True Genius: The Life and Science of John Bardeen: The Only Winner of Two Nobel Prizes in Physics (2002)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

8 The Transistor O n Tuesday, October 9, 1945, the Bardeens closed the door of their Fairfax Village apartment for the last time. They had shipped ahead most of their belongings. Still their Ford was packed full. It was 11:00 A.M when they finally drove off. They were elated to be done with the war and their transient lifestyle. They were driving to Summit, New Jersey, a prosperous resi- dential town known during the Revolutionary War as Turkey Hill. Summit offered excellent schools, parks, and convenient transpor- tation; it was a stop on the Hudson Tubes, the subway to New York City. From Summit one could easily catch a bus to Murray Hill, where Bell Laboratories had built its new research campus. The Bardeens’ house in Summit was their first major financial investment. A comfortable Dutch colonial at 5 Primrose Place, it featured a sun porch, cellar, and detached garage. The children would be able to romp and play in the yard, which had swing sets and “monkey bars.” Betsy was now an active toddler, Billy a mis- chievous four-year-old, and Jimmy a “solemn” seven-year-old. Jane planned beautiful landscaping for the house. Later she would re- port with excitement when the azaleas bloomed and bemoan their fate when they were damaged in a storm. Bad luck dogged the family throughout their move. “The night- mare really started Monday,” Jane wrote to her mother. The moving company had assured Jane “as late as 11 A.M. Monday that they 115 

116 TRUE GENIUS couldn’t come until Tuesday.” So the family continued to pack carefully and methodically. But an hour and a half later the movers phoned, “when we were in the midst of lunch, with much packing remaining to be done.” They said that “they would be right out.” Jane worked herself to a frazzle to meet the shortened schedule, and “after a wild three hours it was all done and we were left with an empty house.” The revised plan also included meeting the moving van at the new house in Summit at 8:30 A.M. on Wednesday. The family members spent Monday night with various friends. After Betsy fell asleep, Jane “crept back to the house to finish pack- ing the remainder.” The next day the nightmare continued. As they drove north- east on Route 1, the Ford developed a flat tire. They “found 35 miles an hour very slow going.” Tired and frustrated after their exhausting drive, they stopped for the night at Princeton Junction, about 50 miles southwest of Summit. On Wednesday morning the family “got up at 6:00 A.M. and drove to Summit before breakfast,” arriving on time. But the moving van did not arrive until midday. Worse than that, “we found the house very dirty.” The previous owner had “not kept his part of the agreement to put the house in good condition before turning it over to us.” His belongings, even some furniture, were “still scattered everywhere.” The sale had not yet closed, and the Bardeens were not sure they could legally take possession of the house. The seller, who Jane described as “a stinker and a complete heel,” was embroiled in a bitter divorce. His wife had left to stay with family in Michigan after “he hit her and broke her jaw last week.” They consulted a lawyer, who advised the family to move in, even though they did not yet have clear title. Bringing the estranged husband and wife together to sign the necessary papers would take time. Removing their belongings and cleaning up their mess took elbow grease. The first evening Jane and John “scoured the bathroom for several hours.” Their scrub- bing was made more difficult by a lack of hot water; the ancient furnace was on its last legs. “Baths are infrequent and water must be heated on the electric stove for dishwashing and cleaning.” Bill Shockley dropped by to help, but mainly distracted them with his storytelling. At first it was difficult to eat meals at home, for the family had no refrigerator. Nor did they have all their furniture. The pieces 

The Transistor 117 Jane had stored in Minnesota were still waiting to be shipped. To top it off, Jimmy, Billy, and John caught terrible colds and needed nursing, while Betsy clung to her mother all the time. Jane wrote stoically that the world would probably “look brighter next time I write.” Ten days later she could report that she had found a woman to help with the cleaning. Although the house still needed consider- able repair, Jane was optimistic that once they had finished “shoveling out the dirt” it would be “attractive and comfortable to live in.” John helped with the housework, but he was not adept. Jane grew exasperated when he repeatedly “let the fire go out in the furnace.” Soon, however, life in Summit would become easy and rich for the Bardeens. A few minutes’ walk brought the children to Brayton Primary School, where Jim entered first grade and Bill attended preschool. On warm summer days the children could swim and enjoy frozen treats at the local pool. The family enjoyed walking along the trails of the Watchung Mountain Reservation, half a block from Primrose Place. The house proved very comfortable. It always seemed filled with young voices and laughter. The children were the center of attention. John was a “comfortable presence” for them. Not inclined to fuss over his children, he made it a point to be at their special events. At Betsy’s fourth birthday party, he “tried to get some color films of the children” playing with their bubble makers out in the sun. John’s half-sister Ann was a regular visitor in Summit. She mar- veled at her brother’s power to concentrate on his journals or the history books he enjoyed while the children played noisily around him, often climbing in and out of his lap. Betsy later said that she and her friends never felt inhibited by her father’s presence or by the fact that he was reading. They did not hesitate to interrupt him. He sat through it all, apparently unperturbed. Ann was completing her medical internship in Jersey City, about 30 miles from Summit. Charles Bardeen, Ann’s (and John’s) father, used to boast of the number of women who graduated from Wisconsin’s four-year medical college, pointing out that out of 25 students in the first graduating class, six were women. But he dis- couraged the women in his own family from studying medicine. “He felt that medicine ‘hardened’ women,” said Ann. “He thought 

