Present and future strategies for neutrinoless double beta decay searches

Abstract

The renewed interest shown in these days towards neutrinoless double beta decay, a lepton number violating process which can
take place only if neutrinos are Majorana particles (ν = $
\bar \nu
$
\bar \nu
) with a nonvanishing mass, is justified by the fact that the Majorana nature of neutrinos is expected in many theories beyond
the Standard Model. We also now know, thanks to the neutrino oscillation experiments, that neutrinos are in fact massive,
as expected in these theories and not requested in the Standard Model. Moreover, since neutrino oscillation experiments measure
only the absolute value of the difference of the square of the neutrino masses, the discovery of neutrinoless double beta
decay would help to disentangle questions that still remain unsolved: what is the absolute mass scale of the neutrinos and
which mass hierarchy (normal, inverted or quasi-degenerate) is the correct one?

The scope of this paper is not only to review the present results reached in the field by the different groups and technologies
worldwide, but also to illustrate and comment on the (near and long-term) future strategies that experimentalists are trying
to pursue to reach the needed sensitivity required to explore the inverted hierarchy neutrino mass scale.

KeywordsNeutrinoless double beta decay-neutrino physics-nonaccelerator particle physics-rare nuclear decays

Full text

Page 1

PRAMANA
— journal of
physics
c ? Indian Academy of SciencesVol. 75, No. 2
August 2010
pp. 271–280
Present and future strategies for neutrinoless
double beta decay searches
C BROFFERIO
Physics Department, University of Milano-Bicocca and INFN, Sez. di Milano-Bicocca,
Piazza della Scienza 3, I-20126 Milano, Italy
E-mail: chiara.brofferio@mib.infn.it
Abstract.
decay, a lepton number violating process which can take place only if neutrinos are Majo-
rana particles (ν = ¯ ν) with a nonvanishing mass, is justified by the fact that the Majorana
nature of neutrinos is expected in many theories beyond the Standard Model. We also
now know, thanks to the neutrino oscillation experiments, that neutrinos are in fact mas-
sive, as expected in these theories and not requested in the Standard Model. Moreover,
since neutrino oscillation experiments measure only the absolute value of the difference of
the square of the neutrino masses, the discovery of neutrinoless double beta decay would
help to disentangle questions that still remain unsolved: what is the absolute mass scale
of the neutrinos and which mass hierarchy (normal, inverted or quasi-degenerate) is the
correct one?
The scope of this paper is not only to review the present results reached in the field
by the different groups and technologies worldwide, but also to illustrate and comment
on the (near and long-term) future strategies that experimentalists are trying to pursue
to reach the needed sensitivity required to explore the inverted hierarchy neutrino mass
scale.
The renewed interest shown in these days towards neutrinoless double beta
Keywords. Neutrinoless double beta decay; neutrino physics; nonaccelerator particle
physics; rare nuclear decays.
PACS Nos23.40.-s; 14.60.Pq; 11.30.Fs
1. Introduction
Double beta decay (DBD) is a nuclear process in which an initial nucleus (Z,A)
with proton number Z and total nucleon number A decays to (Z±2,A). When first
proposed, in 1935 by Goeppert-Mayer [1], the double beta decay (DBD) was thought
of as two simultaneous β-decays in a nucleus, with the emission of two positrons
(electrons) and two (anti)neutrinos, corresponding to a second-order nuclear process
in the framework of Fermi’s theory [2]. The main actor was the unstable nucleus,
with an incredibly long half-life (1021y or more), and not the neutrino itself. Even
when Furry observed [3] that the Majorana theory of neutrino nature [4] would
imply the possibility of a neutrinoless DBD (0νDBD), where only two electrons
are emitted in the decay, the interest was focussed on the possibility to really
271

