OPTO-ELECTRONICS REVIEW 14(1), 59–68 DOI: 10.

2478/s11772-006-0009-x

High-performance IR detectors at SCD present and future

O. NESHER* and P.C. KLIPSTEIN

Semi Conductor Devices (SCD), P.O. Box 2250, Haifa 31021, Israel

For over 27 years, SCD has been manufacturing and developing a wide range of high performance infrared detectors, de-
signed to operate in either the mid-wave (MWIR) or the long-wave (LWIR) atmospheric windows. These detectors have been
integrated successfully into many different types of system including missile seekers, time delay integration scanning sys-
tems, hand-held cameras, missile warning systems and many others. SCD's technology for the MWIR wavelength range is
based on its well established 2D arrays of InSb photodiodes. The arrays are flip-chip bonded to SCD's analogue or digital
signal processors, all of which have been designed in-house. The 2D focal plane array (FPA) detectors have a format of
320×256 elements for a 30-µm pitch and 480×384 or 640×512 elements for a 20-µm pitch. Typical operating temperatures
are around 77–85 K. Five years ago SCD began to develop a new generation of MWIR detectors based on the epitaxial
growth of antimonide based compound semiconductors (ABCS). This ABCS technology allows band-gap engineering of the
detection material which enables higher operating temperatures and multi-spectral detection. This year SCD presented its
first prototype FPA from this program, an InAlSb based detector operating at a temperature of 100 K. By the end of this year
SCD will introduce the first prototype MWIR detector with a 640×512 element format and a pitch of 15 µm. For the LWIR
wavelength range SCD manufactures both linear Hg1–xCdxTe (MCT) detectors with a line of 250 elements and time delay
and integration (TDI) detectors with formats of 288×4 and 480×6. Recently, SCD has demonstrated its first prototype
uncooled detector which is based on VOx technology and which has a format of 384×288 elements, a pitch of 25 µm, and a
typical NETD of 50 mK at F/1. In this paper, we describe the present technologies and products of SCD and the future evolu-
tion of our detectors for the MWIR and LWIR detection.

Keywords: digital detector, 480×384 element detector, 640×512 element detector, focal plane array, MCT, IR detector,
InSb, InAlSb, superlattice, TDI, DDC.

1. Introduction up to a very long TDI DDC for airborne applications with

push-broom imaging, with 2048×16 elements [3].
Second generation infrared (IR) detectors at SCD are based SCD's 2D InSb FPAs have been in production since
on InSb for the mid-wavelength IR (MWIR) atmospheric 1997 with the 320´256 element format and since 2000 in
window (3–5 ìm), and on mercury cadmium telluride the larger format of 640´512 elements. SCD also special-
(MCT) for the long-wavelength IR (LWIR) window (8–12 izes in solutions optimized for various types of application
ìm) [1]. In the past 27 years, SCD has developed and man- from hand held cameras up to missile warning systems
ufactured ten types of infrared detector, both with support (MWS). The various solutions are based on the special de-
from the Israeli MOD and in cooperation with Israeli insti- sign of Dewars which can endure a large range of environ-
tutions and companies such as the Technion, Soreq NRC, mental conditions and on the many operational modes of
RICOR and RAFAEL. SCD's current production line in- the signal processor which supports all the applications.
cludes MCT devices with up to 4806 elements operating in The general requirements of the third generation of IR
time delay and integration (TDI) mode and two dimen- detectors are:
sional (2D) InSb focal plane arrays (FPAs) with up to • high resolution (larger format and smaller pitch),
640´512 elements, all available in various configurations • advanced integrated readouts (digital, low noise),
including fully integrated detector-dewar-cooler (DDC) • higher operating temperatures (> 77 K),
packages. Many DDC configurations were developed, in • high spatial uniformity,
most of the cases to custom design; they range from very • high temporal stability,
small low power and weight DDCs such as “Piccolo” [2] • multi-spectral detection.
In order to be able to meet the third generation require-
ments, SCD has started two main programs about five years
*e-mail: onesher@scd.co.il
ago which are both based on the mature technology of pla-
The paper presented there appears in Infrared Photoelectronics, nar InSb 2D arrays. The first one is focused on the signal
edited by Antoni Rogalski, Eustace L. Dereniak, Fiodor F. Sizov, processor which is bonded directly to the detection material,
Proc. SPIE Vol. 5957, 59570S (2005).

