IEEE

Oral-History:Robert N. Noyce

SHARE |

From GHN

Revision as of 18:07, 5 January 2009 by Nbrewer (Talk | contribs)
Jump to: navigation, search

Contents

About Robert N. Noyce

Robert Noyce was born in Burlington, Iowa, as a son of a preacher. He studied math and physics at Grinnell College in Iowa and went to MIT to study physics. After getting his degree from MIT, he started working with Bill Shockley and later established Fairchild with several people from Shockley. He was a physicist at Fairchild when the concept of the integrated circuit was conceived.

In this interview, Noyce narrates the major events leading up to the development of the working model of the integrated circuit. Noyce also talks about his experiences at Shockley prior to his involvement in Fairchild. He refers to several people who made important contributions to the building of the IC. In the late 1950s and early 1960s, a number of companies were engaged in producing a viable model of the integrated circuit. Unlike other companies, such as Texas Instruments, Fairchild, at which Noyce worked as a physicist, did not receive support from the military and therefore could have more opportunities to directly enter the commercial market. As a tightly knit group, Fairchild provided an atmosphere in which members could experiment and build up important elements of the IC technology. At the end of the interview, Noyce briefly mentions the patent disputes involving Jack Kilby and states the contributions Kilby and he himself made to the development of the IC.


About the Interview

ROBERT N. NOYCE: An Interview Conducted by Michael F. Wolff, IEEE History Center, 19 September 1975

Interview #458 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, Rutgers - the State University, 39 Union Street, New Brunswick, NJ 08901-8538 USA. 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:
Robert Noyce, an oral history conducted in 1975 by Michael Wolff, IEEE History Center, Rutgers University, New Brunswick, NJ, USA.


Interview

INTERVIEW: Robert N. Noyce
INTERVIEWER: Michael F. Wolff
DATE: 19 September 1975

Thought Processes
Leading to Integrated Circuit

Noyce:

One of the things to talk about here would be the thinking process that we went through when the integrated circuit came along. Certainly the first major change was the batch fabrication processes that came along with diffused transistors. The diffused transistors were made hundreds at a time with one slice of silicon and then cut out into individual parts. I always characterized it at that time as taking the beautifully geometrically arranged things where we knew where everything was and cut them into little pieces. Then we had to have girls for them with tweezers and put headers for packages back on them. Then they would wire them all back together again.

Wolff:

Who, when and where were "we"?

Noyce:

The industry. The semiconductor industry would cut them apart and put wires on them and put packages around them and then we'd sell them to our customers and they would plug all of these packages back into a printed circuit board.

Fairchild’s First Transistors

Wolff:

What year are you talking about? Was this when you were at Shockley or Fairchild?

Noyce:

This really started at Fairchild. Although the first discussions of the diffused transistor as opposed to the alloy transistor or junction transistor – the first difused junction transistor – was in the period of 1955-56.


At Shockley we really were not worrying about transistors. The effort at Shockley was primarily in the four-layer diode as a new market area. One of the determinations at Fairchild was that we would not make diodes. There was not a market ready and existing for diodes, whereas there was a market for silicon transistors – and which was primarily a military market at that time. The technology then was applied to the silicon transistor and we were in that business in Fairchild. This is late '57 and going on from there. Fairchild introduced its first transistors in the summer of '58. These were double-diffused silicon transistors made by photoengraving processes.

Photoengraving

Noyce:

Some work was done in photoengraving at Philco when I was there. It had not been perfected, but the concept was beautiful. A design would be done once and one could reproduce very fine geometric patterns just by taking a picture. Certainly was one of the ideas in the back of our minds was, "Gee, that's a neat way to make fine structures." As we were trying to work out a technique for making double-diffused transistors at Fairchild, one of the areas I worked on was developing techniques for doing that. Indeed, we made very crude pictures and used label machines to step and repeat patterns. We then shot them down to the sizes we wanted and made the transistors using these masks. I think those were the first commercially available transistors that were made with photoengraving. I emphasize this because photoengraving is obviously a key element to the whole business. That kind of processing was also being worked on at Bell Labs by Jules Andrus. There is a patent in his name covering the photoengraved transistor. This work at Fairchild was pretty much independent of that. There was probably some information available at that time that they were working on that technique, but I was unaware of who or what.

The Military and
Molecular Engineering

Noyce:

At Fairchild we were working on batch transistors and double-diffused photoengraving. Now that should be looked at against the background of some other work that was being done. The military was interested in the so-called molecular engineering projects at that time, and there were a lot of projects that were funded by the Air Force. Westinghouse was doing some of that work, however the concepts they were using were probably wrong. The basic concepts that were being worked on assumed that we ought to be able to, sort of molecule-by-molecule or atom-by-atom, build up a structure that provides an electronic function. The strength of the electronics industry has always been to synthesize something out of very simple elements rather than trying to invent a complex element. One gets simple circuit elements like capacitors, resistors, amplifying elements, diodes or whatever and then synthesizes the electronic function desired out of those. I think is where the molecular engineering thing went wrong.

