arXiv2015
We describe a set of simulations of super-critical accretion onto a non-rotating supermassive BH. The accretion flow is radiation pressure dominated and takes the form of a geometrically thick disk with twin low-density funnels around the rotation axis. For accretion rates $\gtrsim 10 \dot M_{\rm Edd}$, there is sufficient gas in the funnel to make this region optically thick. Radiation from the disk first flows into the funnel, after which it accelerates the optically thick funnel gas along the axis. The resulting jet is baryon-loaded and has a terminal density-weighted velocity $\approx 0.3c$. Much of the radiative luminosity is converted into kinetic energy by the time the escaping gas becomes optically thin. For an observer viewing down the axis, the isotropic equivalent luminosity of total energy is as much as $10^{48}\,\rm erg\,s^{-1}$ for a $10^7 M_\odot$ BH accreting at $10^3$ Eddington. Therefore, energetically, the simulated jets are consistent with observations of the most powerful tidal disruption events, e.g., Swift J1644. The jet velocity is, however, too low to match the Lorentz factor $γ> 2$ inferred in J1644. There is no such conflict in the case of other tidal disruption events. Since favorably oriented observers see isotropic equivalent luminosities that are highly super-Eddington, the simulated models can explain observations of ultra-luminous X-ray sources, at least in terms of luminosity and energetics, without requiring intermediate mass black holes. Finally, since the simulated jets are baryon-loaded and have mildly relativistic velocities, they match well the jets observed in SS433. The latter are, however, more collimated than the simulated jets. This suggests that, even if magnetic fields are not important for acceleration, they may perhaps still play a role in confining the jet.