development of enhanced double-sided 3d radiation sensors...
TRANSCRIPT
Development of enhanced double-sided 3D
radiation sensors for pixel detector
upgrades at HL-LHC
Marco Povoli1
1Dipartimento di Ingegneria e Scienza dell’Informazione
University of Trento, Italy
Trento, Italy - February 20, 2012
[Work supported by INFN CSN V, projects "TREDI" (2005-2008) and "TRIDEAS" (2009-2011)]
Outline
High Energy Physics experiments at LHC
The Large Hadron Collider
The ATLAS experiment
The current ATLAS silicon tracker
The ATLAS Insertable B-Layer (IBL)
Radiation damage in silicon
Summary
Countermeasures
3D detectors
Originally proposed architecture
The ATLAS 3D sensor collaboration
Enhanced 3D technology at FBK
Design choices and motivations
Detailed electrical characterization
Functional characterization
Possible technological improvements
Additional applications
2 / 46 Marco Povoli
The Large Hadron Collider (LHC)
The Large Hadron Collider (LHC)
• Largest particle collider ever built
• Near Geneva, underneath the
swiss/french border
• Total ring length of about 27 Km
• First started in 2008
• The official physics program
started in 2009
Basic machine parameters
• Proton or Lead Ion collisions
• Nominal proton energy of 7 TeV
(per beam)
• Bunch spacing of 25 ns
• Design luminosity 1034 cm−2s−1
Physics program
• Discovery of new particles/theories
• Particles collide inside the four experiments
• ATLAS and CMS: general purpose
• ALICE: study lead ions collision
• LHCb: specialized in b-physics
3 / 46 Marco Povoli
General purpose experiments
A Toroidal LHC ApparatuS (ATLAS)
Basic parameters
• 45 meters long
• 25 meters in diameter
• weights about 7000 tons
Structure (inside-out)
1. Inner detector
2. Calorimeters
3. 2 Tesla solenoid magnets
4. Muon spectrometers
The inner detector
• Several layers of silicon detectors and one layer of straw tube detectors
• Needed to reconstruct the particle interaction point
• Required to be very fast
• Operates in extremely harsh conditions
4 / 46 Marco Povoli
The current ATLAS silicon tracker
Barrel cross-section
• Total radius of roughly 1 m
• Pixel detector: 3 layers of n-in-n pixel sensors
• Strip detector: 4 layers of p-in-n strip sensors
• Transition Radiation Tracker: straw tubes interleaved
with scintillating fibers
The pixel detector
• Three barrel layers at radiuses 50.5, 88.5, 122.5 mm
• 6 end-caps (three on each side)
• Pixel size 50×400 µm2
• Covers an area of ∼1.7 m2
• Approximately 67 million channels
• Designed to withstand a fluence of
1×1015 1 MeV neqcm−2
5 / 46 Marco Povoli
The ATLAS Insertable B-Layer (IBL)
Planned installation during the first long shutdown (2013-2014)
CURRENT PIXEL DETECTOR
RENDERING OF THE IBL
Addition of a fourth pixel layer close to
the beam pipe
Motivations
• Maintain the event pile-up under
control as LHC luminosity increases
• Add redundancy to recover partial
failure of modules in the other pixel
layers
• Increase tracking and reconstruction
accuracy
Main design parameters
• Need to reduce the beam pipe radius
by 4mm
• Placed at 33.25 mm from the center of
the beam pipe
• Will need to withstand a fluence of
5×1015 neq cm−2
[M. Capeans, (The ATLAS Collaboration), ATLAS-TDR-019]
6 / 46 Marco Povoli
The ATLAS Insertable B-Layer (IBL)
Sensor requirements
Parameter Value unit
Total number of staves 14 -
Pixel size (Φ, z) 50, 250 µm
Dead edge extension 200 µm
Sensor thickness <250 µm
NIEL dose tolerance 5×1015 neq /cm2
Hit efficiency in active area >97% -
Operating bias voltage <1000 V
Operating temperature -15 ◦C
• Reduced pixel size in the z direction
(250 µm) to increase the spatial
resolution
• No tilt possible in the z direction →
need for reduced dead area at the
edges
[A. Clark, et al., (The ATLAS IBL collaboration),
(2012) JINST 7 P11010]
NEED FOR ADVANCED RADIATION HARD DETECTORS
7 / 46 Marco Povoli
Radiation damage in silicon (summary)
Bulk damage
• Due to Non Ionizing Energy Loss (NIEL)
• Displacement of atoms in the Silicon lattice
• Built-up of crystal defects
Consequences
• Change in effective doping concentration
(higher depletion, under-depletion)
• Increase of leakage current
(shot noise, thermal runaway)
• Increase of charge trapping
(charge losses)
Surface damage
• Due to Ionizing Energy Loss (IEL)
• Radiation generates carriers in SiO2
• Electrons can escape while holes get trapped at the
Si/SiO2 interface
Consequences
• Accumulation of charge at the SiO2 /Si interface
(inter-pixel capacitance and isolation and
breakdown behavior)
• Increased charge trapping at the Si/SiO2
• Increased surface recombination velocity
LOWERING OF THE S/N RATIO![Michael Moll - MC-PAD Network Training, Ljubljana, 27.9.2010]
8 / 46 Marco Povoli
