AVS Historical Persons | Robert Waits - 2007

Robert Waits - 2007

Oral History Interview with Robert Waits

Interviewed by Paul Holloway
October 18, 2007

HOLLOWAY: Good morning. My name is Paul Holloway. I am a member of the AVS History Committee. Today is Thursday, October 18, 2007. We are at the 54th International Symposium of the AVS in Seattle, Washington. It's my pleasure today to interview Robert Waits, a long-time member of the AVS community who has participated in a number of activities. Bob, thank you very much for agreeing to be interviewed today.

WAITS: Thank you.

HOLLOWAY: To get us started, perhaps you could give us an indication of your educational background.

WAITS: I'm a graduate of Texas Christian University in Fort Worth, Texas. 

HOLLOWAY: Was that in EE or physics?

waits.jpgWAITS: I majored in chemistry and physics. They didn't have a B.S. degree, so I majored in chemistry and physics, and then noticed that I could probably get a degree in math too because I was taking so much math. I ended up with a Bachelor of Arts degree and a triple major in 1955. Then a friend at TCU in geology had been up to Stanford. I said, "Oh, that looks pretty nice." I applied to Stanford in graduate school in physical chemistry, and I got a master's degree at Stanford.

HOLLOWAY: What year was that roughly?

WAITS: The master's degree, I took as much time as I could, and in 1958 I finally got the degree. I think my best education was during the summers; the summer right after graduating from TCU I worked at Argonne Laboratory, and three summers after that I worked at Los Alamos National Laboratory in New Mexico. I think I learned more at Argonne and Los Alamos than in my graduate work, believe it or not.

HOLLOWAY: Oh, I believe that. Everybody thinks that schools and universities are the only places that you can get an education, but that's far from true. I tell my students, "Be prepared to learn for the rest of your life. We are just trying to get you to the point where you understand how to learn."

WAITS: That's absolutely correct. At Argonne National Lab, I was privileged to work with Robert Thorne. He was doing vapor pressure measurements of refractory oxides, thorium oxide and uranium oxide. My task that summer was to calibrate an optical pyrometer (which I had never heard of before) from first principles; the melting point of copper was one of the standards. I wrote a great report which was classified for no good reason. It was like a graduate school because Dr. Thorne had a student who was getting a Ph.D. from the University of Kansas. That was my first introduction to vacuum, all glass systems, mercury pump, McLeod gauge. Robert Thorne was also the head of the glass shop at Argonne. He was a glass blower and constructed his own vacuum system. Everybody back then had to know how to blow glass.

HOLLOWAY: Right. What year was this roughly?

WAITS: This would have been the summer of 1955. That was before any kind of electronic pumps. I think we had a liquid nitrogen trap, but I'm not absolutely sure. It could have been a dry ice trap of some sort to trap the mercury.

HOLLOWAY: Right. So that was your first introduction to vacuum?

WAITS: Yes, other than filter pump. In chemistry, you use a water aspirator to filter things. 

HOLLOWAY: At Los Alamos, did you work with vacuum there too?

WAITS: Yeah, well actually during 1957, my second summer at Los Alamos, I worked with Dr. Ralph Perkins. That task that summer was to measure the effect of hydrogen on tantalum, which everybody knows now, if you put hydrogen in tantalum it falls apart. We were measuring hardness versus gas exposure, and temperature, again, in a glass vacuum system, the tantalum samples were heated using an RF induction heater.

HOLLOWAY: These were sealed glass systems with mercury diffusion pumps?

WAITS: I don't really remember. I think they were. I do have that report, and I just glanced at it recently. It wasn't classified. It didn't really - I'm trying to remember what kind of pump it was. It was probably mercury pump, but I am not sure.

[Note: The work was reported in Los Alamos Report LA-2136 (unclassified), "Tantalum Annealing and Degassing and Hardness Effects of Dissolved Gases" (Sept. 1957). A two-stage oil diffusion pump was used.]

HOLLOWAY: Yeah, it probably was. So you spent three summers at Los Alamos National Lab?

WAITS: Yes. The first summer I was doing analytical chemistry, which is really boring. Many students will go back to the same group summer after summer, but I actually specifically asked for a different group each summer, which is pretty smart because I learned more things.

HOLLOWAY: Got a broader view of what's going on. So you went back the second summer. Where did you work?

WAITS: The first summer was analytical chemistry. The second summer was the hydrogen work. The last summer was working with a group that was trying to develop the nuclear rocket. The test system was called Kiwi because it was never intended to fly and was to be tested on the testing grounds in the desert. The idea was to heat up hydrogen with the nuclear reactor and blast it through graphite apertures. Of course, hydrogen erodes carbon and turns into methane. I was working the conversion rate of carbon to methane. The way they had simulated this in the lab was to use an atmospheric pressure plasma arc. We had this humongous hydrogen arc blasting out, with huge cables connecting it to a power source. I measured many carbon samples before and after they were run in the hydrogen plasma arc.

HOLLOWAY: You may have been producing diamonds and didn't know it!

WAITS: We could have been. For some reason, I forgot why, we were interested in gas chromatography. In that time, they were homemade, using volcanic tuff (fine volcanic ash common around Los Alamos) as an adsorption medium, and a model airplane engine glow-plug as a gas thermal conductivity sensor on the output. This was really just sandbox playing, but I learned a bit about gas chromatography. 

HOLLOWAY: Funny to go to New Mexico for the volcanic eruption.

