Semiconductor device manufacturing: process – Making device or circuit responsive to nonelectrical signal – Responsive to electromagnetic radiation
Reexamination Certificate
2000-06-14
2002-05-28
Nelms, David (Department: 2818)
Semiconductor device manufacturing: process
Making device or circuit responsive to nonelectrical signal
Responsive to electromagnetic radiation
C438S065000
Reexamination Certificate
active
06395576
ABSTRACT:
BACKGROUND OF THE INVENTION
(1) Field of the Invention
The present invention relates to a highly efficient and adaptive microelectronic fabrication process for adjusting color filter formation to optimize pixel color sensitivity balance and to control the red:green:blue gain-ratio in optoelectronic semiconductor array imaging devices.
(2) Description of Prior Art
Synthetic reconstruction of color images in solid-state analog or digital video cameras is conventionally performed through a combination of an array of optical microlens and spectral filter structures and integrated circuit amplifier automatic gain control operations following a prescribed sequence of calibrations in an algorithm.
Typically solid-state color cameras are comprised of charge-coupled device (CCD), Charge-Injection Device (CID), or Complementary Metal-Oxide Semiconductor (CMOS) structures with planar arrays of microlenses and primary color filters mutually aligned to an area array of photodiodes patterned onto a semiconductor substrate. The principal challenge in the design of solid-state color camera devices is the trade-off between adding complexity and steps to the microelectronic fabrication process wherein color filters are integrally formed in the semiconductor cross-sectional structure versus adding complexity and integrated electronic circuitry for conversion of the optical analog signals into digital form and signal processing with color-specific automated gain-control amplifiers requiring gain-ratio balance. The trade-off between microelectronic fabrication process complexity versus electronic complexity is determined by a plurality of factors, including product manufacturing cost and optoelectronic performance.
The problem in semiconductor array devices for color imaging fundamentally arises because different color pixels in the matrix exhibit varying spectral sensitivity to the different wavelengths or frequencies contained in the incident image light. For example, a photodiode element or sensor in the matrix array that is more sensitive to red light than blue light creates an imbalance in the captured image.
The imbalance can be corrected either by precompensation in the color-filter array, achieved by designing the fabrication process adaptively, or, digitally after the analog-to-digital conversion (ADC) step. It is well known, however, that post-processing signals after ADC is too late to avoid adding undesired quantization noise. Electrical signal precompensation to correct the color sensitivity imbalance, analogous to the aforementioned alternative color-filter process approach, is possible, and, is performed before ADC by adding to each color-pixel output signal a signal quantity derived from a color-specific gain controlled amplifier and additional circuitry for the control of the red:green:blue gain-ratio.
Typical CMOS image sensors incorporate more functional integration than do CCD sensors, which are manufactured by a specialized process, thereby making it a challenge to add image-processing circuitry to the chips. In contrast, CMOS sensors are made with the same high-volume processes used to build most computer chips, so digital circuitry can be added to enhance sensor functionality. The CMOS sensor chips typically integrate pixel array, timing logic, sampling circuits, amplifiers, reference voltage supplies, and ADC's. The CCD sensor chips require a minimum of two support chips to accomplish the same functions as CMOS. The increased integration offered by CMOS sensors can reduce system complexity and allow smaller camera designs. But, more circuits on CMOS chips increase the potential for noise from one section of the chip to interfere with the operations of another section.
A frequent problem is the noise generated by the digital section can interfere with the highly sensitive front-end analog circuits and degrade image quality. In either case of CCD or CMOS devices, it is clearly seen that the preferred method for improving color balance in semiconductor array imaging devices is the analog optical filter, with attendant simplicity of circuitry and avoidance of the electrical noise interference introduced by added circuitry needed to execute color-compensation functions.
Color-photosensitive integrated circuits require carefully configured color filters to be deposited on the upper layers of a semiconductor device in order to accurately translate a visual image into its color components. Conventional configurations may generate a color pixel by employing four adjacent pixels on an image sensor. Each of the four pixels is covered by a different color filter selected from the group of red, blue and two green pixels, thereby exposing each monochromatic pixel to only one of the three basic colors. Simple algorithms are subsequently applied to merge the inputs from the three monochromatic pixels to form one full color pixel. The color filter deposition process and its relationship to the microlens array formation process determine the production cycle-time, test-time, yield, and ultimate manufacturing cost. It is an object of the present invention to teach color-filter processes which optimize these stated factors.
In addition to the dynamic range and noise of the individual photodetectors, the resolution or fidelity of a CCD or CMOS image is influenced by the overall array size, individual pixel size, spacing, and fill factor. Quantitatively, this performance is described by a wavelength-dependent modulation transfer function (MTF) that relates the two-dimensional Fourier transform of the input image to that of the output. In addition to the obvious effects of pixel aperture and shape, the MTF of an array is affected by spatial carrier diffusion, temporal diffusion, and optical diffraction.
While color image formation may be accomplished by recording appropriately filtered images using three separate arrays, such systems tend to be large and costly. Cameras in which a full color image is generated by a single detector array offer significant improvements in size and cost but have inferior spatial resolution. Single-chip color arrays typically use color filters that are aligned with individual columns of photodetector elements to generate a color video signal. In a typical stripe configuration, green filters are used on every other column with the intermediate columns alternatively selected for red or blue recording. To generate a color video signal using an array of this type, intensity information from the green columns is interpolated to produce green data at the red and blue locations. This information is then used to calculate a red-minus-green signal from red-filtered columns and a blue-minus-green signal from the blue ones.
Complete red-minus-green and blue-minus-green images are subsequently interpolated from this data yielding three complete images. Commercial camcorders use a process similar to this to generate a color image but typically utilize more complicated mosaic-filter designs. The use of alternate columns to yield color information decreases the spatial resolution in the final image.
FIG. 1
exhibits the conventional Prior Art vertical semiconductor cross-sectional profile and optical configuration for color image formation. Microlens
1
residing on a planarization layer
2
which serves as a spacer collects a bundle of light rays from the image presented to the video camera and converges the light into focal cone
3
onto photodiode
8
after passing through color filter(s)
4
residing on planarization layer
5
, passivation layer
6
, and metallization layer
7
.
FIG. 2
illustrates a representative Prior Art example for the generation of a color image by a single photodetector array
9
by using a color filter mask comprised of green stripe
10
, red stripe
11
, green stripe
10
, blue stripe
12
, green stripe
10
, and red stripe
11
. In this scheme, green filters are placed over alternate photodetector columns. Red and blue filters alternate in the spaces between them. Interpolation routines are used to generate three-color data for all pixel positio
Chang Bii-Junq
Chang Chih-Kung
Hsiao Yu-Kung
Pan Sheng-Liang
Ackerman Stephen B.
Dang Phuc T.
Nelms David
Saile George O.
Stanton Stephen G.
LandOfFree
High efficiency color filter process to improve color... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High efficiency color filter process to improve color..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High efficiency color filter process to improve color... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2893774