Shuji Nakamura was angry. It was July 1987 and during his eight years as a salaryman at Nichia, Shuji had done everything he had been told to do. Working alone he had developed compound semiconductors, epitaxial wafers, and red and infrared light emitting diodes. But to no avail, they had all been commercial failures. At this point, his standing within the company could scarcely have been lower. He had been passed over for raises and promotions. His co-workers blamed him for wasting the company's money on his research. And, since in order to protect its know-how the company did not permit employees to publish papers, Nakamura had nothing to show for his efforts professionally, either.
It was all so unfair, he reflected bitterly. After all, it wasn't as if he had been the one who had decided to develop these uncompetitive products. The company's salesmen had him assigned to develop them. Nonetheless, he was the one who had to shoulder the blame for their failure in the marketplace. The more he brooded about it, the angrier he became. And he felt guilty toward his friends at Nichia. Simple country folk that they were, his workmates had expected great things from him, new products that would increase the company's sales. He had let them down.
Now, feeling cornered and ill at ease, Nakamura brooded about what to do next. Finally, he decided that the only way out was to quit Nichia. Every morning he would go to work thinking that, that day, he would hand in his resignation. Two things made him hesitate. One was the very practical consideration that he had a family with three young daughters to support. The other was his desire to pay back what he felt he owed the company, by developing a successful product. Only then would he be able to look his colleagues in the eye. Of course it was very unlikely that he would succeed in such an endeavor. But by this stage, Nakamura was a desperate man.
Thus far, he had not done any research that he himself really wanted to do. He had merely done what he had been told. “Since I'm going to quit anyway,” he said to himself, “from now on, I should take risks and do exactly what I want to. If I fail, then that will be a good excuse to quit.” His goal therefore should be as ambitious as possible, something that was almost impossible to achieve. A bright blue LED would be ideal.
He had been mulling the idea in his mind for a long time. Through his reading of the technical literature, Nakamura was well aware that no one had managed to develop a bright blue LED. Its absence was an affront to any self-respecting LED researcher, and Shuji was no exception in this regard. He had mentioned his ambition to his immediate boss several times, only to be flatly rejected. “Are you stupid?” his boss asked. “There's no way you could develop something like that. The big companies have all tried and they haven't been able to—what makes you think you could do it here, at a small company, with no budget and without proper equipment?”
Now, however, brooding about the matter, Nakamura had come to the conclusion that he should no longer listen to what his boss said. Quite the reverse, in fact: he would do the opposite of what people told him. What the company opposed was something worth doing. He also realized that there was no point telling his immediate boss that he wanted to aim at developing a bright blue LED. The latter would only laugh at his pretension. More to the point, his boss did not have the power to make decisions on such an important issue. Big decisions at Nichia were made by top management. Mostly that meant the president, Nobuo Ogawa.
Though at seventy-six he was no longer involved in the day-to-day running of the company, when it came to crucial decisions, old man Ogawa still called the shots. Nakamura knew that if he was to have any chance of chasing his dream, he would have to gain the president's approval. It was not just a question of money. Once he had secured the president's permission, no one would dare to stand in his way. Shuji was aware from what people had told him that, because of his ability to make things, Ogawa was favorably disposed toward him. Nonetheless, Nakamura was taking a big risk: he determined that if the old man turned him down, then he would have to quit. By that stage, however, he felt that he had nothing to lose by being audacious. If they wanted to fire him, let them; he didn't care any longer. Nakamura had reached the end of his tether. He was not afraid of anything anymore.
So, summoning up all his courage, at ten o'clock one morning Shuji marched into the president's office to make his case. He didn't go into detail—the old man would probably not understand it if he did—he just came out and said: “I want to develop a bright blue LED.” To his amazement, Ogawa simply replied: “Is that what you want to do? Well, in that case—go ahead.” Nakamura could hardly believe his ears. Concerned in case the president had misheard or misunderstood him, he sought confirmation: “Really? Is it OK if I try to develop a bright blue LED?” Then Ogawa repeated, loud and clear this time: “If that's what you want to do, then go ahead.”
Nakamura had joined Nichia in April 1979, along with six other new employees. They were all locals, mostly from Anan; he was the only one not from Tokushima Prefecture. Shuji was assigned to the development section. To his surprise, this comprised just three staff: the section chief, one other researcher, and himself. Soon it would consist of just two, Nakamura and the section chief. Others alerted him to the fact that, at this company, the development section was not exactly the fastest of career tracks.
Nichia at the time had fewer than two hundred employees. The company's main business was the production of phosphors—materials that glow when hit by electrons. Phosphors are used in color televisions, fluorescent lamps, and x-ray intensifiers. All three were mature markets. If Nichia was to grow, then it needed to find new products. Coming up with such products was the mission of the development section, the only part of the company not involved with phosphor production.
Nakamura's first job was to refine high-purity gallium metal. This was a promising start. Gallium is a soft, silvery-white metal that occurs in trace amounts in aluminum ores. When compounded or alloyed with other elements, gallium is the sine qua non of almost all LEDs. Nakamura had worried that, at a company whose business was producing chemical powders, he would not be able to exploit his skill in electronics. Now, because he was to work on gallium, a semiconductor material, he felt somewhat reassured. Nichia had embarked on the refining of gallium metal some years before Nakamura joined. But sales were disappointing. So much so in fact that the company's managing director, Nobuo Ogawa's son-in-law Eiji, was considering disbanding the development section altogether.
