Direct detection of WIMP dark matter

arX

iv:a

stro

-ph/

9912

554v

1 3

0 D

ec 1

999

1

Direct Detection of WIMP Dark Matter ∗

Angel Moralesa

aLaboratory of Nuclear and High Energy PhysicsUniversity of Zaragoza50009 Zaragoza. Spain

The status of the recent efforts in the direct search for Weak Interacting Massive Particle (WIMP) DarkMatter is briefly reviewed and the main achievements illustrated by the contributions presented to TAUP 99.The strategies followed in the quest for WIMPs will be first revisited and then the new results and the futureprospects reported.

1. Introduction

There is substantial evidence, reviewed atlength in this Workshop, that most of the matterof the universe is dark and a compelling moti-vation to believe that it consists mainly of non-baryonic objects. From the cosmological pointof view, two big categories of non-baryonic darkmatter have been proposed: cold (WIMPs, ax-ions) and hot (light neutrinos) dark matter ac-cording to whether they were slow or fast mov-ing at the time of galaxy formation. Withoutentering into considerations about how much ofeach component is needed to fit better the obser-vations, nor about how large the baryonic com-ponent of the galactic halo could be, we assumethat there is enough room for WIMPs in the haloto try to detect them, either directly or throughtheir by-products. Discovering this form of darkmatter is one of the big challenges in Cosmology,Astrophysics and Particle Physics.

The indirect detection of WIMPs proceeds cur-rently through two main experimental lines: ei-ther by looking in the space for positrons, an-tiprotons, or other antinuclei produced by theWIMPs annihilation in the halo, or by search-ing in large underground detectors or underwaterneutrino telescopes for upward-going muons pro-duced by the energetic neutrinos emerging as final

∗Review Talk given at the Sixth International Work-

shop on Topics in Astroparticle and Underground Physics,TAUP 99 (Paris, College de France, September 6-10,1999), to be published in Nucl. Phys. B (Proc. Suppl.).

products of the WIMPs annihilation in celestialbodies (Sun, Earth...)

The direct detection of WIMPs relies in themeasurement of the WIMP elastic scattering offthe target nuclei of a suitable detector. Pervad-ing the galactic halos, slow moving (∼ 300 km/s),and heavy (10 ∼ 103 GeV) WIMPs could make anucleus recoil with a few keV (T ∼ (1−100) keV),at a rate which depends of the type of WIMP andinteraction. Only a fraction of the recoil QT isvisible in the detector, depending on the type ofdetector and target and on the mechanism of en-ergy deposition. The so-called Quenching FactorQ is essentially unit in thermal detectors whereasfor the nuclei in conventional detectors range fromabout 0.1 to 0.6. Because of the low interactionrate [typically < 0.1 (0.001) c/kg day for spin in-dependent (dependent) couplings] and the smallenergy deposition, the direct search for particledark matter through their scattering by nucleartargets requires ultralow background detectors ofa very low energy threshold. Moreover, the (al-most) exponentially decreasing shape of the pre-dicted nuclear recoil spectrum mimics that of thelow energy background registered by the detec-tor. All these features together make the WIMPdetection a formidable experimental challenge.

Customarily, one compares the predicted eventrate with the observed spectrum. If the formerturns out to be larger than the measured one, theparticle under consideration can be ruled out asa dark matter component. That is expressed asa contour line σ(m) in the plane of the WIMP-

2

nucleus elastic scattering cross section versus theWIMP mass which excludes, for each mass m,those particles with a cross-section above the con-tour line σ(m). The level of background limits,then, the sensitivity to exclusion.

This mere comparison of the expected signalwith the observed background spectrum is notsupposed to detect the tiny imprint left by thedark matter particle, but only to exclude or con-strain it. A convincing proof of the detection ofWIMPs would need to find unique signatures inthe data characteristic of them, like temporal (orother) asymmetries, which cannot be faked bythe background or by instrumental artifacts. Theonly distinctive signature investigated up to nowis the predicted annual modulation of the WIMPsignal rate.

The detectors used so far (most of whose re-sults have been presented to this Workshop)are Ge (IGEX, COSME, H/M) and Si (UCSB)diodes, NaI (ZARAGOZA, DAMA, UKDMC,SACLAY, ELEGANTS), Xe (DAMA, UCLA,UKDMC) and CaF2 (MILAN, OSAKA, ROMA)scintillators, Al2O3 (CRESST, ROSEBUD) andTeO2 (MIBETA, CUORICINO) bolometers andSi (CDMS) and Ge (CDMS, EDELWEISS) ther-mal hybrid detectors, which also measure the ion-ization. But new detectors and techniques are en-tering the stage. Examples of such new devices,presented at TAUP are: a liquid-gas Xenon cham-ber (UCLA); a gas chamber sensitive to the direc-tion of the nuclear recoil (DRIFT); a device whichuses superheated droplets (SIMPLE), and a col-loid of superconducting superheated grains (OR-PHEUS). There is also some new projects featur-ing a large amount of target nuclei, both with ion-ization Ge detectors (GENIUS, GEDEON) andcryogenic thermal devices (CUORE).

