Physics of Particle Detection

Claus Grupen
Department of Physics, University of Siegen, D-57068 Siegen,
Germany
e-mail: [email protected]

ABSTRACT
In this review the basic interaction mechanisms of charged and neutral particles
are presented. The ionization energy loss of charged particles is fundamental to
most particle detectors and is therefore described in more detail. The production
of electromagnetic radiation in various spectral ranges leads to the detection of
charged particles in scintillation, Cherenkov, and transition-radiation counters.
Photons are measured via the photoelectric effect, Compton scattering, or pair
production, and neutrons through their nuclear interactions.
A combination of the various detection methods helps to identify elemen-
tary particles and nuclei. At high energies, absorption techniques in calorimeters
provide additional particle identification and an accurate energy measurement.

1 Introduction

The detection and identification of elementary particles and nuclei is of

particular importance in high-energy, cosmic-ray, and nuclear physics [1, 2, 3,
4, 5, 6, 7]. Identification means that the mass of the particle and its charge
are determined. In elementary particle physics most particles have unit charge.
But in the study, e.g., of the chemical composition of primary cosmic rays or at
heavy-ion colliders, different charges must be distinguished.
Every effect of particles or radiation can be used as a working principle
for a particle detector.
The deflection of a charged particle in a magnetic field determines its
momentum p; the radius of curvature ρ is given by
p p γm0 βc
ρ= ∝ = , (1)
zeB z z
where z is the particle’s charge, m0 its rest mass, β = vc its velocity, and B
the magnetic bending field. The particle velocity can be determined, e.g., by a
time–of–flight method yielding
1
β∝ , (2)
τ
2 Claus Grupen

where τ is the flight time. A calorimetric measurement provides a determination

of the kinetic energy
E kin = (γ − 1)m0 c2 , (3)
1
where γ = √ is the Lorentz factor
1−β 2
From these measurements the ratio of m0 /z can be inferred, i.e., for singly
charged particles we have already identified the particle. To determine the
charge one needs another z-sensitive effect, e.g., the ionization energy loss

dE z2
∝ 2 ln(aβγ) (4)
dx β

(a is a material-dependent constant).
Now we know m0 and z separately. In this way, even different isotopes of
elements can be distinguished.
The basic principle of particle detection is that every physics effect can
be used as an idea to build a detector. In the following we distinguish between
the interaction of charged and neutral particles. In most cases the observed
signature of a particle is its ionization, where the liberated charge can be col-
lected and amplified, or its production of electromagnetic radiation which can
be converted into a detectable signal. In this sense, neutral particles are only
detected indirectly, because they must first produce, in some kind of interaction,
a charged particle which is then measured in the usual way.

2 Interaction of Charged Particles

Kinematics
Four-momentum conservation allows to calculate the maximum energy
transfer of a particle of mass m0 and velocity v = βc to an electron initially at
rest to be [2]

max 2me c2 β 2 γ 2 2me p2

Ekin =  2 = 2 , (5)
me me m0 + m2e + 2me E/c2
1 + 2γ m0
+ m 0

here γ = mE0 c2 is the Lorentz factor, E the total energy, and p the momentum
of the particle.
me me
For low-energy particles heavier than the electron (2γ m 0
 1; m 0
 1)
Equation (5) reduces to
max
Ekin = 2me c2 β 2 γ 2 . (6)
Physics of Particle Detection 3

For relativistic particles (Ekin ≈ E; pc ≈ E) one gets

E2
E max = . (7)
E + m20 c2 /2me

For example, in a µ–e collision the maximum transferable energy is

E2
E max = , E in GeV , (8)
E + 11
showing that in the extreme relativistic case the complete energy can be trans-
ferred to the electron.
If m0 = me , Equation (5) is modified to

max p2 E 2 − m2e c4
Ekin = 2
= = E − me c2 . (9)
me + E/c E + me c2

Scattering
Rutherford scattering
The scattering of a particle of charge z on a target of nuclear charge Z is
mediated by the electromagnetic interaction (Figure 1).

Figure 1: Kinematics of Coulomb scattering of a particle of charge z on a target of charge

Z.

The Coulomb force between the incoming particle and the target is written
as
z · e · Z · e ~r
F~ = . (10)
r2 r
4 Claus Grupen

For symmetry reasons the net momentum transfer is only perpendicular to p~

along the impact parameter b direction,
Z +∞ Z +∞
z · Z · e2 b dx
pb = Fb dt = · · , (11)
−∞ −∞ r2 r βc

with b = r sin ϕ, dt = dx/v = dx/βc, and Fb force perpendicular to p.

z · Z · e2 +∞ z · Z · e2 +∞
Z Z
b dx d(x/b)
pb = √ =  (12)
,
βc 2 2 3 βcb
3
q
−∞ ( x + b ) −∞ x 2
1+ b
| {z }
=2
2z · Z · e2 2re me c
pb = = z·Z , (13)
βcb bβ
where re is the classical electron radius. This consideration leads to a scattering
angle
pb 2z · Z · e2 1
Θ= = · . (14)
p βcb p
The cross section for this process is given by the well-known Rutherford formula
2
z 2 Z 2 re2 me c

dσ 1
= 4 . (15)
dΩ 4 βp sin Θ/2
Figure 2 shows the results of scattering α particles on gold foils [8, 9].

Multiple scattering
From Equation (15) one can see that the average scattering angle hΘi is
zero. To characterize the different degrees of scattering when a particle passes
through an absorber one normally uses the so-called “average scattering angle”
p
hΘ2 i. The projected angular distribution of scattering angles in this sense
leads to an average scattering angle of [7]
r   
p 13.6 MeV x x
hΘ2 i = Θplane = z· 1 + 0.038 ln (16)
βcp X0 X0
with p in MeV/c and x the thickness of the scattering medium measured in
radiation lengths X0 (see bremsstrahlung page 30). The average scattering
angle in three dimensions is
√ √
Θspace = 2 Θplane = 2 Θ0 . (17)
Physics of Particle Detection 5

Figure 2: Scattering of MeV α particles on gold foils.

The projected angular distribution of scattering angles can approximately be

represented by a Gaussian

Θ2
 
1
P (Θ) dΘ = √ exp − 2 dΘ . (18)
2πΘ0 2Θ0

Figure 3 shows the results of scattering 15.7 MeV electrons off gold foils [10, 11].
For low scattering angles (≤ 5◦ ) multiple scattering dominates. The distribution
develops a tail for large scattering angles due to single scattering events.

Energy loss of charged particles

Charged particles interact with a medium via electromagnetic interactions
by the exchange of photons. If the range of photons is short, the absorption
of virtual photons constituting the field of the charged particle gives rise to
ionization of the material. If the medium is transparent, Cherenkov radiation
can be emitted above a certain threshold. But also sub-threshold emission of
electromagnetic radiation can occur, if discontinuities of the dielectric constant
of the material are present (transition radiation) [12]. The emission of real
6 Claus Grupen

-1
10

Fractional Scattering per Square Degree

-2
10

-3
10

-4
10

-5
10

-6
10

-7
10
0º 5º 10º 15º 20º 25º 30º

Scattering angle

Figure 3: Scattering-angle distribution for 15.7 MeV electrons on two gold targets of different
thickness [10, 11].

photons by decelerating a charged particle in a Coulomb field also constitutes

an important energy loss process (bremsstrahlung).

Ionization energy loss

Bethe–Bloch formula
This energy-loss mechanism represents the scattering of charged particles
off atomic electrons, e.g.,
µ+ + atom → µ+ + atom+ + e− . (19)
The momentum transfer to the electron is (see Equation (13))
2re me c
pb = z
bβ
and the energy transfer in the classical approximation
p2b 2r2 me c2
ε= = e2 2 z 2 . (20)
2me b β
The interaction probability per (g/cm2 ), given the atomic cross section σ, is
N
φ[g−1 cm2 ] = σ [cm2 /atom] , (21)
A
Physics of Particle Detection 7

Figure 4: Sketch explaining the differential collision probability.

where N is Avogadro’s constant.

