ARTICLE IN PRESS

(1)

Production of low-background CuSn

₆

-bronze for the CRESST dark-matter-search experiment

B. Majorovits^a,, H. Kader^b, H. Kraus^a, A. Lossin^b, E. Pantic^c, F. Petricca^c, F. Proebst^c, W. Seidel^c

aDepartment of Physics, University of Oxford, Keble Road OX1 3RH, UK

bNorddeutsche Afﬁnerie, Hovestr. 50, 20539 Hamburg, Germany

cMPI fu¨r Physik, Fo¨hringer Ring 6, 80805 Munich, Germany

a r t i c l e i n f o

Article history:

Received 5 May 2008 Received in revised form 26 September 2008 Accepted 27 September 2008

PACS:

95.35.+d 29.25.Rm Keywords:

Low-background experiment Radiopure materials Dark matter search

a b s t r a c t

One of the most intriguing open questions in modern particle physics is the nature of the dark matter in our universe. As hypothetical weakly interacting massive particles (WIMPs) do interact with ordinary matter extremely rarely, their observation requires a very low-background detector environment regarding radioactivity as well as an advanced detector technique that allows for active discrimination of the still present radioactive contaminations. The CRESST experiment uses detectors operating at milli-Kelvin temperature. Energy deposition in the detectors is recorded via the simultaneous measurement of a phonon-mediated signal and scintillation emitted by the CaWO4 crystal targets.

The entire setup is made of carefully selected materials.

In this note we report on the development of ultra-pure bronze (CuSn₆) wire in small quantities for springs and clamps that are currently being used in the CRESST II setup.

1. The CRESST dark-matter-search experiment

The cryogenic rare event search with superconducting thermometers—CRESST—is an experiment dedicated to the search for hypothetical WIMP particles as the solution for the dark-matter problem (Goodman and Witten, 1985). The CRESST experiment is placed at the underground site of the Laboratori Nazionali del Gran Sasso (LNGS). It is using scintillating CaWO4 calorimeters that are read out using superconducting tungsten phase-transi- tion thermometers (Meunier et al., 1999). A small rise in temperature creates an appreciable increase of the resistance of the tungsten ﬁlm. This is measured using SQUIDs (Henry et al., 2007). Additionally, and in coincidence, a scintillation signal is measured using specially developed light detectors (Petricca, 2005). The amount of scintillation light produced inside the crystal depends on the energy deposited and on the interaction type, i.e. nuclear recoil or electron recoil. Recorded in coincidence, these two signals permit discrimination of background events that are being produced by radioactivity, largely depositing the energy of emitted radiation or particles via interaction with the electron system of the CaWO4crystal.

A nuclear recoil, through which energy is transferred to a nucleus, produces considerably less light output for the same energy deposited. This is due to the different ionization densities of the two processes.

In the second phase of the CRESST experiment a total of 33 CaWO4

crystals with a total mass of up to approximately 10 kg will be used for the search for dark-matter particles (Angloher et al., 2004).

In order to house such an amount of crystals in the experimental setup a completely new design had to be developed.

The new holder system is shown inFig. 1. The bulk of the material is low-heat leak NOSV copper¹from the Norddeutsche Afﬁnerie. It is the same material that had been used for the earlier detector holders of the ﬁrst phase of the CRESST experiment (Bravin et al., 1999). Since the holders, carrying the crystals, have to be protected from external mechanical vibrations, the whole system rests on up to six compression springs (seeFig. 1).

2. Selection of material for the springs

There exist severe constraints regarding the materials that can be used for production of the holder system. The springs are not

ARTICLE IN PRESS

Contents lists available atScienceDirect

journal homepage:www.elsevier.com/locate/apradiso

Applied Radiation and Isotopes

doi:10.1016/j.apradiso.2008.09.015

Corresponding author. Present address: MPI fu¨r Physik, Fo¨hringer Ring 6, 80805 Mu¨nchen. Tel.: +49 89 32354262; fax: +49 89 32354528.

E-mail address:bela@mppmu.mpg.de (B. Majorovits).

1NOSV is the trade name of Norddeutsche Afﬁnerie for their highest conductivity copper with residual resistivity ratio ofX400.

Applied Radiation and Isotopes 67 (2009) 197–200

(2)

allowed to contain any magnetizable materials, as otherwise the setup could not be cooled down to its operating temperature in the milli Kelvin regime. To maintain a low radioactive background all the materials in the direct vicinity of the CaWO4crystals must be carefully selected for their radiopurities as otherwise they could considerably contribute to the background of the experiment.

