New Fe ii energy levels from stellar spectra

F. Castelli; R. L. Kurucz

doi:10.1051/0004-6361/201015126

Home

All issues

Volume 520 (September-October 2010)

A&A, 520 (2010) A57

Full HTML

Free Access

Issue		A&A Volume 520, September-October 2010


Article Number		A57
Number of page(s)		30
Section		Atomic, molecular, and nuclear data
DOI		https://doi.org/10.1051/0004-6361/201015126
Published online		04 October 2010

A&A 520, A57 (2010)

New Fe II energy levels from stellar spectra

F. Castelli¹ - R. L. Kurucz²

1 - Istituto Nazionale di Astrofisica - Osservatorio Astronomico di Trieste, via Tiepolo 11, 34131 Trieste, Italy
2 - Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA

Received 1 June 2010 / Accepted 29 June 2010

Abstract
Aims. The spectra of B-type and early A-type stars show numerous unidentified lines in the whole optical range, especially in the 5100-5400 Å interval. Because Fe II transitions to high energy levels should be observed in this region, we used semiempirical predicted wavelengths and gf-values of Fe II to identify unknown lines.
Methods. Semiempirical line data for Fe II computed by Kurucz are used to synthesize the spectrum of the slow-rotating, Fe -overabundant CP star HR 6000.
Results. We determined a total of 109 new 4f levels for Fe II with energies ranging from 122 324 cm^-1 to 128 110 cm^-1. They belong to the Fe II subconfigurations 3d⁶(³P)4f (10 levels), 3d⁶(³H)4f (36 levels), 3d⁶(³F)4f (37 levels), and 3d⁶(³G)4f (26 levels). We also found 14 even levels from 4d (3 levels), 5d (7 levels), and 6d (4 levels) configurations. The new levels have allowed us to identify more than 50% of the previously unidentified lines of HR 6000 in the wavelength region 3800-8000 Å. Tables listing the new energy levels are given in the paper; tables listing the spectral lines with $\log~gf \ge -$ 1.5 that are transitions to the 4f energy levels are given in the Online Material. These new levels produce 18 000 lines throughout the spectrum from the ultraviolet to the infrared.

Key words: line: identification - atomic data - stars: atmospheres - stars: chemically peculiar - stars: individual: HR 6000

1 Introduction

In a previous paper (Castelli et al. 2009) (Paper I) we have determined 21 new 3d⁶(³H)4f high energy levels of Fe II on the basis of predicted energy levels, computed $\log~gf$ values for Fe II, and unidentified lines in UVES high resolution, high signal-to-noise spectra of HR 6000 and 46 Aql. Both stars are iron overabundant CP stars and have rotational velocity $v\sin i$ of the order of 1.5 km s^-1 and 1.0 km s^-1, respectively.

In this paper we continue the effort to determine new high-energy levels of Fe II. We used the same spectra and models for HR 6000 that we adopted in Paper I, together with Fe II line lists which include transitions between observed-observed, observed-predicted, and predicted-predicted energy levels. In this paper we increase the number of the new energy levels from the 21 listed in Paper I, to a total of 109 energy levels, which belong to the Fe II subconfigutations: 3d⁶(³P)4f (10 levels), 3d⁶(³H)4f (36 levels), 3d⁶(³F)4f (37 levels), and 3d⁶(³G)4f (26 levels), and 14 levels from the even configurations 4d (3 levels), 5d (7 levels), and 6d (4 levels). The new levels have allowed us to identify more than the 50% of the previously unidentified lines in the wavelength region 3800-8000 Å of HR 6000 (Castelli & Hubrig 2007). The method that we adopted to determine the new energy levels is the same as described in Paper I. It is recalled here in Sect. 3. The comparison of the observed spectrum of HR 6000 with the synthetic spectrum which includes the new Fe II lines is available on the Castelli web site.

2 The star HR 6000

According to Paper I, the CP star HR 6000 (HD 144667) has an estimated rotational velocity of 1.5 km s^-1. The model stellar parameters for an individual abundance ATLAS12 (Kurucz 2005) model are $T_{\rm eff}$ = 13 450 K, $\log~g$ = 4.3. In addition to the large iron overabundance [+0.9], overabundances of Xe ([+4.6]), P (>[+1.5]), Ti ([+0.55]), Cr ([+0.2]), Mn ([+1.5]), Y ([+1.2]), and Hg ([+2.7]) were observed. This peculiar chemical composition, together with the underabundances of He, C, N, O, Al, Mg, Si, S, Cl, Sc, V, Co, Ni, and Sr gives rise to an optical line spectrum very rich in Fe II lines, with transitions involving upper energy levels close to the ionization limit (Johansson 2009). Also numerous Fe I and Fe III lines are observable in the spectrum.

