In 2005 the world’s solar energy–generating capacity grew by 44 percent. If that pace can be sustained over the next few decades, by 2050 the sun could supply ten times as much energy as the earth needs. Such a growth rate might seem like wishful thinking. But it is worth remembering that the semiconductor industry has grown at an even faster rate for a similar length of time. Innovators and investors took the personal computer industry from zero machines to almost a billion in thirty years, doubling processing speeds every twenty-four months or less and cutting costs in half each time the speed doubled. As Oliver Morton wrote in the September 7, 2006, issue of the science journal Nature, “If Silicon Valley can apply Moore’s law* to the capture of sunshine, it could change the world again.”
To change the world again, and to get very rich doing so—that’s why venture capitalists are pouring billions into start-ups developing photovoltaic cells, which convert sunlight directly to electricity. (The other way to use the sun to make power is by tapping its heat, a strategy explored in Chapter 3.) For investors who made their first fortunes from semiconductors and the Internet, the learning curve on photovoltaics is not terribly steep. Solar power has grown up alongside the chip industry, borrowing its materials and processes and, increasingly, its talent. The geographies of the two industries overlap. Many of the solar start-ups are in California’s Silicon Valley, in Cambridge, Massachusetts, in Phoenix, Arizona, and in Austin, Texas. And many have close relations with the same universities: Stanford; University of California, Berkeley; the California Institute of Technology; and MIT.
For those who believe that individuals around the world have gained more control of their destiny through the computing and networking revolution, solar power has a further appeal. Because photovoltaic cells produce electricity where they are used, they have the potential to reshape the centralized energy economy into something more like the network created by the advent of personal computers, cell phones, and the Internet. Owners of photovoltaic systems can become power producers themselves, sell energy into the grid at a profit, even gain near-term access to electricity in parts of the world that do not yet have energy infrastructure; already, photovoltaics are bringing electricity to poor villages in Africa for about $250 per household. “Distributed energy” erases the strategic advantage of big energy companies, says Andrew Beebe, president of Energy Innovations.* “With solar, they can’t control the resource. That’s a real shift of power.”
EVERY HOUR, the sun provides the earth with as much energy as all of human civilization uses in an entire year. At just 10 percent efficiency—that is, if only 10 percent of that solar energy were converted to electricity—a square of land 100 miles on a side could produce enough electricity to power the entire United States. Those two facts would seem sufficient in themselves to map the solution to global warming. Yet a century after Albert Einstein explained the photoelectric effect (for which he won a Nobel Prize), and fifty years after Bell Labs invented the first semiconductor-based device to convert sunlight into electricity, solar technology remains a trivial player in global energy. In 2007 the total solar capacity worldwide was just 6.6 gigawatts, compared to more than 1,000 gigawatts for coal; in the United States, solar cells provided less than 0.05 percent of the electricity supply.
Part of the reason is the complexity of that supply. Behind every electrical outlet there is a vast web of resources, and introducing new technologies into that web has cascading and sometimes counterintuitive impacts. Still more important is the nature of sunlight itself. The sun does not shine twenty-four hours a day or every day of the year—far from it in some regions. And that creates the need for cost-effective storage: some way to capture and save that intermittent energy so that it is available on demand, even when the sun is not shining. The same problem confronts other renewable energy sources as well—especially wind.
Traditionally, energy storage has meant batteries. But while batteries can inexpensively provide short bursts of power, using them to store the large volumes of energy needed when the wind stops or the sun doesn’t shine, possibly for many hours or days, is prohibitively expensive. Researchers are therefore working not only to improve batteries but also to develop alternative storage technologies—using excess electricity to pump water up into reservoirs for use later in hydroelectric generators, for instance. Until such technologies are greatly improved, however, storage will remain a major impediment to widespread use of solar energy.
The biggest obstacle is that photovoltaics are not yet cheap enough to compete at scale. Three paths could potentially get them there. The first is to continue stepping up efficiencies of existing technologies, primarily crystalline-silicon cells, while lowering costs. The second is to leap to cheap next-generation technologies that can be produced in quantity: making as many square miles of photovoltaic foil or fabric (or paint or Astroturf) as possible, even if it generates less energy per square foot. The third is to pay a price premium for quality: cramming the most efficiency onto the smallest possible cells, then wrapping those cells in optics that concentrate the sunlight, multiplying its intensity five hundred or one thousand times.
To make sense of these paths, it helps to understand the basic principles of photovoltaics. Solar cells are made of semiconducting elements, most commonly silicon, which hold on to their electrons until hit with the necessary oomph, or “band-gap energy.” When a photon bearing light energy from the sun strikes with that threshold amount of energy, it can kick an electron free from the silicon atom and up to the conduction band, leaving a vacancy known as a hole. Each electron-hole pair, bearing opposite charges, is called an exciton. The electrons flow through the conduction band to reunite with the holes. That’s an electric current.
