This dissertation explores the current research methods and analysis adopted for the study of Active Galactic Nuclei in all wavelengths of the electromagnetic radiation. Being the most violent objects that one can see in the present Universe, AGNs have been attributed to emitting radiation in all wavelengths and still exhibit various unexplained phenomena, alongside with being the probes to the very early Universe. The unification of the AGN model is also included for completeness, albeit not confirmed in its entirety.
3. cate course
in
Astronomy & Astrophysics
by
Sameer Patel
M.P. Birla Institute of Fundamental Research
Bangalore, India
December 2014
4. Declaration
I, Sameer Patel, student of M.P. Birla Institute of Fundamental Research, Bangalore,
hereby declare that the matter embodied in this dissertation has been compiled and
prepared by me on the basis of available literature on the topic titled,
Multi-Wavelength Analysis of Active Galactic Nuclei
as a partial ful
6. cate Course in Astronomy and Astro-physics,
2014. This dissertation has not been submitted either partially or fully to any
university or institute for the award of any degree, diploma or fellowship.
Date:
Place:
Signature
Director,
M.P. Birla Institute of Fundamental Research,
Bangalore
i
7. M.P. Birla Institute of Fundamental Research
Bangalore, India
Abstract
Multi-Wavelength Analysis of Active Galactic Nuclei
by Sameer Patel
This dissertation explores the current research methods and analysis adopted for the
study of Active Galactic Nuclei in all wavelengths of the electromagnetic radiation.
Being the most violent objects that one can see in the present Universe, AGNs have been
attributed to emitting radiation in all wavelengths and still exhibit various unexplained
phenomena, alongside with being the probes to the very early Universe. The uni
8. cation
of the AGN model is also included for completeness, albeit not con
11. nish my dissertation without help from friends, and
support from the team at MPBIFR, Bangalore.
I would also like to thank Dr. Babu for constantly reminding us to complete the dis-sertation
timely, and Ms. Komala for guiding me to coast through countless papers
online for reference. I would like to thank Rishi Dua, who as a good friend, was always
willing to help me and give his best suggestions, and Aakash Masand, who helped me
correct typographical errors and grammatical mistakes after painfully proofreading the
12. nal draft.
I would also like to thank my parents. They were always supporting me and encouraging
me with their best wishes.
iii
27. Abbreviations
AGN Active Galactic Nuclei
SMBH Super Massive Black Hole
QSO Quasi Stellar Objects
IRAS Infrared Astronomical Satellite
NLRG Narrow Line Radio Galaxies
BLRG Broad Line Radio Galaxies
WLRG Weak-Emission Line Radio Galaxies
BLR Broad Line Region
SSRQ Steep Spectrum Radio Quasars
FSRQ Flat Spectrum Radio Quasars
FR-I Fanaro Riley Type I
FR-II Fanaro Riley Type II
BL Lac BL Lacertae
OVV Optically Violently Variable Quasars
LINER Low Ionization Nuclear Emission-Line Region
LLAGN Low Luminosity Active Galactic Nuclei
SED Spectral Energy Distribution
SF Star Formation
RIAF Radiatively Inaccurate Accretion Flow
SSC Synchrotron Self Compton
BH Black Hole
NIR Near Infrared
MIR Mid Infrared
FIR Far Infrared
RM Rotation Measure
x
28. Abbreviations
IC Inverse Compton
UV Ultra-Violet
HST Hubble Space Telescope
MHD Magnetohydrodynamics
FWHM Full Width at Half Maximum
S/N Signal to Noise Ratio
NLS1 Narrow Line Seyfert Type I
SDSS Sloan Digital Sky Survey
BAL Broad Absorption Line
BEL Broad Emission Line
XRB X-Ray Binary
SAS Small Astronomy Satellite
OSO Observing Solar Observatory
HEAO High Energy Astronomy Observatory
2MASS 2 Micron All Sky Survey
XMM X-Ray Multi-Mirror Mission
RGS Re
ecting Grating Spectrometer
CCD Charge Coupled Device
ESA European Space Agency
NASA National Aeronautics and Space Agency
COSMOS Cosmic Evolution Survey
EW Equivalent Width
HIG Highly Ionized Gas
BAT Burst Alert Telescope
ROSAT Roentgen Satellite
INTEGRAL International Gamma-Ray Astrophysics Laboratory
CGRO Compton Gamma-Ray Observatory
LAT Large Area Telescope
VLBI Very Long Baseline Interferometry
HESS High Energy Spectroscopic System
MERLIN Multi-Element Radio Linked Interferometer Network
VLA Very Large Array
HBLR Hidden Broad Line Region
29. Abbreviations
OSSE Oriented Scintillation Spectrometer Experiment
EXOSAT European X-Ray Observatory Satellite
PDS Planetary Data System
HBL High-Frequency Peaked BL Lac Objects
LBL Low-Frequency Peaked BL Lac Objects
RMS Root Mean Squared
30. Chapter 1
Introduction
1.1 The History of AGNs
Unusual activity in the nuclei of galaxies was
31. rst recognised by Minkowski and Humason
(Mount Wilson Observatory), when in 1943 they asked a graduate student Carl Seyfert
to study a class of galaxy with an emission spectrum from the compact bright nucleus.
Most normal galaxies show a continuum with absorption lines, but the emission in the
Seyfert galaxies betrayed the presence of hot tenuous gas. In some cases the emission
lines were broad (Type 1 Seyferts) indicating gas moving with high velocities and in
other objects, the emission lines were narrow (Type 2 Seyferts) indicating that the gas
was moving more slowly. In the 1950s, as radio astronomy became a rapidly developing
science a whole new range of discoveries were made in astronomy. Amongst these were
the Radio Galaxies, which appeared to be elliptical galaxies that were inconspicuous at
optical wavelengths but were shown to have dramatically large, prominent lobes at radio
frequencies, stretching for millions of light years from the main galaxy
1.2 Active Galactic Nuclei
The names active galaxies and active galactic nuclei (AGNs) are related to the main
feature that distinguishes these objects from inactive (normal or regular) galaxies |the
presence of accreting supermassive black holes (SMBHs) in their centers. As of 2011,
there are approximately a million known sources of this type selected by their color and
1
32. Chapter 1. Introduction
several hundred thousand by basic spectroscopy and accurate redshifts. It is estimated
that in the local universe, at z 0.1, about 1 out of 50 galaxies contains a fast-accreting
SMBH, and about 1 in 3 contains a slowly accreting SMBH. Detailed studies of large
samples of AGNs, and the understanding of their connection with inactive galaxies and
their redshift evolution, started in the late 1970s, long after the discovery of the
33. rst
quasi-stellar objects in the early 1960s. Although all objects containing active SMBH
are now referred to as AGNs, various other names, relics from the 1960s, 1970s, and
even now, are still being used.
The most powerful active galaxies were discovered with radio telescopes in the 1960's
and named `Quasi-Stellar Radio Sources', later shortened to QSOs or quasars. Their
huge luminosities ( 104246 erg s1) could not be attributed to starlight alone, and the
rapid variability observed (from months down to days) implied that the radiation was
emitted from very small volumes with characteristic linear size of the order of light days.
At the time, it was proving dicult to reconcile these two properties. As more detailed
observations were performed it became clear that AGNs were most likely powered by
accretion of matter onto a central SMBH (105 M
34. ).
It is considerable to add that not all galaxies are active. Our Milky-Way is one of the
numerous galaxies that hosts a SMBH at its galactic center (Schodel et al., 2002), with
MSMBH 4:6 0:7 106M
35. , but is not considered to be an active galaxy due to
the fact that there is no apparent accretion on to the SMBH. On contrary, the central
regions of an AGN are likely not static, but very dynamic and violent.
1.3 The Taxonomy of AGNs
The observational classi
36. cation of AGNs is not so clear because of observational limi-tations,
heavy source obscuration (in most cases) and usually varying accretion rate on
many orders of magnitude. Classically, an object is classi
37. ed as an AGN if :-
It contains a compact nuclear region emitting signi
38. cantly beyond what is expected
from stellar processes typical of this type of galaxy.
It shows the clear signature of a non-stellar continuum emitting process in its
center.
2
39. Chapter 1. Introduction
Its spectrum contains strong emission lines with line ratios that are typical of
excitation by a non-stellar radiation
40. eld.
It shows line and/or continuum variations.
1.3.1 Seyferts
Owing the name to Seyfert (1943) who was the
41. rst to discover these types, the major-ity
of AGN with visible host galaxies fall under this class, known as Seyfert Galaxies.
Seyfert, in his
42. rst observation, had reported a small percentage of galaxies had very
bright nuclei that were the source of broad emission lines produced by atoms in a wide
range of ionization states. These nuclei were nearly stellar in appearance (no powerful
telescopes at that time were available).
Today, these are further divided into two more subcategories :-
Type I Seyferts: Spectra contain very broad emission lines that include both
allowed lines (H I, He I, He II) and narrower forbidden lines (O [III]). They
generally also have narrow allowed lines albeit being comparatively broader than
those exhibited by non-active galaxies. The width of these lines is attributed to
Doppler broadening, indicating that the allowed lines originate from sources with
speeds typically between 1000 and 5000 km s1
Type II Seyferts: Spectra contain only narrow lines (both permitted and forbid-den),
with characteristic speeds of about 500 km s1
1.3.2 Quasars and QSOs
The terms Quasar (Quasi Stellar Radio Source) and QSO (Quasi Stellar Object), often
used interchangeably, are scaled up versions of a Type I Seyfert, where the nucleus has
a luminosity MB 21:5 + 5 log h0 Schmidt Green (1983). Maarten Schmidt
recognized that the pattern of the broad emission lines of 3C 273 (the
43. rst detected
quasar) was the same as the pattern of the Balmer lines of Hydrogen, but were
severely redshifted to z = 0.158 to unfamiliar wavelengths, thus alluding astronomers
from understanding it.
