Annu. Rev. Astron. Astrophys. 1999. 37: 409-443 Copyright © 1999 by Annual Reviews. All rights reserved

5. ACCRETION DISK INSTABILITIES AND JET FORMATION

Collimated jets seem to be systematically associated with the presence of an accretion disk around a star or a collapsed object. In the case of black holes, the characteristic dynamical times in the flow of matter are proportional to the black hole's mass, and the events with intervals of minutes in a microquasar could correspond to analogous phenomena with duration of thousands of years in a quasar of 10⁹ M (Sams et al 1996, Rees 1998). Therefore, the variations with minutes of duration observed in a microquasar in the radio, IR, optical, and X rays could sample phenomena that we have not been able to observe in quasars.

X-rays probe the inner accretion disk region, radio waves the synchrotron emission from the relativistic jets. The long term multiwavelength light curves of the superluminal sources show that the hard X-ray emission is a necessary but not sufficient condition for the formation of collimated jets of synchrotron radio emission. In GRS 1915+105 the relativistic ejection of pairs of plasma clouds have always been preceded by unusual activity in the hard X-rays (Harmon et al 1997), more specifically, the onset of major ejection events seems to be simultaneous to the sudden drop from a luminous state in the hard X-rays (Foster et al 1996, Mirabel et al 1996a). However, not all unusual activity and sudden drops in the hard X-ray flux appear to be associated with radio emission from relativistic jets. In fact, in GRO J1655-40 there have been several hard X-ray outbursts without following radio flare/ejection events. A more detailed summary of the long term multifrequency studies of black hole binaries can be found in Zhang et al (1997).

The episodes of large amplitude X-ray flux variations in time-scales of seconds and minutes, and in particular, the abrupt dips observed (Greiner et al 1996, Belloni et al 1997, Chen et al 1997) in GRS 1915+105 are believed to be evidence for the presence of a black hole, as discussed below. These variations could be explained if the inner ( 200 km) part of the accretion disk goes temporarily into an advection-dominated mode (Abramowicz et al 1995, Narayan et al 1997). In this mode, the time for the energy transfer from ions (that get most of the energy from viscosity) to electrons (that are responsible for the radiation) is larger than the time of infall to the compact object. Then, the bulk of the energy produced by viscous dissipation in the disk is not radiated (as happens in standard disk models), but instead is stored in the gas as thermal energy. This gas, with large amounts of locked energy, is advected (transported) to the compact object. If the compact object is a black hole, the energy quietly disappears through the horizon. In constrast, if the compact object is a neutron star, the thermal energy in the superheated gas is released as radiation when it collides with the surface of the neutron star and heats it up. The cooling time of the neutron star photosphere is relatively long, and in this case a slow decay in the X-ray flux is observed. Thus, one would expect the luminosity of black hole binaries to vary over a much wider range than that of neutron star binaries (Barret et al 1996). The idea of advection-dominated flow has also been proposed (Hameury et al 1997) to explain the X-ray delay in an optical outburst (Orosz et al 1997) of GRO J1655-40.

During large-amplitude variations in the X-ray flux of GRS 1915+105, remarkable flux variations on time-scales of minutes have also been reported at radio (Pooley & Fender 1997, Rodríguez & Mirabel 1997, Mirabel et al 1998) and near-infrared wavelengths (Fender et al 1997, Fender & Pooley 1998, Eikenberry et al 1998a, Mirabel et al 1998). The rapid flares at radio and infrared waves are thought to come from expanding magnetized clouds of relativistic particles. This idea is supported by the observed time shift of the emission at radio waves as a function of wavelength and the finding of infrared synchrotron precursors to the follow-up radio flares (Mirabel et al 1998). Sometimes the oscillations at radio waves appear as isolated events composed of twin flares with characteristic time shifts of 70 ± 20 minutes (e.g. Pooley & Fender 1997, Dhawan et al 1999). The time shift between the twin peaks seems to be independent of wavelength (Mirabel et al 1998), and no Doppler boosting is observed. This suggests that these quasiperiodic flares may come from expanding clouds moving in opposite directions with non-relativistic bulk motions.

In Figure 7 are shown simultaneous light curves in the X-rays, infrared, and radio wavelengths, together with the X-ray photon index during a large amplitude oscillation. These light curves can be consistently interpreted to imply that the relativistic clouds of plasma emerge at the time of the dips and follow-up recovery of the X-ray flux. In adiabatically expanding clouds the maximum flux density at short wavelengths (i.e. the near infrared) should be observed very shortly after the ejection (10^-3 sec), and it is only in the radio wavelengths that significant time delays occur (Mirabel et al 1998). Figure 7 shows that the onset of the infrared flare occured 200 sec after the drop of the X-ray flux, during its recovery from the dip, probably at the time of the appearance of an X-ray spike (t = 13 min) which is associated to a sudden softening of the (13-60 keV) / (2-13 keV) photon index due to the drop in the hard X-ray flux. Similar phenomena have been observed in this source by Eikenberry et al (1998a). In the context of the unstable accretion disk model of Belloni et al (1997), these observations suggest that the ejection of plasma clouds takes place during the subsequent replenishment of the inner accretion disk, well after the disappearance of the soft component at the sudden drop. The ejection of the clouds seems to be coincident with the soft X-ray peak at the dip. Furthermore, the slow rise of the infrared flux to maximum seen in Figure 7 indicates that the injection of relativistic particles is not instantaneous and that it could last up to tens of minutes.

Figure 7. Radio, infrared, and X-ray light curves for GRS 1915+105 at the time of quasi-periodic oscillations on 1997 September 9 (Mirabel et al 1998). The infrared flare starts during the recovery from the X-ray dip, when a sharp, isolated X-ray spike is observed. These observations show the connection between the rapid disappearance and follow-up replenishment of the inner accretion disk seen in the X-rays (Belloni et al 1997), and the ejection of relativistic plasma clouds observed as synchrotron emission at infrared wavelengths first and later at radio wavelengths. A scheme of the relative positions where the different emissions originate is shown in the top part of the figure. The hardness ratio (13-60 keV) / (2-13 keV) is shown at the bottom of the figure.

Mirabel et al (1998) have estimated that the minimum mass of the clouds that are ejected every few tens of minutes is ~ 10¹⁹ g. On the other hand, the estimated total mass that is removed from the inner accretion disk in one cycle of a few tens of minutes is of the order of ~ 10²¹ g (Belloni et al 1997). Given the uncertainties in the estimation of these masses, it is still unclear what is the fraction of mass of the inner accretion disk that disappears through the horizon of the black hole. Anyway, it seems plausible that during accretion disk instabilities consisting of the sudden disappearance of its inner part, most of it is advected into the black hole, and only some fraction is propelled into synchrotron-emitting clouds of plasma.

Energy outbursts in the flat synchrotron spectrum over at least four decades of frequency have also been observed in Cygnus X-3 (Fender et al 1996). The optical polarization observed in GRO J1655-40 (Scaltriti et al 1997) could also be related to the presence of synchrotron emission at optical wavelengths. The study of GRS 1915+105 led to the realization that besides the energy invested in the acceleration of the plasma clouds to their bulk motions, the oscillations of the type shown in Figure 7 require synchrotron luminosities of at least 10³⁶ erg s^-1. This synchrotron luminosity is not negligible with respect to the thermal luminosity radiated in the X-rays. These results give support to the observation of synchrotron infrared jets reaching distances of a few thousand AU from GRS 1915+105 (Sams et al 1996).