The Future Role of Inorganic Crystal Scintillators in Dark Matter Investigations

: Crystal scintillators and in particular inorganic scintillators play an important role in the investigation of Dark Matter (DM) and other rare processes. The investigation of a DM signature, as the annual modulation, or the directionality technique requires the use of highly radiopure detectors able to explore the very low energy region maintaining a high stability of the running conditions. In this paper, the cases of NaI(Tl), ZnWO 4 and SrI 2 (Eu) crystal scintillators are described in the framework of our activities at the Gran Sasso National Laboratory of the INFN. Their role, the obtained results in DM investigation, as well as their potential and perspectives for the future are reviewed.


Introduction
Astrophysical observations have revealed the presence of DM on all astrophysical scales and much experimental and theoretical evidence supports arguments that a large fraction of DM should be in the form of relic DM particles. Crystal scintillators and in particular inorganic scintillators play an important role in the investigation of DM and other rare processes. In particular, the investigation of a DM signature requires the use of highly radiopure detectors able to explore the very low energy region with a high stability of the running conditions. The paper will review the cases of the NaI(Tl), ZnWO 4 and SrI 2 (Eu) crystal scintillators, developed and used in the framework of our activities at the Gran Sasso National Laboratory of the INFN.
The best qualities offered by inorganic scintillators are due to their physical characteristics (high Z materials i.e., high detection efficiency for gamma quanta, physical form, chemical and mechanical stability) and to their luminescent characteristics (emission wavelength, high light yield, good linearity, suitable decay time, and good radiation hardness). In particular, the high radio-purity, high light yield, the possible choice of different target nuclei, the stability of performance and the reproducibility of the results make these detector particularly suitable for the study of DM annual modulation (DM-AM) signature [1,2].
Ultra-low-background (ULB) NaI(Tl) crystal scintillators have been used by the DAMA/NaI, DAMA/LIBRA-phase1 and DAMA/LIBRA-phase2 [3][4][5][6][7] experiments at the Gran Sasso National Laboratory of INFN (LNGS) for the direct search for DM. These experiments have obtained evidence at a 12.9 σ confidence level of a signal satisfying all the requirements of the model independent DM-AM signature. The selected approach and target assure sensitivity both to DM candidate particles inducing nuclear recoils, as well as to those for which the products of the interaction have electromagnetic nature (mainly electrons). Moreover, the observed signal is compatible with a wide set of scenarios regarding the nature of the DM candidates and related astrophysical, nuclear, and particle physics (see, e.g., Ref. [8] and references therein).
Other features of inorganic scintillators, such as channeling and anisotropy, make them very promising for the investigation of the directionality of DM [9], potentially effective for those DM candidate particles able to induce nuclear recoils. The directionality approach is based on the study of the correlation between the DM particles' arrival direction, through the induced nuclear recoils, and the Earth motion in the galactic rest frame. A detector sensitive to nuclear recoils direction is then required and a promising option is represented by the anisotropic scintillators, where the scintillation properties (quenching and scintillation decay kinetics) depend on the direction of the detected heavy particles relatively to the crystal axes. Among the anisotropic crystal scintillators, the zinc tungstate (ZnWO 4 ) offers very interesting features and its potential for DM investigation with the directionality technique is explored in the framework of the ADAMO project (Anisotropic detectors for DArk Matter Observation) [10][11][12].
Finally, other scintillators became attractive in the last few years, such as SrI 2 (Eu) crystal scintillators because of their high light output (∼100,000 photons/MeV) and of the absence of natural long-living radioactive isotopes. These and other features make SrI 2 (Eu) scintillators promising for low counting experiments such as the search for double β decay ( 84 Sr is a candidate) and the investigation of DM, which requires a very low energy threshold. In particular, in 84 Sr, the double electron capture (2ε) and electron capture with emission of positron (εβ + ) are energetically allowed. These decays have never been observed and the half-life limits are of the order of 10 16 y [13]. Studies of 2ε and εβ + decays are important because the observation of neutrinoless mode of such a process could help to distinguish between the mechanisms of neutrinoless 2β − decay: whether it is due to the light Majorana neutrino mass or due to the right-handed currents' admixture in the weak interaction [14].
In the following section, the strategies to develop ULB scintillators will be discussed. Then, the cases of NaI(Tl), ZnWO 4 and SrI 2 (Eu) crystal scintillators and their application in the investigation of DM will be reviewed, focusing the attention on the results obtained by DAMA/LIBRA using NaI(Tl) detectors and on the recent measurements of the ADAMO project, utilizing ZnWO 4 anisotropic detectors.

Radiopurity of Crystal Scintillators
A wide set of prerequisites has to be fulfilled to obtain the best sensitivity in the investigation of DM, e.g.,: (i) the choice of a suitable underground laboratory for cosmic background reduction; (ii) a suitable shield to be used underground to prevent background from environmental radioactivity (gammas, muons, neutrons, radon, etc.); (iii) the choice of a suitable set-up configuration, e.g., with multiple detectors that offers the possibility to discriminate background events; (iv) selection of low radioactivity materials;, etc.. Nevertheless, the most effective way to obtain the highest sensitivities is closely related to the achievement of an extreme radio-purity in the used crystal scintillators. This process can be slow and challenging.
In inorganic scintillators, the presence of natural radionuclides from 232 Th, 238 U and 235 U families and 40 K, and other cosmogenic and/or anthropogenic radionuclides is common. Some materials contain elements with radioactive isotopes, like 113 Cd in CdWO 4 , 138 La in LaCl 3 and LaBr 3 , and so on [15].
