MUSE observations of NGC330 in the Small Magellanic Cloud. Helium abundance of bright main sequence stars
DDraft version February 10, 2020
Typeset using L A TEX twocolumn style in AASTeX62
MUSE observations of NGC 330 in the Small Magellanic CloudHelium abundance of bright main sequence stars ∗ R. Carini, K. Biazzo,
2, 1
E. Brocato, L. Pulone, and L. Pasquini INAF - Osservatorio Astronomico di Romavia Frascati 33, 00078 Monte Porzio Catone, Italy INAF - Osservatorio Astrofisico di Cataniavia S. Sofia 78, 95123 Catania, Italy ESO - European Southern ObservatoryKarl Schwarzchild Str. 2, Garching b. Munchen, Germany (Accepted)
Submitted to AJABSTRACTWe present observations of the most bright main sequence stars in the Small Magellanic Cloud stellarcluster NGC 330 obtained with the integral field spectrograph MUSE@VLT. The use of this valuableinstrument allows us to study both photometric and spectroscopic properties of stellar populations ofthis young star cluster.The photometric data provide us a precise color magnitude diagram, which seems to support thepresence of two stellar populations of ages of ≈
18 Myr and ≈
30 Myr assuming a metallicity of Z = 0 . R eff = 20 (cid:48)(cid:48) of NGC 330, thus leading to an estimation of (cid:15) ( He ) = 10 . ± .
05 (1 σ ).The helium elemental abundances of stars likely belonging to the two possible stellar populations,do not show differences or dichotomy within the uncertainties. Thus, our results suggest that the twostellar populations of NGC 330, if they exist, share similar original He abundances.If we consider stellar rotation velocity in our analysis, a coeval (30 Myr) stellar population, expe-riencing different values of rotation, cannot be excluded. In this case, the mean helium abundance < (cid:15) ( He ) > rot obtained in our analysis is 11 . ± .
05 dex. We also verified that possible NLTE effectscannot be identified with our analysis because of the spectral resolution and they are within our derivedabundance He uncertainties.Moreover, the analysis of the He abundance as a function of the distance from the cluster center ofthe observed stars do not show any correlation.
Keywords: stars: abundances — techniques: spectroscopic — galaxies: Magellanic Clouds — galaxies:individual: NGC 330 INTRODUCTIONNGC 330 has been early recognized as one of the mostpopulated (total mass of ∼ × M (cid:12) , Mackey Corresponding author: Roberta [email protected] ∗ Based on observations made with the MUSE integral-fieldspectrograph operating at the Very Large Telescope (SierroParanal, Chile) during the commissioning of the instrument. & Gilmore (2003); McLaughlin & van der Marel (2005))and brightest young stellar cluster ( ∼
30 Myr; e.g. Siri-anni et al. (2002); Keller et al. (2000))of the SmallMagellanic Cloud (SMC) (Arp 1959; Robertson 1974).The color-magnitude diagram (CMD) of the cluster dis-closes a blue main sequence and two clumps of super-giant stars, located in the red and blue part of the di-agram, clearly recognizable and distinct from the mainsequence stars. These features can be understood interms of massive stars experiencing the core H-burning a r X i v : . [ a s t r o - ph . S R ] F e b Carini et al. phase (main sequence) and the core He-burning phase(clumps). Moreover, the presence of a quite high frac-tion of Be stars is also well established (Feast 1991),which makes the interpretation of this cluster compli-cated and, at the same time, challenging (e.g., Caloiet al. 1993; Grebel & Richtler 1992; Grebel et al. 1996;Keller et al. 1999, 2000; Martayan et al. 2007b,a; Tanab´eet al. 2013). Nevertheless, NGC 330 remains a very at-tractive laboratory to improve our knowledge about thestellar evolution theory and the physics of intermediatemass stars born in a low metallicity environment such asthe SMC. For instance, from an evolutionary point-of-view, the ratio between the number of blue and red su-pergiants represents a fair indicator of the relative timean intermediate mass star spends along the He-burningloop. The quoted large fraction of Be stars ( ∼ ∼ . − . Y (e.g.D’Antona & Caloi 2004; Piotto 2008; Pasquini et al.2011). This is confirmed by Marino et al. (2014) whoprovided the first direct spectroscopic measurement ofhighly He-enhanced ( Y ∼ .
34) second generation starsin the Blue Horizontal Branch of the globular clusterNGC 2808. The new paradigm for the interpretation ofthese observations is that the GGCs are composed bytwo (or more) populations of stars, the SG stars wereborn from matter expelled by evolving stars of the FGstars and nuclearly processed through the hot CNO cy-cle. According to the theoretical scenarios that try toexplain the phenomena of the multiple populations inGCs, the process of formation of the second generationstars happened within 100-150 Myr from the formation of the FG stars (D’Ercole et al. 2008, 2016; Decressinet al. 2007; Bastian et al. 2013; Gieles et al. 2018).In the recent years, the presence of multiple popu-lations in Magellanic Clouds (MCs) Globular Clusters(GCs), similar to those found in the Milky Way, hasbeen demonstrated by observational evidences. It wasindeed found that many star clusters in the MagellanicClouds show bimodal or extended main sequence turnoff(eMSTO) and dual clump in their color magnitude dia-grams, suggesting the presence of the multiple popula-tions of stars with possible different ages (e.g., Mackey& Broby Nielsen 2007; Glatt et al. 2008; Girardi et al.2009; Milone et al. 2009; Goudfrooij et al. 2014). Morerecently, in young stellar clusters (age less than 600 Myr)have been observed not only the eMSTO, but also thesplit of the MS, similar to that detected in Milky WayGCs (Milone et al. 2015, 2016, 2017, 2018). These pho-tometric evidences could be interpreted with the pres-ence of stellar populations with different ages, but alsowith coeval populations with different stellar rotationvelocity (e.g., Bastian & de Mink 2009; Brandt & Huang2015; D’Antona et al. 2015; Marino et al. 2018). Un-til now, spectroscopic studies of stars in MCs clustersyounger than ∼ Y ∼ . ± .
006 dex) in the secondpopulations stars of four ∼ USE observations of NGC 330 Table 1.