Motivation to
Synthesize Elements of the IC

Noyce:

The great example of an integrated device that was used there were things like p-n p-n diodes, which is equivalent of two transistors – but unfortunately not as useful as two transistors. Bell Labs claims that really the first integrated device was really a little stepping counter that was made out of pn pns, sort of like a gas tube. However the real key to being able to make some progress was to say, "Let's see if we can't devise the various elements needed to build a circuit and put them together." We had the two things going on here. There was that background of desire, and there had been a lot of discussion about, "Wouldn't it be nice to be to be able to make the whole thing in one piece instead of having to fabricate it out of a lot of different pieces." The motivation was there to find a way to do it. The next real key to the development of the integrated circuit—

Wolff:

Was photoengraving the first key?

Noyce:

The first was batch processing – making a lot of transistors at once. That could be done by evaporation through masks or other techniques, but photoengraving was a very convenient technique for it. The second process that was very important was the so-called planar transistor that Jean Hoerni invented. The motivation behind inventing the planar transistor was to get surface stability. The interface between the semiconductor and the outside world is critical in terms of how the minority carriers behave. It's an interface. It is the glue to the MOS transistors of today and it was the glue to all sorts of reliability problems of the transistor of that day. The major motivation was to get the silicon dioxide coating on the junction and stabilize that interface so that it would stay put – where it wouldn't drift, the surface potentials wouldn't change, and so on.

Again, I was not the inventor of the planar transistor, but I had worked on surface stability problems at Philco and it was a problem in which I had been quite interested. I think that the climate around Fairchild at that time was to immediately see the value of that for making high-reliability transistors. It solved one of the major problems with transistors in making stable devices. Therefore Fairchild made an immediate commitment to convert the entire production to that type of a technique. The fallout benefit here was that the surface of the silicon then had one of the best insulators known to man covering it. Holes could be etched through it to make contacts to the underlying silicon. At one point in that process, obviously, there were a whole bunch of transistors embedded in an insulating surface than could be wired together easily. There were a few other things that then had to be done. We had that background at Fairchild earlier than other places because the planar transistor was available at Fairchild and we were familiar with it and working with it.

The next thing to do was to cut them apart electrically – rather than cutting them apart physically – and add onto them the other components needed to make them into circuits and add the interconnection wiring. There were several techniques on which we worked. One was the technique that is used today, which was to put back-to-back diodes between any two transistors. Then no current can flow between the two in either direction. That is built into the silicon as well. It's all buried in the structure. There was another technique we worked on at Fairchild at that time, which was very similar to some of the air gap isolation work that was later publicized. We would etch through between them after the transistors have been glued together so the interconnect between them was preserved, but there was no physical connection between the pieces of floating little pieces of silicon. Jay Last worked on that and had some success at it.

Another technique we thought of using but did not do much work on was to use intrinsic silicon between the various transistors, which in itself essentially carries no current. It's an insulator. That silicon would be made intrinsic by introducing mid-band-level traps, which made it go to very high-resistance material. The technique that was finally decided upon of course was back-to-back diodes between the elements. The other element needed besides the amplifying element in the transistor was a resistor. It was relatively simple to introduce into the silicon – a diode-isolated piece of silicon that acts as a resistor. From there on out we would start building logic circuits because those can be made out of resistors and transistors. The isolation techniques originally used a triple-diffused structure. In other words, we took an N-type slice of silicon and diffused it in a P-type collector and then diffused in an N-type base and then diffused in a P-type emitter for the inverse polarity. That gave a series of isolated and fully functional transistors. Three contacts had to be made to each. Later the technique was revised to use an epitaxial layer instead of the triple-diffused for that first isolation junction, which worked very well. The problem with the triple-diffused was that the collector resistance was high. That could be avoided by using a structure built up with epitaxial growth. It would diffuse in a heavy high-conductivity collector region and then grow a lightly-doped base region over the top or something like that. It was motivated by miniaturization cost, sort of a gee-whiz Dick Tracy type of thing. It was definitely sought out because the need had been established before. I think that's part of the whole scheme of invention. One has to be in the middle of the milieu and have the motivation for using what one knows about that milieu to come up with an answer to the problem. Fairchild then immediately made quite a commitment to go into that.

First IC Demonstrations

Noyce:

I guess the first ICs were demonstrated at Wescon in 1960. It was sort of as a "gee-whiz, see what you can do" thing, and we started selling them almost immediately thereafter.

Wolff:

Was this the first IC at Fairchild that was demonstrated or was it the first IC in the world? TI had some kind of showing in 1958 or '59 for the Air Force.

Noyce:

I don't believe they actually showed it. I may be wrong. They may have written up the stuff that Jack Kilby was doing on cutting up bars and binding wires around and so forth and showing that to be an IC during that Air Force contract. However as far as I know the first public display of anything was in 1960. I think it was 1960 anyway. And Fairchild wasn't the only one doing it. Westinghouse had a good program. They were on what I considered to be the right track with a similar technique.

Wolff:

Did they also display an IC?

