From the Early Days until the Gauge Theory Era

1

Chapter 1

Double Beta Decay - Historical Retrospective and Perspectives

The history of double beta decay using the nucleus as a complicated laboratory for a wide range of particle physics starts more than 70 years ago. This history is closely connected with fundamental discoveries of particle physics, such as the discovery of parity non-conservation [Lee56, Wu57) and of the gauge theories [Gla79, Sa179, Wei79), and double beta decay research has become one of the most important fields of non-accelerator particle physics. Table 1.1 sketches this historical development. Starting from purely nuclear physics questions, double beta decay yields at present a broad access to many topics of physics beyond the Standard Model of particle physics at the TeV range, on which new physics should manifest itself (see Fig. 1.1 and Table 1.2). Recent experiments complement in many ways the search for new physics at the Large Hadron Collider (LHC) and at Next Linear Colliders (NLC), and could serve as an important bridge between the physics that will be gleaned from high energy accelerators such as LHC and NLC on the one hand, and satellite experiments such as MAP and PLANCK on the other. Concerning neutrino physics, without double beta decay there would never have been a solution of the nature of the neutrino (Dirac or Majorana particle) and of the structure of the neutrino mass matrix, since neutrino oscillation experiments measure only differences of neutrino mass eigenstates. Only investigation of neutrino oscillations and double beta decay together can lead to an absolute mass scale. It is the aim of this book to describe the way double beta research went, and in particular the progress achived in the last 15 years, and to give an insight into the contribution and the further potential of double beta decay for present and future particle physics and cosmology.

1.1 1.1.1

From the Early Days until the Gauge Theory Era

The First Steps in Double Beta Research

The long and close association between the phenomenon of nuclear double beta decay, the question of lepton number conservation and the nature and mass of the neutrino began shortly after the 'discovery' of the neutrino by W. Pauli in 1930

Seventy Years of Double Bet a Decay

2

1=20 billion years Today to

";3 K 11 meV)

Galaxy formation Recombination CMS (MAP&PLANCK)

Relic radia tion decouples (CBR)

Matter domination

Nucleosynthesis

Quark-hadron transition

Electroweak phase transition T=10'GeV The Particle Desert

Axions. supersymmetry?

DARK MAnER SEARCH LHC Range RpSUSY Leploquarks. Composltenee mv > ml is the mass of the heavy right-handed neutrino, associated with the large grand unified mass scale governing the breaking of leptonnumber conservation. Thus double beta experiments may probe mass scales M of the order of 10 3 - 10 9 Ge V, i.e. ranges far beyond the highest energies directly accessible in present and future colliders (see Fig. 1.1) .

Fig. 1.8 (right).

Tsung-Dao Lee in 1956 (left) , Chien-Shung Wu (middle) and Chen-Ning Yang in 1956

Another attractive and most natural way to produce small neutrino Majorana masses has been shown by R. Mohapatra and F. Senjanovic [Moh80*-I], [Moh81aj. They obtain in a left-right symmetric model with spontaneous parity non-conservation, based on the gauge group BU £(2) x SU R(2) x U(l), a relation between the neutrino mass and the mass mWR of the right-handed W boson mVe ~ me 2 / gmwR' A similar formula holds for leptons in each generation. This formula is suggestive in the sense t hat , in the limit of mWR going to infinity, the neutrino mass goes to zero and we have at the same time a pure V- A theory of weak interactions. In superstring models the smallness of the Majorana neutrino masses must be traced back to much more complicated mixing mechanisms than the see-saw mechanism, since there are no Higgs fields to generate a sufficiently large Majorana mass for the right-handed neutrino. For this, an intermediate mass scale of order of magnitude Te V is required. This may be identified with the scale of supergravity breaking [Moh86b*-I], [Moh88a**-I], [Va187], [Ber87j. The essential role of stringderived symmetries for ensuring light neutrino masses has been stressed by J.e. Pati [Pat96j.

From the Early Days until the Gauge Theory Era

Neutrino masses in superstring-derived standard-like models have been discussed in [Far93] . Ultralight neutrinos and R-parity in superstring models were considered by [Va187a]. Superstring GUTs lead - since they do not contain high-dimensional Higgs representations - naturally to t he flipp ed SU(S) theory [Ant87], [Ant88], based on a SU(5) x U(l) group [Ant92]' [Lop94a]. The discussion of neutrino mass patterns in a flipped SU(S) model derived from a 'realistic' string theory has revived in the light of recent Superkamiokande data [Ell98] .

Fig. 1.9 Left: Murray Gell-Mann in Harvard University. Center: Mir iam Cvetic wit h t he author, at t he BEYOND'99 Conference, Castle Ringberg, Germany, 1999 (foto author) . Right: Tsutomu Yanagida (foto kindly presented by T . Yanagida).

Fig. 1.10 Left: John Ellis (CERN) and t he author at the DARK'2002 Conference, Cape Town, South Africa, 2002. In middle: Rabindra N . Mohapatra, at t he Dark'98 Conference, Heidelberg, Germany, 1998. Right: Zurab Berezhiani, at t he LEPTON- BARYON '98 Conference, Trento, Italy, 1998 (with the author) (fotos left and right aut hor) .

Neutrino masses in SUSY models have been widely discussed, e.g. by [Leo95], [Vem2000], [Che2000**-I], [Der2000**-I]' [Bla99**-I]' [Cho97a], [Ach95]' [Asa96]. It has been noted that SUSY extensions of the Standard Model with a see-saw mechanism have a general tendency to (strongly) enhance lepton flavour violating (LFV) processes. Therefore it is expected t hat observation of neutrinoless double beta decay and of some charged LFV phenomena would indicate realization of SUSY see-saw in nature [Mas04] .

