John Bardeen

Hoddeson:

The purpose of this discussion is to answer a few questions that occur to me on re-reading the first draft of my article on the transistor. Sorry to have bothered you with it. OK, first of all, I wonder if you could just go over once more the difference between what Shockley meant by holes, and what you and Brattain, when you first realized that holes were being, were entering —

Bardeen:

There wasn’t any difference in what we meant by holes. The only difference is how the holes flowed from the emitter to the collector. In our earlier experiments, the surface barrier of [semi] conductivity type, the inversion layer, played a very important role. It plays a role in the field of practical experiments, and when the transistor in fact was first observed, the influence on the collector current of opposite sides to that expected for the field effect, we knew we were observing something new, and the obvious explanation is that through the metal contact at the emitter, the emitter contact, holes were being introduced into the semi-conductor, and flowing to the collector.

Hoddeson:

Yes. And Brattain had a note in his notebook, using the words that he realized that it was holes.

Bardeen:

Yes. The words emitter and collector were introduced very early, also indicate that we had the idea that holes were being emitted at the emitter and flowing to the collector. Originally it was not called a base contact. We called it a control contact, as it plays a role somewhat analogous to a grid in a vacuum tube. But it wasn’t too satisfactory, because control and collector both start with the same letter, so if you use subscripts to designate them, if you designate them by the first letter, you have a problem. So we decided to call what we first called a control electrode, the base electrode. So I don’t think there’s any, as far as this part I’ve discussed, there is no problem. The difference between ourselves and Shockley came in the picture of how the holes flow from the emitter to the collector. They could flow predominantly through the inversion layer at the surface, which does contain holes. And the collector would be draining out the holes from the inversion layer. They could also flow through the bulk of the — another possibility, that they could flow through the bulk of the semi-conductor, with their charge compensated by the increased number of electrons in the bulk. Shockley's idea for the junction transistor was based on the fact that holes introduced by the emitter in the junction transistor flow by diffusion to the collector, which is very close by the emitter, rather than through the base electrode, or what was earlier called the control electrode. In order to distinguish this from the flow in the surface barrier region, he introduced the concept of injection. Or he introduced the word injection. I wouldn’t say concept so much, but he, to distinguish the hole flow through the bulk from hole flow through the surface layer, he used the term injection to designate hole flow through the bulk.

Hoddeson:

In Shockley’s long article on the discovery of the transistor, he writes that, "when the transistor effect was observed, in the sense of the birth of the point contact transistor, no one had any idea that minority carrier injection, injection was being observed, any more than had been true for Bray’ s results. That is I guess just wrong? He goes on, "So far as I know, the first suggestion that minority carrier injection might be an important mechanism was made five weeks later in my disclosure, 23rd of January." I guess he just had it confused?

Bardeen:

Larry is talking about f low through the bulk. Brattain was making some experiments described in the PHYSICAL REVIEW LETTER published with the Letter on the Transistor — of the way holes flow from an emitter contact, by measuring the potential as a function of distance from the emitter, to see whether the holes are flowing permanently, predominantly through the surface layer, which would be two dimensional flow, or through the bulk, which would be three dimensional flow. If you have an inversion layer at the surface, as we certainly had, in these early experiments, you increase the Fermi level for holes in the vicinity of the surface, and the Fermi level is the same as, in the bulk as in the surface layer, you’ re going to get, introduce, increase the concentration of holes more in the surface than in the bulk. So it was not clear until more experiments were done just how the holes were flowing. In fact I think that they could flow - from what we know now, they could flow either way. The primitive experiment which showed they were flowing through the bulk was by John at the meeting you describe in your account there. He placed two contacts on opposite sides of a very thin wedge and showed that it worked just like a point contact transistor, with two contacts on the same side of a slab, which showed definitely that the holes were flowing through the bulk of the germanium, and that you could make a transistor this way. His transistor had characteristics which were very similar to those of two contacts on the same side. So there couldn’t be any essential difference in the way they operated.

Hoddeson:

Now, the Purdue group had been —

Bardeen:

— in our patent disclosure, we mentioned both mechanisms. We said that the transistor could work either way.

