Particle dark matter: From DAMA/NaI to DAMA/LIBRA

ISSN 1063-7788, Physics of Atomic Nuclei, 2006, Vol. 69, No. 12, pp. 2056–2067. c© Pleiades Publishing, Inc., 2006.

Proceedings of the 5th International Conferenceon NONACCELERATOR NEW PHYSICS

Dark Matter

Particle Dark Matter: From DAMA/NaI to DAMA/LIBRA*

R. Bernabei1), P. Belli1)**, F. Montecchia1), F. Nozzoli1), F. Cappella2),A. d’Angelo2), A. Incicchitti2), D. Prosperi2), R. Cerulli3),

C. J. Dai4), H. L. He4), H. H. Kuang4), J. M. Ma4), and Z. P. Ye4)

Received October 13, 2005

Abstract—DAMA is an observatory for rare processes and it is operative deep underground at the GranSasso National Laboratory of the INFN. Several low background setups have been realized and manyrare processes have been investigated. In particular, the DAMA/NaI setup (100 kg of highly radiopureNaI(Tl)) has effectively investigated the model-independent annual modulation signature over seven annualcycles (total exposure of 107 731 kg day), obtaining 6.3σ C.L. model-independent evidence for the presenceof a dark matter particle component in the galactic halo. Some of the many possible corollary model-dependent quests for the candidate particle have been investigated and others are in progress. At present,the second generation DAMA/LIBRA setup (250 kg of highly radiopure NaI(Tl)) is in data taking deepunderground.

PACS numbers : 95.35.+d, 29.40.McDOI: 10.1134/S1063778806120088

1. INTRODUCTION

DAMA is an observatory for rare processes basedon the development and use of various kinds of ra-diopure scintillators. Several low background setupshave been realized, in particular, (i) DAMA/NaI(100 kg of highly radiopure NaI(Tl)), which tookdata underground over seven annual cycles andwas put out of operation in July 2002 [1–14]; (ii)DAMA/LXe (6.5 kg of liquid xenon) [15]; (iii)DAMA/R&D, which is devoted to tests on pro-totypes and small scale experiments [16]; and (iv)the new second generation DAMA/LIBRA setup(250 kg of highly radiopure NaI(Tl)) in operationsince March 2003. Moreover, in the framework ofdevoted R&D for radiopure detectors and PMTs,sample measurements are carried out by means of thelow background DAMA/Ge detector, installed deepunderground for more than 10 years, and, in somecases, by means of Ispra facilities.

∗The text was submitted by the authors in English.1)Dip. di Fisica, Universita’ di Roma Tor Vergata and INFN,

sez. Roma2, Rome, Italy.2)Dip. di Fisica, Universita’ di Roma La Sapienza and INFN,

sez. Roma, Rome, Italy.3)INFN, Laboratori Nazionali del Gran Sasso, Assergi (AQ),

Italy.4)IHEP, Chinese Academy, Beijing, China**E-mail: [email protected]

In the following, we will focus our attention onlyon the results obtained by the DAMA/NaI exper-iment exploiting the annual modulation signature:a very distinctive model-independent signature forthe investigation of the presence of a particle darkmatter (DM) component in the galactic halo. Thissignature (originally suggested in [17]) requires thesimultaneous satisfaction of all the following re-quirements: (1) the rate must contain a compo-nent modulated according to a cosine function with(2) one-year period T and (3) a phase t0 aroundJune 2; (4) this modulation must only be foundin a well-defined low-energy range, where a DMparticle can induce signals; (5) it must apply tothose events in which just one detector of manyactually “fires” (single-hit events), since the DMparticle multiscattering probability is negligible; (6)the modulation amplitude in the region of maximalsensitivity is expected to be 7% for usually adoptedhalo distributions, but it can be larger in the case ofsome possible scenarios such as, e.g., those in [18,19]. To mimic such a signature, either spurious effectsor side reactions should be able not only to account forthe whole observed modulation amplitude but also tocontemporaneously satisfy all the requirements; nonehas been found [2, 3].

In the following, we will first present themodel-independent result obtained by DAMA/NaIand then the corollary interpretations of this resultin terms of some of the many possible DM particle

2056

PARTICLE DARK MATTER 2057

candidates in given model frameworks. Further sce-narios are also under investigation considering othercandidates and other kinds of interactions.

Finally, in the last part of the paper, some of thefuture prospects of the new DAMA/LIBRA setupwill be briefly summarized. For detailed discussions,see [2, 3] and references therein. For a descrip-tion of the DAMA/NaI setup and its performance,see [1–3, 10].

2. THE MODEL-INDEPENDENT RESULTOF DAMA/NaI

A model-independent approach to the data col-lected by DAMA/NaI over seven annual cycles offersimmediate evidence of the presence of annual mod-ulation of the measured rate of single-hit events inthe lowest energy region. In particular, in Fig. 1 (left)is reported the experimental residual rate of single-hit events in the 2–6-keV energy interval, given by〈rijk − flatjk〉jk, where rijk is the rate in the con-sidered ith time interval for the jth detector in thekth considered energy bin and flatjk is the rate ofthe jth detector in the kth energy bin averaged overthe cycles. The average is made for all the detectors(j index) and for all the energy bins in the consid-ered energy interval. The data favor the presence ofa modulated cosine-like behavior (A cos[ω(t − t0)])at 6.3σ C.L. with a modulation amplitude equal to0.0200 ± 0.0032 cpd/kg/keV, t0 = 140 ± 22 day, andT = 2π

ω = 1.00± 0.01 yr, when fitting the experimen-tal points keeping all parameters free. The period andphase agree with those expected in the case of aneffect induced by DM particles in the galactic halo(T = 1 yr and t0 roughly at 152.5th day of the year).The hypothesis of unmodulated behavior can be ex-cluded, being the χ2/d.o.f. = 71/37 on the 2–6 keVresidual rate, which corresponds to a probability of7 × 10−4. The same data have also been investigatedby a Fourier analysis considering the treatment ofthe experimental errors and of the time binning. Theresult is shown in Fig. 1 (right), where a clear peakcorresponding to a period of 1 yr is present.

