The Man Behind the Microchip Read online

  That Noyce’s notebook entry served to codify what virtually everyone in the lab would have said had they thought about it does not diminish the importance of his work. Anyone walking through a display of modern art has privately thought, “I could have done that.” Perhaps, but the relevant point is that we did not. The artist did.

  JUST SIX WEEKS AFTER Noyce made his integrated circuit notebook entry, Ed Baldwin, the general manager found through the Wall Street Journal, announced that he was decamping to form his own semiconductor operation, which would operate as a wholly owned subsidiary of a larger firm, Rheem Manufacturing. Baldwin had probably not planned to leave in March, but Richard Hodgson, who had gotten word of Baldwin’s talks with Rheem, flew out to Mountain View and summarily fired him with the immortal words, “I wish you lots of luck—all of it bad.” Baldwin did not leave alone. He took with him eight senior Fairchild operations people, including five key engineers, and a process manual detailing how to build Fairchild’s mesa transistors. Shortly after his departure, Baldwin met with an employee of William Shockley who wanted to gather more details on Fairchild Semiconductor’s operations and their possible link to Shockley Labs.23

  The Fairchild Semiconductor founders, who felt the sting of having done unto them what they had done unto Shockley, recognized that Baldwin’s departure was “a disaster” for the company, not yet two years old. Baldwin had functioned as the firm’s de facto CEO, coordinating technical, manufacturing, and business operations, and serving as the official emissary to the parent company—an exquisitely important role at a time when Semiconductor was still receiving its monthly “allowance” from Camera and Instrument.24

  Noyce was the obvious choice to replace Baldwin. “God, he was head and shoulders above anybody in the business,” recalls Tom Bay. “By this time, we’d met most of them.” If Noyce were general manager, Moore could more than ably manage R&D, Bay had marketing under control, and Kleiner and Blank knew manufacturing. Furthermore, as Bay put it, “everybody had still reported what they did to Bob” even when Baldwin had been around. Noyce not only knew more than anyone else about what was happening at Fairchild, but he also had a flair for corporate politics and a patience for dealing with nontechnical people that the other founders lacked (and that a few founders found downright distasteful). This skill would be particularly important in working with John Carter and the other Camera and Instrument executives.25

  As he had a year before, Noyce again resisted the move into management. He had no manufacturing experience, no financial experience, no sense of even rudimentary business tools such as balance sheets and profit-and-loss statements, and no desire to leave the lab, where he was on an inventive roll and “felt sure of [him]self.” He had not changed his managerial approach since his days at Shockley: he still displayed the same tendency to make suggestions (“why don’t you try” or “have you considered”) rather than to issue commands, and he offered the same impression that he was not really in charge of the other men so much as simply happening to do a bit of administrative work in addition to his bench work. He led the weekly lab meeting, but to all outward appearances, the meetings ran themselves. He reviewed monthly progress reports from each of the half dozen research groups and determined which innovations were sufficiently novel and potentially lucrative enough to merit the $1,500 in attorney’s fees and hours of work necessary to patent them. But he did all of this in such close consultation with his subordinates that they felt they worked with him, not for him.26

  In the lab, Noyce had trusted people to fulfill their commitments, and he rarely followed up in a systematic way to confirm that promised work was actually delivered. “If you’re looking at a long-term research program, in general, the people that are doing it [the research] are in the best position to evaluate it, not the people that are supervising it,” he said. “The people that are supervising it are more dependent on their ability to judge people than they are dependent on their ability to judge the work that is going on.” Noyce believed that most people, given enough freedom, will choose to do the right thing. This message had been etched in his mind by his father and was further reinforced during his semester’s banishment at the Equitable insurance company while in college. He had spent his days absorbed in actuarial tables, but he thought he noticed that the people who bought the most life insurance tended to die younger than the actuarial data predicted they should have. Noyce interpreted this to mean that somehow people just instinctively knew the right things to do and should be left alone to do them.27

