Oral-History:Jack Kilby

From ETHW

About Jack St. Clair Kilby

Kilby.jpg

Jack Kilby is best known as the co-inventor of the integrated circuit. He conceived and built the first integrated circuit at Texas Instruments in 1958, simultaneously with Robert Noyce's independent integrated circuit work at Fairchild in California. Kilby then built the first computer using integrated circuits at Texas Instruments in 1961. Six years later he and two co-workers invented the first pocket calculator to show how useful integrated circuits would be in daily life, not just powerful government or military applications. Kilby held more than 60 patents on various inventions. In 1982 Kilby was elected to the National Inventors Hall of Fame. He received the National Medal of Science in 1970, the National Medal of Technology in 1990, and the Nobel Prize in Physics in 2000. Kilby’s work and contributions inspired the IEEE to name one of their most prestigious awards after Kilby. The award is given to those who, like Kilby, make outstanding contributions to the field.

Kilby was born in 1923 in Missouri. His father was an executive with the Kansas Power Company. Kilby traveled with his father during vacations and learned that cost was an important variable in engineering solutions, a lesson he kept with him all his life. By the time Kilby left Texas Instruments in 1970 to become an independent inventor, he was widely recognized for his engineering knowledge and creative inventions. In 1978 he became a professor at Texas A & M, retiring in 1984. He died in 2005.

In the interview, he focuses on the conception and early production of the integrated circuit, revealing that he invented it largely in the space of two weeks — while most of the rest of Texas Instruments was on vacation.

About the Interview

JACK ST. CLAIR KILBY: An Interview Conducted by Michael Wolff, IEEE History Center, 2 December 1975

Interview #455 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc. and Rutgers, The State University of New Jersey

Copyright Statement

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.

Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, IEEE History Center, 445 Hoes Lane, Piscataway, NJ 08854 USA or ieee-history@ieee.org. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.

It is recommended that this oral history be cited as follows:

Jack Kilby, an oral history conducted in 1975 by Michael Wolff, IEEE History Center, Piscataway, NJ, USA.

Interview

Interview: Jack St. Clair Kilby

Interviewer: Michael Wolff

Date: 2 December 1975

Place: Dallas, TX

[Interview originally done for "The Genesis of the Integrated Circuit", IEEE Spectrum (August 1976)]

Early Attempts at Component Integration

Wolff:

I am interviewing Jack Kilby about the invention of the integrated circuit.

Kilby:

Let me see if I can set the stage for this thing a little bit. The whole electronics industry, particularly in the vacuum tube days, was very much one of assembly of bits and pieces. There were some early efforts made to automate or simplify the component fabrication process, or form components in situ at any rate. For instance, there was a large British project run by a guy named Sargrove in which they were going to produce rather integrated, very low cost radio receivers. They were going to swamp India with them.

Wolff:

Was Sargrove with a big company?

Kilby:

He had government backing. I am sure there was some company relationship, but I don't think it was a big one. Of course that would have been vacuum tube work. That project never really got off the ground. I think it started before the war and ran in through the '50s. During World War II there began to be some efforts particularly for proximity fuses.

Wolff:

Sargrove was before World War II?

Kilby:

He started before World War II. I think he stopped during the war. There were a lot of components to get in there, but they had to be small. Under the sponsorship of the Bureau of Standards, in a group which has since become Diamond Ordinance Fuse Labs, work was started on silkscreen techniques for the deposition of conductors and resistors. Capacitors were added in chip form, vacuum tubes were soldered in place and things of that sort. That work was continued after the war at a company called Central Lab, which is still very much in existence. It is still making products somewhat of that type. They did not have a good way to cope with the active device thing, but those were probably the earliest attempts at miniaturization and that kind of thing of which I am aware.

The printed circuit board came into being around the early '50s. At the time it was certainly considered a major step in that direction. In the early 1950s people began to be concerned about what Jack Morton at Bell Labs in particular called the tyranny of numbers. The kinds of systems they wanted to build were so complex that they had two worries. One that it would cost too much to ever get them put together, and second they feared they would not be reliable enough to ever run continuously. For the telephone system or large computers, for instance, this was a very real fear.

Miniaturization and Molecular Electronics

Kilby:

After the transistor came on the scene there came to be a renewed interest in what at the time was usually called miniaturization. That was not a very descriptive phrase. The primary intent was never just to make it small. What miniaturization meant to most people at that time was a way to automate the interconnections in forms, with at least most of the components in situ and make structures with some degree of integration. Somehow all of this was lumped in the term "miniaturization," primarily because nobody had a better phrase.

A couple of things happened. After the war there was an urge to make use of some of this proximity fuse technology, and the Navy started a program called Tinker Toy. This was an arrangement in which basically they were going to repackage all of the components, put everything on a wafer about an inch square, and then stack up the wafers and connect them with vertical riser wire. You can probably find pictures of these around. In spite of the fact that they spent a fair amount of money on that, it did not get off the ground. This was in part because the timing was a little awkward. It didn't really get moving until the transistor was visible. It was also too big. If you are really only going to put one component per wafer, the wafers should not be an inch square.

