novel measurement method of heat and light detection for neutrinoless double beta decay

Novel measurement method of heat and light detection forneutrinoless double beta decay G.B.. So, Novel measurement method of heat and light detection for neutrinoless double beta decay,

Novel measurement method of heat and light detection for

neutrinoless double beta decay

G.B Kim, J.H Choi, H.S Jo, C.S Kang, H.L Kim, I.W Kim,

S.R Kim, Y.H Kim, C Lee, H.J Lee, M.K Lee, J Li, S.Y Oh,

J.H So

DOI: 10.1016/j.astropartphys.2017.02.009

To appear in: Astroparticle Physics

Received date: 29 July 2016

Revised date: 23 January 2017

Accepted date: 28 February 2017

Please cite this article as: G.B Kim, J.H Choi, H.S Jo, C.S Kang, H.L Kim, I.W Kim, S.R Kim,Y.H Kim, C Lee, H.J Lee, M.K Lee, J Li, S.Y Oh, J.H So, Novel measurement method of

heat and light detection for neutrinoless double beta decay, Astroparticle Physics (2017), doi:

10.1016/j.astropartphys.2017.02.009

This is a PDF file of an unedited manuscript that has been accepted for publication As a service

to our customers we are providing this early version of the manuscript The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form Pleasenote that during the production process errors may be discovered which could affect the content, andall legal disclaimers that apply to the journal pertain

ACCEPTED MANUSCRIPT

Novel measurement method of heat and light detection

for neutrinoless double beta decay

G.B Kima,b, J.H Choia,b, H.S.Joa, C.S Kanga,b, H.L Kima, I.W Kima,b,

S.R Kima,b, Y.H Kima,b,∗, C Leea, H.J Leea,b, M.K Leeb, J Lia, S.Y Oha,b,

de-48Ca-depleted elements This new detection method employs metallic magneticcalorimeters (MMCs) as the sensor technology for simultaneous detection of heatand light signals It is designed to have high energy and timing resolutions toincrease sensitivity to probe the rare event The detector, which is composed of

a 200-g40Ca100MoO4crystal and phonon/photon sensors, showed an energy olution of 8.7 keV FWHM at 2.6 MeV, with a weak temperature dependence inthe range of 10-40 mK Using rise-time and mean-time parameters and light/heatratios, the proposed method showed a strong capability of rejecting alpha-inducedevents from electron events with as good as 20σ separation Moreover, we dis-cussed how the signal rise-time improves the rejection efficiency for random co-incidence signals

background

∗ Corresponding author

Email address: yhk@ibs.re.kr (Y.H Kim)

ACCEPTED MANUSCRIPT

1 Introduction

Neutrinos are one of the elementary particles that compose the universe Inthe standard model (SM) of particle physics, they are considered to be masslessand electric-chargeless and to have half-integer spin However, a series of obser-vations on neutrino oscillation phenomena suggest that neutrinos have non-zeromass, and oscillate from one flavor state to others [1] The flavor states can beexpressed by a neutrino mixing matrix with mass eigenstates Although the mix-ing angles in the matrix and square mass differences of the three mass eigenstateshave been obtained recently [2], neutrino oscillation experiments do not providethe absolute mass scale of neutrinos Moreover, their fundamental particle type(Dirac or Majorana) remains unanswered [3]

Double beta (ββ) decay is an allowed nuclear transition for the isotopes for thatthe mass of the initial nucleus (A, Z) is larger than that of the final state nucleus(A, Z+2), but smaller than that of the intermediate state (A, Z+1) According tothe SM, a ββ decay process is always accompanied by emission of two electronsand two neutrinos expressed as 2νββ,

However, in the case that a neutrino is both massive and its own anti-particle (i.e.,

a Majorana particle), the following ββ decay process without neutrino emissioncan be allowed:

Observation of this neutrinoless double beta (0νββ) decay would provide anunambiguous answer to the Dirac-or-Majorana question Allowing the physicalprocess violating lepton number conservation would be a strong clue for matter-antimatter asymmetry in the present universe Moreover, the absolute mass scale

of neutrinos can be confined based on observation of 0νββ decay

The half-life of the 0νββ process can be expressed as

ACCEPTED MANUSCRIPT

where U ei are the PMNS (Pontecorvo-Maki-Nakagawa-Sakata) matrix elements

In the 0νββ process, the total decay energy (Q) is mostly carried by two electronswith negligible amount carried by a recoiled daughter nuclide Hence, the 0νββprocess will result in a peak at the end point of the ββ spectrum

