RCA Transistor History

 

Copyright 2008 Mark P D Burgess

 

The Radio Corporation of America (RCA) was formed in 1919 by General Electric, AT & T and United Fruit with the encouragement of the US Federal Government which wanted to maintain US leadership in long distance communications. During World War I General Electric’s Alexanderson wireless transmitters and American Marconi’s shore stations were regarded as strategic assets having proved their value when Germany cut the trans-Atlantic telegraph cables and wireless was the only alternative. Post war American Marconi was bought by General Electric and transferred to RCA.  General Electric and Westinghouse made RCA their exclusive marketing channel for receivers and tubes in return for cross licenses to their patents.

The alliance was disestablished in 1932 following successful Federal antitrust actions against RCA and General Electric.

In 1941 RCA had decided to bring its R&D on to a single 260 acre site it purchased near Princeton University. The chosen site was close to its manufacturing locations at Harrison and Camden and was opened in 1942 with a staff of 125 engineers and scientists. Early on its programmes were dominated by war-time military contracts which did not necessarily relate to RCA manufacturing. Engineers worked on radar antennas, phosphors for radar screens, acoustic fuses for anti-submarine munitions, navigation, infrared cameras and microwave communications as well as television, an important consumer product for RCA.

Post war the laboratory research programme needed radical redirection in order support the innovation demanded by RCA Chairman, David Sarnoff. He told the Radio Manufacturers Association in 1947:

“The industry does not pick up where it left off before the war... The radio manufacturer is the logical producer of radio-heating equipment, radar, loran, shoran, teleran, and hundreds of allied radio-electronic devices. He must push on to new ventures. To be successful he must not only manufacture, but he must encourage research to create new methods, new devices, new services.” [cited by Kilbon 1964]

Jack Saddler who worked in sales and marketing roles at RCA from 1952-62 commented in a retrospective from 1990:

“The company was in 1950 the dominant maker of vacuum tubes in the world. Add to that the fact that General David Sarnoff, president, board chairman, and spiritual leader was a technical visionary who could smell innovation and you will find the climate for invention. Witness the fact that RCA was very much present in global commercial record communications, sound on film for movies, television, color television, scanning electron microscopy and business oriented computers. It isn't too difficult to see that RCA had to get into the semiconductor business.” [Saddler 1990]

But to do so RCA had to solve a dearth of basic knowledge in the emerging field of semiconductor materials science not having engaged in the war time semiconductor diode programme which gave other corporates such as Bell, General Electric and Westinghouse an advantage.

Post war the RCA Laboratories were reorganised and expanded working on consumer products such as color television, hi-fi audio, computers and components such as transistors, lasers, integrated circuits and advanced vacuum tubes. By the end of 1947 it had eight sections:

 Radio Systems Research (Beverage)
 Radio Tube Research (Wolff)
 Electronic Research  (Zworykin);
 Acoustical Research (Olson)
 Physical Research (Smith);
 Contract Projects (Holmes)
 Industry Service (Seeley)
 Laboratory Services (Nelson) [Kilbon 1964]

 

Prelude

Edward Herold then researcher at RCA Laboratories read the announcement of the Bell transistor in the New York Times and noted in his laboratory notebook:

“For many years I have been interested in the possibility of making a three-element amplifying crystal which bears the same relationship to the silicon and germanium diode as does the triode to the Fleming thermionic diode...An announcement in yesterday’s newspapers indicates that Bell Telephone Laboratories have indeed developed such a device, and in simpler form than I had believed workable.” [cited from Choi 2007]

Early in his career Herold had worked for Bell Laboratories, but mid career enrolled at the University of Virginia and after graduating joined RCA as a development engineer at Harrison New Jersey in 1930. After earning a post-graduate degree in 1941 RCA transferred him to its new laboratory at Princeton NJ in 1942 where he worked on tube microwave problems. After the war he was appointed head of the RCA tube research group. [IEEE 1976]

The RCA Laboratories had indicated in its Annual Research Report for 1947 that solid state research should be on its long term agenda. [Choi 2007] Neither Herold’s note nor this agenda was very farsighted: Eight years earlier William Shockley had noted in his laboratory book in December 1939, after reviewing the work of Mott, Schottky and Wilson: “It has today occurred to me that an amplifier using semiconductors rather than a vacuum is in principle possible.” [cited by Riordan 1997] And these were not isolated musings: Shockley was following the strategy of  Bell Laboratories Director, Dr Mervin Kelly to find an electronic switch for use in communications.

This is a repeated theme. Frederick Seitz tells an anecdote relating to comments by Karl Lark-Horovitz who led the Purdue research programme in radar diode germanium and silicon and who supplied the germanium that was used at Bell Laboratories for the first point-contact transistor: “In 1947, after Bardeen and Walter Brattain had discovered transistor action at Bell Laboratories using a point-contact electrode with germanium, and while their success was being kept confidential during the filing of patents, they visited Purdue as guests of the physics department there. Lark-Horovitz, still deeply involved in the study of germanium, said to them in effect: "There must be some way in which we can make a triode from these semiconductors. Do you have any suggestions?" [Seitz F 1995]

Thus in the post wartime era RCA moved to re-align its research from military contract research to its own needs. The Laboratory noted in its 1948 Annual Report that it had launched eight applied research projects of which three were directed to “crystal amplifiers.”

 

 

RCA Point-Contact Transistor Development

The head of Radio Tube Research, Irving Wolff was initially responsible for the transistor research projects. He appointed Herold to set up a small group to carry out the research. A key person in Herold’s team was Jerome Kurshan who was recruited from the tube Research Group. Kurshan had been working for RCA since receiving his PhD from Cornell in 1943. Kurshan recalls the announcement of the Bell Laboratories point-contact transistor a week after the annual Electron Devices conference (a closed industry conference). “On news of the transistor, I dropped that work and immediately switched to solid state research.” [Ward 2001]

Kurshan began by trying to recreate the Bell Laboratories point-contact transistor. Herold recalls “We already had set up equipment for measuring amplification in any new type of device and it took Dr. [Jerome] Kurshan of my group only one day to cut open a germanium diode, add a second point-contact and confirm the Bell Laboratories result.” [Herold 1983]

In order to achieve precise positioning of the two contacts Kurshan used micro-manipulators more commonly used in microscopy and also reminiscent of the of the turn of the century cats whisker detectors made by the Wireless Specialty Apparatus Company

These enabled him to search for an active spot on the germanium and to bring the wires close enough together to obtain measurable gain. In this way Kurshan produced the first RCA “transistor” in July 1948 although given the bulk of the apparatus this might be more accurately described as a demonstration of the transistor effect.

In the following six months Kurshan and two production engineers assigned to the Laboratory worked on optimising a design for a point-contact transistor. This was an empirical approach: Kurshan and his co-workers tried to understand the effects of modifying designs and materials that they could control. (For example, preparation of the germanium and its surface treatment, spacing of the point-contacts, the shape of the contacts, the conductivity of the germanium, encapsulation and the effects of heat and humidity. [Ward 2005] The experience was frustrating since in the absence of any theoretical understandings they had no conceptual framework. [Choi 2007]

Bernard Slade started as an Engineer at the Harrison plant in July 1948 and was assigned to work on semiconductors in conjunction with the Kurshan and the scientists at the Princeton Laboratories. Slade made experimental transistors in cooperation with Princeton.  “The first task was to read every bit of published information on the subject. This didn’t take very long because at that time there was very little available. However, Bell Laboratories generously shared much of what it had learned, and this information was very valuable. I formed a small laboratory with one technician and fairly rudimentary equipment for assembling and measuring the devices I planned to fabricate. The equipment consisted of tweezers, a welder to attach the tiny wires, a hot plate and soldering iron, a microscope, and some pots and pans for etching and washing the germanium crystals.” [Ward 2005]

Dick Endres started at Princeton in 1948 in applications and was therefore the recipient of the new Bell Laboratories transistors:  “Because of RCA's close patent license tie-in with BTL, we had access to some of the first transistors made available by them.  At first, each transistor which they shipped to us came complete with its measured performance parameters, and I came to know most of them personally - at least, so long as they lived.” [Ward 2003]

 

The Bell Laboratories Transistor License

There were three powerful imperatives that influenced Bell Laboratories’ approach to licensing other manufacturers.

(1)  Its tradition of cross licensing: trading patent rights across broad fields on a non-exclusive basis with other major entities such as RCA;

(2) The military importance of a new disruptive technology promised by semiconductors; and

(3) The competition law environment in the United States prosecuted by the US Department of Justice.

