Performance Study of a Wide-Area SiPM Array, ASICs Controlled

advertisement
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON NUCLEAR SCIENCE
1
Performance Study of a Wide-Area
SiPM Array, ASICs Controlled
A. J. González, S. Majewski, J. Barberá, P. Conde, C. Correcher, L. Hernández, C. Morera,
L. F. Vidal, F. Sánchez, A. Stolin, and J. M. Benlloch
Abstract—In this paper, the capabilities of a wide-area gamma
ray photosensor based on a SiPM array are investigated. For this
purpose, we have mounted an array of 144
SiPMs with
mm and a pitch of
mm, thus
individual active area of
covering an active area of
mm . The measurements
were performed by coupling the SiPM array to LYSO crystal armm ,
mm , and
rays of different pixel size (
mm ) and 10–12 mm thicknesses. The SiPM array was controlled by means of three ASICs, and the SiPM signals were multiplexed in order to determine the gamma ray impact position by
means of implementing the Anger logic algorithm in the ASIC. The
optimum bias voltage and temperature dependence of the gamma
ray sensor were determined. An energy resolution as good as 8%,
for individual crystal pixels, were reached at 5 V overvoltage. The
ASICs design allows one to “activate” different photosensor array
areas. This feature has been used to evaluate the detector performance as a function of the crystal pixel size and the photosensor
dark noise contribution. In this work we also show the system capability to provide depth-of-interaction (DOI) information by means
of implementing a two-layer staggered approach. We have found
that accurate DOI information is obtained when the ASICs enmm (
SiPMs).
abled an SiPM active area as high as
Index Terms—Application-specific integrated circuits (ASICs),
Gamma-ray detectors, photodetectors, positron emission tomography (PET) instrumentation, scintillators.
I. INTRODUCTION
P
HOTOMULTIPLIER tube (PMT) technology has been
widely explored for a variety of applications, and nuclear
medicine has made use of them for about 50 years. Recently, the
routine use of magnetic resonance imaging (MRI) in medical
practice and especially when combined with other functional
imaging techniques such as positron emission tomography
Manuscript received February 21, 2014; revised July 10, 2014; accepted
September 16, 2014. Project funded by the Spanish Ministry of Economy and
Competitiveness and co-funded with FEDER’s funds within the INNPACTO
2011 program. This work was supported by the Spanish Plan Nacional de
Investigación Científica, Desarrollo e Innovación Tecnológica (I+D+I) under
Grant FIS2010-21216-CO2-01 and the Valencian Local Government under
Grants PROMETEOII/2013/010 and ISIC 2011/013.
A. J. González, P. Conde, L. Hernández, L. F. Vidal, F. Sánchez, and J. M.
Benlloch are with Institute for Instrumentation in Molecular Imaging (I3M),
CSIC -Universidad Politécnica de Valencia—CIEMAT, 46022 Valencia, Spain
(e-mail: agonzalez@i3m.upv.es).
S. Majewski and A. Stolin are with the Department of Radiology, West Virginia University, Morgantown, WV 26506 USA.
J. Barberá, C. Correcher, and C. Morera are with Oncovision S. A., 46013
Valencia, Spain.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TNS.2014.2359742
(PET) has boosted the research efforts of several groups to
demonstrate the capability of photosensors to work in strong
magnetic fields without impact on performance.
There are mainly two types of photosensors capable of this
compatibility, the so-called avalanche photodiodes (APDs) and
silicon photomultipliers (SiPMs). Both are, in general, large arrays of micro-APDs, but the first type works in the avalanche
regime, whereas the SiPMs operate in the Geiger mode, over the
self-quenched breakdown voltage. Each micro-APD is referred
to as a cell and produces a signal when it detects one photon. The
SiPM output is provided as the sum of all output cells. SiPMs
have better timing response compared to APDs [1] as well as
. However, both devices present the limiting
high gain
factor of dark noise.
Due to the incompatibility of PMTs to work in magnetic fields
such as those presented in MRI systems, SiPMs (also APDs)
have been suggested to replace PMTs in the design of PET systems [2]–[4]. There are several efforts focusing on the appliSiPMs
cation of individual and small arrays of SiPMs (
mm area each) for this purpose with significant sucof
cess [5], [6]. A large-area, silicon-based detector of approxicm was recently presented [4]. These tests have
mately
demonstrated high spatial and energy performances but also
temporal performance [7]. In particular, SiPMs are proposed for
time-of-flight applications due to its fast response [8]. However,
there are not yet that many developments using large-area arrays of SiPMs achieving good performanc. In this paper, we will
show the good overall performance of an array of 144 SiPMs
with final array dimensions of roughly
cm .
There are primarily two SiPM array readout approaches,
namely networks based on analogue devices [9], [10], but very
recently application-specific integrated circuits (ASICs) have
also appeared to be good candidates [11]–[14]. There are several projects in which different resistor networks (also diodes)
have been studied in detail for both PMTs and SiPMs [9]–[16].
