Spitzer's Contribution to the AGN Population

Abstract

Using large multiwavelength datasets, we study obscured AGN in the distant universe that have been missed via traditional selection techniques (e.g. UV/ optical/X-ray). To do so, we take particular advantage of the mid-IR, which is minimally affected by obscuration. We first select as AGN candidates those objects whose radio emission is significantly brighter, relative to the mid-IR, than would be predicted by the well known radio/infrared correlation, indicating that the radio emission originates in the central engine. We find that of the 27 such sources identified in the CDF-N, 60% lack solid X-ray detections and 25% lack even 2σ X-ray emission. The absorbing columns of the faint X-ray–detected objects indicate that they are obscured but unlikely to be Compton thick, whereas the radio-excess AGN which are X-ray non-detected are Compton-thick candidates. We similarly use the infrared emission to select IRAC (3.6-8.0 μm) power-law AGN. In these luminous AGN, the hot dust emission from the AGN fills in the gap in a galaxy’s SED between the 1.6 μm stellar bump and the long-wavelength dust emission feature. While sources selected in this way are more luminous than the radio-excess AGN, we find a similar X-ray detection fraction. Of the 62 power-law galaxies in the CDF-N, only 55% are detected in the X-ray, and 15% lack evidence for even weak 2σ X-ray emission. A study of their X-ray properties indicates that ∼ 75% are obscured. Finally, we test IRAC color-color and infrared-excess selection criteria. We find that while these selection techniques identify a number of obscured AGN, they may also select a significant number of star-forming galaxies. By combining only the secure AGN candidates selected via all methods discussed above, we estimate that the addition of Spitzer-selected AGN candidates to the deepest Xray selected AGN samples directly increases the number of known AGN by 54- 77%, and implies a total increase to the number of AGN of 71-94%.