Contact us
My IOPscience
3D imaging of radiation damage in silicon sensor and spatial mapping of charge collection
This content has been downloaded from IOPscience. Please scroll down to see the full text.
2013 JINST 8 C03023
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address:
This content was downloaded on 31/03/2015 at 04:10
Please note that terms and conditions apply.
R ECEIVED: October 3, 2012
R EVISED: January 23, 2013
ACCEPTED: February 8, 2013
P UBLISHED: March 27, 2013
1–5 J ULY 2012,
M. Jakubek,a,1 J. Jakubek,a J. Zemlicka,a M. Platkevic,a V. Havranekb and V. Semianb
a Institute
of Experimental and Applied Physics, Czech Technical University in Prague,
Horska 3a/22, CZ 12800 Prague 2, Czech Republic
b Nuclear Physics Institute of the Ac. of Science of the Czech Republic,
Husinec — Rez 130, 250 68 Rez, Czech Republic
E-mail: [email protected]
A BSTRACT: Radiation damage in semiconductor sensors alters the response and degrades the performance of many devices ultimately limiting their stability and lifetime. In semiconductor radiation detectors the homogeneity of charge collection becomes distorted while decreasing the overall
detection efficiency. Moreover the damage can significantly increase the detector noise and degrade
other electrical properties such as leakage current. In this work we present a novel method for 3D
mapping of the semiconductor radiation sensor volume allowing displaying the three dimensional
distribution of detector properties such as charge collection efficiency and charge diffusion rate.
This technique can visualize the spatially localized changes of local detector performance after radiation damage. Sensors used were 300 µm and 1000 µm thick silicon bump-bonded to a Timepix
readout chip which serves as an imaging multichannel microprobe (256 × 256 square pixels with
pitch of 55 µm, i.e. all together 65 thousand channels). Per pixel energy sensitivity of the Timepix
chip allows to evaluate the local charge collection efficiency and also the charge diffusion rate. In
this work we implement an X-ray line scanning technique for systematic evaluation of changes in
the performance of a silicon sensor intentionally damaged by energetic protons.
K EYWORDS : Charge transport and multiplication in solid media; Radiation damage evaluation
methods; Pixelated detectors and associated VLSI electronics; Radiation damage to detector materials (solid state)
1 Corresponding
c 2013 IOP Publishing Ltd and Sissa Medialab srl
2013 JINST 8 C03023
3D imaging of radiation damage in silicon sensor and
spatial mapping of charge collection efficiency
Scanning principle
3D mapping of charge diffusion of a 1000 µm thick silicon detector
3.1 Measurement in “counting” mode
3.2 Measurement in energy mode
3D mapping of a 300 µm thick silicon sensor intentionally damaged by protons
4.1 Scan of the region A (damaged by 2.5 MeV protons) in counting mode
4.2 Scan of the region B (damaged by 4 MeV protons) in counting mode
4.3 Scan of the region B (damaged by 4 MeV protons) in energy mode
The semiconductor pixel detectors of the Medipix type [1, 2] are used, thanks to their superior
imaging properties, in many applications ranging from X-ray and neutron radiography [3] to hadron
therapy [4] and dosimetry [5]. The growing number of applications of this technology brings the
necessity to test and evaluate their detection performance such as local charge collection efficiency
and radiation hardness and their change during their whole life cycle. Especially the problem of
radiation damage plays an important role in applications with difficult access such as the ATLAS
experiment at the LHC [6] or applications on board satellites in space [7].
Existing methods evaluate changes only of global properties of single-pad silicon sensors after
irradiation e.g., [8–10], and lack any imaging or spatial information. In this paper we present a
novel approach enabling three dimensional response mapping of silicon sensors which we tested
with intentional radiation damage by an energetic proton beam. The method allows for the local
characterization of the sensor response using a highly collimated X-ray beam and the Timepix
readout chip. This capability would be useful also for numerical simulation and model studies
e.g., [11–13].
