CEO interview: Semeria ushers in change at Leti
Leti, based in Grenoble, will be 50 years old in 2017. Over its history it has been substantially funded by the French government and has built up a reputation for excellence in fundamental electronics and semiconductor research as well as generating numerous spin-off startups. But with Semeria taking the CEO position there has also been a re-evaluation of the best way to achieve its purpose. The conclusion is that Leti is still about R&D to achieve miniaturization and energy efficiency but with a recognition that the industry is global and has moved to the system-on-chip and Internet era and that applications are changing rapidly.
Semeria is an experienced administrator but at heart an engineer. In her early career she worked on magnetic memory with Sagem and then moved to become chief architect with PixTech Inc., a now-closed 1992 Leti spin-off that developed flat panel displays based on arrays of micro-tip electron emitters.
Semeria then spent from 1996 to 2012 within Leti, rising to the position of deputy director before spending three years as chief scientist at Leti’s parent organization CEA Tech. The French government owned and controlled CEA is the Commissariat à l’énergie atomique et aux énergies alternatives.
“We need to take a global view and not just about the device. The technology-push approach is not enough,” Semeria told eeNews Europe.
The Leti campus in Grenoble.
The near 50-year history of Leti can be characterized in two phases, Semeria said. The first was one of scientific and engineering study to develop electronic technologies of strategic interest to France and to underpin the French communications and computer industry sectors. The second phase added to that the responsibility to help move such technology out into society partly through the creation of startup companies.
“Now there is a third-phase; of customer pull. We need to understand the applications and use cases, build the specification and then supply the technical building blocks to meet system-level requirements,” Semeria said.
That doesn’t mean any turning away from developing hardware technologies, Semeria said. “We still need to be first – to pioneer technologies – but you have to bring value to your partners.” Leti’s work will include much more software and system-level design, she agreed.
Last year Leti’s budget was €315 million (about $350 million) of which only 17 percent was paid by the CEA. Of the remaining 83 percent about half came from collaborative research grants either supplied by the French national government or the European Union. The remainder comes from the industrial companies Leti partners with. “We are free to grow but the direct support is flat,” said Semeria. Leti has 350 industrial partners and about 55 percent of them are small or medium-sized enterprises (SMEs). There are 1,300 people on the Leti payroll but when you include assignees the number of researchers goes up to 1,900, Semeria said.
The company has also produced more than 60 startups and is currently creating them at a rate of 5 or 6 a year. But has the level of startup success achieved by Leti justified the hundreds of millions of francs and euros spent over Leti’s history? In response to this question Semeria references STMicroelectronics, Soitec and Sofradir as just a few examples of the spin-off of Leti technologies. One can dispute the levels of success these and many other Leti spin-offs have achieved, including PixTech, but Semeria argues that the training of engineers in leading-edge technology also brings direct and indirect benefits.
“The ecosystem around Grenoble is now part of the offer of CEA-Leti. We have now reached the point where 50 percent of local technology startups are out of Leti and 50 percent are independent and come for access to our technology platform and services, such as engineering on 200mm and 300mm wafers,” argued Semeria. “This Grenoble clustering effect is also generating interest to invest in startups created by Leti,” she added.
And such boldness creates its own momentum. So now entrepreneurship is entering into second and third generations. Semeria gives the example of eLichens SA (Grenoble, France) formed by managers who had previously formed Movea SA (Grenoble, France) and then sold it to US company InvenSense Inc. (see Audience, InvenSense buy up sensor fusion software firms). Movea, founded in 2007, developed sensor-hub software and firmware that gathers data from inertial sensors. Now eLichens is developing optical MEMS technology for measuring air quality.
European expertise
Leti’s areas of microelectronics expertise currently include: fully-depleted silicon-on-insulator (FDSOI) chip production and circuit design; MEMS and NEMS; magnetics, packaging and 3D technologies and photonics. And all of these areas are active, Semeria said. “FDSOI we have demonstrated down to 7nm. MEMS we are moving in to biology platforms. Magnetics is evolving into spintronics and how to use spin for memory, logic and sensors; and photonics is coming on chip,” she said.
Looking forward: “We are working on the next platform for semiconductors based on nanowire transistors, we have also demonstrated a quantum bit [computing element] on a 300mm CMOS wafer. Quantum computing is part of high performance computing and it is now up to Leti to begin transferring quantum computing to industry. But it can’t be a technology-push approach. For example embedded security has to be considered at the very beginning. And embedded computers will be a mix of quantum, neuromorphic and von Neumann computer architectures. However, we must start with a vision of the application.”
