PHYSICAL REVIE% D VOLUME 14, NUMBER 12 15 DECEMBER 1976

Particle emission rates from a black hole. II. Massless particles from a rotating hole*
Don N. Page
W. K. KeE.ogg Radiation Laboratory, California Institute of Technology, Pasadena, California 91125
(Received 12 May 1976)
The calculations of the first paper of this series (for nonrotating black holes) are extended to the emission
rates of massless or nearly massless particles from a rotating hole and the consequent evolution of the hole.
The power emitted increases as a function of the angular momentum of the hole, for a given mass, by factors
of up to 13.35 for neutrinos, 107.5 for photons, and 26380 for gravitons. Angular momentum is emitted
several times faster than energy, so a rapidly rotating black hole spins down to a nearly nonrotating state
before most of its mass has been given up. The third law of black-hole mechanics is proved for small
perturbations of an uncharged hole, showing that it is impossible to spin up a hole to the extreme Kerr
configuration. If a hole is rotating fast enough, its area and entropy initially increase with time (at an infinite
rate for the extreme Kerr configuration) as heat flows into the hole from particle pairs created in the
ergosphere. As the rotation decreases, the thermal emission becomes dominant, drawing heat out of the hole
and decreasing its area. The lifetime of a black hole of a given mass varies with the initial rotation by a factor
of only 2.0 to 2.7 (depending upon which particle species are emitted). If a nonrotating primordial black hole
with initial mass 5 && 10'" g would have just decayed away within the present age of the universe, a hole
created maximally rotating would have just died if its initial mass were about 7 X 10' g. Primordial black holes
created with larger masses would still exist today, but they would have a maximum rotation rate determined
uniquely by the present mass. If they are small enough today to be emitting many hadrons, they are predicted
to be very nearly nonrotating.

I. INTRODUCTION er energy than the hole mass mill be considered,

so that the adiabatic approximation used in the
Black holes, as Hawking and others have quantum calculations of the emission' ' will be
shown, ' ' emit particles like thermal bodies. Paper valid. The quanta emitted have typical energies of
I' reported numerical calculations of the emission the order of the black-hole temperature or of M '
rates from a nonrotating black hole. This paper [with (10" g) ' =266 MeV in conventional units],
gives the rates for the known particles of zero or which we want much less than M, so we need
negligible rest mass from a. rotating (Kerr) black ' g.
hole and shows how such a hole would evolve as it
M»1 (Planck mass) =2. 18 && 10
emitted these particles. These results are of in- Then roughly M' quanta are needed to carry away
terest in testing the validity of the simplifying as- the energy of the hole; i. e. , the entropy in the ra-
sumption that most black holes which emit signifi- diation, which is roughly the number of quanta
cantly today are not rotating (see, for example, when thermally distributed, is of the same order
Refs. 7 —9). as the initial entropy of the hole, which is one-
fourth the area" or roughly M'.
Paper I noted that although a small black hole
mill quickly give up its electric charge, ""
it is When a black hole is charged and/or rotating so
much less certain whether the rotation will also that Q„and/or a„are significantly different from
become small. The main difference in the time zero, and when it has temperature or electrostatic
scales of the two processes can be seen in the fol- potential high enough to permit emission of elec-
lowing way (using henceforth the dimensionless trons or positrons, it tends to emit most of its
Planck units spelled out in paper I): quanta with the same sign of the charge and/or
The parameters that determine the shape of a angular momentum as the hole. A charged particle
black hole are carries off charge
a„= J/M' and Q —
= Q/M, I
+ql =e =0.0854, (4)
where J' is the angular momentum, Q is the which is roughly of order unity, and a typical quan-
charge, and M is the mass (which sets the scale of tum also carries off an angular momentum
the size). These quantities have a domain limited
(5)
by the constraint
of order unity. Since a~ and Q must have absolute
values not greater than unity, the number of
Only black holes which emit quanta of much small- charged particles needed to neutralize the hole is

14 3260,
14 PARTICLE EMISSION RATES FROM A BLACK HOLE ~ II. . . 3261

Q/e, which is only of order M, whereas the num- masses of primordial black holes (PBH's) that
ber of particles needed to carry off the angular would be just disappearing today, and the maximum
momentum J is of order M', Thus the charge can rotation parameters that PBH's of various masses
be emitted fairly quickly, but the loss of angular today could have. The remainder of the paper will
momentum requires roughly the same number of derive the mathematical formulas for the quantities
particles as the loss of mass. Therefore, in this desired, describe the numerical methods used to
paper we will assume that the charge neutraliza- calculate them, give the results, and discuss their
tion has already occurred but that the angular mo- properties.
mentum may still be significant.
Though one expected a black hole to give up its II. MATHEMATICAL FORMULAS
angular momentum in the same order of time as it
gives up its mass, it has not been known whether Since the total number of particles emitted during
a tends to zero as the black hole evolves. Car- the black-hole evolution, roughly M', is assumed
ter argued that it would tend asymptotically to- to be very large, the emission may be approxi-
ward a fixed value less than unity, but he gave no mated as a continuous process with negligible fluc-
indication of what that value would be. Numerical tuations due to particle discreteness. Then the
calculations were needed to show whether ln al- J rates are mell-determined functions of M and a~
ways decreases faster than lnM', pushing a to- alone (assuming that Q~ = 0, which was justified
ward zero, or whether these two quantities de- above). The rest masses of the particles emitted
crease equally fast at some nonzero limiting value are assumed to be negligible, and the particle
for a~. There is some indirect evidence, to be species emitted are assumed to be fixed (indepen
given below, that if there were a large enough num- dent of M), so the only scale in the problem (other
ber of massless scalar fields (unknown at present than the Planck units, which are here defined to
and therefore not calculated in this paper) to domi- be unity) is determined by M. All quantities to be
nate the emission, a~ might indeed get hung up at calculated scale as some power of M and can
some nonzero value. However, this paper shows therefore be put into a scale-invariant form (e. g. ,
that emission of the known massless fields can only depending only on a~) by dividing out this power
decrease a~ toward zero, and that in fact the de- of M — or, when one calculates the evolution of a
crease is rather rapid compared with the mass hole [Eqs. (11) ff. below], by dividing out the value
decrease. of M at some particular point on the evolutionary
Because black holes that died in recent epochs track.
or that are emitting significantly today spend al- First, let us consider the rates at which the
most all their lives with temperatures of order mass and angular momentum of a black hole de-
20 MeV, which is well above the mass of the elec- crease, which are given in paper I by Eq. (I.12).
tron but well below that of each known heavier par- Since the time f, scales as M', we may define the
ticle, it is reasonable to do the calculations for scale-invariant quantities
the idealized case of emission of a fixed set of
species with negligible rest mass. For example, f—
= M'd lnM/dt = —M'dM/dt, (6)
the "canonical combination" used below is the set g—
= M'din J/dt = —Ma~ 'd J/dt.
of known species with masses less than 20 MeV:
These can be seen to be functions of a~ alone: If
gravitons, photons, electron and muon neutrinos
we define the scale-invariant energy of an emitted
with one helicity each, electrons, and the corre
particle as
sponding antileptons. However, the results will
also be given for other sets of species, to include x Mco,
=—
(s)
some of the possibilities (to be discussed below)
then Eq. (I. 12) gives
of other near-massless particles in nature or of
the emission from black holes too cold to emit
(g)
electrons and positrons.
The quantities to be calculated in this paper are
the rates at which energy and angular momentum where the expected number of particles of the jth
are radiated, the evolution of the mass, rotation species of spin s emitted in the mode or state with
parameter, and area of the hole, the lifetimes of energy M 'x, spheroidal harmonic l, axial angular
holes with different initial angular momenta, the momentum m, and polarization P is

