The MEG experiment at PSI

1
The MEG experiment at PSI

• Fabrizio Cei
• INFN University of Pisa
• On behalf of the MEG Collaboration
• NOW2008 Workshop

2
Outline

• Introduction on physics motivations
• The MEG experiment
• - Goals
• - Experimental set-up
• - Detector calibration performances
• Some results of 2007 engineering run
• Conclusions and perspectives

3

• Physics motivations

4
Physics motivations 1)

• Lepton Flavour Violation is an experimental fact
  in the neutral sector. Neutrino oscillations
  observed for solar
• and atmospheric neutrinos by SK, SNO, MACRO,
  Soudan, Gallex, Borexino, for reactor neutrinos
  by Kamland and
• for accelerator neutrinos by K2K.
• Some mixing matrix parameters already measured.
• Physics beyond the standard model is a reality !
• Next step Lepton Flavour Violation expected by
• strong theoretical motivations in
  the
• charged sector too (see previous
  talk)

5
Physics motivations 2)
Predictions of LFV reactions particularly
interesting in supersymmetric and
GUT-supersymmetric schemes. For m?eg process, the
expected branching ratio with respect to normal
muon decay ranges mainly between 10-15 and
10-12 (Barbieri, Hall, Strumia, Hisano, Isidori,
Ellis, Masiero ). Note that also a negative
result by a high sensitivity experiment would
significantly constrain the SSM parameters.
Important impact in any case on LHC
experiments. Example of recent prediction (G.
Isidori, talk _at_ Neutrino Oscillations in Venice,
Apr. 2008)
6
Past LFV experiments
Previous LFV searches with muons.
Muons are very sensitive probes to study LFV

• intense muon beams
• can be obtained at
• meson factories
• muon lifetime is
• rather long (2.2 ms)

7

• The MEG experiment

8
The MEG goal
FWHM
Need of a DC beam
Exp./Lab Year DEe/Ee () DEg /Eg () Dteg (ns) Dqeg (mrad) Stop rate (s-1) Duty cyc.() BR (90 CL)
SIN 1977 8.7 9.3 1.4 - 5 x 105 100 3.6 x 10-9
TRIUMF 1977 10 8.7 6.7 - 2 x 105 100 1 x 10-9
LANL 1979 8.8 8 1.9 37 2.4 x 105 6.4 1.7 x 10-10
Crystal Box 1986 8 8 1.3 87 4 x 105 (6..9) 4.9 x 10-11
MEGA 1999 1.2 4.5 1.6 17 2.5 x 108 (6..7) 1.2 x 10-11
MEG 2010 0.8 4 0.15 19 2.5 x 107 100 1 x 10-13
Improvement by two orders of magnitude ! A tough
experimental challenge possible, but excellent
detector resolutions are needed.
9
The MEG Collaboration
? 80 people
10
MEG signal and background
background
signal ? ? e g
accidental ? ? e n n ? ? e g n n ee ? g g eZ ?
eZ g

• physical
• m ? e g n n
• (radiative decay)

qeg 180 Ee Eg 52.8 MeV Te Tg
Accidental background limits the sensitivity
g
11
MEG detection technique
Stopped beam of 3x107 ?/sec in a 150 mm
target Liquid Xenon calorimeter for ? detection
(scintillation) fast 4 / 22 / 45 ns
high LY 0.8 NaI
short X0 2.77 cm
? decay at rest
Solenoid spectrometer (COBRA) drift chambers
for e momentum measurement Scintillation
counters for e timing
12
A simulated event
Hits on Xenon Calorimeter
Positron track on Drift Chambers
Hits on Timing Counter
13
The Paul Scherrer Institute (PSI)

• The most powerful continuous machine in the
  world
• Proton energy 590 MeV
• Power 1.1 MW
• Nominal operational current 2.0 mA.

14
MEG beam line 1)
PSI pe5 surface muon beam
Maximum muon flux

• 27.7 MeV/c muons from p stop at rest (surface
  muons)
• Provides a DC beam of ? 108 m/s.

