Sunday, June 02, 2013

Sony Awarded for BSI CMOS Sensor Development

Sony Japan received Prime Minister Invention Award for "development of CMOS image sensor technology for the next generation by the invention of back-illuminated CMOS image sensor" (Microsoft translation, Google translation). Prime Minister Invention Award is a prize for the invention that have greatly contributed to the science and technology and the development of the industrial economy.

41 comments:

  1. Maybe it is just the translation, but "invention of back-illuminated CMOS image sensor" is just wrong. Aside from backside CCDs, I think the first US patent (by priority date) on backside illuminated CMOS image sensors was by Photobit based on the Shellcase sort of process. But the big mass-production breakthrough was by TSMC for Omnivision and mass production. Perhaps there is a Japanese Sony patent that predates all these but since they don't cite a patent number it is hard to tell. If it was a "development" award then it would make more sense. Sorry Sony, no kudos from me on this one.

    ReplyDelete
  2. Dr. Fossum is a little bit arrogant here. The backside illumination is concept well known for all the sensor expert. I didn't see any obstacle to pass from CCD to CMOS for the backside illumination scheme. This kind of patent should be rejected because it's obvious. The real knowledge and development are inside the detailed way to do the substrate thinning, there are so many ways to do and some of them work and some of them don't work. This activity involves a lot of engineering effort and also financial effort too. Visibly SONY is the best at this moment.

    ReplyDelete
    Replies
    1. Seems you agree with me, an "invention" award is wrong, but a "development" award would be sensible.

      Delete
  3. Sony might be a little bit earlier.

    Sony announced the first BSI CIS on 06/11/2008. (1.75um pixel, 60fps)
    http://www.sony.net/SonyInfo/News/Press/200806/08-069E/

    OV introduced the first 8MP BSI CIS on 09/23/2008 (OV8810, 1/3-inch, 1.4um, OmniBSI)
    https://secure.ibsystems.com/NANO/nbc/articles/1/592806/OmniVision-Premieres-World-First-1-3-Inch-8-Megapixel-CameraChipTM-Sensor-With-1.4-Micron-OmniBSITM-Technology


    A Nikkei Electronics article on 04/14/2009 also described Sony as “earlier” to commercialize BSI CIS:
    “OmniVision is the company that announced a plan to commercialize backside-illuminated (BSI) CMOS sensors for the first time in the industry. But Sony commercialized a BSI CMOS earlier than OmniVision, which plans to ship its production model in the summer of 2009.”
    http://techon.nikkeibp.co.jp/english/NEWS_EN/20090414/168778/

    ReplyDelete
    Replies
    1. Thanks for digging this information up. Could be Nikkei Electronics is a little biased towards Japanese electronics makers which explains the defensive posture of the 4/14/09 article. Or perhaps it is dead on. Seems like the mass production development was pretty much contemporaneous between the two companies and I stand corrected on that one.

      Based on papers at IISW, it seems the community was waking up to BSI as a necessary evil to overcome in the 2005-2007 time period and by 2006 the development of BSI for mass production was well underway (see papers at 2009 IISW BSI symposium).

      It is nice that Sony gets a Japanese national award. I just object to calling it an "invention" award, as if they invented the BSI CMOS image sensor.

      Delete
  4. Above opinions are just like I felt when I read Dr. Fossum's CIS patent and Kodak's pinned PD patent.

    ReplyDelete
  5. Annoying lack of understanding of CIS history. OmniVision shipped it's first production BSI sensors in January 2009 and it was more advanced than Sony's ! OmniVision's technology was based on standard bulk Si epi and by June 2009 the OmniVsion-TSMC alliance was production suppling major high volume mobile customers. TSMC and OmniVision were already sharing their breakthrough development at the June 2009 IISW. Sony's technology was based on SOI which could only deliver small quantities and so Sony's production was limited to DSC camera while OmniVision was delivering to the world into the high volume mobile marketplace. Immediately after OmniVision's public announcement an embarrassed Sony quickly came out with their own announcement that they were doing BSI too. But Sonys fell far short as OmniVision was already had a high volume mass production in place with TSMC. It was 1-2 years before Sony got BSI in bulk Silicon working in mass production.

