Particle identification
Katharina Müller, autumn 14 1 Particle identification (PID) important task for all detectors in particle and astro particle physics
Particle physics: B-physics, rare decays, CP violation, exotic hadronic decays quark-gluon plasma:identification of as many particles as possible Astro particle physics: distinguish different nuclei, identify charged particles, photons neutrino detection
Distinguish π/K K/p, e/π, π0/γ .. but also neutrino/muon, ν / ν . μ e
Method for PID depends on energy range Optimisation: efficiency and / or misidentification rate
tag Efficiency: ε = Nx / Nx n o i
tag t
Misidentification rate εmis = Ny / Ny c e j e r
Methods: d n
Mass determination u o r
Lifetime g k
Decay products c a
Missing energy B Shower profile Special detectors signal efficiency
Katharina Müller, autumn 14 2 Particle ID: Example HERA-b
Search for Φ→ KK
physics drowned in background Φ→ KK decay only visible after particle identification
Mass Φ= 1019 MeV Φ→ KK BR 48.9% 0 0 Φ→ K LK S BR 34.2% Φ→ π+π-π0 BR 15.3%
Katharina Müller, autumn 14 3 Particle ID
Red: signal B→D0K Yellow: background B→D0π Green: combinatorial background → particle identification needed: select the right tool
Katharina Müller, autumn 14 4 Particle identification
● dE/dX Energy loss of charged particles → PID if momentum is known ● Flight time (TOF) → velocity βc ● Cherenkov radiation (RICH) ● Transition radiation (TR) ● Cluster shape
most detectors use several methods
Katharina Müller, autumn 14 5 Example: ALICE
Collisions of heavy ions (Pb) at 5.5 TeV QuarkGluon Plasma
Hits in TOF
Red hits belong to one particle: pion Identify as many particles as possible!
Katharina Müller, autumn 14 6 Example: ALICE
ITS: Tracker dEdX TPC: dEdX TOF: Time of flight TRD: Transition radiation HMPID: RICH PMD: Photons PHOS: Photons Muon Arm: Muon ID
Hits in TOF
http://www.lhc-facts.ch/index.php?page=alice
Katharina Müller, autumn 14 7 Time of flight measurement (TOF)
TOF: Time of flight good time resolution → scintillators length L p
2 particles (m , m ), momentum p 1 2 distance D L 1 1 t= − c 1 2
=1/ 1−2=E /m c2 L 1 1 L c 2 2 2 Δ t= ( − )≃ (m −m ) relativistic particles E>>mc 2 1 2 c + ( 2/ )2 + ( 2/ )2 2 p √1 m1c E1 √1 m2 c E2 (E ≃ pc and root expansion) L c non relativistic particles t= m −m p 1 2 Δt ~ 1/p2Δm2 : important for small velocities, large mass differences
Katharina Müller, autumn 14 8 Time of flight: measurement (TOF)
L c ≃ 2− 2 t m1 m2 Difference in TOF after 1m 2 p2
time resolution of scintillators 300 ps
→ kaon-pion separation up to 1 GeV with L = 3 m
TOF limited for particles p < GeV
better time resolution: ● plastic scintillators: 80-300 ps ● parallel plates counters: 100-200 ps
Katharina Müller, autumn 14 9 TOF measurement
MeV 125 Phenix at RHIC Heavy ion physics
250
500 1000
flight distance 5 m 1000 plastic scintillators resolution 85 ps 4 σ K/Pion separation p<4 GeV
http://www.phenix.bnl.gov/WWW/tof/
Katharina Müller, autumn 14 10 Particle identification with TOF particle identification NA49 TOF TOF and dE/dX
n o i P T / T
=
l e r T
BELLE
Mass from TOF measurement
Katharina Müller, autumn 14 11 ALICE (TOF)
TOF with very high multiplicity radius 3. 6 m → 150 m² ! Scintillators too expensive → gas detectors Multi gap resistive plate chambers (MGRPC) 160000 channels 2.5 x 3.5 cm² Time resolution better than 100 ps Small gap: good time resolution many gaps: high efficiency
2 x 5 gaps 250 μm 0.4 mm glass plates Spacer: fishing line Width 7 cm (2 Pads) Length 120 cm (48 Pads)
http://aliceinfo.cern.ch/Public/en/Chapter2/Chap2_TOF.html
Katharina Müller, autumn 14 12 ALICE (TOF)
TOF with very high multiplicity Radius 3. 6 m → 150 m² !
