Astron. Astrophys. 329, 291-314 (1998)

13. The Iron Ions

The iron ions are the most useful of the coronal ions as they retain many of their n=3 electrons at coronal temperatures and the resulting complexity of the ions ensures that there are many emission lines found in the EUV, including many potential density diagnostic line pairs. This complexity, however, also makes the iron ions difficult to model and so it is important to use solar emission line spectra as testing grounds for the atomic data as well as for plasma parameter determinations.

Brickhouse et al. (1995) have provided a comparison of theory with the SERTS-89 observations, and we refer to this work where appropriate. We note that the atomic data used by Brickhouse et al. (1995) are not identical to that found in the CHIANTI database - new calculations for Fe X and Fe XI have appeared since this work, for example - and we refer the reader to Paper I for details.

The Iron Project (Hummer et al. 1993) aims to provide improved data for all the iron ions and in particular the Fe IX-XVII ions looked at here. These data are, as yet, unavailable and so the current work summarises the "pre-Iron Project" status.

13.1. Fe IX

Fe IX is argon-like and so has a single ground level. The first excited configuration is 3p 3d and the most important EUV emission lines of Fe IX come from decays from this configuration to the ground. These are the single allowed transition at 171.07 Å, the two intercombination transitions at 217.10 Å and 244.92 Å, and the forbidden transition at 241.75 Å.

All four lines are observed in the SERTS-89 spectrum, although the 171.07 Å and 217.10 Å lines are observed in second order, while the 241.75 Å and 244.92 Å lines are close to the low wavelength end of the SERTS-89 bandpass. Additionally, the 171.07 Å line is close to the end of the SERTS-89 second order bandpass as mentioned in Brickhouse et al., creating additional uncertainties in the line intensity. Despite this, we note that there is reasonable agreement with the spectrum of Behring et al. (1976) for the intensities of the four lines.

There is no strictly density insensitive ratio between these four lines, but we show the 244.92/171.07 ratio in Table 19 as the error bars on the observed data are larger than the theoretical variation of the ratio. Clearly the ratio is discrepant with theory by a factor of between 2 and 7, with the 171.07 Å line being over-estimated by theory.

We form two density sensitive ratios that are displayed in Table 20 with predicted densities, although the large error bars on the data mean that the values suggest little more than that the three lines are in broad agreement with theory (in contrast with the 171.07 Å line), with perhaps the 217.10 Å line - seen in second order at 434.20 Å - being too weak compared to theory.

Inconsistencies with the Fe IX line intensities have been seen previously: Feldman (1992) has noted that the 241.75/244.92 ratio gave densities significantly higher than other species in flares observed by the S082A spectrograph on Skylab ; while Laming et al. (1995) point out that the SERTS-89 spectrum, the Malinovsky & Heroux (1973) quiet Sun spectrum and an EUVE spectrum of the star Procyon all have the same intensity pattern for the 171.07 Å and intercombination lines. Feldman (1992) proposed that the flare observations could be explained by high temperature transient bursts during the flare which prevent the 241.75 Å line decaying before Fe IX has ionised. However, this does not explain the allowed/ intercombination line ratios being discrepant in quiet Sun conditions.

We suggest here that the discrepancies are due to the atomic model used by Fawcett & Mason (1991) not including the 3s3p 3d and 3s3p 3d configurations which may be significant in de-populating the metastable levels of the 3s3p3d configuration. Additionally the SSTRUCT data used in calculating the transition probabilities (see Paper I) predicted a 3s3p 3d G level as being the 28th excited level of the ion. This level has no allowed transition connecting it to lower levels and so one would anticipate a significant population at coronal densities.

Bearing in mind these points, we recommend that the Fe IX line intensities should be interpreted with caution and that the 244.92/241.75 ratio should not be used for detailed density diagnostic work until further theoretical work is done on this ion.

