5. Concluding remarks
We have shown that the hypothesis of deep mixing in stars ascending the RGB can explain self-consistently the anomalous abundances of C (as well as the 12 C/13 C ratios), N, O and Na in Cen giants and the global anticorrelation of [O/Fe] versus [Na/Fe] most clearly seen in M 13 giants. There is much other observational evidence of deep mixing in GCRGs, the most direct being the progressive decline of [C/Fe] with increasing luminosity on the RGB seen in several globular clusters (see references in Sect. 1). Similar evidence has been obtained recently by Pilachowski et al. (1996), who report that red giants in M 13 become more enriched in Na as they approach the RGB tip.
We did not discuss the nature of the mechanism driving the deep mixing. The most promising candidate seems to be some kind of rotationally induced instability (baroclinic and/or shear instability etc.). In this case clusters with especially strong star-to-star abundance variations are expected to contain a lot of rapidly rotating stars. Indeed, horizontal branch stars in M 13 are found to possess unusually fast rotation (Peterson et al. 1995), and Cen seems to have one of the bluest horizontal branches among the globular clusters (Whitney et al. 1994) which may be a manifestation of surface He enrichment during RGB evolution (Rood 1973; Sweigart 1997). Of course, the latter observational fact does not exclude a primordial origin of any He enrichment.
Unfortunately, the deep mixing scenario alone cannot account for the MgAl anticorrelations in the M 13 and Cen giants in the absence of additional ad hoc assumptions. Possible modifications include a strong but still undetected low energy resonance in the reaction 24 Mg(p, Al, and episodical increases of the HBS temperature up to the value (the standard models predict ) in stars ascending the giant branch. The first assumption, however, will most probably raise objections by nuclear physicists, whereas the second disagrees with the results of S96's magnesium isotopic analysis (Sect. 4.3). As regards the second possibility, the work of taking into account feedback from deep mixing on the structure and evolution of red giants still remains to be done before one can draw any further conclusions. Finally, given the importance of Shetrone's results for the present discussion, we emphasize the importance of their confirmation and amplification.
On the other hand, there are very convincing observational arguments in favour of the primordial scenario. To those mentioned in Sect. 4 one can add the well-known CN bimodality in 47 Tuc stars which has been traced down to the MS turn-off (Briley et al. 1994). Moreover, Briley et al. (1996) have recently shown that the strong CN (and weak CH) molecular band widths are accompanied by spectroscopic signatures of increased Na abundance in stars just below the MS turn-off in 47 Tuc. We have tried to determine whether mixing in a low mass MS star can produce surface C depletion accompanied by N and Na enhancements without dredging up too much He (which would conflict with the very narrow CM diagram of 47 Tuc at the MS turn-off; VandenBerg & Smith 1988), but we failed even with the new NeNa-cycle reaction rates of El Eid & Champagne (1995). Hence, the CN bimodality and the Na enhancements in 47 Tuc stars are most likely of primordial origin and AGB stars might be responsible for this.
We have proposed a primordial plus deep mixing scenario in which intermediate mass AGB stars are considered as the source of a primordial anticorrelation of 24 Mg versus 25 Mg in GCRGs, with 24 Mg being depleted in HBB, and 25 Mg being increased in He shell burning during pulses. The anticorrelations of [O/Fe] versus [Al/Fe] in M 13 and Cen giants are then very well reproduced in the deep mixing calculations with Al synthesized at the expense of 25 Mg. To produce extremely large Al enhancements, which are observed in some GCRGs, this scenario requires that the following two necessary conditions be fulfilled: (1) the star must have a high initial 25 Mg abundance, and (2) a deep mixing mechanism must be switched on in the star on the RGB.
An observational confirmation of our combined scenario would be finding a red giant which did not show signs of deep mixing (i.e. had large [O/Fe] and low [Na/Fe] and [Al/Fe]), but at the same time possessed depleted [Mg/Fe] and increased [25 Mg+26 Mg/Fe]. At present there is only one star in S96's sample, L598, showing no sign of efficient (if at all) mixing. Unfortunately, with 24 Mg/25 Mg/26 Mg = 94/3/3, it does not possess any trace of pollution from intermediate mass AGB stars. Hence, for this particular object both of the above conditions are not met. Another test would be to determine whether the sum 25 Mg+26 Mg in fact consists of 26 Mg only, because in our calculations the 25 Mg isotope is destroyed completely in GCRGs with efficient mixing.
We comment, finally, on the differences in the abundance patterns between globular cluster red giants, on the one hand, and the field halo giants, on the other. These groups differ in two well documented ways. First, anomalies involving C and N are much more prevalent in cluster stars (see Langer, Suntzeff & Kraft 1992, and references therein), and, second, it appears that enhancements of [(25 Mg+26 Mg)/24 Mg] exist preferentially in clusters rather than in the field. In the context of the two above-noted conditions, it appears that for field stars one would then require that neither of them be met. That is to say, one requires not only that field stars have not experienced enrichment from AGB stars but also that they have not experienced deep mixing. One might argue that economy of hypothesis suggests it is more palatable to assume, in the context of the alternatives discussed above, that an agency exists in globular cluster giants which leads both to deep mixing, on the one hand, and which operates in conjunction with either the postulated elevated temperatures () in the HBS or an accelerated rate for the reaction 24 Mg(p, Al, on the other, and that this does not exist in the field stars.
Recently Smith & Kraft (1996) have considered another combined scenario where the initial overabundance of 25 Mg in GCRGs is assumed to come from Ne novae. Our paper extends the list of primordial plus deep mixing scenarios. What will be next?
© European Southern Observatory (ESO) 1998
Online publication: April 28, 1998