The role of super-asymptotic giant branch ejecta in the abundance patterns of multiple populations in globular clusters

Star formation from matter including the hot CNO processed ejecta of asymptotic giant branch (AGB) winds is regarded as a plausible scenario to account for the chemical composition of a stellar second generation (SG) in globular clusters. The chemical evolution models, based on this hypothesis, so far included only the yields available for the massive AGB stars, while the possible role of super-AGB ejecta was either extrapolated or not considered. In this work, we explore in detail the role of super-AGB ejecta on the formation of the SG abundance patterns using yields recently calculated by Ventura and D'Antona. An application of the model to clusters showing extended Na-O anticorrelations, like NGC 2808, indicates that an SG formation history similar to that outlined in our previous work is required: formation of an extreme population with very large helium content from the pure ejecta of super-AGB stars, followed by formation of an intermediate population by dilution of massive AGB ejecta with pristine gas. The present models are able to account for the very O-poor, Na-rich extreme stars once deep-mixing is assumed in SG giants forming in a gas with helium abundance Y > 0.34, which significantly reduces the atmospheric oxygen content, while preserving the sodium abundance. On the other hand, for clusters showing a mild O-Na anticorrelation, like M 4, the use of the new yields broadens the range of SG formation routes leading to abundance patterns consistent with observations. Specifically, our study shows that a model in which SG stars form only from super-AGB ejecta promptly diluted with pristine gas can reproduce the observed patterns. We briefly discuss the variety of (small) helium variations occurring in this model and its relevance for the horizontal branch morphology. In some of these models, the duration of the SG formation episode can be as short as ∼10 Myr; the formation time of the SG is therefore compatible with the survival of a cooling flow in the cluster core, previous to the explosion of the SG core collapse supernovae. We also explore models characterized by the formation of multiple populations in individual bursts, each lasting no longer than ∼10 Myr each.