An X-ray view of the INTEGRAL/IBIS blazars
S. Gianni', A. De Rosa, L. Bassani, A. Bazzano, A.J. Dean, P. Ubertini
aa r X i v : . [ a s t r o - ph . C O ] O c t Mon. Not. R. Astron. Soc. , 1– ?? (2002) Printed 2 May 2017 (MN L A TEX style file v2.2)
An X-ray view of the INTEGRAL/IBIS blazars
S. Giann´ı , ⋆ , A. De Rosa , L. Bassani , A. Bazzano , T. Dean and P. Ubertini Universit´a di Roma Tor Vergata INAF-IASF Roma INAF-IASF Bologna School of Physics and Astronomy, University of Southampton, SO17 1BJ, UK
Accepted 2009 ... . Received 2009 ... ; in original form ...
ABSTRACT
Aim of this work is a broad-band study with INTEGRAL, Swift and XMM-Newton satellites of a sample of 9 blazars (7 FSRQ and 2 BL Lac) with redshift up toabout 4.The spectral analysis has shown clear evidence of a flattening of the continuum to-wards the low energies (
E < N H ∼ -10 cm − in the rest-frame ofthe sources) or a broken power-law continuum model (with an energy break below 3keV in the observer-frame). No Compton reflection features, Fe Kα line and hump athigh energies, have been detected, with the exception of the source IGR J22517+2218that shows the presence of a weak iron line.In this work we also investigate a possible correlation between the absorptioncolumn density N H and the red-shift.We confirm the existence of a N H -z trend, with the higher absorption at z > N H andthe photon index Γ is also presented. The hard X-ray data allow us to detect highlyabsorbed sources (with N H > cm − in rest-frame of the source) characterizedby photon index distribution peaked at harder values (Γ ∼ .
4) with respect to thatobtained with XMM data only (Γ ∼ Key words: galaxies: active – X-rays: galaxies – quasars: general.
Blazars are one of the most intriguing class of objects amongthe Active Galactic Nuclei (AGN) and include BL Lac-ertae objects (BL Lac) and Flat-Spectrum Radio Quasars(FSRQ). The intrinsic differences between these two typesof sources are in their optical spectra and in the intensityof their emission: a BL Lac source does not show any op-tical emission lines (
EW < A ) and it is basically a low-power blazar ( L Bol ∼ − erg s − ), whereas a FSRQsource shows significant emission line equivalent widths andcorresponds to a high-power blazar ( L Bol ∼ erg s − )(Stickel et al. 1991; Padovani 1997). In the AGN unifiedscheme (Antonucci 1993), blazars are interpreted as radio-sources with a relativistic jet aligned along the line of sight(Urry & Padovani 1995); in other words, the class of blazarsis represented by Radio Loud (RL) AGN observed very closeto the direction of the relativistic jets ( < o ).The Spectral Energy Distribution (SED) of blazars ⋆ E-mail:... is usually modelled with two large humps produced viaSynchrotron (peaking in the infra-red to soft X-ray en-ergy band) and Inverse Compton (IC) emission, dom-inating the hard X-ray to gamma-ray regimes, respec-tively. The lower energy emission component is Synchrotronradiation from the jet whereas the high-energy compo-nent arises through the inverse Compton of soft pho-tons by highly relativistic electrons in the jet plasma.These soft photons originate from the local Synchrotronradiation within the jet or the nuclear optical/UVemission, namely Synchrotron Self-Compton (SSC) andExternal Compton (EC) components, respectively (seeMaraschi, Ghisellini & Celotti 1992; Dermer & Schlickeiser1993; Sikora, Begelman & Rees 1994).The X-ray spectrum of an AGN is usually well describedby a power-law with a specific flux (i.e. per unit energy inter-val) of the form N ( E ) ∝ E − Γ , where E is the energy, N(E)is the number of photons in units of s − cm − keV − and Γis the photon index . However, X-ray spectra of blazars showsome deviations from the simple power-law. These spectralsignatures appear in the soft X-ray band with a curvature (a c (cid:13) flattening or a steepening of the spectrum); in addition, thepresence of Compton reflection components in RL objects isstill debated.A clear physical interpretation of these features has notyet been found. In the present work, we investigate these stillopen questions in order to achieve a better understanding ofthe general properties of the blazars spectra. The state-of-art of the study of these important issues can be summarizeas follows:(i) Two physical interpretations to the flattening of theprimary intrinsic continuum with respect to a simple power-law exist so far (Page et al. 2005; Galbiati et al. 2005;Yuan et al. 2006): absorption in excess of the Galactic com-ponent (e.g. Cappi et al. 1997; Yuan et al. 2006) and a breakin the intrinsic continuum (Sikora, Begelman & Rees 1994;Tavecchio et al. 2007). For the absorption excess, a depen-dence between the hydrogen column density of the absorber( N H ) and the redshift z has been proposed. In fact in a for-mer work by Yuan et al. (2006) based on XMM observationsof a sample of 32 RL sources, a N H - z correlation has beenfound.(ii) Contrary to Radio Quiet (RQ) sources the pres-ence of reflection features - most notably the Fe Kα line and the associated Compton reflection ”hump” atabout 20 −
30 keV - from the cores of the RL AGNis not well established. In particular, the reprocessedfeatures are generally intense and always detected in RQAGN (Matt 2001), whereas in RL AGN the iron lineand the reflection component can be absent or weak.X-ray observations of RL AGN with ASCA, RXTE andBeppo SAX (e.g. Sambruna, Eracleous & Mushotzky 1999;Eracleous, Sambruna & Mushotzky 2000; Grandi et al.2001; Ballantyne, Ross & Fabian 2002; Grandi & Palumbo2004; Grandi, Malaguti & Fiocchi 2006) have shown thatsome of these objects (defined as Broad Line RadioGalaxies) seem to have weak hard X-ray reflection features.In this paper we present a broad-band (0 . −
100 keV)study of a sample of 9 blazars up to redshift ∼ The INTEGRAL/IBIS total sample is derived from the thirdIBIS/ISGRI survey catalog that included 421 sources for atotal exposure time of 40Ms; 131 sources out of 421 havebeen identified as AGN. The survey input data set consistsof all pointing data available at the end of 2006 May, fromrevolutions 12-429 inclusive, covering the time period fromlaunch (17 October 2002) to the end of 2006 April. Detailsabout this survey are presented by Bird et al. (2007). A sub- set of sources has been selected according with the followingselection criteria: • sources classified as blazars, optically identified (e.g.Masetti et al. 2004, 2008) • sources also observed in the soft X-ray energy domain(0 . −
10 keV);Two further sources 3C 273 and 4C 04.42 have been ex-cluded since widely discussed elsewhere (Chernyakova et al.2007; De Rosa et al. 2008).Our sample of INTEGRAL selected blazars, is com-posed of 9 sources: 7 FSRQ objects with 0 . < z < . z ∼ .
