Transistor

• 1. Introduction

  ›

  The transistor sits inside your phone, your laptop, your car, your refrigerator. It is so small that billions of them fit on a chip the size of a fingernail. By 2018, more than 13 sextillion MOSFETs had been manufactured, making them the most numerous artificial objects in human history. Yet most people cannot say what a transistor is, what it does, or how it came to exist. This documentary answers those questions. It traces the transistor from a failed experiment in a New Jersey lab to a device that the US Patent and Trademark Office calls a groundbreaking invention that transformed life and culture around the world. Along the way it asks what the inventors were actually trying to build, why they kept failing, and how a small germanium crystal changed the answer to both questions.

• 2. The Problem Before

  ›

  In 1907, the thermionic triode arrived, and for the first time engineers could amplify a radio signal across a continent. The triode was a vacuum tube, and it worked, but it glowed orange with heat and consumed power at a rate that made large installations expensive and fragile. Physicist William Eccles discovered the crystal diode oscillator in 1909, hinting that solid materials could do some of what vacuum tubes did. Julius Edgar Lilienfeld was the first to see the real prize. In 1925, he filed a patent in Canada for a field-effect transistor, a solid-state replacement for the triode that would use a controlled electric field rather than a heated filament. He filed identical patents in the United States in 1926 and 1928. Lilienfeld never published research on his devices, never demonstrated a working prototype, and the semiconductor materials his idea required would not be available for decades. Inventor Oskar Heil filed a similar European patent in 1934. Both men had correctly described a device that the world could not yet build. The gap between a valid idea and the physical materials to realize it would remain open for another generation.

• 3. The Germanium Experiment

  ›

  From the 17th of November to the 23rd of December, 1947, John Bardeen and Walter Brattain worked through a series of experiments at AT&T's Bell Labs facility in Murray Hill, New Jersey. Their goal was to make a field-effect transistor, the device Lilienfeld had described, but the surface states of the semiconductor material kept blocking the electric field from penetrating deep enough to work. In trying to understand why their FET failed, they pressed two gold point contacts into a crystal of germanium. The output power they measured was greater than the input. That was amplification. William Shockley, who led the Solid State Physics Group, saw the result and spent the following months expanding the theoretical understanding of semiconductors. The word transistor itself came not from the inventors but from John R. Pierce, who coined it as a contraction of transresistance. When Shockley proposed that Bell Labs' first patent should list him as the sole inventor, the company's lawyers refused. They had found Lilienfeld's earlier patents and judged that a field-effect claim would not survive scrutiny. What Bardeen, Brattain, and Shockley had built was a point-contact transistor, and all three shared the 1956 Nobel Prize in Physics for their researches on semiconductors and their discovery of the transistor effect.

• 4. Parallel Discoveries

  ›

  Herbert Mataré did not work at Bell Labs. In 1948, while employed at the Compagnie des Freins et Signaux Westinghouse, a Westinghouse subsidiary in Paris, he and Heinrich Welker independently produced a working point-contact transistor they called the transistron. Mataré had spent the war developing crystal rectifiers from silicon and germanium for German radar systems, and that experience gave him an unusual head start. By June 1948, he was achieving consistent results with germanium samples prepared by Welker. When news arrived that Bell Labs had beaten them to the invention, the Paris team accelerated production so the transistron could be used in France's telephone network. Mataré filed his first transistor patent on the 13th of August, 1948. William Shockley, for his part, had already moved past the point-contact design. He applied for a patent on the bipolar junction transistor on the 26th of June, 1948. Bell Labs chemists Gordon Teal and Morgan Sparks produced a working bipolar NPN junction amplifying germanium transistor on the 12th of April, 1950, and Bell announced it to the world in a press release on the 4th of July, 1951.

