a

r
X
i
v
:
h
e
p
-
t
h
/
9
5
1
1
2
2
3
v
2


2
9

D
e
c

1
9
9
6
FOUR-DIMENSIONAL HIGHER-DERIVATIVE SUPERGRAVITY
AND SPONTANEOUS SUPERSYMMETRY BREAKING
AHMED HINDAWI, BURT A. OVRUT, AND DANIEL WALDRAM
Department of Physics, University of Pennsylvania
Philadelphia, PA 19104-6396, USA
Abstract. We construct two classes of higher-derivative supergravity theories generalizing
Einstein supergravity. We explore their dynamical content as well as their vacuum structure.
The rst class is found to be equivalent to Einstein supergravity coupled to a single chiral
supereld. It has a unique stable vacuum solution except in a special case, when it becomes
identical to a simple no-scale theory. The second class is found to be equivalent to Einstein
supergravity coupled to two chiral superelds and has a richer vacuum structure. It is
demonstrated that theories of the second class can possess a stable vacuum with vanishing
cosmological constant that spontaneously breaks supersymmetry. We present an explicit
example of this phenomenon and compare the result with the Polonyi model.
PACS numbers: 04.50.+h, 04.65.+e, 11.30.Qc
1. Introduction
In a recent series of papers [1-3], we studied the general structure of higher-derivative
gravitational theories, both for bosonic gravity and for two-dimensional (1, 1) supergrav-
ity. Although a wide range of technical phenomena, such as consistent coupling of spin-2
elds to gravity and the method of super-Legendre transformations, were explored, the main
thesis of these papers lay elsewhere. We systematically showed that both bosonic and super-
gravitational higher-derivative theories exhibit a rich structure of non-trivial vacua. These
vacua can be expressed as non-vanishing vacuum expectation values of the scalar elds that
generically arise as new degrees of freedom in such theories. While these vacua always have
non-vanishing cosmological constant in pure bosonic gravity and, hence, are of restricted in-
terest in particle physics, we showed that non-trivial vacua with zero cosmological constant
do arise in two-dimensional supergravity. Furthermore, these vacua spontaneously break
supersymmetry [3]. It follows that higher-derivative supergravity theories can potentially
play an important role in particle physics, including introducing what is essentially a new
method of supersymmetry breaking. To realize this goal, however, it is essential to extend
our results to four-dimensional, N = 1 supergravity. This will be the content of this paper.
In Section 2, we will briey present generic results in D = 4, N = 1 supergravity that we
will need later in the paper. Section 3 is devoted to a short exposition of the methods and
Published in Nuclear Physics B 476 (1996), 175199.
1
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 2
results in purely bosonic higher-derivative gravitation that we obtained in [1, 2]. This sets the
stage for the extension to four-dimensional, N = 1 supergravity. We present a simple higher-
derivative, D = 4, N = 1 supergravity theory in Section 4 and explore its vacuum structure.
We show that a subset of such theories is equivalent to no-scale supergravity with a single
chiral supermultiplet [4]. Finally, in Section 5 we give the generic four-dimensional higher-
derivative, N = 1 supergravity theory involving the scalar curvature supereld R only. We
show that these theories contain non-trivial vacuum states, with vacuum expectation values
of the order of the Planck mass M
P
, that have vanishing cosmological constant and that
spontaneously break supersymmetry. Furthermore, all elds around these vacua propagate
physically; that is, there are no ghosts. Finally, we then show that in the limit of large M
P
,
these theories become equivalent to a generalized type of Polonyi model [5].
These results open the possibility that in phenomenological supergravity models and,
perhaps, in superstring theories [6], supersymmetry is spontaneously broken not by a hidden
sector or by gaugino condensates, for example, but simply by the new degrees of freedom
that arise in higher-derivative terms in the super-eective Lagrangian.
2. D = 4, N = 1 supergravity
In this section we give a brief review of four-dimensional N = 1 supergravity from the
supereld point of view. We will discuss both pure supergravitation and supergravity cou-
pled to chiral matter. This is done both to set our notation and to present some of the
explicit formulas we will need later in the paper. Here and elsewhere we follow the supereld
formulation of supergravity presented in [7].
Consider a supermanifold with coordinates z
M
= (x
m
,

) where x
m
are the ordinary
spacetime commuting coordinates while

and

are fermionic anti-commuting coordinates.
Hereafter, Einstein indices that transform under coordinate transformations are denoted by
(M, N, . . . ), whereas Lorentz indices that transform under the structure group are denoted
by (A, B, . . . ).
The geometry of the superspace is determined by the supervielbein E
M
A
and the Lie
algebra valued connection one-form
MA
B
. The torsion is dened as the covariant derivative
of the supervielbein
T
A
= dE
A
+E
B

B
A
=
1
2
dz
M
dz
N
T
NM
A
. (2.1)
The curvature tensor is a Lie algebra valued 2-form dened in terms of the connection by
R = d + . (2.2)
The number of components in the torsion and curvature is very large and a set of con-
straints is required to reduce it. The old minimal supergravity theory is dened by applying
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 3
the following constraints
T

= 0, T

c
= T

c
= 0,
T

c
= T

c
= 2i

c
,
T
b
c
= T
a
c
= 0,
T
ab
c
= 0,
(2.3)
where denotes either or . The torsion and curvature satisfy the Bianchi identities
DDE
A
= E
B
R
B
A
,
DT
A
= E
B
R
B
A
.
(2.4)
One has to solve these identities subject to the constraints (2.3) to nd the reduced set of
elds. We merely state the results here. It turns out that all supereld components of the
torsion and curvature can be expressed in terms of three superelds, a chiral supereld R,
an hermitian vector supereld G

and a chiral supereld W

totally symmetric in its

indices. The component elds of the reduced supergravity multiplet are found to be the
graviton e
m
a
(x), the gravitino
m

(x) and two additional elds, M(x), a complex scalar,

and a real vector eld b
m
(x).
Beside the supergravity multiplet, we need to introduce matter elds. We will restrict our
discussion to chiral superelds which are superelds satisfying the covariant condition

D

= 0. (2.5)
Chiral superelds contain three component elds. The expansion of chiral superelds in
terms of and

is complicated and coordinate-dependent, since and

carry Einstein
indices. It is better to dene the component elds of a chiral supereld by
A = |
=

=0
,

=
1

2
D

|
=

=0
,
F =
1
4
D

|
=

=0
.
(2.6)
These components carry Lorentz indices. New fermionic coordinates

are dened such

that the expansion coecients of chiral superelds are precisely the covariant components
A,

, and F. That is
= A(x) +

(x) +

F(x). (2.7)
These are referred to as chiral superspace coordinates.
We are now in a position to write down the pure supergravity theory as well as theories
of supergravity coupled to chiral matter. The Einstein supergravity Lagrangian is given by
L =
3