3
44. Chapter 1. Introduction
Figure 1.1: The spectrum of NGC 1275. The emission features seen at 5057 A
and
6629 A
are [O III] 5007 and H, respectively.
(Sabra et al., 2000)
Figure 1.2: The visible spectrum of Mrk 1157, a Seyfert 2 galaxy.
(Osterbrock, 1984)
4
45. Chapter 1. Introduction
In 1963, the Dutch astronomer Maarten Schmidt recognized that the pattern of the
broad lines of 3C 273 was the same as the pattern of the Balmer lines of Hydrogen,
only severely redshifted to z = 0:158, hence alluding astronomers from identifying its
spectrum. The continuous spectrum of a quasar may span nearly 15 orders of
magnitude in frequency, very broad compared with the sharply peaked blackbody
spectrum of a star. Quasars emit an excess of UV light relative to stars and so are
quite blue in appearance. This UV excess is indicated by the big blue bump in
(nearly) every quasar spectrum. A quasar's radio emission may come either from radio
lobes or from a central source in its core.
Figure 1.3: The visible spectrum of 3C 273, a Quasar.
(Francis et al., 1991)
1.3.3 Radio Galaxies
These galaxies are very luminous at radio wavelengths, with luminosities up to 1039 W
between 10 MHz and 100 GHz. The observed structure in radio emission is determined
by the interaction between twin jets and the external medium, modi
46. ed by the eects
of relativistic beaming. These are further subdivided into two categories.
5
47. Chapter 1. Introduction
1.3.3.1 Radio Quiet
Similar in many aspects to Type I Seyferts, these galaxies show both broad and narrow
lines, the only dierence being that they are much more luminous than Type I
Seyferts. They are observed in the absence of relativistic jets, which contribute the
most energies in the radio wavelength spectrum.
Radio Quiet Type I AGNs: These have relatively low-luminosities and therefore
are seen only nearby, where the host galaxy can be resolved, and the
higher-luminosity radio-quiet quasars, which are typically seen at greater
distances because of their relative rarity locally and thus rarely show an obvious
galaxy surrounding the bright central source.
Radio Quiet Type II AGNs: These include Seyfert II galaxies at low luminosities,
as well as the narrow-emission-line X-ray galaxies (Mushotzky, 1982). The
high-luminosity counterparts are not clearly identi
48. ed at this point but likely
candidates are the infrared-luminous IRAS AGN (Hough et al., 1991, Sanders
et al., 1989, Wills et al., 1992), which may show a predominance of Type II
optical spectra.
1.3.3.2 Radio Loud
Usually attributed to AGNs with unipolar/bipolar, relativistic jets beaming out of
their centers, the radio emission from radio-loud active galaxies is synchrotron
emission, as inferred from its very smooth, broad-band nature and strong polarization.
This implies that the radio-emitting plasma contains, at least, electrons with
relativistic speeds (Lorentz factors of 104) and magnetic
49. elds. However,
synchrotron radiation not being unique to radio wavelengths, if the radio source can
accelerate particles to high enough energies, features which are detected in the radio
may also be seen in the infrared, optical, ultraviolet or even X-ray.
Radio Loud Type I AGNs: These are called Broad-Line Radio Galaxies (BLRG)
at low luminosities and radio-loud quasars at high luminosities, either Steep
Spectrum Radio Quasars (SSRQ) or Flat Spectrum Radio Quasars (FSRQ)
depending on radio continuum shape.
6
50. Chapter 1. Introduction
Radio Loud Type II AGNs: Often called Narrow-Line Radio Galaxies (NLRG),
these include two distinct morphological types: the low-luminosity Fanaro-Riley
type I (Figure 1.4) radio galaxies (Fanaro Riley, 1974), which have
often-symmetric radio jets whose intensity falls away from the nucleus, and the
high-luminosity Fanaro-Riley type II (Figure 1.5) radio galaxies, which have
more highly collimated jets leading to well-de
51. ned lobes with prominent hot
spots.
Figure 1.4: The total intensity distribution of 3C 338, a FR I classi
52. ed AGN.
(Ge Owen, 1994)
Figure 1.5: The total intensity distribution of 3C 173P1, a FR II classi
54. Chapter 1. Introduction
1.3.4 Blazars
Originally named after what was thought to be an irregular, variable star BL Lacertae,
these are AGNs which are characterized by rapid and large-amplitude
ux variability
and signi
55. cant optical polarization. When compared to quasars with strong emission
lines, blazars have spectra dominated by a featureless non-thermal continuum. The
most well known object in this class is the BL Lacertae. Joining the BL Lac objects in
the blazar classi
56. cation are the optically violently variable quasars (OVVs), which are
similar to the BL Lacs except that they are typically much more luminous, and their
spectra may display broad emission lines. Blazars are AGNs viewed head on and hence
often have jets associated with them (Figure 1.6)
Figure 1.6: The X-ray image of 3C 273's jet.
(3C273 Chandra by Chandra X-ray Observatory - NASA. Licensed under Public
domain via Wikimedia Commons)
1.3.4.1 BL Lacerate Objects
BL Lacertae Objects, or BL Lacs for short, are a subclass of blazars that are
characterized by their rapid time-variability. Their luminosities may change by upto
30% in just 24 hours and by a factor of 100 over a longer time period. BL Lacs are also
distinguished by their strongly polarized power-law continua (30% 40% linear
polarization) that are nearly devoid of emission lines, suggesting that there are very
powerful EM
57. elds at play. BL Lacs, like quasars, are at cosmological distances. Of all
the BL Lacs that have been resolved, 90% of those appear to reside in elliptical
galaxies.
8
58. Chapter 1. Introduction
1.3.4.2 Optically Violent Variable Quasars
Almost similar to BL Lacs, OVVs are typically much more luminous and may display
broad emission lines in their spectra. The currently best known example of an OVV is
3C 279.
1.3.5 LINERs
LINERs (Low Ionization Nuclear Emission-line Regions) are types of active galaxies
that have very low luminosities in their nuclei, but with fairly strong emission lines of
low-ionization species, such as the forbidden lines of [O I] and [N II]. The Spectra of
LINERs seem similar to the low-luminosity end of the Seyfert II class, and LINER
signatures are detected in many (most of) spiral galaxies in high-sensivity studies.
These low-ionization lines are also detectable in starburst galaxies and in H II regions
and hence it is sometimes dicult to distinguish between LINERs and starburst
galaxies. In the local universe, they are found in about one-third of all galaxies brighter
Figure 1.7: The UV spectrum of NGC 4594 LINER observed using the HST FOS.
(Nicholson et al., 1998)
than B = 15.5 mag. This is larger than the number of local high-ionization AGNs by a
factor of 10 or more. Local high-ionization AGNs and LINERs are present in galaxies
with similar bulge luminosities and sizes, neutral hydrogen gas (H I) contents, optical
colors, and stellar masses. Given a certain galaxy type and stellar mass, LINERs are
9
59. Chapter 1. Introduction
usually the lowest-luminosity AGNs, with nuclear luminosity that can be smaller than
the luminosity of high-ionization AGNs by 1-5 orders of magnitude. An alternative
name for this class of objects is low-luminosity AGNs (LLAGNs). The strongest
Figure 1.8: The spread of emission-line galaxies from the SDSS on one diagnostic
diagram that uses four strong optical emission lines, H, H
60. , [O III] 5007, and [N
II] 6584, to distinguish galaxies that are dominated by ionization from young stars
(green points) from those that are ionized by a typical AGN SED (blue points for high-ionization
AGNs and red points for low-ionization AGNs). The AGN and SF groups
are well separated, but the division between the two AGN groups is less clear. The
curves indicate empirical (solid) and theoretical (dashed) dividing lines between AGNs
and star-forming galaxies.
(Groves Kewley, 2008)
optical emission lines in the spectrum of LINERs include [O III] 5007, [O II] 3727,
[O I] 6300, [N II] 6584, and hydrogen Balmer lines. All these lines are prominent
also in high-ionization AGNs, but in LINERS, their relative intensities indicate a lower
mean ionization state. For example, the [O III] 5007/H
61. line ratio in LINERs is 3-5
times smaller than in high-ionization Type-II AGNs. Line diagnostic diagrams are
ecient tools to separate LINERs from high-ionization AGNs. One such example is
shown in Figure 1.8. The exact shape of a LINERs SED is still an open issue. In some
sources, it is well represented by the SED shown in Figure 5.3. Such an SED has a
clear de
62. cit at UV wavelengths compared with the spectrum of high-ionization AGNs.
However, some LINERs show strong UV continua and, occasionally, UV continuum
variations, and it is not entirely clear what fraction of the population they represent.
10
63. Chapter 1. Introduction
Figure 1.9: (left) Radio luminosity vs. optical (B-band) luminosity for various types
of AGNs. (right) The radio loudness parameter R vs. (L=LEdd).
(Sikora et al., 2007)
This is related to the issue of Radiatively Inecient Accretion Flows (RIAFs) and the
relationship between the mass-accretion rate onto the BH and the emitted radiation.
Point-like X-ray sources have been observed in a large number of LINERs. These
nuclear hard X-ray sources are more luminous than expected for a normal population
of X-ray binaries and must be related to the central source. Many LINERs also
contain compact nuclear radio sources similar to those seen in radio-loud
high-ionization AGNs but with lower luminosity comparable to WLRGs (Figure 1.9).
