Neutrino Mass Spectrum and

Neutrinoless Double Beta Decay

arXiv:hep-ph/0003219v4 8 Oct 2000

H.V. KlapdorKleingrothaus, H. Pas

MaxPlanckInstitut f
ur Kernphysik,
P.O. Box 103980, D-69029 Heidelberg, Germany
A. Yu. Smirnov
The Abdus Salam International Center of Theoretical Physics,
Strada Costiera 11, Trieste, Italy
Institute for Nuclear Research, RAS, Moscow, Russia

Abstract
The relations between the effective Majorana mass of the electron neutrino, mee ,
responsible for neutrinoless double beta decay, and the neutrino oscillation parameters are considered. We show that for any specific oscillation pattern mee can take
any value (from zero to the existing upper bound) for normal mass hierarchy and
it can have a minimum for inverse hierarchy. This means that oscillation experiments cannot fix in general mee . Mass ranges for mee can be predicted in terms
of oscillation parameters with additional assumptions about the level of degeneracy
and the type of hierarchy of the neutrino mass spectrum. These predictions for mee
are systematically studied in the specific schemes of neutrino mass and flavor which
explain the solar and atmospheric neutrino data. The contributions from individual
mass eigenstates in terms of oscillation parameters have been quantified. We study
the dependence of mee on the non-oscillation parameters: the overall scale of the
neutrino mass and the relative mass phases. We analyze how forthcoming oscillation
experiments will improve the predictions for mee . On the basis of these studies we
evaluate the discovery potential of future 0 decay searches. The role 0 decay
searches will play in the reconstruction of the neutrino mass spectrum is clarified.
The key scales of mee , which will lead to the discrimination among various schemes
are: mee 0.1 eV and mee 0.005 eV.

Introduction

The goal of the search for neutrinoless double beta decay (0 decay) is to establish the
violation of (total) lepton number L and to measure the Majorana mass of the electron
neutrino, thus identifying the nature of the neutrino [1, 2]. Both issues are related: Even if
the main mechanism of 0 decay may be induced by e.g. lepton number violating righthanded currents, R-parity violation in SUSY models, leptoquark-Higgs couplings (for an
overview see e.g. [3]), the observation of 0 decay implies always a non-vanishing
effective neutrino Majorana mass (0-mass ) at loop level [4].
1

If 0 decay is induced dominantly by the exchange of a light Majorana neutrino

(m < 30 MeV), the decay rate is proportional to the Majorana mass of the electron
neutrino mee squared:
m2ee .
(1)
Thus, in absence of lepton mixing the observation of 0 decay would provide an information about the absolute scale of the Majorana neutrino mass.
The situation is changed in presence of neutrino mixing when the electron neutrino is
not a mass eigenstate but turns out to be a combination of several mass eigenstates, i ,
with mass eigenvalues mi :
e =

Uei i ,

i = 1, 2, 3, ... .

(2)

Here Uej are the elements of the mixing matrix relating the flavor states to the mass
eigenstates. In this general case the mass parameter (0-mass ) which enters the
0 decay rate is not the physical mass of the neutrino but the combination |mee | of
physical masses:


|mee | =




X
2 ij

|Uej | e mj  .


 j

(3)

Apart from the absolute values of masses mj and mixing matrix elements, the effective
Majorana mass depends also on new parameters: phases j which originate from a possible
complexity of the mass eigenvalues and from the mixing matrix elements. Thus searches
for double beta decay are sensitive not only to masses but also to mixing matrix elements
and phases j .
Notice that in the presence of mixing mee is still the ee-element of the neutrino mass
matrix in the flavor basis 1 . In this sense it gives the scale of elements of the neutrino mass
matrix. However, in general, mee does not determine the scale of the physical masses.
If 0 decay will be discovered and if it will be proven to proceed via the Majorana
neutrino mass mechanism, then the mee extracted from the decay rate will give a lower
bound on some physical masses. As it is easy to see from Eq. (3)), at least one physical
mass, mj , should be
mj mee
(4)
for the three-neutrino case.
Can mee be predicted? According to (3) the mass mee depends on absolute values
of masses, mixings and phases j . Certain information about masses and mixing can
be obtained from (i) oscillation searches, (ii) direct kinematical measurements and (iii)
cosmology. Let us comment on these issues in order.
1
In general the experimental value of mee depends on the process being considered. It coincides with
the theoretical mee of eq. (3) if all masses mi Q, where Q is the energy release of a given process.
This fact may become important for comparing heavy neutrino contributions in 0 decay and inverse
neutrinoless double beta decay at colliders, see e.g. [5].

1). The oscillation pattern is determined by mass squared differences, moduli of

elements of the mixing matrix, and (for three neutrino mixing) only one complex phase
which leads to CP violating effects in neutrino oscillations:
m2ij |mi |2 |mj |2 ,

|Uej |2 ,

CP .

(5)

(We indicated here only mixing elements which enter mee .) In what follows we will call
(5) the oscillation parameters.
Neutrino oscillations and neutrinoless double beta decay, however, depend on different
combinations of neutrino masses and mixings. In terms of the oscillation parameters the
mass (3) can be rewritten as
|mee | =



q
X
2 ij

m2j1
|U
|
e
ej

 j




2
+ m1  ,


(6)

where we assumed for definiteness m1 to be the smallest mass. We also put 1 = 0 and
consider the other j as the relative phases.
According to (6) the oscillation parameters do not allow one to determine uniquely
mee . Apart from these parameters, the mass mee depends also on the absolute value of
the first mass (absolute scale) and on the relative phases:
m1 ,

j ,

j = 2, 3, ...

(7)

These parameters can not be determined from oscillation experiments and we will call
them non-oscillation parameters.
The mass squared difference gives the absolute
q value of the mass only in the case of
strong mass hierarchy: mj m1 , when |mj | m2j1 . However, even in this case the
lightest mass (which can give a significant or even dominant contribution to mee ) is not
determined.
The relative phases j which appear in mee (eq. (3)) differ from CP and can not be
determined from oscillation experiments, since the oscillation pattern is determined by
moduli |mi |2 . On the other hand, the phase relevant for neutrino oscillations does not
enter mee or can be absorbed in phases of masses.
2). Apart from neutrino oscillations, informations on neutrino masses and especially
on the absolute scale of masses can be obtained from direct kinematical searches and
cosmology.
There is still some chance that future kinematical studies of the tritium beta decay
will measure the electron neutrino mass, and thus will allow us to fix the absolute scale
of masses. Projects are under consideration which will have a sensitivity of about 1 eV
and less (ref. [6]).
3). The expansion of the universe and its large scale structure are sensitive to neutrinos
with masses larger than about 0.5 eV. The status of neutrinos as the hot dark matter
(HDM) component of the universe is rather uncertain now: it seems that the present
cosmological observations do not require a significant contribution and therefore a
3

large O(1eV) neutrino mass. However in some cases massive neutrinos may help to get a
better fit of the data on density perturbations.
In order to predict mee one should not only determine the oscillation parameters
but make additional assumptions which will fix the non-oscillation parameters. If the
oscillation parameters are known, then, depending on these assumptions, one can predict
mee completely or get certain bounds on mee .
What are these assumptions?
It was pointed out in [8] that predictions on mee significantly depend on two points:
The level of degeneracy of the neutrino mass spectrum, which is related to the
absolute scale of neutrino masses.
The solution of the solar neutrino problem; this solution determines to a large extent
the distribution of the electron neutrino flavor in the mass eigenstates, that is, |Uej |2 .
The assumptions about the level of degeneracy allow one to fix the absolute scale of the
neutrino mass. In fact, at present even the oscillation parameters are essentially unknown,
so that further assumptions are needed. Evidences of neutrino oscillations (atmospheric,
solar neutrino problems, LSND result) allow us in principle to determine the oscillation
parameters up to a certain ambiguity related, in particular, to the existence of several
possible solutions of the solar neutrino problem.
A number of studies of the 0-mass have been performed, using various assumptions
about the hierarchy/degeneracy of the spectrum which remove the ambiguity in interpretations of existing oscillation data. In fact, these assumptions allow one to construct the
neutrino mass and mixing spectrum, and some studies have been performed for specific
neutrino spectra. Most of the spectra considered so far explain the atmospheric neutrino
problem and the solar neutrino problem assuming one of the suggested solutions. Some
results have also been obtained for schemes with 4 neutrinos which also explain the LSND
result. Let us summarize the main directions of these studies.
(1) Three-neutrino schemes with normal mass hierarchy which explain the solar and
atmospheric neutrino data have been studied in [8, 9, 10, 11, 12, 13]. Various solutions
of the -problem were assumed. These schemes give the most stringent constraints on
0-mass in terms of oscillation parameters.
(2) The 0-mass in three-neutrino schemes with inverse mass hierarchy has been
considered in [14, 12, 13]. These schemes favor mee to be close to the present experimental
bound.
(3) Three-neutrino schemes with partial degeneracy of the spectrum and various solutions of the -problem were discussed in [8, 13]. In these schemes mee can also be close
to the present experimental bound.
(4) Large attention was devoted to the three-neutrino schemes with complete degeneracy [8, 15, 16, 17, 18, 19, 20, 21, 22, 23, 13] since they can explain solar and atmospheric
neutrino data and also give a significant amount of the HDM in the universe. In these
schemes the predictions of mee depend mainly on the absolute mass scale and on the
mixing angle relevant for the solar neutrinos.
4

Some intermediate situations between hierarchical and degenerate spectra have been
discussed in [19, 13].
5. The 0-mass in scenarios with 4 neutrinos which can accommodate also the
LSND result have been analyzed in Ref. [14, 10, 11].
Some general bounds on the 0-mass under various assumptions have been discussed
in [24, 25, 26, 19, 27, 28].
In a number of papers an inverse problem has been solved: using relations between
the 0-mass and oscillation parameters which appear in certain schemes restrictions
on oscillation parameters have been found from existing bounds on mee . In particular the
3-schemes with mass degeneracy [15] and mass hierarchy [12] have been discussed.
An important ingredient for the prediction of mee are the phases (see eq. 7). Unfortunately, there is no theory or compelling assumptions which allow to determine these
phases.
In this paper we will analyze the discovery potential of future 0 decay searches in
view of existing and forthcoming oscillation experiments. We will clarify the role 0 decay searches will play in the identification of the neutrino mass spectrum. In the previous
studies, implications of mee for oscillation parameters and the other way around, implications of oscillations searches for mee have been discussed. In contrast, we focus here on the
impact of results from both neutrino oscillations and double beta decay on the reconstruction of the neutrino mass spectrum. We put an emphasis on possible future experimental
results from long-baseline experiments, CMB explorers, supernovae measurements, precision studies of properties of the solar neutrino fluxes (day-night asymmetry, neutrino
energy spectra, etc.).
For this we first (sect. 2) consider the general relations between the effective Majorana
mass of the electron neutrino and the oscillation parameters. We will study the dependence of mee on the non-oscillations parameters. The crucial assumptions which lead to
predictions for mee are identified.
In sects. 3 - 8 we present a systematic and updated study of predictions for mee for
possible neutrino mass spectra. In contrast with most previous studies using oscillation
data we quantify the contributions from individual mass eigenstates and we keep explicitly
the dependence on unknown relative mass phases. The dependence of predictions on nonoscillation parameters the absolute mass value and the phases i is studied in detail. We
consider 3schemes with mass hierarchy (section 3), partial degeneracy (section 4), total
degeneracy (section 5), transition regions (6), inverse hierarchy (section 7) and schemes
with sterile neutrinos (section 8). We analyse how forthcoming and planned oscillation
experiments will sharpen the predictions for mee . In sect. 9, comparing predictions of
mee from different schemes we clarify the role future searches for 0 decay can play in
the identification of the neutrino mass spectrum.

