There are two general issues concerning the possible relation between the BALs and the narrow absorption lines that in my opinion should be addressed. The first is the question of any observed similarity between the properties of the BALs and the properties of the so-called narrow absorption line systems. Is there any link between these two categories of systems? It would seem only natural given the variety of the BAL systems, their velocity extent and their frequency of occurrence, to have some lines-of-sight passing through the edge of a QSO's BAL region giving rise to relatively narrow absorption line systems in objects which are classified as non-BAL QSOs. To illustrate this, I discuss some spectra which are not of the standard BAL type, but which have properties which are reminiscent of the BAL type. I also discuss the question of whether the BAL flow itself may contaminate the intergalactic medium and give rise to intervening narrow absorption line systems. Both Briggs, Turnshek and Wolfe (1984) and Hazard et al. (1984) have suggested that a scenario in which BAL remnant ejecta contaminate the intergalactic medium may be important.
4.1. Similarity Between the BALs and the Narrow Absorption Lines
In studying the different narrow absorption line systems that are observed in QSO spectra, four different types of systems are sometimes considered:
4.1.1. Zabs Near Zem Systems
As discussed by Foltz et al. (1988) in the next paper, these systems tend to preferentially occur in radio-loud QSOs which suggests a dichotomy between them and the BAL systems. However, the strength and velocity extent of some of these systems suggests that they may have a relation to the BAL systems. In Figure 8 three of the strongest of these systems are illustrated. Aside from being generally weaker than the BAL systems and occurring in radioloud QSOs, these systems tend to have relatively stronger Ly- absorption and are of lower ionization than the BAL systems. Williams et al. (1975) and Morris et al. (1986) have derived lower limits on the distance between the central source QSO and the absorbing clouds using CII fine structure line measurements and have concluded that these clouds are at least tens of kiloparsecs from the central source. 3C191 exhibits evidence for absorption-absorption line-locking of 2 CIV doublets.
Figure 8. Three examples of radio-Loud QSOs showing strong absorption. The absorption is generally not quite strong enough to qualify for the classificat ion of BALs.
4.1.2. Highly Displaced Metal Line Systems
The study of Weymann et al. (1979) suggested that between 0-18,000 km s-1 there may be an excess of absorption line systems which cannot be explained by merely excluding the BAL systems and the zabs near zem systems. Of course, the excess of absorption line systems that Weymann et al. talked about could be single isolated intrinsic systems. An important question is: are there any absorption line systems which are not typical BAL systems, but which obviously have a rich absorption line spectrum indicating that spectroscopically, in terms of amount of absorption present, they appear to lie between the BAL type of intrinsic system and any single isolated type of intrinsic system that may exist? Figure 9 shows two spectra which may fit into this category. The property which may distinguish these systems from the majority of other metal line systems is the presence of NV absorption. Of course, a small fraction of the metal line systems show low ions and are inferred to be more than hundreds of kiloparsecs from the central source QSO based on the lack of CII fine structure lines (e.g., Turnshek, Weymann and Williams 1979), leading to energetic arguments which make it difficult to imagine how at least the low ionization systems in this class could be ejected.
Figure 9. Two examples of radio quiet QSOs exhibiting absorption which is intermediate in strength between typical narrow absorption lines seen in non-BAL QSOs and the broad absorption lines seen in BAL QSOs.
4.1.3. Damped Ly- Systems
The damped Ly- systems are probably the systems which are most secure in terms of the intervening interpretation being correct (see Wolfe 1988). However, it is annoying that three of the spectra of damped systems appear to exhibit evidence for absorption-absorption line-locking. As a reference, Figures 10 and 11 illustrate examples of absorption-absorption line-locking of narrow metal lines in two BAL QSO. Figure 10 shows the most striking case which occurs in the SiIV absorption lines in Q1303+308 (see also the center panel of Figure 5). A more subtle case occurring in Q1413+113 is shown in Figure 11. In this case the zabs = 1.870 narrow line system includes absorption components due to the CIV doublet and the AlII singlet, and so does the zabs = 1.659 narrow line system. However, the AlII singlet in the zabs = 1.659 system appears line-locked with the blue member of the CIV doublet in the zabs = 1.870 system, as can be inferred because the blue member of the CIV doublet has considerably more than twice the strength of the red member of the doublet. In Figure 12 three cases of absorption-absorption line-locking that appear to occur in objects having damped Ly- absorption are illustrated. These occur in the SiIV lines of PKS0458-020, the CIV lines of Q1244+347 and the MgII lines of MC1331+170. The probability that this occurs by chance is ~ 10%, although it is not completely clear how selection bias affects this calculation. In estimating this probability it was assumed that clustering of systems did not occur in the presence of damped Ly- systems.
Figure 10. The spectrum of the BAL QSO Q1303+308 which exhibits evidence for absorption-absorption line-locking in the SiIV absorption lines.
Figure 11. The spectrum of the BAL QSO Q1413+113 which exhibits evidence for absorption-absorption line-locking between CIV1548.2 and AlII1670.8.
Figure 12. Three examples of absorption-absorption line-locking in QSOs that have damped Ly- absorption.