Around 1956-57 the Signal Corps looked at that and decided to revise and update it and get it to accommodate the transistor. In the process they scaled it down so that the wafers that had been an inch square became maybe 3/8" square. This was a major program at the time. It is my understanding that something like $25 million was committed it. It was deployed by RCA with the intent that it could draw in most of the electronics industry eventually. This was to become a universal technique for improvement in low-cost electronics. This probably got started in '56 or so, and was very much in evidence between '57 and '58.

A group at the Air Force around that time decided that was not really the way to go and had begun to consider what they called molecular electronics. This did not become visible. I really was not aware of the program until around '59, but there is no question that the work was started and underway. Their concept was quite different. They thought they ought to forget about the existing components and invent new devices or new ways to perform circuit functions. There was some existent theorem they had drawn up. I think their best example at the time was, "It's hard to make conductors, and it's hard to make precision capacitors, so let's look at a quartz crystal which performs the same function but does not have any turns or foils — no plates and a capacitor isn't visible either."

Bell Labs designed an integrated structure which was basically a chip register somewhere along in there, and that was used. It was their premise that things like resistors were wasteful and all they did was dissipate power, and to conserve power they wouldn't use any resistors. There weren't any in a quartz crystal. This was a pretty reasonable idea in some respects. The trouble was that they put it on a very short timescale. They got what was for the time very massive amounts of funding. Two to four million dollars were committed to the program over a period of a couple years. Since each new deal required a rather basic invention, and since there was no process that synthesized these things, they couldn't relate it to anything, and the progress on that program was rather slow. What they really proposed to do was to invent a brand new device or structure for every required application.

Wolff:

Oh, so by deal you meant an application.

Kilby:

That's right. It would have taken literally thousands to cover the whole electronics sector. As I say, this didn't really begin to become visible until — or least I was not aware of it — until late 1959. Then of course, since they proposed to use semiconductors in their basic technique, we tried to interest them in what we now call monolithic integrated circuits. There was a serious question in their mind as to whether this was really molecular electronics or not. We had the analogy with existing circuitry and we had resistors. We could see them, and capacitors and conventional components. This was our major approach. The Navy didn't want to agree with either of the other two services and wanted to preserve their independence. While they didn't quite know what they wanted, what little funding they had was spent on thin film circuits.

What I'm getting at with this is that — certainly by the beginning of 1958 and maybe before — that there was an awareness that there needed to be some changes in component manufacturing technology. Lower cost is always a valid objective and that was a part of it, as well as higher reliability, better performance and smaller size and weight. All of these things were beginning to seem burdensome. The military was the largest single user.

Wolff:

The military. Okay.

Kilby:

Yes. They were the largest single user in the business of funding programs to do something about their problems. Each of the services had their own program.

Central Lab: Silk Screening & Transistors

Kilby:

Until 1958 I was working in Central Lab.

Wolff:

When did you join Central Lab?

Kilby:

In 1947.

Wolff:

Let's get a little of your background there.

Kilby:

I spent two years before the war at the University of Illinois. I was in the Army for thirty months and came back and finished school at Illinois between '45 and '47.

Wolff:

What was your major?

Kilby:

Double E. At the time, double E was primarily power sources. I had considerably more power courses than electronics courses. It was a time when they were really having a transition between the two. I graduated in 1947 and started to work with Central Lab. I was working on their silkscreen circuit.

Wolff:

Did they call you a junior engineer?

Kilby:

When I started I had a title that was low enough that I don't remember it. I'm sure it was not any better than junior engineer. After the transistor was announced, Central Lab was interested and became one of the twenty-five licensees.

Wolff:

Why did they put you on silk screening? Did you start to work there before the transistor was announced?

Kilby:

At that time they were making two classes of things. They were making amplifiers for hearing aids and still doing proximity fuse work and things of that nature that were very directly descended from the National Bureau of Standards' effort during the war. There is an argument within Central Lab and NBS as to who really started that silkscreen component work. It was a small but I guess viable activity at that time. Then Central Lab began to use this for parts for radio and TV and things of that sort. There, instead of screening on an insulator, they used high dielectric constant capacitor bodies.

Wolff:

Capacitor what?

Kilby:

Capacitor ceramics. They made things called RC networks. Some of these had a half-dozen to a dozen resistors and capacitors. The most successful of these was a vertical integrator for TV sets, which is a three-stage RC filter. This was and still is a business activity, though not a very big one.

Wolff:

Are they still a company all by themselves?

Kilby:

Yes. They have always been part of Globe Union [Inc.]. Globe Union basically makes all the storage batteries for Sears, and they have electronic components activity, which is at the moment all passive components: capacitors, volume control switches, things of that sort.