WAITS: Los Alamos is right next to the largest extinct volcanic crater in the United States, Valle Grande.

HOLLOWAY: So where did you go after you finished your master's degree and summer at Los Alamos?

WAITS: I was looking for a job. Of course since I had been isolated from the outside world, as you tend to be in graduate school, I didn't realize there was a recession going on at the time. It was hard. Someone interviewed me from a company called Eitel-McCullough, again, which I hadn't heard of, but they made klystron tubes. They had a project from Sandia that was classified, and were looking for somebody that had an Atomic Energy Commission Q-clearance which I did because I had to have a clearance to work at Los Alamos and Argonne. Someone who had a recent clearance could be renewed quickly, so they hired me.

HOLLOWAY: This is where at?

WAITS: Eitel-McCullough (Eimac), San Bruno, California. Years later they became a division of Varian Associates. They had a contract with Sandia; the work was totally classified and still is as far as I know. First, I had to wait for clearance, and so I did some little projects. When the clearance came through, my first assignment was mass spectrometry  -  back to vacuum again and analyzing the gases from electron tubes. When the Sandia project was cancelled, I ended up working for Lowell Noble who we called a "gentleman scientist," as he had a couple of businesses on the side. He dropped in every once in a while. He was a great guy. He set us, me and a technician, to work on analyzing residual gases during the actual processing of these huge klystron tubes. Of course we needed a small bakeable mass spectrometer, which we didn't have, so we convinced CEC, Consolidated Electrodynamics Corporation, because we had been using their large analytical mass spectrometer previously, to modify their Diatron mass spectrometer tube with an Eimac all-ceramic header so it could be baked out. They used a standard Eimac vacuum tube header and changed the Teflon gasket the seals to gold-wire seals to make a fully bakeable Diatron. During bake-out you have to remove the magnet. We baked the Diatron tube along with the klystron tube. The power klystrons were six or eight feet high  -  major tubes.

HOLLOWAY: Large, very large. So what sort of vacuum pressures were you attempting?

WAITS: They would go down into immeasurable pressures, because they baked them out at 450?, 500? C. The diffusion-pumped system used an Albert copper-foil trap, which was baked also, and copper gasketed knife-edge seals, they would pump as low as they could get. During the process, they "formed" the cathode, which heats up the cathode and turns the carbonates to oxide. It evolved enormous amounts of carbon dioxide plus CO and methane. You see all of these with a mass spectrometer. After a while, it became boring because you see the same old stuff. After processing, the kystron was "nipped off" which means a compression seal of the copper tubulation connecting the kystron to the vacuum station Since the interior surface was so clean and under high vacuum, it would make vacuum-tight copper weld seal. 

HOLLOWAY: When you say immeasurable vacuum...?

WAITS: 10-9 Torr. At that point, you can only measure the vacuum by some sort of auxiliary measurement. They actually put Philips glow discharge gauges, which they made themselves, on the tubes. The fellow I worked for, Lowell Noble, had devised what they called the dynamic getter. Turns out that cathode is a dish-shaped thing at the end of the tube, which could be four or five inches in diameter. As it is dish-shaped, it tends to focus positive ions into the center which wears out the cathode at that point. So the strategy was  -  you just put a hole in it. Then behind the hole, you put a titanium plate. There you go. Now, you've got a sputter-ion pump. So these things pumped themselves as they ran.

HOLLOWAY: Did they know that?

WAITS: Yes, oh yes.

HOLLOWAY: They certainly knew it.

WAITS: Yes. Because ion pumps had been invented by then. Again, they also put what they called appendage ion pumps, which they also made themselves, on the large klystron tubes. They just copied the Varian design. So I learned about ultra-high vacuum.

One of the interesting stories, this technician and I, we were talking with one of the guys that was the hands-on vacuum production expert who saw all these instruments and stuff and he asked us what we were doing. We said, "We are trying to find out what gases are in the tube." He said, "I know what's in the tube." We said, "Oh, what?" He said, "Nitrogen." [Chuckles] I've learned much, much later that Edison knew exactly what was in his electric light bulbs during processing because he'd done residual gas analysis, looking at the glow discharge around the filament with an optical spectrograph and talking to his chemist friend in Princeton to identify the spectral lines.

HOLLOWAY: So Edison didn't have to do RGA analysis; he did optical emission analysis.

WAITS: Yes. He did residual gas analysis, by emissions spectroscopy from a glow discharge [chuckles].

HOLLOWAY: Something we've forgotten about.

WAITS: And have to rediscover.

HOLLOWAY: So were you active in the AVS? Were you a member of the AVS at that time?

WAITS: Lowell Noble sent me to my first AVS conference in Philadelphia to find out what kind of residual gas analyzers we could get that could be bakeable. We looked at things that were available then. We looked at the omegatron sold by Carl Herman Associates, when Carl Herman was there. I was there at the AVS meeting to find out about RGA analysis of vacuum tubes. There was much interest in that subject at that time. We also thought about hiring Paul Redhead to consult on analysis by thermal desorption from a tungsten filament.

HOLLOWAY: What year is this?

WAITS: This would have probably been '59, something like that. Whenever AVS met in Philadelphia [it was 1959]. So we did the work and we gave a paper at the AVS in Washington, D.C., 1961. My paper was the last paper on a Friday. I was terrified. Looking out into the blackness at an audience of maybe a dozen attendees. [Laughter] It was interesting that after the paper, these two guys came up to me. They worked for a getter company. It might have been SGS salesmen. They were really interested in the paper. That was nice to talk to a few people afterwards who were interested in it. We published another paper in the same 1961 ASTM Proceedings that John Vossen, I noticed later, had a paper on cleaning electronic parts. He was starting out. 

