Energy-Dependent Light Quenching in CaWO 4 Crystals at mK Temperatures
R. Strauss, G. Angloher, A. Bento, C. Bucci, L. Canonica, W. Carli, C. Ciemniak, A. Erb, F.v. Feilitzsch, P. Gorla, A. Gütlein, H. Hagn, D. Hauff, D. Hellgartner, J. Jochum, H. Kraus, J.-C. Lanfranchi, J. Loebell, A. Münster, F. Petricca, W. Potzel, F. Pröbst, F. Reindl, S. Roth, K. Rottler, C. Sailer, K. Schäffner, J. Schieck, S. Scholl, S. Schönert, W. Seidel, M.v. Sivers, L. Stodolsky, C. Strandhagen, A. Tanzke, M. Uffinger, A. Ulrich, I. Usherov, S. Wawoczny, M. Willers, M. Wüstrich, A. Zöller
aa r X i v : . [ a s t r o - ph . I M ] J a n Eur. Phys. J. C manuscript No. (will be inserted by the editor)
Energy-Dependent Light Quenching in CaWO Crystals at mKTemperatures
R. Strauss , G. Angloher , A. Bento , C. Bucci , L. Canonica , W.Carli , C. Ciemniak , A. Erb , F.v. Feilitzsch , P. Gorla , A. G¨utlein , D. Hauff , D. Hellgartner , J. Jochum , H. Kraus , J.-C. Lanfranchi , J. Loebell , A. M¨unster , F. Petricca , W. Potzel , F. Pr¨obst , F.Reindl , S. Roth , K. Rottler , C. Sailer , K. Sch¨affner , J. Schieck ,S. Scholl , S. Sch¨onert , W. Seidel , M.v. Sivers , L. Stodolsky , C.Strandhagen , A. Tanzke , M. Uffinger , A. Ulrich , I. Usherov , S.Wawoczny , M. Willers , M. W¨ustrich , A. Z¨oller Physik-Department, Technische Universit¨at M¨unchen , D-85748 Garching, Germany Max-Planck-Institut f¨ur Physik, D-80805 M¨unchen, Germany CIUC, Departamento de Fisica, Universidade de Coimbra, P3004 516 Coimbra, Portugal INFN, Laboratori Nazionali del Gran Sasso, I-67010 Assergi, Italy Walther-Meißner-Institut f¨ur Tieftemperaturforschung, D-85748 Garching, Germany Physikalisches Institut, Eberhard-Karls-Universit¨at T¨ubingen, D-72076 T¨ubingen, Germany Department of Physics, University of Oxford, Oxford OX1 3RH, United Kingdom Institut f¨ur Hochenergiephysik der ¨Osterreichischen Akademie der Wissenschaften, A-1050 Wien, Austria Maier-Leibnitz-Laboratorium, Ludwig-Maximilians-Universit¨at M¨unchen, D-85748 Garching, GermanyReceived: date / Accepted: date
Abstract
Scintillating CaWO single crystals are apromising multi-element target for rare-event searchesand are currently used in the direct Dark Matter exper-iment CRESST (Cryogenic Rare Event Search with Su-perconducting Thermometers). The relative light out-put of different particle interactions in CaWO is quan-tified by Quenching Factors (QFs). These are essentialfor an active background discrimination and the iden-tification of a possible signal induced by weakly inter-acting massive particles (WIMPs). We present the firstprecise measurements of the QFs of O, Ca and W atmK temperatures by irradiating a cryogenic detectorwith a fast neutron beam. A clear energy dependence ofthe QF of O and, less pronounced, of Ca was observedfor the first time. Furthermore, in CRESST neutron-calibration data a variation of the QFs among differentCaWO single crystals was found. For typical CRESSTdetectors the QFs in the region-of-interest (10-40 keV)are QF ROIO = (11 . ± . QF ROICa = (5 . ± . QF ROIW = (1 . ± . having moderateinfluence on the WIMP analysis. Their relevance forfuture CRESST runs and for the clarification of pre- a e-mail: [email protected] viously published results of direct Dark Matter experi-ments is emphasized. Keywords