Page 2

C Brofferio
observe this kind of rare nuclear process, since the expected half-life in this case
was calculated to be at least 105y shorter than what predicted by Goeppert-
Mayer. That conclusion however had to be revised with the discovery of parity
nonconservation in weak interactions: the simple decay channel proposed by Furry
required the Majorana neutrino to be massive, to adjust its helicity at the second
vertex.
With the development of the effective-field theory and grand unification schemes
in the late 1970s and early 1980s [5], the expectation that neutrinos are Majo-
rana particles, therefore identical to their antiparticles, and that they have nonzero
mass, rose in the scientific community and engendered renewed interest in 0νDBD
experiments. In the 1990s, the first direct evidences of 2νDBD of a bunch of nuclei
were collected, starting from82Se [6], followed by100Mo,150Nd,48Ca and oth-
ers, but there was still no confirmation of the existence of 0νDBD. People became
skeptical and looked at DBD searches as difficult precise measurements to better
understand nuclear physics of some particular nuclei, but of no particular interest
for fundamental particle physics.
It is with the discovery of neutrino oscillations, definitively demonstrated by Su-
perkamiokande and SNO at the beginning of this century [7], that imply finite mass
neutrinos, that 0νDBD search has gained the centre stage as the only experiment
able to disentangle the real nature of neutrinos: are they Dirac or Majorana parti-
cles? Moreover, since neutrino oscillation experiments are sensitive to the squared
mass differences among the three mass eigenstates (m2
absolute values, their evaluation will have to come from elsewhere, and 0νDBD
could play an important role in it. Provided the sensitivity of 0νDBD searches is
good enough, they could give fundamental information on the neutrino, even with
the nonobservation of the decay. But all these subtleties and details have already
been exhaustively discussed by other authors, and the interested reader is invited
to refer to [8,9].
i− m2
j), but not to their
2. Experimental approaches in DBD studies
The experimental parameter, which can be extracted from the data is the 0νDBD
rate which, in the case of pure neutrino mass mechanism, can be expressed as
1
T0ν
1/2
= G0ν|M0ν|2?mν?2,
(1)
where G0νis the phase-space term, calculable with high accuracy and scaling ap-
proximately with the fifth power of the Q-value of the reaction, |M0ν| is the nu-
clear matrix element (NME) [10], responsible for the systematic uncertainties on
the global calculation and strongly dependent on the particular nuclide under ex-
amination and ?mν? is the effective Majorana mass, which can in turn be expressed
by
?mν? =
3
?
j=1
mjU2
ejexp(iαi),
(2)
272
Pramana – J. Phys., Vol. 75, No. 2, August 2010

Page 3

Present and future strategies for 0νDBD searches
where mj are the three neutrino mass eigenvalues, Uej forms the first line of the
PMNS neutrino mixing matrix and exp(iαi) represents the neutrino intrinsic CP
parities.
Equation (2) constitutes the bridge between 0νDBD and neutrino flavour oscilla-
tions [8]. What we know about Uejis enough to state that, even in the case of van-
ishing of the lowest of the three mj’s, ?mν? remains finite, with ?mν? = O(10−2) eV
in the case of inverse mass hierarchy and ?mν? = O(10−3) eV in the case of direct
mass hierarchy: it is the first time in neutrino history that a definite goal is set for
0νDBD search, not so far from the achievable experimental sensitivities, at least in
the inverse hierarchy case, and if neutrinos are quasidegenerate, then the situation
would be even better.
The expected signature in a 0νDBD experiment, measuring the sum of the kinetic
energy of the two electrons emitted, is a peak at the transition Q-value, since the
recoil of the daughter nucleus is always negligible, to be extracted from a continuum
background. In order to evaluate the sensitivity of a given experiment to the 0ν half-
life of the investigated nucleus, the following factor of merit can be defined:
?
corresponding to the maximum half-life detectable against 1σ background fluctua-
tion. Here, n is the number of active atoms per source molecule, η is the isotopic
abundance, ? is the detector efficiency, NAis the Avogadro number, A is the mole-
cular mass of the source, M is the detector mass, T is the measurement time, ∆E
is the energy resolution and B is the background per unit mass, time and energy.
Note that for zero-background experiments, F0νscales linearly with MT and loses
any dependence from B and ∆E. Keeping eq. (3) in mind, we can now discuss
the strategies of DBD experimentalists in the past, when the information on the
neutrino oscillations was not yet available or at least not so well defined, dark
matter could still be (partially) composed of neutrinos, and direct neutrino mass
measurements were even suggesting the possibility of a mass around few eV.
F0ν= ln 2nη?NA
A
MT
B∆E(68% CL) (3)
3. The past: Strategies up to 1–10 kg of DBD source
Aiming at reaching sensitivities of the order of 1023y, that seemed enough to
discover 0νDBD if the neutrino was to have a mass of the order of 1–10 eV, the
historical approach of the experimentalists was first of all to choose a high Q-value
decay, since this increases the phase space strongly and therefore the expected
decay rate. If the Q-value is above 2.6 MeV, then there is an extra bonus: natural
background radioactivity, principally coming from238U and232Th chains, falls
down more than two orders of magnitude after the208Tl peak.
The second step was to select an isotope for which the theoreticians had calcu-
lated a favourable NME. Until recently, many theoreticians were not involved in
this difficult task and, moreover, little effort was done to try to collaborate in un-
derstanding the origin of the differences. Therefore, till a few years ago you could
find up to a factor 10 difference among different calculations (which is a lot, since
NME is linearly related to the effective Majorana mass). Even if, more correctly,
Pramana – J. Phys., Vol. 75, No. 2, August 2010
273

