2. The data
The considered sample is constituted of BAL QSOs with measured radio flux and good quality broad-band polarization data compiled from the literature. It also includes polarization measurements specifically obtained for the present study. This sample is given in Table 1. Table 2 contains additional BAL QSOs for which only radio flux and absorption line indices are available.
Table 1. The sample of BAL QSOs with measured radio flux and polarization.
Table 2. Additional BAL QSOs with measured radio flux and BAL indices.
Tables 1 and 2 list the QSO position-name (B1950 or J2000), the redshift z, and the object type / classification which depends on the presence of low-ionization BAL troughs. The balnicity index BI, the maximum velocity , and the detachment index DI are quantities which characterize the BALs, while is the continuum power-law index, and the debiased polarization degree. is the K-corrected radio-to-optical flux ratio, and the radio power at 5 GHz. Details on these quantities are given below.
2.1. The radio data
Most radio measurements are from the Stocke et al. (1992) BAL QSO VLA radio survey at 5 GHz. This sample provides a homogeneous set of K-corrected radio-to-optical flux ratios, and of radio powers in W Hz-1, K-corrected to 5 GHz in the QSO rest-frame.
For only one object of our previous polarization survey (Hutsemékers et al. 1998, hereafter Paper I), B1120+0154 (= UM425), an additional radio measurement is found in the literature (Meylan & Djorgovski 1989). It is included in Table 1 after computing and following the prescriptions of Stocke et al. (1992).
Since then, a handful of radio-loud BAL QSOs has been discovered by Brotherton et al. (1998). These five unusual objects (with J2000 coordinates) are also included in our sample. The K-corrected radio-to-optical flux ratios are from Brotherton et al. (1998), while the radio powers have been computed and K-corrected to 5 GHz according to Stocke et al. (1992). None of these formally radio-loud 1 BAL QSOs appear to be powerful radio sources (i.e. all have ; see also the discussion by Kuncic 1999).
2.2. The polarization
Most polarization data come from our previous survey (Paper I). However this sample was not chosen to investigate possible correlations with radio properties, and additional measurements were needed. Using the ESO 3.6m telescope + EFOSC, we then obtained new broad-band linear polarization data for 16 BAL QSOs, most of them with extreme radio properties, i.e. with the highest values or with stringent upper limits. These data are presented in Lamy & Hutsemékers (2000) with full account of the observation and reduction details. We also refer the reader to Paper I and to Lamy & Hutsemékers (1999) for details related to our previous survey and to reduction procedures. The polarization degree reported in Table 1 is debiased according to the Wardle & Kronberg (1974) method. Typical uncertainties of the polarization degree are 0.2-0.3%.
Independently of our survey, Schmidt & Hines (1999) have recently published a large number of BAL QSO polarization data, obtained mostly in white light. For the sake of homogeneity, we consider in Table 1 only the BAL QSOs of their sample with polarization degrees sufficiently accurate, i.e. with , denoting the uncertainty of the observed polarization degree p. This constraint, also applied to our data, is important since several BAL QSOs have low polarization levels (). 8 BAL QS0s from the Schmidt & Hines (1999) sample and with available radio measurements are then added to our sample. Their polarization degrees have been similarly debiased. The data of Schmidt & Hines (1999) also confirm our previous measurement of the polarization of B0145+0416, which was questioned in Paper I. Note finally that their spectropolarimetric data clearly show that broad-band polarization measurements represent fairly well the polarization of the continuum.
2.3. The spectral indices
Weymann et al. (1991) provide a series of useful spectral indices to characterize the absorption features of BAL QSOs. They define the balnicity index BI which is a modified velocity equivalent width of the CIV BAL, and the detachment index DI which measures the onset velocity of the strongest CIV BAL trough in units of the adjacent emission line half-width, that is the degree of detachment of the absorption line relative to the emission one (cf. Weymann et al. 1991 for more details). CIV BI and DI are reported in Tables 1 and 2. When BI are also given by Korista et al. (1993), we adopt an average of these values and those of Weymann et al. (1991). For the radio-loud BAL QSOs, BI are from Brotherton et al. (1998). We do not consider BI measured from other BAL troughs.
