Transcript of Interview with Jerry Suran, Jan 26, 2005
Interviewer: Jack Ward, Curator – TransistorMuseum.com
Copyright © 2005 by Jack Ward. All Rights Reserved.
1) When you started with GE E-Lab in 1952, was the unijunction transistor applications work your first assignment?
No. When I joined GE in 1952, our particular group under Dick Shea, was working under a tri-service contract, sponsored jointly by the Air Force, the Army Signal Corps and the Navy. The basic purpose of that contract was, amazingly, to advance the state of the art of transistors. Obviously, the military was quite interested in this new device, and they understood from the start that the transistor was going to revolutionize at least military electronics. They wanted to push as quickly as possible the practical applications. We had a very general contract.
At that time the field had pretty well switched from point contact to junction transistors and GE was at the forefront, and maybe a close second to Bell Telephone Laboratories in the development of junction devices. Part of my job was to understand the performance and characteristics, if not the physics, of transistors. As a matter of fact, the first article I wrote when I was at GE Electronics Laboratories was received in 1953 by the Journal of Applied Physics. It was entitled “The Effect of a Transverse Electric Field on Carrier Diffusion in the Base Region of a Transistor”.
As a result of this theoretical work on field effects in the base region of transistors, I became interested in the possibility of building a transistor tetrode - taking a cue from simulating the vacuum tube tetrodes, and wondering what would happen if one were to apply a field across the base. The thing that crossed my mind at that time was that maybe we could reduce surface recombination and increase the gain of a transistor. And, if it were at all possible to get the field to act transversely, we could even increase the diffusion speed across the base and improve the gain-bandwidth product by that dual effect.
2) At the time you were doing this work, you were using junction transistors?
We were using junction transistors, and at that time, all of the work was in germanium. These (devices) were being built in John Saby’s laboratory in the Electronics Laboratory. That group was just upstairs in the same building. We were the circuits group, under Shea, and Arnie Lesk was in that group, reporting to John Saby. He was the one I was working with and I asked him if he could maybe build some tetrodes for us to experiment with. Just try to put two ohmic contacts across the base of a transistor, and try to build the transistor just the way he would normally build a triode. We were just curious to see how these things would work.
I believe that this particular experiment was run about a year after I cam to the Electronics Laboratory, so it was probably about the middle of 1953, or maybe a little bit later in that year. What happened was that the first couple of tetrodes that we got we could notice very little effect of the electric field. Our theory just wasn’t born out., that this fourth electrode was going to do anything at all. On the other hand, one of those tetrodes curiously had a hysteresis effect on the input, and when we put an oscilloscope on it we found that the thing was oscillating. There was no effect of input voltage or current on the output, so it became apparent quickly that something had happened to the collector contact and that this one had a broken lead. Remember these germanium transistors in those days, the experimental ones that we were building in the laboratory, were put in a little vial, a tiny test tube filled with silicone oil to stabilize the surface. Then it was sealed with a wax seal on the input where the wires came into the transistor. The wax seal was just to stabilize the wires. So, we were experimenting with devices that were built that way, and I guess a collector lead had broken off, and in a place we couldn’t see it.
3) So, at this point, with the collector lead broken, then the two remaining ohmic contacts, along with the remaining emitter lead start to look like a unijunction transistor. Is that right?
That’s how we got the name, “double base diode”. When you measure the dc characteristics of the input, when you tie the two bases together, it was just like a diode, but when you put a field across the two base contacts, this thing oscillated, and it was oscillating with the parasitic capacitance of our instrumentation on the input side.
So, we were quickly then able to determine that we had a device that was very different from what we expected, and it was a serendipity effect because the collector contact had opened up quite unexpectedly – as far as I can recall, that was how the unijunction transistor was discovered.
After that, with the flexibility we had with the tri-service contract, we simply pursued this (technology). We decided that this was an interesting device and we began to explore it and we began to have Saby’s group (my contact there was Arnie Lesk) to build these things purposefully. And I began to write papers on the so-called “double base diode”.