As can be seen fromFig. 1the compression springs holding the experimental table are positioned in close proximity to some of the detector modules. Therefore, requirements regarding tolerable levels of radiopurity of the springs are rather stringent.

Similarly, severe requirements apply to the clamps holding the CaWO4crystals and the light detectors (seeFig. 2).

In earlier measurements the CaWO4crystals were held in place by clamps made from polytetraﬂuoroethylene (PTFE). However, in these data taking runs events occurred where energy deposition could only be detected by phonons (phonon channel), but the light detectors did not respond (phonon-only-events). These phonon-only-events are attributed to the extreme stiffness of of the PTFE clamps at very low temperatures. In such operating condition, the PTFE clamps can cause tiny fractures in the crystal.

Such fractures can be observed in the crystals that were held in place by the PTFE clamps. Whenever a new fracture process occurs, this produces phonons, which can be observed in the phonon channel but the fracture does not create scintillation light, or at least not enough scintillation for detection (A˚stro¨m et al., 2006).

In the two last runs of CRESST, before the start of the upgrade in 2004, the PTFE clamps had been replaced by Ag-coated copper–beryllium (CuBe2) clamps. As a consequence, the phonon-only-events disappeared. However, as a new feature in the low-energy spectrum a peak around 47 keV appeared (Angloher et al., 2005) originating from the beta decay of²¹⁰Pb (seeFig. 3).

As the change of the crystal clamps was the only major change in the setup we attribute the 47 keV peak to a contamination of the CuBe2used for the production of the clamps or its silver coating.

As the most promising suitable material for spring production we found CuSn6-bronze, where both components of the alloy—

copper and tin—can be obtained in a controlled high-purity quality.

ARTICLE IN PRESS

Fig. 1.Left picture: part of the CRESST detector holder structure during mounting at the LNGS. In total 33 detector modules with up to 300 g detector material per module can be placed into the holder structure. Each detector module is accessible individually without having to remove other crystals from the setup. The detector holder structure rests on six CuSn6-bronze springs for decoupling from vibrations.

Two detector modules are visible on the top of the left tower, two further on the right tower. The six CuSn6springs of which one is visible here are at the height of the upper detectors. Right picture: two CuSn6springs with 3 mm (left) and 2.5 mm (right) wire diameter.

Fig. 2.Upper picture: Cresst II detector module. On the left, the CaWO4crystal can be seen as it is held inside the copper holder system. It is surrounded with a reﬂective scintillating foil. The crystal is held inside the structure with the help of CuSn6-bronze springs that are wrapped in scintillating foil. The crystal diameter is 40 mm. On the right, the light detector can be seen. Similarly it is held inside the copper structure using CuSn6springs. Lower pictures: left: CuSn6clamps used to ﬁx the CaWO4crystals. Right: CuSn6clamp wrapped in scintillating foil to hold the light detectors in their copper holder structure.

Fig. 3.Low-energy spectrum of one of the detectors taken from the previous run (Angloher et al., 2005). Clearly visible is the peak at 46.5 keV that is attributed to a contamination of the CuBe2clamps with²¹⁰Pb.

B. Majorovits et al. / Applied Radiation and Isotopes 67 (2009) 197–200 198

(3)

3. Production of CuSn6wire

In order to obtain CuSn6 springs that are of controlled purity we produced our own alloy out of electrolytically produced NOSV copper from Norddeutsche Afﬁnerie AG and 99.9999% Sn procured from Alfa Aesar.

The copper pieces were melted in a SiC crucible of about 30 cm diameter. The crucible was heated in an electric resistance furnace up to 12001C at ambient air. The tin metal was fed into the copper melt and the melt was homogenized by agitation for a short time.

Afterwards, a sample was taken to assess and control the correct alloy composition.

In order to keep the oxygen content of the melt at low level (about 5 ppm), ultra-pure graphite from Superior Graphite was used as top layer above the melt. The graphite protected the melt against oxygen from ambient air and, in addition, acted as reduction material.

A Top Cast cooler consisting of a graphite nozzle with 8 mm diameter at the point and a shaft made from a metal pipe with cooling function was used to cast 8 mm wire directly from the melt. The point of the Top Cast cooler was pressed about 5 cm into the bronze melt and the wire was pulled directly with a speed of about 1 m/min.

The Top Cast procedure resulted in 23 kg of high purity CuSn6

wire. This wire was homogenized for 16 h at 7501C.

