Interview: Vincent Donnelly
2011 Thornton Award Recipient
Interviewed by Paul Holloway, November 2, 2011
HOLLOWAY: My name is Paul Holloway. I’m a member of the AVS History Committee. Today is Wednesday, November 2nd, 2011. We’re at the 58th International Symposium of the AVS in Nashville, Tennessee, and today I have the privilege and pleasure of interviewing Professor Vincent M. Donnelly from the University of Houston. Vince is the Thornton Award winner for 2011, and his citation reads, “For innovation of surface and plasma diagnostics to evaluate the complex kinetics of plasma processing, and for the development of fundamental reaction mechanisms to explain that complexity.” So Vince, congratulations on the Thornton Award. Well deserved, and welcome to the interview.
DONNELLY: Well thank you.
HOLLOWAY: Let’s begin, please, by giving us your date of birth and place of birth.
DONNELLY: I was born March 13, 1950 in Philadelphia, Pennsylvania.
HOLLOWAY: Okay. And could you tell us a little bit about your educational background, starting as early as you want?
DONNELLY: [Laughs] Well, maybe I’ll start with undergraduate school. I went to LaSalle University, which was then LaSalle College, in Philadelphia. I started there in 1968. I got a Bachelor of Arts degree, in chemistry. And then I went to the University of Pittsburgh for my Ph.D. I started there in 1972 and got a degree in Physical Chemistry, and then went to the Naval Research Lab as a National Research Council postdoc fellow, and spent almost two years there.
HOLLOWAY: Could you tell us a little bit about who you worked with at Pitt, and mentors that helped you through that process?
DONNELLY: My advisor at the University of Pittsburgh was Professor Frederick Kaufman, whose background was gas phase kinetics. He was very well known in that field. He measured reaction rates and identified fundamental mechanisms for gas phase reactions, mainly ones that were of importance for stratospheric chemistry. His interests were in understanding the complex chemistry of the stratosphere, and in particular at that time, how the introduction of pollutants into the stratosphere would affect the ozone layer. There was a lot of concern in the possibility of flying a supersonic transports in the stratosphere, so reactions of NOx with oxygen atoms, for example OH reactions and so on, were ones that were of key interest. And then following that very closely was the chlorofluorocarbon issue. It was discovered that these chemical used in spray cans and refrigerators were so inert that they would diffuse to the top of the stratosphere and would be photodissociated and make chlorine atoms. So suddenly reactions of chlorine atoms became important. Also important was the reaction of O and NO to make NO2, sometimes in an excited electronic state. That became part of my thesis project, understanding the dynamics of the excited state of NO2.
HOLLOWAY: So did you predict the occurrence of the ozone hole before it…?
DONNELLY: No. I mean our group was very much involved in that. We also had several students that went on to work in that area. Carl Howard went to Boulder and did a lot of very well-known work; and Jim Anderson, who was a postdoc then, and had some of, his own funding by the time he left eventually went to Harvard in the Chemistry Department, and was the chair there for some time. So we was involved in all that, making measurements in the stratosphere, but the actual ozone hole was something that almost discovered by accident some years later. But they were all very much at the center of that issue. My project was a little bit off to the side, but I was, of course, very well aware of what everybody else was doing.
HOLLOWAY: So what was your project?
DONNELLY: Well, we were interested in understanding why nitrogen dioxide had such a long radiative lifetime. Simple theory would predict that it should have been much shorter, and this had some relevance to stratospheric chemistry, but it was more of a basic interest in understanding why that would be the case. It turns out that triatomic molecules were sort of an interesting intermediate case in quantum mechanics between very small where things are simple, and very big where things are so complicated that they again become kind of simple. So we studied some of the details of how the excited electronic states of NO2 decay, and how they’re deactivated by colliding with other gases. We got some insights into what was going on. That was my project.
HOLLOWAY: You say long for a lifetime. What are you talking about, milliseconds?
DONNELLY: No, these were hundreds of microseconds, which was still long. You could calculate a lifetime from the absorption coefficient, and if you do that, you end up calculating a lifetime of a microsecond, roughly. And yet the lifetime was hundreds of times longer than that. It has to do with a breakdown of the Born-Oppenheimer approximation. The electronic and vibrational wave functions are strongly mixed, so the levels of the ground state interact very strongly with the levels of the excited state, and the excited state takes on most of the ground state character, which basically dilutes the oscillator strength and causes the lifetime to be much longer. It’s been a long time since I’ve thought about that kind of stuff. [Laughs] I think what I said is more or less correct. So that was my Ph.D. project.
HOLLOWAY: So where did you go after your Ph.D.? You finished your Ph.D. in ’78, I believe?
DONNELLY: ’77. Yeah. Almost five years. I think it was close to four years and nine months or something like that. Then I went to NRL, Naval Research Lab, in Washington D.C.
HOLLOWAY: What did you do there?
DONNELLY: I worked in a group in the chemistry branch. Jim McDonald was my advisor, and I started to look at laser photodissociation of small molecules, polyatomic molecules. That was just about the time when excimer lasers were commercially developed. As a matter of fact, NRL was one of the first places to have commercial excimer lasers; the technology kind of came out of NRL. So we had one of the first lasers that kind of worked. It was a little bit of a bear, but we could get it to work. The pulse energies were so large, the number of photons in a very short burst of light were so large that you would get all of these multiple photon effects, so you could rip apart molecules and make small radicals from it. So we studied how that happened, simultaneous versus sequential photo excitation and that kind of stuff. And then we used this as a source to generate radicals and then study the reactions of NH2 and NH and C2O and other unusual molecules that we could get access to,
HOLLOWAY: These were all gas phase?
