1. Comparison of P- and N-TYPE structures for both un-irradiated and irradiated MSSD sensors

2. Cint of un-irr./irradiated 200μm devices
f = 1 MHz
Cint = 2*[AC(1,2)+DC(1,2)+AC(1)DC(2)+DC(1)AC(2)]
Defects: Proton T = 253 K,
F = 1.5e15 cm-2
200N/P device, region 5
Double p-stop: Np = 1e16 cm-3, depth = 1.5 μm,
width = 4 μm, spacing = 6 μm
Minimum Cint of irradiated n-type is proportional to the value of QF
Irradiated p-type device reaches geometrical value at V ~300V when QF = 5e11 cm-2,
At higher values of QF electrons are not removed from the inversion layer
→ proton model does not generate enough negative space charge at F = 1.5e15 cm-2 to compensate for Si/SiO2 interface electrons
N-type device has lower Cint noise for most probable values of QF of an irradiated real detector
At F = 1.5e15 cm-2, Cint values of non-irradiated devices are not reached within 1 kV for Qf > 5e11 cm-2 → two defect model is not generating enough negative space charge between strips to compensate for inversion layer electrons

3. Simulation of MSSD : Cint vs. Vbias (un-irradiated)
N type
P type
Simulation is mostly in good agreement with measurements for both P and N-type.

4. Five Trap model Two shallow acceptors and one
Energy Level
Intro.
σe (cm-2)
σh (cm-2)
Acceptor
0.525eV
3.0
1x10-14
1.4x10-14
0.45eV 40
8x10-15
2x10-14
0.40eV
Donor
0.50eV
0.6
4x10-14 20
Two shallow acceptors and one
shallow donor in addition to two deep levels
Able to remove accumulation e-
Produce very high E field near n+
Reproduce experimental observed good
Rint and Cint
With one deep acceptor, it is not possible to create enough E field (similar to measurement)
near n+ strip along with correct current.
We can not use deep acceptors with higher introduction rates as it will change space charge
significantly leading to very high avalanche multiplication & simulated current become very
high compare to measured one.
Moreover, in reality also, shallow levels are created in much more amount compare to deep
trap levels

5. Simulation of MSSD : Cint vs. Vbias (Irradiated)
Red- Experimental result (flux-5e14)
Blue - Flux=5e14neq, & QF =8e11cm-2
Green – Flux=1e15neq, & QF=1.2e12cm-2
Red- Experimental result (flux-5e14)
Blue - Flux=5e14neq, & QF =8e11cm-2
Green – Flux=1e15neq, & QF=1.2e12cm-2
N type
P type
Simulation is mostly in good agreement with measurements for both P and N-type
Cint changes slightly with change in combination of bulk damage (flux) + surface damage (QF) for low bias values.

6. Rint of P-type device(un-irradiated)
3 strip structure, Vstrip1 = Vstrip3 = 0, Vstrip2 ~ 3 V and 0 V
V = -HV at the backplane
Interstip resistance (Rint ) is defined as (Induced Current Method):
Rint is plotted as a function of applied voltage V = -HV
320P
Bulk doping = 3.4e12 cm-3
p-stop depth = 1.6 μm
p-stop width = 6 μm , 12 μm
p-stop spacing = 6 μm
implant depth = 2.2 μm
Rbias = 1 MΩ
Fig. 1
Fig. 2
Oxide charges Qf of the Si/SiO2 interface are varied
Minimum Rint determined by Rbias values of the
two strips
400 V
200 V
950 V
680 V
No bulk defects:
No dependence on pitch and p-stop spacing observed
Double p-stop width (figs. 1, 2):
QF = 7e11 cm-2 isolation reached
~200 V lower voltage
→ Rint has strong dependence on p-stop width
Double p-stop doping:
Strips are isolated at Qf ≤ 1e12 cm-2
Higher electric fields at p-stop edges
(~40% increase)
Spacing = 6 μm
width = 6 μm
Np = 1e16 cm-3
spacing = 6 μm
width = 12 μm
Np = 1e16 cm-3
Timo Peltola, Phase 2 Sensors, 29 August 2013

