You are on page 1of 72

28.

Particle detectors 1
28. PARTICLE DETECTORS . . . . . . . . . . . . . . . . . . . . . . . 2
28.1. Summary of detector spatial resolution, temporal resolution,
and deadtime . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
28.2. Photon detectors . . . . . . . . . . . . . . . . . . . . . . . . . 3
28.2.1. Vacuum photodetectors . . . . . . . . . . . . . . . . . . . . 5
28.2.1.1. Photomultiplier tubes . . . . . . . . . . . . . . . . . . . 5
28.2.1.2. Microchannel plates . . . . . . . . . . . . . . . . . . . . 5
28.2.1.3. Hybrid photon detectors . . . . . . . . . . . . . . . . . . 6
28.2.2. Gaseous photon detectors . . . . . . . . . . . . . . . . . . . . 7
28.2.3. Solid-state photon detectors . . . . . . . . . . . . . . . . . . 7
28.3. Organic scintillators . . . . . . . . . . . . . . . . . . . . . . . . 9
28.3.1. Scintillation mechanism . . . . . . . . . . . . . . . . . . . . 10
28.3.2. Caveats and cautions . . . . . . . . . . . . . . . . . . . . . 11
28.3.3. Scintillating and wavelength-shifting fibers . . . . . . . . . . . . 12
28.4. Inorganic scintillators: . . . . . . . . . . . . . . . . . . . . . . . 13
28.5. Cherenkov detectors . . . . . . . . . . . . . . . . . . . . . . . . 17
28.5.1. Threshold counters . . . . . . . . . . . . . . . . . . . . . . 19
28.5.2. Imaging counters . . . . . . . . . . . . . . . . . . . . . . . 19
28.6. Cherenkov tracking calorimeters . . . . . . . . . . . . . . . . . . . 22
28.7. Gaseous detectors . . . . . . . . . . . . . . . . . . . . . . . . . 25
28.7.1. Energy loss and charge transport in gases . . . . . . . . . . . . . 25
28.7.2. Multi-Wire Proportional Chambers . . . . . . . . . . . . . . . 30
28.7.3. Micro-pattern Gas Detectors . . . . . . . . . . . . . . . . . . 34
28.7.4. Time-projection chambers . . . . . . . . . . . . . . . . . . . 37
28.7.5. Transition radiation detectors (TRD’s) . . . . . . . . . . . . . . 41
28.7.6. Resistive-plate chambers . . . . . . . . . . . . . . . . . . . . 43
28.8. Silicon semiconductor detectors . . . . . . . . . . . . . . . . . . . 46
28.9. Low-noise electronics . . . . . . . . . . . . . . . . . . . . . . . . 49
28.10. Calorimeters . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
28.10.1. Electromagnetic calorimeters . . . . . . . . . . . . . . . . . . 54
28.10.2. Hadronic calorimeters . . . . . . . . . . . . . . . . . . . . . 56
28.10.3. Free electron drift velocities in liquid ionization sensors . . . . . . 60
28.11. Superconducting magnets for collider detectors . . . . . . . . . . . . 61
28.11.1. Solenoid Magnets . . . . . . . . . . . . . . . . . . . . . . . 61
28.11.2. Properties of collider detector magnets . . . . . . . . . . . . . 63
CITATION: C. Amsler et al., Physics Letters B667, 1 (2008)
available on the PDG WWW pages (URL: http://pdg.lbl.gov/) July 24, 2008 18:04
2 28. Particle detectors
28.11.3. Toroidal magnets . . . . . . . . . . . . . . . . . . . . . . . 64
28.12. Measurement of particle momenta in a uniform magnetic
field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
28. PARTICLE DETECTORS
Revised 2007 (see the various sections for authors).
28.1. Summary of detector spatial resolution, temporal resolution,
and deadtime
In this section we give various parameters for common detector components. The
quoted numbers are usually based on typical devices, and should be regarded only as
rough approximations for new designs. More detailed discussions of detectors and their
underlying physics can be found in books by Ferbel [1], Grupen [2], Kleinknecht [3],
Knoll [4], Green [5], and Leroy & Rancoita [6]. In Table 28.1 are given typical
resolutions and deadtimes of common detectors.
Table 28.1: Typical resolutions and deadtimes of common detectors. Revised
September 2003 by R. Kadel (LBNL).
Resolution Dead
Detector Type Accuracy (rms) Time Time
Bubble chamber 10–150 µm 1 ms 50 ms
a
Streamer chamber 300 µm 2 µs 100 ms
Proportional chamber 50–300 µm
b,c,d
2 ns 200 ns
Drift chamber 50–300 µm 2 ns
e
100 ns
Scintillator — 100 ps/n
f
10 ns
Emulsion 1 µm — —
Liquid Argon Drift [7] ∼175–450 µm ∼ 200 ns ∼ 2 µs
Gas Micro Strip [8] 30–40 µm < 10 ns —
Resistive Plate chamber [9] 10 µm 1–2 ns —
Silicon strip pitch/(3 to 7)
g
h h
Silicon pixel 2 µm
i
h h
a
Multiple pulsing time.
b
300 µm is for 1 mm pitch.
c
Delay line cathode readout can give ±150 µm parallel to anode wire.
d
wirespacing/

12.
e
For two chambers.
f
n = index of refraction.
g
The highest resolution (“7”) is obtained for small-pitch detectors ( 25 µm) with
pulse-height-weighted center finding.
July 24, 2008 18:04
28. Particle detectors 3
h
Limited by the readout electronics [10]. (Time resolution of ≤ 25 ns is planned for
the ATLAS SCT.)
i
Analog readout of 34 µm pitch, monolithic pixel detectors.
28.2. Photon detectors
Updated August 2007 by D. Chakraborty (Northern Illinois U) and T. Sumiyoshi (Tokyo
Metro U).
Most detectors in high-energy, nuclear, and astrophysics rely on the detection of
photons in or near the visible range, 100 nm λ 1000 nm, or E ≈ a few eV. This
range covers scintillation and Cherenkov radiation as well as the light detected in many
astronomical observations.
Generally, photodetection involves generating a detectable electrical signal proportional
to the (usually very small) number of incident photons. The process involves three distinct
steps:
1. Generation of a primary photoelectron or electron-hole (e-h) pair by an incident
photon by the photoelectric or photoconductive effect,
2. Amplification of the p.e. signal to detectable levels by one or more multiplicative
bombardment steps and/or an avalanche process (usually), and,
3. Collection of the secondary electrons to form the electrical signal.
The important characteristics of a photodetector include the following in statistical
averages:
1. Quantum efficiency (QE or
Q
): the number of primary photoelectrons generated per
incident photon (0 ≤
Q
≤ 1; in silicon more than one e-h pair per incident photon can
be generated for λ
<

165 nm),
2. Collection efficiency (CE or
C
): the overall acceptance factor other than the
generation of photoelectrons (0 ≤
C
≤ 1),
3. Gain (G): the number of electrons collected for each photoelectron generated,
4. Dark current or dark noise: the electrical signal when there is no photon,
5. Energy resolution: electronic noise (ENC or N
e
) and statistical fluctuations in the
amplification process compound the Poisson distribution of n
γ
photons from a given
source:
σ(E)
E
=
¸
f
N
n
γ

Q

C
+
_
N
e
Gn
γ

Q

C
_
2
, (28.1)
where f
N
, or the excess noise factor (ENF), is the contribution to the energy
distribution variance due to amplification statistics [11],
6. Dynamic range: the maximum signal available from the detector (this is usually
expressed in units of the response to noise-equivalent power, or NEP, which is the
optical input power that produces a signal-to-noise ratio of 1),
7. Time dependence of the response: this includes the transit time, which is the time
between the arrival of the photon and the electrical pulse, and the transit time spread,
which contributes to the pulse rise time and width, and
8. Rate capability: inversely proportional to the time needed, after the arrival of one
photon, to get ready to receive the next.
July 24, 2008 18:04
4 28. Particle detectors
Table 28.2: Representative characteristics of some photodetectors commonly used in
particle physics. The time resolution of the devices listed here vary in the 10–2000 ps
range.
Type λ
Q

C
Gain Risetime Area 1-p.e noise HV Price
(nm) (ns) (mm
2
) (Hz) (V) (USD)
PMT

115–1100 0.15–0.25 10
3
–10
7
0.7–10 10
2
–10
5
10–10
4
500–3000 100–5000
MCP

100–650 0.01–0.10 10
3
–10
7
0.15–0.3 10
2
–10
4
0.1–200 500–3500 10–6000
HPD

115–850 0.1–0.3 10
3
–10
4
7 10
2
–10
5
10–10
3
∼2 ×10
4
∼600
GPM

115–500 0.15–0.3 10
3
–10
6
O(0.1) O(10) 10–10
3
300–2000 O(10)
APD 300–1700 ∼0.7 10–10
8
O(1) 10–10
3
1–10
3
400–1400 O(100)
PPD 400–550 0.15–0.3 10
5
–10
6
∼ 1 1–10 O(10
6
) 30–60 O(10)
VLPC 500–600 ∼0.9 ∼5 ×10
4
∼ 10 1 O(10
4
) ∼7 ∼1
∗These devices often come in multi-anode configurations. In such cases, area, noise, and
price are to be considered on a “per readout-channel” basis.
The QE is a strong function of the photon wavelength (λ), and is usually quoted at
maximum, together with a range of λ where the QE is comparable to its maximum.
Spatial uniformity and linearity with respect to the number of photons are highly
desirable in a photodetector’s response.
Optimization of these factors involves many trade-offs and vary widely between
applications. For example, while a large gain is desirable, attempts to increase the gain
for a given device also increases the ENF and after-pulsing (“echos” of the main pulse).
In solid-state devices, a higher QE often requires a compromise in the timing properties.
In other types, coverage of large areas by focusing increases the transit time spread.
Other important considerations also are highly application-specific. These include the
photon flux and wavelength range, the total area to be covered and the efficiency required,
the volume available to accommodate the detectors, characteristics of the environment
such as chemical composition, temperature, magnetic field, ambient background, as well
ambient radiation of different types and, mode of operation (continuous or triggered),
bias (high-voltage) requirements, power consumption, calibration needs, aging, cost, and
so on. Several technologies employing different phenomena for the three steps described
above, and many variants within each, offer a wide range of solutions to choose from. The
salient features of the main technologies and the common variants are described below.
Some key characteristics are summarized in Table 28.2.
July 24, 2008 18:04
28. Particle detectors 5
28.2.1. Vacuum photodetectors : Vacuum photodetectors can be broadly subdivided
into three types: photomultiplier tubes, microchannel plates, and hybrid photodetectors.
28.2.1.1. Photomultiplier tubes: A versatile class of photon detectors, vacuum
photomultiplier tubes (PMT) has been employed by a vast majority of all particle physics
experiments to date [11]. Both “transmission-” and “reflection-type” PMT’s are widely
used. In the former, the photocathode material is deposited on the inside of a transparent
window through which the photons enter, while in the latter, the photocathode material
rests on a separate surface that the incident photons strike. The cathode material has
a low work function, chosen for the wavelength band of interest. When a photon hits
the cathode and liberates an electron (the photoelectric effect), the latter is accelerated
and guided by electric fields to impinge on a secondary-emission electrode, or dynode,
which then emits a few (∼ 5) secondary electrons. The multiplication process is repeated
typically 10 times in series to generate a sufficient number of electrons, which are collected
at the anode for delivery to the external circuit. The total gain of a PMT depends on
the applied high voltage V as G = AV
kn
, where k ≈ 0.7–0.8 (depending on the dynode
material), n is the number of dynodes in the chain, and A a constant (which also depends
on n). Typically, G is in the range of 10
5
–10
6
. Pulse risetimes are usually in the few
nanosecond range. With e.g. two-level discrimination the effective time resolution can be
much better.
A large variety of PMT’s, including many just recently developed, covers a wide span
of wavelength ranges from infrared (IR) to extreme ultraviolet (XUV) [12]. They are
categorized by the window materials, photocathode materials, dynode structures, anode
configurations, etc. Common window materials are borosilicate glass for IR to near-UV,
fused quartz and sapphire (Al
2
O
3
) for UV, and MgF
2
or LiF for XUV. The choice
of photocathode materials include a variety of mostly Cs- and/or Sb-based compounds
such as CsI, CsTe, bi-alkali (SbRbCs, SbKCs), multi-alkali (SbNa
2
KCs), GaAs(Cs),
GaAsP, etc. Sensitive wavelengths and peak quantum efficiencies for these materials are
summarized in Table 28.3. Typical dynode structures used in PMT’s are circular cage,
line focusing, box and grid, venetian blind, and fine mesh. In some cases, limited spatial
resolution can be obtained by using a mosaic of multiple anodes.
PMT’s are vulnerable to magnetic fields—sometimes even the geomagnetic field causes
large orientation-dependent gain changes. A high-permeability metal shield is often
necessary. However, proximity-focused PMT’s, e.g. the fine-mesh types, can be used even
in a high magnetic field (≥ 1 T) if the electron drift direction is parallel to the field.
28.2.1.2. Microchannel plates: A typical Microchannel plate (MCP) photodetector
consists of one or more ∼2 mm thick glass plates with densely packed O(10 µm)-diameter
cylindrical holes, or “channels”, sitting between the transmission-type photocathode
and anode planes, separated by O(1 mm) gaps. Instead of discrete dynodes, the inner
surface of each cylindrical tube serves as a continuous dynode for the entire cascade
of multiplicative bombardments initiated by a photoelectron. Gain fluctuations can be
minimized by operating in a saturation mode, whence each channel is only capable of a
binary output, but the sum of all channel outputs remains proportional to the number of
photons received so long as the photon flux is low enough to ensure that the probability
of a single channel receiving more than one photon during a single time gate is negligible.
July 24, 2008 18:04
6 28. Particle detectors
MCP’s are thin, offer good spatial resolution, have excellent time resolution (∼20 ps), and
can tolerate random magnetic fields up to 0.1 T and axial fields up to ∼ 1 T. However,
they suffer from relatively long recovery time per channel and short lifetime. MCP’s are
widely employed as image-intensifiers, although not so much in HEP or astrophysics.
28.2.1.3. Hybrid photon detectors: Hybrid photon detectors (HPD) combine the
sensitivity of a vacuum PMT with the excellent spatial and energy resolutions of a Si
sensor [13]. A single photoelectron ejected from the photocathode is accelerated through
a potential difference of ∼20 kV before it impinges on the silicon sensor/anode. The gain
nearly equals the maximum number of e-h pairs that could be created from the entire
kinetic energy of the accelerated electron: G ≈ eV/w, where e is the electronic charge,
V is the applied potential difference, and w ≈ 3.7 eV is the mean energy required to
create an e-h pair in Si at room temperature. Since the gain is achieved in a single step,
one might expect to have the excellent resolution of a simple Poisson statistic with large
mean, but in fact it is even better, thanks to the Fano effect discussed in Sec. 28.8.
Low-noise electronics must be used to read out HPD’s if one intends to take advantage
of the low fluctuations in gain, e.g. when counting small numbers of photons. HPD’s can
have the same
Q

C
and window geometries as PMT’s and can be segmented down to
∼50 µm. However, they require rather high biases and will not function in a magnetic
field. The exception is proximity-focused devices (⇒ no (de)magnification) in an axial
field. With time resolutions of ∼10 ps and superior rate capabiility, proximity-focused
HPD’s can be an alternative to MCP’s. Current applications of HPD’s include the
CMS hadronic calorimeter and the RICH detector in LHCb. Large-size HPD’s with
sophisticated focusing may be suitable for future water Cherenkov experiments.
Hybrid APD’s (HAPD’s) add an avalanche multiplication step following the electron
bombardment to boost the gain by a factor of ∼50. This affords a higher gain and/or
lower electrical bias, but also degrades the signal definition.
Table 28.3: Properties of photocathode and window materials commonly used in
vacuum photodetectors [12].
Photocathode λ Window Peak
Q
(λ/nm)
material (nm) material
CsI 115–200 MgF
2
0.15 (135)
CsTe 115–240 MgF
2
0.18 (210)
Bi-alkali 300–650 Borosilicate 0.27 (390)
160-650 Quartz 0.27 (390)
Multi-alkali 300–850 Borosilicate 0.20 (360)
160-850 Quartz 0.23 (280)
GaAs(Cs)

160–930 Quartz 0.23 (280)
GaAsP(Cs) 300-750 Borosilicate 0.42 (560)

Reflection type photocathode is used.
July 24, 2008 18:04
28. Particle detectors 7
28.2.2. Gaseous photon detectors : In gaseous photomultipliers (GPM) a photoelec-
tron in a suitable gas mixture initiates an avalanche in a high-field region, producing a
large number of secondary impact-ionization electrons. In principle the charge multiplica-
tion and collection processes are identical to those employed in gaseous tracking detectors
such as multiwire proportional chambers, micromesh gaseous detectors (Micromegas), or
gas electron multipliers (GEM). These are discussed in Sec. 28.7.3.
The devices can be divided into two types depending on the photocathode material.
One type uses solid photocathode materials much in the same way as PMT’s. Since it is
resistant to gas mixtures typically used in tracking chambers, CsI is a common choice.
In the other type, photoionization occurs on suitable molecules vaporized and mixed in
the drift volume. Most gases have photoionization work functions in excess of 10 eV,
which would limit their sensitivity to wavelengths far too short. However, vapors of
TMAE (tetrakis dimethyl-amine ethylene) or TEA (tri-ethyl-amine), which have smaller
work functions (5.3 eV for TMAE and 7.5 eV for TEA), are suited for XUV photon
detection [14]. Since devices like GEM’s offer sub-mm spatial resolution, GPM’s are
often used as position-sensitive photon detectors. They can be made into flat panels
to cover large areas (O(1 m
2
)), can operate in high magnetic fields, and are relatively
inexpensive. Many of the ring imaging Cherenkov (RICH) detectors to date have used
GPM’s for the detection of Cherenkov light [15]. Special care must be taken to suppress
the photon-feedback process in GPM’s. It is also important to maintain high purity of
the gas as minute traces of O
2
can significantly degrade the detection efficiency.
28.2.3. Solid-state photon detectors : In a phase of rapid development, solid-state
photodetectors are competing with vacuum- or gas-based devices for many existing
applications and making way for a multitude of new ones. Compared to traditional
vacuum- and gaseous photodetectors, solid-state devices are more compact, lightweight,
rugged, tolerant to magnetic fields, and often cheaper. They also allow fine pixelization,
are easy to integrate into large systems, and can operate at low electric potentials, while
matching or exceeding most performance criteria. They are particularly well suited for
detection of γ- and X-rays. Except for applications where coverage of very large areas
or dynamic range is required, solid-state detectors are proving to be the better choice.
Some hybrid devices attempt to combine the best features of different technologies while
applications of nanotechnology are opening up exciting new possibilities.
Silicon photodiodes (PD) are widely used in high-energy physics as particle detectors
and in a great number of applications (including solar cells!) as light detectors. The
structure is discussed in some detail in Sec. 28.8. In its simplest form, the PD is a
reverse-biased p-n junction. Photons with energies above the indirect bandgap energy
(wavelengths shorter than about 1050 nm, depending on the temperature) can create e-h
pairs (the photoconductive effect), which are collected on the p and n sides, respectively.
Often, as in the PD’s used for crystal scintillator readout in CLEO, L3, Belle,BaBar,
and GLAST, intrinsic silicon is doped to create a p-i-n structure. The reverse bias
increases the thickness of the depleted region; in the case of these particular detectors,
to full depletion at a depth of about 100 µm. Increasing the depletion depth decreases
the capacitance (and hence electronic noise) and extends the red response. Quantum
efficiency can exceed 90%, but falls toward the red because of the increasing absorption
July 24, 2008 18:04
8 28. Particle detectors
length of light in silicon. The absorption length reaches 100 µm at 985 nm. However,
since G = 1, amplification is necessary. Optimal low-noise amplifiers are slow, but, even
so, noise limits the minimum detectable signal in room-temperature devices to several
hundred photons.
Very large arrays containing O(10
7
) of O(10 µm
2
)-sized photodioides pixelizing a
plane are widely used to photograph all sorts of things from everyday subjects at visible
wavelengths to crystal structures with X-rays and astronomical objects from infrared to
UV. To limit the number of readout channels, these are made into charge-coupled devices
(CCD), where pixel-to-pixel signal transfer takes place over thousands of synchronous
cycles with sequential output through shift registers [16]. Thus, high spatial resolution
is achieved at the expense of speed and timing precision. Custom-made CCD’s have
virtually replaced photographic plates and other imagers for astronomy and in spacecraft.
Typical QE’s exceed 90% over much of the visible spectrum, and “thick” CCD’s have
useful QE up to λ = 1 µm. Active Pixel Sensor (APS) arrays with a preamplifier on
each pixel and CMOS processing afford higher speeds, but are challenged at longer
wavelengths. Much R&D is underway to overcome the limitations of both CCD and
CMOS imagers.
In avalanche photodiodes (APD), an exponential cascade of impact ionizations initiated
by the initial photogenerated e-h pair under a large reverse-bias voltage leads to an
avalanche breakdown [17]. As a result, detectable electrical response can be obtained
from low-intensity optical signals down to single photons. Excellent junction uniformity is
critical, and a guard ring is generally used as a protection against edge breakdown. Well-
designed APD’s, such as those used in CMS’ crystal-based electromagnetic calorimeter,
have achieved
Q

C
≈ 0.7 with sub-ns response time. The sensitive wavelength window
and gain depend on the semiconductor used. The gain is typically 10–200 in linear and
up to 10
8
in Geiger mode of operation. Stability and close monitoring of the operating
temperature are important for linear-mode operation, and substantial cooling is often
necessary. Position-sensitive APD’s use time information at multiple anodes to calculate
the hit position.
One of the most promising recent developments in the field is that of devices consisting
of large arrays (O(10
3
)) of tiny APD’s packed over a small area (O(1 mm
2
)) and
operated in a limited Geiger mode [18]. Among different names used for this class
of photodetectors, “PPD” (for “Pixelized Photon Detector”) is most widely accepted
(formerly “SiPM”). Although each cell only offers a binary output, linearity with respect
to the number of photons is achieved by summing the cell outputs in the same way as with
a MCP in saturation mode (see above). PPD’s are being adopted as the preferred solution
for various purposes including medical imaging, e.g. positron emission tomography (PET).
These compact, rugged, and economical devices allow auto-calibration through decent
separation of photoelectron peaks and offer gains of O(10
6
) at a moderate bias voltage
(∼50 V). However, the single-photoelectron noise of a PPD, being the logical “or” of
O(10
3
) Geiger APD’s, is rather large: O(1 MHz/mm
2
) at room temperature. PPD’s are
particularly well-suited for applications where triggered pulses of several photons are
expected over a small area, e.g. fiber-guided scintillation light. Intense R&D is expeced
to lower the noise level and improve radiation hardness, resulting in coverage of larger
July 24, 2008 18:04
28. Particle detectors 9
areas and wider applications. Attempts are being made to combine the fabrication of the
sensors and the front-end electronics (ASIC) in the same process with the goal of making
PPD’s and other finely pixelized solid-state photodetectors extremely easy to use.
Of late, much R&D has been directed to p-i-n diode arrays based on thin polycrystalline
diamond films formed by chemical vapor deposition (CVD) on a hot substrate (∼1000
K) from a hydrocarbon-containing gas mixture under low pressure (∼100 mbar).
These devices have maximum sensitivity in the extreme- to moderate-UV region [19].
Many desirable characteristics, including high tolerance to radiation and temperature
fluctuations, low dark noise, blindness to most of the solar radiation spectrum, and
relatively low cost make them ideal for space-based UV/XUV astronomy, measurement of
synchrotron radiation, and luminosity monitoring at (future) lepton collider(s).
Visible-light photon counters (VLPC) utilize the formation of an impurity band only
50 meV below the conduction band in As-doped Si to generate strong (G ≈ 5 × 10
4
)
yet sharp response to single photons with
Q
≈ 0.9 [20]. The smallness of the band
gap considerably reduces the gain dispersion. Only a very small bias (∼7 V) is needed,
but high sensitivity to infrared photons requires cooling below 10 K. The dark noise
increases sharply and exponentially with both temperature and bias. The Run 2 DØ
detector uses 86000 VLPC’s to read the optical signal from its scintillating-fiber tracker
and scintillator-strip preshower detectors.
28.3. Organic scintillators
Revised September 2007 by K.F. Johnson (FSU).
Organic scintillators are broadly classed into three types, crystalline, liquid, and plastic,
all of which utilize the ionization produced by charged particles (see Sec. 27.2) of this
Review) to generate optical photons, usually in the blue to green wavelength regions [21].
Plastic scintillators are by far the most widely used. Crystal organic scintillators are
practically unused in high-energy physics.
Densities range from 1.03 to 1.20 g cm
−3
. Typical photon yields are about 1 photon
per 100 eV of energy deposit [22]. A one-cm-thick scintillator traversed by a minimum-
ionizing particle will therefore yield ≈ 2×10
4
photons. The resulting photoelectron signal
will depend on the collection and transport efficiency of the optical package and the
quantum efficiency of the photodetector.
Plastic scintillators do not respond linearly to the ionization density. Very dense
ionization columns emit less light than expected on the basis of dE/dx for minimum-
ionizing particles. A widely used semi-empirical model by Birks posits that recombination
and quenching effects between the excited molecules reduce the light yield [23]. These
effects are more pronounced the greater the density of the excited molecules. Birks’
formula is
dL
dx
= L
0
dE/dx
1 +k
B
dE/dx
, (28.2)
where L is the luminescence, L
0
is the luminescence at low specific ionization density,
and k
B
is Birks’ constant, which must be determined for each scintillator by measurement.
Decay times are in the ns range; rise times are much faster. The combination of high
light yield and fast response time allows the possibility of sub-ns timing resolution [24].
July 24, 2008 18:04
10 28. Particle detectors
The fraction of light emitted during the decay “tail” can depend on the exciting particle.
This allows pulse shape discrimination as a technique to carry out particle identification.
Because of the hydrogen content (carbon to hydrogen ratio ≈ 1) plastic scintillator is
sensitive to proton recoils from neutrons. Ease of fabrication into desired shapes and
low cost has made plastic scintillators a common detector component. Recently, plastic
scintillators in the form of scintillating fibers have found widespread use in tracking and
calorimetry [25].
28.3.1. Scintillation mechanism :
Scintillation: A charged particle traversing matter leaves behind it a wake of excited
molecules. Certain types of molecules, however, will release a small fraction (≈ 3%)
of this energy as optical photons. This process, scintillation, is especially marked in
those organic substances which contain aromatic rings, such as polystyrene (PS) and
polyvinyltoluene (PVT). Liquids which scintillate include toluene and xylene.
Fluorescence: In fluorescence, the initial excitation takes place via the absorption of a
photon, and de-excitation by emission of a longer wavelength photon. Fluors are used
as “waveshifters” to shift scintillation light to a more convenient wavelength. Occurring
in complex molecules, the absorption and emission are spread out over a wide band of
photon energies, and have some overlap, that is, there is some fraction of the emitted
light which can be re-absorbed [26]. This “self-absorption” is undesirable for detector
applications because it causes a shortened attenuation length. The wavelength difference
between the major absorption and emission peaks is called the Stokes’ shift. It is usually
the case that the greater the Stokes’ shift, the smaller the self absorption—thus, a large
Stokes’ shift is a desirable property for a fluor.
Ionizat ion excit at ion of base plast ic
For st er ener gy t r ansfer
γ
γ
base plast ic
pr imar y fluor
(~1% wt / wt )
secondar y fluor
(~0.05% wt / wt )
phot odet ect or
emit UV, ~ 340 nm
absor b blue phot on
absor b UV phot on
emit blue, ~ 400 nm
1 m
10
−4
m
10
−8
m
Figure 28.1: Cartoon of scintillation “ladder” depicting the operating mechanism
of plastic scintillator. Approximate fluor concentrations and energy transfer
distances for the separate sub-processes are shown.
Scintillators: The plastic scintillators used in high-energy physics are binary or ternary
July 24, 2008 18:04
28. Particle detectors 11
solutions of selected fluors in a plastic base containing aromatic rings. (See the appendix
in Ref. 27 for a comprehensive list of components.) Virtually all plastic scintillators
contain as a base either PVT or PS. PVT-based scintillator can be up to 50% brighter.
Ionization in the plastic base produces UV photons with short attenuation length
(several mm). Longer attenuation lengths are obtained by dissolving a “primary” fluor in
high concentration (1% by weight) into the base, which is selected to efficiently re-radiate
absorbed energy at wavelengths where the base is more transparent.
The primary fluor has a second important function. The decay time of the scintillator
base material can be quite long—in pure polystyrene it is 16 ns, for example. The
addition of the primary fluor in high concentration can shorten the decay time by an
order of magnitude and increase the total light yield. At the concentrations used (1%
and greater), the average distance between a fluor molecule and an excited base unit is
around 100
˚
A, much less than a wavelength of light. At these distances the predominant
mode of energy transfer from base to fluor is not the radiation of a photon, but a resonant
dipole-dipole interaction, first described by Foerster, which strongly couples the base and
fluor [28]. The strong coupling sharply increases the speed and the light yield of the
plastic scintillators.
Unfortunately, a fluor which fulfills other requirements is usually not completely
adequate with respect to emission wavelength or attenuation length, so it is necessary
to add yet another waveshifter (the “secondary” fluor), at fractional percent levels, and
occasionally a third (not shown in Fig. 28.1).
External wavelength shifters: Light emitted from a plastic scintillator may be absorbed
in a (nonscintillating) base doped with a wave-shifting fluor. Such wavelength shifters are
widely used to aid light collection in complex geometries. The wavelength shifter must
be insensitive to ionizing radiation and Cherenkov light. A typical wavelength shifter
uses an acrylic base because of its good optical qualities, a single fluor to shift the
light emerging from the plastic scintillator to the blue-green, and contains ultra-violet
absorbing additives to deaden response to Cherenkov light.
28.3.2. Caveats and cautions : Plastic scintillators are reliable, robust, and conve-
nient. However, they possess quirks to which the experimenter must be alert.
Aging and Handling: Plastic scintillators are subject to aging which diminishes the light
yield. Exposure to solvent vapors, high temperatures, mechanical flexing, irradiation, or
rough handling will aggravate the process. A particularly fragile region is the surface
which can “craze”—develop microcracks—which rapidly destroy the capability of plastic
scintillators to transmit light by total internal reflection. Crazing is particularly likely
where oils, solvents, or fingerprints have contacted the surface.
Attenuation length: The Stokes’ shift is not the only factor determining attenuation
length. Others are the concentration of fluors (the higher the concentration of a fluor, the
greater will be its self-absorption); the optical clarity and uniformity of the bulk material;
the quality of the surface; and absorption by additives, such as stabilizers, which may be
present.
Afterglow: Plastic scintillators have a long-lived luminescence which does not follow a
July 24, 2008 18:04
12 28. Particle detectors
simple exponential decay. Intensities at the 10
−4
level of the initial fluorescence can
persist for hundreds of ns [21,29].
Atmospheric quenching: Plastic scintillators will decrease their light yield with increasing
partial pressure of oxygen. This can be a 10% effect in an artificial atmosphere [30]. It
is not excluded that other gases may have similar quenching effects.
Magnetic field: The light yield of plastic scintillators may be changed by a magnetic field.
The effect is very nonlinear and apparently not all types of plastic scintillators are so
affected. Increases of ≈ 3% at 0.45 T have been reported [31]. Data are sketchy and
mechanisms are not understood.
Radiation damage: Irradiation of plastic scintillators creates color centers which absorb
light more strongly in the UV and blue than at longer wavelengths. This poorly
understood effect appears as a reduction both of light yield and attenuation length.
Radiation damage depends not only on the integrated dose, but on the dose rate,
atmosphere, and temperature, before, during and after irradiation, as well as the
materials properties of the base such as glass transition temperature, polymer chain
length, etc. Annealing also occurs, accelerated by the diffusion of atmospheric oxygen
and elevated temperatures. The phenomena are complex, unpredictable, and not well
understood [32]. Since color centers are less intrusive at longer wavelengths, the most
reliable method of mitigating radiation damage is to shift emissions at every step to the
longest practical wavelengths, e.g., utilize fluors with large Stokes’ shifts (aka the “Better
red than dead” strategy).
28.3.3. Scintillating and wavelength-shifting fibers :
The clad optical fiber is an incarnation of scintillator and wavelength shifter (WLS)
which is particularly useful [33]. Since the initial demonstration of the scintillating fiber
(SCIFI) calorimeter [34], SCIFI techniques have become mainstream [35].
SCIFI calorimeters are fast, dense, radiation hard, and can have leadglass-like
resolution. SCIFI trackers can handle high rates and are radiation tolerant, but the
low photon yield at the end of a long fiber (see below) forces the use of sensitive
photodetectors. WLS scintillator readout of a calorimeter allows a very high level of
hermeticity since the solid angle blocked by the fiber on its way to the photodetector is
very small. The sensitive region of scintillating fibers can be controlled by splicing them
onto clear (non-scintillating/non-WLS) fibers.
A typical configuration would be fibers with a core of polystyrene-based scintillator or
WLS (index of refraction n = 1.59), surrounded by a cladding of PMMA (n = 1.49) a few
microns thick, or, for added light capture, with another cladding of fluorinated PMMA
with n = 1.42, for an overall diameter of 0.5 to 1 mm. The fiber is drawn from a boule
and great care is taken during production to ensure that the intersurface between the
core and the cladding has the highest possible uniformity and quality, so that the signal
transmission via total internal reflection has a low loss. The fraction of generated light
which is transported down the optical pipe is denoted the capture fraction and is about
6% for the single-clad fiber and 10% for the double-clad fiber.
The number of photons from the fiber available at the photodetector is always smaller
than desired, and increasing the light yield has proven difficult. A minimum-ionizing
July 24, 2008 18:04
28. Particle detectors 13
particle traversing a high-quality 1 mm diameter fiber perpendicular to its axis will
produce fewer than 2000 photons, of which about 200 are captured. Attenuation may
eliminate 95% of these photons in a large collider tracker.
A scintillating or WLS fiber is often characterized by its “attenuation length,” over
which the signal is attenuated to 1/e of its original value. Many factors determine
the attenuation length, including the importance of re-absorption of emitted photons
by the polymer base or dissolved fluors, the level of crystallinity of the base polymer,
and the quality of the total internal reflection boundary. Attenuation lengths of several
meters are obtained by high quality fibers. However, it should be understood that the
attenuation length is not necessarily a measure of fiber quality. Among other things, it
is not constant with distance from the excitation source and it is wavelength dependent.
So-called “cladding light” causes some of the distance dependence [36], but not all. The
wavelength dependence is usually related to the higher re-absorption of shorter wavelength
photons—once absorbed, re-emitted isotropically and lost with 90% probability—and to
the lower absorption of longer wavelengths by polystyrene. Experimenters should be aware
that measurements of attenuation length by a phototube with a bialkali photocathode,
whose quantum efficiency drops below 10% at 480 nm, should not be na¨ıvely compared
to measurements utilizing a silicon photodiode, whose quantum efficiency is still rising at
600 nm.
28.4. Inorganic scintillators:
Revised September 2007 by R.-Y. Zhu (California Institute of Technology) and
C.L. Woody (BNL).
Inorganic crystals form a class of scintillating materials with much higher densities
than organic plastic scintillators (typically ∼ 4–8 g/cm
3
) with a variety of different
properties for use as scintillation detectors. Due to their high density and high effective
atomic number, they can be used in applications where high stopping power or a high
conversion efficiency for electrons or photons is required. These include total absorption
electromagnetic calorimeters (see Sec. 28.10.1), which consist of a totally active absorber
(as opposed to a sampling calorimeter), as well as serving as gamma ray detectors over a
wide range of energies. Many of these crystals also have very high light output, and can
therefore provide excellent energy resolution down to very low energies (∼ few hundred
keV).
Some crystals are intrinsic scintillators in which the luminescence is produced by a
part of the crystal lattice itself. However, other crystals require the addition of a dopant,
typically fluorescent ions such as thallium (Tl) or cerium (Ce) which is responsible for
producing the scintillation light. However, in both cases, the scintillation mechanism is
the same. Energy is deposited in the crystal by ionization, either directly by charged
particles, or by the conversion of photons into electrons or positrons which subsequently
produce ionization. This energy is transferred to the luminescent centers which then
radiate scintillation photons. The efficiency η for the conversion of energy deposit in the
crystal to scintillation light can be expressed by the relation [37]
η = β · S · Q . (28.3)
July 24, 2008 18:04
14 28. Particle detectors
where β is the efficiency of the energy conversion process, S is the efficiency of energy
transfer to the luminescent center, and Q is the quantum efficiency of the luminescent
center. The value of η ranges between 0.1 and ∼ 1 depending on the crystal, and is
the main factor in determining the intrinsic light output of the scintillator. In addition,
the scintillation decay time is primarily determined by the energy transfer and emission
process. The decay time of the scintillator is mainly dominated by the decay time of the
luminescent center. For example, in the case of thallium doped sodium iodide (NaI(Tl)),
the value of η is ∼ 0.5, which results in a light output ∼ 40,000 photons per MeV of
energy deposit. This high light output is largely due to the high quantum efficiency of
the thallium ion (Q ∼ 1), but the decay time is rather slow (τ ∼ 250 ns).
Table 28.4 lists the basic properties of some commonly used inorganic crystal
scintillators. NaI(Tl) is one of the most common and widely used scintillators, with
an emission that is well matched to a bialkali photomultiplier tube, but it is highly
hygroscopic and difficult to work with, and has a rather low density. CsI(Tl) has
high light yield, an emission that is well matched to solid state photodiodes, and is
mechanically robust (high plasticity and resistance to cracking). However, it needs careful
surface treatment and is slightly hygroscopic. Compared with CsI(Tl), pure CsI has
identical mechanical properties, but faster emission at shorter wavelengths and light
output approximately an order of magnitude lower. BaF
2
has a fast component with
a sub-nanosecond decay time, and is the fastest known scintillator. However, it also
has a slow component with a much longer decay time (∼ 630 ns). Bismuth gemanate
(Bi
4
Ge
3
O
12
or BGO) has a high density, and consequently a short radiation length X
0
and Moli`ere radius R
M
. BGO’s emission is well-matched to the spectral sensitivity of
photodiodes, and it is easy to handle and not hygroscopic. Lead tungstate (PbWO
4
or
PWO) has a very high density, with a very short X
0
and R
M
, but its intrinsic light yield
is rather low. Cerium doped lutetium oxyorthosilicate (Lu
2
SiO
5
:Ce, or LSO:Ce) [38],
cerium doped lutetium-yttrium oxyorthosilicate (Lu
2(1−x)
Y
2x
SiO
5
, LYSO:Ce) [39] and
cerium doped gadolinium orthosilicate (Gd
2
SiO
5
:Ce, or GSO:Ce) [40] are dense crystal
scintillators which have a high light yield and a fast decay time. Only properties of
LSO:Ce and GSO:Ce are listed in Table 28.4 since the properties of LYSO:Ce are similar
to that of LSO:Ce except a little lower density than LSO:Ce depending on the yttrium
fraction in LYSO:Ce [41].
Beside the crystals listed in Table 28.4, a number of new crystals are being developed
that may have potential applications in high energy or nuclear physics. Of particular
interest is the family of yttrium and lutetium perovskites, which include YAP (YAlO
3
:Ce)
and LuAP (LuAlO
3
:Ce) and their mixed compositions. These have been shown to be
linear over a large energy range [42], and have the potential for providing extremely good
intrinsic energy resolution. In addition, other fluoride crystals such as CeF
3
have been
shown to provide excellent energy resolution in calorimeter applications.
Table 28.4 gives the light output of other crystals relative to NaI(Tl) and their
dependence to the temperature variations measured for crystal samples of 1.5 X
0
cube
with a Tyvek paper wrapping and a full end face coupled to a photodetector [43]. The
quantum efficiencies of the photodetector is taken out to facilitate a direct comparison
of crystal’s light output. However, the useful signal produced by a scintillator is
July 24, 2008 18:04
28. Particle detectors 15
usually quoted in terms of the number of photoelectrons per MeV produced by a given
photodetector. The relationship between the number of photons/MeV produced and
photoelectrons/MeV detected involves the factors for the light collection efficiency L and
the quantum efficiency QE of the photodetector:
N
p.e.
/MeV = L · QE · N
γ
/MeV (28.4)
L includes the transmission of scintillation light within the crystal (i.e., the bulk
attenuation length of the material), reflections and scattering from the surfaces, and the
size and shape of the crystal. These factors can vary considerably depending on the
sample, but can be in the range of ∼10–60%. The internal light transmission depends on
the intrinsic properties of the material, e.g. the density and type of the scattering centers
and defects that can produce internal absorption within the crystal, and can be highly
affected by factors such as radiation damage, as discussed below.
The quantum efficiency depends on the type of photodetector used to detect the
scintillation light, which is typically ∼15–20% for photomultiplier tubes and ∼70% for
silicon photodiodes for visible wavelengths. The quantum efficiency of the detector is
usually highly wavelength dependent and should be matched to the particular crystal of
interest to give the highest quantum yield at the wavelength corresponding to the peak of
the scintillation emission. Fig. 28.2 shows the quantum efficiencies of two photodetectors,
a Hamamatsu R2059 PMT with bi-alkali cathode and quartz window and a Hamamatsu
S8664 avalanche photodiode (APD) as a function of wavelength. Also shown in the
figure are emission spectra of three crystal scintillators, BGO, LSO:Ce/LYSO:Ce and
CsI(Tl), and the numerical values of the emission weighted quantum efficiency. The
area under each emission spectrum is proportional to crystal’s light yield, as shown in
Table 28.4, where the quantum efficiencies of the photodetector has been taken out.
Results with different photodetectors can be significantly different. For example, the
response of CsI(Tl) relative to NaI(Tl) with a standard photomultiplier tube with a
bialkali photocathode, e.g. Hamamatsu R2059, would be 45 rather than 165 because
of the photomultiplier’s low quantum efficiency at longer wavelengths. For scintillators
which emit in the UV, a detector with a quartz window should be used.
One important issue related to the application of a crystal scintillator is its radiation
hardness. Stability of its light output, or the ability to track and monitor the variation of
its light output in a radiation environment, is required for high resolution and precision
calibration [44]. All known crystal scintillators suffer from radiation damage. A common
damage phenomenon is the appearance of radiation induced absorption caused by the
formation of color centers originated from the impurities or point defects in the crystal.
This radiation induced absorption reduces the light attenuation length in the crystal, and
hence its light output. For crystals with high defect density, a severe reduction of light
attenuation length may cause a distortion of the light response uniformity, leading to a
degradation of the energy resolution. Additional radiation damage effects may include
a reduced intrinsic scintillation light yield (damage to the luminescent centers) and an
increased phosphorescence (afterglow). For crystals to be used in the construction a high
precision calorimeter in a radiation environment, its scintillation mechanism must not be
damaged and its light attenuation length in the expected radiation environment must be
July 24, 2008 18:04
16 28. Particle detectors
long enough so that its light response uniformity, and thus its energy resolution, does not
change [45].
Most of the crystals listed in Table 28.4 have been used in high energy or nuclear
physics experiments when the ultimate energy resolution for electrons and photons is
desired. Examples are the Crystal Ball NaI(Tl) calorimeter at SPEAR, the L3 BGO
calorimeter at LEP, the CLEO CsI(Tl) calorimeter at CESR, the KTeV CsI calorimeter
at the Tevatron, the BaBar and BELLE CsI(Tl) calorimeters at PEP-II and KEK.
Because of its high density and low cost, PWO calorimeters are widely used by CMS
and ALICE at LHC, by CLAS and PrimEx at CEBAF, and are the leading option for
PANDA at GSI. Recently, investigations have been made aiming at using LSO:Ce or
LYSO:Ce crystals for future high energy or nuclear physics experiments [41].
Table 28.4: Properties of several inorganic crystal scintillators. Most of the notation is
defined in Sec. 6 of this Review.
Parameter: ρ MP X

0
R

M
dE/dx λ

I
τ
decay
λ
max
n

Relative Hygro- d(LY)/dT
output

scopic?
Units: g/cm
3 ◦
C cm cm MeV/cm cm ns nm %/

C

NaI(Tl) 3.67 651 2.59 4.13 4.8 42.9 230 410 1.85 100 yes −0.2
BGO 7.13 1050 1.12 2.23 9.0 22.8 300 480 2.15 21 no −0.9
BaF
2
4.89 1280 2.03 3.10 6.6 30.7 630
s
300
s
1.50 36
s
no −1.3
s
0.9
f
220
f
3.4
f
∼0
f
CsI(Tl) 4.51 621 1.86 3.57 5.6 39.3 1300 560 1.79 165 slight 0.3
CsI(pure) 4.51 621 1.86 3.57 5.6 39.3 35
s
420
s
1.95 3.6
s
slight −1.3
6
f
310
f
1.1
f
PbWO
4
8.3 1123 0.89 2.00 10.2 20.7 30
s
425
s
2.20 0.083
s
no −2.7
10
f
420
f
0.29
f
LSO(Ce) 7.40 2050 1.14 2.07 9.6 20.9 40 420 1.82 83 no −0.2
GSO(Ce) 6.71 1950 1.38 2.23 8.9 22.2 600
s
430 1.85 3
s
no −0.1
56
f
30
f

Numerical values calculated using formulae in this review.

