Review Article AGN Obscuration and the Unified Model - Hindawi

18 Pages·2014·2.72 MB·English

HindawiPublishingCorporation
AdvancesinAstronomy
Volume2012,ArticleID782030,17pages
doi:10.1155/2012/782030
Review Article
AGN Obscuration and the Unified Model
StefanoBianchi,1RobertoMaiolino,2,3andGuidoRisaliti4,5
1DipartimentodiFisica,Universita`degliStudiRomaTre,ViadellaVascaNavale84,00146Roma,Italy
2INAF-OsservatorioAstronomicodiRoma,ViadiFrascati33,00040MontePorzioCatone,Italy
3CavendishLaboratory,UniversityofCambridge,19J.J.ThomsonAvenue,CambridgeCB30HE,UK
4INAF-OsservatorioAstrofisicodiArcetri,L.goE.Fermi5,Firenze,Italy
5HighEnergyAstrophysicsDivision,Harvard-SmithsonianCenterforAstrophysics,60GardenStreet,Cambridge,MA0218,USA
CorrespondenceshouldbeaddressedtoStefanoBianchi,bianchi@fis.uniroma3.it
Received31August2011;Accepted29December2011
AcademicEditor:IsabelleGavignaud
Copyright©2012StefanoBianchietal. This is an open access article distributed under the Creative Commons Attribution
License,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperly
cited.
UnificationModelsofActiveGalacticNucleipostulatethatalltheobserveddifferencesbetweentype1andtype2objectsaredue
toorientationeffectswithrespecttothelineofsighttotheobserver.Thekeyingredientofthesemodelsistheobscuringmedium,
historicallyenvisagedasatoroidalstructureonaparsecscale.However,manyresultsobtainedinthelastfewyearsareclearly
showingtheneedforamorecomplexgeometricaldistributionoftheabsorbingmedia.Inthispaper,wereviewthevariouspieces
ofevidenceforobscuringmediaondifferentscales,fromthevicinityoftheblackholetothehostgalaxy,inordertopicturean
updatedunificationscenarioexplainingthecomplexobservedphenomenology.Weconcludebymentioningsomeoftheopen
issues.
1.Introduction:TheStandardUnifiedModel forbiddenlines(suchas[OIII],[NeIII],[OII],[OI],[NII],
[SII]), and several very high ionization coronal lines (such
Inthispaper,wediscusstherecentdevelopmentsontheAGN as[FeX],[FeXI],[SiIX],[SiX])iscommontobothtypesof
unified models, specifically for what concern the geometry, Seyfertgalaxiesandwithsimilarlineratios.Thelatterfinding
location, and physics of the absorbing medium. Before suggested that all Seyfert galaxies are powered by the same
discussingthemorerecentresults,inthissectionweshortly intrinsicengine.
review the early, classical arguments that historically led to Astrongobservationalevidenceofaunificationbetween
the formulation of the standard Unified Model. An early type 2 and type 1 Seyfert nuclei has been the discovery
reviewoftheinitialresultswasalsogiveninAntonucci[1]. of broad optical lines in the polarized spectrum of the
The first unification attempts have been focussed on archetypalSeyfert2,NGC1068,obtainedbyAntonucciand
polarization measurements. In particular Antonucci [2] Miller[4].ThisfindingrevealedthepresenceofaBroadLine
found a perpendicular alignment of optical polarization Region (BLR) in this Seyfert 2 nucleus, which is hidden to
relative to the radio axis in a sample of radio galaxies, ourlineofsight,butwhoselightisscatteredinourdirection
whichwasinterpretedasduetoscatteringofphotons,whose frommaterial(inthecaseofNGC1068,likelyfreeelectrons
direction before entering the line of sight was primarily inionizedgas)distributedonscaleslargerthantheabsorber.
in the vertical direction. Shortly after, additional evidence Suchreflectedlightisveryweakcomparedtothelightofthe
was found in low luminosity, local AGN, and specifically galaxy,hencedifficulttodetectinthetotalspectrum,butit
Seyfert galaxies [3]. Seyfert 1 galaxies are characterized by ishighlypolarized,andthereforedetectableinthepolarized
the presence of broad optical permitted lines (FWHM > spectrum.ThebasicideaoftheUnifiedModelisthattype2
1000km/s), such as Hα and Hβ, that are not observed in andtype1AGNareintrinsicallythesameclassofobjectsand
Seyfert2galaxies.However,thepresenceofbothstronghigh theirdifferencesareonlyduetoorientationeffectsrelativeto
ionizationandlowionizationnarrow(FWHM<1000km/s) anobscuringmedium.
2 AdvancesinAstronomy
Theneedfora“toroidal”,axisymmetricstructureofthe defined by the nuclear absorber. High-resolution, narrow
absorber was initially inferred from the fact itself that the bandimaging(orintegralfieldspectroscopy),especiallywith
reflected broad lines are polarized and from the measured theadventofHST,hasindeedrevealedsuchionizationcones
polarization angles. If the absorber was a simple cloud on scales ranging from a few 10pc up to several 100pc, in
along the line of sight, then reflection should come from manynearbyAGN(e.g.,[22–25]).Theopeningangleofthe
all directions, hence by averaging from all angles the total cones gives the fraction of the sky hidden to our line of
polarizationshouldbezero.Inordertobreakthesymmetry sight, which is in reasonable agreement with what inferred
of the polarization angles the absorber should prevent the from the relative fraction between type 1 and type 2 AGN
nuclearlighttobescatteredinasignificantrangeofangles, in the local universe [26]. It is interesting that the axis of
and a “torus” is the most natural configuration that can theionizationconesisoftennotalignedwiththeminoraxis
achievethiseffect. ofthehostgalaxy,meaningthatthecircumnuclearabsorber
Thesizeofthetoroidalabsorberwasinitiallypostulated is not necessarily aligned with the gaseous disk of the host
tobeontheparsecscale[5,6].Suchtypicalsizewassimply galaxy.Generallytheorientationoftheionizationconeaxis
inferred by the need for the absorber to be largeenough to is in the same direction as the radio jet, but often the two
obscuretheBLR,whichinSeyfertnucleihasasizewellbelow arenotexactlyaligned,meaningthatevenonsmall(parsec
a parsec, based on reverberation studies (e.g., [7, 8]), but or subparsec) scales there is a slight misalignment between
smallenoughnottoobscuretheNarrowLineRegion(NLR), the dusty absorbing medium and the central engine (the
whichisdistributedonthe10–100pcscale.However,aswe accretiondisc)[24,27].
