Pulse Shape Discrimination in liquid argon and its implications for Dark Matter searches using depleted argon
PPulse Shape Discrimination in Liquid Argon and itsImplications for Dark Matter Searches Using DepletedArgon ∗ Pawel Kryczynski on behalf of the WArP R&D Group
The Henryk Niewodniczanski Institute of Nuclear Physics Polish Academy ofSciences, [email protected] brief outline of Dark Matter detection experiments using liquid Ar-gon technology is presented. The Pulse Shape background discriminationmethod (PSD) is described and the example of its use in 2.3 l R& D detec-tor is given. Methods of calculating sensitivity of a Dark Matter detectorare discussed and used to estimate the possible improvement of sensitivityafter introduction of isotopically depleted liquid Argon.PACS numbers: 95.35.+d,61.25.Bi, 25.40.Fq, 42.79.Pw
1. Dark Matter Detection
Since the beginning of the 20th century the number of experimentalfacts suggesting the insufficiency of using solely luminous matter in thedescription of the Universe has been growing. Several measurements haveshown that the baryonic matter provides only the 4.5% of the mass - energyof the Universe. Of the remainder: about 73% is associated with DarkEnergy (Cosmological Constant) and over 22% with Dark Matter [1].One of the prominent observations suggesting the existence of DM wasthe measurement of the shape of the rotation curves of galaxies. Thestars rotate around the center of the galaxy with such velocities as if themass was distributed in an uniform way, and not in the central bulge andthe surrounding disk. The observation of so called ”polar ring galaxies” [4](with additional stars rotating perpendicularly to the main disk) leads to asimilar conclusion. ∗ Presented at the Epiphany 2012 Conference (1) a r X i v : . [ phy s i c s . i n s - d e t ] O c t v43p1509 printed on August 30, 2018 Fig. 1. One of the evidences for the Dark Matter existence: the Bullet Cluster [2].Fig. 2. Another important evidence for the Dark Matter existence: the map of theCMB distribution from the WMAP experiment [3].
Similar conclusions can be drawn from the observations of objects suchas the Bullet Cluster (Figure 1 right) where the movement of the luminousparts does not trace that of the main mass component of the colliding clus-ters as measured by weak lensing. The measurements of Cosmic MicrowaveBackground Radiation (CMB), Figure 2, also support the hypothesis ofDark Matter. The anisotropies in the CMB can help calculate of the per-centage of the baryonic matter in the Universe and the share of the DarkMatter in its total mass.Today, most astrophysicists agree that Dark Matter exists - it is its na-ture that is the question. The number of candidate particles is great. Thesecan be divided into warm DM (the relativistic particles which freezed out atthe early stages of the Universe formation), which will not be discussed here,and the cold DM which includes particles with non - relativistic velocitiesand yet not known form. The last is included in the preferred cosmolog-ical model - ΛCDM. The most popular and promising DM candidate is aWIMP (a Weakly Interacting Massive Particle). Its allowed mass range ex- tends from tens of GeV to above 1 TeV [6]. The excluded cross sections areof over 10 − cm for spin independent interactions and over 10 − cm forspin - dependent interactions [5]. As the velocities are non relativistic, theexpected energies of recoiling nuclei in a detector are of the order of tensof keV [7]. Candidates for WIMPs may be supersymmetric particles (e.gneutralino). The expected event rate is very low - 0.02 events/day/kg ofdetector mass. The issue of Dark Matter detection has to be dealt with byemploying efficient detecting techniques and and reducing the backgroundto unprecedented levels. The methods of Dark Matter detection can be divided into indirectand direct. The former uses products of annihilation or decay of the DMparticles and is not discussed here, whereas the latter is connected withthe observation of DM interactions in a detector. The transfered energycan be detected as ionization of the detector medium, heat in the crystallattice of the detector or scintillation (examples of experiments are given inFigure 3). Typically, a combination of two of the mentioned signals is usedto obtain a better background discrimination. Cryogenic detectors usingGe or Si crystals (CDMS) detect ionization and heat, detectors based onCaWO use the scintillation signal and heat and liquid noble gas detectors- ionization (when working in double phase mode) and scintillation. Fig. 3. Classification of experiments according to their detection techniques [8] .
