Oral-History:Charles W. Mueller
About Charles W. Mueller
Mueller, a pioneer in solid-state electronics, studied electrical engineering at Notre Dame and received a master's degree in engineering and a doctorate in physics from MIT. Upon receiving his Ph.D., Mueller took a position with RCA's tube department in Harrison, New Jersey.
The interview covers Mueller's work in solid-state technology and the early development of RCA's alloy junction transistor. Mueller discusses how the early transistors were made and the initial efforts in full-scale production. The interview also covers the development of the MOS transistor and integrated MOS arrays, the silicon vidicon, the storage tube, the tunnel diode and S.O.S. (Silicon on Sapphire)
About the Interview
Dr. Charles W. Mueller: An Interview Conducted by Mark Heyer and Al Pinsky, IEEE History Center, July 11, 1975
Interview # 025 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc.
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, 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:
Dr. Charles W. Mueller, an oral history conducted in 1975 by Mark Heyer and Al Pinsky, IEEE History Center, New Brunswick, NJ, USA.
Interview: Dr. Charles W. Mueller
Interviewer: Mark Heyer and Al Pinsky
Place: Princeton, New Jersey
Date: July 11, 1975
Early Transistor Work
Please tell us your name and where you work.
When we talked over the phone, you indicated that you were mostly interested in talking about the transistor.
I think that would be the most exciting thing to talk about. It started in vacuum tubes before solid state was invented, one might say. Even the terminology was all different then. We started studying in the laboratory things we had not learned when we went to school. We organized a course, principally started by Al Rose, and we discussed solid-state theory. It contained new theories that had not been taught when we went to school. The early work was on vacuum tubes, this turned out not to be fairly [?] except for the deflection-type tube, which was an interesting new development. It was fairly expensive. It was used during the war on Navy radar but not used much on anything else.
What kind of tube, exactly?
Beam Deflection Tube
This was a beam deflection tube. It was developed with Ed Herold and Walter and those people.
The work was done at which places?
It was done starting in [Electron Tube Division] Harrison [NJ] and then here in the Princeton Laboratories.
What was it used for?
It was one of the high-frequency mixer tubes at that time. It was for radar input and was developed because of its low noise and high sensitivity. It had a rectangular beam, whereas most tubes had a circular beam, which was deflected on and off a wire. This is what gave the control for detecting signals at the input of radar receivers.
I think the most interesting work was the transistor work. Of course, in those days, no one knew what a transistor was. This is something that present engineers do not at all think of. That is, they have grown up with the transistor, and the idea of being in a world in which the transistor was unknown is, of course, something that most people don't appreciate. Early work was done of course by Shockley, Bardeen, and Brattain at Bell Laboratories. We did some of their experiments and we went into something that is known as "the alloy transistor," which is a thing that RCA developed especially to do the work of bringing it to a commercial production stage. Many people had made transistors, but there was a great deal of argument as to whether they would ever be made good enough to be sold as a commercial device. One could not foresee at that time that there could be something made uniform enough to make one after another by selection, much less what we do today in making 10,000 on a single chip.
Point-Contact and Junction Transistors
Were the first ones point-contact transistors?
The first ones were point-contact transistors, but they were exceedingly noisy. The point-contact was something that was difficult to maintain — the size of the point; the shape of the point was always varying over any load. This made the devices unstable and predicting them or designing them was very difficult. When the idea of the junction transistor was expostulated by Shockley, it was evident that this was the one. If the transistor was going to come — and people were not sure that it was going to come at that time — this would be the one. We did a great deal of development work, especially in the preparation, the etching of the devices, which was the important part of their construction.
We then had a symposium here at the laboratories for about two weeks to a month. These were the days in which the laboratories did a great deal of consulting for industry in general. Our Industry Service Laboratory served as a licensing unit for many different companies. They would come here to learn about new developments, and so a symposium was organized in which we had people who were invited for one-day or two-day periods. We had sessions in the morning and in the afternoon. The morning session would be a discussion of transistors and how they were made. The afternoon session would be on the uses of the devices, that is radios and even television sets.
The first television set was made by [George] Sziklai, Jerry [Gerald] Herzog, and Bob [Robert] Lohman. In those days it had a face about six inches square. The major difficulty was getting transistors that would work in the deflection circuits. We went through, maybe, 500 transistors before we found two of them that would work to deflect the beams sufficiently for this device. Accelerating voltage was maybe one-third or one-fourth of what it was now, and of course the picture size was very small, but this was quite a feat to find transistors that would withstand the high voltage to make these deflections.
<flashmp3>025 - mueller - clip 1.mp3</flashmp3>
About when was this symposium?
Let's see, I don't really remember exactly — 1954, 1955 someplace in there ; it was very early. Our chief advantage was that the alloy transistor method was relatively simple compared to some of the grown-junction techniques. One of our greatest sources of pleasure was to have Bill Shockley here and have him put together a transistor by the alloy technique. His comment was that this was the first time he had really made a transistor and that he was very happy to do so. There was always someone else in the laboratories making transistors with very elaborate machines. In this system that we had, it was relatively straightforward. One took germanium — everything was germanium in those days — put a dot of indium on one side, and heated it. It dissolved some of this germanium. When cooled, it refroze and the germanium would grow as a single crystal with indium doping in there.
