Particle Identification Techniques Basic definitions and introductory remarks. Ionization energy loss. Time of Flight. Cherenkov radiation

Transcription

1 Particle Identification Techniques Basic definitions and introductory remarks Ionization energy loss Time of Flight Cherenkov radiation Transition radiation Prof. J. Engelfried (next week) Advised textbooks: R. Fernow, Introduction to Experimental Particle Physics, Cambridge University Press R.S. Gilmore, Single particle detection and measurement, Taylor&Francis, 199 G. F. Knoll, Radiation Detection and Measurement, John Wiley and Sons, New York W. R. Leo, Techniques for Nuclear and Particle Physics Experiments, Springer, 1994

2 Introduction Complete event analysis (based on the reconstruction of conservation laws): 4-momenta of secondary particles ( p,e) r Deflection in a magnetic field (+ sign of particle s charge) Calorimetry (destructive measurement) PID measurement 4 E = m c + m uniquely identifies the internal quantum numbers of the particle p c Example:

3 A Look at the Past 70 s: Hydrogen bubble chamber 1978: BEBC

4 A Modern Approach to PID ALICE at LHC (talk by Prof. Gerardo Herrera Corral) Silicon trackers +TPC (PID with energy loss) TOF TRD Ring Imaging Cherenkov detector

5 RICH magnetic field p, charge TOF and TRD e.m. calorimeter Basic Layout Layers of silicon detectors with excellent position (0(10 µm)) and double track (0(100 µm)) resolution near the primary collision region detection of secondary vertices (short-lived strange and heavy flavor particles) impact parameter resolution σ(rφ) ~ 50 µm for p t ~ 1 GeV/c primary vertex resolution: ~ 10 µm momentum resolution improvement PID with energy loss TPC, away from the interaction region, at more moderate particle densities tracking (δp/p at the level of 1% for low momenta) PID with energy loss

6 PID Usefulness: Examples No PID one K identified two Ks identified φ Κ + Κ m φ =100 MeV/c M 1 ( E ) = invariant mass = i i p jc c j

7 Background Reduction

8 Measuring the Particle Velocity p = mc γβ ; γ = E mc (Lorentz factor) dm m = γ dβ β + dp p dp p m 1 0 m = p β c ( β1 + β ) ( β β ) 1 β β ( m 1 m p ) c

9 Particle Identification Techniques PID techniques are based on the detection of particles via their interaction with matter: TR+dE/dx ionization and excitation (Cherenkov light & Transition Radiation) The applicable methods depend strongly on the particle momentum (velocity) domain of interest π-k identification ranges electron identification Cherenkov de/dx TOF 150 ps FWHM p (GeV/c)

10 Efficiency & Contamination (example: ALICE-ITS simulation) efficiency = ε A A = N A identified N A total contamination = ε B A = B, B A N B A identified / N A identified higher efficiency -> larger contamination

11 Separation Power Detector length (m) σ separation for π/k ToF (100ps@FWHM) TR+dE/dx de/dx Liquid-Solid Aerogel Gases RICH Momentum (GeV/c) N.B. in case of samples with different population: at a given separation power, the resulting contamination of the largest populated sample of particles in the other nσ = separation power species will be larger by a factor equal to the ratio between the relative populations = S A S B σ AB

12 Energy Loss Mechanisms δ-ray Basic processes occurring when a charged particle traverses a medium being surrounded by a cloud of virtual photons that interacts with atoms in the medium ionization and excitation of the atoms of the medium (secondarily produced electrons could further ionize the medium) radiative phenomena! Cherenkov radiation! Transition radiation Overall effect: the particle loses energy Detection of the energy lost is the physical basis of many of the techniques used in charged particle detectors Ionization trail: particle s trajectory and velocity information

13 Charged Particle-Matter Interactions Modern approach (unitary description in terms of matter properties): Allison and Cobb (1980) Charged particle moves in a dielectric medium through which virtual photons propagate The particle loses energy by doing work against the field created by the medium polarization particle p,m 0 virtual photon h ω, hk θ e - atom time fast charged particle A (photon) B atom de dx = n 0 de ω/ v d σ dpe dedp p = hk E = hω density of atoms: n=ρn A /A rel. probability below excitation threshold excitation ionization by distant collision Schematically! (after Gilmore) ionization by close collisions δ-electrons energy transfer/photon exchange space photons are virtual, their energy and momentum are independent E pc the integral must be performed over both energy and momentum separately virtual photon behaviour approximated with a combination of cross sections for the interactions of real photons allowing to perform the momentum integration for virtual photons de = dσ n E de dx 0 de

14 Ionization Energy Loss de dx Classical approach : Bethe-Bloch equation modified to include the Fermi effect Average specific energy loss: = Z A z ρ t β 1 mec ln Valid only for particles with m>m e δ de/dx does not depend on m but on the charge z Non relativistic region: de/dx 1/β (more precisely as β -5/3 ) Minimum: at βγ = 3 4 (Minimum Ionizing Particle) de 1 mip 1 MeV g cm dx At high βγ: de/dx lnγ (relativistic rise) Density effect: δ(βγ) (medium polarization reduces long range effects)! saturates at βγ sat : 30 Ar 68.4 CH C H C 4 H Si β I γ E max β C Z Z=atomic number of the medium; I~Z 1 ev=effective ionization potential; E max =max energy transfer (I de E max )