Opto-Electron. Rev., 14, no. 1, 2006 O. Nesher

in this program we developed a family of digital readout in- LWIR detection. First, the basic concept and the products
tegrated circuits (ROIC) with a 20-µm pitch which is called of the Sebastian range of detectors will be described. Then,
Sebastian [4,5]. The first product of this program has a for- the technology and characterization results of the ABCS
mat of 640×512 and it was first introduced in 2003. Since program will be presented and future plans will be de-
then many detectors were produced and have been inte- scribed.
grated successfully into systems. The second product con-
sists of 480×384 elements and since 2004 it has been manu- 2. SCD MWIR FPA roadmap
factured in SCD's production line [6].
The second program is focused on the detection mate- Two dimensional focal plane arrays based on InSb
rial and more specifically antimonide based compound photodiodes bonded to a CMOS readout circuit have been
semiconductor (ABCS) materials which are the basic tech- produced in SCD since 1997. The basic InSb technology
nology for future detectors at SCD [7–9]. This technology consists of planar ion implanted photodiodes, hybridized to
which uses epitaxial thin film growth in SCD’s MBE ma- a Si focal plane processor (FPP) by means of indium
chine enables band gap engineering of the detection mate- bumps, and then backside thinned with surface passivation
rial and the design and optimization of various structures and anti-reflection coating. The roadmap of the two dimen-
for high operating temperature, a small pitch size and sional FPAs is described in Fig. 1. Three FPAs for the mid
multi-colour detection. The first product of this program is format with 30-µm pitch were developed, two of them are
based on an epitaxial InSb film grown on top of an InSb based on SCD signal processors (Gemini and Blue Fairy
substrate, enabling an operating temperature above 90 K [10]) and one is based on an Indigo signal processor. Since
with the same performance as achieved with implanted pla- 2000 SCD has been producing larger format FPAs with
nar InSb at 80 K. The second product, based on InAlSb 640×512 elements and smaller pitch, in which the first two
film with a 4.8 micron cut-off, was first introduced this FPAs are based on Indigo ROICs with 25- or 20-µm
year at the Orlando SPIE conference, operating inside a pitches. Since 2003 SCD has been producing the Sebastian
camera at a temperature of 100 K. detector which is based on SCD's digital signal processor
This year SCD is introducing the first detector with a which has a 20-µm pitch. This detector will be described in
15-µm pitch and a format of 640×512. SCD's future prod- more detail in the following section. In order to improve
ucts will integrate all these technologies together in the the resolution of the mid format detectors (320×256, 30-µm
same detector, i.e., digital signal processor bonded to pitch) SCD has been developing two types of detector
antimonide array. which have exactly the same active area as the mid format.
In this paper, we describe the technology and the prod- The first one belongs to the Sebastian family and has been
ucts in the present and future both for MWIR and for produced since 2004. It has a format of 480×384 elements

Fig. 1. SCD MWIR 2-D FPA roadmap.

60 Opto-Electron. Rev., 14, no. 1, 2006 © 2006 COSiW SEP, Warsaw

with a 20-µm pitch. The second one has a format of on the FPA. The technology is based on liquid phase
640×512 elements with a 15-µm pitch. Prototypes of these epitaxial (LPE) growth of a MCT layer on a CdZnTe
detectors which are based on the Indigo signal processor (CZT) substrate and implantation of photodiodes into the
ISC0402 will be available this year, and a detector which is epitaxial MCT layer. The first product of this program is a
based on SCD's digital signal processor will be available linear array of 256×1 elements which consists of two rows
next year. Based on the 15-µm technology, SCD will in the of photovoltaic elements for redundancy. The first proto-
future develop detectors with larger formats, such as type of this detector was supplied in 1996 and during the
1000×1000. In parallel to all these activities, SCD is devel- current decade this detector has been produced in high vol-
oping the new technology of the ABCS program which will ume with about 5,000 detectors in the field. On the basis of
be described in detail below This year we introduced the the photovoltaic MCT technology a time delay integration
first prototypes of InAlSb (with 4.8-µm cut-off) bonded to (TDI) detector was developed at SCD, this detector consist-
a Blue Fairy signal processor operating at a temperature of ing of 480×6 elements which enables the operation of
100 K and integrated inside a hand held camera. By the end 480×4 at the system level (the four good diodes are chosen
of the year, prototypes of InAlSb bonded to Sebastian 480 out of six) with improved signal to noise for each channel.
will be available. In the near future (2006), a detector with The 480×6 element TDI detector was integrated success-
an operation temperature above 130 K will be developed, fully into a scanning system demonstrating a high level of
and this detector will have a cut-off of 4.2 µm. On the basis performance with high image resolution on the system
of these technologies a bi/multi-colour detector can be de- level. During 2004 SCD finished the development of a
signed for the 3–5 µm regime. For the long term, based on smaller format TDI detector with 288×4 elements.
the technology of MBE epitaxial thin films and ABCS, During 2002, SCD started to develop a micro-bolo-
InAs/InGaSb superlattice structures can be designed for meter detector which is based on VOx technology. Re-
any wavelength of operation starting from 3 µm, including cently, SCD has demonstrated its first prototype of an
2 dimensional arrays for the MWIR and LWIR regimes. uncooled detector which is based on VOx technology. It is
based on a format of 384×288 elements, with a pitch of 25
3. SCD LWIR FPA roadmap µm and it exhibits a typical NETD of 50 mK at F/1 [11]. In
addition to its high level of radiometric performance, this
During the 90's, SCD started the development of backside detector has two unique features, power-save mode for low
illuminated detectors which are based on Hg1–xCdxTe power consumption and on-line signal drift compensation
(MCT) for detection in the 8–12 µm regime. These MCT due to ambient temperature changes, which reduces the
detectors are flip-chip bonded to a signal processor located need for frequent one point corrections.