It also helps to understand how the energy produced by photovoltaics is priced. Though electricity is usually priced by the kilowatt-hour,* solar manufacturers talk about the “price per peak watt,” referring to the maximum output of their cells during peak sunlight. In 2007 the peak watt price averaged about $4 (or just under $7 installed—a price that includes the cost of all hardware, mounting and electrical systems, engineering, and installation). The manufacturers are not just being difficult, but rather are trying to take account of the fact that the sun shines more and brighter in some places; in those places the very same system will produce a lot more energy and will therefore cost less per kilowatt-hour.
Take a typical single-family rooftop installation. For a system capable of generating 3 kilowatts, you would currently pay about $21,000. And in an hour of peak sunlight, you would get 3 kilowatt-hours. If you lived in Las Vegas, however, you would get many more of those peak sunlight hours than if you lived in, say, Fairbanks, Alaska. In fact, the yearly output of your 3-kilowatt system in Vegas would be about 6,500 kilowatt-hours, nearly twice the 3,300 kilowatt-hours you would get in Fairbanks. If your rooftop panels lasted thirty years, your price per kilowatt-hour would be 11 cents in Vegas, 21 cents in Fairbanks. That might be cheaper than the electricity you buy from your utility, but again that depends on where you live. Analysts for the French brokerage Crédit Agricole estimate that in Tokyo, where retail electricity prices are extremely high and there is moderate sunshine, “solar is cost competitive at $5 a watt [more than $8 installed]. Los Angeles is close behind (even more sunshine; nearly as costly energy), while solar will not become cost competitive in Portland [Oregon] any time soon.”
Though solar power is generally judged on the basis of whether it can beat the retail price of coal-generated electricity, the comparison misses a key point. The greatest value of solar power is that it is most productive when the weather is sunny and hot—precisely when consumer demand forces a utility to operate at full throttle. For PG&E, for instance, peak demand is growing 25 percent faster than overall electricity needs, but the last quarter of capacity is needed less than 10 percent of the time. “Peak shaving” with electricity produced by rooftop solar panels relieves that pressure on utilities, damping the prime driver to build new gas-fired plants. It also reduces costs. Utilities “turn on” power plants in order of their variable cost, starting with the least expensive plant to operate and moving to the most expensive as demand rises. Solar installations produce electricity at no extra cost just when units with very high variable costs would otherwise be called upon to run, providing tremendous savings for the utility, the consumer, and the atmosphere. (To date, most owners of solar photovoltaic systems do not get the discount they deserve on their electric bills. Typically, utilities charge customers the average cost of generating electricity over the course of a day. Some are now experimenting with “real-time pricing,” charging customers hour by hour for the true cost of the electricity they use.)
Given all that, almost everyone in the industry agrees that when the price per peak watt falls to $1 and the storage problem is solved, solar-generated electricity will compete with coal-fired electricity virtually everywhere.
Producers of traditional crystalline-silicon solar cells, which have dominated the market for thirty years and in 2007 still controlled 93 percent of it, believe they will get to that price with the help of the world’s low-cost manufacturers. In 2007 China became the third biggest producer of solar cells, behind Japan and Germany, and raised billions from public stock offerings to expand capacity further still, creating, along the way, several new billionaires. Its leading company, Suntech, is worth $5.5 billion, employs thirty-five hundred people, and sells 90 percent of its output to Germany. Kyocera Solar, the Arizona unit of Japan’s Kyocera Group, plans to build a factory in Mexico able to produce 150 megawatts of solar cells annually, enough to put new 3-kilowatt systems on fifty thousand homes. Japan’s Sharp Corporation, the biggest solar-cell maker in the world, intends to increase its production capacity, all by itself, to 100 gigawatts by 2030.
India, where the government pays up to the full cost of solar projects in nonelectrified rural areas, is also an emerging powerhouse. BP has a solar joint venture with Tata in Bangalore, and Moser Baer India has contracted with Applied Materials to build a solar factory in New Delhi by 2009. Hong Kong and Taiwan are increasingly important players as well. In addition to manufacturing crystalline-silicon cells, some Asian companies are also making lower-efficiency thin films on glass from amorphous (noncrystal-line) silicon, employing a technology similar to that used in manufacturing flat-screen liquid crystal displays.
Though this global expansion has caused a shortage of crystalline silicone, industry research firm Clean Edge predicts that revenues in the solar photovoltaic industry will grow to $50 billion a year by 2015, reaching a total installed base of 75 gigawatts, a tenfold increase from today. But that would still supply just 0.5 percent of the total amount of electricity needed for 2015. A more rapid expansion will almost certainly require the next-generation photovoltaic technologies now emerging from labs into the commercial market.
