вход по аккаунту



код для вставкиСкачать
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
Contents lists available at ScienceDirect
Nuclear Inst. and Methods in Physics Research, A
journal homepage:
Charge-collection properties of irradiated depleted CMOS pixel test
I. Mandić a, *, V. Cindro a , A. Gorišek a , B. Hiti a , G. Kramberger a , M. Zavrtanik a , M. Mikuž a,b ,
T. Hemperek c
Jožef Stefan Institute, Jamova 39, Ljubljana, Slovenia
University of Ljubljana, Faculty of Mathematics and Physics, Jadranska 19, Ljubljana, Slovenia
Physikalisches Institut, Universität Bonn, Nußallee 12, 53115 Bonn, Germany
Particle-tracking detectors (Solid-state
Radiation-hard detectors
Si microstrip and pad detectors
Solid-state detectors
The edge transient-current technique (Edge-TCT) and charge-collection measurements with passive test structures made with the LFoundry 150-nm CMOS process on a p-type substrate with an initial resistivity of over
3 kΩcm are presented. The measurements were made before and after irradiation with reactor neutrons up to
2⋅1015 neq /cm2 . Two sets of devices were investigated: unthinned (700 μm) with the substrate biased through the
implant on top and thinned (200 μm) with a processed and metallised backplane.
The depletion depth was estimated with the Edge-TCT and the collected charge was measured with a90 Sr
source using an external amplifier having a 25-ns shaping time. The depletion depth for a given bias voltage
decreased with the increasing neutron fluence, but it was still larger than 70 μm at 250 V after the highest
fluence. After irradiation a much higher collected charge was measured for the thinned detectors with a processed
backplane compared to the unthinned devices, although the same or an even larger depletion depth was measured
in the unthinned devices. The most probable value of the collected charge of over 5000 electrons was measured
with a thinned device also after irradiation to 2⋅1015 neq /cm2 . This is sufficient to ensure the successful operation
of these detectors at the outer layer of the pixel detector in the ATLAS experiment at the upgraded HL-LHC.
1. Introduction
The upgrade of the Large Hadron Collider (LHC) to the High Luminosity LHC (HL-LHC) [1] foreseen in the next decade will significantly
increase the rate of proton collisions at the interaction points. As a
result, the number of charged tracks generated in each bunch crossing
will increase, leading to a harsher radiation environment [2]. This will
require the replacement of the present pixel detectors in general-purpose
experiments like ATLAS [3,4] and CMS [5] because their granularity and
radiation hardness are not sufficient for the HL-LHC. In the upgraded
ATLAS experiment [6] large areas will be covered with silicon detectors,
so the cost, production time, complexity of assembly and tracking
performance are critical for the success of the project. All these issues
could be greatly improved with the use of monolithic depleted pixel
detectors produced in a commercial CMOS process on large wafers in
high-volume foundries.
In the HL-LHC proton collisions will occur every 25 ns [6] and this
sets the requirement for the charge-collection time. Good performance
in a time of 25 ns can be achieved if the charge is collected by the
drift from the depleted region of the detector. In addition, the radiation
will significantly deteriorate the charge collected by diffusion from the
undepleted region because of charge trapping and carrier recombination
already after the exposure to hadron fluences equivalent to 1013 n/cm2
of 1-MeV neutrons [7], which will be reached early in the lifetime of
the ATLAS detector [2]. Therefore, charge collection from the depletion
region is necessary to achieve the speed and radiation hardness needed
for applications in high-luminosity hadron colliders.
The research on this detector technology for application at the LHC
was initiated by developments in the HV-CMOS process [8] and has
recently become very intensive. Promising results showing sufficient
radiation hardness of the depleted CMOS detectors from various designs
and producers were published in a large number of publications.
Refs. [9–18] represent just some of the more recent ones. CMOS
technology is being investigated as an option for the outermost layer
of the pixel detector in the upgraded ATLAS experiment [6] where
The work was partly done in the framework of the RD50 collaboration.
Corresponding author.
E-mail address: (I. Mandić).
Received 22 January 2018; Received in revised form 14 June 2018; Accepted 22 June 2018
Available online 30 June 2018
0168-9002/© 2018 Elsevier B.V. All rights reserved.
