Seven new CCDs are presented. The devices will be used in a variety of applications ranging from generating color cinema movies to adaptive optics camera systems to compensate for atmospheric turbulence at major astronomical observatories. This paper highlights areas of design, fabrication, and operation techniques to achieve state-of-the-art performance. We discuss current limitations of CCD technology for several key parameters.

A four-port 512 X 512 charge coupled device (CCD) imager hardened against proton displacement damage and total dose degradation has been fabricated and tested. The device is based upon an established thinned, backside illuminated, triple polysilicon, buried channel CCD process technology. The technology includes buried blooming drains. A three step approach has been taken to hardening the device. The first phase addressed hardening against proton displacement damage. The second phase addressed hardening against both proton displacement damage and total dose degradation. The third phase addresses final optimization of the design. Test results from the first and second phase efforts are presented. Plans for the third phase are discussed.

CCDs have been used in spectroscopy for a number of years and for all the obvious reasons. Unfortunately most scientific CCDs are square arrays and are not ideally formatted for spectroscopic applications. This paper discuses the design and fabrication of two CCD arrays specifically intended for use in spectroscopic applications. The devices have 1100 X 330 and 1752 X 532 pixel formats, and are fabricated using the three phase, overlapping polysilicon gate technology and they are available in both front-illuminated as well as back- illuminated versions. In addition, devices with enhanced UV sensitivity are being fabricated. Characterization data are presented. The architecture of these devices leads to some interesting applications which we discuss briefly.

We describe the performance of a 2k X 2k, 15 (mu) pixel, buried channel MPP-CCD (Loral FA2048) using different operating modes and the implementation of an anti-blooming clocking technique. The CCD is part of a camera system developed at Bonn University for astronomical wide field photometry and polarimetry. Besides two basic operating modes (partially and non-inverted mode) the multi pinned phase (MPP) design additionally allows a totally inverted mode providing the strongest reduction of dark current. The disadvantage is a low full well capacity of 120000 electrons/pixel which only depends on the small implanted potential offset. As an example for optimization by choosing adequate voltages we show how this original full well capacity can be raised almost by a factor of 2 without decreasing the quality of the read-out. Finally we discuss the physical understanding and technical implementation of anti-blooming and its future application in astronomical photometry. Using clocking rates up to 2 kHz we achieve a minimum anti-blooming efficiency of 400 electrons/sec/pixel/Hz and a low spurious charge generation further reduced by using ramps or intermediate steps in the anti-blooming clocking waveforms.

A 512 X 512 frame transfer CCD area array image sensor has been developed. The active area of this sensor is subdivided into 4 sections, each with its own output structure that increases the effective output data rate by a factor of four. With an output data rate of 15 MHz, an aggregate data rate of 60 MHz is obtained. This subdivision of the sensor area has been achieved while maintaining a pixel pitch of 7 micrometers . To obtain this pitch, the sensor employs three phase clocking for both vertical transfer and horizontal readout. A sensitive output structure was developed to sense the reduced charge packets from the small pixels. This output structure has a charge conversion efficiency of 7.85 (mu) V/e. The sensor has an excellent charge transfer efficiency and sensitivity making it ideal for high resolution scanning applications.

A frame transfer image sensor has been fabricated for 2/3' optical format. It has a high resolution (1024 X 1024), overexposure handling by means of vertical anti-blooming and square pixels of 7.5 micrometers by 7.5 micrometers . The data-rate is 30 frames per second. Its single output register can run up to 40 MHz. The register can be clocked bi-directionally, which makes it possible to produce a mirrored image. The noise level of the output amplifier is 31 electrons (rms value in 18 MHz bandwidth). These properties make the imager perfectly suited for machine vision applications where full resolution in a single shot is required. The paper describes the sensor, simulation results and measurements.

