6. General discussion
Strong Li lines appear in a variety of stars. However, a reliable Li abundance can only be derived through a detailed analysis. This is illustrated in Fig. 2, where the equivalent width of Li lines are plotted against stellar atmospheric parameters effective temperature () and gravity (log g), for different values of the Li abundance.
According to predictions of the standard models of stellar evolution, Li should be strongly diluted in giant stars. The observations show that Li abundance is even lower than expected in most giant stars of the galactic disk (Brown et al. 1989). Pasquini & Molaro (1996) and Castilho et al. (2000) found the same in globular clusters, but Boesgaard et al. (1998) have studied a sample of seven sub-giants in the globular cluster M92, finding one object with high Li abundance, the remaining ones showing a dispersion covering a range of a factor 3 in abundance. Several authors have also reported the eventual occurrence of one LRG in globular clusters, where the other observed giants show no Li detection (Carney et al. 1998; Smith et al. 1999; Kraft et al. 1999).
Pilachowski (1986) studied the abundance of Li in the moderately old galactic cluster NGC 7789, revealing an exceptionally high Li abundance in one of the cluster giants. Hill & Pasquini (1999) found one Li-rich star in an old open cluster, among several normal (Li-poor) giants.
Different explanations have been discussed: differential depletion, accretion of another object (brown dwarf or planet) by giant stars, or the Li production during a particular phase on the AGB evolution.
The occurrence of Li abundance 200 times larger than the mean values indicates that either these Li-rich giants have not destroyed their original Li, or, on the other hand, there is a process of Li production during the stellar evolution. Smith & Lambert (1989, 1990) showed that stars in the asymptotic giant branch have a large abundance of Li in agreement with the theoretical predictions of Sackmann & Boothroyd (1992) for the Li production in AGB stars of intermediate mass (3 - 7 ). Recently the same authors extended the range of mass (1 - 3 ) for which the Li production is possible, by considering the "cool bottom processing" in stars of the RGB (Sackmann & Boothroyd 1999).
Determinations of 12C/13C ratios are available for a number of Li-rich giants (e.g. Brown et al. 1989; Gratton & D'Antona 1989; da Silva et al. 1995; Berdyugina & Savanov 1995). Although some Li-rich giants present 12C/13C ratios in agreement with the standard mixing models, most of them require an extra-mixing mechanism relative to the first dredge-up predictions.
Two LRG and a normal giant have been observed for Be lines in the UV range (Castilho et al. 1999; Castilho 2000), providing evidence that the observed Li has been added in a later phase: the initial Li and Be have been both depleted in the two LRG. The low Be abundances found for the 2 LRG are not in agreement with the proposition that high Li abundance observed in K giants originate from the accretion of a giant planet or a brown dwarf (e.g. Kraft et al. 1999; Siess & Livio 1999; Denissenkov & Weiss 2000). Such an accretion should also enrich the giant star with Be, while this element is very depleted as we already noted.
The abundance pattern of the Li-rich giants do not differ significantly from the solar pattern nor from star to star, within the errors. In Fig. 8 we present a plot of log N(X) normalized to the solar [Fe/H] vs. atomic number Z for all giants analysed.
In order to verify a possible correlation between depletion in grains and infrared excess, the IRAS colours of our sample have been checked. Most of the stars are located in a well defined part of the [60-25] vs. [25-12] diagram, which is coincident with the region where are located both normal and Li-rich giants showing a recently expelled dust envelope (see region III in Fig. 3 presented by de la Reza et al. 1997). The ejected mass could be mostly in the form of carbon grains, decreasing C, but depletions of Na, Al, Ca, and Si would also be expected. The observations do not support the hypothesis for a depletion pattern in connection with the mass loss in Li-rich giants, given that no clear correlation is found between IR-excess and abundances of C and other metals.
© European Southern Observatory (ESO) 2000
Online publication: January 29, 2001