We were lucky that many things were in our favor, especially the metallurgy of the system. It took quite a few years to know why and much more work to know exactly why that was a good system, but it just happened to be that we chose the right one. The indium was self-annealing, so there were no strains introduced to the juncture. We didn't know that was important at the time, but were lucky that the indium did not oxidize very easily. Germanium oxide could be removed rather easily, so the transistors worked.
One of the main things of course in those days was to make a transistor that would work in radios. Ed Herold and Doc [Irving] Wolff were running the laboratories in those days and decided that an important part that RCA had to do was to get a transistor in the high-frequency region. A high frequency, at that time, was anything above the audio range. No transistors had been made in any quantities that would operate at the IF frequency for radios — 455 kilocycles.
So, a project was set up. A team was constructed that would try to make the transistors in a way that could be done at the factory. People came down from our factory, which was then at Harrison, to the laboratories; and a group from here was organized, which I led, to try to make this transistor. We had already made some radios and some transistors here. Loy Barton and I went around and gave talks at the Philadelphia IEEE section and the New York IEEE section, on portable transistor radios and on the making of the transistors that went into these radios. The idea was to make something that could be made in the manufacturing plant.
This was very difficult at this time, and the main reason that it was difficult were that there were several ways of making transistors, especially since Bell Laboratories had just announced the transistor tetrode, which is a grown-junction transistor, which worked very well at high frequencies, but which was considerably more complex. The question was, "Should this one be the one we choose to make or should we try to make the alloy transistor?" After a great deal of discussion--one could hardly call it "analysis" at that time--we decided on the alloy transistor, mainly because of the ease with which the various manipulations of making the transistor could be done. It looked to us as if it were something that could be made in large quantities.
The interesting thing about a new device, especially one so completely new as the transistor, is that the manufacturing people do not welcome this type of intrusion into their business easily. Their job is to get out production. Something that is new and untried, and which they think may not succeed, is not something that they welcome with open arms. You have to sell it to them, work very patiently with them until they become interested enough to make it their own project. There were no tools for measuring, no tools for manufacturing, and no know-how as to how this should be done, so everything was to be done from scratch, which, of course is what made it interesting. I can perhaps describe some of the things we did, which are almost laughable by the present standards of technology, but in those days these were things that we could put together to make reasonable quantities in a hurry. One of the first jobs was to make enough transistors so that people could evaluate them and decide whether the transistor was really going to fulfill a need for portable radios.
Do you want to talk a little about some of this?
I think I might describe how the transistor was made. As I said before, it was a germanium wafer with an indium dot on top of it. The indium dot melted into the wafer. The depth in which it melted depended on the size of the indium dot. We made these dots 1/5,000ths in diameter and 1/10,000ths in diameter. At that time, many people said that this was such a small dimension; you would never be able to make anything like this in production. These things are exceedingly big compared with what one does now. But the way we did this is what we called "the cookie cutter-French fry technique." Indium is a very soft metal; it's almost like butter. So you can roll out it rather easily like dough. We took razor blades and stacked them together with different shim stock in between, and then we could roll the indium out to different thicknesses rather easily. Then we could cut them into squares. It was important to control the volume, which we did by hand, controlling the dimensions in all three directions rather quickly. This was done to get a certain quantity of indium. We wanted these to be round, so we took these little squares and heated them up in a column of oil, something like the old-style shot towers. The indium would then melt and drop into a ball because of the surface tension and would settle to the bottom.
The initial ways of grading them was to measure each one separately with a microscope in two directions. One time, we figured it would cost us about five dollars for a pair of these that we used to make a transistor. Of course, the same method was used for quite a long time except we refined the shot-tower technique. Molded indium was injected through a fine hypodermic needle into something and settled out of the bottom and this was then measured with scives. For quite a while this technique was used for making the emitter dots and the collector dots.
After making some of those devices here in the laboratories, maybe 100 or 200, we then went up to the tube plant in Harrison to set up a group to make the transistors. The tube division, while it had, supposedly, the responsibility in making transistors, was not very enthusiastic about this. They had seen devices in the past that were going to replace tubes, but they had never done so and their attitude was that this was just another one of those efforts that was not going to make it. The rest of us were all very enthusiastic, and of course the management in the laboratory here was sure that this was going to work. We went up there to Building 15, it was called, which was shared with some outside concerns, one of which was "Chock Full o'Nuts" [a coffee company], which seemed appropriate considering the state of the art at that time.
We finally got a corner in the laboratory there where we set up our line, which consisted of ten girls doing the job. We wanted things very clean, so they said, "Okay, they would clean the place for us." They got in the janitors and they went through the ducts, which were very big, about four or five feet in diameter, and they scraped the dust off the walls. But then, of course, they never really got it out of the pipes for two months. It just kept draining out of the pipes. Some of it was so big that it was hard to tell the difference between this and the indium dots. Ethel Moonan would look at the shape. If it was round, it was indium and if it was a particular shape, it was dust. Finally we got that dust out of there. The building's supervisor was located in a building three blocks away and to him we were just another room to clean. We really wanted to get better cleaning. After some negotiations, he said he would give us another janitor. Fortunately, I came in one day when this janitor was at work. We were in one corner of the laboratory and he used the push broom to sweep the dust from the whole laboratory down into this corner. Every time he tapped the broom, huge volumes of dust went into the air. Then he would take the dust and dump it into a can in our corner of the room. Nobody told him in which direction to sweep the dust or anything like that. We wanted him to use the vacuum cleaner, but the people that did the cleaning said that this did not fit into their budget, that they couldn't do anything like that. We finally bought a vacuum cleaner on the engineering budget.
