Development of Scintillating Bolometers for the Search of ... · a smaller volume (3x1.5x0.4 mm3)...

Development of Scintillating Bolometers

for the Search of the Neutrinoless Double Beta Decay

PhD Candidate: Laura CardaniSupervisor: prof. Fernando Ferroni

Wednesday, February 15, 2012

The Importance of the 0νDBDν: Dirac or Majorana?

(Only) if ν is a Majorana particle (0ν)DBD can happen

Not allowed by SM; up to now never observed T 0ν1/2 > 1024y

2

nnppeeWWx

1

TDBD0ν1/2

= G0ν(Q,Z)|M0ν |2m2ββ


The Importance of the 0νDBDν: Dirac or Majorana?

(Only) if ν is a Majorana particle (0ν)DBD can happen

Not allowed by SM; up to now never observed T 0ν1/2 > 1024y

3

What are we looking for?

nnppeeWWx

β+β spectrum

If we plot the sum of the energies of the two e- we expect a monochromatic peak at the Q - value of the decay

Nuclear process: emission of two e- without ν


The Bolometric Detectors

4

A bolometer can be considered as an “ideal” calorimeter operated at cryogenic temperatures (~10 mK)

It can be sketched as absorber + sensor

An energy release in the absorber gives rise to a temperature increase (read by the sensor)

They can be grown with the isotope of interest (ex: TeO2 for 130Te) --> source =detector

Thermal Bath (~10 mK)

Thermal coupling

sensor(ntd ge thermistor)

absorbercrystal

Time (s)0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Volta

ge (m

V)

2600

2800

3000

3200

3400typical bolometric pulse


The Bolometric Detectors

5

Excellent resolution (5 keV @ 2615 keV)

Good efficiency (>80%)

Radio- pure --> low intrinsic background

Scalability

Which advantages?

A bolometer can be considered as an “ideal” calorimeter operated at cryogenic temperatures (~10 mK)

It can be sketched as absorber + sensor

An energy release in the absorber gives rise to a temperature increase (read by the sensor)

They can be grown with the isotope of interest (ex: TeO2 for 130Te) --> source =detector

Thermal Bath (~10 mK)

Thermal coupling

sensor(ntd ge thermistor)

absorbercrystal


The Sensitivity

6

i.a. = isotopic abundance

A = mass number

ε= detector efficiency

M = mass [kg]

T = measurement time [y]

ΔE = energy resolution [keV]

b = backround [counts/keV/kg/y]


The Sensitivity

7

i.a. = isotopic abundance

A = mass number

ε= detector efficiency

M = mass [kg]

T = measurement time [y]

ΔE = energy resolution [keV]

b = backround [counts/keV/kg/y]

ε > 80 %

M ~ 1 ton

T ~ 5 y

ΔE ~ 5 keV

b ~ 10-2 counts/keV/kg/y

For Bolometers: how can we improve?


The Challenge for Background Reduction

muon flux: (2.58 ± 0.3) x 10-‐8 μ/(s cm2)Mei and Hime, Phys. Rev. D 73, 053004, 2006 [astro-‐ph 0512125]

neutron flux [<10MeV] : ≈ 4x10-‐6 n/(s cm2)F. Arneodo et al., Il Nuovo Cim. 112A, 819, 1999

gamma flux: ≈ 0.73 γ/(s cm2) C. Bucci et al., Eur. Phys. J. A, 41, 155-‐168, 2009

8

The detectors are located in deep underground laboratories

Proper shielding and vetoes

Development of proper cleaning/storage procedures

Ultimate limit given by the contaminations of the detector itself!

LNGS(Laboratori Nazionali del Gran Sasso)

3650 m.w.e.


The Cuoricino Experiment

9

Largest bolometric experiment ever realized

Study of

40.7 kg of TeO2 (11.6 kg of 130Te)

Limit:

130Te ! 130Xe + 2e-

Study of the background : main problem surface α contaminations of the materials!

m!! < 300 ÷ 710 eV!

α contaminationsSIGNAL Black =

background measurement

Green = calibration

measurement (normalized on

208Tl peak)


The Solution: Scintillating Bolometers

10

α and β/γ can release a similar heat in the bolometer BUT they emit a different amount of light --> we detect heat + light

The α background can be eliminated!

Our light detector is a thin germanium slab operated as bolometer

Unfortunately, TeO2 does not scintillate, so we need to grow other crystals....


The Solution: Scintillating Bolometers

11

130Te 76Ge 100Mo 116Cd

Background γ ambientale

82Se

A large Q - value provides a further background reduction (less β/γ contributions at higher energies)

However, we need also a large isotopic aboundance!