will discuss extensively in this paper, there is clear evidence We conclude this introduction by shortly mentioning
thattheabsorbingmediumisalsodistributedonsmallerand that several theoretical works have modelled the physics
largerscales. and the structure of the “toroidal” absorbing medium
SincetheseminaldiscoveryofAntonucciandMiller[4], in the attempt of reproducing the observable properties.
polarized broad lines have been discovered in several other Initialmodelshadassumedasimpletoroidalstructurewith
Seyfert2nuclei[9,10],contributingtothegeneralizationof a uniform distribution of gas and dust with a parsec-
theunifiedmodeltothewholeclassoftype2AGN.Further scale radius (e.g., [6, 28]), while other models suggested
evidenceforaunifiedtheorybetweentype2andtype1AGN more extended geometries, up to 100pc, to explain the
hasbeenobtainedfromvariousmultiwavelengthstudies.In broad infrared spectral energy distribution observed in
a number of type 2 AGN the presence of an obscured BLR AGN [29]. One of the main problems of these models
was inferred from the detection of a broad component of is their dynamical stability. Often radiation pressure from
hydrogen recombination lines in the near infrared, such as infrared photons within the torus is invoked as a solution
Paβ, Paα, Brγ, where dust absorption can be several times to keep the torus geometrically thick [30]. Other authors
lowerthanintheoptical[11–15]. ascribethegeometricalthicknessofthetorustoturbulence
Hard X-ray observations of AGN have provided addi- introduced by supernovae or stellar winds [31, 32]. It has
tional,unambiguousevidenceinfavoroftheunifiedmodel, alsobeenproposedthatawideangleofobscurationdoesnot
obtained already with the early hard X-ray satellites such necessarily require a geometrically thick torus, but can also
asExosat, Ginga,ASCA, and BeppoSAX [16–19]. Although beachievedwithawarpedortilteddisk[33–36].
generally weak or even undetected in the soft X-ray band Recently,thehypothesisofuniformgasanddustdistri-
(<2keV), most type 2 AGN are detected in the hard X- butionhasbeenabandonedbymanymodels,byintroducing
rays (>2keV) and are characterized by a power-law spec- a clumpy structure of the absorbing medium [37–40].
trum similar to Sy1s, favoring a common central engine, These models can account for several of the observational
but affected by a photoelectric absorption cutoff directly properties of AGN and, most importantly, are strongly
demonstrating the presence of an absorbing medium along supported by recent X-ray observations that directly reveal
the line of sight, with a column density typically well in theclumpinessoftheabsorbingmedium,asdiscussedinthe
excess of 1022cm−2. In a number of Seyfert 2s the hard X- followingsections.
ray spectrum does not show the presence of a prominent
absorption cutoff, but it is characterized by a very strong
Fe Kα line at 6.4keV, with an equivalent width larger than 2.FromGalactictoSub-PcScale:Absorptionat
500eV (e.g., [19–21]). Such high equivalent widths of the DifferentScales
FelinecanonlybeexplainedbyassumingthatthedirectX-
rayradiationistotallyabsorbedbyaComptonthickmedium One of the most significant new aspects on the structure
(N > 1024cm−2)andthattheobserved(weak)continuum of AGN, as emerged in last few years, is that the standard,
H
and Fe Kα are due to reflection from the circumnuclear parsec-scale“torus”isnotenoughtoexplainallthecomplex
medium. A detailed discussion of these effects and on the absorptionfeaturesdiscoveredbymanyobservations.While
distributionofthereflectingmediumwillbegiveninthenext the unified picture remains valid in its more general sense
sections. (i.e., the presence of nonspherically symmetric absorbers
IftheNLRintype2AGNiscreatedbythephotoioniza- at the origin of the type 1/type 2 dichotomy) several new
tionofanuclearUV/Xsourcehiddenfromourlineofsight observations and models, mostly in the X-ray and infrared
byatoroidalabsorber,thentheexpectationisthattheNLR domain,suggestthatmultipleabsorbersarepresentaround
shouldhavea(bi)conicalmorphology,duetothelightcones the central source, on quite different physical scales. In the
AdvancesinAstronomy 3
following we review the main observational evidence for key observational parameters of the cloud (Figure2): its
each of them, together with some brief discussion on their covering factor to the X-ray source (suddenly increasing
physicalinterpretation. at the beginning of the occultation, then slowly increasing
overarelativelylongtimeinterval),anditscolumndensity
(highest at the beginning of the occultation, and then
2.1.AbsorptionwithintheSublimationRadius. Theevidence
decreasingsteadily).
for gas absorption within the sublimation radius comes
Occultations have also been observed in the bright
mostly from X-ray observations. The most direct way to
Narrow Line Seyfert 1 Mrk 766 [49], revealing that such
probe the presence of such gas component is through
eventsarepossible(thoughrare)inon-averageunobscured
absorptionvariabilitymeasurements.
sources. While the data quality was not enough to perform
X-ray absorption variability is a common feature in
a column density/covering factor deconvolution as the one
AGN. An analysis of a sample of nearby obscured AGN
describedaboveforNGC1365,additionalinformationcame
withmultipleX-rayobservations,performedafewyearsago
from the detection of highly ionized iron absorption lines.
