A high-resolution spectroscopic study of two new Na- and Al-rich field giants -- likely globular cluster escapees in the Galactic halo
Avrajit Bandyopadhyay, Sivarani Thirupathi, Timothy C. Beers, A.Susmitha
MMNRAS , 1–8 (2020) Preprint 10 March 2020 Compiled using MNRAS LaTEX style file v3.0
A high-resolution spectroscopic study of two new Na- andAl-rich field giants - likely globular cluster escapees in theGalactic halo
Avrajit Bandyopadhyay, (cid:63) Sivarani Thirupathi, Timothy C Beers and A.Susmitha Indian Institute of Astrophysics, Bangalore, 560034 India Department of Physics and JINA Center for the Evolution of the Elements, University of Notre Dame, Notre Dame, IN, 46656, USA Tata Institute of Fundamental Research, Mumbai, 400005, India
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
The stars SDSS J0646+4116 and SDSS J1937+5024 are relatively bright stars thatwere initially observed as part of the SDSS/MARVELS pre-survey. They were selected,on the basis of their weak CH G -bands, along with a total of 60 others, in the rangeof halo globular cluster metallicities for high-resolution spectroscopic follow-up as apart of the HESP-GOMPA survey (Hanle Echelle SPectrograph – Galactic surveyOf Metal Poor stArs). The stars exhibit typical nucleosynthesis signatures expectedfrom the so-called second-generation stars of globular clusters. The light-element anti-correlation of Mg-Al is detected, along with elevated abundances of Na. Carbon isfound to be depleted, which is compatible with expectation. Lithium is also detectedin SDSS J0646+4116 and SDSS J1937+5024 ; the measured abundances are similar tothose of normal halo giant stars. These bright escapees provide a unique opportunityto study the nucleosynthesis events of globular clusters in great detail, and shed lighton their chemical-enrichment histories. Key words: stars: abundances stars: chemically peculiar stars: Population II Galaxy:halo (Galaxy:) globular clusters: general
Globular clusters (GCs) are expected to lose a large amountof stellar mass during their interactions with the tidal fieldof the Milky Way (e.g., Baumgardt & Makino 2003; Krui-jssen 2014; Martell 2018, and references therein). There areseveral processes by which stars can escape from GCs. Theprimary reason is their lower bounding energy, as they areof low mass due to the equipartition of energy, which inturn causes mass segregation. The various effects of masssegregation on the mass of escaping stars are investigated inBalbinot & Gieles (2018). Among many others, Baumgardt& Makino (2003) also discusses such effects on clusters em-bedded in tidal fields. Other processes, such as disc shock-ing and dynamical friction, can also contribute to the lossof stars from GCs. However, mass loss in Milky Way GCsdue to dynamical friction would be negligible, as the massloss in this scenario primarily depends on the Galactocen-tric distance of the cluster. Although disc/bulge shocks doproduce enhanced mass loss (e.g., Dehnen et al. 2004), mostof the mass loss is due to secular evolution. The importance (cid:63)
E-mail: [email protected] of each mass-loss process depends upon the properties ofthe cluster, as shown in the “vital diagram” of Gnedin &Ostriker (1997).Chemical tagging is one of the important tools for iden-tifying stars of GC origin among the halo field stars (Free-man & Bland-Hawthorn 2002). Although both the clusterand field populations exhibit similar elemental abundancesof α -, Fe-peak, and neutron-capture elements (Gratton et al.2004; Pritzl et al. 2005; Lind et al. 2015), many stars inGCs exhibit certain unique trends for their light elements(C, N, O, Na, Mg, Al), uncharacteristic of the vast major-ity of halo stars (e.g., Kraft 1979; Kraft et al. 1979; Nor-ris & Freeman 1979, and numerous references since). Thesetraits are thought to emerge as a result of self pollutionwithin the cluster, where the gas from the first generationof stars does not escape from the cluster, but instead pol-lutes the second-generation stars with the products of ad-vanced hydrogen burning (Kraft et al. 1997; Carretta et al.2009b,a). In hotter regions (T >
40 MK), the Ne-Na chainbegins converting Ne to Na, while simultaneously O isdepleted via the ON cycle (Gratton et al. 2012). At stillhigher temperatures (T >
