Transcript Design, simulation, production and initial characterisation of 3D silicon detectors David Pennicard

Design, simulation, production and initial
characterisation of 3D silicon detectors
David Pennicard – University of Glasgow
Richard Bates, Celeste Fleta, Chris Parkes – University of Glasgow
G. Pellegrini, M. Lozano - CNM, Barcelona
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
3D Detector Structure
•
•
•
•
Array of electrode columns passing through substrate
Electrode spacing << wafer thickness (e.g. 30m:300m)
Benefits
– Vdepletion  (Electrode spacing)2
– Collection time  Electrode spacing
– Reduced charge sharing
More complicated fabrication - micromachining
Planar
+ve
+ve
3D
+ve
-ve
n-type
electrode
+ve
n-type
electrode
electrons
electrons
Lightly
doped
p-type
silicon
holes
300
µm
300
µm
holes
p-type
electrode
p-type
electrode
Particle
-ve
Particle
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Around
30µm
Background
•
•
•
•
Invented in 1997 - S. Parker, C. Kenney, J. Segal
– First produced in 1999 - Stanford Nanofabrication facility
Recent development: R&D towards experimental use
– Improvements in micromachining make larger-scale, reliable production
more feasible
– Application: radiation-hard detectors for Super-LHC
3D detector collaboration between Glasgow and CNM (Centro Nacional de
Microelectronica, Spain)
– Optimisation of 3D design through simulation
– Fabrication of 3D detectors in CNM cleanroom
– Initial characterisation
Overview of other 3D detector projects
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Super-LHC and Radiation Damage
•
•
•
RD50 collaboration – see G. Casse talk
Upgrade to LHC, planned for sometime after 2017
– 10x increase in luminosity
10x increase in radiation damage
– Inner layer of ATLAS pixel tracker will receive 1016neq/cm2 damage
over SLHC running time
Ian Dawson, University of Sheffield
ATLAS upgrade workshop,
Valencia, December 2007
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
3D Detectors and Radiation Hardness
•
Increase in effective p-type doping with damage
– Increased depletion voltage
– 300μm planar detectors cannot be fully
depleted far beyond 1015neq/cm2
– 3D detectors have short depletion distance,
reducing Vdep
•
Charge trapping
– Free electrons and holes trapped by defects,
reducing CCE n   n
 eff ,e
t
See M. Moll thesis, Hamburg 1999
– Dominant effect at very high fluences
– 3D structure reduces collection time – less
trapping
•
Increased leakage current
– Need to cool detectors
G. Kramberger, Aug. 23-24, 2006, Hamburg, Germany
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Simulation of 3D detectors after radiation damage
Simulations performed using Synopsys TCAD
Predict higher collection efficiency for 3D than for planar sensors
– Model uses pessimistic values for trapping rates
25000
strips pixels
signal [electrons]
•
•
20000
p-in-n
n-in-p
3D simulation
15000
140m p-FZ
10000
[1] 3D, double sided, 250m columns, 300m substrate [Pennicard 2007]
[2] p-FZ, 280m, (-30oC, 25ns), strip [Casse 2007]
[3] p-FZ, 280m, (-30oC, 25ns), strip [Casse 2004]
[4] p-MCZ, 300m, (-30OC,  s), pad [Bruzzi 2006]
[5] p-MCZ, 300m, (<0OC, s), strip [Bernadini 2007]
[6] n-MCZ, 300m, (-30OC, 25ns), strip [Messineo 2007]
[7] p-FZ, 140m, (-30oC, 25ns), strip [Casse 2007]
[8] n-EPI, 150m, (-30OC, 25ns), strip [Messineo 2007]
[9] n-epi Si, 150m, (-30oC, 25ns), pad [Kramberger 2006]
[10] n-epi Si, 75m, (-30oC, 25ns), pad [Kramberger 2006]