Wolff:

They started you off in silk screening work.

Kilby:

That's right. At the time they were one of the few companies in the country, and therefore by definition a leader, if you want to think of it that way, involved in what we would call circuit integration techniques in a broad sense. In 1952 they became a transistor licensee at Bell. I was one of two from there who went to Bell Labs. We went through their program for licensees, in which they showed us how to build transistors. Then I went back and started working on simple transistors, basically on cased units that could be attached to the thick film substrate. We set up a small production facility and sold a fair number of these for hearing aids and half a dozen other applications.

Wolff:

In what position were you by then? Were you a section head?

Kilby:

Yes, I was in charge of this transistor project. We made the transistors and attached them to the substrates and things of that sort. This was a reasonably promising approach. I think that the IBM SLP is at least an indirect descendant of that work. Basically their micromodule program was based on it.

Wolff:

How did you get from there to the integrated circuit (IC) then?

Kilby:

I decided to leave Central Lab early in 1958 and began to look around.

Wolff:

What did you do in those six years from 1952 to 1958? Were you were working on this transistor?

Kilby:

That's right. It took us a fair part of that time to learn how to do it. We had a very small group. Even at the end there were never more than three or four engineers on the thing. There were thirty or forty girls doing the assembly, at the end.

Wolff:

Was that why you left?

Kilby:

The transistor industry was beginning to take shape by this time, and it was clear that Central Lab was not in a position to keep up or to spend enough money to maintain any kind of position in it. I spent most of my six years there with a couple other guys trying to learn how to make alloys remain in transistors. We had done that pretty well by 1956, and by 1958 it was pretty clear that alloyed germanium was not really going to go that far. Central Lab was not in a position to fund the switch to diffuse devices or silicon or anything of that sort. I don't want to suggest that there are any hard feelings. I worked under the chief engineer, who at that time was R. L. (Bob) Wolff. He was extremely capable and a very good mentor.

Wolff:

Is he still there?

Kilby:

He retired a few years ago. I think he was perhaps the best engineer I have worked for at any time.

Wolff:

If I wanted to give him a call to talk about those days, do you know where I could find him?

Kilby:

He's still in Milwaukee and he lives on North Links Circle or Links Way.

Arrival at Texas Instruments

Wolff:

Okay. You came to Texas Instruments (TI) in 1958.

Kilby:

Yes.

Wolff:

How did you connect with them?

Kilby:

It was pretty much pure chance. I sent out about a dozen letters describing the kind of work I had been doing and suggesting that I thought miniaturization was a desirable area. As a result of that I had interviews with IBM, Motorola and TI. There may have been one or two others. By that time IBM was well launched on SLP, and I don't really think I got an offer from them. Motorola was interested, and thought I could work on that sort of miniaturization on a halftime basis and do some other things on the side. TI did not have any reservations. They thought it was fine and said, "Come on and let's start. We're going with it." I really didn't feel that I had a better option.

Wolff:

Under whom did you work there?

Kilby:

Willis Adcock hired me. I reported to him at that time, and off and on after that.

Wolff:

Isn't he still there?

Kilby:

Yes. Willis was the R&D manager for the Semiconductor Division. I'd like to go back and pick up on some other threads. Although I did not know it until very much later, sometime in the mid-1950s Willis and [Pat] Haggerty had talked about the desirability of doing more with semiconductors. Frankly I don't really know what form those conversations took or exactly what more they might have had in mind. At any rate, as a result there was an interest and awareness that it might be possible to do something there — which I am sure is one reason why they were receptive when I came along. I started there in May 1958. At that time their micromodule program was very much in evidence. TI had an interest in following that. At any rate, I began to look around for a potential area at TI.

Wolff:

Excuse me. You began to look around?

Kilby:

Yes.

Wolff:

Did you have a specific assignment?

Kilby:

It was not specific really. What I was told to do was to see if we could find some areas in which some of these techniques could be used, without being very specific. At that time radio was still a significant part of the electronics business, and I began to look at the possibility of making it an IF strip. I proposed to do this by making all of the components in tubular form. I think that it was an easier and preferable technique to the flat wafers in the micromodule deal. I spent a few months on that and built up a couple of models that worked. In looking at this, it became apparent that the labor costs, particularly within a semiconductor company, were much higher than those at Central Lab. We really couldn't afford very much hand assembly of anything. Clearly it would have been a fairly marginal or even losing product for them, so it wouldn't work.

Conception of Integrated Circuit

Kilby:


Audio File
MP3 Audio
(455_-_kilby_-_clip_1.mp3)

At about that time TI had a mass vacation policy. That is, they shut down tight during the first two weeks of July and everybody who had any vacation time coming took it. Since I had just started a couple of months before, I had no vacation time and was left in a pretty much deserted plant that had relatively low activity during this mass vacation period. I began to think about the message from the lessons with this IF strip. It became clear that were some things that semiconductor houses could do very well, that they had some very potent techniques and that this IF strip did not make very good use of them. Therefore I began to cast around for alternatives. The monolithic concept really occurred during that two-week vacation period. I had it pretty well written up and by the time that Willis got back at the end of the vacation period I was able to show him some sketches. I had outlined the idea pretty well and we added process sequence on how one might go about building them.

