Changing the World through Materials Innovation


Geoff Beach: Welcome, welcome everybody welcome to the 2016 Fall Wulff Lecture. My name is Geoff Beach, I’m a professor in the Department of Materials Science and Engineering and the Chair of our Committee on Undergraduate Studies so it’s a pleasure to welcome you to see Professor Gene Fitzgerald but before we begin, I want to give you a little bit of background about this lecture series and John Wulff whom this lecture really honors. So John Wulff was a professor in the Department of Metallurgy, the predecessor to DMSE from 1937 until his retirement in 1973. Among his accomplishments, he was an outstanding researcher but he’s really known for teaching and education, really defining how we teach materials science and engineering to this day. One of the most important things that he did for MIT and for materials education in general was the founding of 3.091. So 3.091 was developed in 1969. It’s been taught in this lecture room for many, many years since then. Many people have taken 3.091. How many are taking it now? OK, so we’ve got a few freshmen here. Are there other freshmen who are not taking it? OK, well, we’re hoping to give you some good introduction to materials engineering today. So John Wulff was a very inspiring lecturer and inspiring teacher. You can see him here. The camera can’t quite capture him or capture his excitement; he really inspired imagination and interest in materials science and engineering both through the the lectures that he gave at MIT but also throug a book series of four volumes called The Structure and Properties of Materials that defined how materials was taught, really around the nation, redefined how we got this field. The inaugural Wulff Lecture was given by John Wulff in 1977. The aim of this lecture was to present an engaging lecture to the community but targeted at undergraduates, in particular freshmen, to give them a sense of what materials science and engineering and why it excites us as faculty and to inspire you to consider materials science and engineering as a way to go out and change the world in significant ways. We hope that this lecture will give you some some idea of what do we do in materials science and engineering. As engineers, and many of us, in fact, all of us, on the faculty here are engineers, and what engineers do. Engineers build things. Engineers build systems. Engineers build aircraft and engineers build bridges. Engineers build transistors, but one could really argue that the most fundamental engineering is materials engineering, materials science and engineering, because every complex system, every device that we try to build from components ultimately rests upon the materials and the limits of what we can achieve, and what we can achieve in these materials is defined by their properties, so really the materials performance determines what can be achieved. When we look at materials science and engineering in our department, we look at both the science and engineering aspect and we do it in a vertically integrated way. As scientists, we’re interested in processing of materials, how processing defines the structure of materials, and how the structure defines properties, and importantly how these properties can be useful. We’re not just interested in an abstract sense of what materials do and how they behave but what can we do with them, how can we change society. We also start from the other end; we try to identify the big problems of energy and information and health— what are the big problems? As you often find, in most cases, you find that solving those problems comes down to the materials, the properties, and the performance that we can get from them. Big advances often come from big advances in materials so by identifying these properties that would allow us to transform these fields, identifying the kind of structure that is required to give these properties, and finally processing allows us to make these materials, not just a laboratory-scale but a scale that can really change the world. That’s what we do and that’s what we are about and I think you’ll see some of that today. So Professor Fitzgerald joins a long line of lectures beginning in 1977 with the first Wulff lecture, who was of course John Wulff himself. Professor Fitzgerald is the Merton C. Flemings—SMA Pofessor of Materials Engineering at MIT and the lead Principal Investigator of the Low-Energy Electronic Systems Center of the Singapore MIT Alliance for Research and Teaching (SMART) Center. His research interests are related to new materials devices and circuits. One of his most important accomplishments was the demonstration of high mobility strained silicon in the integration of useful electronic and optoelectronic devices on silicon. Professor Fitzgerald is a practicing researcher, innovator, and entrepreneur. He is the founder of many companies which you’ll get some insight into today but he started all of this here at MIT. He was an undergraduate here, got his degree in materials science and engineering in 1985 at MIT, went out made important accomplishments and has come back to tell us about them so welcome. Gene Fitzgerald: Well, it’s a great honor to be here to talk to you today in the Wulff lecture and for those of you undergrads here at MIT, I was sitting in your chair a long time ago, not that long ago, and I’m hoping that I can relate to you by going through my history and what we’re going to do and what we do in materials science, which is really just what Geoff mentioned, going from atoms to commercial impact and it doesn’t happen overnight. It’s a different story than the latest quick app but the long ranging impact is impressive and that’s the kind of thing that we can do in materials science. I’ll finish with what we’re doing today because we continue to innovate and we’re continuing to think about whatever the next thing could be and so hopefully we’ll be able to come back to you in a few years with even greater impact. So I wanted to start at the beginning and I started thinking about it. Here I am with a blank page, and I had to go back and think about why I really got interested in materials science and engineering and that’s always hard to do. Because you forget about what you were thinking at the time. So I went back to actually what I was doing in high school and before and in elementary, junior high, and highschool and this turns out to be one of the more important moments. I realized that it’s a real chemistry set. I say “real chemistry set” because it actually has grams of material. I have kids (they’re in college now) and when they were young , I went out what to see kind of chemistry set I could get. Now they give you these little things like micro materials in there, it’s like you could do little tiny… but you can’t really achieve anything, you can’t blow anything up, you can’t… I was like, “well, how are my kids supposed to get excited about science when you can’t do that?” And the best part about this kit was not even