IC/95/29 TUM-HEP 215/95 hep-ph/9503233

INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS

AXINO MASS IN SUPERGRAVITY MODELS

INTERNATIONAL ATOMIC ENERGY AGENCY

E..T. Chun and A. Lukas

UNITED NATIONS EDUCATIONAL, SCIENTIFIC AND CULTURAL ORGANIZATION

MIRAMARE-TRIESTE

T"

r f

IC/95/29 TUM-HEP 215/95 hep-ph/9503233 International Atomic Energy Agency

and United Nations Educational Scientific- and Cultural Organization INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS

The implications of axions have been examined oxtensively since their existence was suggested by an attractive mechanism for resolving the strong CP problem [I. 2, 3]. Even though axions are very weakly interacting, their astrophysical and eosmological effects are strong enough to narrow down the window of the Peccei-Quinn scale //.g to 10 in GeV < jpq < ldu GeV [5], On the same footing the axino as the supersymmrtric partner of the axion can play an important role in astrophysics and cosmology [U]. An interesting feature is that axinos may receive a mass of order keV which would render them a good candidate for warm dark matter. If axinos are heavier than a few keV they

AXINO MASS IN SUPERGRAVITY MODELS

have to decay fast enough not to upset any standard prediction of big bang cosmologyGiven the weakness of their interactions, a constraint on their lifetime put a rather severe limit on the lower bound of their mass. Therefore it is very important to know the axino

E. .1. Chun 1 International Centre for Theoretical Phvsics, Trieste, Italy

ami

mass in discussing the cosinologicai implications of supersynnnetric axion models. In global supcrsymmetrv (SUSY) the < alculatiou of the axino m<is.s was performed in refs. [7. 8]. In this paper, we will provide the computations in model?, with local supersymmetry (supcrgravity). Some partial results have been obtained in refs. [!), 10]. In the case of spontaneously broken global SUSY t he axino mass is of the order m'*,.,/ fpq ~ keV where m : ^ ( S ITeV) is taken to be the global SUSY breaking scale [7, H]. On the contrary, in the context of supergravitv. the axino mass can be of order niA/-2 as lirst noticed in ref. [9]. Soon after this if vvas realized that the axino mass is truly model dependent and the global SUSY value ni\rljfnj also including radiative corrections. The prime motivation for supergravity is well known. Realistic supersyminetrie generalizations of the standard model are based on local SUSY spontaneously broken in a so called hidden sector at a mass scale of order W s ~ K)"GeV [12]. The induced SUSY breaking scale in the observable sector is determined by a value of the order of the gravitino mass in^'-i -- M^jMp where ,!/; is the Planck scale. Axionic extensions of the minimal siipersytiunctric standard model (MSSM) inevitably incorporate an extra sector which provides spontaneous breaking of the Percei Quinn U(\) symmetry at the scale fi'Q, This sector (PQ sector) is considered as a part of the observable sector. In the framework of effective supergravity theories with a Lagrangian composed out of a global SUSY part and soft terms the hidden sector dependences are encoded in the soft terms. We will rely mostly on this effective approach as it makes the calculations tractable. may be obtained in supergrnvity models as well [10]. We will extend those results in a generic treatment of supergravity models

A. LukasJ Physik Institut, Teehniseho Universitiit Miinrhen, D-857'17 Garching, Germany and Max Planck Institut i'iir Physik, Werner Heisenberg lustitut, P. O. Box 41) 12 12, Munich, Germany.

ABSTRACT We analyze Ihe mass of the axino, the fennioiiic superpartuer of the axion, in general supergravity models incorporating a Pcccei Quiim symmetry mid determine the cosrnologieal constraints on this mass. In particular, we derive a simple criterion to identify models with an LSP axino which has a mass of O(nifu-,( fpq) = O(keV) atid can serve as a candidate for (warm) dark matter. We point out that such models have very special properties and in addition, the small axino mass has to l>e protected against radiative corrections by demanding small couplings in tin- Pecrei Quimi sector. Geuerieally. we find an axino mass of order J3/a. Such masses air constrained by the requirement of an axino decay which occurs before the decoupling of the. ordinary LSP. Especially, for a large Peceei Quinu scale fPQ > ]()" GeV this constraint might be difficult to fulfill.

