The radio properties of the JWST-discovered AGN

TL;DR

This study probes the radio properties of 37 JWST-selected BLAGN in GOODS-N using multiple frequencies, finding no individual detections and a deep 3σ rest-frame 5 GHz upper limit of ~2×10^39 erg s^-1. By predicting intrinsic X-ray luminosities from Hα and applying Lx–Lrad and Fundamental Plane relations, the authors find most sources are consistent with radio-quiet AGN within the scatter, though several show significant radio underluminosity. They explore contributions from star formation, free-free absorption in dense BLR gas, and possible suppression of the magnetic field or X-ray corona—potentially linked to super-Eddington accretion—to explain the X-ray weakness and lack of radio detections. The results highlight the need for an order of magnitude deeper, higher-resolution radio surveys to constrain this population, with SKA-era observations expected to play a crucial role in elucidating the radio properties of JWST-selected high-z AGN.

Abstract

We explore the radio emission of JWST-selected Broad Line AGN (BLAGN, or type 1) in the GOODS-N field. We use deep radio data at different frequencies (144\,MHz, 1.5\,GHz, 3\,GHz, 5.5\,GHz, 10\,GHz), and we find that none of the {37} sources investigated is detected at any of the aforementioned frequencies. Similarly, the radio stacking analysis does not reveal any detection down to an rms of ${\sim 0.15}μ$Jy beam$^{-1}$, corresponding to a $3σ$ upper limit at rest frame 5 GHz of $L_{5GHz}=2\times10^{39}$ erg s$^{-1}$ at the mean redshift of the sample $z\sim 5.1$. We compared this and individual sources upper limits with expected radio luminosities estimated assuming different AGN scaling relations, {to check whether these are consistent with the standard BLAGN spectral energy distribution}. For most of the sources the radio luminosity upper limits are still compatible with expectations for radio-quiet (RQ) AGN; nevertheless, the more stringent stacking upper limits and the fact that no detection is found {might suggest} that JWST-selected BLAGN are weaker than standard AGN even at radio frequencies. Indeed, the probability of having none of the BLAGN detected in none of the investigated radio images is expected to be on average very low ($P<10^{-4}$). We discuss some scenarios that could explain the possible radio weakness, such as free-free absorption from a dense medium, or the lack of either magnetic field or a corona, possibly as a consequence of super-Eddington accretion. These scenarios would also explain the observed X-ray weakness. We also conclude that $\sim$1 dex more sensitive radio observations are needed to better constrain the level of radio emission (or lack thereof) for the bulk of these sources. The Square Kilometer Array Observatory (SKAO) will likely play a crucial role in assessing the properties of this AGN population.

Figures (9)

• Figure 1: $10"\times 10"$ cutouts of the mean (right) and median (left) radio stacking of the 37 JWST selected BLAGN on the GOODS-N field performed on the images presented in Sect. \ref{['sec:images']}. The values of the rms measured in the stacked images are reported in the panels.
• Figure 2: Distribution of the radio $3\sigma$ upper limits obtained at the frequencies of the images described in Sect. \ref{['sec:images']}. With empty red triangles and filled red triangles we show the sensitivities of the radio images and of the stack, respectively. The green empty circles refer to the sensitivities of the SKAO UD radio survey at $\sim 1$ GHz and $\sim 10$ GHz. The blue empty circle refers to the LOFAR deep radio image of the GOODS-N field. Filled blue and green circles refer to the sensitivities that would be reached considering the stack of the 37 BLAGN investigated in this work on the deep LOFAR and UD SKAO images, to be compared with the stack achieved in the currently available images. The black triangle refers to the radio stack of the 8 high-z BLAGN performed in Mazzolari24c on the AEGIS20 1.4GHz image Ivison07. Instead, the gray and empty tringles refer to the radio stacks of large samples of photometrically selected LRD performed in Perger24 and Akins24.
• Figure 3: Observed rest-frame 5GHz radio luminosity versus expected rest-frame 5GHz radio luminosity of different samples of high-z and local BLAGN according to the four RQ AGN relations described in Sect. \ref{['sec:Lradexp']}. The black data points represent the JWST-detected BLAGN analyzed in this work, the red diamond their stack, the gold star the BLAGN GN-28074 reported in Juodzbalis24_rosetta. The green symbols indicate sources from different local analog samples. The green square represents the position of SBS 0355-052 Hatano24Johnson09, the thick crosses represent the metal-poor dwarf BLAGN reported in Burke21, the thin crosses the X-ray weak BLAGN reported in Paul24, the filled triangles the radio detected and RQ NLS1 reported in Berton18. The light blue data point in the top-left panel represents the stack of the X-ray selected but radio undetected AGN on the GOODS-N field performed by Radcliffe21_rstack. The red diamond and the black data points in the bottom-right panel, are computed considering $\log R=1$, while the red circle indicate the expected radio luminosity assuming $\log R=0$. The gray shaded areas represent the scatter of the relations, while the black dashed line is the 1:1 relation.
• Figure 4: Distribution of the total 5 GHz radio luminosities ($\rm L_{5GHz,AGN} + L_{5GHz,SFR_{MS}}$) of the JWST selected BLAGN investigated in this work normalized by the 3$\sigma$ 5 GHz luminosity upper limit obtained for each of these sources. For each source we show the fractional contribution to the total radio luminosity of the AGN and SF, as described in Sect. \ref{['sec:sfr']}. Sources crossing the black dashed line ($\rm L_{5GHz, tot}= L_{5GHz, obs}$) are those in tension with the radio undetection.
• Figure 5: Ratio between the SFR derived from the radio luminosity upper limits and the SFR derived assuming the whole H$\alpha$ emission (narrow+broad) to be due to SF. The red data point corresponds to the value of the stack. The black dashed line traces the threshold below which the $SFR_{H\alpha}$ becomes incompatible with the radio undetection.