You are on page 1of 10

PHYSICAL REVIEW D VOLUME 41, NUMBER 12 15 JUNE 1990

Can black holes support a gravitating massive scalar field?


Valeria Ferrari
International Center for Relativistic Astrophysics, Dipartimento di Fisica "G. Marconi, "
Uniuersita di Roma, 1-00185 Rome, Italy

Basilis C. Xanthopoulos
Department of Physics, Uniuersity of Crete and Research Center of Crete, Iraklion, Crete, Greece
(Received 1 1 January 1990)
We have used the Kaluza-Klein theory to study a class of possible couplings between gravity and
massive scalar fields. The Einstein equations have been integrated and the conformal structure of
the spacetime has been explicitly determined.

I. INTRODUCTION This is usually referred to as a "minimally coupled" (MC)


scalar field. m =O corresponds to the massless case.
One of the most remarkable predictions of the theory Thus, black holes cannot support an exterior M C scalar
of general relativity is that the collapse of astrophysical field. These conclusions confirmed a result obtained in
configurations, as complicated as massive stars, eventual- 1968 by Janis, Newman, and ~ i n i c o u r . ' They found a
ly leads to the formation of a macroscopical object, a static spherically symmetric solution of Einstein's equa-
black hole, which is identified by only three parameters: tions coupled to a M C scalar field, which is a generaliza-
the total mass, the angular momentum, and the charge. tion of the Schwarzschild solution, and exhibits a naked
The corresponding geometry is described by the Kerr- singularity instead of the horizon. In 1970 chase4 more
Newman family of solutions. Conjectured by Wheeler, generally showed that in static and asymptotically flat
and proved in successive steps by the work of Israel, Car- solutions coupled with a M C massless scalar field, the
ter, Hawking, Wald, Robinson, and Mazur, this is the field is always singular on a simply connected event hor-
essence of the "no-hair theorem" for black holes. izon. The behavior of the M C massless scalar field was
The no-hair theorem has now been established (1984), also analyzed in terms of scalar perturbations of a
but only in the framework of the two classical interac- Schwarzschild solution by price5 and ~ e i t e l b o i m and
, ~ it
tions, gravity and electromagnetism. However, since was shown that the field must vanish in the exterior of
black holes are formed out of strongly interacting materi- the star.
al, as the nuclei of stellar components, and since scalar But the story does not end here. Subsequent investiga-
and vector mesons are responsible for mediating the tions by ~ e k e n s t e i n ' somehow weakened his previous
strong interactions, there could exist a field exterior to conclusions about the incompatibility of scalar fields with
black holes associated with these particles. black holes. H e proceeded as follows. H e developed a
The question whether the existence of black holes can technique to generate from any solution of Einstein-
be compatible with the presence of a scalar field was con- Maxwell equations in the presence of a minimally cou-
sidered by ~ e k e n s t e i n "in~ 1972. H e showed that a static pled massless scalar field, a new solution coupled with a
o r stationary black hole cannot support an exterior mass- conformal massless scalar field. The novelty of these
less or massive scalar field, if the solution is required to solutions is that a new coupling gravity-scalar field is con-
be asymptotically flat and free of naked singularities. (He sidered. In fact, a conformal scalar field is described by
also ruled out vector and spin-2 fields.) an energy-momentum tensor different from ( I ) ,
Bekenstein's conclusions are reached on the basis of
the following assumptions: (i) The physical scalars must
be bounded everywhere in the exterior region, and on the
nonsingular horizon; (ii) the scalar field is a neutral, spin-
less meson field described by the energy-momentum ten-
sor and the specific equations satisfied by the conformal sca-
lar field are

where V p indicates covariant differentiation, and p


satisfies the Klein-Gordon equation as generalized in where R is the scalar curvature. Incidentally, it might be
curved spacetimes by replacing the ordinary derivative noted that since the scalar field is massless, as a conse-
with the covariant derivative: quence of the Einstein equations, R is zero in these solu-
tions. Among them, he found one of particular interest.
It is a charged solution described by the metric

3652 '8
1990 The American Physical Society
CAN BLACK HOLES SUPPORT A GRAVITATING MASSIVE.