Sounds more like the idea of a clean room and a development room.
That's right. Actually, the tube industry was perhaps one of the cleanest industries at that time, but still clean rooms were not something that were available. No one knew any means of measuring dust particles and monitoring them. Even the air conditioning was very poor. The building was, supposedly, air conditioned but for several years after that the tubes suffered from summer sickness. In the summer the humidity would go up and systems couldn't handle it. The number of rejected transistors was always larger then the number in the winter. It took maybe three to five years to get rid of summer sickness. The moisture in the air made the processes difficult — the leakage was high, things of that sort.
Then we had to get the production line going, and that was another interesting problem. We had to decide how one should wed the indium to the germanium, how to keep it clean enough. We did develop a soldering technique with a flux to clean it, but of course not all fluxes worked. We finally found one that did work, but after a few days this was causing troubles. The girls complained the odor was something that they didn't like. One of the fellows had tried putting various harmless perfumes into the flux. Worked for a day or two and then the girls said, "It was just like somebody with B.O. trying to cover it up." Actually, of course, it's quite important, that while people are doing things in which the skill at hand work is important, that they be comfortable.
Finally we learned how to eliminate flux completely, using hydrogen and cleaning and things of that sort. This was a big break for the transistor industry. It kept the girls happy and the yields of good transistors were very much better. Of course, one has to be careful because you never know what is governing the production line. We had here in the laboratory people who were improving the quality of the germanium, especially Fred Rose. He developed a system of growing germanium free from dislocations, free from defects and at first glance everybody said, "Gee! That's wonderful." They began to feed these into our production line, and things got spotty and got worse and worse. It turned out, as we found out — after about two months of blood, sweat, and tears by various people doing different experiments — that the germanium could not be too perfect. If it was too perfect, the indium dot would spread much too much. That is, it would not penetrate, and the area of the dot that would determine the capacitance of the collector would vary so tremendously that it was an accident when it came in right. What we had to do was specify that the material could not be too good and not too bad, and this finally got us through that rather traumatic period.
How long would it take one of these girls on the line to actually assemble a transistor?
They didn't each do the whole transistor but they did parts of it. At the best, ten operators could make maybe 100 transistors a day. Each one would do a different part of the process. Gradually we got more jigs to hold things. Originally, maybe ten or twenty transistors per operator when one operator did all of the work was a good day's work. Then, gradually, the productivity improved as more and more tools were developed, such as jigs to hold things because things were small. One had to pick up everything with tweezers or vacuum chucks and learn how to do this.
Roughly how much were the transistors selling for?
The original ones were very expensive. Maybe, twenty-five dollars a piece, something like that. One had all sorts of transistors. I can remember one time we bought some French transistors through a concern out in Los Angeles [CA]. They had quite an advertising campaign. It turned out there were absolutely no guarantees. If you connected them in a circuit to any voltage, all guarantees were off. They turned out to be very poor, but one never knew exactly what one was getting when you tried to buy transistors in those days. Transistors were only available from our group to various systems groups who then used them to develop portable transistor radios. One of the big problems was setting the limits on the transistor that would be operable in the transistor radios. This took quite a lot of work.
So they could, repeatedly, build radios?
Yes. They had to determine which group of transistors would work in their circuit. They had to make their circuit as broad as possible and we had to adjust our parameters. The adjustment was very loose, of course. The collector was on one side of this disk of germanium — it's all silicon now, but in those days it was germanium. It was about two mils thick and the penetration would be governed by the temperature, which is the dots would go in — would eat into both sides — and dissolve from both sides into this material. We would set the depth by going up in temperature until they would turn sharp together, then we would back off the temperature until we got too poor a yield. Then we would increase the temperature. This was a very rough means of setting the spacing between the emitter and the collector. One could cross-section the devices, but it took a couple of years until we learned how to do that and how to stain them to show up the junctions. Do the measurement techniques that everybody starts out with these days.
I think that is the most interesting part of the alloy transistor. It was developed to the point that, at one time, RCA made about a hundred million of these transistors. This lasted for quite a few years. Gradually, they were replaced by silicon, mainly because the silicon gave better high-temperature performance. There were many different devices in those days, and gradually of course the alloy transistor was replaced by the silicon-diffusion transistor.
Another interesting development was the MOS [metal oxide semiconductor] transistor, which was largely developed in the laboratories here by people such as Fred Heiman and Steve Hofstein, young engineers who worked with me and for me for some of these times. These devices were made and again were only available in the laboratories. They were then given to various circuit people to work out their use in circuits. One needs the stimulus from a use to evaluate a new device, to really see whether it is going to have a place in the industry. One, of course, can and some people do develop a device and then go out and try to sell it.
This is frequently called, "trying to find a problem for which you got the solution." Of course, that sometimes works, but many times it doesn't because, if the systems people do not have a use for the transistor, no matter how elegant it is, it's not going to last very long. Transistor development was tremendous and still is. This occurs because there are tremendous demands. The transistor does things that can't be done in other ways. There are still things that need to be done that even integrated circuits have difficulty doing. It takes this interaction between the people who make the devices and people who use them to really get the developments done rapidly.