2. Strategies for WIMP detection

The rarity and smallness of the WIMP signaldictate the experimental strategies for its detec-tion:

Reduce first the background, controlling the ra-diopurity of the detector, components, shieldingand environment. The best radiopurity has beenobtained in the Ge experiments (IGEX, H/M,

COSME). In the case of the NaI scintillators, thebackgrounds are still one or two orders of magni-tude worse than in Ge (ELEGANTS, UKDMC,DAMA, SACLAY). The next step is to use dis-crimination mechanisms to distinguish electronrecoils (tracers of the background) from nuclearrecoils (originated by WIMPs or neutrons). Var-ious techniques have been applied for such pur-pose: a statistical pulse shape analysis (PSD)based on the different timing behaviour of bothtypes of pulses (DAMA, UKDMC, SACLAY); anidentification on an event-by-event basis of thenuclear recoils by measuring at the same timetwo different mechanisms of energy deposition,like the ionization and heat capitalizing the factthat for a given deposited energy (measured asphonons) the recoiling nucleus ionizes less thanthe electrons (CDMS, EDELWEISS).

A promising discriminating technique is thatused in the liquid-gas Xenon detector with ion-ization plus scintillation presented to this Work-shop (see D. Cline’s contribution to these Pro-ceedings). An electric field prevents recombina-tion, the charge being drifted to create a secondpulse in the addition to the primary pulse. Theamplitudes of both pulses are different for nuclearrecoils and gammas allowing their discrimination.

Another technique is to discriminate gammabackground from neutrons (and so WIMPs) usingthreshold detectors—like neutron dosimeters—which are blind to most of the low Linear En-ergy Transfer (LET) radiation (e, µ, γ). A newtype of detector (SIMPLE) which uses super-heated droplets which vaporize into bubbles bythe WIMP (or other high LET particles) energydeposition has been presented to this Workshop(see J. Collar’s contribution to these Proceed-ings).

The other obvious strategy is to make detectorsof very low energy threshold and high efficiencyto see most of the signal spectrum, not just thetail. That is the case of bolometer experiments(MIBETA, CRESST, ROSEBUD, CUORICINO,CDMS, EDELWEISS).

Finally, one should search for distinctive sig-natures of the WIMP, to prove that you are see-ing a WIMP. Suggested identifying labels are: anannual modulation of the signal rate and energy

3

spectrum (of a few percent) due to the seasonalJune-December variation in the relative velocityEarth-halo and a forward-backward asymmetryin the direction of the nuclear recoil due to theEarth motion through the halo. The annual mod-ulation signature has been already explored. Pi-oneering searches for WIMP annual modulationsignals were carried out in Canfranc, Kamiokaand Gran Sasso. At TAUP 97, the DAMA ex-periment at Gran Sasso, using a set of NaI scin-tillators reported an annual modulation effect in-terpreted (after a second run) as due to a WIMPof 60 GeV of mass and scalar cross-section on pro-tons of σp = 7 × 10−6 picobarns.

The implementation of these strategies willbe illustrated by a selection of the experimentspresented at TAUP 99. The characteristic fea-tures and main results of these experiments areoverviewed in the following paragraphs.

3. Germanium Experiments with conven-

tional detectors

The high radiopurity and low backgroundachieved in Germanium detectors, their fair lowenergy threshold, their reasonable QuenchingFactor (25%) and other nuclear merits make Ger-manium a good option to search for WIMPs withdetectors and techniques fully mastered. The firstdetectors applied to WIMP direct searches (asearly as in 1987) were, in fact, Ge diodes, asby-products of 2β-decay dedicated experiments.The exclusion plots σ(m) obtained by former Geexperiments [PNNL/USC/Zaragoza (TWIN andCOSME-1), UCSB, CALT/NEU/PSI, H/M] arestill remarkable and have not been surpassed tillrecently.

There are three germanium experiments cur-rently running for WIMP searches (COSME-2,IGEX and H/M).

COSME-2 (Zaragoza/PNNL/USC) is a small(240 g) natural abundance germanium detectorof low energy threshold (1.8–2 keV) and energyresolution of 400 eV at 10 keV. It has been under-ground for more than ten years and so is ratherclean of the cosmogenic induced activity in the8–12 keV region. It is currently taking data inCanfranc (at 2450 m.w.e.) for WIMPs and so-

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 10 20 30 40 50

Energy (keV)

coun

ts/k

eV/k

g/da

y

Heidelberg-Moscow (250 kg·d)

IGEX (60 kg·d)

IGEX threshold = 4 keVHeidelberg-Moscow threshold = 9 keV

Figure 1. IGEX and H/M spectra in the low en-ergy region.

lar axion searches (see I.G. Irastorza’s contribu-tion to these Proceedings). The background is 0.6(0.3) c/(keV kg day) averaged between the 2–15keV (15–30 keV) energy region.