α and β/γ can release a similar heat in the bolometer BUT they emit a different amount of light --> we detect heat + light

The α background can be eliminated!

Our light detector is a thin germanium slab operated as bolometer

Unfortunately, TeO2 does not scintillate, so we need to grow other crystals....


The LUCIFER Experiment

12

LUCIFER EXPERIMENT

Array of scintillating bolometers

R & D on different crystals to find the best candidate for a large mass experiment

Baseline for the LUCIFER project: ZnSe

Study of 82Se decay

Q - value = 2995 keV

i.a. ~ 9.2 % but enrichment feasibility proved!

ZnSe has a large LY (~ 7 keV/MeV)

Good energy resolution

Radio - pure


Current Activity...

I’m involved in the assembly and operation of the ZnSe bolometers (LNGS)

I’m analyzing the first results of these runs: internal contaminations, background rejection power, resolution and so on...

13

Energy (keV)2 4 6 8 10 12

310!Li

ght (

keV

)0

50

100

150

200

250

The first tests for the “official” ZnSe production begun a few weeks ago

Several tests are needed to characterize and understand the ZnSe features

particle energy [keV]

detected l

ight [keV]

α events

β/γ events signalQ-value


14

...R&D on Possible AlternativesZnMoO4

(arXiv:1202.0238)cles shows a slightly different time development. This featureseems to be explained [16] by the relatively long scintillationdecay time (of the order of hundreds of µs) observed in somescintillating crystals (e.g. molybdates). This long decay, com-bined with an high percentage of non-radiative de-excitation ofthe scintillation channel, will transfer phonons (i.e. heat) to thecrystal. This extremely tiny, but measurable, time dependentphonon release has a different absolute value for isoenergetic αand β/γ particles due to their different scintillation yield.

The possibility to have a bolometer in which the α back-ground is identified without the need of an additional light de-tector is particularly appealing since it translates in a simplifiedset-up.

After an exhaustive description of the α vs β/γ discrimina-tion power, presented in Sec. 3, in Sec. 5 we present a detailedMonte Carlo simulation of the background performances of apossible future experiment based on a ZnMoO4 crystal array.Within the Lucifer Project [17], an array of enriched Zn82Secrystals operated as scintillating bolometers, there is also theoption of having part of the detectors made of 100Mo enrichedcrystals. The present work, though based on small size crys-tal samples, shows the performance that could be reached by asmall scale DBD decay experiment based on Zn100MoO4 crys-tals.

2. Experimental details

High quality ZnMoO4 crystals were developed in the Niko-laev Institute of Inorganic Chemistry (NIIC, Novosibirsk, Rus-sia). Starting material for the crystal growth were High PurityZnO (produced by Umicore) and MoO3, synthesized by NIIC.Crystals up to 25 mm in diameter and 60 mm in length weregrown by the low-thermal-gradient Czochralski technique in aplatinum crucible with a size of 40 mm ×100 mm [18].

In this paper we present the result of two separate runs, bothcarried out in an Oxford 200 3He/4He dilution refrigerator lo-cated deep underground in the Laboratori Nazionali del GranSasso (average depth≈3650 m w.e. [19]).

The first run was dedicated to the study of the discriminationcapability of ZnMoO4 bolometers, while the second one wasdevoted to the evaluation of the internal contaminations of thesample.

In the first run two different ZnMoO4 crystals were tested inorder to study the reproducibility of the background rejectionover different samples. The samples were a 27.5 g colourlesscylinder ( 18.5 mm ×22.3 mm) and a 29.9 g slightly orangeparallelepiped (28.5×18.4×13.2 mm3).

The ZnMoO4 crystals were held by means of four S-shapedPTFE supports fixed to cylindrical Cu columns. They were sur-rounded (with no direct contact) by a plastic reflecting sheet(3M VM2002). The Light Detector (LD) [20] is constituted bya 36 mm diameter, 1 mm thick pure Ge crystal absorber. TheGe wafer is heated up by the absorbed photons and the temper-ature variation is proportional to the scintillation signal. Theset-up of the detectors is schematized in Fig. 1.

