Modeling Mid-infrared Diagnostics of Obscured Quasars and Starbursts

We analyze the link between active galactic nuclei (AGNs) and mid-infrared flux using dust radiative transfer calculations of starbursts realized in hydrodynamical simulations. Focusing on the effects of galaxy dust, we evaluate diagnostics commonly used to disentangle AGN and star formation in ultraluminous infrared galaxies (ULIRGs). We examine these quantities as a function of time, viewing angle, dust model, AGN spectrum, and AGN strength in merger simulations representing two possible extremes of the ULIRG population: one is a typical gas-rich merger at z ~ 0, and the other is characteristic of extremely obscured starbursts at z ~ 2-4. This highly obscured burst begins star-formation-dominated with significant polycyclic aromatic hydrocarbon (PAH) emission, and ends with a ~10^9 yr period of red near-IR colors. At coalescence, when the AGN is most luminous, dust obscures the near-infrared AGN signature, reduces the relative emission from PAHs, and enhances the 9.7 μm absorption by silicate grains. Although generally consistent with previous interpretations, our results imply none of these indicators can unambiguously estimate the AGN luminosity fraction in all cases. Motivated by the simulations, we show that a combination of the extinction feature at 9.7 μm, the PAH strength, and a near-infrared slope can simultaneously constrain the AGN fraction and dust grain distribution for a wide range of obscuration. We find that this indicator, accessible to the James Webb Space Telescope, may estimate the AGN power as tightly as the hard X-ray flux alone, thereby providing a valuable future cross-check and constraint for large samples of distant ULIRGs.