GEPI

Being so close and so bright with respect to other stars, one could expect that the Sun has no more secrets for us. Naively, one could think that the solar chemical composition is well known, at least for elements that leave a trace in solar spectrum or are enclosed in meteorites. But this is not the case ; for many elements there is still a debate on which is the solar abundance. Among these debated elements, oxygen is the most important example, and as a consequence one of the most studied elements.

After hydrogen and helium, oxygen is the most abundant element in the Universe, and its abundance has been extensively studied in the Galaxy and beyond. In such studies, the solar oxygen abundance serves as natural reference. The solar oxygen abundance has repercussions on solar and stellar physics. For instance, oxygen is a major contributor to the opacity in the convective envelope of the Sun. As such, the solar oxygen abundance has a direct impact on the internal structure and evolution of the Sun and solar-like stars. However, the spectroscopic determination of the solar oxygen abundance is not an easy task : very few atomic lines are available in the solar photospheric spectrum, and most of them are blended with lines of other elements. Unfortunately, the meteoritic oxygen abundance cannot be used for guidance either, since oxygen — as rather volatile element — was only incompletely condensed during the cooling of the proto-solar nebula. Much effort has been devoted to spectroscopic determination of the photospheric oxygen abundance without having led to a convergence on a definite value.

The solar oxygen abundance determinations of the last 40 years are shown in Figure 1. After a phase of « concordance », the oxygen abundance experienced a severe downward trend over the last ten years. This prompted Ayres et al. (2006) to remark jokingly that in view of this trend the Sun could be expected to have no more oxygen left around 2015. Joking aside, a low photospheric oxygen abundance is incompatible with the structure of the solar interior inferred from helioseismology.

In an effort to contribute to the dispute of whether the solar oxygen abundance is « high » or « low », a team of researchers at Paris Observatory in collaboration with others re-derived the solar photospheric oxygen abundance independently of previous analyses. For this purpose, they used the best solar spectra available and employed a realistic 3D hydrodynamical model of the solar atmosphere, computed with the CO5BOLD code. In contrast to a traditional 1D model atmosphere, a 3D simulation provides an ab initio physical description of the convective energy transport and, as a consequence, a self-consistent model of the 3D temperature structure and velocity field in the solar surface layers where the spectral lines are formed. In Figure 2 the (horizontally averaged) CO5BOLD model (solid line) is compared to the Asplund et al. (2004) 3D model (dashed line), and to the semi-empirical Holweger-Müller model.

As a result of their analysis, the authors find an oxygen abundance in the range A(O) = 8.73 — 8.79, encompassing the value obtained by Holweger (2001), and being somewhat higher than the result obtained by Asplund et al. (2004). In Figure 1, the green asterisk represents the measurement of Holweger (2001), the blue one the value of Asplund et al. (2004), and the anti-trend red point shows the present oxygen abundance determination.

The present detailed analysis reveals that the decrease in the derived solar oxygen abundance from the value of about A(O)=8.9 to the one of Holweger (2001), A(O) = 8.73, is due to an improvement in the atomic data and a result of taking into account deviations from local thermodynamic equilibrium (NLTE effects). The further lowering of A(O) due to Asplund et al. (2004) was attributed to the use of a 3D hydrodynamical solar model. In the present work we could not confirm this result : according to this analysis, the oxygen abundance from the 3D model is slightly higher than the one obtained from a reference 1D model. The main differences found in comparing the present analysis to the one by Asplund et al. (2004) are :

• the measurement of the equivalent widths of the spectral lines, which are larger in the present analysis, implying a larger oxygen abundance ;
• the assumptions in the calculations of the NLTE-corrections, which come out smaller now ;
• differences in the 3D model atmosphere (see Fig. 2).

The recommended solar oxygen abundance is A(O)=8.76, implying a solar metallicity in the range Z=0.014 - 0.016, depending on the choice for the abundance of the other elements, mainly carbon and nitrogen. The authors consider Z=0.015 to be the most likely value, to be compared to the Asplund et al. (2004) value of Z=0.012. These solar metallicity values obtained from spectroscopic abundance determinations must be confronted with the results obtained from helioseismology, such as Z=0.0172 of Antia & Basu (2006) or Z=0.016 of Basu & Antia (2008). We conclude that the new result happens to be in much better agreement with helioseismology than the one by Asplund et al. (2004), but the oxygen abundance is still a bit too low to put the solar `oxygen crisis’ to an end.

Reference

• The photospheric solar oxygen project : I. Abundance analysis of atomic lines and influence of atmospheric models
• Elisabetta Caffau (GEPI), Hans-G. Ludwig (GEPI, CIFIST), Matthias Steffen (AIP), Thomas R. Ayres (CASA), Piercarlo Bonifacio (GEPI, CIFIST, INAF-OAT), Roger Cayrel (GEPI), Bernd Freytag (CRAL), Bertrand Plez (GRAAL),
• Astronomy & Astrophysics, in press

Contact

• Elisabetta Caffau (Observatoire de Paris, GEPI, et CNRS)