Identifying the subtle signatures of feedback from distant AGN using ALMA observations and the EAGLE hydrodynamical simulations

TL;DR

This study combines ALMA observations and hydrodynamical simulations to investigate the subtle signatures of AGN feedback on star formation, finding that AGN feedback broadens the distribution of specific star formation rates across galaxies.

Contribution

It provides the first detailed comparison of observed star formation rate distributions with EAGLE simulations, highlighting the impact of AGN feedback on galaxy evolution.

Findings

Good agreement between observations and EAGLE with AGN feedback for sSFR distribution widths

EAGLE without AGN feedback predicts narrower sSFR distributions

AGN feedback results in broad sSFR distributions across galaxies with stellar mass > 10^10 M_sun

Abstract

We present sensitive 870$\mu$m continuum measurements from our ALMA programmes of 114 X-ray selected AGN in the CDF-S and COSMOS fields. We use these observations in combination with data from Spitzer and Herschel to construct a sample of 86 X-ray selected AGN, 63 with ALMA constraints at $z=1.5-3.2$ with stellar mass $>2\times10^{10}M_{\odot}$. We constructed broad-band spectral energy distributions in the infrared band (8 - 1000$\mu$m) and constrain star-formation rates (SFRs) uncontaminated by the AGN. Using a hierarchical Bayesian method that takes into account the information from upper limits, we fit SFR and specific SFR (sSFR) distributions. We explore these distributions as a function of both X-ray luminosity and stellar mass. We compare our measurements to two versions of the EAGLE hydrodynamical simulations: the reference model with AGN feedback and the model without AGN. We find good agreement between the observations and that predicted by the EAGLE reference model for the modes and widths of the sSFR distributions as a function of both X-ray luminosity and stellar mass; however, we found that the EAGLE model without AGN feedback predicts a significantly narrower width when compared to the data. Overall, from the combination of the observations with the model predictions, we conclude that (1) even with AGN feedback, we expect no strong relationship between the sSFR distribution parameters and instantaneous AGN luminosity and (2) a signature of AGN feedback is a broad distribution of sSFRs for all galaxies (not just those hosting an AGN) with stellar masses above $\approx 10^{10}$M$_{\odot}$.