Astron. Astrophys. 337, 714-720 (1998)

2. Data

There is a three-fold advantage in studying Be. First, abundant data exist over a large range of metallicities. It is a pure spallation product not contaminated by neutrino-spallation as is and or by the stellar production as . And finally, its measured abundance is not significantly altered by NLTE effects which in the case of B are difficult to estimate. For these reasons, we will focus primarily on Be.

The last decade has seen considerable progress since the early observations of Rebolo et al. (1988) and Ryan et al.  (1990) of a total of ten low metallicity halo dwarf stars, all yielding upper limits with three potential determinations of a Be abundance. Since then, there have been at least 50 new observations of 25 additional halo dwarfs (Gilmore et al. 1992, Boesgaard & King 1993, Ryan et al. 1994, Primas 1995, Rebolo et al.  1993, Garcia-Lopez, Severino, & Gomez 1995, Thorburn & Hobbs 1996, Molaro et al. 1997). We have compiled the Be data from the literature and show the Be abundances as a function of [Fe/H] in Fig. 1. The data have been combined systematically so that each point corresponds to a single star. Where multiple observations of a star are found, the Be abundances are first adjusted by taking a common set of stellar parameters (surface temperature, surface gravity and metallicity) followed by a weighted average of the different observations. When possible, we have assumed temperatures as given by Fuhrmann, Axer, & Gehren (1993). For example, we have found seven distinct measurements of HD 140283. The Be abundances range from (Be/H) = -13.25 to -12.85 with assumed surface gravities running from 3.2 to 3.56, temperatures from 5540 to 5814, and metallicities from -2.2 to -2.77. Here, we have taken g = 3.4, T = 5814, and [Fe/H] = -2.6, which reduces the range for (Be/H) to -13.11 to -12.87 and yields an average (Be/H) = -12.97 0.07 for this star.

Fig. 1. The evolution of (Be/H) and (B/H) with respect to [Fe/H]. The data points for beryllium and boron have been described in the text. The solid curves represent the reference model discussed in Sect. 3, corresponding to a mass range of Be producers between 60 - 100 , model a). The dotted curve shows the resulting evolution when the source of BeB is extended down to 8 , model b). The dashed curve shows the effect of varying the iron yields with the stellar mass (Woosley & Weaver 1995) in model b).

The large number of Be observations in low metallicity halo stars have shown that the Be/H abundance ratio increases approximately linearly with [Fe/H] up to at least one tenth of the solar metallicity. Because of the multiple observations of many of the halo dwarfs, the errors in the determined Be abundances are relatively small. In contrast, the Fe abundances are particularly uncertain, and in one case, the assumed values of [Fe/H] differ by as much as 0.6 dex. As a conservative estimate for the error in [Fe/H], we have taken 0.2 dex. A linear regression on the data for (Be/H) vs. [Fe/H] (for [Fe/H] -1) then yields

Clearly, this regression indicates a predominantly primary origin for beryllium (secondary Be would give a slope of 2 rather than 1.18 as in Eq. (1)). As yet, the data show no signs of revealing a plateau which could be interpreted as a primordial value for Be as in the case of Li (though see below for a complication on this interpretation). This is of course not a surprise since in standard big bang nucleosynthesis calculations the primordial value of Be/H is - (Thomas et al.  1993, Delbourgo-Salvador & Vangioni-Flam 1994). Also of interest, is the ratio (Be/Fe) vs. [Fe/H] as is shown in Fig. 2. Adopting a solar value of (Fe/H) = -4.465, the weighted mean of the values in Fig. 2 (again for [Fe/H] -1) is (Be/Fe) = -5.84 0.05.

Fig. 2. As in Fig. 1 for the evolution of (Be/Fe) as a function of [Fe/H]. Also shown by the dot dashed curve is the case of a variable iron yield in model a).

The boron data is taken from Duncan et al. (1997) and Garcia-Lopez et al. (1998) and is also shown in Fig. 1. For those stars in which Be observations can be found, stellar parameters were again chosen uniformly. A fit to the (NLTE) boron data for [Fe/H] -1 yields

This fit is actually somewhat flatter than what one would expect due to a simple primary explanation of the origin of B and is due to the two somewhat discrepant points at the lower metallicities. These points show a higher B abundance in part due to the NLTE corrections at low metallicity (Kiselman 1994, Kiselman & Carlsson 1996). Fig. 3 shows the ratio B/Be as a function of [Fe/H] taking Be abundances from the previously described compilation. Because of the low statistics and because of the relatively large errors in the ratio B/Be, determining a mean value from the data is difficult. Converting an average of the log values of B/Be gives B/Be, whereas a straight average of the unlogged ratio gives B/Be. Alternatively, if one assumes that the departure from a linear relationship (in their logs) between B,Be and Fe is simply statistical, assuming a linear fit to the data, gives B/Be . The data at present is clearly open to interpretation and more data particularly boron data is needed. As we will argue below more data of both B and Be at low metallicity ([Fe/H] -3) is needed to learn more about the origin of elements. Thus, in the following we will consider B/Be ratios of 10, 20, and 30.

Fig. 3. The data for B/Be as a function of [Fe/H].

© European Southern Observatory (ESO) 1998

Online publication: August 27, 1998
helpdesk.link@springer.de