The first published paper I had, according to my records, was in March 1955 and was published in Electronics magazine, and shortly after that, in April 1955, I had one published in the IRE Transactions. These were theoretical and experimental papers, and they were (written) to describe the physics and electronic effect of this unijunction transistor. For example, the first paper in the IRE was entitled “Small Signal Wave Effects in the Double Base Diode”, and “Low Frequency Circuit Theory of the Double Base Diode”, a little bit later. The whole idea was to try to find out how this thing was working, so that it could be reproduced physically in a systematic way.
4) There must have been quite a bit of communication and feedback to the device group as you developed a better understanding of the physics of the device.
Yes, that’s right. As we discovered more interesting features and as we thought we knew a little more about how these things worked, we had Arnie make them a little bit differently. Of course much later, I think around 1957, we switched from germanium to silicon.
5) Did General Electric ever sell any of the germanium unijuinction transistors commercially, before switching to silicon? And what were the first applications?
As far as I know, yes. The unijunction transistor was a interesting device. We immediately recognized that this was a relaxation oscillator, and was the analog of the vacuum tube thyristor. We decided it could be used in any area where one needed a non-linear oscillator. If you wanted a saw-tooth, or you wanted a pulse, or you wanted to use this as a pulse width regenerative amplifier, or you wanted to build a multivibrator, this was good device. It would simplify the fact that you would normally have to use two transistors, and here you only had to use one. It had a built-in positive feedback effect.
6) This is a device that exhibits what is called negative resistance, and I believe this is key to its operation. Is that right?
Yes. When you looked at the input terminals, between the junction and one of the bases, and applied an electric field across the other base contact to ground, what you would have is the equivalent of a relaxation oscillator. It would have a negative resistance , an N-Ttype” negative resistance, which was called a short circuit, unstable resistance, because if you put a capacitor across that kind of a device, it will oscillate. It generates sawtooth waves, square waves, pulses, depending on external circuitry, and you can also use it as a regenerative amplifier. Theoretically, the closer you bias these things to the peak point, the less energy it takes to trigger it, and so you can almost get infinite gain out of these things at the price of stability.
7) The point contact transistor also exhibited negative resistance.
That’s right, but they were much more fragile and much less predicable because of the way they were manufactured, than the junction transistor. And the unijunction transistor was, in fact, a junction transistor, and was not based on point contact technology.
8) I thought I saw a reference to the fact that William Shockley did some research related to the unijunction transistor.
I think he analyzed field effects in semiconductors and decided that you could get this internal feedback, but I don’t think that Bell ever built one. As far as I know, the discovery and development of the unijunction was done solely at the GE E-Labs, as I’ve described it.
Now, our original intent to come up with a better transistor, using an electric field across the base, never panned out. The initial idea was not a failure, and by virtue of a failed contact, we came up with a rather interesting and new device.
I never heard of a (commercial) transistor made by GE that used a fourth contact. Now we put an extra junction on the unijunction transistor to really measure it. We were even thinking about modulating it, maybe having a single device that would act as a mixer between two signals. We put a second junction on it, emitting two different signals into this field. As a matter of fact, this is described in one of the papers I wrote. I was using that second junction as a measuring contact as well as an emitting contact to see if we could build a mixer. It would look like a tetrode but it would have two junctions instead of one, and an electric field across the base. We thought that we could use it in a small signal mode as a mixer. So, in a way, it almost the beginnings of an integrated circuit, but this (device) didn’t pan out either.
9) It seems there was quite a bit of experimentation with different device architectures this time.
Yes, there was much experimentation. What made this possible was that we had this amazing contract that had an objective to advance the state of the art. You don’t get those (contracts) very often. This lasted three or four years, and eventually, I think the Senate interfered with this type of contract and felt that the government was paying for a lot of research that was benefiting corporations instead of the military and they passed a law (I think Proxmire was the architect) to make the contracts more specific to military applications. As I recall, we had quarterly reports, which were fairly big reports, and we also wrote monthly, letter-like reports. The quarterly report was fairly descriptive of all the circuits and all the techniques we were using. One of the things the contract encouraged us to do was to publish. After all, this was in the public domain. The whole idea of this was to accelerate the development of transistor technology, both devices and applications.
