Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 344, 263-276 (1999)

Previous Section Next Section Title Page Table of Contents

4. The CS and SC stars

4.1. Introduction

The stars of spectral type S are identified among cool giants by the bands of ZrO and LaO instead of the TiO-ones conspicuous in the spectra of the M-types stars. Their spectra also show strong enhancements of the s-process elements and a high C/O abundance ratio however less than unity (e.g. Keenan & Boeschaar 1980). The intrinsic S stars (e.g. Jorissen & Mayor 1992) are high luminosity AGB objects. Nucleosynthetised elements were recently brought to the surface by convective mixing. A piece of evidence is the observed technetium which has no isotope with period longer than 1 million years. It has often been argued that those stars might be a short-lived phenomenon between the stages of oxygen-rich and carbon-rich giants. The choice of the evolutionary sequence to be adopted is however controversial (Zuckerman & Maddalena 1989, de Jong 1989). A catalogue of galactic S stars has been published by Stephenson (1976).

4.2. The analysis of CS and SC stars

Stars like FU Ori or UY Cen show spectra intermediary between those of C and S stars. They are characterized by weak molecular bands, ZrO and CN being both present.

We selected 23 bright SC and CS stars, but only 21 stars in this sample proved sufficiently documented. Our pair method (see Sect. 2) was again applied, but good solutions in terms of carbon CV-types could be obtained for only 7 stars (1 unreddened, 6 reddened: see Table 2) out of 20. Five of them are Miras. The extinction law is again the mean law adapted from Mathis (1990). There is evidence of CVi-variations with phase.


Table 2. The CS stars in the present study. The stars C-entries from Stephenson (Stephenson 1989) are given. The Miras are referenced by (M) and the phase of the used observations is quoted. The number CN is the intensity index of the red system of the CN-bands in the near IR from Baumert (1972). Our solution is G, E(B-V) as quoted; (a) the group is HC5 (just earlier than CV1: see Paper II) at phase 0.9, but evolve to CVi at later phases.

The SEDs of 6 of the remaining 13 stars (including UY Cen and AM Cen), presumably unreddened or slightly reddened, were found roughly similar to each other. We made a differential analysis of couples of stars. Our conclusion is that, within the errors, the 13 SEDs can be interpreted as a single intrinsic (unreddened) SED affected by interstellar extinction in various amounts. These stars are thus found as members of a homogeneous group we named SCV (see Table 3). The zero of the scale was then determined. As mentioned above, a few SCV stars are slightly reddened but finally, none of them was found unreddened in the present analysis (this is why SCV is not quoted in Table 7). This is a regrettable situation which will not change since the brightest spectroscopic SC stars on the whole sky were studied here. It is worth noting however that AM Cen and U Vol are attributed small colour excesses (0.03 and 0.04 respectively).


Table 3. The SC stars in the present study. The stars S-entries from Stephenson (Stephenson 1976) are given (S1093 is also C4152, V346 Aur is C797, and the IRAS-entries of S674 and S935 are given). Two possible Miras are referenced by (M:). The CN index is as in Table 2. Our solution is SCV for every star and E(B-V) as quoted.

4.3. The obtained extinctions

The results are given in Tables 2 and 3. The obtained reddenings were compared to the fields colour excesses E(B-V) taken from Neckel & Klare (1980), Burstein & Heiles (1982) and FitzGerald (1968). The corresponding diagram is shown in Fig. 3 for stars in common with the former two references, while the last one is not used since only ranges in colour excesses are provided. The linear fit shown in Fig. 3 is:


with a correlation coefficient of 0.964. The standard deviation on the slope is 0.047, while 0.05 is found for a single ordinate estimate. The first bisector falls within the error domain in the diagram. This good agreement lends support to our approach, making use of the additional SCV-SED whose spectral shape differs markedly from the CVi-SEDs.

[FIGURE] Fig. 3. A comparison of E(B-V) excesses from the maps and graphs of Neckel & Klare (1980, NK80) and Burstein & Heiles (1982, BH82) with values from the present paper (KB) for CS and SC stars. The regression line (9) is also shown.

4.4. Discussion

Having completed the above analysis, we searched for independent arguments in favour of the grouping obtained in our study. First of all, Catchpole & Feast (1971) defined the SC stars as the ones closely similar in spectroscopic features to UY Cen the brightest member of this class. Out of our 13 SCV stars (Table 3), 10 at least belongs to their SC-class. They stated that those spectra are readily distinguished from the spectra of CS stars like R CMi, R Ori and W Cas (5 out of the 7 stars in Table 2 are classified as spectroscopic CS stars). The ZrO bands are absent or very weak in the latter stars, and CN much stronger than in the spectrum of UY Cen.

This feature is confirmed by the [CN]-index of Baumert (1972), an estimate of the intensity of the red system of the CN bands as observed in the red-near IR part of the spectrum. The values of the [CN]-index for the CS stars range from 68 to 78 in Table 2, which is very close to the 71-87 range of the mean values for the carbon variables which are classified as HC5, CV1 or CV2 stars (the range is 84-98 for CV3 to CV6 stars). The values quoted in Table 3 for the SC stars range from 29 to 64, the S stars displaying still lower values. Consequently, the stars in Table 2 can be found in the catalogue of carbon stars (Stephenson 1989) while only 2 stars in Table 3 are found in it, precisely those with the highest [CN]-values (64 for C797 = V346 Aur and 60 for C4152 = VX Aql). The transition between the CS and SC-classes probably falls in the 64-68 range of the [CN]-index. This is the transition between the SCV and carbon-rich SEDs. As can be seen in Tables 2 and 3, our CS stars are predominantly Miras (5 out of 7 stars) while our SC stars are predominantly semi-regular and irregular variables.

Finally, the remaining star (HD 121447, [CN]=26) is also classified as a BaII star (number 228 in Lü 1991). We found in our analysis the group to be K5g (i.e. an oxygen-type SED: see Bergeat & Knapik 1997) with E(B-V)=0.05; According to Jorissen et al. (1985), this is the coolest barium star, and this spectroscopic binary with [FORMULA] might be an elliptical variable: IT Vir with V=7.82-7.86 in the 73rd name-list of variable stars (Kazarovets & Samus 1997).

According to Catchpole & Feast (1971), both classes could be closely related and possibly indicate the existence of a continuous sequence between the C and S giants. Our analysis proves that the CS and SC stars should be distinguished. The former stars exhibit SEDs close to those of carbon variables (CV) while the latter ones display a SED intermediary between those of C and S stars we name SCV. Its colour indices explore the same ranges as the CV-stars but the resulting SED is markedly different from any of the CVi-distributions. Thus we are unable at present to propose an estimate of the corresponding effective temperatures.

An additional remarkable feature is the following: we found no IR excess relative to the SCV-solution, starting from the filter H, K or L.., in the diagrams of the SCV stars. Weak or absent IR excesses are thus indicated in SCV stars, except possibly in the far IR. Such excesses are eventually observed with respect to CV-solutions in the diagrams of CS stars as it is the case for normal C stars. They are attributed to thermal emission from CS carbonaceous grains. Our CS stars are predominantly Miras while the SCV stars are usually non-Miras.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: March 10, 1999