Punch-through Protection of SSDs
in Beam Accidents
H. F.–W. Sadrozinski
with
C. Betancourt, A. Bielecki, Z. Butko, A. Deran, V. Fadeyev,
S. Lindgren, C. Parker, N. Ptak, S. Sattari, J. Wright
SCIPP, Univ. of California Santa Cruz, CA 95064 USA
•
•
•
•
•
Large Voltages on Readout Implants in Beam Accidents
Punch-through Protection against Large Voltages on Implants
Simulation of Field-breakdown with an IR Laser
Explanation of large voltages: DC 4-R Model
Role of Implant Resistance
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
1
Damage from Beam Losses to AC-coupled SSD
• ALEPH at LEP observed break-down of the coupling
capacitors on AC-coupled SSD in a beam accident. The
Al readout trace of AC coupled sensors are held at
ground by the readout ASIC.
• We proposed that this was due to the breakdown of
the field inside the sensor when the deposited charge
made the sensor conductive. At that point, the bias
voltage can reach into the sensor bulk and can impart
large voltages to the implants.
• To check this, we used IR lasers to mimic the beam
loss and we indeed observed large voltages on the
implants
T. Dubbs et al., IEEE Trans. Nuclear Science 47, 2000:1902 – 1906.
• The reach-through (punch-through) effect was considered an elegant and effective way to
limit the voltages on the implants. Special punch-through protection (PTP) structure can be
designed where the geometrical layout determines the voltage limits on the implants.
J. Ellison et al., IEEE Trans. Nuclear Science, 36, 1989: 267 - 271.
• PTP structures were implemented in the p-on-n SCT sensors.
• But the PTP structure in SCT sensors were shown not to guarantee protection against
large voltages across the coupling capacitors when IR laser pulses were used.
K. Hara, et al., Nucl. Instr. and Meth. A 541 (2005) p. 15-20.
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
2
PTP in Upgrade Sensors ATLAS07
N-on-P full-size sensors and “mini’s” to investigate PTP (in Zone 4A-4D) Y.
Nobu, et al., NIMA A (2010)doi:10.1016/j.nima.2010.04.080
No PTP structure
PTP structures
BZ4A
BZ4B
BZ4C
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
BZ4D
3
PTP Structure Effectiveness
•
•
The effectiveness of PTP
structures is determined in DC i-V
measurements between the strip
and the bias ring. One measures
the “integral” effective resistance
Reff, which is the bias resistor Rbias
in parallel to the PTP resistor RPTP.
The measure of the effectiveness
of the PTP structure is the punchthrough voltage VPT , defined as
the voltage at which
RPTP = Rbias, i.e. Reff, = 0.5* Rbias .
The PT voltage VPT was
measured to be a few 10’s of Volt.
2.0E+06
W51 p-stop=4e12, Vbias=200V
W51-BZ2-P8
W51-BZ3-P3
W51-BZ4A-P4
W51-BZ4B-P10
1.5E+06
Reff (Ohms)
•
1.0E+06
5.0E+05
0.0E+00
-50
-40
-30
-20
Vtest (Volts)
-10
0
S. Lindgren, et al., NIM A (2010) doi:10.1016/j.nima.2010.04.094
L [mm]
VPT
BZ4A
21
20
BZ4B
21
24
75 / 67
21
But:
VPT shows very little variation with the
dimension of the PTP structure
(channel length L):
BZ2 / BZ3
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
4
Testing Large Implant Voltages with Laser
T. Dubbs et al., IEEE Trans. Nuclear Science 47, 2000:1902 – 1906.
• Alessi IR cutting laser deposits large amounts of charge inside the detector which
collapses the field(>1010 e/h pairs ~ 106 MIPs ~ 1 Rad / pulse).
• Intensity given by number of laser triggers ~ 4 msec apart (we used up to 3).
• Laser spot ~ 10 mm, but large DC voltages
extend over few mm.
• Peak voltage independent of laser intensity,
and AC grnd/floating
-180
W51-BZ4C Laser Profile Strip 10, 200Vbias
-160
Voltage On Implant (V)
• Bias ring is held to ground
• Voltage on a DC pad and/or AC pad are read
via an high impedance voltage divider
into pico-probe or digital scope.
• DC pad reflects biasing of strips,
• AC pad reflects instantaneous collected charge
( ~ depleted region)
1 L Puls
3 L Pulses
-140
-120
-100
-80
-60
-40
-20
0
-10
0
10
20
30
40
Distance Laser - Strip (# of Strips)
Voltages on strips are large, much larger than DC VPT , comparable to bias voltage.
