VoLUME $9 25 JULY 1977 NVMSER4

Cosmological Lower Bound on Heavy-Neutrino Masses

Benjamin W. Lee&'~
Eenni National Accelemtox Labo~ato~, +~ Batavia, Illinois 60510
and
Steven Weinberg '~
Stanford University, Physics Department, Stanford, California 94305
(Received 13 May 1977)

The present cosmic mass density of possible stable neutral heavy leptons is calculated
in a standard cosmological model. In order for this density not to exceed the upper lim-
it of 2x 10 2~ g/cm, the lepton mass would have to be greater than a lower bound of the
order of 2 GeV.

There is a mell-known cosmological argument'
against the existence of neutrino masses greater
than about 40 eV. In the "standard" big-bang
cosmology, ' the present number density of each
kind of neutrino is expected' to be ~» the number
density of photons in the 3'K black-body ba, ck-
ground radiation, or about 300 cm '; hence if the
neutrinos are known to be lighter than 1 MeV.
However, heavier stable neutral leptons could
easily have escaped detection, and are even re-
quired in some gauge models. ' In this Letter, we
suppose that there exists a neutral lepton L' (the
"heavy neutrino") with mass well above 1 MeV,
and we assume that J0 carries some additive or

density would be greater than 2 &&10 "

neutrino mass were above 40 eV, their mass
g/cm',
which is roughly the upper limit allowed by pres-
ent estimates4 of the Hubble constant and the de-
celeration parameter.
However, this argument would not apply if the
neutrino mass were much larger than 1 MeV.
Neutrinos are generally expected' to go out of
thermal equilibrium when the temperature drops
to about 10' 'K, the temperature at which neu-
trano coll~sion rates become comparable to the
expansion rate of the universe. If neutrinos were
much heavier than 1 MeV, then they would al-
ready be much rarer than photons at the time
when they go out of thermal equilibrium, and
hence their number density would now be much
less than 300 cm '.
Of course, the familiar electronic and muonic
multiplicative quantum number which keeps it
absolutely stable. We will present arguments
based on the standard big-bang cosmology to show
that the mass of such a particle must be above a
lower bound of order 2 GeV.
At first glance, it might be thought that the
present number density of heavy neutrinos would
simply be less than the above estimate of 300
cm ' by the value exp[-m~/(1 MeV)] of the
Boltzmann factor at the time the heavy neutrinos
go out of thermal equilibrium. If this were the
case, then an upper limit of 2X10 "g/cm ' on
the present cosmic mass density would require
that m~ exp[-m~/(1 MeV) ] should be less than 40
eV, and hence that m~ should either be less than
40 eV or greater than 13 MeV,
However, the true lower bound on the heavy-
neutrino mass is considerably more stringent.

165
VOLUME )9, NUMBER 4
The heavy neutrinos are assumed to carry a con-
served quantum number, so their number density
can relax to its equilibrium value n0 only by an-
nihilation of a heavy neutrino with a heavy anti-
neutrino, in processes such as
L0L0-vP, e e+, p p,
But although the energy distribution of the heavy
neutrinos is kept thermalized by collisions with
v, e, etc. , down to a temperature of 10'0'K, Bt
that temperature they are so rare that their an-
nihilation rate is already much less than the cos-
mic expansion rate. Thus the heavy neutrinos go
out of chemical equilibrium (in the sense that
their number density begins to exceed its equil-
ibrium value) at a "freezing" temperature Tz
which is much higher than 1 MeV. The condition
on m~ is then that m~ exp( m~/k-T&) should be
less than 40 eV, and the resulting lower bound on
m~ must therefore be greater than 13 MeV.
To make this quantitative, we use the rate
equation
dn 3A
dt 8 n-(m—)n'+ (ov)n 0
Here n is the actual number density of heavy neu-
trinos at time t; R is the cosmic scale factor;
(vv) is the average value of the L'L~ annihilation
cross section times the relative velocity; and n0
is the number density of heavy neutrinos in ther-
mal (and chemical) equilibrium':
2 ~ +
(nT) =
), 4&p'dp exp
(We use units with 5 = c = 1 throughout. )
At, the temperatures we are considering here,
the energy density and the entropy are dominated
by highly relativistic particles, including y,
v&, and e. It follows that 8,
T, and t are related
by
It/It = -T/T = (e~pG/3)'~2
where p is the energy density
p =N~T~ =N~w2(k T)~/15,
with N~ an effective number of degrees of free-
dom, counting ~ and ~I6, respectively, for each
boson or fermion species and spin state. For
temperatures in the range of 10-100 Me& (which
most concern us here) we must include just y,
v„~u, vv, e,
and e', soN~=4. 5, a value
we will adopt for most purposes. However, if
166
P H Y 5 I C A L R E V I E %' I. E T T E R 8 25 JUx, v 1977
current ideas about the strong interactions are
correct, then N~ rises steeply at a temperature
equilibrium because, as we have seen, the upper
limit on the present cosmic mass density requires
the Boltzmann factor exp(-mz, /k T&) to be very
small. For nonrelativistic velocities, the cross-
', & &+, etc. section v(v) for the exothermic processes (1) be-
haves like 1/v, so we can take (ov) as a tem-
perature-independent constant. If there were on-
ly a single annihilation channel open (say L'L'
- vv) and if the annihilation process were due to
an ordinary "V-minus-A" charged-current Fermi
interaction, then (ov) would be just GF'mz, '/2&,
where GF is the Fermi coupling constant 1.15
&10 ' GeV '. To take account of more general
possibilities, we shall write
(cv) = G, 'm, 'N„/2~, (6)
where N„ is a dimensionless fudge factor which
depends both on the number of annihilation chan-
nels open and on the details of the weak L'L' an-
nihilation interaction. (In general, the annihila-
tion proceeds both through exchange of charged
and of neutral intermediate vector bosons. )
(2)
Where a numerical value is needed, we will take
N„=14. [This corresponds to the annihilation
channels L'L'- v, ~„ ~„v„, &, &„e e', p. p, ', uu,
dZ, ss, with all fermions (antifermions) of helic-
' (+-,'), and three colors for each of the
ity ——,
quarks u, d, and s. ] New channels open for m1
above about 5 GeV, leading to a moderate rise in
1V~.
2 2)1/2 1
Equations (3)-(6) allow us to rewrite the rate
yl
kT equation (2) in the convenient form
df/dx=~V'(f'-f ') (7)
where
x =kT/m~, p'=mz, 'N„/vNF, f(x; p) =n/T',
v„ f,(x) = n, /T'
=m 2k2f duu'[exp(x '+u')'~'+1]
(4)
(5)
cm K
Equation (7) is to be solved subject to the initial
condition that, as x-~, f(x) approaches the equil-
ibrium value f, (x). For x «I,
f(x) approaches an
asymptotic value, which evidently depends only on
 M& 39 &UMg~R g PHYSICAL R EVIEW LETTERS 25 Jvi, v 1977

