Suoper Massive Black Hole Mass
Suoper Massive Black Hole Mass
Suoper Massive Black Hole Mass
Abstract Supermassive black holes reside at the centers of most, if not all, massive
galaxies: the difference between active and quiescent galaxies is due to differences in
accretion rate relative to the Eddington rate and thus radiative efficiency. In this contribution, methods for measuring the masses of supermassive black holes are discussed,
with emphasis on reverberation mapping which is most generally applicable to accreting
supermassive black holes and, in particular, to distant quasars where time resolution
can be used as a surrogate for angular resolution. Indirect methods based on scaling
relationships from reverberation mapping studies are also discussed, along with their
current limitations.
Keywords active galactic nuclei black hole reverberation mapping
1 Introduction
As recently as 20 years ago, whether or not active galactic nuclei (AGNs) were powered
by accretion onto supermassive black holes was widely regarded as an open question.
The theoretical arguments supporting gravitational accretion as the primary source of
power in AGNs were in place within a few years of the discovery of quasars (Zeldovich
& Novikov 1964; Salpeter 1964; Lynden-Bell 1969) and some two decades later Rees
(1984) convincingly argued that supermassive black holes were the inevitable endpoint
of any of the scenarios proposed to account for activity in galactic nuclei. But definitive
observational proof remained illusive.
Now, however, it is now generally accepted that black holes reside at the center of
most, if not all, massive galaxies, both quiescent and active. And, perhaps ironically, the
first convincing proof of the existence of supermassive black holes was not in AGNs, but
in quiescent galaxies. It was primarily the high angular resolution afforded by Hubble
B.M. Peterson
Department of Astronomy, The Ohio State University
Columbus, OH 43210 USA
Tel.: +1-614-292-2022
Fax: +1-614-292-2928
E-mail: [email protected]
Space Telescope that enabled determination of the masses of nuclear black holes; for
the first time, the dynamics of stars and gas in the nuclei of nearby galaxies could be
studied on scales smaller than the black hole radius of influence, RBH = GMBH /2 ,
where MBH is the black hole mass, is the velocity disperion of the stars in the hostgalaxy bulge, and G is the gravitational constant. Detection of supermassive black holes
in quiescent galaxies showed that AGNs are different from other galaxies not because
they harbor supermassive black holes in their nuclei, but because their supermassive
black holes are actively accreting mass at fairly high rates, typically more than 0.1%
of the Eddington rate.
The realization that supermassive black holes are ubiquitous led to improved understanding of quasar evolution. The AGN population at the present epoch is a small
fraction of what it was at its peak at 2 < z < 3 or so. The quiescent supermassive
black holes in most galaxies are clearly the remnants of the quasars of the distant past.
The demographics of black holes (e.g., Shankar 2009; Vestergaard & Osmer 2009; Shen
& Kelly 2012, and references therein) are thus of keen interest for understanding the
accretion history of the universe.
central mass as a free parameter. This has enabled measurement of some of the largest
known supermassive black holes (Macchetto et al. 1997; Bower et al. 1998; de Francisco,
Capetti, & Marconi 2008; Dalla Bont`
a et al. 2009).
All together, the number of supermassive black holes whose masses have been determined by modeling stellar or gas dynamics numbers over 70 (McConnell & Ma 2013).
This number is not likely to increase dramatically in the near term on account of the
difficulty in resolving RBH beyond the Virgo Cluster. However, reverberation mapping
presents a viable alternative for measuring black hole masses at large cosmological
distances via gas dynamics by substituting time resolution for angular resolution. The
major limitation is that it is only applicable to Type 1 (broad emission-line) AGNs,
a trace population, but one that was more predominant at higher redshift. Since the
focus of this discussion is on accreting black holes, emphasis on reverberation mapping
of AGNs does not seem misplaced.
2 The discrepancy between theory and observation was due to an oversimplified theory; it
was implicitly assumed in photoionization equlibrium modeling that all BLR clouds are
intrinsically identical. A successful photoionization model was one that correctly predicted the
emission-line intensity ratios in the emitted spectrum of some standard cloud. A key intensity
ratio is C iv 1549/C iii] 1909. Moreover, the very presence of C iii] 1909 set an upper limit
to the density as this line is collisionally suppressed above 109.5 cm3 . The first high-sampling
rate reverberation program (Clavel et al. 1991) showed that C iv 1549 and C iii] 1909 arise
at different distances from the central source, thus obviating the earlier arguments.
5
10
9
8
7
6
7
6.5
6
5.5
5
5450
5500
5550
Fig. 1 Optical continuum (top) and broad H emission-line (bottom) light curves for Mrk
335. The variations in H follow those in the continuum by 13.9 0.9 days. Grier et al.
(2012a,2012b).
the relationship between the continuum and emission-line variations can be written as
L(VLOS , t) =
(VLOS , )C(t ) d,
(1)
which is usually known as the transfer equation and (VLOS , ) is the transfer
function. Inspection of eq. (1) shows that (VLOS , ) is the observed response to a
-function continuum outburst.
