4. CNO abundances
Spectrum synthesis calculations were carried out to derive the carbon, nitrogen and oxygen abundances. A description of the spectrum synthesis code is given in Cayrel et al. (1991) and details on the molecular data in the wavelength range treated here are given in Milone et al. (1992). The code consists of a version improved along the years of the original code for atomic lines by Spite (1967) complemented for molecular lines by Barbuy (1981).
As in Paper I, the well studied galactic supergiant Arcturus (with a temperature very similar to the program stars) was used to check the quality of the physical data of the lines used in the abundance determination. The effective parameters used for Arcturus were: (, log g, [Fe/H]) = (4300,1.5,-0.51) and =1.7 . The C, N and O abundances found for Arcturus are displayed in Table 2.
Table 2. CNO abundances
Moreover, in all the investigated regions, we checked the oscillator strengths of the atomic lines nearby the molecular features studied by fitting the Arcturus spectrum (using the elemental abundances found for Arcturus in Paper I).
Concerning the observed spectra, thanks to the overlapping between the orders, two independent spectra were available for most of the studied spectral regions. All spectra were checked and the best S/N spectrum was used.
Oxygen abundances were derived using the forbidden [OI] 6300.311 line. The oscillator strengths used for the oxygen line and the ScII line nearby are those given in SBS89, whereas the Sc abundance has been determined from other lines (Paper I).
The oxygen abundance was then derived by fitting the observed [OI] line with the line computed with different oxygen abundance values; let us note that the [OI] and the nearby ScII lines form roughly in the same atmospheric layers (Lambert et al. 1974; Barbuy 1988) and that these lines are both very little affected by overionisation.
The final abundances were obtained iteratively, by taking into account the carbon and nitrogen abundances (Sec. 4.2 and 4.3) found for each star. For the star PMMR 27, since we could not derive the carbon abundance (due to the lack of observed spectrum in the corresponding wavelength domain) we adopted [C/Fe] = -0.3 in the calculations of dissociative equilibrium, similar to the underabundance found for the other stars (Sect. 4.2). This carbon deficiency, in turn, leads to a decrease of the oxygen abundance by about 0.1 dex relative to the solar carbon-to-iron case.
The C2 (0,0)5165.24 and C2 (0,1)5635.50 bandheads of the Swan (A - X ) system were used to derive the carbon abundances. The C2 (0,1) feature is weak in all the stars for which it is available. The C2 (0,0) feature is stronger and, although blended with atomic lines and MgH, is more reliable. The values tabulated in Table 2 are thus the ones derived from the C2 (0,0) feature. In most stars, this feature gives lower abundances (by 0.1 dex), and in two stars, both determinations agree.
The carbon abundances, available for five sample stars are given in Table 2.
For three of our most recent spectra, the region at 8003 was available, enabling us to distinguish the 12 CN from the 13 CN features. Fig. 2 displays the CN features in PMMR 27, computed with [C/Fe]=-0.3, [N/Fe]=+0.1 dex and a 12 C/13 C ratio of 10 and 29. For this star, the signal to noise ratio is high and the uncertainty on the ratio is estimated to be of around 5. For the other two stars, the uncertainty is higher, around 10.
Nitrogen abundances were derived from the CN red system (A - X ) bandhead (6,2) 6478.48 . A check of several CN lines in the Arcturus spectrum indicated that this is essentially the only reliable CN feature in the wavelength region available. For the nearby atomic lines, the abundances derived in Paper I were adopted.
For PMMR 27 for which the carbon abundance was not available, [C/Fe] = -0.3 was adopted. However, as we show in Fig. 3, the resulting CN feature is strongly dependent upon the carbon abundance; this figure displays the CN feature computed with [N/Fe] = 0.1 and [C/Fe] = -0.1 and -0.3, and shows that a [C/Fe] = -0.2, leads to a [N/Fe] = +0.25. In Fig. 4 the CN feature in PMMR 145 is shown for [C/Fe]=-0.3 and [N/Fe] =0.0, 0.1, 0.2, 0.4.
The lithium abundance was derived from the 6707.8 line, taking into account the nearby atomic lines (and gf values) following Lambert et al. (1993) and the CN features. The oscillator strengths of the two ionised atomic lines Sm II and Ce II, particularly important in these low gravity stars, were checked on the Arcturus spectrum and the gf of the Sm II line was fitted with =-1.54 dex (instead of -1.04 advised by Lambert et al. 1993). Fig. 5 shows two stars of our sample (with the same temperature): the richest and the poorest in lithium. For PMMR 27, we show two lithium abundance values (0.5 and 0.6 dex), as well as (in dotted line) the variation of the CN features upon a change of +0.2 dex in carbon abundance.
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998