3. Observed properties
Cap (HD 205 637, HR 8260) is a well-known shell star with a considerable record of long-term variations, cf., e.g., Porri & Stalio (1988) who were also the first to report central quasi-emission features in HeI 4471, 5876, 6678, and MgII 4481 in observations obtained in Aug./Sept., 1986. H exhibited emission at a peak height of 1.3 in units of the local continuum, while the absorption core reached a normalized flux of only 0.45.
HEROS spectra from 1995 show similar structures in these lines, except that the CQE was hardly visible in MgII 4481. The features are seen in most, but not all HeI lines. CQE's were not at all detected in HeI 3927, 4009, 4121, 4144, 5047, and hardly in HeI 4388. But in HeI 3820, 4026, 4471, 4713, 4921, 5015, 5876, 6678 and 7065 they were well visible. (cf. Figs. 1 and 2 for examples). While the lines without CQE exhibit purely rotationally broadened profiles, those with CQE display some unmistakable circumstellar contribution, especially in the line wings. Pure shell lines were found in FeIII 5127, 5156, SiII 6347, 6371 OI 8446, and weakly also in FeII . Shell absorption is as well seen in CaII and NaI lines, in addition to the sharp interstellar absorption cores. No photospheric wings were apparent in Paschen lines. The SiII 6347, 6371 absorptions showed a remarkable profile that is described best as a broad, shallow trough with flat bottom (Fig. 1, right panel). No CQE could be detected in typical bona fide photospheric lines such as SiIII 4553, CII 4267, or SII 5454.
In 1996, the strength of the H emission and the shell signature but also the CQE's were weaker. The remnants of these features, seen best in HeI 4471, are now shallow and broadened. However, the MgII 4481 line seems to have developed the CQE only now, although it is very weak and extremely broad compared to HeI . In FeIII 5127, 5156, and to some small extent also in SiII 6347, broad line emission has developped. Photospheric wings now do appear in the Paschen lines.
By 1998 September, the spectrum had changed rather drastically. While FeIII absorption was nearly absent, shell lines of FeII had become strong. Other probable shell lines (cf. above) also returned, but the CQE had vanished completely, so that all HeI lines showed purely rotationally broadened profiles. The shell absorption of H was weaker, while the photospheric absorption wings were filled up by line emission.
Only two months later, on November 20, the absorption core of H was again deep. As the shell absorption lines strengthened, the CQE's not only returned in the HeI and MgII 4481 lines, but for the first time in Cap they were also seen in purely circumstellar lines such as SiII and FeII lines. Quantitative data are given in Table 2.
Table 2. Parameters of CQE's in different stars and lines observed with HEROS and FEROS . Strength denotes the CQE peak height above the lowest point of the parent absorption profile in units of the local continuum. The peak heights are accurate to about 0.2 percentage points, the FWHM's to about 5 km/s. The error estimate of the radial velocities (RV) is about 10 % of the respective FWHM. Note that in Car, the sharp absorption core of FeII 5169 was measured that replaces the CQE's in FeII profiles. The RV of the broad absorption was derived by measuring the wings at about half-depth of the profile
HEROS and FEROS data, together with unpublished spectra with a resolving power of 70 000 or higher obtained with the CAT/CES on La Silla in July, 1985 and November, 1994 illustrate the long-term evolution of the CQE in HeI 6678 over many years. They are shown in Fig. 3. In this line the CQE's can exceed 4 % of the continuum flux and their strength is variable on a time scale of months to years. Both the visual inspection of the profiles and our quantitative measurements show a clear correlation between their strength and the depth or FWHM of the parent absorption profile. The CQE is stronger when the absorption line is deeper and narrower, i.e. when the envelope's contribution to the absorption is larger. A similar trend can be seen also in MgII and HeI in both Cap and Cen.
Three CAT/CES profiles from 1983 June of HeI 4471 also exhibit CQE's (in MgII 4481 even the main absorption is barely recognizable).
The decrease in the FeIII /FeII ratio suggests a change in temperature (Fig. 1). If the concommittant increase of the H line emission (Fig. 2) is due to a replenishment of the disk, the disk would be cooler when denser. This could be plausible but can be concluded with some justification only if a more detailed modelling has ascertained that the lines due to the two ionization stages do not originate from different locations at different times.