Hoddeson:

The Purdue group had observed the spring resistance, in the high vac voltage germanium . Was there any understanding at that time that this was due to hole injection?

Bardeen:

I think when we first learned about the Purdue experiments, it became obvious. The explanation became obvious that conductivity was being increased in the vicinity of the emitter contact, the point contact, by emission of holes, which increased the number of carriers. And this made us somewhat concerned that maybe the Purdue group would also think of the transistor, and so, anxious to publish as rapidly as possible, to get priority on the discovery.

Hoddeson:

Did you feel that you knew all of the work that they were doing at Purdue? Was there any lack of information? on their side?

Bardeen:

Not as far as I know. More than is normal, when you do research. You generally wait until you’ re fairly well convinced that what you have is a complete and reasonably convincing story, before you publish it or talk about it at meetings. I don’t think the way, they handled it was any different than any other scientific discovery involving publishable research. During that period, between the discovery and the first publication, just about six months later, we were of course confined with what we could talk about to other people, outside this small group in the company who were in on the discovery.

Hoddeson:

I’ m interested in how you learned about the details of the work at Purdue and the other laboratories. Was it through publications or written reports or visits?

Bardeen:

Well, first information, when we first started working, is, we had reports from the Radiation Lab at MIT and also from Purdue Pennsylvania, which were progress reports of work done during the war. And we read those. Later, the work of Tory and Whitmore on crystal rectifiers came out. I think it was published in 1948, which is giving the complete account of the work done in that Radiation Lab, somewhat also discussed the work done at the other laboratories, although not in as much detail. We depended at first primarily on these reports, and then later on publications, as after the war they began to publish their work. We depended on publications or discussions at meetings, papers presented at meetings.

Hoddeson:

Old you all read all of the reports, or did you divide up the material amongst you and report on it?

Bardeen:

We didn’t divide up the material. I think people just read as much as they wanted, until they thought they had a reasonably complete picture, as much as they wanted, of what work was done. Some of the mathematical work was, I’ll say that by Vivian Johnson, was done in considerable detail. I don’t know that anybody plowed through it all. But we got a pretty good picture in general of what they were doing.

Hoddeson:

John Bardeen on the period prior to the transistor being made public and the fear that the military might classify it.

Incidentally, you were talking about the secrecy period. Just before the transistor was made public, let’s see, I think first the military were informed.

Bardeen:

Yes, they were informed.

Hoddeson:

And then?

Bardeen:

— and we hoped they wouldn’t classify it, so we couldn’t publish. But we had already sent in manuscripts to the PHYSICAL REVIEW for publication. They were told to hold them back until we got word from the military that they wouldn’t be classified.

Hoddeson:

Why was there some concern that the military would classify them, the information?

Bardeen:

It wasn’ t done under any military contract or anything like that.

Hoddeson:

Was there any support from the military?

Bardeen:

There wasn’t any support from the military. It was entirely supported by Bell Labs. But any important discovery which might have military applications, it was desirable to clear it with the military.

Hoddeson:

I see, just a preventive —

Bardeen:

— just as a courtesy.

Hoddeson:

I see. OK.

Bardeen:

I don’t think the military recognized the importance of it. Perhaps not. Besides, it was so broad that, the possible applications would be so broad that it shouldn’t be classified.

Hoddeson:

OK.

Bardeen:

I believe that maybe in your story, that people who went down to talk with the military were told that similar work was being done in one of the government laboratories. It turned out not to be the case.

Hoddeson:

I don’t know that story at all. Where can I learn more about that? I don’t know about that.

Bardeen:

You don’t have anything on that?

Hoddeson:

I don’t think so. No. Which government laboratory?

Bardeen:

I think it was the Naval Research Lab. I think it may have been in some of Brattain’s accounts. I think he was one of those who — went down —

Hoddeson:

I’ll have to pursue that. OK, now I’m going to skip around a little bit. I have some —

Bardeen:

OK.

