Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 322, 554-564 (1997)

Previous Section Next Section Title Page Table of Contents

4. Spectral characteristics

4.1. HD 5980 before 1994

The spectra taken from 1989 to 1993 show similar main features, without any significant major changes. This is why we concentrate here on the description of the slit and echelle spectra obtained on 1993 September 21 and 22 ([FORMULA] = 0.30 and 0.36, according to the ephemeris of Breysacher & Perrier 1980 ). These spectra seem to show some time variations, but since this is not the main purpose of the present work, they were co-added to increase the S/N ratio. Fig. 2 presents the mean spectrum, while Table 1 gives the line measurement results for all the spectra from 1989 to 1994. The equivalent width (EW) accuracies are [FORMULA] for the brighter, non-blended lines and [FORMULA] for weaker, or blended, features. For heavily blended or very weak lines, we provide only upper or lower limits of the EW. The ions are identified in column 1, and the corresponding theoretical wavelengths listed in column 2. Column 3 indicates the emission (E) or absorption (A) nature of the lines and column 4 the corresponding EWs. Note that the sign "-" refers to emission features and that [FORMULA] and [FORMULA] refer to the absolute values of the EWs. Also, for the He i [FORMULA] [FORMULA] 5876, 6678 lines in 1994 the EW is relative to the total profile (emission + absorption).

[FIGURE] Fig. 2. Spectra of HD 5980 obtained in 1993 September using the ESO NTT+EMMI with grating [FORMULA] (top) and echelle mode (below). The normalized intensities of the strongest lines are 6.8 for He II [FORMULA] 4686, 2.1 for He II [FORMULA] 5412 and 4.0 for H [FORMULA].


Table 1. Lines detected in the spectrum of HD 5980


Table 1. (continued)

In order to classify the spectra taken from 1989 to 1993, we use the criteria given in the W-R catalog of van der Hucht et al. (1981 ), which is based on the Smith (1968a ) system. The classification essentially uses the relative strengths of nitrogen lines N iii, N iv, and N v at [FORMULA] [FORMULA] 4634-4640, 4057, and 4604-4620 respectively. We derive a spectral type of WN 6 since N iii [FORMULA] N iv, and N v is present. Assuming that SMC WN stars can be classified according to the criteria established for Galactic and LMC WN stars, we have also used the new 3-D spectral classification of Smith et al. (1996 ). In this new classification scheme, the primary indicator of ionization, i.e. the ratio between the line intensities of He II [FORMULA] and He i [FORMULA] (2.25), also indicates a WN 6 spectral type.

This means that HD 5980 had essentially the same spectral characteristics from 1989 to 1993 September including the 1990 observations by Koenigsberger et al. (1994 ). However, we note some differences between these spectra. While N iv and N v lines have comparable strengths in the 1990 and 1993 spectra, the N iii lines appear stronger in 1993. In fact, according to Smith et al. (1996 ), the ratio of N v [FORMULA] 4604 and N iii [FORMULA] 4640 on our 1993 spectra (0.09) is more typical of a WN 8 star. We also notice significantly stronger He i and He II lines with respect to the N iv and N v lines in the 1993 spectra. Particularly, the weaker He i lines at [FORMULA] [FORMULA] 4009 and 4144 are practically absent in the 1990 spectrum, whereas the He i [FORMULA] 4144 line is weakly present in the 1991 spectrum and both lines are seen in weak emission in the 1993 observations. These differences might be related to the more substantial change in spectral type that occured only two months after our 1993 September observations when HD 5980 displayed a WN 8 spectral type (Barbá & Niemela 1995 ).

Smith et al. (1996 ) use the oscillations of the Pickering decrement to infer the presence of hydrogen in the atmospheres of WN stars. In the case of the 1993 September spectrum of HD 5980 the two criteria of Smith et al. comparing respectively the H [FORMULA] and H [FORMULA] lines to the surrounding unblended He II Pickering lines, lead to contradictory results: while the H [FORMULA] /He II line intensity ratio indicates no hydrogen ([FORMULA]), the H [FORMULA] line appears stronger than expected from the He II [FORMULA] 4200 and 4541 lines ([FORMULA]) and is therefore indicative of a small but significant amount of hydrogen. However, one has to bear in mind that in the case of a rather complex system, where at least two stars contribute to a composite spectrum, the reliability of such criteria might be questionable. Finally, one has to be careful since our spectra were obtained around the secondary minimum and the line strengths could be affected by the eclipse.

On what concerns the radial velocities of the N iv [FORMULA] 4058 line and the emission component of N v [FORMULA] 4604, we find a good agreement with the radial velocity curves of Niemela (1988 ) for the N v emission and Barbá & Niemela (1995 ) for the N iv line.

4.2. HD 5980 in September 1994

The spectroscopic observations conducted in 1994 September (Fig. 3) were obtained only a few weeks after the maximum of the visual light curve of the LBV-like outburst (Bateson & Jones 1994 ) and more than one month before the observations reported by Barbá et al. (1995 ). These observations are unique since they provide information about the spectrum of HD 5980 during the very early stages of the eruption. The spectrum is dominated by strong H i and He i lines. Weaker lines of N ii, Fe iii, Si ii and Si iii are also detected either as pure emission lines or P Cyg profiles. Some detailed information on the 1994 spectra are listed in Table 1. Columns 10 and 11 give the EWs of the absorption or emission components respectively in the P Cyg features. The radial velocities of the most important lines of hydrogen and He i are given in Table 2, where [FORMULA] and [FORMULA] refer to the heliocentric velocity of the absorption minimum and emission peak respectively. Column 4 contains the relative intensities of the emission peaks of the P Cyg profiles.

