Mu3e Experiment The
Niklaus Berger
PI Palaver,
April 2013
• The Challenge:
Finding one in 10
16muon decays
• The Technology:
High Voltage Monolithic Active Pixel Sensors
• The Mu3e Detector:
Minimum Material, Maximum Precision
Overview
Niklaus Berger – PI Palaver, April 2013 – Slide 3
All there, works beautifully, but...
• Why three generations?
• Why the mixing patterns between generations?
• Is there more to it?
(the dark universe...)
The Standard Model of Elementary Particles
e
-ν
eμ
-ν
μτ
-ν
τu c t
d s b
γ g Z W
+/-Higgs
All there, works beautifully, but...
• Why three generations?
• Why the mixing patterns between generations?
• Is there more to it?
(the dark universe...)
The Standard Model of Elementary Particles
e
-ν
μ
-ν
τ
-u c t
d s b
γ g Z W
+/-Higgs
Leptons
Niklaus Berger – PI Palaver, April 2013 – Slide 4
Lepton Bookkeeping
Normal Muon Decay:
μ
-e
-ν
eν
μLepton Bookkeeping
Normal Muon Decay:
μ
-Muon number 1
e
-Electron number 1
ν
eElectron number -1
ν
μMuon number 1
Niklaus Berger – PI Palaver, April 2013 – Slide 4
Lepton Bookkeeping
Normal Muon Decay:
μ
-Muon number 1
e
-Electron number 1
ν
eElectron number -1
ν
μMuon number 1 Before:
Muons 1 Elektrons 0
After:
Muons 1
Electrons 0
Cooked books?
ν
eElectron number 1 ν
μMuon number 1
Niklaus Berger – PI Palaver, April 2013 – Slide 6
Cooked books?
How about charged leptons (Muons)?
μ
+Muon number -1
e
+Electron number -1
Before:
Muons -1 Electrons 0
After:
Muons 0 Electrons -1
e
-Electron number 1 e
+Electron number -1
This
(charged lepton flavour violation)
has never been seen
Niklaus Berger – PI Palaver, April 2013 – Slide 8
• Neutrinos have mass
• Leptons do change flavour
• However: Standard Model
branching ratio for μ → eee < 10-50
Charged Lepton Flavour Violation
µ + e +
W +
ν µ ν e
γ
e - e +
*
• Neutrinos have mass
• Leptons do change flavour
• However: Standard Model
branching ratio for μ → eee < 10-50
• Can be much bigger with new physics
Charged Lepton Flavour Violation
µ + χ ~ 0 e +
µ e~
~
γ
e - e +
*/Z
Niklaus Berger – PI Palaver, April 2013 – Slide 10
• We want to find or exclude μ → eee at the 10-16 level
• 4 orders of magnitude over previous experiment (SINDRUM 1988)
The Goal: 10
-161940 1960 1980 2000 2020
Year
90%–CL bound
10–14 10–12 10–10 10–8 10–6 10–4 10–2 100
μ eγ
μ 3e
μN eN
τ μγ
τ 3μ
10–16
SINDRUM SINDRUM II
MEG
MEG plan Mu3e Phase I
Mu3e Phase II
(Updated from W.J. Marciano, T. Mori and J.M. Roney, Ann.Rev.Nucl.Part.Sci. 58, 315 (2008))
Search with SINDRUM (1988)
Less than one in 10
12muon decays is to three electrons Corresponding to one gray hair in the population of
Baden-Württemberg
Niklaus Berger – PI Palaver, April 2013 – Slide 12
Our goal
Check whether more than one in 10
16muon decays is to three electrons
Corresponding to one gray hair in all humans that ever
lived
• Observe more than 10
16muon decays:
2 Billion muons per second
• Suppress backgrounds by more than 16 orders of magnitude
• Be sensitive for the signal
The Challenges
Niklaus Berger – PI Palaver, April 2013 – Slide 14
Muons from PSI