Refractive index at the wavelength of the emission maximum.

Relative light output measured for samples of 1.5 X
0
cube with a Tyvek paper
wrapping and a full end face coupled to a photodetector. The quantum efficiencies of the
photodetector is taken out.

Variation of light yield with temperature evaluated at the room temperature.
f = fast component, s = slow component
July 24, 2008 18:04
28. Particle detectors 17
7
Figure 28.2: The quantum efficiencies of two photodetectors, a Hamamatsu R2059
PMT with bi-alkali cathode and a Hamamatsu S8664 avalanche photodiode (APD),
are shown as a function of wavelength. Also shown in the figure are emission spectra
of three crystal scintillators, BGO, LSO and CsI(Tl), and the numerical values of
the emission weighted quantum efficiency. The area under each emission spectrum
is proportional to crystal’s light yield.
28.5. Cherenkov detectors
Revised September 2007 by B.N. Ratcliff (SLAC).
Although devices using Cherenkov radiation are often thought of as particle
identification (PID) detectors, in practice, they are widely used over a much broader
range of applications; including (1) fast particle counters; (2) hadronic particle
identification; and (3) tracking detectors performing complete event reconstruction. A
few examples of specific applications from each category include; (1) the polarization
detector of the SLD [46]; (2) the hadronic PID detectors at the B factory detectors
(DIRC in BaBar [9] and the aerogel threshold Cherenkov in Belle [47]) ; and (3) large
water Cherenkov counters such as Super-Kamiokande [49]. Cherenkov counters contain
two main elements; (1) a radiator through which the charged particle passes, and (2) a
photodetector. As Cherenkov radiation is a weak source of photons, light collection and
detection must be as efficient as possible. The presence of the refractive index n and the
path length of the particle in the radiator in the Cherenkov relations allows tuning these
quantities for a particular experimental application.
Cherenkov detectors utilize one or more of the properties of Cherenkov radiation
discussed in the Passages of Particles through Matter section (Sec. 27 of this Review): the
prompt emission of a light pulse; the existence of a velocity threshold for radiation; and
July 24, 2008 18:04
18 28. Particle detectors
the dependence of the Cherenkov cone half-angle θ
c
and the number of emitted photons
on the velocity of the particle.
The number of photoelectrons (N
p.e.
) detected in a given device is
N
p.e.
= L
α
2
z
2
r
e
m
e
c
2
_
(E) sin
2
θ
c
(E)dE , (28.5)
where L is the path length in the radiator, (E) is the efficiency for collecting the
Cherenkov light and transducing it in photoelectrons, and α
2
/(r
e
m
e
c
2
) = 370 cm
−1
eV
−1
.
The quantities and θ
c
are functions of the photon energy E. However, since the
typical energy dependent variation of the index of refraction is modest, a quantity called
the Cherenkov detector quality factor N
0
can be defined as
N
0
=
α
2
z
2
r
e
m
e
c
2
_
dE , (28.6)
so that
N
p.e.
≈ LN
0
sin
2
θ
c
. (28.7)
We take z = 1, the usual case in high-energy physics, in the following discussion.
This definition of the quality factor N
0
is not universal, nor, indeed, very useful for
situations where the geometrical photon collection efficiency (
coll
) varies substantially for
different tracks. In this case, separate factors for photon collection and detection (
det
),
so that =
coll

det
, are sometimes included on the right hand side of the equation. A
typical value of N
0
for a photomultiplier (PMT) detection system working in the visible
and near UV, and collecting most of the Cherenkov light, is about 100 cm
−1
. Practical
counters, utilizing a variety of different photodetectors, have values ranging between
about 30 and 180 cm
−1
. Radiators can be chosen from a variety of transparent materials
(Sec. 27 of this Review and Table 6.1). In addition to refractive index, the choice
requires consideration of factors such as material density, radiation length, transmission
bandwidth, absorption length, chromatic dispersion, optical workability (for solids),
availability, and cost. Long radiator lengths are required to obtain sufficient numbers of
photons when the momenta of the particle species to be separated are high. Recently,
the gap in refractive index that has traditionally existed between gases and liquid or solid
materials has been partially closed with transparent silica aerogels with indices that range
between about 1.007 and 1.13.
Cherenkov counters may be classified as either imaging or threshold types, depending
on whether they do or do not make use of Cherenkov angle (θ
c
) information. Imaging
counters may be used to track particles as well as identify them.
July 24, 2008 18:04
28. Particle detectors 19
28.5.1. Threshold counters : Threshold Cherenkov detectors [50], in their simplest
form, make a yes/no decision based on whether the particle is above or below the
Cherenkov threshold velocity β
t
= 1/n. A straightforward enhancement of such detectors
uses the number of observed photoelectrons (or a calibrated pulse height) to discriminate
between species or to set probabilities for each particle species [51]. This strategy can
increase the momentum range of particle separation by a modest amount (to a momentum
some 20% above the threshold momentum of the heavier particle in a typical case).
Careful designs give
coll
90%. For a photomultiplier with a typical bialkali cathode,
_

det
dE ≈ 0.27, so that
N
p.e.
/L ≈ 90 cm
−1
sin
2
θ
c
(i.e., N
0
= 90 cm
−1
) . (28.8)
Suppose, for example, that n is chosen so that the threshold for species a is p
t
; that is,
at this momentum species a has velocity β
a
= 1/n. A second, lighter, species b with the
same momentum has velocity β
b
, so cos θ
c
= β
a

b
, and
N
p.e.
/L ≈ 90 cm
−1
m
2
a
−m
2
b
p
2
t
+m
2
a
. (28.9)
For K/π separation at p = p
t
= 1(5) GeV/c, N
p.e.
/L ≈ 16(0.8) cm
−1
for π’s and (by
design) 0 for K’s.
For limited path lengths N
p.e.
can be small, and a minimum number is required to
trigger external electronics. The overall efficiency of the device is controlled by Poisson
fluctuations, which can be especially critical for separation of species where one particle
type is dominant. The effective number of photoelectrons is often less than the average
number calculated above due to additional equivalent noise from the photodetector. It is
common to design for at least 10 photoelectrons for the high velocity particle in order
to obtain a robust counter. As rejection of the particle that is below threshold depends
on not seeing a signal, electronic and other background noise can be important. Physics
sources of light production for the below threshold particle, such as decay of the above
threshold particle or the production of delta rays in the radiator, often limit the separation
attainable, and need to be carefully considered. Well designed, modern multi-channel
counters, such as the ACC at Belle [47], can attain good particle separation performance
over a substantial momentum range for essentially the full solid angle of the spectrometer.
28.5.2. Imaging counters : The most powerful use of the information available from
the Cherenkov process comes from measuring the ring-correlated angles of emission of
the individual Cherenkov photons. Since low-energy photon detectors can measure only
the position (and, perhaps, a precise detection time) of the individual Cherenkov photons
(not the angles directly), the photons must be “imaged” onto a detector so that their
angles can be derived [52]. In most cases the optics map the Cherenkov cone onto
(a portion of) a distorted circle at the photodetector. Though this imaging process is
directly analogous to the familiar imaging techniques used in telescopes and other optical
instruments, there is a somewhat bewildering variety of methods used in a wide variety
of counter types with different names. Some of the imaging methods used include (1)
July 24, 2008 18:04
20 28. Particle detectors
focusing by a lens; (2) proximity focusing (i.e., focusing by limiting the emission region
of the radiation); and (3) focusing through an aperture (a pinhole). In addition, the
prompt Cherenkov emission coupled with the speed of modern photon detectors allows
the use of time imaging, a method which is used much less frequently in conventional
imaging technology. Finally, full tracking (and event reconstruction) can be performed in
large water counters by combining the individual space position and time of each photon
together with the constraint that Cherenkov photons are emitted from each track at a
constant polar angle (Sec. 28.6 of this Review).
In a simple model of an imaging PID counter, the fractional error on the particle
velocity (δ
β
) is given by
δ
β
=
σ
β
β
= tanθ
c
σ(θ
c
) , (28.10)
where
σ(θ
c
) =
σ(θ
i
)
_
N
p.e.
⊕C , (28.11)
where σ(θ
i
) is the average single photoelectron resolution, as defined by the optics,
detector resolution and the intrinsic chromaticity spread of the radiator index of
refraction averaged over the photon detection bandwidth. C combines a number of other
contributions to resolution including, (1) correlated terms such as tracking, alignment,
and multiple scattering, (2) hit ambiguities, (3) background hits from random sources,
and (4) hits coming from other tracks. In many practical cases, the resolution is limited
by these effects.
For a β ≈ 1 particle of momentum (p) well above threshold entering a radiator with
index of refraction (n), the number of σ separation (N
σ
) between particles of mass m
1
and m
2
is approximately
N
σ

|m
2
1
−m
2
2
|
2p
2
σ(θ
c
)

n
2
−1
. (28.12)
In practical counters, the angular resolution term σ(θ
c
) varies between about 0.1 and
5 mrad depending on the size, radiator, and photodetector type of the particular counter.
The range of momenta over which a particular counter can separate particle species
extends from the point at which the number of photons emitted becomes sufficient for
the counter to operate efficiently as a threshold device (∼20% above the threshold for
the lighter species) to the value in the imaging region given by the equation above. For
example, for σ(θ
c
) = 2mrad, a fused silica radiator(n = 1.474), or a flourocarbon gas
radiator (C
5
F
12
, n = 1.0017), would separate π/K’s from the threshold region starting
around 0.15(3) GeV/c through the imaging region up to about 4.2(18) GeV/c at better
than 3σ.
Many different imaging counters have been built during the last several decades [53].
Among the earliest examples of this class of counters are the very limited acceptance
Differential Cherenkov detectors, designed for particle selection in high momentum beam
lines. These devices use optical focusing and/or geometrical masking to select particles
having velocities in a specified region. With careful design, a velocity resolution of
σ
β
/β ≈ 10
−4
–10
−5
can be obtained [50].
July 24, 2008 18:04
28. Particle detectors 21
Practical multi-track Ring-Imaging Cherenkov detectors (generically called RICH
counters) are a more recent development. They have been built in small-aperture
and 4π geometries both as PID counters and as stand-alone detectors with complete
tracking and event reconstruction as discussed more fully below. PID RICH counters are
sometimes further classified by ‘generations’ that differ based on performance, design,
and photodetection techniques.
A typical example of a first generation RICH used at the Z factory e
+
e

colliders [54,55] has both liquid (C
6
F
14
, n = 1.276) and gas (C
5
F
12
, n = 1.0017)
radiators, the former being proximity imaged using the small radiator thickness while
the latter use mirrors. The phototransducers are a TPC/wire-chamber combination
having charge division or pads. They are made sensitive to photons by doping the TPC
gas (usually, ethane/methane) with ∼ 0.05% TMAE (tetrakis(dimethylamino)ethylene).
Great attention to detail is required, (1) to avoid absorbing the UV photons to which
TMAE is sensitive, (2) to avoid absorbing the single photoelectrons as they drift in the
long TPC, and (3) to keep the chemically active TMAE vapor from interacting with
materials in the system. In spite of their unforgiving operational characteristics, these
counters attained good e/π/K/p separation over wide momentum ranges during several
years of operation. In particular, their π/K separation range extends over momenta from
about 0.25 to 20 GeV/c.
Later generation counters [53] generally must operate at much higher particle rates
than the first generation detectors, and utilize different photon detection bandwidths, with
higher readout channel counts, and faster, more forgiving photon detection technology
than the TMAE doped TPC’s just described. Radiator choices have broadened to include
materials such as lithium flouride, fused silica, and aerogel. Vacuum based photodetection
systems (e.g., single or multi anode photomultiplier tubes (PMT), multi channel plate
PMTs (MCP-PMT), or hybrid photodiodes (HPD)) have become increasingly common.
They handle very high rates, may be used with a wide choice of radiators, and may be
sufficiently fast to allow time imaging or the use of time of flight information. Other
fast detection systems that use solid cesium iodide (CSI) photocathodes or triethylamine
(TEA) doping in proportional chambers are useful with certain radiator types and
geometries.
A DIRC (Detector of Internally Reflected Cherenkov light) is a third generation
subtype of a RICH first used in the BaBar detector [48]. It “inverts” the usual principle
for use of light from the radiator of a RICH by collecting and imaging the total internally
reflected light, rather than the transmitted light. A DIRC utilizes the optical material of
the radiator in two ways, simultaneously; first as a Cherenkov radiator, and second, as a
light pipe for the Cherenkov light trapped in the radiator by total internal reflection. The
DIRC makes use of the fact that the magnitudes of angles are preserved during reflection
from a flat surface. This fact, coupled with the high reflection coefficients of the total
internal reflection process (> 0.9995 for higly polished SiO
2
), and the long attenuation
length for photons in high purity fused silica, allows the photons of the ring image to
be transported to a detector outside the path of the particle where they may be imaged
in up to three dimensions (two in space and one in time). The BaBar DIRC uses 144
fused silica radiator bars (1.7 × 3.5 × 490 cm) with the light being focused onto 11 000
July 24, 2008 18:04
22 28. Particle detectors
conventional PMT’s located about 120 cm from the end of the bars by the “pinhole” of
the bar end. DIRC performance can be understood using the formula for (N
σ
) discussed
above. Typically, N
p.e.
is rather large (between 15 and 60) and the Cherenkov polar angle
is measured to about 2.5 mrad. The momentum range with good π/K separation extends
up to about 4 GeV/c, matching the B decay momentum spectrum observed in BaBar.
28.6. Cherenkov tracking calorimeters
Written August 2003 by D. Casper (UC Irvine).
In addition to the specialized applications described in the previous section, Cherenkov
radiation is also exploited in large, ring-imaging detectors with masses measured
in kilotons or greater. Such devices are not subdetector components, but complete
experiments with triggering, tracking, vertexing, particle identification and calorimetric
capabilities, where the large mass of the transparent dielectric medium serves as an active
target for neutrino interactions (or their secondary muons) and rare processes like nucleon
decay.
For volumes of this scale, absorption and scattering of Cherenkov light are non-
negligible, and a wavelength-dependent factor e
−d/L(λ)
(where d is the distance from
emission to the sensor and L(λ) is the attenuation length of the medium) must be
included in the integral of Eq. (28.5) for the photoelectron yield. The choice of medium
is therefore constrained by the refractive index and transparency in the region of
photodetector sensitivity; highly-purified water is an inexpensive and effective choice;
sea-water, mineral oil, polar ice, and D
2
O are also used. Photo-multiplier tubes (PMTs)
on either a volume or surface lattice measure the time of arrival and intensity of
Cherenkov radiation. Hemispherical PMTs are favored for the widest angular acceptance,
and sometimes mounted with reflectors or wavelength-shifting plates to increase the
effective photosensitive area. Gains and calibration curves are measured with pulsed laser
signals transmitted to each PMT individually via optical fiber or applied to the detector
as a whole through one or more diffusing balls.
Volume instrumentation [56] is only cost-effective at low densities, with a spacing
comparable to the attenuation (absorption and scattering) length of Cherenkov light
in the medium (15–40 m for Antarctic ice and ∼45 m in the deep ocean). PMTs are
deployed in vertical strings as modular units which include pressure housings, front-end
electronics and calibration hardware. The effective photocathode coverage of such arrays
is less than 1% but still adequate (using timing information and the Cherenkov angular
constraint) to reconstruct the direction of TeV muons to 1

or better. The size of such
“neutrino telescopes” is limited only by cost once the technical challenges of deployment,
power, signal extraction and calibration in an inaccessible and inhospitable environment
are addressed; arrays up to (1 km)
3
in size are under study or development.
Surface instrumentation [57] allows the target volume to be viewed with higher
photocathode density by a number of PMTs which scales like (volume)
2/3
. To improve
hermeticity and shielding, and to ensure that an outward-going particle’s Cherenkov cone
illuminates sufficient PMTs for reconstruction, a software-defined fiducial volume begins
some distance (∼ 2 m) inside the photosensor surface. Events originating within the
fiducial volume are classified as fully-contained if no particles exit the inner detector, or
July 24, 2008 18:04
28. Particle detectors 23
partially-contained otherwise. An outer (veto) detector, optically separated from the inner
volume and instrumented at reduced density, greatly assists in making this determination
and also simplifies the selection of contained events. The maximum size of a pure surface
array is limited by the attenuation length (∼ 100 m has been achieved for large volumes
using reverse-osmosis water purification), pressure tolerance of the PMTs (< 80 meters
of water, without pressure housings) and structural integrity of the enclosing cavity,
if underground. In practice, these limitations can be overcome by a segmented design
involving multiple modules of the nominal maximum size; megaton-scale devices are
under study.
Cherenkov detectors are excellent electromagnetic calorimeters, and the number of
Cherenkov photons produced by an e/γ is nearly proportional to its kinetic energy.
For massive particles, the number of photons produced is also related to the energy,
but not linearly. For any type of particle, the visible energy E
vis
is defined as the
energy of an electron which would produce the same number of Cherenkov photons.
The number of photoelectrons collected depends on a detector-specific scale factor, with
event-by-event corrections for geometry and attenuation. For typical PMTs, in water
N
p.e.
≈ 15 ξ E
vis
(MeV), where ξ is the effective fractional photosensor coverage; for other
materials, the photoelectron yield scales with the ratio of sin
2
θ
c
over density. At solar
neutrino energies, the visible energy resolution (∼ 30%/
_
ξ E
vis
(MeV)) is about 20%
worse than photoelectron counting statistics would imply. For higher energies, multi-
photoelectron hits are likely and the charge collected by each PMT (rather the number
of PMTs firing) must be used; this degrades the energy resolution to approximately
2%/
_
ξ E
vis
(GeV). In addition, the absolute energy scale must be determined with
sources of known energy. Using an electron LINAC and/or nuclear sources, 0.5–1.5% has
been achieved at solar neutrino energies; for higher energies, cosmic-ray muons, Michel
electrons and π
0
from neutrino interactions allow ∼ 3% absolute energy calibration.
A trigger can be formed by the coincidence of PMTs within a window comparable to
the detector’s light crossing time; the coincidence level thus corresponds to a visible energy
threshold. Physics analysis is usually not limited by the hardware trigger, but rather the
ability to reconstruct events. The interaction vertex can be estimated using timing and
refined by applying the Cherenkov angle constraint to identified ring edges. Multi-ring
events are more strongly constrained, and their vertex resolution is 33–50% better than
single rings. Vertex resolution depends on the photosensor density and detector size, with
smaller detectors performing somewhat better than large ones (∼ 25 cm is typical for
existing devices). Angular resolution is limited by multiple scattering at solar neutrino
energies (25–30

) and improves to a few degrees around E
vis
= 1 GeV.
A non-showering (µ, π
±
, p) track produces a sharp ring with small contributions from
delta rays and other radiated secondaries, while the more diffuse pattern of a showering
(e, γ) particle is actually the superposition of many individual rings from charged shower
products. Using maximum likelihood techniques and the Cherenkov angle constraint,
these two topologies can be distinguished with an efficiency which depends on the
photosensor density and detector size [58]. This particle identification capability has
been confirmed by using cosmic-rays and Michel electrons, as well as charged-particle [59]
and neutrino [60] beams. Large detectors perform somewhat better than smaller ones
July 24, 2008 18:04
24 28. Particle detectors
with identical photocathode coverage; a misidentification probability of ∼ 0.4%/ξ
in the sub-GeV range is consistent with the performance of several experiments for
4% < ξ < 40%. Detection of a delayed coincidence from muon decay offers another, more
indirect, means of particle identification; with suitable electronics, efficiency approaches
100% for µ
+
decays but is limited by nuclear absorption (22% probability in water)
for µ

.
Reconstruction of multiple Cherenkov rings presents a challenging pattern recognition
problem, which must be attacked by some combination of heuristics, maximum likelihood
fitting, Hough transforms and/or neural networks. The problem itself is somewhat
ill-defined since, as noted, even a single showering primary produces many closely-
overlapping rings. For π
0
→γγ two-ring identification, performance falls off rapidly with
increasing π
0
momentum, and selection criteria must be optimized with respect to the
analysis-dependent cost-function for e ↔ π
0
mis-identification. Two representative cases
for ξ = 39% will be illustrated. In an atmospheric neutrino experiment, where π
0
are
relatively rare compared to e
±
, one can isolate a > 90% pure 500 MeV/c π
0
sample with
an efficiency of ∼ 40%. In a ν
e
appearance experiment at E
ν
≤ 1 GeV, where e
±
are
rare compared to π
0
, a 99% pure 500 MeV/c electron sample can be identified with an
efficiency of ∼ 70%. For constant ξ, a larger detector (with, perforce, a greater number of
pixels to sample the light distribution) performs somewhat better at multi-ring separation
than a smaller one. For a more detailed discussion of event reconstruction techniques, see
Ref. 49.
Table 28.5: Properties of Cherenkov tracking calorimeters. LSND was a hybrid
scintillation/Cherenkov detector; the estimated ratio of isotropic to Cherenkov
photoelectrons was about 5:1. MiniBooNE’s light yield also includes a small
scintillation component.
Detector Fiducial mass PMTs ξ p.e./ Dates
(kton) (diameter, cm) MeV
IMB-1 3.3 H
2
O 2048 (12.5) 1% 0.25 1982–85
IMB-3 3.3 H
2
O 2048 (20 +plate) 4.5% 1.1 1987–90
KAM I 0.88/0.78 H
2
O 1000/948 (50) 20% 3.4 1983–85
KAM II 1.04 H
2
O 948 (50) 20% 3.4 1986–90
LSND 0.084 oil+scint. 1220 (20) 25% 33 1993–98
SK-1 22.5 H
2
O 11146 (50) 39% 6 1997–01
SK-2 22.5 H
2
O 5182 (50) 18% 3 2002–
K2K 0.025 H
2
O 680 (50) 39% 6 1999–
SNO 1.0 D
2
O 9456 (20+cone) 55% 9 1999–
MiniBooNE 0.445 oil 1280 (20) 10% 3–4 2002–
July 24, 2008 18:04
28. Particle detectors 25
28.7. Gaseous detectors
28.7.1. Energy loss and charge transport in gases : Written April 2008 by F. Sauli
(CERN) and M. Titov (CEA Saclay).
Gas-filled counters detect and localize the ionization produced by charged particles,
generally after charge multiplication. The peculiar statistics of ionization processes, with
asymmetries in the ionization trails, affect the coordinate determination deduced from
the measurement of drift time or of the center of gravity of the collected charge. For
thin gas layers, the width of the energy loss distribution can be larger than the average,
requiring multiple sample or truncated mean analysis to achieve particle identification
(see Sec. 28.7.4).
The energy loss of charged particles and photons in various materials as a function
of energy is discussed in Sec. 27. Table 28.6 provides values of relevant parameters
in commonly used gases at NTP for unit charge, minimum-ionizing particles [61–67].
Numbers often differ depending on source, and values in the table should be taken only
as approximate. For different conditions and mixtures, and neglecting internal energy
transfer processes, one can use gas-density-dependent composition rules.
Table 28.6: Properties of rare and molecular gases at normal temperature and
pressure (NTP: 20

C, one atm). E
X
, E
I
: first excitation and ionization energy;
W
I
: average energy per ion pair; dE/dx|
min
, N
P
, N
T
: differential energy loss,
primary and total number of electron-ion pairs per cm, for unit charge, minimum
ionizing particles.
Gas Density, E
x
E
I
W
I
dE/dx|
min
N
P
N
T
mg cm
−3
eV eV eV keVcm
−1
cm
−1
cm
−1
Ne 0.839 16.7 21.6 30 1.45 13 50
Ar 1.66 11.6 15.7 25 2.53 25 106
Xe 5.495 8.4 12.1 22 6.87 41 312
CH
4
0.667 8.8 12.6 30 1.61 37 54
C
2
H
6
1.26 8.2 11.5 26 2.91 48 112
iC
4
H
10
2.49 6.5 10.6 26 5.67 90 220
CO
2
1.84 7.0 13.8 34 3.35 35 100
CF
4
3.78 10.0 16.0 54 6.38 63 120
When an ionizing particle passes through the gas it creates electron-ion pairs, but often
the ejected electrons have sufficient energy to further ionize the medium. On average, the
total number of electron-ion pairs (N
T
) is between two and three times larger than the
number of primaries (N
P
) (see Table 28.6).
The probability of releasing an electron of energy E or larger follows an approximate
1/E
2
dependence (Rutherford law), shown in Fig. 28.3 for argon at NTP (dotted line,
left scale). More detailed estimates taking into account the electronic structure of the
July 24, 2008 18:04
26 28. Particle detectors
medium are shown in the figure, for three values of the particle velocity factor βγ [62].
The dot-dashed line provides, on the right scale, the practical range of electrons of energy
E. As an example, there is a 1% probability of creating of an electron of 1 keV or more
in 10 mm of argon, substantially increasing the ionization losses. The practical range of 1
keV electrons in argon (dot-dashed line, right scale) is 70 µm; this contributes to the error
in the coordinate determination.
Figure 28.3: Probability of production of an electron of energy equal or larger
than E (left scale), and range of electrons in argon at NTP (dot-dashed curve, right
scale) [62].
The number of electron-ion pairs per primary interaction, or cluster size, has an
exponentially decreasing probability; for argon, there is about 1% probability for primary
clusters to contain ten or more electron-ion pairs [63].
Once released in the gas, and under the influence of an applied electric field, electrons
and ions drift in opposite directions and diffuse towards the electrodes. The value of drift
velocity and diffusion depends very strongly on the nature of the gas, namely on the
detailed structure of the elastic and inelastic electron-molecule cross-sections. In noble
gases, the inelastic cross section is zero below excitation and ionization thresholds. Fast
gas mixtures are achieved by adding polyatomic gases (usually CH
4
or CO
2
), having
large inelastic cross sections at moderate energies, which results in “cooling” electrons
into an energy range of the Ramsauer-Townsend minimum (located at ∼0.5 eV) of
the elastic scattering cross-section of argon. The reduction in both the total electron
scattering cross-section and the electron energy results in a large increase of electron
drift velocity (for a compilation of electron-molecule cross sections see Ref. 64). Another
principal role of the polyatomic gas is to absorb the ultraviolet (UV) photons emitted
by the excited inert gas atoms. The quenching of UV photons occurs through the
photo-decomposition of polyatomic molecules. Extensive collections of experimental
July 24, 2008 18:04
28. Particle detectors 27
data [65] and theoretical calculations based on transport theory [66] permit estimates of
drift and diffusion properties in pure gases and their mixtures. In a simple approximation,
gas kinetic theory provides the following relation between drift velocity v and mean
electron-molecule collision time τ (Townsend’s expression): v = eEτ/m. Examples for
commonly used gases at NTP are given in Fig. 28.4 and Fig. 28.5, computed with the
program MAGBOLTZ [67] as a function of electric field. For different conditions, the
horizontal axis has to be scaled with the gas density, proportional to 1/P, where P is
the pressure. Standard deviations for longitudinal (σ
L
) and transverse diffusion (σ
T
) are
given for one cm of drift, and scale with the square root of distance. Since the collection
time is inversely related to the drift velocity, diffusion is smaller in fast-counting gases, as
can be seen, for example, in CF
4
. In presence of an external magnetic field, the Lorentz
force acting on electrons between collisions deflects the drifting swarm and modifies the
drift properties. The electron trajectories, velocities and diffusion parameters can be
computed with the program mentioned above. A simple theory (the friction force model)
provides an expression for the vector drift velocity v as a function of electric and magnetic
field vectors E and B, of the Larmor frequency ω = eB/m and the mean collision time τ
between electrons and molecules:
v =
e
m
τ
1 +ω
2
τ
2
_
E+
ωτ
B
(E ×B) +
ω
2
τ
2
B
2
(E· B)B
_
(28.13)
To a good approximation, and for moderate fields, one can assume that the energy
of the electrons is not affected by B, and use for τ the values deduced from the drift
velocity at B = 0 (Townsend expression). For E perpendicular to B, the drift angle to
the electric field vector is tan θ
B
= ωτ and |v | = (E/B)(ωτ/

1 +ω
2
τ
2
). For parallel
electric and magnetic fields, drift velocity and longitudinal diffusion are not affected,
while the transverse diffusion can be strongly reduced: σ
T
(B) = σ
T
(B = 0)/

1 +ω
2
τ
2
.
For example, the large values of ωτ ∼ 20 at 5 T have been measured for Ar/CF
4
/iC
4
H
10
(95:3:2). The dotted line in Fig. 28.5 represents σ
T
for the classic P10 mixture Ar/CH
4
(90:10) at 4 T. This reduction is exploited in time projection chambers (Sec. 28.7.4) to
improve localization accuracy.
In mixtures containing electronegative molecules such as oxygen, water or CF
4
,
electrons can be captured to form negative ions. Capture cross-sections are strongly
energy-dependent, and therefore the capture probability is a function of applied field
and may differ to a great extent for the same amount of additions other mixtures. As
an example, at moderate fields (up to 1 kV/cm) the addition of 0.1% of oxygen to an
Ar/CO
2
mixture results in an electron capture probability about twenty times larger than
the same addition to Ar/CH
4
.
As they experience increasing fields, electrons get enough energy to undergo ionizing
collisions with molecules. Above a gas-dependent threshold, the mean free path for
ionization λ
i
, decreases exponentially with the field; its inverse, α = 1/λ
i
, is the first
Townsend coefficient. As most of the avalanche growth occurs very close to the anodes,
simple electrostatic consideration show that the largest fraction of the detected signal is
due to the motion of ions receding from the wires; the electron component, although very
fast, is rarely seen. This determines the characteristic shape of the detected signals in
July 24, 2008 18:04
28 28. Particle detectors
Figure 28.4: Computed electron drift velocity as a function of electric field in
several gases at NTP [67].
proportional mode, with a fast rise followed by a gradually slower increase. The so-called
ion tail that limits the time resolution of the counter is usually removed by differentiation
of the signal. In uniform fields, N
0
initial electrons multiply over a length x forming an
electron avalanche of size N = N
0
e
αx
; N/N
0
is the gain of the counter. Fig. 28.6 shows
examples of Townsend coefficients for several gas mixtures, computed with MAGBOLTZ.
Figure 28.5: Electron longitudinal diffusion (σ
L
) (dashed lines) and transverse
diffusion (σ
T
) (full lines) for 1 cm of drift. The dotted line shows σ
T
for the P10
mixture at 4T [67].
July 24, 2008 18:04
28. Particle detectors 29
Figure 28.6: Computed first Townsend coefficient α as a function of electric field
in several gases at NTP [67].
Ions released by ionization or produced in the avalanches drift and diffuse under the
influence of the electric field; their drift velocity in the fields encountered in gaseous
counters (up to few kV/cm) is typically three orders of magnitude lower than for electrons.
The ion mobility µ, the ratio of drift velocity to electric field, is constant for a given
ion up to very high fields. For a different temperature and pressure, the mobility can
be computed from the expression µ(P, T) = µ(P
0
, T
0
)(T/P)(P
0
/T
0
). Values of mobility
at NTP (µ
0
) for ions in their own and other gases are given in Table 28.7 [68]. For
mixtures, due to a very effective charge transfer mechanism, only ions with the lowest
ionization potential survive after a short path in the gas. The diffusion of ions follows
a simple square root dependence on the drift time, with a coefficient that depends on
temperature but not on the ion mass. Accumulation of ions in the gas counters may
induce gain reduction and field distortions.
Table 28.7: Mobility of ions in gases at NTP [68].
Gas Ion Mobility (cm
2
V
−1
s
−1
)
He He
+
10.4
Ne Ne
+
4.7
Ar Ar
+
1.54
Ar CH
+
4
1.87
Ar CO
+
2
1.72
CH
4
CH
+
4
2.26
CO
2
CO
+
2
1.09
July 24, 2008 18:04
30 28. Particle detectors
28.7.2. Multi-Wire Proportional Chambers : Written April 2008 by Fabio Sauli
(CERN) and Maxim Titov (CEA Saclay).
Single-wire counters that detect the ionization produced in a gas by charged particle
energy deposit, followed by charge multiplication and collection around a thin wire, have
been used for decades. Good energy resolution is obtained in the streamer mode, and
very large saturated pulses can be detected in the Geiger mode. For an overview see
e.g. Ref. 4.
Multiwire proportional chambers (MWPC’s) [69,70], introduced in the late sixties,
detect and localize energy deposit by charged particles over large areas. A mesh of
parallel anode wires at a suitable potential, inserted between two cathodes, acts (almost)
as an independent set of proportional counters (see Fig. 28.7). Electrons released in the
gas volume drift towards the anodes and avalanche in the increasing field. Analytical
expressions for the electric field can be found in many textbooks; the fields close to the
wires E(r) and in the drift region E
D
are given by the approximations:
E(r) =
CV
0

0
1
r
; E
D
=
CV
0
2
0
s
; C =

0
π(/s) −ln(2πa/s)
where r is the distance from the center of the anode, s the wire spacing, and V
0
the
distance and difference of potential between anode and cathodes. C the capacitance per
unit length of the wires and a the anode wire radius.
Because of electrostatic forces, anode wires are in equilibrium only for a perfect
geometry. Small deviations result in forces displacing the wires alternatively below and
above the symmetry plane, often with catastrophic results. These displacement forces
are countered by the mechanical tension of the wire, up to a maximum stable length
L
M
[71]:
L
M
=
s
CV
0
_

0
T
M
The maximum tension T
M
depends on the wire diameter and modulus of elasticity.
Table 28.8 gives approximate values for tungsten, and the corresponding maximum stable
wire length under reasonable assumptions for the operating voltage V
0
[72]. Internal
supports and spacers have been used in the construction of longer detectors.
Table 28.8: Maximum tension T
M
and stable length L
M
for tungsten wires with
spacing s.
Wire diameter (µm) T
M
(newton) s (mm) L
M
(cm)
10 0.16 1 25
20 0.65 2 85
July 24, 2008 18:04
28. Particle detectors 31
−0.20
−0.15
−0.10
−0.05
0.00
0.05
0.10
0.15
x-axis [cm]
−0.2 −0.3 −0.1 0.0 0.1 0.2 0.3
−0.15
−0.10
−0.05
0.00
0.05
0.10
0.15
0.20
y
-
a
x
i
s

[
c
m
]

y
-
a
x
i
s

[
c
m
]

(a) Multiwire proportional chamber
(b) Drift chamber
Figure 28.7: Electric field lines and equipotentials in (a) a multiwire proportional
chamber and (b) a drift chamber.
Detection of charge over a predefined threshold on the wires provides the event
coordinates with the accuracy of the wire spacing; longitudinal localization can be
obtained by measuring the ratio of collected charge at the two ends of resistive wires.
Making use of the charge profile induced on segmented cathodes, the so-called center-of
gravity (COG) method permits localization of tracks to sub-mm accuracy. Due to the
statistics of energy loss and asymmetric ionization clusters, the position accuracy is
50 µm rms for tracks perpendicular to the wire plane, but degrades to ∼250 µm at 30

to the normal [73]. The intrinsic bi-dimensional characteristic of the COG readout has
found numerous applications in medical imaging.
Drift chambers, developed in the early ’70’s, can estimate the position of a track by
exploiting the arrival time of electrons at the anodes if the time of the interaction is
known. They can achieve rms localization accuracies of 50 µm or better. The distance
between anode wires, several cm, allows covering large areas at reduced cost. Many
July 24, 2008 18:04
32 28. Particle detectors
Figure 28.8: Electric field lines and equipotentials in a multiwire drift module.
Each anode wire is surrounded by six cathode wires, and each cathode wire is
surrounded by three anode wires.
Equipotential
planes
Field-shaping
electrodes
α
Figure 28.9: Schematics drawing of a JET Chamber sector.
designs have been introduced, all aimed at improving performance. In the original design,
a thicker wire at proper voltage between anodes (field wire) reduces the field at the middle
point between anodes, improving charge collection (Fig. 28.8) [74]. Symmetrically
decreasing potentials applied on cathode wires make more uniform and reinforce the
drift fields, resulting in a linear space-to-drift-time relation [75]. In some drift chambers
design, and with the help of suitable voltages applied to field-shaping electrodes, the
electric field structure is adjusted to compensate the distortions.
Sampling the drift time on rows of anodes within the same gas volume, together
July 24, 2008 18:04
28. Particle detectors 33
Figure 28.10: Charge rate dependence of normalized gain in a MWPC [82].
with longitudinal localization through charge division on wires led to the concept of
multiple arrays such as the multi-drift module [76]( Fig. 28.8) and the JET chamber [77](
Fig. 28.9). The time projection chamber [78] combines a measurement of drift time
and charge induction on cathodes to obtain excellent tracking for high multiplicity
topologies occurring at moderate rates (see Sec. 28.7.4). In all cases, a good knowledge
of electron drift velocity and diffusion properties is required. This has to be combined
with the knowledge of the electric fields in the structures, computed with commercial
or custom-developed software [67,79]. For an overview of detectors exploiting the drift
time for coordinate measurement see Refs. 2 and 71. Although very powerful in terms
of performance, multi-wire structures have reliability problems when used in harsh or
hard-to access environments, as a single broken wire can affect the whole detector.
Introduced in the eighties, Straw and drift tube systems make use of large arrays of
wire counters encased in individual envelopes, each acting independently [80]. Suitable
techniques for low-cost mass production have been developed for the needs of large
experiments, as the Transition Radiation Tracker and the Monitored Drift Tubes array
for CERN’s LHC experiments (for a review see Ref. 81).
The production of positive ions in the avalanches and their slow drift before
neutralization result in a rate-dependent accumulation of positive charge in the detector,
with consequent field distortion, gain reduction and degradation of spatial resolution. As
shown in Fig. 28.10 [82], independently from the avalanche size, the proportional gain
drops above a charge production rate around 10
9
electrons per second and mm of wire;
for a proportional gain of 10
4
and 100 electrons released per track, this corresponds to
a particle flux of 10
3
s
−1
mm
−1
(1 kHz/mm
2
for 1 mm wire spacing). Despite various
improvements, position-sensitive detectors based on wire structures are limited by basic
diffusion processes and space charge effects in the gas to localization accuracies of
50–100 µm (see for example Ref. 83).
July 24, 2008 18:04
34 28. Particle detectors
Multiwire and drift chambers have been operated with a variety of gas fillings and
operating modes, depending on experimental requirements. The so-called “Magic Gas”
(a mixture of argon, isobutane and freon) [70] permits very high and saturated gains
(∼ 10
6
), overcoming the electronics limitations of the time, at the cost of reduced survival
due to aging processes. With present-day electronics, proportional gains around 10
4
are
sufficient for detection of minimum ionizing particles, and noble gases with moderate
amounts of polyatomic gases, such as methane or carbon dioxide, are used.
At high radiation fluxes, a fast degradation of detectors due to the formation
of polymers deposits (aging) is often observed. The process has been extensively
investigated, often with conflicting results; a number of culprits has been identified
(organic pollutants, silicone oils); addition of small amounts of water in many (but not
all) cases has been demonstrated to extends the lifetime of the detectors. Addition of
fluorinated gases (e.g., CF
4
) or oxygen results in an etching action that can overcome
polymer formation, or even eliminate deposits, but the issue of long-term survival of gas
detectors in these gases is controversial [84]. Under optimum operating conditions, a
total collected charge of a few coulombs per cm of wire can be reached before degradation,
corresponding, for one mm spacing and at a gain of 10
4
, to a total particle flux of
∼ 10
14
MIP’s/cm
2
.
A new generation of wireless gaseous detectors, micro-pattern gas detectors, has been
developed in the recent years and promises to overcome many of the limitations of MWPC
devices—namely the multi-particle resolution and the rate capability. Their operating
principles and performances are described in Sec. 28.7.3.
28.7.3. Micro-pattern Gas Detectors : Written October 2007 by M. Titov (CEA
Saclay)
Modern photolithographic technology has enabled a series of inventions of novel
Micro-Pattern Gas Detector (MPGD) concepts: Micro-Strip Gas Chamber (MSGC) [85],
GEM [86], Micromegas [87] and many others [88], revolutionizing cell size limits for
many gas detector applications. The MSGC, a concept invented in 1988 by A. Oed, was
the first of the micro-structure gas detectors. Consisting of a set of tiny metal strips
laid on a thin insulating substrate, and alternatively connected as anodes and cathodes,
the MSGC turned out to be easily damaged by discharges induced by heavily ionizing
particles and destroying the fragile electrode structure [89]. The more powerful GEM
and Micromegas concepts fulfill the needs of high-luminosity colliders with increased
reliability in harsh radiation environments. By using fine pitch size compared to classical
wire chambers, these detectors offer intrinsic high rate capability (fast signals with
risetimes of a few ns and full widths of 20–100 ns), excellent spatial resolution (∼ 30 µm),
double track resolution (∼ 500 µm), and single photo-electron time resolution in the ns
range.
The GEM detector was introduced by Fabio Sauli. It consists of a thin-foil copper-
Kapton-copper sandwich chemically perforated to obtain a high density of holes. The hole
diameter is typically between 25 µm and 150 µm, while the pitch varies between 50 µm
and 200 µm. Application of a potential difference between the two sides of the GEM
generates the electric fields indicated in Fig. 28.11. Each hole acts as an independent
proportional counter; electrons released by the ionization in the gas drift into the hole
July 24, 2008 18:04
28. Particle detectors 35
140 µm
50 µm
Figure 28.11: Schematic view and typical dimensions of the hole structure in
the GEM amplification cell. Electric field lines (solid) and equipotentials (dashed)
are shown. On application of a potential difference between the two metal layers
electrons released by ionization in the gas volume above the GEM are guided into
the holes, where charge multiplication occurs in the high field.
HV1
HV2 Micromesh
1
0
0