MIRAMARE

TRIESTE

A color anomaly in the PQ sector can be introduced in two ways. The fields S in this sector can be coupled to the standard Higgs doublets /f,, H? of the MSSM like (DFSZ axion) [2, 4] or to new heavy quarks Q,, Q-2 like gSQ,Q2 gSH^Hi (KSVZ axion) [1]. Our

March WJ5

'E mail:chuin<*iet|>.triestc.it E mailialukasidpliysik.tu-mucncSicn.iie

analysis of axino mass will lie concerned with the tree level result in the effective theory which is obtained after breaking the PQ symmetry and SUSY and should therefore not

depend on which implementation of the axion is chosen. We will find that this mass crucially depends on the structure of the 1'Q-sector as well as on the hidden sector. It should, however, he mentioned that in addition the above couplings and the corresponding trilinear soft terms lead to a one-loop radiative mass of order m^mp 'Ag2Aniv,/l(inz (A is the trilinear soft, coupling) [11. 9]. For the DFSZ axion this contribution is clearly negligible since the coupling g has to be quite small (y ~ 10' 7 ) in order lo generate reasonable Higgs masses. In the KSVZ-implementation there is no such restriction on g and the above one loop contribution can modify the result for tin- axino mass to be obtained below. An interesting observation is that the axino mass depends on whether the PQ sector admits additional accidental zero modes on the global SUSY level. As a general statement we find thni the axiuo muss cannot lie bigger thiin Otiirlyj) if the axion niodc is the onlv global zero mode in Ilie PQ sector. Lagraugian approach. The PQ secmr consists of an arbitrary number ul singlets S" wit h charge r/,, under the PQ symmetry and provides its spontaneous breaking at the scale /,.y = (5Z,,i/,'||i'"|'!)1'a. where ?'" = {.*"). The jixioii multiplct <I' is given bv 'I> Tl: i}at'"S" i frqIts component field content leads < ~ {a + in,a) with the axion <:. the saxion ,s- ami the axino a. [ > Supersynimclrv is broken mainly by the hidden sector with siuglel liclds Z' To simplify the argument we assume minimal kinetic UTin for the PQ lields as well as tor the liiddcli sector fields. The scalar potential reads In order to see this, it is useiul to analyze the full supergravitv Lagraugian. Later, we will confirm I his result by using the eflective

since {Wt) ~ Mpm3/2

maximally arid (H;,) S fj,Q.

Therefore the condition (Va) - 0 we also have (G*) =

gives (Ga) ~ JPQ- In the axion direction G* = EaW^Ga/fPQ,

(E n <laVaK,,[f,>q) = Y.a 'hV'"^I fi'Q - fi'Q- Since the absence of accidental zero modes is required in the PQ sector one finds that (6',,) ~ JFQ for all the fields S". It is now straightforward to estimate the maximal order of the axino mass by using the fermiou mass matrix in supergravity models. MtJ= Along the axino direction, Gn Gb Mpf:\\i{G/2M'p)\G,li, + GaGi,/MP].

(5) Hence the

FVom the ('(1) invariance of (I, we gel Yi,,<lit''<'(Gni,) = fl!,{i'b (Gj,)) S axiuo mass is niaximally of order ra;i,-;.

The above order of magnitude estimation is indeed insensitive to the specific forms of the kinetic term or the superpotential as long as the hidden sector fields have only non renonmilizable couplings ID the oliservable se(t"i". This hapjieus because higher power terms in A' or W are naturally suppressed by powers of Mp which renders their contribution negligible. Therefore we conclude that the a.xino mass in general siipergravity models is at most of the order m:i/2 if no other zero mode than the axion is present in the PQ sector. On the other hand if there arc extra zero modes we are not able to constrain further the order of (G,,) for a zero mode direct ion <i so that the above argumentation breaks down, hi fact, axino masses ;> m;t;.L> ale possible in those models ;is we will sec below. We will now calculate the actual value of the axino mass relying on the effective supergravity Lagraiigian with arbitrary soft terms. This allows to consider axino masses down to O{m\r2lfrq) where the next to leading order in the I/A/;.-expansion of supergravity

w h e r e G = I\ + M'f, In [ U ' / A / ^ p and IV - If ($") + \V[Z'). b r e a k i n g implies {6",) ~