70 MK), the Mg-Al cycle is ini-tiated, which steadily depletes Mg and then Mg to Al © a r X i v : . [ a s t r o - ph . S R ] M a r Denissenkov & Weiss 1996; Salaris & Weiss 2002). The siteswhere the C-N-O, Ne-Na, and Mg-Al cycles occur is the hot-bottom burning (HBB) regions (Bloecker & Schoenberner1991; Boothroyd & Sackmann 1992) of the outer convectiveenvelope of intermediate-to-high mass (3-8 M (cid:12) ) AGB (IH-AGB) stars. Sufficiently high temperatures are not attainedin the H-burning shells of low-mass RGB stars to sustainthese reactions, hence no such variations are expcted to beobserved in halo field stars (Gratton et al. 2000).Fast-rotating massive stars (FRMS) are also a proba-ble site for similar nucleosynthesis reactions to take place(Decressin & Charbonnel 2007). The winds from these earlygenerations of IH-AGB stars and FRMS alter the chemicalcomposition of the birth clouds of subsequent generationsof stars. Consequently, a large fraction of present-day GCstars are enhanced in Na and Al, along with depleted levelsof C, O, and Mg. Almost all the Galactic GCs have beenfound to host multiple generations of stars (Bragaglia et al.2017), as traced by the C-N-O anomaly (Kayser et al. 2008;Smolinski et al. 2011) and the anti-correlations of Na andO, and Mg and Al (e.g., Carretta et al. 2009a,b; Lee 2010;Martell & Grebel 2010; Martell 2011; Bragaglia et al. 2015;Carretta et al. 2017).The scenarios leading to the onset ofthe multiple stellar populations are highly debated; therecould be several scenarios, as comprehensively discussed inBastian & Lardo (2018).Martell & Grebel (2010) demonstrated that ∼ ∼ ∼ High-resolution ( R ∼ , < V < B − V > .
6) to select targetssuitable for the MARVELS RV survey. The survey fields aremostly low-latitude fields, which is suitable for exo-planetsearches, but not ideal for detecting metal-poor stars. Wehave used synthetic spectral fitting of the pre-survey datato identify new metal-poor candidates (in the domain ofGC metallicity; [Fe/H] > − .
5) with weak CH G -bands, awell-known feature of GC stars that can be studied fromlow-resolution data (e,g., SDSS). We have obtained high-resolution data for 60 metal-poor stars in the metallicityrange of GCs (the survey paper is in prep.). Two starsamong them were found to be likely GC escapees – showingall the expected chemical signatures in their high-resolutionspectra. The stars were observed at a spectral resolution of R ∼ ,
000 over the wavelength range 380nm to 1000 nm.Details of the observations, along with the signal-to-noiseratios and V magnitudes, for these two stars are listed inTable 1.Data reduction was carried out using the IRAF echellepackage, as well as the publicly available data reductionpipeline for HESP developed by Arun Surya. A cross-correlation analysis with a synthetic template spectrum wascarried out to obtain the radial velocity (RV) for each star,listed in Table 1.Photometric as well as Spectroscopic data have beenused to estimate the stellar atmospheric parameters for thesestars. T eff , log( g ), [Fe/H], and microturbulent velocity. Theabundances of individual elements present in each spectrumwere determined using standard procedures, as describedin Bandyopadhyay et al. (2018). Photometric temperatureswere obtained using the available data in the literature andthe standard T eff -color relations derived by Alonso et al.(1996) and Alonso et al. (1999). T eff estimates have also beenderived spectroscopically, demanding that there be no trendof Fe I line abundances with excitation potential, as well asby fitting the H α profiles. The wings of H α are also sen-sitive to temperature. Estimates of surface gravity, log( g ),for these stars were determined by the usual technique thatdemands equality of the iron abundances derived for theneutral (Fe I) lines and singly ionized (Fe II) lines. Paral-laxes from Gaia have also been employed to derive the log(g) for individual stars. The wings of the Mg I lines, whichare sensitive to variations of log( g ), have also been fitted toobtain the best-fit value. The plots for ionization balanceand using Fe I and Fe II lines are shown in the top pan-els of Figure 1; fits for the Mg triplet and H α are shownin the middle panels for the two program stars. The wings [ M / H ] M / H F F F F N o r m a li s ed f l u x N o r m a li s ed f l u x J0646J1937 J0646J1937J0646 J1937J 0646 J 1937
Figure 1.
