Anda di halaman 1dari 20

Nuclear Physics B193 (1981) 297-316

North-Holland Publishing Company


UNCONVENTI ONAL MODEL OF NEUTRI NO MASSES*
Howard M. GEORGI
Lyman Laboratory of Physics, Harvard University, Cambridge, MA 02138, USA
Sheldon Lee GLASHOW 1 and Shmuel NUSSINOV 2
Center for Theoretical Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
Received 25 June 1981
Gelmini and Roncadelli have proposed a model of neutrino masses in which B - L symmetry
is spontaneously broken by a small vacuum expectation value of a Higgs triplet. We give an
exegesis of this model. We show that the massive neutrinos in this model cannot be cosmologically
relevant today and that conflicting analyses of double beta decay experiments can be reconciled.
1. Introducti on
The advent of gr and uni fi ed t heori es, whi ch predi ct obser vabl e vi ol at i ons of
ba r yon- numbe r conser vat i on, has di rect ed t heoret i cal at t ent i on t owards ot her
t aci t l y assumed but unpr oven symmet r i es: t he several l ept on number s, bar yon
numbe r mi nus l ept on number , vani shi ng neut r i no masses, col or SU(3), el ect ri c
charge, and even Lor ent z i nvar i ance. In this paper, we address t he quest i on of
neut r i no masses and t hei r phenomenol ogi cal i mport . We see ma ny accept abl e ways
to modi f y t he st andar d t heory so as to gener at e neut r i no masses. We pur sue in
dept h onl y one of t hese appr oaches, t he scheme r ecent l y pr oposed by Gel mi ni and
Roncadel l i [1] (which we encount er ed i ndependent l y but subsequent l y) . Her e,
neut r i no masses appear by vi rt ue of t he s pont aneous br eakdown of gl obal B- L
symmet r y and are accompani ed by t he necessary appear ance of a t rul y massless
Gol ds t one part i cl e. In this respect , t he model r esembl es an earl i er scheme of
Chi kashi ge, Mohapat r a, and Peccei [2], who i nt r oduced t he t er m maj or on to
descri be t he Gol ds t one boson. The GR model differs from t he CMP model in t hat
it does not i nvol ve t he exi st ence of heavy unobs er ved neut r i nos, whi l e' i t does
i nvol ve a much ri cher Hi ggs confi gurat i on. We fol l ow t he GR model wi t h no
compel l i ng ar gument t hat this is Nat ur e' s way to neut r i no mass, save t hat t he
phenomenol ogi cal sequel ae are most bi zar r e and fasci nat i ng.
* This research is supported in part by the National Science Foundation under contract no. PHY77-
22864 and the Department of Energy under DE-AC02-76-ERO-3069.
1 On leave from Harvard University.
2 On leave from Tel-Aviv University; supported in part by the Israel Academy of Sciences.
297
298 H.M. Georgi et al. / Neutrino masses
To set the stage, we present in sect. 2 a brief sampling of the experimental data
which suggest, but do not prove, the existence of non-vanishing neutrino masses.
In sect. 3 we discuss the special role of B- L symmet ry in the understanding of
neutrino masses. In sect. 4 we present the model of Gelmini and Roncadelli in
order to fix a notation and make this work self-contained.
The remaining sections are devoted to a discussion of the phenomenol ogi cal
implications of the model. In sect. 5 we discuss neutrino cosmology in terms of the
GR model, and show the impossibility of a neut ri no-domi nat ed universe. In sect.
6, severe constraints on the model are deduced from our knowledge of stellar
structure. In sect. 7 we derive bounds on the coupling of the maj oron to fermions.
In sect. 8 we discuss implications on the physics of ee - and e e collisions, and
upon decay of the Z .
2. Hi nt s of ne ut r i no mas s e s
Recent papers address the possibility that neutrinos have measurable masses,
and indeed, that effects due to these masses have already been seen. Here is a
sampling of current literature.
Endpoi nt spectra. A finite neutrino mass will affect the shape of the electron
spectrum from beta decay near its endpoint. In a recent study of tritium decay [3],
evidence for a non-zero neutrino mass has been claimed:
46 e V> m~ > 14 eV. (1)
Interpretation of this careful experiment is clouded by uncertainty due to atomic
corrections of the same magnitude as the alleged neutrino mass. Among experiments
which could verify this result is the study of the endpoint of inner bremsstrahlung
photons in radiative electron capture, a process now under consideration [4].
Neut ri nol ess doubl e beta decay. Doubl e beta decay is a rare but conventional
second-order weak process, which has been established to take place for at least
three different nuclides, 82Se, 128Te and 13Te. If the electron neutrino has a
non-vanishing Maj orana mass, their double beta decay can proceed in a neutrinoless
mode. In a recent reanalysis of the experimental data on Te double beta decay,
Doi et al. [5] conclude that the neutrinoless mode does take place, and indeed
dominates the double beta decay of 128Te. In terms of a Maj orana mass of the
electron neutrino, they conclude
My =30+ 1 eV, (2)
in agreement with (1). Ot her studies of double beta decay [6] yield results much
like (2), but in the form of upper limits to the neutrino mass.
Neut r i no oscillations. If neutrinos have mass, they may display the phenomenon
of neutrino oscillations. These have been and are being searched for. There are
two positive indications. One emerges from the apparent deficit of solar neutrinos,
H.M. Georgi et al. / Neutrino masses 299
an effect whi ch c o u l d be expl ai ned by neut r i no osci l l at i ons over an as t r onomi cal
unit*. Pr es ent dat a can admi t of al t er nat i ve expl anat i ons, and exper i ment s sensi t i ve
t o l ow- ener gy neut r i nos (e.g., t he gal l i um exper i ment ) must be done. If neut r i no
osci l l at i ons ar e t he expl anat i on of t he sol ar neut r i no pr obl em, smal l neut r i no masses
woul d suffice (at l east mi cr ovol t s) , but l arge mi xi ng angl es ar e r equi r ed. The second
posi t i ve i ndi cat i on ari ses f r om t he r eact or exper i ment s of Rei nes and Sobel [8].
Di f f er ences of neut r i no s quar ed masses of or der 1 (eV) 2 ar e i ndi cat ed, but t he
r esul t is not deci si ve.