  Such faith in his fellow man, which he maintained throughout his life, led Noyce to give his employees free rein. This approach worked particularly well in the Fairchild lab because the people reporting to him—the other Fairchild founders and a few choice newly hired employees—neither needed nor wanted to be told exactly what to do or how to do it. Indeed, Noyce was “a very good supervisor of technical people,” according to Jean Hoerni, precisely because he was “casual” and “didn’t interfere” with his researchers’ work. Creative freedom and collaboration, which proved crucial to the young company’s technical success, blossomed under Noyce’s laissez-faire management of the lab.28

  Even when he was managing the lab, however, Noyce thought of himself as a scientist, not a businessman. “I could direct the work and see that it was channeled properly, so there wasn’t any great personal trauma involved in that switch [from scientist to head of R&D]. The switch from directing a research program into directing a complete commercial program, however, was quite a traumatic one.” He added, “It was with a great deal of fear of inadequacy, if I can put it that way, that I got into [an] administrative role”—so much fear, in fact, that he would agree only to a six-month trial run as general manager, after which he planned to return to the lab.29

  During this trial period, Noyce oversaw the introduction of seven new transistors, the building of a new diode plant in Santa Rosa that quintupled the company’s manufacturing space, and a ten-fold increase in the size of the employee base, which reached 1,260 at the end of 1959. The R&D operation alone was now as big as the entire company had been only a year before. He also approved the filing of two lawsuits: one charged Rheem with theft of trade secrets; the other, against Baldwin personally, alleged a “breach of confidential relationship.” Both suits were settled out of court.30

  Beyond his work on internal Semiconductor matters, Noyce served as the company’s public face. He met with representatives from Japanese electronics firms in town for a conference. He granted interviews and made presentations to the financial community in San Francisco and in New York. He attended several meetings in Syosset and declared “being in on major actions before they were taken” to be “good for my ego.”31

  As general manager, Noyce tried to track the activities in the lab as closely as he could. Back in R&D, Jean Hoerni had spent the six weeks after his January 1959 patent disclosures attempting to translate his intriguing ideas about oxide layers into silicon reality. He asked Jay Last, with whom he shared an office, to build the extra masks he needed to coat transistors with a layer of silicon dioxide. On March 12, one week after Baldwin left, Hoerni invited several people to join him for a dramatic demonstration of his oxide ideas. On his lab table lay a transistor made with the oxide icing and not yet in its canister. Under a microscope, this transistor looked dramatically different from the mesa devices the group had been building. There were no elevated surfaces in Hoerni’s transistor; it was flat, with all the electrically active regions terminating in the same plane. Its shape resembled a bull’s-eye with one part of the outer ring pulled out a bit to make room for a wire—almost like a teardrop.32

  Eying the assembled group, Hoerni spat directly on the device he, Last, and their technicians had spent weeks building. Heresy! Transistors were handled with tweezers in rooms as clean as possible. This transistor, however, weathered its nasty baptism with no ill effects, as an attached oscilloscope confirmed. This was remarkable. Any device hardy enoug
h to survive saliva directly on its surface could undoubtedly withstand a pencil tap on its canister. The buzz that arose from the assembled group was surprisingly matter of fact: is this transistor a fluke, or can we make a million of them? Because if we can make a million, then we can bypass this tap problem completely—and Baldwin’s stolen process manual will describe how to build a soon-to-be-obsolete product.

  A mightily impressed Noyce likened Hoerni’s oxide layer to “building a transistor inside a cocoon of silicon dioxide so that it never gets contaminated. It’s like setting up your jungle operating room. You put the patient inside a plastic bag and you operate inside of that, and you don’t have all the flies of the jungle sitting on the wound.” Gordon Moore, the new head of R&D, was more wary. A new process with another masking layer—that would not be easy to implement. Yields would definitely be lower than for mesa devices. Once again, where Noyce admired the inspiration, Moore foresaw the perspiration.33