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
Fig. 1. (a) Photo of the thinned LFoundry chip biased through the backplane prepared for measurement. Arrow points to the investigated structure B. (b) Schematic
cross-section of test structure B. (c) Layout of (part of) structure B with bonding pads.
The chips are produced on 700-μm thick wafer and the unthinned
samples with no backplane processing were diced from the wafer. A set
of samples was taken from a wafer thinned to 200 μm with a processed
and metallised backplane.
Before the measurements each chip was fixed to an aluminium
support using conductive glue (see Fig. 1). For the unthinned samples
the substrate was contacted via a wire bond to the p-type ring, while
for the thinned samples contact with the substrate was achieved via
the backplane through the conductive glue. The reverse bias voltage
between pixel implant and the substrate at which the current started to
increase rapidly (breakdown) was about 290 V.
The chips were irradiated with neutrons in the TRIGA reactor in
Ljubljana [23,24] to 1 MeV neutron equivalent fluences ranging from
1⋅1013 neq /cm2 to 2⋅1015 neq /cm2 . In the reactor the irradiation is
performed by inserting the samples into the core through irradiation
tubes. The maximum power of the TRIGA reactor in Ljubljana is 250 kW.
At this power the 1 MeV neutron equivalent flux in the irradiation tube is
1.5⋅1012 neq /cm2 /s; therefore, the fluence of 1⋅1014 neq /cm2 is reached in
65 s. Irradiation at lower fluences is performed at lower reactor powers
to increase the irradiation time and thus reduce the irradiation-time
uncertainties due to insertion and extraction. The neutron flux in the
irradiation channel is periodically controlled by measurements of the
leakage-current increase in dedicated silicon diodes irradiated in this
irradiation channel [25]. Before the measurements the samples were
annealed for 80 min at 60 ◦ C.
the expected displacement damage in the silicon will be equivalent to
exposure to 2⋅1015 1-MeV neutrons per cm2 [2].
One of the investigated versions of the depleted CMOS detectors is
produced by LFoundry [19–22] in a 150-nm process on p-type substrates
with initial resistivities exceeding 2 kΩcm and typically in the range
between 4 and 5 kΩcm [19]. Such a high-resistivity material ensures
a large depletion depth even at moderate bias voltages. An irradiation
study using the Edge-TCT with passive test structures manufactured by
LFoundry was published in [18]. It was shown that a depletion depth
of over 50 μm is achieved at a 120 V bias also after irradiation with
8⋅1015 neq /cm2 . In this paper we report on measurements with a new
set of samples produced using the same process. The depletion depth
was measured with the Edge-TCT and the collected charge deposited in
the detector with the passage of a fast electron from the 90 Sr source was
measured and compared to the value estimated from the Edge-TCT. Two
sets of devices were investigated: devices from an unthinned wafer with
a substrate biased through the implant on the top and devices from a
thinned wafer with processed and metallised backplane for the substrate
2. Samples and irradiation
The photograph in Fig. 1a shows the LFoundry chip with several
test structures. Measurements were made with the test structure B,
schematically shown in the drawings in Fig. 1b and c. Structure B is
an array of 15×6 passive n+ p pixels with a pixel size of 50×520 μm2 .
Structure B in the photograph in Fig. 1a is connected with two bond
wires, while the remaining 9 of 11 bonding pads are not connected.
The first bonding pad is connected to all pixels, except one, connected
together, and the second bonding pad is for this one single pixel (see
Fig. 1c). The remaining unbonded pads are for the p-type implants
between the n-wells in the pixel area, the n-ring surrounding the pixels
and several guard rings. The outermost p-type implant ring also has
a separate bonding pad and was used to bias the substrate in devices
biased from the top.
3. Experimental techniques
The Edge-TCT is a variant of the Transient Current measurement
Technique in which sub-nanosecond pulses of infra-red ( = 1064 nm)
laser light are directed to the edge (i.e., the laser beam runs parallel
with the surface, see arrow in Fig. 1a of the investigated device. A
narrow laser beam with a diameter of less than 10 μm is used so that
the charge carriers are released at a known depth in the investigated
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
Fig. 2. Scheme of the experimental setup for charge-collection measurements
with a 90 Sr source.
device positioned in the beam with high-precision moving stages. The
Edge-TCT method was first described in [26] and has become a standard
tool for investigating the radiation effects in silicon detectors. The
measurements as part of this work were carried out with an EdgeTCT system produced by Particulars [27]. The Edge-TCT data-collection
and analysis techniques are very similar to those described in [18] and
details can be found there.