2
_
d
4
E, (2.8)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 4
where E is the super-determinant of the supervielbein and
2
= 8G
N
is the gravitational
coupling constant which we will set equal to unity unless stated otherwise. This Lagrangian
can be written as an integral over chiral superspace as
L = 6
_
d
2
ER + h.c., (2.9)
where E and R have the following expansions
E =
1
2
e
_
1 +i
a

a
[M

a

ab

b
]
_
,
R =
1
6
_
M +
_

a

b

ab
i
a

a
M +i
b
b
b

+
_

1
2
R +i

a

b

ab
+
2
3
MM

+
1
3
b
a
b
a
ie
a
m
D
m
b
a
+
1
2

M
1
2

c
b
c
+
1
8

abcd
_

a

b

cd
+
a

cd
_
_
.
(2.10)
Here e is the determinant of the vielbein e
m
a
, R is the spacetime curvature scalar, and

mn

=

D
m

D
n

, (2.11)
where

D
n

=
n

+
m

,
n

(x) =
n

(z)|
=

=0
. (2.12)
The component eld Lagrangian for Einstein supergravity can be obtained by substituting
(2.10) into (2.9) and performing the -integral. The result is
e
1
L =
1
2
R
1
3
MM

+
1
3
b
m
b
m
+
1
2

kmn
_

D
m

D
m

n
_
. (2.13)
Clearly M and b
m
are auxiliary elds and can be eliminated from the Lagrangian using their
equations of motion. These are given by
M = 0,
b
m
= 0,
(2.14)
respectively. The remaining bosonic part of the Lagrangian is just Einstein gravity which
describes the propagation of a massless graviton. The fermionic part is a kinetic-energy term
for the massless gravitino
m
.
We now turn our attention to the coupling of supergravity to chiral matter superelds
i
.
This coupling is determined by two functions, the hermitian Kahler potential K(
i
,

i
) and
the holomorphic superpotential W(
i
). The Lagrangian of Einstein supergravity coupled to
the chiral elds
i
is given by
L =
_
d
4
E
_
3e

1
3
K(
i
,

i
)
+
W
2R
+
W

2R

_
. (2.15)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 5
This is an integral over the full superspace. It can be written as an integral over chiral
superspace of the form
L =
1
4
_
d
2
E(

D
2
8R)
_
3e

1
3
K
_

1
4
_
d
2
E(

D
2
8R)
_
W
2R
+
W

2R

_
+ h.c. (2.16)
Simplication can be achieved using the following fact [8]
_
d
2
E(

D
2
8R)
_
L L

_
+ h.c. = total divergence, (2.17)
which holds for an arbitrary supereld L. Hence, a spacetime integral of this expression can
be dropped from any action because it does not contribute to the eld equations. Applying
this to the superpotential term in (2.16) yields
L =
1
4
_
d
2
E(

D
2
8R)
_
3e

1
3
K
_
+ 2
_
d
2
EW + h.c., (2.18)
where we have used the fact that W is chiral.
Using the -expansion of E, R, and
i
, we can evaluate the component eld Lagrangian.
One can show that M, b
m
and F
i
are auxiliary; that is, they are not propagating elds.
Their equations of motion are purely algebraic and can be used to eliminate them from the
Lagrangian. After doing so, we get a Lagrangian for the propagating elds e
m
a
,
m
, A
i
,
and
i

which has a non-canonical coupling of both the scalar curvature R and the gravitino
kinetic-energy term to the matter elds. This non-canonical coupling can be transformed
away by Weyl rescaling the vielbein and making a eld-dependent redenition of the spinors
in an appropriate way. Performing these transformation, we get a canonically normalized
but complicated component eld Lagrangian. We will not reproduce this entire Lagrangian
here, but rather will refer the reader to [7]. Of special importance in this paper is the bosonic
part of the Lagrangian. It is found to be
e
1
L
Bosonic
=
1
2
R +e
1
L
KE-Bosonic
V (A
i
, A

i
), (2.19)
where the kinetic energy Lagrangian for the matter elds L
KE-Bosonic
is given by
e
1
L
KE-Bosonic
= K
ij

m
A
i

m
A
j
(2.20)
and the scalar potential V (A
i
, A

i
) is given by
V (A
i
, A

i
) = e
K
_
K
ij

D
i
W(D
j
W)

3|W|
2
_
, (2.21)
where
D
i
W = W
i
+K
i
W,
K
ij

= K
1
ij
.
(2.22)
Here subscripts denote derivatives with respect to the matter elds, so that, for instance,
K
ij
=
2
K/A
i
A

j
. We will be particularly concerned with local minima of (2.21) with
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 6
vanishing cosmological constant. Supersymmetry is spontaneously broken at such a mini-
mum if and only if D
i
W = 0 for some i. Under these conditions one can show that the
gravitino mass is given by
m
2
3/2
=

e
K
|W|
2
_
. (2.23)
The gravitino mass is non-vanishing if and only if supersymmetry is spontaneously broken.
3. Higher-Derivative Bosonic Gravitation
Einstein bosonic gravitation is specied by the Lagrangian
L =
1
2

g R, (3.1)
where g = det(g
mn
) and R is the spacetime curvature scalar. Although R contains second
derivatives of the metric tensor g
mn
, the Einstein eld equations are nonetheless second-order
dierential equations. Perhaps the simplest modication of Einstein theory is to include
higher powers of the curvature scalar R in the Lagrangian. Once we include any power of R
greater than unity, the equations of motion become fourth-order dierential equations. Since
more initial conditions are required to solve the Cauchy problem, one must conclude that the
higher-derivative theory has more degrees of freedom than the second-order Einstein theory.
A general class of higher-derivative bosonic gravity theories involving the scalar curvature
only is specied by
L =
1
2

g f(R), (3.2)
where f is an arbitrary real function. These theories do indeed have more degrees of freedom
than the simple Einstein theory (3.1). The dynamical content of Einsteins theory is a single
massless spin-2 eld, the graviton. The higher-derivative theory (3.2) contains, beside the
graviton, one real scalar degree of freedom. This degree of freedom can be made explicit using
the method of Legendre transformations. Here we will follow the discussion of [2], though
such transformations were originally discussed in [9-12]. This method reduces the higher-
derivative theory to an equivalent second-order form. The eld equations of the second-order
theory are second-order dierential equations, where the extra degree of freedom is explicitly
represented by a new eld variable. To apply this procedure to the higher- derivative theory
(3.2), we introduce a real scalar eld X into the Lagrangian
L =
1
2

g
_
f

(X)(R X) +f(X)
_
, (3.3)
where f

(X) = df/dX. The equation of motion of the auxiliary eld X is

f

(X)(R X) = 0. (3.4)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 7
Provided that f

