Friday, February 06, 2015

Sony Proposes FinFET Transistors on Top of PD

Sony patent application US20150029374 "Image sensor, manufacturing apparatus and method, and imaging apparatus" by Yoshiaki Kitano proposes a back-illuminated sensor where at least some of SF, SEL and RST and even TG transistors are placed in a layer grown on top of the photodiode - possibly a local epi, although Sony does not use this word. The transistors are FinFET-like, sort of. Since FinFETs have almost no bulk, the SF gain should be much closer to unity and its effective gate cap should be minimal. The PD area can be expanded to almost whole pixel area:

Update: While we are at Sony news many times this week, WSJ quotes Masahiro Ono, an analyst at Morgan Stanley MUFG Securities, saying "Sony’s image sensors are the best in the world and are at least four years ahead of their competitors." If Sony manages to release this FinFET pixel into production, I would agree with that 4 years ahead claim.


  1. Well this is only a patent application, so hard to tell if it is just the work of one guy daydreaming or a real thrust at Sony. The fact that there are a number of embodiments proposed suggests it is more the former than latter.

    That said, Sony is certainly out in front on a number areas. Nothing like having a lot of revenue for enabling massive R&D, which in turn came from their huge investment years ago in CMOS image sensors. Good business strategy at top, and of course great contributions from many talented engineers.

  2. Samsung is at least as capable as Sony in semiconductor technology and ahead of them in FinFET production implementation, so unless Sony has very strong patent protection there will not be a four year gap.

  3. Patents are practically always written with multiple embodiments, at least of the lawyers writing them are any good. I am always surprised at the legal profession's capability to take a specific straightforward idea and twist into a diffuse poorly defined jumble with and endless string of dependent claims that are near impossible for rational brains to comprehend.

    1. You need to understand the role of patent. If you want to protect yourself and you think that your prefered embodiment is the best one, then you can write a very simple application text. But if you want to block "all" the potential competitors to exploit your idea (not patentanle), then you need to imagine all the possible embodiments even dayream-like ones thinking that maybe some technical issues can be resolved.
      -yang ni


All comments are moderated to avoid spam.