What was the bright idea in the MOS? How does that differ from the others?
<flashmp3>025 - mueller - clip 2.mp3</flashmp3>
The MOS transistor is really a transistor in which one uses only one type of carrier, what is called "majority carrier." In the bipolar transistor, say we have a P-N-P transistor, where we inject holes from the emitter into the base. Into the MOS transistor, which is almost a controllable resistor, we send the current along the surface and control the current flowing between two elements called the source and the drain by a metal electrode on the top, which does not contact the main area of the device at all. Through a condenser above an insulator we control the flow of the current in the transistor.
The interesting part about this development is of course that it had many applications that were not foreseen initially. Its chief advantage has been and still is, it's much simpler to make than the bipolar transistor. Where you use large numbers of transistors, they can be made into integrated arrays quite easily. I think the development of these integrated MOS arrays was led by RCA, and was one of the developments that the laboratories can be proud of. People never knew that the device would really make it, and it was a very difficult device because we did not know what to control. The insulator that is used there is only about 1,000 angstroms thick, so you can see that we have come down from the bipolar transistor, which was mils, to making the bipolar transistor with close spacings to a point where we have an insulator that is only 1,000 angstroms thick and which has to withstand ten million volts per centimeter and not be destroyed.
There were some interesting problems with respect to this. Because it was so small and did not have any leakage whatsoever, it was affected by static electricity. The girls that worked on these devices had to be grounded. If they shuffled their feet and picked up one of the devices, the static electricity was large enough to break down the insulation. This was a quite a difficulty for quite a few years. One would put life tests on, for instance, and they would fail in peculiar fashions. It was discovered that every time they took them off the life test, plugged them in the socket, handled leads and carried them across the room, this burnt them out.
Now they have protective diodes in series and in many cases, so this sort of thing can't happen. One also puts them in conducting rubber for storing. You store them in these very high insulating glass wall-type bats. If you put them in for a while and pull them out, you can ruin them. One of the other important developments was to make the oxides clean enough that the devices would not drift. Many of the devices were used initially by an Englishman here. He would describe the tendency for the devices to drift, for them to change when you applied the voltage. He was a cricket player, and he called this "sticky wickets." So, we had a failure mode called "sticky wickets." You applied one voltage and then you applied another voltage — if the results were different, that is if the sequence in which you applied voltage affected the device. We used the term because we thought it was so interesting. Nobody else knew what it meant. I think the MOS transistors had a great development.
Another interesting development was the thyristor. This was developed here more or less by an accidental discovery by Loy Barton. Loy was a farmer by trade and then an inventor, one of these fellow who always tried things. You could never convince him by a theoretical argument that said he should not try something. He was a good engineer, but he would always try everything. We had devices in which the spacings were quite close between the emitter and the collector, and we always warned people not to put too high currents to these things. Loy put lots of current all the time and suddenly discovered that if he did things right, he could control his current. That is, he put on a certain current and the device would switch from high-impedance state to a low-impedance state. If you limited things properly, you had a very good switch. We found this could be done without really destroying the things. If you weren't careful, you switched them over and so much current flowed that it melted them and that was the end of that experiment. But if you did things right, you could keep this. We learned how to control the characteristics of the thyristor and make all sorts of devices. Eventually this was applied by the people in our production plant and in other plants to high-voltage devices. The thyristor of course is a very important, high-current, high-voltage device these days.
Let's go back to the MOS transistor. Maybe you can explain just briefly the differences in the construction between a regular bipolar transistor and a MOS transistor?
The bipolar transistor essentially consists of three types of material: a P-type, an N-type and a P-type material. All are grown in a single crystal. This was the original great contribution of the Bell Laboratories people — to use single crystals. Without single crystals one couldn't make these devices. You injected holes of P-type into a N-type region, and these would traverse across this region. One controlled their flow by the potential of the base. The base was connected in here and always drew current, so it was essentially a low-impedance input device — at least low compared to the MOS. A MOS device is essentially a capacitor input. You put a voltage across it on a metal electrode, which is insulated from the path in which the current flows, and in this way you can control essentially electrostatically the current flow through this area. It is like controlling the current flowing through a very thin wire. So you can have a very high impedance input, 1016 ohms, in that neighborhood. You have a device, which can be controlled by very, very small amounts of power.
This was one of the main advantages of the MOS over the bipolar transistor. That the input impedance is very high immediately gives the designer another tool to work with. This is the type of thing that more closely approaches the tube when people went from thinking about vacuum tubes to transistors. One of their big problems was that the vacuum tubes were high-impedance devices and transistors were low-impedance devices in the input. This caused lots of difficulties, and there were many arguments as to which way one should analyze these circuits, how one should compare them to a vacuum tube. Of course, nowadays no one ever compares the transistor with a vacuum tube anymore. They learn the transistor first and then they go the other way.
They can't even remember what a vacuum tube is!
Yes. One then has a device with a very high impedance. Besides being simpler to make, it has many uses that the bipolar transistor finds difficult to do.
When Hofstein and Heiman set out on this project, they didn't say they are going to make a MOS transistor. What did they think they were going to do?
