Development of an array of HPGe detectors with 980% relative efficiency
D. S. Leonard, I. S. Hahn, W. G. Kang, V. Kazalov, G. W. Kim, Y. D. Kim, E. K. Lee, M. H. Lee, S. Y. Park, E. Sala
DDevelopment of an array of HPGe detectors with 980%relative efficiency
D. S. Leonard a , I. S. Hahn b,c , W. G. Kang a , V. Kazalov d , G. W. Kim a ,Y. D. Kim a,e,f , E. K. Lee a , M. H. Lee a,f, ∗ , S. Y. Park g , E. Sala a,1, ∗∗ a Center for Underground Physics, Institute for Basic Science, Daejeon 34126, Korea b Department of Science Education, Ewha Womans University, Seoul 03760, Korea c Center for Exotic Nuclear Studies, Institute for Basic Science, Daejeon 34126, Korea d Baksan Neutrino Observatory, Institute for Nuclear Research of the Russian Academy ofScience, Kabardino-Balkaria 361609, Russia e Department of Physics and Astronomy, Sejong University, Seoul 05006, Korea f IBS School, University of Science and Technology, Daejeon, 34113, Korea g Department of Physics, Ewha Womans University, Seoul 03760, Korea
Abstract
Searches for new physics push experiments to look for increasingly rare inter-actions. As a result, detectors require increasing sensitivity and specificity, andmaterials must be screened for naturally occurring, background-producing ra-dioactivity. Furthermore the detectors used for screening must approach thesensitivities of the physics-search detectors themselves, thus motivating itera-tive development of detectors capable of both physics searches and backgroundscreening. We report on the design, installation, and performance of a novel,low-background, fourteen-element high-purity germanium detector named theCAGe (CUP Array of Germanium), installed at the Yangyang undergroundlaboratory in Korea.
Keywords: radiopurity, germanium counting, low background, ultra-traceanalysis, double beta decay, dark matter
PACS:
1. Introduction
The Center for Underground Physics (CUP) at the Institute for Basic Science(IBS) and collaborating institutions are involved in developing and operatinga number of rare-event physics-search experiments, including searches for dark ∗ Corresponding author. ∗∗ Corresponding author.
Email addresses: [email protected] (D. S. Leonard), mhlee @ ibs.re.kr (M. H. Lee), [email protected] (E. Sala) Now at Center for Axion and Precision Physics Research, Institute for Basic Science,Daejeon 34051, Korea
Preprint submitted to NIM Section A September 2, 2020 a r X i v : . [ phy s i c s . i n s - d e t ] S e p atter and neutrinoless double-beta decay. Probably the largest challenge to allsuch experiments is controlling and understanding the experiment backgroundsignals arising from trace levels of naturally occurring radioactive contaminantsin the detector construction materials. In collaboration with CANBERRA (nowMIRION Technologies), we have customized and installed a fourteen-elementhigh-purity germanium (HPGe) array detector named the CAGe (CUP Arrayof Germanium) that is intended for screening detector materials for trace levelsof radioactivity [1, 2, 3, 4] and for directly performing new physics searches suchas searches for the decay of Ta [5, 6].Because of their large penetration depths, gamma emissions in particular,including indirect bremsstrahlung photons from beta-decays, place contamina-tion constraints on all materials in and around a low-rate detector, making thesebackgrounds one of the most general classes of concerns for such experiments. Inparticular, backgrounds arise from gamma-emitting decays within decay chainssupported by long-lived naturally occurring radioactive isotopes such as K, Th, and
Ra, as well cosmologically activated isotopes such as Co.The background rates detectable and relevant to development of the mostsensitive detectors in the world tend to be lower, by design, than the sensitivitylevels of existing detectors, making it very challenging to screen materials forconstruction of the new detectors. For this reason, the screening detectors mustbe developed along with the physics detectors, and can also then serve bothpurposes.An optimal gamma-screening detector should have a high efficiency andstrong discrimination of signals from background, also qualities of a good physics-search detector. By using an array of fourteen HPGe detectors with 70% relativeefficiency each, we can obtain a large overall efficiency. The segmented natureof the arrangement also allows for detection of coincident signals, which whencorrelated to gamma energy can often be used to enhance signal detection orbackground rejection. In particular, signals arising from internal samples are,in comparison to those from external shielding elements, relatively more likelyto produce multiple-detector, multiple-gamma coincidence events, identifiablein cases where at least one of the multiple detector signals corresponds to a full-energy gamma peak. We note similarities to the single-crysotat array reportedin Ref. [7], which itself was inspired by the even more similar fourteen-elementconceptual design presented in Ref. [8].In order to optimize the utility of the array, we carried out an extensivecampaign to characterize a prototype single-element detector and to screen andcustomize array components to reduce the final array backgrounds. The arraywas installed in the A5 tunnel at the Yangyang underground laboratory, havinga measured muon flux of 328 ±