• The Paul Scherrer Institut (PSI) in Villigen, Switzerland has the world’s most
powerful DC proton beam (2.2 mA at 590 MeV)
• Pions and then muons are produced in rotating carbon targets
e +
e + e -
• μ+ → e+e-e+
• Two positrons, one electron
• From same vertex
• Same time
• Sum of 4-momenta corresponds to muon at rest
• Maximum momentum: ½ mμ = 53 MeV/c
The signal
Niklaus Berger – PI Palaver, April 2013 – Slide 16
• Combination of positrons from ordinary muon decay with electrons from:
- photon conversion, - Bhabha scattering, - Mis-reconstruction
• Need very good timing, vertex and momentum resolution
Accidental Background e
+e
+e
-• Allowed radiative decay with internal conversion:
μ
+→ e
+e
-e
+νν
• Only distinguishing feature:
Missing momentum carried by neutrinos
Internal conversion background
µ+ νμ
e+
e- e+ νe
γ*
W+
}
Emiss}
EtotBranching Ratio
m - E (MeV)
0 1 2 3 4 5 6
10-12
10-16 10-18 10-13
10-17 10-15 10-14
10-19
• Need excellent μ3e
momentum resolution
(R. M. Djilkibaev, R. V. Konoplich,
Niklaus Berger – PI Palaver, April 2013 – Slide 18
• 1 T magnetic field
• Resolution dominated by multiple scattering
• Momentum resolution to first order:
Σ
P/P ~ θ
MS/Ω
• Precision requires large lever arm (large bending angle Ω) and low multiple scattering θMS
Momentum measurement
Ω MS
θ
MSB
High voltage monolithic active pixel sensors
• Implement logic directly in N-well in the pixel - smart diode array
• Use a high voltage commercial process (automotive industry)
• Small active region, fast charge collection via drift
• Can be thinned down to < 50 μm
• Invented by Ivan Peric at ZITI Mannheim
(I.Peric, P. Fischer et al., NIM A 582 (2007) 876 )
Fast and thin sensors: HV-MAPS
Niklaus Berger – PI Palaver, April 2013 – Slide 20
HV-MAPS chips: AMS 180 nm HV-CMOS
• MUPIX2:
Characterization during 2012
Single pixel Time-Over-Threshold Binary pixel matrix
• MUPIX3:
Column logic with address generation
Extensive test beam campaign 2013
The MUPIX chips
MUPIX2
36 x 42 pixels
30 x 39 μm pixel size 1.8 mm2 active area MUPIX3
40 x 32 pixels
80 x 92 μm pixel size 9.4 mm2 active area For Mu3e:
256 x 256 pixels
80 x 80 μm pixel size 4 cm2 area, 95% active
• Measurements with 55Fe source
• Good energy measurement
• Very good signal to noise
Details in theses:
A.K. Perrevoort: Characterization of HV-MAPS for Mu3e (Master thesis, 2012)
H. Augustin: Charakterisierung von HV-MAPS (Bachelor thesis, 2012)
MUPIX 2 Results
ToT [µs]
0 1 2 3 4 5
10-4
10-3
10-2
10-1
1 55Fe peak
SNR
5 10 15 20 25 30 35 40
Signal to Noise
Niklaus Berger – PI Palaver, April 2013 – Slide 22
MUPIX3 Set-up
Niklaus Berger – PI Palaver, April 2013 – Slide 24
Introduction
Y
• X
• 50 μm silicon
• 25 μm Kapton™ flexprint with aluminium traces
• 25 μm Kapton™ frame as support
• Less than 1‰ of a radiation length per layer
Mechanics
Niklaus Berger – PI Palaver, April 2013 – Slide 26
Niklaus Berger – PI Palaver, April 2013 – Slide 30
• Add no material:
Cool with gaseous Helium
• ~ 150 mW/cm2 - total 2 kW
• Simulations: Need ~ 1 m/s flow