Page 4

C Brofferio
you looked at the same nuclear structure model and calculation, you could find a
factor up to 5 between two isotopes. So this was also taken into account in the
choice of the isotope to be studied.
Experimentalists had then another ace to play: the choice of the detector. Two
main philosophies can be used: either you try to enhance the efficiency as much
as possible, choosing a detector that is made of, or contains homogeneously, the
DBD source (active detector, AD), or you bet on the background discrimination
with a tracker detector that has the DBD source outside (passive detector, PD).
Ge diodes, as the ones used in the Heidelberg-Moscow (HdM) or IGEX experi-
ments, and bolometers, like CUORICINO, were of the first type (AD) and these
have excellent energy resolution and their detector mass is totally active for DBD.
Calorimetric tracking detectors, like NEMO3, are of the second type (PD) and have
chosen event topology, therefore background reduction, at the cost of efficiency and
energy resolution. Many other experiments have been realized, trying to exploit
several of the known techniques in elementary particle detection, and different fea-
tures of the event, like energy, momenta, event topology, were registered, with
higher sensitivity on this or that parameter, according to the detector properties of
each specific search. A reasonably complete review of the past experiments can be
found in [11–14].
Among them, four deserve a short comment. The most sensitive experiment that
has been running is HdM, that was stopped in 2003 and reached an exposure of 71.7
kg·y. Part of the Collaboration claimed an evidence of 0νDBD with a half-life of
2.23×1025y [15]. This result has been heavily controversial, as reported in [12,14].
The first claim came in fact in 2001 with only half of the statistics and a signal at
the level of 2.2σ on a background of 0.17 c/keV·kg·y. Only when integrated on a
doubled exposure, with signal shape analysis to discard Compton events, a heavy
use of Monte Carlo to understand and subtract background and the use of neural
networks to discriminate background events, the statistical significance grew above
six sigmas (11 counts on an almost null background).
Unfortunately, the direct competitor (IGEX) was terminated in 1999, with a
collected exposure of less than 9 kg·y and could not confirm nor contradict the
result. CUORICINO, a bolometer experiment with TeO2crystals, that ran from
2003 to 2008, could have observed a signal in accordance with the HdM result,
even with its smaller exposure of 18 kg·y, thanks to the more favourable phase
space and a similar background (0.18 c/keV·kg·y) and resolution (0.3% FWHM
compared with 0.15%), but the null result obtained is not in contrast with the
claimed evidence because of the uncertainties on the NME of the two different
nuclei and the presence of background fluctuations. Therefore, the HdM peak is
still waiting for a confirmation (or a disclaim...). CUORICINO, with its limit on
130Te half-life of 2.94 × 1024y at 90% CL, has for the first time, after the two Ge
experiments, entered clearly the region below 1 eV for the effective Majorana mass,
with a limit of ?mν? < 0.2–0.7 eV, where the uncertainty comes from different
calculations of the NME.
The only other experiment that will be able to enter this region before the next
generation detectors start to take data is NEMO3, that is still running. It is a
20-sector calo-tracker PD: the energy released is measured with plastic scintilla-
tors read by photomultipliers and the track is reconstructed through a system of
274
Pramana – J. Phys., Vol. 75, No. 2, August 2010

