starsare taken from the compilation of Venn et al. (2004, and ref-erences therein) and complemented with the recent results ofIshigaki et al. (2012a,b). The sample of dSph stars also comesfrom Venn et al. (2004) and includes the abundances of individ-ual stars from Carina, Fornax, Leo, Scl, UMi, Sex, and DracodSphs. The mean abundances of various GCs are taken fromPritzl et al. (2005) and complemented with the more recent re-sults for NGC 1851 (Carretta et al. 2011b), M 5 (Yong et al.2008), and the outermost halo clusters Pal 3 (Koch et al. 2009),and Pal 4 (Koch & Cˆot´e 2010).We chose to plot in Figure 10 three key element ratios, im-portant to trace the chemical evolution of M 75. The α -elementabundance ratio in M 75 is fully compatible with the Galactichalo stars at the same metallicity and consistent with the α -enhanced old stellar populations of the Milky Way halo. Thissuggests that it experienced the same (fast) star formation his-tory, dominated by SNe II and only late, delayed Fe enrichmentby SNe Ia, as most MW GCs. A connection with the dSph galax-ies and their low star formation rates (hence low [ α / Fe]) can beruled out. The latter scenario has been suggested for some GCswith low α -abundance like Pal 12, Ruprecht 106, and possiblyTer 7, associated with the Sgr dwarf (Pritzl et al. 2005).The [Ba / Y] ratio compares the abundance yields betweenthe main s-process, which takes place in intermediate- to low-mass AGB stars and the weak s-process, which is associatedwith very massive ( M > M ⊙ ) stars (Burris et al. 2000). TheGalactic halo and disk stars have a roughly solar [Ba / Y] ratioover a broad range of metallicities, which starts to decrease at[Fe / H] . − / Y] ratio israther high (Venn et al. 2004; Tolstoy et al. 2009), owing to theirtypically lower star formation rates. However, all GCs, includingM 75, present typical of the halo field [Ba / Y] ratios around thesolar value, with again, the notable exception of Pal 12, whichhas unusually high mean [Ba / Y] ratio, more similar to the starsfrom dSph galaxies.Finally, the [Ba / Eu] ratio compares the yields from the mains- and main r- neutron capture processes. The latter operatesmainly in massive stars during the eruptions of SNe II. Typically,all stars in the field and in GCs present abundances that are con-sistent with r-process enrichment plus some fraction of the solars-process contribution from AGB stars. The s-process fractionvaries for di ff erent stars but the general trend is that it could beentirely missing for metal poor stars and rises until it becomesdominant for the metal rich population. As we noted above, M 75has an unusually low, mean s-process contribution for its metal-licity, but its [Ba / Eu] = − . ± .
05 dex ( − . ± .
01 dex ifwe consider only the P-generation) is still consistent with somehalo stars. Only a few GCs, studied to date, have been notedto present [Ba / Eu] ratios compatible or lower than M 75. Themajority of them are amongst the most metal poor GCs in ourGalaxy with [Fe / H] below − . / Eu] = − .
50 (Lee et al. 2005), M 92, [Ba / Eu] = − .
55 (Shetrone et al.2001), M 15, [Ba / Eu] = − .
87 (Sneden et al. 2000), and M 30,[Ba / Eu] = − .
53 (Shetrone et al. 2003). On the more metal-richend, we note the GCs NGC 3201 with [Fe / H] = − .
58 dexand [Ba / Eu] = − .
54 dex (Gonzalez & Wallerstein 1998), Pal 3with [Fe / H] = − .
52 dex and [Ba / Eu] = − .
73 dex (Koch et al.2009), and M 5 with [Fe / H] = − .
30 dex (most similar to M 75)and [Ba / Eu] = − .
60 dex (Ram´ırez & Cohen 2003; Yong et al.2008). −0.20.00.2 [ α / F e ] − − − [ B a / Y ] − − − − − − [ B a / E u ] pure r-processSolar r+s Fig. 10.
