SLAC-PUB-16157 PreprinttypesetusingLATEXstyleemulateapjv.5/2/11 THE NUSTAR VIEW OF NEARBY COMPTON-THICK AGN: THE CASES OF NGC 424, NGC 1320 AND IC 2560 M.Balokovic´1, A.Comastri2, F.A.Harrison1, D.M.Alexander3, D.R.Ballantyne4, F.E.Bauer5,6, S.E.Boggs7, W.N.Brandt8,9, M.Brightman10, F.E.Christensen11, W.W.Craig7,12, A.DelMoro3, P.Gandhi3, C.J.Hailey13, M.Koss14,15, G.B.Lansbury3, B.Luo8,9, G.M.Madejski16, A.Marinucci17, G.Matt17, C.B.Markwardt18, S.Puccetti19,20 C.S.Reynolds21,22, G.Risaliti23,24, E.Rivers1, D.Stern25, D.J.Walton1, W.W.Zhang18 4 ABSTRACT 1 0 We present X-ray spectral analyses for three Seyfert 2 active galactic nuclei, NGC 424, NGC 1320, 2 and IC 2560, observed by NuSTAR in the 3–79 keV band. The high quality hard X-ray spectra allow detailed modeling of the Compton reflection component for the first time in these sources. g Using quasi-simultaneous NuSTAR and Swift/XRT data, as well as archival XMM-Newton data, we u A find that all three nuclei are obscured by Compton-thick material with column densities in excess of ∼ 5×1024 cm−2, and that their X-ray spectra above 3 keV are dominated by reflection of the 2 intrinsic continuum onCompton-thick material. Due to the very highobscuration,absorbedintrinsic 2 continuum components are not formally requiredby the data in any of the sources. We constrainthe intrinsic photon indices and the column density of the reflecting medium through the shape of the ] reflection spectra. Using archival multi-wavelength data we recover the intrinsic X-ray luminosities E consistent with the broadband spectral energy distributions. Our results are consistent with the H reflecting medium being an edge-on clumpy torus with a relatively large global covering factor and . overallreflection efficiency of the order of 1%. Given the unambiguous confirmationof the Compton- h thicknatureofthesources,weinvestigatewhethersimilarsourcesarelikelytobemissedbycommonly p usedselectioncriteriaforCompton-thickAGN,andexplorethepossibilityoffindingtheirhigh-redshift - o counterparts. r Subject headings: galaxies: nuclei — galaxies: Seyfert — galaxies: individual (NGC 424, NGC 1320, t s IC 2560) — X-rays: galaxies — techniques: spectroscopic — facilities: NuSTAR, a Swift, XMM-Newton [ 1 v 4 1Cahill Center for Astronomy and Astrophysics, Caltech, 1 Pasadena, CA91125, USA 4 2INAFOsservatorioAstronomicodiBologna,viaRanzani 1, I-40127, Bologna,Italy 5 3Department of Physics, Durham University, Durham DH1 8. 3LE,UK 4CenterforRelativisticAstrophysics,SchoolofPhysics,Geor- 0 giaInstitute ofTechnology, Atlanta,GA30332, USA 4 5Instituto de Astrof´ısica, Facultad de F´ısica, Pontificia Uni- 1 versidadCat´olicadeChile306,Santiago22,Chile : 6SpaceScienceInstitute,4750WalnutStreet,Suite205,Boul- v der,CO80301, USA i 7Space Sciences Laboratory, University of California, Berke- X ley,CA94720, USA r 8DepartmentofAstronomyandAstrophysics,ThePennsylva- a niaStateUniversity,525DaveyLab,UniversityPark,PA16802, USA 9InstituteforGravitationandtheCosmos,ThePennsylvania State University,UniversityPark,PA16802, USA 10Max-Planck-Institutfu¨rextraterrestrischePhysik,Giessen- bachstrasse1,D-85748GarchingbeiMu¨nchen,Germany 11DTUSpace,NationalSpaceInstitute,TechnicalUniversity ofDenmark,Elektrovej327,DK-2800Lyngby, Denmark USA 12Lawrence Livermore National Laboratory, Livermore, CA 19ASI-ScienceDataCenter,viaGalileoGalilei,I-00044 Fras- 94550, USA cati,Italy 13Columbia Astrophysics Laboratory, Columbia University, 20INAF OsservatorioAstronomico di Roma, via Frascati 33, NewYork,NewYork10027, USA I-00040MonteporzioCatone,Italy 14Institute forAstronomy,UniversityofHawaii,2680Wood- 21DepartmentofAstronomy,UniversityofMaryland,College lawnDrive,Honolulu,HI96822, USA Park,MD,USA 15Institute for Astronomy, Department of Physics, ETH 22JointSpace-ScienceInstitute,UniversityofMaryland,Col- Zurich,Wolfgang-Pauli-Strasse27,CH-8093Zurich,Switzerland legePark,MD,USA 16Kavli Institute for Particle Astrophysics and Cosmology, 23INAFOsservatorioAstrofisicodiArcetri,LargoE.Fermi5, SLACNationalAcceleratorLaboratory,MenloPark,CA94025, I-50125Firenze,Italy USA 24Harvard-Smithsonian Center for Astrophysics, 60 Garden 17DipartimentodiMatematicaeFisica,Universit`adegliStudi St.,Cambridge,MA02138,USA RomaTre,viadellaVascaNavale84,I-00146Roma,Italy 25JetPropulsionLaboratory, CaliforniaInstitute ofTechnol- 18NASAGoddardSpaceFlightCenter,Greenbelt,MD20771, ogy,Pasadena, CA91109, USA This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-76SF00515. 2 M. Balokovi´cet al. 1. INTRODUCTION possible to fully constrain using only soft X-ray data. These Compton-thick AGN demonstrate how NuSTAR It is well established that a significant fraction of spectroscopyofthe nearbytargetscancharacterizetheir active galactic nuclei (AGN) are intrinsically obscured X-ray properties better than previously possible, and by gas and dust surrounding the central supermassive how the new constraints may lead to improved under- black holes (SMBH). Obscured AGN are needed to ex- standing of both local AGN and their distant counter- plain the ∼30 keV peak of the Cosmic X-ray Back- parts. Ultimately, the NuSTAR surveys will allow us to ground(CXB;e.g.,Churazov et al. 2007;Frontera et al. directlydeterminethefractionofCompton-thicksources 2007; Ajello et al. 2008; Moretti et al. 2009), however, in the AGN population and the distribution of the ob- theirspacedensityisobservationallypoorlyconstrained. scuring column density for heavily obscured AGN. AGN obscured by gas with column density of N . 