Hoddeson:

— questions from several years earlier. Well, maybe they’re not that— this particular one has to do with the relevance of All’s ? work. Now, All was the man who picked up on Southwick’s return to the tee of cat’s whiskers, and —

Bardeen:

Not just a return — point contact diodes.

Hoddeson:

Right.

Bardeen:

As detectors for radar.

Hoddeson:

Right. And he experimented with many materials, and discovered that silicon was the best one to use. And then, in working with the metallurgists to obtain pure silicon, he — Jack Scaff and Henry Tonura prepared an ingot with junction — which Awl then discovered.

Bardeen:

Yes. He tried to make them more uniformly, with more uniform characteristics, with the old Lena crystal, you had to hunt around for a so called hot spot, you see. To find a sensitive spot where the rectification would occur. Just not very satisfactory for a production item.

Hoddeson:

No.

Bardeen:

And I think the use of silicon was mainly to try to get better uniformity in the product.

Hoddeson:

Now, how did, how would you evaluate Old’s contributions with — in regard to the development of the transistor? And how, could -

Bardeen:

It’s background for work done at the Radiation Lab, on the development of point contact diodes. And the reason we used point contacts in our experiments is, really just because it was easier to do, not because it was the ideal geometry or anything of that sort, but just that we weren’t interested in going up to very high frequencies, so we didn’t have to use a — point contacts for that reason, and so, we just used point contacts because there had been some art about using them and we knew what metals to use, and how to contact the semiconductor. And how to make a rectifying contact. So both the first field effect experiment using point contacts and dropped electrode and the first transistor experiments which followed from that we used the point contacts, just for convenience, and the geometry chosen was just for convenience, just because it was something easy to do. The sort of experiment you can set up and do in one day. And, there was no other reason for it than that.

Hoddeson:

Well, you were the one who suggested that geometry.

Bardeen:

Yes.

Hoddeson:

After the —

Bardeen:

— one reason for suggesting it, is the earlier— for the field effect — whereas the other field effect work which by and large failed, was not only because of surface states, which we, could be gotten around by going down to very low temperatures, but because the mobility of the carriers in the thin vap resin layers which were being used was very poor. In other words, the electrical character properties of the thin deposited film, on which the first field experiments were done, was very poor. And so I got the idea of using an inversion layer, to get the effect of a thing layer by using the inversion layer on the surface of bulk germanium, and we hoped to — that you’d have good bulk properties but still have the effect of a very thin layer. And we took that first work with silicon because Brattain had contact with All and knew that with proper treatment you could cause an inversion layer on the surface of silicon. The experiments were first done, the first field effect experiments were done with silicon. When we did the experiments with germanium, we didn’t know there was an inversion layer there, but that was one of the— it was easy to do the experiment and find out from the experiment that there was an inversion layer there. And it turned out that there was. There was some suggestion of it before, because of the very good rectification characteristics, we knew that there was a high barrier, perhaps sufficient to make an inversion layer so we had hopes that it would work, and it did work. I said earlier that, in the field effect experiments, the inversion layer was very important, in that what you were modulating was the conductance of minority carriers in the inversion layer. Aril we probably had that too much in our minds, in interpreting the results, the first results of the one contact transistor, in that we knew that there was an inversion layer there which would provide a channel for holes going from the emitter to the collector. The question is, how important that aspect was, compared with the flow of the holes through the bulk, which Geoffrey [???] calls injection, to differentiate it from the word emitter which we’d already introduced. We wanted a different word to describe the flow in the bulk. And in our case, the flow was by the electric field of the conductor, and in his junction transistor, it was called diffusion through bulk, so there's a difference in the way the flow takes place. In Shives’ experiment, he showed the holes were flowing through the bulk again? The electric field of the collector current no doubt played a role there too. So it wasn’t purely diffusion.

Hoddeson:

Was the discovery of the PN junction important, essential to the work on the transistor?

Bardeen:

I guess it depends on what you mean by PN junction. The inversion layer at the surface is really a PN junction. And an inversion from N type say in the bulk to P type in the surface, in a sense, that’s PN junction. In that way it played an important role.