Noyce:

I believe so. It may have been a later show, but I think it was along about that same time. Westinghouse had a molecular engineering project that had been supported by the Air Force. Incidentally, the work done by Fairchild had no government support behind it. The military has often taken the credit for this invention, but I don't think there was any direct support there. They can take credit for it in that one of the primary motivations for getting miniaturization was the effort of the military to get miniature components.

Wolff:

That was one of your primary motivations at Fairchild.

Noyce:

Precisely. However there was no direct support of it from the military. There had been things like the Tinkertoy Project of standard modules to get high-packing density. There was a great deal of talk about packing density of electronic functions in the late '50s. It was the Missile Age, and transportation costs from here to Russia were very high and so on. I would say that it was sought deliberately in the sense that one was aware of the background need for it, but then it pops out of the progress of the technology. You asked whether it was primarily a marketing decision to pursue it. I think not. I think that most of the advances of this sort have not been predicted by the marketing people or deliberately sought by the marketing people. Rather, they have arisen from a technical development. It was a "Now it is possible to do this. Now we can do this, why don't you go sell it?" kind of thing rather than the marketing group coming in and saying, "If we had this we could sell it."

Motivations to Produce the IC

Wolff:

However you recognized that there was a need. It wasn't just that it was something technologically exciting.

Noyce:

Yes. I think the only thing that's technologically exciting is something for which there is a need. That's a question of personal motivation perhaps.

Wolff:

That's interesting. You were motivated and considered it exciting because you saw that something could be done with it.

Noyce:

Yes. I think that most industrial research has got that as the motivation behind it. There is nothing quite as discouraging to someone in industrial research than to come up with what he thinks is a great and wonderful thing and find that it has no commercial viability, because then his idea dies.

Wolff:

Do you recall whether your sense of need went beyond the obvious military applications?

Noyce:

Oh yes. It was clear that that was just a stepping-stone. In other words, it was clear in the beginning that this would be expensive to do. There was no easy way to compete in the television set or in the commercial computer with something expensive. On the other hand there was a military need that was willing to pay anything to cut down the size and weight of equipment. They had been going from tubes to transistors. It was pretty obvious that a transition from transistors into integrated circuits was something they would use. Indeed, the first integrated circuits that Fairchild ever made were the RTL gates that were used in the Apollo flight in Soyuz Apollo. That was the guidance computer.

Wolff:

That was just a few months ago, [in July 1975].

Noyce:

Yes. And that was the guidance computer on all of the Apollos. It was built by MIT.

Wolff:

Those were made using Fairchild's first RTL?

Noyce:

Using the first RTL design, yes. It is also interesting from the standpoint of how long it takes to get new innovations onto military equipment. That one needed to be pretty reliable, so it was tested many times. That computer could have almost been built on a chip by the time the thing flew. The first effort was to sell these to military equipment people. At about this time TI was coming in, and the entire technology for doing this was coalescing into the one that Fairchild had used. People were extending it in various directions. For instance complementary symmetry circuits doing p-n-ps and n-p-ns from the same thing. That is a much a much more complex process, but could be done. I would say that the thrust at Fairchild, though it was using the military as a stepping-stone, was very definitely to get the cost down as fast as possible to get it into the commercial market. The commercial market was the big one in terms of unit volume and so on – in the commercial computer market in particular. The performance of the first integrated circuits was not all that great in terms of speed and so on, but it improved as we refined the technology. Indeed the major market for integrated circuits now and for the last ten years has been the commercial computer market. The first integrated circuits were coming out about the time the first silicon transistors were finally going into commercial computers. Up to point, as you may recall, it was all germanium. The 2N404 type of transistor was used for logic. That was cheap, and there was a military market for silicon, and silicon was going to be used only for military.

Scientific Data and Control Data were the first people to adopt silicon transistors for logic transistors. The basic business that was supporting all of this integrated circuit business was that kind of commercial transistor market – which allowed Fairchild to make a profit and use those profits to extend the development efforts in the integrated circuit. Again, the integrated circuits were first sold to military systems people and then later to the commercial people. In distinguishing the efforts at Texas Instruments and Fairchild, I would say that TI had seen the integrated circuit as primarily something to sell to the military whereas we felt that it was primarily something to sell to the commercial computer market. The military was a stepping-stone to get there, but the big market was the commercial computer. The military market there was perhaps an opportunistic thing. It was a way of getting higher prices than we could out of the commercial market to subsidize the effort as it was developing the commercial products. The maximum rate of support of the laboratory by any military group was 4% of the R&D expenditures. Those were mostly on special studies.

There was one big Minuteman reliability program, which was simply to prove reliability. However that was on transistors, not integrated circuits. Along about this time the Minuteman II and Minuteman III programs came along and they decided they would make the Minuteman out of integrated circuits rather than individual transistors. We looked at the program and essentially did not bid on it. This was because we felt that we could not possibly achieve the timescale that they wanted on it. They were coming up with requests for circuits with all of the bells and whistles technology that really had not yet been proven in yet and were sort of the dreams that people had as to what could be done with integrated circuits. All of those things have been done since, but not in that timeframe. TI took that contract, and that tied up their entire capability for a couple of years and left the commercial marketplace wide open to Fairchild. Fairchild made enormous strides in the commercial marketplace at TI's expense at that point in time. We got to the one-dollar integrated circuit by 1963 or somewhere along in there.

