222 Rn emanation measurements for the XENON1T experiment
E. Aprile, J. Aalbers, F. Agostini, M. Alfonsi, L. Althueser, F. D. Amaro, V. C. Antochi, E. Angelino, J. R. Angevaare, F. Arneodo, D. Barge, L. Baudis, B. Bauermeister, L. Bellagamba, M. L. Benabderrahmane, T. Berger, P. A. Breur, A. Brown, E. Brown, S. Bruenner, G. Bruno, R. Budnik, C. Capelli, J. M. R. Cardoso, D. Cichon, B. Cimmino, M. Clark, D. Coderre, A. P. Colijn, J. Conrad, J. P. Cussonneau, M. P. Decowski, A. Depoian, P. Di Gangi, A. Di Giovanni, R. Di Stefano, S. Diglio, A. Elykov, G. Eurin, A. D. Ferella, W. Fulgione, P. Gaemers, R. Gaior, A. Gallo Rosso, M. Galloway, F. Gao, L. Grandi, M. Garbini, C. Hasterok, C. Hils, K. Hiraide, L. Hoetzsch, E. Hogenbirk, J. Howlett, M. Iacovacci, Y. Itow, F. Joerg, N. Kato, S. Kazama, M. Kobayashi, G. Koltman, A. Kopec, H. Landsman, R. F. Lang, L. Levinson, Q. Lin, S. Lindemann, M. Lindner, F. Lombardi, J. A. M. Lopes, E. López Fune, C. Macolino, J. Mahlstedt, L. Manenti, A. Manfredini, F. Marignetti, T. Marrodán Undagoitia, K. Martens, J. Masbou, D. Masson, S. Mastroianni, M. Messina, K. Miuchi, A. Molinario, K. Morå, S. Moriyama, Y. Mosbacher, M. Murra, J. Naganoma, K. Ni, U. Oberlack, K. Odgers, J. Palacio, B. Pelssers, R. Peres, J. Pienaar, V. Pizzella, G. Plante, J. Qin, H. Qiu, et al. (43 additional authors not shown)
EEur. Phys. J. C manuscript No. (will be inserted by the editor)
Rn emanation measurements for the XENON1T experiment
E. Aprile , J. Aalbers , F. Agostini , M. Alfonsi , L. Althueser , F. D. Amaro ,V. C. Antochi , E. Angelino , J. R. Angevaare , F. Arneodo , D. Barge , L. Baudis ,B. Bauermeister , L. Bellagamba , M. L. Benabderrahmane , T. Berger , P. A. Breur ,A. Brown , E. Brown , S. Bruenner , G. Bruno , R. Budnik , C. Capelli ,J. M. R. Cardoso , D. Cichon , B. Cimmino , M. Clark , D. Coderre , A. P. Colijn ,J. Conrad , J. P. Cussonneau , M. P. Decowski , A. Depoian , P. Di Gangi ,A. Di Giovanni , R. Di Stefano , S. Diglio , A. Elykov , G. Eurin , A. D. Ferella ,W. Fulgione , P. Gaemers , R. Gaior , A. Gallo Rosso , M. Galloway , F. Gao ,L. Grandi , M. Garbini , C. Hasterok , C. Hils , K. Hiraide , L. Hoetzsch ,E. Hogenbirk , J. Howlett , M. Iacovacci , Y. Itow , F. Joerg , N. Kato ,S. Kazama , M. Kobayashi , G. Koltman , A. Kopec , H. Landsman , R. F. Lang ,L. Levinson , Q. Lin , S. Lindemann , M. Lindner , F. Lombardi , J. A. M. Lopes ,E. López Fune , C. Macolino , J. Mahlstedt , L. Manenti , A. Manfredini ,F. Marignetti , T. Marrodán Undagoitia , K. Martens , J. Masbou , D. Masson ,S. Mastroianni , M. Messina , K. Miuchi , A. Molinario , K. Morå , S. Moriyama ,Y. Mosbacher , M. Murra , J. Naganoma , K. Ni , U. Oberlack , K. Odgers ,J. Palacio , B. Pelssers , R. Peres , J. Pienaar , V. Pizzella , G. Plante ,J. Qin , H. Qiu , D. Ramírez García , S. Reichard , A. Rocchetti , N. Rupp ,J. M. F. dos Santos , G. Sartorelli , N. Šarˇcevi´c , M. Scheibelhut , S. Schindler ,J. Schreiner , D. Schulte , M. Schumann , L. Scotto Lavina , M. Selvi , F. Semeria ,P. Shagin , E. Shockley , M. Silva , H. Simgen , A. Takeda , C. Therreau ,D. Thers , F. Toschi , G. Trinchero , C. Tunnell , M. Vargas , G. Volta , O. Wack ,H. Wang , Y. Wei , C. Weinheimer , M.Weiss , D. Wenz , J. Westermann ,C. Wittweg , J. Wulf , Z. Xu , M. Yamashita , J. Ye , G. Zavattini , Y. Zhang ,T. Zhu , J. P. Zopounidis (XENON Collaboration h ). Physics Department, Columbia University, New York, NY 10027, USA Oskar Klein Centre, Department of Physics, Stockholm University, AlbaNova, Stockholm SE-10691, Sweden Department of Physics and Astronomy, University of Bologna and INFN-Bologna, 40126 Bologna, Italy Institut für Physik & Exzellenzcluster PRISMA, Johannes Gutenberg-Universität Mainz, 55099 Mainz, Germany Institut für Kernphysik, Westfälische Wilhelms-Universität Münster, 48149 Münster, Germany LIBPhys, Department of Physics, University of Coimbra, 3004-516 Coimbra, Portugal INAF-Astrophysical Observatory of Torino, Department of Physics, University of Torino and INFN-Torino, 10125 Torino, Italy Nikhef and the University of Amsterdam, Science Park, 1098XG Amsterdam, Netherlands New York University Abu Dhabi, Abu Dhabi, United Arab Emirates Physik-Institut, University of Zürich, 8057 Zürich, Switzerland Department of Physics, Applied Physics and Astronomy, Rensselaer Polytechnic Institute, Troy, NY 12180, USA Department of Particle Physics and Astrophysics, Weizmann Institute of Science, Rehovot 7610001, Israel Max-Planck-Institut für Kernphysik, 69117 Heidelberg, Germany Department of Physics “Ettore Pancini”, University of Napoli and INFN-Napoli, 80126 Napoli, Italy Department of Physics and Astronomy, Purdue University, West Lafayette, IN 47907, USA a r X i v : . [ phy s i c s . i n s - d e t ] N ov Physikalisches Institut, Universität Freiburg, 79104 Freiburg, Germany SUBATECH, IMT Atlantique, CNRS/IN2P3, Université de Nantes, Nantes 44307, France Department of Physics and Chemistry, University of L’Aquila, 67100 L’Aquila, Italy INFN-Laboratori Nazionali del Gran Sasso and Gran Sasso Science Institute, 67100 L’Aquila, Italy LPNHE, Sorbonne Université, Université de Paris, CNRS/IN2P3, Paris, France Department of Physics & Kavli Institute for Cosmological Physics, University of Chicago, Chicago, IL 60637, USA Kamioka Observatory, Institute for Cosmic Ray Research, and Kavli Institute for the Physics and Mathematics of the Universe (WPI), theUniversity of Tokyo, Higashi-Mozumi, Kamioka, Hida, Gifu 506-1205, Japan Kobayashi-Maskawa Institute for the Origin of Particles and the Universe, and Institute for Space-Earth Environmental Research, NagoyaUniversity, Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8602, Japan Université Paris-Saclay, CNRS/IN2P3, IJCLab, 91405 Orsay, France Department of Physics, Kobe University, Kobe, Hyogo 657-8501, Japan Department of Physics, University of California San Diego, La Jolla, CA 92093, USA Department of Physics and Astronomy, Rice University, Houston, TX 77005, USA Physics & Astronomy Department, University of California, Los Angeles, CA 90095, USAthe date of receipt and acceptance should be inserted later
