Interview: Dr. Rudolf Tromp
2004 Medard W. Welch Award Recipient
November 2004
HOLLOWAY: Hello, I'm Paul Holloway. I'm a member of the AVS History Committee. As part of the Society's Historical Archives, today I'm going to talk with Dr. Rudolf Tromp from IBM T. J. Watson Research Center. He is the 2004 recipient of the Medard W. Welch Award. His citation is, "For fundamental discoveries in epitaxial growth and elucidation of their applications to technological problems." Today is November 17, 2004, and we are at the 51st International Symposium of the AVS in Anaheim, California. So Ruud, it's a real pleasure to have you here today and thanks for agreeing to this conversation.
TROMP: Happy to be here.
HOLLOWAY: I understand you got your start in Europe.
TROMP: Yes I did. I was born and raised in Holland, and did my pre-graduate and graduate education there. I went to the University of Twente, which at that time was a technical university, and got a degree in physics engineering there.
HOLLOWAY: Ah, so you're a down to Earth guy.
TROMP: I'm a down to Earth guy, yes. You know, as part of their curriculum, there was a lot of quantum mechanics and stuff like that, as in regular universities, but there was also a lot of work on equipment design, building vacuum apparatus, and I think as a second- or third-year student you built vacuum equipment and built things like multilayer dielectric mirrors and things like that to learn how to do that. And in general, a broad background in engineering, which has been quite important in my work ever since. Then I guess in 1977 or '78, thereabouts, I went to the FOM Institute for Atomic and Molecular Physics in Amsterdam to pursue a PhD in physics, with Frans Saris as my thesis advisor. I spent altogether five or six years there working on and developing medium energy ion scattering (MEIS), a few hundred kilovolt type scattering.
HOLLOWAY: Did you apply that mainly to semiconductors?
TROMP: Mainly to semiconductors, yes. We were a small group of students working together on this big piece of apparatus, and so some people worked on, Rob Smeenk in particular, and Friso van der Veen , whom you may know, worked on metal surfaces. At that time nobody really used it on semiconductors, so my assignment was to see if we could make sense of clean silicon.
HOLLOWAY: Clean reconstructed silicon?
TROMP: Clean reconstructed silicon, yes, silicon (111). And so I spent many years learning to do that.
HOLLOWAY: So how does the medium ion scattering fit into the picture of LEED (low energy electron diffraction), STM (scanning tunneling microscopy) and high resolution TEM and getting a solution to that 7x7 reconstruction?
TROMP: Well, I think it was one of, as you say, different techniques, and it has the advantage of being quantitative, relying on Rutherford scattering, which is old and we understand very well. So if you do an ion scattering experiment, you can actually run a Monte Carlo simulation of the process and get very quantitative comparisons between what you measure and what you think it means. And so I spent years doing experiments as well as building Monte Carlo code, together with Joost Frenken who is well know in the STM field right now at the University of Leiden, writing tens of thousands of lines of Monte Carlo code to do quantitative simulations, and then build models as well as you could and run these computer simulations. I think it was, as you said, one factor in a number of experiments. Of course Takayanagi's experiments, he is getting the Gaede-Langmuir Award today for his work on silicon among other things. And it was scanning tunneling microscopy (STM), which of course in that same timeframe was important, as well as LEED work by Ian McRae at Bell Labs. These were powerful methods, and the time was right, I guess. It was really at a point where a lot of these techniques that had been developed over several decades came to a level of maturity where you could get quantitative information.
HOLLOWAY: Good. So then you moved to the United States.
TROMP: Right, I moved to New York in 1983 and joined the IBM Watson Research Lab, which at that point had a very large contingent of scientists doing surface science. There must have been 50 or 60 people there.
HOLLOWAY: Yes, they were a real force in the field.
TROMP: Yes. And there was lots of money around! [Laughs]
HOLLOWAY: You mean there's not still? [Laughs]
TROMP: Lots of people, lots of money. My first assignment there was to set up medium energy ion scattering, which I'm proud to say is still operational every day of the week even today-still being used, a lot of it in Matt Copel's work, so a bit more technologically applied. So I did that, and got involved with Joe Demuth and Bob Hamers and Mark Welland, and to some degree with Randy Feenstra in scanning tunneling microscopy in its early days.
HOLLOWAY: Was there a lot of interaction and visitation back and forth with Europe and Binnig and Rohrer during that time?