rt, p(a~, x)
&x™ exp (477[1 + (1 a ji (10)
. ~) i~]x 27(a (1 a ) ~) ( 1) ~
3262 DON N. PAGE

Here Eq. (I.4) has been used, with the values of the T(0) =0. (18)
surface gravity, angular frequency, and electro-
Then Eq. (6) combined with Eq. (16) gives
static potentia, l of the hole obtained from Eqs. (I. 8),
(I.9), and (I.lo). I'» & is the absorption probability e e
' (19)
for an incoming wave of the mode considered and dy fh g —2f
equation. ' '
can be found by numerically solving the Teukolsky
It can easily be seen to depend only
on a~ and x in addition to the subscripts. The de-
From the solutions g (y) and v(y) of the coupled
differential equations (16) and (19), one can get
y(w) and e (v), and hence a„and M/M„as a func
j
pendence on the species and polarization p is only
tion of time. From these, one can find how other
through the spin s (assumed positive) and the num-
ber of polarizations P that the species has; then l quantities evolve, such as the area
and I can take on any values such that l —s and
I —~'m are non-negative integers.
A =8mM'[1+(1 a ')'~2]
~

Once one has the evolution of a black hole from

Next, let us consider the evolution of the black
hole. Equations (6) and (7) give the rates of change
a =1, one can consider holes with other initial
J
of M and with respect to time once and g have f values a~, of the rotation parameter. They will
follow the same solution e(y) and v(y) but with dif-
been calculated. However, since f
and g are
ferent initial values:
functions of a~, it is easier to solve the equations
if a is considered as the independent variable. y. —= —lna gz y (21)
Furthermore, dividing Eq. (7) by Eq. (6) shows us
that e,. =e(y,.) = —ln(M, . /M, ), (22)
-
din J
-=~(y, )=M, '.f, . (23)
dlna~ g
dlnM dlnM f These equations determine M, and t,. such that the
hole would have mass M,. and rotation a~,. Rt time
which approaches a constant value as a approaches
zero (assuming that the value is positive so indeed t, if it had started with M =M, and a~ =1 at time
a -0as M-0). Because of the logarithms in Eq. t =0. In terms of M,. and a~„ the evolution follows
(11), it is convenient to define the independent va. r M =M~e 8=M,-e'&", (24)
iable to be
t, =M, '(v . w,. ) =M,
'e" ~(7' —.w,.) . (25)

To cover the greatest range of possibilities, the Equation (25) and the "standard evolution law"
evolution will first be calculated from a =1 or y e(y) and r(y) can be inverted to get w and hence y
=0 to a =0 or y =~; a black hole starting at a dif- and a as functions of time, and then Eq. (24) gives
ferent value of a~ will simply follow the evolution- the mass.
ary track from that point onward. A particular quantity desired is the lifetime
Now the object is to find how the mass and time a,
T(M, , .) of a black hole with initial mass M,. and
vary with y. Let the starting mass at a = 1 be rotation parameter a„, It can be seen from Eq.
(25), assuming that the black hole does evolve to
M, -=M(y =o); Q 0 Dr y ~RS M Os thRt this ls
this will be the mass that sets the scale. With an
eye back on Eq. (11), set T(M, , a~,.) =t(y = ~) —t, =M, 'e"~(v& —v,.). ,. (26)
'

ln(M/M, ), (14) where

which has the initial value = y(y —ao) (27)
e(0) =o (15) is the lifetime in units of M, ' of a hole that started
and evolves according to the reciprocal of Eq. (11) with a =1. The mass dependence of the lifetime
RS can be divided out to get the scale-invariant quanti-
ty
f
de/dy = I/h = /( g —2 ) f . (16)
It has been noted that the time scales as the mass 0,. = M, 3T(M, , a,.) =e "—(v&. —r,.),
cubed, so define the scale-invariant time parame- thus written in terms of quantities previously cal-
ter as culated. Once the lifetime of any black hole is
v=—M, t known, one can calculate the initial mass of a pri-
mordial black hole that has just disappeared within
with initial value the present age t~ of the universe:
14 PARTICLE EMISSION RATES FROM A BLACK HOLE. II. . .
M,. (a„,, f,) =f,'/3e. -'/' with spin 2, 1, and 2, respectively, assuming
1/3e'g(~ g)-&/3 that there are no massless particles of other spins.
0 f i (29)
A total of 463 angular modes (a combination of
Since PBH's would have been spinning down since s, I, m, and a*) were calculated and integrated
their creation at time t, ago, their present values over frequency: 170 modes for s = 2, 155 for s =1,
of a should have anupper limita (M, t, ) less and 138 for s =2. For example, at low a~ all the
than unity, depending upon the present mass M. It modes up through $ = 2 for s = ~ and through l = 3 for
is simpler to solve for the inverse function s =1 and s =2 were calculated. At high a~ the l =-m
M „(a*,to), the minimum mass of a PBH with a modes were calculated up to l = '-,' for s = ~, / = 11
today. By combining Eqs. (24) and (25) with t —t, for s =1, and 7 =9 for s =2, and several l =~@+1
=to, one finds that modes were calculated (with considerably smaller
M=f 0 "'{' ')-"'e-' results), but no modes with I —m&l. At interme-
diate values of a, some combination between
where 7 are evaluated at the present value of
and z these two extremes was taken. The modes calcu-
y or a . Clearly, the minimum occurs at the lated appeared to include nearly a11 of the radia-
smallest value of v, , which is zero if PBH's can tion, though estimates for the small contributions
be created with a~,. up to unity, so in that case of all the other modes were a.dded in, assuming that
M „(a*,t)=t, ' '[1(-lila )]' *', (3l) 'e"' the sum over n~ dropped off exponentially in /
roughly as the calculated modes did.
where —lna~ is shown explicitly as the argument of Once the functions f,and g, were found at 14
r(y) andz(y). If a„,. has a smaller maximum val values of a~, an interpolation algorithm was needed
ue, the corresponding minimum for 7',. is to be to evaluate them at other values of a~ or y. These
used in Eq. (30) to give M „(a„,to). One can see functions varied by factors of up to 25 000 from
that for fixed 7, , M in Eq. (30) is a, monotonically a~ =0. 01 to a~ =0.999 99, and the variation with a~
increasing fllllctlon of Q*q assuiillllg tllatg —2f ls was particularly rapid at the upper end. To find
always positive so that 7 is a decreasing function smooth relationships, various functions of the f's
of a*by Eq. (19). Thentheinversea~ ~(M, t, ) and g's were plotted against various functions of
is uniquely defined and is a monotonically increas- a~. Of the combinations tried, a small fractional
ing function of M. f
power of the 's and g's versus the surface gravity