Fig. 2. SCD LWIR FPA roadmap.

Opto-Electron. Rev., 14, no. 1, 2006 O. Nesher 61

 Since 2003 SCD has been producing detectors which the digital detector Dewar cooler (D3C) with its proximity
are based on quantum well infrared photo-detector (QWIP) card is shown in Fig. 3. A special proximity electronics
structures for LWIR detection, according to customer de- board was developed for this detector, which consists of a
mands. SCD buys the QWIP chip from suppliers, bonds it FPGA with a local oscillator. The basic arrangement is
to a signal processor and integrates in into a Dewar. Based shown in Fig. 3, where the FPGA operates the DFPP di-
on the ABCS program, SCD is planning to implement fu- rectly, collects the data from the DFPP, formats it and
ture LWIR two dimension arrays with a superlattice made sends it out to the system. All other system operation
of periodic InAs/InGaSb layers. This device enables the de- modes are controlled by the system via the communication
sign of detectors which are sensitive to wavelengths from channel including timing of operation. The interface be-
3 µm up to > 14 µm. tween the FPGA and the system can either be standard (e.g.
camera link), or specific as per system requirements. This
4. SCD digital detectors concept of a simple interface between the D3C and the sys-
tem leads to easy and fast integration of the D3C into the
During 2003 SCD completed the development of the first system.
detector which is based on a digital signal processor on the
focal plane array (FPA) itself. The main challenge in de-
signing a high performance signal processor for a cooled
IR detector with digital output was to maintain power con-
sumption similar to that in an analogue processor. Predic-
tions showed that the conventional design for analogue to
digital conversion (ADC) results in power consumption
over 1 W under the operation conditions of a standard IR
detector. However, the special design at SCD of the ADC
and the whole signal processor has resulted in a power con-
sumption of the digital signal processor which is even
lower compared to existing analogue processors.
Detectors based on a digital FPA are considered to be Fig. 3. Block diagram of system-proxy-detector.
very attractive due to their many advantages over detectors
with an analogue FPA, which are expressed especially on 4.2. D3C performance
the system level. These include:
• lower level of readout noise due to immunity of the an- The special design of the signal processor yields an excel-
alogue signal to external noise, lent linearity of less than 0.01% deviation/full range over a
• higher linearity, regime that starts at 2% and continues to 90% well-fill ca-
• less sensitivity to external ambient conditions, pacity. A direct outcome of this high linearity is a low
• higher long term stability of the residual non uniformity RNU which is less than 0.015% std/DR for a range of
(RNU), 2–90% well fill capacity. Figure 4 shows the linear rela-
• removal of the requirement to develop low noise elec- tionship between the squared measured noise and the signal
tronics in the system, which testifies to the clean sampling of the signal inside the
• distance between the detector and the system electron- FPA due to the onboard A/D conversion in the FPA.
ics can be increased up to several meters without per-
The low level of the total noise achieved with SCD's
formance degradations,
digital detector enables the attainment of a very good sys-
• integration of the detector into the system is much sim- tem NETD of 10 mK for 50% well fill capacity. Com-
pler and faster.
paring the spatial and the temporal noise a similar value of
Over the past two years we have presented at the spring-
10 mK is achieved in SCD's Digital detectors. The diver-
time SPIE conferences two types of our digital detector
sity of modes together with the direct control of the FPGA
(Sebastian). The first one with a format of 640×512 in 2003
on the detector enables various operational modes which
and the second one with the format of 480×384 in 2004. All
are not available for analogue detectors. For example, the
the above advantages of digital detectors were demonstrated
detector can be operated with different integration times for
by the performance measured on the Sebastian detectors.
each frame (elaboration of combined mode) where the
In this section we present the general structure of the
length and timing of the integration can be changed from
signal processor and its main features, together with some
frame to frame regardless of the previous frame. The detec-
typical performance results that were measured.
tor can also be operated with multiple integration pulses in
4.1. D3C basic structure the same frame with a pixel saturation level control, this
mode enables high dynamic range together with high frame
The digital focal plane processor (DFPP) is fabricated with rate detector operation. A TDI mode of operation is also
a 0.5 µm double-poly, triple metal CMOS process and it available in the detector with control of the TDI depth to
consists of 4.5 million transistors. The basic appearance of allow operation at different speeds.

