The JR detector industry is sharply segregated into a very large military sector and a very small commercial sector. The sectors are separated by significant differences in cost, technical complexity, quality, and underlying technology. In response to dramatic changes in the worldwide politico—military environment, there is increased interest in commercial diversification. The challenge facing the military—dependent sector of the JR detector industry is less one of diversifying into the commercial market than of creating a sustainable "dual use" market capable of utilizing advanced JR technology. Industry appears to be groping to identify solid commercial applications for its more advanced technologies. Government representatives are no doubt anxious about potential erosion of the industrialbase. The authors suggest that some serious thought be given to innovative intra-industry or government-industry activities focused on lowering cost and providing useful information about dual use applications.
Infrared focal plane array (IRFPA) specifications can drive IRFPA cost and yield. Specifically, IRFPA operability affects IRFPA producibility in two ways, each having an opposite impact. Operability improvements resulting from purer, more perfect materials (i.e., fewer defects) become increasingly difficult to achieve as the operability approaches 100%. Thus, yielded IRFPA material costs increase with an increasing operability specification. On the other hand, IRFPAs with higher operability have a greater probability of meeting system specifications for defective pixels, and thus fmal test yields are better. This effect tends to reduce IRFPA cost as operability increases. This paper reports on efforts to quantify the latter effect. The defect specification is examined through the development and exercise of two computer models. The first is a Monte Carlo simulation to predict the probability that an IRFPA will meet system global and local defective pixel specifications as a function of operability. The second is a manufacturing cost model exercised to determine the cost impact of the "pixel yield." Model output is presented for two IRFPA configurations and for various defect specifications. For each case, relationships are determined for the probability of passing the defect specification versus operability and for (normalized) cost versus operability. The analysis shows that pixel defect specifications can become severe cost drivers for larger arrays. However, even slight relaxation of specifications can be very beneficial in increasing array producibility and reducing IRFPA cost.
Liquid phase epitaxy(LPE) of (Hg,Cd)Te thick films from large tellurium-rich solutions was used to produce very large quantities of high quality films used for the fabrication of focal plane arrays. The composition uniformity is found to be ix=±O.OO1 for x=O.223 over an area ranging from 43cm2 to 54cm2. The reproducibility for 2OO growth runs, <8000cm2, is x=0.226±0.0033 or equivalently Xco9.8Si44Pm. The electrical quality of the material for diodes is evaluated by growing n-type films with an indium back-doping of —5x1014cm3. The carrier concentratio:ri is reproducible and in good agreement with the indium doping level. The average carrier concentration at 77K is 5.3±2.4x10'4cm3 with a mobility of 13 1±0.31x105cm2/V sec. Th.e minority carrier lifetime for the n-type films is Auger limited with an average value of 68±2.0isec.
In situ monitoring provides numerous advantages in the fabrication of HgCdTe-based infrared devices. Two in situ monitoring techniques are currently being investigated in our laboratory: optical emission spectroscopy (OES) and ellipsometry. OES is ideal for end point detection, for monitoring reactor integrity, and it also provides chemical information. Ellipsometry is a technique useful for the determination of film thickness and surface roughness. Process control can be readily achieved through the implementation of these two in situ measurement techniques. Examples of the use of OES for end point detection during the plasma etching of }fgCdTe and ZnS are discussed. In situ ellips.ometry is being pursued for monitoring the mild plasma etching of bromine/ methanol polished HgCdTe surfaces prior to in situ passivation and insulator deposition. To support the utility of in si.u ellipsoriietry, our initial studies using ex situ ellipsoinetry measurements of plasma etched HgCdTe are highlighted. The advantages of in situ monitoring for multistep vacuum processing, including contamination reduction and improved process control, are presented.