Achieving the high radiopurity goal relies on a highly specialized experience and a long, delicate work. In fact, in addition to the prerequisites already mentioned, it is important to have experience on natural, cosmogenic and anthropogenic contaminants, which can be different even in batches of the same material. For each used material, it is important to know origin, history, and manipulation processes. Furthermore, particular attention has to be dedicated to the procedures and protocols for the project/storage/cleaning/installation of the experiment, and to the specific experimental optimizations, which can be pursued for the radio-purity goal. Finally, a deep experience on the measurements and analysis techniques, especially developed for this field, is mandatory. Similar developments and measurements follow different subsequent steps, which can be summarized for inorganic scintillators in: In case of a positive outcome producing an improvement in radio-purity, the production of detectors can start. Otherwise, the critical problems have to be solved with targeted solutions.
Regarding the measurement techniques of radioactive contaminations, it is worth mentioning that the ULB HP-Ge detectors can reach sensitivity at the level of mBq/kg (generally ppb) for 228 Th ( 232 Th), 226 Ra ( 238 U), 227 Ac ( 235 U) and 40 K. This method is commonly used to measure gamma emitters in samples of materials used in the detectors or surrounding areas.
Analytical methods of analysis are also applied, like the Atomic Absorption Spectrometer (AAS) and the Inductively Coupled Plasma Mass Spectroscopy (ICP-MS). Both methods require pre-treating of the samples. In the case of the AAS, the wavelengths of light absorbed by the different elements are used to evaluate their concentrations in the solutions, and a good sensitivity (at ppb, sub-ppb level) can be reached, while an even better sensitivity at ppb-ppm and ppq-ppt level can be obtained with the ICP-MS. In this technique, the mass-to-charge ratio (m/e) is used to identify the isotopes of an element. Each peak in the mass spectrum has an intensity proportional to the isotope concentration. However, only a final low background measurement can assure the highest sensitivity to the identification of the internal contaminants in the crystal, and it is essential to collect the information for the full assembled detector configuration.
The presence of a wide range [15] of radioactive nuclides has been explored in inorganic crystal scintillators. When made for low-background purposes, like NaI(Tl), CaF 2 (Eu), CsI(Tl), ZnWO 4 , CdWO 4 , and Li 2 MoO 4 , they can have standard contaminants in the range 0.01-1 mBq/kg.
The initial impurity concentrations in the used materials, the used processes, and the handling protocols contribute to determine the achievable levels of radiopurity. The tools and abrasives used for crystals' cutting and polishing should be carefully selected. In particular, after the ingots production, all the materials used for the detector production and assembly have to be selected for radiopurity and their number should be minimised. Moreover, all the operations on the bare crystals require selected and controlled environment and procedures, in order to avoid possible surface pollution (see, e.g., Refs. [15,16]).
Nevertheless, slight differences in the residual contaminants of detectors may still arise, even if the same procedures are applied. In fact, some casual pollutions during the growth and handling procedures may in principle be possible, due to the industrial environment of the production sites. Differences may also depend on the selected part of the boule/ingot used to build the detector (a possible non-uniformity of contaminant distribution can play a role [17]).
In addition, during assembly, transport, storage, installation, etc., further pollution may also occur (e.g., at any step there can be the danger of radon contamination, if there is a contact with atmospheric air; the road-transport is preferred, in the shorter time possible, to minimize the cosmogenic activation; the installation is performed in HP nitrogen atmosphere, etc.). However, if accurate protocols are followed, all the critical aspects can be reduced, and the reduction of contaminants and the optimization of the detector performances can be achieved.
Focusing on DM investigation, a major effort has been done for the R&D of NaI(Tl) crystal scintillators. A first advantage of NaI(Tl) is the absence of long lived isotopes. Other fundamental aspects to be considered are the method of crystallisation, the purification of the powders, the specific production line, the ovens, the crucible and so on. The growth of ULB NaI(Tl) crystals was made by the Kyropoulos method using selected platinum crucibles, with dedicated protocols [18]. In case of ULB NaI(Tl) crystals, it does not seem useful to increase the numbers of re-crystallisations to improve radiopurity, whereas specific actions on the chemical/physical purifications of the powders are performed. As an example, in the case of DAMA/LIBRA, the final residual contaminants in the powders of the ULB NaI(Tl) were for 238 U: 0.02 ppb, for 232 Th: 0.02 ppb, and for nat K: <0.1 ppm (95% C.L.), respectively, by MS and AAS measurements. Even if the Tl concentration in the detectors is 0.1%, the TlI powders were specifically selected and dedicated chemical/physical purifications were applied. The residual contaminants were for 238 U: 0.8 ppb, for 232 Th: 1.2 ppb, and for nat K: <0.06 ppm (95% C.L.), respectively, by MS and AAS measurements. The Kyropoulos crystallisation process (platinum crucibles are preferred) by itself represents a further important purification step of the residual contamination in the powders. In fact, the residual background directly measured underground with the DAMA crystals used as detectors is at ppt level for all three radionuclides 238 U, 232 Th and 40 K [18].

DAMA/LIBRA Investigation on DM with NaI(Tl) Detectors
In the DAMA/LIBRA experiment, NaI(Tl) crystals are employed as scintillator detectors to investigate the presence of DM particles in the Milky Way. For this purpose, the DM-AM signature is exploited. This signature is due to the revolution velocity of the Earth around the Sun which is in movement with respect to the galactic frame. The composition of the velocity of the Earth and of the Sun in the galactic frame results in the modulation of the flux of DM particles crossing the Earth. In particular, the flux is expected to be larger around 2 June, when the two velocities are aligned, while it is expected to be smaller around 2 December when the two velocities are opposite. Slight variations may occur in case of the presence of non-thermalized DM components in the galactic halo.