Crystalline silicon has limitations. For example, it absorbs light slowly, meaning that wafers have to be thick (and heavy and expensive) enough to capture photons before they slip through. That thickness, in turn, requires highly purified silicon, with no more than one impurity per trillion parts.*
Such limitations have led most next-generation solar-cell makers to abandon silicon in favor of other semiconductor materials, usually several in combination: by mixing or stacking elements with different band gaps (threshold energies), they can harvest a wider range of wavelengths of light, wasting less incoming solar energy. Many have also abandoned wafers in favor of “thin films,” replacing expensive batch processing and heavy, unwieldy modules with cheap, fast, roll-to-roll manufacturing of acres of flexible materials that can go anywhere. Others are developing solar concentrators, which combine small, ultra-high-efficiency cells with low-cost optics. Like the magnifying glass that scorches the hapless ant, these make diffuse solar energy hundreds of times more powerful.
INNOVALIGHT, AN EARLY-STAGE COMPANY based in Santa Clara, California, is making thin films but is one of the rare cutting-edge solar companies still using silicon. CEO Conrad Burke believes it is the only material benign and abundant enough (it makes up 15 percent of the earth’s crust) to supply the staggering amount of electric power the world needs. “God put silicon on this planet for a reason,” he says. “Not really, that’s just me being Irish and Roman Catholic. I don’t think He figures in it. But when those guys at AT&T built the first transistor with silicon, well you know what that set off. Fast-forward ten years…and I’m telling you silicon will win this war.” What Burke has in mind is not, however, conventional silicon wafers. Innovalight has instead made something brand new, using nanotechnology—the engineering of materials at the atomic scale—to overcome nearly all the limitations silicon has in bulk form.
While the purified silicon used for both microchips and solar cells has been plagued with shortages and price spikes, for instance, Innovalight has found a way to bypass that supply chain, making its nanosilicon powder from cheap, unpurified sources. “People have figured out how to make micrograms of nanosilicon in days,” says Burke, standing outside the closed doors of the labs where the closely guarded work is being done. “We can scale to kilograms.” Innovalight also reports that it has reduced the amount of silicon needed per watt from the 15 grams for a conventional solar cell to just 0.04 gram.
Nanotechnology has also made possible high-throughput manufacturing, which Innovalight expects to cut costs by a factor of ten compared to growing ingots and sawing silicon wafers. The company chemically solubilizes its silicon nanocrystals (one-billionth of a meter wide) in ink, adds impurities (a process called “doping”) to get the right electrical properties, and prints the ink onto any surface with an off-the-shelf industrial printer. While silicon normally melts at temperatures above 1,400ºC (2,550ºF), at 2 nanometers it melts at 300ºC (570ºF), cool enough to print onto stainless steel films. Though by the end the material looks almost like crystalline silicon, its ability to harvest light energy is vastly improved. At nanoscale, silicon can be made to perform like many semiconductors in one. By changing the size of the particles, called quantum dots, Innovalight can tune them to tap the sun’s full spectrum, which conventional silicon only partially uses.
Right now Innovalight’s prototypes look like old-fashioned rolls of Kodak film, but by the end of 2009 the company aims to produce each year enough flexible solar material to generate 100 megawatts at the unimaginably cheap price of 30 cents a watt. In the space of five months in 2007, it pushed efficiencies from 2 percent to more than 9 percent, meaning that a percent of incoming solar energy comes out as electricity.
Most remarkably, Innovalight’s silicon quantum dots seem to have found the solar holy grail. Although many photons carry enough energy to unleash several electrons, photovoltaic materials have never been able to produce more than one excited electron for each incoming photon (instead, the excess energy is squandered as heat). But in a July 2007 paper in the American Chemical Society journal Nano Letters, Arthur Nozik and a team of scientists at the National Renewable Energy Laboratory confirmed that Innovalight’s quantum dots are the first to get “multiple exciton generation” in silicon nanocrystals. While the theoretical maximum efficiency for a crystalline-silicon cell is 33 percent, Nozik calculates that this breakthrough could push Innovalight’s efficiencies to 44 percent and as high as 68 percent with concentrated sunlight. As yet, the electrons have only been seen, not harvested; Innovalight still must figure out how to extract those extra electrons to generate electrical power. “But should this approach prove technically viable,” concluded a 2007 Deutsche Bank report, “it would eclipse the conversion efficiency aspirations” of all other thin films.
Burke appears to have the fearlessness for such a gamble. His early career is a blur of leaps, each of them to bigger responsibilities and more cutting-edge experiments. He was twenty-one and studying physics at Trinity College, Dublin, when he saw a job posting to work with optical lasers and amplifiers at the Nippon Electronics Corporation (NEC). He took an immersion course in Japanese and by summer’s end had moved to Tokyo. It was 1989, and Japan was an electronics powerhouse at its peak before the bubble burst—as he says, “a glorious time to be in the country surrounded by technology.” He began publishing papers and making a name for himself, so much so that the Irish embassy arranged with Trinity to award him a masters degree for his NEC research. “It was a very economical way to get a masters, getting paid a salary and having my trips to Ireland financed. Very efficient.”