Wolff:

Do you recall how the concept occurred to you, or when or what you were doing, or a more fine-grained recollection than just that it was during that two weeks?

Kilby:

No. I have read some on the history of inventions and things of that sort, and as I understand the process, basically what one does is spend some period of time — I guess going back to one's childhood — such as watching the water coming over a waterfall while accumulating random facts and thoughts and things of that sort. This all goes into a pot in one's mind and every once in a while you fish down in there and grab some kind of random collection of these and bring them up and look at them and throw them back. You keep at that until something comes up that makes some sense. Frequently people talked about waking up in the middle of the night with a complete solution, or they talked about things like stepping off the train, and as his foot touches the ground this all flashes into his mind in a complete picture. I don’t think this one really worked that way. I think the actual period of time involved during which this unfolded to me was a couple of days. However I don't think that I could say that there was any one instant before which I didn't know about it and after which I did, either.

Wolff:

Do you remember what started the thought about the concept? What was the beginning?

Kilby:

That goes back to these things that we talked about. It was kind of an analysis of the capabilities or a look at the kinds of things that semiconductor people could do and could do well. There were things like data definition, and at that time the photographic techniques were just beginning to come in which were much more powerful than anything we had with silkscreen circuitry. On the other hand, the hand operation, hand labor and things of that sort were incredibly expensive. This was an attempt to find a set of techniques that matched those capabilities.

Wolff:

A naïve assumption might be that you thought of what you wanted. If I understand correctly, you were thinking of technologies for fabrication and wondering how to put them together. Another way to do it might have been to have thought about or to have had the inspiration for the idea of diffusing all the components into a single chip of silicon — which is what it ended up being, right? You didn't think of that?

Kilby:

Let's talk about what I thought at that time. No, the techniques available were diffusion, oxide masking and good photographic processes. The important thing was to use these on a flat piece of silicon or basically on one side of the silicon.

Wolff:

In other words, you recognized early that was something you wanted to do.

Kilby:

Yes. There is no doubt about that. I began trying to think of how various components could be formed and be interconnected as integral part of the manufacturing process using silicon as the material. This was really the first part of it. During those first passes I tended to feel we would want to use both bulk resistors and diffuse resistors, and both of these were included in that original set of sketches. As it turned out the diffuse resistor have been displaced by bulk resistors almost completely, but diffuse resistors were a part of the original package.

Wolff:

Do these sketches still exist, and would it be possible to get hold of them?

Kilby:

Some of them. Actually the front page of the patent represents a start in that.

Wolff:

That is something that is more refined and came later, right?

Kilby:

Not much. Actually it changed very little through that period.

Wolff:

The sketches you showed to Willis Adcock looked very much like what was on the front page?

Kilby:

Yes. There were others in which we went to specific configurations. I think this one with the phase shift oscillator was very early.

Wolff:

Which one?

Kilby:

Both Figure 8a and Figure 8b.

Wolff:

Figure 8 of the patent was in the original sketches you showed to Adcock?

Kilby:

Yes.

Wolff:

It might be nice to get hold of those simply because they were earlier. Does he have them? If I wanted to get hold of them, where would I look?

Kilby:

I think they still exist. I don't know whether you can do much with them. I am not sure that TI would want to release them, but I'll check on it.

Wolff:

I could call up whoever you think is the appropriate person.

Kilby:

I will check on it.

Wolff:

Okay. Describe to me what these sketches you gave to Willis Adcock revealed. You said they showed diffused and discrete resistors.

Kilby:

As I said, this started with the realization that all kinds of components could be made from a single material. A semiconductor such as germanium or silicon has to be used. Figures 1 and 5 of the patent represent a set of those building blocks. Although this happens to shows them as individual or discrete elements, I think it's clear from the specification that it was never intended that they be used that way — although the oxide capacitor was a new component that had never been made. We filed claims on that as a discrete device, but the rest were all considered for use as a part of an integrated structure.

In the simplest form, Figure 1 represents a bulk resistor. That is just a section of the wafer with only contacts on the end of it. Figure 2 is the diffused junction which is used as a capacitor. Figure 3 is really a distributed RC network in which the bulk resistor has a capacitor junction along it so that this would really be represented by an infinite-section filter, which is a fairly useful gadget. Figure 1a is a diffused resistor. Figure 2a is a capacitor with an oxide dielectric. Figure 5 is a mesa transistor structure that was common during that period. Figure 5a shows a spiral for inductance.

Wolff:

You thought through all those components and ways of putting them onto a single chip in those two weeks.

Kilby:

Right.