Countermeasures
Material engineering
Deliberate incorporation of impurities or defects into the
silicon bulk to improve radiation tolerance of the
detectors
• Oxygen rich Silicon (FZ, DOFZ, Cz, MCZ)
• Pre-irradiated Silicon
New Materials
• Silicon carbide (SiC)
• Amorphous Silicon
• Diamond
Surface isolation
• Important to assure inter-electrode isolation
• Typically achieved using p-spray, p-stop or a
combination of the two
Device engineering
• p-type silicon detectors (n-in-p)
• Thin detectors
• 3D detectors
[Michael Moll - MC-PAD Network Training, Ljubljana, 27.9.2010]
9 / 46 Marco Povoli
Full 3D detectors
Original idea - PROS and CONS
Features of 3D detectors
• First proposed by S. Parker and
collaborators in the mid ’90s
[NIMA 395 (1997), 328]
• Decouple the active volume from the
inter-electrode distance
• Low full depletion voltage (<10 V)
• Short collection distances (∼50 µm)
• Low trapping probability after
irradiation
• Small dead area along the edges
Disadvantages of 3D detectors
• Columns are partially dead regions
• Non uniform response (low field
regions are present)
• Higher capacitance (higher noise)
• Fabrication process complex and more
expensive
10 / 46 Marco Povoli
The ATLAS 3D sensor collaboration
Institutes and processing facilities
• 18 Institutes
• 4 processing facilities:
◮ SNF (Stanford, USA)◮ SINTEF (Oslo, Norway)◮ CNM (Barcelona, Spain)◮ FBK (Trento, Italy)
Available 3D technologies
• Full 3D with active-edges
(SNF and SINTEF)
• 3D-DDTC with slim-edges (FBK, CNM)
• Full 3D-DDTC with slim-edges (FBK)
Main targets
1. Speed-up the test and industrialization of 3D silicon sensors
2. Production and testing of 3D sensors for the IBL
[C. Da Vià, et al., NIMA694 (2012), 321]
11 / 46 Marco Povoli
3D technology for the IBL developed at FBK
Main geometrical features
• Double-type column approach
• Fully double-sided
• Columns etched from both wafer sides
• Fully passing through columns
• Empty electrodes (no polysilicon filling)
• Surface isolation by means on p-spray
implantations on both wafer sides
SEM cross-section
• Very good etching uniformity
• Columns are all passing through
• Slight shrinking of the column tip (not
affecting device behavior)
[G. Giacomini, et al., to appear in IEEE TNS (2013)]
12 / 46 Marco Povoli
3D technology for the IBL developed at FBK
The common wafer layout and pixel layout
Common wafer layout
• 4 inches wafers
• 8 ATLAS FE-I4 pixel detectors
• 9 FE-I3 pixel detectors
• 3 CMS pixel detectors
• 4 strip detectors (80µm pitch)
• Several planar and 3D test structures
−→ z direction
Single pixel layout
• Pixel size: 50×250 µm2
• 2E configuration: 2 n+ columns per pixel
• Inter-electrode distance (d): ∼67 µm
• With field-plate
[C. Da Vià, et al., NIMA694 (2012), 321]
13 / 46 Marco Povoli
Motivation for FE-I4 pixel layout
Results from previous technologies
!
Previous 3D-DDTC technology at FBK
• Non-passing through columns
• Pixel size 50×400 µm2 (FE-I3)
• Three pixel layouts (2E, 3E, 4E)
Best performances from 3E devices (71 µm)
• Noise ∼205 e−
• Good CCE up to 1×1015 neq /cm2
• Tracking efficiency >98% at 1×1015 neq /cm2
Φeq =1×1015 neq /cm2
[A. Micelli, NIMA650 (2011), 150]
Φeq =1×1015 neq /cm2
0
2
4
6
8
10
12
14
16
18
20
0 20 40 60 80 100 120 140
Ch
arg
e [
ke
- ]
Reverse voltage [V]
2E - simulated2E - measured3E - simulated3E - measured4E - simulated4E - measured
[A. La Rosa, NIMA681 (2012), 25]
14 / 46 Marco Povoli
Edge termination (SLIM-EDGE)
Motivation and design
Slim-edges in the z direction
• Requirement: 6200 µm in the beam (z) direction
• Motivation: not possible to tilt modules in the z
direction due to space constrains
• Fence of ohmic columns to prevent the depletion
region to reach the scribe line
• Designed with the aid of numerical simulations
Computer aided design
• Simulation of a structure including the last junction
column and the ohmic fence
• Simulation domain highlighted with the dashed
rectangle
• Scribe line model with a low lifetime region (<1 ns)
• Monitor the current of the last junction column
• No avalanche models
15 / 46 Marco Povoli
Edge termination (SLIM-EDGE)
Motivation and design
[M. Povoli, et al.,
JINST 7 (2012) C01015]
5
10
15
20
25
30
35
40
45
0 100 200 300 400 500
Cu
rren
t [p
A/c
olu
mn
]
Reverse voltage [V]
Nd=1.0x1011
cm-3
Nd=2.5x1011
cm-3
Nd=5.0x1011
cm-3
Nd=1.0x1012
cm-3
[G.-F. Dalla Betta, et al., NSS10 Conf. Record, pp. 382-387]
Simulation results
• Different bulk doping concentrations tested
• No signs of current increase up to 500 V (well above
expected operation voltage)
• The depletion region extends outside the active area by
about 75 µm at 300 V
• Safe device operation with a 200 µm slim-edge
• Note: conservative design!
