Astron. Astrophys. 333, 125-140 (1998)

4. Phases of a line emission outburst

4.1. Relative quiescence

According to Fig. 3, the emission strength of µ Cen is changing most of the time. Therefore, one can only define phases of relative quiescence when the variations are minimal. Even during such relative quiescence the Balmer emission slowly decays. This definition is independent of the emission strength.

In our observations, we have found the end of such a phase in 1996 at Modified Julian Date (MJD JD-2 400 000.5) , nearly 80 days after the last major event had occured. We will base the following description on this period, but the spectra during the relative quiescence before MJD 49 785 look essentially the same. The small differences might be explained by the shorter time that had elapsed since the preceding outburst.

• Balmer lines: At this stage, the emission profiles have sharply defined edges, and the central reversal is very pronounced. The -ratio is only slightly variable about unity.
  The contribution of extended emission to the line wings is quite weak, so that the photospheric absorption wings are still visible. The process responsible for the extended wings might be Thomson scattering in this phase. The quiescent -line is shown in Fig. 5.
• Paschen lines: The emission in the Paschen series is shallow in the spectral range observed (Paschen discontinuity to Pa₁₄). The peak height of Pa₁₅, the blue side of which is slightly blended with Ca ii 8542, does not exceed = 1.1. This level of the Paschen emission during quiescent phases may be crudely recurrent whereas the strength of the Balmer emission during quiescence takes on a much larger range of values. The central reversal is much less pronounced than in the Balmer series, though detectable. Apart from the Paschen lines, three other lines within the Bracket continuum were detected in the observed spectral range: Ca ii 8498,8542 and O i 8446. The first two are weak and moreover blended with Paschen lines. But the latter is stronger than one would expect judging from the O i 7774 blend. It is a fluorescence line of the transition (Briot, 1981), and its strength should, therefore, scale with . The quiescent Pa₁₅ line profile is shown in Fig. 5.
• Silicon: Weak Si ii 6347 emission is present (Fig. 5) whereas Si ii 4131,5056 are missing. Of the two main phases of relative quiescence in 1995 and 1996, the later one was weaker in Si ii 6347 emission. As can be seen in Fig. 6, also the peak separation was smaller.
• Iron: During relative quiescence, Fe ii emission is not detectable down to the noise limit which, given the large number of potential lines, corresponds to a peak height of half a percent above the continuum level.
• Helium: The emission profiles of He i are highly variable in both shape and strength even at quiescence. However, the absolute level of activity and the velocities of the emission peaks are generally lower than during an outburst, and the emission peaks are narrower than in other stages.

4.2. Precursor phase

An imminent outburst announces itself by a short (5-13 days) significant (see Sect.  3and also Fig. 1) drop in the peak height of all emission lines present at the time. A second defining constituent of this phase is an increase of the extended emission wings which is best visible in higher Balmer lines. In these wings may compensate the change in equivalent width caused by the drop of the peak height, however not quite in all cases. The balancing is more perfect in H , and from H on the wings grow more strongly than the peaks decrease.

In most outburst cycles, this phase is the most distinctly recognizable feature; a proto-typical example is displayed in Fig. 3 around MJD 50 510, but the reduced -peak height is also well visible in Fig. 5. This behaviour is not peculiar to µ Cen as is shown by the case of the Be star HD 76 534 in which Oudmaijer & Drew (1997) observed the full recovery of a previously drastically reduced H emission in only 3 hours.

