3. A solution to the problem of Li-rich giants
Following their first discovery that the "7Be-mechanism" can naturally work in luminous intermediate-mass asymptotic giant branch (AGB) stars (Sackmann & Boothroyd 1992), Sackmann & Boothroyd (1999) have demonstrated that under certain conditions the same process can produce Li on the first giant branch, too. In AGB stars, 7Be freshly minted in the reaction 3He(Be is quickly mixed away to a cooler region (where Li produced in the reaction 7Be(eLi can survive) by ordinary convection whereas in RGB stars some extra-mixing is required for this.
The majority of field LIRGs have circumstellar dust shells (De la Reza et al. 1996) and a large number of additional LIRGs have been discovered among stars with IR excess. This feature seems to be the only one to distinguish the LIRGs from ordinary K giants and led De la Reza et al. (1996) to propose a scenario linking the high Li abundances in these stars to the evolution of circumstellar shells. In this scenario every low-mass red giant passes through a short phase during which some internal mechanism initiates atmospherical Li enrichment accompanied by a prompt mass-loss event. De la Reza et al. (1996) have calculated evolutionary paths (in the IRAS color-color diagram) of the detached shells and inferred that the whole cycle completes in about years, the very fast initial increase of the surface Li abundance (during the first several thousand years) being followed by the much longer period (up to years) of Li depletion.
Recently, Siess & Livio (1999) have considered an original external scenario: a red giant engulfs an orbiting body of sub-stellar mass (brown dwarf or giant planet) which has the initial abundance of Li left unprocessed. This body deposits its Li into the giant's envelope and also causes a shell ejection as a consequence of associated processes (mass accretion near the BCE where the body is expected to dissolve and subsequent thermal expansion of the overlying layers; for details see the cited paper). This scenario has an obvious disadvantage: it cannot account for Li abundances exceeding the initial one.
In this Letter we propose a combined scenario in which engulfing of a giant planet by a red giant initiates the internal "7Be-mechanism": It takes into account results of quite recent publications where for the first time extremely high Li abundances have been measured in cluster giants. These are the stars IV-101 ([Fe/H] ) in the globular cluster M 3 (Kraft et al. 1999) and T33 ([Fe/H] ) in the metal-poor open cluster Berkeley 21 (Hill & Pasquini 1999). In both cases a Li abundance of has been reported. The LIRGs IV-101 and T33 are plotted in Fig. 3 in comparison with 5 Li-normal giants from the same studies. One realizes that an episodical Li-enrichment can happen at any time on the upper-RGB, independent of the red giant's evolutionary state, thus indicating an external source. Fig. 2 supports this conclusion: Both the Li-rich giant IV-101 (open square and arrow) and another Li-normal one (open square) close to it have Na increased and O decreased and fit well to the global [Na/Fe] vs. [O/Fe] anticorrelation. At the same time extra-mixing with the parameters adjusted to reproduce this anticorrelation (Fig. 2, solid line) fails to make LIRGs (Fig. 3, lines 3a and 3b for two different values of initial Li). It appears that after having been exposed for a rather long time ( years) to the "ordinary" extra-mixing which is responsible to the Na-O-anomalies, the star IV-101 - but not the other one - experienced something which suddenly changed its extra-mixing parameters to values appropriate for Li-production.
From our model calculations we found that the "7Be-mechanism" can efficiently synthesize Li and after that maintain its high abundance for a long time only if and, even more important, only if (Fig. 3, lines 1c and 2). Denissenkov & Tout (2000) have proposed Zahn's rotation-driven meridional circulation and turbulent diffusion (Zahn 1992; Maeder & Zahn 1998) as a physical mechanism for extra-mixing in low-mass red giants. It turns out, however, that with Zahn's mechanism, values of can be obtained only as upper limits for rotation close to the Keplerian one. In a scenario with engulfing a planet such fast rotation is explained naturally as a result of transferring the planet's orbital angular momentum to the giant's envelope (Siess & Livio 1999). The next question then is how to get the correct mixing depth in this scenario.
The dashed lines in Fig. 3 are similar to those shown in Fig. 10 of Sackmann & Boothroyd (1999). They are the result of calculations under the assumption that mixing depth and rate favourable for the Li-production are constant on the upper-RGB. One of them (like our line 2) has even been used to interpret a LIRG near the RGB tip in the globular cluster NGC 362 by Smith et al. (1999). However, such a straightforward interpretation is not so simple because: (i) mixing under these conditions does not produce Na nor deplete O as is observed in IV-101; (ii) it explains neither the Li-depletion immediately following the Li-production nor the rather short time-scale for the whole cycle; (iii) it requires a very unusual, precise and long-term tuning of the mixing parameters; the tuning appears to be unusual because it assumes shallow but extremely fast mixing compared to that reproducing the [Na/Fe] vs. [O/Fe] anticorrelation; it would be more natural to expect that faster mixing should be deeper as well.
Thus we propose the following explanation of how the correct mixing depth could appear in the engulfing scenario. According to Siess & Livio (1999) the giant planet (or brown dwarf) dissolves near the BCE in a red giant. After that the rotation profile in the radiative zone takes a step-like shape with a steep increase of the angular velocity up to about a local Keplerian value at the point of deepest penetration by the planet. In the course of the subsequent evolution the HBS moves outwards in mass and after years will reach the step in the rotation profile. During a time interval of years this step will be crossing a zone where and when the Li-production becomes efficient. Thus we do not need to fix the mixing depth to a preferred value. Instead, the natural growth of the helium core assures that suitable depths will be encountered and very fast mixing (due to the planet's engulfing) produces Li during this passage (Fig. 3, line 3c).
© European Southern Observatory (ESO) 2000
Online publication: June 8, 2000