New book looks to the past to determine the future of SSL (MAGAZINE)
After sharing his work about LED pioneer Shuji Nakamura, journalist Bob Johnstone followed the path carved through new technology terrain by other LED innovators in his new book, L.E.D.: A History of the Future of Lighting — where he chronicles the origins of the light-emitting diode, its evolution, and the applications it promises to enable, illustrated in this excerpt.
After sharing his work about LED pioneer Shuji Nakamura, journalist BOB JOHNSTONE followed the path carved through new technology terrain by other LED innovators in his new book, L.E.D.: A History of the Future of Lighting — where he chronicles the origins of the light-emitting diode, its evolution, and the applications it promises to enable, illustrated in this excerpt.
The ability to foresee the future, especially when crucial parts of that future remain to be invented, is rare. Ton Begemann acquired the oracular ability through serving as a science attaché at the Dutch embassy in Washington during the 1970s. One of his reports, on American doubts about the prospects for fast-breeder nuclear reactors, caused an uproar in his native Holland. "There were seventeen questions in parliament and headlines in the newspapers," Begemann said, chuckling as he recalled the furore he had inadvertently caused. "So I decided, Well, this is the end of my career as a diplomat." On his return to Holland in 1975, he was offered a job with Philips Lighting in Eindhoven, Holland. In 1990, losing money and stung by criticisms that it had become staid and hidebound, Philips brought in new management. The incoming team included Einar Kloster, a canny Norwegian, who took over as CEO Philips Lighting. One of Kloster's first actions was to appoint Begemann as his resident soothsayer. Though lighting had not changed much in the past hundred years, complacency was dangerous because it left the business vulnerable to unexpected innovation. "Ton," Kloster instructed Begemann, "you are going to sit in the office next to me and look at what is going to happen with lighting technologies."
In his new role one of the first items that caught Begemann's eye was an advertisement in a trade journal for a new, fourth-generation light emitting diode, a device that emitted bright amber light. This was of interest because one of Philips best-selling products also emitted bright amber light: high-pressure sodium-vapor street lamps. Begemann saw that street lights based on LEDs would consume much less energy than sodium-vapor lights. He also recognized that light emitting diodes would have another advantage over conventional sources. LEDs were directional, shining only where pointed, in this case, downwards. To achieve the required illumination at street level, you would not need to produce so much light. Begemann ordered some of the new bright orange LEDs. Technicians in Philips' lab soldered couple of hundred devices onto a circuit board then hooked it up to an electric current. The tiny lights shone brightly. Not nearly enough to compete with sodium, but doubtless they would get brighter.
In 1993, seeking to share his findings with colleagues, Begemann staged a strategic directions workshop. He understood that LED lighting was not going to be an easy sell. Only a few years previously, believing the technology to be a dead-end, Philips had disbanded its LED group and ignominiously exited the field. In 1992 a working group at the company investigating "unconventional light sources" had concluded that LEDs were not suited for general illumination. But in the interim the efficiency of LEDs had doubled, then doubled again. It was time for another look. Begemann began his presentation with a cautionary tale. Imaging had long been dominated by Kodak. But the recent advent of digital cameras meant that the writing was on the wall for the venerable US firm. Light emitting diodes, he asserted, represented the same kind of existential threat in the lighting market. Admittedly, LEDs were as yet not capable of producing white light. But that was not the issue, Begemann argued, he fully expected that they would eventually acquire this capacity. In the long term LEDs had the potential to radically change illumination. For the moment, the important thing was to get cracking with the amber, then continue with the white when it became available. "LED lighting is something we need to look at," he concluded.
His story fell on deaf ears. "They didn't really believe it," Begemann told me. "Nobody said, Yes — that is something." Fortunately, he still enjoyed the backing of his boss. Kloster had been disturbed by the presentation and its implications for the future of Philips Lighting. It was imperative to find out how much brighter amber LEDs would become. The best ones were made by Hewlett Packard. As it happened, one of Kloster's close friends, a fellow Philips board member called Frank Carubba, had formerly been a senior manager at Hewlett Packard. It was arranged that in January 1994 Begemann would travel to San Jose in Silicon Valley to meet with executives at HP's Optoelectronics Division, the source of the best little orange lights.
