5. Relative element abundances
5.1. Prominence plasma
Making use of the intensity maps from 1005.70 Å and 1191.64 Å, we determined the Ne/Mg ratio across the whole field of view and compared it with the 1005.70 Å intensity map, which shows the complex plasma distribution of the active region. Results are shown in Fig. 3 and, for solar X=45", in Fig. 4, where also the uncertainties are reported. It is to be noted that outside the intense central region, the and intensity is mainly due to instrument-scattered light, so that the ratio has not been calculated.
The variation of the Mg/Ne relative abundance seems to be strongly correlated with the plasma structures in the emitting source. Although the instrument resolution is not enough to resolve individual structures precisely, we find that some structures with strong and line emission in Figs. 2 and 3 indicate a normal FIP-bias in Mg/Ne varying from 1.6 to 3.2. However, in regions with strong and weak line emission the FIP-bias ranges from 3.1 to 8.8.
These values show that the complex plasma structuring of the active region can have a strong effect on element abundances. Although the FIP bias is present in all the active region, its variability seem to suggest that the plasma in each individual structure might have its own peculiar composition. Such a possibility has been already suggested by a number of authors, who report abundance variations in active regions similar to ours (see for example Young & Mason 1997; Widing & Feldman 1993, 1995; Landi & Landini 1998a). Unfortunately, the lack of sufficient spatial resolution prevents a more detailed study.
It is important to note that and ion fraction curves (as shown in Fig. 5) are quite different. This causes the relative intensity ratio between lines of these two ions to be temperature dependent, so that it is necessary to take into account the plasma electron temperature when these two ions are used for measurements of Ne/Mg relative abundance.
5.2. Active region corona
In the unstructured, hot coronal plasma we studied the low-FIP/high-FIP pairs S/Ar, Si/Ar and K/Ar and the height dependence of their intensity ratios. The FIPs for K, Si, S and Ar are 4.3, 8.2, 10.4 and 15.8 eV respectively. We have investigated experimental , and ratios, determining the FIP variation with height in the corona outside the structures. The results are displayed in Fig. 7.
The results for the K/Ar, Si/Ar and S/Ar abundances outside the structures show different behaviours.
The coronal relative abundance of S/Ar shown in Fig. 7 (top) stays almost at its photospheric value.
Si/Ar shown in Fig. 7 (middle) varies between 18 and 26 compared to its photospheric value of 9. Thus we find a FIP factor of 2 to 3 for Si/Ar in the corona. It is interesting to note that, apart from the last three positions, the Si/Ar relative abundance seems to decrease with height. The difference is slightly greater than the uncertainties, and show that the Si abundance tends to decrease relative to the Ar value.
The case of K/Ar shown in Fig. 7 (bottom) requires some more detailed discussion.
K has a very low abundance, apart from having the lowest FIP among the elements whose coronal abundance can be determined spectroscopically. Using the relative intensity of a line ratio, observed by the Solflex X-ray spectrometer on the NRL P78-1 satellite, Doschek et al. (1985) established that the abundance ratio of K/Ca=0.10. Anders & Grevesse (1989) give a photospheric abundance ratio of K/Ca =0.058 which is a factor of 1.7 lower than the ratio obtained from the flare data. Using the photospheric abundance of Ca/H=2.24, we get K/H=1.3, which we assume as the reference K photospheric abundance. In our study we find that the K/Ar relative abundance varies from 0.7 to 1.6 in the off limb plasma, showing a clear, strong FIP bias. The values of the FIP bias varies from 16 to 36, stronger than any FIP bias result quoted in the literature.
The line belongs to the nitrogen isoelectronic sequence. The CHIANTI atomic data for this ion (Landi et al. 1999; Landi & Landini 1998b) consist of an interpolation of the old Bhatia & Mason (1980) collisional and radiative transition probabilities, and not of ab-initio calculations. This leaves a greater uncertainty in the calculation of the theoretical K/Ar ratio. Future CHIANTI releases will include new K transition probabilities from the calculations of Zhang & Sampson (1999), which could help in providing a more accurate value for the K FIP bias. However, it is unlikely that uncertainties in the CHIANTI data for K are larger than a factor 2, so that the K bias is still significantly greater than Si values from our work, and also than the values usually quoted in the literature for Mg, Al, Si ions.
Recently, Kink et al. (1999) report the identification of a few and high excitation lines, measured using laboratory beam-foil techniques, and the SUMER quiet Sun spectral atlas taken above a quiet area outside the solar limb by Feldman et al. (1997).
The wavelength of the - at 994.58 Å is very close to the line measured in the present work, so that this line is expected to blend the line. Also, in the present dataset other unidentified lines are found matching the wavelengths of the - transitions reported by Kink et al. (1999), confirming the presence of the blending line at 994.58 Å.
In order to investigate the importance of the contribution to the line, we have tried to investigate these lines. The relative intensities of these transitions, as measured in the present work, are substantially different than those reported by Kink et al. (1999), so that it is not possible to estimate the contribution by simple line ratios with the other line of the - multiplet. This is probably due to the different plasma conditions in the present dataset and in the one analyzed by Kink et al. (1999).
Theoretical estimates carried out with the HULLAC code (Klapisch 1971; Klapisch et al. 1977; Liedahl et al. 1995 - courtesy of Dr. J.M.Laming) have permitted us to estimate the theoretical emissivity of these lines under the physical conditions of the present dataset. However, HULLAC results show that is not expected to provide significant intensity to . However, since these transitions have very high excitation energies, the HULLAC code could underestimate their emissivity (J.M.Laming, private communications). Consequently, these results do not rule out possible blending problems, and no definitive conclusions cannot be drawn from the present dataset on the magnitude of the K FIP bias.
5.3. Uncertainties in the FIP bias measurements
In order to calculate theoretical line ratios, we need to use not only the correct Ne and T values, but also an ion fraction dataset. This parameter represents a further source of uncertainty in our FIP measurements as well as in most diagnostic techniques involving UV-EUV lines (Gianetti et al. 2000). Thus it is important to assess the uncertainties provided by the ion fractions.
We compared the FIP biases obtained adopting in turn the two most recent different ion fraction datasets available in the literature for the ions used in the present work: Mazzotta et al. (1998), and Arnaud & Rothenflug (1985) (with K from Landini & Monsignori Fossi 1991). Ion fractions are displayed in Fig. 5 for the solar ions used for spectroscopic diagnostics in the present study. Fig. 5 shows that all the ions that we are considering present differences between the two ion fraction datasets. This results in a bias to the theoretical ratio which is purely given by the ion fraction dataset used. The uncertainty ranges between 20% and 40%, and must be added to the experimental uncertainty from measured line intensities.
© European Southern Observatory (ESO) 2000
Online publication: January 29, 2001