1. Status and prospects of Lepton Flavor Violation searches
Marco Grassi
INFN Sezione di Pisa
DISCRETE 2010
Rome, 6-11 December

2. Outline Physics motivation for cLFV searches The muons The taus
m  eg (MEG)
m  e e e (SINDRUM)
mA  eA (Mu2e, COMET, PRISM/PRIME)
The taus
t  mg, eg (BABAR, BELLE)
t  lll (BABAR, BELLE)
Conclusions
Rome - 10 December 2010

3. Physics motivations
The Standard Model is believed to be a low-energy approximation a of a more fundamental theory
DM, number of parameters, unexplained symmetries, flavor structure, …
All models beyond the SM contains new particles that could
either be directly discovered
(high energy frontier)
or seen through their contributions in loops
(high precision frontier)
Rome - 10 December 2010

4. Physics motivation
1) The cLFV decay is undetectably small in the extended Standard Model, which takes into account the neutrino masses and mixings
Example m  e g decay
BR ~ 10-54
2) New Physics scenarios usually enhance the rate of cLFV decay by orders of magnitude, through loops of new particles
cLFV decays are clean, no SM contaminated, evidence of new physics
The expected rates are close to the experimental upper bound and within the capabilities of present and near future experiments
Rome - 10 December 2010

5. Physics connections →γ EDM μ→eγ g - 2 μZ→eZ
G.Isidori et al. Phys. Rev. D75 (2007)
Example
In a specific model MSSM /GUT extension with large tan
→γ
μ→eγ
EDM
g - 2
μZ→eZ
This is still a rare process, as proton decay or dark matter search or neutrino oscillation.
Also here one searches for the violation of one of the SM accidental symmetries
If the E821 anomaly is real (3.2 )
It should be
Rome - 10 December 2010

6. Physics connections g-2 , EDM
Within a given NP scenario different processes are linked together with a specific pattern
We could even have signal on one channel and nothing on others μ→eγ , μZ→eZ , →eγ , →γ , →lll , →h
g-2 , EDM
or in other words
The structure and the couplings of NP beyond the SM could be constrained by multiple precision measurements
unfortunately
No clean and direct evidence of New Physics has been established so far
Rome - 10 December 2010

7. The muons Muons are excellent probes for high precision measurements
intense continuous and pulsed beams can be obtained;
long lived;
simple final states at low energy : small detectors;
Rome - 10 December 2010

8. Historical perspective
Hinks & Pontecorvo ? !
Crystal Box
MEGA
Each improvement linked to the technology either in the beam or in the detector
Always a trade-off between various sub-detectors to achieve the best “sensitivity”

9. μ→eγ : MEG experiment Picture
~60 physicists, 12 Institutions, 5 Countries
Italy, Japan, Russia, Switzerland, USA

10. Easy signal selection with μ+ decay at rest
Experimental method
Detector outline
• Stopped beam of >107 μ /sec in a 175 μm target
• g detection
Liquid Xenon calorimeter based on the scintillation light
- fast: 4 / 22 / 45 ns
- high LY: ~ 0.8 * NaI
- short X0: 2.77 cm
• e+ detection
magnetic spectrometer composed by solenoidal magnet and drift chambers for momentum
scintillation counters for timing
Easy signal selection with μ+ decay at rest
e+ + g
Ee = Eg = 52.8 MeV
qeg = 180°
Rome - 10 December 2010

11. Signal and Background qeg = 180° FWHM m+ e+ g n m+ e+ g n m+ e+ g
Signal Prompt Accidental m+ e+ g n m+ e+ g n m+ e+ g
qeg = 180°
Ee = Eg = 52.8 MeV
Te = Tg
at 3x107 m-stop/s
The accidental background is dominant and it is determined by the experimental resolutions
FWHM
Exp./Lab
Year
ΔEe/Ee
(%)
ΔEγ /Eγ
Δteγ (ns)
Δθeγ
(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 10
6.7
2 x 105
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
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 x 1
0.15 19
3 x 107
23 x 10-13

