Sensitivity to half life of 0νββ decay (left) and to effective Majorana neutrino mass (right) of an experiment as a function of exposure for different background indices. The sensitivity to 0v  decay of 76 Ge strongly depends on the background index of the experiment. Neutrinoless double beta decay Neutrinoless double beta decay (0νββ) is an extremely rare second order weak process which is predicted to occur if the neutrino is a Majorana particle with non-zero mass. In this process the lepton number is changed by two units. The half-life T 1/2 of 0νββ decay is a function of the neutrino masses, their mixing angles and CP-phases. Right: Schematic diagram of neutrinoless double beta decay. The nucleus changes its charge by two units. If the neutrino is a Majorana particle, the neutrino emitted by one W boson can be absorbed by the other W boson. Towards the ton scale experiment: The GErmanium Detector Array (GERDA) experiment The GERmanium Detector Array, GERDA, will search for neutrinoless double beta decay in the isotope 76 Ge [2]. Its main design feature is to submerge and operate high purity germanium detectors, enriched in 76 Ge to 86%, directly in liquid argon. The latter serves as coolant and shield against external radiation simultaneously. The cryostat is placed inside a buffer of ultra-pure water which serves as additional shield and as Čerenkov cosmic muon veto detector. A background index of counts/(kg keV yr) is envisioned for phase II. Simulations have shown that this goal is realistic, if the detectors can distinguish the different event topologies of signal and backgrounds. Pulse shape simulation and analysis Measurement of 0νββ decay would imply that the neutrino is a Majorana particle. The half-life of 0νββ decay provides the effective Majorana neutrino mass, m  = where m j is the neutrino mass eigen-value and U ej the neutrino mixing matrix elements. With enough sensitivity on m , the neutrino mass hierarchy as well as the absolute neutrino mass can be probed. Left: effective Majorana neutrino mass as a function of the lightest neutrino mass. The green band shows the case for the inverted hierarchy, the red band shows the normal hierarchy. The width of the bands is determined by the unknown Majorana phases and uncertainties in measured neutrino parameters. Search for 0v  decay with 76 Ge detectors Double beta decay can be observed for 76 Ge. Germanium enriched in the isotope 76 Ge can be used to produce High Purity Germanium Detectors (HPGeDs) with excellent energy resolution. Thus the signal detection efficiency is high and the energy region of interest is very small. Germanium can also be purified to very high levels, allowing the production of large mass detectors. Thus the intrinsic background of the detector is very small. These advantages are crucial for 0v  decay searches. Right: Expected signal for 2v  (gray area) and 0v  (red line). For the 2v  the 2 neutrinos in the final state carry away energy without being detected. For the 0v  decay the two electrons and the nucleus carry the full energy of the decay and can thus be detected. The Q-value for 76 Ge is 2040 keV. Lightest neutrino mass (eV)‏ m  (eV)‏ 18-fold segmented n-type and p-type true-coaxial HPGeDs have been developed together with the suppliers. The performance and background identification capabilities of the n-type detectors have been tested in vacuum and cryoliquid environments [4,5,6]. n-type technology has been chosen as baseline for further development as contacting of the segments with a minimum amount of material proved feasible. The segmentation is achieved by implantation rather than by mechanical grooving. Left: Two 18-fold segmented n-type prototype detectors before immersion into LN. Middle: 18-fold segmented p-type detector mounted in the test stand. A solid contact needs to be made to the outer surface. Thus additional holder material needs to be used. Right: Test stand for immersion of detectors in LN or LAr. Water / Myon-Veto (Č)‏ Clean room / lock Cryo-tank To distinguish between the allowed mass hierarchies an exposure of a few 1000 kg yr and a background index of counts/(kg*keV*yt) need to be obtained [1]. To reach this goal a two fold strategy is chosen: Avoid background by minimizing and selecting materials. Identify background by building detectors capable of distinguishing between different event topologies. Background reduction by minimizing and selecting materials To minimize radio impurities in the vicinity of the detector, material that could contain contaminations is being carefully selected and minimized. A low mass copper detector holder contains 31g of copper and 7g of PTFE. The Kapton substrate for a low mass contact cable is being replaced by radiopure PEN (left). A 60 Co spectrum (right) taken from a detector with the PEN readout shows no degradation in resolution achievable. Cu holder Cu on Kapton cable Low mass (~2.5 g per cable) copper on Kapton cables have been developed and successfully tested on prototype HPGeDs [4,6]. The typical Kapton radio impurities of 238 U, 232 Th with typically 5 mBq/kg would significantly contribute to the background. As replacement a batch of radio pure PEN material has been identified with A( 238 U)<2.0 mBq/kg and A( 232 Th)<1.4 mBq/kg. Germanium Array Left: Schematic drawing of the GERDA experiment. An array of HPGeDs is placed in the center of a cryostat filled with LAr. The cryostat is surrounded by a water tank 10m in diameter. The detectors are mounted and moved into the cryostat through a lock system in a clean room. Right: Picture of the GERDA experiment at LNGS. Background tagging by segmentation and scintillation light read out segments electrons photons Schematic view of two segmented detectors with an electron energy deposit (left) and Compton scattering γ-particle. The location of energy deposits are indicated by yellow dots, the segments with energy deposits are shown in green. For the signal process (left) only one segment is hit, whereas for Compton background processes (right) photons scatter multiple times (red lines) and deposit