1.5×1024 cm−2 havebeenidentified inlargenumbeHrs in Thispaperisorganizedasfollows. In§2wepresentthe target selection, and the new data obtained from quasi- deep soft X-ray (<10 keV) surveys (Brandt & Hasinger simultaneousobservationswithNuSTARandSwift. §3.1 2005), whicharepowerfulmeans foridentifying the bulk and §3.2 demonstrate that reflection is the dominant ofthe AGNpopulationathighredshiftandthus provid- componentofthehardX-rayspectraofthesethreeAGN. ing valuable constraints on the growth history of SMBH §3.3providesamoredetailedspectralanalysisincluding (e.g.LaFranca et al. 2005; Aird et al. 2010). However, the XMM data. A comparison with the previously pub- the heavily obscured, Compton-thick sources (N > H lished X-ray results, as well as a discussion of the multi- 1.5× 1024 cm−2; see, e.g.Comastri 2004 for a review) wavelength properties and constraints on the AGN ge- required by the CXB models remain elusive. ometry, is presented in §4. We summarize our results in Recent surveys with the hard X-ray (>10 keV) tele- §5. In this work we use standard cosmological parame- scopes Swift/BAT (Burst Alert Telescope; Gehrels et al. ters (h =0.7, Ω =0.73)to calculate distances. Unless 2004) and INTEGRAL (International Gamma-Ray As- 0 Λ noted otherwise, all uncertainties are given as 90% con- trophysics Laboratory;Winkler et al. 2003)indicatethat fidence intervals. in the local Universe the fraction of obscured AGN (with NH > 1022 cm−2) is approximately 80%, while 2. TARGETSELECTIONANDOBSERVATIONS Compton-thick sources likely contribute about 20% of 2.1. Target Selection the total number of AGN (estimated from the ob- served .10% fraction corrected for survey complete- The NuSTAR Extragalactic Survey program includes ness, e.g., Malizia et al. 2009; Burlon et al. 2011). Ob- a wide-field shallow component (average exposure of scured AGN therefore contribute significantly to the lo- 20 ks) in which the observatory is pointed towards a calsupermassiveblackholespacedensity(Marconi et al. known AGN previously detected with Swift/BAT, or 2004) and may be even more important at ear- selected because of high obscuration inferred from soft lier epochs (LaFranca et al. 2005; Ballantyne et al. X-ray (< 10 keV) observations. The wide field of 2006; Treister & Urry 2006; Brightman & Ueda 2011; view of the Swift/BAT instrument and its nearly uni- Iwasawa et al. 2012). The peak of the CXB at form coverage of the whole sky down to a sensitivity of ∼30 keV can be reproduced by invoking a signif- & 1.0 × 10−11 erg s−1 cm−2 in the 14–195 keV band icant number of Compton-thick sources at moder- (Baumgartner et al. 2013), provide a reasonable sample ate redshift (e.g.,Gilli et al. 2007; Treister et al. 2009; ofpredominantlylocal(z ∼0.03)AGN.Itsmoreuniform Ballantyne et al. 2011; Akylas et al. 2012), however, and deeper exposure of the sky away from the Galactic to date only a few percent of the CXB has actually plane compared to INTEGRAL makes the Swift/BAT been resolvedat its peak energy (e.g.,Ajello et al. 2008; survey source catalog an excellent starting point for se- Bottacini et al. 2012). Thedistributionoftheobscuring lecting targets for more detailed spectroscopic studies, column density and the degeneracy in relative contribu- suchaspossiblewithNuSTAR. Thetargetswereselected tionsofabsorption-andreflection-dominatedhardX-ray for NuSTAR observations from the catalog utilizing 54 spectraarethereforepoorlyconstrainedwiththecurrent months of BAT operation (Cusumano et al. 2010). Un- data. less preventedby technicalconstraints,all NuSTAR tar- A primary goal of the NuSTAR (Nuclear Spectro- gets in this program receive on average 7 ks of quasi- scopic Telescope Array) hard X-ray mission is to study simultaneouscoverage(with delayof.1day)inthe soft the evolution of obscuration in AGN at 0 < z < 2 X-raybandfromthe Swift/XRTinorderto enablespec- through its comprehensive extragalacticsurvey program tral analysis over the broad 0.3–79 keV band. (Harrison et al. 2013). In addition to blank-field obser- Two of the targets presented here, NGC 424 vations(Mullaneyetal.,inprep.;Civanoetal.,inprep.), (Tololo 0109–383) and NGC 1320, were selected from the programincludes a surveyof knownsourcesselected the 54-month Swift/BAT catalog.26 The third tar- from the Swift/BAT catalog with two goals: (i) ob- get, IC 2560, was selected from a sample of relatively tain high-quality spectroscopy of the nearby Swift/BAT faint AGN with some indication of Compton-thick ob- -selected AGN, and (ii) perform a wide-field search scuration from previous observations (e.g.Risaliti et al. for serendipitous background sources (Alexander et al. 1999; Tilak et al. 2008). Soft X-ray spectroscopy, as 2013). In this paper we present observations and mod- well as multi-wavelength data, indirectly suggest that eling of the hard X-ray spectra of three local AGN: NGC 424 and NGC 1320 are also likely to be Compton- NGC 424, NGC 1320, and IC 2560, two of which are thick (Collinge & Brandt 2000; Marinucci et al. 2011; selected from the program outlined above. All three show spectradominated by reflectionfromcold, distant, 26Duetodifferentmethodologyandlowsignificance,NGC1320 isnotincludedinthelatest70-monthcatalog(Baumgartner et al. Compton-thick material, the properties of which are im- 2013) The NuSTAR View of Nearby Compton-thick AGN 3 cular regions 40′′ in radius, centered on the peaks of the Table 1 point-source images. The background spectra were ex- BasicdataontheAGNpresentedhere. tracted from regions encompassing the same detector as the source,27 excluding the circular region 50′′ around NGC 424 NGC 1320 IC 2560 the source. The background region sampling the same detector as the source provides the best estimate of the GalaxyTypea SB0/a Sa SBb underlyingbackground. ForIC2560thebackgroundwas AGNTypea Sy1/Sy2 Sy2 Sy2 extracted from two adjacent detectors due to its posi- MBH [M⊙]b 6.0×107 1.5×107 2.9×106 Redshift(z)c 0.0117 0.0091 0.0096 tion in the focal plane. All fluxes reported in this paper dL [Mpc]c 50.6 39.1 41.4 have been automatically corrected for the finite extrac- NH,G [cm−2 ]d 1.7×1020 4.3×1020 6.8×1020 tionapertureusingthebestpointspreadfunctionmodel currently available. We do not use NuSTAR data be- a Summary of classifications from the NASA Extragalactic low 3 keV, since the calibrationis currently uncertain in Database(NED;http://ned.ipac.caltech.edu/). that energy range. The upper end of the bandpass is b BlackholemassfromGreenhilletal. (2008)andBian&Gu mostly limited byphotonstatistics and the NuSTAR in- (2007). c Adoptedredshiftandluminositydistance(calculatedassum- strumentalbackground. AllNuSTARspectraarebinned ing h0 = 0.7, ΩΛ = 0.73) based on published measurements to a minimum of20 photons per bin using HEAsofttask availablethroughNED.Notethatthedistancesusedinthelit- grppha. erature differ . 10% for NGC 424 and NGC 1320, and up to 40%forIC2560. 