A hard look at local, optically-selected, obscured Seyfert galaxies
E. S. Kammoun, J. M. Miller, M. Koss, K. Oh, A. Zoghbi, R. F. Mushotzky, D. Barret, E. Behar, W. N. Brandt, L. W. Brenneman, J. S. Kaastra, A. M. Lohfink, D. Proga, D. Stern
DD raft version A ugust
26, 2020Typeset using L A TEX twocolumn style in AASTeX62
A hard look at local, optically-selected, obscured Seyfert galaxies ∗ E. S. K ammoun , J. M. M iller , M. K oss , K. O h , A. Z oghbi , R. F. M ushotzky , D. B arret , E. B ehar , W. N. B randt , L. W. B renneman , J. S. K aastra , A. M. L ohfink , D. P roga , and D. S tern
Department of Astronomy, University of Michigan, 1085 South University Avenue, Ann Arbor, MI 48109-1107, USA Eureka Scientific, 2452 Delmer Street Suite 100, Oakland, CA 94602-3017, USA Korea Astronomy & Space Science institute, 776, Daedeokdae-ro, Yuseong-gu, Daejeon 34055, Republic of Korea Department of Astronomy and Joint Space-Science Institute, University of Maryland, College Park, MD 20742, USA IRAP, Universit´e de Toulouse, CNRS, UPS, CNES, 9, Avenue du Colonel Roche, BP 44346, 31028 Toulouse Cedex 4, France Department of Physics, Technion, 32000, Haifa, Israel Department of Astronomy and Astrophysics, 525 Davey Lab, The Pennsylvania State University, University Park, PA 16802, USA Institute for Gravitation and the Cosmos, The Pennsylvania State University, University Park, PA 16802, USA Department of Physics, 104 Davey Lab, The Pennsylvania State University, University Park, PA 16802, USA Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138, USA SRON Netherlands Institute for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, the Netherlands Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, the Netherlands Department of of Physics, Montana State University, P.O. Box 173840, Bozeman, MT 59717-3840, USA Department of Physics & Astronomy, University of Nevada Las Vegas, Las Vegas, NV 89154, USA Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, MS 169-221, Pasadena, CA 91109, USA (Received xxx; Revised xxx; Accepted)
Submitted to ApJABSTRACTWe study the X-ray spectra of a sample of 19 obscured, optically-selected Seyfert galaxies (Sy 1.8, 1.9 and2) in the local universe ( d ≤
175 Mpc), drawn from the CfA Seyfert sample. Our analysis is driven by thehigh sensitivity of
NuSTAR in the hard X-rays, coupled with soft X-ray spectra using
XMM-Newton , Chandra , Suzaku , and
Swift / XRT. We also analyze the optical spectra of these sources in order to obtain accurate massestimates and Eddington fractions. We employ four di ff erent models to analyze the X-ray spectra of thesesources, which all result in consistent results. We find that 79-90% of the sources are heavily obscured withline-of-sight column density N H > cm − . We also find a Compton-thick ( N H > cm − ) fraction of37 − N H and negatively correlatedwith the intrinsic, unabsorbed, X-ray luminosity (in agreement with the Iwasawa-Taniguchi e ff ect). Our resultssupport the hypothesis that radiation pressure regulates the distribution of the circumnuclear material. Keywords: accretion galaxies: active galaxies — galaxies: Seyfert — X-rays: general — X-ray active galacticnuclei INTRODUCTIONIt is generally accepted that active galactic nuclei (AGN)are powered by accretion onto supermassive black holes(SMBHs) with masses M BH ∼ − M (cid:12) through a geomet- Corresponding author: Elias [email protected] ∗ Based on observations made with ESO Telescopes at the La SillaParanal Observatory under program ID 099.A-0403. rically thin, optically thick disk (e.g., Shakura & Sunyaev1973). According to the unification scheme, originally pro-posed by Antonucci & Miller (1985) (see also Netzer 2015),all AGN are relatively similar in terms of physics. However,some key parameters such as orientation (e.g., Marin 2016),mass accretion rate (e.g., Fanidakis et al. 2011), and feed-back (see Fabian 2012, for a review) may di ff er, resultingin the di ff erent families of AGN classes (Antonucci 1993;Padovani et al. 2017). Within this general scheme, type-2objects are the AGN in which the absorber (broad line re- a r X i v : . [ a s t r o - ph . H E ] A ug K ammoun et al .gion and dusty torus), located at distances ∼ . −
10 pc(e.g., Ja ff e et al. 2004; Ramos Almeida & Ricci 2017), in-tercepts the line of sight (LOS). The actual morphology andcomposition of this material is still uncertain. Several resultssuggest a clumpy distribution of optically thick clouds ratherthan a homogeneous structure (e.g., H¨onig & Beckert 2007;Risaliti et al. 2007; Balokovi´c et al. 2014; Marinucci et al.2016). Recently, Giustini & Proga (2019) have explored therole of the inclination angle but more importantly the role ofthe luminosity in driving a disk wind that in turn can a ff ectthe AGN appearance (see their figure 5 for a comprehensivesummary).A significant fraction of AGN are obscured by the torus orsometimes by the host galaxy (e.g., Buchner & Bauer 2017;Circosta et al. 2019). The most rapid BH growth by accre-tion likely occurs in Compton-thick (CT; with an equivalentneutral hydrogen column density N H ≥ . × cm − )quasars at moderate / high redshift (1 ≤ z ≤
6, e.g., Draper& Ballantyne 2010; Treister et al. 2010; Vito et al. 2018).Mutli-wavelength studies (e.g., Risaliti et al. 1999; Gould-ing et al. 2011) suggest that a large fraction (20 − ∼ −
35% in di ff erent models and as a function of redshift.More recently, the cosmic X-ray background modeling byAnanna et al. (2019) predicts that ∼ ±
9% ( ∼ ± z = . ∼ ffi culty in identifying CT sources is the attenuation of thedirect emission produced by the central engine (in the soft X-rays, utlraviolet, and optical) by the obscuring material. Thehard X-rays ( ≥
15 keV) and the mid-infrared (5 − µ m) arethe only spectral bands where this material is optically thinup to high column densities.The X-ray emission in non-jetted AGN is widely acceptedto be due to Compton up-scattering of ultraviolet (UV) / softX-ray disk photons o ff hot electrons ( ∼ K; e.g., Shapiroet al. 1976; Haardt & Maraschi 1993), usually referred toas the X-ray corona. The resulting energy spectrum can bewell described by a power law with a high-energy cuto ff . Inobscured AGN, this component is heavily attenuated in thesoft X-rays (depending on the column density of the obscur-ing material). However, this attenuated emission is repro-cessed by the obscuring material. This reprocessing is im-printed in the energy spectrum in the form of narrow fluores-cent emission lines (most prominently the neutral Fe lines at ∼ . ∼ −
30 keV(e.g., Ghisellini et al. 1994). In order to accurately deter- mine the column density of the reprocessor and the propertiesof the intrinsic emission, a self-consistent modeling of thephysical processes, accounting for the aforementioned fea-tures, is required. Several e ff orts have been made to modelthese e ff ects, considering various geometries (e..g, spherical,toroidal, patchy; Murphy & Yaqoob 2009; Brightman & Nan-dra 2011; Liu & Li 2014; Balokovi´c et al. 2019; Buchneret al. 2019).Thanks to its unprecedented sensitivity covering the 3 −
79 keV band,
NuSTAR is playing a key role in identifying themissing fraction of CT sources and determining their prop-erties (e.g., Koss et al. 2016a; Annuar et al. 2017; March-esi et al. 2018, 2019; LaMassa et al. 2019, and referencestherein). In this work, we present a
NuSTAR survey observ-ing 21 optically-selected, obscured Seyfert galaxies. This pa-per is organized in the following way: in Section 2 we presentour sample. In Section 3 we present the analysis of the opti-cal spectra of the sources. In Section 4, we present the X-raydata reduction. In Section 5, we present the various modelsused to fit the X-ray spectra. The results of the X-ray spectralmodeling are presented in Section 6. Finally, we present ourconclusions in Section 7. THE SAMPLEThe CfA Redshift Survey (CfARS; Huchra et al. 1983)obtained the optical spectra of a complete sample of 2399galaxies down to a limiting magnitude of m Zw ≤ . δ ≥ ◦ and b II ≥ ◦ , and δ ≥ − . ◦ and b II ≤ − ◦ . Its classifications are robust owing to exhaustivehigh-resolution optical spectroscopy and narrow line classifi-cations (Osterbrock & Martel 1993). A subset of the CfARS,the CfA Seyfert Sample (CfASyS; Huchra & Burg 1992)originally identified 25 Seyfert 1 (i.e., unobscured) and 23Seyfert 2 galaxies. However, it was later shown that two ofthe Seyfert 2 sources (NGC 3227 and Mrk 993) were misclas-sified (see Salamanca et al. 1994; Corral et al. 2005; Trippeet al. 2010, and references therein, respectively). The re-maining 21 obscured sources in the CfASyS were observedthrough the NuSTAR Obscured Seyferts
Legacy Survey (PI: J.M. Miller), as the first volume-limited ( d ≤
175 Mpc) sampleobserved in the hard X-rays with a high-focusing and high-sensitivity instrument. The 21 sources, their coordinates, andredshifts are listed in Table 1. OPTICAL SPECTROSCOPYFrom our sample, 9 sources were observed with opticalspectroscopy as part of the BAT AGN Spectroscopic Survey(Koss et al. 2017) DR2 (Koss et al., in prep, where details onthe observations and data reduction can be found). In addi-tion to this, another 9 sources were observed as part of theSloan Digital Sky Survey (SDSS) and two sources (Mrk 573and Mrk 334) were observed with the Palomar Double Spec-trograph (DBSP) on the Hale 200-inch telescope. The latter hard look at local , optically - selected , obscured S eyfert galaxies Figure 1.
Optical excitation diagnostic (BPT) diagram to separateAGN from star-forming galaxies. The solid black curve is the theo-retical boundary for the region occupied by starburst derived by themaximum starburst model from Kewley et al. (2001). The dashedline is the empirical SF line from Kau ff mann et al. (2003). Thesolid straight line is the empirical Seyfert-LINER separation from(Schawinski et al. 2007). The Black dots mark our targets. Thegrey filled contours indicate the distribution of the SDSS narrowemission-line galaxies ( z < . N ∼ observations took place on UT 2018 January 14, with bothgalaxies observed for 300 s at the parallactic angle. ThePalomar observations were taken with the D55 dichroic andthe 600 / / (cid:48)(cid:48) slit, cov-ering the wavelength range of 3200 − = . = . / Xshooter(Chen et al. 2014) was used to fit the spectra with optimalstellar templates following the general procedure in Kosset al. (2017). These templates have been observed at higherspectral resolution ( R = , + K λ , i λ ii triplet spectral region (8450–8700 Å). We notethat reverberation mapping (e.g., Peterson et al. 2005) andOH megamasers (e.g., Staveley-Smith et al. 1992) o ff er moreprecise BH mass measurements. We adopted those esti-mates whenever available (megamaser: NGC 4388, reverber-ation mapping: NGC 4395). For UM 146, we use a veloc-ity dispersion measurement of 66 ± − from Garcia-Rissmann et al. (2005). However, we found this measure-ment was very near the instrumental limit (57 km s − ) wherethe changes in spectral resolution related to seeing and otherobserving conditions may significantly a ff ect the measure-ment. We therefore assume the velocity dispersion is lessthan 80 km s − , corresponding to a black hole mass of lessthan log M BH / M (cid:12) = .
7. We could not obtain velocity dis-persion measures for NGC 4395 and Mrk 334. In fact, usingthe available instrumental resolution, it is hard to measurethe stellar velocity dispersion in the dwarf galaxy NGC 4395(Woo et al. 2019). As for Mrk 334, the low AGN obscurationleads to contamination in the form of broad lines and contin-uum, which makes measuring a velocity dispersion di ffi cult.For emission line measurements, we performed spectralline fitting for 21 optical spectra using the Gas AND Ab-sorption Line Fitting ( gandalf ) IDL code (Sarzi et al. 2006)which has been extensively applied in measuring spectralline strengths of SDSS galaxies (Oh et al. 2011) and AGN(Oh et al. 2015). Stellar population synthesis model li-braries (Bruzual & Charlot 2003) and empirical stellar li-braries (S´anchez-Bl´azquez et al. 2006) are used to fit thecontinuum at rest-frame ∼ − III ] λ / H β versus [N II ] λ / H α .According to this diagnostic, all sources fall in the Seyfertregime except NGC 5256 and Mrk 461, which are consistentwith being in the composite region of the diagram. UM 146is in the LINER regime, though with a lower-limit on the[O III ] λ / H β ratio, which could be in the Seyfert regime.Throughout the rest of this analysis, we limit our studyto Seyfert galaxies only, excluding the composite sources(NGC 5256 and Mrk 461). It is also worth noting that mostof the sources are found to have broad lines. However, somesources reveal the presence of weak broad H α lines, whichmay be a signature of outflows. X-RAY DATA REDUCTIONIn this analysis, the high sensitivity of
NuSTAR above10 keV plays the major role, as it allows us to determinehigh column densities approaching the CT regime. How-ever, adding soft X-ray spectra is crucial to accurately esti-mate the spectral slope over the full X-ray band. For thatreason we also considered the soft X-ray spectra obtainedby
XMM-Newton , owing to its high sensitivity and excellentcalibration. In cases when
XMM-Newton observations were K ammoun et al . Table 1.