Cap is believed to be a binary with an estimated period of about 0.3 years (Abt & Cardona, 1984). The combination of all 72 spectra spanning 3.5 years supports this and indicates a period of roughly 95 days. The estimated peak-to-peak amplitude in radial velocity as derived from the CQE's is about 20 km/s.
A meaningful determination of the orbital parameters, however, is complicated by the shell-type line profile variability on similar time scales and is beyond the scope of this paper. Because of the underlying long-term variability, all measured quantities provided in Table 2 therefore refer only to a single high-quality FEROS spectrum obtained on Nov. 20, 1998.
The first CQE in Cen (HD 127 972, HR 5440) was reported by Baade (1983). However, as in the case of o And (see below), they were detected in a Balmer rather than in a metallic line. Furthermore, rapid variations seemed to be present. Sorting the published data with the photometric period of 0.6424 day (e.g., tefl et al. 1995), however, does not contradict the hypothesis of an intrinsically stable feature merely reflecting the underlying photospheric variability.
In the present data, Cen is almost a twin of Cap (cf. Figs. 1 and 2). Before 1999, the CQE's were best visible in HeI lines, in which also some shell contribution can be recognized in the line wings. The strongest shell line is FeIII 5156 in which also weak genuine emission is detected. Measurements of the temporal development of the CQE in HeI 4471 are compiled in Table 3.
In January, 1999 the H emission had become stronger. Similarly to Cap, lowly ionized shell lines of FeII and SiII had developed (see also Table 2).
3.3. o Andromedae
o And (HD 217 675 HR 8762) was the first star for which CQE was reported (Doazan 1976) but at the time taken to be genuine emission. These features were seen in Balmer lines and rapidly variable. As in the case of Cen, it is not clear if the structures were intrinsically variable or reflected underlying stellar variability. No subsequent study reported such features. However, shortly after Doazan's observations of CQE's, the star entered into a new shell phase (Fracassini et al. 1977).
CQE's can be recognized again in the spectra taken in 1998. But this time they are present in shell rather than in Balmer lines (see Fig. 4, right panel). The weak H emission with a peak height of about 1.1 looks similar to that in Cap or Cen (Fig. 2). The absorption core drops as low as 0.2 of the ambient continuum. Lines of the Balmer series can be detected up to H26 at 3666 Å in the high-quality, averaged spectrum.
As part of a long term-monitoring programme at Ritter Observatory, K. Bjorkman (1999, private comm.) observed o And in fall 1997 and fall 1998 and found both narrow shell absorption and emission peaks. In late 1998, the emission strengthened somewhat. Peters (1999, private comm.) concluded from observations of the H and H regions that "a weak shell was present in mid-January" (1999). Noteworthy is also a recent report by McDavid (1999), who within only a few months of the present observations measured the highest linear polarization ever published for this star.
Previous shell episodes are known to have started in 1966, 1975, 1983, 1988 (cf. Sareyan et al. 1992), and 1994 (Harmanec 1994). For the latter event it is also known that it was accompanied by a similar rise in polarization (McDavid 1995) as observed recently. Therefore, it seems possible that a new shell phase, or an enhancement of shell characteristics, started only a short while ago. Since shell phases of o And typically last a few years, observations in the next observing season should easily confirm or reject this hypothesis.
o And is the star in which a further typical property of CQE's can be seen best. Although the absorption component of the various parent lines may have different radial velocities, the CQE's always occur at one radial velocity common to all CQE's (see Fig. 4 and Table 2).
The multiplicity of o And (Hill et al. 1988) has most probably negligible physical influence on the phenomenon. The 33-d period only concerns companion B which orbits the Be star and another, much fainter companion a, in about 30 years at least. The period of component a is still of the order of 4 years. Therefore, averaging spectra over some weeks to months does not grossly affect any conclusion about the CQE's of the Be star. Only for measurements with very high precision do the values of the derived parameters depend on the disentangling of the component spectra and would require a dedicated study.