2.3. Swift/XRT Data d Galactic column density averaged between Dickey&Lockman (1990) and Kalberlaetal. (2005). Each NuSTAR observation was accompanied by a short observation with Swift, typically delayed by less than24hours. The purposeoftheseobservationswasto provide coverage on the soft X-ray end of the spectrum, Brightman & Nandra 2011a). The selection of the sam- where the NuSTAR sensitivity drops off, and to facili- ple presented here is based on a basic spectral analysis tate a comparison of the soft X-ray flux with the data of all NuSTAR 20-ks snapshot observations of AGN up available in the literature. Since the sources are not ex- to May 2013. Out of 34 observed AGN we selected 3 pectedtobehighlyvariableontimescalesofhours,quasi- that show the most prominent Compton reflection com- simultaneous exposures with NuSTAR and Swift/XRT ponent signature: very hard spectrum (Γ<1, assuming provide a broadband snapshot covering the range from asimplepowerlawmodel), strongComptonhump(high approximately 0.5 to 70 keV. The Swift/XRT obser- reflection fraction, R>10, assuming the simplest reflec- vations were performed in the Photon Counting mode tion model) and iron emission(large equivalent width of (Hill et al. 2004; Burrows et al. 2005). The data were the neutral iron Kα line, & 1 keV). This is not a uni- reduced using the task xrtpipeline (version 0.12.6), formly selected, statistically complete sample – we will whichisapartoftheXRTDataAnalysisSoftware(XRT- address such samples in future work. However,the hard DAS) within HEAsoft. Spectra were extracted from X-ray properties of these three targets can be consid- circular regions 20′′ in radius centered on the targets ered representative of a larger class of heavily obscured and the backgrounds were extracted from large annular AGN, which make up approximately 10% of the sample source-free regions around them. We used response file of nearby AGN being surveyed with NuSTAR. The ob- swxpc0to12s6 20010101v013.rmf from the Swift/XRT servedspectraofallthreesourcesareshowninFigure1. calibration database, while auxilliary response files were SomebasicdataonthetargetsissummarizedinTable1. generated using the task xrtmkarf. Table 2 provides the complete list of observations. Unfortunately, the ob- 2.2. NuSTAR Data servation of IC 2560 was too short to yield a detection NGC 1320 was observed on two occasions: an initial in the Swift/XRT band, so we use it here only to place 15-ks snapshot and additional follow-up to improve the an upper limit on the soft X-ray emission. Due to low signal-to-noise ratio for detailed spectral anaysis. As a count statistics, the Swift/XRT spectra are binned to a firststepinouranalysiswecheckforvariabilitybetween minimum of 10 photons per bin. the observations and find that they are consistent with no change; hereafter we analyse them jointly, but with- 2.4. Archival XMM-Newton Data outco-adding. Theothertwosourceswereobservedonce In addition to the quasi-simultaneous NuSTAR and each. Table 2 gives a summary of all NuSTAR observa- Swift/XRT data, we use archival data from XMM- tions. The raw data were reduced using the NuSTAR- Newton to additionally verify our models. The XMM DAS software package (version 1.2.1), distributed with spectra are the highest-quality soft X-ray spectra cur- the HEAsoft package by the NASA High Energy As- rently available for these sources. Descriptions of the trophysics Archive Research Center (HEASARC). The dataanddetailsregardingtheirreductioncanbefoundin raw events were cleaned and filtered for South Atlantic Marinucci et al. (2011), Brightman & Nandra (2011a) Anomaly (SAA) passages using the nupipeline task. and Tilak et al. (2008) for NGC 424, NGC 1320 and The cleaned events were further processed for each of IC 2560, respectively. the two focal plane modules (FPMA and FPMB) using the nuproducts task, which generates the spectra and 3. MODELINGOFTHEX-RAYSPECTRA the corresponding response files. These procedures are presented in detail in Perri et al. (2013). 27 In each module the focal plane consists of 4 detectors; for Spectra for all of the sources were extracted from cir- details,seeHarrisonet al. (2013). 4 M. Balokovi´cet al. Table 2 Summaryofthequasi-simultaneousNuSTAR andSwift observations. Sequence StartTime Duration Exposure CountRatea Target ID [UTC] [ks] [ks] [10−2 counts s−1 ] NuSTAR Observations NGC1320 60061036002 2012-Oct-2521:50 25.7 14.5 2.8±0.2/2.4±0.2 NGC424 60061007002 2013-Jan-2606:35 26.7 15.5 4.7±0.2/4.6±0.2 IC2560 50001039002 2013-Jan-2822:05 43.5 23.4 1.1±0.1/1.1±0.1 NGC1320 60061036004 2013-Feb-1007:15 49.9 28.0 3.0±0.1/2.6±0.1 Swift/XRTObservations NGC1320 00080314001 2012-Oct-2602:48 75.6 6.8 1.04±0.03 NGC424 00080014001 2013-Jan-2606:34 23.9 6.6 2.55±0.03 IC2560 00080034001 2013Jan2917:50 2.0 2.0 <0.42 NGC1320 00080314002 2013-Feb-1007:19 23.9 6.6 1.04±0.03 a CountratesforNuSTAR modulesFPMAandFPMB(3–79keV),orSwift/XRT(0.3–10keV). residuals reveal the presence of a strong X-ray reflection 10-2 NGC 424 componentintheNuSTARspectra. Theprominentneu- 1V ] − NICG 2C5 610320 tpreaalkierdonColimneptaornisihnugmfproimntflhueo2r0e–sc3e0nkceeVanredgitohneabrerotaydpliy- e 1s k−10-3 c1a9l9s4i;gMnaattutreestoafl.su2c0h00a,c2o0m03pbo)n.ent(e.g.,Ghisellini et al. s nt Detailed models of the soft X-ray spectra, which is u [ coE10-4 cliogmhtp,opseladsmofaaiocnoimzebdinbaytiotnheofATGhNomasnodn-sstcaarttfeorremdaAtiGonN, F are not the focus of this paper. We refer the reader to Marinucci et al. (2012), Brightman & Nandra (2011b) and Tilak et al. (2008) for more details on such models. 