Hugo J. De Man received his Electrical Engineering and Ph.D. degrees in Applied Sciences from the Katholieke Universiteit. Since 1974, he has been a full professor at the University of Leuven. De Man is one of the co-founders of the Interuniversity Micro-Electronics Center (IMEC). (Please see ISD Magazine's profile of De Man, January 2000, p. 8)

ISD : Besides your studies and teaching, what significance do you place on your time at the University of California at Berkeley?

De Man : At Berkeley, I had contact with all of the other Phil Kaufman award winners. First, I performed post-doctoral research and studies under Don Pederson.

I learned about the basic model for industrial and academic interaction that Don Pederson espoused. I arrived [at Berkeley] in 1969 in the middle of the flower-power world, but also found myself in a heaven of wild creativity. It was the time that ICs and Spice were born alongside each other. Don understood the subtle way in which to form the famous Berkeley team of Cory Hall's 5th floor from which the CAD world was born. He showed that CAD should always be written with a small C, a large A and a huge D - and that you can't make that happen alone. I learned from him that good CAD can only result from doing design yourself and by talking to industry to find out what the real problems are - even if they don't fit into the "ivory tower" of academia.

Don always maintained that the benefits of industrial interaction come from injecting reality and focusing on results, in what would otherwise be isolated research.

I have also always liked the Berkeley interaction model, where researchers from many disciplines work together and cross boundaries. The group interaction model works well to solve the more difficult problems, because the harder problems usually aren't easily broken down into discrete elements that fit cleanly into separate disciplines. In most academic environments, people tend to stay within their specialty and don't venture into other areas.

At Berkeley, in 1970, I'll never forget that Don sent me to attend ISSCC in cold Philadelphia. There I met the first Kaufman winner, Herman Gummel. We had some very exciting discussions about device modeling, which later inspired Roger Van Overstraeten, Robert Mertens, and myself to develop the first bipolar transistor TCAD program - a program that included heavy-doping effects and that received a best paper award at ISSCC in 1973.

I was impressed that such a famous and important person as Herman Gummel would talk to a "lowly" post-doctoral candidate. Here I found that the ability to work in a cross-disciplinary group contributed to better overall results in both design and in CAD

Then MOS came along and suddenly switched-capacitor circuits became important. So, I remember very well meeting another Kaufman winner, Jim Salomon - then at National - responsible for pushing us into developing standard cell (SC) simulation tools. As a part of the design and simulation group, I helped to develop more appropriate algorithms for switched-circuit simulations over the equation-based solutions used in Spice. We did so with my Ph.D. students, Guido Arnout and Jan Rabaey. The Diana-SC tool was, for a long time, a commercial product of Silvar Lisco that came out of the layout work of Willy Sansen and our mixed-mode simulation work in the late ý70s.

That was also around the time that Carver Mead and Lynn Conway published their famous book that, for me, became from then on the bible for teaching custom IC design, Introduction to VLSI Systems. Nelson Goncalvez and I were inspired by the work to develop the NORA CMOS circuits now used in most high-speed designs.

I always tried to mix the circuit design and layout activities because I saw how they were interacting.

In my second visit to Berkeley, I worked with Ernest Kuh, who had just finished his term as dean of engineering and was starting to work in the CAD area. He invited me to be on the advisory board of Cadence in those days. From Ernie, I learned that the only way to have a lasting impact in the EDA world is to base things on the fundamentals of math and physics - a fact often forgotten by today's "digi-kids" who expect that reality is a color plot out of Matlab.

UC Berkeley is fairly unique in having strong industrial connections. The consultants and professors are trying to fix real problems. The sharing of ideas across departments is very strong and helps create an environment of problem solving, rather than individual fiefdoms to be protected.

ISD : What do you see as some of the next important directions for research?

De Man : The research at IMEC is moving towards higher levels of language and synthesis for retargetable DSP-type functions. The language and tools should enable efficient generation of C code that can be reused as an IP component. The design world needs another layer of abstraction for system-level design. The language should be something like C++, added to the concept of concurrency, to enable the system design. For a working term, this will be a SOC++ language that will allow hardware and software optimizations at the architectural level. The language needs to be object-oriented to allow the design of the concurrent-communicating functions within the system.

IMEC is moving away from the PC as a driver of technology and doing more work in communications. An interesting area for research is the wireless, wearable systems. This class of system will need broadband RF capabilities, but the digital parts will use most of their energy for data transfer and not for computation. The personal-area network (PAN) will use from 10 to 100 ALUs with distributed memory and computing capabilities. The challenges will be to optimize the code for low power and minimum memory size. Power has to go down by a factor of 10 to enable the PAN.

The new systems will need new global architectures. They won't be SOCs, but system-on-packages; the passive components will be on multi-chip modules. The problems will be in getting the tools, design, and technology all concentrated in the right mix of people. The next generation of designs will need all of the design and requisite disciplines in one place - even if that place is a virtual one - spread throughout the world.

ISD : Do you see a future for research in big organizations?

De Man : The large consortia, like Joint European Submicron SIlicon Program (Jessi), MicroElectronic Developments for Euorpean Appliances (Medea), and Esprit, are only able to apply their efforts to part of the problems and functions. They all lack critical mass and only focus on a single part of the problem at a time, such as new technology for CMOS processing. IMEC, on the other hand, is organized to look at the full range of system-level issues in their research. They interact with other research at universities worldwide and work with industrial organizations like the Alba Center in Scotland to leverage the other work.

Research organizations need to incorporate more interdisciplinary studies and change from a large number of narrow specialties that don't interact to an integrated whole where the different areas interact. The academic system isn't conducive to developing useful applications and cross-disciplinary work, because the departments and researchers feel that they must be very focused on a narrow research specialty, which their peers can properly assess. The next generation of research needs to address more fundamental issues within a global system and across many methodologies. In most research organizations, the goal of the research is formalization. Otherwise, the work is just industrial experiments.

Wide-ranging research needs to respond to the global picture and champion the push for goals other than the simple solutions. The measure of quality for true innovation isn't the technical papers, but the demonstration of how to do something. The existing structures can't solve all the problems because they are trying to fit the solution into an existing flow. Researchers need to work on solving more than just today's problems. The academics have to give up their territorial prejudices and share. Professors are judged as individuals and their studies are also judged as individuals. This needs to change to have people and their work judged as a part of a team.

ISD : What will be the next important area for research studies?

De Man : Because the scale of devices is approaching the molecular or atomic level, we need to start looking at the possibility of biological interactions in the circuits. Sensors can be small enough to be directly injected into the body. The possibility of this level of sensing means that the fields of bioengineering, electronics, wireless, and MEMS will converge. This type of device will go far beyond the capabilities of today.

EDA has focused on the digital parts of the design process. The new fields will have to address biological, mechanical, and sensor functions. Only multi-disciplinary groups can have the range of expertise to solve the very large problems. The industry needs to look at the technology receivers - the users - to find the proper applications. Engineering schools will also need to have non-engineers involved in the curriculum, including areas like psychology, biology, and medicine. The whole electronics community needs to look back and see the pathfinders in the industry, because they also had to develop a new industry from scratch.

ISD : How can we manage the research in the future?

De Man : The very challenging environment for next-generation electronics design will require new manners of management. The group dynamics will become key to progress, so results will come more from suggestions than from orders. The need for ever-more creativity and freedom in the work is best served by an open environment.

A historical perspective of the previous Kaufman award winners indicates that all of them contributed something in addition to their main work. Pederson developed the industrial-academic interaction model and made research results freely available. Solomon worked on design and on EDA tools. Gummel understood models as a vehicle to improve designs. Meade was working on NMOS and saw the transition to CMOS and the performance advantage of CMOS. He helped the industry make the change from nodal CMOS to dynamic logic. Kuh changed from a system focus to a design-layout focus after being the dean of engineering at U.C. Berkeley. The really great inventions usually come from synthesized refinements. Often the principle researcher's greatest contribution comes from putting together a group that has the right mix of people in areas of expertise to make the breakthrough.