X-ray spectral properties of seven heavily obscured Seyfert 2 galaxies

0.87 MB·

by  Stefano Marchesi

Accepted for publication in the Astrophysical Journal on January 14, 2017
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
X-RAY SPECTRAL PROPERTIES OF SEVEN HEAVILY OBSCURED SEYFERT 2 GALAXIES
S. Marchesi1, M. Ajello1, A. Comastri2, G. Cusumano3, V. La Parola3, A. Segreto3
Accepted for publication in the Astrophysical Journal on January 14, 2017
ABSTRACT
We present the combined Chandra and Swift-BAT spectral analysis of seven Seyfert 2 galaxies
selected from the Swift-BAT 100-month catalog. We selected nearby (z 0.03) sources lacking of a
ROSAT counterpart and never previously observed with Chandra in th≤e 0.3–10keV energy range,
and targeted these objects with 10ks Chandra ACIS-S observations. The X-ray spectral fitting over
7
1 the 0.3–150keV energy range allows us to determine that all the objects are significantly obscured,
0 having NH 1023cm−2 at a >99% confidence level. Moreover, one to three sources are candidate
2 Comptonth≥ickActiveGalacticNuclei(CT-AGN),i.e., haveNH 1024cm−2. Wealsotesttherecent
≥
“spectralcurvature”methoddevelopedbyKossetal.(2016)tofindcandidateCT-AGN,findingagood
n
agreementbetweenourresultsandtheirpredictions. Sincetheselectioncriteriaweadoptedhavebeen
a
effective in detecting highly obscured AGN, further observations of these and other Seyfert 2 galaxies
J
selectedfromtheSwift-BAT100-monthcatalogwillallowustocreateastatisticallysignificantsample
4
of highly obscured AGN, therefore better understanding the physics of the obscuration processes.
1
Subject headings: galaxies: active — galaxies: nuclei — X-rays: galaxies

]
E
H 1. INTRODUCTION ton scattering. Therefore, the 5% fraction of obscured
∼
AGN observed in most works in literature could not be
The Cosmic X-ray Background (CXB), i.e., the dif-
h. fused X-ray emission observed in the energy range going intrinsic, but could actually being related to an observa-
p from1to 200–300keV,ismainlycausedbyunobscured tional bias. In fact, the modelling of this selection effect
- andobscu∼redActiveGalacticNuclei(AGN;e.g.,Alexan- in Burlon et al. (2011) suggests that the intrinsic CT-
o AGN population accounts for about 20% of the whole
der et al. 2003; Gilli et al. 2007; Treister et al. 2009).
r
AGN population. A similar intrinsic fraction of Comp-
t However,whiletheunobscuredAGNpopulation,respon-
s ton thick sources has been computed also by Brightman
a sible for the CXB emission at <10 keV, has been almost
& Nandra (2011) and Vasudevan et al. (2013).
[ completely detected, at the present day only 30% of
∼ While detecting CT-AGN remains a challenging task,
the CXB at its peak ( 30keV, Ajello et al. 2008) has
1 ∼ in the last ten years several works were able to detect at
been directly detected, thanks to several surveys with
v least a part of these extremely obscured objects. Since
the NuSTAR telescope (Aird et al. 2015; Civano et al.
8 at z >1 the >10keV energy range is observable in the
2015; Harrison et al. 2015; Mullaney et al. 2015). More-
3 0.5–10keV band, deep-field surveys with Chandra and
over,whiletheoreticallyComptonthick(CT-)AGN,i.e.,
9 XMM-Newton havebeenabletodetectCT-AGNatz >1
sources with absorbing column density N 1024cm−2,
3 H ≥ (see, e.g., Georgantopoulos et al. 2013; Vignali et al.
0 are expected to be numerous (see, e.g., Risaliti et al. 2014; Buchner et al. 2015; Lanzuisi et al. 2015). A fur-
. 1999) and responsible for 10–25% of the CXB emis-
1 ∼ ther step in detecting and analyzing candidate CT-AGN
sion at its peak, the fraction of CT-AGN detected with
0 was the launch of NuSTAR, which allowed to study sin-
respect to the total number of AGN is so far 5% (Co-
7 ∼ glesourceswithunprecedentedstatisticsinthe5–50keV
mastri 2004; Della Ceca et al. 2008).
1 band(see,e.g., Ar´evaloetal.2014;Balokovi´cetal.2014;
ForAGNwithcolumndensities 1025cm−2,afraction
v: of direct nuclear emission can be≤observed at 10keV; Gandhietal.2014;Annuaretal.2015;Baueretal.2015;
i for larger column densities, instead, the only o≥bservable Koss et al. 2015; Lansbury et al. 2015; Boorman et al.
X 2016; Puccetti et al. 2016; Ricci et al. 2016).
component in the same energy range is the scattered,
AGN obscuration is usually linked to the presence of
r indirect one (see, e.g., Matt et al. 1999; Yaqoob et al.
a a so-called “dusty torus”, i.e., the gas and dust mate-
2010). Consequently, a complete census of the CT-AGN
rialsurroundingtheAGNaccretiondiskandresponsible
populationrequiresdeepobservationsabove10keV.Bur-
for the obscuration. This material is not continuously
lonetal.(2011)analyzedthe15–55keVspectralproper-
distributed, but more likely a clumpy distribution of op-
ties of a complete sample of AGN at z <0.1 detected
ticallythickdustyclouds(e.g.,Elitzur&Shlosman2006;
by Swift-BAT, founding that the observed fraction of
Risalitietal.2007;H¨onig&Beckert2007;Nenkovaetal.