Invention of the IC

Inception

Wolff:

Let's go back to the actual invention. When did it get invented? I don't mean the date, but what was the critical step? It was after you had the planar technology.

Noyce:

Yes.

Wolff:

Can you retrace who thought what or said what to whom or did what and where you fit in this?

Noyce:

We were having a review of the patent application on the planar transistor, as I recall. John [Lippincott J.] Ralls was there. He was Fairchild's patent attorney at the time. We were thinking that we ought to try cover as broad an area as possible out of this development. We sort of assigned our selfs to the task of doing what we could, in conceptualizing, to cover as broad an area as we could with that. That was in the back of our minds anyway, but I suppose that the actual invention came down to the point where one day one comes in to the lab and says, "Hey, here's a good way to do this. Why don't we do it this way?" I'd have to look up the actual data, but it was probably back in '59.

Wolff:

Without trying to recall the date can you recall the circumstances and tell me about that day?

Noyce:

I did not see it as a very unusual day. I just went in and probably had a discussion with Gordon Moore or Vic Grenich or somebody like that initially. I said something like, "Hey, here's a way to do the whole job of making some logic circuits instead of making individual transistors." There wasn't any need for describing what the utility of the thing would be around that shop. That was well known. It was a question of coming up with a compatible set of schemes to come up with the structures that would work electrically. Then we proceeded to try to realize the structures and file patent applications on it.

Wolff:

Let's try to tie down the organization here. What was your position at that time and who was "we"?

Noyce:

Fairchild was still a pretty tightly knit little group at that time. There were the eight founders that had been in there. Obviously the staff had expanded, but most of the technical work was being done by that group of eight founders, which included Sheldon Roberts who was a metallurgist, Jean Hoerni who was a physicist, Gordon Moore who was a chemical physicist, Vic Grenich who was in electronics, Jay Last who was a physicist and myself. And there were a couple of engineers.

Wolff:

You were a physicist.

Noyce:

I was a physicist by training, and had my training in solid-state physics. More accurately it was physical electronics. Solid-state physics was not yet a respectable field at MIT. We were a tightly knit little group. We had worked together at Shockley and knew each other pretty well. There was a awful lot of batting of ideas back and forth and criticism of other people's ideas, "Hey, that won't work because of this." Then we'd go back and say, "Well, let's devise a scheme that will work." It was a question of having rather vague concepts of insulators, isolation, interconnection and the photoengraving for the pattern. We drew on our bag of tricks these particular elements. Those elements in combination made the integrated circuit. Perhaps there was the “I’ve got it” type of thing, there is a way that will work. Once you have a way that will work conceptually then you start refining it to make it work better. Up until this point it was sort of, "Gee, wouldn't it be nice if you could?"

John Rall’s Challenge

Wolff:

Was there such a moment like that on this day that you recall as being ordinary?

Noyce:

Yes. I remember sketching it out on the blackboard for John Ralls, who was the patent attorney. Did I do it some other time for the other people? Very likely not on the blackboard, but I probably said, "Hey, why don't we do this."

Wolff:

What did you sketch for Ralls?

Noyce:

The sketch I did for Ralls was very similar to what is in the patent application.

Wolff:

Do you mean the patent for the planar transistor?

Noyce:

No. I mean the one for the integrated circuit patent. It shows the various elements and how they are constructed conceptually. It shows the transistor, the diode and the resistor as being three elements that we could make easily at that time.

Wolff:

I just lost the thread here because you had been talking about a meeting with Ralls to discuss the planar transistor.

Noyce:

Yes. No, this was a later one then. It was after he was throwing out the challenge as to what could be done with it beyond that – as any good patent counsel does.

Wolff:

You remember him throwing out the challenge.

Noyce:

Yes.

Wolff:

Then you began thinking about circuits.

Noyce:

Yes. Thereafter there was the insight that with the addition of that insulating layer the circuits could be done. There were a couple of other incidental inventions that came along at that time which were more critical to the immediate task that we had at Fairchild. We were trying to make high-frequency transistors. One of the problems with high-frequency transistors was that the structures were extremely small and contacts had to be made to extremely small areas. One of the first concepts in there that was patented was, "Okay, let's make a big contact but have the metal film over this silicon dioxide – which is the insulator – and have a hole in the silicon dioxide that is very small to make a contact through this very small area." That patent was filed, and after that immediately came the idea, "A metal strip can be put over the PN junction without shorting it out." As soon as that can be done the entire circuit can be interconnected. It was a progressive buildup of the bits and pieces of the technology that made the entire thing possible.

Wolff:

Let me run over this again to make sure I understand it. You had this initial meeting in terms of the planar transistor patent application with John Ralls. And he challenged you tell him what else could be done with this.

Noyce:

Yes.

Wolff:

Out of that challenge came the idea, "Well, we could have circuits if we had an insulating layer"?

Noyce:

We had the insulating layer then, and so we could have the circuits. I cannot recall exact timing of that. I would have to go back to notebook entries, and those are all at Fairchild.

