4. Chemical composition and comparison with the other R CrB stars
4.1. Elemental abundances
The derived abundances from the LTE analysis are listed in Table 1. The given errors are the formal standard deviations from the different lines; for elements with only a single line, no error is given. Errors introduced by the uncertainties in the adopted stellar parameters are given in Table 2, from which we judge the accuracy of most determined abundances to be better than 0.3 dex, except for Ca, Ni, Zn, Sr and Ba where the errors may be slightly larger. Abundance ratios such as [X/Fe] will in general be much less vulnerable to errors in the parameters.
According to Table 1 V854 Cen is metal-poor. Assuming C/He=10%, the Fe mass fraction is 0.6 dex below solar if the spectroscopic C abundance is adopted (1.4 dex with the input C abundance). In fact, the Fe/C ratio is the lowest of all analysed R CrB stars with the exception of DY Cen. Had the usual C/He=1% ratio (Lambert et al. 1998) been assumed, the derived metallicity might have been problematic considering the galactic location of the star: Z pc (assuming , Lambert et al. 1998). The latter suggests that it may belong to the thick disk population, and thus not metal-poor by a large factor. If C/He=1% is still to be preferred, V854 Cen may have acquired the metal-poorness (then 1.6 dex below solar) through chemical processes. One chemical process - the separation of dust and gas - will be investigated in detail below.
Compared to other R CrB stars, V854 Cen is only mildly hydrogen-deficient. Only the hot R CrB DY Cen with log has a higher H abundance (Jeffery & Heber 1993). The H abundance of V854 Cen is log for C/He=10% and 8.9 for C/He=1% which is the spectroscopically determined C/He value for DY Cen. The next least H-deficient R CrB star is V CrA with log for an adopted C/He=1%. Synthetic spectra of the H profile in V854 Cen is shown in Fig. 2. A solar hydrogen abundance is clearly excluded since it would require unreasonably low and log g. V854 Cen accentuates the anti-correlation between the H and Fe abundances found for R CrB and EHe stars (Heber 1986; Lambert et al. 1998), which Sakurai's object also follows (Asplund et al. 1997b).
Nitrogen when considered as a [N/Fe] ratio is slightly less overabundant in V854 Cen ([N/Fe]=1.4) than in the R CrB majority stars (mean of 14 stars [N/Fe] = 1.6). The [N/Fe] ratio in V854 Cen is roughly consistent with a complete conversion of the original CNO nuclei in a slightly metal-poor star to N through CNO-cycling. Some additional N may also have been produced subsequently by proton-capture on 12 C synthesized from He-burning. The [O/Fe] ratio for V854 Cen, which is greater than seen in the R CrB majority stars, would seem to require additional production of O through He-burning.
[Na/Fe], [Al/Fe] (see Fig. 3), [Si/Fe], [S/Fe], and to some degree [Ca/Fe], are all overabundant relative to the Sun. In particular, [Na/Fe]=1.6 is very high, which suggests that Na has been synthesized through 22 Ne(p, )23 Na; of the analysed R CrB stars only V CrA has a higher [Na/Fe]. At the same time Al should have been produced by 25 Mg(p, )26 Al, but [Al/Fe] is not unusually high compared with the other R CrB stars. A possible explanation is that the proton captures occurred in gas enriched in 22 Ne. This is quite possible as CNO-cycling converts all C, N, and O to 14 N, which at higher temperatures is converted by successive -captures to 22 Ne. Ne is destroyed in He-burning but, in a convective situation as may occur when H-rich gas is mixed into the final He-shell, some may survive and be available for conversion to Na. In steady conditions, Ne is converted to Mg in a He-shell. Unfortunately the low prevents a determination of the Ne abundance, but since [Mg/Fe] is only solar in V854 Cen the explanation seems plausible.
The ratios [Si/Fe]=1.0 and [S/Fe]=0.6 are higher than expected for a dwarf star with the metallicity of V854 Cen (i.e.[Si/Fe] [S/Fe] 0.2) but similar to what has been determined for the R CrB majority. Apparently, either Si and S have been synthesized or Fe has been depleted (Lambert et al. 1998). The slight Ca overabundance ([Ca/Fe]=0.2) is as expected for a mildly metal-poor dwarf (Edvardsson et al. 1993).
Of the Fe-group elements, Sc and Ti are overabundant while Cr has a solar abundance relative Fe: [Sc/Fe]=1.2, [Ti/Fe]=0.5 and [Cr/Fe]=0.0. In particular Sc is very overabundant, which suggests synthesis by the s -process such that the Sc abundance is raised by neutron captures on the much more abundant Ca nuclei. This is supported by the observed enhancements of the s -elements. A similar phenomenon has been observed in the related stars FG Sge (Acker et al. 1982; Gonzalez et al. 1998) and Sakurai's object (Asplund et al. 1997b). [Ti/Fe] is slightly higher than for metal-poor dwarfs, for which [Ti/Fe] is expected (Edvardsson et al. 1993), though it could be due to observational errors. Cr behaves similarly to Fe, as anticipated for a low metallicity star.