13.2. Fe X

Four Fe X lines are reported in the SERTS-89 catalogue, the strongest being 174.52 Å which is observed in second order. This line is density insensitive relative to the first order 345.74 Å line (although there is some temperature sensitivity) and a comparison with theory is presented in Table 19. We find marginal agreement with theory, but note that we require the first order Fe XI 349.04 Å line to contribute very little to the blend. In the following section we show that this is not the case, and that it actually contributes around 50% to the blend.

The 365.57/345.74 branching ratio does not agree with observations (Table 19) and we believe that this is due to an unreported blend of the 365.57 Å line with Ne V 365.60 Å. Indeed, if we assume the extra flux in the 365.57 Å line is due to Ne V, then we find excellent agreement amongst the Ne V lines, as described in Sect. 10.2.

The line reported at 257.26 Å is actually a blend of two Fe X lines: the reported P - D transition and the P - D transition. The 345.74/257.26 ratio is density sensitive and the measured density is given in Table 19.

The Bhatia & Doschek (1995) calculations used in CHIANTI are the first to include electron excitation data for the 3s3p 3d configuration, and there are a host of weak lines excited from the metastable levels in the 3s 3p 3d configuration predicted in the 250-400 Å region. All of these are predicted weaker than the observed 345.74 Å line and many have never been observed. We draw attention to two transitions for which we do have observed wavelengths: 324.76 Å and 353.44 Å, both of which are density sensitive relative to 345.74 Å with relative intensities of around 0.1-0.3 in the density regime 10⁹ -10¹⁰ cm^-3. Such intensities make the lines potentially observable by SERTS-89, but neither are reported in the catalogue. We feel that, because of this, the Bhatia & Doschek (1995) data over-estimate the strength of these 3s 3p 3d-3s3p 3d transitions, due to limitations of the 4 configuration model for Fe X.

13.3. Fe XI

The SERTS-89 catalogue lists an unidentified line at 406.791 Å that Brickhouse et al. identify as being the 3s 3p D - 3s3p P transition of Fe XI, which is then related to the 352.67 Å line by a branching ratio. The comparison with theory presented in Table 18 shows good agreement, as do the 356.63/341.11 and 369.16/352.67 ratios. However, there is a major discrepancy for the 358.67/341.11 ratio, with the 358.67 Å line being around a factor of 3 too strong. We attribute this to blending and there are two candidates suggested by CHIANTI. One was discussed in Sect. 8.3 as being due to Si XI with a predicted wavelength of 358.65 Å and intensity of around 14 erg cm^-2 s^-1 sr^-1 ; the other is a Ne IV line at 358.69 Å with an intensity of about 5 erg cm^-2 s^-1 sr^-1 as described in Sect. 11.1. Another line, not in the CHIANTI database, but suggested by Bhatia et al. (1994), is a line of Fe XIV which we estimate from their paper to have an intensity of around 5-20 erg cm^-2 s^-1 sr^-1. With our prediction that the Fe XI component should be near 24 erg cm^-2 s^-1 sr^-1, the observed intensity at 358.67 Å of 72.6 erg cm^-2 s^-1 sr^-1 seems to be well accounted for by a sum of these four lines.

Table 18. Branching ratios for the iron ions

Table 19. Density insensitive ratios for the iron ions

Table 20. Density sensitive ratios for the iron ions

Although the 341.11/352.67 ratio shows some density sensitivity, the variation of the ratio over 10 - 10 cm^-3 is of the same order as the error bars on the SERTS-89 data, and so we list this ratio in Table 19 - good agreement is found with theory. The insensitive 352.67/188.21 ratio does not agree with theory, with the 188.21 Å line being apparently a factor of 2 too strong. This may be as a result of a blend of the second order line with a weak first order line. Note that one may have expected the strong 180.40 Å line to have been seen in the SERTS-89 spectrum in second order, however it is completely lost under the very intense Fe XVI 360.75 line.

