A multi-wavelength continuum characterization of high-redshift broad absorption line quasars
D. Tuccillo, G. Bruni, M. A. DiPompeo, M. S. Brotherton, A. Pasetto, A. Kraus, J. I. Gonzalez-Serrano, K.-H. Mack
MMon. Not. R. Astron. Soc. , 1– ?? (2011) Printed November 12, 2018 (MN L A TEX style file v2.2)
A multi-wavelength continuum characterization of high-redshiftbroad absorption line quasars
D. Tuccillo, , (cid:63) G. Bruni, M. A. DiPompeo, , M. S. Brotherton, A. Pasetto, A. Kraus, J. I. Gonz´alez-Serrano, and K.-H. Mack GEPI, Observatoire de , CNRS, Universit´e Diderot, 61, Avenue de l’Observatoire F-75014, , France Instituto de F´ısica de Cantabria (CSIC-Universidad de Cantabria), Avda. de los Castros s/n, E-39005 Santander, Spain Max Planck Institute for Radio Astronomy, Auf dem Huegel, 69, D-53121 Bonn, Germany Department of Physics and Astronomy, Dartmouth College, 6127 Wilder Laboratory, Hanover, NH 03755, USA Dep. of Physics and Astronomy 3905, University of Wyoming, 1000 East University, Laramie, WY 82071, USA INAF-Istituto di Radioastronomia, via Piero Gobetti, 101, I-40129 Bologna, Italy
Accepted for publication on MNRAS, 2017 February 6
ABSTRACT
We present the results of a multi-wavelength study of a sample of high-redshift Radio Loud(RL) Broad Absorption Line (BAL) quasars. This way we extend to higher redshift previousstudies on the radio properties, and broadband optical colors of these objects. We have se-lected a sample of 22 RL BAL quasars with . (cid:54) z (cid:54) . cross-correlating the FIRST radiosurvey with the SDSS. Flux densities between 1.25 and 9.5 GHz have been collected with theJVLA and Effelsberg-100m telescopes for 15 BAL and 14 non-BAL quasars used as compar-ison sample. We determine the synchrotron peak frequency, constraining their age. A largenumber of GigaHertz Peaked Spectrum (GPS) and High Frequency Peakers (HFP) sourceshas been found in both samples (80% for BAL and 71% for non-BAL QSOs), not suggestinga younger age for BAL quasars. The spectral index distribution provides information aboutthe orientation of these sources, and we find statistically similar distributions for the BAL andnon-BAL quasars in contrast to work done on lower redshift samples. Our sample may be toosmall to convincingly find the same effect, or might represent a real evolutionary effect basedon the large fraction of young sources. We also study the properties of broadband colors inboth optical (SDSS) and near- and mid-infrared (UKIDSS and WISE) bands, finding that alsoat high redshift BAL quasars tend to be optically redder than non-BAL quasars. However,these differences are no more evident at longer wavelength, when comparing colors of thetwo samples by mean of the WISE survey. Key words: methods: data analysis; catalogues: observations; quasars:general; quasars: ab-sorption lines; galaxies: high-redshift; galaxies: active
Intrinsic absorption lines are a common feature in quasar (QSO)spectra, and they are likely produced by outflowing winds along theline of sight that are launched from the accretion disk around thecentral supermassive black hole (Murray et al. 1995; Proga, Stone& Kallman 2000). Broad absorption lines (BALs) in ultraviolet andvisible spectra of quasars are the most spectacular manifestationof such outflows, with absorption velocity width larger than 2000 km / s (Weymann et al. 1991). BAL quasars are often classified intothree subtypes depending on the presence of absorption lines inspecified transitions: (1) High-ionization (Hi) BAL quasars showabsorption lines in high-ionization transitions such as Si IV , and (cid:63) E-mail:[email protected] C IV ; (2) Low-ionization (Lo) BAL quasars possess Mg II and/orAI III absorption lines, in addition to the high-ionization transi-tions. (3) Iron low-ionization (FeLoBAL) BAL quasars show ad-ditional absorption from excited states of Fe II and Fe III (e.g.,Hall et al. 2002). About 15% (Knigge et al. 2008) of the wholepopulation of quasars are classified as BAL, although some authors(Ganguly et al. 2007) suggest that the intrinsic fraction could bemuch higher.Powerful outflows can enrich the quasar host galaxy interstel-lar medium, contributing to a variety of feedback effects such asregulating host galaxy star formation rates (Hopkins & Elvis 2010),and limiting quasar lifetimes by removing fuel from the nuclear re-gions (Silk & Rees 1998). Understanding the physics behind BALoutflows may be fundamental to disclose the connection betweenthe evolution of quasars and their host galaxies. According to the c (cid:13) a r X i v : . [ a s t r o - ph . GA ] F e b D. Tuccillo et al. orientation scenario , BAL QSOs are normal quasars seen at highinclination angles, and the outflows at the origin of the BAL fea-tures are located near the equatorial plane of the accretion disk.This model fits naturally in the AGN unification models (Antonucci1993; Urry & Padovani 1995) and it is supported by the claim thatboth BAL and non-BAL QSOs appear to be drawn by the same par-ent population (Reichard et al. 2003). However, the orientation sce-nario is contradicted from the studies on the radio-loud (RL) BALquasars, studied for the first time in Brotherton et al. (1998) and inBecker et al. (2000). In fact an edge-on geometry should lead tolobe-dominated emission and thus steep radio spectrum but, usinglarge samples of RL BAL QSOs (Montenegro-Montes et al. 2008,Fine, Jarvis & Mauch 2011, DiPompeo et al. 2011, Bruni et al.2012), it has been confidently established that these objects havea variety of spectral indices for their radio Spectral Energy Distri-bution (SED hereafter). On the light of these results, a pure ori-entation scenario is clearly inadequate to explain the full extent ofthe BAL phenomena, especially considering that most of the con-tinuum and emission line properties of RL and RQ BALs appearstatistically identical and therefore suggest that the results obtainedon RL QSOs quasars can be extended to their RQ counterparts(Rochais et al. 2014). The other main model proposed to explainthe nature of BAL QSOs is evolutionary model as discussed for ex-ample by L´ıpari & Terlevich (2006). According to this, BAL QSOsrepresent a particular evolutionary stage of quasars, during whichabsorbing material with a high covering fraction is being expelledfrom the central regions of the quasar. The radio-loud systems maybe associated with the later stages of evolution (Becker et al. 1997),when jets have removed the clouds responsible for the generationof BALs. The early claim that BAL QSOs have redder continua hasbeen confirmed by a number of different studies in the UV/opticaldomain (Weymann et al. 1991; Sprayberry & Foltz 1992; Brother-ton et al. 2001; Reichard et al. 2003; Urrutia et al. 2009) and it hasbeen pointed out in support of the evolutionary scenario because itis in agreement with the dustier environment expected in a youngAGN. However, the results of the studies at higher wavelength arecontroversial (Gallagher et al. 2007, DiPompeo et al. 2013) It hasalso been proved that RL BAL QSOs show a variety of possibleradio morphologies (Montenegro-Montes et al. 2008; DiPompeoet al. 2011; Bruni et al. 2012, Bruni et al. 2013) and are not alwaysunresolved, as it may be expected for young radio sources (Fantiet al. 1990).One of the most successful and studied BAL catalogs is theone from Gibson et al. (2009) that includes 5,039 BAL QSOs,and the most recent Quasar Catalog (Paris et al, in prep) based onthe Sloan Digital Sky Survey Data Release 12 (SDSS DR12) in-cludes ∼ z > .RL QSOs account only for 8-13% of QSOs (Ivezi´c et al. 2002;Jiang et al. 2007; Balokovi´c et al. 2012), and the density of quasarsis a strong function of redshift that peaks at z ∼ − and declinesexponentially toward lower and higher redshift (e.g., Boyle et al.2000). Finally the identification of BAL QSO is limited in SDSSat z (cid:46) . , since absorption in high-ionization species like C IV ˚ A is required for the classification of a quasar as a BAL. Asa result the limit in redshift for the work of Montenegro-Monteset al. (2008) was z = 3 . while the samples of DiPompeo et al.