Flatons and Peccei-Quinn Symmetry

IEMFT12x/99
INFNFE-02-99
KIASP99016
E. J. Chun
Korea Institute for Advanced Study
207-43 Cheongryangri-dong, Dongdaemun-gu, Seoul 130-012, Korea
D. Comelli
INFN, sezione di Ferrara
Ferrara, via Paradiso 12 44100 Italy
David H. Lyth
Department of Physics,
Lancaster University,
Lancaster LA1 4YB. U. K.
March 1999
Abstract
We study in detail a supersymmetric Peccei-Quinn model, which has a DFSZ and
a KSVZ version. The elds breaking the Peccei-Quinn (PQ) symmetry correspond
to at directions (aton elds) and have unsuppressed couplings when PQ symmetry
is unbroken. The models have interesting particle physics phenomenology. The PQ
scale is naturally generated through radiative corrections; also, in the DFSZ case, the
problem can be solved and neutrino masses can be generated. Cosmologically they
lead to a short period of thermal ination making the axion an excellent dark matter
candidate if one of the aton elds has a positive eective mass-squared at early times
but with too low a density in the opposite case. A highly relativistic population of
axions is produced by aton decay during the subsequent reheating, whose density
is constrained by nucleosynthesis. We compute all of the relevant reaction rates and
evaluate the nucleosynthesis constraint. We nd that the KSVZ model is practically
ruled out, while the DFSZ model has a sizable allowed region of parameter space.
1 Introduction
With the discovery of the instantons it was realized that the pure gradient topological term
of the QCD Lagrangian
QCD
g
2
S
/32
2
F

F can generate important physical consequences. In
fact the induced CP violation aects the electric dipole moment of the neutron suggesting the
limit
QCD
10
10
. The most attractive explanation for the origin of such a small parameter
would be the Peccei-Quinn mechanism [1]. There is supposed to be a spontaneously broken
global U(1) symmetry (PQ symmetry), which is also explicitly broken by the color anomaly.
The corresponding pseudo-Goldstone boson is called the axion [2].
The PQ symmetry acts on some set of elds
i
with charges Q
i
,

i
e
iQ
i

i
. (1)
To generate the require U(1)
PQ
SU(3)
c
SU(3)
c
anomaly we must choose appropriate
particle spectra. Depending on the charge of the SM matter content, we can have the KSVZ
(hadronic) models [3] in which only some extra heavy quark elds are PQ charged or the
DFSZ [4] models in which, beyond the extra matter content (at least two Higgs elds), also
the Standard Model (SM) matter is PQ charged.
Denoting the vacuum expectation values of the scalar elds charged under PQ by v
i
/

2,
the PQ symmetry breaking scale F
PQ
is dened as F
2
PQ
=

i
Q
2
i
v
2
i
, and the axion mass is
given by F
PQ
m
a
= (79 MeV)
2
N where N is the number of quarks with PQ charge. Col-
lider and astrophysics constraints require F
PQ

> 10
9
GeV, and dening as usual M
Pl
=
(8G
N
)
1/2
= 2.4 10
18
GeV this allows the range
10
9
GeV

< F
PQ

< M
Pl
. (2)
With typical assumptions about the cosmology, the requirement that axions give at most
critical density places F
PQ
towards the bottom of this range. In the particular case that the
axions are radiated by strings with no subsequent entropy production, one probably requires
[5] F
PQ
10
10
GeV.
In a model with unbroken supersymmetry, the holomorphy of the superpotential ensures
that PQ symmetry Eq. (1) is accompanied by a symmetry acting on the radial parts of the
PQ charged elds

i
e
Q
i

i
. (3)
The corresponding pseudo-Goldstone boson is called the saxion (or saxino), and the spin-half
partner is called the axino. Soft supersymmetry breaking gives the saxion a mass of order
100 GeV, and the axino typically has a mass of the same order [6] though it may be very
light in special cases.
1
Let us consider the potential of the elds
i
which break PQ symmetry. In a supersym-
metric model there have to be at least two, but let us pretend for the moment that there is
1
We assume gravity-mediated supersymmetry breaking, which typically gives soft scalar masses in the
range 100 to 1000 GeV, and use the former estimate for deniteness.
1
only one. Its potential will be of the form
V = V
0
m
2
[[
2
+
1
4

4
+

n=1

n
[[
2n+4
M
2n
Pl
. (4)
The non-renormalizable terms are expected to have coecients
n
1. If the renormalizable
coupling is also of order 1, m F
PQ
and the non-renormalizable terms are negligible.
(Remembering that there are at least two complex elds, only the particular combination of
elds corresponding to the saxion will have the soft mass of order 100 GeV.) In this sort of
model one can hope to understand a value F
PQ
10
10
GeV since that is the supersymmetry
breaking scale [7], but it may be hard to understand a bigger scale.
We are concerned with a dierent class of models [8, 9, 10, 11, 12], in which represents
a at direction of supersymmetry. The quartic term is then absent, while m is a soft
mass of order 100 GeV. If the
6
term is present with unsuppressed coecient the vev
is )
1/4
1
10
10
GeV. If instead the
8
dominates one has )
1/6
2
10
13
GeV and so
on. For future reference, note that the mass of [[ in the vacuum is also of order of m, and
that the height of the potential is given by
_
_
V
1/4
0
10
6
GeV
_
_

_
)
10
10
GeV
_
1/2
. (5)
Fields of this kind, characterized by a large vev and a at potential are called aton elds,
and the particles corresponding to them are called atons [13].
In a supersymmetric model there are n 2 complex elds which all acquire vevs. We
are interested in the case where these are aton elds. Then there are 2n1 aton particles
with mass of order 100 GeV, and n atinos with typically similar masses. The saxion (axino)
is a linear combination of the aton (atino) elds, with no particular signicance.
The rest of this paper is as follows. In the next section we dene our models and
summarize the cosmology. In section 3, we give the general structure of the aton and
atino masses. In section 4 we analyze the general self interactions between atons and
atinos. In section 5 we see the eect of the interaction of the atons with the matter elds
(The KSVZ atons interact only with gluons and gluinos whereas DFSZ atons interact also
with ordinary matter and supermatter.) In section 6 we nd the parameter space regions
that can satisfy the cosmological constraints. We conclude in Section 7.
2 The model and its cosmology
We consider a DFSZ and a KSVZ (hadronic) model.
In both models, there are two aton elds P and Q, interacting with the superpotential
[11, 12]
W
aton
=
f
M
Pl

