Signatures
of
black
holes
in
string theory
Metin Gurses*Department
of
Mathematics, Faculty ofScience, Bilkent Uniuersity, 06533 Ankara, Turkey(Received 20 April 1992)
The eftect ofstring theory on the four-dimensional classical Einstein equations is investigated. Itis shown that the throats ofnonrotating charged black holes are exact solutions ofthe gravitational field
equations with string correction terms.
PACSnumber(s): 11.17.
+
y,04.20.Jb, 04.60.+
nGravitational field equations with string correction
terms have recently drawn much attention in several
respects. The most important contribution
of
these terms is believed to change the singularity structureof
the spacetime geometry.For
this purpose there has been much interest in the black-hole solutions [1—10]of
the higher-dimensional Einstein field equations containing higher-order curvature terms. Although not yet known,itisbelieved that the low-energy string action isa pertur-bation expansion in inverse powers
of
the string tension parameter. This expansion contains, in addition to the usual Einstein-Hilbert action, corrections quadratic and higher-order invariants in the massless fields, curvature tensor, and in Maxwell and dilaton fields.Recently it has been shown that the plane-wave metrics
of
the Einstein theory preserve their form under string corrections at all orders[11
—13].
The question arises as to whether or not there exist sphericallysym-metric spacetimes with the same property. In particular,
nonrotating black-hole geometries are spherically sym-metric and it is believed that they do not preserve their form
[1].
Since gravitation is weak far from the holes, the
black-hole solutions
of
the Einstein theory can be considered as approximate solutionsof
the e6'ective field theory men-tioned above. On the other hand, near the singularities,the contribution
of
curvature terms in the extended theory becomes important. Black-hole solutionsof
the Einstein theory are not any more exact solutionsof
the extended theory in these regions. In the general case, itis very hard, almost impossible,to
consider the full extend-ed field equations and find their exact solutions with the property that they asymptotically approach the black-hole solutionsof
the Einstein theory. Recently[9]
it has been conjectured that metrics describing the neighbor-hoodof
the event horizonof
the extreme charged blackholes may solve the extended field equations exactly.
Such candidates are
of
extreme Reissner-Nordstrom type and the recently found metric with adilaton charge [8,9].
In this paper we shall show that the throatsof
these black-hole solutions are, in fact, exact solutionsof
the ex-tended field equations.Electronic address: Gurses at trbilun.
%e
assume a Batinternal space, an Abelian gauge fie)d with zero components in the internal directions, and set the three-form field equal to zero.%e
also assume that the four-dimensional metric, the MaxweH, and dilaton fields do not depend on the internal coordinates.%ith
these assumptions the four-dimensional low-energy
ac-tion obtained from string theory is
[9]
d4~
—
g—
R+2
g
2+g
2~+2+L(R;Jt,
t,F~,
p)],
where the Maxwell field
F;,
is associated with aU(1) sub-groupof
Es X Es orSpin(32)/Z2 and we set the remaining gauge fields to zero. P isthe dilaton field. The contribu-tionof
string theory to the classical gravitational actionis through the function
L.
It
is a perturbation expansion in inverse powersof
the string tension parameter. Theterms in this expansion may depend on all possible invari-ants constructed out
of
the curvature tensor R,.kt,Maxwell field
F,
, dilaton field iI), and their covariantderivatives. Under these assumptions, extremizing this
action with respect to the U(1)potential A,, dilaton field
P, and the metric, we obtain the four-dimensional low-energy limit
of
the string theory (extended gravitational field equations):6;
—
2T;=E;,
, P (e 2$Fii) —Fi
P'P+
2'e
~F=F.
(3) (4)The energy-momentum tensor T'J corresponds
to
a Maxwell field coupled to the dilaton (t[9,10].