Hoddeson:

Would you have hit on that all by yourself without it having been previously discovered?

Bardeen:

It doesn’t require any doping of the — it's dependent on the inversion layer and the surface properties there, rather than introducing impurities into the bulk, so it didn’t make a — if you think of a PN junction as due to donors and acceptors, in the bulk, in the first transistor we didn’t have that. We just had inversion layer due to the surface properties.

Hoddeson:

So in fact, although I don’t want .... (off tape)

Hoddeson:

…but I’m just trying to figure out whether or not the first transistor could have been developed without the previous work, though probably the work on silicon wouldn’t have gone on as rapidly, if the earlier work on silicon hadn’t been initiated.

Bardeen:

The fact that we had people in metallurgy able to make fairly pure silicon — it was also known how to dope it, either — I think that fact that we had metallurgists available was certainly very essential to our program. And that also supplies us with germanium, again, bulk polyfissionable material.

Hoddeson:

To what extent were you looking for a semiconductor amplifier in the period 1945, ‘46, from the time you came to Bell Labs, up until that first crucial experiment which Brattain did which made him realize that the electrolyte was having an effect, which took place on November, 1947? In the period — did anybody ask you to look in that direction, either Kelly or Shockley or?

Bardeen:

I think it was in the back of everybody’s minds, that that was one of the goals of the program, and could be attained, and Shockley was pushing the field effect, and showed that at least theoretically, it should be possible to make it.

Hoddeson:

Then he brought his calculations to you, and you figured out that the calculation was OK, but he had left out the effect of surface states. Right?

Bardeen:

That was later. First, he — at first I just checked his calculations, and they looked OK, but should be observing the field effect, but the experiments which were carried out — we weren’t able to see anything. We tried a number of times, a number of different people, and that was some time later, I don’ t know just how long, after these reported failures, that — I’m sure it was done after I first checked his calculations - that I came up with the idea of surface states.

Hoddeson:

Do you remember how you came upon the idea?

Bardeen:

Some time in 1946.

Hoddeson:

Let’s see, your paper came out in March, 1946. On the surface state.

Bardeen:

‘46 was it?

Hoddeson:

Yes, March. Shockley says that sometime in the winter or fall, the previous fall or winter, he had brought his calculations to you.

Bardeen:

I think that’s fair.

Hoddeson:

So you were thinking about it for quite some time at least three, four —

Bardeen:

— well, a few months.

Hoddeson:

Yes.

Bardeen:

It wasn’t anything immediate, because I think it was hoped initially that we would find a scheme which would work, but .... was unable to.... The silicon or germanium that they were using, deposited films, these materials — I suggested cooling them down so the surface states wouldn’t come into equilibrium, and trying the experiment — with small effect, but smaller than you’ d expect from the bulk mobility. This indicated that the mobility of the carriers was far smaller in these deposited films than effect it was in the bulk, and so that the effect you might expect to observe even in the absence of the surface states was much smaller than the initial calculations, indicated.

Hoddeson:

By initial calculations, you’re referring to calculations done using the mop shop key theory?

Bardeen:

Initial calculations were made using the mobility which is known for bulk material, and introducing thin film, so that we had a thin film bulk material with the properties of bulk material, no surface states, you should have gotten a reasonable field effect, which wasn’t observed. And the two reasons for it were, one, the surface states, and, the other, poor mobility. And so, when Brattain and Gibney came up, showed that you could modulate the surface barrier with an electrolyte, I thought of this point contact geometry as a way of trying to observe a field effect on bulk material. I think that that was an important step in going from thin films to the bulk material with the good electrical properties. And then, we used the inversion layer, to get the effect of a thin film, and it was essentially the principle of modern MOS transistors.

Hoddeson:

Now, there was a period of one year and eight months between your paper on the surface states, and the experiment by Brattain to show the modulation....?

Bardeen:

All these experiments were done during that period.

Hoddeson:

In that period was — I mean, what were the group’s expectations regarding the possibility of building a semiconductor?

