ATL
-PH
YS-
SLID
E-20
09-1
4312
June
2009
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Mauro Donegàon behalf of the ATLAS Collaboration
Department of Physics and Astronomy University of Pennsylvania
Photon Reconstruction and Identification with
the ATLAS detector
Mauro Donegà - Penn Susy 09 LHC/Tevatron - New Physics Signatures
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Photons at the LHC
2
di-jet
Main sources of high pT isolated γ at the LHC: QCD processes.Main background: jets
The search for new phenomena with photons requires:- efficient photon reconstruction- accurate photon energy and direction measurement- particle identification: jets rejection
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
LHC environment and photon driven detector specifications
LHC design luminosity: 1034 cm-2 s-1 (~20 proton-proton interactions per bunch crossing):
- Pile up in space: high granularity of all sub-detectors (in particular the tracker)- Pile up in time: (25 nsec BC) fast read out- High radiation environment over 10 years of data taking 1015 neutrons/cm2 200 kGy
Photon driven detector design goals:
- mass resolution of 1% on H→γγ : - σ/E ~ 10%/√E with a constant term < 1 %- linearity: (Emean-Etrue)/ Etrue better than 0.5 % - absolute precision of the e.m. scale 0.1 %- angular resolution 50 mrad/√E
- jets rejection:- ~5 orders of magnitude
3
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
!" = 0.0245
!# = 0.02537.5mm/8 = 4.69 mm!# = 0.0031
!"=0.0245x436.8mmx4=147.3mm
Trigger Tower
TriggerTower!" = 0.0982
!# = 0.1
16X0
4.3X0
2X0
1500
mm
470 m
m
#
"
# = 0
Strip cells in Layer 1
Square cells in Layer 2
1.7X0
Cells in Layer 3!"$!# = 0.0245$0.05
middle
back
Δφ = 0.0245
Δη = 0.025Δη = 0.003
(4 mm)
strips
4
Coverage |η| < 2.0 3 pixel layers 8 strip layers (4 space points)~36 measurements on track straw detector electron PID capability
Coverage |η| < 3.2 (precision < 2.5)Total thickness: >24X0
170k channelsATLAS Tracker
Liquid Argon e.m. calorimeter
ATLAS sub-detectors used in photon reconstruction
2T Solenoid
4.3 X0
16 X0
2 X0
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
LAr calorimeter upstream material
|!|0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
) 0Ra
diat
ion
leng
th (X
0
0.5
1
1.5
2
2.5
|!|0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
) 0Ra
diat
ion
leng
th (X
0
0.5
1
1.5
2
2.5ServicesTRTSCTPixelBeam-pipe
LHC trackers: heavier than any previous one (LEP, TeVatron):ATLAS Inner Detector ~ 4.5 ton
Large amount of material in front of the EM calorimeter means: - large energy losses of electrons through bremsstrahlung - high photon conversion probability - a complex calibration procedure for the EM calorimeter
physics : in reconstruction and identification need a dedicated treatment
as a tool to reconstruct the 3D material map of the detector 5
Rad
iatio
n le
ngth
(X0)
The conversion probability ranges from 40 % (η =0) up to 80% (η =1.8)
TRT
SCT
Pixel
beam pipe
η =
0
η =1.8
Knowing the exact amount and position of the upstream material is of paramount importance !
conversions
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Conversions as a tool
Silicon Tracks: - inside-out: it seeds in the silicon detector and extends outward- outside-in: it use as a seed a TRT segment and extends inward
TRT Tracks: TRT segments without silicon hits extensions
6
Precise 3D mapping of the material with low-energy conversions from minimum bias data Enough statistics for a 1% material map in a few months of data taking
Tracking efficiency
Silicon Tracks
Conversion ingredients: Tracking + Vertexing
TRT Tracks
Photons ET= 5 GeV
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector 7
Conversion reconstruction efficiency
• Radial resolution: ~ 3 mm for low pT photons
• Other tools are available to understand the material distribution in the detector:- tail of the E/p distributions- dedicated brem-fitting tools- resonances (J/ψ, K0s ...) mass stability as a function of the tracks pT
Material in ID has been checked to a few % by weight and component lists.
Conversion as a tool (II)
Photons
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
From cells to clusters to photon candidates
8
Clustering algorithm for photons:• slide a window ΔηxΔφ = 5x5 in the middle layer (Molière radius ~ 2 cm); find local maximum• track/vertex matching to disentangle electrons from unconverted and converted photons• re-build the cluster: ΔηxΔφ = 3x5 cells for unconverted photons ΔηxΔφ = 3x7 cells for converted photons (wider phi to account for B-field opening; like electrons)
Energy and position reconstruction: See Olivier Arnaez “Electron Reconstruction and Identification with the ATLAS Detector” this conference
• cells in the clusters are summed• position dependent energy corrections are applied.