(X) = 0, this gives X = R. Substituting back into the Lagrangian (3.3)

yields the original higher-derivative form (3.2). The eld redenition
= ln f

(X), (3.5)
puts Lagrangian (3.3) into the form
L =
1
2

g
_
e

R +f(X(e

)) e

X(e

)
_
. (3.6)
To remove the non-canonical factor of e

multiplying R, we make the conformal transforma-

tion
g
mn
= e

g
mn
. (3.7)
Under such a transformation, Lagrangian (3.6) takes the canonical form
L =

g
_

1
2

R
3
4
_

_
2

1
2
e
2
_
f(X(e

)) e

X(e

)
_
_
. (3.8)
This Lagrangian describes Einstein gravity coupled to a physically propagating real scalar
eld with a specic potential energy dependent on the function f(R). Note that scalar
eld has a non-ghost-like kinetic energy.
We have studied the vacuum structure of a wide class of higher-derivative bosonic gravita-
tional theories in previous papers [1, 2]. We found that they exhibit an interesting vacuum
structure, but these vacua generically have non-vanishing cosmological constant. We ex-
tended our investigations to supergravitation in [3], where we studied the vacuum structure
of higher-derivative, D = 2, N = (1, 1) supergravitation. We found that these theories
possess an even richer vacuum structure and, more importantly, that non-trivial vacua can
occur with vanishing cosmological constant. This makes higher-derivative supergravity the-
ories more important for particle physics than their bosonic counterparts. In the present
paper we would like to extend our study to D = 4, N = 1 higher-derivative supergravity
theories.
4. A Simple Generalization of Einstein Supergravity
As stated in Section 2, Einstein supergravity is specied by the Lagrangian
L = 3
_
d
4
E. (4.1)
We would like to generalize this Lagrangian to include higher-derivative terms. A simple
generalization would be to consider
L = 3
_
d
4
E
_
f(R) +f(R

)
_
, (4.2)
where f is a real function of its argument, and R is the chiral scalar curvature supereld.
Note that the apparently more general Lagrangian of the form f + f

, where f is now an
analytic function of R, in fact reduces to (4.2) by virtue of the indentitiy (2.17). We will
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 8
study action (4.2) in this section and will consider other generalizations in the following
section.
What is the dynamical content of Lagrangian (4.2)? We need to compute the component
eld Lagrangian to answer this question. We start by writing (4.2) in chiral superspace as
L =
3
4
_
d
2
E(

D
2
8R)
_
f(R) +f(R

)
_
+ h.c.
=
3
2
_
d
2
E(

D
2
8R)f(R) + h.c., (4.3)
where we have used the identity (2.17). Since R is chiral, so is f(R) and, hence,

D
2
f(R) = 0.
Lagrangian (4.3) can then be written as
L = 12
_
d
2
EF(R) + h.c., (4.4)
where F(R) = Rf(R). It is more convenient to carry out the following analysis in terms of
the function F(R).
The full component-eld Lagrangian can be evaluated using the -expansions of E and
R given by (2.10). However, this component Lagrangian is quite involved. For the present
purpose, it will suce to consider only the bosonic part of the Lagrangian. In this case, the
-expansions of E and R are given by
E =
1
2
e
_
1 M

_
,
R =
1
6
_
M +
_

1
2
R +
2
3
MM

+
1
3
b
m
b
m
i
m
b
m

_
.
(4.5)
Consequently we have
F(R) = F(
1
6
M)
1
6
F

(
1
6
M)
_

1
2
R +
2
3
MM

+
1
3
b
m
b
m
i
m
b
m

. (4.6)
Substituting (4.5) and (4.6) into (4.4) and performing the integral yields
e
1
L
Bosonic
=
1
2
(F

+F

)R +
1
3
b
m
b
m
(F

+F

) i
m
b
m
(F

)
+
2
3
MM

(F

+F

) + 6(M

F +MF

), (4.7)
where, for simplicity, we denote F(
1
6
M) by F and F(
1
6
M

) by F

. There are two ways

to think about the dynamical content of this Lagrangian. The rst is to integrate the term
involving
m
b
m
by parts. Doing this clearly makes b
m
an auxiliary eld. After making the
integration by parts, varying the Lagrangian with respect to b
m
yields
b
m
=
3i
2

m
(F

)
F

+F

. (4.8)
Substituting this expression back into the Lagrangian gives
e
1
L
Bosonic
=
1
2
(F

+F

)R +
3
4
(F

+F

)
1
[
m
(F

)]
2
+
2
3
MM

(F

+F

) + 6(M

F +MF

). (4.9)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 9
We would like to remove the non-canonical coupling of R to the complex eld M. To do
this, we perform the following conformal transformation
g
mn
= (F

+F

)g
mn
. (4.10)
Under such a transformation, Lagrangian (4.9) takes the form
e
1
L
Bosonic
=
1
2

R
3
4
(F

+F

)
2
[
m
(F

+F

)]
2
+
3
4
(F

+F

)
2
[
m
(F

)]
2
+
2
3
MM

(F

+F

)
1
+ 6(F

+F

)
2
(M

F +MF

), (4.11)
or equivalently
e
1
L
Bosonic
=
1
2

R K
MM

m
M
m
M

V (M, M

), (4.12)
where K
MM
=
2
K/MM

, K is a real function of M and M

given by
K = 3 ln
_
F

+F

_
(4.13)
and V is the potential energy given by
V (M, M

) =
12
(F

+F

)
2
_

1
2
M

F
1
18
MM

+ c.c
_
. (4.14)
The bosonic structure of the generalized supergravity Lagrangian (4.4) is now clear. It follows
from (4.12) that it describes Einstein gravity coupled to a complex scalar eld with a specic
non-linear sigma model kinetic energy and self-interactions. It is interesting to note that this
simple generalization of supergravity, instead of adding higher-derivative gravitational terms
such as R
2
as discussed in the previous section, promotes the previously auxiliary eld M to
a propagating eld, thus adding two real bosonic degrees of freedom to the theory. This is
the simplest example of a generic phenomenon that occurs in higher-derivative supergravity,
and will be discussed in more detail in the next section. This phenomenon was rst observed
at the linearized level in a component eld construction of higher-derivative supergravity by
Ferrara, Grisaru and van Nieuwenhuizen [13].
Another way to analyze Lagrangian (4.7) is not to perform the integration by parts. Then
the Lagrangian does not involve any derivatives of M and, hence, M is an auxiliary eld. We
can then eliminate M using its equation of motion. Unfortunately, this equation of motion
cannot be solved in closed form. Nevertheless, we can see that M will be a function of the
curvature scalar R and
m
b
m
. When substituting back into the Lagrangian (4.7), we get
a non-trivial function of the scalar curvature R and
m
b
m
. The non-triviality in R means
this is a higher-derivative Lagrangian for the metric tensor. These higher-derivative terms
describe, according to the analysis of Section 3, the usual graviton plus one real propagating
scalar degree of freedom. The
m
b
m
terms imply that the longitudinal mode of b
m
, once
an auxiliary eld, is now propagating. We again conclude that Lagrangian (4.7) describes
Einstein gravity coupled to two real degrees of freedom, in agreement with the above analysis.
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 10
Since the theory is supersymmetric, there should be superpartners for the two new bosonic
degrees of freedom. This may seem odd at rst, because the supergravity multiplet does not
contain any auxiliary fermionic elds to start propagating along with the bosonic degrees
of freedom. But having auxiliary elds that acquire kinetic-energy terms upon generalizing
the Lagrangian is only one way of getting new degrees of freedom. Another way is to have
higher-derivatives acting on the propagating elds. In Einstein supergravity the gravitino
kinetic energy is given by
e
1
L
KE-Fermionic
=
1
2

