2001ApJ...562..521C

Stellar evolution models that include the effect of helium and heavy-element diffusion have been calculated for initial iron abundances of [Fe/H]=-2.3, -2.1, -1.9, and -1.7. These models were calculated for a large variety of masses and three separate mixing lengths, α=1.50, 1.75, and 2.00 (with α=1.75 being the solar calibrated mixing length). The change in the surface iron abundance for stars of different masses was determined for the ages of 11, 13, and 15 Gyr. Iron settles out of the surface convection zone on the main sequence; this iron is dredged back up when the convection zone deepens on the giant branch. In all cases, the surface [Fe/H] abundance in the turnoff stars was at least 0.28 dex lower than the surface [Fe/H] abundance in giant branch stars of the same age. However, Gratton et al. recently found, based on high-dispersion spectra of stars in the globular cluster NGC 6397, that the turnoff and giant branch stars had identical (within a few percent) iron abundances of [Fe/H]=-2.03. These observations prove that heavy-element diffusion must be inhibited in the surface layers of metal-poor stars. When diffusion is inhibited in the outer layers of a stellar model, the predicted temperatures of the models are similar to those of models evolved without diffusion, while the predicted lifetimes are similar to those of stars in which diffusion is not inhibited. Isochrones constructed from the models in which diffusion is inhibited fall halfway between isochrones without diffusion and isochrones with full diffusion. As a result, absolute globular cluster ages based upon the absolute magnitude of the turnoff are 4% larger than ages inferred from full-diffusion isochrones and 4% smaller than ages inferred from non-diffusion isochrones.