Deep secrets of intermediate-mass giants and supergiants

Context. Recent observational results have demonstrated an increase in the surface Na abundance that correlates with stellar mass for red giants between 2 and 3M⊙. This trend supports evolutionary mixing processes as the explanation for Na overabundances seen in some red giants. In this same mass range, the surface Al abundance was shown to be constant.Aims. Our main aim was to extend the investigation of the Na and Al surface abundances to giants more massive than 3M⊙. We sought to establish accurately whether the Na abundances keep increasing with stellar mass or a plateau is reached. In addition, we investigated whether mixing can affect the surface abundance of Al in giants more massive than 3M⊙.Methods. We obtained new high-resolution spectra of 20 giants in the field of 10 open clusters; 17 of these stars were found to be members of 9 clusters. The giants have masses between 2.5M⊙and 5.6M⊙. A model atmosphere analysis was performed and abundances of up to 22 elements were derived using equivalent widths. Additionally, abundances of C, N, and O were determined using spectrum synthesis. The abundances of Na and Al were corrected for non-local thermodynamic equilibrium (non-LTE) effects. Moreover, to extend the mass range of our sample, we collected from the literature high-quality C, N, O, and Na abundances of 32 Galactic Cepheids with accurate masses in the range between 3M⊙and 14M⊙.Results. The surface abundances of C, N, O, Na, and Al were compared to predictions of stellar evolution models with and without the inclusion of rotation-induced mixing. The surface abundances of most giants and Cepheids of the sample can be explained by models without rotation. For giants above Ȉ2.5M⊙, the Na abundances reach a plateau level of about [Na/Fe] ~ 0.20–0.25 dex (in non-LTE). This is true for both Cepheids and giants in open clusters. Regarding Al, the non-LTE [Al/Fe] ratios are mostly close to solar and suggest that Al is not affected by the first dredge-up up to ~5.0M⊙. Our results support previous works that found models with rotation to overestimate the mixing effects in intermediate-mass stars.