The temperature sensor of the ZnMoO4 crystals is a 3x3x1mm3 Neutron Transmutation Doped (NTD) germanium ther-

Thermistor

Ge!LD

S!PTFEReflecting Foil

Cu !Column

Heater Male pin

Female pin

Figure 1: Set-up of the detectors. The ball-bonded Au wires are crimped into“male” Cu tubes (pins) and inserted into ground-insulated “female” Cu tubes.Custom wires from detectors towards cryostat are not drawn. A section of thelight detector is not drawn for a better understanding.

mistor, the same used in the Cuoricino experiment. It is ther-mally coupled to the crystal via 9 glue spots of ≈ 0.6 mm diam-eter and ≈50 µm height. The temperature sensor of the LD hasa smaller volume (3x1.5x0.4 mm3) in order to decrease its heatcapacity, increasing therefore its thermal signal. A resistor of≈300 kΩ, realized with a heavily doped meander on a 3.5 mm3

silicon chip, is attached to each absorber and acts as a heater tostabilize the gain of the bolometer [21, 22]. The details of theelectronics and the cryogenic facility can be found elsewhere[23, 24, 25].

The heat and light pulses, produced by a particle interactingin the absorber and transduced in a voltage pulse by the NTDthermistors, are amplified and fed into a 18 bit NI-6284 PXIADC unit. The trigger is software generated on each bolometerand when it fires waveforms 0.6 s long, sampled at 2 kHz, aresaved on disk. Moreover, when the trigger of a ZnMoO4 crys-tal fires, the corresponding waveform from the LD is recorded,irrespective of its trigger.

As the main goal of the measurements was to test the discrim-ination capability of the detectors between α and β/γ events, a238U/234U α source was faced to the crystals, on the oppositeside with respect to the LD. The source was covered with a 6µm thick Mylar film, in order to “smear” the α’s energies downto the Qββ energy region. The γ calibration of the ZnMoO4crystals is performed through a movable a 232Th source insertedbetween the Dewar housing the cryostat and the external leadshield. The energy calibration of the LD is achieved thanks toa 55Fe X-ray source, producing two X-rays at 5.9 and 6.5 keV,faced to the LD.

In the second run only the best performing bolometer (thecylindrical crystal) was characterized. The α source was re-moved and the crystal was surrounded by several layers of ul-trapure polyethylene in order to shield the bolometer againstthe surface contamination of the copper structure. In this mea-

2

Promising candidate for the study of 100Mo

Q - value ~ 3034 keV

i.a. ~ 9.7 %

surement the lack of space prevented us from mounting the LD.However this was not a problem, as the previous run convinc-ingly demonstrated that the pulse shape analysis can provide anextremely good α background rejection without the need for thelight detection.

2.1. Data Analysis

To maximize the signal to noise ratio, the pulse amplitudeis estimated by means of an optimum filter technique [26, 27].The filter transfer function is built from the ideal signal shapes(t) and the noise power spectrum N(ω). s(t) is estimated byaveraging a large number of triggered pulses (so that stochasticnoise superimposed to each pulse averages to zero) while N(ω)is computed averaging the power spectra of randomly acquiredwaveforms not containing pulses. The amplitude of a signal isestimated as the maximum of the filtered pulse. This procedureis applied for the signal on the ZnMoO4 bolometer. The am-plitude of the light signal, instead, is estimated from the valueof the filtered waveform at a fixed time delay with respect tothe signal of the ZnMoO4 bolometer, as described in detail inRef. [28]. The detector performances are reported in Table 1.The baseline resolution, FWHMbase, is governed by the noisefluctuation at the filter output, and does not depend on the abso-lute pulse amplitude. The rise (τR) and decay times (τD) of thepulses are computed as the time difference between the 10%and the 90% of the leading edge, and the time difference be-tween the 90% and 30% of the trailing edge, respectively.

Table 1: Technical details for the ZnMoO4 bolometers (cylinder and paral-lelepiped) and for the LD. The cylindrical ZnMoO4 was measured twice, so wereported the parameters also for the background run (Cylinder∗). Rwork is theworking resistance of the thermistors. Signal represents the absolute voltagedrop across the termistor for a unitary energy deposition.

Crystal Rwork Signal FWHMbase τR τD[MΩ] [µV/MeV] [keV] [ms] [ms]

Cylinder 3.7 140 0.6 17 50Parallel. 4.7 320 1.2 8 33LD 8.8 1700 0.16 4 11Cylinder∗ 2.5 200 0.7 17 48

After the application of the optimum filter, signal amplitudesare corrected for temperature and gain instabilities of the set-upthanks to a monochromatic power injection in the Si heater tak-ing place every few minutes. The ZnMoO4 is calibrated usingthe most intense γ-peaks from the 232Th source, while the LDis calibrated using the 55Fe X-ray doublet.

