The Future of Light Is the LED

The heir to the inefficient incandescent bulb isn't the mercury-laden CFL. Instead, it's a liquid-cooled creation that's radically different from anything ever screwed into a socket. The future will be lit by LED.
Photo Yasu  Junko
Photo: Yasu + Junko

Brett Sharenow is presiding over the Pepsi Challenge of lightbulbs. The CFO of Switch, a Silicon Valley startup, Sharenow has set himself up in a 20-by-20 booth at the back of the Pennsylvania Convention Center in Philadelphia, and he's asking passersby to check out two identical white shades. Behind one hides a standard incandescent bulb, the familiar lighting technology that has gone largely unchanged since Thomas Edison invented it 132 years ago. Behind the other is a stunning, almost art- deco-style prototype that holds 10 LEDs and a secret fluid. It's a liquid-cooled bulb, as radically different from Edison's invention as anything that's ever been screwed into a standard socket and, Sharenow hopes, the next big thing in the $30 billion lighting industry. The challenge: Can you tell which is which?

It's day one of Lightfair, the annual international trade show for everything that glows, glares, flickers, or shines—500 exhibitors and 24,000 visitors prowling row after row of light after light.

The Future of Light Is the LED
by Dan Koeppel (44.8 MB .mp3)Subscribe: Wired Features PodcastThis is the last Lightfair before new regulations governing lightbulb efficiency begin to take effect in the US in January, and there's a real sense of history and urgency on the show floor. Ready or not, the way we light our homes and offices is about to change, and the technology that will lead the way is somewhere in this hall.

If all goes according to plan, the provisions of 2007's Energy Independence and Security Act will effectively ban 100-watt incandescents starting in 2012. Seventy-five-watt bulbs will depart in 2013, followed by 60- and 40-watt lamps a year later. So the race to find a suitable replacement technology is coming down to the wire. The industry is banking on LED lighting as the way forward, and it's virtually the only bulb technology on display: There is barely a single incandescent or sickly compact fluorescent to be seen. Just 200,000 square feet of companies racing to fill their share of the world's billions of standard sockets—and betting on LEDs as the way to do it.

There's an excellent reason LEDs have taken on the aura of inevitability: LEDs are semiconductors, and like all solid-state technology, they are getting better and cheaper on a predictable curve. In 1999, a researcher named Roland Haitz, then heading up semiconductor R&D at Hewlett-Packard, coauthored a paper that became the lighting industry's manifesto. By charting the historical prices of LEDs and projecting forward, Haitz estimated that the amount of light they produced would increase by a factor of 20 per decade, while the cost would correspondingly drop by a factor of 10.

Haitz's law has proven remarkably accurate. But the lighting industry still has major hurdles to clear before LEDs gain acceptance by consumers. Beyond the very real technical issues—cooling, costs, light color—there's the public's lingering distaste for compact fluorescent lamps, which failed miserably in their projected role as bulb of the future. That sentiment has fed into a Tea Party-fueled backlash against the new regulations, and there have been attempts in Congress to roll them back entirely.

The reasoning behind the lighting provisions in the Energy Independence and Security Act is pretty straightforward: Incandescents convert less than 10 percent of the energy pumped into them into light, losing the rest as heat. More-efficient bulbs could save billions of dollars, decrease dependence on foreign oil, and significantly reduce greenhouse gases.

Still, the consumer backlash resonates, and not simply because CFLs are horrible, flickery, ugly, and unreliable. Evolutionary biologists believe that human lighting preferences are the result of our trichromatic vision—rare in nonprimates—which makes us particularly suited to daylight and the perception of primary colors. There's an anthropological component as well: For 400,000 years, humankind has been banishing darkness with fire. And Edison's bulb is, at its core, a burning filament that casts the glow of a flame. Abandoning incandescent bulbs means abandoning fire as our primary light source for the first time in human history.

As with all trade shows, the companies and booths at Lightfair become progressively smaller and more patchwork the farther back you go. The front of the hall is the province of the industry's big three—Philips, Osram Sylvania, and General Electric, founded by Edison himself. The behemoths give way to a second tier occupied by the likes of Toshiba, Samsung, Leviton, and Honeywell. Behind them, smaller companies—accessory makers, vendors, and Asian component suppliers—fill out the rest of the available floor space, right up to the back wall. Here, within smelling distance of the hot dog stand, Sharenow is presiding over the Switch booth.

