AVS Historical Persons | Jack Kilby - 1997

Jack Kilby - 1997

Oral History Interview with Jack Kilby

Interviewed by Don Meyer, Oct 21, 1997
 
MEYER: My name is Don Meyer. I'm a member of the American Vacuum Society and a former employee of Texas Instruments during that time period that Jack Kilby, the subject of our interview, invented the integrated circuit. The occasion for this interview is the presence of Mr. Kilby at our 44th Symposium of the American Vacuum Society in San Jose, California. 

It's quite an honor for me, Jack, to interview you on behalf of the History Committee of the American Vacuum Society. This is particularly true because next year is going to be the 40th anniversary of your invention of the integrated circuit, or as perhaps we more commonly call it today, the chip. To me, it's almost inconceivable what has happened during these past 40 years, particularly as our lives have been affected by the chip.

I could cite all sorts of data and examples, but there is one thing I would like to begin our interview with. That's just the idea, back in 1958 when you invented the integrated circuit, this one-inch silicon wafer was the standard of the semiconductor industry. Today, this 8-inch silicon wafer is the standard. You can see the difference. Jack, when you invented the monolithic integrated circuit in 1958, did you in your wildest dreams ever anticipate something from this to this?

kilby.jpgKILBY: No, I certainly did not. That's evolved somewhat gradually over the years, and that process is still going on. Last week I saw some 12- and 16-inch wafers, so that process isn't over yet.

MEYER: Before we go much further, would you tell us something about yourself? What led up to your employment at Texas Instruments in early 1958?

KILBY: When I got out of school, I took a job with a company called Central Lab in Milwaukee. I didn't know it at the time, but it was a rather fortunate choice because they were involved in making what was probably the country's first integrated circuits of any type. They were making thick -film hybrids. Originally, these had been developed for use in proximity fuses. The time I got there, we were much interested in hearing aid amplifiers, resistor-capacitor combinations for television, things of that sort. 

Central Lab was interested in transistors. We took out a Bell license in 1952, and I began making small quantities of germanium alloy transistors, which were designed to be incorporated in the thick film circuits. By 1958, it was pretty clear that they didn't really have much interest in that business. I began to look around for another opportunity. Texas Instruments was the one I chose.

MEYER: I'm sure you didn't come to Texas with the idea in mind of inventing the integrated circuit. Could you expand a little bit more on what your original goal was there, coming to TI?

KILBY: Well, in 1958, there was much interest throughout the country in something called miniaturization. This was not just for the sake of making things smaller, but the intent was to find automated ways to assemble electronics. I was hired to work on miniaturization and eventually chose the monolithic approach as what I thought was the best way to do it. 

MEYER: So you were making what was called the micromodule then. Could you explain a little bit about the micromodule?

KILBY: The micromodule was probably the biggest program of its kind in the country. It was funded by the 1Signal Corps through RCA. Their intent was to put all of the components on roughly a one-centimeter square chip. These could then be automatically stacked and interconnected into complete circuits. I didn't like the micromodule approach and never really worked on it, but it was current at the time. 

MEYER: What gave you the idea to make this quantum jump from the micromodule to the monolithic IC?

KILBY: As I began to look at other techniques and began to understand Texas Instruments, I realized that the only thing that a semiconductor company could make cost effectively were semiconductors. This led to the thought that we really did have all the elements we needed within a semiconductor, and could therefore put them all together. I think it came about in that way.

MEYER: Could you tell us a little bit about the fabrication techniques of these first prototype integrated circuits, and if indeed vacuum science and technology played much of a part?

KILBY: The semiconductor process equipment in those days was very crude. We were just switching from the closed-tube diffusions that had been used on the earliest devices to open-tube diffusions, which was a major change. Actually, the first germanium transistor wafers were a centimeter square, so they were considerably smaller than that. Lithography equipment was also simple. We used film photo mask and a photoflood bulb. Chances are the exposure station didn't cost more than $100. We did use vacuum equipment for evaporation of the contact materials. But all of the equipment was infinitely simple, cheaper, cruder than anything that would be used today. 

MEYER: As I recall, your first circuits - and correct me if I'm wrong on this - the interconnect leads, didn't you glue them down with what we called 2Aquadag, which was a silver paste. 

