A Chandra Survey of the X-ray Properties of Broad Absorption Line Radio-Loud Quasars
B. P. Miller, W. N. Brandt, R. R. Gibson, G. P. Garmire, O. Shemmer
aa r X i v : . [ a s t r o - ph . C O ] J u l A Chandra
Survey of the X-ray Properties of Broad AbsorptionLine Radio-Loud Quasars
B. P. Miller, W. N. Brandt, R. R. Gibson, G. P. Garmire, and O. Shemmer ABSTRACT
This work presents the results of a
Chandra study of 21 broad absorption line(BAL) radio-loud quasars (RLQs). We conducted a
Chandra snapshot surveyof 12 bright BAL RLQs selected from SDSS/FIRST data and possessing a widerange of radio and C IV absorption properties. Optical spectra were obtainednearly contemporaneously with the Hobby-Eberly Telescope; no strong flux orBAL variability was seen between epochs. In addition to the snapshot targets,we include in our sample 9 additional BAL RLQs possessing archival
Chandra coverage. We compare the properties of (predominantly high-ionization) BALRLQs to those of non-BAL RLQs as well as to BAL radio-quiet quasars (RQQs)and non-BAL RQQs for context.All 12 snapshot and 8/9 archival BAL RLQs are detected, with observedX-ray luminosities less than those of non-BAL RLQs having comparable opti-cal/UV luminosities by typical factors of 4.1–8.5. (BAL RLQs are also X-rayweak by typical factors of 2.0–4.5 relative to non-BAL RLQs having both com-parable optical/UV and radio luminosities.) However, BAL RLQs are not asX-ray weak relative to non-BAL RLQs as are BAL RQQs relative to non-BALRQQs. While some BAL RLQs have harder X-ray spectra than typical non-BALRLQs, some have hardness ratios consistent with those of non-BAL RLQs, andthere does not appear to be a correlation between X-ray weakness and spectralhardness, in contrast to the situation for BAL RQQs. RLQs are expected tohave X-ray continuum contributions from both disk-corona and small-scale jetemission. While the entire X-ray continuum in BAL RLQs cannot be obscuredto the same degree as in BAL RQQs, we calculate that the jet is likely partially Department of Astronomy and Astrophysics, The Pennsylvania State University, 525 Davey Laboratory,University Park, PA 16802; bmiller, niel, [email protected] Department of Astronomy, University of Washington, Physics-Astronomy Bldg Room C319, Seattle, WA98195; [email protected] Department of Physics, University of North Texas, 1155 Union Circle, [email protected]
Subject headings: galaxies: active — quasars: absorption lines — galaxies: jets
1. Introduction
Quasar outflows help regulate the accretion structure about the central supermassiveblack hole and propagate kinetic energy into the surrounding environment. Apparently themost extreme manifestation of outflows observed in radio-quiet quasars (RQQs) is the blue-shifted broad absorption lines (BALs) present in the rest-frame UV spectra of ≃ N H > cm − (e.g.,Gallagher et al. 2002). Although UV and X-ray absorption are clearly linked (e.g., Brandt,Laor, & Wills 2000), the higher column density of the X-ray absorber (e.g., Arav et al. 2003)suggests the X-ray absorption arises interior to the UV BALs, perhaps in the “shielding gas”postulated by Murray et al. (1995) and generated naturally in the simulations of Proga etal. (2000).A lack of detected BALs in radio-loud quasars (RLQs) led to early suggestions thatquasars could possess either a jet, or a BAL wind, but not both simultaneously (e.g., Stockeet al. 1992). However, an increasing number of individual BAL RLQs began to be discovered(e.g., Becker et al. 1997; Brotherton et al. 1998; Wills et al. 1999; Gregg et al. 2000; Ma 2002;Benn et al. 2005), and systematic optical spectroscopic coverage of quasars detected in the VLA We follow the convention that “radio-loudness” ( R ∗ ) is defined by the ratio of monochromatic luminosi-ties at rest-frame 5 GHz and 2500 ˚A (e.g., Stocke et al. 1992), where optical/UV luminosities have beencorrected for any strong intrinsic reddening. RQQs have R ∗ <
10 while RLQs require at least R ∗ > < R ∗ <
50 to be radio-intermediate and those with R ∗ >
50 to be definitivelyradio-loud. > ∼
100 (e.g., Becker et al. 2000,2001; Menou et al. 2001; Shankar et al. 2008). The fraction of quasars with BALs doesdecrease with increasing radio luminosity (e.g., Shankar et al. 2008). Several of the discoveredBAL RLQs have flat or convex radio spectra and/or compact morphologies (e.g., Becker etal. 2000; Liu et al. 2008; Montenegro-Montes et al. 2008), similar to the radio properties ofcompact steep spectrum (CSS) or GHz peaked spectrum (GPS) radio sources (e.g., O’Dea1998). The association of BAL RLQs with GPS/CSS radio sources, commonly presumedto be young (e.g., Stawarz et al. 2008 and references therein), has revived evolutionaryscenarios (e.g., Gregg et al. 2006), as has the prevalence of objects with low-ionization BALsamong dust-reddened quasars (Urrutia et al. 2009) which are plausibly newly active (e.g.,Urrutia et al. 2008). Further, Zhou et al. (2006) identify six BAL RLQs ( R ∗ < ∼
250 aftercorrecting for intrinsic extinction) with high brightness temperatures suggesting the nucleusis observed from a polar perspective. The presence of BALs in low-inclination RLQs wouldseem inconsistent with the quasi-equatorial disk-wind model often applied to RQQs.X-ray observations of BAL RLQs can provide insight into the nature of the BAL outflow,through quantifying any X-ray weakness or spectral hardening associated with BAL-linkedabsorption. Unfortunately, there have been only a handful of X-ray studies of BAL RLQs todate. Brotherton et al. (2005) conducted a
Chandra study of five BAL RLQs and found thatthey were X-ray weak but had relatively soft spectra, consistent with complex absorption ora jet-dominated continuum. These sources were of intermediate radio-loudness (with nonehaving a dereddened R ∗ > XMM-Newton observations of four BAL RLQs believed to be viewed at low inclinations (three of whichhave low-ionization BALs and only one of which has R ∗ > Chandra snapshot survey of a well-defined sample of 12BAL RLQs primarily selected from the SDSS Data Release 3 (DR3) BAL quasar catalog ofTrump et al. (2006). The objects were selected to be distinctly radio-loud ( R ∗ > ∼ ) and velocities; bothcore-dominated and lobe-dominated radio sources are included in the sample. For most ofthese objects we were able to obtain optical spectra and photometry with the Hobby-EberlyTelescope ( HET ) within ∼ Chandra pointing, to check for BALand continuum variability. We also make use of
Chandra archival data for an additional 9BAL RLQs (including those observed by SDSS in DR4 and DR5), all of which have C IVabsorption EW > R ∗ > ∼
50. Even taking redshift censoring into account, our sampleis dominated by high-ionization BAL quasars, which represent the majority of SDSS-selectedBAL quasars (e.g., Trump et al. 2006). We have carefully constructed comparison samplesof non-BAL RLQs, BAL RQQs, and non-BAL RQQs observed with SDSS/FIRST/
Chandra in order to provide proper context for our results and to enable interpretation of the physicalnature of BAL outflows in RLQs. Such comparisons are necessitated by the presence of aradio-linked component in the X-ray emission of RLQs, apparent both through increased X-ray luminosity (e.g., Worrall et al. 1987) and X-ray spectral flattening (e.g., Wilkes & Elviset al. 1987) with increasing radio loudness, and commonly presumed to arise in a small-scalejet. This paper is organized as follows: § § HET opticaland
Chandra
X-ray observations and provides notes on individual objects, § § § H = 70 km s − Mpc − , Ω M = 0.3, and Ω Λ = 0.7 is assumed throughout.Unless otherwise noted, errors are given as 90% confidence intervals for one parameter ofinterest (∆ χ = 2.71). Radio, optical/UV, and X-ray monochromatic luminosities l r , l uv ,and l x have units of log ergs s − Hz − , at rest-frame frequencies of 5 GHz, 2500 ˚A, and 2 keV,respectively. Object names are typically given as SDSS J2000.
2. Sample properties
Our sample of BAL RLQs consists of 21 objects, greatly increasing the number of BALRLQs with high-quality X-ray coverage. 12 of these BAL RLQs have X-ray data from a
Chandra snapshot survey (PI Garmire) and 9 have archival
Chandra coverage. 20/21 BALRLQs are detected in the 0.5–8 keV band. We also make use of comparison samples of RLQs, We use positive values throughout for C IV absorption EW; emission line properties are not consideredin this work. All EW values are rest-frame.
Chandra coverage.
The targets for the
Chandra snapshot survey were selected from the Trump et al. (2006)BAL quasar catalog, which includes SDSS quasars with spectra as of DR3. To ensureconsistent consideration of BAL properties, only C IV absorption measurements were used.The Absorption Index ( AI ; Hall et al. 2002) was required to exceed 1500 km s − to removeborderline BALs from further consideration. Note that AI is defined from zero velocitywith a minimum velocity width of 1000 km s − , and is consequently less restrictive thanthe traditional Balnicity Index ( BI ; Weymann et al. 1991); several objects in our samplehave BI = 0 km s − . Optical/UV luminosities were determined from SDSS photometry(corrected for Galactic extinction) through redshifting the composite quasar spectrum ofVanden Berk et al. (2001), convolving it with the SDSS filters, and then using the nearestmagnitude (or nearest two magnitudes) to 2500 × (1 + z ) ˚A to determine the continuum flux(assuming an optical/UV power-law continuum slope of α ν = − .
5, which is reasonable atthese wavelengths). We verified that alternative methods of calculating luminosities (e.g.,the spectral-fitting method of Gibson et al. 2009) yield similar results.These BAL quasars were then checked against the FIRST radio catalog to generate alist of BAL RLQs. Because FIRST has angular resolution sufficient to detect extended radioemission (when present) as distinct sources in many cases, it is necessary to consider thenearby environment to include all components (which could be some combination of core,lobes, and jet) and determine the full radio flux. Candidate matches were considered fromall fields in which there was either a FIRST source within 2 ′′ of the SDSS optical position,or two or more FIRST sources within 90 ′′ . All candidate fields were then examined toexclude intruding background sources (often identifiable due to an optical counterpart seenin the Digitized Sky Survey image). Radio luminosities were calculated at rest-frame 5 GHz,assuming radio power-law continuum slopes of α ν = − . α ν = − . Chandra snapshot targets were required to be distinctly radio-loud, defined as having R ∗ > ∼
100 and l r >
33. The optical spectra were checked for obvious signs of intrinsic reddening (see § . Chandra survey was then constructed from the brightest(in SDSS m i ) BAL RLQs. As can be seen in Figure 1, the survey is substantially completewithin DR3 BAL RLQs to m i < .
6. (The single DR3 object near m i = 17 . Chan-dra coverage is J144707.41+520340.0, which was considered for inclusion in the target list but 6 –dropped as lowest priority due to having the lowest absorption index, AI = 1517 km s − .)One BAL RLQ (J112506.95 − m i ≃ . Chandra coverage (see § .