10 muons / m / day [9], with installation completedin March of 2017.We report here on the development and design of the array and initial signalanalysis and performance results. 2 . Detector Design and Development The array is constructed of fourteen p-type coaxial HPGe elements with70% relative efficiency each. The elements are attached to two opposing coppercryostat bases, with seven detectors on each base. Each base approximates acylinder with a diameter of about 35 cm, and each is cooled through a coppercold finger, extending through the detector shielding and connected to its ownliquid nitrogen (LN2) dewar. The detector elements are surrounded by individ-ual copper cylinders extending from the cryostat bases, and are arranged in ahexagonal arrangement with the seventh element for each placed in the center.The elements are individually encapsulated and sealed to the cryostat base witho-ring flanges to allow easy maintenance of individual detector elements with-out breaking vacuum for the remaining elements. A single large o-ring sealseach of the two main cryostat bases. The elements within each hexagon ar-rangement are positioned with approximately 2.5 cm of space between adjacentcryostat housings, allowing placement of sample material between the elementsto optimize efficiency.The entire array assembly is kept in a room with a volume of 32 m , just largeenough for people to access the hardware. The room has no forced ventilationunder normal operation. We have installed a stand-alone air filter that recircu-lates air and filters it through a Samsung AX90M air-cleaner utilizing an H13class HEPA filter operated at maximum flow rate, specified as 680 m / h. Usinga particle counter located on the opposite end of the room, we have observedclass 1000 level of particulates with this filter operating, even while people areworking in the room. Figure 1: Diagrams of the entire array construction, including shielding and LN2 dewars,without movable doors (left), and with movable doors (right). The shielding layers are visiblearound the array, from inside out as copper, ancient lead, and Goslar lead. Cold fingers extendfrom the upper and lower array bases to the dewars. The shielding doors on the front andback are rolled open and closed on tracks using motors mounted on the top frame. The upperarray can be adjusted along with the cold finger, upper LN2 dewar, and a moveable sectionof shielding. igure 2: The outer array shielding (left) and the array in the inner copper shielding (right).The top array LN2 dewar is seen facing the center section of lead shielding with lifting appa-ratus attached from the top. The wall of lead toward the left of the photo is one of the twomovable side ”doors” which roll along a track on the floor. The door is rolled open in theview of the array on the right. The two array halves are positioned with the coplanar end-caps of the lowerhalf facing upwards and towards the downward-facing end-caps of the upperhalf, creating a space in between for samples, with about 30 cm in useful diam-eter and a vertical separation which can be adjusted from 2.5 cm up to 7.5 cm.A drawing of the entire assembly is shown in Fig. 1 and photos of the com-pleted structure are shown in Fig. 2 Since the detector housings cannot supportsignificant weight, two shelves are mounted to the vertical alignment posts andcan be used for supporting samples. Alternatively sample arrangements can beconstructed with supports from the lower cryostat base or the shielding floor.To achieve the adjustable separation used to optimize sample volume andefficiency, the entire upper assembly can be raised along with its cold finger andLN2 dewar. Since the upper cold finger penetrates the shielding, a column ofshielding, including the copper and lead, above the cold finger can be raised,and shims can be inserted under the cold finger to fill in the gap produced whenraising it. Fig. 3 shows a top view of the shielding without the doors, with themovable column removed. Threaded rods can be used to raise the top array anddewar together, eliminating stress on the cold finger. The dewars are mountedpermanently on adjustable supports, while a purpose-built adjusting tool ismounted between the two array halves during lifting or lowering and is replacedwith fixed-length supports after achieving the desired height. Since the detectoris targeted at high-sensitivity measurements with long-duration of counting, itis not expected that this operation will be performed very frequently.4 igure 3: The diagram (left) shows a conceptual design of the array with the central shielding.Movable shielding doors on the sides are not shown. The top array and its cryostat are notshown in the diagram. The shielding penetration for the top array’s cold-finger is visiblein the upper-left of the shielding. The vertical shielding column allows the cold-finger andshielding inserted above it to raise and lower with adjustments to the top array height. Thephoto (right) shows the missing shielding section above and below the upper cold-finger, witha temporary support chain in place, later replaced with a permanent lifting apparatus.