118 TRUE GENIUS women would have an expensive medical education and then get married and have a family and not really use it,” John later ex- plained. When Helen expressed interest in studying medicine, Charles diverted her into nursing. But Charles passed away before Ann entered college and she felt free to study medicine. In 1945 she graduated from the medical school Charles had created; she then served a residency there in anesthesiology. Afterwards she accepted the internship in Jersey City. Eventually she would follow her father’s example and teach at the University of Wisconsin. In New Jersey, John had many opportunities to keep up with his sports. He often played golf with his old friend Walter Brattain, and he found time for squash. Joining a Bell Labs bowling league, he bowled often enough to make it worthwhile for him to invest some Christmas gift money on a custom bowling ball. As in Minnesota, bridge was one of the ways in which John and Jane entered Summit’s social community. In their early days there, they belonged to a bridge club that included Brattain and his wife Keren. Jane found Walter’s competitive attitude so stressful, how- ever, that she used the fact that she and John caught colds one January as an excuse to quit the club. Jane enjoyed Keren, with whom she had much in common. Like Jane, Keren had a graduate science training. She had completed her doctorate in physical chemistry at the University of Minnesota, where Walter took his Ph.D. in physics. Bardeen later described the Brattains as “a good match,” in which “her calm demeanor” served as an antidote to Walter’s “somewhat volatile temperament.” The Brattains were attracted to a simple and wholesome way of life. Keren dressed simply, and “her cooking would have been toward corn muffins,” Louise Herring recalled. In their home, the decorations included “a scale and a scythe and a few things like that from western living.” About a year after Bardeen came to Bell Labs, Conyers Herring arrived there with his bride, Louise Preusch, a Barnard College graduate with a mathematics major and physics minor. Herring was assigned to the Physical Electronics Division of the Physical Research Department. Bardeen and Herring continued the warm professional relationship they had established at Princeton and Harvard. Outside work, Bardeen spent less time with Herring than he did with Brattain because Herring’s sport was tennis. In their first years at Bell, the Bardeens occasionally socialized 

The Transistor 119 with Bill Shockley and his wife Jean, a warm and gracious woman who was well loved in the community. Fred Seitz vividly recalled an incident in 1938 that illustrated how Jean tried “in a gentle way to hold the reigns on Bill.” Shockley had grown a beard, which Jean detested. Just before a visit to Bell Labs by the distinguished British physicist Nevill Mott, Jean asked Bill to please shave off his beard. When he refused, “she grew two big pigtails for the event and tied pink bows on the ends.” When Bill expressed displeasure with the pigtails and bows, Jean said, “ ‘I’ll get rid of them if you get rid of your beard,’ which he did.” John and Jane saw the Shockleys less often as Jane grew less tolerant of Bill’s boastful displays. She was offended by his habit of bad-mouthing Jean in public. Jane began avoiding gatherings where she expected to see Shockley. She recalled a time when Shockley told a group that Jean’s inferior genes were the reason their children were less talented than he. Years later Shockley told an interviewer, “In terms of my own capacities, my children represent a very sig- nificant regression. My first wife—their mother—had not as high an academic-achievement standing as I had.” Small talk remained difficult for John. At social gatherings he often said nothing at all. Being in a conversation with Bardeen was often more difficult for the wives of physicists, because of the lack of a professional connection. “If you said, ‘Aren’t we having a nice winter season?’ he would say ‘Yes.’ That would be the end.” But while Bardeen’s social partner would often be at a loss, “he didn’t act as if he was uncomfortable.” Bardeen continued the habit he had acquired at the Naval Ord- nance Laboratory of separating his private and professional worlds. In wartime the compartmentalization had been dictated by national security. Now it was self-imposed. He told Jane he didn’t want to risk revealing proprietary information, but she knew he just pre- ferred to keep his spheres separate. Jane understood that “he wanted some part of himself for himself.” He was nearly always home in the evenings and on weekends, and he made time for family outings and visits. He rarely phoned colleagues from home. Jane also recalled that “when he’d get this far away look in his eyes you knew very well that he was somewhere else.” He had drifted into his other world. “Well, it spared him a lot of wasted time,” she muttered. Like his father, John was happy to let his wife rule the family 

120 TRUE GENIUS sphere. Jane did not object. She understood how centrally impor- tant family and home were to John and to his work. She was genu- inely pleased when he said he was glad to “be doing some physics again after four years of abstinence.” But she could not help feeling excluded from his “first love,” especially when he was embroiled in a problem. John’s low whispery voice plus Jane’s continuing struggle with hearing loss added to her growing sense of isolation. In later years she resorted to reading lips to deduce what others said on the basis of context. She was so adept that most people never knew she was almost completely deaf. On workdays John usually left Jane the family car for shuttling the children to their activities and running errands. He made his own way to Murray Hill, either by bus or in a carpool, where he enjoyed catching up on local news. Pulling up to the entrance of the Murray Hill facility on Mountain Avenue was the daily ritual by which he shifted between his two spheres. In the first decades after its incorporation on January 1, 1925, the Bell Telephone Laboratories had occupied an elegant, twelve-story brick and sandstone building at 463 West Street in lower Manhattan. The building flanked the western edge of Greenwich Village, a section of the city populated by artists, musicians, and writers. Within walking distance of galleries, theaters, coffee shops, and ethnic restaurants, Bell Labs scientists and engineers could feel the pulse of the great city. Those whose labs overlooked the Hudson River could watch steamboats and barges at work, or the crowded ferries transporting commuters between New York and Hoboken, New Jersey. Many commuters stood in line on the ferry to grab a doughnut and steaming cup of coffee in an effort to save a few minutes of their busy mornings. In the early years, the West Street building was a spacious home for Bell Labs. But even before the mid-1930s, the facility began to outgrow its accommodations there. The dust and grime of the city, and even its noise, became serious problems, especially for those working on electronics. Many projects expanded into nearby buildings, such as the Nabisco Build- ing, which had been designed for “baking purposes.” A few executives began to dream of a “country lab experiment,” a place outside the city aimed at supporting creativity. The idea emerged to build a facility in New Jersey reminiscent of Edison’s 