Fig. 5. The variations of (leftmost), Pa15 (mid left), Si ii 6347 (mid right), and He i 6678 (rightmost), from relative quiescence (lowermost spectrum, MJD 50 180) over the very first trace of circumstellar variability, the beginning of the precursor (second, MJD 50 184), the high velocity absorption event as described in Sect.  6(third, MJD 50 185), and the appearance at the end of the burst phase (, fourth, MJD 50 195) to the late relaxation phase (, fifth, MJD 50 204). The dates of the spectra are indicated as arrows in Fig. 3. Two similar high-velocity absorption events (Table 4 and Sect.  6) in January 1997 are overplotted as dotted lines on He i 6678 (MJD 50 457 in the lower and MJD 50 461 in the upper spectrum). In the lower row we show how the dynamical spectra develop from relative quiescence through a major burst into the relaxation phase. The dynamical spectra also cover two smaller bursts around MJD 50 223 and MJD 50 232. The apparent long-term radial-velocity variations in the He i 6678 line are due to the highly uniform sampling of one spectrum per day, which results in a strong beat pattern with the photospheric periodicity we describe in Paper II. The feature in the Pa15 line seen in the dynamical spectrum at is a CCD artifact

Fig. 6. The peak separation of the Si ii 6347 () and Fe ii () lines. For 1995 (top), the peak separations of Fe ii 5169 and Fe ii 5317 are averaged. For 1996 (bottom), only Fe ii 5169 was measurable with sufficient accuracy. Note that Fe ii emission becomes detectable only at the peak of an outburst (Fig. 2)

Because no photometry was obtained parallel to the HEROS observations, the possibility cannot be directly dismissed that the precursor drop of the ratio is caused predominantly by a corresponding increase of the continuum flux. In the precursor phase of the outburst on MJD 50 515, which was the deepest observed, the H value temporarily dropped from 3.0 to 2.4. If the decrease was caused only by a change of the continuum flux, a brightening by nearly 0. would be implied. This is more than virtually all optical peak-to-peak light variations reported by Cuypers et al. (1989), Balona (private communication), and the HIPPARCOS photometry (Perryman et al., 1997) for the period 1987 to 1992. Taking into account also the frequency of the outbursts and the density of the photometric data, such a large change in the continuum flux appears improbable.

Another test is to check the amplitudes of various lines for consistency with pure continuum variability. Unfortunately, there are too few metal lines that are of sufficient strength throughout the outburst cycle and to which the test can be applied. However, accompanying major Balmer decrement variations (Rivinius et al., in preparation) indicate that the results of this test, too, would be negative.

4.3. Outbursts

Baade et al. (1988) present arguments that previous rapid increases in line emission from µ Cen were due to the ejection of material by the star to its circumstellar disk. This justifies the notion of outbursts. Hanuschik et al. (1993) distinguish between major and minor outbursts which can be identified by their strength. We find two major and many minor bursts in our data but no compelling evidence that they are genuinely different events. The difference between bursts of different strengths may be blurred further by the dependency of the appearance of bursts on the mean emission strength. Outbursts tend to look more uniform at times of stronger underlying disk emission, and we note that Hanuschik et al. derived their classification from data obtained when there was not yet a new persitent disk.

Of the two major outbursts, the one in 1995, the major burst unfortunately started shortly before a gap in our observing schedule (cf. the description of the behaviour of the Fe ii lines given below). Both major outbursts were sampled with one spectrum per day. At higher temporal resolution only minor bursts were observed with HEROS ; the series of Boller&Chivens spectra was probably also obtained during a minor outburst (Sect.  5.2).