15
MEG beam line 2)

• The beam elements
• Wien filter for m/e separation
• Degrader to reduce the momentum stopping in a 150
  mm CH2 target
• Transport Solenoid to couple beam with COBRA
  spectrometer

• Results (4 cm cyclotron target)
• Z-version
• Rm (total) 1.3108 m/s
• Rm (after W.filter Coll.) 1.1108 m/s
• Rm (stop in target) 6107 m/s
• Beam spot (target) s ? 10 mm
• m/e separation (at collimator) 7.5 s (12 cm)

Maximum beam stop rate ? 108 m/s, but we will use
only 3 x 107 because of accidental background
(proportional to (muon rate)2)
16
The COBRA magnet
Magnet equipped with 5 superconducting and 2
(compensating) conventional coils. Solenoidal
magnetic field with an axial gradient central
value 1.26 T. Field arranged to - sweep away low
Pz positrons - reduce the dependence of the
bending radius from
emission angle (COnstant Bending RAdius).
17
Drift Chambers 1)
MEG target
DC planes
18
Drift Chambers 2)
Sixteen drift chambers (ten degrees interval),
each one equipped with 18 staggered wires and
cathodic kapton foils. Wires r,f
coordinates Cathode z coordinate ?(X,Y) 200
mm (drift time) ?(Z) 300 mm (charge
division)
19
Timing Counter 1)
APD F.E. Board
APD Cooled Support
Fibers
APD
PM
Main Support
Divider Board
Scintillator Slab
Scintillator Housing
PM-Scintillator Coupler
20
Timing Counter 2)
Two layers of scintillation counters placed at
right angles with each other. Outer layer
scintillator bars, mainly devoted to timing
measurement. Two sections of 15 bars each,
read by PMTs, before and after DCH system. Inner
layer scintillating fibres, devoted to provide
trigger and z information. 5 x 5 mm2 fibres,
read by APDs.
Measurements of TC bars timing resolution in
dedicated test beams at several positions and
impact angles
FWHM
100 ps
FWHM(T) 91 ps (? 5) MEG Goal 100 ps
The best TC in the world !
Impact angle
21
Timing Counter 3)
TC after installation in MEG
22
Liquid Xenon calorimeter 1)

• 800 l of Liquid Xenon equipped with 846 PMTs
• Homogeneous detector
• Only scintillation light
• Large light yield ( NaI).
• PMT quartz windows to match LXe scintillation
• UV spectrum

Properties of Liquid Xenon
Measured gt 300 cm
23
Liquid Xenon calorimeter 2)
The PMTs
The cryostat
Hamamatsu R9288ZA Specifically developed for
low-temperature operation.Needed long RD
work (surface resistance, performances in high
photon background ). QE ? 12?14
24
Liquid Xenon calorimeter 3)
LXe purification used during 2007 run

• Xenon circulation in liquid phase through a
  molecular filter
• Pump 70 liter/hour
• Molecular filter to remove water contamination

Since we observed a light level lower than
expected, we tried to improve the purification
system for this year run by adding a O2 getter
to remove possible Oxygen contaminants. A
higher light level was obtained, even if not yet
the expected one. Continuous gaseous phase
purification is now used too.
25
Trigger digital system
2 boards
LXe front face (216 PMTs)
DAQ trigger rate 5 ? 10 Hz
2 x48
552 boards
LXe lateral faces back (216 PMTs) 4 in 1 lat.
(144x2 PMTs) 4 in 1 up/down (54x2 PMTs) 4 in 1
. . .
1 board
9 x 48
2 x48
1 board
9 boards
. . .
Timing counters curved (512 APDs) 8 in 1 bars
(30x2 PMTs)
9 x 48

• Based on simple quantities
• ? energy (QSUM)
• e - ? coincidence
• in time and direction
• LXe Timing Counter
• information (no DC).
• Built on a FADC-FPGA
• architecture.

