Interstrip Characteristics of n-on-p FZ Silicon
Detectors
S. Lindgren
c
, C. Betancourt
c
, A. Chilingarov
b
, N. Dawson
c
, V. Fadeyev
c
, H. Fox
b
, K. Hara
d
, H. Hatano
d
, T.
Kohriki
a
, Y. Ikegami
a
, S. Mitsui
d
, H. F.-W. Sadrozinski
c
, S. Terada
a
, Y. Unno
a
, J. Wright
m
, M. Yamada
d
a
KEK, High Energy Accelerator Organization, INPS, 1-1 Oho, Tsukuba, Ibaraki 305-0801, Japan
b
Physics Department, Lancaster University, Lancaster LA1 4YB, United Kingdom
c
SCIPP, UC Santa Cruz, CA 95064, USA
d
School of Pure and Applied Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8571, Japan
Abstract– We report on the measurement of interstrip
parameters of p-type silicon strip sensors which we are
developing in a large collaboration to be used in a future tracker
for the LHC upgrade. We measure on test structures with about 1
cm long strips the interstrip resistance, interstrip capacitance (at
1 MHz) and punch-through protection both pre-rad and after
irradiation with 70 MeV protons to a fluence of 1.5*10^13
p/cm^2, corresponding to about 1 MRad, from prototyping runs
with Hamamatsu Photonics and Micron Semiconductors. We
report the values for a variety of isolation scenarios of p-stops, p-
spray and a combination of both.
I. INTRODUCTION
N
the need for radiation-hard tracking detectors in
forthcoming elementary particle physics experiments, silicon
is regarded to be the best choice as sensor material because of
its unsurpassed material quality, mature technology and low
cost for mass production [1]. Recently, a luminosity upgrade
to 10
35
cm
−2
s
−1
has been proposed to the Large Hadron
Collider (LHC) ("SuperLHC") [2]. To exploit the physics
potential of the upgraded LHC, an efficient tracking down to a
few centimeters from the interaction point will be required,
where fast hadron fluences above 10
16
cm
−2
will be reached
after five years operation [3]. The CERN-RD50 project
“Development of Radiation Hard Semiconductor Devices for
Very High Luminosity Collider” has been formed to explore
detector technologies that will allow to operate devices up to,
or beyond, this limit [4], [5], and the two large all-purpose
experiments ATLAS and CMS have started to plan for an
upgrade of their detectors to exploit the expected higher
luminosity.
One of the most pressing issue for n-in-p strip sensors are
the interstrip characteristics before and after ionizing radiation,
since the electron accumulation layer on the surface needs to
be compensated for large fluences and dose levels. Both
neutrons and protons displace silicon atoms via non-ionizing
energy losses, which results in a bulk damage. Protons in
addition ionize the atoms in their path that leads to permanent
damage at the sensor surface, which makes them the main
radiation species to study for surface damage
.
The interstrip resistance and capacitance are important
parameters used to characterize the effects of surface radiation
damage of silicon strip detectors. The interstrip resistance is
important for strip isolation, so that a sufficiently high
interstrip resistance can prevent signal sharing between
neighbors which could lead to degradation of the position
resolution. The interstrip capacitance is the main contributor of
noise in between strips. A properly functioning detector
should thus try to minimize the interstrip capacitance in order
to have a higher signal-to-noise ratio, while maximizing the
interstrip resistance to minimize crosstalk between strips.
Ac-coupled sensors are susceptible to very large voltages
between the metal readout traces (held to ground through the
front-end electronics) and the strip implants in the case of
large charge accumulation in the bulk, for instance in the case
of beam losses [6]. Since the field inside of the sensor breaks
down, these large voltages on the implants can reach the order
of half the bias voltage, and thus can exceed the specification
for the hold-off voltage of the coupling capacitor, which are
typically tested to 100V. In order to prevent these large
voltages, the punch-through (reach-through) effect is used [6],
where implants in close proximity will effectively be shorted
together if the voltage between them exceeds a geometry
dependent voltage. This provides an effective over-voltage
protection for single strips, which get shorted to the bias line in
the case of voltages in excess of the punch-though protection
("PTP") voltage.
