Transcript IVD CLASP 010 - Science and Technology Facilities Council

Introduction to Silicon Detectors
G.Villani
STFC Rutherford Appleton Laboratory
Particle Physics Department
Outlook
• Introduction to physics of Si and detection
–
Si electronic properties, transport mechanisms, detection
• Examples of detectors
–
Strips, CMOS,CCD,MOS
• Radiation damage
• Conclusions
2
Introduction
The Si detection chain
E
Sensing/
Charge creation
Charge transport
and collection
Si physical properties
Conversion
Si device properties
Signal
processing
Data TX
Si device topologies properties
almost all the boxes of the detection chain process based upon Silicon
3
Silicon properties
After Oxygen, Silicon is the 2nd most abundant element
in Earth’s crust (>25% in mass)
The crystalline structure is diamond cubic (FCC), 8 atoms/cell with lattice
spacing of 5.43 A ~ 5x1022 cm-3
* In electronic industry all crystallographic forms are used (Single crystal,
Polysilicon, α-Si)
Si
The key to success of Si is related to its abundance and oxide SiO 2, an
excellent insulator (BV ~ 107 V/cm).
1.48A
4
* Micro crystals but the flexible bond angles make SiO2 effectively an
amorphous: its conductivity varies considerably (charge transport in SiO2
via polaron hopping between non-bonding oxygen 2p orbitals)
Silicon electrical properties
Silicon Band structure
The electronic band structure can be obtained within the independent electron approximation (normally
1 electron SE in periodic potential neglecting electron interactions) in terms of Bloch functions
T  U   E  n,k r   un,k r e jkr  En k 
~ a wave associated with free motion of electrons modulated by the periodic solution u n,k. The energy
E is periodic in k so is specified just within the 1 st unit WS cell of the reciprocal lattice (the Brillouin
zone).
* The appearance of Band Gap, separating CB
VB
and VB
* The 6 CB minima are not located at the center
of 1st Brillouin zone, INDIRECT GAP
CB
1st Brillouin zone of Diamond lattice
5
CB
VB-H
VB-L
Silicon electrical properties
The detailed band structure is complicated: usually quasi-equilibrium simplifications
are sufficient to study the charge transport.
Assuming that the carriers reside near an extremum, the dispersion relationship
E(k) is almost parabolic:
E k  
2 k
2
2m0*
2m 
 g E  
* 3/ 2
0
2 3
2 
E  Eo
(3D)
2
2 k
1
k  p
E k  
 v   k E k   *  *
*
2m0

m0 m0

dk t 
d  p

   rV  F
dt
dt
* Under the assumptions of small variation of the electric field, the carrier dynamics
resembles that one of a free particle, with appropriate simplifications.
* The effective mass approximation takes into account the periodic potential of the
crystal by introducing an effective carrier mass ( averaged over different longitudinal
and transverse masses). The lower the mass, the higher mobility (µ  1/m*)
* Similar approach used to calculate the E(k) for phonons.
6
Silicon electrical properties
The carrier density is calculated from:
• The density of states g(E), which depends on dimension;
• The distribution function F(E);
CB
Only partly filled bands can contribute to conduction: carrier
density in CB and VB.
At equilibrium the carrier density is obtained by integrating
the product:
VB
nD   g D E F E dE  NC /V e  Ec Ev / kT  ni  pi
3
F E  
2
2
1
 E  EF 
1  exp 

 kT 
pn  ni  N C N V e
1
 
 E g / kT
10
20
@ T  300K
Fermi level: energy level @ 50% occupancy
0
In intrinsic Si a creation of e in CB leaves behind a hole in VB,
that can be treated as an e with positive charge and mobility of
the band where it resides
The density of states gD(E) depends on the dimension
7
Silicon electrical properties
Conduction of Si intrinsic @ T = 300K:
σ = q(μn +μp) ni = 3.04x10-6mho-cm ->329kOhm-cm
By adding atoms of dopants, which require little
energy to ionize( ~10’s mEV, so thermal energies @
ambient temp is enough) we can change by many
odg the carrier concentration.
Doping concentration: 1012 to 1018 cm-3
In crystalline Si ~ 5*1022atomscm-3
In equilibrium and for non degenerate case
the relationship between carrier concentration and E
is the same as in the intrinsic case:
pn  ni2  N C NV e
 
