© 1998 RAS

Mon. Not. R. Astron. Soc. 297, 399–404 (1998)

Solar EUV line identifications from delayed beam-foil spectra

Elmar Trabert

Experimentalphysik III, Ruhr-Universitat Bochum, D-44780 Bochum, Germany

Accepted 1998 January 26. Received 1998 January 26; in original form 1997 April 2

A B STR ACTA number of electric dipole transitions from low-lying long-lived levels in Si-, P- andS-like ions of Fe and neighbouring elements have been identified by their longevityin delayed EUV spectra after foil excitation of fast ion beams. Some of these newidentifications pertain to previously unidentified lines in solar flare spectra.

Key words: atomic data – line identification – Sun: corona.

1 INTRODUCTION

More than 50 years ago, Edlen (1942) identified the visiblecorona lines with electric-dipole forbidden transitionsbetween fine-structure levels of highly charged ions of irongroup elements. This insight revolutionized models of thesolar corona. The electric-dipole allowed ground-statetransitions in the same ions to relate to lines in the extremeUV part of the spectrum and became observable only afterrockets and spacecraft moved spectrographs beyond theEarth’s atmosphere. Extreme ultraviolet (EUV) spectra ofthe corona and of solar flares have been presented by Behr-ing et al. (1976) and Dere (1978). For both sets of spectra,line identifications have been given that were based onwavelength coincidences with data from laboratory spectra.However, the laboratory data base at the time was incom-plete (and still is), leaving a number of solar lines unclassi-fied. Furthermore, within the spectral resolution of the solardata some spectral blends are likely, and misidentificationsfrom lack of suitable reference data seem almost unavoid-able.

A notable problem for laboratory work is posed by long-lived levels. In most terrestrial light sources, such long-livedexcited levels are quenched collisionally rather than given achance to decay radiatively. Two very different light sourceshave been employed to improve on this. One is the low-density plasma discharge in tokamak nuclear fusion testdevices, offering very precise spectroscopy on highlycharged ion species that are present as impurities. The otheris a beam of fast ions from an accelerator, which are excitedby being passed through a thin carbon foil and which thenproceed and decay in a practically collision-free environ-ment. The latter light source has the additional benefit ofisotopic purity and inherent time resolution (Trabert et al.1988). These features helped to identify intercombination

transitions in Mg-, Al- and Si-like spectra of iron groupelements in solar spectra (Trabert et al. 1987).

Unfortunately, the fast-ion beam light source is feeble,and observation suffers from Doppler shifts and broaden-ing. Consequently the attainable spectral resolution islimited, and line-rich spectra are partly resolvable at best.The situation improves for observations after, say, a fewnanoseconds delay (observation of the excited ion beam at aposition of few cm downbeam of the foil, for the typical 0.5MeV nucleonÐ1 ion beam energies). Although there theoverall light level is even lower, the decays of short-livedlevels are exhausted, and the delayed spectra are then domi-nated by the much less numerous decays of long-lived levels.Furthermore, it is possible to obtain decay curves on promi-nent lines and thus measure atomic lifetimes, a helpful toolfor line identifications as well as a key to aspects of atomicdynamics beyond the atomic level structure.

Ideally, data from different light sources are combinedfor unambiguous line classification. In such work, observa-tions of low-density laboratory and extraterrestrial plasmasprovide high-resolution spectra and precise wavelengths.However, as there are many elements (possible impurities)present, the identity of a given line with an element or evena specific charge state often needs to be established inde-pendently. This can be the unique contribution of fast-ionbeam spectroscopy.

As an example of such a combination of data from low-density plasmas like the tokamak controlled-fusion devices,solar corona observations and delayed beam-foil spectra,Jupen, Isler & Trabert (1993) assigned a number of indivi-dual lines of Fe X–Fe XIV. However, at the time beam-foilmeasurements were available for only some of the data. Ournew data suggest that some (few) of the identifications thatwere based on time-integrated spectroscopy are possibly inconflict with new evidence from time-resolved spectroscopy.

The new data also point to several new identifications ofpreviously unclassified solar lines.