Previous studies with p-type sensors, e.g. within the context
of RD50, were done using p-spray to isolate the n-strips. This
study uses “mini-SSDs” (~1 cm long) produced by Hamamatsu
Photonic (HPK) within the ATLAS upgrade program. The
isolation is done with p-stops of varying geometry, p-spray and
both combined with p-doses (concentration) varying from 0 up
to 2*10
13
p/cm
2
.
This paper is presented as follows. Section II describes the
samples and irradiation. Section III describes interstrip
capacitance and resistance measurements. Section IV
describes punch-through protection. Section V presents results.
Section VI presents conclusions as well as discussion.
II. S
AMPLES AND IRRADIATION
The sample sensors were fabricated using 15 cm wafers with
<100> crystal orientation and 320 μm thickness. The wafers
we report in this paper are FZ grown (FZ p wafers in [7])
having fewer defects than normal FZ wafers. The R&D group
continues to evaluate other commercially available p-type
wafers [7]. The strip pitch is 74.5 μm. The characteristics of
irradiated sensors are studied using miniature samples of 10
mm square, where there are 104 strips of 8 mm length.
I
2009 IEEE Nuclear Science Symposium Conference Record N08-2
U.S. Government work not protected b
y
U.S. cop
y
ri
g
h
t
157
Authorized licensed use limited to: Fermilab contact x3401/library@fnal.gov for help. Downloaded on February 24,2010 at 12:55:08 EST from IEEE Xplore. Restrictions apply.
Fig. 1. Different p-stop structures as indicated by the different zone numbers
It is thought that n-on-p detectors are more sensitive to
surface effects than p-on-n detectors. One concern is the risk
that the fixed oxide charges in the Si-SiO
2
interface would lead
to a conductive layer of electrons at the surface [8]. Within
the project ATLAS07 for the ATLAS upgrade different
structures for mini-SSDs have been produced. The different
structures use the concept of preventing those damages by
surface treatments, positive doped implants (p-impurities) in
form of p-stop or p-spray, or combinations of both.
The p-stops are implanted to the detectors with a mask while
p-spray is sprayed on. Several doses and combinations of p-
stop and p-spray have been applied to different sensors.
Different structures to apply the p-stops have also been used,
which is indicated by different zone number and seen in Fig. 1.
Detectors with zone 1 have no structure, i.e. they have only
p-spray since the p-stop mask was left out. Detectors with zone
2 have individual p-stops, i.e. each strip is surrounded by p-
implants in opposite to the other structures which only have a
line of common p-implants between the strips. Zone 3 shares
the p-stops between the strips and zone 4 has additional punch-
through protection structure, which will be discussed more
later in the paper. Detectors with zone 5 have narrow metal,
meaning the aluminum layer over the strips do not reach
outside the strip itself and finally zone 6 is similar to zone 3
but with a wider strip pitch.
The proton irradiation was performed at Cyclotron Radio
Isotope Center (CYRIC) of Tohoku University. Details of
irradiation facility and methods are described in [7, 9]. 70-
MeV protons were uniformly irradiated by scanning
periodically the sample sensors. The irradiation took typically
a few 10 minutes to a few hours depending on the fluence. The
sensors were kept cooled at -10
o
C during irradiation and the
irradiated samples were immediately stored in refrigerator to
suppress any post-irradiation annealing to take place. The
fluence we refer is in 1-MeV neutron equivalent value taking
into account the NIEL factor of 1.4. The fluence uncertainty is
determined by the
27
Al(pn) reaction cross section, which is no
more than 10%.
III. I
NTERSTRIP CAPACITANCE AND RESISTANCE
MEASUREMENTS
Interstrip resistance measurements were performed in a
probe station. As seen in Fig. 2, the DC pad of a test strip was
connected to an Agilent 4156C Precision Semiconductor
Parameter Analyzer and grounded while a voltage V
2
was
Fig. 2. Measurement set-up for interstrip resistance. Visual illustration of the
measurement (top), showing the connections to the strips via the DC pads and
to the bias ring. An equivalent circuit diagram (bottom) is also displayed.
applied to the DC pads of two of its closest neighbors, which
were also connected to the analyzer. The detectors were
biased using a Keithley 2410 HV Source Meter and each
measurement were performed at several bias voltages ranging
from 5V to 300V. The voltage V
2
applied to the neighbor
strips was varied through the parameter analyzer from -1V to
1V in 100mW steps. Each measurement was performed at 22°
C with nitrogen gas flowing over the detector for moisture
control.
