Multiple Populations in Globular Clusters: The Possible Contributions of Stellar Collisions

TL;DR

This paper investigates how stellar collisions and other phenomena in dense globular clusters contribute to the formation of multiple stellar populations, finding that collisions alone cannot fully explain observed abundance variations.

Contribution

It combines models of stellar collisions, rotation, and AGB stars to assess their roles in creating multiple populations in globular clusters.

Findings

Runaway collisions produce some observed abundance patterns.

Collisions contribute to extreme abundances but are insufficient in mass.

Additional effects are likely needed to fully explain multiple populations.

Abstract

Globular clusters were thought to be simple stellar populations, but recent photometric and spectroscopic evidence suggests that the clusters' early formation history was more complicated. In particular, clusters show star-to-star abundance variations, and multiple sequences in their colour-magnitude diagrams. These effects seem to be restricted to globular clusters, and are not found in open clusters or the field. In this paper, we combine the two competing models for these multiple populations and include a consideration of the effects of stellar collisions. Collisions are one of the few phenomena which occur solely in dense stellar environments like (proto-)globular clusters. We find that runaway collisions between massive stars can produce material which has abundances comparable to the observed second generations, but that very little total mass is produced by this channel. We then add the contributions of rapidly-rotating massive stars (under the assumption that massive stars are spun up by collisions and interactions), and the contribution of asymptotic giant branch stars. We find that collisions can help produce the extreme abundances which are seen in some clusters. However, the total amount of material produced in these generations is still too small (by at least a factor of 10) to match the observations. We conclude with a discussion of the additional effects which probably need to be considered to solve this particular problem.