DONNELLY: Yeah, this was all gas phase.
HOLLOWAY: So how long did you stay at NRL?
DONNELLY: Almost two years. About a year and ten months.
HOLLOWAY: Were there other people there that you collaborated with, besides the group leader?
DONNELLY: Yeah. Well McDonald, I think he was a co-author in all the publications. We had another staff member who was on most of the papers that I wrote, and then we had another postdoc. Andy Baronoski was the staff member, and we had a couple other postdocs that were there at the same time as me.
HOLLOWAY: So then after NRL you went to?
DONNELLY: I went to Bell Labs.
HOLLOWAY: Bell Labs. Murray Hill?
DONNELLY: Yeah, Murray Hill.
HOLLOWAY: And had a good time there?
DONNELLY: Yeah. So that’s interesting. When I went there, I was thrown into a completely different field, at least I thought it was completely different. And like I said in my talk, I kind of recognized something from my past. And there’s this strong tendency for new graduate students and postdocs to want to continue what they were doing: they know a lot about it and they think it’s interesting. So I went from this gas phase kinetics and spectroscopy and photochemistry to semiconductor lasers and integrated circuits and how they operated, and solid-state physics, which I knew nothing about, and surface chemistry, which I knew nothing about. So this new field of plasmas, I knew nothing about plasmas either, was sort of thrown at me, and I was happy to find fairly quickly that there was something that I was familiar with. It kind of gave me a toehold on that field;, that I could then do something fairly quickly, but at the same time, learn about it more.
HOLLOWAY: So you’ve started off in plasma chemistry of etching?
DONNELLY: Etching of silicon, yeah. Like I said in the talk, I met Dan Flamm, who was a collaborator and colleague and still a friend that I communicate with from time to time. He was in the field a few years before me at Bell Labs, and he had set up a lab and had a couple apparatuses. So he told me about this interesting glow that was present above silicon when it was being etched in the presence of fluorine atoms, and that kind of then got me interested. And the system that he had put together was, to my surprise, very much like the discharge flow systems that the other students in my group at the University of Pittsburgh used. All the kinetics that Kaufman did were by a technique called the discharge flow technique, where you make radicals in a small microwave discharge, and then you send them down a tube. And they take some time to flow down a tube, so distance is converted into time. If you move an injector with a reactant into that tube and change the distance over which they can have a chance to react before you detect the product, then you can basically measure a rate constant. And so the system that Dan Flamm had put together looked a lot like that—it was a tube with a discharge at one end and the silicon at the other. So that was a good place to kind of jump in. I was also building up my own laboratory at the time, and wanted to get into using lasers in plasmas, because I had done quite a bit of that, but not in plasmas, and I saw a good opportunity to learn more details than you can get from optical emission, by using lasers to do laser-induced fluorescence. So still gas phase, but in the plasma now, so that’s kind of where I went next.
HOLLOWAY: The lasers you were using were all visible lasers, or excimer lasers?
DONNELLY: Well these lasers were excimer laser pump dye lasers. They were tunable, and they were visible, or you could double them into the ultraviolet so you could cover a fairly wide range and detect a number of species.
HOLLOWAY: Were they fast lasers?
DONNELLY: They weren’t super fast. The excimer laser pulse was like fifteen nanoseconds, and the dye laser was a little bit shorter. But not picosecond. At the time there were picosecond lasers, but for what we wanted to do we didn’t really need that kind of speed; we just needed something that was fast enough to be able to discriminate against the light from the plasma. So that was kind of the main concern.
HOLLOWAY: Now how do you distinguish between fluorescence and optical emission?
DONNELLY: Well that’s the thing. If you use a short pulse-length laser, then you can gate the detection electronics to look only for a short time during and after the laser pulse, and so you could knock down the background from the plasma by quite a bit. If all of the signal is coming in bursts of a hundred nanoseconds, you know, a hundred times a second or ten times a second, then that gives you a huge advantage in discriminating against the plasma emission.
HOLLOWAY: Now yesterday in your very nice presentation, you showed papers from Coburn and Winters. [Right.] There must have been a pretty healthy competition between IBM and Bell Labs. I remember that competition very well in surface science area.
DONNELLY: Yes. Right, right.
HOLLOWAY: Dry plasma processing, I’m sure it was pretty intense at times.
DONNELLY: I always felt that, at least for me, it never got personal. I liked the people that we were competing with. But I enjoyed the competition—it made it more interesting and exciting to maybe be looking at things a little differently than they were. They had a particular approach to things, and we had an approach, and they were sort of complimentary in that our approach was to try to go more directly into the plasma, which admittedly has its complications, and sometimes the results aren’t quite as easy to as interpret or as definitive. Their approach was to simplify things by using a beam apparatus, which gives you maybe more definitive information, but it’s a little bit removed from the plasma environment. So those were kind of the two approaches that we had, and the way that we pose them when we were kind of in dispute over something. I think in the end, we mostly looked at things the same way.
HOLLOWAY: Right. I didn’t mean to imply that it was personal or attacking one another.
DONNELLY: No, no, no. Well sometimes these kinds of things can get to that, but it never really got to that point. We were always good friends.
HOLLOWAY: Yeah. Well, competition amongst approaches and ideas for explanation are what makes us revise our thoughts and move forward.