(2011) and Bruni et al. (2012) had a maximum of z = 3 . and . , respectively. Extension of the samples to higher redshift is animportant test to look for evolutionary trends in BAL properties.Indeed, Allen et al. (2011) already found a dependence on redshiftof the BAL QSO fraction, the latter decreasing by a factor 3.5 ± ∼ ∼ . (cid:54) z (cid:54) . . For this purpose, we selected and collected the ra-dio flux densities of a sample of 15 RL BAL QSOs and 14 normalRL QSOs matched in redshift and magnitude. We observed theseQSOs at 6 different frequencies in order to reconstruct their SED.The age and the orientation of the radio jet is estimated from theSED. In particular the spectral index is used as a statistical indica-tor of the orientation of the jet with respect to the observer’s lineof sight, thus helping in verifying a possible preferred orientation.The number of GigaHertz Peaked Spectrum (GPS) sources is usedas indicative of whether these objects are typically younger thana sample of ”normal” QSOs. Imaging synthesis, possible with theinterferometric observations, is used to give indications about theextension and morphology of these sources.We also used a larger sample of 22 RL high-z BAL QSOsto compare their broadband optical colors with those of 113 RLQSOs with . (cid:54) z (cid:54) . . This way we investigate the question ofthe continuum differences between BALQSOs and non-BALQSOs,comparing our results with those from other authors (Menou et al.2001, Tolea, Krolik & Tsvetanov 2002, Reichard et al. 2003) atlower redshift or for samples of radio-quiet (RQ) BALs.In the same range of redshift . (cid:54) z (cid:54) . we also used alarger sample of 22 RL high-z BAL QSOs to compare their broad-band optical colors with those of 113 normal RL QSOs with com-parable redshift. This way we investigate the question of the contin-uum differences between BALQSOs and non-BALQSOs, compar-ing our results with those from other authors (Menou et al. 2001,Tolea, Krolik & Tsvetanov 2002, Reichard et al. 2003) at lowerredshift or for samples of radio-quiet (RQ) BALs.In this paper we adopt the Λ CDM cosmology with Ω Λ = 0 . , Ω m = 0 . , and H = 70 km s − Mpc − . We selected a sample of spectroscopically identified RL BALQSOs with . (cid:54) z (cid:54) . and detected in both SDSS DR7 andFIRST. The range of redshifts was chosen in order to mostly over-lap the one used in Tuccillo, Gonz´alez-Serrano & Benn (2015),where we used a Neural Networks machine method to select RLquasars candidates in the redshift range . (cid:54) z (cid:54) . . This selec-tion method was proven to be highly complete (97%) and it allowedthe spectroscopic identification of 15 QSOs missed in SDSS DR7(Carballo et al. 2008; Tuccillo, Gonz´alez-Serrano & Benn 2015).The reduced biases in the selection lead to a very accurate estima-tion of the optical luminosity function (LF) of radio-loud quasarin this range of redshift. Exploring a similar range of redshift, wesearch for BAL QSOs within a very complete spectroscopic sampleof RL QSOs. Moreover, the careful determination of the quasar LF,assures that the RQ and the RL QSOs populations do not evolvedifferently in this range of redshift. Therefore any possible differ-ence found in this research between BAL and non BAL QSOs, arenot a-priori biased by different evolutions between RL and RQ pop-ulations. c (cid:13) , 1– ?? multi-wavelength continuum characterization of high-z BAL QSOs Table 1.
Sample of 22 RL BAL selected from sources matched in FIRST and SDSS DR7SDSS ID ID RA DEC i AB S . z log P . log R AI* Note(J2000) (mJy) (W/Hz) (mag) (km/s)(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)J000051.57+001202.5 0000+00 00:00:51.57 +00:12:02.5 19.95 2.99 4.00 26.15 2.16 3762.29 (a), (g), (p)J074738.49+133747.3 0747+13 07:47:38.49 +13:37:47.3 19.15 6.62 4.04 26.53 2.13 5879.09J094003.03+511602.7 - 09:40:03.03 +51:16:02.7 18.77 13.91 3.60 26.76 2.33 166.42J100645.59+462717.2 1006+46 10:06:45.59 +46:27:17.2 19.81 6.32 4.44 26.58 2.65 185.31J102343.13+553132.4 1023+55 10:23:43.13 +55:31:32.4 19.31 2.80 4.45 26.23 2.25 4820.07 (a), (g)J103601.03+500831.8 1036+50 10:36:01.03 +50:08:31.8 19.20 9.22 4.47 26.75 2.65 1737.88 (a), (g)J110946.44+190257.6 1109+19 11:09:46.44 +19:02:57.6 20.03 6.95 3.67 26.47 2.46 1678.28 (p)J111055.22+430510.1 1110+43 11:10:55.22 +43:05:10.1 18.21 1.21 3.82 25.75 1.11 193.72 (p)J112938.73+131232.3 1129+13 11:29:38.73 +13:12:32.3 18.76 1.33 3.61 25.74 1.22 206.46 (p)J113330.91+380638.2 - 11:33:30.91 +38:06:38.2 19.66 0.87 3.63 25.56 1.42 176.29 (g), (p)J115731.67+225726.4 1157+22 11:57:31.67 +22:57:26.4 20.14 3.81 3.92 26.26 2.32 979.66 (p)J120447.15+330938.7 - 12:04:47.15 +33:09:38.7 18.38 0.92 3.61 25.58 1.25 6534.97 (a), (g), (p)J124658.83+120854.7 - 12:46:58.83 +12:08:54.72 19.86 1.44 3.80 25.82 1.76 1519.70 (a), (g), (p)J130348.94+002010.5 - 13:03:48.94 +00:20:10.51 18.66 0.99 3.65 25.62 1.15 1032.38 (a), (g), (p)J133234.18+000921.7 1332+00 13:32:34.18 +00:09:21.7 20.30 1.49 3.66 25.80 1.95 149.60 (t)J134428.55+625608.2 1344+62 13:44:28.55 +62:56:08.2 19.23 2.58 3.67 26.04 1.77 4967.82 (t), (g)J134854.37+171149.6 1348+17 13:48:54.37 +17:11:49.6 18.91 1.89 3.62 25.90 1.52 356.61 (p)J135554.56+450421.0 - 13:55:54.56 +45:04:21.0 19.31 2.07 4.09 26.03 1.62 330.64 (g)J150643.81+533134.5 1506+53 15:06:43.81 +53:31:34.5 18.77 14.63 3.79 26.82 2.34 166.89 (t)J151146.99+252424.3 1511+25 15:11:46.99 +25:24:24.3 19.76 1.39 3.72 25.78 1.72 3034.72 (a), (p)J161716.49+250208.1 - 16:17:16.49 +25:02:08.1 19.83 2.35 3.94 26.06 1.95 4439.58 (a), (g), (p)J165913.23+210115.8 1659+21 16:59:13.23 +21:01:15.8 20.11 28.81 4.78 27.30 3.37 787.91The columns give the following: (1) SDSS object-ID; (2) shortened name assigned to source for the radio observations, sources without this ID were not usedfor our radio studies ; (3) SDSS J2000 coordinates; (4) SDSS i -magnitude corrected for galactic extinction; (5) FIRST integrated radio flux density; (6)QSOredshift; (7) radio luminosity at . GHz; (8) R-parameter of radio-loudness (Kellermann et al. 1989); (9) Modified absorption index AI* (Bruni et al. 2012);(10) note indicating if the QSO is classified as BAL in previous catalogs, (a) Allen et al. 2011, (g) Gibson et al. 2009 (t) Trump et al. 2006 (p) Pˆaris et al. 2012.
Table 2.