HOLLOWAY: It's not how many are in the audience, it's how many are interested.

WAITS: The people who were there were not very many, but they were interested. Being totally unknown, you are always the last paper on a Friday. Right? [Chuckles] 

HOLLOWAY: So did you stay there very long?

WAITS: I stayed at Eimac about four years. That's when I ran into a friend, from TCU, believe it or not, in a bar in San Francisco. He told me about this new company called Fairchild Semiconductor. I had been just missed being laid-off at Eimac. I mean the line was drawn right below me; I was right above the line. Things were not good, and I thought I had pretty much learned all I was going to learn there. I told my friend that, and he said he was working for this new company called Fairchild and thought I might be interested. I said sure. I interviewed him and interviewed with Gordon Moore and got hired into Fairchild R&D, which at that time occupied the first Fairchild building in Mountain View, California. (Now marked by an historical plaque).

HOLLOWAY: You were interviewed by Gordon Moore, eh?

WAITS: He was head of the R&D lab all the while I was there until he left to form Intel and Jim Early came in from Bell Labs to head the lab.

HOLLOWAY: So what year was this, now?

WAITS: January 1963 is when I joined Fairchild. They had been in been in business since - they just had their 50th anniversary this October, and so they started in '57. Right after I joined, the lab moved from Mountain View to a brand new lab in the Stanford Industrial Park, in Palo Alto, right down the road from Stanford. 

HOLLOWAY: They were probably one of the first to move out into that area. 

WAITS: Yes they were. [Shockley, Hewlett Packard, and Varian had preceded Fairchild.] 

HOLLOWAY: Were they making discrete transistors or integrated circuits at that time?

WAITS: Integrated circuits, transistors, and discrete diodes. In Palo Alto, it was all research, and research on integrated circuits. Our task, again vacuum, was to look at thin film resistor processes compatible with integrated circuits. We worked with nichrome resistors, which are notoriously unstable and very, very thin, and you couldn't really make high value resistors. The project evolved into a project to make high value resistors, so we ended up using sputter deposition to make silicon-chromium resistors, which other people had done. Signetics, a spin off of Fairchild, actually had patent on it. Now, I believe it's one of the standard processes in the circuits that convert analog to digital signals using precision resistor arrays.

HOLLOWAY: So this is DC Sputtering?

WAITS: Yeah, DC Sputtering, no water-cooled cathode. Very primitive.

HOLLOWAY: No magnetron.

WAITS: No magnetron. That's another story. [Laughter] So we were at Fairchild. I stayed there 11 years. Turns out that was quite a long time. After Gordon Moore and Robert Noyce founded Intel, it was down hill from then on. They finally actually shut-down the R&D lab, so everybody in R&D was laid off at that point, unless you could find a job somewhere else within the company. I had been sort of consulting on a government contract in Mountain View, again on thin film nichrome resistors, so I scrambled and convinced them that they really needed me to help them get this contract done because they needed somebody to at least write the final report. I got a job in the Radiation Resistant Integrated Circuit group at Fairchild. The parts were used in Trident missiles. 

A few years later, a friend who had left Fairchild the same day I did, called me up and said he was at Hewlett-Packard and they had an interesting possibility there. 

I skipped a few things. Actually while I was at Fairchild in Mountain View, a fellow called me up who had founded a company called Precision Monolithics. His name was Garth Wilson. They had used the Fairchild silicon-chromium process to make resistor arrays for analog to digital convertors. That was a very difficult decision, to leave Fairchild. It took me a long time to convince myself that I could go to this small company, but eventually I did, and helped to make their process a little more reproducible than it was before I came. I just stayed there three years. Again, another recession hit and they were laying off people. Then I got the call from the friend who left Fairchild the same day I did and he had gone to Hewlett Packard and said he had an interesting job. Then I went Hewlett Packard. That decision was very easy, as opposed to the decision to leave Fairchild after 11 years. 

HOLLOWAY: Yeah. So you were still doing thin-film precision resistor work at Precision Monolithics?

WAITS: You bet.

HOLLOWAY: This was all nichrome (NiCr alloy)?

WAITS: No, it was silicon chromium. 

HOLLOWAY: Oh, silicon chromium.

WAITS: When I was at Precision Monolithics (PMI), I looked at this new plasma etching process. Because it was difficult to etch silicon-chromium; you had to use a nitric acid / hydrofluoric acid mixture. For plasma etching, you only had the barrel reactors then. Turns out, one of the innovators and inventors of plasma processing, Steve Irving, was selling barrel etchers for Tegal Corporation. I bought a barrel reactor from Steve and we started using a plasma process to etch the Si-Cr resistors.

HOLLOWAY: What time frame was this now for the plasma?

WAITS: This would have been 1973 or 1974.

HOLLOWAY: Well, that's very early for...

WAITS: Oh yeah, researchers were just discovering the loading effect and realizing that half the circuits would be done before the other half were done in a barrel reactor. Again, wafers were only about, at that time, maybe three inches in diameter. That was also about the same time I gave my first AVS course on film resisters. Well, actually, I didn't feel confident to do a course all by myself, so for the course I talked two other people into helping out, so we had three people teaching one of the first AVS short courses that was given. As I remember it was near Disneyland in Anaheim, California [it was in 1974]. 

