4.1. How many low-mass stars do destroy 3He?
We show in Fig. 1 the 12C/13C ratio as a function of [Fe/H] for all the stars of our sample. The evolutionary status is also indicated by the values.
The distribution of our sample stars in these three luminosity ranges is indicated in Table 1. In each domain, we give the number of stars which present "normal" (i.e., higher than 15) and "low" (i.e., smaller than 15) carbon isotopic ratio.
Table 1. Repartition of our sample stars in the three luminosity ranges discussed in the text, as a function of their observed 12C/13C ratio
The statistics we are interested in concern the most luminous stars. They are presented in the histograms of the Fig. 2 as a function of their metallicity. We obtain that 93% of evolved stars undergo the extra-mixing on the RGB, and are thus expected to destroy, at least partly, their 3He . This high number satisfies the galactic requirements, as discussed by GSTP97. Let us note that if we take -0.5 as a limit for , we obtain that 96% of evolved stars show a 12C/13C ratio in disagreement with the standard predictions .
In Fig. 3, where we present the carbon isotopic ratios as a function of only for field stars, we underline the binary population among our sample. Clearly, binaries and single stars have the same behavior.
4.2. Need for consistent yields
The present non-standard models that attempt to explain the low 12C/13C ratios all predict a severe destruction of 3He on the RGB. However, the task for stellar evolution theory is now to propose a physical process that explains consistently the various chemical anomalies observed in low mass red giant stars. Indeed, in addition to the 12C/13C problem, the behaviour of C, O, N, Al and Na on the RGB remains unexplained in many cases (see CBW98 for references). On the other hand, the extra-mixing process on the RGB has to destroy 3He in more than 90% of the low-mass stars, and preserve it in the others, for the high (and standard) 3He abundance observed in NGC 3242 to be explained. Stars with different histories (different rotation, mass loss, ...) could suffer different mixing efficiency and thus display different chemical anomalies. It is only when all these constraints will be explained that 3He yields by low mass stars will be reliable.
4.3. The "deviant" stars
In this context, one has to raise the question of the statistical significance of Balser's et al. sample. As discussed by the authors, this source sample is indeed highly biased, due to selection criteria. The PN showing high 3He abundance should belong to the 7% of low-mass stars which do not suffer from extra-mixing on the red giant branch. They should also show "normal" carbon isotopic ratios.
This crucial test has already been verified for one PN of Balser's et al. sample for which 3He detection is probable: NGC 6720 shows a 12C/13C ratio of 23 (Bachiller et al. 1997), in agreement with the "standard" predictions. One has however to be cautious with this star for which the estimated initial mass is 2.20.6 (GSTP97). This PN could have indeed been too massive to have reached the bump before igniting helium and thus to have experienced the extra-mixing on the RGB. This is not the case for NGC 3242, which estimated initial mass is 1.20.2 . Observations of 12C/13C in this PN are now necessary.
Let us focus on the stars of the present sample that have passed the bump and do not behave as the majority. If we consider the objects with lower than -0.5, only 3 stars appear to be really "deviant" (see Table 2).
Table 2. Characteristics of the "deviant" stars
V8 (NGC6656) . As can be seen in Figs. 1 and 2, as soon as they are more luminous than the bump, almost all the stars with [Fe/H] lower than -0.5 present carbon isotopic ratios lower than 10. Only one star shows a 12C/13C ratio higher than the standard predictions : V8 in NGC 6656 (M22). This star presents enhanced SrII and BaII lines (Mallia 1976) and is probably a star enriched in 12C and s-elements.
HD 95689 (Uma). The relatively high Li abundance (logN(Li) = 1.26, Lambert et al. 1980) observed in this spectroscopic binary is another indication that this star did not suffer any extra-mixing on the RGB. It corresponds to the expected post-dilution Li for a star more massive than 1.7-2 (see Charbonnel & Vauclair 1992). This is consistent with the observed 12C/13C ratio and with the stellar value we obtain with the HIPPARCOS data. For such an initial stellar mass, the extra-mixing is not expected to occur on the RGB, and this star should be excluded from the present statistics.
HD 112989 . When [Fe/H] is higher than -0.5, the lower envelope of the 12C/13C ratio lies around 12. One star however, HD 112989, lies well below this limit. For this binary star, important differences appear in the various estimations of [Fe/H] available in the literature (-0.44, Yamashita 1964; +0.3, Lambert & Ries 1981; +0.14, McWilliam 1990; -0.05, Dracke & Lambert 1994). A smaller value of [Fe/H] would replace this star in the "normality".
On the other hand, this weak-G band star presents a Vsini value of 11km.sec-1, which is much higher than the mean rotational velocity for giants with the same spectral type (De Medeiros 1990, 1998). Among the stars of our sample for which rotational velocity has been measured with the CORAVEL spectrometer, this star is the fastest rotator, all the others having Vsini lower than 5 km.sec-1 (see Fig. 4).
© European Southern Observatory (ESO) 1998
Online publication: July 27, 1998