Though I’m interested in both history and technology, it’s unusual for both of those interests to be served side by side. But upon arriving in Leuven, Belgium, recently for the Interuniversity Microelectronics Centre’s (IMEC’s) annual technology review meeting, I could not help but be struck by contrasts between the very old and the very new.
Leuven, the provincial capital of Flemish Brabant, is an utterly charming college town dominated by its Catholic University, which dates to 1425 and is the oldest Catholic university still extant. It is also home to the historic Grand Béguinage, a medieval walled compound for semi-monastic women that’s older still, having been founded as early as 1205. Walking the quaint cobblestone alleys of the Béguinage on a mild Sunday evening conjured vivid images of a much simpler time.
But moving from the cobblestones of the Béguinage to the clean rooms of IMEC presented me with a vision of the future that was as bracing as the glimpse of the past had been soothing. IMEC is a micro- and nano-electronics research center founded in 1984 that has as its mission the exploration of technologies that are anywhere from three to 10 years ahead of industrial requirements.
A LOOK AT THE FUTURE
Over two days of intense presentations, demos, and talks, IMEC covered the full breadth of its activities, ranging from sub-32-nm CMOS work to intelligence in software-defined radios to body-area networks and neuroprobes. The effect was of a broad technology fair at which one could peruse a range of offerings that were only slightly sci-fi flavored. Many of IMEC’s research endeavors are well along and demonstrable. We’re not talking about Star Trek’s transporters.
Consider, for example, IMEC’s work in 3D stacked ICs (SICs), which it demonstrated in the form of the first functional 3D ICs created using die-to-die stacking using 5-µm copper through-silicon vias (TSV). According to Eric Beyne, IMEC’s scientific director for 3D technologies, his team will go on to further develop 3D SICs on 200-mm and 300-mm wafers, integrating test circuits from partners participating in its 3D integration research program. Eventually, the technology will find its way into production.
Looking down the road, some industry experts foresee mobile handsets encompassing as many as six separate radio standards by 2014. With that in mind, IMEC has devised a highly power-efficient, extremely flexible multiprocessor platform that’s already caught the eye of Toshiba Corp., which has licensed the technology. According to Serge Vernalde, technical business director of IMEC’s Nomadic Embedded Systems program, the goal is to develop cognitive reconfigurable radios with gigabit/second throughput.
“Such systems will have sensing engines to find signals across the spectrum and to automatically determine which radio type to use for transmitting,” says Vernalde. For baseband processing, Vernalde sees an evolution toward more than two baseband engines and toward flexible error-correcting codecs.
An impressive part of the technology package licensed by Toshiba is the MPSoC toolset. One portion, called CleanC, allows designers to write sequential, high-level code that is optimized for parallelization. The resulting sequential C code is subsequently mapped onto a multiprocessor platform.
MPSoC relieves the designers of having to code synchronization, data communication between threads, and memory organization. Thanks to the parallelization tools, for instance, designers can explore several multithreaded versions of the same application in a short time.
The crown jewel of IMEC’s microtechnology research is its “More Moore” program, intended to extend Moore’s Law to its theoretical limits. IMEC is probing lithography options for the 22-nm process node. According to Kurt Ronse, director of IMEC’s advanced lithography program, it’s now clear that high-index immersion lithography is not the long-term answer. “Development of high-index materials and fluids other than water is not going to happen in time,” says Ronse.
So with double-patterning techniques as a stopgap measure for 45 and 32 nm, IMEC is moving full-speed-ahead toward extreme ultraviolet (EUV) lithography for the 22-nm node. Already up and running in IMEC’s clean room is an alpha demo EUV system from Dutch lithography giant ASML.
With an upgraded illumination source from Philips, the system is capable of four wafers/hour, but that’s still ramping up. A further illumination upgrade is expected later this year to improve throughput. Evaluation results have so far exceeded expectations.
EUV is just in its infancy, though, and is bound to make rapid progress. At ASML’s Veldhoven facilities, however, the company is continuing to improve its technology for doublepatterning immersion lithography with new TwinScan NXT steppers boasting lighter, faster wafer stages that will boost stepper productivity by over 30%. This impressive technology will carry through to ASML’s production EUV equipment, currently scheduled to ship in 2010.
So, the cooperative research-center model is alive and well for the development of emerging technologies in a wide range of spheres. With the help of partners such as ASML, here’s hoping that the organization will deliver the goods to keep Moore’s Law afloat for some years to come.