• First measurements: Need several m/s
• Full scale prototype on the way
Cooling
Details in thesis:
M. Zimmermann: Cooling with Gaseous Helium for the Mu3e Experiment
(Bachelor thesis, 2012)
available from www.psi.ch/mu3e
Niklaus Berger – PI Palaver, April 2013 – Slide 32
• 1 T magnetic field
• Resolution dominated by multiple scattering
• Momentum resolution to first order:
Σ
P/P ~ θ
MS/Ω
• Precision requires large lever arm (large bending angle Ω) and low multiple scattering θMS
Momentum measurement
Ω MS
θ
MSB
Precision vs. Acceptance
50 MeV/c 25 MeV/c 12 MeV/c B→
Niklaus Berger – PI Palaver, April 2013 – Slide 33
Precision vs. Acceptance
50 MeV/c 25 MeV/c 12 MeV/c B→
Precision vs. Acceptance
50 MeV/c 25 MeV/c 12 MeV/c B→
Niklaus Berger – PI Palaver, April 2013 – Slide 33
Precision vs. Acceptance
50 MeV/c 25 MeV/c 12 MeV/c B→
Precision vs. Acceptance
50 MeV/c 25 MeV/c 12 MeV/c B→
Ω ~ π MS
θMS
B
Niklaus Berger – PI Palaver, April 2013 – Slide 34
Detector Design
Target μ Beam
Detector Design
Target Inner pixel layers
μ Beam
Niklaus Berger – PI Palaver, April 2013 – Slide 34
Detector Design
Target Inner pixel layers
Outer pixel layers μ Beam
Detector Design
Target Inner pixel layers
Scintillating fibres
Outer pixel layers μ Beam
Niklaus Berger – PI Palaver, April 2013 – Slide 34
Detector Design
Target Inner pixel layers
Scintillating fibres
Outer pixel layers Recurl pixel layers
μ Beam
Detector Design
Target Inner pixel layers
Scintillating fibres
Outer pixel layers Recurl pixel layers
Scintillator tiles
μ Beam
Niklaus Berger – PI Palaver, April 2013 – Slide 35
• 280 Million pixels (+ fibres and tiles)
• No trigger
• ~ 1 Tbit/s
• FPGA-based switching network
• O(50) PCs with GPUs
Data Acquisition
Pixel Sensors
up to 108 800 Mbit/s links
FPGA FPGA FPGA
...
...
RO Boards 1 3 Gbit/s
link each
GPU
PC GPU
PC
GPU ... PC
12 10 Gbit/s ...
links per RO Board 4 Inputs each
Data Collection
Server
Mass Storage Gbit Ethernet
Pixel DAQ
Online software filter farm
• Continuous front-end readout (no trigger)
• ~ 1 Tbit/s
• PCs with FPGAs and Graphics Processing Units (GPUs)
• Online track and event reconstruction
• 109 3D track fits/s achieved
• Data reduction by factor ~1000
• Data to tape < 100 Mbyte/s
Online filter farm
Niklaus Berger – PI Palaver, April 2013 – Slide 37
Simulated Performance
2] Reconstructed Mass [MeV/c
101 102 103 104 105 106
Events per muon decay and 0.1 MeV
10-20
10-19
10-18
10-17
10-16
10-15
10-14
10-13
10-12
10-11
10-10 µ→ eeeνν generated
simulated ν
ν
→ eee µ
Signal BF 10-12
Signal BF 10-13
Signal BF 10-14
Signal BF 10-15
Signal BF 10-16
Signal BF 10-17
Sensitivity
Target Inner pixel layers
Outer pixel layers μ Beam
Niklaus Berger – PI Palaver, April 2013 – Slide 38
Sensitivity
Phase IB: 2016+
Target Inner pixel layers
Scintillating fibres
Outer pixel layers Recurl pixel layers
Scintillator tiles
μ Beam
Sensitivity
Phase II: 2017+
Target Inner pixel layers
Scintillating fibres
Outer pixel layers Recurl pixel layers
Scintillator tiles
μ Beam
Niklaus Berger – PI Palaver, April 2013 – Slide 39