[41]revealedthatcolumndensity(N )variationsarealmost
H These lines, due to FeXXV and FeXXVI ions, are present
ubiquitousinlocalSeyfertgalaxies.Morerecentobservations
only in the same time intervals where the occultations due
performedwithXMM-Newton,Chandra,andSuzakufurther
to the neutral cloud are also present. Furthermore, their
confirmed this finding. The physical implications of these
energy clearly reveal that the absorbing gas is outflowing
measurementsarethatthecircumnuclearX-rayabsorber(or, with velocities of several 103kms−1. The straightforward
atleast,onecomponentofit)mustbeclumpyandlocatedat
interpretationofthesemeasurementsisthattheabsorption
subparsecdistancesfromthecentralsource.
is due to an outflowing cloud, with a high-density, low-
The comparison between different observations, typi- ionizationhead,andalow-density,high-ionizationtail.
cally performed at time distances of months-years, only Theaboveexamplesremainuniqueamongtheavailable
provides upper limits to the intrinsic time scales of NH X-rayobservationsofAGN,buttheysuggestthatwithfuture,
variations. An improvement of these estimates could only largerareaX-rayobservatories,X-rayabsorptionvariability
be obtained through observational campaigns within a few may become a powerful, relatively standard way to directly
weeks/days, and/or through the search for NH variations measurethephysicalpropertiesoftheabsorbingBLRclouds.
withinsinglelongobservations.Suchshorttime-scalestud- One of the most direct consequences of the presence
ies have been performed for a handful sources: NGC 1365 of gas inside the dust sublimation radius is a decrease
[42–45], NGC 4388 [46], NGC 4151 [47], NGC 7582 [48], of the expected dust-to-gas ratio, as measured from the
andMrk766[49]. ratiobetweenoptical/near-IRreddening,andX-raycolumn
Inparticular,inthecaseofNGC1365,Chandra,XMM- density. This is a well-known observational evidence in
Newton,and Suzaku observations revealed extreme spectral nearby Seyfert galaxies (e.g., [50, 51]), which is therefore
changes, from Compton thin (N in the range 1023cm−2) naturally explained in the context of X-ray absorption by
H
to reflection dominated (N > 1024cm−2) on time scales BLRclouds.
H
from a couple of days to ∼10 hours (see Figure1). Such The presence and structure of “cometary” clouds poses
rapideventsimplythattheabsorptionisduetocloudswith challengesandpossiblesolutionsforlong-standingproblems
velocityv>103kms−1,atdistancesfromtheBHoftheorder in modeling theBLRin AGN. Maiolino etal.[45]estimate
of 104 gravitational radii (assuming that they are moving that the cloud head loses a significant fraction of its mass
with Keplerian velocity around the central black hole). The through the cometary tail, which is expected to cause the
physical size and density of the clouds are estimated to be totalclouddestructionwithinafewmonths.Iftheseclouds
oftheorderof1013 cmand1010-1011cm−3,respectively.All arerepresentativeofmostBLRclouds(oratleastthehigh-
thesephysicalparametersaretypicalofBLRclouds,strongly ionization ones), this implies that the BLR region must
suggestingthattheX-rayabsorberandthecloudsresponsible be continuously replenished with gas clouds, possibly from
for broad emission lines in the optical/UV are one and the theaccretiondisk.Sucha“dynamicalequilibrium”scenario
same. would solve the problem of the long-term stability of BLR
These results are obtained from the analysis of the clouds,forwhichaconvincingsolutionhasnotbeenfound
observed absorption variations in the sources mentioned yet.
above, assuming a simple scheme, where the clouds are A further possible interesting aspect of the “cometary”
homogeneous, with a constant column density and are structure(notyetexploredintheliterature)isthereduction
moving across the line of sight with Keplerian velocity. of the number of clouds needed to explain the smoothness
However,inafewcaseswithparticularlyhighsignal-to-noise of the profiles of the broad absorption lines. High-quality
ratio, the analysis of X-ray “eclipses” can provide further observations of the profile of optical broad emission lines
informationonthegeometricalandphysicalstructureofthe suggest that the minimum number of clouds needed to
cloud. reproduce the observed smoothness is quite high: one of
In the case of NGC 1365, a careful analysis [45] of the most striking examples is that of NGC 4151, with an
the spectral X-ray variability during two eclipses revealed estimated number of at least 108, based on high S/N Keck
a “cometary” shape of the obscuring cloud, consisting of spectra[52].Consideringthattheestimatedblackholemass
a high-density head, and an elongated, lower-density tail. ofNGC4151isoftheorderof107M(cid:3),andthatthedistance
This structure is revealed by the time evolution of the two ofBLRcloudsisoftheorderof104gravitationalradii,ahuge
4 AdvancesinAstronomy
0.1
0.1
−2 0.05
m
c
V −1
nts/sec/ke 0.01 Counts s 0.02
u
o 0.01
C
5×10−3
5
0.001
1 5 0
2
χ
Energy (keV) Δ
−5
OBS 1
OBS 2 (+2days) 1 2 5
OBS 3 (+4days) Energy (keV)
(a) (b)
Figure 1: Spectra of NGC 1365, showing X-ray occultations due to BLR clouds: (a) spectra obtained from three Chandra snapshot
observations performed every two days [43]. The reflection-dominated spectrum in the second observation is due to an occultation by
aCompton-thickcloud(theonlyoneeverdetectedinshorttimescales).(b)Spectraobtainedfromdifferentintervals(lengthofabout10ks)
ofasingleSuzakuobservation,revealingabsorptionchangesduetoacloudwithacolumndensityoftheorderof1023cm−2.
X-ray source
R
x
Cometary
θ<1.2◦
cloud A
VC>3000km/s
NH∼1023cm−2 NH∼31022cm−2
Ltail>2 1013cm
Figure2:StructureoftheabsorbingcloudasobtainedfromaSuzakuobservationofNGC1365.Theestimatesarebasedonthehypothesis
ofKeplerianmotion,andonablackholemassof2×106M(cid:3)[45].Thecloudsizeisnotinthecorrectscale:thetailismuchlongerwhen
comparedwiththesourcesize,whichisoftheorderofafew1011cm.
