by Jon Gertner
Semiconductors—as their name implies, neither conductors nor insulators—are a curious case. In their outer band, atoms that comprise these substances have somewhere between three and five electrons, and they seem to exhibit qualities that are different from those of either a conductor or an insulator. Early in the twentieth century, physicists noted that these materials became better electrical conductors as their temperature increased—the opposite of what happened with metals (and good conductors) like copper. In addition, they could in some circumstances produce an electric current when placed under a light—what was known as a photovoltaic effect. Perhaps most compelling, the materials could rectify, meaning they allowed electric signals to pass in one direction only (they could, in other words, convert alternating electric current, AC, to direct current, DC).13 This was a useful property familiar to almost any Bell engineer. The early crystal wireless radios that so many Labs scientists grew up with depended on semiconductor crystals like silicon. Silicon crystals would process the incoming radio signal, transforming a weak AC signal into DC, so it could be heard through a headphone.
Three Bell Labs researchers in particular—Jack Scaff, Henry Theurer, and Russell Ohl—had been working with silicon in the late 1930s, mostly because of its potential for the Labs’ work in radio transmission. Scaff and Theurer would order raw silicon powder from Europe, or (later) from American companies like DuPont, and melt it at extraordinary temperatures in quartz crucibles. When the material cooled they would be left with small ingots that they could test and examine. They soon realized that some of their ingots—they looked like coal-black chunks, with cracks from where the material had cooled too quickly—rectified current in one direction, and some samples rectified current in another direction. At one point, Russell Ohl came across a sample that seemed to do both: The top part of the sample went in one direction and the bottom in the other. That particular piece was intriguing in another respect. Ohl discovered that when he shone a bright light on it he could generate a surprisingly large electric voltage. Indeed the effect was so striking, and so unexpected, that Ohl was asked to demonstrate it in Mervin Kelly’s office one afternoon. Kelly immediately called in Walter Brattain to take a look, but none of the men had a definitive explanation.14 “In discussing these mysteries Ohl and I decided we needed to characterize them in some way,” the metallurgist Scaff later explained.15 During a phone call, the two men decided to call one type of silicon p-type (for positive conduction) and the other n-type (for negative).
It wasn’t necessarily clear at the start why this was so—or whether it was even important. By the early 1940s, however, Scaff and Ohl were increasingly sure that the two differing types of silicon were the product of almost infinitesimal amounts of different impurities. Atoms within semiconductors bond easily with a number of other elements. Scaff and his colleagues knew that when they cut n-type silicon (atomic number 14) into smaller pieces on a power saw, for instance, they could smell something they were sure was phosphorus (atomic number 15). None of the measurement equipment could pick up the taint, but their noses could.16 Later, the men also determined that p-type silicon often had faint traces of the elements aluminum (13) or boron (5).
This was the beginning of a larger insight. Ultimately the metallurgists Scaff and Ohl agreed that certain elements added to the silicon (such as phosphorus) would add excess electrons to its outer band of electrons; those extra electrons could, in turn, move around and help the silicon conduct current, just as they might in a conductor such as copper. This was n-type silicon. On the other hand, certain other elements added to the silicon (such as boron) created additional empty spaces for electrons in the outer band—these became known as holes. These so-called holes, much like electrons, could also move about and conduct current, like a stream of bubbles moving air through a liquid. This was p-type silicon. For Scaff and Theurer—and, in time, the rest of the solid-state team at Bell Labs—one way to think of these effects was that purity in a semiconductor was necessary. But so was a controlled impurity. Indeed, an almost vanishingly small impurity mixed into silicon, having a net effect of perhaps one rogue atom of boron or phosphorus inserted among five or ten million atoms of a pure semiconductor like silicon, was what could determine whether, and how well, the semiconductor could conduct a current. One way to think of it—a term that was sometimes used at the Labs—was as a functional impurity.
When the war arrived, Ohl, Scaff, and some of the other metallurgists continued to hone their techniques with silicon. So, too, did scientists at a number of other universities and industrial labs. Silicon crystals, in particular, were found to be a vital component in the airborne radar sets that the Labs was designing and that Western Electric was making. Just as in crystal wireless sets, the silicon diodes, as they were known, could perform a vital function in radar receivers for their ability to process (that is, rectify) incoming radar signals. The Labs’ metallurgists were now able to buy highly purified silicon from DuPont and reliably integrate traces of impurities by adding them to the “melt” before a substance cooled.
Indeed, this was the situation Shockley confronted upon his return to Bell Labs, as he began working with his solid-state physics group. His team could have continued working with a semiconductor such as copper oxide, which was what Shockley and Brattain had tinkered with before the war. But early on, the solid-state group decided they were interested in understanding better what seemed to them the most fundamental and representative materials in the class. “Although silicon and germanium were the simplest semi-conductors, and most of their properties could be well understood in terms of existing theory, there were still a number of matters not completely investigated,” an internal history of the solid-state research project later explained. Germanium had been discovered in Europe in the late 1800s, and as with silicon, wafers made from the element had proven useful during the war for radar detectors. The small amounts the men worked with came from tiny military stockpiles in Missouri, Oklahoma, and Kansas.