150m n-EPI
5000
Double-sided 3D, 250 m, simulation! [1]
n-in-p (FZ), 280 m [2,3]
n-in-p (MCZ), 300m [4,5]
p-in-n (MCZ), 300m [6]
n-in-p (FZ), 140 m, 500V [7]
p-in-n (EPI), 150 m [8,9]
p-in-n (EPI), 75m [10]
75m n-EPI
See also: [M. Bruzzi et al. NIM A 579 (2007) 754-761]
[H.Sadrozinski, IEEE NSS 2007, RD50 talk]
1014
1015
eq [cm-2]
1016
M.Moll 2007
Plot compiled by M. Moll
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Optimisation of ATLAS 3D structure
•
ATLAS pixel is 400μm * 50μm
– Different layouts available
– Trade-offs between Vdep, CCE,
capacitance, column area…
3 column
8 column
Charge collection with
1016neq/cm2 radiation damage
Capacitance at each pixel
600
14
8
Total C per pixel
Interpixel C
ATLAS 3D CCE
7
500
8 7
10
Bars show variation in
CCE with hit position
6
5
8
4
6
3
Capacitance (fF)
Charge collection (ke-)
12
6
400
5
300
4
200
3
4
2
Smaller electrode
spacing improves CCE
2
2
100
0
0
0
20
40
60
80
Electrode spacing (m)
100
0
20
40
60
80
Electrode spacing (m)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
100
Double-sided 3D detectors at CNM
•
•
•
Alternative 3D structure proposed by IMB-CNM
N- and p-type columns etched from opposite sides of substrate
– Columns do not pass through full substrate thickness (in first production
run)
– 250μm deep in 300μm substrate
Recently finished production with p+ column readout and n-type substrate
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D Detector production
• Column fabrication introduces
extra steps
• Begin with columns on back side
SiO2
Si, n-type, 300 um
Al/Cu
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D Detector production
Hole etching
•
Deep Reactive Ion Etching
– F plasma etches away base of hole
– CF2 coating protects sidewall
– Limit on depth : diameter ratio
– 250m depth, 10m diameter
SiO2
Si, n-type, 300 um
250μm
Al/Cu
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
10μm
Double-sided 3D Detector production
Column filling and doping
•
•
•
Deposit 3μm poly-silicon
Phosphorus doping through poly
Passivate inside of column with SiO2
Junction
SiO2
TEOS
Poly
(p+)
Si-n+
Poly-n+
2.9m
n-Si
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D Detector production
Finished detector
•
•
•
P+ columns fabricated on front side
Contacts on front
Backside coated with metal for biasing
Al/Cu
Passivation
Si-p+
250μm
Si-n+
10μm
Poly-n+
Al/Cu
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Finished 3D devices
Devices include: Pads, strips, pixels detectors, test structures
Typical device layout –
Strip detector, 80μm pitch
3D guard
ring
Bond pads
Collecting
electrodes
Bias
electrodes
(back
surface)
80μm
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Finished 3D devices
SEM after polysilicon
deposition and etching
Pixel on Medipix detector
Dry etching
of the poly
Polysilicon
Polysilicon and column
(under passivation)
SiO
2
9.4m
Passivation (SiO2
and SiN)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Bump-bond
contact
Lateral depletion around
column (~2V in sim.)
Initial tests - CV
•
Pad detector – 90 * 90 columns, 55μm pitch
P+
Pad detector CV
2.0E-09
1.8E-09
2.3V
lateral
depletion
Capacitance (F)
1.6E-09
1.4E-09
1.2E-09
1.0E-09
8.0E-10
6.0E-10
4.0E-10
2.0E-10
N+
0.0E+00
0.0
5.0
10.0
15.0
Bias (V)
Depletion to back surface from
tip of column (~8V in sim.)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
20.0
Lateral depletion around
column (~2V in sim.)
