Constraints on Inelastic Dark Matter Signal using ZEPLIN-II Results
aa r X i v : . [ a s t r o - ph . C O ] N ov Constraints on Inelastic Dark Matter Signal usingZEPLIN-II Results
D.B. Cline, W. Ooi ∗ , H. Wang Department of Physics & Astronomy, University of California, Los Angeles, USA
Abstract
There has been an increasing interest on the concept of Inelastic Dark Matter(iDM) - motivated in part by some recent data. We describe the constraintson iDM from the results of the two phase dark matter detector ZEPLIN-II, which has demonstrated strong background discrimination capabilities( > >
99% C.L., for WeaklyInteracting Massive Particles (WIMPs) masses >
100 GeV.
Key words:
ZEPLIN-II, dark matter, WIMPs, liquid xenon, radiationdetectors
PACS: > ∗ Corresponding author; address: Department of Physics & Astronomy, University ofCalifornia, Los Angeles, USA
Email address: [email protected] (W. Ooi)
Preprint submitted to Elsevier November 3, 2018 . ZEPLIN-II Operation
ZEPLIN-II is a dual phase (gas and liquid) Xe detector designed specif-ically to identify nuclear recoils induced on Xe atoms by the Weakly Inter-acting Massive Particles (WIMPs); one of the popular candidates for darkmatter in the universe [4], [5]. The detector is located in the UK Bou-bly underground laboratory which provides about 2800 m water-equalventshielding for a 10 reduction in cosmic muon flux [6]. The central detec-tor was constructed out of low-radioactive materials at UCLA, with under-ground ambient radioactivity carefully studied and modeled at Boulby [7].Additional background rejection is afforded by the dual phase design whichmeasures both scintillation and ionization components for each event in theliquid xenon. Since background electron recoils differ from the signal nuclearrecoils in the ratio of these components by about a factor of 3, backgroundrejection can be achieved at >
98% efficiency [3]. ZEPLIN-II detector tookdata for 31.2 days with 7.2 kg fiducial mass and obtained the final data setreported in Ref. [3]. The WIMP signal acceptance box is defined between5 to 20 keV electron-equivalent (keV ee ), encompassing 50% of the total sig-nal region determined by prior calibration runs. 29 events were observed inthe acceptance box at the end of first dark matter run. However, since theacceptance box is flanked on both sides by background events, it is to beexpected that some of the events in the box come from the tail distributionsof the background events. A careful analysis of the background distribu-tion was performed, and it was learned that 28.6 ±
2. ZEPLIN-II Background Measurement
The background events in the S2/S1 ∼
300 region above the acceptancebox are γ events, as learned from calibration runs with γ source Co andamericium/beryllium (AmBe) neutron source. To quantify the event ratedistribution, events with energy between 5 to 20 keV ee were selected andbinned according to its Log(S2/S1)-k(E) values, where k(E) is the 50% nu-clear recoil acceptance value of log(S2/S1) for the energy of each event. Inanother words, events with Log(S2/S1)-k(E) < γ event rate distribution is well characterized by a Gaussian with an off-set, where the offset accounted for the coincidental events arising from the2igh trigger rate ( ∼
70 Hz) encountered during calibrations. These coinci-dental events were not expected to contribute to the background level in theacceptance box, since they were not observed in the science data, which oc-curred at a much lower trigger rate ( ∼ ± γ events were estimated tobe in the acceptance box between 5 to 20 keV ee .The second background events found below the acceptance box were nuclearrecoil events with diminished S2 originated from radon contamination inSAES getter, as confirmed by an dedicated Rn measurement. These recoilsoccurred near the charged PTFE walls which stripped some of the chargesfrom the electron cloud that constituted S2. The diminished S2 subsequentlylowered the position reconstruction accuracy, which resulted in a fraction ofthese events being wrongly placed within the fiducial volume. To quantifythe number of misplaced events, a distribution of reconstructed radii of allevents within the acceptance regions in (S2/S1)-energy parameter space wasstudied. This distribution is well fitted with a Gaussian and is extrapolatedinto the fiducial volume. Finally the extrapolated distribution is integratedto give the expected number of radon-related events, which is 12.5 ± γ -ray ( Co) Rn-initiated Total5-10 keV ee
14 4.2 ± ± ± ee
15 11.9 ± ± ± Table 1: Overall expectation values in the nuclear recoil acceptance window compared toobserved counts. The errors are derived directly from fit uncertainties.
