Evidence for an InterMediate Line Region in AGN's Inner Torus Region and Its Evolution from Narrow to Broad Line Seyfert I Galaxies

0.68 MB·English

by  Ling Zhu

Proceedingsof10thAsian-PacificRegionalIAUMeeting2008, 1–4(2008) Printed22January2009 (APRIMLATEXstylefilev1.0)
Evidence for an InterMediate Line Region in AGN’s Inner
Torus Region and Its Evolution from Narrow to Broad
Line Seyfert I Galaxies
9
0
0
2 Ling Zhu1, Shuang Nan Zhang1,2,3, and Sumin Tang4
n 1Department of Physics and Tsinghua Centerfor Astrophysics, Tsinghua University,Beijing 100084, China;
a [email protected], [email protected]
J 2KeyLaboratory of Particle Astrophysics, Institute of HighEnergy Physics, Chinese Academy of Sciences,
5 P.O. Box 918-3, Beijing 100049, China
1 3PhysicsDepartment, Universityof Alabama inHuntsville, Huntsville, AL 35899, USA
4Harvard-Smithsonian Centerfor Astrophysics, 60 Garden St, Cambridge, MA 02138, USA

]
A
G
.
h
ABSTRACT
p
We have decomposed the broad Hα, Hβ and Hγ lines of 90 Active Galactic Nuclei
-
o (AGNs)intoasuperpositionofaverybroadandanintermediateGaussiancomponents
r (VBGC and IMGC) and discovered that the two Gaussian components evolve with
t
s FWHM of the whole emission lines. We suggest that the VBGC and the IMGC are
a produced in different emission regions, namely, Very BroadLine Region (VBLR) and
[
Intermediate Line Region(IMLR). The details of the two components of Hα, Hβ and

1 Hγ lines indicate that the radius obtained from the emission line reverberation map-
v ping normally corresponds to the radius of the VBLR, but the radius obtained from
7 theinfraredreverberationmappingcorrespondingtoIMLR,i.e.,theinnerboundaryof
6 the dusty torus. The existence of the IMGC may affect the measurement of the black
1 holemassinAGNs.Therefore,thedeviationofNLS1sfromtheM-sigmarelationmay
2
be explainednaturallyinthisway.The evolutionofthetwoemissionlineregionsmay
.
1 be related to the evolutionary stages of the broad line regions of AGNs from NLS1s
0 to BLS1s. Other evidences for the existence of the IMLR are also presented.
9
0 Key words: line: profiles; quasars: emission lines; galaxies: structure; Galaxy: evo-
: lution; galaxies: active; galaxies: nuclei
v
i
X
r
a
1 INTRODUCTION wellasthefeeding,fuelingandgrowthofsupermassiveblack
holes.Inthispaperweattempttoaddressbothproblemsin
a coherent way by studying a sample of SDSS AGNs with
TypeIAGNsareoftenclassifiedintotwosubclassesaccord-
strong broad emission lines.
ing to the Full Width at Half Maximum (FWHM) of their
broad Hβ lines, i.e NLS1 and BLS1. There is a tight cor-
relation between the black hole mass and the stellar veloc-
ity dispersion of the bulge of normal galaxies, the so called
2 LINE DECOMPOSITION AND
M-sigma relation (e.g. Magorrian et al. 1998; Tremaine et
STATISTICAL ANALYSIS
al. 2002;). BLS1s are found also to follow this relation well
(Greene&Ho2005). However,NLS1sseem todeviatefrom Thesample contains 90 objects with clear Hα,Hβ, and Hγ
this relation; they seem to have much smaller masses or line profiles that can be decomposed. They are selected by
much higher stellar velocity dispersions (e.g., Wang & Lu La Mura et al (2004) from SDSS 3 based on their balmer
2001; Zhou et al. 2006). We call this under-massive black line intensities. 21 of them are classified as NLS1s. A dou-
hole problem of NLS1s. It is not clear if the luminosity de- ble Gaussian component model, i.e IMGC and VBGC can
pendence of dust torus and broad line region is simultane- fit the broad line very well. Hα, Hβ and Hγ lines are fitted
ously related to the under-massive black hole problem and withtheirFWHMfixedwitheachother;adeviationof10%
therecedingtorusmodelthatusedtoexplaintheluminosity areallowed.Severaldecompositionsamplesarepresentedin
function of AGN (Simpson 2005). Progress towards solving Fig.1.ThestatisticalanalysisarepresentedinFig.2.Hα,Hβ
these problems may shed new lights to the understanding andHγlinesbehavesimilarly.OuranalysisisfocusedonHβ.
