Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of John Bardeen by Lillian Hoddeson with Gordon Baym on 1978 April 4,
Niels Bohr Library & Archives, American Institute of Physics,
College Park, MD USA,
For multiple citations, "AIP" is the preferred abbreviation for the location.
Systematically recorded autobiographical highlights from childhood through research at Bell Laboratories in 1947-1948 culminating in the discovery of the transistor. Discovery of transistor discussed in detail in fourth and fifth sessions.
Last time, we left off just at the beginning of the discussion of the work of the semi-conductor group. Soon after you started working in this group, along with Brattain and Pearson, you turned to consider copper oxide film...
Well, we weren't studying copper oxide films, but Walter had some data on copper oxide dating from before the war, which he showed me. And it was through trying to understand that data, that we worked out the theory for the oxidation of copper. What he was measuring was the rate of growth of the oxide on copper at different temperatures., And I was able to interpret this in terms of a mechanism, as I recall, involving a vacancy, motion in the oxide layer.
How did you move from that work to the work on silicon and germanium?
The main activity was on silicon and germanium, but of course it took a while to get organized and start getting data, and the data on copper oxide was already available, so...
So that the work on copper oxide was done after the decision to focus on silicon and germanium had already been made?
Yes, it was just kind of a sidetrack. But it did introduce some of the ideas on diffusion, the way diffusion takes place and things of that sort. The paper is the only one with our three names on it which I know of.
You commented very briefly on that paper last time and I think you mentioned that actually Shockley didn't do terribly much on this.
We had a little discussion as to whether we should have his name on it. Since he was head of the group, he decided to put his name on it. He didn't really have a whole lot to do with it.
At this time, this is very early on, was the group already focussed on finding a semi-conductor amplifier?
It was something just in the back of our minds. The research program was directed toward understanding the properties of semiconductors. The only important aspect of the work was Shockley's ideas on the field effect transistor, which was tried out by several people without success. That led to the surface states paper.
Now, Shockley had already done some work with semiconductor amplifiers before the war. He built one using the piezoelectric effect and then a copper oxide rectifier...
I was just looking at his paper on the history of the transistor (W. Shockley, "The Path to the Conception of the Junction Transistor," IEEE Trans. on Electron Devices, Vol. ED-
Which one of them?
I've seen it, that's an excellent paper, I think --
Well, he put in a lot of research, went to the old notebooks. I think it's pretty good except for a few minor points, I might have some disagreement with him. On the whole I think he did a good job of reconstructing the history.
Well, let's go through it step by step. I'm particularly interested in places where you disagree with his interpretation. Before we get involved, I want to ask one question about Kelly. Was he aware of the details of what you were doing in the early stages of the work?
No, he wasn't. He
... he set up the group and got it organized, but he didn't follow in any detail what we were doing.
And did the semiconductor subgroup of the solid state group interact very much with the other members of the group who were not working on semiconductors... Morgan...
... oh yes, everyone interacted ...
... I mean with the Morgan subgroup, Bozorth's group the Goucher group, and Mason's group. I'm wondering what the interactions between those subgroups were like.
Well, it varies between the different groups. Mason's group, with the exception perhaps of Walter Bond, didn't interact too strongly with the other groups. But all of the other groups interacted.
Were you aware of the details of their experiments and vice versa? Was there much input from other individuals?
These study section would have people from all these different groups.
These are study sections where you went through Pauling's [Nature of the Chemical Bond] book for example?
The interactions were closest among the people working in semiconductors, but we did have close interactions with all of the others.
Let's turn to the Shockley article (Shockley, "Path", op.cit.]. Now, he says that it was in April, 1945, on page 604, April 16, 1945 is the date of the notebook entry for the first two fields effect transistor designs.
Then apparently Brattain did the experiment in which, according to Shockley, he didn't see anything. According to Brattain, he didn't get a very large effect. He saw something but it was negligible. Shockley then said in his IEEE paper that on June 23rd, he estimated quantitatively the degree by which the effects fell short, and brought his calculations to you, and then...
This was in the fall...
He says it was June...
...that may have been the time he did this, but I didn't join the group until, must have been October, I guess.
I see. It does seem strange that there should have been such a large time delay before you came back with an answer. He dates your answer as the 19th of March.
He says he brought it to my attention, but that was not in June 1945. That was much later.
I don't know what the date was. It couldn't have been at that time because I was still in Washington.
