MEG Experiment

Motivation & Event Signature

Only allowed after KamLAND

LFV process Forbidden in the SM Sensitive to SUSY-GUT, SUSY-seesaw etc. Our goal : Br(->e)>10-13~10-14

52.8 MeV

52.8 MeV

180°

Clear 2-body kinematics

->e branching ratioPresent limit:1.2x10-11

Michel decay (μ ＋→ e ＋ νeνμ) + random γ　　　　　　　　　 Background Rate ~ 10-

14

Radiative muon decay (μ ＋→ e ＋ νeνμγ)　　　　　　　　　 Background Rate < 10-

14

MEG Experiment & DetectorApproved in 1999,at Paul Scherrer Institut, in Switzerland

Physics run in 2006Initial goal at 10-13, finally to 10-14

+ beam : World’s most intense 　 DC Beam 108 + /s

detector : 800liter liquid xenon scintillation detector with 830 PMTs

e+ detector : solenoidal magnetic spectrometer with a gradient magnetic field (COBRA)

COBRAXe Detector

Drift　 Chamber

TimingCounter

262cm

252cm

Paul Scherrer InstitutPaul Scherrer Institut

ExperimentalHall

• The most powerful machine in the world.• Proton energy: 590MeV• Nominal operation current: 1.8mA.• Max > 2.0mA possible.

Small Prototype

Small PrototypeSmall Prototype

• 32 2-inch PMTs surround the active volume of 2.34 liter

• -ray sources of Cr,Cs,Mn, and Y

• source for PMT calibration• Operating conditions

– Cooling & liquefaction using liquid nitrogen

– Pressure controlled

– PMT operation of 1.0x106 gain•Proof-of-Principle Experiment

•PMT works in liquid xenon?

•Light yield estimation is correct?

•Simple setup to simulate and easy to understand.

•Proof-of-Principle Experiment

•PMT works in liquid xenon?

•Light yield estimation is correct?

•Simple setup to simulate and easy to understand.S.Mihara et al. IEEE TNS 49:588-591, 2002

S.Mihara et al. IEEE TNS 49:588-591, 2002

Small PrototypeSmall PrototypeEnergy resolutionEnergy resolution

• Results are compared with MC prediction.1. Simulation of int. and energy de

position : EGS4

2. Simulation of the propagation of scint. Light

EGS cut off energy : 1keV

Rayleigh Scattering Length: 29cm

Wph = 24eV

Small PrototypeSmall PrototypePosition and Timing resolutionsPosition and Timing resolutions

• PMTs are divided into two groups by the y-z plane

– g int. positions are calculated in each group and then compared with each other.

– Position resolution is estimated as

　　　　 sz1-z2/√2

• The time resolutionis estimated bytaking the difference

between two groups. • Resolution improves

as ～ 1/√Npe

Large Prototype

Large Prototype

• 70 liter active volume (120 liter LXe in use)– 372x372x496 mm3: 17.3 R.L. in depth

• Development of purification system for xenon

• Total system check in a realistic operating condition:

– Monitoring/controlling systems• Sensors, liquid N2 flow control, refrigerator operat

ion, etc.– Components such as

• Feedthrough,support structure for the PMTs, HV/signal connectors etc.

– PMT long term operation at low temperature• Performance test using

– 10, 20, 40MeV Compton beam– 60MeV Electron beam– from 0 decay

TERAS Beam

• Electron beam (TERAS, Tsukuba in Japan)– Energy: 764MeV– Energy spread: 0.48%(sigma)– Divergence: <0.1mrad(sigma)– Beam size: 1.6mm(sigma)

• Laser photon– Energy: 1.17e-6x4 eV (for 40MeV)– Energy spread: 2x10-5 (FWHM)– Divergence: unknown– Beam size: unknown

Compton Spectrum

•(E-Ec/2)2+(Ec/2)2

Collimator size

10MeV

20MeV40MeV

Energy Spectrum Fitting

• Principle…

E Npe

Convolution of

Compton Spectrum

Response Function

Suppose Compton Spectrum around the edge

(E-Ec/2)2+Ec2/4Detector Response Function

Gaussian with Exponential tailf(x) = N*exp{t/2(t/2-(x-x0)}, x<x0+t N*exp{-1/2((x-x0)/)2}, x>x0+t

ConvolutionIntegration +/- 5

E~1.9%

1. TERAS Beam TestInverse Compton Energy : 10, 20, 40 MeV Incident Position : Detector CenterEstimate Energy Resolution using Compton Edge