clouddensityisobtained,withanearlycompleteoccupation which is in perfect agreement with the value measured for
of the available volume. If instead every single cloud has the broad component of the Hα in a simultaneous optical
a cometary structure, analogous to that inferred for NGC observation [56]. Moreover, the observed equivalent width
1365, each cloud would contribute to the observed profile (EW) of the iron line is in agreement with an origin in
withasmall,butnotnull,width,whichwouldgreatlyreduce the BLR, under reasonable assumptions on the geometrical
the total number of clouds in order to reproduce the same distribution of the clouds, their covering factor, and their
smoothness. columndensity[56].Therefore,NGC7213istheonlySeyfert
Finally, if the covering factor and the optical depth of 1 galaxy whose iron Kα line is unambiguously produced in
the BLR are large enough, a significant fraction of the iron theBLR.Itisdifficulttoreachsimilarconclusionsforother
Kα line should be produced there. In order to test this objects, because of the presence of the Compton reflection
hypothesis,NGC7213representsauniqueopportunity.The component never allows us to exclude the contribution,
X-rayspectrumofthissourcehasnoevidenceforCompton from the more extended torus, and because it is difficult to
reflection,auniqueresultamongbrightSeyfert1s[53–55]. measure the FWHM of the iron line and compare it with
TheobservedneutralironKαline,therefore,cannotbepro- theopticallines(seealsonextsection).Inthefuture,high-
ducedinaCompton-thickmaterial,likethediscorthetorus. resolution X-ray spectroscopy with microcalorimeters and
Indeed, the iron line is resolved in a Chandra High-Energy X-ray reverberation studies will be extremely powerful in
TransmissionGratingobservation,withameasuredFWHM tacklingthisissue.
AdvancesinAstronomy 5
2.2. Absorption form Pc-Scale Tori. Early evidence for a between type 1 and 2 sources and the size of the dusty
circumnucleardustymediumonparsec,orsubparsecscales, emitterscaleswiththesquarerootoftheluminosity[73–75].
asinitiallyinvokedbytheUnifiedModel,wasobtainedfrom Comparisons with tori models suggest that, in principle, it
near-IR studies, which revealed the presence of very hot wouldbepossibletodisentanglebetweenface-onandedge-
dust, close to the sublimation temperature, in the nuclei ondistributionsbycomparingthecompactnessofthedusty
of Sy1s [57–59]. The dust sublimation radius in Seyfert structure to the AGN luminosity, but uncertainties on the
nuclei is on subparsec scales and on parsec scales at quasar observedmeasuresarestilltoolarge.
luminosities.Thesubparseclocationofthehotdustemitting The presence of Compton-thick neutral material with
inthenear-IRhasbeenconfirmedbyextensivereverberation large covering factor in the environment of AGN is also
observational campaigns [60], which also confirmed the supportedbytheubiquitouspresenceoftheironKαlineand
expected L1/2 dependence of the sublimation radius. The the Compton reflection component in the X-ray spectra of
coveringfactorofthecircumnucleardustymedium,inferred Seyfert galaxies (e.g., [54, 76, 77]). Although a component
from the near-IR observations, is very high in most of broadened by strong gravity effects arising in the accretion
the nearby Sy1s (exceeding 0.8: [61, 62]), and generally in disk is observed in at least a third of the sources (see, e.g.,
agreementwiththeobservedtype2/type1ratio[26]. [78, 79]), a narrow core of the iron line is a much more
RadioVLBIobservationswerethefirstonestoeffectively commonfeature.Theline,typicallyunresolved(withupper
image the AGN circumunclear medium on parsec and limits of several thousand km/s for its FWHM), must be
subparsec scales. Greenhill et al. [63] obtained VLBA water producedfarfromthenucleus,eitherintheBLR,thetorus,
maserimagesatsubparsecresolutionofNGC1068,revealing or the NLR. Apart from single exceptional cases (like NGC
arotatingwarpeddiskstructure.Thewarpingofthemaser 7213:seeprevioussection),currentX-raysatellitesallowus
disk orientation may indicate that the inclination of the toresolveitsFWHMonlyinafewobjectsandwithlimited
nuclear molecular disk is likely responsible for the large information, generally leading to inconclusive estimates
coveringfaction,ratherthanageometricallythicktorus[36]; on the location of the material producing the lines (see,
however, one must also take into account that the maser e.g., [80, 81]). Future X-ray missions taking advantage of
emission does not necessarily trace the global morphology microcalorimeterswillrepresentabreakthroughinthiskind
ofthecircumnuclearmoleculargas,butonlytheequatorial of analysis, allowing us to deconvolve all the components
edge-on medium (within ∼15◦ from the line of sight) possiblypresentintheironline,asroutinelyperformedfor
where maser amplification is highest. VLBI water maser theopticallines.
observations were subsequently obtained for other AGN, However, clues in favour of a parsec-scale distance
finding similar structures [64–67]. Gallimore et al. [68] of the material producing the narrow iron line and the
obtained VLBA images of the nuclear continuum radio Comptonreflectioncomponentalreadycomefromthelack
emission of NGC 1068. Beside the nuclear nonthermal of variability of these features. X-ray spectra of Compton-
emission, tracing the nuclear engine, they resolved two thick sources (i.e., obscured by a column density larger
symmetricradiofree-freeblobsataradiusof∼0.3pc,which than (cid:5)1024cm−2) are completely dominated by reflection
were interpreted as the inner ionized edge of the obscuring features, and they typically do not show any variability
torus. even on long time scales. This is particular clear in sources
The possibility to effectively “image” the dusty compo- wherethecentralenginefadesawayforalongtimeinterval
nent of the parsec-scale torus has become possible in 2004, (years),whilethereflectioncomponent(includingtheFeKα
when mid-infrared interferometry allowed Jaffe et al. [69] line) remains stable over the same time scale [82, 83]. This
to map for the first time the dust at parsec resolution in suggests that the obscuration/reflection occurs on (at least)
theSeyfert2,NGC1068.Theirresults,refinedbyfollowing pc-scale,likethestandardtorusenvisagedintheUnification
observations[70],areconsistentwithatwo-componentdust Models and mapped by interferometry. In principle, the
distribution:aninner(0.5pcofthickness),ratherelongated geometry and distance of the torus could be estimated
hot(T > 800K)component,andamoreextended(3-4pc), by doing accurate X-ray reverberation analysis of the iron
lesselongatedcolder(T (cid:5)300K)component.Thecompact line and the Compton reflection component, in order to
component is coincident, in size and orientation, with the take into account in detail how the material reacts to the
nuclear water maser. Most of the absorption appears to be intrinsicvariabilityofthecentralsource.Unfortunately,such
located outside 1pc. A similar result was found, with the astudyisextremelydifficultanduncertainwithcurrentX-
same technique, for another Seyfert 2, Circinus: again two raymissions.