 E g / kT
 1020 @ T  300K
e.g. : N D  1017  pn  pN D  p 
8
ni2
 103
ND
Charge transport
Charge transport:
The charge transport description in semiconductors relies on semi-classical BTE
(continuity equation in 6D phase space)



f r , k , t 1
F
f r , k , t
  k E k   r f r , k , t     k f r , k , t  
t


t
 
n r, t 
1
V
 f r , k , t 

 S r , k , t 
coll
Q conservation
k
 
q
vk  f r , k , t 
V k
P conservation
 
1
E k  f r , k , t 
V k
E conservation
J r, t  
W r, t 
The distribution function f(r,k) can be approximated near equilibrium:
Near equilibrium
equilibrium
0
9
k

f r , k , t
t


coll
f  f0
f
Charge transport
Under (many) simplifying assumptions the 1st moment of BTE gives the DD model
(The semiconductor equations):
J n  qn n E r   qDnn
J p  qp p E r   qDp p
Drift term
Diffusion term
n 1
   Jn Un
t q
p
1
    J p U p
t
q
  V   p  n  N D  N A 
Transport of charge is a combination of drift and diffusion mechanism
* DD expresses momentum conservation: it becomes invalid when sharp variation in energy
of carriers occur (due to F for example: deep submicron devices)
When feature size is 0.’sμm the DD model becomes invalid: higher momentum required
10
Detection principles
A: Ionization: by imparting energy to break a bond, electrons are lifted from VB to CB then
made available to conduction ( ionization chambers, microstrip, hybrid pixels, CCD, MAPS…)
α
MIP
Photon interaction
Bethe formula for stopping power gives the rate of energy loss/unit length
for charged particles through matter
11
I z   I oe E z
Detection
MIP charge density
n
I z   I oe E z
dE 1 1

 3 1015 cm3
2
dx  i   R
R
 v
I
 110nm
A MIP forms an ionization trail of radius R
when traversing Si, creating ~ 80e-/μm
Low injection regime:
the generated charge is too small to affect the
internal electric field
h
 1015 cm
2m E
L  0.5  1107 cm

The associated wavelength is much smaller than mean free path:
Each charge is independent from each other;
Carrier dynamics does not need QM
12
Photoelectric charge density
z  

n
Pin
   z
  e
 5.6 10 6 e  / m
h
An optical power of -60dBm (= 1nW) of 1keV
photons generates ~ 6*106e-/μm
High injection regime:
Plasma effects
The internal electric field can be affected by
the generated charge
Detection
B: Excitation: Charge or lattice (acoustic or optical phonons) some IR detectors, bolometer
60meV
Poly Si
SiO2
Si
Ec
EF
~10’s meV
EV
EF
Eigenvalues separation in quantized structures ~ 10’s meV
Dispersion relation for phonons in Si
Phonon excitation energy ~ 10 meV : much lower threshold
13
Signal conversion: The pn junction
Homojunction: consider two pieces of same semiconductor materials
with different doping levels:
In equilibrium, the Fermi level equalizes throughout the structure
The thermal diffusion of charge across the junction leaves just
the ionized dopants : an electric potential, and a field F, develops
across the junction
In equilibrium J = 0: using DD model
0  qnn E r   qDnn  n  n0e