1 x48
1 board
4 boards
Drift chambers 64 channels
4 x 48
2 x 48
Auxiliary devices 16 channels
2 x48
CR counters 32 channels
26
Readout electronics

• Waveform digitizing for all channels (pile-up
  rejection)
• Custom domino sampling chip designed at PSI
• 2.5 GHz sampling speed _at_ 40 ps timing
  resolution
• Sampling depth 1024 bins
• Readout similar to trigger
• Several versions
• presently installed version 2 3 well
  behaving with few per cent
• temperature dependence and ghost pulses
  problems
• version 4 chips delivered and under testing
  for future replacement.

27
Detector calibration and performances

• MEG is a precision experiment
• High experimental resolutions are mandatory
• (background level depends on resolutions)
• Good detector performances must remain
• stable for a 3 year scale.
• Electromagnetic calorimeter uses an innovative
• technology
• ? Frequent and reliable calibration procedures
• represent one of the fundamental quality
  factors
• for MEG.

28
Detector calibration and performances beam
intensity

• Measurement of muon beam intensities

• Positron rate on Timing Counter
• X-ray emission induced by muons
• traversing thin Ti,Cu .. sheets

• Ionization chamber using
• COBRA helium a-source
• for calibration

29
Detector calibration and performances DC target

• Target position and drift chambers alignment
  checked
• by optical survey. Target orientation known
  within 0.2o.
• Final alignment based on tracks
• - cosmic rays
• - Michel positrons ? x-t calibration, algorithm
  refinement

30
Detector calibration and performances DC hits
DC performances hit rate on single wire
Cyan histogram MC Red dots data

• Some problems during
• 2007 engineering run.
• HV Trips
• Some dead chambers ? low tracking efficiency (
  40 ) and tracking quality. Repaired or replaced
  with spares
• 2/32 planes still bad.

Absolute rate under control. Some MC details to
be revised.
31
Detector calibration and performances DC
resolution
DC performances momentum resolution

• Expected improvements in
• detector resolution
• tracking efficiency

FWHMp (meas.) 2.1 FWHMp (MC) 0.9
Decay vertex resolution sV ? 1 mm as
required (holes on target)
32
Detector calibration and performances TC 1)

• Timing counter parameters
• PMT Gain ? charge response
• Effective velocity
• Effective attenuation length.

PMT Gain
leff (cm)
40 Spread
Bar
Bar
33
Detector calibration and performances TC 2)
Effective velocity measurement (cm/ns)
Z (cm)
Veff determined with lt 1 precision by using
bar and fibre information simultaneously
Dt (ns)
Z (cm)
Dt (ns)
Dt (ns)
34
Detector calibration and performances TC
resolution
Timing resolution (timing difference between
adjacent bars)
In 2008 laser system. Two wavelengths 532nm
(TC) and 266nm (LXe) TC bar-bar and
TC-LXe timing calibration Laser under installation
sT ? 54 ps, not far from project goal (100 ps
FWHM)
35
Detector calibration and performances LXe PMTs

• LED
• Gain
• LED pulses with varying
• light intensity
• ?2 1/Nphe
• Relative timing
• Led pulse detected by
• several PMTs
• Alpha sources on wires
• Quantum efficiency
• In liquid and gas
• Measured vs expected Q
• Absorption length
• Fully tested on prototype
• NIM A 565 (2006) 589

36
Detector calibration and performances C-W
accelerator
The Cockroft Walton proton accelerator
The C-W team
37
Detector calibration and performances C-W
reactions
Reactions induced by C-W protons (E 1 MeV) on
Li and B targets

• Boron target
• Lower cross section
• Two gammas emitted
• simultaneously ? tool for
• relative timing calibration

Reaction Peak energy s peak g-lines
Li(p,?)Be 440 keV 5 mb (17.6, 14.6) MeV
B(p,?)C 163 keV 2 10-1 mb (4.4, 11.6, 16.1) MeV
Lithium spectrum on NaI
17.6 MeV line 14.6 MeV broad resonance
gt11.6 MeV 4.4 MeV
gt16.1 MeV
38
Detector calibration and performances LXe purity
Response to 17.6 MeV g-line as a function of
time
CW calib Li(p,g)Be 17.6 MeV, 14.6 MeV
Attenuation length measurement with a sources
(Data/MC) vs (PMT-source distance) labs gt 300 cm
_at_ 95 C.L.
39
LXe calibration with p0

• Charge Exchange Reaction
• p-p ? p0n
• 54.9 MeV lt E(g) lt 82.9 MeV

Liquid Hydrogen Target 124 cm3 Cooled with
liquid Helium (2.5 liters/hour)

• Auxiliary NaI detector
• 9 crystals
• Movable on (f,q) angles
• Opposite to LXe
• calorimeter to select
• photons at spectrum
• boundaries

• ?