2) and tocharacterize the BAL properties of the snapshot and archival samples (this catalog was notavailable at the time of our
Chandra target selection). All but one (J074610.50+230710.8)of the snapshot BAL RLQs are listed in Gibson et al. (2009), and the listed BAL RLQstargeted in the snapshot survey all have C IV EW > Two lists of BAL RLQs were checked for archival
Chandra
ACIS non-grating coverage;the first (134 BAL RLQs) was generated by cross-matching quasars with C IV absorptionmeasurements from the BAL catalog of Gibson et al. (2009) with the FIRST catalog, in amanner analogous to that outlined in § .
1, while the second ( ∼
50 BAL RLQs) was drawnfrom mentions of individual BAL RLQs in the literature. Naturally, BAL RLQs can appearin both of these lists. Candidates were required to be definitively radio-loud ( R ∗ > ∼
50 and l r >
32) with strong C IV absorption (EW > R ∗ = 10 border.Other X-ray telescopes cannot match the angular resolution of Chandra , important for BAL RLQs identified in the following references were checked for archival
Chandra coverage: Becker etal. 1997; Brotherton et al. 1998; Wills et al. 1999; Gregg et al. 2000; Becker et al. 2000; Becker et al. 2001;Menou et al. 2001; Brotherton et al. 2002; Lacy et al. 2002; Ma et al. 2002; Willott et al. 2002; Jiang &Wang 2003; Benn et al. 2005; Brotherton et al. 2005; Gallagher et al. 2005; Urrutia et al. 2005; Brothertonet al. 2006; Gallagher et al. 2006; Gregg et al. 2006; Miller et al. 2006; Schaefer et al. 2006; Zhou et al. 2006;Just et al. 2007; Kunert-Bajraszewska et al. 2007; Liu et al. 2008; Montenegro-Montes et al. 2008.
XMM-Newton archives for observations pointed to within 15 ′ of any of the BAL RLQs described above, andfind coverage of only three objects that would meet our selection criteria: J081102.91+500724.5(Wang et al. 2008), J101614.25+520915.4 (Schaefer et al. 2006), and J151630.30 − Some BAL RLQs targeted by
Chandra or mentioned in the literature have unusual andextreme characteristics, and caution is warranted before including such objects in a statisticalstudy. In particular, objects with heavy intrinsic reddening may have artificially elevatedapparent radio-loudness values. The BAL RLQs J100424.88+122922.2 (Lacy et al. 2002;Urrutia et al. 2005) and J155633.77+351757.3 (Becker et al. 1997; Brotherton et al. 2005)have radio-loudness values below our selection criteria after correcting for intrinsic reddening(J100424.88+122922.2 is also gravitationally lensed), and are therefore excluded from thearchival sample. Both these objects have low-ionization BALs, as do two additional BALRLQs presented in Brotherton et al. (2005) which are also strongly reddened (such thattheir corrected radio loudness values are below our selection threshold, although both werealready excluded from consideration here due to their low redshifts precluding observation oftheir C IV absorption properties); this is not unexpected, as low-ionization BAL quasars areknown to be generally redder than high-ionization BAL quasars (e.g., Reichard et al. 2003).We use the relative color indicator ∆( g − i ) (calculated by subtracting the medianquasar color at a given redshift) to check for large intrinsic reddening (e.g., Hall et al. 2006),keeping in mind that RLQs generally show slightly redder colors than do RQQs (e.g., Ivezi´cet al. 2002). The snapshot and archival BAL RLQs have relative colors that are on averageredder than those of SDSS quasars (although most of our BAL RLQs do have relative colors 8 –within the range spanned by 90% of SDSS quasars) but consistent with those of BAL RLQsin general (Figure 4). They do not appear to have strongly distorted radio-loudness values.The only established low-ionization BAL RLQ in our sample is the archival objectJ081426.45+364713.5, and although it does have a notably red relative color its ∆( g − i )value is within the tail of the BAL RLQ distribution and is significantly less than that of thestrongly reddened J155633.77+351757.3 (Figure 4). The CSS BAL RLQ J104834.24+345724.9suffers from intrinsic reddening (Willott et al. 2002), but its corrected radio-loudness is stillquite high, so it is retained in our archival sample but with a dereddened optical luminosity. In order to interpret the X-ray properties of BAL RLQs, it is necessary also to analyzecomparison samples of non-BAL RLQs (e.g., to gauge the expected X-ray luminosities, in-cluding the contribution from an unresolved jet to the X-ray nuclear emission), of BAL RQQs(e.g., to provide context for X-ray absorption relative to UV properties), and of non-BALRQQs (e.g., to give a baseline for measuring X-ray weakness in BAL RQQs). We caution thatthe comparison samples we use are specifically chosen to permit comparative investigationof our samples of BAL RLQs and should not necessarily be used to infer general propertiesof non-BAL RLQs, BAL RQQs, or non-BAL RQQs, particularly those having luminositiesoutside of the ranges studied here. The optical/UV luminosities and redshifts of the BALRLQs observed with
Chandra and of the comparison samples are shown in Figure 5.We constructed a comparison sample of RLQs by matching the SDSS DR5 QuasarCatalog (Schneider et al. 2007) to FIRST in a manner analogous to that described in § Chandra coverage with constraints of off-axis angle less than12 ′ , exposure greater than 1 ks, ACIS-S or ACIS-I used as the detector, and no grating. Thislist was then filtered to include only RLQs with R ∗ >
50 and l r >
32 so as to match theselection criteria for the BAL RLQ archival sample. X-ray luminosities were determined from
Chandra count rates using the method described in § .
2. There are 68 RLQs selected in thismanner, of which 67 (99%) are detected in X-rays. Additional luminous RLQs were addedfrom the sample of Worrall et al. (1987) based on
Einstein observations; after correcting A Kolmogorov-Smirnov (KS) test comparing the BAL RLQs observed with
Chandra to SDSS/FIRSTBAL RLQs with R ∗ >
50 and
EW > A gives p = 0 .
29 (comparing to BAL RLQs with R ∗ >
10 and
EW > A gives p = 0 . p = 0 . l uv > .
3, which yields a further 36RLQs, 32 (89%) with X-ray detections. The total RLQ comparison sample comprises 104RLQs, 99 (95%) with X-ray detections. Although some of the RLQs have redshifts too lowto permit ready observation of the C IV region, the fraction of BAL RLQs is small enough(see references in §
1) that any contamination of the comparison sample is minor and doesnot impact later analysis; we often refer to the RLQ comparison sample as “non-BAL RLQs”throughout.A comparison sample of BAL RQQs is taken from the BAL catalog of Gibson etal. (2009), combining their Table 1 (absorption properties) with their Table 5 (X-ray data).All high-ionization BAL RQQs with
Chandra coverage were selected, a total of 37 objectsof which 28 (76%) have X-ray detections. We also include those high-ionization BAL RQQslacking SDSS spectra (i.e., not available for inclusion in the Gibson et al. 2009 catalog) fromthe Large Bright Quasar Survey (LBQS; e.g., Foltz et al. 1987) observed with
Chandra byGallagher et al. (2006), an additional 15 objects, 13 (87%) detected by
Chandra . The totalBAL RQQ comparison sample comprises 52 BAL RQQs, 41 (79%) with X-ray detections.A comparison sample of non-BAL RQQs is taken from Gibson et al. (2008a); this samplehas an excellent combination of size, high-quality X-ray coverage, and well-characterizedUV properties. It is composed of the 139 non-BAL RQQs in their sample B, which ismade up of optically-selected quasars (SDSS objects targeted exclusively based on FIRSTor ROSAT properties excluded) with serendipitous (off-axis angle constrained to be 1 ′ <θ < ′ ) Chandra coverage having exposure > . . < z < .
7, with the lower-limit set to permit detection of C IV absorption if present(and thereby exclude BAL RQQs) and the upper limit set to permit direct measurement ofthe 2500 ˚A continuum flux. We also include 21 highly luminous non-BAL RQQs from Justet al. (2007), taking all objects in their “clean” sample with SDSS and
Chandra data, tomatch better the luminosity range of the BAL RLQs. The total RQQ comparison samplecomprises 160 RQQs, all of which have X-ray detections.
3. Observations and Notes3.1.
HET observations
We obtained optical photometry and spectroscopy of 10/12 of the snapshot BAL RLQsnear-contemporaneously with the
Chandra observations, using the queue-scheduled Hobby-Eberly Telescope (Ramsey et al. 1998). The Low-Resolution Spectrograph (LRS; Hill et al. 1998)was used for the spectroscopic observations, generally with a 1.5 ′′ slit and the g2 grating, 10 –providing a resolution of R ≃
867 (sufficient for productive comparison to SDSS spectra,which have a typical resolution of R ≃ (IRAF) software system using standard techniques, and theresulting spectra are presented in Figure 6. (The HET spectrum for J102258.41+123429.7is not shown; unfortunately the BAL region fell too close to the edge of the chip to providea useful comparison to the SDSS data.) None of the objects displays strong absorption-linevariability, although a few objects show minor changes in BAL structure (see § . . R -band images and looked for any flux variability via comparison to fieldstars and galaxies; none of the observed objects showed large ( > . HET observing log is provided in Table 1.
