19 92MNRAS.259. .

209B
Mon. Not. R. Astron. Soc. (1992) 259, 209-217

Angular momentum of a supermassive black hole in a dense star cluster

A. M. Beloborodov,1 A. F. Illarionov,1 R B. Ivanov1 and A. G. Polnarev1,2

lAstrospace Centre, Lebedev Physical Institute, 84-32 Profsoyuznaya St, Moscow 11810, Russia
2
Institute ofAstronomy, Madingley Road, Cambridge CB3 OHA

Accepted 1992 April 22. Received 1991 December 30

ABSTRACT
We consider the evolution of the angular momentum of a supermassive
(Mh ~ 107-109 M0) black hole in the centre of a dense star cluster of mass Mc > Mh.
Our treatment of this problem is based on general relativity. We take into account the
spin-up of the black hole due to disc accretion of the gas component, and the spin-
down due to direct capture of stars from the cluster, and find the equilibrium value of
the Kerr angular momentum parameter. We calculate the rate of gas supply to the disc
due both to tidal disruption of stars by the black hole, and to liberation of gas through
non-elastic star-star collisions far from the black hole. We calculate numerically the
cross-section of tidal disruption, compare it with the capture cross-section and
average both these cross-sections over the cluster. In the range of black hole masses
under consideration and in the case of a dense star cluster (R~ l pc), the interaction
of the black hole with the stars occurs in the regime of the depleted loss cone. In the
limit of a supermassive black hole (Mh>108 M0), when tidal disruptions are
suppressed, we calculate the equilibrium angular momentum analytically. It is shown
that the black holes surrounded by the most luminous accretion discs should rotate
with an angular momentum parameter close to the extremum.

Key words: accretion, accretion discs - black hole physics - relativity - celestial
mechanics, stellar dynamics - galaxies: nuclei - galaxies: star clusters.

per cent if 0=1 (extreme Kerr black hole). There is a

1 INTRODUCTION
tendency for spin-down of the black hole (decreasing a)
According to the hypothesis first put forward by Lynden-Bell because the black hole preferentially swallows stars with
(1969) (see also recent reviews by Rees 1988, 1990), quasi- negative angular momentum (Doroshkevich 1966; Godfrey
stellar objects (QSOs) and active galactic nuclei (AGN) 1970; Young 1976). However, spin-up (increasing a) can
consist of a supermassive black hole (Mh - 107-109 M0) sur- occur by transfer of angular momentum during disc accre-
rounded by a compact (Rc= 1-10 pc) star cluster of mass tion of gas (Young 1977, hereafter Y). If the rate of disc
Mc~ 108-1010 M0. The mass Mh of the black hole, its accretion, Mdisc, is considerably larger than the rate of star
angular momentum Lh, and the corresponding dimension- capture, Mcapt, then the black hole will spin up and approach
less Kerr parameter a = cLJ{GMl){0<a<l) vary in the the extreme Kerr black hole, 0 -► 1. More detailed analysis,
course of accretion of the gas and stars of the cluster (c is the taking into account the capture of photons emitted by the
velocity of light and G is the gravitational constant). The accreting disc, leads to the conclusion that the maximal
efficiency of energy release, the dynamics of stars in the achievable value of the angular momentum parameter is
cluster, the capture and tidal disruption of stars by the black 0Th = 0.998 (Thorne 1974, hereafter Th). In the opposite
hole and also electrodynamic processes in the black hole’s case, Mc^x> Md^c, the capture of stars causes the black hole
vicinity (Thorne, Price & Macdonald 1986; Novikov & to spin down toward the Schwarzschild case, 0 0.
Frolov 1989) crucially depend on the gravitational field of In the case of steady (stationary) accretion of the gas and
the black hole, which is determined only by the values of its stars the dimensionless parameter of rotation tends to an
mass Mh and parameter a. For this reason, we concentrate on equilibrium value, aeq, which characterizes the balance
the evolution of Mh and a. The efficiency of energy release in between the competing effects on the black hole rotation
the course of disc accretion (see, e.g., Bardeen 1970) varies rate. The value of 0eq depends on the parameters of the cluster
from 5.7 per cent if a = 0 (Schwarzschild black hole) up to 42 as a whole, on the set of parameters of the individual stars of

