3.1. Radio-frequency spectral region
Quasars and radio galaxies are the most luminous radio sources known. The emission process is well known to be synchrotron emission by relativistic electrons in a magnetic field, leading to power-law spectra. The most luminous of all are the Fanaroff-Riley `class II' radio sources, double-lobed structures with the strongest emission at the outer edges of the lobes. Long linear structures lead out to the lobes from the nuclei and are evidently the channels along which the energy flows out to where the plasma is braked and radiates. The linear structures can be traced as well aligned down to very small linear structures near the nucleus, generally called jets. The subject is a very extensive one, treated in whole books and detailed review articles, for example by Miley (1980), Bridle and Perley (1984), and Begelman et al (1984). We cannot hope to go into it in depth here. However, it should be noted that many of these galaxies are broad-line radio galaxies which, like the quasars, have optical emission-line spectra similar to Seyfert 1 galaxies. Others are narrow-line radio galaxies, with similar optical spectra to Seyfert 2s. The broad-line radio galaxies are invariably cD, D or E galaxies in form, rather than spirals, chiefly Sa and Sb, as the Seyferts are. It should also be noted that the entire radio luminosity, and the energy that goes into the jets is only of the order of 1% of the total luminosity of the AGNs in these objects (Caganoff 1989). Finally, these class II radio galaxies have a linear relationship between [O III] 5007 luminosity in their nuclei and total radio power, as do Seyfert galaxies (Rawlings et al 1989).
Seyfert galaxies are much weaker radio sources, but within recent years much observational data have been obtained on them (Wilson and Ulvestad 1982a, b, Ulvestad and Wilson 1984a, b). From many early measurements it appeared that Seyfert 2 galaxies are on the average stronger radio sources than Seyfert 1s, and Ulvestad and Wilson (1984a) confirmed this for their sample. They also found that the Seyfert 2 radio sources in their sample on the average are larger (typical diameter ~ 600 pc) than Seyfert 1 radio sources (typical diameter ~ 100-200 pc). In all cases the high-resolution VLA radio maps show that the nucleus of the radio source coincides with the optical nucleus, and in most of them the remaining structure is linear and jet-like. Thus it seems that the same type of processes which generate two-sided plasma jets in radio galaxies do so, perhaps in modified form, in Seyfert galaxies also, but interactions with interstellar matter in these spirals stop the jets in relatively short distances. Probably the smaller typical sizes of the jets in Seyfert 1s result from the interaction of the jet material with the denser BLR gas, as highly simplified calculations suggest. There is also no correlation of the direction of the axis of the jet with the directions of either the long or short axes of the optical galaxy, as projected on the sky, indicating that the jets are neither parallel to nor perpendicular to the plane of the main body of the galaxy (Ulvestad and Wilson 1984b).
In particular, in the nearby, bright Seyfert 2 galaxy NGC 1068, a jet is observed, centered on the nucleus, and about 1 kpc (13") long. It lies in position angle 33°. There is also radio emission on a larger scale in this same direction over a distance of at least 9 kpc (2'), probably mostly associated with the disk of the galaxy. Several small, bright ionized-gas condensations, discovered by Walker (1968), lie very close to the jet, and strengthen the connection between the plasma (radio-emitting) structure and the ionized-gas (optical-line emitting) structure (Wilson and Ulvestad 1982b). A similar, highly elongated radio jet in NGC 4151, in position angle 77°, is also closely connected with the optical emission lines. These are but two examples of cylindrically symmetric structures which so many of the observations indicate.
More recently, radio measurements have been made of a complete, unbiased sample of bright spectroscopically selected Seyfert galaxies at 1.5, 6 and 20 cm (Edelson 1987). All of them have mpg = 14.5 or brighter, and the number of Seyfert 1 and Seyfert 2 galaxies are very nearly equal in this sample. All of them were detected as radio sources. The Seyfert 2 galaxies tend to have radio spectra F -n with n = 0.7 over the range 6 to 20 cm. The Seyfert 1 galaxies have nearly the same index (0.66 ± 0.28 dispersion compared with 0.71 ± 0.23 for the Seyfert 2s), but with a slight tendency to have flatter power laws (5 = 25% of the Seyfert 1s have n < 0.5 but only 2 = 10% of the Seyfert 2s do). Edelson has concluded that at high frequencies the Seyfert 1s and 2s appear to differ; about 25% of the Seyfert 1 galaxies have curved spectra which tend to flatten out at high frequencies while none of the Seyfert 2s do. Thus the Seyfert 2 galaxies have radio spectra similar to those of other `normal' optically thin synchrotron sources, clustered around n = 0.7 to 0.8. The Seyfert 1 galaxies have a broader range of radio spectra, with flatter spectra and more tendency to flatten at high frequency. This may result from a flat-spectrum, optically thick core in some Seyfert 1s, becoming visible near 1.5 cm.
