Status of the MARE experiment in MilanE. Ferri∗, C. Arnaboldi∗, G. Ceruti∗, C. Kilbourne†, S. Kraft-Bermuth∗∗, A.

Nucciotti∗, G. Pessina∗ and D. Schaeffer∗

∗Università di Milano–Bicocca and INFN di Milano–Bicocca, Milan, Italy†Goddard Space FlightCenter, NASA, USA

∗∗Institut für Physik, Johannes-Gutenberg-Universität Mainz, Germany

Abstract. An international collaboration has grown around the project of Microcalorimeter Arrays for a Rhenium Experiment(MARE) for a direct and calorimetric measurement of the electron antineutrino mass with sub-electronvolt sensitivity.

MARE is divided into two phases. The first phase (MARE-1) consists of two independent experiments using the presentlyavailable detector technology to reach a sensitivity of mν ≤ 2 eV/c2. The goal of the second phase (MARE-2) is to achieve asub-electronvolt sensitivity on the neutrino mass.

The Milan MARE-1 experiment is based on arrays of silicon implanted microcalorimeters, produced by NASA/GSFC,with dielectric silver perrhenate absorbers, AgReO4. We present here the status of MARE-1 in Milan which is starting datataking with 2 arrays (72 detectors). In this configuration a sensitivity of about 5 eV can be achieved in two years. We describe indetails the experimental setup which is designed to host up to 8 arrays (288 detectors). With 8 arrays, two years of measurementwould improve the sensitivity to about 3 eV. This talk reports on the activity of the group for the MARE project in Milan.

Keywords: Cryogenics detector, Neutrino massPACS: 23.40.-s, 14.60.Pq, 7.20.Mc

1. INTRODUCTION

Neutrino oscillation experiments have shown that neutri-nos are not massless particles, but they are not able to de-termine the absolute neutrino mass scale. Therefore, theabsolute value of neutrino mass is still an open questionin elementary particle physics.

The experiments dedicated to effective electron-neutrino mass determination are the ones based onkinematic analyses of electrons emitted in singleβ -decay. The most stringent results come from electro-static spectrometers on tritium decay (E0 =18.6 keV).The Mainz collaboration has reached mν ≤ 2.3 eV/c2

[1]. With the Troitsk neutrino mass experiment, an upperlimit on electron-antineutrino mass of 2.5 eV/c2 has beenobtained [2]. The next generation experiment KATRINis designed to reach a sensitivity of 0.2 eV/c2 in fiveyears.

To scrutinize the current results of Mainz, Troitsk andfuture results KATRIN, an entirely different method todetermine the neutrino mass from single β -decay hasbeen investigated. This complementary approach is thecalorimetric one. With this technique, the beta source isembedded in the detector so that all the energy emittedin beta decay is measured, except for the one taken awayby the neutrino. The study of the β -decay spectrum of187Re provides a suitable method to determine the massof the anti-neutrino. The method consists in searchingfor the deformation caused by a non-zero neutrino massto the spectrum near its end point. Previously, a sensi-

tivity of mν ≤ 15 eV/c2 was achieved with the experi-ments MIBETA in Milan and MANU in Genoa [3]. Inthese experiments, the systematic uncertainties are stillsmall compared to the statistical errors. The main sourcesof systematics are the background, the pile-up, the theo-retical shape of the 187Re β spectrum, the detector re-sponse function and the Beta Environmental Fine Struc-ture (BEFS) [4], [5] . The MANU and MIBETA resultstogether with the constant advance in the performanceof low-temperature detectors open the door to a newlarge scale experiment able to explore the sub-eV neu-trino mass range.

2. THE MARE PROJECT

MARE is a new large scale experiment to measure di-rectly the neutrino mass from the end-point of the 187Rebeta decay (E0 = 2.47 keV). The final aim of this projectis to explore the sub-eV neutrino mass range. To achievethis goal, a statistic up to 1014 187Re beta decays and de-tectors with excellent energy and time resolutions (i.e.about 1 eV and 1 µs) are required. These requirementshave been studied in details both with Monte Carlo sim-ulations and with a numerical statistical analysis. Thislatter analysis, limited to the first order expansions andin absence of background, gives the following expressionfor the statistical sensitivity on mν at 90% C.L.

565

Σ90(mν) = 1.13 4

√∆EE3

0Aβ tMNdet

+0.3τE5

0tMNdet∆E

(1)

where ∆E, tM , Aβ , Ndet and τ are the energy range ofinterest near the end-point (∆E cannot be smaller thanthe detector energy resolution), the measuring time, the βdecay rate of each detector, the number of detectors andthe detector time resolution respectively. The first term inequation (1) represents the contribution to the noise fromthe statistical fluctuations of the beta spectrum and thesecond one the pile-up contribution. The time resolutionis the ability to resolve two β decays happening in a shorttime interval, two events with a separation smaller thanτ are interpreted as one single decay with energy equalto the sum of the two decay energies. In practice, τ isof the order of the detector rise time. The number oftotal events is the product of Aβ , tM and Ndet and thefraction of pile-up events ( fpp) is given by the productbetween Aβ and τ . The pile-up spectrum extends from 0to 2E0 (it is assumed to be simply the convolution of thebeta spectrum with itself, Npile−up(E) ∝ N(E) ∗ N(E),where N(E) is the 187Re beta spectrum). The greatestreduction on the limit of mν happens when increasingthe statistics. In fact if the pile-up is negligible, the 90%confidence limit sensitivity is proportional to 4

√1/Nev.