HOLLOWAY: This was sponsored by the National AVS or this was the Southern California Chapter?

WAITS: This was the National Symposium. This was before there were any AVS chapters in California. We couldn't seem to get a [Northern California] chapter going, but Frank Ura, on his own initiative, started what he called the Thin Film Society. Every month they had meetings at Hewlett Packard - Frank Ura was at Hewlett Packard. 
I am getting ahead of myself, because now I'm at Hewlett Packard. The project there was to transfer a process that Frank Ura's group had developed for thin film thermal print-heads from H-P's Palo Alto Research Labs to production in Cupertino, California. I lived two miles from the plant in Cupertino, so a very nice commute. During the gas crisis I actually rode a bicycle to work. The resistor film was tantalum-aluminum, deposited by sputtering. Frank Ura and his group had developed a process using magnetron sputtering. So that was one of the first commercial uses of magnetron sputter deposition. He had already contracted to have built, and I think it had been completed, a system to sputter an aluminum oxide protective layer by conventional [diode] RF sputtering. One of the people in the Palo Alto lab was an engineer from the thin-film print-head production in Loveland Colorado. Colorado was making very large pre-heads also, something like 10-12 inches wide; we were just making one that was only about a half inch wide for a portable calculator, for the Advanced Products Division that made the HP35 calculator in Cupertino. That process for deposition of aluminum oxide by RF sputtering was being used in Colorado. I visited there, and they were not happy with the process. The process was really shaky. But, again, the engineer from Colorado working in Frank's lab was looking at using magnetron sputtering. I took one look at the RF magnetron deposited films and said, "That's the way we're going to go." Frank Ura was supportive of that, so we switched to RF magnetron sputtering for the aluminum oxide deposition and ordered a new custom sputtering system.

HOLLOWAY: So what was the advantage that magnetron sputtering gave you?

WAITS: It was basically semi-reactive sputtering, and (by conventional RF sputtering) the film would come out either brown or it would come out yellow, or it would come out clear. It was not very well controlled. If it was brown, it was too soft. You really wanted it clear, but if you got it too clear, then it was brittle. By RF-magnetron deposition, it just came out hard and clear, no problem. It was very, very controllable - although it was slow, it was controllable. 

HOLLOWAY: Do you think it was the stability of the plasma that...?

WAITS: I don't know. Again, it was RF magnetron. It just worked too well. We didn't know a lot about what we were doing, but it worked well. It was slow. It took all day to do about a micron or two thick film deposition. Frank's lab, the people in his lab were terrific. They actually developed an inline magnetron deposition system that was built locally by the Wilder brothers at Circuit Processing Apparatus Co. At this time, you were building your own equipment. At Fairchild, you were building our own vacuum equipment from standard and custom parts. We had a group at Fairchild Labs that built equipment not just for the labs but for production also. 

HOLLOWAY: You would buy or build your bell jar?

WAITS: You would buy your bell jar. You had a lot of infrastructure in that area which helped all of these new integrated circuit companies. We had a machine shop for small jobs. We had all this available outside, and you could have larger jobs contracted out. One of the companies that we dealt with the most, from Fairchild to PMI through Hewlett Packard, was a company called Davis & Wilder. The people that started it were the Wilder brothers. There were three or four of them. They formed Davis & Wilder, and then they sold that and formed a company called Circuits Processing Apparatus (CPA). By the time we were at Hewlett Packard, they were making the in-line systems for Frank Ura. Pretty much Frank's group's design. It was really a cooperative thing. They were designing things, HP labs were designing things, and also at that time MRC was doing magnetron sputtering with an in-line system. CPA in-line systems are still in use and still around. Some of the people that worked for Circuits Processing Apparatus started other companies to make and service those systems.

HOLLOWAY: So this was oil diffusion pumps?

WAITS: Yep, oil diffusion pumps. We used the so-called Milleron trap that CPA designed based on Norman Milleron's theoretical design. Milleron was at Livermore National Lab and an active AVS member. I attended one of his University of California extension vacuum classes in the 1960s. 

HOLLOWAY: These were liquid nitrogen traps.

WAITS: Liquid nitrogen, right. It was designed so that it wasn't that sensitive to the level of liquid nitrogen.

HOLLOWAY: So your gauging was a Philips gauge or Bayard-Alpert gauge? 

WAITS: Bayard-Alpert gauges. All the Bayard-Alpert gauges then were standard ¾ inch tubulation, but I always asked for a one-inch tubulation. We had one in the history booth that had a ½ inch tubulation. They didn't believe in reasonable conductance to an ion gauge, I guess.

HOLLOWAY: [Chuckles] Thermocouple gauges were used on the fore lines?

WAITS: Yes, thermocouple gauges. I used thermocouple gauges and Pirani gauges in some of the work at Eitel-McCullough. I didn't really like using them [Pirani] because they were very difficult to calibrate.

HOLLOWAY: So you had a factory job that got you into microcircuits, microelectronics, and into thin films. Is that a new activity for you?

WAITS: Well, at H-P that was basically sputtering, and sputtering systems, which I worked with in these companies before, and it was great. So there were some challenges. We had problems dealing with the cracking of the oxide wear layer on the H-P print-heads. Turned out to be mainly due to the paper and not due to our process. We had problems with adhesion of the tantalum-aluminum film resistive film. If you don't have problems, it's not much fun.

HOLLOWAY: [Chuckles] It gives you a challenge.

WAITS: Yep, that had the adrenaline going. 
Another innovation was introduced by Steve Muto, one of the engineers working in Frank Ura's group. Between Frank and Steve, they developed all of the thin film processes. They had developed what they called SEPE, Sputter Edge Plasma Etch, which actually was reactive ion etching, and that's how the tantalum-aluminum film was etched. Then we ran into the same things that the integrated circuit people ran into: the built-up material on the side walls and all those problems that involved. When that project was transferred and the portable calculator division moved to Corvallis, Oregon, I took the family to Corvallis to check it out. We never saw the sun while we were there, and we decided we really didn't want to move to Oregon. At that time H-P was a very progressive company, and they weren't going to let you go if you didn't want to transfer to Oregon. You were guaranteed a job within H-P, but you had to interview for the jobs and you had to find them. Managers were given the responsibility of absorbing a given number of people. They could interview them, and they could choose them, but when the chips were down, they had to hire people. 

I transferred to the Cupertino Integrated Circuits, which was right across the street from where I was working anyway. I worked on projects that didn't go so well, like trying to plasma deposit silicon nitride. I purchased the worst possible system to do it with. This division made silicon on sapphire, which was referred to as "cruddy EPI" because it was very difficult to deposit good epitaxial silicon on sapphire. 

This is another case where AVS came to the rescue, because I came to know John Vossen at RCA. Some of the circuits were very, very sensitive, and I wanted to switch them over from electron beam evaporation (of aluminum) to magnetron sputtering in the new fab that they were building, so we did some experiments where we did magnetron sputtering on split lots, The product engineer found that there was a very slight but a significant yield difference between the two. The normal yield so poor there was no way was he going to accept magnetron sputtering. That small difference was enough to impact his yield. So I called John and I said, "What's happening here?" He said, "Well, the problem is probably the ultraviolet light from the plasma." And he said, "Do an experiment. You'll prove it. Deposit a very thin electron beam layer which will mask the ultraviolet light, and do the rest of it by magnetron. See how that works." That answered the question absolutely. That was the problem. John suggested that the only way to solve that problem was to lower the magnetron voltage so the ultraviolet light energy was less, so you could do a very slow deposition at the beginning. I think we just didn't do magnetron deposition on that particular circuit. It was done with electron beam because it was too much trouble to do a special magnetron process. That's another way that the AVS helps. You get to know other knowledgeable people and get a different take on your problems.

HOLLOWAY: Right. It's remarkable the expertise that's residual in the society. And the other thing is how frequently and how willingly knowledge is shared amongst...

WAITS: Yeah, John Vossen was extremely good at that. Of course, the other John, John Thornton, was working on magnetron sputtering too, and he was a good source of information, mainly from his writings.

HOLLOWAY: His thin film deposition papers are considered to be classic in that area. So you were then in the middle of microelectronics again.

WAITS: Yes, Then one of my bosses - I had several bosses over my time at HP CICO (Cupertino Integrated Circuits Organization), but one of them joined a company called Trilogy, which was another start-up. Of course, we were all jumping to start-ups to get rich. He convinced me to go to Trilogy and help set up their integrated circuit production. Their objective was to do wafer-scale integration - a whole computer on a single wafer, or on several wafers.

HOLLOWAY: And this was what year?

WAITS: This was in 1981. The idea was, needless to say, ahead of its time. [At that time a megabyte of memory on a single wafer was a big deal.] The processes were not there yet. We absolutely had to have processes for etching aluminum by plasma, which didn't exist at that time. People were working on them, but they weren't production-ready. At that time, I knew David Lam from Hewlett Packard, and I was actually working on aluminum plasma etching at Hewlett Packard in the integrated circuit group as well as on silicon nitride plasma deposition.

HOLLOWAY: What sort of plasmas were you trying to...?

WAITS: For aluminum, it was chlorine, basically. That's was before people were using BCl3. We used carbon tetrachloride to etch polysilicon. We were actually trying to use a magnetron etcher, trying to keep that a secret. It worked pretty well, but it fried the photoresist. [Chuckles] So we never really got that into production. Considering we were in the production area, we were doing a lot of R&D work. 

I'm trying to figure out where I am now. I am at Hewlett Packard. My boss went to a company called Trilogy, which was founded by Gene Amdahl who had founded the Amdahl Corporation which had been very successful in making IBM type computers. He came originally from IBM. He had raised $180 million to start Trilogy. It was a lot of money then in 1981 to form a major computer corporation to make computers using wafer-scale integration. Again, it was ahead of its time. We didn't have the processes. David Lam (from HP) had started the Lam Research Co. (LRC) to make plasma etchers and we bought the first or second system that they built. One of LRC's main engineers was Steve Muto, who had come from Frank Ura's group at Fairchild. Steve had also transferred from Frank's lab to Cupertino Integrated Circuits. I think he might have been my boss for a while. Everybody worked together so closely it was hard to distinguish the boss from... [laughter]. 

HOLLOWAY: Paths cross many different ways, too. 

WAITS: Lam said they were going to ship me their first plasma etcher, but I have a feeling they shipped the first one to IBM. Anyway, it was the first single wafer etcher. Basically, the design of plasma etchers evolved from Lam's single-wafer etcher.

HOLLOWAY: This etching was done at what sort of pressures?

WAITS: Low-pressure, as I remember.

HOLLOWAY: So you produced the vacuum and then back fill it with etching gas?