8 4 ratio to a fittedpower law model 463752 ilrinoens Cohmupmtopns 123 fIttmonooroattdhbpheeopeltrahfosnopxarNeilmcytGthsarCietasepS4atrwh2be4eoisfevctan/eontXned3tRdrkNTihbeGeuVsrtCoe.i,ofAtna13Xsog2if-mo0rtoahpidyselesapdpohafopettwoancweoor(mem0rla.ep3wnloa–onw3ilseonkguwetisic(Vetsahd))l 1 Γ ≈2.7(Γ =2.5±0.8andΓ =2.7±0.4,respectively). s s s 3 5 7 10 20 30 50 70 Weadoptthisaveragevalueandkeepitfixedinallmod- energy [ keV ] els,varyingonlythe normalization,inorderto avoidthe degeneracy associated with its large uncertainty. For all Figure 1. Upper panel: The observed NuSTAR spectra of NGC 424 (red circles), NGC 1320 (green rectangles) and IC 2560 three targets we verify that this value is consistent with (blue diamonds). Spectra for two focal plane modules have been the higher-quality XMM-Newton data. The XMM data co-added, and only the longer observation of NGC 1320 is shown clearly require a more complex spectral model in order here for clarity. The typical background level and its uncertainty areshownbythefilledgreysymbolsandlines. Lowerpanels: Ratio tofitthe data well,asadditionalfine structureontopof of the spectra and a simple power-law model fitted to each spec- the slope is apparentinthe <3 keVresiduals. However, trum(symbolsasinthepanelabove). Notethattherightpanelis the simple power law represents a good approximation. rescaledverticallywithrespecttotheleftpanelbyafactorof2in More detailed modeling is not warranted for the hard ordertobetter showtheComptonhumps. X-ray analysis presented in this paper. In the following subsections we present results from applying three different types of hard X-ray spectral The NuSTAR hard X-ray spectra (3–70 keV) of models. First we apply simple, phenomenological mod- NGC 424, NGC 1320 and IC 2560 are qualitatively sim- els, which have been extensively used in previous work ilar, as demonstrated in Figure 1. In the lower panel (§3.1). Wealsoapplyphysicallymotivatedtorusmodels the figure we also show the ratios of the spectra to their (§3.2) and reflection-only models (§3.3). We use Xspec respective best-fit power law model simply to highlight version 12.8.1 (Arnaud 1996) for all our modeling. In the most important features. The best-fit photon in- addition, we: take into account redshifts and Galactic dices in all three cases are lower than unity; these fits absorption column density listed in Table 1; assume a are rather poor (reduced χ2&3) and intended only for contributionfromasoftpowerlawcomponent(Γ =2.6) s demonstration. The spectra exhibit very hard continua with free normalization;assume Solarabundances anda withaconvexshapebroadlypeakingaround20keV,and high-energycutoffinthenuclearcontinuumat200keV;28 a prominent emission feature at 6.4 keV, matching the rest-frame energy of the neutral iron Kα emission line. Thehardeffectivephotonindicesandthestructureofthe 28 Weshowlaterin§4.1thattheNGC 424dataareconsistent The NuSTAR View of Nearby Compton-thick AGN 5 NGC 424 NGC 1320 IC 2560 -2 10 -3 10 s b a c -4 pl 2m ]−10 χΓ2/=ν0=.78±5/06.20 χ2/Γν==11.725±/102.22 χ2Γ/ν==1.207−+/002..568 c10-5 χ2/ν=147/61 χ2/ν=200/123 χ2/ν=31/29 1− s V e k ) [ 10-2 E v F( a r 2E10-3 ex p + s 10-4 Γ=2.1+0.3 Γ=1.3+0.3 Γ=2.5+0.5 ab 0.2 0.2 0.7 c χ2/ν=58−/60 χ2/ν=149/−121 χ2/ν=21−/28 pl 10-5 χ2/ν=60/61 χ2/ν=163/122 χ2/ν=23/29 3 5 7 10 20 30 5070 3 5 7 10 20 30 5070 3 5 7 10 20 30 5070 energy [ keV ] Figure 2. SimpleapproximatemodelsfittedtotheNGC424,NGC1320andIC2560Swift/XRTandNuSTARdata(see§3.1fordetails). Thickcoloredlinesshow modelsforwhichthephoton indexwasdetermined fromthe data; thebest-fit value, its90% confidence interval and the best fit χ2 are given inthe lower right corner of each panel. The thin black lines show the same models with the assumption of Γ=1.9;itsχ2 isgiveninblackletters. Withthesolidlinesweshowthetotalmodel,whilethedashedanddottedlinesshowthereflection andthetransmissioncomponents,respectively. Upper panels: Transmission-onlymodelsbasedonplcabs. Lower panels: Two-component modelsbasedontheplcabs andpexravcomponents. leave the cross-normalization constants between instru- line exceeds 2 keV. These values of Γ are much harder mentstovaryfreely. Weperformallparameteroptimiza- than typical for the coronal continuum of AGN, while tionsusingtheCashstatistic(Cash1979),butreportthe the equivalent width of the iron line strongly indicates χ2 valuesofeachbestfitdue totheir straightforwardin- presence of a reflection component. The photon indices terpretability. can be assumed to take on a typical value of 1.9 at a cost of increasing the χ2, however the best fits remain 3.1. Phenomenological Models qualitatively the same. They are shown for each of the three AGN in the top panels of Figure 2. We first fit an absorbed power law model using the Wenextaddareflectioncomponent,whichweapprox- Xspec component plcabs (Yaqoob 1997). This model imate using the pexrav model (Magdziarz & Zdziarski representsanabsorbedpower-lawspectrumincludingthe 1995). This model produces the reflected continuum of effects of Compton scattering. This model is approxi- an infinite slab of infinite optical depth, and is there- mated in order to be computationally fast, and it is also foreonlyanapproximationforthereflectionoffadistant limited to N < 5×1024 cm−2. In all three cases the H torus. We apply this component to produce only the re- fits are poor and show a strong narrow residual feature flectedcontinuum. Theincidentspectrumissettobethe around6.4keV.TheadditionoftwounresolvedGaussian same intrinsic cut-off power law as in the plcabs com- components(σ =10−3 keV)at6.40and7.06keV,corre- ponent. We start our fitting procedure with both the sponding to neutral iron Kα and Kβ lines, improves the transmitted and the reflected component. We also in- fits significantly. Although the reduced χ2 (χ2/ν, where cludenarrowGaussiancomponentsat6.40and7.06keV ν is the number of degrees of freedom) reaches ≃1 for and the soft Γ = 2.7 power law. The basic result of the thecaseofIC2560and≃1.5fortheothertwoAGN,the fittingisthatthecontributionofthetransmittedcompo- fits are difficult to justify physically. The best fits are nents are minor for all three AGN: the data requires ei- allqualitativelythe same: the intrinsicpowerlawwitha therveryhighabsorptioncolumn(N &5×1024cm−2, photon index which tends to Γ.1 is absorbedby a col- H,A which is the upper limit of the plcabs model), or zero umn density of 1023−24 cm−2, and the equivalent width normalization. of the Gaussiancomponent at the energyof the ironKα For NGC 424 the best-fit column density is N & H,A 5 × 1024 cm−2 and the intrinsic power law continuum with this value; for the other two targets this parameter is un- constrained. This choice is consistent with the recent literature, slope is Γ = 2.1+0.3 (χ2/ν = 58/60). If we fix the ab- −0.2 e.g.,Ballantyne(2014)andMaliziaetal. (2014). sorption column density at a lower value, the normal- 6 M. Balokovi´cet al. NGC 424 NGC 1320 IC 2560 -2 10 d e pl -3 u 10 o c s, u 2m ]−10-4 χ2Γ/ν==2.114±4/06.21 χ2/Γν==21.903±/102.14 χΓ2/=ν2=.5556 (/f3)1 MYtor c10-5 χ2/ν=141/63 χ2/ν=195/125 