[FIGURE] Fig. 3. Echelle spectra of HD 5980 obtained in 1994 September using the ESO NTT+EMMI. The normalized intensities of the strongest lines are listed in Table 2.


Table 2. Heliocentric radial velocities of the absorption minimum and emission peak and emission peak intensity of the most important P Cyg profiles seen on our September 1994 spectra. For the lines labelled with an asterisk there exists no clear absorption component and [FORMULA] is the velocity of a weak residual absorption dip in the blue wing of the emission line.

The profiles of the lower Balmer lines from H [FORMULA] to H [FORMULA] are shown in Fig. 4. Whereas H [FORMULA], H [FORMULA] and H [FORMULA] display shallow, complex P Cyg absorption components, H [FORMULA] and H [FORMULA] are nearly pure emission lines with only a weak absorption dip remaining in the blue wing of the emission line. The relatively sharp emissions in the Balmer lines suggest that an important fraction of the stellar wind moves at velocities below [FORMULA] km s-1. Most of the He i lines have P Cyg profiles displaying some morphological differences between different transitions: the lines at [FORMULA] 4120 and 4920 present a rather flat and shallow absorption component similar to the absorption of the H [FORMULA] line, whereas the lines at [FORMULA] 4144 and 4388 have a very steep red absorption wing and a lower velocity of the absorption minimum. The red edges of the emission components are between 0.7 and 0.9 times less displaced with respect to the position of the emission peak than the blue edges of the absorptions. A similar situation was found for the Balmer lines in R 127 and was ascribed to a decelerated velocity field (Stahl et al. 1983 ).

[FIGURE] Fig. 4. The lower hydrogen Balmer line profiles in the spectrum of HD 5980 as observed in 1994 September.

A closer inspection shows that the absorption component of H [FORMULA] and of some of the He i lines is split into two subcomponents, the most blueshifted one being the strongest. We measure a velocity difference of [FORMULA] 100 km s-1 for the two subcomponents of the H [FORMULA] line. From the complex profiles of H [FORMULA] and He i [FORMULA] [FORMULA] 5876 and 7065, Barbá et al. (1996 ) deduce the presence of an expanding shell at 300 km s-1 surrounding the binary system. Such absorption features have been seen in some LBVs, and described first by Stahl et al. (1983 ) in the case of R 127. Other known cases are AG Car (Wolf & Stahl 1982 , Leitherer et al. 1994 ), and HR Car (Hutsemékers & Van Drom 1991 ). The feature is generally explained as being due to the presence of shells with different velocities in the stellar atmosphere. A new explanation, put forward recently by Stahl (1996 ), ascribes it to the remnants of a higher velocity wind. In the case of a binary system, such as HD 5980, an alternative explanation for the shape of these absorption components could be some kind of interaction between the wind of the erupting component and its componion.

The strongest lines of H i and He i display very extended wings superimposed on the narrow emissions. Although the exact extent of these wings critically depends on the normalization procedure, we can measure a full width of more than about 3000 km s-1 for the strongest lines (H [FORMULA], H [FORMULA], He i [FORMULA] 5016, 5876, 6678, 7065). Such broad features are often encountered in the spectra of LBVs and related stars (see e.g. Wolf & Stahl 1982 , Stahl et al. 1983 , Hutsemékers & Van Drom 1991 ). Different interpretations such as electron scattering or the existence of a high velocity motion in the atmosphere have been suggested. In the present case, the wings are slightly asymmetric, usually more extended to the red, as would be expected from electron scattering (Auer & Van Blerkom 1972 ).

The only outstanding variations in the spectrum of HD 5980 during the 1994 September observing run ([FORMULA] = 0.68-0.79) concern the strength and to a certain extent the velocity of the He II [FORMULA] 4686 line. The EW of this line increases from -0.98 Å (Sept. 10) to -1.52 Å (Sept. 12) whereas its radial velocity as measured by gaussian-fitting varies from 66 to 129 km s-1.

Comparison of these spectra with those obtained by Barbá et al. (1995 ) one month later reveals some differences. In 1994 September the He i lines show well-developed P Cyg profiles, while in October this feature is missing in most of the He i lines. Interestingly, the He i lines that have conserved a rather weak P Cyg profile in 1994 October (at [FORMULA] 3926, 4009, 4144, 4169, 4388, 4438, 4922, 5048) are generally those with significantly less blue-shifted absorption components in our spectra. P Cyg profiles are lacking also for the H [FORMULA] and H [FORMULA] lines in the October spectrum. This is probably due to an enhancement of the emission components in October. In fact, we notice a substantial increase of the relative intensity of the emission lines between September and October. For example the H [FORMULA] emission has a normalized intensity of 6.3 in our spectra whereas it reaches 10.1 one month later. The He II [FORMULA] 4686 line that has a mean EW of -1.31 Å on our spectra has increased its EW to -2.6 Å in 1994 October. Subsequent spectroscopic observations obtained during the declining phase of the outburst indicate a much faster increase in the EW of the He II [FORMULA] 4686 line between October and December 1994 (Barbá et al. 1996 ).

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998