After Nakamura had been at Nichia a couple of months, the company issued instructions that the development section was to move on to a new project. Nichia's salesmen had picked up a promising hint during a visit to a big customer in Osaka. Refining gallium might be a dead end, but if they could go one step further and produce gallium phosphide, a compound semiconductor material used to make red and yellow-green LEDs, they could sell ingots of that instead.
Nakamura was relieved that the development section was to be spared. He would have been less so if he had understood what was really going on. Namely, that the customer was just looking to outsource the production of gallium phosphide to a cheaper supplier. Nakamura had worked on compound semiconductors during his master's course. It was natural that he should be assigned to the project. He began research, throwing himself into the work with great determination. Knowing next to nothing about growing LED materials—which comes under the heading of chemical, rather than electrical engineering—he had to start from scratch. Since he did not get on with the section chief, he had to work mostly on his own.
Shuji spent six months reading the literature of LEDs, immersing himself in the theory, a task that he greatly enjoyed. But when the time came to do experiments, Nakamura discovered that, just like at Tokushima University, there was no budget for equipment. He had to make do by scavenging obsolete bits and pieces that were lying around the place, fixing broken parts, cobbling together everything he needed by hand.
To make gallium phosphide, you place your starting materials—gallium metal (which, like mercury, is liquid at room temperature) and phosphorus—at either end of a sealed, evacuated quartz tube. You heat them up above their melting point, causing them to react with each other. After the reaction is complete, you gradually reduce the temperature. A D-shaped ingot of single-crystal gallium phosphide forms. Finally, you cut off the end of the tube with a diamond saw and remove the ingot.
To build his equipment, Nakamura scrounged some heatproof bricks, cables, a vacuum pump, and an old electric furnace that had formerly been used to make phosphors. The quartz tubes he had to order. Quartz is necessary because it can withstand high temperatures (glass would just melt). But the tubes came open-ended. In order to seal them so that they could be evacuated, he had to learn how to weld quartz.
The production process involves heating up the tube until it reaches about 1,100 degrees Celsius at the gallium end and about 600 degrees Celsius at the phosphorus end. At such temperatures, the tube glows bright red. Controlling the temperature is crucial, because if it gets too high, the phosphorus vapor expands, causing the quartz to crack. This lets in oxygen, which reacts with the phosphorus, causing an explosion. Such explosions became a feature of Shuji's life at Nichia. Following them, the whole room where he was working would instantaneously fill with dense white smoke. The phosphorus, which had ignited, would fly everywhere, all over the walls and the floor, along with shards of broken quartz tube. Nakamura would run around desperately pouring water over the burning phosphorus, trying to douse the flames, which sometimes reached as high as the ceiling.
The blasts happened several times a month, typically around five o'clock in the evening. The shock wave from the explosion would hit his fellow workers as they were heading for their cars in the parking lot about a hundred yards away. The first few times it happened, they dashed into his lab shouting, “Nakamura—are you still alive?” By the fifth or sixth time, however, they had become so used to the bangs that they no longer came to check on his well-being. They would just say to each other, “Sounds like he's done it again.”
After the initial explosions, Shuji constructed a primitive protective aluminum shield about three feet square. He erected it in front of the electric furnace in order to contain the blast. Years later, Nakamura would shake his head in disbelief as he recalled how reckless he was and how dangerous his circumstances had been. And he would ask himself why Nichia had been so negligent about the safety of its employees.
The explosions were a setback in another way, too. Whenever equipment broke, Nakamura would have to rebuild it. He became expert at welding quartz tubes so that they could be reused after the ingots had been cut out. Some days he would spend from eight in the morning until three in the afternoon just welding tubes. It was hot and sweaty work, using an oxyacetylene burner. It was also tense, because any crack in the weld would cause an explosion. As a graduate student, Shuji had felt like a worker in a sheet metal factory. Now, it seemed, he had become a welder. He felt his life as a researcher had ended before it hardly even had begun.
All the time, he had to endure the company putting pressure on him. “Haven't you made anything yet?” his boss would say. “Commercialize a product quickly and contribute to sales.” Eventually, after about three years’ effort, Nakamura succeeded in developing commercial-grade gallium phosphide. More satisfying than producing the material, however, was how he felt when the company's salesmen reported back telling him that they had made a sale. It gave him pleasure to think that he had finally managed to contribute something to the company's bottom line. But only a little: the market for gallium phosphide was already crowded with suppliers. As a late entrant, Nichia was only able to win a sliver of the pie. Monthly sales of gallium phosphide, a relatively low-cost material, never amounted to much more than a few tens of thousands of dollars.
Nakamura's next assignment was to produce gallium arsenide. Like gallium phosphide, gallium arsenide is used to make LEDs, typically infrared ones such as those found in television remote controls. But GaAs also has other applications, such as the semiconductor lasers used in fiber-optic communications. Thus, the potential market for the material was larger.