Actually, the devices were first made by myself and Carl Zaininger at the suggestion of [William] Bill Webster. I think Bill was visiting with some people and they were talking about controlling devices by bipolar and other means. At that time we were trying to make the varactors, i.e. capacitance variation devices, by making an insulator with a metal on top of a semiconductor. We worked perhaps a year, something like that. It turned out never really to be a successful commercial device.
We discussed the possibilities of making two connections on the silicon and making the connection controlling the current flow between them through an oxide. One of the difficulties of course was making things small enough. We were fortunate there again because [unintelligible--Ethel Moonan?] was very skillful with her hands. She was able to make a device that was very small. We hooked these together, put some voltage on them, and found what we called "transistor action." That was the thing that was used in those days when you could see that there was some gain there and that the thing was behaving a little bit different than if it was simply a straight resistor.
I went to Europe after that — I had the RCA fellowship in Zurich. We had just barely shown that control was possible, and by this means. Fred Heiman and Steve Hofstein went to work on it. In about a year and a half they had developed quite a bit of theory of what was actually going on. The first one we made we didn't really know anything about how it was really working. That was quite a good development, but as we learned later, of course, these developments were going on also in several laboratories in addition to RCA. As this happens frequently in these cases, people develop things independently, and its not until they get to a certain point that everyone begins talking about them. I think Hofstein and Heiman's paper published in the IEEE was perhaps the first paper that really discussed any reasonable number of the devices being made. They were at the point that one would expect them some day to be made in engineering quantities, produced in commercial quantities. They did excellent work in producing early ideas for using them, especially for integrated circuits.
Was this roughly in the mid 1960s?
This was about 1963 or 1964. I went to Europe in 1962, and they did the work in 1962. We did the first test in 1961, and they did a lot of work in 1962 and 1963. It took three or four years before they became commercially possible, that is, so people could use them. It turned out computers were one of the chief uses.
I think the development of the silicon vidicon might be worth talking about. The silicon vidicon was announced by Bell Laboratories. It is really something, in which one takes a tremendous number of diodes — about 750,000 of them maybe in one centimeter squared area — and charges them with an electron beam from one side and discharges them with light from the other side. One then uses this signal and passes it into a transmitter. It is one way of picking up a television signal. Bell Telephone Company was interested in this for what they intended to be their Picture-Phone. They did a tremendous amount of work on this, but it turned out not to be economical. It would cost too much for the customers. They had quite a lot of trouble making them. We had various meetings here with our silicon vidicon people, Ralph Simon in particular. Ralph was very interested in having us do some work on silicon. At that time, he and many others had hopes that gallium arsenide was going to be the thing. He was working on that himself, and he wanted us to work on the silicon vidicon.
I think my major contribution was to say that these really weren't diodes, they were MOS devices. They were diodes in a silicon area, but they were isolated by insulator between each diode, so the control of that insulator had to be proper so that the diodes wouldn't become connected together just like in the MOS's. Having realized that, then applying our MOS technique which was done again by Freddy Heiman, we very quickly made some devices that worked — you could pick up a picture.
But the difference between making a device that works and making a usable one is getting rid of all of the defects. I mean an opera singer can't have one defect on her nose while she is singing because the human being and the human eye tend to concentrate on defects. This is something in which you can get all of the information from the picture without worrying about if these defects are there or not. But when you are worrying about a quality for entertainment, it must be absent. It took a great deal of work to get rid of that, and that was started here in the laboratories.
We had a set-up down in the integrated circuit facility that made these vidicons. The structure of these things was quite outstanding in that one took a silicon disk about three-quarters of an inch in diameter and then made about 750,000 to a million diodes on one side of it. From the other side, one thinned this down until it was about a half a mil in diameter, 10 to 50 microns. It was one of those things that we thought we were going to have a great deal of trouble with, but it turned out to be not that difficult at all. The silicon is very, very strong when it's in single crystal form, and it's also so light that you can drop one of these things and it would sort of float down in the air like a snowflake. They don't break. The main thing with the vidicon were the developments at [RCA's Industrial Tube Division] Lancaster [PA] of keeping the number of defects at a very minimum. It had one advantage that we knew from the beginning — it was not affected by high lights. We had quite an interesting project, there, making this for the "man on the moon" project. If you remember on one of those trips, the camera went bad. What happened was that the astronaut inadvertently shined it at that sun, the sun was focused back on the target, it burned a hole in the target and set off the automatic iris, which closed the lens down. There wasn't anything more they could do about it.
That was one of the things that was stimulated from on high. [Robert] Sarnoff, the president of the company, said, "We were the specialists in this field. Why couldn't we make something that really was good and would last?" We knew that the silicon vidicon could do that. However, the ability to make something that was reliable enough in the whole system was a major undertaking. We made the silicon vidicons without any trouble, but the rest of the system was a major effort for the Astro [Electronic] Division. They did a tremendous amount of work on making the system and the cameras. You had to demonstrate the high reliability to get something out of that system. It was a great deal of work, but this was finally done. They used to have a poster on the wall that showed the astronauts on the moon, not worrying about whether they pointed the camera into the sun or not. The camera worked very well.
