2001AJ....122.3115K

The largest, highest-quality, and most near-homogeneously treated extant available samples of Be, B, Fe, and O abundances are analyzed on four different stellar parameter scales, considering different O abundance indicators and deriving uncertainties in their relation with the required aid of jackknife and bootstrap simulations/resampling. Despite large slope and zero-point differences, the various Fe-poor ([Fe/H]≲-1) BeB-FeO relations are found to be independent of parameter scale within the present, sometimes substantial, uncertainties. Variations in the BeB-O relations (as large as 1.12 dex/dex and 1.24 dex in slope and zero point) from differing O indicators do significantly differ; surprisingly, the largest differences are within the same parameter scale and not across different ones. The well-defined mean Be-Fe relation is Be∝Fe^1.16±0.04; the B-Fe relation is virtually identical, B∝Fe^1.17±0.08. The BeB-mean O relations show smaller dispersion than BeB-OH or BeB-O I relations alone, because of the significant reduction in parameter uncertainties, and are in remarkable agreement, indicating Be∝mean O^1.51±0.05 and B∝mean O^1.61±0.12. The latter is in good agreement with the slope (B∝O^1.39±0.08) derived for metal-rich dwarfs by Smith et al. utilizing enhanced Mg I b-f opacity and presumed reliable λ6300 [O I] and λ6158 O I features. The BeB-FeO slopes are also all in excellent agreement with the reanalysis of Garcia Lopez, who utilizes a Hipparcos-based gravity scale. The equivalence of the Be- and B-FeO slopes limits prodigious ν-process ¹¹B production at low metallicity and suggests little Galactic evolution of the B/Be ratio. The BeB-mean O slopes differ significantly from pure ``primary'' and ``secondary'' values, requiring a combination of production mechanisms. The differing behavior of [O/Fe] and [Be/Fe] with [Fe/H] seems to rule out production by accelerated CO-rich grain debris in ejecta of Type II supernovae having progenitor masses M≳8 M_☉. Instead, the data are in fine accord with near-primary/intermediate BeB-FeO slopes produced by various two-component models, including standard GCR and superbubble production. Such models with a low-energy cosmic-ray source from supernovae restricted to very large progenitor mass may be consistent with the large Be abundance in the ultra-metal-poor dwarf G64-12 found by Primas et al.; however, they predict unobserved maxima in B/Be evolution near [Fe/H]~-2, produce too much total Li at intermediate metallicity, and have been suggested to be energetically untenable. Superbubble models considering a range of supernova progenitor mass and a constant cosmic-ray source composition predict the inferred modest or flat slopes in B/Be evolution. These models face possible difficulties in reproducing any nonprimordial Be plateaus at very low [Fe/H], and not underproducing ⁶Li for [Fe/H]≲-2; additional data are required to provide firmer observational constraints. The BeB/FeO ratios do not show consistent evidence for two metal-poor populations expected from bimodal (isolated supernovae and collective supernovae in superbubbles) production mechanisms, though these signatures may be lost in the scatter or have drastically different contributing fractions. Finally, comparison of the metal-poor BeB-Fe and BeB-mean O slopes suggests that [O/Fe]∝-0.25 [Fe/H]–not constant, but not as steep as suggested in some recent analyses and in agreement with the shallow [O/Fe] increase with declining [Fe/H] suggested by King.