Transcript poster_PTP_HSTD8_HFWS - Indico

The Punch-through Effect in Silicon Strip
Detectors
C. Betancourt1,2, A. Bielecki1, Z. Butko1, A. Deran1, S. Ely1, V. Fadeyev1, C.Parker1, N. Ptak1, H. F.-W. Sadrozinski1, J. Wright1
1 Santa Cruz Institute for Particle Physics, Univ. of California Santa Cruz, CA 95064 USA
2 Albert-Ludwigs Universität, Freiburg 79104, Germany
Introduction
The typically used AC-coupled silicon
sensors at the LHC can develop very large
voltages across the SiO2 layer, referred as
the coupling capacitance. The metal
readout traces are held close to ground
due to low input impedance of the readout
amplifier, ~1 kOhm. The strip implants
could reach a significant voltage in the
case of large charge accumulation in the
bulk, which can occur in the instance of
the beam loss. In extreme cases the
electric field can collapse causing the
implants of the sensor to float to unknown
voltages. These large voltages on the
implants can reach the order of the bias
voltage, and thus can exceed the
specification for the hold-off voltage of the
coupling capacitor, which are typically
qualified to 100V.
• Readout electronics are connected to AC pads, which are coupled to the DC strip
implants via a layer of SiO2 and nitride.
• The punch-through effect is used to short the strip implants to the grounded bias rail,
thus preventing excess voltages between the DC and the AC pads.
• Special punch-through protection structures (PTP) have been developed to enhance the
punch-through effect in an effort to bring down the punch-through voltage to a relatively
safe value (below the 100V qualification of the coupling capacitors).
• These sensors, part of the ATLAS07 sensor production [1], are broken up into several
zones, which refer to the surface treatment used for strip isolation structures and their
geometry. In addition, zone 4 sensors (A-D) are built with specially designed PTP
structures.
Z1
!
Z2
Theory and Method
Z3
L aser N ear
Z4D
L aser F ar
R bi as
For large current, IV becomes space-charge limited and
is given by
R P T n ear
• Kinks in the Reff vs Vtest
curves represent punchThe Resistance for PTP structures [5]
through to the next
For large currents we are always in the SCL mode, and the
nearest neighbor strips,
resistance of a PTP structure can be written as a function of
which is clearly seen
voltage by
when Measuring the
voltage and current of the
neighboring strips while
And as a function of current by
doing DC tests.
where
Ri m p
Vn ear
-+
-+
-+
-+
-+
-+
Rb
Vf ar
-+
-+
-+
-+
-+
-+
R P T f ar
Rb
Vbi as
• Using the 4-resistor model we can calculate all relevant currents and
resistances from measured implant voltages.
• PTP structures “work”, i.e. limit the implant voltages at a saturation
voltage Vfb’ much larger than VPT.
• At high currents, the implant voltage can be written as
Laser Injection PT Measurements [3]
• DC measurements involve applying a test voltage
between the DC pad of a test strip and the grounded bias
rail, and then measuring resulting current.
• The effective resistance Reff between the implant and the
bias rail, which consists of the bias resistance in parallel
to the punch-through resistance is then calculated and is
given by
Z4C
The 4-R Model and the Implant Voltage
IV characteristics for PTP structures
DC Punch-through Measurements [2]
Z4B
Z4A
• To mimic beam accidents, charge is injected into the
sensor through the use of an IR cutting laser [4].
• The large amount of charge injected by the laser
(~2x107 MIPs/cm2 over a 2mm x 2mm spot) is able to
collapse the electric field inside the sensor, thus
allowing the strip implants to float to large voltages
that are only limited by the bias voltage and the
effectiveness of PTP.
• The parameter α = Rbulk/Rsat determines if we see voltage saturation:
small α => V~Vbias (large Rsat), large α => V~Vfb (low Rsat).
• Sensors with PTP structures (Z4x) show saturation of the implant
voltage since α is large (low Rsat).
• At lower laser intensities, the implant voltage depends on the ratio of
bulk resistance to effective PTP resistance Rbulk / Reff, NOT on the
effective PTP resistance ALONE.
• At lower particle fluxes, Rbulk /Reff increases and thus reduces the
implant voltages and the risk for sensor damage.
4
Rbulk/Reff vs. Attenuation
Rbulk/Reff
3
• The punch-through voltage VPT for strip detectors is then
defined as the point where Reff(VPT)=1/2 Rbias, which is
not the same punch-through voltage as defined in the
literature for npn or pnp structures.
