LArGe - Active background suppression using argon scintillation for the GERDA $0\nu\beta\beta$-experiment

0.58 MB·English

by  M. Agostini

Eur.Phys.J.CmanuscriptNo.
(willbeinsertedbytheeditor)
LArGe – Active background suppression using argon scintillation for the
nbb
GERDA 0 -experiment
MAgostini1,MBarnabé-Heider1,2,DBudjáš1,CCattadori3,AGangapshev2,4,K
Gusev1,5,6,MHeisel2,a,MJunker7,AKlimenko2,5,ALubashevskiy2,5,KPelczar8,S
Schönert1,ASmolnikov2,GZuzel2,8
5
1 1TechnischeUniversitätMünchen,Germany
0 2Max-Planck-InstitutfürKernphysik,Heidelberg,Germany
2 3UniversitàdegliStudidiMilanoeINFN,Milano,Italy
n 4InstitutforNuclearResearch,Moscow,Russia
u 5JointInstitutforNuclearResearch,Dubna,Russia
J 6NationalResearchCenterKurchatovInstitut,Moscow,Russia
7LaboratoriNazionalidelGranSasso,Assergi,Italy
1
8JagellonianUniversity,Cracow,Poland
1
Received:date/Revisedversion:27/05/15

]
t
e
d Abstract LArGe is a GERDA low-backgroundtest facility 1 Introduction
- to study novel background suppression methods in a low-
s
n backgroundenvironment,forfutureapplicationintheGERDA GERDAisanexperimenttosearchfortheneutrinolessdou-
i experiment.SimilartoGERDA,LArGeoperatesbaregerma- ble beta (0nbb ) decay of 76Ge. Bare high-purity germa-
.
s niumdetectorssubmersedintoliquidargon(1m3,1.4tons), nium(HPGe)detectorsenrichedin76Ge,whichserveboth
c which in addition is instrumented with photomultipliersto assourceanddetectorforthe0nbb -decays,aresubmersed
i
s detectargonscintillation light. Thescintillation signalsare inliquidargon(LAr).TheLArservesasahighpurityshield
y
h usedinanti-coincidencewiththegermaniumdetectorstoef- againstexternalradiationandasacoolantfortheHPGede-
p fectivelysuppressbackgroundeventsthatdepositenergyin tectors.The0nbb -signalisasharppeakintheenergyspec-
[ theliquidargon.Thebackgroundsuppressionefficiencywas trum at Qbb = 2039keV from the sum energy of the two

3 studied in combination with a pulse shape discrimination beta particles in a single HPGe detector. Details of the ex-
v (PSD)techniqueusingaBEGedetectorforvarioussources, perimental setup and performance are summarized in [1].
2
whichrepresentcharacteristicbackgroundstoGERDA.Sup- TheGERDA experimentfollowsastagedapproach:PhaseI
6
pressionfactorsofafewtimes103havebeenachieved.First hasbeenrecentlycompletedafteracquiringanexposureof
7
5 background data of LArGe with a coaxial HPGe detector 21.6kg·yr and a background count rate at Qbb of 1·10−2
0 (withoutPSD)yieldabackgroundindexof(0.12−4.6)·10−2 cts/(keV·kg·yr)after pulse shape analysis [2, 3]. No signal
1. cts/(keV·kg·y)(90%C.L.),whichisatthelevelof GERDA was observed and a limit for the half-life of T0n > 2.1·
1/2
0 PhaseI.Furthermore,forthefirsttimewe monitorthenat- 1025yr (90% C.L.) was achieved [4]. PhaseII is currently
5 ural42Arabundance(parallelto GERDA), andhaveindica- underpreparation:the goalis to explorehalf-livevaluesin
1
: tion for the 2nbb -decay in natural germanium. These re- therangeof1026yrbyfurtherreducingthebackgroundby
v sultsshowtheeffectivityofanactiveliquidargonvetoinan one order of magnitude to ≤10−3 cts/(keV·kg·yr), and by
i
X ultra-low background environment. As a consequence, the collectinganexposureofupto100kg·yrquasibackground
r implementationofaliquidargonvetoinGERDAPhaseIIis free.
a
pursued.
1.1 BackgroundsuppressioninGERDAPhaseII
PACS 23.40.-sb decay; double b decay; electron and
muoncapture·27.50.+emass59≤A≤89·29.30.KvX- Toreachthisdemandingbackgroundcountrate,severalex-
andg -rayspectroscopy·29.40.McScintillationdetectors perimental modifications with respect to PhaseI are being
implemented:the mostimportantare (1)the additionalde-
ploymentofapproximately20kgnovelthick-windowBroad-
EnergyGermanium(BEGe) detectors with highly efficient
ae-mail:[email protected] pulse shape discrimination (PSD) performance [5, 6], and
2
(2)theimplementationofasensorsystemtodetecttheliq-
uidargonscintillationlightinanti-coincidencewiththeger-
maniumdetectorsforbackgroundsuppression,firstdemon-
stratedinreferences[7,8]. lock
LArGe, the Liquid Argon Germanium test facility of forGe-detectordeployment ✲
GERDA, was constructed to study these novelactive back- coppercryostat
groundsuppressionmethodsin a low-backgroundenviron- withWLSmirrorfoil
ment[9, 10]. Similar to GERDA, bare Ge-detectorsare op- LArvolume1m3(1.4t) ✲
eratedinLArGein1m3 (1.4tons)ofliquidargon,whichin PMTs
additionisinstrumentedwithphotomultipliertubes(PMT). 9×8"ETL9357 ✏✏✶
✏
Thesetupislocatedundergroundat3800mw.e.attheLab- coatedwithWLS ✏✏
oratoriNazionalidelGranSasso(LNGS),andhasbeentak- detectorstrings ✏✏✏✶
✏
ing data since May 2010. The data presented in this pa- upto9Ge-detectors ✏✏
perdemonstratesthattheargonscintillationvetotechnique gradedshield
works very efficiently both alone and in combination with 15cmcopper,10cmlead,
PSDappliedtoBEGedetectorsignals. 23cmsteel,20cmpolyethylene ✏✏✏✶
1.2 Conceptofliquidargonscintillationveto
Fig.1:CutawayviewoftheLArGesetup.