Modulation is not observed above 6 keV [2, 3].We recall that DAMA/NaI took data up to the MeVenergy region despite the fact that the optimizationwas done for the keV energy range. Finally, a suitablestatistical analysis has shown that the modulationamplitudes are statistically well distributed in all thecrystals, in all the data taking periods and consideredenergy bins. For a complete discussion, see [2, 3]. Acareful investigation of all the known possible sourcesof systematic and side reactions has been regularlycarried out and published at the time of each datarelease and quantitative discussions can be found

Table 1. Summary of the results obtained by investigatingthe possible sources of systematics or of side reactions [2,3] (No systematics or side reaction has been found ableto give a modulation amplitude different from zero; thus,very cautious upper limits (90% C.L.) have been calcu-lated and are shown here in terms of the measured model-independent modulation amplitude, Sobs

m . As can be seen,they cannot account for the measured modulation ampli-tude and, in addition, cannot satisfy the peculiar require-ments of the signature [2, 3, 10])

Source Cautious upper limit (90% C.L.)

Radon <0.2% Sobsm

Temperature <0.5% Sobsm

Noise <1% Sobsm

Energy scale <1% Sobsm

Efficiencies <1% Sobsm

Background <0.5% Sobsm

Side reactions <0.3% Sobsm

Note: In addition, no effect can mimic the signature.

in [2, 3, 10]. No systematic effect or side reaction ableto account for the observed modulation amplitude andto satisfy all the requirements of the signature hasbeen found. Thus, very cautious upper limits (90%C.L.) on the possible contributions to the modulatedamplitude have been calculated as summarized in Ta-ble 1; for detailed and quantitative discussions, see [2,3, 10].

A further relevant investigation on the multi-ple-hit events collected during the DAMA/NaI-6and seven running periods has been performed. Inthese two running periods, in fact, each detectorwas equipped with its own transient digitizer with adedicated renewed electronics and an analysis usingthe same identical hardware and the same identicalsoftware procedures as for the case of single-hitevents (Fig. 2) was feasible.

The multiple-hit event class—in contrast tothe single-hit one—does not include events in-duced by DM particles since the probability thata DM particle scatters off more than one detectoris negligible. The fitted modulation amplitudes areA = 0.0195 ± 0.0031 cpd/kg/keV and A = −(3.9 ±7.9)× 10−4 cpd/kg/keV for single-hit and multiple-hit residual rates, respectively. Thus, evidence ofannual modulation is present in the single-hit resid-uals (event class induced by DM particle), while itis absent in the multiple-hit residual rate (eventclass due only to background events). Since the sameidentical hardware and the same identical software

PHYSICS OF ATOMIC NUCLEI Vol. 69 No. 12 2006

2058 BERNABEI et al.

500

–0.05

1000 1500 2000 2500Time, day

–0.10

0

0.05

0.10Residuals, cpd/kg/keV 2–6 keV

I II III IV V VI VII

0 0.002 0.004 0.006 0.008Frequency, day

–1

2

4

6

8

10Normalized power

Fig. 1. (Left) Experimental residual rate for single-hit events in the cumulative 2–6 keV energy interval as a function of timeover seven annual cycles (total exposure 107 731 kg day). The experimental points present the errors as vertical bars and theassociated time bin width as horizontal bars. The superimposed curve represents the cosinusoidal function behavior expectedfor a DM particle signal with a period equal to 1 yr and phase exactly at June 2; the modulation amplitude has been obtainedby best fit. See [2, 3]. (Right) Power spectrum of the measured single-hit residuals for the cumulative 2–6 keV energy interval(see text). As can be seen, the principal mode corresponds to a frequency of 2.737 × 10−3 day−1, that is, to a period of 1 yr.

300–0.05

400 500 600Time, day

0

0.05Residuals, cpd/kg/keV

Fig. 2. Experimental residual rates () over seven annualcycles for single-hit events (class of events to whichDM particle events belong) and () over the last twoannual cycles for multiple-hit events (class of events towhich DM particle events do not belong) in the 2–6 keVcumulative energy interval. They have been obtained byconsidering for each class of events the data as collectedin a single annual cycle and using in both cases thesame identical hardware and the same identical softwareprocedures. The initial time is taken on August 7. Seetext.

procedures have been used for the two classes ofevents, the obtained result offers additional strongsupport for the presence of DM particles in thegalactic halo, further excluding any side effect eitherfrom hardware or from software procedures or frombackground.

In addition, we mention the investigation on thebackground behavior. As reported above, no modula-tion has been observed in the background; in fact, thewhole energy spectrum up to the MeV energy regionhas been analyzed and the presence of a backgroundmodulation in the whole energy spectrum has beenexcluded at a level much lower than the effect found

in the lowest energy region [2, 10]. This result alreadyaccounts also for the background component due tothe neutrons and to the radon; nevertheless, furtheradditional independent and cautious analyses to es-timate their possible contribution have been givenin [2, 10]. Moreover, in no case can the neutronsmimic the signature since some of the peculiar re-quirements of the signature would fail. Similarly, anypossible effect of the muon flux modulation reportedby the MACRO experiment [20] is excluded both bya quantitative investigation [2, 10] and by an inabilityto fulfill all the peculiarities of the signature.

In conclusion, the data of the seven annual cyclessupport at 6.3σ C.L. the presence of an annual mod-ulation in the residual rate of the single-hit eventsin the lowest energy interval 2–6 keV, satisfying allthe features expected for a DM particle component inthe galactic halo. In addition, no systematic effect orside reaction able to account for the observed effecthas been found. This is the experimental result ofDAMA/NaI; it is model-independent. No other ex-periment whose result can be directly compared withthis one in a model-independent way is available sofar in the field of DM investigation.

