Dusty Tori in Seyfert Galaxies

The unified model for Seyfert galaxies critically depends upon the existence of a geometrically and optically thick dust torus surrounding the AGN core. Theoretically it is very difficult to sustain a thick dusty torus in a galaxy, as it should either become a thin disk or dissipate over a very short period of time. But evidence has been accumulating which points to the existence of thick dusty tori in AGN. Indirect evidence for a thick torus is given by the numerous ionisation cones observed in Seyfert galaxies (eg. Pogge 1989; Colbert et al 1996; Simpson et al 1997). Amongst the best direct evidence for a thick torus is the HST observation of NGC 4261, which clearly shows a large gaseous disk 250 parsecs in diameter, and more recently, the direct observation of a parsec scale torus using high resolution radio continuum imaging (Gallimore, Baum & O'Dea 1997).

Spectropolarimetry

Motivation

The detection of broad emission lines scattered into our line of sight by free electrons and/or dust in NGC 1068, the prototypical Seyfert 2 galaxy (Antonucci & Miller 1985), and other Seyfert 2 galaxies (Young et al 1996; Inglis et al 1993; Hines & Wills 1993; Tran, Miller & Kay 1992; Miller & Goodrich 1990) has strengthened the unified model for Seyfert galaxies. Unfortunately, all previous studies are known to be biased in their continuum polarization properties, with samples chosen to be either highly polarized or having rising polarization in the red. Such characteristics are not typical of Seyfert galaxies in general. In addition to these biases, there have been very few non-detections published, and there has been no consistency in ensuring non-detections are at the same signal-to-noise ratio in percentage polarized flux. To overcome these problems, Heisler, Lumsden & Bailey (1997) undertook a new spectropolarimetry survey, based on a 60 m flux limited sample consisting of 16 Seyfert 2 galaxies. The 60 m emission is typically a measure of global emission from the galaxy, so is more isotropic than most properties that were previously used to define samples of AGN.

What causes polarisation?

There are several possible mechanisms that may contribute to polarisation in AGN.

a). Non-thermal Polarisation:
Many high polarisation AGN (20-50% polarised) display apparently nonthermal emission spectra and typically variable polarisation (eg. blazars and powerful radio galaxies) which is attributed to synchrotron emission. According to the unification scenario, blazars are radio-loud AGN that contain a relativistic jet which is viewed along our line of sight (eg. Urry & Padovani 1995). Synchrotron emission occurs when high speed (close to velocity of light) charged particles rapidly revolve around a strong magnetic field, emitting highly plane-polarised electromagnetic waves with their polarisation plane perpendicular to the magnetic field. In blazars the synchrotron emission is believed to be produced in the relativistic jets.

b). Thermal Polarisation:
Low polarisation AGN, such as Seyfert galaxies, are believed to be dominated by thermal polarisation processes.

i). Scattering:
Scattering of flux into the observers line of sight naturally produces polarisation. Scattering may be caused by dust grains or by electrons in AGN depending on where the scattering occurs. In the central regions of an AGN the gas that directly faces the AGN is highly ionised, and photons are scattered by free electrons, producing wavelength independent polarisation. In regions slightly further away from the central engine, dust scattering by circumnuclear molecular clouds may contribute to polarisation. The dependence of polarisation with wavelength for dust scattering depends on the scattering angle (Webb et al 1993) and does not necessarily follow the simple Rayleigh scattering law in which the ionised flux rises towards shorter wavelengths.

ii). Dichroic extinction:
Large scale alignment of nonspherical dust grains along the line of sight to the AGN can produce polarisation via dichroic extinction. Consider the simplified classical view of light where the electric and magnetic field vectors are perpendicular to each other and to the direction of propagation of the light. In the presence of a magnetic field, elongated dust grains will align their long axis perpendicular to the direction of the magnetic field. Light with electric vectors in the same direction as the elongated dust grains will be preferentially absorbed, producing a net polarisation which is parallel to the direction of the magnetic field. Such magnetic fields in AGN may be part of the internal structure of dense molecular torus clouds (Krolik & Begelman 1988).

Results of Spectropolarimetry Survey

The most striking result of the Heisler, Lumsden & Bailey survey is displayed in Figure 1. Those galaxies with detected polarised broad lines (represented by filled circles) have smaller values for the far-infrared (FIR) flux ratio f/f, and smaller Balmer decrements as measured by the ratio of the narrow lines, H/H, compared to Seyferts for which polarised broad lines are not detected (represented by open squares). The segregation between Seyfert 2 galaxies with and without polarised broad lines is most obvious for the shorter wavelength ratios, f/f and f/f (Figures 2a,2b).

