Abstract. This paper presents a review of the characteristics of the multiple stellar populations observed in globular clusters, and of their possible origin. The current theoretical issues and the many open questions are discussed.
Importance of globular clusters for cosmology and for galactic physicsGlobular clusters (GC) are magnificent astronomical objects that provide insight to a broad variety of astronomical and cosmological questions. These compact systems that contain between several hundreds of thousands and a million stars are among the oldest structures in the Universe. They are unique relics, witnesses, and clocks of the formation, assembly, dynamics, and evolution of galaxies and of their substructures, from the early to the present-day Universe. They play a key role in hierarchical cosmology, and potentially also in the reionization of the Universe, and they provide fundamental benchmarks for stellar evolution theory. However, their formation, evolution, and survival in a galactic and cosmological context are far from being understood.Determining the age of the oldest GCs through isochrone fitting of their observed color-magnitude diagram (CMD) was crucial for cosmology until the beginning of the 21 st century, as it provided the best lower limit to the age of the Universe (e.g. Carney, 2001, and references therein). Meanwhile, a robust determination of the age of the Universe was made possible thanks to precise observations of the cosmic microwave background (CMB) by the Wilkinson Microwave Anisotropy Probe and Planck satellites. The age of the universe is now estimated (EES 2015) to be ∼ 13.8 according to flat Λ CMD model that accurately reproduces the observed temperature of the power spectrum of the CMB (Planck Collaboration, Ade et al., 2015). This agrees with the current estimate of the age of the oldest Galactic GCs that were among the first baryonic structures to form in galaxies about 13 Gyr ago (e.g. VandenBerg et al., 2013).Determining the absolute and relative ages of GCs is still very important today, for various reasons. This helps constraining the models of the formation of these systems in the more general context of the formation and early evolution of galaxies (e.g. Fall and Rees, 1985;Harris, 1991;Forbes et al., 1997; van den Bergh, 2011). Different scenarios make indeed different predictions for the age of the oldest GCs, as well as for the dispersion in age at a given metallicity (or rather [Fe/H]) and/or galactocentric distance. There is a wealth of literature on these questions, and we quote here only a few illustrative examples. Brodie and Strader (2006) argue that the most metal-poor GCs, which inhabit the haloes of spiral galaxies, were created in low-mass dark matter haloes at redshift higher than 10 (see also Bromm and Clarke 2002), and that metal-rich GCs formed during later mergers of gas-rich structures that build up the parent galaxies. On the other hand, cosmological N-body simulations and semi-analytic models of galaxy and GC formation accounting for the impo...