The visit did not go well. Begemann wanted to discuss the role that LEDs might play in lighting. But the HP guys simply didn't get what he was talking about. Though frustrated, Begemann stubbornly persisted. Back in Eindhoven, he kept pestering Hewlett Packard to send someone over to Holland so that he could show them what he meant. Eventually, in March 1994, keen to get the pesky Dutchman off his back, the division manager capitulated. Roland Haitz, the number-one technology guy, not just of the optoelectronics division but of HP's entire Semiconductor Products Group, happened to be in Japan. Haitz was scheduled fly to his native Germany for a trade show. Would he mind dropping in on Philips in Holland on his way home? Reluctantly, Haitz agreed.
The visible light emitting diode was invented in 1962 by Nick Holonyak, a researcher then working for GE. It glowed a dull red. The tiny size of the device made the LED ideal for its first application, as an on/off indicator. With the advent of digital watches and calculators, LEDs were pressed into service as displays, to provide alphanumeric read-outs. Hewlett Packard got into LEDs because, as a maker of test and measuring kit, read-outs were vital to its core business. The conventional display device was the Nixie tube. Nixies were cumbersome, power hungry, hard to read, prone to breakdown and — worse — an anachronistic embarrassment at a time when the rest of the electronics in HP's products had shrunk to low-power transistors and chips. The capability of LEDs was thus closely aligned with HP's corporate strategy.
In 1970 Bill Hewlett committed his company to building a handheld scientific calculator, an electronic slide-rule that would fit in a shirt pocket. Since production was limited by how many displays HP could make, the company poured resources into the research, development, and manufacturing of LEDs. The calculator turned out to be a huge hit commercially. "This was a big deal for HP," Haitz recalled. "After that, Hewlett was so proud we had helped the company to get such a big success that we were the fair-haired boys for a long time to come." HP continued to innovate in the materials from which LEDs are fabricated. In 1979 Haitz hired George Craford, a former student of Holonyak's who had made the first yellow LED. The potential for LEDs to play a role in general illumination had been clear to Holonyak more or less from day one. "We believe there is a strong possibility of developing [the LED] as a practical white source," he told an interviewer back in 1963. Holonyak and Craford continued bandy around the notion light emitting diodes as the shape of lighting things to come. But most people simply laughed this off. So long as LEDs could only glow dull red they would continue to be a niche market. LEDs were merely light to be seen, not light to see by, which is what Ton Begemann was proposing.
Then, beginning in the mid eighties, new light-emitting compounds were investigated. For the first time an unfamiliar adjective began to be applied to describe the output of LEDs. At long last the tiny lights were becoming bright. Serendipitously, a new application popped up for which bright red LEDs were a perfect fit. In 1985, to reduce read-end collisions — one of the most common forms of road accident — the US National Highway Traffic Safety Administration required that thenceforth all automobiles must be equipped with a third brake light. This would be mounted at eye-level in the center of the rear windshield. By the early 1990s it was clear that bright red LEDs would prevail in the brake light market. Armed with its new bright orange lights, Hewlett Packard was eyeing a related niche — automotive turn-signals. Then, in December 1993, came a bombshell. An obscure, previously unknown Japanese firm named Nichia announced that it had achieved what many of the biggest names in the electronics industry had been trying to make for the past two decades: namely, a bright blue LED. Overnight, the game had changed. With red, green, and blue LEDs, it would now be possible to make white light — as just a few months earlier, with remarkable prescience, Ton Begemann had anticipated. Roland Haitz told his boss that HP urgently needed to set up a blue LED development program. Haitz himself took off for Japan to meet with Nichia. Which is where he was when the request for him to visit Philips Lighting arrived.
"It was the most nerve-wracking trip I ever did," was how Haitz remembered his late-evening drive through the Ruhr from Hanover to Holland. After a horrendous three hours negotiating heavy traffic amid torrential rain, he finally reached Eindhoven. Next morning Ton Begemann gave him a guided tour of Philips Lighting's laboratory. It was a revelation for Haitz who until that point, as Begemann recalled with a chuckle, had known nothing about lighting. Haitz himself described his tour as an introduction to a completely different world. "I learned so much on that day about what counts in lighting," he told me.