Dark Matter · Scintillators · CaWO · Cryogenic detectors · Neutron scattering
Rare-event searches for Dark Matter (DM) in the formof weakly interacting massive particles (WIMPs) [1,2] have reached impressive sensitivities during the lastdecade [3]. Well motivated WIMP candidates withmasses m χ between a few GeV/ c and a few TeV/ c might be detectable via nuclear recoils of few keV interrestrial experiments [4]. While the DAMA/LIBRA[5], and recently the CoGeNT [6], CRESST [7], andthe CDMS(Si) [8] experiments observed excess signalsthat might be interpreted as induced by DM particleswith m χ ∼
10 GeV/ c at WIMP-nucleon cross-sectionsof ∼ − pb, this scenario is ruled out by the LUX [9]and XENON100 [10] experiments, and almost excludedby the CDMS(Ge) [11,12], the EDELWEISS [13,14]and the SuperCDMS [15] experiments. It is stronglydisfavoured by accelerator constraints [16,17] and inmild tension with an extended analysis [18] of publishedCRESST data [19]. The CRESST experiment [7] employs scintillatingCaWO crystals [20,21] as a multi-element target ma-terial. The key feature of a CRESST detector moduleis the simultaneous measurement of the recoil energy E r by a particle interaction in the crystal (operatedas cryogenic calorimeter at mK temperatures [22]) andthe corresponding scintillation-light energy E l by a sep-arate cryogenic light absorber. Since the relative lightyield LY = E l /E r is reduced for highly ionizing particlescompared to electron recoils (commonly referred to asquenching) nuclear-recoil events can be discriminatedfrom e − / γ and α backgrounds. The phenomenologicalBirks model [23] predicts this quenching effect to bestronger the higher the mass number A of the recoilingion, which allows to distinguish, in general, betweenO ( A ≈ A ≈
40) and W ( A ≈ m χ &
20 GeV/c . However, the light targets O and Camake CRESST detectors particularly sensitive to low-mass WIMPs of 1 GeV . m χ .
20 GeV. Furthermore,the knowledge of the recoil composition of O, Ca and Wallows a test of the assumed A -dependence of the spin-independent WIMP-nucleon cross-section [2]. In addi-tion, background neutrons, which are mainly visible asO-scatters (from kinematics [24]), can be discriminatedstatistically. The mean LY of e − / γ events ( LY γ ) is energydependent and phenomenologically parametrized as LY γ ( E r ) = ( p + p E r )(1 − p exp( − E r /p )) [25]. Byconvention, LY γ (122 keV) is normalized to unity. Theparameters p , p , p and p are derived from a maxi-mum likelihood (ML) fit for every detector module indi-vidually. For the module used in this work the fit yields: p = 1 . p = − . · − keV − , p = 6 . · − and p = 147 keV (errors are negligible for the following anal-ysis). The exponential decrease towards lower recoil en-ergies (quantified by p and p ) accounts for the scin-tillator non-proportionality [26]. The Quenching Fac-tor (QF) of a nucleus x - in general energy dependent- is defined as QF x ( E r ) = LY x ( E r ) /LY γ,norm where LY x is the mean LY of a nuclear recoil x and LY γ,norm is a detector-specific normalization factor which corre-sponds to the LY of e − / γ events. By convention, LY γ,norm = LY γ ( E r ) / (1 − p exp( − E r /p )) ≈ p is usedfor the analysis since the scintillator non-proportionalityis not observed for nuclear recoils and p ≪
