Kai Handel Manuscript „Uses and Limits.doc“

The Uses and Limits of Theory: From Radar Research to the Invention of the Transistor Kai Handel Institute for History of Technology, Technical University Aachen (Germany)1 Paper presented at the Annual Meeting of the History of Science Society at Kansas City (Missouri), 21-25 October, 1998

Introduction The invention of the transistor at the Bell Labs in 1947 has been well documented and is often used as a prime example of the interplay between science and technology as well as of the “war’s impact on science”. Special focus is thereby given to how the technical advances during the war directly influenced the invention of the transistor.2 In addition to undeniable material and technical advances which contributed to the invention of the transistor, I focus on the role of the theoretical framework developed before and strengthened within the war. Using the example of the somewhat failed and lesser known invention of a transistor by Heinrich Welker and Herbert Mataré in Europe I demonstrate how the theoretical preoccupations with the then leading semiconductor theory hindered their research. •

I will show that the Mott-Schottky-Theory enabled both, German and British-American scientists, to create working crystal detectors for Radar applications during the second World War.

•

The theory also led in the final months of the war to a refined concept of a crystal amplifier - the so called field effect transistor.

•

Unable to produce a working device on this basis, John Bardeen and Walter Brattain at the Bell Labs, who had not been closely related to wartime research on crystal rectifiers, abandoned the Mott-Schottky-Theory and invented the point-contact transistor.

•

Working with practically identical observations as their American counterparts, Welker and Mataré, retained their allegiance to the Mott-Schottky-Theory and therefore failed to understand the invention they had at their hands.

Radar Project(s)3 At the beginning of the war the British defensive Radar system (“chain home”) operated with waves of about 10 to 12 meters. However, it was soon recognized that the use of shorter waves would enable the detection of smaller objects with higher precision.4 After a convenient method to generate waves of about 10 cm was developed by British physicists at the End of 1939 attention turned to developing an effective receiver for these ultra short waves. 1

2 3

4

Copyright 1998, Kai Handel. All rights reserved. New Address: Dr. Kai Handel, CHE Centrum für Hochschulentwicklung, Gütersloh (Germany); [email protected] See for example Hoddeson, Crystal Rectifiers, 1994. For the history of the British-American Radar project see Callick, Meters to Microwaves, 1990, S. 55-77; Guerlac, Radar, 1987; Kern, Radarverfahren, 1984; and Torrey/Whitmer, Crystal Rectifiers, 1948. See Kaiser, Case Study, 1996.

1

Kai Handel

„Uses and Limits.doc“

Electron tube detectors, used previously, were too insensitive, suffered from a bad signal to noise ratio and because of their size were unable to rectify waves of very short wavelengths. Thus, the more sensitive crystal detector from the early days of radio was reemployed. In a collaborative British-American effort, complete radar sets using centimeter waves were developed and finally put in use in the beginning of 1943. (The new radar sets were composed mainly of a resonant cavity Magnetron, a crystal detector, and a “plan position indicator”.) After capturing a crashed British bomber plane in Feb. 1943, the German military was surprised to discover how far advanced allied radar technology was. This was especially shocking because high ranking Military members had decided only a few month earlier in November 1942 that centimeter waves could not be used for Radar purposes and therefore all efforts in development should be stopped.5 Now, those efforts were restarted with increased intensity.

German crystal detectors During 1943 German scientists and engineers were called back from active military service to work at the research institutes of the armed forces and in industrial laboratories on the problems of centimeter wave radar. One of those scientists was the theoretical physicist Karl Seiler who joined the Telefunken laboratories, where he worked together with the physicist-engineer Herbert Mataré and the Physical Chemist Paul Ludwig Günther to design Silicon detectors. At research laboratories of the air force and the university at Munich another group worked on the development of Germanium detectors. The main actors here included the engineer Wolfgang Büll, the Physical Chemist Klaus Clusius, and theoretical physicist Heinrich Welker, who had studied at the famous institute for theoretical physics at Munich under Arnold Sommerfeld. Crystal detectors for rectifying radio waves had been in use in the early years of the century, but during the 1920s and 1930s they had been nearly entirely replaced by electron tubes, which had been more reliable than the enigmatic and often erratic crystal detectors. But at the end of the 1930s a new theory was developed which enabled scientists to understanding in qualitative and quantitative terms, what happened in a crystal rectifier. Thus, the crystal detector, then finally better understood as a Semiconductor device, came back into use. The theory which enabled this, today know as the Mott-Schottky-Theory, was used and entirely accepted by scientists in both the British-American and the German Radar projects. At that time interdisciplinary working groups investigated the material, physical and electrical properties of semiconductor detectors on the basis of this theory. 6 The Mott-Schottky-Theory was the first theory which explained successfully the functioning of crystal rectifiers in terms of rectification at the metal-semiconductor contact.