µ
m
Anode plane
e

E
2
40 kV/cm
Particle
Drift gap
Amplification
gap
Figure 28.12: Schematic drawing and typical dimensions of the Micromegas
detector. Charges produced in the drift gap are drifting to the small amplification
region, limited by the mesh and the anode, where they are amplified.
and multiply in the high electric field (50–70 kV/cm). Most of avalanche electrons
are transferred into the gap below the GEM. Distributing the avalanche multiplication
among several cascading electrodes allows the multi-GEM detectors to operate at overall
gas gain above 10
4
in the presence of highly ionizing particles, while eliminating the
risk of discharges (< 10
−12
per hadron). This is the major advantage of the GEM
technology [90]. A unique property of the GEM detector is the complete decoupling of
the amplification stage (GEM) and the readout electrode (PCB), which operates at unity
gain and serves only as a charge collector.
July 24, 2008 18:04
36 28. Particle detectors
Ioannis Giomataris introduced the micro-mesh gaseous structure (Micromegas), which
is a parallel-plate avalanche counter (Fig. 28.12). It consists of a few mm drift region
(electric field ∼1 kV/cm) and a narrow multiplication gap (25-150 µm, 50–70 kV/cm),
located between a thin metal grid (micromesh) and the readout electrode (strips/pads
of conductor printed on an insulator board). The electric field is homogeneous both in
the drift and amplification gaps. Due to the narrow multiplication region in Micromegas,
locally small variations of the amplification gap are compensated by an inverse variation
of the amplification coefficient and therefore do not induce gain fluctuations. The small
amplification gap is a key element in Micromegas operation, giving rise to its excellent
spatial resolution: 12 µm accuracy (limited by the the micromesh pitch) for MIPs [91],
and very good energy resolution (∼12% FWHM with 6 keV x rays).
Over the past decade GEM and Micromegas detectors have become increasingly
important. COMPASS is a first high-luminosity experiment at CERN which pioneered
the use of large-area (∼ 40 × 40 cm
2
) GEM and Micromegas detectors for high-rate
particle tracking, reaching 25 kHz/mm
2
in the near-beam area. Both technologies have
achieved a tracking efficiency of close to 100% at gas gains of about 10
4
, a spatial
resolution of 70–100 µm and a time resolution of ∼ 10 ns. GEM’s have entered the
LHC program; they will be used for triggering in the LHCb Muon System and in the
TOTEM Telescopes. A time projection chamber (TPC) using GEM or Micromegas as a
gas amplification device is also one of the main options for high-precision tracking at the
International Linear Collider (ILC).
The performance and robustness of MPGD’s have encouraged their applications
in high-energy and neutrino physics, astrophysics, UV and visible photon detection,
nuclear physics and neutron detection domain, radiation therapy and electronic portal
imaging devices. A big step in the direction of the industrial manufacturing of large-size
MPGD’s is the development of the “bulk” Micromegas technology [92]. The basic idea
is to build the whole detector in a single process: the anode plane with copper strips,
a photo-imageable polyimide film and the woven mesh are laminated together at a
high temperature forming a single object. Employing the “bulk” technology, 72 large
Micromegas (34 ×36 cm
2
) planes with an active area of 9 m
2
are being built for the T2K
TPC detector.
Sensitive and low-noise electronics will enlarge the range of the MPGD applications.
Recently, GEM and Micromegas were read out by high-granularity (50 µm pitch) CMOS
pixel chips assembled directly below the GEM or Micromegas amplification structure
and serving as an integrated charge collecting anode [93,94,95]. With this arrangement
avalanche electrons are collected on the top metal layer of the CMOS ASIC; every
input pixel is then directly connected to the amplification, digitization and sparsification
circuits integrated in the underlying active layers of the CMOS technology. GEM coupled
to a VLSI pixel array could serve as a highly efficient x-ray polarimeter, which is able to
reconstruct simultaneously initial direction and dynamics of photoelectron energy loss for
the low energy (<10 keV) x rays. A fine-pitch GEM matching the pitch of pixel ASIC
(50 µm) allows to achieve a superior single-electron avalanche reconstruction accuracy
of 4 µm, which makes it suitable candidate for fast gas photo-multipliers. For minimum
ionizing particle tracks a spatial resolution down to 20 µm was achieved with Medipix2
July 24, 2008 18:04
28. Particle detectors 37
and Timepix CMOS chips coupled to GEM devices. Similar performance is expected for
Micromegas. An attractive solution for the construction of MPGD’s with pixel anode
readout is the integration of the Micromegas amplification and CMOS chip by means of
the “wafer post-processing” technology. The sub-µm precision of the grid dimensions and
avalanche gap size results in a uniform gas gain; the grid hole size, pitch and pattern can
be easily adapted to match the geometry of any pixel readout chip.
Recent developments in radiation hardness research with state-of-the-art MPGD’s are
reviewed in Ref. 96. Better properties of MPGD’s, which are rather insensitive to aging
compared to wire chambers, can be explained by the separation of multiplication (GEM
or parallel plate Micromegas amplification) and anode readout structures, and the lower
electric field strength (∼ 50 kV/cm) in the multiplication region, compared to the anode
wire surface field (∼ 250 kV/cm).
28.7.4. Time-projection chambers : Written September 2007 by D. Karlen (U. of
Victoria and TRIUMF, Canada)
The Time Projection Chamber (TPC) concept, invented by David Nygren in the late
1970’s [78], is the basis for charged particle tracking in a large number of particle and
nuclear physics experiments. A uniform electric field drifts tracks of ions produced by
charged particles traversing a medium, either gas or liquid, towards a surface segmented
into 2D readout pads. The signal amplitudes and arrival times are recorded to provide
full 3D measurements of the particle trajectories. The intrinsic 3D segmentation gives
the TPC a distinct advantage over other large volume tracking detector designs which
record information only in a 2D projection with less overall segmentation, particularly for
pattern recognition in events with large numbers of particles.
Gaseous TPC’s are often designed to operate within a strong magnetic field (typically
parallel to the drift field) so that particle momenta can be estimated from the track
curvature. For this application, precise spatial measurements in the plane transverse to
the magnetic field are most important. Since the amount of ionization along the length of
the track depends on the velocity of the particle, ionization and momentum measurements
can be combined to identify the types of particles observed in the TPC. The estimator for
the energy deposit by a particle is usually formed as the truncated mean of the energy
deposits, using the 50%–70% of the samples with the smallest signals. Variance due to
energetic δ-ray production is thus reduced.
Gas amplification of 10
3
–10
4
at the readout endplate is usually required in order to
provide signals with sufficient amplitude for conventional electronics to sense the drifted
ionization. Until recently, the gas amplification system used in TPC’s have exclusively
been planes of anode wires operated in proportional mode placed close to the readout
pads. Performance has been recently improved by replacing these wire planes with
micro-pattern gas detectors, namely GEM [86] and Micromegas [87] devices. Advances in
electronics miniaturization have been important in this development, allowing pad areas
to be reduced to the 10 mm
2
scale or less, well matched to the narrow extent of signals
produced with micro-pattern gas detectors. Presently, the ultimate in fine segmentation
TPC readout are silicon sensors, with 0.05 mm × 0.05 mm pixels, in combination with
GEM or Micromegas [97]. With such fine granularity it is possible to count the number
of ionization clusters along the length of a track which, in principle, can improve the
July 24, 2008 18:04
38 28. Particle detectors
particle identification capability.
Examples of two modern large volume gaseous TPC’s are shown in Fig. 28.13 and
Fig. 28.14. The particle identification performance is illustrated in Fig. 28.15, for the
original TPC in the PEP-4/9 experiment [98].
Outer containment volume
Inner containment volume
Central electrode
End-plate
Figure 28.13: The ALICE TPC shown in a cutaway view. The drift volume is 5 m
long with a 5 m diameter. Gas amplification is provided by planes of anode wires.
The greatest challenges for a large TPC arise from the long drift distance, typically
100 times further than in a comparable wire chamber design. In particular, the long drift
distance can make the device sensitive to small distortions in the electric field. Distortions
can arise from a number of sources, such as imperfections in the TPC construction,
deformations of the readout surface, or the presence of ions in the active medium.
For a gaseous TPC operated in a magnetic field, the electron drift velocity v is defined
by Eq. (28.13). With a strong magnetic field parallel to the electric field and a gas
with a large value of ωτ (also favored to reduce transverse diffusion as discussed below),
the transverse displacements of the drifting electrons due to electric field distortions
are reduced. In this mode of operation, it is essential to precisely map the magnetic
field as the electron drift lines closely follow the magnetic field lines. Corrections for
electric and/or magnetic field non-uniformities can be determined from control samples
of electrons produced by ionizing the gas with UV laser beams, from photoelectrons
produced on the cathode, or from tracks emanating from calibration reactions.
The long drift distance means that there is a delay, typically 10–100 µs in a large
gaseous TPC, for signals to arrive at the endplate. For experiments with shorter intervals
between events, this can produce ambiguities in the starting time for the drift of
ionization. This can be resolved by matching the TPC data with that from an auxiliary
detector providing additional spatial or timing information.
July 24, 2008 18:04
28. Particle detectors 39
Inner wall and field cage
Outer wall
E, B
directions
Front end
cards
Beam
direction
Central cathode
Central cathode HV
Figure 28.14: One of the 3 TPC modules for the near detector of the T2K
experiment. The drift volume is 2 m×2 m×0.8 m. Micromegas devices are used for
gas amplification and readout.
In a gaseous TPC, the motion of positive ions is much slower than the electrons, and
so the positive ions produced by many events may exist in the active volume. Of greatest
concern is the ions produced in the gas amplification stage. Large gaseous TPC’s built
until now with wire planes have included a gating grid that prevent the positive ions from
escaping into the drift volume in the interval between event triggers. Micro-pattern gas
detectors release much less positive ions than wire planes operating at the same gain,
which may allow operation of a TPC without a gating grid.
Given the long drift distance in a large TPC, the active medium must remain very
pure, as small amounts of contamination can absorb the ionization signal. For example,
in a typical large gaseous TPC, O
2
must be kept below a few parts in 10
5
, otherwise
a large fraction of the drifting electrons will become attached. Special attention must
be made in the choice of construction materials in order to avoid the release of other
electronegative contaminants.
Diffusion degrades the position information of ionization that drifts a long distance.
For a gaseous TPC, the effect can be alleviated by the choice of a gas with low intrinsic
diffusion or by operating in a strong magnetic field parallel to the drift field with a gas
which exhibits a significant reduction in transverse diffusion with magnetic field. For
typical operation without magnetic field, the transverse extent of the electrons, σ
Dx
, is
a few mm after drifting 1 m due to diffusion. With a strong magnetic field, σ
Dx
can be
reduced by as much as a factor of 10,
σ
Dx
(B)/σ
Dx
(0) =
1

1 +ω
2
τ
2
(28.14)
July 24, 2008 18:04
40 28. Particle detectors
µ π K p D
e
E
n
e
r
g
y

d
e
p
o
s
i
t

p
e
r

u
n
i
t

l
e
n
g
t
h

(
k
e
V
/
c
m
)
Momentum (GeV/c)
8
12
16
20
24
28
32
0.1 1 10
Figure 28.15: The PEP4/9-TPC energy deposit measurements (185 samples, 8.5
atm Ar-CH
4
80:20). The ionization rate at the Fermi plateau (at high β) is 1.4
times that for the minimum at lower β. This ratio increases to 1.6 at atmospheric
pressure.
where ωτ is defined above. The diffusion limited position resolution from the information
collected by a single row of pads is
σ
x
=
σ
Dx

n
(28.15)
where n is the effective number of electrons collected by the pad row, giving an ultimate
single row resolution of order 100 µm.
Diffusion is significantly reduced in a negative-ion TPC [99], which uses a special gas
mixture that attaches electrons immediately as they are produced. The drifting negative
ions exhibit much less diffusion than electrons. The slow drift velocity and small ωτ of
negative ions must be compatible with the experimental environment.
The spatial resolution achieved by a TPC is determined by a number of factors
in addition to diffusion. Non-uniform ionization along the length of the track is a
particularly important factor, and is responsible for the so-called “track angle” and
“E×B” effects. If the boundaries between pads in a row are not parallel to the track,
the ionization fluctuations will increase the variance in the position estimate from that
row. For this reason, experiments with a preferred track direction should have pad
boundaries aligned with that direction. Traditional TPC’s with wire plane amplification
suffer from the effects of non-parallel electric and magnetic fields near the wires that
July 24, 2008 18:04
28. Particle detectors 41
rotate ionization segments, thereby degrading the resolution because of the non-uniform
ionization. Micro-pattern gas detectors exhibit a much smaller E×B effect, since their
feature size is much smaller than that of a wire grid.
28.7.5. Transition radiation detectors (TRD’s) : Written August 2007 by
P. Nevski (BNL), A. Romaniouk (Moscow Eng. & Phys. Inst.)
Transition radiation (TR) x rays are produced when a highly relativistic particle

>

10
3
) crosses a refractive index interface, as discussed in Sec. 27.7. The x rays,
ranging from a few keV to a few dozen keV, are emitted at a characteristic angle 1/γ
from the particle trajectory. Since the TR yield is about 1% per boundary crossing,
radiation from multiple surface crossings is used in practical detectors. In the simplest
concept, a detector module might consist of low-Z foils followed by a high-Z active layer
made of proportional counters filled with a Xe-rich gas mixture. The atomic number
considerations follow from the dominant photoelectric absorption cross section per atom
going roughly as Z
n
/E
3
x
, where n varies between 4 and 5 over the region of interest, and
the x-ray energy is E
x
.* To minimize self-absorption, materials such as polypropylene,
Mylar, carbon, and (rarely) lithium are used as radiators. The TR signal in the active
regions is in most cases superimposed upon the particle’s ionization losses. These drop a
little faster than Z/A with increasing Z, providing another reason for active layers with
high Z.
The TR intensity for a single boundary crossing always increases with γ, but for
multiple boundary crossings interference leads to saturation near a Lorentz factor γ
sat
= 0.6 ω
1

1

2
/c [100,101], where ω
1
is the radiator plasma frequency,
1
is its
thickness, and
2
the spacing. In most of the detectors used in particle physics the
radiator parameters are chosen to provide γ
sat
≈ 2000. Those detectors normally work
as threshold devices, ensuring the best electron/pion separation in the momentum range
1 GeV/c
<

p
<

150 GeV/c.
One can distinguish two design concepts—“thick” and “thin” detectors:
1. The radiator, optimized for a minimum total radiation length at maximum TR yield
and total TR absorption, consists of few hundred foils (for instance 300 20 µm thick
polypropylene foils). A dominant fraction of the soft TR photons is absorbed in the
radiator itself. To increase the average TR photon energy further, part of the radiator
far from the active layers is often made of thicker foils. The detector thickness, about
2 cm for Xe-filled gas chambers, is optimized to absorb the shaped x-ray spectrum.
A classical detector is composed of several similar modules which respond nearly
independently. Such detectors were used in the NA34 [102], NOMAD [103], and are
being used in the ALICE [104] experiment.
2. In another TRD concept a fine granular radiator/detector structure exploits the soft
part of the TR spectrum more efficiently. This can be achieved, for instance, by
distributing small-diameter straw-tube detectors uniformly or in thin layers throughout
the radiator material (foils or fibers). Even with a relatively thin radiator stack,
radiation below 5 keV is mostly lost in the radiators themselves. However for photon
* Photon absorption coefficients for the elements (via a NIST link), and dE/dx|
min
and
plasma energies for many materials are given in pdg.lbl.gov/AtomicNuclearProperties.
July 24, 2008 18:04
42 28. Particle detectors
energies above this value the absorption becomes smaller and the radiation can be
registered by several consecutive detector layers, thus creating a strong TR build-up
effect. Examples of the detectors using this approach can be found in both accelerator
(ATLAS [105]) and space (PAMELA [106], AMS [107]) experiments. For example,
in the ATLAS TR tracker charged particles cross about 35 effective straw tube layers
embedded in the radiator material. The effective thickness of the Xe gas per straw is
about 2.3 mm and the average number of foils per straw is about 40 with an effective
foil thickness of about 20 µm.
Both TR photon absorption and the TR build-up significantly affect the detector
performance. Although the values mentioned above are typical for most of the plastic
radiators used with Xe-based detectors, they vary significantly depending on detector
parameters: radiator material, thickness and spacing, the construction of the sensitive
chambers, their position, etc. Thus careful simulations are usually needed to build a
detector optimized for a particular application.
The discrimination between electrons and pions can be based on the charge deposition
measured in each detection module, on the number of clusters—energy depositions
observed above an optimal threshold (usually in the 5 to 7 keV region), or on more
sophisticated methods analyzing the pulse shape as a function of time. The total energy
measurement technique is more suitable for thick gas volumes, which absorb most of the
TR radiation and where the ionization loss fluctuations are small. The cluster-counting
method works better for detectors with thin gas layers, where the fluctuations of the
ionization losses are big. Cluster-counting replaces the Landau-Vavilov distribution of
background ionization energy losses with the Poisson statistics of δ-electrons, responsible
for the distribution tails. The latter distribution is narrower that the Landau-Vavilov
distribution.
The major factor in the performance of any TRD is its overall length. This is
illustrated in Fig. 28.16, which shows, for a variety of detectors, the pion efficiency
at a fixed electron efficiency of 90% as a function of the overall detector length. The
experimental data, covering a range of particle energies from a few GeV to 40 GeV, are
rescaled to an energy of 10 GeV when possible. Phenomenologically, the rejection power
against pions increases as 5 · 10
L/38
, where the range of validity is L ≈ 20–100 cm.
Many recent TRDs combine particle identification with charged-track measurement
in the same detector [104,105]. This provides a powerful tool for electron identification
even at very high particle densities. Another example of this combination is described in
Ref. 108. In this work Si-microstrip detectors operating in a magnetic filed are used both
for particle and TR detection. The excellent coordinate resolution of the Si detectors
allows spatial separation of the TR photons from particle ionization tracks with relatively
modest distances between radiator and detector.
Recent TRDs for particle astrophysics are designed to directly measure the Lorentz
factor of high-energy nuclei by using the quadratic dependence of the TR yield on nuclear
charge [109,110]. The radiator configuration (
1
,
2
) is tuned to extend the TR yield rise
up to γ
<

10
5
using more energetic part of the TR spectrum (up to 100 keV). Exotic
radiator materials such as aluminum and unusual TR detection methods (Compton
scattering) are used such cases [109].
July 24, 2008 18:04
28. Particle detectors 43
0.01
0.1
0.001
10 20 50 200 100
NA34 (HELIOS)
C.Fabjan et al.
R 806
A. Bungener et al.
ZEUS
KEK
UA2
H.Butt et al.
D0
M.Holder et al.
H.Weidkamp
H.Gr ssler et al.
ATLAS
NOMAD
AMS
Total detector length (cm)
P
i
o
n

e
f
f
i
c
i
e
n
c
y
ALICE
PAMELA
:
Figure 28.16: Pion efficiency measured (or predicted) for different TRDs as a
function of the detector length for a fixed electron efficiency of 90%. The plot is
taken from [102] with efficiencies of more recent detectors [105–106] added (ATLAS
to PAMELA).
28.7.6. Resistive-plate chambers : Revised September 2007 by H.R. Band (U.
Wisconsin).
The resistive-plate chamber (RPC) was developed by Santonico and Cardarelli in the
early 1980’s [111] as a low-cost alternative to large scintillator planes.* Most commonly,
an RPC is constructed from two parallel high-resistivity (10
9
–10
13
Ω-cm) glass or phenolic
(Bakelite)/melamine laminate plates with a few-mm gap between them which is filled
with atmospheric-pressure gas. The gas is chosen to absorb UV photons in order to limit
transverse growth of discharges. The backs of the plates are coated with a lower-resistivity
paint or ink (∼10
5
Ω/), and a high potential (7–12 kV) is maintained between them.
The passage of a charged particle initiates an electric discharge, whose size and duration
are limited since the current reduces the local potential to below that needed to maintain
the discharge. The sensitivity of the detector outside of this region is unaffected. The
signal readout is via capacitive coupling to metallic strips on both sides of the detector
which are separated from the high voltage coatings by thin insulating sheets. The x and
y position of the discharge can be measured if the strips on opposite sides of the gap are
orthogonal. When operated in streamer mode, the induced signals on the strips can be
quite large (∼300 mV), making sensitive electronics unnecessary. An example of an RPC
structure is shown in Fig. 28.17.
* It was based on earlier work on a spark counter with one high-resistivity plate [112].
July 24, 2008 18:04
44 28. Particle detectors
Aluminum
x pickup strips
y pickup strips
Insulator
2 mm
Graphite
Insulator
Spacers
Aluminum
H
V
Foam
Bakelite
Bakelite
Gas
Foam
Graphite
2 mm
2 mm
1 mm
1 cm
Figure 28.17: Schematic cross section of a typical RPC, in this case the single-gap
streamer-mode BaBar RPC.
RPC’s have inherent rate limitations since the time needed to re-establish the field
after a discharge is proportional to the chamber capacitance and plate resistance. The
average charge per streamer is 100–1000 pC. Typically, the efficiency of streamer-mode
glass RPC’s begins to fall above ∼0.4 Hz/cm
2
. Because of Bakelite’s lower bulk resistivity,
Bakelite RPC’s can be efficient at 10–100 Hz/cm
2
. The need for higher rate capability
led to the development of avalanche-mode RPC’s, in which the gas and high voltage have
been tuned to limit the growth of the electric discharge, preventing streamer formation.
Typical avalanche-mode RPC’s have a signal charge of about 10 pC and can be efficient
at 1 kHz/cm
2
. The avalanche discharge produces a much smaller induced signal on the
pickup strips (∼1 mV) than streamers, and thus requires a more sophisticated and careful
electronic design.
Many variations of the initial RPC design have been built for operation in either
mode. Efficiencies of
>

92% for single gaps can be improved by the use of two or
more gas gaps with shared pickup strips. Non-flammable and more environmentally
friendly gas mixtures have been developed. In streamer mode, various mixtures of argon
with isobutane and tetrafluoroethane have been used. For avalanche mode operation,
a gas mixture of tetrafluoroethane (C
2
H
2
F
4
) with 2–5% isobutane and 0.4–10% sulfur
hexafluoride (SF
6
) is typical. An example of large-scale RPC use is provided by the muon
system being built for the ATLAS detector, where three layers of pairs of RPC’s are used
to trigger the drift tube arrays between the pairs. The total area is about 10,000 m
2
.
These RPC’s provide a spatial resolution of 1 cm and a time resolution of 1 ns at an
efficiency ≥99%.
Developments of multiple-gap RPC’s [113] lead to RPC designs with much better
timing resolution (∼ 50 ps) for use in time-of-flight particle identification systems. A
pioneering design used by the HARP experiment [114] has two sets of 2 thin gas gaps
(0.3 mm) separated by thin(0.7 mm) glass plates. The outer plates are connected to high
voltage and ground while the inner plate is electrically isolated and floats to a stable
equilibrium potential. The observed RPC intrinsic time resolution of 127 ps may have
July 24, 2008 18:04
28. Particle detectors 45
been limited by amplifier noise. Fonte provides useful review [115] of other RPC designs.
Operational experience with RPC’s has been mixed. Several experiments (e.g., L3 and
HARP) have reported reliable performance. However, the severe problems experienced
with the BaBar RPC’s have raised concerns about the long-term reliability of Bakelite
RPC’s.
Glass RPC’s have had fewer problems, as seen by the history of the BELLE chambers.
A rapid growth in the noise rate and leakage current in some of the BELLE glass RPC’s
was observed during commissioning. It was found that water vapor in the input gas was
reacting with fluorine (produced by the disassociation of the tetrafluoroethane in the
streamers) to produce hydrofluoric acid. The acid etched the glass surfaces, leading to
increased noise rates and lower efficiencies. The use of copper gas piping to insure the
dryness of the input gas stopped the problem. The BELLE RPC’s have now operated
reliably for more than 5 years.
Several different failure modes diagnosed in the first-generation BaBar Bakelite RPC’s
caused the average efficiency of the barrel RPC’s to fall from
>

90% to 35% in five years.
The linseed oil which is used in Bakelite RPC’s to coat the inner surface [116] had not
been completely cured. Under warm conditions (32

C) and high voltage, oil collected
on the spacers between the gaps or formed oil-drop bridges between the gaps. This led
to large leakage currents (50–100 µA in some chambers) which persisted even when the
temperature was regulated at 20

C. In addition, the graphite layer used to distribute the
high voltage over the Bakelite became highly resistive (100 kΩ/ → 10 MΩ/), resulting
in lowered efficiency in some regions and the complete death of whole chambers.
The BaBar problems and the proposed use of Bakelite RPC’s in the LHC detectors
prompted detailed studies of RPC aging and have led to improved construction techniques
and a better understanding of RPC operational limits. The graphite layer has been
improved and should be stable with integrated currents of
<

600 mC/cm
2
. Molded gas
inlets and improved cleanliness during construction have reduced the noise rate of new
chambers. Unlike glass RPC’s, Bakelite RPC’s have been found to require humid input
gases to prevent drying of the Bakelite (increasing the bulk resistivity) which would
decrease the rate capability. Second-generation BaBar RPC’s incorporating many of the
above improvements have performed reliably for over two years [117].
With many of these problems solved, new-generation RPC’s are now being or soon will
be used in about a dozen cosmic-ray and HEP detectors. Their comparatively low cost,
ease of construction, good time resolution, high efficiency, and moderate spatial resolution
make them attractive in many situations, particularly those requiring fast timing and/or
large-area coverage.
July 24, 2008 18:04
46 28. Particle detectors
28.8. Silicon semiconductor detectors
Updated September 2007 by H. Spieler (LBNL).
Semiconductor detectors are widely used in modern high-energy physics experiments.
They are the key ingredient of high-resolution vertex and tracking detectors and are also
used as photodetectors in scintillation calorimeters. The most commonly used material is
silicon, but germanium, gallium-arsenide, CdTe, CdZnTe, and diamond are also useful
in some applications. Integrated circuit technology allows the formation of high-density
micron-scale electrodes on large (10–15 cm diameter) wafers, providing excellent position
resolution. Furthermore, the density of silicon and its small ionization energy result in
adequate signals with active layers only 100–300 µm thick, so the signals are also fast
(typically tens of ns). Semiconductor detectors depend crucially on low-noise electronics
(see Sec. 28.9), so the detection sensitivity is determined by signal charge and capacitance.
For a comprehensive discussion of semiconductor detectors and electronics see Ref. 118.
Silicon detectors are p-n junction diodes operated at reverse bias. This forms a
sensitive region depleted of mobile charge and sets up an electric field that sweeps charge
liberated by radiation to the electrodes. Detectors typically use an asymmetric structure,
e.g. a highly doped p electrode and a lightly doped n region, so that the depletion region
extends predominantly into the lightly doped volume.
The thickness of the depleted region is
W =
_
2 (V +V
bi
)/Ne =
_
2ρµ(V +V
bi
) , (28.16)
where V = external bias voltage
V
bi
= “built-in” voltage (≈ 0.5 V for resistivities typically used in detectors)
N = doping concentration
e = electronic charge
= dielectric constant = 11.9
0
≈ 1 pF/cm
ρ = resistivity (typically 1–10 kΩ cm)
µ = charge carrier mobility
= 1350 cm
2
V
−1
s
−1
for electrons
= 450 cm
2
V
−1
s
−1
for holes
or
W = 0.5 [µm/

Ω-cm· V] ×
_
ρ(V +V
bi
) for n-type material, and
W = 0.3 [µm/

Ω-cm· V] ×
_
ρ(V +V
bi
) for p-type material.
The conductive p and n regions together with the depleted volume form a capacitor with
the capacitance per unit area
C = /W ≈ 1 [pF/cm] /W . (28.17)
In strip and pixel detectors the capacitance is dominated by the fringing capacitance. For
example, the strip-to-strip fringing capacitance is ∼ 1–1.5 pF cm
−1
of strip length at a
strip pitch of 25–50 µm.
July 24, 2008 18:04
28. Particle detectors 47
Measurements on silicon photodiodes [119] show that for photon energies below 4 eV
one electron-hole (e-h) pair is formed per incident photon. The mean energy E
i
required
to produce an e-h pair peaks at 4.4 eV for a photon energy around 6 eV. It assumes
a constant value, 3.67 eV at room temperature, above ∼ 1.5 keV. It is larger than the
bandgap energy because phonon excitation is required for momentum conservation. For
minimum-ionizing particles, the most probable charge deposition in a 300 µm thick silicon
detector is about 3.5 fC (22000 electrons). Since both electronic and lattice excitations
are involved, the variance in the number of charge carriers N = E/E
i
produced by an
absorbed energy E is reduced by the Fano factor F (about 0.1 in Si). Thus, σ
N
=

FN
and the energy resolution σ
E
/E =
_
FE
i
/E. However, the measured signal fluctuations
are usually dominated by electronic noise or energy loss fluctuations in the detector.
Charge collection time decreases with increasing bias voltage, and can be reduced
further by operating the detector with “overbias,” i.e. a bias voltage exceeding the value
required to fully deplete the device. The collection time is limited by velocity saturation
at high fields (approaching 10
7
cm/s at E > 10
4
V/cm); at an average field of 10
4
V/cm
the collection time is about 15 ps/µm for electrons and 30 ps/µm for holes. In typical
fully-depleted detectors 300 µm thick, electrons are collected within about 10 ns, and
holes within about 25 ns.
Position resolution is limited by transverse diffusion during charge collection (typically
5 µm for 300 µm thickness) and by knock-on electrons. Resolutions of 2–4 µm (rms) have
been obtained in beam tests. In magnetic fields, the Lorentz drift deflects the electron
and hole trajectories and the detector must be tilted to reduce spatial spreading (see
“Hall effect” in semiconductor textbooks).
Electrodes can be in the form of cm-scale pads, strips, or µm-scale pixels. Various
readout structures have been developed for pixels, e.g. CCD’s, DEPFET’s, monolithic
pixel devices that integrate sensor and electronics (MAPS), and hybrid pixel devices that
utilize separate sensors and readout IC’s connected by two-dimensional arrays of solder
bumps. For an overview and further discussion see Ref. 118.
Radiation damage occurs through two basic mechanisms:
1. Bulk damage due to displacement of atoms from their lattice sites. This leads to
increased leakage current, carrier trapping, and build-up of space charge that changes
the required operating voltage. Displacement damage depends on the nonionizing
energy loss and the energy imparted to the recoil atoms, which can initiate a chain of
subsequent displacements, i.e., damage clusters. Hence, it is critical to consider both
particle type and energy.
2. Surface damage due to charge build-up in surface layers, which leads to increased
surface leakage currents. In strip detectors the inter-strip isolation is affected. The
effects of charge build-up are strongly dependent on the device structure and on
fabrication details. Since the damage is proportional to the absorbed energy (when
ionization dominates), the dose can be specified in rad (or Gray) independent of
particle type.
The increase in reverse bias current due to bulk damage is ∆I
r
= αΦ per unit volume,
where Φ is the particle fluence and α the damage coefficient (α ≈ 3 × 10
−17
A/cm for
minimum ionizing protons and pions after long-term annealing; α ≈ 2 × 10
−17
A/cm for
July 24, 2008 18:04
48 28. Particle detectors
1 MeV neutrons). The reverse bias current depends strongly on temperature
I
R
(T
2
)
I
R
(T
1
)
=
_
T
2
T
1
_
2
exp
_

E
2k
_
T
1
−T
2
T
1
T
2
__
(28.18)
where E = 1.2 eV, so rather modest cooling can reduce the current substantially (∼ 6-fold
current reduction in cooling from room temperature to 0

C).
Displacement damage forms acceptor-like states. These trap electrons, building up a
negative space charge, which in turn requires an increase in the applied voltage to sweep
signal charge through the detector thickness. This has the same effect as a change in
resistivity, i.e., the required voltage drops initially with fluence, until the positive and
negative space charge balance and very little voltage is required to collect all signal charge.
At larger fluences the negative space charge dominates, and the required operating voltage
increases (V ∝ N). The safe limit on operating voltage ultimately limits the detector
lifetime. Strip detectors specifically designed for high voltages have been extensively
operated at bias voltages >500V. Since the effect of radiation damage depends on the
electronic activity of defects, various techniques have been applied to neutralize the
damage sites. For example, additional doping with oxygen increases the allowable charged
hadron fluence roughly three-fold [120]. Detectors with columnar electrodes normal to
the surface can also extend operational lifetime Ref. 121. The increase in leakage current
with fluence, on the other hand, appears to be unaffected by resistivity and whether the
material is n or p-type. At fluences beyond 10
15
cm
−2
decreased carrier lifetime becomes
critical [122,123].
Strip and pixel detectors have remained functional at fluences beyond 10
15
cm
−2
for
minimum ionizing protons. At this damage level, charge loss due to recombination and
trapping also becomes significant and the high signal-to-noise ratio obtainable with low-
capacitance pixel structures extends detector lifetime. The occupancy of the defect charge
states is strongly temperature dependent; competing processes can increase or decrease the
required operating voltage. It is critical to choose the operating temperature judiciously
(−10 to 0

C in typical collider detectors) and limit warm-up periods during maintenance.
For a more detailed summary see Ref. 124 and and the web-sites of the ROSE and RD50
collaborations at RD48.web.cern.ch/rd48 and RD50.web.cern.ch/rd50.
Currently, the lifetime of detector systems is still limited by the detectors; in
the electronics use of standard “deep submicron” CMOS fabrication processes with
appropriately designed circuitry has increased the radiation resistance to fluences > 10
15
cm
−2
of minimum ionizing protons or pions. For a comprehensive discussion of radiation
effects see Ref. 125.
July 24, 2008 18:04
28. Particle detectors 49
28.9. Low-noise electronics
Revised August 2003 by H. Spieler (LBNL).
Many detectors rely critically on low-noise electronics, either to improve energy
resolution or to allow a low detection threshold. A typical detector front-end is shown in
Fig. 28.18.
OUTPUT
DETECTOR
BIAS
RESISTOR
R
b
C
c
R
s
C
b
C
d
DETECTOR BIAS
PULSE SHAPER PREAMPLIFIER
Figure 28.18: Typical detector front-end circuit.
The detector is represented by a capacitance C
d
, a relevant model for most detectors.
Bias voltage is applied through resistor R
b
and the signal is coupled to the preamplifier
through a blocking capacitor C
c
. The series resistance R
s
represents the sum of all
resistances present in the input signal path, e.g. the electrode resistance, any input
protection networks, and parasitic resistances in the input transistor. The preamplifier
provides gain and feeds a pulse shaper, which tailors the overall frequency response
to optimize signal-to-noise ratio while limiting the duration of the signal pulse to
accommodate the signal pulse rate. Even if not explicitly stated, all amplifiers provide
some form of pulse shaping due to their limited frequency response.
The equivalent circuit for the noise analysis (Fig. 28.19) includes both current and
voltage noise sources. The leakage current of a semiconductor detector, for example,
fluctuates due to electron emission statistics. This “shot noise” i
nd
is represented by
a current noise generator in parallel with the detector. Resistors exhibit noise due to
thermal velocity fluctuations of the charge carriers. This noise source can be modeled
either as a voltage or current generator. Generally, resistors shunting the input act as
noise current sources and resistors in series with the input act as noise voltage sources
(which is why some in the detector community refer to current and voltage noise as
“parallel” and “series” noise). Since the bias resistor effectively shunts the input, as the
capacitor C
b
passes current fluctuations to ground, it acts as a current generator i
nb
and its noise current has the same effect as the shot noise current from the detector.
Any other shunt resistances can be incorporated in the same way. Conversely, the series
resistor R
s
acts as a voltage generator. The electronic noise of the amplifier is described
fully by a combination of voltage and current sources at its input, shown as e
na
and i
na
.
Shot noise and thermal noise have a “white” frequency distribution, i.e. the spectral
power densities dP
n
/df ∝ di
2
n
/df ∝ de
2
n
/df are constant with the magnitudes
July 24, 2008 18:04
50 28. Particle detectors
DETECTOR
C
d
BIAS
RESISTOR
SERIES
RESISTOR
AMPLIFIER +
PULSE SHAPER
R
b
R
s
i
i i
e
e
nd
nb na
ns
na
Figure 28.19: Equivalent circuit for noise analysis.
i
2
nd
= 2eI
d
,
i
2
nb
=
4kT
R
b
,
e
2
ns
= 4kTR
s
, (28.19)
where e is the electronic charge, I
d
the detector bias current, k the Boltzmann constant
and T the temperature. Typical amplifier noise parameters e
na
and i
na
are of order
nV/

Hz and pA/

Hz. Trapping and detrapping processes in resistors, dielectrics and
semiconductors can introduce additional fluctuations whose noise power frequently
exhibits a 1/f spectrum. The spectral density of the 1/f noise voltage is
e
2
nf
=
A
f
f
, (28.20)
where the noise coefficient A
f
is device specific and of order 10
−10
–10
−12
V
2
.
A fraction of the noise current flows through the detector capacitance, resulting in a
frequency-dependent noise voltage i
n
/(ωC
d
), which is added to the noise voltage in the
input circuit. Since the individual noise contributions are random and uncorrelated, they
add in quadrature. The total noise at the output of the pulse shaper is obtained by
integrating over the full bandwidth of the system. Superimposed on repetitive detector
signal pulses of constant magnitude, purely random noise produces a Gaussian signal
distribution.
Since radiation detectors typically convert the deposited energy into charge, the
system’s noise level is conveniently expressed as an equivalent noise charge Q
n
, which is
equal to the detector signal that yields a signal-to-noise ratio of one. The equivalent noise
charge is commonly expressed in Coulombs, the corresponding number of electrons, or
the equivalent deposited energy (eV). For a capacitive sensor
Q
2
n
= i
2
n
F
i
T
S
+e
2
n
F
v
C
2
T
S
+F
vf
A
f
C
2
, (28.21)
July 24, 2008 18:04
28. Particle detectors 51
where C is the sum of all capacitances shunting the input, F
i
, F
v
, and F
vf
depend on the
shape of the pulse determined by the shaper and T
s
is a characteristic time, for example,
the peaking time of a semi-gaussian pulse or the sampling interval in a correlated double
sampler. The form factors F
i
, F
v
are easily calculated
F
i
=
1
2T
S
_

−∞
[W(t)]
2
dt , F
v
=
T
S
2
_

−∞
_
dW(t)
dt
_
2
dt , (28.22)
where for time-invariant pulse-shaping W(t) is simply the system’s impulse response (the
output signal seen on an oscilloscope) with the peak output signal normalized to unity.
For more details see Refs. 126 and 127.
A pulse shaper formed by a single differentiator and integrator with equal time
constants has F
i
= F
v
= 0.9 and F
vf
= 4, independent of the shaping time constant. The
overall noise bandwidth, however, depends on the time constant, i.e. the characteristic
time T
s
. The contribution from noise currents increases with shaping time, i.e., pulse
duration, whereas the voltage noise decreases with increasing shaping time. Noise with a
1/f spectrum depends only on the ratio of upper to lower cutoff frequencies (integrator
to differentiator time constants), so for a given shaper topology the 1/f contribution to
Q
n
is independent of T
s
. Furthermore, the contribution of noise voltage sources to Q
n
increases with detector capacitance. Pulse shapers can be designed to reduce the effect
of current noise, e.g., mitigate radiation damage. Increasing pulse symmetry tends to
decrease F
i
and increase F
v
(e.g., to 0.45 and 1.0 for a shaper with one CR differentiator
and four cascaded integrators). For the circuit shown in Fig. 28.19,
Q
2
n
=
_
2eI
d
+ 4kT/R
b
+i
2
na
_
F
i
T
S
+
_
4kTR
s
+e
2
na
_
F
v
C
2
d
/T
S
+F
vf
A
f
C
2
d
.
(28.23)
As the characteristic time T
S
is changed, the total noise goes through a minimum,
where the current and voltage contributions are equal. Fig. 28.20 shows a typical example.
At short shaping times the voltage noise dominates, whereas at long shaping times the
current noise takes over. The noise minimum is flattened by the presence of 1/f noise.
Increasing the detector capacitance will increase the voltage noise and shift the noise
minimum to longer shaping times.
For quick estimates, one can use the following equation, which assumes an FET
amplifier (negligible i
na
) and a simple CR–RC shaper with time constants τ (equal to
the peaking time):
(Q
n
/e)
2
= 12
_
1
nA· ns
_
I
d
τ + 6 ×10
5
_
kΩ
ns
_
τ
R
b
+ 3.6 ×10
4
_
ns
(pF)
2
(nV)
2
/Hz
_
e
2
n
C
2
τ
.
(28.24)
Noise is improved by reducing the detector capacitance and leakage current, judiciously
selecting all resistances in the input circuit, and choosing the optimum shaping time
constant.
July 24, 2008 18:04
52 28. Particle detectors
E
q
u
i
v
a
l
e
n
t

n
o
i
s
e

c
h
a
r
g
e

(
e
)
10000
5000
2000
1000
100
500
200
1 0.1 0.01 10 100
Shaping time (µs)
1/f noise
C
u
r
r
e
n
t

n
o
i
s
e
V
o
l
t
a
g
e

n
o
i
s
e
Total
Total
Increasing V noise
Figure 28.20: Equivalent noise charge vs shaping time. Changing the voltage or
current noise contribution shifts the noise minimum. Increased voltage noise is
shown as an example.
The noise parameters of the amplifier depend primarily on the input device. In field
effect transistors, the noise current contribution is very small, so reducing the detector
leakage current and increasing the bias resistance will allow long shaping times with
correspondingly lower noise. In bipolar transistors, the base current sets a lower bound on
the noise current, so these devices are best at short shaping times. In special cases where
the noise of a transistor scales with geometry, i.e., decreasing noise voltage with increasing
input capacitance, the lowest noise is obtained when the input capacitance of the
transistor is equal to the detector capacitance, albeit at the expense of power dissipation.
Capacitive matching is useful with field-effect transistors, but not bipolar transistors. In
bipolar transistors, the minimum obtainable noise is independent of shaping time, but
only at the optimum collector current I
C
, which does depend on shaping time.
Q
2
n,min
= 4kT
C

β
DC
_
F
i
F
v
at I
c
=
kT
e
C
_
β
DC
¸
F
v
F
i
1
T
S
, (28.25)
where β
DC
is the DC current gain. For a CR–RC shaper and β
DC
= 100,
Q
n,min
/e ≈ 250
_
C/pF . (28.26)
Practical noise levels range from ∼ 1e for CCD’s at long shaping times to ∼ 10
4
e
in high-capacitance liquid argon calorimeters. Silicon strip detectors typically operate
at ∼ 10
3
e electrons, whereas pixel detectors with fast readout provide noise of several
hundred electrons.
July 24, 2008 18:04
28. Particle detectors 53
In timing measurements, the slope-to-noise ratio must be optimized, rather than the
signal-to-noise ratio alone, so the rise time t
r
of the pulse is important. The “jitter” σ
t
of
the timing distribution is
σ
t
=
σ
n
(dS/dt)
S
T

t
r
S/N
, (28.27)
where σ
n
is the rms noise and the derivative of the signal dS/dt is evaluated at the trigger
level S
T
. To increase dS/dt without incurring excessive noise, the amplifier bandwidth
should match the rise-time of the detector signal. The 10 to 90% rise time of an amplifier
with bandwidth f
U
is 0.35/f
U
. For example, an oscilloscope with 350 MHz bandwidth
has a 1 ns rise time. When amplifiers are cascaded, which is invariably necessary, the
individual rise times add in quadrature.
t
r

_
t
2
r1
+t
2
r2
+... +t
2
rn
Increasing signal-to-noise ratio also improves time resolution, so minimizing the total
capacitance at the input is also important. At high signal-to-noise ratios, the time
jitter can be much smaller than the rise time. The timing distribution may shift with
signal level (“walk”), but this can be corrected by various means, either in hardware or
software [10].
For a more detailed introduction to detector signal processing and electronics see
Ref. 118.
28.10. Calorimeters
A calorimeter is designed to measure the energy deposited in a contained elec-
tromagnetic (EM) or hadronic shower. The characteristic interaction distance for an
electromagnetic interaction is the radiation length X
0
, which ranges from 13.8 g cm
−2
in
iron to 6.0 g cm
−2
in uranium.* Similarly, the characteristic nuclear interaction length
λ
I
varies from 132.1 g cm
−2
(Fe) to 209 g cm
−2
(U). In either case, the calorimeter must
be many interaction lengths deep, where “many” is determined by physical size, cost, and
other factors. EM calorimeters tend to be 15–30 X
0
deep, while hadronic calorimeters
are usually compromised at 5–8 λ
I
. Moreover, in a real experiment there is likely to be
an EM calorimeter in front of the hadronic section, and perhaps a more poorly sampling
catcher in the back, so the hadronic cascade is contained in a succession of different
structures. In all cases there is a premium on high density, to contain the shower as
compactly as possible, and, especially in the EM case, high atomic number.
There are homogeneous and sampling calorimeters. In a homogeneous calorimeter the
entire volume is sensitive, i.e., contributes signal. Homogeneous calorimeters (usually
electromagnetic) may be built with inorganic heavy (high-Z) scintillating crystals such as
BGO, CsI, NaI, and PWO, non-scintillating Cherenkov radiators such as lead glass and
lead fluoride, or ionizing noble liquids. Properties of commonly used inorganic crystal
scintillators can be found in Table 28.4. A sampling calorimeter consists of an active
medium which generates signal and a passive medium which functions as an absorber.
* λ
I
≈ 35 g cm
−2
A
1/3
; for actual values see pdg.lbl.gov/AtomicNuclearProperties.
July 24, 2008 18:04
54 28. Particle detectors
The active medium may be a scintillator, an ionizing noble liquid, a gas chamber, a
semiconductor, or a Cherenkov radiator. The passive medium is usually a material of
high density, such as lead, iron, copper, or depleted uranium.
28.10.1. Electromagnetic calorimeters : Written August 2003 by R.-Y. Zhu
(California Inst. of Technology).
The development of electromagnetic showers is discussed in the section on “Passage of
Particles Through Matter” (Sec. 27 of this Review).
Formulae are given which approximately describe average showers, but since the
physics of electromagnetic showers is well understood, detailed and reliable Monte Carlo
simulation is possible. EGS4 [128] and GEANT [129] have emerged as the standards.
The energy resolution σ
E
/E of a calorimeter can be parametrized as a/

