Outline Lecture - Introduction Lecture - Tracking Detectors Lecture 3 - Scintillation and Photodetection Lecture 4 Calorimetry Lecture 5a - Particle Identification C. Joram, L. Ropelewski L. Ropelewski, M. Moll C. D Ambrosio, T. Gys C. Joram C. Joram de/dx measurement Time of flight Cherenkov detectors Transition radiation detectors Lecture 5b - Detector Systems/ Design C. D Ambrosio CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/
Particle Identification Particle identification is an important aspect of high energy physics experiments. Some physical quantities are only accessible with sophisticated particle identification (B-physics, CP violation, rare exclusive decays). One wants to discriminate: π/k, K/p, e/π, γ/π 0. The applicable methods depend strongly on the interesting energy domain. Depending on the physics case either ε xx or ε xy has to be optimized: Efficiency: Misidentification: Rejection: ε xx xy = N tag x x tag N y N x N y The performance of a detector can be expressed in terms of the resolving power D x,y D x, y = S x S σ S y ε Rxy = = ε xx ε xy ε xx S x and S y are the signals provided by the detector for particles of types x and y with a resolution σ S. ε xy CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/
Particle Identification - an example A charmless B decay: displaced secondary vertex B-meson Who is who? K + π in final state DELPHI CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/3
Particle ID through de/dx p = m βγc de dx 0 ln β ( β γ ) Simultaneous measurement of p and de/dx defines mass m 0, hence the particle identity (arbitrary units) e µ π K p π/k separation (σ) requires a de/dx resolution of < 5% Not so easy to achive! de/dx is very similar for minimum ionising particles. Energy loss fluctuates and shows Landau tails. Average energy loss for e, µ, π, K, p in 80/0 Ar/CH 4 (NTP) (J.N. Marx, Physics today, Oct.78) CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/4
Particle ID through de/dx How to reduce fluctuations? subdivide track in several de/dx samples calculate truncated mean, i.e. ignore samples with (e.g. 40%) highest values Also increased gas pressure can improve resolution ( higher primary statistics), but it reduces the rel. rise due to density effect! Don t cut the track into too many slices! There is an optimum for a given track length L. wire 4 wires L: most likely energy loss A: average energy loss (B. Adeva et al., NIM A 90 (990) 5) (M. Aderholz, NIM A 8 (974), 49) CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/5
Example ALEPH TPC Gas: Ar/CH4 90/0 N samples = 338 wire spacing 4 mm de/dx resolution ~5% for m.i.p. s log scale! CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT linear scale! 5a/6
High resolution de/dx by cluster counting E most probable < E> Landau curve Experimental observation (first order) W (FWHM) E m.p. E Remember (lecture a): the number of primary electron - ion pairs is Poisson distributed! What would be the resolution in E if we could count the clusters? W n primary cm Ar n primary 8 =.35 = 0. 44 E n d m. p. primary Average distance d 360 µm t = d/v drift few ns In addition diffusion washes out clusters Principle of cluster counting has been demonstrated to work - Time Expansion Chamber - but never successfully applied in a particle physics experiment. (A.H. Walenta, IEEE NS-6, 73 (979)) CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/7
High resolution de/dx by cluster counting Cluster counting with a hybrid gas detector: pixel readout chip + micromegas δ electron another event He / isobutane 80/0 5 mm 50 µm ~ 4 x 4 mm micromegas Medipix chip foil 56 x 56 pixels, 55 x 55 µm, each CERN Academic Training Programme 004/005 M. Campbell et al., NIM A 540 (005) 95 track by cosmic particle (mip): 0.5 clusters / mm, ~3 e - /cluster C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/8
5a/9 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT CERN Academic Training Programme 004/005 t for L = m path length σ t = 300 ps π/k separation up to GeV/c c L t β = 0 = L t c p m start stop Combine TOF with momentum measurement + + = L dl t dt p dp m dm γ Mass resolution TOF difference of particles as f(p) ( ) / / m m p Lc p c m p c m c L c L t + + = = β β Particle ID using Time Of Flight (TOF) p = m 0 βγ L tc L = β
Example: NA49 Heavy Ion experiment detail of the grid High resolution TOF requires C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT Small, but thick scint. 8 x 3.3 x.3 cm 3 Long scint. (48 or 30 cm), read out on both sides fast detectors (plastic scintillator, gaseous detectors, e.g. RPC (ALICE)), appropriate signal processing (constant fraction discrimination, corrections) continuous stability monitoring. 5a/0 CERN Academic Training Programme 004/005
Example: NA49 Heavy Ion experiment System resolution of the tile stack From γ conversion in scintillators NA49 combined particle ID: TOF + de/dx (TPC) T rel. = T / T π L = 5 m CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/
