COMPACT RADIO CORES IN SEYFERT AND STARBURST GALAXIES

(in "Workshop on Relationship between AGN and Starburst galaxies", Taipei, 1991, (ASP conf. series vol. 31). ed. A. Filippenko)

R.P.NORRIS

Australia Telescope National Facility, PO Box 76, Epping 2121, Australia

A.L.ROY

School of Physics, University of Sydney, 2001, Australia and Australia Telescope National Facility, PO Box 76, Epping 2121, Australia

D.A.ALLEN

Anglo-Australian Observatory, PO Box 296, Epping, 2121, Australia

M.J.KESTEVEN, E.R.TROUP, and J.E.REYNOLDS

Australia Telescope National Facility, PO Box 76, Epping 2121, Australia

ABSTRACT

We have examined samples of Seyfert and starburst galaxies with a 275-km radio interferometer, which discriminates sharply between the quasar and starburst mechanisms, and is essentially a tool for detecting AGN cores. Here we present two distinct results from this work.

One is that we find compact radio cores are common in our Seyfert samples, but rare in our starburst samples, confirming an intrinsic difference between Seyfert and starburst galaxies. Even amongst selected infrared-luminous galaxies, we find that compact radio cores are common in those with Seyfert-like spectra, but rare otherwise, implying that some of even the most luminous ones are powered by starburst activity.

An even more surprising result is that we find a clear observational difference between Seyfert 1 (Sy1) and Seyfert 2 (Sy2) galaxies, in that compact radio cores are much more common in Sy2 galaxies than in Sy1 galaxies. This result at first seems to contradict the unified model of Seyfert galaxies, but we suggest a mechanism which can reconcile it with, and even lend support to, the unified model.

INTRODUCTION

It appears that a significant fraction of the energy emitted by some Seyfert galaxies is provided by starburst activity (e.g. Telesco et al., 1984), and that at least some starburst galaxies have compact cores. This therefore raises the question of whether there is any real difference between Seyfert and starburst galaxies. Are starburst galaxies simply Seyferts with obscured cores, or are some Seyferts just a particular form of starburst galaxy?

This debate is accompanied by another concerning the source of energy for the extremely luminous far-infrared galaxies. There is strong evidence (e.g. Norris, 1988) that at least one of them, Arp220, is powered by a quasar core and it has been suggested (e.g. Sanders et al., 1988) that most such objects are powered by quasar cores.

A third such debate concerns the nature of the two main classes (Sy1 and Sy2) of Seyfert galaxies. In the quest for a "unified model" of active galaxies, a model which has gained popularity is one in which the apparent difference between Sy1 and Sy2 is caused simply by an orientation effect, in which the broad line region is surrounded by a torus of optically thick material. Seen end-on, the broad line region is visible and so the source appears as a Sy1. When viewed from the side, however, the broad line region is occulted and the galaxy appears as a Sy2. This model is especially supported by the observations of Antonucci and Miller (1985) and Miller and Goodrich (1990) who have found that several apparently Sy2 galaxies do in fact have a Sy1 spectrum when viewed in the polarised light scattered by hot electrons outside the active nucleus. We call this model the standard unified model.

Radio observations with arcsec resolution cannot easily distinguish between these sources of radio emission. The radio emission from both starburst regions and from Seyfert outflows have similar spectra and morphology, and the steep spectrum of even the compact cores seen in VLA maps of Seyferts suggests that they might contain a significant nuclear starburst component. Long baseline radio interferometry, on the other hand, which is sensitive only to compact, high brightness objects, acts as an acute filter, discriminating between the two spatial scales.

Specifically, the 275-km Parkes-Tidbinbilla Interferometer (PTI) used here is sensitive only to structure with a size of 0.1 arcsec, corresponding to 20 pc at z=0.01, with a brightness temperature > 105 K. Thus the PTI can detect parsec-sized quasar cores, which normally have brightness temperatures ~ 108 K, but is blind to the extended kpc-sized starburst regions or extended radio beams, which have typical brightness temperatures of 104 K. Details of the observations and techniques can be found in Norris et al. (1990, 1991).