Itiswellknownthatliquidargonemitsscintillationlightin
response to ionizing radiation.Up to approximately40000
photons,peakedatawavelengthof128nm(XUVphotons), 2.1.1 Shielding
areemittedper1MeVbeta/gammaenergydeposition[11].
The graded shield of increasing radiopurity is designed to
Backgroundeventsoftenhaveenergydepositionoutsidethe
have the gamma background dominated by the innermost
Ge-detectorinthesurroundingmedium,inourcasethescin-
tillating LAr. Conversely,0nbb -eventsare confined to the layerof15cmelectrolyticcopper.Thisisfollowedby10cm
oflow-activityleadand23cmofsteel.Theoutermostlayer
Ge-detector, so that no scintillation light is triggered. An
consists of 20cm polyethylene to attenuate neutrons. To-
observation of the light is therefore a good indicator for a
gether,allthreelayersattenuateanexternalgamma-rayfrom
background event, and can be used to veto the coincident
Ge-signal.For thatpurpose,the LAr mustbe instrumented the2615keVlineof208Tlto5·10−8 oftheinitialflux.The
to detect the light. In case of LArGe the light is shifted in purestshieldingisprovidedby>40cmLArinsidethecryo-
stat. At the inside of the cryostat, the PMT glass and the
itswavelengthtomatchthesensitiverangeofthePMTsand
VM2000 mirror foil yield the highest radioimpurities. To
isguidedtothePMTswithmirrorfoilontheinnercryostat
maintainsufficientshielding,thedistanceofthePMTstothe
wall.
Ge-detectorsischosenaslargeas90cm.Forthemirrorfoil,
thelowareadensityleadstoasurfaceactivitybelowthatof
2 Experimental thecopper.TheradiopurityscreeningresultsoftheseLArGe
componentsaregiveninTable1.Throughoutthemeasure-
2.1 Detectordesign ments presented here, a part of the steel and polyethylene
shield was not completed yet. Only a small impact on the
AnillustrationoftheLArGesetupisshowninFig.1.Avac-
backgroundrunisexpectedthereof.
uuminsulatedcoppercryostatatthecentercanhold1000l
(1.4t)ofLAr.Nine8"ETL9357photomultipliertubesare
2.1.2 Cryostat&cryogenics
immersedintotheLArfromthetop.Theinnercryostatwalls
arelinedwithVM2000radiantmirrorfoil1.PMTsandmir-
The main body of the double-wall cryostat is made from
rorfoilarecoatedwithTPB/polysterenewavelengthshifter.
electrolyticcopper.Ithasaninnerdiameterof90cmanda
Thecryostatissurroundedbyapassiveshieldagainstexter-
height of 210cm. Heat loss is primarily prevented by the
nalbackground.A doublechamberlock system on the top
insulation vacuum of 10−5 to 10−6mbar. The uppermost
servesasanaccessportforthedeploymentofGe-detectors
40cm collar and bellow are made from stainless steel to
and internal radioactive sources for calibration. The cryo-
minimizethethermalconductivitytothetop.Thecryostatis
genicinfrastructure,aslowcontrolsystem,andtheDAQare
closedbya38mmthickflangefromelectrolyticcopper.In-
locatedadjacenttothissetup.
fraredshieldsfromthincopperfoilaremountedbothinside
1productofthecompany3M the insulation vacuum, and below the top flange. A PMT
3
Table 1:Gamma-spectrometrymeasurementsofradioimpuritiesin LArGematerials. Givenisthe specific activityin units
[mBq/kg].Theoriginofthematerialsis;CopperLENS:NOSVcopperfromNAAHamburg.LeadLENS:‘DoeRun’-quality
fromJLGoslar.Steel(shield):carbonsteelfromDillingerHütteGTS.PE(shield):polyethylenefromSimonaAG.VM2000:
radiantmirrorfoilfrom3M(measuredwithglue,mountedwithglueremoved).PMTglass(bulb):ultra-lowbackgroundglass
fromETL(nowETEnterpriseLtd.).
material specificactivity
226Ra 228Ra 228Th 40K 60Co others
copperLENSa,1 <0.016 <0.019 <0.088 <0.01
leadLENSa,1 <0.029 <0.022 0.44(14) 0.18(2) 27(4)·103210Pb
steel(shield)b,2 2.04(33) 1.63(41) 5.34(69) <4.2 <0.3
PE(shield)b,2 <2.5 11.2(32) <3.4 10.8(67)
VM2000a,2 <1.6 <2.2 <1.2 140(10) <0.44 <0.45137Cs
PMTglass(bulb)b,2 2010(190) 2010(480) 210(60) 1750(430) <370137Cs
ameasuredwiththeGEMPIspectrometeratLNGS.[12] 1upperlimitsgivenfork=2,97.7%C.L.
bmeasuredintheLow-LevelLabatMPIKHeidelberg.[13] 2upperlimitsgivenfork=1.645,95%C.L.
supportstructure of copperand PTFE rests on a rim at the
copper-steel transition. The filling level of the LAr is ad-
justed slightly abovethe PMTs’ equator,leaving the upper
partofthecryostatinthegasphaseofargon.Astrongtem-
peraturegradientbuildsupfromthecopperatLArtempera-
ture,acrossthestainlesssteeltothetopflange,whichitself
staysabovethefreezingpoint.Thetotalheatloadis∼90W.