The ionization charge created by an X-ray beam is sensed by the Timepix read-out chip which
acts as a multichannel imaging microprobe. The Timepix chip was developed by the Medipix
collaboration [14] and consists of a semiconductor radiation sensor bump bonded to a matrix of
256 × 256 pixels with 55 µm pitch size. Each pixel contains independent signal electronics (amplifier, discriminator, counter). Each pixel can be operated in one of three modes: Medipix mode
(counter counts incoming particles), Timepix mode (counter works as a timer and measures the
2013 JINST 8 C03023
Figure 1. (a) Schematic principle of 3D scanning: the detector is shifted perpendicularly to the beam direction maintaining a constant irradiation angle. At each position the depth of interacting X-rays in each
particular pixel along the beam is therefore known with high precision. (b) View of the experimental setup
consisting of a microfocus X-ray tube with filter, shielding plate, controllable slit stand and detector positioning system.
time of the particle detection) and Time over threshold (TOT) mode (counter is used as a Wilkinson type ADC allowing direct energy measurement in each pixel) [2].
The main purpose of this article is to introduce a new instrumental method which can be used
as a tool for sensor characterization with imaging capability and high spatial resolution. Therefore,
we are not giving interpretation of measured data for particular sensors.
Scanning principle
The principle of the system is based on the use of an X-ray beam collimated by a narrow slit [15].
This beam is sent onto a pixel detector at a sharp angle as depicted in figure 1. The tilted narrow
beam penetrating the thick sensor creates a signal along several pixels. Pointing such beam to a
particular sensor position we can determine the depth of interaction for each pixel along the beam
path. Shifting the detector along the axis perpendicular to the plane of the beam (hitting other
detector columns) we can deliver radiation to all depths of each pixel. Data recorded during such
scan have to be normalized for the exponentially decreasing absorption of X-rays in the sensor.
This method allows obtaining a 3D map of the detector response (see figure 1).
The precise positioning of the slit and detector is essential. The system has seven degrees of
freedom: detector position, rotation and tilt, slit position and width and global shift and rotation
of the whole platform (with respect to the X-ray tube). The detector position and slit width is
controlled by motorized stages with precision of 1 µm. The X-ray 3D scanning system and method
was described in more detail in our previous paper [16].
3D mapping of charge diffusion of a 1000 µm thick silicon detector
We demonstrate the usage of the scanning technique for the measurement of charge diffusion in
a new silicon sensor without any radiation damage. The extent of charge diffusion depends on
2013 JINST 8 C03023
local properties of the semiconductor material, the applied bias voltage and the depth of charge
collection. The depth affects the time of charge collection and consequently also the extent of
charge diffusion (a charge collected at a larger distance is spread over more pixels).
Two measurement techniques for charge diffusion mapping are described in the following
counting and energy based.
Measurement in “counting” mode
The Timepix chip is operated in counting mode which means that the pixel counter is incremented
whenever the ionization charge collected to this pixel is greater than a certain global threshold.
This fact does not mean that the pixel counting “one” was necessarily hit by particle. The collected
charge can come from adjacent pixel(s) due to charge diffusion. The extent of this effect changes
with depth of interaction, therefore, the 3D scan described in section 2 shows a different count rate
at different depths of the sensor. Such measurement allows mapping of charge diffusion within the
sensor volume.
The threshold settings play a very important role for this type of scan. For example if the
threshold is set very close to the energy of the used X-ray photons then photons are counted only
when the complete charge is collected by a single pixel. If the ionization charge is shared due to
diffusion by more pixels then the signal in each of them is below the threshold and no pixel counts.
An example of a 3D scan showing the influence of charge sharing in a 1 mm thick silicon
sensor (P-on-N type) is shown in figure 2(b) and (c). The detector bias voltage was 140 V which
was not enough for complete depletion. The scan was performed with a quasi monochromatic Xray beam with mean energy of 40 keV. The slit size was 10 µm. Two scans were performed with
two threshold levels. For a threshold of 33 keV the pixel has to collect more than 3/4 of the photon
energy to register an event. That is why we see many more events deeper in the sensor (closer to
the pixelated electrode) in figure 2(c). Conversely, in case of a threshold of 17 keV we see a much
smaller vertical gradient.