kmn
(

D
m

D
m

n
), (4.15)
as can be seen from (2.13). The associated equation of motion of
m
is rst-order in
derivatives, as it should be for a fermionic degree of freedom. Now note that the middle
component of supereld R contains a rst derivative of the gravitino. This can be seen
from the -expansion of R given in (2.10). Hence, R
n
, for any integer n greater than unity,
will contain in its highest component terms quadratic in the rst derivative of the gravitino

m
. Such terms are generically contained in Lagrangian (4.4). Thus the component eld
Lagrangian contains higher powers of the rst derivative of the gravitino. Therefore, the
fermionic equation of motion will be second-order, requiring additional initial conditions to
solve the Cauchy problem. Hence, there will be more fermionic degrees of freedom in the
theory described by (4.4) than just a gravitino. In fact, as far as the fermionic degrees of
freedom are concerned, Lagrangian (4.4) always describes a higher-derivative theory. These
additional fermionic degrees of freedom act as the superpartners for the two new bosonic
degrees of freedom discussed above.
It is clear that analyzing the theory specied by (4.4) is by no means simple. In analogy
with the analysis of the bosonic theory (3.2), we would like to make the extra degrees
of freedom explicit. Therefore, we would like a supersymmetric analog of the Legendre
transform method. In the bosonic case we need to introduce one real eld to account for
the new degree of freedom. Here we have two real scalar degrees of freedom, or equivalently
one complex scalar, along with their fermionic partners. These degrees of freedom can only
arrange themselves into a chiral supermultiplet. Therefore, we introduce a chiral supereld
and try to rewrite Lagrangian (4.4) in an equivalent second-order form in which the new
propagating degrees of freedom are contained in . In analogy to the bosonic case (3.3),
consider a new Lagrangian
L = 12
_
d
2
E
_
F

()(R ) +F()
_
+ h.c. (4.16)
The equation of motion of the supereld is
F

()( R) = 0. (4.17)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 11
Therefore, provided that F

() = 0, this gives
= R. (4.18)
Substituting (4.18) back into (4.16) yields the original Lagrangian (4.4). Lagrangian (4.16)
can be written as
L = 12
_
d
2
E
_
RF

()
_
F

() F()

_
+ h.c. (4.19)
or, equivalently
L =
3
4
_
d
2
E(

D
2
8R)
_
F

() +F

)
_
+ 12
_
d
2
E
_
F

() F()
_
+ h.c., (4.20)
where we have used the identity (2.17) and the fact that F(R) is chiral. Comparing this
form with (2.16), we nd that this is the Lagrangian of Einstein supergravity coupled to a
single chiral supereld with a Kahler potential given by
K(,

) = 3 ln
_
F

() +F

)
_
, (4.21)
and a superpotential given by
W() = 6
_
F

() F()
_
. (4.22)
In this equivalent formulation of the theory, we can explicitly demonstrate the new propa-
gating degrees of freedom. Recall that in chiral superspace
= A +

F. (4.23)
From the equation of motion of , (4.18), we see, using (2.10), that the new propagating
component-eld degrees of freedom are
A =
1
6
M,

2
12
_

m

n
(

D
m

D
n

m
) i
m

m
M +i
m
b
m
_

.
(4.24)
The scalar bosonic degree of freedom is the complex scalar eld M, and the fermionic partner,
although a complicated combination of
m
, M, and b
m
, is essentially the rst derivative of
the gravitino
m
. This is in accord with the component eld discussion presented earlier.
We now turn our attention to the vacuum structure of the theory. To do this, we need
to evaluate the scalar eld potential energy using the Kahler potential and superpotential
given in (4.21) and (4.22) respectively. A straightforward calculation using (2.21) yields
V (A, A

) =
12
_
F

(A) +F

(A

)
_
2
_
3A

F(A) 2AA

(A) + c.c.
_
. (4.25)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 12
Using (4.24), we see that this potential is exactly the same as the potential for the eld M
in (4.14), as it must be. The kinetic energy for A can be obtained by substituting Kahler
potential (4.21) into (2.20) which gives
e
1
L
KE-Bosonic
=
3
_
F

(A) +F

(A

)
_
2

m
A
m
A

. (4.26)
Again, noting that A =
1
6
M, we see that this is exactly the kinetic energy term for M in
(4.12).
An important question is whether or not the potential (4.25) has stable vacua, denoted
by A, with zero cosmological constant. By a vacuum we will mean a local minimum of the
potential. The cases A = 0 and A = 0 require separate treatment.
Case I: A = 0
It is clear from (4.25) that V (0, 0) = 0, so the cosmological constant vanishes. In order for
A = 0 to be a stationary point of the potential, the rst-order derivatives of the potential
evaluated at A = 0, given by
_
V
A
_
=
9F(0)
F

(0)
2
, (4.27)
should vanish. It follows that we must have
F(0) = 0. (4.28)
To be a local minimum, the scalar mass matrix should be positive denite. In fact, all the
second-derivatives of V vanish at A = 0 except
_

2
V
AA

_
=
6
F

(0)
, (4.29)
This should be positive in order to have a local minimum at A = 0. This can be achieved
by choosing F

(0) to be positive. Hence, for a wide choice of the function F in (4.4), there
is a stable vacuum with vanishing cosmological constant at A = 0. Does this vacuum break
supersymmetry spontaneously? As stated in Section 2, supersymmetry is spontaneously
broken if and only if
D
A
W =
W
A
+
K
A
W
= 6AF

(A)
18F

(A)(AF

(A) F(A))
F

(A) +F

(A

)
, (4.30)
does not vanish when evaluated at the vacuum. In this case, using equation (4.28), we can
easily see that D
A
W = 0. Therefore, supersymmetry is not broken at the A = 0 vacuum.
We call the A = 0 vacuum solution the trivial vacuum.
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 13
Case II: A = 0
Away from A = 0 we can rewrite the potential (4.25) in the form
V (A, A

) =
12|A|
2
_
F

(A) +F

(A

)
_
2
U(A, A

), (4.31)
where
U(A, A

) = 3
F(A)
A
2F

(A) + c.c.. (4.32)

In general, a local minimum of V does not need to be a minimum of U. However, away from
A = 0, V is a strictly positive multiple of U. Thus, it is true that a local minimum of V ,
with the added condition that V = 0, is a local minimum of U with U = 0.
We see from (4.32) that U is a harmonic function. Therefore, it cannot have a local
minimum, except in the special case when it is a constant. We conclude that if U is not a
constant function, there exists no A = 0 vacuum with zero cosmological constant. Now
let U be constant. In order to have vanishing cosmological constant U must also vanish. It
follows from (4.32) that F() must satisfy
3
F()