THE SAMPLES

We have already published results on optically-selected Seyfert and starburst galaxies and on a sample of infrared-selected ELF (extremely luminous far-infrared) galaxies (Norris et al., 1990). To that earlier infrared-selected sample we now add a sample selected from the infrared-selected list of Seyferts compiled by de Grijp et al. (1987), and a sample of infrared-selected starbursts chosen from the list of Allen et al. (1991).

The advantage of using infrared-selected samples is that such a sample will be untainted by selection effects due to orientation, since any obscuring dust should be optically thin at the IRAS wavelengths used for the selection. On the other hand, we acknowledge the faint possibility that infrared-selected galaxies might differ in some intrinsic way from their optically-selected counterparts, although present indications are that the infrared-selected sample has properties similar to the optically-selected sample.

In order to avoid possible confusion from borderline classifications, we use here only those galaxies with a reliable classification of starburst (or HII), Sy1, or Sy2, rejecting all intermediate classes.

RESULTS

The detection rates in the combined sample are shown in Figure 1(a). It is clear that the detection rate of starbursts (6%) is very much lower than that in Seyferts (38%). When we divide the Seyferts into Sy1 and Sy2, we get a surprising result, in that the detection rate for Sy2 (46%) is much greater than that for Sy1 (17%).

Figure 1.

PTI detection rates for starburst galaxies and Seyfert galaxies (separated into types 1 and 2). Each column represents the number of sources observed, and the filled section indicates the number of detections. (a) All infrared-selected data. (b) Infrared-selected sample from Norris et al.,1990. (c) New data

To check on this result, we have shown the two subsamples (that from Norris et al. (1990) and that from the new data) separately in Figures 1(b) and 1(c). It is clear that the result described above is also obtained in each of the subsamples, confirming its significance. Thus our observations convincingly demonstrate that starburst galaxies rarely have radio cores, and that compact radio cores are much more common in Sy2 galaxies than in Sy1 galaxies.

We note that, as suggested by Matt Malkan at this conference, this result could be artificially induced if the Sy1 galaxies had systematically higher redshifts than Sy2 galaxies. However, with the exception of one high-redshift Sy1 galaxy (04505-2958, with z=0.286) there is no significant difference between the redshift distributions of the Sy1 and Sy2 galaxies. Even including this high-redshift galaxy does not have sufficient effect to cause the observed result.

We therefore do not believe this potential selection effect to be significant.

Another complication, suggested by Julian Krolik (private communication), could arise if the samples of Sy1 and Sy2 galaxies had different bolometric luminosity distributions. However, low-resolution radio data (Roy et al., 1991) and the IRAS fluxes do not indicate a significant difference between the two samples, so we have no reason to suspect a significant difference in the bolometric luminosities.

SEYFERTS AND STARBURSTS

The result above demonstrates a clear difference between Seyfert and starburst galaxies: Seyfert galaxies frequently contain radio cores while starburst galaxies rarely do so. Thus, in the absence of any convincing model in which starburst galaxies could shield the core, we infer that starburst galaxies are intrinsically different from Seyfert galaxies.

This result applies even at the highest infrared luminosities, and demonstrates that not all of the infrared-luminous galaxies contain obscured quasar cores. Those infrared-luminous galaxies that have starburst spectra rarely contain a radio core, and so are probably powered primarily by starburst activity.

THE UNIFIED SEYFERT MODEL

We have obtained the surprising result that Sy1 galaxies have compact radio cores significantly less often than Sy2 galaxies. This appears at first to be difficult to reconcile either with the standard model or with the alternative, and common, hypothesis that Sy1 galaxies are in some way more like the powerful radio galaxies and quasars than are Sy2 galaxies. Here we offer a model to explain this result.

Figure 2

The proposed model to explain why we see radio cores in Sy2 galaxies but not in Sy1 galaxies. The NLR clouds are optically thick at the observing frequency of 2.3 GHz, and so the radio core cannot be seen in Seyfert 1 galaxies.