KILBY: The first circuit that I made was intended just to demonstrate that you could get a variety of components to work within the same block of material. To do that, I took an existing germanium transistor wafer and had one of the technicians hand-mask the areas that I wanted to retain in mesas. The external leads were done with gold-bonded wires. The first one had only one internal connection, and that was also made with gold bond.

MEYER: Could you express, get your feeling, looking at today's technologies, today's processing, what impact vacuum science and technology has had? Could you just give us your feeling on that?

KILBY: I think the vacuum processes are extremely critical. Today much more of the complexity of the circuit is above the silicon than below it, and vacuum processes have a major role in those layers. 

MEYER: When you came to Texas Instruments, you became part of what we know as the semiconductor industry. What are some of your early recollections of this industry in general and Texas Instruments in particular?

KILBY: We were making transistors in small quantities at Central Lab, so I'm not sure just when you date my entry into the industry. But the semiconductor business was a very small one. In 1958, I think TI's total billings were less than $100 million dollars. Probably half of that was semiconductors. They had at that time perhaps 70% of the market. So the total market was very small. I remember we were almost through the '60s before semiconductors reached the $300 million dollar level, which made it almost the size of the dog food industry in this country. [Laughs]

MEYER: Looking back on those early days, do you recall - I'm sure you do - some of the movers and shovers in the industry? Could you comment on some that you knew and how you interacted with them and so forth? You know, the3Pat Haggertys and what have you.

KILBY: Well, I think the first that you'd have to mention certainly from the inception of the transistor up through the '60s were the people of Bell Labs, Jack Morton in particular, who drove those programs very effectively. But there were dozens of others, very capable Bell people who really laid the foundations for the industry as evolved. 

At TI, we also have some strong figures. Gordon Teal, who had been the first to grow single-crystal semiconductors, was in charge of the central research activity. Willis Adcock, who had invented the grown diffuse transistor, was in charge of semiconductor R&D. Bob Pritchard, Jay Thornhill, Archie Broodo and many others were involved at that time.

MEYER: So things were moving when you came to TI, but your invention of the integrated circuit put an entire different direction to it, as I recall. Was there much excitement at Texas Instruments at the results of this?

KILBY: TI was the most successful semiconductor company in the world at the time I started. That didn't change. They had very successful germanium alloy and silicon transistor programs and were moving fairly rapidly. The integrated circuit did produce a change over the next five years or so. But it was not dramatic; as people began to buy and use circuits, we built production capacity. But that didn't happen overnight.

MEYER: As I recall, it sort of started slowly. It wasn't like some of our - today, like I remember when the MOS transistor started out. Quite rapidly, it evolved into a competitor with the bipolar. That happened quite-- 

KILBY: I'm not sure you're quite right about that. The MOS transistor was invented in the early '60s. People didn't know how to make stable MOS transistors until the early '70s. So there was a ten-year period there where they couldn't do much except talk about it. So that was a difference. 

MEYER: I can recall, though, there in the lab where you were working, we did not…I guess I could say we didn't really appreciate what you were doing. We honestly didn't. Those of us who were sort of looking over your shoulder really - and it's ironic looking back on that time. If you would have asked me, “Was anything great going on there?” I would have said, “Somebody is here wasting their time.” Was this a general feeling, or was that just my feeling?

KILBY: The use of monolithic semiconductors to make integrated circuits was one of perhaps two dozen approaches going on around the country. For a number of years, it wasn't obvious which the winner would be. TI, for example, spent some money on micromodules as well as some money on monolithic things. So perhaps there was less to get excited about. Today, people tend to assume that the advantages of the integrated circuit must have been obvious to everyone on the day that they first saw them. That really wasn't true. 

There were a number of criticisms made of the monolithic approach. The first one was that it didn't make good use of materials; that you could make better resistors from tantalum nitride and better capacitors with Kapton or something. The second was that the yields were always defeated. If you add ten parts with a 90% yield, everyone with a slide rule could run that out. I think the answer came out at 18%, and everybody knew that you couldn't make components with a 90% yield. And the third thought was that it would put all the circuit designers in the world out of business. These critiques were never really satisfactorily solved. We provided the technical entertainment at meetings around the country while we debated them. But it wasn't until we began to produce circuits in some volume that those questions began to go on. 

