3. The Fe II spectrum
Fe II is, as discussed above, present with sharp absorption and emission features in the investigated IUE spectra of KQ Puppis, and is by far the most prominent of the iron spectra. This is partly explained by the fact that the strong Fe II resonance lines fall in the IUE region, whereas resonance lines of Fe I and Fe III fall in the visible and well below the lower wavelength limit of IUE, respectively.
In this work we have chosen to concentrate on the Fe II lines and their excitation mechanisms. There are different line shapes for the Fe II lines, ranging from pure absorption to pure emission through P-Cygni character of various degrees. We observe that lines with small gf values appear in emission but seldom in absorption. Lines with larger gf values appear with any of the three profiles depending on the level population. We concentrate on pure emission lines and do not discuss the P-Cygni type profiles.
A pure blackbody intensity distribution for K peaks at around 1000 Å. The main part of the continuum radiation from the hot companion star ( K) is therefore emitted in the region of the Fe II resonance lines. This led us to the hypothesis that photoexcitation by continuum radiation (PCR), see Fig. 1, is the dominating excitation process for the observed Fe II emission lines in KQ Puppis.
We have looked through the IUE spectra of KQ Puppis in detail, and all Fe II emission lines identified are presented in Tables 1-4. For each of the lines we give the laboratory wavelength and a wavelength shift, which is the difference between the measured position of an emission feature relative to the wavelength scale defined by absorption lines, see Sect. 2. We also give the excitation potential for the upper level and the photoexcitation channels, e.g. a6D - x6P in Fig. 1. Excitation channels with wavelengths below the lower limit of IUE ( Å) are proposed, but have evidently not been observed. Wavelengths in Table 1 above 3100 Å are from the observation LWP 23117 recorded at a different phase. The wavelength shift between absorption and emission features is smaller in LWP 23117.
Table 1. Fe II PCR emission lines. is the difference between the observed emission wavelength and the wavelength scale of the sharp absorption lines. Wavelengths above 2000 Å are in air.
Table 1. (continued)
Table 1. (continued).
Table 2. Fe II emission lines from the lowest odd parity levels, populated by PCR, some also by cascading from higher levels. is the difference between the observed emission wavelength and the wavelength scale of the sharp absorption lines. Wavelengths above 2000 Å are in air.
Table 3. Fe II lines excited by H Ly PAR process. is the difference between the observed emission wavelength and the wavelength scale of the sharp absorption lines. Wavelengths above 2000 Å are in air.
Table 4. Fe II lines excited by secondary fluorescence. is the difference between the observed emission wavelength and the wavelength scale of the sharp absorption lines. Wavelengths above 2000 Å are in air.
The PCR emission lines in KQ Puppis are formed through the absorption of continuum radiation in the Fe II resonance region. All absorption lines that are expected to be present, based on oscillator strengths and lower level excitation potential and population, are observed and all the observed PCR lines are given in Table 1 and Table 2. We have found no inconsistencies with the idea of PCR as the dominating excitation process. The other processes forming Fe II emission lines is the PAR by H Ly (Table 3) and secondary fluorescence (Table 4).
We have found absorption lines from levels up to around 2.0 eV, where the 3d7 a2G levels are barely populated. The weak emission lines at 2982, 2997 Å (b2D - y2F) appear as fluorescence due to the a2G - y2F PCR process. Levels above 2.0 eV appear not to have a thermal population sufficient to give rise to absorption lines. However, there is one term (b4F) at 2.8 eV that shows absorption, but this term (b4F) we propose to be populated by fluorescence from x4D, y4D and z4G, see Table 1. Furthermore, a high 4G term ((4G)4s4p(1P) 4G) is populated from this b4F term. All the strong (diagonal) lines of this b4G - (4G)4s4p(1P) 4G multiplet are observed, and this suggests a broad PCR process. Notice, that the somewhat unexpected population of the b4F term leads to emission from this high 4G state. There could be other excitation channels involved in this, but we have not been able to establish any.
The emission lines within multiplets UV 193 and UV 191 appear with different relative intensities, see Fig. 2. The intensities are expected to be proportional to for the upper 6P levels. In UV 193 the intensities are reversed with the line having the largest J (1473 Å) being the weakest, but this is not unusual for this multiplet, as it has also been seen in laboratory spectra. In UV 191 the 1785 and 1786 Å lines appear to be equally strong and the 1788 Å line is almost missing. However, this can be explained by the 1785 Å being saturated and that the 1788 Å is perturbed by both a Ni II absorption line and an IUE reseau mark. Notice also that the parasite line to UV 191 discussed in Johansson et al. (1995) does not appear in the tables. This line is not resolved in the IUE spectrum.
We have chosen to treat emission multiplets having a4D or a4P as the lower term separately and present them in Table 2. These lines have pure emission profiles, in spite of a large population of the lower levels. It is the relatively small gf values for these transitions that give rise to the pure emission profiles. The upper levels of these transitions in Table 2 are consequently populated by photoexcitation from "another" of the four lowest terms, but in some cases also by cascading from e4D and e6D, see Table 2.
A third category of emission lines are presented in Table 3. These are the H Ly PAR process lines. H Ly pumps the upper levels of the three transitions in Table 3. The 1869.5 Å line is produced by pumping from a4G into the high (a2F)4s4p(3P) 4G state. This process was first described in Johansson & Jordan (1984) where the process enabled the authors to establish the (a2F) 4s4p(3P) 4G term. The strong emission line - b4G ( 9997.56 Å) was later suggested by Johansson (1990) to be a secondary fluorescence and thus a confirmation of this process. See also Judge et al. (1992) and references therein for a detailed discussion of Fe II excitation mechanisms for low gravity cool stars.
