6. CONTINUITY OF AGN PROPERTIES

6.1. QSOs and Seyfert 1 galaxies

Seyfert galaxies were first recognized as galaxies, then as galaxies with unusually luminous nuclei with strong, broad emission lines. Quasars were first identified as quasi-stellar radio sources with similar strong, broad emission lines. All the evidence since then has tended to indicate these are but two somewhat distance-dependent names for similar physical entities. Many of the most luminous radio galaxies are morphologically classified as N galaxies, meaning bright, near-quasi-stellar objects, with only a faint, barely detectable extended component or `fuzz'. Many objects such as I Zw 1, II Zw 1, Mrk 876, and 3C 120 have appeared both as Seyfert galaxies and QSOs in various catalogues. This continuity of physical appearance has been particularly emphasized by Morgan and Dreiser (1983). In recent years, using high-quality digital detectors and subtraction techniques, various authors have been able to detect faint galaxy images about the AGNs that had previously been described as quasars or QSOs. Out to about z 0.5, nearly every such object has yielded to these techniques. The emission-line spectra show no consistent differences. Furthermore, high-quality spectra show an underlying integrated stellar absorption-line spectrum in many objects, in addition to the strong featureless continuum of the central source. In many cases these spectra are earlier in spectral type than normal spirals, indicating the results of recent star formation in the active galaxies (see also section 7.5). Two very good reviews of this material, with many references to the original papers, are by Hutchings (1983) and Miller (1985).

As mentioned in section 2.4, the observed ratio of H luminosity to continuum luminosity, L(H) / L_C is constant over a wide range of luminosity from Seyfert 1 galaxies through QSOs, suggesting that the form of the spectrum and the amount of the luminosity absorbed (covering factor / 4) is also constant (Yee 1980, Shuder 1981). Attempts have been made to measure the covering factor ( / 4 subtended at the source by the photoionized clouds) from the observed ratio of soft x-ray flux (which is absorbed in the photoionization processes) to the hard x-ray flux (which may pass through unabsorbed if the column density is not too large). These suggest that the covering factors are larger for low-luminosity objects than for higher-luminosity ones (Lawrence and Elvis 1982, Reichert et al 1985). However, more recent, higher-resolution measurements of more objects show that the x-ray situation is quite complicated. Nevertheless they tend to support this decrease in covering factor with luminosity, but with considerable scatter (Mushotzky 1988, Turner and Pounds 1989). Similarly, comparison of the equivalent widths of both L and C IV 1549 in high-luminosity QSOs with those in Seyfert 1 galaxies, show a decrease (by about a factor two) in covering factor over this range of luminosity (Wu et al 1983). Again, the scatter is large (Wilkes 1986).

6.2. Quasars and radio galaxies

Quasars and broad-line radio galaxies appear to be related much as QSOs and Seyfert 1 galaxies are. The broad-line radio galaxies are almost invariably N, cD, D or E galaxies rather than spirals, and are rare among the E galaxies. There are few if any published surveys of D, cD or N galaxies that are not radio galaxies, but practically no radio-quiet Seyfert galaxies of these morphological types are known. Most probably the radio plasma that escapes from the nucleus along the jets in relatively gas-free cD, D and E galaxies is stopped by interaction with interstellar gas in Seyfert galaxies.

The outstanding spectral difference between the radio-quiet and radio-loud objects is that many Seyfert 1 galaxies and QSOs have strong, broad Fe II emission features in the optical region, while few quasars and broad-line radio galaxies do. This subject has been reviewed by Osterbrock (1985), who gives many references to the original papers. The ultraviolet Fe II multiplets arising from the same upper levels as the observed optical features are very strong in all Seyfert 1 galaxies (Veron-Cetty et al 1983). They are observed in the broad-line radio galaxies as well, indicating that Fe⁺ is present in their nuclei, but in lower abundance or less highly excited (Wills et al 1985). The physical interpretation is not known at present. Models which schematically take into account increased heating by relativistic electrons seem to reproduce some but not all of the observed effects: they predict weakening of the Fe II optical multiplets but do not give the observed H I relative line strengths (Ferland and Mushotzky 1984, Cesar et al 1985).

