J.-M. Wang1,2,3 - L. C. Ho4 - R. Staubert1
1 - Institut für Astronomie und Astrophysik, Abt. Astronomie,
Universität Tübingen, Sand 1, 72076 Tübingen, Germany
2 - Laboratory for High Energy Astrophysics, Institute of High
Energy Physics, CAS, Beijing 100039, PR China
3 - Alexander von Humboldt Fellow
4 - The Observatories of the Carnegie Institution of Washington,
813 Santa Barbara Street, Pasadena, CA 91101-1292, USA
Received 7 October 2002 / Accepted 29 July 2003
Abstract
We have assembled a sample of 37 radio-loud quasars that have been
imaged with the Hubble Space Telescope in order to investigate
their black hole masses, accretion rates, and the structure of their
accretion disks. The black hole masses were estimated from the
luminosities of the host galaxies, and the accretion powers were extrapolated
from the emission-line luminosities. The majority of the quasars have masses
in the range
.
Their accretion rates,
times the Eddington rate, suggest that
most of the objects possess standard optically thick, geometrically thin
accretion disks, in some cases perhaps accompanied by an optically thin
advection-dominated component. The coexistence of strong radio emission
and a standard disk conflicts with recent models for jet formation. We
discuss modifications of the standard model that can resolve this discrepancy.
We find there is a strong correlation between the accretion rate and
the extended radio luminosity. This lends support to the idea that the extended
radio emission is somehow linked to the accretion disk. Lastly, we combine
the present sample of radio-loud quasars with the sample of BL Lac objects
studied by Wang et al. (2002) to reevaluate the unification picture
for radio-loud active galactic nuclei. Consistent with
current ideas for the unification of radio-loud sources, we
find that flat-spectrum radio quasars and FR II radio galaxies indeed
seem to belong to the same population, as do BL Lac objects and FR I radio
galaxies on the opposite end of the luminosity spectrum. However, some
members of the low frequency-peaked BL Lac objects may be more closely
associated with FR II rather than FR I radio galaxies. We describe how the
various subclasses of radio-loud sources can be viewed as a continuous
sequence of varying accretion rate.
Key words: BL Lacertae objects: general - galaxies: active - galaxies: jets - galaxies: Seyfert - quasars: general
According to the radio-loudness criterion proposed by Kellermann et al. (1989), quasars whose ratio of radio (5 GHz) to optical (B band) flux exceeds 10 comprise a special population of radio-loud objects (hereafter RLQs). As reviewed by Urry & Padovani (1995), the powerful radio emission is thought to originate from relativistic jets expelled by the accretion disk around a central supermassive black hole (BH). One of the unsolved problems in astrophysics is the exact physical relation between the jet and the accretion disk. Two observational approaches have been taken to investigate this relation in RLQs and active galactic nuclei (AGNs). On large (kpc to Mpc) scales the relation between the jet radio/kinetic luminosity and the narrow-line luminosity has been examined by many authors (e.g., Baum & Heckman 1989a,b; Rawlings & Saunders 1991). It has been found that the narrow-line luminosity is correlated with the kinetic luminosity from the radio lobes. On smaller scales, Celotti & Fabian (1993) and Celotti et al. (1997) used radio data from very long-baseline interferometry to estimate the kinetic luminosities in a large sample of radio-loud objects. Both attempts reached the same conclusion: the relativistic jet is linked with the accretion disk. A more direct connection between the jet and the accretion disk has been found in the RLQ 3C 120 (Marscher et al. 2002). However, how a relativistic jet actually forms from the accretion disk remains an outstanding puzzle (Meier 2001; Blandford 2002).
Table 1: The sample of radio-loud quasars.
The situation in Galactic BH candidates suggests that the formation of the jet
is somehow linked with the accretion rate. The radio emission from these
objects becomes quenched when the system is in a soft (high accretion rate)
state, as seen in GX 339-4 (Fender et al. 1999), GRS 1915+105
(Harmon et al. 1997; Fender et al. 1999), GRO J16655-40 (Harmon et al.
1995), Cynus X-3 (McCollough et al. 1999), and Cygnus X-1 (Brocksopp et al.
