Mass loss law for red giant stars in simple population globular clusters
M. Tailo, A. P. Milone, E. P. Lagioia, F. D'Antona, S. Jang, E. Vesperini, A. F. Marino, P. Ventura, V. Caloi, M. Carlos, G. Cordoni, E. Dondoglio, A. Mohandasan, J. E. Nastasio, M. V. Legnardi
MMNRAS , 1–10 (xxx xxx) Preprint 25 February 2021 Compiled using MNRAS L A TEX style file v3.0
Mass loss law for red giant stars in simple population globularclusters
M. Tailo ★ , A. P. Milone , , E. P. Lagioia , , F. D’Antona , S. Jang , E. Vesperini ,A. F. Marino , P. Ventura , V. Caloi , M. Carlos , G. Cordoni , E. Dondoglio ,A. Mohandasan , J. E. Nastasio , M. V. Legnardi Dipartimento di Fisica e Astronomia “Galileo Galilei”, Univ. di Padova, Vicolo dell’Osservatorio 3, Padova, IT-35122 Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5, Padova, IT-35122 Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Roma, Via Frascati 33, I-00040 Monteporzio Catone, Roma, Italy Department of Astronomy, Indiana University, Bloomington, IN 47405, USA Istituto Nazionale di Astrofisica - Osservatorio Astrofisico di Arcetri, Largo Enrico Fermi 5, Firenze, I - 50125 INAF – IASF Roma, Via Fosso del Cavaliere, Roma, Italy, IT-00133
Accepted 2021 February 22. Received 2021 February 22; in original form 2020 December 30
ABSTRACT
The amount of mass lost by stars during the red-giant branch (RGB) phase is one of the mainparameters to understand and correctly model the late stages of stellar evolution. Nevertheless,a fully-comprehensive knowledge of the RGB mass loss is still missing.Galactic Globular Clusters (GCs) are ideal targets to derive empirical formulations ofmass loss, but the presence of multiple populations with different chemical compositions hasbeen a major challenge to constrain stellar masses and RGB mass losses. Recent work hasdisentangled the distinct stellar populations along the RGB and the horizontal branch (HB) of46 GCs, thus providing the possibility to estimate the RGB mass loss of each stellar population.The mass losses inferred for the stellar populations with pristine chemical composition (calledfirst-generation or 1G stars) tightly correlate with cluster metallicity. This finding allows us toderive an empirical RGB mass-loss law for 1G stars.In this paper we investigate seven GCs with no evidence of multiple populations andderive the RGB mass loss by means of high-precision
Hubble-Space Telescope photometryand accurate synthetic photometry. We find a cluster-to-cluster variation in the mass lossranging from ∼ ∼ 𝑀 (cid:12) . The RGB mass loss of simple-population GCs correlates withthe metallicity of the host cluster. The discovery that simple-population GCs and 1G stars ofmultiple population GCs follow similar mass-loss vs. metallicity relations suggests that theresulting mass-loss law is a standard outcome of stellar evolution. Key words: stars: evolution, (stars:) Hertzsprung-Russell and colour-magnitude diagrams,stars: horizontal branch, stars: low-mass Stars, stars: mass-loss,(Galaxy:) globular clusters:general
A proper understanding of the late stages of stellar evolution dependson the precise knowledge of the law ruling mass loss along the redgiant branch (RGB). Hence, determining the RGB mass loss law isa crucial step to fully understand stellar evolution.To date, we still lack a conclusive theoretical description ofRGB mass loss, and we mostly rely on empirical determinations.Historically, the law by Reimers (1975) based on Population I starshas represented for decades the state of the art for describing RGB ★ E-mail:[email protected], [email protected] mass loss (see also Fusi-Pecci & Renzini 1978; Catelan 2000,and references therein). More recently, new formulations, basedon magneto-hydrodynamics, have been proposed (see Schröder &Cuntz 2005, 2007; Cranmer & Saar 2011) either as new law or asmodifications of the Reimers (1975) one. These new formulationstie mass loss to the interactions between surface turbulence and themagnetic field of the stars which are most relevant in the last part ofthe RGB where, indeed, most of the mass loss is predicted to takeplace.A constantly updated observational framework is therefore cru-cial in the calibration and construction of the theoretical framework.Origlia et al. (2007, 2014) estimated the mass loss of 47 Tucanae © xxx xxx The Authors a r X i v : . [ a s t r o - ph . S R ] F e b Tailo, M. et al. and other fourteen GCs based on the excess of mid-IR light andsuggested that a fraction of stars can lose mass at any RGB lumi-nosity. To do this, they exploited multi-band photometry from theSpitzer space telescope and from NIR ground-based facilities (butsee Boyer et al. 2010; McDonald et al. 2011a, for a different inter-pretation of the photometric results by Origlia and collaborators).Momany et al. (2012) on the other hand did not detect any NIRexcess among RGB stars of 47 Tucanae. Mass loss has been alsoestimated by using Spitzer Infrared Spectrograph spectra of RGBstars, a technique adopted by McDonald et al. (2011b) to investigatethe brightest RGB stars in 𝜔 Centauri.The comparison between the stellar mass of horizontal branch(HB) and RGB stars may provide an efficient approach to infer theRGB mass loss in a simple stellar population. Indeed, after reachingtheir tip luminosity, RGB stars undergo the so-called helium flash,namely an abrupt ignition of their degenerate helium core. Afterthis violent process, HB stars reach their position along the branchwith different effective temperatures. The total mass deficit of theresulting stars with respect to the RGB progenitors represents thesought-after mass loss. Based on this idea, Gratton et al. (2010)estimated the RGB mass loss of 98 Galactic GCs, but the presenceof multiple stellar populations with different chemical compositionprovides a significant challenge to their conclusions.Indeed, in addition to mass loss, the color and the magnitudeof a star along the HB depend on its age, metallicity, and heliumcontent. While the majority of GCs hosts stars with the same age andmetallicity, the majority of them are composed of two or more stellarpopulations with