components, an inner and more compact (0.4pc), and an WhileX-rayabsorptionvariabilitystudieshavedelivered
outer (2pc) component [71]. However, the temperature of exquisite information on the structure of the absorbing
theinnercomponentinCircinus(T =330K)issignificantly medium on scales of the BLR, in particularly for what
lower than in NGC 1068, and far from the sublimation concerns the clumpiness of the absorber, the same kind of
temperature(seeFigure3).Thefirstobservationcarriedon analysisis difficult to performon theparsec scaleabsorber,
atype1object,NGC4151,ledtoresultsinagreementwith both because the time scale for variation is much longer,
thepreviousonesonSeyfert2s[72]. and because it is totally overwhelmed by the variability
When such interferometric studies were performed on introducedbythecloudsintheBLR.However,asmentioned
a sizeable sample of objects in mid- and near-infrared, in the introduction, recent models have shown that even
it was observed that no significant differences are found for the dusty pc-scale torus a clumpy structure (Figure4)
6 AdvancesinAstronomy
75 DEC in mas
Disk Ionisation cone
Clumpy masers (visible,θi=90◦)
torus
50
Outflow
masers Outflow
25 (θm=120◦)
RA in mas
75 50 25 −25 −50 −75
−25
Coneedge
in H2and Hα
1.9pc −50
0.4pc
Disk
Ionisation cone
−75
(obscured)
(a) (b)
Figure3:TheCircinusGalaxy(fromTristrametal.[71]).(a):Sketchofthebest-fitphysicalmodelwhereawarm(T (cid:5) 330K)emission
region(yellow)issurroundedbyacooler(T(cid:5)300K)one(brown).Thewatermaseremittersareover-plotted(blueandredlines),together
withtheionisedcone(dashedanddottedlightblueline).(b):thedustytorusasderivedfromtheMIRinterferometricobservations.
canbetteraccountfortheinfraredobservationalproperties To observer
i
[38, 84–86]. In particular, the very broad infrared Spectral
EnergyDistribution(SED)ofAGNrequiresdustatmultiple
temperature, which is hardly achieved by models with a
compact(pc-scale)uniformtorus[28].Large-scale(100pc)
dusty torii [87] can reproduce the broad IR SED, but can
be hardly reconciled with the small sizes observed in mid-
IR interferometric observations. A pc-scale, but clumpy
σ
torus can at the same time match the observed mid-IR size
and reproduce the broad range of dust temperatures, since
within each dense clump dust does span a wide range of
temperatures. A crucial test of these models will be feasible
with ALMA, which will allow us to image the cooler dust Rd Ro
thermal emission of torus at sub-mm wavelengths at sub-
pcresolution.Indeed,asdiscussedinMaiolino[88],aclear Figure 4: Sketch of the dusty torus as in the model of Nenkova
prediction of this model is that the morphology of the et al. [38], which can reproduce the observed dust temperature
“torus” at far-IR/sub-mm wavelengths (tracing cold dust) distributionsderivedfrominfraredobservations.
should be very similar to the morphology observed in the
mid-IR(tracingwarmdust).Anindependentobservational
indication of the clumpiness of the dusty absorber is, as
the optical reddening associated with the medium in the
discussedinShietal.[89]andinNikuttaetal.[90],thelarge
scatterofthedepthofthe9.7μmsilicateabsorptionfeature hostgalaxy,whichisthereforeanadditionalingredientthat
mustthereforebeconsideredinaglobalUnificationModel
(directlytracingdustabsorptionatmid-IRwavelengths)asa
(see,e.g.,[26,91]).Earlyevidenceofobscurationon100pc
functionoftheX-raygaseouscolumndensity,aswellasthe
scalesbythehostgalaxygaseousdiskcamefromthefinding
findingthatthesamefeatureisobservedinemissioninsome
that optically selected AGN samples tend to avoid edge-on
type2AGN.
systems[26],aresultwhichhasbeenconfirmedandrefined
with much higher statistics by using the SDSS survey [92].
2.3. Absorption by Gas in the Host Galaxy. Although clear Furthermore, it has been suggested that the gas in the host
evidenceinfavourofabsorptionfromtheBLRandthepc- galaxydiskcanpartiallyobscurealsotheNLR.
scale“torus”hasbeenpresentedintheprevioussections,in Furtherdirectevidenceforobscurationon“large”scales
some cases the lowest column densities are consistent with wasobtainedthroughhigh-resolutionHSTimages,showing
AdvancesinAstronomy 7
that dust lanes at distances of hundreds of parsecs are weakAGN,possiblyincapableofgeneratingaclassicalBLR
very common in Seyfert galaxies [93]. The presence of [9,10].
these structures is correlated with Compton-thin X-ray In a model proposed by Nicastro [106], the creation of
obscuration,evenifnotnecessarilybeingdirectlyresponsible the BLR is connected with disk instabilities occurring in
of the obscuration of the nucleus [94]. In some cases, the proximity of a transition radius at which the accretion disk
effectofdustlanescanbeseendirectlyasX-rayobscuration changesfromgas-pressuredominatedtoradiation-pressure
towardsthesoftX-rayemissionfromtheNLR(e.g.,[95])see dominated.Sincethistransitionradiusbecomessmallerthan
Figure5. theinnermoststableorbitforverylowaccretionrates(and
Interferometric maps of the molecular gas distribution thereforeluminosities),veryweakAGNareexpectednotto
haveprovidedadditionalevidenceforlargeamountofdense be surrounded by a BLR. More recently, the disappearance
gas on the 100pc scale surrounding AGN, which certainly of the BLR in low-luminosity AGN has been predicted by
contributes to the obscuration of the central engine along ElitzurandHo[107]andTrumpetal.[108]withinadisk-
some line of sights (e.g., [96–98]). Certainly, within this windscenario,wheretheBLRisembeddedinanoutflowing
context,theadventofALMAisgoingtobeabreakthrough wind, which is no longer supported by the disk below a
byprovidingdetailedmapsofthemoleculargasdistribution criticalvalueoftheEddingtonratio.