The UV-to-X-ray luminosity ratio in LINERs is, again, not very well known. In
LINERs with strong UV continua, ox is smaller than in low-redshift, high-ionization
AGNs, consistent with the general trend between ox and Lbol. However, ox is not
known for most LINERs because of the diculty in measuring the UV continuum.
Like other AGNs, LINERs can be classi
64. ed into Type-I (broad emission lines) and
Type-II (only narrow lines) sources. The broad lines, when observed, are seen almost
exclusively in H and hardly ever in H
65. . This is most likely due to the weakness of the
broad wings of the Balmer lines that are dicult to observe against a strong stellar
continuum. Some, perhaps many, LINERs may belong to the category of real Type-II
AGNs |those AGNs with no BLR. The phenomenon is expected to be more common
among low-luminosity sources and hence to be seen in LINERs. Because of all this, the
classi
66. cation of LINERs is ambiguous, and the relative number of Type-I and Type-II
objects of this class is uncertain even at very low redshift.
11
67. Chapter 2
Non-Thermal Processes
Much of the electromagnetic radiation emitted by AGNs is very dierent from a simple
blackbody emission or a stellar radiation source. The general name adopted here for
such processes is non-stellar emission, but the term non-thermal emission is commonly
used to describe such sources. There are several types of non-stellar radiation
processes.
2.1 Basic Radiative Transfer
Describing the interaction of radiation with matter requires the use of three basic
quantities: the
69. c intensity I, which gives the local
ux per unit time,
frequency, area, and solid angle everywhere in the source. The second quantity is the
monochromatic absorption cross section, (cm1), which combines all loss
(absorption and scattering) processes. The third quantity is the volume emission
coecient, j, which gives the locally emitted
ux per unit volume, time, frequency,
and solid angle. The three are combined into the equation of radiative transfer,
dI
ds = I + j;
where ds is a path length interval. The
70. rst term on the right in this equation
describes the radiation loss due to absorption, and the second gives the radiation gain
due to local emission processes. One usually de
72. Chapter 2. Non-Thermal Processes
dI
d
= I + S;
where S = j= is the source function. The formal solution of the equation of
transfer depends on geometry. For a slab of thickness in a direction perpendicular
to the slab, it is
I() = I(0)e +
R
0
e(t)S(t)dt:
For any other direction , both and dt must be divided by cos .
The general equation of radiative transfer is dicult to solve and requires numerical
techniques. However, there are simple cases in which the solution is straightforward.
In particular, the case of a slab and a constant source function that is independent of
allows a direct integration and gives the following solution:
I = I(0)e + S(1 e ):
For an opaque source in full thermodynamic equilibrium (TE), the optical depth is
large, and both I and S approach the Planck function
B(T) = 2h3=c2
eh=kT1
2.2 Synchrotron Radiation
2.2.1 Emission by a Single Electron in a Magnetic Field
Considering an electron of energy E that is moving in a uniform magnetic
73. eld B of
energy density uB = B2=8, the energy loss rate, dE=dt, which is also the power
emitted by the electron, P, is given by
P = 2T c
2
74. 2uB sin2 ;
where T is the Thomson cross section,
c is the speed of light,
13
76. = v=c, and
v is the speed of the electron.
The angular term sin2 re
ects the direction of motion, where is the pitch angle
between the direction of the motion and the magnetic
78. 2uB:
The radiation emitted by a single electron is beamed in the direction of motion. The
spectral energy distribution (SED) of this radiation is obtained by considering the gyro
frequency of the electrons around the
79. eld lines (!B = eB=
mec) and the mean interval
between pulses (2=!B). The calculation of the pulse width is obtained by considering
the relativistic time transformation between the electron frame and the observer frame.
This involves an additional factor of
2. Thus, the pulse width is proportional to
3
or, expressed with the Larmor angular frequency, !L = eB=mec (which diers from !B
by a factor of
), to
2. Fourier transforming these expressions gives the mean
emitted spectrum of a single electron, P
, which peaks at a frequency near
2!L.
2.2.2 Synchrotron Emission by a Power Law Distribution of Electrons
Assuming now a collection of electrons with an energy distribution n(
)d
that gives
the number of electrons per unit volume with
in the range
(
+ d
), the emission
coecient due to the electrons is obtained by summing P
(
) over all energies:
j = 1
4
1 R
1
P(
)n(
)d
:
There is no general analytical solution to this expression since n(
) can take various
dierent forms. However, there are several cases of interest where n(
) can be
presented as a power law in energy:
n(
)d
= n0
pd
:
The additional assumption that all the radiation peaks around a characteristic
frequency,
2L, where L is the Larmor frequency, gives the following solution for j:
14
80. Chapter 2. Non-Thermal Processes
3T n0uB1
4j = 2
L
L
p1
2 :
Figure 2.1: A comparison of a synchrotron source with p = 2:5 (solid line) and a 105
K blackbody source (dotted line).
(Netzer, 2013)
2.2.3 Synchrotron Self-Absorption
The source of fast electrons can be opaque to its own radiation. This results in a
signi
82. cation of the emergent spectrum especially at low frequencies, where
the opacity is the largest. It can be shown that in this case,
/ p+4
2 ;
that is, the largest absorption is at the lowest frequencies. Using the equation of
radiative transfer for a uniform homogeneous medium, we get the solution at the large
optical depth limit, I / 5=2, which describes the synchrotron SED at low energies.
This function drops faster toward low energies than the low-energy drop of a blackbody
spectrum (I / 2). The overall shape of such a source is shown in Figure 2.1.
2.2.4 Polarization
Synchrotron radiation is highly linearly polarized. The intrinsic polarization can reach
70%. However, what is normally observed is a much smaller level of polarization,
15
83. Chapter 2. Non-Thermal Processes
Source B (G) (Hz)
tcool (yr) E (erg)
Extended radio sources 105 109 104 107 1059
Radio jets 103 109 103 104 1057
Compact jets 101 109 102 101 1054
BH magnetosphere 104 1018 104 1010 1047
Table 2.1: Synchrotron Sources in AGNs.
(Netzer, 2013)
typically 3-15%. This indicates a mixture of the highly polarized synchrotron source
with a strong non-polarized source. For AGNs, especially radio-loud sources, this
polarization is clearly observed. There is also a correlation between high-percentage
polarization and large-amplitude variations. AGNs showing such properties go under
the name blazars. In the NIR-optical-UV spectrum of radio-loud AGNs, the region
around 1 m shows most of the polarization. The percentage polarization seems to
drop toward shorter wavelengths, in contrast to what is expected from a pure
synchrotron source. This is interpreted as an indication of an additional thermal,
non-polarized source at those wavelengths.
2.2.5 Synchrotron Sources in AGNs
It is thought that most of the non-thermal radio emission in AGNs is due to
synchrotron emission. There are various ways to classify such radio sources using the
slope, (p 1)=2, and the break frequency below which it is optically thick to its own
radiation. Table 2.1 gives a summary of the properties of several observed and
expected synchrotron sources in AGNs. It includes the typical strength of the
magnetic
84. eld, B (in gauss), the Lorentz factor,
, and the total energy generated in
the source, E, which is obtained by integrating uB over the volume of such sources.
The table also shows the typical cooling time of the source, tcool, which is a
characteristic lifetime de
86. Chapter 2. Non-Thermal Processes
2.2.6 Faraday Rotation
Michael Faraday discovered in 1845 that the angle of polarization of an
electromagnetic wave changes when the wave is sent through a medium with a
magnetic
87. eld. The so-called Faraday rotation can also aect the synchrotron
emission. Faraday rotation can be understood as the dierent eect the magnetized
plasma has on the left and right circularly polarized light. Depending on the
orientation with respect to the magnetic
88. eld, the components will see a dierent
refractive index. Thus, the phase velocity of the two components will be aected
slightly dierently and lead to a shift of their relative phases. This causes the plane of
polarization to rotate, depending on how strong the magnetic
89. eld is and what
distance the wave has to travel through the plasma. A similar eect is also observed
with linearly polarized light. Once the linearly polarized synchrotron light is emitted
and travelling towards the observer, it can pass through magnetized material causing
Faraday rotation. This can be the emitting plasma itself, or any magnetized gas along
the line of sight. In astrophysical applications, one can simplify the problem by
considering only free electrons in magnetic
90. elds.
The amount of rotation in the polarization angle depends on the magnetic
91. eld
strength and density of the electrons along the line of sight, but also on the frequency
of the electromagnetic wave one observes:
= 2RM:
Here, is the wavelength of the polarized radiation, and RM is the rotation measure
which is a function of the electron density ne and of the component of the magnetic
92. eld Bjj parallel to the line of sight:
= 2 e3
2m2c4
R
ne(s)Bjj(s)ds:
Thus, the rotation is larger for low frequencies. This is because the frequency of the
wave is much larger than the gyro-frequency of the electron. The closer the light and
the electron are to a resonant state, and thus the larger the energy transfer from the
wave to the electron. The light from extragalactic sources will not only have to cross
the intergalactic medium, but the interstellar medium of our galaxy as well on its path
17
94. eld along the line of sight will not be constant, and
importantly, it will not be of the same orientation throughout the path of light. To
determine the net eect of Faraday rotation, it is necessary to measure polarization at
closely spaced frequency interval over many frequencies. Because the rotation aects
the high frequency the least, the best way to get an estimate of the intrinsic
polarization of a synchrotron source is to measure at high frequencies.