The stellar parameters for the two program giants. Theupper panels display the ionization equilibrium plot, in which thebest fit with the minimum slope and standard deviation amongall the computed model stellar atmospheres for each star has beenshown. Black crosses denote the Fe I abundances, while red filledtriangles indicate the Fe II abundances. The middle panels showthe fits in the spectral region of the Mg triplet for different valuesof log(g) in steps of 0.75 dex; the best-fit value is marked in redfor each star. The lower panels show the fits in the H- α region fordifferent values of temperature in steps of 300K; the best-fit valueis marked in red for each star. The bottom panels show the fitsfor the wings of H β ; the best fit is marked in red, and is adoptedas the stellar parameter of the star. The green and blue lines showthe departure variation of temperature by 200K. of the H β line is sensitive to both variation in temperatureand log(g). Spectral fits of H β line are shown in the bottompanel of Figure 1, with the best fit marked in red. For deter-mination of the stellar parameters, 81 and 99 Fe I lines and12 and 8 Fe II lines could be measured for J0646+4116 andJ1937+5024 , respectively. The adopted values of the stellarparameters for both stars are shown in Table 1.We have employed one-dimensional LTE stellar atmo-spheric models (ATLAS9; Castelli & Kurucz 2004) and thespectral synthesis code TURBOSPECTRUM (Alvarez &Plez 1998) for determining the abundances of the individ-ual elements present in each spectrum. We have consideredthe equivalent widths of the absorption lines present in thespectra that are less than 120 m˚A, as they are on the linearpart of the curve of growth. Version 12 of the turbospec-trum code for spectrum synthesis and abundance estimateshas been used for the analysis. We have adopted the hy-perfine splitting provided by McWilliam (1998) and Solarisotopic ratios.We have used the method of equivalent-width analy-sis for the clean, strong, and unblended lines of the lightelements (such as C, N, and O), the α -elements, and theFe-peak elements. Spectral synthesis was carried out forthe weaker features of these elements, and all lines of theneutron-capture elements, taking into account the hyperfinetransitions where they are present. Solar abundances for theindividual elements are taken from Asplund et al. (2009). α -Elements Lithium was detectable for both J0646+4116 andJ1937+5024 . The strong Li doublet at λ A (Li) = 0.95 and 1.05,respectively, similar to other evolved giants observed in thefield.Carbon abundances were measured by performinga spectral synthesis for the CH G -band region around λ − − , and found to be +0.01 dex and +0.50 dex forJ0646+4116 and J1937+5024 , respectively.Nitrogen abundances were obtained by measuring theCN molecular band at λ G -band were used, with a wide range of N abun-dances, and the best fit of the spectral band head was takenas the value of the N abundance. However, being close tothe extreme blue end of the spectrum, the signal-to-noise inthis region is poor, and thus only an upper limit could bederived for one of our stars, J1937+5024 ; it is found to beslightly enhanced, with a value of [N/Fe] < +0.42. Oxygenabundances were measured from the weak lines at λ λ < +0.51,respectively.Among the α -elements, Mg and Ca could be measuredfor both the stars, while the Si lines were too weak to de-rive any meaningful abundances for either. Several lines forMg and Ca could be obtained throughout the spectra, ofwhich only the clean lines were used to derive the abun-dances. The very strong lines, such as the Mg triplet around λ α -element enhancement of +0.4 dex in halo stars.Calcium could be taken as the true representative of the α -elements, as other species, such as O, Si, and Mg are oftenaltered due to the recycling of the products of an earliergeneration of stars during subsequent star formation insideGCs (Kraft et al. 1997; Gratton et al. 2004; Carretta et al.2010; Gratton et al. 2012).Na and Al are