field coupled with a massive scalar field in four dimen-


sions. The nature of the couplings will be discussed in
detail in the next section.
The equations have been integrated and the conformal
structure of the resulting spacetime has been explicitly
determined. The general solution is asymptotically flat,
provided the conformal factor is chosen to tend to unity
where at infinity.
For a particular selection of the parameters and of the
conformal factor, we find the Bekenstein metric ( 5 ) in the
uncharged case. In our context, however, the interpreta-
e and q are, respectively, the electric and the scalar tion of the solution is different, since the scalar field is
charge of the black hole. (The conventions of the present massive, and it possesses an energy-momentum tensor
paper are Rab=R ma,,,b, Gab= T a b , T a b ( e ' = ~ Fbm am different from that of the conformal massless scalar field
- ,ga,F,,,,Fmn.)
1 The divergence of the scalar field on the given in Eq. (3). The geometry is the same in both solu-
surface r = M was first regarded as a proof of the physical tions, but the sources are different.
unreality of this solution. Subsequently, however, Beken- Although we d o not give a definite answer to the ques-
stein8 showed that the divergence of q, is not pathologi- tion put in the title of this paper, we develop a procedure
cal, since it is not associated with an infinite potential which could be used to find solutions which have the
barrier for test scalar particles. In addition, it is not con- properties of a black hole surrounded by a gravitating
nected with an unbounded tidal acceleration between massive scalar field.
neighboring trajectories, and n o particle terminates its It should probably be clarified that, in considering
life on the horizon in a finite amount of proper time. In different couplings between gravity and scalar fields, we
conclusion, r =M should not be considered a physical d o not introduce any illegitimate element into the classi-
singularity. Therefore, he predicted the existence of cal theory of gravity. Needless to say that all couplings
charged black holes with an associated conformal mass- we are considering are consistent with the Bianchi ihenti-
less scalar field. ties, that the energy-momentum tensor is conserved, and
The no-hair theorem eventually overcame the chal- that the equation satisfied by the scalar field, the massive
lenge provided by the conformal scalar field: in 1978 Klein-Gordon equation, is a consequence of the
Bronnikov and ~ i r e ~ eshowed v ~ that the Bekenstein divergence-free condition of the energy-momentum ten-
solution is unstable under radial perturbations. However, sor. We only investigate different energy-momentum ten-
the legacy of this solution remained: (i) Different cou- sors compatible with the standard field equations.
plings between gravity and scalar fields produce different One might ask whether we are entitled to choose ad li-
results, and therefore they should be investigated; (ii) bitum any sort of couplings, although allowed by the
mere divergent behavior of the scalar field, in the absence mathematical structure of Einstein's equations, between
of naked singularities in the geometry, does not exclude gravity and scalar fields. This is, of course, a very deli-
the possibility of the existence of black holes with an as- cate question. In fact, contrary to the Einstein-Maxwell
sociated massless scalar field. theory, the theory of scalar fields in the framework of
More recently, the consequences of the coupling be- general relativity is not complete: we d o not know what
tween gravity and massless scalar fields have been further the form of the Lagrangian and of the corresponding
investigated. For instance, the general static spherically energy-momentum tensor should be. In the absence of
symmetric solution of Einstein's equations coupled with a any definite prescription, we can only try "experiments,"
massless M C scalar field has been determined by Wy- and analyze the consequences of different couplings.
man,'' and generalized by Xanthopoulos and Zannias"
to the case of arbitrary dimension D 2 4 of the spacetime. 11. FROM FIVE TO FOUR DIMENSIONS
In addition, the general exact nonstatic solution for a
conformal massless scalar field coupled to a Robertson- It is known that the Kaluza-Klein theory provides a
Walker geometry has been found by Deng and unified geometrical description of the gravitational and
~annheim.'~ the electromagnetic fields. In 1973 Belinskii and Khalat-
However, no comparable attention has been paid to the nikov15 showed that scalar fields can be included in this
case when the scalar field is massive. The only result is description, and that the Kaluza-Klein theory can be un-
the one quoted above, asserting the incompatibility of derstood also as a scalar-vector-tensor theory of gravity
black holes with a minimally coupled massive scalar field. in the spirit of the Brans-Dicke ideas. We shall here spe-
Since, as we have seen, different assumptions on the na- cialize to the case when only a scalar field is coupled to
ture of the interaction between gravity and scalar fields gravity, addressing the reader interested in more general
can lead to different results, it is interesting to investigate cases to their paper.15
some other possibilities. Belinskii and Khalatnikov start their investigation by
Following the idea introduced by Kaluza and considering the five-dimensional metric
~ l e i n , ' ~ we
, ' ~ have used a dimensional reduction of
Einstein's equations written for five dimensions in a vacu-
um, to obtain the equations describing a gravitational The components of the metric are assumed to be indepen-
3654 VALERIA FERRARI AND BASILIS C. XANTHOPOULOS !I
dent on the fifth coordinate x4. In addition, they intro- It should be noted that the Bianchi identities for the
duce a scalar function p ( x i )and assume that five-dimensional vacuum metric (6) imply that the entire
family of theories we are considering-Eqs. (10) and ( 1 1)
for any choices of A ( q ) and B ( p ) - i s consistent: the
Throughout the paper we shall assume that A is not a divergence-free condition on the energy-momentum ten-
constant. N o assumptions are made at this level on the sor ( 1 3 ) is equivalent to the field equation (10).
symmetries of the spacetime, and gik = g i k ( x i ) . The Ein- In some respect, these theories are "naturally
stein equations written for five dimensions in vacuum are designed" to describe massless scalar fields. In fact Eq.
(10)can easily be put in the form
v 2 (. . )=O,
where the parentheses contain a combination of deriva-
tives of the functions A, B, and p. One could therefore
introduce a new scalar field $ such that