Strips: length 240 cm 96 readout pads
http://aliceinfo.cern.ch/Public/en/Chapter2/Chap2_TOF.html
Katharina Müller, autumn 14 13 ALICE (TOF) cleaning
storage
http://aliceinfo.cern.ch/Public/en/Chapter2/Chap2_TOF.html
Katharina Müller, autumn 14 14 ALICE (TOF)
efficiency > 99.9% Efficiency and time resolution as function resolution better than 60 ps (design 80 ps) of particle flow
http://aliceinfo.cern.ch/Public/en/Chapter2/Chap2_TOF.html
Katharina Müller, autumn 14 15 ALICE (TOF) Data
Cosmic rays: two tracks: two TOF signals Δt(exp) = L/c Resolution Δt(meas)-Δt(exp) σ=125 ps
Two independent measurements → resolution for one track: σt =σ /√2 = 88.5 ps
2 σ k-π separation up to 5 GeV in pT
http://aliceinfo.cern.ch/Public/en/Chapter2/Chap2_TOF.html http://indico.cern.ch/materialDisplay.py?contribId=191&sessionId=15&materialId=slides&confId=181055
Katharina Müller, autumn 14 16 Energy loss dE/dX
Reminder Bethe-Bloch formula -K separation 2 2 2 −dE Z 1 1 2 me c T max C 5% resolution! =K z2 [ ln −2− − ] dX A 2 2 I2 2 Z allows to determine βγ if momentum is known
Difficulties: ● Crossings of bands in dE/dX vs p! ● Saturation ● Landau-Tail ● Control measurement uncertainties ● Single measurements not usable
K- have a relative difference of 10% for βγ>3 → high precision (few percent) needed for significant results
Katharina Müller, autumn 14 17 http://arxivweb3.library.cornell.edu/pdf/1209.5637 ALICE dE/dX Nucl. Instr. Meth. A622 (2010) 316
TPC: σ dE/dx = 5 % (Design) Inner tracker: σ dE/dx = 10-11 % (Design)
Resolution vs # TPC track points
Katharina Müller, autumn 14 18 Measurement of dE/dX
Problem: Bethe Bloch formula only gives the mean → single measurements have large variations (Landau distribution) → multiple measurements of dE/dX needed (sampling)
Better method: truncated mean x% of the measurements with highest dE/dX values are neglected (typically 20-30%), or restricted dE/dX
Improvement of resolution with „truncated mean“ (KLOE)
Katharina Müller, autumn 14 19 Separation Power
Important measure Separation power= Separation/Resolution strong momentum dependence
Opal: require 2σ Separation: e-Pion p<14.3 GeV Pion-Kaon p<20.5 GeV
Katharina Müller, autumn 14 20 Measurement of Landau-distribution
Several measurements of dE/dX: calculate probability that measured dE/dX distribution belongs to pion, kaon, p etc
i P π(x) probability that pion produces a signal x in detector i i P K(x) kaon
each particles produces a set of xi signals. Probability that this set of signals originates from a pion is i i Pπ = ∏i P π(xi) or for a kaon PK = ∏i P K(xi)
Probability that particle is a pion
P = Pπ/(Pπ+PK)
Already few measurements are enough to reach an effective pion-kaon separation up to 100 GeV.
Many measurements: fit Landau distribution
Katharina Müller, autumn 14 21 Systematic errors of measurement
• Non-linearities of readout electronics
• Stability of discriminator threshold
• Purity of chamber gas. Small impurities (10-6!) change gas amplification
• Stability of geometry, mechanical tolerances
• Pressure dependence of gas amplification
• Charge distribution depends on scattering angle
• Track multiplicity changes gas/amplification
• Noise
• Crosstalk
•.....etc
• Has to be understood at the 1% level!