The results obtained from the study of the velocity dispersion. (1): Source name, (2): Right ascension, (3): Decli-nation, (4): redshift, (5): Stellar dispersion, (6): Instrumental dispersion, (7): Mass estimate, (8):Velocity dispersion fit range,(9): AGN classification. The superscripts in column 1 indicate the instruments used to estimate the velocity dispersions;Palomar ( P ), SDSS ( S ), Xshooter ( X ).Source RA Dec z σ ∗ (km s − ) σ inst (km s − ) log M BH / M (cid:12) Range (Å) Type(1) (2) (3) (4) (5) (6) (7) (8) (9)Mrk 573 P . ± .
06 116.64 7.65 3880-5550 Sy2NGC 1144 X . ± .
06 64.17 8.76 3800-5550 Sy2NGC 3362 S . ± .
39 65.22 7.58 3880-5550 Sy2UGC 6100 S . ± .
35 69.78 7.86 3880-5550 Sy2NGC 3982 S . ± .
29 63.75 6.89 3880-5550 Sy2NGC 4388 S . ± .
04 64.53 6.92 a S . ± .
38 65.67 6.88 3880-5550 Sy1.8NGC 5252 S . ± .
84 65.49 9.00 b S . ± .
73 65.82 7.02 3880-5550 Sy2NGC 5695 X . ± .
97 72.03 7.89 3880-5550 Sy2NGC 5929 S . ± .
50 62.07 7.64 3880-5550 Sy2NGC 7674 X . ± .
96 28.35 7.73 3800-5550 Sy2NGC 7682 X . ± .
11 65.15 7.42 3800-5550 Sy2NGC 4395 186.4538 33.5468 0.0011 5.56 c Sy1.8Mrk 334 P P . ± .
14 66.28 7.48 3880-5550 Sy2UM 146 28.8417 6.6117 0.0174 < . d NGC 5256 S . ± .
03 64.89 8.55 3880-5550 CompositeNGC 5674 P . ± .
15 82.59 7.47 Ca Triplet Sy2NGC 1068 P . ± .
26 85.94 6.93 d Ca Triplet Sy1.9Mrk 461 S . ± .
91 68.78 7.59 3880-5550 CompositeN ote —The values of σ ∗ represent the stellar velocity dispersion after subtracting the instrumental resolution (in quadrature).Black hole mass from the literature: a Kuo et al. (2011) (megamaser), b van den Bosch et al. (2015) (stellar velocity disper-sion), c Peterson et al. (2005) (reverberation mapping), d Garcia-Rissmann et al. (2005) (stellar velocity dispersion). Theother masses are from the BASS DR2 (Koss et al., in prep.; see text for more details). not available, we considered either
Suzaku or Chandra obser-vations, if possible. We also note that
Swift / XRT snapshotswere performed simultaneously with each
NuSTAR observa-tion. However, those observations resulted in low signal-to-noise ratio (S / N) spectra, due to the faintness of the sourcesin our sample. We considered the XRT spectra only in thecases when other soft X-ray spectra are not available, and thequality of the XRT spectra allows us to perform spectral fits.For NGC 1068, we limit our analysis to only one
NuSTAR observation. Adding the soft X-ray spectra of this source andaccounting for spectral variability has been discussed in fur-ther detail in previous studies (e.g., Bauer et al. 2015; Zainoet al. 2020) and is beyond the scope of our analysis. In thissection, we present the data reduction of the various instru-ments used in this analysis. The log of the observations ispresented in Table 3. 4.1.
NuSTAR observations
The
NuSTAR (Harrison et al. 2013) data were reducedfollowing the standard pipeline in the
NuSTAR
Data Anal-ysis Software (
NUSTARDAS v1.8.0), and using CALDBv20180814. We cleaned the unfiltered event files with thestandard depth correction. We reprocessed the data usingthe saamode = optimized and tentacle = yes criteriafor a more conservative treatment of the high backgroundlevels in the proximity of the South Atlantic Anomaly. Weextracted the source and background spectra from circularregions of radii 45 − (cid:48)(cid:48) and 100 (cid:48)(cid:48) , respectively, for bothfocal plane modules (FPMA and FPMB) using the HEASOFT task
Nuproducts , and requiring a minimum S / N of 4 perenergy bin. The spectra extracted from both modules areconsistent with each other. The data from FPMA and FPMBare analyzed jointly in this work, but they are not combined hard look at local , optically - selected , obscured S eyfert galaxies ff ect in any of the sources.4.2. XMM-Newton observations
We reduced the
XMM-Newton data using
SAS v.17.0.0 (Gabriel et al. 2004) and the latest calibration files. Wefollowed the standard procedure for reducing the data ofthe EPIC-pn (Str¨uder et al. 2001) CCD camera The datawere processed using EPPROC. Source spectra and lightcurves were extracted from a circular region with a radiusof 25 − (cid:48)(cid:48) . The corresponding background spectra and lightcurves were extracted from an o ff -source circular region lo-cated on the same chip, with a radius approximately twicethat of the source. We filtered out periods with strong back-ground flares. The spectra were then binned requiring a min-imum S / N of 4 per energy bin.4.3.
Suzaku observations
For NGC 5347 and NGC 5929 we used the XIS (Koyamaet al. 2007) spectra from
Suzaku (Mitsuda et al. 2007).The data were reduced following standard procedures using
HEASOFT . The initial reduction was done with aepipeline ,using the CALDB calibration release v20160616. Sourcespectra were extracted using xselect from circular regions3 (cid:48) and 2.5 (cid:48) in radius centered on the sources, for NGC 5347and NGC 5929, respectively. Background spectra were ex-tracted from a source-free region of the same size, away fromthe calibration source. The response files were generated us-ing xisresp . We do not consider the spectrum from XIS1,owing to its poor relative calibration. Spectra from XIS0 andXIS3 were checked for consistency and then combined toform the front-illuminated spectra. The spectra were thenbinned requiring minimum S / N of 3 per energy bin.4.4.
Chandra observations
For NGC 5347 and NGC 5283 we used the archival
Chan-dra (Weisskopf et al. 2000) spectra. The data were reducedusing CIAO version 4.9 and the latest associated calibrationfiles. Source and background spectral files and response fileswere all created using the CIAO tool specextract . We ex-tracted source counts from a circular region, centered on theknown source coordinates, with a radius of 5.2 (cid:48)(cid:48) and 3 (cid:48)(cid:48) forNGC 5347 and NGC 5283, respectively. The background The inclusion of the EPIC-MOS data would have increased the signalto noise in the soft X-rays (below ∼ ff use emission (see Section 5 formore details). This will not lead to any improvement in measuring the LOScolumn density or the intrinsic emission. For this reason, and to avoid anyuncertainties due to instrument cross-calibration, we decided not to use theMOS data. spectra were extracted from circular source free regions, withradii equal to the ones of the source region. The resultant datawere grouped to require at minimum S / N of 3 per energy bin.4.5.
Swift observations
For Mrk 334 and NGC 5674 we used the X-ray telescope(XRT; Burrows et al. 2005) spectra from
Neil Gehrels SwiftObservatory (hereafter
Swift ; Gehrels et al. 2004). We com-bined all the XRT observations for each source in order toincrease the number of counts. The data were reduced fol-lowing standard procedures using
HEASOFT . The initial re-duction was done with xrtpipeline . Source spectra wereextracted using xselect from circular regions 20 (cid:48)(cid:48) in radiuscentered on the source. Background spectra were extractedfrom an o ff -source sky region of the same size. We usedthe default redistribution matrix file (RMF) and ancillary re-sponse file (ARF), available in the calibration database. Thespectra were then binned requiring minimum S / N of 3 perenergy bin. X-RAY SPECTROSCOPYThroughout this work, spectral fitting was performed usingXSPEC v12.10.1o (Arnaud 1996). We apply the χ statisticfor spectra with more than 20 counts per bin, and the Cashstatistic ( C -stat; Cash 1979) otherwise. The assumed statis-tics for each source are presented in the last column of Ta-ble 3. The best-fit parameter values were determined throughthe minimization of the χ ( C -stat) . These values were usedfor initial values for the prior distributions in Markov chainMonte Carlo (MCMC ). We used the Goodman-Weare al-gorithm (Goodman & Weare 2010) discarding 300 ,
000 ele-ments as part of the burn-in period. The final chains contain500 ,
000 elements. The values presented in this paper repre-sent the mean values across the chain samples. Unless statedotherwise, uncertainties on the parameters are listed at the 1 σ confidence level, as derived from the chain samples.In the following, we present the di ff erent models that weused in order to describe the reprocessed emission in the X-ray spectra. We note that the various model set-ups used inthis paper have been already presented in Kammoun et al.(2019a), where we discuss in detail the case of NGC 5347.5.1. Pexmon
We initially fit the spectra using the neutral reflectionmodel Pexmon (Nandra et al. 2007). This model is based on It has been argued in the literature that C -stat, contrary to χ -statistic,yields unbiased results (e.g., Mighell 1999; Kaastra 2017). However, at thelimit of 20 counts / bin, the use of χ may produce a bias of the order of 5%in the best estimate of the flux for those bins. This will have a minor e ff ecton the results of our analysis, which are dominated by statistical errors. We use the
XSPEC EMCEE implementation of the
PYTHON EMCEE (Foreman-Mackey et al. 2013) package for X-ray spectral fitting in XSPEC,provided by A. Zoghbi (https: // github.com / zoghbi-a / xspec emcee). K ammoun et al .the Pexrav model (Magdziarz & Zdziarski 1995), but addsthe Fe K α fluorescence line based on the Monte Carlo simu-lations by George & Fabian (1991). In addition, Fe K β andNi K α fluorescence lines lines are added in Pexmon, in a self-consistent manner, with their fluxes fixed at a fraction of theFe K α flux. This model accounts also for the Compton shoul-der of the Fe K α , following the prescription of Matt (2002).The model that we used to fit the spectra can be written (inXSPEC parlance) as follows: model Pexmon = phabs [ ] ∗ ( zphabs [ ] ∗ zcutoffpl [ ] + constant [ ] ∗ zcutoffpl [ ] + pexmon [ ] + Apec [ ] + Apec [ ]) . In this model, the phabs[1] component represents theGalactic absorption, zcutoffpl[3] represents the primaryemission of the source assumed to be a power-law with ahigh-energy cuto ff (fixed to 500 keV), which is intrinsicallyabsorbed by zphabs[2] . A fraction ( constant [ ] ∗ zcutoffpl [ ],where 0 ≤ constant [ ] ≤
1) of the primary emission couldbe scattered into our LOS by optically thin ionized gas in thepolar regions. This fraction could also partly account for un-absorbed emission in the case of a partially covered source.We do not instead assume a partially covering absorber forconsistency with other models that are used in this work (seelater for more details). In the rest of the work we refer to constant[4] as C sc . All the parameters of zcutoffpl[6] are tied to the ones of zcutoffpl[3] . The photon index,cuto ff energy and normalization of pexmon[7] , which de-scribes the reprocessed emission, are tied to the same pa-rameters of zcutoffpl[3] . We leave the reflection fraction R of pexmon[7] free to vary, taking negative values only toaccount for the reflected emission without the contribution ofthe power-law component. This component is used to modelthe reprocessed emission from neutral material without as-suming any specific geometry or location. Finally, whenneeded, we describe the soft emission with Apec 1 [ ]. Thiscomponent, which mainly arises from extended regions, ismost likely due to di ff use thermal emission and / or photoion-ized emission. In some cases, modeling the soft X-rays re-quire the addition of a second component ( Apec [ ]) with ahigher temperature. In the case of multi-epoch observations,we left the normalization of the power-law component freeto vary, to account for potential flux variability. If the best-fitvalues of normalization are consistent between the di ff erentepochs, indicating no significant variability, we then tie themand repeat the fit.NGC 3362 and UGC 8621 were not detected by NuSTAR .In addition, their
XMM-Newton spectra are background-dominated above ∼ pexmon[7] component for these two sources. However, to avoid further complexity in presenting our analysis, weconsider the models for these sources (i.e., by removing thePexmon component) in the “Pexmon” set (as opposed to theMYTorus fits, see next section). The results assuming thismodel are presented in Table 4.5.2. MYTorus
In the previous section, we presented a phenomenologicalmodel to fit the spectra. However, this model does not as-sume any geometry. Moreover, it does not account for thescattering within the obscuring material. Thus, we next at-tempt to model the obscuration and the reprocessed emis-sion using the MYTorus spectral-fitting suite for modelingX-ray spectra from a toroidal reprocessor (Murphy & Yaqoob2009). This model fixes only the geometry of the reprocess-ing material without necessarily implying a pc-scale location.We first consider the “coupled” configuration of MYTorus(hereafter MYTC). This configuration assumes that intrin-sic emission is self-consistently absorbed and reprocessedby toroidal material with a circular cross section and half-opening angle of 60 ◦ and solar abundance. The viewing an-gle ( θ ) and the equatorial (global) column density ( N H , eq ) ofthe torus are free parameters. We refer the reader to Yaqoob(2012) for more details and an extensive discussion about thevarious configurations of MYTorus and their implications.The model can be written as follows: model MYTC = phabs [ ] ∗ ( MYTZ [ ] ∗ zpowerlaw [ ] + constant [ ] ∗ zpowerlaw [ ] + constant [ ] ∗ ( MYTS [ ] + MYTL [ ]) + Apec [ ] + Apec [ ]) . The phabs [ ], constant [ ] ∗ zpowerlaw [ ], Apec [ ],and Apec [ ] components are equivalent to the ones in thePexmon fit. MYTZ[2] represents the attenuation of the in-trinsic emission.