2 THE EXPER IMENT

The experiment was performed with the same set-up asused previously by Trabert et al. (1987, 1988). Previously,the emphasis was on Si-, Al-, Mg- and Na-like ions. Lowerion energies than before have now been employed to shiftthe charge-state distribution to lower charge states, that isto excite the spectra of the Ar-, Cl-, S- and P-like ions of Fe,Ni and Cu. EUV light emitted by the ions was dispersed bya 2.2-m grazing-incidence scanning monochromator(McPherson Mod. 247) equipped with a 600 line mmÐ1

ruled grating, and detected by a channeltron. Sample spec-tra (Fig. 1) show how the relative intensities of lines change

with the incident ion energy. In our spectral range the indi-vidual spectra of the aforementioned charge states (ionspecies) of each element overlap and are difficult to disen-tangle. We therefore concentrate on some prominent lines(in the delayed spectra, marked in Fig. 1) which belong tothe isoelectronic sequences of Si (transition 3s23p2 3P2–3s23p3d3 Fo

3), P (transitions 3s23p3 2Do5/2–3s23p23d2,4L7/2) and

S (3s23p33d 5Do transition array).Our beam-foil wavelength measurements were tied to the

3p–3d transitions in the Na-like ions and to several otherwell-known lines. However, most of these lines originatefrom relatively short-lived levels. Hence their intensity ishigh when observed near the foil (prompt spectra), and lowin delayed spectra. In the prompt spectra, the lines ofinterest are regularly blended with a multitude of otherlines. After a delay time sufficient to let the line of intereststand out, the reference lines tend to be rather weak. Insuch delayed spectra, the overall intensity of the lines is toolow to permit working with high spectral resolution (con-sidering accelerator time, finite exciter-foil life and the darkrate of the detector). The wavelength transfer then dependson the mechanical properties and measurement precision ofthe scanning monochromator, and it suffers from the needto use rather wide (100 to 200 lm) spectrometer slits whichresult in relatively wide spectral lines (FWHM 0.07 to 0.14nm). Furthermore, calibration lines from few-electron spec-tra are weak or not present at all when the ion energy islowered to produce the spectra of ions with more electronsin the outermost shell. Thus stepwise calibration may needspectra taken at intermediate energies, which adds furtheruncertainties. An illustrative example: at the lowest ionenergies used here, Ar I-like spectra feature three transi-tions between the J\0 ground state and J\1 levels of the3p5 3d configuration. Of these three upper levels, one is veryshort lived (and has a wavelength falling short of the regionof interest); one has a lifetime in the few-nanosecond rangeand was used here, referring to wavelength data by Sugar,Kaufman & Rowan (1987) and Ekberg & Litzen (privatecommunication). The third line, with a many-nanosecondlifetime, was observed, but was too weak to be exploited fora calibration. As a consequence of the multistep calibrationproblem, our measured wavelengths bear typical uncertain-ties of 0.015 to 0.02 nm.

In order to facilitate line identification, time-resolvedspectra were simulated on the basis of theoretical informa-tion on wavelengths and transition probabilities and thenvisually compared with beam-foil data. The characteristicpatterns of relative line intensities changing with increasingdelay after excitation helped to single out the decays ofparticular interest.

3 R ESULTS

3.1 Si I sequence

In the Si isoelectronic sequence, the short-lived levels of then\3 manifold are known, while their lifetimes are difficultto measure because of cascading problems (Trabert et al.1989, 1996b). The most long-lived 3s3p3 level with an elec-tric dipole (intercombination) decay channel, 3s3p3 5So

2, hasbeen identified and studied in moderately charged ions bybeam-foil spectroscopy (Trabert et al. 1988). For low charge

400 E. Trabert

© 1998 RAS, MNRAS 297, 399–404

Figure 1. Delayed spectra of Fe at various ion beam energies anddelay times of observation (for a time interval of about 0.15 ns)after excitation. The prominent lines discussed in the text arelabelled by isoelectronic sequence symbols Si, P and S. In the rangemarked Cl/S/P various transition arrays from the individual ioniza-tion stages overlap.

states, radio-frequency ion-trap data (Calamai, Han & Par-kinson 1992) have been essential to guide theory to therecognition of problems in the treatment of the intercombi-nation decay transition amplitude (Hibbert 1993). In then\3 level manifold, however, there are also triplet levels ofnotably long lifetime which are amenable to lifetimemeasurement (Trabert et al. 1989, 1996b). Of these, the3s23p3d 3 F0

3 level is presently of particular interest, becauseits unbranched decay (to 3s23p2 3P0

2) must give rise to aprominent line that would extend to some distance from thefoil. The line, in fact, features rather prominently in delayedbeam-foil spectra of elements from Ca to Cu (Trabert et al.1993a,b, 1996a,b), and the charge-state assignment hasbeen corroborated by energy variations of the ion beam.

