5 C/O ratio vs. effective temperature for the three populations of carbon-rich giants

The existence of at least three populations of Galactic carbon giants was demonstrated in Paper II on the grounds of vertical space distribution and kinematics. Most HC-stars are members of the thick disk, and their sample is contaminated by a spheroidal component, namely the HC-stars classified as CH stars by spectral criteria (see also Hartwick & Cowley 1985). Most CV-stars are members of the old (thin) disk. We have found this result again in Sect. 6 of Paper III, where the former two populations correspond to the maximum in the LF at $M_{\rm {bol}}\simeq -1.75$ or -1.25 (see Fig. 5 of Paper III) and the latter one can be associated with the second maximum at $M_{\rm {bol}}\simeq -4.75.$ With a separation of more than three magnitudes, the maxima are clearly disentangled. The transition in the HR diagram between the former two populations (HC-stars) and the latter one (CV-stars) is shown in Fig. 9 of Paper III, and nearly coincides with the CSFL for TP-AGB stars having experienced TDU. When compared with the theoretical evolutionary tracks, the former locus (HC-stars) appears shifted leftward if compared to the latter one (CV-stars). We interpret this shift in terms of lower metallicity on average, in the former low-mass old stars (see references cited hereafter in this section), as suggested by the comparisons of tracks for Z= 0.008 to those for Z= 0.02.

We display now further evidence in favor of the three above-mentioned populations. This is the carbon to oxygen (C/O) abundance ratio shown as a function of effective temperature in Figs. $\ref{co1}$, $\ref{co2}$ and $\ref{co3}$ respectively for the CH stars, the HC-stars not classified as CH stars, and finally the CV-stars. The effective temperatures, essentially from Paper I, can be found in Table 2 of Paper III, at CDS. The C/O ratios are taken from Dominy (1984), Lambert et al. (1986), Vanture (1987), Kipper & J$\phi$rgensen (1994), Abia & Isern (1996), and Kipper et al. (1996). We concentrated on values derived from comparisons of observed spectra to model atmospheres, and ignored values found in the literature when deduced from the calibration of a color index from photometry. Only CH stars identified from detailed spectral analysis were kept, excluding the objects mentioned as "CH-like'' or "CH-candidates''.

Figure 5: The C/O abundance ratio as a function of effective temperature for the CH stars (classified in an oxygen-rich or an HC-group). A curve was drawn by hand through the eight available points, a maximum being apparent near 4900 K.

Figure 6: The same diagram as Fig. $\ref{co1}$ but for the HC (Hot Carbon) stars not classified CH stars on spectroscopic grounds. A curve was drawn by hand through the eleven points available, a maximum of the C/O ratio being apparent near 3870 K.

Figure 7: The same diagram as Fig. $\ref{co1}$ for the CV (Carbon Variables) stars which are the coolest ones. There are 39 stars available populating an area below the curve plotted by hand. This "curve of maximas'' shows a maximum near 2500 K.

The most remarkable feature in Figs. $\ref{co1}$, $\ref{co2}$ and $\ref{co3}$ is the existence of maxima whose positions and amplitudes are

• $\left({\rm C/O}\right)\rm {max}\simeq8$ at $\left(T_{\rm {eff}}\right)\rm {max}\simeq\left(4900\pm100\right)\rm {K}$ for the CH stars,
• $\left({\rm C/O}\right)\rm {max}\simeq3.3$ at $\left(T_{\rm {eff}}\right)\rm {max}\simeq\left(3870\pm100\right)\rm {K}$ for the HC-stars not classified as CH stars,
• and $\left({\rm C/O}\right)\rm {max}\simeq1.5$ to 1.75, at $\left(T_{\rm {eff}}\right)\rm {max}\simeq\left(2500\pm100\right)\rm {K}$ for the CV-stars, but with most values close to unity.

Thus, the maximum of the C/O ratio shifts toward lower effective temperatures from one population to the next and its amplitude decreases correspondingly. It should be noted that in Figs. $\ref{co1}$ and $\ref{co2}$ documented with only eight and eleven stars respectively, it is not clear whether the tentative curves shown are truly mean curves or only upper limits for the loci. In Fig. $\ref{co3}$, there are 39 CV-stars and the case is clear. It is possible to estimate an approximate maximum C/O value for every effective temperature, resulting in the curve shown. The result of Paper I is thus confirmed (increasing mean values and dispersions of the C/O ratio along the sequence of the photometric CV-groups until CV6). Making use of effective temperatures instead of photometric groups (and of an enlarged sample), however, reveals a decrease of photospheric C/O for $T_{\rm {eff}}\la 2500~\rm {K}.$ This is precisely the temperature below which circumstellar dust (carbon and SiC) efficiently condenses, as shown by infrared excesses of circumstellar origin. The model atmospheres used in the above-mentioned photospheric studies did not include dust species. The maximum may thus be apparent in Fig. $\ref{co3}$, referring to gas C/O only. We have not included here the very evolved object C2619 = CW Leo = IRC+10216 (CV7, $\left<T_{\rm {eff}}\right>\simeq 2000~\rm {K}$) whose C/O ratio is estimated to be 1.4-1.5. Due to large non-spherical obscuration (e.g. Knapik et al. 1999 and references therein) this is a circumstellar value obtained well above the region of dust condensation with large C-depletion in the gas. Thus this is not in contradiction with our results in Fig. $\ref{co3}$, that correspond to nearly photospheric values.

As expected, low-mass HC-stars $\left(M_{\rm {i}}\le1.15~M_{\odot}\right)$ with low metallicities do exhibit on average higher C/O ratios than the CV-stars with higher masses and metallicities close to solar value. The main results and general trend are consistent with evolutionary tracks shifted leftward (increasing effective temperatures) in the HR diagram, with decreasing metallicity. It is usually assumed that CH stars are binary members whose carbon enrichment occurred through mass transfer from a former TP-AGB primary, now a white dwarf (McClure 1989; McClure & Woodsworth 1990). No evidence for binarity was found in the HC-stars not classified as CH stars, and their status is not clear at all (Sect. 7 of Paper III). The CV-stars are clearly identified with stars enriched in carbon through TDU in TP-AGB objects (Sect. 7 of Paper III). We can summarize those facts by the comments "presumably extrinsic'', "unknown'' and "intrinsic'' for the three categories respectively. The high C/O ratios observed in most CH stars are however somewhat surprising for stars that are usually considered as the equivalent of BaII stars ("extrinsic'' objects) in the spheroidal component.