Astron. Astrophys. 344, 639-646 (1999)

5. Additional consequences and conclusions

Our discussion here was carried out under the assumption that the radial advection of energy is negligible and the material rotates with Keplerian velocities. Could it be that as the thermal balance ceases to exist the disc will choose to develop a flow in the radial direction leading to a non negligible radial advection term rather than develop a wind? Our analysis indicates that this is impossible. Actually, advection aggravates the instability that we have discussed because when advection is assumed to occur, the temperature in the disc is higher and approaches the virial temperature (Narayan & Yi 1994; Shaviv et al. 1998). When discs with an advection term and without it are compared, one finds that the disc with advection is much more bloated in the z direction violating the 1D approximation (Wehrse et al. 1998) and therefore more prone to developing outflows in the z direction. The claim that radial advection stabilizes the thermal instability and produces another stable branch (Narayan & Yi 1994), is based on a 1-D analysis which ignores the thermal instability and its implications for the vertical direction.

It is possible that in the binary context, for certain ranges in the mass transfer rate from the companion, one may encounter a situation where the disc structure oscillates between two different states, one with a hole in the centre, and another with a more complete disc structure extending to the central object. Such structures appear to be common in the CV and the black hole x-ray transient contexts. We consider a situation where the mass transfer rate from the companion is in the range which cause the outer regions of the disc to cycle between two stable branches, driving the disc into high and low mass transfer states, as in the disc instability model (for a review cf. Cannizzo 1996). In the low state (quiescence) the mass transfer rate through the central regions of the disc is low, and the disc may develop a hole in the central regions for the reasons described previously. Accretion from a hot coronal wind will continue during this phase. As mass transfer from the companion continues, the surface density in the outer regions will build up to the critical surface density which triggers thermal instability, and causes the outer regions of the disc to transform to the second stable branch. A heat front will propagate through the disc, which now has a hole in the centre, and transform the entire disc into the hot high viscosity (outburst) state. During the hot phase, the viscosity is higher, and the mass transfer rate through the disc increases. The hole will then fill up and the disc will extend all the way to the star. X-rays from the inner disc will change the boundary condition on the surface of the disc, and may result in enhanced coronal winds from the disc during this phase.

We have argued that considerations of vertical structure lead one to conclude that a vertical outflow is an avoidable consequence. For high discs, thermally driven winds will play a minor role in the energetics and structure of the disc, although additional effects, such as radiation pressure driven outflows may come into play in certain regimes. However, as decreases, thermal winds may begin to play the dominant role, and below a certain mass transfer rate, a hole may develop in the centers of discs. In extreme cases, the bulk of the disc may be lost in the wind. We have shown that this could occur if free-free opacity dominates over a significant portion of the disc, as may be the case for black hole soft x-ray transients. Thermal winds will also be increasingly important in low CV and AGN black hole discs where other opacities dominate, but more detailed calculations are required to estimate the critical mass transfer rate below which holes develop.

The following are likely consequences of our model:

1. Thermally driven winds are a natural consequence of the shearing motion that occurs in discs, and disc viscosity, and should be common place in astrophysical systems which exhibit discs.

2. Due to the prevalence of wind mass loss from discs, the disc temperature profile is expected to be flatter than predicted by the Shakura & Sunyaev (1973) thin disc models, particularly in systems with low .

3. Low discs should exhibit centrally evacuated regions and lack the high temperature thermal component usually attributed to the central regions of discs.

4. Direct accretion via a disc is not expected to occur in low systems, so that a classical boundary layer around the equator will not form. The accretion in classical nova is apparently spherically symmetric.

5. The wind is expected to heat to temperatures of the order of at most 0.4 of the virial temperatures during its expansion phase, and be detected at x-ray energies.

The above characteristics may explain the following observations:

1. The overwhelming evidence for holes in the centers of accretion discs around white dwarfs in dwarf novae during quiescence (Warner 1995), and in the centers of black hole x-ray transients such as GRO1655-40 during quiescence, from the measured UV and Soft x-ray delays (Hameury et al. 1997).

2. The evidence for a hard x-ray spectrum without an accompanying soft x-ray black body component in the spectra of some black hole transients during quiescence (Tanaka & Lewin 1995).

3. The evidence for `discs' which are underluminous in comparison to deduced mass input rates, around black holes in binaries (Narayan et al. 1996), and in the centers of some AGN and our galactic centre (Narayan et al. 1995).

   Finally, we note that other models have been developed to explain central holes in CV discs. We note in particular the coronal siphon model of Meyer & Meyer-Hoffmeister (1994). These authors assume the existence of a corona and calculate the evaporation of the disc by means of an electron conduction flux from the corona back to the disc. In the present case the thermal instability generates the expanding hot layers which eventually become an expanding corona. We expect that thermal winds will be more effective in creating central holes in cases where viscous dissipation () plays a relatively more important role in determining the vertical structure of discs.

© European Southern Observatory (ESO) 1999

Online publication: March 18, 1999
helpdesk.link@springer.de