Annu. Rev. Astron. Astrophys. 2005. 43: xxx-xxx Copyright © 2005 by Annual Reviews. All rights reserved

5.5. ULIRGs and Active Galactic Nuclei at High Redshifts

SMGs are massive ULIRGs at high redshift. One of the key question discussed above for the z 1 galaxies is to distinguish whether starburst or AGN activity powers the dust heating and associated infrared emission. The presence of an AGN in galaxies can be investigated using optical/near-infrared, emission line diagnostics and/or X-ray observations. But the identification of the presence of an AGN does not mean that it is the dominant source of the far-infrared emission. Alexander et al. (2003; see also Almaini et al. 2003) use Chandra observations of the CDF-N to constrain the X-ray properties of 10 bright SMGs. Half of the sample has flat X-ray spectral slopes and luminous X-ray emission, suggesting obscured AGN activity. However, a comparison of the AGN-classified sources to the well-studied, heavily obscured AGN NGC 6240 suggests that the AGN contributes on average a negligible fraction (about 1.4%) of the submillimeter emission. For the MAMBO sources, similar results are found: only one out of the nine MAMBO sources studied by Ivison et al. (2004) has an X-ray counterpart. It has, as expected from low redshift ULIRGs observations (e.g., Rigopoulou et al. 1999), a different mid-infrared SED than the starburst dominated sources. About 75% of their sample has rest-frame mid-infrared to far-infrared SED commensurate with obscured starburst. Swinbank et al. (2005), using AGN indicators provided by near-infrared spectra, estimate that AGNs are present in at least 40% of the galaxies in their sample of 30 SMGs. Emission-line diagnostics suggest that star formation is the dominant power source. However, the composite spectrum for the galaxies that individually show no signs of an AGN in their near-infrared spectra appears to show an underlying broad H line. This suggests that even these galaxies may host a low-luminosity AGN that is undetectable in the individual spectra. All these studies tend to show that starburst activity is the dominant source of power of dust emission in the far-infrared. Still, it is rather difficult to estimate the true "contamination" by the AGN. To go deeper, Chapman et al. (2004) tried an original approach. They observe a sample of identified SMGs at high angular resolution in the radio and use the radio emission as a proxy for the far-infrared emission. This assumption is based on the well-known very tight far-infrared/radio correlation mentioned above. If detected, an extended radio (and thus far-infrared) component is likely to arise from the star formation. The detection of extended emission requires sub-arcsec resolution to map emission on kpc-scales. These are accessible by radio interferometry (they are well beyond far-infrared and submillimeter facilities capabilities). They find that for 70% of the SMG sample, the MERLIN/VLA radio exhibits resolved radio emission which mirrors the general form of the rest frame UV morphology seen by HST. The galaxies are extended on scales of about 10 kpc. They interpret this as a strong support for the hypothesis that radio emission traces spatially extended massive star formation within these galaxies. This is clearly different from what is seen in local ULIRGs where the far-infrared/radio emission is concentrated in the compact nuclear region with an extend less than 1 kpc. In the remaining 30% of the SMG sample, the radio emission is more compact (essentially unresolved). This is a signature of either a compact nuclear starburst and/or an AGN.

In conclusion, the exact fraction of distant submillimeter and millimeter galaxies containing an energetically dominant AGN is difficult to extract from observations. However, even in the systems containing an unambiguously powerfully AGN, the far-infrared emission seems to be powered by the star formation. Surprisingly, this seems to be also the case in distant QSOs. Recently, Beelen (2004) has shown that the far-infrared and blue luminosities from the host galaxies of distant radio-quiet QSOs, are slightly correlated. The far-infrared and radio emission of these quasars follow the radio-infrared correlation observed in local ULIRGs (Yun et al. 2001), providing a first indication that the dust is predominantly heated by the star-formation activity rather than by the AGN. Moreover, the non-linearity between the far-infrared and blue luminosities is also an indication that the heating mechanism of the dust is not directly linked to the AGN. However, the presence of this correlation could suggest a causal connection between the formation of stars in the host galaxy and the activity of the central super massive black hole. This connection has been successfully modeled by Granato et al. (2004).

Finally Houck et al. (2005) and Yan et al. (2005) demonstrate the potential of using mid-infrared spectroscopy, especially the aromatic and silicate features produced by dust grains to directly probe distant L ~ 10¹³ L ULIRGs at z ~ 2. Spitzer / IRS observations provide a unique and direct access to high-z ULIRG physical properties. It will definitively open the route toward a complete census of the distant infrared-luminous Universe. A first study on two distant SMGs using Spitzer / IRS by Lutz et al. (2005) finds for one SMG an equal contribution from star formation and AGN. The second galaxy is dominated by star formation.