40
LXe calibration linearity

• Linearity check over a wide photon energy
  spectrum
• CW and Lithium target 17.6 MeV
• CW and Boron target 4.4, 11.6 and 16.1 MeV
• p0 from liquid hydrogen target 54.9, 82.9 MeV

p0 reactions
CW reactions
41
LXe Energy Resolution _at_ 55 MeV

• Position selection
• Events in central NaI crystal
• Events reconstructed in LXe opposite to NaI
  center (5 cm)
• Shallow events rejected (2 cm)
• Pile-up rejected using LXe light distribution
  (PMT charge clusters)
• Not applied QEs

55 MeV ?
?up 2.4 FWHM 6.5

• 2 residual contribution
• from pedestal

In Large Prototype (4.8 ? 0.2) FWHM
42
LXe intrinsic timing resolution

• p0 data (54.9 MeV)
• Used 12 PMTs around the
• LXe calorimeter center, back
• to back with NaI.
• Low photoelectron number
• Compatible with proposal
• resolution after rescaling
• with sqrt(Nphe)

sT 115ps
(TT-TB)/2

• TT,B Weighted average of Top (T) or Bottom
  (B) PMT times after correcting for photon
  propagation time.

43
LXe TC relative timing calibration with Boron
1)

• 11.6 MeV and 4.4 MeV coincident gammas
• (small angular correlation)

LXe charge
11.6 MeV
11.6 MeV
4.4 MeV
4.4 MeV
TC charge
44
TC LXe relative timing calibration with Boron
2)
TC-LXe timing difference at target Correlation
between energy and timing resolution
4.4 MeV in LXe s 1180 ps
For coincident gammas at the target TTC-TLXe
-32 ns
11.6 MeV in LXe s 885 ps
Some problems with the cosmic ray background and
DRS2 chip
45

• Selected results of
• 2007 engineering run

46
General comments

• Beam lines and detectors were installed and were
  in
• operation together with DAQ, trigger and
  read-out
• We experienced some problems, but most of them
• were solved or under solution
• Several results already shown on detector
• performances, resolutions
• Two more
• - measurement of photon spectrum in LXe
  calorimeter
• - hints for radiative decay events detection.

47
Global simulation of photon energy spectrum in LXe
Red Radiative decay Green Annihilation
In Flight Black Cosmics
(approximated) Blue Total (including pile-up)
Rm 3.2 x 107 s-1
Knowledge and full understanding of this
spectrum is very important to evaluate
accidental background (dominant).
Energy (MeV)
48
Data vs MC comparison
Data (Blue Points) Beam _at_ 3.2 x 107 m s-1,
threshold ? 45 MeV MC (Black line) full
background simulation. Absolute rate
reproduced Pile-up subtracted by charge
distribution cosmics rejected. Final pile-up
rejection by using waveforms (not here).
Energy (MeV)
Energy (MeV)
49
Hints for a first observation of radiative decay
events

• Very weak signal
• Otherwise, it would be a very dangerous
  background for MEG
• All analysis tools are needed (its like a m?eg
  search )
• Energy, timing and position in LXe calorimeter
• Positron tracking reconstruction and DCH-TC
  matching
• Timing coincidence between TC and LXe
• Very large accidental background
• Flat distribution in TTC TLXe
• Preliminary analysis
• Expected signal small excess in the TTC TLXe
  distribution
• (after cuts on energy, track matching ),
  enhanced by angular
• correlation.