different helium content (e.g. D’Antona et al. 2002,2005; Milone et al. 2018; Lagioia et al. 2018). As a consequence,different parameters with degenerate effects determine the effectivetemperature of an HB star. In particular, increased helium and RGBmass loss would both increase the star temperature.Recent work has introduced an innovative approach to inferthe mass loss of the distinct stellar populations in GCs (Tailo et al.2019a,b). These papers are based on the theoretical and empiricalevidence that stellar populations with different helium content pop-ulate distinct HB regions (e.g. D’Antona et al. 2002; Marino et al.2011, 2014; Gratton et al. 2011; Dondoglio et al. 2020). Once stel-lar populations are identified along the HB and their helium contentis independently constrained from the MS and RGB (Milone et al.2018), it is possible to disentangle the effect of helium and massloss along the HB.Tailo et al. (2020, hereafter T20) has extended this method toa large sample of 46 GCs. They identified their stellar populationwith pristine helium abundance (hereafter first generation or 1G)along the RGB and the HB and inferred the RGB mass loss byusing appropriate theoretical models. Similarly, they estimated themass loss of stars with extreme helium content (hereafter extremesecond generation or 2Ge). Tailo and collaborators found that themass loss of 1G stars changes from one cluster to another and istightly correlated with the cluster metallicity. Based on these resultsthey defined an empirical mass-loss law for 1G stars.In this work, we analyze seven clusters with no evidence ofmultiple populations, namely NGC 6426, Palomar 12, Palomar 15,Pyxis, Ruprecht 106, Terzan 7 and Terzan 8. Hence, they are eithersimple stellar populations or host stars with very small internalhelium variations. We compare the RGB mass losses inferred fromthese clusters and from 1G and 2Ge stars of the multiple-populationclusters studied by T20. The main goal is to shed light on whetherthe mass loss law by T20 describes 1G stars of multiple-populationGC alone or is a universal property of stellar evolution.The paper is organized as follows: in § 2 we present the pho- tometric catalogues and the stellar evolution models. In § 3 wedescribe the method to infer mass loss and provide the results forall clusters, in § 5. Finally, we compare the results from this paperand from T20 and summarize the main findings of the paper in § 6.
The clusters studied in this paper are seven Galactic GCs older than ∼ ∼ × 𝑀 (cid:12) , which is considered the mass threshold toform multiple populations in GCs (Milone et al. 2020). In this workwe consider these clusters as SSP GC candidates. We included inthe sample Palomar 15 and NGC 6426, whose HBs have short colorextension but are more massive than ∼ × 𝑀 (cid:12) . Despite thereis no evidence that these two clusters have homogeneous chemi-cal composition, the small color extensions of their HBs suggestthat their 2G stars, if present, would not exhibit extreme chemicalcompositions.In addition, we analyzed the CMDs of the candidate simple-population GCs AM 4, E 3, Palomar 1, Palomar 13 (Monaco et al.2018; Milone et al. 2020). By using literature photometric cata-logues (Sarajedini et al. 2007; Anderson et al. 2008; Dotter et al.2011; Milone et al. 2016), we verified that no HB stars are presentin these last four clusters. Nevertheless, we used their photometry toderive other quantities that are relevant for our analysis, includingcluster age and the stellar mass at the tip of the RGB.In the following, we summarize the photometric data-set andthe stellar models that we employ to derive the RGB mass loss inthe seven clusters with HB stars. To infer the RGB mass loss we derived stellar photometry and propermotions by using two-epoch images collected through the F606Wand F814W filters of the Wide Field Channel of the AdvancedCamera for Surveys (WFC/ACS) on board
HST . The main propertiesof the images are provided in Table 1.Stellar magnitudes and positions have been derived for eachexposure separately by using the img2xym_WFC computer pro-gram from Anderson & King (2006). In a nutshell, we identifiedas a candidate star every point-like source whose central pixel hasmore than 50 counts within its 3 × MNRAS000
HST . The main propertiesof the images are provided in Table 1.Stellar magnitudes and positions have been derived for eachexposure separately by using the img2xym_WFC computer pro-gram from Anderson & King (2006). In a nutshell, we identifiedas a candidate star every point-like source whose central pixel hasmore than 50 counts within its 3 × MNRAS000 , 1–10 (xxx xxx) ass loss in simple-population globular clusters the Vega system by using the photometric zero points provided bythe Space Telescope Science Institute website .Stellar position have been corrected for geometrical distortionby using the solution by Anderson & King (2006) and transformedinto a common reference frame based on Gaia early data release 3(Gaia EDR3, Gaia Collaboration et al. 2020). Stellar coordinatesderived from images collected at different epochs are then averagedtogether and these average positions have been compared with eachother to derive the stellar proper motions relative to the averagecluster motion (see Anderson & King 2003; Piotto et al. 2012, fordetails).These relative proper motions have been transformed into anabsolute reference frame by adding to the relative proper motion ofeach star the average motion of cluster members. The absolute GCproper motions are listed in Table 2 and are derived from stars whereboth Gaia DR3 absolute proper motions and HST relative propermotions are available. Finally, photometry has been corrected fordifferential reddening by following the procedure by Milone et al.(2012) .The vector-point diagrams of proper motions and the 𝑚 F606W vs. 𝑚 F606W − 𝑚 F814W
CMDs corrected for differential reddening ofthe seven GCs in our sample are plotted in Figures 1 and 2, wherewe indicated cluster members, selected on the basis of their propermotions (e.g. Cordoni et al. 2020), with black points and field starswith gray crosses.
