Exploration of s-process elemental abundances in globular cluster stars using medium- and high-resolution spectra.
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
This thesis has used medium- and high-resolution spectral data to derive elemental abundances, in particular light and heavy s-process elemental abundances, for groups of giant stars in the globular clusters 47 Tuc, NGC 6388 and NGC 362. These analyses were undertaken using both curve-of-growth and spectrum synthesis techniques. The techniques were calibrated with respect to the metal-poor giant star Arcturus in order to reduce systematic errors in the analysis process.
A feasibility study was undertaken that compared synthetic spectra at different resolutions throughout the colour-magnitude diagram (CMD) of a metal-rich ([Fe/H] = -0.5 dex) globular cluster. This study identified where on the CMD light and heavy s-process elemental abundances could be derived at medium resolution (R ~ 10,000). Abundance analyses could be undertaken on the giant branches down to just below the horizontal branch and then again on the main sequence below Teff ~ 4500 K. At all other places on the CMD high-resolution spectra (R ~ 30,000) are required to derive these abundances.
Performance verification data at R ~ 5,000$ was obtained using the Robert Stobie Spectrograph (RSS) on the Southern African Large Telescope (SALT) and showed that there were no large scale s-process elemental abundance variations in 47 Tuc giant branch stars. The level of precision in this preliminary data was [X/Fe] ~ 0.5dex. A resolution of R ~ 10,000 should be achievable with SALT RSS in the future which will improve this limit. The AAOmega survey of 47 Tuc stars at R ~ 6,500 was more promising in certain aspects of elemental abundance determination. The observed wavelengths included the key features of CN and CH molecular bands, and light (Z < 30) and heavy (38 < Z < 63) element spectral lines. CN indices were measured and calibrated to previous results. The well-known CN bimodality was observed in the 47 Tuc stars, as well as a radial gradient in CN strength. A preliminary subset of ten of the survey stars have undergone an abundance analysis for which the abundances of Fe, Si, and Ca were found to be homogeneous within this cluster. The Na abundances had a large range in values that were observed to correlate with CN strength. The s-process elemental abundance results were inconclusive. The Zr abundances showed little to no enhancement in the sample and the Ba abundances varied considerably due to strong lines of Ba II being extremely sensitive to microturbulence.
Various high-resolution studies were carried out using spectra of giant stars in 47 Tuc, NGC 6388 and NGC 362 observed on the Australian National Observatory (ANO) 2.3 m echelle spectrograph and the Ultra-Violet Echelle Spectrograph (UVES) on the Very Large Telescope (VLT). The high-resolution analysis of 47 Tuc giant star Lee 2525 found an enhanced Zr abundance in this star which resolved a discrepancy between two previous 47 Tuc elemental abundance studies (Brown & Wallerstein 1992; Wylie et al. 2006). The stars in the VLT dataset that were analysed here included five giant branch stars in 47 Tuc, two in NGC 6388 and thirteen in NGC 362. The low temperatures and gravities of these stars caused departures from local thermodynamic equilibrium in low excitation potential neutral species, particularly Fe and Zr, that needed to be taken into account before reliable stellar parameters and elemental abundances could be determined for these stars. Veiling effects due to circumstellar dust were postulated to have produced artificially low metallicities for the infra-red excess stars in this sample, particularly for the 47 Tuc stars.
The element abundance analyses of 47 Tuc, NGC 6388 and NGC 362 stars found the derived metallicities to be homogeneous for each cluster (<[Fe/H]>(47Tuc) = -0.88 +/- 0.09 dex; <[Fe/H]>(NGC6388) = -0.60 +/- 0.06 dex; <[Fe/H]>(NGC362) = -1.21 +/- 0.08 dex). The 47 Tuc sample included Lee 2525 and the five VLT stars. The derived metallicities were in reasonable agreement with previously reported values. The light (ls) and heavy (hs) s-process element abundances were enhanced and homogeneous in the stars of each cluster. The abundances determined for 47 Tuc and NGC 6388 were in good agreement, reflecting the similarity in metallicity of the stars in these two clusters (<[ls/Fe]>(47Tuc) = +0.53 +/- 0.02 dex; <[hs/Fe]>(47Tuc) = +0.40 +/- 0.06 dex; <[ls/Fe]>(NGC6388) = +0.58 +/- 0.13 dex; <[hs/Fe]>(NGC6388) = +0.39 +/- 0.07 dex). The more metal-poor cluster NGC 362 was less enhanced in ls elemental abundances and slightly more enhanced in hs elemental abundances (<[ls/Fe]>(NGC362) = +0.32 +/- 0.10 dex, <[hs/Fe]>(NGC362) = +0.46 +/- 0.09 dex).
The clear enhancement in the s-process elemental abundances and homogeneity in the results for each globular cluster is evidence that these stars have been enhanced extrinsically in s-process elements. Pollution events in the history of each cluster has resulted in the abundance distribution in both the light elements and the heavy elements that has been observed in the stars analysed in this thesis. The enhancements in Na, ls and hs elemental abundances favours intermediate mass AGB stars as the source of the pollution.