III. NUMERICAL METHODS "

g of the hole was the most linear. Therefore,
cubic spline fits, minimizing the sum of the
The major part of the numerical calculations
consisted of computing the functions (a„) and f the 14 values of a*, were made of
versus
"
squares of the third derivative discontinuities at
f, andg, o'
g(a ) by Eqs. (9) and (10), which was done at 14
values of a~ from 0.01 to 0. 99999 to an accuracy 4M~=2[1+(1 a«') '/'] ', (34)
of one part in roughly 104 or better at low a and
10' at high a„. The basic method is briefly sum- which varies from 0 at a~ =1 or y =0 to 1 at a =0
or y = ~. The fits of these variables indeed were
marized in Sec. III of paper I. In order to cover
quite smooth, with the slopes never changing by a
different possibilities for the set of particle spe-
factor of more than 3. 6 (even though the values of
f
cies, the contributions to and g from each spe-
the fractional powers themselves changed by fac-
cies were calculated separately. Thus /, and f, tors exceeding 50) and with only four of the 84 val-
g, /„ f, f
and g„and , and g, .were calculated as
ues of the second derivatives of the splines at the
the contributions from one species with two polari-
knots exceeding unity in magnitude.
zations of spin &, 1, and 2, respectively:
The functions f,
and g, were evaluated at 363 val
ues of a~ from 1 down to 0.0005 by the cubic spline
.
x&m ag -, ~
interpolation algorithm and then were combined by
f
Eq. (33) for some combination of n's to get and g
Here the dependence on the species is only through at each point. A fourth-order Runge-Kutta method
its spin s, and the sum over the two polarizations was used to integrate Eqs. (16) and (19) simultane-
has already been taken, since the expected num- ously over the corresponding range of y with the
ber emitted in a mode labeled by x, l, and m is initial values set by Eqs. (15) and (18). At every
independent of the polarization. Then other point (since the integration requires two
~~' +n
points per step), the values of M/M„e™, . (a*„f,),
=n ~
+n (33) M „(a*,t,), andA/A, (whereA, =8vM, 'was the
area at a~=1) were calculated by Eqs. (24), (28),
where ny/2 ny and n, are the number of species (29), (31), and (20). As a check on the a.ccuracy
 DON N. PAGE

of the numerical integration, the step size was M, (a, t,) - (snt, )'~'(sn7~e"a '" —1) "~'
halved, which resulted in agreement to four or five - (t /g )'~'e-'a " =M, (a, = 1, t, )e 'a ~",
decimal places.
For a„smaller than 0.0005, the values of and f (45)
W/X, - 2M'/M, '- 2[so
g a~=. 0 or y=-~ were used:
at
(~, —~)]'".
a =' f (a ~ = 0), (35)
IV. RESULTS
---
g(a = 0) . (s6)
f„
P
The values of the scale-invariant power in a
Then Eqs. (16) and (19) become two-helicity particle species of spin s, and of g„
de/dy ~ &/(6 »-) =-x, (37)
the scale-invariant torque per angular momentum
of the hole, are listed at the 14 values of a~ in
d~/dy ~ e "/(6 —o'), Table I, along with the extrapolated values for a~
= 1. The cubic spline interpolations are graphed in

so the solution is
Figs. 1 and 2, which show that below roughly a~
=0. 6 the neutrino (s =-,-) power dominates, followed
by photons (s =1) and finally gravitons (s =2).
However, at greater values of a~ the order is re-
versed, with gravitons dominating the emission
and photons and neutrinos coming second and third,
respectively.
——y dy (41)
This behavior can be explained qualitatively in
the following way: For a slowly rotating hole, the
is a constant that was simply estimated as z —yy coupling depends most strongly on the spheroidal
harmonic index t (which reduces to the total, not
at a~ =0.0005. The solution for large y or small
the orbital, angular momentum when a~ = 0) rather
a~ gives the asymptotic forms
than on the axial angular momentum rn or the spin
PI/M e-~v-6 e-6~ r- [sa.(7 ~)]x/s s. The coupling is greater at lower l values (e.g. ,
(42) paper I showed that the emission rate at low fre-
quencies goes as uP'"), but l~ s, so the emission
is greater at lower values of s, which allow lower
values of /. On the other hand, a rapidly rotating
[cf. Eq. (I.26), where M, =M,. and ~ =T(M, , a,. =0)],
M, (a„,—0, t, ) -. (s~t, )'~', (44) mentum and also with the spin, "
hole couples strongly with the axial angular mo-
so the s =l =m

TABLE I. Power and torque emitted by a black hole. For each spin s, (~~) and g, (a~) are f,
the contributions of one species with two polarizations to f=-j/13dlnM/dt and g=-M3dln&/dt at
that value of the rotation parameter ~~. The first 14 rows were calculated by Eq. (32); the
l. ast row came from a cubic spline extrapolation. The values for &~=0 are nearly the same as
for &+= 0.01; see Table II for more precise values.