3. Inelastic Dark Matter Model
The inelastic dark matter (iDM) [8], [9], [10] model explored the possibil-ity of an altered kinematic of the WIMP-nucleus interaction in an attempt toreconcile the detection of annual modulation observation [11] and null resultsfrom all other experiments such as CDMS [12], XENON10 [13] and ZEPLIN-II [3]. iDM assumes two basic properties: a) WIMP χ has an excited state χ *, with a mass m χ ∗ − m χ = δ ∼
100 keV recoil-energy (keV r ). b) Elastic3catterings of χ off target nucleus N are suppressed compared with the in-elastic scattering χ N → χ ∗ N. Refer to Ref. [10], [14] and [15] for discussionson particle models that may give rise to this scenario.Compared to the elastic case, the introduction of splitting δ in the scatter-ing kinematics increases the minimum velocity β min of χ to scatter with adeposited recoil energy E R . The new β min can be shown to be β min = r m N E R (cid:18) m N E R µ + δ (cid:19) (1)where m N is the mass of the target nucleus and µ is the reduced mass of the χ /target nucleus system. For direct detection experiments, the key conse-quences of iDM are:1. Differential event rate is lowered since minimum velocity β min is in-creased for a given recoil energy.2. Low-energy events are suppressed in the spectrum of events. In theelastic case, event rate is the highest in the low recoil energy, while thesplitting δ ∼
100 keV r in iDM scenario can suppress or even eliminatelow energy events.3. Annual modulation of signal is enhanced. In the extreme case, wherethe χ particles only have enough minimum velocity in the summer butnot in the winter, the modulation can be 100%, albeit at a significantlower overall event rate.4. Heavier target nuclei are even more favored. While it is true that inthe elastic case heavier target nuclei gives rises to a higher differentialevent rate, inelastic scattering favors heavier target more since β min islowered significantly for heavier elements.These key characteristics are quantified in Fig 1. An effective test to theWIMPs discovery signals reported in Ref. [11] with Iodine (I) target (A=127)is to use a different target with comparable atomic mass. ZEPLIN-II withXe (A=132) target is a good experiment for this test.
4. ZEPLIN-II Inelastic Dark Matter Limits
The first ZEPLIN-II dark matter run recorded 29 events in the accep-tance box with an expected background of 28.6 ± Max method detailed in Ref. [17], may not be the best approach tointerpret ZEPLIN-II results since p
Max is in general more suitable for ex-periments with small number of events where a statistical characterizationof background is more challenging. We maintain that we understood ourbackground accurately enough to use Feldman Cousins’ method, which hasan additional advantage of guaranteed correct coverage.Fig. 2 shows iDM ZEPLIN-II exclusion limits and DAMA signal contoursfor different WIMP masses. DAMA contours of 90% (red, △ χ =6.25) and99% (green, △ χ =11.34) C.L. were constructed following the treatments byRef. [8] using iodine quenching factor (qI) of 0.085, where E R = E ee /qI. ForZEPLIN-II, the zero field quenching factor qXe is modified by the presence ofthe electric field, and the conversion equation becomes E R = E ee /qXe · (Se/Sn),where the scintillation quenching of electron and nuclear recoils due to theelectric field are S e =0.54 ± n =0.95 ± R = E ee /0.36), the resultant ZEPLIN-II in-elastic dark matter limits are shown in blue solid lines, for the WIMP massesbetween 70 GeV ≤ m χ ≤
250 GeV. Using these parameters, it can be seenthat these limits constrained a larger iDM parameter space than those pre-viously reported in Ref. [8], suggesting the exclusion of published claims foriDM signals at >
99% C.L. for WIMP masses >