Da r k ma s s i n t he uni ver s e. At var i ous l evel s, t her e ar e i ndi cat i ons of t he exi st ence
of subst ant i al quant i t i es of non- l umi nous mass in t he uni ver se. Neut r i nos wi t h
masses - 30 eV whi ch ar e rel i cs of t he bi g bang coul d compr i s e t he mass of gal act i c
hal oes [9]. Li ght er neut r i nos, wi t h masses of 6 + 3 eV, may pe r me a t e gal act i c cl ust ers
and expl ai n t hei r appar ent mi ssi ng mass [10]. In ei t her case, t hese expl anat i ons
woul d i mpl y t hat t he mass of t he uni ver se is domi na t e d by neut r i nos. I ndeed, a
st abl e rel i c neut r i no wi t h a mass - 100 eV woul d pr ovi de suffi ci ent mass to cl ose
t he uni ver se. Al t hough t he case for subst ant i al amount s of dar k mass in t he uni ver se
is firm, it is by no means cl ear t hai this mass must r esi de in t he f or m of massi ve
neut r i nos.
We have me nt i one d f our di st i nct phe nome nol ogi c a l mani f es t at i ons of non-
vani shi ng neut r i no masses. In a s t r ai ght f or war d model wi t h Ma j or a na neut r i no
masses, all of t hese effects can be pr es ent si mul t aneousl y. But t he gener al si t uat i on
is much mor e compl ex. For exampl e, if non- zer o masses ar e concl usi vel y obs er ved
in endpoi nt s pect r os copy, it does not f ol l ow t hat any of t he ot her effects must be
obs er vabl e. Osci l l at i ons may be s uppr es s ed by smal l or zer o angl es. The mass may
be a l e pt on- numbe r conser vi ng Di r ac mass, in whi ch case neut r i nol ess /3/3 decay
is f or bi dden. The massi ve neut r i nos may decay or anni hi l at e and t hus not sur vi ve
as dar k mass in t he pr es ent uni ver se.
The fi nanci al a ut onomy of var i ous por t i ons of Ha r va r d Uni ver s i t y is of t en put
as an aphor i s m; " Ea c h t ub on its own bot t om" . It seems equal l y apt to t he cur r ent
si t uat i on in neut r i no physi cs. Let t he cosmol ogi st , t he endpoi nt spect r oscopi st , t he
neut r i no osci l l at or , t he doubl e bet a and t he sol ar ent husi ast s each do his (or her)
t hi ng.
It is pr e ma t ur e to at t empt t o confi ne onesel f to one speci fi c t heor et i cal f r amewor k,
for this may easi l y l ead to conf usi on or cont r adi ct i on. In t he next sect i on, we
c omme nt on t he wi de var i et y of t heor et i cal opt i ons t hat ar e avai l abl e. Of cour se,
in t he r e ma i nde r of t hi s paper we shal l commi t our sel ves to an expl i ci t and
par t i cul ar l y unconvent i onal t heor et i cal model . It is not so much fai t h in t hi s model
t hat mot i vat es us, as it is t he unconvent i onal nat ur e of t he phenomenol ogy. For
exampl e, in this t heor y t he neut r i no mass as obs er ved in endpoi nt exper i ment s
needs not agr ee wi t h t he neut r i no mass as deduced f r om neut r i nol ess /3/3 decay.
* For a recent review of neutrino oscillations and solar neutrino flux measurements in the Davis
experiment, see ref. [7].
300 H.M. Georgi et al. / Neutrino masses
And, the oft -quot ed cosmologist limit upon the sum of the neutrino masses is entirely
irrelevant in this theory. Kilovolt neutrinos are perfectly acceptable, for they would
not have survived as relics of the big bang.
3. The importance of B- L
Take as a starting point the standard gauge theory of strong and electroweak
interactions based on G = SU(3) SU(2)x U(1). Three (or more) fermion families
are each 15-dimensional chirat representations of G, and there are no ri ght -handed
counterparts of neutrinos. A Higgs doublet implements the spont aneous breaking
of G to SU(3) x U(1). This system admits a u n i q u e current which is free of gauge-
gauge-current triangle anomalies, which is not a current of G, which commut es
with the elements of G, and which treats the families alike. A linear combination
of its generat or with the U(1) gauge generator yields the generator of B - L
symmetry, baryon number minus lepton number. Neither B nor L is a candidate
for an exact symmetry. Both have triangle anomalies with two weak SU(2) gen-
erators. The combination is not merely anomaly free, but it is, in fact, an exact
global symmet ry of the standard theory. It remains an exact symmet ry in the
extension of G to the unifying group SU(5). This simplest version of a grand unified
theory predicts observable violations of B and of L, but conserves the combination
B - L as an exact global symmetry.
We wish to ring changes upon the standard theory in order to produce non-zero
neutrino masses. The various possibilities may be classified in terms of what has
become of B - L symmetry. In O(10), for example, the generator of B - L lies
within the gauge group and has become a local symmet ry of the lagrangian. It
cannot be an exact symmetry, for there does not exist a second phot on with a
sensible coupling to B- L. It is spontaneously broken. In any model where B- L
is a spontaneously broken local symmetry, the fermion families must be extended
in order to assure that there are no (gauge) [3] triangle anomalies. A minimal
extension involves the introduction of one right-handed neutrino state into each
fermion family, and so it is in O(10). In the simplest O(10) theories [9], the masses
of observed neutrinos are Maj orana masses, and neutrinoless double beta processes
are allowed.
Perhaps B - L is a global symmet ry of the lagrangian and not a local symmetry.
This is the case for conventional SU(5) unification, wherein neutrino masses are
zero. In this scenario, while right-handed neutrinos need not be introduced, they
may be. Couplings may be set to make small Dirac neutrino masses, and to maintain
global B- L conservation. In such a scheme, neutrinoless /3fl processes are pro-
hibited. The adjustments required to make Dirac neutrino masses small compared
to quark and lepton masses do not seem particularly attractive to us. However,
unless neutrinoless t i f f decay or some other B- L violating process is confirmed
experimentally, this scenario cannot be excluded on any but aesthetic grounds.
H. M. Ge or gi et al. / Ne u t r i n o ma s s e s 301
More interesting is the possibility that global B- L symmet r y is spont aneousl y
broken, produci ng a true Gol dst one boson. Two recent papers [1, 2] have i nt roduced
model s of this kind. Even mor e interesting is the fact that the existence of this
Gol dst one boson cannot be excluded by available experi ment al data. In the next
section, we exhibit the Gel mi ni - Roncadel l i model [1] in detail.