3. SOME COROLLARYMODEL-DEPENDENT QUESTS

FOR A CANDIDATE

On the basis of the 6.3σ model-independent evi-dence for the presence of a DM particle component inthe halo, corollary investigations on the nature of thecandidate have also been pursued. These investiga-tions are instead model-dependent and—considering



the large uncertainties which exist in the astrophys-ical, nuclear, and particle physics assumptions andin the parameters needed in the calculations—haveno general meaning (as is also the case of exclusionplots and of the DM particle parameters evaluatedin indirect detection experiments). Thus, it should behandled in the most general way, as we have pointedout over time [2, 3, 6–13].

Candidates, kinds of DM particle couplings withordinary matter and implications, cross sections, nu-clear form factors, spin factors, scaling laws, halomodels, etc., are discussed in [2, 3]. The reader canfind in this latter paper and in references therein de-voted discussions to correctly understand the resultsobtained in corollary quests and the real validity of anyclaimed model-dependent comparison in the field. Weremind the reader that the results briefly summarizedhere are not exhaustive of the many scenarios possibleat the present level of knowledge, including thosedepicted in some more recent works such as [19, 21].

DAMA/NaI is intrinsically sensitive to both lowand high DM particle mass having both a light (23Na)and a heavy (127I) target nucleus; in previous corol-lary quests for the candidate, DM particle massesabove 30 GeV (25 GeV in [6]) have been presen-ted [7, 9, 11–13] for a few (of the many possible)model frameworks. However, that bound holds onlyfor the neutralino when supersymmetric schemesbased on GUT assumptions are adopted to analyzethe LEP data [22]. Thus, since other candidates arepossible and also other scenarios can be consideredfor the neutralino itself as recently pointed out5),the present model-dependent lower bound quotedby LEP for the neutralino in the supersymmetricschemes based on GUT assumptions (37 GeV [25])is simply marked in the following figures. It is worthnoting that this model-dependent LEP limit—whenconsidered—selects the DM particle–iodine elasticscatterings as dominant.

For simplicity, the results of these corollary questsfor a candidate particle are presented here in termsof allowed volumes/regions obtained as a superposi-tion of the configurations corresponding to likelihoodfunction values distant by more than 4σ from thenull hypothesis (absence of modulation) in each ofthe several (but still a limited number) of the possiblemodel frameworks considered in [2, 3]. These allowedregions take into account the time and energy be-haviors of the single-hit experimental data and havebeen obtained by a maximum-likelihood procedurerequiring the agreement (i) of the expectations for the

5)In fact, when the assumption on the gaugino-mass unifica-tion at the GUT scale is removed, neutralino masses down to6 GeV are allowed [23, 24].

modulated part of the signal with the measured mod-ulated behavior for each detector and for each energybin and (ii) of the expectations for the unmodulatedcomponent of the signal with respect to the measureddifferential energy distribution and, from [9], also withthe bound on recoils obtained by pulse shape dis-crimination from the devoted DAMA/NaI-0 data [4].The latter one acts in the likelihood procedure asan experimental upper bound on the unmodulatedcomponent of the signal and—as a matter of fact—as an experimental lower bound on the estimate ofthe background levels. Thus, the C.L.’s that we quotefor the allowed volumes/regions already account forcompatibility with the measured differential energyspectrum and with the measured upper bound onrecoils. Finally, it is worth noting that the best fitvalues of cross sections and DM particle mass spanover a large range when varying the considered modelframework.

Figures 3–5 show some of the obtained allowedslices/regions (see [2, 3] for details and descriptionsof the symbols). Here, we only remind the reader thattanθ is the ratio between the DM particle–neutronand the DM particle–proton effective spin-dependentcoupling strengths and that θ is defined in the interval[0, π). Obviously, larger sensitivities than those re-ported in these figures would be reached when includ-ing the effect of other existing uncertainties on theastrophysical, nuclear, and particle physics assump-tions and related parameters; similarly, the set of bestfit values would also be enlarged as well. For details,see [2, 3].

In Fig. 6, the theoretical expectations in the purelySI coupling for the particular case of a neutralinocandidate in MSSM with gaugino mass unification atthe GUT scale removed [24] are shown. The markedcurve surrounds the DAMA/NaI purely SI allowedregion as in Fig. 4 (left).

4. THE PRESENT SITUATION IN THE FIELD4.1. Direct Detection

As already mentioned, no other experiment whoseresult can be directly compared in a model-indepen-dent way with that of DAMA/NaI is available so farin the field of DM direct detection.

Since some experiments have claimed to have “ex-cluded” DAMA/NaI [26–29], we will briefly bring tothe attention of the reader only a few arguments; inTable 2, some items for comparison are summarized.

Firstly, let us preliminarily assume as fully correctthe “selected” number of events, the energy thresh-old, the energy scale, etc., quoted by those experi-ments (see Table 2) and let us consider that, also un-der this hypothesis, their claims are obviously not jus-tified; in fact, (i) they give a single model-dependent



10

–3

10

–1

10

–3

10

–3

10

–1

10

–3

10

–1

ξσ

SD

, pb

10

–8

10

–6

10

–8

10

–6

10

–8

10

–6

10

–8

10

–6

10

–8

10

–6

ξσ

SI

, pb

θ

= 0

θ

=

π

/4

θ

=

π

/2

θ

= 2.435

40 GeV

50 GeV

70 GeV

90 GeV

110 GeV

10

–1

Fig. 3. Case of a DM particle with mixed SI–SD interaction for the model frameworks given in [2, 3]. Shaded areas: example ofslices (of the four-dimensional allowed volume) in the plane ξσSI vs. ξσSD for some of the possible mW and θ values. Inclusionof other existing uncertainties in parameters and models would further extend the regions; for example, the use of more favorableform factors and/or of more favorable spin factors than the ones considered here would move them towards lower cross sections.For details, see [2, 3].

