Context. Massive AGB ( hereafter super- AGB or SAGB) stars ignite carbon off- center and have initial masses ranging between M-up, the minimum initial mass for carbon ignition, and M-mas the minimum mass for the formation of an iron core collapse supernova. In this mass interval, stars more massive than M-n will undergo an electron capture supernova ( EC- SN). Aims. We study the fate and selected evolutionary properties of SAGB stars up to the end of the carbon burning phase as a function of metallicity and core overshooting. Methods. The method is based on the analysis of a large set of stellar models covering the mass range 5- 13 M circle dot and calculated for 7 different metallicities between Z = 10(-5) and twice solar. Core overshooting was considered in two subsets for Z = 10(-4) and 0.02. The models are available online at http:// www- astro. ulb. ac. be/similar to siess/ database. html. The fate of SAGB stars is investigated through a parametric model which allows us to assess the role of mass loss and of the third dredge- up. Results. Our main results can be summarized as follows: a) prior to C- burning, the evolution of SAGB stars is very similar to that of intermediate- mass stars, being more luminous, b) SAGB stars su. er a large He enrichment at the end of the second dredge- up, c) the limiting masses M-up, M-n and M-mas present a nonlinear behavior with Z, characterized by a minimum around Z = 10- 4, d) the values of M-up, M-n and M-mas are decreased by similar to 2 M circle dot when core overshooting is considered, e) our models predict a minimum oxygen-neon white dwarf mass of similar to 1.05 M circle dot, f) the determination of Mn is highly dependent on the mass loss and core growth rates, g) the evolutionary channel for EC- SN is limited to a very narrow mass range of <= 1- 1.5 M circle dot width and this mass window can be further decreased if some metallicity scaling factor is applied to the mass loss rate, h) the final fate of SAGB stars is connected to the second dredge- up and this property allowed us to refine the initial mass range for the formation of EC- SN. We find that if the ratio of the mass loss rate to the core growth rate averaged over the post carbon- burning evolution zeta = broken vertical bar <((M) over dot)over bar >(loss broken vertical bar)<((M) over dot)over bar >broken vertical bar is greater than about 70- 90, the evolutionary path to EC- SN is not accessible.
