How complex is the obscuration in AGN? New clues from the Suzaku monitoring of the X-ray absorbers in NGC7582

0.27 MB·English

by  Stefano Bianchi

How complex is the obscuration in AGN? New clues from the
Suzaku monitoring of the X-ray absorbers in NGC 7582
9 Stefano Bianchi
0
Dipartimento di Fisica, Universita` degli Studi Roma Tre, via della Vasca Navale 84, 00146 Roma, Italy
0
2

n Enrico Piconcelli
a
Osservatorio Astronomico di Roma (INAF), Via Frascati 33, I-00040 Monte Porzio Catone, Italy
J

4
1
Marco Chiaberge

] Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218
A
INAF - IRA, Via P. Gobetti 101, I-40129 Bologna, Italy
G
.
h Elena Jimenez Bail´on
p
Instituto de Astronom´ıa, Universidad Nacional Auto´noma de M´exico, Apartado Postal 70-264, 04510
-
o Mexico DF, Mexico
r
t LAEFF Apd. 78 Villanueva de la Can˜ada - 28691-Madrid, Spain
s
a
[
Giorgio Matt

1
Dipartimento di Fisica, Universita` degli Studi Roma Tre, via della Vasca Navale 84, 00146 Roma, Italy
v
3
7
Fabrizio Fiore
9
1 Osservatorio Astronomico di Roma (INAF), Via Frascati 33, I-00040 Monte Porzio Catone, Italy
.
1
0
9
0
: ABSTRACT
v
i We present the results of a Suzaku monitoring campaign of the Seyfert 2 galaxy, NGC 7582.
X
The source is characterized by very rapid (on timescales even lower than a day) changes of the
r
a column density of an inner absorber, together with the presence of constant components arising
as reprocessing from a Compton-thick material. The best fitting scenario implies important
modificationsto the zerothorderview ofUnified Models. While the existence ofa pc-scaletorus
isneededinordertoproduceaconstantComptonreflectioncomponentandanironKαemission
line, in this Seyfert 2 galaxy this is not viewed along the line of sight. On the other hand,
the absorption of the primary continuum is due to another material, much closer to the BH,
roughly at the distance of the BLR, which can produce the observed rapid spectral variability.
On top of that, the constant presence of a 1022 cm−2 column density can be ascribed to the
presenceofa dustlane,extended ona galacticscale,aspreviouslyconfirmedbyChandra. There
is now mounting evidence that complexity in the obscuration of AGN may be the rule rather
than the exception. We therefore propose to modify the Unification Model, adding to the torus
the presence of two further absorbers/emitters. Their combination along the line of sight can
reproduce all the observed phenomenology.
Subject headings: galaxies: active - galaxies: Seyfert - X-rays: individual: NGC7582
1
1. Introduction 1989),Ginga(Warwick et al.1993),ASCA(Schachter et al.
1998; Xue et al. 1998). The picture that emerged
The presence of absorbing material along the
from these studies was that of a flat X-ray spec-
line of sight is generally believed to be the only
trum dominated by heavy obscuration. Thanks
difference between Type 2 and a Type 1 Active
to the BeppoSAX broad bandpass, Turner et al.
Galactic Nuclei (AGN). This material obscures
(2000) reported for the first time the detection of
both the emission lines from the Broad Line Re-
a more complex geometry of the absorbing ma-
gion (BLR) and the X-ray spectrum, being the
terial, likely constituted by two different compo-
mainingredientoftheso-calledUnificationModel.
nents, one of which Compton-thick. This sce-
It is usually envisaged as a compact ‘torus’, lo-
nario was confirmed by a combined imaging anal-
cated ata pc scale distance from the nucleus (e.g.
ysis performed with Chandra and HST, which
Antonucci 1993). This distance is basically con-
suggested that the Compton-thick torus coexists
firmed both by indirect techniques, such as con-
with a large-scale Compton-thin material associ-
siderations based on photoionization codes (e.g.
ated with the dust lane and circumnuclear gas is
Bianchi et al. 2001; Massaro et al. 2006), and di-
photoionizedby the AGN along torus-freelines of
rect ‘imaging’ of the torus itself (e.g. Jaffe et al.
sight (Bianchi et al. 2007a).
2004).