TROMP: Well Joe had visited for sure, and was closely in touch. I was just the inexperienced young guy down the block [laughs], and so I was sort of initially working a bit from a distance, but got gradually more and more involved with the tunneling
microscopy. And certainly Gert (Binnig) and Heine (Rohrer) visited us in Yorktown and tried to help and see what we were doing. I remember that when Heine visited, we were struggling to get this microscope to behave properly and we had an oscilloscope with a tunneling current on it that you could see, and Heine walked into the lab and said hello, and he saw the oscilloscope from across the room, and he has sort of a loud and booming voice, so as he said hello, he saw the tunneling current record his voice, and he said, "Boy, you guys still have got some problems here." [Laughter] That was funny. But we got atomic resolution at some point late in the evening. Of course we were competing with the guys at Bell Labs, Gene Golovchenko and his group. But we got past that. We got 7x7 images and all that.
HOLLOWAY: So you applied that mainly to semiconductor surfaces?
TROMP: That was my interest. That's what I knew really well, and also where we felt at the time that it could have a real impact. Si (111)-(7x7) had sort of been understood, but I was interested in understanding where the electronic states come from that you see in photoemission. You know, I'd worked on ion scattering, and I concluded that the dimer structure was the structure of the silicon (001) surface, but it hadn't been looked at with the STM at all, so there was a real chance there to do something new. And so after seeing 7x7 in the STM, we almost immediately switched to (001) because nobody had done it yet. And so we got atomic resolution of that, and we saw dimers, which was really gratifying. Of course we found out how many defects there are in that surface, which I think early on was one of the real surprises of STM, you know, how bad many of those
surfaces are, how many defects there are that we really have no clue of, and how hard it is to really prepare these surfaces well. So I worked with Bob Hamers and Joe Demuth on these silicon surfaces. Evert van Loenen joined us and we started looking at metal films
on surfaces.
HOLLOWAY: Then you added germanium back into that mix?
TROMP: I was not involved in that actually, but that happened soon after. I worked with Evert van Loenen on the silver square-root three structure on silicon (111). You know, Shirley Chiang and her group were working on the same thing at the Almaden Research Lab. It was a proud moment when we published two side-by-side PRLs (Physical Review Letters) on the square-root three silicon structure with diametrically opposed interpretations of what was basically the same data. [Laughter]
HOLLOWAY: That always makes for excitement!
TROMP: Yes, that generated a lot of interest and amusement. It became clear then that in order to understand tunneling microscopy images you really have to do the electronic structure calculations. Because you aren't looking at the atoms, you're looking at the electronic states, and so it becomes a much harder game then. So that's pretty much when I left tunneling microscopy. I was in it for only two years or something, and went back to my ion scattering machine, which I had been building up in the meantime, since there were a lot of good experiments to do there. And I became more interested in understanding how crystal growth works. You know, the STM was pretty much a static technique, and ion scattering up to that point had been looking at basically stable structures. You know, you look at a surface that doesn't change. And I got more interested in crystal growth and wanted to do work there. And ion scattering was a good method, so I went back to that and started working on silicon germanium epitaxy, in that context.
HOLLOWAY: Now, that progressed to the point where it was actually a fabricated device.
TROMP: Yes. So at that time, that fit very well in IBM. Bernie Meyerson was very active, and was working on the development of his CVD method, UHV CVD (ultrahigh vacuum chemical vapor deposition) strain relaxed silicon germanium alloy films with Francoise LeGoues, who joined IBM at about the same time as I did. She did a lot of transmission electron microscopy, trying to understand strain relaxation phenomena. Together with Matt Copel and Mark Reuter, and Tim Kaxiras who was a post-doc at IBM at the time, we worked on the role of surfactants, how we can tame this wild silicon germanium growth and use growth modifiers with group three or five elements as surfactants to really control the growth modes.
HOLLOWAY: So when do you use group three versus group five, then?
TROMP: You can use either.
HOLLOWAY: So either is appropriate for silicon?
TROMP: Yes, for silicon and silicon germanium. And now of course surfactants are used in metallic systems and a large variety of different metals and semiconductors, and super lattices and quantum metals and what not. That sort of evolved into a field of its own over time, but again, one I haven't worked on in many years now.
HOLLOWAY: So then you moved on to wires and quantum dots and that sort of thing?