11

Seventy Years of Double Beta Decay

12

A stringy origin of neutrino masses and a 'gravity-induced ' see-saw mechanism within the minimal supersymmetric standard model has been discussed by [Cve92], [Cve92a *-I]. It is based on an interplay of nomenormalizable and renormalisable terms in the superpotential and may be realized in a class of string vacua. Light neutrino masses are naturally explainable also in theories with large TEVscale extra dimensions [Die2000**-I], [Die2000a]. Here the heavy right -handed neutrino will reside in the bulk of Kaluza-Klein states which leads t o a power-law suppression of the neutrino mass relative to the masses of all other fermions. Oscillations could become possible without v masses in D > 5 theories. Double beta decay in such scenario has been considered by [Moh99**-IV]. Another kind of string origin of neutrino masses, and of sterile neutrinos was discussed by Z. Berezhiani and R.N. Mohapatra, and R. Foot and R.R. Volkas [Ber95*-I], [Fo095], [Ber99a**-I]. The request for a light sterile neutrino naturally leads to the concept of a shadow world, assuming exact duplication of the standard model in both the gauge field and the fermion content. The only connection connecting known and shadow sector is here gravitation, and the mirror neutrinos mix with the ordinary ones through Planck scale induced higher-order operators.

BLACK BOX I

d '}e i

: :I

f.: 0 [Al103**-III] hinges on assuming that galaxy clusters require low as-values (shaded horizontal band) and dissolves when using more reasonable uncertainties in the cluster constraints (see http:// space.mit.edu/home/tegmark/).

Using the critisized way of analysis with respect to the value of L: mv from WMAP, an artificial problem has been raised, which now is discussed in quite a number of papers, which consequently see a 'tension' or a disagreement between the cosmological and the (3(3 data (HEIDELBERG- MOSCOW result) (see, e.g. [Han05], [Fog04], [Me105], [Pie04]). This 'tension' is not existent when using the less biased analysis given by [Teg04**-III]. Also usually in this context a trivial point often has been ignored (see, e.g. [Lin06], [Fog04], [Me105], [Pie04]). It is known since more than 20years (and has been mentioned earlier - see Eq. 1.4) that in principle double beta experiments yield a value of the (effective) neutrino mass only under the assumption of vanishing right-handed weak currents (and of other more exotic contributions to the 01.1(3(3 amplitude) (see e.g. [Mut88**-I], [Pae99a*-II, Pae2000*-II]' [Kla06**-V, Kla06a*V]). In general a 01.1(3(3 experiment can give only an upper limit for the effective mass. So, there is in any case no problem of any discrepancy between cosmological (or other) neutrino masses and results from double beta decay. Neutrino Mass and Cosmologica l Concordance Model: It might be mentioned finally, that a recent alternative cosmological "concordance model" based on an Einstein-de Sitter universe with zero cosmological constant, requires relic neutrinos with the mass of order of eV [Bla03*-III].

Seventy Years of Double Beta Decay

34

1

0.8 0.6 0.4 0.2 0.2 0.4 0.6 0.8

1

F ig . 1.28 Upper part: Left: Max Tegmark (see http: //space.mit.edu/home/tegmark/album. html). Right : Constraints on individual cosmological quantities using WMAP alone (shaded yellow d istributions) and including SDSS information (narrower red distribut ions). Each distribution shown has been marginalized over all other quantities in the class of 6-parameter (T, Sl J\,Wd,w b, A s, n s ) "vanilla" models as well as over a galaxy bias parameter b for the SDSS x2 case (courtesy M. Tegmark). The horizont al dashed lines indicate e- / 2 for x = 1 and 2, respectively, so if the distribution were Gaussian, its intersections with these lines would correspond to la and 2a lim its , respectively. Lower part: Left: Shown above are the spiral galaxies NGC 3788 (top) and NGC 3786 (bottom) in the constellation Ursa Major (home of the Big Dipper) . These two galaxies, like m a ny found throughout the SDSS survey, are gravitationally interacting. Interactions such as these are commonly observed by t he SDSS. They are t hought to lead to the formation of elliptical galaxies, and it will probably be the fate of the Milky Way a nd M31 in the distant future (from ARC & the SDSS Collaboration, www.sdss.org). R ight: The SDSS Survey telescope stands out against the breathtaking backdrop of the Sacramento Mountains (permission from www-v isualmedia.fna l. gov).

1.3.3

Other Key Experiments Contributing to the Determination of the Neutrino Mass and the Neutrino Mass Matrix: 1. Supernova 87A neutrinos; 2. Solar neutrinos; 3. Atmospheric neutrinos; 4. {3 Decay and {3{3 Decay; 5. v-Oscillations at Accelerators

Finally we mention some other historical key experiments which contributed to our understanding of t he 1/ mass.

Double Beta Decay, Neutrino Mass Models and Cosmological Parameters -

Status and Prospects

35

(1) Supernova 87A neutrinos: A laboratory independent limit for mVe of rvl2eV has been deduced ([Arn87, Ko187, Sat87] etc.) from the millenium event of observation of neutrinos from the supernova SN87A (see, e.g., [Ale88**-III], [Lor02**-III], [Kri04 ** -III]). (2) Solar neutrinos: The first observation of solar neutrinos was performed by the Davis experiment [Dav68, Cle98**-III]' which was operated over rv25years in the Homestake mine. This historical experiment delivered the first evidence for neutrinos from the Sun, and for the so-called solar neutrino problem, i.e. for neutrino oscillations. The related history of solar models has been described in [Bah02a**-III]. The final papers of the SNO experiment proving solar neutrino oscillations independently of the solar models have been published by [Ahm04]. The 'robustness' of the interpretation of solar v data as oscillations is investigated in [Va106**-III], [Bah02]. The connection of neutrino oscillations, fluctuations, solar magneto-gravity waves, and solar interior has been investigated by [Bur03**-III, Bur03a**-III]. How Many I:'s in the Solar Neutrino Effect?: It may be historically of interest that the sum of the five experiments prior to SNO (Chlorine, Kamiokande, SAGE, GALLEX, SuperKamiokande) are consistent with a deviation from non-oscillation models within < 2.5a (99% c.l.) only (see [BahOl**-III]). This means that for a decade or more theorists and experimentalists took serious a rv 2a effect in inventing their 'theories' - which is usually not accepted as evidence! The same is true for the 2.2a signal for 2v{3{3 decay reported by [Ell87a*-V]. Only by SNO the confidence level for solar v oscillations has been raised to an acceptable level. (3) Atmospheric neutrinos: After several earlier experiments, finally SuperKamiokande confirmed evidence for neutrino oscillations from atmospheric neutrino data [KajOl], [Ash04, Ash05]. They led under the assumption of degenerate neutrinos to a lower limit mv > O.04eV. (4) {3 Decay and {3{3 Decay: In contrast to Eq. 1.5 single beta decay allows to determine from the spectral shape of the emitted electrons the quantity (see, e.g. [FarO 2a]) ,

(1.7) Other than double beta decay which requires Majorana neutrinos, single beta decay can occur for both Majorana and Dirac neutrinos. Both experiments thus yield complementary information. Figure 1.29-right gives some feeling for the relative potential of both types of experiments (see [FarOl]). The best present limits come from the Mainz and Troitsk experiments [Kra04 **III, LobOl **-III] and are mVe

~ 30

NEMO 3 (f'hose I)

~ -25

I •

20

~ §

15

'0

+

D,"

~

lOOMo

•

Radon simulation

.