P
3

Q. (6)
2
In the hadronic version, the interaction with matter is
W
atonmatter
= h
E
i

E
i

E
c
i

P (7)
where E
i
and E
c
i
are additional heavy quark and antiquark superelds.
In the DFSZ version, the interaction is
W
atonmatter
=
1
2

N

N

P +
g
M
Pl

H
1

H
2

P

Q (8)
where

N are the right handed neutrino superelds and

H
1,2
the two Higgs doublets. Due to
the second term we can provide a solution to the problem [14]. In such case we can add to
the superpotential of the minimal supersymmetric standard model also the terms h

l

H
2

N
that generate the necessary mixing between left and right neutrinos to implement a see saw
mechanism which can explain the solar and atmospheric neutrino decits.
The cosmology of the DFSZ model has already been considered in a rough way [15]. Here
we relatively complete treatment of both models.
To study the cosmology of the aton elds, we can safety analyze only the superpotential
W
aton
for both models. With the inclusion of the soft susy breaking terms, the potential is
V = m
2
P
[
P
[
2
+m
2
Q
[
Q
[
2
+
f
2
M
2
Pl
_
9[
P
[
4
[
Q
[
2
+[
P
[
6
_
+
_
A
f
M
Pl
f
3
P

Q
+h.c.
_
. (9)
The soft parameters m
P
, m
Q
and A
f
are all of order 10
23
GeV in magnitude. It is assumed
that m
2
P
and m
2
Q
are both positive at the Planck scale. The unsuppressed interactions of
P
give radiative corrections which drive m
2
P
to a negative value at the PQ scale, triggering a
vev for
P
. As a result, when
P
gets a vev, the tadpole term proportional to A
f
generates
automatically a vev for the
Q
eld, both of them are v
Q
v
P
10
1012
GeV.
In the early Universe when H

> 100 GeV, there will be eective values m

2
P
(t) and m
2
Q
(t).
Supergravity interactions will make both of them at least of order H
2
in magnitude.
2
The
cosmology depends on the signs of these eective values. If m
2
P
(t) is positive,
P
is held at the
origin by the nite-temperature potential until T [m
P
[. But the potential V
0
dominates
the energy density in the regime [m
P
[

< T

< V
1/4
0
, leading to about ln
_
V
1/4
0
/[m
P
[
_
10
e-folds of thermal ination [13].
To discuss what happens after thermal ination, we suppose rst that m
2
Q
(t) is also
positive in the early Universe, so that
Q
is also trapped at the origin during thermal
ination. When thermal ination ends,
P
moves away from the origin, which destabilizes

Q
. The elds
P
and
Q
move around an orbit in eld space, which would be closed if there
were no energy loss. If the only energy loss came from Hubble damping the elds would
oscillate back and forth many times around an almost-closed orbit. However, the parameter
determining the strength of parametric resonance is q g
0
/m 10
8
, where g 1 is a
typical coupling,
0
10
10
GeV is the amplitude of the oscillation and m 10
2
GeV is its
angular frequency. One therefore expects that parametric resonance will eciently damp
2
During ination this result might be avoided (say by D-term ination) but it should still hold afterwards.
3
the orbit, converting most of the energy into aton particles. At rst these particles are
marginally relativistic, but after a few Hubble times they become non-relativistic. The rest
of the energy resides in the homogeneous oscillating aton elds, now with small amplitude
and therefore almost simple harmonic motion corresponding to some more non-relativistic
atons. The atons decay leading to nal reheating at a temperature [13]
T
RH
1.2g

1
4
RH
_
M
Pl

3
_
10
11
GeV
F
PQ
_
_
m
300 GeV
_
3
GeV.
where m is the mass of the lightest aton and g
RH
100 is the eective number of relativistic
species at reheat.
Two dierent axion populations are produced. One population is radiated by the PQ
strings that form after thermal ination. They become dark matter with abundance [5]

a
(F
PQ
/10
10
GeV)
1.2
. The present scenario predicts F
PQ
10
10
GeV, making the axion
an excellent dark matter candidate. Note that in contrast with the general case, the axion
density in this scenario cannot be reduced by entropy production after the epoch T 1 GeV
when the axions acquire mass; pre-existing long-lived particles that might do the job have
been diluted away by the thermal ination.
The other axion population, that is our main concern, comes from the decay of the atons
[13, 6]. This population is still relativistic at nucleosynthesis, and its density must satisfy
the constraint
_

_
NT
N

0.1 1.5 , (10)

where

denotes the energy density of a single species of relativistic neutrino and N

the
number of extra neutrino species allowed by nucleosynthesis.
If there were just one species of aton eld , this would give the bound [15]
_

_
NT
=
43
7
_
43/4
g
RH
_
1/3
B
a
1 B
a

43
7
_
43/4
g
RH
_
1/3
B
a
, (11)
where B
a
=
a
( a +a) /
tot
is the branching ratio, and in the last line we assumed
B
a
1. From Eq. (10) we get the bound
( a +a)
( X)
0.24
_
N
1.5
__
g
RH
43/4
_
1/3
. (12)
Varying T
RH
from 6 MeV to m 100 GeV we get a factor two variation coming from the
number of degrees of freedom. For N

varying from 0.1 to 1.5 we get B

a
< 1/3 to B
a
< 0.02.
Our model has three aton particles and the quantity B
a
to be used in Eq. (11) is
B
a
=