CT-AGN in this sample is 5%. However, in the same
∼ 2008). In the disk outflow scenario (Emmering et al.
workitwaspointedoutthatevenabove10keVtheAGN
1992) the obscuration is caused by a clumpy disk wind
emissionisstronglysuppressedbyabsorptionandComp-
and can be described using two main parameters: the
AGN luminosity and Eddington ratio. This model pre-
1Department of Physics & Astronomy, Clemson University, dicts that in the low-luminosity regime (i.e., at bolomet-
Clemson,SC29634,USA
2INAF–OsservatorioAstronomicodiBologna,viaRanzani1, ric luminosities Lbol 1042erg s−1) the cloud outflow
≤
40127Bologna,Italy process cannot be sustained by the AGN and the broad
3INAF-IstitutodiAstrofisicaSpazialeeFisicaCosmica,Via line region (BLR) and the dusty clouds should both dis-
U.LaMalfa153,I-90146Palermo,Italy
2 Marchesi et al.
appear (Nicastro 2000; Elitzur & Ho 2009). This pre- BAT spectral redistribution matrix5.
diction was observationally confirmed by Burlon et al. In Table 1 we report the list of sources we analyze
(2011),whichfoundadecreaseinthefractionofobscured in this work. We selected from the Swift-BAT 100-
AGN detected by Swift-BAT at L 1043erg month catalog seven nearby AGN (z <0.03), associated
15−55keV
s−1. Asimilarbehaviourmaybelinkedtoacha≤ngeinthe toSeyfert2galaxies, lackingofaROSATcounterpartin
AGNaccretionregime,switchingfromanopticallythick, the 0.1–2.4keV band. Two out of seven sources (ESO
geometrically thin disk (e.g. Shakura & Sunyaev 1973) 116-G018 and NGC 5972) have never been observed be-
to a radiatively inefficient accretion flow (e.g., Narayan fore at 0.3–10 keV energies, and none of them has been
& Yi 1994; Blandford & Begelman 1999). However, a previously observed using Chandra. The nearest source
recent work by Brightman et al. (2015) did not find ev- has z=0.0122 (i.e., d 52 Mpc), while the farthest has
∼
idence of decrease of the obscured AGN fraction at low z=0.0297 (d 130 Mpc). The 15–55keV luminosity
luminosities. Consequently, the detection and study of range spans fr∼om L 2 1042erg s−1 to L
BAT BAT
low-luminosity, highly obscured AGN becomes strategic 2 1043erg s−1. We se∼lecte×d low-redshift sources be∼-
to better understand the physics of the torus and the ca×use the vast majority (>85%) of Swift-BAT-detected
accretion disk. CT-AGN have so far been discovered at z <0.04 (see,
In this work, we analyze the 0.3-150 keV spectra of e.g., Burlon et al. 2011; Ricci et al. 2015). Moreover,
sevennearbySeyfert2galaxiesdetectedwithSwift-BAT the lacking of a ROSAT counterpart is already a hint
in the 15–150 keV band and reported in the Swift-BAT of at least moderate obscuration, with N (cid:38) 1022cm−2
H
100-month catalog. These sources have L (see, e.g., Figure 2 of Koss et al. 2016).
15−55keV
1043ergs−1. Weobtaineda10ksfollow-upwithChandr≤a Allthesourceshavebeentargetedwith10ksChandra
ACIS-Sinthe0.3–7keVbandforeachobjectinthesam- ACIS-SobservationsduringChandra Cycle17(Proposal
ple. The combination of Chandra and Swift-BAT obser- number 17900432, P.I. Marco Ajello). The data have
vations provides us an ideal dataset to properly charac- been reduced with the CIAO (Fruscione et al. 2006) 4.7
terize the main physical features of these sources, which softwareandtheChandra CalibrationDataBase(caldb)
are all expected to be heavily obscured and even Comp- 4.6.9,adoptingstandardprocedures;nosourceshowssig-
ton thick. In Section 2 we describe our sample and the nificant pile-up, as measured by the CIAO pileup map
datareductionprocedureadoptedfortheChandra data. tool. Sourceandbackgroundspectrahavebeenextracted
In Section 3 we present the result of the spectral fitting using the CIAO specextract tool. Source spectra have
procedure, a physical analysis of the best-fit models and beenextractedincircularregionsof4(cid:48)(cid:48),whilebackground
a comparison with previous results for two of the seven have been extracted from annuli having internal radius
sources. Finally, in Section 4 we discuss our results, fo- r =7(cid:48)(cid:48) and external radius r =20(cid:48)(cid:48), after a visual in-
int out
cusing on the effectiveness of our selection procedure in spection to avoid contamination from nearby sources.
detecting highly obscured AGN. Point-source aperture correction was applied during the
Throughout this work, we adopt a cosmology with source spectral extraction process. Finally, the spectra
Ω = 0.27, Ω =0.73, and H =71 km s−1 Mpc−1. Er- werebinnedinordertohaveatleast15binsintheChan-
m Λ 0
rors are quoted at 90% confidence level unless otherwise dra spectrum: each bin contains from 8 to 20 counts,
stated. according to the source brightness.