OK. And then it still took a few months before you came out with your theory. Or did perhaps your theory come out before you actually wrote it up?
It was not too long after I heard about the experiments. I think that it was probably the result of a retrial of the experiments using silicon. Well, he tried it with several people. Using thin films of silicon.
..."had been deposited by Gordon Teal"... This is page 605...
[reading] "showed there is a substantial"... so this must have been after he made these quantitative calculations because he talks about his calculations. But I know that Ohl deposited some films, measured by Pearson and others. This was one trial, but it was not the only one, by any means.
I see. So there was a series of attempts to measure the field effect.
Particularly to make a quantitative estimate. We need to know the mobilities of the carriers in these thin films. And that requires a Hall effect measurement or something like that.
I see. It seemed to me at first when I looked at the literature that Shockley should have perhaps gotten the idea of surface states himself because he had written an earlier paper before the war on surface states for an ideal situation.
When he was at MIT as a graduate student, he got the idea of dangling bonds, thinking of surface states in terms of the chemistry picture of dangling bonds, and showed that you get dangling bonds if say you start with a ?????? P level, separated, as they are in the atomic state, if they cross over, so you find what you're doing is finding the tetrahedral bonds when the band cross over and the lower band corresponds to the electrons in the tetrahedral bonds from a chemical point of view. And what he showed is that if you have crossing bands of this sort, the surface states you also have primitive ideas of surface states where you assume some potential in the crystal and you have to assume where it's broken off, as it goes from the crystal to the vacuum and see if you have any bound states at the surface. He made calculations of that sort. This is one which made more sense, tied in with the chemical picture -- free bonds for electrons at the surface which could accommodate electrons --. But this, of course, you're talking about an ideal surface in this case, an ideal surface in a vacuum and such states associated with broken bonds have been observed. In actual surfaces you always have an oxide layer and with a surface exposed to air, most of the dangling bonds will be filled up with something or other because they're chemically active.
When you told Shockley about your surface state concept, did he put that picture together with his earlier theory?
Well, he accepted it right away.
He refers to I think in his article, as "one of the most active research topics in the transistor group."
And as a result of it, we organized the experimental efforts so that Brattain concentrated his efforts on surface properties and attempts to measure surface state properties and things of that sort. While Pearson, who was the other experimentalist in the group, concentrated on bulk properties. Although it was an experiment of Pearson's which gave the first indication that the field effect really exists, because you have surface states and you have a time constant associated with them, which presumably increases at lower temperatures. I suggested doing an experiment at low temperatures, which he did, and he observed an effect, but much smaller than we'd expected because of the low mobility of the carriers in the film, it was a deposited film -- it was probably crystalline and the crystals were extremely small, so the mobility of the carriers was very small, very low, was kind of on the borderline between amorphous and crystalline. And so you couldn't expect, a field effect based on bulk mobilities, because of the low mobility of the carriers in these thin films.
Would you say that Pearson's experiment at that time with low temperature was one of the crucial experiments to demonstrate the surface states?
Well, it showed that there were two difficulties with the thin films. One was surface states and the other, the low mobility of the carriers. The idea suggested explanations for other things, already known, which hadn't been explained.
Yes, such as the fact that there is no contact potential between semiconductors of different conductivity type, for example.
Or the fact that their rectification characteristics didn't depend on the nature of the point contact. You could even quite effectively make a point contact between two pieces of germanium and use that as rectifiers back to back so that you have to have a place, were charge could be placed at the interface.
I noticed in the on surface states that you discussed the work with Shockley, Herring and Brattain. I was wondering what contributions they made.
Well, I don't remember specific ones now. There is often confusion as to what's in this, what's meant by surface states in this paper. Some people think I was talking about states on an ideal surface. Those are what Shockley was talking about, the theoretical point of view. But I was using the concept more generally, as t any states that might exist at the interface, particularly the ?????? of surface states, but also surface in fractions, foreign on the surface. On general grounds there's good reason to suppose that the ratio of the number of surface levels to surface atoms may be much bigger than the ratio of the number of impurity levels to atoms in the interior.
... page 719.
To get enough surface states to effectively shield the interior, you need about 1 per 100 surface atoms order of magnitude. Now, in the interior, you might have one, the interior at that time, you might have one per million bulk atoms. You'd expect many more imperfections at the surface. So it's evident from this paragraph I was thinking of the concept much more generally than ideal sorts of surface states. Just any sort of states at the surface which could serve to trap electrons. One of the important points is, you don't need very many to shield the interior, 1 to 100, so with the surface atoms the interior is effectively shielded.