Removing the redundancy provided by the density insensitive ratios, we are left with the two density sensitive ratios shown in Table 20. The 308.58/352.67 ratio gives a very high density which is at odds with results from other ions formed at the same temperature and may indicate that the 308.58 Å line is blended. Although the density derived from the 349.04/352.67 ratio seems more reasonable, we have not taken into account blending of the 349.04 Å line with second order Fe X 174.52 Å. A lower estimate for the 349.04 Å intensity gives a lower density, and to be consistent with other density measurements we require Fe XI to contribute at least 50% to the blend at 349.04 Å.

13.4. Fe XII

A detailed comparison of the SERTS-89 Fe XII line intensities with theory is provided in Keenan et al. (1996). We note that Keenan et al. use the electron collisional excitation calculations of Tayal & Henry (1988) for the 3s-3p transitions whereas for the CHIANTI database we retain the Flower (1977) calculations for these transitions, ensuring that we have a consistent set of data for the 3s-3p and 3p-3d transitions.

In the CHIANTI database we have used the recent laboratory identifications of Jupén et al. (1993) for some weak transitions of Fe X-XIV. One such identification was of a Fe XII line at 283.45 Å and indeed we find an unidentified line in the SERTS-89 spectrum at 283.486 Å. However, given this identification, we would then expect another line, related to 283.45 Å by a branching ratio, at around 312.29 Å. The 312.29/283.45 theoretical value is 0.42, implying a 312.29 Å intensity of 27 erg cm^-2 s^-1 sr^-1, yet this line is not reported.

Keenan et al. suggest that the Fe XII transition which CHIANTI puts at 283.45 Å is instead a weak feature at 283.70 Å (not reported in the original SERTS-89 catalogue; the fit to this feature is given in Table 24), and that the 283.49 Å line is actually due to N IV. We confirm that the 283.49 Å line is largely due to N IV, but it is still possible that Fe XII could be providing a weak component to this line. In Table 19 we give the theoretical value for the CHIANTI 283.45/338.27 ratio, which gives an indication of the likely intensity of the 283.45 Å line (the 338.27 Å intensity is 76.6 erg cm^-2 s^-1 sr^-1 ), but we note below that the 338.27 Å line blended. On account of the above discussion, we feel it is not possible to determine whether the CHIANTI 283.45 Å forms a component to the SERTS-89 283.49 Å feature, or whether it it actually occurs at 283.70 Å as suggested by Keenan et al. In the latter case, the low intensity is consistent with theory, but the wavelength disagrees with the laboratory work of Jupén et al. (1993).

Keenan et al. also suggest the identification of a line at 196.618 Å as being the 3p D - 3p3d D transition. The measured intensity of this line on the old calibration scale (that used in the Thomas & Neupert paper) is . A density insensitive ratio can be formed with the observed 219.43 Å line and a comparison with theory is shown in Table 19 - the insensitivity has been assessed over 10 - 10 cm^-3. We have agreement with theory within the error bars, giving confidence in the identification, which we give in Table 24.

In the SERTS-89 spectrum, there is a Fe XII line identified at 335.04 Å but this transition in CHIANTI appears at 335.34 Å. This is because we have used the energy levels from Jupén et al. (1993) who have used a line at 376.405 Å to derive the energy for the 3s3p D level, which gives rise to the line at 335.34 Å. We note, however, that the line at 376.405 Å is predicted to be around an order of magnitude weaker than the 335.34 Å line, which is in itself a weak line. This raises question marks over whether the 376.405 Å is really a Fe XII line, and so we will assume that the observed SERTS-89 line at 335.04 Å is in fact the identified Fe XII transition.

To take account of the weak Si XI line referred to in Sect. 8.3, we revise the 364.47 Å line intensity to 216 erg cm^-2 s^-1 sr^-1. This value is used in Tables 18 and 19.

Only one pair of lines in the observed Fe XII spectrum is related by branching ratio and we find a significant discrepancy with theory - see Table 18. As the observed ratio is lower than theory, we have to surmise that the 338.27 Å line may be blended.