3.4. 4 Herculis
The first report on CQE's in the shell star 4 Her (HD 142 926, HR 5938) was given by Koubský et al. (1993), a more detailed description by Koubský et al. (1997). Being visible in many shell lines, these features are present at the beginning of new shell phases, when the star is faintest. They vanish when the H emission strengthens. It should also be noted that the shell lines are reported to be unusually broad. Compared to classical shell stars such as Tau or EW Lac (cf. Sect. 3.7), this is true of all CQE stars.
Koubský et al. (1997) do not report any indication for a relation between the 46 day period of the probable companion and the appearance of the CQE.
In the two HEROS spectra of 1998 August, CQE's are not detectable. However, it can hardly be judged if this is so as a result of the limited or because they have really disappeared (cf. Fig. 6).
Central quasi-emission peaks in Car (HD 89 080, HR 4037) were first found by Baade (1989) in his search for line profile-variability in late-type B stars.
The H profile was stable during the monitoring from 1995 to 1999. Of the stars investigated here, Car has the strongest emission with an H peak height of 2 in units of the local continuum. The central absorption is rather deep, but the minimum flux is still above the continuum. CQE's are seen in partly photospheric lines like MgII 4481, being broadest there, as well as in shell lines of TiII and CrII . However, in FeII , the most typical shell lines of Be stars, only deep pure absorption cores are visible (Fig. 4, left panel). Strengths and widths of the CQE's as observed with FEROS are given in Table 2.
Pup (HD 47 670, HR 2451) was also one of the stars investigated by Baade (1989). It exhibited a CQE quite clearly in MgII 4481. No indication of circumstellar matter was, however, found in previous studies.
The HEROS and FEROS observations do not show any trace of CQE's. In comparison to the earlier spectra of Baade, it is, however, not excluded that the MgII 4481 line was deeper during his observations due to additional shell absorption, similar to, e.g., the one in Cap and Cen (cf. Fig. 1) or to HeI 6678 in Cap (Fig. 3). But this could also be the result of imperfect normalization. Nevertheless, there is strong evidence that this star is a not previously recognized bright Be star. The variations of the H profile observed from 1995 to 1999 clearly point towards a variable amount of circumstellar matter (cf. Fig. 5); this indication is weaker but still significant in H.
3.7. Comparison with other well-known shell stars
In order to permit a comparison with other, partly more intensively studied shell stars to be made, a few additional spectra were obtained during the Aug.-Oct., 1998 observations from Calar Alto. The spectra of EW Lac and Tau show Balmer line profiles with nearly black absorption cores. The FeII 5169 absorptions reached down to only 0.5 of the local continuum and were very narrow but still resolved. Compared to the CQE stars, the other shell lines were also more strongly developed (cf. Fig. 6)
In contrast, Gem hardly shows any shell signature except in the Balmer lines and the lines of FeII multiplet 42 (4925, 5018, 5169). These FeII lines are too narrow to be resolved. The shell lines of the CQE stars are broader, incl. the FeII 5169 absorptions of EW Lac and Tau (cf. Fig. 6, left panel)
The comparison with the CQE stars reveals that, although CQE's are intimately linked to shell absorptions, CQE's are not detected in all shell stars. On the one hand, this is probably due to temporal variability as not even the known CQE stars display CQE's at all times. On the other hand, CQE's, if present, could not by the current observations of EW Lac, Tau, and Gem have been detected because their shell lines were not resolved.
Unpublished observations (by DB) at a resolving power of 100 000 and a of better than 300 with the CAT/CES on La Silla of various small wavelength regions in 48 Lib (1982 July and 1983 June) do not show traces of CQE's in the shell lines contained in these spectra, nor do unpublished HEROS spectra. Similar observations of the same star in 1989 June by Floquet et al. (1996) and in 1995 by Hanuschik & Vrancken (1996) are consistent with this finding.
By contrast, the FeII profiles for some of the stars observed in 1990 February by Ballereau et al. (1995) do include indications of a marginal CQE. Therefore, a small survey of very narrow-lined shell stars at very high spectral resolution would still be worthwhile.
© European Southern Observatory (ESO) 1999
Online publication: August 13, 199