Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 326, 1111-1116 (1997)

Previous Section Next Section Title Page Table of Contents

3. The emission line spectrum

The present spectrum of P Cygni is notable for the great number of pure emission lines. Most of them are permitted lines of FeIII, NII, SiII and probably AlIII. Further pure emission lines are the forbidden lines of [FeII], [NII], [NiII] and probably [TiII].

A simple qualitative comparison of the identification lists of Beals (1950), de Groot (1969), Ozemre (1978), Markova (1994) and Markova & Zamanov (1995) as well as an inspection of the spectral atlas published by Stahl et al. (1993) show that more than 70% of the pure emission lines in P Cygni's spectrum (wavelength range from 3500 to 6800 Å) appear to be new.

3.1. Permitted emission lines

Central emission velocities, Ve, averaged within the framework of each ion and multiplet, are summarized in Table 1. One can see that the first three values of Ve fall within the 3 [FORMULA] range while the emission radial velocity of the FeIII lines with P Cygni-type profiles is systematically higher. Also higher, i.e. more negative, are the velocities of the pure emission lines.


Table 1. Mean values and standard deviations for radial velocities of emission lines of different ions

According to Barlow et al.(1994) and Stahl et al. (1991) P Cygni's radial velocity is -22.6 [FORMULA] 1.5 and -22 [FORMULA] 4 km/s, respectively. These values were obtained by analysing flat-topped profiles of [NiII] and [FeII] forbidden lines. Both estimates are higher than Ve derived from HI, HeI and NII lines but in excellent agreement with Ve derived from FeIII lines (see Table1). The observed differences are indeed small but appear to be real as they are above the 3 [FORMULA] level. A possible explanation for this result is that the central emission peak of the P Cygni profiles of the lines of abundant ions (HI, HeI and NII) are distorted by "photospheric" absorption. At the same time the FeIII lines with P Cygni - type profiles have not been affected by such absorption since they form entirely in the higher levels of the wind. Speaking about "photosphere", we mean those wind layers where a mean optical depth [FORMULA], derived from an average over all wavelengths longer than 912Å, drops to 2/3, namely R [FORMULA] = R ([FORMULA]) for [FORMULA] 912Å. Irrespective of its simplicity, the above given explanation meets with some difficulties in the case of a star as extreme as P Cygni, whose wind is much denser than in normal supergiants. For example, we know from the observations that in the photographic region of P Cygni's spectrum no line of photospheric origin, i.e. centred at Vsys, has been observed. In the UV the situation is similar with the exception of the resonance lines of the highly ionized species, CIV and SiIV, whose absorptions are centred at their laboratory wavelength (Cassatella et al. 1979) thus pointing to a possible formation in wind layers near R [FORMULA]. Therefore, the central emission peak of the observed P Cygni profiles of the HeI, HeI and NII lines would be distorted by "photospheric" absorption only if during our observations the wind was more transparent. Another possibility is the existence of large turbulent motions in the region where the expansion velocity is still small.

Adopting Vsys = -22km/s, we also conclude that the permitted pure emission lines in the spectrum appear to be blue shifted (see Table 1). In fact, different multiplets of the same ion can have different Ve. The observed differences can not be fully explained by inaccuracy of the position measurements since most of the studied profiles have a triangular shape with a well defined and sharp central peak. No obvious correlation between the velocity and excitation energy of the upper level of multiplets of the same ion was found.

Various parameters of the studied pure emission lines are listed in Table 2. The full width at half maximum, FWHM, and the edge velocities, V [FORMULA] and V [FORMULA], of the lines were determined from a gaussian fit. These last two quantities were measured with respect to the position of the corresponding emission peak. Most FeIII and NII lines included in the sample, show similar values of FWHM, V [FORMULA] and V [FORMULA]. The estimates range from about 65 to about 100 km/s for FWHM and from about 70 to about 100 km/s for the modulus of the edge velocities. A number of lines, however, deviate from these values of the line parameters (In Table 2 these lines are marked with an asterisk). In the majority of those cases there is a simple explanation for the aberrant behaviour: influence of blends or rather strong photographic noise interferes with both the fitting of the profiles and the determination of the continuum level. By averaging the data for the FeIII and NII lines, mean values of 84 [FORMULA] 3, -80 [FORMULA] 2 and 84 [FORMULA] 2 km/s were obtained for FWHM, V [FORMULA] and V [FORMULA], respectively.