It is still used quite a bit at the present for surveillance because it is very good in the infrared spectrum. You can pick up with the image intensifier tube, which is really the silicon vidicon used in a slightly different manner. Instead of generating carriers by light, you have another surface in front of it that gives off electrons. These electrons are accelerated and they create additional electrons in the silicon vidicon. This gives a tremendous amplification. With this type of pick-up tube you can go into a dark room, turn on a soldering iron, and have the light from the soldering iron be enough to give a picture inside the room. This is an advantage for many applications where you don't have an intense light source available.
The silicon vidicon work was done about 1970, roughly?
Actually, we started here in the laboratories about 1968. It took about a year and a half here in the laboratories, and it really took about two years at Lancaster.
How about the storage tube?
The storage tube is an interesting development because it resulted from the silicon vidicon. The silicon vidicon has between these diodes an insulator. When this insulator is hit with electrons, it will charge up and affect the current that flows between the diodes. This will stay charged and cause lag. With the vidicon you wanted to be able to change every thirtieth of a second so you could get a continuous motion picture. If there is anything there that doesn't change, you work very hard to get rid of it. That's what we call "lag." Everything else in the vidicon works very hard to get rid of it.
We had an engineer here, [Robert] Bob Silver, who was a very ingenious fellow and interested in many things. He observed this lag and thought he would try to use it. We had some tubes that were a little bit different, which we discussed and which looked as if they would show this lag much worse than others. He connected up these devices. I can still remember when he did it, it was the Friday before the 4th of July. That Friday afternoon, he got everything working and took some nice pictures that just showed that one could store an image which was just a piece of something laid over a white piece of paper, put this on the target, read it out at a later time. One can get possibilities of good resolution. There are many types of storage tubes. Nearly anything that's an insulator will store charge — in effect, an electron beam.
Developments in Silicon Technology
The importance of the silicon vidicon was that again we were making use of the tremendous developments of silicon. What we really use in the silicon storage tube is the silicon oxide. One says, "Why do you need a single piece of silicon when all you need to evaporate is silicon oxide and make a storage tube?" The reason is that the tremendous technology of silicon allows one to make things much more controllable than by any other means. We take a piece of silicon, we have a single crystal, and there are no defects to speak of. We know how to polish it very well. We have done that for all sorts of transistors and integrated circuits, so we know how to make the surface very clean and very pure, how to put it in the furnace and pull it out, and how to grow the oxide to any thickness we want. We can make a very perfect insulator because we are starting with very perfect silicon. This is of course the key to making it. If you evaporate it into things with oxide, it's very hard to evaporate and even the evaporated does not have the same degree of purity and the same consistency. It turns out to that, by using silicon technology, we can make a very difficult process rather simple. The beauty of it is that we have the silicon technology.
The people that do not have the silicon technology never started trying to solve the device in this way. They would use the techniques of trying to evaporate a material, but evaporating a compound is a very difficult thing to do if you want to keep it stoichiometric and pure. This was a very convenient way of making a device. We could fashion it into any form we wanted using the photoresist technology. This is a photoengraving system that one does right on the silicon. RCA has the best one in the business now, and it probably will be used until replaced with charge-coupled devices [CCDs], which are a completely new development. In five years or so this will replace the silicon vidicon, which replaced the electron-evaporated type of structure.
The Tunnel Diode
Do you want to hit the S.O.S.?
I don't know how interesting the tunnel diode is because it has never been the commercial success that many once thought it might be. The original physics of this device was done by Leo Esaki in Japan. It's one of the few devices that has been developed outside of this country initially. He developed the physical principles and found the correct explanations for it; however, he could never really make the device in any large quantities. I remember one time, when he came to this country; he brought his boss along from Japan. He told me he brought him particularly to RCA because he wanted to show him that these devices could be made in quantity.
I think the major contribution there — and it was a very interesting one — was making these things in a reproducible way. [Henry S.] Sommers here did a great deal of work on the measurement, the characterization, and the analysis of what was necessary, and then we developed the ways of alloying them. This was something that developed out of our original knowledge of alloying.
The problem here was to make an alloy that was exceedingly thin — the tunneling occurs only when the junction is very thin, 20 to 50 angstroms, say. One had to form this alloy and essentially freeze it to make it a very thin junction, then reduce it in size by automatic etching methods which we developed but which we later found out many other people had developed. They also thought they were the first ones to do it.
The device is very interesting from the physical point of view. It suffers what many devices do in that it is a two-element device. It's just too hard to use a two-element device in control circuits. You can make many circuits that work, in fact people made complete television sets using only the tunnel diodes. At first it was thought that these were going to be very cheap to make, but we very quickly learned that they were very difficult devices and not at all inexpensive. They were not going to offer any advantages as price was concerned, and they were actually more difficult to use in circuits. They are still used in some special applications, but they never really received large-scale use.
Is it because of the fact there is only two leads of three leads.
Essentially that, yes. One likes to separate an input from an output. When you have the input and the output on the same leads, it's a very hard thing to do. You can do it with lots of impedance, tricks and things, but that's hard. It can be done, but it just causes more external equipment to do that. So, it’s not as economical to do it by this means. It's obvious now why the tunnel diode failed. Of course, it competed with much poorer transistors when it started, and it was possible that it could do things that transistors couldn't do because they couldn't go up to a high enough frequency. When that was overcome, the tunnel diode lost its usefulness.
When did you work on tunnel diodes?
Let's see. In 1966 or 1967 [1961-62]. I have a great deal of difficulty of putting these dates exactly. I usually find later on that they are much earlier. Time goes by.