For a few BAL QSOs of our sample, the CIV DI are not given by Weymann et al. (1991). Therefore, as in Paper I, we have computed them by using good quality published spectra, when available. The spectra were digitally scanned, and the measurements done following the prescriptions by Weymann et al. (1991). The new measurements make use of spectra published by Korista et al. (1993) and Brotherton et al. (1998), and are reported in Tables 1 and 2 together with values from Paper I. Only one object (B0004+0147) with a spectrum in Korista et al. (1993) has no measured DI, due to an unusual emission line profile.
In addition to BI and DI, we have also reported in Tables 1 and 2 the maximum velocity in the CIV BAL trough, , which provides an estimate of the terminal velocity of the flow. For the radio-loud BAL QSOs, values of are given by Brotherton et al. (1998). For other objects, is evaluated from the Korista et al. (1993) and Steidel & Sargent (1992) spectra, by measuring, from the blue to the red, the wavelength at which the absorption first drops below the flux level defined by the local continuum (cf. also Lee & Turnshek 1995). Narrow or weak high-velocity absorption features are not taken into account. Further, at velocities higher than 25000 km s-1, CIV BALs may be contaminated by the SiIV emission line, such that the measurement of becomes inaccurate. We therefore limit to 25000 km s-1 from the CIV emission centroid, in agreement with the definition of BI (Weymann et al. 1991). This limit constitutes, in a few cases, a lower limit to the true .
Finally, we have measured the slope of the continuum as in Paper I. Using spectra published by Weymann et al. (1991) and by Brotherton et al. (1998), a power-law continuum was fitted blueward of CIII] (cf. Paper I). The resulting index is provided in Tables 1 and 2.
2.4. The LIBAL / HIBAL classification
Approximately 15% of BAL QSOs have deep low-ionization BALs (MgII 2800 and/or AlIII 1860) in addition to the usual high-ionization troughs (Weymann et al. 1991, Voit et al. 1993). These objects could be significantly reddened by dust (Sprayberry & Foltz 1992). Since they have no or very weak [OIII] emission compared to other objects, Boroson & Meyers (1992) have argued that LIBAL QSOs are not seen along a preferred line of sight but could constitute a physically different class of BAL QSOs.
In Paper I, we have defined three categories of LIBAL QSOs: strong (S), weak (W), and marginal (M) LIBAL QSOs. Indeed, in addition to the strong and weak LIBAL QSOs, first classified as such by Weymann et al. (1991), several authors have reported faint LIBAL features in a number of other objects (Hartig & Baldwin 1986, Hazard et al. 1984). We have classified the latter objects as marginal LIBAL QSOs. They are characterized by very weak MgII and/or AlIII BALs. The asymmetry of the MgII or CIII] emission lines, when cut on the blue side, is also considered as evidence for marginal LIBALs (Hartig & Baldwin 1986). BAL QSOs with no evidence for low-ionization features are classified as high-ionization (HI) BAL QSOs. Objects with poor quality spectra, or objects with no AlIII BAL and MgII outside the observed spectral range, remain unclassified (cf. Paper I for additional details and examples). It is important to note that the present classification somewhat differs from other classifications found in the literature. Most often, the BAL QSOs defined in the literature as low-ionization BAL QSOs are the S-LIBALs or the S-LIBALs + W-LIBALs, while the high-ionization BAL QSOs are anything else, including M-LIBALs and unclassified objects.
Our classification is summarized in Tables 1 and 2. It is based on a careful inspection of good-quality spectra available in the literature (Brotherton et al. 1998, Foltz et al. 1987, 1989, Hewett et al. 1991, Steidel & Sargent 1992, Turnshek et al. 1985, Turnshek & Grillmair 1986, Turnshek 1988, Wampler 1983, Weymann et al. 1991). Several BAL QSO sub-types were already given and discussed in Paper I.
© European Southern Observatory (ESO) 2000
Online publication: June 20, 2000