Charge-collection measurements with charged particles were made
with the experimental setup shown in Fig. 2. The DUT was placed
between two aluminium collimators with 1-mm diameter holes with
a 90 Sr source on one side and a small scintillator coupled to a photomultiplier on the other. Only electrons from the high-energy end of
the 90 Sr spectrum have sufficient energy to pass the silicon chip and
trigger the readout by depositing sufficient energy in the scintillator.
The collimators minimise the contribution of the electrons scattered
by a large angle in the setup. This makes it possible to measure the
charge released in the detector by the passage of a particle that is a
close approximation to a Minimum Ionising Particle (MIP). But it should
be noted that because of the lower energy and the larger scattering of
electrons from the 90 Sr compared to the MIPs, the measured collected
charge will be of the order of 10% larger than in the case of the MIPs.
When triggered by the photomultiplier, a digital oscilloscope records
the waveform from the custom-made shaping circuit with a 25 ns
peak time, processing the signal from the Ortec 142 charge-sensitive
preamplifier. The active area of the structure B used in this work (see
Fig. 1b is 0.75 mm × 3.1 mm. This is slightly too small compared to
the collimator’s hole diameter to ensure a sample with a large fraction
of waveforms with the charge deposited in the structure B. So there
are about 50% of the waveforms in the sample where the electrons
from the source triggered the readout but did not deposit charge in the
structure B. Therefore, the collected charge can be estimated if the peaks
of the measured distribution with and without the charge signal in the
structure can be clearly separated. This was the case if the Most Probable
Value (MPV) extracted from the fit of the convolution of the Landau and
Gaussian distributions to the signal spectrum was larger than about 4000
The system was calibrated by measuring the MPV of the collected
charge for a standard 300 μm-thick, fully depleted silicon detector and
confirmed with 59.5 keV photons from a 241 Am source. More details
about this measurement setup can be found in [13,28].
Fig. 3. Figure (a): charge-collection profiles for an unthinned device before
irradiation at bias voltages from 0 to 250 V in 10 V steps. The narrowest profile
was measured at 0 V. Figure (b): FWHM of charge-collection profiles vs. bias
work a single pixel (see Fig. 1) was connected to the wide bandwidth
amplifier, while other pixels were connected at the same potential, but
not to the amplifier. A high voltage was connected to the pixels and
decoupled from the readout by a Bias-T while the substrate contact was
at ground potential.
Fig. 3a shows charge as a function of the substrate depth at an
increasing bias voltage for an unthinned device before irradiation. The
charge is given in arbitrary units because the signal depends on the laser
power, the laser beam focusing, the edge surface quality, etc., and thus
cannot be compared between devices. It is clear how the width of the
charge-collection profile, which is a measure of the depletion depth,
increases with the bias voltage. In Fig. 3b the depleted depth, defined
here as the FWHM of the charge-collection profile, is plotted vs. the bias
voltage and fitted with:
2 0
(bias ) = 0 +

0 ef f bias
4. Measurements
where ef f is the effective space charge concentration, bias is the bias
voltage, 0 is the elementary charge, 0 is the dielectric constant and
 is the relative permittivity of silicon. In the approximation of an
abrupt junction and uniform doping the parameter 0 would be zero
at zero bias voltage (neglecting the built-in voltage). However, it is
known [12,13,18,29] that in Edge-TCT measurements at shallow depletion depths, a significant offset is observed due to a finite laser-beam
4.1. Edge-TCT
In the Edge-TCT the current pulses induced on the readout electrodes
by the movement of the charge carriers released in the detector by a
short laser pulse are recorded and analysed. The time integral – the
integration time was 25 ns in this study – of the current pulse is defined
as the collected charge. In the Edge-TCT measurements shown in this
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
Fig. 5. Effective doping concentration as a function of 1 MeV equivalent neutron
unthinned devices, as a function of the neutron fluence, and it can be
seen that it increases linearly with the fluence, having the slope  ∼ 0.03
cm−1 . This is somewhat larger than the stable damage-introduction rate
usually found in p-type silicon ( ∼ 0.02 cm−1 [30]), but it is consistent
with a similar measurement using LFoundry samples from [18]. No
decrease of ef f with fluence – a strong indication of initial acceptor
removal – was observed in this work. In the measurements in [18] a
decrease of ef f was measured at 1⋅1013 neq /cm2 and 5⋅1013 neq /cm2 ,
and the acceptor removal parameters could be estimated. But it should
be noted that the initial resistivity of the samples measured in [18] was
∼2 kΩcm, while for samples measured in this study it is over 3 kΩcm (see
Fig. 3). It can be expected, based on the observations in [13,31], that
the acceptor-removal effects would be observable at lower fluences than
those measured in this study because of the higher initial resistivity.