Thesecond-rank symmetric tensor
E;,
the vector F.„and
sca-lar
E
coming from the variationof L
are the stringcorrection terms to the classical gravitationa1 field
equa-tions. They are believed to be composed
of
the curvature tensorR,
"ki, the Maxwell fieldF;,
the dilaton field P,andof
their covariant derivatives at all orders.Although we consider the low-energy limit
of
string theory, our discussion in this paper applies toany theory derivable from a variational principle where the Lagrang-ian is an arbitrary smooth functionof
the Riemannten-sor, Maxwell fie1d tensor, dilaton field, and
of
theirco-variant derivatives.
The metric
of
astatic and spherically symmetrictime is given by
ds
=
—
A dt+B
dr+C
(d8
+
sin Hdg),
(5)(AB)
„
P3 (22)
R'/,/
=
r'.
//,+ r'
/,r
/—
(k I),
R
j
RkjR
R
k The Riemann tensor correspondingto this metric in a compact form isgiven by
RgJkl glS]k gjkSil+gjkSJl gJ1SkJ+'92H&jHkl where
(6)
S/i
=
rtoM/i+
rtikik/+
,
rI3—g/1.
Here the scalars are given by
A
„C
—
AC,
AB
B
(7)
AB C,
C AB
where A,
B,
andC
depend only onr.
Our convention is as follows:The covariant derivatives
of H;.
andk;,
as seen in Eqs. (17)and (18), are expressed only by themselves and themetric tensor. Hence, any higher-order covariant deriva-tives
of
these tensors must obey this rule. Since in Eq. (6) the Riemann tensor is given in termsof
H;,
g,-,and k,-,and the scalars depending on r, its covariant derivatives
at any order obey the same rule. Hence, we have the fol-lowing theorem.
Theorem
1.
Covariant derivativesof
the Riemannten-sor R;Jkl, the tensor H;J, and the vector k; at any order
are expressible only interms
of
H;,
g,",
and k,.
Since the contraction
of
k'
with H,"
vanishes, the only symmetric tensors constructable outof
H;,
g;.
, and k;are
M;,
the metric tensorg;,
andk;k
.
Then the follow-ing theorems hold.Theorem
2.
Any second-rank symmetric tensorcon-structed out
of
the Riemann tensor, the antisymmetric tensorH;,
the dilaton field /t/=/t/(r), and their covariant derivatives is a linear combinationof
M;,
g;,
and k, k.
Let this symmetric tensor beE
.
Then we haveC4 A
„C
—
AC„
92 7r (10)E,
J=0
jM,J+
2g;J+ (23) A;=
cos(e)5'&,HJ=V;Aj
—
VjA;
.
(12) (13)A„C„
AB CThe tensor H; is antisymmetric and derivable from a
vector potential
A;:
E
=crk,
, (24)where 0.&, 0.2, and
03
are scalars which are functionsof
the metric functions, invariants constructed out
of
the curvature tensorR,
kl,H;,
and on the dilaton field.Theorem
3.
Any vector constructed outof
the Riemann tensor R;ik/, H;J, the dilaton field /t/=P(r), and their covariant derivatives is proportional tok;.
Let thisvector be
E,
'.
Hence, itreadsThe symmetric tensor M; isdefined as
k & 2
MJ=H;Hkj
—
—,H
glJ whereC„
c
' pi A„C+AC
„
2AB'C
H
=H'H
ijThe spacelike vector k;isgiven by k,
=V;r
.
The covariant derivatives
of H;.
and k; are given asH,
,
=p(
2k, H,,
+k,
H, ,—
k,H//)~—
Vikj Plgij+P2
ij+P3
ij
where (14) (15) (16) (17) (18) (20)where 0. isascalar as0&,0.2,and 0.3.
We first discuss the solutions
of
the Einstein fieldequa-tions. The Einstein tensor is found as
6
/'=
(2'/I0+
'i/~)M
/2B C
4(6B
Ci)3+B
gq—
4C
gati)g//.+2'/k/kj
. (25) In the Einstein theory, the formof
(25) gives us an idea about the formof
the energy-momentum tensor as the source for the field equations. The source may include the electromagnetic fieldF;,
a dilaton field P,and possi-bly acosmological constant. In this case, the electromag-netic field may have an electric part in addition to the magnetic part, that is,F;
-may have the formF
1=e(r)H
1+qoH,
(26)where H, is the dual
of
the tensor H;.