2p,2vSimulation

NEMO J (Pho~ IJ

,. 12

!!

~ '0

+

5

Z

82S e

•

Aadon simulatlon

.

2ll2Ysimulallon

10

>

:J

14

ci -

~

12

+

NEMO J (Phase I )

•

D",Cu+Te Radon slmule\lon

~ 10

8

'0

6

z

~

~ Z

Data

8

5

10

2.7

2.6

2.9

(a)

3.1

3.2

E,. (MeV)

o

3

2.4

(b)

(c)

3.1 3.2 E2e (MeV)

Fig. 1.43 Spectra of the sum energy of the two electrons in the Ov(3(3 energy window after 389 effective days of data collection by NEMOIII from February 2003 until September 2004 (Phase I): (a) with 6.914 kg of 100Mo; (b) with 0.932 kg of 8 2 Se; (c) with Copper and Tellurium foils. The shaded histograms are the expected backgrounds computed by Monte-Carlo simulations: dark (blue) is the 2v(3(3 contribution and light (green) is the Radon contribution. The solid (red) line corresponds to the expected Ov(3(3 signal if T l / 2 (Ov(3(3) = 5 x 10 22 y (courtesy X. Sarazin).

< 0.1-0.3 e V [Bar97** -V], [NEM2000] i.e. of the order of the present value obtained from the HEIDELBERG-MOSCOW experiment (see below). Figure 1.43 shows the results obtained by NEMO-III for 01/(3(3 decay after 389 days of data collection in the period February 2003-September 2004 [Arn05*-V]. The limits given for 01/(3(3 decay are 4.6 x 1Q23 y for looMo and 1.0 x 1023 y for 82S e (90% c.l.). The background was about 1.5 counts/kg y in the ±200 keY energy window around Q{3{3, and was dominated by radon contamination (see also the HEIDELBERG-MOSCOW 76Ge

Seventy Years of Double Beta Decay

54

experiment (Section 1.5.5) and GENIUS-TF (Section 1.6.2.1)). The limits after about one more year of data taking are given as > 5.8· 1023 Y for 100M0 and > 2.1·1023 y for 8 2 S e [Vas*06-VI]. Since these values are less by a factor 20 than the half-life required to reach the sensitivity of the HEIDELBERG-MOSCOW experiment for the effective neutrino mass, this means that NEMO-III will reach the sensitivity of the Heidelberg-Moscow experiment (see below) only after at least rv 400 years of continuous operation and then only on a 20" level! In other words it unfortunately can never check the present result of the HEIDELBERG-MOSCOW experiment [Kla06c* -V]. When comparing the signal-to-background ratios in this experiment with other experiments, it should be noted, that the 2//(3(3 half-lives investigated by NEMO II, III are around 10 19 years, i.e. more than a factor of 100 less than for example the 2//(3(3 half-1ife of 76Ge. From the NEMO-III measurement of the 2//(3(3 half-life of 82Se (see Fig. 3 in [Arn05*-V]) it can be extrapolated, that the background would hardly be sufficiently low to measure the half-life of the 2//(3(3 decay of 76Ge of rv 1.7 X 10 21 y [Kla03c* -V]. A magnetic tracking detector using a drift chamber, DCBAI and II, with the goal to reach a neutrino mass limit of 0.1-0.5 e V in the decay of 150 N d has been proposed at KEK, Japan by [Ish96].

1.5.3

Scintillation Detectors -

Crystals and Liquids

Scintillation counters have an energy resolution lower than semiconductors and in addition, the photomultipliers are often a source of background. In Brookhaven as the first active source experiment the decay of 48Ca was studied exploiting the fact that calcium in the form of CaF 2 may be used as a scintillation crystal [Mat66*-I]. 48Ca benefits from a very large Q-value of 4.271 Me V, however this isotope is very rare (0.187% natural abundance). The original experiment of 1966 had a source strength of 11.4 9 48Ca. New measurements with 37.4 kg CaF 2 scintillation crystals (43 9 of 48Ca), were carried out in a 512 m deep colliery shaft near Beijing (China) [You91], and an experiment (ELEGANT VI) using 31 9 of 48Ca had been planned at Osaka (Japan) [Kum96], [Eji97] [Eji98] in the Oto tunnel. ELEGANTS VI uses CaF 2 scintillators surrounded by CsI scintillators (see Fig. 1.40). A 1.1 kg CaF 2 detector partially surrounded by low-background NaI(TI) detectors has been used by the DAMA group to investigate all(3(3 decay of 48Ca to excited daughter states [Ber02b]. In an analogous way, 116 Cd can be built into scintillator material (CdW 0 4 ). This line of research has been followed by Yu. Zdesenko and coworkers in Kiev [Zde91], [Dan89, 95], [Bur96*-V], [Dan96**-V]. This experiment operated four enriched CdW0 4 crystals (total mass 339 g) surrounded by an active shield of 14 natural CdW0 4 crystals (see Figs. 1.44-1.45) in the Solotvino salt mine [Dan98, 99, 99a], [Dan2000**-V, Dan03]. The half-life limit obtained for all(3(3 decay of 116Cd is 1.7 X 10 23 years (90% c.l.), and the deduced neutrino mass limit is (mv! < 1.7 eV.

T he Experimental Race: From the Late E ighties to the Discovery of

Ov/3/3 Decay

Fig. 1.44 Left: Yurij Zdesenko and the a uthor , at the Solar Neu trino Conference, Heidelberg, Germany, 1997. Center: The four enriched CdW04 crystals surrounded by 14 natural CdW04 crystals used in the Solotvino salt mine in the Ukraine (foto Yu. Zdesenko). Right: Yurij Zdesenko in Heidelberg, with the author and h is son, 1993 (fotos a uthor) .

Fig. 1.45 Left : Solotvino salt mine in the Ukraine (foto Yu. Zdesenko). Center : Igor Kondurov from Gatchina (right) and the author near t he SISTEMA I/ll installation at the Baksan Underground laboratory, Kaukasus, USSR, in 1987, A. Smoln ikov (middle) . Ludvik Popeko (first from the right) and the author with colleagues at Baksan Valley (Kaukasus) in 1987 (fotos author).

The nuclei 100 M 0 and 150 N d have been investigated also by using plastic scintillators sandwiching the isotopically enriched samples (SISTEMA I/Il) by the Baksan group [Kli86], [Kli89], [Vas93] (see F ig. 1.45). CeF3 crystal scintillators have been applied in a pioneering way for investigation of two neutrino double electron capture in 136Ca and 138Ce [Be103]. The at present most sensitive limit for Ov(3(3 decay of 136 X e comes from the liquid Xenon scintillator set-up (see Fig. 1.46, [Be190] and [Ber02**-V]) of the DAMA experiment performed in the Gran- Sasso underground laboratory [Ber02a*V, Ber02c], which obtained the half- life limit T 1/ 2(0 -----> 0+) > 1.2 x 10 24y (90% c.l.) , which is, together with CUORICINO and IGEX (see below) , the most sensitive after the HEIDELBERG- MOSCOW experiment. The same setup has been used also for a search for various nucleon and di-nucleon decay modes [BerOO]. 1.5.4

Semiconductor Detectors

Semiconductor detectors because of t heir outstanding energy resolution are suited excellently for the study of double beta decay, and in particular for neutrino less

55

56

Seventy Years of Double Beta Decay

double beta decay which would be manifest by a sharp line in the total energy spectrum. Semiconductor devices may be used to study external probes, or as active detectors where the detector material at t he same time is the double beta emitter.

Fig. 1.46 Left: Rita Bernabei during her t alk at BEYOND 2003 at Castle Ringberg, Tegernsee, Germany, June 2003. Center: The view of the inner vessel of the DAMA LXe experimental set-up running at t he Gran- Sasso since several years (see [Ber02a*-V, Ber02c]). Right: Piero Belli at the Dark Matter 2002 conference in Cape Town, South Africa, June 2002 (fotos left and right a uthor).

An example of the first type is a setup consisting of a sandwich detector from silicon counters separated by molybdenum sheets [Oka88], [Als89], [Als93*-V]. This setup yielded a half-life limit of > 4.4 x 10 22 years for Ov{3{3 decay of loo Mo. A particularly favourable case of the second type is represented by the double beta candidate 76Ge. This germanium isotope occurs with an abundance of 7.8% in natural germanium, from which large high-resolution detectors can be manufact ured. Thus, germanium can be used simultaneously as source and detector allowing for large source strength without spoiling t he high energy resolution, which is about 3 ke V in the region of the decay energy of 2. 04 Me V. This method was first used by the Milan group in t he Mont Blanc tunnel [Fi067* -I], [Bel83], [Bel83a] . The most sensitive experiment using detectors from natural Germanium was that of D. Caldwell (see F ig. 1.47) et al. [Cal86], [Cal86a*-V], [Cal87*-V], [Cal91**-V] in California located in the Oroville dam (600 m. w. e. underground) leading to a halflife limit of 1 .2 x 10 24 years , corresponding to a neutrino mass limit (mv ) < 1- 3 eV. This was the most sensitive first generation experiment t hrough many years , until in 1990 the Russian- Armenian, and in 1992/93 the HEIDELBERGMOSCOW experiment took over (see below). An in principle very powerful detector setup suggesting use of enriched Ge detectors in a shielding of silicon was discussed , and the shielding was constructed by Popeko et al. (see Fig. 1.48) [Pop86*-V], [Pop89], but finally was used only with a natural Ge detector supplied by the Heidelberg group. This shielding was one

The Experimental Race: From the Late Eighties to the Discovery of Ovf3f3 Decay

Fig. 1.47 Upper part : Center: Set-up of the USCB- LBL Ge experiment of David Caldwell. Right : David Caldwell, at the BEYOND '97 Conference, Castle Ringberg, Germany, June 1997. Left: David Caldwell, the author and A. Smirnov, at Beyond'97. Lower part: left: Igor Kirpichnikov, with the author, at the NEUTRINO '92 Conference, Granada, Spain, 1992. R ight: Masato Morita, Irina Krivosheina, Frank Avignone and Yurij Lyutostanskii (from left to right) on the Volga river during LEWI Conference, Dubna, Russia, September 1990 (fotos author) .

of the early, however not realized, options for the later HEIDELBERG-MOSCOW experiment. A Russian/Armenian group [Kir87] under 1. Kirpichnikov used two 0.5 kg enriched Ge detectors (enriched in 76Ge to 85%) in a salt mine in Yerevan at a depth of 645 m. They observed for the first time two-neutrino decay of 76Ge [Vas90*-V], confirmed by a Russian- American collaboration [Avi91] and later by the HEIDELBERG- MOSCOW collaboration [Ba194], [HM97*-V] , [Kla03c*-V]. The Heidelberg- Moscow experiment [Leg86*-A], [Kla87**-A], [Mem88*-A], [Add98*-A], [Kla90**-V], [Kla91b], [Kla92*-V], [Kla93**-V], [Kla94], [Ba195], [Kla96a**-V], [Kla96b**-IV], [HM97*-V], [HM97a**-V], [Kla98], [Kla98a], [Kla98g], [HM99], [Kla99], [Kla99b**-VI], [Kla9ge**-IV], [Kla2000c**-A], [Kla2000e*-IV]' [KlaOlc** -V] using for the first time large amounts of enriched double beta emitter material, was the first 'second generation' experiment , starting a new era of double beta experiments. It found already in its first years of operation large public attention, e.g. [Rad86*-A], [Rad87*-A], [Kla87a*-A], [Bun88*-A], [Kla89*-A], [Bun90*A], [NEU90*-A], [Kla91 *-A], [Kla91c*-A], [CER91 *-A], [MPG94*-A], [Sci97*-A], [Kla98b**-IV], [CER2000*-A]. After some exploration of the most suitable location for the experiment, it finally started operation with the first detector in 1990 in the Gran- Sasso underground laboratory (3500 meter water equivalent) (see Figs. 1.49- 1.50) where since 1995 a setup of five detectors with in total 11.5 kg