TROMP: I became more interested in microscopy, actually. At some point, you know, I'd worked on Si (111) (7x7) for a long time, and on and off I still do. But I went to a conference, and Ernst Bauer, who won the Welch Award maybe ten years ago or so, he showed a movie that was made by Wolfgang Telieps, who was one of his graduate students, using low energy electron microscopy (LEEM) to look at the Si (111) (7x7) to (1x1) phase transition. And I was interested in crystal growth. I was doing this ion scattering work, but you don't get a real-time live view of growth in that experiment. But what Ernst Bauer's movie showed is that you can look at surfaces in real time and make movies of surface dynamic phenomenon. So when I saw that, I basically went to see Joe Demuth, who was my manager at the time, and I told him, "I'm ready to drop what I'm doing and go to
Germany and learn how to build a LEEM (low energy electron microscope), because that looks far more exciting than anything I'm doing now." So he said, "Well, that's fine. Go to Germany." So I went to visit Ernst there, I went to Berlin where Bradshaw and his group was constructing a LEEM, and visited for a couple of days, you know, peeked around in their kitchen. And went back to New York and taught myself electron optics, and started building a LEEM. So that took about two years, and I got it together. That was really the most fantastic thing I've ever done. That was even more exciting than STM to me at that time,
because when you turn that thing on-Actually there's a photo I have at home where Peter Bennett from ASU was visiting, and we turned this microscope on, and that's the marvel of engineering really. You turn the microscope for the first time, and within ten minutes we had an image on our screen.
HOLLOWAY: That's exciting! You must have broken out the champagne.
TROMP: Yes, we went home and got drunk. [Laughter]
HOLLOWAY: Celebration time.
TROMP: Yes. That was really amazing. And that project has been sort of at this same level of excitement ever since. It's fantastic.
HOLLOWAY: So how many LEEMs are there?
TROMP: There are probably about 20 or so worldwide.
HOLLOWAY: So it's a very powerful technique. Still limited distribution, though.
TROMP: Yes, limited. It's considered to be a hard thing. You know, it's not cheap. You build your own and it's a bit of a challenge, or you buy one in which case you need a fair bit of money to do so. But seeing is believing. You make these movies, and what you learn is that we all come out of school with a fair bit of experience and you think you know how things ought to work. Like STM, real experiments teach you a fair bit of humility, because more often than not you get it wrong.
HOLLOWAY: You mean that the surface is not flat and perfectly clean?
TROMP: Yeah. And things certainly work in different ways than the conceptions that you take into a project. And that's the exciting part. You actually do things, and you find out that all your good theories go out the window when you get confronted with reality. But that's fun and that's exciting.
HOLLOWAY: An experiment that doesn't prove something different to you and gives you a different direction is not a very exciting experiment.
TROMP: It's not a very exciting experiment, yes. So that has been really interesting, that, and some of the UHV-TEM stuff that I've gotten involved with over the years. That has been really fantastic.
HOLLOWAY: So what are you applying this LEEM technique to today, then?
TROMP: Well, with the LEEM we've worked on crystal growth, silicon germanium surfaces, as well as phase transitions. I told you we still do a little bit of work on 7x7 once in a while, and that has been a very rich system to work on. Over the last couple of years I've worked primarily on growth of organic thin films, pentacene. I gave a talk yesterday on what we've done in pentacene growth over the last few years, and that again has been really interesting, because you know, surface science has mostly dealt with inorganic surfaces, silicon, metals for catalysis and surface phase transition. There have been a few people, Antoine Kahn one of them, who have done a lot of work over the years on organic surfaces. Pentacene growth has been interesting because it's really allowed us to understand in more detail what happens when you grow organic films on inorganic surfaces, and what controls the molecular interactions there, how you can manipulate crystal growth, and what's the basic physics behind the formation of these interfaces. And that's been a lot of fun.
HOLLOWAY: Sort of interesting that people a long time ago wouldn't have agreed to organics going into their vacuum systems.
TROMP: Yeah, it's one of those things. You know, I've always felt that if you go through all the trouble of building a piece of equipment, you might as well do something interesting with it, even if you run the risk of breaking the tool! You know, I haven't broken it yet.
HOLLOWAY: A lot of interesting things to learn out of that field.
TROMP: Yes. Being afraid of the consequences doesn't get you anywhere, so I've never been afraid of just putting crazy stuff in my vacuum system, and sometimes you pay, right? You've got to take it apart and clean it and stuff like that.