1999). These objects may be instructive for the case of AGNs, although we
should be cautious that some of their properties do not scale linearly with
BH mass. The relation between the "states'' of the accretion disk and jet
production in supermassive BH accretion systems is very poorly understood.
![]() |
Figure 1: Distribution of redshift, MR(host), and MR(QSO) for the present sample. |
Open with DEXTER |
There are two impediments in solving this problem. First, the mass of the
central BH is highly uncertain. BH masses derived from fitting the "big blue
bump'' (Shields 1978; Malkan & Sargent 1982), for example, are coupled to the
accretion rate, which is not known independently. Second, the observed
continuum of RLQs, even at optical and ultraviolet wavelengths, may be
significantly contaminated by nonthermal emission from the relativistic
jet. With the exception of objects with prominent big blue bumps, it is
nontrivial to estimate the thermal component of the continuum. Thus, it is
difficult to reliably deduce the state of the accretion disk, which can
be described conveniently by the dimensionless accretion rate
,
where
is the accretion rate,
yr-1 is the Eddington accretion
rate defined by
,
erg s-1 is the Eddington luminosity, and
is the radiative efficiency (Frank et al. 1992).
The situation has improved dramatically with the availability of Hubble Space Telescope (HST) observations of quasar host galaxies. Relatively reliable luminosities for the host galaxies are now available. From studies of nearby inactive galaxies, it has been established that the mass of the BH scales roughly linearly with the luminosity (or mass) of the bulge component of the host galaxy (Kormendy & Richstone 1995; Magorrian et al. 1998; Ho 1999; Kormendy & Gebhardt 2001). Thus, we can use the luminosities of the quasar host galaxies, which are largely dominated by the bulge component, to estimate BH masses for quasars.
Table 2: Line luminosities.
Accretion disks have complicate structures for different
and the
viscosity
(Chen et al. 1995). For a disk with a low accretion rate,
three possible solutions have been suggested: optically thin advection-dominated
accretion flows (ADAF; Narayan & Yi 1994), advection-dominated inflow-outflows
(ADIOs; Blandford & Begelman 1999) and convection-dominated accretion flows
(CDAF; Narayan et al. 2000; Quataert & Gruzinov 2000).
Whether low
necessarily leads to low
,
however, remains
controversial.
When
,
the disk has a standard optically thick,
geometrically thin structure (Shakura & Sunyaev 1973, hereafter SS),
which has been extensively studied. When
,
an optically thick ADAF or super-Eddington accretion flow (SEA)
with slim geometry is formed (Begelman & Meier 1982; Abramowicz et al. 1988;
Chen & Taam 1993). Such a disk has two possible structures: an inhomogeneous
structure (split into rings) due to the photon-bubble instability (Begelman
2002), or a homogeneous disk with photon trapping (Wang & Zhou 1999; Ohsuga
et al. 2002). Once the mass of the BH and the accretion luminosity are known,
we can constrain the state of the accretion disk, which provides possible
clues to jet formation.
Recently, a number of authors have suggested that flat-spectrum radio quasars, based on their very high kinetic luminosities, are powered by super-Eddington accretion flows (Ghisellini & Celotti 2001; Cavaliere & D'Elia 2002; Maraschi & Tavecchio 2003). It would be highly desirable to test this and other hypotheses by obtaining a reliable estimate of the accretion rates of RLQs. This is the purpose of this paper.
The most reliable measurement of quasar host galaxy luminosities currently
come from HST observations. Our analysis uses RLQs with available
HST imaging, based mainly on three samples (see Table 1): (1) Hamilton
et al. (2002), (2) McLure et al. (2001) and Dunlop et al. (2003), and
(3) Boyce et al. (1998, 1999). The B and V magnitudes are converted into
the R band following Fukugita et al. (1995).
We adopt the results from Hamilton et al. (2002)
if the objects have been measured several times in order to decrease the
systematic errors. Unfortunately, systematic errors in host galaxy
measurements are difficult to quantify, but uncertainties on the order
of 0.1-0.3 mag are not unexpected.
Table 1 also lists the absolute magnitudes of the quasar nuclei,
based on spatial decomposition of the HST images. Finally, we give
published values of the extended radio luminosity.