07 and 0 .
09, but only fiveout of seven FSRQ were available at the time of our analy-sis. For the 2 other objects we adopted the hard X-rays BATdata.Relevant information on the sample are reported in Ta-ble 1 where we list the source name, type, optical coordi-nates, redshift, Galactic absorption along the line of sightaccording to Dickey & Lockman (1990). In the last columnof Table 1 we also list the IBIS 20 −
100 keV flux as reportedby Bird et al. (2007) and the BAT 14 −
195 keV flux as re-ported by AGN Catalog of the first 9 months available atthe time we started this project .We remark that source with IGR name (IGRJ22517+2218) is discovered in hard X-ray with INTEGRAL.It is the farthest object so far detected by INTEGRAL whosenature was determined a posteriori through optical spec-troscopy (Bassani et al. 2007). It is worth to note that twosources (FSRQs 3C 279 and PKS 1830-211) have been de-tected in the Gamma Ray band with the Astro-rivelatoreGamma a Immagini Leggero (AGILE) (Giuliani et al.2009; Tavani et al. 2008; Pittori et al. 2009) and the LargeArea Telescope on board the Fermi Gamma-ray Large AreaSpace Telescope (GLAST) . The latter satellite also re-vealed a third source belonging to our sample, the objectBL Lac (Abdo et. al 2009). The INTEGRAL data presented here are based on point-ings with the IBIS instrument (Ubertini et al. 2003), col-lected over the period from end of 2002 up to April 2006(revolution 12 up to 429). Images from the ISGRI detector(Lebrun et al. 2003) for each pointing have been genereatedin different energy bands using off-line scientific AnalysisSoftware (Goldwurm et al. 2003) OSA version 5.1. Count The BAT data have been taken from the on-line archive at:http://swift.gsfc.nasa.gov/docs/swift/results/bs9mon. For theBAT spectra (related to PKS 2149-306 and 0537) we stress thatin the 22-months survey now available, the 15-45 keV fluxes aredoubled with respect to the 9-months survey. This effect is takeninto account with the cross calibration constant that has beenadded when dealing with the fit between soft-X and hard-X raysspectra. An Italian Space Agency (ASI) mission launched on 23 April2007 with a key scientific project: the Gamma-Ray observationsof blazars. An international and multi-agency mission launched on 11 June2008; one of its major scientific goals is to provide new data onthe Gamma-ray activity of AGN.c (cid:13) , 1– ?? n X-ray view of the INTEGRAL/IBIS blazars Table 1.
Data for our sample of blazars observed with INTEGRAL, Swift and XMM-Newton.
Source Type Broad − Band
R.A. Dec redshift NGalH FIBIS − keV Lrest − frame − keV [1022 cm −
2] (
FBAT − keV ) ( Lrest − frame − keV )[10 − erg · cm − · s −
1] [1046 erg · s − rates at the source position have been extracted from in-dividual images to provide light curves in different energybands. From light curves the average fluxes have been de-rived and combined to produce an average source spectrum(for details see Bird et al. 2007) in the 20 −
100 keV band.
XMM data of the six blazars (1ES 0033+595, 4C 04.42, PKS1830-211, QSO B0836-710, PKS 0537-286 and PKS 2149-307) are a combination of proprietary data and public obser-vations obtained from the XMM-Newton Science Archive .The raw EPIC Observation Data Files (ODFs) were ob-tained from the XMM Science Archive and reduced usingthe standard Science Analysis System (SAS) software pack-age (v.7.1.0) and the most recent calibration files availableat the time of the data reduction. We used the EMCHAINand EPCHAIN task for the pipeline processing of the ODFsto generate the corresponding event files. The spectra werecreated using X-ray events of pattern 0-12 for MOS and 0-4for PN. The source counts were extracted from a circular re-gion centred on the source with a radius of 20-40 arcsec andthe background was derived from two nearby source-free cir-cular regions of the same size. Spectra were re-binned usingGRPPHA to have a minimum of 20 counts in each bin , sothat the χ statistic could reliably be used. Details of theXMM-Newton observations and the source observed countrates are summarized in Table 2. Swift/XRT data reduction of four blazars (3C 279, BL Lac,IGR J22517+2218, Swift J1656.3-3302) was performed usingthe tool XSELECT v. 2.4. Events for spectral analysis wereextracted within a circular region of radius 20 arcsec centredon the source position. The background was extracted from http://xmm.esac.esa.int/xsa/index.shtml.
