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

5. The carbon stars with IR silicate emission

5.1. Introduction

There are a few carbon stars whose low resolution IRAS spectra (LRS; IRAS Science Team 1986) do exhibit the characteristic signature of silicates at and eventually (Little-Marenin 1986, Willems & de Jong 1986 and Benson & Little-Marenin 1987). Two models have been proposed:

• a binary system, the silicate shell having been produced by an oxygen-rich giant (possibly obscured) not observed yet,

• an only giant whose atmospheric composition recently switched from oxygen-rich to carbon-rich (third thermal pulse in TP-AGB models; e.g. Straniero et al. 1997and references therein).

Skinner et al. (1990) however argue that M stars with SiC dust, C stars with IR silicate dust signature, and S stars with either grains, are all objects whose photospheric C/O abundance ratio is close to unity, the type of dust which condenses being unpredictable. We note that the SC and CS stars studied in Sect. 4 are involved as well.

5.2. Our analysis

We have collected in Table 4, the results for 7 carbon stars (or formerly classified so) with silicate infrared emission. We compare them to AC Her, the RV Tau star which also displays those signatures. Its detailed analysis was presented in Paper II. We have included C136 = W Cas, a CS Mira (studied in Sect. 3), whose LRS spectrum was commented as "noisy silicate emission" by Chan (1994). Except for AC Her, the only oxygen-type SED found here is M6g for V496 Car which had the entry 1633 in the 1973 edition of the General Catalogue of Cool Carbon Stars (Stephenson 1973) and was then rejected from the next edition on the grounds of a M5 III spectrum observed by N. Houk (Stephenson 1989, Table 2). Skinner et al. (1990) suggest that it may be an S star. The other six stars are classified as HC5, CV1 or CV2, that is late HC or early CV stars.

Table 4. The stars with IR silicate emission. The C-entries from Stephenson (1989) are given and/or a variable name from the GCVS. Our solution is the group G and the colour excess E(B-V) while E'(B-V) is the field value from the literature. Also given the excesses at IRAS wavelengths. Finally, W Cas may not be an IR silicate carbon star; (a) entry 1633 in Stephenson (1973) the first edition of the catalogue of cool carbon stars.

There is thus no doubt that their SEDs show the signatures of carbon stars. From preliminary calibrations, the corresponding range in effective temperatures should be 3600-3000 K at most. It would be interesting to enlarge the studied sample and check whether they all fall in this narrow range in our classification. Our values of the E(B-V) colour index are remarkably close to the E'(B-V) values from maps in the literature (see Table 4). Practically, no room is left for a selective circumstellar extinction. A CS emission is however present. In order to document the case, the extra-emission in the IRAS [12] and [25]-bands are quoted in Table 4, as deduced from the extrapolation of our model (group and colour excesses). The corresponding entries of W Cas are found negligible. The colour corrections were applied to the IRAS photometry and the [12]=1.5 value we obtained is consistent with the LRS spectrum re-processed by Chan (1994). The classification of this CS Mira as an IR silicate emission star is thus very questionable, and we do not consider it further. The other IRAS excesses range from about 0.5 mag. in V496 Cyg to 8.1 mag. in AC Her, with no large gap left. The large values quoted for AC Her are probably due to the marked temperature contrast between the star (group G0g, that is about 6000 K) and the dust (say less than 1800 K). The much cooler carbon star C2011 which has very large excesses at and , also shows substantial excesses in H, K and L quite similar to those of AC Her. The other stars with lower IRAS excesses show possible weak excesses or no excess at all at shorter wavelengths. The magnitude of near IR excesses are thus positively correlated with the intensity of their IR silicate features.

5.3. Discussion

The present data is not sufficient to settle a definitive conclusion. The silicate-rich envelopes seem directly associated to the observed stars, with no binarity required. The Mira W Cas which has no appreciable excess should be rejected from the sample. The RV Tau-type star AC Her is quoted only for comparison in Table 4. The data in Table 4 are consistent with the suggestion of Skinner et al. (1990) on unpredictability of dust type at It arises the question whether "dirty" or "astronomical" silicates, as invoked to model the SEDs of extreme oxygen-rich cool giants (e.g. Jones & Merrill 1976, Rowan-Robinson & Harris 1982, Lefèvre et al. 1982), may explain those near IR excesses. Except in the UV, the complex part of the index of refraction is negligible for terrestrial silicates shortward of (e.g. Pollack et al. 1973). Therefore, the complex refractive index is set to a large value (say 0.1) if dirty silicates are considered, while ad hoc wavelength-dependent values (however acceptable) are adopted for astronomical silicates. Preliminary Monte Carlo simulations of radiative transfer in spherical symmetry (such as those of Lefèvre et al. 1982) for two prescribed dust components (such as in Lorenz-Martins & Lefèvre 1994) favour carbonaceous grains as an explanation for the IR excesses which are not necessarily located at the same place as the silicate grains. Such a study is however hampered by the fact that the true spatial distribution of the dust is unknown at present. The other main conclusion is the interstellar origin of the selective extinction observed. To power the strong silicate dust signature observed in the IR, a substantial absorption increasing in the UV is expected from or eventually for basalt (Lamy 1978), if a sufficient number of silicate grains are seen on the line of sight to the star. The U-observations of our stars show no obvious deficiency relatively to our solution (essentially AC Her and C2011 which are well-documented and exhibit strong IRAS excesses). The data at hand is however insufficient to warrant UV silicate absorption is absent. Simultaneous spectrophotometric UV observations down to at least, should be secured and compared to our (extended) solutions. The CS extinction whose absorption part powers the excesses has to be:

• essentially independent of wavelength at least up to which points to large grains (radii or even larger),

• and/or strongly non-spherical in distribution (e.g. a disk or torus seen at a large inclination angle, nearly pole-on), or even patchy (blobs formation).

We reached similar conclusions in Paper II for HD 100764, AC Her and a sample of RCB-variables at maximum light.

© European Southern Observatory (ESO) 1999

Online publication: March 10, 1999
helpdesk.link@springer.de