27/jan/2005 k. ozone 2004 年度博士論文審査会 contents 1. meg experiment 2. lxe purification...

27/Jan/2005 K. Ozone2004年度博士論文審査会

Contents

1. MEG experiment

2. LXe Purification

3. Prototype Performance

4. Expected Performance

5. Toward the MEG experiment

6. Summary

Liquid Xenon Scintillation DetectorLiquid Xenon Scintillation Detectorfor the New for the New ｅｅ Search ExperimentSearch Experiment

Kenji Ozone（ ICEPP, Univ. of Tokyo,

Japan）


Physics Motivation and event Physics Motivation and event signaturesignature

SU(5) SUSY

LFV process Forbidden in the SM Sensitive to the new physics

such as SUSY-GUT, SUSY-seesaw etc.

Our goal : Br(->e)>10-13~10-14

Present limit < 1.2x10-11

(MEGA)

Clear 2-body kinematics Ee = E = 52.8 MeV Back to back event Time coincidence


MEG DetectorMEG Detector

LXe scintillation detector

Drift chamber

Timing Counter

Compensation Coil

COBRA Magnet

Surface beam

262cm

252cm

Initial goal at 10-13, finally to 10-14


Signal and backgrounds Signal and backgrounds

+e+signal– Back to Back– Ee=E=52.8MeV

Two major Backgrounds– Accidental overlap

e+from Michel decayfrom radiative decay,

e+-annihilation in flight

　　　 Down to ～ 10-14

– Radiative + decay　　 Down to <10-14

e

e

e

?


Polarized muon beam Polarized muon beam • Muons from “Surface” of + production target + decay at rest near the surface

Primary Proton +

stop

p29.8MeV/c100% polarized

+/e+

separator

MEGdetector

PrimaryProton

Quadruplemagnets

Beam TransportSolenoid

Bendingmagnets

Degrader

E5 area@PSI

+ Beamline

target


COBRA spectrometerCOBRA spectrometer

COBRA magnet was already installed into the E5 area in PSI.

Constant bending radius independent of the emission angle e+ momentum easily used at trigger level

Michel positrons are quickly swept out reduce the hit rate for stable operation


Gamma-ray detector in MEG Gamma-ray detector in MEG detectordetector

The gamma-ray detector is a key detector in MEG experiment.

Positron spectrometer can’t distinguish Michel positrons from ones of MEG event.

Ph

oto

n y

ield

per

decay

Gamma-ray energy / 52.8MeV

1.0

10-4

0.9

10-6On the other hand,52.8-MeV gamma ray in gamma-ray detector is only MEG event.


Why Xe ?Why Xe ?

High density and High light yield 1st conversion depth: 2 cm ~ 10 cm Wph: 21.7 eV (75% of NaI(Tl))

Good Energy, Position, Time resolutions

Fast Decay reduces pile-ups. τ(recombi.) = 45 nsec

Low temperature: 165 K requires refrigerator and special PMT Wavelength: ~178 nm requires VUV-sensitive PMT Light absorption requires contaminant level of 100 ppb (water)

T.Doke and K.Masuda, Nucl. Instr. And Meth.A420 (1999) 62.


LXe scintillation detector for the MEG experimentLXe scintillation detector for the MEG experiment

g

Detector concepts Observe as many photoelectrons as possible with a great accuracy.

Many PMTs are directly immersed in LXe. Kamiokande-like detector.Thin material on the incident face. Al honeycomb, compact PMT… 52.8MeV 800 liter liquid xenon 120o, |cos|<0.35, depth~50cm 830 PMTs

67cm

Energy Total number of photons Timing Arrival time of photons Position Next page……


Position reconstructionPosition reconstruction

52.8MeV

52.8MeV

LXe

LXe

PositionReconstructed by outputs of 4PMTs ~ 10 PMTs

Conversion position (X,Y) weighted mean in a region Conversion depth spread of distribution


