High-energy properties of the high-redshift flat spectrum radio quasar PKS 2149-306
aa r X i v : . [ a s t r o - ph . H E ] O c t Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 26 October 2018 (MN L A TEX style file v2.2)
High-energy properties of the high-redshift flat spectrumradio quasar PKS 2149 − F. D’Ammando , ⋆ , M. Orienti Dip. di Fisica e Astronomia , Universit`a di Bologna, Viale Berti Pichat 6/2, I-40127 Bologna, Italy INAF - Istituto di Radioastronomia, Via Gobetti 101, I-40129 Bologna, Italy
Accepted. Received; in original form
ABSTRACT
We investigate the γ -ray and X-ray properties of the flat spectrum radio quasar PKS2149 −
306 at redshift z = 2.345. A strong γ -ray flare from this source was detected bythe Large Area Telescope on board the Fermi Gamma-ray Space Telescope satellite in2013 January, reaching on January 20 a daily peak flux of (301 ± × − ph cm − s − in the 0.1–100 GeV energy range. This flux corresponds to an apparent isotropicluminosity of (1.5 ± × erg s − , comparable to the highest values observed bya blazar so far. During the flare the increase of flux was accompanied by a significantchange of the spectral properties. Moreover significant flux variations on a 6-h time-scale were observed, compatible with the light crossing time of the event horizon ofthe central black hole.The broad band X-ray spectra of PKS 2149 −
306 observed by
Swift -XRT and
NuSTAR are well described by a broken power-law model, with a very hard spectrum (Γ ∼ break = 2.5–3.0 keV, and Γ ∼ . ∼ NuSTAR and
Swift joint observations.As for the other high-redshift FSRQ detected by both
Fermi -LAT and
Swift -BAT, thephoton index of PKS 2149 −
306 in hard X-ray is 1.6 or lower and the average γ -rayluminosity higher than 2 × erg s − . Key words: galaxies: active – quasars: general – quasars: individual (PKS 2149 − Blazars are radio-loud active galactic nuclei (AGN), withpowerful relativistic jets observed at a small viewing angle.For this reason their emission is strongly enhanced due toDoppler boosting and they are expected to be detected upto high redshift. The most distant blazar identified so faris Q0906+6930 (Romani et al. 2004), located at redshift z = 5.47. A possible excess of γ -ray photons from a positioncompatible with this flat spectrum radio quasar (FSRQ) wasobserved by EGRET (Romani 2006), but has not been con-firmed by Fermi
Large Area Telescope (LAT) observationsso far. Recently, two FSRQ at redshift z >
5, B2 1023+25and SDSS J114657.79+403708.6, were detected in hard X-rays by
NuSTAR and identified as blazars (Sbarrato et al.2013; Ghisellini et al. 2014a). Both objects have never been ⋆ E-mail: [email protected] detected in γ rays. The most distant FSRQ reported in theThird Fermi -LAT catalogue (3FGL; Acero et al. 2015) isPKS 0537 −
286 at redshift z = 3.104, indicating the diffi-culty in detecting quasars at z > γ -ray regime.PKS 2149 −
306 (RA = 21h51m55.5239s, Dec. = − ◦ ′ . ′′ z = 2 .
345 (Wilkes 1986). The source isbright in X-rays, showing substantial variability both inintensity and spectral slope, as indicated by
ROSAT (Siebert et al. 1996),
ASCA (Cappi et al. 1997),
XMM-Newton (Ferrero & Brinkmann 2003), and
Swift observa-tions (Sambruna et al. 2007; Bianchin et al. 2009). A ten-tative detection of an emission line at ∼
17 keV in thesource frame by
ASCA was interpreted as highly-blueshiftedFe K α (Yaqoob et al. 1999). This finding was not con-firmed by Fang et al. (2001) and Page et al. (2004) using Chandra data. As for other FSRQ, a low-energy photondeficit in X-rays was suggested for PKS 2149 − c (cid:13) F. D’Ammando, M. Orienti bly due to an absorbing cloud in the source rest frame (e.g.,Sambruna et al. 2007) or to a low-energy tail of the electronpopulation (e.g., Tavecchio et al. 2007). PKS 2149 −
306 wasdetected in hard X-rays with a hard spectrum by
Beppo
SAX(Elvis et al. 2000),
Swift -BAT (Baumgartner et al. 2013),
INTEGRAL -IBIS (Beckmann et al. 2009), and lately
NuS-TAR (Tagliaferri et al. 2015).Among the high-redshift ( z >
2) blazars, 64 were re-ported in the Third
Fermi
LAT Catalog (3FGL; Acero et al.2015). Only two of these objects are at redshift z >
3. In con-trast, ten blazars at redshift z >
Swift -BAT (Baumgartner et al. 2013),
INTEGRAL -IBIS (Bassani et al. 2012) and
NuSTAR (Sbarrato et al.2013; Ghisellini et al. 2014a). In particular, seven blazarsat redshift z >
Swift -BAT. Therefore ob-servations in the hard X-ray band seem to be more effectivethan the γ -ray band for finding blazars at redshift z > γ -ray flare from a high-redshift blazar maybe even more interesting with respect to the flaring activityfrom other blazars.On 2013 January 4 a strong γ -ray flare from PKS 2149 − Fermi -LAT (preliminary results were re-ported in D’Ammando & Orienti 2013). The aim of this pa-per is to discuss the γ -ray and X-ray properties of this sourceand to make a comparison with the other high-redshiftblazars detected by Fermi -LAT and
Swift -BAT using the3FGL catalogue (Acero et al. 2015) and the 70-month
Swift -BAT catalogue (Baumgartner et al. 2013).This paper is organized as follows. In Section 2 we re-port the LAT data analysis and results, while in Sections 3,4, and 5 we present the results of the
Swift , XMM-Newton ,and
NuSTAR observations, respectively. We discuss theproperties of the source in Section 6, while in Section 7 wesummarize our results. Throughout the paper, a Λ–cold darkmatter cosmology with H = 71 km s − Mpc − , Ω Λ = 0 . m = 0 .
27 is adopted (Komatsu et al. 2011). The cor-responding luminosity distance at z = 2 .
