I got an email last week from R. Colin Johnson, of EE Times, looking for my impression on the collaboration between IBM and 3M, and their announcement to begin development of an adhesive material that would allow building “silicon towers” or “silicon bricks” that would be 100 chips high. Johnson wanted to know my thoughts on this, could 3M’s thermally conductive adhesive really be, as they claimed, the “key material” for the commercialization of multi-chip 3D ICs.
Well Colin, I have a confession to make. Although I did read the initial press release when it came out, it mustn’t have made much of an impression on me, because after reading it, I went on to something else and it completely slipped off my radar until you brought it to my attention. So I spent the later part of last Friday doing some research to bring myself up to speed. Honestly, the more I read about this development, the more questions I have. So if I were to put my initial thoughts on this into one word it would be ‘skeptical.’ Not because I don’t think IBM and 3M have the combined skills, but because there’s just not enough information out there about this to really have an opinion.
For sure, there were lots of speculative reports. (The press release did its job.) Some made sense. Some just made things more confusing. One even lumped these developments in with Intel’s TriGate technology, under the catch–all phrase ‘similar 3D semiconductor efforts by Intel’. (I guess we haven’t fully grasped the difference between 3D transistors and 3D stacked ICs yet.)
From what I understand, the “key material” being developed by 3M is a thermal underfill adhesive material that will address thermal issues caused by 3D stacking. While that’s definitely addressing one of the remaining limitations needing to be addressed before 3D ICs go to volume production, I don’t think anyone believes it’s the only/final key to unlocking remaining issues for 3D stacks.
The original press release claims that these new materials co-developed by 3M and IBM (3M creates the material, IBM does the prototype stacking) could “leapfrog” over current attempts at 3D stacking. But beyond that, the rest of the release is pretty vague. Is this to replace different types of bonding? How will interconnects be formed? Details, folks, we need details.
Phil Garrou (Insights from the Leading Edge) wrote a brief blog post on Sept 10; he had similar questions and so went straight to the source and found out that there’s not much to say yet because the program hasn’t even started yet. So it’s kind of hard to believe the leapfrog claim…or that they’re going to go to volume production with 3D ICs before, say, Samsung or TSMC.
I used to get really fired up about all the reported breakthroughs and new developments; and certainly there have been many over the years who claim to have THE answer. Everyone also claims to have a launch date that’s earlier than someone else’s. It’s called “effective marketing”. And while each development and breakthrough technology COULD be the one to bring it to commercialization, I believe it’s really about the collective breakthroughs coming together to create the perfect storm. While each new development is important, what’s really notable is that all these advancements are putting 3D in the limelight, which helps bring us closer to market adoption. Anyone involved in this 3D revolution will tell you that COLLABORATION toward the common goal is the key. It’s not so much that there’s a magic material that might be THE answer, but that everyone is in this race because they realize that the 3rd dimension holds the key to furthering Moore’s Law, and they want to be the ones to get there first. – F.v.T.
