At 2:30 a.m. on the morning of Tuesday, October 7, 2014, just before the official announcement that he had won the Nobel Prize for Physics, Shuji Nakamura called to tell the news to his close friend, neighbor, and UC Santa Barbara colleague Steve DenBaars. “You know what,” DenBaars advised him, “we should probably go to the university.” Once word got out, the media would come pounding on Nakamura's front door, demanding comment. Shuji was reluctant to leave, reasoning that he lived in a gated community with a fence that would keep the hacks at bay. But DenBaars insisted, driving over to pick him up. He arrived in the nick of time. Just as Shuji emerged from his house, a pack of camera-toting Japanese journalists had scaled the fence and were running down the road toward them. “So we drive to the university, and they're following the car like paparazzi,” DenBaars recalled, laughing. “I've never seen anything like it!”
Not content with grilling the new laureate, the Japanese media subsequently headed off to the local ramen place where Shuji and his co-workers often eat lunch. When DenBaars dropped by later to pick up some sushi, he was astonished to discover Japanese TV crews filming bowls of Nakamura's favorite noodles on the table. Such craziness was understandable. Together, Shuji and his old rivals Isamu Akasaki and Hiroshi Amano had jointly won the 8 million kronor ($1.1 million) prize for their invention of the bright blue LED. It was a unique all-Japanese trifecta.
Shuji was flabbergasted by the news of the award. “Unbelievable!” was his initial reaction. “I was not sure whether I could win the Nobel Prize, because the physics prize is awarded to people for basic theory,” he confessed later. “But in my case it was not basic theory, it was just making the device.” To be sure, some commentators sniffed that the Nobel should not be given for a mere invention, it was intended for esoteric discoveries like the Higgs boson. In fact, however, there was plenty of precedent. Every few years the physics prize is awarded for an invention that has conferred great benefit to humankind. Recent examples include the integrated circuit (2000), and fiber optics and the charge-coupled device (2009). “Alfred Nobel wanted his prize to be given to inventions that benefitted mankind and that's what we have recognized today,” Nobel Committee for Physics chairman Per Delsing said at a press conference in Stockholm following the announcement. “I think Alfred Nobel would be really happy about this prize.”
But every rose has its thorn: a Nobel Prize can only have three winners. In the case of LEDs, that meant Nick Holonyak, who invented the world's first visible LED way back in 1962, missed out. Holonyak, now eighty-five and in poor health, was understandably upset that his pioneering work had been overlooked. Shuji had met Holonyak for the first time in 2012, at a seminar to mark the LED's fiftieth anniversary held at the University of Illinois (where the latter was professor emeritus). Shortly after the announcement of the prize, he made a point of sending his illustrious predecessor a letter thanking him for his early work and for the motivation it had provided.
Otherwise, reaction to Nakamura's winning the Nobel was overwhelmingly positive. “It's fabulous,” Herb Kroemer, his fellow laureate and former colleague at UCSB, told a reporter from the Los Angeles Times. “Shuji has the courage to try things that other people wouldn't even dream of doing.” Coverage of the Nobel in the national media would undoubtedly raise mainstream awareness of solid-state lighting. “I think it's wonderful for the LED community and the industry,” said veteran analyst Bob Steele. Press coverage in Japan was uniformly positive. Two weeks after the Swedish announcement, the Japanese government followed suit, nominating Nakamura as a recipient of the Order of Culture for his contributions to the nation. One of Japan's highest honors, the medal would be presented by Emperor Akihito in a ceremony at the Imperial Palace in Tokyo.
Shuji had cleared the last hurdle in his race, leaving controversy far behind. Winning the Nobel was “like vindication, the final stamp,” Steve DenBaars told me. “Nobody can ever take it away from him.” But in fact, the race was not over. And true to form, Shuji was not slowing down; he was still, as we shall see, showing the way.
Research remains Shuji's passion. He still comes into his lab at the university most mornings between 7:30 and 8:00 and works ’til 5:00 or 6:00. In academia, Nakamura's responsibilities include training the next generation of researchers. Plus the never-ending grind to drum up grants to support their work. Most professors will mentor two or three postgraduates at a time. Shuji always had around ten doctoral students in his group. In almost fourteen years since Nakamura came to UC Santa Barbara, he reckoned he must have graduated between fifty and sixty PhDs.