Motivations for IC Concept

Wolff:

Did you start out with the desire to put the components on a flat piece of semiconductor?

Kilby:

What I started with was a desire to use a batch fabrication process; that is, to make many components simultaneously and make them in place so that they are connected.

Wolff:

That was the problem you set for yourself.

Kilby:

Yes. The easy way to do a batch fabrication deal is to have them all on one surface so that one set of imaging techniques and things of that sort can be used. The urge for batch fabrication was probably the biggest single thought. In other words, "How can batch fabrication of components be done using semiconductor techniques?"

Wolff:

That gets back to the labor costs.

Kilby:

Right.

Wolff:

That was your starting point. Then you started thinking about the availability of photographic techniques.

Kilby:

Yes, although at that time we were thinking of somewhat bigger structures, and even silkscreen techniques might have been a conceivable way to do it and I think were probably mentioned in there.

Wolff:

I think Jean Hoerni was the inventor of the planar transistor.

Kilby:

Right.

Wolff:

Did that precede this?

Kilby:

No. Well, I don't really know those dates. You would have to get those from Gordon or someone. After your request for early photographs I looked and didn't find much, but you might find this interesting. This was done in '68. It is probably a pretty good description of the integrated circuits development.

Wolff:

That's very helpful.

Kilby:

It starts on page 64. This also covers Hoerni and the planar stuff.

Wolff:

Can I borrow that?

Kilby:

Yes. Two of these are extra copies you can have. I have a couple others that you might want to page through.

Wolff:

That will be very helpful. Do you recall being aware of the planar work?

Kilby:

No, I was not and in general TI was not. That work was going on pretty much simultaneously with this. If I had been aware of it we sure as hell would have had a picture of it in there, because it could have been added very easily.

Use of Existing Semiconductor Technology

Wolff:

What were the technologies that you thought of incorporating?

Kilby:

The virtue of this whole scheme and its real intent was to draw on the existing semiconductor technology — to start with, purifying the material, crystal growing and things like sawing and diffusion. Epitaxy did not come until later.

Wolff:

I was wondering about that. The main one was diffusion then.

Kilby:

And crystal growth and pattern formation technique. We proved that many of the patterns existed at that time. It was because this thing grew on the mainstream semiconductor technology that it was able to move fairly fast once it got started. It did not require a complete new set of inventions in terms of processes and things of that sort. Again, because it was intended that way, it was able to take advantage of these other improvements in technology that came along later — like epitaxy and planar technique.

Wolff:

They are not synonymous, are they? They are two different things?

Kilby:

Yes, they are quite different.

Wolff:

That's a very important point: that it drew on the technologies that already existed.

Kilby:

It was intended to do more than that. I am sure it was never expressed at that time, but because it was using mainstream semiconductor technology it would turn out later that it would be able to profit by all of the improvements in that area. Epitaxy is a good example. The epitaxial work then was certainly not done because anybody thought it would help integrated circuits. It was done because it was a good way to get better transistors. Yet it became a very important improvement.

Development of Prototypes

Wolff:

What was Adcock's reaction when you showed him these sketches?

Kilby:

He was extremely interested. I think that it clearly fitted his idea of what a good scheme ought to look like. He rather quickly presented it to [Mark] Shepherd and [Pat] Haggerty and they were enthused too. I did not have a major selling job to do. The stories you hear about people practically giving their life to sell an idea certainly did not happen here. It was enthusiastically received right from the start.

Wolff:

This invention was what you showed to him, and it was invented in two weeks.

Kilby:

Yes. On the other hand certainly what I showed him at that time was all paper, sketches and talk. I am not sure it even was all on the paper at that point.

Wolff:

What came next? I realize there was a period between the paper and the first prototype.

Kilby:

Yes. This meeting with Adcock when he got back from vacation was in July. The first thing he wanted was some evidence that all parts could really be made with silicon as the semiconductor. I really don't quite understand even now why he was skeptical about that, but at any rate he was. Therefore I built up a group of discrete elements that looked much like the pictures. That is, I made some silicon bars and used these for resistors, and we got some silicon transistors and cut up some other wafers and made capacitors from them and put these together and made a flip-flop with all silicon components but not integrated. That was done in August of that year, I suspect.

Wolff:

Was there ever any serious thought of doing this in germanium? Was it just assumed that you would do it in silicon?

Kilby:

It was obvious from the start that silicon was the best material with which to work, in part because the first customers for this — the ones who could gain most from it — were the military, and they needed silicon. In addition, TI's emphasis was on silicon and had been for a very long time. Therefore silicon was always an integral part. This first gadget or component, which was not an integrated circuit, was done in all silicon. We began to look around in order to get something else going that was integrated.

At that time TI did not have very much in the way of diffused silicon transistors. They were very big on grown junctions, and they were a little late because they had some diffused silicon power transistors and things of that sort, but there really were not any transistor wafers available. They did have diffused germanium wafers, so I got some wafers in which transistors had been evaporated or formed. I'm sorry — diffused and the contacts evaporated. Then we masked those not even using photographic stuff, but a girl dropped wax on it. We built two circuits.