The Transistor 121 Menlo Park. This was, however, the era of the Depression, and a lack of funds postponed the expensive experiment. The idea to move the laboratory into a cleaner and larger space reemerged when the Second World War brought new funds and a much larger staff to Bell Labs. In 1941 the massive structure in Murray Hill known as “Building 1” was hastily completed in time to house the wartime exodus from the cramped West Street build- ing. The new mazelike array had been designed for several thou- sand scientists, engineers, and technicians. Many of its interior walls were movable, so that spaces could be repositioned to fit the needs of particular projects. To the physicist and writer Jeremy Bernstein, the functional space appeared as “a gigantic technologi- cal warren within which, at least at first sight, everything re- sembles everything else.” Initially Bernstein experienced the place as so menacing that he worried “in the process of going from one laboratory to another, I would take the wrong turn and never find my way out.” On Bardeen’s first day of work at Bell Labs on Monday, October 15, 1945, he showed up with the head cold he had caught during the family’s move to Summit. He spent the morning at West Street completing paperwork and having a routine medical checkup. Then he passed through the regular initiation ritual for all new employ- ees, which included selling his patent rights to the company for one brand new dollar bill. “I really feel that this is only fair,” Bardeen later told a reporter. “People can cooperate without worry- ing who is going to get the patent rights and this promotes a much freer exchange of information.” In the afternoon Bardeen found the small fourth-floor office to which he had been assigned in the B wing of Building 1. Although the Murray Hill facility was much roomier than West Street, there was still a premium on space. Many temporary workers were on site, completing the construction or continuing wartime projects. Bardeen’s two office mates, the experimentalists Walter Brattain and Gerald Pearson, had been studying semiconductors for over a decade. Sharing an office with experimentalists would have bothered some theorists, but Bardeen welcomed the opportunity to discuss practical matters with them regularly. Bardeen and Brattain would become close partners at work. Bardeen valued Brattain’s integrity and physics know-how. He also thoroughly enjoyed Brattain’s unrestrained and colorful use of lan- 

122 TRUE GENIUS guage, reminiscent of his brother Bill, or of Dutch Osterhoudt. The outgoing Brattain, in turn, appreciated Bardeen’s modesty, thought- fulness, and brilliance. He did not mind his reticence. Brattain’s graduate school friend Walker Bleakney had cautioned him about Bardeen. “You’ll find that he doesn’t open his mouth very often to say anything, but when he does, you listen!” It took Bardeen little time to become fully immersed in the problems of semiconductors. He approached them in both ways Wigner had trained him to begin any problem: breaking down larger problems, and reading all he could about what others had done before him. Of particular interest to Bardeen was what happens at the interface between two different semiconductors, or between a semiconductor and a metal. There was a large body of literature on this “surface” question. To answer well required the application of quantum mechanics. Some of the relevant physics was being discussed by the members of a weekly Bell Labs study group that had started to meet in the early 1930s. In those years, employees used the time freed up during Depression-era “layoff days” to educate themselves about the new quantum theory of solids. The group, which now included Bardeen, Shockley, Robert Gibney, Pearson, and Brattain, studied all the important papers on semiconductors, including those written up as military reports during the war. A great deal had been learned about semiconductors since Alan Wilson’s pivotal work in the early 1930s. Silicon and germanium had become the focus, partly because they are elements rather than compounds, making them the two simplest semiconductors. Dur- ing the war they were used as rectifiers in radar detectors to convert high-frequency alternating signals into direct-current signals. The conversion to lower frequencies was necessary so that the signals could be amplified using simple equipment. Much of the wartime research had focused on improving the properties of the so-called “rectifying interface,” the tiny point of contact between a thin metal wire called a “cat whisker” and a slab of semiconductor, either silicon or germanium. Electrically, this metal–semiconductor interface played the same role in an elec- tronic circuit as a vacuum tube diode, but it could function at higher frequencies. The research also dealt with “forming” the metal whiskers, producing pure semiconductor crystals, adding impurities (“doping”) to enhance conduction, manufacturing good 

The Transistor 123 crystals of silicon and germanium, and gaining an understanding of the physics involved. The physics of the central device in the radar detector, the crystal rectifier, turned out to be virtually the same as that of the first transistor. Bardeen reread the important prewar papers on semiconductor rectification by Nevill Mott in England, Walter Schottky in Germany, and Boris Davydov in the Soviet Union. He also read as much of the wartime literature that he could find. Because Bell Labs had been one of the major contractors to the wartime MIT Radiation Laboratory (Rad Lab), the Labs had on hand an almost complete set of the wartime reports on rectifiers. They included classic papers on rectification by Hans Bethe, who had been on the staff of the Rad Lab before moving to Los Alamos, and reports on germanium by members of Karl Lark-Horowitz’s group at Purdue. Using a specially developed superpure germanium, the Purdue group had built a rectifier having exceptionally low conductivity in the “back direction,” in which current flow was lower. Also in- cluded in the collection were papers on silicon by Fred Seitz at DuPont or at the University of Pennsylvania. The next time John saw Fred, he said, “Now I know what you did during the war.” The essential difference between a triode and a diode vacuum tube is the single crucial element known as the “grid” (see Figure 8- 1). In a diode, a heated, negatively biased filament emits electrons that travel to the positive plate. Because the negative electrons can flow only in one direction (i.e., from filament to plate), current can pass only one way. The diode functions as a one-way valve. The triode can function as a valve too, but it has in addition the possibility of amplifying signals. An element called the grid is placed between the filament and the plate in such a way that elec- trons must pass through it. Much as a faucet controls the flow of water, the grid controls the flow of electrons by virtue of its volt- age. When its voltage is negative with respect to the filament, the electrons’ acceleration from the filament is retarded. If the grid is positive with respect to the filament, it accelerates the electrons. In a circuit, this device can cause a signal applied to the grid to become amplified in a circuit through the plate. Around 1910, AT&T needed a technology that could amplify telephone signals. Theodore N. Vail, AT&T’s president, had made the decision to build a transcontinental telephone line connecting the East and West coasts of the United States. Diode tubes, already 

124 TRUE GENIUS Grid P Plate Plate Grid G Filament Filament F F Schematic FIGURE 8-1 The triode tube. in use in telephone switching circuits, could not amplify signals. Triode tubes could in principle do so, but they had only recently been invented, in 1906. When one of the triode’s inventors, Lee de Forest, demonstrated his device (which he called the “audion”) at AT&T, Harold Arnold, a physicist on the staff, realized that it held the answer to the company’s amplification problem. Modifying de Forest’s audion, Arnold designed the “repeater” needed for the transcontinental line. Arnold’s research group, the first such group at AT&T, subsequently evolved into Bell Telephone Laboratories. Unfortunately, the very feature that allowed the vacuum tube to function both as a diode and an amplifier—its heated filament— was also its fatal flaw. The filament’s work function (the energy needed to remove an electron) had to be overcome so that electrons could be emitted. It was the same problem that Bardeen and Brattain were studying at the time they first met in the 1930s. Thermionic emission of electrons consumes a huge amount of power, especially when many tubes are present in the system. Moreover, the filaments in tubes eventually burn out. For these reasons and others (including the cost, bulkiness, and fragility of glass tubes), Bell Labs looked to semiconductors to replace vacuum tube diodes and triodes. 