In the generalcase, we have five equations for five functions A,
B,
C,P(r),
ande(r).
In the spherically symmetric case, the function e(r)
is not independent. Equation (3) forces us to choose itasA,
C
—
AC,
P2C
AB2 (21)e(r)=e
e~,
AB 0 (27)g)
=F3=0,
(28) 2 g2 gQ+CQ (29) 4 CQ AB A„
B
(30)Since the dilaton field P is set to zero, the vector k,.does
not show up in the set
of
tensors discussed in the above theorems. Hence, the tensorE;-
is a linear combinationof
M, and g; . This means thatcr3=0
in (2). The vectorE,
or o in Eq. (24) vanishes identically. Therefore, we have three equations for two functions A,B.
This is again an overdetermined system. We shall further set gQ= const. Under these assumptions, the extended field equations reduce to2
2=
~ Ilo+co
—
2qo=ai('tlo,
co,qo;a
),
1 2 ~ I
4(rjo co)=cT2('go)co,go',
a
),
2C0
(31)
(32) where eQ is a constant. The remaining four functions A,
B,
C, and P can be solved consistently. In fact, there are various typesof
solutions reported so far (see, forin-stance,
[9]).
In the presence
of
string correction terms in Eqs.(2)—(4), the inclusion
of
the electric part is problematic.The dual H,
"
introduces a timelike vector u,=5,
'- into thetensor algebra discussed in theorem
1.
This increases the numberof
symmetric tensors to five and numberof
vec-tors to two. Comparing the coefficients
of
these tensors and vectors in Eqs. (2) and (3), respectively, we obtain the equations for the metric functions A,B,
C,e(r),
and the dilaton fieldP=P(r)
T.he total numberof
equations is raised to eight for these functions. In the Einstein case,since the right-hand sides
of
Eqs. (2)—(4)are absent, one gets a correct numberof
equations for these functions. In the caseof
the extended field equations it seems unlikely that this overdetermined systemof
ordinary differential equations has a solution.In order to overcome this difhculty, we first let
e(r)
=0
andF, =qQH;,
where qQ denotes the magnetic charge.In this case, the tensor
E;
and the vectorE,
have exactly the same forms asE
andE,
respectively. Therefore, in the sequel we shall drop the primes over these tensors.By the theorems given above, we have now five equations
for four functions. Again, the existence
of
a solutionof
this system
of
differential equations is not guaranteed.Hence, we further choose the electromagnetic field in such a way that the right-hand side
of
Eq. (3) automati-cally vanishes. This can be achieved in two ways: eitherF;
can be set equal to zero orF,
J and P can be takenco-variantly constants. In the first case the Einstein field equations couple to ascalar field
[1].
We have four equa-tions and four unknown functions. Hence, we have a well-defined setof
ordinary differential equations which may have exact solutions.When the electromagnetic field is covariantly constant and
/=0,
Eq. (19)leads to C=co=
const. This simplifies the curvature functions gQ,g„g2,
and g3. They read2
qp
2
=E(rlo,
co qo,a'),
CQ
(33)
where o.
"
is the inverse string parameter. Here we have three algebraic equations for three constants CQ, g0, and the constant q0. The scalars 0.&,o.2, andE
are now alsoconstants depending upon cQ, qQ, and qQ, and on the
pa-rameters
of
the theory under consideration, such as the string tension. Since rjo is a constant, Eq. (30)is exactly soluble.It
readsA,
B=
'1/ (qo/co)A
+a,
(34)
where a& is a constant. This is the only equation which
determines the metric functions A and
B.
Although the function A appears to be arbitrary, by changing r,B
can be set to any functionof r
We c.hoose it as1/A.
Withthis choice
of
B,
the above differential equation can be in-tegrated easily and we determine the spacetime metric asds
=
—
A dt+
1 dr+co(d8
+
sin Odg),
(35)where
90
A
=r
r+a&
+a2
.