The next crucial step towards the point contact transistor seems to be a particular experiment that Brattain did, one of a series, in which he was measuring the temperature dependence of the contact potential at the surface of a sample of germanium or silicon. He noticed -- this is in Brattain's account -- he notice that condensation of water from the air on the surface caused a considerable hysteresis effect as the apparatus was brought from high to low temperatures and back.
And the humidity and things of that sort had large effects.
Yes. The fact that the water was in some way affecting the experiment then led him to put the whole apparatus in water -- which looks like a somewhat drastic step. I wonder if you'd comment on it. First of all, was that a natural step to take, simply to put the whole apparatus under water, in order to avoid hysteresis?
Well, he was doing a lot of experiments which would give information about the surface barrier. And measuring the contact potential and the way the contact potential changes with light -- one of the most direct ways you can get information about the surface barrier. When you shine a light on it, you can see how the surface barrier is affected, how the contact potential is affected by the light. You can get information about the sign of the surface barrier.
But the accidental hysteresis seems to have let to a new area of exploration.
He was doing a lot of experiments regarding the surface barrier, and I guess he thought dunking it in was another way you could get information.
Did he realize that the fact that the water was affecting the experiment offered an important clue? Or was it just regarded as a general nuisance?
Well, it indicated that humidity was having an effect, certainly; water condensing on the surface was having an effect on the experiment.
In a sense that was exactly what you were looking for, such effects.
We were looking for such effects, ways to control the surface barrier. Eventually we wanted to control it with an electric field, the field effect. That led to the experiments with Gibney in which Gibney, I think suggested using ethylene glycol, wasn't it?
Well, Brattain and Shockley say it was glycol borate. When I spoke with Gibney last January, he thought it had been monitol.
We'll just have to go back to the notebooks to find out.
I think the notebooks say glycol borate. But, in any case, it was the condensation that led to the dunking of the whole thing in water.
... it was the indication that water was having an effect on the surface.
And then Brattain noticed a much larger effect, when it was entirely immersed in water, a large change in the photo emp. And at that point, he showed the experiment to Gibney who apparently realized that the mobile ...
...that the water was acting as an electrolyte, and putting in a real electrolyte, would lead to a larger effect.
What did you think was happening at that time? How did you think the water was affecting the experiment?
Well, just I guess the water -- I didn't have too much of an idea whether it was a chemical effect at the surface, then when you put in a electrolyte and apply an electric field, yet got a big effect. It was evidently a large field produced by ions at the interface which overcomes the surface states.
There was no talk of anything like hole injection at that time?
No, not at that stage. There were no indications we were doing anything except getting a large electric field at the surface. And they could measure the contact potential and change the contact potential of ????
Were holes being injected at that time?
This is November 17th, 1947.
There was no rush of conduction between the electrolyte and the ??????? Just static effects. If you apply a field, you can get a slow electrolytic decomposition of the germanium. That's obviously not very large currents.
Did you have any idea of trying to make a contact at this point? This is November 17 we're still at.
In this article of Shockley's ("Path", or. cit.] talks about Thursday, the 20th of November -- this is three days after they first observed this large effect of the electrolyte -- they wrote a disclosure proposing that this field effect with an electrolytes could be used to build a field effect amplifier. But they didn't -- in the patent application, they showed as an example of how this could be done, the point contact on the surface -- this is one of the things which is not too clear in Shockley's article. The disclosure they made was the 20th of November, and it was on the 23rd of November Hoff tape)].
November 20th, 1947.
This was a notebook entry which has three days after they first observed this large effect of an electrolyte, applying a field through an electrolyte on the surface barrier of -- let's see...
In connection with this, both Brattain and Shockley claim it was Gibney's suggestion to vary the potential bias. When I spoke with Gibney, he didn't remember whether that was his idea.
I don't remember because I wasn't directly involved in it. But, in any case, they were both involved in it because they wrote this disclosure concept which eventually led to a patent. But originally this was, when they wrote this disclosure, it was just a kind of general idea that you could get by the surface states using an electrolyte and thus presumably, make a field effect transistor. But the idea of using a point contact to observe it on bulk material was mine, and it's in this entry on 23rd November.
Yes. Now, the entry on the 23rd of November was the entry that suggested the point contact arrangement in which the point contact is surrounded by wax and a drop of water.