grams of material but I convinced my mother to go to the drugstore and buy some saltpeter for me, which is potassium nitrate, which …you know. Just by looking through old chemistry books you could tell that, if I had the nitrate group I could do things like make gunpowder, guncotton, and stuff like that. And so I think it’s a really important thing to have real actionable things you can do that gets you excited. It gives you, in some sense, it gives you power over adults, right? You can do some stuff that makes your parents nervous and and that’s all very exciting. Another thing that I really do remember this, and I don’t know if you guys… probably they still have these things but nobody’s interested anymore but this was a huge thing— a hundred in one electronic kit and I don’t know if you can see because it’s a digitized image but it features space-age integrated circuit, right? So that little square thing in the middle was one of the first TI integrated circuits and I remember thinking this, I actually was playing with it, “you know, this isn’t that much better than discretes, you know, what do I care about this thing, right?” But it’s really telling, because you’re going to see at the beginning of all these innovations, they’re oftentimes, people aren’t really valuing what’s going to happen in the future. They can’t see the connection from the thing. As a little kid, I was looking at this thing I could connect up to all the other things and and why is this important, and so that’s going to be a continuing story and being able to look forward and get the basics here at MIT so you can kind of look into the future a little bit is a really important thing to try to figure out. “Why is that integrated circuit going to be really important in the future?” My greatest thing I ever did this with this wasn’t planned, wasn’t in the manual. I learned the magic, without realizing it, of oscillating circuits and then I connected larger power sources to this thing and was able to block television reception in my neighborhood so that was really exciting and again power over adults. So the next thing was when I was in high school, this thing came out; it was the Apple 2 plus. Now that was an amazing machine. This is high-end, man. It had color, right? Now it turns out I didn’t have a a color monitor and so I couldn’t use it initially but I wanted to show you that when this first came out you had to store the information as audio signals on a cassette player. So think about that, that’s kind of cool. Storing digital information that way because this drives being small or too… well, technology wasn’t there. The larger ones were were too expensive at the time, but I guess one of the interesting things is to show you the perspective on what people were thinking about back then. I actually programmed this for a local company and I thought how ridiculous, they’re gonna buy this machine, they’re going to pay me to program and I want to sell this whole thing and they’re just going to use that computer to run one piece of software, and it’s always like why would anybody do that? but it doesn’t matter, I wanted to make some money, so I had this beginning entrepreneurial streak which I completely ignored because I was really interested in just figuring out how to program this thing for what the customer wanted. It was a personnel agency, actually, it’s a long story what they wanted, but they wanted a database and anyway I was able to buy a stereo when I came to MIT with the money that I earned from that, but I think I don’t care so much about that because when I looked inside I was really curious about this thing which I programmed like crazy at the assembly code level it is the motorola 6502 there it says mos 6502 but of course it’s because people left motorola and created a company called MOS and then they you know got into lawsuit with Motorola but the amazing thing about this thing is that people have figured out how to create a depletion mode MOSFET so they could have enhanced mode, a depletion mode, and it saves a lot of power. You can shrink the device and it’s a very powerful small microprocessor and so I sat there saying okay, i can do all this programming but anybody who knows algebra can program, right? so I was thinking about what’s inside the black thing, right? What’s inside there? And you’ll see I really didn’t think this back then but we ended up actually affecting all these integrated circuits today so it’s really an amazing thing that MIT gave me that opportunity so I really wanted to learn more about what’s inside those things. I went to a public high school; I think I’m still only the second person from my high school over to get accepted to MIT and when I came here the reason I chose Course 3, you could tell from the previous slide an interest in electronics, interest in physics, interest in chemistry, and well materials science— I could have it all, right? I didn’t have to choose, I could work on all those things. It’s one of the main reasons, actually, I really was interested in Course 3. Now here’s our graduation picture. Here I am over here. it turns out that I (actually this isn’t working) I lost a lot of weight and and facial hair not— I’m joking, this is me over here. So what is interesting, we still chose to wear wear suits even though this was a time period when people didn’t wear suits too much but I guess we wanted to look professional. Chi Phi. Anybody here from Chi Phi? So I was alpha man, look at that, president of the fraternity. At some point— fools choosing me though. It’s funny the people you meet at MIT. So this guy over here, … I should be able to write on this, I know what I gotta do I can already see that’s… yeah, this guy Bill Irving is now the bond king for Fidelity so bond B-O-N-D, right? So he was an electrical engineer and and he’s on CNBC once in awhile Mike Wershowsky is the CEO of a Swiss bank. I mean, you’ve got to picture, like I knew these guys when, you know, we’re just trying to figure out how to eat, and so then Mark Lambert was the one of the first 500 employees of small company called Oracle so he doesn’t really have to work anymore and then …I think that I got to keep moving here. So it was great; I loved living off campus and our high-tech at the time was you could have terminals to go into the main computers on campus. We were a cool fraternity; we had a terminal, one terminal in the fraternity, so we could go into MIT campus. We were doing this new thing to different research facilities called email and you know it’s just incredible all this. Now here’s the real funny one. So this is Lewis Scott, a fraternity brother from Mexico, and what he’s talking about here is a protest about tuition— see nobody laughed even. That number is so small. It was protesting tuition back then, saying that in 1985 we’re going to reach $10,000 for tuition, it’s like an incredible, you know, a huge number, right? So protests about tuition increasing have been going on forever if you didn’t know that. So some things