Comparison sample of non-BAL QSOs observed with the Effelsberg-100m single dish and/or the JVLA interferometer.SDSS ID ID RA DEC i AB S . z log P . log R (J2000) (mJy) (W/Hz) (mag)(1) (2) (3) (4) (5) (6) (7) (8)J030025.23+003224.2 0300+00 03:00:25.23 +00:32:24.2 19.68 7.69 4.18 26.62 2.31J083322.50+095941.2 0833+09 08:33:22.50 +09:59:41.2 18.60 125.76 3.73 27.74 3.18J084044.19+341101.6 0840+34 08:40:44.19 +34:11:01.6 19.58 13.59 3.89 26.81 2.64J090129.23+104240.4 0912+10 09:01:29.23 +10:42:40.4 20.08 2.14 3.96 26.02 2.01J091824.38+063653.4 0918+06 09:18:24.38 +06:36:53.4 19.18 26.50 4.19 27.16 2.88J101747.76+342737.9 1017+34 10:17:47.76 +34:27:37.9 19.99 2.64 3.69 26.06 2.02J110201.91+533912.6 1102+53 11:02:01.91 +53:39:12.6 20.31 5.57 4.30 26.50 2.51J112530.50+575722.7 1125+57 11:25:30.50 +57:57:22.7 19.44 2.99 3.68 26.11 1.85J115045.61+424001.1 1150+42 11:50:45.61 +42:40:01.1 19.87 1.51 3.87 25.85 1.72J124943.67+152707.1 1249+15 12:49:43.67 +15:27:07.1 19.05 2.00 3.99 26.00 1.61J131121.32+222738.7 1311+22 13:11:21.32 +22:27:38.7 20.19 6.53 4.61 26.62 2.84J142326.48+391226.3 1423+39 14:23:26.48 +39:12:26.3 20.04 6.51 3.92 26.50 2.46J144643.37+602714.4 1446+60 14:46:43.37 +60:27:14.4 19.74 1.80 3.78 25.91 1.77J161105.65+084435.4 1611+08 16:11:05.65 +08:44:35.4 18.84 8.82 4.54 26.74 2.31The columns give the following: (1) SDSS object-ID; (2) shortened name assigned to source for the radio observations ; (3) SDSS J2000 coordinates; (4)SDSS i -magnitude corrected for galactic extinction; (5) FIRST integrated radio flux density; (6) QSO redshift; (7) radio luminosity at . GHz; (8)R-parameter of radio-loudness (Kellermann et al. 1989)
A simple one-to-one match between FIRST and SDSS willmiss double-lobe QSOs without detected radio cores. de Vries,Becker & White (2006) found that for a sample of 5515 FIRST-SDSS QSOs with radio morphological information within 450 arc-sec, the fraction of FIRST-SDSS double-lobe QSOs with unde-tected cores is 3.7 per cent. Since the starting samples of SDSSQSOs in de Vries, Becker & White (2006) and in this work obey similar SDSS selection criteria, we estimate that our sample is sim-ilarly affected for this source of incompleteness.All the sources of our sample are radio-loud in agreement withboth the main definitions adopted in literature to consider a quasaras ”radio loud”. In particular they meet the criteria based on thetotal radio luminosity of the source adopted by Gregg et al. (1996),i.e. log P . , GHz (W / Hz) > . . They also have R > , com- c (cid:13) , 1– ?? D. Tuccillo et al. monly used threshold to define radio-loudness on the basis of the R -parameter, defined as the rest-frame ratio of the monochromatic6-cm (5 GHz) and 2500 ˚A flux densities (Kellermann et al. 1989 ;Stocke et al. 1992). We also note that, given the high-z of our sam-ple of BAL QSOs, we can not measure the transition lines neededto distinguish our BAL QSOs sample in Hi-,Lo- or FeLo- BALQSOs, and therefore our sample is likely to include more than justone subclass of BALs.In section 2.1 we give the details of our selection of BALQSOs. We started our selection considering the sample of 222,517 sourcesobtained cross matching each source of the FIRST survey (2003April 11 version), not flagged as possible sidelobe or nearby brightsource ( ∼ . of the sources in the FIRST catalogue), with theclosest optical object in the ”PhotoPrimary” view of the SDSS DR7catalogue within a . (cid:48)(cid:48) radius. In this sample there is no selection byradio flux density or radio morphology other than the requirementthat the radio source has at least a weak core component. From thissample we discarded all the sources not classified as ”point-like”or tagged with the ”fatal” error flags by the SDSS pipelines. In thisway we pre-selected a sample of 36,267 sources.At this point we searched for all the available optical spec-tra for the sources of this sample, either from the SpecObj viewof SDSS-DR7 or from the 5th edition of the SDSS Quasar Cata-logue (DR7 QSO Catalogue; Schneider et al. 2010). We integratedthe sample with all the new quasars discovered with our neural-network selection strategies (Tuccillo, Gonz´alez-Serrano & Benn2015). We visually inspected all the spectra of the sources classi-fied as QSOs at z > in the DR7 QSO catalogue or in SDSS-DR7,looking for broad absorption trough in the Ci IV line.Finally we measured the absorption in the Ci IV for a rigorousand homogeneous selection of the final sample of BAL QSOs. Inliterature there is more than one metric used to separate BALQSOsand non-BAL quasars on the basis of the measured absorption. Themost widely used definitions are: (i) the balnicity index (BI, firstpresented by Weymann et al. 1991); (ii) the absorption index (AI)(Hall et al. 2002; Trump et al. 2006), designed to include narrowertroughs than the BI; (iii) and the modified balnicity index BI (Gib-son et al. 2009), which extends the integration region to zero veloc-ity. In this work, to discriminate between BAL and not-BAL QSOs,we choose to apply the same criteria used in Bruni et al. (2012), inorder to compare our results with their studies at lower redshift.Therefore we calculated the absorption index (AI), as defined byHall et al. (2002): AI = (cid:90) / s0 km / s (1 − f ( ν )0 . Cdν (1)where f ( ν ) is the continuum-normalized flux (unsmoothed when-ever possible) as a function of velocity ν , relative to the line centre.We integrate the spectral region between the peaks of the Ci IV and Si IV emission lines to up to / s from the former. Theconstant C is posed to be equal to one when f ( ν ) is less than 0.9 forat least / s (as in Trump et al. 2006), and it is posed to beequal to zero elsewhere. We considered as genuine BAL QSOs onlyobjects with an AI >
100 km / s . Applying this criteria we ended-up with a sample of 22 BAL QSOs within . (cid:54) z (cid:54) . . Compar-ing (see Table 1) this sample with larger catalogues of BALs thatare also based on the SDSS and therefore overlapping the sky area of our sample, we verified that 9 of the sources included in our sam-ple are included also in the catalogue of Allen et al. (2011), 10 areincluded also in the Gibson catalogue (Gibson et al. 2009), and 3 inthe catalogue from Trump et al. (2006). 12 QSOs are also classifiedas BAL in the DR12 QSO Catalog Pˆaris et al. (2012), publishedafter our selection. 9 BALs of our sample were not classified assuch in literature at the moment of the selection, and 4 have beenidentified as BAL QSOs in this work for the first time.Table 1 provides a catalog of the 22 BAL QSOs of our sample.In the table we include the measure of the AI for each BAL QSOsand we indicate which of these sources were used for the radiostudies that we will discuss in the next section. As a whole, andas shown in Fig. 1, our sample of 22 RL BAL quasars extend thesamples used in DiPompeo et al. (2011) and Bruni et al. (2012),for the redshift range covered (out to z = 4 . ) and extend to loweroptical and radio flux densities. With the aim of repeating at high-z radio-studies similar to thoseof Montenegro-Montes et al. (2008), DiPompeo et al. (2011) andBruni et al. (2012) we originally proposed to observe the 22 BALQSOs of our sample and a ”comparison sample” of 22 unabsorbedQSOs in multiple frequency bands, including 1.4 GHz (L-band),4.9 GHz (C-band), 8.4 GHz (X-band), and 22 GHz (K-band) atthe 100-m Effelsberg Telescope and at the EVLA radio-array inthe A configuration. Not all sources of the samples were observed,either because they were too faint in radio or for technical prob-lems related to bad weather. Nonetheless the final observed sam-ple, composed of 15 (out of the 22) BAL QSOs, and 14 QSOs ofthe ”comparison sample”, provided sufficient statistic to completethe planned studies.In Table 1 of section 2.1 we have indicated the 15 BAL QSOsfor which the data provide significant frequency coverage to ana-lyze the full shapes of the radio spectrum. In section 3.1 we givethe details of the criteria used to built the ”comparison sample” ofnormal RL QSOs and we will present the catalog of the 14 used forthe radio studies discussed in this section.