345 (i.e. the sourceredshift) is d L = 19240 Mpc. Throughout this paper thequoted uncertainties are given at 1 σ level, unless otherwisestated. For power law spectra dN/dE ∝ E − Γ X ,γ we denoteas Γ X and Γ γ the spectral indices in the X-ray and γ -raybands, respectively. FERMI -LAT DATA: SELECTION ANDANALYSIS
The
Fermi -LAT is a pair-conversion telescope operatingfrom 20 MeV to more than 300 GeV. It has a large peakeffective area ( ∼ for 1 GeV photons), and a fieldof view of about 2.4 sr with an angular resolution (68 percent containment angle) of 0 . ◦ E =1 GeV on-axis. Details about the Fermi -LAT are given inAtwood et al. (2009).The LAT data reported in this paper were collectedfrom 2008 August 4 (MJD 54682) to 2014 August 4 (MJD56873). During this period the LAT instrument operated al-most entirely in survey mode. The analysis was performed with the
ScienceTools software package version v9r33p0 .Only events belonging to the ‘Source’ class were used. Inaddition, a cut on the zenith angle ( < ◦ ) was appliedto reduce contamination from the Earth limb γ rays, whichare produced by cosmic rays interacting with the upper at-mosphere. The spectral analysis was performed with the in-strument response functions P7REP SOURCE V15 using an un-binned maximum likelihood method implemented in the Sci-ence tool gtlike . Isotropic (‘iso source v05.txt’) and Galac-tic diffuse emission (‘gll iem v05 rev1.fit’) components wereused to model the background . The normalisations of bothcomponents were allowed to vary freely during the spectralfittings.We analysed a region of interest of 10 ◦ radius centredat the location of PKS 2149 − γ -ray signal from the source by means of amaximum-likelihood test statistic TS = 2 × (log L − log L ),where L is the likelihood of the data given the model with( L ) or without ( L ) a point source at the position of PKS2149 −
306 (e.g., Mattox et al. 1996). The source model usedin gtlike includes all the point sources from the 3FGL cat-alogue that fall within 15 ◦ of PKS 2149 − <
10 and/or the predicted number of counts basedon the fitted model N pred <
1. A second maximum likeli-hood analysis was performed on the updated source model.In the fitting procedure, the normalization factors and thespectral shape parameters of the sources lying within 10 ◦ ofPKS 2149 −
306 were left as free parameters. For the sourceslocated between 10 ◦ and 15 ◦ from our target, we kept thenormalization and the spectral shape parameters fixed tothe values from the 3FGL catalogue.Integrating over the period 2008 August 4–2014 Au-gust 4 (MJD 54682–56873) using a PL model, dN/dE ∝ ( E/E ) − Γ , the fit results in TS = 2096 in the 0.1–100 GeVenergy range, and a photon index Γ γ = 2.79 ± ± × − ph cm − s − . In order totest for curvature in the γ -ray spectrum of PKS 2149 − dN/dE ∝ ( E/E ) − α − β log ( E/E ) , was used for the fit. We obtain aspectral slope α = 2.36 ± E = 221 MeV, a curvature parameter around the peak β =0.29 ± ± × − ph cm − s − (Table 1). We used a likelihoodratio test (LRT) to check the PL model (null hypothesis)against the LP model (alternative hypothesis). FollowingNolan et al. (2012) these values may be compared by defin-ing the curvature Test Statistic TS curve =TS LP –TS PL , whichin this case results in TS curve = 87 ( ∼ . σ ), meaning thatwe have evidence of significant curvature in the average γ -ray spectrum.Fig. 1 shows the γ -ray light curve for the first 6 years of Fermi -LAT observations of PKS 2149 −
306 using a LP modeland 1-month time bins. For each time bin, the spectral shape http://fermi.gsfc.nasa.gov/ssc/data/analysis/software/ http://fermi.gsfc.nasa.gov/ssc/data/access/lat/BackgroundModels.html c (cid:13) , 000–000 igh-energy properties of the FSRQ PKS 2149 − Figure 1.
Integrated flux light curve of PKS 2149 −
306 obtainedby
Fermi -LAT in the 0.1–100 GeV energy range during 2008 Au-gust 4–2014 August 4 (MJD 54682–56873) using a LP model with30-day time bins. Arrows refer to 2 σ upper limits on the sourceflux. Upper limits are computed when TS <
10. The dashed linerepresents the mean flux. parameters of PKS 2149 −
306 and all sources within 10 ◦ ofit were frozen to the values resulting from the likelihoodanalysis over the entire period. If TS <
10, 2 σ upper limitswere calculated. The statistical uncertainty in the fluxes arelarger than the systematic uncertainty (Ackermann et al.2012) and only the former is considered in the paper.During the first two years of Fermi operation, PKS2149 −
306 was observed in a low-activity state, with an av-erage (0.1–100 GeV) flux of (6.4 ± × − ph cm − s − and a spectrum described by a PL model with photon in-dex of Γ γ = 3.00 ± −
089 andS5 0836+710; Abdo et al. 2010b; Akyuz et al. 2013). Flaringactivity from this source was first observed in 2011 Febru-ary, and subsequently an even stronger flare was detected in2013 January (Fig. 1).
