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

3.4.5. Warm ULIRGs and QSOs

"Warm" ULIRGs (S₂₅ / S₆₀ 0.2) probably represent transition objects between ULIRGs and QSOs (Sanders et al 1988b). The ISO spectrophotometry discussed above strongly supports the conclusion of the earlier IRAS work that warm ULIRGs indeed contain powerful AGNs. For instance, the warm ULIRG Mrk 231 (the most luminous ULIRG within z 0.1) is similar to the infrared-loud but optically selected QSOs I Zw1 and Mrk 1014. Most of the warm ULIRGs have Seyfert spectra (Veilleux et al 1997, 1999). High-resolution HST and ground-based observations show that warm ULIRGs typically have a compact hot dust source at the nucleus, surrounded by a luminous host galaxy with embedded compact star clusters and large-scale, tidal tails (Surace et al 1998, Surace & Sanders 1999, Lai et al 1998, Scoville et al 2000).

Although the AGN is probably dominating, circumnuclear star formation may nevertheless significantly contribute to the total luminosity of warm ULIRGs (Condon et al 1991b, Downes & Solomon 1998). In the case of Mrk 231 Taylor et al (1999) found a smooth, approximately circular radio continuum source of diameter ~ 1" (1 kpc), centered on the core-jet structure of the bright AGN. The extended radio source is associated with a massive rotating molecular disk seen at low inclination (Downes & Solomon 1998) and with a luminous (2L_* for no extinction correction) near-IR stellar disk (Lai et al 1998). If the far-IR-radio correlation for star-forming galaxies (Condon et al 1991a, Condon 1992) is applied to the parameters of Mrk 231's disk, the implied far-IR luminosity is 1.3 × 10¹² L, or ~ 70% of the 60 + 100 µm IRAS luminosity. If the red color of the disk's near-IR emission results from extinction (A_V ~ 12; Lai et al 1998), the corrected K-band luminosity of the disk is L_K(0) ⁹ ~ 2 × 10¹⁰ L. If the K-band emission comes from an old stellar population with M/L_K ~ 100, the mass of the disk within the central kpc has to be 2 × 10¹² M (an 8 M disk!). This mass is two orders of magnitude greater than the dynamical mass within the central kpc derived from the CO rotation curve (Downes & Solomon 1998). It is thus more likely that the K-band light comes from a population of (young) red supergiant or AGB stars. The required mass then is 20 to 100 times smaller, and the ratio of bolometric to K-band luminosity is 10² (Thatte et al 1997). The red disk may then be responsible for a significant fraction of the far-IR luminosity of Mrk 231, in accordance with the radio data. The relative weakness of the mid-infrared line emission observed by ISO-SWS would then require very high extinction or, perhaps more likely, an aged starburst. The situations for the IR-loud QSOs I Zw 1 and Mrk 1014 are similar, but even more extreme stellar masses would be required there to account for the large K-band stellar luminosities in both galaxies (~ 1.4 × 10¹¹ L) by an old stellar population. In both cases the dynamical masses implied by the molecular disk rotation excludes this possibility (Schinnerer et al 1998, Downes and Solomon 1998).

In summary, ISO has made substantial contributions to our understanding of ULIRGs. The resulting picture is fairly complex; there is no simple story to tell. ULIRGs do not seem to undergo an obvious metamorphosis from a starburst-powered system of colliding galaxies to a buried and then finally a naked QSO at the end of the merger phase. Average luminosity and molecular gas content do not change with merger phase. AGN activity and starburst activity often occur at the same time. On average, starburst activity does seem to dominate the bolometric luminosity in most objects. However, the most luminous ULIRGs appear to be AGN dominated. "Warm" ULIRGs are AGN dominated, and "cold" ULIRGs are star-formation dominated. The relative contributions of starburst and AGN activity are in part controlled by local effects. Warm ULIRGs/infrared-loud QSOs are advanced transition objects with powerful central AGNs. In a few well-studied cases, these AGNs are surrounded by luminous (aged?) starburst disks. Locally, (U)LIRGs are a spectacular curiosity. They contribute only ~ 1% to the local infrared radiation field, and ~ 0.3% to the total bolometric emissivity (Sanders & Mirabel 1996, Heckman 2000). We will show in the next chapter that this fraction increases dramatically (factor 10²) at higher redshift.

⁹ L_K is here defined as the luminosity emitted in the K-band (in units of solar luminosities) Back.