by T. R. Reid
Unable to get out of the rut, Noyce turned his attention to another technical dilemma. Double-diffusion transistors—the tiny three-layer chips of N-P-N silicon—were highly susceptible to contamination. A piece of dust, a stray electric charge, a minute whiff of contaminating gas would break down the P-N junctions and impair transistor action. One day in 1958, Jean Hoerni came to Noyce with a theoretical solution: He would place a layer of silicon oxide on top of the N-P-N chip, like icing atop the three-layer cake. The oxide would hold fast to the silicon and protect it from contaminants. “It’s building a transistor inside a cocoon of silicon dioxide,” Noyce explained, “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.”
The people at Fairchild recognized Hoerni’s idea—it was called the planar process because it left a flat plane of oxide atop the silicon—as an important advance in transistor technology. Noyce quickly called in the firm’s patent lawyer, John Ralls, to put together a patent application. Ralls, sensing that this planar idea might have other applications in electronics, wanted to write the application in the broadest language. He told Noyce that Fairchild should make the application as expansive as it possibly could. Every time they talked about the planar patent, Ralls would pose a challenge: “What else can you do with this idea?”
Looking back years later, Noyce could see clearly that it was the lawyer’s question that pushed him out of his mental rut and provoked the leap of insight that became the monolithic idea. What else? What else could you do? In the first weeks of 1959, Noyce was thinking hard about that question, scratching pictures in his notebook, talking things over hour after hour with his sedate, cautious friend Gordon Moore.
As Noyce looked over Hoerni’s planar idea, he realized that it had another useful property. It was quite difficult in those days to make precise electrical connections to the separate regions of an N-P-N transistor, because wires were relatively large compared to the tiny regions of the chip. Hoerni’s oxide icing, spread atop the three-layer silicon cake, helped solve this problem. Connecting wires could be poked down through the icing like candles on a cake, and they could be inserted at the exact spot on the chip where the connection was needed. The oxide would keep them firmly in place. “Remember, what I’m trying to do is make this [transistor] extremely small,” Noyce explained. “Well, I can’t attach a wire to that, because it’s too small. But now, with the planar coating, I can attach a big old wire—big old wire, this is, you know, a quarter of a human hair—running it on top of the oxide.”
And that realization led to a new idea, something that was even better: wires wouldn’t be needed at all. Noyce now saw that tiny lines of copper or some other metal could be printed on top of the oxide layer rather than poked down through. The advantage of this approach was that printing is a faster industrial process than aligning and inserting tiny wires. With the “wires” printed on top of the oxide coating, all the transistor’s interconnections, all the wires, could be made in one fell swoop in a single manufacturing process.
But wait a minute, Noyce thought. We can carry this even further. If you could connect the separate regions of a single transistor with these printed metal lines, then you could put two separate transistors on a single piece of silicon and connect them with the printed lines. And why stop with transistors? If you could put two transistors on a single chip of silicon, couldn’t you build some other circuit components on the same chip? How about a resistor built into the same chip? How about a capacitor in there? Couldn’t you, in fact, build a complete circuit, an integrated circuit, all on a single chip of silicon? Wouldn’t that overcome the tyranny of numbers?
“I don’t remember any time when a light bulb went off and the whole thing was there,” Noyce said. “It was more like, every day, you would say, Well, if I could do this, then maybe I could do that, and that would let me do this, and eventually you had the concept.”
One day, Noyce walked into Moore’s office and showed him, on the blackboard, that two transistors in a single silicon block could be connected by printed copper lines on the oxide layer. A few days later, he was back at the blackboard, showing Moore how he could use a channel of undoped silicon in the same block as a resistor. A few days later, he was drawing a silicon capacitor on the blackboard. It was all completely new, but Moore raised no serious objections.