Wolff:

I just want to get the time sequence chronologically. You credit John Ralls as sort of forcing you to think of things you probably would have thought of it anyway.

Noyce:

I credit John Ralls for forcing us to get our ideas written down and documented.

Wolff:

How do you spell his name?

Noyce:

It was R-a-l-l-s. He was a patent attorney in San Francisco. He died of cancer a few years thereafter.

Wolff:

He pushed you to think about a broader application and sometime soon thereafter came the realization that, "We have a technique for insulation here."

Noyce:

Yes.

Wolff:

And that was what would be necessary then.

Noyce:

That was one necessary thing. We had to have the insulators and we had to have the p-n junction isolation of the individual transistors.

Progressive Buildup of Bits
and Pieces of Technology

Wolff:

First you had the recognition you had the insulation. Did these other developments come soon after?

Noyce:

If one sat down as a semiconductor physicist and said, "How can I do this job?" that was the answer. Indeed, on the p-n junction isolation for multiple diodes or whatever, that patent was issued finally to Kurt Lehovec at Sprague. Clearly these were ideas that were coming up in the industry simultaneously. As people faced that same problem they came up with that solution to it.

Wolff:

In the sequence here with you at Fairchild, there was this realization that we have insulation. You talked about the high-frequency transistor.

Noyce:

Yes. The large contacts on small contacts, if you will. I think that application was done before the integrated circuit application. There was an application on the junction isolation, but it was lost in interference because it turned out that Lehovec had done that work earlier. Then it was really about putting the bits and pieces together. There is no doubt in my mind that if that invention had not arisen at Fairchild that it would have arisen elsewhere in the very near future. It was an idea whose time had come and the technology had developed to a point where it was viable.

Arrival of the IC Concept

Wolff:

I want to try to get clear on the sequence of these things at Fairchild. Can you recall what you talked about that day when you went in and told people, "Well, here's a way to do it"? What exactly did you tell people? It's fuzzy. After the realization of insulation what came next?

Noyce:

The Desiderata would be a way of making individual transistors that were electrically isolated from each other, having the individual resistors electrically isolated from the transistors or from each other, and a scheme for interconnecting them in any desired pattern. Basically what we were talking about that day was, "Okay, if we take the planar transistors and we add junction isolation under the transistor so it's isolated from the substrate. We have the oxide film on top, which is the insulator. Then a metal film can be evaporated on top to interconnect it by making a printed circuit board on top of the silicon." That was a minor extension, if you will, of the ideas that had a major impact.

Wolff:

Do you recall sitting down and discussing it in that way on that ordinary day?

Noyce:

Yes. It was precisely in that way. There was no huge light bulb flashing.

Wolff:

There was a day when you came to work and had these ideas.

Noyce:

I had those ideas worked out in my mind. Yes. That's right.

Wolff:

One could say that was the day when the integrated circuit was essentially invented conceptually.

Noyce:

That's right.

Wolff:

And putting these together was your concept.

Noyce:

Yes. Again, I was very much building on work that had already been done there.

Wolff:

Had the metallization been done by that time?

Noyce:

Yes.

Wolff:

Had the insulation also been done?

Noyce:

The insulation had been done. The junction isolation had been done, but I was unaware of that.

Wolff:

Had the junction isolation been done at Fairchild?

Noyce:

We had not made transistors that way. Obviously one could use bases of transistors and use them isolated diodes so that the integrated diode structure was in the transistor structures.

Wolff:

It could be figured out.

Noyce:

Yes. It's a simple extension.

Wolff:

On that critical day when you came to work and talked about the whole concept, do you recall with whom you discussed it?

Noyce:

I would have to refer back to my Fairchild notebook, and I don't have that.

Wolff:

I don't need it now, but that is something that I'd like to find out about.

Noyce:

I could ask Roger [S.] Borovoy. Since he fought all of the interference on that he might have some of those files here. Let me ask him if he does. [recording turned off and then back on at this point] Probably I discussed that with whoever witnessed that notebook page. That would be the way to find out.

Wolff:

That will be easy enough to find out. I would like to get their recollections. If there was a Eureka moment, that was it. Right?

Noyce:

That would have been it. Yes.

Wolff:

We'll find out what day that was, etcetera.

First Logic Forms of RTL

Wolff:

What happened after that?

Noyce:

We were a small enough group then that it was a very simple matter to say, "Okay, that's a great idea. Let's work on it and see if we can make the structures." We were looking for the simplest logic forms we could make that would be useful. That's why the first logic forms were RTL. Most of the computer forms had used both resistors and capacitors. We couldn't make a good capacitor in the integrated form cheaply, but the resistor-transistor logic form was resistors and transistors only. Therefore that was the device that was chosen. There is a lot of conceptualization of what you would have to make, and there was a lot missionary selling frankly.

Wolff:

To whom?

Noyce:

To the outside customers. The point here is that the semiconductor device manufacturer was stepping up one level in their design activities. Previously it was common to make a transistor and then give it to someone else to design a circuit. Now we were in the mode of taking the circuit and deciding what function we wanted from the circuit and then trying to design something that would provide that function. We did that out of the rather poor building blocks that we had. They weren't by any means optimum transistors or resistors. The frustrations in that were probably two-fold. One was that it was far more expensive to make than it should have been. We had not anticipated the whole yield problem. Here is the first one, but not the full integrated circuit pattern.