Fig. 4. Figure (a): charge-collection profiles at Vbias = 250 V and (b) width of
profiles vs. bias voltage at different fluences for thinned and unthinned devices.
For each fluence a different device was measured.
4.2. Charge-collection measurements
For the charge-collection measurements with the 90 Sr source all the
pixels of device B (see Fig. 1) were connected together, forming an
effective pad detector. The spectra measured before the irradiation at
250 V for the unthinned and thinned samples are shown in Fig. 6a
and b. A convolution of the Landau and Gaussian functions is fitted
to the spectra. The most probable value of the Landau function (MPV)
estimated from the fitted function is a measure of the collected charge.
It can be seen in Fig. 6a that a larger MPV was measured with the
unthinned devices. It corresponds to a charge of 20000 electrons, while
with the thin sample a MPV of 13700 electrons was measured. This
is slightly smaller than expected from the Edge-TCT measurements of
depletion depth, made with the same sample, shown in Fig. 3. An
electron from the90 Sr releases ∼85 electron–hole pairs per μm of silicon
with the largest probability.
Fig. 6c and d show the spectra after irradiation with 5⋅1013 neq /cm2
and it can be clearly seen that a significantly larger charge is measured
with the thinned detector, although the depletion depth obtained from
the Edge-TCT is significantly larger in the unthinned sample, as shown
in Fig. 4.
The depletion depth at 5⋅1013 neq /cm2 and the 260 V bias voltage
is 260 μm in the unthinned sample and 180 μm in the thinned sample,
which is fully depleted (see Fig. 4) at this bias voltage. Therefore, a
fast electron selected from the 90 Sr spectrum releases ∼ 22000 electrons
MPV in the depletion region of the unthinned device and ∼ 15000
electrons in the thinned sample. In Fig. 6c and d it is clear that the
MPV of collected charge is 9000 electrons in the unthinned device and
13000 electrons in the thinned sample. So, while in the thinned device
almost all the charge is collected, in the unthinned device the value is
less than 50%.
The measurements were also made at other fluences and bias
voltages, as shown in Fig. 7a. At the highest fluence a bias voltage of
width, the charge collected by diffusion and by laser-beam reflections
from the metallised surfaces.
We observe that the Eq. (1) function fits the data in Fig. 3 well if
ef f and 0 are free parameters. ef f returned by the fit corresponds to
an initial resistivity of 3 kΩcm, in agreement with the manufacturer’s
Fig. 4a shows the charge-collection profiles of the thinned and
unthinned devices after different neutron fluences at a 250 V bias. For
each irradiation fluence a different device was measured and the charge
profiles for different devices were rescaled to have equal maximum
charge. It is clear that for unthinned devices the charge profile becomes
narrower with increasing fluence. For the thinned devices the profiles
up to the fluence of 1⋅1014 neq /cm2 have similar widths of about 180
μm because they are fully depleted at this bias voltage. It is also
important to note that at higher fluences the profile widths for the
thinned and unthinned devices are similar (except at the fluence of
1⋅1015 neq /cm2 – the reason for this deviation is not understood).
