L.E.D.

by Bob Johnstone

  The issue came to a head in March 2011 during hearings chaired by Jeff Bingaman on the proposed repeal at the Senate’s Committee on Energy & National Resources. As we have seen, Rand Paul led the attack, accusing the government of hypocrisy in being pro-choice when it came to abortion but anti-choice when it came to light bulbs. Paul was seconded by James Risch, the junior senator from Idaho. The latter asserted that, back home, people were “just astonished” by the federal government telling them what sort of light bulbs they could use. “I mean, where has this country gone?” Risch ranted, “It’s absolutely amazing.” Especially the mandate for putting “mercury bulbs” in every home in America. Bingaman did his best to clear up the confusion. He patiently explained that there was in fact no mandate for CFLs, the law merely set minimum efficiency standards, then left it up to manufacturers to determine which technologies to implement. While the hated CFLs were the target of most right-wing ire, LEDs also took a pounding. Especially when, in May 2011, Philips unveiled an LED light bulb which had an estimated life of 25,000 hours, but a price tag of $50. Fox News pounced gleefully on the announcement, running a story headlined: “As Government Bans Regular Light Bulbs, LED Replacements Will Cost $50 Each.” Only out-of-touch Washington bureaucrats could think that such an expensive light bulb was a good thing.

  In July 2011, following frantic efforts by Randy Moorhead and his fellow lobbyists, the Better Use of Light Bulbs act won a majority in the House but not the two thirds super-majority it needed to pass. (Republicans running for reelection in primaries against Tea Party rivals were understandably reluctant to sacrifice their careers by having to explain to angry voters why they had supported something many people perceived as a ban on their precious light bulbs.) Conservatives did however manage to exact a consolation prize. They passed mean-spirited legislation to prevent any funds appropriated for EISA being used to promote or enforce the new standards. In practice what that meant was no money was available to educate a confused public on the transition to energy-efficient lighting alternatives. But the Tea Party’s victory was fleeting. Once the law went into effect in January 2012, consumer anxiety seemed to fizzle. As in California the previous year, the sky did not fall, people were still able to buy lamps, there was no light bulb apocalypse. Back in 2007 it seemed unlikely that, come the deadline, LED manufacturers would have a sufficiently bright alternative ready to replace 100-watt bulbs in the marketplace. Or, more importantly, a solidstate replacement for the half-billion-odd 60-watt incandescent bulbs that were sold in the US each year. Thus far, the best LED makers had managed was a replacement for 25-watt bulbs. Incentives were clearly needed to speed up the pace of development. Happily, buried in the text of the Energy Independence & Security Act, as we shall see in the next chapter, was just such an incentive.

  C H A P T E R S I X

  And the Winner Is L ike many good ideas, the inspiration for what would subsequently become known as the L Prize came to Mike Carr while he was taking a

  shower. One of Jeff Bingaman’s senior senate staffers, Carr had been remodelling his house. In attempting to replace the halogen downlights in his ceiling with compact fluorescents, he had become frustrated, because “the CFLs just cooked, they just burnt right out.” Carr was aware that LEDs would eventually solve the efficiency problem, but not soon enough for him. Was there not some way of compressing the time it would take to develop and deploy LED lights, he wondered, of accelerating the shift to more energy-efficient lighting? Then the idea of offering a prize popped into his head.

  Carr had been fascinated to read about the X Prize, a $10 million incentive for the first non-governmental organization to launch a reusable manned spacecraft. “From my perspective the most fascinating thing was, a relatively small prize was able to generate spending by the private competitors well in excess of the amount of the prize,” he told me. This was back in 2006, when Carr and his colleagues were working on the Energy Independence & Security Act. Having run the idea past his fellow staffers, Carr pitched the prize to Bingaman. “I roughed out a memo to Jeff, said I think we should try to put [the prize] in the energy bill … and he thought it was a good idea.” The next step was to draw up challenging specifications that were based on trends in LED research. “We brought in the industry at that point and we basically said, We want to have specs that are not achievable today, but will be achievable in the next three to five years.”