The most important advantages of using this approach are good
linearity, high dynamic range, and also good timing resolution.
However, these methods are limited to provide information
further than planar impact position and time. Moreover, they
may result in a high system cost when designing a large field
of view scanner. Depth-of-interaction (DOI) information is
possible to be also determined using analogue readouts with
methods such as the phoswhich [17] or additional photosensors,
but also by methods in which the resistor network is upgraded
to provide information related to the DOI [18].
The alternative of using ASICs has recently showed good
results for photosensor readout, especially SiPMs, due to their
0018-9499 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
2
IEEE TRANSACTIONS ON NUCLEAR SCIENCE
entrance and exit faces were also polished, where the entrance
additionally included an ESR reflector layer.
A. Application-Specific Integrated Circuit
Fig. 1. (Left) Stack of three boards with the SiPM array in the front. (Right)
Photograph of the middle board showing the three ASICs.
versatility [19]–[21]. They can be designed to provide high temporal performance but also accurate determination of the photon
impact. They have been used for both crystal arrays and monolithic scintillators.
In this work, we will describe a scalable ASIC that has been
designed to control SiPMs [12]. We have put together arrays of
144 SiPMs that are ASICs controlled. In particular, three ASICs
are required to control each 144-SiPM array. We will describe
the design of the SiPM array and their potential use as a photodetector for gamma rays. We will show the first results obtained
with this assemble using crystal arrays.
II. MATERIALS
Our research team has accumulated experience in the design and construction of dedicated PET systems, starting with
the small animal PET called Albira [22], and then the dedicated breast PET named MAMMI [23]. The photosensors used
in these systems were based on the well-established PSPMT
technology. As with many other research groups, it was our recent goal to replace PMT sensors by SiPM technology in order
to improve the photodetector performance. Arrays of SiPMs
are envisaged to deliver high intrinsic spatial resolutions and
good timing response, but also immunity to magnetic fields. We
carried out a preliminary work with arrays of 256
MPPCs of the type S10362-11-050 from Hamamatsu, with an
active area of
mm and 50
m cell size [11]. They were
mounted on a
cm PCB with a pitch of 3 mm. Due to the
dead area and the scintillation light coupling to the photosensor
using special light guides [24], [25], we encountered limited detector performance. In order to improve operation, one alternative has been to use larger active-area SiPMs with a reduced gap
between them. In particular, we have designed and mounted an
array of 144
SiPMs from SensL. In this case we used
SiPMs of the MicroFB-30035 SMT series that have an active
area of
mm with a total outside dimension of
mm .
These sensors are packaged in a
mm clear molded reflow
lead frame. This package is soldered in standard reflow ovens.
The SiPMs were mounted with a pitch of 4.2 mm. The flatness
of all 144 SiPMs was measured to be below
m. The SiPM
array active area covers
mm . See Fig. 1.
The measurements were performed with crystal arrays of different pixel size (
mm ,
mm and
mm ) and
10–12 mm thickness. In all cases, the individual lateral crystal
pixels were polished, and ESR (3M) reflectors were used. The
The SiPM array was controlled by means of three ASICs [12],
as shown in Fig. 1, right. These ASICs are a CMOS integrated
front-end architecture. The chip is PSPMT and SiPM compatible, presents low gain dispersion among inputs, low noise, and
high speed response [13]. The underlying architecture calculates the moments of the detected light distribution in an analog
mode. Due to the additive nature of the moment calculation,
the operation can be carried out on a single device or split it
into several devices, adding the partial results afterward. All the
operations are carried out in current mode, and the weighting
operations are implemented using linear MOS current dividers.
Weighting coefficients are programmable via an I2C bus and
stored in 8-bit registers. Finally all the weighted currents are
added together and introduced in an output current buffer [13].
Selecting the proper set of weights allows one to estimate
many characteristic parameters of the light distribution, e.g.,
the centroids of the light distribution, their standard deviations,
skewness, etc. The 144 ASIC inputs are replicated eight times.
All the input signals are added forming eight linear combinations of the 144 input signals. In these experiments, the ASIC
output 1 provided the data acquisition Ssystem (DAS) with the
trigger signal. This output is obtained by loading identical coefficients to all SiPMs. ASIC output channels from 2 to 5 were
programmed to deliver planar photon impact information. These
signals were loaded with coefficient matrices representing the
Anger logic (A, B, C, D), that is, diagonal gradient coefficients
[26]. The sixth output provided DOI information, since it was
calculated as the variance of the light distribution [27]. The remaining two channels were not programmed and used in these
experiments.
