The anti-correlation of soft X-ray emission with optical brightness is reminiscent of the behaviour of the SSB RX J0513.9-6951 (Reinsch et al. 1996; Southwell et al. 1996). RX J0513.9-6951 turns on as a supersoft X-ray source only during mag optical dips, which occur every 100-200 days and last about days. This behaviour has been explained by assuming that the shell-burning white dwarf in RX J0513.9-6951 has normally expanded to a few cm and radiates its luminosity in the extreme-ultraviolet. During the optical faint states, the white dwarf contracts with almost constant bolometric luminosity to cm, and radiates in the soft X-ray band.
The model that has been suggested for RX J0513.9-6951 cannot explain the observational data of V Sge. First, the optical brightness changes of V Sge are very rapid: both the faint-/bright-state transitions as well as the succession of different faint states may occur on timescales of day (compared to the smooth decline of several days in RX J0513.9-6951). Such very rapid changes are only possible if the white dwarf envelope expands and contracts on the Kelvin-Helmholtz timescale and the mass of the expanding envelope is rather small (). Such a small envelope mass is difficult to accept for a white dwarf with stable shell burning (e.g. Prialnik & Kovetz 1995). Second, the expected optical eclipse would become deeper when the system becomes brighter, opposite to what has been observed (Patterson et al. 1998).
Before we speculate on a possible explanation for the observed X-ray properties of V Sge, we note the similarity of the V Sge behaviour to that of VY Scl stars (as has been noted with respect to the optical behaviour also by Robertson et al. 1997). In a recent survey of the available ROSAT data of VY Scl stars (Greiner 1998) a relatively hard X-ray spectrum was found for VY Scl stars during optical bright state (see also van Teeseling et al. 1996). Moreover, observations of the VY Scl star V751 Cyg during its 1997 optical faint state have revealed luminous and very soft X-ray emission (Greiner et al. 1998), similar to that of V Sge in its faint state.
Inspection of the change in eclipse depth from faint to bright state (e.g. Fig. 5 in Patterson et al. 1998) shows that it is possible to reproduce this change by an increase of uneclipsed flux, while the eclipsed light (presumably from the irradiated accretion disk) remains almost constant. If the flux from the irradiated disk (and therefore also from the irradiated secondary) remains unchanged, this would suggest that a brightening of V Sge is caused by an increasing amount of extended luminous (outflowing?) matter. This would also be consistent with the emission lines, which only show partial eclipse effects, and which increase in strength when V Sge brightens (e.g. Herbig et al. 1965), indicating either an increasing amount of line emitting matter or an increasing amount of ionizing flux. (Mauche et al. (1997) note that the He II /C IV and N V /C IV line ratios decrease when V Sge brightens.)
Additional extended gas could increase the amount of soft X-ray absorption and make the soft X-ray component fainter and harder and make it even completely undetectable. Because it is impossible to produce the hardness ratios and count rate observed in Nov. 1991 by simply adding more absorption to an absorbed hot-white-dwarf spectrum, we conclude that in Nov. 1991 an additional harder X-ray component was present. The hardness ratios and count rate of the Nov. 1991 PSPC observation can be explained perfectly with a thermal bremsstrahlung spectrum, absorbed with cm-2, a temperature of a few keV, and a 0.1-2.4 keV luminosity of erg s-1 (at 1 kpc). The same bremsstrahlung component may also have been present during the soft X-ray state in Nov. 1992 without significantly affecting the confidence intervals in Fig. 2 of the spectral parameters of the soft component. We note that such a bremsstrahlung component is not inconsistent with the X-ray flux of an evolved secondary in a 12 hr binary (cf. Dempsey et al. 1993).
A simple wind model for the recently observed radio flux density of V Sge implies a mass-loss rate of the order of 10-6 /yr (Lockley et al. 1997). With their (assumed) terminal velocity of 1500 km/s this wind zone is completely opaque for X-rays up to 0.7 keV, even if the wind is assumed to be circumbinary instead of arising from one component. Since the radio measurement has been obtained during optical high state, it supports the above described scenario. We note that the colliding wind scenario as discussed in Lockley et al. (1997) would predict a positive correlation between optical and X-ray emission, contrary to our finding. Vitello & Shlosman (1993) have modeled the UV line shapes of V Sge assuming a biconical accretion disk wind, and need high mass-loss rates to explain the observations, while they did not consider the possibility of a luminous shell-burning white dwarf. We also note that irradiation-induced winds with a rate of 10-7-10-6 /yr are expected in SSB (Van Teeseling & King 1998).
We conclude that it is possible to explain the optical and X-ray behaviour of V Sge by assuming a variable amount of extended uneclipsed matter which contributes significantly in the optical (by reprocessing of soft X-rays?) and may completely absorb the `supersoft' X-ray component during the optical bright/hard X-ray state. The X-ray properties of V Sge, in any case, support the presence of a hot luminous white dwarf, possibly with hydrogen shell-burning, and do not hinder the addition of V Sge to the class of supersoft X-ray binaries. More detailed modelling of the changing optical light curves and ultraviolet and optical spectra are necessary to test this scenario.
© European Southern Observatory (ESO) 1998
Online publication: September 30, 1998