The people have spoken! The results of last week’s poll are in, and it looks like the majority of those who participated think we should keep it simple. When it comes to heterogeneous integration nomenclature for package architectures, it should be 2D or 3D.
The Back Story
The term “2.5D” has been a topic of debate in the advanced packaging world ever since it was first added to the industry lexicon to define the interconnection of two die placed side-by-side on a silicon interposer using through silicon via (TSV) technology. The legend goes that ASE’s Ho Ming Tong coined it as a joke, and well, like some bad jokes, it took off and it stuck.
Recently the Heterogeneous Integration Roadmap Committee took it upon themselves to come up with a better term to replace 2.5D. In March at IMAPS DPC, they unveiled the term, ‘2D enhanced’ to distinguish architectures that weren’t mainstream 2D packages like wire bond and flip chip but encompassed the more advanced system-in-package (SiP), fan-out (FO) technologies, embedded die and interposer architectures. 2D enhanced is further categorized as 2DO (2D organic substrate) and 2DS (2D silicon substrate). This was intended to simplify things. Seriously.
Before things get too out of hand, we decided to take it to the people’s vote last week during SEMICON West. But while 2D and 3D won by a significant margin, it wasn’t a complete shut-out. Newcomers 2DS and 2DO held their own against 2.5D. Supporters of all the choices have their reasons why and, and a few of them were willing to be interviewed to share their reasoning.
Keep it Simple: It’s either 2D or 3D
EV Group is Team 3D all the way; which didn’t surprise me. They’ve been card-carrying members from the very beginning. Thomas Uhrmann, the company’s director of business development, explained it logically: from a terminology perspective, “2.5D is the same mid-end of line. If the line ends in the middle, there is a problem.”
Experts from some of the leading European research institutes agreed with Uhrmann. Hubert Lakner, Fraunhofer Mikroelectronik and Jürgen Wolf, Fraunhofer IZM-ASSID think we should stick with 2D and 3D and emphasized 3D in terms of systems.
Why not 2.5D? “It’s difficult. Only physicists understand half dimensions,” said Lackner. “2DS/2DO is confusing. We should just forget about it. Organic electronics are needed, but they will be part of 3D.”
“2D means one level and 3D means more than one level,” added Wolf. “3D systems do not just include an active die on an active die, but passives too. There can be different architectures. We want to see complex systems connecting everything.” In the Fraunhofer world, 3D heterogeneous systems integrate ICs, sensors, data processing, memory, compound semiconductors, photonics, and more.
Emmanuel Sabonnadiere, Leti, also agrees that we should keep it simple. “It’s good that we are getting rid of 2.5D, but don’t bother with 2DS and 2DO,” was his advice.
2DS and 2DO are a great idea
There was, however, moderate support – slightly more than 2.5D from the 2DS2/2D0 team, particularly for those who were never fans of 2.5D, but not quite ready to classify everything as simply as 2D or 3D.
Acknowledging that it’s difficult to change the established terms, Jeff Demmin, DARPA, said he always thought 2.5D “was bogus. It doesn’t mean anything.” H also said he thinks calling something that is only a few thin layers high “3D” isn’t quite right. It’s worth a try to change the terminology to 2DS/2DO because “a silicon vs. organic substrate is a notable division that captures the most variations. Fan-out can be included.”
Tim Kryman, Rudolph Technologies, is all for switching to 2DS/2DO terms. He said it provides more granularity when talking with a customer. “We’ll understand what they are saying about the substrate and what issues could arise and helps me determine what the customer needs from us,” he said. The key is getting customers to see enough value in it to change. Bottom line: standardization on nomenclature helps suppliers, he said. The issue is getting the whole industry to adapt and change.
Veeco’s Rezwan Lateef also likes that 2DS/2DO clears up the silicon vs. organic question. 2.5D is more generic, and it doesn’t clarify the difference between high and low-density packages. His colleague, Laura Rothman Mauer, thinks none of them cover the bases well enough. 2.5D means more than just two dimensions but not quite 3D. But she said 2D/2DO leaves out glass as a substrate, which is now important to some applications. “What about 3DS, 3DO, and 3DG?” she said. Hmmm. We’ll leave that one for another poll.
Is 2.5D here to stay?
While I didn’t get any justifications for keeping 2.5D, it still got almost 25% of the votes. Unfortunately, once a term has been established and adopted by the industry, it’s hard to change. We’ll probably get to poke fun at 2.5D for years to come.
Thanks to all who participated in our poll. We’ll do another one soon! ~ FvT