Given the context of the times: the Cold War and the Korean War, Bell Laboratories was obliged to work with the Military with respect to US national interests in regard to both the point-contact transistor and, in 1951, the junction transistor. It was agreed not to classify these transistors and that the Military would be best served by accelerating applications research. However Bell Laboratories was enjoined “to guard the special manufacturing processes so essential to the success of the transistor development and production with all possible care short of actual military classification.” [cited by Riordan 1997]

Thus Bell Laboratories held a symposium on transistor applications in September 1951 that was co-sponsored by the Military for some 300 government or institutional engineers and scientists with military clearance.

In 1949 the US Department of Justice initiated an antitrust suit against AT&T (parent of Bell Laboratories) alleging an unjustified monopoly in telephone equipment and telephone services and seeking to separate its manufacturing arm (Western Electric) from AT&T telephone services. Prosecution of the suit was delayed by the Korean War. In the Cold War era that followed AT&T was seen as a strategic asset and the final consent decree (1956) provided that AT&T would be restricted to telephone services, could retain Western Electric but was obliged to license all its existing patents at no charge and all future patents at reasonable rates.

Thus while it is clear that most of the developments at issue here took place prior to the consent decree Bell Laboratories must have considered strategies to defend the suit from 1949 onwards: a decision that led it to anticipate the consent decree by making its transistor patents available to interested parties from 1951.

Initially companies such as RCA and General Electric set out to develop their own transistor technologies taking care not to infringe Bell Laboratories. For example, John Saby, inventor of the alloy junction transistor at General Electric comments in his oral history:

“For a long time when we first started, we didn’t say we had made a transistor because the Bell Labs people invented the transistor, and the legal people in GE felt that if we used the word transistor we would be automatically assuming liability under some patents that we wouldn’t otherwise be assuming liability under.” [Morton 2000]

In November 1951 Bell Laboratories ran a second symposium directed to US companies and again devoted to transistor applications.

Participants at the 1951 symposia received a substantial volume from Bell Laboratories which stated in its preface: “In this volume, as in the symposium, the treatment follows the general sequence of: theory, transistor properties, circuit design principles, applications and, finally characteristics of the different transistor types now under development.”  [Bell 1951] What is conspicuously missing is any discussion of transistor manufacturing technologies of value to the licensees. Saby confirms the sentiments of Bell Laboratories’ licensees around the poor availability of manufacturing know-how:

“In crystal growing, for example, Gordon Teal wrote papers on crystal growing, but never disclosed a lot of the details of the process to get the crystals to grow. People who grew crystals generally had to discover themselves, and people in academia were teed off by this because Bell would print all these things, but they didn’t really tell you how to make crystals that you could perform independent research on, unless you got down on your knees and ask them for a piece of crystal.” [Morton 2000]

Stuart Seeley, head of RCA’s Industrial Service Laboratory had no kinder views towards Bell Laboratories and its 1951 symposium: “[They] brought the group in, they sat them down in some hard chairs, and began throwing theory at them until they were just too fatigued to listen to any more.” [cited by Choi 2007]

Western Electric began offering licenses to the Bell Laboratories transistor patents after the Shockley junction transistor patent issued in September 1951.  By agreement with the Military, these were restricted to NATO countries. Licensees paid a $25,000 upfront deductible against future royalties and were entitled to attend the third symposium in April 1952.

RCA took a license to Bell Laboratories’ transistor patents as did many transistor manufacturers [Ward 2001, Choi 2007] but it is not clear when this was. Early on the inferences are that RCA set out to invent around the Bell Laboratories’ patents although at first it set out to understand what Bell Laboratories had achieved.

The 1952 Bell Laboratories’ Symposium covered semiconductor materials science,  transistor fabrication and testing. Participants received a two volume reference set entitled “Transistor Technology” that was classified and disclosed production principles that had been missing earlier. Its contents were: Technology of Germanium Materials; Preparation of Single Crystals; Principles of Device Fabrication; Principles of Device Characterization; Design for Manufacture; A Manufacturing Procedure. [Bell 1952]

But by 1952 RCA had solved the key production issues for the production of viable point-contact transistors and was in development of an alloy junction transistor which Bell Laboratories did not have. The 1952 symposium may have had value in regard to semiconductor purification although this remained an issue for RCA through to the mid 1950s.

However, in 1948, RCA had been obliged to recreate Bell Laboratories’ knowhow and develop manufacturing processes. Examples follow in relation to:

(1) Germanium materials technology

(2) Electrode geometry

(3) High frequency performance

(4) Forming of point-contacts

The work described below was carried out in the period 1948 to 1951 although RCA did not begin full commercial production of point-contact transistors until May 1953 (when it also launched its alloy diffusion transistor).

Germanium Materials Technology

Performance of the early point-contact transistors was variable. One reason for this related to the quality of the germanium: small pieces were cut from a polycrystalline ingot and consequently there was variability from one piece to the next.

RCA had no experience in the production of germanium of adequate purity. Late in 1948 a delegation led by the Radio Tube Research section manager Irving Wolff visited Lark-Horovitz at Purdue University and arranged for the supply of germanium samples. [Choi 2007]

Purdue was a very appropriate institution to seek advice from. Since March 1942 they had worked on germanium for point-contact radar diodes and were part of a major war time effort coordinated by the Radiation Laboratory at MIT. Participating institutions working on radar detectors met every two months throughout the war, most often the Radiation Laboratory, the University of Pennsylvania, Purdue and General Electric with less frequent participation by the Naval Research Laboratory, Bell Laboratories, Westinghouse, Du Pont and Carnegie Tech. Thus Purdue was kept abreast of developments by others in the field.

Collectively the programme worked on high purity silicon and germanium and its doping. Lark-Horovitz at Purdue led work on high purity germanium and dopants identifying the importance of the Group III and Group V elements. They showed that germanium doped with either phosphorus or antimony was an effective detector. The Purdue programme produced a high back voltage doped germanium point-contact diode that was commercialised by Bell Laboratories and Western Electric, discovered the value of formed point-contacts (where the point-contacts are fused to the germanium by current pulses) and made numerous theoretical advances. The work at Purdue on germanium informed the work at Bell Laboratories that led to their point-contact germanium transistor. [Petritz  1962, Seitz 1995, Henriksen 1987]

RCA picked up information where they could including valuable hints through informal interactions at conferences. Choi notes an example where RCA technologist, Schuyler Christian reported his meeting with Shockley at the National Research Council conference in November 1950. He recorded in his notebook that Shockley had advised “one will not obtain uniform, consistent, reproducible transistor action except with single-crystal germanium, because of random grain-boundary effects in the ordinary ingot.” [cited in Choi 2007]

Electrode Geometry

George Rose worked on the problem of how to keep the emitter and collector electrodes closely aligned at the desired spacing without the need for post assembly adjustment. Western Electric, for example, produced its A1698 point-contact transistor with a port that permitted access to the electrodes once the transistor had been assembled, filling the space with wax and protecting it with a plastic sleeve. Apart from the additional complexities of manufacture, this left the transistor vulnerable to environmental effects that plagued all manufacturers of germanium devices until they introduced hermetic sealing.

Rose points out the nature of the problem: “The emitter and collector electrodes may consist of point electrodes portions which are spaced apart a predetermined distance. The spacing, generally two mils or less, between point electrodes is critical and determines to a large extent the gain and other characteristics of the device. It is accordingly important that that this distance be maintained at a desired value. In the past this has been a difficult problem.”




Because the base was usually polished or etched the electrodes could easily move across its surface on assembly. Rose developed the electrode geometry shown in the drawing and this is characteristic of RCA point-contact transistors in production. In his patent application he describes how the base may be mounted on a plunger that is moved against the point-contacts until the desired contact pressure has been achieved. He also describes methods of treating the points by heating to oxidise them and etching off excess oxide to leave a sharp point. A patent was filed by Rose in 1949 covering this work. [Rose 1949] Photo of an early point-contact transistor courtesy of R McGarrah.

