Copyright Mark P D Burgess 2008
Two giants of the history of science act as poles in the story of the science of semiconductors. Michael Faraday is generally credited with making the first observation of one of the defining properties of a semiconductor: In 1833 he discovered that the “conducting power” of silver sufide increases with increasing temperature whereas the conductivity of metallic conductors decreases as their temperature rises. By the time William Shockley received the Nobel Prize for his work on semiconductor devices in 1956 the theory of semiconductors was well understood. This together with the industrial technologies needed to manufacture semiconductor devices, provided a secure platform for the rapid technological advances seen in that decade.
The history can be framed around three periods
It took a century after Faraday’s work to establish a coherent view of what semiconductors actually were and to explain their important properties. Progress was slow while a platform for later work was created from the astute observations of skilled chemists and physicists. In 1873 Willoughby Smith discovered photoconductivity while working on selenium and three years later William Adams and Richard Day discovered its photovoltaic effect. Charles Fritts used selenium to produce a low efficiency photovoltaic cell. In 1874 Braun observed the rectification properties of galena and Schuster made similar observations in copper oxide. Lastly Henry Round discovered that some carborundum detectors emit light in 1907.
Thus by the beginning of the 20th Century the five properties now associated with semiconductors had been discovered but scarcely characterised:
The Early Commercialisation Era
Then follows the first commercialization era led by the need for detectors for wireless telegraphy by Marconi and the pioneers of commercial wireless. Jagadis Bose patented the cats whisker crystal detector in 1901 and in the same decade Greenleaf Pickard undertook large scale screening of new detector materials including a new entrant: Silicon in 1906.
Virtually no further development occurred for nearly 30 years. The crystal detector was an unreliable device requiring careful adjustment for best sensitivity. The vacuum tube had developed on a parallel track since Edison in 1883 and its first use by Fleming for the detection of radio frequencies in 1904. By the 1920s the vacuum tube was king and the crystal detector was relegated for the low end of the wireless receiver market: the crystal set.
In 1922 Union Switch and Signal, a Westinghouse company filed its first patent on the copper copper-oxide rectifier which was used both high power devices and small current applications. The selenium rectifier was introduced in 1933 and found a niche in high tension power supplies. Photocells based on selenium were used in light meters
The development of radar required the re-invention of the cats whisker detector in the late 1930s and through World War II in the form of cartridges and glass encapsulated units.
These gains were technologically modest depending on good engineering, new purification technologies (germanium and silicon for example) and pressing war time demands. To make further progress a better understanding of the nature of semiconductors was needed.
Controversy around the intrinsic nature of semiconductors remained through to the 1930s. It was thought some observations were simply artifacts of the role of uncontrolled impurities. Indeed many were: progress remained impeded both by the lack of high purity experimental materials and a theoretical framework that could describe their properties.
In the 1931 Alan Wilson published the first quantum mechanical description of semiconductor solids and later in 1938 Walter Schottky, Boris Davydov and Nevill Mott advanced the theoretical description of rectification.
Then the efforts of solo investigators gave way to the power of the industrial laboratory: Bell set up the Solid State Project with the objective of producing a solid state switch that could be used in telephone exchanges to replace mechanical switches. This led to the development of the point contact transistor in 1947 and the junction transistor in 1950. The Bell team were led by William Shockley. Perhaps for the first time in the history of semiconductors the theoretical framework preceded a practical device: Shockely developed his theory of the P-N junction in 1948 and that led to the development of the junction transistor two years later. But the semiconductor era was heralded by the point contact transistor which Bell announced on July 1st 1948:
“An amazingly simple device, capable of performing efficiently nearly all the functions of an ordinary vacuum tube, was demonstrated for the first time yesterday at Bell Telephone Laboratories where it was invented.”
In the following section the work of the early experimentalists is described. Its theme is that for the most part their work was purely observational and quite isolated. Worse, experiments were confounded by the inability of the early pioneers to control the materials they worked with. It is only with the understandings from the mid 20th Century that one can make sense of earlier observations.
But their contribution was important: it created the experimental body of knowledge that the physicists of the 20th Century needed to motivate their work.
The work by Faraday on the unusual conductivity of silver sulfide is an obscure postscript in a remarkable investigatory career. He is notable for his studies in electricity, electrochemistry and magnetism.
He reported his work on silver sulfide as part of a much wider treatise on conduction to the Royal Society in 1833 drawing on it as an exception to the body of his work. He points out that it was already known in electrostatics that insulators show increasing conductivity with temperature and that metals show decreasing conductivity as their temperature rises:
“ The effect of heat in increasing the conducting power of many substances, especially for electricity of high tension, is well known. I have lately met with an extraordinary case of this kind, for electricity of low tension, or that of the voltaic pile, and which is in direct contrast with the influence of heat upon metallic bodies, as observed and described by Sir Humphry Davy.”
Faraday then describes his method of preparing silver sufide (suphuret of silver) by the repeated high temperature fusion of silver and sulphur. He demonstrates the conductivity of the fused suphuret by connecting it to a voltaic pile battery of about 30 volts and observing the current using a galvanometer:
“On applying a lamp under the sulphuret between the poles, the conducting power rose rapidly with the heat, and at last-the galvanometer needle jumped into a fixed position, and the sulphuret was found conducting in the manner of a metal. On removing the lamp and allowing the heat to fall, the effects were reversed, the needle at first began to vibrate a little, then gradually left its transverse direction, and at last returned to a position very nearly that which it would take when no current was passing through the galvanometer.
Occasionally, when the contact of the sulphuret with the platina poles was good, the battery freshly charged, and the commencing temperature not too low, the mere current of electricity from the battery was sufficient to raise the temperature of the sulphuret; and then, without any application of extraneous heat, it went on increasing conjointly in temperature and conducting power, until the cooling influence of the air limited the effects. In such cases it was generally necessary to cool the whole purposely, to show the returning series of phenomena.”
“There is no other body with which I am acquainted, that, like sulphuret of silver, can compare with metals in conducting power for electricity of low tension when hot, but which, unlike them, during cooling, loses in power, whilst they, on the contrary, gain. Probably, however, many others may, when sought for, be found.”
Thus in his account Faraday makes it clear that he has discovered a compound with poor room temperature conductivity which at elevated temperatures is comparable to metallic conductors. He also reports the thermal runaway effects that plagued the early germanium transistors. He does not report any quantitative data and it is clear from his description that the effects while consistent were dependent on the precise manner of the preparations. This is an issue that dogged the science of semiconductors through to the 1930s.