Perhaps his biggest achievement at the university—and a testament to the continued widespread belief in his ability to deliver—was the Solid-State Lighting and Energy Electronics Center. A global industrial consortium, the center was founded in 2002, but its activities did not really take off until 2007. The consortium consists of twelve members, mostly Japanese corporations, but it also includes American, Korean, and European firms. What makes the center unique is the almost unheard-of way the companies donated their money. “It was not a contract,” DenBaars explained, “it was just an unrestricted grant. So it was like, Here's two million dollars over the next five years—just go make whatever you want, and we'll trust your innovation and judgment.”
Historically, the lion's share of funding for research and development in compound semiconductors has come from the US Department of Defense. Gallium nitride devices are of interest to military contractors because of, among other things, their applications as amplifiers of radar and high-frequency wireless signals. As a foreign national, Nakamura could not even attend reviews at the Defense Advanced Projects Research Agency. In order for his group to continue receiving government money, in 2006 he applied for American citizenship.
Like any good engineer, Shuji sees himself primarily as a problem solver. The biggest problem in gallium nitride work remained the lack of what is known in the jargon as a “native” substrate. All other semiconductor materials are grown on like material. For instance, silicon devices are fabricated on silicon wafers. With gallium nitride, however, because it was extremely difficult to produce salami-like “boules” of bulk crystal, researchers were forced to fall back on “foreign” substrates like sapphire or silicon carbide. Devices grown on such substrates were, as we have seen, riddled with defects. This in turn meant that, to avoid catastrophic failure, light emitters could not be run at high power density.
Even at lower currents, researchers had encountered a puzzling issue that they dubbed “droop.” Crank up the current, and the efficiency—how much electricity is converted into light—of the LED would sag. This was a big deal because researchers are forever striving to improve the efficiency of their devices. Today's commercial LEDs are typically only about 50 percent efficient, and, with the addition of a phosphor, efficiency drops further. When I spoke to him a few days after the Nobel announcement, Shuji told me that his ultimate goal is to push efficiency “as close to one hundred percent as possible.”
But though bulk gallium nitride was hard to make, it was not impossible. Giant Japanese corporations like Sumitomo Electric, Hitachi Cable, and Mitsubishi Chemical had deep-enough pockets to invest in production facilities. For such firms the reason for pouring resources into manufacturing this material was the lure of a lucrative niche market. Blu-ray digital video disk players and game machines like Sony's popular PlayStation used violet-blue lasers, descendants of Shuji's work at Nichia. Unable to tolerate the defect densities of gallium nitride on sapphire, lasers had to be grown on GaN.
Shuji had long dreamed of growing gallium nitride devices on native substrates, “GaN-on-GaN,” as it would later come to be known. Now, in 2006, with the availability of the new Blu-ray substrates, even though a single two-inch wafer of gallium nitride might cost several thousand dollars, it was finally possible for him to put his ideas into practice. At the time, the prevailing notion was that GaN-on-GaN devices, even if practicable, would be prohibitively expensive. With sapphire wafers selling for a few dollars a pop, it was crazy even to think about making such things. But as ever, once Nakamura had determined that there was a better way to do things, it was impossible to dissuade him. Here again, his leadership would inspire others. “Shuji's always thinking of the newest thing to try,” DenBaars said. “GaN-on-GaN is something that people wouldn't have tried unless Shuji had pushed Jim [Speck] and me, and pushed our students, too.”
In January 2007, Nakamura and his colleagues announced via a press release that they had used the new material to make “a major breakthrough,” a new low-power blue-violet laser. Among the most interested readers of the press release was the venture capitalist Vinod Khosla. A Silicon Valley legend, Khosla has a track record of funding successful start-ups. Latterly, his company Khosla Ventures has focused on investing in the cleantech sector. In other words, on technologies like LEDs that, as well as providing commercial benefits, also address environmental concerns such as climate change. Khosla had originally approached Shuji back in 2001, just after his move to Santa Barbara. He proposed that Nakamura, DenBaars, and Speck should start an LED company. “Shuji looked straight at this guy, who's a multibillionaire,” DenBaars recalled. “And he said, No, the time's not right yet—we need to come up with our own new technology.”
Six years on, Khosla flew down to Santa Barbara to meet with the three professors. “He said, OK, now you have a new technology, are you ready to start a company? We thought about it for about a week, and then we said, Yeah.” They named their start-up Soraa, after the Japanese word for “sky.” The trio had no business plan, they had not put together a cash-flow analysis or any of the other stuff that is supposed to be sine qua non for a start-up. What they did have was impressive results. “That's what Shuji's so good at, he produces the results that can let you [go ahead],” DenBaars said. “It was just, spin-out the technology and good things will happen.”
