The Switch

On the afternoon of December 23, 1947, in a cluttered laboratory in Room 1E455 of Building 1 at Bell Telephone Laboratories in Murray Hill, New Jersey, a small group of men gathered around a wooden bench and listened as a human voice came out of the loudspeaker louder than it had gone in. The voice was being amplified by a device the size of a thumb. There were no glowing filaments, no warm-up time, no glass envelope thick enough to need replacing every few months. The amplifier was a sliver of germanium pressed up against two gold contacts held in place by a plastic wedge and a paper clip. It looked like something a graduate student had built over a weekend.

Walter Brattain, the experimental physicist who had assembled it, was a tall, plainspoken man in his mid-forties who had grown up on a cattle ranch in eastern Washington and never quite shed the unhurried cadence of someone used to long winter chores. His Bell Labs notebook for that week ran to dozens of pages of careful schematics and pencil-drawn waveforms. The page he started the morning of December 24, after the demonstration, summed up the previous day in a sentence so understated it has been quoted ever since: “This circuit was actually spoken over and by switching the device in and out a distinct gain in speech level could be heard and seen on the scope presentation with no noticeable change in quality.” Beneath the entry, Brattain listed the men who had been there to witness it: Robert Gibney, Hilbert Moore, John Bardeen, Gerald Pearson, and William Shockley. Bell Labs lawyers and senior research staff signed in alongside them. A microphone fed signal into one end of the contraption; a loudspeaker hung off the other. When someone spoke, the signal came out the far side with a power gain measured at about eighteen.

The men in the room understood, almost certainly, what they were looking at. They had spent the better part of two years trying to build it. They had spent the better part of December failing in increasingly clever ways. What they were not in a position to know was that the half-inch lump on the bench would, over the course of the next eight decades, scale down to features measured in atoms and multiply itself by the trillion, that the entire architecture of the modern world would be erected on top of it, and that the three men whose names would be attached to its discovery would barely speak to one another by the time their Nobel medals were struck.

The story of the transistor begins with a problem the Bell System could not solve with vacuum tubes. The American long-distance telephone network was the largest and most complex machine on earth. Every cross-country call passed through repeater amplifiers that boosted the voice signal as it weakened along thousands of miles of copper. Each of those amplifiers, in 1945, contained vacuum tubes. Tubes burned out. Tubes ran hot. Tubes were the size of a small light bulb and drew enough power to run one. Mervin Kelly, who became Bell Labs’ director of research in 1936 and would eventually run the place, had spent the prewar years watching his switching engineers add tube after tube to the network and concluding that the entire architecture was eventually going to collapse under its own thermal load. By the late 1930s he had begun recruiting young solid-state physicists with a quiet conviction that the future of telephony belonged to crystals, not gas-filled glass.

He was not alone in suspecting it. As early as 1925, a Polish-American inventor named Julius Edgar Lilienfeld had filed Canadian and U.S. patents describing an “amplifier” in which a thin layer of semiconductor would have its conductivity modulated by a third electrode held nearby. Nothing Lilienfeld actually built ever worked, and his patents languished. But the concept was correct. A semiconductor was, in principle, a substance whose conductivity could be steered by relatively small electrical signals, which is precisely what an amplifier needed to do. The trouble was that nobody in the 1920s or 1930s could produce semiconductor material pure enough or thin enough to behave the way the theory said it ought to.

By the end of the Second World War that had begun to change. Wartime radar work had pushed engineers to develop high-purity germanium and silicon crystals as detectors for microwave receivers, the kind installed in night-fighter aircraft and shipboard search radars. The technology of crystal-pulling and zone refining was suddenly years ahead of where it had been in 1939. The materials, in other words, were nearly ready. What was missing was the right team and the right idea.

Kelly assembled the team in 1945. He put William Shockley, who had earned an MIT physics PhD in 1936 and spent the war years running the Navy’s antisubmarine warfare research group at Columbia, in charge of a new solid-state physics subgroup, sharing leadership with the chemist Stanley Morgan. Into that group Kelly recruited some of the most talented minds in postwar American physics. There was Gerald Pearson, an experimentalist who had been at Bell since the 1920s. There was Robert Gibney, a physical chemist who knew electrolytes better than anyone in the building. There was John Bardeen, who arrived from a wartime stint at the Naval Ordnance Laboratory and who, of everyone in the room, may have understood the quantum mechanics of solids most deeply. And there was Walter Brattain, who had been hired in 1929 and had spent fifteen years studying surface physics on copper-oxide and other semiconductors, with a patience that occasionally bordered on the geological.

