The chemical evolution of the Milky Way: pushing APOGEE to higher precision and accuracy

Sammanfattning: In the recent decade, a rapid increase in all-sky spectroscopic surveys has occurred, and with it, the wealth of observational data and information on our Galaxy. Consequently, numerous detections of new structures, such as new open clusters, accreted and in-situ components, are discovered. However, the accuracy and results of industrialised pipelines to derive stellar parameters and abundances have been shown to be problematic for different types of stars, and their accuracy and precision are debated. In this thesis, spectra released by APOGEE are re-analysed and compared with results from the high-resolution spectrograph IGRINS available on GEMINI South. A new pipeline to derive elemental abundances for numerous stars is designed by analysing the effects of resolution and stellar parameters. Here the focus is on stars in the APOKASC Kepler sample, which have highly accurate asteroseismic log g and thereby eliminate log g uncertainties. I calculate elemental abundances for more than 3000 giant stars using the developed automated pipeline. The abundance trends are generally similar to the abundances reported by APOGEE but include smaller differences. They differ primarily in the metal-poor end and, for some, at super-solar metallicities. For cerium, the computed abundances are of higher precision and follow abundances derived from optical spectra more closely than APOGEE abundances. For Ti and iron-peak elements V and Co, the lower resolution and the APOGEE stellar parameters cause problems for metal-poor stars, generally resulting in larger scatter and too low abundances. For α-elements, their respective abundance trends are similar as their spectral lines remain their strength at the metal-poor end. In addition, multiple stars enhanced in s-process elements are detected, for which additional s-process elements show promise to be derived from APOGEE spectra. To conclude, with careful spectral analysis, the abundances derived shows a minor effect from the spectral resolution. However, the higher resolution has been shown to perform better for blended spectral lines. The major difference on abundances derived in this thesis are caused by the adopted stellar parameters. As a result spectral lines with high dependence on stellar parameters become problematic. It generates too low abundances and increases the scatter in the abundance trends. Finally, by re-analysing APOGEE spectra, the possibility of deriving abundances for neodymium looks promising.