SilC Silicon Sensor Baseline and Proposal for full 6” Design

Download Report

Transcript SilC Silicon Sensor Baseline and Proposal for full 6” Design

SiLC Silicon Sensors:
Status of different activities
Thomas Bergauer
for the SiLC Collaboration
Outline:
• SiLC “Sensor Baseline”
• DESY TPC
• Synergies with SLHC
• Status of companies
April 25th, 2007
1
Linear Collider Constraints
• Future Linear Collider Experiment will have a large
number of silicon sensors
– Few hundreds m2 (CMS has 200 m2)
• Radiation damage will be no issue (e+e-)
• Concept for strip tracker:
– long strips (10-60cm)
– low material budget
– Light active cooling only due to power cycling of FE electronics
(1/100ms duty cycle)
– Time structure
of beam:
2
SilC Silicon Sensor Baseline
• SilC sensor baseline
– FZ p-on-n sensors: n-bulk material, p+ implants for strips
– high resistivity (5-10 kOhm cm)
– Readout strip pitch of 50µm
• Possibly intermediate strips in between (resulting 25µm pitch)
• Smaller pitch becomes very complicated (Pitch adapter, bonding, charge
sharing,…)
– Thickness around 100-300µm
• mostly limited by readout chip capabilities (S/N ratio)
– Low current: <1nA per strip
(Due to long integration time noise mostly defined by current and resistors)
• Baseline for inner layers:
– 6” inch, Double sided, AC coupled
• Baseline for outer layers:
– 8” (12”?) inch, Single sided, Preferably DC coupled (cheaper)
3
Sensor Baseline Details
Biasing Possibilities:
• bias resistor with poly-silicon
(20 to 50 MOhm)
•
•
punch-through (upper picture)
or FOXFET biasing structure
(lower picture)
– Latter two have non-linear
behavior
– But are cheaper
4
Minimize material budget
•
Multiple scattering is crucial point
for high-precision LC experiment
•
Minimize multiple scattering by
reduction of material budget
– avoid old-fashioned way (pitch
adapter, FE hybrid, readout chip)
Bump bonding
pads
Strixel
• Integrate pitch adapter into
sensor
– Connectivity of strips to readout
chip made by an additional oxide
layer plus metal layer for signal
routing
– Readout chip bump-bonded to
sensor like for pixels
Via (DC coupling)
AL routing & pad area
Oxide
Silicon
5
Similar to SiD Concept:
Slide taken from Timothy Nelson‘s Presentation of SiD detector concept at Bejing ILC GDE
Meeting (Feb 6, 2007)
…..very elegant!
6
“Inline pitch adapter” for SiLC UMC Chip
IEKP Karsruhe
• UMC 130nm chip
successfully tested with
long ladder out of 10 HPK
GLAST sensors
– Wire bonding
• Next UMC chip version
– 128 channels
– Bump-bonding
LPNHE Paris
7
Long ladder with 2 sensor types
PA on sensor with
bumpbonded chips
Standard Wire Bonding
F. Hartmann
Chip to stip routing on sensor
8
SilC work program for sensor R&D
• Step 1 (2007)
–
–
–
–
Use long strips (50 µm pitch)
Wafer thinning (100, 200, 300µm)
Test new readout chips (DC coupling, power cycling)
Improve standardized test structures and test setups
• Step 2a (2008-)
– Move from pitch adapter to in-sensor-routing
– Test crosstalk, capacitive load of those sensors
• Step 2b (2008-)
– Test 6” double sided sensors
• Step 2c (2008-)
– 8” (12”) single sided DC wafer
9
Step 1 and 2a:
•
•
•
•
•
•
Bump-bondable 128-channel chip available end 2007
HPK agreed to provide a sensor design
SiLC adapts strip to pad area
HPK will process the sensor
SiLC (Paris) provides chip
HPK could bump bond chip to sensor
– HPK is very interested to strengthen inhouse bumpbonding
• In Bump
• Flipchip
– Stud-bonding (Jean-Francois Genat)
• Testing begins 2008
10
DESY TPC
11
Silicon “Envelope” for TPC
Ties Behnke, 4th SiLC Meeting Barcelona (Dec 2006):
12
Superconducting Magnet
• Magnet already at DESY
• TPC Support structure
movable; Si modules
should be mounted onto
this structure
• Timescale:
– Construction of TPC Field
Cage until autumn 2007 by
commercial company
– First beam test until end of
the year
13
Silicon Envelope
• Long ladder in z Direction
• Stereo modules with two
sensors
• Resolution requirements still
unclear:
– R-Phi: 10-50 um
– Z: 50-250 um
• Stereo angle responsible for z
resolution
– Optimal z resolution when
stereo sensor perpendicular to
R-phi sensor
• Crucial Point: Space between
magnet and TPC: 2cm
– Large stereo angle needs
more space
– Sensors perpendicular?