A collaboration has formed and submitted a research proposal to PSI
• University of Geneva
• University of Heidelberg: PI and KIP
• Paul Scherrer Institut (PSI)
• University of Zurich
• ETH Zurich
Also in contact with other interested groups
Collaboration
• Mu3e aims for μ → eee at the 10-16 level
• First large scale use of HV-MAPS
• Build detector layers thinner than a hair
• Reconstruct 2 billion tracks/s in 1 Tbit/s on ~50 GPUs
• Start taking first data in 2015
Conclusion
1940 1960 1980 2000 2020
Year
90%–CL bound
10–14 10–12 10–10 10–8 10–6 10–4 10–2 100
μ eγ
μ 3e
μN eN
τ μγ
τ 3μ
10–16
SINDRUM SINDRUM II MEG
MEG plan Mu3e Phase I
Mu3e Phase II
Niklaus Berger – PI Palaver, April 2013 – Slide 41
Backup Material
• One loop term and one contact term
• Ratio κ between them
• Common mass scale Λ
• Allows for sensitivity comparisons between μ → eee and μ → eγ
• In case of dominating dipole couplings (κ = 0):
B(μ → eee) = 0.006 (essentially αem) B(μ → eγ)
Comparison with μ → eγ
L
LFV= A m
μ Rμ
Rσ
μνe
LF
μν+ (μ
Lγ
μe
L) (e
Lγ
μe
L) (κ+1)Λ
2κ (κ+1)Λ
2µ+ χ~0 e+
µ e~
~
γ
e- e+
*/Z µ+
e+ e-
e+ Z’
Niklaus Berger – PI Palaver, April 2013 – Slide 43
• Z-Penguins can be important
• Lots of ongoing theory activity
Comparison with μ → eγ
L
LFV= A m
μ Rμ
Rσ
μνe
LF
μν+ (μ
Lγ
μe
L) (e
Lγ
μe
L) (κ+1)Λ
2κ (κ+1)Λ
2µ+ χ~0 e+
µ e~
~
γ
e- e+
*/Z
Radiation Hardness
• Requirements not as strict as at LHC
• Irradiation at PS
• After 380 MRad (8×1015 neq/cm2)
• Chip still working
Niklaus Berger – PI Palaver, April 2013 – Slide 45
MUPIX electronics
• Inductively heated sample
• Helium flow cooling
More on Cooling
Niklaus Berger – PI Palaver, April 2013 – Slide 47
• 3D multiple scattering track fit
• Simulation results:
280 keV single track momentum 520 keV total mass resolution
Simulated Performance
Hits fitted per track
0 1 2 3 4 5 6 7 8 9
103
104
Reconstructed Momentum [MeV/c]
0 10 20 30 40 50 60
1 10 102
103
Rec. Momentum - Gen. Momentum [MeV/c]
-3 -2 -1 0 1 2 3
1 10 102
103
104 RMS: 0.28 MeV/c
Reconstructed track polar angle
0 0.5 1 1.5 2 2.5 3
1 10 102
103
2] Reconstructed Mass [MeV/c
1020 103 104 105 106 107 108 109 110 200
400 600 800 1000 1200 1400 1600
RMS: 0.52 MeV/c2
: 0.31 MeV/c2
s1
: 0.71 MeV/c2
s2
: 0.37 MeV/c2
sav
MUPIX 2 results
• Test beam at CERN SPS (170 GeV/c pions)
• Timepix telescope
• 2 hours data taking
• Mostly single pixel clusters
• Resolution as expected (pixel size/√12)
• March test beam data (DESY, electrons) currently being analysed
• Next beam week in June Resolution for 30 × 40 μm pixels
Niklaus Berger – PI Palaver, April 2013 – Slide 49
• Measurements with LED pulses
• High-Voltage important for fast signal
• Amplification above ~70 V
Details in theses:
A.K. Perrevoort: Characterization of HV-MAPS for Mu3e (Master thesis, 2012)
H. Augustin: Charakterisierung von HV-MAPS (Bachelor thesis, 2012)
available from www.psi.ch/mu3e
MUPIX 2 Results
HV [V]
0 20 40 60 80
Latency [ns]
300 350 400 450 500 550 600 650 700
ToT [µs]
4 5 6 7 8 9 10
11 12
HV [V]
0 20 40 60 80