15 de dx p = 1 ln β m βγ ( β γ ) Particle ID using the specific energy loss de/dx m from simultaneous measurement of p and de/dx n σ = de / dx de σ A ( de / dx) B B / dx (de/dx)/(de/dx) min Fermi plateau is a few percents above the minimum in solid and liquid media, 50-70% in high Z noble gases at STP -> PID in the relativistic rise region only possible in gases! π/k separation (σ) requires a de/dx resolution of few percents : cross-over regions (as wide as MeV/c) ambiguites-> complementary PID mandatory Average energy loss in 80/0 Ar/CH 4 (NTP) (J.N. Marx, Physics today, Oct.78)

16 Fluctuations in the energy loss de/dx <de/dx> is practically measured by evaluating E in a short interval δx this is not necessarily the average energy lost in the given slice of material-> the distribution shows large fluctuations and Landau tail Most interactions involve little energy exchange -> the total energy loss from these interactions is a Gaussian (central limit theorem). Few interactions involve large energy exchange-> Landau tail Because of the high energy tail, increasing the thickness of the detector or choosing high Z material does not improve σ(de/dx). Indeed the relative width of Gaussian peak reduces but probability of high energy interaction rises -> tail gets bigger L: most likely energy loss A: average energy loss 1 wire 4 wires (B. Adeva et al., NIM A 90 (1990) 115)

17 Improve de/dx resolution and fight Landau tails Thick absorber: large chance of high energy δ ray production cancels the reduction of fluctuations -> (de/dx) A (de/dx) B < Landau fluctuations Usual method of measuring de/dx is: Choose material with high specific ionization Divide detector length L in N gaps of thickness T. Sample de/dx N times Calculate truncated mean, i.e. ignore samples with (e.g. 40%) highest values Also pressure increase can improve resolution. Drawback: reduced relativistic rise due to density effect! Samples must not be too many: for each total detector length L, there exists an optimal N Rule of thumb: at least N=100 for a total track length of 3-5 m/atm (M. Aderholz, NIM A 118 (1974), 419)

18 de/dx & Separation Power Particle Separation ( A; B) = de / dx( A) de / dx( B) σ N S. D. / ( de dx) de/dx resolution (A.H. Walenta et al. Nucl. Instr. and Meth. 161 (1979) 45) σ ( ) de / dx n ( t p) n: number of sampling layers, t: thickness of the sampling layer (cm) p: pressure of the gas (atm) Remarks: σ does not follow the n -0.5 dependence owing to the Landau fluctuations; if the total lever arm (nt) is fixed, it is better to increase n; so long as the number of produced ion-pairs is enough in each layer.

19 TPC: basic principle Time Projection Chamber full 3-D track reconstruction x-y from wires and segmented cathode of MWPC z from drift time -> precise knowledge of v D (LASER calibration + p,t corrections) de/dx Drift over long distances very good gas quality required Gate open Gate closed V g = 150 V Space charge problem from positive ions, drifting back to medial membrane gating

20 Tracker evolution 80 s: 6.4 TeV Sulphur - Gold event (NA35) 000: STAR STREAMER CHAMBER TPC

21 STAR TPC 40 cm Gas: P10 ( Ar-CH 4 90%-10% 1 atm, 50,000 Liters Voltage : - 31 kv at the central membrane 148 V/cm over 10 cm drift path Self supporting Inner Field Cage: Al on Kapton using Nomex honeycomb; 0.5% rad length

22 STAR TPC A Central Event Typically 1000 to 000 tracks per event into the TPC Two-track separation.5 cm Momentum Resolution < % Space point resolution ~ 500 µm Rapidity coverage 1.5 < η < 1.5

23 PID via de/dx with the STAR TPC 1 de/dx PID range: de/dx (kev/cm) µ e π K p d ~ 0.7 GeV/c for ~ 1.0 GeV/c for K/p Anti - 3 He Gas: P10 ( Ar-CH 4 90%-10% 1 atm

24 NA49 TPCs 158 GeV/nucleon

25 ALICE TPC gas volume 88 m 3 drift gas 90% Ne - 10%CO 560 cm 6x10 5 channels, corresponding to 6x10 8 pixels in space Field Cage Inner Vessel Central membrane frame

26 ALICE TPC: performance σ de/dx =10% N ch (-0.5<η<0.5)=8000

27 TPC: experimental issues mechanical tolerances (gain and electrical field) stability of high voltage power (gain) space charge effects (track distortion) gating efficiency (background) temperature, pressure (drift velocity) Drift velocity (cm/µs) examples from STAR Pressure (mbar) Drift Velocity Control: Lasers for coarse value Fine adjustment from tracking

28 de/dx Detector Performance σ ( de / dx) = 0.41n ( t p) calc Belle Type Drift ch. n 5 t (cm) 1.5 p (bar) 1 Gas He/C H 6 =50/50 Calc.(%) 6.6 Meas.(%) 5.1 Babar Drift ch He/C 4 H 10 =80/ CLEO II Drift ch Ar/C H 6 =50/ ALEPH* TPC Ar/CH 4 =90/ PEP* TPC Ar/CH 4 =80/ OPAL* Jet ch Ar/CH 4 /ic 4 H 10 =88./9.8/ MKII/SLC* Drift ch Ar/CO /CH 4 =89/10/ * Data from M. Hauschild (NIM A 379(1996) 436) Higher pressure gives better resolution, however, the relativistic rise saturate at lower βγ. 4 5 bar seems to be the optimal pressure Higher content of hydro-carbons gives better resolution (Belle and CLEO II). Landau distribution (FWMH); 60 % for noble gas, 45% for CH 4,33% for C 3 H 6