Yes. And the reason for doing that -- suggesting this arrangement -- it was originally just a very simple way of testing the idea to see whether this field effect idea would really work. Which they hadn't presumed that you could put on electrodes and carry out the experiment. This was a simple way in which you could do it. And you'd get around two problems, with thin films, if you made use of an inversion layer on the surface. I'm talking here about a thin n-type layer on the surface of a block of p-type silicon. Instead of using an actual thin layer, about the only way we had to make them at that time was by deposited film which had poor electrical properties -- you could use bulk material with an inversion layer and thus you'd get the advantage of the high mobility in the bulk.
And then the point contact geometry is just a very simple way of doing it.
I see. So using the inversion layer was simply a way of getting...
...a way of getting effectively a very thin film but in bulk material so you'd have a high mobility...
I see. Why did you have to use such thin films?
Oh, you have to use a thin film because with the sort of electric fields you can use, the number of carriers. The amount of change in resistivity of a film depends on the charge you can induce by the field, which is a fixed amount, and the relative change is greater the thinner the film.
So you needed a very thin film to get a sizable effect. So this arrangement would get around the problem of the very low mobilities in deposited films.
Was this the first time that an inversion layer was used this way?
This is the first time an inversion layer was used.
That is in the concept of Brattain and Gibney, it was just a way you could vary the surface barrier up and down, but they hadn't given any specific arrangement in which this could be done.
I see. And then this actually worked, this apparatus with the point and the water drop. You observed amplification of power on the 21st of November. But I understand there were two problems, one being that the water drop tended to evaporate, and the second, that you only had low frequency response. Is that correct?
I'm getting this all from Brattain's account.
Well, I wrote it up on the 23rd which was a Sunday so I must have had the idea before that [laughter]. I didn't go into the laboratory on Sunday. It may have been on the 20th or 21st. It was the 17th when it was observed. It was the 20th, three days later -- it must have been earlier than that -- when they wrote their disclosure and it was the 23rd when I wrote my notebook entry.
Then there are a number of experiments that are described in Brattain's notebook and reprinted by Shockley on page 609, figure 13.
Well, this is their figure 3, in the patent...
...figure 14 on page 609.
Figure 14 is one they used in their patent to illustrate that you could use this effect to get a field effect transistor. But it was my idea of using the inversion layer in this geometry, although my name wasn't on the patent. They had the idea earlier that you could make a field effect transistor, but using an inversion layer with this geometry was my idea. They just used this as a kind of reduction to practice.
Now, two changes were made soon after. One to change the electrolyte, apparently to glycol borate, then called "gu".
Yes, it didn't evaporate so rapidly.
And also to change to high back-voltage germanium, and that apparently was your suggestion. I'd like to know why you came up with that.
Well, I don't remember the details, but according to Brattain's memory, that came as the result of a lunch discussion with Shockley. Brattain and Shockley and I apparently had lunch together.
Was Shockley very closely involved in these experiments?
Not very. He was heavily involved after the discovery of the point contact transistor. But not so much before.
OK, well, then it changed to using....
This is more just a kind of a simple experiment to see whether it would work, to see whether there was an inversion layer. We didn't know that there was an inversion layer in germanium. It's called high back-voltage germanium, it made good rectifiers. But...
Was it more pure than the silicon?
It's essentially pure. That was the essential reason.
It made good rectifying contacts, is the reason it's called high back-voltage germanium.
And you knew this from the work of the Purdue group?
From the Purdue group, yes. They had the germanium available so the experiment was very easy to do. So, we thought we'd try it.
I see, but you weren't sure you'd get an inversion layer.
And that worked much better than the silicon. It indicated that there was a p-type layer on the surface of the germanium. Then of course, we also knew from other experiments on germanium rectifiers that the barrier was present on the free surface, that is you have an inversion layer on the free surface. Because if you made contact between two pieces of germanium, effectively a point contact, and he was doped very highly so it's essentially a metal, you got about the same rectification characteristics as you would get with a metal point contact. And from then on, most of the experiments were on germanium because the effects were larger on germanium.
I'm a little bit confused about this experiment. After you switched to germanium and to glycol borate, Brattain writes in one of his articles that you noticed a reversal of sign. And were surprised.
We used P-type silicon with and N-type inversion layer, and in germanium, it was N-type germanium with a P-type inversion layer, so the signs were reversed.
You didn't realize that when you started?