The 3s23p3d3 Fo3 level has been covered computationally

by Biemont (1986a,b) and by Fawcett (1987). Biemont andFawcett used the Cowan code with semi-empirical adjust-ments of the Slater parameters to match the positions of theknown energy levels of the Si sequence spectra. However, asexperimental data on the 3s23p3d 3Fo term were not avail-able at the time (see also Sugar, Kaufman & Rowan 1990),it is not surprising to find a slight mismatch of Biemont’spredictions with our experimental wavelengths for this par-ticular line (Table 1). Biemont’s and Fawcett’s calculationsalso give oscillator strengths which translate into level life-times. Our lifetime data for Fe XIII, Co XIV, Ni XV andCu XVI closely match the predictions. Attempts at measur-ing the lifetime of this level in Ca VII corroborated, withroughly 30 ns, the long lifetime expected for this low charge-state ion (Trabert et al. 1996b).

For Fe XIII our data indicate a wavelength of about 23.92nm, which makes the solar flare line at 23.903 nm a suitablecandidate for identification with the Fe XIII 3p23p2 3P2–3s23p3d 3Fo

3 transition. This line identification, supported bycharge-state variation, delayed spectra and decay-curveanalysis, yields a line position that notably differs from thatobtained by Jupen et al. (1993) (23.761 or 23.7597 nm) onthe basis of time-integrated (tokamak and solar) spectraonly. Whereas the line suggested by Jupen et al is very closeto the result of the calculations by Biemont (1986b), thepresent study draws from beam-foil data for many elementsin the isoelectronic sequence that show a systematic offsetbetween predicted and measured wavelength. By isoelec-tronic comparison with Biemont’s (1986a,b) calculations,our beam-foil data point to corresponding candidate lines in

Cr XI, Ti IX and Ca VII (Table 1). Refined calculationsincluding the newly determined level would be helpful toestablish the remaining fine-structure levels of the 3s23p3d3Fo term, from solar or tokamak spectra.

3.2 P I sequence

The ground configuration of P-like ions has levels with totalangular momentum J values up to J\5/2. Electric dipoletransitions connect these to excited levels with J values up toJ\7/2, but proper-term values of the latter are not availablein the literature. Mori, Otsuka & Kato (1977) note theexistence of these levels by their connections to even higher-lying levels, indicating unknown offsets of the level values.In their Grotrian diagrams, they tentatively place the J\7/2levels in the 700 000 cmÐ1 range of excitation energies,whereas recent calculations by Fritzsche, Froese Fischer &Fricke (1997) place them near 450 000 cmÐ1. The experi-mental problems of identifying the J\7/2 level decaysrelate to two facts. First, the decays of the J\7/2 levels areunbranched, as there is only a single J\5/2 level in theground configuration, and positive identifications of singlelines are often daring. Secondly, of the four levels in ques-tion, one is extremely long-lived [a prediction of 13.6 ls inFe XII (Fritzsche et al. 1997) and therefore not observablewith the present fast-ion beam technique], and two are inthe many-nanosecond lifetime range not often observedwith dense plasmas. However, this latter lifetime range isexactly the most suitable range for the delayed-spectra tech-nique employed here. Indeed, in spectra recorded afterdelay times of order 10 ns and longer, two lines appear thatfeature the predicted wavelength and wavelength separa-tion. For Fe XII the predicted lifetimes of – in LS couplingnotation – the 2,4F7/2 levels are about 29 and 39 ns, respec-tively, and thus the associated decay lengths are longer thanthe vacuum chamber which houses the displaceable exciterfoil. However, there is another J\7/2 level (2G7/2) with apredicted lifetime of 3.8 ns, which is visible at shorter delays.The observed decay component of about 2 ns at the positionof that line is of the right order of magnitude, but the decaycurve indicates a multiple line blend, including short-lived(less than 0.1 ns) and long-lived (9.5 ns) contributions.