Abstract
The selection of low-radioactive construction ma-terials is of utmost importance for the success of low-energyrare event search experiments. Besides radioactive contami-nants in the bulk, the emanation of radioactive radon atomsfrom material surfaces attains increasing relevance in the ef-fort to further reduce the background of such experiments.In this work, we present the
Rn emanation measurementsperformed for the XENON1T dark matter experiment. To-gether with the bulk impurity screening campaign, the re-sults enabled us to select the radio-purest construction mate-rials, targeting a
Rn activity concentration of 10 µ Bq / kgin 3 . Rn sources allowed us to selectively eliminate problem-atic components in the course of the experiment. The pre-dictions from the emanation measurements were comparedto data of the
Rn activity concentration in XENON1T.The final
Rn activity concentration of ( . ± . ) µ Bq / kgin the target of XENON1T is the lowest ever achieved in axenon dark matter experiment. a Also at Simons Center for Geometry and Physics and C. N. YangInstitute for Theoretical Physics, SUNY, Stony Brook, NY, USA b Also at Institute for Subatomic Physics, Utrecht University, Utrecht,Netherlands c Also at Institute for Advanced Research, Nagoya University, Nagoya,Aichi 464-8601, Japan d Also at Coimbra Polytechnic - ISEC, Coimbra, Portugal e Also at INFN, Sez. di Ferrara and Dip. di Fisica e Scienze della Terra,Università di Ferrara, via G. Saragat 1, Edificio C, I-44122 Ferrara(FE), Italy f [email protected] g [email protected] h [email protected] Many cosmological observations suggest that a large frac-tion of the total matter density of the Universe is made upof an unknown form of dark matter [1]. However, despite alarge experimental effort, dark matter has not yet been dis-covered. XENON1T [2] was the largest and most sensitiveso far in the series of XENON direct dark matter search ex-periments [3, 4]. Its successor XENONnT will start data-taking in 2020. The primary aim of these experiments is thedetection of Weakly Interacting Massive Particles (WIMPs),a promising dark matter candidate [5]. As in other astroparti-cle physics experiments [6–8] liquid xenon is used as an effi-cient target for particle detection. The XENON detectors aredual-phase time projection chambers (TPCs) with a gaseouslayer of xenon on top of the target. Particles interacting withxenon nuclei or the atomic electrons create scintillation lightand ionization electrons. The light is detected by two arraysof UV-sensitive photomultiplier tubes (PMTs) on top andbottom of the detector. The ionization electrons are driftedupwards to the liquid-gas interface, where they are extractedand create a second, delayed scintillation light signal. Bothsignals are used to gain information about the location andenergy of the interaction. They are also used to reject back-ground either by the event multiplicity (multi-scatter versussingle-scatter events) or by the type of interaction (electronicrecoil versus nuclear recoil events).XENON1T operated for two years, starting from De-cember 2016. Similarly to other astroparticle physics ex-periment looking for rare events, it required an extremelylow background level. Throughout the different generationsof the XENON experiments, external background sourceshave been suppressed, e.g. by an improved external shieldand xenon self-shielding and by the mitigation of radioactiv-ity from materials. Their level was marginal in XENON1T and intrinsic background sources became dominant. Amongthem, the radioactive isotope
Rn induced the leading back-ground component [9]. Its long-lived mother nucleus
Ra(T / = U decay chainand thus present in most materials. Once
Ra decays, thecreated noble gas isotope
Rn may emanate from innersurfaces into the xenon volume.As
Rn has a relatively long half-life (T / = . β -decaysof its daughter isotope Pb can mimic signal events. Toachieve the scientific goal of XENON1T, a
Rn activityconcentration of 10 µ Bq / kg was required [9]. Other radonisotopes may also lead to background events. However, theircontribution was strongly suppressed due to their small abun-dance in the detector and much shorter half-lives, that didnot allow for their dispersion within the target volume.XENON1T has performed a comprehensive bulk impurityscreening campaign to select radio-pure materials using HighPurity Germanium (HPGe) spectroscopy and InductivelyCoupled Plasma Mass Spectrometry (ICP-MS) [10]. How-ever, the measured Ra bulk activity can in general not beused to predict how much
Rn emanates from the mate-rial, because surface impurities may become dominant. Thismade dedicated
Rn emanation measurements necessary,which are described in this article.There are two ways for a radon atom to leave the mate-rial in which it is produced: by recoil or by diffusion [11]. Inthe first case, the decay occurs directly on or below the ma-terial’s surface. The recoil energy, which the radon nucleusreceives during the α -decay of its mother radium nuclide(85 keV [12] in the case of Rn), is sufficient to eject itfrom the material. In the second case, the radon atom dif-fuses inside the bulk of the material. If it reaches a boundarysurface before its decay, it will be emanated. Data on radondiffusion in metals are rare, but its diffusion coefficient iseven smaller than the minuscule one of xenon in metals [13].Thus, it is reasonable to assume that radon diffusion plays asignificant role only in soft or porous materials. As a conse-quence, a radon emanation measurement of metals is mostlya test of surface impurities.The article is structured as follows. Section 2 discussesthe employed
Rn assay techniques. In section 3, we pre-sent the screening results of investigated materials, most ofwhich were eventually used for the construction ofXENON1T. Section 4 describes the overall
Rn emana-tion measurements of the assembled XENON1T detectorand gives a summary on the inferred
Rn budget, as wellas a comparison to XENON1T data. We also describe radonreduction methods that were applied during detector opera-tion. We close with a summary and outlook in section 5.