 


 

High Frequency Performance

The point-contact transistor maintained a niche due to its superior radio frequency performance even after the advent of the alloy junction transistor. Bernard Slade worked on high frequency point-contact transistors.  “At the early stage of transistor development, the high frequency characteristics of specially designed point-contact transistors were used in the experimental high frequency applications such as oscillators at frequencies as high as  300 megacycles, and TV video amplifiers at 20 megacycles, speeds which, by today’s standards, seem unusually slow, but in 1952, virtually revolutionary.” [Ward 2005] The most important determinant of high frequency performance was the emitter-collector spacing: the transit time of hole carriers from emitter to collector is proportional to the cube of their separation. Minimising emitter-collector separation leads to improved high frequency performance but at the expense of stability which must be corrected by increasing germanium conductivity. The curve,  indicates that point spacing of down to one mil gives an alpha cut-off of 20MHz. [Slade 1952]

Formed Point-Contacts

In 1950 Bardeen and Pfann summarised work at Bell Laboratories in the previous year reporting that the performance of N-type point-contact transistors could be improved by passing large reverse currents through the collector contact. Conversely P-type transistors could be improved by passing a large forward current through the emitter. They suggested that the effect was due to migration of donor or acceptor ions from the body germanium to the surface modifying the rectifying barrier at the point-contact junction. [Bardeen 1950]

A year later Pfann revised this view proposing that forming could involve the migration of donors or acceptors from the metal contact demonstrating the effect by using electrodes with a range of low levels of antimony. He showed that increasing the antimony content improved the current gain of the transistor. [Pfann 1951]

At RCA Stelmak produced similar results for phosphor bronze contacts containing varying levels of phosphorous. His results were more quantitative showing forming produced power gains of an additional 10 db with up to 0.1% phosphorous (over “pure” copper).

Stelmak was able to conclude that “In view of these results, pulsing response appears to increase with phosphorus concentration in the contact material and with pulse power input, and, as such, is a function of the heat generated at the point-contact. Some fusion at the metal-germanium junction is observable under a microscope after successful forming. A superficial diffusion of phosphorus, or other donor, into the germanium surface under the contact point may account for the change in the height of the potential barrier observed to give increased power and current gain.” [Stelmak 1951]

Point-Contact Production Prototypes

Each prototype production lot was given a TA (Transistor Amplifier) number and numbers from TA-150 onwards were reserved for transistors. Early transistors were inscribed with its TA number, date and production number by hand. The following data has been drawn from a compilation from the records of RCA by Joe Knight.

In the period to the first commercial transistor release in May 1953 Knight has reported 22 point-contact TA types. A further 23 were made in the period following this to March 1954. The table is a summary of highlights only. [Knight 2008] Pictured is a TA-151C made in December 1949 at Harrison. [Courtesty J Knight]

 

Date

Number

Description

November 1948

TA-150

First lot of 200 made at Princeton.

December 1948

TA-151

First lot made at the Harrison plant.

Both lots were made from Western Electric diode germanium pellets

August 1949

TA-161

Large pre-production runs. Variants “A” and “B” were made through 1950. Glass base

January 1950

TA-165

Prototype of the 2N32 which was commercialised May 1953. Medium frequency

June 1952

TA-166

RF amplifier (for example used in the Symposium portable TV)

July 1952

TA-172

Prototype of the 2N33 which was commercialised May 1953. High frequency oscillator (used in the Symposium portable TV)

March 1954

TA-178

Prototype of the 2N72 switching type which was developed on a Signal Corps Contract.  

 

Jack Saddler, who had a sales and marketing role at RCA,  recalls details of the forming process and the choice of materials for the electrode ropes after he joined the company in 1952: “These wires were made of material which would provide dopant when fused to the base. My recollection was that the wires were phosphor-bronze. The wire furnished the required n+ material for compensating the boron (p) doping of the base material. The fusing of the wires was accomplished by discharge of a capacitor between emitter and base then collector and base. The machine which formed these junctions was then used to check DC forward current gain (beta). While touring the operation one day I watched the junction formation operators work. Beside each machine were two plastic buckets. The reject container was much larger than that for good parts. The potential transistors were encapsulated in clear plastic before junction formation. Apparently, the importance of light as a carrier generator wasn't of concern at that time. It wasn't long before RCA stopped making germanium transistors that way.” [Saddler 1990]

 

The RCA Symposium: Technology Transfer

Herold began planning a transistor symposium at the time of the last of the three Bell Laboratories Symposia in April 1952 in order to keep its licensees and customers informed and showcase RCA’s developments. Although RCA had rights to license others under Bell patents, our licensing management was eager to show RCA’s own prowess in this exciting new field. We were asked to organize a major symposium in the fall, to which the hundreds of RCA licensees would be invited. As the producer of the key ingredient, the transistor, my group and I became directly involved.” [Herold 1983]

The Symposium was organised by Industrial Service Laboratory which was responsible for licensing and held at Princeton from November 18th 1952. Unlike Bell Laboratories (prior to 1952) RCA did not have an ambivalent attitude to its licensees: the laboratories were supported by royalties and RCA wanted its licensees to be successful. There was no “throwing theory” at the attendees: For example Choi cites Mueller’s explanation of how a transistor works drawing on the general understanding of tube technology: “This emitter is like a cathode of a vacuum tube, to inject carriers into a region where there weren’t any before. The collector here is like a large basket to collect more easily all the holes that are put in by the emitter” [Choi 2007].

Bell Laboratories had made no such concessions at its 1952 Symposium recording in the Preface of Transistor Technology: “It assumes on the part of the reader a familiarity with the theory of conductivity in semiconductors and the functioning of transistors, such as presented in Dr. William Shockley’s book “Electron and Holes in Semiconductors” as well as with certain other material relating to circuit properties and application of transistors, as embodied in the volume The Transistor – Selected Reference Material on Characteristics and Applications.” [Bell 1952]

“The November 1952 demonstrations were an outstanding success. We demonstrated 24 different applications, covering just about every conceivable type of electronic device using transistors, even including the first all-transistor television set (except for the picture tube). For our licensees in the device field, we even set up a small transistor production line to show exactly how they were made. I think it’s fair to say that, in practical application of transistors, we were far ahead of anyone else in the field.” [Herold 1983]

The production line consisted of three operators. The first carried out acid etching of the wafers; the second applied the indium dots and loaded them into a furnace and the third attached the leads. Transistor royalty attended as Mueller recalls. “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.”  [Heyer 1975]

Star of the Symposium was a portable transistor television with a five inch screen which is an icon in the development of advanced transistor applications now on display at the David Sarnoff Library. George Sziklai a senior applications engineer at Princeton thought it would be feasible. A colleague, Gerald Herzog recalls “George proposed building a television set using transistors as a demonstration of what might be possible in the future.  However, he met a lot of resistance because, after all, no transistors could operate anywhere near the frequencies that were necessary for a regular TV set.   However, George persevered and the project was approved.” [Ward 2004]

Herzog worked on the RF sections while George Sziklai and Bob Lohman worked on the complementary audio sections. The RF sections were built with point-contact transistors where the greatest challenge was the front end.  Point-contact transistors were used in the RF stages.  A TA172 (50 Mhz) was selected and while not stable was usable as a local oscillator. TA166s were used in the first IF section (7.5 Mhz) and second IF section (4.5 Mhz). The greatest challenge was finding power transistors that would drive the yoke and withstand several hundred volts. The transistors came from Mueller’s program: “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.” [Heyer 1975]

To make power transistors using germanium required efficient heat conduction to keep the junction temperatures viable. RCA tried adding fins to the transistor case, attaching the base support directly to the metal case and filling the cans with toluene to improve conduction of heat from the junction to the case. [Knight 2007] Herzog found two experimental TA155 toluene filled alloy junction power transistors that did not blow up at the operating levels required and this type were used as the yoke driver.

Overall the television set had 37 transistors and represented the absolute limit of the technology in 1952. [Ward 2004]  

Symposium all transistor TV and Radio on display at the David Sarnoff Library. [Photo courtesy Joe Knight]

An RCA Laboratory Bulletin covering the new applications presented at the Symposium can be downloaded here.

Novel Power Transistors


An interesting developmental dead end explored by RCA and used in the experimental television set was the toluene filled power transistor which made use of a suitable liquid to conduct heat away from the surface of the transistor. RCA found that flammable solvents such as benzene, toluene or xylene were effective and were sufficiently inert not to interfere with the transistor.

These transistors were described  by Larry Giacoletto who worked on them in an article in Electronics for January 1954.

Picture shows: left a liquid filled transistor and right a design where the collector is connected directly to the can to improve heat transfer (which later became common practise).

Giacoletto and joined RCA labs in 1946 as a research engineer. He played a central role in the development of RCA's new color TV system. He later developed the hybrid-pi transistor for RCA.


Amateur Radio

In the spirit of pushing all available boundaries, George Rose, a key figure in the development of the RCA point-contact transistor built a simple single transistor CW transmitter for the 2 metre band successfully communicating with local ham operators up to 25 miles away. This feat was first reported in the February 1953 edition of QST and covered comprehensively the following month. The event was also covered in the RCA publication Ham Tips which featured George Rose and his rig on the front cover. The transmitter is preserved in the AWA Museum with copies of the Ham Tips and February QST [Photo courtesy of Joe Knight].



