The Transistor

NATURE ABHORS A vacuum tube," cracked Bell Labs physicist John Pierce. So did almost everyone else by the 1940s. Sure, vacuum tubes boosted the power of the phone network's electrical signals, which weaken as they travel. But vacuum tubes were too bulky, unreliable and inefficient to support what AT&T expected to be a boom in demand for telecommunications after the end of World War II. So as peace loomed, in the summer of 1945 Bell Labs established a group to forge the future out of semiconductors, materials whose properties lie midway between an electrical insulator's and a conductor's. In the fall of 1947, in a month-long burst of inspiration, Bell scientists invented the device that came to embody, even create, the future, The tiny transistor changed the way we bank, drive, cook, communicate, listen to music, watch television and otherwise work, play and live.

The inventors were an unlikely trio. William Brattain was a farm boy and a born tinkerer. William Shockley, hard-driving, ambitious and impatient, was named manager of semiconductor research in 1945. His, ego would eventually fracture the team. "Whispering John" Bardeen the low-key, famously self-effacing theorist, would become the only person ever to win two Nobels in physics.

In 1925 a British scientist had theorized that if an electric field enveloped a semiconductor, then the semiconductor would conduct electricity differently. In some cases, it would amplify incoming current. That appealed to the Bell scientists charged with finding replacements for vacuum tubes. But the phenomenon remained maddeningly theoretical; try as they might, no one could make semiconductors jack up a signal. Finally, in March 1946, Bardeen hit on the reason. Electrical fields were not having the desired effect on, say, a bar of silicon because the surface of the silicon is riddled with cul-de-sacs, he suggested, making it possible for electrons to enter but not to leave. The surface seemed to be positively charged on the outside, attracting electrons in, but negative on the inside, repelling them when they start to move. The electrons were stuck. As a result the flow of electrons--which is all a current is--could not increase. Applying an electric field did nothing.

For the next 20 months Bell's team turned to the most basic of research--the quantum properties of solids--as it sought ways to liberate the electrons in the semiconductor. "Without understanding solids from a quantum mechanical point of view," says William F. Brinkman, vice president of R&D at Bell Labs, "the transistor could not have been invented." On Nov. 17, 1947, Brattain launched the experiments that would bring success. He began with a splash--literally: he bathed silicon in various electrolytes (liquids that contain electric charges), such as acetone and distilled water. The electrolytes changed the electrical properties of silicon's surface. When Brattain shined a light on the treated silicon, a larger current flowed than from untreated silicon (silicon was known to produce a current in response to light; today that is the basis for solar cells). Apparently, the dectrolytes set up an electric field that bulldozed the cul-de-sacs, allowing the electrons to escape.

On Nov. 21 Bardeen went to Brattain with a new suggestion for making silicon amplify a signal. "Come on, John!" Brattain exclaimed. "Let's go out in the laboratory and make it!" They put a drop of distilled water on a slab of silicon. They pushed a tungsten wire through the drop and onto the silicon. They used a battery to apply one volt to the drop, hoping to stir up the positive and negative charges in the silicon just below the wire. Current through the contact point increased 10 percent: positive charges in the distilled water pulled the silicon's electrons to the surface, making more electrons flow and thus amplifying the current. Carpooling home that evening, Brattain said he'd "taken part in the most important experiment that I'd ever do in my life," according to an AT&T oral history quoted in "Crystal Fire" (352 pages. W. W. Norton. $27.50), a new book on the transistor by Michael Riordan and Lillian Hoddeson.

But hurdles remained. The silicon boosted current only 10 percent, not enough to outdo vacuum tubes. Brattain and Bardeen tried every variation they could think of to better their results. Germanium instead of silicon. Gold foil instead of tungsten. A viscous liquid called glycol borate--"gu"--instead of distilled water. On Dec. 16 they jury-rigged their final contraption. "It was marvelous!" Brattain recalled: their transistor boosted power 450 percent. The key realization was that "holes"--weird quantum-mechanical entities that are the absence of electrons--carried current in silicon. When Bardeen returned home that evening, he mumbled to his wife, Jane, as she peeled carrots, "We discovered something important today."

But Shockley was far from elated at Bardeen and Brattain's success. He argued that work he had done in 1945 had sparked their invention, but AT&T's lawyers had filed a patent only on Brattain and Bardeen's device. Shockley worked obsessively on his own. On Jan. 23, 1948, he had his brainstorm: a sandwich. The bread would be semiconductor material with an excess of electrons; it was dubbed "n-type." The meat would be "p-type," with an excess of positively charged holes. When he attached wires and applied a voltage, holes streamed across the n-material into the p-area. His "junction transistor" amplified current just like Brattain and Bardeen's "point-contact" transistor. He tinkered with it in total secrecy. The rift in the team was now a canyon. Bardeen, fed up with Shockley, resigned in 1951.

In 1952 Bell Labs offered to license the point-contact transistor for $25,000 against future royalties. They had few takers apart from a small Japanese start-up called Sony. Its first transistor radio sold, in 1954, for $49.95 (more than $300 in 1995 dollars). Bell Labs produced the point-contact transistor for 10 years. But by 1954 production of the junction transistor had overtaken it. In 1956 Brattain, Bardeen and Shockley shared the Nobel in physics.

As the price of a transistor plunged--from $45 to $2 in the 1950s to .00001 cent today--the applications mushroomed. In 1959 sales of solid-state transistors overtook sales of vacuum tubes, and there has been no going back. Before the transistor "the whole phone network was analog and the switches were electromechanical; the transistor changed that to digital transmissions and electronic switches," says Ian Ross, Bell Labs president from 1979 to 1991. Today transistors packed by the millions onto microprocessors run car engines, cell phones, missiles, satellites, gas pumps, ATM machines, microwave ovens, computers, CD players and every other modem electronic toy and tool. In 1997 more than half a billion transistors will be manufactured. Every second.