  Hoerni’s jungle transistor clearly merited a patent. Two, in fact: one for the bullseye structure and one for the process to build it. It also needed a name. Fairchild Camera and Instrument vice president Richard Hodgson, who immediately came to Mountain View to see the device, suggested calling the nearly flat transistor and the process that produced it “planar.” He wanted to copyright the name, but Noyce disagreed, explaining that he thought Fairchild would get more advertising value if the industry adopted “planar” as a generic descriptor while Fairchild advertising stressed that the process was invented there. Hodgson deferred to Noyce’s opinion.34

  What happened next is unclear. Something motivated Noyce to dust off his integrated circuit notebook entry in March 1959. The precipitating event may have been Texas Instrument’s announcement, in mid-March, of a breakthrough in “Solid Circuits,” which purported to be an entire circuit on a single semiconductor chip—precisely what Noyce had described in his notebook entry.

  In the fall of 1958, a young Texas Instruments researcher named Jack Kilby set out to build an integrated circuit. By early 1959, he had built a complete circuit on a single germanium substrate. Kilby’s circuit was meticulously hand assembled with a network of gold wires connecting the components to each other. The wires precluded the device from being manufacturable in any quantity, a fact of which Kilby was well aware, but his was undoubtedly an integrated circuit of sorts. As soon as the patent work was filed, Texas Instruments proudly announced its invention.

  It is not hard to imagine this announcement triggering Noyce’s competitive ire. He thought the Texas Instruments circuit was cumbersome—“not aesthetic” was his description, a fairly harsh criticism in a technical world that values elegant, clean solutions to messy problems. Noyce’s notebook entry, by contrast, was highly “aesthetic.” It required no wires to interconnect components on the chip; it eliminated much of the tweezers’ work of connecting chips to each other by hand; and it took advantage of the planar process, one of the most elegant breakthroughs in semiconductor history. Gordon Moore remembers that Noyce called a meeting specifically to discuss the Fairchild response to the Texas Instruments circuit, and that it was during this meting that Noyce introduced his ideas about integrated circuits. No one else remembers the meeting, however, and no record of it survives.35

  A second version of events holds that shortly after Hoerni demonstrated the planar process, Fairchild’s patent attorney asked Noyce and a few key technical men to think as broadly as possible about how the process could be used, so that the patent could be written to cover the greatest number of potential applications. This seems the most likely explanation for Noyce’s having resurrected his integrated circuit ideas in March. Good patent attorneys often make these sorts of requests of their clients, and the timing, with Noyce’s integrated circuit work coming to light almost immediately on the heels of the successful planar demonstration, makes sense. Moreover, if the attorney’s request were the motivation, it would offer a key explanation for why Noyce’s version of the integrated circuit ultimately proved so successful. Noyce was not trying to solve an abstract problem. He was completing a specific task: determine a practical and profitable way to use the planar process. Motivation is often underplayed in discussions of invention—the unintended consequences of technological innovation are often of most interest to researchers—but in the case of the integrated circuit, motivation mattered.

  Whereas Kilby at Texas Instruments asked “how can I build an integrated circuit?” Noyce wondered “how can this planar process be used?” (Noyce: “I was trying to solve a production problem. I wasn’t trying to make an integrated circuit.”) Noyce was thus focused on production from the beginning; with this intellectual launch pad, it would have been difficult for him to consider seriously any device that could not have been mass produced. He also thought about selling the device from the moment he conceived of it; his notebook entry considers “cost per active element.” For his entire life, Noyce saw the integrated circuit as essentially a process breakthrough, not a scientific achievement. His children loved to tease him about his most famous invention—when would he get his Nobel Prize for it? His answer was always the same, always tinged with his disdain for abstract theory, and invariably delivered with a smile: “They don’t give Nobel Prizes for engineering or real work.”36

  Noyce filed his integrated circuit patent on July 30, 1959. The “principal objects” of the device, according to the patent, were “to provide improved device-and-lead structures for making electrical connections to the various semiconductor regions; to make unitary circuit structures more compact and more easily fabricated in small sizes than has heretofore been feasible; and to facilitate the inclusion of numerous semiconductor devices within a single body of material.” The patent application also included a figure that contained within it all the basic elements of the modern complex microchips. Noyce later said that the mother of this particular invention was not necessity, but laziness, that he conceived of the integrated circuit simply because he did not “want to go through all that work [of interconnecting components by hand].”37