CQ
(36)
Here a& and a2 are constants. According to whether g0
iszero ornot there are two distinct solutions. Type (a):
Q=ai=0
(37)It
is a direct productof
S
and a two-dimensional Hat Minkowski space. The vector k, is covariantly constant;hence, itisalso a spacelike Killing vector.
Type (b):
t0
a2=0,
A=r
r+a&
CQ
(38)
This metric describes a spacetime which is adirect
prod-uct
of
a two-dimensional pseudosphere andS
.We now state that the metrics given above are exact
solutions
of
the gravitational field equations with stringcorrections. They are all nonsingular and homogeneous spacetimes. These metrics are also the solutions
of
the Einstein-Maxwell equations.For
instance, the second solution [type (b)]with'g
=C
=q
is the Levi-Civita Bertotti-Robinson metric. Under the string corrections the form
of
these metrics is preserved but the parameters appearing in the metrics no longer satisfy the above equations (39). In this case the relations among these constants are given by Eqs. (31)—
(33).The above metrics, in particular type (b),have interest-ing features. They describe the geometry
of
black-hole solutions in the neighborhoodof
their outer horizons.These regions are the throats
of
the Einstein-Rosen bridgesof
the corresponding hole solutions.For
exam-ds
= —
W dt+
1 dr+r
(d8
+
sin8dg
),
8
where (40) W=1
—
2m+
Q—
A,or p2 (41)Here m, Q, and A,p are, respectively, the mass, charge,
and the cosmological constant. In the neighborhood
of
the outer horizonr=r&+e,
whereW(r=rz)=0,
themetric and the function
8'take
the form ds= —
W dt+
de
+r
(d8
+
sin8dp
)1
h (42)
W
=e(ae+P)
.
(43)This metric remains to be the solution
of
the Einstein-Maxwell field equations with a cosmological constant providedand
a,
P,
and A,pare given bya=
(2Q—
rt,),
1 6 (45) (46) Ap= 2 (rq2Q)
.
—
(47)Notice that the cosmological constant is no longer an in-dependent parameter. This limiting metric (42)isexactly
of
type (b)given above with the identifications90
=a,
a,
=P,
Co
(48)
cp rh~ gp Q (49)
The extended field equations (31)
—
(33) alter the definitions given in Eqs. (44)—(47)of
a,
Q, rz, and A,p. Bythe utility
of
this identification they now satisfyrza+r&
—
2Q=cr,
(a,
r&,Q;a'),
(50)1
(rz
a
—
1)=
oz(a,
rz,Q;
a'),
276 (51)
pie, the Levi-Civita Bertotti-Robinson universe is known
to
be the throatof
the extreme charged black hole[14—
16].
Here, extending the same idea let us find themetric describing the region near the outer horizons
of
the charged black holes in de Sitter space
[17].
In the spherically symmetric case such a black-hole solution is given by2
2 4
=E(a,
r&,Q;a') .
Tg
(52)
Hence, the above equations relate the mass and the charge to the string tension. The type (a) solution corre-sponds tothe throat
of
the extreme charged hole with di-laton discussed recently in[9].
We conclude that the regions in the neighborhood
of
the outer horizons
of
the charged black holes are preserved under string corrections. String theory only changes the relations among the parametersa,
Q,rz, and the cosmological constant A,o. They are related to the in-verse string tension parametera'
through Eqs.(50)—(52).We have found two distinct metrics which may consti-tute exact solutions
of
the gravitational field equations with string correction terms at all orders. The existenceof
solutions,of
course, depends on the functional depen-denciesof
0&,0.2, andE
on the parameters co, go, qo oron rz,
a,
and Q. Since the extended field equations or the explicit formsof
E;,
E;,
andE
are not known yet, it isnot possible to give an answer to this existence problem. On the other hand, the gravitational field equations with the quadratic curvature terms, such as the Gauss-Bonnet term, are commonly known to be the one-loop correc-tions. The tensors
E;,
E;,
andE
in this case are simpler and Eqs. (31)—
(33) have consistent solutions.We have shown that the metrics that solve the extend-ed field equations are asymptotic forms
of
the charged nonrotating black holesof
the Einstein theory. The mass and the charge are related to string tension. When we consider only the first-order corrections, the tensorE,
"
is the Gauss-Bonnet term which vanishes identically forboth types
of
metrics we found in this paper. The effectof
string corrections at this order comes fromEq.