The FWHM energy resolution obtained on the cylindrical(parallelepiped) crystal ranges from 2.5 ± 0.1 (2.4 ± 0.1) keVat 238 keV to 3.8 ± 0.9 keV (7.6 ± 1.3) at 2615 keV. Theenergy resolution on the 5407 keV α + recoil line (due to aweak internal contamination of 210Po ) can be evaluated onlyon the long background run for the cylindrical crystal and gives5.3 ± 1.1 keV. The FWHM energy resolution on the LD, eval-uated on the 55Fe X-ray doublet, is 321 ± 9 eV. Experimentalresolutions are worse than theoretical ones in agreement withthe observed performance of macro-bolometers [29].

Energy [keV]1000 2000 3000 4000 5000

LY

[keV

/MeV

]

-0.5

0

0.5

1

1.5

2

210Po

2615 keV line!"

smeared !source#

Figure 2: The light-to-heat energy ratio as a function of the heat energy obtainedwith the cylindrical crystal in the first run, during a 62 h 232Th calibration. Theupper band (ascribed to β/γ events) and lower band (populated by α decays)are clearly separated. The 2615 keV 208Tl γ-line is well visible in the β/γband as well as a the continuum rate induced by the degraded α source and the5407 keV 210Po doublet in the α band. The discrimination power is reported inSec. 3.

The light-to-heat energy ratio as a function of the heat en-ergy is shown for the calibration spectrum in Fig. 2. β/γ andα decays give rise to very clear separate distributions. In theupper band, ascribed to β/γ events, the 2615 keV 208Tl γ-lineis well visible. The lower band, populated by α decays, showsthe continuum rate induced by the degraded α source as well asthe 210Po doublet.

The Light Yield (LY), defined as the ratio between the mea-sured light (in keV) and the nominal energy of the event (inMeV), was measured for the most intense γ-lines giving 1.10± 0.03 keV/MeV and 0.78 ± 0.02 keV/MeV for the cylinderand for the parallelepiped, respectively. These values are con-stant from 0.2 to 2.6 MeV and are not corrected for the lightcollection efficiency. The LY of the cylindrical crystal is wellin agreement with the one reported in [14, 30], while the paral-lelepiped shows a lower LY.

The Quenching Factor (QF), defined as the ratio of theLYα/LYβ/γ for the same energy release, was evaluated on the5407 keV α-line and results 0.18 for both crystals.

3. α vs β/γ discrimination

As reported in [14, 15], Molybdate crystals can provideα vs β/γ discrimination by making use of the thermal infor-mation only. In Fig. 3 the ideal signal shape s(t) for the twoevent classes is shown together with the percentage differences(t)α-s(t)β/γ. Pulse shapes are obtained by averaging pulses (ob-tained in the same calibration measurement of Fig. 2) in the en-ergy range 2610-2620 keV and aligned at the maximum. Dif-ferences at a level of a few per mille are visible both in the riseand decay of the thermal pulse 1.

1We will refer to pulses from the cylindrical bolometer throughout the restof the text. However, the parallelepiped bolometer showed consistent results.

3


lig

ht y

ield [keV/M

eV]

signalQ-value


15

...R&D on Possible AlternativesZnMoO4

(arXiv:1202.0238)cles shows a slightly different time development. This featureseems to be explained [16] by the relatively long scintillationdecay time (of the order of hundreds of µs) observed in somescintillating crystals (e.g. molybdates). This long decay, com-bined with an high percentage of non-radiative de-excitation ofthe scintillation channel, will transfer phonons (i.e. heat) to thecrystal. This extremely tiny, but measurable, time dependentphonon release has a different absolute value for isoenergetic αand β/γ particles due to their different scintillation yield.

The possibility to have a bolometer in which the α back-ground is identified without the need of an additional light de-tector is particularly appealing since it translates in a simplifiedset-up.

After an exhaustive description of the α vs β/γ discrimina-tion power, presented in Sec. 3, in Sec. 5 we present a detailedMonte Carlo simulation of the background performances of apossible future experiment based on a ZnMoO4 crystal array.Within the Lucifer Project [17], an array of enriched Zn82Secrystals operated as scintillating bolometers, there is also theoption of having part of the detectors made of 100Mo enrichedcrystals. The present work, though based on small size crys-tal samples, shows the performance that could be reached by asmall scale DBD decay experiment based on Zn100MoO4 crys-tals.

2. Experimental details

High quality ZnMoO4 crystals were developed in the Niko-laev Institute of Inorganic Chemistry (NIIC, Novosibirsk, Rus-sia). Starting material for the crystal growth were High PurityZnO (produced by Umicore) and MoO3, synthesized by NIIC.Crystals up to 25 mm in diameter and 60 mm in length weregrown by the low-thermal-gradient Czochralski technique in aplatinum crucible with a size of 40 mm ×100 mm [18].