the hierarchical evolution of AGN and its host galaxy, as The FWHM ratios range from 1.5 to 4 in the NLS1 group
2 Zhu, Zhang & Tang
80 160 40
6700 SDSHSJβ1344+4416 (a1) 112400 SDSHSJα1344+4416 (a2) 3305 SDSHSJγ1344+4416 (a3)
50 100 25
Intensity234000 468000 1125050
10 20 0
0 0 −5
−10 −20 −10
−2406504700 4750 4800 48504900 4950 5000 5050 −460300 6400 6500 6600 6700 6800 −145200 4250 4300 4350 4400 4450
12 25 6
10 SDSSJ2351−0109 (b1) 20 SDSSJ2351−0109 (b2) 5 SDSSJ2351−0109 (b3)
8 15 4
6 3
Intensity24 150 12
0 0 0
−−42 −5 −−21
4650 4750 4850 4950 5050 −160300 6400 6500 6600 6700 6800 −43200 4300 4400
2205SDSSJ1344+0005 (C1) 5600SDSSJ1344+0005 (C2) 180 SDSSJ1344+0005 (C3)
15 40 6
Intensity150 123000 024
0 0 −2
−5 −10 −4
−1406504700 4750 4800wav4e8l5e0ngth4900 4950 5000 5050 −260300 6400 6w50a0velengt6h6 0A0 6700 6800 −46200 4250 430w0avelen4g3t5h0 A 4400 4450 Figure 2. Statistical analysis of broad Hβ, Hα and Hγ lines.
WiththeFWHMincreasing,FWHMratioofVBGCtothewhole
Figure1.DecompositionexampleofbroadHβ,HαandHγlines.
linebecomessmallerandfinallyreachesunity,theintensityratio
From (a) to (c), FWHM of the line increases, and the IMGC
of the VBGC to the whole line becomes larger and finally also
becomesweakerandmaydisappearsometimes,e.g.,insubpanels
reachesunity,andtheIMGCbecomes broaderandweaker.
c1 & c3. Note that for in subpanel c2, the IMGC of Hα line
is still detectable, in contrast to Hβ and Hγ lines, because the
intermediate component of Hα is normally much stronger than
Hβ andHγ lines. 9
Circles:Black hole mass before correction (a)
Dots: Black hole mass after correction
8.5
which lie on the left of Fig.2(a1). The intensity ratio of the n
u
very broad Gaussian component to the whole line is about Ms
0.6 in these objects. This means that for NLS1s the inter- s) 8
mediate Gaussian component has an intensity comparable s
a
m 7.5
to the very broad one, and consequently the FWHM of the
e
wholelineofNLS1sisdominatedbytheintermediateGaus- ol
h 7
sian component. With FWHM increasing, the intermediate k
c
Gaussiancomponentbecomesweakerandfinallydisappears, a
which can be seen from Fig.2(b1). The whole line can be g(bl 6.5
simply described by a single very broad Gaussian compo- Lo
nentwhenFWHMreachesabout5000kms−1.InFig.2(c1), 6
·
FWHM ratio of the intermediate Gaussian component to
the very broad Gaussian component becomes larger with 5.15000 2000 3000 4000 5000 6000 7000
increasing FWHM. FWHM of the whole Hβ lines km/s
Figure 3. Black hole masses (in units of solar mass M⊙) are
plottedinlogarithmscale.Circlesareblackholemasscalculated
3 BHM CORRECTION
with equation (1). Dots are that after correction using equation
Reverberation-based black hole mass is calculated by the (3).ClearlythecorrectionismoreeffectiveforAGNswithsmaller
FWHM,i.e.,NLS1s.