6.3. Few high-luminosity Seyfert 2s

Although Seyfert 1 nuclei and QSOs form a continuous sequence, there are no, or only very few, Seyfert 2 nuclei with luminosities as bright as M_B = -23. Only one `Seyfert 2 QSO' has been reported, E 0449-184, originally discovered as an x-ray object. Its redshift is z = 0.338 and its absolute magnitude M_B = -23.2 (assuming H₀ = 50 km s^-1 Mpc^-1 as the value of the Hubble constant). There is no sign of any broad emission lines in its spectrum, but it has strong, typical Seyfert 2 narrow emission lines, with FWHM 600 km s^-1 (Stocke et al 1982). This is larger than average for a Seyfert 2, but fits well on the correlation (with considerable scatter) between L([O III]) and FWHM.

The fact that high-luminosity Seyfert 2 AGNs are so rare must be incorporated into any overall model of AGNs. On the traditional picture it means that a high-luminosity, strong continuum from the central source, and an extensive BLR always go together. On the `hidden-BLR' picture, it means that the occulting torus becomes very weak or disappears around M_B = -23 (Miller and Goodrich 1990).

6.4. LINERs

Although QSOs and Seyfert galaxy nuclei make up most of the known AGNs, lower-luminosity ones certainly also exist. One is the nucleus of M 81, in which very weak, broad H emission was detected by Peimbert and Torres-Peimbert (1981) and confirmed by Shuder and Osterbrock (1981). Based on the correlation between broa H and x-ray luminosity, Elvis and Van Speybroeck (1982) predicted that it should be detectable as an x-ray source and did find it. Further x-ray data have shown that the nuclear source is relatively soft, with a power-law form spectrum ^-n with n > 2. This is steeper than the characteristic value n 0.7 for AGNs, but within the range of a few others. It can be very approximately fitted by an accretion-disk model corresponding to black-hole mass M 10⁴-10⁵ M and 10^-4-10^-3 M y^-1, in which the x-ray emission arises in the disk itself (Fabbiano 1988). Long-slit spectra show narrow emission lines of [N I], [O III] and [Fe VII] (only a small range in wavelength was observed) in the NLR close to the nucleus, and give only an upper limit to the black-hole mass M 10⁷ M (Keel 1989).

Many such `Low-Ionization Nuclear Emission-Line Regions' or LINERS have been identified. They are nuclei with emission-line spectra showing a generally lower-level of ionization than Seyfert 2s, but with the characteristic relatively strong [O I], [S II] and [N II] lines of an AGN. Many LINERs have relatively weak emission lines, and to detect them and to measure their strengths with any precision it is necessary to subtract the underlying galaxy absorption-line spectrum fairly accurately.

When the class of LINERs was first isolated by Heckman (1980), it was suggested that they might be objects in which shock-wave heating is the main energy-input mechanism, rather than photoionization by a hard spectrum. At relatively low levels of ionization (below log [O III] / H = + 0.5) the diagnostic ratios of figures 2, 3 and 4 do not distinguish between these two possibilities. However, it now seems much more likely that LINERs are simply the extension of Seyfert 2 nuclei to lower luminosities, smaller ionization parameters, and somewhat larger exponents n in the representative power-law photoionizing spectrum. Figures 1, 2 and 3 extend smoothly down to LINERS, and several nuclei with ionization near the Seyfert 2-LINER `boundary' log [O III] / H + 0.5 show weak He II 4686 emission. It can only be understood as resulting from photoionization by a hard spectrum. The models calculated on this basis fit LINERs well (Ferland and Netzer 1983, Halpern and Steiner 1983, Keel 1983a, Stasinska 1984a, Binette 1985).

Very strong confirming evidence of photoionization is available from IUE observations of one LINER, NGC 4579. They show that there is a point-like ultraviolet source at the nucleus of this galaxy, with an observed flux which, with assumed power-law index n = 1.4, fits the measured x-ray flux and provides the required number of ionizing photons to power the observed H flux (Goodrich and Keel 1986).