χ2/ν=100/31 1− s V e k E) [ 10-2 ed F( pl u 2E -3 o 10 c - e d 10-4 χΓ2/=ν2=.05±8/06.20 χ2/νΓ==125.30/−+1002..314 χΓ2/=ν2=.5257 (/f3)0 Ytorus, 10-5 χ2/ν=59/61 χ2/ν=177/125 χ2/ν=28/30 M 3 5 7 10 20 30 5070 3 5 7 10 20 30 5070 3 5 7 10 20 30 5070 energy [ keV ] Figure 3. Torus models fitted to the NGC 424, NGC 1320 and IC 2560 Swift/XRT and NuSTAR data (see §3.2 for details). Thick coloredlinesshowmodelsforwhichthephotonindexwasdeterminedfromthedata;thebest-fitvalue,its90%confidenceintervalandthe best fit χ2 are given inthe lower rightcorner of each panel. Thethin black lines show the samemodels with the assumption of Γ=1.9; itsχ2 isgiveninblackletters. Withthesolidlinesweshowthetotalmodel,whilethedashedanddottedlinesshowthereflectionandthe transmissioncomponents, respectively. Upper panels: Literal torus models represented by MYtorus in the coupled mode, with equatorial columndensityfixedto5×1024 cm−2. Lower panels: Generalizedtwo-component modelsbasedonde-coupledcomponents ofMYtorus. ization of the transmitted component decreases until it primarilydue to the transmissionofthe intrinsiccontin- becomes consistentwith zero for N =2×1024 cm−2. uum through mildly Compton-thick material. We fur- H,A Ifthetransmittedcomponentisremovedfromthemodel ther examine a set of more appropriate physically moti- altogether, the best fit (χ2/ν = 59/61) is found for vated models. Γ = 1.71± 0.09. We find qualitatively similar results for IC 2560. For a fixed NH,A = 5 × 1024 cm−2 the 3.2. Torus Models best-fit photon index is 2.5+−00..57 (χ2/ν = 21/28). For ei- An improvement over the pexrav approximation, ther a lower NH,A or a lower Γ, the plcabs component which assumes infinite optical depth, is offered by the- vanishes, and the best fit is found for a pexrav-only oretical models that use Monte Carlo simulations of model (χ2/ν = 22/29) with Γ = 2.2+0.3. In the case of the propagation of X-ray photons through material of −0.4 NGC 1320 the best-fit model (χ2/ν = 149/121) is dom- finite optical depth in a physically motivated geome- inated by the reflection component, but does include a try. The first Xspec model of that kind is MYtorus 29 transmitted power-law component with Γ=1.3+0.3 and (Murphy & Yaqoob 2009; Yaqoob 2012). The basis of N = (2+2)×1024 cm−2. Simply removing−t0h.e2 lat- the MYtorus model is a literal torus with a 60◦ half- H,A −1 opening angle. It consists of two main spectral com- ter component degrades the fit only to χ2/ν =152/123. ponents: a transmitted continuum component (formally An alternative model, dominated by the plcabscompo- called zeroth-order continuum, MYTZ, by the authors of nentabove10keV,canbe foundwith the assumptionof the model) and a scattered one (MYTS; also referred to Γ=1.9 (χ2/ν =163/122). as reprocessed, or reflected). The former is produced by The models presented in this subsection are summa- scatteringphotons away fromthe line of sight, while the rized in Figure 2. For all three AGN we find that a latterisformedbyphotonsscatteredintothelineofsight statisticallygooddescriptionoftheir hardX-rayspectra of the distant observer. can be achieved using models consisting almost entirely We start with the complete MYtorus model, which of reflection components. However, with the quality of is characterized by a single column density (N = the data acquired in short 20-ks exposures it is not pos- H,R N = N , corresponding to the column density in sibleto excludeaminorcontributionfromatransmitted H,A H intrinsic continuum. The large equivalent width of the 29 We use the version of MYtorus model that is publicly avail- iron lines point towards strong reflection and essentially able at http://www.mytorus.com. Specifically, we use the tables rule out the possibility that the hard X-ray spectrum is calculatedwithaprimarypowerlawwithacutoffat200keV. The NuSTAR View of Nearby Compton-thick AGN 7 the equatorial plane of the torus). The internal nor- a direct power law absorbed by N ≈ 5×1024 cm−2 H,A malizations between the components are fixed. The first dominating above that. Removal of the latter compo- model we test is an edge-on torus with inclination fixed nent, however, leads to a slightly better reflection-only at 90◦. This model does not fit any of the NuSTAR model(χ2/ν =25/30),whichweelaborateonin§3.3. In and Swift data considered here: the reduced χ2 values conclusion,the physically motivated models of the AGN do not get any lower than 2–3. Note that these models torus, in addition to the phenomenological models pre- are transmission-dominatedand therefore formally simi- sented in §3.1, demonstrate that the observed NuSTAR lar(butphysicallymoreappropriate)totheplcabs-only spectra are consistent with being reflection-dominated. model examined in §3.1. Nextwe fitfor the inclinationangle ofthe torusunder 3.3. Reflection-dominated Models different assumptions of the equatorial column density, The conclusion of both of the preceeding two subsec- N , since a straightforward fit for both of those param- H tions is that the reflection-dominatedmodels provide ei- etersis highly degenerate. The resultsareagainqualita- ther better, or statistically equivalent but simpler, de- tively the same for all three AGN, regardlessof N : the H scriptions of the observed hard X-ray spectra compared best-fit inclination angles are found to be close to 60◦, to transmission-dominated or two-component models. matching the opening angle of the torus. For example, Here we summarize the results obtained with the simul- at N = 5×1024 cm−2 (shown in the upper panels of H taneous Swift and NuSTAR data, and also consider the Figure 3), the best-fit inclinations are 69+5, 68+3 and −4 −2 higher-quality,non-simultaneous,archivalXMM-Newton 66+7 degrees for NGC 424, NGC 1320 and IC 2560, re- data. In order to avoid the complexities associated with −4 spectively. The best-fit photon indices are 2.1±0.1 for the detailed modeling of the soft X-ray emission unre- NGC 424 and 2.0±0.1 for NGC 1320,while for IC 2560 lated to the AGN, we only use the XMM data above the fit runs into the upper domain limit of MYtorus at 3 keV. The model parameters of interest are listed in Γ = 2.6. The fits are slightly better for higher assumed Table 3. N , but they never reach χ2/ν < 1.5. Most of the χ2 H contribution comes from