I
r
I
B
I
where r
I
=
n
I

J
n
J
,
B
I

(I a +a) +
1
2
(I a +X)

tot
(I)
. (13)
4
Here n
I
is the number density of the Ith aton just before reheating, related to the mass
density
I
by n
I
=
I
/m
I
. In principle one could calculate the n
I
(coming from parametric
resonance and some residual homogeneous aton oscillations decay) but we have not done
that, and to estimate the allowed parameter range for the model we shall simply assume
r
I
= 1/3.
Next consider what happens after thermal ination in the case that m
2
Q
(t) is negative.
During thermal ination, while
P
is trapped at the origin, [
Q
[ will have some value F
PQ
determined by a higher-order non-renormalizable term. At the end of thermal ination, all
three aton particles will be produced in the manner we have described, and the nucleosyn-
thesis constraint still holds. The dierence from the previous case is that axionic strings are
not produced, so that dark matter axions are produced only by the quantum uctuation of
the axion eld during ination. The density is now [16, 17]
a
3(/)
2
(F
PQ
/10
12
GeV)
1.2
,
where < is the misalignment angle. Again, thermal ination means that the abundance
will not be diluted by entropy production, so discounting an accidentally small we again
have a rather denite prediction, which is too low. For the axions to be the dark mat-
ter in this scenario, one would have to increase F
PQ
by generating it from a higher-order
non-renormalizable term.
Finally, if m
2
P
(t) is negative in the early Universe, there is no thermal ination and one
is back with all the uncertainties of more general models [17, 18, 19]. Axion cosmology now
depends on the scale of the inaton potential, the reheat temperature after ination, the
decoupling temperature of the atons and so on. We have nothing to say about that case.
3 Flaton and atino spectrum
3.1 Flaton spectrum
We write the aton elds as

P
=
v
P
+P

2
e
i
A
P
v
P

Q
=
v
Q
+Q

2
e
i
A
Q
v
Q
, (14)
and we shall take v
P
, v
Q
, A
f
and f as the independent parameters in the potential Eq. (9).
The main components of the axion eld are
a =
v
P
F
PQ
A
P
+ 3
v
Q
F
PQ
A
Q
(15)
where F
2
PQ
= v
2
P
+ 9v
2
Q
and we have neglected the O(v
EW
/F
PQ
) components along the H
1,2
directions. The orthogonal eld to the axion (both are CP Odd) corresponds to a aton
particle. It is

=
v
P
F
PQ
A
Q
3
v
Q
F
PQ
A
P
(16)
5
which has a mass
M
2

=
fA
f
v
P
F
2
PQ
2M
Pl
v
Q
=
f
g
A
f
_
x
2
+ 9
_
. (17)
where

g

v
P
v
Q
2M
Pl
(18)
x
v
P
v
Q
. (19)
For future convenience we have introduced a quantity , related to the g appearing only in
the DFSZ model. At this stage results depend only on the ratio /g dened by (18) and
they apply to both models. Since M
2

is positive, A
f
and f must have opposite signs.
The other two aton particles correspond to the CP even elds P and Q. They have a
22 mass matrix whose components are
M
2
QQ
= M
2

x
2
9 +x
2
(20)
M
2
PQ
= 9f
2
v
4
P
M
2
Pl
x
3
M
2

x
9 +x
2
= 3x
_
12
f
2
g
2

M
2

9 +x
2
_
M
2
PP
= 3f
2
v
4
P
(x
2
+ 3)
M
2
Pl
x
2
3
M
2

9 +x
2
= 12
f
2
g
2
_
x
2
+ 3
_

2
3
M
2

9 +x
2
.
Here two mass parameters m
2
P
, m
2
Q
in Eq. (9) are replaced in favor of v
P
, v
Q
. Performing the
rotation from the avor basis [[P Q[[ to the mass basis [[F
1
F
2
[[
P = cos F
2
sin F
1
Q = sin F
2
+ cos F
1
, (21)
the mixing angle results
tan2 =
2M
2
PQ
M
2
QQ
M
2
PP
= 6
x
x
2
+ 3
a . (22)
Note that tan2 ranges from

3 for x =

3 to zero for x 1 and x 0. For each x

there are two solutions given by
cos
2

1
=
1
2
_
1
1

1 +a
2
_
(23)
cos
2

2
=
1
2
_
1 +
1

1 +a
2
_
(24)
with sin
1,2
0. The two solutions will be relevant when we discuss the parameter space
for the decay of atons into axions.
6
The two eigenstates have masses
M
2
F
2,1
=

2
2
_
f
g
(12(x
2
+ 3)
f
g
+ (3 x
2
)
A
f

) [
f
g
(12
f
g
+
A
f

)[

9 + 42x
2
+x
4
_
(25)
The requirement of a positive denite spectrum (m
2
F
1
> 0) gives the constraint
y
1
< y =
g A
f
f
9 +x
2
4x
2
< y
2
, (26)
where
y
1,2

9 +x
2
8x
2
_
21 +x
2

9 + 42x
2
+x
4
_
(27)
or
1
2
_
21 +x
2

9 + 42x
2
+x
4
_
<
g A
f
f
<
1
2
_
21 +x
2
+

9 + 42x
2
+x
4
_
(28)
and requiring positive diagonal elements implies also
y < y
3

(x
2
+ 3) (x
2
+ 9)
x
2
(29)
3.2 Flatino spectrum
From the superpotential W
aton
we can directly extract also the atinos mass matrix whose
eigenvalues are
M
2

F
2,1
=
9
4
M
2

y x
2
[x
2
+ 2

x
2
+ 1] = 9
f
2
g
2

2
[x
2
+ 2

x
2
+ 1] (30)
The eigenstates

F
1
,

F
2
are related to the avor states

P,

Q by

F
1
= cos

P + sin

Q

F
2
= sin

P + cos

Q (31)
where tan 2 = x,
A parameter space analysis indicates that we have always M
F
1
2M

F
1
. This automat-
ically forbids the decay of F
1
to atinos leaving open only the decay into atinos of the
heavier F
2
and

atons.
4 Interactions between atons
Now that we know the general mass matrix structure of the atons and atinos, common
both to the DFSZ and KSVZ models, we can start analyzing the various decay rate between
7
atonic elds.
3
We begin with the decay channels induced by the kinetic term and the
superpotential W
aton
, which are common to the KSVZ and the DFSZ models.
4.1 Derivative and cubic interaction terms between atons
The aton interaction terms with at least one derivative are given by the lagrangian
L