The Target Material and the Experimental Set-Up
The highly radiopure NaI(Tl) scintillators are a very good target detector to investigate the DM particle component in the galactic halo. They have many advantages: well known technology; 2.
high radiopurity level that can be obtained by material selections, the design of growth and handling protocols, by applying chemical/physical purification procedures, etc.; 3.
crystal of large mass can be grown; 4.
the experimental duty cycle is very high; 5.
it is possible to perform routine calibrations in the keV energy range without perturbing the data taking; 6.
the operational conditions can be controlled and stabilized; 7.
no microphonic noise is present; 8.
the discrimination between scintillation signal and noise (mainly photomultiplier (PMT) noise, Cherenkov light in the light guides and in the PMT windows, and afterglows) close to the energy threshold is effective thanks to: the high number of photoelectrons/keV, the different time decay of PMT noise pulses (of the order of tens of ns) and of the scintillation pulses (of the order of hundreds of ns); 9.
high light yield (∼40,000 photons/MeV) that allows for having a threshold at keV range; 10. pulse shape discrimination feasible at a reasonable level when of interest [19]; 11. no re-purification or cooling down/warming up procedures needed (implying high reproducibility, high stability, etc.); 12. it is possible to have a multi-detectors set-up that can have an important role for DM investigations and for background recognition; 13. sensitivity to both high (by iodine target) and low (by Na target) mass DM candidates; 14. high sensitivity to a DM particles' candidate with spin-independent, spin-dependent and mixed couplings; 15. high sensitivity to several other existing scenarios and to many other possible candidates including those producing just electromagnetic radiation in the interaction; 16. effective to investigate the DM-AM signature by considering all its characteristics; 17. "ecologically clean" apparatus, having no safety issues; 18. more affordable than other detectors considered in the field; 19. compact in size, requiring small occupancy in an underground laboratory; The DAMA/LIBRA experiment can be considered as the successor of the pioneer DAMA/NaI set-up, and it has been taking data in two configurations; the first one was DAMA/LIBRA-phase1 and the second, presently running, is DAMA/LIBRA-phase2 [20]. DAMA/LIBRA-phase2 is an improved version of the experiment in which, as a main upgrade, all the PMTs were changed to be replaced with the Hamamatsu R6233MOD PMTs; these devices have higher quantum efficiency and lower background with respect to those used in phase1 [21,22]. The upgrade has been performed in 2011 and the goal to reach a software energy threshold at 1 keV was achieved.
DAMA/LIBRA-phase2 set-up is made of 25 highly radio-pure NaI(Tl) crystal scintillators (5-rows by 5-columns matrix) having 9.70 kg mass each one; the residual contaminants of all the parts of the apparatus have been measured in details and they are reported in Ref. [18]. The schematic view of the building hosting the DAMA/LIBRA apparatus and of its shield is shown in Figure 1. About 1 m concrete (made from the Gran Sasso rock material) almost fully surrounds (mostly outside the building) the apparatus. In particular, the room A hosts the glove-box for calibrations; some facilities are in the room B; the sensitive parts of the apparatus are in the room C inside the multicomponent shield as well as the electronics; the computer for DAQ is in the room D. Right: schematic view of the passive shield of the DAMA/LIBRA apparatus. Reprinted with permission from [18]. Copyright 2008 Elsevier. Two 10-cm long UV light guides of Suprasil B quartz are coupled to the detectors on the two end faces of the crystal; they connect each crystal to the low background PMTs working in coincidence at a single photoelectron level. The detectors are placed inside a sealed low-radioactive copper box installed in the center of a low-radioactive Cu/Pb/Cd-foils/polyethylene/paraffin shield. About 1 m concrete, obtained from the Gran Sasso rock material, is present outside the detector barrack and completely encloses it. The concrete layer acts as an external shield to moderate neutrons. A layer of neoprene lies down on the floor of laboratory to separate it from the concrete basement. A thickness of several tens of cm of paraffin fills the space between the basement and the metallic structure.
To isolate the set-up from the environmental air of the underground laboratory, a threefold-level sealing system is in use. In addition, the set-up is maintained in HP (highpurity) nitrogen atmosphere. To assure the stability of the temperature, two independent and redundant systems of air conditioning are active. The huge heat capacity of the multitons passive shield (≈10 6 cal/ • C) guarantees further stability of the detectors' operating temperature. The operating conditions are monitored continuously with a dedicate data acquisition system connected to several probes and sensors. The running conditions are very stable at a level better than 1% (see, e.g., Ref. [23]).
The light yield of the detectors in the phase2 ranges from 6 to 10 photoelectrons/keV. To calibrate the energy scale of the detectors, various X-rays/γ sources are used. The calibration operations are carried out without any change of the experimental conditions (for details see, e.g., Ref. [18]). In addition, some gamma lines due to the presence of residual radioactive contamination are also considered. As an example, the 40 K, present in the crystal at ppt level, produces double coincidence events due to internal X-rays emission after its EC decay. A line at 3.2 keV is close to the software energy threshold. The DAQ system records both single-hit events (where just one of the detectors fires) and multiple-hit events (where more than one detector fires) up to the MeV region, although the trigger optimization is performed for the lowest energy events.
The DM-AM signature is a very powerful signature because the expected features of the related signal must satisfy several conditions: the rate has a component modulated according to a cosine function (1), the period of the modulation is one year (2), and its phase is 2 June (3); this modulation should be present in the keV energy range considering the energy released by the DM particle in the interaction with target detectors (4); the modulated signal is characterized by events in which just one detector "fires" (single-hit events), since the interaction of a DM particle with more than one detector has a negligible probability (5); in the region where the signal is maximum, the modulation amplitude is expected to be 7% when Maxwellian-like halo distributions are considered, but it can be larger (even up to 30%) in some scenarios [25][26][27][28][29] (6). The NaI(Tl) target detectors are sensitive to many DM candidates, interaction types, and scenarios (see, e.g., Ref. [30] and references therein). A possible fake effect or systematic should not only account for the measured modulation amplitude but in addition simultaneously satisfy all the above-mentioned conditions (see, e.g., Ref. [30] and references therein).

The Model-Independent Results on DM-AM
Six full annual cycles of DAMA/LIBRA-phase2 have been considered in the analysis of the DM-AM signature; the whole exposure is 1.13 ton × yr, while including also DAMA/NaI and DAMA/LIBRA-phase1, the total exposure is 2.46 ton × yr. The same procedures already adopted for the DAMA/LIBRA-phase1 [4][5][6]18,24] have also been exploited in the analysis of DAMA/LIBRA-phase2 [7,23,31].
