Hybrid focal plane arrays for infrared detection have progressed rapidly over the past seven years from the point of proof of feasibility to demonstration of imagery. Hybrid arrays generally use two different semiconductor materials; an infrared detection material such as HgCdTe or InSb, and a silicon substrate chip which performs initial signal processing on the detector signal. Hybrid arrays have also been fabricated using extrinsic silicon detectors on a silicon substrate (SOX). A third important element in hybrid arrays is the interconnect technology used to electrically connect the detectors with the silicon substrate chip. Mechanical and thermal interaction of the two materials must be considered as well as the electrical. Hybrid arrays fabricated with InSb and HgCdTe detectors have demonstrated nearly BLIP operation. This paper will discuss some hybrid chips that have been developed and future trends in hybrid arrays.

This paper discusses mosaic focal plane evaluation techniques useful in module end-to-end performance evaluation, i.e., responsivity, NEP and in detector channel parameter characterization. A selected detector channel block diagram configuration is used as a model for parameter characterization. The channel performance is expressed in terms of a transfer function which is evaluated by a series of radiometric and electrical signal injection tests. Time and frequency domain solutions are used as computer algorithms to solve for channel parameters. Data is presented in a histogram format for several parameters.

The next generation of infrared systems will provide dramatic increases in sensitivity and resolution due to the larger number of detectors that can be included on the focal plane. Current IR imaging systems utilize a relatively small number of detectors (typically fewer than 200) and each is connected to an individual video processing circuit. The higher detector densities afforded by new IR semiconductor technologies clearly require signal processing on the focal plane, if for no other reason than simplification of the cryostat/system interface. However, the relative immaturity of these new detector technologies currently requires conditioning of the detector signals in order to achieve a signal-to-noise ratio dictated by shot noise on the incident radiation (so called "background limited performance", or BLIP). This paper describes the application of silicon n-channel MOS (NMOS) technology to the development of analog circuitry for IR signal processing.

The MIDASP program has been aimed at development of the fundamental technology necessary for fabrication of hybrid (Hg,Cd)Te detector/silicon signal processor focal planes for second generation FL 1R systems. Effort has been concentrated in four areas: 1) 3-5 micrometer (Hg,Cd)Te photodiodes, 2) detector/CCD mechanical inter-connect, 3) detector/CCD electrical coupling, and 4) fabrication and evaluation of small focal plane structures. Detector/CCDs with several 8-TD1 channels multiplexed to a single output have been successfully demonstrated, with an overall D* of up to 1x10¹⁴ cm-Hz1/2/W.

The large amounts of resources spent thus far on IR mosaic focal plane assemblies (FPA) have unfortunately as yet produced little of operational utility. The main problem areas are dynamic range, on-focal plane sianal conditioning, uniformity of performance, and producibility. These problems cannot be isolated to either the detectors or the read-out electronics individually but must be resolved in an integrated manner. New FPA architectures are called for to provide large electronics real estate while retaining low cost and producibility. One such configuration, HYMOSS, has been under development at Irvine Sensors Corporation for the past two years. HYMOSS is designed to provide the requisite dynamic range, uniformity, and noise characteristics by utilizing the third dimension in an extremely compact form.

Two-dimensional x-y addressed mosaic arrays consisting of more than 12,000 lead-sulfide detectors in a package measuring 1/4 x 3/4 inch have been successfully developed and tested by the Aerojet ElectroSystems Company. The pulse-bias modulation concept, design, fabrication methods, test techniques, and test results for individual modules (192 x 64 detectors) are presented. The successful assembly of four of the individual modules into a subassembly consisting of approximately 50,000 detectors is described. The requirements for testing to achieve a greater throughput rate are discussed.

Mercury cadmium telluride, Hgi_ x Cd x Te, a direct band-gap semiconducting alloy with the band-gap adjustable by varying the alloy composition, has become the material-of-choice for innumerable infrared-detection applications over the 2 to 20 micrometer spectral range. N-type, p-type and intrinsic material can be readily prepared, either by doping or by adjustment of alloy stoichiometry, greatly enhancing the versatility of this alloy system. Material requirements for the next generation of infrared detectors, large area two-dimensional arrays, have placed unprecedented demands on Hgi_ x Cd x Te materials technology. Major research efforts are being conducted which address this need including epitaxial and bulk crystal growth methods, both on earth and in space.