2. SAMPLEPROPERTIESANDDATAREDUCTION 3. SPECTRALFITTINGRESULTS
The Swift satellite (Gehrels et al. 2004) is equipped In Table 2 we report the results of the spectral fitting
with the wide-field (120 90 deg2) Burst Alert Telescope of the seven Swift-BAT Seyfert 2 galaxies in our sample.
×
(BAT; Barthelmy et al. 2005). This telescope performs The spectral fitting has been performed using XSPEC
imaging in the 15-150 keV energy range and has been v. 12.9.0 (Arnaud 1996). Galactic absorption was deter-
continuously observing the whole sky since its launch. mined using the Heasoft tool nh (Kalberla et al. 2005).
The most recent catalog of sources detected in the BAT All sources have first been fitted with a basic power
survey data is the 100-month catalog4, which contains law with photon index Γ and intrinsic absorption N
H,z
sourcesdetecteddowntoafluxlimitf 3.3 10−12 erg caused by the dusty torus surrounding the AGN accre-
s−1 cm−2 inthe15-150keVband. Wit∼hitsc×ombination tion disk. The intrinsic absorption is computed using
of good sensitivity and all-sky coverage, this instrument the MyTorus model (Murphy & Yaqoob 2009), which as-
provides an excellent view of the hard X-ray low lumi- sumes a toroidal distribution of the absorbing material
nosity population in the nearby Universe. We point out and is particularly effective in fitting spectra of heavily
that the 100-month BAT catalog is yet to be published obscured objects.
(Segreto et al. in prep.), but all the data used in this Wedeterminedthatforallsevensourcesinthesample
work have been fully reprocessed. morecomplexmodelsthanthebasicabsorbedpowerlaw
To reprocess the BAT survey 100-month data avail- one were required to significantly improve the fit. The
able in the HEASARC public archive we used the significance of the additional components has been ver-
BAT IMAGER code (Segreto et al. 2010). This soft- ified performing a F-test, when statistically allowed; in
ware has been developed to analyze data from coded theremainingcases(i.e.,whenaGaussianorareflection
mask instruments and performs screening, mosaicking componentwereadded,see Protassovetal.2002)thefits
and source detection. The spectra used in this work are without the additional components always had reduced
background subtracted and have been obtained averag- χ2, χ2/(degrees of freedom)>1.5, while the best-fit with
ing over the whole BAT exposure. We used the official
5 http://heasarc.gsfc.nasa.gov/docs/heasarc/caldb/data/swift/
4http://bat.ifc.inaf.it/100m bat catalog/100m bat catalog v0.0.htm bat/index.html
X-ray spectral properties of seven heavily obscured Seyfert 2 galaxies 3
Swift-BATID Sourcename R.A. Dec z L ObsID ObsDate
15−55keV
deg(J2000) deg(J2000) ergs−1
J1253.5–4844 NGC4785 193.3639 –48.7492 0.0123 2.3×1042 18074 2016Apr16
J1024.6–2332 ESO500-G034 156.1310 –23.5530 0.0122 1.8×1042 18075 2016Feb08
J0333.7–0504 NGC1358 53.4153 –5.0894 0.0134 2.0×1042 18076 2015Nov21
J0324.8–6043 ESO116-G018 51.2210 –60.7384 0.0185 2.5×1042 18077 2016Aug05
J0448.9–5739 ESO119-G008 72.2364 –57.6592 0.0229 4.4×1042 18078 2016Jul09
J2234.9–2543 ESO533-G050 338.7076 –25.6769 0.0264 1.1×1043 18079 2016May22
J1539.0+1701 NGC5972 234.7257 17.0262 0.0297 1.0×1043 18080 2016Apr04
Table 1
Summary of the sources in our sample. All the Swift-BAT IDs come from the Palermo BAT Catalog fourth version
(4PBC). “Obs ID” and “Obs Date” refer to the Chandra observations.
the additional component has reduced χ2 1. tics of the spectrum of NGC 5972, we fixed the re-
∼
These are the different components we used in the fit, flectionscalingfactorRto1. SinceitisR=Ω/2π,
inadditiontothebasicabsorbedpowerlaw. Thebest-fit where Ω is the solid angle of the cold material,
models for each source are reported in Table 2. responsible for the reflection, visible from the hot
corona, responsible from the X-ray emission, R=1
1. Sixspectrarequireasecondpowerlawcomponent,
is the case where the reflection is produced by an
having photon index Γ =Γ , the photon index of
2 1 infiniteslabisotropicallyilluminatedbythecorona
the main power law, and not affected by the in-
emission.
