# Putting it all together - Particle Detectors

## Presentation on theme: "Putting it all together - Particle Detectors"— Presentation transcript:

Putting it all together - Particle Detectors
Writeup for 3rd section:

Measurements Non-Destructive Destructive
Particle only minimally perturbed Generally involves electrically charged particles depositing energy through many soft scatters Aim for low mass detector Destructive Initial particle absorbed or significantly scattered Detection generally by energy deposited by charged particles produced Can detect neutral particles

Types of Measurement Position Timing Velocity Energy (tracking)
Time of flight Event separation Velocity Cerenkov/Transition radiation Energy Total energy dE/dx

Position measurement All detectors give some indication of particle position ( even if it is only that the particle passed through the detector ) Most detectors have better resolution in one (or two) directions than the other two (or three). Hodoscope ~ cm (2D) Silicon strip detector ~5mm (1D) Silicon pixel detector ~5mm (2D) Photographic emulsion ~1mm (3D)

Position Measurement - Tracking
Measuring two (or more) points along the path of a particle allows its direction as well as is position to be measured. Measuring a number of points along the path of a particle allows any curvature to be measured.  Radius of curvature in a magnetic field gives the momentum

Position Measurement – Tracking
Pattern recognition can be tricky….

Timing Measurement The time at which a particle passed through a detector can be measured to better than 1ns (10-9s) Scintillator tends to be good ( 100ps ) Can measure velocity of particle “time-of-flight” (ToF ) from interaction to detector. Measuring b and p or E gives particle mass ( E=bm, p=bgm) and hence (usually) identity ToF only useful for fairly low energy particles ( “slightly relativistic” ) since highly relativistic particles all have b=1 within the bounds of error.

Timing Measurement Distinguish particles from different “events”
The interval between interactions generating the particles being measured is is often short. Need good timing resolution to separate tracks from different events. Measure start time for drift chambers. … and other devices that rely on measuring signal propagation times.

Timing Measurement – Particle ID

Timing Measurement – Particle ID

Timing Measurement – Particle ID
Hermes experiment uses TOF as one means of particle identification. Bunches of electrons hit fixed target. Measure time between collision and particles reaching scintillation detectors. m2 = (1/2 – 1) p2

Dead Time Most detectors take a finite time to produce a signal and recover before they can detect another. This is the dead-time Dead time varies with detector e.g. Si-strip detector ~ ns , Geiger-Muller tube ~ ms If the dead-time is Td and particles arrive at a mean rate of r per unit-time then probability that the detector is “dead” is ~ rTd I.e. efficiency is e = e0(1 – rTd )

Timing Coincidence Where a detector has a high background it is common to use two or more detectors in coincidence Output from combined detector only if all parts detect a particle. ( or 3 of 4, ….. etc.) If two detectors have a background rate of B1, B2 and a signal is produced if both detectors “fire” within the coincidence time, Dt then the background rate from the combined detector is B = B1 B2 Dt

Rate of energy loss – dE/dx Total energy - Calorimetry
Energy Measurement Rate of energy loss – dE/dx Total energy - Calorimetry

Energy Measurement – dE/dx
Measure the rate of energy loss of a charged particle through detector by ionization - dE/dx dE/dx Depends on bg ( particles with same bg but different masses give ~ same dE/dx ) Measuring bg and one of E,p, gives particle mass.

dE/dx Data Data from gaseous track detector.
dE/dx (keV/cm) Data from gaseous track detector. Each point from a single particle Several energy loss samples for each point “Averaged” to get energy loss Fluctuations easily seen p K p e m p (GeV/c)

Energy Measurement – Calorimetry
Measure total energy of a particle by stopping the particle in a medium and arranging for the energy to produce a detectable signal. This process is called calorimetry Detector needs to be thick enough to stop the particle Can measure energy of neutral particles using calorimetry

Energy Measurement – Regions of Applicability
Particle Energy Momentum by tracking (charged particles) Calorimetry keV - MeV Track length too short to measure curvature Absorb energy of initial particle 100’s MeV Good measurement of curvature Fractional error large due to fluctuation in Particle showers 100’s GeV Track too straight even with high B field and long path Fractional error small

Measuring Velocity Use a process such as Cerenkov radiation or transition radiation where the threshold/intensity of the radiation depends on the velocity of the particle Cerenkov radiation: angle and intensity are functions of b Transition radiation: intensity is a function of g (useful for highly relativistic particles) dE/dx by ionization ( already mentioned)

Sources of measurement error
Fluctuations of underlying physical processes “Statistical” fluctuations of numbers of quanta or interactions Variation in the gain process Noise from electronics etc.

