AGN accretion disk vs. torus The torus is the donut of dust encircling the Active Galactic  nucleus.  The accretion disk is inside the torus. Is there a boundary between the two?  At what point does a torus become an accretion disk?  What are their differences?  
 A: 
The Figure above is Figure 1 of this paper. It shows the scales of various features of the unified AGN model in units of $\log_{10}(r/\text{pc})$. Remember that a parsec is about 3.26 light-years or 200,000 AU (or 5000 times Pluto's mean orbital distance). The $10^{-5}$ mark is roughly the Schwarzschild radius for a $10^8\text{M}_\odot$ black hole.
The accretion disk and torus are basically different in their vertical extents. That is, the torus is about as thick as its radial extent whereas the accretion disk is theoretically infinitesimally thin. Why the difference? I don't actually know but I imagine its because the gas in the accretion disk is dense enough to transport angular momentum efficiently. Where this starts happening is probably the point at which gas starts to fall inward and it then joins and gets smeared into the disk. The accretion disk is only a few hundred Schwarzschild radii or a few hundred AU. The torus is 1000 times further out so that's the distance gas has to cover.
Other than thickness, the torus and disk are different because the torus absorbs and scatters light whereas the disk shines and produces light. That's because the torus is "dusty" (cold and probably containing molecular gas) but the disk is basically a hot plasma at thousands (or tens of thousands) of Kelvin.
This isn't my area of expertise so I stand to be corrected, though.
A: If I understand the literature right, the disk "ends" when it reaches the self-gravity radius $R_{sg}$ where local gravity from the disk exceeds the vertical component of the black hole gravity and the disk becomes unstable. $$R_{sg}\approx 1680 \left[\frac{M_{BH}}{10^9 M_\odot}\right]^{-2/9} \alpha^{2/9} \left[\frac{L_{AGN}}{L_{Edd}}\right]^{4/9} \left[\frac{\epsilon}{0.1}\right]^{-4/9} R_S$$ where $\alpha$ is the viscosity parameter $\approx 0.01-0.1$. For a $10^9 M_\odot$ black hole with $L_{AGN}/L_{Edd}\approx 0.1$ $R_{sg}\approx 0.04$ pc. This appears to be in rough agreement with observations of quasars.