16 / 46 Marco Povoli
On wafer selection
Motivation and procedure
BUMP-BONDING
1.
2.
3.
4.
[G. Giacomini, et al., to appear in IEEE TNS (2013)]
On-wafer sensor selection
• The bump-bonding is complex and very expensive
• Assemble only good sensor tiles
• Wafer with more than 3 good sensors are sent for bump-bonding
Temporary metal layer
• Deposited on top of the frontside passivation
• Strip-like metalization
• 80 strips connecting 336 pixels each
• Automatic current measurements
• The sum of all strip currents gives the total sensor current
17 / 46 Marco Povoli
On wafer selection
Example test results
GOOD WAFER (ATLAS12)
100
101
102
103
104
105
106
107
0 10 20 30 40 50 60 70 80
Cu
rren
ts [
nA
]
Reverse voltage [V]
W2
S1S2S3S4S5S6S7S8
100
101
102
103
104
0 10 20 30 40 50 60 70 80
Cu
rren
ts [
nA
]
Reverse voltage [V]
ATLAS09 - W14-S3
On waferFlip-chipped
[C. Da Vià, et al., NIMA694 (2012), 321]
Parameter Symbol Value
Operation temperature Top 20-24◦C
Depletion voltage Vdepl <15 V
Operation voltage Vop >Vdepl +10 V
Leakage current at Vop I(Vop) <2 µA
Breakdown voltage Vbd >25 V
Current "slope" I(Vop )/I(Vop -5V) <2
Wafer/sensor selection
• Good sensors always have currents much lower than the
set limit
• Breakdown voltages are typically higher than 30 V for
good detectors
• Confirmation of the selection method comparing current
pre/after bump-bonding
• Yield of IBL production at FBK: 56.82%
NOTE: further investigation needed on some aspects
• The breakdown is lower than for standard planar detectors
(can be critical after irradiation)
• In some cases the behavior is not very uniform
• Necessary to perform a thorough electrical
characterization
18 / 46 Marco Povoli
Detailed electrical characterization
3D diodes
FE-I4 with field-plate FE-I4 80 µm pitchCMS - 1E
3D diode
• Two terminal device having an area of roughly 10 mm2
• Electrodes of the same type are shorted together
• All the geometries of larger detectors are reproduced
• Eases the characterization
Performed tests
• I-V and C-V measurements as a function of the temperature
• Numerical simulations to confirm the findings and gain a deeper understanding of
the device behavior
[M. Povoli, et al., NIMA699, (2013), 22]
FE-I4 (backside)
19 / 46 Marco Povoli
IV measurements with variable temperature
IV Curves
0
1e-08
2e-08
3e-08
4e-08
5e-08
6e-08
7e-08
8e-08
0 10 20 30 40 50 60 70 80
Cu
rren
ts [
A]
Reverse voltage [V]
I-V 80BIG
35°C30°C25°C20°C15°C10°C
5°C0°C
-5°C-10°C-15°C-20°C
0
1e-08
2e-08
3e-08
4e-08
5e-08
6e-08
7e-08
8e-08
0 10 20 30 40 50 60 70 80
Cu
rren
ts [
A]
Reverse voltage [V]
I-V FEI4
35°C30°C25°C20°C15°C10°C
5°C0°C
-5°C-10°C-15°C-20°C
0
1e-08
2e-08
3e-08
4e-08
5e-08
6e-08
7e-08
8e-08
0 10 20 30 40 50 60 70 80
Cu
rren
ts [
A]
Reverse voltage [V]
I-V CMS
35°C30°C25°C20°C15°C10°C
5°C0°C
-5°C-10°C-15°C-20°C
0
1e-08
2e-08
3e-08
4e-08
5e-08
6e-08
7e-08
8e-08
0 10 20 30 40 50 60 70 80
Cu
rren
ts [
A]
Reverse voltage [V]
I-V FEI4-FP
35°C30°C25°C20°C15°C10°C
5°C0°C
-5°C-10°C-15°C-20°C
Setup
• Devices coming from wafer
W20 of the ATLAS09 batch
• Devices were diced and wire
bonded on small PCBs
• Wire bonding contribution is
negligible
• Temperature variation
between −20◦C and 35◦C
inside a climatic chamber
• Measurements performed
with HP4145
Preliminary results
• Each type of device has its own characteristic behavior
• Breakdown voltages between 40 and 50V
• Different current slope for different devices
• More details in the next slide...
