Mon.Not.R.Astron.Soc.000,000–000(0000) Printed2February2008 (MNLATEXstylefilev2.2) The Compton-thick quasar at the heart of the high-redshift giant radio galaxy 6C0905+39 M. C. Erlund,1⋆ A. C. Fabian1, Katherine M. Blundell2 and Carolin S. Crawford1. 1InstituteofAstronomy,MadingleyRoad,CambridgeCB30HA 8 2UniversityofOxford,Astrophysics,KebleRoad,OxfordOX13RH 0 0 2 n 2February2008 a J 2 ABSTRACT 2 OurXMM-Newtonspectrumofthegiant,high-redshift(z = 1.88)radiogalaxy6C0905+39 showsthatitcontainsoneofthemostpowerful,high-redshift,Compton-thickquasarsknown. ] h Itsspectrumisveryhardabove2keV.ThesteepXMM-Newtonspectrumbelowthatenergyis p showntobeduetoextendedemissionfromtheradiobridgeusingChandradata.Thenucleus - of6C0905+39hasacolumndensityof3.5+1.4×1024cm−2andabsorption-correctedX-ray o −0.4 r luminosityof1.7+−00..91×1045ergs−1 inthe2−10keVband.Alowerredshiftactivegalaxy t inthesamefield,SDSSJ090808.36+394313.6,mayalsobeCompton-thick. s a [ 1 INTRODUCTION sorption along the line-of-sight istaken to be 1.91×1020cm−2 (Dickey&Lockman1990). 1 Thepowerful,classicaldoubleFRII(Fanaroff&Riley1974)radio v galaxy 6C0905+39 is unusual as it is one of the highest redshift 1 4 (z = 1.88) giant radio galaxies known, spanning 111arcsec on 2 DATAREDUCTION 3 thesky(Law-Greenetal.1995)correspondingtoaprojectedsize 3 of945kpcinthecosmologyassumedinthispaper.Radiogalax- XMM-Newton observed 6C0905+39 on 2006 October 30 for 1. ies have long been associated with obscured, (optically defined) 58.0ks with EPIC-pn and 61.9ks with each of the EPIC-MOS 0 Type 2 quasars, because they are considered to be the same ob- cameras.Thedatawerereduced usingthestandardpipeline, run- 8 jects as radio-loud quasars seen through an obscuring torus (e.g. ningtheSAStoolsEPCHAINandEMCHAINfortheEPIC-pnand 0 Barthel 1989 and Antonucci 1993). X-ray observations are par- bothEPIC-MOScamerasrespectively.Theresultingfileswerefil- : ticularly efficient at detecting obscured AGN as they provide the teredtoremoveflares,leaving42.2ksofgood-timeforthePN,and v observer with several diagnostics, such as the photoelectric cut- 53.1ksforbothMOS1andMOS2whendead-timeintervalsare i X off, the iron line equivalent width (EW) and the X-ray to optical alsotakenintoaccount. r or [O III] λ5007 ratio (e.g. Bassanietal. 1999 or, for a review, Spectraforthebackground(anareaofskyfreefromsources a Comastri 2004). Sources with an obscuring column greater than near 6C0905+39 on the same chip), the central source (within a 1.5×1024cm−2,atwhichtheThomsonscatteringopticaldepth 15arcsec circle) and the extended X-ray emission which is dis- reachesunity,areknownasCompton-thickAGN.Itisnon-trivial cussed elsewhere (Erlundetal. 2007) were extracted separately to calculate the column density of such sources as there is little for each instrument and stacked and fittedusing XSPECv11.3.2. (seeWilman&Fabian1999)directemissionbelow10keVmean- GiventhelownumberofcountsintheMOSspectra,onlythePN ing that analysis relies on interpreting the reflected spectrum at spectrawereusedforanalysisofthecore. lowerX-rayenergies.Suzakuisstartingtoprovideawindowinto theX-rayregimeabove10keVenablingdirectdetectionofnuclear emissioninthebrightestCompton-thicksourcessuchasNGC5728 (Comastrietal.2007).SuchsourcesareimportantsincetheX-ray 3 RESULTS background gives strong evidence for a significant population of TheX-rayemissionextendingalongtheradioaxisoverhundreds Compton-thickAGN(e.g.Comastri2004,Worsleyetal.2005). of kpc to the east and west, which was first detected by Chan- In this paper, we present recent XMM-Newton observa- dra and presented in Blundelletal. (2006), is clearly visible in tions of 6C0905+39 (which is located at RA 09h08m16.9s, Dec theXMM-Newtondata(seeFig.1).Intotalthereare2566counts +39d43m26sinJ2000).Throughoutthispaper,allerrorsarequoted (1783 background) inthe source, 2125 (1624 background) inthe at 1σ unless otherwise stated and the assumed cosmology is extended emission and 441 counts (159 background) in the core H0 = 71kms−1Mpc−1, Ω0 = 1 and Λ0 = 0.73. One arc- inthe0.2−8keVbandaftersummingtheMOS1,MOS2andthe second represents 8.518kpc on the plane of the sky at the red- PNdata.ThesmallnumberofcountsinthenucleusinourChandra shift (z = 1.883 ±0.003) of 6C0905+39 and the Galactic ab- observation(Blundelletal.2006)precludedanydetailedanalysis. (cid:13)c 0000RAS 2 M. C. Erlundet al. Figure1.ThreecolourplotofXMM-Newtondata(188′′×116′′imagewith2arcsecpixels;theimagehasbeensmoothedbyaGaussianwithσof2pixels): 0.2−1keVband(red),1−3keVband(green)and3−10keV band(blue)allonasquare-rootscale. Thewhiteoverlaidcontours areVLAA-array 1.425-GHzradiocontours(0.4,1.7and7mJy/beam),displayingthecoreandtwolobesoftheradiosource.Thebluesourcetothewestof6C0905+39is anobscuredAGN(seesection3.3). Allspectraldatadescribedinthepresentpaperwerefittedoverthe 0.5−10keVband. 3.1 Theextendedemission 10-3 TheextendedemissionisdiscussedindetailinErlundetal.(2007); however,webrieflymentionithereinthecontextofitseffectasa potential contaminant of the core spectrum. Fig. 1 clearly shows -1V ke the extended emission from the bridge of 6C0905+39 and that -1c ttehnedneudmembeirssoiofnXi-srnayotpjehtoetomnisssiinocnre(saisnecsettohwisawrdosutlhdebceovreer.yTnhaerreoxw- unts se 10-4 o and sowouldnot beresolved byChandra), but inverse-Compton c up-scatteringofthecosmicmicrowavebackgroundbyspentradio plasmafromthelobes(Blundelletal.2006). Thebridgespectra(excluding a15arcsec regionaround the core) were extracted from the EPIC-pn, MOS1 and MOS2 data partial covering model andwerefittedwithaGalactic-absorbedpower-lawgivingapho- power law model ton index of Γ = 1.75+0.27 and a normalisation of 2.5+0.3 × 10-5 −0.24 −0.3 10−6ctkeV−1cm−2s−1withareduced-chisquaredofχ2ν =0.9 0.5 1 2 5 with 90 degrees of freedom (d.o.f.). The absorption-corrected X- Energy [keV] rayluminosityoftheextendedemissioninthe2−10keVbandis LX =1.8+−00..33×1044ergs−1. Figure2.Thespectrumofthecoreof6C0905+39.Thesolidlineshowsthe bestfitpartialcoveringmodelandthedottedlineshowsthecontinuationof thedirectlyobservedpower-lawcomponent. 3.2 Thecore OnlythePNdatahavebeenusedinthefollowingfitsbecauseeach MOSspectrumcontainsroughlyaquarterofthecountsinthePN an excess in the hard X-rays which can, in principle, be repro- spectrum but a similar amount of noise. Spectral analysis shows ducedbycoldreflection(XSPECmodelwabs*zwabs*[powerlaw thattheMOSspectraareconsistentwiththePNspectrum. +pexrav]).Thismodelisnotagoodfithowevergivingχ2 =2.2 ν The XMM-Newton spectrum of the core is strikingly hard. for7d.o.f.