Neutrino oscillations and neutrinoless double beta

decay

As has been pointed out in the introduction, the prediction of mee depends on oscillation
(|Uei|, m2 ) and non-oscillation (m1 and j ) parameters. In this section we will consider
general relations between mee and the oscillation parameters. We analyse the dependence
of these relations on non-oscillation parameters. We quantify ambiguities which exist in
predictions of mee . Our results will be presented in a way which will be convenient for
implementations of future oscillation results.

2.1

Effective Majorana mass and oscillation parameters

The oscillation pattern is determined by the effective Hamiltonian (in the flavor basis):
H=

1
MM + V ,
2E

(8)

where E is the neutrino energy, M is the mass matrix and V is the (diagonal) matrix of
effective potentials which describe the interaction of neutrinos in a medium.
The oscillation pattern is not changed if we add to H a term proportional to the unity
matrix:
MM MM m20 I .
(9)

Indeed, the additional term does not change the mixing, it leads just to a shift of the
mass eigenstates squared by the same value without affecting m2ij :
m2i m2i m20 .

(10)

(we consider m2i m20 0 for all i to keep the Hermiticity of the Hamiltonian). The
additional term changes, however, the 0-mass . Thus, for a given oscillation pattern
there is a freedom in mee , associated with m20 .
Let us study how arbitrary mee can be for a given oscillation pattern. According to
(9,10) for the three-neutrino case we get
i2
i3
mee = |m(1)
|m(2)
|m(3)
ee | + e
ee | + e
ee | ,

(11)

(i)
|m(i)
where mee
ee | exp (ii ) (i = 1, 2, 3) are the contributions to mee from individual mass
eigenstates which can be written in terms of oscillation parameters as:
2
|m(1)
ee | = |Ue1 | m1 ,

2
m221 + m21 ,
|m(2)
ee | = |Ue2 |

2
|m(3)
m231 + m21 .
ee | = |Ue3 |

(12)
(13)
(14)

and i are the relative phases of the contributions from masses mi and mj (the mass m20
has been absorbed in definition of m21 ). The contributions m(i)
ee can be shown as vectors
in the complex plane (fig. 1).
6

U2 m
e3

<m>

min

2
e2

U2e2 m2

U2e3 m3

Ue1 m1

U2 m
e1

a)

b)

Figure 1: The effective Majorana mass mee in the complex plane. Vectors show contributions to mee from individual eigenstates. The total mee appears as the sum of the three
vectors. Allowed values of mee correspond to modulies of vectors which connect two points
(2)
(3)
on the circles. Here = 3 , = 2 . a). |m(1)
~ (i)
ee | > |mee | + |mee |: the vectors m
ee
(1)
(2)
can not form a triangle and no complete cancellation occurs. b) |mee | |mee | + |m(3)
ee |:
in this case complete cancellation occurs in the intersection points of the circles, so that
mee = 0.
Without loss of generality we assume m3 > m2 > m1 0, so that m1 is the lightest
state. Then normal mass hierarchy corresponds to the case when the electron flavor
prevails in the lightest state: |Ue1 |2 > |Ue2 |2 , |Ue3|2 . We will refer to inverse hierarchy as
to the case when |Ue1 |2 < |Ue2 |2 or/and |Ue3 |2 , i.e. when the admixture of the electron
neutrino flavor in the lightest state is not the largest one.
Let us consider the dependence of mee on non-oscillation parameters mee = mee (m1 , j ).
It is obvious that due to the freedom in the choice of m1 there is no upper bound for mee .
However, in some special cases lower bounds on mee exist.
Let us start with the two neutrino case which would correspond to zero (or negligibly
small) e admixture in one of mass eigenstates, e.g. |Ue3 |. We consider first the case of
2
2
normal hierarchy Ue1
> Ue2
. For m1 = 0 the effective mass mee is uniquely fixed in terms
of oscillation parameters:
q

m0ee = |Ue2 |2 m221 sin2 m221 ,

(15)

where sin Ue2 . For non-zero m1 , the maximal and minimal values of mee correspond
to 2 = 0 and 2 = .
The upper bound (2 = 0) on mee increases with m1 monotonously from m0ee at m1 = 0
and approaches the asymptotic dependence mee = m1 for large m1 (see fig. 2 a). The
lower limit (2 = ) decreases monotonously with increase of m1 starting by m0ee . It
reaches zero at
sin2 q
m221 ,
(16)
m1 = q
| cos 2|

and approaches the asymptotic dependence mee = | cos 2| m1 at large m1 (see fig. 2 a).
Thus, for arbitrary values of oscillation parameters, no bound on |mee | exists.
7

ee

3
2.5

a)

2
1.5
1
0.5

0.5

1.5

2.5

m1
ee

3
2.5

b)

2
1.5
1
0.5

0.5

1.5

2.5

m1

Figure 2: Qualitative dependence of the effective Majorana mass mee on m1 in the two
neutrino mixing case. a) corresponds to the case of normal mass hierarchy, b) to the case
(2)
of inverse hierarchy. Shown are the contributions m(1)
ee (dashed) and mee (dotted). The
max
min
solid lines give mee and mee , which correspond to 21 = 0 and 21 = , respectively.
While in the case of normal hierarchy a complete cancellation is possible, so that mee = 0,
for inverse hierarchy a minimal value |mmin
ee | > 0 exists.
(2) In the case of inverse hierarchy, |Ue2 | > |Ue1 |, the function mee (m1 ) has a minimum
which differs from zero,
q
mmin
=
| cos 2|m221
(17)
ee

at

cos2 q
(18)
m1 =
m221 .
cos 2
At large m1 it has the asymptotics mee = | cos 2|m1 . (fig. 2 b). As we will see, the
8

existence of a minimal value of |mee | can play an important role in the discrimination of
various scenarios.
Let us consider now the three-neutrino case. The mass mee is given by the sum of
~ (2)
~ (3)
three vectors m
~ (1)
ee , m
ee and m
ee in the complex plane (see fig. 1), so that a complete
cancellation corresponds to a closed triangle. The sufficient condition for having a minimal
value of |mee | which differs from zero for arbitrary non-oscillation parameters is
|m(i)
ee | >

X
j6=i

|m(j)
ee |.

(19)

That is, one of contributions m(i)

ee should be larger than the sum of the moduli of the two
others.
Let us prove that this condition can not be satisfied for the normal hierarchy case.
Indeed, in eq. (19) i can not be 1. For m1 = 0 we have m(1)
ee = 0, at the same time the
2
right-handed side of eq. (19) is larger than zero as long as Ue1
6= 1, (i.e. any non-zero
mixing of e exists). The condition (19) can not be satisfied for i 6= 1 either. In this case
(1)
for a large enough m1 , so that m1 m2 m3 , we get m(i)
ee < mee since Ue1 > Uei . This
proof holds also for schemes with more than three neutrinos. Thus, one can conclude that
neither an upper nor a lower bound on |mee | exists for any oscillation pattern and normal mass hierarchy. Any value |mee | 0 can be obtained by varying the non-oscillation
parameters m1 and ij .
For inverse mass hierarchy we find that condition eq. (19) can be fulfilled for i = 3.
Since now both m3 > m2 , m1 and |Ue3 | > |Ue2 |, |Ue1 | one can get
m3 |Ue3 |2 > m2 |Ue2 |2 + m1 |Ue1 |2

(20)

for any set of values of non-oscillation parameters provided that the mixing of the heaviest
state fulfills
|Ue3 |2 > 0.5.
(21)
Indeed, for large enough m1 , such that m1 m2 m3 , the condition (20) reduces to
|Ue3 |2 > |Ue2 |2 + |Ue1 |2 1 |Ue3 |2 , and the latter is satisfied for (21). For smaller
values of m1 the relative difference of masses m3 > m2 , m1 increases and the inequality of
contributions in eq. (20) becomes even stronger. Thus, the inequality (21) is the sufficient
condition for all values of m1 . This statement is true also for any number of neutrinos. It
is also independent of the relative size of Ue2 and Ue1 .
Summarizing we conclude that
No upper bound on |mee | can be derived from oscillation experiments.
A lower bound exists only for scenarios with inverse mass hierarchy when the heaviest state (3 ) mixes strongly with the electron neutrino: |Ue3 |2 > 0.5.
For normal mass hierarchy certain values of the non-oscillation parameters m1 , j
exist for which mee = 0.
9

The cases of normal and inverse mass hierarchy (they differ by signs of m2 once the
flavor of the states is fixed) can not be distinguished in vacuum oscillations. However, it is
possible to identify the type of the hierarchy in studies of neutrino oscillations in matter,
since matter effects depend on the relative signs of the potential V and m2ij . This will
be possible in future atmospheric neutrino experiments, long base-line experiments and
also studies of properties of the neutrino bursts from supernova [37].

2.2

Effective Majorana mass and the degeneracy of the spectrum

As follows from fig. 2, predictions for mee can be further restricted under assumptions
about the absolute scale of neutrino masses m1 . With increase of m1 the level of degeneracy of the neutrino spectrum increases and we can distinguish three extreme cases:
m21 m221 m231 , in this case the spectrum has a strong mass hierarchy.
m221 m21 m231 , this is the case of partial degeneracy;
Inequality m221 m231 m21 corresponds to strong degeneracy.
There are also two transition regions when m21 m221 and m21 m231 . In what
follows we will consider all these cases in order.

2.3

Effective Majorana mass and present oscillation data

Present oscillation data do not determine precisely all oscillation parameters. The only
conclusion that can be drawn with high confidence level is that the muon neutrino has
large (maximal) mixing with some non-electron neutrino state. The channel is
the preferable one, and it is the only possibility, if no sterile neutrino exists. Thus, in 3
schemes the atmospheric neutrino data are described by oscillations as dominant
mode with
m2atm = (2 6) 103 eV2 , sin2 2atm = 0.84 1,
(22)
and the best fit point
m2atm = 3.5 103 eV2 ,

sin2 2atm = 1.0,

(23)

[29], see also [30]. A small contribution of the e mode is possible and probably
required in view of an excess in the e -like events in the Super-K experiment.
As it was realized some time ago [8], predictions for mee depend crucially on the solution of the solar neutrino problem. The solution of the solar neutrino problem determines
the distribution of the e -flavor in the mass eigenstates, and this affects considerably expectations for the 0-mass . Up to now the unique solution is not yet identified and
there are several possibilities [31], see also [32]:

10

1. Small mixing angle MSW solution with

m2 = (0.4 1) 105 eV2 ,

sin2 2 = (0.2 1.2) 102

(24)

2. Large mixing angle MSW solution with

m2 = (0.1 1.5) 104 eV2 ,

sin2 2 = (0.53 1)

(25)

sin2 2 = (0.8 1)

(26)

3. Low mass MSW (LOW) solution with

m2 = (0.3 2.5) 107 eV2 ,

4. Several regions of vacuum oscillation (VO) solutions exist with

m2 < 109 eV2 ,

sin2 2 > 0.7 .