Noyce’s intellectual journey to the monolithic idea began at a different starting point from Jack Kilby’s, but reached the same destination. Kilby had first hit on the (crazy) idea of building all the circuit elements in a single semiconductor block; as an addendum to that notion, he realized that the various elements could be connected by “wires” printed onto the same block. Noyce, on the other hand, arrived first at the idea of printing the wires on the semiconductor chip, and went from that level to the idea of putting all the circuit elements on a chip. Both routes led to the monolithic integrated circuit.
On January 23, 1959, “all the bits and pieces came together,” and Noyce filled four pages of his lab notebook with a remarkably complete description of an integrated circuit. “In many applications,” he wrote, “it would be desirable to make multiple devices on a single piece of silicon, in order to be able to make interconnections between devices as part of the manufacturing process, and thus reduce size, weight, etc. as well as cost per active element.” Noyce went on to explain how resistors and capacitors could be fabricated on a silicon chip, and how the whole monolithic circuit could be connected by metal contacts printed right onto the chip. He also set forth a rough sketch of a computer circuit—a basic “adder” circuit that would add two numbers—realized in integrated form. Six months after Jack Kilby had arrived at the monolithic idea, Bob Noyce sailed into the same port. Kilby’s journey had been slightly quicker, but the use of the planar process made the Noyce route somewhat more viable.
News travels quickly in the electronics industry. By the spring of 1959, rumors about a major new development at Texas Instruments had reached the people at Fairchild. Nobody knew exactly what TI had done, but it was not impossible to guess which problem this breakthrough was designed to solve. Noyce again called in John Ralls and asked the lawyer to prepare a patent application for a new idea—“a unitary circuit structure . . . to facilitate the inclusion of numerous semiconductor devices within a single body of material.” This time, Ralls decided to write a detailed, precise patent application, a document that could serve as a shield to protect Fairchild against any possible legal action by Texas Instruments. This strategic decision would become the decisive factor in a bitter ten-year legal battle fought all the way to the United States Supreme Court.
5
KILBY V. NOYCE
The terrifying rumor that raced through the semiconductor lab at Texas Instruments on the morning of January 28, 1959, turned out, eventually, to be wrong on almost every count. But like many false alarms, it had the salutary effect of scaring people into action. More than four months had passed since Jack Kilby had successfully demonstrated his prototype integrated circuit, but since then further development of the concept had been almost nil. Kilby’s superiors had been hoping to introduce this fantastic new product in March, but as of the end of January, there really wasn’t any product. The only integrated circuits in existence were the crude models Kilby had built by hand for his demonstration; nobody had figured out yet how to turn out a production version. Even the lawyers were behind schedule; they had failed to take the most basic steps to protect Texas Instruments’ right to the new invention.
That’s why the rumor was so frightening. At a technical meeting, somebody from TI thought he heard somebody else saying that he had been told that another company had come up with an integrated semiconductor circuit that would overcome the tyranny of numbers. There was, in fact, a germ of truth in this report; just five days earlier, Bob Noyce, in his office a
t Fairchild, had scratched his first sketchy concept of the monolithic idea in his notebook. But the rumor that reached Dallas had nothing to do with Noyce. The word at TI was that somebody at RCA had come up with an integrated circuit, and that—even worse—RCA was soon going to file for a patent. When this unsettling news reached Samuel M. Mims, TI’s senior staff lawyer, he didn’t hesitate a second before putting through an emergency call to Mo Mosher.
Ellsworth H. Mosher, name partner in the Washington, D.C., patent law firm of Stevens Davis Miller & Mosher (cable address: INVENTION) and an elder statesman of the patent bar, knew immediately that he had to act fast. He dispatched a junior lawyer to Dallas and told Mims to sit down with Kilby and find out precisely what the inventor thought his monolithic idea would be good for. It normally took two or three months to put together all the paperwork, prose, and pictures required for a patent application; in this case, though, Mosher promised to deliver a completed application to the Patent Office within a week.
Mosher’s advice—that Texas Instruments had better apply for a patent, and fast—was not quite so obvious as it might appear. One of the most important rules that patent lawyers try to get across to their clients is that, in some cases, it is better not to apply for a patent at all. For an inventor a patent is a sort of Faustian bargain.
