Future Potential of Image Sensor

Future market potential of consumer-use cameras has focused mainly on compact equipment that combines portability with the high resolution needed to capture subjects in fine detail. Developers have worked to meet such expectations by reducing pixel size while maintaining the photographic characteristics of image sensors. However, in addition to continuing demand for compact, high-resolution cameras, the past few years have also brought an increasing emphasis on improvements in other areas, such as minimum subject lighting and high-speed photography. As a result, there is now increasing demand for a general improvement in the definition provided by image sensors, including signal-to-noise ratios.

Issues Surrounding Conventional CMOS Sensors

Pixels in CMOS sensors consist of photodiodes formed on the silicon substrate and on-chip lenses formed above the photodiodes to collect light. When an incident light beam passes through an on-chip lens and reaches a photodiode, photoelectric conversion turns the light into electronic data, which is output as a voltage. This process allows light to be collected in the form of electrical signals.

Conventional CMOS sensors have a multi-layered structure that includes a layer of transistors and metal wiring between on-chip lenses and photodiodes. These are used for voltage input and output. As a result of this structure, incident light may bounce back if it strikes metal wiring in the light path, or it may be bent by refraction at the boundary of the insulating film layer. Therefore, the transistor and metal wiring can prevent light (gathered by the on-chip lenses) from reaching the photodiodes efficiently. When the number of pixels is increased to enhance definition performance, the area per pixel is reduced. If the metal wiring is increased to create a high-speed drive system, the amount of light collected by the photodiodes is reduced, causing further deterioration in the light collection ratio.

"Exmor R"---Back-illuminated CMOS Sensor

Characteristics of CMOS sensors include low power consumption and high speed. While maximizing these advantages, Sony has eliminated the aforementioned problems associated with conventional front-illuminated CMOS sensors by developing breakthrough technology. In contrast with the conventional CMOS pixel structure, Sony's back-illuminated CMOS sensor is exposed to light from the back of the silicon substrate. The result is a dramatic improvement in photographic performance, including approximately double the sensitivity ^(*1) and a reduction in noise. Sony has succeeded in the development of a CMOS sensor with 1.75µm square pixels, a resolution of 5 megapixels (effective), and a speed of 60 frames per second. Sony has now commenced mass-production of this new chip, the "Exmor R" back-illuminated CMOS sensor.

In a back-illuminated sensor, the silicon substrate is reversed to allow exposure to light on the reverse side. Because the light is not blocked by the wiring and transistor layer, comparatively more light enters each pixel and there is no loss of sensitivity caused by changes in the angle of incidence. However, the structure of a back-illuminated sensor and the manufacturing processes can result in a number of problems that can reduce the resolution and signal-to-noise ratio of the image sensor. These include noise, dark current, defective pixels and color mixing.

Sony overcame these problems by developing a new photodiode structure and on-chip lens optimized for this back-illuminated CMOS sensor. These efforts increased the signal-to-noise ratio by 8dB compared with earlier Sony CMOS sensors with pixel structures based on pixels of the same size. Sensitivity has been improved by 6dB ^(*1), while noise, dark current, and defective pixels have all been reduced. This is reflected in a 2dB reduction in random noise under dark conditions. The problem of color mixing was overcome using high-precision superposition technology.