E ⊕b ⊕c/E,
where ⊕ represents addition in quadrature and E is in GeV. The stochastic term
a represents statistics-related fluctuations such as intrinsic shower fluctuations,
photoelectron statistics, dead material at the front of the calorimeter, and sampling
fluctuations. For a fixed number of radiation lengths, the stochastic term a for a sampling
calorimeter is expected to be proportional to
_
t/f, where t is plate thickness and f
is sampling fraction [130,131]. While a is at a few percent level for a homogeneous
calorimeter, it is typically 10% for sampling calorimeters. The main contributions to the
systematic, or constant, term b are detector non-uniformity and calibration uncertainty.
In the case of the hadronic cascades discussed below, non-compensation also contributes
to the constant term. One additional contribution to the constant term for calorimeters
built for modern high-energy physics experiments, operated in a high-beam intensity
environment, is radiation damage of the active medium. This can be minimized by
developing radiation-hard active media [45] and by frequent in situ calibration and
monitoring [44,131]. With effort, the constant term b can be reduced to below one
percent. The term c is due to electronic noise summed over readout channels within a
few Moli`ere radii. The best energy resolution for electromagnetic shower measurement is
obtained in total absorption homogeneous calorimeters, e.g. calorimeters built with heavy
crystal scintillators. These are used when ultimate performance is pursued.
The position resolution depends on the effective Moli`ere radius and the transverse
granularity of the calorimeter. Like the energy resolution, it can be factored as a/

E ⊕b,
where a is a few to 20 mm and b can be as small as a fraction of mm for a dense
calorimeter with fine granularity. Electromagnetic calorimeters may also provide direction
measurement for electrons and photons. This is important for photon-related physics
when there are uncertainties in event origin, since photons do not leave information in
the particle tracking system. Typical photon angular resolution is about 45 mrad/

E,
which can be provided by implementing longitudinal segmentation [132] for a sampling
calorimeter or by adding a preshower detector [133] for a homogeneous calorimeter
without longitudinal segmentation.
Novel technologies have been developed for electromagnetic calorimetry. New heavy
crystal scintillators, such as PWO, LSO:Ce, and GSO:Ce (see Sec. 28.4), have attracted
much attention for homogeneous calorimetry. In some cases, such as PWO, it has received
broad applications in high-energy and nuclear physics experiments. The “spaghetti”
structure has been developed for sampling calorimetry with scintillating fibers as the
July 24, 2008 18:04
28. Particle detectors 55
sensitive medium. The “accordion” structure has been developed for sampling calorimetry
with ionizing noble liquid as the sensitive medium. Table 28.9 provides a brief description
of typical electromagnetic calorimeters built recently for high-energy physics experiments.
Also listed in this table are calorimeter depths in radiation lengths (X
0
) and the achieved
energy resolution. Whenever possible, the performance of calorimeters in situ is quoted,
which is usually in good agreement with prototype test beam results as well as EGS
or GEANT simulations, provided that all systematic effects are properly included.
Detailed references on detector design and performance can be found in Appendix C of
reference [131] and Proceedings of the International Conference series on Calorimetry in
Particle Physics.
Table 28.9: Resolution of typical electromagnetic calorimeters. E is in GeV.
Technology (Exp.) Depth Energy resolution Date
NaI(Tl) (Crystal Ball) 20X
0
2.7%/E
1/4
1983
Bi
4
Ge
3
O
12
(BGO) (L3) 22X
0
2%/

E ⊕0.7% 1993
CsI (KTeV) 27X
0
2%/

E ⊕0.45% 1996
CsI(Tl) (BaBar) 16–18X
0
2.3%/E
1/4
⊕1.4% 1999
CsI(Tl) (BELLE) 16X
0
1.7% for E
γ
> 3.5 GeV 1998
PbWO
4
(PWO) (CMS) 25X
0
3%/

E ⊕0.5%⊕0.2/E 1997
Lead glass (OPAL) 20.5X
0
5%/

E 1990
Liquid Kr (NA48) 27X
0
3.2/%

E ⊕0.42%⊕0.09/E 1998
Scintillator/depleted U 20–30X
0
18%/

E 1988
(ZEUS)
Scintillator/Pb (CDF) 18X
0
13.5%/

E 1988
Scintillator fiber/Pb 15X
0
5.7%/

E ⊕0.6% 1995
spaghetti (KLOE)
Liquid Ar/Pb (NA31) 27X
0
7.5%/

E ⊕0.5%⊕0.1/E 1988
Liquid Ar/Pb (SLD) 21X
0
8%/

E 1993
Liquid Ar/Pb (H1) 20–30X
0
12%/

E ⊕1% 1998
Liquid Ar/depl. U (DØ) 20.5X
0
16%/

E ⊕0.3%⊕0.3/E 1993
Liquid Ar/Pb accordion 25X
0
10%/

E ⊕0.4%⊕0.3/E 1996
(ATLAS)
July 24, 2008 18:04
56 28. Particle detectors
28.10.2. Hadronic calorimeters : [1–6,131] Written April 2008 by D. E. Groom
(LBNL).
Most large hadron calorimeters are sampling calorimeters which are parts of compli-
cated 4π detectors at colliding beam facilities. Typically, the basic structure is plates of
absorber (Fe, Pb, Cu, or occasionally U or W) alternating with plastic scintillators (plates,
tiles, bars), liquid argon (LAr), or gaseous detectors. The ionization is measured directly,
as in LAr calorimeters, or via scintillation light observed by photodetectors (usually
PMT’s). Waveshifting fibers are often used to solve difficult problems of geometry and
light collection uniformity. Silicon sensors are being studied for ILC detectors; in this
case e-h pairs are collected. There are as many variants of these schemes as there
are calorimeters, including variations in geometry of the absorber and sensors, e.g.,
scintillating fibers threading an absorber [134], and the “accordion” LAr detector,
with zig-zag absorber plates to minimize channeling effects. Another departure from the
traditional sandwich structure is the LAr-tube design shown in Fig. 28.21(a).
(a) (b)
W (Cu) absorber
LAr filled
tubes
Hadrons
z
r
φ
scintillator
tile
waveshifter
fiber
PMT
Hadrons
Figure 28.21: (a) ATLAS forward hadronic calorimeter structure (FCal2, 3).
Tubes containing LAr are embedded in a mainly tungsten matrix. (b) ATLAS
central calorimeter wedge; iron with plastic scintillator tile with wavelength-shifting
fiber readout.
A relatively new variant is the use of Cerenkov light in hadron calorimetry. Such
a calorimeter is sensitive to e
±
’s in the EM showers plus a few relativistic pions. An
example is the radiation-hard forward calorimeter in CMS, with iron absorber and quartz
fiber readout by PMT’s.
Ideally, the calorimeter is segmented in φ and θ (or η = −lntan(θ/2)). Fine
segmentation, while desirable, is limited by cost, readout complexity, practical geometry,
and the transverse size of the cascades. An example, a wedge of the ATLAS central barrel
calorimeter, is shown in Fig. 28.21(b).
In an inelastic hadronic collision a significant fraction f
em
of the energy is removed
from further hadronic interaction by the production of secondary π
0
’s and η’s, whose
July 24, 2008 18:04
28. Particle detectors 57
decay photons generate high-energy electromagnetic (EM) cascades. Charged secondaries

±
, p, . . . ) deposit energy via ionization and excitation, but also interact with nuclei,
producing spallation protons and neutrons, evaporation neutrons, and recoiling nuclei
in highly excited states. The charged collision products produce detectable ionization,
as do the showering γ-rays from the prompt de-excitation of highly excited nuclei. The
recoiling nuclei generate little or no detectable signal. The neutrons lose kinetic energy
in elastic collisions over hundreds of ns, gradually thermalize and are captured, with
the production of more γ-rays—usually outside the acceptance gate of the electronics.
Between endothermic spallation losses, nuclear recoils, and late neutron capture, a
significant fraction of the hadronic energy (20%–35%, depending on the absorber and
energy of the incident particle) is invisible.
In contrast to EM showers, hadronic cascade processes are characterized by relatively
few high-energy particles being produced. The lost energy and the π
0
→γγ fraction f
em
are highly variable from event to event. Until there is event-by-event knowledge of both
the invisible energy loss and EM deposit (to be discussed below), the energy resolution of
a hadron calorimeter will remain significantly worse than that of an EM calorimeter.
It has been shown by a simple induction argument and verified by experiment that the
decrease in the average value of the hadronic energy fraction (f
h
= 1 − f
em
) as the
projectile energy E increases is fairly well described by the power law [135,136]
f
h
≈ (E/E
0
)
m−1
(for E > E
0
) , (28.28)
up to at least a few hundred GeV. The exponent m depends logarithmically on the
mean multiplicity and the mean fractional loss to π
0
production in a single interaction.
It is in the range 0.80–0.87, but must be obtained experimentally for each calorimeter
configuration. E
0
is roughly the energy for the onset of inelastic collisions. It is 1 GeV or
a little less for incident pions.
In a hadron-nucleus collision a large fraction of the incident energy is carried
by a “leading particle” with the same quark content as the incident hadron. If the
projectile is a charged pion, the leading particle is usually a pion, which can be neutral
and hence contributes to the EM sector. This is not true for incident protons. The
result is an increased mean hadronic fraction for incident protons: in Eq. (28.29b)
E
0
≈ 2.6 GeV [135,137].
The EM energy deposit is usually detected more efficiently than the hadronic energy
deposit. If the detection efficiency for the EM sector is e and that for the hadronic sector
is h, then the ratio of the mean response to a pion to that for an electron is
π/e = f
em
+f
h
h/e = 1 −(1 −h/e)f
h
(28.29a)
≈ 1 −(1 −h/e)(E/E
0
)
m−1
. (28.29b)
If h = e the hadronic response is not a linear function of energy. Only the product
(1 − h/e)E
1−m
0
can be obtained by measuring π/e as a function of energy. Since 1 − m
is small and E
0
≈ 1 GeV for the usual pion-induced cascades, this fact is usually ignored
and h/e is reported.
July 24, 2008 18:04
58 28. Particle detectors
The discussion above assumes an idealized calorimeter, with the same structure
throughout and without leakage. “Real” calorimeters usually have an EM detector in
front and a coarse “catcher” in the back. Complete containment is generally impractical.
By definition, 0 ≤ f
em
≤ 1. Its variance changes only slowly with energy, but perforce
f
em
→1 as the projectile energy increases. An empirical power law σ
f
em
= (E/E
1
)
1−
(where < 1) describes the energy dependence adequately and has the right asymptotic
properties. For h/e = 1, fluctuations in f
em
significantly contribute to the resolution, in
particular contributing a larger fraction of the variance at high energies. Since the f
em
distribution has a tail on the high side, the calorimeter response is non-Gaussian with a
high-energy tail if h/e < 1. Noncompensation (h/e = 1) thus seriously degrades resolution
as well as producing a nonlinear response.
It is clearly desirable to compensate the response, i.e., to design the calorimeter such
that h/e = 1. This is possible only in a sampling calorimeter, where several variables can
be chosen or tuned:
1. Decrease the EM sensitivity. Because the EM cross sections increase with Z,* and
the absorber usually has higher Z than does the sensor, the EM energy deposit
rate, relative to minimum ionization, is greater than this ratio in the sensor. Lower-Z
inactive cladding, such as the steel cladding on ZEUS U plates, preferentially absorbs
low-energy γ’s in EM showers and thus also lowers the electronic response. G10 signal
boards in the DØ calorimeters have the same effect.
2. Increase the hadronic sensitivity. The abundant neutrons have a large n-p scattering
cross section, with the production of low-energy scattered protons in hydrogenous
sampling materials such as butane-filled proportional counters or plastic scintillator.
(When scattering off a nucleus with mass number A, a neutron can at most lose
4/(1 + A)
2
of its kinetic energy.) The down side in the scintillator case is that the
signal from a highly-ionizing proton stub can be reduced buy as much as 90% by
recombination and quenching (Birk’s Law, Eq. (28.2)).
Fabjan and Willis proposed that the additional signal generated in the aftermath
of fission in
238
U absorber plates should compensate nuclear fluctuations [138]. The
production of fission fragments due to fast n capture was later observed [139]. However,
while a very large amount of energy is released, it is mostly carried by low-velocity fission
fragments which produce very little observable signal. The approach seemed promising
for awhile. But, for example, the compensation observed with the ZEUS
238
U/scintillator
calorimeter was the result of the two mechanisms discussed above.
Motivated very much by the work of Brau, Gabriel, Br¨ uckmann, and Wigmans [140],
several groups built calorimeters which were very nearly compensating. The degree of
compensation was sensitive to the acceptance gate width, and so could be somewhat
tuned. These included (a) HELIOS with 2.5 mm thick scintillator plates sandwiched
between 2 mm thick
238
U plates (one of several structures); σ/E = 0.34/

E was
obtained, (b) ZEUS, 2.6 cm thick scintillator plates between 3.3 mm
238
U plates;
σ/E = 0.35/

E, (c) a ZEUS prototype with 10 mm Pb plates and 2.5 mm scintillator
sheets; σ/E = 0.44/

E, and (d) DØ, where the sandwich cell consists of a 4–6 mm thick
* The asymptotic pair-production cross section scales roughly as Z
0.75
, and |dE/dx|
slowly decreases with increasing Z.
July 24, 2008 18:04
28. Particle detectors 59
238
U plate, 2.3 mm LAr, a G-10 signal board, and another 2.3 mm LAr gap.
A more versatile approach to compensation is provided by a dual-readout calorimeter,
in which the signal is sensed by two readout systems with highly contrasting h/e.
Although the concept is more than two decades old [141], it has only recently been
implemented by the DREAM collaboration [142]. The test beam calorimeter consisted
of copper tubes, each filled with scintillator and quartz fibers. If the two signals Q and S
(quartz and scintillator) are both normalized to electrons, then for each event Eq. (28.29)
takes the form:
Q = E[f
em
+h/e|
Q
(1 −f
em
)]
S = E[f
em
+h/e|
S
(1 −f
em
)] (28.30)
These equations are linear in 1/E and f
em
, and are easily solved for estimators of the
corrected energy and f
em
for each event. Both are subject to resolution effects, but effects
due to fluctuations in f
em
are eliminated. The solution for the corrected energy is given
by [136]:
E =
RS −Q
R −1
, where R =
1 −h/e|
Q
1 −h/e|
S
(28.31)
R is the energy-independent slope of the event locus on a plot of Q vs S. It can
be found either from the fitted slope or by measuring π/e as a function of E. The
DREAM collaboration expects to build a “triple-readout calorimeter” in which a neutron
signal, proportional to the the missing energy, is measured as well on an event-by-event
basis. It is hoped that such a hadronic calorimeter can approach the resolution of an
electromagnetic calorimeter.
The fractional resolution can be represented by
σ
E
=
a
1
(E)

E

¸
¸
¸
¸
1 −
h
e
¸
¸
¸
¸
_
E
E
1
_
1−
(28.32)
The coefficient a
1
is expected to have mild energy dependence for a number of
reasons. For example, the sampling variance is (π/e)E rather than E. (E/E
1
)
1−
is the
parameterization of σ
f
em
discussed above. At a time when data were of lower quality, a
plot of (σ/E)
2
vs 1/E was apparently well-described by a straight line (constant a
1
) with
a finite intercept—the square of the right term in Eq. (28.32), then called “the constant
term.” Modern data show the slight downturn [134].
The average longitudinal distribution rises to a smooth peak about one nuclear
interaction length (λ
I
) into the calorimeter. It then falls somewhat faster than
exponentially. Proton-induced cascades are somewhat shorter and broader than pion-
induced cascades. In either case, the falloff distance increases with energy. In Fig. 28.22
experimental results for 90% and 95% containment are shown, as are calculations using
Bock’s parameterization [143]. A slightly modified form has been used to fit recent
measurements, e.g. to fit profiles measured in the ATLAS central barrel wedges [144].
July 24, 2008 18:04
60 28. Particle detectors
99%
95%
Single Hadron Energy (GeV)
D
e
p
t
h

i
n

I
r
o
n

(
c
m
)
D
e
p
t
h

i
n

I
r
o
n

(
λ
I
)
5 10 50 100 500 1000
50
100
150
200
3
4
5
6
7
8
9
10
11
12
/ Bock param.
/ CDHS data
/ CCFR data
Figure 28.22: Required calorimeter thickness for 95% and 99% hadronic cascade
containment in iron, on the basis of data from two large neutrino detectors and
Bock’s parameterization [143].
The transverse energy deposit is characterized by a central core dominated by EM
cascades, where the neutral meson parents are themselves produced at a variety of
angles. There is a wide “skirt” produced by wide-angle hadronic interactions. The energy
deposited in an annulus dA is adequately described by an exponential core and a Gaussian
halo: dE/dA = (B
1
/r) exp(−r/a
1
) + (B
2
/r) exp(−r
2
/a
2
) [145].
28.10.3. Free electron drift velocities in liquid ionization sensors : Velocities as
a function of electric field strength are given in Refs. 146–147 and are plotted in
Fig. 28.23. Recent precise measurements of the free electron drift velocity in LAr have
been published by W. Walkowiak [148]. The new measurements were motivated by the
design of the ATLAS electromagnetic calorimeter and by inconsistencies in the previous
literature. Velocities are temperatuer depedent and are systematically higher than those
shown in Fig. 28.23.
July 24, 2008 18:04
28. Particle detectors 61
Field St r engt h (kV cm
−1
)
D
r
i
f
t

V
e
l
o
c
i
t
y

(
µ
m
/
n
s
)
TMS
LAr +CH
4
(0.5%)
LAr
2,2,4,4 TMP
0 2 4 6 8 10 12 14 16
0
1
2
3
4
5
6
7
8
9
10
Figure 28.23: Electron drift velocity as a function of field strength for commonly
used liquids.
28.11. Superconducting magnets for collider detectors
Revised September 2005 by A. Yamamoto (KEK); revised October 2001 by R.D. Kephart
(FNAL)
28.11.1. Solenoid Magnets : In all cases SI unit are assumed, so that the magnetic
field, B, is in Tesla, the stored energy, E, is in joules, the dimensions are in meters, and
µ
0
= 4π ×10
−7
.
The magnetic field (B) in an ideal solenoid with a flux return iron yoke, in which the
magnetic field is < 2 T, is given by
B = µ
0
nI (28.33)
where n is the number of turns/meter and I is the current. In an air-core solenoid, the
central field is given by
B(0, 0) = µ
0
nI
L

L
2
+ 4R
2
, (28.34)
where L is the coil length and R is the coil radius.
In most cases, momentum analysis is made by measuring the circular trajectory of the
passing particles according to p = mvγ = q rB, where p is the momentum, m the mass, q
the charge, r the bending radius. The sagitta, s, of the trajectory is given by
s = q B
2
/8p , (28.35)
where is the path length in the magnetic field. In a practical momentum measurement
in colliding beam detectors, it is more effective to increase the magnetic volume than the
July 24, 2008 18:04
62 28. Particle detectors
Table 28.10: Progress of superconducting magnets for particle physics detectors.
Experiment Laboratory B Radius Length Energy X/X
0
E/M
[T] [m] [m] [MJ] [kJ/kg]
TOPAZ* KEK 1.2 1.45 5.4 20 0.70 4.3
CDF Tsukuba/Fermi 1.5 1.5 5.07 30 0.84 5.4
VENUS* KEK 0.75 1.75 5.64 12 0.52 2.8
AMY* KEK 3 1.29 3 40 ‡
CLEO-II Cornell 1.5 1.55 3.8 25 2.5 3.7
ALEPH* Saclay/CERN 1.5 2.75 7.0 130 2.0 5.5
DELPHI* RAL/CERN 1.2 2.8 7.4 109 1.7 4.2
ZEUS* INFN/DESY 1.8 1.5 2.85 11 0.9 5.5
H1* RAL/DESY 1.2 2.8 5.75 120 1.8 4.8
BaBar INFN/SLAC 1.5 1.5 3.46 27 ‡ 3.6
D0 Fermi 2.0 0.6 2.73 5.6 0.9 3.7
BELLE KEK 1.5 1.8 4 42 ‡ 5.3
BES-III

IHEP 1.0 1.475 3.5 9.5 ‡ 2.6
ATLAS-CS

ATLAS/CERN 2.0 1.25 5.3 38 0.66 7.0
ATLAS-BT

ATLAS/CERN 1 4.7–9.75 26 1080 (Toroid)
ATLAS-ET

ATLAS/CERN 1 0.825–5.35 5 2 ×250 (Toroid)
CMS

CMS/CERN 4 6 12.5 2600 ‡ 12

No longer in service

Detector under construction

EM calorimeter is inside solenoid, so small X/X
0
is not a goal
field strength, since
dp/p ∝ p/B
2
, (28.36)
where corresponds to the solenoid coil radius R.
The energy stored in the magnetic field of any magnet is calculated by integrating B
2
over all space:
E =
1

0
_
B
2
dV (28.37)
If the coil thin, (which is the case if it is to superconducting coil), then
E ≈ (B
2
/2µ
0
)πR
2
L . (28.38)
For a detector in which the calorimetry is outside the aperture of the solenoid, the coil
must be thin in terms of radiation and absorption lengths. This usually means that the
coil is superconducting and that the vacuum vessel encasing it is of
July 24, 2008 18:04
28. Particle detectors 63
minimum real thickness and fabricated of a material with long radiation length. There
are two major contributors to the thickness of a thin solenoid:
1) The conductor consisting of the current-carrying superconducting material (usually
NbTi/Cu) and the quench protecting stabilizer (usually aluminum) are wound on the
inside of a structural support cylinder (usually aluminum also). The coil thickness
scales as B
2
R, so the thickness in radiation lengths (X
0
) is
t
coil
/X
0
= (R/σ
h
X
0
)(B
2
/2µ
0
) , (28.39)
where t
coil
is the physical thickness of the coil, X
0
the average radiation length of the
coil/stabilizer material, and σ
h
is the hoop stress in the coil [151]. B
2
/2µ
0
is the
magnetic pressure. In large detector solenoids, the aluminum stabilizer and support
cylinders dominate the thickness; the superconductor (NbTI/Cu) contributes a smaller
fraction. The coil package including the cryostat typically contributes about 2/3 of
the total thickness in radiation lengths.
2) Another contribution to the material comes from the outer cylindrical shell of the
vacuum vessel. Since this shell is susceptible to buckling collapse, its thickness is
determined by the diameter, length and the modulus of the material of which it is
fabricated. The outer vacuum shell represents about 1/3 of the total thickness in
radiation length.
28.11.2. Properties of collider detector magnets :
The physical dimensions, central field stored energy and thickness in radiation lengths
normal to the beam line of the superconducting solenoids associated with the major
collider are given in Table 28.10 [150]. Fig. 28.24 shows thickness in radiation lengths as
a function of B
2
R in various collider detector solenoids.
The ratio of stored energy to cold mass (E/M) is a useful performance measure. It
can also be expressed as the ratio of the stress, σ
h
, to twice the equivalent density, ρ, in
the coil [151]:
E
M
=
_
(B
2
/2µ
0
)dV
ρ V
coil

σ
h

(28.40)
The E/M ratio in the coil is approximately equivalent to H,* the enthalpy of the coil,
and it determines the average coil temperature rise after energy absorption in a quench:
E/M = H(T
2
) −H(T
1
) ≈ H(T
2
) (28.41)
where T
2
is the average coil temperature after the full energy absorption in a quench,
and T
1
is the initial temperature. E/M ratios of 5, 10, and 20 kJ/kg correspond to ∼65,
∼80, and ∼100 K, respectively. The E/M ratios of various detector magnets are shown
in Fig. 28.25 as a function of total stored energy. One would like the cold mass to be
as small as possible to minimize the thickness, but temperature rise during a quench
must also be minimized. An E/M ratio as large as 12 kJ/kg is designed into the CMS
* The enthalpy, or heat content, is called H in the thermodynamics literature. It is not
to be confused with the magnetic field intensity B/µ.
July 24, 2008 18:04
64 28. Particle detectors
0
0.5
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7 8
T
h
i
c
k
n
e
s
s

i
n

r
a
d
i
a
t
i
o
n

l
e
n
g
t
h
s
B
2
R [T
2
m]
ZEUS
VENUS
BESS,
WASA
ALEPH
H1
DELPHI
TOPAZ
CDF
D0
CLEO-II
SSC-SDC
prototype
ATLAS-CS
CELLO
PEP4-
TPC
Figure 28.24: Magnet wall thickness in radiation length as a function of B
2
R for
various detector solenoids. Gray entries are for magnets not listed in Table 28.10.
Open circles are for magnets not designed to be “thin.” The SSC-SDC prototype
provided important R&D for LHC magnets.
solenoid, with the possibility that about half of the stored energy can go to an external
dump resistor. Thus the coil temperature can be kept below 80 K if the energy extraction
system work well. The limit is set by the maximum temperature that the coil design can
tolerate during a quench. This maximum local temperature should be <130 K (50 K +
80 K), so that thermal expansion effects in the coil are manageable.
28.11.3. Toroidal magnets :
Toroidal coils uniquely provide a closed magnetic field without the necessity of an iron
flux-return yoke. Because no field exists at the collision point and along the beam line,
there is, in principle, no effect on the beam. On the other hand, the field profile generally
has 1/r dependence. The particle momentum may be determined by measurements of
the deflection angle combined with the sagitta. The deflection (bending) power BL is
BL ≈
_
R
0
R
i
B
i
R
i
dR
R sin θ
=
B
i
R
i
sin θ
ln(R
0
/R
i
) , (28.42)
where R
i
is the inner coil radius, R
0
is the outer coil radius, and θ is the angle between
the particle trajectory and the beam line axis . The momentum resolution given by the
deflection may be expressed as
∆p
p

p
BL

p sin θ
B
i
R
i
ln(R
0
/R
i
)
. (28.43)
July 24, 2008 18:04
28. Particle detectors 65
0
5
10
15
1 10 100 1000 10
4
E
/
M


[
k
J
/
k
g
]
Stored Energy [MJ]
D0
ZEUS
VENUS
BABAR
CDF
BELLE
DELPHI
ALEPH
H1
CMS
BES-III
TOPAZ
CLEO-II
ATLAS-CS
SSC-SDC
Prototype
Figure 28.25: Ratio of stored energy to cold mass for thin detector solenoids.
Open circles indicate magnets under construction.
The momentum resolution is better in the forward/backward (smaller θ) direction. The
geometry has been found to be optimal when R
0
/R
i
≈ 3–4. In practical designs, the coil
is divided into 6–12 lumped coils in order to have reasonable acceptance and accessibility.
This causes the coil design to be much more complex. The mechanical structure needs to
sustain the decentering force between adjacent coils, and the peak field in the coil is 3–5
times higher than the useful magnetic field for the momentum analysis [149].
28.12. Measurement of particle momenta in a uniform magnetic
field [152,153]
The trajectory of a particle with momentum p (in GeV/c) and charge ze in a constant
magnetic field
−→
B is a helix, with radius of curvature R and pitch angle λ. The radius of
curvature and momentum component perpendicular to
−→
B are related by
p cos λ = 0.3 z BR , (28.44)
where B is in tesla and R is in meters.
The distribution of measurements of the curvature k ≡ 1/R is approximately Gaussian.
The curvature error for a large number of uniformly spaced measurements on the
trajectory of a charged particle in a uniform magnetic field can be approximated by
(δk)
2
= (δk
res
)
2
+ (δk
ms
)
2
, (28.45)
where δk = curvature error
δk
res
= curvature error due to finite measurement resolution
δk
ms
= curvature error due to multiple scattering.
If many (≥ 10) uniformly spaced position measurements are made along a trajectory
July 24, 2008 18:04
66 28. Particle detectors
in a uniform medium,
δk
res
=

L
2
_
720
N + 4
, (28.46)
where N = number of points measured along track
L

= the projected length of the track onto the bending plane
= measurement error for each point, perpendicular to the trajectory.
If a vertex constraint is applied at the origin of the track, the coefficient under the radical
becomes 320.
For arbitrary spacing of coordinates s
i
measured along the projected trajectory and
with variable measurement errors
i
the curvature error δk
res
is calculated from:
(δk
res
)
2
=
4
w
V
ss
V
ss
V
s
2
s
2
−(V
ss
2
)
2
, (28.47)
where V are covariances defined as V
s
m
s
n = s
m
s
n
− s
m
s
n
with s
m
=
w
−1

(s
i
m
/
i
2
) and w =

i
−2
.
The contribution due to multiple Coulomb scattering is approximately
δk
ms

(0.016)(GeV/c)z
Lpβ cos
2
λ
¸
L
X
0
, (28.48)
where p = momentum (GeV/c)
z = charge of incident particle in units of e
L = the total track length
X
0
= radiation length of the scattering medium (in units of length; the X
0
defined
elsewhere must be multiplied by density)
β = the kinematic variable v/c.
More accurate approximations for multiple scattering may be found in the section on
Passage of Particles Through Matter (Sec. 27 of this Review). The contribution to the
curvature error is given approximately by δk
ms
≈ 8s
rms
plane
/L
2
, where s
rms
plane
is defined
there.
References:
1. Experimental Techniques in High Energy Physics, T. Ferbel (ed.) (Addison-Wesley,
Menlo Park, CA, 1987).
2. C. Grupen, Particle Detectors, Cambridge Monographs on Particle Physics, Nuclear
Physics and Cosmology, # 5, Cambridge University Press (1996).
3. K. Kleinknecht, Detectors for Particle Radiation, Cambridge University Press
(1998).
4. G.F. Knoll, Radiation Detection and Measurement, 3nd edition, John Wiley &
Sons, New York (1999).
5. D.R. Green, The Physics of Particle Detectors, Cambridge Monographs on Particle
Physics, Nuclear Physics and Cosmology, # 12, Cambridge University Press (2000).
6. C. Leroy & P.-G. Rancoita, Principles of Radiation Interaction in Matter and
Detection, (World Scientific, Singapore, 2004).
July 24, 2008 18:04
28. Particle detectors 67
7. [Icarus Collaboration], ICARUS-TM/2001-09; LGNS-EXP 13/89 add 2-01.
8. E. Albert, et al., Nucl. Instrum. Methods A409, 70 (1998).
9. B. Aubert, et al., [BaBar Collaboration], Nucl. Instrum. Methods A479, 1 (2002).
10. H. Spieler, IEEE Trans. Nucl. Sci. NS-29, 1142 (1982).
11. K. Arisaka, Nucl. Instrum. Methods A442, 80 (2000).
12. Hamamatsu K.K. Electron Tube Division, Photomultiplier Tubes: Basics and
Applications, 2nd edition (2002);
Can be found under “Photomultiplier Tube Handbook” at
sales.hamamatsu.com/en/products/electron-tube-
division/detectors/photomultiplier-tubes-pmts.php.
13. A. Braem et al., Nucl. Instrum. Methods A518, 574 (2004).
14. R. Arnold et al., Nucl. Instrum. Methods A314, 465 (1992).
15. P. Mangeot et al., Nucl. Instrum. Methods A216, 79 (1983);
R. Apsimon et al., IEEE. Trans. Nucl. Sci. 33, 122 (1986);
R. Arnold et al., Nucl. Instrum. Methods A270, 255, 289 (1988);
D. Aston et al., Nucl. Instrum. Methods A283, 582 (1989).
16. J. Janesick Scientific charge-coupled devices, SPIE Press, Bellingham, WA (2001).
17. R. Haitz et al., J. Appl. Phys. 36, 3123 (1965);
R. McIntyre, IEEE Trans. Electron Devices 13, 164 (1966);
H. Dautet et al., Applied Optics, 32, (21), 3894 (1993);
Perkin-Elmer Optoelectronics, Avalanche Photodiodes: A User’s Guide.
18. P. Buzhan et al., Nucl. Instrum. Methods A504, 48 (2003);
Z. Sadygov et al., Nucl. Instrum. Methods A504, 301 (2003);
V. Golovin and V. Saveliev, Nucl. Instrum. Methods A518, 560 (2004).
19. M. Landstrass et al., Diam. & Rel. Matter, 2, 1033 (1993);
R. McKeag and R. Jackman, Diam. & Rel. Matter, 7, 513 (1998);
R. Brascia et al., Phys. Stat. Sol., 199, 113 (2003).
20. M. Petrov, M. Stapelbroek, and W. Kleinhans, Appl. Phys. Lett. 51, 406 (1987);
M. Atac, M. Petrov, IEEE Trans. Nucl. Sci. 36 163 (1989);
M. Atac et al., Nucl. Instrum. Methods A314, 56 (1994).
21. J.B. Birks, The Theory and Practice of Scintillation Counting, (Pergamon, London,
1964).
22. D. Clark, Nucl. Instrum. Methods 117, 295 (1974).
23. J.B. Birks, Proc. Phys. Soc. A64, 874 (1951).
24. B. Bengston and M. Moszynski, Nucl. Instrum. Methods 117, 227 (1974);
J. Bialkowski, et al., Nucl. Instrum. Methods 117, 221 (1974).
25. C. P. Achenbach, “Active optical fibres in modern particle physics experiments,”
arXiv:nucl-ex/0404008v1.
26. I.B. Berlman, Handbook of Fluorescence Spectra of Aromatic Molecules, 2nd edition
(Academic Press, New York, 1971).
27. C. Zorn, in Instrumentation in High Energy Physics, ed. F. Sauli, (1992, World
Scientific, Singapore) pp. 218–279.
28. T. Foerster, Ann. Phys. 2, 55 (1948).
July 24, 2008 18:04
68 28. Particle detectors
29. J.M. Fluornoy, Conference on Radiation-Tolerant Plastic Scintillators and
Detectors, K.F. Johnson and R.L. Clough editors, Rad. Phys. and Chem., 41 389
(1993).
30. D. Horstman and U. Holm, ibid, 395.
31. D. Blomker, et al., Nucl. Instrum. Methods A311, 505 (1992);
J. Mainusch, et al., Nucl. Instrum. Methods A312, 451 (1992).
32. Conference on Radiation-Tolerant Plastic Scintillators and Detectors, K.F. Johnson
and R.L. Clough editors, Rad. Phys. and Chem., 41 (1993).
33. S.R. Borenstein and R.C. Strand, IEEE Trans. Nuc. Sci. NS-31(1), 396 (1984).
34. P. Sonderegger, Nucl. Instrum. Methods A257, 523 (1987).
35. Achenbach, ibid.
36. C.M. Hawkes, et al., Nucl. Instrum. Methods A292, 329 (1990).
37. A. Lempicki, et al., Nucl. Instrum. Methods A333, 304 (1993);
G. Blasse, Proceedings of the Crystal 2000 International Workshop on Heavy
Scintillators for Scientific and Industrial Applications, Chamonix, France, Sept.
(1992), Edition Frontieres.
38. C. Melcher and J. Schweitzer, Nucl. Instrum. Methods A314, 212 (1992).
39. D.W. Cooke et al., J. Appl. Phys. 88 , (2000) 7360 (2000);
T. Kimble, M Chou and B.H.T. Chai, in Proc. IEEE Nuclear Science Symposium
Conference (2002).
40. K. Takagi and T. Fakazawa, Appl. Phys. Lett. 42, 43 (1983).
41. J.M. Chen, R.H. Mao, L.Y. Zhang and R.Y. Zhu, IEEE Trans. Nuc. Sci. NS-54(3),
718 (2007) and NS-54(4), 1319 (2007).
42. C. Kuntner, et al., Nucl. Instrum. Methods A493, 131 (2002).
43. R.H. Mao, L.Y. Zhang and R.Y. Zhu, presented in The 9th International
Conference on Inorganic Scintillators and their Applications, Winston-Salem, USA,
June (2007), will be published in Nucl. Instrum. Methods.
44. G. Gratta, H. Newman, and R.Y. Zhu, Ann. Rev. Nucl. and Part. Sci. 44, 453
(1994).
45. R.Y. Zhu, Nucl. Instrum. Methods A413, 297 (1998).
46. P. Rowson, et al., Ann. Rev. Nucl. and Part. Sci. 51, 345 (2001).
47. A. Abashian, et al., Nucl. Instrum. Methods A479, 117 (2002).
48. I. Adam, et al., Nucl. Instrum. Methods A538, 281 (2005).
49. M. Shiozawa, [Super-Kamiokande Collaboration], Nucl. Instrum. Methods A433,
240 (1999).
50. J. Litt and R. Meunier, Ann. Rev. Nucl. Sci. 23, 1 (1973).
51. D. Bartlett, et al., Nucl. Instrum. Methods A260, 55 (1987).
52. B. Ratcliff, Nucl. Instrum. Methods A502, 211 (2003).
53. See the RICH Workshop series: Nucl. Instrum. Methods A343, 1 (1993); Nucl.
Instrum. Methods A371, 1 (1996); Nucl. Instrum. Methods A433, 1 (1999); Nucl.
Instrum. Methods A502, 1 (2003); Nucl. Instrum. Methods A553, 1 (2005).
54. M. Cavalli-Sforza, et al., “Construction and Testing of the SLC Cherenkov Ring
Imaging Detector,” IEEE 37, N3:1132 (1990).
July 24, 2008 18:04
28. Particle detectors 69
55. E.G. Anassontzis, et al., “Recent Results from the DELPHI Barrel Ring Imaging
Cherenkov Counter,” IEEE 38, N2:417 (1991).
56. H. Blood, et al., FERMILAB-PUB-76-051-EXP.
57. L. Sulak, HUEP-252 Presented at the Workshop on Proton Stability, Madison,
Wisc. (1978).
58. K.S. Hirata, et al., Phys. Lett. B205, 416 (1988).
59. S. Kasuga, et al., Phys. Lett. B374, 238 (1996).
60. M.H. Ahn, et al., Phys. Rev. Lett. 90, 041801 (2003).
61. L. G. Christophorou, Atomic and Molecular Radiation Physics (Wiley, 1971);
I.B. Smirnov, Nucl. Instrum. Methods A554, 474 (2005);
J. Berkowitz, Atomic and Molecular Photo Absorption (Academic Press, 2002);
http://pdg.lbl.gov/2007/AtomicNuclearProperties.
62. H. Bichsel, Nucl. Instrum. Methods A562, 154 (2006).
63. H. Fischle et al., Nucl. Instrum. Methods A301, 202 (1991).
64. http://rjd.web.cern.ch/rjd/cgi-bin/cross.
65. A. Peisert & F. Sauli, “Drift and Diffusion of Electrons in Gases,” CERN 84-08
(1984).
66. S. Biagi, Nucl. Instrum. Methods A421, 234 (1999).
67. http://consult.cern.ch/writeup.
68. E. McDaniel & E. Mason, The Mobility and Diffusion of Ions in Gases (Wiley,
1973);
G. Shultz et al., Rev. Phys. Appl. 12, 67(1977).
69. G. Charpak et al., Nucl. Instrum. Methods A62, 262 (1968).
70. G. Charpak & F. Sauli, Ann. Rev. Nucl. Sci. 34, 285 (1984).
71. W. Blum & L. Rolandi, Particle Detection with Drift Chambers (Springer-Verlag,
1993).
72. F. Sauli, “Principles of Operation of Multiwire Proportional and Drift Chambers,”
in Experimental Techniques in High Energy Physics, T. Ferbel (ed.) (Addison-
Wesley, Menlo Park, CA, 1987).
73. G. Charpak et al., Nucl. Instrum. Methods A167, 455 (1979).
74. A.H. Walenta et al., Nucl. Instrum. Methods A92, 373 (1971).
75. A. Breskin et al., Nucl. Instrum. Methods A124, 189 (1975).
76. R. Bouclier et al., Nucl. Instrum. Methods A265, 78 (1988).
77. H. Drumm et al., Nucl. Instrum. Methods A176, 333 (1980).
78. D.R. Nygren & J.N. Marx, Phys. Today 31 Vol. 10 (1978).
79. http://www.ansoft.com.
80. P. Beringer et al., Nucl. Instrum. Methods A254, 542 (1987).
81. J. Virdee, Phys. Rep. 403-404,401(2004).
82. H. Walenta, Phys. Scripta 23, 354 (1981).
83. M. Aleksa et al., Nucl. Instrum. Methods A446, 435 (2000).
84. J. Va’vra, Nucl. Instrum. Methods A515, 1 (2003);
M. Titov, “Radiation damage and long-term aging in gas detectors,” arXiv:
physics/0403055.
85. A. Oed, Nucl. Instrum. Methods A263, 351 (1988).
July 24, 2008 18:04
70 28. Particle detectors
86. F. Sauli, Nucl. Instrum. Methods A386, 531 (1997).
87. Y. Giomataris, et al., Nucl. Instrum. Methods A376, 29 (1996).
88. F. Sauli and A. Sharma, Ann. Rev. Nucl. Part. Sci. 49, 341 (1999).
89. Y. Bagaturia, et al., Nucl. Instrum. Methods A490, 223 (2002).
90. F. Sauli, http://gdd.web.cern.ch/GDD/.
91. J. Derre, et al., Nucl. Instrum. Methods A459, 523 (2001).
92. I. Giomataris, et al., Nucl. Instrum. Methods A560, 405 (2006).
93. R. Bellazzini, et al., Nucl. Instrum. Methods A535, 477 (2004).
94. M. Campbell, et al., Nucl. Instrum. Methods A540, 295 (2005).
95. A. Bamberger, et al., Nucl. Instrum. Methods A573, 361 (2007).
96. M. Titov, physics/0403055; Proc. of the 42nd Workshop of the INFN
ELOISATRON Project, “Innovative Detectors For Super-Colliders,” Erice, Italy,
Sept. 28–Oct. 4 (2003).
97. P. Colas et al., Nucl. Instrum. Methods A535, 506 (2004);
M. Campbell et al., Nucl. Instrum. Methods A540, 295 (2005).
98. H. Aihara et al., IEEE Trans. NS30, 63 (1983).
99. C.J. Martoff et al., Nucl. Instrum. Methods A440, 355 (2000).
100. X. Artru et al., Phys. Rev. D12, 1289 (1975).
101. G.M. Garibian et al., Nucl. Instrum. Methods 125, 133 (1975).
102. B. Dolgoshein, Nucl. Instrum. Methods A326, 434 (1993).
103. G. Bassompierre et al., Nucl. Instrum. Methods 411, 63 (1998).
104. A. Andronic, ALICE, Nucl. Instrum. Methods A522, 40 (2004).
105. T. Akesson et al., ATLAS, Nucl. Instrum. Methods A412, 200 (1998).
106. M. Ambriola et al., PAMELA, Nucl. Instrum. Methods A522, 77 (2004).
107. Ph. V. Doetinchem et al., AMS, Nucl. Instrum. Methods A558, 526 (2006).
108. M. Brigida et al., Nucl. Instrum. Methods A572, 440 (2007).
109. M.L. Cherry and G.L. Case, Nucl. Instrum. Methods A522, 73 (2004).
110. D. M¨ uller et al., Nucl. Instrum. Methods A522, 9 (2004).
111. R. Santonico and R. Cardarelli, Nucl. Instrum. Methods A187, 377 (1981).
112. V. V. Parkhomchuck, Yu. N. Pestov, & N. V. Petrovykh, Nucl. Instrum. Methods
93, 269 (1971).
113. E. Cerron Zeballos et al., Nucl. Instrum. Methods A374, 132 (1996).
114. V. Ammosov et al., Nucl. Instrum. Methods A578, 119 (2007).
115. P. Fonte, IEEE. Trans. Nucl. Sci. 49, 881 (2002).
116. M. Abbrescia et al., Nucl. Instrum. Methods A394, 13 (1997).
117. F. Anulli et al., Nucl. Instrum. Methods A552, 276 (2005).
118. H. Spieler, Semiconductor Detector Systems, Oxford Univ. Press, Oxford
(2005) ISBN 0-19-852784-5.
119. F. Scholze et al., Nucl. Instrum. Methods A439, 208 (2000).
120. G. Lindstr¨ om et al., Nucl. Instrum. Methods A465, 60 (2001).
121. C. Da Via et al., Nucl. Instrum. Methods A509, 86 (2003).
122. G. Kramberger et al., Nucl. Instrum. Methods A481, 297 (2002).
123. O. Krasel et al., IEEE Trans. Nucl. Sci NS-51/6,3055 (2004).
124. G. Lindstr¨ om et al., Nucl. Instrum. Methods A426, 1 (1999).
July 24, 2008 18:04
28. Particle detectors 71
125. A. Holmes-Siedle and L. Adams, Handbook of Radiation Effects, 2nd ed., Oxford
2002, ISBN 0-19-850733-X, QC474.H59 2001.
126. V. Radeka, IEEE Trans. Nucl. Sci. NS-15/3, 455 (1968);
V. Radeka, IEEE Trans. Nucl. Sci. NS-21, 51 (1974).
127. F.S. Goulding, Nucl. Instrum. Methods 100, 493 (1972);
F.S. Goulding and D.A. Landis, IEEE Trans. Nucl. Sci. NS-29, 1125 (1982).
128. W.R. Nelson, H. Hirayama, and D.W.O. Rogers, “The EGS4 Code System,”
SLAC-265, Stanford Linear Accelerator Center (Dec. 1985).
129. R. Brun et al., GEANT3, CERN DD/EE/84-1 (1987).
130. D. Hitlin et al., Nucl. Instrum. Methods 137, 225 (1976). See also W. J. Willis and
V. Radeka, Nucl. Instrum. Methods 120, 221 (1974), for a more detailed discussion.
131. R. Wigmans, Calorimetry: Energy Measurement in Particle Physics, International
Series of Monographs on Physics, vol. 107, Clarendon, Oxford (2000).
132. ATLAS Collaboration, The ATLAS Liquid Argon Calorimeter Technical Design
Report, CERN/LHCC 96-41 (1996).
133. CMS Collaboration, The CMS Electromagnetic Calorimeter Technical Design
Report, CERN/LHCC 97-33 (1997).
134. N. Akchurin, et al., Nucl. Instrum. Methods A399, 202 (1997).
135. T.A. Gabriel et al., Nucl. Instrum. Methods A338, 336–347 (1994).
136. D.E. Groom, Nucl. Instrum. Methods A572, 633–653 (2007); Erratum to be
published: see arXiv:physics/0605164v4.
137. N. Akchurin, et al., Nucl. Instrum. Methods A408, 380 (1998).
138. C.W. Fabjan et al., Nucl. Instrum. Methods 141, 61 (1977).
139. C. Leroy, J. Sirois, and R. Wigmans, Nucl. Instrum. Methods A252, 4 (1986).
140. J.E. Brau and T.A. Gabriel, Nucl. Instrum. Methods A238, 489 (1985);
H. Br¨ uckmann and H. Kowalski, ZEUS Int. Note 86/026 DESY, Hamburg (1986);
R. Wigmans, Nucl. Instrum. Methods A259, 389 (1987);
R. Wigmans, Nucl. Instrum. Methods A265, 273 (1988).
141. P. Mockett, “A review of the physics and technology of high-energy calorimeter
devices,” Proc. 11th SLAC Summer Inst. Part. Phy., July 1983, SLAC Report No.
267 (July 1983), p. 42, www.slac.stanford.edu/pubs/confproc/ssi83/
ssi83-008.html.
142. R. Wigmans, “Quartz Fibers and the Prospects for Hadron Calorimetry at the
1% Resolution Level,” Proc. 7th Inter. Conf. on Calorimetry in High Energy
Physics,Tucscon, AZ, Nov. 9–14, 1997, eds. E. Cheu, T. Embry, J. Rutherford, R.
Wigmans (World Scientific, River Edge, NJ, 1998), p. 182;
N. Akchurin et al., Nucl. Instrum. Methods A537, 537 (2005).
143. D. Bintinger, in Proceedings of the Workshop on Calorimetry for the Supercollider,
Tuscaloosa, AL, March 13–17, 1989, edited by R. Donaldson and M.G.D. Gilchriese
(World Scientific, Teaneck, NJ, 1989), p. 91;
R.K. Bock, T. Hansl-Kozanecka, and T.P. Shah, Nucl. Instrum. Methods 186, 533
(1981);
Y.A. Kulchitsky and V.B. Vinogradov, Nucl. Instrum. Methods A455, 499 (2000).
July 24, 2008 18:04
72 28. Particle detectors
144. M. Simonyan et al., ATL-TILECAL-PUB-2007-008 and Nucl. Instrum. Methods
A, to be submitted.
145. D. Acosta et al., Nucl. Instrum. Methods A316, 184 (1997).
146. E. Shibamura et al., Nucl. Instrum. Methods 131, 249 (1975).
147. A.O. Allen, “Drift Mobilities and Conduction Band Energies of Excess Electrons in
Dielectric Liquids,” NSRDS-NBS-58 (1976).
148. W. Walkowiak, Nucl. Instrum. Methods A449, 288 (2000).
149. T. Taylor, Phys. Scr. 23, 459 (1980).
150. A. Yamamoto, Nucl. Instr. Meth. A494, 255 (2003).
151. A. Yamamoto, Nucl. Instr. Meth. A453, 445 (2000).
152. R.L. Gluckstern, Nucl. Instrum. Methods 24, 381 (1963).
153. V. Karim¨ aki, Nucl. Instrum. Methods A410, 284 (1998).
July 24, 2008 18:04

2

28. Particle detectors
28.11.3. Toroidal magnets . . . . . . . . . . . . . . . . . . . . . . . 28.12. Measurement of particle momenta in a uniform magnetic field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 65 66

28. PARTICLE DETECTORS

Revised 2007 (see the various sections for authors).