An oversimplification, as turned out. Defying the conventional wisdom that GaN-on-GaN technology was ridiculously expensive, hence commercially unviable, made it challenging for the fledgling firm to convince other investors. In the company's early days, employees faced month-to-month payroll cuts. Many times the firm came close to going under. Substantial personal loans were needed to keep Soraa afloat. In a sense, Vinod Khosla was like a latter-day Nobuo Ogawa, a patron who was prepared to back Shuji's vision all the way. By 2014, Soraa had managed to raise over $100 million, most of it from Khosla Ventures.
The allure of Nakamura's name enabled the start-up to attract top talent. One of Soraa's first recruits was Mike Krames, who had led the Advanced Laboratories at Lumileds, a subsidiary of Philips. This group had developed the LEDs used in Apple's iPhone. It had also identified the mechanism that caused the droop phenomenon, as “Auger recombination” (don't ask). In 2009 Krames signed on as Soraa's chief technology officer. The company moved its base up from Goleta, near Santa Barbara, to Fremont, Silicon Valley's cleantech hub.
Though lasers had stimulated Soraa's founding, it soon became clear that GaN-on-GaN was ideal for better LEDs, too. The company's focus switched to making what it would term “LED 2.0,” a second generation of light emitting diodes. In February 2012, after four years beavering away in stealth mode, Soraa startled the lighting industry with the announcement of its first product. It was a replacement for the 50-watt MR16 halogen lamp. This was a product whose small size and intense, high-quality brightness LED makers had thus far failed to match.
The ideal for any form of artificial light is to replicate daylight. As everyone knows, this consists of the colors of the rainbow, violet through red. Incandescents, of which halogen is a subset, did an effortlessly good job of rendering natural light. Fluorescents, at least in their early versions, did not. Consumers hated the “green ghoul” effect that tubes produced, which made skin look unhealthy. The reason was that fluorescents were not capable of reproducing the entire spectrum, merely some parts of it. Despite improvements, tubes never managed to recover public trust. Lighting industry sages warned LED manufacturers that they must learn the lesson of history and not repeat the same mistake. But Haitz's law only measures two variables: brightness and cost. In their haste to produce ever-cheaper, ever-brighter lights, makers were neglecting a third key variable: quality of light.
First-generation LEDs produce white by pumping blue light through a yellow phosphor. Their output lacks a violet component. At the other end of the spectrum, deep red is also absent. For certain applications, however, these colors are vital. In particular, the hospitality and retail sectors. Restaurants want their food to look appetizing, butchers their meat juicy, fashion stores their clothes sexy. But without deep red, everything just looks washed-out. Likewise, makers of fabrics and high-end paper incorporate optical brightening agents to give their products what designers call “pop.” In detergents, such agents make clothes look “whiter than white.” But without violet, they end up looking dingy yellow. And, as anyone who has ever been to a disco knows, short-wavelength violet light also excites naturally occurring fluorescence, for example, in teeth.
So why not simply start with a violet light emitter and add another phosphor to turbocharge the red? Because with conventional GaN-on-sapphire technology, the intrinsic high defect density means that devices cannot be driven that hard. The penalty in terms of diminished light output is simply too great for such a full-spectrum emitter to be viable. GaN-on-GaN has up to a thousand times fewer flaws than GaN-on-sapphire. (Comparing electron micrographs of the two crystals reveals a dramatic difference in appearance: GaN-on-sapphire looks like a turbid maelstrom, while GaN-on-GaN resembles gentle ripples on clear water.) Near-flawless material enables GaN-on-GaN devices to handle ten times more power density. That translates into much better performance, especially at shorter (violet) wavelengths. GaN-on-GaN's ability to tolerate higher current also greatly reduces droop. Moreover, five times less material is required to produce the same amount of lumens. Soraa's chips were about a fifth of the size of rival LEDs. To maximize light extraction, they boasted a unique, triangular shape. Shuji is convinced that, in the future, GaN-on-GaN will be the material of choice for all LEDs. “The performance is so much better,” he explained simply.
Emboldened by the robustness of their new technology, Soraa's researchers resolved, as Mike Krames told me, “to thumb our nose at the whole idea of broken-spectrum light.”