=
2 v
P
P +P
2
2v
2
P
[
v
2
P
F
2
PQ
(a)
2
+ 9
v
2
Q
F
2
PQ
(

)
2
+ 6
v
Q
v
P
F
2
PQ

a] +
2 v
Q
Q+Q
2
2v
2
Q
[9
v
2
Q
F
2
PQ
(a)
2
+
v
2
P
F
2
PQ
(

)
2
6
v
Q
v
P
F
2
PQ

a] (33)
From this expression we can extract the following terms expressed in mass eigenstates.
The trilinear derivative interactions with no axions
(

)
2
1
F
PQ
x

x
2
+ 9
[
_
9 sin +x
3
cos
_
F
1
+
_
9 cos +x
3
sin
_
F
2
] , (34)
The trilinear derivative interactions with only one axion
L
F
i

a
=

a
6
F
PQ

x
2
+ 9
[(cos x sin ) F
2
(sin +x cos ) F
1
] , (35)
The trilinear derivative interactions with two axions
L
F
i
aa
= [

a[
2
1
F
PQ

x
2
+ 9
((9 cos x sin ) F
1
+ (9 sin +x cos ) F
2
) . (36)
All the above derivative interactions can be transformed in scalar interactions if we are
working at tree level and with on-shell external particles

1
(

2
) (

3
) =
1
2
_
M
2

1
M
2

2
M
2

3
_

3
(37)
The cubic interactions come also from the atonic superpotential and the soft terms
L

3 =
_
9
2
f
2
v
P
v
2
Q
M
2
Pl
+
5
2
f
2
v
3
P
M
2
Pl

M
2

v
Q
x
v
2
P
(x
2
+ 9)
_
P
3
+
_
27
2
f
2
v
2
P
v
Q
M
2
Pl
3
M
2

x
v
P
(x
2
+ 9)
_
P
2
Q
+
_
9
2
f
2
v
3
P
M
2
Pl
_
PQ
2
+
_
3f
4
A
f
F
2
PQ
M
Pl
v
Q
_
P

+
_
f
4
A
f
F
2
PQ
v
P
M
Pl
v
2
Q
_
Q

. (38)
3
The two body decay rate is given by the expression
(i f
1
f
2
) =
1
16
S[P[
M
2
i
_
1
1
d cos A(E
1
= E
2
, cos ) (32)
where S=1 if f
1
,= f
2
and S = 1/2 is f
1
= f
2
; E
1
= E
2
=
_
M
2
i
M
2
f1
M
2
f2
_
/2M
i
, [P[ =
1/ (2M
i
)
_
_
M
2
i
(M
f1
+M
f2
)
2
__
M
2
i
(M
f1
M
f2
)
2
_
and A(..) is the square amplitude of the decay.
8
In the mass basis, we get
L
F
2
F
1
F
1
=
_
3
M
2

sin
F
PQ
x

9 +x
2
_
2x cos
2
+x sin
2
+ cos sin
_
+
6 f
2

x
2
+ 9
g
2
xF
PQ
_
18 cos
2
sin x + 9 sin
3
x + 3 cos
3
x
2
cos sin
2
(9 +x
2
)
__
F
2
1
F
2

A
F
2
F
1
F
1
2
F
2
1
F
2
(39)
For the full trilinear F
2

interaction, we have to add up the terms in Eqs. (34) and (38)

to obtain
L
F
2

=
M
2
F
2
2x

9 +x
2
F
PQ
_
(3 sin +x cos )
_
x
2
+ 9
_
M
2

M
2
F
2
+ (40)
_
9 cos + sin x
3
_
_
1 2
M
2

M
2
F
2
__
F
2

A
F
2

2

2
F
2
Collecting the above formulae one nds the decay rates among atons and axions
(F
2
aa) =
1
32
M
3
F
2
F
2
PQ
(x
2
+ 9)
(x cos + 9 sin)
2
(F
1
aa) =
1
32
M
3
F
1
F
2
PQ
(x
2
+ 9)
(x sin + 9 cos )
2
(F
2
F
1
F
1
) =
1
32M
F
2

_
1 4
M
2
F
1
M
2
F
2
[A
F
2
F
1
F
1
[
2
(F
2

) =
1
32M
F
2

_
1 4
M
2

M
2
F
2
[A
F
2

[
2
(41)
(F
2
a

) =
1
16
M
3
F
2
F
2
PQ
(x
2
+ 9)
_
1
M
2

M
2
F
2
_
3
(3 cos 3x sin )
2
(

aF
2
) = (F
2
a

M
2

M
2
F
2
(

aF
1
) = (F
1
a

M
2

M
2
F
1
Energy conservation will of course forbid some of these reactions, depending on the aton
masses. As M
F
1
< M

the channels F
1

and F
1

a are always forbidden.

4.2 Interaction terms between atons and atinos
The trilinear Lagrangian terms responsible for the decay of atons or atinos are
L

=
3f
2M
Pl
_
(v
P
Q+v
Q
P)

P

P +i
v
2
P
+ 3v
2
Q
F
PQ

P
5

P i2
v
P
v
Q
F
PQ
a

P
5

P
_
+
9
3f
2M
Pl
_
2Pv
P

P

Q + 2i
v
P
F
PQ
(3v
Q

+v
P
a)

P
5

Q
_
(42)
(the tilded elds are the fermionic superpartner of the respective P and Q scalars). Let us
denote the Yukawa couplings between the aton (or the axion) and the atinos in mass basis
by L
Y uk
= Y
ijk

F
j
(1,
5
)