1. For typ-ical CRESST detector modules, the uncertainties in en- ergy and LY are well described by gaussians [7] consis-tent with photon-counting statistics in the energy rangeconsidered in this work.Since the resolution of light-detectors operated in theCRESST setup at present is not sufficient to disentangleO, Ca and W recoils unambiguously, dedicated exper-iments to measure the QFs of CaWO are necessary.Earlier attempts yield inconclusive results, in particu-lar for the value of QF W [27,28,29]. B beam of ∼
65 MeV in bunches of 2-3 ns (FWHM) produces mo-noenergetic neutrons of ∼
11 MeV via the nuclear reac-tion p( B,n) C in a pressurized H target [30]. Theseneutrons are irradiated onto a CRESST-like detectormodule consisting of a ∼
10 g cylindrical CaWO sin-gle crystal (20 mm in diameter, 5 mm in height) and aseparated Si light absorber (20 mm in diameter, 500 µ mthick) [31]. Both are operated as cryogenic detectors ina dilution refrigerator at ∼
20 mK [32]. Undergoing elas-tic (single) nuclear scattering in CaWO the neutronsare tagged at a fixed scattering angle Θ in an array of40 liquid-scintillator (EJ301) detectors which allow fasttiming ( ∼ γ discrimination.4.2 Working principleDepending upon which of the three nuclei is hit a dis-tinct amount of energy is deposited by the neutronin the crystal. Triple-coincidences between (1) a Bpulse on the H target, (2) a neutron pulse in a liquid-scintillator detector and (3) a nuclear-recoil event inthe CaWO crystal can be extracted from the data set.A neutron time-of-flight (TOF) measurement betweenneutron production and detection combined with a pre-cise phononic measurement of the energy depositionin the crystal (resolution ∼ ∼ µ s) compared to typical neutron TOFs ( ∼
50 ns) anoffline coincidence analysis has to be performed [25].
Fig. 1
Schematic experimental setup of the neutron-scattering facility. Neutrons produced by the accelerator arescattered off a CRESST-like detector module (operated at20 mK) and tagged in liquid-scintillator neutron detectors ata fixed scattering angle Θ . t (ms) ∆ -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 e v e n t s / ( . m s ) Fig. 2
Histogram of the time difference ∆t between neu-tron events with the correct TOF and the closest W-recoil inthe CaWO crystal ( E r = 100 ±
20 keV). A fit to the distri-bution (solid black line) including a constant for the acciden-tal background (shaded area) and a gaussian for the triple-coincidences on W (dashed red line) is shown. 158 W-scattersare identified with a signal-to-background ratio of ∼ QF W The experiment was optimized for the measurement of QF W [25,33]. To enhance the number of W-scatters ascattering angle of Θ = 80 ◦ was chosen due to scatter-ing kinematics [27]. For this specific angle, the expectedrecoil energy of triple-coincident events is ∼
100 keV forW, ∼
450 keV for Ca, and ∼ . ∼ ∼ cryodetector pulses wererecorded. Fig. 2 shows the time difference ∆t betweenneutron events with the correct TOF identified in one ofthe liquid-scintillator detectors and the closest W-recoil(in time) in the CaWO crystal ( E r = 100 ±
20 keV). Agaussian peak of triple-coincidences on W (dashed redline) at ∆t ≈ .