57

58

Seventy Years of Double Beta Decay

Fig. 1.48 Upper part : Left: Ludvik Popeko (left) with the a uthor and Igor Kondurov at LNPI, Gatchina , Russia, February 1987. Right: Ludvik Popeko with part of his silicon shielding construction at Leningrad Institute of Nuclear Physics, in February 1987 (fotos author). Low part : Left: The author with Ludvik Popeko. Right: from the left to the right: The author , Igor Kondurov and L. Popeko near E lbrus, Baksan, Kaukasus, investigating the optimal site for the later HEIDELBERG- MOSCOW experiment, in 1987 (fotos author).

were operated [HM97*-V]. Dat a taking was going on until November 2003. (Until September 2004 calibration measurements for the pulse shape analysis were still performed.) In total 71. 7 kg y of statistics have been accumulated. The amount of enriched material makes the experiment at least as sensitive as an experiment using 1.2 tonnes of natural germanium. The experiment has the lowest background of all double beta experiments of this type, O.l1 counts/ kgykeV [Kla04d*-V] . This has been further reduced by new methods of pulse shape discrimination distinguishing between single site (double beta) and multiple site events [HM97a**-V], [HelOO*V], [Maj99*-V], [Kla04c*-V, Kla04d*-V], [Kla06a*-VI, Kla06b*-VI, Kla06c*-V] to about 5x 10- 3 counts/kg year keV. The half-life limit for the neutrinoless decay mode obtained until the year 2001 was 1.3 x 1025 years (90% c.l.) from the full spectrum , yielding the most restrictive limit on the effective Majorana neutrino mass of rv 0.4 2 (0.33) e V (90% c. l. and 68% c. l. , respectively) [HM97a**-V], [KlaOlc**V], (see also [Kla9ge**-IV, Kla2000c* *-A, KlaOlb**-III]). Only later a signal for 01/{3{3 decay has been observed. We refer for this to the next Chapter 1.5.5. Just in the right time a high-precision measurement of the Q(3(3 value has been published by the group of Ingmar Bergstrom (see Fig. 1.37) [Dou01]. Earlier measurements of Q(3(3 were performed by [All85a], [Hyk91]' [Aud95].

The Experimental Race: From the Late Eighties to the Discovery of

1.5.5

Ov/3/3

Decay

The First Evidence for Neutrinoless Double Beta Decay

The first evidence for neutrino less double beta decay came from the HEIDELBERGMOSCOW experiment (see Figs. 1.49-1.50) whose history until the year 2000 was described in the previous section.

Fig. 1.49 Left: Gran-Sasso Underground Laboratory, schematical view, Italy. Right: The HEIDELBERG-MOSCOW /3/3-experiment setup Nr.1 in Gran- Sasso with four 76Ge crystals and antimuon shield.

Fig. 1.50 The Heidelberg- Moscow experiment: Upper part: Left photo: The first high-purity enriched 76Ge crystal worldwide. Center: A. Muller with the author in the Gran-Sasso laboratory in the time of discussion the first results , November 1990. Right: One of the enriched 76Ge detectors in its shielding of electrolytic copper. Lower part: Left: In the time of installation of the Ge detectors. Right: The four installed detectors in setup 1. Center: The second enriched detector of the Heidelberg-Moscow experiment, the biggest detector worldwide at time of its production (1992). (Fotos author.)

59

60

Seventy Years of Double Beta Decay

Until the year 2000 the statistics had increased such, that instead of just assuming a const ant background around Qf3f3 (as still done in [Kla01c**-V]), one now could try to take into account the lines in the background around Qf3f3 in the fit. This would automatically decrease the background. This was performed after in 2001 the Kurchatov part of the collaboration left the collaboration and proposed to make analysis and papers independently. (All analysis and all publications until then had been performed by the Heidelberg group .) The result was occurrence of a signal for Ov(3(3 at Qf3f3 [KlaOlj, KlaOla *- V, Kla02b**- Vj, on a 3.1CJ level. Some critical comments [Com02], [Zde02a], [Fer02], [Kur05] to the early result of 2001 have been rejected in all points in a series of papers [Kla02c*-A, Kla02k, Kla02d**-V, Kla02e, Kla03b*-V, Kla03c*-V, Kla03d , Kla04b*-V, Kla04d*-V] . The result presented by [KlaOla*-V, Kla02b**-V] was confirmed by an independent analysis of the same dat a by a famous spectroscopic group from Dubna [Gro06*-V] (see also Fig. 1.51).

Fig. 1.51 Left: Kirill J akovlevich Gromov , head of La boratory of Nuclear Problems, JINR, Dubna, Russia. Right: Vadim Bednyakov, LNPI, Dubna, Russia.

The result obtained from the full data t aken from 1990-2003 yielded (as result of the impr-oved statistics after fu rther three years of measurement and a refined analysis) a signal at a 4.2CJ c.l. , and a half-life of Ov(3(3 decay of 76Ge of 1.19 x 25 10 y, corresponding to a 3CJ range of the effective neutrino mass of 0. 24 - 0.58 e V ([Kla04c*-V, Kla04d*-V]) (Figs. 1.52- 1.53). This result was discussed naturally in many places, e.g., in [Kla04**-IV, MPG03], [Kla04f*-A, Kla04g, Kla04h], [Kla05*V, Kla05a**-V, Kla05b**-V, Kla06e*-V, Kla07a, Kla07c*-VI] . In the years 2004-2006 large efforts went into the improvement of pulse shape analysis. In addit ion to the pulse shape analysis method using a neuronal net developed earlier [Maj99*-V], and the method given in [HelOO*-V], another method was developed. For this purpose a first investigation of the dependence of the tracks of (3(3 events in a Ge detector on particle physics parameters (neutrino mass, righth anded weak current parameters and nuclear matrix elements) has been performed by Monte Carlo calculations [Kla06**-V, Kla06a*-V] (Fig. 1.54).