This particular ASIC allows one to connect SiPM devices
with a terminal capacitance below 40 pF. The new SensL SiPMs
provide, in addition to the standard output
pF , an additional output (named FAST) with a terminal capacitance of
only 30 pF. It has been shown the convenience of using this
output for high time resolution measurements. Unfortunately, it
was not possible to directly use this signal with the ASIC, because it has a positive signal polarity, not matching the negative
input requirement of this particular ASIC. The solution we came
with was to connect the FAST output to ground. In order to limit
the current flowing through the outputs in this new implementation, an additional serial resistor was mounted with resistance
values ranging from 0.1 to
k ( k was the current selected value). In this way the final terminal capacitance of the
anode-cathode signal is 30 pF, matching well with the ASIC requirements. Fig. 2 depicts the circuitry schematic used for this
array and ASIC compatibility.
B. Data Acquisition System
All output signals (trigger and impact positioning) are
transferred to the DAS by means of coaxial cables with
impedance. The DAS used in the present experiments
was mainly composed of a trigger board and multichannel
analog-to-digital converters (ADCs) boards, both in CAMAC
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
GONZÁLEZ et al.: PERFORMANCE STUDY OF A WIDE-AREA SIPM ARRAY, ASICS CONTROLLED
Fig. 2. (Left) Circuit schematic with an output capacitance of
pF in park and 700 pF. (Right) ASIC output trigger signals, 100 mV and
allel with
50 ns plot divisions.
format. The trigger signal (calculated from the sum of all
SiPMs signals) uses a double leading-edge approach, and the
main threshold was set to
mV, suppressing most of the
electronic noise. The trigger signal threshold reduces low-energy impacts that have been multiplexed along the SiPM array.
As an energy window around the 511 keV peak is set when
working with PET applications, the effect of the threshold level
on the spatial resolution is low. The integration time for each
ADC channel was set to 192 ns. All measurements presented in
this report were performed in singles mode (no coincidence).
Thus, all trigger signals above the selected threshold enabled
a synchronized gate signal that started the digitization of the
output ASIC channels on the ADC. These data were sent
through Ethernet to a personal computer workstation and stored
in a list mode. Specially designed applications were used for
data presentation and analysis.
III. MEASUREMENTS AND RESULTS
We have evaluated the performance of a wide-area SiPM
array in terms of spatial resolution, energy resolution, and DOI
capabilities, as a function of temperature, Vop, and crystal size.
Since temporal resolution is also an important parameter for
PET applications, a deep analysis concerning timing performance with these sensor arrays has been recently carried out
by the authors in a separate work [28].
A. SiPM Array Bias Adjustment
The used SiPMs have their breakdown voltage at about
24.5 V. The whole SiPM array was biased to a common
voltage. By analyzing the energy and intrinsic detector spatial
resolution, we determined the optimum operational overvoltage
for these arrays. In these experiments, the SiPM arrays were
coupled to a crystal array of
elements with
mm
size each and through an acrylic spreader window of 1.7 mm
thickness. Optical grease with an index of refraction of 1.46
(Visilox V-788) was used to couple all these elements. In the
experiments carried out to define the best Vop and temperature parameter, the ASICs were programmed to only return
signal information from one quadrant of the SiPMs array. This
configuration offers sufficient information about optimum bias
and temperature without the need of programming the entire
SiPM array. The detector block assembly was located inside
a thermal controlled environment at a detector temperature of
3
Fig. 3. (Left) Contour plot of
mm crystal array pixels at 30 Vop, for
at the 511 keV photopeak. (Top right) Energy resan energy window of
olution of nine pixels as a function of the Vop. The red line is the first-order
exponential decay fit to the data. (Bottom right) Individual pixels energy resolutions and average are plotted.
about 17 C). A gamma ray source was used based on a matrix
of
sources with a total activity of about
Ci.
The source array was located close to the entrance face of the
crystal array.
As observed in Fig. 3 all crystal pixels could be distinguished.
The operational voltage (Vop) of the array was sequentially
varied from 25 V to 31 V and the data stored in list mode files.
Several regions of interest (ROI) around one crystal pixel and a
group of nine pixels were analyzed in order to study the energy
resolution. On the right side of Fig. 3 we have depicted the determined energy resolution (%) as a function of the Vop in volts
for the group of pixels (top) and for individual pixels (bottom).
The energy resolution for each histogram was determined by
fitting a Gaussian curve. The percentage is calculated as the
ratio of full width at half maximum (FWHM) to the centroid
value. In these figures, we observe an exponential decay of the
energy resolution, improving with the increase of the Vop. The
ratio of the 1274/511 keV peaks in
as a function of the
Vop resulted on a standard deviation of only 2%, suggesting a
negligible nonlinearity effect of the detector assembly. Energy
resolution values for individual pixels as good as 9% were
measured, at Vop of 29–30 V.
Additionally to the energy resolution, the spatial resolution
was also evaluated as a function of the Vop. Here, profiles
through one row of crystal pixels were analyzed. We observed
an improvement of pixel identification for Vop higher than
28.5 V, with performance remaining almost constant up to
30.5 V. Fig. 4 depicts profiles through one row of crystal
pixels as a function of different Vop. Based on these results,
we selected an optimal Vop value of
V for the next
experiments.
