first results of soi-based integrated detector and...
TRANSCRIPT
![Page 1: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/1.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 1 -
LCWS 2008 Chicago, November 2008
M.Trimpl, G.Deptuch, R.YaremaFermilab
First Results of SOI-based integrated detector and electronics at Fermilab
Outline- Features / Motivation SOI
- Counting pixel matrix: MAMBO
- 3T test array to evaluate cross talk in SOI
- Summary and Outlook
![Page 2: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/2.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 2 -
Features: SOI detector
Features• monolithic approach (prime example)
• small diode / input capacitance
• access to full CMOS (nmos/pmos)
• no bumping: low mass, low cost
• thinable, no backside implant
(50um depletion at ~15V)
• rad. tolerant (for ILC doses)
Joint SOI Effort
• Commercial SOI process on high resistivity substrate (700 Ωcm) made available from OKI via KEK
• SOI-Pix collaboration started 2007 (coordination Y.Arai, KEK): 2 OKI-runs 2007, 2008
![Page 3: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/3.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 3 -
second run (2008) via OKI:
Process 0.2μm Fully-Depleted Low-Leakage, SOI CMOS, tox=4.5/7 nm, 1P4M, MIM, DMOS, Vcore=1.8 V (OKI Electric Industry Co. Ltd.).
SOI wafer Wafer: 200 mmφ, Top Si : Cz, ~18 Ω-cm, p-type, ~40 nm thick Buried Oxide: 200 nm thick, Handle wafer: Cz、700 Ω-cm (n-type), 650 μm thick (SOITEC)
Backside Thinned to 260 μm, and sputtered with Al (200 nm).
OKI - process overview
diode implant
as small as ~ 1μm (Φ)
![Page 4: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/4.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 4 -
Fermilab designs (2008 run)
47μm
3T readout (MAPS like)• investigate detector propertiesand cross talk effect
• 8x10 pixel array, 20x20μm pixel size
5x5 mm2 submitted Jan. 2008, returned Sep. 2008 (after corrected diode implantation)
MAMBO (Monolithic Active Pixel Matrix with Binary Counters )
• 97x97 pixel matrix, 47x47μm pixel size• counting pixel for imaging application
47 μm
![Page 5: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/5.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 5 -
each pixel:
• integrating CSA w/ p-z network,
• shaping filter CR-RC2 with
τp =150ns; gain=~200 μV/e-
(for small Q),
• multiple diodes/pixel,
• ripple counter (reconfigurable
as shift register)
Cc =28 fF, Cfs =3.3 fF, Rfs =50 MΩ, Gm =5.8 mS, Ch =30 fF(compact design excluding use of physical resistors)
Mambo - pixel overview
![Page 6: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/6.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 6 -
Lack of precise SOI models is design challenge
later measurements
showed very small eq.
resistance
(SPICE models did not
indicate such effect)
PMOS biased at VGS =const as a feedback resistor (50 MΩ)
0.15 μm(2007)
0.20 μm (2008)
2008 run: SB–connected transistor & ‘half-H-gate’ design for feedback transitor
required values of feedback
resistor were obtained
(this was empirical approach,
not driven by device models )W/L=0.63u/0.3u
2007 run: used transistor
with floating body
![Page 7: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/7.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 7 -
back gate effect
13 diodes / pixel
Significant threshold voltage shifts observed, however, design techniques canreduce effects:
transistor from the shaper’s feedback path(separate structure -> no biasing diodes)
adjustment of bias voltages and currents (referred to VGS ) up to several tens of mV is still required for back gate voltages from the range from 0 to 10V.