5

6

Reuter, Funkmeß, 1971, S. 197; Kern, Radarverfahren, 1984, S. 237-238, Schubert, Industrielaboratorien, 1987, S. 278, and Trenkle, Funkmeßverfahren, 1979, S. 43-44. The research on crystal detectors at the MIT Radiation Laboratory was guided by the theory of the British physicist Neville Mott with refinements by the German émigré Hans Bethe. (See Torrey/Whitmer, Crystal Rectifiers, 1948, p. 68-110) Both theories had been unknown to the German physicist, but the Schottky Theory presented mainly the same concepts.

2

Kai Handel

„Uses and Limits.doc“

Semiconductor (n-type)

Metal

diffusion potential

Work function

metal-semiconductor Fermi energy

D+

D+

D+

D+

D+

D+

D+

barrier layer

D+

Figure 2: point-contact device

•

Single charge carrier theory – only electron or holes are considered.

•

Theory of Metalsemiconductor contact – importance of the work functions.

Electron

10-5 cm

Donator atom

Figure 1: Schottky-Theory (see Schottky, Vereinfachte Theorie, 1942)

The main features of the theory are: •

Due to a difference in the work functions of the metal and semiconductor, a potential difference in the semiconductor is formed. More electrons diffuse out of the semiconductor than come in. Thus, a zone near the surface emerges, where there are less electrons than in the bulk of the semiconductor. This zone is called the barrier layer, because electrons have to pass the barrier before a current can flow.7

•

It is important to note that this theory is a single charge carrier theory – either electrons or holes are considered – meaning that the possibility of both charge carriers, electron and holes, carrying the current at the same time was outside the theoretical framework of the Mott-Schottky-Theory.8

Equipped with this theory, the German physicists could (like their British and American counterparts) calculate the size, the resistance and the capacitance of the barrier layer. Since the physical properties of the metal-semiconductor contact seemed successfully dealt with by the Mott-Schottky-Theory, researchers concentrated on the electrical properties of the detector as a whole. From this it was obvious that for a high frequency (small wavelength) application it was necessary to work with a point-contact device.

7

8

The forward direction is from semiconductor to metal, as observed. In the shown case this means, that a positive voltage is applied to the metal, than the barrier is lowered because the Fermi energy of the metal is lowered. (The difference in the work functions stays the same.) (As you see, the work function between metal and semiconductor is of decisive influence on the characteristics of the rectification.) Walter Schottky (like most physicist at the end of the 1930ies) was perfectly conscious about the fact that there are different charge carriers in a semiconductor. In fact his prime example is the semiconductor Selenium, which is a “hole”-carrier (or modern called “p-type”). Further: The Theory is “semi classical” in the sense that the basic concepts originate from the classical diffusion theory and only a few features of quantum theory are employed. That means it is not formulated in the mathematical form of quantum theory: the wave mechanics. (It does not use the full mathematical apparatus of quantum theory.)

3

Kai Handel

„Uses and Limits.doc“

Further it became clear how one could influence the physical properties of the semiconductor – following the Mott-Schottky-Theory – to reach satisfying electrical performance.9 With this knowledge, German researchers were able within a few months – a very short time for the limited manpower involved – to construct working semiconductor rectifiers to detect the allied airplanes over Germany. In the eyes of German Physicists like Heinrich Welker and Herbert Mataré, this success had been made possible because of the good agreement of the Mott-Schottky-Theory with experiment. Therefore they concluded after the War, that the general framework of the MottSchottky-Theory was essentially correct.10 By the End of the war German specialists had the same understanding of the physics of semiconductors and their technical use as their British-American counterparts.11

Field-Effect-Transistors After having been able to replace the electron tube rectifier successfully by a semiconductor device the next almost logical step was to try to replace the overly large and energy consuming electron tube amplifier (“triode”) by a crystal amplifier. This had been attempted several times before the Second World War, but those attempts didn’t lead to the invention of any useful device because the physics of semiconductors had not yet been adequately developed. 12 However, after the war with improved knowledge and experience on semiconductors the situation seemed to have changed. On this basis, in the final months of the war, both in Germany and the United States, the concept of a crystal amplifier, the later so called field effect transistor, was developed. Welker’s and Shockley’s Field Effect Transistor Nearly at the same time, William Shockley, a physicist at the Bell Labs, and Heinrich Welker at Munich designed in a straight forward application of the Mott-Schottky-Theory a very similar device.