61

The Experimental Race: From the Late Eighties to the Discovery of Ov{3{3 Decay

>~ ,'" :lc

3000 mAc

>Q) ....

22IAC214 Pb

2500 214Pb

~ 2000 > ~

214Bi

~

...1/1C Q)

> Q)

214Bi

214Bi

500 0 100

150

200

250

7000

>

.l! ,

6000-

•

5000-

E

137

350

400

CS

o ~~~~~~~~~~ l~l~l~l~l~l~l~l~~OO

~ 4000:300o ~c

"1"l

"'n

mAc 125 I25 2000 Sb Sb

1000

300

I I I

0

214

81

214Pb

' 'BI ~A

~I~ "'r\ \

> 160 ~ 140

....

...1/1C

214Bi

mrl{SE)

120

~Q) 100

214Bi

214Bi

80

'mAc 21'sF'Ac 2MmPa

2GlBi

'7:n+Xray &II eo

60 40 0 2000 2050 2100 2150 2200 2250 2300 2350 2400

0 800

>

4000 3500

,'"

3000

~

850

950

900

1000

1050

1100

1150

1200

~K

300

20'11

~

250-

...1/1C

200 -

Q)

> Q) 150-

:lc 2500

100-

~

> 2000 ~ 1500 1000 500 0 1200

>Q) ....

212Si

I -A .

1250

·Co tU

Bi

I 1300

~ 1350

214

81

7 1;BI 1

r

"'j' l "'Ai

A

1400

1450

1500

1550

1600

energy [keV]

50

214Bi

SOCO

I

~J

,~

0 2400 2450 2500 2550 2600 2650 2700 2750 2800

energy [keVj

Fig. 1.52 Low- and high-energy parts of the full spectrum, measured in the Heidelberg-Moscow experiment in the period November 1995- May 2003. The spectrum is the sum of about 10000 individual spectra (from [Kla04c*-V, Kla04d*-V]). The lines are 'Y-lines from the perfectly understood background (see [Kla03c*-V]). The excellent energy resolution in the range of Q/3/3 may be compared to that in Fig. 1.43.

62

Seventy Years of Double Beta Decay

20

18

25

16 20

14

12 15

10

, ,,

10

, '

looo

2010

, ,,

" , 2020

2030

2040

2050

2~O

2060

Energy, keV

5

II

INNI

4

.

>

~ ~

0

3

2

CJ

energy, keY 15'---~--'---'---'---~--~--e====c==~

- - SSE

~10 E

8"

5

150 '---'---~--~--~--~----r---.-~~==~

- - Full

2000

2050

Energy, keV

2100

2150

2200

2250

2030 Energy, keV

2040

2050

2060

Fig. 1. 53 Heidelberg- Moscow experiment -- FINAL spectra. First lin e: The total sum spectrum of all five detectors (after subtraction of the first 200 days of measurement of each detector) (in total 1O.96kg enriched in 76Ce), in the range 2000-2060 keY and its fit, for the period: August 1990 to May 2003 (71.7kgy) (left). The Bi lines at 2010.7,2016.7,2021.8 and 2052.9 keY are seen, and in addition a signal at ~ 2039 keY. Right: The total sum spectrum of all five detectors (in total 10.96 kg enriched in 7GCe), for the period November 1995 to May 2003 (56.66kgy) in the range 20002060 keY and its fit (see [Kla04c*-V]). Second line: The pulse shape selected spectrum (selected by neuronal netNN) with detectors 2,3,4,5 from 1995 to 2003 in the energy interval 20002100 keY see [Kla04d*-V, Kla04c*-VJ, [Kla05b**-V, K la06c*-V]. The signal at Q(3(3 has a confidence level of 6.40" (7.05±1.l1 events) . This figure is a zoom of the figure below. Last two figures: The pulse shape selected spectrum (subclass NN of events selected) (upper part) and full spectrum (bottom part) measured with detectors 2,3,4,5 from 1995- 2003 in the energy range 1800-2250 keY (from [Kla04d*V, Kla04c* -V]) , see text. The background 'Y-lines are drastically reduced in the pUlse-shape selected spectrum, while the line at Q(3(3 is standing out.

The Experimental Race: From the Late Eighties to the Discovery of Ovf3f3 Decay

REAL POTENTIAL

..." i ...

WEIGHTING POTENTIAL

10

·2 .5E3

·2 .0E3 ·1 .5E3

63

ss, / .mln_0 52 E- 1592k,V

ANO.

-i .0E3 -S.OE2

5

5

IIma.o::h.rv>et

N

1.o-_ m•••uNdp.sIl.

u· u - MSE

1\1\ j/ \

07~ l'-m/nII1'U

1

!o-1eeZaV

nil. AHQ& OJ.

M· D>. D>.

" \\.,N.o'V ••

0.0 0.5 1

R, em

R, em

2. ••

o~

.

~ 0,1 0,0

0,'

01

0,' 0,3

Iml, Dlllnll,1

..

q0~

11, 1t

,.,.

\d

c.

0,3

1,1

~~

o~

0,1

~~

0,0

.0,1

.0,1

·0,2

.0,8

.o,e

.0"

.0.2

it

b

,

o keY

y (511 keY)

•

C ,)

0,0

x,mm

I,BOO _,180

13,00 11.28

11.'0 22,78 21,00

9

9 .o~

0.rA.'~ ....

y(2814keV)

O~

0,'

0,8

.0,3 1i ,1i

20,0

20,1

20,2

20,3

20,'

20,'

x,mm

Fig. 1.54 Upper part: Left: Electric field as calculated by the Poisson Superfish code for the enriched detector ANG5 of the HEIDELBERG-MOSCOW experiment in Gran- Sasso (real - (left) and weighting (right) fields). The chosen ionized impurity density is 1.875.10 9 cm- 3 (included in the left, but set to zero in the right part). Operational volt age is 2500V. Right : Result of comparing experimental pulse shapes (black) in the range around Q {3{3 with the library shapes (red) for an SSE (single site event) event (upper part) , and for an MSE (multiple site event) event (down part). Down part: Typical calculated track in the Ge detector for a Ovf3f3 event without photon emission (bremsstrahlung) (left). Calculated photon event (the energy of the initial photon is 2614 keY) for the double escape case (leading to a line in the detector at 1592 keY) (right) (from [Kla06**-V, Kla06a*-V]).