The most interesting results came from the
However, there is evidence that this simple
analysis of the two XMM-Newton observations,
scenario may not hold for all objects. The
taken4yearsapart,in2001and2005(Piconcelli et al.
co-existence of a Compton-thick torus and a
2007). Both clearly show a completely different
Compton-thin material, extended on a much
spectral and flux state with respect to the 1998
larger scale, seems to better account for the ob-
BeppoSAX observation. The XMM-Newton spec-
served phenomenology (e.g. Matt 2000). The lat-
trum can be well described by a model consisting
ter absorber may be naturally associated to dust- ofacombinationofaheavilyabsorbed(N 1024
H
lanes (e.g. Malkan et al. 1998), or to molecular cm−2)powerlawandapurereflectioncomp∼onent,
gas in the galactic disks (Lamastra et al. 2006). both obscured by a column density of few 1022
The presence of obscuring matter on large (pc- cm−2. Notably, Piconcelli et al. (2007) de×tect a
kpc) scales and detached from the nuclear torus
significant increase by a factor 2 in the column
is also supported by Spitzer studies of MIR lu- ∼
density of the inner, thicker absorbercoveringthe
minous quasars at high z which are very likely
primary X-ray source, between 2001 and 2005.
hostedby dusty galaxies(e.g.Polletta et al.2008;
In this paper, we present a Suzaku monitoring
Mart´ınez-Sansigreet al. 2006)
campaign of NGC 7582, which, together with a
Moreover, Risaliti et al. (2002) showed that a
newXMM-Newton observation,confirmsthevari-
largenumberofSeyfert2spresentssignificantvari-
abilityofthecolumndensityoftheinnerabsorber,
ability of the absorbing column density (N ) on
H but down to timescales smaller than a day.
timescales as low as months, thus suggesting that
the absorbing material should be much closer to 2. Observations and data reduction
the nucleus than assumed for the torus, possi-
bly in the BLR itself. This picture seems the 2.1. Suzaku
only tenable for the objects, which present the
During the second Suzaku Announcement of
most rapid N variations ever observed, within
H
Opportunity (AO2), we proposed a strategy to
daysorevenhours: NGC 4388(Elvis et al.2004),
observe NGC 7582 at different timescales, from 1
NGC 1365 (Risaliti et al. 2005) and NGC 4151
weekto about6months, allowingus to probedis-
(Puccetti et al. 2007). Is this the end of the torus
tancesascloseastheBLRandalmostasfarasthe
paradigm? Orisitonlyanexceptiononahandful
traditional torus. Moreover, this campaign com-
of peculiar objects?
plemented the scales of the order of years already
NGC 7582 (z=0.0053), being included in
tested with XMM-Newton. Therefore, NGC 7582
the Piccinotti et al. (1982) catalog, has been
wasobservedfourtimesbySuzaku in2007(PI:M.
targeted by most X-ray telescopes: Einstein
Chiaberge): on May 1st and 28th, and November
(Maccacaro & Perola1981),EXOSAT (Turner & Pounds
9th and 16th. X-ray Imaging Spectrometer (XIS)
2
and Hard X-ray Detector (HXD) event files were April 29th. Both observations were discussed in
reprocessed with the latest calibration files avail- Piconcelli et al. (2007). Moreover, the source is
able (2008-07-09 release), using ftools 6.5 and within the EPIC field of view of another target,
Suzaku software Version 9, adopting standard fil- observed on 2007 April 30th, accidentally just a
tering procedures. Source and background spec- day before the first Suzaku one. In this paper,
tra for all the three XIS detectors were extracted we presentfor the firsttime the 2007observation.
from circular regions of 2.9 arcmin radius, avoid- The observation was performed with the EPIC
ing the calibration sources. Response matrices CCDcameras,thepnandthetwoMOS,operated
and ancillary response files were generated using inFullWindowandMediumFilter. Datawerere-
xisrmfgen and xissimarfgen. We downloaded ducedwithSAS8.0.0andscreeningforintervalsof
the“tuned”non-X-raybackground(NXB)forour flaring particle background was done consistently
HXD/PIN data provided by the HXD team and with the choice of extraction radii, in an iterative
extractedsourceandbackgroundspectrausingthe process based on the procedure to maximize the
same good time intervals. The PIN spectrum was signal-to-noise ratio described by Piconcelli et al.
then corrected for dead time and the exposure (2004). After this process, the net exposure time
time of the background spectrum was increased wasabout15 ksfor the pnspectrum, adoptingan
by a factor 10, as required. Finally, the contri- extractionradius of19 arcsecandpatterns 0 to 4.
bution from the cosmic X-ray background (CXB) In the following, we conservatively decided not to
was subtracted from the source spectrum, simu- use MOS data, which may be affected by possible
lating it as suggested by the HXD team. For the cross-calibration issues related to the obscuration
sake of simplicity, we will always refer to observa- by the Reflection Grating Array (Mateos et al.,
tion S1, S2, S3 and S4 in this paper, as listed in in preparation). The background spectra were
Table1,wherethefinalnetexposuretimesforthe extracted from source-free circular regions with a
threeXISspectraandtheHXD/PINarereported. radius of 50 arcsec. Finally, spectra were binned
As a final note, let us discuss the possible con- inordertooversampletheinstrumentalresolution
tamination of other sources in the field of view by at least a factor of 3 and to have no less than
(FOV) of the XIS and, most of all, the PIN. The 25 counts in each background-subtractedspectral
two brightest X-ray sources close to NGC 7582 channel. The latter requirement allows us to use
are the Seyfert 2 galaxy, NGC 7590, and the BL the χ2 statistics.