The patent expressly guarantees the inventor “the right to exclude others from making, using, or selling” the idea for the twenty-year life of the patent. The patent holder can, if he chooses, issue licenses to others to make, use, or sell the idea. The license fees can bring in large sums of money. If anybody tries to market the patented product without obtaining a license, the inventor can go into federal court to get an injunction and money damages. Not a bad deal at all for the inventor. In exchange for those benefits, though, the patent holder has to reveal all the secrets of his success. The patent law says that an inventor must provide “a written description of the invention, and of the manner and process of making and using it, in . . . full, clear, concise and exact terms.” The inventor and his company might have expended a dozen years and a hundred million dollars perfecting the idea; once a patent is granted, anybody in the world can acquire the plans— full, clear, concise, and exact—from the Patent Office for $3.
If, for example, John S. Pemberton had applied for a patent for the formula he whipped up in his backyard in Atlanta one day in the mid-1880s, the product that he invented—a soft drink that he named Coca-Cola—would have entered the public domain in 1903, when the patent expired. Anybody in the world would have been free from that day forward to brew and sell the drink without paying a penny to the Coca-Cola Company. But Pemberton kept his formula unpatented, and thus secret. Even without a patent, Coca-Cola has been able to defend its formula under a body of law known as trade secret protection, which makes it illegal to copy deliberately somebody else’s commercial idea.
From the inventor’s viewpoint, the flaw with the trade secret laws is that they apply only to purposeful stealing of an idea. They do not prevent anybody from marketing a product that he has invented on his own, even if an earlier inventor has been selling the same product for years. Lacking a patent, Coca-Cola would have no recourse against a company selling exactly the same drink if the second firm could prove in court that its chemists had been messing around with sugar, flavorings, and cola nuts and just happened to hit on the precise formula that Coca-Cola uses. The holder of a patent, in contrast, can go to court to stop any competitor from selling the same product, even if the competitor developed the product completely on his own. The strategic decision facing every inventor, then, is whether he wants twenty years of the stronger protection provided by a patent, or permanent protection under the trade secret laws against only those who deliberately steal the idea.
The choice has to be made, because an inventor can take either patent protection or trade secret protection, but not both. This principle was established once and for all in the landmark case of Kellogg v. Nabisco— the great shredded wheat decision. The familiar shredded wheat biscuit was invented in 1895 by a Colorado baker named Henry Perky, who promptly—and foolishly, as it turned out—took out a patent (No. 548,086) on his new breakfast cereal. The National Biscuit Company subsequently bought the rights to Perky’s invention. Despite extensive advertising, shredded wheat sales never took off until the 1920s—well after the patent had expired. Somebody in Battle Creek realized that the formula was free for the taking, and grocers everywhere began carrying a new product—Kellogg’s Shredded Wheat. Nabisco went to court, claiming, under the trade secret laws, that Kellogg had deliberately copied its product. The shredded wheat litigation wound its way for years through the legal system, and in 1938 it finally reached the United States Supreme Court. In a decision by Louis Brandeis, the Court sided with Kellogg all the way. Once the patent had been issued, Brandeis wrote, Nabisco lost its right to claim that
shredded wheat was its private secret. The basic goal of the patent system, after all, was to encourage public disclosure of technological advances.
This goal was so important to the founding fathers that the Patent Office (the name comes from the Latin verb patere, “to open”) was one of the earliest federal agencies created by the First Congress. Among the small group of men who constituted the U.S. government then, jobs were assigned largely on the basis of personal interest. The secretary of state, Thomas Jefferson, was an inventor, so Congress gave the Patent Office to the State Department. In the evening, after a long day of diplomacy, Jefferson would review patent applications. Among the patents he personally granted was one to Eli Whitney in 1794 for the cotton gin.
Inevitably, bureaucracy reared its head. In 1804 the Patent Office hired an employee of its own, and that proved to be the first step along a steep slope; today the Patent and Trademark Office (now a wing of the Commerce Department) has about 7,000 employees scattered around several huge buildings in a suburban mall in Arlington, Virginia.