Chandra observations
All snapshot BAL RLQ
Chandra observations were carried out using the AdvancedCCD Imaging Spectrometer (ACIS; Garmire et al. 2003) with exposure times of 4–7 ks. Thetargets were positioned at the aimpoint of the S3 chip, and data were collected in Very Faintmode. The pipeline processing includes automatic application of both the ACIS charge-transfer inefficiency correction and the time-dependent gain adjustment, and it is carriedout using the calibration database version CALDB v3.4.2. The data were analyzed usingCIAO version 4.0.2.The archival BAL RLQ J081426.45+364713.5 has two
Chandra observations of com-parable quality. These were stacked for the purposes of determining source detection andextracting counts, and the resulting increase in signal-to-noise is helpful for more accuratelydetermining the X-ray properties of this faint off-axis source.Source extraction for the BAL RLQ snapshot and archival objects and for the non-BALRLQ comparison sample (see § . Chandra sources was performed using 90% encircled-energy radii, using nearby source-free regions for background determination. We evaluatesource detection through comparison of the observed aperture counts to the 95% confidenceupper limit for background alone. Where the number of background counts is less than10 (as applies in almost all cases) we use the Bayesian formalism of Kraft et al. (1991) todetermine the limit; else, we use equation 9 from Gehrels (1986). If the aperture countsexceed the 95% confidence upper limit we consider the source detected and calculate the netcounts by subtracting the background from the aperture counts and then dividing by the http://iraf.noao.edu/iraf/web/
11 –encircled-energy fraction; else, the source is considered undetected and the upper limit isused. All snapshot BAL RLQs are detected, with net 0.5–8 keV counts ranging from 17 to ∼
170 (Table 1), and 8/9 archival BAL RLQs are detected (Table 2). We confirmed sourcedetections for the BAL RLQs by running the CIAO wavdetect routine on 200 ×
200 squarepixel images centered at the SDSS object coordinates, with wavelet scales of 1, 1.41, 2, 2.83,4, and 5.66 pixels; most sources are detected using a significance threshold of 10 − , whileJ081426.45+364713.5 is detected (in the stacked image only) using a significance thresholdof 10 − .All BAL RLQs were examined for variability within the Chandra observation using theGregory-Loredo algorithm implemented by CIAO . This method filters by relevant good timeintervals and accounts for any dither near chip edges for off-axis sources. The probabilitythat a source is variable can be indicated with a variability index, ranging from 0 to 10; mostBAL RLQs had values of 0 (“definitely not variable”) with only 3 objects having variabilityindices as high as 2 (“probably not variable”). Longer exposures could more tightly constrainvariability on ks timescales, while repeat observations could assess variability between epochs. Optical/UV properties (including absorption characteristics) of the BAL RLQs are listedin Table 3, while radio fluxes and spectral indices are given in Table 4. Below, we brieflycomment on interesting aspects of the BAL RLQs. has a relatively large C IV absorption index of AI = 2955 km s − (Trump et al. 2006), and has wide and deep BAL-like absorption structure (Figures 3 and 6)despite being the only snapshot BAL RLQ not included in the Gibson et al. (2009) BALquasar catalog. The HET spectrum shows enhanced absorption in the higher-velocity BAL(Figure 6), perhaps qualitatively consistent with the tendency of BAL RQQs to vary withinnarrow discrete regions (Gibson et al. 2008b). It is the reddest snapshot BAL RLQ with∆( g − i ) ≃ . g − i ) ≃ . VLA
X-band image (program AB862, observation date 1998-05-04) suggests the radio spectrumof this compact-morphology source peaks near 5 GHz. http://cxc.harvard.edu/ciao/ahelp/glvary.html
12 –
J083749.59+364145.4 has particularly strong C IV, Si IV, Ly α , and O VI BALs.The C IV EW of 34.6 ˚A is the largest in the snapshot or archival BAL RLQ sample. Itappears to have a GHz-peaked (possibly variable) radio spectral shape and is unresolved atmilliarcsecond scales (Montenegro-Montes et al. 2008, 2009). J085641.58+424254.1 displays strong N V absorption. The HET spectrum suggests theC IV emission line might be slightly variable.
J092913.96+375742.9 (also FBQS J092913.9+375742) appears to be resolved in an X-band VLA image (program AG0574, observation date 1999-07-12). There is a jet-like featurewith a flux of 2.1 mJy located 0 . ′′ West of the core.
J102258.41+123429.7 is resolved into a double-lobed morphology by FIRST (see Fig-ure 7), and the southern lobe shows extended diffuse emission past the primary hotspot.The
Chandra image does not show any extended X-ray emission, but there are only ∼ J105416.51+512326.0 has a flat radio spectrum that steepens to α r = − .
35 above1.4 GHz.
J112506.95 − is resolved into a double-lobed morphology by FIRST (de Vrieset al. 2006; see Figure 7). The Chandra image does not show any extended X-ray emission;there are ∼
80 X-ray source counts. The primary C IV absorption trough is at low velocityand splits the emission line.
J115944.82+011206.9 (also B1157+014) was identified as a BAL RLQ by Menou et al. (2001),who noted that in addition to the primary low-velocity BAL there is an additional absorptiontrough near 8000 km s − . The depth of this secondary absorption may have increased slightlybetween the SDSS and HET observations. The radio spectrum appears to be double-peaked(Montenegro-Montes et al. 2008). The source shows symmetric jet-like extended emission onmilliarcsecond scales and a one-sided misaligned sequence of faint knots stretching to ∼ m i inthe snapshot sample and has sufficient X-ray counts ( ∼
170 from 0.5–8 keV) for basic spec-tral analysis (Figure 8). The relatively hard X-ray spectrum suggests intrinsic absorption; aneutral absorber has a best-fit column density of N H = 3 . +2 . − . × cm − with an unusualflat photon index of Γ = 1 . +0 . − . required. J123411.73+615832.6 has an atypical BAL structure, with a deep and wide troughthat decreases gradually in depth until smoothly meeting the base of the Si IV emission line.Narrow redshifted C IV absorption is also present. A C-band VLA image (program AP450,observation date 2003-02-27) indicates the radio spectral index is α r ≃ − .
5. 13 –
J133701.39 − has the highest measured X-ray hardness ratio in the snapshotor archival BAL RLQ sample. A C-band VLA image (program AG400, observation date1994-01-08) suggests this is a flat-spectrum RLQ with α r ≃ − . J141334.38+421201.7 (also FBQS J141334.4+421201) was identified as a BAL RLQ byBecker et al. (2000). The radio spectrum is complex (Montenegro-Montes et al. 2008) whilethe morphology is compact with a one-sided jet on milliarcsecond scales (Liu et al. 2008).
J162453.47+375806.6 is described in detail by Benn et al. (2005), and we use their valueof BI = 2990 km s − and estimate V max = 28300 km s − rather than taking measurementsfrom Gibson et al. (2009) (for which BAL absorption was integrated to 25000 km s − ). Thelarge minimum ( V min = 20560 km s − ) and maximum velocities of the C IV BAL in this sourceare unusual for BAL RLQs and unique within our snapshot and archival samples. There isalso low-velocity absorption, described by Benn et al. (2005) as a mini-BAL (defined as totalvelocity range < − ; the mini-BAL is shaded gray along with the primary BAL inFigure 6 for identification). The radio spectrum is GHz-peaked (steep at high frequencies)and milliarcsecond imaging reveals a one-sided jet (Benn et al. 2005; Montenegro-Montes etal. 2008, 2009). − (also FBQS J0200 − ∼
18 ks ACIS-I image (ObsID 3265; PIEbeling) and is discussed by Gallagher et al. (2005). The source has a radio-loudness valueof R ∗ = 48, on the border for inclusion in the archival sample. FBQS J0256 − was identified as a BAL RLQ by Becker et al. (2001). Flux measure-ments by Montenegro-Montes et al. (2008) indicate a steep radio spectrum; those authorsalso note the increased flux measured by FIRST relative to NVSS may be due to variability.The ∼ − g − i ) = 0. J081426.45+364713.5 has a low radio luminosity ( l r = 32 .
7) and the reddest relativecolor [∆( g − i ) = 1 .
1] in the snapshot or archival BAL RLQ sample. The optical spectrumshows deep and wide BALs in both high and low-ionization lines (Trump et al. 2006 categorizeit as an FeLoBAL) and only weak emission lines. It was observed serendipitously in two ∼
10 ks ACIS-I exposures (ObsID 3436, 3437; PI Fox) and is X-ray weak. 14 –
J091951.29+005854.9 has a non-zero BI = 673 . − and a C IV EW of 6.9 ˚A, butthe BAL is relatively narrow and the absorption index is low ( AI = 1268 km s − ). The radioloudness is also borderline for our sample ( R ∗ = 51). It was observed serendipitously in a ∼ J100726.10+124856.2 (also PG 1004+130) is a low-redshift ( z = 0 .
24) RLQ in whichBALs were discovered by Wills et al. (1999); we use their values of BI = 850 km s − and V max = 10000 km s − since the SDSS spectrum does not cover the C IV region. It is alsoa hybrid-morphology radio source (Gopal-Krishna & Wiita 2000), with an edge-brightenedlobe opposite a broadening edge-darkened jet. It is perhaps the best-studied BAL RLQ atX-ray frequencies: deep XMM-Newton and
Chandra observations show X-ray absorptionvariability and also reveal X-ray jet emission (Miller et al. 2006). PG 1004+130 is X-rayweak relative to comparable non-BAL RLQs.
J104834.24+345724.9 (also 4C +35.23) is a CSS RLQ with a C IV BAL identifiedby Willott et al. (2002). It is radio luminous, and even after correcting for some intrinsicreddening it remains notably radio-loud (Kunert-Bajraszewska et al. 2007). It was targetedby
Chandra in a ∼ J122033.87+334312.0 (also 3C 270.1) is a double-lobed steep-spectrum RLQ with thesecond-highest radio-loudness (log R ∗ = 4 .
2) in the snapshot or archival BAL RLQ sample.Low-velocity C IV absorption has been known to be present in this object for some time(e.g., Anderson et al. 1987) although it has not necessarily been described as a BAL quasar;however, the balnicity index measured from the SDSS data is non-zero ( BI = 52 .
5; Gibsonet al. 2009). J122033.87+334312.0 was observed serendipitously in a ∼ J131213.57+231958.6 (also FBQS J131213.5+231958) was identified as a BAL RLQby Becker et al. (2000) and shows a wide and deep C IV absorption trough that extendsto 25000 km s − (from FBQS data; the DR7 SDSS spectrum does not have sufficient shortwavelength coverage to see the BAL). It shows two-sided extended radio emission on mil-liarcsecond scales but is core dominated (Jiang et al. 2003) and likely variable (Montenegro-Montes et al. 2008); these radio characteristics do not provide a self-consistent orientationmeasure. Liu et al. (2008) suggest this object is similar in some respects to CSS sources.The ∼ LBQS 2211 − was identified as a “marginal” BAL quasar by Weymann et al. (1991),with a non-zero but low BI = 27 km s − , and is radio-loud based on an NVSS flux mea- 15 –surement. The ∼ g − i ) = 0 .
32 based ondata from Gallagher et al. (2007).
4. Data Analysis
Because RLQs are generally more X-ray luminous than comparable RQQs (e.g., Worrallet al. 1987), a direct comparison of the X-ray properties of BAL RLQs to those of BAL RQQsis of limited value. To gain additional insight, we quantify the degree to which BAL RLQsare X-ray weak relative to non-BAL RLQs, then compare this to the X-ray decrement forBAL RQQs relative to non-BAL RQQs. Much of the X-ray weakness of BAL RQQs may beexplained by low-energy X-ray absorption (e.g., Gallagher et al. 2006), which can produceX-ray spectra that are harder than typical; examination of the X-ray spectral propertiesof BAL RLQs compared with those of non-BAL RLQs can clarify whether a similar effecttypically applies to BAL RLQs.