Hittorf studied the temperature effects on conductivity of silver sulfide and cupric sulfide as part of much wider investigations into electrolytes and gases. His preparations were made in the manner of Faraday: by direct fusion of the respective elements. Unlike Faraday he did make meticulous measurements of resistance and temperature and published these in 1851 showing very significant increases in conductivity for both materials.
Reviewing the work over a century later Georg Busch has re-presented Hittorf’s data in conventional logσ against reciprocal temperature and notes that Hittorf’s results for cupric sulfide gives good agreement with the expected results for a conventional semiconductor but silver sulphide less so. Hittorf’s results indicate his silver sulfide was impure: an expected phase change at 170oC is not observed. Above this temperature silver sulfide is no longer semiconducting.
Thus while it is customary to credit Faraday with the first discovery of a semiconducting effect it took Hittorf to obtain precise data that demonstrates this in an unequivocal manner.
Faraday M 1833 On a New Law of Electric Conduction Phil Trans R Soc 123 507-15
Busch G 1989 Early History of the Physics and Chemistry of Semiconductors- From Doubts to Fact in a Hundred Years Eur J Phys 10 254-64
Smith was an electrical engineer who worked on several major under-sea cable laying projects. While developing a system of testing the cables while they were being laid Smith needed a means by which the high resistance of a long cable could be simulated reliably. He obtained some resistors which were made from selenium wire hermetically sealed in glass tubes connected to platinum wires at each end of the tube. Selenium has a high resistance and Smith notes that wires of up to 100 mm long and 1.0 mm in diameter gave a resistances of up to1400MΩ. The construction of Smith’s resistors should have given reliable results but this was not the case as he notes in his letter to Nature:
“...there was a great discrepancy in the tests, and seldom did different operators obtain the same result. While investigating the cause of such great differences in the resistance of the bars, it was found that the resistance altered materially according to the intensity of light to which they were subjected. When the bars were fixed in a box with a sliding cover, so as to exclude all light, their resistance was at its highest, and remained very constant, fulfilling ail the conditions necessary to my requirements; but immediately the cover of the box was removed the conductivity increased from 15 to 100 per cent, according to the intensity of the light falling on the bar. Merely intercepting the light by passing the hand before an ordinary gas-burner, placed several feet from the bar, increased the resistance from 15 to 20 per cent. If the light be intercepted by glass of various colours, the resistance varies according to the amount of light passing through the glass.
To ensure that the temperature was in no way affecting the experiments, one of the bars was placed in a trough of water so that there was about an inch of water for the light to pass through, but the results were the same; and when a strong light from the ignition of a narrow band of magnesium was held about 9 in above the water the resistance immediately fell more than two-thirds, returning to its normal condition immediately the light was extinguished.
Thus Smith discovered a potent photoconductivity effect and inspired extensive research into selenium and commercial applications.
Smith concludes quite wrongly that “temperature was in no way affecting the experiments” because he did not look for a temperature dependent effect. Had he done so he would have found a significant temperature effect over a few degrees: enough to create new errors in his measurements.
Selenium presents in several allotropic forms of which only one, the most stable and common trigonal form is a semiconductor. Accordingly it is highly temperature sensitive.
Later, in 1885 Shelford Bidwell reported doubts about the proposition that these observations were due to the intrinsic properties of selenium speculating that the results could relate to impurities introduced by the electrodes and the annealing process.
Noting that selenium and sulfur are closely related he constructed cells of sulfur with silver added to improve its conductivity.
In doing so he prepared cells of silver sulfide suspended in excess sulfur and obtained readily measurable photoconductivity effects.
“How far this may be considered to prove anything with regard to selenium I do not know; but in any case the discovery of another substance possessing the same remarkable property seems in itself to be a matter of some interest.”
Smith W Effect of Light on Selenium during the passage of an Electric Current
Short Biography of Willoughby Smith http://www.geocities.com/neveyaakov/electro_science/smith.html
Bidwell S 1885 On the Sensitiveness of Selenium to Light and the Development of a Similar Property in Sulphur Proc. Phys. Soc. London 7 129-145
Most reviews of the history of semiconductor research credit Becquerel with noting the first photovoltaic effect. He observed potential differences between electrodes where one is illuminated and one is not. But it is difficult to deduce whether he was observing photoelectric or photochemical reactions and therefore whether his work should be given credit in the context of the history of semiconductors. For example, he concludes that he has observed “decomposition of chloride, bromide and iodide of silver under the influence of light, produces electrical effects which can be used to determine the quantum of the active “chemical radiation”.
Adams and Day prepared their selenium cells by taking a short piece of vitreous selenium 5 mm to 25 mm in length and pressed red hot platinum electrode wires formed into a circle into each end of the rod thereby forming a good connection to each end. The assembly was enclosed in a glass tube and sealed with corks prior to annealing to convert the selenium to the conducting (trigonal) form.
In 1876 they first report three rectifying junction properties: Reducing resistance with increasing junction potential (at constant temperature), unidirectional resistivity and photoconductivity. They attributed the direction of the effect to an initial polarization current creating a
“set of the molecules, in consequence of which the passage of the current through the selenium during the experiment is more resisted in that direction.”
They noted that exposing the conducting cell to light either increased or decreased the current which led to a new hypothesis: The question at once presented itself as to whether it would be possible to start a current in the selenium merely by the action of light.
Accordingly the same piece of selenium was connected directly with the galvanometer. While unexposed there was no action whatever. On exposing the tube to the light of a candle, there was at once a strong deflection of the galvanometer needle. On screening off the light the deflection came back at once to zero.”
They go on to describe further investigations designed to eliminate confounding effects. Interpreting their results as indicating “electrolytic” conduction after Faraday that would lead to polarization of the cell (electrolysis) they carried on with investigations on freshly prepared cells. With these they eliminated possible thermoelectric effects and thermal heating from their light source. By focusing the light on various parts of the cell they showed variable sensitivity of the selenium and that the direction of the current depended on which end of the cell was illuminated.
While some of their observations are difficult to interpret the central finding is clear and convincing: the first photovoltaic effect had been demonstrated in the first widely recognized semiconductor.
Becquerel A.E 1839 Recherches sur les effets de la radiation chimique de la lumiere solaire au moyen des courants electriques Comptes Rendus de L´Academie des Sciences 9 145-149
Becquerel A E 1839 memoire sur les effets d’electriques produits sous l’influence des rayons solaires Comptes Rendus de L´Academie des Sciences 9, 561-7
Adams W Day R 1876 The Action of Light on Selenium Proc R Soc 25 113-7
Braun was prompted to report some experimental observations in 1874 on the conductivity of metal sulfides after reading a paper by Herwig in the same year. He states that Herwig in Several Observations of Iron and Steel Rods in Galvanic Currents found variations of up to 0.03% in resistance according to the direction, intensity and duration of a current.