We exploited the stellar-evolution models and the isochrones used byT20, which have been computed with the stellar-evolution programATON 2.0 (Ventura et al. 1998; Mazzitelli et al. 1999). The grid ofmodels used in this paper includes different ages, metallicity (Z), andhelium mass fractions (Y). Specifically, the iron abundance rangesfrom [Fe/H]= − − 𝛼 /Fe] are0.0, + + 𝛼 /Fe]= +0.0 havebeen calculated specifically for this work. The HB models includea small correction to their helium mass fraction to account for thefirst dredge up effects. The HB evolution is followed until the endof the helium burning phase. Gravitational settling of helium andmetals is not included.To derive the RGB mass loss of each cluster we compare theCMD of the observed HB stars with a grid of synthetic CMDs,obtained following the recipes of D’Antona et al. (2005, and refer-ences therein). Briefly, the mass of the each HB star (M HB ) in eachsimulation is obtained as: M HB = M Tip ( Z , Y , A ) − Δ M ( 𝜇, 𝛿 ) . HereM Tip is the stellar mass at the RGB tip, which depends on age (A),metallicity (Z) and helium content (Y); Δ M is the mass lost by thestar and described by a Gaussian profile with central value 𝜇 andstandard deviation 𝛿 .Once the value of M HB is obtained the star is placed on its HBtrack via a series of random extraction and interpolation procedures.Each simulation in the grid is composed of few thousands star toavoid problems due to high variance. The values of M Tip are ob-tained from the isochrones that provide the best fit with the observedCMD. We refer to D’Antona et al. (2005); Tailo et al. (2016) andT20 for additional details on the procedure. The photometric and astrometric catalogs will be available at thehttp://progetti.dfa.unipd.it/GALFOR web page and at the CDS (cdsarc.u-strasbg.fr).
In this section we summarize the procedure to derive the RGB massloss in simple-population clusters, using Palomar 15 as a template.After extending the same analysis to the other analyzed GCs, wewill discuss the results.
To infer the RGB mass loss experienced by HB stars we followthe procedure introduced by T20 (see also Tailo et al. 2019a,b) andillustrated in Figure 3 for Palomar 15. This procedure is based on thecomparison between the observed HB stars and a grids of simulatedHBs.At odds with T20, who studied clusters with multiple popu-lations, our sample consists in candidate simple-population GCs.Hence, we assumed that all their stars have the same chemical com-position and adopted pristine helium mass fraction for all clusters.The first step is to evaluate the age of the population, needed asinput to generate the synthetic HB grids. We do that via the isochronefitting of the turn off region. We produce an array of isochrones with[Fe/H]= − 𝛼 /Fe]=+0.4 (following the indication by Kirby et al.2008) and Y=0.25, with age ranging from 8.0 to 14.0 Gyr in stepsof 0.25 Gyr. We adopt E(B − V)=0.40 and (m − M) V =19.51 from the2010 version of the Harris (1996) catalogue .The isochrone that provides the best match with the turn-offregion in the m F814W vs m
F606W − m F814W
CMD (orange isochronein Figure 3a) gives us the best estimate for cluster age, in this case13 . ± .
75 Gyr. The uncertainty corresponds to the age range thatallows the isochrones to envelope 68.27% of stars in the turn offregion.We verified that our age estimate is not significantly affectedby unresolved binaries and blue stragglers (BSS). To investigate thepossible effect of binaries and BSSs on the inferred cluster ages, wesimulated two mock CMDs of Palomar 15 with the same age andmetallicity values adopted here, but different fraction of binariesand BSSs. In the first CMD we assumed no binaries and BSSs,while in the second one the same fraction of binaries derived byMilone et al. (2016) and the same number of BSSs as observed inthe actual CMD. We derive the GC age of both simulated CMDsby using the same methods described above and we obtain that theage values are consistent within 0.25 Gyr. Hence, we conclude thatbinaries and BSSs do not affecr our age determination.Our second step is to identify, by eye, the HB stars in thecluster. The selected HB stars of Palomar 15 are enclosed in theblue rectangle of Figure 3a. We take extra care to verify that thestars are identified on the HB in both m
F814W vs m
F606W − m F814W and m
F606W vs m
F606W − m F814W
CMDs.The selected HB stars are compared with appropriate grids ofsimulated HBs corresponding to different values of mass loss ( 𝜇 )and mass-loss dispersion ( 𝛿 ; see T20 for details). In the case ofPalomar 15 𝜇 varies from 0.010 M (cid:12) to 0.180 M (cid:12) in steps of 0.003M (cid:12) , and 𝛿 ranges from 0.002 M (cid:12) to 0.009 M (cid:12) in steps of 0.001M (cid:12) . For each simulation in the grid we compare the normalizedhistogram of the color distribution of observed and simulated stars.To quantify the goodness of the fit, we calculate the 𝜒 -squareddistance between the two histograms, 𝜒 (see Dodge 2008, and ∼ harris/mwgc.datMNRAS , 1–10 (xxx xxx) Tailo, M. et al.
Figure 1. 𝑚 F606W vs. 𝑚 F606W − 𝑚 F814W
CMDs (upper panels) and vector-point diagrams of proper motions (lower panels) of NGC 6426, Palomar 12,Palomar 15 and Pyxis. Candidate cluster members and field stars are colored black and gray, respectively.