20 / 46 Marco Povoli
Data analysis
Intrinsic electric behavior
Breakdown voltages (all devices)
35
40
45
50
55
60
-20 -10 0 10 20 30 40
Bre
akd
ow
n v
olt
ag
e [
V]
Temperature [°C]
UVBD= 48.50 mV/°C
UVBD= 76.82 mV/°C
UVBD= 50.72 mV/°C
UVBD= 55.42 mV/°C
80BIGFEI4-FP
CMSFEI4
Breakdown voltages
• Breakdown between 40 and 50V
• Linear increase with temperature
• The increase is between ∼ 50 and
∼ 80mV/◦C
• In agreement with the expectation
[Crowell, C. R. and S. M. Sze, Appl. Phys. Lett. 9, 6 (1966)
242-244.]
1e-11
1e-10
1e-09
1e-08
1e-07
3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4
Cu
rren
ts [
A]
1000/T [°K-1
]
Arrhenius plot - FEI4-FP
Meas. at 10VCalc. at 10V
Meas. at 30VCalc. at 30V
Purpose
Distinction between thermal and avalanche
generation
Equation
I(T) = I(TR)(
TTR
)2
exp[
E2kB
(
1TR
− 1T
)]
Results
• TR=293.15 ◦K
• Very good agreement at low biases
• The agreement is lost as breakdown
approaches21 / 46 Marco Povoli
Investigation through numerical simulations
Simulated structure
• A quarter of elementary cell (thanks to symmetry)
• Bulk doping: 7×1011 cm−3 (p-type, measured)
• One columnar electrode per type
• Measured p-spray profiles
• Measured n+ and p+ surface implantations
• Device layout fully reproduced
Incremental addition of the layout details
• Used to estimate the contribution of each component of
the device capacitance
• Allows discriminate between inter-electrode and surface
capacitance
Simulation of the full structure
• Estimation of the expected breakdown voltage and
current levels
• Analysis of the distribution of electrical quantities (e.g.
Electric field and Electrostatic potential)
22 / 46 Marco Povoli
C-V measurements vs. C-V simulations
FE-I4 diode with field-plate
0
50
100
150
200
250
300
350
400
450
500
0 10 20 30 40 50
Cap
acit
an
ce [
pF
]
Reverse voltage [V]
Columnsp-spray
P+ implantation
N+ implantation
FullMeasured
Results
• Electrodes contribution: ∼ 51.4pF (constant)
• p-spray causes an increase of ∼ 30pF (basically constant)
• P+ implantation and metal on the back side do not cause much increase
• N+ implantation and metal on the front side cause an increase of ∼ 63pF at a bias voltage of 20V
• At higher biases the contribution of front side saturates to a value similar to the one obtained only with electrodes
and p-spray (∼ 103pF)
• Measured capacitance does not fully saturate at 40V (instrument limitations)
23 / 46 Marco Povoli
Distribution of electrical quantities
FE-I4 diode - Electric field
FE-I4 with FP FE-I4
1
1.5
2
2.5
3
0 5 10 15 20 25 30 35
Ele
ctr
ic f
ield
[10
5 V
/cm
]
Distance [µm]
N+
column
Backside
peaks
Frontside
peaks
Field-plate
Vbias=40V
• Large field peaks on both the upper
and lower surfaces
• Peaks are placed at the n+ to
p-spray junction
• Particularly critical due to the high
dose p-spray implantation
• Both structures have similar
field-peaks on the backside
• The field-plate redistributes and
lowers the field on the frontside
24 / 46 Marco Povoli
Front-side surface irradiation
X-Rays - 2 Mrad (60 minutes irradiation)
P-spray compensation to increase the breakdown voltage
How large is the increase?
91
92
93
94
95
96
97
-300 -200 -100 0
Cap
acit
an
ce [
pF
]
Gate voltage [V]
Qox=1.8x1011
cm-2
Qox=1.2x1012
cm-2
Qox=3.9x1012
cm-2
12
PRE
POST (un-biased)POST (Vgate=20V)
• Irradiation performed at "Laboratori Nazionali di
Legnaro" (LNL, Padova, Italy, thanks to Serena
Mattiazzo)
• Use of MOS capacitors to monitor the increase
in oxide charge
• Two irradiations: without and with bias (20 V)
• Considerably larger charge when irradiation is
performed under bias
0
100
200
300
400
500
600
700
800
0 10 20 30 40 50 60 70 80 90
Cu
rren
t [n
A]
Reverse voltage [V]