This paper reviews the status of two Indium Antimonide Charge Injection Device (CID) focal planes, one composed of multiple line arrays, the other made with an area array using random access addressing. After an introduction to CID operation, factors governing Performance Efficiency (PE) in both line and area arrays are considered and compared to experimental results. The charge sharing mode of operation of a two dimensional array is presented, along with its perform-ance implications. The linear focal plane consists of 48 linear arrays, each with 64 elements. Each InSb array is addressed by a silicon scanner chip. The scanner outputs are serviced by preamp chips mounted perpendicular to the focal plane. The 32x32 element area arrays have been coupled to combinatorial decoders permitting random selection of any site of the array. Focal planes consisting of a single 32x32 array with two silicon decoders mounted in sealed off glass dewars are described.

Both monolithic and hybrid silicon focal planes are presently considered for future IR system applications. Significant progress has been made in the development of two-dimensional arrays because the IR detection properties of extrinsic silicon are well developed and the state of the art of manufacturing CCD devices is within the commercial industrial domain. Monolithic focal planes (MFP) combine the function of signal radiation conversion and electronic signal readout and processing in one device, whereas in a hybrid focal plane (HFP), the detection function is physically separated from the charge injection and transfer function; both devices are mechanically interconnected via a metallic bump matrix. The monolithic approaches benefit from the utilization of only one device. The hybrid approaches, on the other hand, benefit from the, ability to optimize and select independently the sensing and processing elements.

A short review is presented of the development of monolithic platinum silicide Schottky-barrier focal plane arrays with CCD readout registers. The development of the 256 line sensor and the 25x50 area imager are described.

A multiplexing technique applicable to mosaic arrays of (2N-1)xM infrared photoconductors is described. This method employs bias coding. It represents an alternative funda-mentally different from the sequential-readout time-division multiplexing approach in general use. Some of the relevant theory of coding is summarized in both analytic and matrix form. Rationale is presented for the selection of binary codes in general, and maximal-length (pseudo-random) shift-register sequences in particular. Results obtained from a prototype, 31 x 31-element doped-silicon mosaic and data processing computer, oper-ating under low-background conditions, are described.

The invention of the charged coupled device (CCD) almost ten years ago and the utility of the charge transfer device (Call) in general has made development of mosaic focal planes possible. The nature of the metal-oxide semiconductor (MOS) manufacturing processes utilized in most Calls posed special requirements for material, process control, and device evaluation which have been and still are, the object of intensive technology development programs in the past several years. Achieving the long-standing goals of high-volume, low-cost production potential of the technology for mosaic focal planes is still in the future. The degree of achievement of these goals will depend a great deal upon adoption of some standardization of device architecture, operating conditions, test parameters, and evaluations techniques. Production of these devices will require capital investments for dedicated specialized facilities necessary to achieve volume production.

Performance evaluation of focal planes begins with arrays, continues through modules, and concludes with integrated focal planes. The first part of this article clarifies the functions and complexity of four different classes of evaluation, R and D, Prototype, Production, and Integrated focal plane. The second part discusses several major issues currently confronting the focal plane community - some of these are readily resolvable while others await further technological advances. A conclusion is made that automation of a facility having a high throughput capability is essential for production of arrays and that future focal plane designs must incorporate built-in diagnostic and fault-tolerant structures to ease the burgeoning burden of evaluation.

Focal plane production testing to be cost and schedule effective involves many tradeoffs which appear very difficult to optimize. This optimization appears particularly difficult for the more sophisticated devices which have detailed and interactive specifications. These tradeoffs will be broken down into major categories with clear optimization procedures and guidelines provided so that their individual impacts will be more easily understood. Production yields of focal planes can be significantly increased and costs reduced - the technology and techniques are at hand. The manufacturers ability can be further optimized through wise, concise testing at cryogenic temperatures which has long been an industry bottleneck. The six major categories to be discussed are: 1) Go/No Go Testing; 2) Characterization; 3) Specification vs Performance and Cost Tradeoffs; 4) Earliest Significant Testing; 5) Room and Cryogenic Temperature Testing and 6) Self and Diagnostic Testing. Practical examples will be shown to demonstrate the basic techniques so that a firm foundation of each of the major categories will be obtained.