(χ2 = 15.4) withanimplausibly largeamount ofre- A simple Galactic and intrinsic absorbed power-law is a poor flection(reflectionfraction∼33)andaphotonindexofΓ=1.6. fit with χ2 = 2.7 for 8 d.o.f. (χ2 = 21.7). The best-fit pa- Apartial-coveringmodelwitharedshiftedironlineisagood ν rameters give a photon index of Γ = 0.93, a normalisation of fit(Fig.2).Includinganironline,themodelgivesacoveringfrac- 1.3×10−6ctkeV−1cm−2s−1 andnointrinsicabsorption.The tion of 0.97+0.02, intrinsic absorption of 2.6+1.0 × 1024cm−2, −0.03 −0.7 spectrum (Fig. 2) appears to have a power-law component with photon index Γ = 2.3+0.2 and a normalisation of 5.0+0.7 × −0.3 −0.8 (cid:13)c 0000RAS,MNRAS000,000–000 3 10−5ctkeV−1cm−2s−1 for thepower-law. Theredshifted iron line (modelled as a Gaussian) was fixed at 6.4keV in the rest- frame of the source with σ = 0.1keV, giving an equivalent width EW ∼ 0.9keV; the normalisation of this component is 2.2+1.5 ×10−5ctcm−2s−1. Thisgives χ2 = 0.9 with6d.o.f. −1.4 ν (χ2 = 5.4).Withoutanironline,thefitvaluesaresimilar:acov- ering fraction of 0.95+0.01; an intrinsic absorption of 2.1+1.0 × −0.04 −0.6 1024cm−2; a photon index Γ = 2.1+0.2 and a normalisation of −0.2 3.2+0.6×10−5ctkeV−1cm−2s−1,givingχ2 =1.1for7d.o.f. −3.2 ν (χ2 =7.8).Includinganironlineimprovesthefitby∆χ2 =2.4, sothedatamarginallyfavouritsaddition.Theabsorption-corrected luminosityinthe2−10keVbandforthepartialcoveringmodel with an iron line is LX = 3.2+−00..32 × 1045ergs−1 and the ob- served flux is FX = 3 × 10−14ergcm−2s−1 (the errors are poorly constrained). The column density of 6C0905+39 is that of a Compton-thick AGN and its luminosity implies that it is a quasar.SuchCompton-thickquasarsarerare,Fig.3(adaptedfrom Gandhietal.2006)illustratesthispoint.ItshowsaselectionofX- Figure3.ThisfigureisadaptedfromGandhietal.(2006)(theirFig.7,see rayselectedtype-2quasarswhicharenotreflectiondominated.We theircaptionfordetailsofhowthisplotwasproducedandtherelevantref- infer from this that 6C0905+39 isparticularly intrinsically lumi- erencesforeachsource).TheX-rayluminosityaxisshowstheabsorption- nousandhighlyobscured. correctedluminosityinthe2−10keVbandofeachsource.B20902+39 The paucity of counts in the Chandra data (14 ± 4 in the (B20902,Fabianetal.2002),SWIREJ104406.30+583954.1andSWIRE 0.5−7keV band) means that no analysis was performed on the J104409.95+585224.8(SW104406,SW104409respectively;Pollettaetal. 2006)and6C0905+39havebeenadded.ThisselectionofX-rayselected originaldetectionofthecore(Blundelletal.2006).TheChandra type2quasarsfromtheliteratureexcludessourceswhichareclearlyreflec- dataareconsistentwiththeXMM-Newtondatainasmuchasthey tiondominated.Thisimpliesthat6C0905+39isoneofthemostintrinsi- are incapable of constraining the spectral shape of the nucleus. callyluminousandoneofthemostobscuredquasarsyetdetected. However,inthe2arcsecregionextractedfortheChandraanalysis therearesignificantlyfewersoftphotonsthanintheXMM-Newton data(15arcsecregion).Toquantifythiswemeasuretheobserved count-ratebelow1.5keV,whichis1.5±0.9×10−4counts−1,and