• Balmer lines: In 1996, the peak height of dropped after the short precursor phase within a few days by 0.3 in units of the local continuum, while the equivalent width change is roughly compensated by the increasing wings (indicated by the dotted line in Fig. 3, 1996 panel). A comparison between different Balmer lines shows that the increase of the wings is not correlated with the strength of the central emission. Therefore, Thomson scattering is questionable as a major contributor to the emission wings. The rapid variability may occasionally have started already even in the precursor phase, depending on the strength of the burst. The emission wings still continued to rise and, depending on the level of the quiescent emission, may finally also in overcompensate the decrease in emission strength caused by the loss in peak height.
  During the 1995 HEROS observations a major burst occurred, but unfortunately very close to a gap in the observing run, when the instrument was for two weeks off the telescope. Nevertheless, the onset of the burst on MJD 49 801 can be detected both in the equivalent widths (the dotted line in Fig. 3, 1995 panel) and in the dynamical spectra.
  Owing to the increasing optical thickness of during the past few years, the distinction between major and minor bursts becomes on the basis of the variation of the H emission less clear than before when only a non-persistent emission disk surrounded the star. However, this classification can still be retained if applied to other lines within the HEROS wavelength range, for instance to those of Fe ii described below.
• Paschen lines: The Paschen series exhibits the most drastic changes during an outburst. The emission strengthens by several Å, which is due not only to a strengthening in peak height, but also to strongly enhanced wings. The central reversal has nearly disappeared and the peak height of the line is of the order of . The O i 8446 line is not showing this strong variability. It has nearly not changed its peak height and is only slightly broader than before so that after an outburst it is weaker than the neighbouring Pa₁₈ line, whereas it is stronger during quiescence. Similarly to , major and minor outbursts are distinguishable by the amplitude of the increase in emission as long as the persistent emission is not too strong. This was the case in 1995 and 1996, while in 1997 the distinction was less clear.
• Silicon: The Si ii-emission becomes visible in 4131,5056 and strengthens for 6347. The peaks are at higher velocities compared to the stage of relative quiescence. In Si ii 6347 also remnants of the emission from the previous outburst are still visible at lower velocities. This behaviour is displayed in Fig. 5 as a dynamical spectrum and is reflected in the peak separation which seems to reach high values immediately (Fig. 6). It can also be seen that major outbursts are far more efficient in driving up the peak separation than are minor ones.
• Iron: At the peak of major outbursts, the Fe ii emission appears rather suddenly and within 10 days attains its maximum strength (cf. Fig. 2). It then persists throughout the outburst phase and shows a similar behaviour as the SiII -emission. The distinction between major and minor outbursts is clearest for these iron lines: only a major burst leads to persistent emission, while the emission of a minor burst, if present at all, vanishes as the burst ceases.
• Helium: In He i 6678 additional emission shows up at relatively high velocities. The differences between major and minor bursts are less strong in helium than they are in lines formed farther away from the star.

Finally, it should be pointed out that the pre-outburst loss in emission peak height (Sect.  4.2) is not always quite compensated for during the outburst and that the recovery also takes part of the relaxation phase (Sect.  4.4). In rare cases, a small net loss may remain well into the next outburst cycle.

4.4. Relaxation

In the two small bursts observed with HEROS at sufficiently high temporal resolution (see Sect.  5) we find that the variability is most pronounced on the ascending branch of the emission strength curve. We therefore adopt the settling of the rapid -variability as the end of a burst and the beginning of the following phase to which hereafter we refer as the relaxation phase.

• Balmer lines: The wings of the emission lines attain their maximum strength at the beginning of this phase. No trace of the photospheric profile is detectable.
  The decline of the wings sets in slowly while the emission peak height finally increases. This increase may continue for weeks, eventually turning over into a slow decrease when relative quiescence develops.
  Although the -variability has stopped, the ratio is still not unity. There is a general trend for the red peak to remain lower than the blue one by several percent. Only after 10 to 20 days (for major outbursts), the average -ratio finally reaches unity. Line profiles that are representative of these stages are provided in Fig. 5.
• Paschen lines: They are initially also still dominated by the broad wings (Fig. 5). During the course of a couple of days the wings fade slowly, the peak height decreases quite linearly, and so the shapes of the profiles asymptotically approach their quiescence appearance.
• Silicon and iron: Fig. 6 shows the measured peak separation for Si ii 6347 and two Fe ii lines. The separation attains high values immediately after a major burst and then shrinks until the next major burst. Subsequent minor bursts only have little influence on the peak separation. An example of this behaviour is given in Fig. 2 in the form of a dynamical spectrum for Fe ii 5169.
• Helium: Except for a further weakening and narrowing of the emission components and ongoing variability, the helium lines seem to be the first to reach the quiescent phase.

© European Southern Observatory (ESO) 1998

Online publication: April 15, 1998
helpdesk.link@springer.de