Underground Cryogenic Detectors Ettore Fiorini Dipartimento di Fisica dell' Università di Milano Sezione di Milano dell'Istituto Nazionale di Fisica Nucleare

The joint effort of cryogenics, particle physics, and astrophysics to exploit thermal detectors should lead to exciting and perhaps unexpected results.

Fundamental physics, having success­ fully unified the electromagnetic and weak interactions, is now attempting to unify the strong force and hopefully even the gravita­ tional force. The so-called astro-particle physics connection is destined to play an important rôle. Searches with high-energy particle accelerators can be complemented in a wonderful way by studying, for example, the origin of the Universe and Its remnants, Fig. 1 — Techniques based on superconduc­ the physical processes accompanying the tivity for detecting fundamental particles: origin, and the life and the death of a star. schematic illustrations of flux exclusion from The astrophysics approach has led to the Superconducting Normal conducting a granule (a. left) and of a Josephson-type construction of large laboratories placed detector (b, right). deep underground to investigate the inter­ actions of the penetrating component of sider, for instance, a granule of a Type I Performance cosmic rays as well as the rare processes superconductor kept at a fixed temperature When comparing cryogenic detectors with arising from the spontaneous decay of par­ in a magnetic field slightly below the critical standard detectors based semiconductors ticles or nuclei. Noise from cosmic rays is field. The granule is in a metastable super­ and on ionization or proportional counters, suppressed because the rays are absorbed conductor state and as such repels the one should note that the standard devices by the rock overburden. magnetic field owing to the Meissner effect collect only the fraction of energy delivered The astrophysical approach has also (Fig. 1a). Heat delivered by an incoming by the particle in form of ionization (about stimulated the development and application particle raises the temperature of the gra­ 30% of the total energy). The remainder of of techniques which are either totally new or nule, bringing It suddenly to the normal state the energy goes mainly Into phonons (heat), at least seldom or never used In particle and the field is no longer repelled. The result with perhaps a component into lattice dislo­ physics. The thermal detection of nuclear is a magnetic pulse that can be detected. cations in the case of nuclear recoils. Ther­ radiation, perhaps the most promising of Another technique, based on the Joseph- mal detectors collect directly and in the form these techniques, played an Important rôle son effect, exploits a detector comprising a of heat at least 70% of the energy delivered, in the development of nuclear physics. In pair of superconductors separated by a thin, and up to 100% if the electron-ion and 1903, Curie and Laborde published a paper insulating layer. The superconductors in­ electron-hole pairs recombine within a suffi­ entitled Le chaleur dégagée spontanément duce breaking of the Cooper pairs to give a ciently small time interval. Bolometers have par les sels de radium (The heat released current pulse passing through the insulator in fact proved to be excellent detectors of by radium salts) and in 1927 Ellis and which is amplified and measured (Fig. 1b). nuclear recoils, which deliver a much lower Wooster found in the decay of 210Bi missing More appropriate for "underground cryo­ fraction of energy as ionization than fast, thermal energy which was subsequently genics” would appear to be another and in charged particles, and consequently a much attributed by Pauli to the emission of neu­ some ways simpler technique, which like the larger fraction in the form of heat. However, trinos. The thermal detection of single par­ others, is generally called bolometric. Con­ thermal detectors are “slow” since the rise ticles was suggested independently in sider a very pure diamagnetic and dielectric time is related to the velocity of sound (the Europe [1] and the USA [2] for experiments crystal: its heat capacity Cv (in joules per in elementary particle physics some eight degree Kelvin) at low temperatures is given years ago. Since then, the development of quantitatively by the expression: these techniques has been impressive [3] Cv = 1944 (V/Vm) (T/TD)3 and their application In experiments of the where V and Vm are the crystal and molecu­ passive type [4], namely without accelera­ lar volumes, and T and TD are the operating tors, seems almost within reach. and Debye temperatures. It is clear that for TD sufficiently large and T sufficiently small, Techniques the heat capacity can be so small that even Some of the thermal techniques for de­ the minute amount of energy delivered by tecting elementary particles are based on an incident particle in the form of heat can the effect that the thermal “action” of a par­ produce, in the crystal, a sizable increase of ticle may play on a metastable system. Con- temperature. This temperature pulse can be transformed Into an electric pulse using a suitable thermistor in thermal contact with Ettore Fiorini is Professor of Advanced Phys­ the crystal (Fig. 2). The thermistor, whose ics and Director of the high-energy physics group in the Physics Department, University of resistance we call Rb, is biased by a battery Milan. He studied at the university and has and a load resistor with resistance RL pos- worked at Duke University in the USA, and at sibly considerably larger than Rb. The ener­ the Paul Scherrer Institute and CERN in Swit­ gy delivered by the particle to the crystal Fig. 2 — A schematic illustration of a bolome­ zerland. His group participates in Gran Sasso’s produces a negative voltage pulse across ter and its electronic circuit which transforms GALLEX experiment but is mainly involved in the thermistor which can then be amplified a thermal pulse into a detectable electrical developing cryogenic thermal detectors. and measured. signal. Europhys. News 23 (1992) 207 Unfortunately, we are still quite far from detectors placed underground. Results are these limits owing to various important prob­ still negative. We have already established, lems Including: however, that thermal detectors are likely to - non-uniformity in collecting phonons, es­ be more effective than the classical ones for pecially in large detectors; revealing nuclear recoils, and therefore dark - spatial non-uniformity of the recombina­ matter [8]. tion of electron-hole pairs trapped by va­ rious impurities; Searching for rare events - noise due to electromagnetic sources, es­ Let us take double-beta decay (DBD) as pecially microphonics [5] ; an example of a rare event. This process is - difficulties in keeping constant the temper­ very important In the framework of the con­ ature of the bolometer, and consequently its servation of the lepton number, where two gain. electrons are emitted simultaneously. Neu­ trinoless DBD, If observed, would signal the Results violation of lepton number conservation and Despite these difficulties the performance the existence of massive Majorana neutri­ of some thermal detectors is already very nos — unambiguous evidence of a new promising. In the limit of very small detec­ physics beyond the Standard Model [9]. tors (where the heat capacity is obviously There has been a suggestion to use for this intrinsically very small), a NASA-University process a detector made with a DBD active of Wisconsin collaboration recently achie­ material that would act at the same time as ved a resolution of 7.3 eV (Fig. 3) in the source of the decay and as a detector to measurement of x-rays of a few keV [3]. reveal it. Many materials are good candi­ This resolution is already 20 times greater dates for DBD and possess at the same than for any existing semiconductor detector. time mechanical and thermal properties that More interesting for underground cryo­ make them suitable for constructing high- genics is the successful construction of performance croygenic bolometers. In par­ large detectors using materials with both low ticular, the group from Milan recently instal­ and high atomic numbers Z [6], where led in the Gran Sasso Laboratory, Assergi, Fig. 3 — A comparison between the x-ray masses up to about 340 grams have been Italy, a DBD experiment based on 130Te in spectroscopic performance of a Si(Li) diode reached [7]. the form of a 20.9 g TeO2 crystal mounted in (upper) with that for the NASA-Wisconsin a dilution refrigerator (Fig. 4a). It has achie­ 158Tb detector (lower). The bolometric de­ Applications ved a sensitivity three orders of magnitude vice gave a resolution which is some 20 Dark matter searches better than was reported previously [10]. times better than for the semiconductor de­ It is well known that the density in the Uni­ tector. verse of the visible (more correctly, the bar- High-energy qamma-ray detection yonic) matter is at least one order of magni­ It is well known that gamma rays are nor­ time needed for heat to distribute itself in the tude less than the critical value needed to mally detected by searching for peaks in entire crystal), while the decay time is “close” the expansion of the Universe. More energy due to interactions in a detector related to the thermal resistance between significant, at least for an experimentalist, is involving the photoelectric effect and pair the bolometer and the heat sink. Thermal the fact that various measurements such as formation. The cross-sections for these pro­ detectors are consequently excellent for those on the rotation of galaxies and on the cesses increase strongly with atomic num­ searching for rare events in underground relative motion of single galaxies of the ber. Excellent gamma-ray spectroscopy can experiments, but usually unsuitable for same group, indicate that the average den­ therefore be carried out with thermal bolom­ working at the high counting rates and large sity of matter in the Universe is an order of eters constructed using a high-Z material, backgrounds found at accelerators. magnitude larger than the baryonc density. as has already been demonstrated for a tel­ An important figure-of-merit for thermal The origin of this extra mass and the nature lurium oxide detector (Fig. 4b) made by the detectors is the energy resolution. It can be of dark matter are still unclear. Candidates Milan group [11]. > optimized by adapting the characteristics