In the final operation, the openings to each side of the array (as seen inFig 2) have been covered with 3M brand Vikuiti film, and a boil-off nitrogenline is inserted into the shielding interior. The combination serves to flush Rncontamination from the air quickly and effectively.In an attempt to detect background sources for feedback to the final design,we initially tested a single element of the array in a similar shielding config-uration to the final design, and performed MC background simulations. Theresulting observed backgrounds were inconsistent with backgrounds originatingfrom sources entirely outside of the shielding. In particular, low energy peakswere over-emphasized. We thus screened several potential prototype compo-nents for the detector construction to search for potential reducible sources ofbackgrounds. An initial study of this screening, performed by counting withour underground 100% HPGe detector at the nearby A6 area, was presented inref [10, 11]. In particular
Th was found to be a potential background sourcein aluminum, and O-rings were found to potentially contribute non-negligiblyto all of K, Th, and
Ra sub-chain backgrounds. Table 1 lists the resultsfrom HPGe counting of a number of candidate o-ring types.The final O-rings used were peroxide-cured EPDM seals sold by Marco Rub-ber and made with their E1000 compound. Two batches of these O-rings werepurchased with measurements of each shown in Table 1. Ultimately the finalproduction design removed aluminum in the element housings in favor of copperwhile maintaining acceptable efficiencies to low-energy gammas.5 able 1: Radioactivity results from single-HPGe screening of candidate O-rings for the arrayconstruction. In all cases we assume that
Th and
Ra (half-life of 3.6 days) are inequilibrium. A 20% systematic error is included to account for uncertainty in the detectorefficiencies. Uncertainties are one standard-deviation, and limits are 90% C.L.
Supplier, Matrial,and/or Part Num. K[mBq/kg]
Ac[mBq/kg]
Th[mBq/kg]
Ra[mBq/kg]
COG Viton, Vi 650,green ± <
180 96 ±
25 540 ± Marco Rubber FKM”white viton” V1012-154 ±
500 104 ±
26 68 ±
16 1 800 ± Marco Rubber, FKM75A, Black, P/N:V1000-154 < <
29 79 ±
20 460 ± Samwon, FKM 70, P/N:OR VA75 10154 SW ± <
230 170 ±
40 710 ± O-ring USA Viton, V75 ± < <
92 520 ± Marco rubber EPDME1055 Peroxide cured ±
700 460 ±
100 430 ±
90 580 ± Easterns Seals EPDM,BS154EP70 ± <
61 31 ±
10 31 ± Marco Rubber EPDME1000 Peroxide curedbatch 1 ± < <
11 29 ± Marco Rubber EPDME1000 Peroxide curedbatch 2 ± < < . ± O-ring USA EPDM ± <
75 57 ±
17 74 ± Polymax EPDM SKU:BS154E70 ± ± ± ± ROW Inc. FEPencapsulated silicone ±
900 1 050 ±
220 1 040 ±
210 2 500 ± Marco rubber, AFLASTFE-P < ±
100 120 ±
30 570 ± CANBERRA stocko-ring ±
290 360 ±
90 210 ±
40 1 370 ± . Electronics and Signal Processing The array elements are biased by three iseg-brand NHS6060P positive high-voltage programmable supplies, each supporting up to six-channels of output.The supplied CANBERRA pre-amplifiers provide an HV inhibit signal which istriggered when one of the detector elements becomes too warm. These signalsare connected back to the HV inhibits for the corresponding channels.
