Multiple Populations in Globular Clusters

Globular clusters have been important laboratories for stellar evolution because their single stellar populations represent a model of coeval generation with mass alone as a variable, distributed according to a single IMF. Convective mixing, nucleosynthesis and mass loss are able to be studied as a function of mass. Mod- ern studies have shown that this is only a first approximation, and that detailed spectroscopic and photometric examination has revealed multiple generations of stars within GCs (Gratton et al. 2012, and references therein). The first clus- ters to reveal multiple populations were massive GCs such as 47 Tucanae (e.g. Cannon et al. 1998; Lane et al. 2010), ω Centauri (Bedin et al. 2004; Sollima et al. 2007), and M54 (Carretta et al. 2010). Many GCs have now been shown exhibit more than a single population (Milone et al. 2013).

1.4. GLOBULAR CLUSTERS IN THE MILKY WAY HALO

Liight element Li, C, N abundance variations in GCs have been known for a long time (Kraft 1979; Hesser and Harris 1979; Da Costa 1997). Abundance variations have been seen seen in evolved RGB and un-evolved MS stars (Can- non et al. 1998; Gratton et al. 2001). MS stars do not have the convective mechanisms to bring advanced nucleosynthesis products from the core to the surface. This points to fundamental differences in populations that are hard to ascribe to factors such as self-enrichment, fast rotating stars, poor mixing of primordial gas or cluster merging. While there are valid hypotheses con- cerning the role of peculiar environmental effects and self-pollution (Bekki and Chiba 2007a), the consensus is moving toward multiple generations being the explanation for imhomogeneities (Milone et al. 2013).

While light element variations are common, heavy iron peak element varia- tions are not seen except in massive cluster environments (Cohen 1981; Joo and Lee 2013b). Heavy element Ca and Si enhancements seen in all cluster stars, show that supernovae Type II (SNe II) have been the dominant polluters in the GC environment (Gratton et al. 2004). Massive clusters like M4 (Marino et al. 2008), M22 (Marino et al. 2013) show clear evidence of multiple populations and indicate higher star forming efficiency goes with higher mass.. MW GCs have a single [Fe/H] except for ω Centauri (Sollima et al. 2007) which leads some to think this extremely massive cluster may contain accreted populations or be the remnant core of a dwarf galaxy (Bekki and Norris 2006).

Examples of triple MS have been found within the narrow colour spread MS of NGC 2808 MS (Piotto et al. 2007) and NGC 6752 (Milone et al. 2013). Multiple populations were first noticed in massive clusters but now it appears true for most GCs (Gratton et al. 2012). Those in the Magellanic system show the some extreme multiple populations with several MS TO (Glatt et al. 2008; Milone et al. 2009), where a gas rich environment and tidal interactions may have triggered several episodes of star formation.

Gratton et al. (2012) propose three generations to explain light element abundances within a typical GC; an extreme primordial precursor generation in the very early universe which are no longer extant; a first generation of “polluters” which enriched the cluster to the present day level, of which a per- centage are still visible today; and thirdly a second generation of Population II stars which now form the bulk of stars in GCs at typical low metallicity. The extreme progenitor Population III were massive and quickly pre-enriched the cluster molecular cloud with SNe II iron-peak and αelements, raising the metallicity to the present level (Bekki and Chiba 2007b). Then the first gen- eration “polluters” gave rise to the light element abundance anomalies in the second generation. The first generation represent typically around 30% of the GC populations today (Carretta et al. 2010).

The first generation is proposed to have been more massive and burnt H at higher temperatures in order to generate observed light element abundances through proton capture processes (Denisenkov and Denisenkova 1989). The Na-O and especially the Mg-Al anti-correlation seen in many clusters require higher temperature than achieved by the present population, implying the first generation polluters were more massive, but not so massive as to create heavy

elements (Carretta et al. 2010). This means either the IMF was more top heavy for the first generation or the GCs were more massive (Prantzos and Charbonnel 2006), which would imply that GCs have been “evaporating”, losing stars to the tides of the MW. Carretta et al. (2010) estimate clusters must have been twenty times larger than at present. Evaporation has removed most of the “polluter” generation, while the Population II stars reside in what was once the dense core of the cluster. Mackey and van den Bergh (2005) estimate the present GC population may represent only two-thirds of the original.

The metal-poor halo stars may be the evaporated remnant “polluters” of the once extremely massive GCs (Helmi 2008). The gas as well as the first generation stars evaporated into the halo ending star formation within GCs. The question of dark matter and the “missing satellites” is still a problem for ΛCDM models (Klypin et al. 1999; Moore 1996; Moore et al. 2006). More massive early GCs could represent at least part of the the “missing satellites”. These multiple generations within GCs occurred within a short cosmological time in the early universe. Detailed archaeology of multiple star formation episodes in GCs will reveal important clues as to the role of effects like stellar feedback and mass loss in suppressing the formation of more satellite galaxies in dark matter haloes.

The Na-O anti-correlation is not seen in galactic or halo field stars (Gratton et al. 2012). Only stars in GCs exhibit the O-Na anti-correlation abundance anomaly (Gratton et al. 2001) indicating some peculiar GC environmental effect which remains unexplained. The Na-O abundance signature is proposed as a definition of GC populations by Carretta et al. (2010).

IC 4499, one subject of this thesis, does not show photometric colour spreads that would multiple generations (Sarajedini 1993; Ferraro et al. 1995; Walker and Nemec 1996; Walker et al. 2011). A high resolution spectroscopic study that might identify abundance anomalies in IC 4499 has not yet been undertaken.