Wolff:

You said a girl dropped wax on it?

Kilby:

Rather than defining these things photographically.

Wolff:

Defining what things?

Kilby:

The transistor mesa and this capacitor mesa. You see, what I did was start with a wafer about a half-inch square in which transistors had been fabricated all over it.

Wolff:

This is Figure 8 at which we are looking.

Kilby:

Right. I had a girl mask with wax to form the mesa of one those transistors and another patch, which became the capacitor junction. Then we mesa-etched it and cut it into strips. That configuration was really one of the first two circuits built.

Wolff:

Is this the flip-flop?

Kilby:

No, that's a phase shift oscillator. There are better copies of this around, but this is a photograph of that configuration.

Wolff:

This is on the cover of EE Magazine.

Kilby:

Yes.

Wolff:

There is a photo in the 1968 issue of the circuit you have just been describing.

Kilby:

That's the circuit of Figure 8 in the patent and one of the first two that was built. Also, using comparably crude techniques we built a flip-flop circuit. Again we were working with these existing wafers, because they already had the diffusions in them. Those were the first circuits, truly integrated circuits that were made and they were germanium, purely for convenience.

Wolff:

Were both circuits in germanium?

Kilby:

Yes. There were no available small-signal diffused transistor structures around TI at that time. By the next year there were plenty of them when we began the chip.

Wolff:

Okay. Was this germanium circuit you put together the prototype that demonstrated the feasibility of your invention?

Kilby:

Yes, that's right. This was the first integrated version of it.

Wolff:

Roughly when did you finish these two germanium circuits?

Kilby:

It was certainly in the fall. I would guess it was sometime in September.

Early Circuits and Public Announcement

Wolff:

I see. What was the next phase after that? Did it become a matter of making them practical?

Kilby:

These were really both just feasibility demonstrations and attempts to show that the idea worked. The phase shift oscillator was not designed for any particular requirement and, as a matter of fact, probably wasn't stable enough to have filled any had they existed. After that was over we began to look for specific useful circuits wanted to build and began to make the shift to silicon. Those two things were carried out concurrently.

Wolff:

That took the better part of the next year, didn't it?

Kilby:

Yes. In '59 at the IEEE show, which would have been in March, we made the first public announcement of this thing. There was a press conference in New York. I think Shepherd went through this and described it. That was the first public announcement. Of course that set the date at which the patent had to be filed, and we rather frantically got this on file in February.

Wolff:

Is that the announcement you have there?

Kilby:

No, that's the wrong one. This is a deal from 1960.

Wolff:

That was a little cigarette-sized thing?

Kilby:

No, this was our first attempt at a catalog circuit. This was a group of circuits that were custom designed for ARMA, and they were a factor in the thing.

Wolff:

That made it well established.

Kilby:

No, it wasn't well established.

Wolff:

In terms of the invention anyway in 1960.

Kilby:

Yes.

Wolff:

There was a little computer for the Air Force. That was not March '59, was it?

Kilby:

No. Those things were delivered in 1961. You'll want to check this, but it's my recollection that the IEEE [actually IRE] show in '59 was also where the first public announcement of the planar technology was made. It could have been '60, but I think it was '59.

Developing Commercial Circuits

Wolff:

You have not described any ups and downs or difficult moments. Can I take it that it was pretty much smooth sailing for you through conception and early development?

Kilby:

We spent a good deal of time working on it. Up to this point it was really the conception phase itself. That was not the period where the stumbling blocks usually occur. I had almost no trouble in making this pre-component version work and not too much trouble with the germanium circuits. After that it became quite difficult. It took a hell of a lot of time and effort, and there was a fair amount of pressure involved, in trying to get these first commercial circuits underway. I never heard anybody attribute any throes of agony to the conception part of the process. I guess if they get too great you just don't do it. I don't know.

Wolff:

You were heavily involved in getting the commercial circuits coming too.

Kilby:

Yes.

Wolff:

By then I guess you had several people working with you?

Kilby:

Yes. By then there were other people, in particular Jay Lathrop, who had been at the Bureau of Standards and done the original work on photographic techniques with semiconductors.

Wolff:

Was Gordon Teal involved in this at all?

Kilby:

No, not directly. Gordon was in charge of the central research activity at TI, and this work was done in the Component R&D Group of Research & Engineering.

Wolff:

I see. I think this pretty much covers the narrative I wanted to get. The person most closely involved with you up through the fabrication of the germanium circuits was Willis Adcock?

Kilby:

I reported to him and all of my contacts were there.

Wolff:

You did this pretty much alone.

Kilby:

Yes. It was individual effort at that point.

Wolff:

You answered the question to what extent you sought this deliberately. It was pretty much 100% deliberate. It wasn't just a lucky thing you stumbled upon.