Around 1957, the field switched to silicon. Of course, again, this was accelerated by military contracts because germanium was just not adequate for many military applications - they need to operate at higher, ambient temperatures.
10) Was the military interested in unijunction transistors?
Yes, they were. As a matter of fact, we used one in an Apollo simulator later on to generate a very linear sawtooth wave for a TV output. Anyway, the types of applications were just exactly what you would use a thyratron for, in the vacuum tube days. Anything that a relaxation oscillator could do, the unijunctuion could do. Also, with the advent of digital electronics, we thought these could be used for clocks in computers.
Eventually, the silicon controlled rectifier put it out of business, because the SCR could operate at a much higher power level. Initially, they were used together, with the unijunction used at low power levels for triggering. This was a combination that interested the power companies a great deal, and that is one of the reasons, I think, that GE jumped on the unijunction transistor, and became the leading manufacturer of the device. The unijunction was big seller in those days and it made a lot of money for GE.
11) Were there any surprises when moving from germanium to silicon with the unjunction?
The devices acted exactly the same. Silicon at that time had much higher surface recombination problems. The transistor gain-bandwidth initially in silicon were not as good as in germanium. Of course, the mobility of silicon was lower. Silicon caught up as the processing developed to planar technology, using photolithography on a wafer. Silicon pretty much made that possible, because if you oxidize a silicon surface it becomes glass, and you could easily deposit photolithographic materials on glass.
Up till that point, silicon could operate at higher temperatures and so could operate at higher power levels. We switched the unijunction over to silicon from germanium primarily to increase the power. Of course, the military was very interested in getting these devices to operate at a higher ambient temperature.
12) How did the original name for this device, the double base diode, change to the name “unijunction transistor”?
That’s an interesting story. For a while, we called the unijunction transistor a double base diode because that’s the way it was discovered. We thought we had a tetrode, but in fact we had a diode with two ohmic contacts, and so it was a “double base diode”. That was descriptive term. But the diode terminology was a misnomer because it was a three terminal device, and not a two terminal device, and it had gain – it was an active device and not a passive device like a diode.
I think around 1956, I was on an IRE standards committee, or actually, on the IRE circuits committee – 410, the transistor circuits subcommittee. And most of the people on that subcommittee objected to the terminology, and were trying to standardize terminology. We all knew that this thing was a transistor, and not a diode. So, with pressure from the IRE, I proposed that we call it a single junction transistor, which obviously then became a “unijunction transistor”, and it was a unijunction transistor but that was also descriptive as well as correct , because a transistor was a gain device and this thing had gain. So I think with IRE standardization pressure, we switched the name to unijunction transistor, and there was no resistance within GE or any other company to make that switch.
The interesting thing about that is when we came to the same conclusion with the silicon controlled rectifier - the silicon controlled rectifier was a three junction device with three terminals, but it was also an active device, (the circuit model was really two complementary transistors in tandem). It was a four layer device and it certainly was not a rectifier, because a rectifier in those days was a diode – a two terminal device.
We proposed in the IRE that the silicon controlled rectifier be called a “hook collector transistor”. The hook collector part of it came from the profile of the voltage across the junction - there was a hook in the profile, across the third junction. But, within both Westinghouse and GE, there would have been a turf battle if the terminology “transistor” were used. We had two operating divisions within the company, and Westinghouse the same I believe – one (division) which was manufacturing rectifiers for thw power industry and one which was manufacturing transistors for the electronics industry. The people on the power side recognized that the SCR could eventually replace their vacuum tube circuits and would be a very valuable device to be used in power circuits and they wanted control of it. So, it was an internal turf battle that finally scuttled the IRE recommendation that it be called a transistor instead of a rectifier. In those days, Westinghouse and GE had almost 100% of the power device market. This is an interesting story about why the “double base diode” went to “unijunction transistor” but “silicon controlled rectifier” never did.
13) Was the circuit symbol for the unijunction transistor always the same?