Leaving the spectral shape parameters free to vary dur-ing the first high-activity period (2011 February 3–March5; MJD 55595–55625), using a LP model, the fit results ina spectral slope α = 2.53 ± E = 221 MeV, a curvature parameter around the peak β =0.28 ± ± × − ph cm − s − . Using a PL model, the fit resultsin TS = 521 and a photon index of Γ γ = 2.85 ± curve = 18 ( ∼ . σ ), i.e. a significant curvature of the γ -ray spectrum in 2011February.During the second high activity period (2013 January4–February 2; MJD 56296–56325), using a LP model the fitresults in a spectral slope α = 1.99 ± E = 221 MeV, a curvature parameter around thepeak β = 0.28 ± ± × − ph cm − s − . Using a PL model the fitresults in TS = 1239 and a photon index of Γ γ = 2.45 ± curve = 34 ( ∼ . σ ),indicating a significant curvature of the γ -ray spectrum inthat period.In the following analysis of the light curves on sub-dailytime-scales, we fixed the flux of the diffuse emission com-ponents at the value obtained by fitting the data over therespective daily time-bins. In Fig. 2 we show a light curvefocused on the period 2011 February 3–March 5 (left plot)and 2013 January 4–February 2 (right plot), with 1-day (up-per panel), 12-h (middle panel), and 6-h (lower panel) timebins. For each time bin, the spectral shape parameters ofPKS 2149 −
306 and all sources within 10 ◦ of it were frozento the values resulting from the likelihood analysis over theentire period considered.In 2011 the daily peak of the emission was observed onFebruary 18 (MJD 55610) with a flux of (140 ± × − ph cm − s − in the 0.1–100 GeV energy range, a factorof about 14 higher than the average flux over 6 years of Fermi observations. The corresponding apparent isotropic γ -ray luminosity peak in the 0.1–100 GeV energy range is(5.3 ± × erg s − . On 12-h and 6-h time-scale theobserved peak flux is (157 ± × − and (180 ± × − ph cm − s − , corresponding to an apparent isotropic γ -rayluminosity of (5.9 ± × and (6.8 ± × ergs − , respectively.In 2013 the daily peak of the emission was observed inJanuary 20 (MJD 56312) with a flux of (301 ± × − ph cm − s − in the 0.1–100 GeV energy range, i.e. a fac-tor of about 30 higher than the average flux over 6 years of Fermi observations. The corresponding apparent isotropic γ -ray luminosity peak in the 0.1–100 GeV energy range is(1.5 ± × erg s − . On 12-h and 6-h time-scales theobserved peak flux is (335 ± × − and (385 ± × − ph cm − s − , corresponding to an apparent isotropic γ -rayluminosity of (1.6 ± × and (1.9 ± × erg s − ,respectively. By means of the gtsrcprob tool we estimatedthat during this flare the highest energy photon emitted byPKS 2149 −
306 (with probability >
80 per cent to be associ-ated with the target) was observed on 2013 January 12 withan energy of 4.8 GeV. SWIFT
DATA: ANALYSIS AND RESULTS
The
Swift satellite (Gehrels et al. 2004) performed sixteenobservations of PKS 2149 −
306 between 2005 Decemberand 2014 April. The observations were performed with allthree instruments on board: the X-ray Telescope (XRT;Burrows et al. 2005, 0.2–10.0 keV), the Ultraviolet/OpticalTelescope (UVOT; Roming et al. 2005, 170–600 nm) and theBurst Alert Telescope (BAT; Barthelmy et al. 2005, 15–150keV). c (cid:13) , 000–000 F. D’Ammando, M. Orienti
Table 1.
Unbinned likelihood spectral fit results. PL LPDate (UT) Date (MJD) Γ TS PL α β TS LP TS Curve ± ± ± ± ± ± ± ± ± Figure 2.
Integrated flux light curve of PKS 2149 −
306 obtained by
Fermi -LAT in the 0.1–100 GeV energy range during 2011 February3–March 5 ( left plot ) and 2013 January 4–February 2 ( right plot ), with 1-day time bins (upper panel), 12-h time bins (middle panel),and 6-h time bins (bottom panel). Arrows refer to 2 σ upper limits on the source flux. Upper limits are computed when TS <
10. In themiddle and bottom panels upper limits are not shown.
Table 2.
Log and fitting results of
Swift -XRT observations of PKS 2149 −
306 using a PL model with N H fixed to the Galactic absorption. ( ∗ ) The model that best fit this observation is a broken power-law (see Section 3.2).Date Date Net Exposure Time Photon index Flux 0.3–10 keV χ /d.o.f.(UT) (MJD) (s) (Γ X ) ( × − erg cm − s − )2005-Dec-10 53714 3314 1 . ± .
08 1 . ± .
12 44/452005-Dec-13 53717 2255 1 . ± .
10 1 . ± .
10 32/252009-Apr-23 54944 3114 1 . ± .
11 1 . ± .
07 36/312009-Apr-29 54950 1773 1 . ± .
14 1 . ± .
09 16/162009-May-05 54956 2889 1 . ± .
10 1 . ± .
07 28/282009-May-14 54965 2924 1 . ± .
08 1 . ± .
07 40/412009-May-23 54974 2989 1 . ± .
11 1 . ± .
09 31/282009-May-29 54980 2565 1 . ± .
08 1 . ± .
10 39/362010-May-11 55327 4760 1 . ± .
08 1 . ± .
05 38/442011-May-07 55688 2947 1 . ± .
09 1 . ± .
12 35/382011-Nov-10 55875 3261 1 . ± .
07 1 . ± .
09 39/482013-Dec-16/17 56642/43 7956 1 . ± .
04 2 . ± .
08 143/1392014-Mar-28 56744 2537 1 . ± .
08 2 . ± .
11 47/442014-Apr-18 56765 6331 1 . ± .
05 2 . ± .
07 126/106 ∗ c (cid:13) , 000–000 igh-energy properties of the FSRQ PKS 2149 − Swift -BAT
The hard X-ray flux of this source is below the sensitiv-ity of the BAT instrument for the short exposures of thesingle observations, therefore those data from this instru-ment were not used. On the other hand, the source is in-cluded in the
Swift
BAT 70-month hard X-ray catalogue(Baumgartner et al. 2013). The 14–195 keV spectrum is welldescribed by a power law with photon index of Γ X = 1.50 ± χ /d.o.f. = 4.8/6). The resulting 14–195 keV flux is(8.3 ± × − erg cm − s − . Swift -XRT
The XRT data were processed with standard procedures,filtering, and screening criteria by using the xrtpipelinev0.13.0 included in the
HEASoft package (v6.15) . The datawere collected in photon counting mode for all the observa-tions. The source count rate was low ( < − );thus pile-up correction was not required. The data collectedin observations separated by less than twenty-four hours (i.e.2010 May 11, obsid: 31404008 and 31404009; 2011 May 7,obsid: 31404010 and 31404011; 2013 December 16–17, ob-sid: 31404013 and 31404014) were summed in order to haveenough statistics to obtain a good spectral fit. Source eventswere extracted from a circular region with a radius of 20 pix-els (1 pixel ∼ xrtmkarf , and accountfor different extraction regions, vignetting and point spreadfunction corrections. We used the spectral redistribution ma-trices in the Calibration database (CALDB) maintained byHEASARC . The spectra were rebinned with a minimumof 20 counts per energy bin to allow for χ spectrum fit-ting. Bad channels, including zero-count bins, were ignoredin the fit. We have fitted the spectrum using Xspec withan absorbed power-law using the photoelectric absorptionmodel tbabs (Wilms et al. 2000), with a neutral hydrogencolumn density fixed to