7. Rint of P-type device(un-irradiated)
Defects: Proton T = 253 K,
F = 1.5e15 cm-2
200P device, region 5
Double p-stop: Np = 1e16 cm-3, depth = 1.5 μm = implant depth, width = 4 μm, spacing = 6 μm
200P
Rint = 1/A approx. matches qualitatively with Current Induced Method
Isolation is not reached within 1 kV when
Qf > 1.2e12 cm-2
Voltage where isolation is reached equals the voltage of Cint drop
As seen in Cint behaviour of equal fluence, at higher values of Qf electrons are not removed from the inversion layer
→ proton model does not generate enough negative space charge at F = 1.5e15 cm-2 to compensate for Si/SiO2 interface electrons
At low fluence/non-irradiated region Rint is strongly dependent on p-stop width
Rint shows no dependence on pitch and p-stop spacing
At F = 1.5e15 cm-2 with Qf > 1.2e12 cm-2 strips are no longer isolated within 1 kV depletion voltage → two defect model is not generating enough negative space charge between strips to compensate for inversion layer electrons

8. Simulation of MSSD : Rint vs. Vbias (un-irradiated)
N type
P type
QF= 1x1011cm-2
Structure no-1
QF= Vary
P type
N-type:
All 12 structures follow the similar good Rint characteristics for all values of QF.
P-Type:
Good isolation for all 12 structures for low values of QF.
Strip-isolation decreases on increasing the QF.
QF= 3x1011cm-2

9. Simulation of MSSD: Rint vs. Vbias(Irradiated): P-TYPE
Measurement (Wolfgang)
DC-CAP
Simulation (flux = 1x1015 cm-2)
P and Y types
Different QF
Simulated Rint show trends similar to the Measurements.
Rint decreases on increasing the QF.
Rint is a strong function of the combination of surface damage (QF) and Bulk Damage (flux). Bulk damage compensates for surface damage.
Good isolation even at high flux and high QF.
Simulation
Different Flux
QF = 5x1011cm-2

10. Simulation of MSSD : Rint vs. Vbias (Irradiated) : N-type
Measurement (Wolfgang)
DC-CAP
Simulation (flux = 1x1015 cm-2)
Different Structures
N-Type: Different QF
Isolation remains good for all values of QF.
Simulation shows decrease in Rint for high values of QF at high Bias values. Experimentally different structures show similar behaviour.
Electric field near the curvature of p+ strip is quite high & increases with QF . This high E field can initiate a localized avalanche & can decrease Rint

11. E. Fields – N-type un-irradiated
Efield along the surface (0.1 um below)
Efield along the surface (1.3 um below)
QF=1e11cm-2, V=1000V
Comparing (p90,w20), (p240,w20), (p240,w60),
Highest fields for very small width/pitch
Higher electric fields at the junction in the silicon bulk (at 1.3µm) => No critical fields for pitch 90µm, width 20µm
(Larger pitch to width is negative for Cint)

12. E. Field at the Strips – N-type;
1000V
F=1e15neq/cm2
High QF
Low QF
Increase in E worse with irradiation
Soft breakdown due to very high electric fields at the strips with higher QF
F=3e14neq/cm2

13. E. Fields at the Strip – P-Type (Irradiated)
FZ320P
Higher E field with higher irradiation
No critical fields at the strips
Strip
Al overhang
P-stop
Strip
Al overhang
P-stop

14. E. Fields – P-Type
Comparison between 320µm and 200µm thick FZ p-bulk sensors
Not much higher E. fields than for 320µm devices at strips (center of strip)
Higher fields in the bulk
Lower fields for higher QF– intrinsically good!
Pitch 90, width 20
1000V
320µm
200µm
Higher QF
Higher QF

15. E. Fields – P-Type: Effect of P-stop Doping
FTH200P
No critical fields for usual configuration
High fields can occur at high p-stop doping!
QF = 1x1011 cm-2
QF = 1x1012 cm-2
Strip
Al overhang
P-stop
Strip
Al overhang

16. Simulation of MSSD : E. Field (Irradiated): P & N-types
Efield along the surface (1.3 um below)
Efield along the surface (0.1 um below)
P-Type
P-Type
N-Type
N-Type
Flux = 1x1015cm-2 ; QF = 1.2x1012cm-2; Bias = 500 V
Peak electric field is more for N-type as compared to P-type sensor for a given bias.
Micro-discharge possibility is more in N-type sensors.