Fluctuation in dE/dx by Ionization

Fluctuation in dE/dx by Ionization
Up to now we have discussed the mean energy lost by a charged particle due to ionization. The actual energy lost by a particular particle will not in general be the same as the mean. dE/dx due to a large number of random interactions Distribution is not Gaussian.

Fluctuation in dE/dx by Ionization
Distribution of dE/dx usually called the “Landau Distribution”

Fluctuation in dE/dx – Gaussian Peak
Most interactions involve little energy exchange and there are many of them. The total energy loss from these interactions is a Gaussian (central limit theorem)

Fluctuation in dE/dx – Gaussian Peak
For a Gaussian distribution resulting from N random events the ratio of the width/mean  1/N Increasing the thickness of the detector decreases the relative width of the Gaussian peak: (from Bethe)

Fluctuation in dE/dx – High Energy Tail
The probability of a interaction that involves a significant fraction of the particles energy is low. However such interactions produce a large signal in the medium.

Fluctuation in dE/dx – High Energy Tail
Energy loss is in the form of “d-rays” – scattered electrons with appreciable energy. Energy deposited in a thin detector can be different from the energy lost by the particle – the d-electron can have enough energy to leave the detector. Depending on the thickness of the detector there may not be any d-electrons produced.

dE/dx – High Energy Tail
Because of the high energy tail increasing the thickness of the detector does not improve the dE/dx resolution much. Relative width of Gaussian peak reduces, so would expect to get better estimate of mean dE/dx, but…. Probability of high energy interaction rises, so tail gets bigger. Usual method of measuring dE/dx is to take several samples and fit distribution (or just discard values far from Gaussian peak)

Multiple Scattering Deflection of a charged particle by large numbers of small angle scatters.

Multiple Scattering Looking at dE/dx from ionization, ignore nuclei.
Energy transfer small compared to scattering from (lighter) electrons. However, scattering from nuclei does change the direction of the particles momentum, if not its magnitude. Deflection of particle’s path limits the accuracy with which the curvature in a magnetic field can be determined, and hence the momentum measured.

“Single Scattering” Deflections are in random directions
“Drunkards Walk” Total deflection from N collisions  N The angular deflection caused by a single collision is well modelled by the Rutherford Scattering formula: ds/dW  1/q  ds/dq  1/q3 Most probable scatter is at small angle

Multiple Scattering RMS angular deflection, projected onto some plane:
RMS deflection  x Length scale is the radiation length X0

Multiple Scattering – Probability Distribution
Small scattering angles - many small scatters. Gaussian Large scattering angles from single large scatters. Probability  1/q3

Quantum Fluctuations A signal consists of a finite number of quanta (electrons, photons,….) If at some stage in detection chain the number of quanta drops to N then the relative fluctuation in the signal will be: NB. Any subsequent amplification of the signal will not reduce this relative fluctuation

Quantum Fluctuations – Poisson Distribution
If the number of quanta is small then the probability of producing m quanta when the average is n is: Probability of producing no signal: Efficiency of detector reduced by (1- e-n)

Quantum Fluctuations – Fano Factor
If the energy deposited by a particle is distributed between many different modes, only a small fraction of which give a detectable signal then the Poisson distribution is applicable. E.g. scintillation detector: small fraction of deposited energy goes into photons. Only few photons reach light detector.

Quantum Fluctuations: Fano Factor
If most of deposited energy goes into the signal then Poisson statistics are not applicable. E.g. Silicon detector – energy can either cause an electron-hole pair (signal detection and most likely process) or phonons. In this case the fractional standard deviation: F is the “Fano factor” (F ~ 0.12 for Si detector)

Electronic Noise Most modern detectors produce and electrical signal, which is then recorded. Electronic circuits produce noise – with careful design this can be minimized. Consider different sources of intrinsic noise: Johnson noise Shot noise Excess noise.

Johnson Noise Appears across and resistor due to random thermal motion of charge carriers. k : Boltzmanns constant T : Temperature above absolute zero B : Bandwidth ( range of frequency considered) White noise spectrum (same noise power per root Hz at all frequencies)

Shot Noise Fluctuation in the density of charge carriers ( “rain on a tin roof” ) White noise spectrum

Excess Noise Anything other than Johnson and shot noise.
Depends on details of electronic devices (e.g. transistors) Often has a 1/f spectrum ( same power per decade of frequency )

“Typical” Detector Front-End
Equivalent Circuit:

Noise: Dependence on Amplifier Capacitance.
The input resistance and capacitance of a detector “front end” form a low-pass filter which filters the Johnson noise from the input resistance:

Noise: Dependence on Amplifier Capacitance.
“Filtered” noise: Noise spectrum : Integrate over all frequencies to get total noise energy:

Noise: Dependence on Amplifier Capacitance.
Amplifier noise often expressed in terms of the number of electrons, DN, that would generate the same output. Q = CV = e DN Hence: Johnson noise increases with the input capacitance of the pre-amplifier.