•
Initial tests - CV
Pad detector – 90 * 90 columns, 55μm pitch
P+
1/Capacitance, Pad detector
5.0E+09
4.5E+09
4.0E+09
1/C (F-1)
3.5E+09
3.0E+09
2.3V
lateral
depletion
2.5E+09
2.0E+09
~9V back
surface
depletion
1.5E+09
1.0E+09
5.0E+08
N+
0.0E+00
0.0
5.0
10.0
15.0
Bias (V)
Depletion to back surface from
tip of column (~8V in sim.)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
20.0
Initial tests – Strip detector IV
128 strips, 50 holes/strip, pitch 80um, length 4mm
Measured with 3 strips and guard ring at 0V, and backside biased
Strip currents ~100pA (T=21˚C) in all 4 detectors
Can reliably bias detectors to 50V (20 times lateral depletion voltage)
Capacitance 5pF / strip
1.0E-04
strip detector 4
Guard ring currents vary:
1.0E-05
– Highest 20μA at 10V
– Lowest 0.03μA at 50V
1.0E-06
Guard ring
1.0E-07
I(A)
•
•
•
•
•
•
1.0E-08
1.0E-09
Neighbours
Strip
1.0E-10
1.0E-11
0.0
10.0
20.0
30.0
V(V)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
40.0
50.0
Future work
•
Tests on these detectors
– Charge collection test on strip detector with
beta source and LHCb readout electronics
• Tests before and after irradiation
– X-ray detection test, using Medipix pixel
readout (single-photon-counting)
•
New production run at CNM
– Columns pass through full substrate thickness
– Both p+ readout with n-substrate, and n+
readout with p-substrate
– Includes ATLAS pixel detectors
•
Testbeams at CERN in summer
– Collection performance vs position
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Other 3D detector projects
• Stanford / Manchester / Sintef
• FBK-IRST (Trento, Italy)
• Glasgow / Diamond / IceMOS
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Stanford / Manchester / Sintef
•
•
•
First 3D detectors produced at Stanford Nanofabrication Facility
University of Manchester and CERN testing detectors
– Have demonstrated good charge collection behaviour of
ATLAS 3D pixels after SLHC radiation fluences
Working with Sintef (independent research foundation in Norway)
to reproduce Stanford fabrication process on a larger scale
Charge collection and
signal/noise results
Thanks to Cinzia da Via (Manchester)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Stanford / Manchester / Sintef
•
•
“Active edge” electrode
– Usually, silicon sensors have >100μm insensitive area at edge (need
to avoid current flow from saw-cut edges)
– Instead, plasma etch edge, and add a doped polysilicon layer
– Edge acts as an electrode – dead area just 5μm
Achieve good coverage with fewer overlapping layers
45-54
X-ray microbeam scan
36-45
27-36
18-27
9-18
54
45
0
36
60
27
18
9
0
120
180
microns
240
300
360
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
0-9
Developments in Trento, Italy
CV-diode - W861
35.0
Cdiode [pF]
Double-side Double-Column 3D detectors
30.0
stc100
25.0
dtc100
stc80
dtc80
20.0
15.0
10.0
5.0
0.0
0
1
2
Vrev [V]
3
4
Good results from preliminary
electrical tests (C-V and I-V)
First prototypes (p-on-n) completed,
and n-on-p available soon.