2.3 Thomson Scattering
Thomson scattering describes the non-relativistic case of an interaction between an
electromagnetic wave and a free charged particle. The eect was
95. rst describe by Sir
Joseph John Thomson, who discovered the electron when studying cathode rays in the
late nineteenth century. The process can be understood as elastic or coherent
scattering, as the photon and the particle will have the same energy after the
interaction as before. For this process of the energy E of the photon has to be much
smaller than the rest energy of the particle:
E = h mc2:
Another requirement for Thomson scattering is that the particle must be moving at
non-relativistic speed (v c). In the classical view of this process, the incoming
photon is absorbed by the particle with charge q, which is set into motion and then
re-emits a photon of the same energy.
Using the classical electron radius r0 = q2=mc2 (Bohr radius), the dierential
cross-section of this elastic scattering process can be written as
d
d
= 1
2(1 + cos2 )r2
0:
This is symmetric with respect to the angle , thus the amount of radiation scattered
in the forward and backward direction is equal. The total cross-section is then given by
T = 2
R
0
d
d
sin d = 8
3 r2
0 = 8
3
q2
mc2
2
:
18
96. Chapter 2. Non-Thermal Processes
In the case of electrons, this gives a Thomson cross-section of T ' 6:652 1025 cm2.
The cross-section for a photon scattering on a photon is a factor of
(mp=me)2 ' 3:4 106 smaller.
Since in the classical view of this process, the electron has no preferred orientation, the
cross-section is independent of the incoming electromagnetic wave. The polarization of
the scattered radiation depends, however, on the polarization of the incoming photon
wave. Unpolarized radiation becomes linearly polarized in the Thomson scattering
process with the degree of polarization being
= 1cos2
1+cos2 :
Therefore, polarization of the observed emission can be a sign that the emergent
radiation has been scattered.
Thomson scattering is important in may astrophysical sources. Any photon which will
be produced inside a plasma can be Thomson scattered before escaping in the
direction of the observer. The chance for the single photon to be Thomson scattered
and how many of the photons will be scattered out of or into the line of sight is
quanti
97. ed in terms of the optical depth of the plasma:
=
R
T nedx;
where ne is the electron density, and dx is the dierential line element. The mean free
path T of the photon, that is, the mean distance traveled between scatterings will
thus be T = (T ne)1.
2.4 Compton Scattering
The interaction between an electron and a beam of photons is described by the
classical Compton scattering theory. For stationary or slow electrons, one uses energy
and momentum conservation to obtain the relationship between the frequencies of the
coming (
0
) and scattered () photons. If ~n and ~n0 are unit vectors in the directions
of these photons, and cos = ~n ~n0 , we get
19
98. Chapter 2. Non-Thermal Processes
= mec2
0
mec2+h0 (1cos )
:
For non-relativistic electrons, the cross section for this process is given by
d
d
= 1
2r2
e [1 + cos2 ];
where re = e2=mec2 is the classical electron radius. Integrating over angles gives the
Thomson cross section, T . In the high-energy limit, the cross section is replaced by
the Klein-Nishina cross section, KN, which is normally expressed using = h=mec2.
The approach to the low-energy limit is given roughly by
KN T (1 2);
and for 1,
KN 3
8
T
h
ln 2 + 1
2
i
:
2.4.1 Comptonization
The term Comptonization refers to the way photons and electrons reach equilibrium.
The fractional amount of energy lost by the photon in every scattering is
' h
mec2 = :
Considering a distance r from a point source of monochromatic luminosity L in an
optically thin medium where the electron density is Ne, the
ux at this location is
L=4r2, and the heating due to Compton scattering is
HCS =
R
L
4r2NeT
h
h
mec2
i
d:
The cooling of the electron gas is the result of inverse Compton scattering. Like
Compton scattering, this process is a collision between a photon and an electron,
except that in this case, the electron has more energy that can be transfered to the
radiation
99. eld. In this case, the typical gain in the photon energy is a factor of
2
larger than the one considered earlier. This factor is obtained by
100. rst transforming to
the electron's rest frame and then back to the laboratory frame. If x is the fraction of
the electron energy kT which is transferred to the photon,
20
101. Chapter 2. Non-Thermal Processes
*
+
= x kTe
mec2 ;
where Te is the electron temperature. Using this terminology, one can write the cooling
term for the electron gas as
CCS =
R
L
4r2NeT
h
xkTe
mec2
i
d:
A simple thermodynamical argument suggests that if Compton heating and Compton
cooling are the only heating-cooling processes, and if the radiation
102. eld is given by the
Planck function (L = B), the equilibrium requirement, HCS = CCS, gives x = 4.
Because this is a general relation between a physical process and its inverse, the result
must also hold for any radiation
104. eld in luminous AGNs can be very intense, and the energy density of
the photons normally exceeds the energy density due to electrons. The requirement
HCS = CCS gives, in this case, a Compton equilibrium temperature of
TC = h
4k ;
where the mean frequency, , is de
105. ned by integrating over the SED of the source,
=
R
RLd
Ld :
2.4.2 The Compton Parameter
The emitted spectrum of thermal and non-thermal radiation sources that are
embedded in gas with a thermal distribution of velocities is modi
106. ed due to Compton
and inverse Compton scattering. For high-energy electrons, inverse Compton is the
dominant process, and the resulting collisions will up-scatter the photon energy. The
emergent spectrum is modi
107. ed, and its spectral shape will depend on the original
shape, the electron temperature, and the Compton depth of the source, which
determines the number of scattering before escape. Considering an initial photon
energy of hi and the case of thermal electrons with temperature Te such that
hi 4kTe, the scattering of such photons by a fast electron will result in energy gain
21
108. Chapter 2. Non-Thermal Processes
per scattering (inverse Compton scattering). The photon continues to gain energy,
during successive scatterings, as long as hi 4kTe. If the
109. nal photon energy is hf ,
and the number of scatterings is N, we get
hf ' hie
h
N 4kTe
mec2
i
:
For a medium with Compton depth e, the mean number of scatterings is roughly
max(e; 2
e ). Using this, one can de
110. ne a Compton parameter y,
y = max(e; 2
h
4kTe
mec2
e )
i
;
such that
hf hiey:
The factor ey is an energy ampli
111. cation factor. For y 1, one is in the regime of
unsaturated inverse Comptonization. For y 1, the process reaches a limit where the
average photon energy equals the electron thermal energy. This is saturated Compton
scattering.
2.4.3 Inverse Compton Emission
An important example is the case of a source whose spectrum is due to scattering of
soft photons onto relativistic electrons. Again, considering
112. rst the typical energy
following a single scattering and then averaging over the energy distribution of the
photons and electrons, a simple way to estimate the power emitted in the preceding
process is to consider a beam of photons with number density nph and mean energy
before scattering h0. The energy density of these photons is nphh0, and the energy
ux of photons incident on a stationary electron is curad = nphh0c. The mean energy
after scattering, h, is larger than the mean energy before scattering by a factor of
order
2. In the rest frame of the electron, the process can be considered as a simple
Thomson scattering with radiated power given by the classical expression
P = T curad. Thus, the simple L / (p1)=2 estimate for the laboratory frame emitted
power is P =
2T curad. A more accurate derivation of the emitted power must take
22
113. Chapter 2. Non-Thermal Processes
into account the scattering angle and its transformation between frames. The
115. 2urad;
which diers from the simple estimate by a factor of order of unity.
The expression for the power emitted due to inverse Compton (IC) scattering is
basically identical to the power emitted by synchrotron radiation, except that the
energy density of the magnetic
116. eld, uB, was replaced by the energy density of the
radiation
117. eld, urad. Thus, the mean power of the two processes, assuming they take
place in the same volume of space, is simply uB=urad. Also, for the same volume of
space, the energy distribution of the relativistic electrons is given by the same
power-law function used in the synchrotron case, n(
) /
p. Thus, one also gets a
similar dependence of the monochromatic luminosity on the parameter p:
L(IC) / (p1)=2:
2.4.4 Synchrotron Self-Compton
In a compact synchrotron source, the emitted photons can be inverse Compton
scattered by the relativistic electrons that emit the synchrotron radiation. This gives
the photon a big boost in energy. The emergent radiation is synchrotron self-Compton
(SSC) emission. The
ux emitted by this process can be calculated by integrating over
the synchrotron radiation spectrum and the electron velocity distribution. To a good
approximation, the resulting spectral index is identical to the spectral index of the
synchrotron source. The synchrotron self-Compton process can repeat itself, in the
same source, by additional scattering of the emergent photons, which results in an
additional boosting, by a factor
2, to the photons. The natural limit for the process is
when the scattered photon energy extends into the
-ray and the condition of
h
mec2 (the condition for no Compton recoil of the electron) no longer holds. At
this limit, the resulting radiation density decreases dramatically.
23
118. Chapter 2. Non-Thermal Processes
2.5 Annihilation and Pair-Production
The observations of
-ray jets in many AGNs suggest that, under some conditions, the
density of high-energy photons is large enough to result in ecient pair production and
a concentration of both electrons and positrons in some parts of the central source.
Under these conditions, energetic
-ray photons, with energies much above the rest
energy of the electron, can react with lower-energy photons to create electron-positron
pairs. Short-time-scale variations of the X-ray spectrum, in the lower-luminosity AGN,
indicate extremely small dimensions;
-ray photons that are associated with the X-ray
source would not be able to escape these regions and would create electron-positron
pairs. Likely locations where such processes take place are in the corona of the central
accretion disk or inside the
-ray jet.
The process of pair-production and its reverse process (for e e+ pair) is given by
e + e+
+
:
Considering the interaction between a
-ray photon with frequency
, above the rest
mass frequency of the electron, with an X-ray photon of frequency X below this
frequency and using the notation of unit vectors for the photons, one can write the
threshold frequency for pair production as
v
=
mec2
h
2
2
X(1~n
~nX) :
The
cross section is given by
= 3
16T (1
125. for the electron and the positron is measured in the center of
momentum frame. The typical value of
near threshold is 0:2T , and it declines
with frequency as 1
.