Lac PKS 2316-423. Both are largely outside the
extraction regions adopted for the three XIS de- 3. Spectral analysis
tectors. AsforthePIN,wereconstructedtheFOV
In the following, errors correspond to the 90%
of each observation, using the tool aemkreg and
confidence level for one interesting parameter
the actual Euler angle of each pointing. In all
(∆χ2 = 2.71), where not otherwise stated. The
the cases, both sources are inside the PIN FOV,
adopted cosmologicalparameters are H =70 km
but, as already noted by Turner et al. (2000) for 0
s−1 Mpc−1, Ω = 0.73 and Ω = 0.27 (i.e. the
the BeppoSAX PDS (whose FOV is larger than Λ m
default ones in xspec 12.4.0: Arnaud 1996). In
thatofthePIN),theyaresignificantlysofterthan
all the fits, the Galactic column density along the
NGC 7582 and their hard X-ray fluxes are much
line of sight to NGC 7582 is included (1.9 1020
dimmer. Moreover, the BL Lac lies very close to cm−2: Dickey & Lockman 1990). ×
theborderofthe34x34arcminsquarewhichrep-
All Suzaku XIS instruments were used, but, in
resents the Full Width Half Maximum of the PIN
order to avoid inter-calibration issues at low en-
FOV,thuscontaminatingonlyfor 50percentof
its flux1. ≃ ergies, the full band (0.5-10 keV) was employed
only for the back-illuminatedXIS1, whichhas the
2.2. XMM-Newton largest effective area at low energies, while a re-
strictedband(2-10keV)waspreferredforthetwo
XMM-Newton observed NGC 7582 in two tar-
front-illuminatedXIS0andXIS3. Weaddedanor-
geted exposures on 2001 May 25th and 2005
malization constant for each instrument, fixing to
1 the value for XIS0, andto 1.18that for the PIN
1SeeSect. 8.3of“TheSuzakuTechnical Description”.
3
Table 1: Log for the 2007 XMM-Newton and the four Suzaku observations.
Obs Obsid Date ∆t PN XIS0 PIN
(1) (2) (3) (4) (5) (6) (7)
XMM07 0405380701 2007-04-30 – 15 – –
S1 702052010 2007-05-01 <1 – 24 20
S2 702052020 2007-05-28 27 – 29 25
S3 702052030 2007-11-09 165 – 29 23
S4 702052040 2007-11-16 7 – 32 24
(1)Thenamewhichidentifiestheobservationinthiswork;(2)XMM-Newton orSuzakuobservationidentifier;
(3)Observationdate;(4)Timeelapsedfrompreviousobservation(days);(5)NetexposuretimefortheEPIC
pn (ks); (6) Net exposure time for the XIS0 (ks); (7) Net exposure time for the HXD PIN (ks)
(18-50 keV), as appropriate for data taken at the
HXD nominal position2. The values for the XIS1
and XIS3 instruments were left free to vary.
A first look at the Suzaku XIS spectra of the
four observations immediately reveals that the
source varied dramatically above 3 keV (see Fig.
1). On the other hand, the HXD pin spectra,
although in a less conclusive way because of the
muchlowerstatisticsandmuchlargeruncertainty,
5
0
appearfairlyinagreementwitheachother. There- 0.
fore, in the next Sections we will investigate the
origin of the variability by fitting separately the
available datasets. 2
0
0.
3.1. The ‘low-flux’ state V−1
e
We started our analysis from observation S4, nts s k−1 0.01
u
which closely resembles the ‘low-flux’ state al- Co
ready observed in the 2005 XMM-Newton spec- 10−3 S1
trum. We therefore decided to adopt the model 5× SS23
S4
used by Piconcelli et al. (2007) for the above-
mentioned dataset, which, given the longer ex-
posure, has a better signal-to-noise. The model 10−3
×
is constituted by an highly obscured powerlaw3 2 1 2 5
Energy (keV)
2Although the normalization factor we adopted
refers to the PIN band of 12-40 keV (see Fig. 1.— NGC 7582: Suzaku XIS0 spectra for
http://www.astro.isas.ac.jp/suzaku/doc/suzakumemo/suzakumemo-t2h0e08f-o0u6.rpdofbservationsofNGC7582: S1(black), S2
andhttp://heasarc.gsfc.nasa.gov/docs/suzaku/analysis/watchout.html),
(red), S3 (green) and S4 (blue).
