Annu. Rev. Astron. Astrophys. 1984. 22: 471-506 Copyright © 1984 by Annual Reviews. All rights reserved

6. JET FORMATION

Directed outflow is a ubiquitous feature of active galactic nuclei, and it is also seen in some small-scale prototypes of AGNs in our own Galaxy (e.g. SS 433). This is in itself evidence that a spherically symmetric model cannot be entirely realistic. For a full review of theories of jet propagation, with special relevance to radio galaxies, the reader is referred to Begelman et al. (17). The direct evidence for jets pertains exclusively to scales much larger than the primary power source. The scales probed by VLBI are typically a few parsecs ( 10⁴r_g for plausible central masses); the only evidence for smaller-scale beaming comes from indirect arguments about the physics of optically violent variables (OVVs), or "blazars" (6, 87, 88). There are theoretical reasons for postulating that the relativistic outflow is initiated on scales of order r_g, but there are really no grounds for believing that a narrow collimation angle is established until the jets get out to VLBI scales or beyond: indeed, conditions in the medium 1 pc from the central source cannot readily provide the kind of pressure-confined "nozzles" (27) that could best collimate them (107).

The radiation from the jets - the emission detected by VLBI and other radio techniques, as well as the emission in other wave bands from (for instance) the M87 jet - is presumably synchrotron radiation from electrons accelerated in situ. Plainly, any high- random motions produced at r r_g would have been eliminated by radiative and adiabatic losses before the jet got out to 1 pc. In the superluminal sources, there is direct evidence for bulk relativistic outflow (_b 5). We do not know whether this outflow involves ordinary matter, electron-positron plasma, or even Poynting flux, and various authors have suggested schemes involving each of these options.

Any disk structure near a black hole provides a pair of preferred directions along the rotation axis; moreover, within the Lense-Thirring effect's domain of influence, this axis is maintained steady by the hole's gyroscopic effect. Magnetically driven winds from tori or from thin disks (23, 26) could generate outflowing jets with the attractive attribute of a self-confining toroidal field.

The evacuated vortices along the axes of thick accretion tori, which can be very narrow for an angular momentum distribution close to = constant, suggest themselves as possible preexisting channels for directed outflow. The most widely discussed version of this idea, first proposed by Lynden-Bell (74), utilizes radiation pressure. A simple order-of- magnitude argument shows that a test particle (electron plus ion) released from rest outside a source with r r_g and (L - L_E) / L_E 1 would attain a relativistic speed; a radiation-supported torus whose vortex has cone angle emits within this cone a greatly enhanced luminosity ~ ^-2 L_E per unit solid angle, which suggests that this photon beam might impart high Lorentz factors to any matter in its path.

Detailed study reveals flaws in this superficially attractive idea (4, 5, 90, 119). The main problem is that the radiation field within a long, narrow funnel is almost isotropic: there may indeed be a super-Eddington outward flux along it, but the radiation density far exceeds (flux / c) because of scattering, or absorption and reemission, by the walls. Consequently, a test electron travels subrelativistically along the funnel, at a speed such that the radiation appears nearly isotropic in its moving frame. The radiation flux only becomes well collimated by the time the particle escapes from the funnel, at r = r₀. Even for the (probably unstable) = constant tori, r₀ is at least ^-2 r_g; and out there the dilution (because r is now >> r_g) cancels out the factor gained from the beaming. The net result is that -values of only ~ 2 can be reached for an electron-ion plasma, and maybe up to ~ 5 for electron-positron plasma. A second difficulty is that the Thomson depth along the funnel would become > 1, vitiating the test-particle approach adopted in the calculations, if the particles were numerous enough to carry a substantial fraction of L. [However, in the limit of very large optical depths, where radiation and matter can be treated as a single fluid, radiation pressure around a supercritical central source - a "cauldron" (21) - could efficiently generate a jet of ordinary matter with high _b.]

Quite apart from these theoretical difficulties, models involving radiation-supported tori cannot be relevant to the objects where the most spectacular jets are seen (radio galaxies, M87, etc.). We have upper limits to the thermal luminosity from these AGNs; we also have lower limits to the energies involved in producing large-scale radio structure and, hence, to the masses involved. Combining these limits precludes there being any object emitting a thermal luminosity L_E (the level of isotropic emission that would be an inevitable concomitant of a radiation-supported torus with a narrow funnel).

An ion-supported torus maintained by accretion with low can provide funnels along the rotation axis, just as a radiation-supported torus can. The expelled material would then be an electromagnetically driven wind of electron-positron plasma (99, 108). The rest mass energy of the pairs could be << L/c² - indeed, most of the outflow could be in Poynting flux rather than being carried by the pairs themselves - making high beam Lorentz factors _b no problem. An energy flux of this kind could readily be converted into relativistic particles at large distances from its point of origin and is thus an attractive model for radio sources.

Two factors constrain the content and the Lorentz factor of jets emerging from scales of ~ r_g (99, 107). First, an e⁺-e^- jet that started off with too high a particle density would suffer annihilation before moving one scale height: this means that an energy flux L_E in pair kinetic energy, rather than in Poynting flux, is impossible unless _b is high. [The particle flux is then less for a given L; furthermore, the time scale available for annihilation, measured in the moving frame, is only _b^-1(r / c).] But radiation drag effects give a second countervailing constraint that precludes particle jets with very high values of _b. Radiation pressure provides an acceleration only if it comes from the backward direction after transforming into the moving frame (97). If radiation comes from a source of finite size r_s, then the acceleration at a distance r would always saturate for _b (r / r_s), no matter how high the luminosity of the source. Moreover, in a realistic model for a galactic nucleus, some fraction of the luminosity is scattered or reemitted on scales out to ~ 1 pc. This quasi-isotropic flux exerts a Compton drag force on any beam, and it is particularly serious for e⁺-e^- beams, which have the least inertia relative to their scattering cross section.

The interaction of jets with the material at ~ 1 pc in AGNs is an interesting topic that has only recently been seriously discussed (86). Possibly, the beams generally deposit their energy in the emission-line region, and only in especially favorable cases does the jet material get collimated sufficiently to penetrate beyond r 1 pc.