Extended main sequence turnoff as a common feature of Milky Way open clusters
G. Cordoni, A.P. Milone, A.F. Marino, M. Di Criscienzo, F. D'Antona, A. Dotter, E.P. Lagioia, M. Tailo
DD raft version N ovember
6, 2018Typeset using L A TEX default style in AASTeX62
EXTENDED MAIN SEQUENCE TURNOFF AS A COMMON FEATURE OF MILKY WAY OPEN CLUSTERS
G. C ordoni , A. P. M ilone , A. F. M arino , M. D i C riscienzo , F. D’A ntona , A. D otter , E. P. L agioia , and M. T ailo Dipartimento di Fisica e Astronomia “Galileo Galilei” - Univ. di Padova, Vicolo dell’Osservatorio 3, Padova, IT-35122 Research School of Astronomy & Astrophysics, Australian National University, Canberra, ACT 2611, Australia Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Roma, Via Frascati 33, I-00040 Monteporzio Catone, Roma, Italy Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA (Accepted 31 / / Submitted to ApJABSTRACTWe present photometric analysis of twelve Galactic open clusters and show that the same multiple-populationphenomenon observed in Magellanic Clouds (MCs) is present in nearby open clusters. Nearly all the clustersyounger than ∼ ff s (eMSTOs) and all the clusteryounger than ∼
700 Myr show broadened / split main sequences (MSs). High-resolution spectroscopy has re-vealed that these clusters host stars with a large spread in the observed projected rotations.In addition to rotation, internal age variation is indicated as a possible responsible for the eMSTOs, makingthese systems the possible young counterparts of globular clusters with multiple populations. Recent work hasshown that the eMSTO + broadened MSs are not a peculiarity of MCs clusters. Similar photometric featureshave been discovered in a few Galactic open clusters, challenging the idea that the color-magnitude diagrams(CMDs) of these systems are similar to single isochrones and opening new windows to explore the eMSTOphenomenon. We exploit photometry + proper motions from Gaia DR2 to investigate the CMDs of open clustersyounger than ∼ / or broadened MSs, that cannot bedue neither to field contamination, nor binaries; (ii) split / broadened MSs are observed in clusters younger than ∼
700 Myr, while older objects display only an eMSTO, similarly to MCs clusters; (iii) the eMSTO, if interpretedas a pure age spread, increases with age, following the relation observed in MCs clusters and demonstrating thatrotation is the responsible for this phenomenon.
Keywords: globular clusters: general, stars: population II, stars: abundances, techniques: photometry. INTRODUCTIONIn the past years, work based on high-precision
Hubble Space Telescope ( HST ) photometry discovered that the color-magnitudediagrams (CMDs) of most star clusters younger than ∼ ff s(eMSTOs, e.g. Mackey & Broby Nielsen 2007; Glatt et al. 2008; Milone et al. 2009), and clusters younger than ∼
700 Myr displayboth eMSTOs and split main sequences (MSs, e.g. Milone et al. 2013; 2015; 2018; Li et al. 2017; Correnti et al. 2017).The comparison between the observed and synthetic CMDs from the Geneva database (e.g. Georgy et al. 2014) suggests thatsplit MSs are consistent with two stellar populations with di ff erent rotation rates. A group of stars with rotation close to thebreakout value ( ω ∼ . ω cr ), which corresponds to the red MS and includes about two thirds of the total number of MS stars,and a population of slow rotators with ω ∼
0, which populate the blue MS (e.g. D’Antona et al. 2015; Milone et al. 2016). Onthe turn-o ff region, rapidly and slowly rotating stars distribute on brighter and fainter magnitudes, respectively. Measurementsof rotational velocities in MS stars of the LMC cluster NGC 1818 from high-resolution spectra collected with the Very Large Corresponding author: G. [email protected] a r X i v : . [ a s t r o - ph . S R ] N ov G. C ordoni , et al .Telescope (VLT) has recently provided direct evidence that the red-MS and the blue-MS stars exhibit di ff erent rotation rates(Marino et al. 2018a). Similarly, high-resolution Magellan spectra confirm that the bright and the faint MSTO of NGC 1866 aremostly populated by slow and fast rotators, respectively (Dupree et al. 2017).Although it is now widely accepted that rotation is one of the main