When insufficient counts are available for productive spectral modeling, as is unfortu-nately the case for the majority of our data, the relative contributions of hard and soft X-rayemission to the overall spectrum can be assessed from the hardness ratio HR = ( H − S ) / ( H + S ),where H and S are the net hard-band (2–8 keV) and soft-band (0.5–2 keV) counts, respec-tively. Large values of the hardness ratio can indicate intrinsic absorption, or alternatively anunusually flat power-law (or both effects together). Since all our data are taken from Chandra (including the non-BAL RLQ, BAL RQQ, and non-BAL RQQ comparison samples) we cancompare hardness ratios without concern for intrumental cross-calibration effects. Hardnessratios for objects observed with one of the front-illuminated CCDs have been adjusted bysubtracting 0.14 to enable direct comparison to the hardness ratios for the back-illuminatedCCDs (such as S3, which covers the ACIS-S aimpoint).The probability distribution for the hardness ratio can be calculated using the Bayesianformalism detailed in Jin et al. (2006), using a uniform prior (i.e., their equation 13). Themaximum-likelihood hardness ratio is simply ( H − S ) / ( H + S ). We use this method tocalculate 1 σ errors on the value of HR , defined such that 68% of the area above (below) themaximum likelihood hardness ratio is enclosed within the range of the upper (lower) bound(e.g., Wu et al. 2007). Where the total number of counts exceeds 100 symmetric errors are 16 –calculated from equation 8 of Jin et al. (2006).X-ray luminosities are calculated from the 0.5–8 keV count rates, which are converted toobserved-frame 2 keV flux densities with PIMMS , in all cases assuming Galactic absorptionand a power-law spectrum with Γ=1.5. This model is typical of RLQs: for example, Reeves& Turner (2000) found h Γ i = 1.66 with σ = 0.22 for an ASCA sample of 35 RLQs, whilePage et al. (2005) determined h Γ i = 1.55 with σ = 0.29 for an XMM-Newton sample of 16RLQs at z >
2. However, reasonable alternate choices for Γ have only a few percent impactupon the calculated X-ray fluxes. Count rates for archival observations were converted to fluxdensities using the calibration appropriate to that cycle, in order to account for the temporalchanges in ACIS sensitivity. The ACIS-I model in PIMMS was used for all front-illuminatedchips.The bandpass-corrected X-ray luminosities l x are given in units of log ergs s − Hz − at rest-frame 2 keV in Table 5. We also calculate X-ray luminosities l x , S and l x , H at rest-frame 2 keV determined from the soft and hard-band count rates, respectively, in order toinvestigate the influence of spectral shape (and intrinsic absorption) on the X-ray luminosity. X-ray and optical/UV luminosities are correlated in quasars, and so to evaluate thedegree of X-ray weakness in BAL quasars it is necessary to compare to non-BAL quasars ofsimilar optical/UV luminosity. Extensive studies (e.g., Avni & Tananbaum 1986; Strateva etal. 2005; Steffen et al. 2006; Just et al. 2007; Kelly et al. 2007) have demonstrated that X-rayluminosity in non-BAL RQQs may be parameterized as l x ∝ β × l uv , where β ≃ . − . l uv ). There is continuing debate (e.g., Just et al. 2007; Kelly etal. 2007) as to whether the optical/UV-to-X-ray properties of individual RQQs are alsosignificantly dependent upon redshift, but for our purposes the l x ( l uv ) parameterization isfully satisfactory to explore the large deviations from predicted X-ray luminosity that areseen in BAL quasars. We make use of the relation l x = 0 . × l uv + 7 .
055 (a linear fit to logluminosities) found by Just et al. (2007) taking l x as the dependent variable and fitting theirlarge sample of non-BAL RQQs using the Astronomy Survival Analysis Package (ASURV;Lavalley et al. 1992). Using the Bayesian maximum-likelihood method of Kelly (2007), whichaccounts for both upper limits and errors (we presume uncertainties are dominated by typicalquasar variability; cf. § . http://cxc.harvard.edu/toolkit/pimms.jsp
17 –yields a similar relation of l x = (0 . ± . × l uv + (8 . ± . l x ( l uv ) for non-BAL RLQs as for non-BAL RQQs, finding a best-fit correlation of l x = (0 . ± . × l uv − (0 . ± . ≃
15% of BALRQQs. The sample BAL RLQs also tend to be X-ray weaker than non-BAL RQQs at lowoptical/UV luminosities.The X-ray luminosities of non-BAL RLQs can also be parameterized as a function ofradio luminosity, with a best-fit result for the comparison non-BAL RLQ sample of l x = (0 . ± . × l r + (6 . ± . l x ( l r ) for BAL RLQs reflects not only X-ray weakness but also lower radio-loudness val-ues for BAL RLQs than for the comparison non-BAL RLQs; the median value of log R ∗ is2.2 for the BAL RLQs and 3.0 for the RLQs. The outlier with high radio luminosity andlow X-ray luminosity is the CSS source J104834.24+345724.9 (see also § and radio luminosity yields a relation with re-duced scatter: l x = (0 . ± . × l uv + (0 . ± . × l r − (1 . ± . l x ( l uv ) and l x ( l r ) for l x ( l uv , l r ). Obviously more sophisticated models are possible, but this provides auseful quantitative measure of X-ray luminosity in BAL RLQs relative to RLQs taking intoaccount both optical/UV and radio properties.The difference between the observed X-ray luminosity in BAL RLQs and that predictedfrom non-BAL RLQs with comparable optical/UV luminosities, ∆ l x , uv = l x − l x ( l uv ), is plot-ted as a histogram in Figure 10a. The scatter for non-BAL RLQs (Figure 10b) is smallerthan the degree to which BAL RLQs are X-ray weak. However, BAL RLQs do not extendto extreme values of X-ray weakness (log offsets of < − l x , uv is calculated from the Just et al. 2007 relation for non-BAL RQQs) relativeto non-BAL RQQs (Figure 10d). Note that the underlying distribution of ∆ l x , uv for BAL Here and for subsequent model fits the quoted parameter values are the median of draws from theposterior distribution and the errors are 1 σ .
18 –RQQs is even X-ray weaker than the histogram in Figure 10c suggests, as there are a largenumber of X-ray upper limits. There appears to be a limit to how X-ray weak BAL RLQscan become. The Kaplan-Meier estimates of the median ∆ l x , uv values are − . − . − .
06 for BAL RLQs, non-BAL RLQs, and BAL RQQs, respectively. A Peto-Prenticetwo-sample test indicates that the distribution of ∆ l x , uv values for BAL RLQs is significantlydifferent from that of non-BAL RLQs (test statistic 6.621, p < × − ) and BAL RQQs(test statistic 2.249, p = 0 . and radio luminosities, ∆ l x = l x − l x ( l uv , l r ). BAL RLQs are a fac-tor of 2.0–4.5 (median 3.2) weaker in X-rays than comparable non-BAL RLQs. The resultsare similar if the relative X-ray luminosity is instead calculated from the soft or hard-bandluminosities, l x , S and l x , H (Figures 10f and 10g).There is a general trend in quasars relating C IV absorption to X-ray weakness (e.g.,Brandt, Laor, & Wills 2000; Laor & Brandt 2002; Gallagher et al. 2006). We plot the rela-tive X-ray luminosity for BAL RLQs [presented as ∆ α ox for ease of comparison to previouswork, where α ox = 0 . × ( l x − l uv ) and ∆ α ox = α ox − α ox ( l uv ) with α ox ( l uv ) calculated fromthe l x ( l uv ) relations given above] versus C IV EW (Figure 11a) and maximum outflow ve-locity (Figure 11b). It is apparent that even BAL RLQs with large C IV equivalent widths(10–40 ˚A) do not have ∆ α ox < − .
5, as do many strongly absorbed BAL RQQs. BAL RLQsappear to follow the correlation between maximum outflow velocity and relative X-ray lu-minosity that holds for BAL RQQs, but only to a limiting value of ∆ α ox , near which BALRLQs are observed with a wide range of outflow velocities.Additional context for interpreting the relative X-ray luminosities of some BAL RLQsis provided by their radio morphologies or spectral properties, which can constrain source in-clination (e.g., Wills & Brotherton 1995) or age (e.g., Stawarz et al. 2008). Lobe-dominatedBAL RLQs (nested symbols in Figures 9 and 11), which presumably lie at larger an-gles to the line of sight than do core-dominated BAL RLQs, show a range of behav-ior: J112506.95 − − Most of the quasars in the snapshot and archival BAL RLQ samples and in the non-BAL RLQ, BAL RQQ, and non-BAL RQQ comparison samples lack sufficient counts forproductive spectral fitting, so we investigate basic X-ray spectral properties using hardnessratios (see § HR , although this effect can be diluted bycomplex (partial covering or ionized) absorption, such as is established to occur in BAL RQQs(e.g., Gallagher et al. 2002, 2006) and has been suggested for BAL RLQs (e.g., Brothertonet al. 2005). Absorption spectral effects are also diluted by increasing redshift pushing therest-frame soft band to lower observed-frame energies.The non-BAL RLQs in our comparison sample have relatively uniform hardness ratiosthat do not appear strongly dependent upon redshift (Figure 12a), suggesting that thespectra for these RLQs are generally dominated by a simple power-law component witha standard photon index and insignificant intrinsic absorption. Many BAL RLQs havehardness ratios similar to those of non-BAL RLQs, but there are several BAL RLQs withharder X-ray spectra, although none with measured HR > .
2. A Peto-Prentice two-sampletest indicates that the distribution of HR values for BAL RLQs is significantly differentfrom that of non-BAL RLQs (test statistic 3.704, p = 2 × − ). BAL RQQs typicallyhave harder X-ray spectra than non-BAL RQQs (the distributions are statistically different,with p < × − ) and can have extreme hardness ratios (Figure 12b). The distributionof hardness ratios for the snapshot and archival sample of BAL RLQs is not statisticallyinconsistent with that of BAL RQQs (test statistic 1.396, p = 0 . HR values for the BAL RLQs, non-BAL RLQs, BAL RQQs,and non-BAL RQQs in our samples are − . − . − .
26, and − .
57, respectively; thedistribution for BAL RLQs is skewed, with a Kaplan-Meier estimate of the mean HR of − .
34. The slightly higher median HR for non-BAL RLQs relative to non-BAL RQQsmight be expected from prior X-ray spectral studies, but the distributions for our comparisonsamples of non-BAL RLQs and non-BAL RQQs are not statistically inconsistent ( p = 0 . Einstein would have slightly larger
Chandra hardness ratios than the RLQs plotted here). For reference, a photon index ofΓ = 2 approximately corresponds to HR = − . . HR = − . HR > − . − XMM-Newton
20 –spectrum is softer), and J104834.24+345724.9 (the CSS source 4C+35.23). Although thisstudy is not designed to investigate the X-ray spectral properties of various subcategories ofBAL RLQs, we note briefly that the GPS sources J083749.59+364145.4 and J162453.47+375806.6have soft X-ray spectral shapes, and the FeLoBAL J081426.45+364713.5 has an intermediate HR = − .
29, slightly harder than the median for the BAL RLQs studied here.The correlation between X-ray weakness and hardness ratio in BAL RQQs (Figure 12d)is reflective of (often complex) absorption reducing the soft-band X-ray flux in BAL RQQs(e.g., Gallagher et al. 2006). A similar trend is not obviously apparent for BAL RLQs (Figure12c), for which there are several X-ray weak objects with low hardness ratios (or soft X-rayspectra), and essentially no BAL RLQs with ∆ l x ( l uv ) < ∼ −
1. Note that our observations aresensitive to low values of ∆ l x ( l uv ) (only one undetected BAL RLQ is not plotted, and manyof the rest could be detected if they were even X-ray weaker by a linear factor of 5–10; seenet counts in Tables 1 and 2); the sample simply lacks notably X-ray weak BAL RLQs. Itdoes not appear possible to ascribe X-ray weakness in BAL RLQs to intrinsic absorption(with properties as in BAL RQQs) obscuring the entire nuclear X-ray continuum source,although such an interpretation could hold for some particular BAL RLQs.