Figure 4: Electronics diagram. Outputs of shaping amplifiers are digitized by four flash ADCmodules, all synchronized by a trigger control board that generates global triggers to initiatedata transfer to the acquisition computer (PC).
The readout electronics are shown in Fig. 4. The pre-amp outputs are con-nected to shaping amplifiers (CANBERRA 2026) with 6 µ s shaping time. Theoutputs of the shaping amplifiers are connected to analog inputs of four 12-bitflash analog to digital converters (FADC500, NOTICE) capable of operating atup to 5 × samples per second. Each module takes four channels of inputand two modules are combined within a single physical hardware unit. A NO-TICE supplied trigger control board (TCB) maintains synchronization of the4 GHz clocks of the FADC modules, and interprets local triggers from each mod-ule to generate a global trigger. A global trigger results in data frames being7ent directly from the FADC’s to the data-acquisition (DAQ) PC by USB 3.0connections.
4. Analysis and Performance
Waveforms are processed by fitting a Guassian peak to the data in a rangewithin 200 bins before and after the maximum peak height. The pedestal issampled and subtracted. A typical waveform fit is shown in Fig. 5. Calibrationsare applied separately for each detector element, and a ROOT analysis file isconstructed such that each event contains the energy recorded by every detector.
Figure 5: A typical waveform (solid curve) from the shaping amplifier of an array element,as digitized by the FADC500. A Gaussian fit (dashed curve) is used to determine the pulseheight. Time bins span 16 ns per bin.
Background data were initially obtained with untreated air surrounding thedetector and an LN2 boil-off flush line inserted into the detection chamber forflushing. Studies of radon and dust levels in the detector room were reported inRef. [1]. To investigate effects of radon or dust, as described in Sec. 2, we coveredthe detector sides with 3M brand Vikuiti film in an attempt to improve the sealand flushing of the detection chamber. A second background run was taken inthis configuration and indeed resulted in significantly reduced background levels.Both background spectra are shown in Fig. 6. Reductions were noticed in boththe
Ra and
Th decay-chain activities as shown in Table 4. While
Radaughters can be transported far in air through the movement of
Rn, theshorter half-life of the thorium-related Rn isotope,
Rn, creates an expectationthat such effects would be reduced. One possible explanation is the reduction of8ust achieved both by the application of the Vikuiti film and by the introductionof the room air filter described in Sec. 2.Detector resolutions were measured in the first month of full operation usingthe intrinsic 1 460 keV peak from K. FWHM values were determined fromGaussian fits and are reported in table 4.
Figure 6: Background spectra for the detector array before (solid) and after (dashed) addingthe Vikuiti window over the door area. The spectra are for the total of all energies recordedin any single event.
5. Conclusion
An array of HPGe detectors with a total efficiency much higher than a stan-dard single-element 100% HPGe detector, and with competitive backgrounds,was developed and installed at the underground lab in Yangyang, Korea. Thehigh solid-angle of coverage also results in high probability for detection ofcoincident gammas for sources placed in the detector interior, adding a power-ful background discrimination tool. The detector resolution and backgroundsmeasured are well suited to extended low-background counting. Sensitivitiesto activities in sample materials are very dependent on the sample details andthe optimization of each sample geometry. Details of coincidence analysis andresulting sensitivities for the first measurements will be reported separately.With optimized counting with the array, we can expect to achieve sensitivitiesto
Th and
Ra several times lower than those achieved by typical screeningwith our standard 100% detectors. Furthermore the high coincidence efficiency9 able 2: Count rates for gamma peaks generated from the U, Th, and K decay chains.Results are shown before and after covering the door area with a Vikuiti window, effectivelyproducing a better air seal, while still allowing a nitrogen gas flush.