compareittothatpredictedfromamodeloftheXMM-Newtondata inthesameband,whichgivesacount-rateof5×10−4counts−1. This deficit is significant at ∼ 4σ level and implies that the soft photonsarefromtheextendedemissionwhichislessofacontam- inantinthesmallerChandraaperture. Giventhat thereisaconsiderable amount ofextended emis- sion from the lobes of the radio source (clearly seen in Fig. 1) thatislikelytobecontaminatingtheXMM-Newtonnucleusspec- trum,wenowtryahighlyobscuredAGNwithapower-lawcom- ponentatlowenergiesfromtheextendedemission.Inordertotest this, we fit a model consisting of Galactic absorption; a power- law with a photon index fixed to the value found for the ex- tended emission (Γ = 1.75, see Erlundetal. 2007); an intrinsi- cally absorbed power-law (with a photon index fixed to Γ = 2 as the data are unable to constrain it); and a redshifted iron line with it energy fixed at 6.4keV and σ = 0.1 (XSPEC model wabs[powerlaw + zwabs(powerlaw + zgauss)]). The best- fitting model is shown in Fig. 4. The intrinsic column density is 3.5+1.4 ×1024cm−2 andthenormalisationofthiscomponent is −0.4 2.7+1.4×10−5ctkeV−1cm−2s−1.Theredshiftedironlinehas −0.1 anormalisationof2.7+11.2×10−4ctcm−2s−1.Thismodelgives −2.5 a good fit to the data with χ2ν = 1.1 for 7 d.o.f. (χ2 = 7.4); Figure4.ThebestfitobscuredAGNwithcontaminationisshowninblack. statisticallythe partial covering model isa slightlybetter fit with ItconsistsofanabsorbedGalacticandintrinsicabsorbedpower-law(green χ2 = 5.4 for 6 d.o.f. which gives ∆χ2 = 2. The observed line); anadditional power-law component fixedtohavethesamephoton flux is FX = 2.8+−10..45 × 10−14ergcm−2s−1. The absorption- indexastheextendedemission(Γ = 1.75,redline)andaredshiftediron correctedluminosityinthe2−10keVbandofthecentralsourceis line(blueline). LX,core = 1.7+−00..91×1045ergs−1.Thenormalisationofthecon- taminatingpower-lawis1.5+0.2×10−6ctkeV−1cm−2s−1and −0.2 itsabsorption-correctedX-rayluminosityinthe2−10keVband is LX,ext = 1.1+−00..12 ×1044ergs−1. The luminosity of the ex- tendedX-rayemissionisLX = 1.9+−00..33×1044ergs−1 implying Compton scattering of the cosmic microwave background comes that roughly a third of the power in the emission due to inverse- fromclosetothecore. (cid:13)c 0000RAS,MNRAS000,000–000 4 M. C. Erlundet al. Themeasuredobscuringcolumnandabsorptioncorrectedlu- V minosity suggests that 6C0905+39 is a genuine Compton-thick e ec/k 10−4 quasar. (Wenote that the column density measured directly from nts/s X-rayspectra depends onthegeometry oftheobscuring medium ed cou 10−5 apsleshtoorwusngbeyomMeattrtyesteael.m1s9r9e9le.vHaonwt.)evTehre,rienathreevperreysefnetwcaCsoemapstiomn-- normaliz 10−6 athgicokoqduCasoamrspktonno-wthnicakthqiugahsarerdcsahnidfti.dNatoerminatnheetCahl.a(n2d0r0a2d)edeepteficteeldd South with z > 3. Apart from such clear examples, very few 1 Compton-thickquasarsareknownespeciallyatredshiftz∼2,de- χ 0 spitethefactthatobscuredAGNareexpectedtomakeup∼30per 1 − centofthehardX-raybackgroundand70percentshouldbefound 1 2 5 between 1 < z < 3, according to populations synthesis models channel energy (keV) (seee.g.Normanetal.2002andGillietal.2001). Figure5.ThePNspectrumofthesecondobscuredAGNinthefield(the