 
Figure 1: Correlation between Balmer decrement log(H/H), FIR flux ratio and the detectability of polarised broad lines in the IRAS-selected Seyfert 2 sample. The symbols have the following meaning: filled circles - polarised broad lines detected; open squares - polarised broad lines not detected. Optical spectropolarimetric observations were obtained at the Anglo-Australian Telescope during 16 - 18 August 1995. The half-waveplate module, built by the University of Hertfordshire, was used in conjunction with the RGO spectrograph, the 270R grating and the 25 cm camera, giving a spectral resolution of 8.4 Å. Our observations yield a three sigma detection at 0.5% polarization.

 
Figure 2: Far-infrared two-colour plots of
a). f/f against f/ f;
b). f/f against f/ f;
where the symbols have the same meaning as in Figure 1. The dashed line represents the boundary below which the majority of galaxies are classified as Seyfert 1 galaxies based on the broad line region being directly visible in optical spectra (Spinoglio et al 1995)

Is it possible that Seyferts with FIR flux ratios f/f4 and lacking polarised broad lines in optical spectra are simply pure starburst galaxies without an AGN core? To address this question we have made use of data from the literature (Heisler, Lumsden & Bailey 1997; Roy et al 1994; Norris et al 1990) that was obtained using the Parkes-Tidbinbilla Interferometer (PTI) at 2.3 GHz. The PTI is sensitive only to structures with brightness temperatures > 10 K, and sizes less than 0.1", which corresponds to < 35 parsecs at the mean redshift, z=0.018, of our sample (assuming a Hubble constant H75 km s Mpc). Thus, the PTI is blind to extended star formation regions with typical brightness temperatures < 10 K and ideal for detecting compact radio cores associated with AGN. There is no significant difference in the detection rate for Seyfert 2 galaxies with and without polarised broad lines, with 71% and 67% containing compact radio cores, respectively. These observed detection rates are consistent with that found for other samples of Seyfert 2 galaxies, and indicate that they are drawn from the same parent population of galaxies containing an active galactic nucleus. X-ray emission is a powerful way for distinguishing AGN from starbursts, and it will soon be possible to undertake such a project with the Advanced X-ray Astrophysics Facility (AXAF).

The above results can not be explained by a distance or luminosity effect. The redshift and FIR luminosity distributions are similar for the Seyferts with and without polarised broad lines. The Kolmogorov-Smirnov statistical test for the cumulative distributions of the redshift and FIR luminosity, shows that we can reject the hypothesis that the two samples are different at the 96% and 98% confidence level, respectively.

Heisler, Lumsden & Bailey (1997) interpret these results within the context of the unified model, with the assumption that the torus has a significant opacity, so that only the longer wavelengths (m) could be assured of escaping directly. The inner side of the torus directly facing the active galactic nucleus has a high dust temperature (TK), contributing mostly at the shorter FIR wavelengths of 12 and 25 m, and if we do not view the inner face of the torus almost directly, the 12 and 25 m radiation will be reprocessed to longer wavelengths, and the galaxy will have larger FIR flux ratios, f/f, compared to a face-on orientation. This is consistent with theoretical models (Pier & Krolik 1993; Granato & Danese 1994; Efstathiou & Rowan-Robinson 1995) which indicate that an increase in the inclination of dust tori causes steepening of spectral indices at FIR wavelengths, and observational data (Giuricin et al 1995; Maiolino et al 1995; Spinoglio et al 1995) which indicate that Seyfert 1 galaxies have smaller FIR flux ratios, f/f and f/f, compared to Seyfert 2 galaxies. A major implication of these results is that the scattering particles producing the observed polarized broad lines must lie very close to or possibly within the plane of the torus. A schematic diagram illustrating how our results fit within the context of unification between Seyfert 1 and Seyfert 2 galaxies is displayed in Figure 3.

 
Figure 3: Schematic diagram (not to scale) of the central region of a Seyfert galaxy illustrating the effect of viewing angle on infrared colours and ability to detect the broad line region (BLR). We have indicated the BLR which has a typical size of <1 parsec, the narrow line region (NLR) clouds located <1 kiloparsec from the BLR, and the scattering particles located < 300 parsecs from the BLR. The temperature of the dust in the torus decreases radially outwards. The viewing angles producing a Seyfert 1, a Seyfert 2 with polarised broad lines and a Seyfert 2 without polarised broad lines, are illustrated. An edge-on view will result in an optical spectrum which is characteristic of a Seyfert 2 with ``cool" infrared colours, and relatively large values of internal extinction, due to obscuration by the cool dust in the outer regions of the torus. A face-on viewing angle will provide a direct view of the BLR and of the hot dust at the inner edge of the torus resulting in a Seyfert 1 spectrum and ``hot" infrared colours. At intermediate viewing angles, the BLR is detected only by scattered light, the FIR colours have ``warm" dust temperatures, and the internal extinction to the NLR will be small relative to an edge-on orientation.