Performance on 2 metres (146 Mhz) required selected transistors. “Transistors were not supposed to work at frequencies as  high as 146 megacycles. Ordinary transistors do not, but we in RCA managed to put together some special ones which not only oscillated at 146 Mc but  which continued to do so above 300 Mc.” [Rose 1953]

 

RCA Junction Transistor 2N33 & 2N34

Herold attended the IRE Electron Devices closed conference at Durham NC in June 1951 where John Saby of General Electric presented his alloy junction transistor and Bell Laboratories presented their grown junction transistor work.

“I was so persuaded of the advantages of the junction device that my group dropped all work on the point-contact transistor, turning over any further development to our Harrison subsidiary. It was a correct decision because the point-contact device has disappeared altogether. Having been outdone in making the first units, I organized a crash program to find a new approach for quantity production. By the end of 1951, we had succeeded so well that 1952 became a revolutionary year in transistor work at RCA.” [Herold 1983]

Knight’s records partly confirm this: There are two new point-contact types registered in 1950 and two in 1951 compared to seven from the end of 1948 through 1949. But the alloy junction transistor could not deliver the performance necessary for a portable radio and it was necessary for RCA to invest more development effort in 1952.

Charles Mueller was a key person associated with the development of the RCA alloy junction transistor. He graduated PhD in physics from MIT in 1942 and commenced work at the RCA plant at Harrison. In 1951 he was transferred to the RCA semiconductor programme.

Reflecting on the difficulties of manufacturing point-contact transistors and their inherent noisiness Mueller commented: “When the idea of the junction transistor was expostulated by Shockley, [Shockley 1949, Shockley 1951] 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.” [Heyer 1975]

Mueller was asked by Kurshan to lead a team to develop a reliable transistor that could operate reliably at broadcast frequencies and therefore be suitable for portable transistor radios.

Alloy Junctions

RCA opted for an alloy junction strategy based on published work by General Electric. In 1950 Robert Hall and Crawford Dunlap published Hall’s work on P-N junctions made by diffusing donors and acceptors from opposite sides of a germanium wafer. [Hall 1950] This work was directed to applications in rectifiers and while it did not directly anticipate the transistor it was the underlying technology for the transistor programmes undertaken by both General Electric and RCA. Mueller recalls “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.” [Heyer 1975]

Hall’s publication gave very little useful information simply saying…”the non linear impurity distribution which is required may be obtained by thermal diffusion of donor and acceptor impurities into opposite sides of a wafer of semiconductor.”  [Hall 1950] Thus RCA were left guessing which impurities were most suitable and how “diffusion” might be obtained.

The work started in January 1951 when Pankove tried to make a PN junction by diffusing antimony and aluminium into germanium. By June 1951 no progress towards useable junctions had been achieved and Mueller was assigned to the diffusion experiments. The approach adopted was to evaporate the appropriate donors or acceptors onto the germanium surface. This approach was abandoned in September 1951 when Mueller tried alloying a small sphere of indium onto N-type germanium. This was the approach that General Electric had discovered and worked well. By the end of September Mueller had made 40 PN junctions through this approach.

By December 1951 RCA was confident that transistors could be made by the alloy approach. [Choi 2007]

Alloy Junction PNP Transistor

Key members of the alloy junction team were R Law, Jack Pankove and L Armstrong. Their development closely paralleled that undertaken by John Saby at General Electric, utilising indium dots that were alloyed to opposite faces of an N-type germanium wafer producing a PNP transistor after the alloying process.

Small dots were needed: 0.1-0.2 mil in diameter. In the laboratory these were made by slicing up indium wire using a stack of razor blades then dropping the indium pellets into a hot column of oil. This melted the indium which shrank into a sphere thanks to surface tension. The method was refined by injecting molten indium into an oil column from a syringe which functioned like a shot tower. The indium dots were graded by sieving.

After making the first 100-200 transistors at Princeton, pilot production was transferred to the Harrison tube division.  Mueller began to spend a lot of time commuting between Princeton and Harrison. A production line was set up with 10 female operators that could make 100 transistors per day. The room was exceptionally dirty: so much so that there were dust particles as big as the indium dots! Lack of a clean room environment and air conditioning contributed to the high reject rate.

Transistor characteristics were controlled by alloying temperature: Higher temperatures accelerates alloying from the emitter and collector side resulting from a thinner base region.

Mueller reflecting on this time says:

“Convincing factory vacuum tube personnel that transistors had important possibilities was a slow job.  My technician and I spent several months commuting every day from Princeton RCA Laboratories to the RCA plant at Harrison, New Jersey to help them set up a transistor line to make production quantities of transistors.” [Ward 2001]         

 

Materials Processes

Whereas point-contact transistors could be made with polycrystalline germanium, the alloy junction approach required wafers cut from single crystals. This had also been the advice from Shockley given informally to RCA’s Schuyler Christian in November 1950 noted above: single crystals were essential to obtain uniform, consistent and reproducible transistor action.

Thus RCA developed its technology for the production of single crystal germanium. In addition to Christian, Arnold Moore, Bernard Selikson, Fred Rose and Paul Herkart developed the equipment and techniques for the growing of single crystals [Kilbon 1964]

RCA developed sophisticated approaches to zone refining in the mid 1950s. Instead of pulling a germanium crystal through a heated zone, they built a furnace with 34 individually heated cells each 25mm long arranged in three groups. A controller actuated a cell within each group in turn creating three heat waves that travelled along the crystal sweeping the impurities to one end. [Herkart 1956]

Ironically RCA found that defect free germanium would cause increased wetting of the germanium wafer leading to oversized junction area (and hence increased collector capacitance and poor RF performance). [Heyer 1975]

Controlled wetting of the germanium wafer by the indium dots was an important issue in producing consistent transistors. Armstrong wrote in 1952 “Control of the area of p-n junctions made by such combined alloying and diffusion of indium metal into n-type germanium is important to obtain uniformity of results.” [Armstrong 1952] He pointed out that simply applying indium dots of consistent size did not give rise to junctions of consistent size. RCA tried electroplating indium onto the wafers but found that the indium tended to agglomerate on melting. Armstrong developed a method of plating the germanium with a metal such as gold which was readily wet by a melting indium dot applied to it. The area defined by the gold determined the area of the junction.

During the alloying process RCA found that some of the indium evaporated over the surface of the wafer potentially short circuiting the junctions. This was removed by etching. [Law 1952]

Pilot Production

Information extracted from the RCA Laboratories' files and journals by Joe Knight enables the progress at RCA to be charted through 1952. [Knight 2008]

 

Date

Alloy Junction Type

Description

April 1952

TA-153

PNP transistor. Resin case. Prototype of the 2N34

 

TA-154

NPN transistor. Prototype of the 2N35

June 1952

TA-155 & TA-155A

PNP 1 watt transistor including “A” version rated at 100 volts

June 1952

TA-156

NPN 1 watt transistor rated at 100 volts

July-December 1952

 

TA-159 & TA-159A

PNP symmetrical types (same sized emitter and collector indium dots)

 

RCA Alloy Junction Patent

In June 1952 RCA filed on its alloy junction technology. [Pankove 1952] The claims relate to the TA-153 junction transistor described in the following terms:

A transistor body made from N-type monocrystalline germanium typically of 2-5 ohm-cm resistivity (for PNP transistors)

Acid etching diced wafers to restore  the crystalline surface structure

The etched wafers 3-6 mils in thickness (mil = 1/1000 inch)

The collector dot is applied first and adhered by firing at 250C for one minute

The emitter dot is applied and the wafer fired at 400-500C for 10-20 minutes to obtain alloying and diffusion of the indium to the extent required

RCA filed its patent one day prior to John Saby’s General Electric  patent 2,999,195 and was the subject of litigation as to who had priority on the alloy diffusion process. After nine years General Electric won the battle by proving that Saby had invented the alloy junction transistor before RCA. The court truncated RCA’s claims to become an improved version of the General Electric transistor.

Their surviving claims covered the use of a collector of significantly greater diameter than the emitter in order to obtain improved collector efficiency and higher alpha. The data given shows that for emitters and collectors of the same diameter alpha ranges from 0.5-0.85. Where the collector is 8 times the diameter of the emitter then alphas of up to 0.98 are obtained. The RCA patent issued on 17th October 1961.

By the end of 1952 RCA was making pilot quantities of alloy junction transistors in the form that went into small scale commercial production the following year. An n-type germanium wafer is supported on a central base electrode. Fine wires connected the emitter and collector leads to the indium alloy dots on either side of the wafer. The leads are mounted in a resin base and the assembled transistor is embedded in a light proof resin.

In a lot of 118 transistors gain varied from 20-50 db with most around 40 db. Other parameters varied over similarly wide ranges. [Law 1952]

 

 

Product Launch

In May 1953 RCA announced that four transistors were commercially available. These were:

 

RETMA Serial

Application

Price US$

Point-Contact Types

 

2N32

Small signal amplifier to 1.5Mhz

15.40

2N33

Oscillator to 50 Mhz

23.00

Alloy Junction Types

 

2N34

Small signal PNP

13.40

2N35

Small signal NPN

18.40

 

 

 

 

 

 


[Product launch images courtesy of Joe Knight]

Production was at the Harrison plant with a capacity of a few thousand per month: initially small scale production. See here for further detailed information on their production.

The 2N34 was based on the TA-153 prototype as described in Law 1952 and the 2N35 was based on the TA-154 as described in Jenny 1953. Their characteristics were as set out in the following table:

 

TA-153/2N34

TA-154/2N35

Pellet Thickness mil

6

Data sought

Pellet resistivity ohm-cm

3

 

Collector Diameter mil

45

 

Emitter Diameter mil

15

 

Base Width mil

2

 

Alloyed Impurity

Indium

Antimony

 

Pc max m watts

50

50

hfe

40

40

 

The PNP-NPN pairing created new opportunities for novel circuitry not possible with tubes. For example, George Sziklai demonstrated the use of complementary circuitry in the RCA Symposium television set in 1952. [Ward 2004, Sziklai 1953]

In a paper presented to the IRE-RTMA conference in Toronto in October 1953, R Cohen, an applications engineer from Harrison was relatively critical of what had been achieved in this first commercial release. Suggesting a doubtful future for the point-contact transistors he concluded that “the junction types are in a more promising position concerning the problems of reliability and interchangeability.”  But he observes that junction transistors still present formidable obstacles such as “the problem of producing transistors in high volume, at low cost and under processing controls which will ensure a more favourable situation regarding interchangeability.” [Cohen 1953]

The Dick Tracey Wrist Radio

The Dick Tracey radio was a phenomenon in late 1953. It was built by Paul Cooper then at the US Army Signal Corps at Fort Monmouth where the army maintained a group developing transistor receivers and video amplifiers. “On Friday in late August 1953, orders came down to design and build a transistor wrist radio. We designed and build a breadboard model over the weekend, fabricated it with a printed circuit board, and delivered it a few days later.” The design was simple: a single regenerative detector stage using a Western Electric 1729 point-contact transistor followed by two audio amplifier stages using RCA TA-153s. The set was the size of a large matchbox that could be worn on the wrist powering a hearing aid earphone.

The set was made public in September 1953 and earned the “Dick Tracey” nickname in the technical and popular press. For example, the edition of Time Magazine on the 21st of September stated  “In his tireless comic-strip crusade against criminals with brutal habits, and oddly shaped heads, Detective Dick Tracy has had an invaluable mechanical ally: “The two-way wrist radio.” . . . Last week life imitated art again. The U.S. Army Signal Corps announced that it had developed a wrist radio with a receiving range of 40 miles.” [Cooper 1998]

While this was not the brain child of RCA it was a very high profile application of their first commercial transistors creating welcome national publicity.

 

Radio Frequency Junction Transistors 2N139 & 2N140

By August 1952 RCA could make useable low frequency transistors (TA-153) but it could not make high frequency transistors consistently: “The present developmental germanium PNP junction transistor was not designed for high frequency operation, and no attempt was made to control this parameter.” [Law 1952] But there were indications from single outliers that the alloy junction approach could yield transistors that would perform adequately at radio broadcast frequencies. Referring to the possibilities of high frequency performance it was noted by Law: “The variation between units is large and typical values cannot be quoted as is possible for other parameters so far discussed. It will be observed that some units exhibit usable gain to 4 Mhz.”

Norman Ditrick who worked at Harrison and seconded to Princeton to work with Mueller on the development of radio frequency junction transistors recalls “At this time, “high frequency” meant IF frequency for radios.  The original junction transistors would only work at audio frequencies.  Working with Charlie [Mueller], we got the transistors to function all the way up to 2 Mhz, which was good enough for the standard AM radio.  We accomplished this by varying the size of the alloyed indium dots and by changing the physical structure of the device to reduce the parasitic resistance.” [Ward 2001]

There are three important factors that limit high frequency performance in alloy junction transistors:

(1) The width of the base layer

(2) Series base lead resistance

(3) High collector capacitance

In the common base configuration (used to maximise high frequency gain) there is an emitter-base capacitance, Cbe across the input known as the diffusion capacitance that is significantly greater than the capacitance associated with the emitter-base junction. The diffusion capacitance is proportional to the square of the effective base width or the emitter current and inversely proportional to the diffusion constant for the minority carriers. Qualitatively there is a low rate of change in the base potential due to the slow migration of the minority carriers from the emitter to the collector. This phenomena is independent of the junction area.

Thus it is clear that reducing the base width will bring a significant reduction in diffusion capacity. Minority carrier diffusion rates cannot be improved in an alloy junction transistor of a particular kind. NPN transistors of the same geometry perform better than PNP transistors (the diffusion capacitance is halved) due to the higher mobility of electrons.

The base lead may be connected to the wafer some distance from the base region and consequently there is a series resistance in the base circuit. “This combination of series resistance and emitter-to-base capacitance constitutes a low-pass filter which is one of the most important limiting factors of transistors such as those previously described.” [Mueller 1954]

Collector capacitances are also important and as expected can be reduced by reducing the size of the collector junction. Planar junctions improve performance by reducing variations in the base width. A viable base width for a high frequency alloy transistor is 0.5mils.

A narrow base width can be achieved by starting from a thick wafer (such as used in the TA-153) and increasing the alloying period to create greater impurity penetration. But control is difficult leading to highly variable hi frequency performance. In addition, the junctions are hemispherical and not the preferred planar geometry.

 

1953: The Well Geometry: the SX-160

In July 1953 Mueller and Pankove presented some novel work done on RF transistors capable of performing at  up to 20 Mhz at the IRE-AIEE Transistor Research Conference at State College, Pennsylvania.

They described a solution that solved the base lead resistance issue while ensuring a thinner base. Base lead resistance could be reduced by making the base connection as close to the junction region as possible. One approach was to solder an annular base connection to the wafer just big enough to avoid the junctions or using resist and electroplating on a similar connector. These approaches created the potential for shorting the junctions. Instead, Mueller and Pankove used a very thick wafer giving low base lead resistance and obtained a narrow base by creating the transistor in a well as shown.

While a small collector junction improves high frequency performance it also needs to be about 50% larger than the emitter for good amplification. In their prototypes the emitters were 10 mils and the collectors were 15 mils.

Comparison between the SX-160 and the TA-153

SX-160

TA-153

Pellet Thickness mils

20

6

Pellet resistivity ohm-cm

0.7

3

Collector Diameter

15

45

Emitter Diameter

10

15

Base Width

0.5

2

Base Lead Resistance ohms

70

350

Collector Capacitance mmf

10

35

Diffusion Capacitance mmf

700

11000

Collector-Base Resistance m ohms

0.2

2

Alpha Cut-off Mhz

10

0.8

For an Emitter Current ma

1

1

 

RF transistors of this kind were used in an all junction demonstration portable described by Loy Barton late in 1953 [Barton 1954]. In his introduction he points out that previously such sets had been possible only by selecting from batches of AF transistors those that had RF capability. To avoid neutralising the RF stages the new receiver used common base configuration.