Katharina Müller, autumn 14 22 Alice TPC: simulated separation
Katharina Müller, autumn 14 23 Detector Accelerator Type Size B (T) Gas Mixture Pressure Number of Sampling Effective track dE/dx resolution (∅ x L) (bar) samples length (mm) length (bar * m) isol., dense (%) ALEPH LEP TPC 3.6 m x 4.4 m 1.5 Ar/CH4 (91/9) 1 338 4 1.35 4.5 ARGUS DORIS drift cells 1.7 m x 2 m 0.8 C3H8/Methylal 1 36 18 0.65 4.1 BaBar PEP-II drift cells 1.6 m x 2.8 m 1.5 He/i-C4H10 (80/20) 1 40 12 0.48 7.5 BELLE KEK-B drift cells 1.9 m x 2.2 m 1.5 He/C2H6 (50/50) 1 47 16 0.75 5.5 BES BEPC jet cells 2.3 m x 2.1 m 0.4 Ar/CO2/CH4 (89/10/1) 1 54 5 0.27 9.0 CDF TEVATRON jet cells 2.6 m x 3.2 m 1.5 Ar/C2H6/C2H6O (49.6/49.6/0.8) 1 32 12 0.38 7.0 CLEO II CESR drift cells 1.9 m x 1.9 m 1.5 Ar/C2H6 (50/50) 1 51 14 0.71 6.2 CLEO III CESR drift cells 1.6 m x 1.9 m 1.5 He/C3H8 (60/40) 1 47 14 0.66 5.0 CRISIS TEVATRON jet cells 1 m x 1 m x 3 m - Ar/CO2 (80/20) 1 192 15 2.88 3.2 DELPHI LEP TPC 2.4 m x 2.7 m 1.2 Ar/CH4 (80/20) 1 192 4 0.77 5.7 D0 FDC TEVATRON jet cells 1.2 m x 0.3 m - Ar/CH4/CO2 (93/4/3) 1 32 8 0.26 12.7 H1 HERA jet cells 1.7 m x 2.2 m 1.13 Ar/C2H6 (50/50) 1 56 10 0.56 10.0 JADE PETRA jet cells 1.6 m x 2.4 m 0.48 Ar/CH4/i-C4H10 (88.7/8.5/2.8) 4 48 10 1.92 6.5 KEDR VEPP-4M jet cells 1.1 m x 1.1 m 2.0 DME (100) 1 42 10 0.42 10.0
KLOE DAΦNE drift cells 4 m x 3.3 m 0.6 He/i-C4H10 (90/10) 1 58 28 1.62 3.5 MARK II SLC drift cells 3 m x 2.3 m 0.475 Ar/CO2/CH4 (89/10/1) 1 72 8.33 0.60 7.0 NA49 SPS TPC 3.8 m x 3.8 m x 1.3 m - Ar/CH4/CO2 (90/5/5) 1 90 40 3.60 4.7 OBELIX LEAR jet cells 1.6 m x 1.4 m 0.5 Ar/C2H6 (50/50) 1 40 15 0.60 12.0 OPAL LEP jet cells 3.6 m x 4 m 0.435 Ar/CH4/i-C4H10 (88.2/9.8/2) 4 159 10 6.36 2.8 SLD SLC jet cells 2 m x 2 m 0.6 CO2/Ar/i-C4H10 (75/21/4) 1 80 6 0.48 7.0 STAR RHIC TPC 4 m x 4.2 m 0.5 Ar/CH4 (90/10) 1 45 17.2 0.77 8.0 TOPAZ TRISTAN TPC 2.4 m x 2.2 m 1.0 Ar/CH4 (90/10) 3.5 175 4 2.45 4.4 TPC/2γ PEP TPC 2 m x 2 m 1.375 Ar/CH4 (80/20) 8.5 183 4 6.22 3.0 ZEUS HERA jet cells 1.7 m x 2.4 m 1.43 Ar/CO2/C2H6 (90/8/2) 1 72 8 0.58 8.5
Alice: 5% resolution
t Best performance: large detectors & high pressure
Katharina Müller, autumn 14 24 Different approach: Cluster Counting
Traditionally dE/dx measurements integrate all charge deposited on the wire as a proxy for number of primary ionisation
Fluctuations in gas gain and number of primary electrons degrades measurements
Counting primary ionization (clusters) reduces spread around the mean, improving particle identification Number of primary ionisations (Cluster) is Poisson distributed BUT: needs excellent position resolution: Silicon, MSGC
→ large number of clusters needed to reduce statistical error → Density of clusters in track must be low enough to see individual clusters
Studied for ILC and SuperB
Katharina Müller, autumn 14 25 Different approach: Cluster Counting
Number of primary ionisations (Cluster) is Poisson distributed needs excellent position resolution: Silicon, MSGC
Micro strip gas chambers (MSGC) +GEM
Katharina Müller, autumn 14 26 Cluster Counting
#peaks per cm: Pions: 6.45 Kaons: 5.48
Separation power comparable to dEdX plus truncated mean
L. Cerrito et. al, NIM A 434 (1999) 261270 3 GeV, 120 cm track n o i t u l
o dEdX s e r
dNdX
Studies for ILC: resolution <5% possible (arXiv:0708.0142v2 [hep-ex]) Problems: cluster dissolve Proof of principle in test beams but not used in large scale detectors so far
Katharina Müller, autumn 14 27