Placing Bets On Future Technology

One-On-One: Leti’s CEO talks about what’s next for different markets and why.


Marie Semeria, CEO of Leti, sat down with Semiconductor Engineering to talk about where the French research and technology organization is placing its future technology bets and what’s behind those decisions. What follows are excerpts of that discussion.

SE: It’s becoming more difficult and expensive to shrink features, so where do we go next?

Semeria: We see several areas that we believe are interesting. One is CoolCube (monolithic 3D), which we think is disruptive. Leti pushes new types of technology applications, and in 2008 we looked at a lot of Ph.D. work to figure out what was the probable direction technology was heading. We had two concerns. One was the continuation of low-cost scaling as you suggest. The second was open, full integration. This is why we are working with ST, GlobalFoundries and Qualcomm to scale up FD-SOI to create a full ecosystem.

SE: We’ve been hearing about 3D stacking and FD-SOI for awhile, but so far the uptake has been slow. Why do you believe they are the future?

Semeria: We’ve been able to demonstrate and evaluate the gains made in an array in terms of power and cost. That includes finFET on finFET, and finFET on FD-SOI. We are discussing which one is best. Our real interest is cost. We are looking for low-cost scaling. So CoolCube is the N-plus-1 generation. But to get to N-plus-N, you need to show a great gain in time to market, cost and performance. That requires a full process flow with technology that is compatible with the requirements. We’ve been testing the cost and performance of different architectures.

SE: Where does FD-SOI fit in?

Semeria: FD-SOI is happening this year. So 28nm has similar performance with lower cost compared to bulk at advanced nodes. What’s unique is that it allows us to push Moore’s Law further. With finFETs you get high turns and high performance, but it also has a high cost. With 28nm, you don’t have to switch nodes. FD-SOI opens a new path, and one that we think is especially exciting. It can go further than finFETs with lower cost, which is good for consumer markets.

SE: How far are we talking?

Semeria: FD-SOI can push Moore’s Law below 10nm with low cost and low energy. GlobalFoundries announced that its Dresden fab it can manufacture 22nm FD-SOI, which is based on a license with ST and Leti. We signed an agreement to support them at 22nm in Dresden. So 28nm and 22nm are now in full alignment. Today we have 28nm fabrication at ST and Samsung, and ST has licensed GlobalFoundries’ 22nm technology. Leti is working on 14nm and 10nm with support from the full ecosystem.

SE: What’s Leti’s goal?

Semeria: Our commitment is new technology. More than 20 years ago we created SOI. We are now seeing the transformation of that technology in to products.

SE: Where does CoolCube fit in?

Semeria: We have a full roadmap in 3D. Leti is going vertical. We have a full library of IP and technology. CoolCube is part of the 3D integration.

SE: Which markets will take advantage of that?

Semeria: The market is going two directions. One is data storage, servers and consumer, which requires high performance. Another market is the broad market we call IoT. It still has to be better defined, but it’s the autonomous driving market first, medical devices, which might be second, and wearables, which is third. Those require technology with very good performance, low power and low cost, and FD-SOI can answer this market.

SE: How do you see the IoT playing out?

Semeria: There is an IoT platform approach, where you can add RF and MEMS on a mobile platform. The other trigger is 3D integration, where you pack functions into a minimal system.

SE: Let’s change topics here. Where does Leti get its funding?

Semeria: 20% is from the French government, 80% is from partners. The position of Leti is as a leading center of innovation. But innovation is worldwide. You have to work with the best and keep a sharp eye on the competitiveness of the industry.

SE: What about the design side?

Semeria: We have started the Silicon Impulse Initiative to set up an EDA ecosystem and get access to emerging technology for small companies and startups.  Foundries and designers can assess capabilities on FD-SOI, and they can ask Silicon Impulse for access to design enablement technologies. We help develop PDKs and IP for 28nm and 22nm FD-SOI technologies. So small companies can assess designs in FD-SOI. If they join a design center, they have access to the full design center, which allows them to assess technologies before going to a foundry. It’s a complementary offering to a foundry. We do the same with CoolCube.

SE: One of the big issues today is security. How active is Leti there?

Semeria: We have a national center dedicated to security. We are able to assess everywhere in the world the robustness of chips versus threats. We handle security issues throughout the full chain, from hardware up to the physical layers of software. This is a high priority program for us. You have to secure every channel and determine at which point you develop security for devices and functions. But you have to do this while keeping up speed and performance, which may require changing the pattern of information.

SE: How so?

Semeria: One new approach is neuromorphic. We need to change the paradigm regarding security. We have to define an innovative solution and think globally. Technology is more connected. Some of it is used for taking care of people. If you look at smart watches and wearables, those have to take into account security, connectivity and the sensitivity of data.