Systematic Differences in Spectroscopic Analysis of Red Giants

A spectroscopic analysis of stellar spectra can be carried out using different approaches; different methods, line lists, atomic parameters, solar abundances etc. The resulting atmospheric parameters from these choices can vary beyond quoted uncertainties in the literature. Here we characterize these differences by systematically comparing some of the commonly adopted ingredients; line lists, equivalent width measurements and atomic parameters. High resolution and high signal-to-noise spectroscopic data of one helium-core-burning red giant star in each of the three open clusters, NGC6819, M67 and NGC188 have been obtained with the FIES spectrograph at the Nordic Optical Telescope. The M67 target has been used to benchmark the analysis, as it is a well studied cluster with asteroseismic data from the K2 mission. For the other two clusters we have obtained higher quality data than previously analyzed, which allows us to establish their chemical composition more securely. Using a line by line analysis, we tested several different combinations of line lists and programs to measure equivalent widths of absorption lines to characterize systematic differences within the same spectroscopic method. The obtained parameters for the benchmark star in M67 vary up to ~170 K in T_eff, ~0.4 dex in log g and ~0.25 dex in [Fe/H] between the tested setups. Using the combination of equivalent width measurement program and line list that best reproduce the inferred log g from asteroseismology, we determined the atmospheric parameters for the three stars and securely established the chemical composition of NGC6819 to be ~solar, [Fe/H]=-0.02+-0.01 dex. We have highlighted the significantly different results obtained with different combinations of line lists, programs and atomic parameters, which emphasize the importance of benchmark stars studied with several different methods to anchor spectroscopic analyses.