Braun makes it clear he is reticent in producing his results prior being able to consistently replicate and explain his experiments with any precision. He did, however go to considerable lengths to eliminate poor electrical contacts and stated he found no evidence for a thermoelectric effect or polarization. He found:
“With a large quantity of natural and artificial metal sulfides and greatly varying pieces, the most perfectly formed crystals that I could find as well as course samples, I discovered that their resistance varied with the direction, intensity and duration of the current. The differences amount to up to 30% of the total amount.”
Braun had found a curiosity rather than anything of practical application. While he was convinced he had found uni-directional conductivity he found that the current would often decrease over time. His tables of results show the initial current and then the steady state current and that these may also differ by the order of some 30%.
Schuster published On Unilateral Conductivity also in 1874. With hindsight his discoveries are generally interpreted as being due to tarnished copper or copper oxide.
“While I was engaged in other work I met with an irregularity which seemed to me to be of such a peculiar nature that I subjected it to a separate investigation. The results of this investigation have not been entirely satisfactory. I have not been able to raise the phenomenon to which I allude above the rank of an irregularity; that is to say, I am not able to produce it on my own will, although when it is present I am generally able to destroy it. My experiments, however, leave no doubt as to the facts, and they show clearly that, in a circuit composed entirely of copper wires, joined together by means of binding-screws, the electric conductivity may be different in opposite directions”
“I have called this phenomenon 'unilateral conductivity', and I have tried to bring it into connection with known facts. The most plausible explanation seemed to me to be, that a thin layer of air may sometimes intervene between the two wires which are screwed together.”
Bose undertook investigations into “Herzian Waves” (ie radio waves) using wavelengths down to 2.5 cm. Like Herz, his objective was to show that radio waves belonged to the same class of electromagnetic radiation as did light and experimentally such short wavelengths were simpler to deal with. He repeated the classical experiments from optics in the radio spectrum. In doing so he established a reputation to be the first investigator to develop techniques for very short wavelengths later adopted for radar.
Bose experimented with new detectors and was the first investigator to use crystal detectors for radio. In Figure 1 from his 1901 patent 755840 a design based on a pair of galena crystals is shown that detects both radio and light waves. It is ironical that Bose filed a patent at all since he believed that his research was for the public good but it is notable as the first patent for a semiconductor device.
Bose has traditionally been credited with the first use of crystal detectors for wireless but Seitz and Einspruch have noted that in a publication by Braun in 1906 he reports work going back to the 1870s. In the 1906 paper he also includes experiments carried out on rectification in 1901 using psilomelane, (hydrated manganese oxide).
Braun F 1874 Über die Stromleitung durch Schwefelmetalle Ann Phys Chem 153 4 556-63 translated in Sze S M 1991 Semiconductor Devices: Pioneering Papers World Scientific Publishing 377-80
Braun F 1906 Ein neuer Wellenanzeiger (Unipolar Detektor) Elektrotechnische Zeitschrift 52 1199)
Schuster A 1874 On Unilateral Conductivity Phil Mag 48 556-563
Seits F Einspruch G 1998 First Use of Crystal Rectifiers in Wireless Proc Am Phil Soc 142 4 639-42
Sengupta D Sarkar T 1998 Centenial of the Semiconductor Diode Detector Proc IEEE 86 235-43
Henry Round was employed by the Marconi Company working on many aspects of the development of early commercial wireless. In the course of work on carborundum detectors he found they glowed with orange, yellow green or blue light on applying 10 to 110 volts. In his only recording of this finding he wrote in Electrical World:
“During an investigation of the unsymmetrical passage of current through a contact of carborundum and other substances a curious phenomenon was noted. On applying a potential of 10 volts between two points on a crystal of carborundum, the crystal gave out a yellowish light. Only one of two specimens could be found which gave a bright glow on such a low voltage, but with 110 volts a large number could be found to glow. In some crystals only edges gave the light and others gave instead of a yellow light green, orange or blue. In all cases tested the glow appears to come from the negative pole, a bright blue-green spark appearing at the positive pole. In a single crystal, if contact is made near the center with the negative pole, and the positive pole is put in contact at any other place, only one section of the crystal will glow and that same section wherever the positive pole is placed.
There seems to be some connection between the above effect and the e.m.f. produced by a junction of carborundum and another conductor when heated by a direct or alternating current; but the connection may be only secondary as an obvious explanation of the e.m.f. effect is the thermoelectric one. The writer would be glad of references to any published account of an investigation of this or any allied phenomena.”
This is Round’s verbatim report. He does not proceed with any further work.
From 1920-1936 Losev worked as a radio research technician in several soviet research laboratories. Egon Loebner argues that he is the true inventor of the light emitting diode (LED).
In 1922 he discovered Round’s phenomena independently while working on oscillating point contact crystal detectors reporting a green light from a reversed bias steel-carborundum diode (published in 1923).
But Losev had the ability and resources to investigate the phenomena thoroughly. A year later he published microphotographs of the light emitting region, the spectrum of the emitted light and the current density necessary for emission. Later, through to 1930 he worked on single carborundum crystal slices profiling the active light emitting layer.
Round H 1907 A Note on Carborundum Electrical World 19 February 9 309
Loebner E 1976 Subhistories of the Light Emitting Diode IEEE Trans Electron Dev 23 675-99
In the story to date the Experimentalists were driven by an interest in science and observations rather than any intent about the potentially useful outcomes of their work. From around the beginning of the 20th Century a commercial motive becomes apparent from the use of patent applications to protect the work done by investigators. For the first time some were employees of companies with considerable resources directed to the research. Commercial research led to a reliable crystal detector, the copper-copper oxide rectifier, the selenium rectifier and the selenium photocell for light meters in the period 1906 – 1933. Then in the period from the 1930-1945 the need for reliable detectors for war time radar led to intensive work on silicon and germanium, their purification and doping and the development of robust point contact diodes.
Dunwoody filed a patent on a “Wave Responsive Device” acting as a detector in a tuned or unturned wireless receiver. The detector was one of 10 distinct configurations using carborundum (which might be biased with a battery). Of the various configurations that the patent includes one is close to the familiar “cats whisker” in a rather crude form (Figure 7). Here one end of the carborundum is firmly connected to one terminal while the other “simply rests in contact.” Dunwoody also describes the use of electrolytes as part of the method of making a connection to the carborundum. The patent is directed to the use of carborundum and there is no specification as to the selection of the metal connectors.
Between 1902 and 1906 Pickard worked on wireless communications while employed by AT&T. He is said to have screened up to 30,000 materials in a systematic search for efficient detectors for wireless telephony.