its Galactic value (1.63 × cm − ;Kalberla et al. 2005). The results are reported in Table 2.All errors are given at the 90% confidence level. Symmetricerrors are reported, obtained by averaging the positive andnegative errors calculated with Xspec .For the observations performed on 2013 December 16–17 and 2014 April 18 there is enough statistic for testinga more detailed spectral model with respect to a simplepower law. For the 2013 December observations, using abroken power-law the fit results in Γ = 0.95 +0 . − . below thebreak energy E break = 2.40 +0 . − . keV and Γ = 1.32 +0 . − . above E break . The fit with a broken power-law ( χ /d.o.f =129/137) does not improve with respect to a simple powerlaw ( χ /d.o.f = 143/139). For the 2014 April observationusing a broken power-law the fit results in Γ = 0.97 ± E break = 2.76 +1 . − . keV andΓ = 1.34 +0 . − . above E break ( χ /d.o.f = 116/104). The F-test shows an improvement of the fit with respect to thesimple power law ( χ /d.o.f = 126/106) with a probability http://heasarc.nasa.gov/lheasoft/ https://heasarc.gsfc.nasa.gov/ https://heasarc.gsfc.nasa.gov/xanadu/xspec/manual/manual.html of 97.9%, indicating that the broken power-law is the best-fitting model. Swift -UVOT
UVOT data in the v , b , u , w m
2, and w uvotsource task included in the HEASoft package (v6.15) and the 20130118 CALDB-UVOTA release.Source counts were extracted from a circular region of 5arcsec radius centred on the source, while background countswere derived from a circular region with 10 arcsec radius ina nearby, free region. The observed magnitudes are reportedin Table 3. Upper limits at 90 per cent confidence level arecalculated using the UVOT photometric system when theanalysis provided a significance of detection < σ . XMM-NEWTON : DATA ANALYSIS ANDRESULTS
XMM-Newton (Jansen et al. 2001) observed PKS 2149 − XMM-Newton
Science Analysis Sys-tem (
SAS v14.0.0 ), applying standard event selection andfiltering. Inspection of the background light curves showedthat no strong flares were present during the observation,with good exposure times of 20, 24 and 24 ks for the pn,MOS1 and MOS2, respectively. For each of the detectorsthe source spectrum was extracted from a circular region ofradius 30 arcsec centred on the source, and the backgroundspectrum from a nearby region of radius 30 arcsec on thesame chip. All the spectra were binned to contain at least20 counts per bin to allow for χ spectral fitting.All spectral fits were performed over the 0.3–10 keV en-ergy range using XSPEC v.12.8.2 . The energies of spectralfeatures are quoted in the source rest frame, while plots arein the observer frame. All errors are given at the 90% confi-dence level. The data from the three EPIC cameras were ini-tially fitted separately, but since good agreement was found( < tbabs model. Theresults of the fits are presented in Table 4. As reported alsoin Ferrero & Brinkmann (2003) and Bianchin et al. (2009),a simple power-law model is sufficient to describe the data,although some residuals are present (Fig. 3). The flux ob-served by XMM-Newton in the 0.3–10 keV energy range is afactor of 2–3 lower than those observed by
Swift -XRT during2005–2014.A broken power-law does not improve the fit and the as-sociated uncertainties on photon index and flux are largerthan those from a fit with a simple power-law (Table 4). Inorder to check for the presence of intrinsic absorption, a neu-tral absorber at the redshift of the source was added to thismodel, but it did not improve the fit quality and thus is notrequired. Moreover, no Iron line was detected in the spec-trum, in agreement with Page et al. (2004). The 90% upperlimit on the equivalent width (EW) of a narrow emissionline at 6.4 keV is EW <
17 eV. c (cid:13) , 000–000 F. D’Ammando, M. Orienti
Table 3.
Observed magnitudes obtained by
Swift -UVOT for PKS 2149 − < σ .Date (UT) Date (MJD) v b u w m w ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± > > ± ± ± ± > > ± ± ± ± > > ± ± ± ± > > ± ± ± ± > ± ± ± ± ± > ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± > > ± ± ± ± > ± ± ± ± ± > ± Table 4.
Summary of fits to the 0.3–10 keV
XMM-Newton spec-trum of PKS 2149 − break are given in units of erg cm − s − and keV, respectively.Model Parameter ValuePower law Γ 1 . ± . . ± . × − χ / d . o . f . . ± . E break . +1 . − . Γ . +0 . − . Flux (0.3–10 keV) (7 . ± . × − χ / d . o . f . no r m a li z ed c oun t s s − k e V − χ Energy (keV)
Figure 3.
EPIC spectra and residuals of PKS 2149 −
306 fittedwith a power law model. NUSTAR : DATA ANALYSIS AND RESULTS
NuSTAR (Harrison et al. 2013) observed PKS 2149 − NuSTAR
Data AnalysisSoftware ( nustardas ) package (v1.4.1). Cleaned event files(level 2 data products) were produced and calibrated us-ing standard filtering criteria with the
NUPIPELINE task andversion 20140414 of the calibration files available in the
NuS-TAR
CALDB. Spectra of the sources were extracted fromthe cleaned event files using a circle of 20 pixel (49 arcsec)radius, while the background was extracted from two dis-tinct nearby circular regions of 50 pixel radius. The ancillaryresponse files were generated with the numkarf task, apply-ing corrections for the point spread function losses, exposuremaps and vignetting. The spectra were rebinned with a min-imum of 20 counts per energy bin to allow for χ spectrumfitting. All errors are given at the 90% confidence level.By fitting the NuSTAR spectrum in the 3–76 keV en-ergy range a good fit was obtained using a simple powerlaw for both the observations ( χ /d.o.f. = 797/805 and735/747), with photon index Γ X = 1.37 ± X = 1.46 ± with a broken power-law model over the 0.3–76 keVenergy range (see Section 5.1). NuSTAR and
Swift -XRT analysis
Simultaneously to
NuSTAR observations,
Swift -XRT obser-vations were performed on 2013 December 16–17 and on2014 April 18. This allows us to study the X-ray spectrumof PKS 2149 −
306 over a wide energy range, i.e. 0.3–76 keV.The results of the simultaneous fits of the
NuSTAR and
Swift -XRT data are presented in Table 6. The photoelectricabsorption model tbabs , with a neutral hydrogen columndensity fixed to its Galactic value (1.63 × cm − ) was in-cluded in all fits. To account for the cross-calibration be-tween NuSTAR -FPMA,
NuSTAR -FPMB, and
Swift -XRT a We ignored the zero-variance bins in the spectrum, i.e. the 76–79 keV energy range. c (cid:13) , 000–000 igh-energy properties of the FSRQ PKS 2149 − Table 5.
Summary of the results for the fits of the 3.0–76 keV
NuSTAR spectra collected on 2013 December 17 and 2014 April 18.Date Photon index Flux 3.0–76 keV χ /d.o.f.(UT) (Γ X ) ( × − erg cm − s − )2013-Dec-17 1 . ± .
01 11 . ± . . ± .
01 8 . ± . Table 6.
