X-ray-heated coronae and winds from accretion disks: Time-dependent two-dimensional hydrodynamics with adaptive mesh refinement

We perform for the first time time-dependent, two-dimensional, axisymmetric hydrodynamic simulations using local adaptive mesh refinement of thermally driven rotating winds from X-ray-irradiated accretion disks. The disk is assumed to flare in height with radius allowing direct exposure from the central X-ray source. The heating and cooling are treated strictly in the optically thin approximation. We adopt two spectra characteristic of active galactic nuclei (AGNs) which have Compton temperatures of T-IC approximate to 1.3 x 10(7) K and 10(8) K. We have computed a number of models which cover a large range in luminosity (0.002 less than or equal to L/L(Eddington) less than or equal to 1) and radius (less than or similar to 20 Compton radii). Our models enable us to extend and improve on the analytic predictions of Begelman, McKee, & Shields (BMS) for Compton-heated winds by including non-Compton processes such as photoionization heating and line cooling, typical of X-ray-heated winds. These non-Compton processes can be dominant at low temperatures (less than or similar to 10(7) K), thus being important in the wind regions of AGNs. Our results agree well with a number of predictions given by EMS, even when non-Compton processes dominate, suggesting that their analytic approximations of the hydrodynamics of disk winds are applicable to the more general area of X-ray-heated winds. In the regime in which Compton processes dominate (i.e., T-IC = 10(8) K spectrum), we have used our results to improve the analytic predictions of BMS, providing a new expression for the mass-loss rate and a modified view of the wind solution topology. We find that beginning from a basically static state, the time-dependent flow which develops eventually settles into a steady wind, without any evidence of hydrodynamic instabilities. The solution topology consists of a corona with an exponentially truncated wind at small radii, and a vigorous wind at large radii which can be impeded by gravity for small luminosities. We have constructed radius-luminosity parameter space plots of our numerical results in analogy to EMS for both the high and low T-IC, cases, depicting the range of solutions. The plots are strikingly similar to the analytic predictions, especially for high T-IC. We find the radial extent of the corona to be independent of luminosity, as predicted by BMS, extending out to about 0.25R(IC); this is a direct consequence of Compton heating. The transition from an isothermal to nonisothermal corona occurs at a luminosity within about a factor of 2 of the critical luminosity (L(cr) approximate to 0.03T(IC8)(-1/2)L(E)) predicted by BMS. The mass flux density in the corona shows an exponential rise, peaking at around 0.2R(IC) nearly independent of luminosity. The wind solutions can be characterized mainly by steadily heated, free winds (region B in BMS) and gravity-inhibited winds (region C in BMS). Nearly isothermal winds with temperatures of the order T-IC also exist (region A in BMS) but require higher luminosities than was first estimated by BMS. A necessary condition for the winds to approach isothermality is that the luminosity exceed the critical luminosity. The change in wind solutions from regions B and C is characterized by a nearly discontinuous change in the sonic point location from large heights in region C to small heights in region B. The mass-loss rate, however, appears continuous across this boundary. For a streamline leaving the disk surface at a radius R(0), the sonic point distance along the streamline, s(sonic), is such that s(sonic)/R(0) approximate to 0.6 in region B and s(sonic)/R(0) much greater than 1 in region C. An unexpected conclusion from our numerical results is that the area of a flow tube can actually be smaller at the sonic point than at the disk surface. This is because of the presence of an unbalanced radial pressure gradient of the flow at low heights upon being heated. Incorporating this effect into the simple analytical formulae for the mass-loss rate given by EMS yields results which are typically within about a factor of 2 (3) of our numerical results over a wide range of luminosities and radii for the high (low) Compton temperature models. We provide fitting formulae of our numerical results which give the mass flux density as a function of radius and luminosity. We also discuss briefly the implications of our results for the prediction of Fe K alpha lines which have recently been observed in AGNs.