016 ms and a width of σ t ≈ . µ s (onsetresolution of the cryodetector) is observed above a back-ground due to accidental coincidences uniformly dis-tributed in time (shaded area). Within the 2 σ -boundsof the peak 158 W-scatters are identified with a signal-to-background (S/B) ratio of ∼ LY -0.1 -0.05 0 0.05 0.1 0.15 0.2 e v e n t s / ( . ) Fig. 3
LY histogram of the 158 events identified as triple-coincidences on W. A fit to the distribution (solid black line)is shown which includes a gaussian (dashed red line) account-ing for W-scatters and the background-pdf (shaded area) de-scribing accidental coincidences. The simultaneous ML fit in-cluding the timing distribution yields QF W = (1 . ± . W-scatters is found at a lower value compared to themean LY of all nuclear recoils, i.e., the (overlapping)contributions of O, Ca and W if no coincidence mea-surement is involved. The accidental coincidences havea LY-distribution equal to that which is modelled bya probability-density function (background-pdf) [25].A simultaneous maximum-likelihood (ML) fit is per-formed including (1) the timing distribution which fixesthe S/B ratio and the number of identified W-events,and (2) the LY distribution described by a gaussian(W-events) and the background-pdf. The final resultsare LY W = 0 . ± . QF W = (1 . ± . QF Ca and QF O no coinci-dence signals are necessary, instead, an analysis of thenuclear-recoil data alone is sufficient. Also for this anal-ysis the neutron data obtained at the scattering facilitywere used. Commonly CRESST data is displayed inthe energy-LY plane [7] giving rise to nearly horizontalbands which correspond to different types of particle in-teractions ( LY ≈ LY . . ∼ ∼ · pulses) areshown in Fig. 4 (2-d histogram). From kinematics using ∼
11 MeV neutrons as probes the O-recoil band extendsup to ∼ . ∼ .
05 MeV and ∼
240 keV, respectively [30]. De-
Fig. 4
Histogram of neutron-induced nuclear-recoil eventsplotted in the LY-energy plane. The corresponding 1 σ accep-tance bounds (full red lines) of O, Ca, and W as derived fromthe correlated ML fit (see text) are indicated. e v e n t s / ( . ) datafit = 350keV r E Ca O LY -0.1 0 0.1 0.205001000 datafit = 40 keV r E Ca O W Fig. 5
LY histograms of energy slices (20 keV in width) at350 keV (top) and 40 keV (bottom) fitted by gaussians. spite the strong overlap of the 3 nuclear-recoil bandsthe contributions of O and Ca fitted by two gaussianscan be disentangled at E r &
350 keV (see Fig. 5 top) dueto high statistics and a good light-detector resolution.5.2 Results and discussionIn Fig. 6 the results for QF O and QF Ca (red error bars)derived by these independent one-dimensional (1-dim)fits are shown for selected recoil-energy slices of 20 keVin width. All parameters in the fit are left free except forthe LY-resolutions which are fixed by a ML fit of theelectron-recoil band [25]. While QF O clearly rises to-wards lower recoil energies, this effect is less pronouncedfor QF Ca .Below ∼
350 keV, due to the strong overlap of the nu-clear recoil bands, this simple approach fails. Instead, acorrelated ML fit was performed based on thefollowing assumptions: (1) for the mean LY of O- andCa-scatters the phenomenological parametrization LY x ( E r ) = LY ∞ x (1 + f x · exp ( − E r /λ x )) is proposedwith the free parameters LY ∞ x (LY at E r = ∞ ), f x (fraction of energy-dependent component) and λ x (ex- Table 1
Results for the free parameters LY ∞ x , f x and λ x ofthe ML analysis. The statistical errors are given at 1 σ C.L. LY ∞ x f x λ x O 0 . ± . . ± . . ± . . ± . . ± . . ± . ponential decay with energy), and (2) the mean LY ofW-scatters is approximated to be constant in the rele-vant energy range (up to ∼