Summary of the results for the fits of the 0.3–76 keV
Swift -XRT and
NuSTAR spectra collected on 2013 December 16–17 (Obs1) and 2014 April 18 (Obs 2). All fits also included absorption fixed at the Galactic value.Model Parameter Value (Obs 1) Value (Obs 2)power law Γ 1 . ± .
01 1 . ± . χ / d . o . f . . +0 . − . . +0 . − . E break (keV) 2 . +0 . − . . +0 . − . Γ . ± .
01 1 . ± . χ / d . o . f . . ± .
01 1 . ± . zH (cm − ) 1 . +0 . − . × . +0 . − . × χ / d . o . f . −4 −3 no r m a li z ed c oun t s s − k e V − χ Energy (keV)
Figure 4.
NuSTAR (red and black points) and
Swift -XRT (greenpoints) spectra and residuals of PKS 2149 −
306 collected on 2013December 16–17, simultaneously fitted with a broken power-law. constant factor was included in the model, frozen at 1 forthe FPMA spectra and free to vary for the FPMB and XRTspectra. The X-ray spectrum of the source is not well fit-ted by a simple power law model in both the observations( χ /d.o.f. = 1013/943 and 983/851, for the first and secondobservation, respectively), while a broken power-law modelyielded a good fit ( χ /d.o.f. = 924/941 and 834/849). Theresult of fitting a broken power-law to the spectrum collectedon 2013 December 16–17 is shown in Fig. 4. In this modelthe power law breaks from a slope of Γ = 0 . +0 . − . (Γ =0 . +0 . − . ) below E break = 2 . +0 . − . keV (2 . +0 . − . keV) toΓ = 1 . ± .
01 (Γ = 1 . ± .
01) for the first (second)observation (Table 6). The difference of the cross-calibrationfor the FPMB spectra with respect to FPMA spectra is 1–3 per cent, while for the XRT spectra is less than 10 percent. These differences become larger (10–30 per cent) when a single power law model is used. By applying an F-test, theimprovement of the fit with a broken power-law is significantwith respect to a single power law, with a probability thatthe null hypothesis is true of 1.6 × − and 5 × − forthe first and second observation, respectively. These resultsare in agreement with those reported in Tagliaferri et al.(2015). By adding an extra absorption component at theredshift of the source ( ztbabs ) to the single power law, themodel provides a good fit to the spectrum, with an equiva-lent hydrogen column density of ∼ cm − , but the qual-ity of the fit is worse than the broken power-law model inboth spectra ( χ /d.o.f. = 966/942 and 895/850; Table 6).Sambruna et al. (2007) reported an equivalent hydrogen col-umn density obtained by the fit of Swift
XRT and BAT spec-tra of 0.25 +0 . − . × cm − , that is lower than the values ob-tained by fitting the Swift -XRT and
NuSTAR spectra, buttheir statistics was significantly lower than that presentedhere. γ -ray properties PKS 2149 −
306 was not associated with a γ -ray source,either in the LAT bright source list obtained after threemonths of Fermi operation (Abdo et al. 2009) or in the First
Fermi
LAT source catalogue (Abdo et al. 2010a), indicat-ing that its γ -ray activity was low during the first year of Fermi operation. On the other hand, this FSRQ is asso-ciated with 2FGL J2151.5 − − Fermi
LAT source catalogues(Nolan et al. 2012; Acero et al. 2015). The source is not in-cluded in the First
Fermi
LAT Catalog of Sources above 10 c (cid:13) , 000–000 F. D’Ammando, M. Orienti
Figure 5. γ -ray photon index vs apparent isotropic luminosity inthe 0.1–100 GeV energy range for the blazars with z > Swift -BAT, respectively. The starrepresents PKS 2149 − GeV (Ackermann et al. 2013). During the period 2008 Au-gust 4–2014 August 4 the γ -ray spectrum of PKS 2149 − α = 2.36 ± β = 0.29 ± ± × − ph cm − s − .The source showed a significant increase in its γ -ray fluxin 2011 February, and subsequently a strong γ -ray flare oc-curred in 2013 January (D’Ammando & Orienti 2013). Theflux in 2013 January is about a factor of 8 higher than theaverage flux estimated over 6 years of Fermi observations,with a significant change of the spectral slope ( α = 1.99 ± β = 0.28 ± γ -ray spectrumis well described by a LP model. Considering the extragalac-tic background light (EBL) model discussed in Finke et al.(2010), at the redshift of PKS 2149 −
306 the optical depthshould be τ ∼ z > Fermi -LAT during a γ -ray flare up to now. A significant in-crease of the flux together with a spectral evolution in γ rayswas observed for the high-redshift FSRQ TXS 0536+145(Orienti et al. 2014) and S5 0836+710 (Akyuz et al. 2013).In constrast, no significant spectral hardening was observedduring the γ -ray flares for the high-redshift gravitationallylensed blazar PKS 1830 −
211 (Abdo et al. 2015).During the 2013 flaring activity of PKS 2149 − = (88 ± × − to F = (385 ± × − phcm − s − within ∆t = 12 h, giving a flux doubling time-scaleof τ d = ∆t × ln 2/ ln(F /F ) ≃ τ d /ln2 ≃ lc ∼ r g / c = G M / c ∼ . × M h, where r g is the gravita-tional radius, M = (M/10 ) M ⊙ is the black hole (BH)mass, and c the speed of light (e.g., Begelman et al. 2008).In the case of PKS 2149 − × M ⊙ (Tagliaferri et al. 2015), we obtain a t lc of ∼ h , com-patible with the minimum variability detected in the LATlight curve during 2013 January. This short time variabil-ity observed in γ rays constrains the size of the emittingregion to R < ct var δ/ (1 + z ) = 2.7 × cm (assuming δ = 14, Tagliaferri et al. 2015). This small size of the emit-ting region should correspond to a small distance from thecentral BH, putting the emitting region inside the broad-line region (BLR). This extremely small size is rather diffi-cult to accommodate in the ‘far dissipation’ scenario (e.g.,Tavecchio et al. 2010), where the external Compton scat-tering off the infrared photons from the torus is the maincomponent that produces the high-energy emission, at leastduring flaring activity. This is not in contrast to the SEDmodelling of PKS 2149 −
306 presented in Tagliaferri et al.(2015), where low γ -ray activity contemporaneous to the NuSTAR observations was considered. In fact, different ac-tivity states of the same source may have different γ -rayemitting region locations. In the case of the 2011 Februaryflare the statistics are not good enough to determine theflare shape.High-redshift blazars tend to be the most luminousAGN due to their preferential selection by the LAT causedby Malmquist bias (Ackermann et al. 2015). The daily peakflux observed on 2013 January 20 is (301 ± × − phcm − s − , corresponding to an apparent isotropic γ -ray lu-minosity of (1.5 ± × erg s − . On a 6-h time-scale,the flux reached a peak of (385 ± × − ph cm − s − ,corresponding to an apparent isotropic γ -ray luminosity of(1.9 ± × erg s − . As a comparison, the average γ -rayluminosity over 6 years of Fermi operation is 4.4 × ergs − . The peak values are comparable to the highest lumi-nosity observed from FSRQ so far (i.e., 3C 454.3 and PKS1830 − γ -ray flaring blazarobserved by Fermi -LAT up to now (Orienti et al. 2014). InFig. 5, we compare this value with the γ -ray luminosity ofall blazars with z > Swift -BAT in hardX-rays. In particular, in the 70-month
Swift -BAT catalogue(Baumgartner et al. 2013) there are 17 blazars with redshift z >
2. The filled and open triangles in Fig. 5 representthe blazars detected and not detected by
Swift -BAT, re-spectively. The γ -ray luminosity, L γ , is computed followingGhisellini et al. (2009): L γ = 4 πd S γ (1 + z ) − Γ γ (1) c (cid:13) , 000–000 igh-energy properties of the FSRQ PKS 2149 − Figure 6.