28.1. Summary of detector spatial resolution, temporal resolution, and deadtime
In this section we give various parameters for common detector components. The quoted numbers are usually based on typical devices, and should be regarded only as rough approximations for new designs. More detailed discussions of detectors and their underlying physics can be found in books by Ferbel [1], Grupen [2], Kleinknecht [3], Knoll [4], Green [5], and Leroy & Rancoita [6]. In Table 28.1 are given typical resolutions and deadtimes of common detectors. Table 28.1: Typical resolutions and deadtimes of common detectors. Revised September 2003 by R. Kadel (LBNL). Resolution Dead Accuracy (rms) Time Time 1 ms 50 msa 2 µs 100 ms 2 ns 200 ns 100 ns 2 nse f 10 ns 100 ps/n — — ∼ 200 ns ∼ 2 µs < 10 ns — 1–2 ns — h h h

Detector Type

Bubble chamber 10–150 µm Streamer chamber 300 µm Proportional chamber 50–300 µmb,c,d Drift chamber 50–300 µm Scintillator — Emulsion 1 µm Liquid Argon Drift [7] ∼175–450 µm Gas Micro Strip [8] 30–40 µm Resistive Plate chamber [9] 10 µm Silicon strip Silicon pixel
a b c d e f g

pitch/(3 to 7)g h 2 µmi

Multiple pulsing time. 300 µm is for 1 mm pitch. Delay line cathode readout can give ±150 µm parallel to anode wire. √ wirespacing/ 12. For two chambers. n = index of refraction. The highest resolution (“7”) is obtained for small-pitch detectors ( 25 µm) with pulse-height-weighted center finding.
July 24, 2008 18:04

28. Particle detectors 3
h i

Limited by the readout electronics [10]. (Time resolution of ≤ 25 ns is planned for the ATLAS SCT.) Analog readout of 34 µm pitch, monolithic pixel detectors.

28.2. Photon detectors
Updated August 2007 by D. Chakraborty (Northern Illinois U) and T. Sumiyoshi (Tokyo Metro U). Most detectors in high-energy, nuclear, and astrophysics rely on the detection of photons in or near the visible range, 100 nm λ 1000 nm, or E ≈ a few eV. This range covers scintillation and Cherenkov radiation as well as the light detected in many astronomical observations. Generally, photodetection involves generating a detectable electrical signal proportional to the (usually very small) number of incident photons. The process involves three distinct steps: 1. Generation of a primary photoelectron or electron-hole (e-h) pair by an incident photon by the photoelectric or photoconductive effect, 2. Amplification of the p.e. signal to detectable levels by one or more multiplicative bombardment steps and/or an avalanche process (usually), and, 3. Collection of the secondary electrons to form the electrical signal. The important characteristics of a photodetector include the following in statistical averages: 1. Quantum efficiency (QE or Q ): the number of primary photoelectrons generated per incident photon (0 ≤ Q ≤ 1; in silicon more than one e-h pair per incident photon can be generated for λ < 165 nm), ∼ 2. Collection efficiency (CE or C ): the overall acceptance factor other than the generation of photoelectrons (0 ≤ C ≤ 1), 3. Gain (G): the number of electrons collected for each photoelectron generated, 4. Dark current or dark noise: the electrical signal when there is no photon, 5. Energy resolution: electronic noise (ENC or Ne ) and statistical fluctuations in the amplification process compound the Poisson distribution of nγ photons from a given source: 2 Ne fN σ(E) , (28.1) + = E nγ Q C Gnγ Q C where fN , or the excess noise factor (ENF), is the contribution to the energy distribution variance due to amplification statistics [11], 6. Dynamic range: the maximum signal available from the detector (this is usually expressed in units of the response to noise-equivalent power, or NEP, which is the optical input power that produces a signal-to-noise ratio of 1), 7. Time dependence of the response: this includes the transit time, which is the time between the arrival of the photon and the electrical pulse, and the transit time spread, which contributes to the pulse rise time and width, and 8. Rate capability: inversely proportional to the time needed, after the arrival of one photon, to get ready to receive the next.
July 24, 2008 18:04

In other types. the volume available to accommodate the detectors. together with a range of λ where the QE is comparable to its maximum. offer a wide range of solutions to choose from. as well ambient radiation of different types and. while a large gain is desirable.7–10 102 –105 0. In such cases. Type λ (nm) Q C Gain Risetime (ns) Area (mm2 ) 1-p.15–0. The QE is a strong function of the photon wavelength (λ). Some key characteristics are summarized in Table 28.e noise (Hz) 10–104 0. magnetic field.2. temperature. Particle detectors Table 28.3 102 –104 7 102 –105 O(0. characteristics of the environment such as chemical composition. area. Several technologies employing different phenomena for the three steps described above.1) O(10) O(1) 10–103 ∼1 1–10 ∼ 10 1 500–3000 100–5000 500–3500 10–6000 ∼2 × 104 ∼600 300–2000 O(10) 400–1400 O(100) 30–60 O(10) ∼7 ∼1 ∗These devices often come in multi-anode configurations. The time resolution of the devices listed here vary in the 10–2000 ps range. ambient background.15–0. attempts to increase the gain for a given device also increases the ENF and after-pulsing (“echos” of the main pulse). mode of operation (continuous or triggered). bias (high-voltage) requirements. Optimization of these factors involves many trade-offs and vary widely between applications.15–0. Other important considerations also are highly application-specific. 2008 18:04 .25 103 –107 MCP∗ 100–650 0. cost. These include the photon flux and wavelength range.3 103 –106 GPM APD 300–1700 ∼0. calibration needs. Spatial uniformity and linearity with respect to the number of photons are highly desirable in a photodetector’s response.4 28.10 103 –107 0. and many variants within each. noise. a higher QE often requires a compromise in the timing properties.3 105 –106 VLPC 500–600 ∼0. For example.7 10–108 PPD 400–550 0. July 24. power consumption.1–200 10–103 10–103 1–103 O(106 ) O(104 ) HV (V) Price (USD) PMT∗ 115–1100 0. and price are to be considered on a “per readout-channel” basis. coverage of large areas by focusing increases the transit time spread.15–0. aging. and so on.01–0. In solid-state devices.9 ∼5 × 104 0.2: Representative characteristics of some photodetectors commonly used in particle physics. the total area to be covered and the efficiency required.1–0. and is usually quoted at maximum. The salient features of the main technologies and the common variants are described below.3 103 –104 HPD∗ 115–850 ∗ 115–500 0.

whence each channel is only capable of a binary output. GaAs(Cs).1. and fine mesh. e. July 24. The multiplication process is repeated typically 10 times in series to generate a sufficient number of electrons. proximity-focused PMT’s. and hybrid photodetectors. However. CsTe. while in the latter. A high-permeability metal shield is often necessary. chosen for the wavelength band of interest. the photocathode material rests on a separate surface that the incident photons strike. the inner surface of each cylindrical tube serves as a continuous dynode for the entire cascade of multiplicative bombardments initiated by a photoelectron. and MgF2 or LiF for XUV.1. microchannel plates.2. etc. anode configurations. In the former. etc. 28.2. venetian blind. the fine-mesh types. They are categorized by the window materials. multi-alkali (SbNa2 KCs).1. separated by O(1 mm) gaps.7–0. bi-alkali (SbRbCs. Sensitive wavelengths and peak quantum efficiencies for these materials are summarized in Table 28. Microchannel plates: A typical Microchannel plate (MCP) photodetector consists of one or more ∼2 mm thick glass plates with densely packed O(10 µm)-diameter cylindrical holes. dynode structures.8 (depending on the dynode material). Vacuum photodetectors : Vacuum photodetectors can be broadly subdivided into three types: photomultiplier tubes. Particle detectors 5 28. line focusing. sitting between the transmission-type photocathode and anode planes. PMT’s are vulnerable to magnetic fields—sometimes even the geomagnetic field causes large orientation-dependent gain changes. vacuum photomultiplier tubes (PMT) has been employed by a vast majority of all particle physics experiments to date [11]. The total gain of a PMT depends on the applied high voltage V as G = AV kn . G is in the range of 105 –106 . GaAsP.2. Typically. Instead of discrete dynodes. which are collected at the anode for delivery to the external circuit. Common window materials are borosilicate glass for IR to near-UV. In some cases. A large variety of PMT’s.3. Both “transmission-” and “reflection-type” PMT’s are widely used. With e. and A a constant (which also depends on n). but the sum of all channel outputs remains proportional to the number of photons received so long as the photon flux is low enough to ensure that the probability of a single channel receiving more than one photon during a single time gate is negligible. including many just recently developed. can be used even in a high magnetic field (≥ 1 T) if the electron drift direction is parallel to the field. box and grid. The cathode material has a low work function. covers a wide span of wavelength ranges from infrared (IR) to extreme ultraviolet (XUV) [12]. SbKCs). photocathode materials.g.g. Photomultiplier tubes: A versatile class of photon detectors. fused quartz and sapphire (Al2 O3 ) for UV.28. the photocathode material is deposited on the inside of a transparent window through which the photons enter. limited spatial resolution can be obtained by using a mosaic of multiple anodes. which then emits a few (∼ 5) secondary electrons. two-level discrimination the effective time resolution can be much better. 2008 18:04 . 28.and/or Sb-based compounds such as CsI.2. or dynode. When a photon hits the cathode and liberates an electron (the photoelectric effect).1. The choice of photocathode materials include a variety of mostly Cs. where k ≈ 0. or “channels”. the latter is accelerated and guided by electric fields to impinge on a secondary-emission electrode. n is the number of dynodes in the chain. Typical dynode structures used in PMT’s are circular cage. Pulse risetimes are usually in the few nanosecond range. Gain fluctuations can be minimized by operating in a saturation mode.

and can tolerate random magnetic fields up to 0. Since the gain is achieved in a single step. although not so much in HEP or astrophysics.3. thanks to the Fano effect discussed in Sec. have excellent time resolution (∼20 ps). The gain nearly equals the maximum number of e-h pairs that could be created from the entire kinetic energy of the accelerated electron: G ≈ eV /w.23 0. they suffer from relatively long recovery time per channel and short lifetime. proximity-focused HPD’s can be an alternative to MCP’s. However.g. V is the applied potential difference.20 0. Particle detectors MCP’s are thin. MCP’s are widely employed as image-intensifiers. when counting small numbers of photons. Photocathode material CsI CsTe Bi-alkali Multi-alkali GaAs(Cs)∗ GaAsP(Cs) ∗ Reflection λ (nm) Window material Peak Q (λ/nm) 115–200 MgF2 115–240 MgF2 300–650 Borosilicate 160-650 Quartz 300–850 Borosilicate 160-850 Quartz 160–930 Quartz 300-750 Borosilicate 0.42 (135) (210) (390) (390) (360) (280) (280) (560) type photocathode is used.1 T and axial fields up to ∼ 1 T. but also degrades the signal definition. A single photoelectron ejected from the photocathode is accelerated through a potential difference of ∼20 kV before it impinges on the silicon sensor/anode. Large-size HPD’s with sophisticated focusing may be suitable for future water Cherenkov experiments. where e is the electronic charge. HPD’s can have the same Q C and window geometries as PMT’s and can be segmented down to ∼50 µm.3: Properties of photocathode and window materials commonly used in vacuum photodetectors [12]. 28. and w ≈ 3.27 0. one might expect to have the excellent resolution of a simple Poisson statistic with large mean. The exception is proximity-focused devices (⇒ no (de)magnification) in an axial field. Hybrid APD’s (HAPD’s) add an avalanche multiplication step following the electron bombardment to boost the gain by a factor of ∼50.6 28. they require rather high biases and will not function in a magnetic field.8. offer good spatial resolution. but in fact it is even better. e.2. 28. With time resolutions of ∼10 ps and superior rate capabiility. 2008 18:04 .27 0. This affords a higher gain and/or lower electrical bias.15 0. July 24. Low-noise electronics must be used to read out HPD’s if one intends to take advantage of the low fluctuations in gain.7 eV is the mean energy required to create an e-h pair in Si at room temperature.23 0. However. Table 28.1.18 0. Hybrid photon detectors: Hybrid photon detectors (HPD) combine the sensitivity of a vacuum PMT with the excellent spatial and energy resolutions of a Si sensor [13]. Current applications of HPD’s include the CMS hadronic calorimeter and the RICH detector in LHCb.

Some hybrid devices attempt to combine the best features of different technologies while applications of nanotechnology are opening up exciting new possibilities. 2008 18:04 . It is also important to maintain high purity of the gas as minute traces of O2 can significantly degrade the detection efficiency. which would limit their sensitivity to wavelengths far too short. Most gases have photoionization work functions in excess of 10 eV. the PD is a reverse-biased p-n junction. producing a large number of secondary impact-ionization electrons. as in the PD’s used for crystal scintillator readout in CLEO.and gaseous photodetectors. depending on the temperature) can create e-h pairs (the photoconductive effect). In its simplest form. are suited for XUV photon detection [14].and X-rays. They can be made into flat panels to cover large areas (O(1 m2 )). Many of the ring imaging Cherenkov (RICH) detectors to date have used GPM’s for the detection of Cherenkov light [15].2. solid-state photodetectors are competing with vacuum. but falls toward the red because of the increasing absorption July 24. while matching or exceeding most performance criteria. Except for applications where coverage of very large areas or dynamic range is required. Silicon photodiodes (PD) are widely used in high-energy physics as particle detectors and in a great number of applications (including solar cells!) as light detectors. intrinsic silicon is doped to create a p-i-n structure. and can operate at low electric potentials. photoionization occurs on suitable molecules vaporized and mixed in the drift volume. 28. One type uses solid photocathode materials much in the same way as PMT’s. rugged. Solid-state photon detectors : In a phase of rapid development.BaBar.2.3 eV for TMAE and 7. in the case of these particular detectors. However. Particle detectors 7 28. These are discussed in Sec. Increasing the depletion depth decreases the capacitance (and hence electronic noise) and extends the red response. Since it is resistant to gas mixtures typically used in tracking chambers. In the other type.or gas-based devices for many existing applications and making way for a multitude of new ones.28. Since devices like GEM’s offer sub-mm spatial resolution. Quantum efficiency can exceed 90%.5 eV for TEA). CsI is a common choice.3. which are collected on the p and n sides. In principle the charge multiplication and collection processes are identical to those employed in gaseous tracking detectors such as multiwire proportional chambers. and are relatively inexpensive. The structure is discussed in some detail in Sec. to full depletion at a depth of about 100 µm. 28. 28. They are particularly well suited for detection of γ. L3. which have smaller work functions (5. can operate in high magnetic fields. GPM’s are often used as position-sensitive photon detectors. vapors of TMAE (tetrakis dimethyl-amine ethylene) or TEA (tri-ethyl-amine).2. and GLAST.3. or gas electron multipliers (GEM). lightweight. are easy to integrate into large systems. Belle. micromesh gaseous detectors (Micromegas). Compared to traditional vacuum. Often. Gaseous photon detectors : In gaseous photomultipliers (GPM) a photoelectron in a suitable gas mixture initiates an avalanche in a high-field region. respectively. Photons with energies above the indirect bandgap energy (wavelengths shorter than about 1050 nm. Special care must be taken to suppress the photon-feedback process in GPM’s. They also allow fine pixelization. solid-state detectors are proving to be the better choice.7. and often cheaper. The devices can be divided into two types depending on the photocathode material. The reverse bias increases the thickness of the depleted region. tolerant to magnetic fields. solid-state devices are more compact.8.

The absorption length reaches 100 µm at 985 nm. positron emission tomography (PET). and substantial cooling is often necessary. and a guard ring is generally used as a protection against edge breakdown. resulting in coverage of larger July 24. but are challenged at longer wavelengths. Welldesigned APD’s. Very large arrays containing O(107 ) of O(10 µm2 )-sized photodioides pixelizing a plane are widely used to photograph all sorts of things from everyday subjects at visible wavelengths to crystal structures with X-rays and astronomical objects from infrared to UV. where pixel-to-pixel signal transfer takes place over thousands of synchronous cycles with sequential output through shift registers [16]. One of the most promising recent developments in the field is that of devices consisting of large arrays (O(103 )) of tiny APD’s packed over a small area (O(1 mm2 )) and operated in a limited Geiger mode [18]. The gain is typically 10–200 in linear and up to 108 in Geiger mode of operation. e. “PPD” (for “Pixelized Photon Detector”) is most widely accepted (formerly “SiPM”). rugged. However. Intense R&D is expeced to lower the noise level and improve radiation hardness. detectable electrical response can be obtained from low-intensity optical signals down to single photons. high spatial resolution is achieved at the expense of speed and timing precision. these are made into charge-coupled devices (CCD).g. To limit the number of readout channels. is rather large: O(1 MHz/mm2 ) at room temperature.8 28. Position-sensitive APD’s use time information at multiple anodes to calculate the hit position. and economical devices allow auto-calibration through decent separation of photoelectron peaks and offer gains of O(106 ) at a moderate bias voltage (∼50 V). fiber-guided scintillation light. Although each cell only offers a binary output. Among different names used for this class of photodetectors. Active Pixel Sensor (APS) arrays with a preamplifier on each pixel and CMOS processing afford higher speeds. In avalanche photodiodes (APD). The sensitive wavelength window and gain depend on the semiconductor used. Particle detectors length of light in silicon. since G = 1. However. have achieved Q C ≈ 0. As a result. being the logical “or” of O(103 ) Geiger APD’s. and “thick” CCD’s have useful QE up to λ = 1 µm.g. but. noise limits the minimum detectable signal in room-temperature devices to several hundred photons. Excellent junction uniformity is critical. PPD’s are particularly well-suited for applications where triggered pulses of several photons are expected over a small area. Thus. Typical QE’s exceed 90% over much of the visible spectrum.7 with sub-ns response time. PPD’s are being adopted as the preferred solution for various purposes including medical imaging. the single-photoelectron noise of a PPD. such as those used in CMS’ crystal-based electromagnetic calorimeter. an exponential cascade of impact ionizations initiated by the initial photogenerated e-h pair under a large reverse-bias voltage leads to an avalanche breakdown [17]. Stability and close monitoring of the operating temperature are important for linear-mode operation. amplification is necessary. Custom-made CCD’s have virtually replaced photographic plates and other imagers for astronomy and in spacecraft. These compact. linearity with respect to the number of photons is achieved by summing the cell outputs in the same way as with a MCP in saturation mode (see above). Optimal low-noise amplifiers are slow. 2008 18:04 . e. Much R&D is underway to overcome the limitations of both CCD and CMOS imagers. even so.

A one-cm-thick scintillator traversed by a minimumionizing particle will therefore yield ≈ 2 × 104 photons. including high tolerance to radiation and temperature fluctuations. rise times are much faster. measurement of synchrotron radiation. Very dense ionization columns emit less light than expected on the basis of dE/dx for minimumionizing particles. much R&D has been directed to p-i-n diode arrays based on thin polycrystalline diamond films formed by chemical vapor deposition (CVD) on a hot substrate (∼1000 K) from a hydrocarbon-containing gas mixture under low pressure (∼100 mbar).3. Visible-light photon counters (VLPC) utilize the formation of an impurity band only 50 meV below the conduction band in As-doped Si to generate strong (G ≈ 5 × 104 ) yet sharp response to single photons with Q ≈ 0. (28.2) dx 1 + kB dE/dx where L is the luminescence.to moderate-UV region [19]. blindness to most of the solar radiation spectrum.2) of this Review) to generate optical photons. liquid. L0 is the luminescence at low specific ionization density. Of late.03 to 1. The smallness of the band gap considerably reduces the gain dispersion. Birks’ formula is dE/dx dL = L0 . 28. The combination of high light yield and fast response time allows the possibility of sub-ns timing resolution [24]. usually in the blue to green wavelength regions [21].F. Crystal organic scintillators are practically unused in high-energy physics. which must be determined for each scintillator by measurement. Typical photon yields are about 1 photon per 100 eV of energy deposit [22]. A widely used semi-empirical model by Birks posits that recombination and quenching effects between the excited molecules reduce the light yield [23]. July 24. These devices have maximum sensitivity in the extreme. and plastic. Plastic scintillators are by far the most widely used.20 g cm−3 . Organic scintillators are broadly classed into three types. Plastic scintillators do not respond linearly to the ionization density. The resulting photoelectron signal will depend on the collection and transport efficiency of the optical package and the quantum efficiency of the photodetector. These effects are more pronounced the greater the density of the excited molecules. The Run 2 DØ detector uses 86000 VLPC’s to read the optical signal from its scintillating-fiber tracker and scintillator-strip preshower detectors. Organic scintillators Revised September 2007 by K. Densities range from 1. Johnson (FSU). low dark noise. and relatively low cost make them ideal for space-based UV/XUV astronomy.9 [20]. Particle detectors 9 areas and wider applications. The dark noise increases sharply and exponentially with both temperature and bias. Only a very small bias (∼7 V) is needed. crystalline. Many desirable characteristics. 27.28. Decay times are in the ns range. but high sensitivity to infrared photons requires cooling below 10 K. and luminosity monitoring at (future) lepton collider(s). Attempts are being made to combine the fabrication of the sensors and the front-end electronics (ASIC) in the same process with the goal of making PPD’s and other finely pixelized solid-state photodetectors extremely easy to use. and kB is Birks’ constant. all of which utilize the ionization produced by charged particles (see Sec. 2008 18:04 .

that is. This “self-absorption” is undesirable for detector applications because it causes a shortened attenuation length. Liquids which scintillate include toluene and xylene. 2008 18:04 .1: Cartoon of scintillation “ladder” depicting the operating mechanism of plastic scintillator. The wavelength difference between the major absorption and emission peaks is called the Stokes’ shift. This allows pulse shape discrimination as a technique to carry out particle identification. the absorption and emission are spread out over a wide band of photon energies. the initial excitation takes place via the absorption of a photon. Occurring in complex molecules. and de-excitation by emission of a longer wavelength photon.10 28. Recently. the smaller the self absorption—thus.05% wt/wt ) 1m γ photodetector Figure 28. is especially marked in those organic substances which contain aromatic rings. Fluors are used as “waveshifters” to shift scintillation light to a more convenient wavelength. a large Stokes’ shift is a desirable property for a fluor. however. Approximate fluor concentrations and energy transfer distances for the separate sub-processes are shown. Particle detectors The fraction of light emitted during the decay “tail” can depend on the exciting particle. there is some fraction of the emitted light which can be re-absorbed [26]. ~340 nm γ absorb UV photon emit blue. ~400 nm absorb blue photon secondary fluor (~0. Scintillators: The plastic scintillators used in high-energy physics are binary or ternary July 24. Ease of fabrication into desired shapes and low cost has made plastic scintillators a common detector component. Scintillation mechanism : Scintillation: A charged particle traversing matter leaves behind it a wake of excited molecules.1. Ionization excitation of base plastic 10−8 m Forster energy transfer base plastic primary fluor (~1% wt/wt ) 10−4 m emit UV. will release a small fraction (≈ 3%) of this energy as optical photons. plastic scintillators in the form of scintillating fibers have found widespread use in tracking and calorimetry [25]. scintillation. and have some overlap. Fluorescence: In fluorescence. It is usually the case that the greater the Stokes’ shift. such as polystyrene (PS) and polyvinyltoluene (PVT).3. Certain types of molecules. This process. 28. Because of the hydrogen content (carbon to hydrogen ratio ≈ 1) plastic scintillator is sensitive to proton recoils from neutrons.

A typical wavelength shifter uses an acrylic base because of its good optical qualities. Particle detectors 11 solutions of selected fluors in a plastic base containing aromatic rings. but a resonant dipole-dipole interaction. Attenuation length: The Stokes’ shift is not the only factor determining attenuation length. Aging and Handling: Plastic scintillators are subject to aging which diminishes the light yield. The primary fluor has a second important function. a single fluor to shift the light emerging from the plastic scintillator to the blue-green. However. the quality of the surface. so it is necessary to add yet another waveshifter (the “secondary” fluor). solvents. at fractional percent levels. The addition of the primary fluor in high concentration can shorten the decay time by an order of magnitude and increase the total light yield. Caveats and cautions : Plastic scintillators are reliable. Longer attenuation lengths are obtained by dissolving a “primary” fluor in high concentration (1% by weight) into the base. which strongly couples the base and fluor [28]. high temperatures. or fingerprints have contacted the surface.) Virtually all plastic scintillators contain as a base either PVT or PS. 28. for example. such as stabilizers.3. first described by Foerster. mechanical flexing. 2008 18:04 . PVT-based scintillator can be up to 50% brighter. the optical clarity and uniformity of the bulk material. The decay time of the scintillator base material can be quite long—in pure polystyrene it is 16 ns. they possess quirks to which the experimenter must be alert. 27 for a comprehensive list of components. or rough handling will aggravate the process. the greater will be its self-absorption). A particularly fragile region is the surface which can “craze”—develop microcracks—which rapidly destroy the capability of plastic scintillators to transmit light by total internal reflection. Others are the concentration of fluors (the higher the concentration of a fluor. At the concentrations used (1% and greater).1). 28. Ionization in the plastic base produces UV photons with short attenuation length (several mm). The wavelength shifter must be insensitive to ionizing radiation and Cherenkov light.2. and occasionally a third (not shown in Fig. Crazing is particularly likely where oils. a fluor which fulfills other requirements is usually not completely adequate with respect to emission wavelength or attenuation length. The strong coupling sharply increases the speed and the light yield of the plastic scintillators. Unfortunately. which is selected to efficiently re-radiate absorbed energy at wavelengths where the base is more transparent. External wavelength shifters: Light emitted from a plastic scintillator may be absorbed in a (nonscintillating) base doped with a wave-shifting fluor. the average distance between a fluor molecule and an excited base unit is around 100 ˚. and convenient. Such wavelength shifters are widely used to aid light collection in complex geometries.28. Exposure to solvent vapors. irradiation. which may be present. and absorption by additives. (See the appendix in Ref. robust. Afterglow: Plastic scintillators have a long-lived luminescence which does not follow a July 24. and contains ultra-violet absorbing additives to deaden response to Cherenkov light. At these distances the predominant A mode of energy transfer from base to fluor is not the radiation of a photon. much less than a wavelength of light.

Magnetic field: The light yield of plastic scintillators may be changed by a magnetic field. polymer chain length. so that the signal transmission via total internal reflection has a low loss. The fiber is drawn from a boule and great care is taken during production to ensure that the intersurface between the core and the cladding has the highest possible uniformity and quality. as well as the materials properties of the base such as glass transition temperature.3. and increasing the light yield has proven difficult. and temperature. etc.3.42. Data are sketchy and mechanisms are not understood.29]. This poorly understood effect appears as a reduction both of light yield and attenuation length. Particle detectors simple exponential decay.59). A minimum-ionizing July 24. Scintillating and wavelength-shifting fibers : The clad optical fiber is an incarnation of scintillator and wavelength shifter (WLS) which is particularly useful [33]. radiation hard. SCIFI calorimeters are fast. The number of photons from the fiber available at the photodetector is always smaller than desired.45 T have been reported [31]. Radiation damage: Irradiation of plastic scintillators creates color centers which absorb light more strongly in the UV and blue than at longer wavelengths. atmosphere. The fraction of generated light which is transported down the optical pipe is denoted the capture fraction and is about 6% for the single-clad fiber and 10% for the double-clad fiber. The sensitive region of scintillating fibers can be controlled by splicing them onto clear (non-scintillating/non-WLS) fibers. It is not excluded that other gases may have similar quenching effects. The effect is very nonlinear and apparently not all types of plastic scintillators are so affected. Intensities at the 10−4 level of the initial fluorescence can persist for hundreds of ns [21. Annealing also occurs. for added light capture.. 2008 18:04 . SCIFI trackers can handle high rates and are radiation tolerant. with another cladding of fluorinated PMMA with n = 1. e. utilize fluors with large Stokes’ shifts (aka the “Better red than dead” strategy). Atmospheric quenching: Plastic scintillators will decrease their light yield with increasing partial pressure of oxygen. and not well understood [32]. but on the dose rate. dense. SCIFI techniques have become mainstream [35].g. Increases of ≈ 3% at 0. This can be a 10% effect in an artificial atmosphere [30].12 28. The phenomena are complex. 28. Since color centers are less intrusive at longer wavelengths. or. for an overall diameter of 0. WLS scintillator readout of a calorimeter allows a very high level of hermeticity since the solid angle blocked by the fiber on its way to the photodetector is very small. during and after irradiation.49) a few microns thick. unpredictable. Radiation damage depends not only on the integrated dose. A typical configuration would be fibers with a core of polystyrene-based scintillator or WLS (index of refraction n = 1. but the low photon yield at the end of a long fiber (see below) forces the use of sensitive photodetectors. before. accelerated by the diffusion of atmospheric oxygen and elevated temperatures. Since the initial demonstration of the scintillating fiber (SCIFI) calorimeter [34]. the most reliable method of mitigating radiation damage is to shift emissions at every step to the longest practical wavelengths. surrounded by a cladding of PMMA (n = 1. and can have leadglass-like resolution.5 to 1 mm.

4. they can be used in applications where high stopping power or a high conversion efficiency for electrons or photons is required. However. but not all. re-emitted isotropically and lost with 90% probability—and to the lower absorption of longer wavelengths by polystyrene. it is not constant with distance from the excitation source and it is wavelength dependent. Many of these crystals also have very high light output. A scintillating or WLS fiber is often characterized by its “attenuation length. So-called “cladding light” causes some of the distance dependence [36].L.” over which the signal is attenuated to 1/e of its original value. including the importance of re-absorption of emitted photons by the polymer base or dissolved fluors. either directly by charged particles. The efficiency η for the conversion of energy deposit in the crystal to scintillation light can be expressed by the relation [37] η =β·S ·Q . whose quantum efficiency is still rising at 600 nm. July 24. The wavelength dependence is usually related to the higher re-absorption of shorter wavelength photons—once absorbed. of which about 200 are captured. typically fluorescent ions such as thallium (Tl) or cerium (Ce) which is responsible for producing the scintillation light. 2008 18:04 (28. 28. as well as serving as gamma ray detectors over a wide range of energies. and the quality of the total internal reflection boundary. Zhu (California Institute of Technology) and C. Particle detectors 13 particle traversing a high-quality 1 mm diameter fiber perpendicular to its axis will produce fewer than 2000 photons. However.28. Inorganic scintillators: Revised September 2007 by R. should not be na¨ ıvely compared to measurements utilizing a silicon photodiode. other crystals require the addition of a dopant. which consist of a totally active absorber (as opposed to a sampling calorimeter). Many factors determine the attenuation length. and can therefore provide excellent energy resolution down to very low energies (∼ few hundred keV). Some crystals are intrinsic scintillators in which the luminescence is produced by a part of the crystal lattice itself.3) . These include total absorption electromagnetic calorimeters (see Sec. the level of crystallinity of the base polymer. This energy is transferred to the luminescent centers which then radiate scintillation photons. the scintillation mechanism is the same. Inorganic crystals form a class of scintillating materials with much higher densities than organic plastic scintillators (typically ∼ 4–8 g/cm3 ) with a variety of different properties for use as scintillation detectors.10. in both cases. Attenuation may eliminate 95% of these photons in a large collider tracker. Energy is deposited in the crystal by ionization. Attenuation lengths of several meters are obtained by high quality fibers. Woody (BNL).-Y. Due to their high density and high effective atomic number. or by the conversion of photons into electrons or positrons which subsequently produce ionization. it should be understood that the attenuation length is not necessarily a measure of fiber quality. Experimenters should be aware that measurements of attenuation length by a phototube with a bialkali photocathode. whose quantum efficiency drops below 10% at 480 nm.1). However. Among other things. 28.

it needs careful surface treatment and is slightly hygroscopic. with an emission that is well matched to a bialkali photomultiplier tube. the value of η is ∼ 0.000 photons per MeV of energy deposit.4. The decay time of the scintillator is mainly dominated by the decay time of the luminescent center. Particle detectors where β is the efficiency of the energy conversion process. Only properties of LSO:Ce and GSO:Ce are listed in Table 28. and it is easy to handle and not hygroscopic. which results in a light output ∼ 40. In addition. the scintillation decay time is primarily determined by the energy transfer and emission process. but faster emission at shorter wavelengths and light output approximately an order of magnitude lower. the useful signal produced by a scintillator is July 24. BaF2 has a fast component with a sub-nanosecond decay time. 2008 18:04 . but it is highly hygroscopic and difficult to work with. S is the efficiency of energy transfer to the luminescent center. However. but the decay time is rather slow (τ ∼ 250 ns). but its intrinsic light yield is rather low. These have been shown to be linear over a large energy range [42]. and is mechanically robust (high plasticity and resistance to cracking).5 X0 cube with a Tyvek paper wrapping and a full end face coupled to a photodetector [43]. The quantum efficiencies of the photodetector is taken out to facilitate a direct comparison of crystal’s light output. and is the fastest known scintillator. or GSO:Ce) [40] are dense crystal scintillators which have a high light yield and a fast decay time. Cerium doped lutetium oxyorthosilicate (Lu2 SiO5 :Ce. and Q is the quantum efficiency of the luminescent center.4 lists the basic properties of some commonly used inorganic crystal scintillators. which include YAP (YAlO3 :Ce) and LuAP (LuAlO3 :Ce) and their mixed compositions. Compared with CsI(Tl). Of particular interest is the family of yttrium and lutetium perovskites. and has a rather low density. with a very short X0 and RM . it also has a slow component with a much longer decay time (∼ 630 ns). and is the main factor in determining the intrinsic light output of the scintillator. cerium doped lutetium-yttrium oxyorthosilicate (Lu2(1−x) Y2x SiO5 . LYSO:Ce) [39] and cerium doped gadolinium orthosilicate (Gd2 SiO5 :Ce. NaI(Tl) is one of the most common and widely used scintillators. Beside the crystals listed in Table 28. an emission that is well matched to solid state photodiodes.1 and ∼ 1 depending on the crystal. in the case of thallium doped sodium iodide (NaI(Tl)).14 28. other fluoride crystals such as CeF3 have been shown to provide excellent energy resolution in calorimeter applications. Table 28.4 gives the light output of other crystals relative to NaI(Tl) and their dependence to the temperature variations measured for crystal samples of 1. In addition.4 since the properties of LYSO:Ce are similar to that of LSO:Ce except a little lower density than LSO:Ce depending on the yttrium fraction in LYSO:Ce [41]. and have the potential for providing extremely good intrinsic energy resolution. BGO’s emission is well-matched to the spectral sensitivity of e photodiodes. a number of new crystals are being developed that may have potential applications in high energy or nuclear physics. pure CsI has identical mechanical properties.5. Bismuth gemanate (Bi4 Ge3 O12 or BGO) has a high density. This high light output is largely due to the high quantum efficiency of the thallium ion (Q ∼ 1). Table 28. and consequently a short radiation length X0 and Moli`re radius RM . However. For example. CsI(Tl) has high light yield. The value of η ranges between 0. Lead tungstate (PbWO4 or PWO) has a very high density. or LSO:Ce) [38]. However.

Results with different photodetectors can be significantly different. For scintillators which emit in the UV. Additional radiation damage effects may include a reduced intrinsic scintillation light yield (damage to the luminescent centers) and an increased phosphorescence (afterglow). The internal light transmission depends on the intrinsic properties of the material. the density and type of the scattering centers and defects that can produce internal absorption within the crystal. as shown in Table 28. and the size and shape of the crystal. e. the response of CsI(Tl) relative to NaI(Tl) with a standard photomultiplier tube with a bialkali photocathode. and can be highly affected by factors such as radiation damage.g. is required for high resolution and precision calibration [44].4. the bulk attenuation length of the material). The quantum efficiency depends on the type of photodetector used to detect the scintillation light. or the ability to track and monitor the variation of its light output in a radiation environment. For crystals to be used in the construction a high precision calorimeter in a radiation environment. 28. leading to a degradation of the energy resolution. The quantum efficiency of the detector is usually highly wavelength dependent and should be matched to the particular crystal of interest to give the highest quantum yield at the wavelength corresponding to the peak of the scintillation emission. One important issue related to the application of a crystal scintillator is its radiation hardness.. its scintillation mechanism must not be damaged and its light attenuation length in the expected radiation environment must be July 24. LSO:Ce/LYSO:Ce and CsI(Tl). Particle detectors 15 usually quoted in terms of the number of photoelectrons per MeV produced by a given photodetector. A common damage phenomenon is the appearance of radiation induced absorption caused by the formation of color centers originated from the impurities or point defects in the crystal. Fig. would be 45 rather than 165 because of the photomultiplier’s low quantum efficiency at longer wavelengths. Also shown in the figure are emission spectra of three crystal scintillators. and the numerical values of the emission weighted quantum efficiency. where the quantum efficiencies of the photodetector has been taken out. reflections and scattering from the surfaces.2 shows the quantum efficiencies of two photodetectors. Hamamatsu R2059.e. Stability of its light output. a Hamamatsu R2059 PMT with bi-alkali cathode and quartz window and a Hamamatsu S8664 avalanche photodiode (APD) as a function of wavelength. a severe reduction of light attenuation length may cause a distortion of the light response uniformity. For example. The area under each emission spectrum is proportional to crystal’s light yield. This radiation induced absorption reduces the light attenuation length in the crystal. as discussed below.e. /MeV = L · QE · Nγ /MeV (28. 2008 18:04 . All known crystal scintillators suffer from radiation damage.g. a detector with a quartz window should be used. which is typically ∼15–20% for photomultiplier tubes and ∼70% for silicon photodiodes for visible wavelengths. BGO. and hence its light output. For crystals with high defect density.28. These factors can vary considerably depending on the sample. The relationship between the number of photons/MeV produced and photoelectrons/MeV detected involves the factors for the light collection efficiency L and the quantum efficiency QE of the photodetector: Np. e.4) L includes the transmission of scintillation light within the crystal (i. but can be in the range of ∼10–60%.