driver for the photometric features appearing on the CMDsof young and intermediate-age MC clusters, it might not be able to entirely reproduce the observations. Indeed, as noticed byMilone et al. (2017), a fraction of eMSTO are consistent with being younger than the bulk of cluster stars. It has been suggestedthat some clusters have experienced a prolonged star formation, and that age variation, together with rotation is responsible forthe eMSTOs (e.g. Goudfrooij et al. 2014; 2017). In this case, the MC clusters could represent the younger counterparts of the oldglobular clusters with multiple populations (e.g. Conroy et al. 2011; Keller et al. 2011). As an alternative, D’Antona et al. (2017)suggested that the evolution of braked rapidly-rotating stars can mimic an age spread and contribute to the eMSTO.The recent discovery of eMSTOs in four open clusters, namely NGC 2099, NGC 2360, NGC 2818, and NGC 6705 has chal-lenged the text-book concept that the CMDs of open clusters are proxy of single isochrone and have demonstrated that theeMSTO is not a peculiarity of Magellanic Cloud clusters (Marino et al. 2018b). Spectroscopy of MS stars in NGC 6705 showsthat the blue and the red MS are populated by slow and fast rotators, respectively (Marino et al. 2018b). Similarly, the colorand magnitude of eMSTO stars of NGC 2818 and NGC 6705 are connected with their rotational velocity (Bastian et al. 2018;Marino et al. 2018b). These results suggest that rotation plays an important role in shaping eMSTOs and broadened or split MSsin Galactic open clusters, resembling Magellanic Cloud clusters.In this work we exploit the Gaia data release 2 (DR2, Gaia collaboration et al. 2018) to analyze photometry, parallaxes, andproper motions of a large sample of Galactic open clusters younger than ∼ DATA AND DATA ANALYSISTo unambiguously identify multiple populations along the CMD, if present, we need densely-populated clusters with lowdi ff erential reddening and negligible contamination from field stars. To do this, we selected all the Galactic open clusters of thenew general (NGC), Index (IC), Melotte, and Collinder catalogs that, according to Dias et al. (2002), have E(B − V) < − V) = • We first analyze the vector-point diagram (VPD) of stellar proper motions, and find that NGC 2099 cluster members areclearly clustered around ( µ α cos δ : µ δ ) ∼ (1.9: − µ α cos δ and µ δ ( < µ α cos δ > and < µ δ > ) and to derive the quantity µ R = (cid:112) ( µ α cos δ − < µ α cos δ > ) + ( µ δ − < µ δ > ) ). • We plotted G RP as a function of µ R for the selected stars and divided the analyzed magnitude interval with 10 . < G RP < . µ R ( µ R , med ) and the correspondingrms ( σ ) by rejecting all the stars with µ R > µ R , med + · σ . The mean magnitudes of each bin are associated to the quantities µ R , med + · σ and these points are linearly interpolated to derive the orange line plotted in the upper-left panel of Fig. 1.Stars with deviations from µ R , med larger than 4 · σ are excluded from the sample of probable cluster members. • We plotted G RP as a function of the parallax, π , for the probable cluster members and calculated, for each bin of magnitudedefined above, the median parallax π med and the corresponding rms, σ , by using the same procedure described for propermotions. The orange lines plotted in the central-upper panel of Fig. 1 are derived by adding ± · σ to π med and all the starsthat lie outside these two lines are excluded from the sample of probable cluster members. • The selected stars are used to derive improved estimates of < µ α cos δ > and < µ δ > . This ends one iteration. The procedurerequired three or four iterations to reach the convergence. ultiple populations in G alactic open clusters with G aia Figure 1.
This figure illustrates the procedure that we used to select probable members of NGC 2099. The VPD of proper motions for stars inthe cluster field is plotted in panel a, while panels b and c show G RP against proper motions and parallaxes, respectively. The red lines are usedto separate NGC 2099 members from field stars. The G RP vs. G BP − G RP CMD is illustrated in panel d. Selected cluster members are representedwith red symbols. See text for details.
