1. Metal overhang: an important ingredient against “micro-discharge”
L.Demaria – INFN Torino
21/04/2004 Sensor meeting
Introduction and R&D results
Simulation results
Use of overhang on TK modules
ST simulation results
Conclusions

2. High Stability: critical fields around strips
The breakdown at the strips happens because a very high electric field is present at the edge of the p + implant. The magnitude of the field depends on the geometry: It is expected to be higher for lower values of the w/p ratio.
The breakdown field for silicon is about 30 V/mm. In practice breakdown can occur at lower fields if the high electric field triggers a breakdown in a few imperfect strips and impair, at least locally, proper operation of the detector.
- LD - NIM A447 (2000)
- CERN group (MM,LD,EM et al.)
CMS-Note 2000/011
NIM A

3. Improving stability with newer design
Improvement of the breakdown performances of the detectors can be obtained if the region with the higher electric field is moved away from the p + implant and into the oxide where the breakdown field is approximately 600 V/mm, which is an order of magnitude higher than for the silicon.  This can be achieved by the use of metal strips which are wider than the implant.
METAL
OVERHANG

4. Al readout strips referred to GROUND
TOB1-TIB3
Interesting results were found for irradiated detector comparing standard strip geometries with those with 8-mm metal overhang
Al readout strips referred to GROUND
TIB1
- LD - NIM A447 (2000)
- CERN group (MM,LD,EM et al.)
CMS-Note 2000/011
NIM A

5. Simulation studies Without overhang Pitch=240 mm w/p=0.1 w/p=0.3
Here are evident the effects of w/p and of pitch
D.Passeri, G.M.Bilei et al. – IEEE TNS 48 (2000)

6. Overhang effect w=72 mm, p=240 mm, Vbias=250V.
Highest electrical field is at p+ implant
corner in the n-bulk silicon
Highest electrical field is moved to the
oxide layer close to the end of metal
overhang
E-field in the n-bulk silicon is lower
E-field in the silicon bulk in further
decreased
w=72 mm, p=240 mm, Vbias=250V.

7. Vbias=500V, p=240 mm
w/p=0.1
w/p=0.3
ov=0
ov=4
ov=10

8. V bias = 500V, p=60 mm
w/p=0.1
w/p=0.3
ov=0
ov=4
ov=10

9. Irradiation effects

10. E-max for unirradiated sensors and different pitches and widths
Vbreak
E-max for irradiated sensors
Vbreak
D.Passeri, G.M.Bilei et al. – IEEE TNS 48 (2000)

11. MOS effect of metal overhang
The overhanging contact acts there as a p-channel MOS gate, which, under given bias conditions, suppresses the accumulation layer and tends to deplete the underlying region. This indicates a deep change in the electric field profile at the edge of the p+ implant, depending on the overhang extension.
D.Passeri, G.M.Bilei et al. – IEEE TNS 48 (2000)

12. <100> material <111> material Q~5-10 times than
for <100>
- LD - NIM A447 (2000)
- CERN group (MM,LD,EM et al.)
CMS-Note 2000/011
NIM A

13. TK Modules Unfortunatelly the TK modules have the Al referred to ~0.8V
The overhang effect makes the silicon sensor to be more stable at high voltage bias since it reduces the E-max close to the p+ implant.
All the results seen so far, both real measurements and silumation studies, if the AL lines are referred to GROUND i.e. if the voltage of the p+ implant is positive w.r.t. that of the Al.
This is specially important for the MOS effect that otherwise does not happen or even worse instead of working in depletion mode it works in accumulation mode, making the E-max to increase
Unfortunatelly the TK modules have the Al referred to ~0.8V
(input stage of APV) and p+ basically at GROUND.

14. ST simulations A B C D Vbias Qoxide Val Case + 500 Volts
Pitch = 180 mm
Vbias
Qoxide
Val
Case
+ 500 Volts
1.0· e-/cm2
GND A
+ 10 Volts B
1.5· e- /cm2 C D
Good <100> (low Vfb) material has Q~ 1.0·1010 e-/cm2

16. A: Val=GND E-max(Sil)~8 V/mm Oxide Al W= 40 mm P =180 mm W/P=0.22
Qox=1.0·1010 e-/cm2
Oxide Al
W= 40 mm
P =180 mm
W/P=0.22
Over=6 mm

17. B: Val=10V E-max(Sil)~12 V/mm Al Oxide W= 40 mm P =180 mm W/P=0.22
Qox=1.0·1010 e-/cm2 Al
Oxide
W= 40 mm
P =180 mm
W/P=0.22
Over=6 mm

18. C: Val=GND E-max(Si)~16 V/mm Oxide Al W= 40 mm P =180 mm W/P=0.22
Qox=1.5·1011 e-/cm2
Oxide Al
W= 40 mm
P =180 mm
W/P=0.22
Over=6 mm

19. C: Val=GND E-max(Si)~20 V/mm Oxide Al W= 40 mm P =180 mm W/P=0.22
Qox=1.5·1011 e-/cm2
Oxide Al
W= 40 mm
P =180 mm
W/P=0.22
Over=6 mm

20. Comparison of E-field Vbreak Big dependence on Qox (factor ~2)
D : Val=10V, Q= 1.5·1011 e-/cm2
C : Val=GND, Q= 1.5·1011 e-/cm2
B : Val=GND, Q= 1.0·1010 e-/cm2
A: Val=GND, Q= 1.0·1010 e-/cm2
Big dependence on Qox
(factor ~2)
Val > Vp+ is worse
(factor>30%)

21. Conclusions (1)
E-field is very high close to the p+ implant corner (=E-max)
E-max increases with pitch and decrease with width of p+ implant
MORE critical for TOB1-2 than for TIB
At high bias voltage E-max can become close to E-break (30V/ mm) so to cause strip, small area instabilities with I- leakage increase
Metal overhang decrease E-max  makes silicon sensor more stable at high bias voltage, provided Al at lower bias than p+ implant
Unfortunatelly TK-modules have Val~0.8V and Vp+~0V (!)
Oxide charge (~high Vfb) increase E-max close to implant if no metal overhang is used or if it has a positive voltage w.r.t p+ implant (this is even worse since MOS effect might increase local charge)

22. Conclusions (2) Is microdischarge in TOB modules due to Vbreak
in p+ strips ?
If yes, then:
the process behind is well understood and should have no
dependence on time
sensor testing made with strip floating is not ideal…
can the overhang be better used by increasing Vp+ (to 2.5V) ? This would improve module stability at high voltage
Vfb now is lower in ST sensors. This should decrease
‘microdischarge’ appearance
smaller pitches should be more ‘microdischarge’ resistant