the iron line region, which we 3.3.1. NGC 424 treat in more detail in §3.3. These results can easily be Before using the XMM data from Marinucci et al. understood as the tendency of the fit to maximize the (2011) jointly with the NuSTAR data, we first checked contributionofthereflectedcomponent(whichincreases whetherthe targetchangeddramaticallyinfluxbetween with decreasing inclination, as the observer sees more of thetwoobservations. Weconstructasimplephenomeno- the inner far side ofthe torus), while not completely un- logical model for the 3–10 keV region by fitting the coveringthe sourceofthecontinuumatthecenter(since XMM spectrum with a pexrav continuum and Gaus- inthe line ofsightthe lightsuffers significantabsorption sian components for 8 emission lines, fixed to the fol- by passing throughthe edge of the torus for any inclina- lowing energies:31 3.13, 3.83, 5.37, 6.40, 6.65, 6.93, 7.06 tion greater than 60◦). In all cases the spectra are dom- and 7.47 keV. None of the lines are resolved by XMM inated by the reflection component, with only a minor , except the iron Kα line at 6.4 keV with width of contributionfromtransmissionofthe nuclearcontinuum σ = 0.09 ± 0.01 keV. The 3–10 keV flux calculated along the line of sight. for this model fitted to the XMM data is 8.4 ± 0.2 × Finally, we try a model in which the two spectral 10−13 erg s−1 cm−2, which is . 20% lower than the components of the MYtorus model are treated indepen- 1.1 ± 0.1 × 10−12 erg s−1 cm−2derived for the quasi- dently.30 In this case, the transmitted zeroth-order con- simultaneous Swift/XRT and NuSTAR/FPMA data fit- tinuum, MYTZ, and the scattered/reflection component, ted with the same model (all parameters fixed, except MYTS, have fixed inclination parameters (90◦ for the for- for the overall normalization factor). Given the cross- mer and 0◦ for the latter), separate column densities calibration uncertainty between NuSTAR and XMM of (N and N , respectively) and a relative normaliza- H,A H,R 10% (Madsen et al., in prep.), the fluxes can be consid- tion(A )differentthanunity. This againleadsto solu- rel ered almost consistent. We therefore conclude that the tions in which the reflection component dominates over flux variability is not severe and proceed with a joint thetransmissioncomponent(byafactorofA =5−20, rel spectral analysis. compared to the internal normalization of the complete The best-fit approximate reflection-only model (using MYtorus model). However, the fit parameters are dif- pexrav) for the NGC 424 data from Swift and NuS- ferent for each AGN, as shown in the lower panels of TAR is found for Γ = 1.71±0.09, with χ2/ν = 59/61. Figure3. ForNGC424wefindthatatransmissioncom- ponentwithN =(3±1)×1023 cm−2 contributessig- The XMM data require the Fe Kα line to be broad- H,A ened (σ = 0.09± 0.01 keV, ∆χ2 = 109 for one addi- nificantlyintheironlineregion,whilethereflectioncon- tional free parameter), but lead to a very similar result: tinuumdominatesabove10keV.InthecaseofNGC1320 Γ=1.64±0.09withχ2/ν =152/147. StrongFeKαlines the normalization of the transmitted component is con- withequivalentwidthof≈1keVarefoundinbothcases. sistentwith zero. The lowerquality of the IC 2560spec- The MYtorus model fits the NuSTAR and Swift/XRT tum allows for a number of degenerate solutions that sensitivelydependonthechoiceofassumptions. Assum- data well (χ2/ν = 61/60) for Γ = 2.28+0.03. The best −0.09 ing Γ = 2.55 as before, one interesting possible solution fit is achievedfor reflector column density (N ) at the H,R (χ2/ν =27/30)isreflectionfromCompton-thinmaterial (N ≈5×1023 cm−2) dominating below 10 keV with 31 This is a somewhat simpler model than the one used in the H,R originalanalysis,butitdescribesthe3–10keVspectrumverywell, withχ2/ν =0.95. Foridentification of thevarious emissionlines, 30 Thisisthede-coupledmode,afterYaqoob(2012). seeMarinucciet al. (2011). 8 M. Balokovi´cet al. NGC 424 NGC 1320 IC 2560 ]10-2 v) 2m− xra c e 1V s − 10-3 on (p [ ke10-4 ecti E) efl F( w/ XRT: χ2/ν=59/61 w/ XRT: χ2/ν=137/121 NuSTAR: χ2/ν=22/29 r 2E 10-5 w/ XMM: χ2/ν=152/147 w/ XMM: χ2/ν=135/119 w/ XMM: χ2/ν=47/53 te a m 4 xi 2χ 2 ro ∆ 0 p (cid:0) p ±−2 a −4 ] -2 10 2− n m o c ti 1 − 10-3 ec V s efl e r [ k10-4 on E) e- F( w/ XRT: χ2/ν=61/60 w/ XRT: χ2/ν=135/119 NuSTAR: χ2/ν=25/30 ac 2E 10-5 w/ XMM: χ2/ν=153/148 w/ XMM: χ2/ν=143/118 w/ XMM: χ2/ν=65/55 s f u 4 or t 2χ 2 MY ∆ 0 (cid:0) ±−2 −4 3 5 7 10 20 30 5070 3 5 7 10 20 30 5070 3 5 7 10 20 30 5070 energy [ keV ] Figure 4. Reflection-only models fitted to the NGC 424, NGC 1320 and IC 2560 NuSTAR data jointly with simultaneous Swift/XRT andnon-simultaneousarchivalXMM-Newtondata(see§3.3fordetails). ThincoloredlinesshowbestfitstotheNuSTARandSwift/XRT data (except forIC 2560, whereonlythe NuSTAR data was used), whilethe thicklines show the sameforthe NuSTAR and XMM data lowered by 20% for clarity. χ2 values for the best fits are given in each panel. Smaller panels show the residuals: black empty symbols for NuSTAR (diamonds for FPMA and squares for FPMB), and grey filled symbols for XMM. Upper panels: Models with the reflection continuum approximatedbythepexrav component. Lower panels: Modelsinwhichthereflectionspectrum isrepresented bytheface-on component ofMYtorus. upperlimitoftherangecoveredbyMYtorus,1025 cm−2, in the fit (Case B) leads to E = 6.54±0.06 keV and with a 90% confidence lower limit of 5×1024 cm−2. We σ = 0.33±0.08, removing the need for the previously findthatthelinecomponentnormalizationismarginally includedline at6.93keV.Although this is astatistically lower than unity and that the data favor addition of a good model (χ2/ν = 148/147), it is difficult the inter- narrow Ni Kα line at 7.47 keV (∆χ2 = 7 for one addi- pret the broad Gaussian feature that it includes. We tionalfreeparameter). Thebestfitparametersandtheir find an equally good fit (χ2/ν = 153/147, Case C) by uncertaintiesatthe90%confidencelevelarelistedinTa- broadeningthe MYtorus linecomponent,whichincludes ble 3. ModelcurvesandresidualsareshowninFigure 4. neutralFeKαandKβlines,usingaGaussiankernelwith A straightforward fit of the reflection-only MYtorus σ =0.06±0.01keV.Lettingionizedironlineenergyvary model does not find a statistically acceptable solution does not significantly improve the fit. The best-fit