10 counts are the minimum number of counts required to usethe χ minimization technique. a circular region with the same radius and located far offthe source. In all cases, the spectra were binned using GRP-PHA. We used Ancillary Response Files (ARFs) and Re-sponse Matrix Files (RMFs) available for download from theHEASARC Calibration Database (caldb) calibration files at:http://heasarc.gsfc.nasa.gov/docs/heasarc/caldb/caldb intro.html.Table 3 lists the observation ID, the observation start time,the integration time and the observed count rate. The spectral analysis was performed by combining the dif-ferent data from each instrument using the X-ray SPECtralfitting software XSPEC v. 11.3.2 (Arnaud 1996). The errorsare quoted at 90% confidence level for one interesting pa-rameter. The energy values are always given in the observerframe, if not otherwise specified. Throughout this paper, aWMAP (Wilkinson Microwave Anisotropy Probe) Cosmol-ogy of H = 70 km s − Mpc − , Ω λ = 0 .
73 and Ω m = 1 − Ω λ is assumed.We have combined XMM spectrum with the IBIS onefor the blazars 1ES 0033+595, 4C 04.42, PKS 1830-211,QSO B0836-710; XRT spectrum with the IBIS one for 3C279, BL Lac, IGR J22517+2218, Swift J1656.3-3302 andXMM spectrum with the BAT one for PKS 0537-286, PKS2149-307 (see Table 1).To account for possible cross-calibration mismatches be-tween different instruments as well as to take into accountflux variations between the observing periods, a multiplica-tive constant factor (C) has always been added to the fit. Ingeneral, the calibration between the various satellites is per-formed analyzing the Crab spectrum. In particular, variousstudies of the Crab spectrum (e.g. Kirsch et al. 2004, 2005)have shown that the inter-calibration XMM/INTEGRALand Swift/INTEGRAL is close to 1 (within a few percent).We remark that, since XMM, Swift and INTEGRAL obser-vations were not simultaneous, flux variations are also pos-sible according to the blazar-nature of objects. Being theinter-calibration constant C ∼
1, we suggest that, a differ-ent value for this constant is likely due to flux variability. In c (cid:13) , 1–, 1–
1, we suggest that, a differ-ent value for this constant is likely due to flux variability. In c (cid:13) , 1–, 1– ?? Table 2.
XMM-Newton observations log and source observed count rates.
Source
Obs.ID Date of obs. Exp. time Exp. time Exp. time Counts/s Counts/s Counts/sMOS1 MOS2 PN MOS1 MOS2 PN(0 . − keV ) (0 . − keV ) (0 . − keV )( s ) ( s ) ( s ) ( s − ) ( s − ) ( s − )1ES 0033+595 0094381301 2003-02-01 1204 1535 4209 1 . ± .
04 1 . ± .
04 3 . ± . . ± .
01 0 . ± .
01 1 . ± . . ± .
01 3 . ± .
01 11 . ± . . ± .
002 0 . ± .
002 0 . ± . . ± .
01 0 . ± .
01 1 . ± . Table 3.
Swift/XRT observations log and source observed count rates.
Source
Obs.ID Date of obs. Exp. time Counts/sXRT XRT(0 . − keV )( s ) ( s − )BL Lac 00090042010 2008-08-29 5806 0 . ± . . ± . . ± . . ± . our analysis, this constant C between the two instrumentsat low and high energies has been allowed to vary freelyin the fits. In particular, we have considered the MOS1 andXRT instruments as reference detectors in the case of XMM-INTEGRAL or BAT data and XRT-INTEGRAL data,respectively. Therefore, the inter-calibration constants inour fitting procedure are MOS2/MOS1, PN/MOS1, INTE-GRAL/MOS1, BAT/MOS1 and INTEGRAL/XRT. How-ever MOS2/MOS1 and PN/MOS1 are always ∼ To reproduce the shape of the intrinsic continuum of oursources we fitted the combined spectra (0 . −
100 keV) ofeach source with several models.Firstly, we fit the spectra of our sources in the 3 − ∼ c (cid:13) , 1– ?? n X-ray view of the INTEGRAL/IBIS blazars Figure 1.
The data to model A ratio as obtained by fitting a simple power-law over the observed frame 3 −
100 keV energy range,and extrapolating over the 0 . −
100 keV range. The black, red and green bins up to 10 keV are the data of MOS1, MOS2 and PNinstruments, respectively.
Then, we tested a broken power-law continuum model C(phabs*bknpo model in XSPEC) for the whole sample. Re-sults are shown in the Table 6. This model did not improvethe fit further in most of the cases and, also in those casesfor which a slightly better χ is obtained, the null hypoth-esis probability increases only a few per cent with respectto model B. Although strong statistical evidence does notexist for a broken power-law respect to the absorbed one,however the presence of an intrinsic spectral break cannotbe ruled out.For PKS 1830-211, both models B and C fail to repro-duce source data. PKS 1830-211 is a gravitationally lensedgalaxy and the excess absorption can be attributed to theintervening galaxy at redshift z = 0 .
