Observational evidence of multiple Stellar Populations in Galactic Globular Clusters Giampaolo Piotto Dipartimento di Astronomia Universita’ di Padova Collaborators: L.R. Bedin, I.R. King, J. Anderson, S. Cassisi, S. Villanova, A. Milone, A. Bellini, Y. Momany, A Renzini, and A. Sarajedini + the HST GC Tresaury Project team

Globular clusters are the ideal laboratory for the study of stellar population and stellar evolution Indeed, normal hydrogen burning stars, in the stellar core or in a shell typically behave as canonical stellar evolution models predict. And we have CMDs which are a clear evidence that globular clusters are typically populated by stars with homogeneous composition and born at the same time.

However, we do have a number of problems which have been there, unsolved, for too many years. For example, we never really understood the general behaviour of He core burning sequences. The classical second parameter problem, i.e. the fact that GCs with the same metallicity have horizontal branches with quite different morphologies still lacks a comprehensive explanation. Ferraro et al. 1997, ApJ, 484, L145

…and some HBs are surely more complicate to understand than others. NGC 2808 Momany et al. (2004) GAPS JUMPS EXTENDED HOT BLUE TAILS May be they are telling us that GC stellar population is not as simple as we thought.

We do have another long standing problem, i.e. the large spread in abundances for some elements, like C,N,O, Na, Mg, Al, s-process elements inside the same cluster (see Gratton et al. 2003, ARAA, for a comprehensive discussion), even in clusters which do not have any dispersion in [Fe/H] and Fe peak elements Some of these abundance spreads are present also at the level of main sequence and subgiant branch stars, which gives strong support to the idea that they could be primordial. NGC2808 (from Carretta et al. 2006, A&A, 450, 523)

RGB stars unevolved stars NGC 2808 (from Carretta et al. 2006) Some of these anomalies have a well defined pattern like the NaO anticorrelation, or the MgAl anticorrelation. Both anticorrelations indicate the presence of proton capture processes, which transform Ne into Na, and Mg into Al. These processes are possible only at temperatures of a few 10 million degrees, in the complete CNO cycle (which implies also an O depletion) not reached in present day globular cluster main sequence and red giant stars. Are the HB anomalies and the chemical anomalies related with each other?

Let’s start with my favourite “special” case: Omega Centauri Most massive Galactic “globular cluster” (present day mass: ~4 million solar masses). Well known (since the ’70s) spread in metallicity among RGB stars. Multiple RGBs Lee et al Pancino et al Multiple MSs Bedin et al. (2004) Extended HB Omega Centauri

The main sequence of Omega Centauri is splitted into two “main” main sequences (Anderson, 1997, PhD thesis, Bedin et al. 2004, ApJ, 605, L125). This is the first direct evidence ever found of multiple stellar populations in globular clusters.

Indeed, also a third main sequence is clearly visible Villanova, Piotto, Anderson et al. (2006), in prep.

The double main sequence in Omega Centauri Piotto et al. (2005, ApJ, 621,777) 17x12=204 hours i.t. RedMS: Rad. Vel.: km/s [Fe/H]=-1.56 BlueMS: Rad. Vel.: km/s [Fe/H]=-1.27 It is more metal rich!

The most surprising discovery (Piotto et al. 2005, ApJ, 621,777) is that the bluest main sequence is less metal poor than the redder one: Apparently, only an overabundance of helium (Y~0.40) can reproduce the observed blue main sequence, as anticipated by Norris (2004), and Bedin et al. (2004)

The strong He overabundance is really puzzling, but confirmed by other observational evidence. E.g., Castellani et al (2007, ApJ, 663, 1021) provide further support to the He enhancement scenario from the comparison of the star counts on the MS, RGB, and HB, and theoretical models Castellani et al. found that only a mix of 70% of canonical He content (Y=0.23) stars plus a 30% of He enhanced (Y=0.33, 0.42) stars can reproduce the observed ratio of RGB/MS stars. The same mixture of canonical and He enhanced stars reduces the discrepancy between the predicted and observed ratio of HB/MS stars, though the observed ratio is still 15-25% higher than expected.

The radial distribution of MS stars: We performed a careful analysis of the radial distribution of the bMS and rMS stars, complementing the work by Sollima et al. (2006, ApJ, 654, 915), using HST and FORS/VLT data (Bellini et al., in preparation).