Out of this material 2.1 and 1.7 kg were used to produce the 3.0 and 2.5 mm wire, respectively, for the springs of the holder system. In total 32 and 40 m of 3.0 and 2.5 mm spring wire were produced, respectively. These were annealed for 1.5 h at 2701C in order to achieve the required stiffness of roughly 900 N=mm². It was used to produce the springs shown inFig. 1.

The clamps for the crystals with 0.4 mm thickness and for the light detectors with 0.2 mm thickness were first milled and then rolled down to the required thickness in two stages: first from 8 mm wide wire sheets with 1.3 and 0.7 mm thickness, respectively, were produced. These were annealed for 5 h at 4501C, before the final material with thicknesses 0.4 and 0.2 mm was rolled from the 1.3 and 0.7 mm plates, respectively. A picture of the clamps, installed into the detector holders, is shown inFig. 2.

4. Results: cleanliness of the CuSn6-bronze produced

The materials used for the production of the CuSn6wire as well as two samples of the resulting CuSn6-bronze have been analyzed, using a glow discharge mass spectrometry (GDMS) analysis (Betti and Aldave de las Heras, 2003) performed by SHIVA Technologies Europe.²The results are given inTable 1.

Most of the elements detected in the samples are expected in the material due to its history of production. The following elements could be detected in excess of the intrinsic contaminations of the materials used: Al (0.34/0.33 ppm), Si (22/18 ppm), Cr (0.15 ppm), Mn (0.26/0.30 ppm), Pb (12/14 ppm). Most likely these enter the melt through the Al2O3crucible or the lid, both of which were not new. Al, Si, Cr and Mn could also have entered through the ultra-pure graphite or the tin. Since they do not have any long lived naturally occurring isotopes they are of no concern regarding radiopurity of the produced materials and where thus not analyzed in the GDMS analysis of the tin and the graphite.

As seen inTable 2 for uranium, thorium and potassium only upper limits can be given. These are 0.1 ppb for U and Th and 10 ppb for K. The values can be translated to upper limits of

0.31 mBq/kg for⁴⁰K, 0.4 mBq/kg for²³²Th and 1.2 mBq/kg for²³⁸U.

Limits on the activity of the long-lived radioactive isotopes are given inTable 2.

210Pb could so far not be detected in the CuSn6material. The presence of²¹⁰Pb is usually a consequence of contamination of the raw material with ²³⁸U. ²³⁸U decays in a chain to the stable isotope²⁰⁶Pb via²¹⁰Pb. In the ore the decay chain will typically be in secular equilibrium. Thus the activity of ²¹⁰Pb in the ore material will equal its²³⁸U activity. If the material subsequently undergoes purification, U and Pb are separated from the material to be purified—typically with different efficiencies. The radiopurity of the purified material will be reduced accordingly. The

ARTICLE IN PRESS

Table 2

Limits on activity of several long lived isotopes for the CuSn6-bronze as well as for the materials used for its production

Radioactive isotope

Half-life (10⁹years)

Activity CuSn6and NOSV (mBq/kg)

Sn (mBq/kg)

Graphite (mBq/kg) K ⁴⁰K 1.28 p0.31 p0.31 p1.55 Rb ⁸⁷Rb 48.0 p4.49 p4.49 p8.98 Sm ¹⁴⁷Sm 106 p0.62 p0.62 p1.24 Lu ¹⁷⁶Lu 38 p0.26 p0.26 p0.52 Th ²³²Th 14.1 p0.41 p4.1 p4.1 U ²³⁵U 0.70 p0.06 p0.6 p0.6

238U 4.47 p1.23 p12.3 p12.3 The activities were obtained by converting the measured values fromTable 1to radiocontamination level.

Table 1

Results of the GDMS analysis of the NOSV copper from Norddeutsche Afﬁnerie, the 99.9999% Sn from Alfa Aesar, the graphite used for reduction of the copper from Superior Graphite and of two samples of the resulting CuSn6-bronze

Graphite (ppm)

NOSV copper (ppm)

Sn (ppm)

CuSn6

(ppm)