    ReplyDelete
  6. US 6429036 "Backside illumination of CMOS image sensor" is from 2000. Does that count? Also US 6169319 "Backside illuminated image sensor" is from 1999 and discloses a CMOS sensor (US 6168965 similar).

    ReplyDelete
    Replies
    1. Yes, US 6429036 is the Photobit patent I was referring to. It has a priority date of Jan 1 1999, about 6 months earlier than the two Tower patents.
      Really no surprise about that sequence. Tower was developing the Shellcase packaging process and Photobit was an alpha customer for that. We discovered accidentally that we could image from both sides of the chip. After filing, we discussed this with Tower, our fab partner. But at that the time the Shellcase process was still immature and no one was calling for BSI.
      The idea of doing BSI with CMOS dates back to the earliest CMOS APS image sensor papers. Again, no surprise. At JPL we were well acquainted for BSI in CCDs, from Texas Instruments, to Mike Lesser, and the delta-doping by the Grunthaners internal to JPL.

      Delete
  7. ROWR! FFFT! INVENTIVE CATFIGHT!!!

    :)

    ReplyDelete
  8. US 6331873 (MIT) is from December 1998 and column 18 lines 25-39 discloses the concept of a backside illuminated APS CMOS sensor.

    US 5808329 (Raytheon) dates to 1996 and Fig. 2 shows a BSI CMOS sensor, albeit PPS.

    Still haven't seen anything from Sony predating 2000.

    ReplyDelete
    Replies
    1. I mentioned backside illuminated APS in the infamous dinosaur paper in 1993 and in subsequent papers. (So, yes, even then the idea of a backside illuminated structure was obvious - patents are about the fabrication process itself). The MIT patent doesn't claim any process for BSI CMOS APS.
      Still waiting to hear about the Sony invention!

      Delete
  9. This is simply NO INVENTION at all!

    ReplyDelete
    Replies
    1. Ah, so much response for so short a paragraph.

      Does it really matter who invented BSI at this point? Our history has shown again and again in the past half century: American invented, Asian (first Japanese, then Korean, maybe China next?) figure out how to make it. Who invented CCD? Who is making CCD now?

      Open up iPhone, open up S4, whose sensor is in there? For that, Sony deserve an award. Whether they call it "invention" or not, I don't care. While we have so much time debating this, they just work. Our freedom give us time to think and invent, but no time to make it work.

      Delete
  10. there's more american invention in SONY sensors than disclosed to the public

    ReplyDelete
  11. Let’s give awards to all of them:

    People who came up the BSI ideas first.
    People who file patents on some BSI manufacturing methods first.
    People who make BSI sensors for commercial market first.
    People who make BSI sensors for high-volume mobile-phone market first.
    People who make the most cost-effect BSI sensors first.
    People who currently deliver the highest performing BSI sensors.
    People who is the current market share leader.
    People who make the smallest pixel BSI using the most advanced process nodes.

    ReplyDelete
    Replies
    1. How about the guys in NC who invented the wafer bonding process that made it all possible?

      Delete
  12. After BSI CCD sensors, I don't think a CIS that accept light from its backside is an invention.
    I think the invention of BSI CIS should be those with the structure and production method that are used in today's BSI CISs but have not be seen in BSI CCD sensors and FSI CISs.

    ReplyDelete
  13. People forget IR sensors which are made on transparent substrat and hybridized to silicon ROIC. They are all BSI...

    ReplyDelete
  14. Was this the first publication of BSI CCD?

    1973 IEDM: "Development of a thinned, backside-illuminated Charge-Coupled Device" by S. Shortes et al, from TI.
    30 um thickness, 500 pixel linear array, aluminum gates.

    They published more details and a 400x400 BSI array in Nov. 1976 in IEEE trans. El. Dev, and maybe earlier on conferences.

    So we should celebrate the 40th anniversary of the backside illuminated imager this year...

    ReplyDelete
  15. Well, the authors claim it is the first. But from September 1973, months earlier, Westinghouse published this paper:

    A Tradeoff Analysis for CCD Area Imagers: Frontside Illuminated
    Interline Transfer vs. Backside Illuminated Frame Transfer

    http://www.imagesensors.org/Past%20Workshops/Dick%20Bredthauer%20Collection/1973%20San%20Diego/1973%2003%20Barbe.pdf

    and there is seemed that backside illumination was a well known concept.