Rn assay techniques
For XENON1T we used two methods to study
Rn ema-nation. The most sensitive method applied ultra-low back-ground miniaturized proportional counters [14], developedfor the GALLEX solar neutrino experiment [15]. These de-vices are made of high purity synthetic quartz with an ironcathode and a thin tungsten-wire (13 µ m diameter) anode.The active volume of the counters is around 1 cm and thecounting gas consists of 90 % argon and 10 % methane towhich the radon to be measured is added. Rn atoms de-cay by α -disintegration, followed by two further α -decaysfrom Po and
Po until the long-lived
Pb breaks thesecular equilibrium. Cosmic muons as well as environmen-tal β - and γ -radiation cannot deposit energies above 50 keVin the miniaturized counters. In contrast, α -decays exhibita larger energy deposition, allowing for their clear identi-fication. The detection efficiency for the three α -decays isnot equal, because the gaseous Rn is distributed in theentire counter volume, while polonium ions deposit on sur-faces. On average (49 . ± .
0) % of all α -decays are de-tected, yielding an expectation value of (1 . ± .
06) countsper
Rn decay. The background count rate above 50 keVscatters around one count per day for the different propor-tional counters. Thus, a minimum detectable activity of ∼ µ Bq can be achieved.Prior to a measurement, the emanated
Rn atoms hadto be collected from the samples, concentrated and mixedwith the counting gas. For this purpose, the samples wereplaced in gas-tight emanation vessels made of glass or stain-less steel. We ensured that the pieces of a sample were notstacked in order to let radon escape freely from all surfaces.Ambient air was removed by pumping or flushing with radon-free carrier gases (in most cases helium). Then, the vesselwas filled with the carrier gas and sealed and the emanated
Rn accumulated until the
Rn activity reached a sizablefraction of its equilibrium value. After typically one week,the mixture of carrier gas and
Rn atoms was pumped orflushed through a gas purifier to remove gaseous impurities.The
Rn atoms were trapped in an activated carbon col-umn at liquid nitrogen temperature afterwards.Larger samples, such as subsystems of the entireXENON1T experiment, could not be placed in emanationvessels. In these cases, the carrier gas was usually filled di-rectly into the gas-tight system. Such samples took the roleof both, the emanation vessel and the investigated
Rnsource, whereas the rest of the procedure remainedunchanged. Sometimes only a fraction of the filled carriergas could be extracted due to limited pumping power. Insuch cases, the quoted activities were corrected for this re-duced extraction efficiency, assuming that the radon was ho-mogeneously distributed in the gas.
The concentrated
Rn sample was further processedin a sample purification system. The same system was alsoused to fill the proportional counters [16]. It featured severalcold traps and a non-evaporable hot getter pump to removeunwanted trace-impurities that could impair the counter per-formance. In the final step, the gas sample was mixed withthe counting gas and pushed into the counter by means amercury column.In some cases, measurements had to be done on sam-ples that were previously exposed to xenon. The subsequentxenon out-gassing inhibited the use of miniaturized propor-tional counters due to their small volume and the difficultyto separate radon from xenon. In such cases we used electro-static radon monitors with a significantly larger active vol-ume [17, 18]. The positively charged
Rn daughters werecollected on a silicon PIN diode biased with a negative highvoltage with respect to the vessel’s walls. All subsequent α -decays were recorded by the diode and the signal from Po was evaluated as it has the highest detection efficiency(approximately 30 %). We used two different monitors witha vessel size of 1 liter and 4 liters, respectively. The back-ground due to self-emanation of the monitor was negligiblecompared to the signal in all measurements. Even though thesensitivity of the radon monitors was about four times worsecompared to the measurements with proportional counters,it was sufficient for our applications in XENON1T.All results in this article are given with a combined un-certainty σ including statistical and systematical errors. Ifthe result is compatible with zero within 1 . σ , a 90 %C.L. upper limit is given instead. Whenever a sample wasadditionally screened by γ -ray spectroscopy, we quote theresult obtained in [10] and refer to the identifier used in thatwork as Radioassay-ID (RID). This section presents the results of samples which were mea-sured during the preparation and construction phase ofXENON1T. We also list the supplier of the samples as itwas not always possible to identify the manufacturer of theraw material.3.1 Metal samples
In an early phase of the project a cryostat made of grade-1titanium was considered for XENON1T and the
Rn em-anation rate of several titanium samples was measured. Theresults are given in Table 1.Sample seam on the plates of sample Rn emanation rate. We performed surfacecleaning tests for all samples. Sample . ) solution. Titanium itselfis not soluble in nitric acid, but the acid may remove traceimpurities, in particular Ra, from the surface. In contrast,sample µ m ofthe surface were removed by electro-polishing. All resultsare given as both, an absolute Rn emanation rate and arate normalized to the surface of the sample. For samples
Rn came from the weld seam or from the surface. Thus,both normalizations cannot be true simultaneously and mustbe considered as upper limits.Before any purification, the
Rn emanation rates of thegrade-1 titanium samples exhibited large variations, whichdid not show up in the
Ra activity obtained from γ -rayspectroscopy (last column in Table 1), hinting at a surfacecontamination. The nitric acid treatment did not show anyimprovement ( Rnemanation rate. Under the assumption that the contamina-tion was equally distributed among the 13 plates, we ob-tained a factor 4 . ± . Rn emanationrate. It disappeared completely after electro-polishing (
Ra activity of the sam-ple was located on the surface. Thus, the true bulk activitymust have been lower than reported in [10] (RID γ -ray spectroscopy cannot resolve the spatial distribution ofthe radio-impurities in a sample and usually assumes that allactivity is in the bulk. Sample Rn assay technique and γ -ray spectroscopy, and the importance to apply both meth-ods.The results of the titanium Rn screening campaignsuggested that electro-polishing can suppress the
Rn em-anation rate of titanium to a negligible level. However, ourtitanium samples showed a too high uranium bulk contami-nation. Therefore, the XENON1T cryostat was made of stain-less steel [2].