it was estimated that the variation of this
factor when different bands are used for the
PIN is roughly less than 2 per cent (see
ftp://legacy.gsfc.nasa.gov/suzaku/doc/xrt/suzakumemo-2007-11.pdf),
which is lower than the uncertainty on
the normalization factor itself (see again
http://heasarc.gsfc.nasa.gov/docs/suzaku/analysis/
watchout.html)
3The column densities reported in this paper take into ac-
countbothphotoelectricabsorptionandComptonscatter-
4
andapureComptonreflectioncomponent(model NGC 7582 remain constant over long periods of
pexrav in Xspec), together with a Gaussian time. Considering also the 2001 XMM-Newton
line to model iron Kα emission. The latter two observation,theintrinsicpowerlawindex,thenor-
components, though unaffected by the large col- malization of the Compton reflection component
umn density which obscures the primary com- and the flux of the iron Kα line do not show sig-
ponent, are absorbed by a smaller column den- nificant variability in 6 years. On the other hand,
sity. The soft excess is modeled with a powerlaw it is clear from the two XMM-Newton and the 4
and as many Gaussian lines as required, in order Suzaku observations that the spectrum has dra-
to mimic the emission from extended photoion- matically varied. What is the main driver of this
ized gas commonly found in Seyfert 2s (see e.g. variability?
Guainazzi & Bianchi2007). Asfoundforthe2005 The complexity of the best fit model implies
XMM-Newton observation, this model is highly somedegeneracybetweenthespectralparameters,
satisfactoryalso for the Suzaku data of the fourth most of all in the datasets where the presence of
observation (χ2 = 91/94 dof): see last column a stronger primary continuum does not allow us
of Table 2, where all the best fit parameters are todisentangleunambiguouslytheComptonreflec-
listed. tioncomponent. However,wecannowreasonably
A comparison with the results of the 2005 assumethatthepowerlawindexandtheCompton
XMM-Newton observationis veryinteresting. We reflectioncomponentsarenotaffectedbyvariabil-
recover the same best fit parameters reported in ity. Therefore, in all the following fits we decided
Piconcelli et al. (2007), such as the photon index to use the photon index and the pexrav normal-
andnosignificantvariationisdetectedforthetwo izationfixedtothebestfitvaluesfoundinS4. We
absorbing column densities. Moreover, the nor- leave the neutral iron Kα line flux and centroid
malizations of the Compton reflection component energy free to vary, since it is much easier to de-
are perfectly consistent between the 2005 XMM- tect variability in this sharp feature with respect
Newton observation (9.7+0.8 10−3 ph keV−1 to a continuum component. This is a consistency
cm−2s−1)4andtheSuzaku−1d.8at×a. Theneutraliron test: given the common origin of the iron Kα line
Kαlinefluxdoesnotshowanysignificantvariabil- and the reflection component in our model, any
itybetweenthetwodatasets,either. Itsequivalent variation of the former would invalidate our as-
width (EW) with respect to the Compton reflec- sumption.
tion component is around 1 keV, as expected if Adopting the model described above, we then
arising as reprocessing from the same Compton- fitted the other three Suzaku spectra and the new
thick material (e.g. Matt et al. 1991). Finally, 2007 XMM-Newton data. In all cases, we found
whilewereferthereadertoPiconcelli et al.(2007) very good fits (see Table 2 and Fig. 2). Allow-
foradetailedanalysisofthesoftX-rayemissionof ing the frozen parameters to vary does not sig-
NGC7582,weconfirmherethedetectionofstrong nificantly improve the fits (whose reduced χ2 are
KαlinesfromNeIX,NeX,MgXIandSiXIII,as alreadyveryclose to unity), but only enlargesthe
well as from FeXVII L. The observed 0.5-2 keV statisticaluncertaintiesontheparameters. Allthe
flux is around 1.5 per cent of the nuclear one, af- main parameters of the model appears to be con-
ter correction for absorption, in the range usually stant among the observations. In particular, the
found for Seyfert 2 galaxies(Bianchi & Guainazzi neutral iron Kα line does not show any signifi-
2007). cant variation, thus being consistent with our as-
sumption that the reprocessed components from
3.2. The origin of the variability
Compton-thick material are indeed constant. As
Theresultsintheprevioussectionstronglysug- expected, no significant variability is found in the
gest that some important X-ray parameters in flux and modelization of the soft X-ray emission,
either.
ing(cabs*zwabsinXspec). Therefore,the clearvariabilityobservedamong
4This value was not explicitly reported in Piconcellietal. the Suzaku observations has to be ascribed to the
(2007)andwasfoundre-analyzingthe2005XMM-Newton behaviorofthe innercolumndensity,the only pa-
observation. Alltheotherparametersareperfectlyconsis-
rameter which significantly changes between the
tentwiththeonepublishedbyPiconcellietal.(2007).