For Cu XV the same levels feature predicted (interpo-lated from the data for Ni and Zn) lifetimes of about 13 and17 ns, respectively, and these have been confirmed within 20

Solar EUV line identifications 401

© 1998 RAS, MNRAS 297, 399–404

Table 1. The transition 3s23p2 3P2–3s23p3d 3Fo3 in the Si I isoelectronic sequence. Theoretical wavelengths lth and lifetimes th are from

Biémont (1986a,b). Observed wavelengths lobs are from the solar spectra tabulated by Dere (1978), or from fast-beam spectroscopy (fbs,this work). The wavenumber differences have been divided by the ion core charge (spectrum number) z in the second such column.

Element Wavelengthlth/nm

Wavelengthlobs/nm

Wavenumberdifference(sthÐsobs)/cmÐ1

Wavenumberdifference (scaled)(Ds/z)/cmÐ1

Lifetimetobs/nm tth/nm

Ca 40.299 40.455¹0.003 solar 957¹18 137 51.5Ti 32.629 32.786¹0.003 solar 1468¹28 163 17.3Cr 27.487 27.660¹0.003 solar 2275¹39 207 6.60Fe 23.770 23.903¹0.003 solar 2341¹53 180

23.92¹0.015 fbs 2638¹280 203 3.0¹0.2 2.00Co 22.252 22.38¹0.015 fbs 2570¹300 183 1.8¹0.2 2.02Ni 20.911 21.02¹0.015 fbs 2480¹350 165 1.45¹0.08 1.42Cu 19.73 extrap. 19.83¹0.02 fbs 2556¹500 160 1.01¹0.05 1.03

per cent by decay-curve measurements. Once the lines areidentified for Cu XV and Ni XIV, the isoelectric trend pointsout wavelengths for Fe XII. One of the lines is readilyobserved in delayed beam-foil spectra at (24.94¹0.02) nm,which matches with the solar wavelength of 24.938 nm, butthe other, observed in beam-foil spectra at (25.65¹0.02)nm, would in solar spectra be blended by the strong He II

line at 25.633 nm. The third J\7/2 level decay figuresprominently in beam-foil spectra recorded with little delay,at a wavelength of (22.30¹0.02) nm, a position coincidentwith a solar line previously assigned exclusively to a line inCr (Table 2). Similar candidate lines appear in the beam-foil spectra of Co, Ni and Cu. Lifetime measurements of the2G7/2 level were attempted, but the results remained incon-clusive with respect to the prediction. With the exception ofCo where a single strong component of 4.7¹5 ns wasobserved, the curves showed a composite shape with onecomponent in the 1.5 to 2 ns range and another in the 6 to12 ns range. Such observation may relate to a blend withlines fed by slow cascades, or to a primary lifetime abouthalf as long as that predicted, plus cascade repopulation ofthe level of interest from higher-lying, long-lived levels. Forexample, the lowest excited levels of odd parity above theground configuration are expected to be those of the3s3p33d 6D term. These levels are expected to be ratherlong-lived, because their decay toward 3s23p23d or 3s3p4

levels requires a spin change. The level positions of the 6Dterm are not known yet.

The identification in the solar spectra of such candidatelines for J\7/2 level decays in P-like spectra helps with thesearch for other solar lines from the same transitions inlighter ions. Assuming a wavelength offset roughly similarto that between our measurement for Fe XII and the calcula-tion by Fritzsche et al. (1997) to be valid for Cr X and Ti VIII

as well, one finds solar lines in Dere’s tables at wavelengths

that are longer by 0.05 to 0.07 nm than the predicted ones.However, some of these lines have previously been identi-fied with transitions in other ions (Table 2). More preciselaboratory data as well as semi-empirically adjusted calcula-tions will be needed to determine the isoelectronic datatrends conclusively and to decide which of the possibly con-tributing transitions dominate the individual solarfeatures.

