next up previous contents index
Next: Absorption line spectra Up: QSOsGalaxies, and Previous: Some Recent Results

Science with the Hubble Space Telescope -- II
Book Editors: P. Benvenuti, F. D. Macchetto, and E. J. Schreier
Electronic Editor: H. Payne

Recent Results from HST Observations of Broad Absorption Line QSOs

D. A. Turnshek
Department of Physics & Astronomy, University of Pittsburgh, Pittsburgh, PA 15260 USA



Some recent results from HST observations of Broad Absorption Line (BAL) QSOs are presented. In particular, results in the following areas are discussed: (1) the large-scale BAL region covering factor, (2) information on the relative lateral extent and thickness of the BAL region as deduced from imaging and spectroscopy of the individual components of the gravitationally lensed ``Cloverleaf'' BAL QSO H1413+1143, and (3) ionization stratification and element abundances in the BAL region. These results have significantly clarified and improved our understanding of the BAL QSO phenomenon.

Keywords: AGN/QSOs: - Broad Absorption Lines, PG0043+0354, Q0226-1024; Element Abundances: - Broad Absorption Line Region; Gravitational Lenses: - Cloverleaf, H1413+1143


With the advent of HST, a number of different kinds of programs to study the nature of Broad Absorption Lines (BALs) in QSOs have been possible. Here we review some of them. Prior to the launch of HST, Turnshek (1988) gave a review of BAL QSOs. More recently, Turnshek (1995) reviewed the subject again. Some of the results discussed here are covered in more detail in this latest review or in more recent papers. Briefly, a BAL QSO is one which exhibits usually strong resonance absorption lines indicative of a high-velocity outflow superimposed on an otherwise nearly average QSO spectrum (continuum and emission lines). While it is clear that there are some systematic differences between the average spectra of non-BAL and BAL QSOs, the reasons for these differences have yet to be clearly established. Are we primarily dealing with effects due to aspect angle as some of the most recent models propose, or are there some QSOs with large BAL region covering factors that have different intrinsic properties or are in different stages of evolution? What do the observations tell us about the large- and small-scale BAL region geometry? Finally, independent of the difficult questions involving the geometry, the underlying intrinsic QSO properties, and stages of evolution, what do the observations tell us about the ionization stratification and element abundances of the BAL material that is being expelled at high velocity away from the central source? We consider these questions below.

The Large-Scale BAL Region Covering Factor

The large-scale BAL region covering factor is the fraction of a QSO's sky covered by BAL region material as viewed from the central source QSO. The value of this covering factor has always been one of the major questions with regard to interpreting the BAL phenomenon, and it is central to the question of whether BAL QSOs can be fitted into unified models of QSOs/AGN. The earlier work, which was mostly done using BAL QSOs showing only high-ionization BALs, indicated that there were not many photons in the absorption/emission profiles that could be attributed to resonance line scattering of inner photons by the outflowing BAL clouds. Since observational evidence for a viable mechanism for destroying the scattered photons could not be identified, this deficit of photons was taken to be evidence for a small BAL region covering factor, typically < 0.2 (Turnshek et al. 1980, Hamann, Korista & Morris 1993). This result fits very well into a unification scheme in which viewing angle is important. However, analysis of BAL QSO spectra which show low-ionization BALs as well as the high-ionization ones, indicates that dust reddening may be present (Weymann et al. 1991, Sprayberry & Foltz 1992), and so dust could lead to the destruction of a significant fraction of the photons which are scattered by the BAL region. This result leaves open the possibility that some objects have large BAL region covering factors. Analysis of the HST-FOS observations of PG0043+0354, identified as a BAL QSO during the course of the HST QSO Absorption Line Key Project, supports the idea that some QSOs have large BAL region covering factors (Turnshek et al. 1994).

To further investigate constraints on the value of the large-scale BAL region covering factor we recently completed a search for BALs in a sample of low-redshift weak-[OIII] QSOs (Turnshek et al. 1996a). We were motivated to undertake this study because of the work of Boroson & Meyers (1992). Boroson & Meyers suggested that radio quiet QSOs with weak-[OIII] and strong-FeII emission spectra form a class of QSOs that has a high probability of exhibiting BALs in their spectra. However, the Boroson & Meyers' suggestion was not based on observations of the classic CIV BALs normally used to identify BAL QSOs, but on rare low-ionization BALs due to, for example, NaI. In any case, Boroson & Meyers further argued that since narrow-line [OIII] emission is almost certainly emitted isotropically, an excess of BALs in weak-[OIII] objects could not simply be caused by observing this class of objects at a preferred aspect angle; it must however indicate that this class has larger BAL region covering factors. This would be an important conclusion, because evidence for generally small BAL covering factors in QSOs is consistent with scenarios where most QSOs have BAL regions (i.e., a unification scheme), while evidence to the contrary is consistent with scenarios where there are special classes of QSOs with substantial BAL region outflows. By making HST-FOS observations, and using IUE or HST archival data when available, we have explored the details of this effect by directly searching for classical CIV BALs in a sample of 18 weak-[OIII] QSOs with z < 0.35. This redshift range allows the strength of [OIII] to be deduced from optical spectra. We found that six of the 18 QSOs exhibit CIV BALs. In isotropic [OIII]-emitting models this suggests that the average covering factor in the weak-[OIII] class of QSOs is , which is significantly larger (with a > 99% probability) than the overall fraction of QSOs observed to have BALs ( 10%).

