On the occasion of this, the tenth edition of the MIT MTL Annual Report, it seems appropriate to recall how MTL got established. I was there at the time, along with Professors Richard Adler, Gerald Wilson, and Paul Gray, and I saw it grow to its present state of maturity.
Before 1975, MIT did not have a serious research effort in silicon microelectronics. Rightly or wrongly, the theory had been that industry was making such rapid progress that there was no room for a university, with its inherent long-term outlook, to contribute.
During the first fifteen years after the invention of the integrated circuit in 1960, the research problems had to do with fabrication. That is, how do you actually make integrated circuits -- how do you carry out the processing steps needed to make devices and circuits?
In the mid 1970s the complexity of integrated circuits had become so great that manual techniques for designing them no longer worked. Many universities, including MIT, started research in VLSI design automation, and many of these activities continue to this day. However, at that time we realized that there was yet another class of research that would require attention from universities, and that had to do with what could be put on integrated circuits and systems of the future. This included analog circuits, it included digital systems such as computers, but it also included systems incorporating novel physical, mechanical, electrical and chemical structures.
To answer this need MIT, like many other universities, started research in VLSI computer architecture, and also, unlike most others, research on other types of systems. We knew that to do this we would need to fabricate structures using special-purpose processes, at large scale. It would not suffice to make a small number of structures. It would not suffice to use only conventional processing sequences. And most of all, since special processes usually use many process steps, most of them standard, it would not suffice to run standard steps or processes at less than high yield.
All this led to the MTL facility that today can run special-purpose processes reliably and repeatedly, at large scale. Naturally, to run unusual processes MTL first had to master standard ones, and to make circuits at large scale it first had to make small ones. But it also had to organize itself to maintain the discipline needed to fabricate at high yield, in the presence of different process flows every day. And it needed an information-handling system that welcomed unusual process flows and could provide device designers with both simulation and fabrication, driven from the same process description.
MTL now does all this, and does it really well! The ten Annual Reports in the series you are now reading document many advanced process flows leading to quantum systems, microelectromechanical systems, high-speed devices and systems, and many other systems using unusual process flows. At the same time MTL has contributed much to the understanding of all kinds of process steps. Those of us who helped start MTL take great pleasure in its success, and salute its directors, Professors Dimitri Antoniadis and Rafael Reif.
Originally MTL operated as a laboratory within the Department of Electrical Engineering and Computer Science. However, as it became a resource for people from all over MIT, it seemed appropriate that it become an interdepartmental laboratory, reporting at a higher level within MIT. This administrative change took place a year ago.
I was present at the birth of MTL. I was one of those who persuaded our friends in industry and government to establish and support MTL. For many years, I helped coordinate MTL and other VLSI efforts on campus. Then, between 1989, when I became EECS Department Head, and last year, MTL reported to me.
Some people asked me how I felt about MTL leaving the department. The closest thing that I could think of was the intensely proud feeling I had when my daughter got married.
Paul Penfield, Jr.
Professor of Electrical Engineering
Head, Department of Electrical Engineering and Computer Science