14
Resolution of stereo modules
r-Phi resolution:
re solution
pitch
sigma
[um]
[um]
120
35
80
23
50
14
50*
10
* with intermediate strips
z<100 um hard to archive
CMS: angle 6 deg already a lot!
15
Next Steps
• Sensor procurement from HPK
– Details about HPK offer later
– Sensors tests @ Vienna, Karlsruhe
• Find material for module frame/support
– Rohacell, Silicon foam
– Carbon Fiber, Graphite
• Define module/ladder geometry
– Maximum number of daisy-chained sensors?
• Depends on S/N ratio of chip (Jean-Francois?)
– Layout of FE hybrid and Chip pitch needed (Jean-Francois?)
• Everybody who is interested is welcome to join the
phone/video meeting. Just send me a mail and I will take
you on the list.
• Plan to have a tour to the TPC magnet during LCWS07 in
Hamburg (end of May)
16
Synergies with (S)LHC
LHC Upgrade project is called “super” LHC
17
Synergies with (S)LHC
Slide taken from Frank Hartmanns’ Talk in “CMS Sensor upgrade Workshop” (February, CERN)
piece
sensor
development
reason & realization ILC
reason & realization SLHC
synergy
problems
8"
3D
MCz
large area
not needed
not neeed
YES
NO
NO
no industry standard
Thinning
n-in-p, n-in-n
double sided
multiple scattering
not needed
save MB
Strixel sensor
not needed
chip must cope with switch off
and GND on strip
large area
radiation hardness
radiation hardness
Vdep;power, multiple scattering, no need
for thick cause CCE degradation
Depletion after SCI starts from top
not applicable, radiation damage
power goes into service --> MB
occupancy
current too high
NO
YES
NO
NO
signal
NO
DC
edgeless sensors with 3D
for edges
no need for overlap => large area
high voltage stability
YES
small feature size
90/130/180nm
radiation
timing
low power consumption
no issue; standard libs
1-3µs (slow)
low power consumption
special libs needed
~10 ns FAST
not really
NO
NO
multiple scattering; power
multiple scattering; power
YES
to be decided
perfect
not usable
NO
Noise ~ C critical for SLHC
no industry standard
electronics/chip
misc
electronics on chip
(CMOS,bump bonding)
no extra hybrid
SOI
long ladders
18
Synergies with (S)LHC
• Most interesting match between the projects:
– Large areas (8”)
– SLHC wants to use second metal layer to route signals from
“large pixels” (a.k.a. “short” strips”) “stixels”
• Many points which are not matching: (unfortunately)
– search for other material which is more radiation tolerant:
MCz, Oxygenated Si
– Other “base” material: n-on-n; n-on-p
– Double sided sensors
– Long ladders (high capacitive load) vs. short integration time
• Synergies would help both, SLHC and SiLC to convince
companies about new developments
19
Status of the companies
20
HPK
•
•
During Vienna Conference “VCI” (Feb 2007): Meeting with HPK’s
European Sales representant and Japanese engineer.
We had several questions which have been answered by them (later by
e-mail):
–
–
–
–
–
–
Double sided processing (6”)? Not possible
Cost difference AC-DC coupling: AC is 40-50% more expensive
Cost difference biasing (DC=100): PT: 115; FOXFET : 120; Poly-Si : 130
Minimal sensor thickness? 200 micron (possible in 1 year from now)
Maximum dielectric layer? 1m with SiO2, maybe Polyimide in the future
Trace metallization: only Al with 0.9 m thickness and minimum width of 3
to 4 m; no other material for the time being (we asked for Copper)
– Larger Sensors: 8-12” production possible via sub-contractor, but much
more expensive; limited prototyping
– bump bonding: Indium bump-bonding being developed within 1 year, 20 m
pitch
•
We asked HPK for an offer for 30 pcs single sided detectors with 50um
pitch during this meeting.