Overall Statistical Error
Depends on detector and the quantity measured, but… For quantity like dE/dx which is estimated from the signal size: S = A E S=measured signal E=primary signal , A=amplification

Overall Statistical Error
First term is fluctuation in production of interaction process ( e.g. Landau distribution of –dE/dx) .

Overall Statistical Error
Signal is made up of a number of quanta ( electrons, photons, ions, … ). Second terms comes from the fluctuation in the number of quanta, ns , ( F is the Fano factor) .

Overall Statistical Error
In general, not all the quanta in the signal are collected – there is a “statistical bottle-neck” where the number of quanta, nh , is a minimum. The contribution to the error due from this bottleneck is approximated by the third term: .

Overall Statistical Error
Many detectors have an amplification stage (e.g. drift chambers have gain due to avalanche near the anode wire) The gain process will have some fluctuation, represented by the fourth term Each quanta produces on average A quanta after amplification. .

Overall Statistical Error
There is a contribution to the uncertainty in the signal from the noise in the readout electronics. The noise tends to have the same amplitude regardless of the size of the signal, so contributes to ss/S like D/S Described by the fifth term. .

Energy Measurement Fluctuations
As already remarked: Precision of momentum measurement (tracking) deteriorates at large momentum Energy measurement precision (calorimeter) generally improves as energy increases. Response of PbWO4 calo to 120GeV e-

Energy Measurement fluctuations
Statistical fluctuations (n.quanta)1/2 Contributions from noise ~ constant “systematics”  signal (Fluctuations much smaller for EM than hadronic showers)

The “general purpose” detector

The “General Purpose” Detector
Often a detector has to cope with many different types of particle of many different energies.  Construct a system of detectors allowing measurement of different aspects of different particles.

The “General Purpose” Detector
Typically a general purpose detector will have three main parts: Tracking (charged particles, magnetic field) Calorimeter (electrons, photons, hadrons) Muon tracking (generally only muons get this far)

General Purpose Detector: Photons
Tracking - will generally cross the tracking detector without leaving a signal. Desirable – don’t want to scatter the photon or convert to charged particles. minimize material. But, some will pair convert. Calorimeter -will produce an EM shower. Length scale X0. Contained in EM portion of calo. Muon tracking – won’t reach

General Purpose Detector: Electrons
Tracking – will leave a trail of ionization. Measure curvature to measure momentum. Some will undergo Bremsstrahlung. Calorimeter -will produce an EM shower. Same as for photons. Muon tracking – won’t reach

Tracking – charged hadrons will leave a trail of ionization. Calorimeter -will produce an hadronic shower. Length scale l0 Energy in both EM and hadronic parts of calo. Muon tracking – won’t reach

General Purpose Detector: Muons
Tracking –will leave a trail of ionization. Bremsstrahlung not a problem. Calorimeter – X0 for muons so long that no shower takes place. Still deposits energy by ionization. Muon tracking – crosses, leaving track of ionization

General Purpose Detector: Tau, B-mesons, D-mesons
Tracking – Decay close to interaction point. If daughters are charged may be able to reconstruct decay vertex. Calorimeter, Muon tracking- primary particle never reaches, but daughters may.

CMS (Compact Muon Solenoid)

CMS Cryostat Vacuum Tank

CMS – Transverse Slice

CMS “Event Display”

Zeus (800GeV p, 30GeV e+)

Zeus (800GeV p, 30GeV e+)

ZEUS “ Compensating Calorimeter”
Response to hadrons and electrons of equal energy is not the same ( for hadrons energy lost in nuclear binding energy and nuclear fragments

Compensating Calorimeter
Can produce e/h ~ 1 by making absorber out of Uranium – hadronic shower induces fission, and emission of gamma-rays which deposit energy “compensating” for loss in binding energy etc. ( ZEUS calo) Can also compensate by having fine- grained calorimeter, and trying to separate out EM and hadronic parts of shower ( e.g. H1 liquid argon calo )

Compensating Calo

Scintillation detectors
Produce visible light Transport to a light detector Total internal reflection Wavelength shifting fibres. Convert to an electrical signal S c i n t l a o r P e Total Internal Reflection L g h u d Light Detector

Total Internal Reflection
A ray of light is incident on a boundary between two refractive indices is deflected. If the angle of incidence, qi , is greater than the “critical angle” , qc , the light is totally internally reflected. Sin(qc) = n2/n1