Table 1: Fe II energy levels for the 3d⁶ (³P)4f subconfiguration.

3 The method

To determine the new energy levels we used high-resolution UVES spectra of HR 6000 (see Paper I), the corresponding synthetic spectrum, and the list of the computed transitions with predicted values for levels with no experimentally available energies. Predicted energy levels and $\log~gf$ values were computed by Kurucz with his version of the Cowan (1981) code (Kurucz 2009). The calculation included 46 even configurations d⁷, d⁶4s-9s, d⁶4d-9d, d⁶5g-9g, d⁶7i-9i, d⁶9l, d⁵4s², d⁵4s5s-9s, d⁵4s4d-9d, d⁵4s5g-9g, d⁵4s7i-9i, d⁵4s9l, d⁴4s²4d, and d⁵4p² with 19 771 levels least-squares fitted to 418 known levels. The 39 odd configurations included d⁶4p-9p, d⁶4f-9f, d⁶6h-9h, d⁶8k-9k, d⁵4s4p-9p, d⁵4s4f-9f, d⁵4s6h-9h, d⁵4s8k-9k, d⁴4s²4p-5p, and d⁴4s²4f with 19 652 levels least-squares fitted to 596 known levels. The calculations were done in LS coupling with all configuration interactions included, with scaled Hartree-Fock starting guesses, and with Hartree-Fock transition integrals. A total of 7 080 169 lines were saved from the transition array of which 102 833 lines are between known levels and have good wavelengths.

The computed line list was sorted into tables of all the strong lines connected to every predicted level. When a given predicted level gives rise to at least two Fe II lines having $\log~gf \ge -$ 1.0, we selected one of these transitions and searched in the spectrum for those unidentified lines which have wavelength within $\pm$ 50 Å and residual flux within about $\pm$ 5% of those of the selected predicted line. From the observed wavelength of one of these unidentified lines and from the known energy of the lower or upper level of the predicted transition, we derived a possible energy for the predicted level. If most of transitions obtained with this energy correspond to lines observed in the spectrum, we kept the tentative energy value as a real value, otherwise we repeated the procedure using another line taken from the unidentified ones, and continued the searching until we found that energy for which most of the predicted lines correspond to the observed lines. Whenever one or more new levels were found, the whole semiempirical calculation was repeated to produce improved predicted wavelengths and $\log~gf$ -values. Because all configuration interactions were included, and because the mixing is exceptionally strong in the 4d and 5d configurations, every new level changed the predictions. Mixing between close levels can produce large uncertainties in the $\log~gf$ values for lines that involve those levels.

This procedure is very successfull for levels which produce two or more transitions with $\log~gf >$ 0.0, but becomes more and more difficult as the intensity of the predicted lines decreases. In fact, weak lines are usually blended with stronger components, so that the method may fail in these cases.

4 The new energy levels

The new energy levels of the 3d⁶(³P)4f, 3d⁶(³H)4f, 3d⁶(³F)4f, and 3d⁶(³G)4f subconfigurations and from the even configurations 3d⁶4d, 3d⁶5d, and 3d⁶6d are listed in Tables 1-5. Because the 3d⁶4f states of Fe II tend to appear in pairs we have used the $j_{\rm c}[K]_{j}$ notation of jK coupling for them, where ${\vec j}_{\rm c}$ is the total angular momentum of the core and ${\vec K}={{\vec j}_{\rm c}}+{\vec l}$ is the coupling of ${\vec j}_{\rm c}$ with the orbital angular momentum ${\vec l}$ of the active electron. The level pairs correspond to the two separate values of the total angular momentum ${\vec J}$ obtained when the spin ${\vec s}=\pm$ 1/2 of the active electron is added to ${\vec K}$ . The positive energies are those obtained by comparing observed and predicted line profiles, as described in Sect. 3 and shown in Fig. 2. The energies between parentheses in Tables 1-4 are predicted values for which we have been not able to find the corresponding observed level. The reason for the failure is that either all the lines from the energy level are weak or, even if some of the transitions are predicted as moderately strong ( $\log~gf > 0.0$ ), they are blended with other stronger components, so that their identification is uncertain. The columns with label ``c-o'' in Tables 1-5 show the difference between the predicted and observed energy levels.