50
Time distribution TTC-TLXe 1)
Corrected for positron track length and photon
path
No signal reduction
Black no selection on qeg , 1.53 x 105
events Red cos(qeg) lt -0.7, 8.2 x 104
events Blue cos(qeg) lt -0.9, 3.5 x 104 events
signal reduction
time difference for coincident particles emitted
from target TTC-TLXe -32 ns
51
Time distribution TTC-TLXe 2)
-32 ns, like in the Boron case
540 events (fit)
537 eventi (fit) ltDTgt -32 ns and s ? 1.1 ns

• Expected number (Kuno-Okada, 1999)
• Ee gt 40 MeV Eg gt 30 MeV, Dqeg 135O BR(RD)
  7.4 x 10-8
• N BR x Tlive x Rm stop x DW x eLXe x etrackTC
  x ecuts
• BR(RD) x 1.14105 x 6106 x 0.09 x 0.4 x
  0.5 x 0.7 638 (25)

52
Status and Perspectives
53
Background and Sensitivity
Goal Goal Perspectives for 2008 Perspectives for 2008
Measured Simulated Measured 2007 Possible
Gamma energy 4.5 5.0 6.5 5.0
Gamma Timing (ns) 0.15 0.27 0.15
Gamma Position (mm) 4.5 9.0 15. 9.0
Gamma Efficiency () gt40 gt40 gt40
e Timing (ns) 0.1 0.12 0.12
e Momentum () 0.8 2.1 1.1
e Angle (mrad) 10.5 17. 17.
e Efficiency () 65 65 65
Muon decay Point (mm) 2.1 3. 3.
Muon Rate (108/s) 0.3 0.3 0.3 0.3
Running Time (weeks) 100 100 8 12
Single Event Sens (10-13) 0.5 0.5 6.7 4.5
Accidental Rate (10-13) 0.1 0.3 0.1 0.3 6.0 1.0
Accidental Events 0.2 - 0.5 0.2 - 0.5 0.9 0.4
90 CL Limit (10-13) 1.7 1.7 23 13.0
1 week 4 x 105 s Added 250 ps due to
present estimate of DRS systematics
Very pessimistic The muon rate can be
optimized to improve the limit
54
MEG time schedule

• We had an engineering run in 2007 and a second
  engineering and calibration run between April and
  August 2008
• We started the physics data taking today
• We expect first results in summer 2009
• We are confident to reach a sensitivity of
• few x 10-13 in m?eg BR in 3 years of
  acquisition time.

Revised document
LoI
Proposal
Planning
R D
Assembly
E. R.
Data Taking
E. R.
1998 1999 2000 2001 2002
2003 2004 2005 2006
2007 2008
2009
September
55

• Backup slides

56
Physics motivations 3)
In many SUSY models there is a strong correlation
between BR(m ? eg) sin2(q13) (still unknown
!). In these models BR(m ? eg) is one of the
most sensitive tool to measure sin2(q13).
Sensitivity of future long-baseline experiments
10-11
10-14
A.Masiero et al., hep-ph/0407325
57
SUSY predictions for m?eg
LFV induced by finite slepton mixing through
radiative corrections.
Experimental Bound (MEGA Coll., PRD 65 (2002)
112002)
MEG goal
J. Hisano et al., Phys. Lett. B391 (1997) 341
Small (lt10) tan b values highly disfavoured by
combined LEP data. (ALEPH, DELPHI, L3 OPAL
Collaborations, hep-ex/0107030)
SUSY SU(5) model

• In
  SO(10) BRSO(10) ? 100 BRSU(5)
• (R. Barbieri et al., Phys. Lett. B338 (1994)
  212, Nucl. Phys. B445 (1995) 215)