The basic principles of a new high-sensitive multi-element photosensor on the base of avalanche heterostructures with negative feedback (ANF) are considered. The main difference between such a structure and a conventional avalanche photodiode (APD) is non-stationarity of the electrical field strength in the avalanche region caused by the feedback. It is shown that when the non-stationarity is manifested at amplification of a single photocarrier, it changes radically the basic characteristics of the avalanche process. A qualitative physical model of ANF process as well as the results of numerical simulation and some experiments are presented. They show the sharp gain vs. voltage dependence that restricts manufacturing of multi-element APD is essentially smothered by the negative feedback influence. They also exhibit an ability of the ANF-based device to provide a unique, for a solid state photosensor, combination of low noise, high gain and low response time. Two types of the ANF devices -- MOS avalanche structure with pulse supplied voltage, and SiC-Si-based heterostructure with constant supplied voltage -- were realized experimentally. Both of them are shown to be easy in multi-element design manufacturing. The results illustrating the process of a few-photon light pulse registration by the SiC-Si ANF heterostructure are also presented. An opportunity of a new lead in the development of the high-sensitive multi-element photosensors on the base of the ANF technology is discussed.

Charge-coupled devices have found widespread use in many imaging applications and demonstrate many advantages over other detector technologies such as high speed, low power, low noise, and small physical size. Nevertheless, `off-the-shelf' CCD architectures are frequently not feasible for many unique applications, requiring custom CCD designs and processing. In this paper the authors discuss a CCD sensor developed specifically for a near- infrared linescan-type application. The authors discuss the architecture, design, and performance of the sensor, as well as current work in the project.

TDI sensors are a proven means of increasing the responsivity in a line scan imaging application. This paper describes the development of a family of high speed 96 stage TDI sensors. The sensor is available in a high resolution 2048 element version. A 512 element part will also be made. The number of TDI stages in the sensor can be selected in blocks of 6, 12, 24, 48, and 96 stages thus providing optimum sensitivity and performance over a wide range of illumination conditions. The device is fabricated using double metal, triple poly, buried channel, NMOS CCD process. The imaging region is 4-phase, 2-poly for maximum charge storage and optimum MTF. The pixel pitch on the sensor is 13 micrometers . The sensor employs a dual channel, 2-phase, 2-poly output shift register for high speed read-out. This technique enables halving the driving clock frequency thus reducing the power consumption which can be a severe problem at large data rates. Another benefit of dual channels is that each horizontal CCD (HCCD) pixel corresponds to two vertical CCD (VCCD) registers. As a result the charge storage capacity of the HCCD is doubled without having to increase the register width. The developed sensor operates at a combined data rate of up to 40 MHz. The maximum line speeds are 32,000 and 14,000 lines/sec for the 512 and 2048 element parts respectively. Methods to reduce the fixed pattern noise resulting from transfer inefficiencies between the two HCCD channels have been implemented.

Two types of high resolution CCD linear image sensors have been developed for AVNIR, which will be carried on the Advanced Earth Observing Satellite (ADEOS) in early 1996. One has 10000 pixels (pixel size: 8 micrometers X 8 micrometers ) with spacing of 8 micrometers for the panchromatic band (0.52 - 0.69 micrometers ). The other has 5000 pixels (pixel size: 16 micrometers X 16 micrometers ) with spacing of 16 micrometers for the multispectral bands (Mu1: 0.42 - 0.50 micrometers , Mu2: 0.52 - 0.60 micrometers , Mu3: 0.61 - 0.69 micrometers , and Mu4: 0.76 - 0.89 micrometers ). These bands are separated by a dichroic prism in the optical system. Each sensor has a staggered layout of Si p-n-p buried photodiodes, and an overflow drain for an anti- blooming and an electronic shutter operation. The sensor for the multispectral band (MU) has 2 CCDs and the sensor for the panchromatic band (PA) has 4 CCDs to reduce operating frequency. To prevent degradation of modulation transfer function (MTF) for the Mu4 band, which is a near-infrared region, we used a staggered arrangement and p^- on p⁺ epitaxial wafers. MTF in Mu4 band is 0.62 and 0.80 at Nyquist frequency (fn) and 1/2 fn, respectively. We have developed novel packaging technologies so as to obtain a focal plane flatness of less than 10 micrometers peak-to-peak. These technologies include a newly developed SiC-Al₂O₃-SiC package and a method to correct a warp of the chip surface.