Table 2. Permitted emission lines in P Cygni's spectrum. Ve [in km/s] is the velocity of the emission peak. V [FORMULA] and V [FORMULA] [in km/s] denote the velocities of the blue and red edge of a line, measured with respect to its emission peak. "I" is the central emission intensity, measured in units of the nearby continuum; an asterisk " [FORMULA] " marks values which are uncertain


Table 2. (continued)

At the same time the SiII emission lines (multiplets 2 and 4), with their mean FWHM of 194 [FORMULA] 1 km/s and mean edge velocities of- 236 [FORMULA] 5 and 231 [FORMULA] 7 km/s, are considerably broader than the FeIII and NII lines. Multiplet No.5 of SiII is a little narrower than multiplets 2 and 4 but still broader than FeIII and NII. The line parameters of SiII (multiplet No.1) were found to be quite different from those of the others SiII multiplets. But these estimates are less reliable since both profiles are affected by blends.

According to Beals (1950) and de Groot (1969) the presence of AlIII is possible but not certain. On our spectra (Markova 1994, Markova & Zamanov 1995) and those of Stahl et al. (1993), four lines of AlIII (multiplets 3, 5 and 6) appear to show emission or P Cygni-type profiles that are stronger in emission than in absorption. At the same time, the AlIII [FORMULA] 5696 line (multiplet No. 2) clearly shows a P Cygni-type profile with an absorption component stronger than the emission line. But a possible blend with SiIII [FORMULA] 5696 cannot be excluded in this case. Summarizing, we conclude that the AlIII spectrum is certainly present in emission and possibly also in absorption in P Cygni's present-day spectrum. We obtain a mean value of -40 [FORMULA] 5 km/s for the emission velocity of these lines. Unfortunately, it was not possible to derive other line parameters from our data.

Most of the permitted pure emission lines in the spectrum appear to be symmetric with a sharp maximum. In some cases (FeIII [FORMULA] 5300, [FORMULA] 6032, [FORMULA] 5833 and SiII [FORMULA] 6371) a weak emission shoulder can be seen on the blue or red side of the profiles (Markova & Zamanov 1995). Although no appropriate identifications have been found for all features, we believe that they are not due to emission in the corresponding lines since not all lines of the same multiplet show these features.

3.2. Forbidden emission lines

Data concerning the [FeII] spectrum are not listed in Table 2 since similar data were already published by Stahl et al. (1991, 1993). Here we shall only mention that, based on six [FeII] lines, mean values of -18 [FORMULA] 2 and 231 [FORMULA] 3 km/s were derived for the central velocity, Vc = (V [FORMULA] + V [FORMULA])/2, and the half width at zero intensity, HWZI =  ([FORMULA] V [FORMULA] [FORMULA]  + V [FORMULA])/2, of the lines, respectively. These values are in excellent agreement with the results of Stahl et al. (1991), namely -22 [FORMULA] 4 and 230 km/s. This, then, is further evidence of the reliability of our radial-velocity data.

