Canon engineers have developed a new design for the micro lenses it uses on imaging sensors that it claims will reduce vignetting and false coloration at the edges of the picture. The new designs have a bi-convex lens that uses the upper surface to collect light and the lower to channel the light more effectively to the photodiode. The patent application shows the lower face of the micro lenses with a convex surface featuring an off-center vertex. Canon says these would be placed at the edges of the sensor to direct light approaching from a steeper angle. The idea is to direct more of the light toward the photodiode than can be achieved with standard single-micro-lens designs.

As pixels have depth it can be difficult to channel light from the camera’s lens down the ‘well’ to reach the photodiode unless it approaches straight-on. When a pixel is positioned at the edge of the sensor array it becomes more difficult because light approaches from an extreme angle and can miss the photodiode, as the refractive index of the micro lens isn’t high enough to bend it directly down the well. Since certain colors experience different refractive indices, some wavelengths of light don't make it to the photodiode either. Thus, pixels outside the central area can report less light – and false colors – compared to those in the middle of the sensor.

Canon’s new dual micro lens design aims to take more control of the light as it enters and exits the micro lens, and to channel it in a more vertical direction down the well so that less is lost. This should in theory improve both vignetting and false coloration nearer to the edges of the image.

For more information see Canon's full patent application.

Extract from the patent:

The lower surface 102 of the microlens 103 has an asymmetrical shape with a position nearest to the photoelectric conversion device 104 (a position at which the thickness from a center plane 130 of the microlens 103 is the maximum) shifting from the center position of the microlens 103 to the central side of the pixel array 110A. The lower surface 102 of the microlens 103 has a convex shape with respect to the photoelectric conversion device 104.

Each microlens 103 is formed from a material having a higher refractive index than a material in contact with the lower surface 102 at a position between the microlens 103 and the photoelectric conversion device 104. The microlens 103 is formed from, for example, a color filter material.

The upper surface 101 of each microlens 103 has a convex shape with respect to the incident side of incident light. The incident light 111, incident light 121, and incident light 131 entering the microlens 103 from the same direction as that of the straight line 115 each are refracted by the upper surface 101 of the microlens 103 and focused onto the photoelectric conversion device 104. In this case, if the upper surface 101 of the microlens 103 lacks in refractive power with an increase in curvature radius, the lower surface 102 of the microlens 103 compensates for the refractive power to cause the incident light 111, 121, and 131 to enter the photoelectric conversion device 104. At this time, the refractive power of the lower surface 102 of the microlens 103 is larger than that of the upper surface 101 of the microlens 103. As described above, the microlens 103 can focus incident light onto the photoelectric conversion device 104 by using the upper surface 101 having a convex shape extending upward and the lower surface 102 having a convex shape extending downward with respect to the center plane 130.