Wolff:

Which number is that?

Noyce:

[unintelligible phrase]

Wolff:

That's what I have. That is not the full pattern?

Noyce:

That is the device and lead structure, the expanded contact structure. It may show some of the original documentation.

Wolff:

Is your notebook in your possession here?

Noyce:

No, it's in Fairchild's possession.

Wolff:

Do you think I could look through that?

Noyce:

It may well be that Roger has a copy of those pages. I will ask him about it. Roger Borovoy was Fairchild's patent counsel and is now general counsel for us here.

Wolff:

What are these notebook pages from?

Noyce:

It's probably the support documentation for that patent.

Toward a Working Model

Wolff:

What happened after you outlined the concept of the integrated circuit to some of the other people?

Noyce:

I discussed it with the entire technical staff there, and it quickly became part of the whole concept and the business plan on which we were working. I think it seemed pretty obvious to us at that time that it was going to be a major change in the way electronics was being built. There were obvious labor savings involved in not handling the parts individually but rather handling in the groups and providing entire functions rather than just a simple amplifier or the transistor.

Wolff:

That had to come a point where you had a working model.

Noyce:

That probably would have been about a year after that – I would guess at about the end of 1961.

Wolff:

I wonder if our next chore is to try to reconstruct the major events leading up to that working model. Who were the key people working on it?

Noyce:

The key people were still certainly that original group there. Vic Grenich was in charge of the application – circuit design group, if you will. Gordon Moore was doing the process development at that time and in charge of the total technical effort in process. We simply proceeded to set up a project and go ahead and make them. I wish I still had some of the progress reports from that day, but I don't think I have any of them. I think that I was very careful not to take anything like that when I left Fairchild. There might be some published materials that you could find in some technical articles of that day that I conceivably could find in some of my memorabilia which would establish some dates and who did what and that sort of thing. As I recall, the planar transistor was introduced in 1960. This was during that next year. I think that the first integrated circuits that Fairchild made were shown at Wescon in 1961.

Wolff:

The application for the planar transistor had in a sense been invented before this.

Noyce:

Yes. Absolutely. That application was probably filed in 1960.

Wolff:

What did you do during that year?

Noyce:

I was the general manager of the thing there. I wasn't very involved in the actual fabrication of the circuits. I did a little bit of work on it on the side, but by 1961 Fairchild Semiconductor had gotten to be something like an 18- or 23-million-dollar company in that year. I was really running that whole business, and this was one of the projects being followed up in the laboratory at the time.

Wolff:

You are down as the sole inventor of it.

Noyce:

That's right.

Wolff:

Because of your conceptualizing.

Noyce:

It's a question of where the invention occurs. The patent attorneys want to be sure that the guy that had the idea is the guy whose name goes on the thing, because that can be torn apart easily in court otherwise with the question, "Did you really have this idea or did the other guy?"

Wolff:

There was a general agreement that it was you.

Noyce:

Yes.

Contributions From Other
People Toward the IC

Wolff:

In writing about the invention of the integrated circuit it is easy to focus it on you if we stop at the conceptualization and the actual application, but are there other contributions from other people that you consider important and should not be overlooked when we say what it took to get a circuit built?

Noyce:

Yes. Some of came before and some after. Gordon Moore's work on using one metal to contact both P and N type silicon was an essential piece of it that was done prior to this. The planar transistor was done before this. Those were the building elements. In terms of inventions, at least those two were essential. In terms of working out the technology it was a team effort, as is much of industrial R&D. There were contributors to that team which included that entire original group. Vic Grenich did circuit design. Dave Allison, who was not in the original group, did a lot of the process development. Gordon Moore supervised the process development. Jay Last tried some of the other schemes.

Wolff:

For what?

Noyce:

For isolation rather than diffused junction isolation. I hesitate to mention names in the sense that there was a sizeable technical group there, and a lot of them contributed significantly to the development of that early set. Bob Norman, who established Nortech later, was intimately involved in the circuit design. Bill Ferguson did some of the early process development work and testing of various processes. We had to work out the schemes for doing the triple-diffused isolation scheme, which was the first one that was used, and later the epitaxial isolation scheme. After the concept it fanned out to quite a bit of technical activity to pull all of these pieces together and make them work together. I would say that Fairchild went after that with far more dedication that than there might have been if it were just under contract, because it was commercial purposes.

Wolff:

Was there a contract?

Noyce:

No.

Wolff:

This was still in-house. There must have come a point where you had a working model of the circuit or the first circuit.

Noyce:

The working circuit was probably in mid-1961, but I am not sure.

Wolff:

Do you remember it being tested?

Noyce:

No. As far as I was concerned the thing was settled when we found a conceptual scheme for doing it. From there on out there is obviously always a lot of work to be done. Let me put it this way. When we could finally hold one up and say, "Here is an integrated circuit," those first ones were not particularly useful. The performance was not particularly good, etcetera. Therefore it was not the exhilaration of having something really wonderful. It was a continuing development to try to get it up to the point where we had something that was saleable.