We find that he ratio of bMS/rMS stars is constant in the inner 6-7 arcmin (~1.5 half mass radii), then it constantly decreases. Castellani et al. (2007) found that the ratio of extremely hot HB (EHB) stars/hot HB stars is constant in the inner 7-8 arcmin, then it decreases. By itself, this may be another observational evidence that bMS stars are related to the EHB stars in ωCen. Is this radial distribution primordial or due to dynamical relaxation? Note that log(t rh )=10 Gyr.

Ferraro et al (2006, ApJ, 638, 433) found no evidence of mass segregation among the blue stragglers in Omega Centauri, at variance with what found in other globular clusters. Is this an evidence that also the center of Omega Centauri is not completely relaxed? Are the results on the distribution of bMS/rMS and hot HB stars contradicting the results on the BSS distribution? Or is the bMS/rMS radial distribution telling us something about the distribution of the material from which the two stellar populations formed?

The multiple population scenario in Omega Centauri is even more complex than what expected from the already puzzling multiple MS. There are at least 4 distinct populations, plus other more spreaded stars (Villanova et al. 2007, ApJ, 663, 296)

Stars at a given metallicity have a large magnitude spread at the level of the SGB (>0.1 magnitudes). This is a clear indication of an age spread. The size of age dispersion depends on the assumption on the metal and He content of the different SGBs. A detailed analysis of the metallicity pattern along the SGB is ongoing. ω Cen is becoming a really challenging object… But is it a unique case?

And indeed we find a large age spread, of a few Gyr, among the five stellar components, but assuming only two He populations. Sollima et al. (2005, Apj, 634, 332) find a smaller age spread (~2Gyr), but assuming for each population a different He content. We need a better knowledge of the metallicity and metallicity distribution of the different SGBs, to properly quantify the age spread. ω Cen is becoming a really challenging object… But is it a unique case?

Accurate HST’s ACS photometry shows that the MS of NGC 2808 splits in three separate branches Overabundances of helium (Y~0.30, Y~0.40) can reproduce the two bluest main sequences. We tentatively attribute the three branches to successive round of star formation with different helium content. The TO-SGB regions are so narrow that any difference in age between the three groups must be significantly smaller than 1 Gyr The triple main sequence in NGC 2808 Piotto et al. 2007, ApJ, 661, L35 TO Data from GO10922, PI Piotto

A clear NaO anticorrelation has been identified by Carretta et al. (2006, A&A, 450, 523) in NGC Besides a bulk of O-normal stars with the typical composition of field halo stars, NGC2808 seems to host two other groups of O-poor and super O-poor stars NGC2808 has a very complex and very extended HB (as ω Cen). The distribution of stars along the HB is multimodal, with at least three significant gaps and four HB groups (Sosin et al 1997, Bedin et al 2000)

D’Antona et al. 2005, ApJ, 631, 868 A MS broadening in NGC2808 was already seen by D’Antona et al. (2005). D’Antona et al. (2005) linked the MS broadening to the HB morphology, and proposed that three stellar populations, with three different He enhancements, could reproduce the complicate HB. We found them in the form of three main sequences!!!

In summary, in NGC 2808, it is tempting to link together: the multiple MS, the multiple HB, and the three oxygen groups, as indicated in the table below. NGC 2808 represents the second, direct evidence of multiple stellar populations in a globular cluster.

And Three! The Double Subgiant Branch of NGC 1851 Accurate HST’s ACS photometry reveals that the SGB of NGC 1851 splits into two well defined branches The split may be due to a large age spread (1 Gyr) or to a combination of abundance anomalies and a much smaller age spread Milone et al. 2008, ApJ, in press, arXiv: M

45% of the stars are in the lower SGB; 37% in the blue HB. 63% 37% NGC1851 is known as a prototype of bimodal HB globular clusters. Are faint (older) SGB stars related to the blue HB? Is NGC 1851 case related to the cases of NGC 2808 and ωCen? 45%

Apparently there is no large He spread among the MS stars. A first quick reduction of new HST data (GO11233, collected at the end of November) sets an upper limit to the He spread in NGC 1851 of Delta Y ~ 0.03 (work in progress)

Very recently, Cassisi et al. (2007, ApJ, 672, 115) showed that the two SGBs and the double HB can be reproduced by assuming that the fainter SGB is populated by a strongly CNNa enhanced population, which evolve into the blue HB, while the brighter SGB contains normal composition stars. The age difference between the two groups may be very small ( years). This idea is supported also by the recent finding by Yong and Grundahl 2007, arXiv: )) of two groups of stars, one with normal composition, and one strongly CN, Na, s-process element enhanced, O depleted. In conclusion, the SGB split may be mainly due to the presence of two groups of stars, with two different metal patterns, small age difference.