58
Other SSM predictions for m?eg
59
Drift Chambers 3)
Michel Spectrum
Positron rate at drift chamber position
104 Hz/cm2 Corresponding momentum cut 35
MeV/c
60
Positron track
A simulated m ? eg event
Energy release in Liquid Xenon
Hits on Timing Counter
61
Permanent muon beam intensity measurement
VERY INTENSE BEAM OF LOW ENERGY (it stops in 1
mm Mylar) IT MUST TRAVEL IN VACUUM (OR HE
ATMOSPHERE) Intensity measured by small
scanning device (non permanent) Intensity
measured by positrons reaching the timing
counter (it depends on several efficiency
factors) A direct permanent method X-rays from
thin radiator
62
Magnetic field
Longitudinal Profile (R 0) Radial Profile
Needed lt 50 G OK
63
DAQ system
11 Linux machines
80 GB System Disk RAID 1 (Mirror)
VME-Interface
Online cluster megonlxx
Front-End 1
80 GB System Disk RAID 1 (Mirror)
VME-Interface
Front-End 2
Gigabit Switch
NFS
. . .
80 GB System Disk RAID 1 (Mirror)
SC-FE
1.2 TG Data Disk RAID 5
/home/meg
Back-End
64
Offline computing
Offline cluster lcmegxx

• MEG dedicated cluster
• 16 Linux (SL4) Opteron machines
• 64 CPUs
• 90 Tb hard disk
• Fiber channel switch technology

65
DRS4 design

• DRS2 used 2007 successfully for all 3000
  channels, but remaining issues
• Temperature dependencies (solved by DRS3)
• Poor clock pulse (solved by DRS3)
• Ghost pulse problem (will be solved by DRS4)
• DRS4 design and time scale
• started Jan. 2008
• fixes ghost pulse problem, higher bandwidth,
• smaller package
• daisy-chaining of channels allows
• 1.6 GHz ? 3.2 GHz sampling speed (if needed)
• - design terminated end of March 08
• chip production April June 08
• mezzanine board prototype May 08
• mass production August 08
• DRS4 chips delivered at middle of August 08
  tests under way.

66
Z coordinate resolution

• Anodic wires
• sZ lt 1.06 cm
• Reached resolution
• quoted in proposal

• Wernier strips
• sZ lt 760 µm
• Proposal 420 µm
• Further alignment step
• Noise on signals

67
R coordinate resolution

• Extreme values
• 170 µm lt sR lt 350 µm
• Average
• sR ? 230 µm
• Not far from project resolution
• (150 µm)

68
LASER VISIBLE AND UV LIGHT PRELIMINARY TIMING OF
DETECTORS. LXe Calorimeter and Timing Counters
69
Detector calibration and performances LXe PMTs 1)

• LED
• Gain
• LED pulses with varying
• light intensity
• Nphe ?2
• Relative timing
• Led pulse detected by
• several PMTs
• Alpha sources on wires
• Quantum efficiency
• In liquid and gas
• Absorption length
• Fully tested on prototype
• NIM A 565 (2006) 589

70
Detector calibration and performances LXe PMTs 2)
Measured vs expected charge for each PMT
Alphas in GXe
71
Detector calibration and performances LXe PMTs 3)
a-sources on wires in the calorimeter
prototype. Rings in Liquid Xenon
Spots in Gaseous Xenon
NIM A 565 (2006) 589
72
Americium ?-sources on wires
Thermocompression of an Am-source strip around
a 100 ? diameter wire Developed by SORAD Italy
A lattice Great if used in the bulk of any
Liquid Detector !

• Potentialities
• PMT quantum efficiencies
• Xenon optical properties
• low-energy position and
• energy calibration
• use in Xe gas and liquid
• stability checks

73
CW beam line components
74
XEC stability
Rayleigh scattering length 80 cm
75
Data vs MC comparison 1)
Data (Blue Points) Beam _at_ 3.2 x 107 m s-1,
threshold ? 45 MeV MC (Black line) full
background simulation Pile-up not subtracted
cosmics included. Absolute rate reproduced.
Energy (MeV)
Energy (MeV)
76
Data vs MC comparison 2)
Data (Blue Points) Beam _at_ 3.2 x 107 m s-1,
threshold ? 45 MeV MC (Black line) full
background simulation Pile-up subtracted by
charge distribution cosmics rejected. Final
pile-up rejection by using waveforms (not here).
Energy (MeV)
Energy (MeV)