The mean 6-cm radio luminosities of Seyfert 1s and 2s from this sample are indistinguishable. This is probably correct, for here the sample is complete (to mpg = 14.5 apparent magnitude), while previous samples were largely based on the Markarian objective-prism survey, which selected galaxies by their ultraviolet continuous spectra. It was thus biased against low-luminosity Seyfert 2 galaxies (which have only weak ultraviolet continua), as the subsequent CfA optical spectra survey proved by finding many new low-luminosity Seyfert 2s not included in the Markarian lists. Since the optical and radio luminosities are correlated, the ultraviolet-selected objects are thus not only biased against low optical luminosity but also against low radio luminosity. In the magnitude-limited sample the average ratio of radio-to-optical luminosity is a factor 1.9 larger for Seyfert 2 galaxies than for Seyfert 1s (Edelson 1987).
The most recent collection of radio data (21-cm continuum) of a larger sample of bright radio galaxies, spectroscopically selected, shows little clear evidence of a difference in average radio properties between Seyfert 1 and 2 galaxies (Giuricin et al 1990). Likewise, Ulvestad and Wilson (1989) have extended their 6 and 20 cm survey to include all 57 Seyfert galaxies known with z < 0.0153 that are accessible to the VLA. In this distance-limited survey they find only small differences between the radio properties of Seyfert 1 and 2 galaxies, which are but marginally significant. Thus all the recent radio-frequency data confirm that the previously reported differences were biased by the exclusion of lower-luminosity, redder Seyfert 2s, not found by the Markarian survey.
3.2. Infrared continuum
It has long been well known that Seyfert galaxies and QSOs emit much of their radiation in the infrared continuum. In particular NGC 1068 had been the first galaxy to be observed with almost every new infrared detector or infrared telescope. The main mechanism of infrared continuum emission in Seyfert 2s has long been understood as resulting from thermal reradiation by dust particles, heated by the central source. Extinction by dust is very important in QSOs and Seyfert galaxy nuclei (MacAlpine 1985). NGC 1068 and a few other well observed Seyfert 2 galaxies show a silicate absorption feature at = 10.4 µm, and a silicate emission feature at = 19 µm, similar to those that would and do result from dust in our Galaxy. In NGC 1068 the infrared luminosity is approximately 3 x 1011 L, approximately half of it coming from a warm, compact source with measured diameter 1" ( 80 pc), the other half from a more extended, cooler component, the dust in the galaxy. In Seyfert 1 galaxies, quasars and QSOs some of the infrared radiation also comes from heated dust, but in addition some has long been thought to arise in a power-law source (Rieke and Lebovsky 1979).
The recent IRAS long-wavelength measurements have added greatly to our knowledge of all classes of AGNS and their host galaxies. Indeed, these observations showed that it is possible, from the IRAS measurements of known Seyfert galaxies, to develop criteria in a far infrared colour-colour plot for finding Seyfert galaxy candidates (Miley et al 1985). Subsequent follow-up optical spectra showed a fairly large fraction of these candidates to be actual AGNs, and deep-pointed IRAS measurements to low flux levels thus the most efficient known method for finding faint, distant Seyfert galaxies (Keel et al 1988). However, as both these papers stated, and as Salzer and MacAlpine (1988) confirmed with further observational data, the IRAS colours do not find all Seyfert galaxies, and do include among the candidates many starburst and H II-region galaxies, which turn out from slit spectra not to be Seyferts.
Somewhat contradictory results as to the nature of the infrared continuum have been obtained from the IRAS measurements. One study is based on a sample of 29 of the brightest AGNs for which there are good measurements from the IRAS regions (100-12 µm) through the ground-based infrared and optical to the IUE ultraviolet spectral region (Edelson and Malkan 1986). After removing the underlying galaxy spectrum and the H I and Fe II emission features, the remainder of the continuum from 10-1-102 µm was fitted by a combination of a power law v-n (with strength and index n to be determined), an ultraviolet blackbody (with strength and temperature to be determined - this component could equally well have been fitted by an accretion disk), an optically thin Balmer continuum component and a `5 µm excess', fitted by a parabola centered at 5.2 µm. The three Seyfert 2 galaxies in this sample (including NGC 1068), plus three Seyfert 1 galaxies and one radio galaxy, 3C 84 = NGC 1275, could not be fitted by this combination. Their infrared radiation is clearly thermal emission from dust covering a range of temperatures. Their derived dust temperatures are in the range 40-80 K, and their dust masses 100 M, corresponding to gas masses of approximately 104-105 M, a reasonable estimate for an NLR. The great importance of radiation by dust is well brought out by the analysis of Edelson et al (1987).