Neutrinoless double beta decay is expected to be an extremely rare process

A Majorana neutrino mass of 50 meV corresponds to a half-life of about 1026

years in 100Mo 0νββ decay with some model dependance of the nuclear matrixelement [4] One 0νββ event of 1-kg100Mo would occur in about 20 years One ofthe strategies to enhance the sensitivity is to increase the detector mass because alarger number of source elements provides a higher 0νββ decay event rate Reduc-ing background events in the energy region of interest (ROI) is another key factorthat increases the detection sensitivity for this rare process Moreover, the energyresolution of the detector that defines the ROI can be an efficient and crucial pa-rameter High resolution measurement increases the accuracy of energy detection,and narrows the width of the ROI It results in reduction of number of backgroundevents in the ROI, particularly from irreducible 2νββ background events

For a non-negligible background condition, the experimental sensitivity to theMajorana neutrino mass can be written as,

imental sensitivity for Majorana neutrino mass is proportional to the 1/4 power

of the quantities On the other hand, in the case that the expected background ofthe detector in the ROI is less than 1 event during the measurement period, so-

called zero-background case, the sensitivity for the Majorana neutrino mass can

In order to increase the rate of 0νββ events, isotopically enriched100Mo is used

to fabricate CaMoO4 crystals Moreover, enriched 40Ca from 48Ca depletion is

ACCEPTED MANUSCRIPT

used to minimize interference from 2νββ signals of48Ca Consequently, doublyenriched40Ca100MoO4crystals are used

The choice of100Mo as a 0νββ candidate is advantageous The Q-value of the

100Mo decay is 3034.40(17) keV [8] which is sufficiently high to prevent ence from most of environmental γ-ray backgrounds The expected half-life ofthe100Mo 0νββ process is relatively short compared with other candidates [4, 9].Moreover, the high natural abundance of100Mo of about 9.6% does not requireextraordinary enrichment cost

interfer-Metallic magnetic calorimeters (MMCs) are used as the sensor technology forsimultaneous measurement of heat and scintillation-light signals MMCs havedemonstrated high energy resolutions in X-ray and alpha-particle detections [10,

11, 12, 13] The simultaneous measurement technique makes it possible to rate out background alpha signals in an event by event manner Moreover, the fastresponse time of MMC signals can minimize possible background from randomcoincidences of two 2νββ events

sepa-Several Mo-containing crystals, such as Li2MoO4, CaMoO4, MgMoO4, andZnMoO4, have been used for suitability tests of large-scale 0νββ search exper-iments [14, 15, 16, 17, 18] Using semiconductor-based neutron transmutationdoped (NTD) Ge thermistors as their thermal sensors of phonon and scintillationmeasurement, 6.3-keV FWHM resolution for the 2615 keV γ-line of 208Tl wasfound with a 330-g ZnMoO4 detector, where 18σ and 19σ event discriminationcapability were found from heat/light ratios and pulse shape parameters of β/γ and

αsignals, respectively [15] Although the phonon-scintillation detector with NTD

Ge readout provides high energy resolution and discrimination power, it showedlimited timing resolution where the rise-times of the phonon and light signals were

12 ms and 3.2 ms, respectively This limit on the rise-time of the signals nates from inefficient thermal coupling between phonons in the absorber crystaland conduction electrons in the thermistor Recently, another type of phonon-scintillation detector was tested for a CaMoO4 crystal with a NTD Ge sensor forheat signals and a millikelvin photomultiplier tube (PMT) readout for light signalswhere extreme timing resolution of the light signals was achieved [19], and twodecay constants of 41 µs and 3.4 ms were found for CaMoO4scintillation below

resolu-ACCEPTED MANUSCRIPT

crystal

Photon detector

Figure 1: The detector module with a 40 Ca 100 MoO 4 crystal and phonon/photon sensors The phonon sensor sits at the bottom of the module The connection between gold film and an MMC is

shown in the top-left The light detector, which is made of a Ge wafer and another MMC, covers

the top-side of the crystal.

paper focuses on detector performances and characteristics, such as energy andtiming resolutions as well as event discriminations by pulse shapes and heat/lightratios These characteristics are empirically studied under various temperatureconditions in an above-ground laboratory Other aspects of the AMoRE exper-iments are discussed elsewhere (e.g Monte Carlo background simulation [21],radioactive contamination of40Ca100MoO4crystals [22], and the overall status ofthe AMoRE project [7])