The idea is that the voltage V
2
on the neighbors will induce
a current I
1
through the test strip. The interstrip resistance can
then be determined by
(1)
where the factor of 2 comes from the fact that two neighboring
strips are used. The resulting IV curve is shown in Fig. 3.
To measure the interstrip capacitance a slightly different set-
up was used. Instead of using the DC pads, the test strip and
its neighbors were connected to an Agilent E4980A Precision
LCR meter via the AC pads. In addition, the next two
neighbors (two strips away from the test) were grounded to act
as a shield from the outer laying strips. The frequency of the
AC signal from the LCR meter was varied, and each
measurement was taken at 10 kHz, 100 kHz, 1 MHz and 2
MHz. The bias voltage ranged from 0 to -800V with a step
size of 50V. The measurements were again performed at 22°
C with nitrogen gas flowing over the detector.
IV. P
UNCH-THROUGH PROTECTION MEASUREMENTS
Protection against large voltages between strip implants
and the metal traces can be achieved with the punch-through
effect between strips and bias ring. This is trivially done for p-
spray, but requires sophisticated structures in the case of p-
158
Authorized licensed use limited to: Fermilab contact x3401/library@fnal.gov for help. Downloaded on February 24,2010 at 12:55:08 EST from IEEE Xplore. Restrictions apply.
y = -0.27x + 0.16
y = -0.86x + 0.14
y = -1.31x + 0.11
y = -1.94x + 0.07
y = -2.80x + 0.05
-4
-3
-2
-1
0
1
2
3
4
-1 -0.5 0 0.5 1
I
1
(μA)
V
2
(Volts)
300 volts
100 volts
50 volts
20 volts
5 volts
Fig. 3. The current through the test strip vs. the voltage applied to the
neighbors shown at different bias votlages. The interstrip resistance is given
by the inverse of the slope.
Z4A Z4B
Z4C Z4D
Fig. 4. The different PTP structures designs that have been tested for the
paper.
stops. In this paper, the effects of several PTP structures on
the PTP voltage have been studied. In particular, zone 4
detectors were designed with complicated PTP structures, as
seen in Fig. 4.
Unirradiated detectors were measured at Lancaster
University in a probe station kept at 21° C. Measurements on
irradiated detectors were carried out at the University of
Tsukuba where detectors were placed two at a time on printed
circuit boards which were then placed in a thermostat chamber
set at -20° C with nitrogen gas flowing in for moisture control.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
-50 -40 -30 -20 -10
0
Effective Resistance (MΩ)
Test Voltage (Volts)
w42-bz2
w42-bz4d
w44-bz3
w44-bz4d
w25-bz3
w33-bz3
w02-bz3
w11-bz4a
w11-bz4b
w11-bz4c
w11-bz4d
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
-50 -25 0 25
50
Effective Resistance (MΩ)
Test Voltage (Volts)
Z3
Z4A
Z4B
Z4C
Z4D
Φ = 1.2e13 neq
Total p-dose = 4e12 cm
-2
p-stop only
R
PT
= R
bias
Fig. 5. The effective resistance vs. the test voltage applied for a) unirradiated
detectors and for b) detectors irradiated with protons to 1.2x10
13
neq. The
horizontal arrow indicates where the punch-through resistance is equal to the
value of bias resistor, and the vertical arrow indicates the corresponding
punch-through voltage.
For PTP measurements, the effective resistance is measured
between the DC pad of a strip and the bias ring. The detectors
were reversed biased to 300V with voltage being applied to
the backplane and the bias ring grounded. A test voltage V
test
is then applied to the DC pad, and the subsequent induced I
test
current was measured between the pad and the bias ring. The
neighboring strips were left floating. The effective resistance
is then given by
(2)
The effective resistance can be thought of as the bias
resistor in parallel to another resistance, which we call the
punch-through resistance. The effective resistance can then be
written as
(3)
where R
bias
is value of the bias resistor and R
PT
is the punch-
through resistance.