change, some things don’t change. This is Maurice Cartman; he was another materials science grad and we ended up taking a lot of Course 3 subjects together and then we also went off to IBM to do co-ops and learn about how large corporations did research and you know that’s a significant thing, because back then, I love this advertisement, this was in the MIT yearbook. There’s like 50 corporations with R&D labs trying to define the future, doing long-term investments to figure out the future. Hughes, which doesn’t exist anymore, I mean Hughes Research Lab does but the the large corporation as it was doesn’t exist anymore. If you look at this advertisement, it’s appealing to you to help them to find the future, to go out there and do fantastic things, right? You probably know that maybe there’s three companies today, large corporations probably do that kind of thing and everybody else is worried about quarterly profits. Theres been a huge change in our in our innovation ecosystem which which I’ll talk about for my own personal journey here. So when I entered graduate school, which i don’t know i wanted to go to, but an MIT professor in Course 3 told me, “hey, you should look at this,” and I said “wow, you know, the stipend is pretty large and I get to learn more and i don’t have to pay for it.” Right? I was a middle-class kid so, you know, if I had to pay a huge amount of money, I would never have gone, so, you know, but the system was the system set up by our friend— anyone know who this is? Nobody knows who this guy is — what? You got it. Vanevar Bush (rhymes with beaver, by the way, easy way to remember how to say that) and he set up the idea that nation-states should invest in research, which was a novel thing in the thirties, and really if you think about why what I was thinking was I want to go to this place, and these are the large corporate labs. You could go there, they would find you to help them discover things about the future and all I wanted to do was get that white coat and do something important. Really I wasn’t thinking about “I want to create some company and flip it in two days” or things like that. I was thinking that I wanted to go there and learn how to do something significant with this knowledge. I could see those chips, I could see the chemistry; I wanted to do something like that and make an impact. So really my goal was to try to learn as much at Cornell and go off and join one of those places but even then it was starting to narrow so in the late 1980s and what was happening was a lot of the corporate labs started disappearing so I’m going to take a brief moment here. I want you to think of something that is different. At the time when I was going to grad school, everybody focused on sort of individual pieces of technology and one of the important things you’re going to see is that the reason that we are able to move forward and influence the world and what we did was that there’s always some force telling us about these other pieces which is the application and how the world works. I think if you look at great innovators, they identify the piece of technology science that’s critical to having a big impact through knowing a wide range of applications that are needed out there and also how the current industry works and how the world is, not that it can’t change, but having all those things in your mind makes you focus on one particular problem versus another one. I’m going to go through how I was pulled into that through the people that were around me. I don’t have time to talk about this but there’s a lot of classes now at MIT to talk about this and to teach this. We have some in our department and and I’m going to end talking about the innovation and entrepreneurship minor. So at Cornell how did I get those other pieces? How did I figure out market applications? I had been in school all this time, well, except for my little entrepreneur things I was doing when I was in high school, but essentially how did I know what the right problem is? I had to choose an advisor and graduate school, and it was really important to me because I still wanted to to have that impact and build something, and so the key thing back then was, even though it was tailing off, there were still these corporate labs that overlapped with university interests and my advisor used to go to Palo Alto in the summers to work at HP Labs which looked in semiconductors at the future and they were very interested in this new material called gallium arsenide and trying to use that to build new kinds of integrated circuits, and so he would come back to campus and he was motivated by problems that he saw out at HP. The second person is Jerry Woodall; he’s a National Technology Medalist in the United States now, but at the time he was an IBM fellow which meant he could go anywhere, do whatever he wanted, and he used to hang out on the Cornell campus and so I ended up running into him and there was a topic related to the two which is how you are going to build new kinds of devices using these lattice mismatch materials so in semiconductors up to this point in time all the useful semiconductors in different areas were always lattice-matched, meaning that if you were to deposit layers of one material on another, you make sure that the lattice constants are the same size so you don’t get any stress or defects or anything like that, but it was pretty clear that those markets were getting very mature and you could build all sorts of new stuff if you could figure out how to combine different semiconductor materials that didn’t have the same lattice constant, the big one being that silicon was already the dominant electronic material for circuits. So if you could add stuff to that in some way, that’d be great but the problem is, of course, the lattice matched so big for a lot of materials you want to put on silicon, they become defective and they’re not useful, so it’s a huge kind of barrier to going further so that problem now is motivated by those wide range of applications. We know how things are made because I’m interacting with with with Jerry from IBM and I start to focus on this problem which is so even though there’s all these all these potential applications it gets down to one thing is how can I put one of those lattice mismatch materials on another and control the defect density. People didn’t really understand the relationship here. I’m showing you this you know now a well-defined slip system and and people were aware of that semiconductors kind of interesting the reason that this actually becomes a problem is that at room temperature, they’re brittle. They don’t deform through introducing defects like that called the dislocation so what happens is when I want to deposit those materials i have to go to a high enough temperature that they become plastic like metals. You guys probably know that if you bend copper, it’s dislocations that are moving in response to that and so finally they get all tangled up and then it breaks. So here semiconductors