5
Table 2: Best fit parameters for the 2007 XMM-Newton and the four Suzaku observations.
XMM07 S1 S2 S3 S4
Γ (1) 1.92∗ 1.92∗ 1.92∗ 1.92∗ 1.92+0.24
−0.16
Γ (2) 1.92∗ 1.92∗ 1.92∗ 1.92∗ 1.92+0.24
r −0.16
Ne (3) 6.1+3.7 1.9+1.7 <3.0 5.0+2.5 3.9+3.8
H −3.3 −0.8 −2.2 −3.0
Ni (4) 33+4 44+3 68+6 110+14 120 20
A H (5) 9.3−∗5 9.3−∗2 9.3−∗7 9.3−∗11 9.3 ±2.1
pex
A (6) 10.8+1.7 8.7+0.4 11.7+2.0 17+7 14±+9
po −3.3 −1.0 −2.2 −5 −6
E (7) 6.39+0.05 6.419+0.018 6.408+0.016 6.421 0.018 6.411+0.013
σFe (8) <1−700.03 50+−300.021 <−700.018 <±80 <−600.014
Fe −40
F (9) 2.4 0.9 2.5 0.5 2.2 0.4 2.2 0.4 2.4 0.3
Fe
± ± ± ± ±
F0.5−2 (10) 0.32 0.04 0.34 0.08 0.40 0.13 0.42 0.09 0.43 0.08
± ± ± ± ±
F2−10 (11) 7.6 0.4 5.3 0.3 4.1 0.3 3.2 0.2 2.6 0.5
± ± ± ± ±
L2−10 (12) 2.0 0.1 1.5 0.1 2.1 0.4 3.2 1.1 2.9 1.4
± ± ± ± ±
χ2/dof (13) 87/83 114/110 112/111 85/81 91/94
∗ Fixed. (1) Primary continuum powerlaw index - (2) Reprocessed emission powerlaw index (Compton
reflection component and soft X-ray emission) - (3) External column density (1022 cm−2) - (4) Internal
columndensity(1022 cm−2)-(5)Comptonreflectioncomponentnormalization(10−3 phkeV−1 cm−2 s−1)-
(6) Primarypowerlawnormalization(10−3 ph keV−1 cm−2 s−1) - (7) Iron Kα emissionline centroidenergy
(keV) - (8) Iron Kα emission line physical width (eV) - (9) Iron Kα emission line flux (10−5 ph cm−2 s−1)
- (10) Observed 0.5-2 keV flux (10−12 erg cm−2 s−1) - (11) Observed 2-10 keV flux (10−12 erg cm−2 s−1) -
(12) Unabsorbed 2-10 keV luminosity (1042 erg s−1) - (13) Best fit χ2/dof.
NGC7582: Suzaku observation 4 NGC7582: 2007 XMM−Newton observation
0.1
V−1 0.01 V−1 0.05
Counts s ke−1 10−3 Counts s ke−1 0.02
0.01
5×10−3
2 2
χS 2 0 χS 2 0
∆ ∆
−2 −2
1 2 5 10 20 50 1 2 5
Energy (keV) Energy (keV)
Fig. 2.— NGC 7582: Suzaku observation 4 (Left) and 2007 XMM-Newton observation (Right) data and
best fit model.
6
observations (see Table 2 and Fig. 3). The vari- sistent with the one measured by XMM-Newton,
ation (from 7 1023 to 1.1 1024 cm−2) be- around4 5 1022cm−2. Itcanbeidentifiedwith
× × − ×
tween S2 and S3 occurs in roughly 5 months. A a large scale obscuration, as the dust lanes com-
much shorter timescale, the 23 days separating monly observedingalaxies. Indeed, the combined
the first from the second observation, witnesses analysis of HST and Chandra images clearly de-
another significant variation, from around 4.5 to tected such a dust lane also in the X-rays, with
7 1023 cm−2. A stillshortertimescale (less than a column density consistentwith the one required
a×day) characterizes the variation from 3.3+0.4 to by the spectral fits (Bianchi et al. 2007a). In this
−0.5
4.4+0.3 1023 cm−2 measured between XMM07 case, the presence of a second, Compton-thin ab-
−0.2 ×
and S1. In conclusion, our best fit model allows sorber, as invoked in simple modifications of the
us to ascribe most of the observed spectral vari- Unified Models (as in Matt 2000, and references
ability in NGC 7582 to the rapid changes of the therein), is directly observed.
