1994PASP..106..423K

This dissertation addresses several issues concerning stellar oxygen abundances. The 7774 Å O I triplet equivalent widths of Abia & Rebolo [1989ApJ...347..186A] for metal-poor dwarfs are found to be systematically too high. I also argue that current effective temperatures used in halo star abundance studies may be ∼150 K too low. New color-Teff relations are derived for metal-poor stars. Using the revised Teff values and improved equivalent widths for the 7774A O I triplet, the mean [O/Fe] ratio for a handful of halo stars is found to be +0.52 with no dependence on Teff or [Fe/H].

Possible cosmological implications of the hotter Teff scale are discussed along with additional evidence supporting the need for a higher temperature scale for metal-poor stars. Our Teff scale leads to a Spite Li plateau value of N(Li)=2.28 ± 0.09. A conservative minimal primordial value of N(Li)=2.35 is inferred. If errors in the observations and models are considered, consistency with standard models of Big Bang nucleosynthesis is still achieved with this larger Li abundance. The revised Teff scale raises the observed B/Be ratio of HD 140283 from 10 to 12, making its value more comfortably consistent with the production of the observed B and Be by ordinary spallation. Our Teff values are found to be in good agreement with values predicted from both the Victoria and Yale isochrone color-Teff relations. Thus, it appears likely that no changes in globular cluster ages would result.

Next, we examine the location of the break in the [O/Fe] versus [Fe/H] plane in a quantitative fashion. Analysis of a relatively homogeneous data set does not favor any unique break point in the range -1.7 </= [Fe/H] </= -1.0. The red giant O dataset (from the literature) used in this analysis also indicates that [O/Fe] is constant in the halo (at least for [Fe/H] >/= -3), in agreement with the new results for halo dwarfs. We find that the gap in the observed [O/H] distribution, noted by Wheeler et al. [1989, ARAA, 27, 279], persists despite the addition of more O data and may betray the occurrence of a hiatus in star formation between the end of halo formation and the beginning of star formation in the disk. It is noted that the slope of the [O/Fe] versus [Fe/H] relation for [Fe/H] >/= -1 depends on the statistical regression utilized. Hence, alleged "observed" [O/H] - age relations, which do not use truly observed O abundances (but, rather, adopt O abundances based on Fe abundances), should be regarded with caution.

Systematic effects on O abundances derived from the 6300A [O I] and 7774A O I lines are considered next. While our Solar observations confirm the disagreement between the observed 7774A O I equivalent widths and LTE model calculations at low microns, we stress that Solar O abundance determinations made from flux spectra are in very good agreement with the meteoritic value. We find the 6300A [O I] equivalent width value appears to be uncertain for the Sun. Given this uncertainty, the inability of authors to reproduce each others' 6300A O abundances, and the results of recent quasi-two-stream calculations, we do not believe it can be readily claimed (as is usually done) that these abundances are more reliable than those derived from the 7774A O I triplet. In a sample of relatively metal-rich F and G dwarfs, we find no systematic difference between the 6300 and 7774A O abundances for Teff </= 6200-6300 K. The discrepancy may reside in the model atmospheres used in the analyses.

O abundances are also determined for 8 open clusters or moving groups. A very clear relation between cluster age and O abundance is seen; this is in stark contrast to the lack of any relation between age and Fe abundance in the same clusters. Hence, despite possible a priori objections, O abundances may prove to be a superior chronometer (as others have suggested) in the study of Galactic chemical evolution. Somewhat surprisingly, our our [O/Fe] ratios appear to be larger for the younger clusters. The O abundances in the younger clusters are significantly larger than those seen in H II regions, planetary nebulae, and supergiants. It is suggested, and supported with observational evidence, that this may be due to incomplete stellar models and the possibility that a significant fraction of O in gaseous nebulae is locked up in dust grains.

Examining our results in a broader sense, we suggest that: a) the formation of the Galactic halo was a slow process but b) did not involve the merger of independent "fragments" c) Type Ia supernovae are dominated by CO-He white dwarf systems having Fe production timescales of a few Gyr d) The hiatus between the end of halo formation and the beginning of star formation in the disk, possibly required if CO-He white dwarfs are the dominant source of Type Ia supernovae, may be confirmed as the observed gap in the [O/H] distribution e) The long timescale (10^9 - 10^10 yr) of Fe production by CO-He white dwarf systems is also seen to be consistent with the lack of any correlation between age and [Fe/H] in our open cluster sample and the large scatter present in the [Fe/H] vs. age relations for field stars.