The Unification Model
of Active Galaxies:
Implications from
Spectropolarimetric Surveys
Hien D. Tran
W. M. Keck Observatory
Optical Spectra of Various Types of AGNs
Bill Keel
2
Seyfert 1 & 2 Optical Spectra
Narrow forbidden lines: eg. [O III], [N II]
Broad permitted lines: Ha, Hb
Bill Keel
Narrow lines only
3
Hidden Sey1 Nucleus Inside a Sey2 Galaxy
NGC 1068
Miller &
Antonucci (1983)
Antonucci &
Miller (1985)
Miller,
Goodrich &
Mathews (1991)
Bill Keel
4
NGC 1068
Hidden Broad-Line Region (HBLR)
Mrk 348: Another “HBLR Sey2”
HBLR = Hidden
Broad-Line Region
Typical
Seyfert 2 (S2)
spectrum
Appears like a
Seyfert 1 (S1)!
6
AGN Unification Paradigm: Obscuring Torus
Sey 1 view
Sey 2 view
Nuclear Engine
+ Broad-line
region (BLR):
broad Ha, Hb
Obscuring
torus
Narrow-line
region (NLR):
[O III], [N II]
Scattering e-, dust
Urry & Padovani
7
NGC 5252 Ionization Cones
Morse et al. 1998
Puzzling Differences between Sey1 & Sey2


S2s more likely to be in found in hosts with enhanced star
formation (Maiolino et al. 1995; Gu et al. 1998).
S2s tend to be in hosts with higher frequency of companions
(de Robertis, Yee, & Hayhoe 1998; Dultzin-Hacyan et al. 1999).

S2 hosts tend to show richer dust features (Malkan, Gorjian, &
Tam 1998).
9
Spectropolarimetric Study of Seyfert Galaxies






General applicability of the Unification Model (UM)?
What determines the detectability of hidden broad lines?
Is there a connection to AGN power?
How are the HBLR S2s different from non-HBLR S2s?
Correlation of presence of hidden broad lines with
observational properties?
Spectropolarimetric survey of complete samples of Seyfert 2s
to look for hidden broad-line regions (HBLRs)
10
Brief History of Spectropolarimetric Seyfert 2 Surveys
Young et al. (1996)
• sample: 24 “warm” IRAS galaxies & selected Seyfert 2s
• instrument: AAT 3.9-m
Heisler, Lumsden, & Bailey (1997)
• sample: 16 IRAS-selected Seyfert 2s, S60 > 5 Jy
• instrument: AAT 3.9-m
Moran et al. 2000
• sample: 38 objects from Ulvestad & Wilson (1989); distance-limited (cz < 4600 km s–1)
• instrument: Keck 10-m (“snapshot” mode, ~20min per object)
Lumsden et al. (2001)
• sample: 24 IRAS-selected Seyfert 2s, S60 > 3 Jy, LFIR > 1010 L, S60/S25 < 8.85
• instrument: AAT 3.9-m, WHT 4.2-m
Tran (2001, 2003)
• sample: 49 objects from the CfA and 12 mm samples
• instrument: Lick 3-m & Palomar 5-m
Summary of Surveys


HBLRs detected in  1/2 of Seyfert 2 population
Two possibilities:
– HBLRs are there but hard to detect
– Something wrong with the UM

Some explanations:
–
–
–
–
–
–
–
Only HBLR S2s are true counterparts to Seyfert 1s
Non-HBLR S2s are too weak to possess or sustain any BLRs
Broad-line variability
Non-HBLRs are narrow-line Seyfert 1s
Evolution
Obscuration / torus inclination
Scattering material
Viewing Angle Hypothesis
Miller & Goodrich (1990)
C. Heisler (1997)
13
Diagnostic Diagrams and Luminosity Distributions

Compare various properties among the HBLR, non-HBLR S2s and S1s,
illustrating their similarities and differences.
These properties include:
AGN Strength: Optical: [O III] l5007 luminosity
Infrared: Mid-IR 25mm luminosity (L25); IR color f25/f60
Radio: 20cm radio power (S20); S20/f60
X-ray: Hard X-ray (2-10 keV, HX) luminosity; HX/f60
Obscuration:
Ha/Hb; NH; HX/[O III]; Fe Ka equivalent width [EW(Fe)]
Star formation: Far-IR luminosity LFIR = L60 + L100
14
[O III] Luminosity vs f25/f60

stronger



HBLR Sey 2
Non-HBLR Sey 2
Sey 1
HII, LINERs, Starburst
f25/f60 is the IR color
 dust temp. around AGN
[O III] Lum.  AGN strength
weaker
cooler
hotter
HBLRs are more
luminous and
warmer than nonHBLRs
15
[O III] Luminosity
S1
HBLR S2
Non-HBLR S2
16
IR Color: f25/f60
S1
HBLR S2
Non-HBLR S2
17
Radio Strength: S20cm/f60 vs f25/f60




