Background rejection in P326 (NA48/3)

1
Background rejection in P326 (NA48/3)

• Giuseppe Ruggiero
• CERN
• K-Rare 2005 Workshop
• Frascati 26 / 05 / 2005

2
Overview

• Characterization of the background
• Kinematics and background rejection capability
• Muon rejection and Muon ID
• Requirements and results from simulations
• Photon rejection
• Requirements and results from simulations
• Electron ID
• Results from NA48 studies on data
• Results about background rejection
• Some thoughts about Charged Veto
• Conclusions

3
Background rejection

• Main task of a pnn experiment
• All the K decay modes are potentially dangerous
• Goal of P326 S/B 10 10-12
  rejection
• 2-Steps
• Kinematic rejection
• Veto and Particle ID
• g, m, charged particles
• m p - e separation

As better as possible resolution in charged
particle reconstruction
High hermeticity
4
Background kinematically constrained
Pion track hyp.
Decay BR
Km2 0.634
pp0 0.211
ppp- (p0p0) 0.070
92 of total background
Allows us to define the signal region
5
Background not kinematically constrained
Decay BR
Ke3 0.049
Km3 0.033
Km2g 0.006
pp0g 0.001
Ke4 4 x 10-5
Km4 1 x 10-5
Pion track hyp.
8 of total background
Spoils the signal region
6
Kinematics Gigatracker Double Spectrometer

• Tracking systems operating in vacuum
• Gigatracker pixels
• Spectrometer Straw tubes
• Gigatracker
• 4x10-3 X0 per station
• PK measurement
• qK measurement
• Spectrometer
• 5x10-3 X0 per chamber
• 2 Ptrack measurements
• qtrack measurement
• Resolution limited by MS

7
Kinematic reconstruction
Total
qpK
qp
Ptrack
PK
qK
Two independent measurements of the downstream
track momentum
m2miss resolution 1.110-3 GeV2/c4 Main
contribution from QpK measurement
8
Kinematic rejection
CUTS
Against Km2, pp0 and ppp- Gaussian background
lt 10-6
Against Km2 RICH operational reasons
Simulation and results
pp0

• Simulation of the tracking systems
• GEANT - based
• Accidental PileUp in Gigatracker
• (150 ps resolution per station)
• Kinematic rejection inefficiency
• (Limited by non gaussian tails from MS)
• Km2 5 x 10-6 (Region I mainly)
• pp0 2 x 10-4
• Reconstruction
• Room for 3 gain in rejection power
• (loss in signal acceptance)

Cuts on Ptrack
9
Muon rejection Physics
EM showers (from ICARUS)
Hadronic shower (from ICARUS)
10
Muon rejection MAMUD

• Detector Sampling Calorimeter (m rejection)
  Magnet (beam deflection)
• Goal m rejection inefficiency lt 10-5
• Sensitivity to minimum ionizing particles (MIP)
• Distinguish hadronic and electromagnetic showers
  (longitudinal segmentation)
• Bending power 5 Tm ? 75 GeV/c beam
  deflected by 18 mrad

11
Simulation of muon rejection results

• Complete GEANT simulation of MAMUD
• Rejection
• MAMUD LKr calorimeter
• Rejected events
• MIP deposition in last section only
• EM cluster shape
• Inefficiency 10-5 (gt90 signal acceptance)
• (Inefficiency 10-6 with 50 signal acceptance)

12
Muon Pion ID
Detector RICH Goal Muon Pion separation with
10-2 ineff. over a wide momentum range
As low X0 as possible (RICH before LKr) 1st
option P.S. Cooper FERMILAB-CONF-05-015-CD
Some Brain Storming O. Ullaland (CERN)
0.1
?1 in 2 m Ar 22 pe and ?c23.7 mrad
Argon
0.01
Helium
Dq m / p (rad)
0.001
?1 in 15 m He 21 pe and ?c8.2 mrad
0.0001
5
10
15
20
25
30
35
40
Momentum (GeV/c)
13
Photon Rejection

• Detectors lead-scint sandwich (ANTI), LKr,
  lead-scint sandwich (IRC, SAC)
• Goal 10-8 level of veto inefficiency on p0
  (requirement from pp0)
• Decays with p0 energy correlation between gs
  from p0 decay
• Decays with single photon (radiative)
  hermeticity (0 - 50 mr coverage)

Energy of photons from p0 in pp0 events
ANTI
LKr Eg gt 1 GeV
IRC / SAC Eg gt 6 GeV
1
0
2
1
6
0
2
3
4
5
6
0
1
2
3
4
5
7
8
9
10
Eg (GeV)
Eg (GeV)
Eg (GeV)
14
Simulation of photon rejection results

• Simulation of geometrical layout
• Parametrization of the g inefficiencies
• Inefficiency 2 x 10-8 on p0 from pp0
• 5 x
  10-8 on p0 from Kl3
• 10-3 on g from
  radiative

• GEANT simulation of each
• device started
• Simulation results validating on
• existing detector configurations
• where data are available

Detector E range Inefficiency
ANTI lt 50 MeV 1
ANTI (0.5, 1) GeV 10-4
ANTI gt 1 GeV 10-5
LKR lt 1 GeV 1
LKR (1,3) GeV 10-4
LKR (3,5) GeV 10-4 - 10-5
LKR gt 5 GeV 10-5
IRCs, SAC All 10-6
2mm lead / 6mm Scintillator
DATA S. Ajimura et al., NIM A435 (1999) 408
MC Our GEANT4 simulation
Inefficiency
10-4
10-5
200
400
600
800
1000
Photon Energy (MeV)
15
Electron ID

• Detector LKr
• Inefficiency of e ID studied in
• NA48 with e from p0 Dalitz decay.
• Background from hadronic
• showers
• hID 1 for E/p lt 0.9.
• Improvement of a factor 10 if
• E/p lt 0.85 (about 2 signal lost)
• Room for improvements using
• E/p other informations about
• clusterization (NN technique).
• We assume hID 10-3

16
Signal acceptance and Kaon flux

• Fast simulation of the complete layout
• Signal acceptance (Geometry, kinematic cuts, FF)
• Region I 4.5
• Region II 14.5
• Assumed signal BR 10-10
• Detailed simulation of the beam line
• Expected kaon decays in fiducial region per year
  4.8 x 1012

17
RESULTS Events collected per year
Total Region I Region II
Signal 91.2 21.6 69.6
Km2 1.6 1.4 0.1
pp0 4.4 2.3 2.1
Ke3 1.6 0.2 1.4
Km3 lt0.1 0 lt0.1
Km2g 0.4 0.1 0.3
pp0g lt0.1 0 lt0.1
ppp- In progress 0 In progress
Ke4 In progress 0 In progress
Km4 lt 10-2 Ke4 0 lt 10-2 Ke4
Background 8 charged 4 4 charged
Without Form Factor
18
Thoughts about Charged Vetoes

• Goal at least 5 x 10-3 on a single track
• Reject Ke4, Km4, ppp-
• The most dangerous one Ke4

Task at high angle
2 Gigatracker stations
19
Conclusions

• The experimental layout is being finalised
• Background estimation almost complete
• Region I well understood, a RICH is needed for m
  background rejection.
• S/B 5, but room for improvements.
• Region II
• S/B gt 10, (but with Ke4 and Kp3 contributions
  missing)
• Charged vetoes to be optimised
• Once dead-time and selection cuts are taken into
  account, a 10 signal acceptance is plausible
  (i.e. 40 events/year for Br10-10)