Particle Detectors

Geneva

Switzerland CERN

France

an Introduction

The LHC Accelerator

Particle Detectors Michael Hauschild (CERN), page 1

 Historic Detector: Bubble Chamber
BEBC bubble chamber

Invented 1952 by Donald Glaser (Noble Prize 1960)

chamber with liquid (e.g. H2) at boiling point (“superheated”)
charged particles leave trails of ions
formation of small gas bubbles around ions
take photos of interactions and look for interesting events…
nice pictures, but SLOW!!!
only 1 photo/event every few seconds, need something faster à electronic detectors

tracks in CERN 2m bubble chamber

δ-electrons
(knock-on electrons
from atomic shell)

beam

Particle Detectors Michael Hauschild (CERN), page 2

 LHC Detectors
General purpose detectors
(good for everything...)

ATLAS
CMS

ALICE LHCb

dedicated for dedicated for

Heavy Ion collisions b-physics
Particle Detectors Michael Hauschild (CERN), page 3
 Particle Physics Methods

Einstein
(1905):

Matter is
concentrated energy! accelerated particles

detector to
Matter can be transformed measure
into energy and back! outgoing particles

E = m c2

We use this at a particle accelerator

protons are accelerated ⇒ energy
kinetic energy is transformed into matter at the collision
new particles are being produced (new matter)

Particle Detectors Michael Hauschild (CERN), page 4

 The Perfect Detector...
...should reconstruct any interaction of any type with 100%
efficiency and unlimited resolution
get “4-momenta” of basic physics interaction
= energy / momentum / mass + charge of ALL involved particles
in practice: limitations by detector inefficiency (not all particles detected) + detector
resolution (measurements have statistical + systematic uncertainties)

Fixed target geometry Collider geometry

“Magnetic spectrometer” “4π multi purpose detector”

interaction tracking muon filter

LHCb
point ATLAS
N CMS

S
ALICE
beam magnet calorimeter (both geometries) barrel
(dipole) endcap endcap

• Limited solid angle dΩ coverage • “full” dΩ coverage

• rel. easy access (cables, maintenance) • very restricted access
Particle Detectors Michael Hauschild (CERN), page 5
 High Energy Collider Detectors
Tracking Detector (or Tracker) = momentum measurement
closest to interaction point: vertex detector (often silicon pixels)
measures primary interaction vertex and secondary vertices from decay particles
main or central tracking detector
measures momentum by curvature in magnetic field + charge of particle

Calorimeters = energy measurement

electro-magnetic calorimeters (light particles: e-, e+, g)
measures energy of light EM particles (electrons, positrons, photons) based on electro-
magnetic showers by bremsstrahlung and pair production
two concepts: homogeneous (e.g. CMS) or sampling (e.g. ATLAS)
hadron calorimeters (heavy hadronic particles: p, K, p, n)
measures energy of heavy (hadronic) particles (pions, kaons, protons, neutrons) based
on nuclear showers created by nuclear interactions

Muon Detectors = momentum measurement for muons (more precise)

outermost detector layer, basically a tracking detector

Particle Detectors Michael Hauschild (CERN), page 6

 A typical Particle Detector
Cut-away view of ATLAS: several layers of detectors

Muon Detector
à muon ident.
+ muon mom. p
+ charge

Calorimeter
à energy E

Coil

Tracker
à momentum p
+ charge
Particle Detectors Michael Hauschild (CERN), page 7
 Detector Challenges at LHC
High energy collisions
sufficiently high momentum resolution up to TeV scale
High luminosity (high interaction rate)
high rate capabilities, fast detectors (25 ns bunch crossing rate)
High particle density
high granularity, sufficiently small detector elements to resolve particles
High radiation (lots of strongly interacting particles)
radiation mainly due to particles emerging from collisions, not machine background
radiation-hard detectors and electronics (have to survive ~10 years)

LARGE collaborations!!!
~O(3000) physicists for ATLAS and CMS each
communication, sociological aspects
exponential raise of meetings, phone + video conferences...
Particle Detectors Michael Hauschild (CERN), page 8
 38 countries, 177 institutes,
3000 researchers (incl. 1000 PhD students)
Particle Detectors Michael Hauschild (CERN), page 9
 LARGE Detectors
Everything is LARGE at the LHC…