MYTS[7] and
MYTL[8] represent the scat-tered continuum and the fluorescent emission lines emittedby the torus. The constant[6] factor (hereafter, referred toas A ) corresponds the relative weights of the three MYToruscomponents and are fixed to unity (as suggested by Yaqoob2012), unless stated otherwise. We remind the reader that N H , LOS can be estimated using using eq. (1) in Murphy &Yaqoob (2009): N H , LOS = N H , eq (cid:34) − (cid:18) ca (cid:19) cos θ (cid:35) / , (1)where c is the distance from the center of the torus to theorigin of coordinates, and a is the radius of the circular crosssection. For a half-opening angle of 60 ◦ , the ratio c / a is equalto 2. MYTorus does not have a high-energy cuto ff . Impos-ing a cuto ff energy to the primary emission would break the hard look at local , optically - selected , obscured S eyfert galaxies E T ). We used in our analysisthe tables with E T =
500 keV. Using di ff erent values did nota ff ect the fits. The results obtained by fitting MYTC are pre-sented in Table 5.Next, we considered the decoupled configuration of MY-Torus (MYTD) which is intended to mimic the Pexmon con-figuration. In this configuration, the viewing angle of MYTZ is fixed to 90 ◦ , so its N H corresponds to the LOS value. MYTS and
MYTL are decomposed into two components, one from thenear side of the torus ( θ = ◦ ) and the one from the far sideof the torus ( θ = ◦ ). The column densities of these compo-nents could be either tied to the one of MYTZ , correspondingto a uniform distribution of the material (hereafter MYTD-1NH), or free to vary, corresponding to a patchy structure(hereafter MYTD-2NH). The latter configuration allows usto obtain estimates on both N H , LOS and N H , eq . MYTD can bewritten as follows: model MYTD = phabs [ ] ∗ ( MYTZ [ ] ∗ zpowerlaw [ ] + constant [ ] ∗ zpowerlaw [ ] + constant [ ] ∗ ( MYTS [ ] + MYTL [ ]) + constant [ ] ∗ ( MYTS [ ] + MYTL [ ]) + Apec [ ] + Apec [ ]) . The relative weights for the
MYTS and
MYTL components with θ = ◦ , ◦ ( constant[6,9] ) are tied together, unless oth-erwise stated. Hereafter, we refer to those constants as A and A , respectively. The results obtained by fitting MYTD-1NH and MYTD-2NH are presented in Tables 6 and 7, re-spectively.We did not apply the MYTorus models for the followingsources: NGC 5252, NGC 4395, NGC 3362, UGC 6100,and UGC 8621. For NGC 5252 and NGC 4395, the spectrarequired complex, ionized absorption, in addition to a neutralabsorber. Such a configuration is not trivial to implement inthe framework of the MYTorus models. A detailed analysisof NGC 4395 is presented in Kammoun et al. (2019b) (seealso Nardini & Risaliti 2011). As for NGC 3362, UGC 6100,and UGC 8621, the low quality of the data (for UGC 6100)makes the application of the MYTorus fits non-trivial. Asfor NGC 3362 and UGC 8621, both sources are not detectedby NuSTAR and their
XMM-Newton spectra are backgrounddominated above ∼ Γ to the range 1.4-2.4 becausethe models we used (Pexmon and MYTorus) are defined inthis common range. RESULTS All models throughout this paper resulted in statisticallyacceptable fits. In addition, the results for each source, ap-plying the various models considered in our analysis, agreedin detail (see Section 6.1 for more details). Thus, in Fig-ures 13-14, we show only the spectra fitted with Pexmon,and the corresponding residuals. The Γ − N H confidence con-tours for each model are shown in Figures 15-18. For thecases where Γ was kept tied, we show only the 1-D posteriordistribution N H obtained from the MCMC. We show the re-sults from the MCMC in Tables 4-7. We note that the best-fitresults for the Apec components are consistent between allfour models. Thus, we show the ones obtained by the Pex-mon model only. It is worth mentioning that for all sourcesthe iron abundance in the Pexmon model was fixed to the so-lar value, except for NGC 1068. For this source, we obtaineda best-fit value A Fe = . ± . C sc were found to be smaller than 0.1 for all sources, ex-cept for NGC 7674. For this source, we found C sc = . + . − . and 0 . ± . XMM-Newton and
NuSTAR spectra, re-spectively, fitted with Pexmon.Throughout the paper, we refer to the intrinsic, unab-sorbed, luminosity of the power-law component in the 2-10 keV range as L − . Various methods can be used to esti-mate the bolometric luminosity. One way to do so is by con-sidering luminosity-dependent bolometric corrections (e.g.,Marconi et al. 2004; Lusso et al. 2012). However, in Table 2,we present the Eddington ratios ( λ Edd = L bol / L Edd ) by usingthe mass estimates from Section 3 and adopting a bolometriccorrection, κ bol = L bol / L − =
20, from Vasudevan & Fabian(2009). Koss et al. (2017) noted that L bol inferred from theX-ray luminosity and the values estimated from the 5100 Åluminosity show a scatter of ∼ .
45 dex for type-1 AGN (seetheir figure 26). Assuming di ff erent bolometric correctionwould result in di ff erent bolometric luminosities and Edding-ton fractions. This would cause some later plots to shift (e.g.,Figures 10 and 11), but the overall trends identified wouldremain. The estimated λ Edd values are model-dependent, asthey rely on the measure of L − that may vary between dif-ferent models. The consistency between the models is furtherdiscussed in Section 6.1. In addition, λ Edd span a large rangelog λ Edd [ − , − .
5] that is di ff erent from the one observed inquasars. Thus any direct comparison between our sampleand highly accreting quasars should be addressed carefullyas they probe di ff erent accretion regimes.In Figure 2 we show a corner plot for all the relevant pa-rameters for all the observations. The results are shown forPexmon, MYTD-1NH, and MYTD-2NH, as these modelsmeasure the LOS column density. We plot each measure-ment for sources showing variability. We show in the sameplot the Kendall’s coe ffi cient with the null hypothesis proba-bility (i.e., no correlation), between parentheses, for the Pex-mon estimates. The results are consistent with the ones ob- K ammoun et al . . . . N H , L O S ( c m ) K = 0.24 (0.052) PexmonMYTD 1NHMYTD 2NH . . L X ( e r g s ) K = 0.04 (0.732) K = 0.10 (0.397) . . K = 0.16 (0.203) N H, LOS (10 cm ) K = 0.41 (0.001) L X (10 erg s ) K = 0.04 (0.733) Figure 2.
Triangle plot showing the best-fit parameters plotted versus the others, for Pexmon (blue circles), MYTD-1NH (red diamonds), andMYTD-2NH (open black squares). We also show the Kendall correlation coe ffi cient and the corresponding p -value between parentheses. Wenote that objects with multi-epoch observations are reported multiple times. tained for the other two models. Our results suggest moderatepositive correlations between the N H , LOS and Γ , and between N H , LOS and R . We find no correlation between Γ and R . Thisis in agreement with the recent results of Panagiotou & Wal-ter (2020) who also found no correlation between these twoquantities in obscured AGN, while they found a clear corre-lation in unobscured sources (see also Zdziarski et al. 1999,2003). In Figure 3, we show the histograms for the measured N H , LOS , Γ , L − , and λ Edd , for all models. MYTC providesonly N H , eq . Thus, we used eq. 1 to derive the N H , LOS assum-ing the best-fit inclination angle. The results in Figures 2-3 are shown for each measurement. Hence, more than onedata point will be associated with sources showing variabil-ity. Moreover, some sources were not fitted with MYTorus,which result in having less data points for these models com-pared to Pexmon. Figure 4 shows the scatter plot for the ratio of theextinction-corrected [O
III ] luminosity over the intrinsic2-10 keV luminosity plotted versus L − and N H , LOS ob-tained using the Pexmon model (left and right panels, re-spectively). The binned data points suggest constant ra-tios, below L − < erg s − with an average value of (cid:104) log L [O III] / L − (cid:105) = − .
09. A decrease in L [O III] / L − can be seen at high X-ray luminosity, L − > erg s − .The results obtained by Berney et al. (2015) also show aconstant luminosity ratio, albeit with a lower average of (cid:104) log L [O III] / L − (cid:105) = − .
01. Howerver, Berney et al. (2015)studied hard X-ray selected objects from the
Swift / BAT AGNsurvey, with log L − ∼ [42 , Consistency of the models hard look at local , optically - selected , obscured S eyfert galaxies
22 23 24 25 26 log N H (cm ) N u m be r o f S ou r c e s (a) N u m be r o f S ou r c e s (b) PexmonMYTD-1NHMYTD-2NHMYTC log L (erg s ) N u m be r o f S ou r c e s (c) log Edd N u m be r o f S ou r c e s (d) Figure 3.
The distributions of N H , LOS , Γ , L − , and log λ Edd (pan-els a-d) for Pexmon, MYTC, MYTD-1NH, and MYTD-2NH (bluefilled, green dotted, red solid, and black dashed histograms, respec-tively). We note that objects with multi-epoch observations are re-ported multiple times.
40 41 42 43 44 log L (erg s ) l o g ( L [ O III ] / L )
22 23 24 25 26 log N H (cm ) Figure 4.
Ratio of the extinction corrected [O
III ] luminosity overthe intrinsic 2-10 keV luminosity versus L − and the LOS col-umn density obtained using the Pexmon model (left and right pan-els, respectively). The results obtained for each source are shown inshaded blue. The orange points correspond to the binned results. In this section we address the consistency of the resultsobtained from the di ff erent models that are considered in thiswork. The major di ff erence between these models residesin the way that Pexmon and MYTorus treat the reprocessedemission. Pexmon considers an infinite slab of neutral mate-rial that reflects the incident X-ray emission. However, MY-Torus considers Compton scattering (in its various regimes)in a torus with a fixed opening angle, providing continuouscoverage from the Compton-thin to Compton-thick columndensities. In addition, some di ff erences exist between thedi ff erent configurations of MYTorus, as described in Sec-tion 5.2 (see also Murphy & Yaqoob 2009; Yaqoob 2012).In Figure 5, we plot the measured values of N H , LOS , Γ , and L − (panels a, b, and c, respectively) obtained from the var-ious models. We note that MYTC measures N H , eq , thus weused eq. 1 to obtain the corresponding N H , LOS and assumingthe best-fit values of θ , listed in Table 5. In this case, the un-certainty on N H , LOS is estimated using the one on N H , eq only.As for the uncertainty on L − , we assumed the relative un-certainties obtained from the power law normalization. Inpanel d of the same figure, we plot the values of χ / dof (or C / dof) that are obtained from the best-fit models in XSPEC,and used as an initial guess for the MCMC analysis. Thispanel demonstrates that all models resulted in statisticallygood fits. In the case of Mrk 573, the χ / dof is larger than1.2. However, the fit is still statistically acceptable with anull hypothesis probability of 0.06. In contrast, UGC 6100,NGC 3982, and UGC 8621 show very low C / dof. This isdue to the low number of degrees of freedom for these spec-tra. Only in a few cases, does the MYTorus set of modelsprovide a statistical improvement (in terms of χ / dof) withrespect to Pexmon.The column densities are consistent within error bars be-tween the four models in the majority of the cases, exceptfor NGC 3982 and NGC 7674. For NGC 3982, Pexmon0 K ammoun et al . N H ( c m ) (a) . . . . . . (b) L ( e r g s ) (c) M r k N G C N G C U G C N G C N G C U G C N G C N G C N G C N G C N G C N G C N G C M r k N G C U M N G C . . . . . . . ( C ) (d) PexmonMYTCMYTD-1NHMYTD-2NH
Figure 5.
The best-fit N H , LOS , Γ , and unabsorbed L − (panels a,band c, respectively) obtained using the Pexmon (blue filled circles)and three MYTorus models (MYTC: green open circles, MYTD-1NH: red diamonds, and MYTD-2NH: black open squares), for allsources. Panel d shows the values χ ( C ) / dof of the best-fit mod-els, using XSPEC, that were used as initial guesses for the MCMCanalysis. Power law10 C o un t s c m k e V Reprocessed
PexmonMYTD-1NH
Energy (keV) Total
Figure 6.
The best-fit model components: absorbed power law, re-processed emission, and total model (top to bottom) obtained byfitting the
NuSTAR spectrum of NGC 1144 with Pexmon (solid blueline) and MYTD-1NH (dashed red line). The dotted lines in the toppanel correspond to the intrinsic (unabsorbed) power law compo-nent for each case. predicts a Compton-thin (Cth) column density that is signif-icantly smaller than the CT values obtained by MYTD-1NHand MYTD-2NH. For NGC 7674, MYTD-1NH indicates alarge and CT column density, while the other models suggestlower and Cth values. In this case, for MYTD-1NH, the fitis statistically worse than the ones for the other models (seeSection A for more details). The photon indices are broadlyconsistent within uncertainties between the various models,albeit with a large scatter. The intrinsic 2-10 keV luminosityis also consistent between the four models within uncertain-ties. However, this plot suggests that the luminosity estimatesusing the various MYTorus configurations tend to be largerthan the ones obtained using Pexmon, especially for high val-ues of N H . This di ff erence is particularly large in the case ofNGC 5347.In order to assess the reason behind this discrepancy, weplot in Figure 6 the best-fit Pexmon and MYTD-1NH mod-els obtained by fitting the NuSTAR spectra of NGC 1144.The best-fit results are N H = + . − . (1 . + . − . ) × cm − and L − = . . × erg s − for Pexmon (MYTD- hard look at local , optically - selected , obscured S eyfert galaxies N H, LOS (10 cm ) N u m be r o f S ou r c e s Sources ( N H > 10 cm ) . . . . . . . . p
10 12 14 16 18 20
Sources ( N H > 10 cm ) . . . . . p Figure 7.