17. Simulation of MSSD : E. Field (Irradiated): QF variation
P-Type
N-Type
Flux = 1x1015cm-2 ; QF : Vary ; Bias = 500 V
N-TYPE:
As QF increases = > Peak Efield increases.
Micro-discharge possibility is more in N-type sensors.
P-TYPE:
As QF increases = > Peak Efield decreases.

18. Simulation of MSSD : E. Field (Irradiated): Temp. variation
Efield along the surface (0.1 um below)
N-Type
Flux = 1x1015cm-2 ; QF = 8x1011cm-2 ; Bias = 500 V
N:TYPE : Peak E field increases with increase in Temperature.

19. Design Rules for P-type

20. Simulation: Cint vs. Vbias (un-irradiated)
Cint rises with w/p and with thickness
Cint is largest for p-spray, smaller in n-type and smallest in p-stop

21. Simulation: Ileak vs. Vbias (un-irradiated)
Ileak rises with thickness
Ileak almost independent of material
(except FZ120N, where Ileak is 10% higher)

22. Simulation: Cint vs. Vbias (un-irradiated)
Cint data momentarily only available for one thickness and region
Over- and undershoot may be a simulation artefact
No clear pro or con from Cint

23. Electric Fields – P-Type un-irradiated
Strip
P-stop
Strip
Thickness: 320µm
Strip
E-field dependent on p-stop distance to strip
Lower fields for large distance to strip
E-fielddependent on p-stop width
Small width shows lower electric fields

24. Electric Fields – P & Y-types un-irradiated,
200µm thick sensor
P-stop placement:
Lower fields for smaller width of p-stop
Placements of p-stop should be in the center of the strips
P-stop doping concentration should be larger than 5e15cm-3 to isolate
P-spray depth and doping:
High implant depth & high p-spray doping produce high electric fields at the strip edge
P-spray doping depth should be shallow for low electric fields
Too low p-spray doping does not isolate the strip at higher QF
Tuning difficult

25. Effect of P-stop doping concentration : Rint
Pstop – 5e17cm-3
Pstop – 5e15cm-3
Bias = 200V
Flux=1e15cm-2
QF = 1.5e12cm-2
Cutline is 0.1µm
below SiO2
E field (V/cm)
Pstop-5e17cm-3
Pstop-5e16cm-3
Rint (ohm)
Pstop-5e15cm-3
P-stop doping conc. Variation; QF=1.2e12
Strip pitch : 90 micron (width = 20 micron)
Double Pstops (4µm each, separation - 6µm)
Flux = 1e15cm-3
Increase in Pstop-doping conc. Increases Rint but decreases breakdown voltage.
Higher Pstop doping leads to very high E field at lower biases near Pstop curvature which can lead to sensor breakdown or probably microdischarges also.
Lower Pstop-doping concentration is preferred.

26. Effect of P-stop doping width : Rint
Strip pitch : 90 micron (width = 20 micron)
Single Pstop (14µm and 28µm ); Pstop doping conc. = 5x1016cm-3
Flux = 1e15cm-3
For the values of the Pstop width considered in the simulation, the results of Rint and Efield are almost independent of that.

27. Backups!

28. Irradiated Strip Sensors
Effective Irradiation Model (tuned especially for protons)
Parameter
Donor
Acceptor
Energy
EV eV
EC eV
Concentration (cm3)
5.598 * F – 0.959e14
1.189 * F e14
σ(e)
1.0e-14cm2
σ(h)

29. Simulation Strip Device
For Backup
Simulation Strip Device
Strip / backside doping concentration: 1e19cm-3
Implant depth: 1.5µm
Oxide thickness: 1µm
Oxide thickness at the strip: ~200nm
N-bulk doping: 3e12cm-3
P-bulk doping: 3.4e12cm-3
P-stop doping: 1e16cm-3 if not stated otherwise
Simulation Voltage: 1000V
Oxide charge Qox: 1e11cm-2 before irradiation 7e11cm-2 or 1e12cm-2 after irradiation
Envisaged strip sensor layout: pitch 90µm, strip width 20µm, w/p=0.222