Following standard practice, the objects are divided into two
subclasses according to their radio spectral index n, defined by
:
flat-spectrum radio quasars (FSRQs) are those with
,
whereas steep-spectrum radio quasars (SSRQs) have n<-0.5,
where n is generally calculated between 1.4 and 5 GHz.
There are two objects without spectral index, but
they do not change our conclusions (see Tables 2 and 3).
The redshift range of the sample is
,
and the host galaxy
absolute magnitudes span
(Fig. 1). For our present
application, we use the empirical BH mass-bulge luminosity relation of
McLure & Dunlop (2002):
![]() |
(1) |
The observed optical-UV continuum of RLQs may be contaminated by nonthermal emission from the relativistic jet, rendering it difficult to estimate the thermal contribution to the continuum, and hence to estimate accretion rates. On the other hand, quasars, independent of their radio strength, usually have strong, well-observed broad emission lines, which leads to the possibility of using the broad-line region (BLR) emission-line strength to indirectly estimate the ionizing luminosity. Following Celotti et al. (1997), this is the approach we will take here. Celotti et al. (1997) showed that the BLR luminosities in RLQs are roughly equal to the kinetic luminosities of their jets. We assume that the BLR clouds are most likely photoionized by the thermal emission from the accretion disk, rather than by nonthermal emission from the jet. This assumption is supported by that the beamed radiation from the jet has great difficulties to ionize the clouds distributed within a large solid angle to produce a reasonable range of ionization parameter for emission lines. There is also ample evidence that BLR emission lines are photoionized by a thermal continuum (see, e.g., Netzer 1990). The underlying assumption here should be quite safe in statistical test.
Table 3: Results.
Table 2 lists emission-line fluxes we were able to assemble for our sample. We choose prominent broad emission lines, such as H
The fraction
of the thermal disk emission reprocessed by the BLR clouds
is roughly equal to the cloud covering factor. This assumption should be
statistically valid for large samples since in steady state the energy
absorbed by the BLR clouds is approximately that radiated as line emission.
Thus, the thermal emission from the accretion disk is
![]() |
(2) |
We recognize that, for any individual object, variability and internal
reddening may introduce uncertainties into
.
Source-to-source
variations in the covering factor or reprocessing efficiency may also add
additional scatter.
It is hoped, however, that the overall statistics of a large sample will not
be severely affected and that they are at least internally consistent.
Following Wang et al. (2002), we define the line accretion rate and
its dimensionless form as
![]() |
(3) |
![]() |
(4) |
![]() |
(5) |
Accretion disks in the optically thick, geometrically thin SS regime
obey
![]() |
(6) |
![]() |
(7) |
![]() |
Figure 2:
The distribution of a) ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
(8) |
![]() |
(9) |
Our estimation of the parameter
are subject to at least
three sources of uncertainty. First, the total line luminosity is estimated
from only a limited number of emission lines. Since the relative
strengths of the main strong lines are relatively well determined, and
quasars generally have fairly uniform spectra (e.g., Francis et al. 1991;
Zheng et al. 1997), this should not introduce a large source of uncertainty.
More worrisome is our
assumption of a constant cloud covering factor (10%),
which in reality is likely to have an intrinsic dispersion. And third,
even if we assume that the host galaxy magnitudes are perfectly measured,
the
relation still has a scatter of
dex
(McLure & Dunlop 2002). For concreteness, we assume that any individual
value of
may have an uncertainty of
dex. Although
this is not negligible, fortunately it is not sufficiently large to obscure
gross statistical trends in our sample. However, we caution against a
literal interpretation of the individual values of
and
listed in Table 3.
The three critical values of
define four regimes in accretion states,
as shown in Fig. 2b. The average values of
for FSRQs and
SSRQs are, respectively,
and
.
The corresponding values for
are
and
.
Given the uncertainties just
discussed, the two subsamples appear to be statistically indistinguishable.
Wang et al. (2002) present the distributions of
and
for
BL Lac objects and conclude that all BL Lacs have ADAFs. Comparing those
objects with the RLQs presented here, we find that the distributions of
overlap significantly, increasing smoothly from BL Lacs to RLQs.