9

10

11

12

A standard procedure of electrical engineering to get a hold of the electrical properties of a certain complex and real arrangement is to design an equivalent circuit out of ideal components. From this one could conclude which properties the barrier layer of the metal-semiconductor contact should have to work at very high frequencies. Welker to BIOS, 15. April 1946, HWD 002. (“Die theoretischen Betrachtungen von Schottky, Davidow und Mott halte ich im wesentlichen für zutreffend.”) Compare Torrey/Whitmer, Crystal Rectifiers, 1948. At the MIT Radiation Laboratory in the last years of the war more doping and “forming” was done, but this knowledge didn’t affect the transistor development (prior to the invention). Closely following the design of the electron tube amplifier, in 1939 at the Bell Labs, the physicist William Shockley designed a copper oxide amplifier, in which a controlling grid in the barrier layer perpendicular to the current was implemented. The new device failed because “the experiments did not reveal the predicted effect.” (Hoddeson, Point-Contact Transistor, 1981, S. 61.) A similar but working device on the basis of the ionic crystal KBr had been previously built by Rudolf Hilsch and Robert W. Pohl. The used crystal reacted for technical purposes way too slowly. Thus the concept was not pushed further (Hilsch, Elektronenleitung in Kristallen, 1939; Teichmann, Farbzentren, 1988, S.92-96).

4

Kai Handel

„Uses and Limits.doc“

From a metal condenser plate an electrostatic field is applied to a thin layer of a semiconductor. By the applied field charge carriers within the semiconductor should be induced and therefore the conductivity of the semiconductor should change. Thus, the electrostatic field should control the current along the semiconductor.13 Because this controlling effect belonged to the application of an electrostatic field this device was later called : the Field Effect Transistor (FET). Independently both, Shockley and Welker performed calculations based on the MottSchottky-Theory which showed, that in their experimental arrangements this field effect should be detected.14

Figure 3: Shockley, April 1945 (left); Welker, January 1945 (right), H: thin layer of semiconductor; G: insulating support; D: thin insulating layer; E1, E2: electrodes; E3: metal condenser plate

In March 1945 Welker had been able to perform experiments with this arrangement. The semiconductor layer (copper oxide) was about 10 microns thick, the dielectric about 200 – 500 microns, platinum and silver contacts were used as electrodes. However, “only small effects” were detected.15

Shockley at the Bell Labs performed similar experiments, and concluded from his calculation that the effect he was looking for was at least three orders of magnitude smaller than expected. Disappointed, he gave up these experiments in June 1945 to turn his attention to other matters.16 So far his investigations paralleled that of Welker at Munich. In October 1945 Shockley passed his research on to John Bardeen, who had just joined the Bell Labs. In the following months, Bardeen developed a theory of surfaces states, which explained, why the expected effect in the field effect experiments could not be found. This initiated further transistor experiments, which he performed together with Walter Brattain.17 These experiments culminated in the (well known) invention of the point contact transistor at 13

14

15

16 17

The important (but not new) point about both designs, was that the control electrode was arranged parallel to the current (thus the electrostatic field was perpendicular to the current). The story of the concept of a crystal amplifier using a so called field effect is long and starts considerably before the Second World War. Already in 1926 the Polish-American Physicist Julius Edgar Lilienfeld suggested in a US Patent a devise based on copper sulfide for “controlling electric currents” by “electrostatic influence”. In 1934 the German Physicist Oscar Heil applied without success for a similar patent in Germany, but in 1935 a British patent was awarded on the device. (See Lilienfeld, Controlling Electric Currents, 1926, US Patent 1,745,175; Heil, Non-contact-making relays, 1935, Brit. Patent 439,457. For Futher information see Goetzeler, Halbleiter-Bausteine, 1972, S.39-44 and Hofmeister, 50 Jahre, 1976.) Welker, Steuerung von Elektronenströmen, 1945, HWD 003, Riordan/Hoddeson, Crystal Fire, 1997, S. 112114, Hoddeson, Point-Contact Transistor, 1981, S. 63. (Welker: for a detectable effect a very low charge carrier density in the semi conducting material would be necessary.) Welker, Beeinflussung, 1945, HWD 003, S. 3 („Die im Geschilderten bekannte Anordnung zeigte jedoch nur geringe Effekte, ...“); „Aktennotiz: Herstelllung eines Dreielektrodenkristalls“ 14. März 1945, HWD 003. Riordan/Hoddeson, Crystal Fire, 1997, S. 112-114, Hoddeson, Point-Contact Transistor, 1981, S. 63. Riordan/Hoddeson, Crystal Fire, 1997, S. 120, Hoddeson, Point-Contact Transistor, 1981, S. 63-66.