For their initial product offering, they chose an audacious target: to replace halogen as the light source in MR16 lamps, the most difficult form factor possible. Compact multifaceted reflector (MR) lights are familiar as the spots and downlights found in many household ceilings. They are also widely deployed in high-end commercial and institutional applications where quality of light is crucial. For example, to illuminate artwork in museums and blackjack tables in casinos. Though a small part of the overall lighting market, halogen replacement is not a niche. MR16s occupy perhaps half a billion sockets around the world. Annual sales number in the hundreds of millions of dollars. Makers had tried to replicate halogen MR16s, with unsatisfactory results. “LED lamps still trail legacy lamps [halogens] by a significant margin,” a 2014 report by the US Department of Energy concluded.
To produce enough light, some makers had been forced to resort to using multiple LEDs, giving the lamps the appearance of a shower head. This approach resulted in blurry shadows and in some cases necessitated the inclusion of a fan to get rid of the extra heat. There was also the problem of beam width: to merit the designation, an MR16 must be able to produce a narrow spot. Conventional gallium nitride emitters could not muster the optical oomph. Whereas for Soraa, with its more powerful, much smaller source, keeping the beam nice and tight was not a problem.
Lighting designers had begged Shuji to do something to improve the poor color rendering of solid-state lighting. They were enchanted with his company's response. “This is genuinely different and better technology,” commented Jim Benya of the Portland, Oregon–based Benya Burnett Consultancy. “With the Soraa lamps, we are experiencing ‘cleaner’—more vibrant—color on art,” enthused Jan Lennox Moyer of the International Landscape Lighting Institute in upstate New York. “Clients are loving how beautiful their paintings and photographs look now.” Bill Noble of WowLighting in the United Kingdom added, “Soraa's technology is so far ahead of the game that it is almost a no-brainer to specify their products.”
As the only supplier offering a true MR16 replacement, Soraa's revenues have grown rapidly. Keeping up with demand was hard. The company was able to grab a substantial chunk of the retrofit market, competing against entrenched giants like Philips and GE. And, as with any new technology at the top of its learning curve, there was plenty of room for improvement. In the two years since the initial announcement, Soraa had introduced two new product iterations, each 30 percent more efficient than its predecessor. “[The MR16] was really a flagship product to show what could be done,” Krames said. “Since then, everybody's been chasing.”
The company's next challenge was to expand from the tip of the pyramid and take on the general lighting markets that lie beneath. Would consumers in such markets be interested? Or would Soraa's products appeal only to the 5 percent of projects where lighting was specified by designers, where color quality really mattered? After all, as Bob Steele cautioned, in many market segments, color rendering was simply not a priority. “Most lighting is very functional,” Steele told me, “it's just putting light into a space so people can go about their work or their play, reading or whatever they're doing. We've lived under incandescents and crappy fluorescents for decades, so obviously people aren't that picky.” For many of its fixtures, the lighting industry had moved to what it calls “mid-power” devices. Though not point sources, these LEDs were quite cheap and reasonably efficient. “From the industry's perspective, GaN-on-sapphire works OK,” Steele said, “It's good enough and you don't need the high current densities of the Soraa products.”
Krames countered by pointing to recent studies conducted at Penn State University. In them, participants showed a strong preference for the full-spectrum color and whiteness rendering Soraa's lamps deliver. “Now we have the data to prove that color and whiteness really matter,” Krames said. “Once consumers get a taste for higher-quality lamps, they're not going to want to go back. Quality of light will be a bigger driver of adoption than people believe.”
But in addition to consumer preference, there was also another issue, one that puzzled industry observers like Steele. In the semiconductor industry, when a new chip technology emerges, the custom is to license it as broadly as possible to achieve maximum leverage.
“The problem is, Soraa is not making their chip technology available to anyone; they're using it solely for in-house purposes,” Steele said. “Although [GaN-on-GaN] technology is great and potentially much more useful, you just can't get it; it's being tightly held by this one small company. So the big open question is, Will the technology be available?”
Krames conceded that it was inevitable that Soraa would one day license its technology. The firm would likely partner with others. It was simply a matter of when. He pointed out that, behind the scenes, rivals—some of them large corporations—were scrambling to catch up. They were pouring resources into developing GaN-on-GaN devices. “So the vision that Shuji has, where GaN-on-GaN basically takes over the world, and you have a handful of big players that are dominating is certainly a welcome vision for us—provided that we are at the top of that heap.” Of course, for an upstart entrant, cornering the entire lighting market would not be possible. Soraa's job was to show the rest of the industry the way. As ever, what would determine whether GaN-on-GaN was the next generation technology, as Shuji insisted, or merely a next generation technology was cost. On that front, too, the news was good.