The differential probability to hit an electron in the area of an annulus
with radii b and b + db (see Figure 4) with an energy transfer between ε and
ε + dε is

N
φ(ε) dε = 2πb db Z , (22)
A

because there are Z electrons per target atom.

Inserting b from Equation (20) into Equation (22) gives

2re2 me c2 2 1
b2 = z ·
β2 ε
2 2
2re me c 2 dε
2|b db| = z · 2
β2 ε
N 2re2 me c2 2 dε
φ(ε) dε = π 2
z ·Z · 2
A β ε
2πre2 me c2 N Z 2 dε
= · ·z · 2 , (23)
β2 A ε

showing that the energy spectrum of δ electrons or knock-on electrons follows

an 1/ε2 dependence (Figure 5, [14]).
The energy loss is now computed from Equation (22) by integrating over
8 Claus Grupen

Differential Energy Tranfer Probality

-3 100 GeV muons
10 in iron

per [g/cm 2 GeV and muon]

-5
10

1 2 dependence
-7
10

Kinematical
-9
10 cut-off

0,1 1 10 100
Energy Transfer [GeV]

Figure 5: 1/ε2 dependence of the knock-on electron production probability [14].

all possible impact parameters [6],

Z ∞
−dE = φ(ε) · ε · dx
0
Z ∞
N
= 2πb · db · Z · ε · dx
0 A
Z ∞
dE 2πN
− = ·Z ε · b · db
dx A 0
Z · N 2re2 me c2 2 ∞ db
Z
= 2π · z . (24)
A β2 0 b
This classical calculation yields an integral which diverges for b = 0 as well
as for b = ∞. This is not a surprise because one would not expect that our
approximations hold for these extremes.
a) The b = 0 case: Let us approximate the “size” of the target electron
seen from the rest frame of the incident particle by half the de Broglie
wavelength. This gives a minimum impact parameter of
h h
bmin = = . (25)
2p 2γme βc
b) The b = ∞ case: If the revolution time τR of the electron in the target
atom becomes smaller than the interaction time τi , the incident particle
“sees” a more or less neutral atom,
bmax p
τi = 1 − β2 . (26)
v
Physics of Particle Detection 9

p
The factor 1 − β 2 takes into account that the field at high velocities is Lorentz
contracted. Hence the interaction time is shorter. For the revolution time we
have
1 h
τR = = , (27)
νZ · Z I
where I is the mean excitation energy of the target material, which can be
approximated by
I = 10 [eV] · Z (28)
for elements heavier than sulphur.
The condition to see the target as neutral now leads to

bmax p h
τR = τi ⇒ 1 − β2 =
v I
γhβc
bmax = . (29)
I
With the help of Equations (25) and (29) we can solve the integral in Equa-
tion (24)
dE Z 2r2 me c2 2γ 2 β 2 me c2
− = 2π · N · e 2 z 2 · ln . (30)
dx A β I
Since for long-distance interactions the Coulomb field is screened by the inter-
vening matter one has

2γ 2 β 2 me c2
 
dE 2 Z 1
− = κz · ln −η , (31)
dx A β2 I

where η is a screening parameter (density parameter) and

κ = 4πN re2 me c2 .

The exact treatment of the ionization energy loss of heavy particles leads to [7]

Z 1 1 2me c2 γ 2 β 2 Ekin max

 
dE δ
− = κz 2 · · 2 ln − β 2
− (32)
dx A β 2 I2 2

which reduces to Equation (31) for γme /m0  1 and β 2 + 2δ = η.

The energy-loss rate of muons in iron is shown in Figure 6 [7]. It exhibits
a β12 decrease until a minimum of ionization is obtained for 3 ≤ βγ ≤ 4.
Due to the ln γ term the energy loss increases again (relativistic rise, log-
arithmic rise) until a plateau is reached (density effect, Fermi plateau).
10 Claus Grupen

Figure 6: Energy loss of muons in iron [7].

The energy loss is usually expressed in terms of the area density ds = ρ dx

with ρ – density of the absorber. It varies with the target material like Z/A
(≤ 0.5 for most elements). Minimum-ionizing particles lose 1.94 MeV/(g/cm2 )
in helium decreasing to 1.08 MeV/(g/cm2 ) in uranium. The energy loss of
minimum-ionizing particles in hydrogen is exceptionally large, because here
Z/A = 1.
The relativistic rise saturates at high energies because the medium be-
comes polarized, effectively reducing the influence of distant collisions. The
density correction δ/2 can be described by
δ ~ωp 1
= ln + ln βγ − , (33)
2 I 2
where p
~ωp = 4πNe re3 me c2 /α (34)
is the plasma energy and Ne the electron density of the absorbing material.
For gases the Fermi plateau, which saturates the relativistic rise, is about
60% higher compared to the minimum of ionization. Figure 7 shows the mea-
sured energy-loss rates of electrons, muons, pions, kaons, protons, and deuterons
in the PEP4/9-TPC (185 dE/dx measurements at 8.5 atm in Ar/CH4 = 80 : 20)
[13]. Figure 8 shows the tracks of a 5 MeV proton and a 19 MeV α particle in an
optical avalanche microdosimeter [15]. The increase of the ionization towards
the end of the range is due to the 1/β 2 dependence of the energy loss. The
produced δ electrons show the same feature. The same behaviour is also seen
Physics of Particle Detection 11

Figure 7: Measured ionization energy loss of electrons, muons, pions, kaons, protons and
deuterons in the PEP4/9-TPC (Ar/CH4 = 80 : 20 at 8.5 atm) [13].

5 MeV proton 19 MeV

500 nm 500 nm

Figure 8: Tracks of a 5 MeV proton and a 19 MeV α particle in an optical avalanche micro-
dosimeter [15].
12 Claus Grupen

in a double-gem (gas electron multiplier) micro-strip gas chamber, where the

scintillation is measured using a high-resolution CCD camera (Figure 9, [16]).

Figure 9: Tracks of α particles in CF4 -based gas mixtures recorded via the scintillation in
a double-gem micro-strip gas chamber [16].

The z dependence of the ionization density can also clearly be seen from
the tracks of relativistic heavy ions in nuclear emulsions (Figure 10, [17]). The
strong enhancement of the ionization density at the end of the range, especially
for heavy ions, can be used for cancer therapy (Figure 11, [18]).

Figure 10: Tracks of relativistic heavy ions in nuclear emulsions [17].

Physics of Particle Detection 13

Figure 11: Ionization profiles of 12 C ions in water for different beam energies. The strong
ionization at the end of the range [Bragg peak] represents an ideal “scalpel” for the treatment
of deep-seated tumours [18].

Landau distributions
The Bethe–Bloch formula describes the average energy loss of charged par-
ticles. The fluctuation of the energy loss around the mean is described by an
asymmetric distribution, the Landau distribution [20, 21].
The probability φ(ε) dε that a singly charged particle loses an energy be-
tween ε and ε + dε per unit length of an absorber was (Equation (23))

2πN e4 Z 1
φ(ε) = · . (35)
me v 2 A ε 2

Let us define
2πN e4 Z
ξ= · x , (36)
me v 2 A
where x is the area density of the absorber:

1
φ(ε) = ξ(x) . (37)
xε2
Numerically one can write

0.1536 Z
ξ= · x [keV] , (38)
β2 A

where x is measured in mg/cm2 .

14 Claus Grupen

For an absorber of 1 cm Ar we have for β = 1

ξ = 0.123 [keV] .
We define now
1
ω(λ) f (x, ∆) = (39)
ξ
as the probability that the particle loses an energy ∆ on traversing an absorber
of thickness x. λ is defined to be the normalized deviation from the most
probable energy loss ∆m.p. ,
∆ − ∆m.p.
λ= . (40)
ξ
Namber of events

Pedestal
400

350

300 Landau distributed energy loss

250

200

150

100

50

0
0 1 2 3 4 5
Energy loss [keV/cm]

Figure 12: Energy-loss distribution of 3 GeV electrons in a thin-gap multi-wire drift

chamber [25].