and it is obvious that it would be massless. However, we


can impose an additional condition on the field p , and
select among the solutions of the entire set of Eqs. (10)
and ( 1 1 ) those which satisfy the additional constraint.
The assumption we shall make hereafter is that the scalar
field is massive, and therefore Eqs. (10) and ( 1 1) need to
be supplemented by the Klein-Gordon equation

and p,,=O identically.


In these equations pOB where m is assumed to be positive. Thus the theory will
( ~ , D = O , .. . , 4 ) are the
components of the five- describe a massive scalar field coupled to gravity only for
dimensional Ricci tensor, R i k , (i,k=O, . . . , 3 ) are the those particular solutions which satisfy Eqs. ( l o ) , ( l l ) ,
components of the four-dimensional Ricci tensor, and the and (14). Of course, the imposition of the additional con-
prime indicates differentiation of the functions A and B dition will be traded with the loss of one of the degrees of
with respect to their argument p. freedom associated with the choices of A and B.
We have rewritten the previous equations in a form
suitable for our successive developments:
111. THE STATIC SPHERICALLY SYMMETRIC
EQUATIONS

We shall now specialize the equations obtained in the


previous section to the case when both the geometry and
the scalar field are static and spherically symmetric. We
write the metric in the form

where ( ~ p ) ~ = ( ~ , p ) ( In
~ ' the
p ) reduction
. of Eq. ( 1 1 )
use has been made of Eq. (10). The scalar curvature and where K, P, and A are functions of r, retaining the free-
the energy-momentum tensor associated with the previ- dom to impose a coordinate gauge condition among the
ous equations are metric functions. In this coordinate system the "Lapla-
cian" of the scalar field p is

and

Hereafter, a prime and an overdot will indicate


differentiation with respect to p and r, respectively. Since

the Klein-Gordon equation becomes


CAN BLACK HOLES SUPPORT A GRAVITATING MASSIVE.

Equation (18) must be considered together with the field


equation (101, which is convenient to rewrite
+
v2g, [In( A 'B )]'( ~ g)I =
, 0 . (19) By making use of Eqs. (17) and (261, the right-hand
Using Eqs. (16) and (17) Eq. (19) becomes sides of Eq. (11) can be written as

[ln [w1 /'+[In( A1B2)]'=0 R,, = -[ln( AB2)]'-


KK
' - K 2 ( l n ~)'
p2 g,

Equation (20) can be integrated with respect to g, to give

where p is an integration constant. In order that Eqs.


(18) and (19) be consistent, the following condition must
hold:
[ [
+ ( l n B ) ' In ----
A 5 5 I]'
or, equivalently,
+ r 2 A 2 ( 1 n B ) [ln 1% 1 If$ .

