3. Derivation of lithium equivalent widths from low-resolution spectra
A representative segment of the low-resolution spectra used in the present work is shown in Fig. 1, in which the expected positions of the Li I 6707.8 Å doublet and of the nearby Ca I 6717.7 Å line are shown by vertical dashed lines. Fig. 2 shows an enlargement of the same spectra, trimmed to cover a small region near the Li I doublet, and expanded vertically by a factor of 5, for clarity. All the stars in the sample appear to have a spectral feature in absorption at the expected position of the Li I doublet.
We have measured the equivalent width of the line present in the low resolution spectra at the expected position of the Li I doublet, using the IRAF SPLOT task, by fitting two gaussians, one to the line identified with the Li I doublet itself, and the other to the nearby Ca I feature, which is often blended with the Li I feature. The best-fit equivalent widths are reported in Table 1, together with the difference between the true and measured (on the low-resolution spectra) distance between 6708 Å Li I feature and the 6717 Å Ca I feature. This last quantity can be used as a measure of the reliability of the identification of the feature near 6708 Å with the Li I line, as it should be significantly smaller than the spectral resolution (it indeed is for all the program stars).
We have found the measurement of such weak features (few hundreds of mÅ at most) on spectra of such low resolution to be an uncertain process. The lack of any line-free continuum in the neighborhood of the line in question makes continuum estimation a subjective process, and we estimate the uncertainty due to placement of the continuum alone to be at least some 100 mÅ. Such uncertainty is in line with the uncertainty quoted by Alcala et al. (1995), for their measurements on similar spectra, of 150 mÅ (i.e. comparable, or often larger than the equivalent width being measured. Their quoted uncertainty is independent of the spectral resolution).
All the equivalent widths measured in the low-resolution spectra for the feature at Å are higher than the true equivalent width of the Li I doublet, sometimes by several hundreds mÅ, and would, taken at face value, imply (following the criteria of the RASS-WTTS papers) that all the sources discussed here are PMS, or, more specifically, WTT stars. Yet, none of these stars shows evidence of being a WTTS when real lithium abundances are derived from high-resolution spectra of the same stars. The two later-type high-lithium sources in our sample, 1ES025012.9 and 1ES0457 01.7A, which could be suspected of being PMS stars, while certainly young, are also very close to the main-sequence, and not any longer on the Hayashi track, as unambiguously shown, on the basis of Hipparcos parallaxes, by Micela et al. (1997).
The measurement of the 6708 Å feature in low-resolution spectra of low-mass stars is thus very likely to lead to a significant over-estimate the true lithium abundance. Even worse, two sources (1ES032724.2 and 1ES104449.1) which have no measurable lithium down to less than 10 mÅ in their high-resolution spectra, appear to have a similar feature at 6708 Å as the stars with deep Li I doublets visible in their low-resolution spectra. Furthermore, there seems to be no clear relationship between the equivalent width of the 6708 Å feature measured in low-resolution spectra and the equivalent width of the Li I doublet, as the very large measurement error makes it impossible to simply subtract the (metallicity and effective temperature dependent) "foot" due to the contribution of the Fe I lines to the feature measured in the low-resolution spectrum to derive the "true" equivalent width of the Li I doublet. This is clearly illustrated in Fig. 3, which shows a scatter plot of the equivalent width of the the Li I feature measured in low-resolution spectra as a function of its "true" equivalent width.
Fig. 4 shows, superimposed, the high- and low-resolution spectra of 1ES025012.9, a high-lithium K1 dwarf. Both spectra are shifted to the rest wavelength of the spectral lines. As it is obvious from the plot, both the "Ca I " and the "Li I " feature in the low resolution spectrum are actually a blend of several spectral features, making it impossible to accurately derive the true equivalent width of the parent lines.
Fig. 5 shows the same type of spectra for 1ES104449.1, a G5 giant which has (as evident from the high resolution spectrum) no measurable lithium. Remarkably, the Fe I lines evident in the high-resolution spectrum "bunch" together, in the low-resolution spectrum, to mimic a feature at a wavelength not distinguishable (at these resolutions) from a feature containing a sizable contribution from the Li I doublet, and which would thus be confused with it in the absence of high-resolution spectra.
The strength of the 6708 Å feature visible in the low-resolution spectra of G and K stars thus appears not to be a reliable indicator of the equivalent width of the lithium doublet, and bears little relationship with the actual lithium abundance of the source. The presence of an absorption feature at 6708 Å in low-resolution spectra, should thus not, per se, be taken as an indication of the possible PMS status of a G- or K-type star. Further studies on both the low- and high-resolution spectra of large samples would be needed to asses whether, in the presence of a very accurate wavelenght solution for the low-resolution spectrum, the lithium-mimicking feature visible in Li-poor stars could be reliably distinguished from a true Li I feature. Even if this were the case, however, the problem of the associated large uncertainty in the derived equivalent width (related to the low resolution) would still stand.
3.1. M stars
M stars are likely to be easier targets for spotting WTTS sources from low-resolution spectra. In cooler stars most metallic lines (such as the Fe I lines in the region around the Li I doublet) become weak, and merge in a maze of molecular lines (mostly from TiO and from hydrides, specially MgH), forming, in a low-resolution spectrum, a pseudo-continuum against which a strong Li I doublet is likely to stand out. At the same time, for a given lithium abundance the equivalent width of the Li I doublet will be stronger, because of the lower ionization fraction of lithium at lower temperature. While we do not have available, for M stars, an extensive library of both high- and low-resolution spectra, we had a few low-resolution spectra of active M stars with no measurable lithium in their high-resolution spectra and of bona fide M-type WTTS with a strong lithium line visible in high-resolution spectra. At the low resolution discussed here (also Å) these stars appear to be easily distinguishable. The low-resolution spectra of three of these M stars are shown in Figs. 6 and 7, which are analogous to Figs. 1 and 2 shown earlier for G and K stars. The bottom star in both figures (G10221, which has been discussed by Micela et al. 1995) has no measurable lithium line in its high-resolution spectrum (and has no 6708 Å feature in its low-resolution spectrum), while the two top spectra are from bona fide WTT stars in the Sco-Cen SFR reported by Walter et al. (1994), and have clearly visible 6708 Å features. The presence of strong lithium in absorption in their spectra has been determined by Walter et al. (1994) on the basis of high-resolution spectra, and it is reported in Table 2. The low-resolution spectrum of G10221 is very similar to the spectra of the WTTS, the only visible difference in their spectra being the Li I doublet in absorption.
Table 2. The three M stars, two bona fide WTTS and an active M star with no lithium in its high-resolution spectra, for which spectra are shown in Fig. 6 and 7. The high-resolution Li I equivalent widths and rotational velocity are from Walter et al. (1994) for the two WTTS and from our own work (see Micela et al. 1995) for G10221. is defined as in Table 1.
The measured equivalent width of the Li I doublet in low- and high-resolution spectra for M stars are compatible within the large observational error of the low-resolution measurements. For M stars low-resolution spectra thus appear to be a useful tool to search for high-lithium stars. Again, however, any quantitative attempt at measuring the equivalent width of the Li I doublet on low-resolution spectra is still likely to be affected by large errors, whose amount will depend both on the spectral resolution used and on the precise spectral type and metallicity of the star being observed.
© European Southern Observatory (ESO) 1997
Online publication: October 15, 1997