In Fig. 4b the charge-profile widths are plotted versus the bias
voltage. It is easily seen that in the thinned devices the depletion depth
grows with the bias voltage in accordance with equation Eq. (1) until full
depletion is reached. Before full depletion the dependence on the bias
voltage is very similar for the thinned and unthinned devices (except for
the already-mentioned case of eq = 1⋅1015 neq /cm2 ). It is clear that the
width at the full depletion depths is 190 μm for two fluences and 170
μm for the other two. For each fluence a different device was measured,
so this might be a consequence of the variation in the thickness of the
active depth after the thinning and the backplane processing.
ef f can be estimated from the fit to Eq. (1) also after the irradiation.
Fig. 5 shows ef f , the average of the values from the thinned and
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
Fig. 6. Spectrum of signals caused by electrons from the 90 Sr source for: unthinned (a) and thinned (b) device before irradiation, (c) unthinned and (d) thinned
device after irradiation to 5⋅1013 neq /cm2 . The charge is in kilo-electrons (kel.).
over 350 V had to be applied to collect more than 4000 electrons. The
devices could be biased to such a high voltage because the breakdown
performance improves with the irradiation.
In the thinned devices with a backplane contact the collected charge
after irradiation is about 15% smaller than expected, but it follows
roughly the change of depletion depth measured with the Edge-TCT.
This is not the case for the unthinned devices without a backplane,
where a large drop of collected charge is already observed at the lowest
fluence and a much lower collected charge is measured than is released
in the depletion region by electrons from the 90 Sr. This is shown in
Fig. 7b where the collected charge is plotted as a function of fluence
at a 260 V bias. The measured values and the values calculated from
the depletion depth estimated with the Edge-TCT are shown in Fig. 4.
The charge values were calculated by multiplying the depletion depth
by 84 electrons per μm (i.e., a 10% larger charge for 90 Sr electrons than
77 electrons/μm released by a MIP) and a factor roughly taking into
account the charge trapping. This factor reduces the charge by less than
15% at the highest fluence. The trapping factor was estimated with a
simplified simulation using measured trapping probabilities from [30].
The simulation was made with the KDetSim simulation tool – a ROOTbased library specialised in the simulation of charge drift in a static
electric field in silicon detectors [32]. More details about the simulation
method can be found in [33].
The large difference between the measured and the TCT values for
the unthinned devices follows from the different contacting schemes of
the unthinned and thinned devices. As explained in Section 2 in the
unthinned devices the substrate is biased via the p-type implant ring
on top of the device, while in the thinned devices the backplane was
processed and metallised, enabling contact with the substrate.
In the case of the top bias the drift paths of the holes, from the
depletion region below the positively biased implant to the substrate
contact, are passing through the low-field regions of the detector. This
does not significantly affect the charge collection before irradiation
because the trapping is negligible and the ohmic conductivity of the
undepleted substrate is high enough to bring the zero-weighting potential close to the border of the depletion region. In this case the charge
carriers drifting in the electric field of the depletion region cross a large
part of the weighting field – the condition necessary for a high charge
collection according to Ramo’s theorem. This explanation is supported
by measurements before the irradiation in Fig. 6a and b where the
measured collected charge agrees with the charge deposited in the
depletion depth measured with the Edge-TCT.
The situation becomes different after the irradiation because of the
trapping and the increased ohmic resistivity of the substrate [34]. In
the case of the unthinned devices the zero-weighting field potential is
at the top of the sensor and therefore distant from the region of the
high electric field. Carriers drifting towards the substrate contact are
trapped in the volume with a low electric field before they reach the
contact. For the thinned devices the zero-weighting potential electrode
is at the backplane. In the case of full depletion this is at the border of the
depletion region and in the case of partial depletion significantly closer
than in the unthinned and top biased device. This reasoning implies
that both the thinning and backplane processing improve the charge
collection in partially depleted detectors after irradiation. However, it
has to be repeated that the measurements were made with all the pixels
of the structure connected together forming an effective pad detector in
which the weighting field is significantly different than in the (small)
pixel geometry. It can be expected that the effect of thinning and back
biasing on the charge collection would be significantly smaller in the
pixel geometry where the movement of the carriers near the pixel
electrode contributes most to the collected charge.