Left panel: The N H , LOS distribution obtained from di ff erent simulations. The solid black line corresponds to the distributions obtainedby averaging the realizations for each source (see Section 6.2 for details). Middle panel: The distribution of the number of heavily obscuredsources with N H > cm − . Right panel: The distribution of the number of CT sources with N H > cm − . pexmon [ ] and A ∗ ( MYTS [ ] + MYTL [ ]) + A ∗ ( MYTS [ ] + MYTL [ ]),for Pexmon and MYTD-1NH, respectively (see Section 5).It is clear from this figure that MYTD-1NH requires largerintrinsic luminosity compared to Pexmon. MYTD-1NH in-dicates a lower absorbed power-law flux below ∼
15 keVcompared to Pexmon. This di ff erence in flux is compensatedfor when considering the reprocessed emission. As a re-sult, the total emission is consistent between the two models.This e ff ect becomes larger for sources with higher N H , inwhich the intrinsic power-law component cannot be clearlyidentified.For NGC 1068, Pexmon resulted in large N H , LOS = . + . − . × cm − . Such high values cannot be achievedby MYTorus which has an upper limit of 10 cm − . Forthat reason we limit the fits for this source to Pexmon andMYTC only, that resulted in χ / dof = .
07 and 1.12, re-spectively. In addition, the fits required a Gaussian emissionline at ∼ . − ffi cul-ties in modeling the spectra of this source, we excluded allits best-fit parameters from the rest of our analysis, except for N H , LOS that has been well-established in the literature asbeing CT. 6.2.
Compton-thick fraction
Thanks to the high sensitivity of
NuSTAR in the hardX-rays we were able to identify three new CT sources:NGC 5347, NGC 5695, and UGC 6100. NGC 5695 wasclassified by LaMassa et al. (2009) as a CT candidate basedon its
XMM-Newton spectrum. In a recent work, Zhao et al.(2020) identified Mrk 573 as a CT source, that is also a part ofour sample. We identified also two sources (NGC 3362 andUGC 8621) as being CT, based on their
XMM-Newton spec-tra. In addition, we analyzed a
NuSTAR spectrum of the well-known CT source, NGC 1068. Furthermore, two sources(NGC 3982 and NGC 7674) have been identified as Comptonthin when fitted with Pexmon, but CT using MYTD-1NH.Our results for NGC 7674, using Pexmon, are in agreementwith the ones found by LaMassa et al. (2011) and Tanimotoet al. (2020), but di ff er from the ones of Gandhi et al. (2017)who classified the source as CT. NGC 1144 was found to beat the edge of the CT limit with N H (cid:39) cm − . As a re-sult, based on the N H measures obtained from our spectralfits, we identify seven CT sources, and three CT candidates(NGC 3982, NGC 7674, and NGC 1144). In order to find theunderlying distribution of N H from our sample, taking intoaccount variability seen in some sources and the uncertaintyon our measurements, we performed the following simula-tion.2 K ammoun et al . N H, LOS (10 cm ) . . . . . . S C NGC 7674Mrk 573NGC 1144NGC 3982 UGC 6100NGC 5347NGC 5695 NGC 1068 U G C N G C Figure 8.
Spectral curvature as a function of N H , LOS obtained fromPexmon. The red dotted lines show the CT limit for
S C > . N H > . × cm − . The orange and grey shaded areascorrespond to the N H , LOS estimates of UGC 8621 and NGC 3362,respectively.
Given the general consistency between the di ff erent mod-els, we consider the results obtained using Pexmon only, asthis model could be applied for all the sources in our sam-ple. Using the N H , LOS distributions obtained from the MCMCanalysis, we draw randomly a value for each source. Forsources with variable N H , LOS , we choose one estimate ran-domly. We repeated this 10 ,
000 times and considered the N H distribution for each realization. In the left panel of Figure 7we plot in color the histograms obtained from various real-izations. The solid black line correspond to the histogramsobtained by averaging the estimated N H , LOS for each source.The middle and right panels of the same figure show thedistributions of the number of sources with N H , LOS > and 10 cm − ), respectively, obtained from each realization.As a result, these histograms show that our sample contains6 − −
17 sources are heavily obscured with N H > cm − in 94% of the cases. Table 2 . The Eddington ratios obtained by fitting the data with Pexmon,MYTD-1NH, and MYTD-2NH. The fourth column corresponds to the spec-tral curvature estimated from the
NuSTAR observations. The last two columnsshow whether the source is classified as CT (marked with (cid:51) ) or not (markedwith (cid:55) ) according to the spectral curvature and N H , LOS , respectively. Sourcesthat are considered as CT candidates are marked with a ? mark (see Sec-tion 6.2 for details).Source log λ Edd
S C CT SC CT N H Pexmon M-1NH M-2NH
Mrk 573 -2.73 -2.90 -2.86 0 . ± . (cid:51) (cid:51) NGC 1144 -1.86 -1.88 -1.76 − − (cid:55) -2.39 -2.14 -2.17 0 . ± . (cid:51) ?NGC 3362 -2.76 − − − − (cid:51) UGC 6100 -2.04 − − . ± . (cid:51) (cid:51) NGC 3982 -3.26 -3.60 -3.61 0 . ± . (cid:51) ?NGC 4388 -1.24 -1.23 -1.19 − − (cid:55) -0.82 -0.72 -0.72 − − (cid:55) -0.57 -0.50 -0.50 − − (cid:55) -1.28 -1.24 -1.24 0 . ± . (cid:55) (cid:55) UGC 8621 -0.72 − − − − (cid:51)
NGC 5252 -2.53 − − − − (cid:55) -2.46 − − . ± . − (cid:55) NGC 5347 -3.02 -1.67 − . ± . (cid:51) (cid:51) NGC 5695 -2.64 -1.76 -2.04 0 . ± . (cid:51) (cid:51) NGC 5929 -2.75 -2.74 -2.75 − − (cid:55) -2.91 -2.88 -2.89 0 . ± . (cid:55) (cid:55) NGC 7674 -2.15 -2.24 -2.08 − − (cid:55) -2.32 -2.04 -2.01 0 . ± .
05 ? ?NGC 7682 -2.34 -2.50 -2.57 0 . ± . (cid:55) (cid:55) NGC 4395 -2.12 − − − − (cid:55) -1.89 − − − − (cid:55) -1.72 − − − − (cid:55) -2.04 − − . ± . (cid:55) (cid:55) -1.75 − − . ± . (cid:55) (cid:55) -2.19 − − − − (cid:55) -1.99 − − − − (cid:55) -1.83 − − − − (cid:55) Mrk 334 − − − . ± . (cid:55) (cid:55) NGC 5283 -2.24 -2.26 -2.16 0 . ± . (cid:55) (cid:55) UM 146 -2.44 -2.46 -2.35 0 . ± . (cid:55) (cid:55) NGC 5674 -0.90 -1.11 -1.06 0 . ± . (cid:55) (cid:55) NGC 1068 − − − . ± . (cid:51) (cid:51) hard look at local , optically - selected , obscured S eyfert galaxies NuSTAR detection asfollows:
S C = − . × E + . × F + . × GT , (2)where E , F , G , and T are the count rates in the 8 −
14 keV,14 −
20 keV, 20 −
30 keV, and 8 −
30 keV ranges, respectively.A source is classified as CT if
S C > .
4. The estimated
S C are presented in Table 2. In Figure 8, we plot
S C as a func-tion of the measured N H , LOS obtained using Pexmon. Accord-ing to the
S C metric, the five sources (Mrk 573, NGC 5347,UGC 6100, NGC 5695 and NGC 1068) that were detectedby
NuSTAR and considered as CT, based on the spectral fits,are also found to be CT. Interestingly, the two CT candidates(NGC 3982 and NGC 1144) based on the spectral fits, arefound to be CT based on the
S C metric. NGC 7674 is foundto be at the edge of the CT limit with
S C = . ± .
05. Sowe also consider it as a CT candidate based on the
S C met-ric. Two sources (NGC 3362 and UGC 8621) were found tobe CT sources based on their spectral fits could not be con-firmed by the
S C metric because they were not detected with
NuSTAR .In conclusion, our results suggest that 78 . − .
5% ofthe sources in our sample are heavily obscured with N H > cm − . Furthermore, based on both methods we pre-sented above, the number of CT sources in our sampleranges between 6 and 10 sources. This corresponds to31 . − .
6% of our sample being CT. Thus for the com-plete CfASy sample of 46 sources (accounting for all ob-scured and unobscured Seyferts), our results indicate that32 . − .
9% (13 − . N H > ( > cm − ). Our results are in full agree-ment with the ones of Risaliti et al. (1999) who found that77 . ± .
6% of Seyfert-2 galaxies are heavily obscured with N H > cm − and 47 . ±
10% are CT. Panessa et al. (2006)found a CT fraction of 30 −
50% of the Sy2 sources from thePalomar optical spectroscopic survey of nearby galaxies (Hoet al. 1995, 1997). However, flux-limited hard X-ray surveyshave reported a significantly lower observed fraction of CTsources (below ∼ Swift / BAT catalog to be 7 . + . − . %. We note that the sources inRicci et al. (2015) span broader redshift and X-ray luminos-ity ranges than the sources in our sample. In particular, thesources from the Swift / BAT AGN catalog include more lu-minous sources compared to the one presented in this work.Koss et al. (2016b), using a lower redshift range that is sim-ilar to ours, found 22% of hard X-ray selected BAT AGN asCT utilizing the spectral curvature metric. Our results con- firm the conclusions by Cappi et al. (2006) and Malizia et al.(2009) who showed that a careful selection of a completevolume limited sample reconciles the fraction of heavily ob-scured (and CT) sources obtained from multi-wavelength es-timates and the ones from X-ray spectral fitting. It is worthnoting that our sample of obscured sources is based on opti-cal selection. However, it might be possible that some elusiveobjects have been missed by the optical classification due tomissing narrow H β lines. For example, Koss et al. (2017)found that ∼
30 % of the X-ray selected AGN are missingnarrow H β lines. This is mainly because X-ray selected AGNhave much higher Balmer decrements and more likely to bein edge on galaxies (see e.g., Koss et al. 2011). In this case, itis likely that these sources are highly embedded to be missedwithin the small volume. This would lead to a higher num-ber of highly obscured sources than the ones we identified.For example, Smith et al. (2014) studied four optically elu-sive AGN and four X-ray bright, optically normal galaxies.Five sources of their sample are at z < .
03, comparable tothe redshift range of the sources studied in this work. Theauthors found that four out of those five sources are highlyobscured. 6.3.
Moderately obscured sources
As mentioned in the previous section, we found that 16sources are heavily obscured with N H > cm − . Inthis section, we focus on the three sources that show mod-erate obscuration (10 cm − < N H < cm − ), namelyNGC 4395, UM 146 and NGC 5674. Trouille et al. (2009)have shown that 33% ±
4% of their sample of optically-selected obscured sources are X-ray unobscured. In ourcase, considering the three moderately-obscured sources inour sample, it is well-known that NGC 4395 exhibits strongvariability in its obscuring column. Those variations are as-sociated with obscuring clouds in the BLR (see Nardini &Risaliti 2011; Kammoun et al. 2019b). Instead, UM 146and NGC 5674 are two face-on galaxies. Thus, it is lesslikely that the obscuration is due to gas or dust in their hostgalaxies. Trippe et al. (2010) classified these sources as“true” Sy 1.8 and Sy 2, respectively. Risaliti (2002) analyzedthe
BeppoSAX spectrum of NGC 5674 (observed in Febru-ary 2000, i.e., 14 years prior to the
NuSTAR observation).They reported an obscuring column density of the order of6 × cm − that is consistent with our measurement. Fur-ther monitoring observations of these sources will be thennecessary to search for any possible variability. This will al-low us to understand the nature of the moderate obscurationin these two sources.6.4. On the obscurer structure
It is not straightforward to obtain constraints on the struc-ture of the obscuring material (e.g., geometry, filling factor,4 K ammoun et al . L (erg s ) . NGC 4395 (a) log = ( 1.04 ± 0.20) log L + (41.61 ± 8.08)log = ( 0.35 ± 0.10) log L + (14.96 ± 4.07) F ref / F PL (b) log = (1.32 ± 0.18) log ( F ref F PL ) + (0.65 ± 0.06) log N H N H, LOS (cm ) . (c) log = (0.87 ± 0.20) log N H +( 20.38 ± 4.72) L + 0.36 log N H (d) log = ( 0.28 ± 0.09) log L +(0.36 ± 0.15) log N H + (3.55 ± 6.17) Figure 9.
Panel a: Reflection fraction as a function of the intrinsic (unabsorbed) X-ray luminosity in the 2 −
10 keV range. Two linear fits inthe log-log space has been performed: the data points of NGC 4395 (obtained from flux-resolved spectra extracted from four observations) andthe rest of the sources (dashed and solid lines, respectively). Panel b: Reflection fraction as a function of the reflected flux over power-law fluxratio, obtained using Pexmon. The data is fitted with a linear model in the log-log plot (solid line) that results in a slope consistent with unity.The color map used in this plot corresponds to the fitted value of N H , LOS obtained using Pexmon. Panel c: Reflection fraction as a function ofthe LOS column density, fitted with a linear relation in the log-log space. Panel d: The best-fit
R − L − − N H plane obtained by excludingNGC 4395. opening angle, orientation) using our models. This task be-comes harder given the fact that all models provide statis-tically good fits. However, some valuable information canstill be extracted from our results. In particular, a more care-ful look at the R vs L − plot, shown in Figure 9a (alsoin Figure 2), reveals the presence of a correlation betweenthese two quantities. In fact, one can see that the dwarfgalaxy NGC 4395 does not follow the rest of the sources.Considering the full data set results in a Kendall correla-tion coe ffi cient τ K = − .