'
the sin g 1e unknown parameter p' 0 must be subject t o th e lower bound

lim f (x; p) = P p). (10) m~(N„/WN )

'" 2 5. 2 GeV 4 (14)

i z ion of electr on-positron

The subs equent annihilat'
For N„= 14 and
Allo fo
¹~ = 4. 5, this ives
fac t or-of-4 unce
pairs increases RV'b y a factor
~ ~

i i a ion rate, the low ower bound on rn I

d /T'b y a factor 4 so
11 would lie in th e range 1-4 GeV.
o cavy neutrinos (and antineutrin-
It is enlightening to see ow these
ee how th results can
os) will be given by
r
be obtained b y an analytic a pproximation to the
p~
= 2(~i)m~T
L y
'P p) ( 1 1) asymptoti ic va lue of n/T'. We ex ect t
with T =O'K ia con temperature.
K the present radiati
i mi at e 1y equal to p x until the tem-
perature drop ops t o a freezin va
(For freezining teemperature abovee 100 MeV, the
factor 11 must bee d ecreased to takee account of the
subsequent heatia zng of photons by annih'1 t'
i iaionof P7

pp, wm', etc)

df, /dx=Cp. 'fo', at x=x~=kT
f f /m » (15)
omputer solutions of Eq . (7
Com
special cases in Fixg.. 1.. We find that
over the range
ge oof parameters con ' d re ere
and that thereafter f obe a
o eyss approximately the
equation
~~sp. ) of Eq. (10) ca y
f
d /dx = C p'f ', (16)
E(p,) =1.20 x10 ' (cm
cm 'K)
K ''[ GeV)] '~'. (12) with the initial condition f(xz ) = f, (xf).
x Since kT f
nt mass density of heav
The present
p
n approximate ,
iin Eq. ( )bythe
ti t'o
i th en given by Eq. (11) as
is
p ~ = (4. 2 x 10
" g/cm') fp(x)np/Ts —2k3(2vx) (17)
and take only
onl the exponential intoo
xn

x[m, (Gev)] "'(N /WiV p ~

'" o
obtained from
e freezing tern perature
e is then

If we require4 that pz, &2x10 ~

g/ccm, th
then xq
'" exp(1/x~)
g yn L

= Ck32(2m) 3~2p~

n{ T ) {cmi'
=1.35x10'~ ~ -»2"[m~(Ge V)]'.
10 N„N~

7) and (
IO

f(x, ) =f, (x,) =(Cu'x f') ' ~

I.O
Equation 16 can now be solved:
IO f(x) = [f.(;) '+CAx, —x)] ' (20)

IO
so that, with xf ' -—20
0.5GeV 4.5.
(NF =

GeV(NF=4. 5, NA= l4)

I
NA= l4)
f
E= (0) = (Cp'xq') '(1+ x
xf
IO 2GeV(NF=4. 5, NA=I4)
=1.5x10 ' [p (GeV)] '(cm 'K) ' (21)
5GeV(NF=4. 5, NA= 17)
10 in reasonable
ble agreement with the
lOGeV(NF =30, NA= l7)
is analysis shows that
-lO ' t heavy- g
e y to be about — „he
l
l4 l2
I