Stainless steel was mainly used for vessels and pipes in theXENON1T inner detector system. Parts of the TPC werealso made of stainless steel, however, their surface area were TIG: Tungsten Inert Gas
Table 1
Rn emanation measurements of grade-1 titanium samples. The normalization to surface area and, where relevant, to weld seam lengthis also given. For comparison the
Ra bulk activity taken from [10] is quoted and referred to as RID (Radioassay-ID) defined there.ID Sample Supplier Description Treatment
Rn emanation rate
Ra activity [10] ± µ Bq RID × × ± µ Bq/m < . / 6.9 kg ± µ Bqin 1.8 % HNO (140 ± µ Bq/m ± µ Bq RID ×
20 cm × ± µ Bq/m < . / 4.6 kg (31 ± µ Bq/m17.4 m of TIG-weld seam ± µ BqTotal: 0.6 m / 2.4 kg in 0.6 % HF / 7.6 % HNO (130 ± µ Bq/m ± µ Bq/m ± µ Bq RID ×
20 cm × ± µ Bq/m (1 . ± .
4) mBq/kgTotal: 1.1 m / 4.6 kg < µ BqTotal: 0.5 m / 2.1 kg 30 µ m surface removed < µ Bq/m small and the contribution to Rn emanation was thus ex-pected to be minor. Table 2 summarizes the results of allinvestigated stainless steel samples. Samples
Rn source in stainless steel pipes. Samples
Rn emanation rate of TIG-welded samplesoriginated from the surface or from the weld. Thus, in Ta-ble 2 we give both normalizations in addition to the absolute
Rn emanation rate.Sample
Rn emana-tion rate, although the weld seam length of sample
Rn emanation rate as the welded sample
Rn emanation rate by a factor3 . ± . Rn re-duction by cleaning attempts was not measurable within oursensitivity. Again it was confirmed that stainless steel TIG-welds do not represent a notable additional source of
Rn, which is in tension with findings from other experiments[14, 20].The measurements of the bellows were motivated by therelatively high
Rn emanation rate of the cryogenic pipe(
Rn source of thecryogenic pipe.We also tested two heat exchangers, which were usedto evaporate and re-condense xenon in the purification loop.They were made of stainless steel plates brazed with a cop-per alloy. We measured the larger one (
Rn emanation rate of ( ± ) µ Bq. Subsequently, we cleaned it by exposing all inter-nal surfaces to a 1 . ( ± ) µ Bq. Clearly,the treatment was effective despite the rather weak concen-tration of nitric acid. The heat exchangers were combinedwith a high purity electrical heater (
Rn emana-tion rate was found to be ( ± ) µ Bq.The last sample (
Table 2
Rn emanation results of various stainless steel (SS) samples. Where relevant, the normalization to surface area and to weld seam lengthis also given.ID Sample Supplier Description Treatment
Rn emanation rate ± µ Bq(304 or 316L) 17.9 cm x 17.6 cm x 1 cm (560 ± µ Bq/m Total: 0.33 m / 12.9 kg (23 ± µ Bq/m ± µ Bq(316L) Outer diameter: 10.18 cm (380 ± µ Bq/m Thickness of wall: 0.34 cm (45 ± µ Bq/mLength: 30 cmTotal: 0.56 m / 7.2 kg ± µ Bq(316L) (80 ± µ Bq/m (10 ± µ Bq/m ± µ Bq(316L) Outer diameter: 10.18 cm (290 ± µ Bq/m Thickness of wall: 0.34 cmLength: 30 cmTotal: 0.56 m / 7.2 kg ≤ µ Bq(304L) Outer diameter: 10.16 cm ≤ µ Bq/m Thickness of wall: 0.2 cm ≤ µ Bq/mLength: 34 cmTotal: 0.19 m / 1.5 kg ± µ Bq(304L) Outer diameter: 10.16 cm (270 ± µ Bq/m Thickness of wall: 0.2 cm (14 ± µ Bq/mLength: 34 cmTotal: 0.19 m / 1.5 kg ± µ Bq(304L) Outer diameter: 10.16 cm and (300 ± µ Bq/m Thickness of wall: 0.2 cm electro-polished (16 ± µ Bq/mLength: 34 cmTotal: 0.19 m / 1.5 kg ± µ Bq1m long, inner diameter 35 mm ≤ µ Bq0.4m long, inner diameter 100 mm ± µ Bq60 SS plates (338 mm ×
130 mm)brazed with copper alloy ± µ Bqcombined with 20 SS plates (226 mm ×
86 mm) ∼
12h withlarge heat exchanger brazed with copper alloy 1.8 % HNO plus type FG5X12-60 (see above) ± µ Bqheater with large et Applications des with very high purity standards ∼
15 min withheat transfer surface Techniques de L’Energie) 2 % HNO ± µ Bqpacking material 0.095 m /piece, total surface: 5.2 m (9 ± µ Bq/m removal tests [21] (see section 4.3), we were interested in its Rn emanation rate. Indeed, all 55 packings together onlyemanated ( ± ) µ Bq, which was an excellent result forthe rather large sample surface (5.2 m ). 3.2 Gas purification system In order to maintain its ultra-high chemical purity, the xenonin XENON1T was continuously cleaned by SAES gas pu-rifiers. Two of them were used in parallel to provide therequired purification efficiency. Each purifier contained aporous, highly chemically-active zirconium-alloy in two car-tridges. The larger was operated at 400 ◦ C, while the smaller
Table 3
Rn emanation rates of four noble gas purifiers from the company SAES. PS4-MT50-R535 is identical to PS4-MT50-R2, but receiveda new commercial label.ID Model Mass of active material Cold state activity [mBq] Hot state activity [mBq] Used in XENON1T ∼ . ± .