But the most sophisticated semiconductor makers in the country didn't really have anything to do with integrated circuits for ten years after it was announced. The acceptance was not automatic.

MEYER: I recall I was more or less the champion of the grown junction transistor. I looked at the monolithic circuit as of little use to a company that I felt was moving ahead and making all this money on the grown junction transistor. Let me ask you this: what influence, what effect did the military have on the development of the microelectronics industry in general, but your work with integrated circuits?

KILBY: In the early days, the military had a tremendous effect. In the early transistor days, they funded much of the basic research at Bell Labs. They also were very responsive to the integrated circuit. The two biggest integrated circuit programs of the early '60s were the Minuteman Missile, which was extremely important, not only to us at TI but also to… And the Apollo Program, which used circuits in their guidance computer. 

MEYER: Do you recall one of your first efforts to do massive integration on a wafer for the Air Force? One of the first efforts to do interconnects on a wafer? The Air Force project. I forget now what it was called. This was a contract from the Air Force to-- 

KILBY: We began to interconnect on the wafer pretty early. The first commercial circuits that we announced were the Series 51, which had all interconnections on the surface. That was announced in August of '61. Those were really the first parts that we tried seriously to sell outside. 

MEYER: I seem to recall you were doing work then with what we might call today wafer-scale integration where you were-- 

KILBY: Yes. We did some, not as early as '62, but by '64-5 we were doing some work of that sort. 

MEYER: But this did not pan out, obviously, because it's…

KILBY: Oh, it's hard to say. The approach we were using later became, I think, the gate array technique, which has been useful. I don't think anybody still was talking about whole wafer kinds of things seriously.

MEYER: Of course in those days, the wafers were small enough that you could imagine that it was possible. I would hate to have to do a wafer scale on an eight-inch wafer. Let me ask you this: the very fact that we have gone to larger wafers, this has really helped with the integrated circuit development. Wouldn't that be true to say?

KILBY: I think it has. On the other hand, like some of the other factors in integrated circuit making had, probably coming into a period of decreasing return for larger wafers. You don't save silicon that way anymore because you have to make them thicker. What you do save is the handling through the front-end process. A number of these things seem to be reaching a point of diminishing return - making transistors smaller doesn't provide the same electrical benefits that it used to, things of that sort.

MEYER: If one reads the list of your awards, and it's quite lengthy, this shows that many people have recognized by now what you've done, and not just since you invented the integrated circuit but since then. Could you give us some idea of what you have been doing since these earlier days? Where your interests have laid and so forth? I know you have served on government committees and so forth. 

KILBY: For a number of years after I ceased to be an active employee of TI, I worked as an independent inventor. Not too successful. Some of the products that I came up with sold; others did not. I also got involved with a solar energy project. I've always done a bit of consulting work with industry. 

MEYER: Could you tell us something of what you did for the government, some of the committees that you served on for the government?

KILBY: In the early semiconductor days, the government was a major factor, and still is in funding university research. At that time, each of the three services had independent programs. There was a problem in that they didn't always talk to each other. When a hot new project came along, they all jumped to get their money on it and things of that sort. I was asked to serve on a group called the Advisory Group on Electron Devices, which had originally been set up during World War II to keep people from building the same vacuum tubes. We reviewed the service programs and tried to coordinate them. 

MEYER: You just mentioned leaving TI. Not that there was any connection; there certain wasn't. But basically, we left TI at about the same time for different reasons. One of the reasons, though, was the participation or the entrance of the Japanese into this industry. Could you recall some of the early days about how the Japanese played into the semiconductor industry, to Texas Instruments and so forth? Could you give us a tie-in to that?

KILBY: I'm not sure that I'm the best one to do that, Don. Initially in 1952, 1956 when Bell had its meeting for a semiconductor licensee, I don't think there were any Japanese companies involved. I visited Japan in 1967 and a number of companies then had begun to do work to get into the field, mostly at the test tube and beaker level at that point. Some of them had US connections that provided some technology transfer. They did a lot of it on their own. Certainly by the mid-'70s they were a major force. 

MEYER: I think I might have used the date of '57 as leaving, but it was '75 the year, wasn't it? Yes, that was the year. And by that time, the Japanese had really - I mean, we were hurting, the Japanese were hurting us. I can recall later on in the '70s there was some great concern that we were going to lose everything to the Japanese.