Prototype LXe detectorsPrototype LXe detectors

The strategy to develop LXe detector 10-liter prototype ( 2.34L eff. vol. )

verified the principle LXedetector

100-liter prototype ( 68.6L eff. vol. )Theme of my doctoral thesis


100-liter prototype100-liter prototype

■ 68.6-liter active volume■ 228 PMTs

● a source (241Am) x4 PMT calibration (QE measurement) Stability monitor● LED x8 PMT calibration (gain adjustment)● cosmic ray muon counter (3pairs) Light yield monitor

372mm X 372mm X 496mm


100-liter prototype100-liter prototypePrototype LXe Scintillation Detector

1k liter LN2 tank

Compresser for refrigerator

Xe lineStudent A

window

ＫＥＫ east counter hall

π ２ area

2002 Spring

Student B


Purpose of 100-liter prototypePurpose of 100-liter prototype

Construction of a larger prototype of LXe scintillation detector Never constructed such a large detector.

Test for detector components PMTs, feed-through connectors, Cryostat, PMT holder, DAQ, Slow-control system, Purification system,…

long-term stable operation Pulse tube Refrigerator, monitoring of temperature and pressure

Performance Test for higher energy gamma raysResolutions of energy, time, and position for 40 MeV g


PMT (HAMAMATSU R6041Q)PMT (HAMAMATSU R6041Q)

FeaturesFeatures 2.5-mmt 2.5-mmt quartzquartz window window Q.E.: Q.E.: 6%6% in LXe (TYP) in LXe (TYP) (includes collection efficiency)(includes collection efficiency)

Collection eff.: 79% (TYP)Collection eff.: 79% (TYP) 3-atm 3-atm withstanding pressurewithstanding pressure Gain: Gain: 101066 (900V supplied TYP) (900V supplied TYP) Metal Channel DynodeMetal Channel Dynode thin and thin and compactcompact TTS: 750 psec (TYP)TTS: 750 psec (TYP) Works stablyWorks stably within a fluctuation of 0.5 % at within a fluctuation of 0.5 % at 165K 165K

57 mm

32 mm


Gain adjustment with LEDsGain adjustment with LEDs

Assume the LED output obeys Poisson distribution.

The PMT outputs according to the LED intensity.

LED

200

2 )( MMc

qg

200fC/ch:

charge,electron elementary:

gain,:

cq

g

spectrum, LED ofMean :

spectrum, LED ofMean :

spectrum, pedestal ofdeviation :

spectrum, LED ofdeviation :

0

0

M

M


Q.E. estimation with Q.E. estimation with aa particle data particle data

1. Make MC samples for alpha particle events. (QE = 5.0% fixed)

2. Compare the measured data with MC.

Example

Data=MC*1.42

QE = 3.5%


Light yield monitor by cosmic-ray Light yield monitor by cosmic-ray muonsmuons

Data from cosmic-ray muons is useful for monitoring higher light yield.

The cosmic-ray events are triggered with three pairs of counters (7cm x 7cm) above and below the vessel.


LXe purificationLXe purificationandand

Absorption length measurementAbsorption length measurement

At the early stage,

LXe was contaminated…


Water contaminationWater contamination

In principle, Xe scintillation photons are not re-absorbed by xenon atoms.

Possible contaminants:

water

oxygen


Residual gas analysisResidual gas analysis

H2O

O2 CO2

CO, N2

Xe

H2

Residual gas after recovering xenon was analyzed with Q-mass spectrometer at room temperature.


Purification systemPurification system

Purification : ~1200 hoursflow rate = 5 liter/min (GXe)


Growth of scintillation photon yieldGrowth of scintillation photon yield

The light yield was increased for 600 hours.

Far PMTs

Near PMTs


Growth of photon yieldGrowth of photon yield

Npe~25000 Npe~85000

a particle data

Cosmic-ray m data

Npe~300 Npe~1400

BeforeAfter


Absorption length estimationAbsorption length estimation

labs is estimated to be

12.0±1.8 cm

over 1m (90% C.L.)