Al – p0:08 – 0.34/0.32

Si – p0.02 – 22/18

K p0.05 p0.01 p0.01 p0.01 V 2.3 p0.001 p0.005 0.01

Cr – p0.005 – 0.15

Mn – 0.04 – 0.26/0.30

Fe 0.49 0.72 0.02 30/23

Co p0.05 p0.005 p0.05 0.07/0.03 Zn p0.05 0.03 p0.01 8.0/8.6 Rb p0.01 p0.005 p0.005 p0.005 Sr p0.05 p0.005 p0.005 p0.005 Cd p0.05 p0.01 p0.05 p0.01 In – p0.005 p0.5 p0.005 Sb p0.05 0.65 0.9 5.5/5.4 Te p0.05 p0.01 p0.05 p0.01 Cs p0.01 p0.005 p0.05 p0.005 La 0.25 p0.005 p0.05 p0.005 Nd p0.01 p0.005 p0.005 p0.005 Sm p0.01 p0.005 p0.005 p0.005 Gd p0.01 p0.005 p0.005 p0.005 Lu p0.01 p0.005 p0.005 p0.005 Hf p0.01 p0.005 p0.005 p0.005 Re p0.01 p0.005 p0.005 p0.005 Os p0.01 p0.001 p0.005 p0.001 Pt p0.05 p0.005 p0.005 p0.005 Tl – p0.001 p0.3 p0.001/0.01 Pb p0.05 0.24 0.06 12/14 Bi p0.01 0.08 p0.005 0.13/0.16 U p0.001 p0.0001 p0.001 p0.0001 Th p0.001 p0.0001 p0.001 p0.0001 In cases where the results for the CuSn6-bronze were different for the two samples, both values are given.

2SHIVA Technologies Europe, 94, chemin de la Peyrette, F-31170 Tournefeullie, France.

(4)

238U and²¹⁰Pb activities will not be directly related anymore due to the different separation efﬁciencies. The contamination of the CuSn6 with ²¹⁰Pb is also not related to the observed Pb level detected in the GDMS analysis as the sources of the stable lead isotopes and the ²¹⁰Pb contamination are different. Thus from these data no meaningful limits can be derived.

Only 0.24 and 0.06 ppm of lead could be detected in the NOSV copper and the Sn, respectively. As these are rather low values, there must have been a step in the production history of the NOSV copper and the tin that efﬁciently removed Pb contaminations from the raw materials. This reduction efﬁciency is also true for

210Pb. Thus we do not expect any major²¹⁰Pb contamination from the materials used.

In a previous cryogenic measurement a broad structure between 46.5 and 63.5 keV (Q-value of ²¹⁰Pb) can be seen (Angloher et al., 2008). This is due to an internal contamination of the crystals with²¹⁰Pb (46.5 keV gamma energy plus contin- uous beta spectrum ending at 63 keV). As this internal contamination results in a count rate roughly ten times higher than within the 46.5 keV peak in the measurements with the CuBe clamps (Angloher et al., 2005) no conclusive limit can be given for the contamination of the CuSn6clamps so far.

In conclusion, it can be stated that the bronze produced in the Top Cast procedure, using ultra pure source materials, is well suited for low-background experiments. The analysis of the source materials conﬁrms that also the NOSV copper that has been used in large quantities in the CRESST setup is a very suitable low- background material.

Acknowledgments

We would like to thank Dr. Wehling from Superior Graphite Ltd. for supplying us with the ultra-pure graphite for reduction of the NOSV copper. We are also grateful to Mr. Friebel from Ludwigs–Maximilian Universita¨t Mu¨nchen, Dr. Lanfranchi and Dr. Potzel from Technische Universita¨t Mu¨nchen, who helped us rolling the CuSn6material.

References

Angloher, G., et al., 2004. Cresst-II: dark matter search with scintillating absorbers.

Nucl. Instr. and Meth. 520, 108–111.

Angloher, G., et al., 2005. Limits on WIMP dark matter using scintillating CaWO4

cryogenic detectors with active background suppression. Astropart. Phys. 23, 325–339.

Angloher, G., et al., 2008. Commissioning run of the CRESST-II dark matter search.

Astropart. Phys., arXiv:0809.1829, submitted.

A˚stro¨m, J., et al., 2006. Fracture processes observed with a cryogenic detector. Phys.

Lett. A 356, 262–266.

Betti, M., Aldave de las Heras, L., 2003. Glow discharge mass spectrometry in nuclear research. Spectrosc. Europe 15/3, 15–19.

Bravin, M., et al., 1999. The CRESST dark matter search. Astropart. Phys. 12, 107–114.

Goodman, M.W., Witten, E., 1985. Detectability of certain dark-matter candidates.

Phys. Rev. D 31, 3059–3063.

Henry, S., et al., 2007. The 66-channel SQUID readout for CRESST II. JINST 2, P11003.

Meunier, P., et al., 1999. Discrimination between nuclear recoils and electron recoils by simultaneous detection of phonons and scintillation light. Appl.

Phys. Lett 75, 1335–1337.

Petricca, F., 2005. Dark matter search with cryogenic phonon-light detectors. Ph.D.

Thesis, Ludwigs-Maximilians-Universita¨t Mu¨nchen, Germany.