The size of the radiation source, R, plays an important role in determining the optical
depth of the source and hence the probability of pair-production taking place. This
dependence is usually described by de
126. ning a compactness parameter for the
-ray
source, l
, using the source size and its luminosity, L
. There is an equivalent
24
127. Chapter 2. Non-Thermal Processes
compactness parameter for the X-ray source, lX. Assuming that the typical
-ray
photon energy is = mec2 and that the photon number density is
N
= L
4R2c :
The mean free path of the photons for the pair-production is
=
N
T
1
and for unit optical depths, R , which gives
L
T
4mec3R 1:
This leads to the following expression for the compactness parameter:
l
= L
T
4mec3R;
which is equivalent to the pair production optical depth of the source. In principle, l
can be measured from the variability time scale of the
-ray source. In reality,
however, this is dicult to measure and is occasionally replaced by lX and the X-ray
variability time scale. When lX hX=mec2, it will be dicult for the
-rays to
escape the source without creating pairs.
The rate of the inverse process, pair annihilation, in the non-relativistic limit is
independent of temperature and is roughly 0.4 NeT c per unit volume, where Ne is the
combined electron-positron density. In a steady state, pair production is balanced by
annihilation,
mec3l
4TR2h
=
0:4NeTc;
where l
is the compactness parameter for those
-ray photons for which the source is
optically thick to pair production. This equation can be solved for the mean Thomson
depth in the source, T . For large T , the electrons and positrons thermalize because
their interaction time is short compared with the annihilation time. In AGN gas,
where the conditions allow this thermalization, the temperature of the hot, Compton
thick pair plasma can reach 109 K. Such gas can contribute to the observed
high-energy spectrum. It can up-scatter soft (UV) emitted photons and even produce
some free-free electron-positron radiation.
25
128. Chapter 2. Non-Thermal Processes
2.6 Bremsstrahlung (Free-Free) Radiation
Free-free radiation, formally, is thermal radiation. However in the case of AGNs, the
spectral shape is very dierent from that of a blackbody. The free-free emissivity due
to ion i of an element of charge Z whose number density is Ni is given by
4j = 6:8 1038Z2T1=2
e NeNigff (; Te;Z)eh=kTe ;
where gff (; Te;Z) is the velocity-averaged Gaunt factor, which accounts for
quantum-mechanical eects. This factor is always of the order unity and can change
slightly with frequency, in particular, at X-ray energies gff / 0:1. The
Bremsstrahlung radiation extends over a large range of energies and resembles, over
most of this range, a very
at (small spectral index) power law. One can integrate the
free-free emissivity over frequencies to obtain the total energy per unit volume per
second, Cff , where C indicates that this is also the cooling rate due to free-free
emission. The integration gives
Cff = 1:42 1027Z2T1=2
e NeNigffNeNi erg s1 cm3;
where gff is now the frequency average of the velocity-averaged Gaunt factor. This is
typically in the range 1.1-1.5.
26
129. Chapter 3
The IR and Sub-mm Regime
3.1 History
The use of IR techniques to measure AGN continua started in the 1970s with the
advent of the
130. rst sensitive IR detectors (Low Kleinmann, 1968). However, the IR
colours of Seyfert galaxies are only subtly dierent than those of normal galaxies
(Kuraszkiewicz et al., 2003), and the equivalent widths of the IR lines are not sucient
to use as a
131. nding mechanism. Thus, IR color surveys can have a large fraction of
false AGN, unless great care is taken.
The
133. nd AGN in the IR was based on Infrared
Astronomical Satellite (IRAS) data. de Grijp et al. (1987) showed that AGN had
systematically dierent 60 m / 25 m colours than normal galaxies. An alternative
approach Spinoglio Malkan (1989) was to obtain optical spectra of every IR-selected
galaxy. This was a follow-up of the idea of Huchra Burg (1992) to obtain optical
spectra of every optically-selected galaxy, but was not really a survey technique. The
latest use of the IR to
134. nd active galaxies is with the Two Micron All Sky Survey
(2MASS; Cutri et al. (2002)). In this survey, 60% of the objects with
J - K 2 are found to have the optical properties of AGN. This selection criterion is
bootstrapped by using the near-IR colors of known radio and optically-selected AGN
(Elvis et al., 1994), and thus will tend to
135. nd objects with similar properties. The
large space density of these IR-selected objects makes them a major contributor to the
AGN population.
27
136. Chapter 3. The IR and Sub-mm Regime
The far-IR (FIR) band of thousands of AGNs has been observed by IRAS, with limited
spatial resolution, and by Spitzer, with much improved resolution. The 2009 launch of
Herschel is the most recent development in this area. Broadband images with much
improved spatial resolution are now available between 70 and 500 m. Systematic
surveys have already produced high-quality photometry of hundreds of AGNs and their
host galaxies, up to redshift of 5 and beyond. Lower-sensitivity, high-resolution
spectroscopy over the FIR range is also provided by the Herschel instruments.
3.2 Observations and Detection
Most of the emission in the NIR and MIR bands is due to secondary dust emission.
Secondary in this context refers to emission by cold, warm, or hot dust grains that
are heated by the primary AGN radiation source.Primary refers to radiation that is
the direct result of the accretion process itself. The temperature of the NIR- and
MIR-emitting dust is between 100 and 2000 K. The dimensions of the dusty structure
emitting this radiation, in intermediate luminosity AGNs, is of order 1 pc. Most of the
thermal FIR emission is thought to be due to colder dust that is being heated by
young stars in large star-forming regions in the host galaxy. In powerful radio sources,
at least part of the FIR emission is due to non-thermal processes much closer to the
center. Broad and narrow emission lines are seen in the NIR-FIR part of the spectrum
of many AGNs. They are thought to originate in the broad- and narrow-line regions.
A very important aspect techniques which use one IR band or a combination of two IR
bands is the ability to detect highly obscured (Compton thick) AGNs. A large fraction
of such objects, especially at high redshift, do not show detectable X-ray emission, and
being type II sources, their optical spectrum is completely dominated by the host
galaxy. Such sources would not be classi
137. ed as AGNs based on their optical and X-ray
continuum properties. However, their mid-IR (MIR) spectrum is dominated by warm
dust emission, the result of the heating of the central torus by the central source. A
luminosity ratio like L(24 m)/L(R), where R is the red optical band, will be much
larger in such sources compared with inactive galaxies because the AGN light is
heavily obscured at the R-band. Spectroscopic follow-up of such objects can be used to
look for the unique emission-line spectrum of the AGN. Indeed, systematic searches in
uniformly scanned Spitzer
138. elds reveal a large number of Compton thick AGNs.
28
139. Chapter 3. The IR and Sub-mm Regime
Figure 3.1 shows a composite 0.3-30 m spectrum of intermediate-luminosity type I
AGNs. The emission longword of 1 m is due primarily to secondary radiation from
dust. The dip at 1 m is due to the decline of the disk-produced continuum on the
short-wavelength side and the rise of the emission due to hot dust on the other side.
Figure 3.1: A composite spectrum of type-I AGNs covering the range 0:340 m. The
observations were obtained by several ground-based telescopes and Spitzer and were
normalized to represent a typical intermediate-luminosity source.
(Netzer, 2013)
3.3 The Dusty Torus
Dust is the cornerstone of the uni
140. cation theory of active galactic nuclei (AGNs).
Essentially, all types of AGNs are surrounded by an optically thick dust torus and are
basically the same object but viewed from dierent lines of sight (Antonucci, 1993,
Urry Padovani, 1995). The large diversity in the observational properties of AGNs
(eg., optical emission-line widths and X-ray spectral slopes) is simply caused by the
viewing-angle-dependent obscuration of the nucleus: those viewed face-on are
un-obscured (allowing for a direct view of their nuclei) and recognized as Type I
Seyferts, while those viewed edge-on are Type II Seyferts, with most of their central
engine and broad line regions being hidden by the obscuring dust.
29
141. Chapter 3. The IR and Sub-mm Regime
Apparently, key factors in understanding the structure and nature of AGNs are
determining the geometry of the nuclear obscuring torus around the central engine and
the obscuration (ie., extinction, a combination of absorption and scattering) properties
of the circumnuclear dust. An accurate knowledge of the dust extinction properties is
also required to correct for the dust obscuration in order to recover the intrinsic
optical/UV spectrum of the nucleus from the observed spectrum and to probe the
physical conditions of the dust-enshrouded gas close to the nucleus.
The presence of an obscuring dust torus around the central engine was
142. rst indirectly
indicated by the spectropolarimetric detection of broad permitted emission lines
(characteristic of Type I Seyferts) scattered into our line of sight by free electrons
located above or below the dust torus in a number of Type II Seyferts (Heisler et al.,
1997, Tran, 2003) Direct evidence for the presence of a dust torus is provided by IR
observations. The circumnuclear dust absorbs the AGN illumination and re-radiates
the absorbed energy in the IR. The IR emission at wavelengths longward of 1 m
accounts for at least 50% of the bolometric luminosity of Type II Seyferts. For Type I
Seyferts, 10% of the bolometric luminosity is emitted in the IR. A near-IR bump
(excess emission above the 2 10 m continuum), generally attributed to hot dust
with temperatures around 1200-1500 K (near the sublimation temperatures of
silicate and graphite grains), is seen in a few Type I Seyferts (Barvainis, 1987,
Rodrguez-Ardila Mazzalay, 2006). Direct imaging at near- and mid-IR wavelengths
has been performed for several AGNs and provides constraints on the size and
structure of the circumnuclear dust torus (Elitzur, 2006). Spectroscopically, the 10 m
silicate absorption feature and the 3.4 m aliphatic hydrocarbon absorption feature
are widely seen in heavily obscured Type II Seyferts; in contrast, the 10 m silicate
emission feature has recently been detected in a number of Type I Seyferts.
