3. Li abundances
Lithium abundances were derived using the curves of growth of Soderblom et al. (1993a). The contribution of the 6707.44 Fe line to the Li blend was subtracted using the empirical correction of the same authors. In a few cases we also performed spectral synthesis of a 20 Å region centered at Li, using an updated version of the synthetic code by Gratton and Sneden (1990), finding results that were in excellent agreement with those obtained using the Soderblom et al. (1993a) curves of growth and empirical correction for the Fe blend. Such an agreement is to be expected since both analysis are based on similar codes and on the Bell et al. (1990) model atmosphere. Balachandran (1995) pointed out how a simplified treatment of the Li doublet may result in overabundance estimates for large Li equivalent widths. Since our synthetic code uses a list of lines where the two components of the Li doublet are taken into account separately, our abundance estimates are correct also for high Li values.
How the uncertainties on the derived Li abundances depend on stellar parameters has been extensively discussed elsewhere (e.g. Pasquini et al. 1994). It is well known that the effective temperature is the most critical parameter. In this work we have used the vs. B-V calibration of Böhm-Vitense (1981). The same calibration was also used by Balachandran (1995) in her study of Li-dip stars in M 67 and in her critical reassessment of published data of Li in cluster stars. The reddening was fixed to E(B-V)=0.04 (Cayrel de Strobel 1990). We note however that in order to study the Li scatter in a cluster the choice of the temperature scale is not important, since to this purpose absolute abundances are not needed. A comparison of measured equivalent widths will result more accurate, because it will avoid adding the uncertainties involved in passing from measured quantities to abundances. In such a relative comparison, only two sources of uncertainties are present: errors in measured equivalent widths and errors in photometry. The first can be estimated in 5 mÅ. For the photometry, the differences (standard deviation) may be as large as a few hundreds of magnitude as estimated from a comparison of the results of different authors for the same object (cfr. Figs. 2-6 in Montgomery et al. 1993). However, by using the same source of photometry for all stars as we did (see previous section), these systematic errors are minimized. Montgomery et al. (1993) showed an internal accuracy (standard deviation) of 0.003 mag for (B-V)and (V-I) and of 0.002 mag for V for the magnitude and colour ranges of our sample stars.
A few high resolution studies of Li in dwarf stars in M 67 exist in the literature (Hobbs and Pilachowski 1986, Spite et al. 1987, Garcia Lopez et al. 1988). 17 stars belong to the magnitude/colour range we are interested in. In order to combine the published data with our sample thus doubling it, we followed the same approach of Balachandran (1995), i.e. we collected the published equivalent widths and we converted them to Li abundances by using the same procedure as for our data. The resulting abundances are therefore on one scale. The photometry from Montgomery et al. (1993) was available for all but one object (I160) and this photometry was adopted as for our data. The observations taken from the literature and reanalyzed in this way are summarized in Table 2. As in Table 1, the listed equivalent widths include the contribution of the 6707.44 Å Fe line to the Li blend. When the same star was observed by more than one author, the smaller upper limit has been adopted. Three of the published stars were also observed by us: we were able to give lower upper limits for two of them (S958, S976), while for the third (S988), the published upper limit could only be confirmed.
Table 2. Reanalyzed data from the literature. Columns 1 to 8 give: 1) original identification, 2) V magnitude, 3) (B-V) colour, 4) (V-I) colour, 5) Li equivalent width, 6) effective temperature, 7) Li abundance, 8) Sanders numbers and radial velocity flag.
As already mentioned, uncertainties on the Li abundances are dominated by the uncertainty on the effective temperature whereas for warmer stars errors in the measured equivalent widths become also relevant. An error in of 150 K translates in an error in the Li abundance of 0.12 dex. Errors on the equivalent widths of 5 mÅ translates into Li uncertainties that depend on both the effective temperature and the equivalent width itself, and which are typically of 0.06 to 0.1 dex.
The correction for the 6707.44 blend varies between 7 mÅ for the hottest star to 10 mÅ for the coolest. The uncertainty in this estimate is smaller than the typical errors in the measurement of the equivalent widths.
© European Southern Observatory (ESO) 1997
Online publication: April 28, 1998