LXe data were compared with LXe MC samples(lsca~∞) .

LXe(

data

)/M

C

Before

After

After

Before


Other efforts for pure LXeOther efforts for pure LXe

Replacement

PMT cover: acrylic to Teflon

Filler in incident face:

Silicon rubber to stycast with glass

Filler at the side of PMT holder:

acrylic plate to SUS hollow box

Working environment

open-air to clean-room

Circulating pump (not yet)

gaseous pump to fluid pump


Energy resolution expected for 52.8 MeV Energy resolution expected for 52.8 MeV gg

MC(prototype)

lsca=45cm

If labs is over 1m, su/E ~ 1.2%.

Energy distribution was obtained with PCA(principal component analysis) to be fitted with Gaussian with exponential tail.

labs=1m

lsca45cm

FWHM

u

Response function


Performance test with 40MeV LCS Performance test with 40MeV LCS beambeam

Purpose For 10~40 MeV gamma rays

- Energy resolution

- Position resolution Evaluate performance for 52.8MeV gamma rays.


TERAS Beam Test @ AISTTERAS Beam Test @ AIST

Incident g beam Not monochromatic Compton edge: 10, 20, 40MeV collimator: 1 mm φ


Experimental setup @ TERASExperimental setup @ TERAS


Experimental setup @ exp. roomExperimental setup @ exp. room

LCS beam


Incident beamIncident beam

LCS beam


Incident beamIncident beam

10MeV20MeV 40MeV

Energy spread becomes narrow by the collimator and the sharpness of the Compton edge is kept.

Energy resolution


Energy SpectrumEnergy Spectrum

Dep

th p

aram

eter

The spectrum has lower tail.

The resolution is evaluated by upper sigma.


FittingFitting

Expected incident Compton spectrum

Response function

The spectra were fitted with a convolution of the response function and incident LCS spectrum.


Energy Resolution at P0 & P9 Energy Resolution at P0 & P9 (update)(update)

Depth parameter > 2.0

1.4% in for 52.8MeV


Red circle: (Sum of front outputs) / 2

Depth parameter; Nf := # of PMTs in

Reconstructed (x,y); is weighted mean in

Vertex ReconstructionVertex Reconstruction

depth=4.4cmdepth=15.3cm

Depth ~ spread of output distribution.


Difference between true and reconstructed Difference between true and reconstructed vertexvertex

~6mm

MCN

f

1st Conversion Depth

depth

(x, y)


Event Selection & efficiencyEvent Selection & efficiency

Nf(0.5)>2

67%P0


FittingFitting

Fitted with 2-D Double Gaussian

F(x,y) = a(narrow Gaussian) + (1-a)(broad Gaussian),

a: ratio


Position Resolution at P0 & Position Resolution at P0 & P9(update)P9(update)

~90%

~10%3.6~5.0mm (90%) in

8.0~11.1mm (10%) in

for 40 MeV

and for 52.8MeV


Toward the MEG experimentToward the MEG experiment


Prototype to final designPrototype to final design

What’s the difference?

Shape

box -> C-shape

checked by “Mask analysis”

Data

Q-integrated ADC, discriminator & TDC

-> waveform digitizer

Calibration


Mask analysisMask analysis

– Switch off half of the PMTs in the front faceUse 4x4 PMTs out of 6x6 PMTs

– Switch off PMTs on the side walls

– Compton Edge shifts by 6.2%– Resolutions are almost same (1.84 to

1.85% in ) before and after switching off.

– 1 plane off 2.05% in – 2 planes off 2.22% in – 3, 4 planes off > 3% in


Waveform analysis & high QE PMTWaveform analysis & high QE PMT

2.5GHZ sampling

Energy and timing resolution will be improved hopefully.

In particular pile-up rejection power will be up.

QE: 5% to 15% ave.