KILBY: Yes, that's true. There was a period there where American industry almost panicked and felt that it would all go to the Japanese. That's turned around now, and I don't know that anybody has that feeling anymore. But it certainly was prevalent.

MEYER: Jack, let me ask you to go back in history a little further. When you were in college, you got your degree in electrical engineering. Did you have any understanding at that time what you might be doing?

KILBY: No. I finished my last two years of college right after World War II at the University of Illinois. At that time, they were just making a transition from power engineering to electronics. So I had some vacuum tube courses, but not very many. They were mostly taught by ex-radar officers who were pretty good and knew their stuff all right, but they had very few teaching skills. Well, the power courses had a group of old pro teachers. They really knew how to teach their subject. So I probably learned more from the power faculty than I did from the electronics group.

MEYER: Just following up on that, one of the keys, I think, to our progress in microelectronics has been better colleges. Our universities have really come on board with this, and now you're not a school at all unless you have some clean rooms somewhere within your facility.

KILBY: I think that's been true. That's an outgrowth or course that the military began almost as soon as there was a transistor, I think. Most of those facilities either exist because of government money or because of the hope for government money. They're all competing for dollars from that pool. That's a pool which I think is decreasing. Some of these groups will probably have serious difficulty.

MEYER: But there is certainly a widespread interest, even though, probably, the job market isn't as great as some of these schools indicate. Although right now, at this time period, the market's pretty tight.

KILBY: It's pretty good. In addition to the federal funding, the states have also picked up on this as a source of industrial development. So you find states like Mississippi, South Carolina, have invested major amounts of money in microelectronic courses. I'm sure they're turning out some good graduates, but I would not be hopeful that they'll get any large semiconductor activity.

MEYER: Again, thinking back on the years, I certainly want to be positive because I have to be positive. Look what's happened to our industry. But I can recall some of the major semiconductor companies making big mistakes. I'm not asking this to dwell on these, but maybe you could pick out one or two of them or some of these. The idea is that we learn from our mistakes. So if we sit here and talk about the good things… What are a couple of mistakes that we've made, the industry has made, even Texas Instruments has made in the past that we could learn from?

KILBY: I think they've all made mistakes, Don. When Bell had their Transistor Symposium in 1952, there were 24 American licensees and four Europeans present. I can't find a list of the 24, but I think that there are probably only two or three that are still in the semiconductor business in any form. Actually, the Europeans did better than the Americans. There were four of those, and certainly two and maybe three are still in business. It's been a rough ride for those companies. The technology changes have been upsetting. From 1952-60, we went through half a dozen completely different types of transistors - point contact, grown junction, alloy, surface barrier, micro-alloy, mesa, planar - and no two of those had a common fabrication process. It was a very fortunate company that managed to stay on top of that. 

MEYER: So there were a lot of patents written in those days for a lot of different processes, I guess.

KILBY: Yes. All of those devices were produced at some point. It's hard to remember, but at one time the surface barrier micro-alloy was the best transistor in the world. It probably stayed that way for five years. 

MEYER: This idea - let me just dwell a little bit on the patent. There were other people, obviously, as you were saying, pursuing this concept of monolithic integration. Basically, what, the people at Fairchild were? Wasn't it Fairchild?

KILBY: Yes, Fairchild was involved.

MEYER: And I don't know, was General Electric still going then? Here's an example where they had a tremendous program going and they shut it off, General Electric did.

KILBY: Well, General Electric was one of the early semiconductor makers. At one time they probably made more alloy transistors than anybody else in the world. But they had some choices to make as to where to put their efforts. Rather early, they decided they didn't want to be in the computing business, so they got out of that. I'm sure that had implications on their semiconductor decision. They've been amazingly successful without semiconductors, so I don't think you can criticize that.

MEYER: [Laughs] Okay. You mentioned computers. Wasn't this sort of the driving force of the ability to computer, to replace those-- 

KILBY: The computer was the ideal application for circuits, because the circuits that are required were highly repetitive, and the design procedures were sloppy enough so that there were some margins that could be used in semiconductor making.