…continuing… Multiple Stellar Populations in Globular Clusters. IV. NGC 6388! Piotto et al. (2007, in preparation): GC ACS/HST Tresaury data

NGC 6388 double SGB Present in many, independent databases, and visible in different photometric bands. NGC 6388, as its twin NGC 6441, are two, very peculiar globular clusters.

Since Rich et al. (1997, ApJ, 484, L25) it is known that NGC 6388 and NGC 6441 have an anomalous HB. The HB of these two clusters is very different from the HB of NGC 1851, but similar to the HB of ωCen and NGC 2808 [Fe/H]=-0.6 [Fe/H]=-0.5 The HB is anomalous because of: 1) The blueward extension, with the presence of an EHB; 2) The presence of a tilt Also, the RR Lyr (in NGC6441) tells that the HB is anomalously bright

Canonical HB models (dashed red) cannot reproduce the observed HB (Raimondo, Castellani, V. et al. 2002, A&A, 569, 975) Only a HB model with strong helium enhancement (Y=0.40) can reproduce the the observed HB, at all wavelengths About 13% of the stars have such an extreme He enhancement (Busso et al., 2007, A&A, in press, arXiv: )

In this case, only 8% of the stars have an extreme He (Y=0.38) enhancement Note that this cluster has half the mass of his twin NGC 6388 in which 13% of the stars show extreme He enhancement, And about 1/3 the Mass of Omega Centauri. In the latter, the extreme He populations was ~20% of the MS stars (Busso et al., 2007, A&A, in press, arXiv: )

NGC 6441 Caloi and D’Antona, 2007, A&A, 463, 949 In order to reproduce the anomalous HB, Caloi and D’Antona (2007) propose an even more complicate scenario with 3 distinct populations: 1.a normal population (Y~0.25); 2.a polluted pop. (0.27<Y<0.33); 3.A strongly He enhanced pop. (Y~0.4) Three He populations in NGC 6388 and NGC 6441, as in NGC 2808 and perhaps ωCen?

NGC 6441 Gratton et al. (2007), A&A, 464, 953 Also NGC 6388 shows a NaO and a MgAl anticorrelation. (Carretta et al. 2007, A&A, 464, 967) Interestingly enough, NGC6441 shows clear evidence of the NaO anticorrelation, and the [O/Na] distribution roughly resembles the HB distribution.

four massive globular clusters multiple stellar populations So far, we have identified four massive globular clusters for which we have a direct evidence of multiple stellar populations, and they are all quite different: 1)In Omega Centauri (~4x10 6 solar masses), the different populations manifest themselves both in a MS split (interpreted as a split in He and metallicity abundances) and in a SGB split (interpreted in terms of He, metallicity, and age variations > 1Gyr) which implies at least four different stellar groups within the same cluster. Omega Centauri has also a very extended HB (EHB), as NGC )In NGC 2808 (~1.6x10 6 solar masses), the multiple generation of stars is inferred from the presence of three MSs (also in this case interpreted in terms of three groups of stars with different He content), possibly linked to three stellar groups with different oxygen abundances, and possibly to the multiple HB. Age difference between the 3 groups must be significantly <1 Gyr. It has an EHB. 3)In NGC 6388 (~1.6x10 6 solar masses) we have evidence of two stellar groups from a SGB split (age difference ~1Gyr?). An EHB as in NGC 2808 suggests He enhancement. No information on the MS, yet. NGC6441 may be an analogous case. 4) ) In the case of NGC1851 (~1.0x10 6 solar masses), we have evidence of two stellar groups from the SGB split, which apparently imply two star formation episodes separated by ~1Gyr. No evidence of MS split, yet. Bimodal HB, but no EHB.