4.3. TCAD simulations
A TCAD simulation of a simplified structure was made with the
aim to qualitatively support the above discussion about the differences
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
The reason for the discrepancy with the measurements was not studied
in detail, but it should be attributed to the simplifications used in the
simulation: measurements were performed with an array of pixels, while
in the simulation a single implant was considered. Also, the slow chargecollecting component in the thick structure indicates that the zeroweighting field potential is not near the border of the depletion zone,
which could be a consequence of the underestimated conductivity of the
undepleted bulk. The discrepancy between 13700 electrons, measured
with the thinned device before irradiation with the charge-sensitive
amplifier with a 25-ns shaping time, and 16000 electrons collected in
the simulation in Fig. 8c is partly a consequence of the thickness of the
real device (see Fig. 4b), being only ∼180 μm.
The significance of the thinning and the back processing for chargecollection efficiency can be much better seen after the irradiation. The
properties of the irradiated silicon were modelled in the simulation using
the so-called Perugia model [35,36]. Collected charge as a function
of integration time is shown in Fig. 8d for two fluences at a 250 V
bias. It can be seen that with the unthinned structure about a factor
of 2 smaller charge is collected compared to the thinned sample. The
value of the measured collected charge is smaller than the charge in
the simulation; however, the factor of 2 difference is in agreement with
the measurements in Fig. 7 at a fluence of 1⋅1014 neq /cm2 . At 1⋅1015
neq /cm2 the collected charge in the unthinned device was too small to
be measured with this measurement system. The cause of such a large
difference between the thinned and unthinned devices is the trapping
of charge carriers in the low-field regions of the unthinned sample.
5. Conclusions
Fig. 7. Collected charge vs. bias voltage for unthinned and thinned devices at
different fluences is shown in figure (a). Figure (b) shows charge vs. fluence at
260-V bias. Shown are the measured values and the values calculated from the
depletion depth obtained from the E-TCT.
Measurements with a set of passive pixel detectors made with
LFoundry 150 nm CMOS technology irradiated with neutrons were
presented. Two sets of devices were studied and compared: unthinned
devices without a processed backplane with substrate biased through
the implant on top, and thinned devices with a processed and metallised
backplane enabling contact with the substrate through the backplane.
The depletion depth was measured with the Edge-TCT, and it was
found that it increases with bias voltage following the square-root
dependence, as expected in the case of an abrupt junction and a
uniform doping concentration. From this dependence, the effective
doping concentration ef f was extracted. It was found that in the
range of fluences studied here, ef f increases linearly with the fluence
with a somewhat larger introduction rate than is usual for float-zone
silicon detector materials. This result is consistent with the measurement
in [18]. The measurements confirm that after irradiation with 2⋅1015
neq /cm2 depletion depths exceeding 50 μm can be easily achieved at
bias voltages that can be safely applied with this technology.
Charge-collection measurements with MIPs from a 90 Sr source revealed the major advantage of thinned devices with a processed and
metallised backplane after irradiation. A much larger collected charge
was measured with irradiated thinned devices, which delivers a strong
message that thinning (up to the level to remove the undepleted
bulk) and backplane processing would improve the charge collection of
irradiated CMOS devices by modifying the electric and weighting field in
the detector. However, it should be mentioned that these measurements
were made with a large number of pixels connected together. The
effect in individual pixel readout configuration used for charged particle
tracking is expected to be smaller because of the different weighting
Collected charge was measured up to a fluence of 2⋅1015 neq /cm2 .
The most probable value exceeding 5000 electrons was measured at
the highest fluence if a high enough bias voltage was applied. This
amount of collected charge is sufficient for a successful operation in
the experiments at the HL-LHC.
in the charge-collection efficiencies for the samples with and without
the processed backplane. A n-type implant was positioned in a p-type
substrate with an initial resistivity of 3 kΩcm and two cases were
considered: a 200 μm-thick device with a substrate contacted through
the backplane and a 700 μm-thick structure with the substrate contacted
on the area surrounding the implant on top.
Fig. 8a and b show the distribution of the absolute value of the
electric field in the structure biased with 250 V before the irradiation.
The electric field in the unthinned structure (Fig. 8a extends into the
substrate to a depth of more than 200 μm, as expected for a 3 kΩcm
resistivity, while the thinned device is fully depleted.