He stayed with NEC for three years but found research and development too introverted. “I definitely did not fit the mold. I like interacting with people. I was twenty-two, single, enjoying all the attention, traveling to China, Korea, and Indonesia, a bit wild. It was great fun. One of the surreal things about being part of a small minority in Tokyo was that anytime anyone famous came from home—the prime minister or a rock star—the Irish ambassador would round us all up for the party.”
In 1992 AT&T hired Burke away from NEC to do product marketing for its microelectronics group, then asked him to move to the United States. Six months after his arrival, big AT&T came to an end: the three-hundred-thousand-person company was dismantled and split into three independent companies, and Burke’s division became Lucent. Within a few months he had been promoted to Lucent’s director of marketing for Europe, the Middle East, and Africa, and moved his family to Germany (“my children have many different passports”). Three years later he returned to Allentown, Pennsylvania, to run the company’s $400 million optoelectronics group. He was only thirty-two, the youngest director ever at AT&T and Lucent.
So he took another leap: in 1999 he quit his job to become senior vice president for marketing and business development at a small San Diego optical switching start-up called OMM. “I wanted to build a company and participate in the explosive growth of telecom, and they had a really cool new technology,” he says. By 2001 the company had raised $135 million; filed to go public, with Credit Suisse as underwriters; and started its road show, with its product qualified in all networks, $40 million in orders, and the presidents of Selectron and Gateway on its board.
Then the market unraveled, and the orders and initial public offering disappeared. “We had to manage ourselves through that very hard downturn,” Burke recalls. “We stayed intact, finally sold our IP. From a financial point of view it was not an excellent outcome.
“Silicon Valley forgives failure, if you get up and dust yourself off. In the end, I think it helped me gain credibility. Everyone can do well in an up market where everything’s shooting toward the sky. But I learned more in that downward piece, from ’01 to ’03—letting go of a lot of people, bringing down a very expensive operation—than I’d learned in my whole career.
“Though I would not like to do it again.”
Sevin Rosen Funds, the early-stage venture capital firm that had been OMM’s biggest investor, asked Burke to become a partner. He spent a year there. “They paid me very well, and I looked at great technologies, but I found venture capital a bit boring,” Burke recalls. “You meet smart people with great ideas. But you listen from afar, have no real participation—you’re just getting fed the update. I missed being involved in getting those results.”
Eventually, Sevin Rosen asked Burke to run Innovalight, an early-stage Minnesota-based company. The first thing he did was move Innovalight to California. “Capital is much easier to raise in California. And the Silicon Valley culture rewards risk-taking. People don’t hesitate here taking chances with their career and jumping into unproven technology, knowing it may not work out. I’ve lived in the UK, Japan, Germany, the East Coast, and this twenty-mile radius is unlike anything I’ve ever seen. I’m having an absolute ball, and it’s the smallest thing I’ve ever been involved in.”
Burke has assembled a global team of two dozen, including fourteen PhD physicists and chemists from Italy, Belgium, Mexico, Russia, Ukraine, China, Taiwan, and Greece, who together write about two patents a month. Alf Bjørseth, a Norwegian who founded and until recently ran the world’s most vertically integrated solar company, with a $20 billion market capitalization, recently joined Innovalight’s board. Burke wants to grow the company to fifty people by 2008, and would like to tap the deep expertise concentrated in Germany and Japan, but is hamstrung by U.S. immigration policies. “The U.S. could get the best people in the world, superstar PhDs educated at the expense of taxpayers in other countries, but they’re not allowed to stay.” Because immigration is easier for Australians, he is hiring from the University of New South Wales, another powerhouse in solar energy. In October 2007, Innovalight announced that it had raised $28 million in new capital.
Growing concerns about the safety of nanotechnology may present hurdles to Burke and others in the field. Since 2003, Environmental Defense Fund has been involved in a worldwide effort to strengthen the oversight of nanomaterials. In general, regulators treat these materials as if they were identical to the same chemicals in bulk form, despite the fact that the nanoscale versions are valuable precisely because they behave in radically new ways. The limited data now available suggest that some nanomaterials may be mobile and long-lasting in the environment and organisms, and may be capable of damaging brain, lung, and skin tissue.*
For his part, Conrad Burke seems surprised by questions about safety, which is perhaps understandable given his neighborhood’s general faith in new technology. He says his workers are thoroughly protected, and that Innovalight’s nanoparticles are essentially erased by the time the solar thin film is complete. “We use nanotechnology as a cheap vehicle for manufacturing, but by the end the nanoparticles are stuck together, deformed in processing in a way that they’re not ever able to be released.”