Calibration in MEG experimentCalibration in MEG experiment

Absolute calibration Energy (once a week/month…)

- 0-> decay Position

- radiative muon decay Timing

- radiative muon decay

relative adjustment (daily)

- wire alpha source, cosmic-ray muon


Energy calibration with Energy calibration with 00 decay decay

- (at rest) + p -> 0 + n, 0(28MeV) -> 54.9MeV<E<82.9MeV)

Almost monochromatic is available by selecting opening angle.170°175°

54.9MeV 82.9MeV

Opening angle(deg)

Ene

rgy

(MeV

)

155 180 55 80

5580

Energy (MeV)

0.3 MeV FWHM

1.3 MeV FWHM


test with test with 00 decay @ PSI decay @ PSI

8x8 NaI array opposite to Large Prototype Xe detector to know opening angle each NaI is 6.3x6.3x40.6cm3Liquid H2 target


Energy ResolutionEnergy Resolution

4.5% FWHM

Energy spectrum @ 54.9 MeV This satisfies the requirement

5.04.0

3.0

2.0

1.0

Energy resolution in right (%)

preliminary

TERAS data PSI data

Right is a good function of energy


Proton beam: 1.8mAPion Rate: 2x108 -/secCollimate: 2PMTs x 2PMTs ~ 150cm2

20 /sec

# of PMTs on incident face: 216PMTsrequired: 30,000 evts/run

takes 81,000sec~ 1day

Energy CalibrationEnergy Calibration

Anti Counter

up

tilt

down

Support structure: straightly up and downTilt mechanism at every height for NaI front to face target direction.

target

00

55, 83MeV

55, 83MeV


Expected PerformanceExpected Performance


Expected PerformanceExpected Performance

Generate background events– Accidental overlap

e+from Michel decayfrom radiative decay, e+-annihilation in fli

ghtPile-ups

– Radiative + decay

f

f

Bacc∝δEe ・ δte ・ (δE)2 ・ (δe)2 ・N


90 % C.L. Sensitivity90 % C.L. Sensitivity

5.8x10-14

4.5x107

Solid angle: /4=0.09Positron detection efficiency: e=0.90Gamma detection efficiency: =0.67Event selection efficiency: sel=0.7

Gamma energy resolution: 5.2%Gamma position resolution: 4mm-11mmGamma timing resolution: 179 psecPositron energy resolution: 0.8%Positron timing resolution = 100 psecPositron position resolution = 10.5 mrad


Decay Angular MeasurementDecay Angular Measurement

Conditions:+ beam intensity : 4.5x107/secbeam time : 5.2x107sec (2 years)P=97% (from MEGA)

MEG search with sensitivity down to 10-13 with depolarized μ+ beamMuon Stopping target: polyethylene

Define the model of the new physics with polarized μ+ beam

Muon Stopping target: graphite, Be, Al,…


SUSY models and AsymmetrySUSY models and Asymmetry

SU(5) SUSY-GUT A = +1 (AL0, AR=0)

SO(10) SUSY-GUT A 0 (ALAR)

MSSM with R A = -1 (AL=0, AR0)

Detector acceptanceY.Kuno et al.,Phys.Rev.Lett. 77 (1996) 434

• dN(+e+) / dcose Br(+e+)x(1+APcose)/2 Asymmetric Parameter;A = (|AL|2-|AR|2)/(|AL|2+|AR|

2)


Background distributionBackground distribution

• Radiative + decay– N(+e+) ∝ (1+Pcose)

• e+ from Michel decay ∝ (1+Pcose)

• from radiative decay ∝ (1+Pcos)

• Back to back : =e-

⇒ 　 Accidental B.G. ∝ (1-Pcos2e)

～ 11% suppression @ P=97%, cose=0.35

Detector acceptance

Detector acceptance


68%, 90%, 95% C.L. Sensitive Regions68%, 90%, 95% C.L. Sensitive Regions

68.3%90.0%95.4%

A=+1B(+e+) = 10-12

• Poisson-distributed

2= [2(Nobs-Nexp)+Nobsloge(Nobs/Nexp)]