12. Electrostatic separator
μ beam
Electrostatic separator
COBRA
Transport solenoid
Target
Quadrupoles
Intensity’ (μ-stop/s)
Low 2.5 x 106
Normal 3.2 x 107
Max 2 x 108
characteristics
P = 27.7 MeV/c
ΔP = 0.3 MeV/c
σX = 9.5 mm
σY = 10. mm
MEG target
Material CH2
q = 22.5o
thick = 205 mm
Size = 15x25 cmxcm
Rome - 10 December 2010

13. The positron spectrometer
16 low-mass Drift Chambers in a Helium atmosphere with a graded magnetic field :
very low total material budget (< X0);
fast expulsion of tracks from the spectrometer even at large polar angles.
Graded Field
Design Resolutions
Momentum: 200 keV/c
Direction: 4.5 mrad
Rome - 10 December 2010

14. The timing counter
2 detectors (upstream & downstream) for precise positron timing and trigger;
15 plastic scintillating bars per detector read by PMTs:
timing
phi position
trigger
1 layer of scintillating fibers per detector, read by APDs:
zposition
Design Resolution
Timing : 45 ps

15. The LXe calorimeter The largest LXe calorimeter in the world:
800 liters;
Fast response:
t = 4ns / 22ns / 45ns;
Good light yield:
~ 75% ofNaI(Tl);
Light collected by 846 PMTs.
Design Resolutions
Energy: 2.4 MeV (FWHM)
Conversion Point: 4 mm
Time: 65 ps
Hamamatsu R9288
Rome - 10 December 2010

16. MEG key elements MEG is a precision experiment and demands for
unprecedented detector resolutions
accurate determination of detector response with time
The detector is complemented with an impressive set of calibration tools
However the limiting factor is not the beam intensity but the resolutions
there is space for improvements
the resolutions are not yet the design one, but are constantly improving
Rome - 10 December 2010

17. Calibration tools m radiative decay Laser LED e m n p0 gg alpha
PMT Gain
Higher V with light att.
Can be repeated frequently
alpha
PMT QE & Att. L
Cold GXe
LXe
Laser
(rough) relative timing calib.
< 2~3 nsec
Nickel g Generator
9 MeV Nickel γ-line
NaI
Polyethylene
0.25 cm Nickel plate
3 cm
20 cm
quelle on
off
Illuminate Xe from the back
Source (Cf) transferred by comp air  on/off
Proton Acc
Li(p,)Be
LiF target at COBRA center
17.6MeV g
~daily calib.
+ B target  (4.4, 11.7, 16.1 MeV) lines – Energy + Timing K Bi Tl F
Li(p, 0) at 17.6 MeV
Li(p, 1) at 14.6 MeV
m radiative decay
p0 gg
p- + p  p0 + n
p0  gg (55MeV, 83MeV)
p- + p  g + n (129MeV)
10 days to scan all volume precisely
(faster scan possible with less points)
LH2 target g e+ e- e n m
Lower beam intensity < 107
Is necessary to reduce pile-ups
Better st, makes it possible to take data with higher beam intensity
A few days ~ 1 week to get enough statistics
MEG Detector
Standard
Calibrations
Rome - 10 December 2010

18. Resolutions πº Eg Ee+ Teg µ→eγ runs RMD Michel g bck
Double gaussian momentum resolution
s(p) = 390 KeV (0.74%) (core)
Positron angle resolution measures using multi-loop tracks
s(f) = 7.1 mrad (core)
s(q) = 11.2 mrad
Overall angular resolution combining XEC+DCH+target
s(f) = 12.7 mrad (core)
s(q) = 14.7 mrad πº
µ→eγ runs
RMD
40 MeV < Eg < 48 MeV
Resolution corrected for a small energy-dependence
s(t) = (142  15) ps
Stability along the run
< 15 ps
Michel
g bck
Average upper tail for deep conversions
sR = (2.1 0.15) %
Systematic uncertainty on energy scale < 0.6 %
Rome - 10 December 2010