Anactivecoolingsystemcoolsthecryostatbyevaporat-
ing liquid nitrogen(LN ) in an integratedcoolingspiralin Fig.2:Left:CryostatbeeinglinedwithVM2000mirrorfoil.
2
thesteelcollar.Theinnermostinfraredshield(3mmthick) Right:BottomviewofthePMTsinsidethecryostat.
is thermallycoupledto this heat exchangerto preventheat
lostthroughradiation.Allrelevantcryogenicparametersare
compiledinaslowcontrolsystem,whichregulatesthecool-
ing power via the flow of LN . During normaloperationa
2
LN flowof 2.5kg/his sufficienttoreducetheLAr lossto
2
zero, which allows a continuous operation of LArGe. The
masscopperholdersinaseparatecleanbench.Theyaretrans-
workingpressureofthecryostatiskeptat30−70mbarover-
ferredtothemainlockusingatransportationcontainer,keep-
pressure, to preventgaseousimpuritiesfrom the outside to
ing them in gaseous nitrogen atmosphere at all times. The
entertheLAr.
mainlockandtheopenvolumebetweencryostatandshield
The filling procedure for the cryostat requires special
are permanentlyflushed with gaseousargonat slight over-
precaution to prevent contamination of the LAr. The mea-
pressure. The pressure gradient from the cryostat, via lock
surements presented in this article were performed in LAr
andshield,totheoutsideactsasasafeguardagainstair-trace
5.5 (purity 99.9995%). Against traces of humidity several
contaminationoftheLArandavoidsbackgroundfromair-
pumping-flushing cycles were performed with gaseous ar-
borne222Rn.
gon,whilethecryostatwasheatedto>40◦C.Topreventra-
dioactivebackgroundfromradon,theargonwasfilledthrough
an active-charcoaltrap (602g of CarboAct), followedby a
TheGe-detectorsaredeployedintothecryostatapprox-
PTFEparticlefilter.
imately equally spaced between the bottom and side walls
(distance d≈45cm), leaving 90cm of LAr on top towards
2.1.3 Lock&sourceinsertionsystem the PMTs. Internal‘close-by’wire sourcesare inserted di-
rectlyadjacenttotheGe-diodesintheargon(d≈7cm).Ex-
The lock on top of the LArGe setup servesfor the deploy- ternalsourcesaredeployedthroughaccessportsintheshield
ment of Ge-detector strings and internal radioactive wire directlyadjacenttotheouterwallofthecryostat(d≈50cm).
sourcesintothecryostat.Sofar,onlyonedetectorhasbeen Theverticalpositionofallsourcesmatchesthecenterofthe
inserted at a time. The Ge-detectors are mounted to low- Ge-detector.
4
for this low Y are: (1) only 5 out of 9 PMTs were opera-
tional at low temperature4, (2) trace contaminationsof the
argoncanstronglyquenchthescintillationandsignificantly
Fig.3:Schematicviewof shortentheattenuationlengthofthe128nmphotons.Anin-
the BEGe detector (grey dicatorforthelightquenchingisthelifetimet ofthetriplet
colour) in a low-mass states of the argon excimers, which can be measured from
copper-holder. The CC2 recordedscintillationwaveforms.Duringdatatakingt was
preamplifier is placed in monitoredintherangebetween450to550ns,comparedto
a grounded copper box t =1590ns[14]incleanargon.
mountedabovethedetec-
tor (top right). No elec-
tronic cabling is shown. 2.3 Germaniumdetectors
A close-by wire source
with copper weight is Amodifiedthick-windowbroad-energygermanium(BEGe)
hangingontherightside. detectorhasbeenusedtocarryoutthesuppression-efficien-
cymeasurementsdescribedinsection3.Itisap-typediode
of878g byCanberraSemiconductors,N.V.Olen/Belgium.
2.2 Lightinstrumentation The depletion voltage is +4kV [15]. A small p+ contact
leadsto a strongweightingfield close tothe read-outelec-
Thelightread-outisdoneusingnineETL29357photomul- trode,whichallowsgoodpulseshapediscrimination.Apro-
tiplier tubes. The 8" (200mm) diameter end window with totypeoftheGERDAPhaseImulti-channelcharge-sensitive
alowresistancebialkaliphotocathodeissensitivetowave- preamplifier (CC2) with integrated FET and RC feedback
lengthsfrom275−630nm3.Thepeakquantumefficiencyof component[16] is used and mountedclose-by the detector
thePMTsis18%at370nm.Sincetheglassoftheendwin- – see Fig. 3. A low-mass copper-holderis used to support
dow is not transparent for the 128nm scintillation light, it thediodeandsubmerseit‘naked’intoLAr.SignalandHV
mustbe coveredwith wavelengthshifter(WLS). A picture areconnectedvia pressurecontactsto the detectorsurface.
ofthePMTsandmirrorfoilinthesetupisshowninFig.2. A detector resolution of 1.99keV FWHM at 1332keV is
ThePMTsareequippedwithacustommadevoltagedi- achieved in this setup, compared to 1.63keV in a vacuum
vider with a wide linear dynamic range from 2mV to 4V. cryostatwiththesamedetector[5].