This was noted by Scarpa & Falomo (1997), who used only Mg II as the
luminosity indicator. Our result considerably strengthens theirs. A
similar behavior is seen in the distribution of
.
As we discuss
later, these trends suggest that an evolutionary sequence, characterized by
different accretion states, may connect BL Lac objects to RLQs.
The normalized mass accretion rates,
,
span over a relatively large range, from 10-2 to 100.7. Some of the
FSRQs in our sample fall in the range of objects studied by Cavaliere & D'Elia
(2002), but we suggest that some of FSRQs, like SSRQs, may have hybrid ADAF+SS
disks. With the similarity in
between FSRQs and SSRQs, the only
difference between FSRQs and SSRQs is geometrical orientation, which has been
suggested (Urry & Padovani 1995).
Cavaliere & D'Elia (200202) suggest that an evolutionary sequence may be defined
for blazars, depending on their dimensionless accretion rates. FSRQs have
,
whereas BL Lac objects have
,
the two populations separated by a significant gap.
By contrast, our analysis suggests that BL Lacs and RLQs define a
continuous sequence in accretion rate. Although our conclusion is
based on two heterogeneous samples, we note that it is in general agreement
with the evolutionary scenario proposed by Böttcher & Dermer (2002).
![]() |
Figure 3:
The distribution of extended radio power at 5 GHz versus the a)
![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
We note that values of
and
are sensitive to
.
The most likely value of
is 0.3 (Narayan et al. 1998). However if
,
it will follow that
and
.
All the disks listed as
hybrid models of SS and ADAF will be SS disk in Table 3.
In any case, this does not affect the conclusion that most of the objects have
the standard accretion disks.
There is only one possible super-Eddington accretor (PKS 0454-22) in the
present sample. This nicely agrees with the result from an independent test
by Wang (2003), in which super-Eddington accretors have been searched in the
present sample and no super-Eddington accretor has been found. He used the
theoretical spectra emergent from slim disks (Wang et al. 1999)
and the empirical reverberation
relation to find a limit relation bewteen the BH mass and H
width as
in AGNs with
super-Eddington rates. With the help of Eq. (1),
the BH masses can be obtained, he found all the objects in the present sample
are located below this limit.
It is worth noting that, with the exception of 3C 273, our sample lacks very powerful FSRQs. This is an observational selection effect. It is difficult to measure the host galaxies of the most powerful quasars because of their very bright nuclei. We will later return to the issue of whether there exists super-Eddington RLQs that appear as very powerful FSRQs.
Two processes have been widely advocated for jet formation in AGNs: (1) energy extraction from the spin of the BH (Blandford & Znajek 1977, hereafter BZ) and (2) energy extraction from a disk wind (Blandford & Payne 1982, hereafter BP). Several variants of the BZ and BP models have been proposed. The results of the present study allow us to set some constraints on the theoretical models. In their investigation of the BZ process in the regime of the SS disk, Livio et al. (1999) conclude that both the BZ and BP power, even for Kerr BHs, are negligible compared with the radiative output from the accretion disk itself. The reason is that geometrically thin disks cannot produce a strong enough poloidal magnetic field for the energy extraction to be effective.
Ghosh & Abramowicz (1997) assume that the poloidal magnetic field can be
parameterized as
,
where
is the
maximum pressure in the inner region if the accretion rate
,
with
and
.
In such a region, the radiation pressure dominates over the gas pressure and
is independent of the accretion rate. Then, the BZ power is given by
![]() |
(10) |
Meier (2001) considers energy extraction from the spin of the BH and from the
disk itself, for both an SS disk and an ADAF. His model uses the middle-region
solution for the SS disk and a self-similar solution for the ADAF; it considers
both Schwarzschild and Kerr BHs. In this model the jet power for a Schwarzschild
BH is only
erg s-1 if
and
,
whereas
erg s-1 for a Kerr BH with the same mass and accretion rate. The present data clearly do not support
this model because many RLQs have SS disks that support much higher luminosities.