Our model, shown in Fig 2, invokes the ionised clouds that are observed, principally in narrow-line emission (e.g., Wilson et al., 1988; Pogge, 1989; Evans et al., 1991), to be coincident with radio structures. These clouds of gas appear to be ionised by the outflows of plasma associated with radio jets, and each has a typical electron density of 103 cm-3 and a size of tens to hundreds of pc. The total path length (which may encompass several such clouds) through the ionised gas is less certain, and will depend on the filling factor. Assuming a total path length of 100 pc, we derive an optical depth of 5 at 2.3 GHz.

This result demonstrates that free-free absorption in the NLR clouds can completely obscure the compact radio core at 2.3 GHz. Thus, following the standard unified model in which Sy1 galaxies are observed along the radio axis and Sy2 galaxies are observed orthogonal to it, this model would mean that the NLR clouds would obscure the radio core in Sy1, so that only the extended radio emission from the plasma outflow itself would be observable in Sy1 galaxies. We note that, if the covering factor is not unity, then some small fraction of Sy1 galaxies will be observed to have compact radio cores.

In Sy2 galaxies, the radio core would be observed through the dusty torus. The inner edge of this would certainly be ionised, but, as discussed by Pogge (1989), it is likely that the shape of the accretion disc shadows the bulk of the material, so that only a relatively small column density of gas is ionised. Calculations by Krolik & Lepp (1989) indicate that the optical depth of the torus, because of free-free absorption, might approach unity in the GHz range. However, this optical depth has a large uncertainty, and depends heavily on poorly-known parameters, so we consider that this model needs better observational data before it can be confidently used to calculate the free-free optical depth through the torus.

Therefore we do not consider these calculations to pose a serious problem for our model.

CONCLUSION

We have observed infrared-selected samples of Seyfert galaxies with a long-baseline interferometer and have demonstrated that compact radio cores are much more common in Seyfert 2 galaxies than in Seyfert 1 galaxies, and are rarely found in starburst galaxies. We interpret the result on starburst galaxies to indicate that starburst galaxies are intrinsically different from Seyferts. The result distinguishing between the two classes of Seyfert, on the other hand, can be explained by invoking an orientation effect. By considering the radio optical depth of the recently observed ionised clouds entrained within the radio outflows, we have shown that our observations are not only consistent with the standard unified Seyfert model but may actually lend further support to this model.

ACKNOWLEDGEMENTS

We thank Alex Fillipenko, Matt Malkan, and Julian Krolik for helpful comments.

References

• Allen, D.A., Roche, P.F., & Norris, R.P., 1985, MNRAS, 213, 67p.
• Allen, D.A., Norris, R.P., Meadows, V.S., & Roche, P.F., 1991, MNRAS, 248, 528.
• Antonucci, R.R.J., & Miller, J.S., 1985, ApJ, 297, 621.
• de Grijp, M.H.K., Miley, G.K., and Lub, J., 1987, Astr. Astrophys. Suppl., 70, 95.
• Evans, I.N., Ford, H.C., Kinney, A.L., Antonucci, R.R.J., Armus, L., & Caganoff, S., 1991, Astrophys. J., 369, L27.
• Keel, W.C., 1984, ApJ, 282, 75.
• Krolik, J.H., & Lepp, S., 1989, Astrophys. J., 347, 179.
• Miller, J.S, & Goodrich, R.W., 1990, Astrophys. J., 355, 456.
• Norris, R.P., 1988, MNRAS, 230, 345.
• Norris, R.P., Allen, D.A., Sramek, R.A., Kesteven, M.J., & Troup, E.R., 1990, Astrophys. J., 359, 291.
• Norris, R.P., Roy, A.L., Allen, D.A., Kesteven, M.J., Troup, E.R., & Reynolds, J.E., 1991, in preparation.
• Pogge, R.W., 1989, Astrophys. J., 345, 730.
• Roy, A.L., Norris, R.P., Allen, D.A., Kesteven, M.J., Troup, E.R., & Reynolds, J.E., 1991, in preparation.
• Sanders, D.B., Soifer, B.T., Elias, J.H., Neugebauer, G., and Matthews, K., 1988, ApJ, 328, L35.
• Telesco, C.M., Becklin, E.E., Wynn-Williams, C.G., & Harper, D.A., 1984, ApJ, 282, 427.
• Wilson, A.S., Ward, M.J., & Haniff, C.A., 1988, Astrophys. J., 334, 121.