Cleanpulseshapesareobtainedbyoperatingwithnegative Forthebackgroundmeasurementsdescribedinsection4
HV on the photocathode. The voltage divider is based on theBEGedetectorisreplacedbythecoaxialp-typeHPGe-
a 0.5mm thin CuFlon(cid:13)R printed circuit board with compo- diode GTF44. In contrast to BEGe it has a low intrinsic
nentsselectedforlowmassandradiopurity. background(60Co,68Ge)andahighmass(2465g),whereas
ThewavelengthshifterusedtocoatMirrorfoilandPMTs pulseshapediscriminationis inferiorandnotappliedhere.
consistsof10%fluorescentdye(tetraphenylbutadiene)em- The detector has been modified by the manufacturer Can-
beddedintoapolymermatrixofpurifiedpolystyrene.Both berraforthebareoperationinLAr[17].Itisequippedwitha
substancesaredissolvedintolueneanddehumidifiedusing low-backgroundversionoftheCC2chargesensitivepream-
gaseous nitrogen. A homogeneous coating is achieved by plifier.Withinaninvestigationprogramof42Arbackground
pullingthe Mirrorfoilthrougha bathin 45◦ angle.Incase thediodehasbeenencapsuledin a groundedFaradaycage
ofPMTs,thephotocathodesaresimplybrushedwiththeso- madefromthinlayersofPTFEandcopper.
lution in several thin layers. The coating thickness is esti-
matedfromtheconsumptionofthesolution.Athicknessof
1−4m m is choosen as a compromise between shifting ef- 2.4 Dataaquisitionandanalysismethods
ficiency and mechanical stability in cryogenic liquid. The
coated foil has a specular reflectivity of ∼95%at the peak 2.4.1 CombinedGe-detectorandPMTreadout&
fluorescencewavelengtharound420nm. waveformprocessing
A characteristic quantity of the setup is the photoelec-
AblockdiagramofDAQandfront-endelectronicsisshown
tron(pe)yieldY. For the measurementsdiscussed herewe
in Fig. 4. The Ge-detector is supplied with bias HV (Iseg
hadY≈0.05pe/keV,whichcorrespondstoanenergythresh-
oldof1/Y≈20keVforasinglephotoelectron.Tworeasons 4Some PMTs exhibit light production due to discharges within the
glassbodyatcryogenictemperatures andwereeitheroperated atre-
2now:ETEnterpriseLtd. ducedgainornotatall.ThesefindingsledtoanR&Dprogramtosolve
3wavelengthrangeoverwhichquantumefficiencyexceeds1%,accord- this issue for next generation PMTs being implemented in GERDA
ingtothedatasheet. PhaseII.
5
Table 2: Summary table of source activities and measure-
ment time with corresponding signal rates of Ge-detector,
PMTs,andthepulseracceptance.
source pos. activity meas. n BEGe n PMT e
trig trig acc
[kBq] time[d] [Hz] [kHz] [%]
60Co int 0.314 13.06 26.9 5.8 96.6
ext 94.9 2.49 6.9 39.2 78.7
226Ra
int 0.934 1.54 65.4 8.1 94.2
ext 38.9 8.30 4.5 43.3 78.3
228Th
int 0.63 1.96 47.5 8.6 95.7
thebaselinespread,correspondingto∼20%oftheaverage
Fig. 4: Block diagram of the DAQ with Ge-detector and
singlephotoelectronamplitude.
PMTreadout.VD=voltagedivider;PA=preamplifier.
The veto acceptance is the complementary probability
for an event being vetoed by random coincidences (p ),
rc
NHQ 225M NIM).A pulsersignal(Ortec Mod. 448NIM) eacc=1−prc. It is measured by applying the veto cut on
can be fed into a test inputof the preamplifier. The output pulsersignalsorsinglefullenergypeaks.Therandomcoin-
signal is amplified without shaping by a custom-made lin- cidenceprobabilitycanalsobeestimatedviaprc≈n tPriMgT·D t,
earamplifier,andfedintoaFADC(StruckSIS3301VME; usingthePMTtriggerrateandthevetowindowsize.Itturns
8channels,14-bit,105MS/s).ThePMTHVissuppliedby outthatallmethodsyieldconsistentresults.Vetoacceptance
IsegNHQ204M/225M.Thesignalsareamplifiedbyafac- valuesin measurementswith differentsourcesare listed in
torten(PhillipsScientificMod.776NIM)andmergedina Table2.
linearfan-in(LeCroyMod.428FNIM).Theresultinganalog- Thesuppressionfactoristheratioofeventsintheunsup-
sumisamplifiedinanothercustom-madeanalogshaper(NIM) pressed(N )versusthesuppressed(N )spectrum.Tomake
0 S
with a shaping constant of a few 10ns, to match the dy- SF independentof the source strength in the measurement
namicrangeandsamplingrateofthesubsequentFADC.The it is weighted by e , hence SF=e ·N /N . Suppres-
acc acc 0 S
FADC is internally triggered on the Ge-signal, and simul- sionfactorsfortheROIaredeterminedina70keVwindow
taneouslyrecordsGe- andPMT waveformsof40m strace- aroundQbb .UncertaintiesarecalculatedaccordingtoPois-
lengthwith100MHzsamplingrate.TheFADCaquisitionis soncountingstatistics.
controlledbyacustom-madesoftwarebyMIZZIComputer
SoftwareGmbH[18].