pho- (χ2/ν = 195/148) for the joint NuSTAR and XMM ton index is Γ=2.07+0.11 and the reflector column den- −0.09 data. The XMM residuals point towards a disagreement sity is well constrained to N = (3±1)×1023 cm−2. H,R in the region surrounding the prominent iron lines be- The model parameters listed in Table 3 represent this tween 6 and 8 keV. We show this energy range in more particular case. detail in Figure 5, with several different modeling solu- An alternative two-component model is suggested tions. We first attempt fitting for the energy of the ion- by our modeling in §3.2, as well as the literature ized iron line otherwise fixed at 6.65 keV: in this case (Iwasawa et al. 2001; Marinucci et al. 2011). We add (Case A) the fit is improved to χ2/ν = 179/147 for a second MYtorus component to describe the intrinsic E = 6.57+0.01 keV. Letting the width of the line vary continuumcontributiontransmittedthroughtheabsorb- −0.03 The NuSTAR View of Nearby Compton-thick AGN 9 ]10-3 Case A Case B Case C Case D Case E 1− V ux ke10-4 el fl2m− dc mo1 −10-5 s h. [ p10-6 4 2 2χ ∆ 0 (cid:0) ± −2 −4 χ2/ν=179/147 χ2/ν=148/147 χ2/ν=154/147 χ2/ν=174/147 χ2/ν=148/146 5.5 6 6.5 7 7.5 8 5.5 6 6.5 7 7.5 8 5.5 6 6.5 7 7.5 8 5.5 6 6.5 7 7.5 8 5.5 6 6.5 7 7.5 8 energy [ keV ] Figure 5. Demonstration of the spectral modeling of NGC 424 data in the energy range containing prominent iron lines. Different columns show the model curves (upper panel) andXMM andNuSTAR residuals(lower panel) forspecific cases discussed in§3.3.1. The solidblack lines inthe upper panel show the sum of all model components, dashed lines show the reflection continuum, and dotted lines showtheabsorbed/transmittedcomponentintherightmosttwopanels. InthelowerpanelsweshowtheNuSTARFPMA(FPMB)residuals with dark (light) red symbols, and XMM EPIC-pn residuals with grey symbols. In Case A we fit for the energy of the ionized iron line at≈6.6keV, keeping energies ofall other lines fixed; inCaseB,welet the widthof that lineto varyas well. InCaseC webroaden the FeKαandKβ lineswhilekeepinglineenergies fixed attheirexpected values. Table3liststhe modelparameters forCaseC.InCases D andEweaddatransmittedcomponent withNH,A≈7×1023 cm−2,andfitfortheenergyoftheionizedironlineinthelatter. ing torus (MYTZ×pow in Xspec), since the sharp iron significantly improved upon adding a narrow line com- edge of an absorbed power-law component (NH,A ≈ 1× ponentat6.57+−00..1068 keV,whichismostlikelyanFeXXV 1024 cm−2)couldsignificantlycontribute to the 6–8keV Kαline. UsingtheMYtorus reflection-onlymodel,aline lineregion. Wefindthatwithemissionlineenergieskept is required at 6.6±0.1 keV. In both cases, the equiva- fixed (Case D) the best fit occurs at χ2/ν = 174/147. lent width of the line is 0.3±0.2 keV. All other spectral The model is improved (χ2/ν = 148/146) if we addi- parameters are consistent between the two. Addition tionally let one of the ionized iron lines’ energy to vary of the ionized iron line is essentially the only improve- in the fit; the best-fit energy in that case (Case E) is ment needed over the reflection-dominated models al- 6.57+0.02keV.ThephotonindexfoundinCasesDandE, readymentionedin§3.1and§3.2. Wefindthatthenor- −0.05 malizationofthe linescomponentoftheMYtorus model which include component transmitted througha column densityofNH,A =(7±3)×1023 cm−2,isconsistentwith is mildly elevated, but consistent with unity (1.2+−00..34). the one found in Case C. The reflector column density The MYtorus model additionally provides a constraint is at the upper boundary of the model at 1025 cm−2 for on the column density of the reflecting material, instead Cases D and E, with a 90% confidence lower limit of of assuming it to be infinite. The reflector column den- 5×1024 cm−2. In§4we discussthe physicalplausibility sity, NH,R, is mildly degenerate with the intrinsic pho- ofthesimplesolutionsproposedhere,however,westress ton index, but it can be constrained independently to that the details of the iron line region modeling are en- NH,R =(4+−42)×1024 cm−2 with 90% confidence. tirely driven by the high-quality XMM data, which are The only soft X-ray data available for NGC 1320 be- not the focus of this paper. Although some contribution sides our quasi-simultaneous Swift/XRT data is from a of a heavily absorbed component cannot be completely relatively short 12-ks XMM-Newton observation in 2006 ruled out, all of the X-ray data considered here support (Brightman & Nandra 2011a). Only a minor difference thereflection-dominatedspectrumhypothesiswithinthe of . 30% in the 3–10 keV flux is observed between the statistical uncertainties. XMM and Swift+NuSTAR observations, so we proceed with a joint analysis. No significant difference is found 3.3.2. NGC 1320 betweenthebest-fitintrinsicphotonindicesbasedonthe NGC1320wassimultaneouslyobservedwithNuSTAR Swift or XMM data for either the pexrav or MYtorus and Swift twice. As no significant differences are ap- models. A significant improvement in either case is parent between the two observations, we model both found if a narrow line corresponding to ionized iron is epochs simultaneously and list the best-fit parameters added to the model The best fit energy is 6.55+0.10 keV −0.09 of those fits in Table 3. The best-fit photon index is for the pexrav model and E = 6.77+0.06 keV for the hard (Γ=1.3±0.1), however,assuming a higher cut-off MYtorus model, with equivalent width−s0o.2f30.2±0.1 keV energy for the intrinsic continuum brings it closer to the and 0.3+0.1 keV, respectively. Addition of a Ni Kα line typicalvalue: e.g.,foracut-offat500keV,the bestfitis −0.2 at 7.47 keV does not significantly improve the fit. The obtainedforΓ=1.5±0.1. AsinthecaseofNGC424,we XMM data further constrain the column density of the find that in the MYtorus reflection-only model the best- reflector: N = (4+2) × 1024 cm−2 with 90% confi- fitphotonindexissteeper(Γ=1.9+0.1)thanthephoton H,R −1 −0.3 dence. Both best-fit models and their respective residu- index derived from the pexrav modeling. The joint fit als are shown in the middle column of Figure 4. ofthe approximatereflection-onlymodelwithpexrav is 10 M. Balokovi´cet al. 3.3.3. IC 2560 (Iwasawa et al. 2001) it was possible to infer that the intrinsic power-law continuum is absorbed by a