Vt
0  qp p E r   qDpp  p  p0e
 0


Vt
Near the interface, the carrier concentration exponentially
drops: a depletion region (empty of free charge) is formed.
ASCE (Abrupt Space Charge Edge) approximation:
A ‘positive’ voltage increases (exponentially) the charge concentration: high
direct current.
A ‘negative’ voltage decreases it (down to leakage): the current reduces and at
the same time widens the depleted region.
Unidirectionality of current characteristics
14
Signal conversion: The pn junction
The electric field F in the depletion region of the junction is
sustained by the ionized dopants. When charge is generated is
swept across by the field
PN junction signal converter: A capacitor with a strong F across
A device with a large depleted region W can be used to
efficiently collect radiation generated charge ( Solid state
ionization chamber)
W
W
2Vb N a  N d
q Na  Nd
To achieve large W high field region:
• Low doping (high resistivity) Silicon is needed
• Large voltages
Conversion: Q to V // Q to I
15
Detectors examples
Strip detectors
Scientific applications
Monolithic Active Pixel Sensors (MAPS)
Imaging, consumer applications
Charge Coupled Devices (CCD)
Imaging, scientific and consumer applications
MOS detector
scientific applications
RAL PPD has (is) actively involved with all these detector technologies
16
Detectors examples
Use of Si Strip detectors
Almost all HEP experiments use Si detectors:
The high density track region usually covered by pixel detectors; by strip
at larger radius (cost reason)
17
Detectors examples
module
768 Strip Sensors
RO
ATLAS SCT
4 barrel layers,2 x 9 forward disks
4088 double sided modules
Total Silicon surface 61.1m²
Total 6.3 M channels
Power consumption ~ 50kW
Events rate: 40MHz
Put stave pics of AUG!
18
Detectors examples
Strip detectors
768 Strip Sensors
300μm
80μm
P++
N+ (high res)
F
Wires
RO electronic
Vbias ~100’sV
Power supply
Array of long silicon diodes on a high resistivity silicon substrate
A strong F in the high resistivity Si region helps collect charge efficiently (drift).
The transversal diffusion of charge implies a spread of signal over neighbouring strips
The high resistivity Si is not usually used in mainstream semiconductor industry:
Hybrid solution: detectors connected (wire/bumpbonded) to the readout electronic (RO)
19
Detectors examples
High events rate require fast signal collection:
Estimate of charge collection time in strip detector:
z
1
1 z
t z   
dz  
 z0
z0 v  z 
1
dz
z

Fo 1    F1
 W
For a detector thickness of 300um and overdepleted Vb = 50V and 10kohm resistivity
tcoll(e)≈ 12ns
tcoll(h)≈ 35ns
The fast collection time helps the radiation hardness:
The radiation damage to sensors is a crucial issue in modern HEP experiments
20
Detectors examples
MAPS detectors
≈10’s m
RO electronic
RO electronic
3T ( 3MOS) MAPS structure
2D array of ~106 pixels
Monolithic solution:
Detector and readout integrated onto the same substrate
21
Detectors examples
MAPS detectors
Vbias ~V’s
N++ (low res)
Electronics
0.’s μm
P+ (low-med res)
Active region
‘s μm
P++ (low res)
Mechanical substrate
100’s μm
The charge generated in the thin active region moves by diffusion mainly:
‘Long’ collection time
Small signal
Different implants arrangements for charge collection optimization
Circuit topologies for low noise
22
Detectors examples
Example of MAPS detectors:
10-7
TPAC 1 pixel size 50x50um2
Chip size ~1cm2
Total pixels 28k
>8Meg Transistors
n
l2
2
 Dn n  U n  tcoll 
t
Dn
Charge collection time (s) in MAPS vs. perpendicular MIP hit
23
Detectors examples
CCD detectors
Once the charge has been generated, it accumulates in the potential well, under the capacitor.
The control circuitry shifts the accumulated charge to the end of the row, to the input of a charge
amplifier. The sensor is fabricated in a optimized, dedicated process and the RO on a separate
chip. Superior imaging quality but less integration and speed.
Nobel Prize 2009 for Physics to
inventors Boyle and Smith
24
Detectors examples
5 m
Global Photogate and Transfer gate
In-situ Storage Image Sensor: ISIS
ROW 2: CCD clocks
ROW 3: CCD clocks
On-chip logic
80 m
On-chip switches
ROW 1: CCD clocks
Imaging pixel
ROW 1: RSEL
Global RG, RD, OD
55
Fe  source Mn(K
RG RD
OD RSEL
Mn(Kb 
Column
transistor
25
CCD in CMOS process 0.18μm
Charge collection under a PG then
stored under a 20 pixels storage
CCD
Signal conversion: The unipolar MOS device
NMOS
SiO2
N++
P++
Metal Oxide Semiconductor device are unipolar
devices based on voltage modulation of charge.
The control gate is physically separated by the
active region where the charge moves by a thin
(nm) layer of SiO2.
By applying a voltage to the G with respect to the Substrate
an electric field develops across the SiO2: a charge channel
is formed between Source and Drain.
The Ids characteristics depends on the Vgs applied.
The CMOS process refers to the minimum feature size
achievable i.e. the channel length)
Currently 45nm: the modelling of the characteristics of the
device of this size is non-trivial:
•Quantization effects at the boundary;
•QM tunnelling across the gate;
•Hot carriers near the D/S junction;
•…
26
Signal conversion: The unipolar MOS device
LET in SiO2 for different particles
Generation rate in SiO2 vs. electric field