Many analyses have been performed on the data to study the peculiarities expected from the DM-AM signature. In this publication, a part of the results and examples will be reported. Additional, detailed results have been presented in Refs. [7,23,31]. As an example, in Figure 2, the residual rates of the single-hit scintillation events measured in the former DAMA/LIBRA-phase1 and the new DAMA/LIBRA-phase2 are reported. As described e.g., in Refs. [3][4][5][6]24,32], the residual rates are calculated from the rate of the single-hit events by subtracting the unmodulated part, in the energy interval from the software energy threshold of DAMA/LIBRA-phase1 (2 keV) to 6 keV. The null hypothesis of the absence of modulation is discarded by χ 2 test at very high C.L.: χ 2 /d.o. f . = 199.3/102, and related p-value = 2.9 × 10 −8 . The residual rates for the lower energy threshold of 1 keV are shown elsewhere [7,23,31].
The residual rates of the single-hit events for the full data sets DAMA/NaI, DAMA/ LIBRA-phase1, and DAMA/LIBRA-phase2 was fitted by the function (see Figure 2): A cos ω(t − t 0 ), where the period T = 2π ω = 1 yr and the phase t 0 = 152.5 day (2nd June) are those expected by the DM-AM signature. The fitted modulation amplitude is A = (0.0102 ± 0.0008) cpd/kg/keV (χ 2 /d.o. f . = 113.8/138) [7]. If the period and the phase of the signal are kept free, fitting the full exposure (2.46 ton × yr), the result of the fit is: the modulation amplitude of the single-hit scintillation events A = (0.0103 ± 0.0008) cpd/kg/keV, the period T = (0.999 ± 0.001) yr, and the phase (145 ± 5) days [7], all these values are well in agreement with those expected for DM particles; the C.L. is 12.9 σ.
This result confirms the evidence of a signal satisfying all the requirements of the model independent DM-AM signature. Neither systematics nor side reactions able to mimic the DM-AM signature (i.e., to account for the measured modulation amplitude and to simultaneously satisfy all the requirements of the signature) have been found so far.
Other analyses, not shown here, exclude the presence of any background modulation in the energy regions above 6 keV. Among them, the time behaviour of the integral rate above 90 keV, R 90 , has been analysed [6,7,24,30]; similar conclusions have also been obtained considering data in other energy regions.
Among other performed analyses, a further investigation on DAMA/LIBRA-phase2 data was done-as in the previous DAMA/NaI and DAMA/LIBRA-phase1 experiments [4][5][6]24,32]-in the residual rate of the multiple-hit events. These events are collected and analyzed through the same hardware and software procedures as for the single-hit ones. Because of the negligible probability that a DM particle interacts in more than one detector, a DM signal is expected only in the single-hit residual rate. Hence, the proper comparison of single-hit and multiple-hit residuals is equivalent to compare the cases of DM particles beam-on and beam-off. This implies a further test on the background in the same energy region where the positive effect is observed.
The residual rates of the single-hit scintillation events and of the multiple-hit events, as collected in a single cycle, are reported in Figure 3 for DAMA/LIBRA-phase2; the energy range is (1-6) keV. The single-hit residual rate shows a clear modulation, satisfying all the peculiarities of the DM-AM signature, while the multiple-hit one does not; the fitted modulation amplitude for the multiple-hit residual rate is: (0.0004 ± 0.0004) cpd/kg/keV, well compatible with zero. Therefore, evidence of annual modulation with the proper peculiarities of the DM-AM signature is present in the single-hit residuals (class of events to which the DM events belong), and it is absent in the multiple-hit residuals (class of events to which only background events belong).

1-6 keV
Time (day) Residuals (cpd/kg/keV)  No possible source of systematics or side reactions have been found; therefore, cautious upper limits have been set which are not able to account for the modulation amplitudes measured by DAMA, and are not able to simultaneously satisfy all the requirements of the DM-AM signature; for details, see Refs. [3][4][5][6]32,[34][35][36]. In particular, we recall thaton the contrary to what was claimed in Ref. [37] and sometimes uncritically repeated by some others-no possibility exists for any appreciable contribution from the combination of solar neutrinos and muons (one with a phase much lower and the other larger than the one measured by DAMA experiments) for the many scientific quantitative reasons reported in Ref. [34] and Refs. therein. As evident, the DAMA/LIBRA multi-detectors target mass, 250 kg in total, cannot detect any significant contribution to the modulation of its measured single-hit events from those sources, which are typically investigated underground by multi-hundred-tons experiments. Similar analyses on systematics and side reactions were also performed for the DAMA/NaI data [3,32].
In conclusion, the long-standing annual-modulation evidence observed by DAMA is model-independent, and it can be compatible with a wide set of DM scenarios as discussed in DAMA (see, e.g., Refs. [8,30]) and other literature.

