Basic properties pertinent to the application of bismuth-doped silicon as an infrared-photoconducting detector material are described. Si:Bi provides a viable n-type alterna-tive to the p-type gallium-doped material that has been more extensively used in the past.
Large-area focal planes require mechanical assembly techniques which must be compatible with optical alignment, minimum deadspace, and cryogenic requirements in order to achieve optimum performance. Hybrid extrinsic silicon has been found particularly suitable for such an application. It will be shown that by choosing a large-area extrinsic silicon detector array which is hybrid-mated to a multiplicity of multiplexers a very cost-effective and high-density focal plane module can be assembled. Other advantages of this approach are inherent optical alignment and excellent performance.
Hgl-xCdxTe is becoming increasingly important for the preparation of both monolithic and hybrid infrared focal plane arrays. The preparation of such devices utilizing thin film HgCdTe is very attractive due to its cost-effective feature. Based on a study carried out at General Dynamics, sputtering, which minimizes the vapor pressure problems of HgCdTe deposition, promises to be a relatively simple, inexpensive and versatile technique for preparing high quality single crystal Hgl-xCdxTe films with electronic properties suitable for array (large area) fabrication. This paper reports on the initial results of this study which support this claim.
Recent advances in Hgl-xCdxTe photodiode technology have resulted in increased detector sensitivity at elevated temperatures over the spectral region from 1-20 μm. Short wavelength photodiodes sensitive to 1-3 pm radiation have room temperature peak detectivities, Dλ, between 4 x 1011 and 8 x 108 cm Hz1/2/W, dependent on peak wavelength, A. Medium wavip 2length photodiodes sensitive to 3-5 im radiation have peak detectivities of 1 x 1011 and 1 x 1012 cm Hz1//W at temperatures of 193 K and 130 K, respectively. Long wavelength photodiodes sensitive to 12 μm radiation have peak detectivities of 5 x 1010 cm Hz /W at 65 K, and 5 x 1011 cm Hz ½/W at 10 K.
The Charge Injection Device (CID) is an x-y addressed solid state imager introduced by General Electric in 1973 as a television sensor. A family of sensors has evolved to exploit the singular CID characteristics and capabilities. This paper describes the structure and principle of operation of a typical CID. A TV compatible differential current sensing mode of operation is described. A multiple non-destructive readout mode is described with examples of application to transform code generation and centroid interpolation. The CID radiation hardening potential is briefly reviewed.
The advent of charge transfer devices (CTD's) stimulated the development of a new generation of star sensors. The mosaic focal plane is a major componenet of the sensor. This overview describes beneficial characteristics of CTD's for star sensor applications and several mosaic focal plane configurations.
A technique for building a high resolution linear focal plane using a single low resolution area detector array is described. A silicon area imager is fabricated with an opaque light shield layer having small clear apertures located over the detector elements which define the picture elements. By displacing the apertures sideways from one row of detectors to the next and using the array in conjunction with a multiline data buffer, a line scan resolution equal to the total number of detector elements in the area array is achieved. The design considerations and tradeoffs involved in using such a technique will be discussed. A practical realization of such a system will be shown which uses an array built with a 4 x 1024 element configuration to achieve 4096 resolution elements per line. Multispectral capability is achieved by using 4 such arrays fabricated on the same device with spectral filtering provided by patterned interference filters on the cover slip. Array output and results of device characterization will be shown.
A CCD camera is described which has been designed for single-photon X-ray imaging in the 1-10 keV energy range. Preliminary results are presented from the front-side illuminated Fairchild CCD 211, which has been shown to image well at 3 keV. The problem of charge-spreading above 4 keV is discussed by analogy with a similar problem at infrared wave-lengths. The total system noise is discussed and compared with values obtained by other CCD users.