X-ray photon index from
Swift -BAT vs γ -ray photonindex from Fermi -LAT of the 10 high-redshift blazars detectedby both instruments. The filled upside down triangle representsPKS 2149 − where S γ is the energy flux between 100 MeV and 100 GeV,and Γ γ is the photon index.All these high-redshift blazars are FSRQ, with the ex-ception of SDSS J145059.99+520111.7, PMN J0124-0624,MG4 J000800+4712, and PKS 0437-454 classified as BL Lacobjects. Considering the average luminosity, PKS 2149 − −
211 and PKS 0537 − Swift -BAT catalogue,we note that all the high-redshift blazars detected by both
Fermi -LAT and
Swift -BAT have L γ > × erg s − , sug-gesting that only the most luminous γ -ray blazars are de-tected by both instruments. Most of the LAT sources de-tected by BAT, including PKS 2149 − γ -rayphoton index Γ γ > .
5. This corresponds in hard X-rays toa photon index Γ X < . Fermi -LAT (seee.g., Tagliaferri et al. 2015). However, during strong flaringactivity the IC peak may shift to higher energies as in thecase of the 2013 flare from PKS 2149 − Fermi -LAT have L γ > erg s − . The vast majority of themhave a BH mass > M ⊙ (Ghisellini et al. 2009, 2010,2011, 2014b), confirming that the most powerful blazarshave the heaviest BH (Ghisellini et al. 2013). In particular,PKS 2149 −
306 has a BH mass of 3.5 × , as estimated byTagliaferri et al. (2015). Figure 7.
X-ray photon index vs 0.3–10 keV flux of PKS2149 −
306 during 2001–2014. Filled circles are
Swift -XRT observa-tions; the open triangle represents the
XMM-Newton observation.
We investigated the X-ray properties of PKS 2149 −
306 bymeans of
Swift -XRT,
XMM-Newton , and
NuSTAR obser-vations. The X-ray spectrum collected by
XMM-Newton in2001 is quite well modelled by a simple power law with aphoton index of Γ X = 1.45 ± × − erg cm − s − . During 2005 December–2014 April, Swift -XRT observed the source with a 0.3–10 keV flux in therange (1.2–2.8) × − erg cm − s − , with a photon indexvarying between 1.0 and 1.5. Fig. 7 shows the X-ray photonindex estimated from Swift -XRT and
XMM-Newton obser-vations as a function of the X-ray flux in the 0.3–10 keVrange: despite the large errors, a hint of hardening of thespectrum with the increase of the flux is observed.Unfortunately the
Swift observations did not cover the γ -ray flaring periods detected in 2011 February and 2013January, preventing us from investigating the X-ray be-haviour during these γ -ray flaring events.The NuSTAR spectra collected in the 3–76 keV energyrange during 2013 December 17 and 2014 April 18 are wellfitted by a simple power law with photon index Γ X = 1.37 ± X = 1.46 ± −
306 by
Swift -XRT and
NuSTAR showed that the broad band X-ray spectrum is welldescribed by a broken power-law model, with a very hardspectrum (Γ ∼
1) below the break energy, at E break = 2.5–3.0 keV, and Γ = 1.37 ± ± Swift -BAT and
BeppoSAX observed a photonindex Γ X = 1.50 ± X = 1.40 ± obtained by the two Swift -XRT and
NuS-TAR joint fits. The 3–76 keV flux varied by about 40 percent between the first and second
NuSTAR observation. Atthe same time the 0.3–10 keV flux varied by less than 10per cent. In the same way the photon index below the break c (cid:13) , 000–000 F. D’Ammando, M. Orienti
Figure 8.