(27)

There is a good chance that before the new generation of double beta decay experiments starts operation studies of the solar neutrino fluxes by existing and forthcoming
experiments will allow us to identify the solution of the solar neutrino problem. The
key measurements include the day-night effect, the zenith angle dependence of the signal
during the night, seasonal variations, energy spectrum distortions and the neutral current
event rate.
The LSND result [33] which implies
m2LSN D = (0.2 2) eV2 ,

sin2 2LSN D = (0.2 4) 102

(28)

is considered as the most ambiguous hint for neutrino oscillations. The KARMEN [34]
experiment does not confirm the LSND result but it also does not fully exclude this result (see [33]). The oscillation interpretation of the LSND result will be checked by the
MINIBOONE [35] experiment. A simultaneous explanation of the LSND result and of
the solutions of the solar and atmospheric neutrino problems in terms of neutrio mass
and mixing requires the introduction of a forth neutrino. We will discuss the 4 schemes
in section 8.
Summarizing, there is a triple uncertainty affecting predictions of mee :
1. An uncertainty in oscillation parameters. The oscillation pattern does not determine uniquely the 0-mass . Moreover, not all relevant oscillation parameters are
known, so that additional assumptions are needed.
2. An uncertainty in the absolute scale m1 . Some information on m1 can be obtained
from cosmology and may be from direct kinematical measurements.
3. An uncertainty in the relative phases. Clearly, the dependence on the phases is
small in the case if one of the eigenstates gives a dominating contribution to mee .
In what follows we will consider predictions for the 0-mass in schemes of neutrino
masses and mixings which explain the solar and the atmospheric neutrino data. The
schemes differ by the solution of the solar neutrino problem, the type of the hierarchy and
the level of degeneracy. Relative phases are considered as free parameters.
11

10 0
10 -1

111
000
000
111
000
111
000
111

11
00
00
11
00
11

10 -2
m , eV

ATM

1
0
0
1
0
1
0
1

10 -3
10
10

-4
-5

solar
111111
000000
000000
111111
000000
111111

Figure 3: Pattern of neutrino masses and mixing in the scenario with mass hierarchy and
SMA solution of the solar neutrino problem. The boxes correspond to mass eigenstates,
the sizes of different regions in the boxes show admixtures of different flavors. Weakly
hatched regions correspond to the electron flavor, strongly hatched regions depict the
muon flavor, black regions present the tau flavor.

Schemes with normal mass hierarchy

In the case of strong mass hierarchy,

m21 m221 m231 ,

(29)

the absolute mass values of the two heavy neutrinos are completely determined by the
mass squared differences:
m23 = m231 = m2atm , m22 = m221 = m2 .

(30)

The only freedom left is the choice of the value of m1 . In this case the 0-mass is to
a large extent determined by the oscillation parameters.
Since the heaviest neutrino has a mass m3 0.1 eV, the neutrino contribution to
the Hot Dark Matter component of the universe is small: < 0.01. This neutrino
contribution cannot be seen with present and future experimental sensitivity in the CMB
radiation, unless a large lepton asymmetry exists [36]. Oberservational evidence of a
significant amount of the HDM component 0.01 would testify against this scenario.

3.1

Single maximal (large) mixing

In this scheme and are mixed strongly in 2 and 3 (see fig. 3). The electron flavor
is weakly mixed: it is mainly in 1 with small admixtures in the heavy states. The solar
neutrino data are explained by e , resonance conversion inside the Sun. (Notice
that e converts to and in comparable portions.) A small admixture of e in 3 can
lead to resonantly enhanced oscillations of e to in the matter of the Earth.
Let us consider the contributions to mee from individual mass eigenstates. The contribution from the third state, m(3)
ee , can be written in terms of oscillation parameters
12

as

1q

m2atm sin2 2ee ,

(31)
4
where the mixing sin2 2ee 4|Ue3 |2 determines the oscillations of e driven by the atmospheric m2atm . The parameter sin2 2ee immediately gives the depth of oscillation
of the e -survival probability and it is severely constrained by the CHOOZ experiment.
In fig. 4 the iso-mass lines of equal m(3)
ee in the oscillation parameter space are shown,
together with various bounds from existing and future reactor and accelerator oscillation
experiments. The shaded region shows the favored range of m213 from the atmospheric
neutrino data. As follows from fig. 4 in the Super-K favored region the CHOOZ bound
leads to
3
m(3)
(32)
ee < 2 10 eV.
m(3)
ee

For the best fit value of the atmospheric neutrinos the bound is slightly stronger: m(3)
ee <
3
1.5 10 eV.
The mixing sin2 2ee and therefore mee can be further restricted by searches of e
oscillations in the long baseline (LBL) experiments (K2K, MINOS, CERN-Gran-Sasso).
The effective mixing parameter measured in these experiments equals sin2 2e = 4|Ue3 |2 |U3 |2 ,
so that
sin2 2e
,
(33)
sin2 2ee =
|U3 |2
where the matrix element |U3 |2 is determined by the dominant mode of the atmospheric
neutrino oscillations. Using Eq. (33), the value |U3 |2 = 1/2 and the expected sensitivity
to sin2 2e (m2 ) of K2K and MINOS experiments, we have constructed corresponding
bounds in fig. 4. According to fig. 4, these experiments will be able to improve the bound
2
4
on m(3)
eV for a value of m2atm
ee by a factor of 2 - 5 depending on m and reach 2 10
at the present upper bound. For smaller values of |U3 |2 the bound on mee will be weaker.
Taking the smallest value |U3 |2 = 0.3 allowed by the atmospheric neutrino data, we get
that the bound on mee will be 1.7 times weaker. In any case, future LBL experiments will
be able to probe the whole region of sensitivity of even the second stage of the GENIUS
experiment.
A much stronger bound on m(3)
ee can be obtained from studies of neutrino bursts from
Supernovae [37]. A mixing parameter as small as sin2 2ee = 104 can give an observable
6
effect in the energy spectra of supernova neutrinos. This corresponds to m(3)
ee 2 10
eV.
The contribution from the second mass eigenstate is completely determined by the
parameters being responsible for the solution of the solar neutrino problem:
m(2)
ee

1q
m2 sin2 2 .
4

(34)

Taking the 99 % C.L. region of solution (24) we obtain

7
m(2)
105 ) eV,
ee = (5 10

13

(35)

10

-2

MINOS

10

CHOOZ
K2K

KAMLAND

m [eV]

<m>=0.005 eV

-3

10

-2

-1

10 <m>=0.001 eV

sin 2

Figure 4: Iso-mass (hmi|m(3)

ee |) lines (solid lines) in the scheme with hierarchical mass
spectrum. From the upper right downward |m(3)
ee | decreases from 0.01 eV to 0.0001 eV.
Also shown are the regions favored by the SuperKamiokande atmospheric neutrino data
with current bestfit (solid horizontal line) and Kamiokande (lower and upper shaded areas,
respectively, according to [29]) and the borders of the regions excluded by CHOOZ (solid
line) [38] as well as the expected final sensitivity of KAMLAND and K2K (dashed) [39]
as well as of MINOS [40].
and in the best fit point
6
m(2)
ee = 4 10 eV.

(36)

The contribution from the lightest state is

2
3
m(1)
ee = m1 cos m1 m2 < 2 10 eV,

(37)

which can be even larger than m(2)

ee : if the hierarchy between the masses of the first and
the second state is not too strong, m1 /m2 > 102 (for comparison me /m = 5 103 ), we
5
4
get m(1)
eV, with a typical interval m(1)
eV.
ee > 2 10
ee (0.2 2) 10

14

Summing up the contributions (see fig. 5) one finds a maximal value for the 0-mass
mmax
= (2 3) 103 eV
ee

(38)

which is dominated by the third mass eigenstate. No lower bound on mee can be obtained
from the present data. Indeed, Ue3 and therefore m(3)
ee can be zero. The same statement
is true for m(1)
,
since
no
lower
bound
for
m
exists.
The only contribution bounded from
1
ee
(2)
6
below is mee > 10 eV. However, cancellations with the two other states can yield a zero
value for the total mee (see fig. 5).
The following conclusions on future double beta experiments and neutrino oscillations
can be drawn
1). If future experiments will detect neutrinoless beta decay with a rate corresponding
to mee > 2103 eV, the scenario under consideration will be excluded, unless contributions
to 0 decay from alternative mechanisms exist.
2). As we have pointed out, future long-baseline oscillation experiments on e
2
oscillations (MINOS) may further improve the bound on Ue3
and therefore on mmax
by a
ee
factor of 2 5. A much stronger bound may be obtained from supernovae studies [37].
As follows from fig. 4 and from the fact the SMA solution is realized, KAMLAND should
give a zero-result in this scheme.
3). An important conclusion can be drawn if future LBL and atmospheric neutrino
experiments will observe e -oscillations near the present upper bound. In particular, an
up-down asymmetry of the e-like events at Super-Kamiokande is one of the manifestations
of these oscillations [41]. In this case the 3 contribution to mee dominates, no significant
cancellation is expected and the dependence on the relative phases is weak. One predicts
then the result
q
2
2
mee m(3)

U
(39)
ee
e3 matm .

(3)
The observation of 0 decay with mexp
ee mee would provide a strong evidence of the
scheme, provided that the SMA solution will be established. On the other hand it will
be difficult to exclude the scheme if 0 decay will not be observed at the level which
corresponds to mee (39). In this case the scheme will be disfavored. However one should
(1)
take into account also possible cancellations of m(3)
ee and mee , if the mass hierarchy is
weak.

3.2

Bi-large mixing

The previous scheme can be modified in such a way that the solar neutrino data are
explained by the large angle MSW conversion. Now the e flavor is strongly mixed in 1
and 2 .
The contribution from the third state is the same as in the previous scheme (see eq.
3
(31)) with the upper bound m(3)
eV (32).
ee < 2 10
The contribution from the second level,

q
q
1
(2)
2
mee =
1 1 sin 2
m2 ,
(40)
2
15

[eV]
-3

10

-4

10

-5

10

-6

10

mee(3)

mee(2)

mee(1)

Figure 5: Contributions to mee from the individual mass eigenstates for the single
maximal mixing scheme with mass hierarchy. The bars correspond to allowed regions.
m , eV
10 -1

1111
0000
0000
1111
0000
1111

10 -2

10 -3
10
10
10

ATM

-4
-5
-6

solar
1
0

00000
11111
11111
00000
00000
11111

0000
1111
1111
0000
0000
1111
0000
1111

Figure 6: Neutrino masses and mixing pattern of the bi-maximal mixing scheme with
mass hierarchy.
can be significant: both the mixing parameter and the mass are now larger. According to
fig. 7, in the region of the LMA solution of the solar neutrino problem, the contribution
can vary in the interval
3
m(2)
(41)
ee = (0.5 4) 10 eV .

3
(2)
In the best fit point we get m(2)
ee 1.410 eV. Notice that a lower bound on mee exists in
this scheme, provided that sin2 2 < 1. Notice that a day-night asymmetry of about 6 %
3
indicated by the Super-Kamiokande experiment would correspond to m(3)
ee = (1 3) 10
eV.
The contribution m(1)
ee :
2
m(1)
(42)
ee cos m1 ,

where cos2 0.5 0.84, is smaller than in the previous scheme of sect. 3.1, since now
e is not purely 1 and m1 can be as large as 1 103 eV for m1 /m2 < 0.1. Due to the
16

2
m 2 (eV )

<m>=0.005

<m>=0.01

MSW LMA
<m>=0.001

0.002

0.005

0.010
0.015
0.020
0.030
0.040
0.060

0.080
0.100
0.120
0.140
0.160
0.180
0.200

sin 2 2

Figure 7: The iso-mass hmi = m(2)

ee lines, determining the contribution of the second
2
2
state in the m12 sin 212 plane for the hierarchical scheme with the LMA MSW solution. From the upper right downward: |m(2)
ee | decreases from 0.01 to 0.001 eV. Also
shown is the MSW LMA 99 % C.L. allowed region from the combined analysis of the
Homestake, Gallex, Sage and Super-Kamiokande rates and the Super-Kamiokande and
the day-night asymmetry at Super-Kamiokande. The point indictes the best fit value parameters. The horizontal lines correspond to contours of constant day-night assymmetry
[42]. KAMLAND should observe a disappearance signal in this model.
(2)
mass hierarchy m(1)
ee is much smaller than mee (see fig. 8).
Summing up the contributions, we get a maximal value of mmax
7 103 eV. The
ee
typical expected value for mee is in the range of several 103 eV. However, no lower bound
on mee can be obtained on the basis of the present data, although values of mee being
(3)
smaller than 103 eV require some cancellation of the contributions m(2)
ee and mee .