B. Temperature Dependence
SiPMs are known to be temperature sensitive, showing better
performance at lower temperatures. This effect is characteristic
for solid-state-based detectors due to thermal charge generation.
As for the tests on the optimum bias, we also analyzed here the
intrinsic detector spatial resolution and the energy resolution,
but as a function of the detector block temperature. In this experiment, the detector block temperature was controlled by means
of a closed box made out of Porexpan. This box was also covered with a dark blanket. The temperature inside the box was
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
4
Fig. 4. Profiles through one row of crystal pixels as a function of Vop varying
energy window around 511 keV.
from about 25 V to 31 V, for a
IEEE TRANSACTIONS ON NUCLEAR SCIENCE
Fig. 6. Profiles through one row of crystal pixels as a function of detector temaround the
perature varying from 20 to 30 C. An energy window of
511 keV photopeak was selected. (Left) Twelve-pixel profile. (Right) Detail for
three edge pixels.
temperatures below
C (Fig. 6). As expected, an overall
detector improvement is observed at low temperatures.
C. Spatial Resolution Dependence on SiPM Dark Noise
Fig. 5. (Top) Energy resolution (FWHM in %) as a function of temperature
for one (opened squares) and nine (full circles) crystal pixels, respectively.
(Bottom) Photopeak centroid (ADC channel) as a function of the temperature.
regulated by means of a chiller that cyclically pumped glycolic
water to a cold plate located inside the box. The detectors were
positioned on top of such a cold plate.
The experiments were carried out with a crystal array composed of pixels with
mm transversal size. Two ROI
around one and nine pixels were considered for all temperature
measurements that ranged from 20 to 40 C. Fig. 5, bottom,
shows the 511 keV photopeak ADC channel as a function of
the temperature. As expected, there is a gain increase of the
photon detection efficiency (PDE), proportional to the decrease
of the system temperature. We have determined a gain increase
of
C. In Fig. 5, top, we have plotted the energy
resolution for one and nine crystal pixels, as a function of the
detector temperature. An energy resolution improvement is observed at lower temperatures, reaching values of 7.5% for single
crystal pixels.
A profile of one row of crystal pixels provides a visual
demonstration of the detector intrinsic spatial resolution improvement with the temperature decrease. This is observed
as a more prominent signal-to-noise ratio (SNR) measure at
In this section, we present a unique study in which we have
investigated the intrinsic resolution of the detector block, as a
function of the SiPM array size. The measurements were carried out at 30 Vop. The novelty of this work is the capability
of varying the area of SiPMs contributing to the trigger signal
and center-of-gravity (CoG) determination for the planar photon
impact position (ASICs outputs second to fifth). This was possible by programming the three ASICs with different trigger and
CoG matrices according to the SiPM area under study. Areas of
,
,
, and
SiPMs were selected. All
144 SiPM signals are transferred to the ASICs without the capability for individual amplitude thresholds. Individual thresholds
for each SiPM prior to the ASIC [4], [10]or clustering SiPMs
around the highest peak channels [29] to reduce SiPM noise
contributions have been studied by other groups but not considered in this ASIC development. After ASIC calculations, the
eight calculated signals are fed to the DAS, where only an amplitude threshold can be used on the trigger signals. In order
to isolate selected SiPM array areas both the trigger and the
four impact determination signals outside these areas were programmed to zero, and the new matrices recalculated for the
planar impact characterization.
The experiments were carried out for crystal arrays with pixel
sizes of
,
, and
mm , respectively. In order
to attain the best SiPM array response uniformity, spreader windows of 1.7, 1.7, and 2.1 mm were used, respectively. Fig. 7
(left) shows, using contour plots, the results for crystal pixel
sizes of
mm for different SiPM arrays active areas. The
data were taken in single mode, and the representations shown
in the following figures are depicted for an energy window of
about 400–600 keV (at
C). When
SiPMs are enabled, some border effects and a poor resolution at the array
edges are observed. These effects are significantly reduced for
an active area defined by
SiPMs and almost completely
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
GONZÁLEZ et al.: PERFORMANCE STUDY OF A WIDE-AREA SIPM ARRAY, ASICS CONTROLLED
Fig. 7. Contour plots for crystal arrays of
mm (four plots on the left)
mm (four plots on the right) sizes, as a function of the SiPM
and of
active area.
Fig. 8. Profiles through one row of crystal pixels
mm for different
SiPM areas. The black dashed and red solid lines represent the average noise and
signal values, respectively.
vanished for arrays smaller than
SiPMs. Please note that
the area for
SiPMs has dimension of roughly
mm ,
and the one for
still covers
cm . The latter size is
still large and comparable to sizes used in many studies carried
out elsewhere [29], [30], with active dimensions of
mm
and
mm , respectively.
The experiments carried out with pixels of
mm
size were performed at a lower temperature
C in order to
improve the system’s performance (Fig. 7, right). The results for
an area defined by
SiPMs again showed lower spatial
resolution compared to those obtained for smaller SiPM areas.