Silvery and lustrous in appearance, germanium was so rare that one Bell Labs scientist remarked that before 1940 only a handful of people in the world had actually ever seen it.17 Moreover, it was until that time considered largely useless. A Bell Labs metallurgist named Gordon Teal had wondered if therein resided its virtue. “A research man,” he later remarked, “is endlessly searching to find a use for something that has no use.”18
MEN ARE A KIND OF MATERIAL, TOO.19
“It was probably one of the greatest research teams ever pulled together on a problem,” Walter Brattain would later say. When he first reviewed the list of who would be working with silicon and germanium in the new solid-state group with Shockley at Murray Hill—roughly every month, the Labs’ staff received typed organizational charts of their department’s personnel—Brattain read it over twice. There isn’t an S.O.B. in the group, he thought to himself, pleased with the prospect of joining in. Then after a minute he had a second thought: Maybe I’m the S.O.B. in the group.20
It was entirely possible. Brattain, rarely at a loss for words, was known to be reflexively argumentative. He no longer carried a rifle, but a career in the most rarefied precincts of science had done nothing to smooth the cantankerous attitude he’d brought with him from the frontier of Washington State.21 Wiry in stature, with slender mustache and thinning gray hair combed straight back from his forehead, Brattain nevertheless had an endearing side. He was good company and—unlike Shockley—often self-deprecating; he was also collegial, garrulous, and unusually game for an experiment or a round of golf. He had deft hands as well as deft ideas on how to carry out lab work. So did another experimentalist in the group, Gerald Pearson. The two men shared a large laboratory on the fourth floor of Building 1 in Murray Hill. Brattain was intent on investigating the surfaces of semiconductor materials—what happened on those surfaces, for instance, when small slabs or cylinders of silicon or germanium were exposed to heat or connected to various kinds of circuits. Pearson was intent on
exploring the “bulk” properties of the materials. That is, he wanted to investigate their interiors. The distinctions could seem irrelevant to a layman, but in the subatomic world of solids the behavior of surfaces and interiors was not necessarily the same.
Other scientists from the Labs joined the solid-state group at varying levels of involvement—chemists, circuitry experts, metallurgists, technical assistants. Shockley naturally assumed the role of the leading theorist of the group; at the start it was his ideas that drove experimentation. But Shockley believed the group was incomplete. “Bill Shockley is one of the ablest theoretical people around anywhere and always had been,” Jim Fisk, who directed the wartime magnetron work at the Labs, later told Lillian Hoddeson. “But he recognized that we needed more and we decided jointly that there were probably only three people in the country that qualified here.”22 Of the three, Shockley and Fisk set their sights on one man in particular—John Bardeen, who was just finishing up a stint at the Naval Ordnance Laboratory in Maryland, where he’d been stationed for wartime work on mines and submarines. The field of solid-state physics was new enough, and small enough, that Bardeen already had a number of connections with Bell Labs scientists. He’d met Shockley in the mid-1930s, while he was a postgraduate student at Harvard and Shockley was a grad student at MIT; around the same time, he’d become friends with Fisk, who also held a postgraduate fellowship at Harvard. Bardeen likewise knew Walter Brattain through Brattain’s younger brother, with whom he had gone to graduate school.
Fisk and Shockley convinced Mervin Kelly to hire Bardeen, and Kelly in turn offered Bardeen a salary significantly higher than any university could pay him. Bardeen joined the Labs in October 1945.23 The large volume of the war work had created crowded conditions at Building 1 in Murray Hill; a new building—Building 2—was planned but not yet begun. When Bardeen arrived there was no office ready for him. So he did the next best thing: He set up his theoretical shop in the lab shared by Brattain and Pearson.
One could readily see that the men on the solid-state team were distinct in their talents as well as personalities. There was Shockley, a lightning-quick aggressor; Brattain, a skeptical and talkative experimentalist; and Pearson, an easygoing and steady presence. Bardeen—with thinning dark hair, a medium build, and a soft physicality that belied his athleticism—wasn’t just quiet. He barely spoke. And when he did speak it was often in something best described as a mumble-whisper. Bardeen and Shockley differed not only in temperament but in their scientific approaches. Shockley enjoyed finding a hanging thread so he could unravel a problem with a swift, magical pull and then move on to something else. Bardeen was content to yank away steadfastly, tirelessly, pulling on various corners of a problem until the whole thing ripped open. Sometimes he did loosen up in conversation—a few beers would often do the trick—but mainly he watched and listened and worked various angles. Brattain realized that when Bardeen did choose to interpret data or ask a question, a profundity was likely to tumble forth. The group recognized this not long after. When Bardeen talked, everyone else immediately stopped to listen.