The most probable energy loss is calculated to be [20, 22]

2me c2 β 2 γ 2 ξ
 
m.p. 2
∆ = ξ ln − β + 1 − γE , (41)
I2
Physics of Particle Detection 15

where γE = 0.577 . . . is Euler’s constant. The most probable energy loss in

argon for minimum-ionizing particles is ∆m.p. = 1.2 keV/cm while the average
energy loss amounts to 2.69 keV/cm. Landau’s treatment of f (x, ∆) yields

1 ∞ −u ln u−λu
Z
ω(λ) = e sin πu du , (42)
π 0

which can be approximated by [22]

 
1 1
Ω(λ) = √ exp − (λ + e−λ ) . (43)
2π 2

Figure 12 shows the energy-loss distribution of 3 GeV electrons in an Ar/CH4

(80 : 20) filled drift chamber of 0.5 cm thickness [25].
According to Equation (35) the δ-ray contribution to the energy loss falls
inversely proportional to the energy transfer squared, producing a long tail,
called Landau tail, in the energy-loss distribution up to the kinematical limit
(see also Figure 5).
The asymmetric property of the energy-loss distribution becomes obvious
for thin absorbers. For larger absorber thicknesses or truncation techniques
applied to thin absorbers the Landau distribution gets more symmetric. The
truncated Landau distribution for electrons, pions, kaons, and protons in various
momentum ranges are shown in Figure 13 [19].

Channeling
The energy loss of charged particles as described by the Bethe–Bloch formula
needs to be modified for crystals where the collision partners are arranged on a
regular lattice. By looking into a crystal (Figure 14), it becomes immediately
clear that the energy loss along certain crystal directions will be quite different
from that along a non-aligned direction or in an amorphous substance. The
motion along such channeling directions is governed mainly by coherent scat-
tering on strings and planes of atoms rather than by the individual scattering
off single atoms. This leads to anomalous energy losses of charged particles in
crystalline materials [23].
It is obvious from the crystal structure that charged particles can only be
channeled along a crystal direction if they are moving more or less parallel to
the crystal axis. The critical angle necessary for channeling is small (approx.
1◦ for β ≈ 0.1) and decreases with energy. For the axial direction h111i, it can
be estimated by
ψ [degrees] = 0.307 · (z · Z/E · d)0.5 (44)
16 Claus Grupen

4 4
10 10

3 3 K
10 10
e K e p
2 2
10 10
p

10 10

1 1

2 4 6 8 10 1 2 3
dE\dx dE\dx
P= 0.425 ® 0.45 [GeV/c] P = 0.70 ® 075 [GeV/c]

3
K
10
2
10
2 e
10 e
10
10

1 1

05 1 1.5 2 2.5 3 05 1 1.5 2 2.5 3

dE/dx dE/dx
P = 1.5 ® 1.6 [GeV/c] P = 3.2 ® 3.6 [GeV/c]

Figure 13: Truncated energy-loss distributions for electrons, pions, kaons, and protons in
the ALEPH time projection chamber [19].

Figure 14: View into a crystal lattice.

Physics of Particle Detection 17

where z and Z are the charges of the incident particle and the crystal, E the
particle’s energy in MeV, and d the interatomic spacing in Å. ψ is measured in
degrees [24]. For protons (z = 1) passing through a silicon crystal (Z = 14; d =

40000
600 Random
Aligned [110]

Counts Random
Counts Aligned

30000
400

20000

200
10000

0 0
0 200 400 600 800 1000

Figure 15: The energy-loss spectra for 15 GeV/c protons passing through a 740 µm thick
germanium crystal [24].

2.35 Å) the critical angle for channeling along the direction of body diagonals
becomes p
ψ = 13 µrad/ E [TeV] . (45)
For planar channeling along the face diagonals (h110i axis) in silicon one
gets p
ψ = 5 µrad/ E [TeV] . (46)
Of course, the channeling process also depends on the charge of the incident
particle.
In a silicon crystal the positive nuclear charges produce an electric field of
2·1012 V/cm at a distance of 0.1 Å from an individual silicon nucleus. This field,
however, decreases rapidly (like 1/r2 ) and therefore extends only over small dis-
tances. In contrast, for a string of silicon atoms along the h110i crystal direction
one obtains a field of 1.3 · 1010 V/cm. This field extends over macroscopic dis-
tances and can be used for the deflection of high-energy charged particles using
bent crystals [24].
Channeled positive particles are kept away from a string of atoms and
consequently suffer a relatively small energy loss. Figure 15 shows the energy-
loss spectra for 15 GeV/c protons passing through a 740 µm thick germanium
crystal [24]. The energy loss of channeled protons is lower by about a factor of
two compared to random directions through the crystal.
18 Claus Grupen

Scintillation in materials
Scintillator materials can be inorganic crystals, organic liquids or plastics,
and gases. The scintillation mechanism in organic crystals is an effect of the
lattice. Incident particles can transfer energy to the lattice by creating electron–
hole pairs or taking electrons to higher energy levels below the conduction band.
Recombination of electron–hole pairs may lead to the emission of light. Also
electron–hole bound states (excitons) moving through the lattice can emit light
when hitting an activator center and transferring their binding energy to acti-
vator levels, which subsequently deexcite. In thallium-doped NaI crystals about
25 eV are required to produce one scintillation photon. The decay time in inor-
ganic scintillators can be quite long (1 µs in CsI(Tl); 0.62 µs in BaF2 ).
In organic substances the scintillation mechanism is different. Certain
types of molecules will release a small fraction (≈ 3%) of the absorbed energy as
optical photons. This process is especially marked in organic substances which
contain aromatic rings, such as polystyrene, polyvinyltoluene, and naphtalene.
Liquids which scintillate include toluene or xylene [7].
This primary scintillation light is preferentially emitted in the UV range.
The absorption length for UV photons in the scintillation material is rather
short: the scintillator is not transparent for its own scintillation light. There-
fore, this light is transferred to a wavelength shifter which absorbs the UV light
and reemits it at longer wavelengths (e.g., in the green). Due to the lower con-
centration of the wavelength-shifter material the reemitted light can get out of
the scintillator and be detected by a photosensitive device. The technique of
wavelength shifting is also used to match the emitted light to the spectral sen-
sitivity of the photomultiplier. For plastic scintillators the primary scintillator
and wavelength shifter are mixed with an organic material to form a polymeriz-
ing structure. In liquid scintillators the two active components are mixed with
an organic base [2].
About 100 eV are required to produce one photon in an organic scintillator.
The decay time of the light signal in plastic scintillators is substantially shorter
compared to inorganic substances (e.g., 30 ns in naphtalene).
Because of the low light absorption in gases there is no need for wavelength
shifting in gas scintillators.
Plastic scintillators do not respond linearly to the energy-loss density. The
number of photons produced by charged particles is described by Birk’s semi-
empirical formula [7, 26, 27]
dE/dx
N = N0 , (47)
1 + kB dE/dx
where N0 is the photon yield at low specific ionization density, and kB is
Physics of Particle Detection 19

Birk’s density parameter. For 100 MeV protons in plastic scintillators one has
dE/dx ≈ 10 MeV/(g/cm2 ) and kB ≈ 5 mg/(cm2 MeV), yielding a saturation
effect of ≈ 5% [5].
For low energy losses Equation (47) leads to a linear dependence

N = N0 · dE/dx , (48)

while for very high dE/dx saturation occurs at

N = N0 /kB . (49)

There exists a correlation between the energy loss of a particle that goes into
the creation of electron–ion pairs or the production of scintillation light, be-
cause electron–ion pairs can recombine thus reducing the dE/dx|ion signal. On
the other hand the scintillation-light signal is enhanced because recombination
frequently leads to excited states which deexcite yielding scintillation light.