BURKE’S BIGGEST CHALLENGE, as for most innovators, will be taking his team’s success in the lab out onto the production floor. One of his favorite people—and a model for the challenges he knows lie ahead—is Dave Pearce, the founding (and now former) CEO of thin-film competitor Miasolé, which in early 2007 seemed on the brink of commercial production at its 111,000-square-foot Santa Clara factory and subassembly facility in Shanghai. (Pearce, who is paunchy, balding, and middle-aged, gave Miasolé its quasi-Italian name—a loose translation is “my sun”—soon after the film Under the Tuscan Sun was released, in a vain attempt to lure actress Diane Lane onto his board. “We did get John Doerr,” he jokes.)
Miasolé does not use silicon but a compound semiconductor made up of copper, indium, gallium, and selenium known as CIGS, which scientists have long tried to exploit. A CIGS film as thin as 1 micron has the same photovoltaic effect as a typical crystalline-silicon wafer 200 to 300 microns thick (about the thickness of a human hair), translating to savings both on expensive semiconductors and on weight. CIGS also works better than silicon at low angles of the sun and on hazy or cloudy days. And like Innovalight’s thin film, it can be churned out rolls at a time.
Unlike the youngsters running many of the solar start-ups, Miasolé’s founding team brought decades of experience at industrial-scale manufacturing. Pearce himself arrived in Silicon Valley in 1985, when he became CEO of Domain Technology, a hard-disk maker that had burned through $22 million of its $23 million and was about to go under. In six months Pearce turned the company around, proving out its “sputtering technology” for depositing magnetic films and turning a profit. (Portions of Domain were later acquired by Seagate.)
Pearce likens the sputtering process, which Miasolé now uses to make solar films, to a game of billiards. Inside a vacuum chamber, magnets are used to accelerate argon ions (the cue ball), which then knock off atoms (the pool balls) of the target material (copper or indium on these solar films). The atoms settle one by one, producing precise films as thin as five to ten atoms.
In Miasolé’s sparkling Santa Clara factory, that process takes place in giant U-shaped machines. At one end, a meter-wide roll of flexible stainless steel unscrolls at the rate of two feet a minute; at the other end, out comes the photovoltaic foil. Each production run delivers several square miles of solar foil, which is then encapsulated in a rugged, flexible material. The company builds its own capital equipment to save money (they spend a tenth as much as if they bought it, says Pearce) and to make possible continuous redesign for ever-higher throughput.
Pearce grows particularly animated when he talks about production challenges—for instance, how to get the grid lines, which conduct the electricity, onto the cells. When engineers tried printing silver ink directly onto the film, the ink wicked down into microscopic divots in the foil and shorted out the cell. They realized they needed to bridge the divots, so they designed a decal, with the gridlines printed onto a polymer, which they then stuck onto the cell. The decal, in turn, gave birth to other inventions. By printing the decal with black ink first, they eliminated visible grid lines, and reduced reflection so it absorbs more solar radiation. They also found they could hang the buss bar, which transports the electricity, from the decal’s edge, leaving the whole cell photoactive.
Miasolé’s aim is to make solar technology “simple enough for your average Home Depot customer to do it themselves,” as Pearce put it. Their panels will include mounted sensors and wireless communications capability, to make them smart enough to tell their owners when they need to be cleared of fallen leaves. Pearce explains: “In most systems, if shade falls on one cell you degrade function in the whole series. Installers struggle to avoid that with highly precise positioning. But because these will have discrete electronics for each, a problem with one won’t mess up the rest.”
At the beginning of 2007, Pearce was certain that by 2008 Miasolé would be producing 200 megawatts of new generating capacity each year, at a cost of $2 per watt installed. That “installed” part is important. Unlike fragile crystalline-silicon cells, these photovoltaics needn’t be framed in glass and aluminum and mounted on the roof: they are the roof. The product bypasses the costs of framing and installation, and also the conventional solar distribution channels, which have become badly bottlenecked. Instead, Miasolé’s panels will be part of the building-materials industry, like sheetrock and tarpaper. Encapsulated into composition shingles, which already make up 85 percent of U.S. roofs, they can be glued right on to plywood roof decks all over America. The company has already begun discussions with large-scale commercial and residential developers like Toll Brothers and KB Home, which could offer an electricity-generating roof bundled into the mortgage, like a granite countertop or high-speed Internet access. Because the owner will need much less power from the grid, Pearce says, his combined mortgage and electricity bill will be less with the solar roof than without.
As a member of SolarTech, a new consortium that includes SunPower (a leading crystalline-silicon solar cell maker) and PG&E and aims to turn “Silicon Valley into Solar Valley,” Pearce has worked to expand such financing options to eliminate a chief obstacle to broad consumer adoption. People buying rooftop solar systems today are essentially required to prepay a couple of decades’ worth of energy bills; Pearce compares it to buying a car and having to pay upfront for many years’ worth of gasoline.