+p = ~/~' fPgP(&~} f)(& ~) f2(&~) Z) f2(&~) g((a ) g2(&, )

0.01000 8.185 xlQ 3.366 xlp 3.845 xlp 6

6 161 xlp" 4.795 xlp 1 064 xlp
0.10000 8.343 x10 ~ 3.~80x10 4. 684 xlp 6 6.174xlp 4 4.895 xlp 4 1 167xlp 4
0 20000 8 830 xlp-~ 4 265 xlp-5 7.732 xlO-6 6 218 xlp-4 5.207 xlp- 1 514 xlP-4
0 300 00 9 669 x10 &
5 525 xlp 5 1.494 xlp 5 6.299 xlp 4 5.759 xlo 4 2 233 xlp 4
0.400 00 1.089 xlp 7.570 xlo 3.116 xlo" ~ 6.430 xlp 6. 599 x 10"4 3.603
0.50000 1.258 xlp 4 1.080 xlp 4 6.822 x10 "' 6.631 xlp 7.845 xlp 4 6.236 xlp 4
0.60000 1.487 x10 4 1..594x10 4 1.574xlp 4 6.946 x10 4 9.668 xlp 4 1.155xlp 3
0.700 00 1.804 x 1Q 4
2.450 x 10" 3.909 xlp 7.457 x lp 1.245 x10 2.322 x10
0.800 00 2.284 x10 4.014 x 10 1.104 xlp 8.366 x 10 1 706 xlO-3 5 286 xlp-3
0 90000 3.j95xlp 4
7.520 xlo " 4.107 xlp 3 1.034xlp 2 632 xlQ 3 1.544xlp
0.960 00 4. 567 xlo 4
1.313 X10 1.305 xlp 1.343 xlp 3.976 X10 3 4.057 xlp 2
0 99000 6 708 xlp-4 2 151 xlp 3 3.578 xlp"2 1.810 xlo 3
5.829 xlp 3 9.555xlp
Q. 999 QO 9.253 xlQ 3.057 xlp 7.251 x]Q 2 340 xlp 7, 723 xlp 3 1.753 xlQ
0. 99999 1.O74 x 10-3 3.555 x 10-3 9.785 xlO-2 2 641 x10-3 8 730 xlo-3 2 271 xlo-i
1.QOQOO 1,093 xlp 3 3.616 xlp 1.012 xlp ~
2, 678 xlp 3 8 851 xlp 3 2.338 xlp
 PARTICLE MISSION
EM I ATES FROOM A BLA CK
RA HOLE. 3265

rate. ' ' black holes sma o evapor t

Figures
u 1 and 2 a e power an e are hot eno
1 e ectrons and
ies emitted w'th
h t d g vitons are t
han20M e V as negli gible.
i

&nations
g n. There is a greater
MeV, ass l'isted
'
un y about n er of 2--helicity
s c1es. Th consistent y xn Fig. 9.
raiment is 1/2 r
Figure 3 graphs the lif
th l ft-h l. e 0 ] of Eq.
'c e, and the a ion ar ameter a „
ow ever
on neutrin
y l o
for var'rlous corn
n
so
(n,
gs units is
n, .
(10" '=5 g
e

have th t
a just evaporates today
to other , thus mak
aking
0—
1. I I I I

e "in which
itional neutri ino states. Furthermore

IO5 I
.8—
I I I I

(Io, I, I)
(8, I, I)
(6, I, I)
Io
C3
-(4, II) F

Phot
(2, I, I)

Neutrinos pn ( I, OO) Neut

eutrinos only ernItted
(I, O, O
(0, 0) I Phootons only e~'t e
E (2,
10
( 0,0, I) yj pns only emitted
Gravitp
Ever
ergthing (g
(6,
(B,I, !
(IO, I,
0 i 1 1 } I I i

0 .2 .6 .8 I0
a„=J /M
Initial mass of or ia
I I I I I I
at just goes
IO
0 .2 I.O e combiination
a„(- J/M~ ' '
st mass. The
emis g hose with

.
FIG, 3. Lifetimee o
d
~

+hm itt d"

ole, scaled b y
ole

s e canonical
e i
a ove the (4 1
ou d give a c
ng

wi masses
particles with true curve w ldb 1- htl boove one of th e higher
eV.
3266 DON N. PAGE

I.O I.O

CU

0
II

4 4

.2

0
0 .2 .4 .6 I.O 0 .6 .8 I.O

t/lifetime t/ lifetime

FIG. 5. Time evolution of the mass of a black hole FIG. 6. Time evolution of the rotation parameter ag
which started out maximally rotating. The vertical and of a black hole that started with a~;=1. For any given
horizontal axes have been scaled by the initial Inass M; curve representing the emission of an assumed com-
=M, and lifetime M; 38; =M v'&. For a black hole that
&
bination of species, the evolution from a~; &1 can be
starts with a . & 1, one can use one of the same curves gotten by moving the left vertical axis to the right until
but shrink the axes so that the upper left corner of the it intersects the curve at a~=a~;, meanwhile shrinking
graph is on the curve at a later point (to be determined the horizontal axis appropriately to leave its right end
from the value of t /lifetime at a+=a+; in Fig. 6) and fixed.
the lower right corner stays fixed, at the end point of
the curve.
"
"neutrinos only, since those graphs have the rates
scaled out and depend only on the ratios of 's and f
[Eg. (29)], assuming that the present age of the of g's at different values of a~.
universe is The time evolution of the mass and rotation pa-
rameter are shown in Figs. 5 and 6. The curves
g, = 16 x 10' yr = 9. 37 x 10'o (48)
for neutrinos, photons, or gravitons only cover the
so that purely hypothetical cases in which the black hole
' '=2. 11 x10 =4. 5S x10" g. emits only particles of one spin; they are included
t
to illustrate the different behavior that would re-
For example, a PBH emitting the canonical combi- sult. For example, gravitons cause the mass and
nation of all known species (except for the small particularly a~ to decrease more rapidly at large
amount of muons and heavier particles emitted) a, as compared with the behavior at small a,
would have just given up all its mass by now if its than photons or neutrinos do. Only one combina-
initial mass had been 4. 73 x 10'~ g if nonrotating or tion with species of all three spins being emitted
6. 26 x 10'~ g if initially maximally rotating. The is included, (n, &„n„n,) =(4, 1, 1), since other com-
curves marked "neutrinos only emitted" in Figs. binations gave curves only slightly different. One
3 and 4, as in Figs. 1 and 2, give the results if can see that for this canonical combination, a
only one species of neutrinos is emitted; for suc- black hole which started at a~ = 1 will lose half its
cessive graphs it does not matter how many neu- initial mass in 71~/g of its lifetime but half its initial
trino species there are for the curves labeled a~ in only 21% of its lifetime. (Half the angular
 PARTICLE EMISSION RATES FROM A BLACK HOLE. II. . . 3267