the most important among the odd-Zelements to tag a star of GC origin, and both were detectedfor our stars. Aluminium abundances have been derived byspectral fitting of the strong resonance line at λ λ λ λ http://vplacco.pythonanywhere.com/ lose to +1.0 dex, as discussed by Baumueller & Gehren(1997), Andrievsky et al. (2008), and Nordlander & Lind(2017). The NLTE corrections for Na (Andrievsky et al.2007) have also been taken into account, and incorporatedin the final values. The abundances of Cr, Co, Mn, Ni, and Zn could be mea-sured by the usual equivalent-width analysis of the cleanlines. Non-LTE corrections for each species (where available)have been incorporated in the final abundances. Cr and Coalso suffer from large NLTE corrections, which could be areason for the over-abundance (e.g., Bergemann & Cescutti2010).The two odd-Z elements Mn and Cu show the usualdeficiency with respect to Fe in the metal-poor domain, asfound in previous studies of halo and GC stars. However,Ni is found to be rather high for J1937+5024 . Three linesof Ni could be detected at λ λ λ λ λ The neutron-capture elements Sr and Ba could be measuredfor both the program stars, and found to have normal abun-dances without notable irregularities. The resonance linesof Sr II at λ λ λ λ Both of our program stars exhibit enhancement in [Na/Fe]and [Al/Fe] compared to halo stars of similar metallic-ity. They also show an under-abundance of [C/Fe], alongwith depletion in [O/Fe], which is compatible with second-generation GC stars. [Mg/Fe] is also found to be depleted inJ0646+4116 and J1937+5024 . Since elements such as C,O, and Mg could be significantly altered during quiescentburning in proton-fusion reactions (Gratton et al. 2004),Ca should be adopted as the best representative of the α -elements for comparison with halo-star abundances. The de-gree of α -element enhancement based on the Ca abundancesis found to be slightly lower than is typical for halo stars, at[ α /Fe]= +0.22. N o r m a li z ed f l u x N o r m a li z ed f l u x J0646 - Al J0646 - BaJ1937 - Al J1937 - Ba
Figure 2.
The left panels show the fits for Al, while the right pan-els show the fits for Ba. The red lines denote the best-fit abun-dances, over-plotted with two synthetic spectra of abundances0.25 dex higher and lower in blue and green, respectively. Thenames of the program stars and corresponding synthesized ele-ments are provided in each panel. -2.5 -2.0 -1.5 -1.0[Fe/H]-1.0-0.50.00.51.01.5 [ N a / F e ] -2.5 -2.0 -1.5 -1.0[Fe/H]-1.0-0.50.00.51.0 [ A l / F e ] -2.5 -2.0 -1.5 -1.0[Fe/H]-2-1012 [ ( N a + A l ) / F e ] Figure 3.
A comparative study on the light-element abundancesfor GCs and halo stars. The blue upward triangles indicate meanGC abundances, whereas the open circles indicate the halo stars.The two program stars discussed here are marked with red filledstars. The three panels show the distribution of the key elementsNa and Al with metallicity. Our program stars consistently fall inthe domain of GC abundances in all the plots. The data for GCsare taken from Carretta et al. (2009a), and the abundances forhalo stars are taken from the SAGA database (Suda et al. 2008).
We have conducted a comparative study of the light-element abundances for our program stars with stars fromthe halo population and GC abundances, as shown in Fig-ure 3. The abundances for the GC population are based onthe UVES spectra of 19 Galactic GCs, as reported by Car-retta et al. (2009a) while the abundances for halo stars aretaken from the SAGA database (Suda et al. 2008). The topand middle panels display the enhancement in [Na/Fe] and[Al/Fe], with respect to [Fe/H], for GC stars and halo stars.To more clearly separate the GC population from the halostars, we have also plotted [(Na+Al)/Fe] vs. [Fe/H] in the [ N a / F e ] -0.5 0.0 0.5 1.0[Mg/Fe]-1.0-0.50.00.51.0 [ A l / F e ] -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0[O/H]-3.5-3.0-2.5-2.0-1.5-1.0-0.50.0 [ N a / H ] -3.0 -2.5 -2.0 -1.5 -1.0 -0.5[Mg/H]-4-3-2-10 [ A l / H ] Figure 4.
