Clustering, bias and the accretion mode of X-ray-selected AGN

Abstract

We present the spatial clustering properties of 1466 X-ray-selected active galactic nuclei (AGN) compiled from the Chandra Deep Field (CDF) North and South, extended CDF-S, COSMOS and All-wavelength Extended Groth strip International Survey (AEGIS) fields in the 0.5–8 keV band. The X-ray sources span the redshift interval 0 < z < 3 and have a median value of |$\bar{z}=0.976$|⁠. We employ the projected two-point correlation function to infer the spatial clustering and find a clustering length of r0 = 7.2 ± 0.6 h−1 Mpc and a slope of γ = 1.48 ± 0.12, which corresponds to a bias of |$b(\bar{z})=2.26 \pm 0.16$|⁠. Using two different halo bias models, we consistently estimate an average dark-matter host halo mass of Mh ≃ 1.3(± 0.3) × 1013 h−1 M⊙. The X-ray AGN bias and the corresponding dark-matter host halo mass are significantly higher than the corresponding values of optically selected AGN (at the same redshifts). The redshift evolution of the X-ray-selected AGN bias indicates, in agreement with other recent studies, that a unique dark-matter halo mass does not fit well the bias at all the different redshifts probed. Furthermore, we investigate if there is a dependence of the clustering strength on X-ray luminosity. To this end we consider only 650 sources around z ∼ 1 and we apply a procedure to disentangle the dependence of clustering on redshift. We find indications for a positive dependence of the clustering length on X-ray luminosity, in the sense that the more luminous sources have a larger clustering length and hence a higher dark-matter halo mass. In detail we find for an average luminosity difference of δ log10Lx ≃ 1 a halo mass difference of a factor of ∼3. These findings appear to be consistent with a galaxy formation model where the gas accreted on to the supermassive black hole in intermediate-luminosity AGN comes mostly from the hot-halo atmosphere around the host galaxy.