are like that. When you deposit those films at high temperature, and you have to deposit high temperature because you’re using that substrate to see the crystal growth and so it’s not high enough temperature you don’t get a good crystal, so you gotta figure out well if I put a lattice mismatch material on top of this substrate, here’s a substrate and here’s the two-layer there’s two states you could have. If it’s thin enough, all the items go down and they don’t know that they’re supposed to be at their bulk lattice constant yet and so they could be stretched to fit on that substrate template, but you can imagine eventually if I grow thick enough I’m building up more and more strain energy per area and then eventually I can afford to inject in things like dislocations because when this thing comes in it puts a half plate in the material so it allows the layer to relax. The dynamics of how this comes in become very important. This is the thing that the dislocation sitting at the interface that allows them to relax and then this dislocation that comes up is a consequence that at this location can end in a crystal, so a lot of times you’ll want to keep it very thin, in that case you know not to cross that boundary where dislocations come in but you’re going to see later the big advances I want to bring them and I want it to relax but I don’t want to have many of these things but it’s kind of paradox because well i have to have that because i put the misfit dislocations in the interface then you’re always going to have a certain number coming up and dislocations coming up like this if I build devices up here are very bad. They either hurt the performance of the device or make it unreliable. Now it turns out, getting back to academics here, that it’s kind of interesting that when i was at MIT materials science, the metallurgy was kind of becoming an old topic and electronic materials was the newer topic but everybody in silicon had eliminated dislocations so there’s no need in the current technology to worry about this. There was a class here that end up being very important to me but it was about to be eliminated, which is kind of interesting, going to show you never know what what information you need in the future which why continuous education and online education is so important. At the time, Sam Allen, who’s a professor just retired a little while ago in materials science, he actually taught me the basics of dislocation theory, which ended up being in the semiconductor field where nobody knew it. My key added value to try to figure out this problem. It’s kind of an interesting collection of little threads that led to something important. The key thing in grad school I was able to do is to determine that you can actually manage these defects that come. People thought if you deposited material with too big a lattice mismatch, it ends up being lousy and then you just can’t use it right, so one of the key things this experiment is looking down in it, we realized is a little experience say well what happens if i make smaller substrate with the bigger substrate, and we can see how often nucleation is happening, because if nucleation is only happening in a few areas, and then the dislocations are propagating all long ways to make all these long lines here, then that would be some hope that we could that we could manage them in the future. Either completely relaxed materials without threading coming up or keeping up completely strained. What you see on the right is the same sample as that in some areas that have no nucleation source we could build up strength that’s far above the level that people thought would already bring in dislocation so it starts to separate the sort of thermodynamics from the kinetics in this problem and it starts to give you ideas about how to manage and engineer so science leads to engineering of how you can control all this the stuff. So that was an important area that people recognized at Bell Llabs so off I went to AT&T — i finally achieved my goal, I got the white coat, you know, in the long room of all those scientists there. Back then I still got in a time period where it’s called the model. Bell Labs, at that time, I call it the “venture model” of research: they used to give you, in my case it’s like half a million dollars, and they would tell you like within five years, discover something reasonably important or we’ll put you out to the development areas. That’s kind of how it works. It’s kind of daunting because so many famous people bail out. You’re kind of like “geez, can I really do something important?” and there I am a younger man sitting there and and what I built with characterization was kind of interesting. This is all about materials synthesis, but one of the key problems was seeing those defects in a way that would allow you to count them accurately. It sounds trivial and I spent eight months trying to get this, well, besides building a lab. It turns out, once we had that there are other facilities available at Bell Labs so I could grow materials and stuff like that. It was a well-placed investment and the thing about Bell Labs, that you would you would run into so many people from different areas. This is Yong Zee; he ended up being my co-inventor on strained silicon and he was originally from China but had escaped to the US and he ended up going to UCLA where he learned about the things that we wanted to build in the future, combined with my material science. This is Hans Gosman; he came from Germany, so very international place and he ended up working on germanium tin and tin silicon, things that are again popular today but he was early guy who did all that work and he became an important guy in exelis. Conner Rafferty who ended up doing a start-up later on germanium detector arrays on silicon here in the Boston area so the shift here was now trying to do not the strained layer but to put relaxed layers on top of this substrate and make it completely relaxed. We could shift the lattice constant, build new kinds of devices and what I’m trying to show you here is a lot of times the things you learn in universities, the empirical relationships that you need to get started. So for example, I think with problems that you tend to think like there would be this well-defined problem that appeals to me and then i want to calculate things, we’re gonna figure it out. What really happens here is me thinking “so, OK, I’ve got this problem and I understand the basic relationships in this case. You could say I I know kind of how nucleation of dislocations are going to go and how they’re going to spend on strain and I also know the velocity of this location, how that depends temperature.” So the first idea we had was well I have to have these straining dislocations coming up but remember I told you that this length here is what relieves the the stress of the interface so here I have the case where the amount of misfit dislocations length is the same but in this case over here I’ve nucleated three dislocations