Theabsorption-correctedluminosityinthe2−10keVband bluesourcetothewestof6C0905+39inFig.1).Thesolidlineisthebest canbeconvertedtothebolometricluminosityusingresultsderived fitabsorbedpower-lawmodelwitharedshiftedironline. by Vasudevan&Fabian (2007) who found that low and high Ed- dington ratio1 AGN have a bolometric correction of 19±4 and 55±13respectively.Inordertodeterminewhether6C0905+39isa 3.3 AnotherobscuredAGNinthefield? low-orhigh-Eddingtonsource,theblackholemassisestimatedus- ThehardX-raysourcetothewestof6C0905+39(thebluesource ingtheblack-hole-mass–bulge-massrelationship(Magorrianetal. inFig.1locatedatRA09h08m08.3s, Dec+39d43m12s inJ2000 1998) and the stellar mass (measured by Seymouretal. 2007) as coordinates)isanSDSSsourceandso,followingtheSDSSnam- a proxy for the bulge mass. This gives a black hole mass for ingconvention,isSDSSJ090808.36+394313.6.Ithasaspectrum 6C0905+39 ofMBH ∼ 2×109M⊙.Thisisconsistentwiththe consistentwithitbeinganobscuredAGN(seeFig.5)withcolumn value calculated using the method presented in Marconi&Hunt density of NH ∼ 1.28+−00..4316 × 1024cm−2 when the power-law (2003), who make use of the black-hole-mass–bulge-luminosity componentisfixedatΓ=2becauseitisotherwiseunconstrained relation, and the K-band magnitude (from Seymouretal. 2007) (although varying it between reasonable values does not signifi- alongwithacruderedshiftcorrection.Thisassumesthatsuchre- cantly alter the inferred value of the column density). Adding an lationships hold to out to redshift z = 1.88 which is not nec- iron lineredshifted from6.4keV improves thefit from 1.4 for 4 essarily true, but radio galaxies at lower redshifts tend to be the d.o.f. to 0.2 for 3 d.o.f., i.e. by ∆χ2 = 5. The redshift is mea- hosts of similarly massive black holes (e.g. Liuetal. 2006) so suredasz =0.45+0.02.Thereareatotalof261counts(159back- this estimate seems reasonable. Using these estimates for black −0.08 ground)inthe0.2−8keVband,meaningthatthederivedquantities hole mass, 6C0905+39 appears to fall into the high-Eddington are somewhat uncertain; however, as Fig. 1 shows there are very regime(Lbol/LEdd > 0.1) sotheappropriatebolometriccorrec- few photons below 3keV, this source must be a highly obscured tion is 55±13 which implies that the bolometric luminosity of AGN. The 2− 10keV band observed (absorbed) flux is FX = 6C0905+39isLbol =1.0±0.4×1047ergs−1.Thismeansthat 3.1+5.7×10−14ergcm−2s−1andtheabsorption-correctedX-ray itisaverypowerfulsource.Comparedtootherpowerful,obscured −1.0 luminosityinthesamebandisLX =1.3+−00..32×1044ergs−1. quasars(seeFig.3,assumingthatthemeasuredcolumndensityis accurate), 6C0905+39 is amongst the most powerful, (X-ray de- fined)type2quasarscurrentlyknown. Recent research hasshown that high-Eddington objects tend 4 DISCUSSION not to be jetted (e.g. Panessaetal. 2007; Sikoraetal. 2007 and Thespectralfittingofthenucleusof6C0905+39presentedinSec- Maoz 2007). However, 6C0905+39 does not seem to be follow tion3.2showsthatamodel consistingofanobscured quasar nu- this trend as it is an extremely powerful source which is host to cleuswithacontaminatingpower-lawfromthelobesisagoodfit aradiosourcewhichspans945kpcontheskyandappearstostill tothe data. Itscolumn density and luminosity clearlymeans that beactivelyfed.Itsgiantsizeindicatesthatithasbeencontinuously thisisaCompton-thickquasar. activeforatleastthepasttenmillionyears.Suchhigh-redshiftgiant Seymouretal.