Paul Penfield, Jr., Opening Remarks MTL>30 Symposium, MIT, Cambridge, MA Oct 29, 2014 [Final spoken version] Hello. MTL is now 30 years old. Today its successes are being celebrated. I have been asked to tell you about how MTL got started. I can do this because I was present at the birth, 30 years ago. The story starts earlier, though. The semiconductor revolution began in 1947 with the point-contact transistor. Then in 1952 the junction transistor, then MOS transistors and integrated circuits in 1960. The MIT EE department decided to respond to the semiconductor revolution with the highly successful SEEC courses, as Paul Gray just described. But we made another fateful decision then: MIT did no research on silicon integrated circuits because we felt industry could do it better. Indeed, industry WAS doing very well, going from Small Scale Integration to Medium Scale Integration to Large Scale Integration. Circuit design, the bread and butter of many electrical engineers, was also changing: analog to digital or mixed-signal, bipolar to MOS, and discrete transistors to integrated circuits, even entire computers on a chip. Without research in silicon integrated circuits, we gradually lost touch with the state of the art. We found ourselves still teaching forms of electronics rapidly falling out of favor. And then along came VLSI, Very Large Scale Integration. We needed a strategy. Fortunately, we knew how to get one. We had just made another important strategic decision. The rise in the importance of computers had led some to want to split our department into separate EE and CS departments. Lou Smullin, the department head, faced this issue head-on. He organized a department-wide conversation. We decided in 1974 to stay together as one department, which was promptly renamed EECS. This experience prepared us for another department-wide conversation, this one about VLSI. We could not fabricate chips anywhere near the state of the art. We could not even ask industry to make chips for us because industry's design rules were proprietary and constantly changing. And yet we knew that the future of electronics and indeed most of our department's favorite technologies would involve increasingly complex chips. Industry was also having problems coping with their own success. There seemed to be three categories of questions they, and we, were facing about VLSI chips: 1. How do you make them? 2. How do you design them? 3. What do you put on them? At this point we decided we had to jump in with both feet. Our very leadership position in EE and CS was at stake. For our VLSI program to be true to our department traditions, we would need to 1. include both education and research (our teaching, even undergraduate, could be informed by our research and vice versa) 2. include both EE and CS (that is, both electronics and computation, both hardware and software) 3. encourage people in other disciplines to make creative use of our facilities and results 4. let other universities use any unique facilities 5. cooperate with industry (they want our graduates and our ideas; we want their guidance and their willingness to develop our research results further) So what did we do? In January 1978 ten EECS faculty, half EE and half CS, took an industrially oriented MOS chip design course. This helped bring us up to date. It also gave the CS faculty some ideas of what could be put on a chip (recall category 3, what do you put on VLSI chips). In Fall 1978 Lynn Conway taught the first VLSI design course here at MIT, letting students who knew nothing about semiconductors design chips. This was the beginning of the national VLSI Revolution (recall category 2, how do you design VLSI chips). But what about category 1, how do you make VLSI chips? We concluded that we could not do research in this area without a fabrication facility. We would need to run entire processes to evaluate innovative process steps in context. We could not rely on using industrial lines, which were designed for throughput and yield. We would need to run many processes at the same time, some even involving novel materials. We needed flexibility, not throughput. In January 1980 we announced our intentions and the rest, as they say, is history. By 1984 MTL was up and running. In 1986 we named MTL's building after former Dean of Engineering Gordon Brown (with T-shirts made for the occasion). Wildly innovative chips were designed and fabricated, many incorporating sensors and actuators. The beginnings of modern 3-D printing came from some of the MTL people. And it still goes on. Ideas from semiconductor fabrication keep getting applied to other disciplines. Today it's biological and medical engineering. Tomorrow it will be MTL helping make nanofabrication available to all MIT researchers under the exciting MIT.nano project. Will that be the end? Certainly not. If MIT becomes a leader in the new field of quantum engineering, MTL will make the circuits and devices that store qubits in quantum memories, that apply quantum error correction, and that perform quantum information processing including computation. MTL will also make the circuits that pass qubits between chips and that convert classical bits to qubits and back again. One thing you may have noticed is that I have been speaking from a departmental perspective. Yes, it was EECS that, of all MIT departments, most clearly saw the need for MTL. And it was EECS that first administered MTL, while the technical people worked on what they do best, the technology. But EECS made sure MTL was serving researchers from all over MIT, right from the start. Although technically a department lab, it was behaving like an interdepartmental lab. I know because I was, as I said, present at the birth of MTL. I was also present ten years later when Rafael Reif, then MTL Director, decided MTL was strong enough to be an officially recognized interdepartmental lab. I happened to be serving as EECS Department Head at the time. People asked me how I felt about losing control over what I had worked so hard to create a decade earlier. My reaction was not unlike the intense pride I felt at my daughter's wedding. I just said to myself, "well, I guess I've done my job!" Thank you for your attention.