CMOS Imagers with Embedded CCD

Sensors Insights by Piet De Moor

The last decade revealed exciting advances in image sensor technologies. Integration, speed and low-power advantages offered by the introduction of CMOS technologies, coupled with highly customized photo-sensor fabrication processes, resulted in CMOS based imagers that are superior to CCDs for high-volume consumer applications. While much smaller in production volume, imagers for scientific and industrial applications also benefited from these developments. As a result, CMOS imagers are gradually replacing CCDs in almost all applications.

In 2011, CCDs made up only 8% of all image sensor shipments with less than 200 million parts. CCD unit shipments will likely continue to decline, however, technologies for consumer imagers do not answer all needs of highly specialized custom sensors. Large-area scientific applications such as biomedical, fluorescence microscopy, x-ray crystallography imagers, etc., demand extremely low noise, dark-current, non-uniformity, and high quantum efficiency.

For such niche applications, CCDs are still the device of choice. While in machine vision applications, e.g. inspection, requiring low noise, global shutter mode of operation and high shutter rejection ratio interline CCDs are competitive to CMOS alternatives.

For such applications, custom design (specialized pixels, high performance readout circuits, and chip architectures) and custom silicon processing (optimized devices, implants, and packaging) is necessary. As one example, Imec answers this need by offering custom technology and design of high-end imagers ranging from R&D, development-on-demand to low-volume manufacturing, and transfer to a foundry for high-volume production.

Imec developed a CMOS compatible CCD module and included this into its custom imager 130-nm process. The integration of a CCD module enables monolithic integration of CCD pixel structures and standard CMOS readout in one device. CCDs traditionally utilize thick oxides, multiple poly-silicon layers, and voltage levels not compatible with typical CMOS. These needs are not in line with CMOS trends, aiming at lower supply voltages and thinner oxides.

Imec implemented a number of process modifications to realize a CMOS compatible CCD module in its 130-nm custom image sensor technology platform. The embedded CCD module uses a single layer poly-silicon CCD electrode with narrow gaps to allow efficient charge transfer. It is customizable depending on application needs, and is compatible with back-side illumination.

One of the major advantages of CMOS over CCD is the system integration capability of CMOS technology. A classical CCD requires separate readout electronics containing e.g. analog-to-digital convertors. In CMOS, a large part of the total readout system can be integrated on-chip, leading to a camera-on-a-chip or system-on-a-chip (SOC) imager. This reduces the system complexity and hence the total system cost.

The biggest advantage offered by the embedded CCD technology that we propose is integration. With embedded CCD, CCD-based pixel designs can be combined with on-chip readout electronics, such as clock drivers, charge to voltage converters, and digitizers, all on the same die.

So-called time-delay-integration (TDI) applications (aerospace push-broom imagers, wafer-inspection etc.) primarily use CCDs. These inherently linear devices use a clever synchronization of the linear motion of the scene with multiple sampling of the same image, thereby increasing the signal to noise ratio. Since CCDs operate in the charge domain that allows moving charges without creating excess noise, this technology fits extremely well to the TDI application. To replicate a similar performance in aTDI application, a CMOS active pixel imager must have sub-electron read noise, which is not the case today.

Embedded CCD, being a CCD compatible with CMOS operating voltages, opens possibilities for new and rich pixel architectures. For example, users could utilize an in-pixel CCD to implement a charge-domain global shutter functionality. Although this concept to image a full frame at the same exact timing is possible using CMOS imagers, it generally comes with a noise penalty.

Other possible uses of embedded CCD pixels include extreme high-frame rate burst imagers, where subsequent frames are stored inside the pixel using small CCD based analog memories. Certain time-of-flight implementations could also benefit from this technology. In addition to monolithic TDI applications for industrial inspection, TDI functionality is also used in hybrid infrared and x-ray imagers.

About the Author
After obtaining his PhD in Physics from Leuven University, Piet De Moor joined imec in 1998. First he was involved in research and development of MEMS and MEMS packaging as well as 3D integration technology. Later on he was leading a team developing advanced CMOS imagers such as back-side illuminated and embedded CCD in CMOS imagers suitable for high-end imaging applications. Currently he is coordinating the space activities in imec.

Read more on

Suggested Articles

Nvidia GPUs also behind new Azure NDv2 cloud process

Intel also launched oneAPI initiative to reduce complexity for developers

Logitech and Proxy are early customers of Memfault