It should be noted that the term (Vig,)(Vkg,)contrib-


utes only to the Rrr component of the Ricci tensor.
Equations (21) and (23) can now take the place of Eqs.
(18) and (19) in our forthcoming analysis. IV. THE INTEGRATION OF THE EQUATIONS
Next we have to consider the remaining Einstein equa-
tions. The nonvanishing components of the Ricci tensor We shall now concentrate on the integration of the ( t t )
for the metric (15) are and (88) components of the Einstein equations, which
play the central role in our analysis. Putting together the
left- and the right-hand sides of Eqs. (24) and (27), and
converting the r derivatives into g, derivatives by using
Eq. (211, the ( t t ) component gives

It should be noted that the variable r has completely been


eliminated by the introduction of the prime derivatives,
and this occurrence is crucial for the subsequent integra-
tion of the equation. Equation (28) can be written in an
equivalent form

The last equation can be easily integrated, yielding

The R dd component has not been included, and the (44)


Einstein equation will not be considered in the sequel, be- where y is an integration constant, and we have omitted
,,
cause it differs from R only by a factor sin28. an irrelevant multiplicative constant in the right-hand
side.
The nonvanishing Christoffel symbols which are
relevant for computing the second covariant derivatives Similarly, the 88 component of the Einstein equations
appearing in Eqs. ( 11) are can be integrated. However, it is now necessary to ex-
press everything in terms of r derivatives:

We easily find that


3656 VALERIA FERRARI AND BASILIS C . XANTHOPOULOS

By inserting Eq. (30) into Eq. (21) we obtain

We set
w=A"+' and a = P ( y + l ) , (33)
so that Eq. (32) becomes

We can now use Eqs. (30) and (34) to eliminate K and B where we have chosen multiplicative constants in the
from Eq. (31). The result is right-hand side of the expressions (41) such that
lim ( K g ) = lim (AB ) = I .
r -CS r +m

The last equation we need to satisfy is the rr com-


ponent of Einstein's equations. It turns out that its mere
role is to impose one constraint on the integration con-
stants y and y ' .
The last equation clearly suggests the choice of the gauge: By transforming the second of Eqs. (27) in a way that
only differentiations with respect t o r appear, the right-
hand side of the rr equation becomes
which corresponds to the choice of an isotropic coordi-
nate system. In this gauge the problem reduces to solving
the third order, ordinary differential equation for the
function w(r):

where the left-hand side is meant to be given by the


second of Eqs. (24). We have evaluated the two sides of
By performing the successive substitutions Eq. (42),leaving B unspecified, by using the symbolic ma-
nipulation language MACSYMA. The result is that the
w = e Z and then i = $ , (37) equation is satisfied for any B ( r ) , provided the parame-
we eventually obtain ters satisfy the condition

In the previous equation we have ignored the f sign,


The general solution of Eq. (38) is since the minus sign would merely correspond to chang-
ing rO- - ro. The final solution is

,
(44)
+ro 'y+1)/Vy2+y+l
where e1,e2are integration constants.
I t turns out that the physically interesting solutions
correspond to the choice ele2<O. We shall comment on
AB=PB= I---
:;j - ] ,

the alternative choices ele21 0 at the end of this section. and


From Eqs. (37) and (38) we eventually obtain
,

where have been replaced by the new constants r, Thus, we have been able to determine the conformal
and y and we have ignored a third irrelevant multiplica- structure of the spacetime, B being the unspecified con-
tive constant in w . Without loss of generality we shall as- formal factor. The necessary and sufficient condition for
sume that ro >O (the case ro=O should be investigated the geometry to be asymptotically flat is
separately) and 0 5 r < m .
lim B =const+O
Having obtained w , we reconstruct the solution I-a
CAN BLACK HOLES SUPPORT A GRAVITATING MASSIVE. ..

Therefore, we shall assume, without loss of generality,


that
lim B = l . For the solution to be asymptotically flat we demand that
r-m

It is now straightforward to write the Klein-Gordon ry/(y+l IE? 1


B=e -, where lim Q ( r ) = l .
equation, which is equivalent to Eq. (23): Q(r) r-m