Photometry of cluster members has been corrected for di ff erential reddening by using the method described by Milone etal. (2012, see their Sect. 1) and illustrated in Fig. 2. Briefly, we first defined the reddening direction by using the absorptioncoe ffi cients in the G BP and G RP bands provided by Casagrande & VandenBerg (2018). Then, we derived the fiducial of MS starsand calculated the color residuals from this fiducial. To estimate the di ff erential reddening su ff ered by each star in the analyzedfield of view, we selected a sample of 35 neighbours formed by bright MS cluster members that are not evident binaries. Ourbest di ff erential-reddening estimate corresponds to the median of the color residuals, calculated along the reddening line. Toderive the corresponding error, we subtracted the median from the residual of each star and calculated the 68.27 th percentile ofthe distribution of the corresponding absolute values ( σ ). We considered the quantity 1.253 · σ/ √
35 as the uncertainty associatedto the di ff erential-reddening.As an example, we compare in Fig. 2 the original CMD of NGC 2099 cluster members (upper-left panel) with the CMDcorrected for di ff erential reddening (upper-right panel). We also plot the di ff erential-reddening map for a circular region withradius of 40 arcmin centered on NGC 2099 (bottom left). The bottom-right panel shows the reddening variation as a function ofthe relative right-ascension distance from the cluster center for stars in six declination intervals. G. C ordoni , et al . Figure 2.
Upper panels.
Comparison of the original CMD of selected cluster members of NGC 2099 (left) with the CMD of the samestars corrected for di ff erential reddening (right). The arrow plotted in the left-panel CMD indicates the reddening vector and corresponds to ∆ E(B − V) = Lower panels.
Map of di ff erential reddening, centered on NGC 2099. The levels of gray correspond to di ff erent E(B − V)values as indicated by the scale on the middle (left). Right panels show E(B − V) as a function of the right-ascension distance from the clustercenter for stars in six slices of declination.
A visual inspection of the di ff erential-reddening corrected CMDs reveals that at least twelve open clusters, namely IC 2714,Melotte 71, NGC 1245, NGC 1817, NGC 2099, NGC 2360, NGC 2818, NGC 3114, NGC 3532, NGC 5822, and NGC 6705,clearly exhibit multiple sequences in their CMDs. Their CMDs are presented and analyzed in the next section. MULTIPLE POPULATIONS ALONG THE COLOR-MAGNITUDE DIAGRAMSThe final CMDs, corrected for di ff erential reddening, of the selected cluster members are plotted in Figs. 3– 4, where wealso represent with red error bars the typical observational uncertainties for stars with di ff erent luminosities. A visual in-spection of these figures clearly reveals that IC 2714, Melotte 71, NGC 1245, NGC 1817, NGC 2099, NGC 2360, NGC 2818,NGC 3114, NGC 3532, NGC 5822, and NGC 6705 exhibit the eMSTO. Noticeably, the upper MS of NGC 2099, NGC 2287(M 41), NGC 3114, NGC 3532, and NGC 6705 is broadened, in contrast with the faint MS, which is narrow and well defined.Similarly to what previously observed in MCs clusters, the broadened MS seems to disappear at the luminosity of the MS kinkat T e ff ∼ ultiple populations in G alactic open clusters with G aia Field stars
The CMDs shown in Figs. 3– 4 are mostly populated by cluster members that have been selected on the basis of their parallaxesand proper motions as described in Sect. 2. To estimate the contamination from those field stars that have proper motions anddistances similar to those of cluster members, we applied the procedure illustrated in Fig. 5 for NGC 2099.All the stars plotted in Fig. 5 are located in the “reference field”, which is a circular annulus with the same area as thecluster field, centered on the cluster and with internal radius corresponding to three times the cluster radius provided by Dias etal. (2002). The VPD of proper motions for stars in the reference field is plotted in Fig. 5a; panels b and c show the G RP magnitudeas a function of parallax and proper motions, respectively. We plot in each panel the orange lines derived in Fig. 1 that are nowused to select field stars with cluster-like proper motions and parallaxes, in close analogy with what we did for candidate clustermembers. The stars with cluster-like proper motions have been selected according to their position in the diagrams plotted inpanels b and c and are represented with aqua crosses in all the panels of 5.To statistically subtract the selected field stars from the cluster-field CMD we adopted the same procedure used in our previouspapers (e.g. Milone et al. 2009). In a nutshell, we calculated for each selected star (i) in the reference field a distance in the CMD d i = (cid:113) k (( G iBP , rf − G iRP , rf ) − ( G BP , cf − G RP , cf )) + ( G iRP , rf − G RP , cf ) where G BP , rf(cf) and G RP , rf(cf) are the magnitudes of the selected stars in the reference (cluster) field, and k = Binaries