column density of this internal absorber. On the ThesoftX-rayemission,asreportedbyPiconcelli et al.
other hand, there are some hints of variability of (2007) thanks to the well-exposed XMM-Newton
the primarycontinuumintensity,butthey arenot Reflection Grating Spectrometer (RGS) high res-
conclusive, given the large errors. olution spectra, appears dominated by emission
lines of highly ionized species. This is a gen-
4. Discussion eral characteristic of Seyfert 2 galaxies, as found
by Guainazzi & Bianchi (2007). The lack of any
4.1. NGC 7582: the big picture
variability of the soft X-ray emission is in agree-
The fourth and latest Suzaku observation mentwiththescenario,wheretheseemissionlines
caught the source at the lowest state, but this are produced in a large scale material, spatially
indeed allowed us to have a clearer view of the coincident with the Narrow Line Emission (NLR)
reprocessing components of its spectrum, which and likely dominated by photoionizationfrom the
apparently are those of a typical Seyfert 2. The AGN (e.g. Bianchi et al. 2006).
resulting scenario applies well to all other X-ray However, our monitoring campaign discovered
observations of NGC 7582, the different states a striking feature that characterizes NGC 7582:
being due only to the variability of the column the presence of an absorber,whose rapid variabil-
density of the inner absorber. In this section, we ityimposesalocationfarclosertotheBHthanthe
will discuss in detail the implications on the com- torus. While the most rapid variation occurs be-
plex geometry of the absorbers required in this tweenXMM07andS1(lessthanaday),significant
source. variabilityisalsoobservedonlargertimescalesbe-
Theintrinsicnuclearemissionappearsobscured tween the Suzaku observations. The distance of
by a very large column density (just below the the absorber can be roughly estimated with the
‘canonical’ Compton-thick limit). The spectrum followingreasoning,basedonRisaliti et al.(2002).
below 10 keV is therefore dominated by a Comp- We can assume, for simplicity, that the absorb-
tonreflectioncomponentandtherelativeironline. ing material is constituted of individual spherical
Both the flux of the reflection component and of clouds with column density NH and density n. In
the iron line are consistent with being constant this scenario, the variability in the column den-
during the Suzaku monitoringcampaignand with sity that we are observing is due to the complete
the values found in XMM-Newton observations, passage of a given cloud of dimension D NH/n.
≃
the older dating back to 2001. The material that Thetimerequiredforthiscloudtopasscompletely
producesbothcomponentsislikelytobequite far off the line of sight is not larger than t = D/v.
awayfromthenuclearX-rayemittingsource,pos- Assuming that the clouds are rotating at a dis-
sibly inthe classicpc-scale‘torus’invokedin Uni- tance R from the BH with Keplerian velocities,
−1
fication Models (Antonucci 1993). this relation translates into t = D(GMBH/R) 2.
These reprocessing components appears to be Therefore:
obscured by a second absorber, which must be
located farther away. Its column density is not
wellconstrainedintheSuzaku spectra,butiscon-
7
large-scale absorber with column density of a few
1022 cm−2. On the other hand, we stress again
GM t2n2
R= BH that the torus is not along the line of sightin this
N2
H source. This scenario will be generalized in the
M t 2 next Section.
4 1015 BH
≃ × (cid:18)5.5 107M⊙(cid:19)(cid:18)20h(cid:19)
× 4.2. Do we needanew Unification Model?
n 2 N −2
H
cm (1)
(cid:16)109cm−3(cid:17) (cid:18)1023cm−2(cid:19) X-ray observations have collected much evi-
denceinfavorofthepresenceofthepc-scaletorus
where we have normalized all the parameters to envisagedinUnificationModels. Inparticular,the
our fiducial values for the case we are discussing: ubiquitous presence of a Compton reflection com-
the BH mass was estimated by Wold et al. (2006) ponent, invariably accompanied by a neutral iron
as5.5 107M⊙,NH 1023 cm−2 isthedifference narrowKαemissionline,isaclearsignatureofthe
× ≃
observedbetweenXMM07andS1,t 20hoursis presence of Compton-thick material also in Type
≃
the time elapsed between these two observations. 1 objects, even if it does not intercept the line
A crucial value is the electron density: the value of sight (see e.g. Perola et al. 2002; Bianchi et al.
n = 109 cm−3 corresponds to a cloud dimension 2004, 2007b). The distance of the torus is gener-
D = 1 1014 cm, i.e. roughly 10 r for the BH ally inferred from the lack of variability of these
g
×
mass estimate reported above. A larger density componentsbetweenobservationsseparatedyears
would shift the material to larger distances, but one from the other. Therefore, a pc-scale torus
implyingdimensionsofthe cloudssmallerthan10 must be an fundamental ingredient of any Unifi-
r and, therefore, likely smaller than the X-ray cation Model.