3.2.3 Construction of a VelocityDelay Map
The transfer function can be constructed geometrically as it is simply the six-dimensional phase space of the BLR projected into the observable coordinates, line-of-sight
velocity (i.e., Doppler shift) and time delay relative to the continuum variations. It
is therefore common to refer to the transfer function (VLOS , ) as a velocitydelay
map. It should be clear that each emission line has a different velocitydelay map
because the combination of emissivity and responsivity is optimized at different locations of the BLR for different lines. To map out all of the BLR gas would require
velocitydelay maps for multiple emission lines with different response timescales.
Consider for illustrative purposes a very simple BLR model, a circular ring of gas
orbiting counterclockwise at speed vorb around the central source at a distance R.
Suppose that a distant observer sees this system edge-on, as shown in the upper part
of Figure 2, and define a polar coordinate system centered on the continuum source
with the angle measured from the observers line of sight. Two clouds are shown
at positions (R, ). Photons from a -function continuum outburst travel toward the
observer along the x axis. The dotted line shows the path taken by an ionizing photon
from the same outburst will reach the BLR cloud pictured in the lower half of the figure
after travel time R/c; an emission-line photon produced in response by the cloud and,
by chance, directed toward the distant observer travels an additional distance R cos /c,
where it is now the same distance from the observer as the continuum source. So relative
to the ionizing photons headed directly toward the observer from the continuum source,
the emission-line photons are delayed by the sum of these two dotted segments, i.e., by
= (1 + cos ) R/c.
(2)
The locus of points that all have the same time delay to the observer is labeled as an
isodelay surface in the top part of the figure; a moments reflection will convince the
reader that the isodelay surface is a paraboloid. The corresponding Doppler shifts of
the clouds at coordinate (R, ) are vorb sin . These transformations are general,
and a ring of radius R and orbital speed vorb projects in velocitydelay space to an
ellipse with axes 2vorb centered on VLOS = 0 and 2R/c centered on R/c. Here the
ring is pictured edge-on, at inclination 90o ; at any other inclination i, the projected
axes of the ellipse in velocitydelay space are correspondingly reduced to 2vorb sin i
and 2R sin i/c. Thus, a face-on (i = 0o ) disk projects in velocitydelay space to a
single point at (0, R/c); all of the ring responds simultaneously and no Doppler shift
is detected.
Generalization of this structure to a Keplerian disk is straightforward by simply
adding more rings such that vorb R1/2 . This is illustrated for a system of several
rings in Figure 3.
Isodelay surface
To observer
Time delay
2R/c
= (1+cos) R/c
R/c
VLOS = vorb sin
vorb
vorb
At this point, an assumption about how the individual clouds re-emit line radiation can be intrduced. The simplest assumption is that the line emission is isotropic,
i.e., () = , a constant. To transform this to the observable velocitytime-delay
coordinates,
d
( ) d = ()
d.
(3)
d
From eq. (2).
R
d
= sin .
(4)
d
c
It is simple to show then that
c
d
(5)
( ) d =
1/2
R (2c /R)
(1 c /2R)1/2
and that the mean response time for the ring is
R
( ) d
R
h i = R
= ,
( ) d
(6)
To observer
Time delay
Fig. 3 Similar to Fig. 2, except for a series of rings in circular Keplerian orbits. Note in
particular the Keplerian taper of the velocitydelay map at increasing time delay.
as is intuitively obvious. A more realistic assumption is that much of the line emission
is directed back toward the ionizing source because the BLR clouds are very optically
thick even in the lines. A simple parameterization is that () = (1 + A cos ) /2.
Isotropy is the case A = 0 and complete anisotropy (which is, incidentally, perhaps
appropriate for Ly 1215) corresponds to A = 1 (Ferland et al. 1992).
Using a similar transformation for the case of isotropic re-emission, it is easy to
show that
(VLOS ) dVLOS =
(7)
1/2 dVLOS .
2 /v 2
vorb 1 VLOS
orb
A useful measure of the line width is the line dispersion line ; for a ring, the line
dispersion is
1/2 vorb
2
line = hVLOS
i hVLOS i2
= 1/2 .
(8)
2
For comparison, for such a ring, FWHM = 2vorb .
A velocitydelay map for any other geometry and velocity field can be constructed
similarly.
3.2.4 Reverberation Mapping Results: VelocityDelay Maps
As noted above, both continuum and emission-line variations are usually not very large
(< 20%) on a reverberation timescale. This alone makes it very difficult to recover
a velocitydelay map from spectrophotometric monitoring data. Indeed, the relative
photometric accuracy must be extremely good: for ground-based spectra, for which
absolute spectrophotometry at even the 5% level is notoriously difficult to achieve, this
is usually accomplished by using the [O iii] 4959, 5007 narrow lines as an internal flux
calibrator. These lines arise in the spatially extended low-density narrow-line region, so
both the light-travel time and recombination time are sufficiently long to ensure that
Fig. 4 Left: Velocitydelay maps for the Balmer lines in Arp 151. Note the clear inflow
signature for time delays above 15 light days. Right: Velocitydelay maps for H, He i 5876,
and He ii 4686 in Arp 151. Bentz et al. (2010b).