Unresolved binaries formed by pairs of MS stars are redder and brighter than single MS stars with similar masses while binariesformed by a MSTO star and a MS or a MSTO star are brighter than the corresponding single MSTO stars. In the following, wemeasure the fraction of MS-MS binary systems of each cluster to estimate the contribution of binaries to the eMSTO and thebroadened MS.To estimate the fraction of unresolved binaries with q > G brightRP and G faintRP , that delimit the MS region where the high-mass binaries are clearly separated fromthe remaining MS stars and there is no evidence for broadened or split sequences. We then defined two regions in the CMD,namely A and B, that correspond to the gray shaded areas in the CMDs of Fig. 6: region A includes all the single stars with G brightRP < G RP < G faintRP and all the binaries with a primary component in the same magnitude interval; region B is the sub-regionof A that is populated by binaries with q > = f q > . = N Bcl − N Bfi N Acl − N Afi − N Bsim N Asim (1)where N A , (B)cl is the number of cluster members in the region A (B) of the CMD, N A , (B)fi and N A , (B)sim are the corresponding numbersof field stars with cluster-like proper motions and parallaxes and the number of simulated stars, respectively.The measured fraction of binaries with q > . f q > . is used to extrapolate the total fraction of binaries, f TOTbin . Specifically,by assuming a flat mass-ratio distribution, as observed among binaries with q > . f TOTbin ∼ . f q > . . The total fraction binaries is typically around 0.30 and ranges from ∼ ∼ ordoni , et al . Figure 3. G RP vs. G BP − G RP CMDs, corrected for di ff erential reddening, of cluster members for IC 2714, Melotte 71, NGC 1245, NGC 1817,NGC 2099, and NGC 2287. The insets highlight the eMSTO or the broadened MS. Red bars represent typical observational uncertainties. ultiple populations in G alactic open clusters with G aia Figure 4.
As in fig. 3 but for NGC 2360, NGC 2818, NGC 3114, NGC 3532, NGC 5822, and NGC 6705.
G. C ordoni , et al . Figure 5.
This figure illustrates the procedure that we used to identify field stars with similar proper motions and parallaxes as NGC 2099cluster members. The VPD of proper motions of stars in the “reference field” is plotted in panel a, while panels b and c show G RP as a functionof stellar proper motions and parallaxes, respectively. The orange lines defined in Fig. 1 are overimposed to the diagrams of panels b and c. The G RP vs. G BP − G RP CMD is illustrated in panel d. Selected field stars are marked with aqua crosses. See text for details. confirm previous findings that open clusters typically host larger binary fraction than Galactic Globular Clusters (e.g. Sollima etal. 2010). 3.3.
Simulated CMDs
The obtained total binary fractions, listed in Table 1, are used to simulate the CMD of a simple stellar population with thesame observational errors, age, metallicity, distance modulus and reddening as inferred from the observations. To do this, wefirst associated to each star in the observed CMD of cluster member a synthetic star with the same magnitude and the color ofthe fiducial line. We selected a fraction of single stars equal to f TOTbin and estimated the mass M of each of them by using themass-luminosity relation by Marigo et al. (2017). We associated to each selected star a secondary star with a mass M = q · M and derived its G RP magnitude from the relations by Marigo and collaborators. The corresponding color has been inferred fromthe fiducial line. Finally, we summed up the G BP and G RP fluxes of the two components, derived the corresponding magnitudesreplaced the original star in the CMD with this binary system, and summed up the observational errors to all the stars of theCMD.Results are illustrated in Figs. 7– 9, where we compare for each cluster the CMD of selected cluster members (first column ofpanels), the CMD with cluster-like proper motions and parallaxes of stars in the reference field (second column), the decontami- ultiple populations in G alactic open clusters with G aia Figure 6. G BP vs. G BP − G RP CMD of selected NGC 2099 cluster members in the cluster field (left panel) and CMD of stars with cluster-likeproper motions and parallaxes in the reference field (middle panel). Right panel shows the simulated CMD. The shaded areas indicate theregion A of the CMD, which is populated single MS stars and by MS-MS binary pairs with a primary component in the mass interval between1.06 and 1.63 solar masses. The blue lines represent the fiducial lines of binaries with mass ratio, q = .