1.6% in

40MeVEnergy

40MeVVertex distribution

3.8mm in

0 Beam Test at PSI

180°

Opening angle selection of two ’s monochromatic

- (at rest) + p -> 0 + n, 0(28MeV/c) ->

Concept

54.9MeV

82.9MeV

Energy (MeV)

175°

Opening angle (deg)

Ene

rgy

(MeV

)

155°

55M

eV80

MeV

170°

• Requiring

FWHM = 1.3 MeV

• Requiring > 175o

FWHM = 0.3 MeV0

Beam Test Setup

H2 target+degrader

beam

LPNaI

LYSO

Eff ~14%

S1Eff(S1xLP)~88%

8x8NaI115cm 115cm

-

n LP Xe

0 Beam Test at PSI

- (at rest) + p -> 0 + n, 0(28MeV/c) -> + monochromatic calibration of around 52.8MeV - + p -> n(8.9MeV) + (129MeV) linearity check & neutron response

LP X

e to

tal c

harg

e

NaI Energy (MeV)

-p->n0, 0->2

-p->n To get Energy Resolution　　 Select 0 events　　 Select NaI energy　　 Select incident position in Xe detector　　 Remove too shallow and too deep events

Energy ResolutionsE

Xe[

Nph

]

= 1.23 ±0.09 %FWHM=4.8 %

55 MeV

σ = 1.00±0.08 % FWHM=5.2%

83 MeV

CEX 2004

83 MeV to Xe

55 MeV to Xe

Right is a nice function of gamma energy

PSI 2003TERAS 2003alpha

Energy Resolution vs Energy

Position Reconstruction

• Localized Weight Method

• Projection to x and y directions.

• Peak point and distribution spread

•Position reconstruction using the selected PMT

Examples of Reconstruction

(40 MeV gamma beam w/ 1 mm collimator)

Position ReconstructionResolution

Reconstruction of the event depth

• Using event broadness on the inner face

• Necessary to achieve good timing resolution

3 cm

Liq. Xe

Liq. Xe

14 cm

(a)

(b)

05

1015

2025

3035

0

10

20

30

40

50

0

2000

4000

6000

8000

10000

05

1015

2025

3035

0

10

20

30

40

50

0

200

400

600

800

1000

1200

1400

1600

1800

52.8MeV

52.8MeV

• D: depth

parameterMC simulation Data

Previous Test

This Test

Short abs

Long abs

D=

D

DD

DD

D: 20~100 0~25cm

Energy Resolution vs. Depth Parameter

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

30 40 50 60 70 80 90

Sigma2 threshold

Resolution (%)

Number of Photoelectrons

D

• For incident at the detector center

• D > 35, 45, 55….85• Resolution: < 2% in sigma

except shallow events (D<45).

Nfpmt(0.5)

• Another depth parameter: Nfpmt(0.5)– Sort front-face PMTs in the order of th

e distance to the x-y weighted mean position.

– Sum up PMT output in that order.– Stop summing when the sum exceeds the h

alf of Qsum[Front]. Number of PMTs used in the SUM.

1.0

0.5

QPMT

NPMT

Shallow

Deep

Timing/Z Resolution

• Improving Z resolution is essential to improve timing resolution.

• Intrinsic timing resolution can be evaluated by comparing left and right parts of the detector.

– <T> = (TLTR)/2

XenonNaI S1

LYSO tLP - tLYSO

-

3 cm

Liq. Xe

Liq. Xe

14 cm

(a)

(b)

05

1015

2025

3035

0

10

20

30

40

50

0

2000

4000

6000

8000

10000

05

1015

2025

3035

0

10

20

30

40

50

0

200

400

600

800

1000

1200

1400

1600

1800

52.8MeV

52.8MeV

Left

Right

TL

TR

Absolute timing, Xe-LYSO analysis55

MeV

high gainnormal gain

110 psec 103 psec

LYSO Beam L-R depth reso.

110 64 61 = 65 = 56 33 psec

103 64 61 = 53 = 43 31 psec

No

rma

l g

ain

Hig

h

ga

in

A few cm in Z

Timing ResolutionTiming Resolution• Estimated using Electron Beam

(60MeV) data

• Resolution improves in proportion to 1/sqrt(Npe).

• For 52.8 MeV ~ psec + depth resolution.

• QE improvement and wave-form analysis will help to achieve better resolution.