A linear array thin film position sensitive detector (LTFPSD) based on hydrogenated amorphous silicon (a-Si:H) is proposed for the first time, taking advantage of the optical properties presented by a-Si:H devices we have developed a LTFPSD with 128 integrated elements able to be used in 3-D inspections/measurements. Each element consists on a one- dimensional LTFPSD, based on a p.i.n. diode produced in a conventional PECVD system, where the doped layers are coated with thin resistive layers to establish the required device equipotentials. By proper incorporation of the LTFPSD into an optical inspection camera it is possible to acquire information about an object/surface, through the optical cross-section method. The main advantages of this system, when compared with the conventional CCDs, are the low complexity of hardware and software used and that the information can be continuously processed (analogue detection).

As previously reported, the NCP retina is a programmable smart sensor in which images can be thresholded or halftoned and then processed in binary form by a micro-grained array processor. It is shown here that the observation of the current drawn by the NCP retina on its power supply can provide valuable global information on the observed scene. More generally, it yields an appealing solution to the generic output problem affecting artificial retinas.

This paper reports a new contact-type, full-color reading system for compact color facsimiles. This system mainly consists of a direct-contact type image sensor and bright 3-color LED arrays. As the sensor does not require an optical lens system like a rod lens, no color aberrations occur, making this system suitable for color reading. Recent advances in brightness for blue LED have made a 3-color solid state document illuminator possible. An LED illuminator is superior to a fluorescent lamp in uniform illumination, compactness, and longevity. Specifically two points of system structure have been researched. One is an optical system design based on the light transfer performances. The second is a 3-color illuminating method which can reproduce color accurately with high speed scanning. The main experimental results are: (1) reading speed per line is 15 msec which is acceptable for G3 class color facsimile, and (2) resolution is 0.5 at 8 line/mm of spatial frequency for RGB color bar- chart, (3) color difference is 12 which is slightly higher than that of the ordinary systems but can be easily improved through signal processing techniques. Therefore, the proposed reading system is possible for application in compact color facsimiles.

The digital aerial thermographic imagery system including line scanner in spectral band from 8 to 13 micrometers and personal computer with special software is presented. This system can be used in remote sensing (RS) investigations from both air and spaceborne platforms. The main advantages of this system are: the possibilities to register 1024 X 1024 pixels image and filter it in real time during the flight. Small size, reduced power and lower life cycle are critical decision elements for aerial vehicle environment. Different airborne RS investigations have shown the efficiency of the proposed system. 13

An amorphous silicon photodiode array was fabricated using a thin-film PECVD process. The 256 by 256 array contains 65,536 pixels on 200 micrometers centers and measures 5 cm on a side. A pixel configuration containing switching diodes instead of the more common TFT switch approach is used to address pixels. The resulting optical fill factor for each pixel is 66%. This device geometry can be scaled both to smaller pixels and large image sensor panels. Custom support electronics mounted on the sensor substrate using a chip-on-glass technique provide the row addressing and pixel charge readout. A scintillator film attaches to the array to convert x-rays to visible light. The basic sensor performance is evaluated under visible light without the scintillator. The device exhibits a dynamic range of 60 - 80 dB, low image lag and good uniformity. The low dark current allows integration times of up to one minute at room temperature without saturating the device. With the scintillator screen attached the device MTF is measured under x-ray illumination. Large changes in MTF are observed for two different types of commercial scintillator.

One of the drawbacks of using charge-coupled devices (CCDs) for scientific imaging is their relatively small size compared to many optical systems in which they are used. Telescopes, large format cameras, and medical imaging often require detectors much larger than the few cm dimensions of modern CCDs. One solution to this problem is to closely butt several CCDs together in the focal plane of the optical system, creating a focal plane mosaic. We have developed techniques to produce back illuminated CCDs from commercial front illuminated devices for enhanced quantum efficiency and spectral coverage. In this paper we discuss our development of packages and packaging techniques to butt back illuminated CCDs together, creating mosaics of up to 64 million pixels. We have discovered several critical issues during our development of back illuminated edge-buttable CCDs which we discuss in this paper. These include the development of proper chip carriers and packages, the ability to uniformly heat the devices in the required oxidation process, the ability to uniformly match antireflection coatings for all devices in a mosaic, and the development of alternative bonding methods which allow easy bonding of edge-buttable CCDs, especially as they approach whole wafer size.