DONNELLY: Right, right. And as a matter of fact, within Bell Labs there was a lot of competition as well. I don’t know if IBM was that way so much, but I think they were. I can remember some similar incidents at IBM. But we had a lot of competition between different people doing plasma etching, as to what was right. Not only that, but in other fields too. Some of the more basic physics areas at Bell Labs, there was a lot of competition between people for how they interpret things and crystal growth methods; MBE versus MOCVD and that kind of thing. There was a lot of competition to try to make their method the one that was the best.
HOLLOWAY: Who were some of the other people at Bell Labs that were doing etching?
DONNELLY: Besides Dan Flamm, Rick Gottscho was another, of course, very famous person. And then Dale Ibbotson was there for quite a while. Dale was from Caltech, and he left Bell Labs some years ago. He went into the manufacturing side of AT&T for a while, and then went to another company. I kind of lost track of where he is now. Dan Flamm is now a lawyer. He switched fields. And Rick Gottscho is a Vice President of Etching at Lam. So there were also other people who were maybe there for less time. Gary Taylor was doing stuff for a while. Chuck Jurgenson. Tom Mayer, who was pretty big in AVS for a while, went to NC State, then went to Sandia, I think. And there were some others who were there, maybe not as long, but were active. Mike Vasile did a lot of early stuff on silicon etching. Bob Fruend, who made a lot of cross section measurements related to plasmas. We used to have a lot of people in development, too. Of course the other side was a little bit closer to manufacturing. There are some really good early studies by Joe Mogab and Hi Levestein back in the ’70s for the first time, discovering how to etch things, and finding some of the effects that people rediscovered over and over again. We used to have plasma internal meetings that would go on for several days, and there was a pretty good range of stuff, from very much manufacturing-related issues to very basic stuff that would be presented at these internal meetings.
HOLLOWAY: Your citation says that you evaluated the complex kinetics of plasma processing. Give us some examples of how complex that might become. Is that a reasonable question?
DONNELLY: Well, it’s quite complex at first because you’ve got electrons and negative ions in some plasmas, and then of course positive ions. And the electrons are active in ionizing the gas and associating the gas. Usually you have more than one feed gas, so it’s not just a single gas like chlorine, but it might be chlorine with hydrogen bromide and oxygen and helium added to the plasma. And a lot of this chemistry, these recipes that we end up studying, are ones that the industry finds kind of by trial and error to be better or optimal for a particular application. So with all those reactants, just the plasma chemistry itself, the plasma physics and chemistry can be quite complicated, knowing all the different reaction rates for ionization and dissociation of the gas by collisions with electrons. And then the plasma dynamics themselves are complex, how the electric fields form and evolve in space and time. Then couple that with the interactions of the plasma with the surface that you’re processing, which is generating products that come off the surface and enter the plasma and get fragmented. And then the species that then stick to the walls of the plasma chamber and undergo reactions there. So the combination of all of that is very complex.
HOLLOWAY: So that is almost impossible to solve.
DONNELLY: Well, we don’t try to solve everything. What we try to do is make measurements of what we think might be key things, and then try to back out some conclusions from that. So for example, we’ll look at the rate at which chlorine might refill the surface, if you suddenly remove all the chlorine with a laser pulse, which is one thing that we did. So we could use a laser to very quickly clean all the chlorine off the silicon while it’s etching in the plasma, and then we could go back at some variable time later and do that again, and we could measure how much chlorine came off each time. So we could do it fast enough that we could keep the surface more or less clean of chlorine. Or we could make the time longer, where the chlorine mostly filled back in between laser pulses. And from that, we could then measure how fast chlorination occurs in the plasma and get some information on how Cl and Cl2 are responsible for that, and what the probability is that they would come down and stick and stay on the surface.
HOLLOWAY: If you think about a plasma and injection of a gas into a plasma, you’re going to activate a molecular species by cracking them into product or down to the atomic level, give them kinetic energy, let them react, and then knock them off the surface. In your opinion, which of those processes dominates for the majority of the reactions? Again, is that a fair question?
DONNELLY: I think what dominates the initial fragmentation of the molecules (I’m not sure if I’m going to answer your question exactly), but mostly it’s energetic electrons colliding with the molecule. They have enough energy to break the bond. There are other processes where an electron will attach to a molecule and make it fall apart to form a negative ion, and those can be lower energy process, but they’re relatively rare, and even for electronegative gases they tend to be less important for most conditions than the electron impact association. So once that happens once it can happen again, depending of course on how dense the plasma is. If the electron density is very high, then you can, as you say, take a molecule all the way down to its atoms. Typically, we don’t run into those conditions. For example, C2F6 would be all carbon atoms and fluorine atoms. We would have a mix of fluorine atoms and CF3 and CF2 and maybe some CF—probably not much carbon. So it doesn’t typically go completely to atoms. Then depending on the pressure, they can react in the gas phase. So in a lot of plasma processes, the pressures these days are fairly low. Unless the reaction rate in the gas phase is pretty fast, then they don’t tend to react so much with themselves in the gas phase. But what they do is stick to surfaces and then react there. So that’s probably more important. Then ions can bombard those surfaces. Ions typically hitting the gas phase molecules don’t do much chemistry either. . The ions really don’t have much energy until they get very close to the surface, and then they pick up a lot of energy. And so it’s then quite important that the energetic ions, when they hit the molecules on the surface, will cause them to come off the surface and perhaps fragment, or oftentimes fragment when they come off. So then they go back into the plasma and the whole process starts over again. So it’s a very complex recycling kind of mechanism.