inthecircumnuclearregionofmanyAGN. If the BLR cannot form in weakly accreting AGN, we
As it will be discussed further below, there is evidence expect the existence of “true” Seyfert 2 galaxies, that is,
that the role of absorption from gas in the host galaxy optically classified Type 2 objects, without any evidence
becomes increasingly important in high-z AGN [99, 100], of obscuration of their nuclei. Such unabsorbed Seyfert 2
likelyasaconsequenceofthehighergascontentandhigher galaxies do exist, and the best examples (where the lack
starformationinhigh-zAGNhosts. of the optical broad lines and of the X-ray obscuration
Itisworthremindingthattheobscurationoccurringon are unambiguously found in simultaneous observations
such large scales is limited by dynamical mass constraints. with high SNR) have low Eddington rates: NGC 3147
Risalitietal.[101]showedthatCompton-thickgasmustbe (logL /L (cid:5)−4:[109]),Q2131427(logL /L (cid:5)−2.6:
bol Edd bol Edd
containedonscalessignificantlysmallerthan100pc,inorder [110]), and NGC 3660 (logL /L (cid:5) −2: [111, 112]).
bol Edd
not to exceed the dynamical mass in the same region and WhenasizeablecatalogueofX-rayunobscuredradio-quiet
haveacoveringfactorlargeenoughtoaccountforthehigh AGN is analysed, these sources are among those with the
number of observed Compton-thick sources. This in turns lowest accretion rates (e.g., [113]). It is interesting to note
means that the bulk of the ubiquitous Compton reflection that, although lacking the BLR, these objects do appear to
componentandnarrowneutralironKαlinemustalsocome have a face-on Compton-thick torus, as type 1 sources, as
from a compact region. However, in NGC 1068 hard X-ray evidenced by the presence of silicate emission features in
emission (mostly associated to the reflection component) their IR spectra [114], and of a neutral iron emission line
and the neutral iron line are seen extending up to (cid:5) 2 kpc [109].
fromthenucleus[102].
The existence of a critical threshold in luminosity
and Eddington rate has been confirmed on observational
grounds(e.g.,[113,115–118]).Belowthisthreshold(whose
3.OpenIssues
exactvaluedependsontheadoptedsample,andthemethods
to derive the bolometric luminosities), no broad lines are
3.1. “True” Seyfert 2s. As discussed in Section1, one of
detected (either in total or polarized light). On the other
the most convincing pieces of evidence in favor of the
hand, it is clear that above the threshold the BLR still
Unification Model is the detection of broad optical lines in
cannot be detected in many sources (see Figure6). If the
the polarized spectra of type 2 AGN. However, about half
scenarioproposedbyNicastro[106]iscorrect,thesesources
ofthebrightestSeyfert2galaxiesappearnottohavehidden
should possess a BLR, so there must be something that
broad-lineregionsintheiropticalspectra,evenwhenhigh-
prevents us from observing it. Indeed, all these sources
qualityspectropolarimetricdataareanalysed[9,10].
are Compton-thick, so the nucleus is severely obscured by
A number of these cases may be associated with nuclei
intervening absorbers. It was suggested that this could be
where the mirror reflecting the broad lines either has very
explained within the framework of standard Unification
lowscatteringefficiency(eitherduetolowcoveringfactoror
Models,wherebymoreinclinedsources(withrespecttothe
lowcolumndensity)orisobscured[103].Evidencewasalso
line of sight) should intercept a larger column density of
obtained that the lack of polarized broad lines is associated
the torus and may obscure the medium responsible for the
with a stronger contribution/dilution from the host galaxy
scattering of the BLR photons (e.g., [117, and references
orfromacircumnuclearstarburst,makingthedetection of
therein]).
polarizedbroadlinesharder[104,105].
However, a number of Sy2s without polarized broad
lines may be genuine type 2 Seyferts, in the sense that 3.2.Disk-TorusAlignment. Althoughnotexplicitlyrequired
theyintrinsicallylackaBLR.Indeed,observationalevidence by any Unification Model, the most natural assumption on
suggeststhatSeyfert2swithpolarizedbroadlinesaremore the geometry of the circumnuclear matter in AGN is that
easily associated with truly obscured Seyfert 1 nuclei, while it is coaxial with the spin of the BH. This expectation is
Seyfert 2s without polarized broad lines preferentially host basedonanangularconservationargument:iftheobscuring
8 AdvancesinAstronomy
NGC7582–NIR/optical ratio NGC7582–0.8–1.3/0.3–0.8 keV ratio
1 2.8 4.5 6.2 8 0 0.8 1.5 2.2 3
pc pc
1000 500 0 −500 −1000 −1500 1000 0 −1000 −2000
2000
10
1000
10
1000
5
500
ds ds
n n
o o
c sec 0 0 pc c sec 0 0 pc
Ar Ar
−5 −500 −1000
−10
−1000
−10
−2000
−20
5 0 −5 −10 −15 10 0 −10 −20
Arc seconds Arc seconds
Center: R.A. 23 18 23.6 Dec−4222 13.5 Center: R.A. 23 18 23.6 Dec−4222 13.5
Figure5:TheSeyfert2galaxyNGC7582(fromBianchietal.[95]).(a)HST NIRtoopticalratio,mappingtheamountofdustinthe
circumnuclearregionofNGC7582.ThewhitecontourplotsrefertotheChandraemissionbelow0.8keV,whiletheredonestheratioshown
intherightpanel.(b)Chandra0.8–1.3to0.3–0.8keVratio.Theredcontourplotsshowninleftpanelrefertothisimage,butthescaleis
different,asoutlinedbythebroken-linerectangle.Inbothpanels,theblackcrossindicatesthepositionofthenucleus.