04 with a null hypothesis proba-bility of 0.73. However, splitting the data set in two resultsin τ K = − . / − .
41 with null hypothesis probability of0 . / .
007 for NGC 4395 and the rest of the sources, re-spectively. We fit the two subsets with a linear fit in the log-log space, of the form,log R = α + β log L − , (3) using the Orthogonal Distance Regression (ODR ) method.The best-fits result in consistent slopes β = − . ± . − . ± . π (Nandraet al. 2007). In other terms, R = R as a function of the ratio ( F ref / F PL ) ofreflection flux over the power-law flux in the 3 −
30 keV. Bothquantities are positively correlated with a slope of 1 . ± . σ . Fitting R , obtainedfrom Pexmon, vs. the reprocessed-to-power-law ratio, ob-tained from the MYTorus models, for the sources that couldbe fitted with both models, results in consistent slopes, al-beit with a di ff erent normalization. George & Fabian (1991)showed that F ref / F PL positively correlates with the equiva- https: // docs.scipy.org / doc / scipy / reference / odr.html hard look at local , optically - selected , obscured S eyfert galaxies α line (EW(Fe K α ); see their figure16). As a consequence, the anti-correlation seen between the R and L − can be translated in an anti-correlation betweenEW(Fe K α ) and L − , known as the Iwasawa-Taniguchi ef-fect (or the X-ray Baldwin e ff ect; Iwasawa & Taniguchi1993). This e ff ect has been confirmed in Type-2 AGN byRicci et al. (2014), and later confirmed in a large sample ofCT sources by Boorman et al. (2016). The fact that the slopeof the R − L − correlation for NGC 4395 is consistent withunity is suggestive of a constant reprocessed flux despite theintrinsic variability (see Kammoun et al. 2019b). This mayindicate that despite the change in the LOS column densityin this source, the overall covering fraction of the clouds re-mains constant. This also suggests that a large fraction of thereprocessed emission arises from distant material.Furthermore, Figure 2 shows a positive correlation be-tween R and N H , LOS with τ K = . R versus N H , LOS in thelog-log space, similar to eq. (3), for all sources, resulted ina slope of 0 . ± . curve fit , of the form,log R = ˆ α + ˆ β log L − + ˆ γ log N H , LOS , (4)excluding NGC 4395. We show the best-fit plane in Fig-ure 9d, with ˆ β = − . ± .
09 and ˆ γ = . ± .
15. Thisclearly demonstrates that the reprocessed emission dependssimultaneously on both the intrinsic X-ray luminosity and thecolumn density of the reprocessing material.Figure 10 shows the scattered and binned measurementsof N H , LOS as a function of log λ Edd . We plot the measure-ments for the full sample, the Cth subset, and the CT subset(top to bottom, respectively). We removed NGC 1068 fromthis plot due to the large discrepancy in its L − estimatesusing di ff erent models. Despite the fact that the size of oursample is small, our results broadly agree with the ones ofRicci et al. (2017) based on the 70-month Swift / BAT AGNsample. N H , LOS peaks in the range of log λ Edd ∼ [ − − . λ Edd in ourdata is shallower than the one found by Ricci et al. (2017).This trend depends on the way we estimated the Eddingtonfraction, i.e., spectral modeling and the bolometric correctionwe adopt. The similarity between our results and the onesfrom the BAT AGN sample are clearer for the fraction of thesources versus the Eddington ratio (right-hand side panels inthe same figure). This is suggestive that radiation pressure https: // docs.scipy.org / doc / scipy / reference / generated / scipy.optimize.curve fit.html regulates the distribution of the circumnuclear material (seee.g., figure 5 in Kurosawa & Proga 2009). As the accretionrate increases, less dense material (log N H / cm − <
24) isswept away, leaving only the CT material, thus decreasingthe torus covering factor (e.g., Fabian et al. 2006, 2009). Weshow in the N H − λ Edd plane the e ff ective Eddington limit as-suming the standard ISM grain abundance from Fabian et al.(2009). The e ff ective Eddington limit can be lower for dustygas than for ionized dust-free gas (e.g., Laor & Draine 1993;Scoville & Norman 1995). This means that a sub-EddingtonAGN may be seen as super-Eddington by the dusty toruswith substantial column densities. As a result, long-lived sta-ble clouds can survive radiation pressure only in a regimelower than the e ff ective Eddington limit (upper left part inFigure 10; see also Fabian et al. 2008). All our measure-ments reside in the long-lived cloud area.Within the context of the radiation-driven accretion diskwinds presented by Giustini & Proga (2019), most of oursources fall in the regime of small BH mass ( M BH < M (cid:12) )and moderate accretion rate. This category of sources is ex-pected to have weak / moderate line-driven disk winds. Theauthors show that failed continuum radiation-driven dustywind on radial scales of the order of the dust sublimationradius will be extended in the equatorial plane. Thus, at highinclination (which corresponds to the sources we study in thiswork), this failed wind will obscure the innermost regions ofthe system. Further investigations will be needed to confirmthis hypothesis.In Figure 11 we show Γ as a function of log λ Edd for Pex-mon, MYTD-1NH, and MYTD-2NH. A negative correla-tion between these two quantities can be seen. Applying theKendall tau’s correlation to the scattered data (for the Pex-mon model), we found a correlation coe ffi cient of 0.56 witha p − value of 1 . × − . Binning the data points leads toa clearer correlation. This is in agreement with the resultsof Winter et al. (2009) and Younes et al. (2011) (the latterstudied a sample of 13 LINERs). Our results di ff er from thepositive correlation found for high-redshift and more lumi-nous quasars (e.g., Shemmer et al. 2008; Risaliti et al. 2009;Brightman et al. 2013; Huang et al. 2020, and referencestherein). This is most likely due to a systematic deviationof lower luminosity AGN from the correlation. In fact, a “V”shape in the Γ vs. L X has been reported in various X-ray bi-naries by Liu et al. (2019) and Yan et al. (2020), indicatinga transition at a certain Eddington ratio between the “harderwhen brighter” and the “softer when brighter” states.Finally, in the upper panel of Figure 12, we show N H , eq ,obtained by fitting the data with MYTD-2NH, plottedversus the corresponding N H , LOS values. Five sources(namely, NGC 4388, NGC 7682, Mrk 334, NGC 5929,and NGC 5674) out of the 13 fitted with this model showa Cth N H , eq . For NGC 5674 and Mrk 334, N H , eq is higher6 K ammoun et al . . . . . Total N H , L O S ( c m ) . . . . . f r a c t i on Compton thin log
Edd PexmonMYTD-2NHMYTD-1NH log
Edd . . Compton thick
Figure 10.
Left: N H , LOS as a function of log λ Edd for the full sample,the Cth observations, and the CT observations(top to bottom). Theresults are shown for Pexmon (blue circles), MYTD-1NH (red dia-monds), and MYTD-2NH (black squares). The shaded data pointscorrespond to the individual measurements. The large thick sym-bols correspond to the binned data. The green dashed lines showthe e ff ective Eddington limit below which dusty clouds (with stan-dard ISM grain abundance; adapted from Fabian et al. 2009) closeto the BH see the AGN as being e ff ectively above the Eddingtonlimit (known as the forbidden region, grey area). Long-lived ab-sorbing clouds can only occur above the dashed line. Right: Thecorresponding fraction of measurements in the various log λ Edd bins(for the Pexmon model only), normalized to the total number ofobservations. than 10 cm − . However, for the other three sources N H , eq is significantly lower though consistent with N H , LOS . Thismay be an indication of a uniform distribution of absorb-ing / reprocessing material in these two sources. Interestingly,none of the sources requires a Cth N H , eq while its N H , LOS isCT. In addition, none of the sources require N H , LOS to besignificantly higher than N H , eq . This result adds to the rig-orousness of our analysis. It supports the general scheme inwhich most of the sources are surrounded by Compton-thickmaterial that reprocesses the intrinsic X-ray light into ourLOS (see e.g., figure 1 in Marin 2016, for a recent schematicrepresentation). The alignment of this material with our LOSwill result in observed column densities that are smaller orequal to the global one, being mostly > cm − for thesources that are observed at high inclination. In a toroidal ge-ometry, for a given inclination angle, the ratio N H , LOS / N H , eq may be indicator of the opening angle (in other terms the log Edd . . . . . . PexmonMYTD-2NHMYTD-1NH
Figure 11. Γ as a function of log λ Edd . The results are shown for thePexmon (blue circles), MYTD-1NH (red diamonds), and MYTD-2NH (black squares). The shaded data points correspond to the in-dividual measurements. The large thick symbols correspond to thebinned data. covering fraction; see eq. 1). In the lower panel of Figure 12,we plot the ratio c / a (indicating the half opening angle) as afunction of N H , LOS / N H , eq following eq. 1 for a range of incli-nations between 60 ◦ and 85 ◦ . This plot indicates that largevalues of N H , LOS / N H , eq (close to 1) indicate small opening an-gles (i.e., large covering fractions), while small N H , LOS / N H , eq values indicate a wider range of opening angles, depend-ing on the inclination. Alternatively, some recent modelsthat account for di ff erent geometries (e.g., Buchner et al.2015; Balokovi´c et al. 2018; Tanimoto et al. 2019) can beused to derive direct constraints on the torus properties (e..g,equatorial column density, covering fraction, clumpiness).However, given the fact that the rather simple models thatwe used in this paper give reasonable fits, and given the rel-atively modest data quality in several cases, the applicationof additional models, with more free parameters, is not bejustified by the data. Thus, this approach will result in manyunconstrained parameters. In addition, the consistency be-tween the various parameters that we obtained by applyingthe four Pexmon and MYTorus models strongly suggeststhat the application of an additional model will not alter anyof the main results and conclusions of our work (see e.g.,Kammoun et al. 2019a, for a comparison between variousmodels). CONCLUSIONS hard look at local , optically - selected , obscured S eyfert galaxies N H, LOS (10 cm ) . . . N H , e q ( c m ) NGC 5674NGC 5929 Mrk 334 NGC 7682NGC 4388 N H, LOS / N H, eq . . . . . . c / a Figure 12.
Top: The torus (equivalent) column density plotted ver-sus the LOS column density obtained by fitting the spectra withMYTD-2NH. The green dotted lines correspond to the CT limit.The red solid line corresponds to the identity line. Bottom: The ra-tio c / a (indicating the torus opening angle) versus the column den-sity ratio N H , LOS / N H , eq , for inclinations between 60 ◦ and 85 ◦ (frombottom to top, with an increase by 5 ◦ ). We present analysis of optical and X-ray spectra of 21 ob-scured sources in the CfASyS through the
NuSTAR ObscuredSeyferts
Legacy Survey. This is the first optically-selectedand volume-limited survey of AGN performed in the hardX-rays. Our results are summarized below. • Analyzing the optical spectra of those sources allowedus to obtain accurate estimates of their BH masses.Furthermore, we identified two sources in the com-posite regime of the BPT diagram (Mrk 461 andNGC 5256). Those two sources were excluded fromthe analysis, restricting the final sample to 19 Seyfertgalaxies only. • We have fitted the X-ray spectra of the 19 sources usingfour models. The results from all the models are con-sistent with each other, except for intrinsic luminosity in the cases of CT sources. In these sources, the intrin-sic power law emission is heavily suppressed, and thuscould not be well constrained. Di ff erent models thenresult in di ff erent values of the intrinsic luminosity dueto various physical assumptions of each model. • Our results indicate that 82 −
89% of the sources areheavily obscured with N H > cm − . In addition,36 . − .
6% of the sources in our sample are CT. Ourresults are in agreement with the ones found by Risal-iti et al. (1999) based on multi-wavelength estimatesof N H for obscured sources in the local universe, rec-onciling the long-lasting discrepancy between volume-limited values and those observed from flux-limitedhard X-ray surveys. • We found a tight anti-correlation between the reflec-tion fraction (representing the ratio of reflected overintrinsic flux) and the intrinsic X-ray luminosity, inagreement with the Iwasawa-Taniguchi e ff ect. Fur-thermore, our results showed a positive correlation be-tween the reprocessed emission and the LOS columndensity. • Our results support the hypothesis that radiation pres-sure regulates the distribution of the circumnuclearmaterial.Our results demonstrate the unique power of
NuSTAR inunveiling the properties of obscured AGN thanks to its highsensitivity above 10 keV. Deeper observations in the soft andhard X-rays are required, especially for the sources that couldnot be detected with
NuSTAR in order to confirm our con-clusions, and to explore possible intrinsic and / or absorption-induced variability. We note that future missions carryingmicro-calorimeters such as XRISM / Resolve (Tashiro et al.2018) and
Athena / X-IFU (Barret et al. 2018) will allow usto investigate in more detail the nature, composition, and thedynamic properties of the obscuring material in AGN. Cou-pling the high resolution of
XRISM in the soft X-rays to thehigh sensitivity in hard X-rays of
NuSTAR through simulta-neous observations will be of a particular interest. Exploringthe role of future missions in studying obscured AGN will beaddressed in future work.We thank the anonymous reviewer for their feedback thatimproved the quality of our manuscript. KO acknowledgessupport from the National Research Foundation of Korea(NRF-2020R1C1C1005462). This work made use of datafrom the
NuSTAR mission, a project led by the California In-stitute of Technology, managed by the Jet Propulsion Labora-tory, and funded by the National Aeronautics and Space Ad-ministration. This work also made use
XMM-Newton , an ESA8 K ammoun et al .science mission with instruments and contributions directlyfunded by ESA Member States and NASA. Results presentedin this paper are also based on data obtained with the
Suzaku observatory; and the
Chandra
X-ray Observatory. We ac-knowledge the use of public data from the
Swift data archive.ESK and JMM thank Tahir Yaqoob for useful discussions re-garding the implementation of MYTorus. We would like tothank Karl Forster for scheduling all of the
NuSTAR observa-tions in our program. We would like to thank Xavier Barcons,Fiona Harrison, Tim Kallman, Kirpal Nandra and John Ray-mond for useful conversations. E.B. is supported by a Centerof Excellence of THE ISF (grant No. 2752 / PYTHON li- brary for publication of quality graphics. The MCMC resultswere presented using the GetDist
PYTHON package.