Perspectives
To efficiently disentangle some of the numerous potential candidates and theoretical frameworks, decreasing the DAMA/LIBRA energy threshold below 1 keV is important. In particular, the experimental sensitivity to the DM-AM is connected to the product: where is the overall efficiency, ∆E is the energy region where the DM-AM is present, M is the exposed mass, T is the running time, and (α − β 2 ) term gives information about the distribution of the collected data along each annual cycle. In particular, α = cos 2 ω(t − t 0 ) is the mean value of the squared cosine and β = cosω(t − t 0 ) is the mean value of the cosine (the averages are taken over the lifetime of the data taking and t 0 = 152.5 day, i.e., 2nd June); thus, (α − β 2 ) indicates the variance of the cosine (i.e., it is 0.5 for a full year of data taking). This quantity is relevant to pursue the DM-AM investigation. Lowering the software energy threshold corresponds to increasing the acceptance efficiency. An improved experimental sensitivity can investigate the DM-AM more extensively and improve the modulation parameters' measurement, e.g., the phase, which yields relevant information. Indeed, the DAMA project's goal was to point out model-independent evidence for the presence of DM particles in the galactic halo and study the characteristics of such particles and related astrophysical, nuclear and particle physics' scenarios. In particular, a few primary purposes of achieving a lower software energy threshold are: • to explore the DM-AM at lower software energy threshold, with high overall efficiency, to untangle among some of the many proposed theoretical frameworks; • to investigate, also considering the recent GAIA measurements, feasible presence of streams (as we already did for Sagittarius in the past), and caustics in the Galaxy, including Sun's effects gravitational focusing; • to examine the nuclear parameters assumed in model-dependent corollary analyses; • to investigate the diurnal modulation with an increased sensitivity and other feasible diurnal effects due for e.g., to the Earth shadow and the channelling (refer to the DAMA literature for a complete discussion); • to study rare processes other than DM, such as analyzing either other regions of the energy spectrum or studying specific peculiarities of the processes investigated, as performed earlier and published using the DAMA/NaI and DAMA/LIBRA-phase1 data (from keV up to tens MeV; see Refs. [38][39][40][41][42][43][44][45][46][47][48]).
The DAMA collaboration has been operating towards this objective. The purpose is to further improve the signal-to-noise ratio approaching the software energy threshold to untangle the noise pulses (time decay of order of tens ns) and the scintillation events (time decay of the order of 240 ns), including a high overall efficiency, below one keV [18,21]. Thus, new high Q.E. and low radioactivity PMTs have been developed by the Hamamatsu Co. based on the DAMA requirements, e.g.,: (i) Q.E. at λ = 420 nm 30-40%; (ii) dark current < 100 cps; (iii) 3-inch window diameter; (iv) specific shape and size to fit the already-existing honeycomb-shaped copper shield around PMTs and to minimize the amount of material; (v) multiplication factor > 10 6 ; (vi) peak/valley ratio > 2.5; (vii) PMTs radio-purity at the level of a few mBq/PMT. Moreover, to improve the signal/noise ratio, a new optimized preamplifier-voltage divider system, integrated on the same Pyralux board, was developed and mounted on the new metal PMTs. The ultra low noise operational amplifier LMH6624, developed by Texas Instruments, is the core of the new preamplifier produced. It works at ±5 V, with an input bias current of −15 µA at 300 K and a bandwidth of 1.5 GHz. A view of this new optimized preamplifier-voltage divider system is shown in Figure 5. The preamplifier-voltage divider system is printed on the first and third layer of a Pyralux board; between these two layers, a ground layer is deposited. Such a system is installed on the PMT backside in the shield's copper honeycomb structure. Thus, the preamplifiers are as close as possible to the input source (the anode of the PMT) instead of those used so far in DAMA/LIBRA-phase2, which are installed beyond the inner part of the set-up's shield. The upgrade is now possible considering the improved radiopurity of the new metallic PMTs and the novel miniaturized preamplifier-voltage divider system. In summary, a larger signal/noise factor can be achieved with respect to an external preamplifier and the overall radio-purity is improved.
Analyses with and without the preamplifier have been performed to characterize the metallic PMTs, their response, and the preamplifier-voltage divider system's performance: The preamplifier-voltage divider system (overall mass 14.1 g) has been measured in an HP-Ge detector in the STELLA facility (at the LNGS) during 6.87 days to test its residual radioactivity. In particular, the residual activity of 232 Th, 235 U, 40 K, 137 Cs, and 60 Co is much lower than that of a single photomultiplier. Only the 226 Ra activity is slightly higher, at the level of a 13 mBq per preamplifier-voltage divider system. However, this is a common characteristic for such kind of equipment: the used Pyralux material usually has an activity of about hundred times lower for 226 Ra and about 30 times lower for 228 Ra and 228 Th. Nevertheless, the small activity is principally due to the electronic elements. However, the residual total background due to the preamplifier-voltage divider system is negligible because it is placed behind the PMT and shielded largely by a honeycomb Cu structure.
Four DAMA/LIBRA detectors are currently equipped and in data taking mode using the new metal R11065-20 MOD PMTs and the innovative preamplifier-voltage divider system. Other alternatives and cheaper configurations are under study to lower the detectors' software energy threshold in light of these developments and goals. The work is in progress also considering the use of new Transient Digitizers, and a thorough analysis is being carried out; results will be published.

Investigating the Directionality of DM with ZnWO 4 Crystal Scintillators
Another effective approach, in the case of DM candidate particles that induce nuclear recoils, is the directionality technique. The directionality investigates the correlation between the direction of the nuclear recoils induced by DM particles and the motion of the Earth in the galactic rest frame. In fact, due to the solar motion in the DM halo, a wind of DM particles is expected on the Earth with an apparently opposite direction with respect to the Earth velocity in the galactic rest frame. As a consequence of the Earth's rotation, an observer on the Earth's surface experiences a change in the DM average arrival direction during a sidereal day. The induced nuclear recoils' directions are expected to be strongly correlated with that of the DM particles. Thus, the study of the nuclear recoils directions can highlight the presence of a very distinctive signal, as expected for a large class of DM candidates interacting via elastic-scattering on target nuclei. On the other hand, the nuclear recoil range is very short: typically of the order of µm in solid detectors and of the order of a few mm in low pressure Time Projection Chambers. This makes the direct measurement of the recoil tracks very difficult. However, this limitation can be overcome by using anisotropic scintillation detectors [49][50][51].
In anisotropic scintillators, the detector response to heavy particles depends on their impinging direction with respect to crystal axes. Therefore, in the case of nuclear recoils induced by DM particles, the measured energy spectrum at low energy is expected to change its shape during the sidereal day as a consequence of the changing orientation of the crystal axes with respect to the DM wind. This effect can be pointed out as a peculiar variation of the counting rate measured in a given low energy window and offers the possibility to highlight the DM signal with respect to the electromagnetic background (for details, see, e.g., in Refs. [10,50,51]).
