Outflows driven by cosmic-ray pressure in broad absorption line QSOs

Abstract

We present a model that can explain many features of the very fast (~0.1c) outflows that are observed in broad absorption line QSOs (BALQSOs). In our model the wind is accelerated primarily by the pressure of ultrarelativistic protons. These protons are deposited in the wind, over a range of radii, by the decay of ultrarelativistic neutrons escaping from the central engine. Recent models for nonthermal processes in the central engines of active galactic nuclei (AGNs) predict a neutron luminosity which can exceed a few percent of the radiation flux, which is more than adequate to accelerate the wind. Since the energy deposition is predicted to peak at distances of order 1-10 pc, the main acceleration will occur outside the broad emission line region, as seems to be required by observations. The absorbing gas occupies a very small fraction of the volume of the wind, and must be confined by it. Keeping this gas sufficiently cool in the radiation field from the central source requires that the pressure in the absorbing clouds or filaments be higher than a certain value. Since the relativistic protons can pass right through the clouds, the confinement of the clouds is most likely to be due to thermal pressure in the wind. This is possible only if the energy of the ultrarelativistic protons can be converted to thermal energy with a high efficiency. We suggest that this might be accomplished through the excitation and subsequent damping of Alfven waves, or by a turbulent wind structure containing shocks that will drive the thermal and cosmic-ray energy densities to equipartition. We discuss and evaluate the plausibility of other confinement mechanisms as well. A simple model in which the absorption is caused by a fine spray of cloudlets, comoving with the wind and in pressure equilibrium with it, can reproduce a wide variety of BAL profiles similar to those that are observed. Line profiles prove to be more sensitive to ionization effects than to the details of the wind dynamics, thus making it difficult to derive the dynamical properties of the wind from the line profiles. We find that a self-consistent treatment of internal absorption in the wind can play an important role, both in reproducing the observed ionization levels and in explaining certain line features. Finally we discuss the formation, acceleration, and survival of the absorbing cloudlets. We argue that the absorbing gas is probably entrained into the wind near its base, and must survive the acceleration. The clouds are forced to comove with the wind by a combination of hydrodynamic drag and line radiation pressure, and are susceptible to dynamical instabilities, as well as thermal evaporation in the hot ambient medium. We point out difficulties with models in which clouds are continuously destroyed and re-form through thermal instability, and suggest several effects which may allow clouds to survive the various destructive influences. The small solid angle apparently subtended by absorbing gas in a BALQSO could reflect a nonspherical distribution of sources for the entrained matter (e.g., a disk or a flattened distribution of large clouds).