Particle detectors long enough so that its light response uniformity. Recently.89 1280 2. 6 of this Review.57 3.23 3.4 have been used in high energy or nuclear physics experiments when the ultimate energy resolution for electrons and photons is desired. PWO calorimeters are widely used by CMS and ALICE at LHC.4: Properties of several inorganic crystal scintillators.13 1050 1.3 1123 0.40 2050 1.07 2.3 −1.67 MP 3 ◦ ∗ X0 ∗ RM dE/dx λ∗ I τdecay λmax ns 230 300 630s 0. f = fast component. does not change [45]. the CLEO CsI(Tl) calorimeter at CESR.9f 1300 35 s n C cm cm MeV/cm cm 4.71 1950 1.85 yes no no slight slight no no no −0.10 3.9 22.2 nm 410 480 300s 220f 560 420 s Relative Hygro.85 2. † Relative light output measured for samples of 1.8 9.9 42.14 GSO(Ce) 6.89 LSO(Ce) 7.7 20.38 ∗ 6 30s 10f 40 600 56 s f f 310 425s 420f 420 430 f 1.3 20.2 −0.7 39.2 −0.1 0.6 8.03 4. Table 28. the L3 BGO calorimeter at LEP.8 30. The quantum efficiencies of the photodetector is taken out. 2008 18:04 .6 10.1 7. Because of its high density and low cost.2 9.4f 165 3.12 4. the KTeV CsI calorimeter at the Tevatron.15 1.3 −2.3s ∼0f 0.083s 0.9 22.20 1. the BaBar and BELLE CsI(Tl) calorimeters at PEP-II and KEK. and thus its energy resolution.79 1.6 5. and are the leading option for PANDA at GSI. Most of the notation is defined in Sec.82 1.86 CsI(pure) 4.51 621 1. Most of the crystals listed in Table 28.6 5. Refractive index at the wavelength of the emission maximum.13 2.6 s 651 2. by CLAS and PrimEx at CEBAF.29f 83 3 s f f 30 Numerical values calculated using formulae in this review. Parameter: ρ Units: NaI(Tl) BGO BaF2 CsI(Tl) g/cm 3.d(LY)/dT output† scopic? %/◦ C‡ 100 21 36s 3.95 2.59 1.9 −1. investigations have been made aiming at using LSO:Ce or LYSO:Ce crystals for future high energy or nuclear physics experiments [41].50 1.51 PbWO4 8.16 28.3 39. Examples are the Crystal Ball NaI(Tl) calorimeter at SPEAR.86 621 1. s = slow component July 24.0 6.7 −0.5 X cube with a Tyvek paper 0 wrapping and a full end face coupled to a photodetector.57 2. ‡ Variation of light yield with temperature evaluated at the room temperature.00 2.23 4.

Particle detectors 17 7 Figure 28. the existence of a velocity threshold for radiation. 28. As Cherenkov radiation is a weak source of photons.2: The quantum efficiencies of two photodetectors. in practice. 2008 18:04 . The area under each emission spectrum is proportional to crystal’s light yield. and (2) a photodetector. (2) hadronic particle identification. 27 of this Review): the prompt emission of a light pulse. (2) the hadronic PID detectors at the B factory detectors (DIRC in BaBar [9] and the aerogel threshold Cherenkov in Belle [47]) . including (1) fast particle counters. they are widely used over a much broader range of applications. a Hamamatsu R2059 PMT with bi-alkali cathode and a Hamamatsu S8664 avalanche photodiode (APD). (1) a radiator through which the charged particle passes. LSO and CsI(Tl). Cherenkov detectors Revised September 2007 by B.N. and (3) large water Cherenkov counters such as Super-Kamiokande [49].28. (1) the polarization detector of the SLD [46]. and July 24. Cherenkov counters contain two main elements. and (3) tracking detectors performing complete event reconstruction. light collection and detection must be as efficient as possible. are shown as a function of wavelength. A few examples of specific applications from each category include. and the numerical values of the emission weighted quantum efficiency. BGO. Also shown in the figure are emission spectra of three crystal scintillators. The presence of the refractive index n and the path length of the particle in the radiator in the Cherenkov relations allows tuning these quantities for a particular experimental application. Cherenkov detectors utilize one or more of the properties of Cherenkov radiation discussed in the Passages of Particles through Matter section (Sec.5. Ratcliff (SLAC). Although devices using Cherenkov radiation are often thought of as particle identification (PID) detectors.

α2 z 2 re me c2 dE .6) July 24. Recently. Cherenkov counters may be classified as either imaging or threshold types. Long radiator lengths are required to obtain sufficient numbers of photons when the momenta of the particle species to be separated are high. Imaging counters may be used to track particles as well as identify them. are sometimes included on the right hand side of the equation. so that = coll det . However.e. and collecting most of the Cherenkov light. chromatic dispersion. absorption length.5) where L is the path length in the radiator. A typical value of N0 for a photomultiplier (PMT) detection system working in the visible and near UV. availability. (28. have values ranging between about 30 and 180 cm−1 . Particle detectors the dependence of the Cherenkov cone half-angle θc and the number of emitted photons on the velocity of the particle. nor. the choice requires consideration of factors such as material density.18 28. the usual case in high-energy physics.e. radiation length.7) We take z = 1. (28. depending on whether they do or do not make use of Cherenkov angle (θc ) information. a quantity called the Cherenkov detector quality factor N0 can be defined as N0 = so that Np. the gap in refractive index that has traditionally existed between gases and liquid or solid materials has been partially closed with transparent silica aerogels with indices that range between about 1. The number of photoelectrons (Np. indeed. 27 of this Review and Table 6. (E) is the efficiency for collecting the Cherenkov light and transducing it in photoelectrons. ≈ LN0 sin2 θc . and cost. in the following discussion. The quantities and θc are functions of the photon energy E. α2 z 2 =L re me c2 (E) sin2 θc (E)dE . ) detected in a given device is Np. is about 100 cm−1 . transmission bandwidth. and α2 /(re me c2 ) = 370 cm−1 eV−1 . Practical counters. This definition of the quality factor N0 is not universal. In addition to refractive index. 2008 18:04 . utilizing a variety of different photodetectors.1).007 and 1.e. (28. optical workability (for solids). very useful for situations where the geometrical photon collection efficiency ( coll ) varies substantially for different tracks.13. separate factors for photon collection and detection ( det ). Radiators can be chosen from a variety of transparent materials (Sec. In this case. since the typical energy dependent variation of the index of refraction is modest.

species b with the same momentum has velocity βb . perhaps.5. N0 = 90 cm−1 ) . so cos θc = βa /βb .27. the photons must be “imaged” onto a detector so that their angles can be derived [52]. can attain good particle separation performance over a substantial momentum range for essentially the full solid angle of the spectrometer. This strategy can increase the momentum range of particle separation by a modest amount (to a momentum some 20% above the threshold momentum of the heavier particle in a typical case). can be small. For a photomultiplier with a typical bialkali cathode. For limited path lengths Np. The overall efficiency of the device is controlled by Poisson fluctuations. make a yes/no decision based on whether the particle is above or below the Cherenkov threshold velocity βt = 1/n. Np. It is common to design for at least 10 photoelectrons for the high velocity particle in order to obtain a robust counter. Physics sources of light production for the below threshold particle. modern multi-channel counters. As rejection of the particle that is below threshold depends on not seeing a signal. 28.2.28. A second. Though this imaging process is directly analogous to the familiar imaging techniques used in telescopes and other optical instruments. there is a somewhat bewildering variety of methods used in a wide variety of counter types with different names. and a minimum number is required to trigger external electronics.8) cm−1 for π’s and (by design) 0 for K’s. A straightforward enhancement of such detectors uses the number of observed photoelectrons (or a calibrated pulse height) to discriminate between species or to set probabilities for each particle species [51].e. in their simplest form. Since low-energy photon detectors can measure only the position (and. that is. such as decay of the above threshold particle or the production of delta rays in the radiator. which can be especially critical for separation of species where one particle type is dominant.. for example.e. The effective number of photoelectrons is often less than the average number calculated above due to additional equivalent noise from the photodetector. (28. 90%. lighter. and Np. so that Np. electronic and other background noise can be important. often limit the separation attainable.e.5. Threshold counters : Threshold Cherenkov detectors [50].8) Suppose. such as the ACC at Belle [47]. Careful designs give coll det dE ≈ 0. at this momentum species a has velocity βa = 1/n. a precise detection time) of the individual Cherenkov photons (not the angles directly). and need to be carefully considered. /L ≈ 90 cm−1 sin2 θc (i. /L ≈ 90 cm−1 m2 − m2 a b .e. 2008 18:04 .e. that n is chosen so that the threshold for species a is pt . In most cases the optics map the Cherenkov cone onto (a portion of) a distorted circle at the photodetector. Well designed. /L ≈ 16(0. Particle detectors 19 28. 2 + m2 pt a (28.9) For K/π separation at p = pt = 1(5) GeV/c.1. Imaging counters : The most powerful use of the information available from the Cherenkov process comes from measuring the ring-correlated angles of emission of the individual Cherenkov photons. Some of the imaging methods used include (1) July 24.

for σ(θc ) = 2mrad.10) β where σ(θc ) = σ(θi ) ⊕C . C combines a number of other contributions to resolution including. (28. For a β ≈ 1 particle of momentum (p) well above threshold entering a radiator with index of refraction (n). (3) background hits from random sources.2(18) GeV/c at better than 3σ.11) where σ(θi ) is the average single photoelectron resolution.1 and 5 mrad depending on the size. the angular resolution term σ(θc ) varies between about 0. and multiple scattering. July 24. (28. 28. would separate π/K’s from the threshold region starting around 0.15(3) GeV/c through the imaging region up to about 4. the fractional error on the particle velocity (δβ ) is given by σβ δβ = = tan θc σ(θc ) . The range of momenta over which a particular counter can separate particle species extends from the point at which the number of photons emitted becomes sufficient for the counter to operate efficiently as a threshold device (∼20% above the threshold for the lighter species) to the value in the imaging region given by the equation above.474).. or a flourocarbon gas radiator (C5 F12 . In many practical cases. Many different imaging counters have been built during the last several decades [53]. radiator.e. a velocity resolution of σβ /β ≈ 10−4 –10−5 can be obtained [50]. 2008 18:04 .e. alignment. and photodetector type of the particular counter.20 28. In addition. a fused silica radiator(n = 1. (1) correlated terms such as tracking. (2) proximity focusing (i. (2) hit ambiguities. focusing by limiting the emission region of the radiation). detector resolution and the intrinsic chromaticity spread of the radiator index of refraction averaged over the photon detection bandwidth. Particle detectors focusing by a lens. Finally. In a simple model of an imaging PID counter. For example. the prompt Cherenkov emission coupled with the speed of modern photon detectors allows the use of time imaging. (28. These devices use optical focusing and/or geometrical masking to select particles having velocities in a specified region. n = 1. Among the earliest examples of this class of counters are the very limited acceptance Differential Cherenkov detectors. designed for particle selection in high momentum beam lines. With careful design. the resolution is limited by these effects. full tracking (and event reconstruction) can be performed in large water counters by combining the individual space position and time of each photon together with the constraint that Cherenkov photons are emitted from each track at a constant polar angle (Sec. the number of σ separation (Nσ ) between particles of mass m1 and m2 is approximately |m2 − m2 | 1 √ 2 . Np.12) Nσ ≈ 2 σ(θ ) n2 − 1 2p c In practical counters. and (4) hits coming from other tracks.0017). a method which is used much less frequently in conventional imaging technology.6 of this Review). and (3) focusing through an aperture (a pinhole). as defined by the optics.

This fact. and (3) to keep the chemically active TMAE vapor from interacting with materials in the system.0017) radiators.g. fused silica. and utilize different photon detection bandwidths. n = 1. PID RICH counters are sometimes further classified by ‘generations’ that differ based on performance.05% TMAE (tetrakis(dimethylamino)ethylene). rather than the transmitted light. first as a Cherenkov radiator. (1) to avoid absorbing the UV photons to which TMAE is sensitive. ethane/methane) with ∼ 0. n = 1.7 × 3. the former being proximity imaged using the small radiator thickness while the latter use mirrors. They are made sensitive to photons by doping the TPC gas (usually. (2) to avoid absorbing the single photoelectrons as they drift in the long TPC. 2008 18:04 .28. Later generation counters [53] generally must operate at much higher particle rates than the first generation detectors. Other fast detection systems that use solid cesium iodide (CSI) photocathodes or triethylamine (TEA) doping in proportional chambers are useful with certain radiator types and geometries. and second. The BaBar DIRC uses 144 fused silica radiator bars (1. It “inverts” the usual principle for use of light from the radiator of a RICH by collecting and imaging the total internally reflected light. A DIRC (Detector of Internally Reflected Cherenkov light) is a third generation subtype of a RICH first used in the BaBar detector [48]. design. In particular. more forgiving photon detection technology than the TMAE doped TPC’s just described. Vacuum based photodetection systems (e.55] has both liquid (C6 F14 . coupled with the high reflection coefficients of the total internal reflection process (> 0. and may be sufficiently fast to allow time imaging or the use of time of flight information. simultaneously. or hybrid photodiodes (HPD)) have become increasingly common. Great attention to detail is required. The DIRC makes use of the fact that the magnitudes of angles are preserved during reflection from a flat surface. Radiator choices have broadened to include materials such as lithium flouride.276) and gas (C5 F12 . A typical example of a first generation RICH used at the Z factory e+ e− colliders [54. Particle detectors 21 Practical multi-track Ring-Imaging Cherenkov detectors (generically called RICH counters) are a more recent development. and photodetection techniques. A DIRC utilizes the optical material of the radiator in two ways. In spite of their unforgiving operational characteristics. as a light pipe for the Cherenkov light trapped in the radiator by total internal reflection.9995 for higly polished SiO2 ). with higher readout channel counts. multi channel plate PMTs (MCP-PMT). allows the photons of the ring image to be transported to a detector outside the path of the particle where they may be imaged in up to three dimensions (two in space and one in time). may be used with a wide choice of radiators. their π/K separation range extends over momenta from about 0.25 to 20 GeV/c. and faster. and aerogel.5 × 490 cm) with the light being focused onto 11 000 July 24. these counters attained good e/π/K/p separation over wide momentum ranges during several years of operation. They have been built in small-aperture and 4π geometries both as PID counters and as stand-alone detectors with complete tracking and event reconstruction as discussed more fully below.. and the long attenuation length for photons in high purity fused silica. The phototransducers are a TPC/wire-chamber combination having charge division or pads. single or multi anode photomultiplier tubes (PMT). They handle very high rates.

front-end electronics and calibration hardware. where the large mass of the transparent dielectric medium serves as an active target for neutrino interactions (or their secondary muons) and rare processes like nucleon decay. 2008 18:04 . polar ice.5 mrad. absorption and scattering of Cherenkov light are nonnegligible. The choice of medium is therefore constrained by the refractive index and transparency in the region of photodetector sensitivity. PMTs are deployed in vertical strings as modular units which include pressure housings. DIRC performance can be understood using the formula for (Nσ ) discussed above.5) for the photoelectron yield.6. with a spacing comparable to the attenuation (absorption and scattering) length of Cherenkov light in the medium (15–40 m for Antarctic ice and ∼45 m in the deep ocean). or July 24. (28. particle identification and calorimetric capabilities. power. a software-defined fiducial volume begins some distance (∼ 2 m) inside the photosensor surface. For volumes of this scale. Casper (UC Irvine). Volume instrumentation [56] is only cost-effective at low densities. and sometimes mounted with reflectors or wavelength-shifting plates to increase the effective photosensitive area. The effective photocathode coverage of such arrays is less than 1% but still adequate (using timing information and the Cherenkov angular constraint) to reconstruct the direction of TeV muons to 1◦ or better. matching the B decay momentum spectrum observed in BaBar. ring-imaging detectors with masses measured in kilotons or greater.22 28. sea-water. Cherenkov radiation is also exploited in large. Photo-multiplier tubes (PMTs) on either a volume or surface lattice measure the time of arrival and intensity of Cherenkov radiation. Typically. Hemispherical PMTs are favored for the widest angular acceptance. mineral oil. arrays up to (1 km)3 in size are under study or development. To improve hermeticity and shielding. The momentum range with good π/K separation extends up to about 4 GeV/c. highly-purified water is an inexpensive and effective choice. but complete experiments with triggering. Particle detectors conventional PMT’s located about 120 cm from the end of the bars by the “pinhole” of the bar end. and to ensure that an outward-going particle’s Cherenkov cone illuminates sufficient PMTs for reconstruction. The size of such “neutrino telescopes” is limited only by cost once the technical challenges of deployment. 28. Surface instrumentation [57] allows the target volume to be viewed with higher photocathode density by a number of PMTs which scales like (volume)2/3 . Events originating within the fiducial volume are classified as fully-contained if no particles exit the inner detector. Such devices are not subdetector components. Np. vertexing. is rather large (between 15 and 60) and the Cherenkov polar angle is measured to about 2. tracking. and a wavelength-dependent factor e−d/L(λ) (where d is the distance from emission to the sensor and L(λ) is the attenuation length of the medium) must be included in the integral of Eq. and D2 O are also used. Cherenkov tracking calorimeters Written August 2003 by D. signal extraction and calibration in an inaccessible and inhospitable environment are addressed. In addition to the specialized applications described in the previous section. Gains and calibration curves are measured with pulsed laser signals transmitted to each PMT individually via optical fiber or applied to the detector as a whole through one or more diffusing balls.e.

28. Particle detectors 23
partially-contained otherwise. An outer (veto) detector, optically separated from the inner volume and instrumented at reduced density, greatly assists in making this determination and also simplifies the selection of contained events. The maximum size of a pure surface array is limited by the attenuation length (∼ 100 m has been achieved for large volumes using reverse-osmosis water purification), pressure tolerance of the PMTs (< 80 meters of water, without pressure housings) and structural integrity of the enclosing cavity, if underground. In practice, these limitations can be overcome by a segmented design involving multiple modules of the nominal maximum size; megaton-scale devices are under study. Cherenkov detectors are excellent electromagnetic calorimeters, and the number of Cherenkov photons produced by an e/γ is nearly proportional to its kinetic energy. For massive particles, the number of photons produced is also related to the energy, but not linearly. For any type of particle, the visible energy Evis is defined as the energy of an electron which would produce the same number of Cherenkov photons. The number of photoelectrons collected depends on a detector-specific scale factor, with event-by-event corrections for geometry and attenuation. For typical PMTs, in water Np.e. ≈ 15 ξ Evis (MeV), where ξ is the effective fractional photosensor coverage; for other materials, the photoelectron yield scales with the ratio of sin2 θc over density. At solar neutrino energies, the visible energy resolution (∼ 30%/ ξ Evis (MeV)) is about 20% worse than photoelectron counting statistics would imply. For higher energies, multiphotoelectron hits are likely and the charge collected by each PMT (rather the number of PMTs firing) must be used; this degrades the energy resolution to approximately 2%/ ξ Evis (GeV). In addition, the absolute energy scale must be determined with sources of known energy. Using an electron LINAC and/or nuclear sources, 0.5–1.5% has been achieved at solar neutrino energies; for higher energies, cosmic-ray muons, Michel electrons and π 0 from neutrino interactions allow ∼ 3% absolute energy calibration. A trigger can be formed by the coincidence of PMTs within a window comparable to the detector’s light crossing time; the coincidence level thus corresponds to a visible energy threshold. Physics analysis is usually not limited by the hardware trigger, but rather the ability to reconstruct events. The interaction vertex can be estimated using timing and refined by applying the Cherenkov angle constraint to identified ring edges. Multi-ring events are more strongly constrained, and their vertex resolution is 33–50% better than single rings. Vertex resolution depends on the photosensor density and detector size, with smaller detectors performing somewhat better than large ones (∼ 25 cm is typical for existing devices). Angular resolution is limited by multiple scattering at solar neutrino energies (25–30◦ ) and improves to a few degrees around Evis = 1 GeV. A non-showering (µ, π ± , p) track produces a sharp ring with small contributions from delta rays and other radiated secondaries, while the more diffuse pattern of a showering (e, γ) particle is actually the superposition of many individual rings from charged shower products. Using maximum likelihood techniques and the Cherenkov angle constraint, these two topologies can be distinguished with an efficiency which depends on the photosensor density and detector size [58]. This particle identification capability has been confirmed by using cosmic-rays and Michel electrons, as well as charged-particle [59] and neutrino [60] beams. Large detectors perform somewhat better than smaller ones

July 24, 2008

18:04

24

28. Particle detectors

with identical photocathode coverage; a misidentification probability of ∼ 0.4%/ξ in the sub-GeV range is consistent with the performance of several experiments for 4% < ξ < 40%. Detection of a delayed coincidence from muon decay offers another, more indirect, means of particle identification; with suitable electronics, efficiency approaches 100% for µ+ decays but is limited by nuclear absorption (22% probability in water) for µ− . Reconstruction of multiple Cherenkov rings presents a challenging pattern recognition problem, which must be attacked by some combination of heuristics, maximum likelihood fitting, Hough transforms and/or neural networks. The problem itself is somewhat ill-defined since, as noted, even a single showering primary produces many closelyoverlapping rings. For π 0 → γγ two-ring identification, performance falls off rapidly with increasing π 0 momentum, and selection criteria must be optimized with respect to the analysis-dependent cost-function for e ↔ π 0 mis-identification. Two representative cases for ξ = 39% will be illustrated. In an atmospheric neutrino experiment, where π 0 are relatively rare compared to e± , one can isolate a > 90% pure 500 MeV/c π 0 sample with an efficiency of ∼ 40%. In a νe appearance experiment at Eν ≤ 1 GeV, where e± are rare compared to π 0 , a 99% pure 500 MeV/c electron sample can be identified with an efficiency of ∼ 70%. For constant ξ, a larger detector (with, perforce, a greater number of pixels to sample the light distribution) performs somewhat better at multi-ring separation than a smaller one. For a more detailed discussion of event reconstruction techniques, see Ref. 49. Table 28.5: Properties of Cherenkov tracking calorimeters. LSND was a hybrid scintillation/Cherenkov detector; the estimated ratio of isotropic to Cherenkov photoelectrons was about 5:1. MiniBooNE’s light yield also includes a small scintillation component. Detector Fiducial mass (kton) 3.3 H2 O 3.3 H2 O 0.88/0.78 H2 O 1.04 H2 O 0.084 oil+scint. 22.5 H2 O 22.5 H2 O 0.025 H2 O 1.0 D2 O 0.445 oil PMTs (diameter, cm) 2048 (12.5) 2048 (20 +plate) 1000/948 (50) 948 (50) 1220 (20) 11146 (50) 5182 (50) 680 (50) 9456 (20+cone) 1280 (20) ξ p.e./ MeV 0.25 1.1 3.4 3.4 33 6 3 6 9 3–4 Dates

IMB-1 IMB-3 KAM I KAM II LSND SK-1 SK-2 K2K SNO MiniBooNE

1% 4.5% 20% 20% 25% 39% 18% 39% 55% 10%

1982–85 1987–90 1983–85 1986–90 1993–98 1997–01 2002– 1999– 1999– 2002–

July 24, 2008

18:04

28. Particle detectors 25 28.7. Gaseous detectors
28.7.1. Energy loss and charge transport in gases : Written April 2008 by F. Sauli (CERN) and M. Titov (CEA Saclay). Gas-filled counters detect and localize the ionization produced by charged particles, generally after charge multiplication. The peculiar statistics of ionization processes, with asymmetries in the ionization trails, affect the coordinate determination deduced from the measurement of drift time or of the center of gravity of the collected charge. For thin gas layers, the width of the energy loss distribution can be larger than the average, requiring multiple sample or truncated mean analysis to achieve particle identification (see Sec. 28.7.4). The energy loss of charged particles and photons in various materials as a function of energy is discussed in Sec. 27. Table 28.6 provides values of relevant parameters in commonly used gases at NTP for unit charge, minimum-ionizing particles [61–67]. Numbers often differ depending on source, and values in the table should be taken only as approximate. For different conditions and mixtures, and neglecting internal energy transfer processes, one can use gas-density-dependent composition rules. Table 28.6: Properties of rare and molecular gases at normal temperature and pressure (NTP: 20◦ C, one atm). EX , EI : first excitation and ionization energy; WI : average energy per ion pair; dE/dx|min , NP , NT : differential energy loss, primary and total number of electron-ion pairs per cm, for unit charge, minimum ionizing particles. Gas Density, mg cm−3 0.839 1.66 5.495 0.667 1.26 2.49 1.84 3.78 Ex eV 16.7 11.6 8.4 8.8 8.2 6.5 7.0 10.0 EI eV 21.6 15.7 12.1 12.6 11.5 10.6 13.8 16.0 WI eV 30 25 22 30 26 26 34 54 dE/dx|min keV cm−1 1.45 2.53 6.87 1.61 2.91 5.67 3.35 6.38 NP cm−1 13 25 41 37 48 90 35 63 NT cm−1 50 106 312 54 112 220 100 120

Ne Ar Xe CH4 C2 H 6 iC4 H10 CO2 CF4

When an ionizing particle passes through the gas it creates electron-ion pairs, but often the ejected electrons have sufficient energy to further ionize the medium. On average, the total number of electron-ion pairs (NT ) is between two and three times larger than the number of primaries (NP ) (see Table 28.6). The probability of releasing an electron of energy E or larger follows an approximate 1/E 2 dependence (Rutherford law), shown in Fig. 28.3 for argon at NTP (dotted line, left scale). More detailed estimates taking into account the electronic structure of the
July 24, 2008 18:04

or cluster size. The practical range of 1 keV electrons in argon (dot-dashed line. The quenching of UV photons occurs through the photo-decomposition of polyatomic molecules. the inelastic cross section is zero below excitation and ionization thresholds. this contributes to the error in the coordinate determination. 2008 18:04 . The dot-dashed line provides. there is a 1% probability of creating of an electron of 1 keV or more in 10 mm of argon. As an example. on the right scale. for three values of the particle velocity factor βγ [62].26 28. substantially increasing the ionization losses. having large inelastic cross sections at moderate energies. for argon. and range of electrons in argon at NTP (dot-dashed curve. The value of drift velocity and diffusion depends very strongly on the nature of the gas. the practical range of electrons of energy E. right scale) [62]. Extensive collections of experimental July 24. electrons and ions drift in opposite directions and diffuse towards the electrodes. Fast gas mixtures are achieved by adding polyatomic gases (usually CH4 or CO2 ). namely on the detailed structure of the elastic and inelastic electron-molecule cross-sections. there is about 1% probability for primary clusters to contain ten or more electron-ion pairs [63]. Figure 28. has an exponentially decreasing probability. 64). and under the influence of an applied electric field. Another principal role of the polyatomic gas is to absorb the ultraviolet (UV) photons emitted by the excited inert gas atoms.5 eV) of the elastic scattering cross-section of argon. right scale) is 70 µm. The reduction in both the total electron scattering cross-section and the electron energy results in a large increase of electron drift velocity (for a compilation of electron-molecule cross sections see Ref. In noble gases. The number of electron-ion pairs per primary interaction. Once released in the gas.3: Probability of production of an electron of energy equal or larger than E (left scale). Particle detectors medium are shown in the figure. which results in “cooling” electrons into an energy range of the Ramsauer-Townsend minimum (located at ∼0.

α = 1/λi . Particle detectors 27 data [65] and theoretical calculations based on transport theory [66] permit estimates of drift and diffusion properties in pure gases and their mixtures. 28. and use for τ the values deduced from the drift velocity at B = 0 (Townsend expression). 28. drift velocity and longitudinal diffusion are not affected. This determines the characteristic shape of the detected signals in July 24. decreases exponentially with the field. the horizontal axis has to be scaled with the gas density. In a simple approximation.5. Examples for commonly used gases at NTP are given in Fig. the drift angle to √ the electric field vector is tan θB = ωτ and |v | = (E/B)(ωτ/ 1 + ω 2 τ 2 ). Standard deviations for longitudinal (σL ) and transverse diffusion (σT ) are given for one cm of drift. velocities and diffusion parameters can be computed with the program mentioned above. and scale with the square root of distance. the mean free path for ionization λi . The dotted line in Fig. where P is the pressure.4) to improve localization accuracy. of the Larmor frequency ω = eB/m and the mean collision time τ between electrons and molecules: v= τ e m 1 + ω2τ 2 E+ ωτ ω2τ 2 (E · B)B (E × B) + B B2 (28. as can be seen. For example. For parallel electric and magnetic fields.7. In mixtures containing electronegative molecules such as oxygen.4 and Fig. For E perpendicular to B. In presence of an external magnetic field. the large values of ωτ ∼ 20 at 5 T have been measured for Ar/CF4 /iC4 H10 (95:3:2). and therefore the capture probability is a function of applied field and may differ to a great extent for the same amount of additions other mixtures. although very fast. Since the collection time is inversely related to the drift velocity. As they experience increasing fields.13) To a good approximation. This reduction is exploited in time projection chambers (Sec. 28. in CF4 . the Lorentz force acting on electrons between collisions deflects the drifting swarm and modifies the drift properties.28. As an example.1% of oxygen to an Ar/CO2 mixture results in an electron capture probability about twenty times larger than the same addition to Ar/CH4 . and for moderate fields. its inverse. 28. the electron component. is the first Townsend coefficient. proportional to 1/P . The electron trajectories. Above a gas-dependent threshold. simple electrostatic consideration show that the largest fraction of the detected signal is due to the motion of ions receding from the wires. computed with the program MAGBOLTZ [67] as a function of electric field. is rarely seen. A simple theory (the friction force model) provides an expression for the vector drift velocity v as a function of electric and magnetic field vectors E and B. at moderate fields (up to 1 kV/cm) the addition of 0. gas kinetic theory provides the following relation between drift velocity v and mean electron-molecule collision time τ (Townsend’s expression): v = eEτ /m. electrons get enough energy to undergo ionizing collisions with molecules. water or CF4 .5 represents σT for the classic P10 mixture Ar/CH4 (90:10) at 4 T. √ while the transverse diffusion can be strongly reduced: σT (B) = σT (B = 0)/ 1 + ω 2 τ 2 . Capture cross-sections are strongly energy-dependent. For different conditions. 2008 18:04 . for example. As most of the avalanche growth occurs very close to the anodes. one can assume that the energy of the electrons is not affected by B. diffusion is smaller in fast-counting gases. electrons can be captured to form negative ions.

proportional mode. computed with MAGBOLTZ. N0 initial electrons multiply over a length x forming an electron avalanche of size N = N0 eαx .4: Computed electron drift velocity as a function of electric field in several gases at NTP [67]. 2008 18:04 . Particle detectors Figure 28. July 24. In uniform fields. Figure 28. N/N0 is the gain of the counter. with a fast rise followed by a gradually slower increase. 28. Fig.6 shows examples of Townsend coefficients for several gas mixtures. The dotted line shows σT for the P10 mixture at 4T [67].5: Electron longitudinal diffusion (σL ) (dashed lines) and transverse diffusion (σT ) (full lines) for 1 cm of drift.28 28. The so-called ion tail that limits the time resolution of the counter is usually removed by differentiation of the signal.

54 1. Ion He+ Ne+ Ar+ CH+ 4 CO+ 2 CH+ 4 CO+ 2 Mobility (cm2 V−1 s−1 ) 10. Particle detectors 29 Figure 28. The ion mobility µ.7 [68].09 July 24. due to a very effective charge transfer mechanism. 2008 18:04 . For mixtures.28.72 2. For a different temperature and pressure. their drift velocity in the fields encountered in gaseous counters (up to few kV/cm) is typically three orders of magnitude lower than for electrons.7: Gas He Ne Ar Ar Ar CH4 CO2 Mobility of ions in gases at NTP [68]. The diffusion of ions follows a simple square root dependence on the drift time. is constant for a given ion up to very high fields. Accumulation of ions in the gas counters may induce gain reduction and field distortions.7 1. the ratio of drift velocity to electric field. Table 28.26 1. T ) = µ(P0 . T0 )(T /P )(P0 /T0 ).87 1.6: Computed first Townsend coefficient α as a function of electric field in several gases at NTP [67]. Ions released by ionization or produced in the avalanches drift and diffuse under the influence of the electric field. only ions with the lowest ionization potential survive after a short path in the gas. Values of mobility at NTP (µ0 ) for ions in their own and other gases are given in Table 28.4 4. the mobility can be computed from the expression µ(P. with a coefficient that depends on temperature but not on the ion mass.

7. Small deviations result in forces displacing the wires alternatively below and above the symmetry plane. Table 28.7). Multi-Wire Proportional Chambers : Written April 2008 by Fabio Sauli (CERN) and Maxim Titov (CEA Saclay). anode wires are in equilibrium only for a perfect geometry. have been used for decades.8 gives approximate values for tungsten.2. the fields close to the wires E(r) and in the drift region ED are given by the approximations: E(r) = CV0 1 . up to a maximum stable length LM [71]: s 4π 0 TM LM = CV0 The maximum tension TM depends on the wire diameter and modulus of elasticity. and V0 the distance and difference of potential between anode and cathodes. Single-wire counters that detect the ionization produced in a gas by charged particle energy deposit. Table 28. For an overview see e. 2π 0 r ED = CV0 . Internal supports and spacers have been used in the construction of longer detectors. C the capacitance per unit length of the wires and a the anode wire radius. Maximum tension TM and stable length LM for tungsten wires with Wire diameter (µm) TM (newton) s (mm) LM (cm) 10 20 0. Because of electrostatic forces. A mesh of parallel anode wires at a suitable potential. Analytical expressions for the electric field can be found in many textbooks. Electrons released in the gas volume drift towards the anodes and avalanche in the increasing field. often with catastrophic results. s the wire spacing. and very large saturated pulses can be detected in the Geiger mode.30 28. and the corresponding maximum stable wire length under reasonable assumptions for the operating voltage V0 [72]. acts (almost) as an independent set of proportional counters (see Fig.g.70]. followed by charge multiplication and collection around a thin wire. detect and localize energy deposit by charged particles over large areas. 2008 18:04 . Particle detectors 28. Multiwire proportional chambers (MWPC’s) [69. inserted between two cathodes. 4. These displacement forces are countered by the mechanical tension of the wire. introduced in the late sixties. Ref.8: spacing s.16 0.65 1 2 25 85 July 24. 28. Good energy resolution is obtained in the streamer mode. 2 0s C= 2π 0 π( /s) − ln(2πa/s) where r is the distance from the center of the anode.

20 −0.10 −0. Many July 24. Detection of charge over a predefined threshold on the wires provides the event coordinates with the accuracy of the wire spacing.1 0.7: Electric field lines and equipotentials in (a) a multiwire proportional chamber and (b) a drift chamber.0 0.15 0.10 −0.2 −0. allows covering large areas at reduced cost.3 −0. the so-called center-of gravity (COG) method permits localization of tracks to sub-mm accuracy.15 0. Making use of the charge profile induced on segmented cathodes.3 Figure 28.10 (b) Drift chamber y-axis [cm] 0.00 −0. longitudinal localization can be obtained by measuring the ratio of collected charge at the two ends of resistive wires. They can achieve rms localization accuracies of 50 µm or better. several cm.00 −0. the position accuracy is 50 µm rms for tracks perpendicular to the wire plane.05 −0.05 0.05 −0.2 0. 2008 18:04 x-axis [cm] .05 0. developed in the early ’70’s.28. Drift chambers. Particle detectors 31 0. but degrades to ∼250 µm at 30◦ to the normal [73]. The intrinsic bi-dimensional characteristic of the COG readout has found numerous applications in medical imaging. The distance between anode wires. Due to the statistics of energy loss and asymmetric ionization clusters. can estimate the position of a track by exploiting the arrival time of electrons at the anodes if the time of the interaction is known.10 (a) Multiwire proportional chamber y-axis [cm] 0.15 0.15 −0.1 0.20 0.

Particle detectors Figure 28. and each cathode wire is surrounded by three anode wires.32 28. 28.8: Electric field lines and equipotentials in a multiwire drift module.8) [74]. all aimed at improving performance. designs have been introduced. the electric field structure is adjusted to compensate the distortions. a thicker wire at proper voltage between anodes (field wire) reduces the field at the middle point between anodes. improving charge collection (Fig. Sampling the drift time on rows of anodes within the same gas volume. together July 24. 2008 18:04 .9: Schematics drawing of a JET Chamber sector. Each anode wire is surrounded by six cathode wires. In some drift chambers design. In the original design. resulting in a linear space-to-drift-time relation [75]. Field-shaping electrodes Equipotential planes α Figure 28. Symmetrically decreasing potentials applied on cathode wires make more uniform and reinforce the drift fields. and with the help of suitable voltages applied to field-shaping electrodes.

for a proportional gain of 104 and 100 electrons released per track. with consequent field distortion. July 24. multi-wire structures have reliability problems when used in harsh or hard-to access environments. position-sensitive detectors based on wire structures are limited by basic diffusion processes and space charge effects in the gas to localization accuracies of 50–100 µm (see for example Ref. 28. a good knowledge of electron drift velocity and diffusion properties is required. Although very powerful in terms of performance. The time projection chamber [78] combines a measurement of drift time and charge induction on cathodes to obtain excellent tracking for high multiplicity topologies occurring at moderate rates (see Sec. 83). computed with commercial or custom-developed software [67. 28. As shown in Fig. this corresponds to a particle flux of 103 s−1 mm−1 (1 kHz/mm2 for 1 mm wire spacing).10: Charge rate dependence of normalized gain in a MWPC [82]. 2 and 71. 2008 18:04 .79]. 28. Suitable techniques for low-cost mass production have been developed for the needs of large experiments. each acting independently [80]. Introduced in the eighties. For an overview of detectors exploiting the drift time for coordinate measurement see Refs.9). gain reduction and degradation of spatial resolution.10 [82]. The production of positive ions in the avalanches and their slow drift before neutralization result in a rate-dependent accumulation of positive charge in the detector. 81). the proportional gain drops above a charge production rate around 109 electrons per second and mm of wire.8) and the JET chamber [77]( Fig. Despite various improvements. Straw and drift tube systems make use of large arrays of wire counters encased in individual envelopes. as the Transition Radiation Tracker and the Monitored Drift Tubes array for CERN’s LHC experiments (for a review see Ref. as a single broken wire can affect the whole detector. Particle detectors 33 Figure 28. with longitudinal localization through charge division on wires led to the concept of multiple arrays such as the multi-drift module [76]( Fig. 28.4).28. In all cases. independently from the avalanche size. This has to be combined with the knowledge of the electric fields in the structures.7.

double track resolution (∼ 500 µm). Their operating principles and performances are described in Sec. and noble gases with moderate amounts of polyatomic gases. are used. has been developed in the recent years and promises to overcome many of the limitations of MWPC devices—namely the multi-particle resolution and the rate capability. Application of a potential difference between the two sides of the GEM generates the electric fields indicated in Fig. The so-called “Magic Gas” (a mixture of argon. proportional gains around 104 are sufficient for detection of minimum ionizing particles. Consisting of a set of tiny metal strips laid on a thin insulating substrate. and single photo-electron time resolution in the ns range. a fast degradation of detectors due to the formation of polymers deposits (aging) is often observed. 2008 18:04 . Saclay) Micro-pattern Gas Detectors : Written October 2007 by M. A new generation of wireless gaseous detectors. Oed. Under optimum operating conditions. was the first of the micro-structure gas detectors. such as methane or carbon dioxide. electrons released by the ionization in the gas drift into the hole July 24. these detectors offer intrinsic high rate capability (fast signals with risetimes of a few ns and full widths of 20–100 ns). silicone oils). corresponding. With present-day electronics. excellent spatial resolution (∼ 30 µm). 28. The more powerful GEM and Micromegas concepts fulfill the needs of high-luminosity colliders with increased reliability in harsh radiation environments. Particle detectors Multiwire and drift chambers have been operated with a variety of gas fillings and operating modes. a total collected charge of a few coulombs per cm of wire can be reached before degradation. and alternatively connected as anodes and cathodes. revolutionizing cell size limits for many gas detector applications. a concept invented in 1988 by A. Titov (CEA Modern photolithographic technology has enabled a series of inventions of novel Micro-Pattern Gas Detector (MPGD) concepts: Micro-Strip Gas Chamber (MSGC) [85].3. at the cost of reduced survival due to aging processes. At high radiation fluxes. 28. isobutane and freon) [70] permits very high and saturated gains (∼ 106 ).g. Addition of fluorinated gases (e. The process has been extensively investigated. 28. depending on experimental requirements. Each hole acts as an independent proportional counter. CF4 ) or oxygen results in an etching action that can overcome polymer formation. The MSGC.3. the MSGC turned out to be easily damaged by discharges induced by heavily ionizing particles and destroying the fragile electrode structure [89].34 28. GEM [86]. micro-pattern gas detectors. for one mm spacing and at a gain of 104 . It consists of a thin-foil copperKapton-copper sandwich chemically perforated to obtain a high density of holes. The GEM detector was introduced by Fabio Sauli. often with conflicting results. to a total particle flux of ∼ 1014 MIP’s/cm2 . or even eliminate deposits. The hole diameter is typically between 25 µm and 150 µm.7.7. addition of small amounts of water in many (but not all) cases has been demonstrated to extends the lifetime of the detectors. overcoming the electronics limitations of the time.. while the pitch varies between 50 µm and 200 µm. Micromegas [87] and many others [88]. By using fine pitch size compared to classical wire chambers. a number of culprits has been identified (organic pollutants.11. but the issue of long-term survival of gas detectors in these gases is controversial [84].

12: Schematic drawing and typical dimensions of the Micromegas detector. On application of a potential difference between the two metal layers electrons released by ionization in the gas volume above the GEM are guided into the holes. while eliminating the risk of discharges (< 10−12 per hadron).11: Schematic view and typical dimensions of the hole structure in the GEM amplification cell. where charge multiplication occurs in the high field. 2008 18:04 . A unique property of the GEM detector is the complete decoupling of the amplification stage (GEM) and the readout electrode (PCB). Particle detectors 35 50 µm 140 µm Figure 28. This is the major advantage of the GEM technology [90]. Distributing the avalanche multiplication among several cascading electrodes allows the multi-GEM detectors to operate at overall gas gain above 104 in the presence of highly ionizing particles. which operates at unity gain and serves only as a charge collector. where they are amplified. Most of avalanche electrons are transferred into the gap below the GEM. July 24.28. limited by the mesh and the anode. and multiply in the high electric field (50–70 kV/cm). HV1 100 µm Drift gap Micromesh e− E2 Particle 40 kV/cm HV2 Amplification gap Anode plane Figure 28. Charges produced in the drift gap are drifting to the small amplification region. Electric field lines (solid) and equipotentials (dashed) are shown.

located between a thin metal grid (micromesh) and the readout electrode (strips/pads of conductor printed on an insulator board). GEM’s have entered the LHC program. GEM coupled to a VLSI pixel array could serve as a highly efficient x-ray polarimeter. GEM and Micromegas were read out by high-granularity (50 µm pitch) CMOS pixel chips assembled directly below the GEM or Micromegas amplification structure and serving as an integrated charge collecting anode [93. reaching 25 kHz/mm2 in the near-beam area. Employing the “bulk” technology. The performance and robustness of MPGD’s have encouraged their applications in high-energy and neutrino physics.12). It consists of a few mm drift region (electric field ∼1 kV/cm) and a narrow multiplication gap (25-150 µm. A time projection chamber (TPC) using GEM or Micromegas as a gas amplification device is also one of the main options for high-precision tracking at the International Linear Collider (ILC). and very good energy resolution (∼12% FWHM with 6 keV x rays). digitization and sparsification circuits integrated in the underlying active layers of the CMOS technology. Particle detectors Ioannis Giomataris introduced the micro-mesh gaseous structure (Micromegas). a photo-imageable polyimide film and the woven mesh are laminated together at a high temperature forming a single object. For minimum ionizing particle tracks a spatial resolution down to 20 µm was achieved with Medipix2 July 24. 72 large Micromegas (34 × 36 cm2 ) planes with an active area of 9 m2 are being built for the T2K TPC detector. Sensitive and low-noise electronics will enlarge the range of the MPGD applications. astrophysics. a spatial resolution of 70–100 µm and a time resolution of ∼ 10 ns. The basic idea is to build the whole detector in a single process: the anode plane with copper strips. which is able to reconstruct simultaneously initial direction and dynamics of photoelectron energy loss for the low energy (<10 keV) x rays. 28.95]. Both technologies have achieved a tracking efficiency of close to 100% at gas gains of about 104 . Over the past decade GEM and Micromegas detectors have become increasingly important. locally small variations of the amplification gap are compensated by an inverse variation of the amplification coefficient and therefore do not induce gain fluctuations. nuclear physics and neutron detection domain. With this arrangement avalanche electrons are collected on the top metal layer of the CMOS ASIC. 2008 18:04 . every input pixel is then directly connected to the amplification. which is a parallel-plate avalanche counter (Fig. Recently. The electric field is homogeneous both in the drift and amplification gaps. A fine-pitch GEM matching the pitch of pixel ASIC (50 µm) allows to achieve a superior single-electron avalanche reconstruction accuracy of 4 µm. giving rise to its excellent spatial resolution: 12 µm accuracy (limited by the the micromesh pitch) for MIPs [91]. A big step in the direction of the industrial manufacturing of large-size MPGD’s is the development of the “bulk” Micromegas technology [92]. they will be used for triggering in the LHCb Muon System and in the TOTEM Telescopes. Due to the narrow multiplication region in Micromegas. UV and visible photon detection. radiation therapy and electronic portal imaging devices.94. COMPASS is a first high-luminosity experiment at CERN which pioneered the use of large-area (∼ 40 × 40 cm2 ) GEM and Micromegas detectors for high-rate particle tracking. 50–70 kV/cm). The small amplification gap is a key element in Micromegas operation. which makes it suitable candidate for fast gas photo-multipliers.36 28.