Bardeen:

I was trying Debuy’s [???] experiments, to determine the concentration of surface states. When the surface states paper came out, the program was divided into, one mainly on surface properties under Brattain, and another mainly on bulk properties under Pearson. And the, Brattain’s experiments and — were designed to learn more about surface states and surface barriers, and there were quite a number of experiments carried out which confirmed the idea of surface states and gave some estimate of the concentration.

Hoddeson:

But did you feel that if you understood the phenomena well enough, the surface states and whatever else you would learn about the semiconductors, that eventually you would as a group discover how to build —?

Bardeen:

— yes, then the field effect should work. And the amplifier.

Hoddeson:

I see. So you never really lost hope that the thing should —

Bardeen:

— oh no —

Hoddeson:

— throughout this entire period. Because it depends writing on who’s writing the story. Brattain, when he writes his accounts, tends to stress the basic research aspect of the work, and of course it was basic research —

Bardeen:

— it was basic research, because we were trying to understand surface properties, and we were not trying to make, by any empirical ways, new transistors or anything like that. We were trying to learn about nature and the semiconductor surface.

Hoddeson:

But the purpose of this research — supported by —

Bardeen:

— and then, when we were able to bring the surface states under control, we would be able to make a — so that — it was a program of basic research.

Hoddeson:

Yes, but it still had the ultimate goal —

Bardeen:

It had the long range goal to understand this and bring it under control. Then we would make a major advance. I think that’s sort of basic research which should be done in industry, is to understand more about the products, so it’s not just empirical but real understanding; so if something goes wrong, you know what to do to fix it, or if an important advance can be made, make that too. So you will understand what the limitations are, and what is possible so you know how far present projects are from what’s theoretically possible.

Hoddeson:

What were the most important experiments, as you recall, that were done in that period, between the surface state paper, and the start of the experiments on modulation which went, you know, which led you back to the field effect?

Bardeen:

Brattain did a series of experiments on measuring contact potential and change in contact potential with light shining on the surface, which gives information about the surface barrier. And this was — and then in addition, to experiments of this sort which were very important, experiments were done to try to see whether surface barrier was already present on germanium when you put a metal in contact with it, and this seemed to be the case, because the rectification didn’t depend much on the metal used for the contact, and you could also use for the contact heavily doped piece of germanium which had high conductivity, when, if you made a contact by having two wedges perpendicular to each other, with the edges of the wedges touching one another, that they just touched essentially at one point, so it was like a point contact. And this had characteristics of, similar to metal point contact. And then there were also experiments done having two rectifiers back to back — you could also do this with the wedges; if there are surface states when you apply a voltage, you can build up charge at the interface so that the voltage can divide between the part in which the direction is easy flow, and the part where it’s high resistance. And if you have a charge at the interface, this means that there must be some state to hold that charge, which, another experiment confirmed the idea of surface state. So there were quite a few experiments done, a large number on contact potential measurements, a large number on the rectification characteristics which helped confirm the idea of surface states.

Hoddeson:

Most of the experiments were done by Brattain and Pearson?

Bardeen:

By what?

Hoddeson:

Excuse me? Brattain and Pearson?

Bardeen:

Well, Brattain was doing these experiments on surfaces.

Hoddeson:

On surfaces — and Pearson?

Bardeen:

And Pearson was as I said doing experiments on measuring the bulk properties of silicon. He prepared a series of specimens doped at different levels. One of the things one had to confirm is, doping with group 3 and group 5 compounds, do these really do go in interstitially as a simple picture indicated? The way he did this is to make very accurate X-ray measurements, to show that the impurities really do go in substitutionally not interstitially. They substitute for the, group 3 and group 5 elements substitute for the silicon or the germanium. We later published this in a joint paper I wrote with Gerald Pearson on the electrical properties of silicon, which covered that topic pretty thoroughly, and the temperature dependence mobility, reasons for it, all the - a wide variety of things on the effect of doping. After the transistor came out, and people got interested in semiconductors, this paper got very wide distribution.

Hoddeson:

What about Shockley? Was he doing experiments on that in this period also, aiming to understand surface states, or calculations, or was he working on something different?