2. Liquid Argon as a Dark Matter detector
The analysis presented in this work uses some of the methods developedin [9] and expands on the results presented in [10] and as such, will focus on v43p1509 printed on August 30, 2018 liquid Argon detectors. A typical double-phase Argon DM detector consistsof a dewar filled with liquid argon with photomultipliers observing a fiducialvolume. Over the liquid, there is a layer of gaseous Ar.
Fig. 4. A scheme of a LAr DM detector (based on [12]).
The detector can work in single phase mode, registering only the lightsignal produced in the liquid (S1). In this detection mode the only sig-nal which is measured is the scintillation. When a DM particle hits Argonnucleus, it may become excited or ionized and form a dimer Ar . The relax-ation of the excited dimer may be of two forms - singlet (fast signal) withdecay time about 7 ns and triplet (slow signal) with decay time 1500 ns.Such a significant difference (absent in Xenon, also used in DM detection) of-fers a background discrimination possibility because electrons produce slowand fast components in a different proportion than neutrons (and WIMPs).This results in different signal shapes. When the double phase mode isused, the ionization caused by the recoiling nuclei is extracted to the gaslayer above the liquid by the electrodes situated below and over the surfaceto cause the secondary signal (S2) proportional to the ionization (Figure 4). Using S1 and S2, the difference in the ratio of relative amplitudes ofsignals resulting from electron-like events (main background) and neutron(or WIMP)-like events is an additional discrimination method, resulting ina better rejection of background events (Figure 5). The background in DM detectors can be divided into two subclasses:external and intrinsic. Both are caused mainly by γ rays and neutrons.The effects of the background resulting from the cosmic rays (dominatingduring tests performed on the earth surface) and the natural radioactivityof the rocks (in underground laboratories), can be reduced by precise mea- surements and simulations leading to determination of their signature aswell as by using shielding.The intrinsic background connected with the construction of the detectorcan be diminished by using radiopurified materials, but the most importantpart of it is related to the radioactive isotopes Ar and Ar present in at-mospheric Argon along with non radioactive Ar. Produced in interactionof cosmic rays or due to the neutron capture, these isotopes contribute tothe background due to β − decays [13]. The activity of Ar dominates andwas measured to be 1.01 ± ± F P = S F P S = (cid:82) T FP T i V ( t ) (cid:82) T F T i V ( t ) (2.1)where T i and T F are the limits of the time window for signal V ( t ) and T F P is the time which ensures the best separation of two populations - 100 nsfor analysis of 2011 data and 120 ns for the data from 2009 [10]. The WArP 2.3 liter prototype detector (Figure 7 left) was the firstto report DM search results in liquid Argon [15] and has been since usedfor R &D purposes. In 2009 the WArP collaboration measured the n -likeev./ e − -like ev. separation by irradiating a chamber with an Am/Be source.In the FPrompt distributions calculated using this data, an intermediate v43p1509 printed on August 30, 2018 population was found. It was connected with the inelastic scattering ofneutrons and included in the model which was then fitted to the spectrum.Indeed, such scattering dominates for energies of the particles emitted bythe Am/Be source (Figure 6). The data suffered from a low light yield anddue to some DAQ problems it was not possible to subtract the background.After a detector upgrade, we decided to repeat this measurement profitingfrom a four times higher LY. This upgrade, performed in 2010, is described indetail in [16] and [17]. New photomultipliers with higher quantum efficiencywere installed, which led to a higher light yield (6.1 phel/keV compared to1.52 phel/keV in 2009 runs). The consequence of this was an insight intoa region of lower recoil energies. The new DAQ electronics with a broaderdynamic range, causing less saturated events, made the exploration of higherenergies possible (Figure 7 right). Fig. 6. The spectrum of the Am/Be source (left) [11] and references therein andthe cross section for neutron interaction on argon nuclei (right) [9] and referencestherein.Fig. 7. The the detector scheme (left) [18] and a difference of the energy ranges in2009 and 2011 data (right). In 2011, runs were performed to test the new detector setup. The Am/Beneutron source was used to imitate the WIMP interactions and the
Amto perform detector calibration and to estimate the light yield. The resultsfrom 2009 were used as a reference. The FPrompt spectrum was fitted witha sum of two Gaussian functions (for gamma and neutron population) anda convolution of an exponential and Gaussian function (for the intermediate- inelastic neutron interaction population): G n (cid:77) ( G i (cid:79) E i ) (cid:77) G γ (2.2)where the G n and G γ are the Gaussian distributions for a neutron andgamma population respectively and G i (cid:78) E i is the convolution. One of thepurposes of this analysis was to test whether the intermediate populationwould be observed (one of the energy bins is presented in Figure 8). Fig. 8. The FPrompt spectrum with an intermediate population visible ( this anal-ysis results).