62 Opto-Electron. Rev., 14, no. 1, 2006 © 2006 COSiW SEP, Warsaw

 40×20 or 40×40 µm. This mode is very useful while trying
to detect a sub pixel target where the image is smeared over
four pixels (due to the diffraction limitation of the optics)
and the signal is very weak. By applying the four pixels
merging function the signal to noise improves significantly,
compared to the case where an external addition of the data
is done at the digital level of the image processing. In the
following table there is a comparison between the main
features of the two Sebastian detectors and Blue Fairy.
All these features combined with a high level of perfor-
mance at the system level, make the Blue Fairy and
Sebastian detectors the ideal solution for missile warning
systems (MWS) [12].

5. Antimonide based detectors

5.1. Introduction
Fig. 4. Squared noise versus signal measurement.
Going beyond the systems described above, SCD's 3rd gen-
eration programme is based on epitaxial diode materials.
4.3. Main features These materials have the potential for higher operating
temperatures and multi-band detection. They belong to the
In Sebastian, the conversion resolution is controlled exter- antimonide based compound semiconductor (ABCS) fam-
nally and can be changed continuously from 12–15 bits. ily of III-V materials and are based on InAlSb and InAsSb
There is a trade off between the conversion resolution and alloys and InAs/InGaSb superlattices. To cover the MWIR
the maximum frame rate, such that a higher frame rate can atmospheric window, we recently proposed [7] the
be achieved with lower resolution. An anti-blooming cir- epitaxial alloys: InAs1–y Sby on GaSb with 0.07 < y < 0.11
cuit was implemented at the input stage of the signal pro- and In1–zAlz Sb on InSb with 0 < z < 0.03. These can be
cessor to avoid a very strong light source from disrupting used together with superlattices based on thin layers of
its operation. All the detectors contain a correlated double InAs and InGaSb that provide the basis for detector materi-
sampling (CDS) mechanism inside the signal processor, als operating also in the 8–12 µm long wavelength infra-red
where the CDS data is read outside and is subtracted from (LWIR) atmospheric window as well as in the MWIR [13].
the video data. The use of CDS during operation was found For the alloys, there can be a small lattice mismatch be-
to be very useful for low frequency noise reduction and for tween the alloy and the substrate material, and this can lead
NUC stability enlargement. The pixel binning mode is a to the generation of dislocations inside the alloy. It is there-
connection of every two or four adjacent pixels together, fore necessary to find a strategy to suppress the effect of
which is done inside the signal processor. This feature en- the dislocations so that they are not harmful to detector op-
ables operation with an effective pixel size of 20×40, eration. On the other hand superlattices, in principle, offer

Table 1. The main features of the Blue Fairy and Sebastian detectors.
Feature Blue Fairy Sebastian 480 Sebastian 640
Format 320×256 480×384 640×512
–
Full range (gain) Me 3.5/7/11/15/22/30 1/3/7/10/14 3/7/10/14
Frame rate@full frame > 450 Hz >160Hz@15bit >120Hz@15bit
>240Hz@13bit >160Hz@13bit
>280Hz@12bit >180Hz@12bit
Main integration modes ITR/IWR/combined ITR/IWR/combined/multistep/multiple
Readout dilution —— every 2nd and 6th row every 2nd row
Pixel binning —— 1×2, 2×1, 2×2
Linearity <0.05%@2–90% WF <0.01%@2–90%WF
RNU std/dr <0.025%@2–90%WF <0.015%@2–90%WF
InSb bias operating point 500 pA – 1 µA 70 pA – 100 nA 70 pA – 100 nA
Windowing every 2 rows/16 columns every 2 rows every 4 rows