2 Detector setup

The detector module is designed for simultaneous measurement of heat (phonon)and scintillation-light (photon) signals from a CaMoO4crystal In this present ex-periment, a doubly enriched40Ca100MoO4 was used as the target absorber The

internal background of the crystal, named SB28, was previously studied with

ACCEPTED MANUSCRIPT

room-temperature scintillation measurement at the YangYang Underground oratory (Y2L) [23] The module was structured using oxygen-free high conduc-tivity (OFHC) copper for high thermal conductivity at low temperatures Thecrystal, with a mass of about 200 g, as an oval cylinder, was held by phosphor-bronze springs that were firmly attached to the copper holder A patterned goldfilm of 2-cm diameter and 400-nm total thickness was evaporated on the bottomsurface of the crystal This film collects phonons generated by particle detec-tion in the crystal The phonon signals are read by a temperature sensor (i.e., anMMC sensor) placed on a copper plate with a superconducting quantum interfer-ence device (SQUID) The thermal connection between the gold phonon collectorand the MMC sensor was made with annealed gold bonding wires The majority

Lab-of the energy absorbed in the crystals is converted into heat signals in the form

of phonons The excess phonons make net heat flow from the absorber crystalthrough the gold phonon collector film, gold wires and the MMC sensor Theheat eventually releases to a thermal bath through a weak thermal link made of acouple of gold bonding wires connecting the MMC sensor and the copper sampleholder The MMC with the SQUID read-out measures the temperature change atthe MMC sensor in the heat flow sequence The details of the phonon sensingsystem are described in our previous reports [24, 25]

A light detector composed of a 2-inch Ge wafer and an MMC sensor wasemployed to detect scintillation light from the CaMoO4crystal This light detectorwas constructed in a manner similar to how the phonon sensor was made It hasthree circular gold films, 5 mm in diameter and 300-nm thick, to collect phononsgenerated by light absorption in the wafer [26] A thermal connection betweenthe gold films and an MMC sensor was also made using annealed gold wires Anumber of light detectors can be easily made with this patchable design MMCsand light absorbers can be made separately, and assembled in the final stage Theintrinsic signal rise-time of the light detector was about 0.2 ms independent ofoperating temperature [26] The light detector was installed toward the top-side

of the CaMoO4 crystal The crystal was mounted in a light cavity surrounded byVM2000 light reflector films to increase the collection efficiency of scintillationlight

The detector module was attached to the mixing chamber of a dilution erator capable of cooling the system to 7 mK The cryostat, which was installed

refrig-in an above-ground laboratory, was surrounded by a 10-cm thick lead shield

Temperature (mK)

Figure 2: Averaged signals of full absorption of 2615 keV gammas measured in the heat channel

at 10-50 mK The inset shows the FWHM resolution of the baseline noise without signals obtained

by the amplitude-determining algorithm described in text to the baseline noise records.

3 Experimental Results

3.1 Signal Properties

In the present experiment, the signals in both phonon and photon sensors inate from several sources Because the measurement was performed in an above-ground laboratory, events caused by muons passing the crystal appear over a wideenergy range, up to about 30 MeV Radioactive decays of internal radio-impurities

orig-in the crystal also generated heat and light signals simultaneously Moreover, vironmental gamma rays contributed to the background spectrum as a significantportion of electron-induced events, with energies up to 2615 keV An external

en-232Th source was installed between the external lead shield and the cryostat Thesource was used for the energy calibration of the detector, in addition to detectorresolution and signal shape studies at various temperatures

Fig 2 shows the signal shapes of 2615-keV gamma-rays fully absorbed in the

40Ca100MoO4crystal, measured in 10-50 mK Lower temperatures result in larger

ACCEPTED MANUSCRIPT

but slower signals due to the temperature dependence of the thermodynamic erties of the sub-thermal components of the detector A detector model of MMCmeasurement having a large crystal absorber was established [6, 27] It was furtheroptimized for larger pulse height and faster signal rise [25] The inset shows thebaseline resolution, which is the energy resolution of randomly triggered signals

prop-at each temperprop-ature The measured baseline resolutions represent the irreduciblenoise level of signal amplitudes corresponding to the signal-to-noise ratio for themeasurement condition

