Two competing types of solid state image sensors have been available for some time. CCD imagers have superior low noise, low light level performance, but have relatively poor spectral response characteristics. Self scanned photodiode array (SSPD) imagers can be produced with near ideal silicon spectral response, but have higher readout noise. A new optimized structure, the charge coupled photodiode array (CCPD) imager is described which combines photodiode sensors with CCD readout to provide low light level capability along with ' smooth spectral response and high quantum efficiency.
The three major effects that degrade external responsivity of silicon from the 1/A theoretical curve for a quantum detector are: surface reflectance, surface recombination, and junction depth. Since the p-n junction must be very shallow, problems relating to surface are further enhanced. MOS type of processing is necessary. HC1 oxides and numerous acid clean-ups are utilized in order to obtain a contamination free surface with low Qss levels. Stringent process controls such as CV shifts, spreading resistance measurements, thickness monitoring etc., are used to analyze the surface contaminations, surface mobile charges, surface concentrations, junction depth, oxide thickness etc. Low surface concentrations of 1018 atoms/cm3 are achieved by low temperature Boron Nitride depositions. This helps in reducing surface recombination. Shallow junction depths of the order of a few tenths of a micron are achieved by low temperature controlled diffusions. In order to improve breakdown characteristics of these shallow junction devices, field plate and deep diffused p+ ring geometries are used. This increases the breakdown of these shallow structures. Low temperature annealing is employed to reduce the fast states on the surface. Comparisons are made between epitaxial and float zone material for the best UV response. Optimization of Si02 surface coatings are also studied. Responsivity measurements have yielded 0.18 A/w at 380 nanometers.
A high performance CCD area image sensor has been designed and fabricated with a maximized average quantum efficiency over the wavelength range 500 to 900 nm; an average quantum efficiency of 50 percent was achieved. A multi-deposition polysilicon gate technology, supplemented by silicon nitride, was used. The measured response spectrum has the same major features as the computer generated spectrum for the optimum structure, although the measured spectrum is distinctly smoother. The CCD image sensor is of the time delay and integration (TOT) type. This paper describes the CCD image sensor configuration, the details of the unit cell, and the computer model boundary conditions, such as what layer thicknesses were allowed. A large number of layer-thickness designs were run on computer; the results are summarized. The wafer processing sequence is outlined. Possible explanations for the differences between theory and experiment are presented.
A development program is currently being undertaken to produce photon-counting detector arrays which are suitable for use in both ground-based and space-borne instruments and which utilize the full sensitivity, dynamic range and photometric stability of the microchannel array plate (MCP). The construction of the detector arrays and the status of the development program are described.
Patterned optical filters have been used for color TV cameras and infrared imaging systems. These filters can be manufactured by combining the techniques of micro-photolithography and resist lift-off. Micro patterns containing single-band or multi-band filtering can be produced on the surface of a single element. Current capabilities for the production of patterned filters will be reviewed, along with the practical limitations imposed by manufacturing techniques. Spectral characteristics required for visible and infrared applications will be discussed and compared with the characteristics commonly achieved in routine manufacture. Limitations on filter thickness, pattern size and pattern geometry which are dependent on the production techniques, will be treated. Micrographs of typical patterns will be presented for illustration. Practical manufacturing tolerances will be discussed. Prototype development work in thicker filters, tighter tolerances, and miniaturized patterns will be presented.
Possible system applications of Infrared Charge Transfer Devices are reviewed. It is found that this device technology can have a very significant systems impact. Analyses are performed to calculate the quantum efficiency, quantum yield, frequency response, photoconductive gain, operating temperature, noise and the distinction between parallel and transverse bias configurations of silicon detectors. Tables of silicon detector properties are included. Approaches to the interface circuitry which couples the detectors and the CTD multiplexer are examined. Examples of existing low background and high background IRCTD detector arrays are given.
Three types of Texas Instruments CCD sensors have been successfully incorporated into ICCD (intensified CCD) tubes. Intensifier tubes of varying design and application have been manufactured by three different manufacturers. Typical characteristics of the thinned CCDs incorporated into these ICCD's will be presented. Problems concerning the compatibility of tube processing and CCD performance will be discussed, as well as the procedures used to minimize CCD degradation during tube processing. ICCD tube characteristics will also be presented.
Recent advances in thermoelectric cooler, fiber optic, and image intensifier technology have been combined with advanced solid state optical detectors to produce high sensitivity, extremely wide dynamic range detectors for applications in optical spectral measurement. Spectral data from detectors is digitized and signal averaged for higher precision, then recorded under computer control for advanced data reduction. A general comparison of TN-1223 series array detectors to vidicons is made, with sensitivity and dynamic range advantages of the array detectors described.
The number and variety of digicon-type devices has been increasing at a rapid rate in the last year. This paper describes most of those which are new either in type of array used, design of electron optics, photocathodes, faceplates, or unusual applications. New parallel output digicons have been built using large two-dimensional arrays for UV and visible photometry and spectrophotometry. An intensified charge coupled device with 10,000 channels will be described and test results given.
This paper is a progress report on the development of an Intensified Charge Injection Device (ICID). The basic ICID mechanisms and its experimental verification are presented. A 244x250 element array, specifically designed for the intensified mode, will be described along with its expected performance.