5

Kai Handel

„Uses and Limits.doc“

the Bell Labs in December 1947. The point-contact transistor could only be understood by assuming the interaction of the two different charge carriers, holes and electrons, at the same time in the semiconducting crystal. Thus, John Bardeen and Walter Brattain who had not been part of the British-American Radar Project abandoned the Mott-Schottky-Theory and developed a new understanding of the conduction process in a semiconductor.18 During this period Welker’s work in Munich took an entirely different direction. Welker concluded that his and other “known arrangements” didn’t work because the dielectric insulating layer was too thick. As an expert on the Mott-Schottky-Theory and having several years of experience in applying it successfully to crystal detectors, he looked for a solution in this theoretical framework. In April 1945 he proposed to use the barrier layer of a metalsemiconductor contact to insulate the control electrode from the semiconductor. He stated: If one would use a semiconductor of the contrary conducting type (as the main body) as control electrode, a barrier layer would form between them. This barrier would be extremely thin, in the order of microns (10-4 cm) and should be thin enough to allow the control of the current between E1 and E2.19 In modern terminology Welker suggested in his patent of April 1945 the use of a pn-junction to control the current between E1 and E2. At this point, it is somewhat important to note, that Welker conceptualized the pn-junction – completely within the framework of the MottSchottky-Theory – only as a means of creating a conveniently thin insulating layer, and therefore this was not similar to today’s main understanding of the pn-junction.20 Thus, Welker designed a device which is today known as the “Junction-Field-EffectTransistor”21. This transistor was not manufactured until the 1960s when the understanding of semiconductor physics as well as production technologies had been greatly improved. Both types, the “Junction-Field-Effect-Transistor” and the field effect transistor with insulation layer, belong today to the standard transistors realized in Integrated Circuits (IC).22 But in 1945, when Welker filed his patent no working device could be built. It is highly probable that Welker did not even seriously try to build one. Given the increasingly unstable situation in Germany during the last weeks of the War, this is not astonishing. The laboratory Welker worked in from 1942 was destroyed and evacuated in October 1944. Experiments performed in March 1945 were conducted in collaboration with a small company specializing in optical instruments. In 1946 Welker lived and worked as an “independent engineer” near Munich where a small number of Germanium diodes were fabricated. No further investigations on crystal amplifiers could be done at this time.23 In March 1946 Welker was interviewed by allied Intelligence agencies on his war work in the radar project. In consequence, he received an offer to resume his work in collaboration with

18 19 20

21

22 23

Hoddeson, Point-Contact Transistor, 1981 and Riordan/Hoddeson, Crystal Fire, 1997. Welker, Beeinflussung, 1945, HWD 003. The concept of injecting charge carriers into a semiconductor by a pn-junction was clearly developed by Shockley in 1948, the first junction-transistor could be produced from 1951 on. (Main understanding: charge carriers travel through the pn-junction.) Welker, Halbleiteranordnung, 1945, Germ. Patent 980 084. It is generally believed that such a suggestion was made first be Shockley in 1952. (Shockley, Uni-Polar Field-Effect, 1952). Hofmeister, 50 Jahre, 1976. Welker to Schottky, 14. Nov. 1944, HWD 006; „Aktennotiz: Herstelllung eines Dreielektrodenkristalls“ 14. März 1945, HWD 003; Bosch, Werdegang, 1994.

6

Kai Handel

„Uses and Limits.doc“

Herbert Mataré in a laboratory of the French Westinghouse (Compagnie des Freins et Signaux Westinghouse) at Paris. 24

Welker and Mataré at Paris During the Second World War, Herbert Mataré had worked together with Karl Seiler and others in the Telefunken Radar laboratories. When he came to Paris he had considerable experience in high frequency technology and especially with the experimental setup for measurements at point contact rectifiers. With limited funding, Mataré and Welker constructed a working research laboratory for semiconductors in order to establish the production of Germanium detectors for the French military.25 By the end of 1947, suitable rectifiers for the detection of centimeter waves with acceptable background noise were in production, finally giving Welker and Mataré more time for their own research. While Welker, disappointed with the failure of his previous field effect transistor experiments, turned his attention back to his prewar work on the theory of superconductivity, Mataré went on with his research on the cancellation of noise in microwave receivers, which had been his main task during the early years of the war. At this time, he had shown that duo-diodes reduced the noise in the reception circuit.26 Transferring this the basic concept to crystal detectors led him to experiments using a two whisker crystal. These experiments however, didn’t yield any satisfactory and reproducible results during the war.27