5. Discussion5.1. Physical models
The above results suggest a picture in which BAL RLQs are in some sense intermediatebetween BAL RQQs and non-BAL RLQs: BAL RLQs are X-ray weak, but not to the samerelative degree as are BAL RQQs, and they can have harder X-ray spectra, but often havehardness ratios consistent with those of non-BAL RLQs. A simple physical model could alsoportray BAL RLQs as having X-ray characteristics of both BAL RQQs (an outflowing BALwind that is associated with an X-ray absorber) and non-BAL RLQs (an unresolved X-rayemitting jet that contributes to the total continuum). There are too many free parameters toconstrain such a model in detail, but some insight can be gained by making the simplifyingassumptions that the disk-corona emission in RLQs has the same optical/UV-to-X-ray prop-erties as are observed in RQQs [i.e., that any systematic differences in the accretion structureof RLQs compared to RQQs do not produce dramatic changes in the l x ( l uv ) relation], andthat the optical luminosity in RLQs is dominated by disk-related emission with only a mini-mal jet-linked contribution (certainly plausible for these broad-line RLQs, and often inferredfor even RLQs in which the radio and X-ray emission is established as jet-dominated; e.g.,Sambruna et al. 2006). Then the disk-corona X-ray luminosity in RLQs may be calculatedusing the l x ( l uv ) relation for RQQs, and any additional X-ray luminosity may be ascribed to 21 –jet-linked emission.The ratio of total RLQ X-ray luminosity to equivalent RQQ (disk-corona) X-ray lu-minosity increases with increasing radio luminosity (Figure 13a); this presumably reflectsincreasing jet luminosity at both radio and X-ray frequencies with decreasing inclination.The precise nature of the X-ray jet emission in RLQs (and its dependence upon inclination)remains a matter of debate, although it seems likely that two-zone models are required (e.g.,Jester et al. 2006), in which beamed jet-linked X-ray emission from the fast spine dominatesfor objects viewed at low inclinations while a slower (and less beamed) sheath could generatejet-linked X-ray emission radiated in a more isotropic manner. We refrain from imposinga particular jet model upon the data but quantify the observed increase in the X-ray lu-minosity of RLQs relative to RQQs with increasing radio luminosity through the trendlineshown in Figure 13a. Non-BAL RLQs with optical/UV and radio luminosities similar tothose of the BAL RLQs in our sample would have total X-ray luminosities greater than thenon-BAL RQQ equivalent by a typical multiplicative factor of 1.8–2.7, suggesting roughlyequal contributions from disk-corona and jet-linked X-ray emission.If we further presume that the X-ray absorber in BAL RLQs has characteristics similarto those found for BAL RQQs, then based on Figure 10c the disk-corona emission ought tobe reduced by a factor of ∼
10 in BAL RLQs relative to non-BAL RLQs. If the jet-linkedX-ray emission were also absorbed to a similar degree, the entire X-ray continuum in BALRLQs would be veiled as in BAL RQQs and the relative X-ray luminosities and hardnessratios of BAL RLQs would agree better with those of BAL RQQs, contrary to observation.However, it seems that the jet must be partially covered by the BAL-linked X-ray absorberin order to explain the difference between predicted and observed jet-linked X-ray emissionin BAL RLQs. Specifically, we find that many BAL RLQs have jet-linked X-ray emissiononly 20–80% of that expected, and further those BAL RLQs with hardness ratios harderthan 90% of RLQs tend to have less jet-linked X-ray emission than predicted (Figure 13b).We postulate that the BAL-linked X-ray absorber is of sufficient size to cover some fractionof the X-ray emitting jet in many BAL RLQs.Alternative scenarios are possible; while we cannot rule them out, they are difficultto motivate either physically or from the data. It might be surmised that BAL RLQsare intrinsically X-ray weak relative to RLQs and are also (typically) unabsorbed. TheX-ray absorber is thought to shield the BAL wind from overionization in disk-wind models(e.g., Murray et al. 1995), so such a postulated lack of an X-ray absorber could requireBAL formation and acceleration to occur in a manner distinct from that in BAL RQQs.One mechanism by which BAL RLQs could be intrinsically X-ray weak would be if thepresence of a BAL outflow inhibited the production of small-scale jet-linked X-ray emission. 22 –If the disk/corona system were relatively similar to that of RQQs, then BAL RLQs shouldfollow the non-BAL RQQ luminosity correlations; however, BAL RLQs with lower radioluminosities have X-ray luminosities less than those of non-BAL RQQs with comparableoptical/UV luminosities (Figures 9a and 13a). Another mechanism by which BAL RLQscould be intrinsically X-ray weak would be for the disk/corona system to be an inefficientemitter of X-rays. If the small-scale X-ray emitting jet were relatively similar to that inRLQs, then as the fractional contribution in RLQs from the disk/corona would be expectedto decrease at high radio luminosities, the difference in X-ray luminosity between BALRLQs and RLQs should likewise decrease; however, the offset between BAL RLQs andRLQs appears roughly constant (Figures 9c and 9d) across the two orders of magnitudein radio luminosity spanned by our sample. In any event, it seems most straight-forwardto retain those fundamental features firmly established as present in BAL RQQs or RLQswhen interpreting BAL RLQs, and the simple model associated with Figure 13 and describedabove suffices to explain the current data naturally.
There have been suggestions that BAL outflows occur at low inclinations (e.g., Broth-erton et al. 2006; Zhou et al. 2006), something difficult to explain from simple disk-windmodels. Some of the BAL RLQs in our sample have radio properties consistent with thejet being pointed close to the line of sight, including compact morphologies and flat radiospectra, although GPS sources with sparsely sampled radio spectra can mimic such charac-teristics at a range of inclinations. It does not seem likely that outflows in BAL RLQs mustalways be polar, since some of the BAL RLQs in our sample are steep-spectrum objects dom-inated by extended radio emission, arguing against low inclinations for these objects. We canestimate the inclinations of the core-dominated BAL RLQs using the core-radio-to-opticalluminosity ratio (essentially the radio loudness, excluding lobe emission; Wills & Brotherton1995), and find probable inclinations of ∼ ◦ to > ◦ , but this method is insensitive tolarger inclinations and may not apply to BAL RLQs. The sample BAL RLQs do not tendto have large values of R ∗ ( > ∼ Not all GPS or CSS sources display BALs, and not allBAL RLQs are associated with young sources; if there is no inclination dependence to BALsthen (as assessed by Shankar et al. 2008) strictly evolutionary models require problematicfine tuning of the various phases to match observations. It is possible to compare directly theX-ray properties of GPS and CSS sources to those of BAL RLQs to search for similarities.We plot data from Chandra observations of GPS and CSS radio galaxies and RLQs carriedout by Siemiginowska et al. (2008) on the l x ( l uv ) and l x ( l r ) relations shown earlier (Figure 14).GPS/CSS sources are often X-ray weak relative to RLQs of similar optical/UV luminosity,but they are also often extremely radio-loud (see also the spectral energy distribution plotsfrom Siemiginowska et al. 2008) in a manner that the (non-GPS, non-CSS) BAL RLQs arenot. The CSS BAL RLQ J104834.24+345724.9 is both radio luminous and X-ray weak, witha hard X-ray spectrum atypical of GPS/CSS sources, as might be expected for an object inboth classes.
6. Conclusions
This work presents and discusses the X-ray properties of 21 BAL RLQs observed with
Chandra . The sample of BAL RLQs spans a wide range of C IV absorption properties, isdominated by high-ionization BAL quasars, is restricted to definitively radio-loud quasarswith R ∗ > ∼
50, and includes objects with both core-dominated and lobe-dominated radio mor-phologies. We find the following results:1. BAL RLQs are X-ray weak relative to non-BAL RLQs of similar optical/UV lumi-nosity, but not to as extreme a degree as are BAL RQQs relative to comparable non-BALRQQs. BAL RLQs are also X-ray weak, to a lesser extent, relative to non-BAL RLQs ofsimilar radio luminosity or of both similar optical/UV and radio luminosities.2. BAL RLQs do not show a strong correlation between X-ray spectral hardness andX-ray weakness, as is observed in BAL RQQs, and do not tend to have as extreme hardnessratios as can BAL RQQs.3. The simplest model to explain our results is that the X-ray continuum in BAL RLQsconsists of both disk/corona and jet-linked X-ray emission; absorption of the disk/corona The relatively high fraction of low-ionization BAL quasars among dust-reddened quasars has also moti-vated the association of (at least low-ionization) BALs with young quasars (e.g., Urrutia et al. 2009). As in our sample, BAL RLQs can be found in FR IIs with large projected sizes, although the rarity ofsuch objects is interpreted by Gregg et al. (2006) as support for an evolutionary scenario.