    ReplyDelete
    Replies
    1. BTW, I believe backside illumination was used for silicon diode arrays in vidicon tubes, and with other materials before that.

      Delete
    2. not vidicon Doctor! Vidicon is photoconductor sheet.

      http://www3.alcatel-lucent.com/bstj/vol48-1969/articles/bstj48-5-1481.pdf

      Delete
    3. Vidicon designates the electron gun configuration. Various target materials have been used. The usage of the word was not precisely assigned but it is pretty easy to find examples of the terms "silicon vidicon", "lead-oxide vidicon" (for generic Plumbicon), "chalcogenide vidicon" and others in addition to the use of plain "vidicon" for tubes with antimony trisulfide targets.

      Delete
  16. I ran a lab at Texas Instruments 1968-1970 where we tested silicon vidicon tubes made in a special South building fab. Thesre were originally developed to be used inside a big airborne gunship FLIR sytem. The silicon vidicon looked at a mirror-scanned line array of GaAs LEDs emitting in the near infrared that got its signal from a linear array of HgCdTe detectors.

    In silicon vidicon tubes, the photodiodes were created on the top surface of a silicon wafer in areas about 18 mm in diameter. The silicon was then cut into circles and the backside of the wafer was thinned to 25 microns or less. This formed a silicon membrane supported only by a silicon ring left on the outside that had been pretected from etch. You could easily see red images through these membranes. We made some as thin as 13 microns, which appeared yellow-orange.

    In vidicon tubes, the diode side faced the electron beam, which would periodically scan to restore the reverse bias. The recharge current caused a current to flow from the bulk backside that was measured by a low -noise amplifier exactly as was done with other vidicon tubes. The whole array had to be coated with something to avoid having charge stranded on the oxide between the diodes. We used a sputtered metal-silica coating, RCA used gold pads on each diode.

    The illumination came in the back of the array. in SIT tubes (Silicon Intensified Target) electrons from a photocathode came into the back of the array.

    Our targets were used in the Westinghouse cameras on the moon that replaced the old SEC tube that had its ceramic target cut in half by the electron flux generated by a direct solar image.

    ReplyDelete
    Replies
    1. How about the blooming effect on the silicon target tube??

      Delete
    2. Well-designed targets typically bloomed less than the lenses they were used with. They never had the blooming and tailing problems associated with lead-oxide targets.

      Delete
    3. Awesome information Dave! thanks!

      Delete
    4. the Plumdicon has a super sensitivity and low ark current. But I talked about the silicon target device. I remember that the blooming was a big issue since the pixels are formed by the diffusion islands on a thinned silicon substrat.

      Delete
    5. If the forumers here were patent application eximinators, then no patent application will pass on BSI sensors...

      Delete
  17. Crowell & Labuda described a back-side illuminated, silicon diode array sensor in 1968. (Bell Sys Tech J, May-June 1969)

    ReplyDelete
  18. Suddenly we found that we didn't invent any thing...

    ReplyDelete
    Replies
    1. funny, but not true. It is just that we truly all stand on the shoulders of giants. 2 of those giants will be at the IISW meeting next week. Mike Tompsett and Gene Weckler.

      Delete
  19. History has time and again proven that invention is not done by the inventor, rather by the one who aggressively publicizes (or politicizes) it. Popular example being wireless communication, usually attributed to Marconi.

    ReplyDelete
    Replies
    1. And who is the real inventor according to you??

      Delete
    2. Try this for a starter
      http://www.qsl.net/vu2msy/JCBOSE1.htm

      Delete
    3. Interesting, i guess its all about advertising the real inventors die in oblivion.

      Delete
    4. Yeah, indeed interesting that an indian guy writes an article about an indian inventor (Bose) without any citations, saying he invented the wireless communication before Marconi..That's a trustable source...

      Delete
  20. @Anon, if you want citations, there are plenty of materials on the web, do a search for yourself. This will guide you for the references --https://www.cv.nrao.edu/~demerson/bose/bose.html, written by an "American" if that gives you a sense of "authorita"..

    ReplyDelete

All comments are moderated to avoid spam and personal attacks.