l.o l. 2

.8—

.8

3
II

C3

4
Neutrinos on
Photons only 4
Grovitons onl
Everything

0 .t I I I I I

l.0 0 4 .6 I.Q

M/M ~
a„= J/M'
FIG. 7. Variation of the rotation parameter with the FIG. 8. Evolution of the area A of a black hole, scaled
mass during the evolution of a black hole, which pro- by the initialarea A; in the case a+,. —-1. For general
ceeds from the upper right to the lower left corner. a~;, the evolution starts at that value of ag with the
The evolution from g+, & 1 can be gotten by keeping the vertical axis rescaled to give A/A; =1 there, and pro-
left end of the horizontal axis fixed and shrinking the ceeds to the left along the appropriate curve as a~ de-
scale so that M/M;=1 falls at u~=a&; on the curve con- creases with time. The evolution of A. is plotted versus
sidered. a~ rather than time to spread out the very rapid changes
near a~ =1, where A actually increases with time. Since
the area is four times the entropy of the hole, these
momentum X=M'a~ is lost in only 6. 7% of the life- curves can also be viewed as giving the evolution of
time. ) the entropy.
Figure 7 shows how a~ varies with the mass as
the black hole gives up its angular momentum and value given in paper I. This result gives a fairly
energy. The emission of gravitons causes a~ to strong justification for the usual simplifying as-
decrease at the fastest rate compared with M, es- sumption, mentioned in the Introduction, that emit-
sentially because gravitons have the greatest spin ting black holes are not rotating. ' '
and thus carry off the most angular momentum per One might note that this result was not apparent
quantum. For the canonical combination of spe a Priori, since h(a~) in Eq. (11) might have gone to
cies, Fig. 6 showed that a~ is reduced to 0. 19 after zero at a nonzero value of a~, in which case the
half of the lifetime from a~ =1, but since it takes curves in Fig. 7 would have leveled out at that val-
71% of the lifetime to reduce M to half its original ue of a as M decreased. In fact, although the cal-
value, a~ is further reduced to 0. 06 by then, as culations have not been made for hypothetical mass-
Fig. 7 illustrates directly. A check of the values less spin-0 particles, there are two reasons for
f
represented by Fig. 1 reveals that is then only suspecting that h might indeed go to zero some-
1% greater than its value at a„=0. Therefore, a where if the emission were predominantly in scalar
black hole decaying by the emission of gravitons, radlatlon;
photons, the presently known neutrinos, and ultra- (1) If one defines k, (a~) by Eg. (11) with f
and g
relativistic electrons and positrons will emit more repla. ced by f,
and g„one has the logarithmic
than 50% of its energy when it is so slowly rotat- slope of a~ vs M in the curves for only one spin
ing that its power is within 1% of the Schwarzschild emitted in Fig. 7. These curves thus have a~ go-
3268 DON N. PAGE

alnM &nM J~
.8
= —
a
——2a~
1 m
& 0. (51)

Therefore, if ho(a ) is continuous and is negative

at a =0, it must become zero at some intermedi-
.6 atea .
(2) The dominant angular mode at small a is
pr esumably the l = s mode, as it is for s = 2, 1,
and 2. For s =0 that mode carries off energy but
no angular momentum, so unless higher angular
modes contribute significantly, one would expect
g, (a =0) to be roughly zero and hence ho(a„=0)
to be roughly .-2. The higher angular modes would
raise h, (a =0) above -2 [conceivably to the value
-1.1948 predicted by Eq. (50)] but would probably
.2 leave it negative, so again one deduces that ho(a~)
may be zero for some a between zero and one.
If either (1) or (2) is valid and if scalar radiation
dominates sufficiently at low a~ for the total radia-
tion to give h(a =0) &0, then the black hole will
0 '

spin down only to the nonzero value of a~ at which

10 g (0I 5 (
OI6
h(a~) =0. This does not occur for emission of the
canonical combination of species, which causes the
FIG. 9. Maximum present rotation parameter ag of hole to spin down rapidly toward a~=0, as shown
a primordial black hole with mass M today, assuming in the curve marked "everything" in Fig. 7. Once
it was created 16 billion years ago with unity as the up- a~ is reduced to a small value, it decreases as a
per limit on the rotation parameter then. Under these
assumptions, the actual maximum is probably near
power law of I,
with the exponent being

(particularly for M & 10'5 g) or somewhat below tparti- h(a =0) = 6. 3611 (52)
cularly for M & 10'5 g) the bottom curve given, depend-
ing upon the additional emitted species not covered in in the canonical case.
the canonical combination. Another interesting result is the evolution of the
black-hole area A, which is illustrated in Fig. 8.
ing as some power of M for small a~, where the The area first increases with time at large a and
power is h, (a~ =0). The numerical calculations in- then decreases to zero along with a~ and the mass.
dicate that there is a remarkably linear relation- This can be seen formally by using Eq. (11) to dif-
ship between h, (a„= 0) and the spin s for s = — '„1, ferentiate Eq. (20):
and 2'
dlnA g (1 2)~)2 (53)
h, (a „=0) = 13.4464s —1. 1948 (50) dlna„g 2f
is accurate to one part in 10~ for all three values, One may further use Eq. (6) to express the time
roughly the accuracy of the numerical calculations. derivative as
Although there is no apparent theoretical reason
to suspect such a highly linear relationship, which =AM '[(1-a~') '~'(g —2f) -gj. (54)
comes only after one evaluates integrals over fre-
quency and sums over angular modes in Eq. (32) For small a„, the right-hand side of Eq. (54) be-
and therefore seems to be accidental, it is tempt- comes -2AM 'f.
This means that the area de-
ing to extrapolate it to s =0 to get a negative value creases logarithmically at twice the rate the mass
for h, (a„=0). One can easily see that the emission does from Eq. (6), which is obvious since at small
of any species makes h(a~=1) &0, since Eq. (47) a~ the area is simply proportional to M'. At large
says that the emission from a maximally rotating a~, it was shown above that h &0, and hence g —2 f
hole is entirely in the superradiant regime where &0 since &0. But (1-a~') ' ' diverges as a~-l,
f
each quantum contributes so dA/dt becomes positive and even goes infinite
 PARTICLE EMISSION RATES FROM A BLACK HOLE. II. . . 3269

as a -1 (cf. the vertical behavior of the curves at hole.