In return, Begemann badgered Haitz for answers to his questions: The efficiency of HP's amber LEDs had doubled twice in recent months. How much improvement in their output could be expected by the year 2000? In his role as chief technologist Haitz was, like Begemann, accustomed to peering into the future. HP's customers in the automotive industry were always eager to learn what was in the product pipeline. LEDs like the ones that had caught Begemann's eye were based on technology Hewlett Packard had commercialized two years previously and developed three years before that. Between what Haitz knew was in HP's labs back in Silicon Valley and what Philips had acquired there was thus a five-year gap. In addition to which, he was able extrapolate product development five years hence. Though Haitz kept his cards close to his chest — Philips had yet to sign a non-disclosure agreement with HP — some hasty, back-of-the-envelope calculations left both parties in no doubt by the end of the day that the future of LEDs lay in the lighting business.
On the plane home to California Haitz wrote up his impressions in a trip report. LEDs would clearly be in lighting and lighting was a huge market. If Hewlett Packard wanted to remain a major player in the LED business, the company had no choice but to participate. But HP could not go it alone because it didn't understand the market. "If we are not tied up with a lighting company, we will make mistake after mistake, and we are going to get our ass kicked," Haitz concluded, with characteristic brusqueness. "We have to find a partner." There were three possibilities: Osram-Sylvania, GE, and Philips. Osram already had its own internal LED business; GE was no longer interested in making investments in lighting technology. "So there was really only one girl to dance with, and that was Philips."
LEDs as light to see by required a leap of the imagination. Their output compared to that of conventional lamps was still puny. HP's top brass took a lot of persuading that there would be value in teaming up with Philips Lighting. But Haitz persevered; eventually, he succeeded in winning them over. For his part, Begemann reported back to Philips that Haitz had told him that by 2000 it would be possible to make LED street-lights that were as good as the most efficient conventional products. Most of his colleagues still thought he was crazy. But his boss gave him the go-ahead to set up a feasibility study with HP, to be followed by a demonstration project. The two companies initiated their cooperation in late 1994. It was decided to focus initially on the development of a prototype street-light based on HP's amber LEDs because that market was already accustomed to orange-colored illumination.
"I got two guys from HP and two guys from Philips Lighting and they spent half a year locked up in our lab here in Holland," Begemann recalled. Such a tight deadline was very different from the leisurely way that development programs were usually run at Philips. "This was a project where we were exposed to Bay Area dynamics and the semiconductor world, where things are three times faster if not more," recalled Berry Kock, one of the two Philips representatives. Team members were under strict instructions not to discuss their work with anyone. The benchmark had been defined: the efficiency of high-pressure sodium lights was about one hundred lumens per watt. LEDs could only manage half as many lumens, but by directing the light more efficiently in a much tighter beam, the team showed it was possible to match the performance of conventional products.
By October 1995, they were ready to unveil their work. Begemann made sure that all his colleagues in top management at Philips Lighting attended. Three prototype LED streetlights were set up in a warehouse-like building in Eindhoven. Inside it was like a movie set, complete with props, in this case traffic signals and lines painted on the floor to simulate an actual street. "We tried to make it as real as possible," Kock said. For maximum effect, the curtains were tightly drawn to make it very dark inside. The element of surprise worked in their favour. "Everyone was completely flabbergasted that we were doing such a thing with LEDs which were thought to be tiny lights on your stereo set," Kock told me. "It was definitely a moment when people noticed something which they did not believe would ever exist."
Gerard Harbers, then an optical designer from Philips Research, would never forget the thrill of seeing LED streetlights for the first time. "It made a huge impression on me," Harbers said. "Here you were in this big space and there's this single lamp which illuminates the whole space, the whole street. It showed, for the first time, real solid-state lighting. And I think that everyone who saw the demo got that same feeling: this will be big, this will be huge — if you can do this already now, what will the future bring?"