X-ray photon index vs X-ray apparent isotropic lu-minosity in the 14–195 keV energy range for the blazars with z> Swift -BAT. The filled and open triangles rep-resent the objects detected and not detected by
Fermi -LAT, re-spectively. The star represents PKS 2149 − energy did not change, while the photon index above thebreak energy was harder when the source was brighter.In several high-redshift ( z >
4) blazars a steepening ofthe soft X-ray spectrum has been observed (e.g., Yuan et al.2006, and references therein). This steepening may be dueto either an excess of absorption in the soft X-ray part of thespectrum or due to an intrinsic curvature of the electron en-ergy distribution responsible for the X-ray emission. In PKS2149 −
306 this feature was observed below 3 keV, and animprovement of the fit is observed when an extra absorberat the redshift of the source (N z H ∼ cm − ) is addedto the simple power law model. However, this improvementis not as good as when we use a broken power-law model.The AGN may be surrounded by a dense plasma in form of awind or an outflow (e.g., Fabian 1999). This intervening ma-terial may be responsible for the extra absorption observedin high-redshift quasars (e.g., Vignali et al. 2005). However,in contrast to radio-loud quasars, it is unlikely that for ablazar like PKS 2149 −
306 such a large gas column densityin the line of sight is not removed by the relativistic jet thatshould be well aligned with the line of sight. Moreover, witha hydrogen column density of about 10 cm − obtainedby fitting the X-ray spectrum the corresponding optical ob-scuration (see e.g., Guver & Ozel 2009) would be very high( A V ∼
10) and the source would not be detectable in optical-UV with the short exposures of the
Swift observations. Thisproblem may be solved by invoking a high ionization stateof the gas, but it is not possible to have conclusive evidencefrom its X-ray spectrum. In fact, due to the redshift of thissource the most important spectral features used as a di-agnostic are out of the energy range covered by
Swift and
XMM-Newton .The most likely explanation of the steepening of thespectrum is that the soft X-ray emission is produced by ex-ternal Compton radiation from the electrons at the lower end of the energy distribution (e.g., Tavecchio et al. 2007). Thisis in agreement with the stable photon index below 3 keVand the change of the photon index above ∼ XMM-Newton in 2001 may be relatedto the lower flux with respect to that estimated during the
Swift -XRT and
NuSTAR observations. The intrinsic curva-ture in soft X-rays would be less evident during the low ac-tivity of the source. The situation may be more complex dueto the possible presence in the X-ray band also of the syn-chrotron self-Compton emission (SSC). However, the rela-tive importance of the SSC component should decrease withthe luminosity of the source (e.g., Ghisellini et al. 1998),and therefore should be negligible in powerful FSRQ suchas PKS 2149 − −
306 in-cluding the
Swift -XRT and
NuSTAR data are better fittedby considering the emitting region outside the BLR (e.g.,Tagliaferri et al. 2015), where the contribution of the bulkComptonization should be negligible.Considering the blazars included in the 70-month
Swift
BAT catalogue with a redshift z >
2, 7 out of 17 have notbeen detected by
Fermi -LAT so far. Only sources with aΓ X < . γ rays, while no dependenceon the X-ray luminosity seems to be evident (Fig. 8). This isconfirmed by the fact that the average photon index of theFSRQ detected by LAT, < Γ LATX > = 1.42 ± < Γ no LATX > = 1.85 ± z> Swift -BAT, and another one, IGRJ12319 − INTEGRAL -IBIS (Bassani et al. 2012). Inaddition, two blazars at redshift z >
NuSTAR . Only two of these ten blazars have been detectedby
Fermi -LAT: PKS 0537 −
286 (e.g., Bottacini et al. 2010)and TXS 0800+618 (e.g., Ghisellini et al. 2010), confirmingthat the γ -ray energy range is not ideal for detecting blazarsat redshift >
3. Among the FSRQ detected by both BAT andLAT, PKS 2149 −
306 is the third most luminous after PKS1830 −
211 (e.g., Abdo et al. 2015) and B2 0743+25 (e.g.,Ghisellini et al. 2010).
In powerful high-redshift blazars, such as PKS 2149 − Swift -UVOT is 0.6, 0.7, 1.1, 0.8, 0.6, and 0.6 mag(corresponding to a variation of the flux density of 1.5, 1.8,2.8, 2, 1.5, 1.5) from the v to the w c (cid:13) , 000–000 igh-energy properties of the FSRQ PKS 2149 − In this paper we discussed the γ -ray and X-ray properties ofthe high-redshift FSRQ PKS 2149 −
306 by means of
Fermi -LAT,
NuSTAR , XMM-Newton , and
Swift data. We summa-rize our main conclusions as follows: • PKS 2149 −
306 showed a significant increase in its γ -ray activity in 2011 February and 2013 January. During the2013 flare the flux increase was accompanied by a significantchange of the spectral slope, not observed during the 2011flare. • During the 2013 γ -ray flaring activity significant fluxvariations are observed on a 6-hr time-scale, compatible withthe light crossing time of the event horizon of the super-massive black hole (SMBH). On 2013 January 20, the sourcereached a daily γ -ray peak flux of (301 ± × − ph cm − s − , up to (385 ± × − ph cm − s − on a 6-hr time-scale. These values correspond to an apparent isotropic γ -ray luminosity of (1.5 ± × and (1.9 ± × ergs − , respectively, comparable to the highest values observedfrom FSRQ up to now. • The average γ -ray luminosity of PKS 2149 −
306 over 6years of
Fermi operation is 4.4 × erg s − . This is thethird brightest blazar with z > −
211 and PKS 0537 − • All high-redshift blazars detected by both
Fermi -LATand
Swift -BAT have a L γ > × erg s − , suggesting thatonly the most luminous γ -ray blazars are detected by bothinstruments. Like most of the LAT blazars detected by BAT,PKS 2149 −
306 has a soft γ -ray photon index Γ γ > .
5. Thiscorresponds to a photon index Γ X < . • Among the FSRQ with z > −
306 is the third most luminous in hardX-rays after PKS 1830 −
211 and B2 0743+25. • The broad band X-ray spectrum of PKS 2149 −
306 ob-served by
Swift -XRT and
NuSTAR is well described by abroken power-law model, with a very hard spectrum (Γ ∼
1) below the break energy, at E break = 2.5–3.0 keV, andΓ ∼ . • The steepening of the spectrum below ∼ NuSTAR and
Swift joint observations. An extra absorption due tomaterial surrounding the SMBH is unlikely because the rel-ativistic jet should efficiently remove the gas along the lineof sight. Moreover, this extra absorption should correspondto a very large extinction in optical and UV, in contrast tothe detection of the source by
Swift -UVOT. • Fermi -LAT and
Swift -BAT observations are confirmingthat the hard X-ray band is more effective in selecting brightFSRQ at z > γ -ray band is very effective up to z = 2 (seee.g., Ackermann et al. 2015).Further multiwavelength observations of PKS2149 −
306 will be important for shedding light on theproperties of high − z blazars. In particular, simultaneousoptical-to-X-ray observations during a γ -ray flaring activitywill allow us to compare its broad-band spectral energy distribution during both low and high activity states,constraining the emission mechanisms at work. ACKNOWLEDGEMENTS
The
Fermi
LAT Collaboration acknowledges generous ongo-ing support from a number of agencies and institutes thathave supported both the development and the operation ofthe LAT as well as scientific data analysis. These includethe National Aeronautics and Space Administration andthe Department of Energy in the United States, the Com-missariat `a l’Energie Atomique and the Centre National dela Recherche Scientifique / Institut National de PhysiqueNucl´eaire et de Physique des Particules in France, the Agen-zia Spaziale Italiana and the Istituto Nazionale di Fisica Nu-cleare in Italy, the Ministry of Education, Culture, Sports,Science and Technology (MEXT), High Energy AcceleratorResearch Organization (KEK) and Japan Aerospace Explo-ration Agency (JAXA) in Japan, and the K. A. WallenbergFoundation, the Swedish Research Council and the SwedishNational Space Board in Sweden. Additional support forscience analysis during the operations phase is gratefullyacknowledged from the Istituto Nazionale di Astrofisica inItaly and the Centre National d’´Etudes Spatiales in France.Part of this work was done with the contribution ofthe Italian Ministry of Foreign Affairs and Research for thecollaboration project between Italy and Japan. We thankthe
Swift team for making these observations possible, theduty scientists, and science planners. This research has madeuse of the
NuSTAR
Data Analysis Software (NuSTARDAS)jointly developed by the ASI Science Data Center (ASDC,Italy) and the California Institute of Technology (USA). Wethank Eugenio Bottacini, Luca Baldini, and Jeremy Perkinsfor useful comments and suggestions.