Liquid phase epitaxy (LPE) is the preferred technique for growth of high quality (Hg,Cd)Te films for focal plane array applications. Successful implementation of this process in a manufacturing environment requires development of advanced sensors and process controls to increase yields and reduce costs. This paper will review progress on Ti's Intelligent Processing of Focal Plane Arrays program*. (Hg,Cd)Te LPE films are grown on (111)B-oriented (Cd,Zn)Te substrates in vertical dipping reactors in large (4500 g) molten solutions of mercury and cadmium in tellurium. Multiple (up to 8) substrates, for a total area of 54 cm2, can be run in a single growth cycle. The sensors and controls under development for this process include: (1) eddy current analysis (ECA) for detection of the liquidus temperature of the LPE growth solution, (2) electron beam microprobe/wavelength dispersive x-ray analysis (WDX) for rapid measurement of the film composition immediately after growth, (3) UV/visible optical absorption spectroscopy for determination of the mercury partial pressure over the growth solution, and (4) an RTD based precision temperature control system capable of improving capability from +/0.1 to +I- 0.02 C. ECA was found to be sensitive to the onset of crystallization on cooling in sample LPE melts. The WDX technique was found to be a rapid and accurate method for in-process film composition measurements. The impact of these IPFPA sensors and controls on total yield and producibility is discussed.
Demonstrated detectivity of uncooled infrared focal plane arrays is better than O.1°K noise equivalent temperature difference, Uncooled performance levels are now suitable for military and commercial applications. The manufacturing science of this technology lags that for mercury cadmium tellurium based cooled focal planes. The manufacturing issues and approaches for ferroelectric material based uncooled detectors are discussed. The recommended approach is to address producibility issues based on pareto analysis of the process flow and to address the most highly leveraged process steps in terms of time and labor input. Detailed modeling of the process flow has been undertaken as a part of this analysis. Processes addressed in detail are lap and polish of the detector, photolithography, metallization, etches, hybridization, and test. Several approaches are shown to yield comparable cost focal planes, but a process technology leveraging silicon process technologies is shown to minimize space and equipment.
The SOFRADIR technology is in fact the result of years of studies in the field of CMT photovoltaic detectors and CCD readout circuits on silicon. The resulting components are so good that it has been possible to put them in production. Very good yield have been obtained and now it is necessary to go deeper into the production analysis as the question is no longer basic productibility. Therefore the target of this paper is toshow the very detailed approach we have taken in order to obtain the best cost in production.
This paper reports on continuing improvements in the performance and producibility of InSb focal plane arrays, which together with progress in self-contained dewar/cooler designs have resulted in a new level of portable infrared camera functional capability. A new camera will also take advantage of the latest advances in integrated circuit miniaturization and electronic package fabrication to achieve the lowest possible size, weight, and power consumption while meeting its design goals for ruggedness and reliability.
The various types of the Focal Plane Array JR detectors presently in development or under production are characterized by their relatively large size compared with the traditional discrete element detectors. Most of these FPAs contain "on-board" signal processors allowing the number of electrical leads to the cooled detectors plane to be independent of the number of pixels in the array. In general, the total heat load of such detectors is 1.5 -2.0 times larger than an equivalent discrete element type detector assembly. On the other hand, the size of these FPA's dictates the thermal mass of the detector array itself, and in particular when the whole detector/cooled radiation shield/cold optics is considered to be of an order of magnitude larger than an equivalent discrete elements detector assembly. This fact causes the cooldown time requirement to be the lead factor when cooling power is considered, rather than the steady state heat load. In order to achieve miniaturization, low input power and fast cooldown time, a combined team of Cincinnati Electronics, Mason, Ohio and RICOR Ltd., En Harod, Israel are working on the development phase of an integral cooler/dewar assembly - IDCA, specially tailored for FPA type detectors. The design concepts, technical parameters and some experimental test results are presented.
