Wednesday, November 07, 2007

Yole on 3D Packaging Roadmap

Yole Développement published a market report on adoption of wafer level packaging in image sensors and other areas. As far as I'm able to understand, on the graph below the yellow color means R&D activity, while the red is production.

So, by the end of this year we are supposed to see wafer scale packaged sensors from most of the major sensors manufacturers. Somehow I don't see this happens.


  1. Xintec (TW) and WLCSP (China) are supplying WL-CSP services in high volumes to mainly Omnivision and to some other smaller size CIS fabless players as well. They got licenses from Shellcase/Tessera.

    Now, IDMs are also entering the market, with their own WL-CSP technologies:
    - Toshiba announced last month starting by Q3-2007 production of their TCV "Through chip Via" technology, bringing back in house the packaging of all their CIS activity (source: Toshiba website)
    - ST had very little percentage of WL-CSP products on the market this year (CSP packaging was outsourced), mainly for their dual-sensors CIS found in Nokia camera cell-phones. Mass production was initially scheduled in 2007 but seem to have been delayed to 2008 (source: GSM Barcelona)

    Korea: Samsung & MagnaChip WL-CSP activity starting is imminent. A real packaging infrastructure is being built currently in Korea with new CSP packaging providers such as OptoPac, EPWorks...

    Top CIS products for the WL-CSP technology are low resolution CIS (typically found in camera cell-phones & notebook webcams): mostly CIF and VGA resolutions have founded their place today on the market. With sensors ASP declining over time, 1MPixel & 2MPixel CIS encapsulated in a WL-CSP will become more mainstream in the coming years.

    In 2007, the amount of CIS devices (camera cell-phones + notebooks) encapsulated in a CSP is estimated to be about 25-35%. It should be about 65% by 2012...

  2. I see what you are saying. However, let me present another view on this:

    Since its appearence 4 years ago, Shellcase package adoption is limited by Omnivision. At some point Shellcase was used by few other companies, but not anymore. All the other packages are in small-scale samples, I'm even not sure they have passed quals for mass production.

    ST, Toshiba, Magnachip talks about WL packages are still at the talks stage, no real mass production after many years of talks.

    To me the real change will come once backside illumination becomes mainstream. May be this will happen in 1.2um pixel generation, or the generation after that.

    So, I agree, 3D package is the future, it it's still not quite here yet (sans Omnivision).

  3. Thank you for answering to my previous message. I perfectly agree with your comments.
    By the way, the real driver today in consumer CIS is clearly to enhance the fill factor, which is presently limited to 30%-55%, depending on which CIS designs you talk about. And further scaling to next pixel generation nodes (1,4um , 1,2um, ...) won't help to solve this issue. And you can easily realize that these sensors crually suffers from sensitivity, especially in low/medium light conditions.
    Additionally, people would like to get rid of the microlenses array, which is costly and cause very stringent limitations for the test & manufacturing of those devices.

    Therefore, all conditions are present to drive the adoption of "3D like architectures" in the coming years, either to be based on 3D-TSV, active photo-receptive material layer over DSP chip or lastly backside illumination approaches (see e2V and Tracit Technologies last announcement for industrial applications)
    The main criteria for selecting one of these options will be cost. And today, none of these approaches is cost effective to enter the consumer market.

    Coming back to WL-CSP, Shellcase technology has shown really powerfull and cost effective. The drivers to adopt the technology were form factor, package height and cost reduction benefits by full packaging at the wafer level.
    It applied very well to low resolutions CIS (CIF, VGAs...) that suffered from very weak ASP and it is now very hard for any newcomer to enter such low resolutions. So now the question is who will win the next design generations for higher resolution CSPs? There won't be enough place for more than several players.
    Future will tell ...

  4. Interesting news about Toshiba:

  5. Today news about OKI & ZyCube WL-CSP...

  6. To i-micronews: Yes, indeed the news are very interesting. Regarding Oki entering volume production, I wonder who is Oki's customer on its WL-CSP technology. I'll put it on the front page in a minute.

    To Anonymous: Wirebond COB package probably always will be the cheapest option. What really propelled Shellcase to high profile design wins with Sanyo and Omnivision was yield problems assosiated with the COB package. At one point Shellcase controlled 50% of the CIS market. Since then COB packaging yield has improved, so there are not that many companies continuing with Shellcase these days.

    Similarly, I believe the next generation package should solve some really important problem to be widely adopted. To me this will be the backside illumination used in small pixels. Before that I doubt there will big a large following to TCV or other similar technologies.


All comments are moderated to avoid spam and personal attacks.