1. UQox
2. Surfacerecomb.
3. VBD
FE-I4 #1 - PRE
FE-I4 #1 - POST (unbiased)
FE-I4 #2 - PRE
FE-I4 #2 - POST (biased)
• Two different FE-I4 diodes with field-plate
• Current increase do to surface generation
• Limited breakdown increase
• Pre-irradiation trend maintained after
• Confirm that the breakdown occurs on the
backside
25 / 46 Marco Povoli
Functional characterization of FE-I4 pixel detectors
Readout chip and measurement setups
[A. Clark, et al., (The ATLAS IBL collaboration),
(2012) JINST 7 P11010]
ATLAS FE-I4 readout chip
• Designed to withstand a TID of 250 Mrad
• Leakage compensation
• Double stage charge amplifier with constant current
discharge
• Discriminator after charge amplifier
• Operates in Time Over Threshold (ToT) mode
• The ToT is representative of the collected charge
The USBPix system[http://icwiki.physik.uni-
bonn.de/twiki/bin/view/Systems/UsbPix]
The EUDET Telescope[D. Haas, EUDET-Report-2007-07]
26 / 46 Marco Povoli
Functional characterization of FE-I4 pixel detectors
Radioactive source scans (90Sr, Lab)
5
6
7
8
9
10
11
0 5 10 15 20 25 30 35 40 45 50 55
To
T (
MP
V)
Bias voltage (V)
90Sr - Pre-irradiation lab tests
FBK13 - un-irradiated
FBK13 - Simulation
CNM101 - un-irradiated
PRE-irradiation
• Comparison between FBK and CNM detectors
• Calibration: 10 ToT at 20 ke−
• Most Probable Value of the "all-cluster" charge
distribution
• FBK → charge saturation before 10 V
• CNM → charge saturation at roughly 25 V
• Numerical simulations (dashed line) confirm the
measurement results
0
1
2
3
4
5
6
7
8
9
10
0 25 50 75 100 125 150 175 200 225 250 275 300
To
T (
MP
V)
Bias voltage (V)
90Sr - Post-irradiation lab tests
pre-irrad
FBK13 - un-irradiated
CNM101 - un-irradiated
FBK87 - p-irrad - 5x1015
neq/cm2
FBK - p-irrad - 2x1015
- SIMULATED
FBK - p-irrad - 5x1015
- SIMULATED
CNM100 - p-irrad - 2x1015
neq/cm2
CNM36 - p-irrad - 6x1015
neq/cm2
POST-irradiation
• Lowering of charge collection due to trapping
• FBK detector irradiated at 5×1015 neq /cm2 (red)
shows a hint of charge saturation at roughly 150 V
• Confirmed by numerical simulations (red dashed
line)
• Measurement not available for 2×1015 neq /cm2 but
simulations can be trusted (violet dashed line)
• Satisfactory performances
[G.-F. Dalla Betta, et al., VERTEX2012, submitted to POS]
27 / 46 Marco Povoli
Functional characterization of FE-I4 pixel detectors
Test-beam results (tracking efficiency)
[P. Grenier, et al., "IBL TestBeam Results", presented at the
IBL Sensor Review, CERN, 4-5 July 2011]
Beam tests during 2011-2012
• CERN - SPS: 120 GeV pions
• DESY: 4 GeV positrons
• EUDET Telescope
• Both un-irradiated and irradiated samples
• IBL operating conditions
• Planar sensor always used as reference
(b) PRE-irrad. efficiency map
• Good tracking efficiency (98.8%)
• Electrodes appear as less efficient regions
• Possible to obtain higher efficiency by tilting the
device with respect to the beam
(c,d) POST-irrad. efficiency map
• FBK90 (2×1015 neq /cm2) shows great efficiency
(99.2%) at 60 V of bias when tilted by 15◦
• FBK87 (5×1015 neq /cm2) exhibits not sufficient
efficiency (95.6%) at the chosen bias (140 V)
• NOTE: when the proper bias is used (e.g. 160 V)
the efficiency is back within IBL requirements
(98.2%)
28 / 46 Marco Povoli
Proton irradiated 3D diodes
80µm pitch
0
10
20
30
40
50
60
70
80
90
100
0 25 50 75 100 125 150 175 200 225 250
Cu
rren
ts [µ
A]
Reverse voltage [V]
{eq=2.1x1015
neq/cm2
-20°C-15°C-10°C-5°C0°C5°C
10°C15°C20°C
FE-I4
0
10
20
30
40
50
60
70
80
90
100
0 25 50 75 100 125 150 175 200 225 250
Cu
rren
ts [µ
A]
Reverse voltage [V]
{eq=2.1x1015
neq/cm2
-20°C-15°C-10°C-5°C0°C5°C
10°C15°C20°C
-50
-25
0
25
50
75
100
125
150
175
-30 -20 -10 0 10 20 30
Bre
akd
ow
n v
olt
ag
e [
V]
Temperature [°C]
80µm - {eq=2.1x1015
neq/cm2
FE-I4 - {eq=2.1x1015
neq/cm2
80µm - {eq=5.2x1014
neq/cm2
CMS - {eq=5.2x1014
neq/cm2
CMS - {eq=2.4x1014
neq/cm2
• Irradiation performed at Los Alamos with 800 MeV protons
(thanks to Martin Hoeferkamp)
• Only half of the requested fluence was delivered
• Increase in breakdown between few volts and ∼100 V
Important!
• Devices were selected prior to irradiation
• The FE-I4 diode irradiated at 2×1015 neq /cm2 shows a
breakdown voltage of roughly 125 V
• A proper sensor selection will assure optimal
operating voltages after irradiation!