10

–6

0 50

ξσ

SI

, pb

m

W

, GeV

10

–8

10

–4

10

–2

100 150 200

a

bc

de f

0

ξσ

SI

, pb

m

W

, GeV

10

–8

10

–6

200 400

If, e.g., SDcontribution

≠

0this regiongoes down

10

–4

10

–2

10

0

Fig. 4. (Left) Case of a DM particle with dominant SI interaction for the model frameworks given in [2, 3]. Region allowed inthe plane (mW , ξσSI). The vertical dotted line represents a bound in the case of a neutralino candidate when supersymmetricschemes based on GUT assumptions are adopted to analyze the LEP data; the low-mass region is allowed for the neutralinowhen other schemes are considered (see text) and for every other DM particle candidate. The inclusion of other existinguncertainties in parameters and models would further extend the region; for example, the use of a more favorable SI formfactor for iodine alone would move it towards lower cross sections. (Right) Example of the effect induced by the inclusion of anSD component different from zero on allowed regions given in the plane ξσSI vs. mW . In this example, the Evans logarithmicaxisymmetric C2 halo model with v0 = 170 km/s, ρ0 equal to the maximum value for this model, and a given set of parametervalues (see [2]) has been considered. The different regions refer to different SD contributions for the particular case of θ = 0:(a) σSD = 0, (b) 0.02, (c) 0.04, (d) 0.05, (e) 0.06, (f) 0.08 pb. An analogous situation is found for the other model frameworks.For details, see [2].



200 4000

m

W

, GeV

10

–4

10

–2

10

0

10

2

10

4

ξσ

SD

, pb

θ

= 2.435

If, e.g., SIcontribution

≠

0this regiongoes down

ξσ

p

, pb

10

–6

10

–4

10

–2

10

0

0 100 200 300 100 200 300

δ

, keV

50 GeV 110 GeV

Fig. 5. (Left) Case of a DM particle with dominant SD interaction in the model frameworks given in [2, 3]. Example of aslice (of the three-dimensional allowed volume) in the plane (mW , ξσSD) at a given θ value (θ is defined in the range [0, π));here, θ = 2.435 (Z0 coupling). For the definition of the vertical line and of the shaded area, see the caption of Fig. 4. Values ofξσSD lower than those corresponding to this allowed region are possible also, e.g., in the case of an even small SI contribution(see [2, 3]). (Right) Case of a DM particle with preferred inelastic interaction in the model frameworks given in [2, 3]. Examplesof slices (shaded areas) of the three-dimensional allowed volume (ξσp, δ, mW ) for some mW values. Inclusion of other existinguncertainties in parameters and models (as discussed in [2, 3]) would further extend the allowed regions in the two cases. Fordetails, see [2, 3].

result, while DAMA/NaI gives a model-independentresult; (ii) in the single (of the many possible) modelscenario, they consider that they “fix” all the astro-physical, nuclear, and particle physics assumptions ata single questionable choice; the same, is true even forthe experimental and theoretical parameters valuesneeded in the calculations.

In particular, they use different target nuclei(natGe or natXe) instead of 23Na and 127I targets ofDAMA/NaI and a comparison among experimentsusing different target nuclei can only be model-de-pendent, since several aspects must be considered,such as (i) the nature of the DM particle; (ii) itsreal coupling with ordinary matter (mixed SI–SD,purely SI, purely SD, preferred inelastic, etc.); (iii)the choice of the nuclear SI and SD form factorsand the relative parameters for each involved targetnucleus; (iv) the choice of the spin factor for eachinvolved target nucleus; (v) the real scaling laws forthe nuclear cross sections (see, for instance, [21]);(vi) the real halo model and the related parameters;and (vii) the real values of the experimental andtheoretical parameters within their uncertainties. Asan example, we recall that large differences in themeasured rate can be expected not only when usingtarget nuclei sensitive to the SD component of theinteraction (such as 23Na and 127I) with respect tothose largely insensitive to such a coupling (such asnatGe, natSi, natAr, natCa, natW, and natO), but alsowhen using different target nuclei, although all, inprinciple, are sensitive to such a coupling, dependingon the unpaired nucleon (compare, e.g., the odd-spin

isotopes—with neutron unpaired—of 129Xe, 131Xe,and 125Te and those with small abundance of 73Ge,29Si, and 183W with the proton-unpaired 23Na and127I cases) [2, 3]. Hence, most of the regions/volumesalready allowed by DAMA/NaI in the frameworksconsidered so far are unexplorable, e.g., by Ge, Si, Xe,and CaWO4 present experiments and by their futurepossible projection.

ξσ

scalar(nucleon)

, nb

m

χ

, GeV

10

–11

10

10

–9

5 10050 500

10

–10

10

–8

10

–7

10

–6

Fig. 6. [24]. Theoretical expectations of ξσSI vs. mW

in the purely SI coupling for the particular case of aneutralino candidate in MSSM with gaugino mass uni-fication at the GUT scale removed; the curve is the sameas in Fig. 4 (left).



In addition, DAMA/NaI is generally quoted therein an incorrect, partial, and unupdated way andexisting scenarios to which DAMA/NaI is fullysensitive—in contrast to the others—are ignored.Comments about the experimental aspects of thoseexperiments can be found elsewhere (see, e.g., [2, 3]).

In conclusion, those claims for a contradictionhave intrinsically no scientific bases. Moreover, as re-gards the experiments planning to use, in the keV en-ergy range, a very large mass of liquid noble gases asa target—in particular xenon—we remind the readerthat several technical difficulties exist and sharplyincrease with the size of the detector [32].