Selected parameters of MUSE in a nutshell andrelevant quantities of the observations.Parameter ValueNumber of IFU models 24 (images slicer +spectrograph + CCD)Wavelength coverage 4800-9300 ˚A (nominal range)Field of view 59” ×
60” (WFM)Spatial sampling 0.2” (WFM)Multiplex factors 1152 slices, 90000 spaxels,3700 wavelength binsRA (J2000) 00:56:18.2DEC (J2000) − (cid:48)(cid:48) Airmass 1.5 (cid:48)(cid:48) T exp
200 sMean spectral resolution 2.75 ˚A
The Multi Unit Spectroscopic Explorer (MUSE; Ba-con et al. 2010) operating at the Very Large Telescope(VLT) of the European Southern Observatory (ESO)provided us the opportunity of obtaining simultaneouslyphotometric and spectroscopic data for a large numberof stars in NGC 330. Thanks to the exceptional observ-ing capabilities of this instrument, in this work we inves-tigate the presence of multiple populations and possiblestar-by-star differences in helium abundance.In the present paper, we describe the observations anddata reduction in Sect. 2, while the analysis of thephotometric data are presented in Sect.3, together withthe comparison to isochrones of different ages and rota-tion velocity to obtain hints on the presence of multi-populations. The spectroscopic data analysis and themeasurement of He abundances are presented in Sect.4. The effect of stellar rotation is also considered in thissection. The discussion on the age of the cluster and onthe radial distribution of He abundances are reported inSect. 5, while the final remarks are provided in Sect. 6. OBSERVATIONS AND DATA REDUCTIONNGC 330 was observed with the integral-field spectro-graph MUSE (Kelz et al. 2016) operating at the VLTduring the commissioning run in August 3rd, 2014. Theobservation covered the central area of NGC 330, witha field of view of ∼ × , a pixel scale of 0.20arcsec/pixel and a total exposure time of 200 s. Forthis observation the normal Wide Field Mode (WFM)was used and the parameters of the instrument setupadopted are summarized in Table 1.The data reduction of the instrumental raw data fromthe 24 CCDs was performed by using the MUSE pipeline Figure 1.
MUSE image of NGC330 at λ ∼ (Weilbacher et al. 2014). This procedure provides a ’re-duced’ datacube (two spatial and one wavelength axis)where bias subtraction, flat fielding, flux and wavelengthcalibration are properly taken into account. Correc-tion for instrumental and atmospheric effects, geomet-rical calibration and sky subtraction are also performedwithin the context of this pipeline. As a result of thisdata reduction procedure, Fig. 1 shows the image ofNGC 330 at ∼ V and I pass-bands as defined by Landolt 1992) wereextracted from the datacube. In this way, we were ableto derive the photometry and color-magnitude diagramdirectly from our data.Moreover, we extracted from MUSE database thespectra of a sample of ten B stars (Fig. 2). In par-ticular, we selected one by one the stars suffering lesssevere crowding and, in case of moderate overlappingpoint spread functions (PSFs), we took particular care inextracting the spectra from the central region of the PSFwhere the flux of nearby star was negligible. Clearly, thisimplies a slightly lower signal-to-noise ( S/N ) for the se-lected star but minimizes the contamination of nearby
Carini et al.
E(B-V)=0.08 A4A5 A10A17A21A23A24A34 A40A48
Figure 2.
Observed CMD of NGC 330. The name of thestars (designation from Robertson (1974)) analyzed in thiswork are also reported. star in the extracted spectra. Taking advantage of thePSF analysis performed with daophot (Stetson 1987),we obtained spectra which contain most of the flux ofthe PSF ( ≥ ∼ Molecfit models themost appropriate atmospheric profile (i.e., the variationin temperature, pressure, and humidity as a function ofthe altitude) at the time of the given science observa-tions. As input it requires ambient parameters (e.g.telescope altitude angle, humidity, pressure, ambienttemperature, mirror temperature, elevation, longitude,latitude), instrumental parameters (e.g. slit width, pixelscale), molecular columns (i.e. which molecules haveto be considered, which depend on the spectral regionanalyzed), background and continuum (e.g. flux unit,polynomial fit for continuum), spectroscopic resolution ,and so on. In the wavelength range of MUSE the rele-vant molecules are O and H O . A single atmosphericprofile is compiled from data from three sources: astandard atmospheric profile for a given climate zone Figure 3.
The entire spectrum of the B star A17 in NGC 330obtained with MUSE@VLT. (produced for Michelson Interferometer for Passive At-mospheric Sounding on board the ENVISAT satellite),Global Data Assimilation System (GDAS) model pro-file, and the corresponding ground-based by the ESOMeteo Monitor measurements (EMM). The first one in-cludes information on pressure, temperature, and molec-ular abundances as function of height. The second one isprovided by the National Oceanic and Atmospheric Ad-ministration (NOAA) , which gives information aboutthe weather forecast for the location of Cerro Paranal toan altitude of ∼
26 km. The third one provides informa-tion on the local meteorological conditions in the ESOsite Paranal taken from a local meteo station mountedon a 30 m high. After several iterations of the χ min-imization procedure, Molecfit writes the best-fit spec-trum for telluric correction. This process takes into ac-count the optimization for scaling the wavelength gridand the resolution of the model. On the basis of this fit,the code calculates the atmospheric transmission for thewavelength range of the input spectrum and corrects itfor telluric absorption. We applied this procedure foreach spectrum. PHOTOMETRY AND COMPARISON WITHISOCHRONESThe V and I images have been analyzed by using the daophot package developed to perform stellar photom-etry in crowded fields (Stetson 1987). The detection http://140.90.198.158/pub/gdas/rotating USE observations of NGC 330 σ above the background level was adopted.We selected a dozen of stars external to the core of thecluster to find the best PSF. Instrumental v and i mag-nitudes were converted into the standard V, and I John-son/Cousins system using, as calibrators, 12 stars in theMUSE field already detected and with magnitudes mea-sured by Udalski et al. (1998, 2008) . The resultingadopted calibration equations are: V = v − [0 . ± . v − i ) + [3 . ± . I = i + [0 . ± . v − i ) + [2 . ± . < V − I < daophot , the V and I photometry of about 250 brightstars with a mean uncertainty of ∼ .
01 mag was ob-tained. In this sample, 34 stars are found in commonwith Robertson (1974) and in the following they willbe indicated with the same name (i.e., with the prefixA). In Fig. 4 the CMD obtained with our measures ispresented by assuming a distance modulus µ of 19.0mag and a reddening E ( B − V ) = 0 .