The ”comparison sample” was extracted from the spectroscopicallyidentified non-BAL RL QSOs included in the sample of 36,267 pre-selected sources presented in section 2.1. The sample was selectedusing the same criteria applied in DiPompeo, Brotherton & DeBreuck (2012), i.e. searching for each BAL QSOs of our sample,a correspondent non-BAL RL QSOs matched within 20% SDSSi-band magnitude, 20% of 1.4 GHz radio flux density, and 10% ofredshift.Table 2 provides a catalog of the 14 non-BAL QSOs of thecomparison samples for which the radio-data provide significantfrequency coverage to analyze the full shape of the radio spectrum.Fig. 2 show that these 14 non-BAL QSOs and the 15 observed BALQSOs are still well matched in redshift, optical magnitude and radioflux density.As a whole, the sample of 29 RL quasars (15 BAL and 14 non-BAL) that we observed in radio, extend the samples used in DiPom-peo et al. (2011) and Bruni et al. (2012), for the redshift range cov-ered (out to z = 4 . ) and extend to lower optical and radio fluxes.In particular, for their list of 25 radio bright ( S . >
30 mJy )RL QSOs with . (cid:54) z (cid:54) . , Bruni et al. (2012) collected radio c (cid:13) , 1–, 1–
30 mJy )RL QSOs with . (cid:54) z (cid:54) . , Bruni et al. (2012) collected radio c (cid:13) , 1–, 1– ?? multi-wavelength continuum characterization of high-z BAL QSOs Redshift P . G H z ( W / H z ) Bruni sampleDiPompeo sampleTuccillo sample
Redshift i ( m a g ) Bruni sampleDiPompeo sampleTuccillo sample
Redshift l o g ( R ∗ ) Bruni sampleDiPompeo sampleTuccillo sample
Figure 1.
We show, as a function of the redshift, the distributions in radioluminosity (at log P . (W / Hz) ), i-magnitude, and radio-loudness R for the 25 RL BAL QSOs (red triangles) used in Bruni et al. (2012), forthe 74 RL BAL QSOs (blue squares) used in DiPompeo et al. (2011) andfor the 22 RL BAL QSOs (black dots) used in this work. The large crosseswith 1- σ error bar, represent the location of the mean for each of the plottedsamples. The sample studied in this work was selected in order to extendthe previous cited studies at higher redshift and at lower optical and radiofluxes, probing new parameter space. See section 2.1 - - - - S . GHz ( mJy ) s o u r c e s binwidth = 10 mJy RL BAL QSOscomparison QSOs . - . . - . . - . . - . . - . i ( mag ) s o u r c e s binwidth = 0.5 (AB) RL BAL QSOscomparison QSOs . - . . - . . - . . - . . - . log ( R ∗ ) s o u r c e s binwidth = 0.5 RL BAL QSOscomparison QSOs
Figure 2.
We present the distributions in radio power (FIRST integrated ra-dio flux density), i-magnitude, and radio-loudness R for 15 RL BAL QSOs(red bins) and of 14 normal RL QSOs (blue bins) having . (cid:54) z (cid:54) . , aslisted in Table 1 and 2. For all these sources we had significant frequencycoverage to analyze the full shapes of the radio spectrum. The two samplesare well matched in optical magnitude and radio flux density. See section3.1 flux densities from 1.4 to 43 GHz, including observations with theEffelsberg telescope at the same frequencies (2.6, 4.8, and 8.3 GHz)that we use, and observations at (1.4, 4.86 and 8.46 GHz) with theVery Large Array (VLA) in C configuration. While for their sampleof 74 ( S . > mJy ) RL BAL QSOs with . (cid:54) z (cid:54) . ,DiPompeo et al. (2011) collected flux densities at 4.9 and 8.4 GHzwith VLA (in D- and DnC-array configuration) telescope, and com-pleted their observations with the FIRST integrated 1.4 GHz fluxdensities. Our radio observations including the same, or close, fre-quencies used in the two cited works, significantly increase their c (cid:13) , 1– ?? D. Tuccillo et al.
Table 3.
Observing frequencies and beam sizes (half-power beam-width).Telescope Frequency Bandwidth θ HPBW (GHz) (MHz) (arcsec)Effelsberg-100m 2.64 80 265Effelsberg-100m 4.85 500 145Effelsberg-100m 8.35 1100 80JVLA(A) 1.5 1024 1.3JVLA(A) 5.5 2048 0.33JVLA(A) 9.0 2048 0.20 sample size and therefore determine whether the spectral index dis-tribution differences identified remain robust. This allows us to testnot only the orientation properties for a higher redshift population,but also a larger range of radio brightness.
The observational campaign carried out for this work included bothinterferometric (JVLA) and single-dish (Effelsberg-100m) obser-vations. In the following, details about observing setup and strat-egy, as well as data reduction process, are given. See Table 3 for asummary of observations setups, including observing frequenciesand beam sizes for both telescopes.
JVLA
During November 2012, we performed observations at the JanskyVery Large Array observations at 1.5, 5.5, and 9 GHz, with theaim of adding high-sensitivity flux density measurements, and ob-tain morphological information. All observations were performedin A-configuration, and in dynamic mode, with 1 or 1.5 hours slots.Phase referencing was applied to all sources, and standard flux den-sity calibrators were observed at least once for every slot. A mini-mum RMS of ∼ software (4.1.0), making use ofcustomized reduction scripts run on the MPIfR High-PerformanceComputer cluster (HPC). Flux densities were extracted via bi-dimensional Gaussian fit on the produced maps. Errors were cal-culated assuming a 5% uncertainty for the absolute flux densitycalibration, and quadratically adding it to the map RMS. Given thewide bandwidth available, we split it in two equal intervals, tak-ing as a reference the central frequency, to improve the frequencycoverage. Effelsberg-100m telescope
Observation with the Effelsberg-100m single dish were performedin different runs, initially as a granted-time project in Septem-ber 2012, and later continued as a filler project, to improve fre-quency coverage, until May 2015. The cross-scan observing modewas used, at three different frequencies (2.6, 4.8, and 8.3 GHz),increasing the number of sub-scan repetitions depending on thesource faintness, and reaching a maximum of 20 minutes observ-ing time per frequency, at which a minimum RMS of ∼ software reduction http://casa.nrao.edu/index.shtml https://eff100mwiki.mpifr-bonn.mpg.de/doku.php package. Flux density scale was calibrated on well-known sourcesobserved every ∼ The collected flux densities are reported in Tab. 4 and 5. In total, 20sources were observed with the JVLA, and 26 with the Effelsberg-100m. When no detection at 1.5 GHz were available from our cam-paign, we used the measurement at 1.4 GHz from the FIRST sur-vey. Despite observations at the Effelsberg-100m telescope havebeen carried out on a 3-years time window, the obtained SED donot show significant changes along frequencies, that might suggesta strong variability. Thus, we can safely combine data from thiscampaign to obtain the desired frequency coverage.
All the observed sources are unresolved both with the Effelsberg-100m and the JVLA in A configuration. Considering the angularresolution of the latter for the 8 BAL and 8 non-BAL QSOs de-tected at 9 GHz (0.2 arcsec), we obtain an upper limit for the pro-jected linear size of 1.4 kpc at the mean redshift of 4.02 and 3.91 forour BAL and non-BAL samples. This is compatible with the typicallinear size of High Frequency Peakers (HFP, 0.01-0.5 kpc, Dalla-casa et al. 2000 and also GPS sources (0.5-5 kpc, O’Dea 1998).Further Very Long Baselines Interferometry (VLBI) observationswould be required to resolve sources at these redshifts.