21
HPK offer from April 24th (yesterday):
• Specification:
–
–
–
–
SSD type : single-sided DC type
SSD Chip size : about 95 x 95 mm
Thickness : 320 um
Strip : 50 um pitch (with one intermediate strip)
• Price quotation:
NRE
One time only
EUR 27'000.-
Unit price
30 x 1500
EUR 45'000.-
Total batch
EUR 72'000.-
• Delivery:
–
Lead time takes 5 months aro for first 30 pcs prototype.
• They are now waiting for an official order.
22
VTT
• After Proposal from Simo
Eränen during December’s
SiLC Meeting we (=Vienna)
started to collaborate
• Design almost ready
– Two main sensors on wafer
• One sensor DC coupled
• Other AC coupled with PT or
FOXFET bias because of lack of
polySi processing line
– Vienna provided CMS-like ‘halfmoon’ TS
• Now design verification
• Begin of processing soon
1.
2.
3.
4.
5.
6.
MAIN DETECTOR, 5 X 5 SQCM
MEDIPIX2, 1.5 X 1.5 SQCM
ALIGNMENT MARKS, 1 X 1 SQCM
HALF MOON TEST STRUCTURE
EDGELESS TEST STRUCTURES, 1.5
X 1.5 SQCM
BABY DETECTORS, 1 X 1 SQCM
23
Meeting with Hwanbae Park (Korea), Jan 29th
•
•
Hwanbae showed results of the work of
his group during December’s SiLC
meeting
Frank, Manfred and I had a phone
meeting asking him for larger detectors
– Like it is shown on the right
– Maximize size per wafer in a single sensor
(approx. 10 x 10 cm2 )
– According to SiLC baseline
•
We agreed that we provide a modified
CMS design (with TS) for full 6” wafer
(timescale: several months)
•
Agreed to start in April 2007, when new
financial year starts in Korea since more
money will be available
CMS OB2 sensor
24
IET Warsaw
• Frank Hartmann established contact with Institute for Electron
Technology (Jacek Marczewski) already three years ago
• Karlsruhe and Vienna are both in loose contact with them to develop
test structures
– Based on CMS ‘half-moon’
• They have experience with SOI and chip production, but not with fully
depleted devices yet.
• Production of first batch has just started
25
Canberra
?
(Aurore proposed to establish a contact with them)
26
ON Semiconductor
•
•
•
•
http://www.onsemi.cz
Former “Tesla” company
Located in Czech republic
Presumably high throughput
– 4” production line running
– 6” production line currently commissioned
• “first contact” can be established with help of
Vaclav Vrba from Czech Academy of Sciences
hopefully soon.
27
Summary
• Sensor baseline established:
– FZ, p-on-n, high resistivity, 100-300um thick, 50um pitch
– preferably DC coupled, otherwise biasing via PolySi, PT or FOXFET
• DESY TPC project started
– Sensor procurement with HPK ongoing, but takes longer than expected
– Several design decision to be made:
•
•
•
•
Which stereo angle? (or perpendicular sensors; best resolution)
Material of Support: CF/Rohacell, CF legs like CMS, other material?
Next step: Module Construction
Layout of the FE hybrid unknown, need input from Jean-Francois
• Discussion with several companies / institutes ongoing
–
–
–
–
–
VTT: design almost finished
Korea: 1st design has to come from us within next weeks
IET Warsaw: TS already in production
Canberra: unknown
ON Semiconductors: 1st visit next month?
28
The End.
Backup Slides follow
29
Silicon sensor expertise within SiLC
• Both, Vienna and Karlsruhe worked a long time
already with silicon sensors
– DELPHI @ LEP
– CDF @ Fermilab
– CMS @ LHC
• We have experience in
– Strip-by-strip characterization of Si strip sensor
– Process monitoring with test structures
– Proton irradiation facility in Karlsruhe
• Post irradiation characterization
• Examples of our experience on next few slides
30
Strip-by-Strip Characterization
What do we test?
•
Global parameters:
– IV-Curve: Dark current,
Breakthrough
– CV-Curve: Depletion voltage, Total
Capacitance
•
Strip Parameters e.g.
