Although Fe II ions are about 10 times more abundant than Fe I atoms in the solar photosphere, the solar photospheric spectrum does show more than 10 times more Fe I lines than Fe II lines. More trust is generally placed in abundance derived from the dominant stage of ionization because the results hardly depend on temperature. Suitable solar Fe II lines exist, but accurate transition probabilities are available only for very few of these lines (see Sect. 4). Therefore Fe I lines, in a rather large range of excitation energies and with accurate transition probabilities, have commonly been used to derive the solar abundance of iron.
Early determinations resulted in values an order of magnitude less than the coronal and meteoritic abundances. This led to a large number of new mechanisms for producing such a large fractionation in the solar nebula and in the corona. It was finally discovered in 1969 that the transition probabilities of Fe I lines used in these determinations were erroneous i.e. too large by one order of magnitude! The history of the variations of the solar abundance of iron since its first measurement by Russell (1929) is illustrated in Fig. 1.
Recent "oscillations" of the photospheric abundance of iron are discussed in different recent papers devoted to the analysis of Fe I lines (Blackwell et al. 1984; Holweger et al. 1991; Milford et al. 1994; Blackwell et al. 1995a; Holweger et al. 1995; Blackwell et al. 1995b; Kostik et al. 1996; Anstee et al. 1997) as well as Fe II lines (Pauls et al. 1990; Holweger et al. 1990; Biémont et al. 1991; Hannaford et al. 1992; Raassen & Uylings 1998a; Schnabel et al. 1999) or both (Lambert et al. 1996).
Results obtained from Fe II lines generally agree with the meteoritic abundance (see however Sect. 4) but with uncomfortably large uncertainties approaching 25%!
Results obtained from Fe I lines using the same photospheric model (Holweger & Müller 1974) led to a debate between the Oxford group (D.E. Blackwell and co-workers) and the Kiel-Hannover group (H. Holweger, M. Kock and co-workers) as to whether the solar abundance 1 of iron is high, = 7.63 (Oxford), i.e. larger than the meteoritic value = 7.50 (Anders & Grevesse 1989; Grevesse & Sauval 1998), or low (Kiel-Hannover), i.e. in agreement with the meteorites. We have to point out here that the Oxford group uses rather low excitation Fe I lines for which the gf-values have been accurately measured at Oxford whereas the Kiel-Hannover group uses Fe I lines of higher excitation which have been accurately measured at Hannover.
The reasons for this longstanding puzzling difference are to be found (see the hereabove mentioned most recent Fe I papers) in cumulative effects on the abundance results of slight differences between the equivalent widths, the gf-values absolute scales, the microturbulent velocities and the empirical enhancement factors of the damping constants adopted by the two groups. These effects are small individually, but they are all of the same sign and together they amount to the difference between the "high" and "low" solar iron abundance. But this does not tell us who is right!
Recently, Anstee & O'Mara (1995), Barklem & O'Mara (1997) and Barklem et al. (1998a) (see also Barklem et al. 1998b) have made a decisive progress. They have computed accurate cross-sections for the broadening of s-p, p-s, p-d, d-p, d-f and f-d transitions of neutral atoms by collisions with neutral hydrogen. This allows for the first time to get rid of one of the most uncertain parameters, the so-called enhancement factor, applied for decades to the conventionnally used way to compute the collisional damping constant. Anstee et al. (1997) successfully applied their new cross-sections to the analysis of the wings of very strong solar Fe I lines, never used before for abundance diagnostics. This leads to a very accurate abundance in perfect agreement with the meteorites.
As the broadening parameter plays a non negligible role in the difference Oxford versus Kiel-Hannover, we decided to make a new analysis of a sample of Fe I lines including non blended lines for which accurate transition probabilities have been measured either at Oxford for low excitation lines or at Hannover for higher excitation lines and for which accurate damping constants can be computed from the hereabove mentioned new theory. The results are presented in Sect. 2. Unfortunately, the new damping constants do not help solving entirely the iron problem. In Sect. 3 we propose an alternative solution to this problem with a slightly modified photospheric model which we applied successfully to Fe II lines in Sect. 4. Conclusions are presented in Sect. 5.____________________________
© European Southern Observatory (ESO) 1999
Online publication: June 18, 1999