back to... Interaction of charged particles Remember energy loss due to ionisation There are other ways of energy loss! A photon in a medium has to follow the dispersion relation c n c k c ω = πν = π = k ω = 0 ε = n λ n ε Optical behaviour of medium is characterized by the dielectric constant ε Re ε = n Refractive index regime: effect: ε Imε = k Absorption parameter Re ε Im ε optical absorptive X-ray Cherenkov radiation ionisation schematically! transition radiation ω Assuming soft collisions + energy and momentum conservation emission r ω v k r of real photons: = v k cosθ v r,m 0 cosθ = h ω, hk ω = vk = nβ Emission of photons if β = / n cosθ θ ē β A particle emits real photons in a dielectric medium if its speed β c is greater than the speed of light in the medium c/n β n ε CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/
Cherenkov radiation Cherenkov radiation is emitted when a charged particle passes through a dielectric medium with velocity β β thr = n n : refractive index L tanθ wave front l light =(c/n) t θ l part =βc t cosθ C = nβ with n = n( λ) L θ C β thr = θc n 0 Cherenkov threshold Number of emitted photons per unit length and unit wavelength interval dn/dλ d N π z α π z α = sin dxd n = θc λ λ β λ d N c hc d N with λ = = = const. dxdλ λ ν E dxde C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT θ max = arccos n saturated angle (β=) dn/de λ 5a/3 Ε CERN Academic Training Programme 004/005
Cherenkov detectors medium n θ max (deg.) N ph (ev - cm - ) air*.00083.36 0.08 isobutane*.007.89 0.94 water.33 4. 60.8 quartz.46 46.7 96.4 *NTP Number of detected photo electrons N p. e. Energy loss by Cherenkov radiation small compared to ionization ( 0.%) Cherenkov effect is a very weak light source need highly sensitive photodetectors E α = Lsin θ Q E i ( E)dE c ε ( ) ε h E i N0 = 370 ev cm ε total E = E -E is the width of the sensitive range of the photodetector (photomultiplier, photosensitive gas detector...) N 0 is also called figure of merit ( ~ performance of the photodetector) Example: for a detector with ε total E = 0. ev L = cm o and a Cherenkov angle of = 30 one expects 8 photo electrons N p. e. = θ C E CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/4
Cherenkov detectors Detectors can exploit.... N ph (β) threshold detector (do not measure θ C ). θ(β) differential and Ring Imaging Cherenkov detectors RICH principle mirror Threshold Cherenkov detectors ( m ) N ph = + p n β n Example: study of an Aerogel threshold detector for the BELLE experiment at KEK (Japan) Goal: π/k separation β kaon β ; light yield (a.u.)..0 0.9 0.8 β kaon n=.03 0.7 0.6 0.5 n=.0 0.4 0.3 0. 0. n=.0 0.0 0 3 4 5 6 p kaon [GeV/c] particle radiator medium PM CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/5
Ring Imaging Cherenkov detectors (RICH) RICH detectors determine θ C by intersecting the Cherenkov cone with a photosensitive plane requires large area photosensitive detectors, e.g. wire chambers with photosensitive detector gas PMT arrays (J. Seguinot, T. Ypsilantis, NIM 4 (977) 377).......... π/k π/k/p K/p n =.008 C 5 F gas n =.8 C 6 F 4 liquid DELPHI π/h π/k/p K/p θ C = arccos = arccos nβ n = arccos n p + m p cos θc = nβ E p σ β = tanθ σ β Detect N p.e. photons (photoelectrons) p. e. θ σ p.e. σθ minimize σ θ N p. e. maximize N p.e. θ CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/6
Ring Imaging Cherenkov detectors (RICH) Reconstruction and resolution of Cherenkov angle detector window Determination of θ C requires: space point of the detected photon (x,y,z) photodetector granularity (σ x, σ y ), depth of interaction (σ z ) window radiator (x e,y e,z e ) the chromatic error - an irreducible error θ C n rad = n(ε) c σ θ θ C = σ n n tanθ emission point (x e,y e,z e ) keep radiator thin or use focusing mirror particle direction θ p, φ p RICH requires good tracker dn = σ E n tanθ de σ E is related to the sensitivity range of the photodetector E E N pe good σ E bad E N pe bad σ E good CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/7
Ring Imaging Cherenkov detectors (RICH) C 5 F (40 cm, gas) C 4 F 0 (50 cm, gas) spherical mirror DELPHI and SLD: A RICH with two radiators and a common photodetector plane covers a large momentum range. π/k/p separation 0.7-45 GeV/c: (W. Adam et al. NIM A 37 (996) 40) Photodetector TMAE-based C 6 F 4 ( cm, liquid) Two particles from a hadronic jet (Z-decay) in the DELPHI gas and liquid radiator. Circles show hypotheses for π and K CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/8
RICH detectors in LHCb photodetectors (HPD) flat mirror RICH RICH spherical mirror aerogel radiator L(C 4 F 0 ) ~ 85 cm radiator C 4 F 0 aerogel θ 3.03 3.8 n.004.03 p thresh (π).6 0.6 GeV/c N p.e. 3 6.8 σ θ.9.9 mrad p (3σ) 56 3.5 GeV/c C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT radiator CF 4 θ.8 n.0005 p thresh (π) 4.4 GeV/c N p.e. 3 σ θ 0.6 mrad p (3σ) 98.5 GeV/c 5a/9 CERN Academic Training Programme 004/005