T he dependence of the shape of the electrical pulses produced in a Ge detector on these parameters then has been investigated by calculating the electrical field distribution in the detectors (see Figs. 1.54- 1.55) [Kla06b*-V]. The understanding achieved from both methods allowed to efficiently reduce the background of gamma background events from the measured data leading to extraction of a Ov{3{3 signal on a> 60" c.l. [Kla06c*-V, Kla06e*-V , Kla07a, Kla08b*-V] (see F ig. 1.53 middle and bottom). The background around Q{3{3 is now reduced to ~5·1 0 -3 counts/kgy keY , which is lower t han the goals of many future experiments under preparation (see Chapter 6).

Seventy Years of Double Beta Decay

64

Even some fine details of the composition of ,-lines are 'seen' by the pulse shape selection (see [Kla06c* -V]) . 1 Figure 1.55, shows t he time structure of the events identified in the line (see Fig. 1.53) near Q {3{3 by the neuronal net method, and also shows that these events are consistently also identified by the second (electric field) method as single site events.

U

I'\'

AHG3. 2037.6 kcV

f, "" ~

It

1..

f \ /

u

oAAIOl, 203U"'1I

t " t

\

\ ..-.

n

I '~

..

\

i "

it .." 014

u

t "

,. _* ...... ' * __ ..

" 1'

1\

1rNIJIJ.2037.2 YV

i "

I

1. . . .

:

\

Fig. 1.55 Time structure of those events (black) measured in the HEIDELBERG-MOSCOW experiment in the energy region 2036-2042 keY by the four enriched 76Ge detectors (ANG2, ANG3, ANG4, ANG5) during the period 1995-2003, selected by the neuronal net near Q(3(3 (NN selection, see [Kla06c*-V]) as single-site events (see F ig. 1.53, and Table 1 and Fig. 7 in [Kla06c*-V]). The red lines show fits of the events by the zero range pulse shape approximation library, described in [Kla06b*-V] . The result is consistent with t he neuronal net analysis (from [Kla06c*-V]).

From the intensity of the line in the pulse shape selected spectrum a half-life for Ovf3f3 decay of T~/2 = (2.23~g : ji) . 10 25 y is deduced. This corresponds to (m) = 0.32 ± 0.03(stat) ±O.ll(syst) eV. Since the corresponding matrix element for 2vf3f3 decay underestimates this decay by rv 30% (see [Kla04c*-V]) the lower limit for lE.g. the 2505.7 keY line (see [Kla04d*-V, Kla04c*-V]) originating from summation of the subsequent 1173.2 and 1332.5 keY "I-lines from 60Co - naturally expected be a multiple site event - is erased to 100% by both methods [K la06c*-V]. The neuronal net also reduced the 2016 .7 keY line from 214Bi which is as EO transition also produced by two subsequent transitions (see [Kla04d*-V, Kla04c*-V]) , stronger than normal full energy (FE) "I-peaks.

The Experimental Race: From the Late Eighties to the Discovery of Ovf3(3 Decay

65

the effective neutrino mass is rv 0.22eV [Kla06c*-V, Kla08b*-V] (tacidlyassuming in this discussion vanishing right-handed weak currents or other contributions to the 01l/3/3 amplitude). In such a way the dream behind all more than 50 years of experimental efforts, which was of course less to see a standard-model allowed second order effect of the weak interaction - the two-neutrino-accompanied decay mode {which has been observed meanwhile for about 10 nuclei} - but to observe neutrinoless double beta decay, and with this a clear indication of beyond standard model physics - was fulfilled [Kla06c*- V, Kla06e*- VI Another enriched 76Ge experiment, IGEX, had been set up partially at the Baksan underground laboratory in Russia and at the Canfranc laboratory in Spain (2450 m water equivalent) using 9 kg of 76Ge, and was stopped in 1999. 2!iO 200

8

6 4

2

150

to

50

0

180

810

1440

EMrgy(lolV)

2070

2700

Fig. 1.56 Left: Histogram from 117 effectiv mole years of IGEX data (in 2 keY bins). The darkened spectrum results from the application of PSD to ~ 45% of the total data set. The Gaussian curve represents the 90% C.L. constraint of ::; 3.1 Ov(3(3-decay events and has a FWHM of ~ 4 keY, [IGEOOj. Right: "A sample of the early data prior to gain shifting from all three 2 kg detectors. The spectrum is compressed to 10 ke V per channel" (to be compared to 1 keV per channel in the HEIDELBERG-MOSCOW experiment Fig. 1.52), from [IGEOOj.

IGEX claimed to have collected in 8-9 years of operation altogether 117 mol yr of data. This corresponds to 8.7 kg y (about 10% of the statistics accumulated by the HEIDELBERG-MOSCOW experiment), which means that it took data only in a short part of its time of operation (or only a small part of the data was selected for analysis). It has been shown [Kla04a*-V] that the deduced half-life limit is overestimated. The half-life limit deduced from the full spectrum using a correct procedure is [Kla04a*-V] 1.1 x 10 25 years (as also stated in their earlier paper [IGEOOa]). The Feldman-Cousins [FeI98] sensitivity of the experiment is T~/2 < 0.52 X 10 25 y. These values correspond more naturally to those deduced from an experiment heaving almost one order of magnitude higher statistics [Kla04c*-V] (see above). The numbers given by [IGE99a**-V, IGE02] led to some confusion, since they were rather uncritically cited in several theoretical papers. IGEX shows [IGEOO] its full measured energy spectrum only compressed to 10 keV bins (see Fig. 1.56) (compare with HEIDELBERG-MOSCOW- 0.32 keV and 1 keV bins).