The positions of the gamma and neutron populations resulting fromthese fits are compared with the previous measurements and published re-sults from other groups ([10], [19]) in Figure 9.In order to understand the small discrepancy between the new data andthe 2009 results, a re-analysis of the 2009 data with the use of new scriptsand reconstruction procedures was performed. The preliminary results arepresented in Figures 10 and 11. Figure 11 presents a comparison of thenumber of elastic neutron interactions per unit of energy to the numberexpected from Monte Carlo simulations. v43p1509 printed on August 30, 2018
Fig. 9. Comparison of the FPrompt spectrum for the 2011 and 2009 WArP data[9] and the results published by Lippincott et. al [19].Fig. 10. The results of the re-analysis of the 2009 data with the new reconstruction.
3. Predicted Sensitivity for a Depleted Argon Detector
Even if a DM detector sees no signal, it is still possible to set the ex-clusion limits on the allowed σ vs WIMP mass parameter space. Typically,the model accounting for the Earth galaxy motion, the velocity distributionof WIMPs in the galactic halo, seasonal DM flux change and detector effi-ciency is applied, as described e.g in [7]. The resulting function predictingthe number of observed events as a function of recoil energy is integratedtaking the detector threshold into account. Repeating this procedure for allWIMP masses and taking into account the detector exposure leads to thesensitivity plot. The tools using this approach were prepared and success-fully tested on published data. Another way to calculate detector sensitivitywas proposed by S. Yellin [20] and is based on the distribution of events as a function of their energy. This was also tested but not used in presentedanalysis. One of the crucial parameters in calculating the sensitivity of a DMdetector is its exposure. Although the sensitivity should increase with theexposure, it is limited due to the presence of radioactive Ar isotope. Itsabundance causes the pile - up of the signal and leakage of the gammapopulation (background) into the signal region. The isotopically depletedArgon seems to be a promising solution. The abundance of the Ar can bereduced by depletion in centrifuges but costs of such production method arehigh. The other approach is to search for underground Argon, which maybe found e.g in reservoirs of Helium and then distilled. Using this method,a depletion factor of 25 has been confirmed [21].
The method used in this work is based on the known Ar energy spec-trum and radioactivity. The maximum exposure before the backgroundevents start to ”leak” into the conservatively chosen signal region is calcu-lated. To do this, the events are randomly chosen from the analysis energyrange and according to the known parameters of the Gaussian function inthe corresponding bin the number of entries in the signal window is esti-mated. This, thanks to the known radioactivity of Ar, helps to estimatethe possible exposure (Figure 12). v43p1509 printed on August 30, 2018 Fig. 12. The method used to estimate exposure (left) and the Ar spectrum withthe analysis range marked in green (right).
The calculations of the predicted exposure have been performed assum-ing Ar to be the only background. The results from XENON (2011) wereused as the reference and are presented in Figure 13 ([22], [23]). The casewith the S2/S1 discrimination applied (with conservative assumption of thebackground rejection power of 10 ) is displayed with a green curve and thecase with a depleted Argon (by factor 25) - with a blue curve. Fig. 13. The comparison of sensitivity obtainable in a liquid argon detector with(blue) and without (green) the depleted Argon. Results from XENON 100(2011)[23] were used as the reference (red curve) and plotted using tool [22]. Other curvesare obtained using tools prepared for this analysis.