(Visit “The DRS chip” by S.Ritt)

T

imin

g R

esol

utio

n (p

sec)

104 4x104

45 MeV Energy deposit by 60 MeV electron injection

52.8MeV

(nsec)

=75.62.0ps=75.62.0ps

Number of Photoelectron

<TL>-<TR>

PMT Development

Photomultiplier R&D• Photocathode

– Bialkali :K-Cs-Sb, Rb-Cs-Sb• Rb-Cs-Sb has less steep increase of sheet resistance

at low temperature• K-Cs-Sb has better sensitivity than Rb-Cs-Sb

– Multialkali :+Na• Sheet resistance of Multialkali dose not change so m

uch.• Difficult to make the photocathod, noisy

• Dynode Structure– Compact– Possible to be used in magnetic field up to 100G

• Metal channel Uniformity is not excellent

Ichige et al. NIM A327(1993)144

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 eff.)(includes collection eff.)

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

57 mm

32 mm

1st generation R6041Q 2nd generation R9288TB 3rd generation R9869

228 in the LP (2003 CEX and TERAS)

127 in the LP (2004 CEX)

111 In the LP (2004 CEX) Not used yet in the LP

Rb-Sc-Sb

Mn layer to keep surface resistance at low temp.

K-Sc-Sb

Al strip to fit with the dynode pattern to keep surface resistance at low temp.

K-Sc-Sb

Al strip density is doubled.

4% loss of the effective area.

1st compact version

QE~4-6%

Under high rate background,

PMT output reduced by 10

-20% with a time constant of

order of 10min.

Higher QE ~12-14%

Good performance in high rate BG

Still slight reduction of output in very high BG

Higher QE~12-14%

Much better performance in very high BG

PMT Development Summary

MotivationUnder high rate background, PMT output (old Type PMT, R6041Q) reduced by 10-20%.This output deterioration has a time constant (order of 10min.): Related to the characteristics of photocathode whose surface resistance increases at low temperature.

Rb-Sc-Sb + Mn layer used in R6041QNot easy to obtain “high” gain. Need more alkali for higher gain.Larger fraction of alkali changes the characteristic of PC at low temp.

So, New Type PMTs, R9288 (TB series) were testedunder high rate background environment.

K-Sc-Sb + Al strip used in R9288Al strip, instead of Mn layer, to fit with the dynode pattern

Confirmed stable output. ( Reported in last BVR)But slight reduction of output in very high rate BG

Add more Al Strip

Al Strip PatternLow surface resistance

R9288 ZA series

Yasuko HISAMATSU MEG VRVS Meeting @PSI June 2004

Gas Purification

Possible Contaminants

• Residue Gas Analysis– Vacuum level

• LP Chamber 2.0x10-2Pa• Analyzing section 2.0x10-3Pa

HeH2O CO/N2

O2

CO2Xe

H2O is the most dangerous for Xe scintillation light

Purification System• Xenon extracted from the chamber is

purified by passing through the getter.• Purified xenon is returned to the chamber

and liquefied again.• Circulation speed 5-6cc/minute

Gas return

To purifier

Circulation pump

Purification SystemPurification System• Xenon extracted from the

chamber is purified by passing through the getter.

• Purified xenon is returned to the chamber and liquefied again.

• Circulation speed 5-6cc/minute

• Enomoto Micro Pump MX-808ST-S– 25 liter/m

– Teflon, SUS

Gas return

To purifier

Circulation pump

How much water contamination?

Measured/MCSimulated data

Before purification: ~10 ppm

After purification: ~10 ppb

Purification Performance• Xenon Detector Large

Prototype• 3 sets of Cosmic-ray trigger

counters• 241Am alpha sources on the

PMT holder• Stable detector operation for

more than 1200 hours

Cosmic-ray events events

Xenon Purifier

• Attenuation of Sci light– Scintillation light emission from an excited molecule

• Xe+Xe*Xe2*2Xe + h

– Attenuation• Rayleigh scattering Ray~30-45cm

• Absorption by impurity

Absorption Length• Fit the data with a function :

A exp(-x/ abs)• abs >100cm (95% C.L)

from comparison with MC.• CR data indicate that abs >

100cm has been achieved after purification.

Liquid Phase Purification

Liquid-phase Purification Performance

In ~10 hours, λabs ~ 5m

PET vs MEG

• 511 keV vs 52.8 MeV– Smaller number of photo-electron– Shallower detector

• Read out from back vs from front– Smaller light collection– Efficient detection

• Small amount of material in front• Efficient fiducial volume

• Mild requirement on resolutions

The End