10
Fig. 5 Velocitydelay maps for four AGNs. 3C 120 has a disk-like structure and evidence for
infall is apparent in each of these. Grier et al. (2013).
Fig. 6 Toy models of velocitydelay maps for spherical infall (top two panels) and a Keplerian
disk (lower left) and a thick shell of randomly inclined circular Keplerian orbits (lower right).
Grier et al. (2013).
11
delays larger than 15 days) and possibly a localized hot spot in the BLR disk (based
on the enhanced emission at velocitydelay coordinates (+2000 km s1 , 0 days). More
detailed modeling by Brewer et al. (2011) based on formalism developed by Pancoast,
Brewer, & Treu (2011) favors a thick disk-like BLR geometry at an inclination3 of
22 o , with infall favored, and a value for the central mass MBH = 3.2 (2.1)106 M .
The velocitydelay maps in Figure 5 are so recent that detailed physical modeling
has not yet been done. Some of these strongly hint at a disk-like geometry (e.g., 3C 120)
and in each case there is evidence for infall.
12
These results underpin our efforts to measure the masses of the central sources in these
objects, as discussed in the next session.
V 2 R
G
(9)
where V is the line width and R is the reverberation radius c . The quantity in parentheses that contains the two directly observable parameters has units of mass and is
sometimes referred to as the virial product. The effects of everything unknown
the BLR geometry, kinematics, and inclination are then subsumed into the dimensionless factor f , which will be different for each AGN, but is expected to be of order
unity. Presumably, individual values of f can be determined if there is some other way
of determining the black hole mass. In the absence of a second direct measurement,
it has been common practice to use the MBH relationship for this purpose. The
relationship between central black hole mass and bulge velocity dispersion that is seen
in quiescent galaxies (Gebhardt et al. 2000a; Ferrarese & Merritt 2000) is also seen
in AGNs (Gebhardt et al. 2000b; Ferrarese et al. 2001; Nelson et al. 2004) although
of course, the host-galaxy velocity dispersions are are much more difficult to measure
in AGNs because of the bright active nucleus and because even the nearest AGNs are
typically quite distant (Dasyra et al. 2007; Watson et al. 2008). By assuming that the
MBH relationship is the same in quiescent and active galaxies, it becomes possible
to compute a mean value for the scaling factor, which turns out to be hf i 5 (Onken
et al. 2004; Woo et al. 2010; Park et al. 2012), although it is noted that Graham et
al. (2011) argue that in practice this process has been oversimplified. Figure 7 shows
the MBH relationship for quiescent galaxies and AGNs using the assumption that
hf i = 5.25. The scatter around this relationship amounts to about 0.4 dex, which is
a reasonable estimate of the accuracy of the virial method of estimating black hole
masses.
Sometimes concern is expressed that the empirical value of hf i seems uncomfortably large for a truly virialized system. However, it must be kept in mind that AGN
unification stipulates that Type 1 AGNs are generally observed at low values of inclination, much closer to face-on than edge-on. Actually, the fact that hf i is as small as it
is tells us that the BLR must have a fairly significant velocity component in the polar
direction; it is surely not a flat disk.
To return to a point made earlier, reverberation mapping is a direct measure of
black hole mass, but it is a secondary method because, at the present time, it relies on
an independent method, the MBH relationship, to calibrate the mass scale through
determination of hf i.
13
Fig. 7 The MBH relationship. Quiescent galaxies are shown in black and AGNs are shown
in blue. Woo et al. (2010).
14
Table 1 Comparison of Black Hole Mass Measurements
Method
Direct methods:
Megamasers
Stellar dynamics
Gas dynamics
Reverberation
Indirect methods:
[9]
MBH
RL scaling[10]
NGC 4258
NGC 3227
NGC 4151
(Units 106 M )
38.2 0.1[1]
33 2[2]
25260[5]
N/A
N/A
720[3]
[6]
20+10
4
[7]
7.63+1.62
1.72
N/A
70[4]
[6]
30+7.5
22
46 5[8]
13
N/A
25.0
15
6.1
65
[1] Herrnstein
et al. (2005). [2] Siopsis et al. (2009). [3] Davies et al. (2006). [4] Onken et al.
(2007). [5] Pastorini et al. (2007). [6] Hicks & Malkan (2008). [7] Denney et al. (2010). [8] Bentz
et al. (2006b). [9] G
ultekin et al. (2009). [10] McGill et al. (2008).
ing radiation pressure simply leads to systematic underestimation of AGN black hole
masses.
15
Fig. 8 The RL relationship between the size of the BLR measured by the broad H response
as a function of the host-corrected AGN luminosity. Open circles are new or improved data
since Bentz et al. (2009b). Bentz et al. (2013).
clouds. The similarity of AGN spectra over many orders of magnitude in luminosity
(e.g., Dietrich et al. 1993; Vanden Berk et al. 2004) suggests that U and ne are approximately the same for all AGNs. By assuming that L Q(H),
R L1/2 .