virial equation
RBLR 2 2
MBH = G f FWHMHβ, (1) oftheinnermostemission lineregion,i.e.theVBLRinour
analysis. Consequently,FWHM of theverybroad Gaussian
whereFWHMHβ istheFWHMofthewholeHβ,whichrep- component, instead of FWHM of the whole line, should be
resentsthevirialvelocityoftheBLR.Foranisotropicveloc-
used to calculate the black hole mass. We therefore correct
ity distribution, as generally assumed, f = √3/2. RBLR is
theblack hole mass in this way:
theBLRradiusobtainedbyreverberationmapping,whichis
relatedtoanAGN’scontinuumluminositybytheempirical RVBLR 2 2 FWHMb 2
MBHb = f FWHM ( ) , (3)
Radius-Luminosity relation (Bentzet al. 2006): G FWHM
where FWHMb is the FWHM of the very broad Gaussian
log RBLR =(0.518 0.039)logλLλ(5100˚A) 21.2 1.7.(2) component,andRVBLR istakenastheRBLR intheR-Lre-
lt days ± erg s−1 − ±
− · lation.ItgivesamoresignificantmasscorrectionforNLS1s
Thisrelationhasbeenstarlight-corrected.Weusethisequa- than for BLS1s as shown in Fig.3. After such correction,
tion to calculate RBLR throughout this work. Because re- NLS1s still have smaller black hole masses but normal ac-
verberation mapping is used to measure the radius of the cretion rate in units of the Eddington rate. Fifteen objects
broadlineregion,naturallyitnormallymeasurestheradius in our samples have velocity dispersion measurement data
Evidence for IMLR 3
(sigma)inShenetal(2008).ThemassesofallNSL1arewell
below,butbecomemoreverycloseto,thepredictionsofthe 18.5
M-sigma relation before and after the correction.The effect Radii of IMLR obtained from Hα lines (a)
of the correction is obvious, albeit small number statistics 18 Radii of IMLR obtained from Hβ lines
dueto thelimited sample. 0.37
er
met17.5
4 THE LOCATION OF THE IMLR AND THE enti 0.51
EVIDENCES FOR ITS EXISTENCE dii) c 17
a
4.1 Broad Line Region evolution R
g(16.5
Weassumethatbothofthetworegions(VBLRandIMLR) Lo
are bounded by the central black hole’s gravity. Then, we
16
have
f12RVVV2BBLLRR = RGV2MBBLHR, (4) 15.542 Radii 4o2f. 5VBLR de4Lr3ivoegd( Lfr5o1m430 .E05q) u aetriogn/s4 34 44.5 45
and
f22RVII2MMLLRR = GRMI2MBLRH. (5) Fmdeiirgniuovseridetyf.4roC.mCircoHlreβrsellaianrteeiostnahneodfV+tBhLefrRoramrdaiHidαiouflsiI,nM×esL.rRCepiarrcenlsdeesnVfitBlIlMeLdRLwRwitirhtahddoiluutss-
These two equations lead to represent the objects missingthe intermediate Gaussian compo-
nent.TheradiusoftheVBLRisobtainedfromtheRBLR∼L5100
RIMLR=(f1VVBLR)2RVBLR =C0(VVBLR)2RVBLR, (6) (equation 2 not equation 3). These correlations supports a sce-
f2VIMLR VIMLR narioofhierarchicalevolutionofAGNs.
where C0 =(ff21)2. C0 is a constant which need to be deter-
mined by further studies. The radius measured from rever-
berationistakenasRVBLR.Therefore,wecancalculatethe
radius of IMLR from the above equation. RIMLR is propor-
softerphotons
tional to C0. C0 = 1 is used for all the calculation below. h~R0.69~L0.25
RIMLR is obtained from Hα and Hβ lines separately. Fig.4 h DustTorus harder
is the evolution of RVBLR and RIMLR with luminosity. The θ IMLR photons
radius of the IMLR increases slower than VBLR with lu- VBLR
minosity and black hole mass increasing. The two emission
regions have a trend to merge into one with larger higher
luminosity and black hole mass. as RIMLR L51000.37±0.06
while RVBLR L51000.52±0.04. ∝
∝
Luminosityincrease: RIMLR~L0.37,RVBLR~L0.52
4.2 The location of IMLR
Figure 5.CartoonoftheBroadLineRegionevolution. Contin-
The relation RIMLR L51000.37±0.06 is consistent with the uumphotonsfromthecentralregionhaveslightlyhigherenergies
∝
receding velocity of torus based on infrared reverberation alongthenearlyedge-ondirection.Withblackholemassandlu-
mapping(Suganumaetal.2006)andtherecedingvelocityof minosityincreasing,boththeVBLRandIMLRexpand. Thera-
torusthatwecancalculatebasedonanalysisoftypeIAGN diusofVBLRincreasesfaster,sothetworegionshaveatrendto
fraction(Simpson2005).Ourotheranalysisalsoshowsthat mergeintoone.
