design, simulation, production and initial characterisation of 3d silicon detectors
DESCRIPTION
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. 3D Detector Structure. - PowerPoint PPT PresentationTRANSCRIPT
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
• Array of electrode columns passing through substrate• Electrode spacing << wafer thickness (e.g. 30m:300m)• Benefits
– Vdepletion (Electrode spacing)2
– Collection time Electrode spacing– Reduced charge sharing
• More complicated fabrication - micromachining
3D Detector Structure
+ve+ve
holes
-ve
electrons
Lightly doped p-type
silicon
n-typeelectrode
p-typeelectrode
Particle
+ve+ve
holes
-ve
electrons
Lightly doped p-type
silicon
n-typeelectrode
p-typeelectrode
Particle
Planar 3D
Around30µm
+ve +ve-ve
holes
300µm
n-typeelectrode
p-type electrode
electrons
Particle Around30µm
+ve +ve-ve
holes
300µm
n-typeelectrode
p-type electrode
electrons
Particle
300µm300µm
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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 SheffieldATLAS 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
– Dominant effect at very high fluences– 3D structure reduces collection time – less
trapping
• Increased leakage current– Need to cool detectors
See M. Moll thesis, Hamburg 1999
G. Kramberger, Aug. 23-24, 2006, Hamburg, Germany
eeff
nt
n
,
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
1014 1015 1016
eq [cm-2]
5000
10000
15000
20000
25000
sign
al [e
lect
rons
]
3D simulation
Double-sided 3D, 250 m, simulation! [1]n-in-p (FZ), 280 m [2,3]n-in-p (MCZ), 300m [4,5]p-in-np-in-n (MCZ), 300m [6]n-in-p (FZ), 140 m, 500V [7]p-in-n (EPI), 150 m [8,9]p-in-n (EPI), 75m [10]
75m n-EPI
150m n-EPI
n-in-p
140m p-FZ
M.Moll 2007
[1] 3D, double sided, 250m columns, 300m substrate [Pennicard 2007][2] p-FZ, 280m, (-30oC, 25ns), strip [Casse 2007][3] p-FZ, 280m, (-30oC, 25ns), strip [Casse 2004][4] p-MCZ, 300m, (-30OC, s), pad [Bruzzi 2006][5] p-MCZ, 300m, (<0OC, s), strip [Bernadini 2007][6] n-MCZ, 300m, (-30OC, 25ns), strip [Messineo 2007][7] p-FZ, 140m, (-30oC, 25ns), strip [Casse 2007][8] n-EPI, 150m, (-30OC, 25ns), strip [Messineo 2007][9] n-epi Si, 150m, (-30oC, 25ns), pad [Kramberger 2006][10] n-epi Si, 75m, (-30oC, 25ns), pad [Kramberger 2006]
See also: [M. Bruzzi et al. NIM A 579 (2007) 754-761] [H.Sadrozinski, IEEE NSS 2007, RD50 talk]
pixelsstrips
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
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…
0 20 40 60 80 1000
2
4
6
8
10
12
14
2
76
5
4
3
8
Cha
rge
colle
ctio
n (k
e-)
Electrode spacing (m)
ATLAS 3D CCE
0 20 40 60 80 1000
100
200
300
400
500
600
2
7
6
5
4
3
8
Cap
acita
nce
(fF
)
Electrode spacing (m)
Total C per pixel Interpixel C
8 column
3 column
Charge collection with 1016neq/cm2 radiation damage
Smaller electrode spacing improves CCE
Capacitance at each pixel
Bars show variation in CCE with hit position
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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
Passivation
p+ doped
55m pitch
50m
300m
n+ doped
10m
Oxide
n+ doped
Metal
Poly 3m
Oxide
Metal
50m
TEOS oxide 2m
UBM/bump
n-type Si
Passivation
p+ doped
55m pitch
50m
300m
n+ doped
10m
Oxide
n+ doped
Metal
Poly 3m
Oxide
Metal
50m
TEOS oxide 2m
UBM/bump
n-type Si
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Si, n-type, 300 um
SiO2
Al/Cu
Si, n-type, 300 um
SiO2
Al/Cu
Double-sided 3D Detector production
• Column fabrication introduces extra steps
• Begin with columns on back side
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D Detector production
• Deep Reactive Ion Etching– F plasma etches away base of hole
– CF2 coating protects sidewall – Limit on depth : diameter ratio– 250m depth, 10m diameter
Hole etching
Si, n-type, 300 um
SiO2
Al/Cu
Si, n-type, 300 um
SiO2
Al/Cu10μm
250μm
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D Detector production
• Deposit 3μm poly-silicon• Phosphorus doping through poly
• Passivate inside of column with SiO2