The top two panels show the Na-O and Mg-Al anti-correlations. The two panels at the bottom probe the origin oflight-element anti-correlations by removing the dependence onmetallicity. The red arrow indicates the obtained upper limit forO abundances in the case of J0646+4116 .The data for GCs aretaken from Carretta et al. (2009b), and the abundances for halostars are taken from the SAGA database (Suda et al. 2008). Thesymbols are the same as in Figure 3. -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0[O/Fe]-0.50.00.51.0 [ N a / F e ] M15M68NGC4372NGC 6397This study
Figure 5.
Comparison of the abundances of the two halo fieldgiants with the other metal poor globular cluster in the samerange of metallicity namely M15, M68, NGC 4372 AND NGC6397 along the Na-O anticorrelation. The abundances of the in-dividual stars in these GCs have been compiled as follows - M15were obtained from Carretta et al. (2009b) ; M68 were taken fromLee et al. (2004) and Carretta et al. (2009a); NGC 4372 were ob-tained from San Roman, I. et al. (2015); NGC 6397 were takenfrom Carretta et al. (2009b), Lind, K.et al. (2011) and Pasquini,L. et al. (2008). bottom panel. Both the program stars fall in the domain ofGC abundances in these plots.We note that the observed enhancement in [Na/Fe] and[Al/Fe] could also arise if these stars were members of thethick disc, but such stars are expected to show a much highermetallicity than our program stars, and are also inconsistentwith the C and O abundances. The space velocities have alsobeen determined for the target stars, and are found to beconsistent with space velocities for halo stars (Kinman et al. 2007). The values of u,v and w for J1937+5024 are -165.4,-172.9, and -92.3 while for J0646+4116 they are found to be330.3 -275.3 and -57.0. Figure 4 in Kinman et al. (2007)shows the distribution of stars, based on their space ve-locities, and classifies them into prograde and retrogradeorbits. J1937+5024 is consistent with the halo stars withprograde orbits, while J0646+4116 falls very close to theedge of the line separating prograde motion from retrogrademotion. Both of the program stars are on halo-like orbits.Thus, thick-disc membership appears improbable from theconsideration of both abundances and space velocities.To strengthen the argument, we have also tried to lookfor the Na-O and Mg-Al anti-correlations in Figure 4. Thedata for GCs are taken from Carretta et al. (2009a), andthe abundances for halo stars are taken from the SAGAdatabase (Suda et al. 2008). Though we could only obtain anupper limit for O in J0646+4116 the abundances fit betterwith the GC population. In the bottom panels of Figure 4,we attempt to probe the existence of these anti-correlationsby removing the trends with metallicity. In the [Al/H] vs.[Mg/H] plane, the GCs and the likely escapees still standout from the halo population, and exhibit a different trendin the distribution, whereas an offset could be seen with asimilar trend between the GCs and halo population in the[Na/H] vs. [O/H] plane.Figure 5 provides a comparison of the abundances ofour two program stars with individual GC stars of similarmetallicity from M15, M68, NGC 4372 and NGC 6397; theNa-O anticorrelation is clear. The abundances of the indi-vidual stars in these GCs have been compiled as follows -M15 were obtained from Carretta et al. (2009a) ; M68 weretaken from Lee et al. (2004) and Carretta et al. (2009a);NGC 4372 were obtained from San Roman, I. et al. (2015);NGC 6397 were taken from Carretta et al. (2009a), Lind, K.et al. (2011) and Pasquini, L. et al. (2008). The position ofthe target stars in the Na-O plane is found to be well withinthe scatter of the GC stars of similar metallicity.Lithium could also be detected in both the programstars, and is found to be normal. Li is a fragile element,which is completely destroyed in a temperature rangemuch lower than that required for operation of Mg-Alcycle. Thus, the presence of Li in second-generation starsindicates heavy dilution of the gas processed by p-capturereactions with unprocessed gas that still preserves thestandard Population II lithium abundance (D’Antona et al.2019).In the case of AGB stars, Li could also be produced byCameron & Fowler (1971) mechanism in the envelope of thestar at the early stages of the HBB phase (mass loss dur-ing the Li-rich phase of the AGB is discussed in detail inVentura, P. & D’Antona, F. (2005)). D’Antona et al. (2019)shows the classic dilution scheme for the abundance of Li insecond-generation GC stars using AGB ejecta and mixingwith primordial gas. As observed by D ' Orazi et al. (2014)and D’Orazi et al. (2015), the first generation (FG) andthe second generation (SG) stars in M12 and NGC 362 ex-hibit the same Li abundance, which indicates the presenceof a progenitor population like AGB stars that can produceLi. However, the simple dilution model fails to explain theinternal variation and complex Li abundances in some ofthe GCs, like NGC 1904 and NGC 2808 – the productionfficiency of Li also depends upon the cluster’s mass andmetallicity (D’Orazi et al. 2015). Li have been measured inseveral Galactic globular clusters; the Li abundances exhibita similar distribution as normal halo stars.