04 1 . ± .
15 yes ∼ . ± .
03 yes ∼ . . ± .
03 0 . ± .
03 no ∼ . ≤ one was kept at room temperature and acted as a dedicatedhydrogen removal unit. Altogether, ∼ Rn emanationrate in two different thermal conditions; at room temperature(cold state) and at operating temperature (hot state). Whileonly the hot state was relevant for the experiment, the mea-sured rate of the cold state could have given insight into the
Rn emanation process relevant for these porous materi-als. An enhanced
Rn release rate at elevated temperaturewould have given evidence for diffusion-driven emanation.The two gas purifiers used in XENON1T differed signifi-cantly in their hot state emanation rate (
Rn than
Rn emanation rate.For sample . ± .
27 in comparison with its hot state. In con-trast, for sample
Rn emanation rate of the gas purifiers is not fullyunderstood, but the large difference for identical models sug-gested that it depends on the purity of the raw materials. Thatopens up the possibility for further
Rn reduction by ma-terial screening.
XENON1T used customized QDrive piston pumps fromChart Industries for xenon gas recirculation [22]. Threepumps (
Rnemanation rate. The results are summarized in Table 4. Af-ter a mechanical failure of QDrive pump C204, it was sentback to the manufacturer for repair. Afterwards, its
Rnemanation rate was lower by more than a factor two (
Rn budget(see section 4.2), we performed further investigations to un-derstand the origin of the observed
Rn emanation rate. For this purpose, we screened most of the individual compo-nents of a yet unassembled QDrive pump (see Table 5). Wefound that the stators of the pump’s electrical motor (
Rn sources, followed by the pistons(
Rn sources, wefurther investigated their constituent parts (see Table 6). First,we noticed that the bare magnets did not emanate a lot of
Rn (
Rn emanation rate rather originated fromthe epoxy coating as can be seen from the comparison tosample
Rn emanationrate (
Rn emanation rate. Later, we identified a cleaner alterna-tive (
Rn reduction (
Rn emanation rate ofthe QDrive pumps, we followed an approach of EXO-200[23] to build a cleaner magnetically coupled piston pump.The new pump was developed within the XENON collab-oration together with groups from the nEXO collaboration[24]. Its
Rn emanation rate was found to be (0 . ± . Rn budget is discussedin section 4.3.3.3 Other samplesThis section presents
Rn emanation measurements of TPCcomponents and other samples measured for the XENON1T
Table 4
Rn emanation measurements of xenon recirculation pumps.ID Sample Description
Rn emanation Used Commentrate [mBq] in XENON1T ± ± ± ± ± Table 5
Rn emanation measurements of all parts of a QDrive recirculation pump prior to assembly.ID Sample Description
Rn emanation rate ± ± ± ± < .
053 mBq < .
020 mBq
Table 6
Results of
Rn emanation measurement to identify and replace the dominant
Rn sources in the QDrive pump. < .
021 mBqTotal surface: 112 cm < . × × ± (24.8 ± ± (1.51 ± < .
097 mBq × × ± ± Total: 167 cm / 29.4 g ± / 95.4 g (17.7 ± ± / 51.9 g (0.76 ± experiment. The complexity of the XENON1T TPC madeit impossible to screen every component. Therefore, we fo-cused on samples which either cover a large surface area in-side the TPC or are known to be potential Rn sources. Weinvestigated the light sensors (
Rncalibration source (
Rn em-anation rate was measured in a helium-free environment andwe used neon as carrier gas. Helium would have penetratedinto the PMTs, creating an unacceptably high rate of after-pulses [26]. We also measured the
Rn emanation rate ofthe PMT high voltage divider circuits (base) used to read outthe signal and to supply the high voltage (
Ra activity, but no dedicated
Rn ema-nation test was performed. With the results given in Table7 and taking into account that XENON1T used 248 PMTs,we estimated a total contribution of ( . ± . ) mBq fromthe PMTs and their bases. Note that the Ra activity in thePMTs (RID
Rn emanation rate for all cables was be-low the detection limit. The
Ra activity of the cables, [10]normalized to their mass was at least ten times larger than its
Rn emanation rates for each cable sample. This indicatedthat the
Rn sources were located in the inner part of thecables. The impact of the cables on the
Rn budget is dis-cussed in more detail in section 4.1. Different high voltagecable sections were connected at the top of the TPC and justin front of the vacuum feedthroughs with D-sub type pinswhich showed no measurable
Rn emanation rate (
Table 7
Rn emanation measurements of various other samples. For comparison the
Ra bulk activity taken from [10] is quoted and referredto as RID (Radioassay-ID) defined there.ID Sample Supplier Description
Rn emanation rate
Ra activity [10] ± µ Bq RID ± µ Bq/PMT (600 ± µ Bq/PMTand mechanical samples ± µ Bq RID ± µ Bq/piece (15 ± µ Bq/pieceand capacitors ≤ µ Bq RID ≤ µ Bq/m (4000 ± µ Bq/kgfor high voltage supply ≤
460 k µ Bq/kg(used in XENON1T) Ω cable ≤ µ Bq100 m / 0.55 kg ≤ µ Bq/mfor signal readout ≤ µ Bq/kg(not used in XENON1T) ≤ µ Bq RID ≤ µ Bq/m (1000 ± µ Bq/kgfor signal readout ≤ µ Bq/kg(not used in XENON1T) ≤ µ Bq RID ≤ µ Bq/m (400 ± µ Bq/kgfor signal readout ≤ µ Bq/kg(used in XENON1T) ≤ µ Bqcontact pins made of Cu/Be and Cu/bronze ≤ µ Bq/piece ≤ µ BqSeals 9.5 mm thick, 95 mm diameter ≤ µ Bq/m
506 cm / 191 cm / 0.47 kg / 31.9 kg (97 ± µ Bq RID ± µ Bq/m < µ Bq/kgwidth: 13 cm – 19 cm (3.0 ± µ Bq/kgthickness: 0.5 cm – 1.6 cm ≤ µ Bqand 2 cm diameter ≤ µ Bq/m / 7069 cm / 56.7 kg ≤ µ Bq/kg
Rn calibration PTB electro-deposited Th ≤ µ Bqsource [27] on stainless steel disc The cables were fed to the air-side via potted feedthroughs.The epoxy used for this potting ( ≤ µ Bq.The reflective walls of the XENON1T TPC were formedby diamond-shaved PTFE panels. We measured (nondiamond-shaved) leftover pieces from the fabrication of thesepanels, corresponding to a surface area of ≥
50 % of thePTFE surface in XENON1T. The results (see
Rn source inside the TPC.We also measured copper rods from the same batch asthe TPC field-shaping rings (
Rn releaserate of a
Rn source that was used as a calibration sourcefor XENON1T (
Rn, it would have required a long waiting time af-ter each usage until the
Rn had decayed. Our measure-ment showed that its
Rn emanation rate was below the de- tection limit and negligible with respect to the other Rnsources.