To fix that, various financial institutions have adapted “power purchase agreements,” a common financing instrument for conventional electricity generation, to solar power. Morgan Stanley, for instance, is financing the installation of SunPower solar arrays at twenty-two Wal-Mart facilities. Wal-Mart makes no capital investment and does not own the solar panels. Instead, it has committed to buy the electricity generated by the panels at favorable rates, which are locked in for twenty years.
Bank of America has a similar arrangement with Chevron Energy Solutions and the San Jose public schools. The bank finances and owns the solar installation, earning state and federal subsidies and tax credits. Chevron installs and operate the modules, and the school district buys the green energy at below-market rates, reducing its demand for utility-supplied electricity 25 percent and saving $25 million over the life of the equipment. A few companies offer the full range of services. SunEdison, for instance, finances, builds, owns, and operates rooftop solar installations for Whole Foods and Staples. Industry insiders call it a “distributed utility.”
To make such arrangements available to smaller businesses, SolarTech is creating an online marketplace to match consumers with financing and solar suppliers. Eventually, the group envisions individual homeowners being able to switch to solar without any front-end cost, cutting both their global warming impact and their electricity bills.
The technologies themselves are proving more stubborn. In May 2007 Pearce announced delays in commercial production. “We’re trying to give birth to a new process. The trouble is that we don’t know how long the gestation period is.” Though Miasolé’s research and development lines were still hitting the target efficiency of 8 to 10 percent, that was only on 5-square-foot panels of film; on its big commercial lines, the company was getting just 4 to 6 percent efficiencies—a rate that improved somewhat by October 2007. Miasolé is also contending with limits on the global indium supply, which PHOTON International in July 2006 estimated as enough for just 4 gigawatts of CIGS. That scarcity has already affected the semiconductor’s price, which in 2007 was triple what it had been five years earlier.
In September 2007 Pearce became chairman of the Miasolé board, yielding the job of chief executive to Joseph Laia; by year’s end, Pearce and most of his executive team were gone. Laia was most recently a group vice president at KLA-Tencor, which makes diagnostic tools to increase semiconductor manufacturers’ yields; in 2007 the company had revenues of $2.7 billion.
While confronting the challenges of going to scale, Miasolé and Innovalight are well aware of rivals like Nanosolar, which in late 2007 shipped its first CIGS thin films. Asked about silicon quantum-dot start-up Stion, which raised $21 million in its first year, and Octillion, which says it can spray a transparent coat of nanosilicon film onto glass to make windows that can generate electricity, Conrad Burke shrugs. “They just started working on it, and they’re claiming they went from zero to sixty in ten seconds? We’ve been on it for four years. We know all the processes that don’t work because we’ve tried them. Silicon is a beast. And before you can build the technology you have to build the tools you need. It’s not like designing an iPhone, where all the pieces are available and you just have to put them together. This stuff is hard and takes longer than you think. People think Google was an overnight success, but it took ten years.”
And while it is tempting to judge the likelihood of these newcomers’ success by how much money they have raised from venture capitalists, Burke warns against it. “The VCs have more money than they know what to do with,” he says. “But for every dollar you raise, you need to return a multiple of that. If you raise $100 million, you have to return a billion to your venture investors. That makes it a little more difficult.”
One thin-film company has already had a fairy-tale ending, which is often invoked by other companies still struggling to get the process right. When Phoenix-based First Solar was founded in 1999 with $250 million from Wal-Mart family member John Walton, who was passionately concerned about global warming, it planned to spend three years and $40 million to commercialize its process for condensing cadmium telluride gas onto glass. Six years and $100 million later, the company was still trying. Meanwhile, BP Solar abandoned its own large-scale effort to commercialize cadmium telluride. Concerns about cadmium’s toxicity and potential release in a fire or landfills further delayed First Solar’s development and added the ongoing cost of a company-run program for recycling panels.
By 2006, nonetheless, First Solar had secured the prize. Revenues that year reached $135 million, up from $13.5 million in 2004. Its initial public stock offering raised $400 million; shares issued at $20 were worth $74 by the following June, bringing the company’s total value to $5 billion. By 2007 the company’s plants in Ohio, Germany, and Malaysia could make a two-by-four-foot glass module in two and a half hours, with efficiency exceeding 9 percent. (Though their modules sold in 2007 for more than $2 a watt, they expect to cut costs enough to get the market price down to $1.) Annual revenues were forecast to more than triple, to $480 million, and the stock was trading well above $200. The company had long-term contracts worth $1.62 billion to supply modules capable of generating 795 megawatts to European and Canadian buyers—nearly eight times the total shipped in 2006 from every solar factory in the United States—and was planning to raise as much as $1 billion in another public stock offering.