Table 1.
Description of the
HST images used in the paper.ID FILTER DATE N × EXPTIME PROGRAM PINGC 6426 F606W Aug 04 2009 45s + × + × + × + × + × + × + × + × + + × + + × + + + × + × + × + × + × + × + × + × + × + × + × + + + × + + × + × + × + × + × + × + × + × + × + × + ×000
HST images used in the paper.ID FILTER DATE N × EXPTIME PROGRAM PINGC 6426 F606W Aug 04 2009 45s + × + × + × + × + × + × + × + × + + × + + × + + + × + × + × + × + × + × + × + × + × + × + × + + + × + + × + × + × + × + × + × + × + × + × + × + ×000 , 1–10 (xxx xxx) ass loss in simple-population globular clusters Figure 2.
As in Figure 1 but for Ruprecht 106, Terzan 7 and Terzan 8.
F606W m F814W m F W (a) Palomar 15Age= 13.25±0.75 Gyr
F606W m F814W m F W (d) = 0.126 M/MHB stars0.4 0.6 0.8 m F606W m F814W (c) (b) d Figure 3.
Summary of the procedure we follow to obtain the RGB mass loss for Palomar 15. In the panel (a) we compare the observed CMD of cluster memberwith the best-fit isochrone (yellow line) and the isochrones with ages of ± 𝜎 Gyr from the best age estimate (green line). The blue rectangle highlights HBstars. Panel (b) shows the 𝜒 density plot in the mass-loss dispersion vs. mass loss plane. The best fit values are marked by the orange square. The histogramsof the color distribution of observed (black) and simulated HB stars are shown in panel (c), while panel (d) shows the observed HB and contours correspondingto the best fit simulated HB. See text for details.MNRAS , 1–10 (xxx xxx) Tailo, M. et al.
T20). The resulting density map of 𝜒 values in the 𝜇 vs. 𝛿 planeis plotted in Figure 3b. The best fit simulation is then the one thatminimizes the 𝜒 and is indicated with the orange square on themap of Figure 3b.We evaluate the uncertainty on our estimates of mass lossand mass-loss dispersion by means of bootstrapping. We gener-ated 5,000 realizations of the HB in Palomar 15 and performed thecomparison with the grid of simulated HBs on each iteration. Toestimate the uncertainties on mass loss and mass loss-dispersion wefirst considered the standard deviation of the results.Moreover, we added the contribution to the error from theuncertainties on cluster age, metallicity, and reddening. To do this,we derived mass loss by using the same procedure above but bychanging cluster age by 0.75 Gyr, iron abundance by 0.10 dex andreddening by E(B − V)=0.015 mag. In clusters without spectroscopicdetermination of 𝛼 -element abundance, we also accounted for theeffect of a variation in [ 𝛼 /Fe] by 0.2 dex. By adding in quadratureall the contributions to the total error, we obtain the final estimatefor the mass loss of Palomar 15: 𝜇 = . ± . (cid:12) .The best-fit isochrone provides the mass at the RGB tip, M Tip = .
783 M (cid:12) and for the HB stars (M HB = . ± .
030 M (cid:12) ). Thecomplete list of parameters inferred from the procedure illustratedfor Palomar 15 are listed in in Table 2 for all studied clusters.For completeness, we compare the histogram of the color dis-tribution of the best fit simulation, i.e. the one whose values of 𝜇 and 𝛿 minimize 𝜒 , with the corresponding histogram from ob-served HB stars (Figure 3c). Finally, in the panel d of Figure 3 wesuperimposed on the observed CMD the contours of the best-fitsimulation. In the panel 3d, the contour lines delimit the regions ofthe simulated CMD including (starting from the outermost region)98,95,80,60% of stars. The procedure from T20, summarized in the previous section, hasbeen extended to the entire sample of eleven Galactic GCs that arecandidate to host a simple stellar population.The analyzed GCs are listed in Table 2, together with the valuesof [Fe/H], [ 𝛼 /Fe], E(B − V) and (m − M) V used for their analysis.Specifically, the values of reddening and distance modulus are fromthe 2010 version of the Harris (1996) catalog, while the values of[Fe/H] and [ 𝛼 /Fe] are taken from various literature sources andare derived from spectroscopy (see Table 2 for the complete list ofreferences).Since no spectroscopic determination for 𝛼 elements are avail-able for AM 4, Palomar 15 and Pyxis, we fixed the [ 𝛼 /Fe] valuesbased on their metallicity as suggested by Kirby et al. (2008). Hence,we assumed [ 𝛼 /Fe]=0.2 for AM 4, [ 𝛼 /Fe]=0.4 for Palomar 15 and[ 𝛼 /Fe]=0.2 for Pyxis. We also include in the error budget for theseclusters the effects of a possible shift of 0.2 in [ 𝛼 /Fe] that stemsfrom the results of Kirby et al. (2008).The complete showcase of examined HBs with their best fitsimulations is plotted in Figure 4, while the values of 𝜇 , 𝛿 and M HB derived from the analysis described in Section 3.1 are listed in Table2. Our results show that mass loss changes from cluster to clusterand ranges from 𝜇 ∼ . (cid:12) in NGC 6426 to 𝜇 ∼ . (cid:12) inPalomar 1. We plot in Figure 5, as red squares, the values of massloss for the studied GCs as a function of [Fe/H]. Clearly, mass losscorrelates with metallicity as demonstrated by the high values of the Spearman’s rank and the Pearson correlation coefficient (R s = . p = .
