Another piece to the puzzle: radio detection of a JWST detected AGN candidate

Anniek J. Gloudemans, Kenneth J. Duncan, Anna-Christina Eilers, Emanuele Paolo Farina, Yuichi Harikane, Kohei Inayoshi, Erini Lambrides, Eleni Vardoulaki

First listed 2025-01-09 | Last updated 2025-06-23

Abstract

Radio observations can provide crucial insight into the nature of a new abundant and mysterious population of dust-reddened active galactic nuclei (AGN) candidates discovered by the James Webb Space Telescope (JWST), including Little Red Dots (LRDs). In this study, we search for radio bright sources in a large sample of $\sim$700 JWST discovered AGN candidates ($z\sim2-11$) in the 0.144-3 GHz frequency range, utilizing deep radio imaging in COSMOS, GOODS-N, and GOODS-S. Only one source, PRIMER-COS 3866 at $z=4.66$, is significantly detected in our radio surveys, which has been previously identified as an X-ray AGN. Its radio properties are consistent with both an AGN and star formation origin with a spectral index of $α=-0.76^{+0.11}_{-0.09}$, radio-loudness of $R\approx0.5$, and brightness temperature limit of $T_b \gtrsim 10^{3}$ K. Our stacking results of both spectroscopically and photometrically selected AGN candidates yield non-detections in all fields, with 3$σ$ limits of $L_{1.4\text{GHz}} < 8.6\times10^{39}$ erg s$^{-1}$ (spectroscopic sample) and $L_{1.3\text{GHz}} < 1.3\times10^{39}$ erg s$^{-1}$ (photometric sample). We demonstrate that these results are still consistent with expectations from the empirical $L_X - L_{\text{H}α}$ and $L_X - L_R$ correlations established for local AGN. We argue that current radio observations in these studied fields have insufficient depth to claim JWST discovered AGN candidates are radio-weak. We project that future surveys carried out by the SKA and ngVLA should be able to obtain significant detections within a few hours, providing crucial measurements of their brightness temperature, which would allow for distinguishing between AGN and starburst-driven origins of this new abundant population.

Short digest

Deep 0.144–3 GHz maps across COSMOS, GOODS-N, and GOODS-S are mined for radio counterparts to ~700 JWST-selected AGN candidates (including LRDs). Only PRIMER-COS 3866 at z=4.66 is individually detected; its spectrum (α=−0.76+0.11−0.09), modest radio-loudness (R≈0.5), and Tb≳10^3 K remain consistent with either AGN or star-formation power. Stacks of spectroscopic and photometric subsamples yield non-detections with 3σ limits L1.4GHz<8.6×10^39 erg s−1 and L1.3GHz<1.3×10^39 erg s−1, still compatible with local L_X–L_Hα and L_X–L_R relations. The authors argue present radio depths are insufficient to label the population radio-weak and forecast SKA/ngVLA detections within hours to enable brightness-temperature tests that separate AGN from starbursts.

Key figures to inspect

• Figure 1: Examine the F115W stamp with VLA contours and the multi-band (144 MHz/1.3/3 GHz) cutouts of PRIMER-COS 3866 to verify the positional match and compactness; the radio SED fit (with GMRT upper limits) shows α≈−0.76 and hints of curvature possibly from SSA/FFA.
• Figure 2: The COSMOS/GOODS stacked images all show non-detections; read the panel rms values to translate into average luminosity limits and appreciate how stacking pushes below single-image depths yet still finds no mean signal.
• Figure 3: Black-hole mass vs Hα luminosity with predicted radio output: locate the median of the GOODS sources and the locus implied by PRIMER-COS 3866; compare the overplotted 3σ radio limits (GN single, GN stack, COSMOS stack) to see that the null results are not in tension with expectations.
• Figure 4: Examples of false radio associations illustrate how nearby bright galaxies can mimic detections; use the contours, beam sizes, and offsets to understand the team’s visual-vetting criteria and why many SNR “hits” were rejected.