Sebacks

Wolff:

Do you remember any important setbacks during that year?

Noyce:

Probably the most important setback was the yields. It was very difficult to get yields on these more complex devices. We announced a series of devices that would work together with two input gates, three input gates, flip-flops, etc. In the more complex ones of those we got yields of maybe half a circuit per wafer start or something like that. It was frustratingly low.

Wolff:

I meant in the year up until you got the first working circuit.

Noyce:

That all really went pretty smoothly. That's my impression now. I probably would not have given you that answer then. It was frustratingly slow, but in the global time that year was not very long.

Wolff:

There is an enormous research project that could be done here on the whole team effort leading up to the time you had a working circuit.

Noyce:

Yes, I'm sure there could be.

Wolff:

If I get support for that from somebody someday I'll get into that. However in terms of writing this initial article for Spectrum, it would be useful to focus on this conceptualization day and period.

Noyce:

It might be useful to try to get some of the original documentation on that so that we are sure that we have dates and who was involved and that sort of thing.

Personal Background:
Education and Jobs

Wolff:

I'd like to get that documentation, find out who was involved and talk to them. That would start to put that all together. I think that would work well. Let's take a few more minutes for you to tell me about yourself. You studied physical electronics at MIT. Tell me about your background.

Noyce:

MIT is where I met Wheelon. I graduated with a BA in '49 from Grinnell College. Nevo had an interview with me in Spectrum a few years back incidentally. If I've not mistaken it had a character sketch in it. He also had a story about the starting of Intel.

Wolff:

That of course may work for me.

Noyce:

Then I went to MIT. I had math/physics major as an undergraduate at MIT. I did a thesis in physical electronics – which was not significant work. Then I went to work for Philco Corporation. I guess even then I was trying to be a big fish in a small pond rather than the other way around. I had also interviewed at IBM, RCA, GE, Bell Labs and so on, which were the big companies doing work in physical electronics. Philco was getting into the transistor business. They really needed me was sort of the way I figured it. In many ways they were getting in over their heads. They didn't really have good scientific talent behind what they were doing. They had good technical people but more at the technician level, if I might put it that way.

That year they added half a dozen Ph.D.s and were really trying to put a solid basis under it. I was there from '53 to '56 working in surface barrier transistors and then the micro-alloy diffused-base transistors (MADT). Then I got the call from Bill Shockley, who was then recognized as the inventor of the junction transistor. He was setting up the shop out here. I came out here in April of '56. I worked for Shockley for 18 months and was somewhat frustrated in the working situation, as were a number of the people there. That is where this big splinter group that started Fairchild came from. It was sort of a group of equals as Fairchild started out, but I drifted toward being the director of R&D at that time.

Work at Shockley

Wolff:

Is it true that they voted at Shockley as to who would be the best technical leader?

Noyce:

Shockley had some strange ideas. One of the things that he asked his staff to do was to rate everybody else on the staff and rank order them. I didn't see that I could do that, but I did split them up into four groups of good, pretty good, medium and so on. He published the results of that for everybody to see. That was one of the sorts of things Shockley tried to do on an experimental basis. I'm not sure that it was a good idea, but it was good for my ego.

Wolff:

It's true then that you were rated the as the best?

Noyce:

I was rated pretty high and the only one on the list that was rated one across the board. I had more experience in the field than most of the people there. I had worked in the field longer. Anyway, I drifted into the position as director of R&D at Fairchild. We hired Ed Baldwin from Hughes to be the general manager at Fairchild. He split out in February of '59 to start what was then Rheem Semiconductor and is now Raytheon's operation out there. I then agreed to take on the top administrative job there on a temporary basis until we could replace him. After some looking around we decided that we didn't want to replace him. Ed Baldwin had not been a member of that original group that came into Fairchild, and after he left, presumably with all of the technology this group had developed, there was a great deal of distrust in the idea of bringing somebody else in to fill that slot. I ran the semiconductor division for several years in there. Eventually Charlie Sporck, who is president of National now, became in charge of that division. He had come up through the manufacturing route. I was responsible for that division and one that was in the instrumentation – making test equipment for transistors, etcetera, as well as the old Dumont oscilloscope activity.

Background

Childhood

Wolff:

Where did you grow up?

Noyce:

In Iowa.

Wolff:

Where?

Noyce:

My dad was a preacher. I was born in Burlington. Then we moved to Atlantic, Iowa, which is in the southwest corner of the state. We were there until I was eight years old and then moved to Decora, Iowa, which is the northeast corner, until I was ten. Then in Webster City, Iowa until I was twelve and Grinnell, Iowa through high school. After I graduated from high school the folks moved to a little town in Illinois called Sandwich and I went back to Grinnell to go to college.

Wolff:

Was your mother a housewife?

Noyce:

Yes.

Wolff:

Did you have siblings?

Noyce:

I have three brothers. One teaches chemistry at Cal, one works for IBM in San Jose and the other is in the Divinity School at Yale.