Finally, the strategies to reject electromagneticbackground from the data are limited by their sta-tistical nature because of, e.g., the tail effects fromthe different populations and the noise, and by theexistence of well-known side processes (such as end-range alphas, fission fragments, environmental neu-trons, and, in some cases, instrumental effects assurface electrons, etc.), whose contributions cannotbe estimated and subtracted at the level of precisionrequired for their claimed reachable sensitivity (ascan be easily derived even by simple considerations).Moreover, these techniques are absolutely unsuitableif the CDM candidates would instead directly involveionization/excitation phenomena in the detector.

4.2. Indirect Detection

The DM particles, via their annihilation either incelestial bodies (such as Earth and Sun) or in thegalactic halo, could give rise to high-energy neu-trinos, positrons, antiprotons, and gammas. There-fore, they could be indirectly detected by lookingeither for “upgoing” muons—produced by νµ—inunderground, underwater, or underice detectors orfor antimatter and gammas in space. However, itis worth to remarking that (i) there is no directmodel-independent comparison that can be per-formed between the results obtained in direct andindirect searches; (ii) the results obtained by indirectsearch are strongly model-dependent, requiring notonly astrophysical and particle physics modeling, butalso some additional hypothesis on the experimentalbackground.

In the case of “upgoing” muons in terrestrial de-tectors, the expected µ flux is the key quantity. How-ever, several sources of uncertainties are present inthe related estimates (and, therefore, in the obtainedresults), such as the assumption that a “steady state”has been reached in the considered celestial body andthe estimate and subtraction of the existing compet-ing processes, offered by the atmospheric neutrinos.Model-dependent analyses with a similar approach

have been carried out by large experiments deep un-derground such as MACRO and Super-Kamiokande.In particular, the case of the neutralino candidate inMSSM has been discussed in [33], showing that theirmodel-dependent results were not in conflict withDAMA/NaI.

As we mentioned, the annihilation of the DM par-ticles in the galactic halo could also produce antimat-ter particles and gammas. The antimatter searcheshave to be carried out outside the atmosphere, i.e., onballoons or satellites. In particular, the DM particles,annihilation would result in an excess of antiprotonsor positrons over an estimated background arisingfrom other possible sources. The estimate and sub-traction of such a background together with the influ-ence of the Earth and of the galactic magnetic field onthese particles plays a crucial role in the possibility ofa reliable extraction of a signal. However, at present,an excess of positrons with energy 5–20 GeV hasbeen suggested in [34] and other experiments. Inter-preted in terms of DM particle annihilation, it givesa result not in conflict with the effect observed byDAMA/NaI [34].

As regards the possibility to detect gammas fromDM particle annihilation in the galactic halo, exper-iments in space are planned. However, at present, itis difficult to estimate their possibilities considering,e.g., the background level and the uncertainties inits reliable estimate and subtraction as well as thesmallness of the expected signal (even more, if a sub-dominant component is present) when properly cal-culated with a rescaling procedure. However, in [35,36], the presence of a gamma excess from the centerof the Galaxy in the EGRET data [37] has alreadybeen suggested. This excess matches with a possibleDM particle annihilation in the galactic halo [35, 36]and is not in conflict with the DAMA/NaI model-independent result previously published.

For completeness, we recall that it has recentlybeen suggested [38] that the positive hints fromindirect detection—namely, the excess of positronsand gammas in space—and the effect observed byDAMA/NaI can also be described in a scenario withmulticomponent DM in the galactic halo, made ofa subdominant component of heavy neutrinos of thefourth family and of a sterile dominant component.

In particular (Fig. 7), it has been shown that aheavy neutrino with mass of around 50 GeV canaccount for all the observations, while the inclusion ofpossible clumpiness of neutrino density as well as newinteractions in the heavy neutrino annihilation, etc.,can lead to wider mass ranges: from about 46 GeV upto about 75 GeV (see [38] for details).


PARTICLE DARK MATTER 2063Ta

ble

2.F

eatu

res

ofth

eD

AM

A/N

aIre

sult

son

the

dark

mat

ter

part

icle

annu

alm

odul

atio

nsi

gnat

ure

over

the

seve

nan

nual

cycl

es[2

,3]w

ith

thos

eof

[26–

29]

DA

MA

/NaI

CD

MS

-II

Ede

lwei

ss-I

Zep

lin-I

Cre

sst-

II

Sig

natu

reA

nnua

lmod

ulat

ion

Non

eN

one

Non

eN

one

Targ

etnu

clei

23N

a,127I

nat G

ena

t Ge

nat X

eC

aWO

4

Tech

niqu

eW

ellk

now

nP

oorl

yex

peri

ence

dP

oorl

yex

peri

ence

dC

riti

calo

ptic

alP

oorl

yex

peri

ence

d

liqui

d/ga

s

inte

rfac

ein

this

real

izat

ion

Targ

etm

ass

[kg]

10

00.

750.

32

3

0.6

Exp

osur

e[k

gda

y]

1.1×

105

19.4

30.5

280

20.5

Dep

thof

the

1400

780

1700

1100

1400

expe

rim

enta

lsit

e[m

]

Sof

twar

een

ergy

210

202

12

thre

shol

d[k

eVe.

e.]

(5.5

–7.

5p.

e./k

eV)

(but

σ/E

=10

0%m

ostl

y

1p.

e./k

eV[2

9])

(2.5

p.e.

/keV

for1

6da

ys[3

0])

Que

nchi

ngM

easu

red

Ass

umed

=1

Ass

umed

=1

Mea

sure

dA

ssum

ed=

1

fact

or(s

eeal

so[3

1])

Mea

sure

dev

ent

1

cpd/

kg/k

eV??

,cla

imed

γla

rger

10

4ev

ents

tota

l

100

cpd/

kg/k

eV??