(11)
The great utility of the ionization parameter formulation was first recognized by Davidson (1972), although he is not often credited with this insight, possibly because instead
of expressing U in terms of L and R, he used the ionizing flux F = L/4R2 , so the
distance did not appear explicitly.
Thus from the beginning of reverberation investigations, a relationship like equation
(11) was expected, and it was searched for even with the first marginally sampled and
undersampled reverberation data (e.g., Koratkar & Gaskell 1991; Peterson 1993). The
first observationally well-defined version of the RL relationship was by Kaspi et al.
(2000), who found R L0.7 . A refinement of the database led to a somewhat shallower
slope (Kaspi et al. 2005), but it was only when contamination of the luminosity by hostgalaxy starlight was accounted for (Bentz et al. 2006a, 2009b, 2013) that the slope was
close to the nave photoionization prediction. The most recent version of the RL
relationship for the broad H emission line and the optical continuum is shown in
Figure 8. It is also worth noting that the existence of the RL relationship has been
confirmed independently by gravitational microlensing studies (Guerras et al. 2013).
While the existence of the RL relationship is not surprising, the tightness of
the relationship is. The intrinsic scatter in the relationship seems to be 0.13 dex
(Figure 8). Typical good individual reverberation lag measurements are accurate to
about 0.09 dex, so, at least for H, the RL relationship does almost as well as an
actual reverberation measurement, provided one has an accurate host-corrected AGN
luminosity.
Unfortunately, H is the only emission line for which the RL relationship is empirically well-calibrated. There are, however, a limited number observations that seem to
16
indicate that a relationship similar to that between H and the AGN optical continuum
holds for C iv 1549 and the UV continuum as well (Kaspi et al. 2007).
17
C iv 1549, He ii 1640, O iii] 1663 complex there is an unidentified emission feature at 1600
A that is problematic (Fine et al. 2010). Mg ii 2798 is also difficult
because of contamination by broad blends of Fe ii.
There are intrinsic differences between the widths of certain emission lines in mean
(or single-epoch) and rms residual spectra. As an example, Ferland, Korista, &
Peterson (1990) find that the far wings of H do not vary in Mrk 590 and suggest
that this might be because the highest velocity BLR gas, presumably closest to the
continuum source, is optically thin in the ionizing continuum. As another example,
Denney (2012) finds dramatic differences in the C iv 1549 profiles between mean
and rms residual spectra, but that these are correlated with the emission-line profile.
In general, it seems that it should be possible to one way or another calibrate out
the differences between mean and rms residual spectra.
The most common, and intended, use of indirect methods is for surveys. Survey
spectra, however, are generally quite noisy, typically below a threshold where there
are not systematic errors of various type introduced by low signal-to-noise ratio
(S/N ) (Denney et al. 2009). Again, in principle, these effects are statistically correctable.
There are two other continuing problems in determining masses.
Many studies continue to use FWHM to characterize the line widths. This is, of
course, a very tempting thing to do because FWHM is generally easily measured and
less affected by blending than the line dispersion line . However, the reverberation
masses are calibrated using line , not FWHM. This actually matters a great deal
because the ratio FWHM/line is a strong function of line width, as shown in Figure
9.
A related problem is that many studies, particulary those based on lower S/N
survey data, fit single Gaussians to the emission lines. For a Gaussian, as is well
known, FWHM/line 2.35 which, as can be seen in Figure 9 is on average not a
bad approximation for a quasar emission line, but it is terrible at both the highwidth and low-width ends of the distribution. The problem again is that the line
profile is a strong function of line width.
The impact of using FWHM rather than line is, at fixed luminosity L (or, equivalently,
BLR radius R), to stretch out the distribution in mass, overestimating the highest
masses (with the broadest lines) and underestimating the lowest masses (with the
narrowest lines). This introduces a bias into the mass functions derived from these
measurements. Of course, it is possible to remove this bias through the clear relationship
between line and FWHM shown in Figure 9; in principle, line can be inferred, with
some loss of precision, from FWHM.
A fair question to ask at this point is why not use FWHM to characterize the
line widths in the original reverberation data instead of line ? There are a number of
reasons for preferring line over FWHM as the line-width measure:
1. Multiple reverberation measurements of the same source should always yield the
same mass. NGC 5548 is our ideal test case here, as there are over a dozen reverberation measurements of the lag and line width for H alone, with lags ranging
from a few days to nearly a month. For these measurements, line yields a more
consistent mass than does FWHM.
2. For AGNs in which multiple emission lines have been measured, the virial relationship is tighter for line than for FWHM (Peterson et al. 2004). This is essentially
the same test as above.