Neither of the 200.41 Å and 201.12 Å lines is related to any other by a density insensitive line ratio, yet we believe that both lines are blended, by noting that taking either line relative to the 219.44 Å line yields a density diagnostic. However, neither line is predicted to be stronger than the 219.44 Å line at any density (the maximum values of the 200.41/219.44 and 201.12/219.44 ratios are 0.58 and 0.93, respectively), yet both lines are given significantly higher intensities in the catalogue.

For the 201.12 Å line, there is an identified blend with an Fe XIII line and from the following section on Fe XIII we have an estimate of the Fe XII contribution as 154 erg cm^-2 s^-1 sr^-1. This is still rather a large value and we suggest that there may be a very weak ( 1-2 erg cm^-2 s^-1 sr^-1 ) first order line which may comprise some of the feature's measured response.

Keenan et al. suggest that the 200.41 Å line is blended with a weak first order Ca VI line, but we can not confirm this through CHIANTI as Ca VI is not in the database. However, we can estimate the Fe XII contribution as at most 100 erg cm^-2 s^-1 sr^-1 and probably significantly less, compared to the SERTS-89 value of 365 erg cm^-2 s^-1 sr^-1.

Given the density insensitive 192.37/195.12 ratio in Table 19 , we infer that it is the Mn XV 384.75 Å line that dominates the blend at 384.75 Å-the second order Fe XI 192.37 Å only contributing around 3-4 erg cm^-2 s^-1 sr^-1. The 193.51/195.12 ratio disagrees with theory: we suggest that the 193.51 Å line, seen in second order, may be blended with an unknown weak first order line.

The Fe XII line at 186.88 Å is a self-blend of the reported 3s 3p D - 3s3p F transition with the unreported D - F transition, and the 186.88/195.12 ratio is an excellent density diagnostic. A S XI line is also found at 186.88 Å, but, as stated in Sect. 10.5, it is expected to contribute around 55 erg cm^-2 s^-1 sr^-1 to the blend (only 4% of the total) and so could be neglected. On the other hand, CHIANTI predicts a weak first order O III line at 373.81 Å which, although expected to have an intensity of only around 5 erg cm^-2 s^-1 sr^-1, would produce a response equivalent to a second-order line of 500 erg cm^-2 s^-1 sr^-1. Thus we will take an estimate of the Fe XII contribution as 775 330 erg cm^-2 s^-1 sr^-1, and use this to derive the density for the 186.88/195.12 ratio given in Table 20.

Within the error bars on the data, the density insensitive 364.47/195.12 ratio agrees with theory, although there is a suggestion that the second order 195.12 Å line is too weak relative to the first order 364.47 Å line. We compare this to Keenan et al. (1996) who have a theoretical value for the 364.47/195.12 ratio of 0.09, creating a far larger discrepancy with theory. This is due to the adoption by these authors of the Tayal & Henry (1988) collision strengths.

The 291.01/219.43 ratio also agrees within the error bars, but again there is a suggestion that the second order line is too weak.

With the revised estimate of the Fe XII 364.47 Å intensity mentioned above of 216 erg cm^-2 s^-1 sr^-1, we find excellent agreement between theory and observation for the 346.86/364.47 and 352.11/364.47 ratios in Table 19.

For the three density diagnostic ratios presented in Table 20 we find predicted densities of around 10 cm^-3. The smallest error bars are found for the 338.27/364.47 ratio but, as noted above, there is a suggestion that the 338.27 Å may be blended.

13.5. Fe XIII

There are 23 Fe XIII lines seen in the SERTS-89 spectrum-more than for any other ion-and there is a lot of density sensitivity due to the ground levels approaching Boltzmann equilibrium over the 10 - 10 cm^-3 regime.

The line at 191.23 Å is identified as a blend of a S XI and a Fe XIII line, however the Fe XIII line expected at this wavelength is too weak to provide any appreciable intensity and so we dismiss this identification. As noted in Brickhouse et al. (1995) the unidentified line at 312.87 Å is in fact the Fe XIII 3s 3p P - 3s3p P transition. Confirmation of this can be seen by the excellent agreement with theory of the density insensitive ratio 312.87/359.64 given in Table 19.

