1999A&A...344..263K

The present work is an extension of a recent study by Knapik & Bergeat (1997A&A...321..236K) and Bergeat et al. (1999A&A...342..773B) henceforth called Papers I and II, respectively. The spectral energy distributions (SEDs) of about 440 carbon-rich stars and the interstellar extinction observed on their line of sights were analysed. The methods originally developed for Semi-Regular (SR) and Irregular (L) variables (Paper I: our groups CV1 to CV6) were then extended (Paper II) to the hot carbon (HC) stars (our groups HC0 to HC5) and related objects (RCB, BaII and HdC stars). Shortly, this is a kind of a pair method making use simultaneously of the whole SED from UV to IR. Our approach is applied here to the galactic cool carbon-rich variables which were not considered in Paper I, namely the carbon Miras and very cool non-Miras, and the CS and SC variables. The carbon Miras with infrared silicate emission are also studied. The photometric CV1 to CV6 classification scheme of paper I is implemented, and we add here a later CV7-group and a specific SCV-group which corresponds to spectroscopic SC stars. A continuous S-SC-CS-C sequence is clearly supported by our results. The carbon stars with IR silicate emission included in our study do have carbon-rich SEDs of the three consecutive groups HC5, CV1 and CV2. They stand among the relatively hot carbon variables, in the 3600-3000K range in effective temperature. The carbon Miras are satisfactorily described in this enlarged scheme. No specific extension is required since non-Miras are also found in the CV7 and SCV-groups. The derived group is however frequently phase-dependent in these large amplitude variables. Additional selective extinction of circumstellar (CS) origin is observed in variable amounts. The mean extinction law for the interstellar diffuse medium as tabulated by Mathis (1990) is shown to be relevant. It applies to both interstellar and circumstellar extinction with a possible CS neutral extinction in addition which would remain undetected here. The corresponding colour excess E(B-V) is larger at minimum light or intermediate phases than what it is at maximum light (where it can amount to zero). It is associated to large IR excesses attributed to the emission from CS dust. Long-term variations on thousands of days may be interpreted in terms of varying CS dust opacity on the line of sight. The dust influence is discussed. It is shown that scattering, if substantial on the line of sight in the observing lobe, has to be essentially wavelength-independent, i.e. due to large neutral scatterers, especially in high opacity objects like IRC +10216. Finally, with the HC0 to HC5 classification of HC stars (Paper II), we obtain a fourteen groups sequence (HC0 to HC5 and then CV1 to CV7 from the earlier one to the latest one, and SCV for SC stars apart). The number of studied stars amounts now to about 600 that is about 40stars per group on the average when the oxygen-type SEDs are subtracted. The effective temperature calibration of this classification scheme is currently in preparation.