  The bar was deliberately set high. The goal was to prevent the kind of quality problems that had bedevilled compact fluorescents. The specifications challenged industry to develop a perfect, like-for-like replacement for the A19, the classic pear-shaped Edison screw-in 60-watt incandescent that was the nation’s most widely-used bulb. The Department of Energy estimated that more than 425 million 60-watt incandescents were sold each year in the US alone, representing approximately half of all light bulb sales. Using less than 10 watts, the retrofits would have to generate slightly more light (900 lumens) than a typical 60-watt incandescent. Unlike CFLs, the bulbs would also be dimmable. They would last for more than 25,000 hours, compared to 1,000 hours for incandescents. In addition, the light produced by prize entries would have to be warm white and capable of rendering colors faithfully. And it would have to be omni-directional. The winning entry would receive a $10 million cash prize and considerable prestige in the form of bragging rights.14

  To encourage mass production and early deployment the act also contained another carrot. As part of a mandate to slash its energy consumption by 30 percent, the US General Services Administration was instructed to replace all of its incandescent lighting with bulbs of equivalent specs. The GSA manages over 8,000 buildings nationwide. “Somebody told us how many bulbs are in government buildings, and it was some jaw-dropping number,” Carr explained. “We said to the industry — that’s the big prize you’re gonna get.” It would not be compulsory to use the prize-winning bulb, however, the intention also being to gee-up other lamp makers who did not enter the contest. “So then you’re still gonna have competition that’s gonna pull the price down,” Carr said. “The other thing is, GSA purchasing regs favor domestic production, so it actually pulls the manufacturing to the US.” Beyond government procurement loomed the mass market.

  14 The act also called for two other prizes to be offered. One was for a replacement for the PAR 38 halogen downlight, the other for a “21st Century Lamp” fixture. But in January

  2011, no entries having been received, the former was deemed premature and put on hold. The latter has yet to be announced. The DoE also runs another competition, Next Generation Luminaires, for commercial fixtures used in indoor and outdoor applications. Awards are given annually to multiple winners.

  Of ficially known as the Bright Tomorrow Lighting Prize, though most people called it the L Prize, the competition was launched at LightFair, the industry’s US showcase, in May 2008. The previous month at Light + Building, the equivalent trade fair in Europe, Philips had announced its first retrofit LED product. This was capable of delivering 250 lumens, just over a quarter of the L Prize requirement. It was intended as a replacement for 25-watt bulbs, which are mostly used in small decorative applications, like lava lamps.

  At Philips Lighting’s headquarters in Eindhoven, publication of the terms and conditions of the L Prize competition produced precisely the kind of galvanizing effect that Mike Carr had hoped for. In particular, it caught the attention of two ambitious young research managers, Simon Kuppens and Rob Tousain. Kuppens was a physicist by background. He had studied as a post-doc in Boulder, Colorado, before returning to his native Netherlands to join Philips. After working at Lumileds on backlights for mobile phones he had been promoted to project leader on LED applications such as flashlights for cameras. “When the L Prize came by,” he told me one afternoon as the three of us sat chatting in the laboratory library at Eindhoven, “I grabbed it.” The prize was, Kuppens said, “the coolest project you can imagine. Someone puts the bar up there and everyone says, It cannot be d
one. The R&D guys go back to the lab and they start thinking, Well, maybe it can be done. Then management says, OK, let’s try it, let’s see how far you can get — and that’s fabulous. So we wrote a letter to the Department of Energy and said, We’re entering.”

  Tousain’s exposure to lighting had been shorter. He was one of those at Philips who had been excited to hear about the coming LED lighting revolution. Ready for a change, he embraced the opportunity to move into lighting without the encumbrance of much prior knowledge about conventional approaches. “One part of it was, we are the leading company in lighting,” he told me, “and the L Prize was an excellent opportunity to demonstrate that to the outside world. The other part was, to prove to ourselves that we could do it. We have learned over the past years of doing pre-development innovation in this era that you have to set yourself stretching targets, because things will always go faster than you anticipate.” They wanted to push themselves into “taking a leap that went beyond the evolutionary product development that we were already doing.” Added to which was the lure of lucrative government contracts: entering the L Prize contest was potentially a good business opportunity. “We’re a company,” Kuppens said simply, “we make money.”