ATLAS

Building 40:
ATLAS + CMS
“headquarters”
CMS

Particle Detectors Michael Hauschild (CERN), page 10

 Radiation Doses at LHC
~ 2 x 106 Gray / rT2 / year at LHC design luminosity
where rT [cm] = transverse distance to the beam

Lots of R&D over >10 years to develop rad.-hard silicon

detectors, gaseous detectors and electronics

ATLAS
neutron fluences:

> 1015 n/cm2/yr

close to beam line

interaction point
Particle Detectors Michael Hauschild (CERN), page 11
 Challenging Conditions: Pile-up
2012 event with pile-up: 25 reconstructed primary vertices
~7 cm

Particle Detectors Michael Hauschild (CERN), page 12

 Tracking Detectors
How to measure momentum and charge?

Particle Detectors Michael Hauschild (CERN), page 13

 Tracker Technologies
3 major technologies of tracking detectors
Gaseous detectors
ionization in gas
typically ~100 e–/cm à not sufficient to create significant signal height above noise for
standard amplifiers
typical amplifier noise = some 100…1000 ENC (equivalent noise charge, in electrons)
requires gas amplification ~104 to get enough signal over noise (S/N)
Silicon detectors (solid state detectors)
creation of electron – hole pairs in solid state material
typically ~100 e– - hole pairs/µm = 104 more than in gaseous detectors
300 µm thick detector creates high enough signal w/o gas amplification
~30’000 charge carriers per detector layer, noise ~1000 ENC, S/N ~ 30:1

rarely used: fiber trackers

scintillating fibers
scintillation light detected with photon detectors (sensitive to single electrons)
Particle Detectors Michael Hauschild (CERN), page 14
 Tracking Detector Principles
Typical: several layers of sensitive detector elements
each layer gives a 2D hit coordinate (+ detector position à 3D)
Magnetic field bends (charged) particle trajectories
detector layers

⊗
detector element, size = d
d
can reconstruct + fit track (radius)
charged with at least 3 hits
particle hits

resolution of each hit depends on size d of detector elements

à some uncertainly where the particle passed the detector element d
à probability distribution is “flat” Probability
σx
à take the width of an equivalent Gaussian distribution as resolution

e.g. for d = 30 µm → ~10 µm resolution

d/2

Particle Detectors Michael Hauschild (CERN), page 15

 Magnet Concepts at LHC experiments
solenoid (air-core) toroid

+ large homogenous field inside coil + can cover large volume

- needs iron return yoke (magnetic shortcut) + air core, no iron, less material
- limited size (cost) - needs extra small solenoid for general tracking
- coil thickness (radiation lengths) - non-uniform field
- complex structure
CMS, ALICE ATLAS

µ
µ

Particle Detectors Michael Hauschild (CERN), page 16

 ATLAS and CMS Coils

CMS solenoid
(5 segments)

ATLAS barrel toroid coils

autumn 2005

Particle Detectors Michael Hauschild (CERN), page 17

 Momentum Measurement
(Only) charged particles are deflected by magnetic fields
homogeneous B-field à particle follows a circle with radius r

measurement of pt by
measuring the radius

this is just the momentum component perpendicular to the B-field

= transverse momentum pt
no particle deflection parallel to magnetic field
if particle has longitudinal momentum component
à particle follows a helix
total momentum p to be
ptrans p measured by dip angle l

l
B
l
plong
Particle Detectors Michael Hauschild (CERN), page 18
 Basic Gaseous Detector –
Geiger-Müller Tube
Geiger-Müller tube by Hans Geiger and Walther Müller 1928
tube filled with inert gas (He, Ne, Ar) + organic vapour (e.g. CH4) or CO2
central thin wire (20 – 50 µm Æ) , high voltage (several 100…1000 Volts)
between wire and tube
strong increase of E-field close to
the wire
- electron gains more and more energy
above some threshold (>10 kV/cm)
+ electron energy high enough to ionize
other gas molecules
newly created electrons also start ionizing
avalance effect: exponential
increase of electrons (and ions)
measurable signal on wire
primary electron
starting to ionize organic substances or CO2 responsible
for “quenching” (stopping) the discharge
anode wire ~100µm absorption of UV photons