The manufacturing method was the same as for gallium phosphide. Once again, during the experimental stage, there were many explosions. Happily, unlike phosphorus, arsenic is not inflammable, so it was easier to work with. Unhappily, the material is highly poisonous, releasing lethal arsenic oxide gas every time the furnace blew up. Nakamura would wait until the smoke cleared before going to clear up the mess. He had to wear a homemade “space suit” and breathe through a respirator. By some miracle he was never adversely affected by having to work in such a toxic environment. And all the time he was learning the “black art” of making compound semiconductor materials, the kind of detailed know-how that scientific papers never mention. For example, how to control the temperature of an electric furnace with exquisite precision. His masterless apprenticeship would be invaluable for his later work in developing the bright blue LED.
By 1985 Shuji was producing gallium arsenide in bulk quantities. Looking back, he was inclined to think that this was quite an achievement. A big company would have assigned a team of people to such a project. It is not unusual for development to take five years or more. And yet he had done it in just three years, all by himself. But when it came to selling the product, the market's response was the same. There were plenty of existing suppliers, so why would customers buy from an untried Johnny-come-lately like Nichia?
The next bright idea the salesmen brought back was, instead of making the starting materials for LEDs, why not go the whole hog and make the wafers and light emitting devices themselves? To fabricate a simple LED required mastering a technique known as liquid phase epitaxy, in which successive layers of crystalline material, each with different electrical properties but identical crystal lattices, are deposited on the wafer, one after another. Chop the wafer up into chips, stick electrodes on either end of the chip, and bingo! You've got an LED.
To learn how to do liquid phase epitaxy, Nakamura had to resort once again to the technical literature. Having pored over journal articles and published patents, he performed countless experiments, varying the temperature at which the layers were grown and the time it took to grow them. Small differences in thickness, he discovered, could make a big difference in brightness and lifetime. As usual, there was pressure from the company to produce a saleable product quickly. As usual, there was no budget available to buy equipment. Eventually, by repeated trial and error and making improvements based on the results, he managed to make some prototype LEDs.
Samples of this device were delivered to a client for evaluation. The client responded that the device wasn't very bright and degraded over time. Not having measuring equipment of his own meant that Nakamura was dependent on such evaluations from clients. He had to wait several months to get data back before he could start making improvements. Shuji felt strongly that if the company was going to enter the LED device business, then he should be able to conduct his own evaluations. He tried to argue the case with his boss, but was told, No budget, so not possible.
Previously Shuji would have accepted this answer and given up. By now, however, he had been with the company for more than six years. He was in a position where he could make his voice heard. Also, he had come to realize that Nichia was a company that was run on the say-so of its president and founder, Nobuo Ogawa. He decided that it would be more effective to approach the top man. Shuji went to Ogawa and asked him for what he needed. To his surprise, the old man agreed immediately, giving his permission for Nakamura to buy the equipment to measure luminous intensity and device lifetime. This encounter set a precedent, one upon which Nakamura would draw three years later, when he went to see Ogawa again, this time with a much bolder request.
By 1987 Nakamura had developed commercial-grade infrared and red LEDs. He had done everything from R&D, to manufacturing, to quality control. Now, the company tasked him to add yet another string to his bow: sales. Since Nichia did not have any salesmen who understood semiconductors, it fell to Nakamura to go out and visit customers, large consumer electronics manufacturers that used LEDs in their products, and to try and sell them the devices he had made. Often he would visit the factories of Matsushita and Toshiba, Nichia's biggest customers. There, he would meet researchers with PhDs whose speciality was semiconductor technology. The experts would raise their eyebrows at the samples he brought. “How come someone like you is able to make semiconductors all by yourself in a hick town like Anan?” they asked him. “Especially without introducing technology from somewhere else?” Buoyed by such compliments, Nakamura began to acquire confidence in his abilities as an LED researcher.
Potential customers would also come down to visit Tokushima. Nakamura and Nichia's salesmen would wine and dine them in local clubs and karaoke bars. Shuji spent so many nights entertaining that he began to worry he might be permanently reassigned to the company's sales department. Sociable and friendly by nature, Shuji was popular among the salespeople. He would probably have made an excellent salesman. But no matter how persuasive his pitch, customers were still reluctant to buy LEDs from a company that had little or no track record in the semiconductor business. There were plenty of other, much better-established suppliers in the market. Inevitably, customers asked for reductions in price, one company demanding a discount of 50 percent. It is sometimes said that the Japanese economic miracle was built on the sacrifice of small subcontractors.
Nakamura could never manage to raise sales of LEDs beyond a few tens of thousands of dollars a month. Not much, considering the expensive equipment for measuring, testing, and production that he had persuaded the company to buy him. Or the salaries for the new employees that joined the company each time he had commercialized a new product to do the manufacturing. Even though, as Nakamura realized, because he had done all the product development himself and custom-built much of the equipment, the actual development costs were, by industry standards, minuscule.
Nakamura had made many friends among Nichia's employees. When work finished at five, they would often ask him to make up the numbers for a game of baseball or softball. Afterward, they would all drop by a local bar. After a few drinks had loosened their tongues, his workmates would say to him, “Nakamura-san, develop some good products and make the company grow. We're just country people, farming is all we know, but we have faith in you.” And feeling small, knowing that he had yet to make anything that had an impact on the company's bottom line, Shuji would hang his head. Others, especially older employees, were critical. On an overnight business trip to Tokyo, after a few drinks, Nichia's sales manager started haranguing Shuji. “The development section is just a name: What have you been doing for the past five years? We can't sell what you've developed—you're just wasting the company's money.” Feeling miserable and unable to come up with a rebuttal to these charges, Nakamura just kept apologizing until the sales manager fell into a drunken sleep.