Photon energy resolution
E =100 GeV single photons MC
All photonsUnconverted photons
|η| = 1.075 |η| = 1.075σ = (1.37± 0.05)% σ = (1.26± 0.05)%
conversions
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
From cells to clusters to photon candidates (II)
9
Relative resolutionLinearity within 0.1 %
linea
rity
Shower direction resolution
Unconverted photons(45 ÷ 75) mrad/√E
Converted photonswidth 0.42 mrad
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Photon/Jets discrimination based on their characteristics features:Photons narrow objects, contained in the e.m. calorimeter.Jets broader profile and have significant energy deposition in the hadronic calorimeter
• Hadronic leakage
Photon identification methods
10
η
φ
η
φ
η
φ
η
φ
Rη = 3x7 / 7x7 Rφ = 3x3 / 3x7
• Shower shapes in the middle layer of the electromagnetic calorimeter:
Plots shown for:|η|<0.7 20<ET<30GeV
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
• Example of a variable built with the the strips (first layer) of the electromagnetic calorimeter:
Photon identification methods (II)
11
Emax2
Single photon π0
Emin
ΔEs = Emax2 - Emin
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Track Isolation
12
After shower shape cuts the remaining background is dominated by low track multiplicity jets containing a high pT π0
Track Isolation = sum of the pT of all tracks with pT>1 GeV within ΔR<0.3 (tracks from conversions are excluded)
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Cut based Photon Identification performance
13
Example: photons ET > 25 GeV from H→γγ
After all cuts 70% of the remaining fake photons are high momentum π0
Photons Fakes
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Cut based Photon Identification performance (II)
14
Example: γ ET > 100 GeV from Randall Sundrum G→γγ (mG = 500 GeV )
Calorimeter isolation ΔR = 0.45 around the cluster
Photons Fakes
High pT γ shower shape distributions similar to the ones of H→γγ
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
More ambitious photon identification methods
15
Log-Likelihood-Ratio: combines the same variables used with the cut based method.
Index i runs over the variables
For equal efficiencies the LLR and the cut-based method perform comparably
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Summary
16
• The performance of the ATLAS electromagnetic liquid Argon calorimeter and tracker have been extensively studied with beam tests, with simulations and with cosmics data.
• The large amount of upstream material affects the photon reconstruction and identification. Different approaches are being developed to measure its distribution with collision data.
• Different methods are available to identify photons. Detailed MC studies show that the efficiency and rejection levels are adequate for physics analysis.
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Backup17
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
Liquid Argon e.m. calorimeter
18
HV = 2KV/gap = 10KV/cm450 nsec drift time
Design goals:- Energy resolution: + constant term < 1% to have1% mass resolution in H→γγ- Signal linearity 0.5%- Absolute e.m. energy scale 0.1%- Angular resolution
- Electron reconstrution: [1GeV, 5TeV]- photon/jet, electron rejection 106
- Noise: 1GeV to have ~1% on H→γγ mass: this drive the max granularity ΔηxΔφ = 0.3x0.3 for the precision
regionFast electronics: for noise reduction and for triggerRadiation hardness in10 years: 1015 1MeVneq/cm2
200 kGy
!
E! 10%"
E
50mrad!E
From accordion to cells: cells are created ganging signals from the appropriate electrodes together: Strips: 16 electrodes in φ; L2 4 electrodes in φ
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector 19
Liquid Argon e.m. calorimeter
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
• Hadronic leakage• Variables in the middle layer of the electromagnetic calorimeter:
Rη = 3x7 / 7x7 Rφ = 3x3 / 3x7
w2 = lateral width in η• Variables using the the strips (first layer) of the electromagnetic calorimeter:
Shower shape variables
20
η
φ
η
φ
η
φ
η
φ
ΔEs = Emax2 - Emin
Rmax2 = Emax2/(1+0.009 ET/GeV)Fside = [E(±3)-E(±1)]/E(±1)ws3 = shower width around the strip with the maximal energy depositwstot = shower width over 20 strips
Emax2
Single photon π0
Emin
Mauro Donegà - Penn Photon Reconstruction and Identification with the ATLAS detector
H-matrix
21
Covariance-based (H-matrix): combines different shower shape variables
Indices i,j runs over the variablesN = number of photons used for training = mean value for the variable y
H = M-1
= measured value of the variable
The photon likeness then is defined as