Particle Detectors Michael Hauschild (CERN), page 19

 Solid State Detectors
Basic element of a solid state (silicon) detector is... a diode
p-type and n-type doped silicon material is put together
hole electron
Current flow through current current
diode if connected like this
more holes more electrons

for use as particle detector diode needs P N

to be connected in opposite way
depleted
region around junction of p- and n-type
material depletion region is created
zone free of charge carriers
P N
no holes, no electrons
thickness of depletion region depends on
voltage, doping concentration
charged particle can create new
electron/hole pairs in depletion typically 20'000 – 30'000 electron/hole pairs
area sufficient to create a signal in 300 µm thick material

Particle Detectors Michael Hauschild (CERN), page 20

 Silicon Strip Detectors
Now take a large Si crystal, e.g. 10 x 10 cm2, 300 µm thick
make bottom layer p-type
+ and subdivide the top n-type layer into
many strips with small spacing

many diodes next to each other

- with position information
(strip number)

Advantage compared to
wire/gas detectors
strip density (pitch) can be
rather high (e.g. ~20 µm)
high single point accuracy: σx = 20 μm / √12 = 5.8 μm
but also many electronics channels

Particle Detectors Michael Hauschild (CERN), page 21

 Si-Detector Electronics and Si-Pixels
Silicon strip detectors have a laaaarge number of electronics
channels, ~107 each for ATLAS and CMS Si trackers
requires highly integrated chips for amplification, shaping,
zero suppression (only information of strips with signals is read-out)
and multiplexing (put all strip signals on a few cables only)
electronics is directly connected to the sensor (the “multi-diode”) via
wire bonds
Si-strip detectors provide only 1 coordinate,
Pixel detectors are 2D detectors

wire bonds

Sensor

Pixel detector need

“bump” bonding
and have even more
channels, ~108 - 109
Particle Detectors Michael Hauschild (CERN), page 22
 ATLAS Inner Tracker
4-layer Si Pixel
3 à 4 layers in
2014
4-layer Silicon
Strips
Transition
Radiation
Tracker
(gaseous)

Pixel alignment with

cosmic rays 2008

Particle Detectors Michael Hauschild (CERN), page 23

 The ATLAS Pixel Detector
Re-insertion in December 2013 during Long Shutdown 1

Particle Detectors Michael Hauschild (CERN), page 24

 CMS Full Silicon Tracker
4-layers Si Pixel
3 à 4 layers in 2017
10-layers Silicon
Strips
210 m2, largest
silicon detector
ever built

Tracker Inner Barrel TIB

Particle Detectors Michael Hauschild (CERN), page 25

 Material Budget
Tracking Detectors should be light-weighted and thin
multiple scattering by material degrades resolution at low momenta
unwanted photon conversions in front of calorimeters
material often very inhomogeneous (in particular Si detectors)
Power & cooling adds most of the material
not the Si sensor material
ATLAS CMS

Particle Detectors Michael Hauschild (CERN), page 26

 Calorimeters
How to measure energy?

Particle Detectors Michael Hauschild (CERN), page 27

 Particle Showers in Calorimeters
Initial particle creates electro-magnetic shower of secondary
particles (electrons, positrons, photons) in dense material
by bremsstrahlung and pair production
shower depth depends on radiation length X0
(characteristic material constant)
for e± = length after all but 1/e of energy lost by Bremsstrahlung
for γ = 7/9 of mean free path length for pair production

shower maximum

80 GeV electron in
lead glass block

energy profile

Particle Detectors Michael Hauschild (CERN), page 28

 Energy Resolution of Calorimeters
Number of particles in shower is proportional to energy of
initial particle ∝
Critical Energy (typically ~10 MeV)

error of energy measurement determined by (statistical) fluctuations in

the number of shower particles ∝

1
resulting relative energy measurement error is ∝

More contributions from detector inhomogeneities and

electronic noise convolution

relationship valid for all types of calorimeters

∝ ⊕ ⊕ (homogeneous + sampling,
electromagnetic and hadronic)

