1. Experimental Search for the Decay
K. Mizouchi (Kyoto University)
(1) Physics Motivation
(2) Detector
(3) Event Selection Criteria
(4) Background Estimation
(5) Branching Ratio
(6) Single Photon Inefficiency
(7) Conclusions

2. : Physics Motivation [1] Helicity suppressed decay [2] Decay Form of
: left-handed (in SM)
: spin 0
[1] Helicity suppressed decay
(A) Neutrino mass : implies
(B) Neutrino type : Majorana neutrino  x2 larger branching ratio.
[2] Decay Form of
(A) Sensitive to any hypothetical weakly-interacting neutrals.
(B) Decay into different neutrino flavors :
[3] Cosmological Interests
Neutron star cooling model through pion pole mechanism :

3. Event Detection Strategy
Charged particles from K+ decay at rest
Km2
Hermetic photon detection system
Kp2
Clean Kp2 selection
p0 to invisible final states
Prior best limit :
(E787)

4. E949 Detector (1) Target : Kaon decay at rest
E949 detector side view (upper half)
E949 detector
end view (upper half)
Range Stack
Drift Chamber
Target
(1) Target : Kaon decay at rest
(2) Drift chamber : Momentum
(3) Range Stack (scintillator) : Energy / Range

5. E949 Detector (1) Barrel Veto (BV) : Pb-scintillator sandwich
E949 detector side view (upper half)
E949 detector
end view (upper half)
(1) Barrel Veto (BV) : Pb-scintillator sandwich
(2) Barrel Veto Liner (BVL) : Pb-scintillator sandwich
(3) Endcap Calorimeter : CsI crystals

6. DAQ Summary # of accumulated Kaons Accumulated K+:
Platinum target used in 2002
# of accumulated Kaons
Before data taking
After data taking
Accumulated K+:

7. Analysis Strategy (1) : Number of p0 tagged by Kp2 selection criteria
(2) : Correction for under-estimation of the num. of p0.
(3) : Number of remaining events after photon veto application
(4) : acceptance of the photon veto

8. Analysis Strategy (cont.)
(w/ bkgnd)
1/3 sample
2/3 sample
(1) Kp2 selection
p0 sample
p0 sample ( )
tuning
(2) Find the best photon veto paramters
Signal candidate (N)

9. High Purity Kp2 Identification
Dominant non-Kp2 backgrounds
(1) background rejection
(2) Single beam background rejection
(3) Two-beam background rejection

10. High Purity Kp2 Identification
Dominant non-Kp2 backgrounds
(1) background rejection
(2) Single beam background rejection
(3) Two-beam background rejection

11. Top half of side view m+ n
Km2 backgrounds

12. High Purity Kp2 Identification
Dominant non-Kp2 backgrounds
(1) background rejection
(2) Single beam background rejection
(3) Two-beam background rejection

13. Single beam backgrounds
Top half of side view p+
Cerenkov B4
Single beam backgrounds

14. High Purity Kp2 Identification
Dominant non-Kp2 backgrounds
(1) background rejection
(2) Single beam background rejection
(3) Two-beam background rejection

15. Two-beam backgrounds Beam 1 p+ K+ Beam 2 Beam wire chamber
Top half of side view
Beam 2 p+ K+
veto
veto
veto
Two-beam backgrounds

16. # tagged Kp2 and Backgrounds in the selection
Real data (2/3 sample)
Impurity : ~10-9

17. Disruption Correction Factor Cdis
Overlapping g,e+/- (from p0) may cause disruption in the p+ track reconstruction.
Estimation (Pure MC Study) :
(1) Normal Kp2 decays
(2) Kp2 decays but was forced.
Difference in the p+ recon. efficiency  correction
Disruption correction :

18. Hermetic photon veto

19. Acceptance Measurement Cacc
acceptance loss due to coincident accidentals
Correction factor : p+
accidental n n
How to measure the Cacc ?
Measure acceptance loss of Km2 decays (real data) by the photon veto, after all m+ activities are removed.

20. Maximization of Sensitivity
Photon veto :
Esum > Ethreshold
in the timing window
Effective rejection
= rejection ×Cacc
Final photon veto
Real data “1/3 sample”
[ Hermetic photon veto ]
(1) p0  gg rejection :
(2) p0  nn acceptance :
Find the best parameters providing the largest rejection and the smallest acceptance loss. (for ~20 photon detection subsystems)

21. Opening the Box
Real data “2/3 sample”
A total of 99 candidates were observed in the signal box
Kaon decay time (ns)
p+ momentum (MeV/c)

22. Result New upper limit :
2/3 sample Saturation at 3.5x106
1/3 sample
Conservative upper limit
# signal < 113 at 90%CL (Poisson) and subtracting the non-Kp2 backgrounds;
New upper limit :
A factor of 3 improvement from the previous best result.

