In recent years there has been an increasing interest on the possibility of a second generation of stars in some globular clusters. Some of the observed properties of globular clusters are indeed considered evidence of a second SF event. For instance, the anticorrelation between the sodium and oxygen abundances measured in stars of several clusters (e.g. [Carretta, Bragaglia, & Cacciari (2004)] and references therein) has been interpreted as the consequence of the cluster self-enrichment, if the gas replenishment from the retention of stellar ejecta is sufficient to allow for further SF. In this scenario, the stars born during the second SF episode form from mostly (if not entirely) recycled gas and their initial chemical compositions reflect the yields of the stars mostly contributing to the cluster self-enrichment. [Gratton et al. (2001)] and [D'Antona et al. (2002)] suggested that high Na enrichment and significant O depletion are most naturally explained if the major culprits of the cluster self-pollution are relatively massive AGB stars. The stars with more O and less Na would be those formed in the first generation, while the stars with less O and more Na would be those formed in the second generation. [D'Antona et al. (2002)] further suggested that the second generation would be significantly enriched also in helium (again as a consequence of the predominance of AGB star ejecta in the gas available for SF) and that this could explain the Horizontal Branch morphology of clusters with extreme blue tails. Other authors, however, have argued that the winds of massive stars have more chances than AGB stars to adequately pollute the cluster's medium without the side effects of requiring unusual initial mass functions and stellar remnants (e.g. [Prantzos & Charbonnel (2006)]). Appropriate chemo-dynamical models are required to test advantages and disadvantages of the various hypotheses.
The discovery of a second, bluer Main Sequence (MS) in Centauri [Bedin et al. (2004)], together with that of multiple subgiant and red giant branch (RGB) sequences, also calls for the existence of multiple stellar generations, with the additional striking surprise, provided by high-resolution spectroscopy, that the bluer MS is 0.3 dex less metal poor than the standard red MS [Piotto et al. (2005)]. Piotto et al. argue that the only way to allow for the measured colour shift between the two MSs with the measured metallicity difference is to let the bluer MS be much more helium rich than the other, with a difference in the helium mass fraction Y = 0.14. If we attribute to the red MS a primordial He mass fraction of Y = 0.24, this implies that the blue MS should have Y = 0.38: an abundance higher than in any other observed star cluster or galaxy !
Would AGB stars be able to provide such a huge helium enhancement ? It is very unlikely. However, Centauri is definitely not a normal globular cluster; may be not a cluster at all, but the remnant of a nucleated dwarf galaxy captured and stripped by the Milky Way, with the current cluster actually being the original nucleus of the satellite. [Bekki & Norris (2006)] have recently shown that the observed properties of Centauri cannot be explained without considering the strong dynamical interactions with the Galaxy. The actual question then is whether the medium, out of which subsequent stellar generations have formed, was enriched by Centauri own stars or by the stars of the host dwarf originally surrounding it. Whether the polluters were AGBs, massive stars or supernovae is in this case a second level issue.
We are currently modelling the chemical evolution of Centauri (Romano et al., in preparation) considering it as a dwarf galaxy and following the approach presented by [Romano, Tosi, & Matteucci (2006)] (hereafter, RTM06) for other dwarfs. We adopt the SF history derived from the [Sollima et al. (2005)] data and allow for both galactic winds and infall of metal poor gas. The model predictions are compared with the observational mass, age-metallicity relation, metallicity distributions, chemical abundances and abundance ratios. In agreement with Bekki & Norris results, in no way are we able to obtain model predictions consistent with the data if we consider Centauri as an isolated system. On the other hand, by considering it as the residual of a nucleated galaxy stripped by the Milky Way 10 Gyr ago, we reproduce rather well all the data, except the extremely high He abundance of the blue MS. To achieve this goal, new ad hoc assumptions are needed and will be the subject of further efforts, involving appropriate assumptions on the fate of the stellar ejecta and new stellar yields for both high and intermediate mass stars. Our current, preliminary results are shown in Fig. 1.
Figure 1. Predictions from chemical evolution models (solid lines) for Centauri compared with the corresponding observational data (dots and dashed lines). Left panel: the stellar metallicity distribution from [Sollima et al. (2005)]; central top panel: age-metallicity relation from [Hilker et al. (2004)]; central bottom panel: helium vs iron from [Piotto et al. (2005)], box and in RR Lyraes (dots; from [Sollima et al. (2006)]); right panel: abundance ratios vs iron from several literature sources.