This paper describes many of the terms pertaining to radiant flux that reaches a receptor surface, such as photographic film or an infrared sensor. Note that the words "focal surface" are used in the title; for, indeed, not all receiving surfaces are plane--nor do they have to be.

Infrared mosaic focal plane arrays (MFPA) are complex integrated microelectronic devices whose operation is based on a large number of voltages in which the performance is very sensitive to voltage variations. Due to chip-to-chip variations, a computer-based system and a systematic procedure for efficient individual optimization of these operating voltages is essential. At present, the feedback or control in the loop is handled manually, but the system development is oriented toward a transition to automated control once sufficient understanding of the optimization procedure is gained. A fundamental premise of this optimization procedure is that the MFPA performance data is processed in near real-time and displayed in a concise and vivid manner, thus facilitating quick, easy, and confident interpretation. In the Mosaic Sensor Test and Calibration (MOSTAC) facility at the Lockheed Palo Alto Research Laboratory, a summary page output is immediately presented along with visual display of the spatial distribution of the signal, noise, and signal-to-noise. The procedure has generally consisted of (1) coarse optimization by determining the regions of good device responsivity and (2) fine optimization by minimizing the noise (while not losing the responsivity). This has allowed quick optimization of IR mosaic sensors with several MFPA chips in a single day. This paper describes the MOSTAC facility and how it is used to provide rapid optimization of MFPA performance with some specific examples on infrared charge-coupled imagers.

The computational burden of any image-processing system is on the preprocessing functions. These include such functions as two-dimensional convolution, edge extraction, texture-feature extraction, and nonlinear filtering (e.g., median filtering). This paper describes the fabrication of three CCD chips on which several image-preprocessing functions have been implemented with effective operation rates equivalent to 10,000 MOPS. The paper also briefly reviews the work done on the first two chips and then describes in more detail functions included on the third chip and the experimental results. The functions discussed include a 5 x 5 voltage-programmable convolution, a 26 x 26 convolution, a 7 x 7 mask-programmable convolution, a 5-element sort for median filtering, and a 3 x 3 Laplacian filter. These circuits have all been designed to operate at a 7-MHz pixel rate.

This paper reports the results of digital simulations of low level infrared (IR) signal processing/detection techniques in the presence of zodiacal light background. The study investigates the applicability of pulse shape discrimination techniques to the detection of IR signals where the fast rising target signal pulses are superimposed on slowly varying background. The quantities that characterize the techniques are defined. An improvement of signal-to-background ratio is the essential advantage in applying the new concept.

The noise performance of a 32 channel p-surface channel CSD multiplexer for an infrared focal plane array was measured at room temperature and at 160 K. Theoretical calculations were made of the contributions from kTC noise, dark current noise, and interface state trapping noise. The calculated results agree well with the experimental values. The noise of the on-chip source follower output stage was below the 900 rms holes noise floor established for the measuring system by the sample and hold. The calibration of the equipment was checked by comparing the theoretical and measured noise values for optically injected charge carriers. The agreement was better than 10%. We measured 1 to 2 times theoretical kTC noise with the potential equilibration input. The design goal for these devices was a temp3ral noise voltage of 15 µV (3500 rms noise holes) referenced to the input. Cooled to 160 K we measured a noise voltage of 16 µV (3750 rms noise holeq referenced to the input. The spatial or fixed pattern noise was much larger: 12 mV (2.9x10 rms noise holes) referenced to the input.

Evaluation of flight data from a prototype mosaic sensor is required for characterization of sensor performance and data quality analysis. The high data rate from a mosaic sensor necessitates a processing facility designed to reduce large quantities of data to a few easily interpreted indicaters. A computer facility for this purpose is currently in operation. Several of the sensor and system characteristics found to be significant are discussed as well as computer-aided techniques for their quantification in terms of an operational system. Although data evaluation is addressed in terms of a radiometric staring mosaic sensor, many considerations are applicable to nonradiometric and scanning sensors.

An instrument for performing optical tests on infrared photodiodes has been designed and tested. This instrument uses Pb-salt tunable diode lasers as a source and can record spatial sensitivity maps of detector elements, temporal response to fast pulses, and spatial maps of the temporal response as well as uniformly illuminate large area detector arrays. The spatial maps presented demonstrate a spatial resolution of 100 Tim. The pulse data presented indicates that the optical risetime of these Pb-salt lasers is less than 500 ps for pulses of 1, 10, or 100 ns duration.