Column filling and doping
2.9m
TEOS
PolyJunction
2.9m
TEOS
PolyJunction
n-Si
(p+) Si-n+
Poly-n+
SiO2
Si-n+
Poly-n+
SiO2
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D Detector production
• P+ columns fabricated on front side• Contacts on front• Backside coated with metal for biasing
Finished detector
10μm
250μm
Poly-n+
Passivation
Si-p+
Si-n+
Al/Cu
Al/Cu
Poly-n+
Passivation
Si-p+
Si-n+
Al/Cu
Al/Cu
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Finished 3D devices
3D guard ring
Collecting electrodes
Bias electrodes(back surface)
Bond pads
Typical device layout – Strip detector, 80μm pitch
80μm
Devices include: Pads, strips, pixels detectors, test structures
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Finished 3D devices
m
SiO2
Polysilicon
Dry etching of the poly
SEM after polysilicon deposition and etching
Pixel on Medipix detector
Polysilicon and column (under passivation)
Passivation (SiO2 and SiN)
Bump-bond contact
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Pad detector CV
0.0E+00
2.0E-10
4.0E-10
6.0E-10
8.0E-10
1.0E-09
1.2E-09
1.4E-09
1.6E-09
1.8E-09
2.0E-09
0.0 5.0 10.0 15.0 20.0
Bias (V)
Cap
acit
ance
(F
)
Initial tests - CV• Pad detector – 90 * 90 columns, 55μm pitch
P+
N+
Lateral depletion around column (~2V in sim.)
Depletion to back surface from tip of column (~8V in sim.)
2.3V lateral depletion
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
1/Capacitance, Pad detector
0.0E+00
5.0E+08
1.0E+09
1.5E+09
2.0E+09
2.5E+09
3.0E+09
3.5E+09
4.0E+09
4.5E+09
5.0E+09
0.0 5.0 10.0 15.0 20.0
Bias (V)
1/C
(F
-1)
Initial tests - CV
P+
N+
Lateral depletion around column (~2V in sim.)
Depletion to back surface from tip of column (~8V in sim.)
2.3V lateral depletion
~9V back surface depletion
• Pad detector – 90 * 90 columns, 55μm pitch
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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
• Guard ring currents vary:
– Highest 20μA at 10V
– Lowest 0.03μA at 50V
strip detector 4
1.0E-11
1.0E-10
1.0E-09
1.0E-08
1.0E-07
1.0E-06
1.0E-05
1.0E-04
0.0 10.0 20.0 30.0 40.0 50.0V(V)
I(A
)
Strip
Neighbours
Guard ring
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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
0
60
120
180
240
300
360
0918
27
36
45
54
microns
45-54
36-45
27-36
18-27
9-18
0-9
X-ray microbeam scan
Developments in Trento, Italy
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0 20 40 60 80 100
Vrev [V]
Idio
de
[n
A]
stc2 stc3
dtc2 dtc3
3Ddtc1 - Wafer#861
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
0 1 2 3 4Vrev [V]
Cd
iod
e [
pF
]
stc100
dtc100
stc80
dtc80
CV-diode - W861
First prototypes (p-on-n) completed, and n-on-p available soon.
Double-side Double-Column 3D detectors
Good results from preliminary electrical tests (C-V and I-V)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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
Si-n+
poly-n+
Si-p+poly-p+
SiO2
SiO2
passivation
Si(n--)
Metal
Si-n+
poly-n+
Si-p+poly-p+
SiO2
SiO2
passivation
Si(n--)
Metal
n-electrode (bias)p-electrode (readout)
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
P+
N+
2500
5000
10000
20000
30000
30
00
0
40
00
0
40
00
0
D (um)
Z(u
m)
0 10 20 30 40 50
0
10
20
30
40
50
60
70
Detail of electric field (V/cm) around top ofn-type double-sided 3D device (100V bias)
140000
Electric field, 100V bias
Variation in charge collection with depth
0.00
1.00
2.00
3.00
4.00
5.00
0 50 100 150 200 250 300
Depth (um)
Co
llec
tio
n t
ime
(ns)
50%90%Full
Time (ns) for given %collection:
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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
0.0 2.0 4.0 6.0 8.0 10.0 12.00
5
10
15
20
25
C
harg
e co
llect
ion(
ke-)
Fluence (1015neq/cm2)
Simulated CNM 3D (55m pitch)
Experimental n-on-p results
Simulated n-on-p
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)
Example of a simulated 3D structure
n+ contact
p+ contact
oxide
0.93.23*10-143.23*10-13CiOiEv+0.36Donor
0.95.0*10-145.0*10-15VVVEc-0.46Acceptor