2013 JINST 8 C03023
Figure 2. (a) Schematic picture of charge sharing in the detector. (b) Charge sharing effect across the 1 mm
thick Si sensor for 17 keV threshold. (c) Same scan for 33 keV threshold. The vertical intensity profiles are
shown for each threshold in the insets on the right.
Measurement in energy mode
Pixels operated in Time-over-Threshold (ToT) mode measure the amount of charge they collect.
For a reliable measurement it is necessary to avoid pile-up of several events in a single frame which
can be easily accomplished by reduction of the exposure time. Then each measured frame contains
only isolated events corresponding to individual photons. The charge generated by a photon can
be registered by one or more adjacent pixels forming a cluster. The number of pixels in a single
cluster (cluster size) corresponds to the level of charge diffusion and the sum of charges in all pixels
(cluster volume) corresponds to the total collected charge. This principle combined with the 3D
scanning method allows performing direct measurement of local charge collection efficiency and
of the extent of charge diffusion.
The described technique was used for the same detector (1 mm thick Si) taking 2000 frames
in each slit position. The exposure time of each frame was 20 ms. The scan was performed for two
values of bias voltage 140 V and 450 V. The threshold level was set to 4 keV. Results are shown in
figure 3.
The result of this measurement shown in figure 3 on the right hand side proves that charge
collection efficiency is almost constant in the whole depleted volume of the sensor. The slight
vertical gradient is caused by the same effect discussed in figure 2 but its extent is lower due to low
threshold. It also shows that the sensor was not fully depleted for a bias voltage of 140 V. In the
same figure on the left hand side we see that the average cluster size grows with distance from the
pixelated (charge collecting) electrode. It should be noted that even in the very close proximity to
the pixelated electrode the average cluster size is larger than 1. There are several reasons for this
behavior such as the absence of lateral electric field in the volume between the pixel electrodes and
the nonzero size of the initial charge cloud created by the X-ray photon.
2013 JINST 8 C03023
Figure 3. 3D scan of a 1 mm thick Si sensor in energy mode. Left: average cluster size (charge diffusion)
across the sensor thickness (zero corresponds to the pixelated electrode) for bias voltage of 140 V (blue) and
450 V (yellow). The inset shows the vertical map of the average cluster size across the sensor for 140 V.
Right: average cluster volume (charge collection) across the sensor thickness. The inset shows the vertical
map of the average cluster volume across the sensor for 140 V.
3D mapping of a 300 µm thick silicon sensor intentionally damaged by protons
The method described in the previous section was used for 3D mapping of the charge collection
efficiency in a silicon sensor intentionally damaged by energetic protons. A 300 µm thick silicon
sensor of P-on-N type was damaged in several regions by different and well defined doses of protons with energies of 2.5 MeV and 4 MeV (see figure 4). The detector was scanned immediately
after irradiation and then several times later to see the effects of annealing. All scans were performed with three bias voltages 100 V, 14 V and 7 V. The depleted region just touches the surface
of the sensor at 14 V. The other two voltages present situations with highly over-depleted and
under-depleted sensor.
The scans were as follows:
Scan 0: before irradiation
Scan 1: 12 hours after irradiation
Scan 2: 1 month after irradiation (detector kept at room temperature)
Scan 3: after the second scan and annealing at 70◦ C for 3 hours
Scan 4: after the third scan and annealing at 95◦ C for 5 hours
Most of the 3D scans were performed in counting mode because they are more than 10 times
faster than those in energy mode. The energy threshold was set to 5 keV which is well above the
noise level. In contrast to the technique described in section 3.1 we were more interested in the
charge collection efficiency in irradiated regions than in charge diffusion. Therefore we generated
vertical intensity profiles (see vertical profiles in figure 2) from the scan performed before irradiation for each bias voltage. These profiles incorporate the charge sharing effect for a healthy sensor.
Then we used these profiles for vertical normalization of the scans performed after irradiation. This
way we visualize the changes in the detector response caused by radiation damage only.