With the trend toward higher circuit density and finer lines, printed circuit inspection becomes more important both in terms of assuring adequate functional capability, as well as long term reliability. Visual inspection of such circuits in a high volume production environment becomes more difficult, prone to error, and expensive. Automatic optical inspection appears to provide the only effective means of performing this operation. Such a method has application not only for final product inspection, but also for inspection at intermediate stages of production including art work inspection. A practical system must be able to convert the printed circuit pattern into a binary two-dimensional representation and then to process this representation to detect flaws. Development of the binary representation is hindered by the low optical contrast presented by many printed circuit material systems. The detection of flaws depends on selecting error criteria that are indicative of circuit quality, and that are practical to implement. This paper describes a prototype automatic inspection system which uses minimum line width and line clearance criteria. This system provides high resolution and uses linear solid state arrays for sensors. The conversion to a binary representation is made after correcting for system spatial nonuniformities in the illumination, optics, and sensing elements. Internal registers, which are continually being updated, are used to store successive scans of the image. At the same time, the contents of the registers are processed in parallel using combinatorial logic to implement the error criteria.
An automatic crystal growth control system for the edge-defined film-fed growth (EFG) of silicon ribbon is described. The system uses optical sensing with CCD Linear Imaging Devices to determine the process status and to direct control decisions. In EFG, the crystal grows from a film of melt whose shape is controlled by the dimensions of the perimeter of a die. The thickness of the ribbon is determined by the die top thickness and the height of the solid liquid interface above the die top (meniscus height); the position of the ribbon edges is very sensitive to the temperature distribution near the ends of the die. During growth, it is desirable to measure the meniscus height and the position of the edges to make control decisions. An optical image of the growing crystal is formed by a lens system and three CCDs are placed in the image plane to make these measurements. The signals from these CCDs have been used to control the power supplied to small heating elements near the die top. Appropriate selection of the viewing angle and radiation shields will enhance the delineation of the meniscus height and edge positions. Success has been obtained in controlling the edge positions. However, the presence of carbide particles at the die top constitutes a serious source of noise in interface height measurements.
A thermistor sensor dedicated to measure power guided by an optical fiber is described. It measures optical power in the microwatt to milliwatt range through the near infra-red; it is a high-efficiency device, with absolute calibration, and its input is an optical fiber.
To optimize the design of a thermoelectric heat pump for minimum input power, it is necessary to make a good estimate of the heat load. Multistage thermoelectric heat pumps are required to cool temperature sensitive components to 233°K (-40°C) or colder. In these cases the relationship of Coefficient of Performance, COP, vs Cold Side Temperature, TC indicates that any small error in the heat load calculation is reflected as a large error in input power. The relationship of heat load and input power relative to cold side temperature is discussed in this paper. Graphical presentations are used to estimate the heat loads for these examples and to show the effect on input power. Typical examples are as follows: 1. Cooling of a multielement infrared detector array to 193°K (-80°C), 2. Cooling of a single element silicon laser sensor to a temperature of 163°K (-110°C), 3. Cooling a self scanned array DIP to 233°K (-40°C) in Nitrogen and 4, Cooling a self scanned array chip to 233°K (-40°C) in Zenon. Each example has specific problems in evaluating heat load and input power requirements.
A computer program has been derived to quantitatively compare electro-optical systems through the distribution of charge within discrete sensor arrays. The total effect of optical components and discrete detectors on point source inputs is found by utilizing both theoretical and empirical values of MTF and transmission as a function of wavelength. Data output includes responsivity and effective quantum efficiency without regard to the MTF effects of the system. Then, the total energy spectrum is convolved spatially and spectrally over a 7 x 7 matrix to obtain S/N and total signal in each element. The effects of optics and CCD pixel sizes on point source imaging are shown.
In this work we discuss low brightness contrasts anticipated from bottom reflection changes due to water depth and other factors. Even under good lighting conditions it is necessary to expose film at full aperture and boost it in processing to an equivalent film speed of 1000. The best near term solution to this available light problem appears to be to turn to photoelectric sensors which also have inherent data handling advantages. The second and a closely related problem is depth calibration for different water spectral trans-missities. A red-green ratioing method was early devised by Moore (1945) for this problem - his "optical transparency method." In Southern California coastal waters there is a seasonal as well as individual storm change in coastal water turbidity due to run-off and dumping of terrigenous sediment and its mobilization by coastal currents. The problems of solar brightness variations on the water surface and the spatially complex calibration for inhomogeneous suspended sediment loads are apparently well-handled by the technique of electronically ratioing the difference of red-blue and red-green channels of the 9-channel Bendix EMSIDE scanner. The basic concept is that the red channel (6492 A) does not penetrate the water and hence, contains only surface infor-mation (e.g., solar glitter patterns, sea state variations, surface sediment plumes), whereas the blue and green channels contain a combination of surface and depth penetration information (e.g., blue light absorp-tion, green light scattering, bottom reflectances). Thus, differencing of these separate bands against the red band produces images free of surface effects. Ratioing of the differenced blue and green bands electronically produces a synthetic image whose density values are automatically calibrated in the same way as suggested by Moore in his optical transparency method. As a result we find the corresponding ratioed images of Catalina and Little Harbors, free of sun glitter and density changes. This synthetic image may then be isodensity traced and produce even depth density differences despite sharp differences in suspended sediment load.