The 4d even energy levels listed in Table 5 give rise to some of the transitions listed in the Online Material. The strongest transitions related with the 5d, and 6d even energy levels occur in the 6000-8000 Å region and in the 4000-5000 Å region, respectively. The transitions to the odd energy levels are discussed in Sect. 5

The observed energy levels, the least squares fits, the predicted energy levels, and the line lists can be found on the Kurucz web site. The observed levels come from the following sources: Johansson (1978), Sugar & Corliss (1985), Adam et al. (1987), Johansson & Baschek (1988), Johansson (1988, private comm.), Rosberg & Johansson (1992), Castelli et al. (2008), Castelli et al. (2009), and this work. The calculations on the web site are updated whenever there are improvements to the energy levels.

Table 2: Fe II energy levels for the 3d⁶ (³H)4f subconfiguration.

Table 3: Fe II energy levels for the 3d⁶ (³F)4f subconfiguration. Energies between parentheses are predicted values.

Table 4: Fe II energy levels for the 3d⁶ (³G)4f subconfiguration. Energies between parentheses are predicted values.

5 The new Fe II lines

The new Fe II lines in the 3800-8000 Å region, produced by transitions to the Fe II subconfigurations (³P)4f, (³H)4f, (³F)4f, and (³G)4f, are shown in Tables 6-9, respectively. Only lines with $\log~gf \ge -$ 1.50 are listed, because lines with lower $\log~gf$ values are not observable in this wavelength region of HR 6000. The new Fe II lines are mostly concentrated in the 5100-5400 Å interval. The upper energy levels (Cols. 1-4) were derived as described in Sect. 3, the lower energy levels (Cols. 5-6) are those described in Sect. 4, the calculated wavelength (Col. 7) is the Ritz wavelength in air, the $\log~gf$ values (Col. 8) were computed by Kurucz, the observed wavelengths (Col. 9) are the wavelengths of lines well observable in the HR 6000 spectrum. Most of them were listed as unidentified lines in Castelli & Hubrig (2007). In the last column, comments derived from the comparison of the observed and computed spectra are added for most lines. In a few cases, both computed and observed stellar lines correspond to lines measured by Johansson in laboratory works (Johansson 1978; Castelli et al. 2008). The notes ``J78'' and ``lab'' are added for these lines. When lines are computed weaker than the observed ones the disagreement can be due either to a too low $\log~gf$ value or to some unknown component which increases the line intensity. When lines are computed much stronger than the observed ones, some problem with the energy levels or/and $\log~gf$ computations is very probably present. When we observed a very good agreement between the observed and computed lines, either isolated or blends, we added the note ``good agreement''.

Table 5: Fe II new levels from 3d⁶4d, 3d⁶5d, and 3d⁶6d configurations.

Figure 1 shows the Fe II spectrum in the 5185-5196 Å interval, computed before and after the determination of the new energy levels. Figure 2 compares the observed spectrum of HR 6000 with the synthetic spectrum computed with the line list including the new Fe II lines. When the two figures are considered together, the improvement in the comparison between the observed and computed spectra is evident.

$\begin{figure} \par\includegraphics[width=16cm,clip]{15126fig1.eps} \par\end{figure}$	Figure 1: Upper panel shows the Fe II synthetic spectrum for the parameters of HR 6000 ( $T_{\rm eff}$ = 13 450 K, $\log~g$ = 4.3, $v\sin i=1.5$ km^-1, [Fe/H]=+0.9) computed with the line list availble before this work. The lower panel is the same, but with the new Fe II lines added in the line list.
Open with DEXTER

$\begin{figure}\par\includegraphics[width=16cm]{15126fig2.eps} \par\end{figure}$

Figure 2:

Comparison of the UVES spectrum of HR 6000 (black line) with a synthetic spectrum (red line) computed with a line list including the new Fe II lines. The line identification can be decoded as follows: for the first line, 150 last 3 digits of wavelength 518.5150 nm; 26 atomic number of iron; .01 charge/100, i.e. 26.01 identifies the line as Fe II; 105 123 is the energy of the lower level in cm^-1; 970 is the residual central intensity in per mil.

Open with DEXTER

6 Conclusions

Computed atomic data and stellar spectra observed at high resolution and high signal-to-noise ratio of the iron-overabundant, slow-rotating star HR 6000 were used to extend laboratory studies on Fe II energy levels and line transitions. We identified as Fe II about 500 unidentified spectral lines in the 3800-8000 Å region. A few of these lines were already identified as iron from laboratory analyses (Johansson 2007, private communication), but they were never classified. Because numerous other new lines are components of blends they contribute to improve the agreement between observed and computed spectra. On the other hand, there is a small number of new lines which are not observed in the spectrum. We believe that they are due to computational problems related with the mixing of the even energy levels rather than to incorrect energy values for the new 4f odd levels.