18
Fig. 9 Ratio of two measures of the emission line width for H in reverberation-mapped
AGNs, full-width at half maximum (FWHM) and line dispersion line as a function of line
width for the compilation of Peterson et al. (2004). Top panel is for mean spectra, bottom
panel is for rms residual spectra. Values in top panel are very approximate since no deblending
of contaminants (except narrow emission components) was undertaken.
3. Steinhardt & Elvis (2010) find that when they plot AGN luminosity as a function
of black hole mass for individual redshift bins, the quasars at the high-mass end of
the bin fall farther and farther below the Eddington limit, an effect they refer to
as the sub-Eddington boundary. This is, in fact, exactly the bias that would be
expected by using FWHM (and/or Gaussian fits) instead of line , which was indeed
the case for the line-width measurements used in this study (Shen et al. 2008);
the highest masses are overestimated and the lowest masses are underestimated,
which would appear to rotate a distribution parallel to the Eddington limit (i.e.,
at fixed Eddington rate) to something shallower. Rafiee & Hall (2011) measured
line directly from the line profiles for essentially the same sample of AGNs and
the sub-Eddington boundary effect either vanishes or is greatly reduced.
None of these arguments is iron-clad (e.g., the sub-Eddington boundary could be real),
but they certainly indicate that a more critical examination of how to characterize line
widths deserves consideration.
19
7 The Future
In addition to trying to capture the current state of the art, a review article should also
attempt to identify particularly important directions that might be undertaken in the
future. Based on this discussion, here are a few investigations that might be pursued
to improve measurement of the masses of black holes in AGNs:
Velocitydelay maps for high-ionization lines. As discussed in 3.2.4, the
Balmer lines are emitted by lower-ionization gas that seems to be infalling and at
least sometimes in a disk-like structure. Absorption-line studies suggest, by contrast, that the strong high-ionization lines in the UV arise in outflows. A complete
20
picture of gas flows in active nuclei will therefore require contemporaneous velocity
delay maps for both high-ionization and low-ionization lines.
Modeling BLR kinematics. Velocitydelay maps provide entirely new constraints that should allow tremendous improvements in forward modeling of the
BLR and, consequently, determination of the masses of the central black holes.
The radiusluminosity relationship at high luminosity. This is especially
important for the C iv line because of the potential importance of this line in
determining black hole masses at high redshift. Even the H RL relationship is
sparsely populated at high luminosity (Figure 8) and by sources for which the
light curves are not exceptionally well-sampled. It is important to quantify the
systematic effects and intrinsic scatter because of the potential importance of using
this relationship for cosmological applications (e.g., Watson et al. 2011).
Characterization of line widths. At the present time, there is no consensus
best practice for characterization of broad emission-line widths for black hole
mass determination. How this is done for single-epoch spectra is even more controversial and, indeed, the case that C iv in particular can be used to estimate black
hole masses with any confidence at all needs to be made.
Direct comparisons of methods of mass measurement. The opportunities
for this are limited, of course, by the necessity of resolving RBH (1) to relatively
nearby systems. As the precison of mass measurement improves, even in the immediately foreseeable future the efficacy of the comparisons will be limited by uncertainties in the distances to the nearest AGNs on account of their peculiar motions
relative to the Hubble flow (Bentz et al. 2013). Obtaining accurate distances to the
nearst AGNs is a matter of sone urgency since it will get a lot harder following the
inevitable retirement of Hubble Space Telescope.
Expanding the reverberation-mapped population. The sample of some 50 or
so AGNs that have been the targets of reverberation-mapping programs is certainly
plagued by selection effects: for the most part, the AGNs that have been the targets
of reverberation monitoring campaigns have been selected by some combination
of their apparent brightness, known variability, or favorable position in the sky.
Richards et al. (2011) note, for example, the absence in the reverberation sample
of quasars with blueshifted C iv emission. This is, of course, not surprising given
because such objects are found at high redshift and the reverberation sample is
mostly at z < 0.2. Certainly this is an area of concern that must be addressed in
the future.
Acknowledgements The author is grateful for support by the US National Science Foundation through grant AST-1008882 to The Ohio State University and for the kind hospitality
of the International Space Science Institute in Bern, where this work was first presented. The
author thanks Misty Bentz for providing Figures 4 and 8, Kate Grier for Figures 1, 5 and 6,
and Jong-Hak Woo for Figure 7.
References
D. Alloin, J. Clavel, B.M. Peterson, G.A. Reichert, & G.M Stirpe, in Frontiers of Space and
Ground-Based Astronomy: The Astrophysics of the 21st Century, ed. W. Wamsteker, M.S.
Longair, & Y. Kondo (Dordrecht, Kluwer, 1994).
Y. Andrillat & S. Souffrin, Astrophys. Letters, 1, 111 (1968).
R.R.J. Antonucci & R.D. Cohen, Astrophys. J., 271, 564 (1983).
J.N. Bahcall, B.Z. Kozlovsky, & E.E. Salpeter, Astrophys. J., 171, 467 (1972).
21
A.J. Barth et al., Astrophys. J., 732:121 (2011a).
A.J. Barth et al., Astrophys. J., 743:L4 (2011b).