stochastic constant term noise term

(statistic) term electronics noise
inhomogenities
number of shower particles non-linearities
Particle Detectors Michael Hauschild (CERN), page 29
 Calorimeter Concepts
Homogeneous calorimeters (e.g. CMS)
absorber material (generation of the shower) = detector material
typically an electromagnetic shower is created in an optical transparent absorber,
photons created in the shower are collected and detected with some photo detector

photo detector:
absorber + detection photo multiplier,
photo diode etc.

20 - 25 Radiation Lengths (X0)

Sampling calorimeters (e.g. ATLAS)

passive (heavy) absorber material (iron, copper, lead, tungsten, uranium)
interleaved with active detector material
absorber

detector:
scintillators,
gaseous/liquid Argon etc.

Particle Detectors Michael Hauschild (CERN), page 30

 CMS: Homogeneous EM Calorimeter
Clear advantage: good energy resolution
the entire shower is kept in active detector material
no shower particle is lost in passive absorber

Disadvantages
limited granularity, no information on shower shape in longitudinal
direction (along particle flight direction)
position information is useful to resolve near-by energy clusters,
e.g. single photons versus two photons from π0 decay
Two overlapping CMS PbWO4 crystal
Single photon photons from p0 decay

Dj x Dq = 1º x 1º Dj x Dq = 1º x 1º

ρ = 8.28 g/cm3 X0 = 0.89 cm

dense, transparent materials needed with
short radiation length and high light yield
Particle Detectors Michael Hauschild (CERN), page 31
 ATLAS: Sampling EM Calorimeter
Typical sampling calorimeters use iron or lead absorber
material, variety of detectors in between possible
gas detectors (MWPCs), plastic scintillators, liquid noble gases (LAr, LKr)
LAr with “acordeon” shaped Fe-Pb-Fe absorbers at ATLAS
LAr is ionized by charged shower particles ATLAS LAr calorimeter

Charge collected on pads

ionization chamber, no “gas” amplification
pads can be formed as needed à high granularity

acordeon structure
helps to avoid dead
zones (cables etc.)

simulated shower

Particle Detectors Michael Hauschild (CERN), page 32

 ATLAS/CMS Hadron Calorimeters
Measure energy of heavy hadronic particles: p, K, p, n, …
Energy resolution much worse than for el.-magn. calorimeters
shower created by nuclear interactions (hadronic shower, fewer particles in shower)
usually only a few nuclear interaction lengths deep (5 – 6 λl)

Both ATLAS and CMS use scintillators as detector material

need many optical fibers to transport light from scintillators to photo
detectors ATLAS CMS

Particle Detectors Michael Hauschild (CERN), page 33

 ATLAS Muon Detector
Muon detectors are tracking detectors (e.g. wire chambers)
they form the outer shell of the (LHC) detectors
they are not only sensitive to muons (but to all charged particles)!
just by “definition”: if a particle has reached the muon detector
à it's considered to be a muon
all other particles should have been absorbed in the calorimeters
ATLAS Muon Detector Elements
Challenge for muon detectors
large surface to cover (outer shell)
keep mechanical positioning
stable over time
Aluminum tubes with central
wire filled with 3 bar gas

ATLAS Muon System

1200 chambers with 5500 m2
needs also good knowledge of
(inhomogeneous) magnetic field
Particle Detectors Michael Hauschild (CERN), page 34
 ATLAS Detector Status
(a 100 megapixel camera with 40 MHz framerate = 1 PB/second)

Particle Detectors Michael Hauschild (CERN), page 35

 How to Select Interesting Events?
Bunch crossing rate: 40 MHz, ~20 interactions per BX (109 evts/s)
can only record ~1000 event/s (1 MB each), still ~1 GB/s data rate
Need highly efficient and highly selective TRIGGER
raw event data (1 PB/s) are stored in pipeline until trigger decision

trigger
YES
Detector PIPELINE save
NO
109 evts/s (1 PB/s) trash 103 evts/s (1 GB/s)

ATLAS + CMS triggers have 2 levels

Level-1: hardware (FPGAs), ~3 µs decision time, 40 MHz à 100 kHz
Level-2: software (computer farm), ~4 s decision time, 100 kHz à 1 kHz

Particle Detectors Michael Hauschild (CERN), page 36

 From Physics to Raw Data

Collision Point

Actually recorded are raw data with ~1 GB/s for ATLAS/CMS

mainly electronics numbers
e.g. number of a detector element where the ADC (Analog-to-Digital converter) saw a
signal with x counts...