4. Conclusions
The preliminary results for the analysis of the data obtained with theupgraded detector setup were presented. The insight into a wider energyrange has been reported. The comparison with the data from 2009 has beenperformed. The influence of the possible use of the isotopically depletedArgon has been demonstrated and the significant possible improvement ofsensitivity has been seen.
Acknowledgements
I would like to thank the WArP R &D group for providing me with thedata used in this work and for the help with my analysis. In particular, Iwould like to thank A. Szelc both for the support during the data analysisand for correcting this article. I want also to thank A. Zalewska for readingand correcting this work. REFERENCES [1] K. Nakamura et al. (Particle Data Group), J. Phys. G 37, 075021 (2010) and2011 partial update for the 2012 edition.[2] S. W. Randall et al.
Constraints on the Self-Interaction Cross Section of DarkMatter from Numerical Simulations of the Merging Galaxy Cluster 1E 0657-5 ,[arXiv:0704.0261 [astro-ph]][3]
NASA / WMAP Science Team, Seven Year Microwave Sky ,http://map.gsfc.nasa.gov/media/101080/, accessed 02.09.2011[4] E. Battaner, E. Florido, Fund. Cosmic Phys. 21 (2000) 1-154. [astro-ph/0010475][5] P. J. Fox, R. Harnik, R. Primulando and C. -T. Yu,
Taking a Razor to DarkMatter Parameter Space at the LHC , arXiv:1203.1662 [hep-ph].[6] David G. Cedreno,
WIMPs: a Brief Bestiary , in Proceedings of the 4th PatrasWorkshop on Axions, WIMPs and WISPs June 18-21, 2008 Hamburg, Germany[7] J.D.Lewin, P.F.Smith, Astropart. Phys. 6 , (1996) 87-112.[8] A.M. Szelc, Acta Phys.Polonica B37 (2006) 1997-2004[9] Roberto Acciarri, Nicola Canci, Flavio Cavanna, Pawel Kryczynski, AndrzejM. Szelc,WArP R & D:
Neutron to Gamma Pulse Shape Discrimination inLiquid Argon Detectors with HQE PMTs , Presented at TIPP 2011 InternationalConference, Chicago[10] R.Acciarri, PhD. thesis,
Measurement of the scintillation time spectra andPulse Shape Discrimination of low-energy electron and nuclear recoils in liquidArgon with the WArP 2.3 lt detector , Universita Degli Studi DellAquila, 20102 v43p1509 printed on August 30, 2018 [11] L. Pandola, The WArP Dark Matter Experiment Document 56-v1.[12] R. Bernabei et. al,
Liquid Noble gases for Dark Matter searches: a synopticsurvey. , arXiv:0806.0011v2 [astro-ph], 2009[13] P. Cennini et. al, Nuclear Instruments and Methods in Physics Research A356 (1995) 526-529[14] P. Benetti et.al, Nuclear Instruments and Methods in Physics Research A 574(2007) 8388[15] P.Benetti et al.,
First results from a Dark Matter search with liquid Argon at87 K in the Gran Sasso Underground Laboratory , Astropart. Phys. 28 [astro-ph/0701286] (2008) 495-507.[16] R. Acciarri et al.,
Test of a new Fast Waveform Digitizer for PMT signalread-out from liquid Argon Dark Matter detectors , arXiv:1203.1371[17] R. Acciarri et al.,
Demonstration and Comparison of Operation of Photomul-tiplier Tubes at Liquid Argon Temperature , arXiv:1108.5584[18] A.M. Szelc for the WArP collaboration, AIP Conf.Proc. 1115 (2009) 105-110[19] W. H. Lippincott et al., Phys.Rev.D77:075007, arXiv:0801.1534v4 (2008)[20] S. Yellin, Phys.Rev. D66 032005, arXiv:physics/0203002v2 [physics.data-an],(2002)[21] A. Pocar,
The DarkSide project , presented at Dark Matter Workshop, AspenPhysics Institute, 7-11 February, 2011[22] http://cedar.berkeley.edu/plotter Tool maintained by Danielle Speller. Formermaintenance and development was conducted by Rick Gaitskell, Vuk Mandicand Jeff Filippini.[23] XENON 100 Collaboration,