g
source they should obscure. On the other hand, However, there is now a growing number of
lower densities would result in a still smaller dis- sources which show dramatic absorption variabil-
tance, which is already an upper limit, given that ityintimescalesasshortashours(e.g.Elvis et al.
the actual crossing time must be lower than the 2004; Risaliti et al. 2005; Puccetti et al. 2007).
separation between the two observations. There These objects cannot be described in the frame-
is a physical limit at this distance, which cannot work of classic Unification Models and they are
be smaller than the dimension of the clouds D. generally considered exceptions of an otherwise
This gives a lower limit for the density, which is successful scenario. However, it is likely that
n>3 108 cm−3. much more sources would present the same char-
×
On the basis of the average unabsorbed X-ray acteristics, if only they were observed with aimed
luminosity of NGC 7582, as derived in this work monitoring campaigns, as we did for NGC 7582.
(< L2−10keV > 2.3 1042 erg cm−2 s−1), we We propose that the simplest scenario that fits
≃ ×
can estimate the radius of the BLR in this source the X-ray observations should consider the pres-
to be around RBLR = 0.5 1 1015 cm (see ence of three neutral absorbers/emitters, even if
− ×
Kaspi et al.2005,fortherelationbetweenthetwo not necessarily coexisting or observable in all the
parameters and the relative uncertainties). More- sources. A Compton-thick torus is likely present
over, a typical density for the BLR is believed to in the vast majority of the sources at a distance
be 109.5 cm−3, orlargerfor the inner regions(e.g. fromtheBHroughlyaroundapc. Itisresponsible
Peterson1997). Therefore,thelocationandphysi- fortheproductionoftheComptonreflectioncom-
calpropertiesoftheabsorbingcloudsinNGC7582 ponent and the neutraliron narrowKα line, both
are consistent with being within or immediately invariablypresentinallX-rayspectraofAGNand
outside the BLR. Note that the sublimation ra- generally found not to vary on timescales shorter
dius in this source is beyond 1017 cm (see e.g. thanyears. Ifthetorusinterceptsthelineofsight,
Barvainis 1987), thus the BLR and the X-ray ab- the observer classifies the object as a Compton-
sorbingcloudsmustbedust-free. Thismeansthat thick Seyfert 2.
these clouds are not responsible for the absorp- On a much larger scale, a Compton-thin ab-
tion of the BLR and the consequent classification sorberwithcolumndensityaround1022cm−2may
of NGC 7582 as a Seyfert 2. This is due to the
8
intercept the line of sight, completely or partially jects (e.g. Matt et al. 2003; Guainazzi et al. 2005;
obscuring the BLR in the optical and absorbing Bianchi et al. 2005; Teng et al. 2008, the alterna-
the X-ray spectrum. The effect of this material, tive being a ‘switching-off’ of the nucleus) and,
likely associated to dust lanes, is to force the ob- definitely, NGC 1365 (Risaliti et al. 2007). The
server to classify the object as an intermediate fractionofCompton-thicksourcesbelongingtothe
Seyfert 1 or a Compton-thin Seyfert 2. two classes basically depends on the covering fac-
To this dual-absorber scenario, basically the tors of the torus and the inner absorber.
same proposed by Matt (2000), a third material Most of the Compton-thin Seyfert 2s with col-
shouldbeadded,onascalemuchshorterthanthe umn densities of the order of 1022 cm−2 do not
torus,roughlywheretheBLRislocated. Thisma- intercept at all the torus along the line of sight,
terialcannotbeseeninCompton-thickSeyfert2s, but are absorbed by large scale dust lanes, which
i.e. those sources absorbed by the torus, because are also responsible for the obscurationof the op-
it is obscured by the torus itself. It is responsible tical broad emission lines. On the other hand,
for fast variability of the absorbing column den- Seyfert 2s with larger column densities, of the or-
sity. From an observational point of view, this der of 1023 cm−2, are likely seen through the ab-
materialcan be effectively discriminated from the sorbing clouds located at the BLR, in analogy to
torus if it is patchy. In this case, given the close the ‘changing-look’ objects cited above, the only
distance to the BH, the chance to see a cloud ap- differencebeingthattheinterveningcloudsarenot
pearinganddisappearingalongthe line ofsightis Compton-thick. These sources are probably the
not low, on short timescales. In the exceptional best candidates for monitoring campaigns, since
case of NGC 1365, a clear case of eclipse from a they are those with higher probability of rapid
cloud is actually observed(Risaliti et al. 2007). If column density variations. NGC 7582 is a clear
the cloud is not Compton-thick, you may still ob- example of this class. Again, the fraction of these
serve Compton-thin Seyfert 2 with large column sources among Compton-thin Seyfert 2s depends
densities (aslargeasseveral1023 cm−2, forexam- onthe coveringfactorandthe geometryofthe in-
ple), likely varying on short timescales. ner absorber.