M.C. Bentz, B.M. Peterson, R.W. Pogge, M. Vestergaard, & C.A. Onken Astrophys. J., 644,
133 (2006a).
M.C. Bentz, B.M. Peterson, R.W. Pogge, & M. Vestergaard, Astrophys. J., 694, L166 (2009a).
M.C. Bentz, B.M. Peterson, H. Netzer, R.W. Pogge, & M. Vestergaard, Astrophys. J., 697,
160 (2009b).
M.C. Bentz et al., Astrophys. J., 651, 775 (2006b).
M.C. Bentz et al., Astrophys. J., 662, 205 (2007).
M.C. Bentz et al., Astrophys. J., 705, 199 (2009c).
M.C. Bentz et al., Astrophys. J., 716, 993 (2010a).
M.C. Bentz et al., Astrophys. J., 720, L46 (2010b).
M.C. Bentz et al., submitted to Astrophys. J.(2013).
R.D. Blandford & C.F. McKee, Astrophys. J., 255, 419 (1982).
G. Bower et al., Astrophys. J., 492, L111 (1998).
B.J. Brewer et al., Astrophys. J., 733:L33 (2011).
E.M. Burbidge, ARAA, 5, 399 (1967).
J. Clavel et al., Astrophys. J., 366, 64 (1991).
S. Collin-Souffrin, D. Alloin, & Y. Andrillat, Astron. Astrophys., 22, 343 (1973).
N. Cretton, P.T. de Zeeuw, R.P. van der Marel, & H.-W. Rix, Astrophys. J. Suppl., 124, 383
(1999).
E. Dalla Bont`
a, L. Ferrarese, E.M. Corsini, J. Miralde Escud
e, L. Coccato, M. Sarzi, A. Pizzella,
& A. Beifiori, Astrophys. J., 690, 537 (2009).
K. Davidson, Astrophys. J., 171, 213 (1972).
R.I. Davies et al., Astrophys. J., 646, 754 (2006).
G. de Francisco, A. Capetti, & A. Marconi, Astron. Astrophys., 479, 355 (2008).
K.M. Dasyra, L.J. Tacconi, R.I. Davies, R. Genzel, D. Lutz, B.M. Peterson, S. Veilleux, A.J.
Baker, M. Schweitzer, & E. Sturm, Astrophys. J., 657, 102 (2007).
K.D. Denney, Astrophys. J., 759:44 (2012).
K.D. Denney, B.M. Peterson, M. Dietrich, M. Vestergaard, & M.C. Bentz, Astrophys. J., 692,
246 (2009).
K.D. Denney et al., Astrophys. J., 653, 152 (2006).
K.D. Denney et al., Astrophys. J., 721, 715 (2010).
M. Dietrich et al., Astrophys. J., 408, 146 (1993).
M. Dietrich, F. Hamann, J.C. Shields, A. Constantin, M. Vestergaard, F. Chaffee, C.B. Foltz,
& V.T. Junkkarinen, Astrophys. J., 581, 912 (2002).
A.C. Fabian et al., Nature, 459, 540 (2009).
G.J. Ferland, K.T. Korista, & B.M. Peterson, Astrophys. J., 363, L21 (1990).
G.J. Ferland & R.F. Mushotzky, Astrophys. J., 262, 564 (1982).
G.J. Ferland, B.M. Peterson, K. Horne, W.F. Welsh, & S.N. Nahar, Astrophys. J., 387, 95
(1992).
L. Ferrarese & H. Ford, Space Science Reviews, 116, 523 (2005).
L. Ferrarese & D. Merritt, Astrophys. J., 539, L9 (2000).
L. Ferrarese, R.W. Pogge, B.M. Peterson, D. Merritt, A. Wandel, & C.L. Joseph, Astrophys.
J., 555, L79 (2001).
S. Fine, S.M. Croom, J. Bland-Hawthorn, K.A. Pimbblet, N.P. Ross, D.P. Schneider, & T.
Shanks, MNRAS, 409, 591 (2010).
K. Gebhardt et al., Astrophys. J., 539, L13 (2000a).
K. Gebhardt et al., Astrophys. J., 543, L5 (2000b).
K. Gebhardt et al., Astrophys. J., 583, 92 (2003).
R. Genzel, F. Eisenhauer, & S. Gillessen, Rev. Mod. Phys. 82, 3121 (2010).
M. Gliozzi, L. Titarchuk, S. Satyapal, D. Prince, & I. Jang, Astrophys. J., 735:16 (2011).
A.W. Graham, C.A. Onken, E. Athanassoula, & F. Combes, MNRAS, 412, 2211 (2011).
J.E. Greene et al., Astrophys. J., 723, 409 (2010).
C.J. Grier et al., Astrophys. J., 744:L4 (2012a).
C.J. Grier et al., Astrophys. J., 755:60 (2012b).
C.J. Grier et al., Astrophys. J., 764:47 (2013).