In this paper we present the result of two separate runs, bothcarried out in an Oxford 200 3He/4He dilution refrigerator lo-cated deep underground in the Laboratori Nazionali del GranSasso (average depth≈3650 m w.e. [19]).

The first run was dedicated to the study of the discriminationcapability of ZnMoO4 bolometers, while the second one wasdevoted to the evaluation of the internal contaminations of thesample.

In the first run two different ZnMoO4 crystals were tested inorder to study the reproducibility of the background rejectionover different samples. The samples were a 27.5 g colourlesscylinder ( 18.5 mm ×22.3 mm) and a 29.9 g slightly orangeparallelepiped (28.5×18.4×13.2 mm3).

The ZnMoO4 crystals were held by means of four S-shapedPTFE supports fixed to cylindrical Cu columns. They were sur-rounded (with no direct contact) by a plastic reflecting sheet(3M VM2002). The Light Detector (LD) [20] is constituted bya 36 mm diameter, 1 mm thick pure Ge crystal absorber. TheGe wafer is heated up by the absorbed photons and the temper-ature variation is proportional to the scintillation signal. Theset-up of the detectors is schematized in Fig. 1.

The temperature sensor of the ZnMoO4 crystals is a 3x3x1mm3 Neutron Transmutation Doped (NTD) germanium ther-

Thermistor

Ge!LD

S!PTFEReflecting Foil

Cu !Column

Heater Male pin

Female pin

Figure 1: Set-up of the detectors. The ball-bonded Au wires are crimped into“male” Cu tubes (pins) and inserted into ground-insulated “female” Cu tubes.Custom wires from detectors towards cryostat are not drawn. A section of thelight detector is not drawn for a better understanding.

mistor, the same used in the Cuoricino experiment. It is ther-mally coupled to the crystal via 9 glue spots of ≈ 0.6 mm diam-eter and ≈50 µm height. The temperature sensor of the LD hasa smaller volume (3x1.5x0.4 mm3) in order to decrease its heatcapacity, increasing therefore its thermal signal. A resistor of≈300 kΩ, realized with a heavily doped meander on a 3.5 mm3

silicon chip, is attached to each absorber and acts as a heater tostabilize the gain of the bolometer [21, 22]. The details of theelectronics and the cryogenic facility can be found elsewhere[23, 24, 25].

The heat and light pulses, produced by a particle interactingin the absorber and transduced in a voltage pulse by the NTDthermistors, are amplified and fed into a 18 bit NI-6284 PXIADC unit. The trigger is software generated on each bolometerand when it fires waveforms 0.6 s long, sampled at 2 kHz, aresaved on disk. Moreover, when the trigger of a ZnMoO4 crys-tal fires, the corresponding waveform from the LD is recorded,irrespective of its trigger.

As the main goal of the measurements was to test the discrim-ination capability of the detectors between α and β/γ events, a238U/234U α source was faced to the crystals, on the oppositeside with respect to the LD. The source was covered with a 6µm thick Mylar film, in order to “smear” the α’s energies downto the Qββ energy region. The γ calibration of the ZnMoO4crystals is performed through a movable a 232Th source insertedbetween the Dewar housing the cryostat and the external leadshield. The energy calibration of the LD is achieved thanks toa 55Fe X-ray source, producing two X-rays at 5.9 and 6.5 keV,faced to the LD.

In the second run only the best performing bolometer (thecylindrical crystal) was characterized. The α source was re-moved and the crystal was surrounded by several layers of ul-trapure polyethylene in order to shield the bolometer againstthe surface contamination of the copper structure. In this mea-

2

0 100 200 300 400 500 600

Am

plitu

de [A

.U.]

0

0.2

0.4

0.6

0.8

1

Time [ms]0 100 200 300 400 500 600

[%]

!/"

-s(t)

#s(

t)-1.2

-1-0.8-0.6-0.4-0.2

0

Figure 3: Comparison of the β/γ and α pulse shape. Top figure: thermal pulse

obtained by averaging events in the 2615 keV γ peak (blue line) and events

belonging to the α band with an equivalent energy (red line): their difference is

not appreciable. Bottom figure: percentage difference s(t)α-s(t)β/γ. Differences

at a level of a few per mille are visible both in the rise and decay of the thermal

pulse.