Rn emanation resultsof either individual subsystems or of combined measure-ments among them. More details on the measurements them-selves can be found in [18, 28]. We only assayed those sub-systems that were continuously in contact with xenon duringthe data acquisition periods and, therefore, contributed to the
Rn budget. A schematic view of the subsystems most rel-evant for
Rn emanation is shown in Figure 1. The limit presented here differs slightly from the result published in[27] ( ≤ µ Bq) due to a data re-evaluation.0
Fig. 1
Schematic setup of XENON1T (not to scale). The colors indicate different sub-components and are also used in Figure 2, which showstheir individual contributions to the overall
Rn rate.
Fig. 2
The different sub-system contributions to the overall
Rn emanation rate in XENON1T. The colors correspond to those used in Figure 1.The numbers in the brackets refer to the item numbers. QDrive pump C207 was not measured. Its
Rn emanation rate was estimated (see text).
The TPC was hosted in a double-walled, vacuum-insu-lated stainless steel cryostat, which was closed at the top bya dome. The dome was in turn connected to the cryogenicpipe, which also contained the cables and guided them to theelectric feedthrough. Evaporated xenon which reached thecryogenic system got liquefied and was returned back to thecryostat. In a second loop, the xenon gas passed through thepurification system that contained the recirculation pumpsand the gas purifiers. Note that a major part of the XENON1Tinfrastructure will be re-used for its upgrade XENONnT, sothe obtained results will be relevant for the future as well[29].The results of the measurements are listed in Table 8. Af-ter electro-polishing its inner surface, the cryostat (
Rn emanation rate of ( ± ) µ Bq / m for electro-polished stainless steel under the assumption of no contribu-tion from the weld seam. Thus, we expected ( ± ) µ Bqfor the 7 . surface of the cryostat. This was about onethird of what we have measured ( . . ≤ . Table 8
Results from measurements of several subsystems of the XENON1T setup. The cryogenic pipe and the TPC were not measured directly.Their
Rn emanation rates were inferred indirectly by subtracting the results from two measurements, respectively.ID component activity [mBq] comment . ± . . ± . . ± . . ± . . ± . . ± .
04 from Table 2inner detector volume without TPC 18 . ± . . ± . ≤ . for the cables alone. For the entire cable pipe we measured apositive number of ( . ± . ) mBq ( Rnbudget were found to originate from the cryogenic system(
Rn emanation signal of sample
Rn emanation rate was obtained by measuringit simultaneously with the inner vessel of the cryostat. Bysubtracting the known result of the latter one (
Rn emanation rate of ( . ± . ) mBq for it.Finally, we were interested in the Rn emanation rateof the TPC. From now on we will use the term ‘inner detec-tor volume’ for all subsystems illustrated in Figure 1 exceptthe purification system. The TPC contribution could be ob-tained by subtracting the signal of the inner detector volumeafter and before TPC installation. The latter one was foundto be (18 . ± . ) mBq (see Table 8) by summing up thecontributions from sample Rn distribution in the carrier gas. Thus, we thoroughlymixed the sample gas immediately before the extraction byadding
Rn-free carrier gas. Moreover, we extracted fromvarious ports to ensure that locations with possibly differ-ent
Rn activity concentrations were averaged out. Moredetails on the procedure can be found in [18]. From the ob- tained result of ( . ± . ) mBq we calculated the activityof the TPC alone ( Rn sources of the TPC as quoted in section 3.3.4.2 Overall
Rn budget in XENON1TThe measurements of the subsystems allowed us to calcu-late the expected
Rn budget for XENON1T. Apart fromthe inner detector volume (
Rn emanation rates were presented in section 3.2. Thetwo gas purifiers together emanated (1 . ± .
15) mBq (seeTable 3). During the main dark matter search phases, theso-called science run 0 (SR0 [30]) and science run 1 (SR1[31]), three QDrive pumps were used: C204 (after repair,
Rn emanation ratescan be taken from Table 4 except for pump C207, whichwas not measured. Its signal could be estimated by takingthe average of the highest ( . ± .
9) mBqfor pump C207. Thus, the estimated
Rn emanation rateof all three pumps together was (11 . ± .
9) mBq.Summing up all components, we obtained a total
Rnbudget for XENON1T of (32 . ± .