Opto-Electron. Rev., 14, no. 1, 2006 O. Nesher 63

the advantage of full wavelength tunability without the dis- is typically 3–5 µm thick and not intentionally doped. The
advantages of limited device thickness or excessive num- p-side of the junction is typically 1–2 µm thick and is Be
bers of dislocations. doped at ~1018 cm–3. The epitaxially grown junction and
In the initial stages we have focused in particular on mesa geometry provide some significant advantages over
In1–zAlzSb alloys grown on InSb. Using this approach we SCD's current planar implanted diode technology including
have reached in a relatively short time a number of signifi- a higher quality of p-n junction with lower dark current,
cant breakthroughs. These include 320´256 element mesa and suppression of blooming and cross talk. In order to ex-
FPAs from epitaxial InSb with dark currents at 95 K com- ploit this epitaxial material we have developed a new mesa
parable to those of implanted FPAs at 77 K and mesa FPAs technology. The mesas are highly uniform and are prepared
from InAlSb with dark currents up to an order of magni- by either dry or wet chemical etching to depths of 2–6 µm.
tude lower. For epitaxial InSb the operability and residual In the next two sub-sections we present some key per-
non-uniformity (RNU) at 95 K are respectively more than formance parameters of our diodes and FPAs, first for
99.5% and less than 0.03% (standard deviation/dynamic epitaxial InSb and then for epitaxial InAlSb with cut off
range) after a two point non uniformity correction (NUC). wavelengths up to 1 µm shorter than in the case of InSb.
The RNU remains very low even when the operating tem-
perature is allowed to shift quite significantly. This
5.2.1. InSb
so-called “V-curve” stability is a critical performance pa-
rameter for harsh environments. For epitaxial In1–xAlxSb FPAs with 320´256 pixels have been fabricated by com-
the “V-curve” stability increases substantially relative to bining SCD's epitaxial InSb with SCD's Blue Fairy signal
that in epitaxial InSb due to its lower dark current. In these processor. The operability of the FPA is greater than
materials at 90 K we have achieved operabilities greater 99.5%, both at 80 K and 90 K, based on stringent criteria
than 99% in alloys with x = 0.01 and 98.8% for x = 0.03. routinely applied to our production line FPAs which in-
Due to the rapid evolution of performance during the de- volve the removal of both hard defects and of soft non-uni-
velopment phase we expect that further improvements in formity defects. At both of these operating temperatures the
the operability of the alloys will soon be achieved. FPA has a RNU of < 0.03% for scene temperatures be-
In the remaining part of this paper we will present some tween < 20 C and 75 C after a standard two point correc-
of the key technological achievements for In1–xAlxSb based tion has been performed. Excellent quality images have
FPAs, including both binary and ternary epitaxial materi- been obtained which have been presented previously [14].
als. Work on InAsSb alloys and InAs/InGaSb superlattices, The quality remains high even when taken under the fol-
will be reported in a future publication but some further de- lowing operationally demanding conditions: a two point
tails may already be found in Ref. 7. correction is performed once, at an FPA temperature of 80
K and it is then used to correct the raw signal at 90 K. For
5.2. New FPA materials: epitaxial InSb and InAlSb scene temperatures from < 20 C to 75 C (< 20% to 90%
well fill) the RNU remains below 0.1%, even though the
InSb and InAlSb layers that are suitable for fabrication into temperatures of calibration and operation differ by ~10 de-
mesa diodes are grown at SCD by molecular beam epitaxy grees. The dependence of the RNU on FPA temperature is
(MBE) on (100) InSb substrates, using a Veeco GEN III demonstrated explicitly in Fig. 5, which shows a “V-curve”
MBE machine. The n-type active photon absorbing region of the temperature dependence of the maximum RNU (usu-
ally close to 50% well fill) with an integration time of 1.6
mS at F/number 2.5, after a 2-point NUC (~30 and ~60%
well fill) is performed at an FPA temperature of 95 K. It
shows how the RNU remains below 0.1% for deviations
from the 95 K calibration temperature of up to about –10 K
and +5 K. The curve demonstrates clearly the very high
stability of the FPA performance, even in environments
where the FPA temperature may fluctuate strongly. This
gives a serious advantage, for example in missile applica-
tions, where the good RNU stability is a critical issue.
Figure 6 shows the distribution of the dark currents in
the FPA pixels at 90 K. It may be seen that the distribution
is very narrow with a peak value of 4.2 pA and a full width
at half maximum (FWHM) of 0.4 pA. The high stability of
the RNU at 95 K against temperature variations of ~10 K is
a direct consequence of the high uniformity of the diodes
and the narrowness of their dark current distribution 15.
Fig. 5. “V-curve” showing RNU of FPA vs. FPA temperature for The FPA operating temperature can be increased above
2-point NUC performed at 95 K. 90 K before the dark current grows to a level where it be-

64 Opto-Electron. Rev., 14, no. 1, 2006 © 2006 COSiW SEP, Warsaw

 Fig. 8. Photoresponse spectra for InAlSb diodes at 77 K with lC
from 5.4 µm (InSb) to 4.4 µm.