One possibility is that evolution is being observed, in the sense that QSOs evolve from dusty, lower ionization states with large BAL region covering factors to dust-free, higher ionization states with small BAL region covering factors. An alternative possibility is that we are dealing with classes of QSOs that have different intrinsic properties but are in nearly the same evolutionary stage. Thus, QSOs as a class may possess a large range of BAL region covering factors---a very small percentage of QSOs may have large BAL region covering factors, while a much larger percentage may have small BAL region covering factors.

The Gravitationally Lensed Cloverleaf BAL QSO H1413+1143

Another geometry issue which has relevance in BAL QSO research is the small-scale geometry of the BAL region. The small-scale geometry will dictate how scattering occurs in the BAL region and the degree to which BAL region clouds can cover the continuum source and inner broad emission line region---i.e., is the lateral extent of the BAL region along our sight-line much larger than the size of the continuum source or broad emission line region? HST observations of the gravitationally lensed ``Cloverleaf'' QSO H1413+1143 have proved crucial to considering these issues. First we review results that can be derived from analysis of WFPC and WFPC2 images of the Cloverleaf and then we discuss the geometry issues.

We have recently presented an analysis of HST WFPC and WFPC2 images of the Cloverleaf (Turnshek et al. 1996b). Astrometric and photometric measurements are derived for the four components for five different epochs over a baseline of 2.8yr. The data are taken with a variety of filters at the different epochs and can be pieced together to search for photometric variations among the components. The relative positions for the four components are measured to within 5--7mas (1); these results are more accurate than earlier measurements and agree with the previous work. Photometric measurements at any one epoch are normally accurate to --mag (1). The initial HST WFPC images cover a baseline of 1.3yr (1992.2 to 1993.5) and over this time interval there is little evidence for any significant component brightness variations at levels > 0.06mag ( 2). Photometric measurements with WFPC2 obtained with different filters extends this baseline an additional 1.5yr (to 1995.0). The WFPC2 data also fail to reveal significant brightness variations among the components. However, the WFPC2 data include both UV (F336W) and near-infrared (F814W) images. These color data indicate the presence of sight-line dependent extinction, causing the F336W--F814W color index of component B to be 0.5 mag redder than that of component C, which is the bluest component. Given the fact that the data do not appear to be confused by relative component brightness variations, the extinction corrected results provide information on the relative amplifications of the four image components. Thus, HST astrometry and photometry of the Cloverleaf sets clear constraints on models for the gravitational lens system. While existing models can successfully reproduce the relative positions (Kayser et al. 1990), the relative amplifications have yet to be been successfully modeled.

Concerning the small-scale BAL region geometry issues, we have also derived constraints on the lateral extent of the BAL region along our sight-line and the thickness of the BAL region relative to its lateral extent (Turnshek et al. 1996c). Some knowledge of the lateral extent is needed to assess procedures used to determine BAL region column densities because, e.g., if the lateral extent of the BAL region is too small the continuum source will not be completely covered and column densities will be underestimated. The observations of the Cloverleaf indicate that, while the BAL profiles along the four sight-lines are definitely not identical, they are very similar. Given the typical viewing angle differences of 0.7 arcsec, models of the Cloverleaf suggest that the sight-lines intercept the BAL region over a lateral size which exceeds the expected size of the continuum source. Observational constraints coupled with photoionization equilibrium constraints can then be used to infer an upper limit to the thickness of BAL region clouds. The similar BAL profiles among the four components indicate that average BAL region column densities integrated across the continuum source are relatively uniform (to within 10%) over a lateral extent which is much much larger than the thickness of a cloud and which is approximately the size of the continuum source. This suggests that derivations of BAL region column densities as a function of outflow velocity will normally be relatively accurate.

BALR Level of Ionization and Element Abundances

With HST-FOS it has now become possible to observe faint objects far enough into the UV to observe BALs due to different ions of the same element. Generally it has been impossible to do this simply by observing higher redshift QSOs in the optical, because at higher redshift intervening Lyman limit absorption normally obscures our view of such transitions. Thus, HST-FOS spectroscopy has allowed us to study and model ionization stratification and element abundances in the BAL region for the first time.