HOLLOWAY: In microelectronics, we’ve been following Moore’s Law for a number of years. How critical was the development of plasma processing and dry etching to continue that trend?
DONNELLY: It was very critical. It went hand in hand with photolithography. Photolithography is clearly the first step, so you can’t pattern transfer unless you have the pattern. Photolithography kept getting pushed to smaller and smaller dimensions, more and more tricks, shorter wavelengths, other optical tricks. But at each step along the line, that pattern had to be transferred by plasma etching. And as device dimensions continue to shrink, then suddenly something that was not a problem at all becomes a big problem. So over the course of the years, problems changed. Early on it was very hard to etch silicon without getting what was called black silicon. So you had small contaminants that would make these little micro masks, and then you would etch these tall needles. That problem was solved by getting better reactor materials that didn’t contaminate the surfaces as much. And then other examples are something called notching, which was not a problem when dimensions were on the order of, say, half a micron or a micron, but when they became a couple tenths of a micron, then suddenly these little notches that were cut into the side of an etched feature became a problem. And so there was a lot of effort, then, to try to solve that. Damage was another problem. Plasma induced damage at certain thicknesses of the gate oxide became a real problem with tunneling of electrons through the gate oxide, damaging the transistor. So figuring out ways of operation the plasma to minimize that kind of electrical damage became a big problem. Today I think the problems are in losing more than one monolayer of material, and once you’re finished etching through a layer, because the layers are getting so thin. Probably that’s going to be the next big problem, figuring out how to get the selectivity and the control to be so good that you don’t do any damage to underlying layers.
HOLLOWAY: I remember when dry etching was first used for LED operations. The damage that was induced in the surface region by the plasma processing seriously degraded the efficiency of the device. Those problems, by and large, have now been solved?
DONNELLY: I believe so. I’m not as familiar with the compound semiconductor area, these days. I haven’t followed so much, what’s the current state of the art there. Mostly we’ve focused on silicon, where the dimensions are much smaller and it’s a much bigger area. Most of our industrial funding also is in that area. So we tend to work where they would like us to work.
HOLLOWAY: That’s important. What energies are you typically talking about in silicon, and what range ions have in silicon?
DONNELLY: In the early days, a kilovolt ion energy was kind of what people were using. And there are still some processes today that run in energies almost that high. For silicon dioxide etching, the energies tend to be pretty high. For silicon etching, the energies have now gotten down to being on the order of 50 to 100 volts, and probably are moving toward still lower energies. There are threshold energies that you cannot go below, but in fact, if you work close to the threshold and don’t care so much about rates, then there are big advantages to working near threshold of one material versus another material with a higher threshold, where you can then get very good control.
HOLLOWAY: Now in your bio it was mentioned that you made the first measurement of the reaction probability for fluorine atoms in etching of silicon and silicon dioxide. How important was that?
DONNELLY: I guess that’s a hard question to answer!
HOLLOWAY: [Laughs] They’re very important because you did it, right? [Laughter]
DONNELLY: At the time it was not known, as you said, so we wanted to get that number. I think those values have been used quite a few times in modeling studies to try and understand how silicon etches in a fluorine plasma. That kind of fundamental information is hard to come by; they are difficult measurements to do. There weren’t, even in those days, that many people doing those kinds of measurements. So I don’t know who all has used those numbers in their models, but I think they’re probably still pretty well received and used in various models for predicting profile evolution and so on. There’s a lot of work that’s in the public domain that people have published. There are quite a few good modelers in the field who publish a lot. There are also people within industry that do this kind of modeling that is held within the company to guide them. Both the semiconductor tool manufacturing companies, the companies that make plasma etching equipment, and companies that etch integrated circuits have had, from time to time, many of them have still today active groups that do this kind of modeling for their own competitive edge.
HOLLOWAY: It’s amazing how frequently some fundamental cross-section like that or reaction rate, everybody assumes that it’s measured, but when you go looking for it, it’s hard to find a value.
DONNELLY: Right, right.
HOLLOWAY: And that was the situation in this case?
DONNELLY: Yeah. I think this area of plasma processing is pretty starved for basic information on a lot of these systems. There’s a fair amount known about some of the really common and widely important gases like hydrogen or nitrogen or oxygen. There’s a lot known from atmospheric chemistry, for example, on these species, or astrophysics. But things like chlorine or HBr or bromine or carbon tetrafluoride, these are ones that aren’t so widely used or studied in other areas, so a lot of times the data just aren’t there.
HOLLOWAY: You stayed at Bell Labs for twenty years or so?
DONNELLY: 21 years, 21.
HOLLOWAY: And then moved to the University of Houston? [Yes.] Tell us why you made that move.
DONNELLY: [Laughs] Well, I was actually briefly at a spinoff of Lucent. So the history of Bell Labs, at least from when I was there, is it was initially part of the regulated Bell telephone system, and then it was deregulated and then it was put under the control of AT&T, I believe, in 1984. I might be off one year on that. And then AT&T operated Bell Labs for a while, and then AT&T spun off Lucent Technologies, and then Lucent took Bell Labs under their wing, and then that lasted until 2000. And then Lucent really spun off part of their company into a company called Agere Systems, and then Bell Labs basically got split between Agere Systems and Lucent, with most of the semiconductor related research going to Agere Systems. So the silicon research, a good bit of what was remaining of surface science, and compound semiconductors all went to Agere Systems. And that is about the time of the crash of the dot-com business, so very quickly Lucent and Agere in particular realized that they couldn’t sustain these research efforts. So Agere drastically cut the research part of their company, and that included most of the people that had been transferred over to their research lab from Bell Labs. That included me, so that’s why I went to the University of Houston.