 

Developments on other prototypes including the SX-160 are given in the following table. [Knight 2008]

RF Transistor Development 1952-53

Date

Alloy Junction Type

Description

Jul-52

TA 157

PNP HF

Feb-53

SA-160

PNP RF

Mar-53

SX-160

PNP RF

Apr-53

TA-209

PNP HF

Oct-53

TA-215

PNP IF amplifier with ring base tab

 

 

Developments in 1954

While the well geometry was developed and described in mid 1953 work on this concept continued for another year. In addition the annular base connection was further explored as indicated in the following table:

 

RF Transistor Development

Date

Alloy Junction Type

Description

Jan-54

TA-217

PNP IF amplifier with well in pellet

Mar-54

TA-247

PNP RF amplifier with ring base tab

Mar-54

TA-248

PNP RF amplifier with well in pellet

Mar-54

SX-162

PNP RF amplifier with well in pellet

Mar-54

SX-163

NPN RF amplifier with well in pellet

Mar-54

TA-1508

PNP IF amplifier with ring base tab

Nov-54

SX-164

PNP RF

May-55

TA-1550

PNP RF

Reference Knight 2008

Mar-56

2N139

PNP IF amplifier

Mar-56

2N140

PNP Mix/Osc

 

The well geometry did not survive, Mueller noting: “The SX160 fulfilled its purpose as an early transistor that enabled engineers to carry on transistor circuit development at radio frequencies, however, it was considered too difficult and costly to manufacture.” [Mueller 1956]

 

 

1955: Improved Planar Junctions

Work on improved planar junctions was the key technology underpinning the RCA high frequency transistors.

The advantages of planar junctions particularly for RF transistors were well understood. For example, Jenny with reference to RCA NPN transistors had noted two years earlier: “As it is desirable to obtain a planar junction, it is preferable that the wetting process take place rapidly and at a low temperature at which the solubility of germanium in the liquid impurity substance is still small.” [Jenny 1953] In his paper he illustrates the issue: the impurity disk forms a droplet on melting and then wetting and alloying proceed from the point of contact of the droplet creating an uncontrolled hemispherical junction.

Norman Ditrick working on the development of radio frequency junction transistors recalls working at Princeton. “During this time I developed a technique which later was the basis for one of my patents – we developed a technique to wet the indium dots before alloying…” [Ward 2001 Ditrick 1955]

This work was presented at the IRE-AIEE conference on Semiconductor Research at Philadelphia in June 1955 by Mueller and Ditrick. [Mueller 1956] In their paper the authors describe the problem:

“The alloy junction transistor as been widely used in audio transistors where the rounded junction shape can be accepted without serious undesirable effects. However, for many reasons, flat parallel junctions are especially important to improve the uniformity of transistors with small base widths.”

“At present, the method most generally used for making alloy junction transistors is to position the impurity dot on the base wafer and fire the assembly in a reducing atmosphere. In this process, the dot material first wets at one point from which point the alloying and wetting progress simultaneously. Thus the dot penetrates and spreads out at the same time and a curved junction front…results.”

Hence the solution was to separate wetting and alloying in a three stage process:

(1) The dot material wets the base wafer (soldering)

(2) Alloying takes place without further wetting

(3) The dissolved germanium with a small amount of dot material is recrystalised on the base wafer.

Soldering or wetting was done at 300-350 C and needed new fluxes (hydrogen was normally used at the higher alloying temperatures in the conventional process). Ditrick found that by adding 1% zinc to the indium he could obtain effective soldering using weak fluxes. The first step in the process consisted of taking zinc-indium alloy balls of the desired volume, etching them, adding flux, draining, positioning the balls on the base wafer using jigs and heating rapidly to 340 C for 3 minutes to solder the balls to the wafer.

Alloying was carried out as an equilibrium process in which the depth of alloying is independent of the alloying time thereby eliminating one source of variability. In this process the depth of penetration is determined by the alloying temperature, the dot volume and the wetted area. During this stage, further wetting is prevented by using an oxidising or neutral atmosphere.

On slow cooling, recrystalisation of the germanium occurs depositing most of the dissolved germanium back on the wafer as a single crystal. By using wafers cut on the (111) crystallographic face the junction is highly aligned to that plane due to the strong preference of the melt interface to expose the (111) face providing near equilibrium conditions through gradual cooling are maintained. They obtained highly planar junctions:

“In experimental lots of six units, flat junctions have been made that showed a variation of +- 0.025 mil in the location of the junction. Thus, not only do these new techniques provide for microscopically flat junctions, but also for the precise control of their location.”

 “The net result of these characteristics is a uniform well-defined junction with considerably improved electrical properties. Thus, the upper frequency limit to which the alloy transistor can be economically exploited is considerably extended.”

This approach was used in the IF transistor described in the following section.

Pilot Production of the 2N139

In March 1956 RCA Laboratories issued a comprehensive report on its new IF transistor that had been in development since August 1954. [Mueller 1956] The development had been initiated at Princeton but then transferred to Harrison where the pilot production line was located.

The objective of the project was to develop a transistor that had the characteristics and performance of the SX-160 but without the attendant high costs of production. Tight production tolerances similar to those expected in tube production were sought. This was possible using the improvements that RCA had developed for producing planar soldered-alloyed junctions described in the previous section.

In order to obtain low base resistance a ring base tab was employed. The pellet resistivity selected was 1 ohm-cm and the authors note that obtaining the required tolerance of ±20% “was difficult to meet for some time at Harrison because of large variations in resistivity across a crystal diameter.”  Pellets were cut and etched to a thickness of 2 mils. Initially the collector and emitter dots were punched from indium sheet but this method gave poor reproducibility. Instead dots were made by the “shot tower” method described above. The dots were held in place with aluminium jigs and soldered in place at 300-340C and alloyed at 550C using indium alloys and fluxes as described in the preceding  section “Improved Planar Junctions” giving planar junctions to ±0.2 mils.

Finally the transistors were given an electrolytic etch to create groove around the both emitter and collector junctions in order to clean up any dot smearing or other edge defects.

Mueller notes that the project was a productive exercise in intra firm technology transfer: “One of the intangible but important results of the project was the education of a large segment of the Harrison Engineering organization into the concepts necessary to work in the high frequency transistor field. The Princeton people were also introduced to the problems of production design and process control.”

Data for the new transistor shows that this met the objective that it should have “electrical characteristics similar to the SX-160.” 

Comparison Between the SX-160 and the 2N139

SX-160

2N139

Base Lead Resistance ohms

70

80

Collector Capacitance mmf

10

12

Alpha Cut-off Mhz

10

11.6*

Pellet Thickness mils

20

2

Pellet resistivity ohm-cm

0.7

1

Collector Diameter

15

15

Emitter Diameter

10

10

Base Width

0.5

0.8

* Figure of Merit. Specification was 30dB gain at 0.455Mhz

 

 

Launch of the 2N139 & 2N140

The new transistors were announced in March 1956: the 2N139 and 2N140 PNP transistors intended for IF and converter service respectively in portable and automotive broadcast band receivers.

“The 2N139 designed especially for 455 kc applications, can provide a power gain of 30 db at 455 kc in suitable common emitter circuits. It has excellent stability, low collector cut-off current, is only 0.26 inch in diameter and has a seated height of 0.495 inch.

The 2N140 is mechanically like the 2N139 but has characteristics which especially meet the requirements of converter and mixer-oscillator applications in the standard AM broadcast band. In typical operation in common emitter circuit the 2N140 features a conversion power gain of 27 db at 1 Mc. The circuit is sufficiently stable to permit changing transistors without readjusting the oscillator.

Maximum ratings for the 2N 139 and 2N140 are collector voltage -16; collector current -15ma; collector dissipation 35 mw; emitter voltage -12; emitter current 15 ma.”

RETMA

Type

Pc Mw

Fab Mhz

hfe

2N139

PNP

35

4.7

48

2N140

PNP

35

7

45

 

In July 1956 RCA obtained JETEC registration of the 2N218 and 2N219 which were identical to the 2N139 and 2N140 other than their flexible leads and case dimensions [JETEC 1956]

 

Silicon Alloy Transistor                 

RCA developed a silicon alloy junction NPN transistor, the SX-152 from March 1954. [Knight 2008]  This transistor was never commercialised. The work was presented by Herbert Nelson at the Semiconductor Research Conference of the IRE at Minneapolis in June 1954.

Gordon Teal who led the development of the grown junction transistor at Texas instruments noted in regard to other approaches to a silicon transistor: “A reason that going the grown-junction route would avoid the differential expansion difficulties between silicon and an alloying electrode inherent in the use of alloyed junctions. Most companies took the alloy route.” [Goldstein 1991]

At Minneapolis there was an entire section on silicon: In addition to Nelson’s paper, Texas Instruments presented their silicon grown junction transistor disclosed the previous month at Dayton, Ohio; Raytheon presented silicon grown junction transistors; Bell Laboratories three papers on silicon rate grown transistors, silicon alloy transistors and on wide area junctions by diffusion from the gas phase and Hughes Aircraft on alloy junction transistors. [Kurshan 1954]

In order to develop a viable transistor RCA had to solve two key issues:

(1) The production of high quality silicon

(2) Alloy dot compositions that minimised thermal mismatch strains while permitting good wetting

Early on in the program RCA used low quality silicon having 2 microsecond minority carrier lifetime. Later it had 20 microsecond silicon available which gave considerably improved transistor performance.