3Ddtc1 - Wafer#861
Idiode [nA]
0.10
0.09
0.08
stc2
stc3
dtc2
dtc3
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0
20
40
60
Vrev [V]
80
100
Glasgow / Diamond / IceMOS
•
•
Project between Glasgow and Diamond synchrotron to develop 3D
detectors for X-ray crystallography
– Single-photon-counting pixel sensors (Medipix, Pilatus)
– Lower charge sharing in 3D detectors
– Potential for thick 3D silicon detectors with good performance
Detectors produced in fabrication company IceMOS (Belfast)
– First 3D detectors produced entirely in industry
– Prototype run finished
• Working test structures, but some problems with full devices
– Starting second run with improved fabrication flow
p-electrode (readout)
n-electrode (bias)
passivation
Metal
SiO2
poly-n+
Si(n--)
Si-n+
poly-p+
Si-p+
SiO2
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Conclusions
•
•
•
3D detectors
– Fast collection, low depletion voltage
– Radiation hard – candidate for SLHC inner pixel layers
3D production at CNM
– First set of double-sided 3D detectors produced
– Preliminary tests successful – continuing with charge collection tests
– More production runs underway
Other 3D projects
– Different groups working towards 3D detectors for high-luminosity
colliders
– Other applications possible, such as X-ray crystallography
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Thank you for listening
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
First CNM 3D production run
•
P+ readout, n-type substrate devices on 4” wafer
•
6 Medipix2 pixels
Pitch 55μm, 256x256
– Single-photon counting sensor for medical X-ray detection (CERN)
1 Pilatus pixel
Pitch 172μm, 97x60
– Single-photon-counting sensor for X-ray crystallography (PSI)
6 ATLAS pixels
Pitch 50x400μm, 164x18
– Prototypes (wrong readout polarity)
4 short strip
Pitch 80μm, 50x50
1 long strip
Pitch 80μm, 50x180
Pad detectors, test structures
•
•
•
•
•
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D detector – simulated behaviour
•
•
•
Where columns overlap, same
behaviour as standard 3D
Weaker field near front and back
surfaces – slower collection
Greater device thickness for given
column length
Electric
field,
100V
bias
Detail of electric
field (V/cm)
around
top of
n-type double-sided 3D device (100V bias)
0
10
25
50
20
Variation in charge collection with depth
4.00
50%
90%
Full
Z (um)
Time (ns)
for given %
collection:
00
00
10000
30
40
20000
50
140000
3.00
0
1.00
10
20
40000
70
30000
2.00
3 0000
60
4 0 00 0
Collection time (ns)
5.00
P+
30
40
50
D (um)
0.00
0
50
100
150
200
250
300
Depth (um)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
N+
Simulation of 3D detectors after radiation damage
Simulations performed using Synopsys TCAD
Predict higher collection efficiency than planar sensors
– Model uses conservative values for trapping rates
25
Simulated CNM 3D
(55m pitch)
Experimental n-on-p
results
Simulated n-on-p
20
Charge collection(ke-)
•
•
15
10
5
0
0.0
2.0
4.0
6.0
15
8.0
10.0
2
Fluence (10 neq/cm )
12.0
N-on-p results: PP Allport et al., IEEE
Trans. Nucl. Sci., vol 52, Oct 2005
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Simulation methods
•
•
See presentation from 10th RD50 meeting
Synopsis TCAD finite element simulation
•
Damage model
– Trap dynamics modelled directly
– P-type FZ material
– Based on work at Uni. Perugia – see M.
Petasecca et al., IEEE Trans. Nucl. Sci.,
vol. 53, pp. 2971–2976, 2006
– Modified to match experimental trap times
(V. Cindro et al., IEEE NSS, Nov 2006)
βe=
4.0*10-7cm2s-1, βh=
Type
Energy
(eV)
Trap
4.4*10-7cm2s-1,
σe
(cm2)
σh
1
e
Example of a
simulated 3D
structure
n+ contact
p+ contact
oxide
eeq
(cm2)
η
(cm-1)
Acceptor
Ec-0.42
VV
9.5*10-15
9.5*10-14
1.613
Acceptor
Ec-0.46
VVV
5.0*10-15
5.0*10-14
0.9
Donor
Ev+0.36 CiOi
3.23*10-13 3.23*10-14
0.9
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
N+ on p strip detector: CCE
At high fluence, simulated CCE is lower than experimental value
– Trapping rates were extrapolated from measurements below 1015neq/cm2
– In reality, trapping rate at high fluence probably lower than predicted
25
Simulated strip
Experimental results
20
Charge collection(ke-)
•
PP Allport et al., IEEE Trans.
