How will the 450mm Transition Affect Advanced Packaging and 3D ICs?

How will the 450mm Transition Affect Advanced Packaging and 3D ICs?

That was my kick-off question for Manish Ranjan, Vice President, Product Marketing, Advanced Packaging/Nanotechnology Segment at Ultratech, during our annual SEMICON West Thursday wrap-up discussion. It’s never been planned that way, but I always seem to interview Ranjan at the tail-end of SEMICON West, and subsequently end up bouncing a weeks accumulation of thoughts off of him. What inspired this question was his announcement to me that Ultratech’s 450mm initiative is underway, and the company intends to ship its first laser-annealing tool for front-end processes by the end of next year. The company plans to use the same platform for back-end applications, he said.

I posted a similar post a year ago, wondering why 450mm discussions have focused on front-end processes, chalking it up to the fact that before wafer-level packaging, (WLP) the wafer size didn’t affect the packaging end. But that situation is changing rapidly, and Ranjan predicts that the first packaging step performed on 450mm wafers will be flip chip.

As an example of the future impact of advanced packaging, Ranjan pointed to the latest model of iPhone, whose advancements aren’t just based on front-end scaling. “They are enabled by packaging technologies, which enable feature integration, thereby replacing standalone solutions such as digital camera, GPS and alarm clock.,” he explained, adding that if you look at major fabless companies, they are transitioning from QFN to WLP. “50% of their products will require advanced packaging solutions,” predicted Ranjan. He also says that at 450mm, 100% of devices will be manufactured using advanced packaging technologies, versus 14% of devices today. Backend equipment companies will need to scale out to serve the top tier players (Intel, TSMC, IBM, GLOBALFOUNDRIES) who are responsible for 60% of Capex spending.

This naturally led to discussion about the OSATS, and the sticker shock they are likely to feel adding capacity for 450mm WLP. Will they invest in 450? “If (When) TSMC or GLOBALFOUNDRIES goes to 450, the OSATS will have to invest in the infrastructure to support back end operations,” says Ranjan. This is sure to be a daunting task, as many of them have just barely shifted to 300mm.

When it comes to 3D technologies, Ranjan says he thinks it will be a while before those processes transition to 450mm, but will commercialize first on 3OOmm. It is possible that two parallel paths will coexist: 450mm for extreme leading-edge (7nm) technology; and those who can’t afford to invest in 450mm will leverage the mature 300mm infrastructure to add 3D packaging technologies for delivering superior process to performance ratios.

“Leading-edge products have better gross margins than conventional technologies. Companies who invest in those capabilities are profiting,” noted Ranjan. He compared it to the adoption of flip chip, which has gained mainstream attention, despite the fact that it STILL costs more than wire bond. “Flip chip is more expensive than wire bond, but performance dictates the use of flip chip. TSV is not at the same point yet, but we’ll get there,” he said. “Cost is a concern, but the concern to consider first is what does technology need?”

This brought my thinking full circle to the argument of the cost of 3D compared with conventional packaging. Like Ludo Deferm did early in the week, Ranjan confirmed my thoughts that its not as much about the price coming down as it is about the customers being willing to pay more for higher performance. ~ F.v.T.

1 Comment

  1. Dev Gupta - July 26, 2013, 7:16 am Reply

    At SemiCon West Micron had a paper on their HMC 3-d Memory Cube w/ TSVs. Was that covered ?

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title="" rel=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>