Facilities:
Chandra , NuSTAR , Swift , Suzaku , XMM-Newton
Software: pPXF (Cappellari & Emsellem 2004), CIAO(v9.4 Fruscione et al. 2006), HEASoft (Nasa High EnergyAstrophysics Science Archive Research Center (Heasarc)2014), NUSTARDAS (v1.8.0, https: // heasarc.gsfc.nasa.gov / docs / nustar / analysis / , SAS (v17.0.0 Gabriel et al. 2004),XSPEC (Arnaud 1996), XSPEC EMCEE ((https: // github.com / zoghbi-a / xspec emcee), Matplotlib (Hunter 2007), Get-Dist (https: // getdist.readthedocs.io / en / latest / )APPENDIX A. NOTES ON INDIVIDUAL SOURCESIn this section we present details on the spectral analysis of each of the sources studied in this paper. The spectra fitted with the‘Pexmon’ model are shown in Figures 13-14. The best-fit parameters obtained from the various models are shown in Tables 4-7.The Γ − N H confidence contours for each model are shown in Figures 15-18. Mrk 573:
We fitted the
XMM-Newton
EPIC-pn and the
NuSTAR
FPMA / FPMB spectra of this source using the four models thatare described in Section 5. The source does not show any variability, thus we kept all the parameters tied for both observations.All models result in a CT column density in the LOS, in agreement with the results by Zhao et al. (2020). We note that MYTCdoes not provide a good fit using the standard configuration, i.e., by fixing the weight A to unity. Instead, it requires a large valueof A = + − . The soft X-rays are fitted using two Apec components with temperatures of 0.17 and 0.88 keV.
NGC 1144:
We fitted the
XMM-Newton
EPIC-pn and the
NuSTAR
FPMA / FPMB spectra of this source using the four modelsthat are described in Section 5. The source shows variability. Hence, we fitted both observations assuming the same power lawslope but we let the power law normalization, the column densities, and the reflection fraction (in the case of Pexmon) free tovary. The fits indicate changes in the intrinsic power law luminosity and the absorbing N H , confirmed by all models. The columndensity increases from ∼ × cm − in the XMM-Newton observation (consistent with the results found by Winter et al. 2008,2009) to ∼ cm − in the NuSTAR observation, at the lower limit of CT. The soft X-rays are fitted using two
Apec componentswith temperatures of 0.33 and 1.46 keV.
NGC 3362:
The source was not detected by
NuSTAR . However, it was detected by
XMM-Newton below 4 keV. Thus we couldnot use neither the Pexmon nor the MYTorus configurations. We fitted the
XMM-Newton spectrum assuming an absorbed powerlaw plus a scattered component and two
Apec components. We were not able to constrain the power law photon index, hence wefixed it to 1.8 and 2.2. Both values result in consistent results for the absorbing column density being CT. We adopt the valueobtained for
Γ = .
8, being N H = (5 ± × cm − . We also fixed the fraction of the scattered component to its best-fit valueof C sc = .
08. The temperatures of the
Apec components are 0.05 and 0.79 keV.
UGC 6100:
The source was observed but not detected by
XMM-Newton . However, it was detected by
NuSTAR . We fittedthe FPMA / FPMB spectra using an absorbed power law plus Pexmon, with the reflection fraction fixed at unity. The power lawphoton index was not constrained, so we fixed it to 1.8 and 2.2. Both values result in good fits and consistent results for theabsorbing column density being very high. We adopt the value obtained for
Γ = .
8, being N H = . + . − . × cm − . We alsoapplied di ff erent binning schemes to test the validity of our results, and all resulted in consistent values of N H . Due to the lowquality of the spectra, MYTorus models result in no constraints. NGC 3982:
We fitted the
XMM-Newton
EPIC-pn and the
NuSTAR
FPMA / FPMB spectra of this source using the four modelsthat are described in Section 5. We tied all parameters between the spectra of the two observatories. The Pexmon and MYTCmodels suggest a Cth source with N H = × cm − . However, MYTD-1NH and MYTD-1NH suggest a CT column densitywith N H = (5 . ± . × cm − and (4 . ± . × cm − , respectively. We note that Panessa et al. (2006) found that thesource is a CT candidate based on the F X / F [OIII] vs F [OIII] / F IR diagnostic. The soft X-rays are modeled using a single Apec component with kT = .
36 keV.
NGC 4388:
The source has been observed three times with
XMM-Newton and one time with
NuSTAR . We fit the spectrafrom the four observations above 3 keV in order to avoid the complexity of modeling the soft X-rays in this source (see e.g., hard look at local , optically - selected , obscured S eyfert galaxies C sc to zero in order to avoid overestimating the soft emission. Wekept Γ tied between the four observations and let the power law normalization and N H vary. For MYTC the fit requires A (cid:44) R ( A ) free to vary between the di ff erent observations, for the Pexmon (MYTC) model. All four models indicate a Cth( N H ∼ . − . × cm − ) source, in agreement with the results found by Masini et al. (2016) and Miller et al. (2019). Ourbest fits show hints of a change in N H by ∼ . × cm − . In addition, our modeling suggests variability in both intrinsic andreprocessed emission. The XMM-Newton observation with the highest flux indicates the presence of absorption lines, which wemodeled using three Gaussian lines in absorption. The best-fit suggests absorption lines at 6 . + . − . keV, 6 . ± .
03 keV, and7 . + . − . keV. UGC 8621:
The source was not detected by
NuSTAR . However, it was observed and detected by
XMM-Newton below ∼ Apec . We werenot able to get any constrains on the photon index so we fixed it to 1.8. The scattered fraction could not be constrained either, sowe fixed it to C sc = . N H = . + . − . × cm − . NGC 5252:
We analyzed the
XMM-Newton and
NuSTAR spectra of the source. The source showed intrinsic and absorptionvariability, requiring both ionized and neutral partial covering absorption. The best-fit model, in XSPEC parlance, can be writtenas: zpcfabs × zxipcf × zcuto ff pl + pexmon + Apec + Apec. For the
NuSTAR observation, due to the lack of good-qualitysimultaneous soft X-ray observations, we cannot constrain the parameters of both absorption components. Hence, we removedthe ionized absorption and linked the neutral N H to the one of XMM-Newton , and let its covering fraction vary freely. The best-fit results for the neutral absorption are N H = . + . − . × cm − and (4 . ± . × cm − , with a covering fractionof 0 . + . − . (0 . ± .
05) for the
XMM-Newton ( NuSTAR ) observation. As for the ionized absorption, the best-fit results are N H = . + . − . × cm − , log (cid:16) ξ/ erg cm s − (cid:17) = − . + . − . , f cov = . ± .
01. We tied the photon indices between the twoobservations. We find a best-fit value of
Γ = . + . − . . Our results are in contradiction with the ones by Dadina et al. (2010) whoanalyzed the same XMM-Newton observation presented in the current work, in addition to a
Chandra observation of the source.The authors modeled the spectra of this source with an extremely flat photon index of
Γ = . + . − . . Such a flat photon index canbe ruled out by our data, thanks to the NuSTAR data.
NGC 5347:
We fit the spectra obtained by
Chandra , Suzaku , and
NuSTAR using Pexmon, MYTC, and MYTD-1NH. We cannotget useful constraints on N H , eq by letting it be free (i.e., MYTD-2NH model). A detailed analysis of this source is presented inKammoun et al. (2019a), where we confirm the CT nature of the source. The only di ff erence between the current work andKammoun et al. (2019a) is that now we tie the Pexmon normalization to the one of the power law component and we let R free.This does not alter any of the previous results. NGC 5695:
We fit the spectra obtained by
XMM-Newton and
NuSTAR . The
XMM-Newton spectra are background-dominatedabove 2 keV. We modeled the spectra assuming no variability. The soft X-rays are modeled invoking a single
Apec component.The results from the four models confirm a CT LOS column density in this source. This source was identified as a CT candidatebased on its
XMM-Newton by LaMassa et al. (2009).
NGC 5929:
We fit the
Suzaku and
NuSTAR
FPMA / FPMB spectra for this source. All four models used in this analysisconfirm a Cth column density during this observation. We note that variability in the intrinsic flux can be seen between the twoobservations. Letting the column density vary between the observations resulted in consistent results.
NGC 7674:
We fit the non-simultaneous spectra obtained by
XMM-Newton and
NuSTAR . We allow for both intrinsic andabsorption-induced variability. We also allow the fraction of scattered (or unabsorbed) power law to vary between the twoobservations. The spectra showed an excess at ∼ XXVI . We modeled it by adding aGaussian emission line. The soft X-rays are modeled using two
Apec components. The results from Pexmon and MYTD-2NHsuggest that the LOS column density is consistent and Cth in both observations. However, the MYTD-1NH model suggeststhat the column density in the
XMM-Newton observation is Cth while the
NuSTAR one is CT. We note that the best-fit MYTD-1NH returns χ / dof = /
156 while the ones by Pexmon and MYTD-2NH return a better fit with χ / dof = /
155 and159 / N H , LOS = . + − , . + . − . × cm − using Pexmon and MYTD-2NH, respectively)are in agreement with the ones by LaMassa et al. (2011) who also analyzed the XMM-Newton observations of this source.Tanimoto et al. (2020) fitted the
Suzaku and
NuSTAR spectra of NGC 7674 using XCLUMPU (Tanimoto et al. 2019) and found N H , LOS = . + . − . × cm − , which is consistent with our results. In contrast, Gandhi et al. (2017) analyzed the NuSTAR , Suzaku , and
Swift / XRT spectra of this source and claimed the presence of a CT column density, which we did not recover fromour analysis. We also note that fitting the spectra using MYTC required the weights of the reprocessed components to be di ff erentthan unity (standard configuration) with A = . + . − . .0 K ammoun et al . NGC 7682:
We fit the
XMM-Newton and
NuSTAR spectra of this source, and found no indication of variability. The source isdetected out to ∼ XMM-Newton observation. We fit the spectra using all four models. All models result in consistentand Cth column densities in our LOS. Both FPMA and FPMB spectra show an excess at ∼ . ± . Apec componentwith kT = .
64 keV.
NGC 4395:
We fit the flux-resolved spectra obtained during two
XMM-Newton observations and a
NuSTAR observation. Ourmodel consists of a partial covering neutral absorber and two ionized partial covering absorbers. A detailed analysis is presentedin Kammoun et al. (2019b). The only di ff erence between the current work and Kammoun et al. (2019b) is that now we tie thePexmon normalization to the one of the power law component and we let R be free. This does not alter any of the previousresults. Due to the complexity of the model invoking partial covering and ionized absorption, we were not able to fit the spectrawith MYTorus. Mrk 334:
We fit the
Swift / XRT and
NuSTAR
FPMA / FPMB spectra of this source, assuming an absorbed power law plusreprocessed emission and an
Apec component. The source shows flux variability. Due to the poor quality of the XRT spectrumwe are not able to assess whether the variability is intrinsic or due to a change in absorption. Leaving N H free to vary resultsin unconstrained results. Thus, we assumed a change in the intrinsic power law flux. The best-fit results indicate a change by afactor of ∼ . NGC 5283:
We fit the
Chandra and
NuSTAR spectra of this source. The spectra are fitted assuming an absorbed power lawplus scattered component and reprocessed emission. All four models were applied and suggest consistent and Cth absorption inour LOS. We note that variability in the intrinsic flux can be seen between the two observations. Letting the column density tovary between the observations resulted in consistent results.
UM 146:
The source was detected with XRT only above ∼ NuSTAR
FPMA / FPMB spectra only, assuming an absorbed power law plus reprocessed emission. All four models suggest a Cth columndensity in our LOS.
NGC 5674:
We fit the
Swift / XRT and
NuSTAR spectra of this source. The XRT spectrum was fitted above 1 keV. The spectraare fitted assuming an absorbed power law plus scattered component and a reprocessed emission. All four models were appliedand suggested consistent and Cth absorption in our LOS.
