Monday, February 04, 2008

Kodak Catching-Up in Pixel Race

Yahoo: Kodak announced its 1.4um pixel sensor generation. KAC-05020 is claimed to be the world’s first 1.4 micron, 5MP device. The sensor uses two new technologies:

  1. High sensitivity color filter pattern which adds panchromatic, or clear, pixels to the RGB pixels already on the sensor. I believe this CFA idea was announce in June.
  2. N-type substrate, so that holes are detected instead of electrons. Probably Kodak is going to discuss this technology in a few days on ISSCC 2008.
    Correction: The substrate is still P-type, but the photodiode is located inside deep N-well, thanks to Albert T. for the info from ISSCC.
The sensitivity is said to be "up to ISO 3200". The sensor supports 720p video at 30 fps.

The KAC-05020 will be demonstrated by Kodak at the GSMA Mobile World Congress held Feb 11 14 in Barcelona, Spain. Samples of the KAC-05020 are scheduled to be available in Q2 2008.

Kodak was late with its 2.2um pixels. So it probably decided to forgo 1.75um pixel generation and jump directly to 1.4um. Micron, Samsung and ST Micro already presented 1.4um sensors, so Kodak plays a catch-up game here.

Update Feb. 5: AP reports that the new sensor will be sold for around $5 each in quantities of 1 million or more.


  1. Yesterday Kodak presented the new sensor technology at ISSCC. It is an imager that indeed detects holes in stead of electrons, but it is still made on p-type substrates and the pixels are located in an n-well. Interesting improvements are shown for dark current, cross-talk and quantum efficiency.

    Albert T.

  2. Thank you for the information. I have not seen Kodak presentation yet. What is the depth of this N-well?

  3. Please check your text : it still says n-type substrate, it should be p-type substrate. The n-well seems to be the "normal" n-well of the standard TSMC process. I guess it should have a depth of 1.5 ... 2.0 um, although the exact number was not disclosed.

    Albert T.

  4. Albert, thank you for the correction. I fixed the typo in the post. If the N-well is 1.5-2um deep indeed, it's not "standard" digital Nwell, it is closer to deep Nwell used in mixed-signal processes.

    It look like Kodak decided to sacrifice red QE to reduce color crosstalk - quite a common move in modern small pixels designs, albeit with Nwell twist.


All comments are moderated to avoid spam and personal attacks.