The final logical possibility is that B - L is not a symmet r y of the lagrangian at
all. The phenomenol ogi cal implications of such schemes are sparse.
4 . The G R mo d e l
In the st andard t heory, the Higgs doubl et is al most unique. In order to be
responsi bl e for a quark or l ept on mass t erm, a Higgs multiplet must have Yukawa
couplings to the fermi on fields allowed by the gauge symmet r y and one of its
component s must be electrically neut ral so that it may devel op a vacuum expect at i on
value (VEV) wi t hout breaki ng el ect romagnet i c gauge invariance. Besides the Higgs
doubl et , t here is only one ot her possible Higgs multiplet with bot h these propert i es -
the compl ex triplet which couples to pairs of fermi on doublets and can give Maj or ana
neut ri no masses.
If such a triplet exists, it carries lepton number 2. Thus B - L can be a symmet r y
of the lagrangian only if the scalar Higgs meson self-interactions are invariant with
respect to i ndependent phase rot at i ons on the doubl et and triplet Higgs fields. If
the triplet field has a non-zero VEV, the B - L symmet r y is spont aneousl y broken.
This is the GR model .
We denot e the usual Higgs doubl et by
and the compl ex triplet Higgs by the 2 2 matrix field
(
x= x " "
The covari ant deri vat i ve is
e e
D" = 0 " + i T W" + i S V ~' , (4.3)
sin 0 cos 0
where T (S) are the SU(2) (U(1)) generat ors. On the scalar fields, the generat ors
act as follows:
TX = ~ ~'~ + ~g~" (4.4)
s 6 = - 4 0 , s x = - x .
302 H.M. Georgi et al. / Neut ri no masses
We can now wr i t e down t he most gener al l agr angi an whi ch is i nvar i ant under
SU( 2) x U(1) and i ndependent phas e r ot at i ons on ~b and X in t he f ol l owi ng f or m:
~LP(&, X) = (D"& )*D,& +t r [( D" x ) +D, x ]- V(ob, X) , (4. 5)
wher e
V( 6, X) = hi ( & +o - l u2) 2 + A2(tr ( X+X) - v2)z + A3(&+& - l u2) ( t r 0( +X) - i v2)
-1-/~4(~/~+(/~ t r (X+X)-cb+XX+Cb)+as((tr ( X+X) ) 2- t r (X+XX+X)). (4.6)
In this f or m, it is t ri vi al to find t he mi ni mum of t he pot ent i al . If A1, A2, A4 and A5
ar e posi t i ve and ]A31<2 a-,/a~a~, t hen V(~b, x) is posi t i ve semi def i ni t e. But
V((~b), (x)) vani shes for
We can t ake t hese to be t he VEV' s .
The VEV' s of (4.7) gi ve t he f ol l owi ng W and Z masses:
2
M2 w e ( u2+2v2)
- 4 sin 2 0
2
M2 e ( u2+4v2) " (4.8)
z - 4s i n 2 0 cos 2 0
If v # 0, t he s t andar d r el at i on Mw =Mz cos 0 is not sat i sfi ed, and t he neut r al
cur r ent s do not have t he canoni cal st r engt h. Thus f r om t he success of t he s t andar d
pr edi ct i ons, we know t hat v << u -~ 250 GeV.
Of t he t en r eal fields in (4.1), (4.2) t hr ee ar e eat en by t he Hi ggs mechani s m.
They ar e
[u (~b - 4) *) + 2v0( - x*)]/242u 2 + 8v 2 ,
(u~b +~/ 2v X ) / ~+2v 2 , (4.9)
and her mi t i an conj ugat e. One field is t he Gol ds t one bos on
M = [u (A "o - X *) - 2v (oh o _ ~ o, ) ] / 2# 2 u 2 + 8 v 2. (4. 10)
The ot her fields descr i be massi ve scal ars. The neut r al Hi ggs fields
1 / T i 0 O~x
~/~(<b+~b*), ~'e~kX , X ) , (4. 11)
have a mass s quar ed mat r i x
2Al u AaUV]
A3uv 2A2v2j . (4. 12)
Thus for u >> v, one l i near combi nat i on ( pr i mar i l y 4)) has mass of or der ,fA u and
t he ot her ( pr i mar i l y X) is a l i ght Hi ggs par t i cl e wi t h mass of or der ~/A v.
H. M. Georgi et al. / Neut ri no masses
There is a singly charged field
(u1" - 4 2 v4) ) / 4 u - ~ v 2
with mass squared
303
(4.13)
1)t4(U2 + 2V2) . (4.14)
And there is the doubly charged 1" with mass squared
A4U2+2AsU 2
( 4 . 1 5 )
Note that for u >> v, the t" is heavier than the singly charged field (primarily X-)
by a factor of ,]2.
Notice that none of the spinless mesons in the GR model can have a mass which
is large compared to GF 1/2. This result depends crucially on the assumption that
B- L symmet ry is not explicitly broken, not just on the existence of the t" field.
If we were to add a t" field with a large positive mass squared term
M 2 tr (1"~1") (4.16)
to the standard SU(2) x U(1) model and include the B - L breaking term
m& Tx~ d) + b. c. , (4.17)
the 1"o field would develop of VEV of order mu 2 / M 2. Thus, this t" could be used
to give Maj orana masses to the neutrinos, but we would encounter no other new
physics at moment a small compared to M. In the GR model, with (4.17) ruled out
by symmetry, the new physics is unavoidable.
The extra light charged bosons do not appear in the CMP model. The crucial
difference is that the B- L symmet ry in the CMP model is spontaneously broken
at a very large mass scale. Except for the maj oron, no trace of the B - L symmet ry
remains in the effective low-energy theory (indeed the maj oron is not, properly
speaking, a part of the low energy theory because all its couplings to light fields
are suppressed by inverse powers of a large mass).