The potential offered by the anisotropic properties of ZnWO 4 crystal scintillator in the search for DM with the directionality approach was indicated for the first time in Ref. [49]. Recent measurements and R&D show that the ZnWO 4 detector can offer suitable features for the purpose [10][11][12]52,53]. The light output and the pulse shape indeed depend on the particle direction with respect to crystal axes in the case of heavy particles and nuclear recoils, while no difference is observed for γ/β radiation [49]. The ZnWO 4 detector also ensures a high atomic weight and the possibility to obtain large crystals with masses of some kg [54]. Moreover, the three target nuclei, Zn, W and O, have very different masses and make the detector sensitive to DM candidates with small and large mass. In light of this, the ADAMO project was considered and R&D has been progressing [10][11][12]. DAMA and the Institute for Nuclear Research of NASU (Ukraine) have developed various ZnWO 4 detectors [17,52,53,55,56]), and an R&D activity is presently ongoing with the Nikolaev Institute of Inorganic Chemistry (Novosibirsk, Russia). Large ZnWO 4 detectors, produced with the conventional and the low-thermal gradient (LTG) Czochralski technique in a platinum crucible [54], have been measured underground at LNGS in the DAMA/R&D set-up [52,53,[55][56][57].
The measurements performed at LNGS also confirmed that the ZnWO 4 light output at room temperature is about 20% of the NaI(Tl) scintillator. Previous studies have shown that it is possible to improve the light output of the crystal when working at low temperature [58]. This feature will also be verified at LNGS, where a small cryostat is under testing for implementing and optimizing the cooling system. This system will allow the measurement of the light output at a stable working temperature of about −50 • C.
New measurements, with important implications for the use of ZnWO 4 as a detector for the DM directionality approach, have been recently performed by the ADAMO project using a (10 × 10 × 10.4) mm 3 , 7.99 g mass, crystal scintillator [12]. In the measurements performed with a neutron generator, the small size of the crystal reduces the probability of multiple scattering of the neutron in the detector. The used ZnWO 4 is a second crystallization crystal produced with the LTG Czochralski technique from zinc tungstate crystals made from tungsten oxide additionally purified by double sublimation of tungsten chlorides [59].
The first measurement, carried out at LNGS, allowed the study of the detector response to α particles and γ quanta, and the measurement of the α/γ ratio. The α/γ ratio and, more generally, the quenching factor is the ratio between the detected energy in the energy scale calibrated with γ sources to the energy of the heavy ionizing particle. The ZnWO 4 crystal was coupled to a Hamamatsu H11934-200 PMT (Ultra Bialkali photocathode with quantum efficiency 43% at 400 nm and 25% at 500 nm). The scintillation profiles were acquired with a LeCroy WaveSurf24X-sA oscilloscope (2.5 GSamples/s, 200 MHz) using a time window of 100 µs. A collimated beam of α particles produced by an 241 Am source and mylar absorbers with different thickness were used to repeat the measurement at different α energies. The impinging α energies were preliminarily measured with a CANBERRA Alpha Spectrometer model 7401VR. For each measurement with the ZnWO 4 detector, the energy calibration was determined using 22 Na and 137 Cs γ sources. The measured α/γ ratio as a function of the energy for the three directions of the α beam relatively to the crystal planes is reported in Figure 6 [12].  [12] (colored points) for the three directions of the α beam relatively to the crystal planes and various energies. The values measured in Ref. [49] are also reported for comparison (black points). The anisotropic behaviour of the crystal is evident. The model of Ref. [60] is also reported for each crystal axis. It has been obtained by fitting the data from α's and from recoils or only from α's. Reprinted with permission from [12]. Copyright 2020 Springer Nature.
The data confirm the anisotropic features of ZnWO 4 crystal scintillator in the case of α particles. In particular, the α/γ ratio measured along crystal axis III is about 1.2 times larger than that measured along the axes I and II, where the measured quenching factors are quite similar. The reported uncertainties are mainly due to the evaluation of the impinging alpha energy, which is slightly degraded in air. As shown in Figure 6, the measured quenching factors and anisotropic effect are compatible with those of Ref. [49]. The behaviour of the α/γ ratio as expected for each crystal axis in the model of Ref. [60], based on the semiempirical classical Birks formula [61], is also reported in Figure 6 (see Ref. [12] for details).
The second measurement was carried out at ENEA-Casaccia Laboratory. A schematic view of the experimental set-up is reported in Figure 7.  [12] for details. Reprinted with permission from [12]. Copyright 2020 Springer Nature.
Monochromatic neutrons with energy around 14.7 MeV are generated with a Thermo Scientific portable generator MP 320 through the d(t, α)n reaction. After a single elastic scattering on a target nucleus, the recoiling nucleus is detected by the ZnWO 4 detector, while the scattered neutron can be tagged by one of the two neutron detectors placed at a given scattering angle. When that happens, both the energy and the direction of the recoiling nucleus are determined by the selected scattering angle, while the detected energy in the ZnWO 4 detector, measured in keV electron equivalent (keVee), allows the determination of the quenching factor. Therefore, the quenching factors for nuclear recoils along different crystallographic axes and for different recoil energies can be measured changing the ZnWO 4 detector orientation and the neutron detectors position, respectively.
In this measurement, the ZnWO 4 scintillator was optically coupled to two Hamamatsu H11934-200 PMTs; their signals were recorded by a CAEN DT5720 transient digitizer with 250 MSamples/s. The two neutron detectors, by Scionix, employ an EJ-309 liquid scintillator, which can discriminate neutrons interactions from the γ background by an excellent Pulse Shape Discrimination (PSD) capability. The trigger was generated when a coincidence between the ZnWO 4 detector and the EJ-309 detectors occurred within a time window of ±500 ns. The energy calibration was performed using 133 Ba and 137 Cs sources. The obtained typical energy resolution was σ/E = 4.4% at 662 keV.
