|Fig. 1: pixels in a CMOS pixel image sensor.|
Complementary metal oxide semiconductor (CMOS) image sensors are typical devices to convert light signals to electric signals. Because of its low power consumption, high read-out speed and high integration (which means small size), CMOS image sensors become a main part of phone cameras. [1,2] Some famous camera phones, such as iphone 4 and HTC Evo 4G, use CMOS sensor as the image sensor of their cameras. 
CMOS image sensors can be divided into two groups: passive pixel sensors and active pixel sensors.  Passive pixel sensors are very noisy since there are no amplifiers in the pixels; the electric signals that they generated only come from the power of the incident light. So now people often use active pixel sensors in phone cameras to avoid noise. With CMOS transistors as amplifiers, active pixel sensors perform better in obtaining high-quality images.
A CMOS active pixel image sensor consists of many pixels, and in phone camera case, each pixel is covered by red, green and blue light filters (Fig. 1). Photodiodes are set under the filters to detect light signals, and some circuits with CMOS transistors would amplify and readout the electric signal generated by the photodiodes (Fig. 2). Fig. 3 is a simple 3-transistor (3T) pixel circuit that can show how it works. Basically each pixel needs a controller and voltage source for resetting (RST and VRST), a voltage source for amplification (VDD), a row controller (ROW) and a column detector (COL). It can detect light as follows:
|Fig. 2: structure of a pixel.|
Keep the whole device in darkness so the photodiode is off, only an ignorable dark current can flow through it.
Turn on and off RST (reset controller) to make the voltage of the upper end of the photodiode to be VRST, which means no signal (reset).
Use the sensor to detect light.
A. If no light is shining on the pixel, Msf keeps on, or
B. If light is detected, part of the charge accumulated on the upper end of the photodiode would leak to the ground (light current) and make the voltage decrease. Depending on the intensity of the incident light, the voltage of the upper end of the photodiode becomes different, then the resistance of Msf becomes different.
Turn on ROW (row controller), and detect the current on COL (column detector). Different resistance of Msf gives different current, which indicates different light power.
Now more and more transistors are used in one pixel to reduce the noise, which is the main problem of CMOS image sensor system.  5-transistor (5T) pixels and 6-transistor (6T) pixels are proposed and 4-transistor pixels are widely used now. [5-7]
Compare with its competitor Charge-Coupled Device (CCD) image sensor, the read-out of the data on a CMOS image sensor can be customized arbitrarily (so called "random access"). Just as above, the information on each pixel of a CMOS image sensor can be detected separately, while CCD sensors can only obtain the data of the whole frame.  So if people are only interested in a small region of the image, CMOS image sensor is more efficient, which means costing less time and consuming less energy, than its competitor.
|Fig. 3:A simple 3-transistor (3T) pixel. (Source: Wikimedia Commons).|
The capability that every pixel of CMOS image sensors can be read out separately also gives it a high read-out speed. It is very easy for CMOS sensors to achieve several hundreds of Mega frames per second (Mfps) or even several Giga frames per second (Gfps) for a single pixel.  For a whole CMOS image sensor, 240 frames/s is achieved, where a sensor with an array of 2352 × 1728 pixels (4.1-Megapixel) is used.  CMOS image sensors also have very fast response: they can be used to detect fast modulated light, up to 1 MHz modulation. 
According to a recent report, ultra-low power consumption - 84 pW/Frame per pixel can be achieved.  This design can even be used to make solar cells since 4.85 μW energy can be harvest back after shining light.  So if a phone camera with 8 Megapixels uses this technology, the power it uses for obtaining a frame of image should be around 0.7 mW, that can definitely save energy and keep the camera working for a longer time.
In principle, CMOS image sensor is just an integrated circuit chip, so it can be easily made by mature semiconductor technologies, such as lithography. [1,4] Now people can make very small pixels, typically several μm's diameter. [2,9,11] The size of the pixels in some phone cameras can also be estimated. For example, iphone 4S camera has a CMOS sensor of 1/3.2" size and 8-Megapixels, which is 4.536 mm × 3.416 mm for 8 million pixels.  So the size of one pixel is about
which is comparable with the size in Refs. [2,9,11].
As early as 1995, CMOS image sensors are predicted to be the basis of on-chip cameras.  Nowadays, they are used to detect light as portable devices, such as camera phones and sonething like terahertz electromagnetic wave detectors.  It can be expected that CMOS image sensors would still be widely used in the future because they are cheap, fast and small.
© Di Lu. The author grants permission to copy, distribute and display this work in unaltered form, with attribution to the author, for noncommercial purposes only. All other rights, including commercial rights, are reserved to the author.
 D. Litwiller, "CMOS vs. CCD, Maturing Technologies, Maturing Markets," Photonics Spectra, 39, 54, (2005).
 H. P. Wong et al., "CMOS Active Pixel Image Sensors Fabricated Using a 1.8-V, 0.25-μm CMOS Technology", IEEE Trans. Electron Devices 45, 889, (1998).
 J. Goldman, "Why the iPhone 4 Takes Good Low-Light Photos: BSI CMOS Sensors Explained!," CNET News, 2 Feb 11.
 E. R. Fossum, "CMOS lmage Sensors: Electronic Camera on A Chip", IEEE Trans. Electron Devices, 44, 1689, (1997).
 T. Tsai and R. Hornsey, "A Two-Step Readout CMOS Image Sensor Active Pixel Architecture", Proc. IEEE Sensors 2011, 1941 (2011).
 F. Tang and A. Bermak, "An 84 pW/Frame Per Pixel Current-Mode CMOS Image Sensor With Energy Harvesting Capability", IEEE Sensors J. 12, 720 (2012).
 P. Vu et al.,"Design of Prototype Scientific CMOS Image Sensors", Proc. SPIE, 7021, 702103 (2008).
 M. El-Desouki et al., "CMOS Image Sensors for High Speed Applications," Sensors 9, 430 (2009).
 A. I. Krymski et al., "A High-Speed, 240-Frames/s, 4.1-Mpixel CMOS Sensor", IEEE Trans. Electron Devices 50, 130 (2003).
 N. S. Johnston et al., "2D CMOS Image Sensors For the Rapid Aquisition of Modulated Light and Multi-Parametric Images," Proc. SPIE, 8073, 807303 (2011).
 E. A. G. Webster et al., "A Single-Photon Avalanche Diode in 90-nm CMOS Imaging Technology With 44% Photon Detection Efficiency at 690 nm", IEEE Electron Dev. Lett. 33, 694 (2012).
 J. Goldman, "The Secret Behind Nokia's 41-Megapixel Camera Phone," CNET, 28 Feb 12.
 H. Sherry et al., "A 1kpixel CMOS Camera Chip For 25fps Real-Time Terahertz Imaging Applications." 2012 IEEE International Solid-State Circuits Conference ISSCC, 252 (2012).