The government receives about 270,000 patent applications each year and grants about 170,000 patents for inventions, plus more for plants and designs. The office is, among other things, one of the world’s busiest publishing houses; it keeps every one of the 6.2 million patents issued since Jefferson’s time in print.
Like most traditional institutions, the Patent Office has developed parlance and procedures all its own. There are tens of thousands of pages of statutes, regulations, guidelines, and legal opinions governing the issuance of patents. Among much else, the regulations go on for fifteen long paragraphs describing the kind of paper and ink an applicant should use for drawings of his invention: “The sheets may be provided with two ¼-inch (6.4 mm) diameter holes having their centerlines spaced inch (17.5 mm) below the top edge and 2¾ inches (7.0 cm) apart, said holes being equally spaced from the respective side edges.”
At the core, though, the basic rules governing what kind of inventions will be granted a patent are straightforward. The invention has to be “new.” It has to be “useful,” a term the courts have interpreted to mean that the gadget has to work. The Patent Office receives a few applications each year from people who have invented perpetual motion machines; it rejects them all on the ground that they can’t do what they’re supposed to and so aren’t useful. The inventions also have to be desirable; the government refuses to grant patents for nuclear weapons, no matter how new or how “useful” they might be.
When Mo Mosher heard about the important new invention at Texas Instruments, he knew immediately that TI would need a patent. Unlike soft drinks and breakfast cereals, electronic gear rarely has a market life longer than the term of a patent, so it is almost always a wise move for an electronics inventor to seek a patent. Mosher knew, too, that none of the statutory requirements that govern patentability would pose a problem. Kilby’s monolithic circuit was clearly something completely new, and since it promised a solution to the most important problem facing the industry, it was eminently useful and desirable. And relying on trade secret prote
ction in this case was unlikely to be a smart bet; so many other people around the world were looking for a solution to the tyranny of numbers that somebody was likely to come up with an integrated circuit without any access to TI’s internal secrets. A patent meant TI could own rights to the chip even if somebody else independently hit on the monolithic idea.
Before he could start writing Kilby’s application, though, Mosher had to resolve a fundamental tactical question. Anyone who applies for a patent has to decide whether he needs it for offensive or for defensive purposes—whether, to use lawyers’ favorite metaphor, he wants his patent to be a sword or a shield. The decision usually turns on the novelty of the invention. If somebody has a genuinely revolutionary idea, a breakthrough that his competitors are almost sure to copy, his lawyers will write a patent application they can use as a sword; they will describe the invention in such broad and encompassing terms that they can take it into court for an injunction against any competitor who tries to sell a product that is even remotely related. In contrast, an inventor whose idea is basically an extension of or an improvement on an earlier idea needs a patent application that will work as a shield—a defense against legal action by the sword wielders. Such a defensive patent is usually written in much narrower terms, emphasizing a specific improvement or a particular application of the idea that is not covered clearly in earlier patents.
Probably the most famous sword in the history of the patent system was the sweeping application filed on February 14, 1876, by a teacher and part-time inventor named Alexander Graham Bell. That first telephone patent (No. 174,465) was so broad and inclusive that it became the cornerstone—after Bell and his partners had fought some 600 lawsuits against scores of competitors—of the largest corporate family in the world. In the nature of things, though, few inventions are so completely new that they don’t build on something from the past. The majority of patent applications, therefore, are written as shields—as improvements on some earlier invention. Some of the most important patents in American history fall into this category, including No. 586,193, “New and Useful Improvements in Transmitting Electrical Impulses,” granted to Guglielmo Marconi in 1898; No. 621,195, “Improvements in and Relating to Navigable Balloons,” granted to Ferdinand Zeppelin in 1899; No. 686,046, “New and Useful Improvements in Motor Carriages,” granted to Henry Ford in 1901; and No. 821,393, “New and Useful Improvements in Flying Machines,” granted to Orville and Wilbur Wright in 1906.