08 mag (see Caloiet al. 1993, Keller et al. 2000). In the same figure, wealso overplot the PARSEC (Bressan et al. 2012; Marigoet al. 2017) isochrones with Z=0.002 (Spite et al. 1991;Reitermann et al. 1990) and without rotation.The quite broad main sequence extends up to M V ∼− . m V =14.5 mag), in agreement with Chiosiet al. (1995) and Li et al. (2017) (see also referencetherein), where the red and the blue groups are alsoclearly represented in our sample. These blue and redgroups of stars, which are quite brighter than the mainsequence termination, seem to be core He-burning starspopulating the hot and cool limit of the well known loopof the effective temperature foreseen for stellar modelsof intermediate mass ( ≈ − M (cid:12) ) stars. The loopextends from ( V − I ) ≈ . ≈ . http://ogledb.astrouw.edu.pl/ ∼ ogle/photdb/index.html http://pleiadi.pd.astro.it/ E(B-V)=0.08Z=0.002 18 Myr50 Myr 10 Myr 30 Myr
Figure 4.
Color-magnitude diagram of the selected starsin NGC 330. The isochrones at 10 Myr, 18 Myr, 30 Myrand 50 Myr obtained from the PAdova and TRieste StellarEvolution Code (PARSEC; Bressan et al. 2012; Marigo et al.2017) are overplotted with short green dashed, red solid andblue solid and cyan short-long lines , respectively. The best-fitting isochrones are 18 Myr and 30 Myr. stars. More interestingly, the two extended loops of theisochrones, which are due to stars experiencing the coreHe-burning phase, appear to reproduce quite preciselythe luminosities and the colors of the blue and red groupsof supergiants. It is relevant to recall that stellar modelsof intermediate mass (of the adopted metallicity) predictthat the evolutionary time of the He burning phases isspent part at low T eff ( ≤ T eff ( ≥ V ∼ − . V − I ) ∼ . ≈ . (cid:39) Carini et al.
In order to provide a quantitative example of how theisochrones change if different ages are assumed, we plotin Fig. 4 two isochrones with 10 Myr and 50 Myr. Thefigure shows that the isochrone with 10 Myr does notmatch the location of the observed stars, while the 50Myr one reproduce the MS stars but the coolest (andmore populated) part of the core-he burning phase ismuch redder than the location of the red clump stars.We stress here that our outcome is based on very fewsupergiants and needs further confirmation. Neverthe-less, one single isochrone is not able to fit all the su-pergiants of NGC 330 and at least two isochrones arerequired to fit the stars in the He-burning phase dueto the observed spread/separation in their magnitudes(∆ V (cid:39) . m V ≤ . M V (cid:39) . Effect of stellar rotation
We dedicate a paragraph to the discussion on stellarrotation, because, even if our spectra do not have theresolution high enough to analyze this issue in details,it is known that many OB stars are fast rotators ( v sin i up to 300 km/s). In particular, Hunter et al. (2011)analyzing high-resolution VLT-FLAMES data for starsin LMC and SMC, found that the mean v sin i for 77B-type stars of NGC 330 with masses less than 25 M (cid:12) (as in our sample) is about 150 km/s. Stellar rotation isa parameter directly related to the size of the convectivecore, (for a detailed treatment of the rotation we quotee.g. Maeder & Meynet (2000); Meynet & Maeder (2000)and references therein). Here, we recall only the maineffects of the rotation on the evolution of the B stars atlow metallicity: • the convective core of the stars increases duringthe MS phase; • the luminosity of a rotating star is higher (about0.5 mag) than that of a non-rotating star samemass; • the combination of these two effects implicatesthat the lifetime in the H-burning phase grows,but only moderately (about 20% −
30% for an ini-tial velocity of 200 km/s); • the envelopes of rotating stars (angular velocity ω > ω c , with ω c critical angular velocity) areenriched in CNO-processed material, in particularof He and N for stars with M > M (cid:12) ; • also during the He-burning phase the luminosity ishigher because of the larger He cores, if the massloss is not too strong; • the rotation inhibits the formation of stars in thered clump during the He-burning phase, but thisis contrasted by the mass loss effect that favor theformation of red supergiants. Until now the mod-els can not reproduce the ratio between blue andred supergiants in the solar neighborhood observedat metallicity of SMC (Brunish et al. 1986; Schalleret al. 1992). Some models for the most rapid rota-tors at low metallicity predict the disappearanceof the red clump, and the He-burning occurs onlyin the blue part of the HR diagram (log T eff ) ∼ (Georgyet al. 2014), provided by the SYCLIST code. We tookinto account the work done by Milone et al. (2018), inwhich they show the presence of two families of stars inNGC 330, the bluest formed by non-rotating stars withan age compatible with 32 Myr (in agreement with ourresults), the reddest by rotating stars with angular ve-locity ω = 0.9 ω c and age around 40 Myr, representing40-55% of MS stars.As shown in Fig. 5, it is evident that the evolutionarytrack of Geneva predicts the absence of the red clumpwith this age and rotation, so they can not explain themajority of the He-burning stars. For this reason, thepresence of two populations is required. One composedby non-rotating 30 Myr old stars (green long-dashedline) The second one can be identified by the stars su-perimposed to the isochrones of rotating models at 25Myr with ω = 0.5 ω c (blue line) and 30 Myr with the ω = 0.9 ω c (red-dashed line).In this case, we remain with a (nearly) coeval popula-tion formed by a mixture of rotating and non-rotatingstars. We keep in mind that the claim that the clusterincludes a secondary rotating population is based on thelocation of one blue giant star, like the previous scenariowhere a 18 Myr population has been suggested (see Fig.4). The presence of stars with -5.5 < M v < -4 mag, afterthe overall contraction, could favor the isochrone with USE observations of NGC 330 Ebv = 0.08 Z = 0.002
Figure 5.