A sample of ∼
60 stars in the domain of GC metallicity withweak carbon molecular CH G -bands selected from the low-resolution SDSS/MARVEL pre-survey have been observedat high spectral resolution to identify signatures of second-generation stars in GCs. Two such stars were found to beconsistent with all of the expected light-element anomalies.The stars studied here are most likely to be GC escapees.Binary mass transfer from an intermediate AGB or directpollution from a massive star wind might be unlikely tohave caused the abundance anomaly due to the presenceof lithium in both these objects. Upcoming massive spectro-scopic surveys will identify more such objects. GAIA kine-matics and accurate ages from asteroseismology will throwlight on the origin and frequency of such objects. We thank the referee for useful comments that helped clar-ify the presentation and enhance the quality of the paper.We are particularly thankful to the referee for suggestingthe addition of Figure 5. We thank the staff of IAO, Hanleand CREST, Hosakote, that made these observations pos-sible. The facilities at IAO and CREST are operated bythe Indian Institute of Astrophysics, Bangalore. We alsothank Prof. Piercarlo Bonifacio for his valuable commentsand suggestions. T.C.B. acknowledges partial support fromgrant PHY 14-30152 (Physics Frontier Center/JINA-CEE),awarded by the U.S. National Science Foundation (NSF).T.C.B. also acknowledges partial support from the Lever-hulme Trust (UK), which hosted his visiting professorshipat the University of Hull during the completion of this study. able 1.
Observational Details for our Program StarsObject RA DEC Exposure Frames SNR V Radial Vel. T eff log(g) ξ [Fe / H](secs) (mag) (km s − (K) (cgs)SDSS J064655.6+411620.5 06 46 55.6 +41 16 20.5 2400 6 43 11.14 − − − − Table 2.
Elemental Abundances for SDSS J064655.6+411620.5Name Species Solar lines A(X) [X/H] [X/Fe] σ ∗ Li Li I . . . 1 0.95 . . . . . . 0.03C CH 8.43 . . . 6.25 − − − − − − − − − -1.47 +0.43 0.05Ti II 4.95 14 3.57 − − − − − − − − − − − Table 3.
Elemental Abundances for SDSS J193712.01+502455.5Name Species Solar lines A(X) [X/H] [X/Fe] σ ∗ Li Li I . . . 1 1.05 . . . . . . 0.02C CH 8.43 . . . 6.00 − − − − − − − − − − − − − − − − − − − − − − − EFERENCES
Alonso A., Arribas S., Martinez-Roger C., 1996, A&A, 313, 873Alonso A., Arribas S., Mart´ınez-Roger C., 1999, A&AS, 140, 261Alvarez R., Plez B., 1998, A&A, 330, 1109Andrievsky S. M., Spite M., Korotin S. A., Spite F., Bonifacio P.,Cayrel R., Hill V., Fran¸cois P., 2007, A&A, 464, 1081Andrievsky S. M., Spite M., Korotin S. A., Spite F., Bonifacio P.,Cayrel R., Hill V., Fran¸cois P., 2008, A&A, 481, 481Asplund M., Grevesse N., Sauval A. J., Scott P., 2009, ARA&A,47, 481Balbinot E., Gieles M., 2018, MNRAS, 474, 2479Bandyopadhyay A., Sivarani T., Susmitha A., Beers T. C., Girid-har S., Surya A., Masseron T., 2018, ApJ, 859, 114Bastian N., Lardo C., 2018, Annual Review of Astronomy andAstrophysics, 56, 83Baumgardt H., Makino J., 2003, MNRAS, 340, 227Baumueller D., Gehren T., 1997, A&A, 325, 1088Bergemann M., Cescutti G., 2010, A&A, 522, A9Bloecker T., Schoenberner D., 1991, A&A, 244, L43Boothroyd A. I., Sackmann I.