Investigating the influence of radio-faint AGN activity on the infrared-radio correlation of massive galaxies

TL;DR

This study tests whether residual radio emission from radio-faint AGN could bias the infrared-radio correlation (IRRC) in massive star-forming galaxies. By selecting 500 COSMOS galaxies around the IRRC and conducting deep VLBA 1.4 GHz observations (Tb ≳ 10^5 K), they identify 4 compact AGN cores (≈9% detection rate) and quantify the AGN contribution to the total radio luminosity. After removing the VLBA-detected AGN flux, the population-wide $q_{IR}$ distribution remains unchanged in both median and scatter, indicating that non-radio-excess AGN contamination is unlikely to drive the observed mass dependence of the IRRC. Source-count analyses show the detected AGN align with extrapolated AGN number counts, and even extreme extrapolations yield a maximal $q_{IR}$ shift of <0.1 dex, far smaller than the 0.22 dex mass–dependence observed, pointing to alternative physical mechanisms such as cosmic-ray electron energy losses or Type Ia supernovae contributions as more plausible explanations. The results establish the IRRC as robust against non-radio-excess AGN contamination and suggest focusing on CR transport and ISM conditions to understand the mass dependence of the IRRC at cosmic noon.

Abstract

It is well-known that star-forming galaxies (SFGs) exhibit a tight correlation between their radio and infrared emissions, commonly referred to as the infrared-radio correlation (IRRC). Recent empirical studies have reported a dependence of the IRRC on the galaxy stellar mass, in which more massive galaxies tend to show lower infrared-to-radio ratios (qIR) with respect to less massive galaxies. One possible, yet unexplored, explanation is a residual contamination of the radio emission from active galactic nuclei (AGN), not captured through "radio-excess" diagnostics. To investigate this hypothesis, we aim to statistically quantify the contribution of AGN emission to the radio luminosities of SFGs located within the scatter of the IRRC. Our VLBA program "AGN-sCAN" has targeted 500 galaxies that follow the qIR distribution of the IRRC, i.e., with no prior evidence for radio-excess AGN emission based on low-resolution (~ arcsec) VLA radio imaging. Our VLBA 1.4 GHz observations reach a 5-sigma sensitivity limit of 25 microJy/beam, corresponding to a radio brightness temperature of Tb ~ 10^5 K. This classification serves as a robust AGN diagnostic, regardless of the host galaxy's star formation rate. We detect four VLBA sources in the deepest regions, which are also the faintest VLBI-detected AGN in SFGs to date. The effective AGN detection rate is 9%, when considering a control sample matched in mass and sensitivity, which is in good agreement with the extrapolation of previous radio AGN number counts. Despite the non-negligible AGN flux contamination (~ 30%) in our individual VLBA detections, we find that the peak of the qIR distribution is completely unaffected by this correction. We conclude that residual AGN contamination from non-radio-excess AGN is unlikely to be the primary driver of the M* - dependent IRRC.

Figures (6)

• Figure 1: RGB images of the targets VLBA ID '338' (top left), '351' (top right), '408' (bottom left), and '498' (bottom right). These are obtained by combining the F150W, F277W, and F444W JWST/NIRCam filters franco+2025 covering a 5$^{{\prime}{\prime}}$$\times$5$^{{\prime}{\prime}}$ field of view. The NIRCam pixel scale is 30 mas. The white contours are the VLA 3 GHz levels smolcic+2017 at 3, 4, 5, and 6 times the rms (2.3 $\mu$Jy/beam, the VLA beamsize is 0.75$^{{\prime}{\prime}}$ as shown by the white circle in the bottom left). In the zoom-in red panels, the purple contours are the levels at 2, 3, 4, 5, and 6 times the image rms detected by the VLBA, which come from a $0.2^{\prime \prime} \times 0.2^{\prime \prime}$ field of view, also outlined by the red square at the centers of the galaxies. The dotted purple contour also shows the -3 rms level. The VLBA beam size for each detection is: 26$\times$5 mas$^2$ ('338'), 26$\times$5 mas$^2$ ('351'), 28$\times$5 mas$^2$ ('408'), and 23$\times$6 mas$^2$ ('498'), with a similar position angle PA$\sim$162$^{\circ}$.
• Figure 2: $q_{IR}$ distribution as a function of host galaxy stellar mass for all the 500 targets (grey points), the 4 VLBA-detected AGN (blue stars) and the 277 non-detections part of the $M_\star$-matched sample (light blue points). The $q_{IR}$ of the detections account for the AGN emission detected by the VLBA (e.g., $q_{{\rm IR,corr}}$). The darker blue points are the 42 non-detections observed at sensitivities equal to or lower than the sensitivity of the detections, thus constituting the (M*, rms)- matched CS. The black line is the $q_{IR}$ versus $\log M_\star / M_\odot$ relation from delvecchio+2021 at $<z> \ \sim 1$, and the shaded areas spans the scatter of the IRRC at $M_\star > 10^{10.5}M_\odot$ from delvecchio+2021, of around $\pm$ 0.22 dex and $\pm$ 0.44 dex.
• Figure 3: Normalized distributions of the difference between the observed $q_{IR}$ and the one expected from Eq. \ref{['eq:qIRvsmass']}delvecchio+2021 based on the stellar mass and the redshift of the host galaxy, $q_{IR}$ (M$_\star$,z). The light purple histogram represents the distribution of the (M$_\star$,z)-CS. The stacked dark purple histogram (i.e. dark purple histogram on top of the light purple one) represents the detections, and the hatched orange histogram is the so-called mirror distribution, interpreted as the intrinsic $q_{IR}$ distribution of SFGs. The bootstrapped median values of the mirror distribution were fitted with a Gaussian (red curve; see Sec. \ref{['sec:results']} for further details). Left panel: $q_{IR}$ is computed using the VLA luminosity coming from both star formation and AGN activity. Right panel: $q_{IR,corr}$ is the ratio measured after subtracting the AGN radio luminosity detected by the VLBA from the total radio VLA emission.
• Figure 4: Euclidean normalized radio source number counts obtained by fitting, with the Markov chain Monte Carlo algorithm, the total radio Luminosity Function (LF) from novak+2018 using different evolving analytical LFs for AGN hosts (red line) and SF galaxies (blue line). The shaded areas encompass the 3$\sigma$ errors from the $\chi^2$ fits performed on individual populations novak+2018. Grey circles are the number counts from galluzzi+2025 obtained from a compilation of radio observations with different instruments (e.g. VLA 1-3 GHz, ATCA, GMRT, LOFAR, MeerKAT etc.), without any distinction between AGN or SF. Red and black circles show the number counts from the VLBA CANDELS GOODS-North Survey deane+2024 and VLBA 1.4 GHz COSMOS herrera-ruiz+2018 in case of radio emission from AGN activity. The red star is the number count of the detections from the AGN-sCAN survey presented in this work. The grey arrow point towards the regime below our sensitivity limit, from which the number counts extrapolated from the red curve of novak+2018 of the 42 brightest AGN were extracted, in order to constrain upper limits on our results (see text in Sec. \ref{['sec:cumulative']} for further details).
• Figure 5: $q_{IR}$ as a function of redshift for all the 500 targets (grey points), the 4 VLBA-detected AGN (dark blue stars) and the 277 non-detections that constitute the $M_\star$-matched sample (red points). The green dots are the 42 non-detections which make the ($M_\star, rms$)-matched control sample. The black line is the NLS fit of all the targets described by the power law $(1 + z)^{-0.13}$.