4.1.4. Ly- Forest Systems
Finally, Figure 13 illustrates some Ly- forest data in a BAL QSO which at least suggests the possibility that some component of the Ly- forest may have an intrinsic nature. The BAL QSO Q1413+113 has an identification of resonance line absorption due to FeIII and/or PV in the spectrum. I interpret this as evidence for generally enhanced abundances relative to solar values (see Section 3.6). However, there is another trough in the Q1413+113 spectrum lying in the Ly- forest (indicated as unidentified in Figure 12) which I have not been able to identify in terms of resonance line absorption. This unidentified trough cannot be due to NV or SiIV because there is no evidence for corresponding CIV at the appropriate wavelengths. If the unidentified trough is due to Ly-, it shows no evidence for CIV. Could this apparently smooth absorption be a Ly--only ejected BAL trough? If so, its inferred outflow velocity ranges from ~ 10,000 to ~ 22,000 km s-1. The existence of a Ly--only ejected BAL trough would raise the question of whether any of the narrow Ly--only systems were ejected. I think it is important to keep in mind that because of the frequency of Lyman limit absorption systems and because of the blue cutoff in wavelength produced by the earth's atmosphere, it is normally not possible to explore the Ly- forest properties in QSO spectra at arbitrarily high inferred ejection velocities.
Figure 13. An unidentified smooth BAL trough in the spectrum of the BAL QSO Q1413+113. If this trough is due to Ly-, it has no apparent associated metal lines.
Hopefully, with the eventual launch of the Hubble Space Telescope, sensitive absorption line surveys at ultraviolet wavelengths will be made. In addition to extending the redshift pathlength that can be surveyed in any one QSO (when Lyman limit absorption is absent), low redshift examples of the presumed intervening objects giving rise to some of these four types of systems may then be imaged, eliminating some of the concerns that I have detailed here.
4.2. The Effect of the BAL Region Outflow
The final question I wish to consider is whether material from the BAL region outflow itself may eventually propagate to large distances from the QSO and give rise to some cosmologically intervening narrow absorption line systems seen in background QSO spectra. In particular, the question is whether the (past?) presence of a BAL region in a QSO with redshift za could give rise to absorption with a redshift za in the spectrum of a background QSO having emission redshift zb (i.e., za < zb is assumed).
Assuming that the BAL region distance from the central source QSO is r = 50r50 pc, that the global BAL region covering factor is qc = 0.08q8, that the total C+3 column density is N(C+3) = 1016N16 cm-2, that the typical BAL region outflow velocity is V = 104 V10,000 km s-1, that the ratio of the mass confining the BAL region to the mass in observable BAL region clouds is f = 100f100, that BAL region abundances are enhanced a factor of 10 over solar values by number and that one-third of the carbon is C+3, one finds that:
where MTOT is the total mass of the BAL region, Mmetals is the mass in metals of the BAL region, ETOT is the total kinetic energy of the BAL region flow and tcross is the time required for a cloud to cross the BAL region. Note that the assumption f100 = 1 comes from volume filling factor and cloud evaporation arguments which suggest that the medium confining the BAL clouds is > 100 times more massive than directly observable mass which gives rise to the BALs (e.g., Krolik, McKee and Tarter 1981; Weymann et al. 1982; Schiano 1986). Since tcross is likely to be considerably smaller than a QSO lifetime, BAL region clouds are required to be continuously created or injected. One can then infer the total mass loss rate, dMTOT / dt, and the total power in the outflowing clouds, dETOT / dt, to be:
Finally, taking the typical QSO lifetime to be tQSO = 107 t7 yrs, one can find MDEP and EDEP, the amount of mass and energy lost or deposited due to the BAL region flow,
It is clear that in the absence of local forces which would decelerate the BAL flow in the vicinity of the QSO, such a flow will freely expand into the intergalactic medium until it sweeps up its equivalent mass and begins to decelerate. Given the uncertainties, the resulting MDEP and EDEP are not out of line with the Ostriker and Cowie (1981) view of an intergalactic medium dominated by explosions.
Assuming that the BAL material is not locally decelerated in at least some fraction of QSOs (but see 3.5) and adopting the parameters derived above, the material will initially become very highly ionized due to the shock that is generated and, following the calculations of Ostriker and Cowie (1981), eventually will cool on a time scale of several billion years and come to a stop at a distance of ~ 1 Mpc from the central source QSO. The propagation distance, RPROP, depends weakly on EDEP, RPROP E0.2-0.3DEP. At moderate to high redshift, RPROP corresponds to a radius on the sky of ~ 102 arcsec. Considering the number of observed QSOs per square degree per unit redshift interval down to faint levels, a significant fraction of the sky may be covered by remnant BAL gas and swept up gas. Moreover, if the number of non-active QSOs (e.g. normal galaxies?) that once had BAL regions were much greater than the number of currently observable QSOs, which is consistent with t7 = 1, then the fraction of the sky per unit redshift effected by BAL `explosions' could easily exceed unity. Therefore, the remnant ejecta and swept up gas from the BAL regions of QSOs may cause many of the moderate to high redshift narrow absorption line systems seen in QSO spectra. These may be intergalactic clouds, loosely associated with galaxies that were once active. One constraint on this scenario involves the amount of time that it would take the shocked gas to cool. All of this activity would have had to have taken place at an early enough epoch in order for there to be sufficient time for the remnant BAL region gas and the swept up gas to cool.