Color-magnitude diagram of NGC 330. Theisochrone at 30 Myr is obtained from the PAdova andTRieste Stellar Evolution Code (PARSEC; Bressan et al.2012; Marigo et al. 2017) (the green long-dashed line); theisochrones computed from rotating stellar models at 25 Myrwith ω = 0 . ω c (blue line), and at 30 Myr with ω = 0 . ω c (red dashed line) are taken from Georgy et al. 2014. Theadopted values for the age, distance modulus, reddening, andmetallicity are quoted in the figure. ω = 0 . ω c because they should be He-burning stars.One of these stars is A4, already analyzed by Lennonet al. (2003) who found vsini = 20 km/s, thus favoringthe hypothesis that the second family of stars are notfast rotators. We remark here that the overshooting isstronger in the Padova evolutionary tracks with respectto the Geneva ones, so an error of about 5 Myr shouldbe also taken into account. SPECTRA EXTRACTION AND ANALYSISIn this section, we make use of the spectral data se-cured by MUSE observations to investigate the presenceof a possible spread of helium abundances in the B starsof NGC 330. Such a difference would be extremely inter-esting because it would support the possible existence oftwo or more populations with different He content, andwould also help in alleviating the tension between mod-els and the extended MSTO observed in NGC 330 (Liet al. 2017). Furthermore, He is a key element in thecontext of the multiple populations in globular clusters(D’Antona & Caloi 2004; Piotto et al. 2007; Piotto 2008)and discovering the presence of dichotomy or spread ofHe abundance within stars members of a young clusterslike NGC 330 would be extremely interesting. The hightemperature of these B stars ensures us the presence of He i features in their spectra. We decided to excludeBe stars to minimize uncertainties due to rotation andcomplexity in the spectra analysis.4.1. Spectroscopic Stellar Parameters
The procedure adopted to derive the He abundancesfrom the spectra of the selected B stars is briefly outlinedin this section.We used the
SME : (Spectroscopy Made Easy, ver-sion 522; Valenti & Piskunov 1996; Piskunov & Valenti2017) package to determine the fundamental parameters(effective temperature T eff , surface gravity log g , radialvelocity V rad ) and He abundance. The measurementsof these quantities was performed in few key steps andsynthetic spectra were computed to obtain the best-fitof the observed spectroscopic data. SME is a spectralsynthesis code which allows us to find the best-fit of anobserved spectrum, assuming a wavelength range andinitial input parameters. In particular,
SME needs linelist data for all atomic transitions of interest (i.e. ele-ment, ionization state, wavelength, excitation energy ofthe initial state, log gf ; (Vienna Atomic Line Database3) Piskunov et al. (1995)) and model atmospheres (AT-LAS12, Kurucz (2013)). Plane parallel geometry, negli-gible magnetic field and no bulk flows are assumed by SME .The parameter optimization code uses the Marquardtalgorithm (W. Marquardt 1963; Press et al. 1992) toobtain estimates of the parameters, through minimiz-ing the χ statistic by comparing model and observedspectra (Valenti & Piskunov 1996).We assumed local thermodynamical equilibrium(LTE), as the resolution of the spectra is too low toappreciate the difference in He abundance due to NLTE.The evaluation of the effect of the NLTE on our resultsis treated in Section. 4.4.To derive stellar parameters removing the log T eff − log g degeneracy, we decided to fit from normalized spec-tra the H β and the HeI at 4921.9 ˚A features of each starconsidering the range 4800 − A .Before computing the synthetic spectra, input val-ues of the stellar atmosphere have to be provided. Tothis purpose, we used both the photometric informa-tion obtained from our best isochrones fit and fromthe direct comparison with the colors of the Castelli& Kurucz (2003) models. The input values for eachstars of our sample can be found in Table 2 ( T phot eff andlog g phot ). Moreover, we fixed the microturbulence at 5km/s, which is a reliable value for B stars (see Lennonet al. 2003), and Z = 0 .
002 (i.e. [Fe/H] ∼ −
1; see Spite Carini et al.
Figure 6.
Best-fit of the H β and HeI lines (labelled) for thestar A4. et al. 1991; Reitermann et al. 1990; Lennon et al. 1996,2003; Hill 1999). The spectral resolving power rangesfrom ∼ ∼ T eff , log g and V rad from the best-fit of the spectra in the region H β -HeI.An example of the best-fit of the H β add HeI lines isshown in Fig. 6 for the star A4.The resulting stellar parameters ( T eff , log g , V rad ) withtheir uncertainties, and the typical S/N at λ ∼ SME context,according with Press et al. (2002), refer to intrinsic er-rors, and they could be underestimated due to system-atics, like reddening, theoretical assumption, etc.4.2.
Helium abundance
Once the stellar parameters of each star were eval-uated, we repeated the same steps as in the previoussection but for the He features listed in Table 3, i.e weconsidered the following wavelength ranges: 4880-4960˚A, 5000-5030 ˚A, 5030-5100 ˚A, 5860-5890 ˚A, 6640-6720˚A, 7000-7100 ˚A (see Fig. 7 and Fig. 8). As done forthe determination of the stellar parameters, also for theHe features we considered the atomic and molecular linelist from VALD3 (Piskunov et al. 1995) and the spectrawere normalized to their continuum level.Stellar parameters were constrained to the values de-rived by fitting the H β and HeI lines and then used tocreate the synthetic spectra around each He line. AT-LAS12 model atmospheres (Kurucz 2013) were consid-ered and the best-fit was obtained around each spectralrange through the
SME package to derive the heliumabundance. The values of the He abundance for eachstar and each feature are reported in Table 3. Internalerrors computed by
SME and mostly due to uncertain-ties in atomic parameters, stellar parameters, and con-
Figure 7.
Each panel shows the region of the spectrumwhere the He features used in this work are located. Thespectra are normalized but not red-shifted. In every panelthe spectra of the stars A4, A5, A10, A17, A21 are plottedstarting from the top and shifted of 0.05 in the y-axis forgraphical reasons. tinuum position are also reported. Empty values in thistable refer to features for which the best-fit was not ob-tained, in most cases because the observed line was tooweak or not detectable. We also evaluated the impactof external errors considering the photometric T eff and log g . To this aim we recomputed the best-fit of eachfeature by adopting the photometric values of stellar pa-rameters and derived mean errors in the final heliumabundance of ≈ . ≈ .
05 dex due to differencechoice of T eff and log g , respectively. These errors areindeed within the internal errors listed in Table 3. Sys-tematic uncertainties could affect the helium abundancedue to the effect of the assumption on the microturbu-lence. We tested the impact of varying the microturbu-lence from 0 km/s to 10 km/s. The results show that anenhancement of microturbulence to 10 km/s leads to adecrease of He abundance of about 0.2 dex) with respectthe values determined with ξ = 5 km/s. While a valueof ξ = 0 km/s leads to an increase of He up to 0.1 dex.The systematic uncertainties due to the debated valueof metallicity in NGC 330 may affect the He determina-tion. We calculated the He abundance of the stars in oursample by assuming the lowest ([Fe/H]=-1.8, Richtler& Nelles 1983) and the highest ([Fe/H]=-0.5, Meliani USE observations of NGC 330 Table 2.
Input (photometric) and output (spectroscopic) parameters obtained from our analysis. Errors in spectroscopicparameters are listed for each target. Typical
S/N at about 6000 ˚A are reported in the last column.
