1. New Two-Dimensional ASICs for Solid State Pixel Detectors
Tumay O. Tumer*a
, Victoria B. Cajipea
, Martin Clajusa
, Satoshi Hayakawaa
and
Alexander Volkovskiia
a
NOVA R&D, Inc.,1525 Third Street, Suite C, Riverside, CA USA 92507-3429
ABSTRACT
We have developed high energy and high spatial resolution two-dimensional (2D) solid-state imaging pixel
detectors and their custom integrated circuits (ICs). Solid-state pixel detectors and their readout ICs are now
regarded to be an integral part of position-sensitive semiconductor detectors such as Si, CdTe and CdZnTe for x-ray
and gamma-ray imaging. These detectors have a 2D structure. We have also developed one-dimensional (1D)
detectors, which are mostly used for scanning type imaging. The new 2D pixel detectors we have developed can be
used for both scanning and staring mode imaging applications. Because the requirements of various detector
applications tend to be diverse, a custom IC is typically designed for a specific detector array. This often lengthens
the time and raises the cost of system development. To help close the readout technology gap and facilitate advances
in this field, we have been formulating and implementing strategies for instrumenting different detectors of a given
application category with highly versatile ICs that meet a range of requirements. The solid-state pixel detectors that
have been developed within this effort are presented below.
Keywords: Two-dimensional solid-state detector, Multichannel mixed signal ASIC, X-ray detector, Gamma ray
detector, Detector readout IC, Front-end electronics, Solid-state detector, Imaging detector, Pixel detector. 1
INTRODUCTION
Applications of position-sensitive solid-state radiation detectors have grown rapidly in recent years as a consequence
of parallel advances in detector materials science and low-noise microelectronics design [1], [2]. The sensor and
signal processing components of a detector module have increasingly assumed monolithic forms, i.e., pixel or strip
arrays and multi-channel integrated circuits (ICs), respectively. Whereas the concept for a given detector application
often revolves around the characteristics and capabilities of a sensor array, actualizing an instrument design assumes
that the matching readout electronics are already available or will be developed as part of the program. As such, it
has become generally recognized that readout ICs and their associated support electronics are an essential and
integral part of a detector unit. Ideally, a readout IC would be designed to meet the specific performance
requirements of a given application. Custom integrated circuit development can however be a time-consuming,
high-risk and financially onerous undertaking, which most R&D projects cannot afford. Broadening the applicability
and accessibility of readout ICs that do get development funding would therefore benefit manifold efforts to
prototype emerging detector designs. To this end, one could adopt a “generic IC” approach, premised on the fortune
of direct sponsorship that is virtually independent of instrument development. This would proceed by: identifying
commonalities within an application category; defining corresponding generic IC classes; and, developing an IC for
the requirements of each class. More realistically, since most new IC designs are carried out within an application
development program, one could simply expand the features of the Application-Specific IC (ASIC) in question. This
approach would provide ranges and options for key specifications and accommodate additional functions as far as
technically and logistically reasonable.
We have taken both approaches in developing solid-state detectors with readout ICs. Fig. 1 outlines our IC
development roadmap divided according to detector application class. Under the spectroscopy category is the
RENA™ [3] family of chips. We have developed an all-new “RENA-3” version of this linear multi-channel circuit
as a generic IC [8]; a chip named “DANA™-2” with a similar channel design repeated over a two-dimensional (2D)
array has also been fabricated. Each pixel area of DANA-2 contains one channel of a RENA-3 IC. Energy-binning
applications form another important class for which we developed the FESA™ IC [4] and, more recently, the linear
XENA™-2 [5] and 2D HILDA™-2 IC. For imaging with high spatial resolution, the third application category, we
* Correspondence: Email: Tumay.Tumer@novarad.com; Telephone: (951) 781-7332; Fax: (951) 781-0178.