In this case K is well behaved, but for no choice of the


parameters A could exhibit the correct asymptotic behav-
ior.
It should be useful to summarize the results of this sec- (iii) When e2=0, the rr component leads to the con-
tion. For any choice of B satisfying the condition (461, +
straint y ,( y 2 + y 1 ) = 0 , which again does not admit
the geometry will be completely determined. Then one nontrivial solutions.
has to solve the Klein-Gordon equation (47). Having ob- (iv) The case y = - 1 is not covered by the analysis of
tained a solution cp=p(r), the explicit dependency of A the present section, because w = 1. In this case, for P = A ,
and B on p could be found by eliminating r in favor of p. we find
Then from Eq. (13) the energy-momentum tensor could A K B Z 1 and A B A ~ ~ = P .A
be determined and the nature of the coupling would be
specified. The 86' equation then gives
We have not been able to guess simple choices of B ( r )
for which the Klein-Gordon equation admits a simple
solution. However, as we shall see in the next section, the
conformal structure of the solution allows us to answer which, with the substitution [ln(AB = $ - I , )I'
reduces to
some of the physical questions which motivated our in- a linear equation. Eventually we obtain the solution
vestigation. given by Eqs. (44) and (45) in the case y = - 1.
T o conclude this section some additional technical re-
marks are needed.
(i) When > 0 the solution for w , A, KB, and AB V. HORIZONS AND SINGULARITIES
can be found, and it involves inverse trigonometric func-
tions. However in this case the rr component of We are mostly interested in those solutions which are
Einstein's equations leads to the condition ( y 2 + y regular on r = r o . The conformal invariance of the Weyl
+ +
1 )y:+ ( y 112=0, which does not admit nontrivial tensor suggests that the scalar is the curvature
real solutions. invariant with the simplest dependency on B. For the
(ii) When el =0, then solution (44)we find

where [ . . . ] stands for a polynomial in r. We easily see tion of our analysis will be played by the function Q ( r ) .
that, in order to avoid the occurrence of a curvature We find convenient to introduce a new parameter 6
singularity at r = r o , B should vanish at least as fast as defined as
B- ( r - r o ) 2 - [ ( ~ i ~ ) / d ~ 2 + ~ + l ~, (49)

Therefore we set
It is easy to see that
-2 < 6 < 2 for all real values of y . (53)
In terms of Q and 6 the solution is
where Q is assumed to be finite at r = r o . The necessary
6-2
and sufficient condition for the asymptotic flatness of the 1 ( r+r0)3 1
solution now becomes r 2 ( r- r o ) Q ( r ) '
lim Q = l . (51)
r- m

From now on the role of the function B as the free func- and the Klein-Gordon equation becomes
3658 VALERIA FERRARI A N D BASILIS C. XANTHOPOULOS 41
For mro#O the previous equation has always two real
roots, one positive, one negative,

Next we investigate the behavior of some other curva- and the two linearly independent solutions of the Klein-
ture invariants at r = r o , in order to determine the behav- Gordon equation have the behavior
ior of Q which would avoid the occurrence of curvature
singularities at r =r,. We find that
The first will be finite, the second divergent at r = r o .
The solution which is physically acceptable at infinity,
p,,,, will be matched to a linear combination of pro,,and
Rabcd~ - r pro,,. Therefore p will in general diverge on r =r,.
a bcd -
(r+ro) However the reader should be reminded that an un-
bounded scalar field is not necessarily unacceptable from
a physical point of view (see Refs. 7 and 8 and comments
in the concluding remarks).
In conclusion, we can find solutions which do not pos-
sess any curvature singularity on r =r,, but that surface
cannot, in general, be a horizon.
The energy conditions. Before working out some
specific examples, by making particular choices of the
where [ . . . ] is a polynomial in {Q, Q, Q, r, r,, and 6 ) . function Q, let us find the constraints that the physical re-
Obviously, when Q is smooth at r = r o , the previous sca- quirements of the positivity of the Arnowitt-Deser-
lars will be well behaved there, and, although this is not a Misner ( A D M ) mass and the other energy conditions im-
rigorous mathematical proof, no curvature singularities pose on that function.
are expected to develop. On the other hand, when Q is The A D M mass is determined from the coefficient of
-
finite but not smooth at r = r,, behaving like Q ( r - r o )T the 1/r term in the asymptotic expansion of K. We easily
where 77 is positive but not an integer, curvature scalars find that if Q is chosen to behave as
involving sufficiently high-order derivatives of Q will be
singular.
Thus, there exists a subclass of our general solution, for
which the geometry does not present any pathological be- then from Eq. (54) it follows that
havior on the surface r = r o However it should be noted
that, since -2 < 6 < 2 , Eqs. (54) imply that r = r o could
not be a horizon either.
We shall next concentrate on the behavior of the scalar Therefore M is the A D M mass, and we shall assume
field. From Eqs. (51) and (55) we find that the asymptotic M10.
form of the Klein-Gordon equation is Next we introduce the notation

which admits the two linearly independent solutions where p is the energy density, and p,, p o , pg are the com-
ponents of the pressure. The energy conditions are
p>O and p+p, 2 0 ( i = r , Q , + ) (weak) ,

p + z p , 2 0 (strong) ,
We require that the scalar field tends to zero at infinity; I

therefore, the acceptable solution is p?lp,j ii=r,B,+) (dominant).