Spontaneous supersvinmetry ~

A/,, for some i a n d generation of a g r a v i t i n o mass. iii:ir2

becomes important. The superpotenl ial \Y (if the PQ sector is expanded as

Mp expjfr"/2.1 If.]. We also have to admit the possibility thai (G,,) '* Mr

for some <i. We

now took ul I he iiumtui/.ation condition (V,) - tl In esliinate t h e axino mass. Vanishing of t h e cosiuoiogicnl constant ( I ' ) = It is assumed- T h e n we have

ir = /,,.s-" i- -rL,.S"s'' i-

-jnhrs"sbsr

(6)

(V,,) = (G,J;I + G,I,G; +G:)If we take a massive mode 5" in the PQ sector. (G,,i,) is dominated by

and a d e p a r t u r e from s t a n d a r d soft t e r m s is encoded in

N = djasa + ^Uf
Then the scalar potential reads

since (W) ~ Mf.n>M2- I" afidition, we can estimate the maximal order of magnitude of
(4)

V = ifWOfiW + m.f.v<ir + ';)/2

W + (A - 3)IV + N + h.cJ

(8)

To minimize t h i s p o t e n t i a l we apply t h e following s t r a t e g y : T h e V E V s ?/' a r e split i n t o a global S U S Y value 7<" w i t h 0aW(u) = 0 a n d c o r r e c t i o n s iu a d u e t o t h e soft t e r m s .

= u" + w". Expanding daV around the global minimum (") results in

into its Planck (or GUT) scale value J and corrections .7^' arising from renormalization down to fyq. The generic value of the corrections can be roughly estimated as

A-l)M*t

+ <UMu>b

(9)

(14)

M,

+(1
with the global mass matrix (10) and the definitions /. = m,n.ln + mfil,, , ./ = AUu'1 + ((>,,.V}(//) . with appropriate combinations Xa. X'u of the superpotential couplings. eq. (11) Jf is naively of the order /;. y . ino mass to be much smaller than vi:i/2 I* that Jf = 0.

(15) According to

Therefore a necessary condition for the ax-

In general this condition implies a relation between the structure of the superpotential

(ID

and the soft terms. It is instructive to analyze this relation for a certain subclass of models, namely those with independent soft coupling A.B.C for the triluieai, bilinear and linear terms in the superpotential. respectively. A computation leads lo

Can- should he taken on I he choice o! (it"). With any global minimum also a rotation a cxp(r/,,;)n" with : = .r h ill undri the couiplexitied fVy(l) leads to such a mini mum. Despite the y dependent part, of this symmetry which is clearly present in the full theory the r dependent part (present, because of the holoniorphy of the superpotential) is broken by the soft terms. Therefore an appropriate tixinjf for the .r dependent, part of the symmetry should be applied such that I lie j;lnbal minimum comes close to t.hc local values ('") resulting in small expansion cocfiicients ("). Such a fixing is provided by the condition wn = Y.u'la '' "/frQ
lt lll

.if = (B - C)M,,hi,u + -y(A - 2B f C)fnll,n'V .

(l(i)

Depending on the properties of the superpotenlial (and assuming that at least one coupling / is nonzero to force the symmetry breaking) several cases can be distinguished : fai.'t''- /("'''''' Mni,iih /
:

Tl"'" /,',"' = (I 'I' "! only if A = B - C. No special

= " implying that the correction in the global axion

property of the superpoteutial is required. .f,,i"'' = I'. Mu 4- " : Then ./,(,'" = i) if and only if A = C. A simple superpolcntial which fulfills this requirement is e. g. \Y \(SS' - /i2)Y since all /(, - 0. . fnlll.n''ii'~ = 0, Mn. ? 0 : Then .If = (I if and only if B = C.