The passage of a MIP was simulated by injecting 77 electron–hole
pairs per μm along the straight line through the centre of the structure.
The current induced on the implant electrode by the movement of the
charge carriers in the electric field was calculated. The time integral of
the induced current is the collected charge and it is shown in Fig. 8c
and d as a function of the integration time. It can be seen in Fig. 8c
that before the irradiation in the thinned sample with a backplane
bias all the charge is collected rapidly, while in the unthinned case
a slow increase is seen after a steep rise. The slow component is
a consequence of the drift of carriers through the low-field regions
towards the contact on the top and also the consequence of the charge
entering the field region by diffusion from the undepleted substrate.
The simulation predicts that a larger charge would be collected with
the unthinned device, which is in agreement with measurements, as can
be seen in Fig. 6a and b. On the other hand, in the simulation an equal
charge in the thinned and unthinned devices is collected after ∼50 ns,
while 20000 electrons are collected in 100 ns in the unthinned device.
This is considerably slower than in the measurements, but it clearly
illustrates the importance of backplane biasing for charge collection.
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
Fig. 8. Absolute value of the electric field at a 250 V bias voltage in (a) unthinned structure biased from top and (b) thinned structure with processed backplane
before irradiation. Collected charge deposited along the centre of the structure (simulating a MIP) as a function of the integration time before the irradiation (c) and
after the irradiation to two different fluences (d), at a bias of 250 V.
[10] H. Pernegger, et al., First tests of a novel radiation hard CMOS sensor process for
Depleted Monolithic Active Pixel Sensors, JINST 12 (2017) P06008.
[11] E. Cavalaro, et al., Studies of irradiated AMS H35 CMOS detectors for the ATLAS
tracker upgrade, JINST 12 (2017) C01074.
[12] M. Fernandez, et al., Radiation hardness studies of neutron irradiated CMOS sensors
fabricated in the AMS H18 high voltage process, JINST 11 (2016) P02016.
[13] A. Affolder, et al., Charge collection studies in irradiated HV-CMOS particle detectors, JINST 11 (2016) P04007.
[14] V. Fadeyev, et al., Investigation of HV/HR-CMOS technology for the ATLAS Phase-II
Strip Tracker Upgrade, Nucl. Instrum. Methods A 831 (2016) 189.
[15] K. Kanisauskas, et al., Radiation hardness studies of AMS HV-CMOS 350 nm
prototype chip HVStripV1, JINST 12 (2017) P02010.
[16] B. Hiti, et al., Charge collection properties in an irradiated pixel sensor built in a
thick-film HV-SOI process, arXiv:1701.06324.
[17] T. Hemperek, T. Kishishita, H. Krüger, N. Wermes, A Monolithic Active Pixel Sensor
for ionizing radiation using a 180 nm HV-SOI process, Nucl. Instrum. Methods A 796
(2015) 8. arXiv:1412.3973.
[18] I. Mandić, et al., Neutron irradiation test of depleted CMOS pixel detector prototypes,
JINST 12 (2017) P02021.
[19] LFoundry, a SMIC owned Company,
[20] T. Kishishita, T. Hemperek, H. Krüger, N. Wermes, Depleted Monolithic Active Pixel
Sensors (DMAPS) implemented in LF-150 nm CMOS technology, JINST 10 (2015)
[21] P. Rymaszewski, et al., Prototype active silicon sensor in 150 nm HR-CMOS Technology for ATLAS Inner Detector Upgrade, JINST 11 (2016) C02045. arXiv:1601.00459.
[22] T. Hirono, et al., CMOS pixel sensors on high resistive substrate for high-rate, highradiation environments, Nucl. Instrum. Methods A 831 (2016) 94.
[23] L. Snoj, G. Žerovnik, A. Trkov, Computational analysis of irradiation facilities at the
JSI TRIGA reactor, Appl. Radiat. Isot. 70 (2012) 483.
[24] K. Ambrožič, G. Žerovnik, A. Trkov, Computational analysis of dose rates at the JSI
TRIGA reactor irradiation facilities, Appl. Radiat. Isot. 130 (2017) 140.