To properly interpret the observed IR continuum emission and spectroscopy as well as
the IR images of AGNs, it requires a good understanding of the absorption and
emission properties of the circumnuclear dust. To this end, one needs to know the
composition, size, and morphology of the dust - with this knowledge, one can use Mie
theory (for spherical dust) to calculate the absorption and scattering cross sections of
the dust from X-ray to far-IR wavelengths, and then calculate its UV/optical/near-IR
obscuration as a function of wavelength, and derive the dust thermal equilibrium
temperature (based on the energy balance between absorption and emission) as well as
30
143. Chapter 3. The IR and Sub-mm Regime
its IR emission spectrum. This will allow us to correct for dust obscuration and
constrain the circumnuclear structure through modeling the observed IR emission and
images. The former is essential for interpreting the obscured UV/optical emission lines
and probing the physical conditions of the central regions; the latter is critical to our
understanding of the growth of the central SMBH.
However, little is known about the dust in the circumnuclear torus of AGNs. Even our
knowledge of the best-studied dust - the Milky Way interstellar dust - is very limited.
Figure 3.2: A HST image of the gas and dust disk in the active galactic nucleus of
NGC 4261.
(Ngc4261 by Clh288 at en.wikipedia. Licensed under Public domain via Wikimedia
Commons)
3.4 IR Spectra
The value of spectral index () is (almost) constant in the IR region of the spectrum of
an AGN, evident from Figure 3.3. The thermal IR bump is due to the emission from
warm (T . 2000 K) dust grains.
31
144. Chapter 3. The IR and Sub-mm Regime
Figure 3.3: A depiction of the typical features in the continuum observed for many
AGNs.
(Tengstrand et al., 2009)
3.4.1 The 1 m Minimum
The existence of the IR bump longward of 1 m has led many authors to conclude that
this emission must be thermal, as the required temperatures are in the right range
(T . 2000 K) for hot dust in the nuclear regions. Sanders et al. (1988) have shown
that a minimum in the SED at 1 m is a general feature of AGNs. The hottest dust
has a temperature of 2000 K; at higher temperatures, dust grains sublimate. This
upper bound of the temperature explains the constancy of the frequency where the
NIR spectrum is the weakest, ie., at the Wien cut-o at a 2000 K blackbody.
One can de
145. ne a `sublimation radius' as the minimum distance from the AGN at
which grains of a given composition can exist. The dust grains closest to an AGN
probably are graphite rather than silicate, as graphite has a higher sublimation
temperature. The sublimation radius for graphite grains is
r = 1:3L1=2
uv46T2:8
1500 pc;
32
146. Chapter 3. The IR and Sub-mm Regime
where Luv46 is the central source UV luminosity in units of 1046 erg s1, and T1500 is
the grain sublimation temperature in units of 1500 K (Barvainis, 1987).
3.4.2 IR Continuum Variability
Clear evidence that the hot dust scenario for the origin of the IR continuum has some
merit has been provided by the IR continuum variability characteristics. Unlike
UV/optical variability with little if any time delay, the IR continuum shows the same
variations as the UV/optical continuum, but with a signi
147. cant time delay. This is
interpreted as a light-travel eect which occurs because of the separation between the
UV/optical and IR continuum-emitting regions; whereas the UV/optical emission
arises in a very compact region, the IR emission arises in dust that is far away from
the central source. The variations occur as the emissivity of the dust changes in
response to the UV/optical continuum that heats it. Within the sublimation radius,
dust is destroyed. Farther out, however, it survives and is heated by the UV/optical
radiation from the central source to approximately the equilibrium blackbody
temperature. The IR continuum arises as this energy is re-radiated by the dust. In the
FIR, the only AGNs that are found to vary are radio-loud sources.
3.4.3 The Submillimeter Break
Observations of the FIR to sub-mm portion of AGN spectra have been made in a
limited number of cases (Chini et al., 1989, Edelson Malkan, 1987, Hughes et al.,
1993). These observations show that the sub-mm SED decreases rather sharply as one
goes to longer wavelengths, so abruptly that in at least a few cases the spectral index
longward of the sub-mm break must be less than the value of -2.5 expected in the case
of a synchrotron self-absorbed spectrum (ie., F / v5=2). At these long wavelengths, a
thermal spectrum can produce a cut-o this sharp because the emitting eciency of
small grains is a sensitive function of frequency, Q /
, typically with
2 (Draine
Lee, 1984) so the emitted spectrum can have a very strong frequency dependence,
F / 2+
.
33
148. Chapter 4
The Radio Regime
4.1 History
The discovery of radio galaxies preceded the optical discovery of AGNs. It goes back to
the late 1940s and the early 1950s. Many of these sources were later shown to have
optical-UV spectra that are very similar to the various types of optically discovered
AGNs. The main features of many such sources are single- or double-lobe structures
with dimensions that can exceed those of the parent galaxy by a large factor and
strong radio cores and/or radio jets in some sources that coincide in position with the
nucleus of the optical galaxy.
About 10 percent of all AGNs are core-dominated radio-loud sources. This provides an
additional way to identify AGNs in deep radio surveys by correlating their radio and
optical positions. Stars are extremely weak radio sources, and hence an optical point
source that is also a strong radio source is likely to be a radio-loud AGN. The
positional accuracy of optical and radio telescopes is one arcsec or better, and there is
hardly any problem in verifying that the radio and optical emitters are one and the
same source. Most of the early AGN samples were discovered in this way. A
well-known example is the 3C radio sample, which includes some of the most powerful
radio-loud, early-discovered AGNs such as 3C 48 and 3C 273.
34
149. Chapter 4. The Radio Regime
4.2 The Loudness of AGNs
Like optically classi
150. ed AGNs, there are broad-line radio galaxies (BLRGs), the
equivalent of the Type I sources; narrow-line radio galaxies (NLRGs), the
spectroscopic equivalent of Type II AGNs; and even weak-line radio galaxies
(WLRGs), the equivalent of LINERs. While most AGNs show some radio emission,
there seems to be a clear dichotomy in this property. Hence, usually, the radio
loudness parameter, R, is used to separate radio-loud from radio-quiet AGNs. R is a
measure of the ratio of radio (5 GHz) to optical (B-band) monochromatic luminosity,
R = L(5 Ghz)
L(4400A
)
= 1:5 105 L(5 Ghz)
L(4400A
)
;
where L(5 Ghz) and L(4400 A) represent the value of L at those energies. The
dividing line between radio-loud and radio-quiet AGNs is usually set at R = 10.
Statistics of a large number of AGNs show that about 10 percent of the sources are
radio loud, with some indication that the ratio is decreasing with redshift.
Much of the radio emission in radio-loud AGNs originates in a point-like radio core.
The spectrum of such core-dominated radio sources suggests emission by a
self-absorbed synchrotron source. Except for the self-absorption low frequency part,
the spectrum is represented well by a single power law, F / R. Sources with
R 0:5 are usually referred to as
at-spectrum radio sources, and those with
R 0:5 are steep-spectrum radio sources. There is a clear connection between the
radio structure and the radio spectrum of such sources. Steep-spectrum radio sources
show lobe-dominated radio morphology and are also less variables. Flat-spectrum
sources have in general higher luminosity cores, larger amplitude variations, and weak
or undetected lobes. This dichotomy is interpreted as a dependence on the viewing
angle to the core. In steep-spectrum sources, one is looking away from the direction of
the nuclear radio jet, and the radio emission is more or less isotropic. In
at-spectrum
sources, we are looking at a small angle into the core. The intensity is boosted due to
the relativistic motion of the radio-emitting particles, and the variations are ampli
151. ed.
In many cases, there is evidence for superluminal motion in such sources.
35
152. Chapter 4. The Radio Regime
4.3 The Fanaro-Riley Classi
154. rst noticed by Fanaro Riley (1974) that the relative positions of regions of
high and low surface brightness in the lobes of extragalactic radio sources are
correlated with their radio luminosity. This conclusion was based on a set of 57 radio
galaxies and quasars, from the complete 3CR catalogue, which were clearly resolved at
1.4 GHz or 5 GHz into two or more components. Fanaro and Riley divided this
sample into two classes using the ratio RFR of the distance between the regions of
highest surface brightness on opposite sides of the central galaxy or quasar, to the
total extent of the source up to the lowest brightness contour in the map. Sources with
RFR 0:5 were placed in Type I (FR-I) and sources with RFR 0:5 in Type II
(FR-II). It was found that nearly all sources with luminosity
L(178 Mhz) . 2 1025 h2
100 W Hz1 Sr1
were of Type I (FR-I) while the brighter sources were nearly all of Type II (FR-II).
The luminosity boundary between them is not very sharp, and there is some overlap in
the luminosities of sources classi
155. ed as FR-I or FR-II on the basis of their structures.
For a spectral index of ' 1 the dividing luminosity at 5 GHz is
L(5 Ghz) . 7 1023 h2
100 W Hz1 Sr1
At high frequencies, the luminosity overlap between the two classes can be as much as
two orders of magnitude. Various properties of sources in the two classes are dierent,
which is indicative of a direct link between luminosity and the way in which energy is
transported from the central region and converted to radio emission in the outer parts.