Assumptions


SummarySummary

We proposed a novel LXe scintillation detector for the MEG experiment. A 100-liter detector was constructed to design the final detector. The components such as monitoring system, PMTs, and, especially the cryostat, worked as expected. We developed a purification technique and absorption length reached over 1m at 90 % C.L. The detector performance was estimated to be E/E=2%, pos>4mm for 52.8 MeV in MEG experiment.Sensitivity of MEG detector was estimated to be 5.8x10-14 90%C.L. The final detector was already designed, and is ready for construction toward the physics run in 2006.


Reachable sensitivity by MEG Reachable sensitivity by MEG experimentexperiment

Physics run in 2006

5.8x10-14 90% C.L.


End of SlidesEnd of Slides


PMT calibrationPMT calibration

Gain (HV)∝ 9


LXe operationLXe operationEvacuating: ~10 days downs to an order of 10-3 Pa.Pre-cooling: 1 day The 0.2 MPa GXe in the inner vessel is cooled to 165 K in advance.Liquefaction: 2 days Liquefied with LN2 cooling pipe. GXe is purified before entering the vessel. The pressure is kept <0.13 MPa.LXe-keeping: 2000 hours max. Mainly by refrigerator. By LN2 if over 0.13 MPa.Purification : ~1200 hours flow rate = 5 liter/min (GXe)Recovery: 2 days Cool Xe tank storage with LN2. Warming-up: 3 days


DustboxDustbox

The sE is evaluated to be ~1% from the extrapolation to 52.8 MeV.

The st is evaluated to be ~50 psec for 20,000 photoelectrons estimated from the result simulated for 52.8-MeV g.

● 8 LEDs inside PMT holder● Scan HV to get gain curves for all PMTs


10-liter (small) prototype10-liter (small) prototype

● g sources (137Cs, 51Cr, 54Mn, 88Y) Resolution evaluation● a source (241Am) PMT calibration, stability check● LED PMT calibration

■ 2.34-liter active volume■ 32 PMTs

Purpose First “Kamiokande”-like LXe detector Test for R6041Q in LXe and cryostat for LXe. Estimate of the performance for low energy g rays Energy, time, and position resolutions with < 1.8-MeV g sources.

116mmX116mmX 74mm


Energy ResolutionEnergy Resolution

Fully-contained events in each energy distributions are fitted with an asymmetric Gaussian.


Position ResolutionPosition Resolution

PMTs are divided into two groups.γ int. positions are

calculated in eachgroup and thencompared witheach other.

Events in the central2cmX2cm area are selected.

Position resolutionis estimated as sz1-z2/√2


• PMTs are divided again into two groups.• In each group the average of the time measured by TDC

is calculated after slewing correction for each PMT.• The time resolution

is estimated bytaking the differencebetween two groups.

• Resolution improvesas ～ 1/√Npe

• FWHM<120 psec for 52.8 MeV g.

Time ResolutionTime Resolution


How much water contamination?How much water contamination?

data/MCSimulated data

Before purification: ~10 ppm

After purification: ~100 ppb


Constructed the first LXe scintillation detector.The resolutions are evaluated for low energy g. Energy: 4.2~9.4%, Position: 6.3 ~ 19 mm, Time: ~380 psec (FWHM) If extrapolated to 52.8-MeV, resolutions are be expected: energy; ~1%, position; a few mm, time;~100 psec (FWHM)

Stable operation for the cryostat.PMT output fluctuation: ~ 0.5 %.

Summary on Small PrototypeSummary on Small Prototype


Absorption length estimationAbsorption length estimation

Comparing the two results,

the absorption length is estimated to be

over 3m (97.8% C.L.).

Measu

red /

M

C(a

bs=

inf.

)

MC

(ab

s=var.

