The market is demanding ever-increasing image resolutions for cameraphones, while retaining the same or smaller form factors. This has resulted in increasing the number of megapixels, but reducing the corresponding size of each pixel. Maintaining good image quality using small pixels is becoming very challenging as less light is picked up by each pixel. New, innovative approaches are required to solve issues such as: dynamic range, faithful color reproduction (red reproduction in small pixels is an issue, for example), blurred images taken at low light (unsteady hand motions affect imaging at low light) and low light sensitivity itself.

The widely used 4T pixel design is based on the pinned photodiode and correlated double sampling technique (Figure 1). It is important to transfer the photodiode charge completely at the end of each charge transfer phase, so that no charge remains on the photodiode. Any residual charge causes image lag and excessive noise. These problems are especially evident at low illumination level, where resulting pictures will appear "grainy" and unclear.

Figure 1: Conventional 4T Pixel Readout

If the charge transfer is incomplete and leaves just a few electrons on the photodiode, it results in a major image lag in the sensor. Worse still, the few remaining electrons contribute to increased "noise" and are different from pixel to pixel. Therefore pixel designers make great efforts to achieve complete charge transfer. However, achieving a complete charge transfer is often a trade-off with other, equally important parameters. One of them is full-well capacity: the maximum charge that a pixel can store and transfer for readout.

A larger full-well capacity, translates directly into a wider dynamic range for the image sensor. For complete charge transfer, the photodiode volume should be fully depleted at a voltage of 1-1.5V. This means that the photodiode doping should be light. Lower doping means low capacitance and, therefore, only a small, full-well capacity (FWC).

To improve red color sensitivity, the photodiode junction needs to be placed deeper into the silicon. The deep photodiode needs to be even less doped to be able to deplete fully at 1-1.5V. This results in unacceptably low full-well capacity.

Low full-well capacity becomes most problematic when scaling pixels. Traditional 3.6um pixels often gave full-well capacity in excess of 30K electrons. Scaling them down to 2.2um reduced full-well capacity to about 12K. The modern 1.75um pixels have full-well capacity of less than 10K for the most part (see Figure 2).

Figure 2: Full Well Capacity vs. Pixel Size

The use of feedback in the pixel to reduce image lag and noise overcomes the limitation of having low full-well capacity in small pixels.

Advasense’s innovative Feedback Controlled Pixel (FCP ™) does not require full charge transfer, as feedback ensures that image lag is virtually zero (Figure 3). This enables photodiode storage capability to be greatly improved and pixel scaling down is, therefore, simplified.

Figure 3: Block Diagram of Feedback Controlled Pixel (FCP™)

FCP technology enables the use of Deep PD, resulting in much higher FWC, improved red quantum efficiency (QE) and reduced crosstalk (see Figure 4).

Figure 4: Advasense’s Advanced Pixel Architecture

Advasense's FCP technology overcomes three key challenges associated with ensuring high picture quality using small pixels, to enable vivid, colorful images: