Astron. Astrophys. 342, 831-838 (1999)

3. Analysis

We adopted the same spectral synthesis method as Lèbre et al. (1998) to derive the resonance Li I line () abundances. We refer to this paper for a description of our abundance analysis assuming LTE. Only minor changes were done and are described below.

3.1. Atmospheric parameters

All the red giants that we selected for the present study have well determined spectral types and colors. In order to determine the effective temperature of our program stars we used the effective temperature scales of G and K giants given by Bell & Gustafsson (1989), Dyck et al. (1996) and Perrin et al. (1998). The spectral type, color, and effective temperature calibration for red giants derived by Perrin et al. (1998) and Bell & Gustafsson (1989) are in good agreement. For the present purpose we adopted the effective temperature scale for G8 to K5 giants derived by Perrin et al. (1998). In our sample we do not have giants cooler than K4. For giants earlier than G8 we have used the effective temperature scale given by Bell & Gustafsson (1989). The uncertainty in Teff is about K.

From data in Table 1 (Hipparcos parallaxes, spectral types, bolometric corrections from Allen 1973) we computed the luminosities and gravities of the observed stars using the relation between log g, mass (2 M, for a typical red giant (McWilliam 1990)), luminosity and effective temperature. These log gvalues are given in Table 1 and used in Sect. 3.2; they are affected only by dex if the mass of the star is between 1 and 3 M. We emphasize that the Li abundances of G and K giants derived in Sect. 3.2 are not very sensitive to uncertainties as large as 0.5 dex in log gvalues.

3.2. Spectrum synthesis and lithium abundances

The grid of model atmospheres used in Lèbre et al. (1998) and computed with the code of Asplund et al. (1997) has been extended to lower Teff and gravities. All the models were calculated for a microturbulence parameter   km s -1  and solar metallicity. We checked the consistency of these low effective temperature models by comparing them with some NMARCS models incorporating much improved line opacity (Plez, private comm., described in detail in Bessell et al., 1998). There is a very good agreement between the model structures given by these two grids. Only the coolest models differ by  K in their most external layers while the agreement is kept satisfactory for . We therefore are confident in the models we use for this analysis.

We used the same line list as in Lèbre et al. (1998). Since the stars studied in this paper are slighty cooler, we also added lines from the C2 molecule and its isotopes. Line data of the Phillips red system for these three molecules (¹²C¹²C, ¹²C¹³C, ¹³C¹³C) were predicted as in de Laverny & Gustafsson (1998). Lines of the C2 Swan system tabulated by Kurucz (private communication) were also included. The synthetic spectra of the giants were convolved to mimic the stellar rotation of each star (cf. Table 2) and then with an instrumental profile to match the resolution of the observations. We furthermore assumed for these stars evolving on the red giant branch a carbon isotopic ratio 12 C /13 C = 20 (Charbonnel 1994). Finally, the microturbulent velocity in late type giants is of the order of 2 km s^-1 (McWilliam 1990). For all our program stars we assumed 2 km s^-1 for the microturbulent velocity in computing the synthetic spectra in the Li region, and corrected it when necessary.

Table 2. Lithium abundances (Li I ) of the Li-rich stars of Table 1. Solar chemical abundances are from Grevesse et al. (1996), [Fe/H]=7.50. The abundances of the other metals have been scaled to [Fe/H] for each star.

As already described in Lèbre et al. (1998), the major source of uncertainty in this abundance analysis is due to errors in the determinations of Teff . Uncertainties on log g( dex) and the rotational and microturbulence velocities lead to a total error smaller than  dex in [Fe/H] and have almost no effect on the derived Li abundance. An uncertainty of 200 K on Teff results in a change of [Fe/H] of less than 0.1 dex and around 0.2 dex in log (Li). On another hand, considering model atmospheres with metallicity in the range +0.5 dex to -0.5 dex (extreme cases) lead to an error smaller than 0.1 dex in log (Li). Combining all these sources of errors, we find a final expected uncertainty for the derived Fe and Li abundances close to 0.2 dex. However, let's note that a 200 K error on Teff is pessimistic since this parameter is rather well constrained when fitting the several Fe I lines with different excitation energies found in the synthesized spectral range.

From an analysis of the Li region spectra of our 29 red giants (Table 1) with excess flux at 25 and or 60 microns (Zuckerman et al. 1995) we found eight stars with log (Li) larger than 1.0 (Table 2). In the remaining 22 stars the Li abundances are very low: log (Li). In Table 2 we have given the Li and [Fe/H] abundances of 15 stars derived from the spectrum synthesis calculations. From our limited sample of G and K giants with far infrared excess it appears that about 28 percent of G-K giants with circumstellar dust may have Li abundance log (Li) 1.0. All the eight stars in Table 2 with log (Li) 1.0 have 60 micron excess more than a factor of 3. The IRAS fluxes of all the eight Li-rich stars in Table 2 indicate that they have probably detached cold dust shells. In our sample the infrared excess, the amount circumstellar dust and Li abundance in the case of HD 219025 are significant.

Two stars in our sample are found to have previous Li abundance determinations. Brown et al. (1989) found HD 30834 to be a Li-rich K giant and found the Li abundance to be log (Li) = 1.8. From the analysis of our spectra of HD 30834 we derived the Li abundance log (Li) = 2.4 (Fig. 1). A difference in the atmospheric parameters and the coarse determination of abundances by Brown et al. (1989) in their survey probably explain this discrepancy.

Fig. 1. Observed (filled circles) and synthetic spectra (solid line) of four Li-rich giants.

For HD 146850 Castilho et al. (1995) derived log (Li) = 1.6. From the analysis of our spectrum of HD 146850 we derived the Li abundance log (Li) = 2.0 (Fig. 1). For HD 146850 Castilho et al. (1995) give Teff = 4000 K, log g = 1.5, [Fe/H] = -0.3, log (Li) = 1.6. We computed the synthetic spectrum of HD 146850 with the parameters used by Castilho et al. (1995) and compared it with our observed spectrum. We found that the above parameters can not match our spectrum of HD 146850 and favour our value parameters.

3.3. HD 219025

HD 219025 (m_V = 7.67, K2III) is a dustier K giant with warm and cold circumstellar dust. It is an IRAS source with large far-infrared excess at 25 and 60 microns (12 µm: 17.06 Jy, 25 µm: 10.26 Jy, 60 µm: 3.86 Jy and 100 µm: 1.7 Jy) with good quality flux flags (3) (Zuckerman et al. 1995). Whitelock et al. (1991) find strong near-IR (JHKL) excesses. They speculate on whether or not it might be a RS CVn binary or a pre-main-sequence star. Bopp and Hearnshaw (1983) found moderate Ca II H and K emission. From our observations we found strong and broad Li line (Fig. 1). The very broad absorption lines clearly indicate that HD 219025 is a rapid rotator.

From the analysis of the spectrum of HD 219025 we find log (Li) = , [Fe/H] = and the rotational velocity to be 23 km s^-1 (Table 2) (Fig. 1). The Hipparcos parallax of HD 219025 (Table 1) yields an absolute magnitude of M+0.08, which clearly indicates that HD 219025 is a red giant in the (M_V, T) plane. It is not a pre-main-sequence or T-Tau star. HD 219025 is a high galactic latitude star , and therefore the interstellar reddening is not significant. The E(B-V) is found to be 0.05, and hence the uncertainty in the absolute magnitude is of the order of 0.1 magnitude.

The evolutionary status of HD 219025 seems to be very similar to that of the Li-rich rapidly rotating K giant HDE 233517. Fekel et al. (1996) found HD 233517 to be a Li-rich K2III star with high rotational velocity (V km s^-1). They derived the Li abundance to be log (Li) = 3.3. It is an IRAS source with large far-infrared excess. The giant status of HDE 233517 is determined directly from luminosity-sensitive line ratios and is further supported by a large radial velocity (46.5 km s^-1) and small proper motion. We have not found HDE 233517 in the Hipparcos parallax catalogue. Fekel et al. (1996) suggest that the processes causing rapid rotation, large Li abundance, and infrared excess are triggered at the base of the giant branch when the convection zone reaches the rapidly rotating core of low-mass stars.

We found HD 169689 and HD 176884 also to be rapidly rotating Li-rich K giants (Table 2). The Li abundance and rotational velocities are found to be log (Li) = 1.0 and V km s^-1 and 1.2 and 15 km s^-1 respectively. Both these stars have far-infrared 60 micron excess by a factor of 3. Finally, we find HD 175492 also to be a Li-rich K giant (log (Li) = 1.3) with a rotational velocity of 3 km s^-1 (Table 2).

© European Southern Observatory (ESO) 1999

Online publication: February 23, 1999
helpdesk.link@springer.de