HOLLOWAY: Well, at the university we have graduate students to do that. [Laughs]
TROMP: Yes. So that's been...you know, and the organics are a very interesting field, and I think it connects with the whole area of molecular electronics where worry about how does a molecule interact with a contact exactly, what controls that interaction, how much control do you have over the molecular configuration on the junction, which controls all the transport properties. So if you're interested in making either memory type devices or FET type devices out of these molecular junctions, you have to control these things very well, which means you have to understand what's going on there. And so that is an area of current research, of course. This pentacene work has sort of-in my mind anyway; we'll see how other people think-but has been a model system to try and understand some of the basic issues there of a simple molecule. And I've collaborated with a whole bunch of people in IBM. I got involved with that because Christos Dimitrakopoulos, who is making pentacene FETs, gave me a sample and said, "Why don't you take a look at this. It might be interesting." And it was very interesting, and sparked a whole new research program.
HOLLOWAY: As I understand it, now you're leader of a group of people working in a whole large number of areas.
TROMP: Yes. So in industry, and of course I'm in an industrial environment-I've worked for IBM for 21 plus years now, and so it's not university, right. So the research program that we've had has been connected pretty closely with areas of material science that are of interest to the company for our semiconductor technology manufacturing as well as in display technology and magnetic storage. And that's continuing, so as we see that the evolution of silicon technology starts to hit some pretty hard walls, the question is what comes after that? If Moore's Law comes to an end, where do we take computing infrastructure from there? Is there anything beyond silicon? And of course there are a whole bunch of things that have potential: nanowires, nanotubes, spintronics potentially, molecular electronics. There's a whole catalogue of potential materials, material systems, technologies that may either replace silicon or complement it in some way. And so this group that I am heading now is really focusing on how we may use some of these materials, how we may use nanotubes and how we may use nanowires, how we can use chemistry to fabricate things on length scales that are smaller than we can do with brute force lithography methods, and learn how to take some of these materials and not just put together one FET and see how good or bad it is, but put together more complex structures, put it in an architectural context and learn how to build stuff on one, two, three nanometer length scales and have it be functional. So that's I think where our research will go for the foreseeable future, to really try and apply some of these things that we are learning to do in the general field of nanotechnology and apply it in this field of information processing. That's a big program.
HOLLOWAY: Yes, that's a big program and a big challenge right there.
TROMP: Big challenge. But it's exciting, because there's a lot of good science to be done there, a lot of basic science, as there was with silicon. You know, working at IBM, I found that, and people say that there's no good science going on in industry because it's all applied, and I always find that is so much nonsense, because the application actually focuses your mind onto the science problems that are really important to develop the use of technology. And there is lots of good science. The AVS is an example of that, of how being focused on a technology area really spawns a lot of good basic science. That sounds contradictory, but I think that's how it actually works.
HOLLOWAY: I think they go hand in hand.
TROMP: They go hand in hand, yes. You cannot develop technology if you don't understand the underlying science.
HOLLOWAY: Right. You can't maintain it, you can't keep it growing.
TROMP: No. You can't skip doing the science, either. You've got to do the science in order to develop the technology, and I think that's going to continue. And what's exciting now is that our periodic system has been sort limited to silicon and what serves silicon, and now there's actually a tremendous broadening of that. So our task is not getting easier; it's getting harder. But it's an exciting time, I think.
HOLLOWAY: Well Ruud, it has really been a pleasure having you here today. Are there any other points that you'd like to cover?
TROMP: No, I don't think so. Again, I'd like of course to thank all the people I have worked with and I have the fortune of working with, both inside IBM and my friends in academia that I've worked with over many years. I've been fortunate to work at such a great company, and get all the support for so many years.
HOLLOWAY: Good. Well thanks again for being here today.
TROMP: Thank you. [Recording resumes]
TROMP: ...that's a great and important thing to do. So I think that's what we'll see first, and then we'll see hybrid technology in which we will compliment pieces of silicon technology with whatever nanotechnology works and is useful, and does certain things better than silicon does. This revolutionary vision where we'll throw some of the technology away, we'll have something new. Silicon technology took 50 years to develop, and any significant new technology is going to take decades to develop, not years.
HOLLOWAY: There's so much infrastructure and knowledge base there for silicon.
TROMP: It's not just that. Yeah we know all this stuff, and we'd be dumb not to use it. But to develop new technology from the ground up is just tremendously hard. Silicon technology is great. But if you would have asked the guys at Bell Labs if they thought it possible that 40 years from their invention people would put a billion transistors on a chip, they would send you to the loony bin.
HOLLOWAY: It's like Watson predicting that a low number of computers were going to satisfy the needs of the world. [Chuckles] It's difficult to see where some technologies will go.