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
5
Laser implant voltages near RPT vs. Bias
250
250
Laser near
Vnear vs. Vbias, LP=505, LN, AC Ground
W51-BZ4A-P4
200
200
W51-BZ4B-P10
Vnear [V]
W51-BZ4C-P16
DC Voltage
for RPT
W51-BZ4D-P16
150
150
W51-BZ2-P8
W51-BZ3-P3
100
100
50
50
DC VPT
0
0
50
100
150
200
250
300
0
350
Vbias [ V ]
At high bias voltages, implant voltages for PTP structures saturate.
But the ones without PTP structures do not saturate
(even though DC VPT were very similar)
Can this be reconciled with the DC voltage dependence of RPT ?
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
6
DC Characteristics of PTP Structures
Extend the DC measurements to larger voltages and currents.
Reff vs. Vnear
Reff vs. PT current iPT
1.E+07
1.E+07
W51 p-stop=4e12, Vbias=200V
W51 p-stop=4e12, Vbias=200V
W51-BZ2-P8
W51-BZ3-P3
W51-BZ4A-P4
W51-BZ4B-P10
1.E+06
Reff (Ohms)
Reff (Ohms)
1.E+06
1.E+05
1.E+05
W51-BZ2-P8
W51-BZ3-P3
W51-BZ4A-P4
W51-BZ4B-P10
1.E+04
-150
-130
-110
-90
-70
-50
-30
-10
Vtest (Volts)
RPT never becomes a short
Reff = 20 kW reached at different
voltages for different structures.
Strip Voltage depends on
resistance of bulk and implant.
10
1.E+04
1.E-06
1.E-05
1.E-04
itest (A)
1.E-03
1.E-02
Reff ~ 1/ iPT
Independent of structure!
Reff = 10 kW reached at current of
about 10 mA .
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
7
Space-Charge Limitation in PTP Structures
J.l. Chu, G. Persky, and S.M. Sze, J. Appl. Phys., Vol. 43, No.8, August 1972
J. Lohstroh et al.
Solid-State Electronics.
Vol. 24, No. 9, pp. 805-814, 1981
IPT vs. VPT
iPTnear, LF (open) vs. VPT,
LP=505, AC Ground
6.E-03
5.E-03
SCL I ~ V
iPT near [ ]
4.E-03
3.E-03
2.E-03
W51-BZ4B-P10-N
Punch-Through
W51-BZ4C-P16-N
W51-BZ4D-P16-N
I ~exp(V)
1.E-03
W51-BZ2-P8-N
W51-BZ3-P3-N
W51-BZ4A-P4
0.E+00
0
20
40
60
80
100
120
140
160
180
200
Vnear [ V ]
Punchthrough Region: IPT ~ exp(V), Rpt ~ 1/iPT, VPT saturates
SCL Region: iPT~ V, Rpt ~ const1/iPT+const2, RPT saturates
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
8
More evidence for SCL
1.E+06
RPTnear, LF (open) vs. iPT,
LP=505, AC Ground
RPT near [ ]
RPT ~ 1 / iPT
1.E+05
for all PTP structures,
W51-BZ4B-P10-N
W51-BZ4C-P16-N
W51-BZ4D-P16-N
W51-BZ2-P8-N
W51-BZ3-P3-N
RPTnear, LF (open) and RPTfar, Ln (closed) vs. iPT,
LP=505, AC Ground
W51-BZ4A-P4
1.E-04
1.E+04
1.E-02
1.E-03
3.0E+05
W51-BZ4B-P10-F
iPT near [ A ]
W51-BZ4C-P16-F
2.5E+05
…but saturation of RPT at ~ 40kW
for BZ2 and BZ3
SCL Region: RPT ~ const1 / iPT+ 40kW
RPT near or RPT far [ ]
W51-BZ4D-P16-F
W51-BZ2-P8-N
W51-BZ3-P3-N
2.0E+05
W51-BZ2-P8-F
W51-BZ3-P3-F
1.5E+05
1.0E+05
5.0E+04
0.0E+00
0.E+00
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
1.E-03
2.E-03
3.E-03
iPT near or ifar [ A ]
4.E-03
5.E-03
9
Comparison between RPT(near) and RPT(far)
RPTnear, LF (open) and RPTfar, Ln (closed) vs. VPT,
LP=505, AC Ground
1.E+06
W51-BZ2-P8-N
1.E+06
W51-BZ3-P3-N
RPT near or RPT far [ ]
W51-BZ3-P3-F
W51-BZ2-P8-F
1.E+05
1.E+04
0.00
1.E+05
50.00
100.00
150.00
200.00
1.E+04
250.00
Vnear or Vfar [ V ]
Difference of “identical” near and far
resistances in BZ2 and BZ3:
Strong gate effect of the Poly bias resistor
(c.f. FOXFET)
Also: in BZ4A the Poly resistor has the
best channel coverage H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
10
Propose a DC “4R” Model
• After breakdown of field inside the sensor, deal with DC resistor chain only
RPTnear = Reff (RPTnear ,Rbias )
RPTfar = RPT on the far end of the strip, RPTfar > RPTnear
Rimp = Resistance of implant 15kW/cm
To check 4R Model:
Rbulk = Bulk Resistance
Fire laser both near and far.