  It is not unusual for an invention to appear in two different places almost simultaneously, as happened with Kilby and Noyce. Indeed, Noyce always maintained, “There is no doubt in my mind that if the invention hadn’t arisen at Fairchild it would have arisen elsewhere in the very near future. It was an idea whose time had come, where the technology had developed to the point where it was viable.” Noyce’s design was founded on existing, mainstream efforts in the industry—and that was part of its appeal. Relative to manufacturing a new element or growing a device atom by atom (as attempted in the defense-sponsored efforts) Noyce’s design was easy to build.38

  Noyce liked to say that the real impetus for the integrated circuit came from the realization not that it would be desirable to put all these devices on one chip, but that it was possible to do so. “Both of these are necessary. You’ve got to realize that it would be desirable to reach a given goal, and then you’ve got to have a method of getting to that goal before you can really … jump in with both feet and start dumping the effort into it.” In his plan to use the planar process to build integrated circuits, Noyce moved Fairchild from theoretical wouldn’t-it-be-nice musings to a practical we-can-do-this launch pad.39

  In 1965, Noyce said that he could “recall very vividly” that at the end of 1959 he “call[ed] a group of technical people together and [said], ‘Look this [integrated circuit] is possible [using the planar process]. Now, let’s explore every possible way that we could do it besides this way.’” He continued, “From there on out, we laid out a program to go ahead and do it. It was a very conscious decision at that point.” This may be the meeting that Moore recalls in connection with the announcement of the Texas Instruments circuit. Whether the “go ahead and do it” message was transmitted in this single formal meeting or instead through a series of informal conversations, the key point is that the message was sent and it was Noyce who sent it.40

  The man wh
o bore the brunt of moving the integrated circuit to the third stage, from “it’s possible” to “it’s finished,” was Jay Last. Noyce kick-started Last’s interest in July of 1959, when he wandered into the R&D lab and told Last that he thought Texas Instruments would make much ado about its integrated “solid circuits” at an important industry conference called Wescon, held every August. Noyce said that he wanted Fairchild to demonstrate some sort of integrated device at Wescon, too. It was far too early to try to build any sort of complex integrated circuit using the ideas Noyce had outlined in his patent. Instead, Noyce wanted a “show the flag” circuit, a bulwark against Texas Instrument’s claims to primacy in the field of integrated electronics.

  Last cobbled together a basic flip-flop circuit by putting four transistors on a ceramic plate, making resistors from pencil graphite, interconnecting everything with wires, and putting the plate into a transistor package about a half inch in diameter. This item demonstrated the fundamental concept behind integrated circuits—a complete circuit in a single package—but no one at Fairchild considered it a potential product or even particularly interesting. It was a defensive marketing measure suggested by Noyce and developed and built by Last.41

  Immediately after Wescon, Last began work in earnest on what he called “microcircuits.” He hired Lionel Kattner, who had worked at Texas Instruments, and Jim Nall, who had just won a Defense Department award for fabricating microminiature circuits. Two researchers from other parts of Fairchild—Bob Norman and Isy Haas—also joined Last’s group. The team mounted a ferocious campaign to try to build microcircuits. Noyce’s patent did not provide much guidance. His patent said that it ought to be possible to build integrated circuits using isolation techniques and the planar process. It did not, however, say how to do it. That was what Last’s group needed to figure out. Yes, it ought to be possible to isolate devices with extra junctions, but what did that mean? Did you do it by diffusion, and if so, for how long, and using which chemicals? Did you do it by etching an isolating groove through the back side of the wafer through to the oxide on the front and fill the groove with some sort of inert isolating material? (Last patented this idea.) Did you need to build transistors optimized for use on integrated circuits, or could you use the transistors developed for other purposes? What type of metal would you use to interconnect the components on the chip?