Hard X-Ray, Gamma-Ray, and Neutron Detector Physics X, edited by Arnold Burger, Larry A. Franks, Ralph B. James,
Proc. of SPIE Vol. 7079, 707913, (2008) · 0277-786X/08/$18 · doi: 10.1117/12.799392
Proc. of SPIE Vol. 7079 707913-1

2. Spectroscopy (astrophysics, nuclear medicine, weapons monitoring)
Photon-counting, photon energy acquired (CMOS)
RENATM I RENA-2 I RENA-3/3A ID multi-Channel
- DANA, DANA-2 20 pixel array
Energy-binning (security screening, industrial inspection, medical imaging)
Fast photon-counting, counts binned in several energy windows (CMOS)
FESATM —SXENATM —--tv XENA-2, XENA-2A ID muIti-channel
r—p HILDATM, HILDA-2 20 pixel array
High-resolution imaging (digital mammography, radiography, inspection)
Current mode (TDI CCD)
MARYTM5Opm I MARY-2 -)- MARY-N50 (5Opm) 20 array
- MARY-3 MARY-N 100 (100pm) 2D array
have the MARY™ [6], a TDI CCD readout channel array IC, which has recently evolved into MARY-N50/N100
versions. In the following sections we give brief descriptions of the solid-state pixel detectors [7].
Fig. 1. The development roadmap for NOVA’s 1D and 2D imaging solid-state detectors and their custom front-end ICs
divided according to detector application class.
2D SPECTROSCOPY PIXEL DETECTORS
1.1 The 2D DANA-2 Pixel Detector
The DANA IC (Detector Array for Nuclear Applications Integrated Circuit) is a 16 x 16=256 pixel with 500µm
pitch array of channels with a charge sensitive amplifier/shaper, trigger output and sparse readout intended for use in
flip-chip connection with a spectroscopy-grade 2D detector pixel array. It is in effect a 2D incarnation of the RENA-
3; it also takes heritage from an earlier 2D channel array IC developed for x-ray astrophysics [9] and investigated for
use in molecular imaging. The advantages of such a readout channel array include lower input capacitance and
noise, finer pixel pitch, more compact sensor-readout assembly and ease of tiling array units into a large-area 2D
array. On the other hand, its format entails the challenge of fitting complex circuitry of the RENA-3 channel within
each unit cell area, developing the matching detector pixel array, and finding a solution for reliable, high quality and
cost-effective IC-detector bonding.
In the 2D pixel detector design the detector pixels are connected directly onto the readout IC input pixel pads using a
bump bond technique, such as indium, gold stud, solder and Z-Bond bump bond techniques. In this case the charge
bundles created inside the solid-state detector are directly input into the IC’s preamplifier through the bump bond.
This technique results in ultra low input capacitance for small pixel sizes and prevents charge transportation on a
lossy medium such as a PCB.
Fig. 2 displays images of the DANA-2 IC layout and fabricated die while Fig. 3 shows the block diagram for a
single signal chain of the IC. The key features of DANA are summarized in Table 1. The first version of the DANA-
2 IC is being evaluated.
Fig. 4 is presenting the first results obtained from the DANA IC. DANA IC pulser spectra from a single pixel with
0.8 to 6.0 fC charge input for each pulse are shown. The DANA IC had power distribution issue due to large number
of channels. IN DANA-2 the power distribution is improved to prevent voltage drop. DANA-2 IC has been received
recently from the foundry and testing has just started.
Proc. of SPIE Vol. 7079 707913-2

3. Fig. 2. Layout (left) and photograph (right) of the DANA IC.
Active Feedback
Cf
Bump Input
Pre-Amplifier Gain Stage Shaper Filter
Peak
Detector
Token
Logic
Comparator
Vth
Tri-State
Analog Output
Fig. 3. Block diagram for a single signal chain of the DANA IC.
Table 1. Key features of the DANA-2 pixel detector.