the right edge of Fig. 8). The area is at a maxi- The fact that g —2f &0 at a =1 allows one to
mum where prove the third law of black-hole mechanics" for
i small perturbations of an uncharged black hole.
2f =[1 —(1 —a ) ]g. (55)
(Similar reasoning can presumably be made also
For the canonical combination of species, this oc- for an electrically charged bole. ) The third law
curs at a~ =0. 8868, where tbe area is 17.3% great- states that it is impossible to reduce the surface
er than the original value, after a time of only gravity K to zero by a finite sequence of operations„
6. 729M, ' or 1.7% of the total lifetime 394. 5M, ' of a Using Egs. (6) and (7) to differentiate the expres-
hole with a,- =1. sion for a in Eq. (L8) [cf. Eq. (34)], one finds that
Physically, the behavior of the area can be un- the emission of particles makes
derstood by thermodynamic arguments, since the
[1 —(1-a~')'~'](g -2f)+(1-ag') f g —2f
(as was first suggested by Bekenstein, "
area is proportional to the entropy of the black hole
though
d~
2M'(1 - &,')'"I. l+ (1- ~, )'"]
there were problems with this interpretation for a (56)
black hole immersed in a background of very low
temperature until Hawking discovered that black which diverges as a~ 1or v-0. One cannot bal-
holes not only absorb but also emit thermal radia- ance this divergence with a finite accretion rate,
tion"). At high values of a~, the emission is pri- as shown by the Mlowing argument: Bosons inci-
marily the spontaneous emission discovered by dent in the superradiant regime are not absorbed
Zel'dovich" that corresponds to the stimulated but amplified, increasing z. Fermions in this re-
emission of superradiant scattering. In this pro- gime are absorbed and do decrease z, but the ex-
cess, pairs are created in the ergosphere with clusion principle prevents more than one incident
particles (say) being emitted to infinity with posi particle per mode. Even if each such mode with
tive energies and their antiparticles going domo & &mQ is filled, the excess of absorption over
the hole with negative energies as measured at in- emission goes as I'(1+exp[ 2m' -'(~ —IQ)]}'.
finity but positive energies as measured locally. This dies sufficiently rapidly as lt, -0 that these
In fact, the antiparticles can even be on classical fermion modes cannot balance the effect on x of
trajectories at the horizon. Thus heat flows down the boson superradiant modes. Outside the super-
the hole as well as out to infinity, increasing the radiant regime, each accreting particle of energy
entropy of both. On the other hand, at lower val- dM «M can be shown to contribute dz &- M 'dM,
ues of a~ the emission is primarily thermal, draw- which can only decrease K at a finite rate with a,
ing entropy out of the hole. The process may still finite accretion rate dM/dt. As one tries to reduce
be regarded as the creation of pairs, with antipar- ~ by accretion (at least if the accretion is only a.
ticles going down the hole having negative energies small perturbation at any one time), eventually the
with respect to infinity, but outside the superradi- emission dominates and keeps v away from zero.
ant regime (which becomes negligible at small a~), Thus it is impossible to spin up a black hole adi-
the antiparticles also have negative energy locally abatically to the extreme Kerr configuration.
at the horizon and therefore cannot be on classical Figure 9 gives the maximum present value of the
trajectories. Instead, they are tunneling through rotation parameter a„ for a PBH with present
a classically forbidden region in virtual states mass M that was created 16 billion years ago, as-
that actually bring heat out of the hole. suming no spin up from incident particles. The
There is still some entropy produced by the par- curves resulting from the emission of neutrinos„
tial scattering off the gravitational potential bar- photons, or gravitons only are purely illustrative;
rier surrounding the hole, but outside the superra- the true maxi. mum is probably near or somewhat
diant regime this can only partially cancel the en- below the curve for the canonical combination of
tropy flow out of the hole and serves in effect to particle species, since those speci. es and possibly
increase the entropy emitted to the surrounding a few others are the one predominantly emitted
region for a given entropy loss by the hole. For for the mass range shown. For example, elec-
example, numerical calculations for a nonrotating trons, positrons, and all lighter particles will be
'
hole show that the emission of s = —, particles into emitted with negligible effects from their rest
empty space increases the external entropy by masses over the whole range shown, and muons
1.6391 times the entropy drawn out of the hole, s and heavier particles will also be emitted at: a
=1 particles increase it by a factor of 1.5003, s significant rate for M & 5& 10~~ g, as paper I pointed
= 2 particles by 1.3481, and the canonical combi- out. The graph shows that a PBH with M &10"g
nation of species gives 1. 6233 times as much en- should have a~ &0. 64 today.
tropy in radiation as the entropy decrease of the The asymptotic behavior of the graphs in Pigs.
3270 DON N. PAGE 14

1-9 at small c was given in functional form by spin even more, and the upper limit on - may be a,

'f
Eqs. (35) (46), and the parameters n, 8, y, 5, lower than unity; but unless small black holes
vz, and M, (a~, =-0, tc) are given in Table II for the were formed significantly more recently than 16
various combinations (n, &2, n„n, ) of species oi billion years ago, one may predict that any black
spin ~, 1, and 2. Note that hole found today with M & 10" g will have a
& 0. 000 042'3.
o. (a„= 0)
One can also get asymptotic forms near a =1.
P —2n g(a =~0 2f (a—„=0) The lifetime has already been given by Eq. (26)
with g,. =0, 7,. =0, and 7~ listed in Table II, and
M,. (a~, =1, t, ) wa. s given by Eq. (58). If we set
o. , -=f (a„=1}, (61)
behavior of a~ versus I
is the reciprocal of the exponent of the power-law
at the lower left edge of
Fig. 7. The ratio of the lifetime of a black hole
-
tl, =a(a„= 1), (62)
which can be evaluated by combining the numbers
with a, - =1 to one of the same initial mass with
of the last row of Table I according to Eq. (33),
a, . =-0 is3nv&, so Eq. (29) gives the initial mass
then integrating Eqs. (6) and (7) for a small time
of a PBH with e~,- = 1 that would just go away today '
t «M, from t =-0 at a =1 and M =M, gives
RS

M,. (a„,. =1, t,) =(t,/7, )'" M-M, (1 —o. ,M, st), (63)
Z-M, '(I P,M t),
=(3o'.~~) '~'M, . (a„,. =0, t, ),
here written in terms of the parameters in Table
II„Then M „(a~, t, } can be directly evaluated
(58)
:
a„—Z/M2- 1 —(P, —2o.', )M, st.
(64)
(65)