89. The best-fit of theobserved spectrum requires both an absorber and a brokenpower-law (model D), confirming the results reported byZhang et al. (2008) with three XMM-Newton observations Obs.ID 0204580201, 0204580301 and 0204580401 and theINTEGRAL average data (up to April 29, 2006). In fact, inaddition to the excess cold absorption, PKS 1830-211 alsoexhibits an energy break at about 4 keV in the observerframe. Moreover, our result for the absorption in excess tothe Galactic one, is in agreement with that obtained previ-ously by De Rosa et al. (2005) with Chandra observationsand a subset of the actual IBIS data. Table 7 reports theresults of our best-fit model.Finally, we also checked a Compton reflection compo-nent with a power-law reflected from neutral matter (pexravmodel in XSPEC, Magdziarz & Zdziarski (1995)); the reflec-tion component is parameterized in terms of R =Ω / π , thesolid angle in units of 2 π subtended by the reflecting mat-ter, assumed to be observed face-on). This model resultedin a worst fit and upper limit values on R for all absorbedsources (see Table 8). In our analysis we have chosen the observation with longer ex-posure time.c (cid:13) , 1– ?? In addition, we have searched for the presence of an ironline emission in the spectra of our sources. To this aim, weadopted the following method: we exclude the Fe K band,fit a continuum, then freeze the continuum, after we includethe Fe K band and perform the fit again. The comparisonbetween the values of the χ of the fits performed includingand excluding the Fe K band provides us a clear indicationfor the real presence of this feature of the observed spectra.For all sources, but one (IGR J22517+2218), the appli-cation of this method does not produce any clear evidencefor the presence of this feature in the spectrum. The mainresults for the source IGR J22517+2218 can be summarizedas follows: the first step described before resulted in a pho-ton index Γ=1.5 and an upper limit on N H =5 × cm − with χ r /dof=23/29; in the second step we included the ironchannels and then we added a narrow line ( σ =0.01 keV)and the fit has been repeated obtaining χ r /dof=37/44 witha probability P F test > .
7% implying that this feature isreal with an Equivalent Width EW=80 ±
50 eV; repeatingthe fit with a free photon index a value of Γ = 1 . ± . Our broad-band spectral analysis has shown that a deficitof photons in the soft X-ray energy domain is a commonfeature in our sample. This phenomenon has been previouslyobserved in several samples of RL objects (e.g. Page et al.2005; Galbiati et al. 2005; Yuan et al. 2006) and its originis not yet well understood. It can be produced by eitheran intrinsically curved continuum or an extra absorptioncomponent.The broken continuum hypothesis is linked to theblazar-type source. In fact, in the hypothesis of an intrinsi-cally curved blazar continuum, one can assume that the rela-tivistic electrons in the jet follow a broken power-law energydistribution with a low energy cut-off, producing a flatteningin the observed spectrum. This is a purely phenomenolog-ical form assumed to reproduce the observed shape of theblazar SEDs, without any specific assumption on the ac-celeration/cooling mechanism acting on the particles (e.g.Tavecchio et al. 2007). Alternatively, it can be also assumedthere is an inefficient radiative cooling of lower energy elec-trons of the jet producing a few Synchrotron photons to bescattered at higher energies (e.g. Sikora, Begelman & Rees1994).We emphasize that on basis of our spectral analysis anintrinsic spectral break cannot be ruled out (see the val-ues of the reduced χ and of the null hypothesis probabilitylisted in the Tables 5 and 6). In this context, we just notedthat the existence of a break in the intrinsic continuum istightly related to the electron distribution, but importantconstraints on the curved electron distribution can be de-rived only analyzing both Synchrotron and EC componentsin the SED, that is out of the aim of present work. There-fore we will focus the following discussion on the absorptionscenario. The absorption explanation takes into account an ab-sorbing gas in excess of the Galactic one; in some cases, theabsorbing medium can be cold (not ionized) and locatedeither between the observer and the source along the lineof sight at a different redshift with respect to the blazar(e.g. Cappi et al. 1997; Fiore et al. 1998; Fabian et al. 2001;Masetti et al. 2004), or at the same redshift of the source(e.g. Page et al. 2005; Yuan et al. 2006) and it is namely anintrinsic cold absorber.For an absorber located at several redshifts betweenthe observer and the source, the damped Ly-alpha systemsare the more plausible cause (e.g. Elvis et al. 1994) andthe line of sight would have to pass through two or morevery high column density damped Ly-alpha systems (e.g.Fall & Pei 1995). However, this is a rare occurrence (e.g.O’Flaherty & Jakobsen 1997; Zwaan, Verheijen & Briggs1999) and favours the hypothesis of an intrinsic absorberwith column densities of the order 10 − cm − . Anintrinsic origin is also confirmed by the soft X-ray spec-tral flattening detected so far only in radio loud objects(Bechtold et al. 1994; Reimers et al. 1995; Siebert et al.1996).The intrinsic hypothesis also suggests that the mediumcould be ”warm” at the same stage. In this case, the absorb-ing medium, at the same redshift of the AGN (probably, theinner part of the dense interstellar medium of the young hostgalaxy), is ionized as due to the proximity from the jet (e.g.Fabian et al. 2001; Yuan et al. 2006).For our sample, the fitting procedure supports the pres-ence of an intrinsic (local to the AGN) cold absorber forall sources. Nevertheless a warm absorber model (absori inXSPEC) can be excluded since the model with an ionized-gas produces a worse fit and an inconsistent value for theionization parameter ( ξ ∼ erg cm s − and ξ > erg cms − , values indicating a neutral absorber and a completelyionized absorber, respectively).The information on the absorption is of crucial impor-tance to understand the blazar environment and its interac-tion with the jet.Absorption originating from the material present in theAGN environment seems to be a more convincing explana-tion for many sources belonging to the class of RL AGN(Page et al. 2005; Yuan et al. 2006, and references therein).There is some evidence not only of absorption but also of acorrelation between absorption and redshift, with the moredistant sources being more absorbed and the increase in N H with z occurring starting at z ∼ c (cid:13) , 1– ?? n X-ray view of the INTEGRAL/IBIS blazars Table 4.
