Performance of the Spaghetti Performance of the Spaghetti calorimeter with individual fibercalorimeter with individual fiber--readoutreadout

T.Matsumura, T.Tojo, T.Shinkawa (N.D.A.)T.Yamanaka (Osaka Univ.)Y.Kato, T.Sawada, T.Hotta (RCNP)H.Watanabe (KEK)R.Ogata, S.Suzuki, T.Shimogawa (Saga Univ.)Y.Tajima (Yamagata Univ.)

01 November, 2006Joint Meeting of Pacific Region Particle Physics Community

Contents • Introduction

KL rare-decay experiment at J-PARCSpaghetti calorimeter for the pre-shower detector

• Performance of prototype detectorBeam test with photon beamShower tracking

e+e- conversion-pointAngular resolution

KL rare-decay experiment

Purpose:Purpose: Search for new physics beyond the S.M.Search for new physics beyond the S.M.by measuring the Kby measuring the KLL→→ππ00νννν decay decay

Br (KKL L →→ ππ00νννν) ~) ~ (3.0 (3.0 ±± 0.6)0.6)××1010--1111 (S.M.)(S.M.)

KEK E391a KEK E391a Feb. 2004 - Dec. 2005•• establishing the experimental method establishing the experimental method •• update the upper limitupdate the upper limit

JJ--PARC PARC -- KKL L •• first observation of the first observation of the KKL L →→ ππ00νννν decay decay

Br (KKL L →→ ππ00νννν) <) < 2.12.1××1010--77 (current upper limit: KEK(current upper limit: KEK--E391a 2006)E391a 2006)

Basic design of the experiment

~ 15m~ 15m

Veto counter

End

cap

CsIDecay Region γ

KL

γν

ν

prepre--shower detectorshower detector• Signal identification

2γ + nothing

• Main backgroundKL → π0π0 → 4γ

limited constraint

• 2γ missing• 1γ missing + 2γ fusion

• Pre-shower detector2γ separationphoton directionKL→π0νν decay

Candidate of the pre-shower detector

SpaghettiSpaghetti--type calorimeter (SPACAL)type calorimeter (SPACAL)

can be obtained with the spaghetti calorimeter.

ScitillatingScitillating fibers and lead radiator fibers and lead radiator

grooved lead plate and fibersfiber diameter 1 mmφfiber interval 1.35 mm

Photon energy, direction, position

report of the KLOE experiment

By measuring the fibers individually,

Feature of the SPACAL (1)

2 photons Eγ1 = Eγ2 = 0.3 GeVR12 = 65 mm

y(m

m)

x(m

m)

z (mm)

z (mm)

Fiber-hit distribution

X X

X XY Y

Y

γ1γ2

γ1γ2

End

-cup

CsI

End

-cap

CsI70 mm

70 mm

MC Demonstration(simple XYXY configuration)

Conversion efficiency~ 90 % with 3.2 X0

=5 cm thickness Compact

Energy resolutionσE/E = 2 % for Eγ=1 GeV

( 1.5 % for CsI only )

Photon separation efficiency~100 % for >20 mm distance

Eγ >0.1 GeV

Eγ = 1.00 GeVEγ = 0.50 GeVEγ = 0.25 GeVEγ = 0.10 GeV

num

ber o

f fib

er-h

its /

even

ts

Energy deposit in a fiber (MeV)

MIP(150 keV)

No ADC is needed for energy measurement. Simplification of the readout electronics

The distribution of the energy deposit does not change so much among the all different energies.

Feature of the SPACAL (2)

Photon energy can be estimatedby just counting the number of fibers

Performance of the prototype detector

Prototype Detector10 x10 x 9.7 cm3

fibers(5,850)

fibers(5,850)• trigger timing

• energy deposit 75 cm• shower profile

light guide

~ 6 X0

IIT-1

IIT-2

CCD

γPMT

Image IntensifierImage Intensifier (2 stages)• IIT-1: image reduction (1/16)• IIT-2: amplification(x106), gate (100μs)

CCD cameraCCD camera• 768(H) x 494(V), 30 flames/sec

IIT+CCD system

Fiber bundleTotal: 5850 fibers

bundle unit( 3 x 25 fibers)

photon beam

Prototype detector

Pictures

Image Intensifier

PMT

Setup of the beam test

•• Photon beam @ SPringPhoton beam @ SPring--88– Eγ : 1.5 ~ 3.0 GeV tagged photons

•• CollimatorCollimator (1 mm2 hole)– avoiding the image overlap

CollimatorCollimator1x1 mm1x1 mm22

PMTPMT

IIT+CCDIIT+CCD

SPACALSPACALprototypeprototype

active collimatoractive collimator5 5 mmmmφφ

ACAC

γγ beambeam

scintillatorscintillator

10 kHz10 kHz

•• TriggerTrigger– self trigger (Eγ > 300 MeV)– rate: 16 Hz

TOP VIEW

γ beamrearrangement

top

bottom

3x25 fibers

9.7 cm (~6X0)

10 c

m

IIT effective area (10cmφ)

Detector (side view) IIT surface

Fiber rearrangementto maximize the number of readout-fibers

Fiber bundle

CCD image is not directly related to the fiber position in the detector.