Oh yes we knew that it would if the experiment was going to work properly. Because we knew that it was N-type germanium, and rectified as N-type material, so if we were going to get an effect, it would be of the opposite sign. We knew we had an N-type inversion layer in the P-type silicon.
I see so it was no surprise.
It wasn't any surprise. That's what we expected and that's what we found.
The surprise was perhaps that you could see, through the glycol borate that you were growing interference films in this experiment. Brattain mentions this.
Well, the glycol borate was affecting the surface. It was gradually etching the surface, slowly etching the surface and making an oxide on the surface.
And how did that show up in the measurements? Did that cause the resistance to slowly rise?
Well, it didn't show up in the measurements.
It was just a visible detection?
It may have affected the interface, although the rectification characteristics are just about what you'd expect without the glycol borate, you just measure the reverse characteristics.
I mean the film. The interference films that were being grown under the glycol borate, you say that did not affect the results?
That's probably an oxide.
Yes, the oxide that was being deposited.
Yes grown -- wasn't that insulating the electrolyte from the semiconductor?
I don't think the oxide was that thick. It couldn't have been so thick you'd get interference from it. It was just a few layers of oxide.
Let's see, this is an interview that Brattain did in 1964 with Alan Holden. They discuss, let me find it, he say, this is Brattain, on page 30 of the Holden interview, that "In the process of working with these things, we found that if we put on some steady bias on the electrolyte, we got a bigger effect, put our AC on around some bias point, and then a steady bias on high back- voltage germanium -- it was in the anodic direction -- and we could see through the glycol borate that we were anodizing, growing visible interference film, green film. I can remember the green color under the glycol borate. We were unable to make the thing amplify much above 10 cycles. We reasoned that this was the slowness of the response of the electrolyte, and...." OK, then he talks about the glycol borate being slow, and then he goes on, "So seeing this film we thought, ah, this oxide film must be insulating. If it is, we can form the film and put metal electrodes right on top of the film, get this field effect without the electrolyte, and get the higher frequencies."
Well, that's what we were trying to do, is form a thicker oxide film on the surface. So it may be that you could grow a thick enough film, if you applied a steady voltage on it, but just doing the experiment to show that you got a field effect, I think we had effectively just a very thin oxide layer, not a thick oxide layer, not one comparable to the way it went with ??????
It seems that there was some confusion about how the oxide coat was functioning.
We did want to -- this is described in my Nobel lecture, [J. Bardeen, "Semiconductor Research Leading to the Point Contact Transistor," Les Prix Nobel En 1956 (Stockholm 1957).], which exactly quotes these experiments between the 8th and 16th of December.
Yes, this is on page 611, the first column.
Although it worked well as an amplifier, it could amplify only very low frequencies and we knew that if you have the sort of ????? parts you have in the electrolyte alkali, you're never going to get up to high frequencies. And so this is not a practical way of making an amplifier. So we wanted to get rid of the electrolyte, and at the same time, to get a higher electric field, so we wanted to get a very thin oxide layer.
Now, were you in effect trying to have the oxide layer play the role in the experiment that the electrolyte played previously?
Yes, so you wouldn't have an electrolyte, you'd have just an oxide layer, we'd evaporate metal over the oxide, and hopefully the oxide would be insulating. But the oxide wasn't to that extent.
[Laughter] Before we get to that step, I just want to ask, did you think that the oxide layer would be polarized in the experiment? How would the electrical effect be communicated through the oxide layer?
It would just act as a condenser.
An MOS transistor, metal oxide semiconductor. The oxide was very thin. You can get a high electric field when you apply just a small voltage across a small distance. To go back to the original idea of getting a field effect, with this geometry we knew that we had an inversion layer and we got around the mobility problem of a thin film. So you could get a high electric field, with a moderate voltage, then you could make an amplifier.
Gibney remembers that after you noticed that you were accidentally growing an oxide layer, he had the idea of intentionally anodizing.
Creating a thick oxide on which we could evaporate the gold. See, the transistor here (pointing to a figure in Shockley "Path" article].
This is page 611 and it's the original.
Figure 16, the original point contact transistor, you can still see the gold spot on that piece of germanium although we weren't using it in this experiment.
This little gold spot right here. But before that, Walter Brattain observed a small amplification, voltage amplification or power amplification, by putting the point contact right near the gold spot, and using the gold spot as the emitter and the point contact as the collector.
Now, this I think is the crucial step.