7. The region B of the CMD, which ispopulated by binaries with q ≥ . nated CMD (third column), and the simulated CMDs. A visual inspection of these figures clearly demonstrates that the eMSTOsand the broadened MSs are not due neither to unresolved binaries nor to residual field-star contamination. COMPARISON WITH THEORYThe eMSTOs of Magellanic-Cloud clusters has been interpreted either as the signature of stellar populations with di ff erent ages(e.g. Mackey et al. 2008; Goudfrooij et al. 2011) or as the e ff ect of stellar rotation on a single stellar population (e.g. Bastian &De Mink 2009; Yang et al. 2013; D’Antona et al. 2016; Marino et al. 2018). To disentangle between these two possibilities, inthis section we compare the observed CMDs with isochrones of di ff erent ages and with simulated CMDs of stars with di ff erentrotation rates.To estimate the age spread, in the hypothesis that the eMSTO is entirely due to a prolonged star formation, we comparedthe observed CMDs with isochrones by Marigo et al. (2017). The procedure exploited to derive accurate age distributions isillustrated in Fig. 10 for NGC 2099, and is similar to what we have used in previous work (e.g. Milone et al. 2015).In a nutshell, we first derived by hand the parallelepiped plotted in Fig. 10 with the criterion of selecting the region around theturn o ff where the color and magnitude spread due to age variation are clearly distinguishable. Only stars within the parallelepipedare used to infer the age distribution. Then, we overimposed on the CMD a grid of isochrones with the same metallicity and[ α/ Fe ] and ages between 380 and 700 Myr in steps of 10 Myr (grey lines in Fig. 10) and derived isochrones separated by 1 Myrby linearly interpolating among these isochrones. We associated to each star the age of the closest isochrone and derived the agedistribution shown in the right panel of Fig. 10. Finally, we calculated the median age and the absolute value of the di ff erencebetween the age of each star and the median. We considered the 68.27 th percentile of the distribution of these absolute values asindicative of the observed age spread, σ AGE , obs . To estimate the contribution of observational errors on the inferred age spread,we applied the procedure described above to the simulated CMD of a simple population and derived the corresponding age0 G. C ordoni , et al . G BP G RP G B P IC2714
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P MELOTTE 71
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC1245
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC1817
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED
Figure 7.
From the left to the right. CMD of the selected cluster members in the cluster field for IC 2714, Melotte 71, NGC 1245, and NGC 1817(first column), CMD of reference-field stars with cluster-like parallaxes and proper motions (second column), CMD of cluster members afterthe statistical subtraction of field stars with cluster-like parallaxes and proper motions (third column). Simulated CMD (forth column). spread, σ AGE , sim . The intrinsic age spread is estimated as σ AGE = (cid:113) σ , obs − σ , sim . Uncertainties on σ AGE are derived bybootstrapping with replacements performed 1,000 times on both the observed and the simulated age distributions.Our results are summarized in Table 1 where we provide the full width half maximum of the age distribution, FWHM = . · σ AGE , for each cluster. We find that the FWHM ranges from ∼
70 for NGC 3114 to ∼
260 Myr for NGC 5822 and correlates with ultiple populations in G alactic open clusters with G aia G BP G RP G B P NGC2099
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC2287
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC2360
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC2818
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED
Figure 8.
As in Fig. 7 but for NGC 2099, NGC 2287, NGC 2360 and NGC 2818. the cluster age as shown in Fig. 11, with old clusters having, on average, larger age spread than younger clusters. A similar trendbetween the age spread inferred from the eMSTO and the cluster age is also present among Magellanic Cloud clusters and isinterpreted as the signature of stellar rotation. Indeed, since rotating stars have longer MS lifetime than non-rotating stars withthe same age and mass, they would appear younger than coeval non-rotating stars within the same cluster. In this case, if theresulting eMSTO is interpreted as an age spread, the resulting age spread would correlate with the cluster age (Niederhofer et2 G. C ordoni , et al . G BP G RP G B P NGC3114
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC3532
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC5822
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED G BP G RP G B P NGC6705
CLUSTER FIELD G BP G RP REFERENCE FIELD G BP G RP DECONTAMINATED G BP G RP SIMULATED
Figure 9.
As in Fig. 7 but for NGC 3114, NGC 3532, NGC 5822 and NGC 6705. al. 2015; Bastian et al. 2018). On the other hand, in the case of a true age spread we would expect that the amount of age spreaddoes not depend on cluster age, and therefore a correlation would be very unlikely.To further investigate the e ff ect of rotation on the observed CMDs, we extended the method by Niederhofer and collaborators toGalactic open clusters, and compared the observations with simulated CMDs of coeval stellar populations with di ff erent rotationrates based on stellar models from the Geneva database with Z = ultiple populations in G alactic open clusters with G aia Figure 10.