IGEX is a set of enriched Ge-76 detectors look-ing for 2β decay (see D. Gonzalez’s contributionto these Proceedings) which have been recentlyupgraded for WIMP searches with energy thresh-olds of ≤ 4 keV, and energy resolution of 2 keV (at10 keV). Data from one of these detectors (RG2,2.1 kg ×30 d) show backgrounds of 0.1 c/(keV kgday) in the 10-20 keV region and 0.04 c/(keV kgday) between 20 and 40 keV. It is remarkable thatbelow 10 keV, down to the threshold of 4 keV thebackground is ∼ 0.3 c/(keV kg day) mainly fromnoise which is being removed. The spectrum isshown in Fig. 1. IGEX is also operating two otherGe detectors in Baksan [one natural—TWIN—and other enriched in Ge-76 (RV)] of about 1 kgeach, and thresholds of 2 and 6 keV respectively.After a subtraction procedure a background of∼ 0.1 c/(keV kg day) between 10 and 30 keV wasobtained.

The Heidelberg/Moscow Ge experiment on 2β

decay in Gran Sasso is also using data of oneof their enriched Ge-76 detectors (2.7 kg, andthreshold of 9–10 keV) in a search for WIMPs.The background is similar to that of IGEX (0.16c/(keV kg day) from 10-15 keV and 0.05 c/(keV

4

kg day) from 15-30 keV), although its thresholdis more than a factor two higher. The low-energyspectrum is also shown comparatively to that ofIGEX in Fig. 1.

The exclusion plots obtained from the spectraof Germanium detectors are shown in Fig. 2. TheGe-combined limit is the contour obtained fromthe envelope of all of them—including the lastH/M data and is compared with that derivedfrom the last COSME, IGEX and CDMS datapresented at this Workshop. Also the most strin-gent NaI exclusion plot is shown together withthe (σ, m) region where a seasonal modulation ef-fect in the recorded rate has been reported by theDAMA Collaboration and attributed to a WIMPsignal. The exclusions depicted in this paper referto spin-independent interactions. The sensitivityof the present detectors does not yet reach therates needed to explore spin-dependent couplings.For comparison among different experiments, thecoherent spin independent WIMP-nucleus cross-section is normalized to that of WIMP on nucle-ons. All the Ge and NaI exclusion plots shownhave been recalculated from the original spectraby the author and his collaborators I.G. Iras-torza and S. Scopel, with the same set of pa-rameters. The values used for the halo modelare ρ = 0.3 GeV cm−3, vrms = 270Kms−1 andvesc = 650Kms−1.

There exist some projects with Ge detectors indifferent degree of development: HDMS (Heidel-berg DM Search) is a small (200 g) Ge detec-tor placed in a well-type large (2 kg) Ge crys-tal. The goal is to reach a background of 10−2

c/(keV kg day). GEDEON (Germanium Diodesin One Cryostat, Zaragoza / USC / PNNL) willuse a single cryostat of IGEX technology hostinga set of natural Ge crystals of total mass of 28kg. The threshold of each small detector is < 2keV and the background goal—expected from themeasured radioimpurities—is 10−2−10−3 c/(keVkg day). A set of three cryostats (∼ 80 kg of Ger-manium) is the planned final configuration which,embedded in Roman lead and graphite, will beinstalled in Canfranc. GENIUS (Germanium De-tectors in Liquid Nitrogen in an Underground Set-up) plans to operate 40 natural abundance, nakedgermanium crystals (of 2.5 kg each) submerged

1 5 10 50 100 500 1000MW +GeV/

10�9

10�8

10�7

10�6

V+ n/ +

nrabn/

Figure 2. Combined exclusion plots obtainedfrom previous Ge experiments (thin dashedline) compared with those presented to thisWorkshop—CDMS (dot-dashed), COSME (thickdashed), IGEX (thick solid)—and with theDAMA results (thin solid). Prospects for IGEX(dot-dot-dashed) and CUORICINO (dots) arealso shown

directly in a tank of liquid nitrogen.

4. Sodium Iodine experiments

The full isotopic content on A-odd isotopes(Na-23, I-127) makes sodium iodine detectors sen-sitive also to spin-dependent WIMP interactions.The main recent interest in scintillators is dueto the fact that large masses of NaI crystalsfor exploring the annual modulation are afford-able. There exist four NaI experiments running(UKDMC, DAMA, SACLAY and ELEGANTSV) and two in preparation (ANAIS and NAIAD).

