2014MNRAS.439.2146C

We employ cosmological hydrodynamical simulations to investigate models in which the supermassive black holes powering luminous z ∼ 6 quasars (QSOs) grow from massive seeds. We simulate 18 regions with densities ranging from the mean cosmic density to the highest σ peaks in the Millennium simulation volume. Only in the most massive haloes situated in the most overdense regions, can black holes grow to masses up to ~109M_☉ by z ∼ 6 without invoking super-Eddington accretion. Accretion on to the most massive black holes becomes limited by thermal active galactic nucleus (AGN) feedback by z ∼ 9-8 with further growth proceeding in short Eddington-limited bursts. Our modelling suggests that current flux-limited surveys of QSOs at high redshift preferentially detect objects at their peak luminosity and therefore miss a substantial population of QSOs powered by similarly massive black holes but with low accretion rates. To test whether the required host halo masses are consistent with the observed galaxy environments of z ∼ 6 QSOs, we produce realistic rest-frame UV images of our simulated galaxies. Without strong stellar feedback, our simulations predict numbers of bright galaxies larger than observed by a factor of 10 or more. Supernova-driven galactic winds reduce the predicted numbers to a level consistent with observations indicating that stellar feedback was already very efficient at high redshifts. We further investigate the effect of thermal AGN feedback on the surrounding gas. AGN outflows are highly anisotropic and mostly energy driven, pushing gas at ≳ 1000 km/s out to tens of kpc consistently with observations. The spatially extended thermal X-ray emission around bright QSOs is powered by these outflows and is an important diagnostic of the mechanism whereby AGN feedback energy couples to surrounding gas.