240 keV) at the value pre-cisely measured with the triple-coincidence technique( LY W = 0 . ± . QF W =(1 . ± . A . The nuclear-recoil bands are cut into energy inter-vals of 10 keV (20 keV to 1 MeV), of 20 keV (1 MeV to1.4 MeV) and 50 keV (above 1.4 MeV) and fitted withup to 3 gaussians depending on the recoil energy (e.g.,shown in Fig. 5 bottom for E r =40 keV). Except for theassumptions mentioned above and the LY-resolution allparameters are left free in the fit. The fit converges overthe entire energy range (20-1800 keV). In Table 1 theresults for LY ∞ x , f x and λ x are presented which corre-spond, e.g. at 40 keV, to QF O =(12 . ± . QF Ca =(6 . ± . σ C.L. Errors are dominated by sys-tematics including different choices of the LY parametri-zation. The final results for QF O , QF Ca and QF W arepresented in Fig. 6 and are found to be in perfect agree-ment with the outcome of the 1-dim fits (red error bars).Fig. 4 shows the 1 σ acceptance bounds (full red lines)of O, Ca and W recoils as obtained in the correlatedML fit.These are the first experimental results which clearlyshow a rise of QF O of ∼
28% towards the ROI (10-40 keV) compared to that at a recoil energy of 500 keV.For QF Ca the best fit yields a rise of ∼ were assumedto be constant over the entire energy range [7]. A sta-tistical analysis shows that this simple model is clearlydisfavoured. Employing a likelihood-ratio test in combi-nation with Monte-Carlo simulations gives a p-value of p < − for the data presented here to be consistentwith constant QFs. Furthermore, the derived energyspectra of the individual recoiling nuclei agree with theexpectation from incident 11 MeV neutrons while theconstant QF approach provides non-physical results. E r (keV) Q F Ca OW best fit (correlated) 1 σ σ σ ) Fig. 6
Results of the correlated ML analysis for QF O , QF Ca and QF W (solid lines). The shaded areas indicate the 1 σ and2 σ bounds. For the first time a clear energy dependence of QF O and QF Ca is observed. These results are in agreementwith that of the 1-dim fits of discrete energy intervals (seetext) shown as red error bars. QF W is fixed (in the correlatedfit) at the value measured by the triple-coincidence technique. In the present paper, using the 8 detector modules oper-ated in the last CRESST measurement campaign (run32)an additional aspect was investigated: the variation ofthe quenching behaviour among different
CaWO crys-tals [25]. Nuclear recoils acquired during neutron cal-ibration campaigns of CRESST run32 are completelydominated by O-scatters at E r &
150 keV (from kine-matics) [7]. Despite low statistics (a factor of ∼
100 lesscompared to the measurement presented here) in theavailable data, the mean LY of O-events can be deter-mined by a gaussian fit with a precision of O (1%) forevery module. In this way, the mean QF of O between150 and 200 keV was determined individually for the 8detector modules (index i ) operated in run32 ( QF ∗ O,i )and for the reference detector operated at the neutron-scattering facility ( QF O ). Different values of QF ∗ O,i areobserved for the CRESST detector crystals (variationby ∼ ∼
12% higherthan the mean of QF ∗ O,i ). This variation appears to becorrelated with the crystal’s optical quality. The QF- which is a relative quantity - is found to be lowerif a crystal has a smaller defect density and thus ahigher absolute light output, i.e., the LY of nuclear re-coils is less affected by an increased defect density. Thisis in agreement with the prediction described in a re-cent work [34]. In the present paper, a simple modelto account for this variation is proposed: For every de-tector module which is to be calibrated a scaling fac-tor ǫ i is introduced, ǫ i = QF ∗ O,i /QF O . Then, within thismodel the QFs of the nucleus x can be calculated forevery module by QF ∗ x,i ( E r ) = ǫ i · QF x ( E r ) where QF x Table 2
QF results averaged over the ROI (10-40 keV) andadjusted by the scaling factor ǫ i for the modules Rita andDaisy, and the mean (Ø) of all run32 detectors (1 σ errors). ǫ i QF ROI O [%] QF ROI Ca [%] QF ROI W [%]Rita 0 .
844 10 . ± . . ± .
44 1 . ± . .
939 12 . ± . . ± .
58 1 . ± .
24Ø 0 .
880 11 . ± . . ± .
