The Solar Heavy Element Abundances : II. Constraints from Stellar Atmospheres

Estimates of the bulk metal abundance of the Sun derived from the latest generation of model atmospheres are significantly lower than the earlier standard values. In Paper I we demonstrated that a low solar metallicity is inconsistent with helioseismology if the quoted errors in the atmospheres models (of order 0.05 dex) are correct. In this paper we undertake a critical analysis of the solar metallicity and its uncertainty from a model atmospheres perspective, focusing on CNO. We argue that the non-LTE corrections for abundances derived from atomic features are overestimated in the recent abundance studies, while systematic errors in the absolute abundances are underestimated. If we adopt the internal consistency between different indicators as a measure of goodness of fit, we obtain intermediate abundances [C/H] = 8.44 +/- 0.06, [N/H] = 7.96 +/- 0.10 and [O/H] = 8.75 +/- 0.08. The errors are too large to conclude that there is a solar abundance problem, and permit both the high and low scales. However, the center-to-limb continuum flux variations predicted in the simulations appear to be inconsistent with solar data, which would favor the traditional thermal structure and lead to high CNO abundances of (8.52, 7.96, 8.80) close to the seismic scale. We argue that further empirical tests of non-LTE corrections and the thermal structure are required for precise absolute abundances. The implications for beryllium depletion and possible sources of error in the numerical simulations are discussed