Staring sensor systems require large fields-of-view with fine resolution. This combination requires systems which contain several million detector elements, or pixels, on the focal plane. A modular approach for this focal plane assembly has been deve,loo,:Ld at Rockwell International. The assembly is romnrised of detector chips mounted on carriers which in turn are placed ',TILL, sqbmodules. The submcduLe: ara modules and the modules into a focal plane. It is desirable to have a very low, inactive, or dead, area in the total focal plane array. The assembly must address at least three interfaces: mechanical, electrical, and thermal. The mechanical tolerances in obtaining the appropriate focal surface are very critical. The focal surface must be maintained at a low temperature with a very small variation during operation. Low dead space demands multiple lead fan-in and many signals being transmitted simultaneously thus the electrical interface is quite critical. This paper will describe a prototype assembly which was developed to evaluate the major assembly and interface areas. Tooling and assembly fixtures were built as required. Assembly tolerance measurements made and checked after thermal cycling and launch environment vibrations and shock will be presented.
We have developed a new, systematic approach to analyzing the effects of different infrared mosaic focal plane geometrical configurations in order that a confkguration may be selected which is best for a particular application. This approach has been incorporated into a computer simulation program. The program provides a quantitative and graphical methodology which can be used to assess the performance of user-specified focal plane geometries when used in combination with specified target complexes. To do this, simulated detector signal data are first calculated. These data are used to directly develop an apparent image, or map, of the scene. Recently-developed techniques used to survey and map the sky at X-ray wavelengths have been modified appropriately for the infrared focal plane situation in order to do this. Once a map is developed, it is compared to the known target complex characteristics in order to assess its validity. Various focal planes and different instrumental noise levels can be used with a single reference scene in order to select a best focal plane configuration. The simulation will be described by reviewing a recent example application.
This article reviews current development and issues in detector-CCD interface methods employed in mosaic focal plane arrays,particularly those designed for the infrared spectrum. The functions of inter-faces, such as signal inputing, impedance matching, signal conditioning, and background suppressing, are discussed along with their compatibility with MOS (metal oxide semiconductor) and cryogenic techniques. An attempt is made in classifying the interface methods into five main categories - monolithic (optical), direct injection, gate modulation, buffered circuit, and preamplified circuit - while a qualitative assess-ment is made of the capability of background suppression for several interfaces. The article concludes by identifying four classes of limitation of the interfaces - architectural, electrical, test and control, and processing - of achieving high sensitivity, wide dynamic range, and large linearity.
Although charge coupled staring mosaic FLIRs offer ultimate sensitivity improvement factors of 30 or more over existing scanning FLIRs, dynamic range and uniformity limitations restrict realizable improvement. By dealing only with background shot noise, charge handling, and uniformity, the basic limitations are determined without complicating the analysis with additional electronics reset noise, transfer efficiency, fast interface state noise, and other possible additional noise degradations. It is also shown that describing the detector operation as "BLIP" does not in itself imply that the best sensor electronics configuration is being utilized. Ultimate Noise Equivalent Temperature Difference (NETD) of .0012°C and .0003°C are predicted for 3-5 m and 8-5 im sensors. Further studies of atmospheric thermal uniformity statistics may be required to determine the ultimate usable sensitivity.
Staring mosaic focal planes, which are under development, combined detectors and associated signal processing to perform their mission functions. Much of the development work is centered on planar approaches in which a two dimensional detector array is sandwiched together with a two-dimensional signal processing array. This paper defines the "Z-tech-nology", a three-dimensional architecture as contrasted to the planar architectures. The status of the Z-technology as fabricated and tested at this time is reviewed, and near-term and long range design developments and goals are described. It is concluded that the Z-technology approach represents a low risk alternative to planar approaches, and has an inherent degree of flexibility which should prove to be cost effective.
Future generation, very large-scaled mosaic focal plane assembly concepts based on the Z-technology are described. These advanced concepts represent logical extensions of the current base-line Z-technology by exploiting the inherent advantages of this architecture in the areas of packaging flexibility and usable volume increase. Direct consequences of these attributes on the performance, risk factor and cost effectiveness in mosaic focal plane hardware assembly are emphasized.
The concept of a focal plane mosaic composed of charge coupled devices (CCD's) optically bonded to tapered fiber optic bundles is described. The mosaic eliminates a number of disadvantageous features associated with the application of CCD's to large area focal planes such as the low yield and transfer efficiency related MTF losses inherent in large arrays as well as permitting peripheral signal processing architecture without loss of imagery. Experimental results obtained from a prototype version of the mosaic are presented and realizable improvements to second generation devices are discussed.