24 –emission alone typically will neither reduce the observed X-ray luminosity nor harden theX-ray spectrum of BAL RLQs to the same degree as in BAL RQQs.4. Although jet-linked X-ray emission in BAL RLQs does not generally appear to beabsorbed to the same degree as is the X-ray continuum in BAL RQQs, it does seem likelythat the X-ray emitting small-scale jet is partially covered in many BAL RLQs.Microquasar observations have been interpreted to show that in the soft state a radiatively-driven disk wind develops and becomes the dominant channel for outflow of accreting ma-terial, quenching the jet (Neilsen & Lee 2009). Although the dearth of BALs in stronglyradio-loud objects suggests a similar mechanism may apply to quasars, it is clear that jetsand winds can coexist in at least some RLQs. Further X-ray studies can help clarify the re-lationship between jets and outflows in RLQs: snapshot
Chandra observations of additionalBAL RLQs could permit more quantitative consideration of various physical models, whiledeep
XMM-Newton spectral observations of the brightest BAL RLQs would help elucidatethe properties of the X-ray absorber and perhaps differentiate them from those in BALRQQs.We gratefully acknowledge the financial support of NASA grant SAO SV4-74018 (G. P. G.,Principal Investigator) and NASA LTSA grant NAG5-13035 (B. P. M., W. N. B.). We thankthe referee for useful comments, Mike Eracleous for helpful discussions as well as assistancewith HET/LRS data reduction and analysis, Jianfeng Wu for technical advice, and ChrisWillot and Bob Becker/Rick White for providing us with electronic spectra of 4C +35.23and FBQS 0256 − REFERENCES
Anderson, S. F., Weymann, R. J., Foltz, C. B., & Chaffee, F. H., Jr. 1987, AJ, 94, 278Arav, N., Kaastra, J., Steenbrugge, K., Brinkman, B., Edelson, R., Korista, K. T., & de Kool, M.2003, ApJ, 590, 174Avni, Y., & Tananbaum, H. 1986, ApJ, 305, 83Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ, 450, 559Becker, R. H., Gregg, M. D., Hook, I. M., McMahon, R. G., White, R. L., & Helfand, D. J. 1997,ApJ, 479, L93Becker, R. H., White, R. L., Gregg, M. D., Brotherton, M. S., Laurent-Muehleisen, S. A., & Arav,N. 2000, ApJ, 538, 72Becker, R. H., et al. 2001, ApJS, 135, 227Benn, C. R., Carballo, R., Holt, J., Vigotti, M., Gonz´alez-Serrano, J. I., Mack, K.-H., & Perley,R. A. 2005, MNRAS, 360, 1455Brandt, W. N., Laor, A., & Wills, B. J. 2000, ApJ, 528, 637Brotherton, M. S., van Breugel, W., Smith, R. J., Boyle, B. J., Shanks, T., Croom, S. M., Miller,L., & Becker, R. H. 1998, ApJ, 505, L7Brotherton, M. S., Laurent-Muehleisen, S. A., Becker, R. H., Gregg, M. D., Telis, G., White, R. L.,& Shang, Z. 2005, AJ, 130, 2006Brotherton, M. S., De Breuck, C., & Schaefer, J. J. 2006, MNRAS, 372, L58Condon, J. J., Cotton, W. D., Greisen, E. W., Yin, Q. F., Perley, R. A., Taylor, G. B., & Broderick,J. J. 1998, AJ, 115, 1693Douglas, J. N., Bash, F. N., Bozyan, F. A., Torrence, G. W., & Wolfe, C. 1996, AJ, 111, 1945Foltz, C. B., Chaffee, F. H., Jr., Hewett, P. C., MacAlpine, G. M., Turnshek, D. A., Weymann,R. J., & Anderson, S. F. 1987, AJ, 94, 1423Gallagher, S. C., Brandt, W. N., Chartas, G., & Garmire, G. P. 2002, ApJ, 567, 37Gallagher, S. C., Richards, G. T., Hall, P. B., Brandt, W. N., Schneider, D. P., & Vanden Berk,D. E. 2005, AJ, 129, 567Gallagher, S. C., Brandt, W. N., Chartas, G., Priddey, R., Garmire, G. P., & Sambruna, R. M.2006, ApJ, 644, 709
26 –
Gallagher, S. C., Hines, D. C., Blaylock, M., Priddey, R. S., Brandt, W. N., & Egami, E. E. 2007,ApJ, 665, 157Garmire, G. P., Bautz, M. W., Ford, P. G., Nousek, J. A., & Ricker, G. R., Jr. 2003, Proc. SPIE,4851, 28Gehrels, N. 1986, ApJ, 303, 336Gibson, R. R., Brandt, W. N., & Schneider, D. P. 2008, ApJ, 685, 773Gibson, R. R., Brandt, W. N., Schneider, D. P., & Gallagher, S. C. 2008, ApJ, 675, 985Gibson, R. R., et al. 2009, ApJ, 692, 758Gopal-Krishna, & Wiita, P. J. 2000, A&A, 363, 507Green, P. J., Aldcroft, T. L., Mathur, S., Wilkes, B. J., & Elvis, M. 2001, ApJ, 558, 109Gregg, M. D., Becker, R. H., Brotherton, M. S., Laurent-Muehleisen, S. A., Lacy, M., & White,R. L. 2000, ApJ, 544, 142Gregg, M. D., Becker, R. H., & de Vries, W. 2006, ApJ, 641, 210Gregory, P. C., Scott, W. K., Douglas, K., & Condon, J. J. 1996, ApJS, 103, 427Hall, P. B., et al. 2002, ApJS, 141, 267Hall, P. B., Gallagher, S. C., Richards, G. T., Alexander, D. M., Anderson, S. F., Bauer, F., Brandt,W. N., & Schneider, D. P. 2006, AJ, 132, 1977Hewett, P. C., & Foltz, C. B. 2003, AJ, 125, 1784Hill, G. J., Nicklas, H. E., MacQueen, P. J., Tejada, C., Cobos Duenas, F. J., & Mitsch, W. 1998,Proc. SPIE, 3355, 375Ivezi´c, ˇZ., et al. 2002, AJ, 124, 2364Jester, S., Harris, D. E., Marshall, H. L., & Meisenheimer, K. 2006, ApJ, 648, 900Jiang, D. R., & Wang, T. G. 2003, A&A, 397, L13Jin, Y. K., Zhang, S. N., & Wu, J. F. 2006, ApJ, 653, 1566Just, D. W., Brandt, W. N., Shemmer, O., Steffen, A. T., Schneider, D. P., Chartas, G.,& Garmire,G. P. 2007, ApJ, 665, 1004Kelly, B. C. 2007, ApJ, 665, 1489
27 –
Kelly, B. C., Bechtold, J., Siemiginowska, A., Aldcroft, T., & Sobolewska, M. 2007, ApJ, 657, 116Kraft, R. P., Burrows, D. N., & Nousek, J. A. 1991, ApJ, 374, 344Kunert-Bajraszewska, M., & Marecki, A. 2007, A&A, 469, 437Lacy, M., Gregg, M., Becker, R. H., White, R. L., Glikman, E., Helfand, D., & Winn, J. N. 2002,AJ, 123, 2925Laor, A., & Brandt, W. N. 2002, ApJ, 569, 641Lavalley, M., Isobe, T., & Feigelson, E. 1992, Astronomical Data Analysis Software and Systems I,25, 245Liu, Y., Jiang, D. R., Wang, T. G., & Xie, F. G. 2008, MNRAS, 391, 246Ma, F. 2002, MNRAS, 335, L99Menou, K., et al. 2001, ApJ, 561, 645Miller, B. P., Brandt, W. N., Gallagher, S. C., Laor, A., Wills, B. J., Garmire, G. P., & Schneider,D. P. 2006, ApJ, 652, 163Montenegro-Montes, F. M., Mack, K.-H., Vigotti, M., Benn, C. R., Carballo, R., Gonz´alez-Serrano,J. I., Holt, J., & Jim´enez-Luj´an, F. 2008, MNRAS, 388, 1853Montenegro-Montes, F. M., Mack, K. -., Benn, C. R., Carballo, R., Dallacasa, D., Gonz´alez-Serrano,J. I., Holt, J., & Jim´enez-Luj´an, F. 2009, arXiv:0903.5119Murray, N., Chiang, J., Grossman, S. A., & Voit, G. M. 1995, ApJ, 451, 498Neilsen, J., & Lee, J. C. 2009, Nature, 458, 481O’Dea, C. P. 1998, PASP, 110, 493Ogle, P. M., Cohen, M. H., Miller, J. S., Tran, H. D., Goodrich, R. W., & Martel, A. R. 1999,ApJS, 125, 1Page, K. L., Reeves, J. N., O’Brien, P. T., & Turner, M. J. L. 2005, MNRAS, 364, 195Perley, R. A., Condon, J. J., Cotton, W. D., Cohen, A. S., Lane, W. M., Kassim, N. E., Lazio,T. J. W., & Erickson, W. C. 2006, VizieR Online Data Catalog, 8079, 0Proga, D., Stone, J. M., & Kallman, T. R. 2000, ApJ, 543, 686Ramsey, L. W., et al. 1998, Proc. SPIE, 3352, 34
28 –
Reeves, J. N., & Turner, M. J. L. 2000, MNRAS, 316, 234Reichard, T. A., et al. 2003, AJ, 126, 2594Rengelink, R. B., Tang, Y., de Bruyn, A. G., Miley, G. K., Bremer, M. N., Roettgering, H. J. A.,& Bremer, M. A. R. 1997, A&AS, 124, 259Sambruna, R. M., Gliozzi, M., Tavecchio, F., Maraschi, L., & Foschini, L. 2006, ApJ, 652, 146Schaefer, J. J., Brotherton, M. S., Shang, Z., Gregg, M. D., Becker, R. H., Laurent-Muehleisen,S. A., Lacy, M., & White, R. L. 2006, AJ, 132, 1464Schneider, D. P., et al. 2007, AJ, 134, 102Shankar, F., Dai, X., & Sivakoff, G. R. 2008, ApJ, 687, 859Siemiginowska, A., LaMassa, S., Aldcroft, T. L., Bechtold, J., & Elvis, M. 2008, ApJ, 684, 811Stawarz, L., Ostorero, L., Begelman, M. C., Moderski, R., Kataoka, J., & Wagner, S. 2008, ApJ,680, 911Steffen, A. T., Strateva, I., Brandt, W. N., Alexander, D. M., Koekemoer, A. M., Lehmer, B. D.,Schneider, D. P., & Vignali, C. 2006, AJ, 131, 2826Stocke, J. T., Morris, S. L., Weymann, R. J., & Foltz, C. B. 1992, ApJ, 396, 487Strateva, I. V., Brandt, W. N., Schneider, D. P., Vanden Berk, D. G., & Vignali, C. 2005, AJ, 130,387Trump, J. R., et al. 2006, ApJS, 165, 1Urrutia, T., Lacy, M., Gregg, M. D., & Becker, R. H. 2005, ApJ, 627, 75Urrutia, T., Lacy, M., & Becker, R. H. 2008, ApJ, 674, 80Urrutia, T., Becker, R. H., White, R. L., Glikman, E., Lacy, M., Hodge, J., & Gregg, M. D. 2009,ApJ, 698, 1095Vanden Berk, D. E., et al. 2001, AJ, 122, 549de Vries, W. H., Becker, R. H., & White, R. L. 2006, AJ, 131, 666Wang, J., Jiang, P., Zhou, H., Wang, T., Dong, X., & Wang, H. 2008, ApJ, 676, L97Weymann, R. J., Morris, S. L., Foltz, C. B., & Hewett, P. C. 1991, ApJ, 373, 23White, R. L., Becker, R. H., Helfand, D. J., & Gregg, M. D. 1997, ApJ, 475, 479
29 –
White, R. L., et al. 2000, ApJS, 126, 133Wilkes, B. J., & Elvis, M. 1987, ApJ, 323, 243Willott, C. J., Rawlings, S., Archibald, E. N., & Dunlop, J. S. 2002, MNRAS, 331, 435Willott, C. J., Rawlings, S., & Grimes, J. A. 2003, ApJ, 598, 909Wills, B. J., & Brotherton, M. S. 1995, ApJ, 448, L81Wills, B. J., Brandt, W. N., & Laor, A. 1999, ApJ, 520, L91Worrall, D. M., Tananbaum, H., Giommi, P., & Zamorani, G. 1987, ApJ, 313, 596Wu, J. F., Zhang, S. N., Lu, F. J., & Jin, Y. K. 2007, Chinese Journal of Astronomy and Astro-physics, 7, 81York, D. G., et al. 2000, AJ, 120, 1579Zhou, H., Wang, T., Wang, H., Wang, J., Yuan, W., & Lu, Y. 2006, ApJ, 639, 716
This preprint was prepared with the AAS L A TEX macros v5.2.
30 –Table 1.