trinsicabsorptionN . Thissecondpowerlawac-
H,z
counts for a fraction of AGN emission unabsorbed Since the Swift-BAT spectrum is averaged on 100
bythetorusand/orforascatteredcomponent,i.e., months of observations, while the Chandra one is ob-
light deflected rather than absorbed by the dust tainedbyasingleobservation,thejointfitoftheChandra
and gas. The normalization of this second compo- and Swift-BAT spectra requires the addition of a multi-
nent is always <1% than the normalization of the plicative constant (C ) to the Swift-BAT spectrum,
BAT
main power law. to account for both flux variability and cross-correlation
offsets. We examined the Swift-BAT light curves of the
2. Four fits are significantly improved adding the
seven sources in our sample and none of them shows
MyTorus fluorescent emission line component to
strong flares, the flux variability being limited to 20–
model an excess in the spectrum at energy
40%. For five out of seven sources, the best-fit constant
E 6.4keV. The line in the model was fixed at
6.4∼keV (rest-frame) and accounts for the presence we obtained is consistent (within the errors) to CBAT=1
andcanbeexplainedwiththevariabilityobservedinthe
of the iron Kα line. The iron Kα equivalent width
Swift-BAT light curves. For two AGN (NGC 1358 and
variesintherangeEW=[0.15–0.75]keV(wereport
ESO 116-G018), however, the best-fit constant we find
the EW values in Table 2).
is significantly higher, being 9 and 3, respectively.
∼ ∼
3. Two sources in the sample require the addition Such a high variability cannot being ruled out on the
of a black body component with kT <0.3 keV to basis of our observations, since NGC 1358 and/or ESO
model the characteristic “soft excess” commonly 116-G018couldbechanging-lookAGN(i.e.,sourceswith
observed in the AGN spectra and possibly due to a significant change in flux and/or amount of obscuring
the presence of warm gas close to the AGN accre- material); however, similar flux variations are extremely
tion disk or in the BLR (see, e.g., Risaliti & Elvis unlikely in highly obscured sources, the reflection com-
2004). In order to understand the origin of this ponentwashingoutfluctuationsonshorttimescales(see,
excess, for these two sources (NGC 1358 and NGC e.g.,Gandhietal.2015). Therefore,forthesetwosources
5972) we used the equations reported in Lehmer we fixed CBAT to 1.
et al. (2010) to quantify the expected contribution In Figure 1 we show the NH,z distribution versus red-
of star-forming (SF) processes to the X-ray emis- shiftforthesourcesinoursample,whileinFigures2and
sion. WeusedtheIRAS8–1000µmluminositiesto 3 we report the seven spectra and their best fits. In the
compute the star-formation rate (SFR) and SFR inset, we also report the confidence contours for Γ and
to estimate LSF . For both sources, we find NH,z: ascanbeseen,allthesourcesareheavilyobscured.
2−10keV We will discuss in detail the N measurements in the
LSF 1039 erg s−1, i.e., (cid:46)0.1% of the AGN H,z
2−10keV ≤ next section: here we point out that, as a consequence
luminosityintheenergyrange,thereforesafelyrul- of the heavy obscuration, which affects mainly the 0.3–
ing out a significant SF contribution to the X-ray 7kev band, the ratio between the flux in the 15–150keV
emission. bandandthefluxinthe0.3–7 keVoneis>10forallthe
sources in our sample and even >30 for the three most
4. Onesourceinthesample(NGC5972)showsanex-
obscured objects.
cessintheSwift-BATspectrumatE 30keVwhich
∼
isbest-fittedwiththePEXRAVmodel(Magdziarz&
3.1. Intrinsic absorption and its trend with torus
Zdziarski 1995), which parametrizes a power law
inclination angle
with a significant reflection component. This re-
flection component is caused by cold material sur- One of the MyTorus model parameters is θ, the incli-
rounding the accretion disk. Due to the low statis- nation angle between the observer line of sight and the
4 Marchesi et al.
Swift-BATID NH,gal NH,z,θ=90 NH,z,θ=65 θmax Γ EW kTBB SC χ2/DOF Model
1020cm−2 1022cm−2 1022cm−2 ◦ keV keV
NGC4785 12.2 41.38+10.94 77.37+20.65 – 1.86+0.21 0.48+0.33 – 0.47±0.13 7.9/15 po+MyT*(po+zga)
−8.75 −16.47 −0.19 −0.34
ESO500-G034 5.1 57.41+13.48 107.60+25.26 66 2.30+0.27 – – 0.18±0.06 12.3/15 po+MyT*po
−10.80 −20.24 −0.24
NGC1358 3.8 104.87+42.14 196.44+211.09 90 1.59+0.46 0.75+0.75 0.16+0.12 0.77±0.42 14.8/13 po+MyT*(bb+po+zga)
−35.80 −67.03 −0.45 −0.74 −0.11
ESO116-G018 3.1 94.68+46.06 178.13+86.19 81 1.86+0.51 1.41+2.67 – 0.52±0.29 17.2/14 po+MyT*(po+zga)
−39.97 −75.34 −0.48 −0.94
ESO119-G008 1.3 18.15+6.18 34.0211.57 – 1.77+0.21 – – 0.39±0.10 17.7/20 po+MyT*po
−5.49 −10.28 −0.18
ESO533-G050 1.5 22.56+5.45 42.24+10.23 – 1.65+0.18 0.15+0.17 – 0.54±0.12 25.6/22 MyT*(po+zga)
−4.74 −8.88 −0.14 −0.15
NGC5972 3.0 78.62+16.94 147.27+79.89 72 1.63+0.23 – 0.14+0.07 0.65±0.35 17.6/16 po+MyT*(bb+pexr)
−12.34 −59.22 −0.09 −0.05
Table 2
Best fit properties for the seven Seyfert 2 galaxies in our sample. The XSPEC components reported in the “model”
columns are the following: pow is a power law with photon index Γ; when two power laws are present, their photon
index is the same. zga is the MyTorus Iron Kα component with equivalent width EW. bb is a black-body component
having temperature kT and fitting a soft excess observed below 1 keV. pexr is a power law with a cold matter
BB
reflection component. The intrinsic absorption N is modeled with MyTorus (MyT in the table), with inclination
H,z
angle θ equal to 65◦ or to 90◦. θ is the maximum inclination angle for which is N >1024cm−2. SC is the
max H,z
spectral curvature value computed following Koss et al. (2016)
sources in our sample are heavily obscured. Even in
the worst case scenario, i.e., assuming that all sources
are observed edge-on, (θ=90◦), all the objects have
N >1023cm−2 at a 99% confidence level. Moreover,
H,z
four out of seven sources can in principle be CT-AGN,
assuming that we are observing them with a θ close to
2]− 100 the torus opening angle: in such a scenario, the path of
m thephotonsacrossthetorusandtheintrinsicabsorption
c
2 valuearerespectivelysmallerandlargerthaninthecase
210 with θ=90◦.