We next discuss the couplings of the 1` to leptons. Let
t/.,[ =( ~' e) (4.18)
( L
be the standard charged lepton doublets where ( runs over e, tx and r. Then the
most general coupling of the triplet field X to the leptons is
g e e " e R X 'XVL +h. c. , (4.19)
where ~ec is the charge conjugate field. In a Maj orana basis for the y matrices,
g t ~ . c = g t [ * . ( 4 . 2 0 )
304 H. M. Georgi et al. / Neutrino masses
The coupl i ng mat ri x gee' = ge' e is symmet ri c. The Maj or ana mass mat ri x of the
neut ri nos is
ga~, v. (4.21)
We have no reason to assume that the coupl i ng mat ri x g is di agonal . For that
mat t er, we have no a pr i or i reason to assume anyt hi ng about it at all. For lack of
anyt hi ng bet t er, we will somet i mes assume that the ga.' are not wildly different for
different ( and ( ' , and we may refer to the component s generi cal l y as g.
5. Neutrino cosmol ogy
In this section, we will discuss the bounds on the par amet er s in the GR model
which deri ve from cosmol ogi cal consi derat i ons. We will discuss the range of par a-
met ers in the GR model for which the neut ri nos masses are cosmol ogi cal l y rel evant .
Our discussion starts with the of t - quot ed bound on neut ri no masses*
Z mv~< 100e V, (5.1)
where the sum ext ends over the stable light neut ri no species. This r emar kabl e
bound is many orders of magni t ude bet t er than any which can be obt ai ned in di rect
part i cl e physics experi ment s.
The argument for the bound is part i cul arl y simple and convincing. Just as t here
are phot ons which are relics of the big bang in the famous 2.7 bl ackbody radi at i on,
t here should be relic neut ri nos of each of the st abl e species. The neut ri no t em-
per at ur e ( Tv- 1. 9 K) is slightly lower than that of the phot ons (because e+e
anni hi l at i on added ent r opy to the phot ons after the neut ri no decoupl ed. We expect
a neut ri no number densi t y
n~ ~ 100/ cm 3 (5.2)
for each type. Then the t ot al energy densi t y due to the rest masses is
p, = n~ 2 rn, , . (5.3)
We know from rough observat i ons of the decel erat i on par amet er and from
bounds on the age of the universe to (from radi oact i ve dating) t hat p, cannot be
much bigger than the critical densi t y requi red to close the universe,
pc = 3 H~ / ( 8 r r G) , (5.4)
where Ho is the Hubbl e const ant
Ho - =- - t ~ / R ~ 100 km/ s mpc, (5.5)
and G is Newt on' s constant. This compari son yields the bound (5.1).
* This bound was first pointed out in ref. [11].
H. M. Georgi et al. / Ne ut r i no mas s es 305
Conver s el y, if n ~m~, is less t han a few per cent of pc, t he ener gy densi t y in neut r i nos
of t ype i is less t han t hat in t he obs er ved l umi nous mass in t he uni ver se and t he
cosmol ogi cal i mpor t ance of neut r i no mass m,~ is smal l . Thus we de ma nd as a
necessar y cr i t er i on for cosmol ogi cal r el evance of a neut r i no speci es,
m,~ ~> 1 e V. (5.6)
In t he GR model , t her e ar e sever al pr ocesses whi ch ar e not pr es ent in mor e
s t andar d model s and whi ch cause neut r i nos t o di s appear f r om t he uni ver se. The
most i mpor t ant are: decay of a neut r i no UH i nt o a l i ght er neut r i no uL pl us a maj or on,
UH~ uL+M ; (5.7)
i nt er conver s i on of heavy i nt o l i ght neut r i nos in pai rs due to s- channel M exchange,
PH 4- P H' ~ M- ' ~ PL + PL ;
and anni hi l at i on in pai r s i nt o maj or ons ,
t , +u- ~M+M.
(5.8)
(5.9)
The cosmol ogi cal effect of neut r i no decay has been di scussed by CMP [12]. But
t he decay is onl y r el evant if t he l i f et i me is shor t c ompa r e d to to ( - 10 l years). In
a model of t he GR or CMP t ype, t he neut r i nos ar e l ong- l i ved. The r eason is t hat
t he ma j or on is a Gol ds t one boson. Its coupl i ng t o neut r i nos in t he t r ee appr oxi ma-
t i on is pr opor t i onal to t he neut r i no mass mat r i x and t her ef or e pr es er ve neut r i no
i dent i t i es. The l eadi ng cont r i but i on to t he decay (5.7) comes f r om t he W exchange
di agr am in fig. 1. The r esul t for t he decay r at e is
F(I, H ~ uL+M) = ~sml 2 2c~ uFm~,,rnLv8rr5 In 2 , (5. 10)
wher e mL is t he mass of t he char ged l ept on in fig. 1 and c~ is a mi xi ng angl e. If we

w
~ ' H L - 1
I L - u L
I
I
I
I
M o
Fig. 1. The decay of a massi ve neut ri no into a lighter neut ri no plus maj or on proceeds, in lowest order,
t hrough this Feynman di agram.
306 l l . M. Georgi et al. / Neut ri no masses
t ake L t o be t he r and assume maxi mal mi xi ng, we find
5 X 1027 S
r - g3( v/ 1 keV) s . (5. 11)
The i nt er conver si on and anni hi l at i on pr ocesses (5.8), (5.9) ar e mor e i nt er est i ng.
The cross sect i ons ar e
2 2
O'2VH+2VL = g Hg L / 3 2 r r / 3 s , (5. 12)
O' 2u~2M = g 4 / 3 / 4 8 r r s , ( 5. 13)
wher e ~/s is t he t ot al c. m. ener gy and /3 is t he vel oci t y of t he i ni t i al neut r i nos in
t he cent er of mass. The anni hi l at i on cross sect i on (5. 13) is gi ven in t he appr oxi mat i on
of smal l /3 (at / 3- 1 t her e ar e i mpor t ant l ogar i t hmi c cor r ect i ons, but we will not
need t hem) .
We will focus our at t ent i on on t he anni hi l at i on pr ocess. It oper at es for any
neut r i no speci es. The i nt er conver s i on pr ocess al l ows heavy neut r i nos to di s appear
even f ast er but as we will see t he anni hi l at i on pr ocess al one is enough to si gni fi cant l y
decr eas e t he densi t y of neut r i nos.