The intrinsic 3D segmentation gives the TPC a distinct advantage over other large volume tracking detector designs which record information only in a 2D projection with less overall segmentation.05 mm × 0. 28. Recent developments in radiation hardness research with state-of-the-art MPGD’s are reviewed in Ref. Time-projection chambers : Victoria and TRIUMF. precise spatial measurements in the plane transverse to the magnetic field are most important. 2008 18:04 . in principle. of The Time Projection Chamber (TPC) concept. Performance has been recently improved by replacing these wire planes with micro-pattern gas detectors. and the lower electric field strength (∼ 50 kV/cm) in the multiplication region. Karlen (U. Similar performance is expected for Micromegas. Until recently. Gaseous TPC’s are often designed to operate within a strong magnetic field (typically parallel to the drift field) so that particle momenta can be estimated from the track curvature. Canada) Written September 2007 by D. can improve the July 24. Particle detectors 37 and Timepix CMOS chips coupled to GEM devices. ionization and momentum measurements can be combined to identify the types of particles observed in the TPC. An attractive solution for the construction of MPGD’s with pixel anode readout is the integration of the Micromegas amplification and CMOS chip by means of the “wafer post-processing” technology. Gas amplification of 103 –104 at the readout endplate is usually required in order to provide signals with sufficient amplitude for conventional electronics to sense the drifted ionization. is the basis for charged particle tracking in a large number of particle and nuclear physics experiments. can be explained by the separation of multiplication (GEM or parallel plate Micromegas amplification) and anode readout structures. Presently.05 mm pixels. the gas amplification system used in TPC’s have exclusively been planes of anode wires operated in proportional mode placed close to the readout pads. in combination with GEM or Micromegas [97].4. A uniform electric field drifts tracks of ions produced by charged particles traversing a medium. The signal amplitudes and arrival times are recorded to provide full 3D measurements of the particle trajectories. compared to the anode wire surface field (∼ 250 kV/cm). allowing pad areas to be reduced to the 10 mm2 scale or less. 96. namely GEM [86] and Micromegas [87] devices. Variance due to energetic δ-ray production is thus reduced. pitch and pattern can be easily adapted to match the geometry of any pixel readout chip. using the 50%–70% of the samples with the smallest signals. which are rather insensitive to aging compared to wire chambers. Advances in electronics miniaturization have been important in this development. particularly for pattern recognition in events with large numbers of particles.7. towards a surface segmented into 2D readout pads. For this application. Since the amount of ionization along the length of the track depends on the velocity of the particle. The estimator for the energy deposit by a particle is usually formed as the truncated mean of the energy deposits. well matched to the narrow extent of signals produced with micro-pattern gas detectors. Better properties of MPGD’s. either gas or liquid. With such fine granularity it is possible to count the number of ionization clusters along the length of a track which. the grid hole size. with 0. The sub-µm precision of the grid dimensions and avalanche gap size results in a uniform gas gain. the ultimate in fine segmentation TPC readout are silicon sensors. invented by David Nygren in the late 1970’s [78].28.

With a strong magnetic field parallel to the electric field and a gas with a large value of ωτ (also favored to reduce transverse diffusion as discussed below).15. In particular. the transverse displacements of the drifting electrons due to electric field distortions are reduced.14. typically 100 times further than in a comparable wire chamber design. The drift volume is 5 m long with a 5 m diameter. Particle detectors particle identification capability. Corrections for electric and/or magnetic field non-uniformities can be determined from control samples of electrons produced by ionizing the gas with UV laser beams.38 28. or the presence of ions in the active medium. For a gaseous TPC operated in a magnetic field. Outer containment volume Central electrode Inner containment volume End-plate Figure 28. 28. 28. Gas amplification is provided by planes of anode wires. Distortions can arise from a number of sources. for signals to arrive at the endplate. such as imperfections in the TPC construction.13 and Fig. 28. This can be resolved by matching the TPC data with that from an auxiliary detector providing additional spatial or timing information. For experiments with shorter intervals between events. this can produce ambiguities in the starting time for the drift of ionization. The particle identification performance is illustrated in Fig. July 24. The greatest challenges for a large TPC arise from the long drift distance. or from tracks emanating from calibration reactions. (28. The long drift distance means that there is a delay. deformations of the readout surface. Examples of two modern large volume gaseous TPC’s are shown in Fig.13).13: The ALICE TPC shown in a cutaway view. the electron drift velocity v is defined by Eq. typically 10–100 µs in a large gaseous TPC. for the original TPC in the PEP-4/9 experiment [98]. the long drift distance can make the device sensitive to small distortions in the electric field. In this mode of operation. from photoelectrons produced on the cathode. 2008 18:04 . it is essential to precisely map the magnetic field as the electron drift lines closely follow the magnetic field lines.

Special attention must be made in the choice of construction materials in order to avoid the release of other electronegative contaminants. which may allow operation of a TPC without a gating grid. Particle detectors 39 Outer wall E. the motion of positive ions is much slower than the electrons. For a gaseous TPC. is a few mm after drifting 1 m due to diffusion. the active medium must remain very pure. σDx . Large gaseous TPC’s built until now with wire planes have included a gating grid that prevent the positive ions from escaping into the drift volume in the interval between event triggers. the transverse extent of the electrons. 2008 1 1 + ω2τ 2 (28. the effect can be alleviated by the choice of a gas with low intrinsic diffusion or by operating in a strong magnetic field parallel to the drift field with a gas which exhibits a significant reduction in transverse diffusion with magnetic field. With a strong magnetic field.14) 18:04 . For example. as small amounts of contamination can absorb the ionization signal. Micromegas devices are used for gas amplification and readout. The drift volume is 2 m×2 m×0. σDx (B)/σDx (0) = √ July 24. and so the positive ions produced by many events may exist in the active volume. O2 must be kept below a few parts in 105 .8 m. B directions Inner wall and field cage Beam direction Front end cards Central cathode Central cathode HV Figure 28. σDx can be reduced by as much as a factor of 10. In a gaseous TPC.28.14: One of the 3 TPC modules for the near detector of the T2K experiment. Given the long drift distance in a large TPC. otherwise a large fraction of the drifting electrons will become attached. in a typical large gaseous TPC. Of greatest concern is the ions produced in the gas amplification stage. Diffusion degrades the position information of ionization that drifts a long distance. For typical operation without magnetic field. Micro-pattern gas detectors release much less positive ions than wire planes operating at the same gain.

4 times that for the minimum at lower β. where ωτ is defined above.1 1 10 Figure 28. The spatial resolution achieved by a TPC is determined by a number of factors in addition to diffusion. experiments with a preferred track direction should have pad boundaries aligned with that direction. This ratio increases to 1. 8. giving an ultimate single row resolution of order 100 µm.15: The PEP4/9-TPC energy deposit measurements (185 samples.15) σx = √ n where n is the effective number of electrons collected by the pad row. Traditional TPC’s with wire plane amplification suffer from the effects of non-parallel electric and magnetic fields near the wires that July 24. For this reason.6 at atmospheric pressure. The slow drift velocity and small ωτ of negative ions must be compatible with the experimental environment.40 28. which uses a special gas mixture that attaches electrons immediately as they are produced. The diffusion limited position resolution from the information collected by a single row of pads is σDx (28. The drifting negative ions exhibit much less diffusion than electrons.5 atm Ar-CH4 80:20). The ionization rate at the Fermi plateau (at high β) is 1. Diffusion is significantly reduced in a negative-ion TPC [99]. Non-uniform ionization along the length of the track is a particularly important factor. and is responsible for the so-called “track angle” and “E×B ” effects. the ionization fluctuations will increase the variance in the position estimate from that row. Particle detectors 32 Energy deposit per unit length (keV/cm) 28 µ π K p D 24 20 e 16 12 8 0. If the boundaries between pads in a row are not parallel to the track. 2008 18:04 Momentum (GeV/c) .

Mylar. and the x-ray energy is Ex .7. but for multiple boundary crossings interference leads to saturation near a Lorentz factor γ sat √ = 0. since their feature size is much smaller than that of a wire grid. The TR signal in the active regions is in most cases superimposed upon the particle’s ionization losses. optimized for a minimum total radiation length at maximum TR yield and total TR absorption. 1 is its thickness. about 2 cm for Xe-filled gas chambers. 27. Those detectors normally work as threshold devices. A. ∼ ranging from a few keV to a few dozen keV. The atomic number considerations follow from the dominant photoelectric absorption cross section per atom 3 going roughly as Z n /Ex . consists of few hundred foils (for instance 300 20 µm thick polypropylene foils). providing another reason for active layers with high Z. materials such as polypropylene. as discussed in Sec. Micro-pattern gas detectors exhibit a much smaller E×B effect. Even with a relatively thin radiator stack. are emitted at a characteristic angle 1/γ from the particle trajectory. The x rays. In most of the detectors used in particle physics the radiator parameters are chosen to provide γ sat ≈ 2000. Inst. July 24. 2. Transition radiation detectors (TRD’s) : Written August 2007 by P. To increase the average TR photon energy further. Romaniouk (Moscow Eng.28. by distributing small-diameter straw-tube detectors uniformly or in thin layers throughout the radiator material (foils or fibers). and are being used in the ALICE [104] experiment. Since the TR yield is about 1% per boundary crossing. Nevski (BNL). ensuring the best electron/pion separation in the momentum range 1 GeV/c < p < 150 GeV/c. The TR intensity for a single boundary crossing always increases with γ. where ω1 is the radiator plasma frequency. thereby degrading the resolution because of the non-uniform ionization. carbon. This can be achieved. A classical detector is composed of several similar modules which respond nearly independently. The radiator. and 2 the spacing. a detector module might consist of low-Z foils followed by a high-Z active layer made of proportional counters filled with a Xe-rich gas mixture. radiation from multiple surface crossings is used in practical detectors. 2008 18:04 . and dE/dx|min and plasma energies for many materials are given in pdg. Particle detectors 41 rotate ionization segments. The detector thickness. and (rarely) lithium are used as radiators. Such detectors were used in the NA34 [102]. These drop a little faster than Z/A with increasing Z.101]. where n varies between 4 and 5 over the region of interest. However for photon * Photon absorption coefficients for the elements (via a NIST link). In the simplest concept.* To minimize self-absorption. radiation below 5 keV is mostly lost in the radiators themselves. is optimized to absorb the shaped x-ray spectrum. part of the radiator far from the active layers is often made of thicker foils.gov/AtomicNuclearProperties.) Transition radiation (TR) x rays are produced when a highly relativistic particle (γ > 103 ) crosses a refractive index interface. 28. ∼ ∼ One can distinguish two design concepts—“thick” and “thin” detectors: 1.6 ω1 1 2 /c [100. & Phys.lbl. A dominant fraction of the soft TR photons is absorbed in the radiator itself. In another TRD concept a fine granular radiator/detector structure exploits the soft part of the TR spectrum more efficiently. for instance. NOMAD [103].7.5.

the rejection power against pions increases as 5 · 10L/38 . 28. The major factor in the performance of any TRD is its overall length. Although the values mentioned above are typical for most of the plastic radiators used with Xe-based detectors. on the number of clusters—energy depositions observed above an optimal threshold (usually in the 5 to 7 keV region). for a variety of detectors. Recent TRDs for particle astrophysics are designed to directly measure the Lorentz factor of high-energy nuclei by using the quadratic dependence of the TR yield on nuclear charge [109. Cluster-counting replaces the Landau-Vavilov distribution of background ionization energy losses with the Poisson statistics of δ-electrons. thickness and spacing. The radiator configuration ( 1 . 108. The discrimination between electrons and pions can be based on the charge deposition measured in each detection module. or on more sophisticated methods analyzing the pulse shape as a function of time. which absorb most of the TR radiation and where the ionization loss fluctuations are small.16. they vary significantly depending on detector parameters: radiator material. AMS [107]) experiments. The total energy measurement technique is more suitable for thick gas volumes. 2008 18:04 . covering a range of particle energies from a few GeV to 40 GeV. The latter distribution is narrower that the Landau-Vavilov distribution. The effective thickness of the Xe gas per straw is about 2. Examples of the detectors using this approach can be found in both accelerator (ATLAS [105]) and space (PAMELA [106]. are rescaled to an energy of 10 GeV when possible. where the range of validity is L ≈ 20–100 cm. etc. Both TR photon absorption and the TR build-up significantly affect the detector performance. Exotic ∼ radiator materials such as aluminum and unusual TR detection methods (Compton scattering) are used such cases [109]. The excellent coordinate resolution of the Si detectors allows spatial separation of the TR photons from particle ionization tracks with relatively modest distances between radiator and detector. Particle detectors energies above this value the absorption becomes smaller and the radiation can be registered by several consecutive detector layers. For example. 2 ) is tuned to extend the TR yield rise up to γ < 105 using more energetic part of the TR spectrum (up to 100 keV). The cluster-counting method works better for detectors with thin gas layers. July 24. In this work Si-microstrip detectors operating in a magnetic filed are used both for particle and TR detection. Phenomenologically.3 mm and the average number of foils per straw is about 40 with an effective foil thickness of about 20 µm.42 28. which shows.110]. their position. This is illustrated in Fig. thus creating a strong TR build-up effect. responsible for the distribution tails. the construction of the sensitive chambers. Many recent TRDs combine particle identification with charged-track measurement in the same detector [104. the pion efficiency at a fixed electron efficiency of 90% as a function of the overall detector length. The experimental data. This provides a powerful tool for electron identification even at very high particle densities. in the ATLAS TR tracker charged particles cross about 35 effective straw tube layers embedded in the radiator material. Thus careful simulations are usually needed to build a detector optimized for a particular application.105]. where the fluctuations of the ionization losses are big. Another example of this combination is described in Ref.

D0 M. Particle detectors 43 KEK UA2 H. An example of an RPC structure is shown in Fig. R 806 A.Butt et al.Fabjan et al. The x and y position of the discharge can be measured if the strips on opposite sides of the gap are orthogonal. Band (U. July 24. The passage of a charged particle initiates an electric discharge. ATLAS NOMAD AMS ALICE PAMELA NA34 (HELIOS) C.16: Pion efficiency measured (or predicted) for different TRDs as a function of the detector length for a fixed electron efficiency of 90%. The gas is chosen to absorb UV photons in order to limit transverse growth of discharges. 28. Wisconsin).1 Pion efficiency 0.7. The backs of the plates are coated with a lower-resistivity paint or ink (∼105 Ω/ ).* Most commonly. The resistive-plate chamber (RPC) was developed by Santonico and Cardarelli in the early 1980’s [111] as a low-cost alternative to large scintillator planes. Resistive-plate chambers : Revised September 2007 by H. an RPC is constructed from two parallel high-resistivity (109 –1013 Ω-cm) glass or phenolic (Bakelite)/melamine laminate plates with a few-mm gap between them which is filled with atmospheric-pressure gas.6. * It was based on earlier work on a spark counter with one high-resistivity plate [112]. H. 2008 18:04 50 100 Total detector length (cm) 200 .17. and a high potential (7–12 kV) is maintained between them. ZEUS : 0.28. making sensitive electronics unnecessary. The signal readout is via capacitive coupling to metallic strips on both sides of the detector which are separated from the high voltage coatings by thin insulating sheets. The sensitivity of the detector outside of this region is unaffected. The plot is taken from [102] with efficiencies of more recent detectors [105–106] added (ATLAS to PAMELA).R. whose size and duration are limited since the current reduces the local potential to below that needed to maintain the discharge. When operated in streamer mode.Weidkamp H. Bungener et al.01 0.Gr ssler et al. 28. the induced signals on the strips can be quite large (∼300 mV).001 10 20 Figure 28.Holder et al.

17: Schematic cross section of a typical RPC. Bakelite RPC’s can be efficient at 10–100 Hz/cm2 . in which the gas and high voltage have been tuned to limit the growth of the electric discharge.3 mm) separated by thin(0. RPC’s have inherent rate limitations since the time needed to re-establish the field after a discharge is proportional to the chamber capacitance and plate resistance. the efficiency of streamer-mode glass RPC’s begins to fall above ∼0. Typical avalanche-mode RPC’s have a signal charge of about 10 pC and can be efficient at 1 kHz/cm2 . The average charge per streamer is 100–1000 pC. and thus requires a more sophisticated and careful electronic design. The total area is about 10. a gas mixture of tetrafluoroethane (C2 H2 F4 ) with 2–5% isobutane and 0. For avalanche mode operation. Particle detectors Aluminum x pickup strips HV Foam Insulator Graphite Bakelite Gas Bakelite 2 mm 2 mm 2 mm Graphite Insulator y pickup strips Spacers Aluminum 1 mm 1 cm Foam Figure 28. In streamer mode. Typically. in this case the single-gap streamer-mode BaBar RPC. where three layers of pairs of RPC’s are used to trigger the drift tube arrays between the pairs.000 m2 . Non-flammable and more environmentally friendly gas mixtures have been developed. These RPC’s provide a spatial resolution of 1 cm and a time resolution of 1 ns at an efficiency ≥ 99%. Developments of multiple-gap RPC’s [113] lead to RPC designs with much better timing resolution (∼ 50 ps) for use in time-of-flight particle identification systems. The need for higher rate capability led to the development of avalanche-mode RPC’s. The avalanche discharge produces a much smaller induced signal on the pickup strips (∼1 mV) than streamers. An example of large-scale RPC use is provided by the muon system being built for the ATLAS detector. Because of Bakelite’s lower bulk resistivity. The outer plates are connected to high voltage and ground while the inner plate is electrically isolated and floats to a stable equilibrium potential. The observed RPC intrinsic time resolution of 127 ps may have July 24.4–10% sulfur hexafluoride (SF6 ) is typical.4 Hz/cm2 . Efficiencies of > 92% for single gaps can be improved by the use of two or ∼ more gas gaps with shared pickup strips. various mixtures of argon with isobutane and tetrafluoroethane have been used. A pioneering design used by the HARP experiment [114] has two sets of 2 thin gas gaps (0. preventing streamer formation. Many variations of the initial RPC design have been built for operation in either mode.7 mm) glass plates.44 28. 2008 18:04 .

L3 and HARP) have reported reliable performance.. Second-generation BaBar RPC’s incorporating many of the above improvements have performed reliably for over two years [117]. the graphite layer used to distribute the high voltage over the Bakelite became highly resistive (100 kΩ/ → 10 MΩ/ ). Unlike glass RPC’s. and moderate spatial resolution make them attractive in many situations. The BaBar problems and the proposed use of Bakelite RPC’s in the LHC detectors prompted detailed studies of RPC aging and have led to improved construction techniques and a better understanding of RPC operational limits. Several experiments (e. Glass RPC’s have had fewer problems. In addition. Several different failure modes diagnosed in the first-generation BaBar Bakelite RPC’s caused the average efficiency of the barrel RPC’s to fall from > 90% to 35% in five years. ease of construction. With many of these problems solved. as seen by the history of the BELLE chambers. July 24. Fonte provides useful review [115] of other RPC designs.g. Particle detectors 45 been limited by amplifier noise. However. It was found that water vapor in the input gas was reacting with fluorine (produced by the disassociation of the tetrafluoroethane in the streamers) to produce hydrofluoric acid. Molded gas ∼ inlets and improved cleanliness during construction have reduced the noise rate of new chambers. The graphite layer has been improved and should be stable with integrated currents of < 600 mC/cm2 . high efficiency. oil collected on the spacers between the gaps or formed oil-drop bridges between the gaps. This led to large leakage currents (50–100 µA in some chambers) which persisted even when the temperature was regulated at 20◦ C. particularly those requiring fast timing and/or large-area coverage. Under warm conditions (32◦ C) and high voltage. Operational experience with RPC’s has been mixed. A rapid growth in the noise rate and leakage current in some of the BELLE glass RPC’s was observed during commissioning. The BELLE RPC’s have now operated reliably for more than 5 years. 2008 18:04 . ∼ The linseed oil which is used in Bakelite RPC’s to coat the inner surface [116] had not been completely cured. good time resolution. The use of copper gas piping to insure the dryness of the input gas stopped the problem. leading to increased noise rates and lower efficiencies. resulting in lowered efficiency in some regions and the complete death of whole chambers. the severe problems experienced with the BaBar RPC’s have raised concerns about the long-term reliability of Bakelite RPC’s. new-generation RPC’s are now being or soon will be used in about a dozen cosmic-ray and HEP detectors. Their comparatively low cost. The acid etched the glass surfaces. Bakelite RPC’s have been found to require humid input gases to prevent drying of the Bakelite (increasing the bulk resistivity) which would decrease the rate capability.28.

a highly doped p electrode and a lightly doped n region. Furthermore. The conductive p and n regions together with the depleted volume form a capacitor with the capacitance per unit area C = /W ≈ 1 [pF/cm] /W . the strip-to-strip fringing capacitance is ∼ 1–1. Silicon detectors are p-n junction diodes operated at reverse bias. (28. CdZnTe. providing excellent position resolution. Particle detectors 28. (28. This forms a sensitive region depleted of mobile charge and sets up an electric field that sweeps charge liberated by radiation to the electrodes.8. Semiconductor detectors are widely used in modern high-energy physics experiments. the density of silicon and its small ionization energy result in adequate signals with active layers only 100–300 µm thick.46 28. Spieler (LBNL). Detectors typically use an asymmetric structure. and ρ(V + Vbi ) for p-type material.5 V for resistivities typically used in detectors) doping concentration electronic charge dielectric constant = 11. The thickness of the depleted region is W = where V = Vbi = N= e= = ρ= µ= 2 (V + Vbi )/N e = 2ρµ (V + Vbi ) . so the detection sensitivity is determined by signal charge and capacitance.16) external bias voltage “built-in” voltage (≈ 0. but germanium.17) In strip and pixel detectors the capacitance is dominated by the fringing capacitance. For example. and diamond are also useful in some applications. Integrated circuit technology allows the formation of high-density micron-scale electrodes on large (10–15 cm diameter) wafers. 2008 18:04 . They are the key ingredient of high-resolution vertex and tracking detectors and are also used as photodetectors in scintillation calorimeters. For a comprehensive discussion of semiconductor detectors and electronics see Ref. e. Semiconductor detectors depend crucially on low-noise electronics (see Sec. 118. CdTe.5 [µm/ Ω-cm · V] × √ W = 0.g. so that the depletion region extends predominantly into the lightly doped volume. July 24.3 [µm/ Ω-cm · V] × ρ(V + Vbi ) for n-type material. Silicon semiconductor detectors Updated September 2007 by H.5 pF cm−1 of strip length at a strip pitch of 25–50 µm. so the signals are also fast (typically tens of ns).9 0 ≈ 1 pF/cm resistivity (typically 1–10 kΩ cm) charge carrier mobility = 1350 cm2 V−1 s−1 for electrons = 450 cm2 V−1 s−1 for holes or √ W = 0. The most commonly used material is silicon. gallium-arsenide.9). 28.

where Φ is the particle fluence and α the damage coefficient (α ≈ 3 × 10−17 A/cm for minimum ionizing protons and pions after long-term annealing. Particle detectors 47 Measurements on silicon photodiodes [119] show that for photon energies below 4 eV one electron-hole (e-h) pair is formed per incident photon. Position resolution is limited by transverse diffusion during charge collection (typically 5 µm for 300 µm thickness) and by knock-on electrons. Bulk damage due to displacement of atoms from their lattice sites. or µm-scale pixels. DEPFET’s. and build-up of space charge that changes the required operating voltage.28. monolithic pixel devices that integrate sensor and electronics (MAPS).e. It is larger than the bandgap energy because phonon excitation is required for momentum conservation. carrier trapping. In magnetic fields. The effects of charge build-up are strongly dependent on the device structure and on fabrication details.e. For an overview and further discussion see Ref. The collection time is limited by velocity saturation at high fields (approaching 107 cm/s at E > 104 V/cm). above ∼ 1. 3. e. This leads to increased leakage current. 2008 18:04 . 2. and can be reduced further by operating the detector with “overbias.5 keV. Radiation damage occurs through two basic mechanisms: 1. Thus.1 in Si).g. Electrodes can be in the form of cm-scale pads. The mean energy Ei required to produce an e-h pair peaks at 4. Various readout structures have been developed for pixels. the dose can be specified in rad (or Gray) independent of particle type. which can initiate a chain of subsequent displacements. the measured signal fluctuations are usually dominated by electronic noise or energy loss fluctuations in the detector. which leads to increased surface leakage currents. Displacement damage depends on the nonionizing energy loss and the energy imparted to the recoil atoms. Charge collection time decreases with increasing bias voltage. α ≈ 2 × 10−17 A/cm for July 24. 118. it is critical to consider both particle type and energy. the Lorentz drift deflects the electron and hole trajectories and the detector must be tilted to reduce spatial spreading (see “Hall effect” in semiconductor textbooks). the variance in the number of charge carriers N = E/Ei produced by an √ absorbed energy E is reduced by the Fano factor F (about 0. the most probable charge deposition in a 300 µm thick silicon detector is about 3.5 fC (22000 electrons).” i. It assumes a constant value. Resolutions of 2–4 µm (rms) have been obtained in beam tests. strips.4 eV for a photon energy around 6 eV. σN = F N and the energy resolution σE /E = F Ei /E. and holes within about 25 ns. a bias voltage exceeding the value required to fully deplete the device. at an average field of 104 V/cm the collection time is about 15 ps/µm for electrons and 30 ps/µm for holes.67 eV at room temperature.. The increase in reverse bias current due to bulk damage is ∆Ir = αΦ per unit volume. CCD’s. In strip detectors the inter-strip isolation is affected. However. Since both electronic and lattice excitations are involved. i. Hence. electrons are collected within about 10 ns. damage clusters. and hybrid pixel devices that utilize separate sensors and readout IC’s connected by two-dimensional arrays of solder bumps. Surface damage due to charge build-up in surface layers. Since the damage is proportional to the absorbed energy (when ionization dominates). For minimum-ionizing particles. In typical fully-depleted detectors 300 µm thick.

123]. until the positive and negative space charge balance and very little voltage is required to collect all signal charge.web. For a more detailed summary see Ref. in the electronics use of standard “deep submicron” CMOS fabrication processes with appropriately designed circuitry has increased the radiation resistance to fluences > 1015 cm−2 of minimum ionizing protons or pions. various techniques have been applied to neutralize the damage sites.48 28. 125. These trap electrons. The increase in leakage current with fluence.web. and the required operating voltage increases (V ∝ N ).e. The safe limit on operating voltage ultimately limits the detector lifetime. At this damage level. This has the same effect as a change in resistivity. Detectors with columnar electrodes normal to the surface can also extend operational lifetime Ref. Particle detectors 1 MeV neutrons). At fluences beyond 1015 cm−2 decreased carrier lifetime becomes critical [122. charge loss due to recombination and trapping also becomes significant and the high signal-to-noise ratio obtainable with lowcapacitance pixel structures extends detector lifetime. It is critical to choose the operating temperature judiciously (−10 to 0◦ C in typical collider detectors) and limit warm-up periods during maintenance.ch/rd48 and RD50.2 eV.ch/rd50.18) where E = 1. The occupancy of the defect charge states is strongly temperature dependent. on the other hand. Currently. July 24.cern. competing processes can increase or decrease the required operating voltage. Displacement damage forms acceptor-like states. 2008 18:04 . At larger fluences the negative space charge dominates. which in turn requires an increase in the applied voltage to sweep signal charge through the detector thickness. the required voltage drops initially with fluence.. i. building up a negative space charge. For a comprehensive discussion of radiation effects see Ref. 124 and and the web-sites of the ROSE and RD50 collaborations at RD48. Strip and pixel detectors have remained functional at fluences beyond 1015 cm−2 for minimum ionizing protons. 121. additional doping with oxygen increases the allowable charged hadron fluence roughly three-fold [120]. appears to be unaffected by resistivity and whether the material is n or p-type. Since the effect of radiation damage depends on the electronic activity of defects.cern. The reverse bias current depends strongly on temperature IR (T2 ) = IR (T1 ) T2 T1 2 exp − E 2k T1 − T2 T1 T2 (28. so rather modest cooling can reduce the current substantially (∼ 6-fold current reduction in cooling from room temperature to 0◦ C). Strip detectors specifically designed for high voltages have been extensively operated at bias voltages >500V. the lifetime of detector systems is still limited by the detectors. For example.

Shot noise and thermal noise have a “white” frequency distribution. Since the bias resistor effectively shunts the input. as the capacitor Cb passes current fluctuations to ground. This “shot noise” ind is represented by a current noise generator in parallel with the detector.e. Particle detectors 49 28. Low-noise electronics Revised August 2003 by H. it acts as a current generator inb and its noise current has the same effect as the shot noise current from the detector. Any other shunt resistances can be incorporated in the same way. Bias voltage is applied through resistor Rb and the signal is coupled to the preamplifier through a blocking capacitor Cc . which tailors the overall frequency response to optimize signal-to-noise ratio while limiting the duration of the signal pulse to accommodate the signal pulse rate. the electrode resistance. i.g. The preamplifier provides gain and feeds a pulse shaper. either to improve energy resolution or to allow a low detection threshold. Generally.28. The series resistance Rs represents the sum of all resistances present in the input signal path. and parasitic resistances in the input transistor. DETECTOR BIAS Cb PREAMPLIFIER BIAS RESISTOR PULSE SHAPER Rb Cc Rs OUTPUT DETECTOR Cd Figure 28. 28. Spieler (LBNL). 2008 18:04 . Many detectors rely critically on low-noise electronics. 28.9. all amplifiers provide some form of pulse shaping due to their limited frequency response. Resistors exhibit noise due to thermal velocity fluctuations of the charge carriers. for example. a relevant model for most detectors. fluctuates due to electron emission statistics. e. resistors shunting the input act as noise current sources and resistors in series with the input act as noise voltage sources (which is why some in the detector community refer to current and voltage noise as “parallel” and “series” noise).18. Conversely. The detector is represented by a capacitance Cd .19) includes both current and voltage noise sources. Even if not explicitly stated. the series resistor Rs acts as a voltage generator. This noise source can be modeled either as a voltage or current generator. The electronic noise of the amplifier is described fully by a combination of voltage and current sources at its input. A typical detector front-end is shown in Fig.18: Typical detector front-end circuit. The leakage current of a semiconductor detector. shown as ena and ina . the spectral power densities dPn /df ∝ di2 /df ∝ de2 /df are constant with the magnitudes n n July 24. any input protection networks. The equivalent circuit for the noise analysis (Fig.

Typical amplifier noise parameters ena and ina are of order T √ nV/ Hz and pA/ Hz. Since radiation detectors typically convert the deposited energy into charge. TS 18:04 (28. Id the detector bias current. The total noise at the output of the pulse shaper is obtained by integrating over the full bandwidth of the system.19) where e is the electronic charge. Particle detectors DETECTOR BIAS RESISTOR SERIES RESISTOR AMPLIFIER + PULSE SHAPER Rs ens ena Cd Rb ind inb ina Figure 28. the corresponding number of electrons. the system’s noise level is conveniently expressed as an equivalent noise charge Qn . purely random noise produces a Gaussian signal distribution. i2 = 2eId . Since the individual noise contributions are random and uncorrelated.19: Equivalent circuit for noise analysis. The spectral density of the 1/f noise voltage is e2 = nf Af . 2008 . or the equivalent deposited energy (eV). The equivalent noise charge is commonly expressed in Coulombs. dielectrics and semiconductors can introduce additional fluctuations whose noise power frequently exhibits a 1/f spectrum. f (28. nb Rb e2 = 4kT Rs . For a capacitive sensor Q2 = i2 Fi TS + e2 Fv n n n C2 + Fvf Af C 2 . they add in quadrature. ns (28.21) July 24. Trapping and detrapping processes in resistors. nd 4kT i2 = .50 28. which is equal to the detector signal that yields a signal-to-noise ratio of one. which is added to the noise voltage in the input circuit.20) where the noise coefficient Af is device specific and of order 10−10 –10−12 V2 . resulting in a frequency-dependent noise voltage in /(ωCd ). k the Boltzmann constant and √ the temperature. Superimposed on repetitive detector signal pulses of constant magnitude. A fraction of the noise current flows through the detector capacitance.

the total noise goes through a minimum. Fv . whereas the voltage noise decreases with increasing shaping time.e. Fv = TS 2 ∞ −∞ dW (t) dt 2 dt . For more details see Refs. for example.g.23) As the characteristic time TS is changed. Fi . pulse duration.6 × 10 4 ns (pF)2 (nV)2 /Hz C e2 n 2 (28. e. At short shaping times the voltage noise dominates. judiciously selecting all resistances in the input circuit.e. to 0. 2008 18:04 .28.. one can use the following equation. and Fvf depend on the shape of the pulse determined by the shaper and Ts is a characteristic time.. and choosing the optimum shaping time constant. A pulse shaper formed by a single differentiator and integrator with equal time constants has Fi = Fv = 0.45 and 1. the characteristic time Ts . For the circuit shown in Fig.19. Q2 = 2eId + 4kT /Rb + i2 Fi TS n na 2 2 + 4kT Rs + e2 Fv Cd /TS + Fvf Af Cd . 28. Increasing pulse symmetry tends to decrease Fi and increase Fv (e. The overall noise bandwidth. The form factors Fi . so for a given shaper topology the 1/f contribution to Qn is independent of Ts . which assumes an FET amplifier (negligible ina ) and a simple CR–RC shaper with time constants τ (equal to the peaking time): (Qn /e)2 = 12 1 kΩ τ Id τ + 6 × 105 nA · ns ns Rb + 3.22) where for time-invariant pulse-shaping W (t) is simply the system’s impulse response (the output signal seen on an oscilloscope) with the peak output signal normalized to unity.0 for a shaper with one CR differentiator and four cascaded integrators). independent of the shaping time constant. For quick estimates. i. Fig.20 shows a typical example. Furthermore. the contribution of noise voltage sources to Qn increases with detector capacitance. where the current and voltage contributions are equal. July 24. Pulse shapers can be designed to reduce the effect of current noise. (28. Fv are easily calculated Fi = 1 2TS ∞ −∞ [W (t)]2 dt . depends on the time constant.g. however.. Increasing the detector capacitance will increase the voltage noise and shift the noise minimum to longer shaping times. mitigate radiation damage. i. na (28. Noise with a 1/f spectrum depends only on the ratio of upper to lower cutoff frequencies (integrator to differentiator time constants). the peaking time of a semi-gaussian pulse or the sampling interval in a correlated double sampler. 126 and 127.9 and Fvf = 4. The contribution from noise currents increases with shaping time. τ Noise is improved by reducing the detector capacitance and leakage current. The noise minimum is flattened by the presence of 1/f noise.24) . 28. Particle detectors 51 where C is the sum of all capacitances shunting the input. whereas at long shaping times the current noise takes over.

Silicon strip detectors typically operate at ∼ 103 e electrons. Increased voltage noise is shown as an example. Capacitive matching is useful with field-effect transistors. the lowest noise is obtained when the input capacitance of the transistor is equal to the detector capacitance. which does depend on shaping time. The noise parameters of the amplifier depend primarily on the input device. so reducing the detector leakage current and increasing the bias resistance will allow long shaping times with correspondingly lower noise. In bipolar transistors. In special cases where the noise of a transistor scales with geometry.52 28.26) Practical noise levels range from ∼ 1e for CCD’s at long shaping times to ∼ 104 e in high-capacitance liquid argon calorimeters. Q2 n. i. the base current sets a lower bound on the noise current. Qn.25) 1 10 100 where βDC is the DC current gain.e. the minimum obtainable noise is independent of shaping time.01 0. Fi TS (28. decreasing noise voltage with increasing input capacitance. July 24. but not bipolar transistors. albeit at the expense of power dissipation.20: Equivalent noise charge vs shaping time. Particle detectors 10000 5000 Equivalent noise charge (e) Increasing V noise 2000 1000 Total 500 1/f noise Total 200 C 100 0. Changing the voltage or current noise contribution shifts the noise minimum. whereas pixel detectors with fast readout provide noise of several hundred electrons. but only at the optimum collector current IC .. In bipolar transistors.min = 4kT √ C βDC Fi Fv at Ic = kT C e βDC Fv 1 .1 e urr n nt ois e Vo lta ge no ise Shaping time (µs) Figure 28.min /e ≈ 250 C/pF . so these devices are best at short shaping times. In field effect transistors. the noise current contribution is very small. 2008 18:04 . For a CR–RC shaper and βDC = 100. (28.

In either case.27) (dS/dt)ST S/N where σn is the rms noise and the derivative of the signal dS/dt is evaluated at the trigger level ST . * λI ≈ 35 g cm−2 A1/3 .28. The timing distribution may shift with signal level (“walk”).0 g cm−2 in uranium. either in hardware or software [10].10. When amplifiers are cascaded. which is invariably necessary. tr ≈ t2 + t2 + . while hadronic calorimeters are usually compromised at 5–8 λI . so the hadronic cascade is contained in a succession of different structures. CsI. an oscilloscope with 350 MHz bandwidth has a 1 ns rise time. high atomic number. and perhaps a more poorly sampling catcher in the back.e.35/fU . Calorimeters A calorimeter is designed to measure the energy deposited in a contained electromagnetic (EM) or hadronic shower. the individual rise times add in quadrature. or ionizing noble liquids. cost.. where “many” is determined by physical size. EM calorimeters tend to be 15–30 X0 deep. 2008 18:04 .1 g cm−2 (Fe) to 209 g cm−2 (U). the amplifier bandwidth should match the rise-time of the detector signal. For a more detailed introduction to detector signal processing and electronics see Ref. The 10 to 90% rise time of an amplifier with bandwidth fU is 0. Homogeneous calorimeters (usually electromagnetic) may be built with inorganic heavy (high-Z) scintillating crystals such as BGO. non-scintillating Cherenkov radiators such as lead glass and lead fluoride. especially in the EM case.gov/AtomicNuclearProperties. contributes signal. In a homogeneous calorimeter the entire volume is sensitive. For example. July 24. and other factors. The characteristic interaction distance for an electromagnetic interaction is the radiation length X0 . the characteristic nuclear interaction length λI varies from 132. for actual values see pdg.. There are homogeneous and sampling calorimeters. the calorimeter must be many interaction lengths deep. In all cases there is a premium on high density. so minimizing the total capacitance at the input is also important. to contain the shower as compactly as possible. in a real experiment there is likely to be an EM calorimeter in front of the hadronic section. but this can be corrected by various means. which ranges from 13. To increase dS/dt without incurring excessive noise. Particle detectors 53 In timing measurements.8 g cm−2 in iron to 6. NaI. the slope-to-noise ratio must be optimized.* Similarly. i. 118. The “jitter” σt of the timing distribution is σn tr σt = ≈ . and PWO.. At high signal-to-noise ratios. A sampling calorimeter consists of an active medium which generates signal and a passive medium which functions as an absorber. + t2 rn r1 r2 Increasing signal-to-noise ratio also improves time resolution. so the rise time tr of the pulse is important. Moreover.lbl. (28. and. rather than the signal-to-noise ratio alone. the time jitter can be much smaller than the rise time.4. 28. Properties of commonly used inorganic crystal scintillators can be found in Table 28.

or constant. Formulae are given which approximately describe average showers.-Y. These are used when ultimate performance is pursued. In some cases. Like the energy resolution. For a fixed number of radiation lengths. term b are detector non-uniformity and calibration uncertainty. While a is at a few percent level for a homogeneous calorimeter. and GSO:Ce (see Sec. detailed and reliable Monte Carlo simulation is possible. copper. The energy resolution σE /E of a calorimeter can be parametrized as a/ E ⊕ b ⊕ c/E. With effort. calorimeters built with heavy crystal scintillators.1. LSO:Ce. have attracted much attention for homogeneous calorimetry. EGS4 [128] and GEANT [129] have emerged as the √ standards. Typical photon angular resolution is about 45 mrad/ E. The passive medium is usually a material of high density. Electromagnetic calorimeters : Written August 2003 by R. The best energy resolution for electromagnetic shower measurement is e obtained in total absorption homogeneous calorimeters. where ⊕ represents addition in quadrature and E is in GeV. such as lead. an ionizing noble liquid. the stochastic term a for a sampling calorimeter is expected to be proportional to t/f . 28.131]. The stochastic term a represents statistics-related fluctuations such as intrinsic shower fluctuations. Zhu (California Inst. or depleted uranium. where a is a few to 20 mm and b can be as small as a fraction of mm for a dense calorimeter with fine granularity. or a Cherenkov radiator.131]. New heavy crystal scintillators. is radiation damage of the active medium. photoelectron statistics. Particle detectors The active medium may be a scintillator. The term c is due to electronic noise summed over readout channels within a few Moli`re radii. Electromagnetic calorimeters may also provide direction measurement for electrons and photons. operated in a high-beam intensity environment. since photons do not leave information in √ the particle tracking system. it can be factored as a/ E ⊕ b. non-compensation also contributes to the constant term. which can be provided by implementing longitudinal segmentation [132] for a sampling calorimeter or by adding a preshower detector [133] for a homogeneous calorimeter without longitudinal segmentation. a gas chamber.4). a semiconductor. iron. dead material at the front of the calorimeter. e. In the case of the hadronic cascades discussed below. Novel technologies have been developed for electromagnetic calorimetry. 27 of this Review). such as PWO. One additional contribution to the constant term for calorimeters built for modern high-energy physics experiments. This is important for photon-related physics when there are uncertainties in event origin. the constant term b can be reduced to below one percent. and sampling fluctuations. but since the physics of electromagnetic showers is well understood.10. The “spaghetti” structure has been developed for sampling calorimetry with scintillating fibers as the July 24. This can be minimized by developing radiation-hard active media [45] and by frequent in situ calibration and monitoring [44. of Technology). 2008 18:04 . 28. where t is plate thickness and f is sampling fraction [130. The development of electromagnetic showers is discussed in the section on “Passage of Particles Through Matter” (Sec. The main contributions to the systematic.54 28. it is typically 10% for sampling calorimeters. such as PWO. The position resolution depends on the effective Moli`re radius and the transverse e √ granularity of the calorimeter.g. it has received broad applications in high-energy and nuclear physics experiments.