23. p0  gg Background subtraction
Measurement of the detector single photon inefficiency
Kp2 w/ one photon missing event

24. Single Photon Inefficiency
Measure single photon inefficiency with real data.
Kp2 w/ one photon missing event
(1) : raw photon distribution
(2) : mis-detected photon distribution
(3) : Trigger prescale compensasion,

25. Analysis Strategy Kp2 w/ one photon missing event
Reconstruct (tagging) photon
Extract kinematics of the mis-detected photon.
Correction factors

26. (1) Photon Clustering Method
Reconstruct photons
and extract
their (A) positions
(B) energies and
(C) timings.

27. (2) Kinematical Fitting
[Lagrange Multiplier]
c2 minimization with constraints.
(A) Four Constraints
(B) Five inputs

28. Denominator / Numerator Map
Relaxed photon veto
(acc = 0.80)
Validity of this background subtraction method
Understanding of the detector performance

29. Correction factors CL1.1after : unwanted trigger rejection embedded in
online photon veto
(2)Cacc : over-rejection by photon veto with accidentals
(3)Csplit : self-vetoing effect by splitting tagging photon
CL1.1after = 1.14
Cacc = 0.80

30. Self-vetoing effect due to split photon
Missing-side
MC simulation
Missing photon kinematics

31. Single Photon Inefficiency
Validity of this background subtraction method
Understanding of the detector performance

32. p0  gg detection inefficiency
(1) Single photon inefficiency
(2) Photon kinematics
PSPI=
from MC simulation
(N events)

33. Daughter Table Method w/ Binominal error Convoluted inefficiency
Daughter tables produced by random number generator
300 daughter tables

34. p0  gg background subtraction
Number of candidates with relaxed photon veto
4131 events
Singal (90% C.L) : 2259
A factor of 1.8 improvement

35. Improvement by the subtraction
p0  gg rejection at various photon veto
Improvement (Before/After)
More than ~2 improvements at various photon veto
search
Estimation from photon inefficiency

36. Num of p0  gg backgrounds as a function of cos(qp+)
Single photon inefficiency
Signal discrimination capability from backgrounds

37. Background subtraction with dip angle distribution
Signal candidates

38. Likelihood distribution
Candidates : sraw = 4131
Best fit value : s = 1977
90 % C.L : s90 = 2449
A factor of 1.7 improvement
Ref. w/o subtraction :

39. Conclusions (1) search was performed with 3.02x109 Kp2 events,
where impurity of 10-9 was achieved.
(2) New upper limit of was
obtained with a total number of 99 candidates in
the signal region;
x3 improvement from the previous best limit.
Single photon inefficiency was measured with special data
p0  gg background subtraction was performed with the photon inefficiency;
(A) x1.8 improvement with simple subtraction
(B) x1.7 improvement with cos(qp+) shape discrimination

40. Thank you !

41. Background distribution

42. Background understanding and detector inefficiency
Can we understand the remaining events from a view of photon inefficiency ? ( if possible, subtract them as backgrounds.)
An Idea :
Special trigger
Kp2 but one photon is missed.
(2) Event reconstruction
Missing photon kinematics
(3) Photon inefficiency as a function of its energy and direction ?
NOTE :
Different type of critical backgrounds.　　　　　　　　
Geometrical dependence : Detector hole, dead material
Energy dependence 　 : Photonuclear interaction …

43. K+  p+ + “nothing” in E949 (1) K+  p+nn (above Kp2)
Published in PRL, (2004)
Charged track momentum from various K decay modes
(2) p0 nn (on Kp2 peak)
This report.
Need tighter photon rejection.
(3) K+  p+nn (below Kp2)
Analysis ongoing.
Require more sophisticated treatment in p+ multiple scatterings.

44. Published in PRD as rapid communication
Phys. Rev. D72, (2005)

45. Optimized Photon veto parameters

46. Performance of the clustering Method
MC sample
theta

47. p0 gg backgrounds Photon inefficiency 20<Eg[MeV]<225
Low energy g : sampling fluctuation
High energy g: photonuclear interaction ( hard to simulate reliably.)
Detector photon inefficiency
(measured with real data)
20~40MeV
40~60MeV
60~80MeV
80~100MeV
100~120MeV
120~140MeV
140~160MeV

48. Phase space correction factors
Polar angle distribution
Correction factors
Monte Carlo simulation
Real data