At the high end of the digital still-imaging market, a few companies are developing high-resolution CCD sensors with complexities of 14 million to over 81 million pixels. These sensors serve in applications such as professional still-image cameras, medical imaging systems, and space telescopes. Players in this market include Dalsa, Fairchild Imaging, Hamamatsu, Kodak, and Sarnoff Labs. All of them continue to rely on CCDs for their high-resolution and high-speed imaging devices.
Many sensors have the pixels accessible on the chip's surface, with the metallization arranged for the light to reach the pixels. But the layers of metallization, polysilicon, silicon dioxide, and other materials still can block a significant amount of the light the sensor must capture, especially in low-light applications.
Instead of using the sensor's surface, designers can remove the silicon on the bottom of the chip and have the light source shine on that thinned area. Such bottom-thinned sensors are used in many telescopes to collect every possible photon. In addition to thinning the chips, further steps can improve sensitivity and signal-to-noise ratios. Cooling the chips to near liquid-nitrogen temperatures can minimize the electrical noise caused by thermal activity. In addition, the lower temperatures reduce the dark current, improving sensitivity.
Some of the largest commercial imaging sensors include Fairchild Imaging's 81-Mpixel CCD595, Kodak's 39- and 31.6-Mpixel KAF39000 and KAF31600, Dalsa's 28-Mpixel sensor, and Atmel's 16-Mpixel chips.
The CCD595 full-frame sensor, which contains a square array of 9216 by 9216 pixels, employs a pixel-pitch of 8.75 mm. That large pixel size pushes the total imaging array area to almost 81 by 81 mm (just over 3 in. a side). Four serial readout registers and a two-phase clocking scheme move the image data out of the large array. Potential applications include scientific imaging, space exploration, and aerial reconnaissance.
Replacing 35-mm Film
Designed for professional digital still-image camera backs, the Kodak KAF-39000 and 31600 deliver some of the highest-resolution images to date?resolution comparable to, if not better than, 35-mm film. A new 6.8-mm CCD pixel enables higher sensor resolutions while retaining the performance professional photographers require for sensitivity, dynamic range, and color fidelity.
With 39 million imaging pixels in a 36- by 48-mm imaging area, the KAF-39000 raises the bar for image resolution in professional photography. The 31.6 million-pixel KAF-31600 deploys this same high-performance pixel architecture in a smaller, 33- by 44-mm imaging area that's compatible with the Advanced Photographic Standard (APS).
These image sensors join three other Kodak image sensors in the medium-format market. The KAF-16802 provides 16 Mpixels in a square optical configuration. The 22-Mpixel KAF-22000 set the previous high-resolution benchmark for this market. And, the KAF-18000 18-Mpixel image sensor delivers a unique combination of sensitivity, resolution, and frame rate.
The Dalsa 28-Mpixel full-frame CCD sensor also fits professional digital still-imaging applications. It has an optical format of 44 by 33 mm and a pixel size of 7.2 mm square. The chip's charge-binning scheme can combine 2 by 2 blocks of pixels to provide higher sensitivity (4X), better signal-to-noise ratios (2X), and higher frame rates (3X to 4X) for applications that don't need the full resolution or higher frame rates.
To implement binning, designers had to integrate more circuitry to deal with adding the color pixels in the proper sequence. To manage the large frame size, designers stitched mask sections together. At the same time, they overcame the yield degradation issues of these large chips as well as the large resistance-capacitance time constants of long cross-chip interconnects. Organized as 6096 by 4560 active pixels, the imaging array features a dynamic range of 70 dB and a dark current of just 0.1 nA /cm 2(at 60ºC).
Atmel's 16-Mpixel AT71201 offers a full-frame 4096-by-4096-pixel array that also can operate in a 4/3 aspect ratio (4096 by 3072 pixels) for video and a 2/1 aspect ratio (4096 by 2048 pixels) for industrial applications. The large 11- by 11-mm pixel regions provide high sensitivity, but they also can be binned into clusters of 2 by 2 or 4 by 4 pixels to further improve sensitivity at the expense of resolution. With a dynamic range of 72 dB, minimum, the sensor offers a responsivity of 3.8 V/(mJ/ cm 2 ).
The imaging array is divided into four independent zones, each with its own thermometer to keep track of thermal noise. An independent four-phase clock drives each zone. The independent clocks and zones give designers lots of system flexibility to match the sensor to their application. To read data out quickly, two serial registers and four independent output amplifiers permit the sensor to deliver up to 7.4 frames/s.
A proprietary 12-Mpixel sensor lies at the heart of a camera developed by Fujifilm (Fuji Photo Film U.S.A. Inc.). The FinePix S7000 camera offers 6.3 million effective pixels, which can produce images of 4048 by 3040 recorded pixels at a maximum rate of 5 frames/s. Fujifilm claims its Super CCD HR sensor, a second-generation device, will deliver best-in-class color images with rich, vibrant colors. The CCD sensor also can capture VGA-resolution videos