Wolff:

That's interesting. There is a lot of technical interest in your family. To what do you attribute that?

Noyce:

I don't know.

Wolff:

It's unusual to have three out of four brothers who went into technology. Was there anything special about your high school?

Noyce:

No, I really don't think so. I think it was a drifting. I respected my older brother, the one that teaches at Cal, and growing up in a small town in the Midwest we got exposed to a lot of mechanical gadgets, farm machinery and whatnot. One of the things that we did in the neighborhood was tear apart the local Model T and put it back together and that sort of thing. I always felt comfortable with physical things, so it was an area that came easily.

Studies at Grinnell

Wolff:

What did you study at Grinnell?

Noyce:

Math and physics.

Wolff:

Did you know when you went to college you wanted to study physics?

Noyce:

I took college physics as a senior in high school because I was bored with the general science course. I guess my interest in it sort of resulted from the fact that I mowed the lawn of the local physics professor. I did that as a way to make spending money. He was a very sympathetic character.

Wolff:

Is he still alive?

Noyce:

Yes. He has retired from Grinnell College. His name is Grant O. Gale. He was the head of the physics department at Grinnell. He had gone to school with John Bardeen at the University of Wisconsin in Madison, and the transistor was invented during the time I was in college out there. Grant Gale got hold of one of the first point contact transistors that was ever made. That was during my junior year there. I suppose that was one of the things that influenced me to get involved in transistors.

Wolff:

That's intriguing. You took college physics in your senior year in high school, You knew Gale at that time?

Noyce:

Yes.

Wolff:

Where does he live now?

Noyce:

He's still in Grinnell mostly. He spends quite a bit of time in Florida but still has his house in Grinnell.

Wolff:

That would be useful to talk to him. You grew up in the good old American tradition with many inventive people being and comfortable with machinery and so interested in technical subjects, but probably especially interested because of Gale.

Noyce:

Probably. Yes.

Excitement About the First Transistor

Wolff:



Do you remember anything about your junior year and the excitement about the first transistor?

Noyce:

It seemed like an enormously simpler way of making an amplifier device than to get a light bulb and put elements inside of it. It also seemed that it would be enormously cheaper than making vacuum tubes – and of course it has come out that way. Things never happen as fast as we think they ought. That was '47. We didn't really start getting transistors into television sets until the mid-60s or something like that. There was a long gestation time. Extrapolating, young people can take it on faith that, "Gee, that idea will work" and envision the end result of an idea. And it was a "can do" time. It was the time that the atomic bomb was being exploded the first time, and technology was capable of ruling the universe.

Wolff:

Much different than today.

Noyce:

Yes. In comparison, technology is ill repute now. It was quite different then.

Noyce’s Father

Wolff:

For what denomination was your father a preacher?

Noyce:

Congregational, which is now merged into the United Church of Christ. He is now retired and living in Berkeley.

Wolff:

Okay. This has been extremely valuable and helpful.

Noyce:

Okay. I'll try to get some of that other documentation.

Patent Disputes

Wolff:

Anything else?

Noyce:

[Beginning of sentence not recorded] ...the integrated circuit patent were the ones that have to do with the interconnect scheme of having the insulating layer, putting the metal film down and etching them to the pattern for the interconnect. Jack has the patent on perhaps a more fundamental part of the thing, which is the concept of putting two devices together to make a complex function out of a single one. The desirability of doing that was pretty well known in the industry. I felt the critical question was, "Can it be done commercially?" However the concept for doing it is – at least as far as the patent goes – is Jack's.

Wolff:

I've been through the whole court record, and I'm not a lawyer. I'm a journalist. That was how it seemed to have started out, but then there was a priority awarded to Kilby and then there was an appeals and he was overruled or something. There was a decision reversed. Can you tell me how it all finally came out in the end?

Noyce:

Yes. I think that you have to say that in terms of the way the integrated circuit is built today we split it.

Wolff:

In the end his priorities were not over—?

Noyce:

Not on the interconnect scheme, the method of interconnection, which is critical to it. I think that is the way that turns out.

Wolff:

That's yours.

Noyce:

Yes. Jack's original drawings, if you've looked up his patents, show shaped pieces of a semiconductor with a transistor here and using the bulk piece of semiconductor as a resistor and then a wire bonded from here to there and that kind of thing. It was a scheme of making it with sort of brute force in a single piece of silicon. However, it was probably not viable in terms of an economic way to do it. At least it has never been used economically. The contribution I made to that was to simply work out a scheme where it was economically viable. I think that's a fair statement.

Wolff:

Were all of his priorities overturned?

Noyce:

By no means. He got several of the claims.

Wolff:

Okay. What you are telling me is that you were given the credit or whatever for the interconnect scheme.

Noyce:

Yes. He has the credit for the concept of putting two devices together of the same material.

Wolff:

That is very helpful, because I could not track that through the legal stuff. I couldn't understand what they were talking about.

Noyce:

Basically you can't legally do an integrated circuit without licensing both his and mine.

Wolff:

I see. I was not even going to trouble you with that. I was going to talk to Roger about it. I'm glad you cleared that up for me.