Hydrogenated amorphous silicon (a-Si:H) is an electronically readable semiconductor material which can be inexpensively deposited over large surface areas. It is a technology under active development within the Xerox Corporation's Palo Alto Research Center (PARC) for a variety of possible applications. Our calculations indicate that, properly designed and properly applied, this material has great promise as a two-dimensionally sensitive electronic sensor for x-ray area detectors, useful in protein crystallography. The Argonne and Brandeis groups are currently making CCD area detectors, which are excellent but are expensive to build and have a much smaller area then we would like. We therefore have begun to develop large, inexpensive area detectors for protein crystallography, based upon a-Si:H sensors manufactured by Xerox Corporation. Our calculations suggest we can make them large, efficient, fast, high resolution, and with high dynamic range. These detectors should be much less expensive to manufacture than CCD-based detectors, their active areas should be comparable to or larger than image phosphor plate detectors, and they will be electronically readable directly into computer systems with speeds of 1 second or faster.

Fluorescent lumogen films are now widely used for improving the UV quantum efficiency of CCDs and other silicon photodetectors. Because of the organic nature of lumogen and its low melting and boiling points, stability of the films has been in question. We present results of stability tests in which quantum efficiency and film characteristics are evaluated with respect to exposure to illumination, elevated temperature, and reduced pressure. Our results indicate a high tolerance to UV and visible illumination, and to slightly elevated temperatures (95 degree(s)C) at normal operating pressures. However, high vacuum conditions (10^-6 torr) can produce voids in the films at even slightly elevated temperatures.

Low light level surveillance cameras with significantly higher performance and reduced form factor, than present state of the art are critical for many commercial and military applications. Towards this end, a new approach to low light level cameras was successfully demonstrated. In a cooperative research and development effort between Scientific Imaging Technologies, Inc. of Beaverton, Ore., and Intevac EO Sensors of Palo Alto, Calif., back-illuminated, electron-bombarded CCD (EBCCD) sensors were designed and fabricated. Experiments demonstrated the EBCCD's sensitivity and contrast resolution superior to conventional intensified CCD (ICCD) approaches. Low light level signal to noise (STN) and contrast transfer function (CTF) data are presented. A model is derived that describes the performance of the EBCCD and the back-illuminated CCD relative to conventional approaches to nighttime imaging. A design and simulated performance of a video rate 2/3 inch, back-illuminated, electron-bombarded CCD currently under development for low light imaging applications is also described.

We have designed, fabricated, and tested a modular CCD area detector system for macromolecular crystallography at synchrotron x-ray sources, code-named the `gold' detector system. The sensitive area of the detector is 150 mm X 150 mm, with 3,072 X 3,072 pixel sampling, resulting in roughly a 50 micrometers pixel raster. The x-ray image formed on the face of the detector is converted to visible light by a thin phosphor layer. This image is transferred optically to nine CCD sensors by nine square fiberoptic tapers (one for each CCD), arranged in a 3 X 3 array. Each taper demagnifies the image by a factor of approximately 2. Each CCD has a 1,024 X 1,024 pixel raster and is read out through two independent data channels. After each x-ray exposure period the x-ray shutter is closed and the electronic image is digitized (16-bit) and read out in 1.8s. Alteratively, the image may be binned 2 X 2 during readout, resulting in a 1,536 X 1,536 raster of 100 micrometers pixels; this image can be read out in 0.4s. The CCD sensors are operated at -40 degree(s)C to reduce electronic noise. The detector is operated under full computer control: all operational parameters (readout rates, CCD temperature, etc.) can be adjusted from the console. The image data (18 MByte/image) are transferred via a fast VME system to a control processor and ultimately to disk storage. During April 1994 we carried out a complete set of measurements at the Stanford Synchrotron Radiation Laboratory (SSRL) for a full characterization of the gold detector. Characterization includes quantitative evaluation of the instrument's conversion gain (signal level/x-ray photon); detective quantum efficiency (DQE); point-spread function; sensitivity as a function of x-ray energy; geometrical distortion of images; spatial uniformity; read noise; and dark image and dark image noise. Characterization parameters derived from these measurements show that this detector will be extraordinarily valuable for macromolecular crystallography.