2F

() = 0, (4.33)
which has the unique solution
F() = c
3/2
, (4.34)
where c is a constant which we will set to unity without loss of generality.
The choice of F given by (4.34) is an interesting case. It is a higher-derivative theory of
supergravity that is equivalent to Einstein supergravity coupled to a single supereld with
vanishing potential energy. The Kahler potential and superpotential corresponding to (4.34)
are given by
K = 3 ln
_
3
2

1/2
+
3
2

1/2
_
,
W = 3
3/2
.
(4.35)
Einstein supergravity coupled to matter is invariant under Kahler-superWeyl transforma-
tions, which can be used to scale the superpotential W to unity. Doing this yields a new
Kahler potential K

= K + lnW + ln W

which we nd to be
K

= 3 ln
3
1/3
2
_

1/2
+

1/2
_
, (4.36)
as well as the new superpotential
W

= 1. (4.37)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 14
With the eld redenition
S =
3
1/3
2

1/2
, (4.38)
the Kahler potential and superpotential take the standard form of no-scale supergravity [4]
K

= 3 ln
_
S +S

_
,
W

= 1.
(4.39)
Since the scalar eld potential energy of this theory vanishes, any S = 0 value is a vacuum
solution with zero cosmological constant. It follows from (2.22) that
D
S
W

=
K

S
=
3
S +S

. (4.40)
Therefore, D
S
W

= 0 for any S = 0 and, hence, supersymmetry is always spontaneously

broken. It is amusing to note that the simple no-scale theory is, in fact, completely equivalent
to a higher-derivative theory of pure supergravitation specied by Lagrangian (4.4) with
function F given in (4.34).
5. A Generic Higher-Derivative Supergravity
In Section 4, we studied a class of supergravity theories where the Lagrangian is the sum
of an arbitrary function of the chiral supereld R and its hermitian conjugate. The vacuum
structure of this class of theories was found to be very restricted. If we insist on having
zero cosmological constant, we are generically forced to the trivial vacuum at A = 0. This
vacuum does not spontaneously break supersymmetry.
In this section, we consider a second, more general, class of supergravity theories in which
the Lagrangian is given by
L = 3
_
d
4
Ef(R, R

), (5.1)
where f is a real function. Of course, this class will include the previous one as a special
case. In the following, we will restrict our discussion to functions f(R, R

) that cannot be
split into f(R) +f(R

). Most of the following analysis will break down for the f(R) +f(R

)
case, and we will make occasional reference to where such breakdown occurs.
What is the dynamical content of Lagrangian (5.1)? We need to compute the component
eld Lagrangian to answer this question. We start by writing (5.1) in the chiral superspace
as
L =
3
4
_
d
2
E(

D
2
8R)f(R, R

) + h.c. (5.2)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 15
The full component eld Lagrangian can be evaluated using the -expansions of E and R
given in (2.10). However, this component Lagrangian is very complicated. For the present
purpose, it suces to consider only the bosonic part of the Lagrangian. In this case, we can
use the expressions for E and R given in (4.5). Substituting these into (5.2) and performing
the -integral, we nd that
e
1
L
Bosonic
=
1
2
(f +Mf
M
+M

f
M
4MM

f
MM
2b
m
b
m
f
MM
)R
3
4
f
MM
R
2
+ 3f
MM

m
M
m
M

3f
MM
(
m
b
m
)
2
+i(f
M

m
M f
M

m
M

)b
m
i(f
M
M f
M
M

)
m
b
m

1
3
MM

[f 2(Mf
M
+M

f
M
) + 4MM

f
MM
]
+
1
3
b
m
b
m
(f +Mf
M
+M

f
M
4MM

f
MM
b
m
b
m
f
MM
), (5.3)
where f = f(
1
6
M,
1
6
M

), f
M
= f/M, and f
M
= f/M

. The rst notable feature

of this Lagrangian is the appearance of an R
2
term in addition to R. Hence, this is a higher-
derivative theory of gravity. Note that there are no higher powers of R. Therefore, this is
a theory of quadratic gravitation only. Recall that R +R
2
gravity describes a graviton plus
one real scalar degree of freedom. The second feature is that there are no auxiliary elds.
Both M and b
m
have kinetic energy terms. The eld M has the normal kinetic energy for a
complex scalar eld with a sigma-model factor in front. The eld M represents one complex
scalar degree of freedom or two real degrees of freedom. The vector eld b
m
has a kinetic
energy term for its longitudinal mode. This is one more propagating degree of freedom.
Note that the coecient of R
2
and the kinetic energy terms of M and b
m
are proportional to
f
MM
. This will vanish if f(R) splits into f(R) + f(R

), a case which we exclude from the

discussion as mentioned before. To conclude, the bosonic part of Lagrangian (5.1) describes
the propagation of four real degrees of freedom in addition to the usual graviton of Einstein
theory. What about the fermionic degrees of freedom? Since the theory is supersymmetric,
there must be extra fermionic degrees of freedom that propagate along with the new bosonic
degrees of freedom. These fermionic degrees of freedom should be associated with higher-
derivatives acting on
m
, as it is the only fermionic eld in the supergravity multiplet.
An analysis of the fermionic part of the Lagrangian shows that the gravitino eld equation
is, indeed, not rst-order but third-order. We will exhibit these new fermionic degrees of
freedom explicitly below.
In analogy with the analysis in the previous section, we would like to make the extra de-
grees of freedom explicit. Therefore, we would like a supersymmetric analog of the Legendre
transform method. Here we have four real bosonic degrees of freedom, or equivalently two
complex scalars, along with their fermionic partners. These degrees of freedom can only
arrange themselves into two chiral supermultiplets. Therefore, we introduce two chiral su-
perelds and and try to rewrite the Lagrangian (5.2) in an equivalent second-order form
in which the new propagating degrees of freedom are contained in and . This equivalence
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 16
was rst established in [14] using the compensator formalism of supergravity. It was also
discussed in supergravity theories with Chern-Simons terms, such as those associated with
low energy superstring eective Lagrangians [13, 15]. Consider a new Lagrangian
L =
3
4
_
d
2
E(

D
2
8R)f(,

) + 6
_
d
2
E( R) + h.c.
=
3
4
_
d
2
E(

D
2
8R)
_
f(,

) + +

_
+ 6
_
d
2
E + h.c. (5.4)
The equation of motion of can be obtained by varying Lagrangian (5.4) with respect to
. From the rst line of (5.4) we have

L = 6
_
d
2
E( R) . (5.5)
Setting

L to zero gives the equation of motion

= R. (5.6)
Substituting this back into (5.4) yields the original Lagrangian (5.2). For completeness, we
display the equation of motion which is given by
=
1
8
_

D
2
8R
_
f

. (5.7)
Now let us compare Lagrangian (5.4) with the standard form of the Lagrangian for chiral
matter coupled to supergravity in chiral superspace (2.18). We clearly see that Lagrangian
(5.4) describes Einstein supergravity coupled to two chiral superelds and with the
Kahler potential and superpotential
K = 3 ln
_
f(,
ger
) + +
ger
_
,
W = 6,
(5.8)
respectively. Although the case in which the function f(,