,6

cpd/

kg/k

eV

rate

inlo

wth

anC

DM

S-I

abov

e35

keV

ener

gyra

nge

(60

cpd/

kg/k

eV,

105

even

ts)

Cla

imed

even

tsE

ithe

r0

or1

2(c

laim

edta

ken

20

–50

cpd/

kg/k

eV16

afte

rre

ject

ion

ina

nois

ype

riod

)af

ter

reje

ctio

n

proc

edur

esan

d??

afte

r

stan

dard

PS

D[2

9,30

]

Eve

nts

sati

sfyi

ngM

odul

atio

nam

plit

ude

Inse

nsit

ive

Inse

nsit

ive

Inse

nsit

ive

Inse

nsit

ive

the

sign

atur

ein

tegr

ated

over

the

inD

AM

A/N

aIgi

ven

expo

sure

10

3ev

ents

Exp

ecte

dnu

mbe

rF

rom

few

dow

nto

zero

,F

rom

few

dow

nto

zero

,F

rom

few

dow

nto

zero

,

ofev

ents

from

depe

ndin

gon

the

mod

els

depe

ndin

gon

the

mod

els

Dep

endi

ngon

the

mod

els

depe

ndin

gon

the

mod

els

DA

MA

/NaI

effec

t(a

ndon

quen

chin

gfa

ctor

)(a

ndon

quen

chin

gfa

ctor

)(e

ven

zero

)(a

ndon

quen

chin

gfa

ctor

)



50 60 70 80 90 50 60 70 80 90Neutrino mass, GeV

10

–3

10

–2

10

–1

10

0

Local density fraction

χ

loc

DAMA favorable region DAMA favorable region

Gamma

GC

e

+

p

–

Halo

Fig. 7. [38]. Case of a subdominant heavy fourth neutrino candidate in the plane local density fraction vs. the heavy neutrinomass. The favorable region for this candidate obtained from the DAMA/NaI data (thin dashed line when using the Evans halomodel; thin solid line when using the other halo models) and the best-fit density parameters deduced from cosmic gammaradiation (from halo and galactic center (GC)), positron and antiproton analysis in the given model framework are shown (leftpanel). The effect of the inclusion of possible neutrino clumpiness is also presented (right panel). See [38] for details.

Fig. 8. (Left) The installation of the 25 NaI(Tl) crystals (9.70 kg each) of DAMA/LIBRA in HP nitrogen atmosphere. (Right)One of the final stages of the detectors’ installation. All the procedures, as well as taking these photos, have been carried out inHP nitrogen atmosphere.

5. TOWARDS THE FUTURE:FROM DAMA/NaI TO DAMA/LIBRA

AND BEYOND

The large merits of a highly radiopure NaI(Tl)setup have been demonstrated in practice byDAMA/NaI, which has been the most radiopuresetup available in this particular field. It has effectivelypursued a model-independent approach to investi-gate DM particles in the galactic halo, collectingan exposure several orders of magnitude larger thanthose available in the field, and has obtained manyother complementary or by-product results.

In 1996, DAMA proposed to realize a ton setup [39]and a new R&D project for highly radiopure NaI(Tl)detectors was funded at that time and carried out forseveral years in order to realize, as an intermediate

step, a second generation experiment, successor toDAMA/NaI, with an exposed mass of about 250 kg.

Thus, new powders and other materials were se-lected, new chemical/physical radiopurification pro-cedures in NaI and TlI powders were exploited, newgrowing/handling protocols were developed, and newprototypes were built and tested. As a consequenceof the results of this second generation R&D, the newexperimental setup DAMA/LIBRA (Large sodiumIodide Bulk for RAre processes), 250 kg of highlyradiopure NaI(Tl) crystal scintillators (matrix oftwenty-five 9.70 kg NaI(Tl) crystals), was funded atthe end of 1999 and realized. In fact, after the comple-tion of the DAMA/NaI data taking in July 2002, thedismantling of DAMA/NaI occurred and the installa-



Single triggers 1–25OB1

Gate500 ns

AND14

AND16

AND15

Gate600 ns

Gate50 nsD 1 usGate

openR

AND17

AND18

AND19

Scaler

Reset ACQLine CAMAC

STROBE IR

Main Trigger gate 1Main Trigger gate 2Main Trigger gate 3

Gate3 us

Gate3 us

Gate3 us

trigger TD1

trigger TD2

trigger TD3etc.

AnalogSUM

SCA

From AND15 DGG

IR input

X 25X 25

SA

SA

SA

ADC 3GAMAC

Gate

F3

WA1 VXI

WA1 VXI

L16Crystal1

PMT1

PMT2

A1

PreampGain 10

L1

Lightguide

PreampGain 10

A2

L5F1

L8

L9

L10

Cable line

Cable line

L6L3

F2

L11

L12

L13

Delayline

200 ns

L2

Cable lineL4

L7

Delayline

200 ns

R 1KΩ

TFA 2

TFA 1

GateADC 1

GAMAC

ADC 3GAMAC

Gate

Main Triggergate 1

Main Triggergate 3 L21

L17

Discrim 1

Discrim 2

L15

L18

AND1

Gate 1500 ns

OR enable 1Line CAMAC

Scaler 1CAMAC

Single Triggercrystal 1

Single Triggercrystal 1

L22

L23

L27

L26

L25

L24

L28 ln 1IR pattern

recognition

Fig. 9. (Top) Schema of the electronic chain referred to a single detector and its trigger. (Bottom) Schema of the main triggerof DAQ and of the trigger of the WA.

tion of DAMA/LIBRA started. The experimental siteand many components of the installation itself wereimplemented (environment, shield of PMTs, wiring,HP nitrogen system, cooling water of air conditioner,electronics and DAQ, etc.). In particular, all the Cuparts were chemically etched before their installationfollowing a new devoted protocol and maintained inHP nitrogen atmosphere until the installation. Allthe procedures performed during the dismantling ofDAMA/NaI and the installation of DAMA/LIBRAdetectors were carried out in HP nitrogen atmosphere(Fig. 8).