In both
and
mm pixel cases, when
SiPMs
and smaller areas were selected, the achieved spatial resolution
allowed one to resolve most of the pixels within the active field
of view.
In Fig. 8, the profiles through one row of crystal pixels for
the four measured cases when using
mm pixel sizes
are shown. These results confirm that the level of accumulated
noise is a constant background per unit of area that reduces the
peak-to-valley determination when increasing the SiPM active
area.
The SNR data obtained through the profiles as those exemplified in Fig. 8, for the
mm ,
mm , and
mm
5
Fig. 9. SNR as a function of the number of SiPMs for the different crystal
pixel sizes.
pixel cases are included in Fig. 9. The SNR values were calculated as the peak-to-valley ratio of one profile. The error bars
are the standard deviations of peaks and valleys counts, respectively. We observed an improvement of the signal when reducing the number of SiPMs. Another source of noise such as
the natural radioactivity of the Lutetium in the scintillator is
hardly contributing to the SNR data, since an energy window
around the 511 keV photopeak was applied, in addition to its low rate compared to the
sources.
These tests suggested that smaller crystal pixels of
mm
would hardly be resolved for the entire
SiPM array.
In order to show the system capability to resolve these small
pixels, we directly analyzed the case for
SiPM arrays,
which is comparable with studies carried out by other groups. A
smaller area of
SiPMs was also tried in order to observe if
an improvement could further be observed. However, as Fig. 9
depicts, there is no improvement for areas smaller than
SiPMs. In Fig. 10 we have plotted the results for
mm
pixels and two array areas. In the bottom, a profile of one row
of crystal pixels is plotted.
The behavior of the SNR values for
and
SiPMs
in the case of a
mm pixel area could be explained if we
consider that the relatively small amount of light (i.e., signal)
collected in that case cannot compensate for the noise reduction
produced when the
and
SiPMs configurations are
used. In this sense, the
mm pixel area configuration seems
to limit the SNR improvement when reducing the photosensor
area.
D. Depth of Interaction With Crystal Arrays
In the search of replacing PSPMT by arrays of SiPMs, photon
depth of interaction has to also be preferentially resolved. One
of the most popular approaches when using crystal arrays to return DOI information is the so-called phoswich concept [17].
Here, crystal arrays with different decay times are assembled together in a stack, one on top of the other. The impacted crystal
layer is determined by differentiating the distinct scintillation
decay time of that layer. When high intrinsic spatial resolutions
are reached, an alternative to the phoswich is the staggered approach [31]. In this method, the pixels on one crystal layer are
shifted in the X and Y directions, for example, by the size of half
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
6
IEEE TRANSACTIONS ON NUCLEAR SCIENCE
Fig. 11. Staggered approach reading
1274 keV energy windows.
SiPMs for the 511 keV and
Fig. 10. (Top) Contour plots for crystal arrays of
mm together with
(bottom) the profiles through of one row. (Left) 6 6 SiPMs area. (Right) 3
3 SiPMs. The measures were taken at an assembly temperature of 14 C.
pixel. Therefore, the light of one top pixel is shared among four
pixels on the bottom layer. The layer identification is typically
done by CoG, although other methods like direct pixel identification could be enabled. There is also a possibility to combine
the phoswich approach with the staggered approach to maximize the number of distinguishable scintillation layers in the
package.
As a proof of concept, we mounted two blocks of
mm
LYSO pixel size following the staggered configuration. The
block covered an area of approximately
mm , and the
entire assembly was kept at 13 C. The thicknesses of the two
slabs were 10 mm each. We again run several tests varying the
number of activated SiPMs. When the whole
array was
enabled, it was hard to distinguish between the two crystal slabs.
Note that when using the described ASICs readout, even if only
a portion of the SiPM array is covered by crystals, all SiPM
signals (including noise) contribute to the CoG determination.
In Fig. 11 the results are shown for energy windows
centered at 511 and at 1274 keV, respectively. The resolution
and pixel layer determination improves with the reduction of
the SiPM array area, as expected. When a smaller active area
of 8 8 SiPMs was programmed, they were differentiated.
The top of Fig. 12 shows the sequence of results for the staggered approach with
mm pixels, as a function of detection areas. In the bottom of Fig. 12, the profiles of two consecutive rows (i.e., crystal layers) for the
SiPM case are
shown. Well-differentiated top and bottom scintillation layers
for this configuration are visible. The higher number of events
in the top curve correlates to the location of the gamma ray
source, being above this crystal slab, while the second layer was
shielded by the first array.
Fig. 12. (Top) Contour plots for the staggered configuration at 511 keV window
as a function of the number of activated SiPMs. (Bottom) The profiles for two
consecutive rows of pixels. (Solid line) From top layer. (Dashed line) From
bottom layer.