Total Internal Reflection – Fraction of Light Trapped
Estimate the fraction of light trapped by TIR by integrating over the solid angle E.g. light trapped in a scintillating fibre:

TIR – Fraction of Light Trapped
Fraction trapped , f = (solid-angle, qi>qc)/(total solid-angle) Put q = p- qi dW= df d(Cosq) = Sinq df dq

Light Detectors Typically only get a few photons at light detector due to passage of particle  Need a detector sensitive at the single-photon level. Photomultiplier tube Avalanche photo-diode Hybrid photodiode

Photomultiplier Tube Light falls on a photocathode in an evacuated tube and electrons emitted (photoelectric effect) Quantum Efficiency depends on cathode material and wavelength ( QE ~ 25% ) Photoelectrons focused and accelerated towards the first dynode by electric field.

Photomultiplier Tube When photoelectron strikes dynode several electrons emitted (on average) n ~ 5 Several dynodes ( ~ 10 ) give high gain ( 107) PMT sensitive to magnetic field – need screening in many applications

Photodiode If a photon falls on a semiconductor an electron/hole pair can be created if the photon energy is greater than the band-gap  photodiode.

Avalanche Photodiode Light output from scintillator normally too low to allow the use of photodiodes No gain  output signal lost in noise of readout. Increase bias to a point where electrons/holes collide with lattice with sufficient energy to generate new electron/hole pairs  avalanche photodiode (APD)

Avalanche Photodiode Gain ~ 100 in linear mode ( can be operated in “Geiger Muller” mode) Compact Low sensitivity to magnetic field

Hybrid Photodiode Like photomultiplier tube, has a photocathode in an evacuated envelope Photoelectrons accelerated towards a reverse- biased solid-state diode ( e.g Si)

Hybrid Photodiode When accelerated photoelectron hits diode ( ~ kV ) it liberates several electron-hole pairs. Energy for one electron-hole pair in Si ~ 3.6eV Gain ~ 1000 Can also use avalanche photodiode to get extra gain Less sensitive to magnetic field than PMT Can have bigger light sensitive area than APD Can divide diode in to “pixels” to get position of photons

Hybrid Photodiode Used in e.g. readout of CMS HCAL
Wavelength shifting fibres used to couple light from scintillating sheets.

Scintillating Materials

Scintillating Materials
Emit light when excited by passage of charged particle To be useful should be transparent to the light they produce. Two types ( more or less ): Organic (work at molecular level) Inorganic (work at crystal level)

Organic Scintillators
Scintillation is a property of the individual organic molecules: Cartoon of molecular energy levels

Organic Scintillators-Light Emission
Passage of charged particle excites molecule. Can decay radiatively with photon energy , Eemission = EB1 – EB0 B0 rapidly decays to A0 by exchanging vibrational quanta with surroundings

Organic Scintillators-Light Absoption
Scintillator will absorb light – molecule state A0A1 ( atomic spacing doesn’t have time to change ) Photon energy Eabsorption = EA1 – EA0 Eabsorption> Eemission Emission and absorption spectra not the same.  Scintillator transparent to the light it produces (but usually put in a wavelength-shifter to move out of UV) Bicron BCF-91A Plastic wavelength shifter

Organic Scintillator Plastic common. E.g. polystyrene doped with fluorescent molecules which shift the emission from UV to visible. Can be in solid or liquid form. Atomic number, Z, low. Density low. Good or bad, depending on application Fraction of energy converted to light is lower than for inorganic scintillators ~ 10 photons per keV deposited ( ~ 1% of energy deposited, or about 10,000 photons/cm for MIP)

Inorganic Scintillators
Depend on properties of crystal. Interaction of atoms in lattice broaden energy levels of individual atoms into bands. In an insulator, valence band is full , conduction band is empty. Electrons “locked into position”, (no available energy states) If promoted to conduction band, electrons are free to move Valence Conduction Impurity levels

Inorganic Scintillators
If promoted to conduction band electrons will move through lattice until trapped by an impurity/defect in the lattice or a deliberately introduced dopant

Inorganic Scintillators
For some traps, the electron decays by emitting a photon (scintillaton) Electron decays from some traps without emitting light (quenching) Efficiency ( ~ 10% ) higher than for organic scintillators. Often high Z (low X0) – good for x-ray detection More expensive than organic.

Silicon diodes. Gas ionization chambers.
Position Detectors Silicon diodes. Gas ionization chambers.