  “Of course,” Tousain readily admitted, “nobody had a clue how to do it.” The company had only just launched its state-of-the-art, 250-lumen, 25-watt bulb. Now here was this spec that called for almost four times the amount of light, in the same form factor, using less than forty percent the power. Plus, the bulb had to be dimmable, and shine light omnidirectionally. It was, as Kuppens said, “a challenge on all fronts.” Tousain agreed: “We really had to leap-frog on all disciplines.” Taking on the challenge would mean leveraging all the core strengths that a large multinational corporation like Philips could muster. At its peak the project would employ more than a hundred researchers and support staff spread across five facilities located in the US, Holland, and China. The bulk of the team worked at Lumileds, Philips’ chipmaking facility in San Jose.

  The first priority was to identify really bright LEDs that could produce the required 900 lumens. Light emitting diodes are made by depositing layers of chemicals - a process known as epitaxy - on round wafers made of ultra-pure compound semiconductor material. The processed wafers are diced up into chips which are then packaged so that they can be hooked up to power sources. Performance varies from batch to batch and across individual wafers. To build the prototype L Prize lamp, the brightest diodes were culled from the Lumileds lab’s private stash. But to meet the conditions of the L Prize, Philips had to submit 2,000 sample lamps. That meant the lab could not continue to rely on cherry-picking the best chips. To increase yields of high-output devices the researchers had to beef up epitaxial production “using everything we had cooking in our R&D facility.”

  The design Philips came up with mounted 24 LEDs, mostly blue, but including a few red ones to make the light look warm. To modify the output from blue to white, a yellow phosphor is added. Normally the phosphor is poured in liquid form directly onto the chip, a process known in the industry as “goop in a cup.” But this approach results in optical losses because much of the light is reflected back into the chip. For their bulb, Kuppens and Tousain chose a different approach, called “remote phosphor,” a technology developed by Philips that is used to coat the inside of fluorescent tubes. They took advice from the company’s phosphor specialists in Aachen, a town just across the German border about an hour’s drive from Eindhoven. The result was a curved outer shell made of translucent plastic in which yellow phosphors were embedded. Using remote phosphors also helped reduce thermal load. In addition to how to make lots of light, how to get rid of the resultant heat was another problem that needed to be solved. “The key challenge in LED lights is to keep the electronics cool enough,” Tousain explained, “because often it’s not the LEDs determining the lifetime of the product but the electronic components that you use.”15

  One of the L Prize’s trickier specs was that the light had to shine omni-directionally, just like an incandescent bulb. Until then most LED bulbs had featured an ice-cream cone shape. The “scoop” covers a flat board on which the chips are mounted; the cone underneath is the metal heat sink. But in such designs, the light can only shine above the cut-off between scoop and cone. This is not acceptable in a table lamp, for example, where you want light to shine down as well as up. To get their light to radiate in all directions, the LEDs would have to be stacked vertically. This caused a problem with dissipating the heat. “So what we did is, we had to play a trick,” Kuppens said. Into the bulbous yellow plastic they carved four deep metal-lined slots, to enable cooling air to flow through to the LED stacks. The team was also able to draw on recent research done in Boston by Philips’s recent acquisition, Color Kinetics, with a grant from the DoE. “One fundamental piece of this,” commented Kevin Dowling, the principle investigator at Color Kinetics, “was using numbers of smaller low-power LEDs rather than larger high-power ones. This had the benefit of reducing the heating load and increasing efficiencies.” For the light to shine downwards, the stem of the bulb needed to be narrow. That in turn left only a very small space for the driver electronics.16 For help in solving this problem, Kuppens and Tousain turned to the expertise of their colleagues at Philips Lighting Electronics in Rosemont, Illinois, northwest of Chicago, a facility which worked on developing ballasts for fluorescent lamps. “We were able to build on their knowledge,” Tousain said, “to make efficient, low-cost electronics.” Much of the actual LED lamp product development was done in Shanghai, where Philips maintains a competence center for reliability and quality, skills needed in order to manufacture 2,000 lamps.