[
) Tocompleteouranalysis,were-fittedthespectrawith
H
N 50 two other models to describe the absorbed power-law:
(
n the first one is the standard photo-electric absorption
o
pti (zwabs in XSpec), while the other is pclabs, which de-
or scribes the X-ray emission of an isotropic source of pho-
s
b tons, corrected by the absorption caused by a spherical
a
c distribution of material. In this second model, Compton
nsi 20 scattering is taken into account. In both cases, we find
ntri that the NH,z best-fit values are consistent (within the
I errors)withtheMyTorusbest-fitvaluesobtainedassum-
ing θ=90◦. These results may indicate that reliable θ
values are closer to the θ=90◦ limit than to the θ=65◦
10
one. However, a proper disentanglement of the N –θ
H,z
0.010 0.015 0.020 0.025 0.030 can be broken only increasing the 0.3–7keV band expo-
Redshift sure (see Section 4 for further details).
Figure 1. IntrinsicabsorptionNH,z,computedassuming
θ=90◦, as a function of redshift, for the seven sources 3.2. Comparison with previous results
in our sample. The horizontal dashed line marks the Twosourcesinoursamplehavealreadybeenobserved
Compton-thick regime, i.e., NH,z 1024cm−2. inthe0.5–10keVband. Inthissection,wepresentprevi-
≥
ouslypublishedworksandwecomparetheirresultswith
torus symmetry axis, with θ=0◦ being a face-on observ- ours.
ing angle and θ=90◦ being an edge-on observing angle. NGC1358wastargetedwitha10ksXMM-Newton ob-
θ=60◦ isthetorusopeningangle,whileθ=61◦ isthelow- servation in 2005 (PI: I. Georgantopoulos): the spectral
fitting results for this observation (the Swift-BAT spec-
est angle for which the observer line of sight intercepts
trum was not used in this work) have been published in
the torus.
Marinucci et al. (2012). Their results are in good agree-
The quality of our spectra does not allow us to con-
ment with ours: they find that NGC 1358 is consistent
strain N and θ simultaneously. Therefore, we fitted
H,z
our data twice: the first one assuming θ=90◦, i.e., the withbeingaCT-AGN,althoughwithlargeuncertainties,
edge-onscenario,thesecondonewithθ=65◦, i.e.,anan- having best-fit intrinsic absorption NH,z=1.3+−80..56×1024
gleclosertotheangleforwhichtheobserverlineofsight cm−2.
intercepts the torus. As can be seen in Table 2, assum- NGC 4785 has already been studied by Gandhi et al.
ingθ=65◦impliesobtainingbest-fitN values 2times (2015): they combined the Swift-BAT observation used
H,z
higher than those obtained assuming θ=90◦. ∼ in this work with a 79ks Suzaku observation. They fit-
As already pointed out in the previous section, all ted the data with PEXRAV, with MyTorus and with the
X-ray spectral properties of seven heavily obscured Seyfert 2 galaxies 5
TORUS model by Brightman & Nandra (2011). In all tained without applying any selection criteria is 35–55%
cases, they found the source being Compton-thick, the (Burlon et al. 2011).
measuredintrinsicabsorptionvaryingfromN =1.4+0.5 Koss et al. (2016) developed a technique to identify
H,z −0.9
cm−2 whilefittingthedatawithPEXRAVtoN =2.7+3.8 CT-AGN in AGN with low counts statistics observed
H,z −0.8 with Swift-BAT or NuSTAR, finding that the fraction
cm−2 fitting the data with TORUS. Our N estimate is
H,z of CT-AGN at z <0.03 is 22%. The Koss et al. (2016)
thereforeconsistent,withintheerrors,withtheirPEXRAV ∼
method is based on the curvature of the AGN spectrum
best-fit, and lie below their results with MyTorus and
between 14 and 50 keV, parametrized as follows:
TORUS. While the hypothesis that this discrepancy can
be partially driven by the low statistics in our fit, which 3.42A 0.82B+1.65C+3.58D
SC = − − , (1)
has only 16 degrees of freedom, compared to the 89 of BAT Tot
the Gandhi et al. (2015) one, it is also possible that the
where A, B, C, D and Tot are the count
change in the measure intrinsic absorption is due to in-
rates measured with Swift-BAT in the 14-20keV, 20-
trinsic variability in the amount of obscuring material
24keV,24-35keV,35-50keVand14-50keV,respectively.