The PSD analysis for the selection of neutron events in the EJ-309 detectors was based on the head/tail analysis and on the analysis of the mean time of the pulse. This latter is defined as: τ = ∑ a i · t i / ∑ a i , where a i is the pulse height amplitude at the time t i and the sum is extended over the first 100 ns of the pulse. The used software energy threshold for ZnWO 4 detector was 30 keVee. At this value, only the recoils of oxygen nuclei can be observed. In particular, considering the estimation of the quenching factors of Ref. [60], in these measurements, the Zn and W recoils are expected at energies between 7 keVee and 14 keVee and between 1.6 keVee and 2.5 keVee, respectively. The searched event pattern was given by an event in the ZnWO 4 detector in coincidence with a neutron in the EJ-309 detectors. A clear peak was observed in the ZnWO 4 energy distribution, when a proper time of flight interval was selected for neutrons travelling from the ZnWO 4 to the neutron detectors [12]. The quenching factors were evaluated from the positions of these peaks in the energy scale calibrated with γ quanta. The measurements were repeated studying the nuclear recoils along the three ZnWO 4 crystal axes and with different recoil energies (scattering angle). Since the responses of the crystal axes I and II are expected to be similar, only the case of axis I was considered. The measured quenching factors are shown in Figure 8 along with the models for the considered crystal axes obtained by fitting the α's and oxygen recoil (O) data with the prescription of Ref. [60] (see Ref. [12]).  [12] for oxygen nuclear recoils along the crystal axes I and III as a function of the expected recoil energies E R,O . The models for the considered crystal axes obtained by fitting the α's and oxygen recoil (O) data following the prescription of Ref. [60] are also reported. Reprinted with permission from [12]. Copyright 2020 Springer Nature.
A "mean anisotropy" equal to (1.145 ± 0.027) can be calculated considering the ratio between the quenching factors measured along axes III and I, averaged over the different recoil energies [12]. Thus, the anisotropy in the light response for oxygen nuclear recoils is significantly evident at the 5.4 σ confidence level. Moreover, the trend of the measured quenching factors supports the presence of an anisotropy at a level of ∼15% at 20 keV recoil energy.
In conclusion, the results briefly reported here demonstrate the goodness of anisotropic features of the ZnWO 4 scintillator. Moreover, the high radiopurity level reachable by this detector and its good potential characteristics in terms of light output when operated at a temperature of a few tens of degrees [58] makes it very competitive in the search for rare processes even at low energy. These results open the possibility to realize the pioneer ADAMO experiment and to exploit the directionality technique to investigate those DM candidate particles that induce recoils of target nuclei.
The R&D for further radiopurification of ZnWO 4 crystal scintillators and optimization of the light response is ongoing. In particular, the synthesis of the ZnWO 4 compound from zinc oxides and tungsten oxides of different origin, including commercially available, was carried out and optimized. Optimal synthesis time and temperature were found and the analysis of the synthesized ZnWO 4 compound for each process was done. Many growth processes have been carried out and crystals of single and double crystallizations were grown from the synthesized ZnWO 4 charge from stoichiometric and un-stoichiometric melts. Optimization of ZnWO 4 crystals annealing in an oxygen-containing atmosphere was done and the evaluation of crystal's optical quality dependence on initial Zn and W oxides purity levels, origins, synthesis procedures and stoichiometry was carried out. Further work will be aimed at obtaining larger crystals of high uniformity, optical quality, and free of various inclusions and defects.

SrI 2 (Eu) Crystal Scintillator
The strontium iodide, SrI 2 (Eu), is another inorganic scintillator that has a high amount of potential in the field of DM and rare processes investigation. The strontium iodide scintillator was patented in 1968 by Hofstadter [62]. In the last few years, the high light output (>100,000 photons/MeV), the good energy resolution (≈3% at 662 keV) and the possibility to grow crystals with large-size and uniform light yield have increased the interest in this detector [63][64][65][66]. The main properties of SrI 2 (Eu) crystal scintillators are summarized in Table 1. An important point is that the matrix elements of SrI 2 (Eu) crystal do not have natural long lived radioactive isotopes (an exception is 84 Sr, isotopic abundance 0.56(2)%, which is a double β decay candidate isotope, but this decay has never been observed and the half-life limits are of the order of 10 16 y [13]). It is not the same for other crystal scintillators with high resolution like LaCl 3 (Ce), LaBr 3 (Ce), LuI 3 (Ce), or Lu 2 SiO 5 (Ce), which have long lived radioactive isotopes as 138 La and 176 Lu. Moreover, thanks to the high light output, the SrI 2 (Eu) scintillator has the potential to reach a very low energy threshold. These features make the SrI 2 (Eu) detector promising in various applications, as the low counting experiments for the search of DM and double β decay (the 84 Sr, a double β candidate isotope, is indeed contained in the detector itself).
The internal radioactive contamination of a SrI 2 (Eu) crystal scintillator were studied in Ref. [13]. The used crystal was grown with the vertical Stockbarger method in a quartz ampoule [72]. In particular, the reaction of strontium carbonate and europium oxide with the hydroiodic acid [62] was used to obtain anhydrous strontium iodide activated by europium at 1.2%. The obtained hydrate was dried, placed in a quartz ampoule for the crystal growth and slowly heated with permanent vacuum pumping. When a temperature of 150 • C was reached (after five days), the ampoule was welded and placed in the Stockbarger growing set-up. Then, the temperature was increased up to 538 • C and the crystal was grown with a speed of 20 mm per day [65]. A near to cylindrical crystal 13 × 11 mm was cut from the crystal boule using a dry box filled by pure nitrogen. The crystal was neither milky nor cracked. After wrapping the crystal with PTFE tape, it was encapsulated in an oxygen-free high thermal conductivity (OFHC) copper housing using epoxy glue. A quartz window was present on one side of the housing to allow the collection of the scintillation light. All the materials used in the crystal assembling had a low level of radioactive contamination.