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
210 A. M. Beloborodov et al.
the cluster, and on the mass of the black hole. The present angular momentum, which is of order yc ~ vcRc, where
work is devoted to an analysis of the behaviour of the value
vc « (GMcl2Rcyi2 « 1500 (Mc/109 Mo^^/pc)- ^ km s"1
of aeq in the most interesting range of black hole masses,
Mh~ 107-109 M0, for which Sun-like stars are affected by (i)
the competing processes of tidal disruption and direct
capture. For this range of Mh, tidal disruption of stars occurs is the virial velocity in a cluster of mass Mc and radius Rc.
close to the gravitational radius of the black hole; the Within radius Rh~ GMjvl, the gravitational field of the
problem therefore requires a relativistic treatment. black hole dominates over the gravitational field of the
In Section 2 we introduce a relativistic criterion for tidal cluster itself and stars move almost along parabolic orbits
disruption of stars by a rotating black hole; we compare tidal in the field of the black hole. Parabolic orbits are para-
disruption and direct capture of stars and obtain the range of metrized by azimuthal and polar (floj angles, which
black hole masses in which these two processes compete. determine the direction of star motion at infinity, and also by
In Section 3 we derive the equations for the time evolution the components and jd of angular momentum. Integrals of
of the mass and angular momentum of the black hole due to motion for the point test particle in the Kerr metric, Lz, and
the processes of direct capture of stars and disc accretion of Carter’s integral Q (Carter 1968), can be expressed in the
gas. following form:
In Section 4 we consider the feeding of the loss cone
L^jgsmdn, Q=jj+fecos2Gœ.
(Frank & Rees 1976) which the black hole creates in the
angular momentum space, and we find the minimal black Impact parameters bp bd are related to and je by y^ = f «> be,
hole mass, which occurs for an empty loss cone (Frank 1978; je^Vozbp The orbits corresponding to very small angular
Duncan & Shapiro 1983, hereafter DS; Illarionov & momenta enter the black hole. In the plane of parameters jp j6
Romanova 1986b). Then we find the mass and angular these orbits correspond to a region Sc, the cross-section of
momentum gained by the black hole in the course of capture. Because of axial symmetry, the form of Sc depends on
captures and the rate of gas liberation in the course of tidal the polar angle 0«, only. The cross-section S^Q^) has been
disruption. calculated by Young (1976). Stars that are not captured by
In Section 5 we discuss stellar collisions as an alternative the black hole could, nevertheless, be tidally disrupted by it.
mechanism of gas liberation in the star cluster (Spitzer & The region of the plane of the parameters jp jd correspond-
Saslaw 1966). ing to orbits of tidally disrupted stars we call the cross-
In Section 6 we discuss the results of numerical calcula- section of tidal disruption, S{. For the sake of simplicity we
tions of the equilibrium value of the angular momentum assume that, within Sv total disruption occurs, while outside
parameter ¿zeq. St no stars suffer any tidal disruptions. We call the union
In Section 7 we obtain an analytical solution for «eq in the oss =zSt\JSc the cross-section of losses (see Figs 1 and 2). In
case of an ultramassive black hole when tidal disruptions velocity space, Sloss corresponds to the so-called loss cone
switch off and stellar collisions become the only mechanism (Frank & Rees 1976). To evaluate the cross-section St( 0« ) we
of gas supply. consider tidal accelerations:
In our work we make the following simplifying assump-
d2ra/dr2 = Cfaß, a,ß=l,2, 3.
tions:
Here, x is the proper time of the star, r is the separation between
( 1 ) the masses and radii of individual stars are assumed to an element of the star and its centre of mass, and
be solar;
(2) the mass of the black hole is less than the mass of the
Caß^ RiknmQ(0p(afi{0)eU3)
star cluster;
(3) rotation of the star cluster as a whole is slow enough to is the tidal acceleration tensor (in a comoving system) of a
affect both the dynamics of individual stars and the structure star that moves along a geodesic in the Kerr metric (Pirani
of the star cluster as whole; nevertheless, the specific angular 1956; Marck 1983; Chandrasekhar 1983). Riknm (/, k, n,
momentum of the gas component is high enough for accre- m = 0,1, 2, 3) is the Riemann tensor at the current point of the
tion disc formation (y’^> GMJc)\ orbit and el{a) is the tetrad of the comoving system of ref-
(4) the accreting disc is geometrically thin, i.e., its erence. As a time-like vector e^ we take here the 4-velocity
luminosity is less than the critical Eddington luminosity, and i/' [in Boyer-Lindquist coordinates, U’ is given, e.g., by equa-
(5) external sources of supply of accreting matter, such as tions (33.32) of Misner, Thorne & Wheeler (1973), p. 899].
host galaxy gas, intergalactic gas and matter from neighbour- The problem of tidal disruption of a star has been consid-
ing galaxies, can be neglected. ered numerically, both non-relativistically (Nduka 1971;
Nolthenius & Katz 1982, 1983; Carter & Luminet 1985;
Evans & Kochanek 1989) and relativistically (Mashhoon
2 CAPTURE AND TIDAL DISRUPTION OF
1975; Luminet & Marck 1985; Luminet 1986, hereafter
STARS BY A ROTATING BLACK HOLE
Lm). However, these results are not sufficient to calculate the
A small fraction of stars in the cluster take part in the pro- cross-section of tidal disruption because they represent only
cesses of capture and disruption, namely only those stars a few individual orbits; to calculate the cross-section of inter-
that move along very elongated orbits with periastron radii est we need to take into account orbits with arbitrary para-
rp ~ rg, where rg is the gravitational radius. The value of the meters. We introduce the following rather simple criterion of
specific angular momentum is of the order of rgc for these tidal disruption in the form of an inequality on the value of
orbits; this is considerably smaller than the typical specific. the Cu component of the tidal tensor (see Marck 1983) at