Left Panel.
Grid of isochrones from Marigo et al. (2017) overimposed on the CMD of NGC 2099. The two isochrones representedwith dark-tick lines have ages of 380 and 700 Myr, while the thin isochrones are spaced by 10 Myr in age.
Right panel.
Histogram agedistribution of the eMSTO stars plotted with red crosses in the left-panel CMD. The median age of these stars is marked with a verticalcontinuous line, while the two dashed lines have distances of ± σ from the median value. isochrones, and for the isochrones with rotation equal to 0.9 times the breakout value ( ω = . ω cr ). These data account for thelimb-darkening e ff ect as in Claret (2000), adopt the gravity-darkening model by Espinosa Lara & Rieutord (2011) and assumerandom distribution for the viewing angle. We transformed the synthetic photometry into the observational plane by adopting themodel atmospheres by Castelli & Kurucz (2003) and the transmission curves of the G BP and G RP filters of Gaia. We assumed thatone third of stars in the simulated CMD do not rotate, while two thirds of stars have ω = . ω cr , in close analogy with what isobserved in Magellanic Clouds open clusters (e.g. Milone et al. 2018).We first applied the procedure above to each synthetic CMD, by assuming that the eMSTO is due to age spread, and derivedthe FWHM of the age distribution. Results are represented with grey dots in Fig. 11. As expected, the age spread increases withthe cluster age, in close analogy with what was previously found by Niederhofer et al. (2015) in Magellanic Clouds clusters. Thefact that the FWHM values derived for synthetic CMDs and for Galactic open clusters follow similar trends against the clusterage suggests that rotation is the main responsible for the observed eMSTOs.Finally, we compare in Fig. 12- 13 the CMDs of cluster members (left panels) with simulated CMDs (right panels). SyntheticCMDs are derived from the Geneva database (Georgy et al. 2014) and have metallicity, Z = ∼ M (cid:12) . e note that, while in young clusters like NGC 2287 and NGC 2099 both fast rotators and slow-rotator stars are neededto reproduce the broad MS, the eMSTO of old clusters seems consistent with fast rotators alone. The poor quality of the fitcould be due to the modeling of several second-order parameters that characterize the end of the core hydrogen burning phase,including the parametrization of the inclination angle, which strongly a ff ects the stellar luminosity and e ff ective temperature (seeDAntona et al. 2015 for details). Nevertheless, the comparison between data and simulations corroborates the conclusion thatstellar rotation is the main responsible for the observed eMSTOs and the broadened MSs. SUMMARY AND DISCUSSION4 G. C ordoni , et al . F W H M [] Figure 11.
Full width half maximum of the age distribution as a function of cluster age. Blue dots with error bars refer to the analyzed clusters.Grey dots are derived from synthetic CMDs of coeval stellar population with di ff erent rotation rates. The dashed line is the least-squares best-fitstraight line for the gray dots. See text for details. We have presented the first analysis of twelve open clusters in the Milky Way in the context of multiple stellar populations.Our results suggest that the multiple photometric sequences observed by Gaia in the CMDs of these nearby objects belong to thesame phenomenon present in Magellanic Clouds clusters, and interpreted as due to stellar rotation and / or age spreads.Since the early discoveries, the eMSTOs have been considered a common feature of the CMDs of LMC and SMC clustersyounger than ∼ ∼ ∼ ff erential reddening. We find that all the analyzed clusters show the eMSTO. In addition, all the clustersyounger than ∼
700 Myr exhibit a broadened upper MS, whereas the bottom of the MS is narrow and well defined. The appearanceof certain photometric features depending on age, is similar to that observed in Magellanic Clouds clusters.We statistically subtracted field stars with cluster-like proper motions and parallaxes from the CMD of candidate cluster starsthus demonstrating that eMSTOs and broadened MSs are not due to residual contamination from field stars. We calculatedfor each cluster the synthetic photometry of a simple population of stars with the same age, metallicity, binary fraction, andobservational errors. The comparison between the observations and the simulated CMDs reveals that the eMSTOs and thebroadened MSs are due neither to observational uncertainties nor to unresolved binaries. These facts demonstrate that eMSTOsand broadened MSs are intrinsic features of the CMDs of the analyzed open clusters.To investigate the physical mechanisms that are responsible for the eMSTO we first compared the CMDs of cluster memberswith isochrones with di ff erent ages. The eMSTOs of the analyzed open clusters are consistent with stellar populations withdi ff erent ages in close analogy with what has been observed in Magellanic Cloud clusters with similar ages. The FWHM ofthe age spread ranges from about 70 Myr in the ∼ ∼
260 Myr in ∼ . ultiple populations in G alactic open clusters with G aia G BP G RP G B P IC2714
CLUSTER FIELD G BP G RP SIMULATED =0=0.9 cr G BP G RP G B P MELOTTE71
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC1245
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC1817
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC2099
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC2287
CLUSTER FIELD G BP G RP SIMULATED
Figure 12.