The punch-through voltage V
PT
is then taken to be the
voltage where
(4)
159
Authorized licensed use limited to: Fermilab contact x3401/library@fnal.gov for help. Downloaded on February 24,2010 at 12:55:08 EST from IEEE Xplore. Restrictions apply.
1.0E+05
1.0E+06
1.0E+07
1.0E+08
1.0E+09
1.0E+10
1.0E+11
1.0E+12
1.0E+13
0 50 100 150 200 250 300
Resistance (Ω)
Bias Voltage (Volts)
W46-BZ3-P1 (PRE RAD)
W46-BZ3-P1 (POST RAD)
W44-BZ2-P2 (POST RAD)
W40-BZ2-P2 (PRE RAD)
W40-BZ2-P2 (POST RAD)
W40-BZ3-P1 (PRE RAD)
W40-BZ3-P1 (POST RAD)
W33-BZ3-P1 (PRE RAD)
W33-BZ3-P1 (POST RAD)
W23-BZ3-P15 (PRE RAD)
W23-BZ3-P15 (POST RAD)
W02-BZ2-P2 (PRE RAD)
W02-BZ2-P2 (POST RAD)
W02-B Z4B-P10 (PRE-RAD)
W02-BZ4B-P10 (POST-RAD)
W02-BZ5-P11 (PRE-RAD)
W02-BZ5-P11 (POST-RAD)
Total p-dose = 1*10
12
1/cm
2
Total p-dose = 2*10
12
1/cm
2
Total p-dose = 4*10
12
1/cm
2
Pre-rad
Φ = 1.14x10
13
neq
10
13
10
12
10
11
10
10
10
9
10
8
10
7
10
6
10
5
Fig. 6. Interstrip resistance for irradiated series 3 detectors. There is a clear
dependence on the total p-dose applied after irradiation.
y = 1.43E+06e
1.90E-
12x
1.00E+06
1.00E+07
1.00E+08
1.00E+09
1.00E+10
1.00E+11
1.00E+12
02E+124E+126
E+12
Rint (Ohms)
Total P-Dose (1/cm^2)
Rint
Fit
(1/cm
2
)
R
int
(Ω)
10
12
10
11
10
9
10
8
10
7
10
6
2x10
12
4x10
12
6x10
12
10
10
Φ = 1e13 neq
Fig. 7. Interstrip resistance vs. the total p-dose applied for detectors irradiated
with protons to 1x10
13
neq. Higher p-doses lead to higher interstrip
resistance after irradiation.
or alternatively the point where the effective resistance is equal
to half the value of the bias resistor The effective resistance as a
function of the test voltage for an unirradiated detector is shown
in Fig. 5a while irradiated detectors are shown in Fig. 5b.
V. R
ESULTS
To first order, the interstrip resistance does not seem to
depend on the specific zone, but instead depends only on the total
amount of p-dose applied to the surface (through p-stop or p-
spray), as seen in Fig. 6. There is an obvious correlation between
the total amount of p-dose applied and the value of interstrip
resistance after irradiation, with a higher total p-dose resulting in
better post-rad strip isolation (higher interstrip resistance), which
is illustrated in Fig. 7.
The interstrip capacitance shows little change after irradiation,
and seems to be dependent on the specific zone, but not really on
the amount of total p-dose applied. Still, the dependence is weak,
and no predictions could be made about whether adding a p-stop
mask would increase or decrease, or by which amount, the
interstrip capacitance. Zone 5 is the only exception to this,
consistently showing an increase in capacitance after irradiation.
It is helpful to plot the interstrip resistance vs. the interstrip
capacitance. This helps determine which detectors have the most
favorable behavior (highest resistance and lowest capacitance).
