Last week I was in Belgium for imec's international technology forum (ITF). For me one of the high spots is always An Steegan's presentation on what is going on just over the horizon in process development. I will write about that in a week or two once we are through the insanity of DAC. But in the meantime, I sat down with An and got some details about her background and also her philosophy of what the role of imec (yes, they really do put it all in lower case) is in the semiconductor ecosystem. You can probably guess from her name that An is Belgian. She studied materials science and electrical engineering at Catholic University of Leuven. She then did a PhD at imec. But unlike many people, who a friend described as "they are born at imec and they stay there until they die," An was recruited by IBM in late 1999 and moved to East Fishkill in New York state. Her first boss was TC Chen, who is now head of "Yorktown," the IBM Thomas J. Watson Research Center. The first job she had to do was to take the 0.13um (or what today we call 130nm) and transfer it from technology development in Fishkill to volume manufacturing in Burlington. This was being thrown in at the deep end since her previous experience was all in an academic environment and ramping a process to volume is about as unacademic as it gets. She said it was the best schooling ever. She brought up a whole series of processes after that: 90nm, 65nm, 45nm, 32nm, 28nm. An was working in what was then the common platform alliance with Samsung, STMicroelectronics and GLOBALFOUNDRIES, so the later processes were being transferred to the partners too, not just internally inside IBM. So she learned a lot about foundry too, especially from Samsung. She felt proud when she saw silicon that she had worked on ending up in products like a Samsung phone. In that era, she also started to interact a lot with the advanced fabless ecosystem companies, who also took a deep interest in process, such as Qualcomm, ARM and Cadence. As a result, she build up a great network of contacts. Her last boss at IBM was Gary Patton, who is now CTO for GF (and also talked at ITF). In the first year of bringing up 20nm she left IBM and returned to Belgium and imec. She had been there for almost exactly 10 years. Another great mentor was Lisa Hsu, who is now, of course, CEO of AMD. One of the other key lessons that she learned was that you have to derisk what you are going to do, but at some point you have to take your chances. For example, when Intel and IBM decided the future was copper interconnect, there were still many open issues. For a start, copper was regarded as about the last chemical element you wanted anywhere near your fab. At imec, An feels that her job is to make sure that they have a big funnel of plausible future technologies. However, there is a lot that is about consensus building. The semiconductor industry depends on the equipment industry and the materials industry, and everyone has to agree on the direction. The R&D investment is so high that it is not feasible to build different equipment for each semiconductor company, or use completely different materials. A good example was the consensus that it was best to use Hafnium for HiK metal gate. I asked her about EUV. "We needed it yesterday" was her immediate reply. It is getting more and more critical to have it ready. One big issue is whether to commit to design rules that can only be manufactured with EUV (as opposed to multiple patterning) or be less aggressive and stick with something that will work even without EUV. Despite all An's experience at IBM ramping processes to volume, imec doesn't do that. Basically they work for two years on a process in a pre-competitive environment, getting tools and process and prototype libraries all working. Then the foundries take over and do the detailed work for a couple of years. Then they ramp to volume and start to manufacture commercially. But more and more of the process depends on data and results coming out of imec. One thing that comes over talking to An is how enthusiastic she is about the future, and how proud she is of what the semiconductor industry has achieved in terms of its impact on humanity. She takes her inspiration from what we have already accomplished. Going forward, it is clear that semiconductor process scaling is not the only important vector, and the world needs to be looked at more holistically as technology, systems and applications. Today An is the imec senior vice president of process technology development, responsible for the technical leadership and execution of imec’s CORE programs in the areas of devices, process, lithography, and design and CMORE activities such as MEMS, power, sensors and photonics. An will be speaking at the US edition of ITF, as always the day before SEMICON starts, in San Francisco on July 11th. But you need to apply for a code before you can register (details here ). Two days later on July 13th she will be speaking on the Cadence campus at the Distinguished Speaker Series. But you need to be a Cadence employee to attend that day, it is not open to the public. I'm sure she'll be speaking again at ITF in Brussels next year. If you really like to plan ahead, it will be held from May 16-17th 2017. Previous: Securing the IoT for Billions of Possible Intrusion Points
↧