The Cosmos AGN Survey: The Nature of Faint, Weak, and Obscured AGN

The Cosmic Evolution Survey (COSMOS) provides a unique opportunity for the study of AGN with a combination of deep radio (VLA), infrared (Spitzer MIPS & IRAC), optical (Hubble/ACS & 21-band (Subaru/Suprime-Cam), UV (GALEX), and X-ray (XMM & Chandra) observations over 2 deg². I will present results from a 3-year spectroscopic survey of X-ray and IR selected AGN in COSMOS using the Magellan/IMACS and MMT/Hectospec instruments. This spectroscopic sample reaches the customary quasar/Seyfert luminosity boundary at z 2, and also reveals Type 1 AGN with black hole masses of only 10⁷ M_sun to z 2. The IR selection additionally allows us to observe a large number of heavily obscured AGN, which recent pencil-beam surveys have revealed to be a significant fraction of the total AGN population. The obscured fraction of AGN in COSMOS shows the well-known luminosity dependence, but also shows the most significant evidence to date for redshift evolution. These dependencies suggest a strong case for dusty star formation driving AGN obscuration, with luminous AGN luminosity providing negative feedback. COSMOS also allows for new constraints on fueling mechanisms of obscured and unobscured AGN. A large sample of Type 1 AGN allows us to suggest limits on accretion rate for the presence of a stable BLR. And the largest sample of X-ray bright / optical normal galaxies (XBONGs) to date, complete with bolometric SEDs, suggests that these targets are actually radiatively inefficient accretors with truncated accretion disks. I will try to place these ideas in the framework of the AGN unified model and discuss how they might be especially tested with future Herschel and ALMA observations.