BILL GROSS, CEO AT ENERGY INNOVATIONS, is as interested as the thin-film makers in generating trillions: both the trillions of watts of clean energy the world needs, and the trillions of dollars to be made in the energy markets. “Energy drives 50 to 75 percent of our economy. It’s embedded in everything.” He picks up a water bottle. “It’s embedded in this. And it’s a complete commodity market. All electrons coming out of the wall look the same.”
But rather than make acres and acres of cheap photovoltaic material as Innovalight and Miasolé are doing, Gross believes the way to get to the terawatt scale is by concentrating the sun. “The sun is just too diffuse,” he explains. “It radiates 1,000 watts per square meter of the earth’s surface. A hair dryer is 1,000 watts per square inch. So the sun is sixteen hundred times more diffuse than a hair dryer.” To make 1 gigawatt of power (enough to power San Francisco) using traditional photovoltaics requires four square miles of silicon, according to the National Renewable Energy Laboratory. Concentration reduces that area by orders of magnitude: under Energy Innovations’ sun concentrators, Gross says, 1 square inch of photovoltaics will produce as much energy as 800 square inches without concentration.
Concentration is also, he believes, the cheapest of all options because it leverages the scarce supply of purified materials: “If you concentrate the sun a thousand times, a gigawatt’s worth of silicon will get you a terawatt.” Optical devices for concentrating light, and trackers to follow the sun across the sky, are cheaper than the photovoltaics themselves. “Thanks to Moore’s law,” says Gross, “a full-fledged microprocessor that can track the sun costs 20 cents, down from $2,000 just twenty years ago.”
Gross has been toying with concentrating the power of the sun ever since high school. In 1973, at age fifteen, he began making small parabolic dishes in shop class to power Stirling engines, which use heat to drive a piston. He sold the devices through mail order ads in the back of Popular Science magazine. “Super solar devices,” the ads read. “Catalogue 25 cents.” In 1996, Gross founded Idealab, a Pasadena-based incubator for high-tech start-ups. In a decade it has spun off forty companies, including Internet directories (City Search) and paid search sites (Goto.com). In 2003, Idealab sold Overture Services to Yahoo for $1.7 billion.
Where Dave Pearce would fit right in at a midwestern factory, it is hard to imagine Gross anywhere but California. He has a dreamy, slightly otherworldly air, like a boy prodigy immersed in his own imaginings. He talks quickly, sometimes drawing pictures to explain himself (his paintings hang on Idealab’s white brick walls). All of Idealab’s incubating companies occupy an open workspace that is a study in high-tech cool, with exposed metal conduit and a wide-open “village” of bright yellow workstations meant to encourage creative mingling. In the reception area, an “Energy cam” tracks the performance of the solar modules on the roof, including the kilowatt-hours generated and the pounds of carbon dioxide avoided since December 2006. A quote from Schopenhauer flanks the entry: “All truth passes through three stages: First, it is ridiculed. Second, it is violently opposed. Third, it is accepted as self-evident.”
Energy Innovations’ initial concentrator design, the Sunflower, arrayed a circle of mirrors (called heliostats, because they track the sun) around a Stirling engine. That design was shelved in favor of a lens that could produce higher concentrations by focusing sunlight down through a glass funnel and onto a solar cell. The heliostat technology was taken over by another Idealab company called eSolar, which couples the mirrors with small towers and thermal receivers to make solar power plants, the subject of Chapter 3. In late November 2007, Google named eSolar a partner in its initiative to spend hundreds of millions of dollars developing “renewable energy cheaper than coal” (the program’s name uses mathematical shorthand: RE<C).
The new Energy Innovations modules, which resemble banks of stadium lights laid flat on the roof, come fully assembled for easy installation. Twelve are linked on a frame that pivots up and down and across the horizon; microprocessors guide the movement, keeping the lenses and cells aimed directly at the sun. Energy Innovations designed both the tracking software and much of the hardware—from the chemistry of the plastic and patterns of grooves within the lens, to the gill-like passive cooling system, designed to conduct away the intense heat while using minimal aluminum. In an environmental test chamber, the modules are hammered with one-inch balls of hail and hundred-mile-an-hour winds, to make sure they do not lift off or break or get wet and short-circuit. The commercial modules, which will be made in China, will “origami down” into a compact form to reduce shipping costs.
The real magic comes from coupling the concentrators with the world’s highest-efficiency solar cells. The cells don’t come from Silicon Valley, but from a big, old company, Boeing subsidiary Spectrolab, which for two decades has made the photovoltaics that power NASA satellites and lunar explorers. Spectrolab is now bringing that space technology back to Earth, developing “terrestrial applications” for its cells. In December 2006 it set a new world record of 40.7 percent efficiency, the highest ever achieved by any kind of solar cell.
