Predictions for self-pollution in globular cluster stars

Fully evolutionary models have been built to follow the phases of asymptotic giant branch evolution with mass loss for metal mass fractions from Z = 2 x 10(-4) to Z = 4 x 10(-3). The hot bottom burning at the base of the convective envelope is followed by fully coupled nuclear burning and noninstantaneous mixing. The models also show the occurrence of a spontaneous (i.e., not induced by overshooting) third dredge-up. For the first time, we find that temperatures close to or even larger than 10(8) K are achieved at low Z; the full CNO cycle operates at the base of the envelope, the O-16 abundance for the most metal-poor models of 4 and 5 M-., is drastically reduced, and sodium and aluminum production by proton capture on neon and magnesium can occur. Lithium is first largely produced in the envelope and then burned completely, so the average lithium abundance in the expelled envelope is a factor of up to 5 times smaller than the initial one, but it is never completely depleted. These results may be relevant for the evolution of primordial massive globular clusters; we suggest that the low-mass stars may have been polluted at the surface by accretion from the gas that was lost from the evolving intermediate-mass stars at early ages [(1-2 x 10(8) yr]. In this hypothesis, we should expect that the polluted stars show smaller abundances of oxygen, larger abundances of products of advanced nucleosynthesis (as Na and Al), and lower, but never negligible, abundances of lithium. The abundance spreads should be smaller in clusters of higher metallicities, where the lithium in the polluted stars could be larger than in the nonpolluted stars.