4. Bi-modal LiBeB production
The vast majority of spallogenic Li, Be and B nuclei are produced by particles of relatively low-energy, which are just the most numerous. Now since only the SBEPs of highest energy can diffuse away from superbubbles, through the dense shell, without losing their energy through coulombian losses, the LiBeB production induced by the SBEPs outside the superbubbles is very small. Any isolated supernova exploding in the `unperturbed' ISM (i.e. far from SBs) then enriches the ambient gas with freshly synthesized C and O without being accompanied by an equivallent production of LiBeB. The gas around such a SN can thus show very low L/M ratios, unless another mechanism produces LiBeB in the same region. Several processes can be invoked for that purpose. First, the standard GCRN: the shocks created by isolated SNe accelerate CRs from the unperturbed ISM (mostly protons and -particles) which then interact with the ambient CNO. The ISM abundance of CNO being very low in the early Galaxy, the resulting LiBeB production efficiency is much smaller than in SBs. The corresponding L/M production ratios are represented in Fig. 2: they increase linearly with metallicity, as expected for GCRN.
If this were the only production mechanism of light elements in the unperturbed ISM, one should expect to find extremely low L/M ratios at very low Z. However, we have shown in Parizot & Drury (1999a,b) that most of the metal-free CRs accelerated at the shock of an isolated SN are actually confined inside the supernova remnant (SNR) during the Sedov-like phase, and interact there with freshly ejected C and O nuclei to produce significant amounts of Be and B. This means that isolated SNe also produce LiBeB locally, where it is easily mixed with the fresh CNO. We evaluated the production efficiency for this mechanism to be about one order of magnitude lower than in superbubbles. The resulting L/M ratios are then about 10 to 30 times below the most common values (obtained with the SB model), and should be considered as a lower limit for L/M ratios in halo stars (provided no depletion occurs after star formation, as can be checked from the Li abundance). This is represented by the lower horizontal line in Fig. 2.
At very low metallicity, we thus predict a bimodal production of Be and B, with SBEPs leading to a high efficiency mechanism (any of the SB models) and CRs accelerated at the shock of isolated SNe leading to a low efficiency mechanism (SNR model, Parizot & Drury, 1999a,b). This results in a bimodal distribution of the L/M ratios, as schematically shown in Fig. 3 (left). Note that the relative weight of the two `modes' depends on the fraction of stars exploding in OB associations, and the fraction of stars forming far from SBs. At higher metallicity, when the Be and B production by GCRN exceeds that of the SNR model, the distance between the peaks gets smaller, and it is hard to distinguish between bimodality and the scatter described in the previous section. This is shown in Fig. 3 (right).
The ideal picture described above would be correct if there were no mixing between the gas processed inside SBs (or their shells) and the general ISM. In practice, this is only true during the first few years of Galactic chemical evolution, when the Galaxy is still largely inhomogeneous. Later on, gas with high L/M ratios will `pollute' the gas with low L/M ratios, leading to a broad L/M distribution, rather than two distinct peaks. Therefore, data at very low-metallicity (say at ) are needed to fully test the model. Most importantly, since the Li abundance in the early Galaxy is dominated by the primordial 7Li, and thus unaffected by spallative processes, only Be and B should be underabundant in the low L/M stars. The latter should thus show normal Li abundance, and underabundant Be and B. Interestingly enough, Primas et al. (1998) reported such a behavior for the population II star HD 160617, with a deficiency of dex in B, at [Fe/H] . As recalled by the authors, no stellar depletion process can be responsible for the low B abundance observed, as any such process would deplete Li much more than Be, because of its lower nuclear destruction temperature.
We also wish to draw attention on the recent report by Boesgaard et al. (1999c) on two pairs of stars, (HD 84927, BD +203603) and (HD 94028, HD 219617), having the same stellar parameters (which limit the risk of systematic errors in the derivation of the elemental abundances) but Be abundances differing by as much as 0.3 and 0.6 dex, respectively, at metallicities around [Fe/H] and (or [O/H] and ). This amounts to "depletion" factors of respectively 2 and 4. It is still not clear whether these differences are due to variations in the Be production efficiency or to poor mixing of the SN ejecta with the gas containing the secondary elements produced by spallation (cf. Sect. 3). Additional observations at lower metallicity should allow us to draw more compelling conclusions in the next few years.
© European Southern Observatory (ESO) 2000
Online publication: April 17, 2000