Spectral Matching for Elemental Abundances of Evolved Stars of Globular Clusters

Abstract

In order to understand the origin of globular clusters, large samples of their stars need to be observed and analyzed for their chemical composition. This is especially true for the complex, multimetallic cluster ω Centauri, with its large range of iron, carbon, nitrogen, oxygen, sodium and barium abundances. In order to accomplish this, an automated spectral matching pipeline was developed to determine these abundances. This thesis made use of photometry and low resolution spectroscopy to analyze the chemical composition of evolved stars in three clusters: ω Cen, 47 Tuc and NGC 6752. The latter two clusters are monometallic and selected due to their similar metallicities to the metal-rich and metal-poor stars in ω Cen. This allowed them to be used as test-cases for the spectral matching pipeline.

For ω Cen, two analyses were performed. In the first, 221 giant branch stars were selected that had known [O/Fe]. These stars showed the expected anticorrelation in [C/Fe]to [N/Fe]. In the second, spectral indices were used to estimate the oxygen abundance of the stars, leading to a determination of whether a particular star was oxygen-rich or oxygen-poor. From this a catalogue of abundances of iron, carbon and barium of 848 giant branch stars were determined, of which 557 also had well-defined nitrogen abundances. k-means clustering analysis was used to group the stars in ω Cen into four homogeneous groups based upon these abundances. These groups suggest that there were at least four main periods of star formation in the cluster. The exact order of these star formation events is not yet understood, with some models predicting the groups formed from iron-poorest to iron-richest, while others suggest the potential for iron-poorer groups to form after iron-rich groups.

These results compare well with those found from higher resolution studies and show the value of more extensive lower resolution spectral surveys. They also highlight the need for large samples of stars when working with a complex object like ω Cen.