The SiO2 is a very good insulator: a strong electric
field can be applied to it and the charge
generated in SiO2 by ionizing radiation efficiently
collected
However SiO2 is a polar material: the recombination
processes are stronger than in Si. Furthermore, hole
Transport is non Gaussian (low ‘mobility’) and traps
form near Si interface.
27
Am   K l
 
m0 m! n 0 l  m n 1 l!
Y ( Fox , T , ro )  K 1e(  A)e(  K ) 
K
A
qFox ro
kT
rc
q2
, rc 
ro
4 SiO2 kT
Signal conversion: The unipolar MOS device
Floating Gate
Control Gate
Si
O2
SiO2
Si
O2
FG
FG
Addition of a Floating Gate (FG): the electrical characteristics of the device are controlled by the charge
stored in the FG. The electric field in the SiO2 due to the FG drifts charge towards/away from it.
The discharge of the FG alters the device electrical characteristics
Ids(A)
Pre-rad
Post-rad
Radiation sensitivity
Reprog
Chip #1
100Gy
<∆Vth >
0.6152
Std dev
0.00598
Conversion: Q to I
28
The MOS structure easily allows
excitation based radiation detection
Radiation damage
In HEP and space applications the detectors are exposed to high level of radiation:
LHC: 10’s Mrad (100kGy) over 10years of operation
N.B.: 1 rad/cm3 Si ~1013e/h pairs
TID/Gy/yr
Total Ionizing Dose (rad = 0.01Gy)
fluence)
29
ATLAS
NIEL/cm2/yr
Non Ionizing energy Loss (1MeV neutrons/cm2
Radiation damage
Radiation environment in LHC experiment
TID
Fluence
1MeV n eq. [cm-2] @ 10 years
ATLAS Pixels
ATLAS Strips
CMS Pixels
CMS Strips
ALICE Pixel
LHCb VELO
50 Mrad
7.9 Mrad
~24Mrad
7.5 Mrad
250 krad
-
All values including safety factors.
30
1.5 x 1015
2 x 1014
~6 x 1014
1.6 x 1014
3 x 1012
1.3 x 1014/year
Radiation damage
Microscopic effects: Bulk damage to Silicon :
Displacement of lattice atoms (~ Kinetic Energy Released)
V
EK>25 eV
Atoms scattered by incoming particles leave behind vacancies or atoms in
interstitial positions (Frenkel pairs).
Low energy particle ~ point defects
High energy particles ~ cluster defects
31
I
Vacancy
+
Interstitial
Radiation damage
Energy
deposition
altered
Lattice
periodicity
Atoms
displacement
generation
Donor levels +++
recombination
Band gap
Spurious
states
trapping
Altered
Electrical
characteristics
Conduction band
-
compensation
Band gap
Acceptor levels
Valence band
The appearance of spurious band gap states affects the electro/optical characteristics of the device:
• Thermal generation of carriers (increased leakage current @ same T)
• Reduced recombination time ( quicker charge loss , reduced signal)
• Charge trapping
• Scattering
• Type conversion
32
Radiation damage
Detrimental Macroscopic effects:
• Noise increases because of increase leakage current
• Charge Collection Efficiency (CCE) is reduced by trapping
• Depletion voltage increases because of type inversion