A low background set-up located at sea-level in the Institute for Nuclear Research of NASU (Ukraine) was used to measure the internal radioactive contamination of the SrI 2 (Eu) crystal scintillator. A sketch of the set-up is shown in Figure 9.
The detector was optically coupled to a 3 PMT tube Philips XP2412 through a high purity polystyrene light-guide ( 66 × 120 mm 2 ). A passive shield made of OFHC copper (5-12 cm thick) and lead (5 cm thick) surrounded the detector. Moreover, an anti-muon veto counter (50 × 50 × 8 cm polystyrene based plastic scintillator) was installed above the set-up. The energy spectrum measured with the SrI 2 (Eu) scintillator over 101.52 h together with a model of the background is presented in Figure 10-Left.  (1) is viewed through a polystyrene light-guide (2) by a photo-multiplier tube (3). The detector is surrounded by copper (4) and lead (5). The polystyrene based plastic scintillator of the anti-muon veto counter (6) is viewed by a photo-multiplier tube (7).   3) MeV is due to α activity of trace U/Th contamination of the crystal scintillator. Right: Energy spectra measured by a ULB HPGe γ spectrometer deep underground in Ref. [13]. The measurements were performed with the SrI 2 (Eu) sample (706 h) and without it (BG, 1046 h). The reported γ lines energies are in keV. The BG spectrum was measured before the contamination of the set-up by a 44 Ti (1157 keV peak of 44 Sc). Reprinted with permission from [73]. Copyright 2020 IOP Publishing.
The derived summary on activities is presented in Table 2.
The data analysis also allowed the estimation of the α/γ ratio as 0.55 for the α particles of 214 Po (E α = 7687 keV); moreover, no clear indication on differences was observed in the kinetics of the scintillation decay in the SrI 2 (Eu) crystal scintillator under γ quanta (β particles) and α particles irradiation.
The SrI 2 (Eu) crystal was then measured with the ULB HPGe γ ray spectrometer GeCris (468 cm 3 ) located deep underground at the LNGS. This detector has a rather thin Cu window of 1 mm thickness. The passive shield of the detector consists of 15 cm of OFHC copper and 20 cm of low radioactive lead. The whole set-up is sealed in an air-tight plexiglass box continuously flushed with high purity nitrogen gas to avoid the presence of residual environmental radon. The data accumulated with and without the SrI 2 (Eu) sample are shown in Figure 10-Right. The measured activities are reported in Table 2. Table 2. Radioactive contamination of the SrI 2 (Eu) scintillator used in Ref. [13]. The upper limits are given at 90% C.L., and the uncertainties of the activities at 68% C.L.

Chain
Nuclide The intrinsic radiopurity of the SrI 2 (Eu) scintillator is still far from that of the NaI(Tl) scintillators [18]. On the other hand, it is three orders of magnitude better than that of other high light output scintillators as those containing La, as LaCl 3 (Ce) and LaBr 3 (Ce), and five orders of magnitude better than that of the scintillators containing Lu, as Lu 2 SiO 5 (Ce) and LuI 3 (Ce).
A better radiopurity for the SrI 2 (Eu) crystal can be reached by selecting the starting SrI 2 and EuI 2 materials and applying efficient purification procedures. Moreover, the performances and the ability to grow high quality, uncracked SrI 2 (Eu) crystals also strongly depend on the quality of the starting components. Often, the zone refining is used to enhance the material purity, particularly for the EuI 2 dopant, which cannot be obtained commercially in an ultra-high purity form [74,75].
Another aspect that can affect the light yield and light collection uniformity [15] of SrI 2 (Eu) crystal is the Eu doping concentration [76]. The scintillation properties of pure SrI 2 and SrI 2 doped with different europium concentration were studied e.g., in Refs. [67,69]. In Eu 2+ doped crystals, light-trapping can occur, in which scintillation light is re-absorbed and re-emitted. It complicates the acquisition of high resolution gamma ray spectra but is readily mitigated with a small taper applied to the crystal and/or digital signal processing [66].
As shown in Ref. [77], it is possible to further improve the light output and the energy resolution also lowering the working temperature. In particular, it has been shown that the already excellent scintillation properties are particularly enhanced in the lower temperature range of 130 to 180 K, with a reduction in decay time and increase in light output by approximately 50% and 35%, respectively. In Figure 11, the light output, normalized to its value at 295 K, is shown as a function of temperature. The relative light output increased as the temperature was decreased from room temperature towards 130 K, with the maximum occurring between 130 and 180 K, where the emission intensity was also observed to be the greatest. The increasing light output may be attributed to the reduction in non-radiative processes which typically dominate and reduce luminescence in alkali halides at higher (i.e., ambient) temperatures [77].
In conclusion, the results of these studies demonstrate the applicability of SrI 2 (Eu) detectors to a variety of applications, and show the potential perspective of this highly efficient scintillator for low counting measurements, such as those for the investigation of DM and double β decay of 84 Sr. Figure 11. Evolution of estimated light output as a function of temperature for SrI 2 (Eu) crystal scintillator under 137 Cs gamma-ray excitation, as measured in Ref. [77]. Reprinted with permission from [77]. Copyright 2012 IEEE.

Conclusions
In this paper, some of the important roles and the perspectives of inorganic scintillators in the investigation of DM have been discussed. The many advantages offered by these detectors and the possibility to develop ULB crystals make them particularly suited for the study of model independent DM signatures, as the DM-AM, and for the directionality technique, in the case of detectors with anisotropic response.
In the field of direct DM investigation, three of the most interesting inorganic crystal scintillators are NaI(Tl), ZnWO 4 and SrI 2 (Eu). Their properties, potential and the recent results obtained by DAMA for NaI(Tl) detectors and by the collaboration between DAMA and the Institute for Nuclear Research of NASU (Ukraine), for ZnWO 4 and SrI 2 (Eu) detectors, have been reviewed.
Author Contributions: The authors have been significantly contributing to the conceptualization, and writing of the paper. All authors have read and agreed to the published version of the manuscript.