Weekly Patent Review

I have not written these weekly reviews for a while, in part because I have not seen any noteworthy ideas in the weekly application stream. However, this week I feel I must do it.

Among 28 image sensor related US patent applications published this week, there are two that look quite similar. They are filed just 3 days apart.

TSMC application US20080217659 promises to improve small pixel crosstalk and blooming. Micron's US20080217716 also reduces dark current. Both are achieving their goals by separating photodiode anodes from the substrate by means of N-type epi layer. About the only difference that I was able to spot is that Micron uses P-type substrate, while TSMC uses N-substrate.

Once the anodes are isolated from the substrate, one needs to bring a reliable ground potential to them. This is what Micron application US20080217718 about. It proposes to add ground contact to each pixel or only some of them. I highly doubt it would be granted though, as there is a lot of prior art on that.

TSMC application 20080217719 explains how to reduce crosstalk by adding an additional implant between the pixels. Nice try, but I afraid that prior art prevents this application from being ever granted.

Another interesting application came from Teledyne. Teledyne acquired Rockwell Scientific in 2006 to create Teledyne Imaging Sensors. Then, in February 2008 Judson was acquired and merged with the new business unit. Teledyne application US20080217661 proposes 2D TDI imager based on 4T CMOS pixel with double transfer gate. Very nice idea, except that I could not understand how they read TDI charge packets at the end of integration. Now, if Teledyne manages to scale the pixel size down and adapt its TDI to bayer pattern, we would get a very efficient image stabilization for the mainstream applications.

Teledyne also applies for BSI patent US20080217723. The main idea is to apply a voltage between front and backside of the wafer, so that photocarries would drift, rather than diffuse through the thick substrate. Sounds great to me, apart from the power consumption spent on the maintaining of electric field in the substrate.