"Which one is which?" Sharenow asks. Most people get it wrong, which is significant. When they select what they feel is the superior light, they're selecting the Switch.

Though there are countless 60-watt LED prototypes in evidence at Lightfair, few are available for side-by-side comparisons with incandescents. Most are sealed inside display cases or exist only in press releases or PowerPoint presentations. The Switch bulbs are out in the open: crystalline shells of clear, faceted glass, with aluminum prongs holding yellow-tinted LEDs that shine through the cooling liquid to cast a warm, living-room-quality glow—and draw only 13 watts to do it.

About 80 percent of all bulbs sold in the US are rated at or equivalent to 60, 75, or 100 watts, but nearly every LED brought to market so far has come in the 40-watt-equivalency range, good for little beyond closets and crawl spaces. Brighter bulbs are either too hot, too expensive (upwards of $50, though they can last for 20 years), or both. Switch plans to start selling its 60-watt-equivalent bulb this fall for around $30.

"Light is something people have always needed," Sharenow says. "This is a light source people will want."

Switch uses a liquid-filled dome to keep the LEDs from overheating. Photo: Misha Gravenor

Misha Gravenor

Though fluorescent lighting has been around since the 1930s, GE physicist Ed Hammer was the first, in 1975, to figure out how to twirl that long gas-filled tube so it would screw into a standard socket. (His twisty tube was a major achievement, so tricky to produce that many CFLs are still curled by hand.) The GE CFL produced as much light as an incandescent, while drawing just 20 percent as much electricity. But CFLs couldn't be dimmed, they were fragile, and they produced a light that had all the warmth and appeal of skim milk. Rather than commit the funds to bringing a new technology to market, GE shelved Hammer's design.

HOW TO BUY A LIGHTBULB IN 2011

These days, a browse through the lighting aisle at Home Depot is likely to leave you baffled. Here are the terms you need to know. —D.K.

  • LUMENS
    The amount of light a bulb produces. Depending on which government agency you ask, this is "brightness" or "light output." Your reference point: A standard 100-watt incandescent produces about 1,700 lumens.
  • WATTS
    Not a measure of brightness; instead, it's a measure of how much energy a bulb consumes to reach its claimed brightness.
  • WATT-EQUIVALENT
    Since we've conflated watts and brightness, it's easier to talk about bulbs in terms of watts. So if a 100-watt incandescent produces 1,700 lumens, and a 20-watt LED does the same, the LED will be sold as a 100-watt equivalent.
  • EFFICACY
    The number of lumens a bulb produces for each watt it consumes. The higher the number, the more efficient the bulb. A good number for incandescents is around 18, CFLs around 60, and LEDs around 54.
  • BULB LIFE
    LEDs dim over time. They're considered effectively dead when they produce no more than 70 percent of their original brightness. For LEDs, this lifespan is given in hours or years, the latter an estimate based on three hours of daily use.
  • ENERGY COST
    Based on an assumption of three hours of use per day at 11 cents per kilowatt-hour. For a 60-watt incandescent, it's just over $7 per year. CFLs and LEDs both come in at about $1.50 per year.
  • COLOR TEMPERATURE
    Expressed in degrees Kelvin, this is how we measure things like soft white or daylight. A pleasant soft white will have a color temperature of 3,000 K. White light ranges from 4,100 to 6,000 K, roughly equal to midday sun. Higher numbers get increasingly blue.

In 1980, Philips became the first company to market a CFL, with a design based on a series of bends rather than GE's spiral. But with the 1970s energy crisis over, there was little interest. It wasn't until the 1990s that CFLs gained traction, driven by concerns about energy efficiency. GE finally put Hammer's spiral into production. And by the mid-2000s, CFLs stood package to package with incandescents in store lighting aisles. At first, consumers seemed to embrace the technology, partially because of the advertised energy savings and partially because of massive public bulb-distribution programs. The Los Angeles Department of Water and Power distributed 2.4 million CFLs in 2009, predicting that their use would save 131,000 metric tons of CO2 emissions annually, equivalent to removing 24,000 vehicles from the road.

