Balancing the energy budget between star formation and active galactic nuclei in high-redshift infrared luminous galaxies

We present deep Spitzer mid-infrared spectroscopy, along with 16, 24, 70, and 850 μm photometry, for 22 galaxies located in the Great Observatories Origins Deep Survey-North (GOODS-N) field. The sample spans a redshift range of 0.6 lsim z lsim 2.6, 24 μm flux densities between ~0.2 and 1.2 mJy, and consists of submillimeter galaxies (SMGs), X-ray or optically selected active galactic nuclei (AGNs), and optically faint (zAB > 25 mag) sources. We find that infrared (IR; 8-1000 μm) luminosities derived by fitting local spectral energy distributions (SEDs) with 24 μm photometry alone are well matched to those when additional mid-infrared spectroscopic and longer wavelength photometric data are used for galaxies having z lsim 1.4 and 24 μm-derived IR luminosities typically lsim3 × 1012 L sun. However, for galaxies in the redshift range between 1.4 lsim z lsim 2.6, typically having 24-μm-derived IR luminosities gsim3 × 1012 L sun, IR luminosities are overestimated by an average factor of ~5 when SED fitting with 24 μm photometry alone. This result arises partly due to the fact that high-redshift galaxies exhibit aromatic feature equivalent widths that are large compared to local galaxies of similar luminosities. Using improved estimates for the IR luminosities of these sources, we investigate whether their infrared emission is found to be in excess relative to that expected based on extinction-corrected UV star formation rates (SFRs), possibly suggesting the presence of an obscured AGN. Through a spectral decomposition of mid-infrared spectroscopic data, we are able to isolate the fraction of IR luminosity arising from an AGN as opposed to star formation activity. This fraction is only able to account for ~30% of the total IR luminosity among the entire sample and ~35% of the "excess" IR emission among these sources, on average, suggesting that AGNs are not the dominant cause of the inferred "mid-infrared excesses" in these systems. Of the sources identified as having mid-infrared excesses, half are accounted for by using proper bolometric corrections while half show the presence of obscured AGNs. This implies sky and space densities for Compton-thick AGNs of ~1600 deg-2 and ~1.3 × 10-4 Mpc-3, respectively. We also note that IR luminosities derived from SED fitting the mid-infrared and 70 μm broadband photometry agree within ~50% to those values estimated using the additional mid-infrared spectroscopic and submillimeter data. An inspection of the far-infrared (FIR)-radio correlation shows no evidence for evolution over this redshift range. However, we find that the SMGs have IR/radio ratios which are a factor of ~3 lower, on average, than what is measured for star-forming galaxies in the local universe