When the accretion disk is in an ADAF state, for example, an accretion rate of
onto a Schwarzschild BH radiates
erg s-1,
which may generate a radio power of
W Hz-1 at
1 GHz in the extended lobes, where
is the
fraction of the total jet power radiated in the radio band (Meier 1999). This
can account for the radio emission from BL Lac objects shown in Fig. 3;
it also implies that the BHs of BL Lac objects may not necessarily have large
spins. The jet power from an ADAF surrounding a Kerr BH is approximately
1044.1 erg s-1, corresponding to a radio power of
W Hz-1 at 1 GHz. This can account for RLQs in the
ADAF+SS regime. But how do we explain those RLQs that have much higher
radio powers and that contain pure SS disks?
From the above discussion, it is evident that the power output of the jet, whether generated through the BZ or BP processes, depends both on the model adopted for the accretion disk and the parameterization used for the poloidal magnetic field. Below we derive an improved formula for the jet power.
Following Meier (2001), the magnetohydrodynamic (MHD) power of the jet is
given by
![]() |
(11) |
![]() |
(12) |
The thermal instability arises from insufficient cooling. It is well known that
such an instability in the innermost regions of the disk can be suppressed by
advection, which acts as an efficient cooling process (Abramowicz et al.
1988; Chen & Taam 1993). Advection mainly converts the heat of the accreting
gas into entropy, rather than radiation. Consequently, the sound speed
()
and the scale height H are increased. It would be of interest to
include the role of advection cooling in the calculation of the BZ power.
We follow Chen's (1995) treatment of advection cooling in determining the
structure of the disk. The disk is assumed to undergo Keplerian rotation
at an angular velocity
,
and we assume conservation of mass,
conservation of angular momentum, and vertical equilibrium. Advection is
included as local cooling, and we neglect the corona. The energy balance
between the heat production (
)
and cooling can be written as
![]() |
(13) |
![]() |
(14) |
The solutions of the disk give the sound speed
![]() |
(15) |
![]() |
(16) |
![]() |
(17) |
![]() |
(18) |
![]() |
(19) |
![]() |
(20) |
The output powers given by Eqs. (19) and (20) are different from those found by Ghosh & Abramowicz (1997) and Meier (2001). First, the energy output based on Eqs. (19) and (20) depends on the accretion rate, whereas the model of Ghosh & Abramowicz (1997) does not. Second, the output is much higher than that given by Meier (2001) for the standard accretion disk model.
The MHD power for a Schwarzschild BH is due to the BP process. For a Kerr BH, the BP process also depends strongly on the spin because the metric itself drags the MHD fluid, which then contributes to the rotation of the magnetic field; thus, Eq. (20) covers both the BP and BZ processes. The above levels of MHD power, which is extracted from the BH and the disk in the standard SS regime, are sufficient to account for the highest radio luminosities seen in RLQs.
As shown by Xu et al. (1999), the mass outflow rate from the jet is given by
![]() |
(21) |
We note that Eq. (21) is not self-consistently obtained from
the disk model. Such a high outflow rate may take place in ADIOS models. Here
we stress that the Fe K
line in RLQs may be strongly affected
by the outflow. The case of 3C 120 adds credibility to the picture that
jet formation influences the innermost region of disk: "dips in X-ray
emission are followed by ejections of bright superluminal knots in the radio
jet'' (Marscher et al. 2002). The detectability of the Fe K
line in
radio-loud objects may also be influenced by contamination of
the X-ray continuum by non-thermal (beamed) emission.
The correlation between the extended radio and line emission has been explored
by a number of authors (e.g., Baum & Heckman 1989a,b; Rawlings &
Saunders 1991; Rawlings 1992; Falcke et al. 1995; Xu et al.
1999; Cao & Jiang 1999; Willott et al. 1999). The present sample allows
us not only to further test this correlation, but presents us with an
opportunity to explain it in the context of our improved model.
Figures 3a,b show that the extended radio emission is
evidently quite closely linked with
and
.
The linear
regression for the RLQs alone is
![]() |
(22) |
![]() |
(23) |
For comparison, we include the BL Lac objects studied by Wang et al. (2002)
that have extended radio luminosities given by Urry et al. (2000); there are
12 high frequency-peaked (HBL) and 15 low frequency-peaked (LBL) objects. The
BL Lac objects smoothly extend the sequence defined by the RLQs toward lower
values of
,
,
and
.