Pickard made two significant contributions to commercial wireless: He developed and patented designs for crystal detectors that endured over the following 20 years and developed all the common crystal materials that were used in these detectors (silicon, zincite, molybdenite bornite, chalcopyrite, and molybdenite
His progress can be charted by his patents.
His Silicon patent 836531 proposed that the crystal be of a non-metallic natural element such as silicon. The rectifying contact is a spring loaded metal point set in a ball joint. This arrangement would enable a sensitive spot on the crystal to be found.
In 1907, Pickard formed the Wireless Specialty Apparatus Company to produce and market his detectors. His trade name was the very well known “Perikon” from "perfect Pickard contact."
The patents that follow cover new rectifying crystals and the instruments that they were set in for convenient operation.
Patent 886154 in 1907 covered the use of fused zinc oxide broken open to utilize the crystalline interior. The crystal is mounted in a metal cup using fusible metal such as woods metal which became a standard method of mounting crystals through to the 1920s.
In patent 904222 he claims the use of “thermo-electric regenerative” class of conductors such as molybdenite (suphide of molybdenum). Here Pickard characterises molybdenite as having a high specific resistance and notes that its components had both good conductivity (molybdenum) and excellent insulating properties (sulphur). The contact conductor must give “high thermoelectric power” hinting at some underlying principals of successful operation.
Pickard revisited the carborundum detector following Dunwoody but claiming a single crystal detector with applied external bias (912613 February 1909). Pickard describes a single selected crystal electroplated on one side with copper, for example, to provide a good electrical contact or alternatively set in a fusible metal base. The crystal is biased with 1-3 volts to obtain operation on the steepest portion of its conductance/bias curve (eg where conductance changes by 100% for a change in bias of only 0.1 volt). In his example shown this is around 1.0 volts.
In this same year he introduced the crystal pair detector using zinc oxide with a chalcopyrite crystal contact (912726 Feb 1909) and later in 1914 further examples of numerous crystal pair detectors (1118227)
The well known cat’s whisker was patented in 1914 (1104073) and used a platinised gold contact on a crystal in the familiar horizontal form that became a standard design through to the 1920s. By then its use was relegated to low cost crystal sets by the emergence of the vacuum tube
Lars Grondahl at Union Switch and Signal, a Westinghouse company, was working on electrical switch failures and noticed that tarnished copper had the properties first noticed by Schuster (On Unilateral Conductivity) nearly 50 years earlier in 1874.
In 1922 Lars Grondahl filed his first patent on the copper copper-oxide rectifier which was used both high power devices and small current applications (US Patent 1,640,335).
The patent shows one application as a bridge rectifier for battery charging. The rectifier consists of copper washers with cupric oxide formed on one side, forming the rectifying junction. Each washer was sandwiched by a soft metal contact such as lead foil that gave a low resistance contact. More copper washers stacked together gave improved reverse voltage performance. More current could be carried by increasing the size of the washers and having them serve as cooling fins as shown in the drawing from Grondahl’s patent.
Metal rectifiers were also used as detector diodes in tube radio receivers. The WX6 Westector was a typical example.
Whereas the Russian Losev’s work on light emitting diodes (The Experimentalists of the 19th and early 20th Century) was never commercialized, his crystadyne negative resistance diode and radios were.
His diodes were made from zincite with a steel point contact. Loebner writes Losev “built over 50 radio receivers, incorporating his own tuning, heterodyning, and frequency converting circuits.” This is only marginally “commercial” but Loebner enthusiastically notes “Losev’s crystadyne radios and detectors were exhibited during the mid-twenties at major European radiotechnology expositions.”
The well known publisher and editor, Hugo Gernsback covered the development in sensational terms: “Stated in a few words, the invention encompasses an oscillating crystal. A special form of crystal in a special arrangement is now made to oscillate just exactly as does a vacuum tube. It is now not only possible by means of this invention to receive radio impulses, but to generate and transmit radio waves as well….In other words the crystal now actually replaces the vacuum tube.”
But Losev never made an impact. Loebner quotes a colleague of Losev’s, Professor Ostroumov, saying that there were difficulties with the crystadyne amplifiers. The best zinc oxide crystals came from New Jersey USA and that was problematic for Stalinist Russia. “Furthermore, tubes had become the dominant amplifying device family in radio receivers by that time.” Gernsback’s confident prediction was undone. The excitement was typical of the hype attached to the development of radio in the 1920s.
Selenium rectifiers offered the advantage of a higher reverse breakdown voltage: 20-30 volts compared to around only 5 volts for copper oxide.
They did however, suffer from a high leakage current until Albert Lotz in his 1938 patent 2,121,603 (assigned to Westinghouse) described a method of treating the selenium layer to reduce this.
Selenium rectifiers found a niche in high-tension rectifiers for radio and TV power supplies They were also used in EHT power supplies for TV which needed to operate at 20-30Kv. In this application thousands of tiny selenium-iron discs were assembled in ceramic tubes.
Russell Ohl began at Bell Laboratories in 1927 with responsibilities to keep Bell informed on advances in radio. At this time radio was still quite primitive. For example, RCA had just announced its first commercial tetrode the UX222, a development that reduced inter-electrode capacitance allowing better high frequency performance. By the 1930s Bell was working on high frequency radio that would prove vital for radar and microwave communications and Ohl was assigned to study crystal detectors as an alternative to vacuum tubes. Crystals offered better high frequency performance.In a study of over 100 materials Ohl found that silicon was the best performer in an interesting echo of the work of Pickard in 1906. In order to obtain more reliable performance it was clear that higher purity silicon was needed and in 1937 Ohl began a programme to find methods of its purification. Bell Labs tried melt purification in which silicon is heated above its melting point and allowed to re-crystalise on cooling. This is a classical purification technique in which the first crystals formed have improved purity and the contaminants are left in the melt. It is exceptionally difficult in the case of silicon due to the very high temperatures needed (over 1600C). Bell metallurgists developed crude techniques to do this and delivered Ohl crucibles of fused silica with a high purity zone at the bottom (marked “N” in the drawing).
While it was clear that the effects obtained were artifacts of the separation of impurities no one had any idea what these might be. Riordan credits Theuerer at Bell as recognizing the odour of phosphine (from phosphorus) when cutting N silicon but more work was needed to establish why that was an important observation. The PN junction was kept secret at Bell. War time work at other institutions to develop pure doped silicon and germanium for radar detectors was not informed by this insight.
Ohl patented his work on the light sensitive silicon (U S Patent 2,402,662).
The development of radar began in the 1930s and was intensified hugely as part of the war effort in the 1940s. Conventional tubes could operate at up to around 300Mhz which was barely sufficient for a useful radar image. With the development of the cavity magnetron operation at around 3Ghz was possible but at such high frequencies tube mixers for the detection of the reflected radar signal were impractical.