19. Uniformity Calorimeter uniformity Timing uniformity and stability
55 MeV g (from p0 decay)
Eg resolution vs conversion point
Close to design value
Timing uniformity and stability
Events from raditive m decay (RD)
Requires LXe+TC+track
Close to design value
(if rescaled for the different Eg energy)
Rome - 10 December 2010

20. Analysis principle 2009 data sample
A µ→eγ event is described by 5 kinematical variables
We decided to adopt a likelihood analysis strategy on a wide blind box
The blinding variables are
Hidden until analysis is fixed
Use of the sidebands
our main background comes from accidental coincidences
RMD can be studied in the low Eγ sideband
Left Sideband
Right Sideband
Blind Box
E Sideband
Rome - 10 December 2010

21. Analysis principle
Likelihood function is built in terms of Signal, radiative Michel decay RMD and accidental background BG number of events and their probability density function PDFs
Un-binned likelihood fit of over the entire blind box
Three different analysis for cross check
PDF
Approach (freq. or Bayes)
Rome - 10 December 2010

22. Probability Density Functions
Signal
- calibration data (p0, Michel edge, CW, XEC single events ...)
for photon/positron energy and relative angle
- RD data on sideband for timing (corrected for energy dependence)
RMD
- 3-D theoretical distribution folded with detector response to
take into account kinematical constraints
- direct measurement for timing BG
- Everything measured on sidebands
Important: PDF for BG are measured !
Rome - 10 December 2010

23. Systematic uncertainties
The effect of systematics is taken into account in the calculation of the confidence region by fluctuating the pdfs according to the uncertainty values
overall effect of systematics: ΔNsig ~ 1
Uncertainty
Normalization 8%
Pe+ εγ εTRG
Eγ scale
0.4%
Light yield stability, gain shift
Eγ resolution 7%
Ee scale
50 keV
from Michel edge
Ee resolution
15%
teγ center
15 ps
teγ resolution
10%
RMD peak
Angle
7.5 mrad
Tracking + LXe position
Angular resolution
Ee–φe correlation
50%
MC evaluation

24. Normalization B.R. = Nsig x (1.01 ± 0.08) × 10-12
The normalization factor is obtained from the number of observed m decay positrons taken simultaneously (prescaled) with the m→eg trigger
Cancel at first order
Absolute e+ efficiency and DCH instability
Instantaneous beam rate variations
O(1)
~18k
107
theory
resolution
acceptance
B.R. = Nsig x (1.01 ± 0.08) × 10-12

25. Control samples MEGPreliminary MEGPreliminary MEGPreliminary
Likelihood analysis performed on the Left and Right teγ sidebands gives
Nsig and NRMD = 0
BR < (4 ÷ 6) x 10-12
Experiment sensitivity is computed as the average 90% upper limit on toy experiments
no signal in MC generation
6.1 x 10-12
MEGPreliminary
MEGPreliminary
MEGPreliminary
Blue lines are 1(39.3 % included inside the region w.r.t. analysis window), 1.64(74.2%) and 2(86.5%) sigma regions.
For each plot, cut on other variables for roughly 90% window is applied.
MEGPreliminary

26. Likelihood analysis MEGPreliminary Signal RADIATIVE DECAY ACCIDENTAL
TOTAL
(dashed: 90% U.L)
Best fit:
Nsig = 3.0, NRMD = (expected , NRMD = 32 ± 2)
Nsig < 90% C.L. (Feldman-Cousins)

27. Result From the analysis of 2009 data the MEG Upper Limit is
BR(m→eg ) @90% CL < 1.5 x 10-11
Notes
The three analyses are consistent
The Nsig = 0 hypothesis has a probability of 20-60%
The MEG published result on 2008 data was (Nucl.Phys. B834(2010)1-12)
BR(m→eg CL < 2.3 x 10-11
The best Upper Limit from MEGA is (Phys.Rev. D65 (2002) )
BR(m→eg CL < 1.2 x 10-11
MEGPreliminary
Rome - 10 December 2010