F
k
/2 where
5
is taken for
i
= a,

. We nd from Eq. (42) the

following expressions for the Yukawa couplings
Y
F
1

F
1

F
1
=
6f

x
2
+ 9
gxF
PQ
[(x cos sin ) cos
2
x sin sin 2 ]
Y
F
1

F
2

F
2
=
6f

x
2
+ 9
gxF
PQ
[(x cos sin ) sin
2
+x sin sin 2 ]
Y
F
1

F
1

F
2
=
6f

x
2
+ 9
gxF
PQ
[x sin cos 2 +
1
2
(sin x cos ) sin 2 ]
Y
F
2

F
1

F
1
=
6f

x
2
+ 9
gxF
PQ
[(x sin + cos ) cos
2
+x cos sin 2 ]
Y
F
2

F
2

F
2
=
6f

x
2
+ 9
gxF
PQ
[(x sin + cos ) sin
2
x cos sin 2 ]
Y
F
2

F
1

F
2
=
6f

x
2
+ 9
gxF
PQ
[+x cos cos 2
1
2
(cos +x sin ) sin 2 ]
Y

F
1

F
1
=
6f
gxF
PQ
[
_
3 +x
2
_
cos
2
3x sin 2 ] (43)
Y

F
2

F
2
=
6f
gxF
PQ
[
_
3 +x
2
_
sin
2
+ 3x sin 2 ]
Y

F
1

F
2
=
6f
gxF
PQ
[3x cos 2 +
1
2
_
3 +x
2
_
sin 2 ]
Y
a

F
1

F
1
=
6f
gxF
PQ
[2x cos
2
x
2
sin 2 ]
Y
a

F
2

F
2
=
6f
gxF
PQ
[2x sin
2
+x
2
sin 2 ]
Y
a

F
1

F
2
=
6f
gxF
PQ
[x
2
cos 2 x sin 2 ]
From this we can extract the decay rates for F
i


F
j

F
k
,

F
j

F
k
, or

F
2


F
1
F
i
(

, a)

_
F
i


F
j

F
k
_
=
M
F
i
8
S
_
1
(M

F
j
+M

F
k
)
2
M
2
F
i
_
3
2
_
1
(M

F
j
M

F
k
)
2
M
2
F
i
_
1
2
Y
2
F
i

F
j

F
k

F
j

F
k
_
=
M

8
S
_
1
(M

F
j
+M

F
k
)
2
M
2

_
1
2
_
1
(M

F
j
M

F
k
)
2
M
2

_
3
2
Y
2

F
j

F
k
(44)
10

_

F
2


F
1
F
i
_
=
M

F
2
16
_
_
1
(M

F
1
+M
F
i
)
2
M
2

F
2
_
_
1
2
_
_
1
(M

F
1
M
F
i
)
2
M
2

F
2
_
_
1
2
_
_
_
1 +
M

F
1
M

F
2
_
2

M
2
F
i
M
2

F
2
_
_
Y
2
F
i

F
1

F
2

_

F
2


F
1

_
=
M

F
2
16
_
_
1
(M

F
1
+M

)
2
M
2

F
2
_
_
1
2
_
_
1
(M

F
1
M

)
2
M
2

F
2
_
_
1
2
_
_
_
1
M

F
1
M

F
2
_
2

M
2

M
2

F
2
_
_
Y
2

F
1

F
2

_

F
2


F
1
a
_
=
M

F
2
16
_
_
1
M
2

F
1
M
2

F
2
_
_
_
1
M

F
1
M

F
2
_
2
Y
2
a

F
1

F
2
where S is a symmetric factor (1/2 for identical nal states or otherwise 1).
5 Interaction of atons and atinos with matter elds
Now we study the interactions of the atons with matter and supermatter, specied by
Eq. (7) for the KSVZ case and by Eq. (8) for the DSFZ case. Through these interactions the
over-produced atons or atinos could decay into ordinary matter while the number of the
decay produced axions are suciently suppressed satisfying the nucleosynthesis limit (10).
5.1 KSVZ model: Interactions between atons and gluons
We begin with the hadronic models in which the only decay mode available for the atons is
into two gluons coming from the anomaly (when the space phase will be available, we have
to take into account also the decay into massive gluinos, in this discussion we neglect such
a possibility). The respective one loop corrected decay rates are
(F
1
g +g) =

2
S
(M
F
1
)
72
3
N
2
E
M
3
F
1
x
2
F
2
PQ
_
x
2
+ 9
_
sin
2

_
1 +
95
4

S
(M
F
1
)

_
(45)
and
(F
2
g +g) =

2
S
(M
F
2
)
72
3
N
2
E
M
3
F
2
x
2
F
2
PQ
_
x
2
+ 9
_
cos
2

_
1 +
95
4

S
(M
F
2
)

_
(46)
where N
E
is the total number of the superheavy exotic quark elds (M
E
= h
E
v
P
M
F
i
).
We do not consider the atino decay into a gluon and a gluino which will be irrelevant
for our discussion.
5.2 DFSZ model: Interactions between atons/atinos and ordi-
nary matter
The decay properties of the atons in the DFSZ models involves the direct interactions
between atons and ordinary matter and supermatter In general the interaction between
atons and Higgs elds are quite interesting due to the fact that this two sectors, after the
spontaneous breaking of the PQ and the EW symmetry, mix together. We notice that the
11
Peccei-Quinn symmetry prevents the introduction of a SUSY invariant mass term H
1
H
2
,
solving automatically the so called mass problem as mentioned before.
Let us start by writing the Higgs-aton potential
V (H, ) = [H
1
[
2
_
_
m
2
H
1
+

Q
M
Pl

2
_
_
+ [H
2
[
2
_
_
m
2
H
2
+

Q
M
Pl

2
_
_
+
_
gH
1
H
2
_
A
g

Q
M
Pl
+ 3f

2
P
[
Q
[
2
M
2
Pl
+f

2
P
[
P
[
2
M
2
Pl
_
+ c.c.
_
(47)
+
1
8
_
g
2
+g

2
_ _
[H
1
[
2
[H
2
[
2
_
2
.
When the elds
P,Q
get vevs, the m
2
3
H
1
H
2
mass term is generated dynamically. The size
of such a term is xed by
m
2
3
=
_
A
g
+
f
g

_
x
2
+ 3
_
_
(48)
In the limit [m
2
3
[ M
2
W
the masses of the pseudoscalar A
0
, of the CP even scalar Higgs eld
H
0
and of the charged Higgs elds H

are almost degenerate

m
2
A
0
,H
0
,H

m
2
3
sin cos
(49)
so from the constraint of positivity of such a masses we get
A
g