Cherenkov radiation
A charged particle traversing a medium with refractive index n with a
velocity v exceeding the velocity of light c/n in that medium, emits Cherenkov
radiation. The threshold condition is given by

vthres 1
βthres = ≥ . (50)
c n
The angle of emission increases with the velocity reaching a maximum value for
β = 1, namely
1
Θmax
C = arccos . (51)
n
The threshold velocity translates into a threshold energy

Ethres = γthres m0 c2 (52)

yielding
1 n
γthres = p 2
=√ . (53)
1 − βthres 2
n −1
The number of Cherenkov photons emitted per unit path length dx is
Z  
dN 2 1 dλ
= 2παz 1− 2 2 (54)
dx n β λ2
20 Claus Grupen

for n(λ) > 1, z – electric charge of the incident particle, λ – wavelength, and α
– fine-structure constant. The yield of Cherenkov-radiation photons is propor-
tional to 1/λ2 , but only for those wavelengths where the refractive index is larger
than unity. Since n(λ) ≈ 1 in the X-ray region, there is no X-ray Cherenkov
emission. Integrating Equation (54) over the visible spectrum (λ1 = 400 nm,
λ2 = 700 nm) gives
dN λ 2 − λ1
= 2παz 2 sin2 ΘC
dx λ1 λ2
= 490 · z 2 · sin2 ΘC [cm−1 ] . (55)

Figure 16: Pulse-height distribution for 3.5 GeV/c pions and protons in an aerogel
Cherenkov counter [28].

The Cherenkov effect can be used to identify particles of fixed momentum

by means of threshold Cherenkov counters. Figure 16 shows the pulse height
distribution for 3.5 GeV/c pions and protons in an aerogel Cherenkov counter
[28]. For an index of refraction of n = 1.015 the threshold Lorentz factor for
Cherenkov radiation is, according to Equation (53), γ = 5.84. This threshold
is exceeded for pions (γπ = 24.2) but not for protons (γp = 2.86). Therefore
protons deposit energy only due to ionization while pions produce in addition
Cherenkov light. More information can be obtained, if the Cherenkov angle is
measured by DIRC counters (Detection of Internally Reflected Cherenkov light).
In these devices some fraction of the Cherenkov light produced by a charged
particle is kept inside the radiator by total internal reflection. The direction of
the photons remains unchanged and the Cherenkov angle is conserved during
the transport. When exiting the radiator the photons produce a Cherenkov ring
Physics of Particle Detection 21

on a planar detector (Figure 17).

Figure 17: Imaging principle of a DIRC counter [29].

The pion/proton separation achieved with such a system is shown in Fig-

ure 18 [29].

Figure 18: Cherenkov-angle distribution for pions and protons of 5.4 GeV/c in a DIRC
counter [29].

Ring-imaging Cherenkov counters (RICH counters) have become extraor-

dinarily useful in the field of elementary particles and astrophysics. Figure 19
shows the Cherenkov-ring radii of electrons, muons, pions, and kaons in a
C4 F10 /Ar (75 : 25)-filled Rich counter read out by a 100-channel photomul-
22 Claus Grupen

Figure 19: Cherenkov-ring radii of e, µ, π, K in a C4 F10 /Ar (75 : 25) RICH counter. The
solid curves show the expected radii for an index of refraction of n = 1.00113. The shaded
regions represent a 5% uncertainty in the absolute momentum scale [30].

tiplier of 10 × 10 cm2 active area. The resulting π/µ/e separation for 3 GeV/c
particles is shown in Figure 20 [30].

160

140

120

100

80

60

40

20

0
0 2 4 6
Ringradius (cm)

Figure 20: Cherenkov-ring radii for pions, muons, and electrons of 3 GeV in a C4 F10 /Ar
(75 : 25) RICH counter [30].

An important aspect of neutrino physics with atmospheric neutrinos is

the correct identification of neutrino-induced muons and electrons. A deficit
of neutrino-induced muons would hint at neutrino oscillations, where the νµ
Physics of Particle Detection 23

Figure 21: Neutrino-induced muon with subsequent muon decay in the Super-Kamiokande
experiment [31].

Figure 22: Neutrino-induced muon in the SNO experiment [32].

24 Claus Grupen

Figure 23: Cherenkov ring produced by an electron from muon decay, where the muon was
created by a muon neutrino [32].

neutrinos born in the decay of pions (π + → µ+ + νµ ) or muons (µ− → e− +

ν̄e + νµ ) could have transformed themselves into τ neutrinos (ντ ) or some sterile
neutrino (similarly for ν̄µ ). Figure 21 shows a neutrino-induced event (νµ +
N → µ− + X) with subsequent decay of the µ− (µ− → e− + ν̄e + νµ ) in the
Super-Kamiokande experiment [31]. Both rings due to the µ− and e− are seen.
Figures 22 and 23 show a similar event in the SNO experiment [32]. In the
first frame the muon ring is visible (event time 22.1867771 sec) and 0.9 µs later
the electron ring. These events clearly demonstrate the particle identification
capability of large-volume neutrino detectors.

Transition radiation
Transition radiation is emitted when a charged particle traverses a medium
with discontinuous dielectric constant. A charged particle moving towards a
boundary, where the dielectric constant changes, can be considered to form
together with its mirror charge an electric dipole whose field strength varies in
time. The time-dependent dipole field causes the emission of electromagnetic
radiation. This emission can be understood in such a way that although the
dielectric displacement D ~ = εε0 E~ varies continuously in passing through a
boundary, the electric field does not.
Physics of Particle Detection 25

The energy radiated from a single boundary (transition from vacuum to a

medium with dielectric constant ε) is proportional to the Lorentz factor of the
incident charged particle [7, 26, 33]:

1 2
S= αz ~ωp γ , (56)
3
where ~ωp is the plasma energy (see Equation (34)). For commonly used plastic
radiators (styrene or similar materials) one has

~ωp ≈ 20 eV . (57)

The typical emission angle of transition radiation is proportional to 1/γ.

The radiation yield drops sharply for frequencies

ω > γωp . (58)

The γ dependence of the emitted energy originates mainly from the hardening of
the spectrum rather than from the increased photon yield. Since the radiated
photons also have energies proportional to the Lorentz factor of the incident
particle, the number of emitted transition-radiation photons is

Figure 24: Electron/muon separation in the NOMAD transition-radiation detector [34].

N ∝ αz 2 . (59)

The number of emitted photons can be increased by using many transitions

(stack of foils, or foam). At each interface the emission probability for an X-ray
photon is of the order of α = 1/137. However, the foils or foams have to be
26 Claus Grupen

of low-Z material to avoid absorption in the radiator. Interference effects for

radiation from transitions in periodic arrangements cause an effective threshold
behaviour at a value of γ ≈ 1000. These effects also produce a frequency-
dependent photon yield. The foil thickness must be comparable to or larger
than the formation zone
D = γc/ωp (60)
which in practical situations (~ωp = 20 eV; γ = 5 · 103 ) is about 50 µm. Tran-
sition radiation detectors are mainly used for e/π separation. In cosmic-ray
experiments transition-radiation emission can also be employed to measure the
energy of muons in the TeV range. Figure 24 [34] shows the muon/electron
discrimination power for 10 GeV particles in the NOMAD transition-radiation
detector. Muons fall below the TRD threshold and deposit energy loss due to
ionization only, while electrons of 10 GeV (γe = 20 000) clearly produce transi-
tion radiation.