WAITS: Yes. And used Freon gases for polysilicon. It was various types of Freons. The main difficulty in all this work, including the sputtering, is to actually develop a production worthy process that could be reproduced. 

HOLLOWAY: For the gauging that used at that point in time, was it capacitance manometer gauging at this point?

WAITS: Right. [Also Granville-Phillips Convectron thermocouple gauges]

HOLLOWAY: That had been introduced just in that time frame then?

WAITS: We were using MKS capacitance manometers. Which reminds me that we did attempt to use a Bayard-Alpert gauge during magnetron etching using carbon tetrachloride, and found out right away that it would clean the thorium oxide right off the iridium filament. You got a nice, bright, shiny iridium filament that wouldn't emit electrons. So I had to go back to tungsten filaments for the Bayard-Alpert gauges in those systems.

HOLLOWAY: And it was still oil diffusion pump and liquid nitrogen traps. 


HOLLOWAY: Oil mechanical vacuum pumps. 

WAITS: Yes. I'm trying to think. My other involvement with vacuum was at the end of R&D lab in Fairchild at the time of the R&D lab. Jim Early had come Bell Labs, and he brought with him some other people from Bell Labs, including the fellow who became the head of Apple Computer. His name escapes me right now [Gilbert Amelio]. They wanted to establish ion implantation because they were very familiar with the process advantages. At the time we had no ion implanters. And so I very involved in the setting up...actually specifying the kind of vacuum and vacuum pumping and what not for the first ion implanters at Fairchild R&D. I went to Extrion Co. in Peabody, Massachusetts to consult with them and help specify the vacuum systems, which then used diffusion pumps and liquid nitrogen traps.

HOLLOWAY: Right. What type of vacuum levels were you achieving for that?

WAITS: 10-7, 10-8 Torr; that was about all. You were never got into ultra-high vacuum as we do right now with many production systems that are now being built. 

HOLLOWAY: And these early ion implantation systems were O-ring sealed?

WAITS: Oh, yeah.

HOLLOWAY: Oh, O-ring sealed using Viton or neoprene?

WAITS: Yeah. We switched to Viton way back in the Eimac and Fairchild days.

HOLLOWAY: Oh, is that right? So Viton has been around for a long time. So now you are at Trilogy. 

WAITS: Okay, now I'm at Trilogy. That was a struggle. We never really succeeded in doing what we were trying to do, which was to make very large-scale integration. You just couldn't meet the line-width specs, especially in aluminum. Plasma etching was yet production-worthy, so we ended up using an aluminum lift-off process.

HOLLOWAY: What kind of line widths were you trying to achieve?

WAITS: I really can't remember, but it would seem crude now (a one megabit memory was only a dream then). One of the engineers went to a lift-off which involved depositing aluminum on a photoresist, actually, and then lifting off with ketone type solvents, which were very flammable. The problem with lift-off is if anything re-deposits back on the wafer, you are never going to get it off again. It just was never really going to work. At least we couldn't get it to work. 

Trilogy failed, but at that time one of the investors from Trilogy was Digital Equipment Corporation, and DEC decided that they wanted to build on what we had done. By the way, Trilogy circuits were not CMOS; they were conventional bipolar transistors. That type of circuitry uses a lot of power, so it had to be water-cooled. But DEC was thinking they were going to be able to build on what that we had done, so they absorbed the R&D portion of Trilogy in 1986, which I was a part of at that time. We kept going in the same location until 1992. 

HOLLOWAY: This was in California?

WAITS: Yes, Cupertino. It was actually right across the street from H-P Cupertino Integrated Circuits. Trilogy built a beautiful building there, which is now occupied by Hewlett-Packard [laughs]. We were involved in setting up two fabs there to build circuits. A small one, then they expanded into a larger one. By that time, we were in the ultra clean room era. They brought in some more engineering talent and they actually did develop a process that was semi-wafer scale integration including a water-cooled module that very competitive with what IBM was doing at the same time. using different technology. Our technique was incredibly difficult  -  we ended up making a multi-layer polyimide copper film stack which was deposited on not silicon wafers but aluminum wafers, and then the aluminum was sacrificed or etched away at the end. The basis for the aluminum wafers were the disks that were used in hard drives because the method for making nice, flat, and smooth aluminum came from that technology. These were like three or four inches in diameter. 

HOLLOWAY: So how many layers were used for that?

WAITS: We had about four layers. So it was copper polyimide, so we ended up plasma etching polyimide and pattern-plating copper to define the copper layer.

HOLLOWAY: This was an idea that generated off of hybrid microcircuit boards or?

WAITS: What were really doing was making, yeah, a circuit board that the chips would be put in.

HOLLOWAY: Were hybrid microcircuit boards prevalently used at that time as well?

WAITS: Yes, at that time, there were wide circuit boards. We were making miniature, a micro circuit board. Shrinking the size of the circuit board is what it amounted to. Then DEC realized that MOS was the way to go. They had made several computers with bipolar transistor technology, but they decided to abandon it. DEC was very similar to Hewlett Packard in the way they treated their people. It was a good company, so we had a schedule, we knew exactly when the plant was going to be shut down. We had system for placing people. I was lucky because on the same day I was going to be laid-off, I was also eligible for voluntary early retirement. I had a choice: be laid-off or to take voluntary early retirement. 

HOLLOWAY: Let me think [laughs].