The correlations
are now enhanced, becoming
![]() |
(24) |
![]() |
(25) |
Following Meier (2001), we assume that 1% of the MHD energy will be
radiated as radio emission with a spectral index of -1. Then, for a Kerr BH
and using the relation
,
W Hz-1, or
![]() |
= | ![]() |
|
= | ![]() |
(26) |
In our sample, only PKS 0454-22 is a super-Eddington accretor ( ER = 100.7). The most powerful RLQs, which have powers in excess of 1047erg s-1, potentially may be super-Eddington accretors (Ghisellini & Celotti 2001). In this section we consider the upper luminosity limit of RLQs as a threshold set by super-Eddington accretion.
For a BH accreting well above the Eddington rate, much of the dissipated
energy will be advected into BH because most of the photons are trapped in
the flow by Thomson scattering (Wang & Zhou 1999; Ohsuga et al. 2002).
Photon trapping efficiently controls the radiated luminosity;
the radiated luminosity from the disk depends very weakly on the accretion
rate,
,
tending to be saturated.
For a self-similar disk (Wang & Zhou 1999), Eq. (15) reduces to
![]() |
(27) |
![]() |
(28) |
![]() |
(29) |
From the total rotational energy due to the spin,
,
it follows that the
spin-down timescale due to the BZ process is
![]() |
(30) |
![]() |
(31) |
A useful diagnostic for radio galaxies, as first shown by Owen & Ledlow (1994; see also Ledlow & Owen 1996 and Owen et al. 1996), is the plot of extended radio luminosity versus optical absolute magnitude. A diagonal line well separates the two radio populations: at a given optical luminosity (which in the case of radio galaxies is dominated by the host galaxy), FR IIs have much higher radio powers than FR Is (Fanaroff & Riley 1974). The physical interpretation of the Owen-Ledlow line has been much debated. Bicknell (1995) and Gopal-Krishna & Wiita (2001) suggest that it may arise naturally from the interaction of the jet with the interstellar or intracluster medium. Or perhaps the two classes reflect intrinsic differences in their central engines, such as BH spin (Baum et al. 1995; Meier 1999) or state of the accretion disk. Ghisellini & Celotti (2001) recently suggested that the dividing line corresponds to a transition in the accretion mode, from an SS disk (FR II) to an optically thin ADAF (FR I).
Urry et al. (2000) also examined BL Lac objects and RLQs in the
diagram, but the data for their RLQs are
mostly not from HST observations. Figure 3c gives a new
representation of the Owen-Ledlow diagram for RLQs and BL Lacs, where we
have substituted the optical luminosity on
the abscissa with BH mass
.
We have plotted our sample of RLQs
along with the sample of BL Lac objects studied by Urry et al. (2000). The
diagonal line is the Owen & Ledlow (1994) line for radio galaxies, adapted
from Ledlow & Owen (1996) after converting to our distance scale, the radio
luminosities from 1.4 GHz to 5 GHz assuming a spectral index of -1, and MR to
using Eq. (1).
Several features are noteworthy in this diagram. First, the Owen-Ledlow line that divides FR I and FR II radio galaxies seems to cleanly bisect the population of beamed radio-loud objects. Nearly all HBLs fall below the line, and all but two FSRQ lies above the line. Second, the population of LBLs straddle the divide, occupying the region between HBLs and FSRQs. Third, as with the FSRQs, essentially all SSRQs lie above the line. And fourth, at least some BL Lac objects, mostly of the LBL variety, appear to be more closely associated with FR II radio galaxies; this is in agreement with the suggestion of Kollgaard et al. (1996) that some radio-selected BL Lac objects may be beamed FR II sources.
The availability of accretion rates enables us to consider radio-loud AGNs in
a more unified, physical context. Wang et al. (2002) show that most
BL Lac objects have accretion rates below the critical value
that demarcates the transition from a standard
SS disk to an optically thin ADAF. Moreover, they find that HBLs have
characteristically lower accretion rates than LBLs. This study demonstrates
that nearly all RLQs have
.