The three men whose names would later be carved together on the Nobel diploma were studies in temperamental contrast. Bardeen, born in Madison, Wisconsin in 1908 to the dean of the University of Wisconsin medical school, was the son of academia and read like it. He was so soft-spoken that his colleagues at Bell had nicknamed him “Whispering John.” Outside the office he was unprepossessing to the point of invisibility; the neighbors in Urbana, Illinois, where he later lived for forty years, knew him mostly as a cordial cookout host who happened to teach at the university. Brattain, six years older and from a homestead near Tonasket, Washington, was loud, blunt, easily exasperated, and almost compulsively hands-on. He kept a notebook the way an old farmer kept a weather log. His wife Karen, by all accounts, had to absorb a fair amount of his volatility, and he liked to say later in life that running a research group was, in the end, much like herding cattle.

Shockley was the third type, and the unstable element in the compound. Born in London in 1910 to American parents and raised in Palo Alto, California, he had an extraordinarily quick mind and a nearly limitless confidence in it. His doctoral thesis under John C. Slater at MIT, “Electronic Bands in Sodium Chloride,” was a dense piece of band-structure physics that helped establish him, in his own estimation and in others’, as one of the brighter theoretical minds of his generation. The war turned him into something close to an operations-research czar; at Columbia he learned to design depth-charge patterns and to run rooms full of senior officers. By the time he returned to Bell in 1945, he was thirty-five, ambitious, and increasingly persuaded that he was the central figure in any project he attached himself to. His colleagues respected his intellect without fail. They liked him with reservations, then with more reservations, then not very much at all.

Shockley’s first big idea after the war was wrong. In April 1945 he sketched out what he called a “field-effect amplifier”: a semiconductor slab whose conductivity, he reasoned, ought to vary if you held a charged metal plate nearby and let its electric field reach into the material. This was Lilienfeld’s idea, refined with the better materials of 1945. Shockley was sure it would work. It did not. He and his technicians built version after version through 1945 and 1946, biased and rebiased the plate, swapped germanium for silicon, and got nothing close to amplification. The conductivity would not budge.

Bardeen, with the dispassion that the rest of the group was beginning to identify as his hallmark, eventually offered an explanation in March 1946 that became one of the foundational insights of modern semiconductor physics. The electric field was not penetrating into the bulk of the semiconductor, he proposed, because electrons were getting trapped at the surface itself, in what he called surface states, forming a screening layer that effectively shorted the field out before it could reach the interior. To make a field-effect device work, you would have to find a way through that surface charge.

That diagnosis, more than any individual experiment, was what eventually unlocked the transistor. For most of 1947 Bardeen and Brattain hunted the surface states. Bardeen reasoned about them; Brattain hammered on them in the lab. They tried etching, evaporating thin films, dunking samples in various solutions. Through the summer they got nowhere in particular.

Then came what Bell Labs’ older hands later called the miracle month. In mid-November 1947 Brattain, exasperated with condensation that kept fogging up an experimental sample, dropped the entire apparatus into a thermos of water on a whim. The water was supposed to be a temperature bath. What it turned out to be was an electrolyte that neutralized the surface charge. Suddenly the rig amplified, weakly but unmistakably. The team had stumbled, half by accident and half because Bardeen was poised to interpret it, into a way around the surface-state barrier. On November 17, Gibney, the resident chemistry expert, suggested putting a small drop of electrolyte directly onto the germanium between two electrodes. They spent the next three weeks chasing that thread.

Through the first weeks of December the work went around the clock. The electrolyte version of the device worked at low frequencies but failed above a few hundred cycles per second; the chemistry was too slow to keep up with the signal. On December 8 Bardeen suggested swapping silicon out for germanium and using a thin layer of grown oxide as a substitute for the slow electrolyte. By December 12 they had a sample of germanium with a germanium-oxide skin and gold contacts evaporated on top. Brattain, in the process of cleaning the surface, accidentally washed off the oxide with distilled water. Without quite realizing it, he had also washed off the assumption underneath the experiment. He pressed the gold contact down anyway. The thing amplified across all the frequencies they could throw at it.