There are several discrepancies with the branching ratios given in Table 18, although the only major one is the 204.95/201.12 ratio. The problem here is probably with the second order 204.95 Å line being blended with a first order line at around 409.90 Å since previous first order observations of this line have always shown it to be significantly weaker than the 201.12 Å line - see, e.g., the solar spectrum of Malinovsky & Heroux (1973).

The discrepancy in the 311.56/320.80 ratio is probably caused by a Cr XII line blending with the 311.56 Å line. The Cr XII transition is 3s 3p P - 3s3p P and by analogy with the corresponding transition in Fe XIV we can estimate its contribution to be around 5 erg cm^-2 s^-1 sr^-1, bringing the observed Fe XIII ratio into better agreement with theory.

The 240.72/251.94 branching ratio was noted to be inconsistent with the observations of Malinovsky & Heroux (1973) by Flower & Nussbaumer (1974) and we note that the same discrepancy is apparent in the SERTS-89 spectrum.

Although the observed 413.00/368.16 value matches its branching ratio within the error bars, 368.16 Å is identified as a blend with Cr XIII. Using the theoretical ratio in Table 17, we can estimate the Fe XIII component of this line as 105 erg cm^-2 s^-1 sr^-1, which then implies a Cr XIII contribution of 23 erg cm^-2 s^-1 sr^-1.

The density insensitive ratios given in Table 19 have been assessed over the density range 10 - 10 cm^-3. For the second order lines (those below 222 Å) we note that a line with intensity of around 200 erg cm^-2 s^-1 sr^-1 is predicted at 200.02 Å; however, this value falls below the sensitivity limit of 373 erg cm^-2 s^-1 sr^-1 for SERTS-89 at this wavelength. The Fe XIII component to the 201.12 Å blend can be estimated from the 201.12/(202.04,203.82) ratio given in Table 19 as around 240 erg cm^-2 s^-1 sr^-1 ; the other component is primarily Fe XII with an intensity as much as 154 erg cm^-2 s^-1 sr^-1 (see in Sect. 13.4).

For the lines found above 300 Å, the three ratios found in Table 19 show reasonable agreement with theory. In comparison with the second order lines, however, we find problems. The 320.80/203.82 and 348.18/202.04 ratios are both discrepant with theory, with the second order lines around 50% lower than theory predicts. This may seem like a simple first-order/second-order calibration problem, but we note that the first order line seen at 251.94 Å agrees with the second order lines but not the 300 Å lines as witnessed by the 251.94/(202.04,203.82) and 359.64/251.94 ratios. A detailed discussion of the SERTS-89 calibration is presented at the end of this paper.

After accounting for the redundancy provided by the density insensitive ratios, we are left with the three density sensitive ratios involving lines close in wavelength and the predicted densities are presented in Table 20. While 203.82/202.04 and 359.64/348.18 agree well with each other and are consistent with other ions formed at the same temperature, the 318.12/320.80 ratio gives a higher density. A possible explanation for this discrepancy stems from the fact that the 318.12 Å line is principally excited from the D level in the ground configuration-the collision strengths from the P levels are around 1 to 2 orders of magnitude weaker. However, at densities below 10 cm^-3 the occupation numbers of the P levels are around 1 to 2 orders of magnitude higher than the D level, and so excitations from these levels will be contributing significantly to the intensity of the 318.12 Å line. To bring the 318.12/320.80 ratio in line with the other two, it is necessary that the collision strengths from the P ground levels be higher than their current values.

13.6. Fe XIV

Strong lines of Fe XIV are seen in the SERTS-89 spectrum, and they potentially yield excellent density diagnostics. The five branching ratios presented in Table 18 show good agreement with the observations apart from 257.40/270.52. As the observed ratio is less than the theoretical value, the obvious possibility is that the 270.52 Å line is blended with another line. The only candidate in the CHIANTI database is the Fe XXI 270.57 Å, which is not expected to be seen in the SERTS-89 spectrum. We note that if blending is a problem, the unknown component must have a considerable intensity, of the order of 250 erg cm^-2 s^-1 sr^-1.