HOLLOWAY: That’s a good reason.
DONNELLY: That’s a good reason, and I certainly don’t regret it. That was a great move. I had from time to time thought about going to academia, and almost did it a few years before that, but I always had it too good at Bell Labs where I had postdocs and students, and it was a pretty good life. Right up until the last year, pretty much, and then it very quickly became obvious that I needed to move on. So Houston’s been good. I have a good collaborator. One of the things I liked about Bell Labs over the years is that I always had a lot of good people to collaborate with, and so I have kind of the same feeling at Houston with Demetre Economou being a close colleague, so we work jointly on a lot of stuff.
HOLLOWAY: Are there other people at Houston that you collaborate with?
DONNELLY: Some. There are a few other professors that we’ve done some joint things with, but mostly it’s been Demetre and I.
HOLLOWAY: Is he an experimentalist or a theorist?
DONNELLY: I would say he’s probably more, over his career, more a modeler than an experimentalist, but he’s done both. So we kind of complement each other pretty well in that I don’t do much modeling. The modeling I do is more…I’ll do some chemistry modeling or stuff that’s heavily involved in surface reactions. He’s does the full 2D or 3D modeling with a lot of transport folded into it. But he also does experiments, but I’ve got more experimental experience. Bell Labs was a good place to develop your experimental skills because you had to do a lot of the stuff yourself, so I was pretty hands-on my whole career there.
HOLLOWAY: In terms of Bell’s system, it seems strange to me that it was split up at the time you were mentioning. It seems like it’s coming back together now. But the research is not coming back.
DONNELLY: No, the research is not coming back; that’s gone. To be honest, they’ve broken up so many times and reformed and so on that I can’t quite figure it out anymore. But I know that AT&T is back as a company, but if I start talking about this anymore I’m going to show my ignorance. [Laughs] That’s so long in the past that I just haven’t kept up with it.
HOLLOWAY: It wasn’t critical to your career path.
DONNELLY: No.
HOLLOWAY: Tell me what you enjoy about being at Houston: the students, the classroom teaching, the research?
DONNELLY: Yeah, I enjoy all of that. Surprisingly, I enjoy writing proposals.
HOLLOWAY: You’re the first person that I’ve ever had. [Laughs]
DONNELLY: It’s a good chance to just sit down and think about something new, and just sort of think it through. Like could you do this, how would you do it, and what issues come out? So it’s like doing a thought experiment. I like to go through this process, and in the end come up with a scenario to get to where we want to go, thinking of a lot of the problems along the way. Of course, not everything. That’s relatively easy to do. The hard part is to then actually do it, because then you find all the real problems. But I enjoy that, that process. I enjoy teaching, too. It’s a lot of work, especially the first time you teach a course, but you do learn a lot. I learned a lot that I didn’t know before about all kinds of things from having to teach it. And I enjoy the students, both the undergraduates mostly that I’ve taught and the graduate students that I’ve taught and supervised. There’s a range of personalities and so on, and you kind of have to figure out how to interact with each one in a different way.
HOLLOWAY: Each has their strengths and their weaknesses, just like we do.
DONNELLY: Just like we do. So I’m sure they talk about me that way, too. [Laughs] But anyway, they’ve all gone on to do quite well, and it’s always rewarding. A little bit frustrating, too, when you see them finally graduate. It’s rewarding, of course, because now you’ve trained this person and they’re going to hopefully have a good career. But the frustrating part is that now they’re really efficient and they’re trained and they can really do research, but they’re going somewhere else to do it.
HOLLOWAY: Yeah, and then you start over again.
DONNELLY: Then you start over again. Right. But that’s the name of the game. That’s the job. And so I certainly enjoy it.
HOLLOWAY: Are there particular students that stand out in your mind that you’d like to mention? Or all of them?
DONNELLY I had better not do that, because if I don’t mention every single one, then they’ll wonder. So all of the ones who have graduated have done good work. I mean sometimes the nature of the project means that somebody will get more publications. We’ve had some students build apparatuses from scratch and maybe not get so many publications, but they had to really work hard to do that. And then we’ve had others that maybe followed another student, and they were able to publish more, or worked with a postdoc and were able to publish more. And you know, that’s good, too. But in the end I think they all got the kind of training that they need to go out and work in this field. Most of them have gone into the semiconductor field, but not all. A few of them really fell in love with Houston or had a spouse who had a job in Houston, and they then went to either petroleum or chemical industry there. And even those students that had no training in that ended up doing pretty well.
HOLLOWAY: It’s funny how the students do get accustomed to the local living and tend to adapt to it.
DONNELLY: These were both Chinese students, and there are a lot of Chinese students in Houston and there’s a Chinatown. And I think they like the climate, too. Not too many people like the climate of Houston. [Laughs] It’s maybe more like home to them, coming from humid cities.
HOLLOWAY: Now I understand that you’ve invented a process called nanopantography. [Right.] Tell me about that.
