Feedback from central massive black holes in galaxies using cosmological simulations

Abstract Gas accretion onto central supermassive black holes of active galaxies and resulting energy feedback, is an important component of galaxy evolution, whose details are still unknown especially at early cosmic epochs. We investigate BH growth and feedback in quasar-host galaxies at z ⩾ 6 by performing cosmological hydrodynamical simulations. We simulate the 2R 200 region around a 2 × 1012 Mʘ halo at z = 6, inside a (500 Mpc)3 comoving volume, using the zoom-in technique. We find that BHs accrete gas at the Eddington rate over z = 9–6. At z = 6, our most-massive BH has grown to M BH = 4 × 109 Mʘ. Star-formation is quenched over z = 8–6.


Introduction
Active galactic nuclei (AGN) emit enormous amounts of energy powered by the accretion of gas onto their central supermassive black holes (SMBHs) (Rees 1984). SMBHs of mass 10 9 M are observed to be in place in luminous quasars by z ∼ 6, when the Universe was less than 1 Gyr old (Wu et al. 2015). It is difficult to understand how these early SMBHs formed over such short timescales (Matsumoto et al. 2015).
AGN feedback should operate mostly in the negative form quenching star formation (Schawinski et al. 2006). In the host galaxy of the quasar SDSS J1148+5251 at z = 6.4, Maiolino et al. (2012) detected broad wings of the [CII] line tracing a massive outflow with velocities up to ±1300 km/s. The physical mechanisms by which quasar outflows affect their host galaxies remain as open questions.
We performed zoomed-in cosmological hydrodynamical simulations of quasar-host galaxies at z 6 and study their outflows. Details of this work is in Barai et al. (2018).

Numerical method
The initial conditions are generated using the MUSIC † software (Hahn & Abel 2011). We use the code GADGET-3 (Springel 2005) to perform our zoom-in simulations. First, a dark-matter (DM) only low-resolution simulation is carried out of a (500 Mpc) 3 comoving volume, using 256 3 DM particles, from z = 100 up to z = 6. We select the mostmassive halo at z = 6 to zoom-in, which has a total mass M halo = 4.4 × 10 12 M , and a virial radius R 200 511 kpc comoving. We execute a series of four simulations, all incorporating metal cooling, chemical enrichment, SF and SN feedback. The first run has no AGN included, while the latter three explore different AGN feedback models. † MUSIC -Multi-scale Initial Conditions for Cosmological Simulations: https:// bitbucket.org/ohahn/music 36 P. Barai

Black hole growth
The redshift evolution of the most-massive SMBH mass in the three AGN runs is plotted in Fig. 1 -left panel. Each BH starts as a seed of M BH = 10 5 M , at z ∼ 14 in the runs AGNcone and AGNsphere (z ∼ 10 in AGNoffset). The dominant mode of BH growth occurs over z = 9−6 in runs AGNcone and AGNsphere, corresponding to Eddingtonlimited gas accretion where Eddington ratio = 1. The BH has grown to M BH = 4 × 10 9 M at z = 6 in run AGNcone (red curve). The BH grows 10 times more massive at z = 6 in the AGNcone case than in the AGNsphere run. This is because more gas can inflow along the perpendicular direction to the bi-cone, and accrete onto the BH.

Star formation
The star formation rate (total in the simulation box) versus redshift of the four simulations is displayed in Fig. 1 -right panel. The SFR rises with time in all the runs initially, and continues to increase in the noAGN case without a BH. The SFR in run AGNoffset is almost similar to that in the run noAGN, because the BHs are too small there to generate enough feedback. The models suppress SF substantially from z ∼ 8 onwards, when the BHs have grown massive and generate larger feedback energy. Thus, we find that BHs need to grow to M BH > 10 7 M in order to suppress star-formation. BH feedback causes a reduction of SFR up to 4 times at z = 6: from 800M /yr in the noAGN run, to 200M /yr in run AGNcone, and 350M /yr in run AGNsphere.