3) mBq. Under the as-sumption of a homogeneous radon distribution, we expecteda
Rn activity concentration of ( . ± . ) µ Bq / kg witha total xenon mass of 3 . Rn budget. α -decays of Rn (5 . Po (6 . ∼ Fig. 3
The activity concentration evolution of
Rn and
Po during XENON1T data-taking. In science run 0 and 1 (SR0 and SR1) the activityconcentrations were stable over the entire time and we only show the initial period here. Xenon distillation campaigns to remove
Rn as well asthe exchange of the recirculation pumps lead to a reduction of the
Rn and
Po activity concentration. They gray regions indicate periods ofdetector calibration or maintenance. high energies was sufficient to allow for a clear separationof the two α -peaks. Other background sources in that en-ergy range were subdominant and could be ignored. There-fore, the α -analysis represented a reliable way to measurethe Rn and
Po activity concentrations in the detector.In Figure 3, we show the α -rate evolution during the sci-ence runs. An average activity concentrations of ( . ± . ) µ Bq/kg and ( . ± . ) µ Bq/kg for
Rn and
Po,respectively, was found for SR0 and SR1, excluding the dis-tillation period for radon removal discussed in section 4.3.
Po is often positively charged after its creation [35]. Thus,ion drift as well as convective motion may transport
Poout of the analysis volume and cause its deposition on TPCsurfaces. This may explain the slightly lower observed
Poactivity concentration with respect to
Rn.The discrepancy between the ( . ± . ) µ Bq/kg de-duced from α -analysis and the expectation of(10 . ± . µ Bq/kg from emanation measurements corre-sponds to (10.6 ± . Rn re-lease from the QDrive recirculation pumps. As most sam-ples in this work, they were measured at room temperature.However, during operation, they heated up and the diffusion-driven
Rn emanation could have been enhanced at ele-vated temperature. In addition, it could have been that theunmeasured QDrive pump C207 emanated more than ex-pected. 4.3 Reduction of
RnThere are several possibilities to further reduce the
Rnbudget. The best option is to remove
Ra, the mother nu-cleus of
Rn, from the experiment. This was achieved byreplacing the QDrive pumps in science run 2 (SR2). Theywere exchanged for the magnetically coupled piston pumpdescribed in section 3.2.2 [24]. The decrease of the
Rnactivity concentration in liquid xenon before and after thepump exchange is shown in Figure 3. It corresponded toan absolute reduction of ( . ± . ) mBq in 3 . . ± .
9) mBq . The observed difference of ( . ± . ) mBq agreed within the uncertainty with the differencebetween Rn emanation measurements and α -analysis (seesection 4.2). This supported the hypothesis that the Rnemanation rate of the QDrive pumps was larger than esti-mated. By exchanging the recirculation pumps, the largest
Rn source was successfully removed from the experi-ment yielding a reduced background level [36]. This is verypromising for XENONnT, where magnetically coupled pis-ton pumps will be employed.Another possibility for
Rn reduction is an online radonremoval system in the xenon purification loop, which sepa-rates xenon from radon and retains the latter until its disin-tegration. Such a radon removal system based on cryogenicdistillation was pioneered by the XENON collaboration [21, The sum of the signals of the three replaced QDrive pumps minus thesignal of the newly mounted magentically coupled piston pump.3
Rn reduction of ∼
20 % during SR0, although only a small fraction of the re-circulation flow was distilled. A compatible absolute
Rnreduction was achieved in a second xenon distillation runperformed in SR2, after the pump exchange. The finally ac-complished
Rn activity concentration in XENON1T was ( . ± . ) µ Bq / kg.As a consequence of the promising results, the XENONcollaboration realized two new purification systems forXENONnT which complement the existing ones and signi-ficantly improve the xenon purity. The first one is a novel li-quid purification system which is able to produce and main-tain ultra-pure liquid xenon at a very fast flow rate. The se-cond one is a dedicated distillation column which was de-signed [38] and built for radon removal and which takes ad-vantage of the large flow rate enabled by the liquid purifica-tion system.The impact of the new radon removal system on the ex-periment’s background can be further maximized if Rn isflushed out of the detector before it enters the TPC. A de-tailed mapping of the
Rn sources in XENON1T was ob-tained by the
Rn emanation measurements of the varioussubsystems, presented in this work. Thus, a targeted xenonflow pattern optimization could be studied. Such a flow pat-tern optimization with respect to the radon removal systemwill be applied in the XENONnT experiment [29].
The background rate of current xenon dark matter detectorsis dominated by
Rn-induced events and it is expectedthat
Rn daughters will remain an essential backgroundcomponent in future experiments. Therefore,
Rn emana-tion measurements become increasingly important and pro-vide complementary information to the bulk radioactivityscreening efforts. In this article, we presented the resultsof a comprehensive material screening campaign for
Rnemanation carried out for the XENON1T experiment, usingstate-of-the-art counting techniques. We selected construc-tion materials with the lowest possible
Rn emanation rateand we were able to identify and locate the remaining
Rnsources in the experiment.The predicted activity concentration from these measure-ments was in agreement with the target
Rn activity con-centration of 10 µ Bq / kg in 3 . α -analysis of the XENON1T data were about 30 % higherthan the prediction. The discrepancy could be understood byan underestimation of the recirculation pump’s Rn em-anation rate. With the exact knowledge of the distribution of
Rn sources in XENON1T, it became possible to se-lectively eliminate problematic items. The ultimately mea-sured
Rn activity concentration of ( . ± . ) µ Bq / kg isthe lowest ever achieved in a xenon dark matter experiment.Significant improvements are possible in XENONnT andfurther projects, for instance, by continuous xenon distilla-tion. We gratefully acknowledge support from the National Sci-ence Foundation, Swiss National Science Foundation, Ger-man Ministry for Education and Research, Max Planck Ge-sellschaft, Deutsche Forschungsgemeinschaft, NetherlandsOrganisation for Scientific Research (NWO), Weizmann In-stitute of Science, ISF, Fundacao para a Ciencia e a Tecnolo-gia, Région des Pays de la Loire, Knut and Alice WallenbergFoundation, Kavli Foundation, JSPS Kakenhi in Japan andIstituto Nazionale di Fisica Nucleare. This project has re-ceived funding or support from the European Union’s Hori-zon 2020 research and innovation programme under the Ma-rie Sklodowska-Curie Grant Agreements No. 690575 andNo. 674896, respectively. Data processing is performed us-ing infrastructures from the Open Science Grid, the Euro-pean Grid Initiative and the Dutch national e-infrastructurewith the support of SURF Cooperative. We are grateful toLaboratori Nazionali del Gran Sasso for hosting and sup-porting the XENON project.