His situation was truly depressing. The fact that he had made sophisticated semiconductor products essentially unaided should have been appreciated. In fact, he had been repeatedly passed over for promotion. In part, this was because he had never been involved in manufacturing, the section that typically garners most praise, because that is where sales derive from.
Whenever it came time to commercialize a product, the company would employ someone to take charge of the manufacturing. Shuji would teach him what to do, then he would be assigned his next project. This latecomer, who had not contributed anything to product development, would be promoted ahead of him to become his boss. He felt as if the credit had been stolen from him. And he remained on the lowest rung of the salary scale. The only way a corporate researcher can contribute directly to the bottom line is through patent royalties. But fearful of losing precious trade secrets, Nichia did not permit patent applications on principle. Thus Nakamura's apparent sales were zero.
Elsewhere, too, he felt invisible. Few people in the industry even knew that Nichia made semiconductor materials and devices. Sometimes he would call semiconductor equipment makers and ask them to send catalogs. Perhaps feeling that an inquiry from a remote rural location like Anan was unlikely to lead to a sale, the suppliers would not bother to send them.
In ten years Nakamura had not published a single scientific paper, because of Nichia's policy of keeping its technical know-how a tightly guarded secret. Nor had he ever been to a conference. From a professional point of view, he had no achievements, he did not even exist. He had been a loyal and obedient employee, and where had it got him? Now, in early 1988, having won the president's consent to go ahead with his audacious plan to develop a bright blue LED, Shuji decided it was about time he made his presence felt.
Not for the last time in this story, Shuji's timing was impeccable. During the 1980s, riding the boom in personal computers, which meant increased demand for monitors, hence phosphors, Nichia's sales had been rising steadily. The company was making good money. Nonetheless, president Ogawa was taken aback when Nakamura returned to see him to explain that he would need around three hundred million yen (worth in 1988 about $2.4 million). This was equivalent to 2 percent of the company's sales that year. Compared to previous research costs, it was an unbelievably large amount. “That much? I see,” Ogawa said. “Well, let me think about it.” Nakamura went back several times, explaining about the bright blue LED, why it would be a big market, and why it would cost so much to develop. Eventually, the old man gave Nakamura his blessing.
In a 1997 book he coauthored with Nakamura, Gerhard Fasol put the figures into perspective. “It is rare for a large company to spend [$2.4 million] within essentially one year on a single blue-sky type research project of a single researcher,” Fasol pointed out. “It is even rarer that [2 percent] of annual sales would be spent on a single blue-sky project of initially unknown outcome, such as Nichia [did].” Large R&D-based companies like IBM spend an average of around $250,000 to $400,000 per researcher. Their approach is usually to provide small budgets until commercial success is within reach, or at least until the risk is easier to assess.
In fact, Nakamura's apparently quixotic quest to develop a bright blue LED would end up costing Nichia even more than originally budgeted. By 1990 the company had spent five hundred million yen (about $4 million) funding Shuji's work.
Nakamura had calculated how much money he would require based on the expertise he needed to acquire and the equipment he needed to buy. Two-thirds of the money would go to equipment, together with the laboratory and the clean room facilities to house it. Of the remaining third, one large expense item was time to be spent in the United States, mastering the crystal growth technology that Nakamura reckoned he would need in order to make bright blue LEDs. This was called metal organic chemical vapor deposition, MOCVD for short.
The conventional method of making LEDs, liquid phase epitaxy, could not grow films of high enough quality and sufficient thinness. There were two more modern alternatives. Both were capable of growing films of material just a few atoms thick. One, molecular beam epitaxy, works via vacuum evaporation. MBE is especially popular with academic scientists. It can produce small quantities of material, enough for researchers to extract data based on which they can write and publish papers. But MBE requires an ultrahigh vacuum, has very slow growth rates, and is difficult to scale up. In the opinion of most people, Shuji included, the method is not suitable for mass production.
That left MOCVD, which does not need a high vacuum and can be applied to the factory floor. The choice was thus, as they say in Silicon Valley, a no-brainer. Shuji selected MOCVD without hesitation. But he had little idea of how MOCVD was done. By a stroke of good fortune, it just so happened that one of Japan's leading experts on the technique was an old acquaintance of his from Tokushima University. Though Shiro Sakai had been two years Shuji's senior, they had worked together in the same laboratory, and Shuji knew him well. In the intervening years, Sakai, now a professor at Tokushima, had become well known for his expertise in MOCVD. Now he was on sabbatical at the University of Florida. During the summer holiday of 1987, he returned to Japan. Nakamura went to see Sakai to ask his advice on how to learn MOCVD.
When Sakai returned to Japan for a week at the end of 1987, Nakamura invited him to visit Nichia. There, the professor explained to Ogawa the significance of MOCVD as a crucial tool for the production of state-of-the-art LEDs. At this meeting, blue LEDs were not mentioned. Sakai recommended that Nichia should send Nakamura for a year to the engineering school at the University of Florida, where he was currently on sabbatical. A deal was arranged: Nakamura would learn MOCVD under Sakai's tutelage. In return, Nichia would donate around $100,000 to fund Sakai's research.