49 1 . ± . is the value precisely measured within this work. Thenuclear-recoil behaviour of CRESST modules is welldescribed by energy-dependent QFs. In Table 2 theQFs, averaged over the ROI (10-40 keV), and the scal-ing factor ǫ i are listed for two selected detector mod-ules (Rita and Daisy, with the lowest and highest ab-solute light output, respectively) and the mean of all 8detector modules of run32 (Ø), QF ROIO = (11 . ± . QF ROICa = (5 . ± . QF ROIW = (1 . ± . We now turn to the effect of energy-dependent quench-ing since constant QFs as assumed in earlier CRESSTpublications do not sufficiently describe the behaviourof the nuclear-recoil bands. The value of QF O in theROI was underestimated by ∼
8% while the room tem-perature measurements overestimated the values of QF Ca and QF W by ∼
7% and ∼ ∼ . σ ) above known back-grounds was observed. If interpreted as induced by DMparticles two WIMP solutions were found [7], e.g. at amass of m χ = 11 . with a WIMP-nucleon crosssection of σ χ = 3 . · − pb. The dedicated ML analy-sis was repeated using the new QF values (Ø in Table2) yielding m χ = 12 . and σ χ = 3 . · − pbat 3.9 σ . Beside this moderate change of the WIMP pa-rameters also the background composition ( e − , γ , neu-trons, α ’s and Pb) is influenced. This is mainly dueto the significantly lower value of QF W which increasesthe leakage of Pb recoils into the ROI (by ∼ m χ changes from 25.3 to 25.5 GeV/c , σ χ from 1 . · − to 1 . · − pb and the significance drops slightly from4.7 to 4.3 σ . In conclusion, the first precise measurement of QF W at mK temperatures and under conditions comparableto that of the CRESST experiment was obtained atthe neutron-scattering facility in Garching by an ex-tensive triple-coincidence technique. Furthermore, theQFs of O and Ca were precisely determined by a ded-icated maximum-likelihood analysis over the entire en-ergy range ( ∼ − crystals was observed which is relatedto the optical quality. By the simple model proposedabove the measured QFs can be adapted to every in-dividual crystal. The updated values of the QFs arehighly relevant to disentangle the recoil composition(O, Ca and W) of a possible DM signal and, there-fore, to determine the WIMP parameters. Since theseparation between the O and W recoil bands is higherby ∼
46% compared to earlier assumptions, backgroundneutrons which are mainly visible as O-scatters [24] canbe discriminated more efficiently from possible WIMP-induced events. A reanalysis of the run32 data shows amoderate influence of the new QF values on the WIMPparameters.The results obtained here are of importance for the cur-rent CRESST run (run33) and upcoming measuringcampaigns. Providing a highly improved backgroundlevel run33 has the potential to clarify the origin ofthe observed excess signal and to set competitive limitsfor the spin-independent WIMP-nucleon cross sectionin the near future.For the planned multi-material DM experiment EU-RECA (European Underground Rare Event Calorime-ter Array) [35] the neutron-scattering facility will bean important tool to investigate the light quenching ofalternative target materials in the future.
Acknowledgements
This research was supported by theDFG cluster of excellence: Origin and Structure of the Uni-verse, the DFG Transregio 27: Neutrinos and Beyond, theHelmholtz Alliance for Astroparticle Phyiscs, the Maier Leib-nitz Laboratorium (Garching) and by the BMBF: Project05A11WOC EURECA-XENON.
References
1. G. Bertone, D. Hooper, J. Silk, Phys.Rept. , 279(2005). DOI 10.1016/j.physrep.2004.08.0312. G. Jungman, M. Kamionkowski, K. Griest, Phys.Rept. , 195 (1996). DOI 10.1016/0370-1573(95)00058-5 3. P. Cushman, et al., (2013). URL http://arxiv.org/abs/arXiv:1310.8327
4. J. Lewin, P. Smith, Astropart.Phys. , 87 (1996). DOI10.1016/S0927-6505(96)00047-35. R. Bernabei, et al., Eur.Phys.J. C67 , 39 (2010). DOI10.1140/epjc/s10052-010-1303-96. C. Aalseth, et al., Phys.Rev.Lett. , 131301 (2011).DOI 10.1103/PhysRevLett.106.1313017. G. Angloher, et al., Eur.Phys.J.