IV. DISCUSSION AND CONCLUSIONS
In this paper, we have shown the design of a gamma ray photosensor based on an array of SiPMs, ASICs controlled, in order
to replace PSPMTs in some dedicated PET systems. The detector module covers an active area of about
cm , well
matching the area of H8500/H9500 PSPMT. A special circuit
configuration of the SiPM array had to be implemented in order
to make it compatible with our previously developed ASIC. This
new circuitry has not shown performance degradation at the rate
levels that these experiments were carried out at.
The bias voltage and temperature experiments showed the expected dependence of the SiPMs assembly with these parameters. Nevertheless, we found the present results to be of practical importance, since there are only few reported studies with
SiPM arrays of similar dimensions
cm studying these
effects [32]. An energy resolution as good as 8%, for individual
crystal pixels, was reached at 5 V overvoltage. The temperature
stabilization method that we have used has nicely worked in the
laboratory environment, but other methods are being studied in
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
GONZÁLEZ et al.: PERFORMANCE STUDY OF A WIDE-AREA SIPM ARRAY, ASICS CONTROLLED
7
Summarizing, a detector block combining a SiPM array that
is controlled with three ASICs, has been shown to return a good
tradeoff performance. A unique ASIC versatility individually
controlling the loaded matrix coefficients has allowed us to provide information on the detector behavior as a function of the
chosen SiPMs array area.
REFERENCES
Fig. 13. Three-dimensional representation of the generated field of view for
a PET system formed by ten photosensor arrays (white cylinder) compared to
alternatives approaches. The left sketch shows the superposition of a smaller
field of view in red color with higher image performance when only half array
is activated in the transaxial plane. The right side depicts the example when the
half active array is in the axial direction, defining an improved performance with
shorter axial length (green color).
order to extend the cooling approach to a higher number of detector blocks.
We found an overall good performance of the entire detector
block (crystal array,
SiPMs matrix and ASICs PCB)
when coupled to pixel arrays with
mm size at 30 Vop
and detector temperatures below 20 C. For smaller pixel sizes
starting from
mm , we observed that if the whole SiPM
assembled array is activated, it is hard to reach a good detector
performance. However, for a SiPMs array of
devices and
smaller, there is also a good detector performance. The active
area defined by such an array, would cover about
mm ,
slightly above some of the current PMT-based or APD-based
commercial PET imagers.
One of the most interesting features of the studied detector
block encountered during the performed experiments has been
the capability of the ASICs to allow one to define different photosensor array areas. This option has permitted the evaluation
of the detector performance as a function of the crystal pixel
size and the dark noise contribution from the changing active
detector size.
As a future development work one might think on a system
implementation combining both a moderate spatial resolution
and a high-resolution detector mode. The latter would be constrained for instance by a reduced axial or transaxial field of
view. See Fig. 13. In more detail, a system formed by two rings
of these types of sensors could for instance define adjacent arrays of 6 (axial) 12 SiPMs which would result in a system
with improved spatial resolution in the axial length of
SiPMs, array 1 and array 2, respectively. In the case of defining
arrays of
(radial) SiPMs one could have a high-resolution performance along the two axial arrays but with decreasing
performance in the transaxial field of view.
In this paper, we have also shown the system’s capability to
provide DOI information by means of implementing a staggered
approach. We investigated this feature with crystal pixels of
mm . Here, we observed that the detector block could
not provide accurate DOI information when the entire
SiPM array was enabled, but it might be possible with an active
area of about
mm (
SiPMs).
[1] S. Seifert, R. Vinke, H. T. van Dam, H. Löhner, P. Dendooven, F. J.
Beekman, and D. R. Schaart, “Ultra precise timing with SiPM-based
TOF PET scintillation detectors,” in IEEE Nuclear Science Symp.
Conf. Record, 2009, vol. J01-4, pp. 2329–2333.
[2] S. Mohers, A. D. Guerra, D. J. Herbert, and M. A. Mandelkern, “A detector head design for small-animal PET with silicon photomultipliers
(SiPM),” Phys. Medi. Biol., vol. 51, p. 1113, 2006.
[3] D. R. Schaart, H. T. van Dam, S. Seifert, R. Vinke, P. Dendooven, H.
Löhner, and F. J. Beekman, “A novel, SiPM-array-based, monolithic
scintillator detector for PET,” Phys. Med. Biol., vol. 54, p. 3501, 2009.
[4] R. R. Raylman, A. Stolin, S. Majewski, and J. Proffitt, “A large area,
silicon photomultiplier-based PET detector module,” Nucl. Instrum.
Methods Phys. Res. A, vol. 735, p. 602, 2014.
[5] T. Kato, J. Kataoka, T. Nakamori, T. Miura, H. Matsuda, K. Sato, Y.
Ishikawa, K. Yamamura, N. Kawabata, H. Ikeda, G. Sato, and K. KaMPPC array
mada, “Development of a large-area monolithic
for a future PET scanner employing pixelized Ce:LYSO and Pr:LuAG
crystals,” Nucl. Instrum. Methods Phys. Res. A, vol. 638, p. 83, 2011.