The symbol was always the same – it was a single junction represented by an arrow into a bar, and then two ohmic contacts. Actually, the IRE came up with the standardization for symbol for the transistor; you remember that the arrow pointing down into the base region, and then an ohmic contact just being a line connecting to that base region. So, if you just put two lines and take that collector contact away, you would be perfectly consistent with the symbol of the IRE, and that’s what we did with the unijunction transistor. This was part of the standardization process.
14) I know you did pioneering work while at E-Labs on developing a transistorized pacemaker. Did this device use a unijunction transistor?
No. What we used for the pacemaker device was a complementary transistor circuit using silicon junction transistors. These were selected on the basis of the transistors that had been qualified for use I the U.S. missile program. They had undergone, by the Army, tremendous reliability testing. This was in the late 1950s. One was an NPN and the other was a PNP. We connected them in a configuration that was the equivalent circuit model of the silicon controlled rectifier. This device though would have a negative resistance by virtue of external feedback, and you could control the negative resistance by the external circuit elements.
If we had used the unijunction transistor, the negative resistance characteristics of the transistor was pretty dependent upon the physics of the internal device and the peak point and valley point of the negative resistance were controlled by device characteristics. And the internal device characteristics could not be as controllable as a device that was controlled by external circuit elements, like resistors.
If we had used the unijunction transistor, we would have had to have greater tolerance on where that peak point stabilizing resistor was - in other words, we would have needed to use more power, and this would have shortened the life of the pacemaker. We wanted to reduce the tolerance to where we could run the pacemaker without replacing the battery for about four to five years. If we had used the unijunction, because of tolerance problems, the life would have be two to three years. So we used a two transistor, negative resistance circuit in the GE pacemaker.
15) You never considered germanium transistors for the pacemaker, even though the operating voltage for the germanium devices was lower than for silicon?
No we didn’t because we wanted this to be as reliable as possible, and we were at that time basing reliability on accelerated life testing, which meant raising the temperature, and even though the pacemaker would operate internally in the body, with a homeostasis temperature stabilization built in at about 98 degrees Fahrenheit, and germanium would have operated quite well there – but we felt that with the reliability testing that the Army, Navy and Air Force were doing on silicon, and with the resultant reliability data that we had, we needed to use silicon.
We wanted to use as high a voltage as possible for the pacemaker because we had to deliver a fixed energy pulse to the heart. The capacitor was the storage device for the energy, and the higher the voltage we could use, the smaller the capacitor could be, and hence the physical size of the device could be reduced. Therefore, the function voltage difference between germanium and silicon was not a factor. I think we used four 1.5v batteries for the pacemaker, and that constituted about 85% of the volume of the device.
We built these originally in the laboratory, and these were used in clinical trials. The same device was passed over to the Medical Systems Division of GE and put into manufacture. They used the same circuit we did. Several hundred thousand GE pacemakers were eventually implanted.
16) Were patents generated for the unijunction transistor?
Yes. I can provide a list of some of the patents related to the work on unijunctions at GE E-Lab.
17) One last comment. I recently did an internet search and came across an article describing a recent reunion of E-Labs personnel, with a reference to your name and also some early transistor work there. How did this recent event come about?
What happened was that I had an appointment at Syracuse University and I emailed a couple of my old colleagues at GE – in particular, my ex-secretary, and asked her if she could maybe arrange a luncheon with some of the folks who, were still around from our era. She got together with a couple of other people decide, “Why don’t we make this a big E-Lab reunion?”. What I had anticipated was that I might have lunch with six people, and she began to look through the records and try to find every E-Lab employee they could. This turned out to be a 300 person party! There were quite a few of our people, such as Addision Sheckler and Art Stern, who were in the Laboratory in 1951-53 when I was there. From 1972 to 1979, I was managing the Electronics Laboratory – so practically, my whole staff showed up.
18) Thank you, Jerry, for agreeing to this interview. I’ll follow up with an update to your Oral History at http://www.transistormuseum.com
and will be in touch when this is completed. I’m pleased to be able to document this important semiconductor work done in the early 1950s at the GE Electronics Laboratory.