95, respectively). As the identification of NGC 6426and Palomar 15 as simple-population cluster is not certain, we willrepresent them with a different symbol in this and later figures. Thepoints are fitted with a least-squares straight line (red continuousline in Figure 5): 𝜇 = ( . ± . ) × [ Fe / H ] + ( . ± . ) M (cid:12) (1)The mean dispersion around the best fit line is 0.019 M (cid:12) . We reporta summary of the parameters in the best-fit relation in Table 3. In their recent paper, T20 constrained the RGB mass loss of thedistinct stellar populations of 46 galactic GCs with multiple pop-ulations. In particular, they investigated the mass loss of 1G starsand find a linear relation between the RGB mass loss and the ironabundance of the host GC. The comparison between the findings ofour paper and the results from Tailo and collaborators for 1G stars(Figure 5) reveals that simple-population clusters follow a simi-lar distribution in the mass-loss metallicity plane as 1G stars ofmultiple-population GCs. In particular, the best-fit line of simple-population GCs described by Equation 1 is almost coincident withthe corresponding relation discovered by T20 for the entire sam-ple of 1G stars, 𝜇 = ( . ± . ) × [ Fe / H ] + ( . ± . ) 𝑀 (cid:12) , black dashed-line of Figure 6. In contrast, the RGB mass lossof simple-populations clusters do not match that of 2G stars withextreme helium contents in GCs with similar metallicities.This evidence suggests that the relation between mass-loss andmetallicity by T20 does not depend on the presence of multiple pop-ulations in GCs but may be a standard stellar evolutionary property.This conclusion is corroborated by the results by Salaris et al. (2013)and Savino et al. (2019) who found a similar mass-loss law in dwarfgalaxies. By combining the results by T20 for 1G stars in 46 GCs andthose in this paper for simple-population clusters we derive the im-proved relation: 𝜇 = ( . ± . ) × [ Fe / H ] + ( . ± . ) M (cid:12) ,where the dispersion is ∼ (cid:12) and the correlation coefficientsare R s ∼ R p ∼ .
89. As a matter of fact this is equal to the gen-eral relation in T20 (black-dashed line in Figure 5). We report thisgeneral relation in Table 3.Recent works, based on asteroseismology, provide mass lossestimates in star clusters. Miglio et al. (2016) used Kepler data ofseven RGB and one red-HB stars of the GC M 4 to derive stellarmasses. The resulting mass loss, estimated as the mass differencebetween HB and RGB stars, ranges from ∼ ∼ (cid:12) , de-pending on the adopted scaling relation. The latter value alone isconsistent with the mass-loss inferred by T20 for the 1G stars ofM 4. On the other hand, similar studies on the old, metal rich opencluster NGC 6791 suggest a moderate RGB mass loss for this Galac-tic open cluster ( 𝜇 = . ± .
03 (random) ± .
04 (systematic) ,Miglio et al. 2012, and references therein), whereas RGB and redclump stars of the ∼ ≥ +0.3 and [Fe/H] ∼ MNRAS000
04 (systematic) ,Miglio et al. 2012, and references therein), whereas RGB and redclump stars of the ∼ ≥ +0.3 and [Fe/H] ∼ MNRAS000 , 1–10 (xxx xxx) ass loss in simple-population globular clusters F606W m F814W m F W NGC6426HB data= 0.096 M
F606W m F814W m F W Pal 12HB data= 0.226 M
F606W m F814W m F W Pal 15HB data= 0.126 M
F606W m F814W m F W PYXISHB data= 0.186 M
F606W m F814W m F W RUP106HB data= 0.113 M
F606W m F814W m F W Terzan 7HB data= 0.276 M
F606W m F814W m F W Terzan 8HB data= 0.110 M
Figure 4.
The analysed GCs in alphabetical order. We report in each panel the m
F814W vs m
F606W − m F814W
CMD of the HB stars together with the contourplot of the best fit simulation. The average mass losses of the best fit simulations are quoted in the insets.
Table 2.
Parameters of the GCs analysed in this work. The values of iron abundances, reddening and distance modulus are taken from the 2010 version of theHarris (1996) catalog. The average GC proper motions are derived in this paper based on Gaia EDR3 motions. Cluster age, mass at the RGB tip, RGB massloss, mass-loss spread and average HB mass are derived in this paper. Sources for [ 𝛼 /Fe]: (a) Sbordone et al. (2005),(b) Dias et al. (2015),(c) Monaco et al.(2018),(d) Cohen (2004),(e) Brown et al. (1997),(f) Dotter et al. (2018, and references therein),(g)Pritzl et al. (2005),(h) Jahandar et al. (2017), (i)Koch & Côté(2019), (l) from the indications in Kirby et al. (2008). ID [Fe/H] [ 𝛼 /Fe] E(B-V)(mag) (m-M) V (mag) Age (Gyr) M Tip / M (cid:12) 𝜇 / M (cid:12) 𝛿 / 𝑀 (cid:12) ¯M HB / M (cid:12) 𝜇 𝛼 𝑐𝑜𝑠𝛿 (mas yr − ) 𝜇 𝛿 (mas yr − )NGC6426 − 𝑏 . ± .
00 0.782 0 . ± .
029 0 . ± .
002 0 . ± . − ± − ± − 𝑑,𝑒 . ± .
50 0.909 0 . ± .
035 0 . ± .
001 0 . ± . − ± − ± − 𝑙 . ± .
75 0.783 0 . ± .
030 0 . ± .
002 0 . ± . − ± − ± − 𝑙 . ± .