The Transistor 125 What made it natural to pursue a semiconductor amplifier was the magic of the interface between a semiconductor and a metal or between two different semiconductors. In a circuit this interface behaves in the same way that the crystal rectifier behaves in a radar receiver: like a valve, it allows current to flow much more easily in one direction than the other. The result of a one-way barrier at the surface, this rectification property of semiconductor interfaces had been discovered during the nineteenth century, but it was not yet understood. It had made possible the “cat’s whisker” radio detector, which Bardeen, Brattain, and Shockley had played with as teen- agers. The question arose: If adding a grid to a diode tube can turn the diode into a triode tube capable of amplifying signals, can one add something analogous to a semiconductor diode and create a semi- conductor amplifier? And if this were possible, what would function as the grid? Brattain and Shockley were among many researchers who confronted this question in the 1930s. Mervin Kelly, who had been head of the Vacuum Tube Department at Bell Labs from 1928 to 1936, was very familiar with the problem. He encouraged Shockley to pursue it in the late 1930s. In the late 1930s, Shockley shared a laboratory with Dean Wooldridge (later the “W” of the firm TRW), another physicist that Kelly hired in 1936. Wooldridge vividly recalled a crude experiment in which Shockley tried to convert a copper oxide diode (used as a “click reducer” in the telephone system) into a triode. Arranging two wires so they barely contacted the green oxide on either side of a rusty porch screen and adjusting the voltage applied, Shockley expected the jagged screen to function like the grid of a vacuum tube. The screen, according to Wooldridge, had been “out in the elements for years and years.” Shockley was “orders of magnitude away from anything that would work,” yet he had before him in the late 1930s “the three elements of a transistor.” Bardeen listened with interest to such stories about early semi- conductor experiments, especially as told by Brattain, who would enliven them with his colorful commentary. Brattain loved to tell how one day Shockley came to see him in his lab and asked him to build a workable version of this crude copper oxide amplifier. The seasoned experimentalist knew it was a hopeless project because there was no physical technique for inserting a screen into a barrier region only ten-thousandths of a centimeter thick. “I 

126 TRUE GENIUS laughed at him,” said Brattain. “I was quite sure it wouldn’t work. But I said, ‘Bill, it’s so damned important that if you’ll tell me how you want it made, and if it’s possible, we’ll make it that way. We’ll try it.’” Brattain was right, of course. “The result was nil. I mean, there was no evidence of anything.” Bardeen also heard stories about Russell Ohl’s interesting work just before and during World War II on silicon detectors. A feisty gnome of a man, Ohl was a radio engineer working in Bell’s radio lab in Holmdel, New Jersey. By the mid-1920s most radio buffs had replaced their crystal sets with vacuum tube sets, which could better separate stations whose frequencies were close to one another. A few engineers, however, including Ohl, realized that the old crystal sets were actually far more sensitive at high frequencies. Experimenting with old crystal sets found in secondhand radio shops, Ohl set out to determine which semiconductors work best at the higher frequencies. Trying more than 100 materials, he found that silicon worked best. The trouble was that the individual silicon samples worked erratically, rectifying only when the whisker con- tacted “hot spots” on their surface. The magnitude of the effect, and even its direction, changed unpredictably. In one test conducted in 1940, Ohl noticed that a particular silicon sample showed an enormous photovoltaic effect whenever he would shine a flashlight on it. In further study of this mysterious effect, he identified a junction within the body of the sample com- posed of two kinds of silicon, now called p- and n-type silicon. Further study of this junction might have resulted in an amplifier, but the war diverted Ohl to other problems. The work that led to the transistor began on October 22, 1945. Shockley asked Bardeen to look over a design that he had sketched in his notebook six months earlier for a silicon “field effect” amplifier. An electric field was applied perpendicular to a thin slab of silicon; the field drew charges in the slab to its surface. In a thin sample, Shockley argued, the field would cause a substantial change in the available charge carriers. In this design, the field would play the role of the grid. But the design did not work in actual practice, and Shockley could not explain why. Drawing on the best available theories about semiconductor rectification, those of Mott and Schottky, Shockley 

The Transistor 127 had calculated that for the predicted effect not to be observed, the changes in conductivity would have to be less than 1/1,500 of the predicted amount! Stumped, he showed the calculation to Bardeen, who found nothing wrong with it. Bardeen recognized that the theory must be wrong, or at least incomplete. To explain how was just the kind of puzzle he loved to attack. The problem was both challenging and practically impor- tant. In addition, his competitive spirit was aroused, for the prob- lem had stumped another excellent theorist. He worked on it for the next five months. Bardeen realized that, as in the work function problem, the electrons in this system are more mobile than the ions. This implied that at the surface of the semiconductor the electron wave function (a measurement of the probability of finding an electron there) would extend slightly beyond that of the ions. The result was a small excess of negative charge on the surface, rendering the total charge neutral. By March 1946, drawing on earlier surface- state work by Shockley and Igor Tamm, Bardeen had developed a new theory predicting that a substantial number of the electrons would be trapped in surface states. Unable to contribute to the conduction, the trapped electrons substantially decreased the change in conductivity caused by the applied field. That was why Shockley’s field effect design failed. But were the surface states real? And if so, how did they behave? To progress further, these questions had to be addressed. The solid-state group divided up tasks: Brattain studied surface properties such as the contact potential; Pearson looked at bulk properties such as the mobility of holes and electrons; and Gibney contributed his knowledge of the physical chemistry of surfaces. Bardeen and Shockley followed the work of all members, offering suggestions and conceptualizing the work. “It was probably one of the greatest research teams ever pulled together on a problem,” said Brattain: I cannot overemphasize the rapport of this group. 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. We went to the heart of many things during the existence of this group, and always when we got to the place where something needed to be done, experimental or theo- 