Bardeen and Brattain—brain and hands, introvert and extrovert—discovered they worked well together. Indeed, they sometimes seemed hardly able to function independently. “I was the experimentalist and Bardeen was the theorist,” Brattain would later recall, “and in fact there were occasions where I had to go to another department and Bardeen was left in the lab and he was anxious to get the experiment done and I said, ‘Well, there it is, John, I’ll be back in about an hour.’ And I’d come back in about an hour and John would be gone and I’d ask the other people in the lab what happened and they’d say, ‘Oh, he worked for about five minutes and said, “Oh, damn!” and left.’”24
Their work together was further buoyed by the exchange of ideas within the larger solid-state group, which would gather sometimes once a day—and at least once a week—in meetings often led by Shockley, to exchange thoughts and review experiments. “I cannot overemphasize the rapport of this group,” Brattain said. “We would meet together to discuss important steps almost on the spur of the moment of an afternoon. We would discuss things freely. I think many of us had ideas in these discussion groups, one person’s remarks suggesting an idea to another.”25 The group would carry their discussions into lunch in the cafeteria as well. Or they would get in their cars and drive a few miles south along Diamond Hill Road, a narrow, sinuous county highway, to visit a small hamburger joint called Snuffy’s. The Bell Labs cafeteria didn’t serve beer. Snuffy’s did.26
The formal purpose of the new solid-state group was not so much to build something as to understand it. Officially, Shockley’s men were after a basic knowledge of their new materials; only in the back of their minds did a few believe they would soon find something useful for the Bell System. In many respects their work was orders of magnitude more difficult than what usually went on at the Labs. Some years later, the semiconductor historians Ernest Braun and Stuart Macdonald would note that “everything in a solid happens on such a minute scale that not even a microscope, even an electron microscope, can resolve the elementary processes. The scientist must operate at a level of abstraction of which the untrained mind is not capable in order to visualize processes which cannot be seen.” So the theorists at Bell Labs worked on blackboards, attempting to “see,” at a subatomic level, the surfaces and interiors of semiconductor crystals; the experimentalists, in turn, tested the theorists’ blackboard predictions at their lab benches with carefully calibrated instruments that recorded what happened within tiny pieces of silicon or germanium set on a breadboard—a wooden plank with holes—and wired into battery-powered circuits. Then the theorists would in turn try to interpret the data that emerged from the experimentalists’ attempts to investigate the theorists’ original ideas.
The circle was virtuous but not always illuminating. Early on, for instance, Shockley had formulated a theory on something he called the “field effect.” The team agreed it seemed like a sound idea, suggesting that an electrical current applied to the surface of a suitably prepared semiconductor slice should change (that is, increase) the conductivity of the semiconductor slice and result in an amplifier. But it didn’t. “My calculations showed that very substantial modulation of the resistance should occur,” Shockley later noted. “None was observed. On 23 June 1945, I wrote that the effects were at least 1,500 times smaller than what I predicted should have been observable.” He was vexed. And for close to a year, any attempts to make the field effect work failed. Shockley would eventually call this period “the natural blundering process of finding one’s way.”27
Whatever it was, by January 1946 the solid-state group, venturing far beyond the traditional methods of trial-and-error invention, had found neither enlightenment nor promise. As Brattain’s lab mate Gerald Pearson would later note, they were groping in the dark.
Six
HOUSE OF MAGIC
In the early fall of 1947, Bardeen and Shockley made their short drives to Murray Hill every morning—Shockley from the town of Madison, Bardeen from the town of Summit. Brattain would come to Murray Hill by carpool from nearby Morristown. For more than a year their solid-state group had been scientifically vigorous and socially cohesive—Shockley and Bardeen had even spent several weeks together in the summer of 1947 visiting various European laboratories. By late autumn, with the newly planted trees on the Murray Hill campus nearly bereft of leaves, the solid-state team began an experimental regimen on silicon and germanium slices that offered a steady progression of insights.
The failures of the previous year had been instructive. In the spring of 1946, in an effort to explain the failure of Shockley’s field effect, John Bardeen had spoken up, and something profound had tumbled out. He had offered a theory relating to what he called the “surface states” on semiconducting materials. In simplest terms, he postulated that when a charge was applied to a semiconductor, the electrons on its surface were not free to move the same way the electrons
in the interior might. Instead (in Shockley’s words) these surface electrons “were trapped in surface states, so that they were immobile.” The result was that the surface state created a frozen barrier between any outside voltage and the material’s interior.1 This insight had changed the direction of their research. If one were to make an amplifier out of solid materials—that is, if one wanted to pass a current through a piece of silicon or germanium and boost that current by a third input current, much like what transpires within a vacuum tube—there was now clear agreement: One first had to break through the wall of the surface state. Bardeen’s notion, as Brattain later recalled it, liberated the team.