24

25

26

27

Welker as well as Mataré had been interviewed by both, The British Intelligence Objectives Sub-Committee (BIOS) and the Field Information Agencies Technical (FIAT).Welker/Ringer, Investigations about GeDetectors, 1946, HWD 002; Welker to BIOS/FIAT, 15. April 1946, HWD 001; BIOS-Report 725; Curriculum Vitae of Heinrich Welker in LMU E-II-N Welker; Reisegenehmigung für Welker und seine Familie, 12. April 1947, HWD 005. They both started their work at Paris around the end of 1946 or the beginning of 1947. In the end of 1946, beginning with an empty building, they began together to set up a new laboratory where Welker was concerned with the refinement of Germanium while Mataré was responsible for the high frequency design of the detectors. After setting up the laboratory they provided the necessary information for the production of the Germanium detectors. The general goal was to produce detectors similar to the at this time well known Sylvania detectors. The „Sylvania detectors“ were Germanium detectors and had been produced during the war by the Sylvania Electric Products Company. As possible prototype the diode 1N34 could be mentioned. (see Thorrey/Whittmer, Crystal Rectifiers, 1948, S. 361 ff., Mataré-Interview 1998. See also Bothelo, Industry Policy, 1994.) Mataré, Rauschen von Dioden, 1942, S. 117-122; and Mataré, Empfangsprobleme, 1951, S. 27-28. Mataré was particularly concerned with the “compensation of the oscillator noise in heterodyne receivers”. Mataré, Lesser known history, 1997, unpublished Manuscript; Mataré-Interview 1998.

7

Kai Handel

„Uses and Limits.doc“

Figure 4: “duodiode” (left), double detector (D, right) as circuit element in a push-pull circuit.28

Like other scientific work in Germany Mataré’s research was stopped by changes brought about towards the end of the war and could not be resumed before he came to Paris in the End of 1946. For noise cancellation with a duodiode it had been crucial that both diode characteristics were essentially equal. For the crystal device, this meant that the point contact characteristics should be equal. In this regard, the high purity Germanium crystals provided by Welker at Paris were much more suited than the wartime Silicon crystals available at the Telefunken laboratory in 1943. Mataré finally achieved equal rectifying characteristics at the point contacts of a two whisker crystal, but only when the point contacts were situated very close together on the crystal. 29 In performing these experiments he also detected a control influence on the rectification abilities of one contact by the other. Thus, he was able to control the current in the crystal and therefore he found a crystal amplifier.

Welker’s and Mataré’s Transistor According to Mataré’s own account, he demonstrated these experiments to his colleague Welker in the beginning of 1948. With his interest in the crystal amplifier renewed, Welker attempted to understand how the device worked.30 On August 13, 1948, Mataré and Welker filed a patent on their device, which is the first written and dated proof of their transistor activities at Paris.31 Priority This patent was filed six weeks after the famous press conference of the Bell Labs, at the end of June 1948, where the point-contact transistor was presented. I do not want to go into priority questions here – there seems to be no doubt about Bardeen’s and Brattain’s priority on the invention of the transistor.

28

29

30 31

Mataré, Empfangsprobleme, 1951, S. 27 (left) and Mataré, Lesser known history, 1997 (right). (“Gegentaktschaltung”: oscillator noise suppression mode) Mataré, Push-Pull Converter, 1948, US-Patent 2,552,052, (French application in may 1947). Mataré patented already in May a “push-pull converter of the crystal type”. Although the patented device looks considerably different to the sketch above, it demonstrates, that Mataré went on with experiments with a two whisker crystal. Welker, Es ist allgemein bekannt ... , 1948, HWD 001; Mataré-Interview 1998. Westinghouse, Welker, Mataré , Nouveau système, 1948, French Patent 1.010.427; Mataré/Welker, Crystal Device, 1949, US Patent 2,673,948. The manuscript Welker, Es ist allgemein bekannt ... , 1948, HWD 001 is not dated but has to be at least written a couple day before, more probably week or even month.