1.6139.5*10-149.5*10-15VVEc-0.42Acceptor
η (cm-1)σh (cm2)σe (cm2)Trap
Energy (eV)Type
βe= 4.0*10-7cm2s-1, βh= 4.4*10-7cm2s-1, eqee
1
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
0.0 2.0 4.0 6.0 8.0 10.00
5
10
15
20
25
C
harg
e co
llect
ion(
ke-)
Fluence (1015neq/cm2)
Simulated strip Experimental results
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
PP Allport et al., IEEE Trans. Nucl. Sci., vol 52, Oct 2005
900V bias, 280m thick
From β values used, expect 25μm drift distance, 2ke- signal
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
C. da Via et al., Liverpool ATLAS 3D meeting, Nov. 06
Detectors produced at Stanford
0.0 2.0 4.0 6.0 8.0 10.00
5
10
15
20
25
160V
100V
60V
60V
C
harg
e co
llect
ion
(ke-
)
Fluence (1015neq/cm2)
Simulated ATLAS 3D Experimental results
At high fluences, simulated CCE ~2/3 of experimental value (like with planar detector)
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 (400m * 50m) 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 230m-thick p-type substrate, n+ readout
– Columns have 5m radius, with dopant profile extending ~2m further
– P-spray is used to isolate the columns
50m cell length
8
133m cell length
3 50m
400m
Spacing
Note larger volume occupied by columns
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
0 20 40 60 800
50
100
150
200
250
Fit:V=0.07(X-13.5m)2-1.5
76
5
4
3
8
B
ias
(V)
Electrode spacing (m)
Depletion voltage
ATLAS 3D – Depletion voltage at 1016neq/cm2
• Depletion voltage will depend on substrate material (this model matches p-type FZ, rather than oxygenated silicon)
• No. of n+ columns shown next to each data point
• Vdep proportional to depletion distance squared
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
0 20 40 60 800
50
100
150
200
250
76
5
4
3
8
B
ias
(V)
Electrode spacing (m)
Depletion voltage High field voltage
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
150V safe level
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
Y
X
Z
6.0E+18
8.8E+15
1.3E+13
-1.3E+13
-8.8E+15
-6.0E+18
5-column ATLAS 3D device1016neq/cm2, 150V bias
p-spray
n+ p+ Doping concentration(cm-3)
Y
X
Z
-10
-30
-50
-70
-90
-110
-130
-150
5-column ATLAS 3D device1016neq/cm2, 150V bias
p-spray
n+ p+Electrostatic potential(V)
Doping conc. (cm-3)
5-column ATLAS 3D, 1016neq/cm2, 150V bias
5-column ATLAS 3D, 1016neq/cm2, 150V bias
Electrostatic potential (V)
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
Y
X
Z
6.0E+18
8.8E+15
1.3E+13
-1.3E+13
-8.8E+15
-6.0E+18
5-column ATLAS 3D device1016neq/cm2, 150V bias
p-spray
n+ p+ Doping concentration(cm-3)
Y
X
Z
1.0E+14
4.1E+13
1.7E+13
7.0E+12
2.8E+12
1.0E+12
0.0E+00
5-column ATLAS 3D device1016neq/cm2, 150V bias
p-spray
n+ p+Hole concentration(cm-3)
5-column ATLAS 3D, 1016neq/cm2, 150V bias
5-column ATLAS 3D, 1016neq/cm2, 150V bias
Doping conc. (cm-3) Hole conc.
(cm-3)
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
05
1015
2025
0
5
10
15
20
25
024
6
8
10
12
14
10
12
6
9
11
p+
n+
02.04.06.08.0101214
05
1015
2025
30
0
5
10
15
20
25
0246
8
10
12
14
8
11
6
8
10
p+
n+
02.04.06.08.010.012.014.0
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.
8 columns 6 columns
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
0
510
1520
2530
3540
4550
0
5
10
15
2025
02468101214
7
84
65
p+
n+02.04.06.08.010.012.014.0 0
5
1015
2025
3035
4045
5055
6065
05
1015
2025
02468101214
5
24
64
p+
n+
02.04.06.08.010.012.014.0
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.
4 columns 3 columns
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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)
0 20 40 60 80 1000
2
4
6
8
10
12
14
2
76
5
4
3
8
C
harg
e co
llect
ion
(ke-
)
Electrode spacing (m)
ATLAS 3D CCE
Variation in collection with position larger relative to CCE
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
0 20 40 60 80 1000
100
200
300
400
500
600
2
7
6
5
4
3
8
C
apac
itanc
e (f
F)
Electrode spacing (m)
Total C per pixel Interpixel C
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.