Density insensitive ratios are presented in Table 19, and a major discrepancy can be seen in the 274.21/211.32 ratio. We do not expect the problem to be due to blending because the branching ratios for both lines agree well with the observations (Table 18). The 270.52/211.32 and 334.17/274.21 ratios both agree with theory, suggesting a distinction between those lines found above 274 Å and those found below 274 Å. This is further supported by the density sensitive ratios presented in Table 20; the 219.12/211.32 and 353.83/334.17 ratios both giving densities consistent with other ions formed at similar temperatures, while the 264.78/274.21 ratio suggests an extremely low density.

The resonance lines of Fe XIV have been observed in previous solar spectra and we look to see signs of an inconsistency in the 274.21/211.32 ratio. Malinovsky & Heroux (1973) give a value of (taking error bars of 20%); Behring et al. (1976) give (taking error bars of 30%); Keenan et al. (1991) give several sets of observations from the S082A instrument on Skylab with nine ratios of between 0.36 and 0.49 and another of 0.19. These latter observations seem more consistent with theory but, as noted by Bhatia et al. (1994), the S082A instrument response falls off rapidly below 220 Å and so we have to treat these values with caution. Overall, the data on the 274.21/211.32 ratio is confusing and seem to suggest that it varies with solar conditions. Observations with CDS may resolve these problems.

The discrepancy above and below 274 Å is more clearly demonstrated by the 274.21/270.52 ratio, since these are both strong lines and are close enough in wavelength so that calibration errors are minimized. As can be seen from Table 18, the CHIANTI value is , whereas the SERTS-89 active region catalogue gives . If 270.52 Å is actually a significant blend, as suggested above from branching ratio arguments, then the problem only gets much worse. Other unpublished SERTS observations using different optical components give ratios between 1.3 and 2.3 for quiet sun, off-limb, and various active regions. The Malinovsky & Heroux (1973) and Behring et al. (1976) values are and , respectively. Blaha (1971) summarizes six early full-sun measurements that found ratios ranging from 1.3 to 1.7. The ten Skylab values reported by Keenan et al. (1991) include those for three different active regions, yielding 1.2, 1.8, and 1.9. The remaining seven Skylab values are for various flares or flare sites, and show ratios between 0.73 and 0.95. While these latter observations may seem to match the CHIANTI prediction, Fe XXI 270.57 Å can become very important in flares, significantly reducing the expected 274.21/270.52 ratio. Thus, measurements of this ratio with many different instruments observing a wide variety of sources seem to be remarkably consistent, yet at serious variance with the best theoretical calculations presently available.

In addition to the above allowed transitions, three intercombination lines at 430-450 Å from ²P - ⁴P transitions are identified by Thomas & Neupert. However, the Fe XIV line predicted at 429.54 Å is several orders of magnitude weaker than the observed line and so we dismiss this identification. The remaining two lines are density sensitive and give a density in reasonable agreement with the two allowed ratios mentioned earlier, although we note the error bars on the data are considerably larger (Table 20).

Brickhouse et al. found the intensities of these ²P - ⁴P lines to be more than a factor of 3 too high compared to other Fe XIV lines. We get a similar result, but only when comparing them to lines below 274 Å, such as in the density insensitive ratio 444.24/211.32. In distinct contrast, we find the comparisons to lines above 274 Å are in good agreement with theory, as shown by the insensitive ratio 444.24/334.17 (Table 19). Clearly this is part of the same problem that has been discussed above, and we concur with Brickhouse et al. in suggesting that any problems with the ²P - ⁴P transitions merely indicate difficulties with the atomic physics for Fe XIV and do not reflect inaccuracies in the SERTS calibration.

