Transcript What`s wrong with drift chambers?

Introduction to silicon pixel
detectors
M. Garcia-Sciveres
October 2004
Oct. 2004
Intro to pixel detectors
1
Charged Particle Tracking
Negligible change to
energy and direction
detector
Charged particle
shooting through space
Record of path taken by particle
(:trajectory”). Not the same thing
as a “picture” of the particle.
Oct. 2004
Intro to pixel detectors
2
Flying back and forth to CERN
• Think of jet airplanes.
• Easy to see where they have been by streaks left in the sky
• An insignificant bit of the energy and mass of the airplane
(of the fuel) are used to produce the streak
• Atmospheric conditions affect sharpness, persistence, and
amplification (natural cloud formation) of the streak.
• The earliest charged particle tracking detectors were cloud
chambers
Oct. 2004
Intro to pixel detectors
3
Cloud and Bubble Chambers
•
•
•
A gas-liquid (cloud chamber) or liquid-gas (bubble chamber) is exploited to greatly
amplify microscopic perturbations due to a charged particle passing through gas or liquid.
Both require taking a picture of the cloud streak or bubble trail left behind by the particle.
See 5th floor hallway “art”.
The detection is a 3 step process:
–
–
–
•
Particle goes by, interacting with a few gas or liquid atoms- losing a small amount of energy in
each interaction
A phase change is triggered by the interaction, which eventually involves huge numbers of nearby atoms in a chain reaction (domino effect). Cloud streaks or bubbles form.
A picture is taken (this is the “raw” data that is recorded and must then be analyzed).
Bubble chambers are faster and more precise than cloud chambers, but both are painfully
slow by today’s standards. The gas or liquid must be brought to its critical point for every
event, it takes time for bubbles/clouds to form, and taking pictures is also slow.
(Gargamel bubble chamber on display at CERN)
Oct. 2004
Intro to pixel detectors
4
Drift Chambers: much faster
• Cut out the middleman. Forget bubbles, clouds, and
pictures.
• Wire chambers, time projection chambers.
• Measure the ionization left behind by a passing particle
(through a gas).
– Two step process: use avalanche in gas to amplify the primary
signal (obviously faster than bubbles + pictures)
Ions created by
passing charged
particle. Too small for
electronic detectionjust a few atoms
+
Wire at –ve
voltage
Oct. 2004
-
+
+
+
- -
-
Intro to pixel detectors
When ions get close to HV
wire they trigger an
avalanche- each ion leads
to ~10,000 new ions. Now
you have a charge you can
measure with an electronic
circuit.
Wire at +ve
voltage
5
Why Does This Work?
• (1) An avalanche is a natural phenomenon in gasses (think lightning)
• (2) Gasses consist 100% of neutral molecules- there is no natural
contamination of ions an any level.
– The ions needed to start an avalanche must be externally introduced (drift
chambers) or are created when the electric field is high enough to rip
atoms apart.
– Until ions are introduced the gas will happily ignore the electric field.
• What’s wrong with drift chambers?
– Drift: it takes time for ions to move towards the HV wires
– Rate limit: must wait for all ions to clear away before device is sensitive to
new particles.
– Resolution: The diffusion of ions in the gas and the ionization statistics
(randomness in location of primary ions) limit the ultimate resolution
(~100um).
Oct. 2004
Intro to pixel detectors
6
Silicon Detectors: still faster and more accurate
• Cut out the middleman again!
• Detect the primary ionization directly
• Silicon (a solid) is much denser than gases => more primary ions are
produced (this also means that the charged particle loses more energy
in order to be tracked).
– Just enough charge for direct measurement with electronic circuits.
• Silicon (a solid) has less diffusion than a gas => higher resolution
(~10um).
• The catch: solids in general are not 100% made or neutral atoms, free
of ions like gasses.
– Need a solid that one can prepare to be 100% ion-free
– Has to conduct electricity (so ions can flow to an electronic circuit)
Oct. 2004
Intro to pixel detectors
7
Silicon Detector Basics
•
•
Metals: Atoms arranged in lattice that shares valence electrons. Number of free
charge carriers ~1022/cc (~1,000 Coulomb of charge per cc). Impossible to
remove this free charge.