The offline analysis of the digitized Ge-waveforms is
performedwiththesoftwareframeworkGELATIO[19]and 2.4.3 Pulseshapediscrimination
followingtheproceduredescribedinreference[20].Theen-
ergy deposited in the Ge-detectors is reconstructed by ap- TheobjectiveofPSDistodistinguishthesinglesiteevents
plyinganapproximatedGaussiandigitalfilter.Non-physical (SSE)ofthebb -decayfrommultisiteevents(MSE)ofcom-
signalsduetoinstabilitiesoftheread-outelectronicsorelec- mon gamma-backgroundwith multiple interaction vertices
tromagneticpick-upnoiseare rejectedduringthedata pro- withintheGe-diode.SSEandMSEcanbedistinguishedby
cessingbyapplyingasequenceofqualitycuts;includingthe themaximumamplitude(A)oftheircurrentpulse,normal-
flatnessandnoise-levelofthebaseline,thepositionandrise ized by the energy (E). The so-called cut parameter ’A/E’
time of the leading edge,and the fall time of the exponen- hasbeenestablishedforthispurpose[5].ThePSDcutiscal-
tialdecaytailofthechargesignal.ThePMTwaveformsare ibratedto90%acceptanceonthedoubleescapepeak(DEP)
analysedwithoutfilteringor quality cuts. Merelythe base- ofthe2615keV208Tlline,whichbyitsnatureisdominantly
lineandvetoconditionaredetermined. ofSSEcharacter.Asdiscussedin[2],inaBEGedetectorthe
DEPisagoodproxyfor0nbb andtheiracceptancesagree
withinabout1%. UncertaintiesofPSD suppressionfactors
2.4.2 LArscintillationvetocut
includebothstatisticalandsystematicuncertainties.
Thevetoconditionisfulfilledwhenoneormorephotoelec- In addition,the combinedsuppressionof LAr veto and
trons are detected in a 5m s window aroundthe Ge-trigger. PSDisdeterminedfromthespectra.Duetothestrongsup-
Thresholdand windowsize are optimizedto maximizethe pression of close-by 60Co and 228Th the analysis window
product of suppression factor SF at Qbb and the veto ac- forthesesourcesisextendedfrom70keVwidthto200keV,
ceptancee .Thethresholdisslightlyabovenoiseat5s of excludingthesingleescapepeakof208Tlat2104keV.
acc
6
106
105
104
103
102
eV] 101
k
5 100
#/
s[ 106
nt
ou 105
c
104
103
102
101
100
0 500 1000 1500 2000 2500 3000 3500
energy[keV]
(a)228Thsourceclose-by(top)&external(bottom)
106
105
104
103
102
eV] 101
k
5 100
#/
s[ 106
nt
ou 105
c
104
103
102
101
100
0 500 1000 1500 2000 2500 3000 3500
energy[keV]
(b)226Rasourceclose-by(top)&external(bottom)
106 withoutveto
V] 105 withLArveto
ke 104 withPSDcut
5 withveto&PSD
[#/ 103
s
nt 102
u
co 101
100
0 500 1000 1500 2000 2500 3000 3500
energy[keV]
(c)60Cosourceclose-by
Fig.5:Fullenergyspectraofthesourcemeasurementsin5keVbinning.The228Thand226Raspectrafeatureapulsersignal
at3MeV.Incaseof60Co,asmallbackgroundpeakof208Tlisvisibleat2615keV.Thecolourcodeisshownincanvas(c).
7
3 Measurementofsuppressionfactors
104
Energyspectra of various sources in close-by and external
103
position(see 2.1.3) were recordedwith the BEGe detector.
102 The sources represent characteristic background contribu-
tionsinGERDA.Whilefromexternalsourcesonlygammas
V] 101 canenterthecryostat,close-bysourcesareencapsuledince-
e
1k 100 ramics(D=1mm)andthinsteel(R=0.25−0.5mm),thus
#/ allow some high-energy beta particles to enter the LAr as
s[ 104
nt well.Thesourceactivitiesarechosentobalancehighsignal
u
co 103 rates in the Ge-detector with random coincidences (2.4.2)
– see Table 2. Without a source the pulser acceptance is
102
97.3%,andthePMTtriggerrate(∼5kHz)isdominatedby
101 darknoiseandthedecayof39Ar(1.4kBq).
TheLArvetoandPSDcutsareappliedtoeachmeasure-
100
ment in the whole energy range. While the achieved sup-
1900 1950 2000 2050 2100 2150 pressionfactorintheROIistheultimatelyrelevantnumber,
energy[keV] other energy regions of distinct gamma lines illustrate the
(a)228Thsourceclose-by(top)&external(bottom) differentandcomplementarysuppressionmechanismunder-
lyingPSDandLArveto.
103
3.1 Th-228suppression
102
Even after careful material screening and selection, 228Th
101 anditsprogeniesfromthenaturaldecaychainarepresentat
V] tracelevelsin the constructionmaterialsof GERDA. Back-
e
1k 100 groundfromsourcesclose-bythegermaniumoriginatesfrom
[#/ detectorholders,cablesandfront-endelectronics[3].These
nts 103 components are immersed in the liquid argon and are re-
u
o
c ferredtoas‘close-bysources’.Conversely,‘externalsources’
102
arelocatedinthecryostat,itsneckandthephotomultipliers
oftheLArinstrumentationitself.Thecorrespondingenergy
101
spectraofaclose-byandexternal228Thsourceareshownin
Fig.5a.