Of course, we cannot investigate whether these condi-
tions are satisfied everywhere without specifying Q ex-
and then we analyze the behavior of p near r =r,. As- plicitly. However, we can restrict ourselves to the re-
suming that p- ( r - r O Ik when r -ro, the leading terms quirement that they are satisfied in the asymptotic region,
in ( r - r o in the Klein-Gordon equation give the follow- and this will determine the function Q up to the second
ing indicia1 equation for k: order [one more than in Eq. (6211. U p to the second or-
der in r, Q can be written as
41 CAN BLACK HOLES SUPPORT A GRAVITATING MASSIVE. . . 3659

where E is a new constant to be determined. From the VI. SOME PARTICULAR EXAMPLES
condition p L 0 we find that E should be
In Sec. V we investigated the nature of the geometry at
r = r o > 0 for all physically acceptable functions Q = Q ( r )
under the implicit assumption that Q ( r ) has no roots o r
poles for r = r l # r o . We reached the conclusion that (a)
where e 1 0 . From p + 2 ,pi > 0 it follows that either r = r o is a curvature singularity, or (b) r = r o is reg-
ular, but it cannot be a horizon. The scalar field always
diverges at r =ro.
In this section we shall investigate in some detail a few
In addition, more examples, considering also cases in which Q ( r ) has
poles for r = r , #ro, and r #O. In order to obtain non-
singular solutions, we shall assume also that the expres-
sion in square brackets in Eq. (56)

Thus we shall assume


has roots for r > 0. The roots of the polynomial D are
real provided

and consequently In addition, they will be positive if 6 > 0. Taking into ac-
count the permitted range of 6 given by the expression
(53), we find that D has real positive roots only if

For 6 = 1 , r = r o is a double root. For 0 < 6 < 1, the two


It is now straightforward to verify that all the energy roots r , and r, are
conditions are satisfied (for some of them to the new lead-
ing order in r), provided that we choose

and they satisfy the relation 0 < r , < ro < r,. Let us con-
sider the following cases.
(i) 6 =O.
Thus, we have determined the asymptotic form of the
function Q up to second order, which guarantees that all
the physical requirements are satisfied, at least for
sufficiently large values of r.
Equation (56)gives
We are now in a position to draw some conclusions
about the possibility that black holes could support an
exterior massive scalar field. In this section we have seen
that we can find solutions free of curvature singularities
for r > 0, provided Q has the proper asymptotic behavior Let us assume
and it is smooth at r = r o . At this level, we are still left
with a lot of freedom in the choice of the function Q. In
principle, nothing prevents us to choose Q such that it
satisfies the previous requirements not only for large r, It is apparent that for n > 2 the invariant (78) will be
and it has a pole somewhere, say r = r l . This pole would singular for r =O. Thus we shall assume n =2. It is
be a zero for K, and, by a clever choice of the parameters, straightforward to find that the only choice of the
it could combine with the terms in the parentheses of Eq. coefficients in Eq. (79) which is compatible with the ener-
(48) in such a way that the curvature scalars would not be gy conditions at infinity is
singular. In this case the solution could represent a black
hole supporting a massive scalar field. One might argue
that this procedure is very tricky and that to believe that
all these conditions can be satisfied is to believe in mira- However the resulting solution exhibits a curvature
cles. However we shall show in the next section, as an singularity at r = O which is manifested by the divergency
example, that with a suitable choice of Q and of the pa- of the scalar curvature R . In addition, for all possible
rameters one obtains the Bekenstein solution, which can values of r, and E the energy density p becomes negative
therefore be interpreted, in the framework of our theory, in some region, although it is positive asymptotically.
as representing a black hole surrounded by a massive Therefore, no physical solutions have been found for
field. 6=0.
3660 VALERIA FERRARI AND BASILIS C. XANTHOPOULOS