The NaI experiments serve to illustrate one ofthe strategies for background discrimination men-tioned above. The time shape differences be-tween electron recoils and nuclear recoils pulsesin NaI scintillators can be used to discrimi-nate gamma background from WIMPs (and neu-trons) because of the shorter time constants ofnuclear versus electron recoils. Templates ofreference pulses produced by neutron, gamma

5

(and X, β...) sources are compared with thedata population pulses (in each energy band)by means of various parameters [time constants(UKDMC), momenta of the time distribution(DAMA, SACLAY), integrated time profile dif-ferences (SACLAY)]. From this comparison, thefraction of data which could be due to nuclearrecoils turns out to be only a few percent, de-pending on energy, of the measured background[1 to 3 c/(keV kg day) in DAMA and UKDMC,and of 2 to 10 c/(keV kg day) in SACLAY],with different degree of success, depending of theexperiment and (slightly) on the method used.Due to the drastic background reduction, the ex-clusion plots obtained from the stripped spectrahave surpassed (DAMA, UKDMC) that derivedfrom the (non-manipulated) spectra of the Gedetectors—whose radiopurity is much better thanthat of NaI. Let us briefly mention the main per-formances of these experiments.

The United Kingdom Dark Matter Collabora-tion (UKDMC) uses radiopure NaI crystals ofvarious masses (2 to 10 kg) in various shieldingconditions (water, lead, copper) in Boulby. Typ-ical thresholds of 4 keV and background (beforePSD) of 2–4 c/(keV kg day) (at about thresh-old) have been obtained. Recent results fromNaI crystals of 5 and 2 kg show a small popu-lation of pulses (Fig. 3) of an average time con-stant shorter than that of gamma events and nearto that corresponding to neutron-induced recoils,which is not due to instrumental artifact. (Forrecent results, see I. Liubarsky’s contribution tothese Proceedings). Plans of the UKDMC includeNAIAD (NaI Advanced Detector) consisting of50–100 kg in a set of unencapsulated crystals toavoid surface problems and improve light collec-tion.

The SACLAY group is carrying out a thoroughprogram of investigation about the virtues andlimitations of the pulse shape analysis as far asthe statistical background discrimination is con-cerned. They use, at LSM Frejus, a radiopure 9.7kg NaI crystal with an energy threshold of 2 keVand backgrounds of (before PSD) 8–10 c/(keV kgday) (at 2–3 keV) and of ∼ 2 c/(keV kg day)(and flat) above 5 keV. The high backgroundat threshold, not well understood, has spoiled

Figure 3. Pulse shapes for gammas, alphas, neu-trons and the anomalous events from a 2-kg NaIcrystal (UKDMC)

the exclusion plots of this experiment, comparedwith other NaI searches. On the other hand,their data—as it happened in that of UKDMC—cannot be sharply split into Compton plus nuclearrecoils showing a spurious population (Fig. 4), afact that limits the sensitivity of the PSA theycould perform. Including this peculiar situation,as systematic effects, their PSA background re-duction is only 65% to 85%.

The DAMA experiment uses also NaI crys-tals of 9.7 kg with energy threshold of ∼ 2keV. No spurious population is found in DAMAwhich could spoil the PSA separation back-ground/nuclear recoils. In fact, their backgroundreduction reaches levels of 85% (4–6 keV) and97% (12–20 keV), providing exclusion plots whichhave surpassed that of germanium.

In conclusion, besides the significant reduc-tion of background provided by this statisticalmethod, a most intriguing result, as stated above,is that UKDMC and SACLAY, applying thesePSD techniques to their data, have found thatthey are not compatible with a contribution ofonly Compton nor with nuclear recoil events, andsuggest the existence of an unknown populationor systematic effects. It is also an intriguing coin-cidence that the energy spectrum of that residual

6

Figure 4. < t > distribution for the 10–15 keVbin showing unknown population “U” responsi-ble for shorter decay times in the time profiles(SACLAY)

populations are very similar in both experiments(Fig. 5). (See G. Gerbier’s contribution to theseProceedings).

5. WIMPs Annual Modulation as a dis-

tinctive signature

The Earth rotation around the sun combinedwith the sun velocity through the halo makesthe velocity of the detector with respect to theWIMPs halo a yearly periodic function peakedin June (maximum) and December (minimum).Such seasonal variation should provide an annualmodulation in the WIMP interaction rates andin the deposited energy as a clear signature ofthe WIMP. However, the predicted modulationis only a few percent of the unmodulated, aver-age signal. To reveal this rate modulation, largedetector masses/exposures are needed.

Such type of seasonal modulation in the spec-tra has been reported in the DAMA experimentand attributed by the Collaboration to a WIMPsignal (see P.L. Belli’s contribution to these Pro-ceedings).