In the following we analyze the SEDs in order to get informationabout the orientation and age of our high-redshift sample.
Synchrotron peak frequencies
The collected flux densities between 1.25 and 9.5 GHz allow usto reconstruct the SED of our objects in a reasonable way. Whenno measurement from our campaign were available at 1.5 GHz, weused data from the FIRST survey. In Fig. 3 and 4 the SEDs foreach source are presented. We can get an indication of the presenceof a peak in order to obtain the fraction of young radio sourcesin our samples. Indeed, GPS sources at lower redshifts are usu-ally identified as objects peaking in the range 1 GHz (cid:54) ν peak (cid:54) > ∼ c (cid:13) , 1– ?? multi-wavelength continuum characterization of high-z BAL QSOs Table 4.
Flux densities (mJy) collected in the radio band for the 15 high-redshift BAL QSOs. Measurements at 1.25, 1.75, 5, 6, 8.5, and 9.5 GHz are from theJVLA, while at 2.6, 4.8, and 8.3 GHz are from the Effelsberg-100m single dish. Asterisked values at 1.4 GHz comes from the FIRST survey. In the last threecolumns, the spectral index, fit type, and peak frequency in the observer’s frame (in GHz) are given.
Source S . S . S . S . S . S S S . S . S . α Fit Peak0000+00 2.0 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 5.
Flux densities (mJy) in the radio band collected for the 14 high-redshift non-BAL QSOs. Measurements at 1.25, 1.75, 5, 6, 8.5, and 9.5 GHz are fromthe JVLA, while at 2.6, 4.8, and 8.3 GHz are from the Effelsberg-100m single dish. Asterisked values at 1.4 GHz comes from the FIRST survey. In the lastthree columns, the spectral index, fit type, and peak frequency in the observer’s frame (in GHz) are given.
Source S . S . S . S . S . S S S . S . S . α Fit Peak0300+00 6.4 ± ± ± ± ± ± ± ± ±
12* - 106 ±
11 88 ± ±
14 - - -0.24 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± synchrotron emission, and a parabola in logarithmic scale (‘P’ here-after) as a simplified version of the peaked synchrotron emission,including the optically thin/thick parts. The aim was to determine,via chi-squared minimization, which among our sources presenta peak in the explored frequency range. Only for one source,1511+25, we have two points for the SED, not allowing a parabolicfit. Thus, for this source we can not discard the presence of a peak,however given the clearly inverted spectrum between 1.4 and 8.3GHz, and the upper limit at 4.8 GHz, a peak in this range is im-probable. For some sources we excluded from the fit the measure-ments that clearly belong to a second synchrotron component, atat higher or lower frequency range with respect to the fitted one.These were: 1.4 GHz for 0840+34, 1129+13, 1157+22, 1348+17;1.25 GHz and 1.75 GHz for 1249+15; 8.5 GHz and 9.5 GHz for1023+55, 1125+57. The results are reported in Tables 4 and 5, in-cluding the observer’s frame peak frequency, when present.In total, 12 out of 15 BAL QSOs (80%) and 10 out of 14 non-BAL QSOs (71%) could be fitted with a P, and thus show hints of apeak, not suggesting a clear difference between the two samples. Aselection effect, due to the requirement that sources are detected inFIRST despite the large redshift, could also have a role in selectingsources peaking in the 1-50 GHz (rest-frame) range. Nevertheless,the use of a comparison sample - selected in the same way - allowsus to draw statistically significant conclusions about the differencesbetween BAL and non-BAL QSOs. Spectral indices
The spectral index of the synchrotron spectrum (defined here as α in the expression S = ν α ) can be a useful orientation indica-tor (Orr & Browne 1982) . This has been used in previous worksto characterize the orientation angle distribution of RL BAL QSOs(Montenegro-Montes et al. 2008, DiPompeo et al. 2011, Bruni et al.2012, DiPompeo, Brotherton & De Breuck 2012) finding only amildly preferred orientation with respect to non-BAL QSOs. Werepeat here the same analysis, to test the orientation scenario athigh redshifts. For the calculation, we used the flux density mea-surements at 5 and 8.5 GHz from the JVLA, or the ones 4.8 and 8.3GHz from the Effelsberg-100m dish when the previous were notavailable. In this way, we have an estimate as homogeneous as pos-sible for the different sources, and consider frequencies at the rightside of the peak, when present. The only exceptions are the twosources presenting an inverted spectrum in the observed frequencyrange: BAL QSO 1511+25, and non-BAL QSO 1017+34. In thiscase we are looking at the self-absorbed part of the synchrotronspectrum. The obtained values for the two samples are given in Ta-ble 4 and 5.Considering only the objects that do not present an invertedspectrum (i.e. α < , 14 BAL and 13 non-BAL QSOs) 12/14BAL QSOs (86%) are steep ( α < − . ), while 10/13 non-BALQSOs (77%) are steep. A Kolmogorov-Smirnov (K-S) test on thetwo spectral index distributions results in a p =0.40, that does not c (cid:13) , 1– ?? D. Tuccillo et al. m J y m J y m J y m J y GHz GHz GHz GHz
Figure 3.
SEDs of the 15 high-redshift BAL QSOs studied here (x-axis: GHz, y-axis: mJy). Flux densities from the FIRST catalogue, at 1.4 GHz, are plottedas crosses. Triangles are 3 σ upper limits. allow us to exclude the null-hypothesis that the two samples of val-ues are drawn from the same parent distribution. Also consideringthe whole spectral index distributions from Tables 4 and 5 for theK-S test, we obtain a p =0.45, confirming the previous result. The continuum emission differences between BALQSOs and non-BALQSOs can be investigated by studying the differences betweentheir broadband colors, since they can give an indication of thephotometric spectral index. In this section we investigate the ra-dio loudness, the broadband optical (subsection 5.1) and infrared(subsection 5.2) colors of RL BAL QSOs in the highest redshift range for which the C IV absorption troughs is still visible in theoptical spectra. For this purpose we use all the 22 RL BAL QSOsselected in section 2 and listed in Table 1. We compare them withnormal RL QSOs in the same range of redshift ( . (cid:54) z (cid:54) . )and detected both in SDSS DR7 and FIRST. This comparison sam-ple consists of 99 QSOs from the SDSS DR7 QSO catalogue andit is complemented by another 14 spectroscopical confirmed QSOsselected using a 97% complete selection strategy described in Tuc-cillo, Gonz´alez-Serrano & Benn (2015) and based on the use of anartificial neural network to select quasar candidates within sourceswithout spectra in SDSS. This way resulting a total of 113 RLQSOs. We note that we excluded from the comparison sample 5QSOs showing broad absorption in the o IV emission-line only c (cid:13) , 1–, 1–
SEDs of the 15 high-redshift BAL QSOs studied here (x-axis: GHz, y-axis: mJy). Flux densities from the FIRST catalogue, at 1.4 GHz, are plottedas crosses. Triangles are 3 σ upper limits. allow us to exclude the null-hypothesis that the two samples of val-ues are drawn from the same parent distribution. Also consideringthe whole spectral index distributions from Tables 4 and 5 for theK-S test, we obtain a p =0.45, confirming the previous result. The continuum emission differences between BALQSOs and non-BALQSOs can be investigated by studying the differences betweentheir broadband colors, since they can give an indication of thephotometric spectral index. In this section we investigate the ra-dio loudness, the broadband optical (subsection 5.1) and infrared(subsection 5.2) colors of RL BAL QSOs in the highest redshift range for which the C IV absorption troughs is still visible in theoptical spectra. For this purpose we use all the 22 RL BAL QSOsselected in section 2 and listed in Table 1. We compare them withnormal RL QSOs in the same range of redshift ( . (cid:54) z (cid:54) . )and detected both in SDSS DR7 and FIRST. This comparison sam-ple consists of 99 QSOs from the SDSS DR7 QSO catalogue andit is complemented by another 14 spectroscopical confirmed QSOsselected using a 97% complete selection strategy described in Tuc-cillo, Gonz´alez-Serrano & Benn (2015) and based on the use of anartificial neural network to select quasar candidates within sourceswithout spectra in SDSS. This way resulting a total of 113 RLQSOs. We note that we excluded from the comparison sample 5QSOs showing broad absorption in the o IV emission-line only c (cid:13) , 1–, 1– ?? multi-wavelength continuum characterization of high-z BAL QSOs m J y m J y m J y m J y GHzGHzGHz GHz
Figure 4.