) splits into f() + f(

) is
excluded from the present discussion, we would like to point out that in such a case (5.8) is
still valid. However, one can also perform the eld redenetion
f() + , (5.9)
to render the Kahler potential independent of . Thus, the supereld becomes auxiliary
with an algebraic equation of motion. One can use this equation of motion to eliminate
from the Lagrangian and end up with a theory of Einstein supergravity coupled to a single
propagating chiral supereld. This is in agreement with the discusion of the previous section.
The extra degrees of freedom in the higher-derivative theory (5.1) are now manifest in the
lowest and fermionic components of the superelds and
= A+

F,
= B +

G.
(5.10)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 17
Using the equations of motion (5.6) and (5.7), we nd that to lowest order the new propa-
gating degrees of freedom are given by
A =
1
6
M,

2
12
_

m

n
(
m

n
)
_

,
(5.11)
and
B =
1
6
f
R
(0, 0)M
1
12
f
RR
(0, 0)(
1
2
R +i
m
b
m
),

2
12
f
R
(0, 0)
m


n
(
m

m
)
i

2
24
f
RR
(0, 0)
m

_

m

_
.
(5.12)
As in the previous section, the scalar and fermionic component elds in are the M eld
and, essentially, the rst derivative of the gravitino
m
respectively. On the other hand, the
scalar eld associated with is constructed from M, R and
m
b
m
, whereas the fermionic
eld is, essentially, the second derivative of the gravitino.
To study the vacuum structure of the theory we need to evaluate the bosonic part of the
component eld Lagrangian corresponding to (5.2). The bosonic part of the Lagrangian is
given in (2.19) where K and W are given in (5.8). The kinetic energy for the bosonic elds
A and B is obtained by substituting (5.8) into (2.20). The result is
e
1
L
KE-Bosonic
= 3(f +B +B

)
2
_
(f +B +B

) f
AA

m
A
m
A

+|f
A

m
A +
m
B|
2
_
.
(5.13)
We see that for non-ghost-like propagation of the elds A and B we require that at the
vacuum
(f +B +B

)f
AA
< 0. (5.14)
We can also substitute (5.8) into (2.21) to give the potential energy
V = 12 (f(A, A

) +B +B

)
2
U(A, B), (5.15)
where U is given by
U = |A|
2
_
f 2 (f
A
A+f
A
A

) + 4f
AA
|A|
2
_
f
1
AA

B f
A
A+ 2f
AA
|A|
2

2
. (5.16)
An important question is whether or not potential (5.15) has stable vacua, generically
denoted by A and B, with zero cosmological constant that break supersymmetry spon-
taneously. In general, a local minimum of V need not be a local minimum of U and vice
versa. However, V is a strictly positive multiple of U. As in Section 4, it is simple to argue
that A, B is a local minimum of V with zero cosmological constant if and only if it is
a local minimum of U at which U vanishes. The function U has a simpler structure than
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 18
V and it is possible to make some general statements about its zero cosmological constant
minima. First note that the second term in U is unbounded from below unless

f
1
AA

_
< 0. (5.17)
If this condition is satised, the condition for non-ghost-like propagation (5.14) simplies to
f +B +B

> 0. (5.18)
Assuming that

f
1
AA

_
is negative denite, since B enters only the second term of (5.16),
minimizing U with respect to B gives
B = f
A
A2f
AA
AA

, (5.19)
at which point the second term vanishes. It follows that A must be a local minimum of
the rst term on the right hand side of (5.16) with zero cosmological constant. Thus, the
problem of nding a local minimum of V with vanishing cosmological constant and where
the A and B elds have non-ghost-like propagation is reduced to nding a local minimum
A of
U
0
= |A|
2
_
f 2[f
A
A+f
A
A

] + 4f
AA
|A|
2
_
, (5.20)
at which U
0
= 0 and which satises (5.17) and (5.18). The value of B then follows from
(5.19).
The above discussion enables us to construct a wide class of Lagrangians which have a
rich vacuum structure. We demonstrate this by considering a concrete example. Since we
will be interested in the dierent energy-scales in this model, we restore the gravitational
coupling constant = M
1
P
where M
P
is the Planck mass. Consider
f(R, R

) = 1 2
RR

m
2
+
1
9
(RR

)
2
m
4
, (5.21)
where m is a coupling parameter with mass dimension one. This coupling parameter need
not be related to the Planck mass. We would also like all the elds to have canonical
dimensions. In the above discussion, has mass-dimension one and is dimensionless. In
order to give dimension one, we scale it by M
P
. Furthermore, it is convenient to write
the Kahler potential in terms of M
P
only, relegating mass m to the superpotential. This be
achieved if we scale by M
P
/m. The Kahler and superpotential associated with (5.21) are
then given by
K = 3M
2
P
ln
_
1 2

M
2
P
+
1
9
(

)
2
M
4
P
+
+

M
P
_
,
W = 6m.
(5.22)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 19
Using (5.15) and (5.21) the potential energy becomes
V = 12m
2
M
2
P
_
1 2
|A|
2
M
2
P
+
1
9
|A|
4
M
4
P
+
B +B

M
P
_
2
_
|A|
2
M
2
P
_
|A|
2
M
2
P
1
_
2
+
9
2
_
9 2
|A|
2
M
2
P
_
1

B
M
P

2
3
|A|
2
M
2
P
_
3
|A|
2
M
2
P
_

2
_
, (5.23)
and the associated function U
0
is given by
U
0
=
|A|
2
M
2
P
_
|A|
2
M
2
P
1
_
2
. (5.24)
This function is non-negative and clearly has two local minima at which U
0
vanishes. The
rst is at A = 0. Now
f
AA
=
2
M
2
P
_
1
2
9
|A|
2
M
2
P
_
. (5.25)
Therefore, at A = 0,

f
1
AA

_
< 0 as required. It follows from (5.19) that B = 0. We
conclude that the above potential has a local minimum at
A = B = 0. (5.26)
with vanishing cosmological constant. This minimum is clearly visible at the center of the
potential energy plotted in Figure 1. In order for (5.26) to be a physically acceptable vacuum,
the kinetic-energy terms for the scalar elds must have the correct sign when evaluated at
this point, otherwise elds A and B would be ghost-like. We argued above that the kinetic-
energy terms have the correct sign if and only if f +B +B

> 0. In fact
f +B +B

= 1 > 0 (5.27)
as required. Explicitly, the A and B eld kinetic-energy terms are given in (5.13). When
evaluated around vacuum (5.26) to quadratic order, they become
e
1
L
KE-Bosonic
= 6
m
A
m
A

3
m
B
m
B

, (5.28)
which is clearly non-ghost-like. Is supersymmetry broken by this vacuum? Using (2.22) and
(5.22), we nd that
D
A
W = 6mB
_
_
_
1 + 6
|A|
2
M
2
P
_
1
1
9
|A|
2
M
2
P
_
_
1 2
|A|
2
M
2
P
+
1
9
|A|
4
M
4
P
+
B+B