in the time between the Suzaku observation (taken in
SC =0.4 is the CT-AGN selection threshold: in
July 2013) and the Chandra one (taken in April 2016). BAT
Koss et al. (2016) work, seven of the nine sources with
A significant change of the AGN luminosity seems less
SC >0.4intheirsamplehaveN >1024cm−2,and
likely, since the power-law photon index we measured BAT H,z
(Γ=1.80+0.21) is in good agreement with the one of the remaining two are significantly obscured (NH,z > 5
−0.19 1023 cm−2).
Gandhi et al. (2015) (Γ=2.1+0.4 using MyTorus). Fi- ×
−0.3 In Table 2 we report the SC values for the seven
BAT
nally, Gandhi et al. (2015) find evidence of a strong Iron sources in our sample: as can be seen, the Koss et al.
Kα line in NGC 4785, with EW=1.1+u keV (where u (2016) assumption on SC is supported by our re-
−0.4 BAT
meansunconstrained),whiletheEWofthelineinourfit sults. In fact, the most obscured source in our sam-
is weaker (EW=0.48+−00..3334). This change in line intensity ple (NGC 1358) has SCBAT=0.77±0.42, i.e., the high-
may strenghten the hypothesis that we are observing a est SCBAT value for the sources in our sample, al-
variationintheamountofobscuringmaterial, sinceCT- though the uncertainty on the SC value is quite large.
AGNusuallyshowprominentironlineswithEW>1keV Other two highly obscured sources (ESO 116-G018 and
(see, e.g., Koss et al. 2016). NGC5972),bothhaveSCBAT >0.4(SCBAT=0.52 0.29
±
andSC =0.65 0.35,respectively),althoughboththe
BAT
4. DISCUSSIONANDCONCLUSIONS ±
sources have SC <0.4 at a 90% confidence level.
BAT
In this work, we presented the spectral fitting over the These two objects have N <1024 cm−2 assuming
H,z
0.3–150 keV energy range, obtained combining Chandra θ=90◦, but may have N >1024 cm−2 assuming rea-
H,z
and Swift-BAT observations, of seven AGN at z <0.03 sonable values of θ (81◦ and 72◦, respectively). Finally,
selected from the 100-month Swift-BAT all-sky survey two sources with SC >0.4, and therefore candidate
BAT
catalog. These sources are associated to Seyfert 2 galax- CT-AGN according to the Koss et al. (2016) equation,
ies and lack of a ROSAT counterpart in the 0.1–2.4keV have N 2–4 1023 cm−2 assuming θ=90◦.
H,z
band. The combined Chandra and Swift-BAT spectra The resu∼lts of ×this work represent a first step in the
have then been fitted with different models, to estimate processofidentifyingandanalyzingCT-AGNinthelocal
spectral parameters such as the spectral photon index Γ Universe. Toimproveourfindings,weplantoextendour
and the intrinsic absorption NH,z. analysis in two ways:
As discussed in Section 3.1, all the sources in the sam-
ple are heavily obscured: even in the (unlikely) scenario 1. Increasing the exposure for the objects we de-
of all the sources being edge-on (i.e., with inclination scribed in this work, using XMM-Newton, Chan-
angle θ=90◦) the less obscured object (ESO 119-G008) dra and NuSTAR. The 0.3-7 keV spectra studied
has N =1.82+0.62 1023 cm−2 and all sources have in this work have been obtained with observations
N >H,1z023 cm−−0.255at×a 99% confidence level. Moreover, of only 10ks each, and have between 80 (for the
H,z two most obscured AGN in our sample) and 270
one source (NGC 1358) has best-fit intrinsic absorption
(ESO 533-G050) net counts in the 0.3-7 keV band:
N =1.05+0.42 1024 cm−2, i.e., consistent with being
H,z −0.36 × breaking the degeneracy between the intrinsic ab-
a Compton thick AGN.
sorption N and the inclination angle θ, there-
Furthermore, we performed a further fitting analysis H,z
fore definitely assessing if a source is or is not a
varying θ in 1◦ steps, and we found that five of the
CT-AGN, requires a larger counts statistics. As a
sources in our sample can be CT-AGN, if the inclina-
comparison, Tanimoto et al. (2016) constrained si-
tion angle θ between the observer line of sight and the
multaneously N and θ for the three candidate
torus rotation axis is small enough (see column θ in H,z
max CT-AGN observed with Suzaku for 30-70ks and
Table 2).
having each about 500 net counts in the 0.3-7keV
A first important result of this work is therefore the
band.
