2000A&A...353..528O

The elemental abundances of carbon, nitrogen, and oxygen in three carbon stars (TX Psc, V Aql, and BL Ori) are determined from high-resolution infrared spectra in a self-consistent manner, with the dependence of the atmospheric structure on the chemical composition fully taken into account. For this purpose, we have constructed a grid of line-blanketed model atmospheres using a revised Band Model opacity. The derived carbon, nitrogen, and oxygen abundances show good agreement with the result obtained by Lambert et al. (1986ApJS...62..373L) in most cases. The differences between the two groups are typically within 0.1dex, which is remarkable for abundance analyses, especially for very cool stars. Since our model atmospheres and the method of the analysis are independent of their work, this result lends a support to the consistency of the analyses by the both groups. However, accurate determination of C/O ratio has turned out to be extremely difficult. The C/O ratios which are derived even with the same stellar parameters with those adopted by Lambert et al. (1986ApJS...62..373L) are somewhat larger than their results: 1.07 (TX Psc), 1.47 (V Aql), and 1.07 (BL Ori), while their results are 1.027 (TX Psc), 1.25 (V Aql), and 1.039 (BL Ori). Moreover, the resulting abundances are rather sensitive to the effective temperature and the surface gravity. It is possible that the effective temperature scale is higher by 100 - 200K than that Lambert et al. (1986ApJS...62..373L) adopted, and in that case, C/O ratios in three stars become still systematically higher than those derived by Lambert et al. (1986ApJS...62..373L), showing a contrast to their conclusion that the majority of carbon stars have C/O ratios rather close to 1. In fact, the C/O ratios derived here are 1.17 (TX Psc), 1.74 (V Aql), and 1.19 (BL Ori). A decrease of the surface gravity from logg=0.0 to -0.5 also leads to an increase of C/O ratio by 10%. In other words, given the uncertainties of stellar parameters and molecular data in addition to internal errors, it is still difficult to determine C/O ratios within accuracy of 10%, which can be translated to a difference between C/O=1.02 and 1.1. The ¹²C/¹³C ratios in N-type carbon stars derived in our previous analysis have been revised with the new model grid. The ¹²C/¹³C ratios in three stars have turned out to be 31 (TX Psc), 74 (V Aql), and 35 (BL Ori), larger by about 40% than those we obtained previously. And the C/O ratios and the ¹²C/¹³C ratios derived here fall within the range predicted by the addition of ¹²C to the atmosphere of K and M giants. It supports the scenario in which K and M giants evolve to carbon stars as ¹²C synthesized in the thermal pulse is dredged-up.