direction (denoted by an index <v) vanishes. As can be expected the axino mass is expressible in terms of the corrections ("'") -

(Ma)a

= 1 ,,
(12) [, (!)). Lei n.s work in a basis with

Mn = (I, f,,i,u'' 4 0 : Then ,/,i'" = 0 if and only if A - IB + C = (I. As the only ones these models allow for the full standard pn1t.cn! B = A - 1, (' = A - 2. They possess, however, at least, one additional zero mode Yl,, n"S". An example is provided by the superpoteutial It" = \(SS' - Z2)Y - X'(Z - u)' 1 [III]. vanishes in any model.

These corrections have to be determined fnn

d i a g o . n l g l o b a l m a s r , m a t r i x M,,i, = .A/,/\,j,. W e d e n o t e m a s s i v e m o d e s w i t h i n d i c e s I.J.

and possible additional zero modes with indices ^ . 1 . ' ' Then an important observation is that the collections m, are basically detemiiued by the linear term \Mj\2u<1 ami /, in eq. (9). In zero mode directions the situation might be more complicated since \Mli\2ii'!3 can be small compared to other terms in eq. (!)). Taking this into account we conclude that the expression Observe that in [jarlicnlar for A = B (' I he expression Jf by the first term in eq. (L3). On tree level this means (17) If no additional zero mode is ])resent the axino mass is already completely determined

(M(l}n ~
JVQ

./<"> ^ 0

= O(m 3 / 2 )

We have therefi>re found a simple criterion to decide about the magnitude of the axino mass which for soft terms specified by the couplings A,B.C singles out. the particularly simple patterns listed above. We remark that the small axino mass in no scale models

gives the correct, order of the axino mass. For the moment we leave the values of wtl unspecified. Instead we concentrate on the hist term in eq. (13) and split the expression ./

observed in ref. [9] can also be understood In terms of our analysis since in those models A = B = C = 0. A full supergravity computation of the axino mass in no scale models shows that their tree-level mass even vanishes. Therefore the first line of eq. (17) has to be understood as a generic result which in certain special cases might be too large. In the first case of a light axino mass radiative corrections to the potential parameters can become important. Using eq. (14) this leads to a contributions of rn8 = ktn,,n (18)

supermultiplet arise after integrating out the heavy fields in the PQ-sector :

(I + -zs) -<) aO,,a + -0^0.1 JfQ V2 2

+ ta-j dILa - - / fPQ

(20)

where x = Y.,<livf/frQ IV -. \(SS'

I'1 some cases, 1 1 particular in a model with superpoteutial 1

- ;t J )V and universal scalar soft masses, x is zero at the Planck scale and

receives a contribution .r, ~ A 2 ln(jUp/7ry)/G4?r 2 when the RG improved potential at the PQ scale is considered [13]. Genetically, however, x is of order 1. In this case the self coupling becomes important since a saxion can decay into two axions faster than e.g., into two ghions. Decay produced axions do not heat the universe. Therefore the cosmological effect ot saxion decay is different from what lias been investigated assuming vanishing j [14. 15]. If .r is of order 1, a stronger bound on the saxion mass can be expected since the decay produced axions are not ihcnnalim.l but. red shifted away. The standard nucleosynthesis constrains the energy density of the universe due to this red shifted axions to be less than what is contributed by one species of neutrinos at the time of nucleosynthesis. This gives the constraint O..'/./) < T/> where 7'j> = 0.55jv,/> \/TA?r is the decay temperature of the saxion and I' = .'"''"^/^^/ry i f s decay rate. The relativistic degrees of freedom at Tp are counted by i/.j]. We get > 2.1 TeV (21)