Star T photeff (K) σ ( T photeff )(K) log g phot σ (log g phot) T speceff (K) σ ( T speceff )(K) log g spec σ (log g spec) V rad(km/s) σ ( V rad) (km/s) S/N
A4 20000 1000 3.0 0.5 17000 1000 2.7 0.1 152 8 300A5 24000 1000 4.0 0.5 22500 2000 3.7 0.2 155 40 100A10 22000 1000 4.0 0.5 22500 1000 3.9 0.1 140 20 130A17 24000 1000 4.0 0.5 22000 1000 3.7 0.1 147 15 200A21 23000 2000 4.0 0.5 23500 1500 3.9 0.2 150 20 200A23 33000 2000 4.0 0.5 22000 1000 3.6 0.2 142 11 200A24 26000 2000 4.0 0.5 23000 3000 4.1 0.4 140 25 100A34 23000 1000 4.0 0.5 22500 2000 4.0 0.2 145 20 120A40 16000 1000 3.0 0.5 18500 500 3.2 0.1 145 10 300A48 32000 2000 4.0 0.5 25000 2500 3.6 0.5 152 45 200
Figure 8.
As for Fig. 7 but for the stars A23, A24, A34,A40, A48 (from top to bottom). et al. 1995) values commonly assumed in the literaturefor this cluster. It turns out that the He abundance Yin our stars changes of an amount less than 0.01 dex.In Fig. 9 we show the logarithmic helium abundance( (cid:15) ( He ) determined for each He i line (4921.9 ˚A, blackcircles; 5015.7 ˚A, red squares; 5047.7 ˚A, blue diamonds;5875.6 ˚A, green stars; 6678.2 ˚A, magenta open trian-gles; 7065.7 ˚A, cyan filled triangles) and for each staranalyzed. Uncertainties obtained by SME are also dis-played.We note that the He abundance determined for the6678.2 ˚A is systematically higher than ones obtainedfrom the other lines, except for the stars A5 and A48. (cid:15) ( He ) is defined to be log( N He /N H ) + 12, where N He and N H are the number densities of elements He and H, respectively. This could be due to the contribution of the He isotopeenhancement (Schneider et al. 2018) which can not beresolved in our spectra.We then computed the mean of the He abundance val-ues derived from the best fitted features for each star.This computation was afterwards repeated using onlythe values deviating less than 1 σ from the initial av-erage evaluation. This method allowed us to minimizethe impact of single unpredictable errors provided bythe best fitting procedure of each He feature. The Heabundances averages and the standard deviation σ ofthe average are reported in the 8 th and 9 th columns ofTable 3, while the last column lists their correspondinghelium mass fraction value ( Y ).In order to check the reliability of our measurements,we also evaluated the weighted mean of the He abun-dances, i.e. using as weight the error on the (cid:15) ( He ) valueobtained by fitting each single feature. As expected, wefound that the two determinations of the He abundanceare in full agreement within their internal uncertainties.We notice that Lennon et al. (2003) have obtainedspectra of a sample of 7 stars in NGC 330 and one star(A4) is in common with our work. The T eff and the sur-face gravity we derive in this work are in good agreementwith the values obtained by these authors for the samestar. While, the He abundance found by Lennon et al.(2003) using nine lines is (cid:15) ( He ) = 10 .
66, which is lowerthan the value found by us (i.e. (cid:15) ( He ) = 10 . ± . A4 The spectrum of the star A4 presents one of the high-est
S/N ratio of our sample, with a value of about300. The spectral regions where the He features usedin this work are located are shown in Fig. 7 (first solidlines from the top). The line 5015.7 ˚A is faint butthe shape is regular and the best-fit was obtained, evenif the resulting He abundance for this feature is verylow ( (cid:15) ( He )=10.35). As already mentioned, this is theonly star of our sample for which the He abundance hasbeen reported by other authors (Lennon et al. 2003).0 Carini et al.
Table 3.
Helium abundance ( (cid:15) ( He )) derived from our analysis for each line. Effective temperature and surface gravity areshown in Table 2. Star (cid:15) ( He ) (cid:15) ( He ) (cid:15) ( He ) (cid:15) ( He ) (cid:15) ( He ) (cid:15) ( He ) < (cid:15) ( He ) > σ Y σ (Y)( λ λ λ λ λ λ ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Figure 9.
Logarithmic He abundance obtained for each he-lium line. Each symbol refers to different line. Uncertaintiesobtained by
SME are also shown. x -axes reports the numberof each star following the nomenclature by Robertson (1974). They derived for this target spectroscopic values of T eff =18000 K, log g =2.8 ± ξ = 5km/s. These quantities are in very good agreement withour results ( T eff ∼ ± g =2.7 ± (cid:15) ( He ) = 10.94 ± σ )) is higher (morethan 1 σ ) than the one found by Lennon et al. (2003)( (cid:15) ( He ))=10.66. We remark here that the abundance ofthis star is one of the lowest of the entire sample ob-served by Lennon et al. (2003). A5 S/N ratio of this star is about 100. The features at λ =5047.7 ˚A , λ = 5875 . λ =7065.7 ˚A arenoisy and hardly detectable, therefore no best-fit was obtained. This star shows features larger than mostof the other stars, thus suggesting very high rotation(see, in particular, the features centered at 5875.6 ˚Aand 6678.2 ˚A). The mean value of the He abundance is (cid:15) ( He ) =10.93 ± A10
The effective temperature and the surface gravity de-rived from the synthetic spectra are in very good agree-ment with the photometric values (see Table 3). Forthe computation of the He abundance, we used all lineswith the exception of the features at λ =5015.7 ˚A and λ =5047.7 ˚A. As found for A5, also this star shows verywide features, suggesting high rotation (see solid linesin the middle of each panel of Fig. 7). In this case thederived He abundance resulted to be slightly sub-solar(i.e. 10.90 ± (cid:15) ( He ) (cid:12) = 10 .
93 as he-lium solar abundance (Asplund et al. 2009).