29 / 46 Marco Povoli
SLIM-EDGE characterization
Electrical tests
Performed test
• Several cuts performed by means of a
diamond saw
• Each cut is closer to the active area
• I-V measurement after each cut
[M. Povoli, et al., JINST 7 (2012) C01015]
10-9
10-8
10-7
10-6
10-5
10-4
10-3
0 10 20 30 40 50 60 70 80 90 100
Cu
rren
ts [
A]
Reverse voltage [V]
scribe linecut #1cut #2cut #3cut #4cut #5cut #6
Results
• Intrinsic device behavior is equal to
roughly 60 V
• No increase in reverse current up to
the fourth cut
• Possible to reduce the total edge
extension to roughly 100 µm
• NOTE: critical only before irradiation
30 / 46 Marco Povoli
SLIM-EDGE characterization
Functional tests (FE-I4 diode)
Laser scan Vb=15 V
Layout
Numerical simulation
[M. Povoli, et al., JINST 7 (2012) C01015]
Edge efficiency
(Test beam data, 2 × 1015 neq/cm2)
[G.-F. Dalla Betta, et al., VERTEX2012, submitted to POS]
Laser scan (Lab)
• Laser: λ=1060 nm
• Readout: CSA + 20 ns shaper
• Very good agreement with simulations
Edge efficiency after irradiation (test beam)
• Full efficiency inside the active-area
• Roughly 25 µm of the slim-edge are active
31 / 46 Marco Povoli
Possible technological improvements
Investigation through numerical simulations
• The high field region on the backside can be critical!
• It is of paramount importance to use large operating
voltages!
• Lifting the n+ column tip will eliminate the critical
region on the backside
• At the same time the fabrication will be easier and
faster
[M. Povoli, et al., IEEE NSS12 Conf. Record, pp. 1334-1338]
• Important to also improve the field distribution on
the frontside
• The field-plate is important
• Some of the designed structures include a floating
n+ ring which is intended to interrupt the
electrostatic potential of the p-spray
• The design is performed by means of numerical
simulations
32 / 46 Marco Povoli
Simulation results
5
10
15
20
25
30
35
40
45
50
55
60
15 20 25 30 35 40 45 50-70
-60
-50
-40
-30
-20
-10
0
10
20
Efi
eld
[V
/µm
]
Ep
ot [V
]
distance [µm]
Vbias= -65V
N+
N+ ring
field-plate
Efield - NO RINGEpot - NO RING
Efield - RINGEpot - RING
Effect of the floating ring
• The potential of the inner p-spray is lower
• The field at the main junction is lower and the
field-plate works properly
• Large peak on the outer ring junction → causes
breakdown
• The ring placement is critical due to space
constrains
Simulated I-Vs
• All devices show larger breakdown than in the
previous technology
• The floating ring limits the performances but could
act as additional shielding from surface currents
• Raising the n+ column should deliver breakdown
voltages larger than 100 V
0
0.5
1
1.5
2
2.5
3
3.5
4
0 20 40 60 80 100 120 140 160 180
Cu
rre
nts
[n
A]
Reverse voltage [V]
DTC-4breakdown
RINGNORING
80µm
80µmFE-I4FP
FE-I4, N-ringCMS1ECMS2ECMS3ECMS4E
CMS1E, N-ringCMS2E, N-ringCMS3E, N-ringCMS4E, N-ring
33 / 46 Marco Povoli
New SLIM-EDGE implementation
Pixel detectors slim-edge
• Reduced by 50 µm following the indications obtain from
previous tests
• Conservative design to avoid problems
New slim-edge implementation for some 3D diodes
• Double row of short trenches (mimicking the active-edge)
• Dead area of roughly 50µm
• Maintains the mechanical integrity and does not require support
wafer
• Simulation results at 50 V of bias show how this solutions does
not allow the depletion region to reach the scribe line
[G.-F. Dalla Betta, et al., NSS11 Conf. Record, pp. 1334-1340]
34 / 46 Marco Povoli
Fabricated devices
Preliminary electrical characterization
Available devices
• 4 FE-I4 pixel detectors (2 versions)
• 26 CMS pixel detectors (8 versions)
• 2 MEDIPIX-II detectors
• 3D diodes in several different flavors
0
5
10
15
20
25
30
35
40
45
50
0 25 50 75 100 125 150 175 200
Cu
rren
ts [
nA
]
Reverse voltage [V]
FE-I4 diodesMEASURED
W2
W8
Preliminary results
• Batch completed in mid October 2012
• A few IV measurements on 3D diodes
here reported
• Leakage current higher than expected
but acceptable
• Sizable increase of breakdown voltage
[M. Povoli, et al., IEEE NSS12 Conf. Record, pp. 1334-1338]
35 / 46 Marco Povoli
The ATLAS Forward Physics (AFP)
Motivation and requirements
• Measure protons scattered from the collision
• Located at roughly 200 m both upstreams and
downstreams
• Requires reduced edge extension
• FE-I4 modules are investigated
[The ATLAS Collaboration, CERN-LHCC-2011-012]
Performed tests
• The edge of interest is the one not "IBL-like"
• Same cut and measure tests were performed
• Proper operation up to the 6th cut (75 µm edge)
Aspect to investigate...
• Very un-uniform irradiation
• Tests are being performed on CNM devices
[S. Grinstein, RESMDD12, submitted to NIMA]
0
1
2
3
4
5
6
7
8
9
10
0 10 20 30 40 50 60 70
Cu
rre
nts
[n
A]
Reverse voltage [V]
strip-75
cut-1cut-2cut-3cut-4cut-5cut-6cut-7cut-8
36 / 46 Marco Povoli
Planar detectors with active-edges
Standard vs. Active Edge detectors
Standard detectors
• In standard detectors a dead border
region must be present
• In a good design cracks and damages
on the edges should be at least at a
few hundreds of micrometers away
from the depleted region
• Total dead region a + d > 500µm
How to limit dead region?