Kilby:

There is no question that there was a particular solution for which I was looking. That is, the solution came about because of what I thought was a fairly specific problem. I think you could say that I sought it.

Wolff:

The decision to pursue this was primarily based on a perceived need. Often an invention occurs because there is something too promising to neglect.

Kilby:

It was not a solution looking for a problem. This was clearly the other way around.

Influences and Impacts

Wolff:

To what extent do you feel your work was aided or influenced by the work that was being done at other places like Bell Labs or Fairchild?

Kilby:

I have told you those programs of which I was aware. There was the RCA program, the work at Central Lab and things of that sort. I think those things, particularly the work at Central Lab, greatly assisted in defining the problem that we were trying to solve. It sensitized me to the need and practical requirements of a successful product in this area. That was invaluable. I sure did not get that in the three months I was at TI before this came along. That was clearly a part of the background. On the other hand, the choice of the semiconductor approach and things of that sort definitely occurred after I got to TI.

Wolff:

In terms of your personal motivation and reward, what gave you the greatest satisfaction out of all this?

Kilby:

Collectively the whole experience was very satisfying. It was a nice idea, it worked, and I was able to be rather closely associated with it from the start and stayed with it until it was a fairly large and successful industry. I got a good deal of reward and recognition for it. I would find it hard to pick out any single thing.

Wolff:

Back in 1959, did you think it would be as big a thing as it has turned out to be? Did you envision the tremendous applications for it?

Kilby:

Probably not. On the other hand, in '61, which was not all that much later, we did some work and made some projections that Haggerty used that forecast through 1970 or so. They weren't too bad. In '59 I certainly thought it looked like an attractive solution and the best one I knew, but it would have been foolish to try to put numbers on it at that point — or at least I didn't.

Childhood and Education

Wolff:

Tell me a bit about yourself as a boy. Were you always determined to study electrical engineering or were you a mechanic or a chemist or anything like that?

Kilby:

No. I grew up in Kansas. My father was the head of a power company that was scattered over the western third of the state. He was an electrical engineer, and I sometimes I went around with him to visit the properties.

Wolff:

Did he own this?

Kilby:

No. He was the president, but it was a public company. I suppose that sensitized me to engineering at any rate. I don't have any idea why I first decided I wanted to be an engineer, but I don't remember being in any great doubts about it at any time either.

Wolff:

What did your mother do? Was she a housewife?

Kilby:

Yes.

Wolff:

Do you have any brothers or sisters?

Kilby:

I have one sister a couple of years younger than me.

Wolff:

Were you a ham radio operator or anything like that?

Kilby:

Yes, I did that. I got a license when I was in high school and was on the air for a couple of years.

Wolff:

Where did you attend high school?

Kilby:

Great Bend, Kansas.

Wolff:

Is that where you grew up?

Kilby:

Yes.

Wolff:

Is that a great place or a little place?

Kilby:

It's called that because the Arkansas River makes a major turn as it goes across there. It is where the Santa Fe Trail hits the river.

Wolff:

Was this a small town?

Kilby:

It was about 10,000 people I guess.

Wolff:

You went to the University of Illinois. Why Illinois?

Kilby:

I wanted to go to MIT. They insisted on a score of 500 on the math exam that they had at that time. My score was 497 and they said "no thank you" about three days before school started. Both of my folks had gone to the University of Illinois. The Kansas schools were not too impressive, so I went there.

Wolff:

Did they tell you that they were turning you down because your score was too low on the math exam?

Kilby:

Yes. Which was true. Our high school was short one year of math, and that had something to do with it. And I was not all that much of a math student anyway.

Wolff:

Do you have a family?

Kilby:

Yes, I have a wife and two daughters.

Bob Noyce

Wolff:

Did you know Bob Noyce in 1959?

Kilby:

No.

Wolff:

I guess you have gotten to know him since.

Kilby:

Yes. I'd say we are good friends.

Wolff:

What do you think the two of you share in common? Does anything strike you?

Kilby:

We were probably both in the right place at the right time. It would be hard go much beyond that. To some extent I think we both helped make that place.

Role of Military Market

Wolff:

When you got this enthusiastic response and the program started to move, do you feel that the company's interest was primarily in the military market? Did they see this mainly as a military program or were they equally interested in potential commercial use?

Kilby:

During that period, and to a lesser extent even today, if you have a new technology or product and it is of interest to the military, or can be made to be of interest to the military, you tend to start in that area because there is funding available, their problems are pretty real and rather urgent, and it's a good place to launch a new product. That was true then and basically it is true now.

Wolff:

Your work was funded by the military, wasn't it?

Kilby:

Yes. Military funding was a fairly significant factor in keeping things moving and accelerating the work.

Wolff:

I guess that was one of the differences between the programs at TI and Fairchild. I understand theirs was internally funded.

Kilby:

Yes.