+
f
g
(x
2
+ 3) 0 (50)
In such a limit we also know that the mass eigenstate of the CP even electroweak sector
H
0
, h
0
and of the CP odd one A
0
, G
0
are
H
0
= sin h
0
1
+ cos h
0
2
h
0
= cos h
0
1
+ sin h
0
2
(51)
A
0
= sin A
0
1
+ cos A
0
2
G
0
= cos A
0
1
sin A
0
2
where H
1
=
1

2
(v
1
+h
0
1
+iA
0
1
) and H
2
=
1

2
(v
2
+h
0
2
+iA
0
2
) are the gauge eigenstates and
tan = v
2
/v
1
. To allow the aton decay into A
0
, we want it to be light so that small tan
is preferred in our discussion. Hereafter we will take tan = 1.
From Eq. (48), we nd
V
Fhh
=
1
2

2
_
h
02
1
+h
02
2
+A
02
1
+A
02
2
_
_
P
v
P
+
Q
v
Q
_
+
1
2
__
h
0
1
h
0
2
A
0
1
A
0
2
_
+i
_
h
0
1
A
0
2
+h
0
2
A
0
1
__
(52)
_
A
g

_
P
v
P
+
Q
v
Q
i
x
2
+ 3
xF
PQ

_
+ 6
f
g

2
_
P
v
P
+
Q
v
Q
+i
3
xF
PQ

_
+ x
2
f
g

2
_
4
P
v
P
+i
6
xF
PQ

__
+c.c.
12
It is then simple manner to get the decay rates for the kinematically more favorable decay
channels F
1,2
h
0
h
0
and h
0
A
0

_
F
1
h
0
h
0
_
=
M
3
F
1
32F
2
PQ
(x
2
+ 9)
16 x
2

4
M
4
F
1
_
1
4M
2
h
0
M
2
F
1
_
1/2
[A
F
1
hh
[
2

_
F
2
h
0
h
0
_
=
M
3
F
2
32F
2
PQ
(x
2
+ 9)
16 x
2

4
M
4
F
2
_
1
4M
2
h
0
M
2
F
2
_
1/2
[A
F
2
hh
[
2
(53)

h
0
A
0
_
=
M
3

16F
2
PQ

4
M
4

_
1
(M
h
0 M
A
0)
2
M
2

_
1/2
_
1
(M
h
0 +M
A
0)
2
M
2

_
1/2
[A
hA
[
2
where
A
F
1
hh
= sin 2[
_
A
g

+ 6
f
g
_
(x cos sin ) 4x
2
f
g
sin ] + 2 (x cos sin )
A
F
2
hh
= sin 2[
_
A
g

+ 6
f
g
_
(x sin + cos ) + 4x
2
f
g
cos ] + 2 (x sin + cos )
A

hA
=
_
A
g

6
f
g
_
(x
2
+ 3)
x
(54)
If atons produce a large number of atinos, atino decay into axions has to has to be
suppressed as well. Primary importance is the production of the lightest atino

F
1
which
cannot decay into other atons (axions) or atinos. The atino decay into ordinary particles
comes from the superpotential W =
g
M
Pl

H
1

H
2

P

Q. We nd that the atino decay into a
Higgs and a Higgsino (more precisely, the lightest neutralino
1
) has the rate;

F
i

1
h
0
_
=
M
3

F
i
8F
2
PQ

2
M
2

F
i
_
x
2
+ 9
_
2
C
2

F
i
_
_
1
(M

1
+M
h
0)
2
M
2

F
i
_
_
1/2
_
_
_
1 +
M

1
M

F
i
_
2

M
2
h
0
M
2

F
i
_
_
where C

F
1
= (sin +x
1
cos )N

1
and C

F
2
= (cos x
1
sin )N

1
. Here N

1
denotes the
fraction of lightest neutralino in Higgsinos.
Let us now consider the aton decay into ordinary fermions or sfermions. The mixing
terms between aton and Higgs elds allow a direct tree level coupling (after full mass matrix
diagonalization) between the usual fermions and atons. Parameterizing such a mixing with
the parameter
FH
the eective aton-fermion interaction is h
f

FH
so that the rate of decay
is

_
F
i
f +

f
_
= N
c
h
2
f

2
FH
16
M
F
i

_
_
1 4
m
2
f
M
2
F
i
_
3
(55)
where N
c
is a color factor for the fermion f. Since
FH

_
v
EW
F
PQ
_
, for light fermions
(2m
f
< M
F
)
_
F
i
f +

f
_
/(F
i
a +a) h
2
f
v
2
EW
/M
2
F
i
m
2
f
/M
2
F
i
which is less than
one. Therefore, the rate of the aton decay into ordinary fermion cannot be made suciently
larger than that into axions.
13
For the coupling between sfermions and atons, we have two contributions. One is a
direct coupling coming from the scalar potential
V
F

f

f
=

F
PQ

x
2
+ 9
x
v
1

2
_
h
d
tan

D
L

D
R

+h
e
tan

E
L

E
R

+h
u

U
L

U
R

_
(56)
(F
1
(x cos sin ) +F
2
(cos +x sin )) +h.c.
where

D

denote down-type squark, etc.

The other arises from an indirect coupling induced by the mixing between Higgs and
aton elds as for the fermion case. Taking in consideration the cubic soft A-terms we nd
V
eff
= h
d
A
d

F
i
H
1
F
i

D
L

D
R
+h
e
A
e

F
i
H
1
F
i

E
L

E
R
+h
u
A
u

F
i
H
2
F
i

U
L

U
R
(57)
so that eectively we have couplings of the size
G
F sfermion
h
f
( +A
f
)
v
EW
F
PQ
(58)
Diagonalizing the sfermion mass matrix we can write

f
R

f
L
= a
11

f
1

f
1
+a
22

f
2

f
2
+a
12

f
1

f
2
(where a
ii
h
f
so for h
f
0 we have a
12
1). Considering the decay of the light aton
we get