[25] ROSE Collaboration, G. Lindström, et al., Radiation hard silicon detectors developments by the RD48 (ROSE) collaboration, Nucl. Instrum. Methods A 466 (2001) 308.
[26] G. Kramberger, et al., Investigation of irradiated silicon detectors by Edge-TCT, IEEE
Trans. Nucl. Sci. 57 (2010) 2294.
[27] Particulars, Advanced measurement systems, Ltd.
[28] G. Kramberger, et al., Charge collection properties of heavily irradiated epitaxial
silicon detectors, Nucl. Instrum. Methods A 554 (2005) 212.
The authors would like to thank the staff at the TRIGA reactor in
Ljubljana for their help with the irradiation of the detectors. Part of
this work was performed in the framework of the CERN-RD50 collaboration. The authors acknowledge the financial support of the Slovenian
Research Agency (research core funding No. P1-0135 and project ID PR06802). The work was supported in parts by the Deutsche Forschungsgemeinschaft DFG, grant number WE 976/4-1, by the Ministry BMBF
grant number 05H15PDCA9. This project has received funding from the
European Union’s Horizon 2020 Research and Innovation Programme
under Grant Agreement no. 654168.
[1] F. Gianotti, et al., Physics potential and experimental challenges of the LHC
luminosity upgrade, Eur. Phys. J. C 39 (2005) 293 [hep-ph/0204087].
[2] ATLAS Collaboration, I. Dawson, Radiation background studies for the phase II inner
tracker upgrade, ATL-UPGRADE-PUB-2014-003, CERN, Geneva Switzerland, 2014.
[3] ATLAS Collaboration, The ATLAS experiment at the cern large hadron collider,
JINST 3 (2008) S08003.
[4] ATLAS Collaboration, ATLAS pixel detector electronics and sensors, JINST 3 (2008)
[5] H.C. Kastli, et al., CMS barrel pixel detector overview, Nucl. Instrum. Methods A 582
(2007) 724.
[6] ATLAS Collaboration, Letter of intent for the phase-II upgrade of the ATLAS
experiment, CERN-LHCC-2012-022, CERN, Geneva Switzerland, 2012, [LHCC-I023].
[7] H. Hillemanns, et al., Radiation hardness and detector performance of new 180 nm
CMOS MAPS prototype test structures developed for the upgrade of the ALICE
Inner Tracking System, Nuclear Science Symposium 2013, Seoul, September 2013.
[8] I. Peric, A novel monolithic pixelated particle detector implemented in high-voltage
CMOS technology, Nucl. Instrum. Methods A 528 (2007) 876.
[9] D.-L. Pohl, Radiation hard pixel sensors using high-resistive wafers in a 150 nm
CMOS processing line, JINST 12 (2017) P06020.
I. Mandić et al.
Nuclear Inst. and Methods in Physics Research, A 903 (2018) 126–133
[33] B. Hiti, et al., Charge collection properties in an irradiated pixel sensor built in a
thick-film HV-SOI process, JINST 12 (2017) P10020.
[34] L. Beattie, et al., Charge collection efficiency in heavily irradiated silicon diodes,
Nucl. Instrum. Methods A 412 (1998) 238–246.
[35] D. Passeri, et al., Comprehensive modeling of bulk-damage effects in silicon radiation
detectors, IEEE Trans. Nucl. Sci. 48 (5) (2001) 1688–1693.
[36] M. Petasecca, et al., Numerical simulation of radiation damage effects in p-type and
n-type FZ silicon detectors, IEEE Trans. Nucl. Sci. 53 (5) (2006) 2971–2976.
[29] M. Fernandez, et al., On the determination of the substrate effective doping
concentration of irradiated HV-CMOS sensors using an edge-TCT technique based
on the Two-Photon-Absorption process, JINST 12 (2017) C01038.
[30] V. Cindro, et al., Radiation damage in p-type silicon irradiated with neutrons and
protons, Nucl. Instrum. Methods A 599 (2009) 60–65.
[31] G. Kramberger, et al., Radiation effects in Low Gain Avalanche Detectors after
hadron irradiations, JINST 10 (2015) P07006.
[32] KDetSim webpage,
Без категории
Размер файла
2 284 Кб
niman, 062, 2018
Пожаловаться на содержимое документа