75 0.860 0 . ± .
045 0 . ± .
001 0 . ± .
045 1.04 ± ± − 𝑒, 𝑓 . ± .
50 0.831 0 . ± .
035 0 . ± .
002 0 . ± . − ± ± − 𝑎 . ± .
50 0.954 0 . ± .
045 0 . ± .
002 0 . ± . − ± − ± − 𝑏 . ± .
50 0.782 0 . ± .
022 0 . ± .
002 0 . ± . − ± − ± − 𝑙 . ± .
75 0.829 — — — — —E03 − 𝑐 . ± .
00 0.899 — — — — —PALOMAR1 − ℎ . ± .
25 0.926 — — — — —PALOMAR13 − 𝑖 . ± .
50 0.836 — — — — —
MNRAS , 1–10 (xxx xxx)
Tailo, M. et al. -2.50 -2.00 -1.50 -1.00 -0.50[Fe/H]0.050.100.150.200.250.300.35
Simple GCsNGC 6426Palomar 15M3-like (T20)M13-like (T20)
Figure 5.
Mass loss ( 𝜇 ) of the HB stars in the simple population GCs asfunction of their [Fe/H] values. The red line is the best-fit straight line. Inthe background we plot as orange and blue points the mass loss of the 1Gstars for the M3- and M13-like GCs from T20, respectively, and their bestfit lines. The black dashed lines is the general relation from T20. loss and metallicity inferred for GCs can not be extrapolated toPopulation I stars, being valid up to [Fe/H] ∼ -0.5. As an alternative,uncertainties in stellar evolution models and/or in asteroseismologyscale relations can contribute to the discrepancy between the resultsbased on Kepler data and those of this paper. T20 identified two groups of GCs with different HB morphology: agroup of GCs, that, similarly to M 3, exhibit the red HB (M 3-likeGCs) and a the group of M 13-like GCs with the blue-HB alone (seealso Milone et al. 2014). M 3-like and M 13-like GCs are representedwith orange and azure colors, respectively, in Figure 5. The twogroups of M 3-like and M 13-like GCs define distinct trends in the 𝜇 vs. [Fe/H] plane, with M 13-like clusters having higher values of 𝜇 than M 3-like clusters with similar iron content.Clearly, the best-fit line of simple-population clusters is inagreement within one- 𝜎 with the corresponding relation of M 3-like GCs ( 𝜇 = ( . ± . ) × [ Fe / H ] + ( . ± . ) , orangeline), but exhibits a different slope than the best-fit line defined byM 13-like GCs.To further investigate HB stars in simple-population GCs weshow in the left panel of Figure 6 the stellar mass at the RGB inferredfrom the best-fit isochrone against metallicity. In this figure, we alsoincluded the studied clusters with no HB stars. Clearly, candidate-simple population GCs exhibit higher values of 𝑀 tip than GCs ofsimilar metallicity.The large RGB-tip stellar masses are mostly due to the fact thatthe majority of the simple-population clusters are younger than thebulk of GCs studied by T20. This is illustrated in the right panel ofFigure 6 where we show the age-metallicity relation for the clustersstudied in this paper (squares) and by T20 (circles). Table 3.
Linear fits in the form 𝛼 × [ Fe / H ] + 𝛽 derived in this paper forcandidate simple-population GCs and by combining the results of T20 onmass-loss of 1G stars in 46 GCs and those of this paper. We also providethe Pearson rank coefficient, R P , and the r.m.s of the residuals with respectto the best-fit line.Var. 𝛼 𝛽 R P scatterSimple-population GCs 𝜇 . ± .
013 0 . ± .
026 0.93 0 . HB . ± .
013 0 . ± .
022 -0.89 0 . 𝜇 . ± .
006 0 . ± .
011 0.89 0 . HB − . ± .
007 0 . ± .
013 -0.33 0 . We confirm previous result by Dotter et al. (2010, 2011, andreferences therein) and Leaman et al. (2013, and references therein)of two main branches of clusters in the age vs. [Fe/H] diagram, withsimple population GCs populating the younger branch. From thecomparison between observed GC ages and simulations of Milky-Way formation, Dotter and collaborators suggested that the distinctbranches of clusters in the age-metallicity plane may originate fromtwo different phases of Galaxy formation, including a rapid col-lapse followed by a prolonged accretion (see also Kruijssen et al.2019). Similarly, based on the integrals of motions by Massari et al.(2019), Milone et al. (2020) suggested that simple-population GCsmay form in dwarf galaxies that have been later accreted by theMilky Way. The possibility in-situ and accreted clusters show simi-lar mass loss-metallicity relations further corroborates the evidenceof a universal mass-loss law.Candidate simple-population GCs share similar HB masses asshown in Figure 7, where we plot M HB against [Fe/H]. The HB-mass range of candidate simple-population GCs is comparable withthat of 1G stars in M 3-like GCs and significantly differs from thebehaviour of M 13-like GCs.The evidence that M 13-like GCs exhibit different patterns thansimple-population star clusters in both the 𝜇 vs [Fe/H] and the M HB vs. [Fe/H] planes indicates that their 1G stars behave differentlythan the bulk of stars with similar metallicity. As a consequence, inaddition to metallicity, some second parameter is responsible for thedifferent mass loss required in their 1G stars. Although M 13-likeGCs are, on average, older than M 3-like GCs, age difference aloneis not able to account for the different HB shapes. Results of thispaper, and from T20, indicate that either mass loss is the secondparameter of the HB morphology or 1G stars of all GCs share thesame mass loss, but the reddest stars of M 13-like GCs are enhancedby 0.01-0.03 in helium mass fraction with respect to M 3-like ones.In this case, as suggested by D’Antona & Caloi (2008), M 13-likeGCs could have lost all 1G stars and their red HB tails are populatedby 2G stars with moderate helium enhancement. We derived high-precision ACS/WFC photometry in the F606Wand F814W of seven GCs that are candidates simple stellar popula-tions. We identified probable cluster members by means of stellarproper motions and corrected the photometry for the effects of dif-ferential reddening. The resulting CMDs have been used to inferthe RGB mass loss by comparing the observed HB stars with ap-propriate simulated CMDs. The main results can be summarized asfollows.(i) The RGB mass loss in candidate simple-population GCs
MNRAS000
MNRAS000 , 1–10 (xxx xxx) ass loss in simple-population globular clusters -2.50 -2.00 -1.50 -1.00 -0.50[Fe/H]0.780.800.820.850.880.900.930.95 M T i p Simple GCsNGC 6426Palomar 15No HB starsM3-like (T20)M13-like (T20) -2.50 -2.00 -1.50 -1.00 -0.50[Fe/H]8.09.010.011.012.013.014.0 A g e ( G y r ) Simple GCsNGC 6426Palomar 15No HB starsM3-like (T20)M13-like (T20)
Figure 6.