Bremsstrahlung
If a charged particle is decelerated in the Coulomb field of a nucleus a fraction
of its kinetic energy will be emitted in form of real photons (bremsstrahlung).
The energy loss by bremsstrahlung for high energies can be described by [2]
dE Z2 2 2 183
− = 4αNA · z r E ln 1/3 , (61)
dx A Z
1 e2
where r = 4πε0 · mc2 . Bremsstrahlung is mainly produced by electrons because
1
re ∝ . (62)
me
Equation (61) can be rewritten for electrons,
dE E
− = , (63)
dx X0
where
A
X0 = (64)
4αNA Z(Z + 1)re2 ln(183 Z −1/3 )
is the radiation length of the absorber in which bremsstrahlung is produced.
Here we have included also radiation from electrons (∝ Z, because there are
Z electrons per nucleus). If screening effects are taken into account X0 can be
more accurately described by [7]
716.4 A 2
X0 = √ [g/cm ] . (65)
Z(Z + 1) ln(287/ Z)
Physics of Particle Detection 27

Figure 25: Muon bremsstrahlung event in the ALEPH detector [35].

Figure 25 [35] shows a bremsstrahlung event of a cosmic-ray muon in the electro-

magnetic calorimeter of the ALEPH experiment. The produced photon initiates
an electromagnetic shower and the electrons and positrons in the cascade are
deflected in opposite directions in the transverse magnetic field in the time
projection chamber.
The important point about bremsstrahlung is that the energy loss is pro-
portional to the energy. The energy where the losses due to ionization and
bremsstrahlung for electrons are the same is called critical energy
 
dEc  dEc 
= . (66)
dx ion dx brems

For solid or liquid absorbers the critical energy can be approximated by [7]
610 MeV
Ec = , (67)
Z + 1.24
while for gases one has [7]
710 MeV
Ec = . (68)
Z + 0.92
28 Claus Grupen

The difference between gases on the one hand and solids and liquids on the
other hand comes about because the density corrections are different in these
substances, and this modifies dE
dx ion .

The energy spectrum of bremsstrahlung photons is ∝ Eγ−1 , where Eγ is
the photon energy.
At high energies also radiation from heavier particles becomes important
and consequently a critical energy for these particles can be defined. Since

dE  1
∝ 2 (69)
dx brems m

the critical energy, e.g., for muons in iron is

 2
610 MeV mµ
Ec = · = 960 GeV . (70)
Z + 1.24 me

Direct electron pair production

Direct electron pair production in the Coulomb field of a nucleus via virtual
photons (“tridents”) is a dominant energy-loss mechanism at high energies. The
energy loss for singly charged particles due to this process can be represented
by 
dE 
− = b(Z, A, E) · E . (71)
dx pair
It is essentially – like bremsstrahlung – also proportional to the particle’s
energy. Because bremsstrahlung and direct pair production dominate at high
energies this offers an attractive possibility to build also muon calorimeters [2].
The average rate of muon energy losses can be parametrized as
dE
= a(E) + b(E) · E (72)
dx
where a(E) represents the ionization energy loss and b(E) is the sum of direct
electron pair production, bremsstrahlung, and photonuclear interactions.
The various contributions to the energy loss of muons in standard rock
3
(Z = 11; A = 22; ρ = 3 g/cm ) are shown in Figure 26.

Nuclear interactions
Nuclear interactions play an important role in the detection of neutral
particles other than photons. They are also responsible for the development
of hadronic cascades. The total cross section for nucleons is of the order of
Physics of Particle Detection 29

Figure 26: Contributions to the energy loss of muons in standard rock (Z = 11; A = 22;
ρ = 3 g/cm3 ).

50 mbarn and varies slightly with energy. It has an elastic (σel ) and inelastic
part (σinel ). The inelastic cross section has a material dependence

σinel ≈ σ0 Aα (73)

with α = 0.71. The corresponding absorption length λa is [2]

A
λa = [cm] (74)
NA · ρ · σinel

(A in g/mol, NA in mol−1 , ρ in g/cm3 , and σinel in cm2 ).

This quantity has to be distinguished from the nuclear interaction length
λw , which is related to the total cross section,
A
λw = [cm] . (75)
NA · ρ · σtotal
Since σtotal > σinel , λw < λa holds.
The total and elastic part of the proton–proton cross section are shown in
Figure 27
√ [7] as a function of the laboratory energy Elab and the center–of–mass
energy s, where s = 2 · m · Elab (for Elab  mproton ).
Strong interactions have a multiplicity which grows logarithmically with
energy. The particles are produced in a narrow cone around the forward di-
rection with an average transverse momentum of pT = 350 MeV/c, which is
responsible for the lateral spread of hadronic cascades.
30 Claus Grupen
14 37. Plots of cross sections and related quantities

200
pp
100
Cross section (mb)

50 ⇓ total

20

10 elastic

–1 2 3 4 5 6 7 8 9
10 1 10 10 10 10 10 10 10 10 10

1.9 2 10 100 10 3 10 4
Center of mass energy (GeV)

Figure 27: The proton–proton cross section as a function of the laboratory energy and
center–of–mass energy [7].

A useful relation for the calculation of interaction rates per (g/cm2 ) is

2
φ((g/cm )−1 ) = σN · NA , (76)

where σN is the cross section per nucleon and NA Avogadro’s number.

3 InteractionTotalofand Photons
Figure 37.19: elastic cross sections for pp and pp collisions as a function of laboratory beam momentum and total center-of-mass
energy. Corresponding computer-readable data les may be found at http://pdg.lbl.gov/xsect/contents.html (Courtesy of the COMPAS
Group, IHEP, Protvino, Russia, August 1999.)

Photons are attenuated in matter via the processes of the photoelectric

effect, Compton scattering, and pair production. The intensity of a photon
beam varies in matter according to

I = I0 e−µx , (77)

where µ is the mass attenuation coefficient. µ is related to the photon cross

sections σi by
3
NA X
µ= σi . (78)
A i=1
Physics of Particle Detection 31

Photoelectric effect
Atomic electrons can absorb the energy of a photon completely,

γ + atom → atom+ + e− . (79)

The cross section for absorption of a photon of energy Eγ is particularly large in

the K shell (80% of the total cross section). The total cross section for photon
absorption in the K shell is
 1/2
K 32
σPhoto = α4 Z 5 σThomson [cm2 /atom] , (80)
ε7

where ε = Eγ /me c2 , and σThomson = 38 πre2 = 665 mbarn is the cross section for
Thomson scattering. For high energies the energy dependence becomes softer,

K 1
σPhoto = 4πre2 Z 5 α4 · . (81)
ε
The photoelectric cross section has sharp discontinuities when Eγ coincides with
the binding energy of atomic shells. As a consequence of a photoabsorption in
the K shell characteristic X rays or Auger electrons are emitted [2].

Compton scattering
The Compton effect describes the scattering of photons off quasi-free atomic
electrons
γ + e → γ 0 + e0 . (82)
The cross section for this process, given by the Klein–Nishina formula, can be
approximated at high energies by

ln ε
σc ∝ ·Z , (83)
ε
where Z is the number of electrons in the target atom. From energy and mo-
mentum conservation one can derive the ratio of scattered (Eγ0 ) to incident
photon energy (Eγ ),
Eγ0 1
= , (84)
Eγ 1 + ε(1 − cos Θγ )
where Θγ is the scattering angle of the photon with respect to its original
direction.
32 Claus Grupen

For backscattering (Θγ = π) the energy transfer to the electron Ekin

reaches a maximum value
max 2ε2
Ekin = me c2 , (85)
1 + 2ε
which, in the extreme case (ε  1), equals Eγ .
In Compton scattering only a fraction of the photon energy is transferred
to the electron. Therefore, one defines an energy scattering cross section
Eγ0
σcs = σc (86)
Eγ
and an energy absorption cross section
Ekin
σca = σc − σcs = σc . (87)
Eγ
At accelerators and in astrophysics also the process of inverse Compton scat-
tering is of importance [2].