SEDs of the 14 comparison non-BAL QSOs (x-axis: GHz, y-axis: mJy). Flux densities from the FIRST catalogue, at 1.4 GHz, are plotted as crosses.Triangles are 3 σ upper limits. (and not in C IV ) in order to reduce any bias in the comparisonbetween BALs and non-BALs quasars.The distribution of the radio power, the i-magnitude and theradio-loudness R , for the two samples of 22 BAL and the 113 non-BAL samples is showed in Fig. 5. All the QSOs of our samplesatisfy our criteria to be detected in FIRST ( S . > )and in SDSS DR7, therefore there is no a-priori reason for whichBAL QSOs should be radio- and optical- distributed differentlythan normal QSOs. In this sense the fact that our BAL sampletends to have lower radio-power, is consistent with the tendencyfor strongly radio-loud quasars to lack of BAL QSOs (Becker et al.2001; Richards et al. 2011). In Fig. 6 we show SDSS color-color diagrams of our samples of 22BAL and 113 non-BAL QSO. All the colors have been corrected forgalactic extinction using the maps of Schlegel, Finkbeiner & Davis(1998). In the plots the large crosses with 1- σ error bar, representthe location of the mean for each of the plotted samples, respec-tively red for BAL and blue for normal QSOs. BALQSOs tend tooccupy redder sub-regions of the color-color space determined bythe distribution of the parent population of quasars. This trend isanalyzed with a t-test to compare the means of all the SDSS colorsfor the BALs and non BALs samples. In Table 5.1 we report onlythe colors for which the t-test rejects the null-hypothesis that thecolor-vectors come from normal-distributed samples having equal c (cid:13) , 1– ?? D. Tuccillo et al. log ( P . GHz ) s o u r c e s RL BAL QSOsnormal RL QSOs i ( AB ) s o u r c e s RL BAL QSOsnormal RL QSOs log ( R ∗ ) s o u r c e s RL BAL QSOsnormal RL QSOs
Figure 5.
Distribution in radio power, i - magnitude, and radio-loudness R , for our sample of 22 . (cid:54) z (cid:54) . RL BAL QSOs (red bins) and asample of 113 normal RL QSOs (blue bins) matched in redshift. We usedthese samples to compare the differences between broadband optical andNIR colors of RL BAL and non-BAL RL QSOs at high-z. See section 5. means, and therefore suggesting a significant difference in thosevalues. The t-test is performed in the general conditions of not as-suming equal variances (Behrens-Fisher problem) and the numberof degrees of freedom df is given by the Satterthwaite approxima-tion. The same trend is analyzed in Fig. 7 and 8, where we plotthe normalized distribution of the colors for which the t-test giveslower value for the statistic p, indicating a larger difference in themeans of the colors for BAL and non-BAL QSOs. We matched our sample of 22 RL BAL QSOs and 113 normalQSOs with the LAS-DR9 UK Infrared Telescope Deep Sky Survey(UKIDSS; Lawrence et al. 2007). For each source, we searched for
Table 6. t-test on SDSS colors
Color Population Mean σ Median t df p ( × − )(1) (2) (3) (4) (5) (6) (7) (8)(u-i) BAL 5.13 1.02 5.21 2.3 32.2 3.00non BAL 4.58 1.14 4.40(u-z) BAL 5.38 1.04 5.47 3.0 32.1 0.46non BAL 4.62 1.17 4.48(g-r) BAL 1.96 0.82 1.87 3.0 23.8 0.55non BAL 1.41 0.48 1.31(g-i) BAL 2.36 1.15 2.11 2.8 24.1 1.00non BAL 1.64 0.71 1.43(g-z) BAL 2.61 1.15 2.45 3.6 24.8 0.14non BAL 1.69 0.77 1.46(r-z) BAL 0.64 0.42 0.54 3.7 29.3 0.08non BAL 0.28 0.41 0.18(i-z) BAL 0.25 0.15 0.26 5.7 34.3 0.00non BAL 0.05 0.18 0.05 The columns give the following: (1) SDSS color; (2) subsampleconsidered; (3) mean of the color; (4) standard deviation; (5) median ofthe color; (6) statistic t ; (7) associated degrees of freedom to the t-test; (8)the statistic p associated to the t-test, as usually, the null hypothesis ofequal mean is rejected for values ¡ 0.05 the nearest counterpart within . (cid:48)(cid:48) from the radio-position (as givenin FIRST). In fact, as shown Wu & Jia (2010), 99.6 per cent of theSDSS QSOs having matched counterpart in UKIDSS lie within 0.5arcsec of the SDSS positions, and the maximum radio-optical sep-aration of the QSOs of our sample is less 1.0 arcsec . We have that6 BAL QSOs and 49 normal QSOs lie in the footprint of UKIDSSDR9, however only 5 BAL QSOs and 36 normal QSOs are detectedin all 4 UKIDSS bands (Y,J,H,K). In Fig. 9 we show the compari-son of the two samples in the H − K vs Y − J color-color space.The means of the UKIDSS colors are shown in the plot respectivelyas a red (for BAL QSOs) and a blue (for normal QSOs) cross. Thet-test gives (see Table 5.2) indicate that only the means of the lattercolors are statistically different from each other.We also searched for counterparts of our sample of QSOs inthe Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010)most recent data release (AllWISE; Cutri et al. 2013) within . (cid:48)(cid:48) from the radio position. Thus, we used a slightly higher radius thanthe one of . (cid:48)(cid:48) , used for the SDSS QSO Catalog (from the 9thDR, see Pˆaris et al. 2012) to cross match SDSS and WISE. In thiscase 104 normal QSOs and all 22 BAL QSOs are detected in allfour (W1,W2,W3,W4) WISE bands. However we consider onlythe sources without image artifact contamination flags and detectedwith a flux signal-to-noise ratio > in all W1,W2 and W3 bands.This way we consider two samples of 14 RL BAL QSOs and 47normal RL BAL. Using the same methodology used for the SDSSand UKIDSS colors, we compare the WISE colors of the two sam-ple as shown in Fig.10. In this case, this simple comparison of theNIR colors does not reveal differences between the two samples.The means of the colors are compared with the t-test as resumedin Table 5.2 and none of these means are statistically different forthe two samples. Thus we find no evidence of our high-z RL BALQSOs having redder WISE colors than other optically selected RLQSOs at the same range of redshift. c (cid:13) , 1– ?? multi-wavelength continuum characterization of high-z BAL QSOs g − i r − z RL BAL QSOsnormal RL QSOs g − r i − z RL BAL QSOsnormal RL QSOs
Figure 6.