Stellar mass at the tip of the RGB (M
Tip , left panel) and cluster age (right panel) against iron abundance. Clusters with no HB stars are indicated withopen squares, while the other symbols are the same as in Figure 5. The black dashed line plotted in the right panel is the least-square fit for all T20 clusters,whereas the orange and azure dashed horizontal lines in the right panel indicate the average ages of M 3 and M 13-like GCs, respectively. The corresponding1 − 𝜎 age intervals are indicated by the shaded areas. -2.50 -2.00 -1.50 -1.00 -0.50[Fe/H]0.550.600.650.700.750.80 M H B Simple GCsNGC 6426Palomar 15M3-like (T20)M13-like (T20)
Figure 7.
Average HB mass ( ¯M HB ) as a function of [Fe/H] values for clustersstudied in this paper and by T20. The symbols and the colour coding are thesame as in Figure 5. Red, orange and azure continuous lines are the straightlines that provide the best fit with simple-population candidate, M 3 like andM 13-like GCs. The black-dashed line refers to all clusters studied by T20. varies from cluster to cluster and strongly correlates with thecluster metallicity.(ii) The mass-loss vs. [Fe/H] relation is consistent with a similarrelation inferred by T20 for 1G stars of 46 GCs. We combined the results from this paper and those derived from 1G starsby T20 to derive an improved mass-loss metallicity relation.Moreover, our relation matches the values of mass loss inferredby Savino et al. (2019) for dwarf galaxies but is not consistentwith the mass-losses values inferred for 2G stars by T20.(iii) For a fixed metallicity, the mass losses and the average HBmasses of 1G stars in a subsample of GCs with the blue HBalone (M 13-like GCs) significantly differ from those inferredfrom simple-population GCs and from 1G stars of the remain-ing GCs (M 3-like GCs).These results suggest that the tight correlation between theamount of RGB mass loss and [Fe/H] that we observed both insimple-population GCs and in 1G stars of multiple-population GCsdoes not depend on the multiple-population phenomenon and is agood candidate as a general property of Populations II stars. More-over, the finding that M 13-like GCs exhibit different mass-lossvs. metallicity relation than simple-population clusters suggests thatmass loss is one of the main second parameters that govern the HBmorphology of GCs. DATA AVAILABILITY
The images analysed in this work are publicly availableone the Space Telescope Science Institute
HST
MNRAS , 1–10 (xxx xxx) Tailo, M. et al.
ACKNOWLEDGEMENTS
This work has received funding from the European Research Coun-cil (ERC) under the European Union’s Horizon 2020 research inno-vation programme (Grant Agreement ERC-StG 2016, No 716082’GALFOR’, PI: Milone, http://progetti.dfa.unipd.it/GALFOR). MT,APM and ED acknowledge support from MIUR through theFARE project R164RM93XW SEMPLICE (PI: Milone). MT, APMand ED have been supported by MIUR under PRIN program2017Z2HSMF (PI: Bedin). EV acknowledges support from NSFgrant AST-2009193.