Number of channels: 16x16 matrix = 256 channels/pixels, 500µm pitch
Frond end: Self-resetting charge sensitive amplifiers
Input energy ranges: ≈ 150 keV and 800 keV (Externally selectable)
Input polarity: Positive and negative externally selectable
Count rate capability: > 400 x10
3
counts/sec, all channels in parallel
Gain and offset: Digitally adjustable for each channel
Pulse shaping time: Selectable, 1 to 4 µs
Input referred noise: < 100 e rms
Power consumption: 1,500 mW nominal
Data readout: Controlled by programmable token logic
Daisy chaining: Up to 16 chips
Die size: 8.575 x 9.535 mm
2
2D FAST MULTI ENERGY IMAGING PIXEL DETECTORS
1.2 Multi energy fast imaging CZT pixel detector bump bonded onto the 2D HILDA IC
The HILDA IC (Hyperspectral Imaging with Large Detector Arrays) is a 16x16=256 pixel with 500µm pixel pitch
array of channels designed for high-rate photon counting and multiple-energy binning up to eight energy bands.[10]
Like the DANA, it is intended for use in flip-chip bump bonding with a matching 2D detector pixel array. Such as a
16x16, 500µm pitch CZT pixel array detector. The main applications envisioned for this IC are CT scanning,
baggage and munitions inspection.
Fig. 5 displays the HILDA IC layout and a picture of the fabricated die. The block diagram for a single channel of
the HILDA is similar to that of the XENA-2 IC except that each channel has eight rather than five parallel
Proc. of SPIE Vol. 7079 707913-3

4. iI11jJiiD]
JLH_1000
__LADC Code detected
Pulser Spectra
3000
2500
2000
1500
a
-.—PulseroBIC
—Pulserl.OIC
PuIser2.OIC
Pulser 3.OIC
--PuIser40fC
PuIser 5.OIC
PuIser 6OIC
comparators and counters. Fig. 6 shows a close up view of the CZT pixel detector bump bonded on top of the
HILDA IC. Table 2 summarizes the mai specifications of HILDA-2 pixel detector.
Fig. 4. DANA pixel detector test pulse spectra using a single pixel with 0.8 to 6.0 fC charge input per pulse.
(The settings are: Channel: 241; Gain: 31; and Peaking time: 1 us.)
Table 2. Key features and specifications of the HILDA pixel detector.
Number of channels: 16x16 channels/pixels, 500µm pitch
Front end: Self resetting preamplifier for continuous operation
Input energy range: ≈ 200 or 600 keV (CZT) maximum, globally selectable
Input polarity: Negative
Count rate capability: ≈ 2 x10
6
counts/sec-channel
Energy bins per channel: 8 (Thresholds are externally adjustable)
Gain and offset: Digitally adjustable for each channel
Input capacitance: 0.5pF optimum
Charge collection: 50-100 ns
Input referred noise: < 1000 e rms
Die size: 8.575 x 9.535 mm
2
Number of channels: 16x16 channels/pixels, 500µm pitch
Front end: Self resetting preamplifier for continuous operation
Fig. 5. From left layout, photograph of the die, photograph of the IC mounted inside the CQFP package and
mounted onto the evaluation system daughter card with a CZT pixel detector bump bonded on top of the
HILDA IC.
Proc. of SPIE Vol. 7079 707913-4

5. I
riin
S S A 1 0_//r_____- _•_5 555__
Fig. 6. Two close up photographs of the HILDA pixel detector with a 2D CZT array bump bonded on top of
the readout IC. On the right design drawing of the CZT detector array to be bump bonded on top of HILDA.
Fig. 7 shows the HILDA-2 and DANA-2 pixel detector evaluation system. It consists of a motherboard and a
daughter board, which houses a single HILDA-2 or DANA-2 pixel detector. Both the motherboard and daughter
board designs are shared between HILDA-2 and DANA-2, but the system firmware and software are different.
Fig. 7. HILDA/DANA & HILDA-2/DANA-2 Evaluation System with mother and daughter boards. The HILDA or DANA
pixel detector (HILDA is shown here) inside a CQFP package soldered onto the daughter board shown on the left on top of
the mother board.