Nucl. Sci., vol 52, Oct 2005
900V bias,
280m thick
15
10
From β values used,
expect 25μm drift
distance, 2ke- signal
5
0
0.0
2.0
4.0
6.0
8.0
15
2
Fluence (10 neq/cm )
10.0
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
ATLAS 3D detector: CCE
•
Experiment used n+ readout, with 3 n+ columns per ATLAS pixel
Experiment used defocused IR laser pulse to flood the pixel with charge; the
simulation mimics this
Both experiment and simulation show improved CCE at high fluence
25
Simulated ATLAS 3D
Experimental results
20
60V
Charge collection (ke-)
•
•
C. da Via et al.,
Liverpool ATLAS 3D
meeting, Nov. 06
Detectors produced at
Stanford
15
60V
100V
At high fluences,
simulated CCE ~2/3
of experimental
value (like with
planar detector)
10
160V
5
0
0.0
2.0
4.0
6.0
8.0
15
2
Fluence (10 neq/cm )
10.0
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Overview
• Radiation damage model and comparison with
experiment
• Behaviour of different ATLAS pixel 3D layouts
• Comparison of double-sided & standard 3D
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
ATLAS 3D simulations
•
ATLAS pixel (400m * 50m) allows layouts with different electrode spacing
– No of n+ columns per pixel could vary from ~2-8
• Stanford have produced devices with 2-4 n+ columns
• Previous ATLAS results shown used 3 columns
Simulations use 230m-thick p-type substrate, n+ readout
– Columns have 5m radius, with dopant profile extending ~2m further
– P-spray is used to isolate the columns
•
400m
3
50m
Spacing
133m cell length
8
50m cell length
Note larger volume occupied by columns
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
ATLAS 3D – Depletion voltage at 1016neq/cm2
•
•
Depletion voltage will depend on substrate material (this model matches ptype FZ, rather than oxygenated silicon)
No. of n+ columns shown next to each data point
Vdep proportional to depletion distance squared
250
3
Depletion voltage
Fit:
2
V=0.07(X-13.5m) -1.5
200
Bias (V)
•
150
4
100
5
6
50
8
7
0
0
20
40
60
Electrode spacing (m)
80
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
ATLAS 3D – high-field voltage at 1016neq/cm2
•
As an approximate judge of a “safe voltage”, found the bias at which the
maximum field in each device reached 2.5*105V/cm
Surprisingly, all the devices gave much the same results at 1016neq/cm2
250
3
Depletion voltage
High field voltage
200
150V safe level
Bias (V)
•
150
4
100
5
6
50
8
7
0
0
20
40
60
Electrode spacing (m)
80
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Device structure and high-field regions
•
•
•
•
P-spray links p+ columns to n+
So, the p-spray is at the same potential as the p+, resulting in high field at
front surface where it meets the n+ columns
At higher bias the p-spray around the n+ column becomes depleted
These effects won’t be greatly affected by the electrode spacing itself
5-column
ATLAS 3D,
5-column ATLAS 3D device
2, 150V
1016neq/cm
, 150V
bias
10
neq/cm
bias
16
5-column
ATLAS 3D,
5-column ATLAS 3D device
neq/cm
bias bias
101610neq
/cm2, 150V
, 150V
Z
2
16
Y
Z
2
Y
X
X
p-spray
n+
p-spray
p+
Doping conc.