128 TRUE GENIUS retical, there was never any question as to who was the appropriate man in the group to do it. To better understand the surface states, Bardeen proposed an experiment in which the temperature was lowered enough so that the electrons in the surface states would be “frozen.” That way, he argued, the field might be able to get through. Pearson tried the experiment, which involved using liquid nitrogen to cool the semi- conductor. When he applied 500 volts, he observed a field effect! It was a triumph for Bardeen’s theory of surface states, but the effect was fifty times smaller than predicted. Bardeen deduced that charge carriers in thin films are less mobile than they are in bulk material. This was a second reason why Shockley’s experiment had failed. A successful design would need to get around both obstacles. The group tried many related experiments, but for the next eighteen months they made little progress. During the remainder of 1946 and throughout 1947, Shockley’s engagement with the field effect studies flagged, for he developed new interests during the summer of 1947, when he and Bardeen took an extended tour of European laboratories. It was to be the last time Bardeen and Shockley would share an extended collegial exchange. The two started out late in June 1947, stopping first in London. Bardeen noted that the city still exhibited “much bomb damage.” Food was “in short supply” and suitable lodging was “often diffi- cult to find.” Shockley was stimulated by a conference they attended during the first week of July in Bristol on recent work in the new field of “dislocations,” the study of departures from regu- larity in crystals that take the form of linear displacements of the crystal lattice. The fact that the dislocations can propagate through a crystal so captured Shockley’s interest that on his return to Bell Labs in August he refocused his research around this phenomenon. On a Sunday off during the conference, Bardeen brought Shockley along on a visit to Muriel Kittel and her family in Cheltenham. During the war, the Bardeens had become friends with Muriel and Charles Kittel, when both men had worked at the NOL. Kittel had recently accepted a position at Bell Labs. “I told [Muriel],” John wrote Jane, “that you were expecting to assist them any way you could when they get in.” Muriel wrote to Jane years later that when “John and Bill turned up at my parents’ house to tell me of their pleasure at Charles’s appointment at Bell Labs,” her 

The Transistor 129 parents were surprised and “tremendously touched by such kindness.” While Shockley was learning about dislocations, Bardeen caught up with new research on superconductivity. He discussed the phenomenon on visits in Holland, to the universities in Delft, Amsterdam, Eindhoven, and Leiden. He had productive meetings at Philips Laboratories in Eindhoven with Hendrik G. B. Casimir, Jan Verwey, and others. Unfortunately, there was just one room available in the small hotel Bardeen and Shockley felt lucky to find in Eindhoven. As the room “had only one double-bed,” the two men slept “with our heads at opposite ends.” Bardeen wrote to Jane regularly during the two-month-long trip. The separation was almost as long as the one they had experienced during the war. John wrote from Eindhoven shortly before their wedding anniversary. “Sorry we can’t be together on the 18th. I will be thinking of you. We will have to have a real celebration on our tenth next year.” It was drizzling when Bardeen and Shockley returned to their room at Queens Gate in London on Sunday, July 19. Despite the rain Shockley made plans to go rock climbing with a British friend and then to spend a few days visiting Edmund Stoner at the Univer- sity of Leeds to talk about magnetism. While Shockley was gone, Bardeen went to the theater and saw 1066 and All That, a musical based on British history. The next Monday Bardeen took the train to Cambridge. As it rattled along, he finished a letter to Jane that he had started on the previous day. He indulged in some early morning grumpiness. “One of my shoes is disintegrating,” he wrote. “We have used very little of the food I brought along. I could better have used the space for extra shoes and more bath towels. The latter are seldom furnished in the hotels.” On a more positive note, “Bill says I got a raise starting July 1. He is not sure of the amount, but it may be as much as $60 or $70 a month.” In Cambridge, Bardeen visited the Mond Lab, where he spoke with David Shoenberg, the author of one of Bardeen’s favorite texts on superconductivity. Shoenberg’s graduate student, A. Brian Pippard, was also present. At Shoenberg’s suggestion Pippard had studied Heinz London’s prewar experiments on the penetration depth of magnetic fields in superconductors. He could draw now on microwave techniques to measure the surface impedance of 

130 TRUE GENIUS metals at high frequencies, for during World War II Pippard had worked on microwaves at the British Royal Radar Establishment at Malvern. Pippard realized that the microwave region was a promising one for studying superconductivity, because microwave energies are intermediate in frequency (between zero, at which super- conductors do not dissipate energy, and the optical frequencies at which they dissipate normally). The London theory, with its radical assumption of the long-range ordering between electrons, had pre- dicted the size of the penetration depth of magnetic fields. If Pippard could measure this depth experimentally, it would be a powerful check of the London theory. During his visit, Bardeen also had the opportunity to discuss superconductivity directly with Heinz London himself. They decided to go to Paris next. Bardeen wrote Jane that he hoped “to get to Paris next Sunday [July 26] but we don’t have any reservations yet.” In fact, he confessed, “We don’t have any for the Paris–Zurich leg either, but are hoping for the best.” After they arrived in France—on “the hottest day in Paris’ history”—they visited Louis Néel, one of the experimentalists whose work had put the study of magnetism on a sure footing. Néel “took the not unreasonable attitude that, when in France, you speak French.” So on that hot day, the two sat with Néel at an outdoor cafe and “spoke to him in our high school French.” Their last stop was Switzerland, where Bill and John swam in Lake Zürich before visiting Wolfgang Pauli, who in the 1930s had called solid-state physics a “physics of dirt.” They followed the visit to Pauli with a few days of vacation in the Alps. Bardeen loved the mountains. “As you must have gathered from the postcards,” he wrote Jane, “I like Switzerland very much.” In fact, “If I were coming again strictly for vacation, I think I would skip the other countries and come straight to Switzerland.” He told Jane that Shockley “bought quite a lot of stuff here in Zürich for climbing in the high mountains,” including “nailed boots, windbreaker jacket, and various kinds of mountaineering equipment. The only thing I got is a pair of sport oxfords with corrugated rubber soles for hiking.” He justified the expense saying that he could “use them later for golf or all around wear.” Bardeen wrote that he had “learned a lot during the trip, and have picked up some information that may be useful to the Lab. Whether or not it’s enough to pay for the trip is hard to judge.” In 