A comparison of the α and n-capture element abun-dances of the 16 stars of M 75 (blue asterisks) with Galacticdisk and halo stars (grey crosses), average abundance values ofother Galacatic GCs (filled green circles), and a representativesample of individual dSph stars (open magenta circles).NGC 1851 is probably the GC that shares most com-mon properties with M 75 and is often thought as its twin(Catelan et al. 2002). Here, we investigate the similarities anddi ff erences between the two objects in deeper detail. Both GCsare coeval, share the same metallicity, and show similar HBmorphology. Both are luminous, massive and very concentratedclusters located in the transition region between the inner andouter Milky Way halo. A notable di ff erence between the two ob-jects is the presence of double subgiant and red giant branchesin NGC 1851 (see Milone et al. 2008; Han et al. 2009), whichare not observed in M 75, despite the same photometric qual-ity. Ventura et al. (2009) have suggested that large CNO varia-tions and a small age spread could explain the subgiant branchin NGC 1851. This scenario is supported by Yong et al. (2009)from observations of 4 RGB stars. However, it has been dis-missed by more recent studies by Villanova et al. (2010) of 15RGB stars and by Gratton et al. (2012b), who derived abun-dances for a large sample of HB stars. Both studies found noevidence for significant CNO variations. Gratton et al. (2012b)concluded that the only explanation of the splitting of the SGBand RGB of NGC 1851 is a considerable age di ff erence of about1 . ff erent ages. In any case,the clearly more complex formation history of NGC 1851 is re-sponsible for the very extended HB in this system. In the caseof M 75, it is not yet clear what physical processes drive theformation of such an extended HB.The most comprehensive chemical study of NGC 1851, interms of derived abundances for many di ff erent elements, ispresented in Carretta et al. (2011b). We used it to compare the
13. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75
Table 7.
Comparison with GCs that share some common properties with M 75M 75 NGC 1851 M 4 M 5 47 Tuc Pal 3R GC (kpc) h (arcmin) c = log( r t / r c ) ∼
10 9 . . . . V1 − . − . − . − . − . − . / H] − . − . − . − . − . − . α/ Fe] / Eu] − . − .
19 0 . − .
60 0 . − . / Y] .
04 0 . − . − .
29 0 . − . Notes. (1)
Data taken from Harris (1996, 2010 version) (2)
Relative ages from Mar´ın-Franch et al. (2009). We adopted a reference age of 13 Gyr. (3)
Various sources cited throughout the text. − efg hijk lmn
25 −1.20 −1.15 −1.10 −1.05 −1.00[Fe/H]−1.0−0.8−0.6−0.4−0.20.00.20.4 [ B aLa C e / E u D y ] pure r-process10% s-processsolar r+s-process Fig. 11.
A comparison between the r- and s-process enrichmentin M 75 (blue dots) and NGC 1851 (red dots). The figure clearlyreveals the di ff erent enrichment history between the two GCs.chemical abundances in both clusters. They share the samemetallicity (with no evidence of significant iron spreads in nei-ther of them) and similar α -enhancement. The p-capture ele-ments Na and Al have similar variations in both clusters but Oshows a larger spread in NGC 1851, leading to a more extendedNa-O anticorrelation in the latter GC (See Figure 9). The iron-peak element abundances are identical in both GCs. The largestdi ff erence between the two lies in the n-capture elements. Thisis best illustrated in Figure 11, where we present the total s- tor-process ratio in both GCs. We chose the average abundance ofBa, La, and Ce as representatives of typical s-process elements,where about 80% of their production comes from the s-process,and the average abundance of Eu and Dy as representatives ofelements produced mainly by the r-process (Burris et al. 2000).The stars of NGC 1851 clearly lie above those from M 75 in thisparameter space, indicating di ff erent primordial s-process con-tribution. We note, however, that both M 75 and NGC 1851 hosts-process rich stars, which were enhanced in s-process elementsmost probably by intermediate mass AGB stars during the earlyevolution of these stellar systems via self enrichment mecha-nisms, but the s-process rich stars in M 75 reach [s / r] ratios sim-ilar to the primordial s-process enhancement of NGC 1851. Finally, we note that M 75 is both a unique and a normal GCof the Milky Way’s GCs system. Unique, in the sense that thereis not a GC, which resembles all the same properties of M 75,and normal in the sense that its properties fit well in the gen-eral picture of the Milky Way’s GCs. So far there are not twoclusters found to be exactly alike and each one of them deservesspecial attention. In Table 7 we present some important charac-teristics of M 75 compared to other GCs, which were discussedin this section and that share some important similarities and dif-ferences with M 75.
