Browse Prior Art Database

Shape Correction For Optimal Color Filter Array Mosaics for Color Image Sensors Disclosure Number: IPCOM000125526D
Original Publication Date: 2005-Jun-06
Included in the Prior Art Database: 2005-Jun-06
Document File: 3 page(s) / 34K

Publishing Venue



Disclosed is a method and process of shape correction for optimal color filter array (CFA) mosaics in color image sensors. This disclosure describes a method where the CFA design shapes are optimized using post processing algorithms in order to reduce the necessary process steps and to minimize optical noise due to light leakage.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 3

Shape Correction For Optimal Color Filter Array Mosaics for Color Image Sensors

Many current color filter array processes use four separate photo processes for optimal placement of the red, green, and blue color filters in a typical red-green-blue (RGB) Bayer pattern on wafer. The typical Bayer pattern unit cell consist of two green, one red, and one blue pixel cells. The ideal color filter array Bayer pattern unit cell is shown in Figure 1. The typical layout design has the two green filters with kissing corners, with no absence of color filters or gaps between the filters. This absence of holes in the CFA pattern is necessary in order to minimize optical noise in the pixel array when performing the necessary color interpolation algorithms to output color images.

Figure 1: Ideal CFA Bayer mosaic unit cell.

The problem with current CFA processes, is that the color filter squares tend to round out on wafer using current photolithography processes. The resulting Bayer pattern on wafer can result in shapes on that look similar to Figure 2. As shown in Figure 2, the on wafer shapes have gaps between the corners of the color filters. The concern and disadvantage with the lack of filter in the corner is that incoming light can enter in the hole in the filter patten and add additional optical noise to the pixel array. Those familiar with the art typically combat this rounding of the filter shapes by printing all four squares as separate masks to get them as square as possible, but this increases masks and cost to the CFA process.

Figure 2: Typical CFA Bayer mosaic unit cell on wafer.


Page 2 of 3

This paper discloses a method that uses post processing shape correction of the design layout pattern in order to eliminate the gaps and holes in the CFA pattern on wafer, thus reducing optical noise. One can use a shape processing algorithm on incoming CFA design shapes to apply shape corrections to the design data in order to build masks tha...