and I have many more threading dislocations to deal with where if I use the feature that it’s difficult to nucleate, which I proved in grad school, then you see if you could get this thing to propagate further. You could have much more strain relaxation but have fewer defects coming up to the top. At the time, for anybody in electronic materials (because the most electrical engineers there) are like nobody’s gonna think about this; this is like not within the realm. So it’s again the beginning of realizing interdisciplinary science and engineering. This is all kind of happening concurrently here when I was in school too. What you realize is that how do you create the conditions where you can keep the nucleation rate low and increase the velocity. You can see that we did something was opposite of the time if I keep the stress low, which means I increment the mismatch a little bit at a time, but I go to very high temperatures then you think that you could create dislocations that once one enters it flies across the entire sample relaxing all that area preventing other nucleation near it. So this is actually anti-intuitive because at the time everything’s going to lower and lower temperatures, they’re developing new epitaxy equipment to try to create more and more strain materials; here we are trying to create relaxed materials and now we have this crazy idea and want to go back to older equipment and deposit stuff at really high temperature, and it turns out that — it was an amazing result — you can create these layers and all those black lines or dislocations, you can have massive relaxation, tons and tons of of misfit dislocations being formed but relatively few actual threading dislocations passing through the material to do this. So you can see that because when you stop and go uniform layer at the top there’s no more mismatch, at least in this cross-section TM, you can’t see any black lines threading up through the top layer. This image itself told us that we can build a silicon germanium substrate on top of silicon. You can’t create that any other way, Now that i have a larger lattice constant, we can use the other regime to put silicon back on top the atoms will deposit and they’ll be further apart than its supposed to be in silica which means they’re under tension. If you use epitaxy to create tension like that, you can create a gigapascal of stress, for example, and that’ll change the materials properties of silica. You can’t create a gigapascal of stress by bending a wafer. If you try to do that on a silicon wafer, it will crack. You have to use this process by putting atom by atom the bottom silicon-germanium here is acting like little hands that are all pulling on that silicon creating this tremendous amount of stress. It changed the band structure of it and it turns out electrons and holes go much faster. Of course you know the beauty of this— silicon is is what most transistors are made of. So you start to realize that maybe you could translate this into the application we are thinking about a long time ago. Maybe there’s a way to actually change the other properties of silicon ICs that are built on this. The good days of Bell Labs: it become a very famous thing, because nobody had measured higher mobility in silicon than what was achievable in regular silicon. Exciting in the physics community and then then there’s a practical aspect which could have big commercial impact. People got very excited. By the way, getting on the cover of Bell Labs News was like the internal Nobel Prize or something. It was prize territory so, as a young guy, I thought, “this is fantastic!” There’s my friend Young, now a professor at UCLA, and of course the manager has to be in the photo, part of corporate life. I won’t spend too much time on this because I want to be entertaining and not delve into too much but one of the interesting things was that this is so new and unusual that you could see the simple version of this, which is that strain is going to change the band structure of the conduction band and valence band of silicon, but it turns out that because silicon was such a valuable material and commercial activity there are tons of of scientists that built Boltzmann models of how electrons achieve certain mobilities in the silicon channel. If you actually just put strain in there, it turns out none of those models work. So for the longest time there, they kept telling us well what you’re talking about is impossible, and I would put transistors in front of us say well there’s a transistor. The simple idea to change the band structure seems to be working, there’s something wrong with your scattering theory. What was interesting is because there was never any way to change anything those scattering theories were were held for like 20-30 years in silicate people just assumed they were correct. This is what happens; we have really fundamental innovation, there’s many, many people out there that are are taught with the old rules and you have to tell them, and it’s really amazing to watch, you can put physical things in front of them that violate what they learned in previous textbooks, and they won’t believe you actually. Things like velocity saturation and all these things we’re gonna we’re going to change. Actually to this day, there still are models that the simple model explains it but when you actually go and look at that scattering people still can’t explain it. Even though now virtually all silicone ICs are made with this technology. What’s interesting about Bell Labs is that the reason that you know places like Bells Labs and Xrox were famous for coming with so many of these early innovations is they really had the perfect environment for the innovation process. At Bell Labs, they made everything from transistors to telecom systems so I could go to lunch and there may be people outside of the main labs but they would be working on development in those areas and I could hear about their projects so for me to figure out what what technology and research would probably impact them 10 years from now was was easier right because they were there right there living with us. The other thing is that they made things so we used to go out to Reading and look at how commercial epitaxy was done and things like that, things I needed to deposit thin films, and so you had an understanding of what was possible in real manufacturing so it’s no accident that people kept on discovering important things because they’re not doing it in a vacuum, they were doing it in in sort of unusual environment that that is really hard to stabilize from an economic point of view. What happened here naturally, I figured out that the emperor has no clothes. I was trying to commercialize this within AT&T and nobody knew what to do and and then I started to think why do people have all these research labs but they don’t ever ever do anything with the stuff in here that’s really different. Over time, people, instead of solving that, probably just got rid of labs or