To make a chip the size of a grain of sand takes a mighty big box. A typical MOCVD system is almost as big as a shipping container and costs well over a million dollars. Seen from the outside, MOCVD equipment looks rather dull, like a row of office cabinets. Peek behind the bland-looking doors, however, and you will discover a bewildering assemblage of tanks, pumps, and valves connected by what appears to be several miles of thin, stainless steel pipe. At one end of the cabinets is a rack containing a computer that runs the recipes for growing LEDs. These are programs that, with exquisite precision, control the pressure and flow of gases, while monitoring the temperature and the rate at which the thin films of crystal grow.
The heart of an MOCVD system, visible through a little window in one of the cabinet doors, is its reactor chamber. This is a cylinder about the size of a cookie jar, made of quartz in some systems, of metal in others. It may be positioned either horizontally or perpendicularly. The chamber is remarkably small in comparison to the whole. It occupies perhaps 3 or 4 percent of the total space. Inside the jar is a graphite chuck mounted on a little pedestal. Here sits the wafer on which the thin films are grown. The chuck is connected, via a thermocouple that monitors the growth temperature, to a heater. To grow gallium nitride, the wafer is heated to between 1,000 and 1,200 degrees Celsius, at which point it glows bright golden-orange. In the case of quartz chambers, the heat comes from copper coils wound round the jar. An exhaust system, typically a vacuum pump, completes the process. It sucks the unused gases out of the reactor chamber, flushing them away to a scrubber for disposal.
For more than twenty years LEDs were grown by one of two methods, liquid phase epitaxy (LPE) or vapor phase epitaxy (VPE). Epitaxy simply means stacking crystal layer upon crystal layer with exactly the same orientation, like piles of egg trays. But when it came to growing high-quality thin films and quantum wells, which require abrupt atomic-level transitions from one layer to the next, both processes were too crude. For example, an LPE system consists of a quartz tube in which are lined up little graphite dishes called, because of their cigarlike shape, “boats.” Each boat contains a different semiconductor material that is heated until it melts. You slide your wafer along the tube, leaving it to sit a while on top of each boat. Cooling causes some of the material to precipitate onto the surface of the wafer. LPE produces relatively thick layers, and the boundaries between them are gradual rather than sharply defined. Precise control over thickness is almost impossible to achieve.
MOCVD (sometimes, confusingly, also known as MOVPE) systems became the method of choice for growing high-brightness devices, originally red LEDs, in the mid-1980s. MOCVD accomplishes the abrupt transition between layers by allowing the crystal grower to run two mixes of gases through the system simultaneously. While using mix A to grow a film in the reactor, you have all the gases for mix B flowing directly to the exhaust. Then, at just the right moment, you switch mix A to the exhaust and send mix B to the reactor. The process takes a matter of moments. All you hear is the sound of the compressed air-driven pneumatic valves. They open and close in quick succession—phsst, phsst, phsst, phsst—et voila! You have grown a quantum well.
So much for vapor deposition. Now we come to metal organic chemicals. Why it is necessary to use such fancy-sounding stuff instead of ordinary metal? The answer is that, in their vapor phase, neither aluminum, gallium, nor indium—the three most common metals used in growing bright blue (and red and green) LEDs—can muster sufficient vapor pressure to make it as far as the jar under their own steam. They have to be picked up and carried there, in organic form. To bump up the vapor pressure, organic chemicals such as methyl groups are attached to the metals. Gallium becomes trimethyl gallium; the positive-type dopant magnesium becomes bis(cyclopentadienyl) magnesium, mercifully abbreviated as CP2Mg. The carrier gas is hydrogen. It is kept flowing through the system at the rate of many liters per minute. During its travels, the hydrogen bubbles through the temperature-controlled baths that contain the metal organic mixtures. The gas picks up some of the compounds, which it transports to the reactor. When the compound gases get to the hot zone, they lose their methyl groups. The nitrogen for gallium nitride arrives at the jar in the form of ammonia. The heat decomposes the gas, leaving nitrogen atoms hot to trot with their gallium partners.
The process of growing a gallium nitride LED begins by heating the sapphire wafer to a very high temperature. Once hot, you clean the surface by flowing nitrogen over it. Then you drop the temperature way down to maybe 500 degrees Celsius to grow the first layer, the so-called nucleation, or buffer, layer. This is a thin film, typically of gallium or aluminum nitride and just fifty to one hundred atoms thick, that is grown directly on the wafer. The buffer layer is amorphous, that is, lacking a crystalline structure. When you heat it up, the surface of this amorphous layer becomes very lumpy as nucleation islands oriented to the surface of the sapphire start to form. As you reach higher temperatures, however, these islands grow together laterally, to form a smooth, mirrorlike layer of gallium or aluminum nitride. One of the secrets of growing high-quality GaN is being able to control exactly how this nucleation layer is deposited, how it crystallizes, and how it grows together during the heat-up step.
On top of the nucleation layer, you deposit plain vanilla (i.e., undoped) gallium nitride. Next comes a layer of negative-type gallium nitride, with silane as the electron-donating dopant. That is followed by a layer of negatively doped aluminum gallium nitride, a compound with a wider bandgap than GaN. This layer plus another, positively doped layer of AlGaN on the other side serve to confine the charge carriers within the active (i.e., light emitting) layer of the device. Then you drop the temperature down from 1,000–1,200 degrees Celsius to 750–850 degrees Celsius so that you can grow an indium gallium nitride quantum well. You grow, say, 20 angstroms of InGaN, then maybe 100 angstroms of GaN, then repeat the process for as many quantum wells as your recipe calls for, adjusting the amount of indium to produce the desired wavelength of light. The more indium you include, the greener the output will be. After growing your last combo of InGaN + GaN, you crank the temperature back up and deposit your other confining layer of positively doped aluminum gallium nitride. Then you cap the whole thing off with a layer of positive-type gallium nitride using magnesium as the hole-donating dopant. That completes the device.