M
P
_
_
_
_
,
D
B
W = 6mA
_
_
_
1 3
B
M
P
1
_
1 2
|A|
2
M
2
P
+
1
9
|A|
4
M
4
P
+
B+B

M
P
_
_
_
_
.
(5.29)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 20
It follows that for A = B = 0 both D
A
W and D
B
W vanish. Hence, supersymmetry
is not spontaneously broken. We call vacuum (5.26) the trivial vacuum.
Figure 1. The potential energy as a function of A at B = f
A
A2f
AA
AA

.
Are there any other minima of potential (5.23)? As can be seen from Figure 1, there is a
second local minimum of U
0
given in (5.24) at which U
0
vanishes. It occurs for | A |
2
= M
2
P
or A = M
P
e
i
for arbitrary real phase . Note that this degeneracy in the vacuum is a
direct consequence of the symmetry of the function f in (5.21), and hence the Lagrangian,
under the transformation R e
i
R. It follows from (5.25) that at this minimum

f
1
AA

_
< 0
as required. Equation (5.19) then gives B =
4
3
M
P
. Therefore, the potential energy has a
ring of local minima at
A = M
P
e
i
,
B =
4
3
M
P
,
(5.30)
with vanishing cosmological constant. These are visible as the circular set of minima in
Figure 1. To check that there are no ghosts in the theory around vacuum (5.30), we need to
show that condition (5.18) is satised. We nd that
f +B +B

=
16
9
> 0 (5.31)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 21
as required. Explicitly, expanding A = A +a and B = B +b, it follows from (5.13) and
(5.30) that to quadratic order
e
1
L
KE-AB
=
21
8

m
a
m
a

m
c
m
c

, (5.32)
where c =

3(
9b
16
e
i
a), which is clearly non-ghostlike. The masses of the four real
scalars are easily evaluated at this minimum, but the exact values are complicated and
unenlightening. Suce it to say that, in addition to the one zero-mass eld in the direction
of the circular set of minima, the three remaining masses are numerically dierent but of
order m. Is supersymmetry broken at the vacua (5.30)? Using (5.29), we nd that the
Kahler covariant derivatives evaluated at (5.30) are
D
A
W = 32mM
P
,
D
B
W =
15
2
e
i
mM
P
.
(5.33)
It follows that supersymmetry is indeed spontaneously broken at these vacua (5.30). Sub-
stituting (5.22) into (2.23) gives the gravitino mass
m
3/2
=
27
8
m, (5.34)
which is of the same order as the non-vanishing scalar masses. Note that the supersymmetry
breaking scale is set by D
A
W and D
B
W, which are of order mM
P
. It is conventional to
denote this scale by =

mM
P
. The gravitino mass can then be written in the familiar
form
m
3/2
=
27
8

2
M
P
. (5.35)
With this denition, the non-vanishing scalar eld masses at the non-trivial vacuum are of
order
2
/M
P
.
The preceding discussion is exact and valid at any energy scale. Indeed, as we have shown
above, it is necessary to know the full theory in order to determine the vacuum structure.
Once one has found the non-trivial supersymmetry breaking vacuum (5.30), however, it is
of interest to consider small uctuations around this vacuum, and to determine the eective
theory for these uctuations at energy scales much smaller than the Planck mass M
P
. We
now proceed to do this. Above we did not specify the value of the supersymmetry breaking
scale relative to M
P
. For the remainder of this discussion we will assume that M
P
.
For example, if we would like the gravitino mass to be of the order of the electroweak scale,
that is m
3/2
10
2
GeV, then we must choose 10
11
GeV, eight orders of magnitude
smaller than M
P
. Expanding around the non-trivial vacuum (5.30) with = 0, we dene
= M
P
+,
=
4
3
M
P
+,
(5.36)
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 22
where and are the uctuations of and around their vacuum expectation values, M
P
and
4
3
M
P
respectively. We will henceforth consider and to be of the order of or smaller.
The exact Lagrangian is given by expression (2.15) with Kahler potential and superpotential
(5.22). At the scale , since /M
P
1, the gravitational eects become negligible, and the
supergravity Lagrangian (2.18) simplies to the at superspace Lagrangian given by
L =
_
d
2
d
2

K +
__
d
2
W + h.c
_
. (5.37)
However, in this regime we must also neglect terms in K and W suppressed by powers of
/M
P
. That is the at superspace Lagrangian (5.37) in fact reduces to
L =
_
d
2
d
2

+
__
d
2
W

+ h.c.
_
, (5.38)
where K

and W

are obtained by substituting (5.36) into (5.22) and keeping only the rst
non-trivial terms in /M
P
. Note that any terms in the Kahler potential which are constant
or purely chiral plus antichiral, such as

, do not contribute to Lagrangian (5.38). A

constant term in the superpotential also does not contribute. In this limit, we nd that
K

=
45
8

27
16
(

) +
243
256

,
W

= 6
2
(
4
3
+).
(5.39)
We can diagonalize the Kahler potential by making the following eld redenitions
=
_
21
8
,
=

3(
9
16
).
(5.40)
In terms of superelds and , Kahler potential and superpotential (5.39) become
K

,
W

=
2
_
_
7
2
+
_
,
(5.41)
where
2
=
32

3
9

2
. The scalar eld potential energy corresponding to (5.41) is simply a
constant given by
V =
9
2

4
. (5.42)
Therefore, below the Planck scale, the theory behaves like a simple version of the
ORaifeartaigh model [16] with two chiral superelds, and . The positive denite value
of the scalar potential signals supersymmetry breaking at scale . Note that the Kahler
potential and superpotential (5.41) look like a simple extension of the Polonyi model with
two superelds. For comparison, let us consider the original Polonyi model for supersymme-
try breaking [5]. The Polonyi model has one chiral supereld Z with Kahler potential and
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 23
superpotential given by
K = ZZ

,
W =
2
(Z +),
(5.43)
where is a constant that is chosen to ensure vanishing cosmological constant. The exact
theory is found to have a unique vacuum at
Z = (

3 1)M
P
, (5.44)
which has vanishing cosmological constant if is chosen to be = (2

3)M
P
. Supersym-
metry is spontaneously broken with strength at this vacuum since
D
Z
W =

3
2
. (5.45)
Expanding Z = Z +z, and performing the same analysis as we did above, we nd that the
eective at superspace Kahler potential and superpotential are given by
K