the IMLR may have higher density, more dusty than the
VBGC, and the IMGC show slightly Baldwin effect while
thecorrespondingAGN,butsometimestheyarealmost the
the VBGC does not. The Baldwin effect can be explained
same (Laor 2004; Elitzur 2006). Suganumaet al. (2006).
as a flattened geometry. These results suggest that IMLR
is the inner hot skin of torus. A simple scenario of the
BLR (BLR=VBLR+IMLR) evolution can be constructed
as shown in Fig.5. The inner spherical region is the VBLR
4.3 Evidence for the existence of IMLR
(Very Broad Line Region). It expands to a larger radius
with luminosity increase and perhaps also black hole mass Thefirstconcreteevidencecomesfrom thethestudyofthe
increase. IMLR is the inner part of the torus, which can micro-lensingoftheBroadLineRegioninthelensedquasar
be sublimated by the central radiation and thus its radius J1131-1231(Sluseetal.2007,2008).Inthisworktheyfound
also increases with luminosity increase. When the luminos- evidencethattheHβ emission line(aswell asHα)is differ-
ity is high enough, the broad line region and torus become entiallymicrolensed,withthebroadestcomponent(FWHM
onesingleentity,butwithdifferentphysicalconditions.This 4000 km/s) being much more micro-lensed than the nar-
scenario is consistent with the dust bound BLR hypothesis rowercomponent(FWHM 2000km/s).Becausetheampli-
which also showed that the delay time of infrared emission tude of micro-lensing depends on the size of the emitting
is always longer than the delay time of emission lines of region, this is a clear evidence that the broadest compo-
4 Zhu, Zhang & Tang
rection theNLS1follows M-sigma relation as well asBLS1.
1 1
0.9 response to a δ source 0.9 response to a δ source Thereforetheproblemofunder-massiveblackholeinNLS1s
0.8 0.8 appears to have disappeared. Our results offer strong evi-
spherical shell:R~0.04−0.19 spherical shell:R~0.1−0.25
relative intensity00000.....34567 chchy=y=l0li0inn..1d1d7r7rii c cii=a=al0l1 ss0hheellll:: RR~~00..33−−00..3355 relative intensity00000.....34567 chchy=y=l0li0inn..2d2d5r5rii c cii=a=al0l1 ss0hheellll:: RR~~00..5555−−00..66 dIcMaelnLecRveooalfunttdihoeaneVsvcBoelLnuaRtrio)ionf.roofmthNeLbSr1osadtoliBneLSre1gsioinnt(hceonhsiiesrtasrochfia-
Finally,weshowthatthestudyofmicro-lensingprovide
0.2 0.2
0.1 0.1 concreteevidencefortheexistenceofIMLRandMrk79pro-
00 0.4 0.8 1.2 1.6 2 00 0.4 0.8 1.2 1.6 2 vides possible evidence. Narrower RMS spectrum also sup-
Delay time + 0.1 Delay time + 0.1
1 1 portourmodel.TheexistenceofaweakintermediateGaus-
response to a δ source
0.9 0.9 siancomponentinthebroademission linesofmanysources
relative intensity000000......345678 spchh y=i=el0i n1ri.=i3d0c2 raii0=cl a0shl sehll:eRll~: R0.~205.−70−.04.75relative intensity000000......345678 sRp~h0e.4ri5ca−l0 s.6hell: cRhy=~ l0ii0=n ..i3d1=97r0−2i c00i=a.9l0 s5hell: wtVlieniBmtehLaiRtnrietc,ovIebttriwwbaseoilerlsaGbtfiaeoournhsestmlihpaaenfupmlpcotienomagsdpuemorcneeoemamnsetupnsrotesasmesoeewfnaetctshhhembavrraoeayadddciuoaensumesoeifshssetiyrohsene-,
0.2 0.2 and then do cross-correlation analysis between the contin-
0.1 0.1 response to a δ source uum and each of the two components, in order to measure
00 0.4 De0.l8ay time +1 .02.1 1.6 2 00 0.4 De0.l8ay time +1 .02.1 1.6 2 theRadius-Luminosityrelationsforthetwocomponentsin-
dependently.Thiswouldprovidedecisivetestforourmodel
Figure 6. Different responses of different geometries to a delta presented in this paper, as well as providing more accurate
impulse.TheVBLRisassumedtobeasphericalshellofuniform
measurements of the masses of supermassive black holes in
density, and the IMLR is assumed to be a cylindrical shell of
AGNs.