Best fit parameters of a simple power-law model in 3-100 keV energy range.Model A Name F ob (3 − keV ) 1 F ob (20 − keV ) 3 Γ C χ r /dof P null QSO B0836+710 4.1 5.2 1 . ± .
02 0 . ± .
03 0.91/1521 0.9961ES 0033+595 1.1 1.5 2 . +0 . − . . +2 . − . . ± .
06 0 . ± . . ± .
06 1 . ± . . ± .
04 0 . ± . . +0 . − . . +2 . − . . +0 . − . . +11 . − . . +0 . − . . +6 . − . . +0 . − . . +3 . − . Observed flux in the 2 −
10 keV (or 20 −
100 keV) energy range in units of 10 − erg cm − s − . Photon index, related to thespectral index α (where F ν ∝ ν − α ) by α = Γ − IBIS/MOS, BAT/MOS or IBIS/XRT cross-calibration constant. Reducedchi-squared to degrees of freedom. Null hypothesis probability that is the probability of getting a value of χ as large or larger thanobserved if the model is correct. If this probability is small then the model is not a good fit to the data. Table 5.
Best fit parameters of an absorbed power-law model in 0.2-100 keV energy range.Model B Name F ob (2 − keV ) 1 F ob (20 − keV ) 2 N H Γ C χ r /dof P null QSO B0836+710 4.0 4.9 0 . ± .
02 1 . ± .
005 0 . ± .
03 1.02/2421 0.2061ES 0033+595 1.2 1.5 0 . ± .
02 2 . ± .
05 3 . ± . . ± .
05 1 . ± .
01 0 . ± . < .
05 1 . ± .
01 2 . ± . . +5 . − . . +0 . − . . +1 . − . . +3 . − . . +0 . − . . +6 . − . . ± .
07 2 . +0 . − . . +2 . − . . ± .
02 1 . ± .
06 2 . +0 . − . Observed flux in the 2 −
10 keV (or 20 −
100 keV) energy range in units of 10 − erg cm − s − . Intrinsic column density ofhydrogen in units of 10 cm − . Photon index, related to the spectral index α (where F ν ∝ ν − α ) by α = Γ − IBIS/MOS,BAT/MOS or IBIS/XRT cross-calibration constant. Reduced chi-squared to degrees of freedom. Null hypothesis probability that isthe probability of getting a value of χ as large or larger than observed if the model is correct. If this probability is small then themodel is not a good fit to the data.c (cid:13) , 1– ?? Table 6.
Best fit parameters of a broken power-law model in 0.2-100 keV energy range.Model C Name F ob (2 − keV ) 1 F ob (20 − keV ) 7 Γ ,Γ E b C χ r /dof P null QSO B0836+710 4.0 4.9 1 . +0 . − . . ± .
08 0 . ± .
03 1.02/2420 0.2741 . ± . . +0 . − . . +0 . − . . +0 . − . . +0 . − . PKS 0537-286 0.5 2.3 0 . +0 . − . . ± .
09 0 . ± . . ± . . +0 . − . . +4 . − . . ± . . +0 . − . Swift J1656.3-3302 0.6 2.0 1 . +0 . − . . +45 . − . . +2 . − . . +0 . − . IGR J22517+2218 0.2 3.5 0 . +0 . − . . +0 . − . . +8 . − . . +0 . − . Bl Lac 0.5 2.2 1 . +0 . − . . +0 . − . . +1 . − . . +0 . − .
3C 279 0.5 2.2 1 . +0 . − . . +0 . − . . +0 . − . . ± . Observed flux in the 2-10 keV (or 20-100 keV) energy range in units of 10 − ergcm − s − . IBIS/MOS, BAT/MOS or IBIS/XRTcross-calibration constant. Reduced chi-squared to degrees of freedom. Null hypothesis probability that is the probability of gettinga value of χ as large or larger than observed if the model is correct. If this probability is small then the model is not a good fit to thedata. Spectral index below the break (Γ ) and above the break (Γ ). Observed break energy in units of keV.
Table 7.
Best fit parameters of a broken power-law model with an absorption in excess in 0.2-100 keV energy range.Model D Name F ob (2 − keV ) 1 F ob (20 − keV ) 2 N H Γ ,Γ E b C χ r /dof P null PKS 1830-211 1.4 4.3 1 . +0 . − . . ± .
05 4 . +0 . − . . +0 . − . . +0 . − . Observed flux in the 2 −
10 keV (or 20 −
100 keV) energy range in units of 10 − erg cm − s − . Intrinsic column density ofhydrogen in units of 10 cm − . IBIS/MOS, BAT/MOS or IBIS/XRT cross-calibration constant. Reduced chi-squared to degrees offreedom. Null hypothesis probability that is the probability of getting a value of χ as large or larger than observed if the model iscorrect. If this probability is small then the model is not a good fit to the data. Spectral index below the break (Γ ) and above thebreak (Γ ). Observed break energy in units of keV.
Table 8.