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
Angular momentum of a supermassive black hole 211

Figure 1. Cross-sections oftidal disruption,^, and capture, 5C, in the Figure 2. Capture cross-section and set of tidal disruption cross-
plane of angular momentum components and je for Mh = Mcrit, sections ( curves 1-3) for a = 1 and 0«, = jt/2. The curves 1, 2 and 3
0oo=ji/2 and for two limiting vidues of the angular momentum correspond to Mh/Mcrit = 0.3,1 and 4.
parameter (a = 0 and 1 ). At « = 0 (dotted line), St coincides with Sc. At
a=l, curve 1 is the boundary of St and curve 2 is the boundary of Sc.

(strong inequality in 2) requires a numerical treatment of the

1 l dynamical tidal problem (see, e.g., Lm).
the periastron, where 77 = 0: (iv) Along with the stretch of the star in the radial direc-
Cn=g“{Rl0l0(U(')2 + Rl2n(U2)2 + Rl3n(U3)2+2Rl0l3U0U3} tion, described by the Cn component and the self-gravity
which appears in (2), an important role in star disruption is
played by pressure forces, tidal contraction in the direction
perpendicular to the radial direction, and centrifugal forces,
which result from the star spinning with respect to the
Here, the component of the metric tensor g11 is taken from comoving tetrad. We try to describe roughly all these effects
Boyer & Lindquist (1967), r* and ra* are the radius and the with the help of the introduction of the parameter £. It
mass of the star, and § is a numerical parameter of order should be emphasized that the actual value of £ is important
unity. The inequality (2) means that the tidal force at peri- only for the evaluation of the mass-scale of the black hole
astron exceeds the self gravity of the star. (Hills 1975).
We have selected criterion (2) for the following reasons. The comparison of the cross-sections of tidal disruption
(i) This is the simplest criterion, taking into account the and capture enables us to imagine qualitatively how the
relativistic aspects of the problem, particularly the asym- competition between these processes takes place for differ-
metry of the tidal cross-section in the case of a rotating black ent values of the angular momentum parameter a. In the case
hole (see Fig. 1). ö = 0 both cross-sections are circles with the same centre at
(ii) In the non-relativistic Newtonian limit, criterion (2) is y = 0 (see Fig. 1). The radius of the capture cross-section,
equivalent to the Roche criterion of tidal disruption (Lm). y0 = 3MhG/c, corresponds to the minimal periastron rp = 2rg
(iii) The criterion (2) is applicable in the case of a (see, e.g., Zel’dovich & Novikov 1971). This increases with
depleted loss cone, when the orbital parameters of disrupted the mass of the black hole more rapidly than does the radius
stars correspond to approximate equality in (2), i.e., we deal of the tidal disruption cross-section yt, which varies with
with the situation near the limit of the stationary tidal black hole mass approximately as MU2. In the case a = 0,
problem, whereas tidal disruptions deep inside the loss cone these cross-sections coincide at Mh = Mcrit, where
51/2 l r r2 \3/2
‘The coordinates of periastron, r=rp, 0=0p, are found from (3)
the equations for U’ under the conditions Ul <xdr/dt = 0. For
some orbits the value of Cj j does not reach the maximum at peri-
astron and differs from the maximal value by a few per cent. If an for m*= 1 M0, r*= 1 R0. This estimate was obtained from
orbit is nearly captured then the star spends a long time at radii the criterion (2) with Q^yV-4 (l + rg/2r), evaluated at
close to rp and winds around a barrel-like surface, r~ rp,
|0-jt/2|<;|jt/2-0mJ, where cos 0min= ßl/2/y, j1=j]+jl There- periastron
fore, the maximal value of Cx x might be reached at some value of the r=rv= H c[/7/o + (flñ - 1 )1/2]
angle 0 within the interval 10-jt/2|<jt/2-0min, and this angle
differs from 0p. But if in (2) we use the maximal value of Cn rather
than its value at periastron, the final results will not change and y = y0, where Cn = 5c2/{16rÿ. Note that a naive estimate of
significantly. rp for the case y = y0 with the help of the non-relativistic formula