Pair production
The production of an electron–positron pair in the Coulomb field of a nucleus
requires a certain minimum energy
2m2e c2
Eγ ≥ 2me c2 + . (88)
mnucleus
Since for all practical cases mnucleus  me , one has effectively Eγ ≥ 2me c
2 .
The total cross section in the case of complete screening ε  αZ11/3 , i.e.,
at reasonably high energies (Eγ  20 MeV), is
 
7 183 1
σpair = 4αre2 Z 2 ln 1/3 − [cm2 /atom] . (89)
9 Z 54
Neglecting the small additive term 1/54 in Equation (89) one can rewrite, using
Equation (61) and Equation (64),
7 A 1
σpair = · . (90)
9 NA X0
The partition of the energy to the electron and positron is symmetric at low
energies (Eγ  50 MeV) and increasingly asymmetric at high energies (Eγ >
1 GeV) [2].
Figure 28 shows the photoproduction of an electron–positron pair in the
Coulomb field of an electron (γ +e− → e+ +e− +e− ) and also a pair production
in the field of a nucleus (γ + nucleus → e+ + e− + nucleus0 ) [36].
Physics of Particle Detection 33

Figure 28: Photoproduction in the Coulomb field of an electron (γ + e− → e+ + e− + e− ,

left) and on a nucleus (γ + nucleus → e+ + e− + nucleus0 , right) [36].

Figure 29: Mass attenuation coefficients for photon interactions in silicon [37].

Mass attenuation coefficients

The mass attenuation coefficients for photon interactions are shown in Fig-
ures. 29–31 for silicon, germanium, and lead [37]. The photoelectric effect
dominates at low energies (Eγ < 100 keV). Superimposed on the continuous
photoelectric attenuation coefficient are absorption edges characteristic of the
absorber material. Pair production dominates at high energies (> 10 MeV). In
the intermediate region Compton scattering prevails. A typical experiment
34 Claus Grupen

Figure 30: Mass attenuation coefficients for photon interactions in germanium [37].

Figure 31: Mass attenuation coefficients for photon interactions in lead [37].

setup for the measurement of γ-ray spectra is shown in Figure 32 [38]. Photons
from a radioactive source enter a NaI(Tl) scintillation crystal and deposit energy
either by the photoelectric effect, Compton scattering, or – depending on the
photon energy – pair production. The amount of light produced by the electron
is recorded in a photomultiplier. A typical γ-ray spectrum of the 662 keV γ line
Physics of Particle Detection 35

137 137
from the decay of Cs into Ba is shown in Figure 33 [39]. Clearly visible

Figure 32: Typical setup for the measurement of γ-ray spectra [38].

Figure 33: γ-ray spectrum of 662 keV photons in a NaI(Tl) scintillation counter [39].

is the photopeak (full absorption peak) at E = 662 keV. Compton scattering

produces a continuum with a maximum energy transfer to electrons at 478 keV
(see Equation (85)), the Compton edge. Photons which are backscattered from
shielding material into the detector may undergo photoelectric effect thereby
max
producing the backscatter peak at E − Ekin = 184 keV. At low energies char-
acteristic X-ray lines are visible originating from the source or the shielding
material. The spectra of γ-rays from a 60 Co source recorded in a NaI(Tl) detec-
tor [39] and in a high-purity germanium counter [40] are compared in Figure 34
and 35. All features of γ-ray spectra like photopeaks at 1.17 MeV and 1.33 MeV,
Compton edge, backscatter peaks, and characteristic X rays are clearly visible.
Also the outstanding performance of the high-purity Ge counter with a rms res-
olution of 400 eV shows the superiority of solid-state detectors over scintillation
36 Claus Grupen

Figure 34: γ-ray spectrum of 60 Co measured in a NaI(Tl) scintillation detector [39].

Figure 35: γ-ray spectrum of 60 Co measured in a high-purity Gedetector [40].

Physics of Particle Detection 37

counters.

Figure 36: Spectra of characteristic X rays measured in a superconducting phase-transition

thermometer [41].

Even higher resolutions are obtained in cryogenic detectors. Figure 36

[41] shows the energy resolution for characteristic X rays of Al, Ti, and Mn
in a superconducting phase-transition thermometer. The change of resistance
upon an energy absorption in the thermometer is transformed into a magnetic-
flux change which is read by a superconducting quantum interference device
(SQUID).

4 Interaction of Neutrons

In the same way as photons are detected via their interactions also neu-
trons have to be measured indirectly. Depending on the neutron energy various
reactions can be considered which produce charged particles which are then
detected via their ionization or scintillation [2].
38 Claus Grupen

a) Low energies (< 20 MeV)

n + 6 Li → α + 3H ,
n + 10 B → α + 7 Li ,
n + 3 He → p + 3H , (91)
n+p → n+p .
The conversion material can be a component of a scintillator (e.g., LiI
(Tl)), a thin layer of material in front of the sensitive volume of a gaseous
detector (boron layer), or an admixture to the counting gas of a propor-
tional counter (BF3 , 3 He, or protons in CH4 ).
b) Medium energies (20 MeV ≤ Ekin ≤ 1 GeV)
The (n, p)-recoil reaction can be used for neutron detection in detectors
which contain many quasi-free protons in their sensitive volume (e.g., hy-
drocarbons).
c) High energies (E > 1 GeV)
Neutrons of high-energy initiate hadron cascades in inelastic interactions
which are easy to identify in hadron calorimeters.
Neutrons are detected with relatively high efficiency at very low energies.
Therefore, it is often useful to slow down neutrons with substances containing
many protons, because neutrons can transfer a large amount of energy to colli-
sion partners of the same mass. In some fields of application, like in radiation
protection at nuclear reactors, it is of importance to know the energy of fis-
sion neutrons, because the relative biological effectiveness depends on it. This
can, e.g., be achieved with a stack of plastic detectors interleaved with foils of
materials with different threshold energies for neutron conversion [42].

5 Interactions of Neutrinos

Neutrinos are very difficult to detect. Depending on the neutrino flavor the
following inverse-beta-decay-like interactions can be considered:
νe + n → p + e− ,
ν̄e + p → n + e+ ,
νµ + n → p + µ− ,
ν̄µ + p → n + µ+ , (92)
ντ + n → p + τ− ,
ν̄τ + p → n + τ+ .
Physics of Particle Detection 39

Figure 37: Charged-current interaction of an energetic electron neutrino in a bubble chamber

[44].

The cross section for νe detection in the MeV range can be estimated as [43]
 2
4 ~p
σ(νe N ) = · 10−10
π (mp c)2
= 6.4 · 10−44 cm2 for 1 MeV . (93)

This means that the interaction probability of, e.g., solar neutrinos in a water
Cherenkov counter of d = 100 meter thickness is only

φ = σ · NA · d = 3.8 · 10−16 . (94)

Since the coupling constant of weak interactions has a dimension of 1/GeV2 , the
neutrino cross section must rise at high energies like the square of the center–
of–mass energy. For fixed-target experiments we can parametrize

σ(νµ N ) = 0.67 · 10−38 Eν [GeV] cm2 /nucleon ,

σ(ν̄µ N ) = 0.34 · 10−38 Eν [GeV] cm2 /nucleon . (95)

This shows that even at 100 GeV the neutrino cross section is lower by 11 orders
of magnitude compared to the total proton–proton cross section.
Figure 37 [44] shows the interaction of a high-energy electron neutrino in
a bubble chamber (νe + nucleon → e− + hadrons) producing an electron in a
40 Claus Grupen

charged-current interaction. The electron initiates an electromagnetic cascade

in the bubble chamber.