Having tested a wide range of N-type impurities and diluents, RCA selected an alloy of 54% lead, 45.4% gold and 0.6% arsenic. The alloy is brittle and difficult to form into dots: the solution was to produce a composite by punching disks from a foil consisting of gold and lead-arsenic layers. The arsenic-lead side was applied to the silicon.

Nelson’s paper gives comprehensive process information (cleaning, etching fluxing and alloying) indicating that RCA was not seeking to protect its commercial position around the silicon alloy development. (In contrast, its publications on the TA-153 and the TA-154 give virtually no process information.) An interesting feature of the process was use of a graphite jig to hold the collector, emitter and base tab dots in place prior to alloying “as often employed in making PNP germanium transistors.”

The transistors were described as performing in a similar manner to conventional AF germanium PNP transistors providing good quality silicon was used and that their high temperature performance was as expected. [Nelson 1954]

But RCA did not pursue silicon technology at that time. Saddler attributes this to an influential market assessment  undertaken by RCA that took the view that silicon was a military technology and that  long term the military sector would not be significant. “It could be argued that belief in that report caused RCA’s ultimate small portion of the industry while Texas Instruments whole-heartedly embraced silicon and adapted itself to the technology that followed the point-contact transistor. For silicon that was the grown junction transistor.” [Saddler 1990]

 

Drift Transistors

The development of drift transistors are an important milestone in the history of the RCA transistor programme since they represent the first time the company led the way with a new transistor design.

The drift transistor solved the three issues that inhibited the high frequency performance of the alloy junction transistors of the time:

(1) Slow minority carrier transit across the base

(2) High base lead resistance

(3) High collector capacitance

Drift transistors were made by controlling the doping of the base such that this is high at the emitter and is low at the collector. The graded doping creates an accelerating field which increases carrier transit velocity. Drift transistors have lower input impedance due to the higher conductivity of the base at the emitter junction. Lastly the collector capacity is reduced by the low conductivity of the base at the collector junction. Thus the drift transistor offers a design that combines three strategies to improve high frequency performance.

The drift transistor was “conceived before its time by a young German theorist at Darmstadt.” [Early 1992] Its inventor, Dr Herbert Kroemer, joined RCA in 1954 where it was commercialised. Kroemer was awarded a Nobel Prize in 2000 for his work in semiconductor hetero-structures used in high-speed and opto-electronics.

Kroemer graduated PhD in 1951 after submitting a thesis on electron transport in high fields in germanium transistors and began work in the telecommunications laboratory of the German Postal Service. Initially he worked on how transistor junctions were formed in the alloy process.

In his Nobel Prize lecture Kroemer states:  The early bipolar junction transistors were far too slow for practical applications in telecommunications and I set myself the task of understanding the frequency limitations theoretically – and what to do about them. One approach - not the only one - was to speed up the flow of the minority carriers from the emitter to the collector by incorporating an electric field into the base region. This could be done by using, not a uniform doping in the base, but one that decreased exponentially from the emitter end to the collector end – the so-called drift transistor.” [Kroemer 2001]

In 1953 Shockley visited Kroemer and discussed these ideas. Later the same year, Herold visited from RCA and being impressed with Kroemer’s ability to speculate correctly on how RCA were developing their alloy junction transistors offered Kroemer a job at RCA. A year later in 1954 Kroemer started at the RCA Princeton Laboratories.

At RCA he began theoretical research on diffusion of donors or acceptors into semiconductors. In 1954 his work on diffusion and drift transistors was published.

“The first drift transistor paper introduced the concept of a doping-engineered electric field in the base to reduce the electron base transit time. The paper predicted an 8-fold increase in the theoretical frequency limit as compared to Shockley’s “diffusion” bipolar transistors. The notion of aiding base transport with a built-in electric field resulting from the variation of base doping density is in use in virtually all bipolar transistors fabricated today, and is one of the central concepts in BJT design.” [Brar 2001]

Ditrick who worked on many of RCA’s development transistors recalls: These transistors were made using a special process to prepare the substrate, which was intrinsic germanium that was doped by diffusion with arsenic.  It was difficult controlling the thickness of the substrate, since we had to etch off one side of the diffused arsenic.  I built almost every one of these devices in the development stage.  After we were satisfied with the design, more units were made in the model shop and then finally in production.  This work probably started in Harrison, and then was finalized after we had  moved to Somerville.“ [Ward 2001]      

First record of a prototype drift transistor occurs in 1955 (TA-E-1579  PNP type). This was commercialised as the 2N247 in 1956. [Knight 2008] This was remarkably fast compared to the time to market for the first AF and RF transistors and indicates an increasing maturity in its process technologies:

“As if to emphasize the close cooperation between the research organization and the product divisions in the transistor program, the drift transistor became a part of the commercial line of RCA transistors within months after its development at Princeton.” [Kilbon 1964]

The 2N247 was a small signal RF type with an alpha cut-off of 30 Mhz. In common with many drift transistors it had a fourth lead connected to the can to serve as an RF shield.

Progress was rapid. At the 1957 Electron Devices meeting Schwatrz and Slade reported on high frequency PNP drift transistors made by formation of the base region by diffusion from the alloy emitter. They obtained a cut-off of 200 MHz with high gain and a reverse collector breakdown of 50 volts. [Schwartz 1958] In 1958 RCA said it had drift transistors that could operate up to 300 Mhz [Lenz 1958] At this time drift transistors and surface barrier designs were the only technologies competing in the high frequency arena.

In their abstract Schwartz and Slade do not disclose how they obtained “base doping by diffusion from the alloy emitter.” It appears to describe the post alloy diffusion approach employed elsewhere in which the emitter pellet contains a majority of P-type and in addition, minority N-type impurities. An alloying process is undertaken first to form a P-type emitter. After the un-alloyed impurities are removed by etching, the structure is heated to a higher temperature to obtain diffusion of the P and N impurities. Because the rate of diffusion of N impurities in germanium is about 100 times faster than P-type the region beyond the emitter becomes N-type forming the base with the required profile: High N-type near the emitter junction decreasing towards the collector. [For example, Lamming 1958]

This approach yields PNP drift transistors.  Relative diffusion rates of acceptors and donors in silicon are reversed and this approach yields NPN transistors.

In the following table the commercially released RCA drift transistors are shown with the associated date courtesy J Knight. The 2N384 was released by JEDEC on October 1957 and the 2N379 series in October 1958. The 2N640 series date is inferred from their first appearance in the 4th Edition of the General Electric Transistor Manual.

 

RETMA No

Date

Type

Pc (mW)

Hfe/alpha

Fab Mhz

Typical

Typical

2N247

1956

PNP RF

35

60

30

2N267

1957

PNP RF

35

60

30

2N384

1957

PNP Osc

120

0.984

30

2N370

1958

PNP RF

80

60

30

2N371

1958

PNP RF

80

0.984

30

2N372

1958

PNP RF

80

60

30

2N373

1958

PNP Osc

80

60

30

2N640

1959

PNP RF

80

0.984

42

2N641

1959

PNP IF

80

0.984

42

2N642

1959

PNP Osc

120

0.984

42

2N643

1959

PNP Sw

120

20

20

2N644

1959

PNP Sw

120

20

40

2N645

1959

PNP Sw

120

20

60

 

References and Acknowledgements

 

The author gratefully acknowledges the work of Jack Ward at the Semiconductor Musuem and the Institute of Electrical and Electronics Engineers for the recordings of the oral histories referenced below and that of Hyungsub Choi and Joe Knight in relation to historical research on the RCA transistor programme at the David Sarnoff Library. Additionally to Joe Knight for reviewing this text and helpful suggestions and generous assistance from his personal collection of records in relation to RCA. Lastly thanks are due to the Hagley Museum who kindly provided copies of key reports from their archive of RCA Camden records and the role of this library and the David Sarnoff library in the preservation of important historical records.

