It is generally accepted that emission from heated dust produces the steep far-infrared continua of Seyfert 2's and the IRAS warm galaxies. It is also generally acknowledged that the rapidly variable infrared emission in the BL Lacs and optically violently variable quasars must be produced by non-thermal processes. Both thermal and non-thermal emission evidently occurs in nature, in objects of comparable luminosities. The controversy over the nature of the emission in radio-quiet quasars is thus not over whether either emission process is physically possible or plausible. Both are, and both are probably present at a significant level in all objects. The question is simply which happens to predominate in radio-quiet quasars.
In most optically selected quasars, the 1-100µ infrared luminosity comprises 10-50% of the bolometric luminosity (Sanders et al. 1989). Since the energy liberated per octave in radius in an accretion disk scales as r-1, the high relative infrared luminosity requires that the ultimate source of energy for the infrared radiation be within a few times the inner radius of the accretion disk. If the infrared radiation is emitted from a region comparable in size to that of its energy source, it must be nonthermal: blackbody limits on the source size at 1 µ and at 60 µ are respectively > 0.1 pc and > 100 pc for the most luminous sources in figure 1a and > 0.003 pc and > 10 pc for the least luminous. If the infrared radiation is thermal, and therefore emitted at large radii, it must be reprocessed energy transported to the emission radii by radiation or by mechanical means (e.g., a jet).
Dust heated locally by stars may contribute to some of the infrared emission from quasars. It is unlikely, however, that this dominates in the majority of sources. For stars to produce the typical infrared luminosity of 1013 L = 0.5 M c2 yr-1 would require a star formation rate of at least 500 M yr-1 (this lower limit assumes that only massive O stars are formed; a normal IMF would require a rate approximately 10 times higher). Over the ~ 108 yr lifetime of a quasar, > 5 x 1010 M of gas would have to be processed in massive stars. Dust would form from the metals produced. So much gas and dust, pushed to high latitudes by supernova explosions, would inevitably absorb and reradiate much of the luminosity from the central source. But the optical and ultraviolet radiation from quasars, variable on timescales 10 yr (Usher 1978) must come from such a central relativistic source. Since its luminosity is comparable to the infrared luminosity, we conclude that reradiation from gas and dust heated by it would necessarily be at least comparable to anything contributed by stars. In what follows we therefore ignore the heat input from stars, except insofar as they provide a natural minimum dust temperature 25 K.
Quasars and Seyfert galaxies appear to be located in galaxies amply supplied with interstellar medium. The disks of gas and dust in normal galaxies exhibit warps on all scales, and would intercept and re-radiate ~ 10% of the luminosity from a central source. Several lines of evidence suggest that the warps are even more severe in Seyferts and quasars, so that an even larger fraction of the central luminosity would be reradiated. Provided most of the reradiating material is located in an optically thick disk, the central source will never appear absorbed or reddened - either we have a clear line of sight, or the source is entirely occulted. This picture is thus consistent with the absence of reddening or absorption by broad emission line clouds in quasars.
We describe below the opacities and physical state of gas in the disks at distances from 10-3 pc to 104 pc from the central source. Dust grains can form and grow in the disk within 1 pc. At much smaller radii, however, even graphite grains will sublimate. When this occurs, the gas loses its primary coolant, and heats until it reaches a new thermal equilibrium at 104 K. The superposition of emission from radii inside and outside this transition point naturally explains the minimum in L at = 1014.5 Hz observed in most quasars. The characteristic scale length of dust in galaxies (~ 2-10 kpc) naturally explains both the frequency and steepness of the drop in L at = 1014.5 at 1012 Hz. The normalization of infrared luminosity relative to the UV and X-ray luminosities of quasars is consistent with expected covering factors and space-densities. Free-free emission from the photoionized zones on the illuminated surfaces of the disk naturally provides a flat-spectrum radio flux comparable to that observed in many quasars (Antonucci & Barvainis 1988), and may contribute to the optical continuum emission. It appears therefore that the emission at wavelengths 1-1000 µm from Seyfert galaxies and quasars other than OVV's is naturally explained as thermal reradiation from the nuclear disk and the interstellar medium of the host galaxy. Although non-thermal emission from the central source may contribute in some objects at some times, a significant contribution from thermal emission seems unavoidable.