Bardeen:

He was not actively engaged in it himself. He was interested, and suggested some experiments, which Brattain carried out. He didn’t — he did some of the surface, the field effect experiments himself, or in collaboration with others.

Hoddeson:

Did he continue trying to do field effect experiments in this period after the surface state hypothesis had become accepted?

Bardeen:

The only one I remember, I think it was with silicon - when he lowered the temperature so that the electrons were frozen in the surface states and observed an effect, but much smaller than expected. And attributed that to the low mobility in the films.

Hoddeson:

Once again, in Brattain’s account, he describes these experiments also, and then discusses a single experiment that I also discuss in my paper, where he noticed that the water affected the results. That led to the work with electrolytes.

Bardeen:

Yea, it became apparent that the water affected the surface properties. Later, there was his experiment showing how you modified the surface barrier by humidity. This was after the transistor had come out, but we already knew the format, that modifying the surface properties by various sorts of surface treatments — and humidity and moisture and so on, were things which affected the surface.

Hoddeson:

Anyway, when one reads Brattain’s account or his interview transcript, it looks as though that experiment in which he noticed the effect of the water suddenly, after he discussed the experiment with Gibney and Gibney made the suggestion of modulating it with applied field, which then gave you the idea, that geometry in which one could actually observe field effect - that it was almost an accident that the group returned to active pursuit of the field effect amplifier.

Bardeen:

During that period, Shockley was heavily involved in other experiments, and the theory of dislocations, and also, flow of electrons through insulators like alpha [???] haloids and silver haloids, in which he learned about, got interested in these as a result of the trip we made, in the summer of 1947 — Bill Shockley and I, to Europe. He got much interested in dislocations after visiting Mott at Bristol. Mott and co-workers at Bristol. And so he wasn’t, at the time we were doing these experiments, he wasn’t so actively concerned himself. We discussed the experiments with him, but he wasn’t actively working in them during that period.

Hoddeson:

One would think he would get quite excited about the fact that now, perhaps, field effect —

Bardeen:

— after we got things rolling, of course he did get very excited about it, and put his full time on it.

Hoddeson:

Yes. But this resumption of the field effect research looks, as I say, it looks almost as though it was an accident. But on the other hart —

Bardeen:

Oh, I wouldn’t say it was an accident. It was part of an overall program.

Hoddeson:

Research...

Bardeen:

Perhaps partly accidental, but when you’re doing experiments and have an overall picture in mind of what you’ re trying to do, experiments which might not be relevant ordinarily do become relevant, and are put into the larger picture.

Hoddeson:

Right. OK, my final question on the list concerns the accident at which the point contact was created. You say you had an accidental washing away of the oxide layer.

Bardeen:

I’m not sure whether it was that, or the fact that the oxide wasn’t insulating, because it’ s difficult to grow an oxide to be — we know now that it would be difficult to do that experiment, even under the best conditions, grow a thing oxide on germanium, and then put a layer of metal under it. It can be done, with silicon; to make the MOS transistors, you grow a thin layer of oxide on a silicon crystal, without great metal over it, as the basis for the MOS transistor. It would take about 15 years of advances in technology to be able to learn how to do that without introducing too many surface states at the interface, and make it really, make a really effective device out of it. One of the key things was to remove - we had to learn was how to remove all traces of metal impurities from the oxide.

Hoddeson:

So, the I don’t know where I got this idea, that in washing off the glycol borate, Brattain had washed away the water soluble oxide layer. Is that true?

Bardeen:

Yes, that’s what he says.

Hoddeson:

Yes. But you think that that’s not what happened? That in fact, the oxide was there but it was not insulating? I don’t know that —

Bardeen:

Well, I’m not sure. In any case, the oxide wasn’t insulating. We know now that you can’t grow a thin oxide on germanium with operative metal, and expect to have insulating contacts on it. So I think there’s more to it than that.

Hoddeson:

I have to rewrite that section.

Bardeen:

But I think, growing the oxide and the treatment with the glycol borate helped produce the large inversion layer. At the surface. That was a good surface treatment, for getting transistor effect with point contacts.

