Astron. Astrophys. 347, L39-L42 (1999)

4. Spectral evolution

In comparison to previous outbursts our observations begun much earlier, significantly extended toward later phases and have been performed at a higher resolution and over a broader wavelength range (see Fig. 2).

Fig. 2. Spectroscopic evolution of the 1999 outburst of U Sco from some of our spectra. Numbers are days since maximum brightness. The spectra are normalized to 1.0 at 5500 Å and shifted to avoid overplotting. H profiles are plotted on an expanded scale in Fig. 3.

As for previous outbursts (cf. B81, S88), the spectrum has been characterized by very wide emission lines, with Balmer hydrogen lines being the strongest at earliest phases (in quiescence hydrogen lines are generally absent and mimicked by the Pickering series of HeII, cf. Hanes 1985 and Johnston & Kulkarni 1992).

A feature not reported for previous outbursts (perhaps due to the poorer resolution and looser time coverage) is the monotonic decrease with time of the FWZI (full width zero intensity) of H (and similarly for the other Balmer lines; cf. Figs. 2 and 3). The values in Table 1 are plotted in Fig. 4 where the linear fit is given by the equation:

This linear decrease is difficult to explain in term of ejecta deceleration by circumstellar material because the spectra carry no sign of the typical signatures that characterize the presence of shock fronts (cf. Osterbrock 1989).

Fig. 3. Evolution of the H profile. The first spectrum (dotted line) is actually a H profile (cf. Fig. 1). The transition from a saddle-like profile at earliest stages toward a more Gaussian-like one and eventually to a triple-peaked shape is evident as it is the continuous decrease in width (cf. Fig. 4).

Fig. 4. Decrease of the FWZI of H since outburst maximum.

The identification of emission lines other than the Balmer ones is complicated by the their large width. Surely present close to maximum are OI 7775 and 8446 Å, HeI 5876 and 7075 Å (the presumably weaker 6678 Å line is nearly lost in the H wings) and NII 5675. The complex at 5015 Å could be a blend of HeI and NII, and NIII and CIII are probably the main contributors to the blend at 4630 Å. The ionization degree has increased during the decline, with HeII 4686 Å becoming visible after the first week from maximum. As for previous outbursts, also this time the nebular lines have not shown up in the late spectra of U Sco, reinforcing the notion that a limited amount of material - if any - has been ejected by U Sco.

The Balmer and O I 7775-8446 Å emission lines showed a saddle-like profile at earliest phases, while other lines presented a more Gaussian-like profile. At day +3 Balmer and O I lines turned to single-peaked profiles as well. Later evolution has been characterized by Balmer and HeII lines to split into three components with velocity separation of the order of 1600 km sec^-1. This triple-peak profile was not observed in the 1987 outburst (cf. S88) and can be perhaps only marginally spotted in the latest H profile presented by B81 for the 1979 outburst. The eclipsing nature of U Sco prevents an explanation of the triple peaks at later phases (when the ejecta become presumably optically thin) as collimated beams of material ejected at a large angle from the plane of the orbit or of an accretion disc.

The 1979, 1987 and 1999 outbursts spectroscopically resemble each other only in broad terms, with significant differences from eruption to eruption. Such differences might trace large changes from event to event in the optical depth and kinematics of the ejecta. The optical depth must be connected to the amount of ejected material (the velocity and time extent remaining about the same from outburst to outburst), which should in turn depend on the amount of material accreted between successive outbursts.

© European Southern Observatory (ESO) 1999

Online publication: June 6, 1999
helpdesk.link@springer.de