of for dark matter include massive, but light, the preamplifier to the impedances of the neutrinos, heavy neutrinos, small black bolometer and the load resistors, and by holes, and new but still undiscovered par­ choosing the best value for the difference ticles such as those predicted by theories between the temperatures of the bolometer for supersymmetry that try to unify all the and the heat sink. The overall resolution of forces in Nature. The solar system and the detector (full width at the half maximum: Earth presumably move with respect to FWHM) can be as small as: these particles, which should then be revea­ ΔE = 2.36 ξ √k CVT² led by their interactions in a suitable de­ tector placed underground and shielded where k is Boltzmann’s constant and ξ is against radioactivity [4]. If dark matter con­ a dimensionless parameter usually of the sists of WIMPS (Weak Interacting Massive order of a few unities. The following exam­ Particles) the only way of detecting them ples demonstrate that the resolution should directly is via their elastic scattering on a be, in principle, much better than for any nucleus in a detector where the nuclear other detector: recoil energy is measured. Some experi­ - A cubic crystal of silicon (TD = 645 K) of ments have already been carried out by 1 mm side kept at 20 mK would have a heat searching for these recoils in semiconductor capacity of 5 x 10-15 J/K and a FWHM reso­ lution of 0.1 eV. - A germanium thermal detector (TD = 370 Fig. 4 — Cryogenic TeO2 detectors that have K) with 100 mm side (more than 5 kg in been developed by the Milan group to search weight) operated at the same temperature for double-beta decay (a, upper) and to per­ would have an heat capacity around 2.5 x form gamma-ray spectroscopy (b, lower). 10-8 J/K and a resolution of about 150 eV. The diagram gives a cross-section through Both the overall mass and the resolution the copper frame holding the TeO2 crystal would be much larger than for existing ger­ which was used for a); the photograph shows manium diodes. a similar detector that was employed for b). 208 Europhys. News 23 (1992) Underground Cryogenics surface without special shields, the thermal [1] Fiorini E. and Niinikoski T.O., Nucl. Intr. & As the events being sought in experi­ energies delivered by cosmic rays and by Methods 224 (1984) 83. ments on dark matter or double-beta decay radioactivity are comparable, amounting in [2] Moseley S.M., Mather J.C. and McCam­ are rare, and since the pulses are long, it is total to 0.3 pW/g: this is only an order of mon D., J. Appl. Phys. 56 (1984) 1257. essential to reduce spurious counting aris­ magnitude less than the total heat leak [3] Proc. 3rd Int. Workshop on Low Tempera­ ing from cosmic rays and environmental encountered experimentally in the microkel- ture Detectors for Neutrinos and Dark Matter, radioactivity. The dilution refrigerator being vin region. It will therefore become essential Eds: L. Brogiato, D.V. Camin and E. Fiorini operated at Gran Sasso by the Milan group in the near future to carry out searches for (Editions Frontières) 1990. is in fact the world’s first deep underground rare events in an underground laboratory [4] Fiorini E., Proc. 2nd Conf. of the Euro­ cryogenic plant (Fig. 5). The device is con­ where contributions from cosmic rays and pean Materials Res. Soc., Strasbourg, 1988 structed from specially selected materials radioactivity are reduced by many orders of (Elsevier, 1989). having low levels of radioactivity. A high-Z [5] Camin D.V., Pessina G. and Previtali E., magnitude. This single example shows how Nucl. Instr. & Methods A 315 (1992) 385. bolometer with a TeO2 detector crystal the joint effort of cryogenics, particle physics weighing 340 g has been installed. This may [6] Fiorini E., Physica B 167 (1991) 388 ; and astrophysics should lead to exciting and Proc. Low Temperature Detectors for Neu­ represent the beginning of “underground perhaps unexpected results. cryogenics”, the importance of which could trinos & Dark Matter IV, Eds.: N.E. Both & G. Salmon (Editions Frontières) 1992. extend well beyond the field of nuclear and [7] Alessandrello A., et al, in Proc. 24th Int. subnuclear physics. Conf. on High Energy Physics (Dallas, USA, Consider, as a final example, attempts to 1992), to be published. reach very low temperatures (tens of micro­ [8] Smith P.F. and Levin J.D., Phys. Rep. 187 kelvin) where up to now heat from radio­ (1990) 283; Rich J., Relativistic Astrophysics, activity and cosmic rays has been neglec­ Cosmlogy and Fundamental Physics, Ann. ted. In a normal laboratory at the Earth’s New York Acad. of Sciences (New York, NY, USA) 1991, p. 357.; Fiorini E., ibid., p. 446. [9] Tomoda T., Rep. Prog. Phys. 54 (1991) 53; Moe M.K., Nuclear Phys. B (Proc. Suppl.) 19 (1991) 158; Fiorini E., Nucl. Phys. News 1 Fig. 5 — Schematic illustration of the (1991) 17; Klapdor H.V., J. Phys. G: Part. Milan group’s underground detector oper­ Phys. 17 (1991) 1. ating at Gran Sasso laboratory. The large [10] Alessandrello A., et al., Phys. Lett. B (340 g) tellurium oxide crystal is supported 285 (1992) 176. inside a specially constructed, shielded [11] Alessandrello A., et al , Nucl. Instr. & Me­ dilution refrigerator. thods A 320 (1992) 388.

