Line formation in solar granulation - VII. CO lines and the solar C and O isotopic abundances

CO spectral line formation in the Sun has long been a source of consternation for solar physicists, as have the elemental abundances it seems to imply. We modelled solar CO line formation using a realistic, ab initio, time-dependent 3D radiative-hydrodynamic model atmosphere. Results were compared with space-based observations from the ATMOS space shuttle experiment. We employed weak (CO)-C-12-O-16, (CO)-C-13-O-16 and (CO)-C-12-O-18 lines from the fundamental (Delta v = 1) and first overtone (Delta v = 2) bands to determine the solar carbon abundance, as well as the C-12/C-13 and O-16/O-18 isotopic ratios. A weighted solar carbon abundance of log epsilon(C) = 8.39 +/- 0.05 was found. We note with satisfaction that the derived abundance is identical to our recent 3D determination based on C-I, [C-I], C-2 and CH lines, increasing our confidence in the accuracy of both results. Identical calculations were carried out using 1D models, but only the 3D model was able to produce abundance agreement between different CO lines and the other atomic and molecular diagnostics. Solar C-12/C-13 and O-16/O-18 ratios were measured as 86.8(-3.7)(+3.9) (delta C-13 = 30(-44)(+46)) and 479(-28)(+29) (delta O-18 = 41(-59)(+67) ), respectively. These values may require current theories of solar system formation, such as the CO self-shielding hypothesis, to be revised. Excellent agreement was seen between observed and predicted weak CO line shapes, without invoking micro- or macroturbulence. Agreement breaks down for the strongest CO lines however, which are formed in very high atmospheric layers. Whilst the line asymmetries ( bisectors) were reasonably well reproduced, line strengths predicted on the basis of C and O abundances from other diagnostics were weaker than observed. The simplest explanation is that temperatures are overestimated in the highest layers of the 3D simulation. Thus, our analysis supports the presence of a COmosphere above the traditional photospheric temperature minimum, with an average temperature of less than 4000 K. This shortcoming of the 3D model atmosphere is not surprising, given that it was never intended to properly describe such high layers.