Preceding the DAMA experiment, there havebeen various attempts to search for annual

Figure 5. Energy spectra of the unknown pop-ulation of events found in NaI experiments fromUKDMC and SACLAY

modulation of WIMP signals, starting as earlyas in 1991. COSME-1 and NaI-32 (Can-franc) (with 2 years statistics), DAMA-Xe (GranSasso), ELEGANTS-V (Kamioka), and more re-cently DEMOS (Sierra Grande) (with 3 years ofstatistics), are examples of seasonal modulationsearches with null results. However, these experi-ments produced results which improved the σ(m)exclusion plots and settled the conditions and pa-rameters for new, more sensitive searches.

The DAMA experiment uses a set of 9 radiop-ure NaI crystals of 9.7 kg each, viewed by twolow-background PMT at each side through lightguides (10 cm long), coupled to the crystal. Spe-cial care has been taken in controlling the stabil-ity of the main experimental parameters. A noiseremoval is done by using the timing behaviour ofnoise and true NaI pulses (with 40% efficiency atlow energy). The background after noise removalis, averaging over the detectors: B (2–3 keV) ∼

1–0.5 c/keV kg day and B (3–20 keV) ∼ 2 c/keVkg day. Notice the drop in the first two chan-nels, precisely where the expected signal is moresignificant and, consequently, essential for deriv-ing the reported modulation effect. (The PulseShape Analysis is not used in the annual modula-tion search). The multidetector energy spectrumof single hit events (each detector has twelve de-

7

tectors as active veto) has been reported to thisWorkshop (Fig. 6) (see P.L. Belli’s contributionto these Proceedings).

Figure 6. Energy spectrum recorded with themultidetector set-up of the DAMA experiment(single hit events)

At the time of TAUP 99, DAMA had issuedthe results of two runnings. Run 1 reported atTAUP 97, extended over 1 month in winter and2 weeks in summer, i.e. a total of 4549 kg day.Run 2, which used a slightly different setup, ex-tended from November to July, (one detector for90 days and eight detectors for 180 days) i.e. atotal of 14962 kg day. Both DAMA 1 and 2results are compatible and consistent with eachother. A likelihood method applied to the totalstatistics of 19511 kg day provides a minimumfor: m =

(

59+17−14

)

GeV, σp =(

7.0+17−14

)

× 10−6

pb as the most likely values of the mass and in-teraction cross-section on proton of an “annualoscillating” WIMP (at 99.6% C.L.). Other sta-tistical approaches essentially agree with the like-lihood result. Fig. 7 shows the so-called DAMAregion of the positive modulation effect and thescattered plot of the MSSM prediction of σp infunction of m. The time evolution of the DAMAsignal has been recently issued by the collabora-tion indicating an oscillating trend in spite of thesmall exposure time and the discontinuity of thetwo runnings (Fig. 8). The DAMA results havearoused great interest but also critical comments

related to various aspects of the experiment. Oneof the most frequently heard has to do with thedrop of background in the first energy bins or tothe peculiar look of the residual background aftersubtracting the likelihood-emerging signal. Obvi-ously the delivery of new data is eagerly awaited.

Figure 7. Region (σ, m) of the positive modula-tion effect found by DAMA.

Independently of other considerations and be-yond any controversies whatsoever, it is imper-ative to confirm the DAMA results by other in-dependent experiments with NaI (like ANAIS orELEGANTS) and with other nuclear target. TheZARAGOZA group is preparing ANAIS (AnnualModulation with NaIs), consisting of 10 NaI scin-tillators of 10.7 kg each, stored for more than tenyears in Canfranc, and recently upgraded for DMsearches. It will be placed in Canfranc within ashielding of electroformed copper and a large boxin Roman lead, plus a neutron screen and an ac-tive veto. The tests of a smaller set are underway.Expected performances are an energy thresholdof 2 keV and a background at threshold of ∼ 2–3c/(keV kg day).

8

Figure 8. Time evolution of DAMA signal

The OSAKA group is performing a search withthe ELEGANTS V NaI detector in the new un-derground facility of Otho, with a huge mass ofNaI scintillators (760 kg) upgraded from previ-ous experiments. A search for annual modula-tion with null result has been presented to thisWorkshop (see S. Yoshida’s contribution to theseProceedings).

The DAMA σ(m) region should also be ex-plored by the standard method of comparing the-ory with the total time-integrated experimentalspectrum (without enquiring about possible vari-ations in time). In fact, various experimentsare reaching the DAMA region (below ∼ σp ∼

10−5 − 10−6 pb for WIMPs of 40–80 GeV) whichis itself half-excluded by a previous DAMA-0 run-ning data using PSA discrimination. For in-stance, CDMS has reached the upper left cor-ner and exclude it (see Fig. 2 and R. Gaitskell’scontribution to these Proceedings), whereas theIGEX and H/M germanium experiments are veryclose to it (with direct, non-stripped data).