[6] H. S. Yoon, G. B. Ko, S. I. Kwon, C. M. Lee, M. Ito, I. C. Song,
D. S. Lee, S. J. Hong, and J. S. Lee, “Initial results of simultaneous
PET/MRI experiments with an MRI-compatible silicon photomultiplier PET scanner,” J. Nucl. Med., vol. 53, p. 608, 2012.
digital SiPM array with 192
[7] S. Mandail and E. Charbon, “A
TDCs for multiple high-resolution timestamp acquisition,” J. Instrum.,
vol. 8, p. P05024, 2013.
[8] C. L. Kim, G. C. Wang, and S. Dolinsky, “Multi-pixel photon counters
for TOF-PET detectors and its challenges,” IEEE Trans. Nucl. Sci., vol.
56, no. 5, pp. 2580–2585, Oct. 2009.
[9] P. Dokhale, C. Stapels, J. Christian, Y. Yang, S. Cherry, W. Moses,
and K. Shah, “Performance measurements of a SSPM-LYSO-SSPM
detector module for small animal positron emission tomography,” in
IEEE Nuclear Science Symp. Conf. Record, 2009, pp. 2809–2812.
[10] S. Majewski, J. Proffitt, A. Stolin, and R. Raylman, “Development of a
“resistive” readout for SiPM arrays,” in IEEE Nuclear Science Symp.
Conf. Record, 2011, pp. 3939–3944.
[11] P. Conde, A. J. González, L. Hernández, P. Bellido, A. Iborra, E.
Crespo, L. Moliner, J. P. Rigla, M. J. Rodríguez-Álvarez, F. Sánchez,
M. Seimetz, A. Soriano, L. F. Vidal, and J. M. Benlloch, “Results
of a combined monolithic crystal and an array of ASICs controlled
SiPMs,” Nucl. Instrum. Methods Phys. Res. A, vol. 734, p. 132, 2014.
[12] V. Herrero-Bosch, C. W. Lerche, M. Spaggiari, R. Aliaga-Varea,
N. Ferrando-Jodar, and R. Colom-Palero, “AMIC: An expandable
front-end for gamma-ray detectors with light distribution analysis
capabilities,” IEEE Trans. Nucl. Sci., vol. 58, no. 4, pp. 1641–1646,
Aug. 2011.
[13] V. Herrero-Bosch, J. M. Monzo, A. Ros, R. J. Aliaga, A. González, C.
Montoliu, R. J. Colom-Palero, and J. M. Benlloch, “Programmable integrated front-end for SiPM/PMT PET detectors with continuous scintillating crystal,” J. Instrum., vol. 7, Dec. 2012.
[14] M. D. Rolo, R. Bugalho, F. Gonçalves, A. Rivetti, G. Mazza, J. C.
Silva, R. Silva, and J. Varela, “A 64-Channel ASIC for TOFPET applications,” in IEEE Nuclear Science Symp. Conf. Record, 2012, pp.
1460–1464.
[15] A. J. González, M. Moreno, J. Barberá, P. Conde, L. Hernández, L. Moliner, J. M. Monzó, A. Orero, A. Peiró, R. Polo, M. J. Rodriguez-Alvarez, A. Ros, F. Sánchez, A. Soriano, L. F. Vidal, and J. M. Benlloch, “Simulation study of resistor networks applied to an array of
256 SiPMs,” IEEE Trans. Nucl. Sci., vol. 60, no. 2, pp. 592–598, Apr.,
2013.
[16] S. Cherry, Y. Shao, S. B. Siegel, and R. W. Silverman, “High resolution
detector array for gamma-ray imaging,” US Patent 5 719 400, 1998.
[17] C. M. Pepin, M. Bergeron, C. Thibaudeau, C. Bureau-Oxton, S.
Shimizu, R. Fontaine, and R. Lecomte, “Digital identification of fast
scintillators in phoswich APD-based detectors,” IEEE Trans. Nucl.
Sci., vol. 57, no. 2, pp. 1435–1440, Apr. 2010.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
8
IEEE TRANSACTIONS ON NUCLEAR SCIENCE
[18] C. Lerche, J. Benlloch, F. Sánchez, N. Pavón, B. Escat, E. Giménez,
M. Fernández, I. Torres, M. Giménez, A. Sebastiá, and J. Martínez,
“Depth of -ray interaction within continuous crystals from the width
of its scintillation light-distribution,” IEEE Trans. Nucl. Sci., vol. 52,
no. 1, pp. 560–572, Feb. 2005.
[19] P. Barrillon, S. Blin, M. Bouchel, T. Caceres, C. de La Taille, G. Martin,
P. Puzo, and N. Seguin-Moreau, “MAROC: Multi-anode readout chip
for MaPMTs,” in IEEE Nuclear Science Symp. Conf. Record, 2006, pp.
809–814.