For the relatively dust-free Seyfert 1 galaxies such as II Zw 136, NGC 3516, Mrk 335 and others, the power-law component with n 1.3 does fit and the infrared continuum is therefore interpreted as primarily synchrotron emission. These objects have relatively low internal reddening, as estimated from their 2175 (graphite dust) absorption. In the far infrared the fluxes generally peak near 80 µm, and a `turnover' (decrease in flux at longer wavelengths) due to synchrotron self-absorption is also included among the fitting parameters (Edelson and Malkan 1986). Strong arguments that the non-stellar infrared continua out to 10 µm are primarily power-law, synchrotron emission had previously been given by Malkan and Filippenko (1983). They argued that, if the source were hot thermal dust, the infrared flux should fall steeply (have a `cutoff') at wavelengths shorter than 2 µm. In this picture the 5 µm excess can best be fitted by thermal emission by dust near the nucleus, heated by the optical and ultraviolet radiation of the accretion disk. The short wavelength fall-off is defined by the 1500 K evaporation temperature of graphite grains, and the long wavelength fall-off by complete absorption of the radiation over the limited solid angle subtended by the dust clouds at the nucleus. A torus with the dust between 1-2 pc (inner edge) distance and 200-500 pc (outer edge, T 100-300 K) will fit the observed 5 µm excess (Barvainis 1987). Another good analysis of the continuum radiation in these terms is due to Carleton et al (1987).
However, a large sample of bright QSOs and quasars for which IRAS data are available was analyzed by another group with a different interpretation (Sanders et al 1989). This sample is restricted to the bright QSOs from the PG survey, and does not include any Seyfert 2s. The main observational result is that the gross shape of the energy distributions over the range = 1012-1017 Hz ( = 3 x (105-10-1)nm) is similar for all these objects, and can be fitted plausibly by two broad components of thermal emission. One is the `big blue bump', the radiation of the accretion disk. The other the `infrared bump' between = 2 and 103 µm. The latter is interpreted as entirely resulting from dust, at distances from the central source from 0.1 pc (the absorbing torus) out to more than 1 kpc (in the disk). It includes both the power-law and the 5 µm `excess'. Quantitative calculations show that the amount of dust expected to be present is sufficient to produce the required infrared radiation. Optically-thin emission from dust near the source contributes to the radiation in the region around = 3 x 1014 Hz ( = 1 µm) between the two bumps. It is supposed to come from the most refractory graphite or graphite-coated particles, with sublimation temperature 2000 K. (The theoretical analysis is `in preparation'.) Thus these authors see no convincing evidence for significant power-law radiation in the 1012-1017 Hz region in these objects.
Sanders et al (1989) argue that the dust and gas clouds are distributed in a warped disk, in agreement with section 4.1 of this review and that, as a result of this warp, the covering factor for dust to be heated by the central source is 0.1. As they discuss, there is good evidence for a `tilted disk' (or warp) near the center of our Galaxy, and further out in many Seyfert galaxies.
Sanders et al (1989) emphasize that evidence for power-law (synchrotron) emission in the far infrared spectra of AGNs has come primarily from blazars (BL Lac objects and optically violently variable quasars) and flat-spectrum radio-loud quasars, of which 3C 273 is a prime example. These types of objects are relatively rare. Blazars, with their relatively smooth, near power-law spectra (n -1), high polarization and rapid variability, are best interpreted as synchrotron emission from a relativistic jet over the entire radio-frequency to ultraviolet continuum (Saikia and Salter 1988). 3C 273 does vary in the optical and near-infrared on time scales as short as 1 day, and on time scales of months at = 10 µm. Thus there is good evidence for non-thermal, power-law synchrotron radiation in it. It is an exceptional object, and the process is exceptional according to the interpretation of Sanders et al (1989).
A strong argument that there is a power-law continuum has been that the infrared (3 µm) and x-ray (2 keV) fluxes of AGNs are correlated. This argues that these fluxes are produced by the same process, and therefore that the infrared is not thermal emission from dust (Elvis et al 1986). However, more recently McAlary and Rieke (1988), using more recent data and more nearly complete samples, have found no such correlation for the Seyfert 1 galaxies, and only a weak correlation for PG quasars. Sanders et al (1989) likewise find no strong evidence for a correlation between the entire infrared (2-100 µm) flux the x-ray (2-10 keV) flux, and conclude that the data do not require a common origin or a power law linking these two regions. Thus at present there is no consensus on the origin of the infrared radiation in QSOs and most Seyfert 1 galaxies. Very probably both thermal dust emission and the power law contribute, to different relative amounts in different objects.
