Nucleosynthesis and mixing on the asymptotic giant branch.: III.: Predicted and observed s-process abundances

We present the results of s-process nucleosynthesis calculations for asymptotic giant branch (AGB) stars of different metallicities and different initial stellar masses (1.5 and 3 M-circle dot), and we present comparisons of them with observational constraints from high-resolution spectroscopy of evolved stars over a wide metallicity range. The computations were based on previously published stellar evolutionary models that account for the third dredge-up phenomenon occurring late on the AGB. Neutron production is driven by the C-13(alpha ,n)O-16 reaction during the interpulse periods in a tiny layer in radiative equilibrium at ther top of the He- and C-rich shell. The neutron source C-13 is manufactured locally by proton captures on the abundant C-12; a few protons are assumed to penetrate from the convective envelope into the radiative layer at any third dredge-up episode, when a chemical discontinuity is established between the convective envelope and the He- and C-rich zones. A weaker neutron release is also guaranteed by the marginal activation of the reaction Ne-22(alpha ,n)Mg-25 during the convective thermal pulses. Owing to the lack of a consistent model for C-13 formation, the abundance of C-13 burnt per cycle is allowed to vary as a free parameter over a wide interval (a factor of 50). The s-enriched material is subsequently mixed,with the envelope by the third dredge-up, and the envelope composition is computed after each thermal pulse. We follow the changes in the photospheric abundance of the Ba-peak elements (heavy s [hs]) and that of the Zr-peak ones (light s [ls]), whose logarithmic ratio [hs/ls] has often been adopted as an indicator of the s-process efficiency (e.g., of the neutron exposure). Our model predictions for this parameter show a complex trend versus metallicity. Especially noteworthy is the prediction that the flow along the s-path at low metallicities drains the Zr and Ba peaks and builds an excess at the doubly magic Pb-208, which is at the termination of the s-path. We then discuss the effects on the models of variations in the crucial parameters of the C-13 pocket, finding that they are not critical for interpreting the results. The theoretical predictions are compared with published abundances of s-elements for AGB giants of classes MS, S, SC, post-AGB supergiants, and for various classes of binary stars, which supposedly derive their composition by mass transfer from an AGB companion. This is done for objects belonging both to the Galactic disk and to the halo. The observations in general confirm the complex dependence of neutron captures on metallicity. They suggest that a moderate spread exists in the abundance of C-13 that is burnt in different stars. Although additional observations are needed, it seems that a good understanding has been achieved of s-process operation in AGB stars. Finally, the detailed abundance distribution including the light elements (CNO) of a few s-enriched stars at different metallicities are examined and satisfactorily reproduced by model envelope compositions.