This will undoubtedly place demands on Leti, just as it does on other research institutes, because applications are numerous and diverse, particularly for the Internet of Things (IoT), and markets fragmented. But this is a nettle that Leti has to grasp, Semeria indicated, because system knowledge is now getting integrated within in SoCs and other basic building blocks.
“IoT covers industrial process control, agriculture, food; automotive, autonomous driving and highly connected vehicles; medical devices may be slower to arrive because of legal requirements but still digital medicine and diagnostics will come.” Semeria made the point that just as the electronics industry has moved up to an application-specific platform-style of design that includes hardware and software, so must Leti. “We must make the move to the convergence of physics, technology and embedded software. We have to be in touch with the different application domains. So it is policy to recruit staff from these areas – from agriculture, from banking, from industry and so on,” Semeria said.
European R&D
We asked Semeria if there is a policy to perform R&D that is complementary to that of other European research institutes such as IMEC in Belgium and the Fraunhofer Institutes in Germany.
Semeria said that she had just come back from a meeting with Fraunhofer and Finland’s VTT and that they meet several times a year to exchange understanding about the research environment. And complementarity is clearly more efficient than duplication. “Take the example of the FDSOI process, which STMicroelectronics transferred to Globalfoundries in Dresden. We supported ST so it is natural that we support Globalfoundries which we do in the front-end, while Fraunhofer provides support in the back-end.”
“IMEC is quite different. Proximity to ASML and the pioneering of extreme ultraviolet lithography is their main differentiator. But they also have a different business model.” Whereas IMEC tends to create industry affiliation programs with multiple industrial partners working on pre-competitive research Leti’s partnerships tend to be two-party affairs, Semeria said.
And a partnership with Intel is an example of this, Semeria said. “We decided to work together with Intel on IoT (see Intel, CEA to pitch jointly for European funding). We have a similar partnership with Globalfoundries on FDSOI. We have to deal more and more with international companies. We are leaders in healthcare, automotive, aerospace, and companies need help with all the silicon-based components and not just with the digital part.”
European manufacturing
Leti may help France and Europe punch its weight in technology research but we asked Semeria if she is concerned that Leti is running out of European partner companies that can receive its technology and run with it. After all European tax payers might prefer that their euros end up benefiting European rather than foreign companies. We also pointed out that Leti partners Intel and Globalfoundries, although two of the biggest semiconductor manufacturers in Europe, are controlled from outside Europe and their wafer fab sites in Leixlip, Ireland and Dresden, Germany, are mature.
Semeria said manufacturing is important and Europe has world-class players such as STMicroelectronics and Robert Bosch. “We need manufacturing in Europe. It is a key for innovation,” she said. “Innovation is in the hardware but there is also great space to innovate in integration, in firmware, in sensing and security.” That’s an opportunity for Europe, she said.
Semeria added that part of Leti’s function is to enable a rebirth of industry in Europe. “New businesses can leverage new approaches in things like augmented reality, artificial vision, in micro-displays,” she said. “I want to move Leti to a more solution-oriented approach, a more system-level approach.” Leti will be a technology provider in a system and usage context, she said. “Demonstrators are key to showing that full solution.”
That presents a challenge because different parts of the system-level solution will have different time-lines in R&D. Quantum computing is potentially still years away from commercial deployment whereas a breakthrough application developed in software could be produced in weeks or months. Semeria acknowledged that challenge and said that Leti has to work on these different timescales.
Semeria gave the example of how years of work on FDSOI at Leti has been complemented by the formation in 2015 of Silicon Impulse, a Leti initiative to ease access to FDSOI and help companies move to mass production. Silicon Impulse is a one-stop shop for most of the engineering support needed to make a successful IC design in FDSOI; feasibility studies, foundation libraries, essential IP, multiproject wafer runs, packaging and test. Silicon Impulse can even provide design services to help with IC designs (see Silicon Impulse web page).
France was one of the pioneers of consumer internetworking with the development and launch of the Minitel terminal and services in the 1970s and 1980s. It was a nationally-backed Videotex online service accessible over telephone land lines that allowed users to have electronic mail, make online purchases, check sports results and stock prices, all long before the introduction of the world-wide-web.
We asked Semeria if France is likely to make a similar national commitment to provide infrastructure for the Internet of Things. She answered that the way forward was more likely to be numerous test beds. The key would be to develop pilot schemes for domestic, urban, automotive and industrial IoT, to make them work and then be able to scale them up, she said.
And end-to-end security and security certification would be an essential part of such Internet of Things applications, she added. And that, in turn, was the reason she had been eager to promote cooperation between cyber security players, including Intel, at the Leti Innovation Day in Lyon where our interview was conducted.
Related links and articles:
News articles:
Audience, InvenSense buy up sensor fusion software firms
Intel, CEA to pitch jointly for European funding
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