6. Summary
In this work, we presented the first chemical abundance study ofthe outer halo GC M 75. Our data sample consists of high res-olution spectra of 16 giant stars, obtained with the MIKE spec-trograph at the Magellan Observatory. We derived abundancesthrough EW measurements and spectral synthesis in LTE fora total of 32 di ff erent elements covering a broad range of p-capture, α , iron-peak, and n-capture elements. M 75 is moder-ately metal rich cluster with [Fe / H] = − . ± .
02 dex witha marginal spread of 0 .
07 dex, typical for GCs with similarluminosity. We measured an enhanced average α abundance[ α / Fe] = . ± .
02 dex, based on Mg, Si, and Ca, typical forthe Galactic halo at this metallicity. We found significant vari-ations in the abundances of the p-capture elements O, Na, andAl, which provide evidence for the presence of at least two gen-erations, formed on a short time-scale. Sodium is anticorrelatedwith O and correlated with Al, consistent with simple dilutionmodels. The Na-O anticorrelation appears discrete, suggestingthree chemically distinct populations. Additionally, the two mostNa-rich stars form a fourth, Ba-enhanced population. Based onthe extent of the Na-O anticorrelation, we conclude that the I-population stars were enriched by the ejecta of relatively lessmassive AGB stars in several episodes of star formation, whichended before the SNe Ia began to contribute iron to the cluster’senvironment. We note that the least massive polluters were ableto alter the s-process abundances of the cluster’s ISM.The moderate O-Na anticorrelation (our sample of 16 starslacks an extreme population of stars with very low O abun-dances) and the lack of significant Mg variation are at odd withthe very extended trimodal HB of M 75. We conclude that the pa-rameters that shape the peculiar HB morphology of this GC arestill unclear and more observations are required, in particular aspectroscopic sample of stars, which represents the full span ofthe HB. A careful CNO abundance analysis of the existing spec-
14. Kacharov et al.: A comprehensive chemical abundance study of the outer halo globular cluster M 75 troscopic sample is foreseen in a subsequent paper (Kacharov etal., in prep.).The n-capture element pattern is consistent with predomi-nant r-process enrichment with a marginal contribution (about10% of the scaled solar production) of s-process.The overall chemical, evolutionary status of M 75 is con-sistent with other inner and outer halo GCs and field stars,which suggest a similar origin with the bulk of Milky Way GCs.Despite its large galactocentric distance, coupled with its highmetallicity and younger age, M 75 does not seem to present anyodd, chemical properties, that would indicate extragalactic ori-gin and accretion to the Milky Way halo at a later stage of itsevolution.
Acknowledgements.
The authors thank Miho Ishigaki for providing tables withabundances of heavy elements in the Galactic halo and Ra ff aele Gratton andFrancesca D’Antona for helpful discussions on multiple populations in GCs. NKand AK acknowledge the Deutsche Forschungsgemeinschaft for funding fromEmmy-Noether grant Ko 4161 / References
Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481Bergemann, M., Lind, K., Collet, R., Magic, Z., & Asplund, M. 2012, MNRAS,427, 27B¨oker, T. 2008, ApJ, 672, 111Burris, D. L., Pilachowski, C. A., Armandro ff , T. E., et al. 2000, ApJ, 544, 302Carollo, D., Beers, T. C., Lee, Y. S., et al. 2007, Nature, 450, 1020Carretta, E., Bragaglia, A., Gratton, R., D’Orazi, V., & Lucatello, S. 2009a,A&A, 508, 695Carretta, E., Bragaglia, A., Gratton, R., D’Orazi, V., & Lucatello, S. 2011a,A&A, 535, 121Carretta, E., Bragaglia, A., Gratton, R., & Lucatello, S. 2009b, A&A, 505, 139Carretta, E., Bragaglia, A., Gratton, R. G., et al. 2009c, A&A, 505, 117Carretta, E., Bragaglia, A., Gratton, R. G., et al. 2010a, A&A, 516, 55Carretta, E. & Gratton, R. 1997, A&AS, 121, 95Carretta, E., Gratton, R., Bragaglia, A., D’Orazi, V., & Lucatello, S. 2012, A&A,2012arXiv1212.1169CCarretta, E., Gratton, R. G., Lucatello, S., et al. 2010b, ApJ, 722, 1Carretta, E., Lucatello, S., Gratton, R. G., Bragaglia, A., & D’Orazi, V. 2011b,A&A, 533, 69Castelli, F. & Kurucz, R. L. 2003, IAUS, 210, 20Catelan, M. 2009, Ap&SS, 320, 261Catelan, M., Borissova, J., Ferraro, F. R., et al. 2002, AJ, 124, 364Cayrel, R., Depagne, E., Spite, M., et al. 2004, A&A, 416, 1117Cohen, J. G. & Mel´endez, J. 2005, AJ, 129, 303Cottrell, P. L.; Da Costa, G. S. 1981, ApJ, 245, 79Cowan, J. J., Pfei ff er, B., Kratz, K.-L., et al. 1999, ApJ, 521, 194Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003, VizieR Online DataCatalog, 2246D’Antona, F., Bellazzini, M., Caloi, V., et al. 2005, ApJ, 631, 868D’Antona, F. & Caloi, V. 2004, ApJ, 611, 871D’Antona, F., Caloi, V., Montalbn, J., Ventura, P., & Gratton, R. 2002, A&A,395, 69De Mink, S. E., Pols, O. R., Langer, N., & Izzard, R. G. 2009, A&A, 507, 1Decressin, T., Meynet, G., Charbonnel, C., Prantzos, N., & Ekstrm, S. 2007,A&A, 464, 1029Denisenkov, P. A. & Denisenkova, S. N. 1989, A.Tsir., 1538, 11D’Ercole, A., D’Antona, F., Ventura, P., Vesperini, E., & McMillan, S. L. W.2010, MNRAS, 407, 854D’Ercole, A., D’Antona, F., & Vesperini, E. 2011, MNRAS, 415, 1304D’Ercole, A., Vesperini, E., D’Antona, F., McMillan, S. L. W., & Recchi, S.2008, MNRAS, 391, 825Fabbian, D., Nissen, P. E., Asplund, M., Pettini, M., & Akerman, C. 2009, A&A,500, 1143Frebel, A., Christlieb, N., Norris, J. E., et al. 2007, ApJ, 660, 117Gallino, R., Arlandini, C., Busso, M., et al. 1998, ApJ, 497, 388Garc´ıa-Hern´andez, D. A., Manchado, A., Lambert, D. L., et al. 2009, ApJ, 705,31Geisler, D., Wallerstein, G., Smith, V. V., & Casetti-Dinescu, D. I. 2007, PASP,119, 939Girardi, L., Williams, B. F., Gilbert, K. M., et al. 2010, ApJ, 724, 1030Gonzalez, G. & Wallerstein, G. 1998, AJ, 116, 765Gratton, R., Sneden, C., & Carretta, E. 2004, ARA&A, 42, 385 Gratton, R. G., Bonifacio, P., Bragaglia, A., et al. 2001, A&A, 369, 87Gratton, R. G., Carretta, E., & Bragaglia, A. 2012a, A&ARv, 20, 50Gratton, R. G., Lucatello, S., Carretta, E., et al. 2012b, A&A, 539, 19Gratton, R. G., Lucatello, S., Carretta, E., et al. 2011, A&A, 534, 123Han, S.-I., Lee, Y.-W., Joo, S.-J., et al. 2009, ApJ, 707, 190Harris, W. E. 1996, AJ, 112, 1487Heiter, U. & Eriksson, K. 2006, A&A, 452, 1039Ishigaki, M. N., Aoki, W., & Chiba, M. 2012a, AIPC, 1480, 379Ishigaki, M. N., Chiba, M., & Aoki, W. 2012b, ApJ, 753, 64Kelson, D. D. 2003, PASP, 115, 688Kelson, D. D., Illingworth, G. D., van Dokkum, P. G., & Franx, M. 2000, ApJ,531, 159Koch, A. & Cˆot´e, P. 2010, A&A, 517, 59Koch, A., Cˆot´e, P., & McWilliam, A. 2009, A&A, 506, 729Kraft, R. P. 1994, PASP, 106, 553Kravtsov, A. V. & Gnedin, O. Y. 2005, ApJ, 623, 650Kravtsov, V., Alcano, G., Marconi, G., & Alvarado, F. 2007, A&A, 469, 529Langer, G. E., Ho ff man, R., & Sneden, C. 1993, PASP, 105, 301Lee, J.