With its much smaller chips, GaN-on-GaN had a natural advantage. Fewer wafers would be needed to produce the same number of lumens. Since gallium nitride wafers were still more than a hundred times more expensive than sapphire ones, reducing the cost of the substrate was imperative. The incumbent technology for bulk GaN production—hydride vapor phase epitaxy—left much to be desired in terms of material quality. But there was also another method, known as ammono-thermal growth, that promised to reduce both cost and defect density. This technology had been tried and tested in other areas. For example, it was used to produce thousands of tons of quartz annually, for applications that included crystal oscillators in watches and clocks.
As ever, Nakamura had long been eager to exploit this alternative method. “Shuji's been very big on ammono-thermal growth from day one,” said Krames. In 1999, while still at Nichia, Nakamura had flown to Poland to visit an outfit called Ammono that was attempting to develop ammono-thermal growth of gallium nitride. Just before he left Japan, Shuji had arranged for Nichia to fund development at Ammono. At UC Santa Barbara, he continued to pursue the technology. The Poles had run into trouble adapting conventional autoclaves. Among other issues, growth rates in these reactors were painfully slow. Externally heated autoclaves could not reach high-enough temperatures and pressures. It was necessary to completely redesign the reactors and build new, internally heated autoclaves that were customized for growing bulk gallium nitride. Better equipment would speed up growth rates. Substrate costs would be reduced by an order of magnitude. Bulk growth has become a hot topic. At recent nitride conferences, papers on growth methods outnumber those on any other subject.
Soraa licensed ammono-thermal patents from UCSB. With funding from the Department of Energy, the company's researchers spent five years improving the technology. “It's a lot of heavy lifting for a start-up to do this,” Krames said. By late 2014, Soraa was able to grow boules of very high-purity gallium nitride that were two inches in diameter. For commercial production, however, four-inch boules were needed. Krames estimated that to reach the point where the company could supply its own needs in-house would take another two years. In preparation, Soraa was negotiating with the governor's office in New York State on a deal for a chip and wafer factory that would bring hundreds of high-tech jobs to Buffalo.
With the first generation of his LEDs already ubiquitous and a second generation on the way, was there anything left in solid-state lighting, I asked Shuji, for him to achieve? It turned out there was: laser lighting. For one thing lasers, unlike LEDs, do not suffer from droop. “So the efficiency is very high,” Nakamura explained. “You can make very very bright lights using tiny chips.” UCSB was mounting a big effort on laser lighting. “That's where most of our PhD students work,” DenBaars said.
The first fruits of this new emphasis were already emerging in commercial applications. Having vanquished incandescents and halogens, LEDs were taking on xenon lamps. In partnership with Osram, the German auto maker BMW had developed laser headlights for its new i8 hybrid sports car. These lamps were capable of shining double the amount of light as LEDs and almost three times as much as xenon. The new, ultra-bright laser headlights could illuminate the road up to 600 meters ahead, twice as far as LED equivalents. They used 30 percent less energy than LEDs, weighed less, and occupied less space in the headlamp housing. It was also possible to imagine other road-related applications for lasers, with their much longer reach. “If you're lighting a highway,” DenBaars speculated, “the space in between streetlights becomes a kilometer instead of a hundred meters.”
Another area where lasers were displacing xenon was in cinema projectors. “All projector companies are switching to laser light sources for movie theaters,” DenBaars said. Not only were lasers capable of producing high-brightness light across a wide color gamut, because they are ultra-efficient, they also reduce running costs. Most important, they offered cinema owners a much longer operating lifetime: more than ten years, versus from 500 to 2,000 hours for xenon. As usual with solid-state lighting, up-front costs were higher. However, measured by cost of ownership, laser projectors would pay for themselves in a matter of months. Closer to home, consumer electronics companies like Sony had begun offering laser projectors for the living room. “Ultra-short throw” projectors could be placed against a wall yet still create an image as large as 147 inches across.
For this new technology to become ubiquitous, there were, Shuji acknowledged, still some challenges remaining. Safety was one: out in the real world it would not be possible, as it was in the lab, to wear safety glasses to prevent being blinded by laser light. And, as always, there was cost. “Everybody thinks lasers are expensive,” he said, “so we have to reduce the cost.” Would laser light be the next big thing? Industry observers were skeptical. But during the course of his brilliant career Shuji had confounded the skeptics many times before. The odds were good that, yet again, he would prove them wrong.