References
1. Planck Collaboration
Planck 2018 results. VI. Cosmo-logical parameters , arXiv:1807.06209 (2018).2. E. Aprile et al. (XENON Collaboration),
TheXENON1T Dark Matter Experiment , Eur. Phys. J.C77:881 (2017).3. E. Aprile et al. (XENON10 Collaboration),
Design andPerformance of the XENON10 Dark Matter Experi-ment , Astropart.Phys.34:679-698 (2011).4. E. Aprile et al. (XENON100 Collaboration),
TheXENON100 Dark Matter Experiment , Astropart. Phys.35, 573-590 (2012).5. L. E. Strigari,
Galactic Searches for Dark Matter , Phys.Rep. 531, 1 (2013).6. nEXO Collaboration, nEXO Pre-Conceptual DesignReport , arXiv:1805.11142 (2018).7. B.J. Mount et al. (LZ Collaboration),
LUX-ZEPLIN(LZ) Technical Design Report , arXiv:1703.09144(2017).8. Li Zhao et al. (PandaX Collaboration),
PandaX: A deepunderground dark matter search experiment in China using liquid xenon , Mod. Phys. Lett. A33 (2018) no.30,1830013 (2018).9. E. Aprile et al. (XENON Collaboration), Physics reachof the XENON1T dark matter experiment , J. Cos. As-trop. Phys. 04:027 (2016)10. E. Aprile et al. (XENON Collaboration),
Material ra-dioassay and selection for the XENON1T dark matterexperiment , Eur. Phys. J. C77:890 (2017).11. V. Balek,
Emanation thermal analysis , Thermochim.Acta 222 (1978) 1.12. International Atomic Energy Agency
Nuclear Data Ser-vices
Diffusion of xenon inmetals , J. Nucl. Materials 19(1):90-92 (1966).14. G. Zuzel and H. Simgen,
High sensitivity radon ema-nation measurements , Appl. Radiat. Isot. 67(5):889-893(2009).15. R. Wink et al.,
The miniaturized proportional counterHD-2(Fe)/(Si) for the GALLEX solar neutrino experi-ment , Nucl. Inst. Meth. A 329:3, 541 (1993).16. H. Simgen et al.,
Highly sensitive measurements of ra-dioactive noble gas nuclides in the Borexino solar neu-trino experiment , App. Rad. Isot. 61, 213 (2004).17. J. Kiko,
Detector for
Rn measurements in air at 1mBq/m level , Nucl. Instrum. Methods A, 460 (2001).18. S. A. Bruenner, Mitigation of
Rn inducedbackground in the XENON1T dark matter ex-periment , PhD thesis, University of Heidelberg,https://doi.org/10.11588/heidok.00023261 (2017).19. E. Aprile et al. (XENON Collaboration),
Removingkrypton from xenon by cryogenic distillation to the ppqlevel , Eur. Phys. J. C 77, 275 (2017).20. H. Simgen et al.,
Analysis of radioactive trace impuri-ties with µ Bq-sensitivity in Borexino , Int. J. Mod. Phys.A 29,16 1442009 (2014).21. E. Aprile et al. (XENON Collaboration),
Online
Rnremoval by cryogenic distillation in the XENON100 ex-periment
A magnetically driven piston pumpfor ultra-clean applications , Rev. Sci. Instr. 82, 105114(2011).24. E. Brown et al.,
Magnetically-coupled piston pump forhigh-purity gas applications , Eur. Phys. J., C78:604(2018).25. E. Aprile et al. (XENON Collaboration),
Lower-ing the radioactivity of the photomultiplier tubes forthe XENON1T dark matter experiment , Eur. Phys. J.C75:546 (2015).26. Hamamatsu Photonics,
Handbook Photo-multiplier tubes Basics and Applications
A Rn-220 source for the calibration oflow-background experiments , JINST 11 P04004 (2016).28. N. Rupp
On the detection of
Rn with miniatur-ized propotional counters: background, sensitivity stud-ies and results for XENON1T , Master thesis, Uni-versity of Heidelberg, http://hdl.handle.net/11858/00-001M-0000-0027-797D-D (2015).29. E. Aprile et al. (XENON Collaboration),
ProjectedWIMP Sensitivity of the XENONnT Dark Matter Exper-iment , arXiv:2007.08796 (2020).30. E. Aprile et al. (XENON Collaboration),
First DarkMatter Search Results from the XENON1T Experiment ,Phys. Rev. Lett. 119, 181301 (2017).31. E. Aprile et al. (XENON Collaboration),
Dark Mat-ter Search Results from a One Ton-Year Exposure ofXENON1T , Phys. Rev. Lett. 121, 111302 (2018).32. E. Aprile et al. (XENON Collaboration),
Intrinsic back-grounds from Rn and Kr in the XENON100 experiment ,Eur. Phys. J. C78: 132 (2018).33. P.A. Breur,
Backgrounds in XENON1T
Phd thesis, University of Amsterdam,https://hdl.handle.net/11245.1/7bed5cdb-57f2-4c13-b09c-407936a669de (2019).34. N. Rupp,
PhD thesis in preparation (2020).35. EXO-200 Collaboration,
Measurements of the ion frac-tion and mobility of alpha and beta decay products inliquid xenon using EXO-200 , Phys. Rev. C 92, 045504(2015).36. E. Aprile et al. (XENON Collaboration),
Observa-tion of Excess Electronic Recoil Events in XENON1T ,arXiv:2006.09721v2 (2020).37. S. A. Bruenner et al.,
Radon depletion in xenon boil-offgas , Eur. Phys. J., C77:143, (2017).38. M. Murra,