To keep the order w,-, = ()(mj/.2/f,>c}) an upper bound on the couplings has to be required. For m-t/2 ^ 1()J GeV and / > Q ~ 10" GeV this implies A, -- ID"'1. A systematic way to avoid such small couplings is to consider no scale models. Since gangino masses are the only source of SUSY breaking in those models the standard model singlet fields in the PQ sector will not receive any radiative soft, terms. We see that without additional /.ero modes a complete answer can be given. In particular we recover the result m,-, O(ni.iri)If additional zero modes are present the situation becomes more complicated since the corrections , in the zero mode directions can become large. In addition we have ;.< to consider that the /.ero entry M,, of the imiss matrix receives a radiative contribution Mft = Xi'iifrQ with l':i in analogy t.o eq. (15) unless it is protected by an additional continuous or discrete symmetry. Let us discuss some relevant cases. First we discuss a model with .!}") = 0, e. g. a model of the last type in the above list for ,4, D.C type soft torms. Then a tier level axinn mass O^ur^.Jf,,Q) e. g. occurs for /. t> = (I. / ^ is not guaranteed as opposed to the ) results which causes ease without additional zero modes : If the terms in e<[. (!)) linear in ,., vanish (which = 0 or .4 = /J) n value tvj = O(lJ' '"
mLV

\100J

'

This bound on the saxiou mass which receives ;\ contribution O(m:,/-j) from scalar soft masses might be difficult to fulfill for large values of fi>Q, Cosmological implications of axinos were first discussed in ref. [6] assuming unbroken R parity. The axino mass can be constrained in two ways. First, the axino can be the lightest supersymnictnc particle (LSP). Then it should be lighter than a few keV in order not to overdose the universe. Otherwise, the axino decays into at least one I.SP composed out of the neutraliuos in the MSSM. In this case, the decay produced neutralinoK tend to overdominate the energy density of the universe. To avoid this the axino should be heavy enough to decay before the neutraliuos decouple. Considering the axino decay into photino plus photon, it was obtained that the axino mass should be bigger than a few Te.V [6]. Then, the axino decay into top quark and scalar top can be allowed. From this decay channel one finds a less restrictive bound m,i > 90 GeV I

an nxino mass given by rii^ (H(in-M-i/frij'\'"'"'.(/-J)A = (){l) the axino mass is shifted lo ;,-, = (J(/;;il;L.)

case a small axino mass

< m-.\p has to be .stabilized against radiative corrections. For superpoteiitial couplings Now we assume that ./^"' ^ 0. Then for values A = ()([) the linear term | A/,j | ^ H',J in eq. (9) will dominate and the axino mass is given by m,; = O(m;y,.,/l;i) . It, on the other hand, the couplings A are very small we can have ;; = 0(1,^ (19) ) leading

to an axino maws in,-, = O((TH^, i //-y) 1/ '' t ). We we that axino masses 3> ;n;1/2 are indeed possible. An example featuring all these aspects is the superpotential in the fourth entry of the above list for A,D,C-type. soft terms,

fnq/X, 40GeVy V 1 { ) U G c V / v

(22)

'

Now we turn to a discussion of the cosmological constraints on the masses of the axino and the saxion. We begin by noticing the fact that self couplings among the axion

Here Xt is the PQ charge of the top quark. We see that a wide range of axino masses between O(keV) and O(10 2 GeV) is excluded.

Let us now analyze how the above constraints modify if an inflationary expansion is taken into account, The decoupling temperature of axino or saxion is around thp range of the reheating temperature Tn ~ 1010 GeV which is the maximally allowed value to cure the gravitino problem in supergravity models [IB]. If decoupling of the axino occurs before1 inflation the primordial axino relics are diluted away. The above consideration, then, has to be applied to the regenerated population of axinos. We recall that the axino decoupling is determined by its interactions with gluinos, quarks and anti quarks via gluoti exchange [(>] The axino decouples at the temperature 10" GeV 0.1 HI GeV
11

mode of the globally supersymmetrie theory, i. e. Mu = 0 with the global mass matrix Af, the relation A - 2B + C = 0 which admits the standard pattern B = A-l,C is sufficient for the criterion to hold. We showed that in models without aelditional zero modes our criterion is sufficient, i. e. it. guarantees a tree level axino mass of nt most mf conditions (like e. %. A j= B for mexiels with an A,B,C - 0{m\j2lfpQ}. In the presence of other zero modes it serves as a good indication for such a small mass but aelditional pattern) are required to have = A-2