The Transistor 131 any case, it had been “very hard work—much harder than you can imagine without doing it—but it’s also been a grand experience.” He added, writing on August 1, “In another week or so I will be home—and glad to be there.” The “magic month” that culminated in the transistor began in the middle of November 1947, three months after Bardeen and Shockley returned from Europe. Brattain had encountered an apparently innocent problem during the course of one of his experi- ments. Droplets of water condensing on the apparatus were causing a spurious effect. In an effort to avoid the cumbersome two-week- long job of pumping out all the water, he attempted a quick fix. “I’m a lazy physicist,” he later reported. “I like to do things in the easiest way.” He immersed the system in various liquids including distilled water and was “completely flabbergasted” to find that the photovoltaic effect he was studying increased whenever the liquid was an electrolyte. Soon Brattain was demonstrating these things “to anybody in the group that would listen.” Bardeen suggested that the mobile ions in the electrolytes might be creating a large enough electric field to overcome the surface states. Then Gibney said, “Wait a minute. You’ve got a potential on there, haven’t you?” “Yes,” Brattain responded. “Let’s vary this thing just a bit,” suggested Gibney. When Brattain and Gibney did so (see Figure 8-2), they noticed that when they used either water or an electrolyte, a layer of posi- tive charge would form on one surface and a layer of negative charge on the other. “We could vary the photo emf [electromagnetic force] from anywhere to a very large value to zero to change its sign.” Suddenly the team realized that they might be able to build a field effect amplifier after all! Brattain and Gibney described their find- ings in a patent disclosure dated November 20. The major remain- ing problem concerned how to project the change caused by the varying potential onto a second circuit. By now Shockley was beginning to behave somewhat strangely with respect to the group. He kept to himself more than before. Although he continued to offer suggestions, he saved most of his energy for his own work on dislocations and on the flow of elec- trons through alkali and silver halides. In the privacy of his home, he began to sketch designs for a different approach to the solid- state amplifier. He told no one else in the group about them. 

132 TRUE GENIUS Reference Light electrode – – – – – – – – – – ++ + + + + + + + Potential – – – – – – – – – Electrolyte bias V1 ++ + + + + + + + P-type silicon + FIGURE 8-2 Gibney’s suggestion to vary the potential. In contrast Bardeen engaged himself fully with Brattain and Gibney’s new results. His mind played constantly with the possi- bilities. On Friday morning, November 21, he walked into Brattain’s office and “suggested a geometry for making an actual amplifier.” Brattain responded, “Come on John, let’s go out in the laboratory and make it.” The two had begun to work together regularly in Brattain’s lab. Bardeen loved to peer over Brattain’s shoulder and watch him pre- pare his experiments. Sometimes Bardeen would offer Brattain a hand in routine tasks, such as recording measurements or holding a piece of apparatus in place while Brattain soldered it. Bardeen typically would probe Brattain for clarification on questions of tech- nique or material. In pondering the data Bardeen often made inter- pretive suggestions based on his deep understanding of the physics. Brattain valued that input. The interplay between the two was reminiscent of Bardeen’s work with Bridgman, but it was more fun because of Brattain’s sense of play and adventure. The Bell Labs environment supported the collaboration with countless resources hard to come by in an academic lab—technicians, materials, equip- ment, and even patent attorneys. Bardeen’s proposal of November 21 (see Figure 8-3), although not yet an amplifier, contained most of the elements of the first transistor. Bardeen had modified the basic structure of Brattain and Gibney’s promising experiment, using results from the wartime radar program. As Brattain later recalled the design: The geometry was essentially one of taking a point contact, some- 

The Transistor 133 Output current, I2 Metal point Wax + Water drop V2 – – Control V1 voltage N-type inversion layer + P-type silicon FIGURE 8-3 Bardeen’s proposed field effect amplifier, November 21, 1947. how insulating its surface, and putting the point down on the semi- conductor surface, and then surrounding it with a drop of an electro- lyte to which we made contact with another metal, thus hoping to modulate the flow of current from the point to the semiconductor by a fine electric field through the electrolyte. Bardeen suggested that for the electrolyte they try a drop of distilled water, and for the semiconductor, a slab of p-type silicon with an n-type “inversion layer” (a thin region that forms in cer- tain circumstances near the surface of a semiconductor). He was well aware of the phenomenon of the inversion layer after his surface-states work in 1946. Indeed, he held a patent on the idea. He hoped to use the fact that in such a layer most carriers of charge are opposite in sign to those in the bulk material. (Thus if holes carry most of the current in bulk samples, electrons would carry most of the current in the inversion layer, and vice versa.) In addi- tion, the wartime research had shown that much larger changes in conductivity occur in bulk samples than in thin films. Thus by using an inversion layer contiguous with the bulk material, instead of a thin film deposited on its surface, one could perhaps get around not only the difficulty of depositing a very thin layer of semi- conductor, but also that of the low mobility of charge in thin films. For the contact, Bardeen proposed using a sharp tungsten point. Brattain suggested insulating the point from the water using a thin layer of wax. The big advantage of such a “point-contact” design 