considering.
(iii) 0 < 6 < 1.
In this case
- e~4 ( cr- r l, l2(r
c ~ ~ ~ ~ ,- r 2~ )2 . (88)
and
It is interesting to choose
Q=(r-r,)-'/'~,(r), i = 1 or 2 , (89)
where Q, ( r ) is smooth at r = r,. We shall assume that Q,
The pole in the metric (81) can be eliminated if
Q - ( r -ro )-I. With this choice the invariant (82) is
does not vanish at r = r , , so that K -
( r - ri ) vanishes
smoothly at r = r , . Then we find
finite at r = r o If we choose Q = ( r + k ) / ( r - r , ) , and
k > 0 , we find that the energy conditions cannot be
satisfied throughout the spacetime. If we assume, in-
stead, that Q is the ratio of polynomials of second degree,
such that r = r , is a simple pole, we are forced by the en-
ergy conditions to choose

-
and Qi is assumed to behave like Qi r 'I2 asymptotically.
We have considered the cases ~ , = r ' / ~and
Then
+
Q, = ( r r, )'I2, and in both cases r = r, is in fact a curva-
ture singularity (the scalar curvature R becomes infinite).
(iv) ro = 0.
In this case K = A = Q - I , and the resulting solution is
conformally flat. For a general smooth Q the scalar cur-
and the resulting geometry vature diverges like R -r -I. In order to eliminate this
singular behavior, we need to choose QQ -r. We have
considered three possible different choices: namely,

is isometric to the Bekenstein solution given by the first


of Eqs. (5) when the charge e is equal to zero. This can be
easily seen by changing r + r -M, and r o + M . Thus, in where a > 0, b > 0. In all cases, the four curvature scalars
our coordinate system, r = O corresponds to the regular listed in Eq. (56) are finite at r =0, which is not a hor-
horizon of the Bekenstein solution. The energy density izon. However no choice of a and b allows the energy
and the pressure are given by conditions to be simultaneously satisfied throughout the
spacetime.

VII. CONCLUDING REMARKS

and they satisfy all the energy conditions throughout the The analysis carried out in this paper certainly does
spacetime. not provide a proof of the existence of black holes sur-
It should be stressed that, although the geometry de- rounded by a gravitating massive field. However it gives
scribed by the solution (85) coincides with that of the indications on how these solutions could be found, if they
Bekenstein solution, the two solutions are different from exist.
a physical point of view, since the source which produces The solution given in case (ii) of Sec. VI presents a very
the geometry is different in the two cases. In the Beken- interesting situation. The geometry is that of the Beken-
stein case it is a conformal scalar field which satisfies the stein solution ( 5 ) (when e =0), and therefore it is asymp-
massless Klein-Gordon equation (41, and the correspond- totically flat, it has a smooth horizon at r = r o , and it is
ing energy-momentum tensor is given by Eq. (3). In our free of naked singularities. However, the scalar field is
case the scalar field is massiue, the Klein-Gordon equa- now a massiue Klein-Gordon field, whose energy-
tion being momentum tensor is given implicitly by Eqs. (13) and
(87), and it is singular at r = r o , where it diverges as a
power law. How seriously should we take a singularity
which appears in the scalar field but it is absent in the
However, since we have not been able to solve Eq. (87)) geometry? T o answer this question, we first consider the
compute A and B in terms of the resulting q,, and then motion of test particles which are not coupled with the
the corresponding energy-momentum tensor given by Eq. scalar field. They are affected by the presence of the sca-
(13), we cannot specify the nature of the coupling we are lar field only indirectly, via the geometry which deter-
!?! C A N BLACK HOLES SUPPORT A GRAVITATING MASSIVE. . . 3661