28. Blind box content MEGPreliminary MEGPreliminary MEGPreliminary

29. Blind box content MEGPreliminary MEGPreliminary MEGPreliminary

30. MEG Event display Events in the signal region were checked carefully
An event in the signal region

31. MEG 2010
2010 run prematurely ended by a quench in the Beam Transport Solenoid
2009 statistics only doubled
Found burned feed-through, if nothing else, the 2011 run shall start after the PSI winter shutdown
The collaboration is considering the publication of combined data in this spring
Rome - 10 December 2010

32. MEG perspectives - Seen detector improvements
Not only statistics
MC description of the LXe to implement different
Alignment
more dedicated XEC–DCH coincidence
Matrix of lead dices to improve Lxe position reconstruction
Mott scattering target to calibrate the tracker with monochromatic variable energy positron
tracking improvement
better treatment of rapidly varying magnetic field
cross talk between adjacent Vernier pads
shadow effect from the anode wires on the Vernier pads
- Seen detector improvements
- Two full year run 2011 / 12
- Working hard to push the sensitivity down to 10-13
- New results soon ! ! !
Analysis
Inclusion of information from the sidebands in the likelihood
variations of proton current

33. +e+e+e- background signal   eee accidental   e n n e+e- e+e-
correlated
  e e e n n
e+ + n e- e+
e+ + n
Coplanarity
Vertexing
Ee = m
Te+ = Te+ = Te- e+ n e+ n e- e- +
Rome - 10 December 2010

34. +e+e+e- : SINDRUM I
Present limit from SINDRUM : B(m3e ) < 1x10-12
U.Bellgardt et al. Nucl.Phys. B299(1988)1
Generalities
This channel has only charged particles in the final state
The experiment needs only a tracking system but
Sustain the entire Michel decay rate
Down to low momentum
Large solid angle coverage
SINDRUM I parameters
beam intensity 6x106 m+/s
m+ momentum 25 MeV/c
magnetic field 0.33T
acceptance 24%
momentum res. 10% FWHM
vertex res.  2 mm2 FWHM
timing res.  ns
target length 220 mm
target density 11 mg/cm2
Is there any future?
Expression of interest from Heidelberg for a search aiming at based on ultra thin Si tracker with MAPS readout
Comment: background scales  quadratically with beam
SINDRUM sensitivity  sensitivity 10-15
beam 6x beam 6x108
background  background 10-10

35. RPC (radiative pion capture) MIO (muon decay in orbit)
-e- conversion
Background
Signal
coherent LFV decay
 (A,Z)  e (A,Z)
RPC (radiative pion capture)
 (A,Z)   (A,Z-1)
e- - g
(A,Z) - g
(A,Z)
MIO (muon decay in orbit)
 (A,Z)  e n n (A,Z)
Ee = mm -EB -ER
e- - n
(A,Z)
Rome - 10 December 2010

36. -e- physics Effects from almost all NP scenarios
These do not contribute to eg
Rome - 10 December 2010

37. -e- : SINDRUM II result
SINDRUM II parameters
beam intensity 3x107 m-/s
m- momentum 53 MeV/c
magnetic field 0.33T
acceptance 7%
momentum res. 2% FWHM
S.E.S 3.3x10-13
counts/channel
momentum (MeV/c)
Present limit from SINDRUM II : B(me:Au ) < 7x10-13
E. Bertl et al. Eur.phys.J. C47 (206) 337
Rome - 10 December 2010

38. -e- step forward
1 particle in the final state: no accidental Background
chance to explore very low BR
Event selection based on e- detection only
m lifetime ~.86 ms on Al ms on Ti
e energy ~105 MeV on Al MeV on Ti
Key element: beam !
new and expensive beam line and m channels
Ultra high intensity m beam ~ 1011 m/s
Pulsed P beam
Short beam-on (dt ~ 10ns)
Long beam-off ( Dt ~ 1 ms )
Huge P extinction factor (10-9 or FFAG)
Low momentum m (~ 68 MeV/c)
Narrow momentum spread (<2 % with FFAG)
Rome - 10 December 2010

