1947: how Bell Labs stumbled into the transistor

Last time, we ended on a one-way street . A diode can steer a current, but nothing lets a small signal open or close it, so it can't make a signal bigger . For that you need an amplifier, and in 1947 that meant a glowing glass vacuum tube . Then, in the run-up to Christmas 1947, two physicists at Bell Labs pressed two gold contacts onto a slab of germanium, and the signal came out about a hundred times bigger . And here's the strange part. It worked, but not for the reason they were trying to make it work . So how did they stumble into it? Bell Labs was the research arm of the American phone system, and its problem was scale . Long telephone lines needed amplifiers along the way, and those were vacuum tubes . Tubes work by heating a filament, so they ate power, ran hot, and burned out . Calls were connected by switches with moving metal contacts . The research director, Mervin Kelly, dreamed of doing all that switching electronically instead . He thought the answer might lie in semiconductors . After the war, Bell set up a semiconductor group led by William Shockley . Walter Brattain, an experimenter who'd been at Bell since 1929, studied surfaces . John Bardeen, a theorist, joined in late 1945 . And wartime radar had already pushed people to purify and dope germanium and silicon, as you saw in episode two . Shockley's idea was elegant . Take a thin film of semiconductor and make it one plate of a capacitor . Put a voltage on a metal plate just above it, and that pulls extra charge into the film . If that extra charge is free to move, the film conducts better, or worse, as the voltage changes . A small voltage on the plate would control a much bigger current: an amplifier, using what's called the field effect . His calculations said it should work . It didn't . Tests on thin films of germanium and silicon all came back negative . In one version there was no observable change in current at all . The theory said a big effect. The bench said nothing. Shockley asked Bardeen to work out why . In March 1946, Bardeen had an answer: surface states . Inside a crystal, every atom has neighbours on all sides. At the surface, the pattern just stops . Bardeen proposed that this leaves places where electrons can get stuck, right at the surface . So when the plate pulled extra electrons in, most of them got caught there, unable to move . And their charge shielded the inside of the film from the plate . The field never got in . A theory like that needs a test . Bardeen's model suggested cooling the experiment way down, to freeze the surface states in place . Done at liquid-air temperatures, the field effect gave its first positive result . Then came what's now called the miracle month . On the seventeenth of November 1947, Brattain was fighting condensation on his experiment, so he put the whole thing in water . The effect got bigger . With the chemist Robert Gibney, he found that a voltage applied through the liquid could strengthen the effect or kill it . Bardeen's explanation was that ions in the water piled up at the surface, making a field strong enough to punch through the surface states . It amplified, but liquids are slow, so it only worked at very low frequencies, far too slow for a voice . In early December they switched to germanium, then replaced the liquid with a gold contact over a thin oxide layer . Brattain had accidentally washed the oxide off . And something odd happened. A positive voltage on the gold made the current to a nearby point go up, not down . It looked as if holes were flowing out of the gold, into the germanium, and over to the point . Bardeen called it the first sign of the transistor effect . To get real power gain, Bardeen worked out that the two contacts had to be about five thousandths of a centimetre apart . That's a twentieth of a millimetre . About the thickness of a thin sheet of paper . So Brattain made the gap instead of placing it . On the sixteenth of December, he wrapped a strip of gold foil around the point of a small plastic triangle . Then he slit the foil at the tip with a razor blade, leaving two gold contacts separated by a hairline gap . A spring pressed the triangle down onto a slab of germanium . You'll often hear that the spring was a bent paper clip. In the old photos it does look like one, but the accounts we checked just call it a spring . Brattain recalled an amplifier with gain of the order of a hundred, working right up into audio frequencies . Here's how it worked, in plain terms . One gold contact, called the emitter, is pushed in the easy direction, like the forward bias from last episode, so it pushes holes into the germanium . The other, the collector, is set the hard way round, so on its own it carries only a little current . But it pulls in the holes arriving from next door . Now wiggle the emitter a little, and the stream of holes reaching the collector wiggles with it . The collector sits in a high-resistance circuit, so that change in current becomes a much bigger change in voltage . Overall, the power came out about a hundred times bigger . On the twenty-third of December, they demonstrated it to Bell Labs managers . Brattain's notebook says the circuit was actually spoken over, and the voice could be heard and seen on an oscilloscope . Shockley called it a magnificent Christmas present . But it wasn't his invention, and it wasn't his field effect . It was holes, injected into germanium . Bardeen later wrote that if they had recognised that earlier, the transistor might have come sooner . The first transistor was temperamental . Brattain said it would sometimes stop working, but he could always wiggle it and make it work again . Two metal points pressing on a crystal are hard to build the same way twice . Later point-contact transistors got their gain by blasting the collector with a burst of current, a process called forming, and that was hit-or-miss . Shockley later admitted that his elation was balanced by frustration at not being one of the inventors . He also worried that the delicate point-contact would be hard to make in large numbers . On the twenty-third of January 1948, he wrote down a different design . No points at all. Just a sandwich of three layers, n-type, p-type, n-type, made from one crystal . In February, John Shive showed that holes could travel right through a slab of germanium, not just along its surface, which is exactly what the sandwich needed . Each layer boundary is a p-n junction, like last episode's . The first junction is forward biased, so it pushes electrons into the thin middle layer, called the base . Because the base is so thin, nearly all of them wander straight across, by diffusion, before they can meet a hole, and the reverse-biased junction beyond sweeps them up . Shockley wrote that a small current into the base could control a current a hundred times larger . And it could do that on less than a millionth of a watt . Nothing presses on a surface, so there's nothing to wiggle . Building it took two more years . Gordon Teal and Morgan Sparks grew it as a single crystal, dropping pellets of impurity into the melt as it grew, the trick from episode two . The first working junction transistor was tested in April 1950 . Within about a year, these grown-junction transistors beat the best point-contact ones on almost every count, and Bell announced them on the fourth of July 1951 . Sparks put it simply. It was quieter. It was just a better device . Credit got ugly . Shockley wanted a patent built on his field effect. When he told Brattain so, Brattain shot back that there was enough glory in it for everybody . Then Bell's lawyers found that Julius Lilienfeld had already patented that idea, filing back in nineteen twenty-six . So the point-contact patent went to Bardeen and Brattain alone . But Bell's management ruled that every photo of the inventors had to include Shockley . That's why he's in this famous picture, seated at Brattain's own bench while the two inventors stand behind him . In 1956, all three shared the Nobel Prize in Physics, for their research on semiconductors and their discovery of the transistor effect . That wording fits . Shockley posed the problem and invented the junction transistor . Bardeen and Brattain, theory and bench side by side, found the transistor effect . By then Bardeen had left Bell, writing that his difficulties stemmed from the invention of the transistor . In 1972, he won a second physics Nobel, for superconductivity . And Shockley's failed field effect? As the technology improved, it came back . By 1956, field-effect amplifiers were already being made . So now there are two ways to do it. Inject carriers through a junction, or reach in with a field from a plate. Either way, something small is steering something big . But how, exactly, does a small signal control a big one? That's next time.