DONNELLY: Well, I was actually trying to come up with an idea for a proposal when I was still at Bell Labs. Well Agere Systems, I guess, but the last year at what was still called Bell Labs I was in the process of packing up equipment that they were donating to me, and looking for a job, so I was hired as a nonpaid consultant so I could use their facilities. And so I would sit at work and try to come up with ideas for proposals. One of the things I thought about was can you put nanoparticles in selected places on a surface and not just have them nucleate where they want to nucleate, but actually drop them in a specific location. And so I thought of the game where you have little steel balls that you roll around and have to drop them in a hole and you have to tilt the thing in certain ways, I forget what that’s called. But there’s now an electronic version of that; I have the app on my phone. But it used to be a plastic thing and you dropped the ball in the hole. So I thought, could you charge up a little sphere with one charge and then drop it in a hole and then have it kind of be repelled and eventually settle in the center of the hole, kind of like that game? And then I thought, well that probably wouldn’t work, but why couldn’t you just do that by putting a charge on a particle or maybe just even an atom or an ion, and then having it encounter a hole with some charge on it that would cause the particle to go to the center of the hole and hit the center. And so I had some friends at the labs who had some simulation package called SIMION that I loaded on the computer that I still had there, and I started running some simulations of how you could focus ions to hit the center of a hole. And in fact you could do it. I mean I kind of knew you could, but I wasn’t sure about scaling and things like that, and turns out it scales just fine to small dimensions. So then I realized you could take an ion beam, and I was concerned about the energy spread, so I varied energies and so on, and then realized that you probably could do this. You could take an ion beam, and if it was energetic enough, you could have the ions encounter a hole in a conducting material with an insulator below, and then put a voltage on it and have it focus at the center. And then the thought was can you tilt that substrate to move that focus around? And if you could do that, then you could write patterns in a massively parallel way. My hope was to charge up nickel atoms and then drop them at the center of that hole and then build a little nickel nanoparticle that you could then nucleate carbon nanotubes off of. So that was kind of the initial idea. But I didn’t actually propose that when I went on interviews. It was still a little premature. But when I went to Houston, it turns out that just at that time there was this call for proposals from NSF from a program that was called NIRT, Nano Interactive Research Teams I think it was, or something like that. They were calling for proposals, and certainly teaming up with people was encouraged, so I went to Demetre and said, “Hey, what do you think of this?” And he liked it, and we got somebody else who did ion optics, Paul Ruchhoeft from electrical engineering, and he liked it. So we put together this team, and we wrote a million-dollar proposal, and it got funded. [Laughs] My first proposal, and it got funded, and it was like wow.
HOLLOWAY: This is too easy!
DONNELLY: Well I didn’t want to say that, because I had a lot since that weren’t funded, but I kind of was lucky with that one, and so that was my first proposal. We had a couple of students and a postdoc work on that. And the student who was on it was one of these students that had nothing to start with. “Here’s the idea. Build it and do it,” and he managed to get some results before he graduated, which given the complexity of it, was pretty good.
HOLLOWAY: That’s pretty impressive.
DONNELLY: So we’re still doing that. We have another grant now, and we’re back to doing it. We had several turned down, but we finally got one funded, so we’re trying to push those dimensions now. We did ten nanometers—that was pretty good for the first time for a new technique, ten nanometer resolution, and we’d like to push it to two or three. We think we can with that we’re going to do now.
HOLLOWAY: So this would be an additive technology versus a subtractive.
DONNELLY: It’s both. Mostly what we did was subtractive. So we focused an argon ion beam in a presence of chlorine and etched silicon with it. So we could etch little holes in silicon that were ten nanometers in diameter and a hundred nanometers deep. Which has potential applications for very small holes in membranes for DNA sequencing, for example—something a lot of people were trying to do.
HOLLOWAY: Now you also developed the spinning wall technology. What is that?
DONNELLY: Well, it’s a way of looking at a plasma reactor wall, which is basically a boundary condition that you’d like to have well established. So reactants go to the reactor wall many times in their lifetime within a plasma reactor. If a chlorine molecule comes into a plasma reactor, it gets dissociated in the chlorine atoms. With chlorine atoms, the most likely thing that they’re going to do is strike the walls of the chamber and form a product, maybe Cl2 or maybe some other silicon chloride product, and then they’ll come into the plasma and they’ll get dissociated. They might do that a hundred times before they would etch the silicon or get pumped out of the chamber. So what happens on the wall is very important. And what we were doing at Bell Labs was using a technique by which we would direct a laser at the surface and desorb what was on the surface, and then it would go into the plasma, and we could see either laser-induced fluorescence or emission from the plasma and get sound information on certain species. But it wasn’t all-inclusive, and it couldn’t do the kind of quantitative work you can do with a mass spectrometer or with XPS or with Auger electron spectroscopy. So the challenge, then, was to try to couple mass spectrometry or electron spectroscopy with a plasma to characterize a surface. And of course, it seems impossible. How could you do Auger electron spectroscopy in a plasma that has 1011 electrons per cubic centimeter in electric fields of many kilovolts per square centimeter, and magnetic fields and so on? So the way people have done, kind of, this work is to put a sample into the plasma, expose it, then turn the plasma off and move the sample out under vacuum into an ultra-high vacuum chamber, and then analyze it. That takes a few minutes to get the sample from the plasma reactor to the ultra-high vacuum system. And the thought is, well, things could happen in between and you can’t detect any desorbing species with that, so it’s very limited in what you could learn from that. So the thought would be, then, how do you get the sample from the plasma to analysis really quickly? The idea actually came from this laser desorption work. We were trying to put a mass spectrometer into the plasma, get it very close to the sample so we could hit it with a laser, and then detect product before it had a chance to interact with the plasma. And we tried and we weren’t successful at doing that. And then the postdoc, Keith Guinn, who was working on this project at Bell Labs, had an idea of taking a platter, like a hard disk drive, putting a sample on it, and then spinning it so that the sample would be under a plasma and then under the laser to desorb species. And so kind of taking the one step toward getting the sample out of the plasma. And I thought that was a great idea, but it had a lot of problems with the huge amount of gas load that you would have to get rid of, and it didn’t seem like there was a good way to do that. He didn’t pursue it any further. He was hired at Bell Labs, and then went on to do something else. But the idea stuck in my head. So I was in Houston, and then I kind of thought of, well, how do you do this? And I thought that maybe you could have a flat sample that you could flip back and forth, like flip it first one way and then the other to expose one side to the plasma and one side to analysis. Problem with that is when you flip it, you let a lot of gas through, so it can’t be a flat sample then; it has to be cylindrical. So that was kind of the process. And so I did some simple calculations to figure out how much pressure drop I could get, and then I put together this proposal. The review came back and said, “I don’t think it would work, it seems like a crazy idea. But why don’t you let him try?” [Laughs] I wasn’t asking for a whole lot of money. I had some startup money to look at it, too. So I hired a postdoc, and he and I put it together. And lo and behold, we got huge signals when we started doing this, looking at chlorine.
