Wednesday, October 07, 2020

Sony Expands Polarization Sensors Lineup

Framos reports that Sony launches the new polarizer sensors IMX264MZR (monochrome) and IMX264MYR (color). These 5MP sensors have a 2/3-type format and a pixel size of 3.45 µm. It reaches frame rates up to 35.7 fps at 12-bit (only) and has one ROI window. The new CMOS Pregius sensors are equipped with a four-directional polarization square pixel array which can filter light from all four direction in a single capture.

Sibel Yorulmaz-Cokugur, Sensor Expert at FRAMOS, says: “Polarsens technology has changed the industry. The on-chip polarization brings significantly higher quality analysis and simplified application design“.

Evaluation samples of Sony’s IMX264MZR and IMX264MYR will be available from October 23rd 2020, mass production starts in February 2021.

Framos compares an image coming from a polarization sensor:

with the same image with no polarization filters engaged:


  1. Interesting that Sony is building up its polarized sensor lineup but ... what is the main difference from IMX250MZR apart that it is slower? :)

  2. Imx264 is the same die as imx250, just reduced functionality/speed. Imx252 is the same die as well. Sony sells sorted variants of the same die under different names.

  3. Ok, not so much things new under the sun then. Anyway, I found it very interesting to see how Sony develops its line of products beside "standard" visible sensor (polarized and SWIR sensors mainly).
    Thanks to the really amazing stacking process that they develop, I think they could relatively easily address even MWIR/LWIR spectral ranges. In addition, as they are thinning down the PDA substrate, they could also make multispectral SWIR sensors.
    To make it short,I am sure we will see in the next couple of years very interesting innovative devices at "low cost".

    1. welcome to the club of new world !

    2. Hi Simon, I agree, we will see a lot of progress due to stacking technologies. What we see is just the beginning, CIS is a pioneer, many others will follow. A lot of new options become available once you can stack dies with precision in the 100s nm range and contact pitch in the <um range. You can combine components of different technology nodes into one die. Wafer stacking with hybrid bonding is established technology now in CIS. What comes around the corner is die to wafer hybrid bonding. Companies are working on machines and we will see this in mass production few years from now. As posted before... Besi presented a paper at ECTC about one of this machines:


    4. Hi Simon, even though polarizer and stacking are different topics... another addon on stacking. A question I ask myself is 'which companies will have this machines'? I think there will be/is a paradigm shift in this as well. Huge investments are required to keep the machines/tools up to date for such technologies. You know the "real men have fabs" quote and how the situation in front end foundries have changed since this sentence was said? I think it is less and less meaningful for image sensor companies and even for example industrial camera companies to have own manufacturing equipment. I think more and more will go fabless and large companies will specialize on manufacturing/production, more and more backend stuff as well will move to the TSMCs of this world. Is it useful for a industrial camera manufacturer that produces a few 10000s (the big players make a few 100000s) of cameras a year to solder them in a owned fab? of course not (in my opinion, can be I'm wrong ;-). There are EMS companies that manufacture cameras and focus on manufacturing and they can fill their lines around the clock. To build a camera on top of a image sensor is trivial compared to make the sensor. You need very little guys to make such cameras on your own (e.g. buying IP cores for technical core know how) and let them be soldered/manufactured/tested somewhere. You have a huge cost benefit by doing so (the main cost of a industrial camera is the margin of the camera manufacturer, the main difficulty is the image sensor) and can add features to differeciate your optical/camera assemblies, you can access new sensors early or adapt them in a way that are not marketed by mainstream camera manufacturers. Why should a specialised image sensor manufacturer that serves a niche market package/stack/... on their own machines?

    5. Hi Yang Ni, I suppose you read the pdf ;-) there is a nice sentence for you in the 'outlook' chapter, dont you agree?
      > moving to IR substrate-camera capability
      > will eliminate the need for the up-looking
      > search and enable real in-line
      > C2W position alignment boosting both UPH and accuracy.

      Die alignment in such a machine in the SWIR spectral range is a classical application of the effect that SI becomes transparent above 1100nm, right?

    6. This operation is very common in wafer bonding machine.

  4. btw. they have the IMX253 (12MP) listed as well in a Polarizer-variant 'MZR'


All comments are moderated to avoid spam and personal attacks.