But problems of dimming, flicker, and light color remained. And manufacturer claims of bulb lifetime and quality were wildly exaggerated. Consumers didn't understand that CFLs burn out rapidly when they're not allowed to rest at least 15 minutes between being cycled off and on, or that they'd overheat and fail if they were used in recessed ceiling canisters; they weren't told that there was no standard definition of what a 60-, 75-, or 100-watt incandescent replacement actually was.

"There were products out there that claimed whatever they wanted," says Ed Crawford, general manager of lighting systems for Philips North America. "There were too many promises." Though industry claims that some of the problems resulted from an early influx of cheap imported bulbs are true—as is the fact that CFLs have improved—many issues remain. No-name commodity bulbs still perform poorly, and even the big brands continue to poison the well by selling, for example, CFL floodlights, which often wind up in outdoor motion-sensor systems. This is an absolutely inappropriate use, given how fast the bulbs expire when they have to flick on and off so quickly. The final blow came when consumers learned that CFLs contain mercury, enough that the Environmental Protection Agency's cleanup instructions for a broken bulb run three pages and start with a warning to open windows and evacuate people and pets.

In 2008, as part of the same law that mandated the virtual ban on incandescents, the US Department of Energy began writing the CFL's obituary by launching a competition intended to jump-start the LED transition. The L Prize (short for Bright Tomorrow Lighting Prize) offered $10 million to the first company to come up with a 60-watt-equivalent bulb that met a series of standards designed to avert a CFL-like debacle. (Additional categories will be announced in the future.) It would draw no more than 10 watts of electricity, be truly dimmable, flip on and off without flicker or hesitation, and emit a pleasant light. It would also have to fit into a standard socket and last at least 20 years.

The money wouldn't cover the development costs of such a bulb, but the winner would get name recognition and the inside track on lucrative government contracts. On August 3, the DOE announced that Philips—the only company to even submit an entry—had won the competition. The decision came after almost two years of evaluation.

It's possible to create a bulb that meets the L Prize's 10-watt benchmark, and it's possible to create one that consumers can afford. But it's not possible to do both, according to Sharenow, who says the costs associated with developing an L Prize bulb didn't make sense for a startup with limited funds. So Switch decided to skip the L Prize in favor of concentrating on a consumer product. In fact, while Philips has come to market with a 60-watt equivalent for $40, its L Prize winner was a separate, competition-specific design.

The Switch bulb will cost $30. That's still not cheap. A decent 60-watt incandescent costs less than a dollar. But the simple fact is that coming up with a truly worthy LED bulb is enormously complex, requiring expertise in physics, chemistry, optics, design, and manufacturing. Since nobody has built such a multidisciplinary lighting product before, approaches have varied wildly. So have results.

As with personal computers, photography, and mobile phones, the massive shift that LEDs represent may open the door to fast-moving entrepreneurs. That dynamic seems to be emerging in the form of companies like Switch and Cree—a North Carolina LED maker frequently cited by President Obama for its homegrown innovation—and with the increasing appearance of solid-state lighting programs at universities like UC Davis and the University at Albany-SUNY.

In the same way that Tesla Motors leapfrogged the superpowers of the automotive industry in the race to develop viable and exciting electric cars, so might a nimble startup take the lead in the transition to solid-state lighting. That's what Alan Salzman, CEO of VantagePoint Capital Partners, is hoping. The company—which provided funding for Tesla—won't give exact numbers but admits to an eight-figure investment in Switch. "You'll hear it over and over," Salzman says. "When an industry transforms, some of the incumbents adapt, but a lot don't. That's what's happening in lighting."

Individual diodes sit in vials of cooling fluid for testing at the Switch facility in San Jose, California. Photo: Misha Gravenor

Misha Gravenor

GE invented the light-emitting diode in 1962. The first ones to come into wide use—glowing a space-age red—turned up in the clock radios, pocket calculators, and digital watches of the 1970s. Additional colors came along over the next couple of decades.