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
212 A. M. Beloborodov et al.
''p,n yo/2G:Mh = 4rg (instead of the correct relativistic one,
:=
disruption switches off and only the capture process con-
r =
p 7o/c==2A'g) would result in an incorrect estimate of tinues. Capture is suppressed if Mh < M_ ( 1 ) « 0.5 Mcrit, while
Qi,n = c2/(64^)== Cn/20 and the corresponding estimate if Mh> M+(l)~lMcrk no disruption takes place (compare
of Mcrit>n would be Mcrk/j20. Thus a relativistic treatment of with YSW ).
tidal disruption of stars by a supermassive black hole is
essential not only in the case of a rotating black hole ( Y; Young,
3 EVOLUTION OF THE BLACK HOLE IN
Shields & Wheeler 1977, hereafter YSW ) but also in the case of
THE STAR CLUSTER
a non-rotating Schwarzschild black hole.
In the case a = 0 and Mh> Mcrit, only the process of We assume that the considerable amount of gas generated by
capture needs to be considered because the cross-section of tidal disruptions and star-star collisions (see Section 5)
tidal disruption is totally inside the capture cross-section; settles down to the equatorial plane of the cluster and forms
tidal disruptions take place only along orbits plunging into an accreting disc. [The fate of the gas liberated by tidal
the black hole (the process of gas supply due to disruptions is disruption has been discussed by Hills (1978) and Lacy,
entirely halted). Townes & Hollenbach (1982).] The matter in the disc gradu-
If ö > 0 the situation is more complicated. As the angular ally spirals inward carrying energy and angular momentum
momentum parameter a increases from 0, both cross- into the black hole. The rates of increase of the black hole’s
sections shift towards negative angular momenta y^, but the mass, Mh, and angular momentum, Lh, can be written in
shift of the capture cross-section exceeds the shift of the tidal terms of the rate of disc accretion and the rate of direct
disruption cross-section. This tendency is more prominent in capture of stars:
the equatorial plane of the black hole (0«; = jt/2) (see Fig. 1),
while at 0«= 0 both cross-sections are symmetric with
respect to jd = 0. As ö increases further, there is a transition M, = ^capt + '¡X^d¡sc> A^capI + ^msMlisc- (4)
from domination of tidal disruptions to domination of
captures; this occurs gradually with the increase of the mass Here, Mcapt and Lcapt are the rates of mass and angular
Mh and covers a range of black hole masses (in contrast to momentum supply into the black hole due to swallowing
the unique value of Mcrit in the case a = 0). For any values of a stars; these rates are determined by the cross-section, Sc, of
and 0«, there exists some value of Mh for which the cross- direct capture of stars. The rate of disc accretion
sections of capture and tidal disruption overlap (see Fig. 1).
For 0«, = jt /2 the mass Mh has its smallest value; we denote Mdisc = aMt + as,Mcon, (5)
this minimal value by M_(a). For larger masses Mh the
fraction of the capture cross-section that lies outside the tidal is determined by the rate, Mt, of tidal disruption of stars by
cross-section increases until the overlapping cross-sections the black hole, with cross-section St, as well as by the rate,
touch in the rightmost point (see Fig. 1). For 0«, =;n;/2 this Mco„, of the gas release due to non-elastic collisions between
occurs for the largest value of the mass, M+(a). For stars; these rates are determined by the cross-section Sc, of
Mh>M+{a) (in the case a = 0 for Mh> Mcrit), only direct corresponding fractions of gas swallowed by the black hole. The
capture of stars by the black hole takes place. dependences of the specific energy, Ems, and angular
Fig. 3 shows the dependencies of M_(a) and M+(a) on a. momentum, Lms, of the matter at the internal edge of the
Region I in this figure corresponds to the dominance of the accreting disc (at the marginally stable circular orbit in the
process of tidal disruption; in region II competition between Kerr metric; see, e.g., Shapiro & Teukolsky 1983) on para-
these two processes takes place; and in region III tidal meter a can be expressed conveniently in the parametric

Figure 3. Dependence of M_ and M+ on a.