100
Bothspectraaredominatedbythe2615keVgammaline
1900 1950 2000 2050 2100 2150 of 208Tl and it’s single- and double escape peak (SEP at
energy[keV]
2104keV,DEPat1593keV).Sincethedoubleescapepeak
(b)226Rasourceclose-by(top)&external(bottom) isdominantelyofSSEnature,andassuchusedtocalibrate
thePSDacceptance,itremainspracticallyunsuppressedby
PSD(Fig.7).Ontheotherhand,thetwo511keVannihila-
103
tiongammastriggertheLArvetoreliablysuchthatthepeak
V] vanishes.IncontrasttotheDEP,theneighbouring1621keV
e
k 102
1 fullenergypeakfromthe212Bidecayisaffectednearlyop-
#/
[ posite by both cuts: all the energyof this line is deposited
nts 101
u inthegermaniumdetector,leavingnoneforthesurrounding
o
c
LAr to create scintillation. Moreover, the transition is not
100
part of a gamma cascade, thus no additional energy depo-
1900 1950 2000 2050 2100 2150
sitionintheLArisoccuring.Hence,theLArvetodoesnot
energy[keV] comein.ThesuppressionfactorofPSDisabouttenforboth
(c)60Cosourceclose-by source positions. The situation is the same for PSD on the
2615keVgammaline.However,theLArvetocanalsosup-
Fig.6:Close-upview ofthe ROI atQbb (2039keV)ofthe pressthispeak,becausethegammaisemittedinacascade
sourcespectrain1keVbinning.ForcolourcodeseeFig.5. precededbyothergammas(coincidentgammas),whichcan
8
The predominant isotope in this chain that contributes
to the regionat Qbb is 214Bi, with severalgammalines up
104
V] to 3184keV. The suppression of gamma peaks follows the
ke 103 same logicas showcasedfor228Th:single lines notbeeing
1
[#/ 102 affectedbytheLArveto(e.g.1764keV,2204keV,2448keV),
nts asopposedtolinesemittedaspartofagammacascadeand
ou 101 thereforeincoincidence(e.g.609keV,1120keV).TheLAr
c
100 suppressionfactorsin the ROI are 4.6±0.2(close-by)and
3.2±0.2 (external), and about four for PSD (see Table 3,
1575 1600 1625 1650 2575 2600 2625 2650
Fig. 6b). The LAr suppression is much inferior compared
energy[keV]
to 228Th for mainly two reasons: (1) all gamma lines with
Fig. 7: Left: the double escape peak of 208Tl at 1593keV sufficientenergyto create Compton eventsat Qbb are sin-
and the 1621keV single gamma line of 212Bi. Right: the gle,hencedeprivingtheLArvetoofapossibilitytovetoon
2615keV coincident gamma line of 208Tl. Both examples a coincident gamma. The lack of coincident gammas also
fromtheclose-bysource.SamecolourcodeasinFig.5. makesthe vetoless dependenton the source position.And
(2)thegammashaveonlylittleenergyexceedingQbb ,thus
onlylittlelightiscreatedtotriggertheveto.Despitethein-
themselves trigger the LAr veto. The suppression is much dividualsuppressionofLArvetoandPSDbeingmoderate,
strongerfortheclose-bysource(SF=47)thanfortheexter- theircombinationagainprovidesasuppressionwellbeyond
nal(SF =1.3),becauseintheformercasecoincidentgam- oneorderofmagnitude.
mashavelittlechancetoescapefromtheactiveLArvolume.
This instance can be exploited to identifythe location of a
228ThbackgroundsourceviatheLArsuppressionfactors.
TheROIofthe0nbb -decayisdominatedbyaflatComp-
3.3 Co-60suppression
tonregionof208Tlbeforeandafterthecutsinbothspectra
(Fig. 6a). The suppression factors of the LAr veto cleary
Cosmogenic60Co is formedin Ge-detectorsandtheir cop-
differ for the close-by (1180±250) and external (25±1)
perholdersduringproductionaboveground.While60Coin
sources, while beingquasi-independentof the source loca-
the detectors can create background directly via their beta
tionforPSD(2.4±0.1and2.8±0.1respectively–seeTable
decay, 60Co background from the holders relies on gam-
3).TheLArsuppressionisstronglyenhancedintheComp-
mas: only two gamma lines with significant branching ra-
ton region of the 2615keV line, as compared to the full
tioareemittedafterthedecayof60Co.Sincetheirenergies
energy peak itself: since a fraction of about 2MeV is de-
of 1173keV and 1332keV are below Qbb , they can cre-
positedin the Ge-detectors,an excess energyof ∼600keV
ate backgroundeventsonly via summation. As summation
isdepositedinthe liquidargonin its vicinity,providingan
strongly dependson the solid angle and the angular corre-
additional handle for the LAr veto to act upon. The sup-
lation of the gammas, only 60Co sources close-by the Ge-
pressionfactorsofthecombinedLArvetoandPSDcutsare
detectors are of concern for the GERDA background. The
5200±1300(close-by)and129±15(external),thusprovid-
energyspectrumofclose-by60CoasmeasuredinLArGeis
ing a strong suppression on all 228Th backgroundsources.
showninFig.5c.