On December 15 Brattain wrote in his notebook that with the points “very close together” the device showed voltage gain of about two but no power gain. His distance between the two gold contacts at that point was on the order of a hundredth of an inch. Bardeen, working in his usual half-whisper at the next desk, told him to push them closer. Brattain went home that night and came back with a homemade jig: a small triangular block of plastic with a strip of gold foil glued around its tip. He took a razor blade, in front of the laboratory’s good light, and made a single careful slit through the foil right at the apex of the triangle. He had now manufactured a pair of gold contacts separated by a distance equivalent to the thickness of the cut. Mounted on a spring that pressed the slit firmly against a piece of high-purity n-type germanium, with a separate base electrode underneath, the contraption looked, depending on the angle, either crude or wholly improvised. It was both.

On the afternoon of December 16, 1947, with that jig pressed against germanium, Brattain measured a power gain of about 1.3 and a voltage gain of about fifteen on a fifteen-volt bias. He repeated the measurement, swapped polarities, repeated it again. The thing was real. Bardeen, watching, was satisfied that the surface-state trap was being bypassed: one contact was injecting positive charge carriers, called holes, into the surface region; the other was collecting them, with the field between the two pulling them along the way the gradient of a hill pulls water. A small change in current at the input contact produced a much larger change in current at the output. The crystal was acting as a switch and an amplifier. They had a transistor effect.

The principle was simple enough that schoolchildren now learn it on diagrams. A semiconductor is a material in which electrons can be added to or subtracted from the available pool of charge carriers by a small input. If the input controls a much larger output, you have a switch. If you can make the relationship continuous, you have an amplifier. Vacuum tubes did this with a glowing wire and a cloud of free electrons in evacuated glass; the new device did it with carriers moving through a solid, controlled by an electrode that drew almost no current of its own. There was no filament to burn out. There was no cathode to wear down. The device could in principle be made arbitrarily small, packed arbitrarily densely, run arbitrarily cool. It would be years before anyone outside Bell Labs grasped that this was what had happened in Building 1.

On the afternoon of December 23, Brattain and Hilbert Moore set up the demonstration in earnest. The attendees from the previous week’s tests, plus a small group of senior managers, including the research vice president Ralph Bown, gathered around the bench. A microphone fed into one end of a circuit; a small loudspeaker, salvaged from somewhere in the building, sat on the other. The men took turns speaking into the microphone with the device switched in, then switched out, then switched back in. According to Bell Labs accounts later collected by the Computer History Museum, the volume difference was unambiguous: a whisper went in, and the loudspeaker boomed it back. Shockley, who had been kept partially apprised of Bardeen and Brattain’s progress but had not been in the room for the breakthrough on the sixteenth, watched and listened. He called the device, in a phrase that has stuck to the demonstration ever since, “a magnificent Christmas present.”

It was an oddly mixed compliment. Shockley had spent two years convinced that the right kind of solid-state amplifier would be a field-effect device of his own design. In the lab the day before Christmas Eve, he was looking at a working amplifier that did not use his idea at all and that bore the names of two of his subordinates on the relevant lab pages. The demonstration ended; people went home for the holidays; Shockley did not. He flew alone to Chicago for the annual Physical Society meeting and checked into a hotel room. By his own account and by accounts of those who later examined his notebooks, he stayed in that room for most of the next four weeks, scribbling. The notebook entries he produced in early January 1948 ran to roughly thirty pages of dense pencil work. By January 23 he had sketched out a fundamentally different transistor architecture: a sandwich of three regions of doped semiconductor material, each with controlled excesses or deficits of electrons, in which a small voltage on the central layer could control current flow through the whole stack. He called it the junction transistor. It was, on paper, vastly more elegant and vastly more manufacturable than anything Bardeen and Brattain had built. He returned to Bell Labs convinced that the real transistor was his.

What followed was a slow and unhappy collapse of a working partnership. Shockley, returning to Murray Hill, told the patent attorneys that the invention was rooted in his own field-effect ideas of 1945 and that the patent applications should bear his name alone. The attorneys investigated, found Lilienfeld’s 1925 and 1928 patents covering essentially the same field-effect concept, and concluded that Shockley’s claim could not stand. The patent on the point-contact transistor was filed in the names of Bardeen and Brattain. A separate patent for the junction transistor would, in time, be filed in Shockley’s name alone. He had achieved the credit he wanted, but only on a different invention.

Even that did not settle him. Brattain later remembered that when Shockley first announced his intention to take sole credit, he protested directly. “There’s more than enough glory in this for everybody,” he told Shockley, in one of the few quotes from the period that historians have been able to verify across multiple sources. Bardeen, in his usual register, said almost nothing and went home and brooded. Around the lab, Shockley began to behave like a man who suspected his subordinates of trying to take what was his. He reorganized the project so that he, and not Bardeen and Brattain, would be the lead on the junction-transistor work. He gave orders that publicity photographs of the trio be staged with him in the foreground. Bardeen would later tell colleagues that he could not contribute to the experimental program without working in direct competition with his own supervisor, and that the isolation he experienced was, in his judgment, a deliberate management choice.

