In conclusion, we have calculated the Be production associated with the explosion of a supernova in the ISM, using a time-dependent model, and confirmed the results of Parizot & Drury (1999) stating that isolated SNe cannot be responsible for the Be observed in the metal-poor stars of the Galactic halo. All the qualitative and quantitative features of the two processes investigated (i.e. acceleration of particles at the forward and the reverse shocks of an isolated supernova) have been found to conform to the analytical expectations. This includes the dependence of the Be yields on the ambient density, the evolution of the spallation rates during and after the Sedov-like phase of the SNR expansion, and the influence of the adiabatic energy losses.
The implications of these results for the Galactic chemical evolution of the light elements have been discussed in detail in Paper I. We shall only stress here that it proves very hard for theoretical models to produce the required amount of Be (and similarly 6Li and B) by isolated SNe, according to conventional shock acceleration theory. Indeed, the processes that we investigated tend to optimize the spallation efficiency, in that they either accelerate the freshly synthesized C and O or confine the EPs in an environement much richer in C and O than the surrounding ISM at this stage of chemical evolution. Shock acceleration efficiencies of order 10 percent are also about the maximum that can be expected of any acceleration process. Thinking of a process involving more energy than that released by a SN and/or a higher concentration of C and O than within a SNR is rather challenging.
One promising alternative, however, seems to be a model in which the SNe act collectively, rather than individually, as in the processes investigated in this paper. The idea is that most of the massive stars in the Galaxy are formed in associations (Melnik & Efremov, 1995) and generate superbubbles which expand owing to the cumulated energy released by several consecutive supernovae. This energy leads to strong magnetic turbulence within the superbubble, which is thought to accelerate particles in a very efficient way, according to a specific model developed by Bykov & Fleishman (1992). The interesting feature is that the interior of the superbubble is enriched by significant amounts of C and O previously ejected by stellar winds and SN explosions, so that the accelerated particle should have a primary composition (Parizot et al., 1998; Higdon et al., 1998; Parizot & Knoedlseder, 1998) and therefore be very efficient in producing Be. Moreover, the average energy imparted to the EPs by each supernova is directly related to the explosion energy, instead of only the energy in the reverse shock, as in the process 2 investigated here. Indeed, either that the particles are accelerated directly by the forward shock or that the explosion energy first turns into turbulence and a distrubution of weak secondary shocks (this will be investigated in a forthcoming paper), the total energy imparted to the EPs is expected to be about ten times larger than that assumed for process 2 above (say 10% of the explosion energy, instead of the implied by the use of the reverse shock energy). Further considering that the adiabatic losses would not apply in such a case, we predict an overall factor of about 10 to 30 on the Be yields, depending on the mixing of the ejecta with non enriched ISM within the superbubble. According to the results presented in this paper, this would be enough to account for the [Be/O] ratio observed in the metal-poor halo stars.
Apart from the problem of light element production in the early Galaxy, our calculations have shown that the situation is somewhat different whether we compare Be to Fe or O. This obviously indicates that the Galactic evolution of Fe and O are mutually inconsistent, if one uses the yields of Woosley & Weaver (1995), so that a revision of the SN models should be considered. A similar conclusion has been pointed out by Fields & Olive (1999), who observed that these theoretical yields cannot reproduce the O/Fe slope measured in the abundance diagram. Since the Be problem is found to be less serious when comparison is made with O rather than Fe, we suggest that the Fe rather than the O yields may be responsible for the Fe-O problem. Further observational and theoretical work are however needed to reach a convincing conclusion.
© European Southern Observatory (ESO) 1999
Online publication: May 21, 1999