HOLLOWAY: You love reviewers that come back with an answer like that sometimes.
DONNELLY: Right! [Laughs]
HOLLOWAY: It’s realistic.
DONNELLY: [Laughs] Yeah. So it’s been really interesting to look at some of these things with that system. We have kind of a unique capability, and we learned quite a bit about the way reactions happen on walls. We’re right now seeing some really complex things, that it’s good to know they’re complex, I guess we kind of suspected they would be, but we didn’t think they’d be that complex. So if we’re etching silicon in a reactor and there’s a little bit of oxygen present, which there usually is from the reactor materials, then that ends up coming off the wall is a whole slew of stuff from radicals to big molecules with several silicons and oxygens and chlorines on them. So much more complex than what anybody thought was going on.
HOLLOWAY: I interviewed earlier Mohan Sankaran, and he was talking about a microplasma You’re talking about macro plasmas.
DONNELLY: Right, right.
HOLLOWAY: Is there a connection between the two? Are the fundamental principles the same, or are there strong differences?
DONNELLY: There’s a scaling wall that’s basically pressure time distance. So if the size is very small, then you can have a glow discharge with a very high pressure. If the size is very large, then you’d only have a discharge with a very low pressure. So that pressure times distance is a constant—depending on the dimensions of your units, it’s a different number. So for a microplasma of a few hundred microns in dimension, the pressure range that you need to operate at in order to get a stable glow is roughly atmospheric pressure. So you can make a stable atmospheric glow discharge that’s not like a welding arc, not real hot, not what’s called a thermal plasma. You can do that with a very small gap between the anode and cathode, for example, and at atmospheric pressure. So the kind of stuff that we’ve traditionally done, and many people in our field, the applications has been etching, where you want to etch big silicon wafers, so the chambers have to be on the order of half a meter or so across. They’re 300 millimeter wafers, but they’re chamber has to be a little bigger than that. So that means that the pressure has to be in the order of tens to hundreds of millitorr in order for that to work. If the application is for using a plasma needle to do surgery, which is one of the things that’s up and coming now, then you want something very small, and then of course you also don’t want to put a patient in vacuum, so you need to work at atmospheric pressure.
HOLLOWAY: [Laughs] We used to do that with canaries. [Laughter]
DONNELLY: I guess you could go down a little bit, but not very much. So fortunately, you want both small and high pressure, so that’s where he’s working. And it’s not just for that application. There are many other things that you can do with that kind of a small plasma, including just very simple sensors that are very compact and nanoparticle synthesis, which is what he’s doing. Under these conditions you can synthesize particles in a way that you can’t do at low pressures and larger dimensions.
HOLLOWAY: I’m sure that there’s going to be room for both approaches, but how do you see if you were to take microplasmas and fine tips and arrays, you could directly write patterns of materials, for example, and avoid the deposition of a thin film, the lithography, and the patterning? In principle at least you could.
DONNELLY: Well the problem is when you say micro, you really mean hundreds of micro. So these microplasmas are typically several hundred microns inside, and of course they diffuse when you get away from the plasma source. So you could write several hundred micron-sized features with them, perhaps, but the dimensions are way too big for integrated circuits. So that’s not something you could do.
HOLLOWAY: So you don’t see a competition there in the future?
DONNELLY: No, no. There’s a little bit of overlap for some applications where you might want to strip a polymer off of a wafer, photoresist after it’s been patterned, where you could do that either at low pressure in a more traditional reactor or at high pressure with arrays of plasmas that are running across the surfaces and burning the stuff off at a pretty high rate. So there could be some entry into that market there. But for the most part, they’re really orthogonal in their applications.
HOLLOWAY: Earlier on, you intimated at least that there is a large amount of information that has been derived and understanding developed. What’s in the future for plasma processing and microelectronics? Is there a lot to be done yet?