8

Kai Handel

„Uses and Limits.doc“

However, it is probable that in any laboratory concerned with high purity Germanium and with experience in experiments with point contacts, somebody would perform experiments with two point contacts close to each other. This is what Bardeen and Brattain feared and why they wanted to have a patent filed and a publication out as soon as possible. The main competition seemed to come from researchers at Purdue University who had been the experts for Germanium during the war. But it turned out that the physicists at Purdue had not been on the same track, they did not even hear about the transistor invention until late July 1948.32 Given that researchers at Purdue were so late in hearing of the developments at Bell, it is plausible that the news of the transistor invention didn’t reach Paris before Welker’s and Mataré’s patent application of August 13, 1948. It is certain however that they – unlike Bardeen and Brattain – didn’t succeed in understanding, how their device really worked. Both Welker and Mataré were too compelled by the general theoretical frame work of the Mott-Schottky-Theory and thus tried to understand their crystal amplifier in its terms. Manuscript and patent application Here you see a drawing from an undated manuscript of Welker prior to the patent application in August 1948.33

Figure 5: Welker manuscript, undated, prior to patent application of August 13, 1948. The point effect device (left) is explained as a field effect device (right). (Welker, Es ist allgemein bekannt ... , 1948, HWD 001)

32

33

See Riordan/Hoddeson, Crystal fire, 1997, p. 151- 157, 165-166. At July 20, 1948 a more technical public demonstration took place at Bell Labs. Welker, Es ist allgemein bekannt ... , 1948, HWD 001. “Eine derartige Anordnung mit der man wenigstens prinzipiell Ströme in einem Kristall steuern könnte besteht darin, daß man z.B. auf einem Detektorkristall (Abb. 1) benachbart zwei Spitzen aufsetzt. Der Abstand der beiden Spitzen muß dabei so gewählt werden, daß die eine Spitze in den Bereich der Sperrschicht der anderen fällt und umgekehrt.” The manuscript is in German, the patent claims are German and French, the drawings have French sub titles. A couple of the formulations, most of the drawings and the general arguments are identical with the patent applications. Westinghouse, Welker, Mataré , Nouveau système, 1948, French Patent 1.010.427. Mataré/Welker, Crystal Device, 1949, US Patent 2,673,948.

9

Kai Handel

„Uses and Limits.doc“

Welker closely compared the point contact device (on the left) with his former field effect arrangement (on the right). One electrode (“Electrode de commande”) should control by a field effect the barrier layer of the other electrode (“electrode porteuse de courant”) and therefore should control the current between these two electrodes (“electrode porteuse de courant”). He stated: although one could with this arrangement (figure 1) in principal control a current in a crystal, in practice it is very difficult to actually place a point contact within the barrier layer of another point contact, since the barriers are too small (of the order of 1 micron = 10-4 cm). However, since the device worked with point contacts at a much greater distance (of the order of 100 microns34), Welker and Mataré realized that a different explanation of the device was necessary. Thus, in their patent they concluded that (somehow) two zones of different conductivity characteristics are formed in the Germanium crystal. If on each zone an electrode is placed, control of the current could be explained – again as a field effect – within the general framework of the MottSchottky-Theory.36

Figure 6: “[T]he device ... comprises a semiconductive crystal 1(for example of Germanium) in which two zones 2 and 3 of different conductivity characteristics have been formed. For example, zone 2 has an n-type (excess) electronic conductivity and zone 3 a p-type (deficiency) electronic conductivity or vise-versa.” 35 From H. F. Mataré and H. Welker, US Patent 2,673,948, French Priority August 13, 1948

34 35

36

37

But in fact this device could not be understood in terms of the Mott-Schottky-Theory. The device they created was very probably a point contact transistor, in which holes are injected into the n-type Germanium, influencing the conduction characteristic of the contacts.37

Mataré-Interview 1998. Quotes from Mataré/Welker, Crystal Device, 1949, US Patent 2,673,948 but they are a word to word translation from the same paragraphs in the French patent Westinghouse, Welker, Mataré , Nouveau système, 1948, French Patent 1.010.427. Even clearer this concept is found in Welker’s suggestion (similar to his earlier junction-field-effecttransistor) to apply a p-type semiconductor (such as Selenium) on the n-type Germanium close to the point contact and to control the current trough the forming barrier layer. Mataré admits that this arrangement has been included in the patent applications although “it newer worked.” (See Welker, Es ist allgemein bekannt ... , 1948, HWD 001; Westinghouse, Welker, Mataré , Nouveau système, 1948, French Patent 1.010.427. Mataré/Welker, Crystal Device, 1949, US Patent 2,673,948.) Mataré later claimed that the transistor they created worked on the basis of grain boundaries. But there is no evidence of this in the quoted patents. Mataré’s first publications on grain boundary transistors appeared only a few years later. (Mataré, elektronisches Verhalten, 1954, Kongrenzenstruktur, 1955 und KorngrenzenTransistoren, 1956).

