Low-luminosity AGN and X-ray binary populations in COSMOS star-forming galaxies

TL;DR

This study uses X-ray stacking of around 75,000 star-forming galaxies from the COSMOS survey to analyze low-luminosity AGN and X-ray binary populations, revealing obscured AGN at high redshift and their relation to galaxy properties.

Contribution

It provides new insights into the evolution of X-ray binaries and obscured AGN in star-forming galaxies across a wide redshift range, using deep stacking analysis.

Findings

X-ray binaries' luminosity evolution depends on obscuration assumptions.

Most high-redshift stacks show significant nuclear X-ray excess after XRB subtraction.

Obscured AGN are prevalent at high redshift, correlating with stellar mass.

Abstract

We present an X-ray stacking analysis of $\sim$75,000 star-forming galaxies between $0.1<z<5.0$ using the Chandra COSMOS Legacy survey to study the X-ray emission of low-luminosity active galactic nuclei (AGN) and its connection to host galaxy properties. The stacks at $z<0.9$ have luminosity limits as low as $10^{40}-10^{41}$ erg s$^{-1}$, a regime in which X-ray binaries (XRBs) can dominate the X-ray emission. Comparing the measured luminosities to established XRB scaling relations, we find that the redshift evolution of the luminosity per star formation rate (SFR) of XRBs depends sensitively on the assumed obscuration and may be weaker than previously found. The XRB scaling relation based on stacks from the Chandra Deep Field South overestimates the XRB contribution to the COSMOS high specific SFR (sSFR) stacks, possibly due to a bias affecting the CDF-S stacks because of their small galaxy samples. After subtracting the estimated XRB contribution from the stacks, we find that most stacks at $z>1.3$ exhibit a significant X-ray excess indicating nuclear emission. The AGN emission is strongly correlated with stellar mass but does not exhibit an additional correlation with SFR. The hardness ratios of the high-redshift stacks indicate that the AGN are substantially obscured ($N_{\mathrm{H}}\sim10^{23}$ cm$^{-2}$). These obscured AGN are not identified by IRAC color selection and have $L_X\sim10^{41}-10^{43}$ erg s$^{-1}$, consistent with accretion at an Eddington rate of $\sim10^{-3}$ onto $10^7-10^8 M_{\odot}$ black holes. Combining our results with other X-ray studies suggests that AGN obscuration depends on stellar mass and an additional variable, possibly the Eddington rate.