Why did Willis Adcock succeed when others failed? Everybody in the semiconductor business in the 1950s knew about silicon and most people knew it would possibly overcome the limitations of germanium transistors. Everybody knew that silicon had great potential. But it was just too difficult to manufacture a silicon transistor. It was an impossible process. Lots of people had tried. Along came Willis Adcock, who didn't know any better, so he grew silicon.

A year after Geophysical Service Inc., a seismic-survey company, changed its name to Texas Instruments in 1951, it became one of 20 companies to pay Western Electric $25,000 for a license to produce transistors. Then, in November 1952, it hired Gordon Teal from Bell Labs, where the transistor originated, to head its Central Research Lab.

Teal needed a staff. He visited his old school, Brown University, and asked if the folks there knew anybody who might be interested in electronics. And they suggested he talk to Adcock, who was a chemist, not an electronics engineer. And that proved to be critically important.

Before taking his doctorate in physical chemistry in 1948, Adcock was drafted into the Army which, recognizing his prior academic achievements, assigned him to the atom-bomb project at Oak Ridge. A few months later, Army officials were shocked to learn that Adcock was at Oak Ridge. "You can't be here," they said, "you're not a citizen." They had uncovered an unfortunate detail. He was born in Canada (but later became a U.S. citizen). The Army folks didn't know what to do with Adcock, so they let him continue at Oak Ridge until his term of duty was over.

After he took his doctorate at Brown, he went to work in a research lab of Standard Oil of Indiana (in Tulsa, Okla.). Brown-alumnus Teal showed up one day in 1953 and offered him a job working on transistors at Texas Instruments. Adcock's boss at Standard Oil tried to dissuade him on the grounds that it was risky. "You're out of your mind," he said. "That company is smaller than our research lab. And there's no future in that stuff. Just remember that people will always buy gas." (He was right about the gas part.)

TI, already a major manufacturer of transistors, was also a major manufacturer of radar equipment for the military. It was interested in silicon because germanium transistors failed at the high temperatures that military equipment might have to endure, and it was likely that silicon could survive those temperatures.

Ignorance leads to success

So Teal ordered Adcock to grow silicon and make silicon transistors. Adcock succeeded, he says, because he didn't know how difficult it would be. If it weren't for his heavy background as a chemist with an interest in electronics, rather than as an electronic engineer, he says, he might never have succeeded.

And there was another factor: the availability of pure silicon from duPont.

The problems, Adcock says, depended on doing everything right. He had to find the right dopants, the right doping ratios, the right diffusion and the right temperature. Especially the right temperature. The right high temperature.

Germanium, Adcock points out, melts at about 800°C while silicon melts at about 1,400°C, but a very carefully controlled 1,400°C. At that temperature, everything tends to burn. So he needed a gas that would keep silicon from oxidizing. Fortunately, helium is inert and, in Texas, it's cheap and plentiful.

Beyond that, you have problems if the temperature isn't rigidly controlled. If it gets too hot, the silicon seed crystal melts. And if it gets too cool, the seed freezes and has to be discarded. And at 1,400°C, it's not easy to control the temperature; the furnace wants to heat the environment, so the temperature wants to drop. At high temperatures, heat transfer is very rapid.

Adcock designed circuitry and equipment for very tight temperature control and for tight control of all other factors that make for a good semiconductor, then later make for a good transistor.

Of course, you have to control the temperature during germanium processing, just as you have to with silicon, he points out. But controlling 1,400°C is much more demanding than controlling 800°C because of the faster heat transfer.

After Texas Instruments offered the first silicon transistor in 1954, most competitors responded that you don't really need silicon; it was too expensive (at about five times the price of a comparable germanium transistor); and it was only worthwhile at very high temperatures, where very few circuits had to operate. At the same time, they all worked furiously to develop silicon transistors.

Today, except for some recent efforts in silicon-germanium, which calls for essentially a silicon process, the germanium transistor is all but dead.

The Century of the Engineer: Misunderstood Milestones