In a typical growth run, the whole process takes anywhere between two and a half and four hours. If you load your wafer first thing in the morning just as the coffee is brewing, you will get the growth run out around lunchtime. You could schedule another run around two o'clock and have it out before dinner. Between runs, you have to clean the reactor by baking it out at high temperature. In the R&D lab, two runs is not a bad day. On the production line, four growth runs in a twenty-four-hour period is considered pretty good going. A large production-line reactor may contain a platter with as many as one hundred wafers on board.
The growth process itself is not in the least dramatic. You can hear faint hums and hisses from the pumps and the valves, but that's about it. The only smell MOCVD machines give off is a subtle whiff of burnt reactants that emanates from inside the jar. If you smell anything else—ammonia, for example—that means there is a leak. This is a good time to leave the lab. Quickly.
Though in many ways the epitome of high tech, operating an MOCVD reactor is actually a fine art. Herb Maruska, a gallium nitride pioneer we shall meet in the next chapter, compares it to playing a musical instrument. “Lots of people can play the violin, but only a few have the superior musical abilities that make them virtuosos. Similarly, some people really understand the peculiarities of their reactors: they know just how to position the wafer, to set the flows of various gases, to switch the temperatures. It really comes down to feel.”
In March 1988, three months after getting the green light to go ahead with his plan, Shuji was on a Delta Airways jet flying from Tokyo via Atlanta to Gainesville, home of the University of Florida, the fourth-largest public university in the United States. It was the first time the country boy had ever boarded an airplane. Like many first-time fliers, Shuji fretted that the plane might fall from the sky. It was also the first time he had been abroad. Assuming he did manage to arrive alive, he was worried that his rudimentary and thus-far-untested high school English would enable him to communicate with Americans.
As it happened, language would not be Nakamura's main problem in the United States. Most of his fellow students in Ramu Ramaswamy's laboratory were also from Asia, mainly Korea and Taiwan. English was not their first language, either. Shuji was astonished by the ethnic diversity of the student body at Gainesville. Which country does this university belong to? he wondered. In later life, Nakamura would marvel at the generosity of Americans in inviting knowledge-seekers from all over the world to come study on their campuses. Such openness compared favorably, he felt, with the close-mindedness of their Japanese counterparts.
In 1988 Shuji turned thirty-four, rather long in the tooth for a student. His fellow researchers were mostly in their mid-twenties. All of them had doctoral degrees. A common failing among academics, especially young academics, is that they tend to be overly status conscious. A person with a PhD will typically let you know that he has a PhD within moments of your first meeting. This failing is particularly pronounced among Asian academics, whose hierarchical cultures produce an exceptional degree of status consciousness.
Shuji's status at Gainesville was ambiguous. He was coming to the university at the behest of Sakai, a visiting professor. Since his time was limited to one year and he was not studying for a degree, he was obviously not a student. Nor, since he did not have a PhD, could he be offered a postdoctoral fellowship. As a compromise, he was designated a “guest research associate.”
Nakamura arrived in Florida as something of a mystery man. Nothing was known about him other than what he had written about his previous work experience and his proposed research theme. This of course did not include any mention of bright blue LEDs. Formally, he was there to do research on infrared LEDs made of gallium arsenide. Nichia's real intentions in sending him to the United States would remain a closely guarded secret.
Initially, his colleagues treated Shuji as an equal or even, because he was older, as a senior. However, once they discovered that he had only a master's degree to his name and, worse, that he had not published a single paper, their attitude toward him changed completely. Henceforth they looked down on Nakamura, treating him as little more than a lab technician. Nakamura felt humiliated. It was particularly galling because, from his perspective, these puffed-up PhDs were mere novices. He had years of hands-on experience under his belt. They could not do the simplest experiment without making a fuss. Something would go wrong and they would come crying to him for help. He would show them what to do. But that did not make the brats any less snooty in their attitude toward him.
For Shuji, Fridays were sheer torture. On that day, Ramaswamy held discussion sessions that lasted from eight o'clock in the morning sometimes until quite late in the evening. “We'd talk about the research papers they were writing,” Ramaswamy recalled, “with every student going to the board and discussing their concepts, their problems, how things could be modified, and this and that. I got the feeling that Nakamura was intimidated, because he hadn't published anything, didn't have a PhD. I used to look at his face and he would not be very happy, he would look very perturbed. He didn't have the confidence, he wouldn't ask any questions, he was very shy and [perhaps because of the language issue] he spoke very little.”
Nakamura was by nature a diligent worker. The guilt he felt about his failure to develop a commercially successful product combined with his anger at the way Nichia had treated him served to motivate him further. He had always hated to lose. Now, the arrogance of these greenhorn academics poured fuel on the competitive fires that burned within him.
“I do not like to be defeated,” he wrote. “I feel resentful when people look down on me. At that time, I developed more fighting spirit—I would not allow myself to be beaten by such low-level people.”