= zz

,
W

=
2
z.
(5.46)
The associated scalar eld potential energy is
V =
4
. (5.47)
Comparing (5.41) and (5.42) with (5.46) and (5.47) respectively, we see that our higher-
derivative supergravitation model is, at low energy, simply a two supereld Polonyi model,
with no essential dierences. They only dier at the Planck scale. The Polonyi model can be
thought of as the most trivial extension of the at superspace ORaifeartaigh model. In the
Polonyi extension, one does not change the low energy superpotential or Kahler potential at
the Planck scale, except for the addition of the constant term which is required to set the
cosmological constant of the non-trivial vacuum to zero. However, this extension is by no
means unique. As one goes up in energy, one could start seeing generic modications to both
the superpotential and the Kahler potential consisting of new terms that are suppressed by
powers of the Planck mass. The theory constructed in this section is an explicit example of
such a phenomenon.
We end this section by noting that there is a class of general higher-derivative supergravity
theories which have a no-scale structure, with at directions in the scalar potential. In the
previous section, when f(R, R

) could be decomposed into f(R) + f(R

), we found that
choosing F(R) = Rf(R) = cR
3/2
made the scalar potential identically zero. In fact,
the transformed theory, in terms of the new chiral eld , was identical to pure no-scale
supergravity. In the general case, we see that we cannot chose f(R, R

) such that the whole

scalar potential is zero because the second, quadratic, term in U (5.16) cannot be made to
vanish. However, choosing f(R, R

) can introduce at directions into the potential, giving a

SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 24
continuous set of degenerate vacua with zero cosmological constant. The maximum possible
degeneracy is acheived by setting the rst term in the potential, U
0
(5.20), to zero. The
general form of f(R, R

) with this property is

f
_
R, R

_
=

RR

_
h(R) + (h(R))

_
, (5.48)
where h(R) is an arbitrary complex function of R. The Kahler potential and superpotential
for the transformed theory are then given by
K = 3 ln
_

_
h() + (h())

_
+ +

_
,
W = 6.
(5.49)
and the scalar potential becomes
V =
48|A|
[|A|(h +h

) +B +B

]
2
[h +Ah

+h

+A

B +|A| A

2
, (5.50)
where h

= dh/dA. Considered as a function of B, the potential is clearly only bounded

from below if the prefactor in (5.50) is positive. As before this gives the condition that at
the vacuum

f
1
AA

_
=

|A|
_
h +Ah

+h

+A

__
< 0. (5.51)
As before to ensure that the A and B elds have non-ghost-like propagation we require
f +B +B

> 0. It is clear that setting B = |A|A

sets the potential to zero, inde-

pendent of the value of A. Thus the potential has at directions, with a continuous set of
vacua, stable provided (5.51) is satised, and given by
A = A
0
,
B = |A
0
| A

0
[h

(A
0
)]

.
(5.52)
As in the pure no-scale model, all these vacua break supersymmetry, with zero cosmological
constant. The value of the gravitino mass depends on the vaccum in question and is given
by
m
3/2
=
|A|
3
|h

|
2
h Ah

+h

. (5.53)
The important point here, as in all no-scale models, is that the at directions in the scalar
potential imply that the vacuum is undetermined. As a result the value of the gravitino
mass is also undetermined at tree level.
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 25
6. Conclusion
We have shown in this paper that higher-derivative N = 1 supergravity in four dimensions
has non-trivial vacua with vanishing cosmological constant that spontaneously break super-
symmetry. This result opens the possibility of a new approach to supersymmetry breaking
in phenomenological supergravity theories and, perhaps, in superstring theories. In this
new approach, supersymmetry is spontaneously broken by a non-trivial vacuum of the new
degrees of freedom associated with higher-derivative supergravitation, and does not need
a hidden sector or gaugino condensates. A more complete understanding of this approach
to supersymmetry breaking requires coupling higher-derivative supergravity to matter and
examining the eect of the non-trivial supergravity vacuum on the low energy matter La-
grangian. It is clear that supersymmetry will be broken in this Lagrangian, but the details
of the pattern and strength of this breaking require careful study. This study is presently
underway [17].
Acknowledgments
This work was supported in part by DOE Grant No. DE-FG02-95ER40893 and NATO
Grand No. CRG-940784.
References
[1] A. Hindawi, B. A. Ovrut, and D. Waldram, Consistent spin-two coupling and quadratic gravitation,
Phys. Rev. D 53 (1996), 55835596, hep-th/9509142.
[2] A. Hindawi, B. A. Ovrut, and D. Waldram, Non-trivial vacua in higher- derivative gravitation, Phys.
Rev. D 53 (1996), 55975608, hep-th/9509147.
[3] A. Hindawi, B. A. Ovrut, and D. Waldram, Two-dimensional higher-derivative supergravity and a new
mechanism for supersymmetry breaking, Nucl. Phys. B471 (1996), 409429, hep-th/9509174.
[4] E. Cremmer, S. Ferrara, C. Kounnas, and D. V. Nanopoulos, Naturally vanishing cosmological constant
in N = 1 supergravity, Phys. Lett. 133B (1983), 6166.
[5] J. Poloyni, Budapest Preprint KFKI-93 (1977).
[6] K. Foerger, B. A. Ovrut, S. Theisen, and D. Waldram, Higher derivative gravity in string theory, Phys.
Lett. 388B (1996) 512520, hep-th/9605145.
[7] J. Wess and J. Bagger, Supersymmetry and supergravity, 2nd ed., Princeton University Press, Princeton,
1992.
[8] E. Cremmer, B. Julia, J. Scherk, S. Ferrara, L. Girardello, and P. van Nieuwenhuizen, Spontaneous
symmetry breaking and Higgs eect in supergravity without cosmological constant, Nucl. Phys. B147
(1979), 105131.
[9] G. Magnano, M. Ferraris, and M. Francaviglia, Nonlinear gravitational Lagrangians, Gen. Rel. Grav.
19 (1987), 465479.
[10] A. Jakubiec and J. Kijowski, On theories of gravitation with nonlinear Lagrangians, Phys. Rev. D 37
(1988), 14061409.
SUSY BREAKING IN FOUR-DIMENSIONAL SUPERGRAVITY 26
[11] M. Ferraris, M. Francaviglia, and G. Magnano, Do nonlinear metric theories of gravitation really exist?,
Class. Quantum Grav. 5 (1988), L95L99.
[12] G. Magnano, M. Ferraris, and M. Francaviglia, Legendre transformation and dynamical structure of
higher-derivative gravity, Class. Quantum Grav. 7 (1990), 557570.
[13] S. Ferrara, M. T. Grisaru, and P. van Nieuwenhuizen, Poincare and conformal supergravity models with
closed algebras, Nucl. Phys. B138 (1978), 430444.
[14] S. Cecotti, Higher derivative supergravity is equivalent to standard supergravity coupled to matter, Phys.
Lett. 190B (1987), 8692.
[15] B. A. Ovrut and S. Kalyana Rama, Lorentz and U(1) Chern-Simons terms in new minimal supergravity,
Phys. Lett. 254B (1991), 132138.
[16] L. ORaifeartaigh, Spontaneous symmetry breaking for chiral scalar superelds, Nucl. Phys. B96 (1975),
331.
[17] A. Hindawi, B. A. Ovrut, and D. Waldram, Soft supersymmetry breaking induced by higher-derivative
supergravitation in the electroweak standard model, Phys. Lett. 381B (1996), 154162, hep-th/9602075.