30. Summary
P-stop sensors intrinsically better after irradiation – lower electric fields at the strips
200µm thick sensors perform similar to 320µm thick sensors (in terms of electric field)
High p-stop doping concentration can cause high electric fields after irradiation

31. Effect of QF variation for Pstop = 5e16cm-3
E field (V/cm)
For low values of QF, E field peak is under MO as well as near Pstop also
For higher values of QF, very high E field peak near Pstops, which increases with
increase in QF

32. Back up-One of the Rint measurement (Robert Eber)
Simulation
Simulation indicate toward QF ~ 1.2e11 cm-2
Good measurements can be used to predict value of QF using simulations!

33. Simulation of Rint for MSSD with Double P-stops
For p-type of sensor, three strips structure was used for Rint simulations in which bias of 1V is given to Central DC Anode while two neighboring Anodes are shorted together. Reverse bias is provided from cathode (not shown), below while a very low DC external resistance of 1Ω is used to avoid scaling confusion.

34. Simulation of Rint without any bulk damage
- Three different type of Rint curves were observed.
- For low values of QF, good strip isolation was observed.
For intermediate values of QF, strip isolation is very poor for low biases but improves with higher
reverse biases. Electrons from accumulation layer are progressively removed by higher reverse
bias resulting in better Rint.
But for higher values of QF, Rint remain very low even at higher reverse bias.
Further, it can be observed that pstop doping density 5x1015cm-3 is not sufficient (Fig. 2(a) ) to maintain strip isolation with oxide charge density QF= 5x1011cm-2.
Similarly, it can be inferred from figure 2 (b) that without any isolation structure, strip isolation would not be possible, up to 800V, even for QF = 3x1011cm-2.
Figure 2 (a)
Figure 2 (b)
Without any isolation str
With HPK Double Pstops
1x1011
3x1011

35. Why two more acceptors with higher introduction rates ? – continue…
Ionized Acceptor trap density inside Si sensor
Ionized Shallow levels (green and blue) are much less compare to deep levels (Red color).
Ionized Acceptors just below SiO2/Si
Interface
In some of the region, Ionized shallow
traps (green and blue) are much more
compare to deep one
Cutline is 0.1 um below SiO2
Cutline is perpendicular to n+ strip
(Through middle)

36. Why micro discharge is quenched!
E field inside irradiated sensors is a strong function of space charge charge.
So, when breakdown happen near curved area of p+ strip, a lot of free e/h carriers are produced which will change the nearby space charge significantly, changing the electric field, thus stopping the further breakdown.
Moreover, in irradiated sensors, because of presence of high density of traps, free path length of a charge carrier will be very low, particularly in middle of sensors where E field is very low.
These fact would stop the avalanche from turning global and continuous

37. Measurement of E-field in a irradiated Si strip sensor (n+p-) G
Measurement of E-field in a irradiated Si strip sensor (n+p-) G. Kramberger et all , 2009, IEEE conference
E field profile for a non-irradiated sensors
<8000V/cm for reverse bias = 200V
E field profile for a irradiated sensors
Can be as high as 80000V/cm, near the strips for reverse bias = 200V !
- Formation of high density negative space charge near n+ strips
(flux=5e14cm-2)
- The negative space charge will act as Pspray and increases with irradiation !
Hence, we never had much problem of strip isolations in hadron irradiation expt!

38. Electron conc. in the interstrip region decreases as flux increases

39. Rint vs. Vbias (Irradiated) : Strip pitch and Implant width
Flux=1e15cm-2
QF = 1.2e12cm-2
No significant dependence of Rint on changing the strip pitch and width.

40. Summary
Bulk damage and surface damage models are used to investigate the
strip isolation, micro - discharge problem and higher leakage current
for strip sensors
p+n- sensors are more prone to micro - discharge problem
Because of very high electric fields in curved regions of strips, Strip
sensors can have more leakage current compare to diodes
Rint measurement curves can also be understood qualitatively by
simulations
Further tuning of simulations is going on