RICH detectors in LHCb RICH photodetector plane CERN Academic Training Programme 004/005 beam test in 004 with 6 HPDs C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/0
RICH detectors in LHCb Beam test results with C 4 F 0 radiator gas (autumn 004). Single pion (0 GeV/c) Superimposed events (00 k pions, 0 GeV/c) 70 60 50 40 30 0 0 0 0 0 0 30 40 50 60 70 80 Column 90 number Left Centre Right Run Number: 3 Number of hits: 397589 0 GeV/c pion Number of events: 99477 Row number 70 60 50 40 30 0 0 0 0 0 0 30 40 50 60 70 80 Column 90 number Left Centre Right CERN Academic Training Programme 004/005 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0. 0. 0 5000 0000 5000 0000 5000 0 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/ Row number Run Number: 3 0 GeV/c pion Number of hits: 48 Number of events:
Particle ID by Transition radiation (there is an excellent review article by B. Dolgoshein (NIM A 36 (993) 434)) Transition Radiation was predicted by Ginzburg and Franck in 946 TR is electromagnetic radiation emitted when a charged particle traverses a medium with a discontinuous refractive index, e.g. the boundaries between vacuum and a dielectric layer. ε TR is also called sub-threshold Cherenkov radiation A (too) simple picture medium vacuum Re ε ε <! A correct relativistic treatment shows that (G. Garibian, Sov. Phys. JETP63 (958) 079) Radiated energy per medium/vacuum boundary W = αhω pγ W γ only high energetic e ± emit TR of detectable intensity. 3 particle ID ω p = Nee ε m 0 e plasma frequency hω 0eV p electron (plastic radiators) Im ε regime: optical absorptive X-ray ω effect: Cherenkov ionisation transition radiation radiation CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/
Particle ID by Transition radiation W Number of emitted photons / boundary is small N ph α hω 37 Need many transitions build a stack of many thin foils with gas gaps Emission spectrum of TR = f(material, γ) Typical energy: hω 4 hω photons in the kev range p γ Simulated emission spectrum of a CH foil stack X-rays are emitted with a sharp maximum at small angles θ γ TR stay close to track Particle must traverse a minimum distance, the so-called formation zone Z f, in order to efficiently emit TR. c Z f =, ξ = ω / ω p ω( γ + θ + ξ ) Z f depends on the material (ω p ), TR frequency (ω) and on γ. Z f (air) ~ mm, Z f (CH ) ~ 0 µm important consequences for design of TR radiator. CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/3
Particle ID by Transition radiation TR Radiators: stacks of thin foils made out of CH (polyethylene), C 5 H 4 O (Mylar) hydrocarbon foam and fiber materials Low Z material preferred to keep re-absorption small ( Z 5 ) R D R D R D R D alternating arrangement of radiators stacks and detectors minimizes reabsorption TR X-ray detectors: Detector should be sensitive for 3 E γ 30 kev. Mainly used: Gas detectors: MWPC, drift chamber, straw tubes Detector gas: σ photo effect Z 5 gas with high Z required, e.g. Xenon (Z=54) Intrinsic problem: detector sees TR and de/dx C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT Pulse height ( cm Xe) de/dx 00 e - TR (0 kev) 500 e - Discrimination by threshold 5a/4 t CERN Academic Training Programme 004/005
Particle ID by Transition radiation The ATLAS Transition Radiation Tracker (TRT) 680 cm Straw tubes (d = 4mm) based tracking chamber with TR capability for electron identification. photo of an endcap TRT sector. Active gas is Xe/CO /O (70/7/3) operated at ~x0 4 gas gain; drift time ~ 40ns ( fast!) Radiators Barrel: Propylen fibers Endcap: Propylen foils d=5 µm with 00 µm spacing. Counting rate ~ 6-8 MHz at LHC design luminosity 0 34 cm - s - CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/5
ALICE TRD Time Expansion Chamber with Xe/CO gas (85%-5%) 7.36 m amplification region 5mm 7.34 m TRD fibers drift region 30 mm CERN Academic Training Programme 004/005 TPC C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/6
ALICE TRD performance integ. charge method charge + t D method CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/7
Particle ID by Transition radiation Rejection Power : R π/e = ε π /ε e (90%) one order of magnitude in Rejection Power is gained when the TRD length is increased by ~ 0 cm CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/8
Particle Identification Summary: A number of powerful methods are available to identify particles over a large momentum range. Depending on the available space and the environment, the identification power can vary significantly. A very coarse plot. TOF de/dx RICH TR 0-0 0 0 0 0 3 0 4 p [GeV/c] e ± identification π/k separation π Κ µ p? CERN Academic Training Programme 004/005 C. D Ambrosio, T. Gys, C. Joram, M. Moll and L. Ropelewski CERN PH/DT 5a/9