The linearity of the amplifier response as a function of input pulse height was measured with the help of the on-
board pulser and the pixel detector’s pulse counters. The pixel detector’s six-bit gain DACs were all set to a mid-
level setting of 31 in the 600 keV range setting, the offsets were calibrated to a baseline voltage of 2.5 V. The pulser
amplitude was varied between 5 fC and 20 fC in steps of 5 fC. For each amplitude setting, we varied the comparator
threshold voltages in steps of 10 to 20 mV (depending on the amplitude) and counted for 5 ms at each step. From
these data we determined, for each cell, the threshold at which the number of counts had fallen to half the value
expected from the pulse frequency, using linear interpolation between the nearest steps. The differences between
these voltages and 2.5 V were taken as the amplifiers’ output amplitudes. This approach yields better and more
relevant results than determining the signal amplitudes at the analog test output by observing them on an
oscilloscope, for example. The results show very good linearity. Fig. 8 displays the results for channels 0 to 127 of
the 256 channel pixel detector.
A 3 mm thick CZT detector array is flip-chip bump bonded onto a HILDA IC. The detector was biased to 500 V.
The x-ray source voltage was 160 kV. Before acquiring the data, the offset DACs of the HILDA amplifiers were
calibrated to align the baseline voltages for all 256 cells.
HILDA pixel detector with a CZT array bump bonded on top is tested for its response to x-ray flux. The test system
was placed in the beam of NOVA’s 160 kV x-ray generator and the tube voltage was set to 160 kVp. Threshold
voltages were set to 2.30 V and 2.20 V, that is, 0.1 V and 0.2 V below the baseline. Count rates were measured for
10 ms count periods as a function of x-ray tube currents. Count rates were stable and increased linearly until pile up
thickness
total width
Proc. of SPIE Vol. 7079 707913-5

6. C)
CDD)=-0C)
C)(Q
0
C)
CD
C)
CD
0)
C)
CD
(Cl
0)
C)
CD
co
C)
CD
0
C)
CD
()F')
C)
CD
F')C)
CD
0)C)
CD
CoHLJ
Outputamplitude(V)
14000
12000
U) 10000
E
0
8000
U)
0.
Annn
• Level 0
. Level 1
C
0
0 4000
2000 .
0.20 0.30 0.40 0.50 0.60 0.70
X-ray tube current (mA)
started to become important. The high-flux response as a function of x-ray tube current for the comparators/counters
associated with the two lowest-level comparators of cell 122 is shown in Fig. 9.
Fig. 8. Output amplitude as a function of input pulse height for test pulses applied to cells 0 to 127 of
HILDA D17.
Fig. 9. High-flux response as a function of x-ray tube current for the counters associated with the two lowest-
level comparators of cell 122 on the HILDA pixel detector.
HIGH SPATIAL RESOLUTION IMAGING
1.3 High resolution imaging CZT pixel detectors mounted onto MARY-N50 and MARY-N100 ICs
Like the original MARY IC (MAmmogRaphY) [6], MARY-2 is a solid-state pixel detector readout IC with a 192 x
384 array of 50 x 50 micron (MARY-3 is 64 x 192 array of 100 x 100 micron pitch) pixels operated in current mode
Proc. of SPIE Vol. 7079 707913-6

7. **SAP SF5 PADS *PADREDSPACE2 PADS
**SAPOPN PADS ONE PAD CENTERED IN SPACE OF 2 PADS
with TDI using CCD charge transfer technique. The test results and images produced using the original MARY Si,
CdTe and CdZnTe pixel detectors has been published previously [6]. The solid state detector, such as silicon or
GaAs PIN photodiode array, CdTe, or CdZnTe (CZT) can be bump bonded onto the MARY-N IC. It is also possible
to deposit detector material such as amorphous Se or PbI2 on top of the IC surface. This technique can reduce cost
significantly. The basic element of the array is a TDI group of eight pixels. Charge is integrated through eight shifts
of position and then presented to an output buffer through a multiplexer and eventually to the external system. This
accumulate-and-shift operation occurs continuously.
These integrated circuits also have staring imaging mode. In this mode the full integrated circuit is read out without
TDI, which can be used to produce staring images of objects and it is also useful in testing, calibrating and
monitoring the integrated circuit and the solid state pixel detector mounted on top.
