Oct 18 2007
At Large At IMEC: “To 3D or not 3D? When/How is the Question.”
Editor’s Note: These continuing reports are from Jim Kobylecky – Chip Design’s Technology Editor - at the IMEC in Leuven, Belgium. – JB
Unless you’re going to begin that second career soon as an Alaskan bush pilot, you will probably find yourself becoming a 3D designer. Chip real estate isn’t getting any cheaper, and the decision to start building high-rises on your silicon Manhattan is inevitable. Some applications, such as memory are already there. Past a point, it will just make economic sense.
But “When” will that point be, and “How” it will come about are not easy questions. That was the consensus of a panel discussion at the ARRM2007 Press Preview, an intensive IMEC technology conference in Leuven, Belgium. Moderator Pol Marchal (see figure 1 below), a Senior Researcher at IMEC, poised why/when/how questions to Fred Roozeboom of NXP, Diederik Verkest of IMEC (see figure 2 below), Jochen Resinger of Infineon, and Erick Beyne of IMEC (see figure 3 below). While 3D designing in some categories is already taking place, they’re relatively obvious, almost homogeneous propositions. When it comes to stacking many different kinds of IP and deciding between many different ways to interconnect them, the situations get much more complex. Much too risky to try without really good, thorough exploration and verification tools. Unfortunately, it’s also too risky to develop such complex EDA tools until there are enough users willing to buy them. It will be 2D business as usual for the time being.
Yet the economic advantages will gradually grow, first for special applications, then for more and more uses until the question in ten years, according to one panelist, will change to “why use 2D?” Future historically, it seems that when there are enough potential or experimental users, the tools will appear, and when there are enough tools, enough users will appear.
Chicken? Egg? The answer may be incremental. Long before there is true 3D general purpose design, their will be “2.5D.” Maybe we’re already at 2.1D. Startups and graduates straight out of the universities are already experimenting and developing early, incomplete tool sets. And as they and their initial users collide, the databases and standards necessary to develop more advanced tools will form out of the debris. The question then is “how soon.” Will there be enough “Co-petition” as one IMEC foil put it for these initial developments to quickly take place? Or are we coming into a time of competition that will delay the future? Ten years from now, we’ll know the answer, and it will seem very obvious.
On the “Yeah, but what now?” front, I should mention there were also demonstrations of recent IMEC developments that are just short of going out to become someone’s new products. These included a wireless sleep apnea monitor that can be used for home pre-screening and a body heat powered wireless EEG prototype (see figure 4 below). Coming soon to a Borg near you.
Figure 1: Moderator Pol Marchal set the stage.
Figure 2: Fred Roozeboom and Diederik Verkest gave their perspectives.
Figure 3: Jochen Resinger and Erick Beyne added their views.
Figure 4: Sleep Apnea and Self-Powered Brain Monitors hit the keyboard.