uniformdensity.Timeisnormalizedtoarbitraryunit.Eachfigure
shows a different RIMLR−RVBLR pair. Clearly in all cases the
responsesofIMLRhaveshortervariabilityandthuswillresultin
narrowerRMSspectrathanthatfromVBLR. ACKNOWLEDGMENT
WeareextremelygratefultoG.LaMuraandL.C. Popovic
for sending us the processed spectra we used in this work.
nentoftheemissionlineismorecompactthanthenarrower
SNZ thanks Jianmin Wang for many discussions, as well
component.
as sharing their early results which motivated us to pursue
Then, we turn to the reverberation mapping experi-
the work presented here. SNZ acknowledges partial fund-
ment.IftheIMLRisreally adiscreteemission region apart
ing support by Directional Research Project of the Chinese
from the VBLR, another peak corresponding to the radius
Academy of Sciences under project No. KJCX2-YW-T03
ofthisregionmightappearinthecrosscorrelation function
and by the National Natural Science Foundation of China
between the lightcurves of the emission line and the con-
undergrant Nos.10521001, 10733010,10725313, andby973
tinuum. In fact, a possible example exists in the database
Program of China undergrant 2009CB824800.
ofreverberationmappingobservations(Petersonetal.1998;
2004).Mrk79showsobviouslydoublepeaksinthecrosscor-
relation centroid distribution which has not been explained
wellsofarwhichcanbewellexplainedinourmodel(please REFERENCES
refer to Zhu et al. 2008 for details).
Bentz,M.C.,etal.2006, ApJ,644,133
Anotherfact knowntousisthattheRMSspectrum of
Collin, S., Kawaguchi, T., Peterson, B.M., & Vestergaard, M.
the broad emission line is normally (but not always) nar- 2006,A&A,456,75
rower systematically than the mean spectrum (Collin et al. Elitzur, M. 2006, ASP Conference Series, Vol.***,Xi’an, China,
2006). If the emission lines are produced from two distinct 16-21October,2006
regions, VBLR and IMLR, they will respond to the central Greene,J.E.&Ho,L.C.2005, ApJ,630,122
continuum radiation differently. Because the IMLR has a LaMura,G.,etal.2007,ApJ,671,104
flattened geometry, it can create a narrower response func- Laor, A. 2004, AGN Physics with the Sloan Digital Sky Survey,
ASPConferenceSeries,311,169
tion to a delta impulse even with radius much larger than
Magorrian,J.,etal.1998,AJ,115,2285
theVBLR,whichisconfirmedbyourcalculations shownin
Peterson,B.M.,etal.1998,ApJ,501,82
Fig.6.
Peterson,B.M.,etal.2004,ApJ,613,682
Simpson,C.2005,MNRAS,360,565
Simpson,C.2005,MNRAS,360,565
Shen,J.J.,etal.2008,AJ,135,928
5 SUMMARY
Sluse,D.,etal2007, A&A,468,885
We conclude that the broad Hβ, Hα, and Hγ line of the Sluse,D.,etal2008, RMxAC,32,83
Suganuma,M.,etal.2006,ApJ,639,46
AGNs in our sample can be well fitted by a double Gaus-
Tremaine,S.,etal.2002, ApJ,574,740
sian model, i.e., VBGC and IMGC. The properties of the
Wang,T.,&Lu,Y.2001,AAP,377,52
twocomponentsindicatethatthetwoGaussiancomponents
Zhou, H., Wang, T., Yuan, W., Lu, H., Dong, X., Wang, J., &
originate from two emission regions, i.e. VBLR and IMLR.
Lu,Y.2006, APJS,166,128
The physical properties of IMLR suggest it should be the Zhu, L., Zhang, S.N., & Tang, S.M. 2008, submitted to ApJ
torusinnerboundaryregion.Basedonourmodel,Theblack (arXiv:0807.3992)
hole mass measured by RM should be corrected, after cor-