It is important to note
that, at least for the sample considered here, most of the sources occupy a
relatively narrow range of BH masses,
with a spread
of
0.5 dex. The various subpopulations of radio-loud AGNs can therefore
be cast into a physical, perhaps evolutionary, sequence described by a single
variable - accretion rate. For the beamed population,
decreases
along the sequence FSRQ
LBL
HBL. When viewed at
large angles to our line of sight, the sequence becomes
FR II
(FR II or FR I)
FR I. Since the structure
of the accretion flow is governed by
,
the variation in
along
the sequence directly translates into a variation in the dominant mode of
accretion. The most powerful RLQs and FR II radio galaxies may contain
super-Eddington disks or optically thick ADAFs. The majority of RLQs and
FR IIs have SS disks. The low-power HBLs and some LBLs have pure optically thin
ADAFs. Objects that occupy the fuzzy boundary near the Owen-Ledlow line,
which comprise some LBLs and a minority of FR IIs, may have a hydrid structure
consisting of an SS disk plus an optically thin ADAF.
The basic elements of the above evolutionary scenario, based on complementary but different lines of evidence, have already been advocated by a number of authors (e.g., Ghisellini et al. 1998; Ghisellini & Celotti 2001, 2002; Ghisellini et al. 2002; Böttcher & Dermer 2002; Cavaliere & D'Elia 2002). Böttcher & Dermer (2002) and Cavaliere & D'Elia (2002) argue that RLQs evolve into BL Lacs rather than BL Lacs into RLQs (Böttcher & Dermer 2002). If the evolutionary direction is from RLQs to BL Lacs, the BH masses of RLQS should be systematically smaller than those of BL Lacs. However, the significant overlap in the BH mass distributions of the two groups (Fig. 3c; see also Falomo 2003) does not seem to support this single evolutionary direction. Given the current uncertainties in BH mass determinations for AGNs, it is probably premature to draw strong conclusions. Here, we merely wish to note that both evolutionary paths can, and probably do, occur. The detection of quasar remnants in nearby galaxies in the form of inactive massive BHs conclusively demonstrates that the cosmic evolution of AGN activity must be episodic (Richstone et al. 1998). The duty cycle for accretion is short, such that any individual massive BH is likely to have been activated (and deactivated) many times since it was formed.
We attempt to constrain the accretion rates and the mode of accretion in
RLQs using BH masses estimated from HST imaging of their host galaxies
and accretion luminosities derived from their emission-line spectra. The
quasars have large BH masses,
,
and
substantial accretion rates,
times the Eddington rate.
Their accretion rates suggest that most of the objects possess standard
optically thick, geometrically thin accretion disks, perhaps in combination
with an optically thin ADAF in some cases. Our sample has a deficit of
objects with super-Eddington disks, although this is likely due to a
selection effect in the sample.
Our data can also be used to test current ideas on jet formation. We
present an improved model for a standard optically thick, geometrically
thin disk that incorporates the effects of advection cooling. Our model can
explain the level of radio emission observed in RLQs. We discuss some
consequences of our model for the interpretation of Fe K
lines observed
in the X-rays.
Finally, we combine the present sample of RLQs with a recently analyzed sample of BL Lac objects to reevaluate the unification picture for radio-loud AGNs. The strong correlation found between the luminosity of the extended radio emission and the accretion rate supports the notion that radio jets are directly coupled to the accretion disk. In agreement with prevailing ideas, our results support the proposition that FSRQs and BL Lacs are the beamed counterparts of FR II and FR I radio galaxies, respectively. However, we argue that some fraction of the low frequency-peaked BL Lac sources in fact may be more closely associated with FR II radio galaxies. The various subclasses of radio-loud AGNs - ranging from high-luminosity sources (FSRQs and FR IIs) on the one end to low-luminosity sources (BL Lacs and FR Is) on the other - can be viewed as a continuous sequence of varying accretion rate.
Acknowledgements
The authors are grateful to the referee for helpful comments and suggestions. J. M. W. thanks D. L. Meier for useful discussions. He is grateful to the support from Alexander von Humboldt Foundation, the Hundred Talent Program of Chinese Academy of Sciences, and the Special Funds for Major State Basic Research Projects and NSFC. The research of L. C. H. is supported by the Carnegie Institution of Washington and by NASA grants from the Space Telescope Science Institute (operated by AURA, Inc., under NASA contract NAS5-26555).