Many LINERs have been detected in spiral galaxies. For instance in a well defined sample, observed and with the continuum carefully subtracted to reveal the presence of very weak nuclear emission lines, all showed H and [N II]. Of these nuclei 5% were Seyferts, and of the remainder approximately 80% of the Sa and Sb nuclei showed LINER-type spectra. This percentage dropped steeply through Sc, to essentially zero at Scd, probably at least in part because of the increased strength of the H II region type spectra in the nuclei of these objects, resulting from recent star formation. Furthermore, approximately 10% of these LINERs in this survey show weak, broad H lines (FW0I 3000 km s^-1). They might be classified as Seyfert 1.99 (a very weak, broad component of H barely detectable!) on the traditional scheme. They must be an extension of `typical' (previously known) AGNs to smaller black-hole masses and lower accretion rates (Stauffer 1982, Keel 1983b).

In addition to the spirals, the nuclei of a few elliptical galaxies have LINER spectra. Two bright, well studied examples are NGC 1052 and NGC 4278. In NGC 1052 a weak, broad H emission component has been detected, again indicating an AGN similar in general type to those in Seyfert 1 galaxies (Filippenko and Sargent 1985). Many more similar cases have been found by careful analysis of high signal-to-noise-ratio spectra (Filippenko 1985, Filippenko 1989). Presumably there are `mini-AGNs' with smaller masses and mass accretion rates than Seyfert 1s and QSOs. Probably the values previously cited for the nucleus of M 81 are roughly applicable. There are probably wide ranges in both mass and mass accretion rate, for the luminosity () and FW0I (determined by M and R) are not well correlated and the objects do not form a one-parameter sequence in these observational parameters (Keel 1983a).

The lowest-luminosity AGN currently known is the nucleus of the late-type, nearby, dwarf spiral NGC 4395 (Filippenko and Sargent 1989). Its absolute magnitude is only M_B -10 (in the continuum). Its narrow emission lines have FWHMs < 60 km s^-1, but are otherwise characteristic of a Seyfert 2 nucleus, with [O III] 5007 / H 7, strong [O I] 6300 and [S II] 6716, 6731, and measurable He II 4686, [Ne V] 3426 and [Fe X] 6375. In addition H and H have weak broad components with FW0I ~ 4000 and 7000 km s^-1 respectively. Thus this nucleus fits in most ways the definition of a very low-luminosity Seyfert 1.8 nucleus with unusually narrow lines. The luminosity in the broad H emission component is L(H) 1.2 x 10³⁸ erg s^-1, down by about a factor 10 from M 81. This broad component would probably not be detectable in an earlier-type spiral galaxy with a correspondingly more luminous stellar nucleus.

Note however that all LINER-type spectra are not the results of photoionization. Morphologically, a certain number of LINERs are extended, nebulous objects in which the form strongly suggests that shock-wave heating is in progress. None of these show He II 4686. Clearly LINERs include objects of quite different physical nature; but those in the nuclei of galaxies are AGNs photoionized by a hard but weak spectrum (Heckman 1987, Filippenko 1989).

Finally we may briefly discuss the question as to whether our Galaxy has an AGN at its nucleus. The optical extinction to the center (A_V 35 magnitudes) is so high that there is no chance to observe the nucleus directly in the optical region. There is a compact non-thermal radio source, Sgr A*, at the center, with loops and filaments of non-thermal radio emission apparently emerging from it. Far infrared and radio-frequency velocity measurements suggest but do not prove the presence of a black hole with mass 5 x 10⁶ M. There may also be a small, highly variable x-ray source at the nucleus. However, its x-ray luminosity, about 2 x 10³⁸ erg s^-1 is smaller by a factor of approximately 100 than that of the nucleus of M 81. This is probably the best estimate or upper limit to the activity of a possible AGN in our Galaxy (Genzel and Townes 1987, Fabbiano 1988).