New light on stellar abundance analyses: Departures from LTE and homogeneity

The information on the chemical compositions of stars encoded in their spectra plays a central role in contemporary astrophysics. Stellar element abundances are, however, not observed: to decipher the spectral fingerprints in terms of abundances requires realistic models for the stellar atmospheres and the line-formation processes. Still today, the vast majority of abundance analyses of late-type stars rely on one-dimensional (1D), hydrostatic model atmospheres and the assumption of local thermodynamic equilibrium (LTE). In this review possible systematic errors in studies of F-, G- and K-type stars introduced by these questionable approximations are discussed. Departures from LTE are commonplace and often quite severe, in particular for low surface gravities or metallicities, with minority species and low-excitation transitions being the most vulnerable. Recently, time-dependent, 3D, hydrodynamical model atmospheres have started to be employed for stellar abundance purposes, with large differences compared with ID modeling found in particular for metal-poor stars. An assessment of non-LTE and 3D effects for individual elements as well as on the estimated stellar parameters is presented.