Two-color diagrams presenting the SDSS colors of 22 high-z RLBAL (red dots) and 113 normal RL QSOs (blue dots) in the same range ofredshift. We show the mean colors of the BAL sample as a red cross and themean colors of the normal QSOs as a blue cross, both with 1- σ error bar.See section 5.1 Table 7. t-test on NIR, MIR colors
Color Population Mean σ Median t df p ( × − )(1) (2) (3) (4) (5) (6) (7) (8)(Y-J) BAL 0.73 0.18 0.72 1.90 6.1 10.56non BAL 0.54 0.26 0.54(H-K) BAL 0.74 0.18 0.72 3.2 10.44 0.83non BAL 0.57 0.26 0.63(W2-W1) BAL -0.51 0.16 -0.46 0.5 18.9 63.82non BAL -0.49 0.14 -0.46(W3-W2) BAL -3.23 0.45 -3.13 0.3 23.2 78.92non BAL -3.14 0.50 -3.14 The columns give the following: (1) Near and mid infrared color; (2)subsample considered; (3) mean of the color; (4) standard deviation; (5)median of the color; (6) statistic t ; (7) associated degrees of freedom to thet-test; (8) the statistic p associated to the t-test, as usually, the nullhypothesis of equal mean is rejected for values ¡ 0.05 s o u r c e s RL BAL QSOs r − z s o u r c e s normal RL QSOs Figure 7.
Comparison of the normalized distribution of the (r-z) opticalcolors of high-z RL BAL and non-BAL QSOs, as discussed in section 5.1.The mean of the BAL QSOs colors is indicated by red a dashed lines, whilefor the normal RL QSOs is indicated by blue dashed lines. s o u r c e s RL BAL QSOs i − z s o u r c e s normal RL QSOs Figure 8.
Comparison of the normalized distribution of the (i-z) opticalcolors of high-z RL BAL and non-BAL QSOs, as discussed in sectio 5.1.The mean of the BAL QSOs colors is indicated by red a dashed lines, whilefor the normal RL QSOs is indicated by blue dashed lines.c (cid:13) , 1– ?? D. Tuccillo et al. ( Y − J ) VEGA ( H − K ) V E G A RL BAL QSOsnormal RL QSOs
Figure 9.
Comparison of the NIR UKIDSS colors for a sample of 5 high-zRL BAL (red dots) and 36 normal RL QSOs (blue dots) matched in redshift.The means of the colors are shown in the plot as red cross for the BALs andas a blue cross for normal QSOs, both with 1- σ error bar. See section 5.2 ( W − W VEGA ( W − W ) V E G A RL BAL QSOsnormal RL QSOs
Figure 10.
Two-color diagram presenting the NIR WISE colors of 14 high-zRL BAL (red dots) and 47 normal RL QSOs (blue dots) matched in redshift.The large crosses with 1- σ error bar, represent the location of the mean: redfor the BALs and as a blue for normal QSOs. See section 5.2 In section 6.1 we present the conclusions of our studies in the radioband, in section 6.2 we present the conclusions of our investigationsof optical and infrared colors of our RL BAL QSOs.
In the radio band, we studied a sample of 15 RL BAL QSOs andcompared our results with the ones obtained for a well matchedsample of 14 RL non-BAL QSOs.All objects are unresolved, even at the highest resolution ob- servations performed with JVLA. The upper limit linear size de-duced from the mean redshift of BAL and non-BAL samples is of ∼ normal QSOs, even at this redshift range. A similar re-sult was found at lower redshifts for the GPS fraction (Bruni et al.2012).The fact that the orientation of BAL and non-BAL QSOs donot show a significant difference, is not in line with what found inlower redshift samples (DiPompeo et al. 2011; Bruni et al. 2012).This could be an effect of the higher fraction of GPS/HFP sourcespresent in this work, due to the considered redshift window. In fact,if outflows responsible for the BAL absorption could be later re-oriented to form the jet (Elvis 2000), young radio sources couldpresent an outflows more likely oriented towards the polar direc-tion, compensating for the tendency of BAL QSOs to have equa-torial orientation. If verified with larger samples, this would linkthe jet-collimation mechanism with accretion-disk outflows reori-entation, favoring the hypothesis that accretion disk rotation can beat the origin of magnetic jet launching (Blandford & Payne 1982;Boccardi et al. 2016).
Numerous studies on the continuum and emission-line propertiesof BAL QSOs spectra have been pursued in the last 25 years, andit was soon noticed (Weymann et al. 1991) that they show reddercontinua than those of normal quasars. This claim has been con-firmed by a number of different studies (Sprayberry & Foltz 1992;Brotherton et al. 2001; Reichard et al. 2003), and there is a gen-eral agreement on the fact that the subpopulation of LoBAL aresignificantly redder than HiBALs (Urrutia et al. 2009), and that Hi-BAL are moderately redder than quasars not showing BAL fea-tures. However the origin of the reddening itself is still subject ofdebate (see for instance Krawczyk et al. 2015) and the discussionis complicated from contrasting results in the near infrared (Gal-lagher et al. 2007, DiPompeo et al. 2013). In fact, if the differencesin the mean optical-colors of BAL and non-BAL QSOs are conse-quence of a dustier environment (in agreement with the evolutionscenario) they should be brighter in the infrared, where dust is seenin emission rather than in absorption.In section 5 we studied the SDSS optical color distributionsof our sample of 22 RL BAL QSOs and of a sample of 113 nor-mal RL QSOs with redshift in the same range and selected usingthe same criteria. The colors for BALQSOs and non-BALQSOs in-dicate that BALQSOs are redder than non-BALQSOs, and t-testsconfirm that the means of the two groups are statistically differentfrom each other. This result can not be due simply to absorptionfrom the BAL troughs themselves, since the trough absorption canmake the broadband color of BAL QSOs bluer as well as redder,depending on where the redshift of the quasar places the troughswith respect to the filters. Instead, it gives an indication of an over-all flux deficit. These results confirm that BAL QSOs are redderthan normal QSOs also at high redshift.The UKIDSS colors of the subsample of 5 BAL and 36 non-BAL QSOs detected in this survey, indicate that the excess is likelyto be extended at the wavelength range of . − . µm (Hewett c (cid:13) , 1– ?? multi-wavelength continuum characterization of high-z BAL QSOs et al. 2006). We extend the comparison at longest wavelengths con-sidering the 14 BAL and the 47 non-BAL QSOs detected and withreliable photometry in WISE. However, in this case the comparisonof the colors defined from the 3.4, 4.6 and 12 µm WISE bands, donot point to significant differences in the colors of BAL and non-BAL QSOs.
We have presented multi-frequency properties of the largest sampleof RL BAL QSOs detected in SDSS DR7 and having . (cid:54) z (cid:54) . , i.e. the highest redshift bin that allow the identification of theBAL feature with optical spectra. The sample consist of 22 RL BALQSOs, 4 of them identified as BAL QSOs in this work for the firsttime. We observed a fraction of them (15/22) in the radio band andwe analyzed optical and infrared broadband colors of the wholesample. We can summarize the conclusions of this work as follows: • All sources are unresolved, even when observed with theJVLA at 9 GHz (8 BAL vs • We compared the peak synchrotron frequencies for the BALand non-BAL QSOs samples, not finding a predominance ofGPS/HFP in the former. This does not suggest a particular youngerradio phase for BAL QSOs with respect to non-BAL objects, evenin this redshift range. Nevertheless, more than half of both samplescan be classified as GPS/HFP, that is a larger higher fraction thanthe one found at lower redshift. • We derived the spectral index for the two samples, and foundthat no statistically significant differences in orientation are presentamong BAL and non-BAL QSO objects. Given the high fractionof young radio sources present in this work (GPS/HFP), this couldmean that BAL-producing outflows can have a preferential polarorientation in these objects, compensating the preferred equato-rial orientation confirmed by different authors at lower redshifts.This would favor the hypothesis that the jet can be collimated byaccretion-disk driven magnetic force (Blandford & Payne 1982;Boccardi et al. 2016), since outflow orientation and newly-formedjets would be connected. This should be verified on larger samplesof GPS/HFP BAL QSOs. • We compare the broadband optical and NIR colors of our sam-ple of 22 RL BAL QSOs and of 106 normal RL QSOs matchedin redshift. We find that RL BALs QSOs BALQSOs tend to be lo-cated, on average, in redder regions of the color-color space respectto non-BAL RL QSOs. This trend is found in the optical (SDSS)and in the NIR wavelength of . − . µm (UKIDSS). Howeverwe do not find significant differences in the two populations whencomparing the colors at longer wavelength, i.e. at . − µm (WISE). ACKNOWLEDGEMENTS
This work has been funded by the Spanish Ministerio de Econom´ıay Competitividad (MINECO) under projects AYA2011-29517-C03-02 and AYA2014-58861-C3-2-P. The research leading to theseresults has received funding from the European Commission Sev-enth Framework Programme (FP/2007-2013) under grant agree-ment No 283393 (RadioNet3). This work is partially based on ob-servations with the 100-m telescope of the MPIfR (Max-Planck-Institut f¨ur Radioastronomie) at Effelsberg. We thank the useful help from the Effelsberg operators. D. Tuccillo also thanks the Uni-versity of Wyoming for hosting his useful and nice three monthsvisit at their Department of Physics and Astronomy. Finally, thankyou to the anonymous referee, whose constructive comments as-sisted in clarifying and improving complex parts of the paper.The National Radio Astronomy Observatory is a facility ofthe National Science Foundation operated under cooperative agree-ment by Associated Universities, Inc. This research has made use ofthe NASA/IPAC Infrared Science Archive and NASA/IPAC Extra-galactic Database (NED) which are both operated by the Jet Propul-sion Laboratory, California Institute of Technology, under con-tract with the National Aeronautics and Space Administration. Usehas been made of the Sloan Digital Sky Survey (SDSS) Archive.The SDSS is managed by the Astrophysical Research Consortium(ARC) for the participating institutions: The University of Chicago,Fermilab, the Institute for Advanced Study, the Japan ParticipationGroup, The John Hopkins University, Los Alamos National Labo-ratory, the Max-Planck-Institute for Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA), New Mexico State Uni-versity, University of Pittsburgh, Princeton University, the UnitedStates Naval Observatory, and the University of Washington.