DONNELLY: I think so. And getting back to the expression, as long as the device dimensions continue to shrink. I think when finally the industry reaches the end, wherever this is, then there might be a few more years, and then you would see the problems maybe not be there anymore to work on, at least in plasma processing for microelectronics. But it’s hard to predict where the end of this is. If graphene patterning is something that needs to be done, would we still just be using a plasma to do that? I don’t know. Carbon nanotubes to be cut; do you need a plasma to do that? I don’t know. But I would say at least for the moment, it’s still going strong. There’s a lot of clever device design as well, turning things on their sides and stacking things in different ways to try to increase the density of circuitry. And new materials issues that come about with that—having to etch things in very unusual configurations. So there’s always going to be, I think, well not always, but for at least for the foreseeable future, even if the device dimensions kind of reach a limit, there will be a lot of that kind of innovation that will keep giving rise to a lot of interesting materials and materials processing problems that I think would keep people busy for a while. And plasmas themselves, of course, this is just one application. There’s a whole slew of things they’re being used for, including the high-pressure microplasmas for the things that we’ve discussed. In all of that, I think that it’s a pretty broad portfolio that people have to look at. So I think if the problems start to become fewer in plasma etching, then there’s other areas that are pretty comfortable, I think, to slip into, that would still keep people busy for a while.
HOLLOWAY: So plasmas are still bright?
DONNELLY: I think so. [Laughs] I think so. We have a DOE center now that’s focused on advanced control of energy distributions, which is something that’s sort of a broad swath that goes through a lot of different areas. Like how do you make the electrons and the ions have the energies that you want them to have? For high-pressure plasmas, for low-pressure plasmas, for lighting, for synthesis of solar cell materials, for conformal thin films for all kinds of applications, there’s always a need for that kind of control. And understanding the basic physics behind that and how to do that is also something that’s worthy of study and is being studied now by a number of people.
HOLLOWAY: Let me put you on the spot again and ask you, you’re in a unique situation where you spent 21 years at an industrial fundamental basic research lab, and now you’ve spent ten years at a university. What’s the comparison for a young person who might be thinking about which one of those two he/she might try to pursue?
DONNELLY: Well, I think that it’s hard to answer that question because that question is really a function of time. So when I came out in 1979 and went to Bell Labs, it was a very good choice, especially if you weren’t sure if you wanted to go into academia or you wanted to get industrial experience. Because that was the kind of place, and there were several other labs like that, where you could go and do pretty fundamental stuff, but get a good exposure to what the important relevant problems were in technology. And then you could go either way. You could continue to do basic stuff and not really do much that relates directly to technology, or you could go more in the direction of technology, or even some hybrid staying in between. And then if you decided at some point that you wanted to go to a university, there was a pretty good opportunity to do that after spending some time at a place like Bell Labs or IBM or Xerox or RCA Labs. There were a lot of choices in those days. But today, at least in this field, I can’t comment so much on maybe the medical field or pharmaceuticals, there might still be some models there. But in this field, the industrial labs, the nature of the work is much more tied to the product line and shorter term problems, so it’s not so easy to go to a company and publish a lot and do a lot of basic work that you could then go to a university later on. So I think the person has to make their mind up quicker. And usually, since there are many more industrial jobs than university jobs, usually people go that way. But if the person really wants to do the academic career, then they need to give that a shot. You can still, of course, spend time as a professor, decide you don’t like it and you want to go into industry, then that’s an easier transition to make, I think. It might be a little difficult to adjust at first, but I think there’s more opportunity to do that.
HOLLOWAY: In your opinion, how do the national labs compare to the old Bell Labs for fundamental research?
DONNELLY: Well, I think that’s certainly a good kind of-- Today I guess that’s the analog, that you could go into a national lab and then go to academia later. I mean clearly, a lot of national labs are doing very good work. What Bell Labs had that those labs didn’t have was the focus. So the business was focused on the telephone system, and how you make that better, more reliable. So even the very fundamental stuff that came out of Bell Labs, underneath it there was some justification that could be made based on the company’s long-term needs.
HOLLOWAY: I see that you have a number of publications and patents. Do those patents come from Bell Labs or from the university?
DONNELLY: I have mostly from Bell Labs, but I have I guess two or three patents now from Houston. The nanopantography actually just came out two weeks ago; that patent just issued. And then we have another one that’s joint with Tokyo Electron, perhaps two—I’m a little bit fuzzy now. [Laughs]
HOLLOWAY: It’s hard to keep up with those. Those things go on for so long.
DONNELLY: Yeah. The nanopantography one was I think four or five years back and forth before we finally issued. And I didn’t even know it issued. I just got this letter in the mail saying, “Do you want a plaque with your patent on it?”
HOLLOWAY: [Laughs] They come first, don’t they? [Laughter] Before any other official notification comes.
DONNELLY: Yeah, yeah.
HOLLOWAY: We covered a lot of topics. Are there any topics you’d like to add, Vince?
DONNELLY: No. I guess just to reiterate what I said at the end of my talk. I think it’s such an honor to receive this award, and especially having read a little bit recently about John Thornton, to hear that he had such a strong appreciation for collaborative work and how working with people and having a society that can bring people together fosters a lot of ideas. And so I’ve been pretty fortunate in my career to be at places where I’ve had good people through both colleagues and students and postdocs that I was training and then became colleagues, to interact with, really. They made whatever I’ve accomplished possible.
HOLLOWAY: Well we don’t accomplish anything alone, to speak of. We build upon a number of people that support us, and it’s important to acknowledge that.
DONNELLY: Right. That’s what we do.
HOLLOWAY: I can speak personally about John Thornton. When I first was a new faculty member at Florida, he gave me some solar black coupons. It was part of a study that I was doing at that point in time, and so he was collaborative and interactive on a personal basis, and I always appreciated that very much.
DONNELLY: Yeah. Those are good interactions. Well thank you for doing this. It’s a very complete interview.
HOLLOWAY: Well thanks very much for the interview, Vince.