Page 5

Present and future strategies for 0νDBD searches
GM wire chambers. The DBD sources are in the form of metal foils (100Mo) or
sandwiched between thin mylar foils and are seven per sector. There are seven
isotopes under study with this detector [16], but only two are present in a rea-
sonable amount to put stringent (and competitive) limits on 0νDBD. These are
100Mo, with almost 7 kg occupying more than half of the source slots, and 1 kg of
82Se. The present (preliminary) limits, after 3.85 y of live time, are respectively
1.1×1024y and 3.6×1023y at 90% CL. Thanks to the reconstruction of the event
topology they can quote a background limit of 2 c/keV·ton·y, but at the cost of
an 8% FWHM energy resolution and a 14% efficiency, to be compared with the
95% and 86% of HdM and CUORICINO, respectively. Anyhow, the advantage of
the reconstruction of the event, that can distinguish between the energy released
by one electron and the other, can be of extreme importance in the comprehension
of the decay structure. For instance, the NEMO Collaboration was able to confirm
experimentally [17] the single-state dominance model of the 2νDBD, where only
the first 1+level dominates the decay. In a similar way they could in principle
recognize between a light Majorana neutrino exchange and a right-handed current
scenario as the dominant channel of the 0νDBD, and they were able to measure
with high accuracy the 2νDBD of all the isotopes under study in their detector.
4. The present: Strategies up to 100–200 kg of DBD source
The clear involvement of DBD in neutrino physics after the discovery of neutrino
oscillations has urged the experimentalists in the rush towards a sensitivity at the
level of 1026y, with which the quasidegenerate hierarchy hypothesis can be assessed
and the76Ge claim scrutinized with high accuracy. But this effort would be useless
if the NME were still known with big uncertainties. In an effort to solve this prob-
lem, recently there have been many discussions and exchanges between theoretical
groups, to understand discrepancies and to evaluate errors. When available, β de-
cay and 2νDBD experimental data have been used to fix parameters in QRPA, and
other approaches, like shell model (SM) or interacting boson model (IBM) are now
applied to calculate the same NMEs in an independent way. This resulted in an
improvement on the calculations that ended in 2007 [18] to a very good agreement
between different approaches using QRPA, that were reasonably confirmed also by
the recent work with IBM [19]. The SM approach maintains a somewhat large
difference, that was partially solved just this year. In fact, some experimental data
on the differences in neutron vacancies and proton occupancies between76Ge and
76Se, obtained by the cross-section of p,n adding or removing transfer reactions
showed some contradiction with what was expected by the QRPA approach. Once
the discrepancy is solved, the NME calculated with all the approaches become sim-
ilar within a factor 30% [20]. Similar positive influence on the 2νDBD calculations
was obtained in the last years from the experimental data of charge exchange re-
actions [21]. There are big expectations on the possibility to see reasonably soon a
convergence to a unique result for the calculation of NMEs of interest for DBD.
Meanwhile experimentalists are trying to improve on the sensitivity a factor 100.
Looking at eq. (3) one has to concentrate the efforts on a consistent gain in mass
and background. Among the funded experiments, now under construction, aiming
Pramana – J. Phys., Vol. 75, No. 2, August 2010
275