column IC 2560 is the faintest target considered in this pa- density of & 2×1024 cm−2. Note, however, that only per, and correspondingly has the poorest photon statis- thesimpleapproximatemodelswereusedinthespectral tics. The quasi-simultaneous Swift/XRT observation is analyses leading to the inference of the column density, too short to provide useful soft X-ray data, so for the andthattheyrequiredassumingaphotonindex(Γ=2), initial modeling we use the NuSTAR data alone. From sinceitwasnotpossibleto constrainit directlyfromthe the best fit of the approximate pexrav model we find data. Using the simultaneous Swift/XRT and NuSTAR that the intrinsic photon index is steeper than in the data,wefirmlyestablishthatthehardX-rayspectrumof other two targets (Γ = 2.2+0.3) and that the iron lines −0.4 NGC424canbe describedasbeingdominatedbyreflec- are strong; the equivalent widths of the Fe Kα and Kβ tion. Acontributionfromaheavilyabsorbedcomponent lines are 2.1+−10..35 keV and 0.6+−00..64 keV, respectively. In cannot be completely ruled out, but it is not formally the case of the MYtorus reflection-only model we find required by any of the data considered in this work. that the normalization of the line component is signif- Burlon et al. (2011) analyzed the first three years icantly elevated, 2.2+−00..75. The intrinsic photon index is of Swift/BAT data on NGC 424 and modeled it sim- somewhat degenerate with the reflector column density ply as a heavily absorbed power law with N = H,A andbotharebestfittedbyparametervaluesontheedge (2.0+0.3) × 1024 cm−2 and Γ = 1.9 ± 0.3. Since that of the validity domain of the model. Formally, we are −0.4 model does not include any contribution from a reflec- able to derive only the lower limits on the best-fit pa- tion component, it is not directly comparable to our rameters: Γ>2.6 and N >1025 cm−2. By fixing the H,R results. The Swift/BAT spectrum from the 70-month photon index to 2.55 (which is statistically acceptable, survey (Baumgartner et al. 2013) is entirely consistent with χ2/ν = 25/30), we can estimate a 90% confidence with the NuSTAR and XMM data (χ2/ν =153/153;see lower limit on NH,R to be 7×1024 cm−2 based on the Figure 6) for a cross-normalizationconstant of 1.5±0.3 NuSTAR data alone. relative to NuSTAR/FPMA. We also verify that our The XMM data we use are MOS1 and MOS2 spec- models are consistent with the BeppoSAX data from tra (above 3 keV) from an 80-ks observation published Iwasawa et al. (2001). With the photon index fixed at in Tilak et al. (2008). By calculating 3–10 keV fluxes its best-fit value for that model, Γ = 1.68, we find that basedona simple reflectionmodelrepresentedby a sum thecutoffenergyisE =120+50 keV.Aslightlyhigher of a pexrav continuum and a Gaussian line component cut −30 cut-off energy, E = 190+260 keV, is inferred from the we find that the flux did not change between the XMM cut −80 and NuSTAR observation by more than 20%. The joint equivalent fit using the Swift/XRT instead of the XMM fits to the NuSTAR andXMM data resultinbest-fit pa- data. The 14–195 keV luminosity published in the 70- rameters entirely consistent with those found with the month Swift/BAT catalog (6.5×1042 erg s−1) is calcu- NuSTAR data alone. Again, the best-fit photon index lated by assuming a relatively flat Γ ≈ 2 spectrum; if and column density in the MYtorus model is formally we use our model instead, the 14–195 keV luminosity outside of its validity domain. By assuming statistically based on Swift/BAT data alone is 5.5 × 1042 erg s−1. acceptable Γ = 2.55 (χ2/ν = 65/55) we can constrain The apparently significant normalization offset and its N to be greater than 7×1024 cm−2 with 90% confi- uncertainty are likely due to the limited statistics and H,R long-term averaging of the Swift/BAT data, in addition dence. Bothmodels andtheir residuals areshownin the to simple flux calibration differences. In the rest of the rightmost column of Figure 4. Table 3 provides the list discussionweassume∼30%lowerluminosityasinferred of the best-fit values for all relevant model parameters. from the NuSTAR data, as listed in Table 4. NodedicatedlongX-rayobservationsofNGC1320ex- 4. DISCUSSION istinthe literature: the onlypreviouslyavailablesoftX- 4.1. Comparison with Previously Published raydatawastakenwithXMM-Newtonaspartofarecent X-ray Spectral Analyses survey of infrared-bright AGN (Brightman & Nandra TheearliestX-rayspectrumofNGC424camefroman 2011a). The original modeling by those au- ASCA observation,which revealeda prominentironline thors and later re-analyses (e.g.,Georgantopoulos 2011; andahardspectrumsuggestiveofCompton-thickreflec- Severgnini et al. 2012;Marinucci et al. 2012)agreethat tionofthenuclearcontinuum(Collinge & Brandt2000). the nucleus of NGC 1320 is heavily obscured (NH,A & This result has been confirmed in later observations, by 1×1024 cm−2) and infer the presence of a considerable BeppoSAX, Chandra and XMM-Newton (Iwasawa et al. reflection component on the basis of a strong iron line 2001; Matt et al. 2003a). The soft X-ray spectrum has with an equivalent width of ∼1 keV. Gilli et al. (2010) mostrecentlybeenanalysedindepthbyMarinucci et al. assert that both reflected and transmitted components (2011), using a long ∼100-ksXMM-Newton observation. contribute to the spectrum, but provide very few details The focus of that work was on detailed modeling of the onthemodelingasthe modelparametersarelargelyun- physical state of the plasma dominating below 2 keV, constrained by the data. Our analysis (which includes but the data were also used to model the X-ray con- the same XMM data in addition to the NuSTAR data) tinuum and line emission up to 10 keV. In agreement confirms most of the earlier results and solidifies the with earlier results, they found support for a strong re- dominance of the reflection spectrum above 2 keV. A flection component and a heavily absorbed power-law transmitted component is not formally required by our continuum obscured by nearly Compton-thick material data. A hard X-ray source was detected by Swift/BAT with N = 1.1×1024 cm−2 contributing only above atthe coordinatesofNGC 1320,butthe lowsignificance H,A 5 keV. With the hard X-ray coverage of BeppoSAX of that detection does not provide any additional spec-
Description: