2.2. Evolution - Formation and Sustenance of the Black Hole
Although the presence of the super-massive black hole gives an
explanation for the source of the immense power output from AGNs, it
raises the question of how the black hole itself is created and
continually sustained. An
answer to this was provided by
Hernquist (1989), who proposed that an
active nucleus can result from a star-burst episode
(12)
caused by a
disturbance in the interstellar medium of a galaxy involved in an
interaction with a
second galaxy
(13):
Increased star formation rates had previously been noted during such
mergers
(Larson & Tinsley 1978;
Lonsdale, Persson & Mathews 1984;
Cutri & McAlary 1985;
Kennicutt Jr. et al. 1987)
(14),
and many AGNs are known to host multiple nuclei (e.g.
Petrosian, Saakian & Khachian 1979;
Killatschny, Fricke & Netzer 1986). In this scenario, a flow of gas into
the galactic nucleus occurs when the interstellar gas in one of the
galaxies, which is compressed and shocked by the approach of the
second galaxy (15),
reconfigures itself
into a ``thin bar-like'' structure. As the merger continues, the gas
is forced along the bar until the accumulation of gas causes self
gravity to increase to such a point where, in a reversal of rôles,
the bar will tidally disrupt the colliding galaxy. Further work (e.g.
Combes & Gerin 1985;
Roos 1985;
Barnes & Hernquist 1991;
1992; 1996;
Solomon, Downes & Radford 1992)) has shown that such mergers can transport the
necessary ~ 1010
M of gas and stars to
the nuclear regions,
and Barnes & Hernquist (1991; 1992; 1996)
have shown, through computations, how the gas may be
triggered into a vigorous nuclear star-burst via the formation of an
accretion shock. The star-burst will occur within a ring of molecular
gas located close to the inner Lindblad resonance (ILR)
(16)
(e.g.
Schwarz 1981;
Michalodimitrakis & Terzides 1985;
Buta 1986;
Freeman 1996;
Byrd, Ousley & Dalla Piazza 1998;
Tacconi et al. 1998a; 1998b; 1999a)
(17)
Following this, the gas will be transported (possibly by a
molecular bar, Section 3) into the central
few pc and eventually
accreted via a dense (optically thick to Thomson scattering,
(Dopita et al. 1997) rotating disk into the very centre. If this gas
accretes at super-Eddington rates
(18)
the dust
accumulated will be capable of forming the obscuration required by
unified models (Section 1). As an alternative
to the black hole
being a result of the accretion continuing to its ultimate conclusion,
the black hole may form from the coalescence of stars as well as
instabilities within the nucleus
(Rees 1978). Whatever the case, as mentioned previously,
super-massive black holes seem to be an inherent feature of a galactic
nucleus. The growth of the black hole
(19)
is regulated by the ongoing star-burst with which it competes for the
inflowing material. While the presence of a star-burst will restrict
feeding to the black hole, there is expected to be an inner limit to
the proximity of this star formation, since tidal effects will exert shear
on the parent gas clouds, tearing them apart (e.g. the inner limit for
a molecular cloud of density 104 cm-3 approaching
a 108 M
black hole is 55 pc,
Lang 1980;
Maiolino & Rieke 1995). As well as this, ionising
radiation from the black hole will hinder star formation in molecular
clouds (Silk & Rees 1998). It should be noted, however, that
Sanders (1998) has calculated that compression of the gas clouds will
ignite star formation in the near tidal field of a point mass, thus
permitting the presence of stars relatively close to the black
hole (20):
In NGC 7214 the star formation rate is found to
increase with distance from the nucleus
(Radovich, Rafanelli & Barbou 1998) and in Circinus
the star-burst is seen to occur at ~ 200 pc from the nucleus
(Marconi et al. 1994). In fact
Maiolino et al . (1998) suggest that the star-burst luminosity on this scale is
comparable to the luminosity of the AGN, while being only ~ 3% of
this within 12 pc of the black hole, thus supporting the notion of an
inner limit.
