11.2. Seyfert 2 Galaxies
The narrow emission lines of Seyfert 2 galaxies cover a large range of ionization. Lines from [OI] to [FeXI] are observed, and the spectrum is similar to the narrow line spectrum of Seyfert 1s. The equivalent width of the lines, relative to the nonstellar continuum, is systematically larger than in Seyfert 1s. The line width distribution is broad, from 250 to 900 km s-1, with an average around 400 km s-1. Blue asymmetry is observed in most lines, being stronger in lines of higher critical density. This again is similar to the observations of Seyfert 1 galaxies, and suggests that the higher density clouds move faster with respect to the central source.
The nonstellar optical, ultraviolet and X-ray continua of Seyfert 2 nuclei are very weak but their ratios are not too different from those in Seyfert 1s, thus the overall continuum shape is similar in the two sub-groups. The extrapolation of the observed ultraviolet continuum to high energies, does not seem to give enough ionizing flux to explain the observed emission lines (i.e. the required covering factor is larger than 1). Thus, there are several indications that much of the nonstellar continuum is not observed by us.
The key to the understanding of the difference between Seyfert 1 and Seyfert 2 galaxies is in recent polarization measurements of these objects. The degree of polarization is small, 1-3%, but when the polarized flux is plotted separately from the rest, as in Fig. 36, it shows, very clearly, a typical Seyfert 1 spectrum, with strong broad emission lines of hydrogen and FeII. The small fraction of polarized flux explains why this detection escaped the notice for many years. Currently there is a handful of Seyfert 2 galaxies showing this phenomenon.
Figure 36. Top: the spectrum of the Seyfert 2 galaxy Mkn 348. Bottom: The polarized flux of Mkn 348, showing the typical broad emission lines (after Miller and Goodrich 1990).
The following physical model for Seyfert 2 galaxies has emerged (Fig. 37). A "normal" Seyfert 1 nucleus, surrounded by its BLR, is situated at the center of the system. A thick torus, of inner radius ~ 1pc and similar thickness, is present too. Most of the torus material is in molecular clouds, that are shielded from the central radiation by dust and by free electrons that evaporate from the clouds. The central BLR is obscured from some observers by the molecular torus. Such observers can only see the small fraction of BLR light that is scattered in their direction. The intensity of the scattered light depends on the Compton depth of the medium, and the degree of polarization on the observer's viewing angle. Some viewing angles are not obscured, and a Seyfert 1 type spectrum is observed. The NLR size greatly exceeds the dimension of the torus, and the narrow emission lines are seen by all observers. However, the NLR illumination and ionization is not isotropic, because of the torus, and the observed NLR is likely to attain a jet-like structure. There are radio observations and narrow emission line maps that support the claim for a jet-like NLR in Seyfert 2 galaxies.
Figure 37. A unified model for Seyfert galaxies. The central source and the BLR are surrounded by a thick torus of molecular gas. The Seyfert 1 galaxies are those objects observed from the pole direction. Seyfert 2 galaxies are those sources whose inner parts can only be seen through reflected radiation.
The above model provides a natural explanation for the difference between Seyfert 1 and Seyfert 2 galaxies. One suggestion is that the different viewing direction is the only distinction between the two groups of objects. In this case the thickness and dimension of the torus can be estimated from the relative number of Seyfert 1 and Seyfert 2 galaxies. There are difficulties too, such as several well studied Seyfert 2s where the polarized flux is extremely small and no broad lines are seen. The role of dust is not very clear and heavy reddening is likely to be present, at least in some directions. Light scattered by dust is polarized in a wavelength dependent way, and there are observational ways to test this idea. The Compton depth should be large enough to scatter part of the broad line radiation, but not too large to smear out the line and continuum variability in Seyfert 1 galaxies. There are theoretical uncertainties too, to do with the structure and stability of the torus.
An alternative explanation to the differences between the two groups of galaxies is related to their variability. The broad emission lines of some Seyfert 1 galaxies exhibit a large amplitude variations. In some objects the variability so large that the spectrum at minimum light resembles a Seyfert 2 spectrum. It has been suggested that some Seyfert nuclei spend a large fraction of their active phase in a "turned-off" state, when the ionized flux is too weak to excite the gas. The recombination time of the broad line gas is short and the lines disappear several hours after the continuum decline. The recombination time of the low density gas is long enough to show strong narrow lines many years after the decline. Such an object, with strong narrow lines and weak continuum, will be classified as a Seyfert 2 galaxy. A possible cause for the drop in luminosity is a large decrease in the accretion rate. The big continuum bump observed in broad line AGNs has been attributed to emission from accretion disks. If this is indeed the case then the time scale for its fading is very long. It is hoped that future HST observations will help to decide whether the continuum of Seyfert 2 galaxies shows any sign of such a bump.
To summarize, Seyfert 2 galaxies may be Seyfert 1 nuclei that are hidden in space or in time. This may explain many, perhaps most observed properties of Seyfert 2s. However, we should not neglect the possibility of the presence of genuine narrow line objects, with no BLR at all.