That's the one where first, and this was in the opposite direction, whereas in the field effect, in this case, where you have an N-type germanium with a P-type surface layer, if you apply a positive voltage, you tend to drive out the holes if it's a field effect and decrease the current, that's what we observed. But in this experiment with the gold spot, we observed when we applied a positive voltage, an increase in current, not a decrease.
Then you know you were contacting.
Then we knew that we were not only contacting, but somehow introducing carriers into the layer.
And this is the first time you've talked about hole injection?
Yes. It's the first time.
Now, in Brattain's account, it appears that this contact was an accident. He claims he accidentally washed the oxide layer off when he washed off the glycol borate and that caused the contact. Is that the way you remember it?
Well, as I remember for whatever reason, the gold spot wasn't insulating in the way we'd hoped. We though we'd try and see what happened anyway. [Laughter].
There's a discrepancy between Brattain' account and the official Bell Laboratory account put together by Gorton. [W.S. Gorton, "The Genesis of the Transistor", Case 38139, December 27, 1949, 1100-WSC-XB ]. Gorton ways, on page 5, beginning of the first full paragraph, "In order to increase the frequency response, Bardeen suggested replacing the electrolyte by a metal contact--
-- and an oxide layer.
But Brattain, again in his interview and also in his Physics Teacher article W. Brattain, "Genesis of the Transistor"], claims that, let's see --
Well, there's really no discrepancy. Just that this is not a good way of forming --
Let's see, Brattain said, this is the bottom of page 30 of the Holder interview, "I inadvertently shorted the point of the gold film in the nice center hole that we had fixed up very early in the experiment. So I got practically no data there was disgusted with myself, of course, but decided that there was no reason why I shouldn't go around with the point, around the edge of the gold, to see if there was any effect, even if the gold was covering half the surface. I got an effect of the opposite sign. I got some modulation," etc., and then he realized that he had washed the oxide film off. It sounds like an unexpected effect, a true accident to make the contact. But this account by Gorton's makes it sound as though it was a designed step.
Well, the anodizing and the evaporated gold contact were supposed to design it to make an insulating oxide layer between the metal contact and the germanium surface. But the gold spots were not insulated, so that, of course, you couldn't tell when you just measure that, the resistance across, to see whether they're insulating or not. Whether you must have a few spots where the film was gone or whether it was essentially making contact almost everywhere. And I think the latter was true, as Walter says in his statement. But we thought we'd see what happened anyway, even though it wasn't insulating. Then we observed this small effect.
But there was no power amplification?
No power amplification. There was some voltage amplification because it indicated holes were going into the contact, the gold was supplying holes to the surface layer because the current was enhanced and the reverse current was mostly holes going into the contact rather than electrons going out. And you were enhancing that current so you had to be introducing holes.
Was this clear at the time? Did you understand it as well then as you just explained it to me now?
That's the way I understood it. But what wasn't so clear was just how these holes were flowing in the surface. I'm not sure -- some of the later discussion in Shockley's paper takes up the point about hole injection. We knew we were introducing holes. Recently, came the designation of calling one of the contacts the emitter and the other the collector. Originally, they called the base electrode, the control electrode but then this didn't look very satisfactory because if you're going to see subscript, you have a C for the Collector and C for the Control electrode. And so they started using the base as the control.
Ok, we're on again.
Then, right after this came the suggestion for the first point contact transistor to make the two contacts very close together. We originally thought that we would get a better effect if we had line contacts, that the reason...
I see, these were actually line contacts.
Line contacts, yes, that's the reason for this geometry, to get two line contacts very close together. Bt then it soon became evident that you really didn't need lines if you had enough current to the collector. The field of the collector would draw in the holes so that the point worked just as well. Walter got the idea of grinding off a plain case of a point contact, so you could put two point contacts very close together. This geometry with the quartz wedge and gold on the side wasn't used very long. It served for the first experiments and worked when we tried it, first time we tried it using the same surface which had been anodized with the gold spot on them. And with this line contact, we didn't have extremely high resistance in the reverse direction. With this gold contact, the resistance in the reverse direction wasn't as high as it is on a normal high back-voltage rectifier where it's extremely high so you don't draw very much current. And then, one tried to use point contacts, they worked well if their reverse resistance wasn't too high but in many cases, the reverse resistance was too high and so it didn't work. It depends on the nature of the surface and what contact you put on. What we thought was happening was that if you had a good inversion layer on the surface, then the resistance would be low, but if the inversion layer wasn't all that great, then the resistance would be higher. Later we found out that using tungsten contacts, I think they wee tungsten-bronze, if you passed a large current through a collector, that you so-called "form" the point contact and then it does have more resistance, and then you could do this every time. It isn't so critical as to the nature of the surface and the contacts. You could form, what you call form, the collector. Evidently, what you were doing was forming a small P-type region when you formed the collector. It also reduced the resistance from what you get with a normal point contact.