Doping
-3concentration
(cm
(cm ) )
-3
n+
p+
Electrostatic
potential
(V)
Electrostatic potential
(V)
6.0E+18
-10
8.8E+15
-30
1.3E+13
-50
-1.3E+13
-70
-8.8E+15
-90
-6.0E+18
-110
-130
-150
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Device structure and high-field regions
•
•
•
•
P-spray links p+ columns to n+
So, the p-spray is at the same potential as the p+, resulting in high field at
front surface where it meets the n+ columns
At higher bias the p-spray around the n+ column becomes depleted
These effects won’t be greatly affected by the electrode spacing itself
5-column
ATLAS 3D,
5-column ATLAS 3D device
2, 150V
1016neq/cm
, 150V
bias
10
neq/cm
bias
16
5-column
ATLAS 3D,
5-column ATLAS 3D device
neq/cm
bias bias
101610neq
/cm2, 150V
, 150V
Z
2
16
Y
Z
2
Y
X
X
p-spray
n+
p-spray
p+
Doping conc.
Doping
-3concentration
(cm
(cm ) )
-3
n+
6.0E+18
p+
Hole conc.
Hole concentration
-3
(cm
(cm ) )
-3
8.8E+15
1.0E+14
1.3E+13
4.1E+13
-1.3E+13
1.7E+13
-8.8E+15
7.0E+12
-6.0E+18
2.8E+12
1.0E+12
0.0E+00
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Charge collection vs position at 1016neq/cm2
•
Simulated MIPs passing through detector at 25 positions, to roughly map
the collection efficiency. Charge sharing not taken into account.
8 columns
6 columns
p+
14
12
10
8
9
25
20
)
5
6
8
8
20
25
0 15
20
2.0 10
20
n+
15
15
4.0
10 m )
5
10 m )
6.0
X(
5

(
X
8.0
0 0
6
10
12
D.Pennicard, University of Glasgow,
INSTR08, Novosibirsk
14
)
m
n+
25
m
Y(
10
6
4
2
25
0
10
Y(
15
00
12
10
10
5
14
11
12
11
Charg
e colle
ction
(ke-
Charg
e colle
ction
(ke-
)
)
p+
30
8
6
4
2
0
0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
Charge collection vs position at 1016neq/cm2
•
Simulated MIPs passing through detector at 25 positions, to roughly map
the collection efficiency. Charge sharing not taken into account.
45
7
40
35
5
6
25
20
30
n+
15
15
Y
10
m
(
10
5
)
5
0 0
25 )
20 (m
X
0
6
5
4
2.0
20
4.0 25
n+
20
6.0
15
15
8.0 Y
10
( 10
10.0
m
5
)
5
12.0
0 0
14.0
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
ollec
ti
Char
ge c
4
)
8
Cha
rge c
ollec
t
ion (
ke-)
14
12
10
8
6
4
2
0
50
14
12
10
p+
8
6
4
2
0
65
4
2 60
55
50
45
40
35
0
30
2.0
25
4.0
(
m
p+
on (k
e-)
3 columns
X
4 columns
6.0
8.0
10.0
12.0
14.0
Average ATLAS CCE at 1016neq/cm2
•
Average CCE found by flooding entire pixel with charge
Previous simulations used to find RMS variation from average, as a
measure of nonuniformity. Shown by “error bars”.
CCE improves as electrode spacing is reduced (faster collection)
14
ATLAS 3D CCE
12
Charge collection (ke-)
•
•
8 7
10
6
5
8
Variation in
collection with
position larger
relative to CCE
4
6
3
4
2
2
0
0
20
40
60
80
100
Electrodeofspacing
m)
D.Pennicard, University
Glasgow,(INSTR08,
Novosibirsk
Total capacitance seen at each ATLAS pixel
•
The total pixel capacitance was found with 1012cm-2 oxide charge (a typical
saturated value) but without radiation damage.
C increases rapidly with no. of columns – the column capacitances add in
parallel, and the capacitance per column gets larger as spacing decreases.