Chandra and
HET
Observing Log
Chandra
Observations
HET
ObservationsName (SDSS) ObsID Date Exp (s) Counts a Date Exp (s) λ/ ∆ λ b S/N c +9 . − . +5 . − . +5 . − . +7 . − . · · · · · · · · · · · · +6 . − . +7 . − . − +10 . − . · · · · · · · · · · · · +14 . − . +5 . − . − +5 . − . +8 . − . +5 . − . Chandra on-axis with the ACIS-S array, using Very Faint mode.
HET spectra wereobtained with the Low Resolution Spectrograph; 10/12 targets were able to be observed. a Background-subtracted and aperture-corrected counts in the 0.5–8 keV band. Errors are 1 σ (Poisson errors; Gehrels 1986).All snapshot targets are detected. b Most observations were conducted using the g2 grism with a 1.5 ′′ slit, providing a resolving power of 867. c Signal-to-noise of the continuum near observed-frame 6000 ˚A.
31 –Table 2.
Chandra
Archival Sources
Name (SDSS) Name (Other) ObsID Date Exp (s) θ ( ′ ) a Counts b Sel c Ref020022.01 − − +8 . − . S 1 · · ·
FBQS J0256 − +5 . − . L 2081426.45+364713.5 . . . . . . . . . . . . . . . . . . . . . . . . . 3436 2002 Jan 31 9839 8.3 8.8 +4 . − . S · · · d . . . . . . . . . . . . . . . . . . . . . . . . . 3437 2002 Feb 11 9933 8.3 9.5 +4 . − . S · · · < . · · · +44 . − . L 3104834.24+345724.9 4C +35.23 . . . . . . . . . . . . . . 9320 2008 Jan 20 4658 0.0 5.5 +3 . − . L 4122033.87+334312.0 3C 270.1 . . . . . . . . . . . . . . . . 2118 2002 Apr 03 3087 4.3 177.4 +14 . − . S · · · +8 . − . L 2 · · ·
LBQS 2211 − +8 . − . L 5 a Off-axis angle in arc-minutes; a value of 0.0 indicates an observation targeting that BAL RLQ. b Background-subtracted and aperture-corrected counts in the 0.5–8 keV band. Errors are 1 σ (Poisson errors; Gehrels 1986),while limits for non-detections are at the 95% confidence level (Bayesian statistics; Kraft et al. 1991). c Selection method: S = BAL RLQ identified from SDSS/FIRST data with serendipitous
Chandra archival coverage, whileL = BAL RLQ identified from literature with targeted
Chandra archival coverage. d Since the two observations of 081426.45+364713.5 are of comparable quality, both are shown; these observations are stackedfor later analysis. There are 18.3 +5 . − . net 0.5–8 keV counts in the combined 19.8 ks exposure.References. — Prior analysis of Chandra data: (1) Gallagher et al. (2005); (2) Brotherton et al. (2005); Miller et al. (2006);(4) PI Kunert-Bajraszewska; (5) Gallagher et al. (2006).
32 –Table 3. Optical/UV Characteristics
Name (SDSS) z m i M i ∆( g − i ) BI AI EW V max Type a RefSnapshot BAL RLQs074610.50+230710.8. . . . . . . . . . . . 2.093 18.27 − · · · b − − − − · · · − − − − − − − − − b − − − − · · · b b H 4081426.45+364713.5. . . . . . . . . . . . 2.732 19.81 − − − − · · · b − · · · · · · b b Hi 6122033.87+334312.0. . . . . . . . . . . . 1.532 18.10 − · · · − · · · b − − · · · k -correction for M i assumes a power-law continuum with spectral index α ν = − .
5. The absorption propertiesrefer to C IV measurements. BI, EW, and V max values are primarily from the listed reference, chiefly Gibson et al. (2009),while AI values are taken from Trump et al. (2006) where available. The units for BI, AI, and V max are km s − ; EW is in ˚A. a BAL type following Trump et al. (2006): Hi = HiBAL (no Mg II absorption in spectrum); HLF = FeLoBAL (C IV BAL,Fe II or Fe III absorption in spectrum); HL = HiBAL with some low-ionization absorption; H = HiBAL lacking spectralcoverage of Mg II; n = relatively narrow absorption. Type is taken from Trump et al. (2006) for all quasars with reportedAI measurements; the remainder are classified based on our examination of the SDSS spectrum where available or else by thelisted reference. b EW or V max value measured by us.References. — (1) Trump et al. (2006); (2) Gibson et al. (2009); Benn et al. (2005); (4) Becker et al. (2001); (5) Wills etal. (1999); (6) Willott et al. (2002); (7) Becker et al. (2000); (8) Weymann et al. (1991); (9) Gallagher et al. (2006).
33 –Table 4. Radio Characteristics
Flux-Density Measurements a Spectral Indices b Name (SDSS) Type c ≤
365 MHz 1.4 GHz (F) 1.4 GHz (N) 4.85 GHz α low α high Snapshot BAL RLQs074610.50+230710.8 . . . . . . P · · · ± ± · · · +0.11083749.59+364145.4 . . . . . . P · · · · · · · · · · · · − · · · ± ± · · · +0.30092913.96+375742.9 . . . . . . P 94 W 43.43 42.9 ± ± − − ±
51 T 118.65 d ± ± − − ± ± − − − ±
130 V 65.47 d ± · · · − · · · ±
30 T 268.48 275.6 ± ± − − · · · ± · · · − − · · · ± · · · − ± ± − − ± ± − − − · · · ± · · · · · · · · · FBQS J0256 − · · · ± ± · · · − · · · ± · · · · · · · · · · · · ± · · · · · · · · · · · · d ± d ±
37 G − − ±
29 T 1050.97 1034.4 ± ±
39 G − − ±
134 T 2819.00 d ± ±
75 G − − · · · ± ± · · · − − · · · · · · ± · · · · · · · · · a All flux density measurements are in mJy, taken from the following sources: B = Benn et al. (2005); C = archival VLAC-band imaging; F = FIRST: Faint Images of the Radio Sky at Twenty cm, integrated flux, RMS errors are ≃ − (White et al. 1997); G = Green Bank 6-cm survey (Gregory et al. 1996); M = Montenegro-Montes et al. (2008); N = NVSS:NRAO VLA Sky Survey (Condon et al. 1998); P = Parkes Catalogue 1990, 408 MHz; T = Texas Survey of Radio Sources at365 MHz (Douglas et al. 1996); V = VLA Low-Frequency Sky Survey, 74 MHz (Perley et al. 2006); W = Westerbork NorthernSky Survey, 326 MHz, RMS errors are ≃ − (Rengelink et al. 1997). b Radio spectral indices include extended emission components, are given as S r ∝ ν α r , and use FIRST measurements whereavailable (else NVSS measurements); the quantities α low and α high are calculated from the flux densities presented in thecolumns labeled ≤
365 MHz and 4.85 GHz, respectively, in addition to the 1.4 GHz data. J083749.59+364145.4 has a brightunrelated radio source ∼ ′′ North of the core that contaminates low-resolution maps; the spectral index is from a high-resolution8.45 GHz flux density measurement from Montenegro-Montes et al. (2008). c Radio morphology: P = point source, D = double (lobes summed for flux measurements). See § . d Extended emission: J102258.41+123429.7 has two FIRST components offset from the SDSS position by 0.058 ′ and 0.174 ′ ,with integrated fluxes of 93.98 and 24.67 mJy, respectively; J112506.95 − ′ and 0.201 ′ , with integrated fluxes of 55.36 and 10.11 mJy, respectively; J100726.10+124856.2(PG 1004+130) is over-resolved by FIRST, but has two NVSS components offset from the SDSS position by 0.434 ′ and 1.010 ′ ,with fluxes of 656.4 and 559.7 mJy, respectively; J122033.87+334312.0 (3C 270.1) has two FIRST components offset from theSDSS position by 0.067 ′ and 0.073 ′ , with integrated fluxes of 2096.61 and 722.39 mJy, respectively.
34 –Table 5. X-ray Counts, Luminosities, and Properties of BAL RLQs
X-ray Counts a Luminosities b Derived Properties c Name (SDSS) Soft Hard HR Rate l r l uv l x R ∗ α ox ∆ l x , uv ∆ l x , S ∆ l x , H Snapshot BAL RLQs074610.50+230710.8 49.4 +8 . − . +5 . − . − . +0 . − . +1 . − . − − − − +4 . − . < . < − .
59 2.9 +0 . − . − − − < − +4 . − . +3 . − . − . +0 . − . +0 . − . − − − − +6 . − . +5 . − . − . +0 . − . +2 . − . − − − − +5 . − . +3 . − . − . +0 . − . +1 . − . − − − − +6 . − . +4 . − . − . +0 . − . +1 . − . − − − − − +8 . − . +6 . − . − . +0 . − . +2 . − . − +10 . − . +10 . − . − . +0 . − . +3 . − . − − − +4 . − . +3 . − . − . +0 . − . +0 . − . − − − − − +4 . − . +4 . − . . +0 . − . +1 . − . − − − − +6 . − . +5 . − . − . +0 . − . +1 . − . − − − − +4 . − . +3 . − . − . +0 . − . +1 . − . − − − − − +7 . − . +4 . − . − . +0 . − . +0 . − . − − − − − +5 . − . +3 . − . − . +0 . − . +1 . − . − − − − +4 . − . +3 . − . − . +0 . − . +0 . − . − − − − < . < . · · · < . < < − < − < − < − +34 . − . +29 . − . − . +0 . − . +1 . − . − − − − < . +3 . − . > .
16 1.2 +0 . − . − − < − − +12 . − . +7 . − . − . +0 . − . +4 . − . − − − +8 . − . +4 . − . − . +0 . − . +1 . − . − − − − − +7 . − . +5 . − . − . +0 . − . +1 . − . − − − − a The soft and hard bands are 0.5–2 keV and 2–8 keV, respectively; errors/limits are as in Tables 1 and 2. The hardness ratiois HR = ( H − S ) / ( H + S ), where S ( H ) is the soft (hard) band counts; errors are 1 σ . The HR values for J020022.01 − HR values. Rate is counts ks − in the 0.5–8 keV band. b These monochromatic luminosities have units of log ergs s − Hz − , at rest-frame frequencies of 5 GHz, 2500 ˚A, and 2 keVfor l r , l uv , and l x , respectively. l x is calculated from the 0.5–8 keV count rates for a power-law of Γ = 1 . c The radio loudness (in log units) is R ∗ = l r − l uv and the optical/UV-to-X-ray spectral slope is α ox = 0 . × ( l x − l uv ).The relative X-ray luminosities are ∆ l x , uv = l x − (0 . × l uv − . l x , S / H = l x , S / H − (0 . × l uv + 0 . × l r − . l x , S and l x , H are 2 keV X-ray luminosities calculated from the 0.5–2 keV and 2–8 keV count rates, respectively.
35 –Fig. 1.—
Histogram showing the SDSS m i distribution for radio-loud broad absorption line quasarssatisfying R ∗ > ∼
100 and EW
CIV > Chandra coverage (dark gray, 8objects shown; PG 1004+130 has m i = 15 .