NGC 1068:
Due to the complexity of the spectra of this source in the soft X-rays (Kallman et al. 2014; Bauer et al. 2015) wefit the FPMA / FPMB spectra extracted from only one
NuSTAR observation. The source is known to be consistently highly CT.Detailed analysis of the multi-epoch, X-ray spectra of this source are presented in Bauer et al. (2015) and Zaino et al. (2020).We modeled the spectra using an absorbed power law plus a high-temperature
Apec component and reprocessed emission. Weapplied the Pexmon and MYTC models only. The two models consistently confirm the CT nature of the source. We found in bothmodels an excess in the 6 − NuSTAR spectraof this source. The authors model this feature with a Gaussian line fixed at 6 keV, attributing it to instrumental calibrations.Investigating the origin of this feature is beyond the scope of the current paper. It is also worth noting that the Pexmon modelimplies large LOS column densities that cannot be achieved by MYTorus (having an upper limit at 10 cm − . This resulted indiscrepancy in the intrinsic luminosity that are implied by our best-fits being 4 . × erg s − and 2 . × erg s − for Pexmonand MYTC, respectively. The L − value inferred by Pexmon is in agreement with the results by Marinucci et al. (2016) andZaino et al. (2020), favoring a high luminosity (a few times 10 erg s − ) by adding the constraints from the mid- IR and [O III ]observations of this source. hard look at local , optically - selected , obscured S eyfert galaxies C o un t s c m k e V Mrk 573
Energy (keV) () C o un t s c m k e V NGC 1144
Energy (keV) () C o un t s c m k e V NGC 3362 Energy (keV) () C o un t s c m k e V UGC 6100 Energy (keV) () C o un t s c m k e V NGC 3982
Energy (keV) () C o un t s c m k e V NGC 4388 Energy (keV) () C o un t s c m k e V UGC 8621
Energy (keV) () C o un t s c m k e V NGC 5252
Energy (keV) () C o un t s c m k e V NGC 5347
Energy (keV) () Figure 13.
The
NuSTAR (red / blue for FPMA / FPMB),
XMM-Newton (black),
Swift / XRT (green),
Chandra (cyan), and
Suzaku (orange) spectraof the 19 Seyfert galaxies studies in this work. The best-fit “pexmon” model are shown in grey. We also show the di ff erent spectral components:primary power law (dashed lines), scattered power law (solid thin line), reflection (dash-dotted lines), ‘Apec’ and Gaussian lines (dotted lines).The lower panels show the corresponding residuals defined as ∆ χ = (data − model) /σ . ammoun et al . C o un t s c m k e V NGC 5695
Energy (keV) () C o un t s c m k e V NGC 5929
Energy (keV) () C o un t s c m k e V NGC 7674
Energy (keV) () C o un t s c m k e V NGC 7682
Energy (keV) () C o un t s c m k e V NGC 4395
Energy (keV) () C o un t s c m k e V Mrk 334
Energy (keV) () C o un t s c m k e V NGC 5283
Energy (keV) () C o un t s c m k e V UM 146
Energy (keV) () C o un t s c m k e V NGC 5674
Energy (keV) () C o un t s c m k e V NGC 1068
Energy (keV) () Figure 14.
Same as Figure 13. We note that the gap seen in the ∼ − hard look at local , optically - selected , obscured S eyfert galaxies − N H , LOS (10 cm − )1 . . . . . . Γ PexmonMrk573 10 − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC1144 obs . . − N H , eq (10 cm − )1 . . . . . . Γ MYTC obs . . − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − obs . . − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS , obs . N H , LOS , obs . − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC3982 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − − N H , LOS (10 cm − )1 . . . . . Γ PexmonNGC4388 obs . . . . − − N H , eq (10 cm − )1 . . . . . Γ MYTC obs . . . . − − N H , LOS (10 cm − )1 . . . . . Γ MYTD − obs . . . . − − N H (10 cm − )1 . . . . . Γ MYTD − N H , eq N H , LOS , obs . N H , LOS , obs . N H , LOS , obs . N H , LOS , obs . − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC5347 10 − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − Figure 15. Γ − N H confidence contour for the di ff erent models used in this work obtained from the MCMC analysis. For the sources where Γ was fixed (see text), we show only the 1D probability density of N H . ammoun et al . − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC5695 10 − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC5929 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC7674 obs . . − − N H , eq (10 cm − )1 . . . . . . Γ MYTC obs . . − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − obs . . − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS , obs . N H , LOS , obs . − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC7682 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonMrk334 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS
Figure 16.
Same as Figure 15. hard look at local , optically - selected , obscured S eyfert galaxies − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC5283 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonUM146 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS − − N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC5674 10 − − N H , eq (10 cm − )1 . . . . . . Γ MYTC 10 − − N H , LOS (10 cm − )1 . . . . . . Γ MYTD − − − N H (10 cm − )1 . . . . . . Γ MYTD − N H , eq N H , LOS
300 600 900 N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC1068 N H , eq (10 cm − )1 . . . . . . Γ MYTC
Figure 17.
Same as Figure 15. ammoun et al . . . . . . . N H , LOS (10 cm − )1 . . . . . . Γ PexmonNGC5252 10 − − N H , LOS (10 cm − )1 . . . . . Γ PexmonNGC4395 obs . . . − − N H , LOS (10 cm − )0 . . . . . . P D F NGC336210 − N H , LOS (10 cm − )0 . . . . . . P D F Pexmon UGC6100 10 − N H , LOS (10 cm − )0 . . . . . . P D F phabsUGC8621 Figure 18.
Same as Figure 15 but for the sources where we could not apply the MYTorus models. hard look at local , optically - selected , obscured S eyfert galaxies Table 3 . Log of the observations that we analyzed. The last column indicates the statistics used for thespectral fits.Source Instrument ObsID Obs date Count rate ∗ exp. time Statistics(count ks − ) (ks)Mrk 573 XMM-Newton ± χ NuSTAR / FPMA 60360004002 2018-01-06 9 . ± . NuSTAR / FPMB 60360004002 8 . ± . XMM-Newton ± χ NuSTAR / FPMA 60368001002 2017-10-14 48 ± NuSTAR / FPMB 60368001002 44 ± XMM-Newton ± C -stat NuSTAR N / D
32 .0UGC 6100
XMM-Newton N / D C -stat NuSTAR / FPMA 60465004002 2019-05-20 2 . ± . NuSTAR / FPMB 60465004002 2 . ± . XMM-Newton ± C -stat NuSTAR / FPMA 60375001002 2017-12-06 4 . ± . NuSTAR / FPMB 60375001002 3 . ± . XMM-Newton ±
10 3.9 χ XMM-Newton ±
10 7.3
XMM-Newton ± NuSTAR / FPMA 60061228002 2013-12-27 320 ± NuSTAR / FPMB 60061228002 310 ± XMM-Newton ± C -stat NuSTAR N / D XMM-Newton ± χ NuSTAR / FPMA 60061245002 2013-05-11 412 ± NuSTAR / FPMB 60061245002 384 ± Suzaku . ± . C -stat Chandra . ± . NuSTAR / FPMA 60001163002 2015-01-16 7 . ± .
47 46.6
NuSTAR / FPMB 60001163002 8 . ± .
49 46.6NGC 5695
XMM-Newton . ± . C -stat NuSTAR / FPMA 60368004002 2018-01-16 9 . ± . NuSTAR / FPMB 60368004002 7 . ± . Suzaku ± χ NuSTAR / FPMA 60465009002 2018-09-17 45 . ± . NuSTAR / FPMB 60465009002 39 . ± . XMM-Newton ± χ NuSTAR / FPMA 60001151002 2014-09-30 28 . ± . NuSTAR / FPMB 60001151002 28 . ± . Table 3 continued ammoun et al . Table 3 (continued)
Source Instrument ObsID Obs date Count rate ∗ exp. time Statistics(count ks − ) (ks)NGC 7682 XMM-Newton . ± χ NuSTAR / FPMA 60368002002 2017-10-06 12 . ± . NuSTAR / FPMB 60368002002 9 . ± . † XMM-Newton χ XMM-Newton
XMM-Newton
NuSTAR
Swift / XRT 882529 2019-05-21 5 . ± . χ NuSTAR / FPMA 60465001002 2019-05-21 59 . ± . NuSTAR / FPMB 60465001002 58 . ± . Chandra . ± . χ NuSTAR / FPMA 60465006002 2018-11-17 86 . ± . NuSTAR / FPMB 60465006002 83 . ± . NuSTAR / FPMA 60465002002 2019-02-05 36 . ± . χ NuSTAR / FPMB 60465002002 35 . ± . Swift / XRT 80672 2014-07-10 56 . ± χ NuSTAR / FPMA 60061337002 2014-07-10 211 . ± . NuSTAR / FPMB 60061337002 201 . ± . NuSTAR / FPMA 60002030002 2012-12-18 214 ± χ NuSTAR / FPMB 60002030002 197 ± ote — ∗ : count rates are reported in the full bands analyzed for each instrument. † : we analyze the flux-resolved spectra of this source. The corresponding count rates are reported in Kammoun et al. (2019b). Table 4 . The best-fit parameters obtained by fitting the ‘Pexmon’ model.Source N H , LOS Γ Norm PL C sc R kT Norm
Apec , kT Norm
Apec , L − Mrk 573 2 . + . − . . ± .
13 3 . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . ± . . ± .
04 1 . ± .
09 41 . + . − . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . t . + . − . t . + . − . t t t t ± . f . ± .
16 0 . f − . − . − . . ± . ± .
04 0 . ± . + − . f . ± . − f − − − − . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . ± .. − − . + . − . . ± .
03 36 . + . − . − . + . − . − − − − . ± . t . + . − . − . + . − . − − − − . ± . t + − − . + . − . − − − − . ± . t . + . − . − . + . − . − − − − . + . − . . f . + . − . . f − . + . − . . + . − . − − Table 4 continued hard look at local , optically - selected , obscured S eyfert galaxies Table 4 (continued)
Source N H , LOS Γ Norm PL C sc R kT Norm
Apec , kT Norm
Apec , L − NGC 5252 0 . + . − . . + . − . . + . − . − . ± .
04 0 . + . − . . ± .
09 0 . + . − . . + . − . t t . + . − . − t t t t t . + . − . . + . − . . + . − . . + . − . . + . − . . ± .
07 0 . ± . − − . + . − . . + . − . . + . − . . + . − . . + . − . . − . − . . + − − − . ± .
01 1 . ± .
06 9 . + . − . . + . − . . + . − . . + . − . . + . − . − − t t . + . − . t t t t − − . + . − . . ± .
07 3 . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . t . + . − . . + . − . . + . − . t t t t . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . − − . ± .
003 1 . + . − . . + . − . − . + . − . . + . − . . + . − . . ± .
03 1 . ± . t t . + . − . − . + . − . t t t t t t . ± . − . + . − . t t t t . ± . t . + . − . − . + . − . t t t t t t . ± . − . + . − . t t t t . + . − . . ± .
02 11 . + . − . − . + . − . t t t t t t . ± . − . + . − . t t t t t t . + . − . − . + . − . t t t t . ± .
02 1 . ± .
08 6 . + . − . − . + . − . . + . − . . + . − . − − t t . + . − − t t t − − . ± .
09 1 . ± .
05 13 . + . − . . + . − . . + . − . . + . − . . + . − . − − t t . + . − . t t t t t t . + . − . . ± .
12 6 . + . − . − . + . − . − − − − . + . − . . ± .
05 39 . + . − . . + . − . . + . − . − − − − + − . + . − . + − − . ± .
01 7 . ± .
41 20 . ± . − − ote — t : tied. f : fixed. The column density is in units of 10 cm − . Norm PL / Norm
Apec , and Norm Apec , are in units of 10 − and10 − photon s − keV − cm − . The luminosity is in unit of 10 erg s − . Table 5 . The best-fit parameters obtained by fitting the ‘MYTC’ model. We used eq. 1 to estimate the N H , LOS values.Source N H , eq θ N H , LOS Γ Norm PL C sc A L − Mrk 573 3 . + . − . . + . − . . + . − . . + . − . . + . − . + − . + . − . . + . − . . + . − . + − . + . − . × − f . + . − . t t + − t f . + . − . . + . − . . + . − . . + . − . . + . − . f . + . − . . + . − . . ± .
03 35 . + . − . − . + . − . . + . − . t t . + . − − . + . − . . + . − . t t + − − . + . − . . + . − . t t . + . − . − . + . − . Table 5 continued ammoun et al . Table 5 (continued)
Source N H , eq θ N H , LOS Γ Norm PL C sc A L − NGC 5347 7 . + . − . . + . − . . ± . + − . ± .
001 1 f . + . − . . + . − . . + . − . . + . − . + . − . f . + . − . . + . − . . + . − . . + − . . + . − . f t t t t . + . − . t t . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . . + . − . t t . + . − . . + . − . t . + . − . . + . − . . + . − . . + . − . . + . − . f . + . − . . − . − . . + . − . . + . − . − f t t t t . + . − . − f . + . − . . + . − . . ± .
01 15 . + . − . . + . − . f t t t t . ± . t t . + . − . . + . − . . ± .
23 5 . + . − . − f . + . − . . + . − . . ± .
04 37 . + . − . . + . − . f . + . − . . + . − . . ± .
07 172 + − − . + . − . ote — t : tied. f : fixed. The column density is in unit of 10 cm − . Norm PL is in units of 10 − photon s − keV − cm − .The luminosity is in units of 10 erg s − . Table 6 . The best-fit parameters obtained by fitting the ‘MYTD-1NH’ model.Source N H , LOS Γ Norm PL C sc A A L − Mrk 573 2 . + . − . . + . − . . + . − . . + . − . + − A . ± .
04 1 . + . − . + − . + . − . . + . − . A . + . − . t . + . − . + . − . t t . ± . . + . − . . − . − . . + . − . + − A . ± .
03 1 . + . − . . + . − . − . + . − . A . ± . t . + . − − . + . − . A . ± . t + − − . ± . A . ± . t . + . − . − . + . − . A . + . − . . + . − . ±
13 0 . + . − . . + . − . f . + . − . . + . − . . + . − . + . − . . − . − . A . ± .
01 1 . ± .
06 8 . + . − . . + . − . f A t t . + . − . t t t . + . − . . + . − . . + . − . . + . − . . + . − . A . + . − . t . + . − . . + . − . t A . − . − . . + . − . . + . − . . + . − . f A . + . − . . + . − . . + . − . − f A t t . + . − . − t A . ± .
01 1 . ± .
