BROAD ABSORPTION LINES IN QSOS: OBSERVATIONS AND IMPLICATIONS FOR MODELS

Abstract

Spectroscopic observations of fourteen broad absorption line (BAL) QSOs are presented and analyzed. Other observations are summarized. The following major conclusions are reached. Broad absorption lines (BALs) are probably present in 3 to 10 percent of the spectra of moderate to high redshift QSOs. The BALs exhibit a variety of velocity structures, from seemingly smooth, continuous absorption to complexes of individual absorption lines. Outflow velocities up to 40,000 km s⁻¹ are observed. The level of ionization is high. The minimum total absorption column densities are 10²⁰ to 10²² cm⁻². The emission line properties of BAL QSOs appear to be different from those of non-BAL QSOs. For example, N V emission is generally stronger in BAL QSOs and the emission near C III] λ1909 is generally broader in BAL QSOs. The distribution of multiplicities for isolated absorption troughs suggests that the large-scale spatial distribution of BAL clouds is non-random, possibly described by a disk geometry. The BAL clouds are incapable of accounting for all of the observed broad emission lines, particularly C III] λ1909 and Mg II λ2798. Therefore, if the BAL clouds give rise to observable emission, the generally adopted (optically thick, single component) model for the emission line region must be incorrect. Also, photoionization models, which utilize solar abundances and take the ionizing continuum to be a simple power law, are incapable of explaining the level of ionization in the BAL clouds. By considering the observed percentage of QSOs with BALs and resonance line scattering models, it is found that the absorption covering factor in BAL QSOs is between 3 and 20 percent. This suggests that possibly all, but not less than 15 percent, of the QSOs have BAL clouds associated with them. The amount of observable emission and polarization expected to be produced by the BAL clouds from resonance line scattering and collisional excitation is considered in detail. It seems likely that the BAL clouds contribute to the observed high ionization emission. A model worth exploring is one in which an inner, optically thick component gives rise to the low ionization emission, whereas an outer BAL cloud region gives rise to much of the high ionization emission.