C72 (4), 1 (2012). URL http://dx.doi.org/10.1140/epjc/s10052-012-1971-8
8. R. Agnese, et al., Phys. Rev. Lett. , 251301 (2013).DOI 10.1103/PhysRevLett.111.2513019. D. Akerib, et al., Phys.Rev.Lett. , 091303 (2014).DOI 10.1103/PhysRevLett.112.09130310. E. Aprile, et al., Phys.Rev.Lett. , 181301 (2012). DOI10.1103/PhysRevLett.109.18130111. Z. Ahmed, et al., Science , 1619 (2010). DOI 10.1126/science.118611212. Z. Ahmed, et al., Phys.Rev.Lett. , 131302 (2011).DOI 10.1103/PhysRevLett.106.13130213. E. Armengaud, et al., Phys.Lett.
B702 , 329 (2011). DOI10.1016/j.physletb.2011.07.03414. E. Armengaud, et al., Phys.Rev.
D86 , 051701 (2012).DOI 10.1103/PhysRevD.86.05170115. R. Agnese, et al., Phys.Rev.Lett. , 241302 (2014).DOI 10.1103/PhysRevLett.112.24130216. G. Aad, et al., JHEP , 075 (2013). DOI 10.1007/JHEP04(2013)07517. S. Chatrchyan, et al., JHEP , 094 (2012). DOI10.1007/JHEP09(2012)09418. A. Brown, S. Henry, H. Kraus, C. McCabe, Phys. Rev.D , 021301 (2012). DOI 10.1103/PhysRevD.85.02130119. G. Angloher, et al., Astropart.Phys. (4), 270 (2009)20. T.A. Edison, Nature , 470 (1896)21. V.B. Mikhailik, H. Kraus, S. Henry, A.J.B.Tolhurst, Phys. Rev. B , 184308 (2007).DOI 10.1103/PhysRevB.75.184308. URL http://link.aps.org/doi/10.1103/PhysRevB.75.184308
22. F. Pr¨obst, et al., J. Low Temp. Phys. (1-2), 69 (1995). DOI 10.1007/BF00753837. URL http://dx.doi.org/10.1007/BF00753837
23. J. Birks,
The Theory and Practice of Scintilla-tion Counting (Pergamon Press, 1964). URL http://books.google.com/books?id=0RlRAAAAMAAJ
24. S. Scholl, J. Jochum, J. Phys.: Confer-ence Series (1), 012020 (2012). URL http://stacks.iop.org/1742-6596/375/i=1/a=012020
25. R. Strauss, Ph.D. thesis, TU M¨unchen (2013). URL http://mediatum.ub.tum.de/node?id=1166886
26. R. Lang, et al., (2009). URL http://arxiv.org/abs/0910.4414
27. T. Jagemann, et al., Astropart.Phys. , 269 (2006). DOI10.1016/j.astropartphys.2006.06.01028. J. Ninkovic, et al., Nucl.Instrum.Meth. A564 , 567(2006). DOI 10.1016/j.nima.2006.04.03929. I. Bavykina, et al., Astropart.Phys. , 489 (2007). DOI10.1016/j.astropartphys.2007.09.00630. T. Jagemann, J. Jochum, F. Feilitzsch,Nucl.Instrum.Meth. A551 (2–3), 245 (2005)31. R. Strauss, et al., J. Low Temp. Phys. (5-6),1063 (2012). DOI 10.1007/s10909-012-0536-4. URL http://dx.doi.org/10.1007/s10909-012-0536-4
32. J.C. Lanfranchi, et al., Opt. Mat. (10), 1405 (2009)33. R. Strauss, et al., J. Low Temp. Phys. pp. 1–6 (2014). DOI 10.1007/s10909-013-1075-3. URL http://dx.doi.org/10.1007/s10909-013-1075-3
34. S. Roth, Ph.D. thesis, TU M¨unchen (2013). URL https://mediatum.ub.tum.de/node?id=1172019
35. G. Angloher, et al., Physics of the Dark Universe3