5.1. Lithium abundance
For the stars observed with the CAT, the lithium abundance was determined by fitting synthetic spectra to the Li I 670.78 nm and 610.36 nm lines. Carlsson et al. (1994) and Ortega-Terra (1997) showed that non-local thermodynamical equilibrium (NLTE) effects play an important role in the abundance determination based on the Li I 670.78 nm line. In our calculations, we adopted the corrections for Li abundance proposed by Carlsson et al. (1994). The other Li I line, at 610.36 nm, is not clearly identified in some of the spectra, due to a blend with Fe I lines. This Li line was used for ten sample stars; for six of them only upper limits for abundances were estimated. For the other six stars, abundances could not be derived from the Li I 610.36 nm line.
In Table 5 the adopted NLTE correction and the Li abundance obtained by using both lines are presented. Li abundances from the literature are also presented where available. Fig. 3 and Fig. 4 illustrate two examples of the synthetic spectra fitting to the observed Li I 670.78 nm line and Fig. 5 and Fig. 6 show the examples of the Li I 610.36 nm line fitting. For each log N(Li) determined by the best fit, two additional synthetic spectra are shown, which roughly indicates a confidence test. The mean error in the Li abundance estimated by this test is less than 0.2 dex, in agreement with the errors estimated in Sect. 4.
Table 5. Lithium abundances (logN(Li)) obtained in the present work and from literature for both Li I 670.78 nm and 610.36nm (when available) lines. NLTE corrections were applied to the abundances obtained using the 670.78 nm line (see text).
Note that among the stars showing a strong Li line in our survey, only one (GCSS 577) has actually a low Li abundance, despite the strong (310 m) equivalent width measured for the Li I 670.8 nm line. This is in agreement with the discussion presented in Sect. 4, showing that lower temperatures correspond to lower abundances, for a fixed equivalent width of the Li line.
5.2. Other elements
The abundance of other chemical elements were derived by using curves of growth, for cases where several lines of a species were available, and for all others by using spectrum synthesis. The errors in the abundances are estimated to be about 0.20 dex per element, mostly due to inaccuracies in determining the atmospheric parameters and oscillator strengths.
We compared our results for HD 39853 with those obtained by Gratton & D'Antona (1989). They used atmospheric parameters ( = 3900 K, log g = 1.16, and [Fe/H] = -0.5) consistent with the present work and their estimation of abundances are in good agreement with ours. Their results of oxygen and element overabundances are confirmed by us. They did not find overabundance of the observed s-elements (Zr, La), at contrast with our results of [Zr/Fe] = 0.30 and [La/Fe] = 0.25. These differences are not significant by considering the expected errors on abundances.
McWilliam (1990) determined abundances for HD 787, adopting atmospheric parameters ( = 3980 K, log g = 1.74, [Fe/H] = 0.03), which are in good agreement with those adopted in the present work. For the s-elements a solar value of was found. Based on the results for Ba, the only s-element measured in this star, we also determine a solar value for . For the elements, McWilliam (1990) found underabundances of -0.2, in rough agreement with our results of solar values or slight overabundance for some elements. The results are compatible given the mean errors in abundance determination.
In Table 6 the abundance ratios relative to iron are given. Fig. 7 shows an example of a best fit of synthetic spectra to the Mg I lines with [Mg/Fe] = 0.1 for HD 787. The list of equivalent widths measured for the lines used in the analysis is presented in Table 7 (available only in electronic form).
Table 6. Chemical abundances [X/Fe], and number of lines used in the abundance derivation.
© European Southern Observatory (ESO) 2000
Online publication: January 29, 2001