Due to what is at stake, it is important to knowwhat the prospects of WIMP detection are troughthe annual modulation signature for planning theright experiments. In fact, to find an unam-biguous, reproducible and statistically significantmodulation effect is, by now, the best identifyinglabel of a WIMP. Sensitivity plots for modula-tion searches (presented to this Workshop) give

the MT exposure needed to explore σ,m regionsusing the annual modulation signature or, equiv-alently, needed to detect an effect (should it ex-ist) at a given C.L., due to a WIMP with a massm and cross-section σp. Examples of sensitivityplots for Ge, NaI and TeO2—and the ensuing ca-pability to explore σp,m regions, are given to illus-trate the modulation research potential of somedetectors (see S. Scopel’s contribution to theseProceedings) running or in preparation.

6. Cryogenic Particle Detectors

In the WIMP scattering on matter, only a smallfraction of the energy delivered by the WIMPgoes to ionization, the main part being releasedas heat. Consequently, thermal detectors shouldbe suitable devices for dark matter searches withquenching factors of about unity and low effectiveenergy threshold (Evis ∼ T) as WIMPs searchesrequire. Moreover, bolometers which also collectcharge (or light) can simultaneously measure thephonon and ionization (or scintillation) compo-nents of the energy deposition providing a uniquetool of background subtraction and particle iden-tification.

There exist five experiments searching for di-rect interactions of WIMPs with nuclei based onthermal detection currently running (MIBETA,CDMS, EDELWEISS, CRESST, and ROSE-BUD) another one, CUORICINO, being mountedand a big project, CUORE, in preparation.

The CRESST (Cryogenic Rare Event Searchwith Superconducting Thermometers) (MPI Mu-nich / TUM Garching / Oxford, Gran Sasso) de-tectors are four sapphire crystals (Al2O3) of 262 geach with a tungsten superconducting transitionedge sensor. The energy resolution and thresh-old obtained with an x-ray fluorescence sourceare respectively 133 eV (at 1.5 keV) and 500 eV.The background obtained in the Gran Sasso run-ning is of about 10–15 counts/(keV kg day) above30 keV, going down to 1 c/(keV kg day) above100 keV, whereas below 30 keV the spectrum islargely dominated by noise and other spurioussources preventing to derive exclusion plots.

Recently the collaboration has performed si-multaneous measurements of scintillation and

9

heat, with a 6 g CaWO4 crystal as absorber. Thepreliminary results indicate a rejection of electronrecoil events with an efficiency greater than 99.7%for nuclear recoil energies above 15 keV. Shortterm prospects for CRESST are the implementa-tion of the scintillation-phonon discrimination ofnuclear recoils in a CaWO4 detector of 1 kg (seeJ. Jochum’s contribution to these Proceedings).

ROSEBUD (Rare Objects Search with Bolome-ters Underground) [University of Zaragoza andIAS (Orsay)] is another sapphire bolometer ex-periment to explore the low energy (300 eV–10 keV) nuclear recoils produced by low massWIMPs. It is currently running in Canfranc (at2450 m.w.e.). It consists of two 25g and one50g selected sapphire bolometers (with NTD (Ge)thermistors) operating inside a small dilution re-frigerator at 20 mK. One of the 25g sapphire crys-tals is part of a composite bolometer (2 g of LiFenriched at 96% in 6Li glued to it) to monitor theneutron background of the laboratory. The inner(cold) shielding and the external one are madeof archaeological lead of very low contamination.The experimental setup is installed within a Fara-day cage and an acoustic isolation cabin, sup-ported by an antivibration platform. Power sup-ply inside the cabin is provided by batteries anddata transmission from the cabin through con-venient filters is based on optical fibers. Infrared(IR) pulses are periodically sent to the bolometersthrough optical fibers in order to monitor the sta-bility of the experiment. Pumps have vibration-decoupled connections.

The first tests in Canfranc have shown that mi-crophonic and electronic noise level is quite good,about 2nV/Hz1/2 below 50Hz. The bolometerswere tested previously in Paris (IAS) showinga threshold of 300 eV and energy resolution of120 (at 1.5 keV). Typical sensitivities obtained(in Canfranc) are in the range of 0.3–1 µV/keV.Overall resolutions of 3.2 and 6.5 keV FWHMwere typically obtained in Canfranc with the 50g and 25 g bolometers, respectively, at 122 keV.Low energy background pulses corresponding toenergies below 5 keV are seen. In the test run-nings, the background obtained was as large as120 counts/keV/kg/day around 40–80 keV. Aftervarious modifications in the cryostat components,

the background level of the 50 g bolometer standsabout 15 counts/keV/kg/day from 20 to 80 keV.This progressive reduction is illustrated in Fig.9. Measurements of the radiopurity of individ-ual components continue with an ultralow back-ground Ge at Canfranc and their removal donewhen needed with the purpose of lowering thebackground one more order of magnitude. Thenext step of the ROSEBUD program will dealwith bolometers of sapphire and germanium, op-erating together to investigate the target depen-dence of the WIMP rate (see P. de Marcillac’scontribution to these Proceedings).