134 TRUE GENIUS was that there was “a simple way in which you could do it,” Bardeen explained. It was “the sort of experiment you could set up and do in a day” because the technology of making point contacts had been refined during the war. Brattain and Bardeen knew they were getting close when they tested Bardeen’s design. It worked to about 10 cycles, giving ampli- fication of current and power, but not of voltage. Brattain said, “I told my driving group that night, going home, that I felt that I had taken part in the most important experiment I had ever taken part in my life.” As Bell Labs had not yet released the news, “the next evening I had to swear them to secrecy.” Over the next month Bardeen and Brattain worked steadily on the semiconductor amplifier problem, on occasion even on week- ends. They tried to change only one or a small number of features at a time to keep track of their progress as they explored countless variations of materials and geometry. They tried gold instead of tungsten for the electrodes; Duco lacquer in place of wax for the insulation; and water, tetramethyl ammonium hydroxide, and a series of gels for the electrolytes. They achieved their best results using an electrolyte that was a gooey mixture they called “gu,” made of glycol borate and glycol bori-borate. The viscous fluid, taken from a lead battery, worked better than water, which tended to evaporate. For the geometry they settled on a configuration in which the electrode was a gold ring set down on the gu, with the sharp metal electrode passing through. They tried using a drop of gu on a junction of p- and n-type silicon and pressing down two points instead of one on the silicon slab. The most critical change in the design resulted from a sugges- tion from Bardeen on Monday, December 8. During a lunch-time discussion with Brattain and Shockley, he said he thought they should try germanium instead of silicon. In particular, Bardeen thought that by using the special “high-back-voltage germanium” doped with tin that researchers at Purdue had developed during the war, they might improve the amplification significantly because of its high resistance to currents flowing in the “back” direction. “John Bardeen was great at coming up with approximate guesses of this kind and making the right guess,” Brattain recalled. Brattain happened to have a piece of the special germanium in his laboratory, so he and Bardeen tried the experiment that very afternoon. Brattain pushed the gold point into the germanium 

The Transistor 135 through a drop of the gu, to which they applied a few volts. They were startled to measure both a voltage amplification of two and a dazzling power amplification of 330! They also noticed a mysteri- ous change in the direction of the current. “This is the opposite of what one might expect,” Bardeen noted. Might positive carriers—holes, rather than electrons—be re- sponsible for the transport? Brattain recorded, “Bardeen suggests that the surface field is so strong that one is actually getting P-type [i.e., positive] conduction near the surface.” It was a strange idea, but it fit their data. Could they be producing an inversion layer electrically? Could negative ions in the electrolyte be inducing a layer of positive charge consisting of holes just beneath the sur- face? Partly through accident and partly through good experimen- tation, the two had managed to increase the population of holes at the surface. The negative voltage on the tungsten point was indeed driving the electrons away. Increased by the electric field from the droplet, the hole population had raised the conductivity, giving the observed power gain. Gradually they realized that the holes, the ghostlike particles that had intrigued Bardeen ever since he learned of their existence, were functioning as the grid. These holes, alluded to by Peierls in 1929 and well described by Heisenberg and Wilson in 1931, were the key actors in a new effect they would come to recognize over the next several days. Two days later, on December 10, Bardeen and Brattain repeated the experiment using a specially prepared germanium sample. They saw an even greater power gain—a dramatic factor of 6,000! But as the frequency response remained poor, the device could not yet be used to amplify the range of the human voice. They soon realized that the sluggish frequency response was caused by the ions moving too slowly in the gu. They decided to dispense with the electrolyte and instead take advantage of an effect that Brattain happened to notice earlier that day. When he had applied a steady electric field to the glycol borate, the electrolyte had etched the surface of the germanium and caused a green oxide film (similar to rust) to form. “I can remember the green color un- der the glycoborate,” Brattain recalled. “So, seeing this film, we thought, ‘Ah. This oxide film must be insulating. If it is, we can form the film and put metal electrodes right on top of the film, get this field effect without the electrolyte and get [power gain at] the 

136 TRUE GENIUS higher frequencies.’” Because the film was rigid, unlike the gu, the frequency response should improve. Because it was thin, the elec- tric field should be higher. Thus, they argued, the hole population should be substantial enough to achieve a higher power gain. For the next experiment, performed on December 11, 1947 (see Figure 8-4), they used the oxide film in place of the electrolyte. Brattain reversed the currents using an n-type rather than a p-type sample. To serve as the voltage plate, Gibney carefully evaporated a gold spot on the oxide film, leaving a hole in the center of the spot for the metal point to pass through to the germanium. When an electrical discharge spoiled the hole, Brattain compromised: he separated the two neighboring circuits by placing the point near the edge of the gold. Once again all the elements for observing a field effect seemed to be present. But the experiment did not work as planned. When on Decem- ber 15 they applied a positive voltage to the gold spot and a nega- tive one to a point next to the spot, Brattain saw modulations of both the output current and voltage. There was no power gain, but the voltage doubled! It worked the same at high frequencies. “This voltage amplification was independent of frequency [from] 10 to 10,000 cycles,” Brattain wrote. Curiously, the modulation occurred only when the gold was positively biased: “I got an effect of the opposite sign,” Brattain recalled. Output current Evaporated gold Tungsten point Oxide layer + P-type inversion layer + – Control N-type germanium V1 V2 voltage – + I1 I2 FIGURE 8-4 Bardeen’s and Brattain’s field effect model of December 11, 1947. 

The Transistor 137 “We knew that something different was happening,” said Bardeen. It became clear that “holes were flowing into the germa- nium surface from the gold spot and that the holes introduced in this way flowed into the point contact to enhance the reverse cur- rent.” This, Bardeen recalled, “was the first indication of what later was called the transistor effect.” Based on the holes introduced into the germanium, the device became known as the “bi-polar” transistor. It worked on a completely different principle than the field effect device that Shockley had sketched more than two years earlier. “What we didn’t know then,” Brattain explained, recalling that classic experiment, “was that the oxide that was formed was soluble in water.” They had not expected that “when we washed the gu off, we washed the oxide off too.” This simple property of the material being used—the solubility of the oxide that forms on germanium—would allow Bardeen and Brattain, but not Shockley, to patent the first transistor. Had they not switched from silicon to germanium, the oxide would not have washed off. Had they stayed with silicon, they might eventually have arrived at a field effect transistor similar to the MOS-FET (metal oxide–silicon field effect transistor), in com- mon use today. Bardeen would later joke that his and Brattain’s invention of the point-contact transistor slowed the quest for the field effect transistor for several years. But as Nick Holonyak later remarked, that “really could not have occurred without all the semiconductor work the point-contact transistor set into motion.” Did Bardeen and Brattain fully understand the behavior of the holes in these historic experiments in December 1947? Did they, for example, realize that the holes flow not only along the surface but into the bulk of the semiconductor as well? These questions have become the focus of a recent debate among a small group of engineers and historians. The available documentation cannot completely resolve the issue. One party in the debate argues that Bardeen’s training as an engineer closed his eyes to the possibility of seeing this “hole injec- tion.” We consider this implausible. Bardeen was, after all, fully trained as a theoretical physicist, with a Ph.D. from Princeton and three years of postdoctoral study at Harvard. Moreover, the clue about the flow of holes into the bulk had been published almost a decade earlier, in an abstruse mathematical paper written by Boris 