Point contact diodes were the answer and the first of these to be developed in the modern era were by British Thompson Houston who made them from commercial silicon of 98% purity. The British company GEC also made silicon diodes using higher purity silicon which they doped with aluminium or beryllium. They were known as red spot detectors. Both were made from poly-crystalline silicon.
These early diodes were unreliable and depended on finding a rectifying spot in the manner of the cat’s whisker detector of the early 20th Century. The answer was considered to be in high purity semiconductors and development moved to the USA. US Government research contracts were let to over 30 institutions. Seitz of the University of Pennsylvania worked on high purity silicon with du Pont using a chemically pure precursor (silicon tetrachloride). Lark-Horovitz at Purdue made high purity germanium and tried many dopants identifying the expected Group III and Group V elements. He concluded that germanium doped with either phosphorus or antimony was an effective detector. Schaff and Theuerer of Bell Laboratories made both silicon and germanium for diode production. Researchers at Pennsylvania and Purdue studied dopants and their effect on the performance of the diodes produced.
Western Electric was the manufacturing side of the AT&T Group and Bell Labs was its research wing. Thus Bell radar diode designs were made by Western Electric. War time production rose from 2000 per month in 1942 to 50,000 units monthly by 1945. An example of a diode cartridge from 1945 is shown here.
Brittain J 1995 Greenleaf W Pickard and the Eclipse Network Proc IEEE 83 10
Dunwoody 1906 U.S. Patent 837,616 Wireless telegraph system
Gernsback H 1924 A Sensational Radio Invention Radio News 1924 291; The Crystodyne Principle Radio News 1924 294-5, 431 Reproduced at http://earlyradiohistory.us/1924sens.htm
Grondahl L1922 US Patent 1,640,335 Unidirectional Current Carrying Device
Lee T 2007 The (Pre-) History of the Integrated Circuit: A Random Walk Solid State Circuits Society April 2007
Loebner E 1976 Subhistories of the Light Emitting Diode IEEE Trans Electron Dev 23 675-99
Lotz A 1938 US Patent 2,121,603 Method of Producing Selenium Rectifiers
Morris P 1990 A History of the World Semiconductor Industry Peter Peregrinus Ltd
Ohl R 1946 U S Patent 2,402,662 Light-sensitive electric device filed 1941
Petritz R 1962 Contributions of Materials Technology to Semiconductor Devices Proc IRE 50 1025-38
Pickard G 1906 US Patent 836,531 Means for receiving intelligence communicated by electric waves
Pickard G 1907 US Patent 886,154 Oscillation Receiver
Pickard G 1908 US Patent 904,222 Oscillation Detecting Means for Rectifying Intelligence Communicated by Electric Waves
Pickard G 1909 US Patent 912,613 - Oscillation detector and rectifier
Pickard G 1909 US Patent 912,726 - Oscillation receiver
Pickard G 1914 US Patent 1,118,228 - Oscillation detectors
Riordan M, Hoddeson L 1997 The Origins of the PN Junction IEEE Spectrum June 46-51
It took until the 1930s before there was any general consensus regarding semiconductors and whether their commonly observed properties (negative temperature coefficient of resistance, photoconductivity, rectification photovoltaic effect and electroluminescence) had any common origin.
This is scarcely surprising:
· Observations were confounded by the exceptional sensitivity of semiconductor materials to very low levels of impurities which made it difficult or impossible to replicate results in different laboratories.
· Some observations related to the body properties of the semiconductor and others to its surface properties.
· It was difficult to determine between the properties of a semiconductor material and artifacts created by making electrical connections (which could be ohmic or rectifying).
· It took until the beginning of the 20th Century for a systematic approach to be taken to researching these materials as new tools such as the Hall Effect made it possible to understand the nature of conduction.
· The platform of quantum mechanics was needed to provide a basis for the work of Alan Wilson, Walter Schottky, Boris Davydov, Nevill Mott in the 1930s and their successors.
In 1885 Bidwell commenting on the work on Selenium of Smith and Adams & Day suggested that their observations could be due to the platinum electrode and not the selenium. He suggested that during the annealing process the selenium reacted with the metal electrode to form metal selenides. Platinum selenide is a semiconductor material and Bidwell found it reacted slightly with selenium at annealing temperatures. Thus the proposal that selenide doping could be important was potentially insightful.
Pierce undertook an extensive review of the properties of crystal rectifiers publishing his results over the period 1907-09. He found that there was no evidence for earlier hypotheses that the rectifying effect was due to an internal bias due to thermoelectric effect.
The effects of temperature and light on the conductivity of selenium occupied researchers in a maze of conflicting experiments virtually through to the 1930s. Some of the extraneous effects may have been due to the use of other forms of selenium which are not conductors.
Pierson reviewing the literature of the photoconductive effects of selenium in 1927 noted that “in the past 20 years approximately 100 scientific papers have been published presenting experimental data as to the light sensitive property of selenium. The results have been masked by extraneous effects to such an extent that they have been so contradictory in nature that it has been impossible to establish any one accepted theory.”
Mendeleev in The Periodic Table (1869) predicted the existence of an element that would lie between silicon and tin which he later called “ekasilicon.”
In 1886 Clemens Winkler announced “argyrodite (a recently discovered mineral) contains a new element, very similar to antimony and yet sharply distinguished from it, to which the name "germanium" shall be given.” The first announcement was quickly followed by a correction regarding the supposed similarity with antimony concluding: “the new element is no other than the eka-silicon predicted fifteen years ago by Mendeleev.”
In 1915 Carl Benedicks showed that the conductivity of Germanium lay between that of silicon and tin confirming the predictive power of Mendeleev’s work. Benedicks showed that germanium could be used in a point contact crystal rectifier using platinum and copper contacts (see Jenkins 2005).
Merritt in 1925 studied rectification by germanium noting that Mendeleev put it in the same group as silicon “which is one of the best rectifying materials known.” He used an extensive range of contact metals looking for inferences around the thermoelectric effect, a theory favoured by early workers. He could not deduce any association with the thermoelectric effect but as a result of heating his germanium up to 400C and noting an improvement in rectification of the cooled germanium concluded “The results therefore support the view that the presence of a high resistance film is an important and possibly an essential factor in determining the behavior of contact rectifiers.” Regrettably this proved not to be the case.
within the conductor be similarly affected? And if there was a concentration of current there would be a telltale potential differences that could be measured.
Later it was shown that the size of the effect is inversely proportional to the density of current carriers in the material being investigated and its sign is determined by the net effect of positive and negative carriers. Thus for the first time an experimental tool had been discovered that enabled the investigation of conduction in solids and to express conductivity as a function of the density of carriers, their charge and their mobility.