_
F
1


f
i
+

f
j
_
N
c
G
2
F

f
64 M
F
1
a
2
ij

_
1 4
m
2

f
M
2
F
1
(59)
As observed in Ref. [15], the aton may decay eciently to two light stops as h
t
1 and
a
ij
1 and thus a large splitting between light and heavy stops helps increasing the aton
decay rate to light stops. This kind of mass splitting occurs also in the Higgs sector and
furthermore the light Higgs (h
o
) is usually substantially lighter than the heavy Higgs (H
o
) in
the minimal supersymmetric standard model. This should be contrasted to the case with the
mass splitting for stops which requires some adjustment in soft parameters. In this paper,
therefore, we will concentrate on the aton decay into Higgses as a dominant mode of aton
decays.
6 Parameter space analysis
Our task is now to nd the parameter space for which B
a
gets small enough. As a reference,
let us try to see if B
a
< 0.1 (corresponding to N

< 0.7) can be obtained imposing the

stronger condition that B
I
< 0.1 for each I = F
1,2
,

. We rst note that the decay rates

calculated in the previous sections are functions of the 4 variables x = v
P
/v
Q
, f/g, A
f
/
and A
g
/ disregarding their overall dependence on F
PQ
.
Before starting our discussion, let us make some remarks. We are dealing with two kind
of PQ models with a natural intermediate scale 1) the DFSZ and 2) the Hadronic model
(KSVZ).
14
i) They have a common aton potential and thus the same aton and atino spectrum.
But they have dierent interactions between atons and matter.
ii) The symmetries and parameter space constraints impose that the following decays
are forbidden

aa, F
1

a, F
1

, F
1


F

F.
Neglecting for the moment the model dependent aton-matter interactions, we have to
analyze the decays F
1,2
a a, F
2

a and the orthogonal

F
1,2
a plus the atons-
atinos interactions that through the chain Flaton Flatinos Flatino axion can
also generate a non negligible axionic density at nucleosynthesis time.
The decays of the atons into two axions doesnt have any phase space suppression. If we
choose =
1
(see below Eq. 22) then we can suppress the rate (F
1
a +a) only taking
x
2
= 18 and cos
1
[
x
2
=18
= 0.426 (giving for example (F
2
a +a) = 3.6
M
3
F
2
32F
2
PQ
). On the
other hand, if we choose =
2
we can suppress only (F
2
a +a) in the region x
2
= 18
and cos
2
[
x
2
=18
= 0.905 (giving for example (F
2
a +a) = 3.6
M
3
F
1
32F
2
PQ
).
The decay rate into a single axion has the phase space suppression constraint m
2
F
2
M
2

that translated in our parameters reads

g
f
A
f

12.
When the atons branching ratio to atinos becomes sizable we have also to impose
B
_
I

F
2
_
B
_

F
2
a

F
1
_
< 0.1 . (60)
with B
_
I

F
2
_
= 2 B
_
F
2


F
2

F
2
_
+B
_
F
2


F
2

F
1
_
+2 B
_

F
2

F
2
_
+B
_

F
2

F
1
_
and also the important requirement that

F
1
has to be heavier than the lightest neutralino
(
0
1
) and the light Higgs (h
o
) since it has the unique decay mode to a neutralino and a Higgs.
Assuming m

F
1
is much larger than the light Higgs mass, we impose a strong condition
m

F
1
> that translates into the approximate bound

f
g

>
1
3x
.
6.1 Parameter space of KSVZ models
As discussed already, the light aton F
1
can decay only into two gluons, so the simple
requirement B
1
F
1
= [F
1
gg]/[F
1
aa] 1 impose a quite strong constraint on the
parameter space. This ratio is given by
(F
1
g +g)
(F
1
a +a)
=

2
S
(M
F
1
)

2
4
9
N
2
E
_
sin (x
2
+ 9)
x (9 cos x sin )
_
2
_
1 +
95
4

S
(M
F
1
)

_
. (61)
In order to have a large number we can use only the x parameter and the number of exotic
quark ( to be as large as possible). For
S
(M
F
) 0.1, this ratio can be larger than 10
accepting a tuning of the x parameter as follows

18 0.04 < x <

18 + 0.04 for N
E
= 1,

18 0.3 < x <

18 + 0.3 for N
E
= 9. (62)
However, in such region we dont nd any solution in which, B
a
(F
2
) and B
a
(

) are less
than 0.1 at same time. Therefore the KSVZ model cannot give a satisfactory solution of the
15
nucleosynthesis bound unless some extra ne tuning on the initial densities of the atons
F
2
,

can be dynamically achieved.

6.2 Parameter space of DFSZ models
In the DFSZ model the number of the possible decay channels is much larger than the
other model and it is not an easy task to nd, in the four parameter space, some easy
understandable avaliable region. To be as independent as possible of the soft supersymmetry
breaking parameters we will try to make analytic computations on the rates of the aton
decays into Higgs particles, in particular into the lightest Higgs (h
0
) whose mass has an
automatic upper bound of 140 GeV [20]. We will concentrate on the region with
f
g
negative and

f
g

1 and x 1 which turns out to be required for B

a
< 0.1.
To open the decay channels of the atons into Higgs particles we have, in particular, to
impose M

> M
A
> 0 which requires (from now on we will use the unequal symbols as
strictly satised)
Ag

<

f
g

x
2
with

f
g
A
f

x
2
> 2

A
g

+
f
g
x
2

. (63)
For simplicity we divide into two regimes
[
A
g

[ < [
f
g
[ x
2
,

A
f

> 2 (64)
and

A
g

>

f
g

x
2
,

A
g
A
f

<
1
2
x
2

f
g

A
f

> 2 . (65)
Remember that the positivity of aton masses requires

A
f

< x
2

f
g

. (66)
Combining altogether we get

f
g

>
1
2 x
2
in the regime (64) and

f
g

>
2
x
2
in the regime
(65). Besides, if atino production rates are sizable, we also have to impose R