Bhatia et al. (1994) presented new Distorted Wave calculations for Fe XIV, including for the first time the 3p and 3s3p3d configurations, and presented comparisons with the SERTS-89 spectrum. The addition of the extra configurations has resulted in predictions of intensities for lines at 290.69 Å, 358.67 Å and 363.75 Å which are present in the SERTS-89 spectrum. We dispute that the 363.75 Å line can be ascribed primarily to Fe XIV as the Mg VII line at this wavelength accounts for most of the observed intensity (see Sect. 10.3), though Fe XIV may contribute a little. The 358.67 Å Fe XIV component may be important as the intensity at this wavelength in SERTS-89 can not be accounted for by Si XI, Ne IV and Fe XI lines alone (Sects. 8.3, 11.1, and 13.3). According to the Bhatia et al. (1994) calculations, these three Fe XIV lines are expected to have intensities of around 5-20 erg cm^-2 s^-1 sr^-1.

13.7. Fe XV

The 417.25/284.16 ratio has long been acknowledged as discrepant with theory (see, e.g., Flower & Jordan 1971 and Feldman et al. 1992) and we find this the case in the SERTS-89 spectrum, although we note that Brickhouse et al. (1995) find marginal agreement with theory using the electron collision data from Pradhan (1988). The data in Table 19 show that we find the theoretical value to be around 25-50% below the observed value. The remaining insensitive ratios in Table 19 show good agreement with theory. Interestingly, it is the 327.03/284.16 ratio that agrees with theory rather than the 327.03/417.25 ratio.

The Fe XV 312.55 Å line is blended with a Co XVII line. Given the Co XVII 339.54 Å line intensity, a comparison with the iso-electronic transitions of Fe XVI suggests that the Co XVII contribution to the 312.55 Å line is around 15 erg cm^-2 s^-1 sr^-1, leaving 51 erg cm^-2 s^-1 sr^-1 as due to Fe XV. Taking into account this contribution, the observed 312.55/327.03 value would then agree with the theoretical branching ratio shown in Table 18.

The Fe XV line at 372.76 Å is too weak to account for the whole of the observed line intensity-in the high density limit (i.e., above 10 cm^-3 ), the 372.76/327.03 ratio is predicted to be 0.12, while the observed ratio of the lines is . At the density predicted by the 321.78/327.03 ratio (Table 20) the 372.76/327.03 ratio falls to around 0.03-0.04.

The transitions identified at 292.392 Å and 304.874 Å originate from the same upper level (3p P ) and represent radiative decays down to the 3s3p P and P levels, respectively. The splitting of the these two levels is known from other lines in the Fe XV spectrum and so if we know the wavelength of one line we know the wavelength of the other. Doing this we find that if the 304.874 Å identification is correct (as would be expected since it is the stronger line) then it implies a line at 292.26 Å rather than 292.392 Å. There is in fact a line at 292.251 Å in the SERTS-89 spectrum (mistakenly identified as a Si X line, as shown in Sect. 9.4) and so we suggest that this line is actually the Fe XV 3s3p P - 3p P transition.

With this identification, we can then estimate the contribution of Fe XV to the blend at 304.87 Å as being around 120 erg cm^-2 s^-1 sr^-1. The other two components to this blend are from Fe XVII (not identified) and Mn XIV. The Fe XVII component is negligible and can be estimated to be 4 erg cm^-2 s^-1 sr^-1 (see Sect. 13.9), while the Mn XIV component can be estimated at around 50-60 erg cm^-2 s^-1 sr^-1 by comparison with Mn XV 360.96 Å and the iso-electronic transitions in Fe XV and Fe XVI. The combined total is then in good agreement with the observed blend intensity of 206 erg cm^-2 s^-1 sr^-1.

Keenan et al. (1993b) suggest the identification of a line at 324.97 Å from Skylab spectra, which is not identified in the SERTS-89 spectrum. However we note that this line is density insensitive relative to the 327.03 Å line, with a value of , while the corresponding 324.97/327.03 ratio in the Skylab spectra presented in Keenan et al. has an average value of 0.09, casting doubts on the identification of this line.