Upper limits to R under the assumption of a pexrav model. Name QSO B0836+710 1ES 0033+595 PKS 0537-286 PKS 2149-307 PKS 1830-211 Swift J1656.3-3302 IGR J22517+2218 Bl Lac 3C 279 R c (cid:13) , 1– ?? n X-ray view of the INTEGRAL/IBIS blazars likely the absorbing material is linked to the presence of thiscomponent. This working scenario is supported by a theorypresented, e.g., in Ferrari (1998), in which the confinementof the radio jet needs the presence of a cold medium (localto the AGN) so that jet expansion is suppressed.In our sample we have found neither an evidenceof an iron emission line at 6.4 keV nor evidenceof a Compton reflection ”hump” above 10 keV. Thelack of these spectral features – interpreted as the re-sult of reprocessing of the primary continuum by coldmatter around the X-ray source, presumably the ac-cretion disk (Lightman & White 1988; Guilbert & Rees1988; George & Fabian 1991; Sambruna & Eracleous 1999;Sambruna, Eracleous & Mushotzky 1999) – can be ex-plained either with the out-and-out absence of reprocess-ing features in blazar-type objects or with the hypothesisthat these features are difficult to be detect in strong radiosources being likely dilute by the jet (Sambruna et al. 2006).There is also evidence that the spectral shape does notchange between 0 . −
10 keV and 20 −
100 keV energy ranges,supporting the idea that, above 10 keV, no additional com-ponent to the power-law (e.g. Compton reflection hump) ispresent in the spectrum.The value of the cross-calibration constant (see TablesB and C) shows that a strong flux variability is presentin QSO B0836+710, 1ES 0033+595, PKS 1830-211, IGRJ22517+2218, BL Lac, and 3C 279. In particular, we derivedthat IGR J22517+2218 showed severe intensity changes be-tween the INTEGRAL and the Swift observations with avalue of the XRT/IBIS inter-calibration factor of about 7.This is not a surprise because of the nature of the objectand data being not simultaneous. In addition, we remarkthat INTEGRAL/IBIS data set correspond to an averageperformed on observations spanning on 3 years, while thedata for the low-energy range (XMM and XRT) correspondto observations with a time baseline of few hours.
Although the number of sources of our sample is not large,we have also analyzed the distribution of the spectral pa-rameters (absorbing column density, spectral index, red-shift) and their possible correlations. This type of analy-sis has been performed, so far, only between 0 . −
10 keV(Yuan et al. 2006). Our study is still preliminary and hasbeen performed with a detailed comparison among the re-sults of our own spectral analysis and those already availablein literature in a limited energy range. The reference sampleincludes 35 RL objects – with 19/35 sources identified asblazars – (Galbiati et al. 2005; Page et al. 2005; Yuan et al.2006) and shows an absorption of unknown intrinsic nature.In order to avoid any problem of ”orientation dependencebias” when studying the N H and Γ distribution, we have se-lected, from the quoted sample, only the blazar-type objects.In particular, for the N H study we have accounted for theblazars with a specific value of the intrinsic absorption (wehave not accounted for the upper limits on N H ), obtaininga sub-sample of 16 objects.In the framework of a possible N H -redshift relation,we also combined our results with the larger sample of RLobjects (Galbiati et al. 2005; Page et al. 2005; Yuan et al. Figure 2.
Distribution of the X-ray absorption column density( N H ) for our sample objects over the 0 . −
100 keV energy range.
Figure 3.
The N H distribution of a larger sample of 16 blazarsover the 0 . −
10 keV energy range. See text for more details. N H with the redshift previously found byYuan et al. (2006). If this would be the case, we would havefurther evidence that the absorbing column density, in thesesources, is linked to the distance and thus a confirmation ofa cosmic evolution effect. We made a detailed analysis of the rest-frame hydrogen col-umn density for our selected blazars, in conjunction withthe intervening absorption in excess to the galactic one.The N H distribution shown in Figure 2 appears to beasymmetric, with a peak located at about log ( N H ) = 21and tails extending towards low and high values up to log ( N H ) = 20 and log ( N H ) = 23, respectively. It is worth tonote that the statistical significance of this result is stronglyaffected by the limited number of objects in our sample.Therefore we compare our sample with the larger sub-sample c (cid:13) , 1– ?? Figure 4.
The X-ray absorption column density N H as a functionof the redshift for our sample. Figure 5.
The X-ray absorption column density N H as a functionof the redshift, combining the results of our sources (filled rhombs)and the ones of a larger sample (plus and cross signs) analyzedin the soft X-ray energy range. from the literature observed in the 0 . −
10 keV energy rangefrom XMM-Newton that includes 16 RL objects identifiedas blazars and with a specific value of the absorption. Weunderline that this one includes only one object belongingto our sample (the blazar QSO B0836+710). The N H dis-tribution of the larger sample is shown in Figure 3, with apeak located at log ( N H ) = 22 and a tail extending towardslower values up to log ( N H ) = 20. Comparing the distribu-tions in Figure 3 and in Figure 2, we note that sources withabsorbing column density around log ( N H ) = 23 are ”miss-ing”. This evidence is expected in view of the fact that in thesoft X-ray energy range (0 . −
10 keV) the sources with anabsorbing column density N H > . cm − are almostcompletely obscured.This occurrence strongly suggests that data above 10 keV represent a crucial tool to construct an unbiased selec-tion of heavily absorbed objects.The N H -redshift trend of our sample is shown in Figure4. The comparison of this N H - z plot with those previouslyanalyzed by Yuan et al. (2006) at various redshifts for a sam-ple of 32 RL sources, seems to indicate that the property ofthe X-ray absorbing gas evolves with cosmic time. This trendis shown in Figure 5, where we added 3 objects analyzed byGalbiati et al. (2005). In fact, at first glance, a correlationbetween N H and redshift seems to exist and, compared tolow redshifts ( z < N H seems to be indicated at high redshifts.The sample of 32 sources includes 3 objects of our sam-ple (QSO B0836+710, PKS 0537-286 and PKS 2149-307)and the N H values for these 3 sources as obtained throughour spectral analysis are in agreement with those presentedin the work of Yuan and collaborators (2006).The existence of a correlation is also confirmed by thestatistical analysis: we obtain a correlation coefficient ( R )around 0.7 (Figure 5), when the results corresponding to oursample are added to the sources investigated by Page et al.(2005), Galbiati et al. (2005) and Yuan et al. (2006).All together these results seem to support the hypoth-esis of an intrinsic absorber. Three further hints support itsintrinsic nature: • the existence of a spatial isotropy of the inter-galacticabsorbers is most unlikely; • the X-ray absorption is associated with RL but not withRQ objects (Yuan & Brinkmann 1999). This evidence sup-ports the hypothesis that the absorber are physically asso-ciated with the RL quasars; • the change of the N H distribution that seems to occurat z ∼ N H - z relationat z > z The photon-index distribution was then investigated byadding the FSRQ 3C 273 and 4C 04.42 in our analysis(Γ ∼ . ∼ .
