Consistent metallicity scale for cool dwarfs and giants A benchmark test using the Hyades

Context. In several instances chemical abundances of main-sequence and giant stars are used simultaneously under the assumption that they share the same abundance scale. This assumption, if wrong, might have important implications in different astrophysical contexts. Aims. It is therefore crucial to understand whether the metallicity or abundance differences among dwarfs and giants are real or are produced by systematic errors in the analysis. We aim to ascertain a methodology capable of producing a consistent metallicity scale for giants and dwarfs. Methods. To achieve that, we analyzed giants and dwarfs in the Hyades open cluster, under the assumption that they share the same chemical composition. All the stars in this cluster have archival high-resolution spectroscopic data obtained with HARPS and UVES. In addition, the giants have interferometric measurements of the angular diameters. We analyzed the sample with two methods. The first method constrains the atmospheric parameters independently from spectroscopic method. For that we present a novel calibration of microturbulence based on 3D model atmospheres. The second method is the classical spectroscopic analysis based on Fe lines. We also tested two different line lists in an attempt to minimize possible non-LTE effects and to optimize the treatment of the giants. Results. We show that it is possible to obtain a consistent metallicity scale between dwarfs and giants. The preferred method should constrain the three parameters T-eff, log g, and xi independent of spectroscopy. A careful selection of Fe lines is also important. In particular, the lines should not be chosen based on the Sun or other dwarfs, but specifically to be free of blends in the spectra of giants. When attention is paid to the line list, the classical spectroscopic method can also produce consistent results. In our test, the metallicities derived with the well-constrained set of stellar parameters are consistent independent of the line list used. Therefore, for this cluster we favor the metallicity of +0.18 +/- 0.03 dex obtained with this method. The classical spectroscopic analysis, using the line list optimized for the giants, provides a metallicity of +0.14 +/- 0.03 dex, in agreement with previous works.