ON THE GEOMETRY, COVERING FACTOR, AND SCATTERING-EMISSION PROPERTIES OF QSO BROAD ABSORPTION-LINE REGIONS

We analyze line profiles of the 40 broad absorption-line (BAL) QSOs measured by Weymann et al. (1991) to estimate the geometry, covering factor, and scattering-emission properties of the BAL regions (BALRs). If the line photons are not selectively destroyed. the ratio of emission to absorption equivalent widths in C IV 1550 angstrom indicate that the BALR cannot completely envelope the continuum source. Detailed models of the C IV profiles formed in isotropic, conical, or disk-shaped winds strongly reinforced the requirement for generally low covering factors. Simultaneous modeling of the C IV and N V 1240 angstrom profiles argues again for low covering factors, and shows further that the approximately 1265 angstrom bump near N V is generally not scattered Lyalpha flux, as previously suggested, but rather an unrelated emission line. We conclude that the covering factors are usually small and therefore all radio-quiet QSOs have BALRs. These results apply equally for the sources with low-ionization BALs (Mg II) as long as the line photons are not selectively destroyed. If QSOs are observable over 4pi sr, the mean covering factor for the entire population equals the detection rate of BAL sources, i.e., approximately 12%. Low covering factors imply that the emission lines form mostly in the underlying broad emission-line region (BELR) and not the BALR. As a result, the geometry and orientation of the BALRs cannot be constrained by modeling the profiles of individual sources. However, we test the wind models by generating hypothetical mean spectra for randomly oriented BAL and non-BAL samples, and comparing the results to the mean spectra derived by Weymann et al. The observed approximately 25% larger N V equivalent width in the mean BALQSO spectrum could be caused by mean covering factors that are approximately 0.2 (i.e., approximately 20% of 4pi sr) larger in QSOs with BALRs in our line of sight. However, the narrow profile of the N V excess in BALQSOs is inconsistent with all of the isotropic, conical, and disk-shaped winds considered here. This result holds for a wide range of scattering-emission properties in the BALRs and is not sensitive to any details of the profile calculations. We conclude that (1) the observed N v excess results from differences (possibly orientation effects) in the BELRs and not the BALRs, and (2) the mean BALR covering factors in QSOs with and without observed BALs differ by less than 0.2.