Armstrong L 1952 P-N Junctions by Impurity Introduction Through an Intermediate Metal Layer Proc IRE 40 1341-42

Barton L 1954 An Experimental Transistor Personal Broadcast Receiver Proc IRE 42 1062-66

Bardeen J Pfann W 1950 Effects of electrical forming on rectifying barriers of n- and p-germanium transistors  Phys. Rev  77  401-2

Bell 1951 The Transistor: Selected Reference Material on Characteristics and Applications November 1951

Bell 1952 Transistor Technology Vol 1 July 1952

Brar B 2001 Herb’s Bipolar Transistors IEEE Trans on ED 48 No11 2473-76

Choi H 2007  The Boundaries of Industrial Research    Making Transistors at RCA 1948-1960 Technology and Culture 48 758-82

Cohen R 1953 Application Considerations for RCA Commercial Transistors Proc IRE-RTMA Toronto Meeting October 1953 32-46

Cooper P 1998 The USA Army Signal Corps’ “Dick Tracey” Transistor Wrist Radio Proc IEEE 86 163-9

Ditrick N 1955 US Patent 2,761,800 Method of Forming P-N Junctions in N-Type Germanium Filed May 2 1955  Issued Sept 6 1956

Early J 1992 Classic Semiconductor Devices – Point-Contact Through HSI IEEE Trans on ED 39 2635

Goldstein A 1991 Gordon K. Teal, Electrical Engineer, an oral history conducted in 1991 by Andrew Goldstein, IEEE History Center, Rutgers University, New Brunswick, NJ, USA.

http://www.ieee.org/portal/cms_docs_iportals/iportals/aboutus/history_center/oral_history/pdfs/Teal136.pdf

Hall R Dunlap W 1950 P-N Junctions Prepared by Impurity Diffusion Phys Rev 80 467

Henriksen P 1987 Solid State Physics Research at Purdue Osiris 2nd Series 3 237-260

Herkart P Christian S 1956 A Switched Zone Furnace for Germanium Purification Transistors 1 Radio Corporation of America 59-65

Herold E 1983 Focus on a Career Engineer: Excerpts from Bygone Days: An Autobiography (David Sarnoff Library http://www.davidsarnoff.org/ewh.html)

Heyer M Pinsky A 1975   Dr. Charles W. Mueller, Electrical Engineer, an oral history conducted in 1975 by Mark Heyer and Al Pinsky, IEEE History Center, Rutgers University, New Brunswick, NJ, USA. http://ieee.org/portal/cms_docs_iportals/iportals/aboutus/history_center/oral_history/pdfs/Mueller025.pdf

IEEE 1976 Edward W. Herold, 1907 – 1993 IEEE Biographies http://www.ieee.org/web/aboutus/history_center/biography/herold.html

Jenny D 1953 A Germanium NPN Alloy Junction Transistor Proc IRE 41 1726-34

Knight  J 2007  A Survey of Early Power Transistors: RCA 1950s Germanium Power Transistors

http://www.semiconductormuseum.com/Transistors/LectureHall/JoeKnight/JoeKnight_EarlyPowerTransistorHistory_RCA_Index.htm

Knight J 2008  The First RCA Experimental, Developmental and Production Transistors Tube Collector 10 No 4  7-11 Also on this site RCA Developmental Transistors

Kilbon K 1964 Pioneering in Electronics; A Short History of the Origins and Growth of RCA Laboratories, Radio Corporation of America, 1919 to 1964 (David Sarnoff Library http://www.davidsarnoff.org/kil.html)

Koremer H 2001 Quasi-Electric Fields and band Offsets: teaching Electrons New Tricks Rev Mod Phys 73 783 - 93

Kurshan J et al Report on the Conference of Semiconductor Devices Research Minneapolis MN June 28-30 1954 RCA Technical Report PEM-452

JETEC 1956 Joint Electron Tube Engineering Council Release 1698 July 23rd 1956 Courtesy J Knight

Lamming J 1958 A High Frequency Drift Transistor by Post Alloy Diffusion International Jnl of Electronics 4:3 227-36

Law R Mueller C Pankove J Armstrong L 1952 A Development Germanium PNP Junction Transistor Proc IRE 40 1352-57

Morton D 2000 John Saby, Electrical Engineer, an oral history conducted by David Morton, IEEE History Center, Rutgers University New Brunswick, NJ, USA www.ieee.org/organizations/history_center/oral_histories.html

Mueller C Pankove J 1954 A PNP Triode Alloy-Junction Transistor for Radio-Frequency Amplification Proc IRE  42 386-391

Mueller C 1956 Uniform Planar Junctions for Germanium Transistors RCA Review March 1956 and reproduced in Transistors 1 Radio Corporation of America 1956 121-31

Mueller C Pensak L 1956 Intermediate-Frequency Transistor – Construction and Pilot Production Report PTR-549 RCA Laboratories

Nelson H 1954 A Silicon NPN Junction Transistor by the Alloy Process Transistors 1 Radio Corporation of America 1956 172-81

Pankove J 1952 US Patent 3,005,132 Transistors Filed 13th June 1952 Patented 17th October 1961 (RCA)

Petritz R 1962 Contributions of Materials Technology to Semiconductor Devices Proc IRE 50 1025-38

Pfann W 1951 Significance of Composition of Contact Point in Rectifying Junctions on Germanium Phys. Rev. 81 882

Lenz W Cook W 1958 Transistor Fundamentals & Applications RCA Service Company Camden NJ

Riordan M, Hoddeson L 1997 Crystal Fire W W Norton & Company

Rose G 1949  US Patent 2,538,593 Semiconductor Amplifier Construction Filed 30th April 1949  Patented 16th January 1951

Rose G 1953 The Transistor-Or 25 Miles on a Hunk of Germanium QST March  13-15

Saddler J 1990 The Human Side of Early Electronics and Semiconductors SMEC Vintage Electronics 2 1990

Schwartz R Slade B 1957   A high-speed P-N-P alloy-diffused drift transistor for switching applications  Electron Devices Meeting 1957 3 117

Seitz F 1995 Research in Silicon and Germanium in World War II Physics Today  January 1995 22-7

Sheckler A 2004 The Beginning of the Semiconductor World; General Electric’s Role Proc IEEE Conference on the History of Electronics

Shockley W 1949 The Theory of P-N Junctions in Semiconductors and P-N Junction Transistors Jnl Bell Sys Tech 28 435-89 (reprinted in Sze S M 1991 Semiconductor Devices: Pioneering Papers World Scientific Publishing)

Shockley W Sparks M Teal G 1951 P-N Junction Transistors Phys Rev 83 151-62

Slade B 1952 The Control of Frequency Response and Stability of Point-Contact Transistors Proc IRE 40 1352-84

Stelmak J 1951 Electric Forming in n-Germanium Transistors Using Phosphorous Alloy Contacts Phys Rev 83 165

Sziklai G 1953 Symmetrical Properties of Transistors and Their Applications Proc IRE 41 717-24

Ward J 2001 Early Transistor History at RCA Oral History of Norman H. Ditrick

http://semiconductormuseum.com/Transistors/RCA/OralHistories/Ditrick/Ditrick_Index.htm

Ward J 2001 Early Transistor History at RCA Oral History Israel Kalish

http://www.ck722museum.com/history/rca/OralHistory/Kalish/Kalish_Index.htm

Ward J 2001 Early Transistor History at RCA Oral History of Jerome Kurshan

http://www.semiconductormuseum.com/Transistors/RCA/OralHistories/Kurshan/Kurshan_Index.htm

Ward 2001 Early Transistor History at RCA Oral History of Charles Mueller

http://www.ck722museum.com/history/rca/OralHistory/Mueller/Mueller_Index.htm

Ward J 2003 Early Transistor History at RCA Oral History of Richard Endres

http://www.semiconductormuseum.com/Transistors/RCA/OralHistories/Endres/Endres_Index.htm

Ward J 2004 Early Transistor History at RCA Gerald B. Herzog

http://www.semiconductormuseum.com/Transistors/RCA/OralHistories/Herzog/Herzog_Index.htm

Ward J 2005 Early Transistor History at RCA Oral History of Bernard Slade

http://www.semiconductormuseum.com/Transistors/RCA/OralHistories/Slade/Slade_Index.htm
 
 
 
 
 
Links to RCA data
 
 
 
 
This link takes you to data for early RCA RETMA registered transistors
 
This  link takes you to an extensive complilation of RCA experimental and production transistors produced by Joe Knight from the RCA archives at the David Sarnoff library and other sources. Joe Knight has also extensively documented the history of the development of power transistors to be found at the Transistor Museum site including a history of RCA power transistors.

This link takes you to more information about the first 2 Metre ham radio transmission by George Rose using an experimental RCA point-contact transistor. Images by kind permission of Joe Knight. QST article kindly provided by Jack Ward.

This link takes you to some very detailed information on the production of the first commercial RCA point-contact transistors and junction transistors.

This link takes you to downloads of the RCA Laboratories Bulletin 898 that covers all the applications presented at the RCA Symposium held in November 1952