Since the mass decreases only infinitesimally with-

from the last quantity in Eq. (45) at small a„. One in the age of the universe if My» , to one can
can invert this asymptotic formula to obtain use Eq. (65) with M =M, and t =. t, a.s an asymptotic
a,„(M, t,)-[e'M/i', (a„, =I, t, )]'». approximation to a~ (M, t,) for large M. For ex-
ample, the canonical combination of species gives
for M «M,. (a,. =1, tc). For example, the canonical
combination of species gives a, (M, to = 16 x 10' yr) -1 (M/1. 500 x 10"' g)-'
a, ,„(M, t, ) - (M/4. 870 x 10" g)"" =1 —0.003378(M/10" g) '.
=4 234 x 10 '(M/10'~ g)s '" (60) (66)
The actual maximum is almost certainly somewhat This formula depends only weakly on the number
lower than this, since muons and other particles of spin- —, species, since gravitons dominate the
omitted in the calculation will have decreased the f
emission. However, since andg change so rap-

TABLE II. Pal. ameters in the asymptotic behavior of a slowly rotating black hole: n =f (a~= 0)
i' dM/dt, P —
==g(a„=-0) -
Ma~ ~dZ/dt, -y = n/(P —2— j
a) -dlnM/dlna„, 6 == (h —y)a„~da~ Vina„
ln(~/~M f) Tf ~(+~ = 0) = (lifetime from + ~ = 1) /(initial mass M&), and M'(a ~; = 0, t 0) =
—(initial
mass of a Schwarzschild hole ~ith lifetime &0) =(3+to)» 3. These are given for var ious combina-
tions of n&y2 spin-$, n& spin-l, and &2 spin-2 species.

(n(y2, n(, n2) 104m M, (a„.=p, t, )

(1, 0, 0) 0.818 30 6.161 08 0.180 86 0.323 38 2011.52 2.8728 x10~4 g
(0, 1, 0) 0.336 38 4.793 64 0.081 63 0.295 39 3856.63 2.1360 x10&4 g
(0, 0, 1) 0.038 36 1.062 65 0.038 91 0.280 73 32 560.0 1.0359 x 10&4 g
(0, 1, 1) 0.374 75 5.856 29 0.07338 0.264 05 3449.2 2.2143 x 10'4 g
(1, 1, 1) 1.193 04 12.017 36 0.123 87 0.282 14 1159.5 3.2575 x10~4 g
(2 1 1) 2. 01133 18.178 43 0.142 09 0.240 84 695.08 3.8770 x 10&4 g
(3, 1, 1) 2.829 63 24. 339 52 0.15148 0.246 51 502.44 4. 3442 x 10&4 g
(4, 1, 1) 3.647 93 30.500 59 0.167 21 0.260 47 394.60 4.7280 x10~4 g
(5, 1, 1) 4.466 23 36.661 67 0.161 07 0.256 06 325, 27 5.Q58Q xlp g
(6, 1, 1) 5.284 52 42. 82275 0.163 84 0.269 17 277.01 5„3497 1Q g
(8, 1, 1) 6.92111 55.144 90 0.167 57 0.266 10 214.02 6.8531 x1014 g
(10, 1, 1) 8.557 71 67.467 06 0.169 96 0.27164 174.65 6.2823 x 10&4 g
14 PARTICLE EMISSION RATES FROM A BLACK HOLE. II. . . 3271

idly with a near one (e. g. , decreasing roughly 10'%%uo nonrotating as one the same mass maximally ro-
between a~ =1 and a~ =0.9999), these asymptotic tating initially. The initial mass of a PBH created
formulas are only accurate very near a~ =1. 16 billion years ago that just disappears today
varies from 4. 73 &10' g for a Schwarzschild hole
V. CONCLUSIONS to 6. 26&&10' g for an extreme Kerr hole initially.
(This is for the emission of the canonical species;
The power emitted from a black hole in particles
the emission of muons and heavier particles will
of negligible mass and of spin ~, 1, and 2 are
make these masses somewhat greater, say 5 x 10"
strongly increasing functions of the rotation param-
eter a =J/M', varying in the range a =0 to a =1 g and 6. 6 x 10'» g, respectively. )
A black hole evolving from a~,. = 1 initially has
by factors of 13.35 for spin 2, 107.5 for spin 1,
and 26380 for spin 2. The power increases 299. 3
its area and entropy increase as heat flows into the
hole from particle pairs created in the ergosphere.
times for the "canonical combination" of 4 spin-&,
Then as a~ falls low enough (below 0. 89 for the
1 spin-1, and 1 spin-2 species that represent all
canonical species), the nonsuperradiant thermal
of the presently known particles with rest masses
emission begins to dominate, taking heat out of
less than 20 MeV. The power is greatest in spin-&
the hole and thus causing the entropy and area to
particles for a~ ~ 0. 6, followed by spin 1 and then
decrease. The maximum increase in the area is
spin 2; but for a ~0. 6 the order is reversed.
about 17.3%% for the canonical emission. For a
The emission of angular momentum also in-
Schwarzschild hole that emits its energy into the
creases greatly with a, even after the linear de-
canonical species in empty space, the emission
pendence expected at small a~ is factored out to 'A
process increases the entropy of the universe (—,
get the relative torque or logarithmic rate of de- black hole's
+entropy outside) by 62. 3% of the
crease in the angular momentum of the hole. The
initial entropy.
relative torque g behaves similar to the relative
Finally, it was shown that a black hole cannot
power fwith respect to spin and a~, but it is al-
be spun up to a =1. A PBH today is predicted to
ways sufficiently greater than 2 f, for the three
have a maximum rotation parameter as a function
spins calculated, that a black hole spins down to-
of mass that is given by Fig. 9 for 10' g&M&10 g
ward a Schwarzschild configuration much faster
and by Eqs. (59) and (66) for larger and smaller
than it loses energy. More than half of the energy
values of the mass. Black holes that are small
is emitted after a is reduced below a small value,
enough to emit many muons and heavier particles
less than 0. 06 for the canonical combination of
today are seen to be very nearly nonrotating.
species. At this point the power is within 1%%uo of
its Schwarzschild value, so the assumption that de- ACKNOWLEDGMENTS
caying black holes have negligible rotation is gen-
erally valid. This work was aided by discussions with many
Even though the power emitted is such a strong colleagues. Special thanks are due S. A. Teukolsky
function of a, the fact that a black hole loses u~ and %. H. Press for making available to me their
so rapidly means that the total lifetime for a given computer programs to calculate the absorption
mass varies only by a factor between 2. 02 (for the probabilities for gravitons and photons. F. J. Nagy
emission of spin 2 only) and 2. 67 (for spin 2 only) gave continued guidance on the use of the Lawrence
over all a~, A black hole emitting the canonical Berkeley CDC-7600 computer. K. S. Thorne off-
species has a lifetime 2. 32 times as long if initially ered constructive suggestions on the manuscript.