The second H Ly PAR process is the pumping of the mixed 3d6(5D) and 3d6 (b3F)4p levels from a (Johansson & Jordan 1984), which gives two fluorescence lines, 2506.8 and 2508.3 Å, see Fig. 3. The - 3d6(5D)5p (1217.9 Å) pump channel is in closer resonance with H Ly than the - 3d6(b3F)4p (1218.2 Å), see Fig. 3. This favors the - 3d6(5D)5p (2506.8 Å) fluorescence line. However, in KQ Puppis both emission lines have about the same intensity even though the 2506.8 Å feature in the spectrum is blended with an emission component of UV 175, see Table 1. This shows that the width of Ly in KQ Puppis is large enough to encompass both (1217.9 and 1218.2 Å) excitation channels, see Fig. 3.
The fourth category of emission lines, presented in Table 4, is secondary fluorescence lines. These emission lines from the e4D, e6D terms are often appearing in astrophysical spectra, and they are also present in the case of KQ Puppis. We suggest an explanation for the presence of these secondary fluorescence lines based on two different schemes. Many of the 3d6(5D)5p levels have strong absorption channels around 1000 Å from the ground term and from the lowest quartet terms. This is the wavelength region where the continuum radiation peaks, and we suggest that PCR from these low terms are populating the 3d6(5D)5p levels. Levels of the e4D, e6D terms are then populated by the normally strong IR transitions from these 3d6(5D)5p levels, see Table 4. The other scheme that is adding to the population of the e4D, e6D levels is selective in J. Levels with J-values 7/2 and 9/2 are populated by emission from 3d6(5D)5p , which is populated by PAR, as discussed above.
3.1. Comments on earlier work
In this work we concentrate on Fe II, but during our analysis of the IUE spectra we have noticed that many of the lines left unclassified by Altamore et al. (1982) are due to Ni II. Nickel is next to iron the most prominent of the iron group elements in this spectrum and appears in both absorption and emission.
There is a discussion in Altamore et al. (1982) of the Mg II resonance line where the k line is claimed to be asymmetrical. The reason given is a superimposed absorption by a Fe I line at 2795.01 Å. However, this is the a5D4 - z3G4 transition, which is unlikely to appear in absorption. A more likely absorption in Fe I occurs at 2795.540 Å, a5F4 - y5G3. The Fe I 2795.540 Å line is the absorption part of a Mg II pumped PAR process, first proposed by Thackeray (1937). The PAR process yields primarily the 2843.976 Å and the 2823.276 Å Fe I emission lines of UV 44, see also Gahm (1974) and Johansson & Hamann (1993). However, we have not in this work been able to identify either of these PAR emission lines or other Fe I lines, not even the 2823.28 Å emission line identified by Altamore et al. (1982).
Intensity anomalies in the KQ Puppis spectrum are discussed in Muratorio et al. (1992), e.g. the strong emission lines a6S - x6P (UV 191), - (5D)5p (2506.8 Å) and - (b3F)4p (2508.3 Å). Dielectronic recombination is proposed to enhance UV 191, whereas H Ly pumping generates strong fluorescence in the latter two transitions. We support the latter proposal, but not the former. Dielectronic recombination has earlier been suggested as a possible explanation for the enhancement of UV 191 (Johansson & Hansen 1988), but this is mainly to be considered for cool giants where there is no continuum radiation in the UV 9 channel. Dielectronic recombination is the inverse of autoionization, and requires in this case an adequate abundance of Fe ions and suitable autoionizing states of Fe II. Some Fe III 4s - 4p transitions are present in the spectrum as broad absorption features with a shift of about 0.5 Å compared to the sharp features of Fe II, but the main part of the Fe III opacity is below the IUE limit (Johansson & Cowley 1988). The actual abundance of Fe ions in the region where Fe II emission lines are formed is thereby difficult to estimate. Two autoionizing terms recombining to x6P, with transitions around 1730 Å and 1890 Å respectively, are theoretically predicted by Johansson & Hansen (1988). The calculated wavelength from a theoretical prediction is uncertain, but recombination lines would have to appear somewhere when dielectronic recombination is an active process. However, there are no candidates in the investigated part of the KQ Puppis spectrum. For these reasons we do not support the scheme of dielectronic recombination as the population process for x6P, but rather PCR through UV 9 as is also proposed by Che & Reimers (1983)
Che & Reimers (1983) observe emission in UV multiplets 60, 78, 191, 363, 373, 391, 399. They also claim to observe emission in UV multiplets 375 and 384 but that has not been verified in our work. The upper term in UV 375 (e6F) belongs to the 3d6(5D)4d subconfiguration (Johansson 1988) and we do not observe any lines in the 3d6(5D)4p - 3d6(5D)4d super multiplet. Only transitions in the 3d6(5D)4p - 3d6(5D)5s super multiplet are observed. The PCR process populating the 5p levels leads to a selective population of the 3d6(5D)5s subconfiguration.
In UV 384, as defined by Moore (1952), the upper term (e6P) has been found to be spurious and a new identification was given by Johansson (1978). The lines from this new 6P term are not present in the IUE spectra. The paper by Che & Reimers (1983) also discusses the Fe II level population mechanism in the stellar system and propose that the systematic variation in line-profile (absorption - P-Cygni - emission) with excitation energy suggests recombination followed by cascading. They also point out the importance of the relative oscillator strengths of the different channels from a populated level. We agree in general with Che & Reimers (1983). However, we suggest, as discussed above, that all levels giving rise to emission lines are populated through photoexcitation, PCR and PAR, and that recombination is not necessary for the explanation of Fe II emission lines in KQ Puppis.
© European Southern Observatory (ESO) 2000
Online publication: January 29, 2001