138 TRUE GENIUS Davydov. Bardeen probably read this paper, for he was in the habit of reading all he could of the relevant physics literature. Even had he not grasped all the implications of Davydov’s work, he would have recognized its relevance. Bardeen was among the few physi- cists who at that time had the proper training to understand Davydov’s work. A more reasonable interpretation, supported by much circum- stantial evidence, has been put forth by Holonyak, who has argued that Bardeen understood how the holes behave either immediately or very soon after the invention of the transistor. Characteristi- cally, Bardeen would have been cautious about publicizing his con- jecture until there was adequate proof that holes flow not only in the surface but also in the bulk of semiconductors. The debate seems to boil down to how long it took for Bardeen and Brattain to realize that holes flow into the bulk. “There was a period in which one was not sure whether these holes were flowing in the space-charge layer or whether this was radial injection into the crystal,” Brattain told an interviewer in 1964. A year earlier Bardeen had explained in an interview that “our first thought was that the holes were being introduced just in this surface layer here and that the field of the collector pulled them in, so that the action is mainly confined just to this surface layer. We did a number of experiments to try to find out just what was going on here.” What these experiments indicated, Bardeen said, “was that you’re intro- ducing excess carriers into the surface to lower the conductivity in the neighborhood of the point.” Throughout this period of uncer- tainty, Shockley was conducting his own studies. These “suggested that he could inject holes through a p-n junction into the material.” In any case, this issue of whether the holes flow into the bulk was soon cleared up by the definitive experiments of John Shive (discussed in Chapter 9). By mid-February 1948, Shive had demon- strated “that holes could flow through the bulk, in conformity with Shockley’s ideas.” Bardeen and Brattain spent a few days trying to improve the gain in their experiment. “The observed effect” was not very large and no real power amplification was seen. Bardeen suggested “that if we really were introducing carriers into the surface and wanted to get a real large interaction, you have to get the electrodes ex- tremely close together, within a thousandth of an inch or so.” He therefore proposed a geometry (shown in Figure 8-5) that would 

The Transistor 139 Spring I1 I2 Gold foil Polystyrene Gold foil wedge + Slit – Input V2 Output signal V1 P-type inversion layer signal – + N-type germanium FIGURE 8-5 The first transistor, December 16, 1947. allow the holes to flow closer to the input signal. The wire and gold spot would be replaced by two metallic line contacts set down on the germanium with a separation between them of only a few thou- sandths of a centimeter. (At this stage they believed that line con- tacts would give a stronger effect than point contacts. Later they realized that point contacts would do if enough current flowed in them.) Following the suggestion of Bardeen’s, Brattain created the two line contacts and separated them by about 0.004 centimeters by wrapping a piece of prewar gold foil around the apex of a poly- styrene triangle. “I slit carefully with the razor until the circuit opened” and filled the cut with wax. Using a spring, he pushed the triangle down on the germanium. By wiggling it “just right,” he succeeded in making contact with both points. “I could make one point an emitter and the other point a collector,” Brattain recalled. 

140 TRUE GENIUS This particular experiment worked the first time they tried it, on Tuesday afternoon, December 16, 1947. In one of their first tri- als, at 1,000 cycles per second, they achieved a power gain of 1.3 and a voltage gain of 15. “I had an amplifier with the order of mag- nitude of 100 amplification clear up to the audio range,” Brattain boasted. “It would sometimes stop working, but I could always wiggle it and make it work again.” The transistor was born. That evening, when John came home to Primrose Place, he told Jane about the results. She recalled, “It was one of those days he would drive in and put his car in the garage, which was in the back of the lot, and then come in the back door.” As usual, “the surest way to find me was in the kitchen,” she said. Wading through the “kids all over the floor,” John murmured almost inaudibly, “We discovered something today.” Jane hardly glanced up from the sink. “That’s great,” she responded automatically. But as he typically said nothing about his work, she knew it had been a special day. Some time later she found out “that the something was the transistor.” Bardeen and Brattain could not yet risk showing their inven- tion to Bell executives. Brattain recalled that “at each level of super- vision. . . . there was hesitation about informing the next level for fear an announcement of such importance would turn out to be a fluke and that their faces would be red for prematurely claiming something so important.” They spent another week verifying everything. By Tuesday afternoon, December 23, the group was ready to announce the discovery. They gathered nervously for the demon- stration, held in one of the executive offices at Murray Hill. Brattain had arranged the apparatus so that some observers could speak over the input circuit while others could hear the output signal over earphones, or see it displayed on the screen of an oscilloscope. He described what happened in his notebook on the following day, Christmas Eve: The 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. By measurements at a fixed frequency in, it was determined that this power gain was the order of a factor of 18 or greater. Bardeen also wrote about the demonstration somewhat later in the day. By then he and Brattain were seeing “voltage gains up to 

The Transistor 141 about 100 and power gains up to about 40.” What most interested Bardeen at this point was the mechanism by which the holes were entering the semiconductor, spreading out in the inversion layer and affecting the flow of current. In an effort to explain how the holes in the inversion layer might be causing the amplification, he drew an analogy with the familiar system of the triode: When A [the gold electrode] is positive, holes are emitted into the semi-conductor. These spread out into the thin P-type layer. Those which come in the vicinity of B [the tungsten point] are attracted and enter the electrode. Thus A acts as a cathode and B as a plate in the analogous vacuum tube circuit. The ground corresponds to the grid, so the action is similar to that of a grounded grid tube. The signal is introduced between A (the cathode) and ground (grounded grid). The output is between B (the plate) and ground. The signs of the potentials are reversed from tthose in a vacuum tube because conduction is by holes (positive charge) rather than by electrons (negative charge). “The analogy was suggested by W. Shockley,” he added. This was the climax of Bardeen’s productive involvement with Shockley.