*Work supported in part by the National Science Foun- Perry, Astron. Astrophys. (to be published).
dation under Grant No. AST75-01398 A01 and by the ~
W. T. Zaumen, Nature 247, 530 (1974).
Danforth Foundation. G. W. Gibbons, Commun. Math. Phys. 44, 245 (1975).
S, W. Hawking, Nature 248, 30 (1974). ' B. Carter, Phys. Rev. Lett. 33, 558 (1974).
~S. W. Hawking, Commun. Math. Phys. 43, 199 (1975). S. W. Hawking, Phys. Rev. D 13, 191 (1976).
~B. M. Wald, Phys. Rev. D 13, 3176 (1976). S. A. Teukolsky, Phys. Rev. Lett. 29, 1114 (1972).
4L. Parker, Phys. Rev. D 12, 1519 (1975). ~5S. A. Teukolsky, Astrophys. .J. 185, 635 (1973).
~B. S. DeWitt, Phys. Rep. 19C, 295 (1975). ~
I. J. Schoenberg, Quart. Appl. Math. 4, 45 (1946); 4,
~J. B. Hartle and S. W. Hawking, Phys. Hev. D 13, 112 (1946).
2188 (1976). S. A. Teukolsky and W. H. Press, Astrophys. J. 193,
D. N. Page, Phys. Rev. D 13, 198 (1976). 443 (1974).
D. ¹ Page and S. W. Hawking, Astrophys. J. 206, 1 Ya. B. Zel'dovich, Zh. Eksp. Teor. Fiz. Pis'ma Red.
(1976). 14, 270 (1971) fJETP Lett; 14, 180 (1971)].
9B. Carter, G. W. Gibbons, D. N. C. Lin, and M. J. Ya. B. Zel'dovich, Zh. Eksp. Teor. Fiz. 62, 2076
3272 DON N. PAGE

angular mode dominates greatly and now has an and by Teukolsky" and Starobinsky and Churilov~'
effect increasing with s. for electromagnetic and gravitational waves. That
It is of interest to note that as a -
1, the surface is, the waves gain amplitude on reflection and ex-
gravity and hence temperature of the black hole go tract rotational energy from the hole in the wave
to zero, but the emission does not. In fact, Eq. analog of the Penrose process. ' Bekenstein" has
(10) becomes shown that this result follows from Hawking's
1)""r„„,(a„x)a(m area theorem" for waves with positive-definite en-
( 2x), (47)
ergy density. For fermions (2s odd), I" is always
where H(m —2x) is the Heaviside step function (0
positive, as Unruh" has shown for the classical
neutrino field, which has a negative energy densi-
if m —2x&0, 1 if m —2x&0), so one gets simply
ty near the hole in the superradiant regime. In the
the spontaneous emission (first discovered by
quantum analysis, the amplification of a boson
Zel'dovich") inthe superradiant regime where the
wave corresponds to stimulated emission, whereas
angular velocity ur/I of the wave is lower than the
the Pauli exclusion principle prevents fermions
angular velocity
from being amplified. The fact that this behavior
Q~
a~~1
2M shows up in the solutions of the classical wave

of the hole. For bosons (2s even), I' is negative tween spin and statistics. "
equations is a manifestation of the connection be-
Field-theoretic deriva-
in the superradiant regime, as predicted by tions of the spontaneous emission from a rotating
Zel'dovich" "and confirmed by Misner, Starobin- ' black hole with the appropriate initial state for no
sky, ~ and Press and Teukolsky for scalar waves, thermal emission have been given by Unruh" and
Ford, '0 but one must remember that a black hole
formed by collapse has a nonzero temperature
(except when a~ = 1) and thus emits at a greater

i I I

-I
IO

IO-i

IO C3
D

~ IO-
o

II

I
O-4

IO6 I

0 .2 .6 0
1.
a =J/M
0-6
FIG. 1. Power emitted in various combinations of
I

0 .2 4 .6 .8 I.Q
species by a rotating black hole, expressed in a scale- o~= J/M'
invariant way byf. The symbol (n~yq, n~, n2) denotes a
combination of n&y2 spin-2, n& spin-l, and n2 spin-2 FIG. 2. Relative torque emitted by a black hole (i.e. ,
species, where each species is assumed to have two the rate of emission of angular momentum, divided by
polarizations (e.g. , left-handed neutrino plus right- the angular momentum of the hole), expressed in a
handed antineutrino), scale-invariant form by g.
14 PARTICLE EMISSION RATES FROM A BLACK HOLE. II. . . 3273

(1972) [Sov. Phys. — JETP 35, 1085 (1972)]. L. H. Ford, Phys. Rev. D 12, 2963 (1975).
C. W. Misner, Bull. Am. Phys. Soc. 17, 472 (1972). ~~H.Fritzsch and P. Minkowski, Nuovo Cimento 30A,
'A. A. Starobinsky, Zh. Eksp. Teor. Fiz. 64, 48 (1973) 393 (1975).
—
t.Sov. Phys. JETP 37, 28 (1973)]. R. P. Feynman and M. Gell-Mann, Phys. Rev. 109, 193
W. H. Press and S. A. Teukolsky, Nature 238, 211 (1958).
(1972). ~~H. Fritzsch, Caltech Report No. CALT-68-524, 1975
A. A. Starobinsky and S. M. Churilov, Zh. Eksp. Teor. (unpublished) .
—
Fiz. 65, 3 (1973) [Sov. Phys. JETP 38, 1 (1974)]. ~4H. Fritzsch, M. Gell-Mann, and P. Minkowski, Phys.
~R. Penrose, Riv. Nuovo Cimento 1, 252 (Numero Lett. 59B, 256 (1975).
Speciale 1969). ~~H. Fritzsch and P. Minkowski, Phys. Lett. 62B, 72
~5J. D. Bekenstein, Phys. Rev. D 7, 949 (1973). (1976).
S. W. Hawking, Phys. Rev. Lett. 26, 1344 (1971). J. D. Bekenstein, Phys. Rev. D 9, 3292 (1975).
W. G. Unruh, Phys. Rev. Lett. 31, 1265 (1973). ~
J. M. Bardeen, B. Carter, andS. W. Hawking, Commun.
See, e.g. , W. Pauli, Phys. Rev. 58, 716 (1940). Math. Phys. 31, 161 (1973).
W. G. Unruh, Phys. Rev. D 10, 3194 (1974).