3, respectively, as obtained by our broad-band analysis). The Γ distribution for the complete sampleis shown in Figure 6 and the mean value obtained in 0 . − . ± . ∼ . ∼
2. However, we notehere that our sample is mainly composed by FSRQ, whilethe larger sample contains different types of RL objects -not all identified as blazars. Therefore, in order to performa meaningful comparison, we have studied the photon indexdistribution for sources – in the larger sample – identifiedas blazars (19 objects). The peak of the Γ distribution (seeFigure 8) is, again, around 2.To account for the fact that most of the sources in our c (cid:13) , 1– ?? n X-ray view of the INTEGRAL/IBIS blazars Figure 6.
Distribution of the power-law photon-index for oursample in the 0 . −
100 keV energy range.
Figure 7.
Distribution of the photon-index resulting by the anal-ysis performed in the works of Page et al. (2005), Galbiati et al.(2005) and Yuan et al. (2006) and including 35 RL objects.
Figure 8.
Photon-index distribution for 19 sources identified asblazars.
Figure 9.
Photon-index distribution for 15 sources identified asFSRQ. sample are FSRQ-type objects and that FSRQs in X-raysshow a lower value of Γ with respect to BL Lac objects, westudied the Γ distribution including only the FSRQ sources(15 objects) in the larger sample. In this case, the Γ distri-bution in Figure 9 shows a peak at Γ ∼ .
6, a value steeperthan that for our blazars sample.The differences between the values of Γ of our sampleand those of the larger sample could be due to the hard X-rayselection of our INTEGRAL/IBIS sample, which is clearlybiased towards the flatter values of the photon index.
In this work we presented a broad-band X-ray spectral studyof a sample of 9 blazars (with a redshift range of 0 . < z < .
7) observed with INTEGRAL, XMM-Newton and Swift.The main results can be summarized as follows: • The broad-band spectra of all selected sources are wellreproduced with a power-law model absorbed by an amountof gas in excess to the Galactic one ( N H ∼ -10 cm − in rest-frame of the source; only an upper limit of N H ∼ cm − has been derived for the FSRQ PKS 2149-307). • The absorption seems to be a signature of a cold in-trinsic absorber, confirming and extending to larger sampleprevious results quoted in the literature (Cappi et al. 1997;Page et al. 2005; Yuan et al. 2006). • The present work provides a further confirmation of theexistence of a N H -redshift trend, obtained for a large sampleof RL objects (not only blazars). • The broad-band analysis of our sample of blazars re-vealed a harder spectrum with a photon-index of the orderof Γ ∼ .
4, compared to the value obtained with the distri-bution including the FSRQ sources of the larger sample (15objects). Such a difference could be due to the hard X-rayselection of our INTEGRAL/IBIS sample which is clearlybiased towards flatter values of the photon index. • We have found no evidence of reflection components (re-flection ”hump” and iron emission line), with the exceptionof the source IGR J22517+2218 that shows the presence of aweak iron line. This result is expected in blazar-type objects. c (cid:13) , 1–, 1–
4, compared to the value obtained with the distri-bution including the FSRQ sources of the larger sample (15objects). Such a difference could be due to the hard X-rayselection of our INTEGRAL/IBIS sample which is clearlybiased towards flatter values of the photon index. • We have found no evidence of reflection components (re-flection ”hump” and iron emission line), with the exceptionof the source IGR J22517+2218 that shows the presence of aweak iron line. This result is expected in blazar-type objects. c (cid:13) , 1–, 1– ?? In conclusion, the analysis presented here has shownthat our INTEGRAL-sample selection favours objects heav-ily absorbed and with a flatter value of spectral index. Onthe other hand, the present broad-band analysis of INTE-GRAL/IBIS, XMM-Newton and Swift observations of RLQSOs confirms that the observed flattening is common inthese objects, and is clearly detected in eight quasars ofour sample (8/9). The assumption of an intrinsic origin anda cold nature for the absorber is consistent with previousresults up to 10 keV obtained by Page et al. (2005) andYuan et al. (2006) with XMM-Newton data only. However,a broken power-law model, as an alternative explanation forthe deficit of soft photons observed in the majority of oursources, cannot be ruled out by the data.
ACKNOWLEDGMENTS
Authors acknowledge support from INAF and ASI via con-tract I/008/07.