• Cut lines not sawed but etched with
Deep Reactive Ion Etching (DRIE) and
doped
[C. Kenney, et al., IEEE TNS 48-6 (2001) 2405]
Problems
• Process is more complicated
• Need for support wafer
• Finding the correct "d" to limit early
break-down phenomena[M. Povoli, NIMA658 (2011), 103]
37 / 46 Marco Povoli
Planar detectors with active-edges
Device and wafer layout
Single-sided p-in-n devices
(support wafer)
General device layout (test diode)
1. Distance between n+ and
p+ doping (GAP)
2. Field-plate
3. Bias pad (connected to the
doped trench)
4. Floating p+ ring
Trench etching
• Designed to be 4 µm
• Not well defined at first
• Optimized etching in the
second part of the batch
(roughly 10 µm width)
• Partial polysilicon filling
needed to restore the
surface planarity
Wafer layout
• Strip detectors with
inter-strip pitches of 50, 80
and 100 µm (AC or DC
coupled)
• Pixel detectors compatible
with the readout chips of
the ALICE experiment
• Several test diodes in many
different flavors
• Standard planar test
structures
38 / 46 Marco Povoli
Planar detectors with active-edges
Electrical and functional characterization
0
10
20
30
40
50
0 100 200 300 400 500 600
Revers
e c
urr
en
t [n
A]
Reverse voltage [V]
GAP=10 µmGAP=15 µmGAP=20 µmGAP=25 µmGAP=30 µmGAP=35 µmGAP=40 µmGAP=50 µmGAP=70 µmGAP=80 µmGAP=90 µm
GAP=100 µmGAP=120 µmGAP=150 µm
100
200
300
400
500
0 20 40 60 80 100 120 140 160
Bre
akd
ow
n v
olt
ag
e [
V]
Gap dimension [µm]
no FPFP=5 µm
[G.-F. Dalla Betta, et al., NSS11 Conf. Record, pp. 1334-1340]
Electrical characterization (I-V)
• Good reverse current values and uniformity
• Clear trend with GAP size
• Once again the field-plate proves its
effectiveness
Functional tests
• Bi-dimensional X-Ray scan performed at
Diamond Light Source, Didcot, UK.
• 15 keV X-rays, spot size ∼3 µm FWHM
• Very good signal efficiency up to less than
20 µm away from the edge
x [mm]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05
x [mm]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05
x [mm]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05
x [mm]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05
x [mm]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05
x [mm]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05
Sig
na
l [a
.u.]
x [mm]
D2, GAP=15µm D3, GAP=20µm
D2 - Vb=100V
D2 - Vb=20V
D2 - Vb=5V
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05 4.06 4.07 4.08 4.09 4.1
Sig
na
l [a
.u.]
x [mm]
D2, GAP=15µm D3, GAP=20µm
D3 - Vb=100V
D3 - Vb=20V
D3 - Vb=5V
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
4 4.01 4.02 4.03 4.04 4.05 4.06 4.07 4.08 4.09 4.1
[M. Povoli, et al., NIMA (2012)
http://dx.doi.org/10.1016/j.nima.2012.09.035]
39 / 46 Marco Povoli
Thin 3D detectors with built-in charge multiplication
Evidence and exploitation of this effect
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 25 50 75 100 125 150 175 200 225 250 275
Co
llecte
d c
harg
e [
fC]
Reverse voltage [V]
{=5e14 cm-2
- Simulated {=1e15 cm
-2 - Simulated
{=2e15 cm-2
- Simulated {=5e14 cm
-2 - Measured
{=1e15 cm-2
- Measured {=2e15 cm
-2 - Measured
Charge multiplication (CM)
• Evidence of CM was found in irradiated planar and
3D detectors
• CM was observed in older generations of both FBK
and CNM 3D detectors
[A. Zoboli, et al., NSS08 Conf. Record, 2721]
[M. Köhler, et al., NIMA659 (2011), 272]
• Triggered by high electric field at the tip of the
junction columns
• Confirmed by numerical simulations
[G. Giacomini, et al., VERTEX2011, POS]
• Can charge multiplication be exploited?!?