10

Kai Handel

„Uses and Limits.doc“

Conclusion Although I believe that Welker and Mataré had a working device prior to their knowledge of the Bell invention, this is not the most significant point, historically. Spoiled from their success with the Mott-Schottky-Theory in the production of crystal rectifiers, Welker and Mataré had not been able to understand how the device that they had created really worked. Their successful theoretical framework didn’t allow them to think of minority carriers having any influence at all on the behavior of a semi conducting device and didn’t provide the means to deal with it. As the comprehensive analysis of the Bell invention by Lillian Hoddeson has shown it was necessary for John Bardeen and Walter Brattain to understand more about surface states, to develop the notion of the “inversion layer” 38, and to realize that they had somehow altered the characteristic of the point contact by injecting holes in a n-type semiconductor. Only this enabled them to stabilize their effects and push further development. This development, however, required deviation from the single carrier Mott-Schottky-Theory.39 In conclusion, I would like to argue that while wartime developments provided increasingly pure semiconductor materials, as well as more refined experimental techniques not available before the war, material and experimental progress alone was not enough for the development of the transistor. Reliance on the previously successful theoretical framework of the Mott-Schottky-Theory hindered the understanding and invention of a crystal amplifier at Paris. Although, Bardeen and Brattain had been familiar with the Mott-Schottky-Theory, they did not experience its success on a day to day basis over several years and therefore were not so deeply indebted to this theory. Thus, they invented the point-contact transistor because they had not been involved in wartime efforts on crystal rectifiers, and were therefore not so much hindered by theoretical constraints.

38

39

Normally the concept of an inversion layer is attributed to Walter Schottky and Eberhard Spenke. Although they talked about this concept, they didn’t think of as of any importance at all. They also didn’t give any hints how to deal with it. (Schottky/Spenke, Quantitative Durchführung, 1939, p. 231) Hoddeson, Point-Contact Transistor, 1981 I omitted one more important point in my presentation above concerning the direction of the controlling effect on the current. Assuming a field effect device, the direction of the controlling effect should be just opposite to the observed effect in a point-contact device. This is explained in detail in Hoddeson, PointContact Transistor, 1981, p. 70-74. As I have shown elsewhere, Welker tended to belief in the correctness of the Mott-Schottky-Theory even against the experimental data, if the devices worked. (Handel, Halbleiterforschung und Radar, 1998, in Press)