![]() |
Figure 4. The CO 2 -> 1 first moment map of Circinus (see Curan et al. 1999) shown in relation to that of the HI distribution (Jones et al. 1999). On the right, the grey-scale shows the HI distribution exhibiting the bar and spiral arms. This is also apparent through the twisting in the position angle of the kinematic axis. HI image courtesy of Keith Jones and Bärbel Koribalski. |
In the case of non-interacting galaxies, it is not so clear how the material is transported to the nucleus, although it has been suggested that this may also occur along a bar (Kormendy 1979; Simkin, SU & Schwarz 1980; Jörsäter 1984; Combes & Gerin 1985; Buta 1986; Lindblad & Jörsäter 1987; Norman 1988; Shlosman, Frank & Begerman 1989; Thronson et al. 1989; Elmgreen 1994; Villa-Vilaró et al. 1995; Lindblad, Lindblad & Athanassoula 1996; Maiolino, Risaliti & Salvati 1999). Since Malkan, Gorjian & Tam (1998) find that less than a tenth of Seyfert galaxies exhibit signs of a past interaction (21), the bar is hypothesised to result from disk instabilities (22) Bars are known to exist in some solitary spiral galaxies (23) and even a weak bar is sufficient to funnel gas into the nucleus (Schwarz 1984; Schwarz 1985; Sanders 1989). Although this point is debatable (Ho, Filippenko & Sargent 1997), the transport of gas via a bar is confirmed by the survey of Sakamoto et al. (1998), who find larger gas concentrations in barred than in unbarred galaxies. After large scale bar formation has occurred, again collisions with the gas cause the clouds to lose momentum and consequently form the molecular ring close to the ILR. Since there is no hydrodynamic/gravitational equilibrium, within this ring (Section 3) intense star formation (or gas fountaining out from the centre) commences (Rees 1989; Daly 1990; Struck-Marcell 1991).
It should be noted that there exist many star-burst galaxies in
which no AGNs are observed to occur. As previously described in the
case of AGNs, the
star-bursts which dominate these galaxies' activity
(24),
are believed to be triggered in
disturbed, i.e. merging/barred, systems. One example of such a galaxy
is M82 (25)
which has the canonical star formation rate of
6
M
yr-1 in
its molecular ring
(Young & Scoville 1982). As in the case of Seyfert galaxies
(next section), the extent of the bulk carbon monoxide in this ring is
~ 200 pc
(Nakai et al. 1987), which raises the question of why some disturbed
galaxies continue to become AGNs while others, such as M82, will only
evolve as far as the star-burst stage
(26).
Norman (1988) suggests that
only a limited number of star-bursts reach the luminosity required for
the formation of an AGN, and
Sakamoto et al. (1998) postulate that all
galaxies may host AGNs which may be obscured by star formation in
molecular disks. This would occur when the disks are unstable due to the
high gas mass fractions
(27).
present (Hohl 1971; Ostriker & Peebles 1973) in radio-quiet
galaxies. Finally, since star-bursts are short lived ( ~ 107 yr,
Rieke & Low 1975; Hernquist 1989; Planesas, Colina & Perez-Olea 1997),
there is the possibility that star-burst
galaxies have simply not had sufficient time in which to evolve into
AGNs (28)
(Marconi et al. 1994; Genzel et al. 1995; Maiolino et al 1998). This idea
of star-burst events evolving into AGNs is confirmed by the work of
Oliva et al. (1995), who find that the star-bursts in Sy2s appear to be
older (29)
than those in
star-burst galaxies. Another point is that many Seyferts do not
exhibit any star-burst activity
(Filippenko, Ho & Sargent 1993)
(30)
and indeed may not statistically have
more activity than normal galaxies
(31)
(Pogge 1989; Heckman 1991; Carone 1992). This
result, however, appears to ``hold truer'' for Sy1s
(32)
(e.g. Oliva et al. 1999). In addition to this, surveys of Seyfert galaxies
(Heckman et al. 1989; Sahai et al. 1991; Curran 2000) show that, on
average, the CO luminosity, LCO, in Sy2s may
be as much as double that in Sy1s
(33).
Curran (2000) finds this result to be biased by
the LFIR ~ 1010
L
Seyferts and that for
higher FIR luminous (LFIR ~ 1011
L
)
galaxies, the mean CO luminosity for a given FIR luminosity is similar
between the two main classes. The results may suggest that the type 2
galaxies have more molecular gas (and hence star-burst activity) per AGN
luminosity than the type 1 Seyferts, again suggesting that the
star-burst activity is already exhausted in the Sy1s.