4.3.1 Fanaro-Riley Class I (FR-I)
Sources in this class have their low brightness regions further from the central galaxy
or quasar than their high brightness regions (Figure 1.4 and Figure 4.1). The sources
become fainter as one approaches the outer extremities of the lobes and the spectra
here are the steepest, indicating that the radiating particles have aged the most. Jets
are detected in 80% of FR-I galaxies. A jet can begin as one-sided close to the core,
36
156. Chapter 4. The Radio Regime
but beyond a few kiloparsec it becomes two-sided and continuous, with an opening
angle 8 that varies along its length. Along the jet the component of the magnetic
158. rst parallel to the jet axis, but soon becomes aligned
predominantly perpendicular to the axis.
FR-I sources are associated with bright, large galaxies (D or cD) that have a
atter
light distribution than an average elliptical galaxy and are often located in rich clusters
with extreme X-ray emitting gas (Owen Laing, 1989, Prestage Peacock, 1988) As
the galaxy moves through the cluster the gas can sweep back and distort the radio
structure through ram pressure, which explains why narrow-angle-tail or
wide-angle-tail sources, say, appear to be derived from the FR-I class of objects.
A typical FR-I galaxy is shown in Figure 4.1. This is the radio source 3C 449, which is
optically identi
159. ed with a galaxy of type cDE4 at a redshift of 0.0181, so that 1
corresponds to 255 h1
100 pc. There are twin jets that are straight for 30 from the
core, after which they deviate towards the west and terminate into diuse lobes. These
jets and outer lobes are mirror symmetric about an axis through the core. The jets are
generally smooth in appearance, but higher resolution observations show knots on a
smooth ridge of emission, the southern jet being more knotty than the northern one.
Within 10 of the nucleus, the surface brightness of the jets is much reduced. The
jets widen at a non-uniform rate close to the core, with the greatest expansion
occurring where the jets are faintest. Beyond 10 from the nucleus, the opening
angle is constant at 7. The emission from the jets is highly polarized, the average
polarization over the jets being 30%, and the projected magnetic
160. eld is
perpendicular to the jet axis.
4.3.2 Fanaro-Riley Class II (FR-II)
This class comprises luminous radio sources with hotspots in their lobes at distances
from the center which are such that RFR 0:5. These sources are called
edge-darkened, which was a terminology when the angular resolution and dynamic
range used in observing the classical sources was not always good enough to reveal the
hotspots as distinct structures. In keeping with the overall high luminosity of this type
of source, the cores and jets in them are also brighter than those in FR-I galaxies in
absolute terms; but relative to the lobes these features are much fainter in FR-II
37
161. Chapter 4. The Radio Regime
Figure 4.1: VLA map of the FR-I galaxy 3C 449 at 1465 MHz, with angular resolution
4.8 3.4 arcsec2. The peak
ux is 22.2 mJy per beam, with contours drawn at 5%
intervals, beginning with the -5% contour.
(Perley et al., 1979).
galaxies. Jets are detected in 10% of luminous radio galaxies, but in nearly all
quasars. The jets have small opening angles ( 4) and are knotty; the jet magnetic
162. eld is predominantly parallel to the jet axis except in the knots, where the
perpendicular component is dominant. Figure 4.2 shows an example of an FR-II
galaxy, which is a VLA map of the radio quasar 3C 47 made by Bridle et al. (1994).
The most striking feature of the jets in the FR-II class is that they are often one-sided,
as is clearly seen in Figure 4.2. Jet one-sidedness occurs at large (kpc) scales as well as
in the milli-arcsecond jets which are found in compact cores through VLBI
observations. The feature A in the jetted lobe is a hotspot, while feature H on the
38
163. Chapter 4. The Radio Regime
unjetted side looks like one, but does not qualify for being a hotspot according to the
criteria of Bridle et al. (1994).
Figure 4.2: VLA map of the FR-II quasar 3C 47 made at 4.9 GHz with 1:45
1:13 arcsec2 resolution. G is the core, A the jetted hotspot. H does not meet the
hotspot criteria of Bridle et al.
(Bridle et al., 1994).
FR-II sources are generally associated with galaxies that appear normal, except that
they have nuclear and extended emission line regions. The galaxies are giant ellipticals,
but not
164. rst-ranked cluster galaxies. The environment of FR-II sources does not show
enhanced galaxy clustering over the environment of randomly chosen elliptical galaxies
(Owen Laing, 1989, Prestage Peacock, 1988). Owing to the large dierences in
the nature of the host galaxies and the environments of the FR-I and FR-II sources, it
is possible that they are intrinsically dierent types of source not related to each other
through an evolutionary sequence.
4.4 Radio Lobes and Jets
Radio-loud sources usually consist of a radio core, one or two detectable jets, and two
dominant radio lobes. The radio-quiet sources are less luminous at radio wavelengths
by a factor of 103 to 104, consisting of a weak radio core and perhaps a feeble jet. The
39
165. Chapter 4. The Radio Regime
increased level of activity in radio-loud AGNs is not con
166. ned to radio wavelengths,
however; they also tend to be about three times brighter in X-rays than their
radio-quiet counterparts.
4.4.1 The Generation of Jets
The radio lobes are produced by jets if charged particles ejected from the central
nucleus of the AGN at relativistic speeds. These particles are accelerated away from
the nucleus in two opposite directions, powered by the energy of accretion and/or by
the extraction of rotational kinetic energy from the SMBH via the Blandford-Znajek
mechanism (Blandford Znajek, 1977). The jet must be electrically neutral overall,
but it is not clear whether the ejected material consists of electrons and ions or an
electron-positron plasma. The latter, being less massive, would be more easily
accelerated. The disk's magnetic
167. eld is coupled (frozen in) to this
ow of charged
particles. The resulting magnetic torques may remove angular momentum from the
disk, which would allow the accreting material to move inward through the disk.
The incredible narrowness and straightness of some jets means that a collimating
process must be at work very near the central engine powering the jet. A thick, hot
accretion disk around the SMBH could provide natural collimation by funneling the
out
owing particles, as shown in Figure 4.3. Because the accreting material retains
some angular momentum as it spirals inward through the disk, it will tend to pile up
at the smallest orbit that is compatible with its angular momentum. Inside this
centrifugal barrier, there may be a relatively empty cavity that can act as a nozzle,
directing the accreting gases outward along the walls of the cavity. However, producing
highly relativistic jets, as frequently observed, appear to be dicult to accomplish with
this nozzle mechanism.
Alternatively, magnetohydrodynamic (MHD) eects could play an important role in
accelerating and collimating the relativistic
ows.
4.4.2 The Formation of Radio Lobes
As a jet travels outward, its energy primarily resides in the kinetic energy of the
particles. However, the jet encounters resistance as it penetrates the interstellar
40
168. Chapter 4. The Radio Regime
Figure 4.3: A Sketch of the electromagnetic out
ows from the two sides of a rotating
magnetized accretion disk owing to the unipolar dynamo action.
(CYGAM (CYlindrical GAmma-ray Monitor), Russian Space Research Institute).
medium within the host galaxy and the intergalactic medium beyond. As a result, the
material at the head of the jet is slowed, and a shock front forms there. The
accumulation and deceleration of particles at the shock front cause the directed energy
of the jet to become disordered as the particles splash back to form a large lobe in
which the energy may be shared equally by the kinetic and magnetic energy.
The motion of the charged particles and magnetic
169. elds within the lobes of radio-loud
objects contain an enormous amount of energy. For Cygnus A Figure 4.4, the energy of
each lobe is estimated to be approximately 1053 to 1054 J, equivalent to energy
liberated by 107 supernovae!
41
170. Chapter 4. The Radio Regime
Figure 4.4: Contour images of the Cygnus A radio jet on various scales.
(Carilli et al., 1996).
4.4.3 Accelerating the Charged Particles in the Jets
The observations of jets are made possible by ineciencies in the transport of particles
and energy out to the radio lobes. The spectra of the radio lobes and jets follow a
power law, with a typical spectral index of ' 0:65. The presence of power-law
spectra and a high degree of linear polarization strongly suggest that the energy
emitted by the lobes and jets comes from synchrotron radiation.
The loss of energy by synchrotron radiation is unavoidable, and the relativistic
electrons in jets radiate away their energy after just 10,000 years or so. This implies
that there is not nearly enough time for particles to travel out to the larger radio lobes.
42
171. Chapter 4. The Radio Regime
This long travel time implies that there must be some mechanism for accelerating
particles in the jets and radio lobes. As one possibility, shock waves may accelerate
charged particles by magnetically squeezing them, re
ecting them back and forth
inside the shock. Radiation pressure may also play a role, but is alone not enough to
generate the necessary acceleration.
4.4.4 Superluminal Velocities
Although the standard model of jets and radio lobes requires a steady supply of
charged particles moving at relativistic speeds, evidence for such high velocities is
dicult to obtain. The absence of spectral lines in a power-law spectrum means that
the relativistic velocity of the jet material cannot be measured directly but must be
inferred from indirect evidence. The most compelling argument for relativistic speeds
involves radio observations of material ejected from the cores of several AGNs, with
so-called superluminal velocities. This eect is observed within about 100 pc of the
AGN's center and probably continues farther out.
Figure 4.5: The apparent superluminal motion of the M87 Jet.
43
173. rst hint of the violent heritage of today's galaxies was found by Edward A. Fath
(1908), who was observing the spectra of spiral nebulae. Although most showed an
absorption-line spectrum produced by the combined light of the galaxy's stars, NGC
1068 displayed six bright emission lines. In 1926, Edwin Hubble recorded the emission
lines of this and two other galaxies. Seventeen years later, Carl K. Seyfert reported
that a small percentage of galaxies have very bright nuclei that are the source of broad
emission lines produced by atoms in a wide range of ionization states. These nuclei
were nearly stellar in appearance.