Whe n t he neut r i no t e mpe r a t ur e is l arge c ompa r e d to t he neut r i no mass, t he
i nver se r eact i on
M+M- ~v+u (5. 14)
is i mpor t ant and t he maj or ons and neut r i nos ar e in equi l i br i um. Whe n t he t em-
per at ur e dr ops t o t he or der of rn~, (at a t i me whi ch we will call ti) t he i nver se
r eact i on phases out . Fr om t hat t i me on we will i gnor e it. Let n be t he neut r i no
densi t y as a f unct i on t i me, t hen
N = R 3 n (5. 15)
is t he numbe r of neut r i nos in a comovi ng cube of si de R. Then for ti < t < to, N( t )
sat i sfi es
dN N 2 g4/ 32
dt rtN/3o'2v+2M- R3 192rrm 2 . (5. 16)
u
At t = ti, t he ener gy densi t y p is domi na t e d by t he neut r i nos whi ch ar e becomi ng
non- r el at i vi st i c. For t > t~, t he uni ver se will r emai n mat t er domi na t e d unl ess so
many v' s anni hi l at e t hat t hei r cont r i but i on t o p becomes negl i gi bl e. But we ar e
t r yi ng to det er mi ne t he condi t i ons under whi ch t he neut r i nos still cont r i but e
si gni fi cant l y t o p. Thus, we can assume t hat t he uni ver se is mat t er domi na t e d for
t i < t < to. Then R is pr opor t i onal t o t 2/3, SO we can wr i t e
t . 2/ 3
H. M. Ge or gi et al . / Ne u t r i n o ma s s e s 307
where Ri = R (t i ),
Then (5.16) becomes
where
Int egrat i ng, we find
t ,81
t." 1o/3
1 dN . A( t ) (5.19)
N 2 dt -
A _
4
g
2 3
1927rm ~R i
(5.20)
7/ 3
1 1 =3At , [ 1- ( ~) ] (5.21)
N Ni
If A t i Ni >> 1, the final neutrino number approaches an asympt ot i c value i ndependent
of Ni:
1 9 2 ~ - rn . R i
No~ = 7 / ( 3 A t i ) 7 2 3
z - - _,
3 g4 t i (5.22)
The cont ri but i on of the surviving neutrinos to the energy density is
7 192, m3o/RiV to
/ - - |
(5.23)
P" - 3 g4 to kR o ] t~
W e can use (5.17) and the relation bet ween R and t emperat ure:
Ri kT~ (5.24)
go m, '
and write
1400m 3 / \ ' kTd 3/2
p, - - g4 t ~ - ) . (5.25)
Requi ri ng that pv exceed 1% of Pc, we obt ai n
( mJ 1 keV)3/ g ~ t> 1038 . (5.26)
Not e that we do not have to worry about the cont ri but i on to the energy density
from the maj or ons produced by the annihilation. Eq. (5.21) shows that most of
the annihilation takes place for t = ti so the energy of the maj orons is compar abl e
to the phot on t emper at ur e at the time. Thus, t here is a maj or on background
somewhat larger than we would expect if we i gnored annihilation. But like phot ons,
308 H. M. Georgi et al. / Neutrino masses
v 8
- - ~ - 1 0 3 e K e Y :
1 G e V
I
1 M e V
v
1 K e Y
r v= t 0
1 e V
m u ~- leV
I , , , I I I
] e V 1 K e Y 1 M e V 1 G e V
m ~
Fig. 2. This graph establishes the impossibility of a neutrino-dominated universe in the GR model. If
a given neutrino species is to contribute at least 1% to the critical mass density of the universe (.Q~ > 0.01 ),
then its mass and the value of v must lie in the unshaded domain. The vertical constraint ensures
m~ > 1 eV, while the skew constraint ensures that these neutrinos survive annihilation into massless
Goldstone bosons. In the subsequent section, we show that v must lie below the dotted line, so that
the unshaded region is physically unacceptable.
t he ma j or ons ar e mas s l es s , so t he i r c ont r i but i on t o t he e ne r gy de ns i t y is s i mi l a r t o
t ha t of t he mi c r owa ve ba c kgr ound, a nd t he r e f or e negl i gi bl e.
We s umma r i z e t he di s c us s i on i n t hi s s e c t i on i n fig. 2 whi c h s hows t he r a nge of
v a nd rnv f or whi c h a gi ve n ne ut r i no s peci es c a n be c os mol ogi c a l l y r e l e va nt . The
r e l e va nt r e gi on sat i sf i es t wo bounds ,
my/ > 1 e V, v S / mS v ~ 1038 ke V 3 , ( 5. 27)
whi c h c ome f r om ( 5. 6) a nd ( 5. 26) r e s pe c t i ve l y. We al so s how t he l i ne c or r e s pondi ng
t o a ne ut r i no l i f e t i me e qua l t o t he age of t he uni ve r s e ( f or ma xi ma l mi xi ng) . Fi nal l y,
l et us not e t ha t t he t wo ve r y l i ght s cal ar s c ount as one ne ut r i no . s peci es i n t he
anal ys i s of pr i mor di a l he l i um nuc l e os ynt he s i s .
H.M. Georgi et al. / Neutrino masses 309
6. Astrophysics
The discussion of the previous section does not put any useful bounds on the
paramet ers of the GR model. The " al l owed" region in fig. 1 is the region in
paramet er space for which a given neutrino is cosmologically relevant (in fact, there
is a small part of the cosmologically relevant region which is cosmologically ruled
out, but as we will see it is not interesting). However, there is an astrophysical
consideration which we can use to put a very strong bound on the VEV v,
v < 100 keV. This bound, together with fig. 2, shows that neutrinos in the GR
model cannot be cosmologically relevant.
The bound on v derives from consideration of maj oron emission from the cores
of red supergiant stars. A similar process, the emission of light axions, has been
studied by Dicus, Kolb, Teplitz and Wagoner [13]. The coupling of the GR maj oron
to fermions (except neutrinos) is similar in form to that of the axion, but suppressed
by a factor of 2 v / u . It is a pseudoscalar coupling to mass eigenstates with coupling
2v m (6.1)
b/ b/ '
where m is the quark or lepton mass. This follows from (4.10) which shows that
the GR maj oron has a component (with amplitude - 2 v / u ) which is the neutral
Gol dst one boson of the standard SU(2) U(1) model. Furt hermore, Dicus et al.
extended their calculations to very low mass axions ( - 10 eV) which for stellar core
t emperat ures much greater than 105 K are just like massless majorons.
When axion reabsorption corrections are neglected (which will certainly be a
good approximation for the much more weakly interacting majorons), Dicus et al.
find that light axion emission from the helium core of a red supergiant (M = 95
solar masses, T = 10 s K, p = 104 g/ cm 3) should cause energy loss of 1015 erg/ g s.