In Fig. 9, the analogic schema of theDAMA/LIBRA electronic chain for one crystal withits single trigger, the main trigger of the acquisitionsystem, and the trigger system of the waveformanalysers (WA) are reported. A dedicated electroniccircuit, shown in Fig. 9, allows triggering only the WAwhich correspond to fired detectors; it gives a triggerto each WA when (i) at least one of its correspondinglines has a trigger, (ii) the main trigger is present,

and (iii) the total energy of the events is in the chosenenergy window. Let us recall that, for the events withenergy outside this energy window (e.g., high-energyevents), the ADC values are acquired.

The data acquisition system is composed of aworkstation by Compaq with the Linux operatingsystem, which is interfaced with the hardware systemthrough MXI-2 and GPIB buses. The GPIB bus al-lows communicating with the CAMAC crate housingthe ADCs, the scalers, and the I/O registers, whilethe MXI-2 bus allows communicating with the threeVXI mainframes, where the WA are installed.

As previously for DAMA/NaI, a hardware/soft-ware system to continuously monitor the runningconditions is also operative; in particular, severalprobes are read out by the data acquisition systemand stored with the production data. Moreover, self-controlled computer processes are operational to au-tomatically control several parameters and to managealarms.



0 4000 8000 12000 16000 20000Energy, a.u.

200

400

600

800

1000

1200Frequency

σ

= 0.5%

–0.1 0 0.1(tdcal –

⟨

tdcal

⟩

)/

⟨

tdcal

⟩

0

20

40

60

80

100

120

140

160

180

Frequency

–0.1 0.1(

α

–

⟨α⟩

)/

⟨α⟩

0

Fig. 10. (Left) Energy distribution of the 241Am source as measured by one of the new highly radiopure DAMA/LIBRA NaI(Tl)detectors (σ/E = 6.7% at 59.5 keV). (Right) An example of the stability of the calibration factor (tdcal) and of the ratio of thepeaks’ positions (α) of the measured energy distribution of the 241Am source during about 1 yr of data taking.

DAMA/LIBRA has been taking data since March2003 and the first data release will, most probably, oc-cur when an exposure larger than that of DAMA/NaIhas been collected and analyzed in all aspects. Asa preliminary example of the performance of theDAMA/LIBRA setup, Fig. 10 (left) shows the energydistribution of the 241Am source as measured by oneof the new detectors (σ/E = 6.7% at 59.5 keV), whileFig. 10 (right) shows, as an example, the stability ofthe calibration factor and of the ratio of the peaks’positions of the 241Am source during about 1 yr ofdata taking.

The highly radiopure DAMA/LIBRA setup is apowerful tool for further investigation on the DMparticle component in the galactic halo having manyintrinsic merits [1–14] and a larger exposed mass, ahigher overall radiopurity, and improved performancewith respect to DAMA/NaI. Thus, DAMA/LIBRAwill further investigate the 6.3σ C.L. model-inde-pendent evidence pointed out by DAMA/NaI withincreased sensitivity in order to reach even higherC.L. Moreover, it will also offer an increased sen-sitivity to improve corollary quests on the nature ofthe candidate particle, trying to disentangle at leastfrom some of the many different possible astrophys-ical, nuclear, and particle physics models as well asto investigate other new possible scenarios. As anexample, we recall the effects induced on the DMparticle distribution by the contributions from tidalstreams from satellite galaxies, by the possible ex-istence of caustics, and by the possible existence of“solar wakes” [40]. In particular, recently, it has been

pointed out [19] that contributions to the DM parti-cles in the galactic halo should be expected from tidalstreams from the Sagittarius Dwarf elliptical galaxy.Considering that this galaxy was undiscovered until1994 and considering galaxy formation theories, onehas to expect that also other satellite galaxies do existand contribute as well. In particular, the Canis Majorsatellite galaxy has been pointed out as reported in2003 in [41]; it can, in principle, play a very significantrole being close to our galactic plane. At present, thebest way to investigate the presence of a stream con-tribution is to determine more accurately the phaseof the annual modulation, t0, as a function of theenergy [3]; in fact, for a given halo model, t0 would beexpected to be (slightly) different from 152.5 day andto vary with energy.

Moreover, other interesting topics will be addres-sed by the highly radiopure DAMA/LIBRA, such asthe study (i) on the velocity and spatial distributionof the DM particles in the galactic halo [2, 13]); (ii)on possible structures as clumpiness with small scalesize; (iii) on the couplings of the DM particle with the23Na and 127I target nuclei; (iv) on the nature of theDM particles; (v) on scaling laws and cross sections(recently, it has been pointed out [21] that, even forthe neutralino candidate, the usually adopted scalinglaws could not hold); etc. A large amount of work willbe faced by DAMA/LIBRA, which will also furtherinvestigate with higher sensitivity several other rareprocesses.

Finally, at present a third generation R&D efforttowards a possible NaI(Tl) ton setup has been fundedand related works have already been started.



6. CONCLUSION

DAMA/NaI has been a pioneering experiment in-vestigating for the first time the dark matter particleannual modulation signature with suitable sensitiv-ity and control of the running parameters. Duringseven independent experiments of one year each, ithas pointed out at 6.3σ C.L. in a model-independentway the presence of a modulation satisfying the manypeculiarities of an effect induced by DM particles inthe galactic halo; no systematic effect or side reac-tion able to account for the observed effect has beenfound. As a corollary result, it has also pointed outthe complexity of the quest for a candidate particlemainly because of the present poor knowledge of themany astrophysical, nuclear, and particle physics as-pects. At present, after a dedicated R&D effort, thesecond-generation DAMA/LIBRA (a 250 kg moreradiopure NaI(Tl) setup) has been realized and hasbeen in operation since March 2003, and a third-generation R&D effort toward a possible ton NaI(Tl)setup, which we proposed in 1996 [39], is also inprogress.