In a word Nakamura was, as Ramaswamy put it, driven.
“He was a bulldog worker, he would work around the clock. I used to come back to the lab late at night, sometimes I'd stay until ten, then I'd go home and I'd forget something. I'd get in my car and drive back to the lab and I'd see him working at two o'clock. Next day I'd come back at five or six in the morning, and he'd be still there! I'd say to him, Don't you go home and sleep? And he'd say, Well, I was in the middle of this, so I thought I'd finish it. I think maybe he felt somewhat insecure, because he was so motivated, and so driven.”
Outside of his work at the laboratory, not much distracted Shuji during his time in Florida. Gainesville was not much to his liking. “The end of the South,” as the town is sometimes known, the birthplace of the sports drink Gatorade was, as Nakamura saw it, “surrounded by a swamp full of alligators and mosquitoes, a place where African Americans were still discriminated against.”
He lived in three-hundred-dollar-a-week student accommodation, eating out most nights at a cheap Chinese restaurant nearby. His only friend was a Japanese professor who lived in the neighborhood. The kindly academic would invite his fellow countryman to go fishing with him, then back to his apartment for dinner where he would cook what they had caught. On this first long absence, Shuji was naturally homesick for his family and his native land. During the summer holidays Hiroko and the girls came to visit him. He treated them to a trip to Disney World in nearby Orlando. During his year in the United States, it was the only time he took off.
In addition to wrestling with his inner demons, there was also a pressing practical reason why Shuji worked so relentlessly. He had come to Gainesville to learn how to do MOCVD. When he arrived, however, he had been shocked to discover that getting time to experiment on the university's two existing MOCVD systems was not going to be easy.
State-of-the-art equipment is always fully booked. Inevitably, the lion's share of access time goes to the most powerful. “Professors are like…how shall I put it? Wild animals roaming the mountains,” Ramaswamy said. “You can't get them in a corral and make them go round and round.” Others saw the situation more prosaically. A turf war had been waged over control of the systems at the university, and it seemed like the gentle Ramaswamy had lost out.
Not surprisingly, Ramaswamy wanted a machine of his own for his lab. Commercial MOCVD equipment did not quite suit his purposes, so he had brought in Sakai to custom design a system for him. By the time Nakamura got there, the parts had arrived, but they had yet to be assembled. Shuji realized that if he was ever going to have a chance of using this machine, then he would have to help Sakai build it. He had to spend ten months of his precious year in the United States with his sleeves rolled up, connecting pipes and welding quartz, just like back at Nichia. He threw himself into the task, working sixteen hours a day, seven days a week.
Here again, as with Shuji's long and apparently fruitless apprenticeship learning the basics of LED growth, adversity in the short term would turn out in the long term to be priceless training for his quest to develop the world's first bright blue LEDs. He was willy-nilly gaining an intimate familiarity with the inner workings of MOCVD equipment that few could match.
At last, having managed to assemble the equipment, Nakamura was not about to waste what little time in the United States he had left. He wanted exclusive use of the machine that he had done so much to assemble. This led to clashes with the other students. Ramaswamy was forced to intervene to settle the issue. In the end, Nakamura was only able to do about ten device-growing runs on the system. Driven as never before, he was frantically busy right up until the last moment. Then it was time to go home.
Nakamura returned to Anan in March 1989. While in the United States he had ordered his own basic MOCVD equipment, keeping his goal a secret from the supplier, Japan Oxygen. He stated on the order form that his purpose was to grow gallium arsenide infrared LEDs. Now the huge $1.6 million machine had arrived, but the big question was, what material would he choose to grow in it?
There were, as Shuji knew well from his extensive reading of LED literature, and from conferences he had attended during his time in the United States, three candidates. All were compound semiconductors. Silicon carbide, despite the fact that it was already in limited commercial production, was one he had already rejected. LEDs grown from silicon carbide produced a weedy blue light, rather like the color of denim jeans that have been repeatedly washed. Silicon carbide had an indirect bandgap. In plain English that meant the material would never be able to emit bright blue light.
The other two materials, zinc selenide and gallium nitride, both suffered from the same major deficiencies. One was that, in order to make a proper LED, you need to be able to fabricate two types of material: negative type, doped with impurities to give it an excess of electrons, and positive type, doped with impurities to give it an excess of holes. Thus far, however, it had proved impossible to fabricate either p-type zinc selenide or p-type gallium nitride. At the time he was making his decision, Nakamura could not have known it, but this was about to change: in 1989, researchers would succeed in fabricating p-type gallium nitride; the following year would see the first p-type zinc selenide.
The other, more serious, drawback was the lack of a suitable base material on which to build your LED. Gallium arsenide LEDs could be grown on gallium arsenide wafers sliced from ingots of GaAs. But nobody had been able to grow bulk zinc selenide or gallium nitride. That meant you had to employ wafers of some other material as the substrate. Which in turn meant that there was always going to be a mismatch between substrate and light emitting layers. That is to say, there would be defects such as cracks in the crystal lattice. It was a bit like trying to fit Lego bricks onto a base made by another toy brick company. Defects are bad news for LEDs because they cause devices to dissipate energy in the form of heat instead of light.
With zinc selenide, a very soft material, the problem seemed much less severe. You could grow zinc selenide on a gallium arsenide substrate, and the mismatch was only 0.3 percent, not so far off the ideal value of 0.01 percent. This translated into a defect density of around one thousand per square centimeter.