A17
The effective temperature and the surface gravity arein agreement, within the uncertainties, with the theo-retical values and all the selected He features were usedto derive the final He abundance, which resulted to be (cid:15) ( He )= 10.94 ± A21
Again the effective temperature and log g are in verygood agreement with the photometric values. The linesare particularly broadened for this star, and the uncer-tainties in the determination of (cid:15) ( He ) is between 0.5 and0.6 dex. The mean value is (cid:15) ( He )=10.90 ± A23
From the photometry, we found that A23 is the targetof our sample with the highest effective temperature (i.e.33000 K), while the result of the spectral best fitting ofthe H β and HeI lines is very different (i.e. 22000 K).Also in this case, all the He features were used for thecalculation of the mean He abundance, resulting to besolar. USE observations of NGC 330 A24
Spectroscopic effective temperature and surface grav-ity are in agreement, within the errors, with the theo-retical values, despite the
S/N ratio is one of the lowestwithin our sample ( ≈ λ . ± . A34
Also this star shows a relatively low
S/N ratio ( ≈ ∼ . < (cid:15) ( He ) > is indeed close to the solar value ( (cid:15) ( He ) = 10.95 ± A40
The
S/N ratio of this star is very high, being largerthan 350. The effective temperature from the spectra isslightly higher than the theoretical one, while the log g is in good agreement. All the He features were used toderive the He abundance; the He mean value is, (cid:15) ( He )= 10.91 ± A48
Our spectroscopic best-fit confirms the theoreticalvalue of log g within the uncertainties, but not the T eff ,that is lower of 7000 K. The lines of this target seemto suggest high v sin i (see Fig.8). We fitted each lineexcept for λ (cid:15) ( He )=10.84 ± Stellar Rotation
As already discussed in the previous sections, the starsin our sample could rotate and this may affect the spec-tral features. Unfortunately, the resolution of our spec-tra is too low to evaluate stellar rotation velocities lowerthan 150 km/s. We made an attempt to evaluate thecontribution of the stellar rotation in the helium abun-dance. To this aim, we repeated the same procedureperformed in Sect. 4.1 by adopting a rotating modelframework. We first used the isochrone of 30 Myr withthe angular velocity ω =0 . ω c , already shown in Fig. 5,to evaluate the input parameters ( T photeff ,log g phot , v sin i )for each star. These values are then used to compute thebest fit of the spectra in the spectral regions of the H β and He I at 4921.9 ˚A lines. Once the best fit is reached,we estimated again the abundance of helium by fitting each of the available lines. The results are reported inTables. 4 and 5, where we show the photometric valuesconsidered as input parameters, the ones resulting fromour best fit, and the mean abundance of He, respectively.The photometric and spectroscopic values of the tem-peratures and gravities are in a good agreement, withinthe errors, but the A5 and A48. It is interesting to notethat, the obtained T speceff (K) and log g spec values are alsoin fair agreement, within the uncertainties, with the onesfound from non-rotating models.The fitting procedure disclosed that, for some starsas A5, A17 and A23, the projected rotation velocityobtained from spectroscopy moves away from the inputvalues, typically decreasing their values. The highest vsini are found for the stars A5 and A10, as also evidentfrom Fig. 7. In fact, when the stellar rotation is takeninto account, the quality of the fit improves and the bestfit is obtained for nearly all the available He lines.Fig. 10 shows the helium abundance of each star.Black filled and red empty circles represent respectivelythe mean helium abundance considering or not the stel-lar rotation. The measured mean abundances deter-mined considering the rotation are systematically higherthan those calculated without rotation. This resultagrees with the expectations from the theory discussedin Sect. 3.1. We note that A4 is an exception, with themean value obtained considering the rotation slightlylower than the one determined without rotation. Forthis star is not possible to fit all spectral lines consid-ering the stellar rotation. For this reason, we suggestthat this source is not a rotating star, in agreement withLennon et al. (2003).Even if the resolution of the data and the uncertaintiesof the results require caution, we find that the rotatingframework seems to suggest higher helium abundancesthan the values found using non-rotating models. Themean < (cid:15) ( He ) > rot = 11 . ± .
05 can be compared withthe value obtained for the non-rotating one < (cid:15) ( He ) > =10 . ± . NLTE effect
Conditions of non-LTE could play a role on the esti-mation of the He abundance because they affect bothline cores and wings and therefore the equivalent widthsof the He lines. (Schneider et al. 2018; Przybilla 2005).For this reason, in this section we briefly discuss theNLTE effect.2
Carini et al.
Table 4.
Input (photometric) and output (spectroscopic) parameters obtained from our analysis considering the stellar rotationvelocity. Errors in spectroscopic parameters are listed for each target.
Star T photeff (K) σ ( T photeff )(K) log g phot σ (log g phot ) v sin i input ( km/s ) T speceff (K) σ ( T speceff )(K) log g spec σ (log g spec ) v sin i output ( km/s )A4 16800 1000 3.0 0.5 100 17000 1000 2.9 0.1 100A5 16000 1000 3.8 0.5 350 21400 3000 3.5 0.2 250A10 18500 1000 3.2 0.5 250 20500 1000 3.7 0.1 200A17 20000 1000 3.3 0.5 180 21700 1000 3.6 0.1 100A21 19000 1000 3.3 0.5 170 21000 1000 3.6 0.2 150A23 22000 1000 3.5 0.5 230 21600 2000 3.6 0.2 100A24 20500 1000 3.4 0.5 330 21700 1500 3.9 0.3 150A34 17800 1000 3.8 0.5 300 20500 1500 3.8 0.2 150A40 16500 1000 2.9 0.5 100 17500 500 3.1 0.1 100A48 18000 1000 3.1 0.5 150 22000 1500 3.1 0.3 150 Table 5.
Mean values, within 1 σ of the He abundance for each star of the sample obtained considering stellar rotation. Star < (cid:15) ( He ) > rot σ Y rot σ ( Y )A4 10.90 0.3 0.24 0.25A5 11.02 0.03 0.29 0.03A10 10.96 0.04 0.27 0.04A17 10.98 0.03 0.27 0.03A21 10.98 0.08 0.28 0.07A23 10.98 0.10 0.28 0.10A24 11.08 0.20 0.32 0.20A34 11.11 0.10 0.34 0.10A40 11.02 0.10 0.30 0.09A48 10.88 0.21 0.23 0.17 Figure 10.
Mean value of the He abundance for each targetconsidering ω = 0.9 ω c (black points), and ω =0.0 (red points).Error bars are also shown. Schneider et al. (2018) analyzed
He I lines in the opti-cal range of several B-type main sequence stars, studyingalso the isotopic shift of the He due to the presence ofthe He with the NLTE approach. From high-resolution spectra, they found the strongest departure from LTEapproach for the He I 5875.6 ˚A line. They compared theHe abundance of B-type with 20000 < T eff < n he ) evaluated are reported ofthe order of ± . DISCUSSION
USE observations of NGC 330
Cluster age
The photometric results of the bright region of theCMD of NGC 330 (see Section 3) suggest once more thatthe blue and red groups of supergiants, clearly iden-tified since early studies, are due to core He-burningstars (e.g., Stothers & Chin 1992, Brocato & Castellani1993). In this work, we also suggest that at least twoisochrones are needed to reproduce the position of thesestars in the CMD. In the non-rotation case, the separa-tion in age we find between the two isochrones is ≈ differencial magnitudes of very bright and well measured stars. Thissecures that we are dealing with differential magnitudeuncertainties which are very small ( ≈ .