Both [Ni/Fe]=1.1 and [Zn/Fe]=1.2 (see Fig. 4) are distinctly non-solar, which cannot be attributed to an initial metal-poor composition for V854 Cen. Furthermore, the light s -process elements Y and Zr are significantly enhanced ([Y/Fe]=1.4 and [Zr/Fe]=1.0) and to lesser degree the heavy s -elements ([Ba/Fe]=0.6, [La/Fe]=0.7 and [Ce/Fe]=0.5), which are all more abundant than for metal-poor dwarfs where [s /Fe] is characteristic (Edvardsson et al. 1993). According to Malaney's (1987a) calculations of s -processing in a single exposure, the elements Ni-Ce suggest that V854 Cen has suffered a mild neutron exposure of mb-1, as shown in Table 3. Since also Ni and Zn are well fit by the predictions, the stellar atmosphere may consist predominantly of material exposed to neutrons. If instead the elemental abundances are to be explained as the result of s -processing by an exponential exposure, the derived abundances suggest mb-1 (Malaney 1987b). In this case, the observed [Ni/Fe] is not well reproduced. The estimated neutron exposure is similar to what seems appropriate for most of the R CrB stars (Lambert et al. 1998), and indicates either that the formation of an R CrB star in general produces an environment capable of mild s -processing, or that the atmospheres have retained the s -process characteristics from the previous thermally pulsing AGB-phase. The latter possibility may likely be discounted as s -process enriched AGB stars generally show a much more severe exposure to neutrons, say mb-1 (Busso et al. 1995). A final He-shell flash in a post-AGB star is clearly able to produce these abundance patterns, as demonstrated by Sakurai's object. It is more uncertain whether this is also possible in the merger of two white dwarfs.
4.2. Minority or majority status for V854 Cen?
The first survey of compositions of R CrB stars (Lambert & Rao 1994) led to the definition of the two classes: majority and minority. The latter were principally characterized by high [Si/Fe] and [S/Fe] ratios and a low spectroscopic metallicity. The minority is also distinguished by their high [Na/Fe], [Al/Fe], [Ca/Fe] and [Ni/Fe] ratios. 2
With the abundances determined here, V854 Cen is mainly located in between the three minority stars and the majority group, as shown in e.g. Fig. 3. Only in [Si/Fe] vs [Na/Fe] is V854 Cen distinctly different from the majority. In particular, the [S/Fe] ratio is as expected for the majority and far from the very high characteristic ratios of the minority. The high [Ni/Fe]=1.1 and [Zn/Fe]=1.2 (see Fig. 4) ratios are also atypical of the majority (on average 0.6 and 0.7, respectively) but typical of the minority for which both ratios show a large range. The results for V854 Cen may suggest that there is a gradual difference between the two groups introduced perhaps by varying degree of dust depletion rather than reflecting different evolutionary backgrounds.
4.3. Iron-depleted rather than iron-deficient?
The peculiar abundances relative to Fe of V854 Cen may suggest that the star was not born as metal poor as its Fe abundance indicates. Dust depletion has been proposed to explain the observed abundance patterns in several post-AGB stars (cf. Bond 1991; Lambert 1996) and Boötis stars (Venn & Lambert 1990), as well as in the hot R CrB star DY Cen (Jeffery & Heber 1993); elements that condense readily into grains are now underabundant in these stellar photospheres.
It is tempting to identify the low Fe abundance of the minority stars as the result of a dust-gas separation that either occurred in the post-AGB progenitor or is occurring in the R CrB star. The temptation is especially strong for V854 Cen which is frequently in decline with its surface obscured by dust. A high [S/Fe] ratio is readily explained as S does not easily condense. Moreover, a high [S/Fe] ratio is characteristic of those post-AGB stars for which a severe separation of dust and gas has occurred. However, the high [Si/Fe] is not naïvely expected, in particular not if the dust-gas separation occurred in the C-rich gas of an R CrB star because in such an environment a likely condensate is SiC. The [Si/Fe] ratio of the extreme minority stars DY Cen, V CrA, and V3795 Sgr greatly exceed the [Si/Fe] ratios seen in even those post-AGB stars most severely affected by the dust-gas separation. In Fig. 5 the observed depletions, defined here as the stellar abundance relative to the solar value, are shown for the minority members, as a function of the observed depletions for the Oph main cloud (Cardelli 1994). The latter cloud is taken as being representative of the ISM depletions. Note that the observed depletions using this definition do not necessarily reflect the exact amount of removed material, since the initial metallicity may not have been solar, but indicate the differential depletion between different elements.
Though there is some tendency for the abundances to follow the ISM depletions as seen in Fig. 5, the correlation is not conclusive. The differential depletion for the different elements for each star does not seem to exceed about 1.0 dex. Judging from Fig. 5, all elements, if depleted, seem to have been altered roughly by the same amount, except possibly Si, S, Ni and Zn. In the case of S and Zn this might be expected but not for Si and Ni. A further complication is that the initial abundances of, e.g., the s -process elements, such as Ni and Zn (see above), and the proton-capture elements like Na and Al, have likely been modified by nucleosynthesis. This is exemplified by a slightly greater depletion of S than Al in V854 Cen, while in the ISM S reflects the initial metallicity and Al being one of the most depleted elements.
Before drawing any definite conclusions more condensation calculations for H-deficient and C-rich environments are needed. Such a special composition may well cause significant changes in expected dust depletions compared to what is found in the H- and O-rich ISM.
© European Southern Observatory (ESO) 1998
Online publication: March 23, 1998