paper

Chaotic cold accretion onto black holes

arXiv:1301.3130 · doi:10.1093/mnras/stt692

Abstract

Using 3D AMR simulations, linking the 50 kpc to the sub-pc scales over the course of 40 Myr, we systematically relax the classic Bondi assumptions in a typical galaxy hosting a SMBH. In the realistic scenario, where the hot gas is cooling, while heated and stirred on large scales, the accretion rate is boosted up to two orders of magnitude compared with the Bondi prediction. The cause is the nonlinear growth of thermal instabilities, leading to the condensation of cold clouds and filaments when t_cool/t_ff < 10. Subsonic turbulence of just over 100 km/s (M > 0.2) induces the formation of thermal instabilities, even in the absence of heating, while in the transonic regime turbulent dissipation inhibits their growth (t_turb/t_cool < 1). When heating restores global thermodynamic balance, the formation of the multiphase medium is violent, and the mode of accretion is fully cold and chaotic. The recurrent collisions and tidal forces between clouds, filaments and the central clumpy torus promote angular momentum cancellation, hence boosting accretion. On sub-pc scales the clouds are channelled to the very centre via a funnel. A good approximation to the accretion rate is the cooling rate, which can be used as subgrid model, physically reproducing the boost factor of 100 required by cosmological simulations, while accounting for fluctuations. Chaotic cold accretion may be common in many systems, such as hot galactic halos, groups, and clusters, generating high-velocity clouds and strong variations of the AGN luminosity and jet orientation. In this mode, the black hole can quickly react to the state of the entire host galaxy, leading to efficient self-regulated AGN feedback and the symbiotic Magorrian relation. During phases of overheating, the hot mode becomes the single channel of accretion (with a different cuspy temperature profile), though strongly suppressed by turbulence.

Accepted by MNRAS: added comments and references. Your feedback is welcome

References in corpus (23)

Cited by in corpus (303)