This is thirteen orders of magnitude larger than the energy production due to the
3c~ process in the core. It would lead to the complete exhaustion of the t hermonu-
clear energy available in times much shorter than the red supergiant core lifetimes.
To avoid this effect in the GR model, we must decrease the coupling by taking
2 v / u <~ 10 -6 or v ~< 100 keV. In the remainder of this section, we will give a simplified
analysis of maj oron emission.
At the relevant core t emperat ures - 10 s K- 104 eV, we can estimate the mean
free path of phot ons in the core, Lv, as follows:
LS~ 1 =o' c" ne, (6.2)
where o-c is the cross section for Compt on scattering (which dominates at these
energies),
2 2
cr c ' ~ Tro~ / me , ( 6. 3)
and ne is the electron number density,
n~ = 1028/cm 3 , (6.4)
310 H.M. Georgi et al. / Neutrino masses
corresponding to a mass density
Thus
The radius of the helium core is
P = 10 4 g/ c m3 (6.5)
Lv =4x 10 4 cm. (6.6)
R = ( 3 M/ 4 z r p) 1Is ---3 x 109 cm. (6.7)
The process of phot on diffusion can be approximated as a random walk of step
Lr. On the average ( R / L , ) 2 steps are required in order to diffuse a distance R.
Thus the number of collisions Nc suffered by the phot on in the process of its escape
is
Nc ~-- ( R / L ~ ) 2 . (6.8t
In each such collision there is a small probability P that a maj oron will be
produced instead of a photon, with
P = Or V+e ~M+e / OV c . (6.9)
When this happens, the maj oron will escape, carrying off the typical phot on energy
Ev ~ - - k T = 2 x l O 2 me . (6.10)
Thus, the fraction f of the produced energy carried off by maj orons is of the order
of the probability that a given phot on will convert to a maj oron before it escapes,
roughly
f = P Ne . (6.11)
We demand that f <~ 1, so that maj oron emission will not determine phot on emission.
The maj oron phot oproduct i on cross section can be inferred from the axion
calculation [14]. Expanding to lowest order in E. J me [see (6.10)] we find
t2~2
c~g /z v
O'v+e~M+e 3m 4 , (6.12)
where g' is the el ect ron-maj oron coupling constant given by (6.1) with rn = mo.
Putting all this together, we find that PNc ~< 1 implies
v ~<75 keV, (6.13)
in agreement with the Dicus et al. result.
7. Do u b l e bet a de c a y and bo unds on g
It is important to set limits on this process, for neutrinoless double beta decay
is logically separate from the question of neutrino masses. We may have large
H.M. Georgi et al. / Neutrino masses 311
Dirac neutrino masses, and exact B- L conservation. In this case, neutrinoless
double beta decay is fobidden*. We may (in the context of theories we do not
discuss) have vanishing neutrino masses and yet allow neutrinoless double beta
decay. This tub is on its own bottom.
Let us consider the question of neutrinoless double beta decay in the context of
the GR theory of neutrino masses. Three processes are relevant to double beta
decay. They correspond to the decay schemes
( Z , A ) ~ ( Z + 2 , A ) + e +e - +~+~, (7.1)
( Z , A ) ~ ( Z + 2 , A ) + e +e - , (7.2)
( Z , A ) ~ ( Z + 2 , A ) + e - + e +X . (7.3)
In (7.3), X signifies the maj oron or its very light scalar counterpart. The novelty
of the GR scheme in the occurrence of the third process, wherein no neutrinos are
produced but a very light boson is emitted. If double beta decay is observed in the
laboratory, it is difficult but conceivable to distinguish reaction (7.1) from reaction
(7.2) by the observed shape of the electron spectrum. It is even more difficult to
distinguish reaction (7.3) from reactions (7.1) and (7.2) in this fashion.
It may be possible to infer the occurrence of reactions (7.2) and (7.3) from the
observation of the rates of double beta decays together with a certain confidence
in the underlying nuclear physics. Doi et al. [5] have recently analyzed the Hennecke
et al. [16] data concerning the double beta decay of two isotopes of tellurium. For
13aTe, they conclude that the decay rate for the neutrinoless reaction (7.2) is about
thrice the rate for reaction (7.1). [Needless to say, they do not consider reaction
(7.3).] In terms of a Maj orana mass of the electron neutrino, they conclude that
my =41 eV.
It is straightforward to adapt the Doi calculation to deal with the GR model in
which reaction (7.3) can effectively compet e with reaction (7.2). Bot h of these
processes depend upon the identical nuclear matrix elements. Their respective
decay rates are related in the following manner:
F 3/F 2 = g2(847r2) l ( Q/ mv ) 2 R( x ) , (7.4)
where Fi is the partial rate for reaction (7.1) and g is the Yukawa coupling of the
maj oron to the electron neutrino, rn~ is the mass of the electron neutrino, Q is the
available energy of the decay, and x = Q/ me. We have omitted inessential complica-
tions due to neutrino mixing. The factor R ( x ) is a ratio of phase-space integrals,
Ix 4 + 14x 3 + 84x 2 + 210x + 210]
R ( x ) = [ x ~ - + ~ O ~ - ~ ~ 3. (7.5)
* Early theoretical work on the subject was done by Primakoff and Rosen [15].
312 H. M. Georgi et al. / Ne ut r i no mas s e s
In t er ms of t he GR model , t he t el l ur i um anal ysi s yi el ds t he s t at ement t hat F2 + F3 -~
3F~, and gi ves t he const r ai nt
g~e[v 2 + mZxZR (x)/ (84~-2)] = (30 eV) 2 , (7.6)
wher e x is eval uat ed at x = 1.7, so t hat x 2 R( x ) = 8.4.
Sensi t i ve sear ches have been pe r f or me d for t he no- neut r i no doubl e bet a decays
of 76Ge E17] and 82Se E18], r esul t i ng in upper l i mi t s for r eact i on (7.2) in whi ch t he
el ect r ons car r y off all of t he decay ener gy. Thes e exper i ment s ar e i nsensi t i ve to
r eact i ons (7.1) and (7.3). Haxt on, St ephens on, and St r ot t man [6] have anal yzed
t hese exper i ment s in t er ms of t he Maj or ana mass of t he el ect r on neut r i no, and
concl ude t hat its mass must be less t han 15 eV. In our not at i on, this means
geov < 15 e V. (7.7)
The const r ai nt s on gee and v f r om (7.6) and (7.7) ar e di s pl ayed in fig. 3. It is to be
st r essed t hat in t he cont ext of t he Gr model , t he i nequal i t y (7.7) is not in confl i ct
wi t h t he equal i t y (7.6).