torus is related to the inflowing material, it is natural to Surprisingly, the results obtained by Raban et al. [70]
expect that the torus, the accretion disk, and the black strongly suggest that the two structures are misaligned in
hole rotation (mostly due to the angular momentum of NGC 1068 (Figure7). Moreover, the direction of the radio
theaccretingmaterial)sharethesameaxis.Thishypothesis, jet is also clearly tilted with respect to both the NLR and
though reasonable, may not be verified in several more the torus. Some of the discrepancy can be solved by taking
complexscenarios.Forexample,iftheBHgrowthisdueto into account the detailed kinematics of the outflow, and by
multiple,unrelatedaccretionevents,theactualBHspinmay assuming a clumpy torus that could prevent the ionization
not reflect the rotation axis of the accretion disk. Another of all the gas present in the geometrical opening angle of
possibility is that the obscuring torus is not within the the ionization cones. However, these solutions are not in
gravitational sphere of influence of the BH (e.g., a galactic agreement with the results inferred from the appearance of
dust lane). In this case, no obvious physical relation is theconesintheinfrared[70,andreferencestherein].Similar
expected between the torus axis and the BH spin. If the analysisonothersourcesisclearlyneededinordertoshade
torus-BH spin alignment hypothesis is accepted, it implies somelightonthisissue.Apromising,independent,method
that the accretion disk is aligned with the obscuring torus. to test the torus/ionization cones misalignment is via X-
Any radio jet should be aligned to the same axis. Finally, ray polarimetry, but we will have to wait for a future X-ray
theorientationandtheopeningangleoftheNLRionization missionequippedwithabroadbandpolarimeter[121].
conesarecollimatedbytheinnerapertureofthetorus,thus It is more difficult to estimate the inclination angle of
beingthemselvescoalignedwiththecommondisk/torusaxis. theaccretiondisk.Onepossiblemethodisviatherelativistic
These expectations are very difficult to test, due to the very profileofironemissionlinesproducedintheinnerregions
smallscalesoftheinnerregionsofAGN.Whenlargerscales of the disk, which is expected to be strongly dependent on
areimaged,atleasttheradiojetandtheoptical/X-rayNLR theinclinationangle(e.g.,[122]).Whensizeablesamplesof
appeartobegenerallyinagreementwiththissimplepicture AGN are systematically analysed, it appears that a simple
(e.g.,[119,120]). relation between the inclination of the nuclear obscuring
The mid-infrared interferometric studies described in matter (as measured by the optical type) and that of the
Section2.2 allowed us for the first time to directly image accreting matter should be ruled out, in contrast with the
thegeometryofthetoruswithrespecttotheopticalcones. naive expectations from Unified Models [123]. However, as
AdvancesinAstronomy 9
45 againnotcompatiblewiththepresenceofatoruscoaligned
with the accretion disk, unless the torus covering factor is
extremelysmall.
44
Finally, evidence of a misalignment between disk and
radiojetforafewindividualsourcescomesfromtheanalysis
)
V 43 ofthejet-diskrelationinobjectswherethediskinclination
e
k
0 canbeinferredfrommaseremission.Aprominentexample
1
2– ofthiskindofstudiesisthatofNGC3079[67].
L ( 42
g
o
L
3.3. Luminosity and Redshift Dependence of the Covering
41
Factor. An interesting finding of the recent years has been
the evidence for a dependence of the obscuring medium
40 covering factor on the AGN luminosity. More specifically,
−4 −3 −2 −1 0 the covering factor of the obscuring medium appears to
Log(LBol/LEdd) decrease significantly with luminosity. This effect has been
HBLR compton thin HBLR compton thick shown quite clearly by various hard X-ray studies [125–
130] and by optical surveys [131], which have measured
Non-HBLR compton thin Non-HBLR compton thick
the relative fraction of obscured and unobscured AGN as a
function of the bolometric luminosity. Clearly these works
Figure 6: The separation between Seyfert 2 with detection of are affected by various uncertainties and caveats, primarily
polarizedbroadlines(hiddenBLR,HBLR)andSeyfert2swithout relatedtopossibleincompletenesseffectsandbiasesthatmay
polarizedbroadlines(non-HBLR)isclearbothinX-rayluminosity
preventtheidentificationofobscuredAGNinmoredistant
and Eddington rate, when only Compton-thin sources are taken
galaxies (hence more luminous, as a consequence of the
into account. However, a number of Compton-thick Non-HBLR
Malmquistbias).Indeed,theseresultshavebeenquestioned
sources have high accretion rates and luminosity, suggesting that
bysomeauthors[132–134].However,morerecentextensive
obscuration also plays an important role in the detection of the
hardX-raysurveyshavefurtherconfirmedacleartrendfor
HBLR(fromMarinuccietal.[113]).
a decreasing fraction of obscured AGN at high luminosity
[135–138].Figure8showstheresultsofsomeofthesestudies
[135],whereblackandbluesolidsymbolsshowthefraction
Jet of obscured AGN as a function of X-ray luminosity, while
0.5 pc
red-dotted symbols illustrate the same trend from optical
surveys[131].
The luminosity dependence of the covering factor has
been recently questioned by Lawrence and Elvis [36],
showing that, at least in optical and IR-selected sample,
the luminosity dependence is largely a consequence of
the inclusion of low excitation AGN (which may have
an intrinsically different engine) and on the definition of
“obscured” AGN (i.e., whether including mildly obscured
AGN or not). However, the luminosity dependence of the
covering factor still persists in the X-ray-selected samples,
regardlessoftheclassificationscheme.