Since only 5 counts survive the combined suppression of
The energy regionabove the two gamma lines is dom-
theclose-bysource,theanalysiswindowwasextendedfrom
inated by the summation spectrum, which expands up to
70keV to 200keV width aroundQbb . Still, the numberof
the summation peak at 2505keV5. Similar to 214Bi, LAr
counts after the cut (=15) dominate the uncertainty of the
suppressionfactor. suppression in the ROI (Fig. 6c) happens mainly via the
gammaenergyexceedingQbb ,whichincaseof60Coisal-
most 500keV. This higher energy and multiplicity reduces
3.2 Ra-226suppression thechanceforitbeeingdepositedindeadvolumeratherthan
activeargon,henceleadingtoasuperiorsuppressionfactor
226Ra is a long lived progeny of 238U in the natural decay of27±2comparedto226Ra(Table3).ThePSDcutworks
chain. Similar to 228Th, it is present at trace levels in all veryefficientlywithasuppressionof76±9,asbyconstruc-
constructionmaterials.226Radecaysto222Rn,whichisara- tion summation events are MSE. Again, the combined cut
dioactivenoblegasthatdiffusesintotheLAroftheGERDA canrejectbackgroundbythreeordersofmagnitude.
cryostat and is a potential source of background. Energy
spectra were recorded with LArGe for a close-by and ex- 5Thepeakabove2505keVisthe2615keVlineof208Tlbackground
ternal226Rasource,bothofwhichareshowninFig.5b. fromthefront-endelectronicsusedwiththeBEGedetector.
9
Table 3: Summary table of suppression factors for differ- Table4: Summarytableofthebackgroundindexfordiffer-
ent sources in the ROI around Qbb at 2039keV. The com- entchoicesoftheROI.AllROIarecenteredaroundQbb and
binedsuppression(‘total’)rangesfromafew101 toseveral excludethe2204keVlineof214Bi.Confidenceintervalsare
103 andgivestheorderofmagnitudebywhichbackground givenwith90%.
sourcescanbesuppressed.
region counts[#] BI[cts/(keV·kg·yr)]
source position suppressionfactor ofinterest noveto veto noveto veto
LArveto PSD total 100keV 14 0 0.44(12) <7.2·10−2
60Co int 27±2 76±9 3900±1300 200keV 30 1 0.47(9) 0.17−6.8·10−2
ext 3.2±0.2 4.4±0.4 18±3 300keV 40 1 0.43(7) 0.12−4.6·10−2
226Ra
int 4.6±0.2 4.1±0.2 45±5
perimentaldataofLArGeandwillbereportedinapublica-
ext 25±1 2.8±0.1 129±15
228Th tiondedicatedtotheLArvetoinGERDA.
int 1180±250 2.4±0.1 5200±1300
3.4 Conclusiononsuppressionfactors
4 Backgroundmeasurementsinnaturalgermanium
The suppressionfactorsof the differentsourcesin the ROI
of the 0nbb -decay are summarized in Table 3. The large LArGe has been designed to demonstrate the applicability
variation of the suppression factors is consistent with our of the LAr veto in an ultra low-background environment.
understandingoftheunderlyingphysics,asdescribedinthe Forthatpurposeameasurementwithasemi-coaxialdetector
previoussections. withnaturalisotopiccomposition(GTF44,see2.3)andim-
ThecombinationofLArvetoandPSDprovestobemore provedradiopurefront-endelectronicswasconducted.The
powerfulthanwouldbeexpectedfromindependentcuts:the LArvetowasoperatedunderthesameconditionsasprevi-
mean average of the combined suppression in the ROI is ouslydescribed,exceptforanexchangeoftheLArbyhigh
enhancedbya factor1.84±0.17,comparedto the product purityargon6.0.Thevetoacceptanceof91%isdetermined
oftheindividualsuppressionfactorsofPSDandLArveto. bythepulser.Thedetectorresolutionis3.5keVat1332keV,
This means that event classes leading to rejection by one noPSDwasapplied.
or the other cut are anti-correlated. For example, a Comp-
ton event at Qbb from close-by 208Tl leaves the 2.6MeV
gammatodeposit∼600keVoutsidetheGe-detector,along- 4.1 Backgroundcomponents
sideits583keVcoincidentpartner.However,iftheeventat
Qbb resultsfromthesummationofthetwogammas(amul- Thefullenergyspectrumwithanexposureof116kg·d(life
tisite event likely to be rejected by PSD), a single gamma time 47.05 days) is shown in Fig. 8. The dominant back-
of higher energy ∼1.2MeV can leave the detector, and is groundsourceabove1.5MeVisthe2615keVlineof208Tl
more likely to escape the active LAr volume than two en- and its Compton continuum.The suppression factor of the
ergetically lower gammas. Hence, this event class is anti- fullenergypeakis1.52±0.62,whichisinagreementwith
correlatedinLArvetoandPSD.Ananalogueanalysisoffull thecorrespondingvalue1.28±0.01obtainedintheexternal
energypeaksoftheinvestigatedsourcesreturnsanaverage source measurement. Hence, the suppression factor points
‘enhancement’of1.007±0.015.Thisisconsistent,sinceno towardsadistantoriginofthisbackground,presumablythe
(anti-)correlatedeventtopologiesareexpectedhere. PMTs.Otherprominentbackgroundsourcesare40K(1461keV)
The suppression factors measured here sketch the or- and214Bi(1764keVand2204keV).Theirlinesappearonly
der of magnitude by which active background rejection in inthevetoedspectrum:whilethecontinuousComptonback-
GERDAmaybeachievable,indicatingthatthegoalforPhaseII, groundisvetoedeffectively,singlefullenergypeaksarere-
namely to reduce the background by one order of magni- jected only by random coincidences. Cosmogenic 58Co is
tude,isinreach.Duetoahigherargonpurityweanticipate found at 811keV, likely sitting in the activated copper of
to achievea betterphotoelectronyieldin GERDA, support- thedetectorencapsulation.Atlowenergiesthespectrumis
ingthesuppressionof226Raandotherbackgroundsources dominated by the 39Ar beta spectrum (Qb -value 565keV)
that create little light per event at Qbb . Other determining andaccompanyingBremsstrahlungphotons.