11

Kai Handel

„Uses and Limits.doc“

Bibliography Bosch, Berthold (1994): Der Werdegang des Transistors 1929 -1994: Bekanntes und weniger Bekanntes. Vortrag aus Anlaß des 65. Geburtstages von Magnifizenz Prof. Dr. Eberhart Köhler, 17. Nov. 1994, Unveröffentlichtes Manuskript Bothelo, Antonio José J. (1994): The Industrial Policy that never was: French Semiconductor Policy, 1945-1966. History and Technology 11, S. 165-180 Goetzeler, Herbert (1972): Zur Geschichte der Halbleiter-Bausteine der Elektronik. Technikgeschichte 39:1, S. 31-50 Guerlac, Henry E. (1987): Radar in World War II. 2 Vols. New York: American Institute of Physics, Handel, Kai (1998): Halbleiterforschung und Radar im Zweiten Weltkrieg. In: Meinel, Christoph (1998): Instrument Experiment. Historische Studien (in Press). Bassum, Stuttgart: GNT-Verlag Heil, Oskar (1934): Non-contact-making relays (Abstract). Britisches Patent 439,457 Hilsch, Rudolf (1939): Elektronenleitung in Kristallen. Die Naturwissenschaften 27, S. 489-492 Hoddeson, Lillian H. (1981): The Discovery of the Point-Contact Transistor. Historical Studies in The Physical Sciences 12:1, S. 41-76 Hoddeson, Lillian H. (1994): Research on Crystal Rectifiers during World War II and the Invention of the Transistor. History and Technology 11, S. 121-130 Hoddeson, Lillian H.; Braun, Ernest; Teichmann, Jürgen; Weart, Spencer (1992) (Hg.): Out of the Crystal Maze. Chapters from the History of Solid-State Physics. New York, Oxford: Oxford University Press, Hofmeister, Ernst (1976): 50 Jahre Feldeffekttransistor. Funkschau 20, S. 857-860 Kaiser, Walter (1996): A Case Study in the Relationship of History of Technology and of General History: British Radar Technology and Neville Chamberlain's Appeasement Policy. ICON 2, S. 29-52 Kern, Ulrich (1984): Die Entstehung des Radarverfahrens: Zur Geschichte der Radartechnik bis 1945. Stuttgart: Dissertation am Historischen Institut der Universität Stuttgart, Lilienfeld, Julius Edgar (1926): Method and Apparatus for Controling Electric Currents (Abstract). US Patent 1,745,175 Mataré, Herbert F. (1942): Das Rauschen von Dioden und Detektoren im statischen und im dynamischen Zustand. Elektrische Nachrichtentechnik 19:7, S. 111-126 Mataré, Herbert F. (1948): Push-Pull Converter of the Crystal Type for Ultra-Short Waves. United States Patent Nr. 2,552,052 Mataré, Herbert F. (1951): Empfangsprobleme im Ultrahochfrequenzgebiet unter besonderer Berücksichtigung des Halbleiters. München: Oldenbourg, Mataré, Herbert F. (1954): Elektronisches Verhalten bestimmter Korngrenzen in perfekten Kristallen. Zeitschrift für Naturforschung 9a:7/8, S. 698 Mataré, Herbert F. (1955): Korngrenzenstruktur und Ladungsträgertransport in Halbleiterkristallen. Zeitschrift für Naturforschung 10a:8, S. 640-652 Mataré, Herbert F. (1956): Korngrenzen-Transistor. Elektronische Rundaschau 10:8-9, S. 209-211 und 253-255 Mataré, Herbert F. (1997): The lesser known history of the crystal amplifier. Unveröffentlichtes Manuskript Mataré, Herbert F.; Welker, Heinrich J. (1949): Crystal device for controlling electric currents by means of a solid Semiconductor. US Patent 2,673,948 Reuter, Frank (1971): Funkmeß. Die Entwicklung und der Einsatz des RADAR-Verfahrens in Deutschland bis zum Ende des Zweiten Weltkrieges. Opladen: Westdeutscher Verlag, Riordan, Michael; Hoddeson, Lillian H. (1997): Crystal Fire. The Birth of the Information Age. New York, London: W.W. Norton & Company, Schottky, Walter (1938): Halbleitertheorie der Sperrschicht. (Kurze Originalmitteilungen) . Die Naturwissenschaften 26, S. 843 Schottky, Walter (1939): Zur Halbleitertheorie der Sperrschicht- und Spitzengleichrichter . Zeitschrift für Physik 113, S. 367414 Schottky, Walter (1942): Vereinfachte und erweiterte Theorie der Randschichtgleichrichter. Zeitschrift für Physik 118, S. 539-592 Schottky, Walter; Spenke, Eberhard (1939): Zur quantitativen Durchführung der Raumladungs- und Randschichttheorie der Kristallgleichrichter. Wissenschaftliche Veröffentlichungen aus den Siemens-Werken 18:3, S. 225-291 Shockley, William B. (1952): A Unipolar 'Field Effekt' Transistor. Proceedings of The I.R.E. 40, S. 1365-1376 Sueur, R. (1949): Le Transistron Triode Type P.T.T. 601. L'Onde Électrique 29, S. 389-397 Torrey, Henry C.; Whitmer, Charles A. (1948): Crystal Rectifiers. New York, London: McGraw-Hill Book Company, Trenkle, Fritz (1979): Die deutschen Funkmeßverfahren bis 1945. Stuttgart: Motorbuch-Verlag, Weiß, Herbert (1975): Steuerung von Elektronenströmen im Festkörper. Physikalische Blätter 31, S. 156-165 und 209-212 Welker, Heinrich J. (1945): Beeinflussung und Steuerung von elektrischen Strömen in Halbleitern. Unveröffentlichtes Manuskript, HWD 003 Welker, Heinrich J. (1945): Halbleiteranordnung zur kapazitiven Steuerung von Strömen in einem Halbleiterkristall. Patent DBP 980 084, 21g, 11/02 [Aktenzeichen: P 980 084.4-33 (F Welker, Heinrich J. (1945): Zur Steuerung von Elektronenströmen in einem Dreielektrodenkristall I. Unveröffentlichtes Manuskript, HWD 003 Welker, Heinrich J. (1948): Es ist allgemein bekannt ... Unveröffentlichtes Manuskript, HWD 003 Welker, Heinrich J.; Ringer, Walter (1946): Investigations about Ge-Detektoren (Manuskript für B.I.O.S). Unveröffentlichtes Manuskript, HWD 002 Westinghouse, Compagnie des Freins et Signaux; Welker, Heinrich J.; Mataré, Herbert F. (1948): Nouveau système cristallin à plusieurs électrodes réalisant des effets de relais électroniques. Französiches Patent 1.010.427

12