THE BE-9 ABUNDANCES OF ALPHA-CENTAURI-A AND ALPHA-CENTAURI-B AND THE SUN - IMPLICATIONS FOR STELLAR EVOLUTION AND MIXING

We present high-resolution, high signal-to-noise ratio spectra, obtain ed at the Cerro Tololo Inter-American Observatory 4 m telescope, of th e Be II 3131 Angstrom region in the metal-rich solar analog alpha Cent auri A and its companion alpha Centauri B. Be abundances are derived r elative to the Sun in a consistent fashion via spectrum synthesis. For alpha Cen A, we find [Be/H] = +0.20 +/- 0.15, where the error reflect s random uncertainties at the 1 sigma confidence level; systematic err ors of similar to 0.1 dex are also possible. The analysis of alpha Cen B is more uncertain since inadequacies in the line list, which was ca librated with solar data, may manifest themselves in cool metal-rich d warfs. Our analysis suggests [Be/H] less than or similar to +0.05, whi ch is lower than the value of A, but not significantly so given the un certainties in the A determination alone. In order to derive a conserv atively probable and larger extreme range for the solar photospheric B e abundance, we consider various uncertainties (including those in the gf-values, continuum location, non-LTE effects, model atmospheres, an alysis codes, and contaminating blends) in its determination. We concl ude that the probable range of depletion of photospheric Be from the m eteoritic value is 0.16-0.50 dex. Our larger extreme range is 0.05-0.6 2 dex. Even a slight real depletion in solar photospheric Be itself wo uld strongly contradict the standard solar model (as does the Sun's Li depletion), suggesting the action of additional mechanisms. When coup led with the solar Li depletion of only similar to 2 dex, Be depletion would point to mixing mechanisms (possibly rotationally induced) belo w the surface convection zone acting on a timescale that is much longe r than the convective timescale. If the difference in the Be abundance s of alpha Cen A and B is real, it too would strongly suggest the acti on of additional mixing mechanisms. The study of both stars with highe r resolution data and improved atomic and molecular data is clearly im portant. We conclude that the light-element abundances of the Sun and alpha Cen A (and other solar analogs) are not grossly dissimilar. The idea that standard models and the current solar photospheric Li and Be abundance are discrepant because the Sun is a lone ''oddball'' is dou btful. We also have considered the issue of the unidentified blending feature(s) in the Be II 3131.065 Angstrom region. While a putative Mn I lambda 3131.037 feature has several favorable characteristics, we su ggest that a single significant blending feature likely lies similar t o 0.02 Angstrom blueward of this position.