HBLR Sey 2
Non-HBLR Sey 2
Sey 1
HII, LINERs, Starburst
HBLR S2s are stronger and
hotter than non-HBLR S2s.
HBLR S2s are similar to S1s.
18
Comparison between S1, HBLR S2 & non-HBLR S2
Mean, standard deviation () and K-S statistical test results
for various properties of the S1, HBLR S2s and non-HBLR
S2s in the CFA and 12mm samples.
Property
log [O III]/Hb
log f25 /f60
log S 20 /f60
log L(O III)
log L25
log Lrad
log LFIR
log HX/[O III]
log EW(Fe)
log NH
Ha/Hb
Seyfert 1 (S1)
N
25
46
43
44
46
43
46
Mean
0.971
-0.37
-1.77
7.56
10.2
3.37
10.2

0.223
0.281
0.409
0.637
0.66
0.668
0.614
HBLR Seyfert 2 (S3)
N
22
22
22
22
22
22
22
15
13
13
23
Mean
0.972
-0.27
-1.55
7.56
10.6
3.87
10.5
0.33
2.63
23.84
7.16

0.145
0.179
0.6
0.78
0.506
0.597
0.564
0.915
0.139
0.24
4.7
non-HBLR Seyfert 2 (S2)
N
27
27
27
26
27
27
27
13
8
14
26
Mean
0.831
-0.678
-2.07
6.85
9.92
3.1
10.3
-0.196
2.84
23.86
7.39
K-S Pnull (% )

S1-S3 S1-S2 S3-S2
0.23 24.9 3.6
6.3
0.239 13.6 0.079 8E-04
0.358 11.1 1.1
0.8
0.703 94.8 0.016 0.36
0.735 8.8 26.2 0.23
0.824 4.5 15.5 1.5
0.765 12.7 52.5 58.2
0.971
36.5
0.231
57-68
0.29
88-100
5.51
36
Indicator
AGN S trength
AGN S trength
AGN S trength
AGN S trength
AGN S trength
AGN S trength
S F Activity
Obscuration
Obscuration
Obscuration
Obscuration
Note:
• L in solar luminosity, EW(Fe) in eV, NH in cm-2
• K-S Pnull is the probability for the null hypothesis that the two
distributions are drawn at random from the same parent population.
20
Comparison between S1s, HBLR S2s & non-HBLR S2s
S1 = Sey 1
Property
S3 = HBLR Sey 2
Indicator
S2 = non-HBLR Sey 2
log f 25/f 60
AGN Strength
Comparison
(S1 = S3) > S2
log S20/f 60
AGN Strength
(S1 = S3) > S2
log L(O III)
AGN Strength
log L 25
AGN Strength
(S1 = S3) > S2
S1 ~ S3 > S2 ~ S1
log L rad
AGN Strength
S3 > (S2 ~ S1)
log L FIR
SF Activity
S1 = S3 = S2
log HX/[O III]
log EW(Fe)
Obscuration
Obscuration
log N H
Obscuration
S3 = S2
S3 = S2
S3 = S2
Ha/Hb
Obscuration
S3 = S2
21
Large-scale Properties of Seyfert Galaxies

De Robertis et al. 1998 environment study:
– Very significant difference found in mean environment between S1s and S2s.
– Sample contains few known HBLR S2s (NGC 4388).

Schmitt et al. 2001 study of a far-IR selected sample (Keel et al):
– No statistically significant difference between S1s and S2s in terms of host galaxy
morphology and freq. of companions.
– However, sample was selected for warm IR color, and hence HBLR S2s.
– 70% of S2s in this sample are HBLRs, compared to ~ 50% or less for others.

Clavel et al. 2000 ISO study:
– Mid-IR ISO spectra of HBLR S2s are similar to S1s.
• 7 mm continuum, PAH luminosities and EW.
– Those of non-HBLR S2s are different (indistinguishable from SBs).
• Smaller 7 mm conti. L and higher PAH EW, but similar PAH lum.

Malkan et al. 1998 HST snapshot survey:
– S2s richer in dust than S1s.
– However, when grouped into HBLR and non-HBLR S2s, differences in dust
morphology and incidence disappear.
• non-HBLR S2: 55%, HBLR S2: 27%, S1: 23%
22
Mid-IR ISO Spectra of Seyfert Gals
S1
HBLR S2
PAH features
Non-HBLR S2
From Clavel et al. 2000
23
Two types of Seyfert 2s?


