Abundances of light elements in metal-poor stars. I. Atmospheric parameters and a new T_eff_ scale

We present atmospheric parameters for about 300 stars of different chemical composition, whose spectra will be used to study the galactic enrichment of Fe and light elements. These parameters were derived using an homogeneous iterative procedure, which considers new calibrations of colour-T_eff_ relations for F, G and K-type stars based on Infrared Flux Method (IRFM) and interferometric diameters for population I stars, and the Kurucz (1992) model atmospheres. We found that these calibrations yield a self-consistent set of atmospheric parameters for T_eff_>4400K, representing a clear improvement over results obtained with older model atmospheres. Using this T_eff_ -scale and Fe equilibrium of ionization, we obtained very low gravities (implying luminosities incompatible with that expected for RGB stars) for metal-poor stars cooler than 4400K; this might be due either to a moderate Fe overionization (expected from statistical equilibrium calculations) or to inadequacy of Kurucz models to describe the atmospheres of very cool giants. Our T_eff_ scale is compared with other scales recently used for metal-poor stars; it agrees well with those obtained using Kurucz (1992) models, but it gives much larger T_eff_'s than those obtained using OSMARCS models (Edvardsson et al. 1993). This difference is attributed to the different treatment of convection in the two sets of models. For the Sun, the Kurucz (1992) model appears to be preferable to the OSMARCS ones because it better predicts the solar limb darkening; furthermore, we find that our photometric T_eff_ 's for metal-poor stars agree well with both direct estimates based on the IRFM, and with T_eff_'s derived from Hα wings when using Kurucz models.