Astron. Astrophys. 322, L29-L32 (1997)

2. Prediction of the r-element distribution of CS 22892-052

The close to solar distribution of some 16 r-elements ranging from Ba to Os observed in CS 22892-052 raises quite naturally the question: to what extent can its r-nuclidic abundance distribution be non-solar outside the observed range ? The answer to this question may come from the analysis through r-process calculations of the correlation between the production of the Ba to Os elements and the rest of the r-elements.

We first consider a random superposition of about 10 equally weighted astrophysical events with paths characterized by values in the range ¹. These events are taken at a constant temperature and at constant neutron densities evenly distributed in the range. For each event, a full time-dependent network is used to calculate the final abundances at 20 different irradiation times corresponding to 20 values of the number of neutrons captured by the seed nuclei during the r-process. These values are evenly distributed in the range (for full details of the model, see Goriely and Arnould, 1996). We are thus left with some 200 different time-dependent events. Note that these astrophysical conditions have been chosen in order to cover a large range of possible r-process paths. Paths closer to the valley of -stability, i.e. , are not included, because they require very large (and likely unrealistic) irradiation times (larger than about 30 s) in order to produce the nuclei. The nuclear mass predictions of Aboussir et al. (1995) and the -decay and one-neutron -delayed emission rates of Tachibana et al. (1990) are used in the r-process calculation.

The elemental abundance distribution in the range resulting from a random superposition of some 40 events out of the 200 defined above is shown in Fig. 1. The agreement between the calculated and observed abundance distributions is seen to be surprisingly good, almost all the observed abundances being reproduced within the error bars. Note that the predictions are independent of the mode of random selection of the 40 different events.

Fig. 1. Elemental abundances (relative to the H abundance ) in the range obtained from a random superposition of r-process events belonging to the set defined in the main text. The abundances of r-elements observed in CS 22892-052 (Sneden et al., 1996) are also represented with their corresponding error bars.

As demonstrated in Fig. 2, the good local agreement exhibited in Fig. 1 between the observations and our predictions clearly does not imply a satisfactory agreement with the global solar system content of r-nuclides. Neither the position, nor the width or height of the r-process peaks fit the solar pattern. Such a result is of course expected since the solar system r-abundance distribution can only be matched by a specific superposition of events which is far from being random (e.g. Goriely and Arnould, 1996; Goriely, 1997). Moreover, it is well known that the solar r-abundance distribution originates from r-process paths in the range. An equal contribution of lower -value r-process events tends to broaden the r-process peaks ², as well as to shift their location to lower A -value. This effect is observed in Fig. 2.

Fig. 2. Comparison between the solar system r-process abundance curve (Käppeler et al., 1989) and the distribution predicted by the random superposition of events leading to Fig. 1. Both distributions are arbitrarily normalized.

One might wonder if the conclusion drawn above that the observations in CS 22892-052 do not necessarily imply a global solar system mix of r-nuclides is not just resulting from the non perfect match of the predictions and the observations appearing in Fig. 1. A fully rigorous proof that this is in fact not the case cannot be provided. However, one way to tackle this question is to examine if the calculated distribution in Fig. 1 can be made significantly closer to the CS 22892-052 observations by replacing our choice of random events leading to Fig. 1 by a superposition of events providing the best fit to the solar r-nuclide distribution. The result of this test is found in Figs. 3 and 4. The adopted fitting procedure is described by Bouquelle et al. (1996). The isotopic fit is seen to be relatively good in the whole range and involves principally r-process events, this results being similar to the analysis of Goriely and Arnould (1996). While the predictions of Fig. 4 are by far closer to the whole solar system distribution than those of Fig. 2, it is hard to single out from a comparison between Figs. 1 and 3 which selected set of events really accounts better for the CS 22892-052 observations. We thus confirm our previous conclusion: the observations of CS 22892-052, while indeed being remarkably close to the solar pattern, do not demonstrate that the r-process heritage of that star for all the nuclei is necessarily solar.

Fig. 3. Same as Fig. 1, but the selected r-process events are those leading to Fig. 4.

Fig. 4. Same as Fig. 2, but the selected r-process events provide the best fit to the whole solar system r-nuclide abundance.

This conclusion reflects the fact that the pattern of r-abundances in the range is mainly governed by nuclear physics properties along the different r-process paths rather than by specific astrophysical situations. In order to enlighten this statement, let us assume that all the isotopes of the elements that are not bypassed by the r-process are just produced with equal abundances. In such conditions, the abundance of an element is simply proportional to the number of stable non-bypassed isotopes. This artificial and simplistic abundance distribution reproduces remarkably well the observations, as seen in Fig. 5. This result explains why the random distribution of r-process events considered in constructing Fig. 1 agrees so well with the observations. The clearest deviations between the predictions and observations are obtained in the Ba and Os regions. This is attributable to the (or/and ) and shell effects, which tend to increase the abundances with respect to those of the off-shell elements. Some deformation effects might also alter the predictions from our simplistic model. Also note that the Ba region raises some specific nuclear problems, the and shell closures making this region highly dependent on uncertainties in the nuclear mass and -decay rate predictions.

Fig. 5. Comparison of the elemental abundances in the range observed in CS 22892-052 with values obtained under the assumption that the abundance of each of these elements is proportional to the number of its stable isotopes that are not bypassed by the r-process.

Even if the patterns of abundances shown in Figs. 1 and 3 are dominated by nuclear physics properties, other agents might contribute to the deviations of the predictions from the observed abundances. In particular, it has to be recalled that the theoretical abundances do not rely on any astrophysical background. Moreover, some contribution to the CS 22892-052 abundances from the s-process cannot be excluded, especially in the Ba region ³. In this respect, it is of interest to mention the analysis of the classical metal-poor subgiant HD 140283 ([Fe/H]=-2.7) by Magain (1995), who concludes that its Ba isotopic composition is fully compatible with an s-process origin. Similarly, the ultra-metal-poor stars CS 22948-27 ([Fe/H]=-3.2) and CS 29497-34 ([Fe/H]=-3.5) show large Ba and La overabundances with respect to Eu (Barbuy et al., 1997). This result might also reflect an s-process contribution to the heavy elements content of very metal-poor stars.

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de