energy in more than one segment. scintillation light Effective Majorana neutrino mass Required background index for a ton scale experiment The best limit on the half-life of the 0νββ-process of 76 Ge using HPGe detectors comes from the Heidelberg- Moscow experiment with T 1/2 > 1.9·10 25 years (90% C.L). The expected signal is a sharp peak at the Q-value of the decay. For 76 Ge this is at 2040 keV. In case of 0νββ-decay two electrons are emitted. The total energy of the decay is typically deposited within a radius of a few millimetres. Background gammas of the same energy have a mean free path of the order of cm. Thus they will deposit their energy in multiple interactions separated typically by cm. Events that have energy deposit in more than one segment are thus likely to be background events. References [1] A.Caldwell and K.Kröninger, Phys.Rev.D 74(2007) [2] GERDA collaboration, arXiv:hep-ex/ [3] I. Abt et al., NIM A 570(2007)479 [4] I. Abt et al., NIM A 577(2007)574 [5] I. Abt et al., NIM A 583(2007)332 [6] I. Abt et al., JINST 4(2009)P11008 [7] I. Abt et al., EPJ C 52(2007)19 [8] I. Abt et al., EPJ C 68(2010)609 Simulations have shown that segmentation 6-fold in Φ and 3 fold in z of a true coaxial HPGeD with 75mm diameter and 70mm height yields a background suppression factor of ~5 for the 208 Tl 2.6 MeV line and of ~150 for external 60 Co [3]. Left: Spectrum measured with a 228 Th source. The black spectrum represents all events whereas the red spectrum represent events with only one segment hit. The close-up shows two peaks. On the left is the double escape peak of the 208 Tl line, right is a photon line of the 212 Bi. Events in the double escape peak deposit energy locally whereas events in the 212 Bi peak deposit energy over a larger volume and therefore suppressed more strongly. Right: suppression factor (ratio of total number of events in a specific energy region and the number of events that have energy deposited in only one segment) for the different full absorption peaks. Monte Carlo simulation shows and agreement on a level of 10%. Single-segment spectrum Total spectrum Investigation of HPGeD surfaces Background events with multiple interactions inside the detector can be distinguished from signal like events by their pulse shapes. Due to the different drift times of electron-hole pairs from different locations their recorded pulses differ. For 18-fold segmented detectors it has been shown that background rejection can be improved by an additional factor 1.5 [7]. A technology to deposit copper traces on PEN has been developed. While operation of a detector using this cable shows comparable resolution to the proven Kapton solution, the production line needs further development avoiding significant background contaminations. For an energy deposit in one segment mirror pulses are induced on the electrodes of the neighboring segments. From their form and amplitude additional information on the event topology can be deduced. It has been shown in [4] that the Φ position of an event can be determined by the ratio of mirror pulse amplitudes of the left and the right neighbor segment. A refinement of event localization and methods for identification of different type of event topologies using mirror pulses is planned. To improve the understanding of pulse formation and thus possibly of the rejection efficiency a pulse shape simulation package has been developed and successfully tested on experimental data [8]. Libraries of simulated multi- and single-site pulses will be created and used for training of a neural network to distinguish between the two event types. Also the shape and amplitude of mirror pulses are simulated. These are important for checking capabilities of mirror pulse based background identification methods. Left: Weighting potential of a segment together with the indication of a gamma interaction and the drift path of electrons and holes. Center: Simulated pulses. Right: Simulated pulse compared to a real pulse. Investigation of mirror pulses Surface contaminations on the detector are an important background contribution for low level experiments. Charge collection efficiencies close to the HPGeD surface are poorly known. For a reliable identification of surface background events detector response at the surfaces thus need to be studied. A vacuum test stand has been built that allows for scanning of HPGeD surfaces with collimated α-, β- and low energy γ- radiation. It can be equipped with a wave-length tuneable laser. This allows to modify the penetration depth of the laser and thus to scan the dead layer of the detector. For these investigations a HPGeD with a 19 th segment along the passivation area has been procured. Detailed investigation of the surfaces of this detector will be carried out. Left: Test stand for scanning of HPGeD surfaces. The detector sits on a cooling finger that is in thermal contact with a LN bath. The thermal shield can be turned around its axis. Stepping motors allow to scan the detector with collimated α-, β- or γ-sources or a wavelength tunable laser. Right: Schematic view of the 19-segmented coaxial HPGeD. The 19 th segment consists of a 5mm thick layer on top of the crystal. Cold Finger Thermal Cu shield Stepping motor for movement around axis Preamplifiers Stepping motor Collimators Right: Display of a single segment event. Each panel shows the pulse shape as recorded in the indicated segment. The energy deposit happened in segment 14. Mirror pulses are seen in the neighboring segments. As the event occurred close to the boundary of segments 14 and 15 the mirror pulse in segment 13 is larger than in segment 15. Development of background reduction techniques for a ton scale 0νββ experiment using segmented HPGe detectors I. Abt, H. Aghaei, N. Becerici-Schmidt, A. Caldwell, F. Cossavella, S. Dinter, F. Faulstich, S. Hemmer, J. Janicsko, J.Liu, X. Liu, B. Majorovits, C. O’Shaughnessy, H. Seitz, F. Stelzer, A. Vauth, O. Volynets Max-Planck-Institut für Physik, München Fachbeirat of the MPG, Munich, October 26th - 27th 2010 Development of prototype detectors