36 –Fig. 2.— (a) Radio loudness plotted versus broad absorption line strength (parameterized byC IV EW). The
Chandra snapshot BAL RLQs are shown as diamonds and the archival BALRLQs as squares (nested symbols are lobe-dominated BAL RLQs). The solid line marks the R ∗ >
10 (log R ∗ >
1) boundary, below which quasars are defined to be radio-quiet. The dashedlines show the selection criteria for the archival sample of BAL RLQs, which were required to bedefinitively radio-loud ( R ∗ > ∼
50) and show strong broad absorption lines (
EW > R ∗ > ∼
100 that was used to select the snapshot sample.As reported by previous authors it is rare for quasars to be simultaneously strongly absorbedand strongly radio-loud. (b) Plot of the distribution of C IV EW for objects with R ∗ <
10 (openhistogram), objects with R ∗ >
10 (gray histogram), and objects with R ∗ >
50 and
EW > R ∗ >
10 (the gray and black histograms) have been multipliedby 5 for clarity.
37 –Fig. 3.—
SDSS spectra for the snapshot sample, plotted with rest-frame wavelengths and showingthe L α to C IV BAL region. The dashed line in each panel is at 1549 ˚A, or zero velocity. The dottedline indicates the maximum outflow velocity, primarily taken from Gibson et al. (2009). Flux isgiven in units of 10 − erg cm − s − ˚A − . The sample includes objects covering a range of BALabsorption strengths and outflow velocities. Each panel is labeled with the SDSS DR5 name as wellas the absorption index (a measure of BAL strength), primarily taken from Trump et al. (2006).
38 –Fig. 4.—
Relative color ∆( g − i ), calculated by taking the measured ( g − i ) for a given object andsubtracting the ( g − i ) that is typical for quasars at that redshift (positive values correspond toredder objects). The top panel shows the snapshot and archival sample of BAL RLQs with Chandra coverage (the object shaded in gray is the strongly reddened object FBQS J1556+3517, markedfor comparison but not included in the archival sample). The middle panel shows SDSS/FIRSTBAL RLQs, and the bottom panel shows BAL RQQs. The dashed lines enclose 90% of SDSS DR5quasars. While BAL quasars tend to be redder than non-BAL quasars, the BAL RLQs studied with
Chandra are representative of BAL RLQs in general and do not show excessive intrinsic reddeningthat could significantly elevate radio-loudness values.
39 –Fig. 5.—
Optical luminosity l uv in units of log ergs s − Hz − calculated at rest-frame 2500 ˚A for oursample of BAL RLQs (nested symbols are lobe-dominated BAL RLQs) and for comparison samplesof non-BAL RLQs, BAL RQQs, and non-BAL RQQs, plotted versus redshift. The RLQ compari-son sample is constructed from SDSS/FIRST/ Chandra data, supplemented with some particularlyluminous RLQs observed by
Einstein (Worrall et al. 1987). The BAL RQQ comparison sampleis taken from Gibson et al. (2009) and is supplemented with non-SDSS objects from Gallagher etal. (2006). The RQQ comparison sample is taken from Gibson et al. (2008a) and is supplementedwith luminous RQQs from Just et al. (2007). Quasars from the DR5 Quasar Catalog of Schneideret al. (2007) are shown as gray points. Our
Chandra snapshot sample is biased toward luminousquasars as a consequence of the magnitude-limited selection method. The comparison samples havebeen constructed to overlap and bracket the BAL RLQs in optical luminosity and redshift.
40 –Fig. 6.—
HET/LRS spectra (red lines) taken near the times of the
Chandra snapshot observations,shown compared to the earlier epoch SDSS spectra (blue lines) matched to HET/LRS resolution.Each panel is labeled with the SDSS DR5 name as well as the rest-frame interval between observa-tions (in days). The horizontal axis is velocity in 1000 km s − and the vertical axis is normalizedflux. The C IV absorption regions are shaded gray. There is only minor BAL variability seenin these objects, indicating that variability does not significantly complicate a comparison of UVabsorption properties to X-ray weakness.
41 –
E N30" E N 30"
Fig. 7.—
Chandra images of the two BAL RLQs in the snapshot survey possessing extended radioemission. The left panel shows J102258.41+123429.7; the right panel shows J112506.95 − ′′ is ≃
250 kpc. Adaptively smoothed 0.5–8 keV images are plotted in grayscalewith logarithmic scaling, overlaid with contours from the 5 GHz FIRST survey at levels of 2, 8,and 32 mJy beam − . Peak fluxes for the radio sources are <
85% of the integrated fluxes and thedeconvolved major axes are ∼ − ′′ ; FIRST apparently resolves these components. The crossesmark SDSS photometric sources within the field; none of these aligns with the apparent extendedradio emission, further indicating that these are lobes rather than unrelated sources.
42 –Fig. 8.—
Chandra spectrum of J115944.82+011206.9, the X-ray brightest source in our snapshotsample with ≃
170 counts from 0.3–8 keV ( ∼ × that of the next-brightest snapshot BAL RLQ).The plotted model has intrinsic absorption with column density N H = 3 . +2 . − . × cm − and apower-law photon index of Γ = 1 . +0 . − . . The fit was performed using the cstat statistic and thecosmetic binning is based on a minimum significance of 3 σ within a maximum of 30 bins. Theinset shows the 68, 90, and 99% confidence contours (for two parameters of interest) for the photonindex Γ and intrinsic column density N H .
43 –Fig. 9.—
X-ray luminosities of BAL quasars compared to similar non-BAL quasars. Luminositieshave units of log ergs s − Hz − , at rest-frame frequencies of 5 GHz, 2500 ˚A, and 2 keV for l r , l uv ,and l x , respectively. Arrows indicate X-ray upper limits, and nested symbols are lobe-dominatedBAL RLQs. Blue lines are best-fit correlations for non-BAL RLQs (taking l x as the dependentvariable) calculated using the Bayesian maximum-likelihood method of Kelly (2007). The solidgreen line shows the best-fit correlation for non-BAL RQQs that Just et al. (2007) calculated withASURV for a large sample of RQQs; fitting our comparison sample of RQQs yields a similar result(dotted green line). The l x ( l uv ) relation for RLQs/RQQs is also plotted as a dashed line in (b)/(a),illustrating the well-known tendency for RLQs to be X-ray brighter than comparable RQQs. BALRLQs are X-ray weak relative to non-BAL RLQs with similar optical/UV luminosities (a) but notto the same degree as are BAL RQQs relative to non-BAL RQQs (b). BAL RLQs are also modestlyX-ray weak relative to non-BAL RLQs with similar radio luminosities (c) and to non-BAL RLQswith both similar optical/UV and radio luminosities (d).
44 –Fig. 10.—
Histograms showing the distribution of the difference between actual and anticipatedX-ray luminosity. Arrows indicate X-ray limits. Dotted histograms show subsamples: archival BALRLQs (black), Worrall et al. (1987) RLQs (cyan), Gallagher et al. (2006) BAL RLQs (magenta),and Just et al. (2007) RQQs (light green). The left column (a, b, c, d) shows ∆ l x , uv calculatedfrom optical/UV luminosities using the relations shown in Figures 9a and 9b. BAL RQQs reachmore extreme values of X-ray weakness relative to non-BAL RQQs than do BAL RLQs relativeto non-BAL RLQs with similar optical/UV luminosities. The right column (e, f, g, h) shows ∆ l x calculated from both optical/UV and radio luminosities for RLQs, using the relation shown inFigure 9d. X-ray luminosity is calculated using the full (0.5–8 keV), soft (0.5–2 keV), and hard (2–8 keV) counts. BAL RLQs are typically X-ray weaker than comparable non-BAL RLQs by a factorof 2.0–4.5. The similarity of results derived using full, soft, and hard X-ray luminosities suggestssimple absorption of the entire X-ray continuum source cannot provide a universal explanation forthe X-ray weakness in BAL RLQs.
45 –Fig. 11.—
C IV absorption properties as a function of relative X-ray luminosity, calculatedusing the relations shown in Figures 9a and 9b and expressed in terms of ∆ α ox ( l uv ), where α ox = 0 . × ( l x − l uv ), for ease of comparison with previous work. Panel (a) shows C IV EWand (b) shows maximum outflow velocity. Nested symbols are lobe-dominated BAL RLQs. Thepurple points in (a) are non-BAL RQQs from the BQS, with C IV absorption values from Brandt,Laor, & Wills (2000) and optical and X-ray luminosities from Steffen et al. (2006). X-ray weaknessappears more closely linked to absorption strength in BAL RQQs than in BAL RLQs; even BALRLQs with extreme C IV absorption properties do not have ∆ α ox < − .
5, as do many stronglyabsorbed BAL RQQs.
46 –Fig. 12.—
Hardness ratio plotted versus redshift for the sample of BAL RLQs (nested symbolsare lobe-dominated BAL RLQs) and for comparison samples of RLQs (a) and for BAL RQQs andRQQs (b). The X-ray spectra of BAL RLQs are sometimes harder than those of typical non-BALRLQs, but are often consistent. The X-ray spectra of BAL RQQs are often harder than for typicalnon-BAL RQQs, reaching extreme hardness ratios in some cases, and the distribution of hardnessratios for BAL RQQs is not statistically consistent with that of RQQs. Panels (c) and (d) showhardness ratio plotted versus ∆ l x , uv calculated using the relations shown in Figures 9a and 9b;the X-ray weakness of BAL RQQs is linked to increasing intrinsic absorption of the continuum(illustrated via harder X-ray spectra), while many X-ray weak BAL RLQs do not have extremelyhard X-ray spectra.
47 –Fig. 13.—
The top panel (a) shows the ratio of X-ray luminosity in RLQs to that of RQQs withcomparable optical/UV luminosities, expressed in log units, as a function of radio luminosity. Themedian factor by which the comparison sample of RLQs are X-ray brighter than the comparisonsample of RQQs is 3.4 (cyan dashed line); there is a trend (illustrated with the solid blue line)toward increasing X-ray brightness with increasing radio luminosity that likely reflects increasing jetdominance. BAL RLQs are plotted at their predicted (black) and observed (gray) X-ray luminosityratios (nested symbols are lobe-dominated BAL RLQs). The bottom panel (b) shows the fractionof jet-linked X-ray emission in BAL RLQs relative to that expected for non-BAL RLQs with similaroptical/UV and radio luminosities, assuming the disk/corona X-ray emission (predicted from theoptical/UV luminosity using the RQQ relation) is reduced by a factor of 10, as is typical for BALRQQs. Values for the fractional jet-linked emission near (or above) 1 would suggest the jet isunobscured, while values near 0.1 would indicate the jet is covered and reduced in intensity to asimilar degree as is the disk/corona emission. The median, 10th, and 90th percentile values ofhardness ratio for RLQs are also plotted (blue dashed and cyan dotted lines, respectively).
48 –Fig. 14.—
Comparison of the luminosities of GPS and CSS sources to those of BAL RLQs andnon-BAL RLQs. Data for GPS and CSS sources (red stars) are from Siemiginowska et al. (2008).The legend and caption for BAL RLQs and non-BAL RLQs is identical to Figures 9a and 9c for (a)and (b), respectively. The l x ( l uvuv