(33).When the cosmological constant is set to zero, the mass, charge, and radius
of
the horizon turn outto
be equal and they are proportional to&a'.
Hence, at the Planckscale we have an exact solution
of
the extended field equations which is the signatureof
a classical black hole.It
isexactly the throatof
the extreme charged black hole. Inclusionof
the three-form field into the field equations (2)—
(4) isalso possible. This will not alter our conclusion.Let H;Jk=pk~;Hjk~, where
p
is a functionof
r. One canprove the following theorem: The only antisymmetric second-rank tensor obtainable out
of
Rijkl H;j k H&jk,and their covariant derivatives is proportional to
H;. .
Hence, in addition to the field equations (3) and (4), we have V';(e t'H'~")=
a4HJ", where o4isascalar as o„o
2,o3.
Letting H; be covariantly c.onstant,P=Pp=
constand
p=0,
we have (31)—(33)witho's
also depending onIn addition to these equations we also have
cT4('gp cp gp Pp'a')
=0.
This leads to four algebraicequa-tions forfour constants rtp cp gp and Pp.
Extension
of
the results reported in this paper tohigher dimensions is also possible.
For
instance, aD-dimensional spacetime, which is a direct product
of
a two-dimensional pseudosphere andS,
isa solutionof
Einstein-Maxwell field equations in D dimensions. The
Maxwell field is covariantly constant.
It
is possible toas given in Eq. (6)with slight changes in the definitions
of
7fp 'g~ 'g2 and g3. The theorems given in this work
remain valid. The form
of
the metric ispreserved under string corrections. The extended field equations are again redefinitionsof
the parameters appearing in the theory.I
would like to thank Rahmi Guven for a critical read-ingof
the manuscript and for several suggestions andYa-vuz Nutku for discussions. This work was partially sup-ported by the Scientific and Technical Research Council
of
Turkey(TUBITAK).
[1]C.G.Callan,
R.
C.Myers, and M.J.
Perry, Nucl. Phys.B311,673(1988).
[2]D. G. Boulware and S.Deser, Phys. Rev. Lett. 55, 2656 (1985).
[3]
J.
T.Wheeler, Nucl. Phys. B268,737(1986). [4]J.
T.Wheeler, Nucl. Phys. B273,732(1986). [5]D. L.Wiltshire, Phys. Lett. 169B,36(1986). [6]R.
C.Myers, Nucl. Phys. B289,701(1987). [7]R.C.Myers, Phys. Rev. D 36,392(1987).[8]G. W. Gibbons and
K.
Maeda, Nucl. Phys. B298, 741 (1988).[9]D. Garfinkle, G.
T.
Horowitz, and A. Strominger, Phys.Rev.D 43,31&0(1991).
[10]N.
R.
Stewart, Class.Quantum Grav. 8, 1701(1991).[11)R.Giiven, Phys. Lett.B191,275(1987).
[12]D.Amati and C.Klimcik, Phys. Lett. B219,443(1989). [13]G. Horowitz and A. R. Steif, Phys. Rev. Lett. 64, 260
(1990).
[14]
B.
Carter, in Black Holes, edited by C. DeWitt and B.DeWitt (Gordon and Breach Science, New York, 1973). [15] C.Misner, K.S.Thorne, and
J.
A. Wheeler, Gravitation(Freeman, San Francisco, 1973).
[16]D. L.Wiltshire, Phys. Rev.D31,2445(1988). [17]P.C.Davies, Report No. NCL-89 TP9(unpublished).