Executive Briefing: Soitec CEO

The balancing act between finFETs, SOI and new materials at advanced process nodes, and how that is likely to affect SoC design beyond 20nm.

popularity

By Mark LaPedus
Semiconductor Manufacturing & Design sat down to discuss FD-SOI, solar and various technology trends with André-Jacques Auberton-Hervé, chairman and chief executive of Soitec, a supplier of silicon-on-insulator (SOI) substrates, solar concentrators and other products.

SMD: The digital process roadmap is moving in several directions. Some pure-play foundries will offer bulk planar processes up until the 20nm node. Then, they hope to move to 14nm-class finFETs. FD-SOI provides chipmakers with another technology option. How do you see the landscape?
Auberton-Hervé: The server market will drive finFETs. But going to finFETs is not always the right solution, especially where the market is heading. The market is heading towards system-on-chip designs for mobility. So, there is a segment in the market that requires technology to cope with today’s power and performance needs. In this segment, you don’t need to go to finFETs. FD-SOI is a good option for the industry. This is a cheaper way to offer mobility for system-on-chips.

SMD: Would you say FD-SOI and finFETs are competing with each other?
Auberton-Hervé: I see two parallel paths. These two technologies will co-exist in parallel in some ways. From a user point of view, high-performance applications like servers are where finFETs will fit in. On the other hand, you have systems-on-chip. The design cycles are very fast. There are other criteria like mobility, power, performance and cost. That’s where FD-SOI will play.

SMD: On the roadmap, the plan is to extend FD-SOI planar technology from 28nm to 10nm. Then the plan is to offer a finFET based on FD-SOI at 7nm. Is that correct?
Auberton-Hervé: We’re maintaining planar FD-SOI as long as possible for system-on-chips. ST is looking to extend FD-SOI for three generations. That’s what they have committed to do. We think that FD-SOI provides more value and lower costs for mobile SoC chipmakers. Then, for the server, we think finFET on FD-SOI is a cheaper way to make a finFET. We see this as a better path.

SMD: What about the FD-SOI ecosystem?
Auberton-Hervé: We have partners that use this technology. IBM and ST are sharing the development of the technology. We also have a foundry partner, GlobalFoundries, which is supporting this. In fact, the first fully depleted technology available from the foundries will be FD-SOI. Everybody agrees that we need to be fully depleted. FinFET is a fully depleted device, but the first fully depleted device in the market will be FD-SOI on planar. That’s easier to do. That will be available at the end of the year at GlobalFoundries.

SMD: The traction for FD-SOI has been limited to IBM, STMicroelectronics and perhaps a few others. Is FD-SOI gaining more traction in the marketplace?
Auberton-Hervé: The first FD-SOI product created momentum. For example, ST recently announced a very aggressive demonstration of the value of this technology, which was a system-on-chip for mobility. Separately, IBM and ARM recently indicated where FD-SOI is heading. They talked about a system-on-chip with an ARM core for mobile devices. In addition, ST has announced there are other fabless companies that will bring out products based on this technology. So the momentum is there.

SMD: Who are the new FD-SOI customers?
Auberton-Hervé: Obviously, I cannot disclose it. I’m sure ST will disclose it soon.

SMD: Soitec is leading a 450mm SOI consortium in Europe. By the time 450mm reaches production in the latter part of the decade, the industry may require devices with III-V materials. How do you see the 450mm market playing out?
Auberton-Hervé: It will take a bag of tricks to move to 450mm beyond 10nm. You will require materials engineering. We will require many new elements in the periodic table. You cannot avoid implementing the new materials to help with the functionality of a device. This will bring several layers like III-V and substrate engineering into the mix. That’s where Soitec will play a role. The materials engineering guys like Soitec, which provide substrate engineering, will offer the bag of tricks that will be required to move to 450mm, as well as EUV, 3D and III-V materials. Our technology is part of a toolbox that helps manipulate layers and atoms.

SMD: Can the industry afford to do everything at once? For example, the industry is migrating to 450mm, EUV, III-V finFETs and stacked die at roughly the same time.
Auberton-Hervé: Cost is the key issue. You also need to create profits and have a better return. From one point of view, one way to sustain profits in the industry is to extend certain technologies much longer. For example, I can see 28nm and 20nm technologies for SoCs being used much longer. That’s something that will create a better return. So, remaining on planar, and using existing libraries, is the best path until the industry moves to 450mm. Using technology longer is a good way to avoid falling off a cliff.

SMD: Is Moore’s Law relevant today?
Auberton-Hervé: At one time, roughly 100% of the market revolved around the PC. Now half of the market is in mobility, so now you have two kinds of ecosystems. The industry is moving in parallel with these two markets, which are very different. There are separate needs for the MID (mobile Internet device) and server/PC devices. That’s where I think you need to retune the mobile industry, which has been driven by Moore’s Law. Moore’s Law is still alive and has been driven by the server. But given the economic aspects, it’s difficult to maintain it. You can’t sustain the mobility model with the existing drivers in Moore’s Law. We need to have a mobile Moore’s Law. With this, of course, the average cost is lower at each node.

SMD: Let’s talk about Soitec. The company is not only making a big push in FD-SOI, but it is also making some waves in RF SOI.
Auberton-Hervé: We launched two major products last year. One was for RF. The portfolio has been successful. The other one is obviously for digital, with 2D and 3D FD-SOI. We see 2D FD-SOI ramping up nicely. Regarding 3D FD-SOI, we believe this is a cheaper way of doing finFETs. We see good traction here.

SMD: Where are you seeing the growth?
Auberton-Hervé: Digital takes longer. Meanwhile, 40% of our growth last year involved the RF business. We’ve had huge success with our HR SOI, or high resistivity SOI. For RF, in fact, we have a full range of products, such as gallium arsenide, silicon-on-sapphire and HR SOI. So we see a lot of products moving to one of our substrate technologies. For example, the portion of RF is getting bigger. 4G is coming. And all of these things are pushing for more and more selectivity. That also means the silicon platform requires more integration. In that case, HR SOI is gaining momentum for the integration of switches, PAs and all of the modules.

SMD: Soitec is also making a major push into solar with a PV concentrator. What’s happening in solar?
Auberton-Hervé: The market for solar has completely changed in two directions. One is geographically. At one time, 80% percent of the market was in Europe. Today, 80% of the market is outside of Europe. Most of the growth markets are in regions with more sunlight. The market is also evolving from a top-of-the-roof model, which was restricted to the United States and Europe, to the point where 30% to 50% of the projects and the volumes in terms of kilowatts installed will be in large-scale utilities and solar farms.

SMD: Which solar market is Soitec focusing on?
Auberton-Hervé: We are not a top-of-the-roof player. We are concentrating on the large-scale utilities and solar farms. In that market, the panel cost is not a benchmark. The electricity cost is the benchmark. The market for large-scale utilities is booming. That’s the market we are shooting for.

SMD: What is Soitec’s strategy in the long term?
Auberton-Hervé: In 2015, our target is to have the same activities in solar and electronics. The growth will be balanced between solar and electronics. We expect growth in both. Solar will see very strong growth over the next three years. We are also looking for growth in lighting, which involves LEDs. That’s where the Soitec Group is positioning itself.