39. Full discussion by F. Cervelli this afternoon
Mu2e at Fermilab
-e- : Al
Sign selection and
antiprotons rejection
Graded magnetic field
to select electrons with
P>90 MeV/c and recover
backwards going electrons
Expected tracker resolution
900 keV/c MeV
8 GeV, 100 ns
width bunches
Full discussion by F. Cervelli this afternoon
Rome - 10 December 2010

40. Mu2e summary Goal: explore region down to 10-16 in 5 years
expected 40 evts for Rme =10-15
Critical element: proton beam
Background – free measurement: requires p-extinction of 10-9
Detector magnet; straight tracker far from the bulk of m decay
Detector: again resolutions on momentum and energy 900 KeV/c FWHM @100 MeV/c
Status: got CD-0 approval
Time scale: several years
D. Glezinsky, NuFact 09
Rome - 10 December 2010

41. COMET at J-PARC
-e- : Al
Y. Kuno, Nufact 10
Rome - 10 December 2010

42. COMET summary Similar to Mu2e
Goal: explore region down to 6x10-17 in 2 years
expected 40 evts for Rme =10-15
Critical element: beam
Background – free measurement: requires p-extinction of 10-9
Detector magnet; curved to protect the detector from the bulk of m decay
Detector: high resolutions on momentum and energy
400 KeV/c FWHM @100 MeV/c
Status: Stage 1 approval in 2009
Timescale: (very optimistic)
construction 2012
running 2017
Rome - 10 December 2010

43. PRISM/PRIME : 2nd stage A. Sato, NP08 PRIsm Muon
Electron conversion experiment
Phase
Rotated
Intense
Slow
Muon source
A. Sato, NP08
Rome - 10 December 2010

44. PRISM: motivation Phase rotation
- Energy spread of m reduced to 3 % FWHM
by means of RF in the FFAG ring
- Well defined m momentum 68 MeV/c
- Large aperture, high intensity  1011÷12 m/s
- Long m flight length p suppression
- Kickers in the FFAG : p extinction not crucial
Phase rotation
Small energy spread allows very thin targets
Momentum resolution  350 keV (FWHM) could be possible
Background reduced at 0.1 events
Experiment sensitive down to BR  10-18
Experimental demonstration of phase rotation in FFAG in progress at Osaka
Far in the future …
Rome - 10 December 2010

45. The taus Taus are excellent probes for high precision measurement
with differences with respect to muons
Massive initial state, many channels
Massive initial state, effects enhanced by 35 orders of mag
Intensity could be an issue t/year vs 10+8 m/s
Short lived
Complex detectors
Rome - 10 December 2010

46. LFV in t decays
In NP models limits are linked: μ→eγ at 10-12 implies →γ at >10-9
impressively close to experimental bounds
Rome - 10 December 2010

47. Detection strategy b - factories are also excellent t - factories
Search strategy
Identify tt couples with one of the possible
t – tags on one side
Search LFV on the other side
Background
Generally larger for mg and eg channels, smaller for the other modes
Tails of the tt distributions, mm, uds, Bhabha, cc
Rome - 10 December 2010

48. t  mg/eg BABAR Data sample 468 fb-1 @(4S) 90 fb-1 @(nS)
(963  7) x 106 t decays
Total sample collected by BABAR
BABAR Collaboration hep-ex/ v2
BABAR Collaboration PRL 104 (2010)
Full discussion by A. Cervelli this afternoon
See also E. Guido on Tuesday
Rome - 10 December 2010

49. t  mg/eg BABAR
Recipe
Look for one single m (or e) plus one  in the signal side
Reconstruct (it should be = mt)
Reconstruct (it should be zero)
Blind analysis to evaluate the background
Open a 2s region and count eg
Green ellipse (2 s’s):
events observed 2 (m) 0 (e) eff. (6.1  0.5) % (m)
events expected (m) 1.6 (e) (3.9  0.3) % (e)
Upper 90% C.L.:
BR( t  mg ) < 4.4 x BR( t  eg ) < 3.3 x 10-8 mg
Rome - 10 December 2010