Particle Detectors Michael Hauschild (CERN), page 37

 From Raw Data To Physics

We need to go from raw data back to physics

reconstruction + analysis of the event(s)
Particle Detectors Michael Hauschild (CERN), page 38
 Simulation (Monte Carlo)
Even with best calibration + alignment
some detector influence, e.g. efficiency for track reconstruction etc. will
not be known well enough from data
Use detector simulation (Monte Carlo) to “unfold” detector influence

-1

=
convolution

measured data detector physics

Event “generator” simulates physics processes

Full detector description
geometry, detector volumes, detector response (noise etc.)
physics interactions with matter and tracking particles through detector volumes
also needed for detector design studies before detector actually built

Particle Detectors Michael Hauschild (CERN), page 39

 ATLAS (A Toroidal LHC ApparatuS)
~7'000 t
main assembly in cavern

25 m

46 m
Particle Detectors Michael Hauschild (CERN), page 40
 CMS (Compact Muon Spectrometer)
main assembly on surface,
then lowering into cavern in
210 m2 silicon detectors 5 big parts by ~2500 t crane

15 m
4T

~12'500 t

22 m
Particle Detectors Michael Hauschild (CERN), page 41
 ATLAS/CMS Concept Overview
The two large LHC detectors have somewhat
different concepts
ATLAS
small inner tracker with moderate field (small 2 T solenoid)
electron identification by transition radiation tracker
sampling calorimeter with high granularity outside solenoid
air-core toroid system for good muon momentum measurement
emphasis on granular calorimeter and good muon measurement

CMS
large inner tracker with high B-field (large 4 T solenoid)
no dedicated particle identification detector
homogeneous crystal calorimeter with good energy resolution inside solenoid
emphasis on good general tracking and good energy resolution

However, both detector concepts have very similar

performance for Higgs physics (efficiency, mass resolution…)
Particle Detectors Michael Hauschild (CERN), page 42
 First Digging started in 1998
Gallo-roman remains
on future CMS site

Roman
coins

ATLAS cavern
September 2000

Particle Detectors Michael Hauschild (CERN), page 43

 The ATLAS Site 2005
CERN Main Entrance B LHC Cooling Towers

Globe of Innovation & Science ATLAS Main Hall

ATLAS Control Room
and Visitor Centre
Particle Detectors Michael Hauschild (CERN), page 44
 ATLAS Underground Cavern
huge cavern + surface buildings,
2 access shafts 18m + 12m Ø,
2 small shafts for elevators + stairs

Length = 55 m
Width = 32 m
Height = 35 m

Particle Detectors Michael Hauschild (CERN), page 45

 Start of ATLAS Detector Construction

Transport and lowering of first

superconducting Barrel Toroid coil

Particle Detectors Michael Hauschild (CERN), page 46

 CMS Lowering of 2000 t Central Part

Particle Detectors Michael Hauschild (CERN), page 47

 ATLAS Barrel Toroid Complete (Nov 2005)

Particle Detectors Michael Hauschild (CERN), page 48

 Detector Technology and Arts
Stage Design of Opera “Les Troyens” in Valencia, October 2009

Particle Detectors Michael Hauschild (CERN), page 49

 The first Higgs at LHC (4 April 2008)

Particle Detectors Michael Hauschild (CERN), page 50

 First LHC Collisions at High Energy
2 × 3.5 TeV, 30 March 2010 ATLAS Control Room
not (yet) super spectacular (but these are the first events…)

ALICE

ATLAS

LHCb

CMS
Particle Detectors Michael Hauschild (CERN), page 51