mines their geodesic motions. Since the geometry is reg- upon the assumptions on the interaction action, and it is
ular at r = r o , these particles would never get any hint bounded for the simplest one.
that something pathological is happening on the horizon. In one aspect the eventual outcome of the paper over-
The only probes that could possibly feel the divergence came our initial expectations. In fact we were surprised
of the scalar field on the horizon are those "particles" to realize that for all possible couplings described by Eqs.
that are coupled directly with the scalar field, in the same (101, ( l l ) , (131, and (14),it is possible to determine explic-
sense as charged particles are coupled directly with the itly the geometry of the spacetime, expressed in closed
electromagnetic field. The relevant question therefore is form. However the fact that we have not been able to
whether these particles fall onto the singularity r = r o in a determine explicitly any physically acceptable choice of
finite, or in an infinite, amount of proper time. And one the conformal factor Q for which the Klein-Gordon
should regard r =r, as a physical singularity if that inter- equation could be solved, and the associated coupling ex-
val is finite. For these particles the equations of motion plicitly found, is disappointing. Apart from the Beken-
are stein spacetime, we did not find any other geometry ad-
mitting a regular horizon.
gmV, = force terms It should be stressed, however, that the analysis carried
out in Sec. V I and the examples considered there are by
The right-hand side of this equation is determined from no means exclusive and complete. There are countless
the interaction action we are considering, it is chosen by choices of Q ( r ) which would be compatible with the
hand (it is not determined by the energy-momentum ten- physical requirements, whose systematic investigation is
sor), and its choice derives from considerations which are outside the scope of this paper. We also think that, in
outside the realm of general relativity. In Ref. 8 Beken- demanding that the sought solutions should satisfy all
stein considered the simplest parameter-independent, conceivable energy conditions, we were probably very
conformally invariant interaction action. His analysis strict. With a more flexible attitude, one could seek for
showed that the particles approach the horizon only asymptotically flat geometries with horizons, while being
asymptotically. In Bekenstein's words, all test scalar willing to give up some of the energy conditions.
charge trajectories which approach the horizon from the Note added in prooj In order to describe a massive
exterior are complete. Since the geometry is the same in Klein-Gordon scalar field within the Belinskii-
our solution, the same conclusions hold. Of course, one Khalatnikov framework, we have introduced the nonco-
could have considered different couplings that would variant reduction expressed by Eqs. (18) or (23). We now
transform r =r, into a physical singularity for particles think that Einstein's equations can similarly be integrated
with scalar charge. in the massless case, when the scalar field is subject only
The conclusions are similar when one considers the ti- to Eq. (10). The solution for the geometry is again de-
dal acceleration that a particle experiences near the hor- scribed by Eqs. (54) for an arbitrary Q ( r ) . However,
izon. The tidal acceleration is determined by the Jacobi since the equation which determines p ( r ) in terms of Q ( r )
' which satisfies the geodesic deviation equation
field 7 is different, the solution will describe a coupling different
from the one considered in this paper. In the massless
f m V m ( { n V n ~ a ) = R m n s '.~ m ~ " ~ case the reductions are fully covariant.
For particles which interact only gravitationally, the
smoothness of the geometry implies the boundness of the ACKNOWLEDGMENTS
tidal forces. For particles that do interact directly with
the scalar field, the right-hand side of the previous equa- The research reported in this paper has partially been
tion would contain a force term as well. Whether or not supported by the International Center for Relativistic As-
the acceleration is bounded in these cases depends again trophysics.

'J. D . Bekenstein, Phys. Rev. D 5, 1239 (1972). 'OM. Wyman, Phys. Rev. D 24, 839 (1981).
*J. D . Bekenstein, Phys. Rev. D 5, 2403 11972). "B. C. Xanthopoulos and T. Zannias, Phys. Rev. D 40, 2564
3 ~ I.. Janis, E. T. Newman, and J. Winicour, Phys. Rev. Lett. (1989).
20, 878 (1968). I2Y. Deng and P. D . Mannheim, Astrophys. J. 324, 1 (1988).
4J. E. Chase, Commun. Math. Phys. 19, 276 (1970). I 3 ~ hKaluza,
. Sitzungsber. Preuss. Akad. Wiss. Phys. Math. K1.
5R. Price, Phys. Rev. D 5, 2419 (1972). LIV, 966 (1921).
6C. Teitelboim, Phys. Rev. D 5, 2941 (1972). 140.Klein, Z. Phys. 37, 875 (1926).
'J. D . Bekenstein, Ann. Phys. (N.Y.)82, 535 (1974). I5v.A. Belinskii and I. M. Khalatnikov, Zh. Eksp. Teor. Fiz.
s ~ D.. Bekenstein, Ann. Phys. (N.Y.)91, 72 (1975). 63, 1121 (1972) [Sov. Phys. JETP 36, 591 (1973)].
'K. A. Bronnikov and Y . N. Kireyev, Phys. Lett. 67A, 95 I6v.Ferrari and B. C. Xanthopoulos (in preparation).
(1978).

You might also like