600
8
Total C per pixel
Interpixel C
7
500
Capacitance (fF)
•
6
400
5
300
4
200
3
2
100
Unlike in planar
detectors, interpixel
C is only a small
component of total
0
0
20
40
60
80
100
Electrode
m)
D.Pennicard, University
of spacing
Glasgow, (INSTR08,
Novosibirsk
Signal to noise estimate at 1016neq/cm2
•
Uses noise vs. capacitance data from unirradiated ATLAS sensors (won’t
include high leakage current or damage to readout chip)
– Assume 100fF from preamplifier input and bump bond
– Also 70e- threshold dispersion
Estimated signal-to-noise ratio
40
Noise≈60e-+39e-/100fF
35
8
30
7
6 5
4
Increasing C noise
counteracts
improving CCE
25
20
3
15
2
“Progresses on the
ATLAS pixel detector”,
A. Andreazza, NIMA vol.
461, pp. 168-171, 2001
10
ATLAS 3D SNR
5
0
0
20
40
60
80
100
Electrode
spacing (m)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Overview
• Radiation damage model and comparison with
experiment
• Behaviour of different ATLAS pixel 3D layouts
• Comparison of double-sided & standard 3D
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Comparison of double-sided & standard 3D
n+
readout
•
•
Full 3D (Parker et al., Stanford, Sintef, ICEMOS)
Double-sided 3D (CNM, Trento)
– Readout columns etched from front surface
– Bias columns etched from back surface
– Columns don’t pass through full substrate thickness
•
The maximum column depth that can be etched is about 250m
(with a 5m radius)
– Double-sided 3D simulation uses 250m columns in a
300m substrate
– Full-3D device used for comparison is 250m thick
•
Device structure used for comparison
– N+ columns used for readout, p-type substrate
– 55m* 55m pixel size (Medipix)
– 100V bias
p+ bias
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D field and depletion
•
Where the columns overlap, (from 50m to 250m depth) the
field matches that in the full-3D detector
At front and back surfaces, fields are lower as shown below
Region at back is difficult to deplete at high fluence
A.
•
•
D o u b le - s id e 2d 3 D , p - ty p e ,
16
2, front surface
1 6eq
n e/cm
q /c m , fro
n t s u rfa c e
A. 101 e + n
0
100V
100V
0
n+
n+
200
00
Z (m)
30
50
p+
30000
30000
190000
170000
150000
130000
110000
90000
70000
50000
30000
20000
10000
5000
0
00
E le c tric
F ie ld ( V /c m )
250
250
Z (m)
E le c tric
F ie ld ( V /c m )
60
0
00
00
10
240
20
40
25 00
230
50
10
B.
70000
2500
D o u b le - s id e 2d 3 D , p - ty p e ,
16
2,
10
surface
1 e + 1n6 eq
n e /cm
q /c m , bback
a c k s u rfa
ce
p+
260
10
00
0
270
2500
280
Undepleted
290
190000
170000
150000
130000
110000
90000
70000
50000
30000
20000
10000
5000
0
70
B.
0
25
D ( m )
50
300
0
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
25
D ( m )
50
Collection with double-sided 3D
•
•
Slightly higher collection at low damage
But at high fluence, results match standard 3D due to poorer collection from
front and back surfaces.
25
20% greater substrate
thickness
Standard 3D, 250m substrate
Double-sided 3D, 250m
columns, 300m substrate
Charge collection (ke-)
20
15
10
5
0
0.0
2.0
4.0
6.0
8.0
15
2
10.0
Fluence (10 neq/cm )
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
High-field regions in full and double-sided 3D
•
•
Simulated full and double-sided 3D using p-spray isolation at 1016 neq/cm2
Double-sided 3D is less prone to surface effects because columns are
etched from opposite sides, but high-field regions develop at n+ column tip.