The main specifications of the MARY-N50 and MARY-N100 ICs are shown in Table 3. Fig. 10 shows a
photograph of a fabricated MARY die and the MARY-N layout. MARY-N100 layout is shown in Fig. 11. These ICs
are being tested and evaluated at present.
Table 3. Key features of the MARY-2 and MARY-3 ICs.
Readout Selectable on chip TDI using CCD technique for charge transfer or staring mode imaging.
Pixel Pitch 50x50 and 100x100 µm2
for MARY-2 and MARY-3, respectively.
Pixel Array Size 192x384 and 64x192 for MARY-2 and MARY-3, respectively.
Dynamic Range ≥ 16 Bit
Noise Designed for low noise. (Not yet measured)
Clock Drivers Internal or External
Test Input Fat Zero
Overflow Overflow Controlled
TDI 8 or 24 Independent Readout Taps for MARY-3 and MARY-2, respectively.
Input Optimized for electron collection for detectors such as CZT and Se.
Fig. 10. Photograph of MARY (top) and layout of MARY-N50 (50 µm) IC (bottom).
Fig. 11. Layout of MARY-N100 (100 µm) IC.
Proc. of SPIE Vol. 7079 707913-7

8. I
I
I I
I I I
I J I I I
H 1
Fig. 12. (Left) A Photograph of MARY-N50 and MARY-100 Evaluation System mother and daughter boards shown
together. (Right) MARY-N Evaluation System daughter board with a MARY-N50 IC wire bonded.
A sophisticated MARY-N50/N100 pixel detector evaluation system has been developed (Fig. 12). As in the DANA
and HILDA evaluation system, the MARY-N pixel detector is placed on a daughter board. This has the advantage
that it makes it easier and lower cost to make new daughter boards for different tests and applications.
MARY-N pixel detectors with electron collection are expected to have significantly improved performance
compared to the original hole collecting MARY pixel detector [7]. For example, the excellent contrast obtained in
imaging an ACR phantom (Fig. 13) with the original MARY IC and a 150 micron thick CZT detector – the small
thickness was used to reduce the effects of hole trapping in CZT – is expected to be improved significantly by the
electron collecting MARY-N pixel detectors, because thicker CZT (0.3 to 0.5 mm) can be used. Fig. 14 shows two
images of a 2 cm long Mosquito fish taken with the original MARY IC and 0.15 mm thick CdTe and CZT pixel
detectors.
Fig. 13. Comparison of contrast between the central ACR phantom image taken by a digital mammography
system (University of Toronto) and sections taken by MARY pixel detector (NOVA), which are shown on
the two sides. The improvement in the contrast and the spatial resolution are clearly visible.
Proc. of SPIE Vol. 7079 707913-8

9. 4 - -.
- 0.15 mm thick'L. — - - - - -
CdTe MARY-—.—) _________ -
picel detector
—
-
5- 0.10 mm thick
-- ooc- CZTMARY
'_4_
-i cc., • pioetdetector
Fig. 14. MARY pixel detector image of a 2 cm long Mosquito fish taken at room temperature using CdTe and
CZT arrays with 0.15 mm thickness bump bonded onto MARY. X-ray generator was run at 30 kVp and
40 mA.
SUMMARY
We developed new 2D pixel detectors for three different fields in radiation detection. These are the fields of
spectroscopic, multi-energy, and high spatial resolution imaging. All the pixel detectors developed have custom
mixed signal multi-channel readout integrated circuits, which are an integral part of the pixel detectors. These ICs
are developed to be versatile and generic to respond to varying needs of user. Direct support for the development of
such “generic” readout IC solutions would benefit the radiation detector R&D community by reducing time and cost
for prototyping new, advanced instrumentation concepts. The resulting generic designs can provide excellent
starting points for more application-optimized and/or lower-cost versions of the readout ICs.
The ICs described in this paper are either immediately available for evaluation (RENA-3, XENA-2 and HILDA-2)
or presently undergoing characterization (DANA-2 and MARY-N50/N100). A modular system to build arrays of
these pixel detectors are also under development.
