5 Conclusion

Our results confirm that the AGB evolution depends on the initial mass of the progenitor on the main sequence. The study of LPVs with peculiar properties, often associated with transition states in the stellar evolution, elucidates some points of the very complex evolution along the AGB. The simultaneous study of the behaviour of the circumstellar envelope provides further information on the evolutive state of the stars along the AGB. However, this study is mainly statistical and so results for individual stars can be erroneous because the confidence level of a probabilistic discrimination can never reach 100%.

The proposed evolutive scenario, schematically represented in Fig. 7, is:

• At the beginning of the variability phase, stars belonging to the disk 2 or old disk population (i.e., with a not too large initial mass ${\cal M}_{\rm ms}$) are irregular L or semi-regular SRb O-rich variables. They begin to slowly produce an envelope. Thereafter the envelope expands, and the IRAS luminosities of the star rapidly grow brighter. They become O-rich SRa or Mira. Depending on smaller or greater ${\cal M}_{\rm ms}$, they can evolve either to OH emitters and finally OH-IR sources or, for the more massive, to be enriched in s elements and then in carbon by successive dredge-ups and finally become a C-rich Mira. The least massive stars leave the AGB with no special transformation of their surface abundance;

• The disk 1 and bright disk population LPVs seem to very rapidly build a bright expanding envelope. They can also evolve to OH or C-rich stars. Although this population is not attractive for O-rich LPVs, a tenth of OH Mira belong to it. They are probably not massive enough to be sufficiently carbon enriched. A few may be HBB stars.

  If the mass of the star is high enough, after a number of dredge-ups the external shells of the star can be enriched in carbon. The C/O ratio becomes larger than 1. When C/O is around 1, the star is an S star. At the same time, strong changes take place in the circumstellar envelope, which becomes dominated by carbonated grains, and the 25-12 index increases in conformity with the loop drawn in the IRAS color-color diagram, as predicted by Willems & de Jong (1988) and calculated by Chan & Kwok (1988). Our luminosity calibrations in 12 and 25 clearly show that the star becomes fainter in both luminosities but more in 25 than in 12. The C-rich irregular and SRb stars seem to be the most evolved and massive.

The study of LPVs enriched in Tc confirms that they are at different stages along the AGB but after a recent dredge-up. It allows us to confirm the location of the first dredge-up at $M_{\rm bol}=-3.5$ and the quite early operative third dredge-up on the TP-AGB

Figure 7: Schematic illustration of the proposed stellar and circumstellar evolutive scenario.

The no-Tc S-type LPVs (except R And), are faint in K, 12 and 25, and they are confirmed as extrinsic S stars enriched not by their own nucleosynthesis but by mass exchange from a more evolved companion. The extrinsic enrichment in s-elements may accelerate the evolution along the AGB and lead to the formation of an envelope closer to being carbonated than silicated before any intrinsic enrichment by successive dredge-ups.

The examination of the brightest LPVs allows us to propose a list of stars with peculiar spectral, envelope and luminosity properties that may be Hot Bottom Burning candidates. The most luminous of them, R Cen, a star in a He-shell flash, could become, before leaving AGB, a C-rich LPV brighter than the usual luminosity limit of carbon stars.

Acknowledgements

This study was supported by the PICASSO program PICS 348 and by the CICYT under contract ESP97-1803 and AYA2000-0937. We thank N. Mowlavi and R. Alvarez for constructive remarks and A. Gomez and S. Van Eck for fruitful discussions of our preliminary results.