  15 Unlike incandescents, which waste most of their output in the form of finger-burning infrared heat, LED lamps are cool to the touch. However, like all electronic devices, they do generate heat internally. (The amount depends on the efficiency with which they convert electricity to light. The efficiency of today’s devices is around fifty percent, but researchers reckon that by improving materials it should eventually be possible to push the figure up to around ninety percent.) The heat has to be removed via a process known as thermal management. In practical terms, this mostly means attaching to the back of the LEDs a metal heat sink which has fins that conduct and convect the heat away.

  With facilities located around the world, work on the L Prize bulb continued round the clock, seven days a week. “It really was a global team,” Kuppens said, “because when I entered the office in the morning, Shanghai had already finished, and in the afternoon I got updates from Lumileds.” From start to finish the project took just sixteen months. In late September 2009 Philips delivered 2,000 sample bulbs to the Department of Energy. They also submitted product specifications, and a commercial manufacturing plan. It called for the bulbs to be assembled, with chips made in San Jose, California, at a plant near Milwaukee, Wisconsin. Then the testing began.

  16 Like the electronic ballasts in fluorescent tubes, driver circuits control the power supplied to the LEDs. Drivers convert the incoming alternating current (AC) to direct current (DC). They also handle dimming.

  In the Department of Energy’s determination to meet the letter of the law (and avoid another CFL-style debacle), the samples would be the most rigorously tested bulbs ever. The first step was to take 200 of the samples - a statistically valid representation - and dispatch them for standard photometric testing at two independent labs, one in Boulder, Colorado, the other in Atlanta, Georgia. There, the bulbs were placed first inside a large hollow spherical instrument known as an integrating sphere in order to verify the amount and color of light they produced and then in a splendidly-named gonio-photometer (from the Greek gonia, meaning angle) to measure its dispersion. When they came back, the same lamps were inserted into a custom-made long-term light maintenance test-bed - essentially a big box - at the DoE’s sprawling Pacific Northwest National Laboratory in Richland, Washington. At an elevated temperature (45 degrees C), the bu
lbs were left on for around six thousand hours, or almost nine months of continuous usage. Data from this test was then mathematically extrapolated to assure that it was possible to achieve the required 25,000 hours of life.17 To identify potential failure mechanisms, the lamps were subjected to a series of stress tests. Batches of bulbs were baked in ovens, frozen in freezers, sweated at high humidity, shaken on a vibration table, and subjected to extreme voltages and currents. Others were switched on and off constantly, while still others were zapped with electro-magnetic interference. “Those tests were nasty,” a DoE official chuckled, “I mean, we really put it to ‘em.”

  17 On April 29 2014, the test lamps passed 25,000 hours of continuous operation. Not only had none of the 200 samples failed, they were still producing one hundred percent of their initial light output. By August 17 2016, a selection of 31 lamps had passed 50,000 hours and were still going strong, with not a single lamp having failed over the course of testing. Average light output for the lamps was 93 percent.

  In addition to in-house testing, the department also sent out samples to assess how the bulbs would perform in the real world. For this purpose 31 partners were recruited. They included utilities and regional energy-efficiency organizations. Based on the proposals they submitted fourteen partners were selected. In summer 2010 each received 100 bulbs. More than forty different test locations were chosen. In addition to homes, they included hospitals, hotels, community centers, coffee shops, and grocery stores. During the next few months, data was gathered on metrics such as quality of light. Samples were compared with conventional alternatives. Feedback was solicited using a standard questionnaire from homeowners, occupants, and users of the spaces. That fall, the partners wrote up their results and sent them in. Then these and the results of the lab tests were put before a technical review committee consisting of eight experts.