![Page 8: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/8.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 8 -
array with 3T pixels
architecture
109Cd (22 keV γ): max. signal 200 mV Cin = 4.8 fF (33 μV/e-)Noise (RMS) after CDS ~ 1mV
reset resetintegration
hits
Vback=2.5V Vback=2.5V
Vback = 2.5V -> ~ 20 μm depletion
![Page 9: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/9.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 9 -
cross talk structures
sensitive node: 1.2um diode p-implant, 20um pitch
Aggressor:
- diffusion implant supported by metal1- switching between two voltages- 1.2 um wide located at pixel boundary
grid creates various scenarios, here worst and best case cross section – cross talk arrangement
![Page 10: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/10.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 10 -
cross talk results – pixel array
10
8
6
4
2
0500
10001500200025003000
1
2
3
4
5
6
7
8
inducedcharge [e]
row
column Vback=6.5VΔU = 1.5V
note, effect on falling and rising edge almost equivalent (net charge ~0) -> suggests differential logic
Procedure: - reset pixel and toggle voltage during integration time- calculate difference before and after voltage change
charge induced by falling edge of voltage swing
clear correlation between aggressor lines and induced charge
![Page 11: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/11.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 11 -
10
8
6
4
2
0500
10001500200025003000
1
2
3
4
5
6
7
8
inducedcharge [e]
row
column Vback=6.5VΔU = 1.5V
charge induced by falling edge of voltage swing
cross talk results – pixel array (cont’d)
10 8 6 4 20
500
1000
1500
2000
2500
indu
ced
eq. e
lect
rons
column
row 3
• ~ 80% from left neighbor• ~ 10% from right neighbor• top and bottom contribution:
consistent but small
most significant symmetry in the pixel is left / right ….. (?!) note, effect on falling and rising edge almost equivalent
(net charge ~0) -> suggests differential logic
![Page 12: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/12.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 12 -
cross talk - quantification
2 3 4 5 6 7 8
0
500
1000
1500
2000
2500
3000
indu
ced
eq. e
lect
rons
Vback[V]
pixel 5/6(fully enclosed)
ΔU = 1.5V
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.60
500
1000
1500
2000
2500
3000
indu
ced
eq. e
lect
rons
ΔU [V]
Equation y = a + b*x
Weight No Weighting
Adj. R-Square0.99998 Value Standard Error
J Intercept -43.71429 2.25921J Slope 1883.14286 2.4848
pixel 5/5 (fully enclosed)VB=6.5V
eq. coupling C = 0.3fF
cross talk starts at VBG > 2V -> shows that coupling is through substrate
(and not CMOS)
saturated
linear dependence of charge vs. voltage swing( as expected from capacitive coupling )
(as depletion grows into depthcapacitive coupling saturates)
![Page 13: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/13.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 13 -
cross talk – DC level
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
50
100
150
200
250
300
350
400
indu
ced
eq. e
lect
rons
DC-Voltage U1[V]
ΔU = U2-U1 = 500mV
Equation y = a + b*x
Weight No Weighting
Adj. R-Square 0.99044Value Standard Error
Intercept 401.4 5.5443 Slope -266.10989 7.24885
pixel 5/6(fully enclosed)Vback = 3.3V
2 3 4 5 6 7 80
200
400
600
800
1000
1200
1400
pixel 5/6(fully enclosed)
0..0.8V 1..1.8V
indu
ced
eq. e
lect
rons
Vback [V]
note, this dependency will disappear at higher VBG
-> high VBG (> 6V) desired, otherwise logic states
can influence potential distribution in substrate
• voltage swing constant (800mV)• sweep starting voltage (DC pedestal)
effect starts at higher VBG by shifting DC potential at electrode (same effect possible due to charge build up in BOX)
![Page 14: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/14.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 14 -
• SOI approach pursuit by KEK/OKI offers interesting opportunitiesfor particle detection, yet also new challenges (backgate, crosstalk…)
- These effects need to be understood / analyzed to succeed in complexcircuit designs
• MAMBO – counting pixel matrix- corrected design of 2nd gain stage expected back: End Nov. 2008- USB DAQ-System is build to operate Mambo for imaging application
• Test structures to analyze cross talk
- Parasitic extraction for vertical integration designs (SOI, 3d) is important(effects are huge due to high integration level)
- Available Tools at Cadence Level very limited / not existing to date- Cross check some effects with Silvaco Simulation (dominating left neighbor)
• We appreciate being part of the SOIPIX collaboration around KEK and hope that we can push SOI design to a next level
Summary / Outlook
![Page 15: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/15.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 15 -
Backup Slides
AN INTERNATIONAL LINEAR
![Page 16: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/16.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 16 -
single pixel exposure to single pixel exposure to 109109Cd source Cd source
• Pixel responses to the radiation events, polarity of signal is reversed and amplitude attenuated – it is believed that a depletion transistor (capacitor) is biased too shallow in inversion – changed in corrective submission in August 08,• problems encountered with switching counters to shift register – more thorough tests are underway
Output of the shaping filter
![Page 17: First Results of SOI-based integrated detector and ...ppd.fnal.gov/eed/asic/Presentations/lcws08_trimpl.pdf · • 8x10 pixel array, 20x20. μ. m pixel size 5x5 mm. 2 . submitted](https://reader034.vdocuments.us/reader034/viewer/2022052012/6028be131520c241cc0f1182/html5/thumbnails/17.jpg)
LCWS, Chicago, 17.11.2008 Marcel Trimpl, Fermilab- 17 -