Cincinnati Electronics has developed a high resolution, high sensitivity, commercial infrared camera, the IRC-160ST. This camera incorporates a 160 X 120 element Indium Antimonide (InSb) multiplexed focal plane array (FPA) integrated with a miniature closed cycle Stirling cooler. The resulting product is small enough for handheld applications and can be operated from an AC power source or battery. Highlights of this particular design include a high reliability integrated dewar/cooler assembly, compact electronic viewfinder, image correction electronics, display controls, and low power consumption. Size and power reductions resulted from incorporating an integrated dewar/cooler assembly (IDCA). Thermal performance issues relevant to the packaging of the IDCA are presented. Performance characteristics of the FPA/electronics such as uniformity, NEDT, and stability of corrections also are presented.
The common module family of long wavelength infrared (LWIR) forward looking infrared (FLIR) components, developed in 1974, established the feasibility of building custom infrared (IR) systems from a small family of critical components. Since 1974, Texas Instruments (TI) has produced over 35,000 of 80,000 detector dewar assemblies in 60-, 120-, and 180-channel configurations delivered to various U.S. and foreign government agencies. TI's success has resulted in a joint DoD and industry initiative to develop a family of second generation multi- mission LWIR detector dewar assemblies for the 1990s and beyond. This paper defines the dewar related second generation detector and system requirements, categorizes the requirements into specific dewar classes, and shows TI's approach to achieving cost effective results.
Among the techniques of fabrication focal plane arrays (FPA), hybridizing a multiplexer (MUX) on a detector is of growing interest. Some of the advantages include high pixel count with the least amount of wire bonding and higher operating speed. This paper describes the various FPA available, their advantages and disadvantages, with special emphasis on hybridizing equipment availability and requirements.
To meet the needs of military and commercial markets, the infrared focal plane array industry must develop new, effective and low cost methods of fabricating and testing imaging detectors. This paper describes Texas Instruments new concepts in automated testing and cold probe technology as they apply to volume production.
This paper presents a method for measuring the modulation transfer function (MTF) of a detector array from zero frequency to twice the Nyquist frequency. The equipment is simple and requires no complex optical components. Also, the use of laser speckle circumvents the problems inherent with traditional methods of MTF testing where the phase of the test target with respect to the sampling grid affects the observed contrast. The MTF measured with this method is compared to the MTF measured using sine targets. The results of the two methods agree to within 2%.
Here, we report on our studies of ballistic electron transport across metal layers and metal/semiconductor interfaces using ballistic-electron-emission microscopy (BEEM). This new technique, which uses a scanning tunneling microscope to inject electrons with a controlled energy into a thin metal film, allows measurements (with spatial resolution approaching 1 nm) of (1) the local Schottky barrier height, (2) ballistic mean free paths of energetic electrons (or holes), and (3) transmission probability of hot carriers across the metal/semiconductor interface. We have measured the attenuation length of hot electrons (1.5 eV above the Fermi level) in PtSi to be approximately 4 nm. This should be compared with an attenuation length of 13 nm for similar energy electrons in Au layers. BEEM images of the Au/Si interface show features on a very small length scale suggesting that the inelastic mean free path of electrons in Au is close to the attenuation length. The SB height (as determined by BEEM) is 0.87 eV in good agreement with optical measurements. We have also used BEEM to observe the sharp onset of inelastic scattering mechanisms in Au/Si and in PtSi/Si. It is our belief that these studies of ballistic carrier transport will allow a fundamental determination of how to achieve higher quantum efficiencies in SBIRDs.
We discuss in detail the producibility issues associated with GaAs/AlxGa1-xAs quantum well infrared photodetectors (QWIPs). Excellent uniformity in growth (thickness, doping, and Al concentration) and in processing are expected to lead to high yield, high performance large area infrared imaging arrays.
HgCdTe was grown on Si substrates containing CCD and CMOS readout (R/O) circuits. Evaporated aluminum (Al) thin films were used to interconnect MWIR HgCdTe detector arrays with 1 X 64 scanned R/Os to demonstrate monolithic integration and eliminate indium bump bonds required to fabricate hybrid infrared focal plane arrays (IRFPAs). Conformal electroplated gold (Au) thin films on 32 X 64 staring arrays were used to integrate isolated MWIR HgCdTe detectors in each of the 100 micrometers X 100 micrometers unit cells to the input of the CMOS R/Os. Five micron wide Au thin films were used to make a conformal interconnect to 10 micrometers high HgCdTe layers in 40 micrometers X 40 micrometers unit cells within 256 X 256 arrays. Multiple thin film interconnects do not limit the size of the unit cell for dual band and multispectral staring arrays.