The investigation continues. 38 HST orbits allocated in Cycle 16 (GO 11233, PI Piotto) for the Search of multiple MSs Stay tuned……

Relevant exception to the presence of a double MS in massive clusters: 47Tuc. It is at least as massive as NGC6388 and NGC2808 but it does have neither an evident double main sequence nor an anomalously hot HB. (data from HST GO10775)

Ejecta (10-20 km/s) from intermediate mass AGB stars (4-6 solar masses) could produce the observed abundance spread (D’Antona et al (2002, A&A, 395, 69). These ejecta must also be He, Na, CN, Mg) rich, and could explain the NaO and MgAl anticorrelations, the CN anomalies, and the He enhancement. Globular cluster stars with He enhancement could help explaining the anomalous multiple MSs, and the extended horizontal branches. Proposed scenario (1)

Alternative explanation (2) Pollution from fast rotating massive stars (Decressin et al. 2007, A&A, 475, 859). The material ejected in the disk has two important properties: 1)It is rich in CNO cycle products, transported to the surface by the rotational mixing, and therefore it can explain the abundance anomalies; 2)It is released into the circumstellar environment with a very low velocity, and therefore it can be easily retained by the shallow potential well of the globular clusters.

Open problems Both proposed scenarios have a number of problems. Among them: 1)Both scenarios need either an anomalously flat (top- heavy) IMF or to assume that a large fraction of the original cluster population has left the cluster (e.g., because of the evaporation). 2)There are serious dynamical problems: how is the gas retained? How is the gas re-accumulated? How is the second epoch star formation event triggered? 3)Is (part of) the ejected material He-rich enough to explain the strongly He-enhancement populations?

The case of M54

Data from the ACS Tresaury and from GO10922, Piotto et al. (in preparation) Multiple RGB Multiple SGB Multiple MS? But… Who is who?

M54 coincides with the nucleus of the Sagittarius dwarf galaxy It might be born in the nucleus or it might be ended into the nucleus via dynamical friction (see, e.g., Monaco et al. 2005), but the important fact is that, now, M54 is part of the nucleus of a disaggregating galaxy.

M54 Omega Centauri The CMDs of M54 and Omega Centauri are astonishingly similar! It is very likely that M54 and the Sagittarius nucleus show us what Omega Centauri was a few billion years ago: the center of a dwarf galaxy, now disrupted by the Galactic tidal field. Is this true for all globular clusters?

NaO anticorrelation present also il low mass globular clusters (M71: 3x10 4 solar masses)

The interesting case of M4 Mass: 8x10 4 solar masses Strong NaO anticorrelation Two distinct groups of stars Marino et al., in prep. CN strong CN weak Na rich, O poor stars are CN strong Na poor, O-rich stars are CN weak

M4 The two stellar groups are well distinguishable also in the color-magnitude and two color diagrams: Marino et al., in prep. Noticeable: a double generation of stars in a globular cluster with a present day mass ~2% of the mass of Omega Centauri!!!

Apparently no main sequence split. However, it needs further investigation with the refurbished ACS camera or WF3. Note: this CMD is proper motion cleaned and corrected for diff. redd. M4 saturation

Mackey et al. (2007, MNRAS, 379,151) suggested the presence of two populations with an age difference of ~300Myr in the 2Gyr old LMC cluster NGC The presence of two populations is inferred by the presence of two TOs in the color magnitude diagram of the cluster. Are these two populations the consequence of tidal capture of two clusters, or are they showing something related to the multiple MSs identified in Galactic Globular clusters? Multiple generations of stars in LMC clusters was already proposed in the past (see the case of NGC 1850, Vallenari et al. 1994, A&A, 244, 487)

Conclusions These new results on the multiple populations would have never them,we are now looking at been possible without HST. Thanks to them,we are now looking at globular cluster (and cluster in general) stellar populations with new eyes. De facto, a new era on globular cluster research is started: 1)Many serious problems remain unsolved, and we still have a rather incoherent picture. The new HST cameras that will be available after SM4 will play a main role in composing the puzzle. 2)For the first time, we might have the key to solve a number of problems, like the abundance anomalies and possiby the second parameter problem (which have been there as a nightmare for decades), as well as the newly discovered multiple sequences in the CMD. 3)Finally, we should not forget that what we will learn on the origin and on the properties of multiple populations in star clusters has a deep impact on our understanding of the early phases of the photometric and chemical evolution of galaxies (see Meynet et al. 2007).