Semiconductor Detectors
High ionisation density in solids – particularly semiconductors due to small band gap Small excitation energy  large thermal background p-n junction gives depletion region with few free charge carriers + -

Semiconductor detectors
Diffusion of holes from p-type into n-type Diffusion of electrons from n- type into p-type Results in charge separation.

Semiconductor detectors
Charge separation causes electric field which opposes further diffusion ( and sweeps free charge out of depletion layer)

Semiconductor Detectors
Depletion region widened by reverse bias voltage. Thickness ~ 100’s mm Ionisation in depletion layer collected on strips. In a 300 mm depletion layer will give ~ electron-hole pairs for a MIP Can “mass produce” pre-amplifiers with noise ~ 1000 electron-equivalent.

Semiconductor – Pulse Shape
View junction as a parallel plate capacitor. As ionization charge moves within depletion region, charge flows into the “plates” of the capacitor to maintain constant voltage. If charge, e, moves by distance dx then the charge dQ flows into the diode: Pulse shape determined by drift of charges in junction.

High Precision Silicon Vertex Detectors
This matrix of silicon microstrip detectors was at the heart of the ALEPH detector at LEP

Tracking With Precision Vertex Detectors
1 cm event observed in the Delphi Vertex detector

Tracking With Precision Vertex Detectors
1 cm event observed in the SLD detector

Movement of charges in gases
At modest E fields, electrons and positive ions drift at constant velocity Careful choice of gas to avoid absorbing electrons. If B field is present, drift is at an angle to the lines of E (Lorentz angle) Can use time for electrons to arrive at anode to get distance of track from wire. If electrons gain enough energy between collisions (ie. In one mean free path) multiplication of electron/ion pairs results ( needs large E )

Movement of charges in gases
Behaviour of electrons and ions in gas depends on details of gas mixture and electric (and magnetic) field Drift velocity, Lorentz velocity, gain Ions, being much heavier than electrons, have slower drift velocity and do not give amplification. Like, Si detectors, velocity of charges determines pulse shape. Fast part from electrons, slow “tail” from ions

Gas ionisation detectors
Thin (20–30m diameter) anode wires provide gain, due to electron avalanche in high field region near the wire ( E  1/r ) Many variants Geiger counter Multi-wire proportional chamber MWPC Drift chamber Time Projection Chamber TPC MicroStrip Gas Chamber MSGC

Liquid Ionization Detectors
Usually liquid noble gas – eg. Argon. No gain ( mean free path short, so electrons don’t get enough energy to cause further ionization) Use in calorimeters ( e.g. H1 , Atlas ) Don’t care about multiple scattering High density  many electron/ion pairs No gain  less danger of signal saturation

Cerenkov counters for particle identification
Reminder: Cerenkov angle and intensity depend on particle velocity and refractive index of the medium If momentum is known, measurements of  give b and hence particle mass and type Cerenkov detectors used for p/K separation at medium or high energies

Cerenkov detectors Cerenkov counters consist of radiator medium plus a photon detector Use of liquid, gas or aerogel radiators gives a range of refractive index Photon detectors usually either PM tubes or doped gas ionisation detector Different detector layouts: Threshold Cerenkov Differential Cerenkov Ring-imaging Cerenkov (RICH) Very large liquid filled detectors used for neutrino detection

Aerogel Silica based “foam”.
Tune refractive index by fraction of air ( ~ 99.8% air)

Super-K: Large Water Cerenkov detector for neutrinos

Threshold Cerenkov All charged particle above b threshold will give a signal. Adjust threshold by adjusting refractive index Gas for high threshold (has low , adjust with pressure) Liquid for low threshold (high )

Differential Cerenkov
Use circular annular collimator slit – only accepts light at a range of angles Only get signal if particle in right b range. Only works if particles on-axis Useful for “tagging” particles in a low intensity beam,

Ring-imaging Cerenkov

The DELPHI Detector DELPHI at LEP features extensive particle identification capability from its TPC and RICH counters

Conclusion: Aims To introduce the interactions of fast particles and high-energy photons in materials, particularly those types of interaction which are important for particle detection and measurement.

Conclusion: “Learning outcomes”
Understand those properties of stable and long-lived particles important for their detection. Able to perform calculations of scattering kinematics and mean decay paths for relativistic particles.

Conclusion: “Learning outcomes”
Understand the variation of ionisation energy loss for charged particles as a function of velocity, as given by the Bethe-Bloch formula. Appreciate the physical origin of the various terms in this formula. Able to describe the underlying physics of other important energy-loss processes.

Conclusion: “Learning outcomes”
Understand the operation of certain types of detector. Able to analyse the effect of counting fluctuations on the performance of detectors. Understand the response of detectors to different particle types. Know the design elements of a "general purpose" particle detector.

Similar presentations