(12) A period
of enhanced star formation (e.g. Weedman 1983).
Back.
(13) AGNs are several times more likely
than normal
galaxies to have companion systems (e.g. Dahari 1984; MacKenty 1990).
Back.
(14) And in Seyfert galaxies (e.g.
Wilson 1988).
Back.
(15) Smith, Herter & Haynes (1998)
postulate that the internal
properties of the galaxies, rather than their proximity, has the
greatest effect on the transport of the gas.
Back.
(16) See e.g. Binney & Tremanine (1987).
Back.
(17) In
Circinus the bar has been observed by Jones et al. (1999),
Fig. 4. See the discussion at the end of
Section 3 for an
example on how the inner atomic gas disk compares with the molecular ring.
Back.
(18) When
0.2 (M / 108
M
)
M
yr-1,
i.e. when the
accretion rate exceeds that required to maintain the Eddington
luminosity; the limiting luminosity due to radiation pressure acting
on the in-falling ionised gas
(Blandford, Netzer & Woltjer 1990;
Frank, King & Raine 1992).
Back.
(19) The interested reader is referred to
Dopita (1998) (and references therein) for details on how the black
hole can position itself so as to accrete at an optimal rate.
Back.
(20) The absence of broad lines in Sy2s
may be due to the obscuration being comprised of stars and dust
(Fernandes & Telrevich 1995). Also,
Norman & Scoville (1988) propose that the lines may arise from the stellar
envelopes of red giant stars ionised by the ultra-violet radiation
from the black hole.
Back.
(21) Although
Simkin, Su & Schwarz (1980) find larger numbers than this. Also,
Kukula et al. (1999)
find that all of the four Seyferts which they have observed appear to
have been involved in past interactions.
Back.
(22) See Martin & Friedli (1999) for a review.
Back.
(23) These may be more common in Sy2s than Sy1s
(Pogge 1989), e.g. a bar appears to be absent in the Sy1 NGC 5033
(Thean et al. 1997). This would appear to complicate the unification
scheme (Section 1) but, as we shall see, evidence
suggests that as well as the difference in orientation
(Antonucci 1993),
Sy1s may generally be the more evolved Seyfert class.
Back.
(24) For general reference the reader is
referred to Larson & Tinsley (1978);
Joseph (1986);
Norman (1998); Elmgreen (1994)
Back.
(25) Although this may harbour weak AGN
(Wills et al. 1999).
Back.
(26) The IR energy distribution in M82
is in fact very
similar to that of Circinus, although the ratio of IR to H2O maser
luminosity is about 100 times higher
(Moorwood & Glass 1984).
Back.
(27) The ratio of molecular mass to total
(dynamical) mass. Globally in a typical galaxy, this is expected to be
10%
(Mihalas & Binney 1981).
Back.
(28) On the topic of evolution,
Genzel et al. (1998) argue that for ultraluminous galaxies (see next
section), the accretion rate onto the black hole, which is time
dependent, determines the luminosity of the AGN and thus whether it
can be observed over and above the star-burst luminosity.
Back.
(29) For example, ~ 108 yr in
Circinus (Marconi et al. 1994), although
Davies et al. (1998) determine an age of
107 yr within
the inner ~ 200 pc.
Back.
(30) Although Gozales-Delgado & Heckman
(1999) report that
40% of
Sy2s host a nuclear star-burst.
Back.
(31) The high infra-red luminosities in
Seyfert galaxies may be due to the AGN (e.g. Curran, Aalto & Booth 2000
and references therein).
Back.
(32) In fact Heckman (1987) finds that
while Sy2s are host to enhanced star forming activity, Sy1s are not.
Back.
(33) Heckman et al. (1989)
find that LCO / Lblue (Sy2)
2 LCO /
Lblue (Sy1) and attribute this to a low CO luminosity
rather than high blue luminosity in the Sy1s of the sample. Assuming
that the blue continuum is radiated isotropically (Fernandes & Telrevich
1995 and
references therein), this suggests that the high LCO /
Lblue ratio is due to an excess of CO in Sy2s and that the
star-burst episodes, which occurred as much as 109 years ago, are
even older in Sy1s.
Back.