Seventy Years of Double Beta Decay

66

The claimed limit of 3.1 events for 8.7kgy of measurement [IGE02] is consistent with the 19.6 events observed in the HEIDELBERG- MOSCOW experiment in 51.39kgy measurement, or the 28.8 events found in the HEIDELBERG-MOSCOW experiment in 71. 7 kg y. 1.5.6

Cryogenic Detectors

Another impressive approach is to use cryogenic detectors. The so-called bolomet ers were explored for the use in the search for double beta decay by E. Fiorini and T. O . Ninikoski [Fi084], [Fi091a], [Fi091b], [Giu91] . They involve the use of a pure diamagnetic or dielectric crystal, whose thermal capacity at low temperatures is so small, that even the energy released in a single double bet a decay event causes a measurable increase in the temperature. From the first four 334 g heavy Te02 crystals operat ed as bolometers at 10 mK in the Gran- Sasso laboratory in the search for double beta decay of 130Te [Ale94], a half-life limit for the neutrino less mode of 1. 8 . 10 22 years was deduced. This setup has yielded, after a measuring time of O.35kgy, a half-life limit of 7.7·10 22 y for the neutrinoless mode [Giu99a*-VI], corresponding to an effective neutrino mass limit of 2.5- 5.2eV.

~ 100 ~ 5

o

208TI

/

80 60

40

214

81

\

20

2420

2500

2580

2660 Energy (keV)

Fig. 1.57 Left: Schematic view of the CUORICINO experiment [from CU005J. Right: Andrea Giuliani, at the LEPTON-BARYON Conference at Trento, Italy, April 1998. Right down: Anticoincidence spectrum of the sum of the two electron energies in the region of neutrinoless Ov{3{3 decay for the CUORICINO experiment (see [CU005], [Bro05*-V]).

The Future of Double Beta Decay

The Milano group then operated a 20 detector array whose single modules have a mass of 340g of Te02 [Ale95], [Ale98], [Ale99], [Ale99a]. Since 2003 they operated an array of 44 smaller and 18 larger crystals of Te02, with a total detector mass of 40.7kg (5.2.10 25 nuclei of 130Te), i.e. around 13kg of active mass of 130Te (CUORICINO) (see Fig. 1.57). The energy resolution, around Q{3{3 is 8-12 keV, i.e. worse than for Ge detectors around Q{3{3, but much better than in many other experiments. The background around Q{3{3 was claimed to be 0.18±0.01counts/kgykeV). Data from 10.85 kg y of 130Te yield a limit for Ov(3(3 decay of 1.8.10 24 years [Bro05*V], [CU005]. According to the [Sta90a*-II] matrix elements this corresponds to a limit of (mv) < 0.53eV (90% c.l.). This means that CUORICINO could reach the sensitivity to probe the HEIDELBERG-MOSCOW result in about rv 10 years of permanent measurement. If the matrix element calculated for 130Te would be underestimated by only a factor of 2, it would need already 160 years, to reach the sensitivity of the HEIDELBERG-MOSCOW experiment - but only on a 90% (1.50") c.l. If on the other hand the matrix element would be overestimated, then CUORICINO would have a real chance, to check the Heidelberg-Moscow result in relatively short time. In any case CUORICINO can never rule out the result of the HEIDELBERG-MOSCOW experiment. For the future an extension of this array is planned, in the CUORE experiment [Fio98], [Giu99a*-VI] (see Section 1.6.2.5). An interesting candidate was for some time a N dF 3 bolometer exploiting the relatively large phase space of 150 Nd (see [Moe95]), but the project up to now never surpassed a 100 g crystal (see, however Chapter 1.6.2.11). 1.6

1.6.1

The Future of Double Beta Decay

General

After the first evidence of observation of Ov(3(3 decay, the goal of future experiments should be (1) to confirm this result independently with preferably another isotope, (2) to determine the contribution of neutrino mass, right-handed currents A, TJ (or other more exotic contributions) to the double beta amplitude (see Eq. 1.4). The situation nowadays is unfortunately such, that (1) it seems that most of the experiments at present running or under preparation will hardly be sensitive enough to test the result of the HEIDELBERG-MOSCOW experiment presented in the previous Chapter on a reasonable time scale (see Fig. B.11). The experiments coming closest to this goal are probably CUORE, EXO and SNO+ (see below). (2) In any case none of the presently planned and prepared future experiments - which are all (3- (3- experiments - will be able to determine, together with the information from the HEIDELBERG-MOSCOW experiment, the different contributions to the Ov(3(3 amplitude [Kla06e*-V, Kla06a*-V, Kla07c*-VI]. The reasons for this second statement are outlined in the Section 1.6.3, and a solution will be proposed. (3) After fixing a half-life for Ov(3(3 decay, there is no more chance to improve the limits

67

68

Seventy Years of Double Beta Decay

Fig. 1.58 Conference foto of the BEYOND '97 Conference, Castle Ringberg, Bavaria, Germany, where the GENIUS proposal has first been presented (photo author).

for other beyond SM physics (see Chapter 4), as it was hoped earlier, and was explored in details in connection with the GENIUS project (see e.g. Fig. 1.59). In general it has to be understood that the time of the small smart experiments is over. The big step in sensitivity increase achievable by replacing natural by enriched material with the effect, that with a 'few kg experiment ' the sensitivity of a 'order of ton of natural material experiment' can be reached , has been made. Example is the Heidelberg-Moscow experiment using 11 kg of enriched 76Ge, corresponding to more than 1.2 tonnes of natural Ce. While the improvement in the neutrino mass sensitivity went with the square root of the degree of enrichment of the detector material, after this step having been done, the improvement in the neutrino mass bound now goes with the fourth root of the detector mass, of the background and of the measuring time. The half-life bound reachable in experiment is T ~ a(mt/bB)1 / 2, with percentage of enrichment of the double beta emitter a in the detector, detector mass m, background B, energy resolution 5.8.10 23 14 :j: 1.5 * 15 under construction, t of enriched material, start 2015 ?

t

0.18 :j: 1.8

~O. 75t

~0.2t

~0.1

11

> 1.8. 10 24