Let us consider possible implications of future results from oscillations and 0 decay
searches:
The observation of mee > (f ew) 102 eV will exclude the scheme.
The non-observation of 0 decay will not exclude the scheme due to possible
cancellations.
The situation can, however, change in the future, if oscillation experiments restrict
(3)
strongly one of the contributions m(2)
ee or mee . Let us discuss possible developments in
this direction:
17

[eV]
-2

10

-3

10

-4

10

-5

10

mee(3)

(2)

mee

mee(1)

Figure 8: Contributions from different mass eigenstates to mee for the bi-large mixing
scheme with mass hierarchy.
Within several years solar neutrino experiments will check the LMA-solution. In
particular, further measurements of the day-night asymmetry and zenith angle distribution at Super-K and SNO could give a decisive identification of the solution
of the solar neutrino problem (see fig. 7). Notice that precise measurements of
the day/night asymmetry can sharpen the predictions of m(2)
ee . Moreover, the LBL
reactor experiment KAMLAND should observe an oscillation effect thus providing
an independent check of the LMA MSW solution.
If MINOS or atmospheric neutrino studies will fix m(3)
ee near the present upper bound,
(2)
one can study the interference effects of mee and m(3)
ee in 0 decay determined
by the relative phases 2 and 3 .

3.3

Scheme with Vacuum oscillation solution

The solar electron neutrinos e oscillate in vacuum into comparable mixtures of and
(fig. 6). The fit to the data indicates several disconnected regions in the m2 sin2 2
- plot. We consider the large m2 region, m2 = (4 9) 1010 eV2 , and sin2 2 > 0.8,
where oscillations allow one to explain an excess of the e-like events in the recoil electron
spectrum indicated by Super-Kamiokande. (Obviously a small m2 will give even smaller
contributions to the effective Majorana mass). In this case

m(2)
m2 sin2 < 2 105 eV.
(43)
ee =
Due to the mass hierarchy and large mixing, the lightest mass eigenstate gives an even
(2)
smaller contribution: m(1)
ee mee . The contribution from the third state is the same as
in Eq. (31) and in fig. 4. For the sum we get
3
mee m(3)
ee < 2 10 eV ,

18

(44)

[eV]
-3

10

-4

10

-5

10

-6

10

mee(3)

mee(2)

mee(1)

Figure 9: Contributions from different mass eigenstates to mee for the bi-maximal mixing
scheme with mass hierarchy.
and clearly, m(3)
ee can be the dominant contribution (fig. 9).
The following conclusions may be drawn:
1). The observation of mee > 102 eV will exclude the scheme. On the other hand
there is no minimal value for mee according to the present data, so that negative results
of searches for 0 decay will have no serious implications for this scheme.
2) A positive signal for atmospheric e oscillations or in the MINOS experiment will
allow us to predict uniquely the value of mee . Then searches for mee will give a crucial
check of the scheme. The absolute scale of the neutrino mass will be fixed.
Similar results can be obtained for the LOW MSW solution. Here the mass squared
difference m221 = 3 107 eV2 implies
4
m(2)
ee < 3 10 eV.

(45)

(2)
Again m(1)
ee mee , and the main contribution may arise from the third state.

Thus, models with normal mass hierarchy lead to rather small values of mee , certainly
below 102 eV. Moreover, the largest value can be obtained in the scheme with the LMA
MSW solution of the solar neutrino problem. The lower bound is of the order 103 eV,
unless cancellation (which looks rather unnatural) occurs. Clearly, only the second stage
of the GENIUS experiment can obtain positive results.

3.4

Triple maximal mixing scheme

In
the scheme of [43] all elements of the mixing matrix are assumed to be equal: |Uij | =
1/ 3 (see fig. 10). The 0-mass is dominated by the contribution from the third state:
m(3)
ee =

1q
m2atm .
3
19

(46)

0
1
1
0
0
1
0
1
0
1

m , eV

0
1
0
10 -11
0
1

0
1
1
0
0
1
0
1
0
1
0
1
0
1
0
1
-2
0
10 1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
0
1
10 -3 1
0
1
0
1
0
1
0
1
0
1
0
1
0
0
1
-4 1
0
0
1
10 1
0
1
0
1
0
1
0
1
0
1
0
1
0
0
1
-5 1
0
1
0
11
00
10 1
0
1
0
1
0
1
0
1
0
1
0
1
0
1

111
000
000
111
000
111

1111
0000
0000
1111

0000
1111

111111111111111111
000000000000000000
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
111111111111111111
000000000000000000
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
solar
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
ATM
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000000000000000000
111111111111111111
000
111
000000000000000000
111111111111111111
000
111
000000000000000000
111111111111111111
000
111
000000000000000000
111111111111111111
000
111

111
000
000
111
000
111

Figure 10:

0
1
1
0
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1

Neutrino masses and mixing in the scheme with threefold maximal mixing.

The best fit of the atmospheric neutrino data in this scheme implies that m2atm
8 104 eV2 and thus
mee 102eV .
(47)
The scheme has rather definite predictions for solar and atmospheric neutrinos. It does
not give a good fit of the data and will be tested by forthcoming experiments.

Schemes with partial degeneracy

In the case of partial mass degeneracy,

m221 m21 m231 ,

(48)

the masses of the two light neutrinos are approximately equal to m1 and the heaviest
mass is determined by the atmospheric mass squared difference:
m1 m2 ,

m3

m231 =

m2atm .

(49)

The interval of masses implied by the condition of partial degeneracy (48) is rather
narrow especially for the LMA and SMA solutions of the solar neutrino problem, when
m231 and m221 differ by two orders of magnitude only. A mass value of m1 > 3 102
eV will shift m3 to larger values, and therefore influence the contribution from the third
eigenstate. We will consider this transition case separately in sect. 6.
The contribution from the third state is the same as in hierarchical schemes (see fig.
4). For the two light states, the contribution can be written as
(2)
2
i2
m(1)
sin2 ) ,
ee + mee m1 (cos + e

(50)

and depending on the relative phase 2 it varies in the interval

(2)
m(1)
ee + mee = m1 (cos 2 1).

20

(51)

[eV]
-2

10

-3

10

-4

10

-5

10

mee(3)

mee(2) mee(1)

Figure 11: Contributions from different mass eigenstates to mee for partially degenerate
scenarios with MSW SMA solution.
This contribution can be further restricted, if the solution of the solar neutrino problem
will be identified. In the case of the SMA MSW solution m(1)
ee dominates; the dependence
on the phase practically disappears and one gets
(2)
m(1)
ee + mee m1 .

(52)

The condition of partial degeneracy implies that the mass m1 should be in the interval:
0.5 102 eV < m1 < 3 102 eV,
and therefore, m1 can reach 3 102 eV at most (fig. 11).
Summing up all contributions we expect mee between 103 and 3 102 eV. Notice
that a lower bound on mee exists here. Near the upper bound the mass mee is dominated
by the contribution from the lightest states and therefore the 0 decay rate will give a
direct measurement of m1 : m1 mee .
2
(3)
Observations of mee larger than m(3)
ee = Ue3 m3 (mee can be determined from oscillation experiments) would favor the scheme, although will not allow one to identify it
unambigously.
Future observations of 0 decay with mee > 3 102 eV will exclude the scheme
testifying for spectra with complete degeneracy or inverse hierarchy (see sect. 5 or 7).
For the LMA solution the typical m221 is bigger than in the SMA case and the
condition of partial degeneracy implies an even narrower interval m1 = (1 3) 102 eV.
Moreover, for m1 at the lower limit of this interval, the difference of light mass eigenvalues
2
can give a substantial correction to formula (50). In the lowest approximation of m
we
m21
get:
2
(2)
2
i2
2
i2 m
m(1)
+
m

m
(cos

+
e
sin

)
+
e
sin2 .
(53)

ee
ee
2m1
The correction (last term in this equation) can be as big as 103 eV and may turn out to
(2)
be important when a cancellation of m(1)
ee and mee occurs.
21

[eV]
-2

10

-3

10

-4

10

-5

10

mee(3)

mee(2) mee(1)

Figure 12: Contributions to mee from different mass eigenstates in schemes with partial
degeneracy and LMA, LOW or VO solutions of the solar neutrino problem. The degenerate states give the main contribution. A complete cancellation of contributions is possible.

Summing up the contributions we find, that the maximal value of mee can be about 3
102 eV as in the case of the SMA solution with similar implications for future 0 decay
searches. In contrast with the SMA case, now due to possible strong cancellations of the
contributions no lower bound on mee can be obtained from the present data (fig.12).
Future oscillation results will allow to sharpen the predictions of mee . In particular, the
solar neutrino experiments will allow to measure a deviation of mixing from the maximal
(2)
3
eV. In
value. The bound 1 sin2 2 > 0.1 would imply that m(1)
ee + mee > 3 10
this case no complete cancellation in mee is possible and a minimum value mee 103
eV appears. The searches for e -oscillations driven by m2atm will further restrict (or
measure) m(3)
ee .
Future studies of the 0 decay can have the following implications: (i) A measurement of mee > 2 102 eV will exclude the scheme. (ii) The non-observation of mee at the
level of 103 eV (second stage of GENIUS) can exclude the scheme if future oscillation
(2)
experiments will lead to a determination of the sum m(1)
ee + mee and a lower bound on mee
will be derived. (iii) If mee will be observed at the level (0.3 2)102 eV (and alternative
schemes which yield a prediction in this interval will be rejected by other observations),
then mee measurements will imply a certain bound in the m1 2 plane.
In the case of the LOW solution m2 is much smaller than for the LMA solution and
m1 can be in the interval m1 = (103 3 102 ) eV. Correspondingly, the contribution
from the two lightest states can be in the wider range (104 3 102 ) eV. The maximal
value for mee can reach 3 102 eV. However it will be impossible to establish a lower
bound on mee even if the solar mixing angle will be measured.
Notice that the LOW solution can be identified by a specific enhancement of the
regeneration effects (in particular the day/night asymmetry) in the lower energy part of
the solar neutrino spectrum. An especially strong effect is expected on the 7 Be-line.
22

For vacuum oscillations the situation is similar to the LOW case.

Schemes with complete mass degeneracy

In schemes with a degenerate neutrino mass spectrum the common mass m1 is much
larger than the mass splittings:
m221 m231 m21 .

(54)

This can be realized as long as m1 > 0.1 eV.

Already the present bound mee < 0.2 0.4 eV [44] implies not too strong degeneracy
unless a substantial cancellation of the contributions in mee occurs. Indeed, if m2atm =
3 103 eV2 , and m1 0.2 eV, we get
m23
m2atm

= 4 102 .
m1
2m22

(55)

For m1 = 0.1 eV, the ratio equals 0.15.

In the case of the SMA solution the e flavor is mainly concentrated in 1 and
2
(2)
(3)
m(1)
ee m1 cos mee mee .

(56)

Numerically, we get mee m(1)

ee m1 > 0.1 eV, i.e. close to the present bound. Basically
one measures m1 by measuring the 0-mass (see fig. 13).
Important conclusions follow from a comparison of the 0 decay results with the
cosmological bounds on the neutrino mass [20], as well as from bounds which follow from
observations of the large scale structure of the Universe. In this scenario one expects

and therefore

mi = 3mee ,

(57)

3mee 2
h .
(58)
91.5eV
Thus the effective Majorana mass, and the Hubble constant are related. This relation
may have the following implications:
1). A significant deviation from equation (58) will exclude the scheme. The present
bound on mee implies 3m1 (0.6 1) eV. If, e.g., data on the large scale structure of the
P
universe will require mi 1 eV, this scheme will be excluded [15, 17].
2). The discovery of 0 decay at the level of the present bound, 0.1 - 0.2 eV,
P
will give mi 0.3 0.6 eV. This range can be probed by MAP and Planck. If these
experiments will put a bound on the sum of neutrino masses below 0.3 eV the scheme
will be excluded.
P
3). This scheme will also be excluded, if cosmological observations will require mi >
0.3 eV, but 0 decay searches will give a bound below mee < 0.1 eV.
=

23

[eV]
1

10

10

-1

10

-2

10

mee(3)

mee(1) mee(2)

Figure 13: Contributions from different mass eigenstates to mee for the scenario with
complete mass degeneracy and the SMA solution of the solar neutrino problem.
Apart from a confirmation of the SMA solution future oscillation experiments will not
influence predictions of mee in this scheme.
Observations of mee at the level 0.1 0.4 eV will be in favor of the scheme.
For the LMA solution a significant cancellation of the contributions from the first and
the second state may occur, resembling the situation in the partially degenerate case. We
have
(2)
2
i2
2
m(1)
sin2 ) , m(3)
(59)
ee + mee m1 (cos + e
ee = m1 |Ue3 | .
2
Since Ue3
0.03, the contribution from the first two states dominates (fig. 15), unless
strong mixing, which has an extremely small deviation from maximal mixing, is introduced: |1 sin2 2 | < 103, which leads to strong cancellation. Thus
(2)
mee m(1)
ee + mee = (cos 2 1) m1 = (0.2 1)m1 ,

(60)

and since m1 > 0.1 eV, we expect for sin2 2 = 0.96:

mee 2 102 eV.