WAITS: That was a very difficult choice. 
Then I was looking for a job [during a recession], and it took five months. But I was supported by a professional job-placement organization (paid for by DEC), which helped keep my morale up, but it didn't help me get a job, because I ended up getting a job from someone who I had dealt with as a customer (he was selling vacuum valves at that time). He kidded me about being very disappointed because I never bought his valve. So I ended up working for him at a company called UTI, which made mass spectrometers. So I was back to mass spectrometers, which I had used at Eimac and Fairchild. Again, history repeated itself, and UTI was later acquired by a Massachusetts company, MKS. 

HOLLOWAY: So what year was this now?

WAITS: This was in 1992

HOLLOWAY: '92. Okay. What year did MKS acquire UTI?

WAITS: In 1995. In 1992 it was UTI Instruments, which brought me back to when I first used a mass spectrometer - not only at Eimac but also at Fairchild, where I brought a CEC Diatron and put it on bell jar systems. We were looking at residual gases found in all this water and hydrogen, so it got pretty boring. So at UTI I was able to get up to speed on mass spectrometers pretty quickly. I was an applications engineer. I got to travel around a lot; mainly just in the U.S., though, at places like Bell Labs and IBM, and other non-semiconductor customers. 

HOLLOWAY: Now these were quadrupole?

WAITS: Yes they were quadrupole mass spectrometers, which are much nicer than the magnetic sector instruments, and they could be baked out if you took the electronics off.

HOLLOWAY: So MKS had both magnetic sector mass spectrometers as well as quadrupole spectrometers.

WAITS: Just quadrupole instruments. At that time, in 1995, MKS had their own mass spectrometer division, and they just brought UTI to add to that division. The upshot of the whole thing was that it just eliminated UTI as a competitor. 

HOLLOWAY: So the original MKS mass spectrometer line was a quadrupole line as well?

WAITS: Yes. 

HOLLOWAY: So you just recently had retired from MKS then?

WAITS: Yes about 5 months before. I left DEC in '92, and I worked at UTI-MKS until 1998 Then MKS laid me off. When they did, I was in Albuquerque giving a talk at the AVS local chapter meeting, and when I came back, they said, "Where were you?" [Laughter] 

HOLLOWAY: "We had an important talk while you were gone."

WAITS: But I was, as you know, selling mass spectrometers (in Albuquerque). I had been doing a lot of writing at MKS, writing white papers and articles for magazines. Donna Bakale (another active AVS member), at Technical Marketing Programs (TMP) in nearby Sunnyvale, California (where I lived), had MKS as a client and I had been working with her on articles for trade magazines. I had written an article that was published in May of '97, which was a response to the editor who had called Donna and said, "Help, we need an article right away for the 40th anniversary issue of our magazine." She called me up and said, "Can you do this?" I said, "Sure." So I very quickly wrote an article. Anyway, [after I got laid off] I called Donna and said, "Can we have lunch?" She said, "You know we don't have to have lunch." I said, "I just want a half-time job. I don't want to work full time." I started working for her for half-time, and she said, "Are you sure you don't want a full-time job?" I said, "No, I don't want a full-time job." I worked with her for half-time until finally she'd basically retired and was operating the company from her boat. Finally, in 2004, TMP was dissolved, and that's when I stopped working. I've done some writing and consulting since then, but nothing steady.

HOLLOWAY: So for Donna's company you would write articles about vacuum and general processing?

WAITS: I would edit. I would take articles that had been written. Like Applied Materials was a client, Lam Research, was a client, other companies. I learned a lot about other areas of technology also. They'd get an article that would be written and edit it, or sometimes just interview people and write an article based on an interview. 

HOLLOWAY: That process over the last few years, certainly the semiconductor industry and microelectronics industry has under gone a transition from the wet vacuum pumps to the dry pumps. 

WAITS: Absolutely.

HOLLOWAY: Can you comment on some aspects of that?

WAITS: Well, when I was working for MKS, we were using the mass spectrometers on systems that were being made. For example, in the Endura sputter deposition system for Applied Materials, and I was astounded at what we saw, the degree of the level of the vacuum. Applied Materials used cryo pumps in the ultra high vacuum portions of the Endura systems. They were down in the 10-9 range, and really were dealing with background which we could see was hydrogen. We discovered some interesting things which didn't really affect process. The hydrogen was coming probably from the aluminum sputter cathode. The hydrogen level - we are talking about very, very low levels - would depend on the rate of wafer loading. If the sputter time was relatively short for the wafers to be going in and out of the system, the average hydrogen level would stay fairly low. If the sputter times were longer, the background hydrogen would slowly drift up. It would go up and down with the wafers cycling, but the general level would sort of drift up. You noticed the cryo pumps weren't quite keeping up. But it didn't matter, because it was the normal process, and the process engineer couldn't care less about this.

So yes, I was really impressed with the quality, considering that the systems still use Viton O-rings. They weren't metal seals. The systems weren't really baked at 450° C or anything like that, but they were able to achieve ultra-high vacuum.

HOLLOWAY: The wafers were introduced by a load-lock system?

WAITS: Yes, a load-lock system. They had a three-stage system. One is a load-lock; then there was an intermediate stage where the wafer transfer was done, which was surrounded by the actual ultra-high vacuum process modules.

WAITS: The cryos, of course, didn't have a backing vacuum pump. The turbo pumps may or may not have had a dry pump. The tendency at that time - this was in the late '90s - was to sort of stick with what you had. Since the very late '90s when they started working predominantly, that's when the dry pumps were definitely coming in as the backing pumps for almost all of the turbos, and for the cryos during regeneration.