Motivation
• Increasing interest in reducing the total material
budget
• Reduction of the sensor thickness
• Lower thickness → less detection volume
• Reduction in available charge for particle detection
Idea and investigation through numerical simulations
• A shrinking of all geometries by a factor of ∼3 will
allow to also reduce inter-electrode spacing and
column diameter
• Bulk thickness: ∼70 µm
• Column diameter: ∼4 µm
• Higher field at lower voltages
[M. Povoli, et al., submitted to NIMA (2013)]
40 / 46 Marco Povoli
Thin 3D detectors with built-in charge multiplication
Simulation results
0
1
2
3
4
5
6
7
0 20 40 60 80 100 120 140 160 180 200
Gain
Reverse voltage [V]
1N-1P1N-2P1N-3P
1N-SqP1N-RoP
1N-HexP
Gain before irradiation
• All investigated structures show CM
• The onset of CM changes with
geometry
• Lower operating voltages for structures
having trench ohmic electrodes
• A good gain uniformity was found
within the entire investigated cell
0
1
2
3
4
5
6
7
0 10 20 30 40 50 60 70 80 90 100 110 120
Gain
Reverse voltage [V]
1N-SqP
PRE-irradiation
{eq=1x1015
neqcm-2
{eq=5x1015
neqcm-2
{eq=1x1016
neqcm-2
{eq=2x1016
neqcm-2
Gain after HEAVY irradiation
• Results reported only for the
rectangular cell
• Bulk radiation damage modeled with a
3 level trap model
• Reduction of the collected charge due
to trapping (expected)
• No change in CM onset voltage
• Completely recover charge trapping
41 / 46 Marco Povoli
Thin 3D detectors with built-in charge multiplication
Surface isolation and electrode efficiency
0
1
2
3
4
5
0 2.5 5 7.5 10 12.5 15
Efi
eld
[V
/µm
]
X [µm]
Field-plate
n+
p+
Vbias=110V
TOP (Y=0.01 µm)MIDDLE (Y=27.5 µm)
TIP (Y=55 µm)
Final sensor geometry
• Rectangular cell shape
• P-spray and ∼4 µm field-plate
• Raise the n+ column to avoid critical regions on the
backside
• Modification of the tip shape in order to avoid early
breakdown phenomena
• Extracted 1D electric field profiles show that the
surface and tip field are under control
• Possible to operate in CM mode
0
0.5
1
1.5
2
2.5
3
0 20 40 60 80 100
Gain
Reverse voltage [V]
Tp=10ns
silicon
n+ electrode
p+ electrode
Electrode response
• Crucial to have polysilicon filling and sufficiently
large lifetimes in it
• Lifetimes calibrated to match the electrode
efficiency found for Stanford detectors
Full 3D MIP simulation (proposed shaping time of 10 ns)
• Three hit points investigated (bulk and both
electrode types)
• CM properties similar to the simplified structure
• p+ electrode is fully efficient and show good CM
• n+ electrode is less efficient and shows lower CM
• Multiplication of the charge generated under the tip
42 / 46 Marco Povoli
HYbrid DEtectors for neutrons (HYDE)
Neutron detection
• Bare silicon is not able to detect neutrons
• Need for a converting material
• The most used converter is LiF
• Most of the commercial devices are able to only
detect thermal neutron
Neutron detectors produced with 3D technology
• Purposely designed cavities
• Cavities are filled with the converter
• Increased interaction probability between
reaction products and silicon
The HYDE project (INFN)
• Innovative polysiloxane converter
• Detects both thermal and fast neutrons
• Reaction products: recoil protons and light in the
blue to red range
Realized detectors
• The cavities are connected through columnar pillars
• Both with an without polysilicon filling
• Good leakage currents and breakdown voltages
• The converter is deposited at Laboratori Nazionali di
Legnaro
0
5
10
15
20
25
0 10 20 30 40 50 60 70 80
Cu
rren
t [n
A]
Reverse bias [V]
d=400, Up+=50
d=400, Up+=100
d=400, Up+=150
d=400, Up+=200
43 / 46 Marco Povoli
HYbrid DEtectors for neutrons (HYDE)
SEM pictures
• Fabricated cavity with connecting pillars (LEFT)
• The same after filling with converter (RIGHT)
α-particle measurements
• Measurements from both sides of the sensors
• Main peak correspond to 241Am alphas (except for
the energy loss in air)
• Lower peak from the trench side: geometrical
motivations
Neutron beam measurements
• Calibration with radioactive sources (α,γ)
• Bare sensor as comparison and two sensors with
converter
• Indication of increased statistics in the range from
0.5 to 1.25 MeV (VERY PRELIMINARY)
0.0001
0.001
0.01
0.1
1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
No
rmali
zed
co
un
ts
Energy [MeV]
Vbias = 40 V
METAL CONTACT SIDETRENCH SIDE
100
101
102
103
0 0.25 0.5 0.75 1 1.25 1.5
Co
un
ts
Energy [MeV]
Gamma source calibration, NO-conv.n28-350-150, WITH-conv. (Vbias=50V)
n8-300-50, WITH-conv. (Vbias=35V)n21-400-100, NO-conv. (Vbias=50V)
44 / 46 Marco Povoli
CONCLUSIONS
• An enhanced 3D-DDTC sensor concept with fully passing
through columns was designed at University of Trento and
fabricated at FBK
• All the performed studies allowed to gain a better
understanding of the behavior of 3D detectors both from the
electrical and the functional point of views
• 3D Pixel detectors compatible with the FE-I4 readout chip
proved to operate efficiently in IBL operating conditions
• These devices were chosen, together with CNM 3D detectors
and planar 3D detectors, to populate the ATLAS Insertable
B-Layer which will be installed during the first long shutdown
of the LHC (2013-2014)
• The large amount of activities performed in the framework of
the ATLAS 3D sensor collaboration triggered new ideas that
are currently being investigated and will be soon tested
45 / 46 Marco Povoli
Thank you!
46 / 46 Marco Povoli