(61)

Notice that some recent studies show that even exact maximal mixing is allowed by the
present data, so that the cancellation can be complete.
Precise measurements of in future oscillation experiments will play a crucial role
for predictions of the mass mee .
If the scheme will be identified, measurements of mee will provide a bound in the
m1 -plane.
The same results hold also for the LOW solution.
For the vacuum oscillations of the -problem, the situation is similar to the one with
2
2
2
LMA MSW. In the strict bi-maximal scheme Ue3
= 0 and Ue1
= Ue2
, so that for 2 =
mee 0 in the limit of equal masses. Small deviations from zero can be related to the
24

3 HDM

m 0 (eV)

CHDM h=0.65 =0.2

CHDM h=0.5 =0.2
CHDM

m=0.5

MAP
<m>=0.1 eV

PLANCK
<m>=0.05 eV

<m>=0.01 eV

MSW LMA

Vacuum oscillations

<m>=0.2 eV

sin2 2

2
Figure 14: Plotted are iso-mass hmi = m1+2
ee lines in the m1 sin 2 plane for the case of
cancellation between the contributions m1 and m2 in degenerate scenarios. Mass splitting
is neglected, since m2 m1 m1 and m3 m1 < 0.1m1 Shown are the bestfit values of m0
for cold+hot dark matter (CHDM) for different values of the Hubble constant (according
to [45]) and cosmological-constant+cold+hot dark matter (CHDM) with m = 0.5
(according to [46]) scenarios. Also shown is the sensitivity to m0 of MAP/Planck according
to [47, 48]. Vertical lines indicate the favored regions and best fits of the LMA and VO
solutions according to [49].

mass difference of m1 and m2 :

1
1eV
1 m2
mee (m1 m2 ) =
1010 eV
.
2
4 m1
m1

(62)

Thus, no unique prediction for mee exists. Although a large value mee > 0.1 eV would
favor degenerate scenarios, the non-observation of 0 decay at the level of 0.1 eV will
not rule out the scheme.
The identification of the scenario will require (i) a strong upper bound on Ue3 , (ii) the
confirmation of the vacuum oscillation solution and (iii) a large mee > 0.1 eV. This would
testify for the case of addition of the 1 and 2 contributions. An upper bound on m1 can
be obtained from cosmology.
If however 0 decay will not be discovered it will be practically impossible to exclude the scenario (and distinguish it from the hierarchical cases), unless cosmology will
25

[eV]
1

10

10

-1

10

-2

10

mee(3)

mee(1) mee(2)

Figure 15: Contributions from different mass eigenstates to mee for the scenario with
complete degeneracy and LMA, LOW or VO solution.
be able to measure neutrino masses down to m1 0.1 eV. The key element is the precise
determination of and its deviation from the maximal mixing.
Let us consider how deviations from the exact bi-maximal case affect the predictions
for the rate in double beta decay experiments. The lower bound on the effective mass
turns out to be
q
mee m1 1 sin2 2 .
(63)

We show this result in fig. 14 as lines of minimal values of mee in the m1 sin2 2 plane.
These lines give the lower bound on values of m1 and the upper bound on values of sin2 2
for which a given value mee can be reproduced. We have shown also the favored regions
of the solar MSW large mixing angle solution as well as the Just-so vacuum oscillation
solution. E.g., a CHDM model with a total m = 0.5 of both cold and hot dark matter
as well as a cosmological constant, and a Hubble constant of h = 0.6 would imply an
overall mass scale of about 0.5 eV. Assuming a mixing corresponding to the best fit of
solar large mixing MSW or vacuum oscillations, sin2 2 = 0.76, this yields hmi = 0.2 0.5
eV. Larger mixing allows for smaller values of mee . In fig. 14 also shown is the sensitivity
of CMB studies with MAP and Planck, which have been estimated to be sensitive to
P
m = 0.5 0.25 eV [48, 47]. For not too large mixing already the present 0 decay
bound ontained from the Heidelberg-Moscow experiment [44] is close to the sensitivity of
these cosmological observations.

Transition regions

There are two intermediate regions

of m1 :
q
2
1) The region with m1 m , where the transition between the hierarchical case
and the case with partial degeneracy occurs. Here m1 m2 m3 .
26

2). The region with m1 m2atm which corresponds to the transition between
partial degeneracy to complete degeneracy: m1 m2 m3 . Here the two lightest states
are strongly degenerate and their contributions are described by eq. (51) in sect. 4. with
the only difference that m1 now can be larger. For m21 m2atm , m1 and therefore mee
can reach 0.1 eV.
The contribution from the third state is also modified:
m(3)
ee

= |Ue3 |

m2atm

m21
1+
m2atm

"

#1/2

(64)

Thus now for the same values of the oscillation parameters the contribution m(3)
ee can be
2
2
1/2
[1 + m1 /matm ] (1 2) times larger.
In fig.16 we show the dependence of the individual contributions to mee on m1 . For
(3)
mee only the upper bound is used; the two other lines represent possible values of m(1)
ee
and m(2)
ee for certain neutrino mixing parameters. We show also the maximal and the
minimal possible values of mee .
The position of the first transition region is determined by the specific solution of
the solar neutrino problem: According to fig. 16, m1 = (2 15) 103 eV for the LMA
solution, m1 = (1 9) 103 eV for the SMA MSW solution and m1 = (1 10) 105 eV
for the VO solution.
The position of the second transition region, m1 = (3 20) 102 eV, is similar in all
the cases.
The upper bounds on mee as functions of m1 have a similar dependence for all the cases.
The lower bounds are different and depend on specific values of oscillation parameters.
Thus, for the LMA solutions a lower bound exists in the range of mass hierarchy
(m1 < 103 eV) if the solar mixing angle is sufficiently large (see fig. 16 b)). In this
case the contribution from 2 dominates and no cancellation is possible even for maximal
2
possible m(3)
ee . In contrast, for a lower sin 2 the cancellation can be complete so that
no lower bound appears (see fig. 16 a)).
In the first transition region all states contribute with comparable portions to mee ,
thus cancellation is possible and no lower bound exists.
In the second transition region as well as in the completely degenerate case the first
and the second state give the dominating contributions to mee and the increase of m3
does not influence significantly the total mee . The mass mee is determined by m1 and .
Moreover, a larger sin2 2 implies a larger possible range of mee for a given m1 (fig. 16
a,b)).
Let us consider the SMA MSW solution (fig. 16 d)). In the mass hierarchy region the
third state gives the main contribution and no lower bound exists. A lower bound on mee
appears at m1 > 1.5 103 eV and at m1 > 102 eV the mass mee is given by m1 .
In the case of the VO solution (fig. 16 c) the upper bound on mee is given by m(3)
ee up
to m1 2 104 eV. In the range of partial degeneracy the contribution from the first and
the second states become important. No lower bound on m(3)
ee can be established from the
present data in the whole range of m1 .
27

m ee
0.01

a)

0.001

0.0001

0.00001

0.00001

m ee
0.01

0.0001

0.001

0.01

0.1

m1

0.0001

0.001

0.01

0.1

m1

b)

0.001

0.0001

0.00001

0.00001

mee
0.01

c)

0.001

0.0001

0.00001

m1
0.00001

0.0001

0.001

0.01

28

0.1

m ee
0.01

d)

0.001

0.0001

0.00001

0.00001

0.0001

0.001

0.01

0.1

m1

mee
0.2
0.15

e)

0.1
0.07
0.05

0.03
0.02
0.015
0.01

m1
0.00001

0.0001

0.001

0.01

0.1

m ee
0.2
0.15

f)

0.1
0.07
0.05
0.03
0.02
0.015
0.01

m1
0.00001

0.0001

0.001

0.01

29

0.1

m ee
0.2

g)

0.15
0.1
0.07
0.05
0.03
0.02
0.015
0.01

mee

0.00001

0.0001

0.001

0.01

0.1

0.00001

0.0001

0.001

0.01

0.1

m1

0.2
0.15

h)

0.1
0.07
0.05

0.03
0.02
0.015
0.01

m1

Figure 16: mee (eV) as a function of m1 (eV) for three-neutrino mixing. Shown are
(2)
(3)
the contributions m(1)
ee (dashed), mee (dotted) and mee (interrupted dashes). The solid
lines correspond to mmax
and mmin
and show the allowed region for mee . Panels a)-d)
ee
ee
correspond to the case for normal hierarchy, panels e)-h) inverse hierarchy. Shown are
2
2
the cases a) and e) LMA MSW with Ue2
= 0.2, b) and f) LMA MSW with Ue2
= 0.4, c)
2
2
and g) vacuum oscillation with Ue2 = 0.5 and d) and h) SMA MSW with Ue2 = 7 103 .
2
The mixing of the third state is varied from zero to its upper bound, Ue3
= 2.5 102 .

Scheme with inverse mass hierarchy

Let us consider the partially degenerate spectrum with

m23 m22 = m2atm ,

m21 m22 ,

m223 = m2 ,

(65)

so that the mass of the second and third neutrino are determined from the atmospheric
neutrino data. The e flavor is concentrated in the heavy states (inverse mass hierarchy).
A small admixture of e in the lightest state can exist (fig. 17 ).
The contribution to mee from the first state equals
2
m(1)
ee = m1 Ue1 .

(66)

The inequality m21 m2atm implies m1 < 2 102 eV for m21 /m22 < 0.1. Using then the
2
2
CHOOZ result which restricts (in schemes with inverse hierarchy) Ue1
: Ue1
< 2.5 102 ,
30

10 1
10 0

m , eV

10

0000
1111
0000
1111
1111
0000
0000
1111
0000 0000000
1111
0000
1111
1111111
0000
1111
0000000
1111111
0000000
1111111
0000000
1111111

solar

-1

10 -2
111
000
000
111
000
111

10 -3

HDM
1111111
0000000
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
ATM
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
0000000
1111111
00
11
00
11
00
11

10 -4

Figure 17: The neutrino mass and mixing pattern in the bi-large mixing scheme with
inverse mass hierarchy.
we get
4
m(1)
eV .
ee < 5 10

(67)

The sum of the contributions from the two heavy degenerate states can be written as
(3)
m(2)
ee + mee

m2atm (sin2 + ei23 cos2 ) ,

(68)

where 23 2 3 . For the SMA solution we get from eq. (68)

mee

m(2)
ee

m(3)
ee

m2atm = (4 8) 102 eV

(69)

and in the bestfit point of the atmospheric neutrino data: mee 6 102 eV. This means,
that the predicted value of m2ee coincides with m2atm (fig. 18). This coincidence provides
a unique possibility to identify the scheme (see also, e.g. [12]).
The relation m2ee = m2atm applies also for the case of the LMA solution as long as
2 = 0 2 .
For the LMA solution the sum of the contributions from the two heavy states lies in
the interval
q
(3)
m(2)
+
m
=
(cos
2

1)

m2atm .
(70)

ee
ee

For sin2 2 < 0.98 we get mee > 4 103 eV which is still much larger than m(1)
ee . The
(2)
(3)
compensation can be complete if the mixing is maximal. The value mee + mee < 2 103
eV requires a very small deviation from maximal mixing: 1sin2 2 < 2103. Thus, the
lower bound on mee can be further strengthened, if the deviation from maximal mixing
will be established.
A similar consideration holds for the cases of LOW MSW or vacuum oscillation solutions (see fig. 19).
2

Notice that if the mass degeneracy originates from some flavor blind interactions one may indeed
expect that the masses of 2 and 3 have the same phase.