HOLLOWAY: And when they were using the wet pumps, of course, they had probably had the fore-line traps on there for the cryo pumps, particularly. 

WAITS: Sometimes they would have these, like you said, fore-line traps, yeah. One thing that people discovered, and we found way back at Fairchild using that CEC Diatron mass spectrometer that if you pumped down to below 50 millitorr during your roughing, that's when you were doing to get your hydrocarbons back into the chamber. 

HOLLOWAY: Oh-oh, contamination.

WAITS: Contamination wasn't coming as much from the diffusion pump as it was from the mechanical pump. Of course they were starting to build diffusion pumps that had much higher fore-pressure tolerances so that was much better too, then you didn't have to rough down quite so low before cross-over to the diffusion pump.

HOLLOWAY: During this time of the late '90s and early 2000s, dry etching and the plasma etching processes became more established and very ubiquitous. Could you make a comment about that?

WAITS: Essential. My direct involvement with plasma processes ended sometime during the Trilogy and DEC phase, and became indirect. At DEC we were plasma etching polyimide. So I don't have any direct experience after that; just from helping to edit papers by people at companies like Applied Materials, Lam Research and others. Yeah, both the plasma etching and plasma CVD (chemical vapor deposition) are essential to the IC processes now. 

HOLLOWAY: Applied Materials and companies similar to them have made a living recently off of turn-key systems where they develop a system, test it, demonstrate it, qualify it, and then transfer it into the customer's house. Which is quite different from where you started where you had to build the system completely from scratch yourself. I was wondering if you would comment about that?

WAITS: In the early days of equipment development vendors would claim a process that we all knew they didn't have. They didn't really have the capability and the resources to develop and demonstrate a process. It was very difficult for them to do that. They had no research infrastructure. So we all knew back then in the '60s and '70s that you buy the system, then you rebuild it and re-engineer it. I know that at the time the sales people would say  - Well, IBM has this system. IBM bought our system last week or last month or last year. We'd say, "Okay, yeah, yeah." We knew that IBM probably had rebuilt that system to get to really be production ready. So how much did IBM have to put into it or re-build or whatever to make it work? We would be very skeptical, and try to work with vendors to make the hardware as good as possible, but knowing that we would probably have to make some changes. We didn't ask them to produce a process, to engineer a process. Now that has changed. The process development that used to be done in the companies is now being done by the vendors of the equipment. They do have to guarantee a process. I saw that while working for MKS and UTI when we were selling mass spectrometers. What is a mass spectrometer? It's a trouble-shooting tool. An engineer from one major semiconductor company told us: "If we have to have a mass spectrometer on it to make the process work, we aren't going to buy that tool." We had no good answer for that.

HOLLOWAY: Well, your career has spanned quite a long time and a large change in the technology, Bob. I was wondering if you wanted to add anything else to the end of the interview today. 

WAITS: Well, Fairchild had its 50th anniversary, as I mentioned, last week. They had a three-day party, a lot of panel discussions. Gordon Moore was asked what's impressed him. I would have to paraphrase what he said. He was quoted in The Wall-Street Journal that he is still astounded that you can go online and in seconds search an incredible amount of material and information. That's, of course, not just software; it's through the integrated circuits, the incredible speed of the circuitry. 

HOLLOWAY: Right, it's the complementary interactivity between the hardware and the software capability. Without either of those two, you don't have the end result. 

WAITS: Yes. I'm in awe of the circuitry that is being built now, compared to what it was when our dimensions were in mils - a mil is about 25 micrometers. A tenth mil, 2 microns, that was pretty good if you could get that kind of control. Now we are talking about nanometers and nanometer-type process. You need much more powerful microscopes to look at these things. We used to use modify lab microscopes to do the inspections. You can't do the inspections quite the way we were doing it way back; the circuits are far too complex. It's all got to be automated.

HOLLOWAY: Well, that's right. The characterization tools have come along with the vacuum equipment and everything else.

WAITS: Yeah, characterization was a microscope inspection. Of course scanning spectrometers, reflecting spectrometers measuring the thickness, that was a major breakthrough right there. Then the introduction of the Sloan Dectak profilometer to measure the thicknesses.

HOLLOWAY: Surface profilometers.

WAITS: Surface profilometers - huge. I remember IBM had some very large profilometers that were made in England to measure the thicknesses of magnetic films on their first hard disks. The Dectak was a major breakthrough in making a tabletop instrument, something that could be used in a lab and then in production. [The Dectak was essential for Fairchild's Isoplanar process]. Now measurement tools are all computer controlled to do the calculations that couldn't really be done before. They use a lot of techniques that couldn't possibly have been done without the computer control and the automated calculations. 

HOLLOWAY: The yield analysis that goes on in that, they make it profitable, and they make it profitable at the relatively low costs that we have to pay for the integrated circuits today. That's truly remarkable as well.

WAITS: Yes. You used to worry about losing the process. I've been reading recently about that now. Now, no one is worrying about losing the process. That's why it was difficult to deal with any kind of changes in production - you lose the process. This was at HP, they changed a process. I forgot the details, but they changed the process involving a cleaning process during assembly. All of the sudden the yield went 'kaplooey'. The details don't come to mind. But it was situation with a totally unexpected result - what seemed to be a benign change destroyed the yield.

HOLLOWAY: The old adage, "If it works, don't fix it." Okay, thanks very much for participating in the interview today, Bob.

return to top