finances improve, and that’s where we are today. I realize that with things we’re going to change because if you don’t have a method for commercializing the research, then eventually you’re gonna eliminate it and sure enough, Bell Labs eliminated it. The discovery allowed me to come back to MIT and one of the great things about coming back here was being with so many young people. If you look at the hiring cycle, I was always one of the youngest guys at Bell Labs because there was nobody really hired after me, and so it was kind of depressing in a sense. Getting back to where I could be with young people to look forward and do things with was really rejuvenating. So back to the Institute. It turns out I had a startup package that was worse than the average assistant professor’s because I was well-known now so I should be able to conjure things out of nothing. So fortunately a lab was actually throwing out half of the mocvd machine so I grabbed that and figured out how to how to finish it and we started with one-inch (this is a really old mocvd, that we used to work on one-inch squares). Now in Singapore, we’re working on 200 millimeter wafers in state-of-the-art stuff. Here’s Sri again with another grad student of mine, one of the first ones, Mayank Bulsara. The co-founder of my first company, AmberWaves, and he’s actually here right now— Raise your hand. He’s now chief scientist at SunEdison semiconductor; he’s convinced them to let him still live in Boston even though they’re in St. Louis so Mayank is that kind of guy, important guy. One problem is that we got to get rid of that and put Boston in there. He’s still a Yankees fan, unfortunately. I always liked this picture. This is Mayank again with Andy Kim and Ken Woo. They all look like “wow, this thing is really working, what do we do now?” This is early in my group. We started small. Eventually, Mayank graduated and you could see the economy changing and there really wasn’t anything out there that really made sense for a guy with Mayank’s talent so we said, “hey, we’ve got all this new stuff”— the industry still wasn’t embracing strained silicon or a lot of lattice mismatch stuff —so we said “let’s go do it!” With not enough knowledge about really how hard it’s gonna be, go out there not really knowing. Mayank and I formed a company called AmberWave when he graduated and one of the big challenges was how to commercialize this stuff because The large semiconductor supply chain at that time was huge, people were making lots of circuits but you can’t go to s material supplier and say “make ten wafers for me” so we had to search around to find specialty places that were growing and convince them that they could grow with us if they could produce stuff and work with us. That was really a very interesting experience and we learned how to work with medium-scale producers to create the technology and it had to be much better than we did in the research lab because now it had to go into a silicon fab and we really didn’t know how we would make it smooth enough so it wouldn’t be rejected by the the characterization tools but at the end of the day we did all that. There were three markets we were going after and and again you can’t tell at the beginning: it’s such a basic thing that you can apply to several different areas but one of the early places that in the digital space was willing to work with us was AMD. One of our proudest moments was here: with seven people in the company, we signed with AMD and were able to ship them wafers. Look here it says “AMD or bust” because really at this point if these wafers didn’t work we were going to be finished. This is in Woburn, by the way, (It’s changed now with the Engine just created by MIT) but Woburn is the place of sort of startup slums for MIT. It’s because there’s so much contamination to land there that it can’t be used for anything else so startups go there and rent is cheap. If you look at what a small group of people did, we were able to get this through their flash fab, which is really sensitive to any impurities and things like that and to this day this is really an amazing thing. We built a 25 nanometers physical gate in 2003 because people didn’t believe. Now you have strained silicon, a bit larger devices, and the next thing they says “it’s not going to last, when you make tiny ones, it’s going to go away.” We asked why — and they said “we just think so.” We had to prove it so we did this 25 nanometers thing and of course the thing operates faster and it turns out that this is the last profitable node for most manufacturers in silicon today. We’re doing technology beyond this but the cost per transistor goes up from 25 nanometer transistor as you make them smaller. What’s really cool about this is that back in 2003 we didn’t know it but we actually built sort of the end of road map device in planar form and really quite an accomplishment. During this time, one important thing for creating value is to believe that glimpse in the future. It’s easy to say “we’re going to change this, we’re going to beat intel, we’re going to do all this stuff,” but you actually have to prepare for it and believe you’re going to do it and we created a huge patent portfolio to protect us which is, again you really have to believe when you spend all those resources, very expensive, that you’re really going to do it. We discovered that we were building circuits in strained silicon, and they’re going to find out the different properties for a long time and and the future would be different, that the layouts for the circuits would be different because the p moss and n moss would end up being similar sized transistors. That’s a big deal in circuit design and so we started realizing we can sort of re-patent all of silicon technology. Intel really tried to destroy this patent But Nicole Gerrish is the co-inventor on this so she was a Course 3 student and used to come into my class and sit in the back, she wore a baseball cap because just she rolled out of bed and she looked like she wasn’t listening to me and she aced everything so when we started AmberWave I knew she went to Draper and I went over there and I stole her back. My friends at Draper got mad at me. She joined the company and and ended up being in charge of our intellectual property group. We had projected ultimate power savings of fifty percent which is fairly accurate to what happened. We have to believe in and kind of protect yourself. Intel was first to market and I don’t want to go into the story but they eventually licensed our strained silicon which was a huge deal for the company and then later on TSMC which is a foundry in Taiwan to have parity and technology bought most of the portfolio. There’s the press release. You’d have a hard time finding this on the web, by the way. Intel controlled the press release and really didn’t like this too much. I’m going to go quickly through a few other examples because I want to finish with talking a little bit about what’s happening in materials