LEDs are manufactured more or less like any other semiconductor. Each diode is cut from a wafer of crystals layered over a base of silicon or sapphire. The crystal layer on early LEDs was gallium arsenide or gallium phosphide, which lent that reddish color. Additional colors and increased brightness required more nuanced control of layer composition and depth. Modern LED makers accomplish this by using precise ratios of indium, gallium, aluminum, and nitrogen for the crystal layer, which results in a bluish color.

But on their own, not even advanced LEDs can produce anything suitable for the living room. The blue-tinged illumination is fine for, say, a pen flashlight on a keychain, but it doesn't come close to the warm light the human eye desires.

There are two ways LED makers create a more pleasant white. In the 1990s, the favored technique was to combine red, green, and blue LEDs. But they all have differing efficiencies and operating requirements. Heat management, power supply, and drivers—the bulbs' controlling circuit boards—get more complex.

So the LEDs found in current household applications are blue diodes daubed with a powdered coating called a phosphor, which includes rare-earth elements that filter blue light. The phosphor is generally yellow, and depending on the composition of the phosphor and the ratio of unconverted blue light, the resulting "white" light can range from the warm glow preferred for home use to cooler tints more suited to, say, retail and outdoor use.

Though still rather expensive to produce, LEDs are getting cheaper, just as Haitz's law predicted, due to both technical advances and economies of scale. They're also getting brighter, which means manufacturers can use fewer of them per bulb, further driving down costs. Things are advancing so quickly, in fact, that Haitz, now retired, thinks his law will soon expire. Sometime around 2020, he says, bulbs built for our conventional infrastructure will max out, since at some point additional brightness from a single socket would be overkill. "We'll reach a limit where we can only move forward by producing quantities of luminous flux that nobody needs," he says. After that, progress would continue only if the Edison socket disappeared and the world moved away from bulbs altogether—most likely to flat-panel LEDs, Haitz says.

Before we get to that point, or even to viable, affordable home LED bulbs, engineers have to solve a pair of challenges: cooling the diodes, and the shape of the light. Cooling is essential because hot diodes don't last long. Also, the cooler they are kept, the more wattage they can handle, which translates into more light per diode. Getting an LED to cast a light in a shape that will properly illuminate a room is perhaps an even more difficult challenge. LEDs are point lighting sources; they shine in only one direction. That makes them fine for floodlights or traffic signs. But for LEDs to replace the Edison bulb, the shape of light they emit has to be round.

Once you've got that, you also need a driver to control the individual diodes, which means additional circuitry and firmware. So far, putting all those elements into a single, affordable package has proven to be an immense challenge.

The basis of the company that Ron Lenk called SuperBulbs rests in a 2007 patent filing. In it, Lenk—an MIT-trained physicist who was working at a semiconductor company—described a new technology that might solve the heat and shape problems of LED bulbs. The bulb would "use a thermally conductive fluid, gel, or plastic" to "act as the means to transfer the heat power generated by the LEDs." The substance might also "contain materials dispersed in it to disperse the light."

Working out of their home in Redwood City, California, Lenk and his wife, Carol, an electrical engineer, began constructing prototypes using raw LEDs attached directly to a power source and mounted inside baby food jars. It took two years for the couple to arrive at what they believed was a working coolant, a gel similar to the superabsorbent polymers used in disposable diapers. (Lenk, who speaks with a scientist's nonmetaphoric precision, scoffs at the suggestion that there might be a connection between the baby food jars and diaper materials and the fact that Carol gave birth to their third child in the middle of their bulb experiments.)

The Lenks incorporated SuperBulbs in 2007. Drawing on their personal savings, they built working prototypes and hired a VP of marketing, Ethan Thorman, to help get them off the ground. Thorman suggested a consultation with a member of his theater group, Brett Sharenow.

Thorman and Sharenow had appeared together in an independent film called Pathogen. But Thorman knew that Sharenow was also a trained chemist with an MBA who had spent two decades as a consultant to tech firms like Pacific Bell and Verio. At a meeting in Oakland, Sharenow asked Lenk to pretend he was meeting a venture capitalist. After 15 minutes, Sharenow stopped him. "Are you interested in honest feedback," he asked, "or do you want to see your company vanish?"

Over the next four hours, Sharenow offered a brutal critique. "There was no business model," he recalls. "No financial assumptions, no way that even a revolutionary product could win in a market locked up by three huge companies."

Sharenow began working with the Lenks in February 2008. In May, he took a prototype to Las Vegas. There he met with George Abittan, whose company, Advanced Lighting & Electric Service, supplies lighting products to hotels and casinos. Abittan led Sharenow to a table with three bulb fixtures mounted on it. One held an incandescent bulb, another a CFL. The third was for Sharenow's LED.

The SuperBulbs light was so good, Sharenow says, that Abittan didn't believe it actually used solid-state technology. He attached a meter to the lead supplying power to the bulb to check. The measurements confirmed that it was drawing just 10 watts of power but producing the equivalent of 40 watts in incandescent light. If a product like that could be mass-produced, Abittan said, every casino in Las Vegas would buy in.

Sharenow and the Lenks had already been through several rounds of talks with VantagePoint, the venture capital firm run by Alan Salzman, whose headquarters were just a 10-minute drive from SuperBulbs' converted condominium offices in Redwood City. On the heels of their successful trip to Vegas, they finally got the funding they were after. Besides Tesla, VantagePoint had invested in the first company to build an LED-manufacturing facility in the US in more than two decades, and in BrightSource, a company whose solar energy plants are now being tested in the Mojave Desert. VantagePoint was keen to expand its energy portfolio and decided to invest in SuperBulbs. "We had looked at what was out there," Salzman says. "Bulbs that cost $80, with 64 LEDs. Not a lot's going to come of that. These guys were doing solutions."

Sharenow and the Lenks now had backing, a business plan, and a working prototype. But prototypes aren't products. And the SuperBulb, as a product, was turning out to have some problems.

BATTLE OF THE BULBS

Our panel of testers assessed two LEDs, two CFLs, and two incandescents—all 60 watts or equivalent—to separate the light from the hype. —D.K

INCANDESCENT

  • Philips Soft White 57 Watt
    This is part of a new wave of incandescents that come in at slightly lower wattages than their claimed equivalents (to comply with California law requiring greater—but not great—efficiency). Light quality was best in test, with the classic softness that has made it so hard for us to quit incandescents. But it did seem dimmer than a full 60. And it's rated for only 11 months at three hours per day. WIRED: Price is right. Natural light color. TIRED: Inefficient. A bit dim.
    $2 for four. 6/10

- Feit Electric Vintage Style Carbon Filament Bulb
This replica of an early Edison product wins points for aesthetics, with its elongated shape and ribbonlike loops of carbon filament. It's perfect if you've got lots of antiques or live in a restored Victorian—or if you're a retro-grouch /hipster. It casts a beautiful, warm light, just not very much of it; we rated it dimmest in our test. WIRED: Awesome throwback chic and natural hue. TIRED: Generally sells for three times suggested retail.
$6 6/10

CFL

- GE Energy Smart 13 Watt
One of the most popular CFLs on the market, this twister from GE is Energy Star rated, meaning that its claims of lifetime (8,000 hours), brightness (825 lumens), and color temperature (an incandescent-like 2,700 Kelvin) have been independently tested. We rated it the second-dimmest bulb in our roundup, but the excellent soft-white light was the best we've ever experienced from a CFL. WIRED: Outstanding light color. Long lasting. TIRED: Undimmable. Like all CFLs, it contains mercury.
$4 6/10

  • Ecosmart Daylight A19 60-Watt Equivalent
    Despite a worst-in-test 700 lumens (claimed) on a 14-watt draw, the EcoSmart didn't seem particularly dim. And the globelike casing around the coil gave this CFL a nice, even glow. Too bad the light was such a disagreeable blue-gray that we couldn't wait for the test to be over. It felt like being stuck in a Darren Aronofsky movie. WIRED: The rare bulb that appears brighter than advertised. TIRED: Daylight in name only. Undimmable. Contains mercury.
    $9 for two. 2/10
LED
  • Philips Ambientled
    The first commercially available 60-watt-equivalent LED, this striking bulb has double the life (a claimed 15 years) of a comparable CFL. Our panel ranked it at or near the top for brightness and color of light (a nearly incandescent soft white), though opinions were mixed on its yellow color when the 8-ounce bulb is turned off. WIRED: First-to-market bragging rights. Dimmable. Superb light and longevity. Conversation piece. TIRED: Expensive. Slightly odd shape means it may not fit in certain light fixtures.
    $40 8/10

  • Switch60 Warm White
    The clear winner in terms of aesthetics, the Switch was also the brightest bulb in our test. The warm light it cast was comparable to any incandescents we've seen, but it was hotter to the touch than any of the others—a function of the liquid cooling that transfers the heat up through the glass—and, at 10 ounces, the heaviest. WIRED: 20,000-hour lifetime (claimed). The choice for design nerds. Surprisingly bright. TIRED: Hot and relatively heavy. Ten times heavier than an incandescent.
    $30 8/10

Illustration: Kate Francis

Brett Sharenow with the Switch75. Photo: Misha Gravenor

Misha Gravenor

In February 2009, David Horn, a physicist and engineer who'd trained at California's Lawrence Livermore National Laboratory, joined SuperBulbs as CTO. Horn had worked with VantagePoint-backed companies before, consulting with several solar startups that the firm had added to its portfolio. His first act at his new job was to test the existing 40-watt prototype, which at that point was all SuperBulbs had developed. What he discovered was disappointing. To shake up the industry, SuperBulbs' designs would need to be the equivalent of 60 watts and higher. But when Horn touched the glass part of the bulb, he felt no heat. "The gel wasn't working the way it needed to," he says. The cooling wasn't sufficient to scale up to a 60-watt equivalent.

In the Lenk design, gel fills the glass shell, transferring the heat from the LEDs outward, through conduction. A bulb cooled this way should have a dome that's hot to the touch. That the SuperBulb wasn't suggested to Horn that the gel wasn't an effective coolant, which would mean nearly all of the bulb's thermal energy was being dissipated via the aluminum heat sink at the bottom. Heat sinks work—they're standard on most LED designs—but the idea with the gel was to find something more efficient, since a better coolant would allow more power to be pumped through each LED. That would mean fewer LEDs per bulb, which is how SuperBulbs would hit the lower prices at the center of the company's business plan.

A series of tests confirmed Horn's suspicions. The gel wasn't doing the trick. So he began experimenting with liquids. As a free-moving liquid heated up at the center of the bulb where the LEDs were, it would rise to the glass, transfer the heat outward, cool down, and return to the base of the bulb—a process known as passive convection.

The liquids worked, but the gel retained important advantages. It was nontoxic and inexpensive, and it had wonderful diffusion qualities that the liquids couldn't match.

Lenk believed the gel offered superior enough cooling to produce a commercially viable 40-watt equivalent. "I wanted to get to market quickly," he says. Even at a lower output, they would still have a revolutionary product cheaper than anything any other company had managed to produce. "If we changed to liquid," he argued, "we'd be starting from scratch." But Sharenow and Horn didn't believe a lower-brightness light could produce the market results they were after.

Meanwhile, SuperBulbs was nearly broke. VantagePoint's initial investment had allowed the company to grow to 12 employees. Now half had to be laid off. Horn reported back to the capital firm that the product still had a chance, but only if it got new money to fund research into finding a liquid coolant less viscous than the gel.

The Lenks left the company in late 2009 (though Carol worked as an adviser for a few more months) and moved to Atlanta, where Ron founded a firm that makes LED replacements for circular overhead fluorescents (the product is not liquid-cooled). "We were tired of the exhausting pace and never getting to see our children," Lenk says. The couple remains on good terms with Switch and, as shareholders, stand to profit if the company succeeds. But SuperBulbs, now just Sharenow, Horn, and four employees, dropped into what Sharenow calls "stealth mode."

In September of that year, as Horn continued searching for the ideal coolant, Philips announced its entry in the L Prize competition. Then, in December 2010, the world's biggest lighting company introduced a slightly less energy-efficient version of its L Prize candidate, the $40 AmbientLED (or the EnduraLED in some markets), for consumers. For now, it is the only 60-watt-equivalent LED that people can actually buy.

The Philips bulb looks nothing like the Switch. While the Switch seems like it could have been plucked from the set of Fritz Lang's Metropolis, the AmbientLED has an industrial, almost steampunk aesthetic. A cast-aluminum heat sink extends up from the socket threads, taking up almost half of the product's 4-inch height. It's also more conical than traditional incandescents, but what's most striking about the Philips bulb is that it has no bulb.

The eponymous shape of the product Thomas Edison invented in 1879 was no accident. It was designed for maximum heat dissipation. But the flattish top of the Philips bulb, while warm to the touch when lit, isn't a major part of the product's thermal management system. That job falls to the heat sink and the three clawlike extensions rising up from it. Fitted between these extensions, a trio of bright yellow plastic panels diffuse the raw blue light and give it a pleasant, incandescent-like tone.

The end result looks so different from what we've come to expect of bulbs that every AmbientLED package carries a large sticker: WHITE LIGHT WHEN LIT.

Part of the reason the L Prize judges took two years to declare Philips the winner is that LED bulbs are so novel. Though the performance standards are clear, they're also incredibly high, and devising ways to test them remains elusive. "This category of product works so differently than incandescents or fluorescents," says Kelly Gordon, a program manager with the Department of Energy's Pacific Northwest National Laboratory, which coordinates L Prize evaluations. "Test methods need to be entirely different." Consistent, repeatable tests will be essential, however, if the industry is to avoid the free-for-all of claims that bedeviled the CFL.

Gordon breaks the protocols into four categories: longevity, color, amount of light, and shape of the light. An L Prize contestant is required to submit 2,000 bulbs, a sample big enough to ensure sufficient bulbs for testing.

The biggest hurdle is longevity. LEDs don't simply burn out. Instead, they fade. The current standard, called L70, refers to the point at which an LED is able to generate only 70 percent of the light it originally produced. The L Prize criteria require a minimum of 25,000 hours before L70. That's roughly three years of continuous service, more than two decades if the bulb is used for three hours daily, about the national average. "Nobody has that kind of time," Gordon says.

Switch has been doing its own longevity testing. To simulate years of usage, bulb makers rely on heat. High temperatures accelerate the decay of LEDs to the point that their lifespans can be more easily estimated. The DOE's L70 protocol, which Switch uses, calls for 6,000 hours (250 days) at 113 degrees Fahrenheit. Switch's offices contain a testing area filled with what looks like a series of oversize bread ovens with tinted windows, behind which glow dozens of bulbs. The effect is like a room overloaded with votive candles. In another area, a warning—danger: extremely bright white light source. do not enter enclosed area without dark glasses—is taped to a sheet of blackout canvas.

"Our first prototypes lasted only a few hours," Horn says while looking over the oven room. "We'd leave lights burning on Friday and come in Monday to find them dead." Over seven months, he tried more than 200 liquid coolants, from mineral oil to antifreeze-type glycols. None had the right mix of consistency, diffusion, and price.

Horn began mixing earlier coolants together, trying to see if a blend, rather than a single base substance, would do the trick. In October 2009, he hit on a working brew. Switch won't reveal what's in the proprietary mix, other than to say that it's cheap and environmentally benign. "You can clean it up with an ordinary paper towel," Horn says.

The rest of the final design—the heat sink, drivers, and shape—came together relatively quickly, he says. The bulb's 10 diodes, each with a yellow phosphor, sit along a halo of aluminum fingers. That ring of light contributes substantially to the bulb's design appeal, but it also helps create a very incandescent-like light shape.

Today, Horn says, there are bulbs in the oven that haven't been switched off since August 2010. In April 2011, the company formerly known as SuperBulbs announced its new name, starting the coming-out party that continued at Lightfair and will culminate in October, Sharenow says, when Switch comes to market with its 60-watt equivalent bulb, followed by a 75-watt offering in November and a 100-watt model in February 2012.

"Those bulbs will change the world," Sharenow says.

Dan Koeppel (dan@dankoeppel.com) is the author of Banana: The Fate of the Fruit That Changed the World.