Operational Requirements of CMOS Image Sensors

In different applications of CMOS imaging systems, there exist certain specific requirements for the design of CMOS image sensors. In general, the requirements involve quantum efficiency, conversion gain, saturation level, image lag, smear, crosstalk, anti-blooming, noise, dynamic range, readout rate, array size, and pixel pitch. Some general discussions of these requirements are summarized below.

1) Quantum Efficiency : Quantum efficiency is the probability that a photon incident on the pixel creates a hole or electron that is collected in the pixel’s potential well. Quantum efficiency depends on several parameters, including the reflection coefficient of the optical stack on the pixel, the fraction of the pixel area which is light-shielded, the fraction of the pixel area where photocharge goes elsewhere than the photosite, the photon wavelength, the bulk and surface carrier lifetimes, the depth of the potential well, and the volume and shape of the depletion regions. It is useful to think of quantum efficiency as the product of optical aperture, which is the fraction of pixel area which is optically active, optical stack transmission, which is the fraction of photons incident on optically active areas which reach the silicon surface, and collection efficiency, which is the fraction of photogenerated electrons that get collected.

2) Conversion Gain : The image data starts as charge and is usually converted to voltage. For an imager with a linear charge-to-voltage characteristic, the ratio of output signal to collected charge is a constant, referred to as the conversion gain. It is usually expressed in units of µV/e⁻ and is equal to the reciprocal of the charge sense capacitance, times any system gain.

3) Saturation Level : Pixel saturation occurs when the pixel stores the maximum amount of charge. If more charge is added, it simply spills into the substrate or adjacent pixels. It is usually expressed in terms of electrons, although it is sometimes referred to the output as a voltage. It can also be stated in terms of illumination required to produce the saturation charge. Depending on design, the output circuit can saturate before the pixel saturates. In this case, the saturation level is set by the output circuit.

4) Image Lag : Image lag refers to the persistence of one frame in successive frames. An imager demonstrates lag when, for example, a strong light is suddenly

turned off but the imager output does not immediately return to the black level. Lag is usually caused by incomplete charge reset. It is informally characterized by whether or not it is visible on a video display.

5) Smear : In a CMOS imager, a charge packet passing by a region of high illumination accumulates additional charge due to optical crosstalk, creating vertical stripes in the reconstructed image. A column line can easily collect stray charge from a highly illuminated pixel, corrupting the readout of all pixels in that column.

6) Crosstalk : Crosstalk is any contamination of one pixel’s signal by another pixel’s signal. Smear is an example of optical crosstalk. Examples of electrical crosstalk include ground bounce and capacitive coupling.

7) Anti-Blooming : Blooming is the spread of charge from saturated pixels into surrounding potential wells. It differs from optical crosstalk in that all photogenerated charge, not just a small fraction, spreads out into the surrounding pixels/shift registers.

Most pixels contain some means for shunting excess charge to a drain in order to preserve the information in nearby non-saturated pixels. Anti-blooming characterizes the effectiveness of the technique used to drain excess charge. It is expressed as the ratio of the illumination required to produce blooming to the illumination required to saturate the pixel.

8) Noise : Noise in the imager sensor can be separated typically into two categories, random noise and pattern noise. Random noise varied temporally and is not constant from frame to frame in the imager. Pattern noise is divided into two components, one is fixed pattern noise (FPN) and the other is the photo-response nonuniformity (PRNU). The FPN comes from dimensions, doping concentration, and contamination of photo-detectors and the characteristics of threshold voltage, width, and length in MOSFETS. The PRNU noise comes from the thickness of layers on the top of photo-detectors and wavelength of illumination. These noise in the CMOS

imager sensor are briefly discussed below.

(i) Random Noise

An imager with a constant scene should produce identical output from frame to frame. In practice, the output from a given pixel will vary over time due to thermal noise, charge trapping, and 1/f noise in the devices which comprise the imager. Photonic shot noise is usually not included in this quantity, although this also contributes to noise at the output. Random noise is typically stated in terms of input-referred equivalent electrons, i.e., the root mean square (rms) output voltage noise divided by the conversion gain.

(ii) Fixed-Pattern Noise

Fixed-pattern noise (FPN) is the fixed (constant in time) variation between pixel outputs under spatially uniform illumination. Fixed-pattern noise is typically due to random or mask-induced mismatches in device parameters such as threshold voltage, trap density, and parasitic capacitance.

FPN is usually a function of illumination, and can be written as the sum of a gain term and an offset term for an imager with a linear response characteristic. Offset FPN is constant over illumination, and gain FPN is proportional to illumination.

FPN consists of components that describe variation between columns, and variation between pixels in a single column. Column FPN is the standard deviation of the column-average pixel output values in a time-average, uniformly illuminated frame. The column FPN is expressed as

1

pixel value in the frame. Since a column FPN calculation requires multiple columns, j > 1. Pixel FPN is the standard deviation of pixel output values after column FPN has been removed. In order to calculate pixel FPN, multiple pixels are required. The pixel FPN is expressed as

1

If the diffusion of the photodiode is reset through a MOSFET, this is equivalent to a capacitance being charged through the resistance of the MOSFET channel. The rms (root-mean-square) noise voltage can be expressed as

C

V_rms = kT (2.3)

where k is the Boltzmann constant, T is temperature, and C is the capacitance of photodiode. The reset noise is generally called “KTC” noise.

KTC noise can only be canceled by using the photogate-type active pixel sensor (APS). Currently, reset noise limits the read noise in photodiode-type APS [22].

(iv) Thermal Noise

Thermal noise is a white noise which means the noise power is constant over all frequencies. For a resistor, the thermal noise rms voltage can be expressed as

kTBR 4

V_rms = (2.4) where B is the noise bandwidth and R is the resistor. Since the thermal noise covers the entire frequency range, the bandwidth determines the actual amount measured.

(v) Shot Noise

Shot noise is another white noise that arises from the discrete nature of the electrons, for example, the random arrival of particles of charge. This is the result of the random generation of carriers such as thermal generation within a depletion region (i.e. shot noise of dark current) or the random generation of photon-electrons.

(vi) Flicker (1/f) Noise

The flicker noise occurs at any junction, including metal-to-metal, metal-to-semiconductor, semiconductor-to-semiconductor, and conductivity fluctuations. The flicker noise arises mainly in amplifier circuits where there are numerous such contacts. At low frequency, flicker noise can be the dominant component, but it drops below thermal noise at higher frequency.

9) Dynamic Range : The dynamic range is defined as the ratio of maximum charge capacity to noise floor. The required dynamic range of CMOS imagers is determined by the ratio of the brightest signal level to the weakest. Larger dynamic range is preferred but limited by storage capacitance, linearity, and noise level.

10) Readout Rate : The readout rate is chosen according to the specific imaging system requirements and limited by the allowable chip power dissipation as well as the circuit operation speed. Usually a higher readout rate is needed for multiple sampling applications in image compensation function. Higher readout rate is also

needed to avoid the saturation of the signal after integration.

11) Array Size and Pixel Pitch : Higher image resolution requires larger array size and smaller pixel pitch. However, a larger pixel size is needed to increase the fill factor, photo-senstivity, and dynamic range. Thus the optimal design trade-off should be made between the application flexibility and resolution performance.

It is important to determine the operational requirements in the design of CMOS imager for specific applications. A complete analysis of operational parameters like noise, spectral response, sensitivity, and resolution should be set before the design trade-off. Therefore, all the operational requirements discussed above have unique optimized orientations for CMOS imagers in different applications.