50. See talk of K. Hayasaka on Tuesday
t  mg/eg BELLE
Data sample
Integrated luminosity 535 fb-1
 4.77 x 108 t pairs
Similar search strategy:
look for two opposite charge tracks, accompanied in “signal-side” by one or more photons
reduce other background by cuts on missing quantities;
examine surviving events in the plane
(DE, Mmg);
BELLE Collaboration (K. Abe et al.), PL B666 (2008) 16-22
See talk of K. Hayasaka on Tuesday
Rome - 10 December 2010

51. t  mg/eg BELLE 3 s ellipse mg 2 s ellipse eg Data Signal MC
Maximum likelihood fit to signal & bck. UL:
BR( tmg ) < 4.5 x 10-8
BR( teg ) < 1.2 x 10-7

52. t  lll BABAR Standard 1-prong tag
BABAR Coll. PhysRevD81, (2010)
Standard 1-prong tag
Three tracks of opposite tot charge on the signal
Data sample 468 fb-1
Search based on invariant mass and DE
Very low background on these channels
no excess observed
U.L. Range: (1.8 ÷ 3.3) x (90% C.L.)
Rome - 10 December 2010

53. t  lll BELLE Data sample 782 fb-1 Very low background as for BABAR.
BELLE Coll. PL B687 (2010)
Data sample 782 fb-1
Very low background as for BABAR.
U.L. Range: (1.5 ÷ 2.7) x (90% C.L.)
t3l search as no irreducible background
(no photons no problems with initial state radiation)
Rome - 10 December 2010

54. Future: super B factories
Projects of Super-B factories in
Japan (KEKB upgrade) approved start years
Italy (Frascati) approval requested
Expected luminosities:
1035 cm-2 s-1 (SuperKEKB), 1036 cm-2 s-1 (SuperB)
Both projects would reach an integrated luminosity L = 75 ab-1
SuperB could exploit the beam polarization to reduce the background on t  lg
Detector upgrades are foreseen
Expected sensitivities
- BR(t  lg) < 2 x 10-9
- BR(t  3l) < 2 x 10-10
- BR(t  lh or lhh) < (2 ÷ 6) x 10-10
Specific talks this afternoon
A. Cervelli SuperB
S. Korpar SuperKEKB
Rome - 10 December 2010

55. Conclusions
General arguments suggest that New Physics should be close to the experimental reach, both in particle searches and effects through loops
The structure and the couplings of NP could be constrained by multiple precision measurements
MEG is close to publish new data and it’s ready to explore an interesting region down to a few 10-13
Challenging μ→e conversion searches are expected to start in 2016
Data on t LFV decays are not yet fully analyzed, and new powerful machines are under construction or planned
Rome - 10 December 2010

56. University of Virginia
Graphically
Super-B factories
MEG
Mu2e, COMET
E. Craig Dukes
University of Virginia

57. Bibliography Barbieri et al,, Nucl. Phys B445 (1995) 225
Hisano et al., Phys. Lett. B391 (1997) 341
Masiero et al., Nucl. Phys. B649 (2003) 189
Bettoni et al., Phys. Rep. 434 (2006) 74
Calibbi et al., Phys. Rev. D74 (2006)
Isidori et al., Phys. Rev. D75 (2007)
Raidal et al., Eur. Phys. J. C57 (2008) 13
Rome - 10 December 2010

58. Spare

59. The Paul Sherrer Institute (PSI)
The most powerful continuous
machine (proton cyclotron) in the
world;
Proton energy 590 MeV;
Power 1.2 MW;
Nominal operational current 2.2 mA.
Rome - 10 December 2010

60. Expected performance
Rome - 10 December 2010

61. -e- : sensitivity R.Kitano et al Phys.Rev.D66(2002)096002
Rome - 10 December 2010