Probing the Torus via Far-Infrared Colours

The Heisler, Lumsden & Bailey (1997) results show a clear progression of colour through Seyfert 2 galaxies with polarised broad lines to those in which polarised broad lines are not detected. This suggests a progression of increasing optical depths of dust. Recently, Dopita, Heisler, Lumsden & Bailey (1998) carried out an investigation to test the possibility that the IRAS flux ratios of Seyfert galaxies are attributed to extinction of an intrinsic Seyfert 1 spectrum (due to the emission by hot dust near its sublimation temperature) with optical depth which is variable from one object to another, and which tends to increase with viewing angle.

We began our investigation by plotting the positions of the Seyferts investigated by Heisler et al (1997) in the FIR flux ratio diagrams that are most commonly found in the literature: log(f/f) vs. log(f/f) and log(f/f) vs. log(f/f) (Figure 4a and b, respectively). The solid line in these figures represents a theoretical reddening line computed using the infrared extinctions given by Dwek et al (1997). The zero point for this reddening line was fixed using those Seyfert 1 galaxies with the smallest FIR flux ratios from the literature (Heisler & Vader, 1994; and Rush, Malkan & Spinoglio, 1993). The assumption is that Seyfert 2 galaxies have the intrinsic colours of Seyfert 1 galaxies, reddened by various amounts of an obscuring screen of dust. The symbols have the same meaning as in section 2.1.3 where Seyfert 2s with and without polarised broad lines are represented by filled circles and open squares, respectively. It is apparent that while many of the Seyferts with polarised broad lines lie within the warmer regions of the plots, there is a large scatter.

Stepping outside the boundary of the traditional IRAS flux ratio diagrams, we plotted the diagram of log(f/f) vs. log(f/f). We obtained the surprising result that all of the Seyferts fall in a straight line along the reddening line, with the Seyferts observed to have polarised broad lines lying at the low extinction end of the plot (Figure 4c). As far as we are aware such a plot has not been previously published in the literature.

 
Figure 4: Far-infrared two-colour diagrams of
a) log(f/f) vs. log( f/f).
b) log(f/f) vs. log( f/f).
c) log(f/f) vs. log( f/f)
The symbols have the same meaning as in Figure 1. The solid line represents the reddening line and the optical depths at 12 m are shown by tickmarks in following units (0, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 5, 6).

What causes such a tight correlation in this FIR flux ratio diagram? We know that f/f discriminates between Seyfert 2 galaxies with and without polarised broad lines, presumably due to the dust torus being optically thick for wavelengths m, indicating that this ratio is tied in with the temperature of the hot dust. The correlation between the ratio of f/f and f/f can be explained in terms of inclination of the torus to our line of sight as follows. When we view a Seyfert 2 galaxy nearly face-on and detect polarised broad lines, the FIR is dominated by emission from the hot dust near the nucleus, resulting in small values for f/f and f/f. As the viewing angle is increased (referenced to the polar direction) the hot dust becomes obscured and the ratios reflect the cooler dust observed. As we progress to larger viewing angles (approaching edge-on), the FIR emission from the cooler regions of the outer tours and from the host galaxy itself start to dominate. This cool dust will eventually cause the ratio of log(f/f) to level off at a value (1.6 based on Figure 4c) determined by the ratio of the emissivity of silicates which dominate at 100 m (eg. Puget et al 1985), and the emissivity of poly-aromatic hydrocarbon (PAH) molecules which dominate at 12 m. The effects of star formation on the FIR flux ratio plots are discussed further by Dopita, Heisler, Lumsden & Bailey (1998).

The fact that the torus is optically thick for 50 m will have potentially serious consequences for sample selection criteria of AGN. For example consider samples of Seyferts chosen by their warm FIR colours, or small FIR flux ratios, (eg. de Grijp et al 1985, Heisler 1991). These samples will have a bias towards warmer objects, and so, because most objects in it are close to the IRAS detection limit, the Seyfert 1 galaxies will tend to be slightly more distant than the Seyfert 2 galaxies. This will affect many statistical studies including number counts of Seyfert 1 and Seyfert 2 galaxies, and flux and luminosity distributions due to Malmquist bias. This now raises suspicion on the results of Roy et al (1994) in which Seyfert 2 galaxies were found to have radio core detections more often than Seyfert 1 galaxies, as it may be attributed to the fact that the torus is optically thick at mid-infrared wavelengths (Roy, Norris & Heisler, 1998), and that Roy et al chose Seyfert galaxies from the de Grijp et al sample. As such, caution should be exercised when choosing samples of AGN.

© Copyright Astronomical Society of Australia 1997