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
Angular momentum of a supermassive black hole 213
form(0<je<l): If À is much smaller than the scalesize of the loss cross-
2 1/2 section then the loss cone is depleted. This means that the
ö = (1 + 2x+3x ) , (6)
distribution function,/, inside the loss cone differs from zero in
1 2 1/2
e = Ejc = ( 1 + 3x)/[3( 1 + 2x + 3x )] , (7) a narrow band of width A close to the boundary of S|OSS. In the
l 2 2 1 2 limiting case, A yloss, the integrals ( 11 ), ( 12) are proportional
A = cLJGM^ = 2(3 l x + ae) = 2(1 + 2x)/3 / . (8)
to the arc lengths along the boundaries of the corresponding
3
Note that, in the limiting case x 0, parameter ö « 1 - 2x (see cross-sections. To simplify integration over L we put E = Ec,
Section?). neglecting the dependence of A on L. Then all these integrals
The rotation of the black hole in the cluster is determined by become proportional, with the same normalization factor
equations (4) and (see, e.g., Th) described by the following A = j f(E)/t¿{E) dE, which does not enter into our final results.
equation for the evolution of the angular momentum para- In the opposite limit of completely full loss cone (A /loss),
meter: the integration in ( 11 ), ( 12 ) is reduced to calculating the areas of
the corresponding cross-sections. The transition from full loss
-^capt -^m cone (A > y,oss) to depleted loss cone (A < /,oss) occurs when the
da/d In Mh = -2a + (9)
GMh (Mcapt + £msMdisc/<: ) mass of the black hole exceeds the value
Further analysis provides values of Mcapt, ¿Capt> and Mn which Mf~ 2 x 106(RC/1 pc)3/4 M0,
are determined by the process of loss-cone feeding (Section 4), obtained from the condition A = /loss ( DS ), in the case of a tidally
and the value of Mcou, which is determined by the process of dominated loss cone in the non-relativistic approximation.
stellar collisions (Section 5). Note that, in the case of a compact cluster with /^ < 100 pc, the
value of Mf/Mcrit<l, which means that for the black hole
4 LOSS-CONE FEEDING masses just in the region under consideration the loss cone is
depleted significantly, in particular for Mh = Mcrit. In the case
MCapt, Lcapt and Mx are found by integrating the time derivative when the radius of the cluster is about 1 pc, the depletion of the
of the distribution function of stars in the cluster,/( 0«, ), cross-section is about 90 per cent (see equation 18 below).
over E, jd, 0«,. In the stationary case, and in the frame of the In the range of large black hole masses (Mh Mf) when the
loss-cone approximation, this derivative is zero everywhere loss cone is empty, the mass flow Mcone does not depend at all on
except in the loss cone where it is approximated by//¿d. Here, the parameters of the black hole; it is determined only by the
the orbital period of a star with energy E is parameters of the star cluster (DS; Illarionov & Romanova
1986b):
id(L) = 2 (10)
J'v 3GMC\112 9A
Mcom = Mt + McapX~
8AC
where rp and rap are the periastron and apastron of the orbit.
Integrating over the capture cross-section we get « 1.4 x lO“2(Mc/lO9M0)1/2(7?c/lpc)“3/2Meyr'1, (13)
where A c ~ In (yc/ crg) depends on neither the black hole’s
^capt Ld2/, Leapt /y'flSin 0 d2/, (ID mass nor its angular momentum.

where L = 2jrm*J£dL/0///dd cos 0. Integrating over the

5 COLLISIONS BETWEEN STARS
tidal cross-section, we get
In the case of a dense star cluster, the dominant mechanism of
Mt = Ld y. 2
(12) gas supply is non-elastic, contact, star-star collisions (Spitzer &
Saslaw 1966; Illarionov & Romanova 1986a, b). The rate of
this process is
To calculate integrals (11-12) we evaluate the function /
within the loss cone, taking into account the dynamics of the McoU^l^relOc^c-
feeding of the loss cone, Sloss. All stars with angular momenta The fraction of liberated gas lost per star during a single
within Sloss are swallowed or disrupted by the black hole in a collision, ^ * 0.02t>c/1000 km s"^ depends on the relative
time-scale td. The elastic (gentle) gravitational scattering of velocity of colliding stars, vrel = y[2vc, the cross-section of
stars in the cluster feeds the loss cone and can be described by a non-elastic collisions, oc = 4nrl, and the number density of
diffusive approximation. The dispersion of angular stars, n = 3Mc/(4jTra*R3). Then
momentum variations over the time of motion along the
trajectory (DS; Spitzer 1987) is Mcoll « 0.1 (Mc/109Mo)3(Rc/l pc)_4M0yr“1. (14)
In the case of a supermassive black hole (Mh > Mcrit), the tidal
Dr2dr/v3Gm*RcAIE/Ec channel of gas supply switches off and collisions will be the
sole mechanism for the liberation of gas. For comparatively
small black hole masses (Mh < Mcrit), the collisional mechan-
where the dispersion, D, of deviations of the tangential ism of gas supply dominates over the tidal mechanism
component of the velocity, gained per unit time, is taken for an (equation 13) when the virial velocity is high, vc > 1000 km
isothermal cluster of energy Ec = GMC/3RC, A = In (0.5 Nc) is s~1 (see equation 1), i.e. when the cluster is compact enough:
the Coulomb logarithm, Nc = MJm* is the number of stars and
/ is a numerical factor of order unity. Rc<2(Mc/109Mo)pc.