F
2
(

F
2

1
h
0
)/(

F
2
a

F
1
) > 10 and M

F
1
> M

1
+ M
h
0 to open the decay mode

F
1

1
h
0
. The
condition R

F
2
> 10 gives the restriction [f/g[ < 0.02 x N

1
and [
f
g
[ >
1
3 x
.
Then we study, in our limit, the constraints given by the conditions R

h
0
A
0
)/(

aF
1
) > 10 and R
F
i
(F
i
h
0
h
0
)/(F
i
aa) > 10. The ratios R are
R

1
144
_
g
f
_
2

2
A
2
f
_
A
g

6
f
g
_
2
R
F
1

1
4

4
M
4
F
1
(
A
g

2
f
g
x
2
+ 2)
2
(67)
R
F
2
10
3
x
4

4
M
4
F
2
(
A
g

+ 18
f
g
+ 2)
2
16
where M
2
F
1
[
f
g
A
f

[x
2

2
and M
2
F
2
12
f
2
g
2
x
2

2
for
A
f

< 12[
f
g
[, and M
F
1
M
F
2
for
A
f

>
12[
f
g
[. We can now divide the parameter space into four regions.
I)

Ag

>

f
g

x
2
, II) 2 <

Ag

<

f
g

x
2
, III)

f
g

<

Ag

< 2 , IV )

Ag

<

f
g

.
Depending on
A
f

< 12[
f
g
[ or
A
f

> 12[
f
g
[ we will have region a or region b.
We nd that all of the a regions are forbidden, and so is the IV
b
region. The constraints
for the regions are as follows.
I
b
x > 14, A
g
< 0, 12

f
g

<

A
f

1 <
1
2

A
f

<

f
g

x
2
2
<

A
g

< 2
_

f
g

x
2
2
_
2
.
II
b
x > 9, A
g
> 0,

f
g

< 3 10
2
2 <

A
f

<

f
g

x
2
, 2 <

A
g

<

f
g

x
2
.
III
b
A
g
> 0, 1 < 2

f
g

x
2
,
2

f
g

<

A
f

f
g

< 2 10
2
,

f
g

<

A
g

< 2 . (68)
To summarize, we nd that:
- In all cases, x has to be large (

> 10)
- In cases II and III,

f
g

has to be small (< 3 10

2
) but it has no useful upper bound in
case I. In all cases x
2

f
g

> 1.
- In all cases, [A
f
[ [A
g
[ [[ is a possibility.
7 Conclusions
We have explored the cosmology of a particularly attractive extension of the Standard Model,
which has a Peccei-Quinn symmetry broken only by two aton elds
P
and
Q
, character-
ized by a very large vev (10
1012
GeV) and a relatively small mass (10
23
GeV). These and
their superpartners generalize the saxion and axino, that in the non-at case are the only
elds with soft masses.
In contrast with more general models the density of dark matter axions can be estimated
with essentially no assumption about other sectors of the theory, assuming only that
P
has
a positive eective mass-squared in the early Universe. If
Q
also has a positive eective
mass-squared the axion is an excellent dark matter candidate. In the opposite case the axion
density is probably to low in this particular model.
Our main concern has been with a dierent, highly relativistic, axion population that is
produced by aton decay. We have calculated the rates for all relevant channels and examined
17
the constraint that the energy density of these axions does not upset the predictions of the
standard nucleosynthesis.
For the KSVZ model we nd that it is almost impossible to satisfy such a cosmological
bounds due to the fact that the only atonic decay channel in competition with the axionic
one is into gluons (through axial anomaly) which is too much constrained.
For the DFSZ model there are more decay channels. To evade complicated phase space
suppressions we concentrate on the Higgs decay products, as the mass of the lightest Higgs
has naturally a relatively low upper bound and the mass of the other Higgs and aton elds
are xed by the parameters of the atonic potential itself. We have quantied a portion of
the parameter space available showing the strength of this model.
An interesting question, lying beyond the present investigation, is whether the allowed
region of parameter space can be achieved in a supergravity model with universal soft pa-
rameters.
References
[1] R. D. Peccei and H. R. Quinn, Phys. Rev. Lett. 38, 77 (191440); Phys. Rev. D16, 77
(191791).
[2] For a review see, J. E. Kim, Phys. Rep. 150, 87 (191).
[3] J. E. Kim, Phys. Rev. Lett. 43, 79 (19103); M. A. Shifman, V. I. Vainstein and V. I.
Zakharov, Nucl. Phys. 166, 80 (19493).
[4] A. R. Zhitniskii, Sov. J. Nucl. Phys. 31, 260 (1980); M. Dine, W. Fischler and M.
Srednichi, Phys. Lett. 104, 81 (19199).
[5] R. A. Battye and E. P. S. Shellard, astro-ph/9808221 To be published in the proceedings
of 5th IFT Workshop: Axions, Gainesville, FL, 13-15 Mar 1998. For a dissenting view
see C. Hagmann, S. Chang and P. Sikivie, hep-ph/9807428.
[6] E. J. Chun and A. Lukas, Phys. Lett. 357, 95 (1943).
[7] J. E. Kim, Phys. Lett. 136, 84 (19378).
[8] P. Moxhay and K. Yamamoto, Phys. Lett. 151, 85 (19363).
[9] G. Lazarides, C. Panagiotakopoulos and Q. Sha, Phys. Rev. Lett. 56, 86 (19432).
[10] J. A. Casas, and G. G. Ross, Phys. Lett. 192, 87 (19119)
[11] J. Hisano, H. Murayama and T. Yanagida, Phys. Lett. 291, 92 (19263).
[12] H. Murayama, H. Suzuki and T. Yanagida, Phys. Lett. 291, 92 (19418).
[13] D. H. Lyth and E. D. Stewart, Phys. Rev. D53, 96 (191784)
18
[14] J. E. Kim and H. P. Nilles, Phys. Lett. 138, 84 (191)50; E. J. Chun, J. E. Kim and H.
P. Nilles, Nucl. Phys. 370, 92 (19105).
[15] K. Choi, E. J. Chun and J. E. Kim, Phys. Lett. 403, 97 (19209).
[16] M. S. Turner, Phys. Rev. D32, 85 (19843).
[17] D. H. Lyth, Phys. Lett. 275, 92 (19279).
[18] D. H. Lyth and E. D. Stewart, Phys. Rev. DD46, 92 (19532).
[19] D. H. Lyth Phys. Rev. D48, 93 (194523).
[20] A. Casas, J. R. Espinosa, M. Quiros, A. Riotto Nucl. Phys. 436, 95 (193); Erratum-ibid.
B439, 466 (1995).
19