-J., 1992, ApJ, 393, L21Bragaglia A., Carretta E., Sollima A., Donati P., D’Orazi V.,Gratton R. G., Lucatello S., Sneden C., 2015, A&A, 583, A69Bragaglia A., Carretta E., D’Orazi V., Sollima A., Donati P.,Gratton R. G., Lucatello S., 2017, A&A, 607, A44Cameron A. G. W., Fowler W. A., 1971, ApJ, 164, 111Carretta E., et al., 2009a, A&A, 505, 117Carretta E., Bragaglia A., Gratton R., Lucatello S., 2009b, A&A,505, 139Carretta E., Bragaglia A., Gratton R. G., Recio-Blanco A., Lu-catello S., D’Orazi V., Cassisi S., 2010, A&A, 516, A55Carretta E., Bragaglia A., Lucatello S., D’Orazi V., GrattonR. G., Donati P., Sollima A., Sneden C., 2017, A&A, 600,A118Castelli F., Kurucz R. L., 2004, ArXiv Astrophysics e-prints,D’Antona F., Ventura P., Marino A. F., Milone A. P., Tailo M.,Criscienzo M. D., Vesperini E., 2019, The Astrophysical Jour-nal, 871, L19D ' Orazi V., Angelou G. C., Gratton R. G., Lattanzio J. C., Bra-gaglia A., Carretta E., Lucatello S., Momany Y., 2014, TheAstrophysical Journal, 791, 39D’Orazi V., et al., 2015, Monthly Notices of the Royal Astronom-ical Society, 449, 4038Decressin T., Charbonnel C., 2007, in Kerschbaum F., Charbon-nel C., Wing R. F., eds, Astronomical Society of the PacificConference Series Vol. 378, Why Galaxies Care About AGBStars: Their Importance as Actors and Probes. p. 54Dehnen W., Odenkirchen M., Grebel E. K., Rix H.-W., 2004, TheAstronomical Journal, 127, 27532770Denissenkov P. A., Weiss A., 1996, A&A, 308, 773Freeman K., Bland-Hawthorn J., 2002, ARA&A, 40, 487Gnedin O. Y., Ostriker J. P., 1997, ApJ, 474, 223Gratton R. G., Sneden C., Carretta E., Bragaglia A., 2000, A&A,354, 169Gratton R., Sneden C., Carretta E., 2004, ARA&A, 42, 385Gratton R. G., Carretta E., Bragaglia A., 2012, A&A Rev., 20,50Kayser A., Hilker M., Grebel E. K., Willemsen P. G., 2008, A&A,486, 437Kinman T. D., Cacciari C., Bragaglia A., Buzzoni A., Spagna A.,2007, MNRAS, 375, 1381Koch A., Grebel E. K., Martell S. L., 2019, A&A, 625, A75Kraft R. P., 1979, ARA&A, 17, 309Kraft R. P., Trefzger C. F., Suntzeff N., 1979, in Burton W. B.,ed., IAU Symposium Vol. 84, The Large-Scale Characteristicsof the Galaxy. pp 463–473Kraft R. P., Sneden C., Smith G. H., Shetrone M. D., LangerG. E., Pilachowski C. A., 1997, AJ, 113, 279 Kruijssen J. M. D., 2014, Classical and Quantum Gravity, 31,244006Lee J.-W., 2010, MNRAS, 405, L36Lee J.-W., Carney B. W., Habgood M. J., 2004.Lind, K. Charbonnel, C. Decressin, T. Primas, F. Grundahl, F.Asplund, M. 2011, A&A, 527, A148Lind K., et al., 2015, A&A, 575, L12Majewski S. R., APOGEE Team APOGEE-2 Team 2016, As-tronomische Nachrichten, 337, 863Martell S. L., 2011, Astronomische Nachrichten, 332, 467Martell S. L., 2018, in Chiappini C., Minchev I., Starken-burg E., Valentini M., eds, IAU Symposium Vol. 334,Rediscovering Our Galaxy. pp 38–42 ( arXiv:1710.03858arXiv:1710.03858