Pure silicon: atoms arranged in metal-like lattice, but number of “charge carriers”
is ~1010/cc (~1nC/cc).
– A modest electric field can remove this free charge.
•
Impure silicon can have varying number and kind (+ or -) of charge carriers,
depending on impurity type and concentration.
– N-type has –ve carriers
– P-type has +ve carriers
– Typical concentration in range 1012 – 1018/cc
•
•
After free carriers are removed by electric field, silicon looks electrically like a
gas- 100% neutral. This state is called “depleted”.
A passing charged particle creates 22,000 (avg.) charge carrier pairs per 300mm.
(~4fC).
– If my laptop ran on 4fC/s the battery would last ~1010 years (the age of the universe).
Oct. 2004
Intro to pixel detectors
8
Actual ATLAS Pixel Sensor
Bumps
connect to
implants
P-spray doping to isolate individual pixels
Heavily n-doped pixel implants (doping too heavy to deplete)
Diode
junction
Lightly n-doped bulk
Heavily p-doped back side contact
•
•
Guard rings
A diode junction forms wherever p-doped and n-doped regions touch.
Depletion always begins at the diode junction as reverse external voltage is
applied.
Hadron irradiation introduced p-type defects. Eventually this will cause the
bulk to “type invert” and become p-type. At this point the diode junction shifts
to the top. This was chosen on purpose because it allows to operate without
fully depleting the bulk.
Oct. 2004
Intro to pixel detectors
9
Basic Charge Amplifier
DC “current
source”
compensates for
detector leakage
With Leakage current Compensation
Active amplifier. Increases effective capacitance by
factor of “open loop gain” at the expense of adding
a rise-time (active takes time)
Gain = 1/C(fF) V/fC
Q=CV
Pixel chip input stage gain is ~300mV/fC
C = 3.5fF
Q1/C1 = Q2/C2
Pixel capacitance + parasitics ~ 100x inverse gain!
Oct. 2004
Intro to pixel detectors
10
Pixel Chip Front End
Input from
Pixel sensor
(bump goes here)
Oct. 2004
preamp
Intro to pixel detectors
comparator
11
Front End Output
Threshold (adjustable up and down).
Comparator output is on when red line is above threshold, off if below.
Oct. 2004
Intro to pixel detectors
12
Front End Features
• Programmable threshold = Global Threshold + Pixel Threshold
Can easily change threshold
for whole chip
• Calibration charge injection
Input from
detector
V1
V2
Can fine tune each pixel to
compensate for response
differences (Tuning)
switch
Injection capacitor
(must be small)
Good old charge
amplifier
• Ability to measure leakage current
• Time over Threshold (TOT) charge measurement
– How long the red curve says above threshold depends on the size of the
input charge
Oct. 2004
Intro to pixel detectors
13
CMOS integrated circuits
• CMOS = combination metal oxide silicon
• Combination = p-type and n-type implants on the same wafer.
+ ion implantation
source
drain
source
gate
Gate oxide
p-implant
drain
gate
p-implant
n-implant
photoresist
n-implant
silicon
Oct. 2004
Intro to pixel detectors
14
The MOS transistor
G
• Transistor
S
D
• Diode
S
D
• ~Resistor
S
D
• Capacitor
Oct. 2004
Intro to pixel detectors
15
Complexity in Numbers
• Complex circuits are possible using a large number of
MOS transistors (~3M in FE-I3).
• In practice there are some special circuit elements also
used in small numbers, such as metal-insulator-metal
(MIM) capacitors and polysilicon resistors.
Oct. 2004
Intro to pixel detectors
16
The MOS transistor schematic of the FE-I3
charge amplifier
Oct. 2004
Intro to pixel detectors
17
Pixel Chip Readout Architecture
Data driven
Time stamp
Trigger
Oct. 2004
Trigger driven
Intro to pixel detectors
18