For Philips and Hewlett Packard the next step was to form a limited 50:50 joint venture based in Amsterdam. This was announced in late 1996, with minimum fanfare, so as not to alert the competition. It was a small operation, mustering about thirty engineers. The aim was to take LED component technology from HP, couple it with applications and assembly expertise from Philips, and target sections of the lighting market. The first niche chosen was red and amber traffic signals. This was seen as an initial stepping stone, to get the company started and pave the way for more complex applications like street lights.
In 1999, encouraged by the results from this first niche market, Philips led by Ton Begemann initiated a push to scale up the joint venture. It was decided to locate the new firm, named Lumileds, in San Jose, where most of the technical expertise resided. Also, to make it as much like a startup as possible, complete with stock options for employees. This was an unprecedented move for the Dutch company, but it was done for a good reason: namely, to avoid its stuffy corporate culture stifling the entrepreneurial spirit. For Hewlett Packard the idea of forming an independent joint venture was not such a stretch. It helped that HP was then undergoing a wrenching corporate restructure. In 1999 the firm split itself in two, with the original company retaining the computer-related businesses and a new entity called Agilent formed to retain the test and measurement businesses. Among the latter was the semiconductor products group, which contained the optoelectronics division. Eventually most of the division — around three hundred individuals — left to join Lumileds.
In the joint venture there were, inevitably, clashes of culture. The Dutch were accustomed to calling a spade a spade: their bluntness could upset the milder-mannered Silicon Valley folk. There was also a gulf between the worlds of lighting and semiconductors. The Philips people were top-down in their approach, accustomed to thinking in terms of system solutions, how lighting could be used in offices and retail spaces to create a better atmosphere; or, if it was street lights, how illumination could make roads safer for drivers. The HP folk were by contrast bottom-up. They saw things in terms of nuts and bolts, what increased light output would mean for package design, how thermal performance would need to be improved. In higher-level discussions the components guys would sometimes struggle to comprehend the issues. Nonetheless, the US-Dutch relationship worked remarkably well.
In 2005, when Agilent decided to sell off its semiconductor operations, Philips acquired the US firm's share of the company. In that year Lumileds employed 1,760 people and earned $83 million profit on sales of $324 million. It was, Begemann boasted, the best investment Philips had ever made. "We paid $80 million for the original fifty percent, then five years later, Philips acquired the remaining fifty percent for eleven times that value [$950 million]. If you tell somebody in the financial world that you will multiply the value of your investment by a factor of eleven, you're a hero!"
Republished from May 2017 edition of L.E.D.: A History of the Future of Lighting by BOB JOHNSTONE (shown above). Excerpt from Chapter One, "Only One Girl To Dance With." Copyright May 2017; published by CreateSpace Independent Publishing Platform (available at: http://amzn.to/2x5R1p5). No revisions have been made to the original published text.
Origin of Haitz's Law
In early '99 Roland Haitz started to consider how much energy would be saved by converting incandescent lamps to LEDs based on two decades' worth of historical data he had in his files. "I'm somewhat of a pack rat," Haitz told me. "I dug through my old reports, which went back to 1971 or so. We used to have an annual review with the top brass at HP, when I had to present our accomplishments and performance targets for the next five years. So I had a lot of information about how many LEDs we'd sold in a year, what was the average selling price, and the efficiency at that time. I basically plotted a historical collection of points of what was the most powerful LED you could buy commercially in that year. Then I also dug into what was the cost per lumen."
Haitz was not a big believer in numerical analysis by spreadsheets. For him, a simple formula was a far better tool for understanding. The result was what came to be known as Haitz's Law. In graphical terms, it plotted two lines, price and performance, versus time; the former headed straight down, the latter straight up. Haitz's Law states that every decade the price per lumen falls by a factor of ten, while the amount of lumens output by an LED increases by a factor of twenty. Extrapolating the lines, Haitz estimated that by the middle of the next decade, that is, around 2005, LEDs would begin to become competitive with conventional light sources. He also worked out what the potential energy savings of LEDs would be. The numbers turned out to be huge. "I was shocked by how big they were," Haitz said. "And that's what convinced me: Dammit — we have to do something!"