References
Allen J. T., Hewett P. C., Maddox N., Richards G. T., BelokurovV., 2011, MNRAS, 410, 860Antonucci R., 1993, Ann.Rev.Astron.Astrophys., 31, 473Baars J. W. M., Genzel R., Pauliny-Toth I. I. K., Witzel A., 1977,A&A, 61, 99Balokovi´c M., Smolˇci´c V., Ivezi´c ˇZ., Zamorani G., Schinnerer E.,Kelly B. C., 2012, ApJ, 759, 30Becker R. H., Gregg M. D., Hook I. M., McMahon R. G., WhiteR. 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, 227Blandford R. D., Payne D. G., 1982, MNRAS, 199, 883Boccardi B., Krichbaum T. P., Bach U., Bremer M., Zensus J. A.,2016, A&A, 588, L9Boyle B. J., Shanks T., Croom S. M., Smith R. J., Miller L., Loar-ing N., Heymans C., 2000, MNRAS, 317, 1014Brotherton M. S., Tran H. D., Becker R. H., Gregg M. D., Laurent-Muehleisen S. A., White R. L., 2001, ApJ, 546, 775Brotherton M. S., van Breugel W., Smith R. J., Boyle B. J., ShanksT., Croom S. M., Miller L., Becker R. H., 1998, ApJ, 505, L7Bruni G., Dallacasa D., Mack K.-H., Montenegro-Montes F. M.,Gonz´alez-Serrano J. I., Holt J., Jim´enez-Luj´an F., 2013, A&A,554, A94Bruni G. et al., 2012, A&A, 542, A13Carballo R., Gonz´alez-Serrano J. I., Benn C. R., Jim´enez-Luj´anF., 2008, MNRAS, 391, 369Cutri et al., 2013, VizieR Online Data Catalog, 2328Dallacasa D., Stanghellini C., Centonza M., Fanti R., 2000, A&A,363, 887de Vries W. H., Becker R. H., White R. L., 2006, AJ, 131, 666DiPompeo M. A., Brotherton M. S., De Breuck C., 2012, ApJ,752, 6DiPompeo M. A., Brotherton M. S., De Breuck C., Laurent-Muehleisen S., 2011, ApJ, 743, 71DiPompeo M. A., Runnoe J. C., Brotherton M. S., Myers A. D.,2013, ApJ, 762, 111 c (cid:13) , 1– ?? D. Tuccillo et al.
Elvis M., 2000, ApJ, 545, 63Fanti R., Fanti C., Schilizzi R. T., Spencer R. E., Nan Rendong,Parma P., van Breugel W. J. M., Venturi T., 1990, A&A, 231, 333Fine S., Jarvis M. J., Mauch T., 2011, MNRAS, 412, 213Gallagher S. C., Hines D. C., Blaylock M., Priddey R. S., BrandtW. N., Egami E. E., 2007, ApJ, 665, 157Ganguly R., Brotherton M. S., Cales S., Scoggins B., Shang Z.,Vestergaard M., 2007, ApJ, 665, 990Gibson R. R. et al., 2009, ApJ, 692, 758Gregg M. D., Becker R. H., White R. L., Helfand D. J., McMahonR. G., Hook I. M., 1996, AJ, 112, 407Hall P. B. et al., 2002, ApJS, 141, 267Hewett P. C., Warren S. J., Leggett S. K., Hodgkin S. T., 2006,MNRAS, 367, 454Hopkins P. F., Elvis M., 2010, MNRAS, 401, 7Ivezi´c ˇZ. et al., 2002, AJ, 124, 2364Jiang L., Fan X., Ivezi´c ˇZ., Richards G. T., Schneider D. P., StraussM. A., Kelly B. C., 2007, ApJ, 656, 680Kellermann K. I., Sramek R., Schmidt M., Shaffer D. B., GreenR., 1989, AJ, 98, 1195Knigge C., Scaringi S., Goad M. R., Cottis C. E., 2008, MNRAS,386, 1426Krawczyk C. M., Richards G. T., Gallagher S. C., Leighly K. M.,Hewett P. C., Ross N. P., Hall P. B., 2015, AJ, 149, 203Lawrence A. et al., 2007, MNRAS, 379, 1599L´ıpari S. L., Terlevich R. J., 2006, MNRAS, 368, 1001Menou K. et al., 2001, ApJ, 561, 645Montenegro-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, 1853Murray N., Chiang J., Grossman S. A., Voit G. M., 1995, ApJ,451, 498O’Dea C. P., 1998, PASP, 110, 493Orr M. J. L., Browne I. W. A., 1982, MNRAS, 200, 1067Pˆaris I. et al., 2012, A&A, 548, A66Proga D., Stone J. M., Kallman T. R., 2000, ApJ, 543, 686Reichard T. A. et al., 2003, AJ, 126, 2594Richards G. T. et al., 2011, AJ, 141, 167Rochais T. B., DiPompeo M. A., Myers A. D., Brotherton M. S.,Runnoe J. C., Hall S. W., 2014, MNRAS, 444, 2498Saikia D. J., Salter C. J., 1988, Ann.Rev.Astron.Astrophys., 26,93Schlegel D. J., Finkbeiner D. P., Davis M., 1998, ApJ, 500, 525Schneider D. P. et al., 2010, AJ, 139, 2360Silk J., Rees M. J., 1998, A&A, 331, L1Sprayberry D., Foltz C. B., 1992, ApJ, 390, 39Stocke J. T., Morris S. L., Weymann R. J., Foltz C. B., 1992, ApJ,396, 487Tolea A., Krolik J. H., Tsvetanov Z., 2002, ApJ, 578, L31Trump J. R. et al., 2006, ApJS, 165, 1Tuccillo D., Gonz´alez-Serrano J. I., Benn C. R., 2015, MNRAS,449, 2818Urrutia T., Becker R. H., White R. L., Glikman E., Lacy M.,Hodge J., Gregg M. D., 2009, ApJ, 698, 1095Urry C. M., Padovani P., 1995, PASP, 107, 803Weymann R. J., Morris S. L., Foltz C. B., Hewett P. C., 1991, ApJ,373, 23Wright E. L. et al., 2010, AJ, 140, 1868Wu X.-B., Jia Z., 2010, MNRAS, 406, 1583 c (cid:13) , 1–, 1–