100 GeV. Also shown in pinkdotted lines are ZEPLIN-II limits assuming the the most recent energy de-pendent qXe values reported in Ref. [20], which estimated an increase in thesevalues at high recoil energies (qXe ∼ r ). The new qXe mea-surements lowered the exclusion power of ZEPLIN-II results, but at WIMPsmasses >
180 GeV there are still strong indications of exclusion of discov-ery claims at >
99% C.L. Incidentally, uncertainty in qI, which is reportedas qI=0.09 ± . Conclusion In conclusion, the published ZEPLIN-II background analysis reported inRef. [3] showed that the events observed in the acceptance box were statisti-cally consistent with the tail distributions from two background populations;namely the γ and radon-induced events. As such it is natural to take thebackground subtraction approach to analyze the events in the acceptance re-gion, from which we declared ZEPLIN-II a null experiment with a 90% C.L.upper limit on WIMP signal number at 10.4. This translated to iDM exclu-sion limits that are significantly more stringent than the previously publishedvalues in Ref. [8], which adopted p Max method to derive an upper limit. Usingparameters from Standard Halo Model, ZEPLIN-II limits suggest the exclu-sion of published claims for iDM signals at >
99% C.L., for WIMP masses >
100 GeV.This work has been funded by the US Department of Energy (grant num-bers DE-FG03-91ER40662 and DE-FG03-95ER40917) and the US NationalScience Foundation (grant number PHY-0139065 and PHY-06-53459). Wewould like to thank members of ZEPLIN-II Collaboration for the help andsupport. We are also indebted to N. Weiner for valuable advice on the detailsof iDM model and we acknowledge helpful discussions with P.F. Smith. D.B.Cline would like to thank the Aspen Institute for Physics where part of thispaper was completed.
References [1] D.B. Cline et al. (2000),
Astropart. Phys. , 373.[2] H. Wang, Ph.D. thesis, UCLA (1998).[3] G.J. Alner et al. (2007), Astropart. Phys. , 287.[4] W. Freedman and M. Turner (2003), Rev. Mod. Phys. , 1433.[5] G. Jungman, M. Kamionkowski and K. Griest (1996), Phys. Rep. ,195.[6] M. Robinson et al. (2003),
Nucl. Instrum. Meth.
A511 et al. (2005)
Astroparticle Phys. , 97.68] S. Chang et al. (2009), Phys. Rev. D , 043513.[9] D. Tucker-Smith and N. Weiner (2005), Phys. Rev. D , 063509.[10] D. Tucker-Smith and N. Weiner (2001) Phys. Rev. D , 043502.[11] R. Bernabei et al. (2008), Eur. Phys. J. C , 333.[12] Z. Ahmed et al. (2009), Phys. Rev. Lett. , 011301.[13] J. Angle et al. (2008),
Phys. Rev. Lett. , 021303.[14] J. March-Russell et al. (2009), arXiv:0812.1931v3 [astro-ph].[15] Y. Cui et al. (2009), arXiv:0901.0557v1 [hep-ph].[16] G. Feldman and R. Cousins (1998),
Phys. Rev. D , 3873.[17] S. Yellin (2002), Phys. Rev. D , 032005.[18] E. Aprile et al. (2005), Phys. Rev. D , 072006.[19] A. Manzur et al. (2009), arXiv:0909.1063v1 [physics.ins-det].[20] P. Sorensen et al. (2009), Nucl. Instrum. Meth. A , 339.[21] J.D. Lewin and P.F. Smith (1996),
Astropart. Phys. et al. (2006), Phys. Rev. D , 043531.[23] http://people.roma2.infn.it/dama/7 igure 1: (top) Energy spectrum of WIMP events on Xe target for both elastic scattering(red, dotted) and inelastic scattering (blue, line) models. (middle) Inelastic event ratesuppression, compared to the elastic case, as a function of the splitting δ . (bottom) Xeannual modulation fraction as a function of δ assuming v earth =227 ± escape =500 km/s. igure 2: Preferred regions for DAMA/LIBRA results and ZEPLIN-II exclusion limits.The blue solid lines are ZEPLIN-II exclusion limits at 90% C.L., assuming xenon zero-fieldquenching factor qXe=0.19. The pink dotted lines are the exclusion limits assuming themost recent qXe reported in Ref. [20]. The DAMA/LIBRA contours are at 90%(red) and99% (green) C.L. with standard qI=0.085, following the analysis done in Ref. [8]. Theseplots assumed v escape = 500 km/s.= 500 km/s.