These behaviours are confirmed by the Monte Carlosimulation (for details, see A. Nucciotti et al. in theseproceedings).

The MARE project has a staged approach: the goal ofthe first phase (MARE-1), which consists of two inde-pendent experiments using the presently available detec-tor technology, is to reach a sensitivity of the order of1 eV and to explore the systematic uncertainties specificof the microcalorimetric technique. The goal of the lastphase (MARE-2) is to achieve a sub-electronvolt sensi-tivity on the neutrino mass and consists in the deploy-ment of several large arrays of thermal microcalorime-ters.

3. MARE-1 IN MILAN

One of the MARE-1 experiments is carried out in Milanby the group of Milano–Bicocca in collaboration with theInsubria University in Como, NASA/GSFC at Greenbeltand University of Wisconsin at Madison. We presenthere the status of the experiment installed in one of thedilution refrigerators located in the cryogenic laboratoryof the Milano-Bicocca University.

The Milan MARE-1 arrays are based on semiconduc-tor thermistors, provided by the NASA/Goddard group.These arrays, developed as detectors for the XRS2 ex-periment on the ASTRO-EII mission, consist of 6 x 6

FIGURE 1. Structure of AgReO4 microcalorimeter.

implanted Si:P thermistors with a size of 300 x 300 x1.5 µm3. An energy resolution of 3.2 eV FWHM at 5.9keV has been obtained with these thermistors and HgTeabsorbers [6].

Adapting these thermistors, optimized for X-ray spec-troscopy, to the measurement of the beta spectrum ofthe 187Re, single crystals of AgReO4 replace the HgTeabsorbers and act as absorber. An R&D work has beendedicated to determine the best crystal geometry andthe optimal thermal coupling between Si thermistors andAgReO4 absorbers.

Our studies have shown that with crystals cut in reg-ular shape of 600 x 600 x 250 µm3 (approximatively500 µg, giving 0.27 decays/sec) and gluing silicon piecesof 300 x 300 x 10 µm3 between the thermistor and therather large absorber, it is possible to achieve an energyand time resolution of 25 eV and 250 µs respectively.The AgReO4 are grown by Mateck GmbH (Germany).Mateck has developed a procedure to grow large singlecrystals with high purity and to cut them in regular shapeas precisely as possible. A spectrum of one test detectorcan be seen in Fig. 2. The thermal coupling of this de-tector is made of Araldite Rapid between the thermistorand the silicon spacer and made of ST2850 epoxy be-tween silicon spacer and AgReO4absorber. Fig. 1 showsa sketch of our microcalorimeters.

With 288 detectors and such performances, a sensitiv-ity of 3.3 eV at 90 % CL on the neutrino mass can bereached within 3 years. This corresponds to a statisticsof about 7x109 decays [7].

The performance of such bolometers, which are char-acterized by high impedance at low temperature (around4 MΩ at 85 mK), depends not only on the thermistorsand the quality of the crystals but also on the read-outelectronics and the wiring of the apparatus. A cold unitygain buffer stage, based on JFETs working at 135 K, isinstalled to shift down the impedance of the thermistors.For the expected operative conditions, the JFETs havea white noise of 0.8 nV/Hz1/2. This first stage, placedas close as possible to the detectors, is followed by an

566

h1Entries 4096

Mean 3.213

RMS 2.104

energy[keV]0 1 2 3 4 5 6 7 8

cou

nts

0

100

200

300

400

500

600

700

800

h1Entries 4096

Mean 3.213

RMS 2.104

Energy spectrum

FIGURE 2. A spectrum of one test detector with an energyresolution of 33 eV at 2.6 keV in the presence of a calibrationsource. The calibration is made with the fluorescence lines (Al,Cl, Ca, Ti and Mn) of targets illuminated with 55Fe source.

amplifier stage at room temperature. This amplifier stagemakes the difference between the signal presents at thecold buffer output and the reference ground signal. Thepresence of only one reference ground signal for all thechannels cancels the ground loop interference. The out-put signal is filtered with an active Bessel low pass an-tialiasing filter, placed close to the acquisition system.

The wiring has been optimized to reduce the parasiticcapacitance and the microphonic noise. To electricallyconnect the cold electronics (135 K) to the detectors at 85mK with low thermal conductance wires, microbridgesare produced by the Memsrad/FBK in Trento, Italy. Themicrobridges are thin wires (thickness around 200 nm)made of Ti or Al deposited onto polyamid. The siliconwafer below these wires is completely etched away. Tomanipulate them, the chip presents two wings which con-nect the two ends of the wires. After mounting, the wingsmust be carefully cleaved along the furrows, made dur-ing the microbridges fabrication. Ti or Al microbridgesare glued on the dedicated PCB with epoxy resin. Themicrobridges provide thermal decoupling between de-tectors and JFETs, but they do not guarantee mechani-cal stability. Therefore, materials with very low thermalconductivity are used as a mechanical support, namelyKevlar and Vespel.