Self-scanning pyroelectric infrared detec tor arrays (32 and 128 elements on 100 μm center-to-center spacing) have been developed and characterized. The detectors consist of a monolithic crystal with in dividual elements deliniated by photo-etched electrodes. They are coupled to an integrated circuit array of self-scanning FET multiplexing switches, mounted in the same IC package with the detectors, for periodic signal readout. Current applications are instantaneous beamscans for infrared lasers, high resolution detec-tors for real time spectroscopy, and un cooled operation for thermal imaging. Pulsed and chopped CW radiation applications are to be discussed, as are timing requirements, sensitivity, dynamic range, and circuit considerations.
This paper presents experimental results obtained from an MFPA. The discussion includes: (1) a description of the MFPA, (2) how performance measurements were taken and performance data, (3) a description of the diagnostic techniques used to identify noise mechanism and the diagnostic data, and (4) a presentation of data which demonstrates noise performance improvement.
Surface acoustic waves excited in a Si-Si02-ZnO layered structure can produce a traveling electric field in the silicon substrate. Charges stored in the traveling potential wells can be transferred at high speed and density and with less complexity than conventional coupled devices. The monolithic structure under investigation for the SAW-charge transfer device consists of a silicon substrate, a thin silicon dioxide insulating layer on top of which a ZnO piezoelectric film is deposited by sputtering. The surface acoustic waves are excited by interdigital transducers. The signal charge is injected into traveling potential wells that travel with the velocity of sound. The presence of a thin shorting plate placed on the ZnO film, over the charge transfer region can enhance the acoustoelectric potential at the Si-Si02 interface, thus resulting in a more efficient device. An 80 MHz, 2g-second SAW-CCD has successfully been fabricated. An optical application utilizing such a structure is proposed. It can be used in place of a conventional interline transfer design. Surface acoustic waves are launched before the charges are transferred from the sensor region to the transport region.
A unique process for fabricating HgCdTe photodiodes directly on a variety of substrate materials has been developed. This process, which can achieve fill factors in excess of 90%, has the advantages of minimal thermal mismatch, less than 1% crosstalk and scaling to large array sizes. This structure consists of thin, fully delineated HgCdTe detector ele-ments bonded to a substrate which contains CCD's or individual lead-outs for each detector. The junctions are connected with a common grid electrode. D* (4.5μm) of 1.3 x 1010cmHz2/w has been achieved in 4 x 4 element arrays fabricated directly on silicon substrates. These arrays have immediate applications in mosaic sizes up to 4000 detectors and future development potential of 128 x 128 electronically scanned focal planes.
This paper discusses the theory and practice of tunable spectral filters, particularly the types that operate on the principle of light interference and diffraction in a birefringent crystal. Three methods of tuning are described: acousto-optic (AOTF), electro-optical (EOTF) and electromechanical (Solc). Capabilities and limitations of all three are compared. The AOTF is most versatile, is simple to construct, and can be rapidly tuned. Due to its relatively high drive power for mosaic sensor applications AOTF has been limited to a small portion of the focal plane (e.g., for laser warning). In principle, the EOTF has the advantage of low drive power. However, because of the material problems and fabrication complexity involved, practical implementation on the focal plane is probably many years away. For near-term mosaic sensor requirements, the electromechanically tuned Solc appears to be the most promising solution.
The dependence of the signal-to-noise ratio (SNR) on three important focal plane design functions--frame time, time delay and integration (TDI), and pulse shaping--is examined. A general focal plane model is developed that permits the SNR, to be parameterized separately in terms of the three functions. The focal plane model treats both staring, mosaic sensors and scanning, line sensors as special cases of the model parameters, and accounts for the presence of multispectral detectors arranged in either a side-by-side or a sandwiched configuration. Both noise in the detectors (including photon noise) and CCD noise that arises in transferring the signal off the focal plane are considered, but it is assumed that the detectors are not limited by l/f noise. General parametric expressions for the SNR in terms of frame time, the number of samples per blur spot (pulse shaping), and the number of TDI steps are derived, and specific results for both a mosaic focal plane and a scanning line focal plane are presented. It is found that, regardless of the focal plane design, the SNR increases at least as the square root of the frame time and decreases at least as the square root of the number of samples per blur spot. The behavior of the SNR in terms of the number of TDI steps is more complicated and depends on the specific focal plane configuration. It can be stated generally, however, that the presence of CCD noise ultimately causes the SNR to peak and then decrease with increasing TDI. The peak SNR occurs at large TDI steps for scanning line sensors and at small TDI steps for mosaic sensors.