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
214 A. M. Beloborodov et al.
In the opposite case, Rc > 2 pc, when the virial velocity is
small, vc < 1000 km s-1, the tidal mechanism dominates but
the maximal rate of gas supply is limited by Mt < 5 x 10 ~3
M0 yr-1, which corresponds to a small tidally powered
luminosity, L < 3 x 1043 erg s~ ^

6 EQUILIBRIUM ROTATION OF THE BLACK

HOLE
Having discussed the physical processes in Sections 4 and 5,
we can now specify the equation (9) governing the evolution
of the black hole’s rotation in the cluster:
+
dfl 2J I ft^coneV^capt - K
0
d ln Mh 1 + e{aMt + ßMcom)/Mcapt 5

k= cLçapl
(15)
GMhMcapt

Here, Mcapt, Mcone, and Lcapt are given by equations Figure 4. Dependence of «eq on Mh in the case ot ß = 0.4 and <5 = 0.
( 11 )-( 13). The values X(a) and e(a) are given by equations (7) Curves 1, 2 and 3 correspond to a = 0, 0.5 and 1. The asymptotic
and (8). The dimensionless parameter ß is the ratio of gas value of a at large Mh does not depend on a and is equal to 0.35
supply into the accreting disc by non-elastic collisions to the (see Fig. 8a).
total mass flow into the loss cone:

o ^coll ^ s' k Ç* °C
(16)
Mpnnp oa 1000 km s
Here, og = 4ji{Gm*lvl)2A is the cross-section of elastic
collisions. Note that in the case of a depleted loss cone, i.e.
for large Mh > Mf, parameter ß does not depend on Mh or on
a.1
If the time-scale of black hole evolution, th =
Mh/(Mdisc + McapX is shorter than the time-scale of evolution
of the cluster, tev = MJ(Mcoll + Mcone), the rotation of the
black hole has time to reach equilibrium, a = acq. In the case
of a compact cluster, this happens if the mass of the black
hole is not too large (Mh < astMc).
The dependence of aeq on Mh is found by putting
da/d InMh = 0 in (15); in this way we obtain the following
non-explicit equation for aeq :

a ¿.(aMt + ßMco„e)/Mcm - k
eq
2[1 + e(aMt + ßMcom)/Mapt] '
Figure 5. Dependence of acq on Mh in the case of a = 0.1 and <5 = 0.
The analytical solution of the equilibrium equation for the Curves 1,2, 3,4 and 5 correspond to /ft = 0,0.4,1, 3 and 100.
case Mh > Mcrit when Mt < Mcapt « Mcone is presented in the
next section. For arbitrary values of parameters, we are
forced to use numerical solutions. Figs 4-7 give results of
numerical calculations of aeq for different values of the para- gas generated by colhsions, is accreted into the black hole
meters a, ß and <5. The last parameter describes the through the disc, carrying inside the black hole the positive
fractional fullness of the loss cone at Mh = Mcrit (equation 3): angular momentum and spinning it up almost to the
extremum of the angular momentum parameter,
ô=A
ö = aTh = 0.998 (Th).
^ri,=i7^r“0-07^c/ipc)i/2. (is) At Mh = M_, the cross-section of capture (for 0«, =ji/2)
touches the tidal one at the left edge point. For larger masses,
For Mh< M_, the capture cross-section is inside the tidal M- < Mh < M+, the capture mechanism of spinning down
one. For an empty loss cone this means strong suppression of switches on sharply because the loss cone is empty. As a
captures as a mechanism of spinning down of the black hole result, the value of parameter aeq decreases with the increase
and domination of tidal disruptions. Disruptive gas, as well as of mass Mh (see Fig. 4). For Mh> Mcriv along with stars with
2
For a full loss cone (Mh < Mf) the value of ß does depend on Mh, since negative angular momentum, stars with positive angular
Mcone (see Y) is proportional to the area of the loss cross-section momentum are also captured by the black hole. As a result of
SXoss~My\ß~M-^\ this competition between the capture and tidal disruption of

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
Angular momentum of a supermassive black hole 215
value acq+ (see equation 17) is determined only by the para-
meter ß, which in turn is determined by the virial velocity
only (see equation 16). In the case of an ultramassive black
hole, therefore, acq+ does not depend on the black hole mass.
From equation (17), in the limit ß-+one obtains fleq= 1.
However, taking into account the capture of photons
radiated by the disc, one can rewrite equation (17) in the
following form:
{X-öX)ß-K{a)
^=2[l+(e+yeW (19)