finding of an effective way to detect obscured AGN in
the nearby Universe, taking advantage of the combina- 2. Increasing the sample of candidate CT-AGN. The
tion of low limiting flux and wide field of view of the selection criteria we adopted in this work proved
Swift-BATall-skysurvey. Itisworthnoticingthat,while very effective, and the Swift-BAT 100-month cata-
with our selection technique we detect a population of log contains other 5–10 sources at z <0.04 that
sources with NH,z > 1023 cm−2, the observed fraction are Seyfert 2 galax∼ies lacking of ROSAT counter-
of unobscured AGN (i.e., with N < 1022 cm−2) ob- part and which have not been observed by either
H,z
6 Marchesi et al.
Chandra orXMM-Newton. Weplantotargetthese Gilli,R.,Comastri,A.,&Hasinger,G.2007,A&A,463,79
sources with future Chandra and XMM-Newton Harrison,F.A.,Aird,J.,Civano,F.,etal.2015,ArXive-prints
observations. [1511.04183]
H¨onig,S.F.&Beckert,T.2007,MNRAS,380,1172
Kalberla,P.M.W.,Burton,W.B.,Hartmann,D.,etal.2005,
A&A,440,775
REFERENCES Koss,M.J.,Assef,R.,Balokovi´c,M.,etal.2016,ApJ,825,85
Koss,M.J.,Romero-Can˜izales,C.,Baronchelli,L.,etal.2015,
Aird,J.,Alexander,D.M.,Ballantyne,D.R.,etal.2015,ApJ, ApJ,807,149
815,66 Lansbury,G.B.,Gandhi,P.,Alexander,D.M.,etal.2015,ApJ,
Ajello,M.,Greiner,J.,Sato,G.,etal.2008,ApJ,689,666 809,115
Alexander,D.M.,Bauer,F.E.,Brandt,W.N.,etal.2003,AJ, Lanzuisi,G.,Ranalli,P.,Georgantopoulos,I.,etal.2015,A&A,
126,539 573,A137
Annuar,A.,Gandhi,P.,Alexander,D.M.,etal.2015,ApJ,815, Lehmer,B.D.,Alexander,D.M.,Bauer,F.E.,etal.2010,ApJ,
36 724,559
Ar´evalo,P.,Bauer,F.E.,Puccetti,S.,etal.2014,ApJ,791,81 Magdziarz,P.&Zdziarski,A.A.1995,MNRAS,273,837
Arnaud,K.A.1996,inAstronomicalSocietyofthePacific Marinucci,A.,Bianchi,S.,Nicastro,F.,Matt,G.,&Goulding,
ConferenceSeries,Vol.101,AstronomicalDataAnalysis A.D.2012,ApJ,748,130
SoftwareandSystemsV,ed.G.H.Jacoby&J.Barnes,17 Matt,G.,Pompilio,F.,&LaFranca,F.1999,NA,4,191
Balokovi´c,M.,Comastri,A.,Harrison,F.A.,etal.2014,ApJ, Mullaney,J.R.,Del-Moro,A.,Aird,J.,etal.2015,ApJ,808,184
794,111 Murphy,K.D.&Yaqoob,T.2009,MNRAS,397,1549
Barthelmy,S.D.,Barbier,L.M.,Cummings,J.R.,etal.2005, Narayan,R.&Yi,I.1994,ApJL,428,L13
SpaceSci.Rev.,120,143 Nenkova,M.,Sirocky,M.M.,Ivezi´c,Zˇ.,&Elitzur,M.2008,ApJ,
Bauer,F.E.,Ar´evalo,P.,Walton,D.J.,etal.2015,ApJ,812,116 685,147
Blandford,R.D.&Begelman,M.C.1999,MNRAS,303,L1 Nicastro,F.2000,ApJL,530,L65
Boorman,P.G.,Gandhi,P.,Alexander,D.,etal.2016,ArXiv Protassov,R.,vanDyk,D.A.,Connors,A.,Kashyap,V.L.,&
e-prints[1610.08997] Siemiginowska,A.2002,ApJ,571,545
Brightman,M.,Balokovi´c,M.,Stern,D.,etal.2015,ApJ,805,41 Puccetti,S.,Comastri,A.,Bauer,F.E.,etal.2016,A&A,585,
Brightman,M.&Nandra,K.2011,MNRAS,413,1206 A157
Buchner,J.,Georgakakis,A.,Nandra,K.,etal.2015,ApJ,802, Ricci,C.,Bauer,F.E.,Treister,E.,etal.2016,ApJ,819,4
89 Ricci,C.,Ueda,Y.,Koss,M.J.,etal.2015,ApJL,815,L13
Burlon,D.,Ajello,M.,Greiner,J.,etal.2011,ApJ,728,58 Risaliti,G.&Elvis,M.2004,inAstrophysicsandSpaceScience
Civano,F.,Hickox,R.C.,Puccetti,S.,etal.2015,ApJ,808,185 Library,Vol.308,SupermassiveBlackHolesintheDistant
Comastri,A.2004,inAstrophysicsandSpaceScienceLibrary, Universe,ed.A.J.Barger,187
Vol.308,SupermassiveBlackHolesintheDistantUniverse,ed. Risaliti,G.,Elvis,M.,Fabbiano,G.,etal.2007,ApJL,659,L111
A.J.Barger,245 Risaliti,G.,Maiolino,R.,&Salvati,M.1999,ApJ,522,157