This paper outlines a successful approach to the performance evaluation of competing concepts. Only space--looking systems are considered. No specific results are given; rather a general methodology is presented.
Measurements of system optical modulation functions (MTF, SWR) may be distorted by time-dependent environmental effects (thermal, vibration, flexure) and by electronics drift. Fast data collection may therefore be advantageous by minimizing drift time. The problem of fast data collection is accentuated when modulation data must be taken on a large number of detectors in a focal plane array. A method has been developed for generation and storage of knife edge data from focal plane arrays, where data collection time per detector is in the sub-millisecond range. Once knife edge collects are completed, MTF response is found using conventional convolution techniques. SWR is obtained directly from knife edge response using a computerized simulation algorithm which bypasses use of MTF harmonics. Requirements for detector electronics speed, damping, and dynamic range are considered.
Many of the advanced sensor concepts, which are either recently developed or are under development, utilize one or more large mosaic arrays of detectors. It has been shown through various studies that the only practical way to handle the very high rates of data produced by these large mosaic arrays is to perform a variety of signal processing and 1 data compression functions on the focal plane by using some form of focal plane processing. Tn this paper, we have postulated a focal plane layout and a scheme to carry out various processing functions on the focal plane using Charge Coupled Devices (CCD's). The performance of this proposed focal plane processing scheme under a host of error sources has been evaluated through simulation and emulation. The parametric performance results are presented in this paper.* The simulation was carried out using a representative Monte Carlo statistical method and six error sources were used to evaluate the performance of focal plane processing schemes under various input signal to noise ratios. A simulation of the performance was carried out under the presence of all of the error sources and results such as the relationship between output signal to noise ratio vs the number of TDI stages were obtained.
Staring infrared imagers presently under development possess an inherent fixed pattern noise characteristic at the output which is Primarily determined by the non-uniformities in detector resoonsivity, CCD threshold variations and detector/CCD electrical coupling circuitry. This fixed pattern noise can take up as much as 50% of the focal plane dynamic range and, as such, generally requires high accuracy, high speed compensation electronics which then allows the full NEAT sensitivity oerformance of the focal plane to be realized. This paper describes the requirements for the comoensation electronics and discusses different implementation circuitry for both conventional "shuttered" calibration (i.e., reimaging of a uniform thermal reference source to calibrate the non-uniformity coefficients) and for a concept which has been developed called the "shutterless" compensation which uses scene dynamics and statistics to calculate the correction coefficient values.
In recent years, long wavelength infrared (LWIR) sensor systems have gained significant importance in a wide range of applications such as surveillance, reconnaissance and target acquisition. The interests in the LWIR sensor area can be reflected by the many technology development programs currently being sponsored and supported by NASA and the various DoD agencies.
This paper presents a discussion of some of the consideration in the development of an architecture for post focal plane signal processing using distributed programmable data processors. The resulting architecture employs a series of modules from which a signal processing system can be constructed based on the required sensor data rates. An analysis of a typical scenario is performed to show how a specific processing requirement can be defined using statistical modeling concepts. This analysis relates the number of memory/processing elements in the system to the number of objects in the field of view, the number of sensors and their format, and to the sampling rate of the sensor signal processing system.
A detailed discussion of statistical sensor performance analysis was presented in a previous SPIE publication. The essential point of the paper was to the effect that responsible sensor performance analysis in real world conditions required statistical formulations and techniques in order to obtain complete and unambiguous conclusions. In this earlier paper illustrations were presented of applications using this statistical approach. In addition, a method was developed for circumventing the difficulties inherent in Monte Carlo processes, customarily considered necessary for the statistical analyses of interest. It was mentioned in this previous work that further development of the foundations of the technique were in progress. The primary purpose of the present paper is to convey those elements of progress which have been achieved since the original work was presented.