suppression of the dark current over that possible in high

Fig. 6. Distribution of dark currents at 90 K and at a bias of –168 quality epitaxial InSb diodes.
mV in the pixels of an InSb FPA (peak = 4.2 pA, FWHM = 0.4 pA,
In Fig. 8, we show the photo-response of several
operability 99.6%).
InAlSb diodes where the aluminium composition increases
gins to degrade the performance seriously. For example an from right to left. The decrease in the cut-off wavelength
image taken at 110 K is shown in Fig. 7. Even at this tem- with increasing Al composition may be seen clearly, from
perature, the definition remains high and power lines at a lC ~5.4 µm for InSb to a value of lC ~4.4 µm for the 3%
range of 2 km are still visible. Such performance at high alloy. The steepness of the absorption edge is similar in all
operating temperature provides benefits for example in cases without any significant signs of degradation. Figure 8
hand-held and sighting applications, since it allows signifi- shows the ease with which the bandgap and cut-off wave-
cant power savings to be made, an issue which can some- length of our FPAs can be tuned by varying the aluminium
times be critical. composition.
In the following two subsections we demonstrate the ef-
5.2.2. InAlSb fect of alloying on diode performance. Below we present
results of dark current measurements on single diodes and
The bandgap of In1–xAlxSb is expected to increase at
the collective diode performance for whole FPAs.
roughly 18 meV for every 1% of Al, based on a simple lin-
ear interpolation between the bandgap of InSb and the di-
Single diode properties
rect bandgap of AlSb. This increase provides additional
Figure 9 shows the dark current at a reverse bias of –100
mV as a function of temperature between 77 K and 180 K
for a representative 50´50 µm InAlSb diode with lC ~5.0
µm. In the figure, the logarithm of the current is plotted as a
function of the inverse temperature. The slope for a purely
GR limited diode should be approximately equal to EG/2k,
where EG is the low temperature value of the energy
bandgap of the semiconductor, while for diffusion it should
be nearly equal to EG/k. The two straight lines indicated in
the figure correspond to activation energies of EG ~ 270
meV for the upper line and EG/2 ~140 meV for the lower
line. These values correspond quite well with the expected
bandgap and show clearly that the lower temperature re-
gion with the lower slope is dominated by GR while the
higher temperature region with the steeper slope is domi-
nated by diffusion. The changeover from GR to diffusion
occurs at a temperature of about 130 K.
The low dark current for single diodes at 90 K that may
Fig. 7. Image recorded with the 256´320 pixel InSb mesa-FPA at a be derived from Fig. 9 is a necessary condition for good
focal plane temperature of 110 K. FPA performance. However, as already mentioned above,

Opto-Electron. Rev., 14, no. 1, 2006 O. Nesher 65

 Fig. 10. Distribution of dark currents at 90 K and at a bias of –309
Fig. 9. Arrhenius plot of reverse current at a bias of –100 mV for an mV in the pixels of an InAlSb FPA, lC ~4.4 µm (peak = 0.36 pA,
InAlSb diode with lC ~ 5.0 µm. The straight lines have activation FWHM = 0.03 pA, operability 99.1%).
energies of ~140 meV and ~270 meV, respectively.
cess level. Applying the method described in Ref. 14 to
it is also critical that the distribution of dark currents of all analyse the dark current distribution, we can show that for
of the FPA diodes is as narrow as possible, in order to get a x = 0.03, the In1–xAlxSb alloy composition across the FPA
very low and stable value for the detector RNU, both at 90 has a standard deviation that must lie below 0.0005.
K and above. This in turn requires a high uniformity in the The dependence of the mean FPA dark current at a bias
aluminium composition. We shall demonstrate that a very of approximately –150mV and FPA temperature of 90 K is
uniform composition and a very narrow distribution of dark shown in Fig. 11 as a function of cut-off wavelength (com-
currents can be achieved. position range: 0 < x < 0.03). The results are compared
with the theoretical expectation for a constant number of
GR centres, NGR: I d = aN GR exp( -hc 2l C kT ) where a is a
FPA properties constant of proportionality, and the agreement is found to
Many 320´256 InAlSb FPAs have been made using SCD's be quite good. This shows that increasing the aluminium
Blue Fairy silicon processor. For a cut-off of lC ~5 µm typ- composition does not significantly increase the number of
ical dark currents at 90 K are 1.5–2.5 pA. Operabilities of GR centres. The dark current thus decreases essentially as
99% or better at 90 K can be achieved, based on the same expected from a simple increase in the bandgap. The tech-
selection criteria as described above for InSb. In these nological implications of this behaviour are significant,
cases, the full width at half maximum (FWHM) of the dark since it provides the basis for two colour detectors made
current is typically about 200–300 fA. This is less than the with stacked diodes, each with a different composition.
FWHM reported above for epitaxial InSb and is well
within the range required to provide excellent temperature
stability of the FPA, as already discussed in Sect. 5.2.1.
When the cut-off wavelength is further reduced to lC ~4.4
µm, the dark current is lowered to ~250 fA and the FWHM
to ~30 fA. This is demonstrated in Fig. 10. In fact, the most
striking feature of the dark current distributions in our
InAlSb FPAs is that their relative widths are essentially
constant for the full range of cut-off wavelength up to lC
~4.4 µm (0 < x < 0.03), that is 1–13% of their peak dark
current values.
From a technological point of view, the satisfying out-
come of the tight dark current distributions described above Fig. 11. Dependence of mean dark current at a bias of about –150
is that by going to a more complex ternary material it is mV and at an FPA temperature of 90 K on cut-off wavelength (0 < x
possible to gain the benefits of bandgap tunability with al- < 0.03). Points are mean dark current values. Line is fit to GR
most no penalty in uniformity, both at the material and pro- formula for a constant number of GR centres.