I % v 1 2 " / ~ ~
lO % V -
103eV -
my :15eV \
IOOeV
lO eV -
I e V - -
Doi
Est imat e
I I I
lO -5 lO -4 lO -3 lO -Z lO -I
gee
Fi g. 3. The Doi et al. anal ysi s of Te /3/3 de c a y yi el ds t he sol i d cur ve as a cons t r ai nt on g~ a nd v. Al s o
s hown in t he Ha xt on et al. uppe r l i mi t on g~ev a nd t he Di cus et al. uppe r l i mi t on u. The GR model
is c ompa t i bl e wi t h all t hr ee cons t r ai nt s .
H.M. Georgi et al. / Neutrino masses 313
We turn now to the process of double K capture, for which there are three
relevant reactions,
e - + e + ( Z , A ) - ~ ( Z - 2 , A ) + v + v , (7.8)
e +e + ( Z , A ) ~ ( Z - 2 , A ) , (7.9)
e +e + ( Z , A ) ~ ( Z - 2 , A ) + x . (7.10)
Reaction (7.8) is " convent i onal " , t hough rarely discussed. Reaction (7.9) is not a
reaction at all, and will command separate comment. Reaction (7.10) is characteris-
tic of the GR model, and will be of concern to us.
Process (7.9) corresponds to an off-diagonal mass term which mixes two distinct
atomic species: the normal (Z, A) at om and an excited ( Z- 2, A) atom with two
holes in the atomic S-shell and, possibly, in an excited nuclear state. Conservation
of angular moment um requires that the initial nuclear spin coincides with the final
nuclear spin. The magnitude of the off diagonal mass term, dimensionally, is given
by
3 2
= (Zc~rne)-Gvm~m, ~ (7.11)
and cannot reasonably be expected to exceed 10 -2o eV.
At omi c mixing such as we have just described is analogous to KK mixing and
to neut ron-ant i neut ron mixing. It can lead to observable decay processes in the
following fashion. The (Z - 2, A) atom is in an unstable configuration, due to the
possibility of X-ray emission, the Auger process, or -/-ray emission from the excited
nuclear state. It follows that the original (Z, A) at om will also be unstable with a
decay rate given by
[ ' ( Z, A ) = e 2 F ( A 2 + ( 1/ . ) 2) 1 (7.12)
where F is the decay rate of the excited (Z = 2, A) atom and A is the difference
in mass between the two atoms. We know of no instance where this process can
realistically be detected. Our discussion of double K-capture may therefore be
limited to the conventional reaction (7.8) and the GR process (7.10).
A favorable candidate for double K-capt ure is the decay of SSNi (the principal
isotopic species) to the relatively rare isotope SSFe, where the available energy is
~3 MeV, and reaction (7.10) may possibly dominate reaction (7.8). Al t hough the
lifetime is certainly long, the signal of double K-capture is unique. It consists of
two characteristic Fe X-rays, with slightly different energies, in time coincidence.
A careful calculation of the rates of reactions (7.8) and (7.10), coupled with a
measurement of double K-capt ure in ~SNi may provide a unique test of the GR
model of neutrino masses.
Anot her rare decay mode that is generally permitted in the GR model is the
unobserved process
/ x- - ~e- e e + . (7.13)
314 H. M. Ge or gi et al . / Ne u t r i n o ma s s e s
An effective f our - f er mi on coupl i ng gener at i ng this decay is pr oduced by X
exchange, and its st rengt h is gi ven by
= g e e g , e / M 2 , (7.14)
wher e M is t he mass of t he doubl y char ged X meson. The exper i ment al upper
limit on t he br anchi ng ratio for react i on (7.13) is 10 9. It follows t hat t he GR
model is const r ai ned to satisfy
10 ~<3 x 10 -5 GF. (7.15)
We know f r om t he pr ecedi ng analysis t hat gee < 10 -3. If it is t rue t hat g, e - gee, we
may concl ude t hat M > 50 GeV. However , we cannot be sure t hat gue is not l arger
t han ge~, or cont rari wi se t hat it is not zero. In ei t her case, t here is a large range of
par amet er s in whi ch t he GR model is compat i bl e with t he obser ved suppressi on
of/ . t -+ 3e. Fur t her searches for this decay mode are mandat or y.
8. Lept on- l ept on col l i si ons and Z decay
Lept on- l ept on collisions present t he cl eanest l abor at or y in whi ch t o di scover t he
exi st ence of t he compl ex triplet of Hi ggs bosons, t he essential new component of
t he GR model . This is clearly t rue for X ++ and X +, whose st rongest coupl i ngs are
t hei r el ect r omagnet i c couplings. Wel l above t hreshol d X+X - pairs will be pr oduced
1 + - + ++ - -
i n e+e collisions with a cross sect i on ao' (e e +/ z /x ), while X X pairs will
have a cross sect i on equal to t he tz+/ x- pr oduct i on cross section. The masses of
t hese particles satisfy t he rel at i ons
15 Ge V~< M ++ = x/ 2M+ ~< 250 Ge V, (8.1)
wher e t he l ower limit is experi ment al , t he equal i t y is an essential pr oper t y of t he
GR model , and t he upper limit is pr obabl y necessary if t he t heor y is t o be
sel f-consi st ent . Whet her t hese particles will be accessible to t he next gener at i on of
e+e - machi nes is a mat t er of luck, as it is for t he t op quar k and t he convent i onal
neut ral Hi ggs boson.