universe gives Ω = 0.2. Many astronomers, Cosmology with Supernovae however, favour a higher density (Ω = 1) because it is difficult to form galaxies with a Hans Hippelein from the Max Planck Institute for Astronomy in Heidelberg, Germany, smaller value. This higher value is only pos­ describes the significance of a recent observation of a Type 1a supernova “candle” at a sible by introducing hypothetical “dark” mat­ record distance. ter which interacts with the visible matter through gravitational forces. Type 1a supernovae (SN1a) are the result In April 1992, S. Pearlmutter, C. Penny- of deflagration and detonation of accreting packer, G. Goldhaber, and other astrono­ stars. These exploding stars shine half as mers from Berkeley, USA, together with brightly as a whole galaxy (maximum bright­ scientists from Cambridge and Durham in ness MB = 19.8 mag) for about a week and the UK and from Stockholm used the 2.5 m can be easily detected up to a redshift of z ≈ Isaac Newton Telescope in La Palma, Ca­ 0.5, corresponding to a distance of about nary Islands, for one week to see if they 5.109 light years. The scatter in maximum could apply the new method. They scanned brightness is small (0.5 mag; ≈25%) be­ 10 000 deep-space galaxies in a search for cause all Type 1a supernovae arise from Type 1a supernovae, where the probability exactly the same kind of star: a white dwarf of locating a SN1a is only about one per gal­ which is fed by material from a companion axy in 500 years. Pearlmutter and col­ star until it reaches a critical mass of 1.4 leagues were therefore extremely lucky times that of the sun and explodes. when they had a detection, near the border Because of the high brightness and the between the Hercules and Corona Borealis small scatter in brightness, SN1a are proba­ constellations, with z = 0.457 determined bly the best “candles” for studying the ex­ from the Doppler broadening of spectral pansion of the universe. At distances z = lines. The measured z is considerably larger 0.5, the cosmological effect of deceleration Apparent brightness m of Type 1a superno­ than the previous record of 0.31 for Type 1 of the expansion of the universe due vae versus redshift z (logarithmic scale) for a supernovae. However, this single observa­ to gravitational forces between galaxies is Hubble constant of 50 km S-1 Mpc-1 for diffe­ rent values of the deceleration parameter q0 tion is insufficient for discriminating between comparable to the variation in the maximum which is directly related to a parameter Ω density parameters of 0.2 and 1. brightness of supernovae. describing the mean density of matter in the Pearlmutter says that the detection of 25 By observing a large number of super­ universe. By plotting m for different Type 1a Type 1a supernovae (which requires about novae one can plot a curve describing the supernovae as a function of the observed z one to two years of observation) would allow apparent brightness as a function of the red- one can therefore in principle determine q0, £2 to be pinned down to within about 30%, shift (see figure). The deceleration parame­ and hence Ω, without knowing the value of sufficient to distinguish between a low-den­ ter q0 is given by the slope of the curve, the Hubble constant. sity universe with Ω = 0.2 and a higher den­ without knowing the Hubble constant which sity one with Ω = 1. The number 25 seems relates redshift and the distance of galaxies optimistic considering absorption and other in the expanding universe. Using q0, astron­ For Ω<1, the universe will expand forever effects that have to be allowed for. In princi­ omers can directly estimate £2, the mean and for £2>1 the expansion will eventually ple, however, the method should work and mass density parameter in the universe halt and reverse course into a contraction. be able to solve one of today’s major prob­ (e.g., for the Friedman model Ω = 2q0). The observation of luminous matter in the lems in astronomy within a few years.

Europhys. News 23 (1992) 209