(2007)haveperformedaSpitzersurveyof69 radiogalaxiesmayactasbeaconsforextremelypowerfulAGN. high-redshiftradiogalaxiesandfindthatthemonochromaticlumi- nosityof6C0905+39at8µmisL8µm ∼5×1044ergs−1.They donothavemid-infrareddatafor6C0905+39,butinthemajority 5 CONCLUSIONS ofcaseswheretheyhavelongerwavelengthdata,thespectralen- ergydistributions(SED)oftheirhigh-redshiftradiogalaxiescon- Our XMM-Newton spectrum of 6C0905+39, which is one of the tinuetorise,oftenbyonetotwoordersofmagnitude.Observing highest redshiftgiant radiogalaxiesknown, shows ittocontaina 6C0905+39atthesewavelengthsisimportanttoconfirmitsnature Compton-thickquasar.6C0905+39isoneofveryfewsuchsources asaCompton-thickquasar. currentlyknownandoneofevenfewerathighredshift.Itislikely Radio galaxieshavelong been supposed toberadioquasars tobeahigh-Eddington sourcewithabolometricluminosityof∼ oriented closer to the plane of the sky. Given its immense linear 1047ergs−1,makingitoneofthemostpowerfulsourcesknown. sizeofnearly1Mpc,6C0905+39islikelytolieclosetotheplane of theskyand so,following theunificationmodel, weareseeing thecorethroughthetorusandthuswewouldexpectthesourceto 1 wheretheEddingtonratioisdefinedasthebolometricluminositydivided beCompton-thick.Thisalsomeansthatbeamingisunimportant. bytheEddingtonluminosity (cid:13)c 0000RAS,MNRAS000,000–000 5 ACKNOWLEDGEMENTS MCEacknowledgesPPARCforfinancialsupport.ACFandKMB thanktheRoyalSociety.MCEthanksPoshakGandhiforproviding thedatatorecreatehisfigure(Fig.3)sothat6C0905+39couldbe includedandRanjanVasudevanforhelpfuldiscussions. REFERENCES AntonucciR.,1993,ARA&A,31,473 BarthelP.D.,1989,ApJ,336,606 BassaniL.,DadinaM.,MaiolinoR.,SalvatiM.,RisalitiG.,della CecaR.,MattG.,ZamoraniG.,1999,ApJS,121,473 Blundell K. M., Fabian A. C., Crawford C. S., Erlund M. C., CelottiA.,2006,ApJ,644,L13 Comastri A., 2004, in Astrophysics and Space Science Library, Vol.308,BargerA.J.,ed,AstrophysicsandSpaceScienceLi- brary,p.245 ComastriA.,GilliR.,VignaliC.,MattG.,FioreF.,IwasawaK., 2007,ArXive-prints,704 DickeyJ.M.,LockmanF.J.,1990,ARA&A,28,215 Erlund M. C., Fabian A. C., Blundell K. M., 2007, MNRAS in prep. FabianA.C.,CrawfordC.S.,IwasawaK.,2002,MNRAS,331, L57 FanaroffB.L.,RileyJ.M.,1974,MNRAS,167,31P Gandhi P., Fabian A. C., Crawford C. S., 2006, MNRAS, 369, 1566 GilliR.,SalvatiM.,HasingerG.,2001,A&A,366,407 Law-GreenJ.D.B.,EalesS.A.,LeahyJ.P.,RawlingsS.,Lacy M.,1995,MNRAS,277,995 LiuY.,JiangD.R.,GuM.F.,2006,ApJ,637,669 MagorrianJ.etal.,1998,AJ,115,2285 MaozD.,2007,MNRAS,377,1696 MarconiA.,HuntL.K.,2003,ApJ,589,L21 MattG.,PompilioF.,LaFrancaF.,1999,NewAstronomy,4,191 NormanC.etal.,2002,ApJ,571,218 Panessa F.,Barcons X.,Bassani L.,Cappi M., CarreraF. J.,Ho L.C.,PellegriniS.,2007,A&A,467,519 PollettaM.d.C.etal.,2006,ApJ,642,673 SeymourN.etal.,2007,ArXivAstrophysicse-prints SikoraM.,StawarzŁ.,LasotaJ.-P.,2007,ApJ,658,815 VasudevanR.V.,FabianA.C.,2007,ArXive-prints,708 WilmanR.J.,FabianA.C.,1999,MNRAS,309,862 WorsleyM.A.etal.,2005,MNRAS,357,1281 (cid:13)c 0000RAS,MNRAS000,000–000