Did you understand what was actually happening in the forming process, how the physical properties of the surface were actually being modified?
It melted a little region around the surface which introduces impurities.
Did you understand that at the time?
We had a pretty good idea of what was going on.
Were you aware of Bray's work at Purdue on the "spreading resistance" at the time you first began to see hole injection?
I don't think we tied it in immediately. But in the sequence of events which followed, December 23rd when we were trying to get the patent applications in as rapidly as possible, because we didn't want to get scooped on the discovery, and wanted to get the patent applications in so we could eventually publish. I was assigned the job of working with the patent attorney so a large part of my time for the next month or, next couple of months, was spent with the patent attorney who was one who hadn't known anything about semiconductors. I had to start from scratch to teach him something about semiconductors.
This was Hart?
Yes, this was Hart. And I tried to get the applications written. So, a large part of my time was taken up by that. In the meantime, Shockley jumped in with both feet. That's the time he really got excited and spent all his time on it. And he talks here about it in his article, let's see if I can find it. Well, our thinking up to that time....
... this is page 616 [Shockley, "Path", op. cit.].
This is from December 23rd until, let's see if I can find the date here -- it's the date when John Shive gave a talk to the group, showing that you could make a point contact transistor with points on the opposite sides of a thin wedge, which showed that the holes must be going in from above. That was, I think, early February.
Oh -- was it that late?
Let's see if I can find this. Well, I think it was early February, a little over a month later.
Oh, yes, page 618, he says "Striking experimental evidence for injection...."
...it's reported, 18 February by Shive. And when Shive was talking, I immediately came to the conclusion that that's what was happening and then Shockley says that, following that, he gave his -- his first announcement to the group of a junction transistor structure which he thought of as a way of trying to understand how the point contact work, and this involved the injection in the bulk rather than depending on the surface layer. And this is in late January. It was about a month later that he came up with the junction transistor structure, which would be about the 20th of January. And then this was the 18th of February, about two or three weeks later, when independently Shive decided to see whether he could make a transistor to work with two point contacts, the opposite sign, which showed definitely that holes are being injected into the bulk.
Was this the first definite demonstration, that what you were seeing was holes injection?
That was the first definite evidence. And then Shockley earlier got this juncture transistor structure for which he made calculations.
You pointed out in several places, for example in your Nobel lecture, that the Bell program differed from programs elsewhere in understanding the significance of the minority carriers, the holes. Why hadn't others realized this? Was it such a far out concept at that time?
Dealing with non-equilibrium phenomena. I think people are just not used to dealing with non-equilibrium phenomena of that sort. And our thinking was biased toward the inversion layer on the surface, because that's the way the experiment started and clearly demonstrated that it was there and we were trying to understand why some point contacts would work and others wouldn't; some would have high resistance and others low. It seemed to suggest that it was the inversion layer on the surface and also it seemed to depend on the way the surface was treated. The one we used here was anodized, and that seemed to be a way of getting the low resistance in the reverse direction, in other words, a large inversion layer on the surface. And so it was unclear. Holes were obviously going in to the bulk and in the inversion layer because if they're in the bulk, they're even more in the inversion layer. If you increase the concentration in the bulk, you increase them even more in the inversion layer. What you're doing is essentially raising the Fermi level for holes, increasing the level for holes. If you increase them in one place, you increase them in the other. It was not clear at the beginning which effect predominated. And we were thinking more in terms of the surface. But his experiment of John Shive showed that the surface layer was not essential at all, they could go through the bulk. It was a very clear demonstration of that. But I think that that whole program I think perhaps got distorted with the use of the point contacts which was originally introduced just as an easy way of doing the experiment, just to show the effect. A little later we made a transistor and then the emphasis was on trying to improve the point contact transistors, rather than going back to the beginning, and trying to make better devices which we only did much later. Of course, the whole technology had to develop where you diffuse in impurities of appropriate amounts and the growth of single crystals by Gordon Teal was an important growth of single step. The first good junction transistor was the grown junction, in which junctions are introduced as you grow the crystal from the melt. These had good characteristics, but were expensive to make. I think that if the orientation of the program had been otherwise, say following up, more of what was going on in forming and how to do things on the surface, we might have come to surface barrier transistors and integrated circuits and things like that a lot quicker than actually happened.