In spite of the large number of the new identified lines, several medium-strong lines and a conspicuous number of weak lines remain still unidentified in the spectral region we analyzed. If we examine the list of the Fe II lines which correspond to transitions from predicted energy levels, we can count about 4600 lines with $\log~gf \ge -$ 1.0, where about 400 of them have $\log~gf \ge 0.0$ . Because the transitions producing these lines occur between high-excitation energy levels that are not strongly populated, most of the lines are weak in a star like HR 6000. This large number of weak predicted lines could explain the spectrum of HR 6000 longward of about 5800 Å. The spectrum looks like it is affected by a noise larger than that due to the instrumental effects. Castelli & Hubrig (2007) explained this ``noise'' with the presence of a T-Tauri star affecting the HR 6000 spectrum. After this study, we prefer to state that the spectrum shows the presence of numerous weak Fe II lines from high-excitation levels, probably 4d, 5d, 6d - 4f, 5f, 6f transitions, which still have to be identified. The hypothesis of the presence of the T-Tauri star affecting the HR 6000 spectrum is an example of an incorrect conclusion that can be drawn owing to the use of incomplete line lists. We will extend this study of the Fe II spectrum to the near infrared region in the near future using CRIRES (CRyogenic high-resolution InfraRed Echelle Spectrograph) observations of HR 6000 and 46 Aql. The observations are scheduled in summer 2010 (ESO proposal 41380, P.I. S. Hubrig).

References

Adam, J., Baschek, B., Johansson, S., Nilsson, A. E., & Brage, T. 1987, ApJ, 312, 337 [NASA ADS] [CrossRef] [Google Scholar]
Biémont, E., Johansson, S., & Palmeri, P. 1997, Phys. Scr., 55, 559 [Google Scholar]
Castelli, F., Kurucz, R., & Hubrig, S. 2009, A&A, 508, 401, Paper I [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
Castelli, F., & Hubrig, S. 2007, A&A, 475, 1041 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
Castelli, F., Johansson, S., & Hubrig, S. 2008, J. Phys. Conf. Ser., 130, 012 003 [Google Scholar]
Cowan, R. D. 1981, The Theory of Atomic Structure and Spectra (Berkeley: Univ. California Press) [Google Scholar]
Johansson, S. 1978, Phys. Scr., 18, 217 [NASA ADS] [CrossRef] [Google Scholar]
Johansson, S., & Baschek, B. 1988, Nucl. Instr. Meth. Phys. Res. B, 31, 222 [NASA ADS] [CrossRef] [Google Scholar]
Johansson, S. 2009, Phys. Scr. T, 134, 014 013 [Google Scholar]
Kurucz, R. L. 2005, Mem. Soc. Astron. Ital. Supp., 8, 14 [Google Scholar]
Kurucz, R. L. 2009, Am. Inst. Phys. Conf. Ser., 1171, 43 [NASA ADS] [Google Scholar]
Rosberg, M., & Johansson, S. 1992, Phys. Scr., 45, 590 [Google Scholar]
Sugar, J., & Corliss, C. 1985, J. Phys. Chem. Ref. Data, 14, Supp., 2 [Google Scholar]

Online material

Table 6: Fe II lines in the 3800-8000 Å region with $\log~gf$ $\ge$ -1.5 and 3d⁶(³P)4f energy levels as upper levels.

Table 7: Fe II lines in the 3800-8000 Å region with $\log~gf$ $\ge$ -1.5 and 3d⁶(³H)4f energy level as upper levels.

Table 8: Fe II lines in the 3800-8000 Å region with $\log~gf \ge -$ 1.5 and 3d⁶(³F)4f energy levels as upper levels.

Table 9: Fe II lines in the 3800-8000 Å region with $\log~gf \ge -$ 1.5 and 3d⁶(³G)4f energy levels as upper levels.

Footnotes

... spectra: Tables 6-9 are also available in electronic form at the CDS viaanonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/520/A57
... site: http://wwwuser.oat.ts.astro.it/castelli/hr6000new/hr6000.html
... site: http://kurucz.harvard.edu/atoms/2601
...2007): http://wwwuser.oat.ts.astro.it/castelli/hr6000/unidentified.txt

Copyright ESO 2010

All Tables