Detectors performing the best should lie in the upper-left corner
1.0E+06
1.0E+07
1.0E+08
1.0E+09
1.0E+10
500 600 700 800 900
Resistace (Ω)
Capacitance (fF)
W46-BZ3-P1
W44-BZ2-P2
W40-BZ3-P1
W35-BZ1-P19
W35-BZ2-P17
W35-BZ5-P11
W33-BZ3-P1
W31-BZ5-P11
W25-BZ5-P11
W12-BZ3-P1
W02-BZ1-P7
W02-BZ2-P2
W02-BZ4A-P4
W02-BZ4B-P10
W02-BZ4C-P16
W02-BZ4D-P22
W1-BZ5-P11
W1-BZ2-P17
W1-BZ1-P19
10
10
10
9
10
8
10
7
10
6
Zone 5
Zone 4
Φ = 1e13 neq
Fig. 8. Scatter plot of the interstrip resistance vs. interstrip capacitance for
detectors irradiated to 1x10
13
. Zone 5 (black circle) tends to perform the
worst compared to the other zones.
0
5
10
15
20
25
30
35
40
45
50
0 200 400 600 800 1000
Current (μA)
Bias Voltage (V)
W02-BZ1-P7
W02-BZ3-P1
W02-BZ4A-P4
W02-BZ4B-P10
W02-BZ4C-P16
W02-BZ4D-P22
W02-BZ5-P11
W12-BZ3-P1
W23-BZ3-P15
W33-BZ3-P1
W44-BZ2-P2
W46-BZ3-P1
Φ = 1e13 neq
Fig. 9. Leakage current vs. bias voltage for various detector types irradiated to
1x10
13
neq taken at 22°C. Zone 3 breaks down the earliest at ~900V, which
is still greater than the 600V breakdown specified by the manufacturer.
of the graph, while the least preferable behavior is seen in the
lower-right of the graph. Fig. 8 shows the interstrip resistance vs.
capacitance for several detectors of different zones and total p-
dose. One can see that zone 5 tends to have the worst interstrip
properties after irradiation.
The leakage current as a function of the bias voltage was
measured in order to determine if good strip isolation came with
the advantage of a lower breakdown voltage. The IV curves for
various irradiated detectors taken at room temperature are shown
in Fig. 9. It can be seen that all most detectors have a breakdown
voltage in excess of 1000V, while three detectors (all zone 3)
exhibit breakdown at about 900V. The high breakdown voltage
should not significantly affect our measurements, as the interstrip
resistance is taken at 300V, and the capacitance at 800V. It is
also important to note that all detectors exhibit breakdown at a
much higher voltage than the breakdown voltage of 600V that
was specified by the manufacturer.
The punch-through voltage for unirradiated detectors is
between 10-30V, as seen in Fig. 10a. There seems to be a clear
wafer correlation on the value of the punch-through voltage, i.e.
the voltage depends on the total p-dose and type (p-spray, p-stop).
After irradiation, a dependence of the punch-through voltage on
the total p-dose and type is still evident, which is seen in Fig. 10b.
It is interesting to note that the punch-through voltage for zone
3 detectors and zone 4 detectors is similar before and after
irradiation. This implies that the complicated punch-through
protection structures are not needed, as zone 3, which has a
distance of 70μm from the n+ implant to the bias rail, seems to
provide adequate protection. Further, as seen in Fig. 11, zone 3
160
Authorized licensed use limited to: Fermilab contact x3401/library@fnal.gov for help. Downloaded on February 24,2010 at 12:55:08 EST from IEEE Xplore. Restrictions apply.
0
5
10
15
20
25
30
W42-BZ2-P8
W42-BZ4D-P22
W44-BZ3-P21
W44-BZ4d-P22
W02-BZ3-P6
W11-BZ4A-P4
W11-BZ4B-P10
W11-BZ4C-P16
W11-BZ4D-P22
W25-BZ3-P9
W33-BZ3-P13
PT Voltage (Volts)
Total p-dose
= 2e12
Total p-dose
= 4e12
Total p-dose
= 4e12
Total p-
dose =
2e12
Total p-
dose =
4e12
p-spray only = p-spray + p-stop = p-stop only =
0
10
20
30
40
50
60
W43-BZ3-P3
W43-BZ4A-P4
W43-BZ4B-P10
W43-BZ4C-P16
W47-BZ3-P9
W47-BZ4A-P4
W47-BZ4B-P10
W16-BZ3-P6
W16-BZ4A-P4
W16-BZ4B-P10
W16-BZ4C-P16
W16-BZ4D-P22
W42-BZ3-P6
W45-BZ4A-P4
W45-BZ4B-P10
W45-BZ4C-P16
W45-BZ4D
W46-BZ3-P6
W49-BZ4A-P4
W49-BZ4B-P10
W49-BZ4C-P16
W49-BZ4D-P22
PT Voltage (Volts)
Total p-dose
= 4e12
Total p-
dose =
1e13
Total p-dose
= 1e13
Total p-dose
= 2e13
Total p-
dose =
2e12
Φ=1.2e13 neq
p-spray only = p-spray + p-stop = p-stop only =
Fig. 10. Punch-through voltage for a) unirradiated detectors and b) detectors
irradiated to 1.2x10
13
neq. Both pre and post-rad detectors show a
dependence on the total p-dose applied.