Rate [cnts/day]Decay Gamma Energy [keV] No window With Vikuiti window
Pb 238.6 36 ± . ± . Pb 242.0 511 ± . ± . ± . ± . ±
10 41 . ± . Bi 609.3 1 412 ± . ± .
21 120.3 354 ± . ± .
71 764.5 414 ± . ± .
62 204.8 113 . ± . . ± . K 1 460.8 19 . ± . . ± . Ac 911.2 6 . ± . . ± . . ± . . ± . Pb 238.6 35 . ± . . ± . Tl 583.2 12 . ± . . ± .
42 614.5 8 . ± . . ± . Table 3: FWHM resolutions of all detector elements, obtained from Gaussian fits to the1 460 keV peak of K decay. The detectors are numbers as 1 – 7 for the bottom array and 8– 14 for the top array. The values were obtained in the first month of operation.
Bottom TopFWHM Resolution FWHM ResolutionDetector [keV] Detector [keV]1 1.99 8 1.932 2.01 9 2.043 1.89 10 2.004 1.93 11 2.075 2.05 12 1.996 1.96 13 1.917 1.91 14 2.1010rovides an expectation that we can search for physics at previously unachievedsensitivity levels, for example, searches for decays of
Ta [5].11 cknowledgements
This work was supported by the Institute for Basic Science (IBS) funded bythe Ministry of Science and ICT, Korea(Grant id: IBS-R016-D1). We wouldlike to thank Pascal Quirin and his team at CANBERRA, now MIRION Tech-nologies, for their extensive help and cooperation throughout the design processand installation.
References [1] S. Y. Park, et al., An enriched
Mo powder measurement by an array ofHPGe detectors, Proc. of Sci. 340 (2019) 783.[2] S. Y. Park, et al., Measurement of the Background Activities of a
Mo-enriched powder sample for an AMoRE crystal material by using a singlehigh purity germanium detector, J. Korean Phys. Soc. 76 (2020) 1060. doi:10.3938/jkps.76.1060 .[3] M. H. Lee, et al., An ultra-low radioactivity measurement HPGe facility atthe Center for Underground Physics, Proc. of Sci. 340 (2019) 363.[4] M. H. Lee, et al., Radioassay and Purification for Experiments at Y2Land Yemilab in Korea, J. Phys.: Conf. Ser. 1468 (2020) 012249. doi:10.1088/1742-6596/1468/1/012249 .[5] G. W. Kim, et al., Simulation Study for the Half-Life Measurement of
Ta Using HPGe Detectors, J. Korean Phys. Soc. 75 (2019) 32. doi:10.3938/jkps.75.32 .[6] G. W. Kim, Rare Decay Experiments for
Ta and Pb ∗ Using HPGeDetectors, Ph.D. thesis, Ewha Womans Univ. (2019).[7] M. E. Keilor, et al., CASCADES: An Ultra-Low-Background GermaniumCrystal Array at Pacific Northwest National Laboratory, AIP ConferenceProceedings 1412 (2011) 208. doi:10.1063/1.3665316 .[8] M. E. Keilor, et al., Design and construction of an ultra-low-background14-crystal germanium array for high efficiency and coincidence mea-surements, J. Radioanal Nucl Chem 282 (2009) 703. doi:10.1007/s10967-009-0248-7 .[9] H. Prihtiadi, et al., Muon detector for the COSINE-100 experiment, J.Instrum. 13 (2018) T02007. doi:10.1088/1748-0221/13/02/T02007 .[10] E. Sala, et al., Development of an underground HPGe array facility forultra low radioactivity measurements, AIP Conf. Proc. 1672 (2015) 120001. doi:10.1063/1.4928007 .[11] E. Sala, et al., Development of an underground low background instrumentfor high sensitivity measurements, J. Phys.:Conf. Ser. 718 (2016) 062050. doi:10.1088/1742-6596/718/6/062050doi:10.1088/1742-6596/718/6/062050