MEYER: I think one of the reasons that we do have the computing power and so forth that we do today is because people like you, not realizing what was going to happen, were just looking for new answers to not necessarily resolve problems, but you were looking for new answers and you weren't really concerned about problems. In other words, there was no driving force to replace necessarily the vacuum tube in a computer. Or is that not true?

KILBY: Oh, there was a tremendous driving force to get vacuum tubes out of computers. They burned out, and the early vacuum tube computers had 5000-10,000 tubes. You couldn't run long programs because a tube would burn out before the run was complete. So there was tremendous motivation in that. 

MEYER: And people envisioned from early on that the transistor would do this?

KILBY: It was a hope, at any rate. I think it probably provided part of the incentive to work on it.

MEYER: Was that an incentive for Texas Instruments, as far as getting in the semiconductor industry for seismographic computers? Was that their motivation? Do you recall?

KILBY: When Dr. Kelly began the R&D program at Bell that turned into the transistor, his primary interest was in switching. Today switching and computers are synonymous. I don't think they were at that time. But that was his primary motivation. The military, of course, always had some interest in getting rid of vacuum tubes. 

MEYER: Again, I'm trying to find out your understand of how TI got into semiconductors. Wasn't it because they wanted to replace the vacuum tubes in their seismographic equipment?

KILBY: I think as much as anything - although I'm sure Haggerty saw many more applications than the seismic stuff. I think he felt that that was a way to broaden TI and to bring it into new areas. 

MEYER: Going back then, when you left Central Lab, did you have choices of other companies besides Texas Instruments.

KILBY: Yes. At that time, I only talked to three. I talked to IBM, TI, Motorola. I don't think IBM had much interest because they were pretty much committed to the SLT project by that time. Motorola did offer me a job and I could have gone there.

MEYER: So you chose Texas Instruments, which in many respects was a good choice because they were ready for you, I think.

KILBY: Yes, more so I think than Motorola would have been. 

MEYER: Is there anything you would like to say or to put on the record that might be of value to historians or anyone else in future years regarding the role you played in the microelectronics industry, or anything you see about it that someone 50 years from now would find interesting?

KILBY: It's easy to look for a few individuals that have had a significant influence. The bigger story in going from this (points to a 1inch wafer) to this (points to an 8 inch wafer) is probably the tens of thousands or hundreds of thousands of the world's most capable engineers that have worked on improving it. That's a continuing process and they're really the people who deserve the credit for that kind of thing. 

MEYER: We can look forward to much more from this industry. I'm sure you agree with that, don't you?

KILBY: Sure. On the other hand, the time will come when this approach begins to level off and something will replace it. I don't know what. But that kind of thing is inevitable. 

MEYER: Something, perhaps, in the field of biotechnology or-- 

KILBY: Oh, yes. Biotechnology of some form or optical electronics - almost anything.

MEYER: But this has had some surprising - I shouldn't say surprising, but unanticipated - results in communications. We certainly-- 

KILBY: Well, it's had a good long run. It looks like it's not quite over. 

MEYER: I mean, the idea of the Dick Tracy watch on everybody's wrist - this certainly is almost upon us. 

KILBY: No, there are not many things that you could imagine you might want to do with it or are truly impossible. 

MEYER: I guess to bring this discussion to a close, we of the American Vacuum Society have become very much a part of the semiconductor industry - our members, the sessions we have, the groups we have. We've done our part, I believe, in bringing together scientists and engineers who have been working within the semiconductor industry. I think it's very fitting that we have you here this year as our plenary speaker. I'm looking forward to your talk in a few minutes. On behalf of the History Committee, one of these days we plan on making an effort to develop some type of permanent display, some type of permanent documentation and monograph on the application of vacuum science and technology to the semiconductor industry. So I'm hoping that in future months that we can call upon you again maybe for some, not so much directive, but maybe for some archival material that you may have or may be able to direct us towards. Because we don't want to forget about the contribution that people like you played in the early days, or for that matter, you mentioned the tens of thousands of others. You're very correct there, and many of them have been AVS members. 

Jack, I want to thank you.

KILBY: Thank you very much, Don. I've enjoyed being here, and I wish you good luck with your history project.

Notes:
1. US Army Signal Corps
2. Aquadag was a water-based carbon-black paint used in labs for many purposes. 
3. Pat Haggerty was C.E.O of Texas Instruments.

return to top