Figure 9. ROSEBUD background spectrum

The Cryogenic Dark Matter Search Collabo-ration (CDMS) (CfPA / UC Berkeley / LLNL/ UCSB / Stanford / LBNL / Baksan / SantaClara / Case Western / Fermilab / San Fran-cisco State) has developed bolometers which col-lect also electron-holes carriers for discriminat-ing nuclear recoils from electron recoils. Theelectron-hole pairs are efficiently collected in thebulk of the detector, but the trapping sites nearthe detector surface produce a layer (10− 20µm)of poor charge collection, where surface electronsfrom outside suffer ionization losses and fake nu-clear recoils.

Two types of phonon readout have been de-veloped. In the BLIP (Berkeley Large Ioniza-tion and Phonon) detector, a NTD Ge thermistorreads the thermal phonons in milliseconds. In the

10

FLIP (Fast Large Ionization and Phonon) detec-tors, non-equilibrium, athermal phonons are de-tected (microsecond time scale) with supercon-ducting transition edge thermometers in tung-sten. Current prototypes are BLIPs of 165 g Ger-manium and FLIPs of 100 g Silicon. The FWHMenergy resolutions are 900 eV and 450 eV respec-tively in the phonon and charge channels in BLIPdetectors and about 1 keV in the FLIPs. Theelectron nuclear recoil rejection in both detectorsis larger than 99% above 20 keV recoil energy.Backgrounds below 0.1 c/(keV kg day) in the 10–20 keV energy region have been obtained in recentruns.

Following new developments achieved in the de-tectors, the surface events have been successfullydiscriminated using their phonon rise time: thelow-charge collection events (surface electrons)have been proved to have faster phonon rise timethan the bulk events. A rise time cut is applied toget rid of them. Results from a recent run are de-picted in Fig. 10. In spite of the small masses andshort time runs, CDMS exclusion plots are com-petitive with much larger exposures of other de-tectors. In fact, CDMS is now probing the DAMAregion, as reported to this Workshop (see Fig. 2and R. Gaitskell’s contribution to these Proceed-ings). Projects of the CDMS Collaboration in-clude the transfer of the FLIP technology to ger-manium crystals of 250 g. Planned exposure atStanford (only 17 m of overburden) is of 100 kgday, with a background goal of B = 0.01 c/keVkg day. The experiment will be moved to Soudanalong the year 2000 with twenty FLIP detectorsof Germanium (250 g each) and the backgroundgoal of B = few × 10−4 c/keV kg day.

EDELWEISS (Orsay/Lyon/Saclay/LSM/IAP)has operated two 70 g HP Ge bolometer in theFrejus tunnel with heat-ionization discriminationgetting similar results to that of the BLIP detec-tors of CDMS and so they will not be repeatedhere. The background obtained is B = 0.6 c/keVkg day in the 12 − 70 keV region of the recoilenergy spectrum. The rejection is 98% for sur-face events and > 99.7% for internal events. (Formore details, see G. Chardin contribution to theseProceedings).

The collaboration is preparing a small tower of

Figure 10. CDMS low energy spectrum obtainedin a recent run

three Ge bolometers of 70 g each, to be enlargedto other three of 320 g each. The background goalis to get B = 10−2 c/keV kg day which seemsto be at hand. A second phase of EDELWEISS(2000–2001) will use a reverse dilution refrigera-tor of 100 liters now under construction to host50 − 100 detectors. Twenty Ge detectors of 300g will be placed in the next two years, expectingto improve the rejection up to 99.99% and get abackground of 10−4 c/(keV kg day).

CUORE (Cryogenic Underground Observatoryfor Rare Events) (Berkeley / Florence / LNGS/ Leiden / Milan / Neuchatel/ South Carolina /Zaragoza) is a project to construct a large mass(775 kg) modular detector consisting of 1020 sin-gle bolometers of TeO2 of dimensions 5×5×5 cm3

and 760 g each, with glued NTD Ge thermistors,to be operated at 7 mK in the Gran Sasso Lab-oratory. A tower of 14 planes consisting of 56 ofthose crystals with a total mass of 42 kg, the so-called CUORICINO detector, will be a first stepin the CUORE project.

Preliminary results of a 20 crystal array of tel-lurite bolometers (340 g each) (MIBETA experi-

11

ment) optimized for 2β decay searches show en-ergy thresholds ranging from 2 to 8 keV (depend-ing on the detector) and background levels of afew counts per keV kg day in the 15–40 keV lowenergy region. CUORICINO is planned as anextension of the MIBETA setup featuring moreand larger crystals. The sum of the 20 contempo-rary calibration spectra with a single 232Th sourceshows that the array is indeed acting as a singledetector.