The interaction process of laser and charge-coupled devices (CCD) having MOS structure has been analyzed in brief, and several kinds of damage thresholds have been put forward. On the basis of experimental test, the heat melting threshold, optical breakdown threshold, and direct damage threshold produced by a Q-switched Nd:YAG laser acted the CCD and laser energy threshold causing the whole device to fail have been obtained first.

A 2 micrometers BiCMOS process module has been developed for incorporation into existing high performance 2-phase CCD processes, to enable integration of digital and analog circuits on- chip with the CCD image sensor. The modular process architecture allows the integration of CMOS, NPN bipolar or BiCMOS circuits without affecting the baseline CCD characteristics. A design of experiments approach was employed using process and device simulation tools and selected physical experiments, to optimize CMOS and NPN device performance and process latitude. Both enhancement and depletion mode Poly-1 and Poly-2 CMOS devices were realized and demonstrated good long channel behavior down to 1.6 micrometers drawn. A 12 V, 2.5 GHz, low collector resistance NPN was also produced. Experimental process splits were used to demonstrate and verify that the CMOS and NPN process module incorporation did not affect the CCD device characteristics or yield. CMOS circuit performance was found to be comparable to that of a standard 2 micrometers CMOS process. Finally, a trilinear sensor with on-chip timing generation and correlated double sample was designed and fabricated. To our knowledge this is the first demonstration of high performance CCD, 2 micrometers CMOS, and an isolated vertical NPN, integrated on the same chip.

A 256 X 256 CMOS photo-gate active pixel image sensor is presented. The image sensor uses four MOS transistors within each pixel to buffer the photo-signal, enhance sensitivity, and suppress noise. The pixel size is 20 micrometers X 20 micrometers and was implemented in a standard digital 0.9 micrometers single-polysilicon, double-metal, n-well CMOS process; leading to 25% fill-factor. Row and column decoders and counters are monolithically integrated as well as per column analog signal correlated double-sampling (CDS) processors, yielding a total chip size of approximately 4.5 mm X 5.0 mm. The image sensor features random accessibility and can be employed for electronic panning applications. It is powered from a single 5.0 V source. At 5.0 V power supply, the video signal saturation level is approximately 1,200 mV with rms read-out noise level of approximately 300 (mu) V, yielding a dynamic range of 72 dB (12 bits). The read-out sensitivity is approximately 6.75 (mu) V per electron, indicating a read-out node capacitance of approximately 24 fF which is consistent with the extracted value. The measured dark current (at room temperature) is approximately 160 mV/s, equivalent to 3.3 nA/cm². The raw fixed pattern noise (exhibited as column-wise streaks) is approximately 20 mV (peak-to-peak) or approximately 1.67% of saturation level. At 15 frames per second, the power dissipation is approximately 75 mW.

Two types of image sensors are described that exploit the CCD's capability of spatially moving photocharge simultaneously with the exposure to a scene. The lock-in CCD is a two- dimensional array of pixels, each of which is a synchronous detector for oscillating optical wave fields with a spatially varying distribution of phase, amplitude and background offset. Main applications of this novel CCD type are in time-of-flight and heterodyne interferometric range imaging without moving parts. The convolver CCD consists of a two-dimensional array of pixels, connected with their nearest neighbors through short CCD lines, with which photocharge can be transferred vertically and horizontally during the exposure. By suitably timing these charge shifts, freely programmable convolutions with kernels of arbitrary size become possible. Tap weight accuracies of typically 2% of the largest tap value have been obtained for a variety of linear filters that are commonly used in machine vision. An integral part of these CCDs is a programmable, microcontroller-based driver system, capable of generating dynamic pulse sequences and driving virtually any image sensor available commercially or custom designed for special applications.