31

[eV]
-1

10

-2

10

-3

10

-4

10

mee(1)

mee(2) mee(3)

Figure 18: Contributions to mee from different mass eigenstates in the scheme with inverse mass hierarchy and SMA solution. The degenerate states give the main contribution,
implying a unique prediction for mee .
[eV]
-1

10

-2

10

-3

10

-4

10

mee(1)

mee(2) mee(3)

Figure 19: Contributions to mee from different mass eigenstates in the schemes with
inverse hierarchy and LMA, LOW or VO solutions. Now cancellation between the degenerate states is possible leading to a wide range of values allowed for mee .
The contribution of the two heavier eigenstates to the HDM, = (2m1 ) /(91.5 eVh2 )
0.01, is rather small and below the reach of future projects on measurements of cosmological parameters.
If the e admixture in the lightest state is non-zero, so that the e -oscillations driven
by m2 exist, the scheme can be identified by studying matter effects in atmospheric and
supernova neutrinos as well as in the long-baseline experiments.
Indeed, in the case of inverse mass hierarchy the e 3 level crossing (in matter)
occurs in the antineutrino channel, so that in supernovae the antineutrinos e will be
strongly converted into a combination of , and vice versa. This leads to a hard e s
spectrum at the Earth detector which coincides with the original spectrum [37].
In atmospheric neutrinos the identification of the type of mass hierarchy will be possible if the sensitivity will be enough to detect oscillation effects in e-like events (electron
32

neutrinos and antineutrinos). It will be also important to measure the sign of the electric
charge of the lepton, since the matter effects are different in the neutrino and antineutrino
channels and this difference depends on the type of mass hierarchy.
These matter effects can be studied in LBL experiments [39, 40] with neutrinos from
neutrino factories where beams of neutrinos and antineutrinos are well controlled.
Let us consider the dependence of the predictions for meeqon m1 . In the schemes with
inverse hierarchy there is only one transition region: m1 m2atm , that is m1 m2
m3 or m1 = (1 8) 102 eV. The sum of the contributions from the second and the third
states dominates in the whole range of m1 . It
q is determined by the solar mixing angle
and m2 m3 . The latter changes from m2atm in the hierarchical region to m1 in
the region of complete degeneracy (see fig. 16 e-h). The mass mee is completely predicted
in terms of m2 for the SMA solution (fig. 16 h).
No lower bound on mee appears when the (solar) mixing parameter is maximal or
close to maximal.

Four neutrino scenarios

The introduction of new (sterile) neutrinos mixed with the usual SU(2) doublet neutrinos opens new possibilities for the construction of the neutrino mass spectrum and for
the explanation of the data. It also modifies predictions of mee . Here we will consider
several scenarios which are motivated both by phenomenology and theory. All scenarios
we will discuss contain one or two (degenerate) states in the range relevant for structure
formation in the universe and/or for the LSND oscillations.

8.1

Scenario with small flavor mixing and mass hierarchy

The scheme (fig. 20) is characterized by a mass hierarchy:

m4 = mHDM , m3

m2AT M ,

m2

m2 , m1 m2 .

(71)

The states and s are strongly mixed in the second and fourth mass eigenstates, so
that s oscillations solve the atmospheric neutrino problem. All other mixings are
small. In particular, the solar neutrino problem is solved by small mixing MSW conversion
e , s .
The main motivation for this scheme is to avoid the introduction of large mixing
between flavor states and to keep in this way as much as possible correspondence with
the quark sector.
A clear signature of the scheme is the s oscillation solution of the atmospheric
neutrino problem. The solution can be tested by (i) studies of the neutral current interactions in atmospheric neutrinos, in particular, N N 0 (with N = n, p), which gives
the main contribution to the sample of the so called 0 events (the rate should be lower in
33

10

m , eV

10 -1

1111
0000
0000
1111
0000
1111
0000
1111

s
e

1
0
0
1
0
1

10 -2

0
1
1
0
0
1
0
1

ATM
1
0
0
1
0
1
0
1

10 -3
10 -4

HDM

11111111
00000000
00000000
11111111
00000000
11111111
00000000
11111111

solar
2

Figure 20: Neutrino masses and mixing in the 4 scenario with small flavor mixing and
mass hierarchy. Here the white parts of the boxes correspond to admixtures of the sterile
state.
the s case); (ii) studies of the zenith angle distribution of the upward going muons
(stopping and through-going); (iii) detection of the leptons produced by converted .
Recent Super-Kamiokande data do not show a deficit of 0 events, and moreover the
oscillations give a better fit (of about 23) of the zenith angle distribution thus
favoring the interpretation. However, more data are needed to draw a definite
conclusion (see [30]).
The novel element of this scheme (compared with the 3 - schemes discussed in the
previous sections) is the existence of a heavy state in the HDM range. Its contribution to
mee equals:
2
m(4)
(72)
ee = Ue4 m4 .
The relevant parameters, Ue4 and m4 , can be determined from studies of the short
range
e e oscillations (disappearance) driven by the largest mass splitting m4 m2
mHDM . For this channel the effective mixing angle equals
sin2 2ee = 4|Ue4 |2 (1 |Ue4 |2 ) 4|Ue4 |2 ,

(73)

so that

1
m2 sin2 2ee .
(74)
4
The corresponding iso-mass lines in the m2 sin2 2 plot together with various oscillation
bounds are shown in fig. 21.

In the cosmologically interesting range, m2 mHDM (0.5 5) eV, the mixing

is constrained by the BUGEY experiment: sin2 2ee = (2 4) 102. Therefore we get the
upper bound
2
m(4)
(75)
ee (2 5) 10 eV .
m(4)
ee

There is no strict relation between mee and the parameters of the e oscillations
since both relevant mixing elements Ue3 and U3 are small. Indeed, now the effective

34

m [eV]

(4)

(4)
(4)
m ee =0.01 eV m(4)
ee =0.05 eV m ee =0.1 eVm ee =0.2 eV
10

m(4)
ee =0.005 eV

BUGEY

10

-1

10

-2

m(4)
ee =0.001 eV

10

-1

sin 2

Figure 21: Iso-mass |m(4)

ee | lines in the four-neutrino scenario with small flavor mixing and
mass hierarchy. The shadowed area shows the region for neutrino masses of cosmological
interest as HDM. Also shown are the regions excluded by the reactor experiment BUGEY
(from [50]).
depth of oscillations is determined by sin2 2e = 4|Ue3 |2 |U3 |2 , so that
m(4)
ee =

sin2 2e mHDM
.
4|U4 |2

(76)

It is impossible to infer useful information from this unless the U4 will be determined
from other experiments. Taking the bound |U4 |2 < 0.25 from the 3 - analysis of the
atmospheric neutrino data, we get from eq. (76) the lower bound
2
m(4)
ee > sin 2e mHDM .

(77)

To get an estimation we assume sin2 2e 103 which corresponds to upper bounds

on the elements Ue4 and U4 from the BUGEY experiment and searches for
oscillations. (Notice that LSND result can not be completely explained in this scheme.)
This leads to
3
>
m(4)
(78)
ee 10 eV .
The contributions from the three light states are similar to the contributions in the 3
single maximal mixing scheme with mass hierarchy (sect. 3.1). Inqparticular, the largest
2
contribution may come from the third mass eigenstate: m(3)
< 2 103
m2atm Ue3
ee =
eV (see eq. (32)). The contributions from the two lightest states can be estimated as
7
3
m(2)
105 ) eV and m(1)
eV.
ee = (5 10
ee < 2 10
35

[eV]
10
10
10
10
10

-1

-2

-3

-4

-5

mee(1)

mee(2)

mee(3)

mee(4)

Figure 22: Contributions to mee from different mass eigenstates in the 4 scheme with
small flavor mixing and mass hierarchy.
Thus, the 0-mass (see fig. 22) can be dominated by the contribution of the heaviest
2
state which can reach mee m(4)
eV.
ee 5 10
The contribution depends strongly on the mixing angle sin2 2e . Short baseline experiments (such as the rejected short baseline neutrino oscillation proposal TOSCA) could in
principle test the region of large masses mHDM 10 eV down to sin2 2e = 103 , which
correspond to an improvement of the upper bound on mee by 1 - 2 orders of magnitude.
Due to possible cancellations between the contributions no lower bound on mee can be
obtained from the present data.
Notice that the MINIBOONE experiment will probe the mixing angle sin2 2e down
to 4 104 eV and thus will check the LSND result. A confirmation of the LSND result
will exclude this scheme.

8.2

Scenario with two heavy degenerate neutrinos

The main motivation for this scenario (see fig. 23) is to explain the LSND result along
with oscillation solutions of the solar and atmospheric neutrino problems [51, 52]. The
masses are determined as
m3 m4

m2LSN D ,

m2

m2 ,

m1 m2 .

(79)

The neutrinos and are strongly mixed in the two heavy mass eigenstates 3 and
4 , so that oscillations solve the atmospheric neutrino problem. The two other
neutrinos, e and s , are weakly mixed in the two lightest mass states and the resonance
e s conversion solves the solar neutrino problem.
The two heavy neutrinos with masses m3 m4 can be relevant for cosmology, their
contribution to a hot dark matter component equals:
q

mHDM = 2m3 = 2 m2LSN D .

36

(80)

10

1
1111
0000
0000
1111
0000
1111
0000
1111

m , eV

10 -1

s
e

10 -4

11
00
00
11
00
11
00
11

00
11
11
00
00
11

ATM

10 -2
10 -3

HDM

LSND
11
00
00
11
00
11
00
11

solar

11111111
00000000
00000000
11111111
00000000
11111111
00000000
11111111

Figure 23: The pattern of the neutrino mass and mixing in the scheme with two degenerate neutrinos and one sterile component.
In this scheme the new element is the existence of two heavy degenerate states. Let us
consider in details their contribution to mee (the effect of the two lightest states is small).
Using relations (79) we can write this contribution as
q

(4)
2
2 i34
) m2LSN D
m(3)
ee + mee (|Ue3 | + |Ue4 | e

(81)

and 34 4 3 is the relative phase of the 3 and 4 masses. Let us express the
masses in eq. (81) in terms of oscillation parameters. In short base-line experiments the
only oscillation phases which enter are the ones between heavy states and light states.
One can neglect the oscillation phase between the two light states which is determined by
m2 and the phase between the two heavy states which is determined by m2atm . In this
case the oscillations are reduced to two neutrino oscillations with a phase determined by
m2LSN D and the width for the e channel:

sin2 2e = 4|Ue3
U3 + Ue4
U4 |2 .