3.3 S I sequence

The 3s23p4 3P–3s23p3 5Do transition array in S I-like spectragives rise to multiplet that was found to be dominating thedelayed EUV spectra after foil excitation of iron groupelement ions (Trabert et al. 1993a,b). The multiplet consistsof two groups of lines which are, for Fe XI, at wavelengths ofabout 25.7 and 26.6 nm. The shorter wavelength group con-sists of the transitions to the J\2 ground level and isexpected to be more intense than the other group (consist-ing of decays to levels J\0, 1) by about a factor of 3 (theintensity estimate takes the inherent time resolution of thebeam-foil light source into account). The predicted fine-structure splitting of the upper term, 3s23p33d 5Do, is about300 cmÐ1 between neighbouring levels (Jupen, private com-munication; Bhatia & Doschek 1996). This is too small to beresolved with spectral lines of wavenumbers close to 400 kKin our beam-foil observations, where the signal level is lowand the observation of such long-lived level decays (pre-dicted lifetimes in the range 10 to 50 ns) out of necessityproceeds with wide spectrometer slits. However, from solarobservations data with higher spectral resolution are avail-able. If in such high-resolution spectra the transition arrayof present interest can be recognized, it might also be pos-sible to identify individual components. As it turns out, theoptimum case is Fe XI.

402 E. Trabert

© 1998 RAS, MNRAS 297, 399–404

Table 2. Decays from J\7/2 levels in P-like ions (transitions 3s23p3 2Do5/2–3s23p2 3d2,4L7/2). Data sources: fbs, fast-beam spectroscopic data

from this work, solar, catalogue of Dere (1978) with identification given there; other data, theory (Fritzsche et al. 1997; the numbers forCu are from an interpolation of the theory data).

Level 3s23p23d 2F7/2 3s23p23d 4F7/2 3s23p23d 2G7/2

Wavelength Lifetime Wavelength Lifetime Wavelength LifetimeSpectrum l/nm t/nm l/nm t/ns l/nm t/ns

Cu XV 21.097 18 20.534 13.5 18.18 1.321.16¹0.02 fbs 20¹3 fbs 20.63¹0.02 fbs 11.4¹2 fbs 18.22¹0.02 fbs

Ni XIV 22.365 22.7 21.774 16.3 19.382 1.8622.43¹0.02 fbs 21.83¹0.02 fbs 19.46¹0.02 fbs

Co XIII 23.865 29.5 23.217 20.3 20.675 2.623.95¹0.23 fbs 18¹5 fbs blended 23.29¹0.02 fbsd 10¹2 fbs 20.78¹0.02 fbsblended with 23.89 nmCo XI

3P25D0

Fe XII 25.598 38.8 24.892 28.9 22.229 3.9225.65¹0.02 fbs 24.94¹0.02 fbs 23.31¹0.02 fbs

7.3.¹1.0 fbs25.633 solar He II 24.938 solar 22.299 solar Cr XXII

Cr X 29.897 72.4 29.027 50.7 25.919 8.7929.9612 solar Si IX 20.100 solar Fe XII 25.997 solar

Ti VIII 35.990 106 34.883 1910 31.112 22.336.076 solar Fe XVI 34.913 solar Mg VI

34.988 solar Si IX

31.174 solar Mg VIII

A number of factors contribute to this optimum situation.There is the detailed knowledge on the spectra of thevarious Fe ionization stages (Mori et al. 1997; Sugar &Corliss 1988; Jupen et al. 1993), the solar abundance of Feis favourably high, the levels of interest lie in an apparentlyadvantageous lifetime range and there are incidental gaps inthe distribution of possibly blending known transitions fromother elements and charge states. Whereas the ground con-figuration levels of Fe XI are well established, the upper-term level structure is available from theory only.Calculations of the levels involved and some of the decayprobabilities have been performed by Fawcett (1986), Jupen(private communication), Chou et al. (1996) and Bhatia &Doschek (1996). We find that in the Bhatia & Doschekcalculations several groups of levels differ in position by3000 to 10 000 cmÐ1 from experiment (if known) and alsofrom the results of Jupen’s scaled Cowan code calculations.The larger value pertains to some fairly low-lying levels likethe aforementioned 3s23p33d 5Do levels. Such a deviationilluminates the earlier problem of identifying the intercom-bination transition array in any spectra.

For our search of candidate lines in the solar spectra, wecombine the known ground configuration levels with theobserved beam-foil wavelengths. For the two-part transitionarray, the best match appears to be that of the strongestmultiplet component, 3s23p4 3P2–3s23p33d 5Do

2, with the solarline at 25.755 nm. This in turn provides three further goodmatches (see Table 3), while the line at 25.778 nm is toohigh in wavelength by 0.005 nm for an equally perfectmatch. However, at an estimated measurement error of0.003 nm, this is not bad either (and might even point to theactual upper-level fine-structure interval being differentfrom prediction or to a blend with an unknown further line).The wavenumber difference for the alternative identifica-tions of the 2–3 transition is about 437 cmÐ1, which wouldspoil the systematics of Table 3.