Unlike in planar detectors, interpixel C is only a small component of total
D.Pennicard, University of 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
Noise≈60e-+39e-/100fF
“Progresses on the ATLAS pixel detector”, A. Andreazza, NIMA vol. 461, pp. 168-171, 2001
0 20 40 60 80 1000
5
10
15
20
25
30
35
40
2
76 5
4
3
8
Est
imat
ed s
igna
l-to
-noi
se r
atio
Electrode spacing (m)
ATLAS 3D SNR
Increasing C noise counteracts improving CCE
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
• 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 250m (with a 5m radius) – Double-sided 3D simulation uses 250m columns in a
300m substrate– Full-3D device used for comparison is 250m thick
• Device structure used for comparison– N+ columns used for readout, p-type substrate– 55m* 55m pixel size (Medipix)– 100V bias
p+ bias
n+ readout
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Double-sided 3D field and depletion• Where the columns overlap, (from 50m to 250m 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
30
00
0
300
00
10000
5000
2500
20000
D (m)
Z(
m)
0 25 50
0
10
20
30
40
50
60
70
19000017000015000013000011000090000700005000030000200001000050000
Double-sided 3D, p-type,1e+16neq/cm2, front surface
n+
p+
ElectricField (V/cm)
70
00
0
25000
2500
0
10000
2500
D (m)
Z(
m)
0 25 50
230
240
250
260
270
280
290
300
19000017000015000013000011000090000700005000030000200001000050000
Double-sided 3D, p-type,1e+16neq/cm2, back surface
n+
p+
ElectricField (V/cm)
A.
A.
B.
B.
Undepleted
100V 100V
1016neq/cm2, front surface 1016neq/cm2, back surface
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
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.
20% greater substrate thickness
0.0 2.0 4.0 6.0 8.0 10.00
5
10
15
20
25
Cha
rge
colle
ctio
n (k
e-)
Fluence (1015neq/cm2)
Standard 3D, 250m substrate Double-sided 3D, 250m
columns, 300m substrate
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.
10000
200
00
70000
40000
600
00
30
000
D (m)
Z(
m)
0 10 20 30 40 50
0
20
40
60
19000017000015000013000011000090000700005000030000200001000050000
n+
p+
ElectricField (V/cm)
25
00
01000
00
50
00
0
30
000
60
00
0
25
00
0
D (m)
Z(
m)
0 10 20 30 40 50
0
10
20
30
40
19000017000015000013000011000090000700005000030000200001000050000
n+ p+
ElectricField (V/cm)
Full 3D Double-sided 3D
Field reaches 2.5*105V/cm at 170V
Field reaches 2.5*105V/cm at 130V
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
P-type FZ model – proton irradiation
0.93.23*10-143.23*10-13CiOiEv+0.36Donor
0.95.0*10-145.0*10-15VVVEc-0.46Acceptor
1.6139.5*10-149.5*10-15VVEc-0.42Acceptor
η (cm-1)σh (cm2)σe (cm2)Trap
Energy (eV)Type
• 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
• 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
e
nt
n
e
ethe veqe
e
1
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Comparison with experiment
P-type trap models: Depletion voltages
300
350
400
450
500
550
600
0 1E+14 2E+14 3E+14 4E+14 5E+14 6E+14 7E+14
Fluence (Neq/cm2)
Dep
leti
on
vo
ltag
e (V
)
Default p-type sim
Modified p-type sim
Experimental
“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 α=3.75*10-17A/cm
• 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
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 1E+15 2E+15 3E+15 4E+15 5E+15 6E+15
Fluence (neq/cm^2)
Lea
kag
e cu
rren
t (A
/cm
^3)
Experimentally,α=3.99*10-17A/cm3 after 80 mins anneal at 60˚C (M. Moll thesis)
α=5.13*10-17A/cm
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Example of CCE with varying bias
Collection vs bias in 5-column ATLAS
0
2
4
6
8
10
12
14
16
0 20 40 60 80 100 120 140 160 180
Bias (V)
Ch
arg
e co
llect
ed (
ke-)
10^16neq/cm^2
5*10^15neq/cm^2
Vdep
Vdep
• CCE curves show a smaller gradient after depletion voltage is reached
CCE increases beyond Vdep, due to increasing carrier velocity
D.Pennicard, University of Glasgow, INSTR08, Novosibirsk
Electric field distribution – 8 columns per pixel
110000
50000
10000
70000
1000
0
4000
0
40000
X (m)
Y(
m)
0 5 10 15 20 25 300
5
10
15
20
25
19000017000015000013000011000090000700005000030000200001000050000
ATLAS 3D, p-type, 50m cell, 8 column1e+16neq/cm2, 150V bias
n+
p+
ElectricField(V/cm)
• 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