Recently, we have discussed evidence which indicates that the gas giving rise to the BALs at a specific outflow velocity has non-uniform ionization and enhanced abundances (Turnshek et al. 1996d). In the context of a photoionization model, ionization parameter fluctuations of at least 32--16 are needed to explain ionization stratification which gives rise to the observed column densities of different ions of the same element at a specific outflow velocity. In such a model the gas must have very enhanced metal-to-hydrogen abundance ratios relative to solar composition. However, the actual metal-to-hydrogen abundance enhancements are difficult to constrain because they are so model dependent. For example, in a photoionization model the shape of the photoionizing continuum has a significant influence on the derived metal-to-hydrogen enhancement, but for normally adopted shapes the enhancement is very large, i.e., a nitrogen-to-hydrogen enhancement 120--230 times solar. If collisional ionization is important, the need for metal-to-hydrogen abundance enhancements would be severely reduced. The derived relative metal abundances are somewhat more robust because the relevant metal-line transitions correspond to similar ionization potentials. In a photoionization model, nitrogen enhancements relative to carbon and oxygen of 9:--10: and 4--3 times solar values, respectively, are indicated. If normal stellar nucleosynthesis is important, the results may be indicative of star formation with a relatively flat initial mass function with the nitrogen overabundance produced by secondary processing in massive stars. However, large enhancements of elements like P in some objects suggest that the enrichment scenario may be more complex. In this respect we note that Shields (1996) has recently pointed out that the derived abundance enhancements appear to be similar to those observed in certain classes of novae.

Summary and Discussion

Thus, recent HST studies have significantly clarified and improved our understanding of the BAL QSO phenomenon. There are clear indications that in some classes of objects the large-scale BAL region covering factor is significantly greater than the fraction of QSOs with BALs ( 10%). In particular, in weak-[OIII] QSOs the covering factor may typically be 0.25--0.50. At the same time, recent results from spectropolarimetric observations of BAL QSOs (Glenn, Schmidt & Foltz 1994, Cohen et al. 1995, Goodrich & Miller 1995, Hines & Wills 1995) suggest that the large-scale BAL region geometry is non-spherical and, therefore, aspect angle effects should be important. Studies of the gravitationally lensed Cloverleaf QSO indicate that the small-scale BAL region geometry is highly flattened and fairly uniformly thick perpendicular to the sight-line. While this information is relevant to models for BAL cloud formation and acceleration, it further suggests that procedures used to determine BAL region column densities are generally valid. Attempts to model the column density indicate that at a given outflow velocity significant ionization stratification is present. In the context of a photoionization model the element abundances appear to be enhanced at least 10--100 times solar values relative to hydrogen, with N/C 10 and N/O 3.

Future observational efforts will, in principle, be able to further isolate classes of QSOs that have large BAL region covering factors. However, ionization models of the BAL region need to be better understood. One concern is the indication that constraints placed on the details of a photoionization model for the BAL region gas are unphysical (e.g., the need for fairly constant ionization independent of outflow velocity), indicating that central-source photoionization may not dominant. In any case, the results on the relative enhancements of the metals are providing important information on the chemical evolution of the QSO population in the young Universe.


This research was supported by grants from the Space Telescope Science Institute, which is operated by AURA, Inc., under NASA contract NAS5-26555. I wish to thank Brian Espey, Valery Khersonsky, Mike Kopko, Lincoln Lee, Olivia Lupie, Eric Monier, Patty Morris, Donette Noll, Sandhya Rao, Sheflynn Sherer, and Chris Sirola for useful collaborations and/or discussions during this work. The use of Gary Ferland's photoionization code CLOUDY is also acknowledged.


Boroson, T. & Meyers, K. 1992, ApJ, 397, 447

Cohen, M. et al. 1995, ApJ, 4448, L77

Glenn, J., Schmidt, G., & Foltz, C. 1994, ApJ, 434, L47

Goodrich, R. & Miller, J. 1995, 448, L3

Hamann, F., Korista, K., & Morris, S. 1993, ApJ, 415, 541

Hines, D. & Wills, B. 1995, ApJ, 448, L69

Kayser, R. et al. 1990, ApJ, 364, 15

Shields, G. 1996, ApJL, in press

Sprayberry, D. & Foltz, C. 1992, ApJ, 390, 39

Turnshek, D.A. et al. 1980, ApJ, 238, 488

Turnshek, D.A. 1988, in QSO Absorption Lines: Probing the Universe, eds. J.C. Blades, D.A. Turnshek, & C. Norman (CUP), p17

Turnshek, D.A. et al. 1994, ApJ, 428, 93

Turnshek, D.A. 1995, in QSO Absorption Lines, An ESO Symposium, ed. G. Meylan (Springer), p223

Turnshek, D.A., Monier, E., Sirola, C., & Espey, B. 1996a, ApJ, submitted

Turnshek, D.A., Lupie, O., Rao, S., Espey, B., & Sirola, C. 1996b, ApJ, submitted

Turnshek, D.A., Lupie, O., Monier, E., Espey, B., & Sirola, C. 1996c, in preparation

Turnshek, D.A., Kopko, M., Monier, E., Noll, D., Espey, B., & Weymann, R. 1996d, ApJ, in press (May issue)

Weymann, R. et al. 1991, ApJ, 373, 23

next up previous contents index
Next: Absorption line spectra Up: QSOsGalaxies, and Previous: Some Recent Results