REFERENCES
Abdo A. A. et al., 2009, ApJ, 707, L142Abdo A. A. et al., 2010a, ApJS, 188, 405Abdo A. A. et al., 2010b, ApJ, 721, 1425Abdo A. A. et al., 2015, ApJ, 799, 143Acero F. et al., 2015, ApJS, 218, 23Ackermann M. et al., 2010, ApJ, 721, 1383Ackermann M. et al., 2012, ApJ, 747, 104Ackermann M. et al., 2013, ApJS, 209, 34Ackermann M. et al., 2015, ApJS, 810, 14Ajello M., Alexander D. M., Greiner J., Madejski G. M.,Gehrels N., Burlon D., 2012, ApJ, 749, 21Atwood W. B. et al., 2009, ApJ, 697, 1071Akyuz A. et al., 2013, A&A 556, A71Barthelmy S. D. et al., 2005, SSRv, 120, 143Bassani L. et al., 2012, A&A, 543A, 1Baumgartner W. H., Tueller J., Markwardt C. B., Skin-ner G. K., Barthelmy S., Mushotzky R. F., Evans P. A.,Gehrels N., 2013, ApJS, 207, 19Beckmann V. et al., 2009, A&A, 505, 417Begelman M. C., Fabian A. C., Rees J. M., 2008, MNRAS,384, L19Bianchin V. et al., 2009, A&A, 496, 423Bottacini E. et al., 2010, A&A, 509, 69 c (cid:13) , 000–000 F. D’Ammando, M. Orienti
Burrows D. N. et al., 2005, SSRv, 120,165Cappi M., Matsuoka M., Comastri A., Brinkmann W.,Elvis M., Palumbo G. G. C., Vignali C., 1997, ApJ, 478,492Celotti A., Ghisellini G., Fabian A. C., 2007, MNRAS, 375,417D’Ammando F., & Orienti M., 2012, The Astronomer’sTelegram, 4706Elvis M., Fiore F., Siemiginowska A., Bechtold J., MathurS., McDowell J., 2000, ApJ, 543, 545Fabian A. C., 1999, MNRAS, 308, L39Fang T. et al., 2001, ApJ, 555, 356Ferrero E., & Brinkmann E., 2003, A&A, 402, 465Finke J. D., Razzaque S., Dermer C. D., 2010, ApJ, 712,238Fossati G., Maraschi L., Celotti A., Comastri A., GhiselliniG., 1998, MNRAS, 299, 433Gehrels N. et al., 2004, ApJ, 611, 1005Ghisellini G., Celotti A., Fossati G., Maraschi L., ComastriA., 1998, MNRAS, 301, 451Ghisellini G., Maraschi L., Tavecchio F., 2009, MNRAS,396L, 105Ghisellini G. et al., 2010, MNRAS, 405, 387Ghisellini G. et al., 2011, MNRAS, 411, 901Ghisellini G., 2013, proceedings of “X-ray Astronomy: to-wards the next 50 years!” (Milano, 1-5 October 2012)[arXiv:1302.0011]Ghisellini G., Haardt F., Della Ceca R., Volonteri M., Sbar-rato T., 2013, MNRAS, 432, 2818Ghisellini G., Sbarrato T., Tagliaferri G., Foschini L.,Tavecchio F., Ghirlanda G., Braito V., Gehrels N., 2014a,MNRAS, 440, L111Ghisellini G., Tavecchio F., Maraschi L., Celotti A., Sbar-rato T., 2014b, Nature, 515, 376Guver T., Ozel F., 2009, MNRAS, 400, 2050Harrison F. A. et al., 2013, ApJ, 770, 103Jansen F. et al., 2001, A&A, 365, L1Johnston K. J. et al., 1995, AJ, 110, 880Kalberla P. M. W., Burton W. B., Hartmann D., Arnal E.M., Bajaja E., Morras R., P¨oppel W. G. L., 2005, A&A,440, 775Komatsu E. et al., 2011, ApJS, 192, 18Mattox J. R. et al., 1996, ApJ, 461, 396Nolan P. et al., 2012, ApJS, 199, 31Orienti M., D’Ammando F., Giroletti M., Finke J., AjelloM., Dallacasa D., Venturi T., 2014, MNRAS, 444, 3040Page K. L., O’Brien P. T., Reeves J. N., Turner M. J. L.,2004, MNRAS, 347, 316Romani R. W., Sowards-Emmerd D., Greenhill L., Michel-son P., 2004, ApJ, 610, L9Romani R. W., 2006, AJ, 132, 1959Roming P. W. A. et al., 2005, SSRv, 120, 95Sambruna R. M., Tavecchio F., Ghisellini G., Donato D.,Holland S. T., Markwardt C. B., Tueller J., MushotzkyR. F., 2007, ApJ, 669, 884Sbarrato T. et al., 2013, ApJ, 777, 147Siebert J., Matsuoka M., Brinkmann W., Cappi M., MiharaT., Takahashi T., 1996, A&A, 307, 8Tagliaferri G. et al., 2015, ApJ, 807, 167Tavecchio F., Maraschi L., Ghisellini G., Kataoka J., Fos-chini L., Sambruna R. M., Tagliaferri G., 2007, ApJ, 665,980 Tavecchio F., Ghisellini G., Bonnoli G., Ghirlanda G., 2010,MNRAS, 405, L94Vignali C., Brandt W. N., Schneider D. P., Kaspi S., 2005,AJ, 129, 2519Yaqoob T., George I. M., Nandra K., Turner T. J., ZobairS., Serlemitsos P. J., 1999, ApJ, 525, L9Yuan W., Fabian A. C., Worsley M. A., McMahon R. G.,2006, MNRAS, 368, 985Wilkes B. J., 1986, MNRAS, 218, 331Wilms J., Allen A., McCray R., 2000, ApJ, 542, 914 c (cid:13)000