Silicon-on-Sapphire Technology (SOS)
Will you say a little bit about S.O.S. [silicon on sapphire]?
This is something that is just beginning to become really useful, and it took a really long time. One of the things that one does in making integrated circuits is to have devices and means of insulating them from each other. To do this on a single crystal is rather difficult. One obtains the insulation by putting in an additional junction, say an "N" type material on a "P" type material, but that always introduces additional capacitance. It is obvious to people in the field that one would like to really have individual transistors on a good insulator, so the amount of coupling between the devices is extremely low.
One way to do this is what is called "the hybrid technology" these days. You take a transistor and mount it individually on a piece of ceramic chip, the insulator. The idea we were trying to explore was to grow good silicon on an insulator. This would be a very nice thing to do. It would be rather straightforward if nature or science provided us with an insulator that perfectly matched the silicon crystal. There aren't any that have been found that do this perfectly. It turns out that sapphire, when cut on a peculiar angle, comes fairly close to this.
This is again one of those developments where with a little bit of trouble you can talk yourself completely out of even trying the experiment. No crystallographer would try the experiment because the match is just not good enough that you could hope that the silicon atoms would come down and sit in the right spots so that later ones would again come down and make a single crystal. Usually what one gets when you evaporate on something is a substance that is completely unorganized; that is, the atoms come down randomly like they do in a liquid — it makes the device amorphous. This structure is not useful for any device.
This was a development that was begun by Paul Robinson and myself. I was leader of a group. We had a government contract to explore this field. The big problem was choosing what to work on and what to work with. We talked to lots of people about various chemicals, asking what would be the best way to grow this. Paul was a physical chemist. I was a physicist and a device man. The idea was to try to put all the things together as soon, and as easily, as one could. Strangely enough, we had success rather quickly in the sense that we grew films — although nothing like one can grow today — that were good enough to try to make devices. Other people also began looking at this, but other people generally did not choose the right devices to try it.
Immediately, from our MOS experience, we decided that the MOS was the right device to use here, that is, the bipolar needed minority carrier lifetime and it required much better silicon than the MOS did. We went to [Steven] Hofstein after we had some devices and asked him to reproduce his entire process on the silicon on sapphire. Inside of two weeks, we had some devices that were much better than some people were trying to get, using all sorts of different materials. At that time, it wasn't known whether one should use silicon, cadmium sulfide, cadmium selenide, or any of the large numbers of things that one could evaporate. We had a project on evaporating silicon in addition to trying to grow it, but it went very slowly — the growth was much quicker than the evaporation. We stopped doing evaporation. We had reasonable devices in about 1964 or 1965, but it has taken until now to really get the things where they are commercial. Many people have thought several times that they were commercial. In fact, a couple of our young engineers left and started up a company of their own to make these devices.
Unfortunately, they were about a year or two ahead of their time to make a successful commercial device. To commercialize, so many things have to happen at the same time. You have to have the market, you have to have the system that will pay for the things that you can do better. The number of people that will require these better attributes is not large when you start because art has developed around what is available. So when available devices are cheaper, which is likely, those are the things that you must compete with. And as you are developing your device, all the other devices in the business don't stand still either. They improve, so the targets we were shooting for in 1968 and 1969 all disappeared because the bipolar transistors, MOS transistors, became much better than they were.
There are still quite a few applications for the S.O.S. It's certainly much faster than the bulk devices, but one of the main difficulties has been making it as good as the regular silicon transistor. While people always want higher speed, they are never willing to pay in quality in order to get the higher speed. This has been a development that is still going on there. I have gotten completely out of it. I have been out of it for three or four years. Much of the activity has been in the integrated circuit field because the devices have to be built in such a form that they will fit in with the particular integrated circuit in which they are used.
Is density another factor?
Yes. Where the insulation is necessary, the higher density can be achieved. It has one very great military value: it is very radiation resistant. In fact, this is why the military supported most of the research on it for many years. It has qualities in radiation resistance that can't be done by any other schemes. That's something that will eventually pay off for them.
The military market will fall out eventually to the commercial?
Well, it may if the costs get low enough — and I think they will. There are also certain applications, such as electronic watches, where it can do the job and other devices have great difficulty. It can combine high insulation, high packing density, and low current consumption. Of course, the radiation resistance is something that the military wants for any of its satellites and moving devices.
Mueller's Educational Background
The one thing that we haven't yet discussed is how you got into engineering, and a little bit on your schooling.
I guess I got into engineering because of my high school professor. I went to a very small high school. We had about twenty-five seniors all through the junior and senior year. The science courses were taught by a very good old German professor. He was a very enthusiastic and very good teacher. He made everything very interesting.
What school was that?
In New Athens, Illinois. A small town of about 1,200 or something in southern Illinois. It was a German community, and the professor was German. He was a good scientist, and I think it is because of him that I became interested in science and engineering.
Where did you go after high school?
I did undergraduate work at Notre Dame in South Bend, Indiana. I studied electrical engineering there. When I came out of school in the middle of the Depression, there wasn't much pushing to go to graduate school because the number of the jobs available was almost nil, but I wanted to go to graduate school anyhow. I was interested at that time in MIT because they had publicity as being a very good engineering school, and I went there for a master's degree. I did interesting work under Professor Edgerton on some of the light sources for his high-flash photography.