REFERENCES
Abdo A.A., Ackermann M., Ajello M. et al., 2009, ApJ,700, 597Antonucci R., 1993, ARA&A, 31, 473Arnaud K., 1996, in ASP Conf. Ser. 101, AstronomicalData Analysis Software and Systems V, ed. G. Jacoby& J. Barnes (San Francisco: ASP), 17Ballantyne D.R., Ross R.R. & Fabian A.C., 2002, MNRAS,332, L45Bassani L., Landi R., Malizia A. et al., 2007, ApJ, 669, L1Bechtold J., Elvis M., Fiore F. et al., 1994, AJ, 108, 759Bird A.J., Malizia A., Bazzano A. et al., 2007, ApJ, 170,175Canizares C.R. & White J.L., 1989, ApJ, 339, 27Cappi M., Matsuoka M., Comastri A. et al., 1997, ApJ,478, 492Chernyakova M., Neronov A., Courvoisier T.J.-L. et al.,2007, A&A, 465, 147De Rosa A., Piro L., Tramacere A. et al., 2005, A&A, 438,121De Rosa A., Bassani L., Ubertini P. et al., 2008, MNRAS,388, 54Dermer C.D. & Schlickeiser R., 1993, ApJ, 416, 458Elvis M., Fiore F., Mathur S. & Wilkes B., 1994, ApJ,425,103Eracleous M., Sambruna R. & Mushotzky R.F., 2000, ApJ,537, 654Fabian A.C., Celotti A., Iwasawa K. & Ghisellini G., 2001,MNRAS, 324, 628Fall S.M. & Pei Y.C., 1995, Proceedings of the ESO Work-shop Held at Garching, Germany, 21-24 November 1994,edited by G. Meylan. Springer-Verlag Berlin HeidelbergNew York. Also ESO Astrophysics Symposia, p.23Ferrari A., 1998, A&A, 36, 539Fiore F., Giommi P., La Franca F. et al., 1998,arXiv:astro-ph/9911149Galbiati E., Caccianiga A., Maccacaro T. et al., 2005,A&A, 430, 927George I.M. & Fabian A.C., 1991, MNRAS, 249, 352 Giuliani A., D’Ammando F., Vercellone S. et al., 2009,A&A, 494, 509Goldwurm A. et al., 2003, A&A, 411, 223Grandi P., Maraschi L., Urry C.M. & Matt G., 2001, ApJ,556, 35Grandi P. & Palumbo G.C., 2004, Science, 306, 998Grandi P., Malaguti G. & Fiocchi M., 2006, ApJ, 642, 113Grupe D., Mathur S., Wilkes B. & Osmer P., 2006, AJ,131, 55Guilbert P.W. & Rees M.J., 1988, MNRAS, 233, 475Kirsch M.G.F., Becker W., Larsson S. et al., 2004, ESASP,552, 863Kirsch M.G.F., Briel U.G., Burrows D. et al., 2005, SPIE,5898, 22Lawson A.J. & Turner M.J.L., 1997, MNRAS, 288, 920Lebrun F., Leray J.P., Lavocat P. et al., 2003, A&A, 411,141Lightman A.P. & White T.R. 1988, ApJ, 335, 57Magdziarz P. & Zdziarski A.A. 1995, MNRAS, 273, 837Maraschi L., Ghisellini G. and Celotti A., 1992, ApJ, 397,L5Masetti N., Palazzi E., Bassani L. et al., 2004, A&A, 426,41Masetti N., Mason E., Morelli L. et al., 2008, A&A, 482,113Matt G., 2001, AIPC, 599, 209O’Flaherty K.S. & Jakobsen P., 1997, ApJ, 479, 673Padovani P., 1997, MmSAI, 68, 47Page K.L., Turner M.J.L., Reeves J.N. et al., 2003, MN-RAS, 338, 1004Page K.L., Reeves J.N., O’Brien P.T. & Turner M.J.L.,2005, MNRAS, 364, 195Pittori C., Verrecchia F., Chen A.W. et al., 2009, A&A506, 1563Reimers D., Bade N., Schartel N. et al., 1995, A&A, 296,49Sambruna R.M. & Eracleous M., 1999,arXiv:astro-ph/9911503Sambruna R.M., Eracleous M. & Mushotzky R.F., 1999,ApJ, 526, 60Sambruna R.M., Gliozzi M., Tavecchio F. et al., 2006, ApJ,652, 146Siebert J., Brinkmann W., Morganti R. et al., 1996, MN-RAS, 279, 1331Sikora M., Begelman M.C. & Rees M.J., 1994, ApJ, 421,153Stickel M., Padovani P., Urry C.M., Fried J.W., & KuhrH., 1991, ApJ, 374, 431Tavani M. et al. 2008, Nucl. Instrum. Methods Phys. Res.A, 588, 52Tavecchio F., Maraschi L., Ghisellini G. et al., 2007, ApJ,665, 980Ubertini P., Lebrun F., Di Cocco G. et al., 2003, A&A,411, 131Urry M.C. & Padovani P., 1995, PASP, 107, 803Vignali C., Brandt W.N., Schneider D.P. et al., 2003, AJ,125, 2876Vignali C., Brandt W.N., Schneider D.P. and Kaspi S.,2005, AJ, 129, 2519Zhang S., Chen Y., Collmar W. et al., 2008, ApJ, 683, 400Zwaan M.A., Verheijen M.A.W. & Briggs F.H., 1999,PASA, 16, 100 c (cid:13) , 1– ?? n X-ray view of the INTEGRAL/IBIS blazars Yuan W., Brinkmann W., Siebert J. & Voges W., 1998,A&A, 330, 108Yuan W. & Brinkmann W., 1999, Proceedings of the Sym-posium ”Highlights in X-ray Astronomy”, eds. B. Aschen-bach & M.J. Freyberg, MPE Report 272, 240Yuan W., Fabian A.C., Worsley M.A. & McMahon R.G.,2006, MNRAS, 368, 985 c (cid:13) , 1–, 1–