REFERENCES1. R. Bernabei et al., Nuovo Cimento A 112, 545 (1999).2. R. Bernabei et al., Riv. Nuovo Cimento 26, 1 (2003).3. R. Bernabei et al., Int. J. Mod. Phys. D 13, 2127

(2004).4. R. Bernabei et al., Phys. Lett. B 389, 757 (1996).5. R. Bernabei et al., Nuovo Cimento A 112, 1541

(1999).6. R. Bernabei et al., Phys. Lett. B 424, 195 (1998).7. R. Bernabei et al., Phys. Lett. B 450, 448 (1999).8. P. Belli et al., Phys. Rev. D 61, 023512 (2000).9. R. Bernabei et al., Phys. Lett. B 480, 23 (2000).

10. R. Bernabei et al., Eur. Phys. J. C 18, 283 (2000).11. R. Bernabei et al., Phys. Lett. B 509, 197 (2001).12. R. Bernabei el al., Eur. Phys. J. C 23, 61 (2002).13. P. Belli et al., Phys. Rev. D 66 , 043503 (2002).14. R. Bernabei et al., Phys. Lett. B 408, 439 (1997);

P. Belli et al., Phys. Lett. B 460, 236 (1999); R. Bern-abei et al., Phys. Rev. Lett. 83, 4918 (1999); P. Belliet al., Phys. Rev. C 60, 065501 (1999); R. Bernabeiet al., Phys. Lett. B 515, 6 (2001); F. Cappella et al.,Eur. Phys. J. C 14, 1 (2002); R. Bernabei et al., Eur.Phys. J. A 23, 7 (2005); 24, 51 (2005).

15. R. Bernabei et al., Nuovo Cimento A 103, 767 (1990);Nuovo Cimento C 19, 537 (1996); Astropart. Phys.5, 217 (1996); Phys. Lett. B 387, 222 (1996); 389,783(E) (1996); 436, 379 (1998); 465, 315 (1999);493, 12 (2000); New J. Phys. 2, 15.1 (2000); Phys.Rev. D 61, 117301 (2000); Eur. Phys. J. C 11, 1

(2001); Nucl. Instrum. Methods Phys. Res. A 482,728 (2002); Phys. Lett. B 527, 182 (2002); 546, 23(2002); in Beyond the Desert 03 (Springer, 2004),p. 541.

16. R. Bernabei et al., Astropart. Phys. 7, 73 (1997);R. Bernabei et al., Nuovo Cimento A 110, 189 (1997);P. Belli et al., Nucl. Phys. B 563, 97 (1999); P. Belliet al., Astropart. Phys. 10, 115 (1999); R. Bernabei etal., Nucl. Phys. A 705, 29 (2002); P. Belli et al., Nucl.Instrum. Methods Phys. Res. A 498, 352 (2003);R. Cerulli et al., Nucl. Instrum. Methods Phys. Res.A 525, 535 (2004).

17. K. A. Drukier et al., Phys. Rev. D 33, 3495 (1986);K. Freese et al., Phys. Rev. D 37, 3388 (1988).

18. D. Smith and N. Weiner, Phys. Rev. D 64, 043502(2001).

19. K. Freese et al., Phys. Rev. Lett. 92, 11301 (2004).20. M. Ambrosio et al., Astropart. Phys. 7, 109 (1997).21. G. Prezeau et al., Phys. Rev. Lett. 91, 231301 (2003).22. D. E. Groom et al., Eur. Phys. J. C 15, 1 (2000).23. A. Bottino et al., Phys. Rev. D 67, 063519 (2003);

A. Bottino et al, hep-ph/0304080; D. Hooperand T. Plehn, MADPH-02-1308; CERN-TH/2002-29; hep-ph/0212226; G. Belanger, F. Boudjema,A. Pukhov, and S. Rosier-Lees, hep-ph/0212227.

24. A. Bottino et al., Phys. Rev. D 69, 037302 (2004).25. K. Hagiwara et al., Phys. Rev. D 66, 010001 (2002).26. CDMS Collab., astro-ph/0405033; Phys. Rev. Lett.

84, 5699 (2000).27. EDELWEISS Collab., in Proceedings of NDM03,

Japan, 2003; Phys. Lett. B 513, 15 (2001).28. CRESST Collab., Astropart. Phys. 23 325 (2005).29. N. Smith, Talk given at IDM02, York, 2002.30. R. Luscher, Talk given at Moriond, 2003.31. E. Simon et al., Nucl. Instrum. Methods Phys. Res. A

507, 643 (2003).32. R. Bernabei et al., Cosmology and Particle Physics

(AIP, 2001), p. 189.33. A. Bottino et al., Phys. Lett. B 402, 113 (1997); 423,

109 (1998); Phys. Rev. D 59, 095004, 095003 (1999);Astropart. Phys. 10, 203 (1999); 13, 215 (2000);Phys. Rev. D 62, 056006 (2000); 63, 125003 (2001);Nucl. Phys. B 608, 461 (2001).

34. G. L. Kane et al., hep-ph/0108138.35. A. Morselli, Talk given at Vulcano 2002.36. A. Morselli et al., astro-ph/0211327.37. A. Strong et al., Astrophys. J. 537, 763 (2000).38. K. Belotsky et al., hep-ph/0411093.39. R. Bernabei et al., Astropart. Phys. 4, 45 (1995);

R. Bernabei, The Identification of Dark Matter(World Sci., 1997), p. 574.

40. F. S. Ling, P. Sikivie, and S. Wick, astro-ph/0405231.41. R. A. Ibata et al., Mon. Not. R. Astron. Soc. 348, 12

(2004).


Documents

Particle dark matter: From DAMA/NaI to DAMA/LIBRA