-W., Carney, B. W., & Habgood, M. J. 2005, AJ, 129, 251Lind, K., Asplund, M., Barklem, P. S., & Belyaev, A. K. 2011, A&A, 528, 103Marigo, P., Girardi, L., Bressan, A., et al. 2008, A&A, 482, 883Mar´ın-Franch, A., Aparicio, A., Piotto, G., et al. 2009, ApJ, 694, 1498Marino, A. F., Villanova, S., Milone, A. P., et al. 2011, ApJ, 730, 16Martell, S. L. & Grebel, E. K. 2010, A&A, 519, A14McWilliam, A., Preston, G. W., Sneden, C., & Searle, L. 1995, AJ, 109, 2757McWilliam, A., Rich, R. M., & Smecker-Hane, T. A. 2003, ApJ, 592, 21McWilliam, A. & Smecker-Hane, T. A. 2005, ApJ, 622, 29McWilliam, A., Wallerstein, G., & Mottini, M. 2012, in preparationMilone, A. P., Bedin, L. R., Piotto, G., et al. 2008, ApJ, 673, 241Osborn, W. 1971, OBS, 91, 223Piotto, G., Bedin, L. R., Anderson, J., et al. 2007, ApJ, 661, 53Piotto, G., Milone, A. P., Anderson, J., et al. 2012, ApJ, 760, 39Piotto, G., Villanova, S., Bedin, L. R., et al. 2005, ApJ, 621, 777Pritzl, B. J., Venn, A. K., & Irwin, M. 2005, AJ, 130, 2140Prochaska, J. X., Naumov, S. O., Carney, B. W., McWilliam, A., & Wolfe, A. M.2000, AJ, 120, 2513Pryor, C. & Meylan, G. 1993, APS Conf. Ser., 50, 357Raiteri, C. M., Gallino, R., Busso, M., Neuberger, D., & Kaeppeler, F. 1993, ApJ,419, 207Ram´ırez, I. & Mel´endez, J. 2005, ApJ, 626, 465Ram´ırez, S. V. & Cohen, J. G. 2003, AJ, 125, 224Rodgers, A. W. & Paltoglou, G. 1984, ApJ, 283, 5Ruchti, G. R., Bergemann, M., Serenelli, A., Casagrande, L., & Lind, K. 2013,MNRAS, 429, 126Sadakane, K., Arimoto, N., Ikuta, C., et al. 2004, PASJ, 56, 1041Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525Searle, L. & Zinn, R. 1978, ApJ, 225, 357Shetrone, M., Venn, K. A., Tolstoy, E., et al. 2003, AJ, 125, 684Shetrone, M. D., Cˆot´e, P., & Sargent, W. L. W. 2001, ApJ, 548, 592Sneden, C., Johnson, J., Kraft, R. P., et al. 2000, ApJ, 536, 85Sneden, C. A. 1973, PhD thesis, the University of Texas at AustinTinsley, B. M. 1979, ApJ, 229, 1046Tolstoy, E., Hill, V., & Tosi, M. 2009, ARA&A, 47, 371Travaglio, C., Gallino, R., Arnone, E., et al. 2004, ApJ, 601, 864Valcarce, A. A. R. & Catelan, M. 2011, A&A, 533, 120Venn, K. A., Irwin, M., Shetrone, M. D., et al. 2004, AJ, 128, 1177Ventura, P., Caloi, V., D’Antona, F., et al. 2009, MNRAS, 399, 934Villanova, S., Geisler, D., & Piotto, G. 2010, ApJ, 722, 18Worley, C. C. & Cottrell, P. L. 2010, MNRAS, 406, 2504Worley, C. C., Cottrell, P. L., McDonald, I., & van Loon, J. T. 2010, MNRAS,402, 2060Yong, D., Grundahl, F., D’Antona, F., et al. 2009, ApJ, 695, 62Yong, D., Karakas, A. I., Lambert, D. L., Chie ffi , A., & Limongi, M. 2008, ApJ,689, 1031Zhang, H. W., Gehren, T., Butler, K., Shi, J. R., & Zhao, G. 2006, A&A, 457,645Zinn, R. 1993, ASPC, 48, 38Zinn, R. 1996, ASPC, 92, 211Zinn, R. & West, M. J. 1984, ApJS, 55, 45, A., & Limongi, M. 2008, ApJ,689, 1031Zhang, H. W., Gehren, T., Butler, K., Shi, J. R., & Zhao, G. 2006, A&A, 457,645Zinn, R. 1993, ASPC, 48, 38Zinn, R. 1996, ASPC, 92, 211Zinn, R. & West, M. J. 1984, ApJS, 55, 45