Here, k{a) is numerically integrated using equation (11) and

well approximated (see Fig. 8b) by k{a) * 0.75tf. The relative
error of the linear approximation is less than 10 per cent
everywhere. Taking into account that the value of k enters
into the final result simply by addition to the main term, 2 a
(see equation 20), one can conclude that the corresponding
error in the final result is less than 2-3 per cent. The values
Figure 6. Dependence of «eq on Mh in the case of a = 1 and ó = 0. ôe and ÓA describe the corrections to the value of aeq due to
Curves 1, 2, 3, 4 and 5 correspond to ß = 0,0.4,1, 3 and 100.
the photons radiated by the disc and captured by the black
hole. We rewrite equation ( 19) in a more convenient form:

2a + k{a)
ß = A - 2ae~(ôX + 2aôe) (20)

and find its solution. Numerical calculations (Th) were done

for the case of pure disc accretion, i.e., ß= °o. This calcula-
tion gave ßeq = 0.9982, which corresponds to xTh«0.1 (see
equation 6). Note that ß = oo implies that the denominator in
(20) is zero. Since

X-2ae = 2j3x2 (21)

we have, at ö » aTh,
<3A + 2aJhôe = A(%h) - 2aThe(aTh) = 2y[3xjh « 0.034. (22)
This combination of <5A and ôe changes slowly with aifa~l
(see Th), which is why we approximate this combination with
a constant given by equation (22). Then, substituting
Figure 7. Dependence of fleq on Mh in the case of a = 0.1 and ß = 0.4. equation (21) into equation (20), we obtain
Curves 1,2 and 3 correspond to ô = 0,0.01 and 0.1.
2a + k(a) 0.8(1-x)
ß(x) = (1 + 2x + 3x2)1/2. (23)
2-l3(x2-Xjh) x2-0.01
stars with negative and positive angular momenta, there is a
minimum in the curve aeq(Mh). This minimum is most promi- Note that equations (6)-(8), combined with (21)-(23),
nent at a = 0 (see equation 15) and gradually disappears with describe entirely in parametric form the equilibrium rotation
the increase of a and/or ß (see Figs 4-6). of an ultramassive black hole. The explicit dependence of acq
In Fig. 7, the dependence of aeq on Mh is plotted for dif- on ß is plotted in Fig. 8. In the limit of large ß (compare with
ferent values of parameter ô corresponding to different the case ß = 0 shown in Fig. 9), one can obtain from
degrees of loss-cone fullness. As ô increases, the minimum equations (23) and (6) the following expansion of aeq:
disappears: its value increases and its location shifts towards
smaller masses Mh. aeq=l-2fl2 + ^/+ (24)
Non-monotonic dependence of aeq on Mh isa. direct con-
sequence of the physics of an empty loss cone, and this does where y = 0.8/^ + Jc|h«0.8/^ + 0.01. In the range
not occur if the loss cone is full (see, for comparison, YSW ). l<£ß <£25 the corrections for the photon capture are
negligible and
7 MASS-INDEPENDENT EQUILIBRIUM tfeq~ 1 - 1-4 x ß~3l2. (25)
ROTATION OF AN ULTRAMASSIVE BLACK
For /3» 25,
HOLE
fl a —
eq“ Th 0.24/9"(26)
In the case of ultramassive black holes (Mh> M+), tidal dis-
ruption switches off(Mt«C Mcapt * Mcone) and the equilibrium i.e., the asymptotic value of aeq is aTh, rather than 1.

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
216 A. M. Beloborodov et al.

Figure 8. (a) The asymptotic value of aeq+ at large masses, Mh, versus ß. (b) Dependence of k on a (see equation 15).

mass of the central black hole. The equilibrium value of the

angular momentum parameter (equation 24) is determined
by the value of the parameter ß. Neither ß nor accretion
luminosity, L = [l - e(a)]astMconC2, depends on the mass of
the black hole.
In the case of very dense clusters, with virial velocity
exceeding approximately 1000 km s~l, the rate of gas supply
due to star-star collisions is so high that it could explain
objects of arbitrarily high luminosity. As for the process of
tidal disruption, it is relevant only for short time-scale
phenomena around supermassive (but not ultramassive)
black holes, and it fails to be adequate for the interpretation
of steady luminosity exceeding 1044 erg s “1.
For numerical illustration, suppose ast~0.1, taking into
account two processes: a significant amount of hot gas
escaping (evaporating) from the cluster, and the rest of the
gas partly fragmenting into stars in the outer part of the
accreting disc and partly being expelled from the cluster in
the course of supernova explosions. To explain an arbitrarily
8 CONCLUSIONS high luminosity (L> 1044 erg s 'j, one could assume the
following parameter values for the cluster: virial velocity
In the case of an ultramassive black hole, the sole mechanism í;«4000(L/1047 erg s"1)1/6 km s"1; £«150(L/1047 erg
of gas supply is star-star collisions in the cluster; the rate is s 1)5/6. The black hole rotation in this case is very close to
determined by equation (14) and does not depend on the the Thome limit (see equations 25 and 26).

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System

19 92MNRAS.259. .209B
Angular momentum of a supermassive black hole 217
Luminet, I. P. & Marek, I.-A., 1985. Mon. Not. R. astr. Soc., 212, 57.
ACKNOWLEDGMENTS Lynden-Bell, D., 1969. Nature, 223,690.
We are grateful to M. Rees for hospitality in the Institute of Marck, I.-A., 1983. Proc. R. Soc. London A, 385,431.
Astronomy of Cambridge University, and for helpful discus- Mashhoon, B., 1975. Astrophys. J., 197, 705.
sions. We also thank I. Novikov for critical remarks, and Misner, C. W., Thorne, K. S. & Wheeler, I. A., 1973. Gravitation, p.
D. Earn and C. Pichón for fruitful conversations. 899, Freeman, New York.
Nduka, A,\91l. Astrophys. J., 170,131.
Nolthenius, R. A. & Katz, I. L, 1982. Astrophys. J., 263, 377.
REFERENCES Nolthenius, R. A. & Katz, 1.1., Astrophys. J., 269,297.
Novikov, I. D. & Frolov, V. P., 1989. Physics of black holes, p. 151,
Bardeen, J. M., 1970. Nature, 226,64. Kluwer Academic Publishers, Dordrecht.
Boyer, R. H. & Lindquist, R. W., 1967. J. Math. Phys., 8, 265. Pirani, F. A. E., 1956. Acta Physica Polonica, 15, 389.
Carter, B., 1968. Phys. Rev., 174, 1559. Rees, M. !., 1988. Nature, 333, 523.
Carter, B. & Luminet, J. P., 1985. Mon. Not. R. astr. Soc., 212,23. Rees, M. !., 1990. Science, 247, 817.
Chandrasekhar, C., 1983. The mathematical theory of black holes, Shapiro,S.L. &Teukolsky,S. A., 1983. Black Holes, White Dwarfs and
p. 291, Oxford University Press, New York. Neutron Stars, p. 362, Wiley-Interscience Publication, New York.
Doroshkevich, A. G., 1966. Soviet Astr., 10, 83. Spitzer, L., 1987. Dynamical Evolution of Globular Clusters, p. 47,
Duncan, M. J. & Shapiro, S. L., 1983. Astrophys. J., 268, 565 (DS). Princeton University Press, Princeton, NI.
Evans, C. R. & Kochanek, C. S., 1989. Astrophys. J. Lett., 346, LI3. Spitzer, L. & Saslaw, W., 1966. Astrophys. 143, 400.
Frank, J., 1978. Mon. Not. R. astr. Soc., 184,87. Thorne,K.S., 1974. Astrophys.]., 191, 507(Th).
Frank, J. & Rees, M. J., 1976. Mon. Not. R. astr. 5oc., 176,633. Thorne, K. S., Price, R. H. & Macdonald, D. A. (eds), 1986. Black
Godfrey, B. B., 1970. Phys. Rev. D, 1,2721. Holes: The Membrane Paradigm, Yale University Press, New
Hills,!. G., 1975. Nature, 254, 295. Haven.
Hills,!. G., 1978. Mon. Not. R. astr. Soc., 182, 517. Young, P. G., 1976. Phys. Rev. D, 12, 3281.
Illarionov, A. F. & Romanova, M. M., 1986a. Soviet Astr., 32,148. Young, P. G., 1977. Astrophys. J., 212, 227 (Y).
Illarionov, A. F. & Romanova, M. M., 1986b. Soviet Astr., 32, 356. Young, P. G., Shields, G. A. & Wheeler, I. C, 1977. Astrophys. J., 212,
Lacy, I. H., Townes, C. H. & Hollenbach, D. !., 1982. Astrophys. J., 367 (YSW).
262,120. Zel’dovich, Ya. B. & Novikov, I. D., 1971. Relativistic Astrophysics,
Luminet, !. P., 1986. In: Gravitation in Astrophysics, p. 215, eds Carter, Vol. 1, p. 104, University of Chicago Press, Chicago.
B. & Hartle, I. B., Plenum Press, New York (Lm).

© Royal Astronomical Society • Provided by the NASA Astrophysics Data System