We note that our transitions 2–2 and 2–1 have previouslybeen suggested as 2–3 and 2–2 in the same transition array(Jupen et al. 1993). Jupen et al. (1993) assign the wave-length of the 2–3 component of Table 3 (25.726 nm) to atransition in Fe X exclusively instead, following earlier work(Smitt 1976). The upper level of that transition in Fe X hasa J value of 5/2 and thus an unbranched decay to the J\3/2level of the ground state, which might give rise to a quite

prominent line, comparable to the line intensity of thepresently suggested transition in Fe XI. A blend of the twocandidate lines seems likely. Such ambiguity of identifica-tion can only be settled if laboratory data can be obtainedon all fine-structure components, preferably with photo-electric intensity measurements.

However, for the other ions in this isoelectronic sequence(restricting the discussion to the more abundant even-number element ions), there is no such seemingly perfectline pattern to enable immediate identification. Thereforeall these comparison hinge on the data for Fe XI. Theexpected isoelectronic trend assumes a mismatch betweencalculation and experiment that is almost constant on thewavelength scale (or nearly proportional to ion charge on alevel energy scale, as has been discussed above for the Siisoelectronic sequence). For Ni XIII, the strongest compo-nent of the multiplet, 2–2, may be expected near l\22.75nm. There are many other identified lines in this region thatmight mask the line of present interest. Of the other part ofthe multiplet, the 1–1 component may be expected in therange 23.76 to 23.8 nm, and the previously unidentified lineat 23.761 nm is a possible candidate. For Cr IX, our esti-mates indicate the wavelength range near 30.90 nm. In thisrange, Dere lists unidentified lines at 30.896 and 30.924 nm.If these were correctly identified with two of the three moreprominent 2–Jp transitions, the other part of the multipletwould likely be blended with the 31.702-nm line fromMg VIII or the 31.761-nm line of Fe XV. For Ti VII, theunidentified Dere lines at 37.980 and 37.991 nm having aspacing and position that might fit to the 2–2 and 2–1components of the multiplet of present interest, while the38.723-nm line would be close to the matching position ofthe 1–0 component. However, there is a blending line nearthe position of the 1–0 component, while no candidate lineis available for the 2–3 component. The latter may be weakbecause of its much longer upper-level lifetime. For Ca V,the unidentified Dere line at 48.952 nm might be associatedtentatively with the 2–2 componnt, but such a single line isinsufficient for a decisive identification of the transitionarray.

4 CONCLUSIONS

The time resolution inherent in fast-beam spectroscopy, aswell as the elemental purity of the beam-foil light source,has again proven essential in assigning coronal lines, thistime of transitions in Si I-, P I- and S I-like spectra of irongroup elements. In the process, corrections to previousidentifications from time-integrated spectra were found tobe necessary. A critical re-evaluation of the solar data, onthe basis of the up-to-date laboratory data base, and includ-ing cross-checks for the completeness of the spectra forwhich individual lines are noted, might indicate more aboutsuch inconsistencies.

On the positive side, coronal data with their superiorwavelength precision, in comparison to most beam-foilwavelength data, can guide the analysis of terrestrial data aswell as parametric calculations and isoelectronic intercom-parisons, which will be needed to systematize the atomicstructure data. In a subsequent step it will become possibleto extend the present identifications to lighter ions, the linesof which probably show in the recent longer wavelength

Solar EUV line identifications 403

© 1998 RAS, MNRAS 297, 399–404

Table 3. The Fe XI 3s23p4 3PJ–3s23p33d 5DoJp transition array in S-

like ions. The wavelength estimates combine known groundconfiguration levels (Sugar & Corliss 1988) with calculatedexcited lavels (Fawcet 1986; Bhatia & Doschek 1996). The solarflare wavelengths are previously unidentified lines from Dere’stables (1978).