To fully exploit this feature several shape variables have been

studied: τR, τD, TVL, TVR, χ2

OT. The Test Value Left (TVL)

and Right (TVR) are defined as:

TVL =1

A · wl

iM

i=iM−wl

(yi − A si)2; (1)

TVR =1

A · wr

iM+wr

i=iM

(yi − A si)2. (2)

All quantities in the above equations refer to the optimum

filtered pulses: yi is the pulse, A and iM its amplitude and max-

imum position, si the ideal signal pulse scaled to unitary ampli-

tude and aligned to yi, wl(r) the left(right) width at half maxi-

mum of si. The maximum of the ideal pulse is aligned with the

maximum of the filtered pulse, fractional delays are handled

with a linear interpolation of the data samples.

Finally, χ2

OTwas computed as the sum of the squared fit resid-

uals normalized to the noise RMS, using si as a fit function of

the central part of the pulse. This algorithm is described in de-

tail in [31].

Bolometers are nonlinear detectors [32], i.e. the shape of the

signal slightly depends on the amount of released energy. This

implies that also the shape parameters (evaluated with respect to

an energy independent pulse, si) will show such a dependence.

To remove such a dependency variables were linearized in the

2300-3200 keV energy range.

Energy [keV]0 1000 2000 3000 4000 5000

TVR

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

!µ 0.00055± -0.00052

!" 0.00057± 0.00849 #$

µ 0.0013± -0.2759 #$" 0.0014± 0.0108

TVR-0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0

counts/0.01

0

10

20

30

40

50

60

70

80

90

100

Figure 4: Top Figure: TVR as a function of the energy. The upper band is

populated by α particles (events in the 2300-3200 keV energy range are shown

in red) while β/γ’s contribute to the lower band (events in the 2300-3200 keV

energy range are shown in black). Bottom Figure: TVR histogram for the αsample in red and for the β/γ in black. The mean values and the standard

deviations, as estimated from a Gaussian fit, are reported. Using Eq. 3 a dis-

crimination power of ≈ 20 is obtained.

This procedure, moreover, allowed to enlarge the data sample

on which the pulse shape analysis was performed.

Fig. 4 shows the TVR variable as a function of the energy. As

for the case of Fig.2, β/γ and α events are distinctly separated.

Events in the linearized energy region and belonging to the up-

per (α) band are shown in red while events, in the same energy

interval, belonging to the lower (β/γ) band are shown in black.

The discrimination power (DP) between the two distributions

is usually quantified as the difference between the average val-

ues of the two distributions normalized to the square root of the

quadratic sum of their widths:

DP =µβ/γ − µασ2

β/γ + σ2α

. (3)

The above definition does not imply any probabilistic meaning,

it was used for the sake of comparison with similar quantities

reported in published papers [14, 15].

4

0 100 200 300 400 500 600

Am

plitu

de [A

.U.]

0

0.2

0.4

0.6

0.8

1

Time [ms]0 100 200 300 400 500 600

[%]

!/"

-s(t)

#s(

t)

-1.2-1

-0.8-0.6-0.4-0.2

0

Figure 3: Comparison of the β/γ and α pulse shape. Top figure: thermal pulse

obtained by averaging events in the 2615 keV γ peak (blue line) and events

belonging to the α band with an equivalent energy (red line): their difference is

not appreciable. Bottom figure: percentage difference s(t)α-s(t)β/γ. Differences

at a level of a few per mille are visible both in the rise and decay of the thermal

pulse.

To fully exploit this feature several shape variables have been

studied: τR, τD, TVL, TVR, χ2

OT. The Test Value Left (TVL)

and Right (TVR) are defined as:

TVL =1

A · wl

iM

i=iM−wl

(yi − A si)2; (1)

TVR =1

A · wr

iM+wr

i=iM

(yi − A si)2. (2)

All quantities in the above equations refer to the optimum

filtered pulses: yi is the pulse, A and iM its amplitude and max-

imum position, si the ideal signal pulse scaled to unitary ampli-

tude and aligned to yi, wl(r) the left(right) width at half maxi-

mum of si. The maximum of the ideal pulse is aligned with the

maximum of the filtered pulse, fractional delays are handled

with a linear interpolation of the data samples.

Finally, χ2

OTwas computed as the sum of the squared fit resid-

uals normalized to the noise RMS, using si as a fit function of

the central part of the pulse. This algorithm is described in de-

tail in [31].

Bolometers are nonlinear detectors [32], i.e. the shape of the

signal slightly depends on the amount of released energy. This

implies that also the shape parameters (evaluated with respect to

an energy independent pulse, si) will show such a dependence.