I would like to ask a question on the earlier experiment in which you were using the deposited gold spots. If the shorting out hadn't taken place, would that experiment have worked the way it was set up?
Oh, it worked, yes, like the present MOS transistors.
So either way, you would have observed transistor action.
Yes, mm [laughter].
Wow [more laughter].
In the patent application in which I probably talked about the point contacts too much, we probably could have gotten broader claims. That was essentially this same geometry, but using an insulator layer with a metal deposited over it to get the field effect. That's essentially what the MOS transistor is. Although, as Shockley points out here, eventually, when we didn't have an inversion layer at the start, by applying a large enough voltage, you could swing the surface over, to make an inversion layer, even though it wasn't present initially. And that didn't come along until quite a few years later.
What did you say the first time the circuit was actually spoken over, on December 23rd? Do you remember?
Brattain says on December 24th, in his notebook, that the circuit was actually spoken over on the 23rd. The notebook entry is on the 24th. He says on the 23rd, it was actually spoken over. "The power gain was of the order of 18 or greater." My question is, what did you say when you spoke over the circuit?
Well, I said quite a few things. I've forgotten what they were. I think Shockley says it's too bad we didn't use the words which Bell used, "Watson, or something like that, are you there?" I wouldn't be surprised if we didn't actually use those words.
It must have been a very heady occasion. Before we close, I wonder if you could comment just a bit about the seven month period between the discovery and the public announcement. This is the period when you were involved with Hart.
I was involved with Hart a good bit of that time.
It must have been very difficult to keep it quiet.
Yes, it was "laboratory confidential", so it couldn't be mentioned outside the immediate group and also, patent attorneys were trying to get -- the original case was re filed, not so much in the disclosure, but to get more comprehensive claims. I think the original one had half a dozen claims. I think there were over 50 in the final version. And I learned a lot about patent law. Hart learned a lot about semiconductors during that time and then in the meantime, Shockley was also getting his patent on the junction transistor.
Was Hart working with Shockley also?
No, another patent attorney was working with Shockley.
In this period, Brattain recalls going to a APS meeting in January and hearing a paper by either Benzer or Brey in New York on the spreading resistance, and having to say nothing because of the secrecy. Were you at that meeting also?
I expect so, I don't remember. I remember the paper on the spreading resistance.
Do you remember who gave it? There were two papers given actually, one by Brey and another by Benzer. I think was the Brey paper.
One of those was in Washington.
Oh, that's right, the Bray paper was in April in Washington. You're absolutely right. So it must have been the Benzer paper in January.
The resistance in the forward direction was ten times lower than was accounted for by spreading resistance.
And you couldn't explain it to him.
As in many cases, when you find an explanation for a new effect, or an explanation for one thing, you can explain other things with it too. Now we were very concerned, particularly with those papers, that they would come up with a discovery of the transistor before we did.
Did you go to Washington?
I think so. I don't remember it now.
Did the Purdue group suspect that ..
... I don't think so. I don't think they had any idea. It came as a complete surprise to them. After the announcement, a lot of people were able to repeat the experiment, within a day or so. Once you have the idea, there's no problem repeating
Were there any hard feelings between the Purdue and the Bell people over the transistor?
I don't think so. If they'd had the idea, there might have been. But they didn't so there wasn't any hard feelings. We tried to give them credit for the work they did which was very fundamental to ours of course, on germanium during the war and after the war.
Do you remember any specific reactions to the discovery at Bell, from Kelly, or Bown or anybody else? After all, this was the fruit of all the planning that Kelly had been making for the past ten years, on getting a solid state program at Bell. And it finally paid off now. He must have been...
Of course, they brought Jack Morton in to head up a device development group and expanded the research. Work expanded quite rapidly. One aspect we haven't discussed was the work of Gordon Teal on single crystals, which was very essential to the subsequent development. Gordon Teal has written a good bit about Shockley was not going to back him in making single crystals because he thought he could make single crystals or cut out small single crystals from the poly crystalline material which were enough for research purposes. But Jack Morton eventually gave him backing but he had to do it essentially on his own, bootlegging it.
OK, well, thank you very very much. This was an extremely good session.