The 640 X 480 MOS multiplexer developed for PtSi MWIR focal plane arrays has been adapted to LWIR operation. The multiplexer is very flexible and can be used in various operating modes. The MOS approach, with its high saturation capacity and low-temperature operating capability, is ideally suited for long-wavelength operation. In this paper applications of the multiplexer to IrSi Schottky detectors and SiGe heterojunction detectors are discussed.
Popcorn noise, also called burst noise, manifests itself as a random charge fluctuation in linear In0.53Ga0.47As detector arrays. The noise is not present in all the devices; it may affect only 1 or 2% of the photodiodes at a moderately high reverse bias (5 V) but represents a potential performance limitation for the InGaAs photodetector arrays. The random charge fluctuations can be of the order of a few hundred to a million electrons at room temperature and can also be observed at temperatures as low as 200 K. Our studies suggest that dislocations which originate in the InP substrate are the major cause of popcorn noise. This noise can be significantly reduced or completely eliminated by reducing the reverse bias of the photodiodes from 5 V to 2 V or less. Crystal defects seem to cause the popcorn noise when the diode is substantially depleted. A lower reverse bias prevents the depletion width from reaching these defects and thereby prevents the possible generation of popcorn noise.
The Microelectronics Manufacturing Science and Technology (MMST) program at Texas Instruments is developing a generic semiconductor device manufacturing technology for the mid-1990s. In-situ vacuum processing, modular cluster tool equipment, sensor-based real time process control and computer integrated manufacturing are key components of this activity. Although the program goal is to develop and demonstrate low volume, fast cycle time, cost effective silicon microelectronic manufacturing, the technology also is applicable to HgCdTe, GaAs and other materials. The extension and evolution of MMST concepts and technologies to HgCdTe focal plane array production at TI are discussed. The use of generic process modules, sensors, and energy sources are outlined and compared for silicon integrated circuit and HgCdTe FPA applications. Generic modules and those tailored to specific HgCdTe process requirements are discussed.
Routine cryowafer probe testing of readout integrated circuits at T equals 80 K has been demonstrated at the Hughes Aircraft Company Technology Center. This new method of wafer testing at cryogenic temperatures has greatly reduced the test time needed to qualify infrared (IR) focal plane readout devices for subsequent hybridization to detectors. We have developed methods for handling sawn wafers during auto cryoprobe testing which have eliminated the need to package the devices into 68-pin leadless chip carriers. This reduced handling has improved yields and further decreased test times. In addition, we are upgrading our current cryowafer prober to test hybrid IR sensor chip assemblies (SCAs) at T equals 80 K by incorporation of cold shielding. We are also designing a new 10 K cryoprober which will extend our readout testing capability to low-background applications. This paper describes the 80 K cryowafer prober as well as the advanced 10 K cryoprober now in development and discusses the improved test flow enabled by the cryowafer prober compared to packaged-part dewar testing. Test data is presented showing the excellent correlation between packaged-part dewar testing and cryowafer testing.
Hughes Aircraft Company has successfully demonstrated accurate, high-throughput, automated cryogenic wafer-level testing of a 64 X 64 readout integrated circuit (ROIC) for use in the Army's Advanced Anti-armor Weapon System-Medium (AAWS-M) program (contract #DAAH01-89-C-A012). Data acquired from our cryogenic wafer prober is compared to data acquired utilizing a standard 68-pin dewar. ROIC noise, transimpedance, transimpedance linearity, slew rate, data valid time, operating speed, and dynamic range test data are presented. Delivery system flow diagrams are presented to show the streamlining effect of cryogenic wafer probing on the ROIC test flow.