Determination of the signal amplitude for each measured pulse is one of mostessential works in the analysis The optimal filtering method may provide thebest energy estimation only if all signals are in the same shape but different pro-portionality [28, 29] However, this method is not practically applicable to thepresent measurement The event-rate of the measurement is relatively high com-pared to the long-decay-component of the pulse shape A following signal oftenappeared before the signal is fully decayed In this high event-rate condition thatpile-up signals frequently appear, remained decay-component of a signal affectsthe signal shape of its subsequent events This makes optimal filtering inadequate

to our measurement condition The pulse height parameter, the maximum of the

pulse taking the high frequency noise into account, and the left area (LA)

pa-rameter [24], the weighted partial area of pulse, were considered as alternativeamplitude parameters However, the pulse height parameter is also inadequate torepresent the signal amplitude mainly because of the position dependence of thesignal The position dependence appears because our signal is sensitive to ather-mal phonon signals with good timing resolutions The LA parameter is a favorablealternative for the presence of the position dependence as described in Ref [24].The LA parameter is found for the sum of the initial part of the pulses Theintegration range is determined by the mean-time value of the pulses The mean-time and LA parameters are defined as,

where v t is the measured voltage signal at time t subtracting its base level The

time scale of the triggered signals is redefined to zero at the time when the signal

rises at 10% of the pulse height l and r indicate the time length of the signal

Figure 3: An energy spectrum of the heat channel measured with an external 232 Th source Most

of the peaks resulted from the gamma-rays of the source, except the 511- and 1461-keV peaks The five peak positions marked with arrows are used for the energy calibration of β/γ signals.

toward left and right directions from the redefined trigger time, respectively, used

to calculate the mean-time as described in Ref [24]

3.2 Energy calibration

The energy spectrum shown in Fig 3 was obtained with a 232Th source at

10 mK The energy scale of the spectrum is determined using the LA parameter

as the signal amplitude of the heat channel Five gamma-ray peaks, indicated byarrows, are chosen for the calibration points of the electron-induced β/γ events.The spectrum shows some other peaks that are likely coming from neighboringgamma lines with a few keV apart, and are not added in the calibration to avoidambiguity

Energy calibration of the α signals is separately made with alpha-inducedpeaks originating from internal alpha decay events inside the crystal Alpha-induced signals can be clearly selected based on their pulse-shapes and light/heatratios, as described in the following subsections

Energy (keV)

0 0.5 1 1.5 2 2.5 3

ACCEPTED MANUSCRIPT

The measured peak amplitudes for α and β/γ events are plotted with theircorresponding energy in Fig 4 Quadratic fit functions without a constant term areapplied to calibrate the energy scales of each α and β/γ event, considering smallnon-linearity Fig 4 (b) indicates the proportionality difference of the energy scalebetween the α and β/γ signals where all of the signal amplitudes are divided bytheir corresponding energy of a linear fit found using the electron signals only Thedashed lines are the quadratic calibrations without a constant term for each group

of α and β/γ signals Event-dependent scintillation yield is likely responsiblefor different proportionality of the energy scales for α and β/γ signals Electron-induced events are known to generate more scintillations in CaMoO4crystals thanalpha-induced events with the same energy [30] Scintillation lights coming out

of a crystal are, in fact, regarded as energy loss in the heat measurement channel,resulting in smaller signals for β/γ events The bottom plot shows the residuals ofthe peak locations from each of the quadratic calibrations converted to an energyscale It is shown that the quadratic calibration functions well represent the energy

of the α and β/γ signals Only a small amount of deviations from the calibrationswas found for both the α and β/γ peaks Extrapolation of the energy calibrationcan result in sub-keV accuracy at 3034 keV, i.e., the Q-value of100Mo decay forelectron-induced events

On the other hand, the corresponding quadratic calibration is not perfectlymatched for alpha events However, all the alpha peaks can be identified by com-parison between the peak amplitudes and the decay-energy (Q) values of alphadecaying nuclides In fact, the deviations of the peaks from the calibration aremuch smaller than the energy resolutions of the alpha peaks with a few tens ofkeV in FWHM With this alpha energy calibration alpha decay events of 212Bifollowed by crucial208Tl beta decay with a 4999-keV Q-value can be efficientlyidentified Because208Tl has a half-life of about 183 s, 208Ti decay events in anabsorber crystal can be tagged by212Bi alpha events in a low-background environ-ment

3.3 Energy resolution

After applying the energy calibration as discussed above, the energy tions of known peaks were obtained for the spectra measured at different temper-atures as shown in Fig 5 For 2615-keV gamma rays, an 8.7-keV FWHM wasobtained at 10 and 30 mK

resolu-In an ideal case of thermal calorimetric detection in an equilibrium detector,the energy resolution of the detector is limited by thermodynamic fluctuation noiseand thermometer noise [28] In our detector setup, the noise from the thermometer