Measure both Vnear and Vfar
Calculate Vnear = Vfar- Rimmp*Inear
250
Vnear vs. Vbias, LP=505, LN+LF, AC Ground
W51-BZ4A-P4
W51-BZ4B-P10
W51-BZ4C-P16
W51-BZ4D-P16
W51-BZ2-P8
W51-BZ3-P3
W51-BZ4A-P4
W51-BZ4B-P10
W51-BZ4C-P16
W51-BZ4D-P16
W51-BZ2-P8
W51-BZ3-P3
Vnear (Volts)
200
150
Current regulates the punchthrough resistance through
RPT ~ 1 / iPT
200
150
100
100
50
50
Voltages near RTP structures clamped
independent of laser position
0
What about bulk resistance?
250
0
50
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
100
150
200
250
300
0
350
Vbias (Volts)
11
Bulk Resistance
One more parameter after breakdown:
Bulk resistance measured consistently Rb ≈ 15 kW
(independent of the current)
5.E+04
5.E+04
W51-BZ2-P8-N
Rbulk vs ibulk
An approximately constant
W51-BZ3-P3-N
4.E+04
4.E+04
W51-BZ4C-P16-N
bulk resistance
W51-BZ4D-P16-N
W51-BZ4A-P4
Rb ≈ 15 kW
depending on the inverse
Rbulk [ ]
and a PTP resistance
W51-BZ4B-P10-N
3.E+04
3.E+04
2.E+04
2.E+04
1.E+04
1.E+04
of the current
RPT ~ 1 / IPT
can explain the saturation
of the PTP voltages.
0.E+00
0.E+00
2.E-03
4.E-03
6.E-03
8.E-03
0.E+00
1.E-02
ibulk [ A ]
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
12
Effect of Finite Implant Resistance Rimp
Fire laser at the far end of the 1 cm strip, measure both Vnear and Vfar
Laser far
250
250
250
Vfar vs. Vbias, LP=505, LF, AC Ground
Vnear vs. Vbias, LP=505, LF, AC Ground
200
200
200
W51-BZ4B-P10
W51-BZ4C-P16
W51-BZ4D-P16
150
150
W51-BZ2-P8
W51-BZ3-P3
W51-BZ4A-P4
100
50
50
0
0
50
100
150
200
250
300
0
350
Vfar > Vnear
Saturation in Vnear
No saturation in Vfar
RTP structure not effective
for Vfar
Vfar (Volts)
Vnear (Volts)
W51-BZ4D-P16
100
200
W51-BZ4B-P10
W51-BZ4C-P16
150
250
150
W51-BZ2-P8
W51-BZ3-P3
W51-BZ4A-P4
100
100
50
50
0
0
50
100
Vbias (Volts)
150
200
Vbias (Volts)
250
300
0
350
Limitation of application of PTP structures: finite Rimp
Implant Voltages do not saturate at high bias voltages, if finite implant
resistance Rimp isolates PTP structure from breakdown region.
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
13
Conclusions
• An IR cutting laser simulates the beam accident conditions:
large Q, collapsing E-field.
• Field breakdown size ~ 0.5 mm, with a much larger area with partial breakdown .
• Voltages on the implants V > VPTP (DC), due to finite RPTP.
• The observed voltages can be explained by 4 resistor DC model :
RPTP(near), RPTP(far), Rbulk , Rimplant.
• For PTP structures, the voltage near the laser VPTP(laser, near) saturates.
Then RPTP ~20kW; I ~ 5 -10 mA.
• Saturation voltage V = 50 – 100V
• RPTP ~ 1 / iPT, while Rbulk ~15kW. For no PTP structure, RPTP saturates at ~ 40kW (SCL region)
• Rimplant is very important: the present value, ~15 kW/cm, can effectively isolate the collapsed
field from the PTP structure, increasing voltages on the implants by 100’s of volts => need low
Rimplant!
• The effect of the R-C biasing network at the backplane will be studied next.
2010/11/02
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
14
Acknowledgements
The early work on the PTP measurements on ATLS07 sensors from
HPK was shared within the ATLAS Upgrade Collaboration with:
Y. Unno, S. Terada, Y. Ikegami, T. Kohriki, S. Mitsui
Inst. of Particle and Nuclear Study, KEK, Oho 1-1, Tsukuba, Ibaraki 305-0801, Japan
K. Hara, N. Hamasaki, Y. Takahashi,
Univ. of Tsukuba, Inst. of Pure and Applied Sciences, Tsukuba, Ibaraki 305-9751, Japan
A. Chilingarov, H. Fox
Physics Department, Lancaster University, Lancaster LA1 4YB, United Kingdom
H. F.-W. Sadrozinski, PTP RD50 CERN Nov 2010
15