We report on the integration of computational and control functions with image sensor arrays using commodity CMOS processes. The sensors are highly efficient, high density arrays of photo-diodes, with a minimal overhead of one transistor per pixel. We report production arrays of up to 786 X 576 pixels of 10.5 micrometers pitch in 0.8 micrometers double metal single poly CMOS. We also report techniques presently in development to achieve 7.5 micrometers pitch in the same technology. These sensors are fully self-contained with regard to drive and sense electronics. Typical off-chip needs are limited to a clock crystal, supply regulator and some decoupling capacitors. `Intelligence' is provided in the form of analogue and digital functions integrated on the same chip as the sensors. Such functions can include 8- bit ADC, on-chip exposure control and computational functions to achieve color restoration. Integration of these parts allows us to construct a single chip color camera. Results from a proto-type color CMOS camera system are given. Successful realization of these functions leads to order of magnitude reductions in system cost, size and power-consumption when compared to the CCD alternative.

In this paper, a new CID image sensor architecture, pre-amplifier per pixel (PPP), is presented. In this architecture, an amplifier is dedicated to and integrated within each pixel of the image sensor array. The read-out node for a pixel design employing this architecture is local to its pixel and buffered from any other read-out nodes within the image sensor pixel array. Thus, the capacitance of the read-out node in this design is significantly reduced, resulting in significant improvements of read-out sensitivity and noise. This architecture reduces the read-out node capacitance to an order of magnitude of few tens of fF, which yields an rms read-out noise in the order of few tens of electrons. This brings the CID imaging technology to the main stream of read-out noise levels of other solid-state imaging technologies. This is achieved without compromising the traditional strengths of CID imaging technology such as non-destructive read-out, resistance to blooming, radiation hardness, and random-accessibility. The only compromise is the pixel fill-factor. Depending on the minimum process feature size and the pixel size, the pixel fill-factor can be higher than 50%. A test chip has been designed and fabricated. Evaluation of this test chip is presented.

The design of an adaptive sensitivity CCD image sensor is described. The sensitivity of each pixel is individually controlled (by changing its exposure time) to assure that it is operating in the linear radar range of the CCD response, and not in the cut-off or saturation regions. Thus, even though an individual CCD sensor is limited in its dynamic range. The idea is based on the biological retina principle of operation, enabling the imager to capture details in both the light and dark areas of high-contrast scenes, as in the human eye.

The theory of new imaging processes is proposed based on quenching or gain effects of visible secondary radiation in perfect crystals (e.g. CdS by incident mid IR or far IR radiation). Three types of the transformation processes from IR into visible range are considered both theoretically and experimentally. The first is based on IR-quenching effect for Raman light scattering (RLS) due to collisions of free and bound Wannier-Mott excitons at helium temperatures. To use this effect for imager design it must have some advantages compared with the usual thermovision systems: (1) highest sensitivity, (2) great protection against background IR radiation, (3) the possibility to design far IR TV photocathodes, (4) highest spatial resolution, that is near the same as one restricted by the number of fibers in line served for the transferring of visible image from the crystal immersed in liquid helium to the CCD array beyond the cryostat; (5) low heating of liquid He (power of UV excitation of the crystal is less than 10 mW). The second one is based on IR-quenching effect for probe visible radiation in the crystal relative transparency region at room temperature with participation of spherical excitons. The third one is based on the stimulated gain due to two-photon emission connected with interactions of Wannier-Mott excitons with spherical excitons. In the process the gain effect must be observed for both submillimeter waves with large angular momenta and visible radiation at the frequency of resonant direct transition from A(n equals 1) and B(n equals 1) states of Wannier-Mott excitons.

A 128 X 128 element CMOS active pixel image sensor (APS) with on-chip timing, control, and signal chain electronics has been designed, fabricated and tested. The chip is implemented in 1.2 micrometers n-well process with a 19.2 micrometers pixel pitch. The sensor uses a photodiode-type CMOS APS pixel with in-pixel source follower, row selection and reset transistors. The sensor operates from a +5 V supply and requires only a clock signal to produce video output. The chip performs correlated double sampling (CDS) to suppress pixel fixed pattern noise, and double delta sampling (DDS) to suppress column fixed pattern noise. The on-chip control circuitry allows asynchronous control of an inter frame delay to adjust pixel integration. On-chip control is also provided to select the readout of any size window of interest.