REFERENCES
Anderson J., King I. R., 2003, AJ, 126, 772Anderson J., King I. R., 2006, PSFs, Photometry, and Astronomy for theACS/WFC, Instrument Science Report ACS 2006-01Anderson J., et al., 2008, The Astronomical Journal, 135, 2055Baumgardt H., Hilker M., 2018, MNRAS, 478, 1520Baumgardt H., Hilker M., Sollima A., Bellini A., 2019, MNRAS, 482, 5138Boyer M. L., et al., 2010, ApJ, 711, L99Brown J. A., Wallerstein G., Zucker D., 1997, AJ, 114, 180Carretta E., Bragaglia A., Gratton R. G., D’Orazi V., Lucatello S., SollimaA., 2014, A&A, 561, A87Catelan M., 2000, ApJ, 531, 826Cohen J. G., 2004, AJ, 127, 1545Cordoni G., Milone A. P., Mastrobuono-Battisti A., Marino A. F., LagioiaE. P., Tailo M., Baumgardt H., Hilker M., 2020, ApJ, 889, 18Cranmer S. R., Saar S. H., 2011, ApJ, 741, 54D’Antona F., Caloi V., 2008, MNRAS, 390, 693D’Antona F., Caloi V., Montalbán J., Ventura P., Gratton R., 2002, A&A,395, 69D’Antona F., Bellazzini M., Caloi V., Pecci F. F., Galleti S., Rood R. T.,2005, ApJ, 631, 868Dias B., et al., 2015, A&A, 573, A13Dodge Y., 2008, The Concise Encyclopedia of Statistics. Springer New York,pp 68–70, doi:10.1007/978-0-387-32833-1Dondoglio E., Milone A. P., Lagioia E. P., Marino A. F., Tailo M., CordoniG., Jang S., Carlos M. G., 2020, arXiv e-prints, p. arXiv:2011.03283Dotter A., et al., 2010, ApJ, 708, 698Dotter A., Sarajedini A., Anderson J., 2011, ApJ, 738, 74Dotter A., Milone A. P., Conroy C., Marino A. F., Sarajedini A., 2018, ApJ,865, L10Fusi-Pecci F., Renzini A., 1978, in Philip A. G. D., Hayes D. S., eds, IAUSymposium Vol. 80, The HR Diagram - The 100th Anniversary ofHenry Norris Russell. p. 225Gaia Collaboration Brown A. G. A., Vallenari A., Prusti T., de Brui-jne J. H. J., Babusiaux C., Biermann M., 2020, arXiv e-prints, p.arXiv:2012.01533Gratton R. G., Carretta E., Bragaglia A., Lucatello S., D’Orazi V., 2010,A&A, 517, A81Gratton R. G., Lucatello S., Carretta E., Bragaglia A., D’Orazi V., MomanyY. A., 2011, A&A, 534, A123Handberg R., Brogaard K., Miglio A., Bossini D., Elsworth Y., SlumstrupD., Davies G. R., Chaplin W. J., 2017, MNRAS, 472, 979Harris W. E., 1996, AJ, 112, 1487Jahandar F., et al., 2017, MNRAS, 470, 4782Kirby E. N., Guhathakurta P., Sneden C., 2008, ApJ, 682, 1217Koch A., Côté P., 2019, A&A, 632, A55Kruijssen J. M. D., Pfeffer J. L., Reina-Campos M., Crain R. A., Bastian N.,2019, MNRAS, 486, 3180Lagioia E. P., et al., 2018, MNRAS, 475, 4088Lagioia E. P., Milone A. P., Marino A. F., Cordoni G., Tailo M., 2019, AJ,158, 202Leaman R., VandenBerg D. A., Mendel J. T., 2013, MNRAS, 436, 122 Marino A. F., Villanova S., Milone A. P., Piotto G., Lind K., Geisler D.,Stetson P. B., 2011, ApJ, 730, L16Marino A. F., et al., 2014, MNRAS, 437, 1609Massari D., Koppelman H. H., Helmi A., 2019, A&A, 630, L4Mazzitelli I., D’Antona F., Ventura P., 1999, A&A, 348, 846McDonald I., et al., 2011a, ApJS, 193, 23McDonald I., Johnson C. I., Zijlstra A. A., 2011b, MNRAS, 416, L6Miglio A., et al., 2012, MNRAS, 419, 2077Miglio A., et al., 2016, MNRAS, 461, 760Milone A. P., et al., 2012, A&A, 540, A16Milone A. P., et al., 2014, ApJ, 785, 21Milone A. P., et al., 2016, MNRAS, 455, 3009Milone A. P., et al., 2018, MNRAS, 481, 5098Milone A. P., et al., 2020, MNRAS, 491, 515Momany Y., Saviane I., Smette A., Bayo A., Girardi L., Marconi G., MiloneA. P., Bressan A., 2012, A&A, 537, A2Monaco L., Villanova S., Carraro G., Mucciarelli A., Moni Bidin C., 2018,A&A, 616, A181Origlia L., Rood R. T., Fabbri S., Ferraro F. R., Fusi Pecci F., Rich R. M.,2007, ApJ, 667, L85Origlia L., Ferraro F. R., Fabbri S., Fusi Pecci F., Dalessandro E., RichR. M., Valenti E., 2014, A&A, 564, A136Piotto G., et al., 2012, ApJ, 760, 39Pritzl B. J., Venn K. A., Irwin M., 2005, AJ, 130, 2140Reimers D., 1975, Memoires of the Societe Royale des Sciences de Liege,8, 369Salaris M., de Boer T., Tolstoy E., Fiorentino G., Cassisi S., 2013, A&A,559, A57Sarajedini A., et al., 2007, AJ, 133, 1658Savino A., Tolstoy E., Salaris M., Monelli M., de Boer T. J. L., 2019, A&A,630, A116Sbordone L., Bonifacio P., Marconi G., Buonanno R., Zaggia S., 2005,A&A, 437, 905Schröder K. P., Cuntz M., 2005, ApJ, 630, L73Schröder K. P., Cuntz M., 2007, A&A, 465, 593Tailo M., Di Criscienzo M., D’Antona F., Caloi V., Ventura P., 2016, MN-RAS, 457, 4525Tailo M., D’Antona F., Caloi V., Milone A. P., Marino A. F., Lagioia E.,Cordoni G., 2019a, MNRAS, 486, 5895Tailo M., Milone A. P., Marino A. F., D’Antona F., Lagioia E. P., CordoniG., 2019b, ApJ, 873, 123Tailo M., et al., 2020, MNRAS, 498, 5745Ventura P., Zeppieri A., Mazzitelli I., D’Antona F., 1998, A&A, 334, 953Villanova S., Geisler D., Piotto G., Gratton R. G., 2012, ApJ, 748, 62This paper has been typeset from a TEX/L A TEX file prepared by the author.MNRAS000