SpringerLink
Forum Springer Astron. Astrophys.
Forum Whats New Search Orders


Astron. Astrophys. 353, 557-568 (2000)

Previous Section Next Section Title Page Table of Contents

5. Discussion of the abundances

In Fig. 6 we compare, in the form of a log g  vs. [FORMULA] diagram, the derived atmospheric parameters for our program stars with those of other metal-deficient C-rich stars discussed in the recent literature. The open symbols represent mildly metal-deficient C-rich stars, in the abundance range -2.0 [FORMULA] [Fe/H] [FORMULA] -1.0; the filled symbols represent very metal-poor C-rich stars with [Fe/H] [FORMULA] -2.5. Metal-poor carbon stars are found in a large region of the HR diagram, from main-sequence turnoff stars (log g [FORMULA] 3) to the very evolved cool giants (log g [FORMULA]). The interpretation of the phenomenon depends in particular on the evolutionary stage of the star. Most of these stars are rich in heavy elements from Sr to Sm. However, in the case of one star, CS 22957-27 (Norris et al. 1997b; Bonifacio et al. 1998), these elements have been found to be underabundant relative to iron.

[FIGURE] Fig. 6. log g  vs. [FORMULA] for metal-poor carbon-rich stars, with data points by Vanture (1992), Luck & Bond (1991), Kipper et al. (1996), Kipper & Jorgensen (1994), Norris et al. (1997a), Sneden et al. (1996), Bonifacio et al. (1998), Zacs et al. (1998) and our present stars.

A variety of elemental abundance distributions among the heavy elements are exhibited in metal-poor stars. For example, in the so-called CH stars, where the s-process elements are enhanced, the ratio of the heavy s elements to the light s elements (the [hs/ls] ratio in the definition of Luck & Bond 1991) is larger than in the sun - [hs/ls] is positive and may exceed 1.0 dex (Vanture 1992). In the very metal-poor star CS 22892-52 (Sneden et al. 1996), the heavy elements are enhanced, but the ratio [hs/ls] is only around 0.4 dex, because the enhancement is due to the r-process. Also, owing to the large abundance of Eu (a nearly pure r-process element), [Ba/Eu] is nearly as low as -1.0 dex. The two very metal-poor stars (a main sequence turn-off star and a subgiant star) recently observed by Norris et al. (1997a) exhibit elemental patterns suggesting an s-process production of the heavy elements. The [hs/ls] ratio in these two stars is as high as 1.5 dex, and [Ba/Eu] is around 0.7 dex, in clear contrast with the star CS 22892-52. The most likely interpretation, as suggested by the authors, is the accretion of matter from an AGB (a phase producing both carbon and s elements) companion.

In Fig. 7 we present a detailed comparison of the abundance patterns of several heavy elements in our two program stars with other C-enriched stars, in particular, the metal-poor C-rich stars analysed by Kipper & Jorgensen (1994), Kipper et al. (1996), Norris et al. (1997a), and the very-metal-poor r-process enriched star CS 22892-52 (Sneden et al. 1996; Cowan et al. 1995; Norris et al. 1997b). For clarity, we have restricted the comparison to the elements also present in our results.

[FIGURE] Fig. 7a-f. [X/Fe] abundance ratios for neutron-capture elements (Z[FORMULA]35), versus the Atomic Number Z.

The heavy-element abundance patterns of these C-rich stars exhibit a variety of characteristics. The behaviour of the metal-poor carbon stars, illustrated by that of HD 25408 (Kipper et al. 1996), compared to ours (Fig. 7b), shows that the abundance ratios between the heavy elements (in particular [hs/ls]) are quite different from our program stars. One case closer to ours is that of HD 187216 (Kipper & Jorgensen 1994), where unfortunately Eu and Dy were not measured. Note that HD 189711 (Kipper et al. 1996) exhibits a similar pattern as observed in our two program stars and has an overall heavy element overabundance of the same order (whereas HD 187216 has an even larger enhancement of these heavy species). The abundance pattern of our stars is also rather similar to the stars LP 625-44 and LP 706-7 (Norris et al. 1997a) shown in Figs. 7e,f. Finally, a comparison with CS 22892-52 (Fig. 7a) indicates how different this "r-process star" is from all the others.

In Fig. 8 we show the expected pattern of r-process and s-process enrichment, as described by Norris et al. (1997a), where our abundances are overplotted in Figs. 8a,b, compared to the same plot for CS 22892-52 and LP 625-44 (Figs. 8c,d). It is known (Sneden et al. 1996; Norris et al. 1997b) that the abundances of CS 22892-52 are well fit by a scaled solar r-process pattern (see however Goriely & Arnould 1997).

[FIGURE] Fig. 8a-d. Observed abundance pattern of the neutron-capture elements compared to the solar s- (thick line) and r-process (thin line) patterns, normalized to Ba (Z = 56), as in Sneden et al. (1996). The standard deviations were computed (see Table 8) when more than one line was measured, and the error bars are indicated accordingly.

