Experts At The Table: The Sky Isn’t Falling

Last of three parts: More on modeling; Moore’s Law; imbalances; the challenges and impact of stacked die, and when they’re likely to show up.


By Ann Steffora Mutschler
Semiconductor Manufacturing and Design sat down recently to discuss how the industry is making 3D ICs a reality today with Sylvan Kaiser, chief technology officer at Docea Power; Steve Smith, senior director for 3D-IC strategy at Synopsys; and Ahmed Jerraya, director of strategic design programs at CEA-LETI.

SMD: How far are we from having the models we need to enable 3D ICs? Are we going to get them from the foundries?
Smith: You do inevitably have to get basic models from the manufacturer or whoever is supplying the source of the materials even. We’ve talked about models of manufacturing materials—chemicals even. There are models for everything. But certainly the consumer that we’re talking about in this room is a semiconductor design engineer or a packaging engineer so they’re going to need to know what rules they need to follow to get their device manufactured within enough reliability, performance, longevity and cost in order that I’m going to be able to take responsibility on behalf of my company or my employer. It goes back to the first question of who is responsible. It’s the guy who is writing a check, but there’s trust. And that trust is done through a formal contract, a set of models. The models come from the engineering work that the supplier has done and the work that they do is often done based on work that their equipment manufacturers did alongside research organizations and universities around the world.
Jerraya: In fact what you have today you have many of the pieces. The work today is to put these pieces together and make it simpler, more automated. Process related modeling is almost there. We know all of these have been done for some cases, for some technologies; now the point is how to make this all automated and inserted in the tools, all the backend tools, all the modeling simulation, all the physical simulation, the power, thermal, mechanical simulation—all those things we need. The pieces are there.
Smith: We were talking about collaboration earlier. I see a lot more now even within our customer base that the semiconductor design companies historically have had separate departments responsible for each of these pieces: a packaging department, an IC department, a CAD or EDA department, IP modeling department, and more. For the first time I’m starting to see them have meetings with their vendors together, which hasn’t happened in my experience. I’ve been in EDA for over 30 years and I haven’t seen that before so I think they’re forced to talk more closely to one another inside the organization because the challenges are coming from the combination of the package, the interaction with the ICs, the stacking, the foundries and the vendors. It’s happening naturally.
Kaiser: Something also that is encouraging from my viewpoint is that the possible targeted systems with 3D have been clarified, and that’s very encouraging. Now we can clearly see different products that can be targeted with 3D and that’s very encouraging because it gives the objectives to reach. And as an EDA provider this really gives the direction to follow. Now with 3D we can really see that some products, maybe memory cubes, there are the interposers, there are the memories on digital logic—these are very clear targets. This really gives the path towards the tools development and collaboration.

SMD: What is the biggest weakness other than the integration piece that we talked about?
Smith: It’s not any weakness. I think it’s more just balancing in the force of business. It only becomes unbalanced if technology is no longer able to deliver reliably the next generation. Everybody’s been predicting Moore’s Law is going to die. Well, it’s still going on, and we’re talking about 14nm. With packaging, they’ve been talking about wire bonding eventually giving out or memory design. We’re still dealing with DDR and low-power DDR, DDR2 and DDR3. Each one of those extends the lifecycle of a certain technology so the question is, at what point does it become imbalanced for a particular company who has a need to fill? When that happens they have to fill the void with the next jump, and in this case it will be 3D or 2.5D. It’s always the same question. It’s not a matter of if, but when. The technology is there. I think we are starting to see signs now of life with a certain set of classes of application. We’ve already seen memory design done in stacks so we know that that can be done. We have also seen 2.5 D being done with a number of companies like Altera, Xilinx, TSMC, and the next one seems to be showing signs of life right now so if we are lucky we might see next year at DAC an actual physical example. I know we’ve said that for many years, but it actually as long as the imbalance stays there it’s going to get filled with a real device very soon I think.
Jerraya: 3D has been around for long time, and there has been a lot of hype around it. Today we see the light in the tunnel. Coming back to the question of the strength and the weakness, it relies on the business part because there are two kinds of forces. Some people are willing to control this 3D and take the benefits, and some people are scared of it because it’s going to change business. In fact, what we are near to making, which is 2.5D, will create a change in the PCB business in system integration. It changes in terms of the packaging part but also the procurement of chips, because we are going to buy chips at advanced nodes provided by some design houses or whatever they are, and we will then package them into a system. This is new, and it is a step further in integration.
Smith: Even though we’ve been doing this for years we could be completely thrown off by some random element. We will look at our other favorite topic: smart phones. Who would have thought five or six years ago the phone companies could have a game changer? It wasn’t even new. It was existing technology. But somebody with the right idea and the integration skills that made it happen. The same will be true here.

Leave a Reply