Typical image

Accumulated image

Dead regionX (pixel)

Y (p

ixel

)A shower event

A cosmic-ray event

Reconstruction of the cosmic-ray track

CCD coordinate detector coordinate

251 pixels fired 107 fibers fired

x (pixel)

y (p

ixel

)

z (mm)

y (m

m)

Note: flip vertical

Shower tracking

Eγ = 2.1 GeV, real data

z (mm)

y (m

m)

1. Remove the isolated clusters by requiring cluster-size cut (Nsize)

2. Fit a straight line to the weighted mean of the fiber-hits in each layer

3. Select only shower core-regionby applying the Δ cut

4. Estimate the conversion point (z0)from the most upstream fiber in the selected region

γ

Angle of incident photons was estimated by using information of fiber hits in the core region.

Typical shower events

Δ

z0

Effect of shower-core selection

Eγ ~ 2.1 GeV, Nsize≧4

Δ (mm)

σ θ(r

ad)

MCdata

less shower fluctuation

• By selecting the shower-core region the angle resolution would be improved

• The dependence was not so significant, but improvement (~20 %) was seen in both data and MC.

Δ ~ 8 mm is sufficient

Conversion Point

Eγ (GeV)

z0 (mm)

Eγ ~ 2.1 GeVReal data

ε( |

z 0-z

true|

< 4

mm

) Reconstruction Efficiency

• Vertex resolution in y at the conversionpoint was independent of energies

~ 0.7 mm

• Reconstruction efficiency (<4mm ) is more than 95 % above 1 GeV photons

• Reconstructed z0 distribution shows an exponential curve as we expected

Tight Nsize cut results in the worse efficiencydue to the suppression of the small clusters near the conversion point.

Nsize≧2Nsize≧4Nsize≧6

limited by the position resolution of the fiber hits

MC

even

ts /

bin

Tracking length (Lz)

Eγ ~ 2.1 GeV, Nsize≧4, Δ = 8 mm

σ θ(r

ad)

Lz (X0)

X0 = 16 mm

MCdata

Lz

Large Lz→ Entire shower information

Small Lz→ Only the information of the early stage

of the shower development

z0

Deterioration at very small Lz regionswas due to the limitation by the position resolution of the fiber hits.

minimum at ~2.5 X0

Length from conversion point (z0)for straight line fitting.

• Good agreement between data and MC

Energy dependence

Eγ (GeV)

Nsize≧4, Δ = 8 mm, Lz = 2.5X0

σ θ(r

ad)

MCdata

Data point ranges 1.5 ~ 2.8 GeV( tagged photon energy)

37 mrad @ 2 GeV

• Angular resolution

Nfiber VS Energy deposit

ADC (PMT)

Nfib

er

dE ~ 150 MeV CCD

γ

PMT

Eγ ~ 2.1 GeV

Number of fiber-hits and the energy deposit measured by PMT

Linear correlation

Energy deposit can be measuredby just counting the number of fiber

~ 6 X0

Future plans• Improvement of the sampling ratio

Present setup with round fibers : 12 %

with 1 mm2 square fiber : ~ 30% is possible

Study of the basic performance ofthe M-APD samples (JINR, Russia)has been started.

1 mm

• Study of devices for the fiber readout

Square fibers(1x1 mm2)

Summary

• A pre-shower detector with spaghetti-type calorimeter was studied for– better 2γ separation– photon direction measurement

• Prototype detector of the Spaghetti-type calorimeter was constructed and tested with the photon beam.– angular resolution : 37 mrad @ 2 GeV– position resolution : 0.7 mm (independent of energies)

• For the future plan– Improvement of the photon detection inefficiency– Development of the readout device (M-APD)

Backup Slides

Energy resolutionσ E

/E (%

)

SPACAL thickness (mm)

Eγ = 0.1 GeV

Eγ = 1.0 GeV

Assumption:CsI Npe = 10p.e./MeVSPACAL Npe = 30p.e./MeV

50 1000

X (pixel)

Y (pixel)

1 mm @ IIT entrance → 4.7 pixel @ CCD

2.2 pixels @ CCD (0.47 mm @IIT entrance)

Trigger counterScintillating fiberKuraray SCSF-78IIT

90Sr

Gain 5.0, gate width 100μs

50 cmcollimator

CCD

• Monte-Carlo (MC) simulationGeant4 + pixel simulation was developed.

• Performance (evaluated by the 90Sr bench test)

Efficiency (1 pixel found)90 % for MIP energy (0.15 MeV)

1 fiber is covered by ~ 5x5 pixelsResolution

Performance of the IIT-CCD system

Position resolution of the fiber hits

• position resolution of the IIT-CCD is σ ~ 2.2 pixel (= 0.47mm)

X (pixel)

Y (pixel)

the pixel hits could leak intothe next fiber region

The position resolution of fiber hitswas estimated by MC

σy = 0.6 ~ 0.9 mmσz = 0.45 ~ 0.7 mm

depending on the bundle units

Position distribution @ z0

Eγ ~ 1.5 GeV Eγ ~ 2.1 GeV Eγ ~ 2.8 GeV

y (mm) y (mm) y (mm)

Position resolution for y direction at z0 is independent of energy.~ 0.7 mm

Image to fiber-hits

1. For all fired pixels, determine the fiber cell which the pixel belongs to

2. Summing up the number of pixels containing fibers

3. Fiber position in the detector coordinate is assigned according to the fiber position map.

3x25 fiber cellscell size ~ 5x5 pixels