(82)

Let us consider two extreme situations: suppose an admixture of the e flavor in one of
the heavy states is much larger than in the other one, e.g. |Ue3 | |Ue4 |, then sin2 2e =
|Ue3 |2 |U3 |2 , and therefore |Ue3 |2 = sin2 2e /|U3 |2 . In this case we get from eq. (81)
(4)
2
m(3)
ee + mee |Ue3 | m3 , and consequently,
(4)
m(3)
ee + mee

sin2 2e q
m2LSN D .
|U3 |2

(83)

Since |U3 |2
0.5 is determined by atmospheric neutrino oscillations, taking sin2 2e =
q
(4)
3
eV.
2 103 and m2LSN D = 1 eV we find m(3)
ee + mee 10
2

(4)
Let us
now take Ue3 U3 Ue4 U4 , then sin 2e = 16|Ue3
U3 | and m(3)
ee + mee
q
sin2 2e m2LSN D /2, provided that the two contributions are in phase. This result is
two times smaller than the result in the previous case.
37

[eV]
-2

10

-3

10

-4

10

-5

10

mee(1)

mee(2)

mee(3+4)

Figure 24: Contributions to mee from different mass eigenstates for the 4 scheme with
two degenerate pairs of neutrinos and normal mass hierarchy.
For the e e channel we find the depth of oscillations
sin2 2ee = 4U+ (1 U+ ) 4U+ ,

(84)

where U+ |Ue3 |2 + |Ue4 |2 . At the same time mee U+ m2LSN D /2, so that
m(3)
ee

m(4)
ee

q
1 2
sin 2ee m2LSN D .
4

(85)

(4) <
2
Using the BUGEY bound on sin2 2ee we get m(3)
eV.
ee + mee 10
Since cancellations may show up, no lower bound can be obtained.
The same combination of neutrino mixing matrix elements (84) determines the e
mode of oscillations in atmospheric neutrinos. It will lead to an overall suppression of the
number of the e-like events.
The contributions from the light states are similar to those in the 3 case (see sect.
7
3
3.1): m(2)
105 ) eV and m(1)
eV.
ee = (5 10
ee 2 10
Summing up all the contributions we get that the 0-mass can be at most (f ew)
102 eV being dominated by the contribution of the heavy states at the upper bound
(fig. 24). A coincidence of a 0 decay signal in this range with a confirmation of
the LSND oscillations by MINIBOONE can be considered as a hint for this scheme. At
the same time, since cancellation between different contributions can show up, no lower
bound on mee exists. Thus, a non-observation of 0 decay of the order of magnitude
(f ew) 102 eV does not rule out the scheme.

A similar situation appears in the Grand Unification scenario [53, 54] which is
characterized by strong mixing of and s in the two heavy states and mixing of e and
in the two light states (fig. 25). Here the atmospheric neutrino problem is solved by
s oscillations whereas the solar neutrino data are explained by e conversion.
38

10

111
000
000
111
000
111
000
111

00
11
11
00
00
11
00
11

1
0
0
1
0
1

10 -1
m , eV

s
e

ATM

LSND

10 -2

Figure 25:

8.3

0
1
1
0
0
1
0
1

solar

10 -3
10 -4

2 HDM

11111111
00000000
00000000
11111111
00000000
11111111
00000000
11111111

The neutrino masses and mixing in the Grand Unification scenario.

Scenario with inverse mass hierarchy

The mass hierarchy in the two schemes with two pairs of states with small splitting can
be inverse. In the first case, e and s flavors are concentrated in the two heavy states 3
and 4 , whereas and are in the two light states. The dominating contribution comes
from the third state which almost coincides with e :
mee m(3)
ee

m2LSN D 0.4 1eV.

(86)

Thus in the context of this scheme the double beta decay searches check immediately the
LSND result, and in fact, already existing data disfavor the scheme.
Another possibility of the inverse hierarchy is that the e and flavors are concentrated in the heavy states, whereas and s are in the pair of light mass states whose
splitting leads to the atmospheric neutrino oscillations. The situation is similar to that
for the 3 scheme with inverse hierarchy (see sect. 7) with the only difference that m2atm
should be substituted by m2LSN D :
(4)
m(3)
ee + mee

m2LSN D (sin2 + ei23 cos2 ) .

(87)

The third and the fourth mass eigenstates give the dominating contributions. Thus the
expected interval for the total effective mass is
mee

m2LSN D (cos 2 1).

(88)

This interval can be probed already by existing experiments, although for large mixing
angle solutions of the solar neutrino problem (LMA, LOW, VO) strong cancellation can
occur.

Discussion and Conclusions

We have performed a general analysis of the dependence of the effective Majorana mass
on the oscillation and non-oscillation parameters. Systematic studies of contributions
39

mee (eV)
10

Heidelberg-Moscow

10

-1

CUORE
MOON
GENIUS 1 t

-2

10

GENIUS 10 t

Hierarchy

Degeneracy Partial Degeneracy Inverse Hierarchy

INV

2DEG

HIER

VAC

LMA

SMA

VAC

LMA

SMA

VAC

LMA

SMA

Triple

VAC

LMA

-3

SMA

10

4v

Figure 26: Summary of expected values for mee in the different schemes discussed in
this paper. The expectations are compared with the recent neutrino mass limits obtained
from the HeidelbergMoscoscow [44] experiment as well as the expected sensitivities for
the CUORE [57], MOON [58] proposals and the 1 ton and 10 ton proposal of GENIUS
[13].
from the individual mass eigenstates have been performed. We also have considered
future developments in view of forthcoming oscillation results. A systematic study of
predictions from various schemes allows us to compare these predictions and to conclude
on implications of future double beta decay searches.
In fig. 26 we summarize the predictions for mee in various schemes considered in this
paper. We also show the present upper bound of 0 decay experiments [44] and regions
of sensitivity which can be reached in future double beta decay experiments. Future double
beta decay projects such as GENIUS [55, 13, 56], CUORE [57], MOON [58] will lead to
a significant improvement of the sensitivity. The most ambitious and at the same time
most realistic project, GENIUS, will test mee down to 2 102 eV in the one ton version
with one year of measurement time and down to 2 103 eV in the 10 ton version with 10
years of measurement time.
According to figure 26 there are two key scales of mee , which will allow one to discriminate among various schemes: mee 0.1 eV and mee = 0.005 eV.
1). If future searches will show that mee > 0.1 eV, then the schemes which will survive
are those with neutrino mass degeneracy or 4 schemes with inverse mass hierarchy. All
other schemes will be excluded.
2). For masses in the interval mee = 0.005 0.1 eV, possible schemes include: 3
schemes with partial degeneracy, triple maximal scheme, 3 schemes with inverse mass
40

hierarchy and 4 scheme with one heavy (O (1 eV)) neutrino.

3). If mee < 0.005 eV, the schemes which survive are 3 schemes with mass hierarchy,
schemes with partial degeneracy, and the 4 schemes with normal hierarchy. The schemes
with degenerate spectrum and inverse mass hierarchy will be excluded, unless large mixing
allows for strong cancellations. For mee < 0.001 eV this applies also for schemes with a
partial degenerate spectrum, again unless large mixing occurs.
Future oscillation experiments will significantly reduce the uncertainty in predictions
for mee and therefore modify implications of 0 decay searches. Before a new generation
of 0 decay experiments will start to operate we can expect that
The solution of the solar neutrino problem will be identified. Moreover, m2 and
sin2 2 will be determined with better accuracy. In particular, in the case of the
solutions with large mixing (LMA, LOW, VO) the deviation of 1 sin2 2 from
zero can be established.
The dominant channel of the atmospheric neutrino oscillation ( or s )
will be identified. The mass m2atm will be measured with better precision.
A stronger bound on the element Ue3 will be obtained or it will be measured if
oscillations of electron neutrinos to tau neutrinos driven by m2atm will be discovered
in the atmospheric neutrino or LBL experiments.
The LSND result will be checked by MINIBOONE.
In the following we summarize possible consequences of these oscillation results. The
conclusions obtained are not always stringent enough to exclude or prove any of the
solutions in the whole parameter space. In these cases we use phrases like favor or
disfavor to describe the situation.
1). Let us first comment on how the identification of the solution of the solar neutrino
problem will modify implications of the 0 decay searches.
If the SMA solution of the solar neutrino problem turns out to be realized in nature,
a value of mee > 0.2 eV will imply a completely degenerate neutrino mass spectrum
or schemes with inverse mass hierarchy. The measured value of mee will coincide
with m1 and will fix the absolute mass scale in the neutrino sector. A confirmation
of this conclusion can be obtained from the CMB experiments MAP and Planck, if
the degenerate neutrino mass is larger than 0.1 eV.

For lower values, mee = 2 103 102 eV, a scheme with partially degenerate
spectrum will be favored. Again, we have mee = m1 and the mass scale can be
fixed.
For even lower mass values: mee < 2 103 eV, or, after MINOS improved the bound
4
<
on m(3)
eV, with the contribution m(3)
ee , mee 4 10
ee a new parameters enters,
which for larger mee could be neglected. Thus it will be impossible to quantify the
contribution of each single state to mee , unless m(3)
ee will be fixed in atmospheric or
LBL oscillations.
41

 If the LMA solution of the solar neutrino deficit turns out to be realized in nature,
a value mee > 2 102 eV will testify for a scenario with degenerate mass spectrum.
A confirmation of this result will be obtained from the CMB experiments MAP and
Planck, if the degenerate neutrino mass is larger than 0.1 eV. Using the mixing
angle determined in solar neutrino experiments the range for the absolute mass
scale can be determined from mee according to fig. 14. A value of mee < 2 102
eV will favor schemes with partial degeneracy or hierarchical spectrum. As soon as
mee < 2 103 eV m(3)
ee becomes important and enters as a new parameter and it
will be difficult to reconstruct the type of hierarchy.
If the LOW or VO solution is the solution of the solar neutrino problem, the situation
is similar to the MSW LMA case. The only difference is, that an observed 0mass mee > 2 103 eV will imply a partially or completely degenerate scheme.
Below this value the type of hierarchy can not be identified until bounds on m(3)
ee
will be improved.
For schemes with inverse mass hierarchy the situation can be more definite:
q

If the MSW SMA solution turns out to be true, a value of mee = m2atm =
(5 8) 102 eV is expected. This value coincides with m1 m2 and therefore will
give the absolute mass scale. For larger masses: mee > 8 102 eV the transition to
a completely degenerate spectrum occurs.
If the MSW LMA, MSW LOW or vacuum oscillation solution is realized, a value
of mee = (0.02 8) 102 eV will testify for inverse mass hierarchy. The interval of
expected values of mee can be narrower once the deviation of 1 sin2 2 from zero
will be measured in solar neutrino experiments.
For larger values of masses: mee > 8 102 eV the scheme approaches the degenerate
case.
2). The discovery of a sterile neutrino will have significant impact on the implications
of the double beta decay searches.
The existence of a sterile neutrino can be established by a confirmation of the LSND
result in MINIBOONE, or by a proof of the e s oscillations solution of the solar
neutrino problem by SNO, or by studies of the atmospheric neutrinos.
For 4 scenarios the interpretation of the 0 decay results is rather ambiguous. A
value of mee > (f ew) 102 eV will favor the intermediate mass scale scenario, while
a value of mee < 103 eV will favor a scenario with two degenerate pairs of neutrinos
and normal mass hierarchy. A value of mee > 101 eV will clearly disfavor a strongly
hierarchical scheme with normal mass hierarchy and favor the cases of inverse hierarchy
or degeneracy. In all cases it will be difficult to disentangle the single contributions and
to identify a specific spectrum. Important input in this case may come from the CMB
experiments MAP and Planck by fixing the mass of the heaviest state.