science. There’s a couple of other startups that we did and I don’t have time to talk about them. When you give talks like this for young people, I always — people have asked this — I always get a feeling that a lot of time people just don’t even venture into trying to do all this stuff because they think about work-life balance, which is really important, but you can do it. Notice I have a lot of people I work with; I’m not doing it all myself; I’ve got my students and employees and then other collaborators. I think the thing to realize is that innovation is a social activity, actually, which we often don’t talk about, and you can actually achieve quite a bit if you recognize that up front. I do have a wife— that’s her. She’s wonderful and we ended up hiking up this volcano mountain in Hawaii recently. I have two children: Danielle and Patrick. Danielle is a chemistry major at Holy Cross and Patrick is a mechanical engineer at Drexel University, and they do know who I am and I like me. So what’s happening today? This was a brief interlude, I woke you guys up. What we’re doing today. So Moore’s Law has stopped, like we predicted. People thought that we were going to extend it by putting III-V materials like gallium arsenide into transistors, but we took a different approach, where we said that silicon is great and III-Vs are great; if we ccombine these materials with all of our knowledge accumulated over this time, and we could actually have a transparent or opaque to the circuit designer we could build new kinds of ICs. It’s just putting more flexibility in the circuit designers’ hands. That’s the big vision that we ended up having. Remember that all during Moore’s Law where the industry knows where it’s going, it’s the application end that determines what you do in materials and stuff like that. When you reach the end of that road, it’s a chicken-and-egg problem because yes, we want to create these platforms to give circuit designers more capability but the markets for these things haven’t been defined yet because those circuits haven’t been created yet so you’re kind of back to the beginning stages of innovation where the markets haven’t developed yet. We’re going to have new materials in this new world and that those platforms that define new circuit innovations are a much harder problem and what you realize is that you have to build a research program where you have materials scientists working on materials process and devices and circuit design all affecting each other simultaneously which is impossible to do the United States. There’s no way you can create such a program. However, MIT having some unique things going on, we have something called SMART which is Singapore-MIT Alliance Research and Technology and so we ended up winning one-fifth of that program to do this. Any Singaporeans in the audience? yes! CREATE campus next to NUS and a thousand square meters to go do our thing. The great thing about working out of the country is that you learn about culture and food and so I’m going to make you hungry now, because Singapore’s all about food and we have laksa — bet you guys miss that. I just destroyed your day and mee siam. Besides technology there’s lots of excellent things to have in Singapore. The bottom line is we were able to build infrastructure. If you see these pieces called SMART, those are pieces that we could build that now allow us to bring this kind of big vision together. To give you an idea of how this thing was set up from the beginning: a regular silicon process goes the following path. You have a wafer that goes through what’s called the front-end in a foundry and then it goes to the back end and then your chips are finished. So what we’re doing here is we’re going to work with real foundries in the research process and we have to make it look like we’re a customer. This is a real big innovation both of the research stage but also to to affect commercialization at the other end of the process. We ended up building these two pieces and because we have a separate line coming in here and we join them in SMART, we then process the unusual parts of the wafer and send it back to the foundry so that it looks it never left. So the manufacturing facility isn’t doing anything different than what it does today. This is huge because first it allows us to build real circuits in a loop with materials people. The second thing is it shows you what the new business model is for the circuit industry, which is that you can essentially today build an Intel that produces novel circus with their own process but they don’t have to invest in infrastructure, which means that the capital return for such an enterprise is going to return to the early days of silicon, which is a really exciting proposition. This is our test chip so when I was in grad school, we used to dream about having some sort of silicon design environment where we could put these kinds of exotic materials like III-Vs together with CMOS and here’s the real thing that is going to come out this year. Now that this is more resolved and we can see how it’s going to be commercialized, we created the first foreign company that I’ve been involved with. I never founded a company in a foreign country before, but you know with MIT and all these international relationships, we’re starting to get experience with that and so we formed a company called New Silicon, again a little hubris there, New Silicon Private Limited. We’re going to follow that model I just talked about in parallel to finishing our research in Singapore. So let me finish with talking about kind of big picture stuff and education. We’ve been using these words like materials science and engineering and then innovation and you know Vanevar Bush again here wrote a famous thing called The Endless Frontier, which is that there’s always new ideas, we’re always going to keep on advancing, you should always invest in technology and research. Those elements for innovation that are important besides just studying the technology piece alone: What’s interesting is really coming full circle because Vanevar Bush, with the rise of things in the 1930s and 1940s, found it necessary to define what an engineer is. I think sometimes you come to school you think that things have always been a certain way, defined a certain way so it’s very interesting to go back and look. So this is Vanevar Bush in 1942: Today we would call this an innovator or an entrepreneur, but remember that’s the standard definition he was trying to create for an engineer. In my career, we went from an age of superspecialization at the end of a particular wave and then when I started going through Cornell, interdisciplinary science was the big thing, interdisciplinary fields in the technical realm. What we call innovation today is just folding in other elements that an engineer used to practice because we’re headed back to a time when we need to innovate and it’s skill we haven’t used in a long time. Things are a lot less prescribed and so what is valuable today