factors for photoelectronyield are the geometry of the ac-
tiveLArvolume,theGe-detectorstringswithin,andtheef- 4.1.1 2nbb contribution
ficiency of light detectors and wavelength shifters. A full
MC campaign using photon tracking has been carried out Thestrongsuppressionin thevetoedbackgroundspectrum
to study and optimize a veto design for GERDA PhaseII. makesitpossibletoobservethe2nbb spectrumeventhough
Thesesimulationsareinreasonableagreementwiththeex- thedetectorismadefromnon-enrichednaturalgermanium.
10
103
4
withoutveto
3
withLArveto
2
V] 102
e 1
k
5 0
#/
[ 1800 1900 2000 2100 2200 2300
nts 101
u
o
c
100
0 500 1000 1500 2000 2500 3000
energy[keV]
Fig.8:BackgroundspectrumofGTF44withanexposureof116kg·d(lifetime47.05days).Theinlayshowsaclose-upview
oftheregionofinterestaroundQbb at2039keV.
Thepredictionofthe2nbb spectrumbasedon[21,22]for aPoissonvariableintheabsenceofbackground’usingfre-
thisdetectorinfersthat69outof135observedcountsinthe quentistsstatisticsaccordingto[23].Theratioofcountsbe-
continuum from 500keV (above dominant bremsstrahlung foreandaftertheLArvetoyieldabackgroundsuppression
from39Ar)to2100keV(abovethe2nbb endpoint)areex- byoneorderofmagnitudeormore.
pectedtostemfrom2nbb decays.
4.1.2 42Kabundance
5 Conclusion
AuniquebackgroundtoGERDA andLArGeisthatof42K,
identified through its gamma line at 1525keV. 42K (Qb = TheLArGetestfacilityhasdemonstratedthegreatpotential
3525keV, T1/2 = 12.6h) is a b -emitting progeny of 42Ar of an active liquid argonveto for the suppression of resid-
(Qb =599keV,T1/2=32.9yr)withtracesexpectedinnat- ual background signals which deposit part of their energy
ural argon. In LArGe we observe 7 counts in the interval inLAr.ItisthefirsttimebareGe-detectorsareoperatedin
(1523−1527)keVaroundthepeak,outofwhich(1.35±0.27) a low-background environment with 1m3 of instrumented
countsareexpectedfromComptonbackground.Theproba- LAr.Thebackgroundsuppressionefficiencyhasbeenstud-
bility6toobserve≥7eventsfromthisbackgroundis0.08%. ied in combination with pulse shape discrimination (PSD)
Neglectingpossibleinhomogeneitiesofthe42K spatialdis- of a BEGe detector. Suppression factors have been mea-
tribution, the number of counts observed in the 1525keV suredfor severalsources(60Co, 226Ra, 228Th)representing
peak corresponds to an abundance 42Ar/natAr of about characteristicbackgroundsourcestoGERDAindifferentlo-
2·10−21g/g[10].AlongwithGERDA[3],thedatapresented cations (close-by and external). The strongest suppression
hereisthefirstpositivedetectionofnatural42Ar. factorswere obtainedfor combinedLAr veto and PSD for
close-by 228Th (SF ≈ 5200) and 60Co (SF ≈ 3900). The
combinedsuppressionofLArvetoandPSDismutuallyen-
4.2 BackgroundIndex hanced.Theparticularresponseofthedifferentsuppression
methods is a useful tool to understand the location of dif-
TheinlayofFig.8showsthe ROI ofthe0nbb -decay.Ina ferentbackgroundsevenin thecase oflowcountingstatis-
largeenergywindowof300keVcenteredaroundQbb only tics. In a low background measurement without PSD, the
oneeventsurvivestheLArvetocut.Dependingonthecho- LArvetohelpedtoachieveanexcellentbackgroundindexof
sen width for that region, the achieved background index (0.12−4.6)·10−2cts/(keV·kg·y)(90%C.L.).Theconfidence
after vetois about10−2 cts/(keV·kg·yr)– see Table 4. The intervalcoincideswiththebackgroundlevelofGERDAPhaseI
lowerlimitscoverabackgroundindexof10−2cts/(keV·kg·yr), <10−2 cts/(keV·kg·y), despite LArGe being a much more
whichisthedesigngoalofLArGeandGERDAPhaseI.The compactsetup.LArGehasthesensitivitytomeasurethenat-
90% confidence intervals are determined for ‘the mean of uralabundanceof42Arandthe2nbb -decayinnon-enriched
germanium.Asaconsequenceoftheseresults,anactiveliq-
6Theprobabilityiscalculatedfromagaussiandistributedbackground
g(l |m ,s )andpoissondistributedcountsp(n|l ),usingprob(n≥7)= uid argonveto has been developedfor GERDA and will be
(cid:229) ¥n=7R0¥ g(l |m ,s )p(n|l )dl . usedinPhaseIIoftheexperiment.