04 9 . + . − . . + . − . . + . − . A Table 6 continued hard look at local , optically - selected , obscured S eyfert galaxies Table 6 (continued)
Source N H , LOS Γ Norm PL C sc A A L − t t . + . − . t t f . + . − . . + . − . . + . − . − . + . − . A . ± .
003 1 . ± .
03 28 . + . − . − . + . − . A ote — t : tied. f : fixed. The column density is in unit of 10 cm − . Norm PL is in units of10 − photon s − keV − cm − . The luminosity is in units of 10 erg s − . Table 7 . The best-fit parameters obtained by fitting the ‘MYTD-2NH’ model.Source N H , LOS N H , eq Γ Norm PL C sc A A L − Mrk 573 3 . ± . . + . − . . + . − . . + − . . + . − . + − A . ± .
04 6 . ± . . + . − . + − . + . − . × − . + . − . A . ± .
12 6 . ± . . + . − . + − t . + . − . A . ± . . ± . . + . − . . + . − . + . − . + − A . + . − . . + . − . . ± .
03 42 . + − − . + . − . A . + . − . t t + − − . + . − . A . ± . t t ± − . + . − . A . + . − . t t . + . − − . + . − . A . + . − . . + . − . . + . − . . + . − . + . − . . + . − . A . + . − . . + . − . . + . − . . + . − . . ± .
01 1 f A t t t . + . − . t t A . + . − . . + . − . . + . − . . + . − . + . − . . + . − . A . + . − . t t . + . − . . + . − . . + . − . A . + . − . . + . − . . + . − . . + . − . . + . − . . − . − . A . + . − . . ± .
04 1 . + . − . . + . − . − f A t t t . + . − . − f A . ± .
01 3 . + . − . . ± .
07 19 . + . − . . + . − . . + . − . A t t t . + . − . t t A . + . − . . + . − . . + . − . . + . − . − . + . − . A . + . − . . + . − . . + . − . . . − . . + . − . . + . − . A ote — t : tied. f : fixed. The column densities are in unit of 10 cm − . Norm PL is in units of 10 − photon s − keV − cm − .The luminosity is in units of 10 erg s − . REFERENCES
Akylas, A., Georgakakis, A., Georgantopoulos, I., Brightman, M.,& Nandra, K. 2012, A&A, 546, A98,doi: 10.1051 / / / / aafb77 Annuar, A., Alexander, D. M., Gandhi, P., et al. 2017, ApJ, 836,165, doi: 10.3847 / / / / / annurev.aa.31.090193.002353 ammoun et al . Antonucci, R. R. J., & Miller, J. S. 1985, ApJ, 297, 621,doi: 10.1086 / / / ab578eBalokovi´c, M., Comastri, A., Harrison, F. A., et al. 2014, ApJ, 794,111, doi: 10.1088 / / / / / / aaa7ebBarret, D., Lam Trong, T., den Herder, J.-W., et al. 2018, in Societyof Photo-Optical Instrumentation Engineers (SPIE) ConferenceSeries, Vol. 10699, Space Telescopes and Instrumentation 2018:Ultraviolet to Gamma Ray, 106991GBassani, L., Molina, M., Malizia, A., et al. 2006, ApJL, 636, L65,doi: 10.1086 / / / / / / mnras / stv2181Boorman, P. G., Gandhi, P., Alexander, D. M., et al. 2016, ApJ,833, 245, doi: 10.3847 / / / / / j.1365-2966.2011.18207.xBrightman, M., & Ueda, Y. 2012, MNRAS, 423, 702,doi: 10.1111 / j.1365-2966.2012.20908.xBrightman, M., Silverman, J. D., Mainieri, V., et al. 2013,MNRAS, 433, 2485, doi: 10.1093 / mnras / stt920Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000,doi: 10.1046 / j.1365-8711.2003.06897.xBuchner, J., & Bauer, F. E. 2017, MNRAS, 465, 4348,doi: 10.1093 / mnras / stw2955Buchner, J., Brightman, M., Nandra, K., Nikutta, R., & Bauer, F. E.2019, A&A, 629, A16, doi: 10.1051 / / / / / / / / / / / s11214-005-5097-2Cappellari, M., & Emsellem, E. 2004, PASP, 116, 138,doi: 10.1086 / / / / / / / / / / / / aa9115Draper, A. R., & Ballantyne, D. R. 2010, ApJL, 715, L99,doi: 10.1088 / / / / L99Fabian, A. C. 2012, ARA&A, 50, 455,doi: 10.1146 / annurev-astro-081811-125521Fabian, A. C., Celotti, A., & Erlund, M. C. 2006, MNRAS, 373,L16, doi: 10.1111 / j.1745-3933.2006.00234.xFabian, A. C., Vasudevan, R. V., & Gandhi, P. 2008, MNRAS, 385,L43, doi: 10.1111 / j.1745-3933.2008.00430.xFabian, A. C., Vasudevan, R. V., Mushotzky, R. F., Winter, L. M.,& Reynolds, C. S. 2009, MNRAS, 394, L89,doi: 10.1111 / j.1745-3933.2009.00617.xFanidakis, N., Baugh, C. M., Benson, A. J., et al. 2011, MNRAS,410, 53, doi: 10.1111 / j.1365-2966.2010.17427.xForeman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J.2013, PASP, 125, 306, doi: 10.1086 / / mnras / stx357Garcia-Rissmann, A., Vega, L. R., Asari, N. V., et al. 2005,MNRAS, 359, 765, doi: 10.1111 / j.1365-2966.2005.08957.xGehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611,1005, doi: 10.1086 / / mnras / / mnras / / / / / camcos.2010.5.65Goulding, A. D., Alexander, D. M., Mullaney, J. R., et al. 2011,MNRAS, 411, 1231, doi: 10.1111 / j.1365-2966.2010.17755.xGreene, J. E., & Ho, L. C. 2006, ApJ, 641, 117,doi: 10.1086 / hard look at local , optically - selected , obscured S eyfert galaxies Haardt, F., & Maraschi, L. 1993, ApJ, 413, 507,doi: 10.1086 / / / / / / / / j.1365-2966.2007.12157.xHuang, J., Luo, B., Hu, C., et al. 2020, ApJ, arXiv:2005.01749.https: // arxiv.org / abs / / / / MCSE.2007.55Iwasawa, K., & Taniguchi, Y. 1993, ApJL, 413, L15,doi: 10.1086 / ff e, W., Meisenheimer, K., R¨ottgering, H. J. A., et al. 2004,Nature, 429, 47, doi: 10.1038 / nature02531Kaastra, J. S. 2017, A&A, 605, A51,doi: 10.1051 / / / / / / / / ab1c5fKammoun, E. S., Nardini, E., Zoghbi, A., et al. 2019b, ApJ, 886,145, doi: 10.3847 / / ab5110Kau ff mann, G., Heckman, T. M., Tremonti, C., et al. 2003,MNRAS, 346, 1055, doi: 10.1111 / j.1365-2966.2003.07154.xKewley, L. J., Dopita, M. A., Sutherland, R. S., Heisler, C. A., &Trevena, J. 2001, ApJ, 556, 121, doi: 10.1086 / / annurev-astro-082708-101811Koss, M., Mushotzky, R., Veilleux, S., et al. 2011, ApJ, 739, 57,doi: 10.1088 / / / / / / aa8ec9Koss, M. J., Glidden, A., Balokovi´c, M., et al. 2016a, ApJL, 824,L4, doi: 10.3847 / / / / L4Koss, M. J., Assef, R., Balokovi´c, M., et al. 2016b, ApJ, 825, 85,doi: 10.3847 / / / / / pasj / / / / / / j.1365-2966.2009.15084.x LaMassa, S. M., Heckman, T. M., Ptak, A., et al. 2009, ApJ, 705,568, doi: 10.1088 / / / / / / / / // arxiv.org / abs / / / mnras / stz1622Liu, Y., & Li, X. 2014, ApJ, 787, 52,doi: 10.1088 / / / / / j.1365-2966.2012.21513.xMadsen, K. K., Grefenstette, B. W., Pike, S., et al. 2020, arXive-prints, arXiv:2005.00569. https: // arxiv.org / abs / / mnras / / j.1365-2966.2009.15330.xMarchesi, S., Ajello, M., Marcotulli, L., et al. 2018, ApJ, 854, 49,doi: 10.3847 / / aaa410Marchesi, S., Ajello, M., Zhao, X., et al. 2019, ApJ, 872, 8,doi: 10.3847 / / aafbebMarconi, A., Risaliti, G., Gilli, R., et al. 2004, MNRAS, 351, 169,doi: 10.1111 / j.1365-2966.2004.07765.xMarin, F. 2016, MNRAS, 460, 3679, doi: 10.1093 / mnras / stw1131Marinucci, A., Bianchi, S., Matt, G., et al. 2016, MNRAS, 456,L94, doi: 10.1093 / mnrasl / slv178Masini, A., Comastri, A., Balokovi´c, M., et al. 2016, A&A, 589,A59, doi: 10.1051 / / / / aaa83dMatt, G. 2002, MNRAS, 337, 147,doi: 10.1046 / j.1365-8711.2002.05890.xMighell, K. J. 1999, ApJ, 518, 380, doi: 10.1086 / / / ab3e05Mitsuda, K., Bautz, M., Inoue, H., et al. 2007, PASJ, 59, S1,doi: 10.1093 / pasj / / j.1365-2966.2009.15025.xNandra, K., O’Neill, P. M., George, I. M., & Reeves, J. N. 2007,MNRAS, 382, 194, doi: 10.1111 / j.1365-2966.2007.12331.xNardini, E., & Risaliti, G. 2011, MNRAS, 417, 2571,doi: 10.1111 / j.1365-2966.2011.19423.xNasa High Energy Astrophysics Science Archive Research Center(Heasarc). 2014, HEAsoft: Unified Release of FTOOLS andXANADU, Astrophysics Source Code Library.http: // ascl.net / ammoun et al . Netzer, H. 2015, ARA&A, 53, 365,doi: 10.1146 / annurev-astro-082214-122302Oh, K., Sarzi, M., Schawinski, K., & Yi, S. K. 2011, ApJS, 195,13, doi: 10.1088 / / / / / / / / / / s00159-017-0102-9Panagiotou, C., & Walter, R. 2019, A&A, 626, A40,doi: 10.1051 / / // arxiv.org / abs / / / / s41550-017-0232-zRicci, C., Ueda, Y., Koss, M. J., et al. 2015, ApJL, 815, L13,doi: 10.1088 / / / / L13Ricci, C., Ueda, Y., Paltani, S., et al. 2014, MNRAS, 441, 3622,doi: 10.1093 / mnras / stu735Ricci, C., Walter, R., Courvoisier, T. J. L., & Paltani, S. 2011,A&A, 532, A102, doi: 10.1051 / / / nature23906Risaliti, G. 2002, A&A, 386, 379,doi: 10.1051 / / / / / / / L6Salamanca, I., Alloin, D., Baribaud, T., et al. 1994, A&A, 282, 742S´anchez-Bl´azquez, P., Peletier, R. F., Jim´enez-Vicente, J., et al.2006, MNRAS, 371, 703,doi: 10.1111 / j.1365-2966.2006.10699.xSarzi, M., Falc´on-Barroso, J., Davies, R. L., et al. 2006, MNRAS,366, 1151, doi: 10.1111 / j.1365-2966.2005.09839.xSchawinski, K., Thomas, D., Sarzi, M., et al. 2007, MNRAS, 382,1415, doi: 10.1111 / j.1365-2966.2007.12487.xScoville, N., & Norman, C. 1995, ApJ, 451, 510,doi: 10.1086 / / / / / / / / mnras / / / / ab1b20—. 2020, arXiv e-prints, arXiv:2005.12927.https: // arxiv.org / abs / / / / / L238Treister, E., Urry, C. M., & Virani, S. 2009, ApJ, 696, 110,doi: 10.1088 / / / / / / / / / / / / / / / / / / / / / / / / / j.1365-2966.2008.14108.xVito, F., Brandt, W. N., Yang, G., et al. 2018, MNRAS, 473, 2378,doi: 10.1093 / mnras / stx2486Weisskopf, M. C., Tananbaum, H. D., Van Speybroeck, L. P., &O’Dell, S. L. 2000, in Proc. SPIE, Vol. 4012, X-Ray Optics,Instruments, and Missions III, ed. J. E. Truemper &B. Aschenbach, 2–16Winter, L. M., Mushotzky, R. F., Terashima, Y., & Ueda, Y. 2009,ApJ, 701, 1644, doi: 10.1088 / / / / / / s41550-019-0790-3Yan, Z., Xie, F.-G., & Zhang, W. 2020, ApJL, 889, L18,doi: 10.3847 / / ab665eYaqoob, T. 2012, MNRAS, 423, 3360,doi: 10.1111 / j.1365-2966.2012.21129.x hard look at local , optically - selected , obscured S eyfert galaxies Younes, G., Porquet, D., Sabra, B., & Reeves, J. N. 2011, A&A,530, A149, doi: 10.1051 / / / mnras / staa107 Zdziarski, A. A., Lubi´nski, P., Gilfanov, M., & Revnivtsev, M.2003, MNRAS, 342, 355,doi: 10.1046 / j.1365-8711.2003.06556.xZdziarski, A. A., Lubi´nski, P., & Smith, D. A. 1999, MNRAS, 303,L11, doi: 10.1046 / j.1365-8711.1999.02343.xZhao, X., Marchesi, S., Ajello, M., balokovic, M., & Fischer, T.2020, arXiv e-prints, arXiv:2003.12190.https: // arxiv.org / abs //