is to have not only interdisciplinary science and engineering but to also be interacting with that social process: how do you know what the real applications are? how do you know what people are doing today? If you ignore those things, you can’t restart real innovation in any field. It’s a big shift occurring. Some tools that you can use: we have been online from the very beginning with edX class talking about the innovation process, just how these elements work together and how you could do risky things and Andreas Wankeral is over here so he helps me out with all of our innovation work and this expanded into MIT last year, the innovation and entrepreneurship minor. Are you guys familiar with that? It’s a new minor that talks about large-scale problem solving and how do you solve this problem in a real way. So if you’re interested you should look at this new minor at MIT. I just wanted to summarize that these classes and in Course 3, venture engineering is a core class in the innovation entrepreneurship minor, and if you take that minor we have a class you can take, 3.086. We spend more time with you going through how you do this innovation problem, hoow you can avoid going down dark alleys earlier in the process, and how you can increase your chance for having impact. I’ll stop there and answer any questions that you might have. Geoff Beach: Well, thank you, Gene, for an outstanding talk. Do we have questions in the audience? If you look at my story, a place that I was able to innovate and learn about that was Bell Labs. So these corporate labs were around and it made academia’s life easier because we could even do forward look at these here which is preparing the student how to think in those areas and they could go innovate in those areas. What’s happened over time is a gap between corporate time horizon and then universities were asked to do the entire thing. Clearly a university can’t because it has other goals; it’s got educational goals and things like that. So it’s a great question, what’s going to happen? Well, I can tell you that if nothing happens, you’ll continue to see productivity decrease and you’ll see lots of public investment over here and it will be very difficult to have people have time to figure out how to produce real innovation on the other side. So what do I think will happen? If you look at startups, the ones that really produce things that added to productivity in the economy, they’re in some way tailing a lot of that early corporate work. The other startup set is successfully finishing this sort of information age paradigm. How do we find the new thing? How do we get a new Moore’s Law to kind of change things? It’s very difficult for a start-up to do because of just the resources involved— you’re going to have capital costs to set up, capital costs have to go up, they’ve gone up a lot recently, historically 5 to 7% short-term rates are normal. In that world you have to have a lot of profit to invest in the future, it’s the only way you can do that. So i think the answer is going to be probably that you’ll have, in this consolidation phase we’re going through, a few companies in each sector that are going to have dominant pricing and then as capital costs go up, they’re gonna probably look out into the future because they’re going to the resources to do so and they’ll spend money on R&D to preserve their future and that’s the only thing I could see because in this sort of longer-term chaotic thing, startups can’t last long enough to cover that time horizon. So in a weird way, trying to figure out how to get corporations to invest in the future again for their own preservation, but more importantly, there’s all sorts of spillover effects that are really important for innovation. So in a weird way, you want companies to look forward to spend towards the future because it helps create competition ultimately. But I think nobody really knows, that’s the key issue looking forward. Geoff Beach: You have any more questions? If you look at those elements that I was talking about for innovation, they become corporations that don’t mind investing in markets that they aren’t in today, they don’t mind going into new technology, and they don’t mind changing business models. You have to look at companies that invest in internal projects with that in mind. There’s very few of them, but the number one was Google. Google will invest internally in ways that they’ll go into different markets, go in to different technology, and they’ll change that thing if they’re successful can change the business model fromwhatever business model theyre in. And they’re one of the few companies that practices that but there’s very few. Geoff Beach: We have one more quick question. Question: so a lot of your work focused on information technology and you’ve seen really rapid progression and movement pver the last 20 years, a lot of advancements do you think there’ll be something different that is really rapidly changing? So one thing that’s clear that, A guy who ended his career at MIT, Thomas Kuhn wrote this book called The Structure of Scientific Revolutions and so he’s the guy who came up with the original term “paradigm shift” Go read his book, he’s a philosopher physicist. It reads like a physicist philosopher so you have to be patient with it but one thing that is a fundamental of looking at the shifts that occur is that they will always occur so it’s always these points for people think that nothing will ever happen again, but it’s guaranteed something will happen, and the other thing is that you can’t predict ever when it will happen or what area it will be in. But I think you can see things in information systems there has to be some sort of new thing that happens. I’m trying to play my own little role in it with our collaborators and SMART in that way but probably something has to happen there and obviously people have been waiting for something to happen in the health space if you want to call it that. And I think you’ll probably see more and more devices in medical technology that really probably at some point will take off like an exponential. Just from my own needs, because I don’t mind drinking cocktails, is that you know I think buying organs in the future is going to be a big thing, so if you can create livers, create a heart, whatever, and pop it into me, we’re good to go. People will buy multiple, fueled by multiple things, you can see some sort of exponential thing — it affects people’s lifetime. You get a feedback loop. Once we can start growing organs, there will be another development. There are some things you can see that have the properties of a paradigm but it’s really hard to tell. Geoff Beach: Well, with that I think it’s a cocktail hour so let’s thank Professor Fitzgerald one more time. Thank all of you for coming and you’re all invited to attend the reception right outside right now.

One Comment

  1. Eight December said:

    34:20

    October 15, 2019
    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *