Quasiclassicalscattering and glory effect in the gravitational field of a black hole
A. B. Gaina
Polytechnic Institute, Kishinev
(Submitted 7 January 1989)
Zh. Eksp. Teor. Fiz. 96,25-3 1 (July 1989)
The differential cross section for backward scattering (6- n) of spinless particles with nonzero
rest mass on the background of a Schwarzschild geometry is calculated by quantum-mechanical
methods. It is shown that a glory exists if the gravitational radius exceeds the particle wavelength.
1. One of the remarkable results from the theory of scattering by black holes is the prediction in Ref. 1 of a glory
effect in the backward scattering of light rays. In the framework of geometrical optics, Ford and Wheeler' showed that
the differential scattering cross section in a Schwarzchild
field grows like ( T - 8) as 8-n. Later, an analogous result was obtained for massive probe particles as well.' Glory
scattering by a black hole (BH) is due to the fact that particles with impact parameters slightly greater than the radius
of stability of a circular orbit can execute one or considerably
more than one (half)-revolutions about the black hole before moving away to infinity.
However, it is known from quantum mechanics that
backward (forward) scattering may not be classical if the
classical angle of deviation tends to n (or 0 ) at certain finite
values of the impact parameter. In this case it is necessary to
take account of interference phenomena associated with the
finiteness of the particle wavelength. Thus, a quantum-mechanical investigation of the problem of the scattering of .
particles by a black hole is of interest. The long-wavelength
case (A> R , ) has been considered in sufficient detail for
waves and particles with various spins, both in the approximation of a weak scattering field and in a more rigorous
approach (see, e.g., Ref. 4 and the citations contained therein). In Ref. 5, the opposite case of short-wavelength scattering of electromagnetic waves by a Schwarzschild BH was
considered. It was noted that when polarization properties
are taken into account the glory effect may be absent-the
strictly backward differential scattering cross section for
photons is equal to zero, this being true for both the shortwavelength and the long-wavelength case. A more exact
analysis of the glory for massless waves was carried out in
Refs. 6 and 7.
In the present paper we consider, on the basis of the
Klein-Gordon equation, the scattering of massive spinless
particles by a nonrotating (Schwarzschild) BH in the shortwavelength case:
-'
With the assumption ( 1), everywhere up to the event horizon R , = 2GM /c2 the particle wavelength is short in comparison with the characteristic length scale of the nonuniformity of the gravitational field, and, therefore, it is
convenient to use the phase-shift theory of scattering and the
quasiclassical approximation. In this connection, we note
Ref. 8, in which it was shown that an exact analysis of glory
scattering on the basis of path integrals in flat space leads to a
result coinciding with the quasiclassical result.
2. Writing the Klein-Gordon equation in the Schwarzschild metric and performing a separation of variables, we
13
Sov. Phys. JETP 69 (I), July 1989
obtain the wave function of a particle with energy o , orbital
angular momentum I, and angular-momementum component m along the z axis ( 6 = 0 ) (see Ref. 4):
where the Y are spherical harmonics. (Here and below, we
use a system of units with c = fi = G = 1.) The radial part
satisfies the equation
dr'
2M -'
,
dr"
w , ( ~ . ) R = u , - d- -r= ( 1 L 7 )
z+
with
Choosing at r-+ co (w > p ) a solution of the form
-
and, near the horizon (r* - ), a solution satisfying the
capture condition R a exp(ior*), it is not difficult to show
that the phase shift 6 , is complex,with
2 sh (2 Im 6,) =
1-lSIlL
= T , csp ( 2 Im b , ) ,
(6)
lSll
where S , = exp(2i8, ), and TI is the coefficient of absorption
of an incident partial wave. Expanding the wave function of
a stationary scattering state in the partial waves (2), we obtain the standard expression for the elastic-scattering amplitude. For angles 8 # 0 it has the form
%
.
Here the imaginary part of the phase shift determines the
total absorption cross section:
-
e+
3. Restricting ourselves next by the condition ( 1), we
note that in this case the function W, ( r ) takes the form
Here, for 1% 1 we have made the replacement
1( l + l )
-+
(l+L/z)2=L2.
(10)
The function U, ( r ) coincides with the effective potential
energy of a classical probe particle in a Schwarzschild field
(see, e.g., Ref. 9). In the classical limit, when
where L. = ( 1 - p ' / ~ ~ ) " the
~ , particles fall into the black
0038-5646/89/070013-04$04.00
@ 1989 American Institute of Physics
13
hole, and a state with L = L C corresponds to an unstable
circular orbit.
For ( L- LC1 4 LC the function Wl( r ) can be represented in the form
w,(r)=wte( r )+ ~ W (Lr )
(12)
3
to r - 3 in the effective potential can be treated as a perturbation'':
where
where
2M
SWL ( r ) = - r'
(13)
Using the approximation ( 12), ( 13) and the WKB method,
one can calculate the transmission coefficient TI for orbital
angular momenta close to the critical value (i.e., near the top
of the potential barrier) :
w:" ( r )= 2MLa
Without dwelling on the details of the calculation, we write
out the result, omitting the logarithmically divergent part:
Re 6 i s - v In(L'ir.'/k2)"' + --k
2
+ arcain where the coefficient 0 depends only on the velocity of the
particle at infinity:
=[
5 + 44u2 + 32u4 - 3 ( 1
8(1 -u2)
+8
1.
~ ~ ) ~ ' ~
(15)
It varies slowly from 0.5 at u = 0 to 1 at v = 1. The same
result is obtained if one uses a parabolic approximation of
the potential near the top of the barrier. It follows from ( 14)
that for L - LC> 1 the transmission coefficient is exponentially small, while for LC - L( 1 it is practically equal to 1.
The quasiclassical total absorption cross section takes the
form
nMZ (1+8u2)"+8v'+20v2-1
o*=-v2 vZ
( 1 ) .
(16)
The factor in front of the bracket is the exact expression for
the classical cross section for capture of probe particles in a
Schwarzschild field, with well known nonrelativistic and ultrarelativistic limits. The additional term in the brackets is of
order (R,w)-' and is a quantum correction. It should be
noted that in the first of Refs. 10 an attempt was made, for
the first time, to calculate the glory for spinless particles of
nonzero mass.
4. The relation ( 14) implies that for L > L Cwe can use
the usual WKB formula to calculate the real part of the scattering phase shifts:
r
Re 6,= lim{
I-00
r
j [ W ,(r) 1%dr' -
( k 2 - L Z )d }
.
'rl
It is easy to convince oneself that in the nonrelativistic limit
( k $ p ) , with the replacementpM- - eQ, the formula (22)
goes over into the expression for the WKB phase shifts in a
Coulomb field. In the relativistic case ( k $ p ) this same formula gives the partial phase shifts, calculated in Ref. 12, for
scalar massless waves in a Schwarzschild metric.
In the second case ( 0 < L - LCg L c ) ,when the impact
parameters of the particles are close to the critical values, we
make use of the representation of the function W ,( r ) in the
form ( 12), ( 13), regarding S W , as a perturbation. Then
where
Here Cis a function that depends only on the velocity of the
particles at infinity:
( 17)
L/k
The imaginary part of the phase shift here is small. The second integral in (17) gives the phases of the free motion,
while the first contains the divergent part typical of longrange potentials. The exact form of this part is
The first integral in ( 17) reduces to an elliptic integral, and
analytical calculations are possible only in the limiting cases
L >LC and 0 < L - LCg L C .However, in these particular
cases it is possible to use a perturbation method from the
outset to estimate the contribution of the small terms in the
potential ( 9 ) .
In the first of these cases ( L$ LC) the term proportional
14
11
[ arcain jL2+,z/kz)s
Sov. Phys. JETP 69 (I),July 1989
It varies slowly in the range from 32 at v = 0 to 15.5 at u =
1 [see ( 15) 1. In Eqs. (23)-(25), as above, we have omitted
logarithmically divergent terms.
5. The subsequent calculation of the scattering amplitude does not differ fundamentally from that in the case of
flat space, and this is also true for the glory. However, before
giving the results we remark that in our case the calculational method is applicable only for the condition LC2 lo2
(or, correspondingly, for wR, 2 lo2). In the opposite case
the phase shift 6;'' can be of the order of 1, and the angle of
deviation changes sharply. In this case it would be necessary
to take account of interference of waves with I-I,. Such
interference phenomena have been analyzed numerically for
scalar waves with A R, in Ref. 13.
-
A. B. Gaina
14
Since the phase shifts (23)-(25)give a nonunique relationship between the angle of deviation and the impact parameter (the angle of deviation tends to infinity as L L C),
introducing the number n of revolutions of the particle we
find that the scattering angle is equal to
-
For L >LC the angle of deviation is a one-to-one function of
the orbital angular momentum, and for relativistic particles
the phase shifts ( 2 2 ) always correspond to small angles of
deviation:
At the same time, for nonrelativistic particles with angular
momenta LC9 L < z / k the angle of deviation can be large
( 1 5; 8 < P ) . For this it is necessary that the condition v < 4
1'
be fulfilled.
Expanding the phase shift 6, in a series about the value
L,, ( 8 ) determined from ( 2 7 ) , we represent the scattering
amplitude ( 7 )in the form of a sum over n and go over from
the summation to integration over the continuous parameter
L:
It can be seen that with increase of n the amplitudes f, ( 8 )
decrease exponentially rapidly. Therefore, the small-angle
scattering is classical and is determined by the amplitude
(33), which has a Rutherford character:
We remark that if one calculates the phase shifts using the
scheme ( 19)-(21), keeping terms -- L f / L , differs from the
Rutherford cross section by a term of order L : / k 28"Ref.
1 1 ) . In the nonrelativistic case the angular distribution of
the scattering will be of the Rutherford form up to rather
large angles.
In the range of intermediate angles ( 8 , a - 8 ) % L ;
the amplitudes f, ( 8 ) corresponding to the phase shifts
(23)-(25) are integrated with the aid of the corresponding
representation of the Legendre polynomials:
'
We stress that in the case n = 0 this formula is valid only for
sufficiently large angles, such that the expression (31) is
small in comparison with 1. The corresponding cross section
is determined by the amplitude f, and coincides, to within a
factor approximately equal to 1.4, with the classical cross
section calculated in Ref. 2:
-0 (L-L")
(,,,
- ( L - L n ) 2 ] } ~ , 0) dL.
2L7,'(8)
In order of magnitude, the maximum number N of revolutions is equal to ( I d , )/2afl ' I 2 . The moment L,(8) is determined from ( 2 7 )and ( 2 2 ) for small angles. But if 8-a, or
n > 1 , then L, ( 8 ) can be determined directly from ( 2 7 ) :
However, for angles close to n, the expressions (36)
and ( 3 7 ) are not valid. Using the representation of the Legendre polynomials in the form ( - 1 ) 'J, [ L ( n- 8 )1, we
find that for angles ?r - 8.4 ( L:, ( n )( - ' I 2 the amplitudes f,
( 8 )have the form
The value of the phase shift S,,,( 8 )is then equal to
First we shall consider scattering through small angles.
Using the asymptotic representation of the Legendre polynomials in terms of Bessel functions J,(LO) and taking into
1, we find
account that, by virtue of ( 2 8 ) ,O L z 2 q ~ %
Whenf, ( 8 ) with n % 1 is integrated over a range of angles
8-4 IL, ( 8 )1 - ' I 2 one can take the Bessel function outside the
integral. We obtain
15
Sov. Phys. JETP 69 ( I ) , July 1989
(38)
With increase of n the amplitudesf, decrease exponentially
fast, and so the principal contribution to the backward scattering will be given by the amplitude f o ( 8 ) . Restoring the
dimensions of the constants c, 4, and G, we write the differential cross section for backward scattering in the form
where f and A are dimensionless functions of the velocity of
the particles at infinity. The functionf ( v ) is defined in ( 1 1 ),
and A ( v ) has the form
4 f 3(6p) 'I)"2exp (-2x3 )
A=f3C!3 esp(-2np") =
(40)
(2-8) [ (:3p) '+ (2p-1) I *
The limiting values of this function are equal to 17 at
v = 0 and 4.06 at u = 1 , in agreement with the result of Ref.
7.
A. B. Gaina
15
Unlike the classical cross section, determined by formula ( 3 7 ) ,the quantum cross section is finite when 6 - a. Nevertheless, the total number of particles scattered backward
(more precisely, in the range of angles O < a - B<aL, where
L;
:g a g 1) is, by formula (39 ), twice as large as in the
classical case. By virtue of well known properties of Bessel
functions, the angular distribution ( 3 9 ) possesses a series of
peaks at a - 6 = 0 , 3.83/LC,7.01/LC,etc., alternating with
minima. The ratios of the intensities at the maxima are
1 :O. 161:0.090:0.062,etc. We note also that the differential
cross section ( 3 9 ) obtained in the case of relativistic particles exceeds the Rutherford cross section in order of magnitude by a factor of R , /A. In the range of angles L ;
< T - 6 g L [I the amplitudes ( 3 6 ) and ( 3 8 ) go over into
each other [as, correspondingly, do the cross sections ( 3 7 )
and ( 3 9 ) ] ,and the angular distribution of the scattering
becomes classical.
Thus, the scattering of spinless particles by a black hole,
like the scattering of massless waves, possesses a glory. In
forward scattering the glory effect is masked by the Rutherford scattering, and so the wave and spin properties of the
particles are manifested only in the the backward scattering.
The orbital angular momentum value Lo corresponding to the first ring of the glory is determined from formula
(27):
-
P-+ 1. In this case, in view of the small value of the exponent
, the forthe expression ( 31 ) becomes large ( ~ 0 . 3 8 )and
mula (41) yields a rough estimate for Lo ( 5.5).Developing an iterative scheme on the basis of the original formula
( 2 7 ) ,we find after a few steps that Lo = 4.65,uM. The glory
cross section for nonrelativistic particles is inversely proportional to the square of the momentum and to the Compton
wavelength of the particles: r o ( 6 - T ) I2z103,uM3/u2.
From this we easily obtain the result that the glory for particles of nonzero rest mass is observable against the back( A c / R ) '/'.
ground oftheRutherford scattering for u>
,
'
[
+
1.
96P3e x p (-2nP1")
(2-p)(P5+[(2p-f)j3]"!)i
kBp
and coincides with the value obtained by Ford and Wheeler'
and DarwinI4 in the ultrarelativistic limit w $,u(P-+1 ). We
shall consider more carefully the nonrelativistic limit w +,u,
16
Sov. Phys. JETP 69 (I), July 1989
View publication stats
'K. W. Ford and J. A. Wheeler, Ann. Phys. (N.Y.) 7, 259 (1959).
'P. A. Collins, R. Delbourgo, and R. M. Williams, J. Phys. A Math. Nucl.
Gen. 6, 161 (1973).
'R. G. Newton, Scattering Theory of Waves and Particles, McGraw-Hill,
New York (1966) [Russ. transl., Mir, Moscow (1969) p. 5291.
4A. B. Gaina and G. A. Chizhov, Vestn. Mosk. Gos. Univ., Fiz. Astron.
24, No. 3, 3 (1983); A. B. Gaina and F. G. Kochorbe, Zh. Eksp. Teor.
Fiz. 92,369 (1987) [Sov. Phys. JETP 65,211 (1987) 1.
jB. Mashhoon, Phys. Rev. D 10, 1059 (1974).
'T.-R. Zhang and C. DeWitt-Morette, Phys. Rev. Lett. 52,2313 ( 1984).
'R. A. Matzner, C. DeWitt-Morette, B. Nelson, and T.-R. Zhang, Phys.
Rev. D 31, 1869 ( 1985).
'C. DeWitt-Morette and B. Nelson, Phys. Rev. D 29, 1663 ( 1984).
9Ya. B. Zel'dovich and I. D. Novikov, Relativistic Astrophysics, Vol. I ,
Stars and Relatiuity, University of Chicago Press, Chicago (1971)
[Russ. original (2nd edition), Nauka, Moscow (1974), p. 1221.
"'A. B. Gaina, Dep. No. 1970-80, All-Union Institute of Scientific and
Technical Information (1980); A. B. Gaina, Astron. Zh. 62, 1018
(1985) [Sov. Astron. 29, 589 ( 1985)].
"A. B. Gaina, Izv. Vyssh. Uchebn. Zaved., Fiz. 31, 33 (1988).
"N. G. SPnchez, J. Math. Phys. 17,688 (1976).
I3N. Sanchez, Phys. Rev. D 18, 1798 (1978).
I4C. Darwin, Proc. R. Soc. London Ser. A 249, 180 ( 1959).
Translated by P. J. Shepherd
A. B. Gaina
16