2
1
W71-BZ3
W224-BZ4A (275Vbias)
0
0
1
2
3
4
5
Attenuation (Sheets)
Results
Near Side
Far Side
The Effectiveness of PTP Structures
Conclusions
• Protons and pions both increase the saturation voltage,
while neutrons have no effect. The increase in saturation
voltage already saturates at a relatively low proton fluence of
1e13 neq/cm2 (~1MRad).
•This indicates that the origin of punch-through is mainly from
surface charge, not bulk charge.
• Punch-through protection still works even at high fluences.
P-stop 1e13
P-stop 4e12
P-spray 2e12
P-spray 2e12
p-stop=4e12 cm^-2
130
120
110
Vfb (Volts)
•Zone 4 sensors perform the best (lower
implant voltage and larger α) compared to
zone 2 and 3 sensors due to the reduced
channel length.
• Increasing p-dose density increases the
saturation voltage Vfb.
•RPTP is “low” ~ 10 kW (~ Rbulk) when
current ≥ 1 mA
• Finite implant resistance of 15 kW/cm
isolates the PTP structure from the charge
deposit.
• Large difference between RPTnear and
RPTfar even for non-PTP structures
• The placement of the polysilicon biasing
resistor
significantly
reduces
the
resistance between the implant and the
bias rail.
• This is explained by the fact that the bias
resistor provides a gate in 3 terminal
device (implant, bias resistor, bias rail),
which increases the current flow between
the implant and bias rail.
• Increasing the coverage of the bias
resistor over the channel length increases
the effect of the gate, and hence
increases the effectiveness of PTP
structures.
Radiation Damage
• MICRON sensors have a lower Vfb than HPK
zone 1 sensors, even though they have similar
p-spray concentration and channel length.
• The lower Vfb can be attributed to a larger
gate effect due to the placement of the
polysilicon bias resistor running more directly
over the channel length.
• By placing the biasing resistor directly over
the channel length, a lower Vfb can be
achieved.
Pre-Rad
Fluence=1e13 p/cm^2
100
90
80
70
60
50
Z4A
Z4B
Z4C
Z4D
RC Backplane Filter Network
• Filter resistance at the backplane (10kΩ) gives a drop of
about 10V at 1mA per strip.
• Extending to 100 strips, the voltage drop becomes equal
or greater than the bias, thus dropping the whole voltage
before reaching the sensor.
• Large implant voltages caused by beam losses can cause permanent damage to silicon detectors in high energy particle physics
experiments. Punch-through protection structures are designed to limit these voltages.
• The beam loss conditions are simulated by flooding the sensors with IR laser pulses, and measuring the implant voltages [4]
• The 4 resistor model proposed earlier can accurately determine the punch-through characteristics of a given sensor. The model describes
the detector at breakdown as a circuit with 4 resistors.
• The parameter that determine the effectiveness of PTP structures are the ratio of the bulk resistance to the saturation resistance α [5].
• The finite implant resistance Rimp can isolate the region of electric field collapse from the PTP structure.
• Increasing the p-dose density increases the saturation voltage Vfb for a given zone.
• PTP structures with an effective gate (a la Micron) offer increased protection because the gate effect lowers the saturation voltage.
• Tests on sensors irradiated with protons, pions and neutrons show that Vfb increases arise mainly from surface charges, not bulk charges.
• The PTP technique was developed to deal with particular beam loss scenarios. They relay on large currents flowing through the structures
(mA’s). In cases with distributed particle fluences, the PTP structures are not needed since either the field does not break down (Rbulk/ RPT >>
1), or the filter resistor at the back plane affords sufficient protection, clamping the bias voltage to ground.
Acknowledgments
• We acknowledge the help of the ATLAS and
RD50 collaboration in the design and
procurement of the silicon sensors
• The invaluable assistance of the institutes at
Ljubljana, CERN, PSI and KEK and their staff in
carrying out the radiation.
• We appreciate the discussions with Yoshinobu
Unno and Kazuhiko Hara.
References
[1] Y. Unno et al, NIM A 636 (2011) pp. 24-30
[2] S. Lindgren et al, NIM A 636 (2011) pp. 111-117.
[3] H.F.-W. Sadrozinski et al, NIM A 658 (2011) pp. 46-50
[4] T. Dubbs et al, IEEE Trans. Nucl. Sci. 47 (2000) pp. 1902-1906.
[5] C. Betancourt, Physics M.S. Thesis, UC Santa Cruz (2011)