6 Electromagnetic Cascades

The development of cascades induced by electrons, positrons, or photons

is governed by bremsstrahlung of electrons and pair production of photons.
Secondary particle production continues until photons fall below the pair pro-
duction threshold, and energy losses of electrons other than bremsstrahlung
start to dominate: the number of shower particles decays exponentially.
Already a very simple model can describe the main features of particle
multiplication in electromagnetic cascades: a photon of energy E0 starts the
cascade by producing an e+ e− pair after one radiation length. Assuming that
the energy is shared symmetrically between the particles at each multiplication
step, one gets at the depth t (measured in radiation lengths X0 )

N (t) = 2t (96)

particles with energy

E(t) = E0 · 2−t . (97)
The multiplication continues until the electrons fall below the critical energy
Ec ,
Ec = E0 · 2−tmax . (98)
From then on (t > tmax ) the shower particles are only absorbed. The position
of the shower maximum is obtained from Equation (98),

ln E0 /Ec
tmax = ∝ ln E0 . (99)
ln 2
The total number of shower particles is
tX
max X
S = N (t) = 2t = 2tmax +1 − 1 ≈ 2tmax +1 ,
t=0
E0
S = 2 · 2tmax = 2 · ∝ E0 . (100)
Ec
If the shower particles are sampled in steps t measured in units of X0 , the total
track length is obtained as
S E0 1
S∗ = =2 · , (101)
t Ec t
Physics of Particle Detection 41

Figure 38: Some muon-induced electromagnetic cascades in a multi-plate cloud chamber

operated in a concrete-shielded air-shower array [45].

which leads to an energy resolution of

√ √ √
σ S∗ t t
= ∗ =p ∝√ . (102)
E0 S 2E0 /Ec E0
In a more realistic description the longitudinal development of the electron
shower can be approximated by [7]
dE
= const · ta · e−bt , (103)
dt
where a, b are fit parameters.
Figures 38 [45] and 39 [46] show muon-induced electromagnetic cascades
in a multi-plate cloud chamber.
The lateral spread of an electromagnetic shower is mainly caused by mul-
tiple scattering. It is described by the Molière radius
21 MeV 2
Rm = X0 [g/cm ] . (104)
Ec
95% of the shower energy in a homogeneous calorimeter is contained in a cylinder
of radius 2Rm around the shower axis.
Figure 40 demonstrates the interplay of the longitudinal and lateral de-
velopment of an electromagnetic shower [2].
42 Claus Grupen

Figure 39: Muon-induced electromagnetic cascade in a multi-plate cloud chamber [46].

7 Hadron Cascades

The longitudinal development of electromagnetic cascades is characterized

by the radiation length X0 and their lateral width is determined by multiple
scattering. In contrast to this, hadron showers are governed in their longitudinal
structure by the nuclear interaction length λ and by transverse momenta of
secondary particles as far as lateral width is concerned. Since for most materials
λ  X0 and hpinteraction
T i  hpmultiple
T
scattering
i hadron showers are longer and
wider.
Part of the energy of the incident hadron is spent to break up nuclear
bonds. This fraction of the energy is invisible in hadron calorimeters. Fur-
ther energy is lost by escaping particles like neutrinos and muons as a result
of hadron decays. Since the fraction of lost binding energy and escaping par-
ticles fluctuates considerably, the energy resolution of hadron calorimeters is
systematically inferior to electron calorimeters.
Fig 41 sketches the various fractions of energy in a hadronic cascade and
their variation with the hadron energy.
The longitudinal development of pion-induced hadron cascades is plotted
in Figure 42 [48]. Figure 43 shows a comparison between proton-, iron-, and
photon-induced cascades in the atmosphere [47].
Physics of Particle Detection 43

Figure 40: Sketch of the longitudinal and lateral development of an electromagnetic cascade
in a homogeneous absorber [2].

The different response of calorimeters to electrons and hadrons is an un-

desirable feature for the energy measurement of jets of unknown particle com-
position. By appropriate compensation techniques, however, the electron–to–
hadron response can be equalized.

50

40

30

20

10

Figure 41: Fractions of the total energy in a hadronic cascade that go into nuclear fragments,
binding energy, charged particles, and electromagnetic cascades in their variation with energy.
44 Claus Grupen

Figure 42: Longitudinal development of pion-induced hadron cascades [48].

8 Particle Identification

Particle identification is based on measurements which are sensitive to the

particle velocity, its charge, and its momentum. Figure 44 sketches the different
possibilities to separate photons, electrons, positrons, muons, charged pions,
protons, neutrons, and neutrinos in a mixed particle beam using a general-
purpose detector.
One particle identification technique that has not been discussed so far
in detail is the time–of–flight measurement (TOF). Two particles of different
velocity v1 and v2 passing through a telescope consisting of two scintillation
counters at a distance L will exhibit a time–of–flight difference
   
1 1 L 1 1
∆t = L · − = · − . (105)
v1 v2 c β1 β2
Replacing the pnormalized velocities βi = vi /c by the corresponding Lorentz
factors γi = 1/ 1 − βi2 in Equation (105) one gets
(s s )
L γ12 γ22
∆t = · − . (106)
c γ12 − 1 γ22 − 1
Physics of Particle Detection 45

Figure 43: Comparison between proton-, iron-, and photon-induced cascades in the atmo-
sphere. The primary energy in each case is 1014 eV [47].

For relativistic particles (E  m0 · c2 ), Equation (106) can be approximated by

(s 2 s 2 )
m1 · c2 m2 · c2
 
L
∆t = · 1+ − 1+ . (107)
c E1 E2

Since in the relativistic case E ≈ p · c, one gets for a momentum-defined beam

(E1 ≈ p1 · c, E2 ≈ p2 · c, p1 = p2 = p)
L·c
· m21 − m22


∆t = 2
(108)
2·p

Figure 44: Particle identification using a detector consisting of a tracking chamber,

Cherenkov counters, calorimetry, and muon chambers.
46 Claus Grupen

allowing the identification of particles with different mass. Figure 45 [49] shows
the particle identification power in a scintillator system using TOF and dE/dx
information in a momentum-defined beam containing electrons, muons, and
pions of 107.5 MeV/c.

Figure 45: e/µ/π separation in a momentum-defined beam (p = 107.5 MeV/c) using TOF
and dE/dx techniques [49].

Essential for an effective particle identification with TOF techniques is an

excellent time resolution. Figures 46 and 47 show the time resolution in a scin-
tillator TOF system and a multi-gap resistive plate chamber [28, 50, 51] allowing
for an efficient e/π/p separation. Figures 48 [52] and 49 [53] show the particle
separation power of a balloon-borne experiment using momentum, time–of–
flight, dE/dx, and Cherenkov-radiation measurements for singly charged parti-
cles at an altitude of 1234 m (ground level) and at flight altitudes
Physics of Particle Detection 47

Figure 46: Pion/proton separation in a 2 GeV/c beam with a scintillator TOF system. The
distance of the scintillation counters was 1.3 m [28].

Figure 47: Time resolution in a multi-gap resistive plate chamber [50, 51].
48 Claus Grupen

Velocity b

0.8

p
0.6

d
0.4

-3 -2 -1 0 1 2 3
Rigidity [GV]

Figure 48: Particle composition in a balloon-borne experiment at ground level (altitude

1234 m) [52]. Rigidity is defined as momentum divided by the charge of the particle.

1.2

1.0

p
0.8

0.6

0.4

p d t
z=1
–10 –5 0 5 10
p
Rigidity R = z [GV]

Figure 49: Particle identification in a balloon-borne experiment using momentum, time–of–

flight, dE/dx, and Cherenkov radiation information [53].
Physics of Particle Detection 49

200

energy loss dE/dx [arbitrary units] 160

120
Helium

80

Proton

40

0
2
0 0.2 0.4 0.6 0.8

Figure 50: Identification of proton and helium nuclei in primary cosmic rays in a balloon-
borne experiment [54].

1.2

1 3
He 1200
0.8
Tof (top-niddle)

500

0.6 200
4
He 100
0.4
50
0.2 20

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Rigidity [GV]

Figure 51: Isotopic abundance of energetic cosmic-ray helium nuclei [55].

Relatively easy is the proton/helium separation in such an experiment

(Figure 50 [54]).
Even the abundance of different helium isotopes can be determined from a
velocity and momentum measurement (Figure 51 [55]). This is feasible, because
at fixed momentum the lighter isotope 3 He is faster than the more abundant
50 Claus Grupen

Charge Separation with ISOMAX98

dE/dx top ToF [a.u.] 800

700
600

500
400 O
300 N
200
C
He B
100 Be
Li
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
bToF12

1200

1000 top ToF

0.18e
800

600 0.19e
0.12e 0.15e
400
0.13e 0.21e
200

0
2 3 4 5 6 7 8 9
charge

Figure 52: Identification of light elements in primary cosmic rays in a balloon experiment
[56].