'"s" = 0(mlr,/flv).
From the cosinological point of view axiuo masses O(;;i^ / 2 ///.y) = O(keV) are very (23) attractive. In this ca.se the axino is the LSP and can contribute, a relevant purl of the mass in the universe as (warm) dark matter. Though models wil h such an axino mass can be constructed as we have seen they correspond lo very special points in the space spanncel by the supcrpotential and solt term parameters. Moreover, such small masses are not (24) stable under raeliative corrections arising from rcliormalization effects between A/,, and the PQ scale frQ- Taking these elfects into account the axino mass will be generie'ally given by ml" - <)(hti:tri) with k = A'-'hi( 1 V,'// / .g)/;i2,T i and a typical superpolential coupling A- The one loop contribution horn the cliarneleristic coupling of the axiuo to the Higgs h'e'ds or heavy quarks will be given by /",-,.i,,p St/'1 Am-^^f Hiir' which is only

The regenerated number di usily per entropy is given bv [Hi]

r~2x

Hi" GeV

Tlt

Depending on the lunge of the axino mass we can distinguish three cases as follows. First, for a stable axino. the constraint from overcfosurc gives the following loose bound on the axino mass in terms off,, : m a < 1G0 keV

frQ
10" GeV

K)" GeV

Tit

(25)

relevant m the KVSZ implementation of the axion. As the LSP the axiuo has to be lighter than a few keV which in turn puts a severe limit on the couplings A (and IJ in the KVSZ case), typically A i H)" 4 . If the decoupling temperature of the axiuo is larger than the reheating temperature of inflation the overelosure bound on the regenerated axino population is weakened resulting in a sejineiwhaf weaker bound on A, typically A i 10" 3 . In any case we conclude that a e-osmolivgical relevant LSP axino principle though possible in is not very likely to occur : Special models are needed anel in addition small

Second, an axino with a mass satisfying the lower bound in eq. (22) is still allowed by cosmology. Finally, for an unstable axiuo with mass between the estimations in eq. (25) ami it) eq. (22). one1 gels a hound on the reheating temperature by replacing the axino mass in eq. (25) by the mass of the usual LSP since (lie decay produe'ts of the nxiuo contain at least one LSP: I GeV (20)

couplings have to be chosen in order lo avoid a conflict with the overelosure bound. At this poinl it should be mentioned thai no scale supergravitv models can provide a naturally light axino [!)]. Those1 models are1 ehtiriictcrbicd by a special pattern of the ,soft terms ; The only IKHI vanishing soil t.emis ;it the Planck scale are gaugino musses and therefore A = B = C = I) at tree level. Applying the above statements a small axino mass results in this ca.se. In fact, a full supergravity calculation shows that the mass vanishes on tree level. Other soft terms for the gauge non singlets can be generated due to renormalizatioii group effects below the Planck scale in those models. Since the P Q sector consists of singlets their soft terms are not affected by renormalizatioii effee'ts and the axino (.saxion) remains massless [9j. Non vanishing masses can be however generated via the one loop contribution w.s,irp- In the DFSZ-implementation they are so small that cosmological effects of the axino ami the saxino are negligible. This is clearly different in the KVSZ-case. However, the limit on g necessary to keep the axino mass below the

This represents a quite stringent bound mi T/i. In I his letter we have analyzed the axino mass in general supergravity models and the eosmeilogieal constraints on such models. We have distinguished models with and without additional zero modes in the PQ sector. For I he latter we found the axino mass to be at most of Oiwyi). hi the context of an effective approae-h encoding supersymmetry breaking in sott terms we were able to derive a simple necessary criterion for a small tree level nxiKO mass -C m:l/2 given in terms of superpotential and soft term properties. For uniform t.rilinear, bilinear and linear .soft couplings A. D. C the criterion is always fulfilled for A = D = C whereas for A = D ^ C or ,4 ^ B = C additional properties of the superpotential had to be required. If the global vacuum () represents an additional zero

overclosure bound will be somewhat weakened with respect to the ordinary case since the trilinear coupling A for g$QiQi originates from radiative corrections.