U (DØ) 20.3/E √ Liquid Ar/Pb accordion 25X0 10%/ E ⊕ 0.5X0 16%/ E ⊕ 0.9 provides a brief description of typical electromagnetic calorimeters built recently for high-energy physics experiments. which is usually in good agreement with prototype test beam results as well as EGS or GEANT simulations.7% √ 2%/ E ⊕ 0.5X0 5%/ E √ Liquid Kr (NA48) 27X0 3. Detailed references on detector design and performance can be found in Appendix C of reference [131] and Proceedings of the International Conference series on Calorimetry in Particle Physics.6% spaghetti (KLOE) √ Liquid Ar/Pb (NA31) 27X0 7.7%/E1/4 √ 2%/ E ⊕ 0. Table 28.09/E √ Scintillator/depleted U 20–30X0 18%/ E (ZEUS) √ 13.2/E √ Lead glass (OPAL) 20. E is in GeV.5% ⊕ 0. provided that all systematic effects are properly included.7% for Eγ > 3. 2008 18:04 .7%/ E ⊕ 0.5% ⊕ 0.42% ⊕ 0.3/E (ATLAS) July 24.4% ⊕ 0.9: Resolution of typical electromagnetic calorimeters.3% ⊕ 0. Particle detectors 55 sensitive medium. The “accordion” structure has been developed for sampling calorimetry with ionizing noble liquid as the sensitive medium.3%/E 1/4 ⊕ 1. Also listed in this table are calorimeter depths in radiation lengths (X0 ) and the achieved energy resolution.4% 1.5 GeV √ PbWO4 (PWO) (CMS) 25X0 3%/ E ⊕ 0. the performance of calorimeters in situ is quoted.5%/ E Scintillator/Pb (CDF) 18X0 √ Scintillator fiber/Pb 15X0 5.2/% E ⊕ 0.28.45% Date 1983 1993 1996 1999 1998 1997 1990 1998 1988 1988 1995 1988 1993 1998 1993 1996 Technology (Exp. Table 28. Depth 20X0 27X0 16X0 Energy resolution 2. Whenever possible.) NaI(Tl) (Crystal Ball) CsI (KTeV) CsI(Tl) (BaBar) CsI(Tl) (BELLE) Bi4 Ge3 O12 (BGO) (L3) 22X0 16–18X0 2.5%/ E ⊕ 0.1/E √ Liquid Ar/Pb (SLD) 21X0 8%/ E √ Liquid Ar/Pb (H1) 20–30X0 12%/ E ⊕ 1% √ Liquid Ar/depl.

An example is the radiation-hard forward calorimeter in CMS. with zig-zag absorber plates to minimize channeling effects. practical geometry. Particle detectors 28. as in LAr calorimeters. Groom (LBNL). Most large hadron calorimeters are sampling calorimeters which are parts of complicated 4π detectors at colliding beam facilities. Such a calorimeter is sensitive to e± ’s in the EM showers plus a few relativistic pions. 28. or gaseous detectors. e. and the “accordion” LAr detector. the calorimeter is segmented in φ and θ (or η = − ln tan(θ/2)). tiles. and the transverse size of the cascades. a wedge of the ATLAS central barrel calorimeter. E. Typically. iron with plastic scintillator tile with wavelength-shifting fiber readout. 2008 18:04 .21: (a) ATLAS forward hadronic calorimeter structure (FCal2.131] Written April 2008 by D. including variations in geometry of the absorber and sensors. or via scintillation light observed by photodetectors (usually PMT’s). Cu.21(a). Fine segmentation.2. Hadronic calorimeters : [1–6. or occasionally U or W) alternating with plastic scintillators (plates. Pb.21(b). liquid argon (LAr).10.g. in this case e-h pairs are collected. bars). scintillating fibers threading an absorber [134]. An example. Silicon sensors are being studied for ILC detectors. 3). whose July 24. readout complexity. Waveshifting fibers are often used to solve difficult problems of geometry and light collection uniformity. In an inelastic hadronic collision a significant fraction fem of the energy is removed from further hadronic interaction by the production of secondary π 0 ’s and η’s. while desirable. 28. Tubes containing LAr are embedded in a mainly tungsten matrix.. The ionization is measured directly. with iron absorber and quartz fiber readout by PMT’s. is limited by cost. Ideally. is shown in Fig. A relatively new variant is the use of Cerenkov light in hadron calorimetry.56 28. (b) ATLAS central calorimeter wedge. Another departure from the traditional sandwich structure is the LAr-tube design shown in Fig. (a) W (Cu) absorber (b) PMT waveshifter fiber LAr filled tubes scintillator tile φ r Hadrons z Hadrons Figure 28. There are as many variants of these schemes as there are calorimeters. the basic structure is plates of absorber (Fe.

If h = e the hadronic response is not a linear function of energy. Particle detectors 57 decay photons generate high-energy electromagnetic (EM) cascades. but also interact with nuclei. nuclear recoils.87. p. producing spallation protons and neutrons. In a hadron-nucleus collision a large fraction of the incident energy is carried by a “leading particle” with the same quark content as the incident hadron. gradually thermalize and are captured. ) deposit energy via ionization and excitation. evaporation neutrons. this fact is usually ignored and h/e is reported. 2008 18:04 . the leading particle is usually a pion.28) up to at least a few hundred GeV. In contrast to EM showers. The recoiling nuclei generate little or no detectable signal. Until there is event-by-event knowledge of both the invisible energy loss and EM deposit (to be discussed below).29b) E0 ≈ 2.29b) . depending on the absorber and energy of the incident particle) is invisible. . and recoiling nuclei in highly excited states. Since 1 − m (1 − h/e)E0 is small and E0 ≈ 1 GeV for the usual pion-induced cascades. as do the showering γ-rays from the prompt de-excitation of highly excited nuclei. the energy resolution of a hadron calorimeter will remain significantly worse than that of an EM calorimeter. The exponent m depends logarithmically on the mean multiplicity and the mean fractional loss to π 0 production in a single interaction. with the production of more γ-rays—usually outside the acceptance gate of the electronics. but must be obtained experimentally for each calorimeter configuration. . Charged secondaries (π ± . . It is 1 GeV or a little less for incident pions. If the detection efficiency for the EM sector is e and that for the hadronic sector is h. This is not true for incident protons. a significant fraction of the hadronic energy (20%–35%.29a) (28. Between endothermic spallation losses.28. July 24. (28. It has been shown by a simple induction argument and verified by experiment that the decrease in the average value of the hadronic energy fraction ( fh = 1 − fem ) as the projectile energy E increases is fairly well described by the power law [135. It is in the range 0. The charged collision products produce detectable ionization. which can be neutral and hence contributes to the EM sector. Only the product 1−m can be obtained by measuring π/e as a function of energy. The lost energy and the π 0 → γγ fraction fem are highly variable from event to event. If the projectile is a charged pion. and late neutron capture.136] fh ≈ (E/E0 )m−1 (for E > E0 ) .137]. The neutrons lose kinetic energy in elastic collisions over hundreds of ns. E0 is roughly the energy for the onset of inelastic collisions.6 GeV [135.80–0. then the ratio of the mean response to a pion to that for an electron is π/e = fem + fh h/e = 1 − (1 − h/e) fh ≈ 1 − (1 − h/e)(E/E0 ) m−1 (28. (28. hadronic cascade processes are characterized by relatively few high-energy particles being produced. The result is an increased mean hadronic fraction for incident protons: in Eq. The EM energy deposit is usually detected more efficiently than the hadronic energy deposit.

the calorimeter response is non-Gaussian with a high-energy tail if h/e < 1. Lower-Z inactive cladding. √ σ/E = 0.. the compensation observed with the ZEUS 238 U/scintillator calorimeter was the result of the two mechanisms discussed above. where several variables can be chosen or tuned: 1. This is possible only in a sampling calorimeter. Eq.* and the absorber usually has higher Z than does the sensor. in particular contributing a larger fraction of the variance at high energies. Gabriel.75 . and so could be somewhat tuned. Since the fem distribution has a tail on the high side. 2. with the production of low-energy scattered protons in hydrogenous sampling materials such as butane-filled proportional counters or plastic scintillator. (When scattering off a nucleus with mass number A. but perforce fem → 1 as the projectile energy increases. (b) ZEUS. 2. Its variance changes only slowly with energy. Particle detectors The discussion above assumes an idealized calorimeter. σ/E = 0. i. with the same structure throughout and without leakage. An empirical power law σfem = (E/E1)1− (where < 1) describes the energy dependence adequately and has the right asymptotic properties. However.5 mm scintillator √ sheets. For h/e = 1. and Wigmans [140].2)). (c) a ZEUS prototype with 10 mm Pb plates and 2. for example. to design the calorimeter such that h/e = 1. is greater than this ratio in the sensor.3 mm 238 U plates. a neutron can at most lose 4/(1 + A)2 of its kinetic energy. Noncompensation (h/e = 1) thus seriously degrades resolution as well as producing a nonlinear response. By definition. Fabjan and Willis proposed that the additional signal generated in the aftermath of fission in 238 U absorber plates should compensate nuclear fluctuations [138]. σ/E = 0. relative to minimum ionization. It is clearly desirable to compensate the response. Increase the hadronic sensitivity. These included (a) HELIOS with 2. the EM energy deposit rate. Because the EM cross sections increase with Z. The approach seemed promising for awhile.44/ E. and (d) DØ.) The down side in the scintillator case is that the signal from a highly-ionizing proton stub can be reduced buy as much as 90% by recombination and quenching (Birk’s Law.58 28. preferentially absorbs low-energy γ’s in EM showers and thus also lowers the electronic response. such as the steel cladding on ZEUS U plates. G10 signal boards in the DØ calorimeters have the same effect. it is mostly carried by low-velocity fission fragments which produce very little observable signal.35/ E. July 24. But.34/ E was obtained. Complete containment is generally impractical. fluctuations in fem significantly contribute to the resolution. 2008 18:04 .e. Br¨ckmann. Decrease the EM sensitivity. The production of fission fragments due to fast n capture was later observed [139]. (28. u several groups built calorimeters which were very nearly compensating. “Real” calorimeters usually have an EM detector in front and a coarse “catcher” in the back. while a very large amount of energy is released. where the sandwich cell consists of a 4–6 mm thick * The asymptotic pair-production cross section scales roughly as Z 0. The degree of compensation was sensitive to the acceptance gate width.5 mm thick scintillator plates sandwiched √ between 2 mm thick 238 U plates (one of several structures). and |dE/dx| slowly decreases with increasing Z. The abundant neutrons have a large n-p scattering cross section. Motivated very much by the work of Brau. 0 ≤ fem ≤ 1.6 cm thick scintillator plates between 3.

and are easily solved for estimators of the corrected energy and fem for each event.32). (E/E1 )1− is the parameterization of σfem discussed above. a G-10 signal board.3 mm LAr. 2008 18:04 . 28.32) The coefficient a1 is expected to have mild energy dependence for a number of reasons.” Modern data show the slight downturn [134]. July 24. The average longitudinal distribution rises to a smooth peak about one nuclear interaction length (λI ) into the calorimeter.g. the falloff distance increases with energy. Both are subject to resolution effects. At a time when data were of lower quality. it has only recently been implemented by the DREAM collaboration [142]. The solution for the corrected energy is given by [136]: 1 − h/e|Q RS − Q . It can be found either from the fitted slope or by measuring π/e as a function of E. It then falls somewhat faster than exponentially. the sampling variance is (π/e)E rather than E. then called “the constant term. e. to fit profiles measured in the ATLAS central barrel wedges [144]. A slightly modified form has been used to fit recent measurements. In Fig.3 mm LAr gap. proportional to the the missing energy. in which the signal is sensed by two readout systems with highly contrasting h/e. A more versatile approach to compensation is provided by a dual-readout calorimeter. then for each event Eq. and another 2.30) These equations are linear in 1/E and fem . In either case. It is hoped that such a hadronic calorimeter can approach the resolution of an electromagnetic calorimeter.28. each filled with scintillator and quartz fibers. Particle detectors 59 plate. For example. The fractional resolution can be represented by a1 (E) σ h ⊕ 1− = √ E e E E E1 1− (28. is measured as well on an event-by-event basis. The DREAM collaboration expects to build a “triple-readout calorimeter” in which a neutron signal. The test beam calorimeter consisted of copper tubes.22 experimental results for 90% and 95% containment are shown. Although the concept is more than two decades old [141]. as are calculations using Bock’s parameterization [143]. where R = (28. 2. but effects due to fluctuations in fem are eliminated. (28.29) takes the form: Q = E[fem + h/e|Q (1 − fem )] S = E[fem + h/e|S (1 − fem )] 238 U (28. (28. If the two signals Q and S (quartz and scintillator) are both normalized to electrons. a plot of (σ/E)2 vs 1/E was apparently well-described by a straight line (constant a1 ) with a finite intercept—the square of the right term in Eq. Proton-induced cascades are somewhat shorter and broader than pioninduced cascades.31) E= R−1 1 − h/e|S R is the energy-independent slope of the event locus on a plot of Q vs S.

Particle detectors 200 99% 12 11 10 Depth in Iron (cm) Depth in Iron (λI) 150 95% 9 8 7 100 / Bock param. The transverse energy deposit is characterized by a central core dominated by EM cascades.3. / CDHS data / CCFR data 50 5 10 6 5 4 3 50 100 500 1000 Single Hadron Energy (GeV) Figure 28. The new measurements were motivated by the design of the ATLAS electromagnetic calorimeter and by inconsistencies in the previous literature.23. Velocities are temperatuer depedent and are systematically higher than those shown in Fig. 28. Recent precise measurements of the free electron drift velocity in LAr have been published by W.22: Required calorimeter thickness for 95% and 99% hadronic cascade containment in iron. The energy deposited in an annulus dA is adequately described by an exponential core and a Gaussian halo: dE/dA = (B1 /r) exp(−r/a1 ) + (B2 /r) exp(−r 2 /a2 ) [145]. There is a wide “skirt” produced by wide-angle hadronic interactions. Walkowiak [148]. on the basis of data from two large neutrino detectors and Bock’s parameterization [143]. 28. Free electron drift velocities in liquid ionization sensors : Velocities as a function of electric field strength are given in Refs.60 28.23. 28. 146–147 and are plotted in Fig. 2008 18:04 . where the neutral meson parents are themselves produced at a variety of angles. July 24.10.

in which the magnetic field is < 2 T. the central field is given by L .5%) 2. the dimensions are in meters. Particle detectors 61 10 9 8 Drift Velocity (µm/ns) TMS 7 6 5 4 3 2 1 0 0 2 LAr LAr+CH4 (0.D. 28.28.33) where n is the number of turns/meter and I is the current. The magnetic field (B) in an ideal solenoid with a flux return iron yoke. r the bending radius. E. revised October 2001 by R. of the trajectory is given by s = qB 2 /8p . q the charge. is given by B = µ0 n I (28. Solenoid Magnets : In all cases SI unit are assumed. In most cases. m the mass.4 TMP 4 6 8 10 12 −1) Field Strength (kV cm 14 16 Figure 28. Kephart (FNAL) 28. 0) = µ0 n I √ 2 + 4R2 L where L is the coil length and R is the coil radius. so that the magnetic field. (28. Yamamoto (KEK).1.11. momentum analysis is made by measuring the circular trajectory of the passing particles according to p = mvγ = q rB.34) B(0. Superconducting magnets for collider detectors Revised September 2005 by A. In a practical momentum measurement in colliding beam detectors. 2008 18:04 . (28.23: Electron drift velocity as a function of field strength for commonly used liquids. is in Tesla. it is more effective to increase the magnetic volume than the July 24.4. The sagitta. B. s. the stored energy. and µ0 = 4π × 10−7 . In an air-core solenoid. is in joules.35) where is the path length in the magnetic field.11. where p is the momentum.2.

5 20 30 12 40 25 130 109 11 120 27 5.5 1.84 0.6 7.62 28.2 1.75 1. (28.38) For a detector in which the calorimetry is outside the aperture of the solenoid.3 26 5 12.5 38 1080 2 × 250 2600 E/M [kJ/kg] 4.36) where corresponds to the solenoid coil radius R. (which is the case if it is to superconducting coil).5 4.5 1.25 4.5 5.4 5.7 0.55 2.75 3 1.6 3.8 1.5 2.75 0.0 1.8 1.4 2.37) B 2 dV E= 2µ0 If the coil thin.7 5.35 6 5.475 1. so small X/X0 is not a goal field strength.5 1.10: Progress of superconducting magnets for particle physics detectors. Particle detectors Table 28. Experiment TOPAZ* CDF VENUS* AMY* CLEO-II ALEPH* DELPHI* ZEUS* H1* BaBar D0 BELLE BES-III† ATLAS-CS† ATLAS-BT† ATLAS-ET† CMS† ∗ † ‡ Laboratory KEK Tsukuba/Fermi KEK KEK Cornell Saclay/CERN RAL/CERN INFN/DESY RAL/DESY INFN/SLAC Fermi KEK IHEP ATLAS/CERN ATLAS/CERN ATLAS/CERN CMS/CERN B [T] 1.85 5.29 1.52 ‡ 2.07 5.2 1.2 5.9 1.4 2. the coil must be thin in terms of radiation and absorption lengths.5 0.70 0.3 2. then E ≈ (B 2 /2µ0 )πR2 L .3 5.6 1.0 2.64 3 3. since dp/p ∝ p/B 2 .6 42 9.45 1.8 1. 2008 18:04 . (28.0 7.9 ‡ ‡ 0.5 2.66 (Toroid) (Toroid) ‡ 12 No longer in service Detector under construction EM calorimeter is inside solenoid.8 3. The energy stored in the magnetic field of any magnet is calculated by integrating B 2 over all space: 1 (28.75 3.5 1.46 2.8 ‡ 0.8 7.73 4 3.0 1.5 4.8 3.2 1.825–5. This usually means that the coil is superconducting and that the vacuum vessel encasing it is of July 24.0 1 1 4 Radius Length Energy X/X0 [m] [m] [MJ] 1.5 0.0 0.7 5.75 2.8 1.5 2.7–9.

so the thickness in radiation lengths (X0 ) is tcoil /X0 = (R/σh X0 )(B 2 /2µ0 ) . and σh is the hoop stress in the coil [151]. and it determines the average coil temperature rise after energy absorption in a quench: E/M = H(T2 ) − H(T1 ) ≈ H(T2 ) (28. Fig. The outer vacuum shell represents about 1/3 of the total thickness in radiation length. The ratio of stored energy to cold mass (E/M ) is a useful performance measure.10 [150]. σh . the aluminum stabilizer and support cylinders dominate the thickness. Since this shell is susceptible to buckling collapse. One would like the cold mass to be as small as possible to minimize the thickness.* the enthalpy of the coil. ∼80. Properties of collider detector magnets : The physical dimensions. its thickness is determined by the diameter.11.40) M ρ Vcoil 2ρ The E/M ratio in the coil is approximately equivalent to H. B 2 /2µ0 is the magnetic pressure. It can also be expressed as the ratio of the stress. The E/M ratios of various detector magnets are shown in Fig.2. 10. There are two major contributors to the thickness of a thin solenoid: 1) The conductor consisting of the current-carrying superconducting material (usually NbTi/Cu) and the quench protecting stabilizer (usually aluminum) are wound on the inside of a structural support cylinder (usually aluminum also). length and the modulus of the material of which it is fabricated.25 as a function of total stored energy. The coil package including the cryostat typically contributes about 2/3 of the total thickness in radiation lengths. An E/M ratio as large as 12 kJ/kg is designed into the CMS * The enthalpy. or heat content. In large detector solenoids.28. X0 the average radiation length of the coil/stabilizer material. but temperature rise during a quench must also be minimized. 28. the superconductor (NbTI/Cu) contributes a smaller fraction. The coil thickness scales as B 2 R. 28. (28.39) where tcoil is the physical thickness of the coil. and T1 is the initial temperature. July 24. respectively. 2008 18:04 . in the coil [151]: (B 2 /2µ0 )dV σ E ≈ h = (28. and 20 kJ/kg correspond to ∼65. Particle detectors 63 minimum real thickness and fabricated of a material with long radiation length. 28.24 shows thickness in radiation lengths as a function of B 2 R in various collider detector solenoids. is called H in the thermodynamics literature.41) where T2 is the average coil temperature after the full energy absorption in a quench. 2) Another contribution to the material comes from the outer cylindrical shell of the vacuum vessel. to twice the equivalent density. It is not to be confused with the magnetic field intensity B/µ. central field stored energy and thickness in radiation lengths normal to the beam line of the superconducting solenoids associated with the major collider are given in Table 28. and ∼100 K. E/M ratios of 5. ρ.

24: Magnet wall thickness in radiation length as a function of B 2 R for various detector solenoids. 28. Gray entries are for magnets not listed in Table 28. 2008 18:04 (28. The momentum resolution given by the deflection may be expressed as p p sin θ ∆p ∝ ≈ . so that thermal expansion effects in the coil are manageable. in principle. WASA DELPHI SSC-SDC prototype CLEO-II Thickness in radiation lengths ALEPH H1 0. and θ is the angle between the particle trajectory and the beam line axis .3. Particle detectors 3 2. R0 is the outer coil radius.11. On the other hand.5 1 TOPAZ D0 ZEUS PEP4. solenoid.” The SSC-SDC prototype provided important R&D for LHC magnets.10. R sin θ sin θ (28. This maximum local temperature should be <130 K (50 K + 80 K).64 28. Thus the coil temperature can be kept below 80 K if the energy extraction system work well.CDF TPC ATLAS-CS VENUS BESS. The limit is set by the maximum temperature that the coil design can tolerate during a quench. with the possibility that about half of the stored energy can go to an external dump resistor. The deflection (bending) power BL is BL ≈ R0 Ri B R Bi Ri dR = i i ln(R0 /Ri ) . Open circles are for magnets not designed to be “thin. p BL Bi Ri ln(R0 /Ri ) July 24. The particle momentum may be determined by measurements of the deflection angle combined with the sagitta. there is.5 2 1. no effect on the beam. the field profile generally has 1/r dependence.43) . Because no field exists at the collision point and along the beam line.5 0 CELLO 0 1 2 3 4 5 B 2 R [T2 m] 6 7 8 Figure 28.42) where Ri is the inner coil radius. Toroidal magnets : Toroidal coils uniquely provide a closed magnetic field without the necessity of an iron flux-return yoke.

10) uniformly spaced position measurements are made along a trajectory July 24. The momentum resolution is better in the forward/backward (smaller θ) direction. In practical designs. The mechanical structure needs to sustain the decentering force between adjacent coils. The radius of → − curvature and momentum component perpendicular to B are related by p cos λ = 0. (28.12. The geometry has been found to be optimal when R0 /Ri ≈ 3–4. 28. Measurement of particle momenta in a uniform magnetic field [152.45) curvature error curvature error due to finite measurement resolution curvature error due to multiple scattering. with radius of curvature R and pitch angle λ. The distribution of measurements of the curvature k ≡ 1/R is approximately Gaussian. Particle detectors 65 15 CMS 10 E/M [kJ/kg] SSC-SDC Prototype ATLAS-CS 5 ZEUS CDF BELLE BABAR VENUS TOPAZ D0 CLEO-II BES-III ALEPH H1 DELPHI 0 1 10 100 Stored Energy [MJ] 1000 104 Figure 28.153] The trajectory of a particle with momentum p (in GeV/c) and charge ze in a constant → − magnetic field B is a helix.28. 2008 18:04 . Open circles indicate magnets under construction.25: Ratio of stored energy to cold mass for thin detector solenoids.44) where B is in tesla and R is in meters. and the peak field in the coil is 3–5 times higher than the useful magnetic field for the momentum analysis [149].3 z B R . the coil is divided into 6–12 lumped coils in order to have reasonable acceptance and accessibility. The curvature error for a large number of uniformly spaced measurements on the trajectory of a charged particle in a uniform magnetic field can be approximated by (δk)2 = (δkres )2 + (δkms )2 . where δk = δkres = δkms = If many (≥ (28. This causes the coil design to be much more complex.

Nuclear Physics and Cosmology. 27 of this Review). More accurate approximations for multiple scattering may be found in the section on Passage of Particles Through Matter (Sec. Cambridge University Press (2000).-G. The Physics of Particle Detectors.47) sm = where V are covariances defined as Vsm sn = sm sn − sm sn with −2 w−1 (si m / i 2 ) and w = i . the X0 defined elsewhere must be multiplied by density) β = the kinematic variable v/c. perpendicular to the trajectory. Ferbel (ed. X0 (28. Leroy & P. Radiation Detection and Measurement. 4. Cambridge University Press (1998). The contribution to the curvature error is given approximately by δkms ≈ 8srms /L2 . δkres = L 2 (28.48) momentum (GeV/c) charge of incident particle in units of e the total track length radiation length of the scattering medium (in units of length. 5. Particle Detectors. Kleinknecht.46) where N = number of points measured along track L = the projected length of the track onto the bending plane = measurement error for each point. 2.66 28. (World Scientific. If a vertex constraint is applied at the origin of the track. D. C. # 5. Singapore. Nuclear Physics and Cosmology. where srms is defined plane plane there.R. # 12. 6. 3nd edition. G. Cambridge Monographs on Particle Physics. Cambridge University Press (1996). References: 1. John Wiley & Sons. Rancoita. 1987).016)(GeV/c)z Lpβ cos2 λ L . Knoll. CA. K. July 24. 2008 18:04 . New York (1999). Particle detectors 720 .) (Addison-Wesley. N +4 in a uniform medium. Green. T. 3. Grupen. Cambridge Monographs on Particle Physics. w Vss Vs2 s2 − (Vss2 )2 (28. 2004). Principles of Radiation Interaction in Matter and Detection. Experimental Techniques in High Energy Physics.F. the coefficient under the radical becomes 320. The contribution due to multiple Coulomb scattering is approximately δkms ≈ where p = z= L= X0 = (0. C. For arbitrary spacing of coordinates si measured along the projected trajectory and with variable measurement errors i the curvature error δkres is calculated from: (δkres )2 = Vss 4 . Detectors for Particle Radiation. Menlo Park.

Methods 117. B. Electron Tube Division. Nucl. Electron Devices 13. P. Methods A216. Diam. Matter. B. Arnold et al. Hamamatsu K. Achenbach. J. Nucl. Birks. Proc. 36. 1971). 122 (1986). 164 (1966). Instrum. M. & Rel. Instrum. P. Diam. Nucl. 513 (1998). D. 51. 255.. Sci. 10. Sadygov et al. 221 (1974). SPIE Press. 874 (1951). Petrov. Nucl. Methods A270. McIntyre. Sauli. R. 22. Berlman. P. 15. Applied Optics. R. Spieler. Kleinhans. Methods 117. R.. Appl.. July 24. & Rel. Handbook of Fluorescence Spectra of Aromatic Molecules. in Instrumentation in High Energy Physics. 25.. Methods 117. Instrum. et al. Instrum. Stapelbroek. Particle detectors 67 7. 19. Stat. Can be found under “Photomultiplier Tube Handbook” at sales. 1142 (1982). Golovin and V. 560 (2004). 14. T. Arnold et al. Sol. 32.B. D. Matter. 55 (1948). 11. et al. Instrum.B. Atac. 1 (2002). Methods A518. NS-29. Bengston and M. 80 (2000)..28. Nucl.. 2nd edition (2002). 70 (1998). Atac et al. Nucl.. 21.hamamatsu. Singapore) pp. Foerster. Sci. 295 (1974). Aston et al. C. 1964). The Theory and Practice of Scintillation Counting. Arisaka. Albert. R. Nucl. M. Appl. Sci. Moszynski. Instrum. 218–279. Phys. Phys. 465 (1992). “Active optical fibres in modern particle physics experiments. Nucl. Instrum. 56 (1994). Z. R.” arXiv:nucl-ex/0404008v1. Methods A314. J. 7. Ann. IEEE Trans. M. R. 301 (2003). Buzhan et al. 199. 113 (2003). 8. 3123 (1965).. 9. Mangeot et al. Petrov. Instrum. Methods A518. I. J. 16. Instrum. Lett.. Methods A283. Methods A504. F. A64. V. 289 (1988).. [Icarus Collaboration]. Phys. IEEE Trans. 12. Instrum. Nucl. 18. 2nd edition (Academic Press.. Birks. McKeag and R. IEEE. (1992.. 48 (2003). 36 163 (1989). ed. WA (2001). Landstrass et al. Nucl. 28. Nucl. Brascia et al. Nucl. Phys. Nucl. World Scientific. Methods A314. Braem et al. 2. Nucl. Saveliev. 79 (1983). Instrum. 20. (21).. Methods A409.php. 2. Haitz et al. J. Methods A504. Janesick Scientific charge-coupled devices.. R.. Dautet et al. Nucl. 406 (1987). 574 (2004). C. Aubert. 23. Instrum. et al. Nucl. Instrum. Jackman. H. H.B. (Pergamon. Nucl. 1033 (1993). Bialkowski. Methods A479. 582 (1989). Zorn. M. London. M. Photomultiplier Tubes: Basics and Applications. 227 (1974). ICARUS-TM/2001-09. Soc. 17. Instrum. M. A. Instrum.. Trans. Apsimon et al. E. Bellingham. J.. 27. Methods A442. K. LGNS-EXP 13/89 add 2-01. 33. New York.K. Perkin-Elmer Optoelectronics. Nucl.com/en/products/electron-tubedivision/detectors/photomultiplier-tubes-pmts. Clark. and W. 2008 18:04 13. Phys. [BaBar Collaboration]. 26. 24. Avalanche Photodiodes: A User’s Guide. 3894 (1993). IEEE Trans. .

Instrum. B. Instrum. Fakazawa. Nucl. Shiozawa. ibid. “Construction and Testing of the SLC Cherenkov Ring Imaging Detector. 31. Instrum. Ann.F. Nucl. Sci. 396 (1984). Instrum. Zhu. J. 395. Sept. Methods A371. Sci. Methods A433. 117 (2002). 45.68 28. Blasse. 38. 2008 18:04 . D. 23. Mao. J. 453 (1994). Nucl. S. 35. Nucl. et al.. L.L. Methods A433. July 24. Conference on Radiation-Tolerant Plastic Scintillators and Detectors.H. Chai. 52. Methods. Methods A502. 43. Nucl. R. Methods A257.Y. Rad. Nucl. Instrum. 32. Instrum. C. 718 (2007) and NS-54(4). 44. A. Methods A553.. M.. Chen. Gratta. L..W. Phys. K. 40.Y. 50.. USA. Appl. Fluornoy. 39. Particle detectors 29. 42. Methods A260. 1 (1996). D. Strand. Clough editors.R. Instrum. Sci. Proceedings of the Crystal 2000 International Workshop on Heavy Scintillators for Scientific and Industrial Applications. Instrum. IEEE Trans. 36.Y. Methods A538. Rad. Nucl. Instrum. Zhang and R. Litt and R. D. Nuc. 345 (2001). Ann. 1 (1999). Methods A479. K. Lett.” IEEE 37. 47. 304 (1993). (2000) 7360 (2000). Cavalli-Sforza. 55 (1987). Ann. France. presented in The 9th International Conference on Inorganic Scintillators and their Applications. Nucl. A. NS-31(1). N3:1132 (1990).. G. (1992). Zhang and R. Nucl.H. 54. Nucl. Mainusch. et al. Instrum. 53.. Nucl. and Part.. Rev.Y. See the RICH Workshop series: Nucl. 46. IEEE Nuclear Science Symposium Conference (2002). in Proc. 41 (1993). Clough editors. Nucl. Instrum. and Part. G. and Chem.M. Methods A314. Methods A292.H. Cooke et al. Nucl. Methods A312. Rowson. Methods A502. J. M. 41.Y. Nuc. Nucl. Holm. Johnson and R. Nucl. 240 (1999). 1 (1993).T. Instrum. H. 297 (1998). Newman. ibid. Instrum. et al. and R. Chamonix. Phys. Achenbach. P. Edition Frontieres. et al. R. Sci. Sci. C. Nucl. Rev. Winston-Salem. T.Y. Zhu. R. Nucl. 1 (1973).. Phys. Instrum. Blomker. et al. Methods A413. June (2007). 329 (1990). Hawkes. M Chou and B. Instrum. Ratcliff. 131 (2002). et al. 49. 451 (1992). et al. Appl. Schweitzer. 41 389 (1993).. Kimble. NS-54(3). Horstman and U. Nucl. et al. 281 (2005).L. et al. J. et al. Conference on Radiation-Tolerant Plastic Scintillators and Detectors. Methods A333. Nucl. Instrum. 37. Rev.C. Meunier. Sonderegger. Melcher and J. 211 (2003). Takagi and T. 505 (1992). 44. Bartlett. Instrum. 30. C. K. Nucl.. Zhu. Mao. Johnson and R. 88 . 34. Lempicki. D. 51. 43 (1983). 1 (2005). Instrum.F. I. [Super-Kamiokande Collaboration]. 1 (2003).M. Kuntner. will be published in Nucl. 42. Borenstein and R. Adam. 51. 33.. Zhu. 212 (1992). Abashian. 1319 (2007). Methods A311. Phys. Methods A343. IEEE Trans. P. Instrum.. 48. 523 (1987).M. Methods A493. and Chem. J.

154 (2006). 74. Oed. Sauli. Methods A62. Sulak. Virdee. G. Marx. W. Nygren & J. 76. Particle detectors 69 55. 83. Breskin et al.401(2004). Beringer et al. Lett. H. Phys. Methods A124.. Ahn. G. 1971). 61. The Mobility and Diffusion of Ions in Gases (Wiley. Instrum.S. Kasuga. 66. Appl. Nucl. 59. 79. 65. 403-404. G. 62.” in Experimental Techniques in High Energy Physics.. 67(1977).” CERN 84-08 (1984). Charpak et al. Scripta 23. Wisc. E. Instrum. Instrum. Methods A265. K. Instrum. Methods A92. L. 77. 84. F.lbl. 64. McDaniel & E. Methods A254. 455 (1979). 333 (1980).” arXiv: physics/0403055. 1 (2003). J. 2008 18:04 . et al. Anassontzis. Methods A446. S. I.. Rev. 373 (1971). 416 (1988). 58. 90. “Recent Results from the DELPHI Barrel Ring Imaging Cherenkov Counter. Nucl. CA. 262 (1968). 1993). Phys. http://www. 60. (1978). Nucl. Methods A263. Aleksa et al. Fischle et al.ch/writeup. 57.ch/rjd/cgi-bin/cross. 189 (1975). Instrum. Instrum. Nucl. Bouclier et al. Charpak et al.. 78 (1988). 75.. 435 (2000). Instrum.B. 10 (1978). Phys.H. J. 1973). Nucl. Rolandi. A. 1987). 285 (1984).com. B374. Lett. et al. Methods A167. 81. Methods A515. L. A. H.28.web. 63. Rev. 85. Biagi..cern. Blood. 041801 (2003). 474 (2005). et al. Instrum. 351 (1988). Instrum. G. 238 (1996). T. Nucl. HUEP-252 Presented at the Workshop on Proton Stability. Nucl. H.. Phys. Particle Detection with Drift Chambers (Springer-Verlag. 12. 70. Nucl.. Nucl. Titov. P. et al..gov/2007/AtomicNuclearProperties. Nucl. Phys. Methods A176. “Drift and Diffusion of Electrons in Gases. 69. N2:417 (1991)..N. et al.cern. Peisert & F. A.. M. Methods A562. FERMILAB-PUB-76-051-EXP. Nucl. 82. “Principles of Operation of Multiwire Proportional and Drift Chambers. Menlo Park. 2002). R..H. Va’vra. 72. J. http://pdg. Rep. D. Smirnov. Instrum.. Sauli. Atomic and Molecular Photo Absorption (Academic Press. http://consult. G. Mason. H. 56. A. 202 (1991). Madison. 78. 71. Methods A554. 234 (1999). S. Instrum. 34. 80. July 24. 354 (1981). Walenta et al. Shultz et al.G. “Radiation damage and long-term aging in gas detectors. 67. Berkowitz. Hirata. Phys. Charpak & F. Nucl. 68. Bichsel. Instrum. Atomic and Molecular Radiation Physics (Wiley. Nucl. Ferbel (ed. B205. Rev. Sci. Phys. E. 73.” IEEE 38. Lett.) (AddisonWesley. Nucl. 542 (1987). Blum & L.ansoft. Methods A421. Walenta. H. Sauli. M. http://rjd. M.R. Christophorou. Instrum.. Nucl. Methods A301.. Drumm et al. Today 31 Vol. Instrum. Ann.

106. Instrum. Nucl. Oxford (2005) ISBN 0-19-852784-5. P. Petrovykh. 77 (2004).L. M. Scholze et al. 117.. Nucl.. 297 (2002). Methods A326. 63 (1983). M¨ller et al.web.. Instrum. 91. Garibian et al. Methods A522. physics/0403055. Nucl.3055 (2004). 123. Instrum. M. 105. Sci NS-51/6. 114. Nucl. 526 (2006). ALICE. 208 (2000). Methods A522. 101. Fonte. .. 94. Nucl. Nucl. 405 (2006). Italy. Y. et al. 73 (2004). Instrum. Instrum. Akesson et al. Ambriola et al. AMS. 92. Instrum. Methods 93. 121. Spieler. Rev. R. Instrum. D12. 113. Methods A535. 269 (1971). 881 (2002). 4 (2003).. O. 28. F. 90.70 86. 341 (1999). Instrum. Sauli. H. Bamberger. 223 (2002). Instrum. Nucl. Instrum. 112. 63 (1998). 108. Campbell. et al. Case. Anulli et al. Instrum. Methods A560. Instrum. Colas et al.. 1 (1999).. H. Methods A509. Nucl. 440 (2007). 276 (2005). Aihara et al. IEEE Trans. Nucl.. “Innovative Detectors For Super-Colliders. Nucl. Trans. Instrum. Instrum. Nucl. Parkhomchuck. Nucl. et al.M. Methods A552. 107. 104. Derre. 13 (1997). 200 (1998). Yu. Lindstr¨m et al. Cerron Zeballos et al. Methods A376. http://gdd. Methods A490. IEEE Trans. 120. Methods A540.” Erice. Methods A558. Instrum. Instrum. G. Nucl.. et al. NS30. P. et al. D. 523 (2001). 295 (2005).. 102. Bagaturia.. Nucl. o C. IEEE. Kramberger et al. Instrum. Methods 125. M. et al. Proc. 99. Nucl. Methods A426. et al. A. PAMELA. Instrum. Ann. Methods A440. Da Via et al. Nucl. 122. Semiconductor Detector Systems. Instrum. Nucl. I. E. G. 87. M. Bassompierre et al. 115. V. 89. Nucl. M. 111. Instrum. V. 2008 18:04 97. Methods A578. 477 (2004). Ph. Methods 411. 86 (2003). 98. B. T. 118.J. Nucl. 110. Instrum. F. Nucl. Part. u R. 49. Instrum. 361 (2007). Brigida et al. 506 (2004). 109. Methods A481. Nucl. Sci. Nucl. o July 24. Nucl. M. 96. X. Instrum. Sharma. Andronic. Artru et al. of the 42nd Workshop of the INFN ELOISATRON Project. 95.. C. Campbell et al... Instrum. Oxford Univ. A. Methods A439. 434 (1993). Nucl.. Titov.. Phys. Methods A412. Y. Martoff et al. Instrum. Nucl. Instrum. 531 (1997). 88. Instrum.. Methods A187. 119.ch/GDD/. Giomataris. Pestov. Instrum. Ammosov et al. G. Instrum. V. Methods A459. Methods A394. Methods A522. 103. 93. Methods A465. Cardarelli. Nucl. Dolgoshein. Nucl. Nucl. F. Sauli and A. N. Sci.. 49. & N. ATLAS. 1289 (1975).L.cern. 9 (2004).. Methods A572. F. Sept.. M. Nucl. Giomataris. Press. Methods A374. Nucl. Instrum. 132 (1996)... Doetinchem et al... 28–Oct. 60 (2001). 119 (2007). 40 (2004). 124. 29 (1996). Sauli. Methods A573. Instrum.. Methods A386. Nucl. Instrum. Bellazzini. Particle detectors F. Abbrescia et al. 133 (1975). Methods A535. Rev. J. Cherry and G. G... V. 295 (2005). Nucl.. Nucl. Lindstr¨m et al. Santonico and R.. 100. 116. Methods A522. V. 355 (2000). G. 377 (1981). Krasel et al. Methods A540. Nucl. Nucl.

Hansl-Kozanecka.html. 134. Holmes-Siedle and L. Oxford 2002.” Proc. Brau and T. March 13–17. Methods A408. 91. CMS Collaboration. Br¨ckmann and H. Methods A259. J. N. 1989. 1997. Radeka. 51 (1974). Wigmans. Nucl. Donaldson and M. H. on Calorimetry in High Energy Physics. Nucl. 142. Instrum. 633–653 (2007). IEEE Trans. AL. Oxford (2000). River Edge. Nucl. 202 (1997). Cheu. Methods A572. V.E.E. Nelson. Bock. C. “The EGS4 Code System.S. R. Instrum. Nucl. Kowalski. Hitlin et al. Nucl. CERN/LHCC 96-41 (1996). July 1983. Bintinger. Sci. Wigmans. Akchurin. N. Nucl. p. Mockett. Methods 100. Willis and V. CERN DD/EE/84-1 (1987). Sci. Nucl. Note 86/026 DESY. GEANT3.” Proc. Goulding and D. 137. 129. Nucl. D. Nucl.. NS-29. 4 (1986). T. Instrum. W. Akchurin et al. NJ. ZEUS Int. 7th Inter. SLAC Report No. Handbook of Radiation Effects.A. 11th SLAC Summer Inst. E. 489 (1985). 2nd ed. p. 380 (1998). Hamburg (1986). J. Instrum. Radeka. Instrum. The CMS Electromagnetic Calorimeter Technical Design Report. Nucl. Groom. www. 143.B. Methods A399. F. Instrum. Instrum.slac. R. 499 (2000). N.P. Methods A265. R. Radeka.S. NS-21.stanford.. Tuscaloosa. J. Instrum. “Quartz Fibers and the Prospects for Hadron Calorimetry at the 1% Resolution Level. AZ. Instrum. Rogers. Phy. Fabjan et al. Leroy. 273 (1988). Y. Methods A537. A. Embry. R. Nucl. 1989). Methods 186. 336–347 (1994). Part. The ATLAS Liquid Argon Calorimeter Technical Design Report. NS-15/3. 107.A.H59 2001. Calorimetry: Energy Measurement in Particle Physics. 126. Nucl. vol. IEEE Trans.. Erratum to be published: see arXiv:physics/0605164v4. D. 533 (1981). T. Nov. eds. edited by R. Methods 120. 182.W. Methods 141. H. J. 493 (1972). Stanford Linear Accelerator Center (Dec. 127. 221 (1974). 133. Nucl. F. D. 455 (1968). for a more detailed discussion. Landis. Wigmans. et al. Wigmans (World Scientific. 1985).D. Clarendon. CERN/LHCC 97-33 (1997). Wigmans. Hirayama.A. Gilchriese (World Scientific. 135. 267 (July 1983).Tucscon. 131. V.W.edu/pubs/confproc/ssi83/ ssi83-008. Sci. Vinogradov. Teaneck. R. Instrum. Sirois. et al. R. Instrum.A. Instrum. 2008 18:04 . QC474.. Nucl..” SLAC-265. 537 (2005). and T. 225 (1976).. “A review of the physics and technology of high-energy calorimeter devices. and D. Instrum. Nucl. 138. Brun et al. Methods A455. 128.. Particle detectors 71 125. Nucl. July 24. 136. 140. 132.K. See also W. Instrum. and R. u R. Conf. 61 (1977). 389 (1987). Rutherford. 130.G. Methods A238. NJ. ATLAS Collaboration. Adams. 9–14. C.. Wigmans. Methods 137. in Proceedings of the Workshop on Calorimetry for the Supercollider. 1125 (1982). Shah. Akchurin. IEEE Trans.R. Nucl.O. 42. P. Methods A252. 141. Instrum. T. Gabriel. Gabriel et al.28. Kulchitsky and V. ISBN 0-19-850733-X. Goulding. 1998). Nucl. 139. International Series of Monographs on Physics. p. Methods A338..

Instr. Meth.” NSRDS-NBS-58 (1976). 23. A. Phys. 445 (2000). Methods A316. Nucl. 150. Methods A410. Methods 131. A.O. 146. 149. Nucl. Instrum. a July 24. Meth. Walkowiak. 284 (1998). Methods A449. Yamamoto. 184 (1997). A453. E. A494. Nucl.. Instr.L. Allen. Shibamura et al. 152.. 255 (2003). Nucl. 151. Methods A. R. 288 (2000). “Drift Mobilities and Conduction Band Energies of Excess Electrons in Dielectric Liquids. 147. Nucl. Nucl. T.. Particle detectors 144. Acosta et al. Scr. V. Nucl. 249 (1975). A. Simonyan et al. Taylor. 459 (1980). M. Karim¨ki. D. 145. Methods 24. Instrum. Instrum. 381 (1963). to be submitted.72 28. Instrum. Yamamoto. W. 2008 18:04 . Instrum. ATL-TILECAL-PUB-2007-008 and Nucl. 153. Instrum. Gluckstern. 148.