Left panels. reproduction of the observed CMDs of cluster members of IC 2714, Melotte 71, NGC 1245, NGC 1817, NGC 2099 andNGC 2287.
Right panels.
Comparison of the observed CMDs plotted in the left panels (grey dots) and simulated CMDs of a non rotating stellarpopulation (blue) and of a stellar population with rotation ω = . ω cr (red). We compared the CMDs of cluster members with synthetic diagrams derived from Geneva models and find that the eMSTOsand the broadened MSs are consistent with coeval stellar populations with di ff erent rotation rates. These findings suggest thatrotation is the main responsible for the eMSTOs and the broadened MSs observed in Galactic clusters and corroborate directspectroscopic evidence that stars with di ff erent rotation rates populate the eMSTOs of NGC 6705 and NGC 2818 (Marino etal. 2018b; Bastian et al. 2018) and that the blue and the red MS of NGC 6705 are populated by slow rotators and fast rotators,respectively (Marino et al. 2018b).Our investigation of 12 Galactic open clusters demonstrates that the eMSTO and the broadened MS are not a peculiarity ofMagellanic Cloud star clusters but are common features of Galactic open clusters. Coeval stellar populations with di ff erentrotation rates are likely the responsible for the eMSTO and the broadened MS of the analyzed clusters.ACKNOWLEDGMENTS We thank the anonymous referee for several comments that improved the quality of this manuscript. This work has received funding fromthe European Research Council (ERC) under the European Union’s Horizon 2020 research innovation programme (Grant Agreement ERC-StG 2016, No 716082 ’GALFOR’, PI: Milone), and the European Union’s Horizon 2020 research and innovation programme under the MarieSkłodowska-Curie (Grant Agreement No 797100, beneficiary: Marino). APM acknowledges support from MIUR through the the FARE project ordoni , et al . G BP G RP G B P NGC2360
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC2818
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC3114
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC3532
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC5822
CLUSTER FIELD G BP G RP SIMULATED G BP G RP G B P NGC6705
CLUSTER FIELD G BP G RP SIMULATED
Figure 13.
As in Fig 12 but for NGC 2360, NGC 2818, NGC 3114, NGC 3532, NGC 5822 and NGC 6705Cluster ( m − M ) E(B − V) Z Age [Myr] FWHM [Myr] f q > . f totbin IC 2714 10.60 0.38 0.0205 540 134 ±
57 0.105 0.350MELOTTE 71 11.60 0.22 0.0095 1220 165 ±
38 0.085 0.283NGC 1245 12.45 0.29 0.0183 1000 139 ±
21 0.119 0.397NGC 1817 11.00 0.26 0.0100 1030 165 ±
47 0.083 0.277NGC 2099 10.90 0.26 0.0300 580 125 ±
21 0.085 0.283NGC 2287 9.40 0.04 0.0219 280 76 ±
28 0.034 0.113NGC 2360 10.23 0.16 0.0140 1020 210 ±
35 0.087 0.290NGC 2818 12.35 0.22 0.0100 1110 160 ±
59 0.088 0.293NGC 3114 10.05 0.12 0.0209 180 75 ±
31 0.068 0.227NGC 3532 8.30 0.06 0.0160 430 140 ±
50 0.074 0.247NGC 5822 9.40 0.11 0.0170 1130 270 ±
52 0.131 0.437NGC 6705 11.10 0.46 0.0083 570 245 ±
71 0.153 0.510
Table 1.
Distance modulus, reddening, age, FWHM of the age distribution, fraction of binaries with mass-ratio, q > ultiple populations in G alactic open clusters with G aia R164RM93XW SEMPLICE; AFM has been supported by the Australian Research Council through Discovery Early Career Researcher AwardDE160100851.