) /

MC

(ab

s=in

f.)

measured a data/simulated a data

simulated data


Thickness of incidence faceThickness of incidence face

PATH A: 0.24 X0

PATH B: 0.24 X0

The Most of g-rays transmit to the LXe volume through the incident face.


Energy SpectrumEnergy Spectrum

data MC


Stability of PMT outputsStability of PMT outputs

The a particle data are useful for monitoring the stability of PMTs because it is regarded to be a point-like source.

After the completion of the liquefaction, the PMT output is stabilized within 1.5 % in 50 hours.

a±1.5%


Q.E. estimation with Q.E. estimation with aa particle data particle data

Compared with the simulated data, the a data in LXe can estimate Q.E.s, which include collection efficiencies of the PMTs.


Position Resolution at P0 & P9Position Resolution at P0 & P9

MC

data


Position ReconstructionPosition Reconstruction

depth=4.4cm

depth=15.3cm

qi: number of p.e. observed by i-th PMT

Qf: Total number of p.e. observed by i-th PMT

Ci: Ratio of overlap of the red circle for i-th PMT

Simple weighted mean

Local weighted mean


Position Resolution at P0 & Position Resolution at P0 & P9(update)P9(update)

MC data

If the range of Nf(0.5) is from 2.0 to 20.0 and fitted with 2-D double Gaussian

~10%

~90% ~90%

~10%


Position Fitting (old)Position Fitting (old)

Data; 40 MeV P0Data; 40 MeV P9

Fitted with 2-D Gaussian


MC


Event SelectionEvent Selection

Correction

4<Nf(0.5)<13


TriggerTrigger


Energy Resolution at P0 & P9 (old)Energy Resolution at P0 & P9 (old)

MC data


Energy Resolution at P0 & P9 Energy Resolution at P0 & P9 (update)(update)

MC data

古い

If the range of Nf(0.5) is from 2.0 to 20.0


Fitting (old)Fitting (old)

Data; 40 MeV P0Data; 40 MeV P9

Fitted with 2-D Gaussian


Vertex ReconstructionVertex Reconstruction

depth=4.4cm

depth=15.3cm

qi: number of p.e. observed by i-th PMT

Qf: Total number of p.e. observed by i-th PMT

Ci: Ratio of overlap of the red circle for i-th PMT

Simple weighted mean

Local weighted mean


Sensitive Region for A=+1Sensitive Region for A=+1

• Excluded region for each B(+e+) @ A=+1

68.3%90.0%95.4%

excluded

Allowed


Mask analysis 1Mask analysis 1

– Switch off half of the PMTs in the front faceUse 4x4 PMTs out of 6x6 PMTs

– Switch off PMTs on the side walls


Mask analysis 2Mask analysis 2Only 4x4 PMTs on the front face• Switching off half the front PMTs

– Compton Edge shifts by 6.2%– Resolutions are almost same (1.84 to

1.85% in ) before and after switching off.

• Switching off PMTs on theside wall(s)– 1 plane off 2.05% in – 2 planes off 2.22% in – 3, 4 planes off > 3% in

4 planes off

3 planes off

2 planes off

1 plane off

1/Npe

1/sqrt(Npe)

Number of Photoelectrons



1 plane off

3 planes off

• Deterioration of the energy resolution when switching off PMTs is not mainly caused by loss of Npe.

• PMTs on the side walls compensate 1st conversion point dependence.



D D



• All PMTs on: =1.8%

• Switching off one PMT on the front wall.– the nearest PMT

=2.3%– 2nd nearest PMT

=1.9%– 3rd nearest PMT

=1.9%

• 300 PMTs on the front face in the final detector– ~4/300 = 1.3% loss

of acceptance

F30 off

=2.3%

F22 off

=1.9%

F28 off

=1.9%


Pile-up rejectionPile-up rejection

27/jan/2005 k. ozone 2004 年度博士論文審査会 contents 1. meg experiment 2. lxe purification...

Documents