Double-sided 3D
Full 3D
0
Field reaches
2.5*105V/cm at 170V
70
00
0
10000
50000
20
30
n+
20
30
D (m)
40
20000
n+
40
p+
50
60
25000
Field reaches
0
10
20
2.5*105V/cm
at 130V
D.Pennicard,
University
of Glasgow, INSTR08, Novosibirsk
00
190000
170000
150000
130000
110000
90000
70000
50000
30000
20000
10000
5000
0
30000
60000
30000
10
Electric
Field (V/cm)
0
40
0
p+
20
4 00 0
190000
170000
150000
130000
110000
90000
70000
50000
30000
20000
10000
5000
0
Z (m)
10
Z (m)
Electric
Field (V/cm)
25000
1 00 000
0
30
D (m)
40
50
P-type FZ model – proton irradiation
Type
•
•
•
Energy
(eV)
Trap
σe
(cm2)
σh
(cm2)
η
(cm-1)
Acceptor
Ec-0.42
VV
9.5*10-15
9.5*10-14
1.613
Acceptor
Ec-0.46
VVV
5.0*10-15
5.0*10-14
0.9
Donor
Ev+0.36
CiOi
3.23*10-13
3.23*10-14
0.9
See presentation from RD50 June 2007
Based on work at Uni. Perugia – see M. Petasecca et al., IEEE Trans. Nucl.
Sci., vol. 53, pp. 2971–2976, 2006
Modified to give correct trapping times while maintaining depletion
behaviour
n n

e
t
1
e
eeq
e vthee
•
Experimental trapping times for p-type silicon (V. Cindro et al., IEEE NSS,
Nov 2006) up to 1015neq/cm2
– βe= 4.0*10-7cm2s-1
βh= 4.4*10-7cm2s-1
•
Assume these can be extrapolated to 1016neq/cm2
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Comparison with experiment
•
•
•
Compared with experimental results with proton irradiation
Depletion voltage matches experiment
Leakage current is higher than experiment, but not excessive
P-type trap model: Leakage Current
P-typetrapm
odels: Depletionvoltages
0.30
600
Depletionvoltage(V)
500
450
400
Default p-typesim
M
odifiedp-typesim
350
α=5.13*10-17
A/cm
α=3.75*10-17A/cm
0.25
Leakage current (A/cm^3)
550
“Comparison of Radiation Hardness of P-in-N, N-in-N, and N-in-P Silicon
Pad Detectors”, M. Lozano et al., IEEE Trans. Nucl. Sci., vol. 52, pp. 1468–
1473, 2005
0.20
0.15
0.10
0.05
Experimentally,
α=3.99*10-17A/cm3 after 80 mins
anneal at 60˚C (M. Moll thesis)
Experim
ental
300
0.00
0
1E+14
2E+14
3E+14
4E+14
Fluence(Neq/cm
2)
5E+14
6E+14
7E+14
0
1E+15
2E+15
3E+15
4E+15
Fluence (neq/cm^2)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
5E+15
6E+15
Example of CCE with varying bias
•
CCE curves show a smaller gradient after depletion voltage is reached
Collection vs bias in 5-column ATLAS
16
10^16neq/cm^2
5*10^15neq/cm^2
Charge collected (ke-)
14
12
10
Vdep
8
Vdep
6
CCE increases beyond Vdep, due
to increasing carrier velocity
4
2
0
0
20
40
60
80
100
120
140
D.Pennicard, University of
Glasgow,
Bias
(V) INSTR08, Novosibirsk
160
180
Electric field distribution – 8 columns per pixel
The previous simulations showed an “average” CCE for the pixel, but the
uniformity across the pixel is also important. The following slides show how
the electric field distribution varies with the pixel layout
ATLAS 3D, p-type, 50m
cell, 8 column
2
1e+16neq/cm , 150V bias
25
10
00
0
p+
70 0 00
20
40
0
Electric
Field
(V/cm)
00
15
Y (m)
190000
170000
150000
130000
110000
90000
70000
50000
30000
20000
10000
5000
0
00
0
50 000
10
40
11
00
00
00
0
5
0
10
•
n+
0
5
10
15
20
25
30
D.Pennicard, University of Glasgow,
X (m)INSTR08, Novosibirsk