• 5. Silicon and Speed

  ›

  Germanium dominated the early transistor era, but it ran hot, leaked current, and could not handle high voltages. Morris Tanenbaum at Bell Labs produced the first working silicon transistor on the 26th of January, 1954. Texas Instruments announced the first commercial silicon transistor in May of that same year, crediting Gordon Teal, who had previously worked at Bell Labs and brought his expertise in growing high-purity crystals with him to Texas. Speed was a separate problem. Philco answered it in 1953 with the surface-barrier germanium transistor, capable of operating at frequencies up to 60 MHz. Philco manufactured the device by etching depressions into an n-type germanium base from both sides using jets of indium(III) sulfate until the base was a few ten-thousandths of an inch thick. Indium electroplated into those depressions formed the collector and emitter. AT&T adopted transistors in telecommunications that same year, deploying them in the No. 4A Toll Crossbar Switching System to select trunk circuits from routing information encoded on translator cards. The transistor was no longer a laboratory curiosity. It was moving into infrastructure.

• 6. Into the Living Room

  ›

  INTERMETALL, a company founded by Herbert Mataré in 1952, showed the first prototype pocket transistor radio at the Internationale Funkausstellung Dusseldorf from the 29th of August to the 6th of September, 1953. The first production model came a year later: the Regency TR-1, released in October 1954, built as a joint venture between the Regency Division of Industrial Development Engineering Associates and Texas Instruments of Dallas, Texas, and manufactured in Indianapolis, Indiana. It carried four transistors and one germanium diode. Its industrial design came from the Chicago firm of Painter, Teague and Petertil. The TR-1 launched in six colors: black, ivory, mandarin red, cloud grey, mahogany, and olive green. Chrysler and Philco moved the technology into automobiles, announcing the first production all-transistor car radio in the 28th of April, 1955, edition of The Wall Street Journal. Chrysler made the Mopar model 914HR available as an option for its new 1956 cars, which reached dealership showrooms on the 21st of October, 1955. It was Sony that achieved genuine mass adoption. The TR-63, released in 1957, was the first mass-produced transistor radio. Seven million TR-63s were sold worldwide by the mid-1960s, and Sony's success drove vacuum tubes out of consumer electronics for good.

• 7. The MOSFET Era

  ›

  In 1955, Carl Frosch and Lincoln Derick at Bell Labs accidentally grew a layer of silicon dioxide over a silicon wafer and noticed that it passivated the surface, protecting it and preventing dopants from diffusing inward. By 1957 they had used that discovery to manufacture the first planar transistors, with drain and source adjacent at the same surface. Mohamed Atalla and Dawon Kahng built on this directly. In 1959 they proposed a silicon MOS transistor, and in 1960 their Bell Labs team successfully demonstrated a working device. The MOSFET could be scaled down far more aggressively than bipolar junction transistors, consumed far less power, and packed far more densely. A single integrated circuit could hold more than 10,000 of them. Chih-Tang Sah and Frank Wanlass at Fairchild Semiconductor invented CMOS, the complementary MOS architecture, in 1963. Dawon Kahng and Simon Sze reported the first floating-gate MOSFET in 1967, laying the basis for non-volatile storage. By the 1970s, MOSFETs had captured nearly all market share for digital integrated circuits. Today the MOSFET accounts for 99.9% of all transistors in the world, and an advanced microprocessor can contain as many as 92 billion of them on a single die, or 2.6 trillion transistors in the most exceptional chips as of 2020.

• 8. The Transistor's Reach

  ›

  Digh Hisamoto and his team at Hitachi Central Research Laboratory demonstrated the FinFET in 1989, a three-dimensional multi-gate transistor that allowed chipmakers to keep scaling as flat planar designs reached their limits. The 1947 invention at Bell Labs was named an IEEE Milestone in 2009. The MOSFET invention of 1959 received the same recognition. Some audiophiles still prefer the harmonic distortion profile of vacuum tubes, and certain high-power transmitters and satellite amplifiers still require the electron mobility that only a vacuum can provide. High-electron-mobility transistors, or HEMTs, built on heterostructures of aluminium gallium arsenide and gallium arsenide, are used in satellite receivers operating around 12 GHz, where both speed and low noise matter. Researchers continue to explore carbon nanotube transistors, organic transistors for flexible displays, and nanoscale vacuum-channel transistors. A prototype vacuum-channel transistor built by NASA and the National Nanofab Center in South Korea in 2012 measured only 150 nanometers across. Bardeen, Brattain, and Shockley set out to replace a glowing vacuum tube. What they left behind is a device class that the world has now produced more than 13 sextillion times, with no ceiling yet in sight.