Analternativemethodtoinvestigatethecoveringfactor
oftheobscuringmediumisbymeansofthedustreprocess-
ing by the obscuring medium. More specifically, the ratio
Figure7: Sketch ofthe geometry ofthe circumnuclear matter in
betweenthehotdustemissionobservedinthenear/mid-IR
NGC1068:compactdustinferredfrommid-infraredinterferome-
and the primary AGN bolometric emission (optical/UV/X-
try(red),ionizationconesasmappedbythe[OIII]emission(blue,
reducedinscalebyafactor∼100)andasinferredfromspectroscopy ray), responsible for heating the dust, is proportional to
(yellow).FromRabanetal.[70]. the covering factor of the obscuring medium. This method
can only be applied to type 1 AGN, where the primary
optical/UV radiation is directly detectable. By using ISO
and Spitzer data to trace the dust emission, various studies
noted by the same authors, the inclination angles derived have confirmed that the covering factor of the absorbing
from the profile of a relativistically broadened line are still medium decreases as a function of luminosity, as shown
affected by large systematic uncertainties. Another method in Figure8 [62, 139, 141, 142] (but see also [143]). These
wasrecentlysuggestedbyRisalitietal.[124],whoanalysed IR results obtain a covering factor significantly higher than
the distribution of the equivalent widths of the [OIII] X-ray studies, but this is likely due to the population of
emission line in a large sample of AGN. Their results are Compton-thickAGNthatismostlylostbyX-raysurveys.
10 AdvancesinAstronomy
1 extension of a few hundred pc. The covering factor of this
disc diminishes as the gravitational pull from the central
SMBH and the bulge increases with the BH mass (and,
therefore, the luminosity), producing the observed anticor-
0.8
relation.However,thismodelcanexplaintheanticorrelation
onlyforCompton-thinsources.
Another possible scenario is that the lower covering
n 0.6
o factor in luminous AGN (quasars) is simply a consequence
cti
a of the stronger AGN radiation pressure impinging onto
e 2 fr thecircumnuclearmediumandexpellinglargerfractionsof
p
Ty 0.4 material. In support of this scenario growing evidence for
massiveoutflowsinluminousAGNhasbeenreportedinthe
recentyears[151–155].
Some studies have also claimed an evolution of the
0.2
coveringfactorwithredshift.Morespecifically,byusinglarge
samplesofX-rayselectedobjects,LaFrancaetal.[127]and
Hasinger [135] have found evidence for a strong increase
0 of the fraction of obscured AGN from z = 0 to z =
42 43 44 45 46
2. This result is however more debated, due to the larger
Log(Lx) (erg s−1) uncertaintiesindisentanglingluminosityandredshifteffects
in flux-limited samples and also more critically subject to
Figure8:Fractionoftype2AGNasafunctionofX-rayluminosity selectioneffectsinhigh-zobjects.Indeed,otherauthorsdid
(fromHasinger[135]).ThesolidsymbolsshowhardX-ray-selected
not find significant redshift dependence [125, 140], while
AGN (blue and black symbols illustrate different subsamples in
Ballantyne et al. [156] and Treister and Urry [157] found
terms of redshift completeness). The red-dotted symbols show
a shallow redshift dependence of the fraction of obscured
optically selected AGN (from Simpson [131]). The magenta line
AGN.Hereweonlymentionthatanincreaseofthefraction
shows the type 2 fraction inferred from the dust covering factor
obtainedthroughnear/mid-IRobservations(fromMaiolinoetal. of obscured AGN at high redshift is naturally expected by
[139],notethattheseincludealsoCompton-thickAGN,whichare the larger gas content found in high-z galaxies [158, 159]
not present in X-ray samples). The green-dotted line shows the and associated with the enhanced star formation rates in
expected fraction of Compton-thin type 2 AGN according to the high-z galaxies, also observed in high-z AGN host galaxies
X-raybackgroundsynthesismodelsgiveninGillietal.[140]. [160,161].
Apossibleadditionalcomplicationtothesimpleversion
of the Unified Model is that the covering factor of the
obscuringmediummaynotonlydependontheAGNlumi-
Inasimilarmanner,theequivalentwidthofthe(narrow)
nosity,andpossiblyonredshift,butmayalsobeintrinsically
Fe Kα line has been used as a proxy of the covering factor
different between type 2 and type 1 AGN. Ricci et al. [162]
ofthecircumnuclearmaterialresponsibleforproducingthe
have shown that the X-ray cold reflection “hump” in type
Fe Kα line relative to the primary X-ray continuum. The
2 AGN is stronger than in type 1 AGN, suggesting a larger
EW of the Fe Kα has also been found to anticorrelate with
coveringfactorofthecircumnuclearmediumintheformer
luminosity (“X-ray Baldwin effect” or “Iwasawa-Taniguchi
class.Thisresultis,however,atoddswiththoseobtainedby
effect”),atrendwhichwasgenerallyinterpretedintermsof
mid-IRinterferometrystudies,asdiscussedabove.Moredata
decreasing covering factor of the circumnuclear absorbing
arecertainlyrequiredtotacklethediscrepancy.
mediumasafunctionofluminosity[144–148].
Therefore, although partly questioned, several indepen-
dentobservationalstudiesstronglyfavoraluminositydepen- 3.4.UnusualGeometries. Whilethegeneralpicturediscussed
denceofthecoveringfactorofthecircumnuclearabsorbing in this paper can account for most of the observational
medium. The origin of the anticorrelation between lumi- properties observed in AGN, some subsamples require
nosity and covering factor is unclear. The “receding torus” differentgeometries.Thisisthecaseforanumberofnuclei
scenario[149]isofteninvokedtoexplainthistrend:higher that are unambiguously hosting a relatively powerful AGN,
luminosities imply a larger dust sublimation radius and, if based on their X-ray or mid-IR properties, but which do
the torus has a constant height as a function of radius, this not show any evidence for classical NLR tracers in their
results into a smaller covering factor of the dusty medium. optical spectra [163–173]. A possibility is that in these
However, this scenario cannot explain the results on the objects the NLR, although extended, is heavily obscured by
decreasingcoveringfactorinferredfromX-raystudies,which absorbing medium distributed on large scales in the host
donottracethedustycomponentoftheabsorber. galaxy.Indeed,inanumberofobjectsSpitzer spectroscopic
In the model proposed by Lamastra [150], the covering observationshaverevealedmid-IRhighexcitationlines(e.g.,
factor-luminosityanticorrelationnaturallyarisesiftheX-ray [NeV]14.3μm)typicaloftheNLR[174–176].
obscuration in Compton-thin sources is due to interstellar However,anumberofactivenucleidonotshowNLR-like
gas, distributed in a rotationally supported disk with an lineseveninthemid-IR[177].Inthesecasesalikelyscenario