Repository logo

Black hole growth and AGN feedback under clumpy accretion

Published version



Change log


De Graf, C 
Dekel, A 
Gabor, J 
Bournaud, F 


High-resolution simulations of supermassive black holes in isolated galaxies have suggested the importance of short (∼10 Myr) episodes of rapid accretion caused by interactions between the black hole and massive dense clouds within the host. Accretion of such clouds could potentially provide the dominant source for black hole growth in high-z galaxies, but it remains unresolved in cosmological simulations. Using a stochastic subgrid model calibrated by high-resolution isolated galaxy simulations, we investigate the impact that variability in black hole accretion rates has on black hole growth and the evolution of the host galaxy. We find this clumpy accretion to more efficiently fuel high-redshift black hole growth. This increased mass allows for more rapid accretion even in the absence of high-density clumps, compounding the effect and resulting in substantially faster overall black hole growth. This increased growth allows the black hole to efficiently evacuate gas from the central region of the galaxy, driving strong winds up to ∼2500 km s−1, producing outflows ∼10 × stronger than the smooth accretion case, suppressing the inflow of gas on to the host galaxy, and suppressing the star formation within the galaxy by as much as a factor of 2. This suggests that the proper incorporation of variability is a key factor in the co-evolution between black holes and their hosts.



black hole physics, methods: numerical, galaxies: active, galaxies: haloes, quasars: general

Journal Title

Monthly Notices of the Royal Astronomical Society

Conference Name

Journal ISSN


Volume Title



Oxford University Press
European Research Council (638707)
This work was supported by ISF grant 24/12, by GIF grant G-1052-104.7/2009, by a DIP grant, by the I-CORE Program of the PBC, by ISF grant 1829/12, by NSF grants AST-1010033 and AST-1405962, and supported from the E.C. through an ERC grant StG-257720.