02 mag).Uncertainties on absolute calibrations related to dis-tance modulus, reddening, calibrations and theoreticalassumptions are not expected to affect severely the dif-ference in age between the isochrones fitting the data,while they may change the absolute age evaluations.A weakness of this argument is the fact that blue su-pergiants in the observed CMD are a few and far fromany robust statistical sample, in comparison to the largenumber of main sequence stars. This is consequenceof the evolutionary timescales of the core He-burning,which are a factor of ≈
10 times shorter than the onesof the core H-burning evolutionary phase lifetimes. Nev-ertheless, NGC 330 is one of the youngest and most mas-sive star cluster in the local Universe and, for this rea-son, it remains a fundamental ensemble of stars to beconsidered in studying young stellar populations.It is interesting to highlight here that the mass of thestars in the core He-burning phase obtained from thedata of the two best-fit isochrones are 9 . ± . M (cid:12) and12 . ± . M (cid:12) (Bressan et al. 2012; Marigo et al. 2017).Since the ratio of their evolutionary time during thisphase is ≈ Table 6.
Mean Helium abundance of NGC 330 (see text). < (cid:15) ( He ) > σ <(cid:15) (He) > < Y > σ
18 Myr stars 10.84 0.13 0.22 0.05 ∼
30 Myr stars 10.91 0.08 0.25 0.03
Finally, we recall a further point related to stellar ro-tation. As we show in Fig. 5, isochrones computedwith stellar models which take into account rotation af-fect the determination of the age of the stellar popula-tions of NGC 330. In fact, in Sect. 3.1 we show thatthe CMD can be explained by one nearly coeval stel-lar populations composed by a mixture of rotating andnon-rotating stars.5.2.
He abundance and age
The presence of an age spread in the stellar popula-tion of NGC 330 leads us to investigate the existence ofpossible star-to-star difference in He abundance. Usingthe results obtained in Sect. 4, we now investigate pos-sible relationships between He abundance and generalproperties of the cluster.We evaluated the mean He abundance of NGC 330as obtained considering our homogeneous set of MUSEdata (we call these stars as
MUSE sample in Table 6),i.e. (cid:15) ( He ) MUSE sample = 10 . ± .
05 (see also Fig.11). Moreover, considering the sample of stars analyzedby Lennon et al. (2003), namely A1, A2, A4, B4, B22,B32, and B37, and the star B30 studied by Korn et al.(2000), we obtain an average of (cid:15) ( He ) other works =10 . ± .
13. Considering both He abundance results ofour
MUSE sample and of other works , the global valuewe obtain from the entire sample of stars with availableHe abundances is (cid:15) ( He ) entire sample = 10 . ± . MUSE sample . Therefore can conclude that the averagevalue of the He abundance of NGC 330 obtained withour (homogeneous) sample is consistent, within the un-certainties, with that found considering the entire - nothomogeneous - sample.We now investigate possible relationships between Heabundance and stellar age. To this purpose, we triedto assign to each star of the
MUSE sample the ageof the nearest isochrone according to their CMD loca-tion. Unfortunately, the differences in color between theisochrones are too small to make a safe separation among4
Carini et al.
Figure 11.
Mean value of the logarithmic He abundancefor each target. Error bars are also shown. The red dashedline represents the average of these values, while the yellowshadow region shows the ± σ level. stars located near the blue (older) or the red (younger)isochrones (see Fig. 4). We therefore considered thelog T eff − log g diagram for both the MUSE sample andthe stars analyzed in other works ..The log T eff − log g diagram of the entire sample ofNGC 330 stars with He abundance determination is pre-sented in Fig. 12.Firstly, this plot confirms our previous findingsthat the stars of NGC 330 seem to show age differ-ences/spread of ∼
12 Myr. Then, if we arbitrarily as-sume from the log T eff − log g diagram an age of 18 Myrfor the stars A2, A4, A48, B22, B30, B32, and B37, andan age of 30 Myr for the stars A1, A5, A10, A17, A21,A23, A24, A34, and A40, the mean He abundance foreach group of stars is < (cid:15) ( He ) > = 10 . ± . < (cid:15) ( He ) > ≈ . ± .
07, respectively (seeTable 3). We hence find that, within the errors, theresulting He abundances of the two groups of stars sep-arated in age do not show relevant differences.If the rotational scenario is taken into account, thelog T eff − log g diagram changes (see Fig. 13). Firstlybecause the fits of the spectra provide different valuesof T eff and log g when the rotation of the star is consid-ered, secondly the isochrone obtained by rotating stel-lar model has a different pattern from a ”non-rotating”isochrone.The ”rotating” isochrone of 30 Myr which reproducesthe CMD of NGC 330 in Fig. 5 fails to match the posi- A2B37A4 B30A4L B22 B32 A40A23A17A5A48A1A21 A10A34A24
Figure 12. log T eff − log g diagram for the stars analyzedspectroscopically by us with MUSE (black dots) and by otherauthors (magenta dots). The Z = 0 .
002 PARSEC isochronesof 18 and 30 Myr are also reported with red and blue solidlines, respectively. We report also the name of the stars. Weshow the log T eff and log g of A4 determined by us (A4) andby Lennon et al. (2003)(A4L). tion in the log T eff − log g diagram of several stars (A24,A34, A10, A21, A23, A17, A48). On the contrary, thebright stars (A4, A5, A40) appears to be located alongthe pattern of this isochrone. Thus, this comparisonsuggests once more that a scenario in which the massivestars in NGC 330 could have the same age but experi-encing a different rotation velocity cannot be excluded.In this plot we have not considered the sources analyzedby Lennon et al. (2003) and Korn et al. (2000) becausethe authors considered their stars as non-rotating.If the hypothesis of coeval populations is considered,we find that the mean helium abundance within 1 σ ofthe only ”rotating” stellar population is ∼ ∼ ∼ ∼ R eff = 20 (cid:48)(cid:48) ; Porte-gies Zwart et al. 2010). We can therefore investigate inthis internal cluster region the behavior of the stellar Heabundance with the distance from the center.In Fig. 14, we present our (cid:15) ( He ) values obtained usinghomogeneous MUSE data as a function of the distance USE observations of NGC 330 A4 A40A23A17A5 A48 A21A10A34A24
Figure 13. log T eff − log g diagram for the stars analyzedspectroscopically by us with MUSE (black dots) consideringstellar rotation. The Z = 0 .