With gallium nitride, a rock-hard material, it was an altogether different story. The best available substrate for GaN was sapphire. But even sapphire produced a huge mismatch of 16 percent. That translated into a defect density of a whopping ten billion per square centimeter. It was plausible to imagine that, given time and effort, imperfections in crystal ZnSe could be reduced by an order of magnitude. But ten billion defects? There was no way that figure was going to be significantly reduced in any researcher's working lifetime.
Gallium nitride, as we shall see in the next chapter, had been thoroughly investigated by RCA, Bell Labs, and Matsushita. GaN, or “gan” as it is often called these days, was almost universally perceived to be a dead end. Indeed, worldwide, there were only three or four groups still active in the GaN field, most of them at universities. By the late 1980s, the overwhelming consensus in the research community was that if you wanted to do bright blue, zinc selenide was the way to go. Witness large-scale programs dedicated to developing devices made from the material at universities including Brown, North Carolina State, and Purdue and at companies such as Sony, Matsushita, Toshiba, IBM, and 3M. At the domestic Japanese compound semiconductor conferences Nakamura attended, he noted that participants at the ZnSe sessions numbered in the hundreds. At the sessions on gallium nitride, only a handful of researchers would show up.
Yet zinc selenide devices, such as they were, displayed (and would continue to display) a depressing tendency to fall apart when zapped with current. For a crystal lattice, giving birth to a photon is a stressful event. Zinc selenide simply wasn't robust enough to cope with the stress. What few people could have foreseen in 1989 was that gallium nitride would turn out, for reasons that are even today not well understood, to behave very differently than any previous light emitting material. Any other semiconductor with that density of defects would be dead in the water. Much to everybody's surprise, however, defects just didn't seem to matter with gallium nitride. It would prove a magical material.
Having arrived at what he described as this “fateful fork in the road” just before his return to Japan, Shuji decided to go with gallium nitride. His reason for making this apparently reckless choice was not because he was confident that he could do what no one else had done and make bright blue LEDs. Rather it was because he had repeatedly had the bitter experience of developing products only to find that his company could not sell them because big competitors were already well established in the marketplace. If he chose zinc selenide, since big companies had had several years’ start on him in developing the material, it was likely that history would repeat itself. It was already too late. With gallium nitride, however, in the highly unlikely event that he did succeed, there would be no competition, because as far as he knew no other companies were working seriously on GaN. Nichia would end up in sole possession of the marketplace. And that, remarkably, is exactly what came to pass.
Nakamura was able to make this seemingly foolhardy decision by himself without reference to Nichia's senior management because none of them knew anything about semiconductors. All they knew was that Nakamura's target was to develop a bright blue LED. The choice of methodology to adopt and the material to work on was thus his, and his alone. If he had been working at a large company, his proposal to work on a material known to be a loser would undoubtedly have been shot down before it left the ground. But as he himself would later say, “Breakthroughs are born out of unusual circumstances.”
When Shuji embarked on his harebrained quest, in addition to his attempt to build a bright blue LED, he also had a second, more personal goal in mind. Even if he was only able to scrape together some basic results, at least he would have something to show for his efforts in the form of research papers. He was determined to prove to those swellheaded PhDs back in Florida that he, too, could write papers, despite the fact that doing so would mean breaking Nichia's regulations. By this stage, however, Shuji had had enough. He no longer cared about company rules. He was resolved to write as many papers as he could. With a bit of luck, he might even get his doctoral degree.1
This time, however, there was at least a chance that he would not have to work entirely on his own. He went to see Shiro Sakai, his friend and MOCVD mentor, who had himself recently returned from Florida to his chair at Tokushima University. Shuji suggested to the professor that the pair should collaborate on gallium nitride research. But Sakai rejected the younger man's proposal out of hand. He told Nakamura that for an academic such as himself, it was vital to keep publishing papers. Who knew whether anything would come of gallium nitride research, whether it would generate the kind of results that could be written up and published in journals? Better to stick closer to the mainstream, where the outcomes were more predictable.
Several ironies here. First, starting in 1991, Nakamura would publish the first in a long string of papers. Within a few years of their publication, these papers would become among the most cited pieces of scientific literature in the world. Second, perhaps getting wind of the fact that Nakamura was making great progress, Sakai would come to have second thoughts about the merits of gallium nitride research. He would secretly switch his focus at Tokushima to work on GaN, without telling his former friend and colleague that he had done so. Nakamura would only find out about what was going on in Sakai's lab later, when Nichia hired some of his students. Ultimately, Sakai would even start his own GaN-based company, Nitride Semiconductor. But he was an academic, not an entrepreneur, and the company would not be a success.
In the meantime, Sakai's refusal to collaborate meant that Nakamura, once again, would have to proceed on his own. Shuji was used to that, indeed he found it easier to solve problems if he did all the thinking, without interference from others. But on his lonely marathon, he would soon come up against his most difficult challenge yet. And this seemingly insurmountable obstacle would come, not from the technology, or external rivals, but from within his own company.
But before delving further into Nakamura's travails, let us first turn our attention to the owners of the shoulders Shuji had chosen to climb upon, and the history of the development of the bright blue LED.
1. He did, from his alma mater, Tokushima University, in 1994.