66 Opto-Electron. Rev., 14, no. 1, 2006 © 2006 COSiW SEP, Warsaw

 After a two-point NUC, the RNU of an InAlSb FPA
with lC ~5 µm is typically less than 0.01% between 26%
and 75% well fill. In Fig. 12, we compare the V-curve for
the InSb FPA of Fig. 5 with a V-curve for such an InAlSb
FPA. The conditions are slightly different in that the inte-
gration time has been increased to 4 mS without changing
the well-fill. This naturally narrows the width of the
V-curve (due to the lower photocurrent conditions) but is
sufficient for a straight comparison between the two FPA
compositions. It can be seen that the V-curve for the InAlSb
FPA has nearly doubled the width of that for the InSb FPA.
This can be explained in terms of the lower absolute width
of the dark current distribution in the alloy FPA. In Fig. 13,
we present an image measured with an InAlSb FPA with
lC ~5 µm operated at 100 K, in order to demonstrate the
high quality of the image that can be produced. Fig. 12. V-curves for the FPA of Fig. 5 and an FPA made from
In0.99Al0.01Sb. The integration time is 2.5 times longer than in Fig. 5,
6. Conclusions resulting in a narrower opening of the “V”.

Today semi-conductor devices (SCD) has a high produc-

tion capability of over 5800 IR detectors per year. Pres-
ently, SCD products are based on both InSb technology for
MWIR two dimensional array detectors and mercury cad-
mium telluride technology for LWIR time delay and inte-
gration detectors. All these technologies are provided by a
full suite of in-house facilities such as: crystal growth, full
chip processing including flip-chip bonding, VLSI design,
FPA integration within the Dewar with either Stirling or JT
cooling, and full detector characterization. This complete
capability at each stage of production allows total optimi-
zation of the integrated detector to achieve the best perfor-
mance. Large numbers of SCD IR detectors are used in a
variety of applications such as high quality imagers, hand
held cameras, seeker heads, targeting pods, and air recon-
naissance. SCD's revolutionary high performance digital
detector, the only one of its kind, is spearheading a new
Fig. 13. Image produced by InAlSb FPA at focal plane temperature
family of detectors (640´512, 480´384, and 320´256)
of 100 K.
which can be integrated very quickly into the system, and
which provide excellent performance at the system level
including: low noise, low residual non uniformity (RNU) Acknowledgements
with long term stability, and high linearity of response. In
this paper, we have presented details of our new ABCS The authors, first and foremost, would like to acknowledge
mesa FPAs, based on epitaxial InSb and InAlSb layers the many technicians, engineers and scientists of SCD without
grown in-house by MBE. We have been able to demon- whose long and dedicated work the products and research de-
strate a very high degree of alloy uniformity, with a stan- scribed in this review would never have been realized. We
dard deviation in composition of sz £ 0.0005 for In1–zAlzSb gratefully acknowledge the close and continuing collaboration
with a wavelength cut-off close to 5 µm. We have demon- with the Soreq NRC and also their assistance with the
strated a radiometric performance and stability in InSb photoluminescence measurements presented in this article.
FPAs at 95 K comparable to that previously only achiev- We also thank the Israeli MOD for their support.
able at 77 K, and we have presented high quality images
for FPAs made from both binary and alloy materials oper- References
ating up to 110 K. We are confident that by connecting the
FPAs developed within SCD's ABCS program to SCD's re- 1. O. Nesher, P.C. Klipstein, and E. Weiss, “Advanced IR de-
cently developed Digital FPP, we shall be able to provide tector design at SCD: from D3CTM to ABCS”, Proc. SPIE
our customers with maximum performance advantage, both 5359, 101–110 (2004).
at the level of the diode material and at the level of the pro- 2. T. Markovitz, F. Schapiro, D. Alfiya, S. Hasson, A. Magner
cessor. and O. Nesher, “Piccolo – a high performance IR detector

