Radio and far-IR emission associated with a massive star-forming galaxy candidate atz≃ 6.8: a radio-loud AGN in the reionization era?

Abstract

We report the identification of radio (0.144–3 GHz) and mid-, far-infrared, and sub-mm (24–850μm) emission at the position of one of 41 UV-bright ($\mathrm{M_{\mathrm{UV}}}^{ }\lesssim -21.25$) z ≃ 6.6–6.9 Lyman-break galaxy candidates in the 1.5 deg2 COSMOS field. This source, COS-87259, exhibits a sharp flux discontinuity (factor >3) between two narrow/intermediate bands at 9450 and 9700 Å and is undetected in all nine bands blueward of 9600 Å, as expected from a Lyman alpha break at z ≃ 6.8. The full multiwavelength (X-ray through radio) data of COS-87529 can be self-consistently explained by a very massive (M* = 1010.8 M⊙) and extremely red (rest-UV slope β = −0.59) z ≃ 6.8 galaxy with hyperluminous infrared emission (LIR = 1013.6 L⊙) powered by both an intense burst of highly obscured star formation (SFR ≈ 1800 M⊙ yr−1) and an obscured ($\tau _{_{\mathrm{9.7\mu m}}} = 7.7\pm 2.5$) radio-loud (L1.4 GHz ≈ 1025.4 W Hz−1) active galactic nucleus (AGN). The radio emission is compact (1.04 ± 0.12 arcsec) and exhibits an ultra-steep spectrum between 1.32 and 3 GHz ($\alpha =-1.57^{+0.22}_{-0.21}$) that flattens at lower frequencies ($\alpha = -0.86^{+0.22}_{-0.16}$ between 0.144 and 1.32 GHz), consistent with known z > 4 radio galaxies. We also demonstrate that COS-87259 may reside in a significant (11×) galaxy overdensity, as common for systems hosting radio-loud AGN. While we find that low-redshift solutions to the optical + near-infrared data are not preferred, a spectroscopic redshift will ultimately be required to establish the true nature of COS-87259 beyond any doubt. If confirmed to lie at z ≃ 6.8, the properties of COS-87259 would be consistent with a picture wherein AGN and highly obscured star formation activity are fairly common among very massive (M* > 1010 M⊙) reionization-era galaxies.

Hybrid photometric redshifts for sources in the COSMOS and XMM-LSS fields

“…Our template fitting photo-z thus was consistent with the dropout method, and the ML was consistent with a lower-z solution. In agreement with Endsley et al (2022) we found low χ 2 values (χ 2 galaxy = 0.60 and χ 2 AGN = 3.9) indicat-ing good fits, and that the template fitting solution should be reliable 11 . However, for the ML ν/σ 2 = 0.013, indicating that it wasn't excessively extrapolating, so that the ML prediction ought also be reliable.…”

“…In this section we show how the combined Hierarchical Bayesian method could be used to check the validity of candidate high-z galaxies, where although there are very few high-z training spec-z, the higher accuracy of GPz at lower redshift may give more accurate weighting to low-z solutions. Endsley et al (2022) report the detection of a possible z ∼ 6.8 massive star-forming galaxy. The source, COS-87259, was identified and assigned a redshift using the a Lyman-break narrow-band dropout technique -and is among the COSMOS sources we have analysed.…”

“…and choice of prior. As Endsley et al (2022) note, true confirmation of the redshift of the source will require spectroscopic follow-up. However the authors do present compelling evidence beyond simply the photometry (e.g.…”

“…For context, taken at face value our z template values would indicate 6 percent of our sources are at z > 6, wheras our zHB values would indi-cate closer to 1 percent. Finally, we would also note Endsley et al (2022) identify radio continuum emission associated with COS-87259, which may further alter the appropriate redshift distribution prior.…”

“…Note however that we did not use any of the narrow bands used inEndsley et al (2022), so might expect to have a broader pdf than their results.© 2021 RAS, MNRAS 000, 1-19…”