NOVA welcomes all opportunities for collaboration to fully exploit the capabilities of its integrated readout
electronics and hasten the development of emerging detector applications.
ACKNOWLEDGMENT
We thank D. Ward, G. Kline and G. Visser for their valuable contributions to the various dedicated readout pixel
detector designs and testing. We also acknowledge the DoD/Army SBIR grants (DAAE30-02-C-1015 and
DAAE30-03-C-1074) for the development of the HILDA CZT Pixel detector and its custom readout IC. A DoD
BCRP grant, DAMD17-01-1-0356, for the development of the MARY™ digital mammography pixel detector, an
NCI grant with Aguila Technologies & University of California, San Diego 1 R42 CA110192-01 for the
development of the MARY-N50/N100 pixel detectors and to fabricate an integrated array for digital mammography.
We also acknowledge the technical contributions and support of the Sunnybrook & Women's College Health
Sciences Centre on the MARY pixel detector.
REFERENCES
[1] G. Lutz, Semiconductor Radiation Detectors. Berlin Heidelberg: Springer-Verlag, 1999.
[2] P. Sellin, “Advances in compound semiconductor radiation detectors: a review of recent progress”, at
http://www.ph.surrey.ac.uk/~phs2ps/psd6_sep02.pdf.
[3] S.D. Kravis, D.G. Maeding , T.O. Tümer, G. Visser, S. Yin, "Test results of the Readout Electronics for
Nuclear Applications (RENA) chip developed for position-sensitive solid state detectors," SPIE Symp. Proc.
3445, 374 (1998). See: http://www.novarad.com/pages/documents/RENA_test_results_SPIE_1998.PDF
Proc. of SPIE Vol. 7079 707913-9

10. [4] M. Clajus, T.O. Tümer, G.J. Visser, S. Yin, P.D. Willson, and D.G. Maeding, “Front-End Electronics for
Spectroscopy Applications (FESA) IC,” contribution to the IEEE Nuclear Science Symposium 2000, Lyon,
France, October 15 – 20, 2000.
[5] V.B. Cajipe, M. Clajus, O. Yossifor, R. Jayaraman, B. Grattan, S. Hayakawa, R.F. Calderwood and T.O.
Tumer, “Multi-energy x-ray imaging with linear CZT arrays and integrated electronics”, IEEE RTSD
conference record, Rome, 2004.
[6] Shi Yin et al, "Hybrid direct conversion detectors for digital mammography," IEEE Transactions on Nuclear
Science, Vol. 46, No. 6 (1999) 2093-2097.
[7] This paper focuses on readout chips for direct conversion detectors. NOVA also has ICs designed for
applications using scintillator crystals and PMTs or APDs. See for example, M. Clajus et al, “Compact detector
modules for high resolution PET imaging with LYSO and avalanche photodiode arrays” IEEE MIC conference
record, Rome, 2004.
[8] T.O. Tümer, V.B. Cajipe, M. Clajus, F. Duttweiler, S. Hayakawa, J.L. Matteson, A. Shirley, O. Yossifor. “Test
results of a CdZnTe pixel detector read out by RENA-2 IC,” Presented at the14th International Workshop on
Room-Temperature Semiconductor X-Ray and Gamma-Ray Detectors, Rome, Italy (October 2004) and
submitted to IEEE Trans. Nucl. Sci.
[9] T.O. Tümer, et al., “Preliminary test results of pixel detectors developed for the All-sky X-ray & Gamma-ray
Astronomy Monitor (AXGAM),” Trans. of Nucl. Science, 47, 1938-1944 (Dec. 2000).
[10]Clajus M, Cajipe VB, Hayakawa S, Tümer TO, Willson PD. Multi-Energy, Fast Counting Hybrid CZT Pixel
Detector with Dedicated Readout Integrated Circuit, 2006 IEEE Nuclear Science Symposium Conference
Record, Vol. 6, pp. 3602-3606.
Proc. of SPIE Vol. 7079 707913-10