A Far-IR Characterization of 24 micron Selected Galaxies at 0<z<2.5 using Stacking at 70 microns and 160 microns in the COSMOS Field

TL;DR

This study uses stacking analysis at 70 and 160 microns to analyze the average properties of 24 micron selected galaxies across 0<z<2.5 in the COSMOS field, revealing insights into their infrared SEDs and luminosities.

Contribution

It introduces a stacking-based method to characterize the IR properties of 24 micron galaxies and compares empirical models to improve IR luminosity estimates.

Findings

Galaxies with higher 24 μm flux have warmer SEDs.

24 μm selected galaxies are consistent with normal star-forming and warm IR galaxies.

Previous IR luminosity estimates overpredict for bright sources at high redshift.

Abstract

We present a study of the average properties of luminous infrared galaxies detected directly at 24 $\mu$m in the COSMOS field using a median stacking analysis at 70$\mu$m and 160 $\mu$m. Over 35000 sources spanning 0<z<3 and 0.06 mJy<S_{24}<3.0 mJy are stacked, divided into bins of both photometric redshift and 24 $\mu$m flux. We find no correlation of $S_{70}/S_{24}$ flux density ratio with $S_{24}$, but find that galaxies with higher $S_{24}$ have a lower $S_{160}/S_{24}$ flux density ratio. These observed ratios suggest that 24 $\mu$m selected galaxies have warmer SEDs at higher mid-IR fluxes, and therefore have a possible higher fraction of AGN. Comparisons of the average $S_{70}/S_{24}$ and $S_{160}/S_{24}$ colors with various empirical templates and theoretical models show that the galaxies detected at 24 $\mu$m are consistent with "normal" star-forming galaxies and warm mid-IR galaxies such as Mrk 231, but inconsistent with heavily obscured galaxies such as Arp 220. We perform a $\chi^{2}$ analysis to determine best fit galactic model SEDs and total IR luminosities for each of our bins. We compare our results to previous methods of estimating $L_{\rm{IR}}$ and find that previous methods show considerable agreement over the full redshift range, except for the brightest $S_{24}$ sources, where previous methods overpredict the bolometric IR luminosity at high redshift, most likely due to their warmer dust SED. We present a table that can be used as a more accurate and robust method for estimating bolometric infrared luminosity from 24 $\mu$m flux densities.