In the 19th Century electrochemical effects were readily observable and often confused with the transport of electrical current. Carl Riecke showed in 1901 that the transmission of electricity did not entail any net migration of mass typical of an electrochemical process. He sandwiched a cylinder of copper between two aluminium conductors that powered his laboratory for a year and found no change in mass of the copper.
In 1897 Joseph Thompson discovered the electron through its charge to mass ratio in a vacuum tube. In 1916 Tolman and Steward concluded that the electron was the primary current carrier in metals due to its similar charge to mass ratio to Thompson’s electron.
This confirmed the view that Riecke had advanced that conduction was due to an “electron gas” carrying the current. Busch in his review of these developments notes that Riecke’s theory of conduction provided for conduction by both positive and negative charges of different densities and mobilities in a manner that anticipated later investigations in semiconductors utilising the Hall Effect.
A few years after Riecke, Johann Koenigsberger proposed a theory of conduction in 1908 that assumed that there was an activation energy of dissociation, ΔW, of the electron from the nucleus. This gives an increase of charge carriers with temperature off-set by a decrease in mobility. In 1914 he classified conductors with respect to the value of ΔW: Metals would tend to zero at high temperatures while insulators would have a very high dissociation energy. Semiconductors would have an intermediate value and show the observed exponential increase in conductivity with temperature.
Koenigsberger’s hypothesis was wrong but his mathematics correctly reflected the notion of an activation energy driving conduction in semiconductors. Following the work of Alan Wilson (1931) it became accepted that the energy needed was to promote electrons from their valance band to the conduction band.
Busch notes that Koenigsberger and his student Weiss use the term “Halbleiter” or “Semiconductor” in a publication based on Weiss’ doctoral thesis and a review of theories of conduction in 1911. Busch credits Weis with using the word semiconductor for the first time in the modern sense.
The Hall coefficient in metallic conductors is negative indicating that the current carriers are negative ie electrons. In semiconductors it was found that the carrier density was orders of magnitude lower than metals but mobility higher. Some had a positive Hall coefficient (copper oxide, for example) and inexplicably had mobilities similar to the electron.
Busch summarises important work by Karl Baedker in 1907 on copper iodide thin films prepared by sputtering copper onto glass or mica substrates and converted to the iodide by exposure to iodine. He found two important effects: the Hall Effect was positive and the conductivity was low but extremely sensitive to the level of iodine. Today we would conclude that these observations indicated majority positive carriers with a carrier density influenced by departures from stoichiometry. The influence of stoichiometry on semiconductor properties has been noted for other binary compounds such as zinc oxide and copper oxide.
Thus the Hall effect had been used to demonstrate majority positive or negative carriers. But despite Riecke’s conjecture about the possibility of majority and minority carriers of differing charge there was no direct evidence for this as the Hall Effect gave only the net flow. Failure to understand the importance of minority carriers in semiconductors left a serious flaw in the development of a comprehensive theory until Boris Davydov in 1938.
Max Planck began the quantum revolution in 1900 when he showed that light was emitted or absorbed in “quanta” or discrete amounts in a manner that seemed to contradict the prevailing view that light was a continuous electromagnetic wave.
In 1905 Albert Einstein explained the photoelectric effect observed in vacuum tubes by demonstrating how the energy of incident light is transferred to ejected electrons. Quanta of light became known as photons and visualized as particles thus launching the notion that dual properties, wave-like and particle-like could co-exist.
In 1924 Louis-Victor de Broglie proposed that just as light had particle-like properties ordinary matter too should have a wave-like nature. For most observable particles the wavelength would be too long to be detectable but something on the scale of the electron should have a measurable wavelength.
Diffraction patterns are a classical observation in optics easily explained by the wave-like properties of light. High energy “light” (X-rays) exhibit diffraction patterns on interacting with crystalline solids. In 1927 this effect was demonstrated for the first time in a particle with mass: Clinton Davisson and Lester Germer at Bell Laboratories and George Thomson at Aberdeen University demonstrated wave-like properties of the electron.
A year earlier Schrödinger produced a quantum mechanical solution for the hydrogen atom that correctly predicted its spectrum. As in Bohr’s model of the hydrogen atom of 1913, electrons only occupied discrete energy states.
Heisenberg published his uncertainty principle in 1927: On the atomic scale, the more precisely the position is determined, the less precisely the momentum is known in this instant, and vice versa.
The Pauli exclusion principle from 1925 requires that, for example, electrons in a single atom can only share the same state if they have opposite spin imposing a limit on the number of electrons in any one energy band.
A year later Fermi and Dirac created a statistical description of the energy states of many electrons when subject to thermal equilibrium which Sommerfeld utilized to describe metallic conduction.
Felix Bloch worked out the quantum mechanical solution for a periodic system such as in crystalline or metallic materials where the fields are subject to the regular placement of the component nuclei. His work enabled a theoretical description of many of the observed properties of crystalline materials including semiconductors (Alan Wilson in 1931).
Despite the fact that the predictive power of quantum mechanics was a success even in the 1920s there was no instant acceptance. Einstein who contributed to the work remained a skeptic with many famous quotations. For example: “All these fifty years of conscious brooding have brought me no nearer to the answer to the question, 'What are light quanta?' Nowadays every Tom, Dick and Harry thinks he knows it, but he is mistaken.” (The Born-Einstein Letters)
In his 1931 paper Alan Wilson set out the now familiar conduction band theory of semiconductors and developed an explanation for their negative temperature coefficient of resistance.
In his model electrons occupy discrete energy bands separated by bands of disallowed energy. Implying a definition of a semiconductor he proposes “suppose the number of electrons is just sufficient to fill up the lowest band when all the electrons are at their lowest states.” At absolute zero and for a low applied field there is insufficient energy to promote any electron to the next band. And Pauli’s Exclusion Principle excludes conduction by conducting electrons sharing positions in the occupied bands.
When the temperature is raised “there will be a few electrons in the second band and conduction can take place.” This is the conduction band. The band below the conduction band is usually known as the valance band due to the role of these electrons in covalent bonding.
“The increase in the number of conduction electrons with rise of temperature will tend to increase conductivity, while the excitation of the thermal vibrations of the lattice will tend to decrease it. At low temperatures the first effect must predominate, and so resistance will have a negative temperature coefficient. At higher temperatures the effect of the thermal vibrations will be the more important and resistance will obey the normal law. This is just what is observed for semiconductors.”
Wilson derived the function for conductivity at lower temperatures in the range where conductivity is largely determined by thermal excitation of electrons to the conduction band (where ΔW is the energy difference between the bands):
σ = σ0e-ΔW/2KT
This is the relationship between conductivity first observed by Hittorf in 1851 for cupric sulphide as shown by Busch in his 1989 review.