Bell Labs sat on the announcement for six months while patents were filed and military reviewers cleared the publication. On June 30, 1948, the company held a press conference in New York City. A row of small metallic cylinders, each containing two fine wires touching a pinhead-sized chip of germanium, sat on the table. The Bell Labs executive on stage explained that the new device, which someone in the building had christened the “transistor”—a name coined by the engineer John Pierce by combining “transconductance” and “varistor”—could do nearly everything a vacuum tube could do, and a few things tubes could not. He waved a small radio receiver into which transistors had been substituted for tubes; it played without warm-up. He passed around a transistor oscillator. Reporters listened politely. The New York Times ran the story two days later on page forty-six, deep in the radio column, as a brief paragraph. The popular press, by and large, did not understand what it had been shown.

Inside Bell Labs, however, the response was rapid and serious. The point-contact device was hand-fabricated, fragile, and difficult to reproduce in volume; even within the lab, only certain technicians could get a working sample on a given day. Throughout 1948 and 1949 Bell pursued a parallel track: refining the manufacturing of the point-contact transistor for limited applications, and racing to bring Shockley’s junction design to working hardware. On April 12, 1950 a Bell Labs metallurgist named Morgan Sparks succeeded in producing the first functioning grown-junction transistor; further refinement followed through 1951. By 1952 Bell was licensing the transistor patents broadly, for a flat fee of $25,000, to any company that wanted to try its hand at making them. Among the early licensees were a small Texas firm called Texas Instruments and a newly incorporated Tokyo company that would later rename itself Sony.

The first commercial product to use a transistor was a hearing aid, sold in late 1952 by Sonotone for $229.50, with one transistor and two vacuum tubes. By 1954, ninety-seven percent of all hearing aids in the United States were transistor-based. The earliest transistor radios, including Texas Instruments’ Regency TR-1, reached store shelves in 1954. The military, meanwhile, was already buying transistors as fast as Bell could make them, partly for hearing aids’ sake and partly because solid-state amplifiers could survive the vibration and temperature swings of guided weapons in ways tubes could not.

Inside the lab where it had all started, the inventors went their separate ways. Bardeen, finding conditions under Shockley untenable, accepted a professorship at the University of Illinois at Urbana-Champaign in 1951. He turned, in the next phase of his career, to a problem he had been thinking about quietly for years: the theory of superconductivity. In 1957, with two younger collaborators, he produced what came to be called the BCS theory, and in 1972 he was awarded a second Nobel Prize for it, the only person ever to win the physics prize twice. Brattain stayed at Bell into the late 1960s, did fundamental surface-physics work, and eventually returned to Whitman College, the small liberal-arts school in Walla Walla where he had been an undergraduate, and taught physics there until he retired. Shockley left Bell Labs in 1955, moved back to Palo Alto to be near his ailing mother, and announced that he was going to start a company to commercialize the junction transistor on his own.

In December 1956, the three men met in Stockholm to receive the Nobel Prize in Physics, awarded jointly for “their researches on semiconductors and their discovery of the transistor effect.” It was the first time in fifty-three years, since Becquerel and the Curies, that three physicists had shared the award. King Gustaf VI Adolf handed each of them a gold medal, a diploma, and an envelope containing roughly a third of the year’s prize money. The three posed for the obligatory photograph and, by most accounts, shared a single cordial dinner that week before scattering again. They almost never appeared in a room together for the rest of their lives.

Behind them, in the fall of 1956, the company Shockley was building in a small storefront at 391 South San Antonio Road in Mountain View, California, was hiring its first technical staff. He had recruited, with the help of personal phone calls and cross-country trips, a group of young physicists and chemists from the East Coast and the Midwest. Among them were a Caltech-trained chemist named Gordon Moore, an Iowa-born physicist named Robert Noyce, an electrical engineer named Jean Hoerni, and several others who would later be remembered by an entire industry by the name they had not yet earned. They had come west on the strength of Shockley’s name and the promise of working at the frontier of solid-state electronics. Within a year, most of them would resign en masse, walk a few miles up the road, and start a new company that would do for the integrated circuit what Bardeen and Brattain had done for the single transistor: turn an idea into a thing that could be manufactured by the millions.

The germanium sliver on the bench at Murray Hill had begun to multiply.