Investigating the Far-IR/Radio Correlation of Star Forming Galaxies to z=3

TL;DR

This study examines the far-IR/radio correlation in star-forming galaxies up to redshift 3, revealing a gradual decrease in the flux density ratio with redshift and potential differences in galaxy spectral energy distributions over cosmic time.

Contribution

It provides the first detailed analysis of the far-IR/radio correlation at high redshift, using stacking techniques to probe below detection limits and compare galaxy types and redshifts.

Findings

q70 decreases with redshift for star-forming galaxies

Detected 70um counterparts for 22.5% of radio sources

High-redshift ULIRGs show different SEDs compared to local counterparts

Abstract

In order to examine the far-IR/radio correlation at high redshift we have studied the Spitzer 70um/far-infrared (far-IR) properties of sub-mJy radio sources from the 13^H XMM-Newton/Chandra Deep Field by redshift and galaxy type: active galactic nucleus (AGN) or star forming galaxy (SFG). We directly detect 70um counterparts (at >3sigma significance) for 22.5% (92/408) of the radio sources, while for the rest we perform stacking analysis by redshift and galaxy type. For the sources detected at 70um we find that the median and scatter of the observed flux density ratio, q70, are similar to previous results in the literature, but with a slight decrease in q70 towards higher redshifts. Of the radio sources detected at 70um 8/92 were already classified as AGN, but two of which maybe SFGs. For the stacked sources we obtain a detection for the SFGs at every redshift bin which implies they have mean flux densities a factor ~5 below the original 70um detection limit. For the stacked AGN we obtain a detection only in our highest redshift bin (1<~z<~5) where we may be sampling hot dust associated with the AGN at rest-frame 12-35um. The combined observed mean value of q70 for the SFGs (detected and non-detected at 70um) decreases gradually with redshift, consistent with tracks derived from empirical spectral energy distributions (SEDs) of local SFGs. Upon closer inspection and when comparing with tracks of appropriate luminosity, the values of q70 broadly agree at low redshift. However, at z~1, the observed q70 (for ULIRGs) is 2sigma below the value seen for local ULIRGs tracks, implying a difference in the SED between local and z~1 ULIRGs. At higher redshifts, the convergence of the tracks and the higher uncertainties in q70 do not allow us to determine if such a difference persists.