[5] For a review see : E. Kolb and M. Turner. "The Early Universe'' (Addison Wesley, 1990). [0] K. Rajagopal, M. S. Turner and F. Wilczek, Nucl. Phys. B 358 (1991) 447. [7] K. Tamvakis and D. Wyler. Phys. Lett. B 112 (1982) 451. [8] ,1. F. \ieves, Phys. Rev. D 33 (1080) 1702. [!)] T. Goto and M. Yamaguclu, Phys. Lett. B 276 (1902) 10,1 [10} E. ,1. Chun, .1. E. Kim and H. \\ Nilles. Phys. Lett. B 287 (1992) 123. [11] P. Moxhay and K. Yamamoto, Phys. Lett. / U 5 1 (1985) 363. [12] H. P. Nilles. Phys. Hep. H O (1984) I. [13] E. .). Chun, 11. B. Kim and A. Lukas. Phys. Lit. [14] J. E. Kim. I'hi/s. Rev. Lilt 67 (ISJ91) U4(i.r,. [15] D, H. l.yth, Phys. Hrv. D 48 (1993) 4.723. [16] ,1. Ellis. .1. E. Kim and D. V. Nnnopoulos, Phys. Lett. B 145 (1984) 181. Z 328 (1994) 34G, ?

As a generic situation we consider models which do not fulfill our criterion for a small tree level axiiio mass and possess couplings A O{1). The axino mass in such models is given by nia = O(m 3 /j) (no additional zero modes) or rnA = O(m:(/i./A-) (additional zero modes). First of all this mass has to be linger than the mass of the ordinary LSP in the MSSM to allow for a decay of axinos. Secoml, these decays have to occur before the LSP decouples which translates into a liound on /,-, of typically ma i. 1IKI GeV for JPQ = 10" GeV if the decay channel into top and stop is possible. Otherwise an even stronger bound /,-, ~ TeV is required. We see tliat these generic models are significantly constrained liv cosmological considerations, however, a linal decision depends on details of the model like t lie exact iixino mass, the sfeniiioii masses, t lie PQ scale etc. Thereliiri: it might, he iiiiiTesi ing to hliidy models which put 1 1 it her i oust rail its mi these parameters 1 like e. g. siipcrsvijiiuetnc unified models uicorporotinf; an axion. F<n- a 1'Q scale in the upper hull'of the allowed range fpQ > It!" GeV the lower bounds on niA and the saxion mass m s (which both increase qvmdratieally with //>y) become very stringent and it might, be difficult to construct, viable models, hi this context, having axmo masses larger than TH.ty'2 in models with additional zero modes might be an interesting option. A c k n o w l e d g m e n t s This work was Mipporied liv the EC under contract No. SC] CT91-072!). We would like to thank K. Choi and D. Mainlliolakis lor stimulating discussions in the earlier development of this work, Iv.l.C, would like to thank Professor Abdus Salam, the Intel-national Atomic Energy Agency, 1'NESCO mid the International Centre for Theoretical Physics, Trieste, for support.

References
[1] .1. E. Kim. rh/s. Hrv Lit 43 (I!)T! 10:!: M. A Shihiiiin, V. 1. Vaiustein and V. 1.

Zakliavov, Nuvt, Phya. B 166 (1<)NO) ltK!3. [2] M. Dine, \V. Fisehler and M. Srednicki, Pints. Ltit. /? 104 (I'JWl) I'J'J: A. P. ZliiMiitskii. Sov. .1 Nud, Phys. 31 (1980) 20(1. [3] For reviews see : ,1. E. Kim, Phys. Rep. 150 (1987) 1; H. Y. Cheng. Phys. Rep. 158 (1988) 1; R. D. Peccei. n "CP Violation, 503. (4j For a natural snpersymmetric version of the DFSZ axion see : E. ,1. Chun, pirpi-mt. IC/94/358, hep-ph/9411290 (to appear in Phi/s. Lrll. D). ed. C. ,)m-[skoS (WSPC, Singapore, 1989)

10

11