The chemical composition indicates IRAS 22223+4327 and IRAS 04296+3429 to be `bona-fide' post-AGB stars: they display an enrichment of carbon with C/O higher than solar and especially high overabundances of s-process elements. For both stars the predicted heavy element abundances for an exponential distribution of neutron exposures with low value of the mean neutron exposure fit the observed s-process abundances best.
Note that not all stars, thought to be in-between the AGB and PN stage of stellar evolution, have all these chemical properties. E.g. HD 133656 ([Fe/H]=-1) shows CNO enrichment, but no overabundance of the s-process-elements (Van Winckel et al. 1996b). Also SAO 239853 ([Fe/H]=-0.8--1.0) (Van Winckel 1997) and HR 4912 (Luck & Bond 1989) are thought to be post-AGB stars without an enrichment of the s-process-elements. On the contrary, other 21 µm stars like IRAS 07134+1005 ([Fe/H])=-1.17; Klochkova 1995) and IRAS 05341+0852 ([Fe/H] = -0.9; Reddy et al. 1997) do also show overabundances of s-process-elements.
The two IRAS sources IRAS 22223+4327 and IRAS 04296+3429 belong to the small group of stars displaying the unidentified 21 µm emission feature. The other best studied 21 µm sources are listed in Table 6 (Kwok et al. 1989; Hrivnak & Kwok 1991a, b; Kwok 1993; Hrivnak 1995; Justtanont et al. 1996). Some of the sources are resolved in the 8-13 µm window (Meixner et al. 1997). Note that the table does not include the new possible candidates given by Henning et al. (1996).
Summary of spectral features in the best studied 21 µm sources, emphasizing correlation with carbon.
Table 6 shows that the 21 micron sources have some common
The carbon-rich chemistry, as evidenced by the C2 circumstellar molecular lines and the photospheric analyses corroborates the suggestion of Hrivnak et al. (1989a) that carbon is a major constituent of the molecule producing the 21 µm emission feature. So far the only star that evolved hot enough to display photospheric helium lines is HD 187885 (Van Winckel et al. 1996a) which only has a very weak 21 µm feature. The effective temperature of 8000 K marks probably the region where the carriers of the feature are destroyed. High-resolution spectroscopy of more central stars and certainly accurate determinations of the effective temperatures are needed to test several theories concerning the nature of the IR features and the influence of changing properties of the stellar radiation field on them (Buss et al. 1990; Begemann et al. 1997).
So far, the study of heavy element nucleosynthesis to constrain 3rd dredge-up models, concentrated on two groups of stars: the AGB stars (mainly Carbon stars, but also M stars that display s-process enhancements) and the Ba stargroup (Ba giants, Tc poor S stars, CH stars etc.). In the latter group the s-process enhancement is generally accepted to originate from mass-transfer episodes. These objects are binaries with one component being a cool old white dwarf (WD). The chemical peculiarities were build up when the star, which is now the primary, accreted s-process element enhanced material from the companion, which then was an AGB star and now is a cool white dwarf. The s-process distribution seen in these stars is thus not a direct result of internal mixing of the Ba-star itself, but a result of a not well understood accretion process of s-process enhanced material.
The study of the s-process element distribution of AGB stars themselves is made extremely difficult by the large photospheric molecular opacity that makes the modelling difficult and the detection of several species impossible (e.g. Utsumi 1985).
A systematic study of post-AGB stars in general and of the 21 µm post-AGB stars in particular can in this debate contribute invaluable information. Indeed, not only the temperature-gravity domain of these stars makes it possible to detect a large number of s-elements by their atomic lines, but also the photospheric molecular opacity is negligible. Since the photospheric abundance is a direct result of the dredge-up process (contrary to the case of the Ba-stars) and a fair number of 21 µm stars are now known, a systematic study of the s-process distributions can be used directly to constrain evolutionary models. In this first study the s-process distribution is characerized for IRAS 22223+4327 and IRAS 04296+3429 both with mean neutron exposure and for another object IRAS 05341+0852, Reddy et al. (1997) find the . This value is certainly much lower than what might be expected from their metallicity if the 21 µm stars would follow the same metallicity-neutron exposure relation of the Ba-stars (see Fig. 1 Busso et al. 1995). It is clear that a more homogeneous study of more sources is needed to compare carefully the different groups of stars.
© European Southern Observatory (ESO) 1998
Online publication: March 30, 1998