It is important to stress that such a material Finally,Seyfert1sarethoseobjectswherenone
does not relax our need for a torus. The latter is of the three materials intercepts the line of sight.
needed because nearlyallthe observedAGN have However, depending on the geometry and cover-
a Compton reflectioncomponent and neutral iron ing factor of the inner absorber, they may occa-
Kα line, which do not show significant variabil- sionally show even rapid partial or total occulta-
ity up to quite long timescales. In order to re- tion events. Indeed, this is the case of NGC 4051
produce this observational evidence, the material (Guainazzi et al. 1998), NGC 3227 (Lamer et al.
must be Compton-thick, with large covering fac- 2003)Mrk335(Grupe et al.2007)andH0557-385
tor and, mostofall, quite far fromthe BH,unless (Longinotti et al. 2008). While in some cases an
the nuclear emission remains constant(but this is alternative solution in terms of ‘switching-off’ of
notthecaseformanyAGN).Onlyapc-scaletorus the source is equally viable, in the last two cases
has all these characteristics. the partial covering from an intervening absorber
This scenario (summarized in Table 3) makes is preferred. Evenif the timescales are notpartic-
a number of predictions. Most of the Compton- ularly constraining, the presence of warm absorp-
thick Seyfert 2s are likely still absorbed by the tion signatures unchanged between the two states
torus and are not expected to show any flux or of H0557-385 strongly suggests that the neutral
spectral change on timescales lower than years. absorbing clouds are located close to the BH, in
However,afractionofCompton-thickobjectsdoes agreement with the predictions of the model pro-
not intercept the torus along the line of sight, posedinthis paper. We notehere thatsimultane-
but are caught when a Compton-thick cloud lo- ous optical and X-ray campaigns of these sources
catedin the BLR is passinginfrontof the source. would be revealing for the exact location of the
Such sources may change their status in a follow- absorberwith respectto the BLR,looking for the
ing observation, once the cloud has passed, ex- presence of broad optical lines during the X-ray
plaining some of the so-called ‘changing-look’ ob- absorption states.
9
Table 3: A new Unification Model, based on three absorbers, located at different distances from the BH.
Their presence along the line of sight (highlighted by a √) determines the classification of the object. As
for the torus,itis requiredin allcases,because of the ubiquity ofnot-variablereprocessedcomponentsfrom
Compton-thick material, even if only Compton-thick sources intercepts it along the line of sight. See text
for details.
Classification Dust lane Torus Clouds
(>> pc) (pc) (< pc)
Seyfert 1
‘Changing-look’ Seyfert 1 √
Compton-thin Seyfert 2 √
‘Changing-look’ Seyfert 2 √ √
Compton-thick Seyfert 2 ? √ ?
While the study of the nature of this inner ab-
sorber is well beyond the scope of this paper, we
note here that this scenario is well in agreement
with theoretical models which suggest a strong
link, both geometrical and physical, between the
accretion disk and the BLR, and possibly the
torus itself (see e.g. Elvis 2000; Nicastro 2000;
Elitzur & Shlosman 2006; Elvis 2006). In any
case, we would like to point out that our simplifi-
cationintermsoftwoseparatematerials,onecom-
pact, roughly at a pc from the BH and Compton-
thick(thetorus)andtheotherclosetotheBHand
constitutedbyCompton-thickand/orthinclouds,
can reproduce the observed phenomenology.
5. Conclusions
We have presented a Suzaku monitoring cam-
paign of the Seyfert 2 galaxy, NGC 7582. The
dramatic spectral variability observed during the
4 observations is best explained by changes of the
absorbing column density of a material close to
the X-ray primary source. Given the significant
variation between a new XMM-Newton observa-
tion and the first Suzaku one, separated by only
20hours,itsdistancecanbeestimatedtobeafew
Fig. 3.—NGC 7582: Columndensity ofthe inner 1015 cm, i.e. roughly consistent with the BLR.
×
absorber in the four Suzaku observations (circles) Together with this material, the presence of a
andthelatestXMM-Newton one(triangle). Time Compton-thick materialon larger scale, likely the
bins on the abscissa are 10 days long. ‘torus’ envisaged in the Unification models, is re-
quiredinordertoaccountfortheComptonreflec-
tion component and the neutral iron Kα emission
line, whose fluxes appear constant in years. On
the top of that, a third, Compton-thin material
intercepts the line of sight and can be associated
10