Not only are HBLR S2s stronger than non-HBLR S2s, they are
comparable in a lot of ways to S1s.
Non-HBLR S2 properties, on the other hand, generally do not match
well with those of HBLR S2s or S1s.
Proposal:
 HBLRs are visible largely because their AGNs are intrinsically more
powerful

Many non-HBLR S2s maybe too weak to represent “real” hidden S1s,
and HBLR S2s may be the only true counterparts to normal S1s.
25
Mid-Infrared line ratios
Wu et al. (2011)
Obscuration
L [O III] > 1041 ergs/s
L [O III] > 1041 ergs/s
more obscured
L [O III] < 1041 ergs/s
Luminous S2s:
More likely to
find HBLRs in
less obscured
objects
L [O III] < 1041 ergs/s
less obscured
Wu et al. (2011)
Obscuration plays a role for high-luminosity AGNs
Eddington Ratios
Marinucci et al. (2012)
 Only objects with good quality
data and MBH estimated from
stellar velocity dispersion.
 Clear separation between
HBLRs and non-HBLRs,
especially in Compton-thin
sources.
 Inclusion of Compton-thick
sources blurs out the
separation, but no HBLR falls
below Lbol/Ledd threshold.
Theoretical models: BLR unable to
form when AGN luminosities or
accretion rates are too low to
support outflows from accretion
disks (Nicastro 2000; Elitzur &
Shlosman 2006; Cao 2010)

28
Two Key Questions


Non-detections real or due to limited depth of
surveys?
Lower AGN luminosity of non-HBLRs leads to
intrinsic differences?
29
HBLR Detection: Flux Comparison



HBLR Sey 2
Non-HBLR Sey 2
HII, LINERs, Starburst
 There is no separation in the
observed fluxes for the most
common AGN indicators
between the two S2 types.
 HBLR detection reaches to
very low flux level, below
those of many non-HBLR S2s.
 Non-detections of HBLRs
cannot simply all be due to the
detection limit of the survey.
30
Deep Keck Spectropolarimetric Survey

Same sample as Tran (2003)
– CfA and 12μm samples

Target mainly non-HBLR Seyfert 2s
– ~ 25 objects

LRIS + polarimeter
– Typical exposure times: 80-160 min per object
– Multiple epochs
– 4 - 15 X deeper than previous surveys

Six new southern objects observed at CTIO
Results of the Deep Keck Survey

Six new HBLRs
– Non-HBLR  HBLR: Mrk 573, NGC 3892, NGC 5929, UGC 6100 (Keck)
– ESO 541-IG12, NGC 1125 (CTIO)

Five new non-HBLRs
– HBLR  non-HBLR: NGC 5347 (called “HBLR” by Moran et al.)
– ESO 33-G2, ESO 253-G3, NGC 3147, NGC 4968
New HBLRs: NGC 3982 & UGC 6100
NGC 3982
UGC 6100
A non-HBLR: NGC 5347
[O III] Luminosity – IR Color Diagram
stronger
f25/f60 is the IR
color
 dust temp.
around
AGN
[O III] Lum. 
AGN
strength
weaker
cooler
warmer
Receding Torus: Evolution of AGN Obscuration


Hasinger (2008)
Obscured AGN
fraction decreases
with luminosity
Consistent with
receding torus
model: larger
opening angle for
higher luminosity
AGN
Evolution?
HBLR
Non-HBLR
Yu & Hwang (2011)
Wu et al. (2011)
solar
 Significant difference in [N II]/Ha ratio
 Can be explained by an increase in nitrogen
abundance in non-HBLR S2s
 Stellar evolution  overabundance of N/O
 evolutionary connection between HBLR and
non-HBLR S2s
5x solar
Evolution? (2)
S1s
HBLR S2s
Non-HBLR S2s
Starbursts
Wu et al. (2011)
38
“Naked” View of Type 2 Seyferts

NGC 3147, NGC 4698, 1ES 1927+654

Other candidates in Shi et al (2011) and Bianchi et al (2012)
(IRAS 01428-0404, NGC 4594, Q2131-427 …)


All are Compton-thin, consistent with little or no intrinsic X-ray absorption
above Galactic column density (~ 1020 - 1021 cm-2 )
Rapid, persistent, and strong X-ray variability observed over 12 year time
scale in 1ES 1927+654

X-ray variability also observed in NGC 3147

High hard X-ray to [O III] ratios (1-100) indicate little obscuration

Inferred nuclear optical extinction is less than ~ 1 mag.