[20] M. Ritzert, P. Fischer, V. Mlotok, I. Peric, C.o Piemonte, N. Zorzi, V.
Schulz, T. Solf, and A. Thon, “Compact SiPM based detector module
for Time-of-Flight PET/MR,” in IEEE NPSS Real Time Conf., 2009,
pp. 163–166.
[21] F. Corsi, M. Foresta, C. Marzocca, G. Matarrese, and A. Del Guerra,
“BASIC: An 8-channel front-end ASIC for silicon photomultiplier
detectors,” in IEEE Nuclear Science Symp. Conf. Record, 2009, pp.
1082–1087.
[22] F. Sánchez, A. Orero, A. Soriano, C. Correcher, P. Conde, A.
González, L. Hernández, L. Moliner, M. J. Rodríguez-Alvarez, L. F.
Vidal, J. M. Benlloch, S. E. Chapman, and W. M. Leevy, “ALBIRA:
A small animal PET/SPECT/CT imaging system,” Med. Phys., vol.
40, p. 051906–1, 2013.
[23] L. Moliner, A. J. González, A. Soriano, F. Sánchez, C. Correcher, A.
Orero, M. Carles, L. F. Vidal, J. Barberá, L. Caballero, M. Seimetz, C.
Vázquez, and J. M. Benlloch, “Design and evaluation of the MAMMI
dedicated breast PET,” Med. Phys., vol. 39, p. 5393, 2012.
[24] A. J. González Martínez, A. Peiró Cloquell, F. Sánchez Martínez, L.
F. Vidal San Sebastian, and J. M. Benlloch Baviera, “Innovative PET
detector concept based on SiPMs and continuous crystals,” Nucl. Instrum. Methods Phys. Res. A, vol. 695, p. 213, 2012.
[25] A. J. González, A. Peiró, P. Conde, L. Hernández, L. Moliner, A. Orero,
M. J. Rodríguez-Álvarez, F. Sánchez, A. Soriano, L. F. Vidal, and J.
M. Benlloch, “Monolithic crystals for PET devices: Optical coupling
optimization,” Nucl. Instrum. Methods Phys. Res. A, vol. 731, p. 288,
2013.
[26] M. Spaggiari, V. Herrero, C. W. Lerche, R. Aliaga, J. M. Monzó, and
R. Gadea, “AMIC: an expandable integrated analog front-end for light
distribution moments analysis ,” J. Instrum., vol. 6, Jan. 2011.
[27] V. Herrero, N. Ferrando, J. D. Martínez, Ch. W. Lerche, J. M. Monzó,
F. Mateo, R. J. Colom, R. Gadea, A. Sebastiá, and J. M. Benlloch,
“Position sensitive Scintillator based detector improvements by means
of an integrated front-end,” Nucl. Instrum. Meth. A, vol. 604, p. 77,
2009.
[28] J. Torres, A. Aguilar, R. Garcia-Olcina, J. Martos, J. Soret, A. J. Gonzalez, P. Conde, L. Hernandez, F. Sanchez, and J. M. Benlloch, “Highresolution multichannel Time-to-Digital Converter core implemented
in FPGA for ToF measurements in SiPM-PET,” presented at the 2013
IEEE Nuclear Science Symp. Medical Imaging Conf., 2013.
[29] G. Llosá, J. Barrio, C. Lacasta, M. G. Bisogni, A. Del Guerra, S. Marcatili, P. Barrillon, S. Bondil-Blin, C. de la Taille, and C. Piemonte,
“Characterization of a PET detector head based on continuous LYSO
crystals and monolithic, 64-pixel silicon photomultiplier matrices,”
Phys. Med. Biol., vol. 55, p. 7299, 2010.
[30] C. J. Thompson, A. L. Goertzen, J. D. Thiessen, D. Bishop, F. Retiere,
P. Kazlowski, G. Stortz, and V. Sossi, “Measurement of energy and
timing resolution of very highly pixellated LYSO crystal blocks with
multiplexed SiPM readout for use in a small animal PET/MR insert,”
presented at the 2013 IEEE Nuclear Science Symp. Medical Imaging
Conf., 2013, M12-41.
[31] N. Inadama, T. Moriya, Y. Hirano, F. Nishikido, H. Murayama, E.
Yoshida, H. Tashima, M. Nitta, H. Ito, and T. Yamaya, “X’tal cube
PET detector composed of a stack of scintillator plates segmented by
laser processing,” IEEE Trans. Nucl. Sci., vol. 61, no. 1, pp. 53–59,
Feb. 2014.
[32] J. Du, J. Schmall, M. S. Judenhofer, K. Di, Y. Yang, N. Pavlov, S.
Buckley, C. Jackson, and S. R. Cherry, “Evaluation of a 12 × 12 pixel
SiPM array for small-animal PET,” presented at the 2013 IEEE Nuclear
Science Symp. Medical Imaging Conf., 2013, M12-41, Abstract J1-2.
Download