0
2
4
6
8
10
12
14
16
18
20
pre-rad
1.35x1012
1.35x1012
1.18x1015
PT Voltage (Volts)
Zone 3
Total p-dose = 4e12 p-stop only
1.35x10
12
1.20x10
13
1.18x10
15
Fig. 11. Punch-through voltage for zone 3 measured at various fluences. Total
p-dose = 4x10
12
cm
-2
, fluence values are in neq.
detectors exhibit adequate protection even at high fluences,
having a punch-through voltage in the range of 7-20V. This
further suggests that complicated punch-through protection
structures are not necessary
Measurements of the effective resistance taken at different
temperatures indicate that there is clear temperature dependence,
as seen in Fig. 12. The temperature dependence of the effective
resistance would at first seem to complicate the extraction of the
punch-through voltage, but it has been found that although the
resistance is dependent on temperature, the value of the punch-
through voltage is independent of the temperature. The
dependence on temperature seems to come mostly from the poly-
silicon bias resistor.
0.00E+00
5.00E+05
1.00E+06
1.50E+06
2.00E+06
-70-60-50-40-30-20-10
0
Effective Resistance (Ω)
Test Voltage (Volts)
-20deg
0deg
20deg
Fig. 12. The effective resistance measured at three different temperatures.
There is a clear dependence on the temperature.
The effective resistance R
eff
can also be defined by taking the
integral form of equation 2. The integral definition would have
the added benefit of incorporating the total current that can be
drained from the strip to the bias rail and providing the effective
resistance for charges to escape through. As can be seen from Fig
13, the disadvantage to this definition is that it is less sensitive to
the onset of punch through, so it is less suitable for defining the
punch-through voltage. This leads to a higher punch-through
voltage, culminating in differences between 5-30Volts from the
definition used in this paper.
VI. C
ONCLUSIONS AND SUMMARY
The interstrip resistance decreases and interstrip capacitance
increases after irradiation. To first order, the interstrip resistance
does not depend on the specific zone of the detector, but instead
depends on the total dose of the p-impurities applied. The higher
the total dose of impurities, the interstrip capacitance shows little
change after irradiation and is dependent on the specific zone.
Specifically, zone 5, which has narrow metal, tends to have the
highest interstrip capacitance after irradiation, without providing
any benefits such as a higher breakdown voltage. The higher
interstrip capacitance for zone 5 is not surprising given the
shorter distance between readout strips compared to other zones.
Even if the interstrip resistance is low, one would only expect
to see a loss in signal and increase in noise if the values were on
the same order of magnitude as input impedance of the signal
readout chip, which is typically about 1kΩ. The signal gets
coupled to the metal strip and to the amplifier instead of being
shorted to the neighbor via the DC conductance. So as long as
the interstrip resistance is large compared to the input impedance
of the chip, the signal will end up in the amplifier. Further, there
has not yet been any observed signal degradation or noise
increase due to a decrease in the interstrip resistance.
The leakage current for zone 3 detectors tends to start
increasing at slightly lower voltages than other detectors, but the
voltage is above 900 V and is much greater than the 600V
breakdown voltage specified by the manufacture.
Measurements of the punch-through voltage reveal that there
is a clear wafer correlation. In particular, when looking at a
161
Authorized licensed use limited to: Fermilab contact x3401/library@fnal.gov for help. Downloaded on February 24,2010 at 12:55:08 EST from IEEE Xplore. Restrictions apply.