All classified as type 2 AGNs (no broad lines observed)
 contrary to expectation from the AGN unification model
Spectropolarimetric Results
Deep, repeated
observations to
probe weak lines.
In each case, a
small amount of
polarization is
detected but no
polarized broad
lines indicative of
a hidden broadline region are
seen in the
polarized flux
spectra.
Near-Infrared Spectroscopy Results
The spectra are
dominated by
galactic starlight,
and we do not
detect any
emission in Paβ or
Brγ. No direct
broad emission
lines are present.
If typical broad lines were present, their non-detections would
indicate an extinction of AV ~ 11-26
Why don’t we see any broad lines, given the naked nature of these AGNs?

Misclassified Compton-thin AGNs?
- High spatial resolution Chandra and XMM-Newton observations rule out
confusion from external sources
- Temporal variation in X-ray flux implies X-ray is likely not scattered

Variable broad emission lines?

Hidden narrow-line Seyfert 1 (NLS1) galaxies?

X-ray unobscured, but optically highly obscured?
- Multi-epoch spectropolarimetric observations designed to search for
variability failed to find any
- Available spectra over timescales of years do not show any evidence of
broad-line appearance
- No emission lines of any kind, broad or narrow, are seen in polarized flux
spectra
- No polarized FeII emission
- Narrow-line Balmer decrements are fairly normal, giving AV < 1.6 mag
- Heavy obscuration in BLR itself? High AV/NH ratio?
Most Likely Explanation

Low-powered AGNs with weak or absence of BLRs
- All three objects are low-luminosity AGNs
- Observed Eddington ratios are consistent with being below
minimum threshold needed to support BLRs (L/LEdd ≲ 10-3)
(e.g.: Nicastro 2000)
More recent X-ray & optical analysis of NGC 3147 confirms it as a
best candidate for a “true” S2 (Matt et al 2012, Bianchi et al
2012)
However, NGC 4698 is likely Compton-thick (Bianchi et al 2012)
Correlation with Broad Hα Luminosity
Stern & Laor (2012)
For NGC 3147, Bianchi et al (2012) found upper limit of broad Ha
~ two orders of magnitude smaller!
45
Seyfert 2 Spectropolarimetry
Some Surprises
NGC 2110: A Hidden Double-Peaked Emitter
Moran et al, Dec. 2005
February 2007
FWHM ~ 15,000 km/s
November 2007
FWZI ~ 28,000 km/s
Hidden (Polarized) Hα Profile Variability
NGC 2110
NGC 5252
Timescales ≲ 1 yr (similar to dynamical timescales of
accretion disks)
HDPE: Observed Polarization Variability
NGC 2110: Blue
NGC 5252: Green
 continuum and broadline polarization PAs are
the same  similar
polarization mechanism
and geometry
 polarization PA does not
change with time
 polarization PA (~70º)
perpendicular to ionization
cone axes
 scattering is the
source of polarization
What Causes the Rapid Variations?

Variability timescales ≲ 1 yr  very compact scatterers
– Discrete clouds ≲ 1 ly, NOT filled cones with large filling factor

Dramatic line profile changes in polarized flux
– NOT due to “light echo” or “search light” effect

Non-changing polarization PA  changes in line-emitting
flux, not scattering medium
Constraining the Properties of the Scatterers



No “smearing” of line profile  Te ≲ 106 K
Polarization PA ⊥ to cone axis  polar distribution
Scattering spherical “blob” model:
r
f = Lsc/Lin ≈ σT ne 2r ΔΩ
assume τe = σT ne 2r ~ 1 (optically thin)
ΔΩ ≈ ¼ (2r/d)2 = r2/d2
d
2r ~ 1 ly
ΔΩ
f ~ 1%
 ne ~ 107 cm-3; d ≲ 10 pc


Scatterers lie just outside the obscuring torus between
BLR and NLR
Much more compact and close-in than previously thought
Compact Scattering Region, Close to Nucleus

Line-emitting “ejectiles” from the nucleus
– Ejection (bi-polar) outflow model
– Simple: same material for both scatterer and line-emitting gas
– Well-defined bi-cones

Radio hot spots or material entrained in the base of jets
– Preference of DPEs in radio-loud AGNs
– Problematic for accretion disk model as scattering angle must be small (< 15°)

Material from outskirts of obscuring torus itself
– Clumpy obscuring torus model (Elitzur & Schlosman 2006; Nenkova et al 2008)
– Properties similar to scattering clumps
– Viewing angle (i ~ 65°) more consistent with extended cone morphology
Summary



Unified Model of AGN is undoubtedly correct
Orientation, evolution, luminosity, and obscuration may all
play important roles
Scatterers can be very compact, (≲ 1 ly), and located close
(< 10 pc) to central nucleus, giving rise to variability of
polarized light
Backup Slides
57