A first microbridge stage provides a thermal decou-pling between the detectors and the JFET holder (4 K).These microbridges are made of titanium and the typicalresistance is 3 kΩ. The 4 K parts are suspended by threeKevlar crosses. This structure can be seen in Fig. 3.

A second step of thermal decoupling is made in theJFET box. The microbridges connecting the JFETs totheir holder are made of aluminium and the resistanceof one wire is 10 Ω at 77 K. The dedicated PCB issuspended by two thin Vespel rods. Thanks to the thermalresistance of microbridges and Vespel rods, the JFET

FIGURE 3. The assembly of the first decoupling stage. Thispicture shows the three Kevlar crosses connecting the detectorholder to the 4 K parts. Due to fragility of microbridges, thesuspended parts are reinforced by angles and clips which areremoved just before closing the dilution refrigerator.

FIGURE 4. The assembled JFET box. This picture shows thefour sets of Al microbridges, 40 JFETs soldered on the top ofthe dedicated PCB (the other 40 are soldered on the bottom)and the ceramic board. The PCB is screwed on Vespel rods.Heaters and thermometer are connected via spare lines of themicrobridges of the output.

bias is almost sufficient to warm up the preamplifiertemperature to its optimal working temperature of 135K. The assembled JFET box can be seen in Fig. 4.

The cryostat wiring is thermalized through pieces ofcopper with three slits producted using an electroerosionmachine. Taking a serpentine path, the manganin wiresare glued with epoxy resin inside the copper pieces.

We tested the cold buffer stage: the thermal bath wasat the nitrogen temperature and the JFETs at 130 K. Allthe channels are good and in this condition the whitenoise measured is of 2.1 nV/Hz1/2. This Johnson noise isprimarily due to the wiring between the cold electronicand the amplifier stage. At 1 Hz the noise is about 5nV/Hz1/2.

567

FIGURE 5. The cryogenic setup of MARE-1 in Milan fortwo arrays. The final assembly is 300 mm high.

A calibration source completes the set-up. The energycalibration system, located between the detector holderand the JFETs boxes, consists of fluorescence sourceswith 10 mCi of 55Fe as a primary source movable in andout of a Roman lead shield [8]. The fluorescence targets,made of Al, Si, NaCl, CaCO3 and Ti, allow a preciseenergy calibration around the end-point of 187Re with theKα and Kβ X-rays.

The cryogenic setup of MARE-1 in Milan is shownin Fig. 5. Mounted in a Kelvinox KX400 dilution re-frigerator, it can host up to eight arrays (288 detectors),although only two of them with electronics have beenfunded so far. For that reason, read-out cryogenic wiring,preamplifers, anti-aliasing filters, triggers and DAQ sys-tem have been installed only for 80 channels.

A first test run with two arrays but only 11 AgReO4crystals attached is planned to start shortly. The goals ofthis run are to check the functionality of all the channelsand a final test for the thermal coupling between thermis-

FIGURE 6. The XRS2 array with the first 11 AgReO4 crys-tals. The crystals are cut in regular shape as precisely as possi-ble.

tors and silicon spacers. Therefore two different kinds ofepoxy resins are tested: five silicon spacers are attachedwith Araldite Normal and the other six with ST1266epoxy. ST2850 epoxy is used to glue all the AgReO4absorbers on the silicon spacers. The 11 crystals on thearray can be seen in Fig. 6.

After this test, the remaining AgReO4 crystals will beattached to the thermistors and the complete measure-ment of 72 channels will start. With two arrays equippedwith AgReO4 absorbers, a sensitivity of 4.5 eV at 90%C.L. is expected in three years. Based on the results ob-tained from the two arrays, a decision concerning fund-ing of the deployment of the remaining six arrays can bemade.

4. CONCLUSIONS

The first phase of MARE-1 in Milan is getting ready tostart using silicon implanted thermistors array equippedwith AgReO4 single crystals absorbers. The experiment,which starts with 72 channels, is designed to be expendedto 288 channels to increase the sensitivity on the neutrinomass.

REFERENCES

1. Ch. Kraus, Eur. Phys. J. C 40, 447–468 (2005)2. V. M. Lobashev, Nucl. Phys. B (Proc Suppl.) 91, 280–286

(2001)3. M. Sisti et al., NIM A 520, 125 (2004)4. F. Gatti et al., Nature 397, 137 (1999)5. C. Arnaboldi et al., Phys. Rev. Lett. 96 (2006)6. F. S. Porter et al. Nucl. Inst. and Meth. in Phys. Res.A 559,

436-438 (2006)7. A. Nucciotti, J. Low Temp. Phys. 151, 597–602 (2008)8. D. Schaeffer et al., J. Low Temp. Phys. 151, 623–628

(2008)

568