2013 JINST 8 C03023
Figure 4. Damaged regions of the sensor with indicated doses expressed in protons/cm2 .
Scan of the region A (damaged by 2.5 MeV protons) in counting mode
The range of 2.5 MeV protons in silicon is 70 µm which is nicely displayed in the results of the 3D
scans. The results for bias voltages of 100 V and 14 V are shown in figure 5 and figure 6. For 7 V
the depleted volume did not reach the damage volume. In both figures there are four sections for
four scans (12 hours after, 1 month after, additional annealing at 70◦ C and 95◦ C). In each section
there are three profiles along three planes in the 3D data: horizontal plane parallel to sensor surface
at depth of 20 µm and two vertical planes perpendicular to the sensor surface going through the
damaged spots.
Comparing figure 5 and figure 6 it is obvious that higher bias voltage improves charge
collection efficiency. The effect of annealing is nicely visible in the area damaged by 5.13 ·
1010 protons/cm2 for bias of 100 V and it is even more pronounced in the area damaged by
6.36 · 109 protons/cm2 for 14 V bias.
Scan of the region B (damaged by 4 MeV protons) in counting mode
The range of 4 MeV protons in silicon is 150 µm. The summarized results of all 3D scans are
shown in figure 7. For brevity we selected only results acquired for all bias voltages after the last
annealing step.
2013 JINST 8 C03023
Figure 5. 3D scans of region A damaged by 2.5 MeV protons. Scans are performed in counting mode
with bias voltage 100 V. The top row shows the charge collection efficiency from the horizontal plane going
through the sensor at a depth of 20 µm (from the common electrode). The bottom row shows pairs of vertical
cross sections going through damaged spots along the lines indicated. Four scans are displayed: a) 12 hours
after irradiation. b) 1 month after irradiation, c) additional 3 hours of annealing at 70◦ C, d) additional
5 hours of annealing at 95◦ C. The uniform white color shows areas with high noise. We see that damage
by 2.27 · 1010 protons/cm2 is practically invisible and damage by 5.13 · 1010 protons/cm2 is almost repaired
after annealing.
Figure 7. 3D scans of region B damaged by 4 MeV protons. Corresponding layers at 3 different depths are
shown (top row). The response after the last annealing step is displayed for all three bias voltages. The two
areas damaged by fluence of 1010 protons/cm2 and lower disappeared at 100 V bias. The area damaged by
5 · 1010 protons/cm2 remarkably shrunk at 100 V bias. Note the changing depth of the depletion volume in
the vertical profiles (bottom row).
2013 JINST 8 C03023
Figure 6. Same as figure 5 for bias 14 V. The dark streaks in the horizontal profiles (top row) are regions
where the detector is not fully depleted due to inhomogeneous concentration of dopants in silicon. All
damaged areas are visible. The area damaged by 6.36 · 109 protons/cm2 is almost fully repaired after annealing. During the first scan shown in a) we could not deplete the detector fully (high leakage current). The
corresponding horizontal profile is therefore shown at depth of 60 µm.
Scan of the region B (damaged by 4 MeV protons) in energy mode
The 3D scan in energy mode was performed for 14 V bias only. It showed an unchanged charge
collection efficiency in the non-irradiated areas. The distribution of average cluster size is more
interesting (see figure 8) showing the extent of charge diffusion across the sensor (which in fact
reflects the local intensity of the electric field inside of the sensor). In figure 8 we see a magnified
region near the spot which was heavily damaged by 2.5 · 1011 protons/cm2 with energy of 4 MeV.
Especially in the volume below the damage it is evident that the clusters are smaller which means
that charge collection is faster implying a stronger electric field.
We developed a new instrumental method for 3D response scanning of semiconductor detectors
with X-rays. The method can create 3D maps of charge collection efficiency and charge diffusion
rate in the sensor volume. The method was successfully used for visualisation of changes in the
silicon sensor after damaging by energetic protons and during subsequent annealing process. The
measurements showed two sensor properties changing with irradiation level in damaged regions.
