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

7. SOME COMMENTS ON PHENOMENOLOGY

7.1. The Continuum Spectrum

The only direct clue to physical conditions in the central region (i.e. within a radius of, say, 100r_g) is the rather featureless continuum luminosity: spectral lines originate farther out. The models we have discussed can radiate either thermally or nonthermally: indeed, one of the hardest things to estimate is what fraction of the power dissipated via viscous friction in a realistic flow pattern would go directly into ultrarelativistic particles (via shocks, magnetic reconnection, etc.) rather than being shared among all the particles. Unfortunately, observations are little help in discriminating between various continuum radiation mechanisms: a smooth spectrum could be produced equally well by several alternative mechanisms. For instance (99), there are at least four ways of getting a spectrum with L() ^-1/2.

1. Thermal processes can mimic a power law if the spatial properties of the emitting medium vary in a suitable way (84). This particular slope arises if we consider bremsstrahlung from a spherical distribution of gas with density n r^-3/2 (corresponding to free-fall) and T T_virial r^-1.

2. Relativistic particles may be steadily injected with a high Lorentz factor and then lose energy by synchrotron or inverse Compton emission before escaping, yielding nonthermal radiation with L() ^-1/2.

3. Relativistic particles could be accelerated with an E^-2 differential spectrum. An (over) simple theory of shock acceleration (9, 48) actually yields this law for a compression factor of 4 (the value expected for a strong nonrelativistic shock).

4. Comptonization of injected soft photons yields a power law, which would have the particular value -½ for a value of the parameter y = (kT / m_e c²) _T² that is slightly geometry dependent but typically close to unity.

It is true that theoretical arguments can rule out some of these emission processes in some particular instances: for example, bremsstrahlung can never generate a high luminosity (L L_E) without _T being so large Comptonization reshapes the spectrum (71). These examples of mechanisms, any or all of which could be occurring within a single source, nevertheless highlight the necessity of other indicators (such as polarization or spectral breaks) for discriminating between them.

Obviously the values of M and are crucial in determining the properties of an accreting hole; the angular momentum parameter (J / J_max) is also important. We conclude further, and somewhat less trivially, that it is the value of = / _E that determines the nature of the inflow. The value of M itself only enters explicitly (and with weak fractional powers) when reabsorption effects are important. This means that there is a genuine physical similarity, not merely a crude resemblance, between active galactic nuclei and the stellar-scale phenomena (X-ray binaries, etc.) observed within our own Galaxy.

While it is perhaps foolhardy to put forward any fully comprehensive unified scheme for the various kinds of AGNs, there have been several proposals to relate particular categories of objects, or particular features in their spectra, to specific mechanisms.

7.2. Ordinary QSOs

Most QSOs are radio-quiet and are neither violently variable nor highly polarized. The main bolometric luminosity, in the near-ultraviolet, could come from the photosphere of a radiation-supported torus around a (10⁷ - 10⁸) M hole. Blandford (24) has suggested that the characteristic surface temperature is determined by He recombination, which changes the mean molecular weight. An isentropic torus of the type discussed in Section 5.3 would need to have a very high central density (and a correspondingly low value of in order to be sufficiently optically thick to thermalize radiation out at the putative photosphere - indeed, its central pressure and temperature might have to be so high that nuclear energy released via hydrogen-burning (16) dominates accretion-generated power (see Figure 4).

Figure 4. This diagram shows physical conditions near the pressure maximum of an optically thick radiation-supported torus around a hole of mass M. For a given M, the free parameter is the density (or Thomson optical depth T), which scales as -1. Rather high values of this parameter are needed in order to achieve LTE, even in the inner parts of the torus where the pressure is maximal (and where the temperature would then have the value given on the right-hand scale); if LTE is to extend out nearer the surface, then the torus may be so dense and massive near the center that nuclear burning and self-gravitation become significant. (In the approximate treatment given here, the pressure maximum is taken to occur at r rg; in fact is always at a larger radius than this, by an amount related inversely to the hole's angular momentum parameter.)

Even if one accepts that there is something special about a photospheric temperature T = 20, 000 K, the configuration need not resemble a stable torus. A more tentative and less controversial conjecture would simply be that typical QSOs are objects where the central hole is smothered by plasma clouds at distances (10² - 10³)r_g, which are dense enough to be close to LTE [but which are not necessarily supported quasi-statically by an n r^-3 density distribution (cf. Equation 29) at smaller r]. Such a hypothesis would suffice to explain the "UV bump" in quasar spectra (78, 79). The filaments emitting the broad spectral lines would lie outside this photosphere. Realistically, one expects an additional nonthermal component due to shocks and/or magnetic flaring (by analogy with O star photospheres, except that in AGNs the escape velocity, and probably also the characteristic Alfv'en speed, would be very much higher). The X rays could be attributed to this component, since in such a model no radiation would escape directly from r r_g.

7.3. Radio Galaxies

In radio galaxies, the direct radiative output from the nucleus is typically ~ 10⁴² erg s^-1, less than the inferred output of the beams that fuel the extended radio components. The energy carried by the beams in Cygnus A exceeds the central luminosity by a factor of ~ 10 . These objects must therefore channel most of their power output into directed kinetic energy. Moreover, the mass involved in producing the large-scale radio structure must be large - certainly > 10⁷ M. The thermal output from these AGNs is therefore 10^-3L_E, implying that they cannot involve radiation-supported tori; nor can radiation pressure be important for accelerating the jet material. Such considerations suggest that strong radio sources may involve massive spinning black holes onto which matter is accreting very slowly (maybe 10^-3 M yr^-1) to maintain an ion-supported torus, so that the holes' energy is now being tapped electromagnetically and being transformed into directed relativistic outflow (108).

7.4. Radio Quasars and Optically Violent Variables (OVVs)

Data on OVVs (also known as "blazars") have been reviewed by Angel & Stockman (6; see also 87, 88). For the extreme members of the class, such as OJ 287 and AO 0235 + 164, the case for beaming seems compelling. The less luminous objects might also be beamed, but they could alternatively involve unbeamed synchrotron emission from r 10r_g. More evidence on the hard X-ray spectrum of such objects would help to decide between these options. If gamma rays were emitted and (9.) were fulfilled, the resultant "false photosphere" of electron-positron pairs would scatter the optical photons and destroy any intrinsic high polarization (58). One would then be disposed to invoke relativistic beaming, which would increase the intrinsic source sizes compatible with the observed variability and reduce the luminosity in the moving frame; this would mean that (9.) was no longer fulfilled, and gamma rays could escape without being transformed into pairs.

7.5. Hard X-Ray and Gamma-Ray Sources

Boldt & Leiter (30, 68) have proposed a scheme whereby the output in gamma rays relative to X rays increases as decreases. Low-redshift objects are postulated to have low and to emit gamma rays; their high-z counterparts, however, are fueled at a higher rate, and they yield most of the X-ray background without contributing proportionately to the gamma-ray background.

According to White et al. (128), the characteristic X-ray spectrums of active nuclei depends on whether their primary luminosity in hard photons is 10^-2 L_E. For a source size of ~ 10r_g, this determines whether or not a pair photosphere is produced (9.). In sources with high L / r, where a pair photosphere is produced, the emergent Comptonized spectrum is softer. A small-scale analogue of this phenomenon may be the galactic compact source Cygnus X-1, which undergoes transitions between "high" and "low" states, the spectrum being softer for the former. The fact that many AGNs emit variable X rays with a flat spectrum (energy index 0.6; 89, 131) suggests that e⁺-e- production is inevitable, and that the effects of pairs on dynamics (99) and radiative transfer (130) need further attention.