1

Qe,h (t )  Q0 e,h exp 
t
  eff e,h 


1
 N defects
 eff e,h
1015 1MeV n-eq.
33
Radiation damage
To increase the Radiation Hardness of Sensors:
• Operating conditions (cooler – lower leakage)
• Material engineering ( OFZ - Diamond detectors)
• Device engineering (n in n – 3D detectors)
Electrodes in the bulk – lateral collection
The device achieve full depletion
• Low depletion voltage
• short collection time
• claim reduction in signal 33% after 8.8X1015 1Mevn
• difficult to manufacture
•3D DDTC similar to 3D but easier to manufacture; also
better mechanical strength.
* Radiation damage affects also the RO electronics,
but modern process can address the problem efficiently
( guard rings, sub micron devices)
34
Addendum - Detector systems
HEP experiments: large detector systems
Challenging engineering issues
ALICE
ATLAS
CMS
LHCb
Strips
4.9m2
64m2
210m2
14.3m2
Drift
1.3m2
Pixels
0.2m2
2m2
1m2
0.02m2
6.3 x
106
9.6 x
106
1 x 106
80 x
106
33 x
106
1 x 106
Number of Channels
The ATLAS SCT (semiconductor tracker) detector.
The thick red cables on show feed the detector with half of its
power – adding more will take up even more space
35
Strips
2.6 x
106
Drift
1.3 x
105
Pixels
9.8 x
106
Addendum - Detector systems
Alternative powering schemes:
SP
ATLAS SCT Barrel 3 at CERN. Half of the
384 cables are visible; the rest enters the
other end of the detector.
DC2DC
A serial powering or DC2DC approach can increase efficiency in
power distribution compared to a parallel approach
36
Conclusions
The field of semiconductor detectors encompasses different scientific and technology
fields: solid state physics, nuclear and particle physics, electrical engineering, …
Some of the issues relevant to radiation detectors:
• Radiation hardness
• Topologies optimization (power reduction, noise reduction)
• Development of new detection techniques based on novel and well established
semiconductor material: ( phonon-based detectors, compounds, low dimensional)
• Integration with electronics (monolithic solution to achieve more compactness
and reduce cost),3D structures
37
Backup - Detector systems
Power reduction at detector level
=
At pixel level, power consumption could be optimized by
using a non linear approach:
The positive feedback structure is biased near threshold
(variable)
A small signal triggers the structure
I
Backup - Detection
The variance in signal charge σi associated to the ionization process is related to the phonon excitation
i 
Eo
i
E pn   i

  1
Ei  Ei 
Fano factor ~0.1 in Si
High resolution requires smaller band gap (εi ),
direct or small phonon excitation energy
Intrinsic resolution of Si and Ge based detectors
II
Backup -Detection
Ph: Q~107 m-1
Q~1010 m-1
/a
The indirect BG of Si requires higher energy for charge excitation, because energy and
momentum must be conserved (Phonon-assisted pair creation/recombination)
In Si an average of 3.6 eV is required for pair creation
Put values of photon momentum typ.
III
Backup
Quantization effects due to band bending in Si-SiO2 interface: excitation based detection
SiO2
Si-sub
Si-poly
Q-effects
IV
Backup - The bipolar transistor device
A bipolar transistor can be thought of as a two diode system,
connected in anti series;
•One is forward biased;
•The other is reverse biased
The bipolar transistor can be (and it is) used as a high gain
detector
Main limitations arising from speed: the minority carriers diffuse
through the base ( relatively low speed)
V