The first is the shorter life time of charge carriers in damaged zones which is demonstrated in
figure 7 (charge can be collected from the damaged zones but part of it recombines if the collection
is slow for low bias voltage). The second changed property is the decrease of material resistivity
in the damaged zone which is seen in figure 8 as the increased intensity of electric field below
damage. In the near future we plan to collect more complete data and prepare a semi-empirical
model of irradiated silicon sensors behaviour.
2013 JINST 8 C03023
Figure 8. Detail of distribution of average cluster size in the vertical plane across the detector thickness.
It shows an area near the spot damaged by 2.5 · 1011 protons/cm2 . The measurement was done with bias
voltage of 14 V. The average cluster size increases with distance from the pixelated electrode (bottom) and
it is systematically lower in the space below the damage which indicates a stronger electric field.
This research was carried out in framework of the Medipix Collaboration and has been supported
by project MSM 6840770040 of the Ministry of Education, Youth and Sports, Grant TA01010237
of the Technology Agency and Project 103/09/2101 of the Grant Agency of the Czech Republic.
[2] X. Llopart, R. Ballabriga, M. Campbell, L. Tlustos and W. Wong, Timepix, a 65k programmable pixel
readout chip for arrival time, energy and/or photon counting measurements,
Nucl. Instrum. Meth. A 581 (2007) 485.
[3] J. Jakubek, Semiconductor Pixel detectors and their applications in life sciences,
2009 JINST 4 P03013.
[4] J. Jakubek et al., Imaging with secondary radiation in hadron therapy beams with the 3D sensitive
voxel detector, IEEE NSS/MIC (2011) 2281.
[5] M. Othman et al., Neutron dosimeter development based on Medipix2,
IEEE Trans. Nucl. Sci. 57 (2010) 3456.
[6] Z. Vykydal et al., The Medipix2-based network for measurement of spectral characteristics and
composition of radiation in ATLAS detector, Nucl. Instrum. Meth. A 607 (2009) 35.
[7] D. Turecek et al., Small dosimeter based on Timepix device for International Space Station,
2011 JINST 6 C12037.
[8] K. Gill et al., Radiation damage by neutrons and protons to silicon detectors,
Nucl. Instrum. Meth. A 322 (1992) 177.
[9] RD48/ROSE collaboration, G. Casse et al., Scanning of irradiated silicon detectors using alpha
particles and low-energy protons, Nucl. Instrum. Meth. A 434 (1999) 118.
[10] J. Zhang et al., Study of X-ray radiation damage in silicon sensors, 2011 JINST 6 C11013.
[11] D. Pennicard et al., Simulations of radiation-damaged 3D detectors for the Super-LHC,
Nucl. Instrum. Meth. A 592 (2008) 16.
[12] G. Lindstr¨om, Radiation damage in silicon detectors, Nucl. Instrum. Meth. A 512 (2003) 30.
[13] D. Passeri, P. Ciampolini, G.M. Bilei and F. Moscatelli, Comprehensive modeling of bulk-damage
effects in silicon radiation detectors, IEEE Trans. Nucl. Sci. 48 (2001) 1688.
[14] M EDIPIX collaboration homepage, http://medipix.web.cern.ch/medipix/.
[15] J. Jakubek et al., Spectrometric properties of TimePix pixel detector for X-ray color and phase
sensitive radiography, IEEE Nucl. Sci. Symp. Conf. Rec. 3 (2007) 2323.
[16] M. Jakubek, J. Jakubek, J. Zemlicka, M. Kroupa and F. Krejci, Probe and scanning system for 3D
response mapping of pixelated semiconductor detector with X-rays and the timepix device,
AIP Conf. Proc. 1423 (2011) 461.
2013 JINST 8 C03023
[1] X. Llopart, M. Campbell, R. Dinapoli, D. San Segundo and E. Pernigotti, Medipix2, a 64-k pixel
readout with 55-µm square elements working in single photon counting mode,
IEEE Trans. Nucl. Sci. 49 (2002) 2279.