Hoddeson:

Yes. I think you said last session that, I think Gordon asked you whether, if the — well, if the oxide had been insulating, whether —

Bardeen:

— we’d have probably had a field effect transistor.

Hoddeson:

Do you think that would have worked?

Bardeen:

Sure. That’s what, I don’t know if over half, but I wouldn’t be surprised if over half the present transistors work on that principle. MOS transistors.

Hoddeson:

Yes. Well, was your set-up appropriate for observing that effect? If the oxide had been —

Bardeen:

Well, we should have observed the field effect. As I’ve suggested, the MOS transistor — you modulate the inversion layer of the electric field [???] Now, you can do it, get a surface so you can swing it all the way from the inversion layer to the accumulation layer. That is, you can — there are so few surface states that you can swing the band all the way over from one way to the other. From highly type. At the surface. By application of the electric field. But if we … (off tape)

Bardeen:

thought that the field required over common [???] surface states, and modulate the inversion layer — it depends on the density of the surface states, as long as the density is smaller than the field required. And the experiment with the electrolyte we interpreted as finding very large electric field, because the ions are very close to the surface, and so you find a large electric field right at the surface, and it overcomes the surface states. And we wanted — in order to do a similar thing, we needed to, to have a large field with a moderate voltage, you have to have a very thin insulating layer. And that’s what we tried to do, by going with a thin oxide layer on the germanium.

Hoddeson:

But you said that, in that period, you didn’t know enough about how to grow this oxide layer, to grow —

Bardeen:

— to make an insulating layer. We didn’t know all the — at the start, whether we would be able to or not. We thought we’d try. Then we found it wasn’t insulating. That my have been in part because of a, washing the glycol borate off, but the main thing is that in any case, it probably wouldn’t have gotten us anywhere.

Hoddeson:

You wouldn’t, if you hadn’t washed the glycol borate off, you —?

Bardeen:

No, because I think even today, no one knows how to make an MOS transistor with germanium.

Hoddeson:

So the experiment couldn’t have worked as a field effect transistor, the one that you —?

Bardeen:

On the basis of what we know now, no, it couldn’t. Even so, it took 15 years, even though people knew the principle of it, to be able to make it in practice. To get the materials so perfect that they had a surface free from surface states. The problem is that only very small densities of surface state is enough to spoil the effect. So you have to have very high deal material to conserve the effect. It took 15 years for the technology to get around to learning how to do that.

Hoddeson:

You once said that if the accident, so-called accident hadn’t occurred, the shorting out, the contact, that then you would have had an MOS and perhaps the field would have advanced much more rapidly. But in fact, it probably had to go the way it did, because the technology wasn’t up to creating the proper oxide — at that time.

Bardeen:

No, that would have been hard, but we could have made junction transistors by doping the surface, the way bipolar transistors are now. The first junction transistors are grown junction transistors, in which the doping was done by putting impurities in the meld as the crystal was grown. The early junction transistors were wholly poorly only operated at relatively low frequencies. Even as late as 1952, point contact transistors had, they/we put in a lot of effort to try to improve the point contact transistors, and they had for many applications superior properties to junction transistors. One of the most important of these is the high frequency range. The point contact would operate at something like 15 megacycles, the junction only to 100 megacycles or so, 100,000 cycles, 100 kilocycles. So that junction transistors had to be improved and further developed. They were first applied to hearing aids, where you didn’t have to have a very high frequency, and the next major step was to make them suitable for radio frequencies. This was done with silicon transistors Texas Instruments, and I think some of the Japanese people who worked with Sony had a lot to do with developing germanium transistors which operated in the radio frequency range. Later silicon transistors to operate in the television range. The junction transistors that are made in this country are very costly, which meant that they could only be used for very special purposes, where you need the performance you can get, and most of these are military and space applications. So that the semiconductor manufacturers in this country are looking more to military and space applications, and left the field open to the Japanese to develop products for the consumer market.

Hoddeson:

Well, I think I can go back to work on this paper now. Thank you very much, for answering some questions that have been bothering me for a while.