Then I went to work for Raytheon for a couple of years. I attended some of the electronic conferences at MIT and I soon decided that there were lots of things I ought to still know, so I went back in the physics department under Professor [Wayne B.] Nottingham. That was a great group to work with. He was very stimulating. Each year we would make trips to various laboratories, which was one of the things that really got us all very interested in all of these things that we saw being used. This was something that many physics departments at that time did not do. They were interested in so-called "pure science" and not something that was of use. So I think this visitation to scientists was what made me decide that I wanted to go into an industrial lab such as GE, Bell, or RCA. It was because of B. [Browder] J. Thompson that I applied to RCA for a job. He was the one that hired me in their tube department at Harrison.
That was after you had your doctorate?
That was after I received the doctorate at MIT in physics.
That was a time when physics was really just taking off?
Yes. We had [John C.] Slater there, who was just beginning to open the field of solid-state physics. He was a marvelous graduate teacher but a rather poor undergraduate teacher. He was just too good for the people. I could remember taking a graduate course from him in which there were a dozen professors as students. He finished about five weeks before the end of the course. After about ten minutes, he said, "Well, I've finished the course and I don't know what to do from here. Let's have a vote." People suggested various things and he wrote them out on the blackboard, then we had a vote. "OK!" he says,” start on this." So he started on that and he filled the rest of the blackboard, as if he'd been preparing a lecture for a week ahead of time. This was a great place to study. It was good for graduate students, but it drove the undergraduates up the wall because he was just too good. He never knew when things were difficult or not.
Early Solid-State Theory
Did you have any idea at the time as to what direction things were going? Could you envision effective chip technology?
Yes. We certainly did not see the solid-state as amounting to anything because it was all vacuum technology and thermionic emission study. There were no really solid-state courses that amounted to anything. Slater taught solid-state, but it was very general and everybody was happy that quantum theory was working for the hydrogen atom, that's about as far as we became interested. Because it looked as if this would be a great field for someone that wanted to teach pure physics, but for practical applications it looked as if was something that would take a long time. Of course, it did. It wasn't until after the war and the work at Bell Laboratories that showed what could be done with single crystals. Then the field really took off.
The interest was then in the development of the basic theories in physics, up to quantum theories, that led to the atomic bomb in World War II? A lot of things that we take very much for granted today were developed during the 1910s and 1920s.
Yes. Many of the theories were known, but it was just not possible to make materials good enough as far as the solid state was concerned. In fact, I guess, the MOS transistor to some extent is an example of that. The original idea of controlling current this way was [Oskar] Heil. I think he wrote a patent in the 1930s that pretty well describes the whole thing; except the materials were not capable of doing anything that he thought he wanted to do. He had the whole concept.
I keep wondering if there are theories today that project us that far into the future, and where to look.
I'm sure there are. The question is of knowing where to look and being right. The number of times you are absolutely right on anything like that is pretty small. It's when things begin to develop. There's a great deal of difference between having something that works in the laboratory and something that works commercially and that can be made in commercial quantities at the price people will be willing to pay.
I've always said if you can give enough money to an engineer he can do anything.
Although it doesn't always work that the largest groups do all the development. MOS is a big industry now, but at one time we had one engineer, Steve Hofstein, working on it. We didn't know whether it was going to work or not, but we kept going. You can look at that in two ways: one way you can say, "Gee, you were dumb. You should have put lots of people on it." But we like to say that we were smart. Other people didn't put any people on it.
[Irving Wolff?] mentioned a nice story yesterday about this guy who had a tubeless radio in a black box.
Were you in on that?
Yes. We also had another one. I don't know whether he told you that one or not. This was with the spectra transistors. Right after the war the chief army officer at the Signal Corps [Major General Harry C. Ingles] sent over a transistor to the head of the laboratories here [Elmer Engstrom]. Supposedly, it had been smuggled out of Russia. It was going to be something that no one knew anything about it, it must be a major discovery. We were instructed to examine it very carefully.
We started and couldn't get any current through it. We took X-rays of it, and we couldn't see anything inside it. Then we stopped a little. We weren't going to do anything more unless we were given the authority to go ahead because if you destroy something, you can't examine it anymore. Finally the chief officer in Fort Monmouth gave his permission to go ahead and make any examination we wanted. We made a few more, then we cut it open, and it was just an empty can. I always wonder what the previous history of that sample was because it must have gone through lots of hands. Somebody might have paid a great deal of money to somebody else for it, risked his life for it, and there was absolutely nothing inside of this can. With leads coming out of it, it might have even been prepared by someone for this purpose.
- 1 About Charles W. Mueller
- 2 About the Interview
- 3 Copyright Statement
- 4 Interview
- 4.1 Early Transistor Work
- 4.2 Beam Deflection Tube
- 4.3 Point-Contact and Junction Transistors
- 4.4 Alloy Transistors
- 4.5 Transistor Manufacturing
- 4.6 MOS Transistors
- 4.7 Silicon Vidicon
- 4.8 Storage Tubes
- 4.9 Developments in Silicon Technology
- 4.10 The Tunnel Diode
- 4.11 Silicon-on-Sapphire Technology (SOS)
- 4.12 Mueller's Educational Background
- 4.13 Early Solid-State Theory
- 4.14 Spectra Transistor