DellaCeca,R.,Caccianiga,A.,Severgnini,P.,etal.2008,A&A, Segreto,A.,Cusumano,G.,Ferrigno,C.,etal.2010,A&A,510,
487,119 A47
Elitzur,M.&Ho,L.C.2009,ApJL,701,L91 Shakura,N.I.&Sunyaev,R.A.1973,A&A,24,337
Elitzur,M.&Shlosman,I.2006,ApJL,648,L101 Tanimoto,A.,Ueda,Y.,Kawamuro,T.,&Ricci,C.2016,PASJ,
Emmering,R.T.,Blandford,R.D.,&Shlosman,I.1992,ApJ, 68,S26
385,460 Treister,E.,Urry,C.M.,&Virani,S.2009,ApJ,696,110
Fruscione,A.,McDowell,J.C.,Allen,G.E.,etal.2006,in Vasudevan,R.V.,Brandt,W.N.,Mushotzky,R.F.,etal.2013,
Proc.SPIE,Vol.6270,SocietyofPhoto-Optical ApJ,763,111
InstrumentationEngineers(SPIE)ConferenceSeries,62701V Vignali,C.,Mignoli,M.,Gilli,R.,etal.2014,A&A,571,A34
Gandhi,P.,Lansbury,G.B.,Alexander,D.M.,etal.2014,ApJ, Yaqoob,T.,Murphy,K.D.,Miller,L.,&Turner,T.J.2010,
792,117 MNRAS,401,411
Gandhi,P.,Yamada,S.,Ricci,C.,etal.2015,MNRAS,449,1845
Gehrels,N.,Chincarini,G.,Giommi,P.,etal.2004,ApJ,611,
1005
Georgantopoulos,I.,Comastri,A.,Vignali,C.,etal.2013,A&A,
555,A43
X-ray spectral properties of seven heavily obscured Seyfert 2 galaxies 7
NGC 4785 ESO 500−G034
1.21.2 1.61.6
11 1.41.4
m s keV)−2−1−1 10−3 nH (10)240.40.60.80.40.60.8 11..66 11..88Ph+oton Index22 22..22 0.70.7 NGC 4785 m s keV)−2−1−1 10−3 nH (10)240.811.20.811.211..88 22 22..22Pho+ton In22d..44ex 22..66 22..88
c c
ons 0.60.6 ons 0.80.8
ot ot
h h
keV (P2 10−4 nH (10)24 0.40.50.40.5 + keV (P2 10−4 nH (10)24 0.60.6 +
0.30.3
0.40.4
11..44 11..66 Pho11t..o88n Index 22 22..22 11..88 22 Photo22n..22 Index 22..44 22..66
1 10 100 1 10
Energy (keV) Energy (keV)
NGC 1358 ESO 116−18
1
0
0. 88 33
m s keV)−2−1−1 10−3 nH (10)24 246246 11 P11h..55oto+n Index 22 NGC 1358 m s keV)−2−1−1 1010−4−3 nH (10)24 11.522.511.522.5 11..66 11..88+ Ph22oton Index22..22 22..44
ns c 22 ns c 22
keV (Photo2 10−4 nH (10)24 11.511.5 + keV (Photo2 010−6−5 nH (10)24 11.511.5 +
1
0.50.5
0.50.5
10−5 1 10 11 P11h..55oton Index 22 100 10−7 1 11 10 11..55 Photon Ind22ex 22..55 100
Energy (keV) Energy (keV)
ESO 119−8 ESO 533−50
1 0.60.6
0
0.50.5 0. 0.50.5
V)−1 0.01 nH (10)24 0.40.4 + V)−1 nH (10)240.30.40.30.4 +
ke 0.30.3 ke 11..44 11..55 11..66Photon11 ..I77ndex 11..88 11..99 22
s −1 s −1
ns cm −2 10−3 11..66 Ph11o..88ton Index 22 ns cm −2 10−3 0.30.3
oto 0.250.25 oto
keV (Ph2 10−4 nH (10)24 0.20.2 + keV (Ph2 nH (10)24 0.20.250.20.25 +
0.150.15 11..66 Pho11to..88n Index 22 10−4 0.150.1511..33 11..44 11..55 Pho11t..o66n Index11..77 11..88 11..99
2 5 10 20 50 100 5 10 20 50
Energy (keV) Energy (keV)
Figure 2. Chandra (black) and Swift-BAT (red) spectra of six of the seven sources in our sample. The best-fitting
model is plotted as a solid line, while the single components are plotted as dotted lines. In the inset, the confidence
contoursat68,90and99%confidencelevelforΓandN (inunitsof1024 cm−2)areshown,assumingθ=90◦ (bottom
H,z
right) and θ=65◦ (top left).
8 Marchesi et al.
NGC 5972
33
V)−1 10−3 nH (10)24 1212 +
e
s k−1 10−4 11 11..55Photon Index 22 22..55 NGC 5972
m −2
s c 1.51.5
on 0−5
ot 1
h
V (P2 H (10)24 11
ke 0−6 n +
1
0.50.5
0−7 11 11..55 22 22..55
1 Photon Index
1 10 100
Energy (keV)
Figure 3. Chandra (black) and Swift-BAT (red) spectra of NGC 5972. The best-fitting model is plotted as a solid
line, while the single components are plotted as dotted lines. In the inset, the confidence contours at 68, 90 and 99%
confidence level for Γ and N (in units of 1024 cm−2) are shown, assuming θ=90◦ (bottom right) and θ=65◦ (top
H,z
left).