2018ApJ...866...63A

We present an analysis using the MOSFIRE Deep Evolution Field (MOSDEF) survey on the nature of "MIR-excess" galaxies, which have star formation rates (SFRs) inferred from mid-infrared (MIR) data that are substantially elevated relative to those estimated from dust-corrected UV data. We use a sample of ∼200 galaxies and active galactic nuclei (AGNs) at 1.40 < z < 2.61 with 24 µm detections (rest-frame 8 µm) from MIPS/Spitzer. We find that the identification of MIR-excess galaxies strongly depends on the methodologies used to estimate IR luminosity (L_IR) and to correct the UV light for dust attenuation. We find that extrapolations of the SFR from the observed 24 µm flux, using luminosity-dependent templates based on local galaxies, substantially overestimate L_IR in z ∼ 2 galaxies. By including Herschel observations and using a stellar-mass-dependent, luminosity-independent L_IR, we obtain more reliable estimates of the SFR and a lower fraction of MIR-excess galaxies. Once stellar-mass selection biases are taken into account, we identify ∼24% of our galaxies as MIR excess. However, SFR_Hα is not elevated in MIR-excess galaxies compared to MIR-normal galaxies, indicating that the intrinsic fraction of MIR excess may be lower. Using X-ray-, IR-, and optically selected AGNs in MOSDEF, we do not find a higher prevalence for AGNs in MIR-excess galaxies relative to MIR-normal galaxies. A stacking analysis of X-ray-undetected galaxies does not reveal a harder spectrum in MIR-excess galaxies relative to MIR-normal galaxies. Our analysis indicates that AGN activity does not contribute substantially to the MIR excess and instead implies that it is likely due to the enhanced polycyclic aromatic hydrocarbon emission.