E. Guerras, E. Mediavilla, J. Jimenez-Vicente, C.S. Kochanek, J.A. Mu
noz, E. Falco, & V.
Motta, Astrophys. J., in press [arXiv:1207.2042] (2013).
K. G
ultekin et al., Astrophys. J., 698, 198 (2009).
22
J.R. Herrnstein, J.M. Moran, L.J. Greenhill, & A.S. Trotter, Astrophys. J., 629, 719 (2005).
E.K.S. Hicks & M.A. Malkan, Astrophys. J. Suppl., 174, 31 (2008).
K. Horne, B.M. Peterson, S.J. Collier, & H. Netzer, PASP, 116, 465 (2004).
W. Ishibashi & T.J.-L. Courvoisier, Astron. Astrophys., 540, L2 (2012).
K. Iwasawa, G. Miniutti, & A.C. Fabian, MNRAS, 355, 1073 (2004).
S. Kaspi, W.N. Brandt, D. Maoz, H. Netzer, D.P. Schneider, O. Shemmer, Astrophys. J., 659,
997 (2007).
S. Kaspi, D. Maoz, H. Netzer, B.M. Peterson, M. Vestergaard, & B.T. Jannuzi, Astrophys. J.,
629, 61 (2005).
S. Kaspi, P.S. Smith, H. Netzer, D. Maoz, B.T. Jannuzi, & U. Giveon, Astrophys. J., 533, 631
(2000).
W. Kollatschny, Astron. Astrophys., 407, 461 (2003a).
W. Kollatschny, Astron. Astrophys., 412, L61 (2003b).
J.A. Kollmeier et al., Astrophys. J., 648, 128 (2006).
A.P. Koratkar & C.M. Gaskell, Astrophys. J., 370, L61 (1991).
J. Kormendy & D. Richstone, ARAA, 33, 581 (1995).
C.A. Kuehn, J.A. Baldwin, B.M. Peterson, & K.T. Korista, Astrophys. J., 673, 69 (2008).
H. Landt, M.C. Bentz, B.M. Peterson, M. Elvis, M.J. Ward, K.T. Korista, & M. Karovska,
MNRAS, 413, L106 (2011).
A. Laor, Astrophys. J., 505, L83 (1998).
D. Lynden-Bell, Nature, 223, 690 (1969).
F. Macchetto, A. Marconi, D.J. Axon, A,. Capetti, W. Sparks, & P. Crane, Astrophys. J.,
489, 579 (1997).
J. Magorrian et al., Astron. J., 115, 2285 (1998).
D. Maoz et al., Astrophys. J., 404, 576 (1993).
A. Markowitz et al., Astrophys. J., 593, 96 (2003).
A. Marconi, D.J. Axon, R. Maiolino, T. Nagao, G. Pastorini, P. Pietrini, A. Robinson, & G.
Torricelli, Astrophys. J., 678, 693 (2008).
N.J. McConnell & C.-P. Ma, submitted to Astrophys. J., [arXiv:1211.2816] (2013).
K.L. McGill, J.-H. Woo, T. Treu, & M.A. Malkan, Astrophys. J., 673, 703 (2008).
I.M. McHardy, MNRAS, in press [arXiv:1212.2854] (2013)
I.M. McHardy, E. Koerding, C. Knigge, P. Uttley, & R.P. Fender, Nature, 444, 730 (2006).
R.J. McLure & M.J. Jarvis, MNRAS, 337, 109 (2002).
L. Meyer, A.M. Ghez, B. Sch
odel, S. Yelda, A. Boehle, J.R. Lu, T. Do, M.R. Morris, E.E.
Becklin, & K. Matthews, Science, 338, 84 (2012).
M. Miyoshi, J. Moran, J. Herrnstein, L. Greenhill, N. Nakai, P. Diamond, & M. Inoue, Nature,
373, 127 (1995).
R. Morales & A.C. Fabian, MNRAS, 329, 209 (2002).
C.J. Nelson, R.F. Green, G. Bower, K. Gebhardt, & D. Weistrop, Astrophys. J., 615, 652
(2004).
H. Netzer & P. Marziani, Astrophys. J., 724, 318 (2010).
C.A. Onken, L.Ferrarese, D.Merritt, B.M. Peterson, R.W. Pogge, M. Vestergaard, & A.Wandel,
Astrophys. J., 615, 645 (2004).
C.A. Onken et al., Astrophys. J., 670, 105 (2007).
A. Pancoast, B.J. Brewer, & T. Treu, Astrophys. J., 730:139 (2011).
A. Pancoast et al., Astrophys. J., 754:49 (2012).
I.E. Papadakis, MNRAS, 348, 207 (2012).
D. Park, B.C. Kelly, J.-H. Woo, & T. Treu., Astrophys. J. Suppl., 203:6 (2012).
G. Pastorini et al., Astron. Astrophys., 469, 405 (2007).
M. Pastoriza & H. Gerola, Astrophys. Letters, 6, 155 (1970).
B.M. Peterson, PASP, 105, 247 (1993).
