Mass-loss contrails from giant stars

Mass loss from the giant stars in the central cluster of an active galaxy can provide an ongoing supply of fuel to the central active nucleus and high density material which may be accelerated outwards to produce the BAL features. Mass lost from late type stars is a mixture of both gas and dust and the dust is critical to the fate of this material. If the matter is injected into the interstellar space at a radius from the AGN such that the grain temperature exceeds the sublimation temperature, the grains will evaporate and the radiation pressure will be greatly reduced. For an AGN of luminosity 10(12-13) L., this radius occurs at approximately 1 pc. The material lost from stars inside this radius may then accrete inwards, providing a steady supply of fuel for the central accretion disk. Outside 1 pc, the grains may survive and the mass-loss material will feel strong radiative acceleration away from the AGN. This material, compressed into thin trails behind each giant star, may account for the broad absorption line systems seen in QSOs (Scoville & Norman 1995). Starting at the sublimation radius (1 pc), the mass-loss trail may ultimately be accelerated to speeds of approximately 0.1c, in good agreement with maximum velocities seen in BALs. The trail is compressed initially by the radiation pressure gradient in the radial direction when dust opacity is greater than unity and later the compression can be maintained by ram pressure. Although the initial acceleration is predominantly due to radiation pressure acting on dust, the dust may be sputtered and destroyed, leaving the radiation pressure in ion resonance lines to dominate (as long as the ionization parameter is less than less than or equal to 10). This model involving radiation pressure acting on both grains and resonance lines has the virtue of maintaining a low ionization parameter in the early phases of acceleration when the grains are optically thick, yet enabling line-locking in the resonance lines in the later phases of acceleration, in agreement with recent observations. Line profiles may be computed for this model and a critical test will be measurements of profile variability over time spans exceeding ten years.