To remove such a dependency variables were linearized in the

2300-3200 keV energy range.

Energy [keV]0 1000 2000 3000 4000 5000

TVR

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

!µ 0.00055± -0.00052

!" 0.00057± 0.00849 #$

µ 0.0013± -0.2759 #$" 0.0014± 0.0108

TVR-0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0

counts/0.01

0

10

20

30

40

50

60

70

80

90

100

Figure 4: Top Figure: TVR as a function of the energy. The upper band is

populated by α particles (events in the 2300-3200 keV energy range are shown

in red) while β/γ’s contribute to the lower band (events in the 2300-3200 keV

energy range are shown in black). Bottom Figure: TVR histogram for the αsample in red and for the β/γ in black. The mean values and the standard

deviations, as estimated from a Gaussian fit, are reported. Using Eq. 3 a dis-

crimination power of ≈ 20 is obtained.

This procedure, moreover, allowed to enlarge the data sample

on which the pulse shape analysis was performed.

Fig. 4 shows the TVR variable as a function of the energy. As

for the case of Fig.2, β/γ and α events are distinctly separated.

Events in the linearized energy region and belonging to the up-

per (α) band are shown in red while events, in the same energy

interval, belonging to the lower (β/γ) band are shown in black.

The discrimination power (DP) between the two distributions

is usually quantified as the difference between the average val-

ues of the two distributions normalized to the square root of the

quadratic sum of their widths:

DP =µβ/γ − µασ2

β/γ + σ2α

. (3)

The above definition does not imply any probabilistic meaning,

it was used for the sake of comparison with similar quantities

reported in published papers [14, 15].

4

Shape Parameter

α pulse

γ pulse

α pulse - γ pulse


lig

ht y

ield [keV/M

eV]surement the lack of space prevented us from mounting the LD.

However this was not a problem, as the previous run convinc-ingly demonstrated that the pulse shape analysis can provide anextremely good α background rejection without the need for thelight detection.

2.1. Data Analysis

To maximize the signal to noise ratio, the pulse amplitudeis estimated by means of an optimum filter technique [26, 27].The filter transfer function is built from the ideal signal shapes(t) and the noise power spectrum N(ω). s(t) is estimated byaveraging a large number of triggered pulses (so that stochasticnoise superimposed to each pulse averages to zero) while N(ω)is computed averaging the power spectra of randomly acquiredwaveforms not containing pulses. The amplitude of a signal isestimated as the maximum of the filtered pulse. This procedureis applied for the signal on the ZnMoO4 bolometer. The am-plitude of the light signal, instead, is estimated from the valueof the filtered waveform at a fixed time delay with respect tothe signal of the ZnMoO4 bolometer, as described in detail inRef. [28]. The detector performances are reported in Table 1.The baseline resolution, FWHMbase, is governed by the noisefluctuation at the filter output, and does not depend on the abso-lute pulse amplitude. The rise (τR) and decay times (τD) of thepulses are computed as the time difference between the 10%and the 90% of the leading edge, and the time difference be-tween the 90% and 30% of the trailing edge, respectively.

Table 1: Technical details for the ZnMoO4 bolometers (cylinder and paral-lelepiped) and for the LD. The cylindrical ZnMoO4 was measured twice, so wereported the parameters also for the background run (Cylinder∗). Rwork is theworking resistance of the thermistors. Signal represents the absolute voltagedrop across the termistor for a unitary energy deposition.

Crystal Rwork Signal FWHMbase τR τD[MΩ] [µV/MeV] [keV] [ms] [ms]

Cylinder 3.7 140 0.6 17 50Parallel. 4.7 320 1.2 8 33LD 8.8 1700 0.16 4 11Cylinder∗ 2.5 200 0.7 17 48

After the application of the optimum filter, signal amplitudesare corrected for temperature and gain instabilities of the set-upthanks to a monochromatic power injection in the Si heater tak-ing place every few minutes. The ZnMoO4 is calibrated usingthe most intense γ-peaks from the 232Th source, while the LDis calibrated using the 55Fe X-ray doublet.

The FWHM energy resolution obtained on the cylindrical(parallelepiped) crystal ranges from 2.5 ± 0.1 (2.4 ± 0.1) keVat 238 keV to 3.8 ± 0.9 keV (7.6 ± 1.3) at 2615 keV. Theenergy resolution on the 5407 keV α + recoil line (due to aweak internal contamination of 210Po ) can be evaluated onlyon the long background run for the cylindrical crystal and gives5.3 ± 1.1 keV. The FWHM energy resolution on the LD, eval-uated on the 55Fe X-ray doublet, is 321 ± 9 eV. Experimentalresolutions are worse than theoretical ones in agreement withthe observed performance of macro-bolometers [29].