5.2 Spectrum
Figure 3.3 is a rough schematic of the continuum observed for many types of AGNs.
The most notable feature of this SED is its persistence over some 10 orders of
magnitude in frequency. The wide spectrum is markedly dierent from the thermal
(blackbody) spectrum of a star or the combined spectra of a galaxy of stars, and one
can see that there are many non-thermal radiation processes that are going on at
dierent stages in the AGNs.
When AGNs were
174. rst studied, it was thought that their spectra were quite
at.
Accordingly, a power law of the form : F / was used to describe the
44
175. Chapter 5. The Optical-UV Regime
monochromatic energy
ux, F. The spectral index, , was believed to have a value of
' 1.
The power received within any frequency interval between 1 and 2 is
Linterval /
2 R
1
Fd =
2 R
1
F
d
= ln 10
2 R
1
Fd log10 ;
so that equal areas under the graph of F vs. log10 correspond to equal amounts of
energy. A value of ' 1 re
ects the horizontal trend seen at the IR bump of
176. gure 3.3.
The continuous spectra of AGNs are now known to be more complicated, involving a
mix of thermal and non-thermal emission. However F / is still used to
parameterize the continuum. typically has a value between 0.5 and 2 that usually
increases with increasing frequency, so the curve of log10 L (or log10 F) vs. log10
in Figure 3.3 is generally concave downward. In fact, the value of is constant over
only a limited range of frequencies, such as in the IR and visible regions of the
spectrum. The shape and polarization of the visible-UV spectrum indicates that it can
sometimes be decomposed into contributions from thermal sources (blackbody
spectrum, low polarization) and non-thermal sources(power law spectrum, signi
177. cant
polarization). The thermal component appears as the big blue bump in Figure 3.3,
which can contain an appreciable amount of bolometric luminosity of the source. It is
generally believed that the emission from the big blue bump is due to an optically
thick accretion disk, although some believe that free-free emission may be responsible.
5.2.1 The Optical-UV Continuum and the Accretion Disk
The best-understood disks are thin disks with or without X-ray emitting coronae. The
majority of intermediate- and high-luminosity AGNs found in large surveys are
thought to be powered by such disks. Thin-disk theory suggests that the SED of such
systems contains a broad wavelength band where the spectral slope (L / ) is in
the range 0-0.5. The classical thin-disk slope is = 1=3. X-ray emission from the
hot corona and X-ray re
ection from the surface of the disk are additional important
characteristics of such systems. Much observational eort has been devoted to the
measurement of and the characterization of the part of the continuum showing this
45
178. Chapter 5. The Optical-UV Regime
Figure 5.1: A Composite Optical-UV Spectra of AGNs
(Francis et al., 1991).
Figure 5.2: A General View of the Optical-UV SED of AGNs.
46
179. Chapter 5. The Optical-UV Regime
slope (the big blue bump). For MBH = 109M
180. and L=LEdd = 0:1, the peak of this
emission is predicted to be around 1000 A. Even the most sophisticated calculated
spectra are rather limited, and several models show spectra that dier considerably
from the schematic 1=3 dependence and also, the infrared ( 1 m) part of the SED
is dominated by non-disk emission, in particular, stellar emission by the host galaxy
and thermal emission by warm dust, presumably in the central torus. In radio-loud
AGNs, non-thermal emission can also contribute at this and even shorter wavelength
bands. This obscures the part of the spectrum where the standard thin-disk theory
predicts = 1=3. One must also consider the (yet hypothetical) possibility that
additional processes, related perhaps to disk winds, are taking place and changing the
observed SED.
5.3 Observations in the Optical-UV Region
Optical images of luminous Type-I AGNs show clear signatures of point-like central
sources with excess emission over the surrounding stellar background of their host
galaxy. The non-stellar origin of these sources is determined by their SED shape and
by the absence of strong stellar absorption lines. Type-II AGNs do not show such
excess. The luminosity of the nuclear, non-stellar source relative to the host galaxy
luminosity can vary by several orders of magnitude. In particular, many AGNs in the
local universe are much fainter than their hosts, and the stellar emission can dominate
their total light. For example, the V-band luminosity of a high-stellar-mass AGN host
can approach 1044 erg s1, a luminosity that far exceeds the luminosity of many local
Type-I AGNs. This must be taken into account when evaluating AGN spectra obtained
with large-entrance-aperture instruments. The relative AGN luminosity increases with
decreasing wavelength, and contamination by stellar light is not a major problem at
UV wavelengths. The optical-UV spectra shown in Figure 5.4 and Figure 5.5 represent
typical spectra of high-ionization luminous Type-I and Type-II AGNs. The added
high-ionization is needed to distinguish such sources from low-ionization Type-I and
Type-II sources. The Type-I spectrum is a composite composed of several thousand
spectra of dierent redshift AGNs. This is done to illustrate the entire rest wavelength
range of 900-7000 A
using only ground-based observations. The data used to obtain
this composite at 5000 A
are based on spectra of lower luminosity, low-redshift
47
181. Chapter 5. The Optical-UV Regime
Figure 5.3: Broadband spectral energy distributions (SEDs) for various types of
AGNs.
(Ho, 2008)
Figure 5.4: The average optical-UV SED of several thousand high-luminosity Type I
AGNs
(Vanden Berk et al., 2001)
48
182. Chapter 5. The Optical-UV Regime
objects, and the SED at those wavelengths is aected by host galaxy contamination.
The Type-II spectrum covers a similar range, but this time, the spectrum is a
combination of a ground-based optical spectrum with a space-borne (HST) UV
spectrum. The striking dierences between the high-ionization Type-I and Type-II
Figure 5.5: The spectrum of the low-luminosity, low-redshift type-II AGN NGC 5252.
(Tsvetanov et al., 1996)
spectra, which were the reason for the early classi
183. cation into Seyfert 1 and Seyfert 2
galaxies, are the shape and width of the strongest emission lines. Type-II AGNs show
only narrow emission lines with typical full-width-at-half-maximum (FWHM) of
400 800 km s1. In Type-I spectra, all the permitted line pro
185. les, indicate large gas velocities, up to 5000 10000 km s1 when
interpreted as owing to Doppler motion. The line ratios and line widths of the
49
186. Chapter 5. The Optical-UV Regime
Figure 5.6: Comparison of dierent broad-line pro
187. les in a typical Type-I AGN.
(Netzer, 2013)
forbidden lines in the spectra of Type-I sources are very similar to those observed in
Type-II spectra and indicate that the basic physics in the narrow line-emitting region
of both classes is the same. The broad emission lines can be used to map the gas
kinematics very close to the central BH and to measure the BH mass. Study of the
spectra of many thousand Type-I AGNs shows a considerable range in optical-UV
continuum slope but little if any correlation between the slope and Lbol. Some of the
observed dierences are attributed to a small amount of reddening in the host galaxy
of the AGN or other sources of foreground dust. Although the spectra shown here
clearly illustrate the large dierences in emission-line widths between Type-I and
Type-II sources, observational limitations can make it dicult to detect weak broad
emission lines. Slightly obscured or low-luminosity Type-I AGNs are occasionally
classi
188. ed as Type-II based on their stellar-like continua and narrow emission lines.
This can be the result of reddening of the broad wings of the H
189. line or a relatively
strong stellar continuum, especially in large-aperture, low-spatial-resolution
observations. Higher signal-to-noise (S/N), better-spatial-resolution observations of the
same sources reveal, in some cases, very broad wings in one or more Balmer lines. The
term broad emission lines, which is used to describe the permitted and semi-forbidden
lines in Type-I AGNs, does not necessarily mean similar widths for all lines in all
50
190. Chapter 5. The Optical-UV Regime
objects. The various broad emission lines show typically dierent widths, and in
general, the width re
ects the level of ionization of the gas, the source luminosity, and
the mass of the central SMBH. Historically, it was found that broad emission lines in
some Type-I AGNs are narrower than narrow emission lines in Type-II sources. A
well-known example is the subgroup of narrow-line Seyfert 1 galaxies (NLS1s). This
class of objects was historically de
194. in many Type-II AGNs. Evidently, the distinction between
Type-I and Type-II sources requires other criteria, such as the presence of a non-stellar
continuum; strengths of emission lines typical of Type-I sources such as FeII emission
lines; or the presence of a strong, unobscured X-ray continuum. There are also
dierences in the shape and even the velocity of the same line in dierent objects.
Some examples are shown in Figure 5.6. A similar remark should be made about the
width of the narrow emission lines. For example, the width of the strong [O III] 5007
line can depend on the mass of the central SMBH (or, more accurately, the mass of the
bulge). Thus [O III] 5007 lines with FWHM 1000 km s1 are commonly observed
in high-redshift, large-MBH, large-Lbol AGNs.
5.4 Discovery by Optical-UV Properties
As said, typical AGN SEDs are dierent in several ways from stellar SEDs, in which
they cover a broader energy range and do not resemble a single-temperature blackbody.
This dierence provides a simple and ecient way of discovering AGNs using
broadband multicolor photometry. Several color combinations, based on three-band
and
195. ve-band photometry, are useful in separating AGNs from stars by their color.
Earlier methods were based on a UVB photometry in large areas of the sky. This
three-band system is useful for discovering low-redshift sources but fails to detect many
high-redshift objects because the spectrum gets eectively red and resembles the colors
of nearby stars. In addition, even the low-redshift AGNs can be confused with the
local population of hot white dwarfs. Some AGNs are intrinsically red, or reddened by
dust, which results in colors that are not very dierent from those of stars. Moreover,
intrinsically blue, high-redshift AGNs are eectively red due to the absorption of
their short-wavelength radiation by intergalactic gas. Figure 5.7 illustrates this and
51