JÐJ p Wavelength Solar Wavelengthestimate wavelength differencel/nm l/nm s/cmÐ1

2–3 25.347 25.726 5812 blend Fe X

2–2 25.374 25.755 58302–1 25.391 25.778 59131–2 26.623 blend Ar XV

1–1 26.235 26.642 58231–0 26.250 26.660 5859

EUV solar spectra obtained by the SOHO spacecraft (Feld-man et al. 1997) and which are only incompletely classified.Of particular interest, there are some J\9/2 levels in P-likeions that would decay only (by M1 transition) to J\7/2levels as determined here. Calculations that reproduce thelatter might yield better predictions for the J\9/2 levels aswell, and thus provide a guide to identifications.

ACKNOWLEDGMENTS

The logistic hospitality of Professor C. Rolfs and his chair isgratefully acknowledged. Discussions with and the provisionof computations before publication by C. Jupen (Lund) andS. Fritzsche (Kassel) have been most helpful. C. Jupen alsokindly confirmed that the present assignments comply withhis iso-electronic analyses. Parts of this study were sup-ported by Fysiografiska Selskap Lund, NATO, DeutscheForschungsgemeinschaft and an Exchange Program ofDeutscher Akademischer Austauschdienst (DAAD) withSvenska Institutet (SI).

404 E. Trabert

© 1998 RAS, MNRAS 297, 399–404

R EFER ENCES

Behring W. E., Cohen L., Feldman U., Doschek G. A., 1976, ApJ,203, 521

Bhatia A. K., Doschek G. A., 1996, At. Data Nucl. Data Tables, 64,183

Biemont E., 1986a, J. Opt. Soc. Am. B, 3, 163Biemont E., 1986b, Physica Scripta, 33, 324Calamai A. G., Han X., Parkinson W. H., 1992, Phys. Rev. A, 45,

2716Chou H.-S., Chang J.-Y., Chang Y.-H., Huang K.-N., 1996, At.

Data Nucl. Data Tables, 62, 77

Dere K. P., 1978, ApJ, 221, 1062Edlen B., 1942, Z. Astrophys., 22, 30Fawcett B. C., 1986, At. Data Nucl. Data Tables, 35, 185Fawcett B. C., 1987, At. Data Nucl. Data Tables, 36, 129Feldman U., Behring W. E., Curdt W., Schuhle U., Wilhelm K.,

Lemaire P., Moran T. M., 1997, ApJS, 113, 195Fritzsche S., Finkbeiner M., Fricke B., Sepp W.-D., 1995, Physica

Scripta, 52, 258Fritzsche S., Froese Fischer C., Fricke B., 1997, At. Data Nucl.

Data Tables, submittedHibbert A., 1993, J. Phys. B, 26, L91Jupen C., Isler R. C., Trabert E., 1993, MNRAS, 264, 627Mori K., Otsuka M., Kato T., 1977, Grotrian Diagrams of Highly

Ionized Iron, Fe VIII–Fe XVI. Institute of Plasma Physics,Nagoya Univ.

Smitt R., 1976, Sol. Phys., 51, 113Sugar J., Corliss C., 1988, J. Phys. Chem. Ref. Data, 14, Suppl. 2Sugar J., Kaufman V., Rowan R. L., 1987, J. Opt. Soc. Am. B, 4,

1927Sugar J., Kaufman V., Rowan W. L., 1990, J. Opt. Soc. Am. B, 7,

152Trabert E., 1993, Physica Scripta, 48, 699Trabert E., Heckmann P. H., Hutton R., Martinson I., 1987,

MNRAS, 227, 27P

Trabert E., Heckmann P. H., Hutton R., Martinson I., 1988, J. Opt.Soc. Am. B, 5, 2173

Trabert E., Reistad N., Martinson I., Hutton R., 1989, Z. Phys. D,11, 207

Trabert E. et al., 1993, Physica Scripta, 48, 580Trabert E., Doerfert J., Granzow J., Heckman P. H., Jupen C.,

Hutton R., Takacs E., 1996a, in Tchang-Brillet W.-U., WyartJ.-F., Zeippen C. J., eds, Abstracts 5th Int. Colloq. on AtomicSpectra and Oscillator Strengths for Astrophysical andLaboratory Plasmas (ASOSALP). Publ. de l’Observatoire deParis, Meudon, p. 46

Trabert E., Pinnington E. H., Kernahan J. A., Doerfert J., GranzowJ., Heckmann P. H., Hutton R., 1996b, J. Phys. B, 29, 2647