Patent Appeals and Decisions

Wolff:

Regarding this business about the patent appeals and decisions, would it be fair to say that you and Noyce are considered co-inventors in that the court finally gave him the credit for the metallization of the interconnection side of the integrated chip (IC) and that you are credited for the concept of putting devices together to perform this complex function?

Kilby:

Well, courts do not make that kind of a distinction. They decide questions that they are asked; that is, who was entitled to a claim that says it will do a specific thing. As such they are not a particularly good mechanism for apportioning credit. They have substantial significance when they make those decisions, but the basic issue was really whether gold sticks to silicon or silicon oxide.

Wolff:

If I want to summarize the end result of these patent appeals without going into it very much, how would I do that?

Kilby:

You will find it difficult to do in every phase. I wish you luck with it.

Wolff:

What would you say the court ended up ruling, without worrying about whether it was the truth of the matter?

Kilby:

I don't really recall it well enough, but if you'd like I'll check that out and give you a response.

Wolff:

I would appreciate that — if you would explain to me where it all ended up as far as the court is concerned — and I would also be interested in the extent to which you would take issue with that, of course.

Kilby:

The extent to which we take issue with it is substantial, but I'll give you what I can.

Wolff:

I'd appreciate it. What I was just spouting off to you is sort of my understanding of where it all ended up in terms of the court decision.

Kilby:

The reason that this was at issue is that this happened: Let's say that we would want to replace some of these wires with a metal evaporate and that we would put an insulator such as silicon oxide down and then evaporate a metal over it. What the court held was that that was not an adequate teaching. What I had said was we would use a metal such as gold. Had I said a metal such as aluminum I guess the whole goddamned issue would never have been raised. As it turned out gold does not stick very well to the thing, and therefore the feeling was that that was an insufficient teaching, which is a serious flaw.

Wolff:

That's interesting. I had not understood that.

Kilby:

That's why if you look at a transcript of that case you will see great quantities of testimony by some people saying gold does stick and some saying gold does not stick and things of that sort.

Wolff:

Gold was insufficient or sufficient?

Kilby:

I said that we would evaporate a metal over an insulating layer. I said we would use a "metal such as gold."

Wolff:

That's helpful. I understand that. I've been through all those papers and it would have been a lot easier if I had known this first.

Kilby:

Yes. There are great quantities of testimony in there. We made some circuits in which we had gold interconnect and they did not fall apart. All kinds of crap like that. Fairchild tried to duplicate them and theirs didn't stick. We knew ours would stick before we started and they knew theirs would not stick before they started, and that's the way it turned out. All of that I think had fairly little weight on the final decision, but I don't know.

Inevitability of IC

Wolff:

Bob Noyce was talking about the technological environment that existed at the time, and his feeling was that if the invention had not taken place by either of you, then soon someone else would have invented it. Do you agree that it was idea whose time had come?

Kilby:

Sure. I don't know whether it would have been three months or three years, but sooner or later I think it would have appeared. It may have been in a little different form or something of that sort.

Wolff:

I think the work you really got into would be characterized as industrial physics. It's not traditional engineering and certainly not power engineering.

Kilby:

Yes.

Industrial Physics

Wolff:

I was going to ask you why you went into industrial physics. I guess it was more a matter of just being drawn in by this component market.

Kilby:

That's part of the history of the semiconductor thing. Today there are well-established courses in school that train everyone to be semiconductor engineers, but when the transistor began to get off the ground they didn't exist. Therefore all kinds of disciplines were called into the thing — electrical engineers because they might know something about how to use the product, physicists, chemists, metallurgists, the whole gamut. The people who were involved in those days had to come from somewhere. Jim Early, who is one of the best engineers I know, has a background in paper mills.

Wolff:

I know Jim, but I didn't know that.

Kilby:

Yes. That's where his degree was — I guess maybe his B.S. There was an almost deliberate attempt to pull in people with different backgrounds. I don't think anyone deliberately tried to go as broad as papermaking, but certainly it was recognized that it was an interdisciplinary thing. All the people who were pulled into that, except maybe in the very large research institutions, got kind of forcibly broadened. The electrical engineers have learned something about chemistry, physics and metallurgy, and the metallurgists picked up the same things.

Master's Degree

Wolff:

Do you have a master's degree?

Kilby:

Yes, from Wisconsin. A kind of a mail order proposition.

Wolff:

When did you get that?

Kilby:

In 1950.

Wolff:

Is that in EE also?

Kilby:

Yes. Wisconsin began to offer graduate courses in Milwaukee on an extension basis. They were not really authorized to confer degrees on people who weren't residents of Madison, so they preserved the illusion that we were in residence by asking us to go over there and register. Shortly after that they established a full-fledged operation in Milwaukee.

Wolff:

I think this takes care of it at least for now. If I have any further questions, I can probably clarify with you on the phone. This has been extremely helpful. Thank you for the interview.