Energy [keV]1000 2000 3000 4000 5000

LY

[keV

/MeV

]

-0.5

0

0.5

1

1.5

2

210Po

2615 keV line!"

smeared !source#

Figure 2: The light-to-heat energy ratio as a function of the heat energy obtainedwith the cylindrical crystal in the first run, during a 62 h 232Th calibration. Theupper band (ascribed to β/γ events) and lower band (populated by α decays)are clearly separated. The 2615 keV 208Tl γ-line is well visible in the β/γband as well as a the continuum rate induced by the degraded α source and the5407 keV 210Po doublet in the α band. The discrimination power is reported inSec. 3.

The light-to-heat energy ratio as a function of the heat en-ergy is shown for the calibration spectrum in Fig. 2. β/γ andα decays give rise to very clear separate distributions. In theupper band, ascribed to β/γ events, the 2615 keV 208Tl γ-lineis well visible. The lower band, populated by α decays, showsthe continuum rate induced by the degraded α source as well asthe 210Po doublet.

The Light Yield (LY), defined as the ratio between the mea-sured light (in keV) and the nominal energy of the event (inMeV), was measured for the most intense γ-lines giving 1.10± 0.03 keV/MeV and 0.78 ± 0.02 keV/MeV for the cylinderand for the parallelepiped, respectively. These values are con-stant from 0.2 to 2.6 MeV and are not corrected for the lightcollection efficiency. The LY of the cylindrical crystal is wellin agreement with the one reported in [14, 30], while the paral-lelepiped shows a lower LY.

The Quenching Factor (QF), defined as the ratio of theLYα/LYβ/γ for the same energy release, was evaluated on the5407 keV α-line and results 0.18 for both crystals.

3. α vs β/γ discrimination

As reported in [14, 15], Molybdate crystals can provideα vs β/γ discrimination by making use of the thermal infor-mation only. In Fig. 3 the ideal signal shape s(t) for the twoevent classes is shown together with the percentage differences(t)α-s(t)β/γ. Pulse shapes are obtained by averaging pulses (ob-tained in the same calibration measurement of Fig. 2) in the en-ergy range 2610-2620 keV and aligned at the maximum. Dif-ferences at a level of a few per mille are visible both in the riseand decay of the thermal pulse 1.

1We will refer to pulses from the cylindrical bolometer throughout the restof the text. However, the parallelepiped bolometer showed consistent results.

3


lig

ht y

ield [keV/M

eV]

signalQ-value

This crystal has a low LY (is this a problem for larger crystals?)

Good energy resolution

High radio-purity

Different pulse shape for α and β/γ


...R&D on Possible Alternatives

16

TeO2

(Astropart. Phys. 35 (2012) 558)

!"#$ %&'($)*+&,*&-",*" .&',#/

Test on a small TeO2 crystal: first detection of Cerenkov light

The detected light is not enough BUT

TeO2 is a well known bolometer and its performances are already optimized

1000 crystals already @ LNGS, ready for the CUORE experiment


detected l

ight [keV]

signalQ-value


...R&D on Possible Alternatives

17

Test on a small TeO2 crystal: first detection of Cerenkov light

So...more tests on reflecting foil and light detectors must be performed

The detected light is not enough BUT

TeO2 is a well known bolometer and its performances are already optimized

1000 crystals already @ LNGS, ready for the CUORE experiment

TeO2

(Astropart. Phys. 35 (2012) 558)

!"#$ %&'($)*+&,*&-",*" .&',#/particle energy [keV]

detected l

ight [keV]


Conclusions and Perspectives

18

The background reduction is one of the main issues for (0ν)DBD physics

The LUCIFER experiment will study the (0ν)DBD of 82Se by means of ZnSe bolometers

Thanks to the scintillating bolometers technique, we will achieve a very low background counting rate ( ~ 10-3 counts/keV/kg/y) in the energy region of interest --> a competitive sensitivity on mββ (~100 meV)

I will characterize the ZnSe detectors by means of bolometric and optical measurements, performed in the cryostats of Roma and Gran Sasso laboratories.

At the same time, I will follow a parallel R&D activity on other possible candidates for the study of the (0ν)DBD


Development of Scintillating Bolometers for the Search of ... · a smaller volume (3x1.5x0.4 mm3)...

Documents

Transcript of Development of Scintillating Bolometers for the Search of ... · a smaller volume (3x1.5x0.4 mm3)...