We see that the abundance patterns of our two stars lie roughly half-way between these two predicted nucleosynthesis scenarios, at least for the heavier elemental species. The same conclusion is reached regarding the [Ba/Eu] and [hs/ls] abundance ratios as given in Table 9 - our stars show ratios intermediate between those of CS 22892-52 and the s-element stars LP 625-44 and LP 706-7. Kipper & Jorgensen's star HD 187216 is also rather well fit by an r-process (although Eu is missing in their determinations).


[TABLE]

Table 9. [Ba/Eu] and [hs/ls] computed as in Norris et al. (1997a): [[FORMULA]Ba,Ce,Nd[FORMULA]]/[[FORMULA]Sr,Y,Zr[FORMULA]]


5.1. Interpretation

At first inspection, the huge carbon enhancements found in CS 22948-27 and CS 29497-34 suggest a scenario, originally proposed by Iben (1982), which induces a very efficient production of carbon in zero-metal (Population III stars) of moderate mass. The carbon may even be violently mixed (Fujimoto et al. 1990, 1999; Hollowell et al. 1990), also producing large amounts of nitrogen, and bringing these elements to the stellar surface. This scenario, however, would predict a [FORMULA] ratio near the equilibrium value, whereas our measurements provide a significantly larger value. In addition, the luminosity of a post-He-core flash star, even in this exceptional scenario, is expected to be rather high. Also, the presence of metals has to be explained. Fujimoto et al. (1995) propose accretion from interstellar clouds as one possible process, but any stellar wind during the lifetime of these stars would make this accretion rather inefficient. Moreover, the high observed Eu/Fe ratio would restrict the accretion specifically to those clouds which themselves possess high Eu/Fe ratios - a very improbable process.

Another plausible scenario would be mass transfer from an evolved companion. A Population II binary of low metallicity may have evolved to the point that the (present-day) secondary star has accreted the mass ejected by its companion in the AGB phase, and we observe now the low mass secondary, enriched in C, N and s-elements. Some Pop. II stars are known with a relatively high Eu/Fe ratio. Norris et al. (1997a) also suggest such a transfer scenario for their two stars, which are rather similar to ours (see Figs. 7 and 8). They also find that one of their stars has indeed a variable velocity, suggesting binarity. To check this interpretation we have carefully measured the radial velocity in all the available spectra of our stars. Let us note that in our Paper I there was a mistake in the computation of the heliocentric corrections, and thus all the radial velocities of the red spectra should be replaced by the ones in Table 1. In the red (EMMI spectra for [FORMULA] [FORMULA] 5000 Å), we estimate that the radial velocity precision is about 1.5 km s-1 (the blending of lines in these spectra makes it difficult to obtain more accurate radial velocities at our resolution). In the blue spectra, the radial velocities were estimated from the position of the hydrogen lines (mainly H[FORMULA]) and the uncertainty is larger, of the order of 3 km s-1. From the resulting values given in Table 1, we see that there is no clear evidence of binarity from the radial velocities. One of the Norris et al. (1997a) stars, LP 625-44, which exhibits a rather similar abundance pattern to both of our program stars, has been shown by Ryan (1999, private communication), on the basis of recent high-resolution data, to exhibit a definite binary signature in its radial velocity data, and a period of more than 2000 days. Further long term followup of CS 22948-27, CS 29497-34, and others like them, preferably with higher resolution spectroscopy, should be carried out. Let us recall that in Sect. 4.3, we noted that the "intermediate" lithium abundance is compatible with both a strong dilution of the primordial abundance by convective dilution (as observed in Pop II giants), compensated by an enhancement caused by the transfer of Li-rich mass from the companion. Let us also recall that the "intermediate" value of the [FORMULA] ratio is similar to the intermediate value observed by Aoki & Tsuji (1997) in metal-poor CH stars (mostly known to be binaries). These authors interpret this value by some compensation between mass transfer (bringing additional [FORMULA]) and convective mixing (decreasing the [FORMULA] ratio). The lithium and carbon isotopes observed in our two stars are therefore compatible with the mass transfer scenario.

Another possible interpretation would be the evolution of a single low mass Pop. II star (with Eu already present from some previous enrichment of the ISM) through the AGB phase and the third dredge-up. This is, however, a rather short lived episode in a stellar lifetime, and it would be remarkable to have found two such similar stars in this same evolutionary stage at the same time. Furthermore, a rather high luminosity is then expected, in contradiction with the rather large values of surface gravity derived for our stars (and supported by the observed proper motions).

Finally, it is important to note that the amount of enrichment in C, N and in the heavy elements (relative to iron) for our program stars is among the highest found up to now.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: December 17, 1999
helpdesk.link@springer.de