In the same paper, Wilson recalculates the Hall Effect coefficient assuming conduction is due only to electrons. Minority carriers have not been recognized, yet.
In 1931 Wolfgang Pauli famously wrote to the Physicist, Rudolf Peierls “Ueber Halbleiter sollte man nicht arbeiten, das ist eine Schweinerei, wer weiss, ob es überhaupt Halbleiter gibt.”
“One should not work on semiconductors, that’s a mess, who knows whether there are semiconductors at all.”
This comment is at high risk of being misconstrued given the obvious importance of semiconductors today. Pauli was reflecting the views of other researchers such as Gudden who published a review of the conductivity of semiconductors in 1930. He proposed that no chemically pure material would be a semiconductor. Gudden considered that most semiconductor behaviour was due to impurities. “semiconductors in the scientific sense of the word - if they exist at all - are by far scarcer than originally assumed”
Wilson used his model in 1931 to offer a qualitative explanation of the role of impurities. In his diagram the grey bands are the permitted energy levels of the pure material. The foreign atom has an energy level indicated at A-B, closer to the conduction band (2). In the case of a semiconductor the valance band (1) is full and since it requires less energy to raise A-B to the conduction band the impurity makes a disproportionate contribution to conduction. Wilson concludes “If this is the correct view then the occurrence of semiconductors is purely accidental. This is not in disagreement with the facts as there seem to be no other properties distinguishing insulators from semiconductors.”
“…the observed conductivity of semiconductors must be due to the presence of impurities.”
There is no doubt that this was literally true since there were few if any materials that were “pure.” Pearson and Brattain note that Carl Wagner in the period 1930-33 published work on the conductivity of copper oxide showing that a deficiency of copper of just a few parts per million gave a postivie sign of the Hall Coefficient (defect semiconductor). If there is an excess of metal such as is the case for zinc oxide then the Hall Coefficient is negative and this was known as an excess semiconductor.
Thus while binary compounds could exhibit deviations from ideal stoichiometry what of the semiconducting elements such as silicon, germanium, selenium? They were far from pure. For example, commercial silicon was typically 2% impurities and very difficult to produce in a pure form. While much progress was made, both silicon and germanium defeated even the intense war time efforts to make high purity materials familiar to modern processing.
The importance of minority carriers is fundamental to the theory of rectification and the transistor but were not recognized in the theoretical treatments worked out by most of the physicists in the 1930s. In their comprehensive review, Pearson and Brattain refer to this as a persistent “blind spot.”
In 1931 Dember found a photoelectric effect in cuprous oxide by observing a potential difference between illuminated and dark areas of the same wafer. Frenkel explained this by supposing that the photo effect created hole-electron pairs and suggested that the observed potential resulted from the unequal diffusion coefficients of the holes and electrons. In a theory of the photomagentoelectric effect discovered by Kikoin and Naskov, Frenkel recognized the role of both majority and minority carriers.
Later in 1938 Davydov explained rectification in copper oxide rectifiers by assuming that there was a rectifying junction composed of two forms of copper oxide: a thin layer of copper oxide adjacent to the copper of the excess type (excess copper) next to a layer of copper oxide of the defect type (missing copper). The metal junction was not rectifying at all. In Davydov’s approach minority carriers were accounted for:
Holes in the excess layer (or N-type)
Electrons in the defect layer (or P-type)
In the same year Davydov published an analysis of photoelectromotive force in semiconductors that included the role of minority carriers. But this work was largely ignored by other workers. “…these two theoretical papers attracted little attention from other investigators and this blind spot regarding the role of minority carriers continued to persist.”
In 1949 Pearson and Bardeen published a comprehensive paper on the role of donor and acceptor elements in relation to P and N-type semiconductors. Three years earlier their colleagues at Bell, Scaff, Theuerer and Schumacher had shown that elements in Group III such as boron and aluminium are acceptors and give defect or P-type conductivity. Elements in Group V such as phosphorus, antimony and arsenic are donors and give excess or N-type conductivity.
A Group V element has one more electron than required to bond with the adjacent silicon atoms and is only weakly bound. This is represented by ED in the diagram which is of the order of 0.1 ev. This means that at room temperature most of the impurity atoms are thermally ionised and their associated electrons are in the conduction band. These ideas were first set out by Alan Wilson in 1931.
Group III atoms have one less electron than would be required for covalent bonding with the neighboring silicon. This missing electron is a hole which behaves in all respects like a positively charged electron and requires little thermal excitation to contribute to conduction.
Work at Purdue on germanium directed by Lark-Horovitz showed that boron, aluminium and gallium produce P-type conductivity and nitrogen, phosphorus, arsenic, bismuth and antimony produce N-type conductivity. Lark-Horovitz included tin in his list of N type dopants and tin was literally used in high back voltage germanium rectifiers but was a case of an impure dopant. Tin was simply the carrier.
And Lark-Horovitz found “All of the samples investigated show the existence of large and perfect crystallites indicating that lattice distortions and foreign enclosures are not primarily responsible for the electrical properties of these semiconductors” confirming the work at Bell Labs [Hoddeson et al 1992]. Workers were, however, not certain on this point. For example, Shockley’s junction transistor patent includes the possibility that the dopants did work through lattice defects.
Shockley endured considerable angst after not being directly part of the team that discovered the point contact transistor late in 1947 and his formal exclusion as a named inventor on the associated patents (Shockley 1976).
“The birth of the point-contact transistor was a magnificent Christmas present for the group as a whole. I shared the rejoicing. But my emotions were somewhat conflicted. My elation with the group’s success was tempered by not being one of the inventors. I experienced some frustration that my personal efforts, started more than eight years before had not resulted in a significant inventive contribution of my own.”
These events inspired Shockley to invent a robust device without the delicate point contacts, their manufacturing difficulties and low power capacity.
From New Year’s Eve that year through the early part of 1948 Shockley sketched out his design for the Semiconductor Valve. In an entry in his lab book dated 23 January 1948 He describes this as…”at least three layers having different impurity contents. Suppose there are two layers of N separated by a thin layer of P. Such a device may be produced by evaporation. Ohmic contacts are made to all three layers. Such a structure is indicated diagrammatically on the left. Under the operating conditions (a) is the emitter, (b) is the control and (c) is the collector. Modulation by (b) is affected as follows:
Looking back from 1976 Shockley has a sanguine view of his conception of the junction transistor noting that this was no Eureka moment! He was not setting out to invent the junction transistor at all: “Instead I was trying to invent new experiments to improve the scientific aspects of the research on the P-type surface layer of the point contact transistor… In effect, after doing this planning, the junction transistor structure was staring me in the face, or at least, looking me squarely in my mind’s eye.”