4
He.

In very much the same way works the identification of the light elements
in primary cosmic rays (Figure 52 [56]) by using dE/dx and TOF techniques.
The elements from lithium up to oxygen can be resolved with high resolution.
A similar method can be used to extend the charge spectrum up to the iron
family (Figure 53 [57]).
Physics of Particle Detection 51

Figure 53: Chemical composition of primary cosmic rays as determined by range and dE/dx
measurements [57].

9 Conclusion

Basic physical principles can be used to identify all kinds of elementary

particles and nuclei. The precise measurement of the particle composition in
high-energy physics experiments at accelerators and in cosmic rays is essential
for the insight into the underlying physics processes. If a particle cannot be
directly identified – such as is the case for short-lived particles like Ks0 or Λ0
– their decay products can be measured and identified. Knowing their energy,
momentum, and mass, the four-vectors of the decay products can be combined
to form the invariant mass of the parent particle (Ks0 → π + +π − , Λ0 → p+π − ).
This technique has become very powerful especially for the tagging of very short-
lived mesons like B 0 or D0 in hadronic jets.
Neutrinos or other weakly interacting neutral particles are almost impos-
sible to identify directly. If, however, an event is fully contained in a detector
and the center–of–mass energy is known the direction of the missing momentum
can be associated with the direction of the weakly interacting neutral particle,
thereby identifying it. This technique has been used intensively, e.g., for the
reconstruction of the leptonic decays of the charged vector bosons W + and W − (
W + → e+ + νe , W + → µ+ + νµ ).

10 Acknowledgements

It is a pleasure to thank the organizers of the Instrumentation School for

their kind invitation and special hospitality. I would also like to thank Dipl.
52 Claus Grupen

Phys. Tilo Stroh and in particular Dipl. Phys. Anbang Ngac for their help in
preparing the manuscript.

References

[1] K. Kleinknecht, Detectors for Particle Radiation, Cambridge University

Press 1998
[2] C. Grupen, Particle Detectors, Cambridge University Press 1996
[3] R. Fernow, Introduction to Experimental Particle Physics, Cambridge Uni-
versity Press 1989
[4] D. Green, The Physics of Particle Detectors, Cambridge University Press
2001
[5] W. R. Leo, Techniques for Nuclear and Particle Physics Experiments,
Springer, Berlin 1987
[6] B. Rossi, High Energy Particles, Prentice-Hall (1952)
[7] Particle Data Group, R.M. Barnett et al., Phys. Rev. D54 (1996) 1; Eur.
Phys. J. C3 (1998) 1; Eur. Phys. J. C15 (2000) 1; Phys. Rev. D66 (2002) 1
[8] E. Rutherford, Phil. Mag. 21 (1911) 669
[9] H. Geiger, E. Marsden , Phil. Mag. 25 (1913) 604
[10] A. O. Hanson et al., Phys. Rev. 84 (1951) 634
[11] R. O. Birkhoff in Handbook of Physics XXXIV (1958)
[12] W.W.M. Allison, P.R.S. Wright, Oxford Univ. Preprint 35/83 (1983) and
W.W.M. Allison, J.H. Cobb, Ann. Rev. Nucl. Sci. 30 (1980) 253
[13] D. R. Nygren, J.N. Marx, Physics Today 31 (1978) 46
[14] C. Grupen, Ph.D. Thesis, University of Kiel 1970
[15] U. Titt et al., Nucl. Instr. Meth. A 416 (1998) 85
[16] F. A. F. Fraga et al., Nucl. Sci. Symp. and Medical Imaging Conf., Seattle,
Washington, Oct. 1999, to be publ. in IEEE Trans. Nucl. Sci.
[17] G. D. Rochester, Advancement of Science, Dec. 1970, 183–194
[18] G. Kraft, GSI-Preprint 96–60 (1996) and private communication; Nucl.
Phys. A 654 (1999) 1058c
[19] A. Ngac, Diploma Thesis, University of Siegen 2000
[20] D. H. Wilkinson, Nucl. Instr. Meth. A 383 (1996) 513
[21] L. D. Landau, J. Exp. Phys. (USSR) 8 (1944) 201
[22] S. Behrens, A.C. Melissinos, Univ. of Rochester Preprint UR-776 (1981)
[23] S. P. Möller, CERN 94–05 (1994)
[24] D. S. Gemmel, Rev. Mod. Phys. 46 (1974) 129
Physics of Particle Detection 53

[25] K. Affholderbach et al., Nucl. Instr. Meth. A 410 (1998) 166

[26] R. C. Fernow, Brookhaven Nat. Lab. Preprint BNL-42114 (1988)
[27] J. Birks, Theory and Practice of Scintillation Counting, MacMillan 1964
[28] A. Satpathy et al., BELLE Collaboration, hep-ex/9903045 (1999)
[29] I. Adam et al. SLAC-Pub-7706, Nov. 1997; and hep-ex/9712001
[30] R. Debbe et al. hep-ex/9503006 (1995)
[31] Super-Kamiokande home page
http://www.ps.uci.edu/~tomba/sk/tscan/pictures.html
[32] Sudbury Neutrino Observatory home page
http://www.sno.phy.queensu.ca/event/events.html
[33] S. Paul, CERN-PPE 91–199 (1991)
[34] NOMAD collaboration,
http://nomadinfo.cern.ch/Public/detect/trd/trd.html
G. Bassompierre et al. Nucl. Instr. Meth. A 403 (1998) 363–382
[35] ALEPH Collaboration (1998), unpublished
[36] F. Close, M. Marten, C. Sutton, The Particle Explosion, Oxford University
Press 1987
[37] Harshaw Chemical Company (H. Lentz, L. van Gelderen, Th. Courbois)
1969
[38] The University of Tennessee, Department of Physics and Astronomy,
http://electron5.phys.utk.edu/gamma/detector.htm
[39] Amptek Inc. Bedford MA, USA;
http://www.amptek.com/gammax.html
[40] J. Hausmann, University of Siegen, private communication (2001).
[41] CRESST Collaboration, Nucl. Instr. Meth. A 354 (1995) 408
[42] E. Sauter, Grundlagen des Strahlenschutzes, Thiemig, München 1982
[43] D. H. Perkins Introduction to High Energy Physics, Addison-Wesley, 1986
[44] H. Wachsmuth, CERN, private communication (1998)
[45] W. Wolter, private communication 1999
[46] G. D. Rochester, K.E. Turvev, Contemp. Phys. 22 (1981) 425
[47] J. Knapp, D. Heck Luftschauer-Simulationsrechnungen mit dem Corsika-
Programm, Forschungszentrum Karlsruhe Nachrichten 30 (1998) 27
[48] M. Holder et al., Nucl. Instr. Meth. 151 69 (1978)
[49] E. Fragiacomo et al., Nucl. Instr. Meth. A 439 (2000) 45
[50] F. Sauli, CERN-EP/2000-080
54 Claus Grupen

[51] ALICE technical design report, CERN/LHCC 2000-12 (2000)

[52] J. Kremer, Ph.D. Thesis, University of Siegen 1999
[53] J. W. Mitchell et al., Phys. Rev. Lett. 76 (1996) 3057
[54] O. Reimer, Ph.D. Thesis, University of Siegen 1995
J. W. Mitchell et al., Proc. Cosmic Ray Conf. Rome 1995
[55] M. Simon, H. Göbel, private communication 1999
[56] H. Göbel, Ph.D. Thesis, University of Siegen 2001
[57] A. Hesse, Proc. 22nd Int. Cosmic Ray Conf. Dublin Volume 1 (1991) 596