B.M. Peterson, in Advanced Lectures on the StarburstAGN Connection, ed. I. Aretxaga, D.
Kunth, & R. M
ujica (World Scientific, Singapore, 2001).
B.M. Peterson, L. Ferrarese. K.M. GIlbert, S. Kaspi, M.A. Malkan, D. Maoz, D. Merritt, C.A.
Onken, R.W. Pogge, M. Vestergaard, & A. Wandel, Astrophys. J., 613, 682 (2004).
B.M. Peterson, K.A. Meyers, E.R. Capriotti, C.B. Foltz, B.J. Wilkes, & H.R. Miller, Astrophys.
J., 292, 164 (1985).
B.M. Peterson, R.M. Wagner, D.M. Crenshaw, K.A. Meyers, P.L. Byard, C.B. Foltz, & H.R.
Miller, Astron. J., 88, 926 (1983). d
B.M. Peterson & A. Wandel, Astrophys. J., 521, L95 (1999).
23
B.M. Peterson & A. Wandel, Astrophys. J., 540, L13 (2000).
B.M. Peterson et al., Astrophys. J., 368, 119 (1991).
B.M. Peterson et al., Astrophys. J., 632, 799 (2005).
Popovi
L.C.
c et al., Astron. Astrophys., 529:A130 (2011).
A. Rafiee & P. Hall, Astrophys. J. Suppl., 194, 42 (2011).
M.J. Rees, ARAA, 22, 471 (1984).
G.T. Richards et al., Astron. J., 141:167 (2011).
E.E. Salpeter, Astrophys. J., 140, 796 (1964).
F. Shankar, New Astron. Rev., 53, 57 (2009).
A.I. Shapovalova et al., Astron. Astrophys., 509, 106 (2010).
A.I. Shapovalova et al., Astrophys. J. Suppl., 202:19 (2012).
Y. Shen, J.E. Greene, M.A. Strauss, G.T. Richards, & D.P. Schneider, Astrophys. J., 680, 169
(2008).
Y. Shen & B.C. Kelly, Astrophys. J., 746, 169 (2012).
C.L. Steinhardt & M. Elvis, MNRAS, 402, 2637 (2010).
C. Siopsis et al., Astrophys. J., 693, 946 (2009).
J. Thomas, R.P. Saglia, R. Bender, D. Thomas, K. Gebhardt, J. Magorrian, & D. Richstone,
MNRAS, 353, 391 (2004).
J. Tohline & D.E. Osterbrock, Astrophys. J., 210, L117 (1976).
H.D. Tran, Z. Tsvetanov, H.D. Ford, J. Davies, W. Jaffe, F.C. van den Bosch, & A. Rest
Astron. J., 121, 2928 (2001).
S. Tremaine et al., Astrophys. J., 574, 740 (2002).
M.-H. Ulrich, A. Boksenberg, M.V Penston, G.E. Bromage, J. Clavel, A. Elvius, G.C. Perola,
& M.A.J. Snijders, Astrophys. J., 382, 483 (1991).
M.-H. Ulrich & K. Horne 1996, MNRAS, 283, 748 (1996).
M. Valluri, D. Merritt, & E. Emsellem, Astrophys. J., 602, 66 (2004).
D.E. Vanden Berk, C. Yip, A. Connolly, S. Jester, & C. Stoughton in AGN Physics with the
Sloan Digital Sky Survey, ed. G.T. Richards & P.B. Hall (Astronomical Society of the
Pacific, San Francisco, 2004).
R.P. van der Marel, N. Cretton, P.T. de Zeeuw, & H.-W. Rix, Astrophys. J., 493, 613 (1998).
M. Vestergaard, Astrophys. J., 571, 733 (2002).
M. Vestergaard, Astrophys. J., 601, 676 (2004).
M. Vestergaard & P.S. Osmer, Astrophys. J., 699, 800 (2009).
M. Vestergaard & B.M. Peterson, Astrophys. J., 625, 688 (2005).
M. Vestergaard & B.M. Peterson, Astrophys. J., 641, 676 (2006).
A. Wandel, B.M. Peterson, & M.A. Malkan, Astrophys. J., 526, 579 (1999).
I. Wanders et al., Astrophys. J., 453, L87 (1997).
D. Watson, K.D. Denney, M. Vestergaard, & T.M. Davis, Astrophys. J., 740, L49 (2011).
L.C. Watson, P. Martini, K.M. Dasyra, M.C. Bentz, L. Ferrarese, B.M. Peterson, R.W. Pogge,
& L. Tacconi, Astrophys. J., 682, L21 (2008).
J. Woo et al., Astrophys. J., 716, 269 (2010).
X.-B. Wu, R. Wang, M.Z. Kong, F.K. Liu, & J.L. Han, Astron. Astrophys., 424, 793 (2004).
Ya.B. Zeldovich & I.D. Novikov, Sov. Phys. Dokl., 158, 811 (1964).
Y. Zu, C.S. Kochanek, & B.M. Peterson, Astrophys. J., 735:80 (2011).