Annu. Rev. Astron. Astrophys. 2000. 38: 761-814 Copyright © 2000 by Annual Reviews. All rights reserved

3.3.2. The Source of Far-Infrared Luminosity in Seyfert Galaxies and QSOs

IRAS data have shown that the infrared spectral energy distributions (SEDs) of Seyfert galaxies and (radio-quiet) QSOs are dominated by thermal (dust) emission (Sanders et al 1989, Barvainis 1990, Pier & Krolik 1993, Granato & Danese 1994, Rowan-Robinson 1995, Granato et al 1997). ISOPHOT observations of the 2-200 µm SEDs of Seyferts, radio galaxies, and QSOs have confirmed these conclusions and improved the quality and spectral detail of the data (Haas et al 1999, Wilkes et al 1999, Rodriguez Espinosa et al 1996, Perez Garcia et al 1998). As part of the European Central Quasar Program (~ 70 QSOs and radio galaxies), Haas et al (1998a, 1999) reported ISOPHOT/IRAS SEDs, with additional 1.3 mm points from the IRAM 30 m telescope, for two dozen (mainly radio-quiet) PG QSOs. They conclude that all QSOs in their sample, including the four radio-loud ones, have substantial amounts of cool and moderately warm (20 to 60 K) dust radiating at 60 µm, in addition to warmer (circumnuclear) dust radiating in the mid-infrared. The derived dust masses (10^7±1 M) are typical for the total dust masses in gas-rich, normal galaxies, which suggests that the 60 µm emission is probing the large-scale disks of the host galaxies. Andreani (2000) has observed a complete subset of 34 QSOs from the Edinburgh and ESO QSO surveys that sample the bright end of the (local) QSO luminosity function. From her 11-160 µm ISOPHOT observations, she concludes that the mid-IR and far-IR fluxes are poorly correlated, as are the far-IR and blue-band fluxes. In contrast, the 60, 100 and 160 µm fluxes seem to be well correlated. This finding suggests that the far-IR emission in QSOs is a distinct physical component that may not be physically related to the shorter wavelength emission. Wilkes et al (1999) reported the first results of the US Central QSO program that contains another 70 objects and extends to z = 4.7. These data confirm as well the presence of thermal dust emission with a wide range of temperatures. The objects selected in that survey extend to the high-luminosity tail of the high-z QSO population with infrared luminosities of several 10¹⁵ L. Van Bemmel et al (1999) reported observations of matched pairs of QSOs and radio galaxies (radio power, distance, etc). They found that QSOs are actually more luminous far-IR sources than radio galaxies, contrary to simple unification schemes.

The key question that must be answered next concerns the nature of the energy source(s) powering the IR emission: direct radiation from the central AGN, or distributed star formation in the host galaxy? The near- and mid-IR emission ( 30 µm) is very likely reradiated emission from the AGN accretion disk (the "Big Blue (or EUV) Bump"; Section 3.3.3). The 30 µm SEDs can be well matched with the Pier and Krolik (1992, 1993), Granato & Danese (1994) models of AGN heated dusty tori (scale size ~ 100 pc), with an additional component of somewhat cooler dust.

More difficult is the answer to the question of whether the 30 µm emission is reradiated AGN luminosity as well. A compact and thick torus, as proposed by Pier & Krolik (1992), for instance, definitely does not produce a broad enough SED to explain the far-IR emission. A clumpy, extended, and lower column den- sity torus (Granato et al 1997), a warped disk (Sanders et al 1989), or a tapered disk (Efstathiou & Rowan-Robinson 1995) are more successful in qualitatively accounting for the observed broad SEDs. However, even these models work only marginally for a quantitative modeling of the emission at 100 µm. For the CfA sample, the correlation between 60 + 100 µm band luminosity and (extinction-corrected) [OIII] luminosity is not impressive for Seyfert 1s, and outright poor for Seyfert 2s.

For QSOs, no clear answer has emerged yet regarding the nature of the far-IR continuum. On the one hand, Sanders et al (1989) concluded that the far-IR emission in PG quasars mainly results from AGN reradiation. On the other hand, Rowan-Robinson (1995), Haas et al (1999) argued that the far-IR emission is caused by star-forming activity in the QSO hosts. In the Sanders et al model, a warped disk could intercept at least 10% of the luminosity of the central AGN. In their sample of ~ 50 radio-quiet PG quasars with available IRAS luminosities, the average ratio of total infrared to total UV+visible ["Big Blue (or EUV) Bump"] luminosities is about 0.4. The 30 µm luminosity is about half of the total infrared luminosity, thus requiring that about 15% of the nuclear luminosity be absorbed and reradiated at 10² to 10³ pc from the AGN. This may be possible if substantial warps are present on that scale. In contrast, Rowan-Robinson (1995) cited the cases of three PG QSOs (0157+001, 1148+549, 1543+489) and the Seyfert 1/ULIRG Mrk 231 where the far-IR luminosity exceeds the optical+UV luminosity, a result that is not possible in the reradiation scenario (see also Section 3.4.5). An interpretation of the far-infared emission in terms of star formation is also favored by the far-IR/radio relationship in Seyferts and radio-quiet QSOs. Colina & Perez-Olea (1995, and references therein) show that the ratio of 60 + 100 µm IRAS luminosity to 5 GHz radio luminosity in these objects is in excellent agreement with the ratio found in star-forming spirals of a wide range of luminosities.

For Seyfert galaxies, a clearer picture is emerging. For the CfA Seyfert sample of Clavel et al (1998), the median ratio of 60 + 100 µm luminosity to B-band luminosity is 0.9 for Seyfert 1s and 1.6 for Seyfert 2s. Assuming that about 30% of the total UV+visible luminosity is contained in the B-band (as in the PG QSOs of Sanders et al 1989), the fraction of bolometric luminosity emerging in the far-IR in the Clavel et al Seyferts is about 40%. This is probably too large for a reradiation scenario. Perez Garcia et al (1998) have investigated the infrared spectral energy distribution of ten Seyfert galaxies with a combination of IRAS and ISOPHOT data. They decomposed the 4-200 µm spectra into a sum of blackbodies with ^-2 emissivities. In nine of the ten galaxies studied the infrared spectrum decomposes into three components, each with a similar, narrow range of temperatures. A 110-150 K component dominates the mid-infrared. The far-IR emission comes from a combination of a 40-50 K (30 to 100 µm) and a 10-20 K (150 to 200 µm) component. Perez Garcia et al concluded from the similarity of the temperatures of the three components that they represent well-separated spatial regions (compact torus, star-forming regions, and diffuse ISM), rather than a temperature range in a single physical component (i.e. a circumnuclear torus or warped disk). In this interpretation, the 40-50 K temperature component dominating the 60 + 100 µm IRAS band is a measure of star formation in the disk of the Seyfert galaxies. Finally, the finding (Figure 10) that the ratio of UIB luminosity to 60 + 100 µm luminosity in the Clavel et al (1998) Seyferts is essentially the same as in starburst galaxies also strongly supports an interpretation of the far-IR emission in terms of (disk) star formation.