42

3). Ue3 : further searches for e oscillations in atmospheric neutrinos, LBL and reactor
experiments will allow one to measure or further restrict this mixing element. This, in
turn, will be important for sharpening the predictions for mee especially in the schemes
with strong mass hierarchy.
4). Matter effects and hierarchy: Studies of matter effects on neutrino oscillations will
allow to establish the type of mass hierarchy, which in turn is of great importance for
predictions of mee .
We can conclude from this summary, that in 3 scenarios any measurement of mee >
2 103 eV in 0 decay (corresponding to the final sensitivity of the 10 ton version of
GENIUS) will provide informations about the character of hierarchy of the neutrino mass
spectrum and in some cases also to fix the absolute mass scale of neutrinos. For values of
mee < 2 103 eV no reconstruction of the spectrum is possible until the contribution m(3)
ee
will be fixed or bounded more stringent in atmospheric or LBL neutrino oscillations. For
four-neutrino scenarios it will be not that easy to fix the mass scale of the neutrino sector.
Crucial informations can be obtained from tests of the LSND signal and cosmology.
As has been mentioned before, a non-zero 0 decay rate always implies a nonvanishing neutrino Majorana mass [4]. Let us comment finally on possible ambiguities in
the interpretation of a positive signal in neutrinoless double beta decay in terms of mee , in
view of the existence of different alternative mechanisms, which could induce neutrinoless
double beta decay, such as R-parity violating SUSY, right-handed currents, or leptoquarks. While no absolute unique method to identify the mechanism being responsible
for neutrinoless double beta decay exists, the following remarks can be done:
1). Many of the possible alternative contributions require new particles, e.g. SUSY
partners, leptoquarks, right-handed W bosons or neutrinos having masses in or below the
TeV range, which to date not have been observed. Thus one expects to observe effects of
new particles at future high energy colliders as the LHC or the NLC, giving independent
informations on possible contributions to 0 decay (keeping in mind an uncertainty in
nuclear matrix elements of about a factor of O(2)). Notice that the same new interactions
mentioned here may induce effects in neutrino ocillations and imply ambiguities in the
interpretation of the data also there (see e.g. [59, 60]).
2). Using different source isotopes in different experiments and figuring out the values
of 0 decay nuclear matrix elements for different contributions may help to identify the
dominant one. Also a future experiment being sensitive to angular correlations of outgoing
electrons could be useful in the discrimination of different contributions. Observing a
positive signal in 0 decay should encourage new experimental efforts to confirm the
results.
3). Last but not least and as discussed in this paper, a non-zero 0 decay signal
can be related to some experimental results (both positive and negative) in neutrino
oscillations and cosmology. A coincident and non-contradictory identification of a single
neutrino mass scheme from the complementary results in such different experiments thus
should be respected as a strong hint for this scheme.
43

In conclusion, after Super-Kamiokande has established large mixing in atmospheric

neutrinos, the simplest neutrino spectrum with strong mass hierarchy and small flavor
mixing (which typically predicts an undetectable mee ) is excluded. Now the neutrino
mass spectrum can exhibit any surprise: it can have a normal or inverse mass hierarchy,
be partially or completely degenerate. More than three mass eigenstates can be involved
in the mixing. In view of this more complicated situation a detection of a positive signal
in future double beta decay searches seems to be rather plausible. We have shown that
for a given oscillation pattern any value of mee is possible, which is still not excluded by
the experimental bounds on the neutrinoless double beta decay half life limit. (A lower
bound on mee appears in the case of inverse mass hierarchy.) This means that even after
all oscillations parameters will be measured no unique prediction of mee can be derived.
On the other hand this means that double beta decay searches provides informations
being independent on informations obtained from oscillation experiments. Combining
the results of double beta decay and oscillation searches offers a unique possibility to shed
some light on the absolute scale of the neutrino mass, the type of hierarchy and the level
of degeneracy of the spectrum. If we want eventually to reconstruct the neutrino mass
and flavor spectrum, further searches for neutrinoless double beta decay with increased
sensitivity seem to be unavoidable.
Note added: When this paper has been prepared for submission the papers of ref.
[61] appeared which discuss similar topics.

References
[1] M. Doi, T. Kotani, E. Takasugi, Progr. Theor. Phys. Suppl. 83 (1985) 1;
[2] T. Tomoda, Rep. Progr. Phys. 54 (1991) 53
[3] H.V. KlapdorKleingrothaus, H. Pas, hep-ph/0002109, talk given at COSMO99,
Trieste/Italy; H.V. KlapdorKleingrothaus, hep-ex/9901021, Proc. Int. Conf. on
Lepton- and Baryon Number Non-Conservation, Trento, Italy, April 20-25 1998, IOP
Bristol (1999) ed. H.V. Klapdor-Kleingrothaus and I. Krivosheina, p.251
[4] J. Schechter, J.W.F. Valle, Phys. Rev. D 25 (1982) 2951
[5] D. London, hep-ph/9907419, talk given at Beyond the Desert 99, Castle Ringberg,
Tegernsee, Germany, June 1999, IOP Bristol 2000, ed. H.V. Klapdor-Kleingrothaus
and I. Krivosheina
[6] V. Lobashev et al. Phys. Lett B. 460 (1999) 227; Ch. Weinheimer et al., Phys. Lett
B. 460 (1999) 219
[7] A. A. Klimenko, A. A. Pomansky, A. A. Smolnikov, Proceedings of the Moriond-1985
Workshop on Perspectives in Electroweak Interactions. v. 2 (1985) 303.
[8] S.T. Petcov, A.Y. Smirnov, Phys. Lett. B. 322 (1994) 109
44

[9] S. Bilenky, A. Bottino, C. Giunti, C. Kim, Phys. Rev. D54 (1996) 1881-1890
[10] S.M. Bilenki, C. Giunti, C.W. Kim, M. Monteno, Phys. Rev. D57 (1998) 6981-6988;
hep-ph/9904328
[11] C. Giunti, Phys. Rev. D 61 (2000) 036002 (hep-ph/9906275); S. Bilenky, C. Giunti,
W. Grimus, hep-ph/9809368 Neutrino 98, Nucl. Phys. B 77 (Proc. Suppl.) (1999)
ed. Y. Suzuki and Y. Totsuka; S. Bilenky, C. Giunti, hep-ph/9904328 Proc. WIN99,
Cape Town, south Africa, 24-30 January (1999)
[12] S.M. Bilenky, C. Giunti, W. Grimus, B. Kayser, S.T. Petcov, hep-ph/9907234,
Phys.Lett. B465 (1999) 193-202
[13] J. Hellmig, H.V. KlapdorKleingrothaus, Z. Phys. A 359 (1997) 351; H.V. Klapdor
Kleingrothaus, M. Hirsch, Z. Phys. A 359 (1997) 361; H.V. KlapdorKleingrothaus,
J. Hellmig, M. Hirsch, J. Phys. G 24 (1998) 483;
[14] S. Bilenky, C. Giunti, C. Kim, S. Petcov, Phys. Rev. D54 (1996) 4432-4444
[15] R. Adhikari, G. Rajasekaran, hep-ph/9812361, Phys. Rev. D 61 (2000) 031301
[16] H. Minakata, O. Yasuda, Phys. Rev. D 56 (1997) 1692-1697
[17] H. Minakata, O. Yasuda, Nucl. Phys. B 523 (1998) 597-610
[18] F. Vissani, hep-ph/9708483
[19] F. Vissani, hep-ph/9904349, Proc. of the 6th Tropical Seminar on Neutrino and
Astroparticle Physics, May 17-21 1999, San Miniato, Italy
[20] V. Barger, K. Whisnant, Phys. Lett. B 456 (1999) 194-200
[21] J. Ellis, S. Lola, hep-ph/9904279 Phys. Lett. B 458 (1999) 310-321
[22] G.C. Branco, M.N. Rebelo, J.I. Silva-Marcos, Phys. Rev. Lett. 82 (1999) 683-686
[23] H. Georgi, S. L. Glashow, hep-ph/9808293
[24] T. Fukuyama, K. Matsuda, H. Nishiura hep-ph/9708397
[25] T. Fukuyama, K. Matsuda, H. Nishiura, hep-ph/9708397; Mod. Phys. Lett. A13
(1998) 2279
[26] T. Fukuyama, K. Matsuda, H. Nishiura; Phys. Rev. D57 (1998) 5844
[27] F. Vissani, JHEP 9906 (1999) 022, hep-ph/9906525
[28] M. Czakon, J. Gluza, M. Zralek, hep-ph/9906381 Phys. Lett. B 465 (1999) 211-218

45

[29] T. Kajita, talk given at Beyond the Desert 99, Castle Ringberg, Tegernsee, Germany,
June 1999, IOP Bristol 2000, ed. H.V. Klapdor-Kleingrothaus and I. Krivosheina
[30] N. Fornengo, M.C. Gonzalez-Garcia, J.W.F. Valle, hep-ph/0002147,
[31] J. N. Bahcall, P. I. Krastev, A. Yu. Smirnov, to be published
[32] M.C. Gonzalez-Garcia, P.C. de Holanda, C. Pena-Garay, J.W.F. Valle, hepph/9906469
[33] C. Athanassopoulos et al., Phys. Rev. Lett. 81 (1998) 1774-1777 K. Eitel, hepex/9909036;
[34] T.E. Jannakos, hep-ex/9908043
[35] A.O. Bazarko, hep-ex/9906003
[36] J. Lesgourgues, S. Pastor, S. Prunet, hep-ph/9912363
[37] A.S. Dighe, A.Yu. Smirnov, hep-ph/9907423
[38] M. Apollonio et al. (CHOOZ collab.), hep-ex/9907037, Phys. Lett. B 466 (1999)
415-430
[39] K. Zuber, hep-ex/9810022
[40] MINOS
detectors
technical
design
report,
version
1.0,
October 1998, http : //www.hep.anl.gov/ndk/hypertext/minostdr.html; we thank
Kai Zuber for bringing this reference to our attention
[41] E.Kh. Akhmedov, A. Dighe, P. Lipari, A.Yu. Smirnov, Nucl. Phys. B 542 (1999) 3
[42] J. N. Bahcall, P. I. Krastev, A. Yu. Smirnov, Phys. Rev. D 60 (1999) 093001
[43] P.F. Harrison, D.H. Perkins, W.G. Scott, Phys. Lett. B 196 (1997) 186
[44] HeidelbergMoscow Collab., Phys. Rev. Lett 83 (1999) 41
[45] E. Gawiser, J. Silk, Science 280 (1998) 1405-1411
[46] J.R. Primack, M.A.K. Gross, astro-ph/9810204, Proc. of the Xth Recontres de Blois
The birth of galaxies, June 28 - July 4 1998
[47] R.E. Lopez, astro-ph/9909414
[48] D.J. Eisenstein, W. Hu, M. Tegmark astro-ph/9807130, Submitted to ApJ.
[49] J.N. Bahcall, P.I. Krastev, A.Yu. Smirnov, Phys. Rev. D 58 (1998) 096016
[50] L. di Lella, hep-ex/9912010
46

[51] J.T. Peltoniemi, J.W.F. Valle, Nucl. Phys. B 406 (1993) 409
[52] D.O. Caldwell, R.N. Mohapatra, Phys. Rev. D 48(1993) 3259
[53] J.T. Peltoniemi, D. Tommasini, J.W.F. Valle, Phys.Lett.B298:383-390,1993
[54] A.S. Joshipura, A.Yu. Smirnov, Phys.Lett.B439:103-111,1998
[55] H.V. KlapdorKleingrothaus, in Beyond the Desert 97 Accelerator and NonAccelerator Approaches (Eds. H.V. KlapdorKleingrothaus, H. Pas), Proc. Int. Workshop on Particle Physics beyond the Standard Model, Castle Ringberg, June 8-14,
1997, IOP Publ., Bristol, Philadelphia, p. 485 and Int. J. Mod. Phys. A 13 (1998)
3953
[56] H. V. Klapdor-Kleingrothaus, L. Baudis, G. Heusser, B. Majorovits, H. Pas, hepph/9910205
[57] E. Fiorini et al., Phys. Rep 307 (1998) 309
[58] H. Eijiri et al., nucl-ex/9911008
[59] S. Bergmann, Nucl.Phys. B515 (1998) 363-383
[60] N. Fornengo, M. C. Gonzalez-Garcia, J. W. F. Valle, hep-ph/9906539
[61] K.Matsuda, N.Takeda, T.Fukuyama, H.Nishiura, hep-ph/0003055; W. Rodejohann,
hep-ph/0003149 M. Czakon, J. Gluza, M. Zralek, hep-ph/0003161.

47