8
I

6
neutrino masss, or 0OMeVform =2 GeV.
IO IO IOIO IO IO I.O T
IO lO
Of course if a st s able
bl hea ne
FIG. 1. n/T'v
g/T vs Z' ie y oof special cases of
for a variet ere with a masss oof order 1-15 GeV,
mL, N~, 311d N~. iona fie
the gravitational ield of these heavy t '
vy neutrinos

167
VOLUME 39, NUMBER 4 PHYSICAL REVIEW LETTERS 25 JUx.v 1977

would provide a plausible mechanism for closing
the universe.
One of us (S.W. ) would like to thank Robert
Wagoner for helpful conversations, and the other
(B.W. L.) wishes to thank Robert Pearson for his
assistance in computer calculations.
~~~ Deceased.
&&Operated by Universities Research Association Inc.
under contract with the U. S. Energy Research and De-
velopment Administration.
")On leave 1976-1977 from Harvard University.
'See e.g. , G. Marx and A. S. Szalay, in Proceedings
of the Neutrino 1972 Conf-erence, edited by A. Frankel
and G. Marx (OMDK-TECHNIOFORM, Budapest, 1972),
Vol. I, p. 191; R. Cowsik and J. McClelland, Phys.
Rev. Lett. 29, 669 (1972), and also Astrophys. J. 180,
7 (1973).
~For a review with references to the original litera-
ture, see S. Weinberg, i"ragitation and Cosmology:
Principles and Applications of the General Theory of
Relativity (Wiley, New York, 1972), Chap. 15.
This includes a factor zs for the effects of Fermi
statistics; a factor ~«because electron-positron annihi-
lation has increased the photon temperature by an ex-
tra factor of (~4) '/; and a factor of 2 because both neu-
trinos and antineutrinos are included. The massive neu-
trinos are, like the photons, supposed to exist in two
helicity states. (The 40-eV upper limit given here dif-
fers from the 8-eV bound given by Cowsik and McClel-
land in Ref. 1, because we take account of the heating
of photons by electron-positron annihilation, and be-
cause we do not assume that the muon and electron-
type neutrinos have equal mass. )
To take account of the uncertainty in galactic evolu-
tion, we assume that qo& 2. A value Ilo =50.3+ 4.3 km/
sec/Mpc is given by A. Sandage and G. A. Tammann,
Astrophys J.. 210, 7 (1976). The desntiy 2x 10 2~ g/cm3
corresponds (for zero cosmological constant) to q0=2
and Ho =50 km/sec/Mpc. [These quantities are defined
in terms of the cosmic scale factor 1t (t) as (d11/dt)0
—
= ROHD and (d'R/dt')0=—-Ropo qo, a subscript 0 denoting
the present instant. ]
See, e.g. , B. W. Lee and S. Weinberg, Phys. Rev.
Lett. 38, 1237 (1977). It is possible that the heavy neu-
tral leptons in this model actually have lifetimes which
are shorter than the age of the universe, in which case
the lower bound on ~~ derived here need not be satis-
fied. tFor calculations of neutral-lepton decay rates,
see J. N. Bahcall, ¹
Cabibbo, and A. Yahil, Phys.
Rev. Lett. 28, 318 (1972); W. J. Marciano and A. I.
Sanda, Phys. Lett. 67B, 303 (1977); T. Goldman and
G. J. Stephenson, Jr. , Los Alamos Laboratory Report-
No. LA-UR-77-941 (to be published); and H. Fritsch
and P. Minkowski, Phys. Lett. 62B, 72 (1976).]
6We suppose here that the chemical potentials associ-
ated with any conserved quantum numbers carried by
are vanishingly small. If there were any apprecia-
ble degeneracy, the mass density of the heavy neutri-
nos would greatly exceed reasonable cosmological lim-
its.
This includes contributions of y, v„v, , v„, v„, e, e,
p, p+; plus three color triplets g, d, s, of quarks and
the correspondirg antiquarks; plus eight massless spin-
one gluons.

Cosmological Upper Bound on Heavy-Neutrino Lifetimes

Duane A. Dicus~ ) and Edward %. Kolb '

Center for Particle Theory, University of Texas at Austin, Austin, Texas 78712

and

Vigdor L. Teplitz'")
Department of Physics, Virginia Polytechnic Institute and State University, Blachsburg, Virginia 24061
(Received 31 May 1977)

An upper bound on the lifetime of a massive, neutral, weakly interacting lepton, v~,
is derived from standard big-bang cosmology. Saturation of the bound and reasonable
assumptions about the weak interaction of the v„ then yield a prediction of approximately
10 MeV for its mass.

Recently Lee and Weinberg' have pointed out This density is an order of magnitude greater
that stable neutrinos with masses in the GeV than proven mass reserves such as galaxies and
range are capable of "hiding" a cosmological en- the cosmic microwave background but is suggest-
ergy density on the order of ed by current best values for the Hubble constant
and deceleration parameter. '
It has been shown
'
p =p, =10 MeV/cm'. by Cowsik and McClelland' that stable neutrinos

168