As Stahl et al. have noted, the [FeII] lines are flat-topped showing that they are both optically thin and formed in a region of constant expansion velocity, Vexp =HWZI=230 km/s. The complex structure of some [FeII] lines is probably due to blends with permitted lines of different ions (Markova 1994, Markova & Zamanov 1995; Israelian 1995). However, since the blends are weak and since the spectral resolution of our data is 20 to 40 times higher than the line widths we concluded that the HWZI for these lines could provide an almost exact measure of Vexp. This velocity is considerably smaller than the terminal velocity of the wind, Vterm = 460, 400 and 311 km/s as determined from the blue edge of the UV absorption lines by Hutchings (1979), Underhill (1979) and Cassatella et al. (1979), respectively, and a little higher than the terminal velocity of 210 km/s determined by Lamers et al. (1985). Prinja et al. (1990) studied a significant sample of O and B supergiants and found that the terminal velocity is equal to the sum [FORMULA] VDAC [FORMULA]  + HWHIDAC where VDAC is the mean velocity of the discrete components observed in unsaturated P Cygni profiles, and HWHIDAC is the mean value of the half-width at half-intensity of these components. In the case of P Cygni, discrete components were observed in the UV (Lamers et al. 1985) as well as in the optical part of the spectrum (Markova 1986). The corresponding values of the two quantities, VDAC and HWHIDAC are -206 [FORMULA] 2km/s and about 30 km/s (from the UV FeII lines), and -211 [FORMULA] 1.4 km/s and about 25 km/s (from the higher members of the Balmer series). This gives a value of 236 km/s for the sum [FORMULA] VDAC [FORMULA]  + HWHIDAC and, consequently, for the terminal velocity of the wind. This value is a little higher than the estimate of Lamers but it is practically the same as Vexp derived from [FeII] flat - topped profiles. Thus, we conclude that the [FeII] lines are formed in a region expanding at the terminal wind velocity. Stahl et al. (1991) came to the same conclusion but they used the estimate of Lamers (1985).

Further forbidden lines in P Cygni's spectrum are those of [NII] multiplet Nos. 1 and 3. We confirm the result of Stahl et al. (1993) that the yellow line of [NII] ([FORMULA] 5755) shows a rounded, blue-shifted (Vc =-40 km/s) and slightly narrower (HWZI=200 km/s) profile. The greater strength of the line makes the derivation of its parameters more reliable than in the case of the [FeII] lines. The presence of the two red lines of [NII] is not obvious but, in our opinion, they are present in the averaged spectrum (Markova & Zamanov 1995). Symmetrizing the H [FORMULA] profile with respect to its emission centre and accounting for the existence of water vapour lines, we derive a profile of the [NII] [FORMULA] 6548 line that is centred a little above Vsys and has a HWZI of about 190 km/s. The similarity in the line-parameters of the [NII] [FORMULA] 6548 and [NII] [FORMULA] 5755 is obvious. The extraction of the profile of the other red line, [NII] [FORMULA] 6584, is more complicated since it is influenced, not only by the H [FORMULA] emission wing but also, by the P Cygni profiles of the two lines of CII (mult. No.2). Taking into account:

  • the extent of the red emission wing of the CII [FORMULA] 4267 line (free of blends) and assuming the same extent for the red emission wings of the two CII lines of multiplet 2;
  • the existence of absorption in CII [FORMULA] 6578 and the lack thereof in CII [FORMULA] 6583,

we come to the conclusion that the [NII] [FORMULA] 6584 line is also present in our spectra. Of course, in this way it is not possible to determine the shape of the profiles and their line parameters with confidence. Therefore, the above results have to be regarded as somewhat rough estimates.

3.3. Origin of the permitted emission lines

In Table 3 we list the multiplets seen in emission in P Cygni and the excitation energies of their upper and lower levels. We observe the following:


Table 3. Different multiplets seen in emission in P Cygni; e [FORMULA] and e [FORMULA] denote the excitation energy of lower and upper level, respectively

  • the excitation energies of the upper levels of the listed FeIII, NII and AlIII transitions (except multiplet 1 of AlIII) fall into three groups ranging from 20.32 to 20.79 eV, from 23.14 to 23.57 eV and from 24.98 to 25.35 eV;
  • the excitation energies of the upper level of the SiII transitions are grouped around 10 and 12 eV;

A preliminary result obtained on the basis of the SAC method (Friedjung & Muratorio 1987 and Baratta et al. 1995), applied to the emission lines of FeIII and NII lines with P Cygni and pure emission profiles shows that the NII levels above 23eV as well as FeIII levels above 20eV seem strongly overpopulated, suggesting that NLTE-effects have become important (Markova & Muratorio 1997).

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1997

Online publication: April 8, 1998