There were two key issues with the Shockley invention. It was not at all clear that electrons could diffuse through the P layer without recombination with its holes and there was no way to fabricate and test it. On the former point Shockely notes: “The demonstration that minority carrier injection could occur and be really useful [proved] to be a key development in 1948. Actually, injection of minority carriers by rectifying, metal point contacts had been observed at least once and possibly twice before my disclosure [on 23 January 1948].” Shockley refers to two problems in the properties of high back voltage germanium: a material developed at Purdue University for use in diodes. The Purdue team noticed that high electric fields lowered its resistivity but did not know why. From the vantage of 1976 Shockely writes “What really increased the conductivity was the injection of minority holes, plus neutralizing electrons.” The same material was used in the first point contact transistor. At this time the theory of the point contact transistor was by no means worked out and the second issue was how minority carriers navigated the N-type base layer: it was generally assumed that minority carriers flowed from the emitter along the surface of the base layer (but not the body).
John Shive was part of the Bell team supporting the patent application and carried out an experiment in which the point contacts were positioned on the opposite side of a thin N germanium wedge (Shive 1949). It amplified and at the same time presented striking experimental evidence for injection of minority carriers by the emitter. Shive’s experiment was presented to the select transistor group on 18th February 1948. Shockley realized that this was, in effect, a proof of principle of a key issue in his design. And its linear geometry was alarmingly close to the Shockley invention which he had deliberately kept from the group. At risk of being gazumped on the invention of the junction transistor Shockley immediately presented his ideas on the junction transistor and used them to interpret Shive’s observation.
The importance of Shockley’s ideas were recognized and a patent was filed on 26th June 1948. The only exemplification was in a P-N junction with control exercised through a liquid electrode at the junction. This amplified but only at very low frequencies.
Progress from conception to realization of the junction transistor took over two years. The first “solid state” junction transistor was made on 7th April 1949 by dropping molten P-type germanium onto hot N-type and sawing the resulting junction to make two strips of P-type on an N-type plane.
Throughout this time Shockley developed his seminal paper The Theory of P-N Junctions in Semiconductors and P-N Junction Transistors which was published in 1949 as a Bell monograph.
Progress towards the Shockley design was slow. Following the materials choice for the point contact transistor, Shockley wanted a junction transistor made from polycrystalline germanium. But Gordon Teal, a Bell Labs physical chemist was keen to extend work on single crystal germanium proposing that minority carriers would have longer lifetimes in single crystals. In pursuing this approach he and Morgan SparksMorgan Sparks records the method in his lab book: “Yesterday we pulled an N-P-N junction by doping twice (once with Ga-Ge second with Sb-Ge during a pull of a crystal. The unit was cut longitudinally…by etching away the N sides we were able to solder to the P-type bridge.” [Shockley 1976]
No doubt with a great deal of satisfaction, Shockley noted in his lab book page of the 23rd January 1948 that set out his concept for a junction transistor “An NPN unit was demonstrated today to Brown, Fisk Wilson, Morton” dated 10th April 1950. But Shockely writes in 1976 “This non-photogenic device did perform according to the theory but had a wide base and poor frequency response and provoked little interest. For about nine months, the efforts to improve junction transistors were practically negligible at the laboratories.”
However, by mid 1951 a range of viable transistors had been made (Shockely 1951). These operated with power levels from milliwatts to 2 watts. Their collector junction capacitance limited their operating frequency to the audio range.
The theory of semiconductor devices had, in the hands of Shockley, advanced to a point where it was possible to make paper inventions limited only by the available technology to manufacture them.
Hoddeson L Braun E Teichmann J Weart S 1992 Out of the Crystal Maze: Chapters from the History of Solid State Physics Oxford University
Bidwell S 1885 On the Sensitiveness of Selenium to Light and the Development of a Similar Property in Sulphur Proc. Phys. Soc. London 7 129-145
Benedicks C 1915 Electrischerwiderstand einiger seltener Metalle; Thermokraft und Gleichrichterwirkung des Germaniums Int Z Metallgraphie 7 225-38
Born M Einstein A The Born-Einstein Letters 1916-1955: Friendship, Politics and Physics in Uncertain Times Macmillan
Busch G 1989 Early History of the Physics and Chemistry of Semiconductors- From Doubts to Fact in a Hundred Years Eur J Phys 10 254-64
Hall E H 1879 On a New Action of the Magnet on Electric Currents Am Jnl Mathematics 2 287-292 reproduced at http://www.stenomuseet.dk/skoletj/elmag/kilde9.html
Jenkins T 2005 A Brief History of Semiconductors Phys Educ 40 (5) 430-39
Lojek B 2007 History of Semiconductor Engineering Springer
Mendeleev D 1869 The Periodic Table Zeitschrift für Chemie 12 405-6 Translated at http://web.lemoyne.edu/~giunta/mendeleev.html
Merritt E 1925 On Contact Rectification by Metallic Germanium Proc NAS 11 No. 12, 743-48
Pearson G Bardeen J 1949 Electrical Properties of Pure Silicon and Silicon Containing Alloys Containing Boron and Phosphorus Phys Rev 75 865-83
Pearson G Brattain W 1955 History of Semiconductor Research Proc IRE 43 1794-1806
Pierson R 1927 Photoelectric Conduction in Selenium Phys Rev 30 664-72
Pierce G W 1907-09 Crystal rectifiers for electric currents and electrical oscillations Parts I-III Phys Review 25 31-60; 28 153-87; 29 478-84
Pauli W 1985 1930-39 Scientific Correspondence with Bohr, Einstein, Heisenberg and others vol II (Berlin: Springer) p94
Shive J 1949 The Double-Surface Transistor Phys Rev 75 689–90
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
Shockley W 1951 U S Patent 2,569,347 Circuit Element Utilizing Semiconductive Material (filed 1948)
Shockely W 1976 The Path to the Conception of the Junction Transistor IEEE Trans on Electron Devices ED-23 July 597-620
Riordan M Hoddeson L 1997 Crystal Fire W W Norton & Co Inc
Wilson A H 1931 The Theory of Electronic Semiconductors I and II Proc R Soc 133 No 822 458-91; 134 277-87
Winkler C 1986 Germanium, Ge, a New Nonmetallic Element Berichte der Deutschen Chemischen Gesellschaft 19 210-211.
Winkler C 1886 About Germanium Journal für praktische Chemie 142 (N. F. 34) 177-229 both translated at http://dbhs.wvusd.k12.ca.us/webdocs/Chem-History/Disc-of-Germanium.html