Opto-Electron. Rev., 14, no. 1, 2006 O. Nesher 67

 optimized for handheld applications”, Proc. SPIE 5406, Weiss, and S. Risemberg, “3rd generation infrared detectors
239–248 (2004). at SCD: InAlSb focal plane arrays”, Proc. SPIE 5406,
3. M. Zucker, I. Pivnik, E. Malkinson, J. Haski, T. Reiner, D. 222–229 (2004).
Admon, M. Keinan, M. Yassen, I. Sapiro, L. Bykov, N. 10. O. Nesher, S. Elkind, A. Adin, U. Palty, O. Pelleg, E.
Sapir, and A. Fraenkel, “Long, mid-wave infrared detector Jacobsohn, T. Markovitz, I. Szafranek, Z. Calahorra, and J.
with time delayed integration”, Proc. SPIE 4820, 580–592 Oiknine-Schlesinger, “Performance of BF focal plane array
(2002). 320×256 InSb detectors”, Proc. SPIE 4820, 699–707
4. S. Elkind, A. Adin, I. Nevo, and A.B. Marhasev, “Focal (2002).
plane processor with a digital video output for InSb detec- 11. U. Mizrahi, A. Fraenkel, L. Bykov, A. Giladi, A. Adin, E.
tors”, Proc. SPIE 4820, 751–757 (2002). Ilan, N. Shiloah, E. Malkinson, Y. Zabar, D. Seter, R.
5. O. Nesher, S. Elkind, A. Adin, I. Nevo, A.B. Yaakov, S. Nakash, and Z. Kopolovich, “Uncooled detector develop-
Raichshtain, A.B. Marhasev, A. Magner, M. Katz, T. Mar- ment program at SCD”, Proc. SPIE 5783, 551–558 (2005).
kovitz, D. Chen, M. Kenan A. Ganany, J.O. Schlesinger, 12. O. Nesher, S. Elkind, and T. Markovitz, “Advantages of
and Z. Calahorra, “A digital cooled InSb detector for IR de- SCD detectors for missile warning system”, Proc. SPIE
tection”, Proc. SPIE 5074, 120–129 (2003). 5612, 32–41 (2004).
6. O. Nesher, S. Elkind, I. Nevo, T. Markovitz, A. Ganany, 13. A. Fuchs, L. Bürkle, R. Hamid, N. Herres, W. Pletschen,
A.B. Marhashev, and M. Ben-Ezra, “480×384 element InSb R.E. Sah, R. Kiefer, and J. Schmitz, “Optoelectronic proper-
detector with digital processor”, Proc. SPIE 5406, 214–221 ties of photodiodes for the mid- and far-infrared based on
(2004). the InAs/GaSb/AlSb materials family”, Proc. SPIE 4288,
7. P. Klipstein, E. Jacobson, O. Klin, M. Yassen, Z. Calahora, 171–182 (2001).
E. Weiss, S. Risemberg, and D. Rosenfeld, “Antimonide 14. P. Klipstien, Z. Calahora, A. Zemel, R. Gat, E. Harush, E.
based materials for infrared detection” Proc. SPIE 4820, Jacobson, O. Klin, M. Yassen, J. Oiknine-Schlesinger, and
653–662 (2002). E. Weiss, “Third-generation infrared detector program at
8. P. Klipstien, Z. Calahora, A. Zemel, R. Gat, E. Harush, E. SCD: InAlSb focal plane arrays”, Proc. SPIE 5612, 42–50
Jacobson, O. Klin, M. Yassen, J. Oiknine-Schlesinger, E. (2004).
Weiss, and S. Risemberg, “Third-generation infrared detec- 15. L. Shkedy, O. Amir, Z. Calahorra, J. Oiknine-Schlesinger,
tor program at SCD”, Proc. SPIE 5406, 222–229 (2004). and I. Szafranek, “Temperature dependence of spatial noise
9. P. Klipstien, Z. Calahora, A. Zemel, R. Gat, E. Harush, E. in InSb focal plane arrays”, Proc. SPIE 4028, 481–488
Jacobson, O. Klin, M. Yassen, J. Oiknine-Schlesinger, E. (2000).

68 Opto-Electron. Rev., 14, no. 1, 2006 © 2006 COSiW SEP, Warsaw