002 SYCLIST isochrones of 30Myr and ω = 0 . ω c (Georgy et al. 2014) is reported with redsolid line. We report also the name of the stars.. of each star from the cluster center ( d ). The centerof the cluster was defined according to the definitionof Evans et al. (2006), i.e. α ( J h m . s and δ ( J − ◦ (cid:48) . (cid:48)(cid:48) . The result of our inves-tigation clearly points out that there is no correlationbetween the distance of the stars from the center of thecluster and their He abundances. This can be also quan-tified by a Kendalls’ rank correlation. For this test wefound a significance of about 0.71 for the (cid:15) ( He )-distanceand 0.70 for Y-distance correlations. This means thatthere is no apparent correlation between the two samplesof data (see Fig. 14). CONCLUSIONWe have presented an homogeneous analysis of photo-metric and spectroscopic data of the SMC young clusterNGC 330. The results can be summarized as follows: • We have found a possible difference in the age ofthe stars within NGC 330 with a spread/separationof the order of ≈ M yr . This evidence has beenderived also on the basis of the photometry of coreHe-burning stars. However, further observationalstudies are required to investigate if the brightest(youngest?) stars are binaries or not. • We can not exclude that the age spread is reducedor disappears, if stellar rotation is considered. • We have measured for the first time the He abun-dance of 10 stars placed in the center of NGC 330(e.g., r star ≤ R eff ). We have found a mean valueof < (cid:15) ( He ) > = 10 . ± • Considering also the stars studied in the past inthis cluster, we have found a mean global heliumabundance of < (cid:15) ( He ) > = 10 . ± • We evaluated the effect of rotation on the He abun-dance by fitting our spectra assuming the v sin i value for which the fit is reached. In this case amean global helium abundance of < (cid:15) ( He ) > rot =11 . ± • Finally, for the main sequence B stars with r star ≤ R eff , we have not found a possible correlation ofthe stellar helium abundance with the distancefrom the cluster center.The results reported in this work need more robustconfirmation and we are working to increase the statis-tics and to minimize uncertainties. In particular, weintend to obtain a larger sample of spectra of NGC330B stars to increase the number of He abundance mea-surements of the 18 Myr stellar population. Moreover,we will also derive the He abundance for all the coreHe-burning stars of both populations. Nevertheless, theresults obtained in this paper show that MUSE@VLTis an extremely powerful instrument able to investigateboth photometric and spectroscopic properties of stellarpopulations in young stellar clusters.ACKNOWLEDGEMENTSThis work has made use of the VALD database, oper-ated at Uppsala University, the Institute of AstronomyRAS in Moscow, and the University of Vienna. KBthanks the Osservatorio Astronomico di Roma for thehospitality during the preparation of the paper. Wethank the anonymous referee for valuable commentsand suggestions that improved the quality of the pub-lication. We thank Antonio Sollima for his preciousfeedback.SCIENTIFIC SOFTWARE PACKAGES
Software: daophot (Stetson 1987), PARSEC (Bressanet al. 2012; Marigo et al. 2017), SYCLIST (Georgy et al.2014), molecfit (Smette et al. 2015; Kausch et al. 2015),SME(Valenti&Piskunov2012),ATLAS12(Kurucz1979)6
Carini et al.
Figure 14.
He abundance versus distance from the center ofNGC 330 for our MUSE sample. Dashed line represents thelinear best-fit to the data. The effective radius of NGC 330is at 20 (cid:48)(cid:48) . REFERENCES
Arp, B. H. 1959, AJ, 64, 254Asplund, M., Gravesse, N., Sauval, A. J., & Scott, P. 2009,ARA&A, 47, 481Bacon, R., Accardo, M., Adjali, L., et al. 2010, inProc. SPIE, Vol. 7735, Ground-based and AirborneInstrumentation for Astronomy III, 773508Bastian, N., & de Mink, S. E. 2009, MNRAS, 398, L11Bastian, N., Lamers, H. J. G. L. M., de Mink, S. E., et al.2013, MNRAS, 436, 2398Bodensteiner, J., Sana, H., Mahy, L., et al. 2019, arXive-prints, arXiv:1911.03477Brandt, T. D., & Huang, C. X. 2015, ApJ, 807, 25Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS,427, 127Brocato, E., & Castellani, V. 1993, ApJ, 410, 99Brunish, W. M., Gallagher, J. S., & Truran, J. W. 1986,AJ, 91, 598Caloi, V., Cassatella, A., Castellani, V., & Walker, A. 1993,A&A, 271, 109Carretta, E. 2015, ApJ, 810, 148Carretta, E., Bragaglia, A., Lucatello, S., et al. 2018, A&A,615, A17Carretta, E., Bragaglia, A., Gratton, R. G., et al. 2009,A&A, 505, 117 Castelli, F., & Kurucz, R. L. 2003, in IAU Symposium, Vol.210, Modelling of Stellar Atmospheres, ed. N. Piskunov,W. W. Weiss, & D. F. Gray, A20Chantereau, W., Salaris, M., Bastian, N., & Martocchia, S.2019, MNRAS, 484, 5236Chiosi, C., Vallenari, A., Bressan, A., Deng, L., & Ortolani,S. 1995, A&A, 293, 710D’Antona, F., & Caloi, V. 2004, ApJ, 611, 871D’Antona, F., Di Criscienzo, M., Decressin, T., et al. 2015,MNRAS, 453, 2637Decressin, T., Charbonnel, C., & Meynet, G. 2007, A&A,475, 859D’Ercole, A., D’Antona, F., & Vesperini, E. 2016, MNRAS,461, 4088D’Ercole, A., Vesperini, E., D’Antona, F., McMillan, S., &Recchi, S. 2008, MNRAS, 391, 825Evans, C. J., Lennon, D. J., Smartt, S. J., & Trundle, C.2006, A&A, 456, 623Feast, M. W. 1991, in IAU Symposium, Vol. 148, TheMagellanic Clouds, ed. R. Haynes & D. Milne, 1Georgy, C., Granada, A., Ekstr¨om, S., et al. 2014, A&A,566, A21Gieles, M., Charbonnel, C., Krause, M. G. H., et al. 2018,MNRAS, 478, 2461