–
–
–
–
strip leakage current Istrip
poly-silicon resistor Rpoly
coupling capacitance Cac
dielectric current Idiel
31
Process Monitoring on Test Structures
•
CMS “Standard Half moon”
–
–
•
9 different structures
Use to determine one parameter per
structure
Worked extremely well during CMS
sensor production
–
Example of an identified problem can
be seen in plot: low interstrip resistance
TS-CAP
sheet
•
baby
GCD
CAP-TS-AC
CAP-TS-AC
diode
MOS 2
MOS 1
Improved version for SiLC
–
–
overall size reduction
Structure design improvements
(e.g.better sheet structure)
HEPHY Vienna
32
Strip-by-strip Test Setup
• Sensor in Light-tight Box
• Vacuum support jig is carrying the
sensor
– Mounted on freely movable table in
X, Y and Z
• Cold chuck in Karlsruhe available
• Needles to contact sensor bias line
– fixed relative to sensor
• Needles to contact:
– DC pad (p+ implant)
– AC pad (Metal layer)
– Can contact ever single strip while
table with sensor is moving
33
Test Structures Description
•
TS-CAP:
–
–
•
Aluminium resistivity
p+-impant resistivity
Polysilicon resistivity
GCD:
–
–
–
Gate Controlled Diode
IV-Curve to determine surface
current Isurface
Characterize Si-SiO2 interface
CAP-TS-AC:
–
•
•
Inter-strip Resistance Rint
Diode:
–
–
•
IV-Curve for dark current
Breakthrough
CAP-TS-DC:
–
•
Inter-strip capacitance Cint
Baby-Sensor:
–
–
Sheet:
–
–
–
•
Coupling capacitance CAC to
determine oxide thickness
IV-Curve: breakthrough voltage of
oxide
•
CV-Curve to determine depletion
voltage Vdepletion
Calculate resistivity of silicon bulk
MOS:
–
–
–
CV-Curve to extract flatband voltage
Vflatband to characterize fixed oxide
charges
For thick interstrip oxide (MOS1)
For thin readout oxide (MOS2)
34
Test structures Measurement Setup
•
Probe-card with 40 needles
contacts all pads of test structures
in parallel
–
–
–
•
Half moon fixed by vacuum
Micropositioner used for Alignment
In light-tight box with humidity and
temperature control
Instruments
–
–
–
Source Measurement Unit (SMU)
Voltage Source
LCR-Meter (Capacitance)
•
Heart of the system: Crosspoint
switching box, used to switch
instruments to different needles
•
Labview and GPIB used to control
instruments and switching system
35
Example of identified problems
Inter strip resistance issue during CMS sensor
production
•
•
•
•
•
Limit: Rint > 1GΩ to have a good
separation of neighbouring strips
Each dot in the left plot shows
one measurement
Value started to getting below
limit
We reported this to the company
Due to the long production
pipeline, a significant amount of
~1000 sensors were affected
36
Proton Irradiation @ Karlsruhe
• 35 MeV protons with easy
access to 10^15 1MeV Nequiv
• Irradiation for design check
• Later: Irradiation control during
production
• Standard pre- and post
irradiation measurements
Compact Cyclotron at Forschungszentrum Karlsruhe
37
UMC Technology parameters
•
•
•
•
•
3.3V transistors
Logic supply
Metals layers
MIM capacitors
Transistors
180 nm
130nm
yes
1.8V
6 Al
1fF/mm²
Three Vt options
yes
1.2V
8 Cu
1.5 fF/mm2
Low leakage option
38
SilC Readout Chip
• SilC Collaboration is
developing a new readout
chip designed for the ILC
needs:
– DC coupled readout
– Power cycling
• UMC 130nm technology
– Chip with 4 channels
already available
– Chip with 128 available
early next year
LPNHE Paris
Picture
39
SiTR-130_1 tests results
Gain OK:
30 mV/MIP
Dynamics:
30 MIPs @ 5%
Peaking time: 0.8 – 2s
OK
OK
0.7 - 3 s
LPNHE Paris
Power (Preamp+ Shaper) = 300 W
Noise comparative
130nm @ 0.8 s : 850 + 14e-/pF
130nm @ 2 s :
625 + 9e-/pF
180nm @ 3 s :
360 + 10.5 e-/pF
40
NEW
S/N measured with SiTR-180 (after t.b.)
W. Da Silva, J. David, F. Kapusta, F. Rossel (LPNHE)
VERY VERY PRELIMINARY
GLAST module channels read by VA1 (top) or 180 (bottom)
41
UMC 130nm 4-channel test chip: SiTR-130
Channel n+1
Zero-suppression
Can be used for
a “trigger”
S aiVi > th
Time tag
Channel n-1
reset
reset
Analog samplers, (slow)
Ramp
ADC
Strip
Ch #
Preamp + Shaper
DC servo implemented for DC coupled detectors
UMC
CMOS 130nm: SiTR-130_1
Waveforms
Counter
Received in August 15 2006
Being tested: Analog OK, Digital under tests
Clock 3-96 MHz
42