How do t hese part i cl es decay? They can decay into t wo l ept ons:
4+ + +
) ~# g , (8.2)
+ +_
X --> ( v, (8.3)
or, t hey can decay weakl y:
++ + J +
X ~ A" + ,
+ 0+j + (8.4)
X -+X
wher e J+ denot es t he decay pr oduct s associ at ed with a char ged weak current : ud,
+ + +
cg, e v, b~ v, r v, etc. The weak decays are compl et el y predi ct abl e in t erms of
H. M. Geor gi et al. / Ne ut r i no mas s e s
the mass M of the X ++. The partial rate for these decays is of order
0.01 G~ M 5
if M <~ Mw, and of order
a 2M/sin4 0
if M ~> Mw. The partial width for the leptonic decay (8.2) is
[ga,,12M/64rr ,
and that for the decay (8.3) is
315
(8.5)
(8.6)
(8.7)
5Z [gee,12M/167r~/2. (8.8)
~,
++ + +
It is clear from the double beta decay constraint that the ~ e e partial width
is small compared to (8.5) or (8.6). Apart from that, we can say little about the
direct leptonic decays.
If the weak decays of the X ++ dominate, the decay products which result from
X++X product i on in e+e + annihilation consist of J+J+J J XX . The neutral
mesons are unobservable light Higgs or Gol dst one bosons. Thus, these events will
be highly spherical events involving four quark (or lepton) pairs with more than
20% missing energy. They should be readily distinguished from the larger jetlike
background. If the leptonic decays dominate, we can anticipate the more spectacular
+ + + +
final statestx tz tz /x or z z z r .
If gee is not too small, it may be possible to produce X singly in e- e- collisions.
Here the only hadronic background comes from two phot on exchange and the
purely leptonic background (e /~ , /z tz , etc.) is nil. For M- 100Ge V and
gee--10 3 (saturating the double beta decay bound (7.6)) the e e width of the
X is about 1 keV. The full width is much larger, at least 10 MeV. The X may
well be the most exciting potential discovery to be found in an e e facility. A
reasonable criterion that such a machine should satisfy is that it will reveal the X
if it exists.
It may be possible to observe X ++ and X + as decay products of Z and W + which
are produced in PP or PP colliding beam machines. The partial widths for various
decay modes are given in table 1.
TABLE 1
Decay mode Partial width Mass range
o + +
Z ~X ~ 0. 11Ge V M< < 47Ge V
W + ~ X++)C- 0. 24 Ge V M << 49 GeV
Z ~ x +X 0. 02 GeV M << 66 Ge V
W+~)C+X 0. 24 GeV M<< 117 Ge V
Z-+ X X 0. 34 GeV for any mass
316 H. M. Georgi et al. / Ne ut r i no mas s e s
The Z decay modes in t abl e 1 can be conveni ent l y st udi ed on t he Z r esonance
whi ch will be abundant l y pr oduced and will decay in a known fashi on i nt o all
ki nemat i cal l y accessi bl e channel s. The char ged X' s can be seen if t hey are light
enough t o be pr oduced. However , t he Z can cer t ai nl y decay i nt o XX , or mor e
accurat el y, i nt o our new light Hi ggs boson plus a maj or on, i ndependent of t he
val ue of M. The part i al decay wi dt h for Z t o decay i nt o this dar k channel is
pr eci sel y t wi ce its decay wi dt h i nt o each neut r i no- ant i neut r i no mode. It is envi saged
t hat LEP will succeed in compar i ng t he act ual Z wi dt h wi t h t he " vi si bl e" and
smal l er wi dt h deduced f r om t he l ept oni c wi dt h and br anchi ng rat i o. Thus, LEP
will at t empt to count t he number of f er mi on fami l i es by count i ng t he number of
neut r i no decay modes. In t he GR model , LEP' s count i ng must r eveal at l east five
" neut r i nos " . If t he meas ur ed number is less t han five, t he GR model is shown t o
be false. If it is five or mor e, we shall have a l ovel y ambi gui t y for f ut ur e exper i ment s
to resol ve. A wi nner ei t her way.
We woul d like t o t hank Dua ne Di cus, Dan Fr eedman, Al an Gut h, Ken Lane,
Davi d Schr amm, and Vi gdor Tepl i t z for useful conver sat i ons.
Re f e r e nc e s
[1] G.B. Gelmini and M. Roncadelli, Phys. Lett. 99B (1981) 411
[2] Y. Chikashige, R.N. Mohapatra and R.D. Peccei, Phys. Lett. 98B (1981) 265
[3] V.A. Lubimov, V.Z. Nozsik, E.G. Novikov, E.F. Tertyakov and V.S. Kosik, Phys. Lett. 94B
(1980) 266
[4] A. De Rtijula, Nucl. Phys. B188 (1981) 414
[5] M. Doi, T. Kotani, H. Nishiura, K. Okuda and E. Takasugi, Osaka University preprint number
OS-GE 80-27 (1980), revised version
[6] W.C. Haxton, G.J. Stephenson, Jr. and D. Strottman, Phys. Rev. Lett. 46 (1981) 698
[7] J.N. Bahcall et al., Phys. Rev. Lett. 45 (1981) 945
[8] F. Reines, H.W. Sobel and E. Pasierb, Phys. Rev. Lett. 45 (1980) 1307
[9] E. Witten, Proc. 1st Workshop on Grand unification, Durham, New Hampshire, April, 1980;
H. Sato, Kyoto University preprint RIFP-423 (1981)
[10] D.N. Schramm and G. Steigman, Gen. Rel. Gray., to be published
[11] R. Cowsik and J. McLelland, Phys. Rev. Lett. 29 (1972) 669; Astrophys. J. 180 (1973) 2
[12] Y. Chicashige, R.N. Mohapatra and R.D. Peccei, Phys. Rev. Lett. 45 (1980) 1927
[13] D.A. Dicus, E.W. Kolb, V.L. Teplitz and R. Wagoner, Phys. Rev. D18 (1978) 1829;
M.I. Vysotsskii, M. Yu. Khlopov and V.M. Chechetkin, JETP Lett. 27 (1978) 502;
K. Sato and H. Sato, Prog. Theor. Phys., 51 (1975) 1564
[14] T.W. Donnelly, S.J. Freedman, R.S. Lytel, R.D. Peccei, and M. Schwartz, Phys. Rev. D18 (1978)
1607
[15] H. Primakoff and S.P. Rosen, Phys. Rev. 184 (1969) 1925
[16] E.W. Hennecke, O.K. Manuel and D.D. Sabu, Phys. Rev. Cl l (1975) 1378
[17] E. Fiorini et al., Nuovo Cim. A13 (1973) 247
[18] B.T. Cleveland et al., Phys. Rev. Lett. 35 (1975) 757

Anda mungkin juga menyukai