3.3. X-ray region
From the early days of x-ray astronomy, Seyfert 1 galaxies and QSOs were recognized as luminous objects in this energy range (Wilson 1979). The H (broad emission-line) flux was discovered to be proportional to the x-ray flux, and a striking result was the prediction on this basis of the x-ray flux in the LINER M 81, and its confirmation, as described in section 6.4. The surveys in the 2-20 keV range found AGNs to have power law spectra with a narrow range around n = 0.7 (Mushotzky 1984). A later survey of more objects at lower energy (0.2-4 keV) gave somewhat larger scatter, with a mean n 1.0 for the radio-quiet objects, mostly QSOs but including some Seyfert 1s, and n 0.5 for the fewer quasars and radio galaxies. Furthermore, these data showed an increase above the power law at low-energies, or `soft spectral excess' in many of these objects (Wilkes and Elvis 1987).
The most recent survey, covering an even wider energy range (0.1-10 keV), shows still more complexity. The objects observed are largely x-ray flux-limited samples, mostly Seyfert 1 (and 1.5) with a few `narrow emission-line galaxies' (Seyfert 1.9 or 1.99 in the sense of section 6.3), one quasar (3C 273), and one Seyfert 2 (NGC 1068). The general result is that in the 2-10 keV range a power law with n = 0.7 ± 0.2 (average deviation) fits well. At lower energies many objects have complex spectra. Some have soft x-ray excesses. In addition to the power law, the absorbing column along the line of sight can be determined from the amount of absorption at low energy. Typical values range from 1021 to 1023 H atoms cm-2 (the absorption is actually primarily by elements heavier than H and He and an assumption about abundances is involved in the calibration). As stated in section 6.1 there is a tendency for more of the low luminosity nuclei to have large absorbing columns, suggesting the high luminosity objects may have blown away and dispersed the clouds around them. The soft excesses can be modeled either by a second component with a steeper power law, or by partial absorption by material in clumps, allowing some of the radiation to pass through between them, unattenuated. In most but not all cases, the two-component model fits the data better than the `leaky-absorber' model (Turner and Pounds 1989).
X-ray emission in AGNs with a power-law spectrum with n = 0.7 can be associated with the infrared power-law continuum with n = 1.2 discussed in the previous section, if the latter is actually present. The most plausible process is synchrotron emission in the infrared (low-energy) region, and Compton upscattering of these photons plus synchrotron emission in the x-ray (high-energy) region. Models based on these ideas are called synchrotron self-Compton models and require a broken power-law of electron energies. Fairly natural assumptions give the break in flux power laws (n) by about 0.5 at energy ~ 1 keV. These models are well summarized by Zdziarski (1986) and by O'Dell et al (1987).
The more complicated x-ray spectra resulting from the recent, most accurate data complicate this picture (Turner and Pounds 1989, Kruper et al 1990). The several observed spectral components imply several synchrotron self-Compton regions, or perhaps a combination of one with a high-temperature blackbody spectrum, or perhaps other, non-power-law electron energy spectra (Urry 1990, Urry et al 1990). Very complex models, bridging the range from x-ray to infrared, have been calculated (Band 1987) and fitted to observed AGN spectra (Band and Malkan 1989).
Finally, x-ray variations have been observed in some AGNs on a time scale of hours. A recent, quite well determined case is in Mrk 335, where large amplitude variations have been detected on times of ~ 6 h. This is consistent with variations in the accretion disk at 5 Schwarzschild radii about a black hole with mass 107 M (Turner and Pounds 1988, who give references to earlier similar observations). X-ray variations at the 10% level, which are not simple to interpret theoretically, have been observed with doubling times from 102 s (Seyfert 1s) to 106 s (QSOs) (Barr and Mushotzky 1986).
3.4. -ray radiation
X-rays have been observed from several AGNs to energies above 0.1 MeV, which might be called -rays, and 3C 273 has been measured to over 100 MeV. The flux is not much below an extension of the x-ray spectrum to this energy. Hence a significant amount of energy is emitted in these very high-energy photons. Little is definitely known as to the mechanism, but an extension of the processes responsible for the x-ray emission is the most obvious possibility. Others include pair production and annihilation in a hot, Doppler-broadened region and Penrose-Compton scattering. However, 3C 273 may not be representative of the general class of AGNs in its -ray properties (Ramaty and Lingenfelter 1982).