Of the ten Fe XV lines identified in SERTS-89, we can dismiss three lines as being mis-identifications (256.91 Å, 292.39 Å, and 372.76 Å), and add one that had incorrectly been called Si X (292.25 Å). In addition, Brickhouse et al. have already noted that the unidentified line at 393.97 Å is probably the forbidden Fe XV 3s3p P - 3s S transition, an identification that we also adopt. We can form three groups of insensitive lines. Group 1 (243.78 Å, 284.16 Å, 312.55 Å, 327.03 Å and 417.25 Å) shows internal consistency apart from the well-documented case of 417.25 Å, while Group 2 (292.25 Å, 304.87 Å and 321.78 Å) shows reasonable agreement. Group 3 consists of just the 393.97 Å line.

Taking any ratio between lines of the different groups yields a density diagnostic, and we choose 321.78/327.03 and 327.03/393.97 with results shown in Table 20. The low density derived from the 327.03/393.97 ratio suggests the 393.97 Å line is possibly blended.

13.8. Fe XVI

There are five transitions between the n=3 configurations of Fe XVI and they all appear in the SERTS-89 spectrum. All the ratios are density insensitive, although there is some temperature sensitivity in the 3s-3p/3p-3d ratios. Comparisons with observations are given in Table 18 and Table 19. Surprisingly the only major discrepancy is for the single branching ratio 265.02/251.07. By virtue of the excellent agreement for the density insensitive ratio involving the 251.07 Å line given in Table 19, we surmise that it is the 265.02 Å line causing the problem, in which case the observed intensity of the latter line is lower than theory predicts and so blending can not be invoked. We note that Skylab measurements of this ratio presented by Keenan et al. (1994c) range between 0.13 and 0.23, more consistent with theory and so we suggest there may be a problem with the reported SERTS-89 observation of the 265.02 Å line. In fact, an improved fit to this line in the SERTS-89 data is given in Table 24, and it provides a 265.02/251.07 ratio of , which is now in excellent agreement with the predicted value.

13.9. Fe XVII

There are many weak 3s-3p and 3p-3d transitions expected in the 200-320 Å region, but only the 254.89 Å line is strong enough to be observed by SERTS-89. Above 320 Å, the higher sensitivity of the instrument allows several 3s-3p transitions to be clearly identified.

In addition to the lines in the SERTS catalogue, Brickhouse et al. (1995) identify the feature observed at 367.287 Å as an Fe XVII line. We confirm this and give the transition information in Table 23. The intensity of the line is in excellent agreement with theory, as witnessed by the data in Table 19.

Taking ratios relative to the 350.48 Å line, all the lines agree well with theory except for 409.71 Å, which is around a factor of 2 too low. The reasons for this are unclear. The 389.08 Å line is blended with an Ar XVI line, and we estimate an Fe XVII contribution of around 7.1 erg cm^-2 s^-1 sr^-1 to the total of 12.8 erg cm^-2s^-1 sr^-1, which is consistent with the estimate made for the Ar XVI line in Sect. 7.2. A line is predicted by CHIANTI at 304.94 Å, making it a potential blend with the observed 304.87 Å line identified as Mn XIV plus Fe XV. However, the density insensitive 304.94/350.48 ratio shows that the expected Fe XVII intensity of this line is only around 4 erg cm^-2 s^-1 sr^-1, less than 2% of the other components.

Although good agreement with theory has been found for Fe XVII, we note that four of the transition identifications of CHIANTI disagree with those of Thomas & Neupert. In particular, the 254.892 Å line is given by CHIANTI as a ³P₁ -¹S₀ transition, 347.814 Å as a ³P₁ -¹D₂ transition, 358.247 Å as a ¹P₁ -³P₁ transition, and 389.075 Å as a ¹P₁ -³ D₂ transition. The reason for these discrepancies stem from the use in CHIANTI of atomic data from Bhatia & Doschek (1992), who adopted different term designations than those used in previous work - see, e.g., Jupén (1984).

© European Southern Observatory (ESO) 1998

Online publication: November 24, 1997
helpdesk.link@springer.de