Four bolometers of the future CUORICINOarray have been recently tested in Gran Sasso.The results on the energy resolution in the re-gion of neutrinoless double beta decay of 130Te(2500 keV) are about 5 ∼ 8 keV (see M. Pavan’scontribution to these Proceedings). Other val-ues obtained are ∼ 2 keV at 46 keV and 4.2 keVat 5400 keV. Energy resolutions of 1-2 keV andbackgrounds of 10−2 c/(keV kg day) in the fewkeV region can be expected.

Fig. 11 shows the exclusion contour obtainedfrom running experiments (Ge, NaI), and the pro-jections for GEDEON, CUORICINO and CDMSassuming the parameter values expected in suchexperiments.

7. Where we stand and where we go

Unrevealing the nature of the dark matter isof uttermost importance in Cosmology, Astro-physics and Particle Physics. It has triggered alarge experimental activity in searching for all itspossible forms, either conventional or exotic. Inparticular, there exist various large microlensingsurveys looking for dark baryons (EROS, MA-CHO, OGLE...) and a variety of observationssearching for the baryonic component of darkmatter. (See J. Uson’s contribution to these Pro-ceedings).

As far as the exotic, non-baryonic objectsare concerned, a few experiments are looking(or project to look) for axions (RBF/UF, LIV-ERMORE, KYOTO, CRYSTALS, CERN SolarAxion Telescope Antenna...), as reviewed in P.Sikivie’s contribution to these Proceedings.

In the WIMP sector (see L. Roszkowski’s con-tribution to these Proceedings) to which thisexperimental overview is dedicated, there are

1 5 10 50 100 500 1000MW +GeV/

10�12

10�10

10�8

10�6

10�4

V+n/ +

nrabn/

Figure 11. Projection of exclusion plots expectedfor CUORICINO (thick dashed), GEDEON(thick solid), CDMS Stanford (thick dot-dashed)and CDMS Soudan (dots). Ge-combined (thindashed) and DAMA (thin solid) results are alsoshown as well as the MSSM region (thin dot-dashed)

various large underground detector experiments(MACRO, BAKSAN, SOUDAN, SUPER-K...)and deep underwater (ice) neutrino telescopes(AMANDA, BAYKAL, ANTARES, NESTOR...)looking for (or planning to look for) neutrino sig-nals originated by the annihilation of WIMPs,as well as some balloons and satellite experi-ments looking for antimatter of WIMP origin,most of them included in these Proceedings.About thirty experiments either running or be-ing prepared are looking for WIMPs by the di-rect way (COSME, IGEX, HEIDELBERG/MOSCOW, ELEGANTS-V and VI, DAMA, SACLAY,UKDMC, ANAIS, TWO-PHASE Xe, LqXe,CASPAR, SIMPLE, MICA, DRIFT, CRESST,ROSEBUD, MIBETA, CUORICINO, CDMS,EDELWEISS, ORPHEUS...), with conventionalas well as with cryogenic techniques... andsome large projects with 100 to 1000 detectors(CUORE, GEDEON, GENIUS...) are being initi-ated. Their current achievements and the projec-tions of some of them have been shown in terms ofexclusion plots σp (m), which illustrate the poten-

12

tial to investigate the possible existence of WIMPdark matter in regions pretty close to where thesupersymmetric candidates must appear.

After witnessing the large activity and progressreported to this Workshop, it is clear thatthe main strategies recommended to search forWIMPs have proved to be quite efficient to reducethe window of the possible particle dark matterand to approach the zone of the more appeal-ing candidates and couplings. Examples of theachievements in radiopurity, background identifi-cation or rejection, in low (effective) threshold en-ergy and efficiency, as well as in investigating thegenuine signatures of WIMPs, like modulationand directionality, have been largely reported toTAUP 99 and reviewed selectively in this paper.The conclusion is that these strategies are well fo-cused and should be further pursued. Finally, anannual modulation effect—supposedly producedby a WIMP—is there, alive since the last TAUP97, waiting to be confirmed by independent ex-periments.

Acknowledgements

I wish to thank the spokespersons of the exper-iments presented at TAUP 99 for making avail-able to me in advance their contributions as wellas other useful information about the status andplans of their experiments. I am indebted to mycollaborators I.G. Irastorza and S. Scopel for theircontribution to the making of the exclusion plots.The kindness of my collaborators of COSME,IGEX and ROSEBUD for allowing me to use thedata from these experiments is warmly acknowl-edged. Thanks are due also to my CUORICINOcolleagues for their permission to use internal in-formation on the status and preliminary resultsof the experiment. Finally, I thank MercedesFatas for her patience and skill in the composi-tion of the text. The financial support of CICYT(Spain) under grant AEN99-1033 and the Euro-pean Commission (DGXII) under contract ERB-FMRX-CT-98-0167 is duly acknowledged.