In a big old building in a run-down industrial area east of Los Angeles, Spectrolab technicians in clean rooms feed elements to a crystal growing atop a wafer, layering gallium, indium, and arsenide, each of which absorbs a different wavelength of light. Each layer is extremely thin and semitransparent, allowing light to pass through to one of three active junctions, where the different wavelengths are absorbed. In darkened labs, researchers test the spectral performance of new compounds, trying to push efficiencies higher still. Others work on removing the wafer, to make the cells lighter and flexible, like a thin film.
With its $50 million infrastructure and decades of experience in space, Spectrolab is one of the few places capable of making such super-cells. The difficulty is that the elements that work best together to maximize the conversion of solar energy into electricity do not easily grow together because their atoms are differently spaced. “It’s like trying to put a big peg in a small hole,” explains Nasser Karam, Spectrolab’s vice president for advanced technology products. “It sits uncomfortably.” To ease the strain, Spectrolab adds twenty-seven buffer layers between the sheets of semiconducting alloys, which allow the atom spacing to change slowly, like a series of frames in cartoon animation.
In a big assembly bay, immense barrel vaults and the flat broad wings of satellites wait for their cells to be affixed. On the wall hangs a gift from NASA’s Jet Propulsion Lab: pictures of the Mars Rover chugging along, powered by Spectrolab cells on its flanks—cells that not only have withstood temperature changes from 100ºC to –170ºC (210ºF to–340ºF) in a few seconds but also, to everyone’s surprise, have been undeterred by Martian dust.
By the time it is finished, each Spectrolab wafer can make a kilowatt of power. Every two days, the facility makes one thousand wafers—a megawatt of generating capacity. By late 2007 the company had orders in house for a million wafers—a gigawatt. Among its customers is Concentrating Technologies (C-TEK) in Alabama, which is building a grid-connected “solar farm” for Arizona Public Service.
Most companies combining concentrators with photovoltaics are pursuing the utility-scale market, but Gross prefers to compete with retail price. He is focused on the commercial market, half a megawatt or more, to leverage the expense of installation. “To power your building you typically need a watt per square foot, and at 15 percent efficiency, as much roof as floor. For instance, at Idealab we have forty-five thousand square feet, so we need 45 kilowatts, and we can get that if we cover our whole roof at those efficiencies. If your roof is pitched, you can only cover the south half. But we think there are enough big, flat roofs out there for us to make a big impact on peak demand. You can’t turn off the coal plants but you can make it possible to build no new ones.”
To break through the sales and distribution logjam that hampers many solar start-ups, Energy Innovations purchased a big power and systems integrator, now called EI Solutions. In 2007 the company installed enough Sharp crystalline-silicon cells on three acres of roof at the Googleplex in Mountain View, California, to produce 1.6 megawatts, at the time the biggest commercial installation in the United States (Google cofounder Larry Page is an investor in Energy Innovations). Google expects to save almost $400,000 annually in energy costs, paying for the system in seven years. Real-time information about the system’s output is posted on Google’s corporate Web site. For Puget Sound Energy in eastern Washington, EI Solutions arrayed solar panels capable of generating half a megawatt between windmills, getting the land to do double duty. The beta installations—the first real-world test—for Energy Innovations’ own rooftop concentrators will be in Pasadena.
Meanwhile, in laboratories around the country, photovoltaic innovation continues at a fierce pace. Seven months after Spectrolab set its record, for instance, a Defense Advanced Research Projects Agency and University of Delaware consortium that includes BP Solar, EMCORE, SAIC, National Renewable Energy Lab, and MIT achieved 42.8 percent solar-cell efficiency. Instead of layering different semiconductors onto a single cell, researchers used “spectral splitting” optics to divide the sunlight into three “energy bins” of like colors, directing each onto a different type of cell capable of absorbing that range of wavelengths. The device’s wide acceptance angle, they claim, eliminates the need for tracking devices. DuPont will help transition the lab-scale work to a manufacturing prototype.
Yet another strategy is under development at StarSolar, a Cambridge-based company that was launched by a young MIT graduate student named Peter Bermel when he won a $100,000 prize for his business plan. Bermel uses “photonic crystals,” which can be tuned to reflect specific wavelengths to recycle the photons that slip through silicon, making possible much thinner wafers. Backing the cell, the nanoscale structures also bend the light so that it reenters at an oblique angle and bounces around inside, upping the odds that the silicon will harvest its energy.
Caltech chemistry professor Nate Lewis is looking beyond thin film to solar materials that would scale up “like rolling out carpet or painting your house.” His solution to the thickness problem is a new geometry. What if a solar cell were tall but very thin, like a nanoscale blade of grass or Berber carpet fibers? Then it would be thick enough from top to bottom to absorb the solar radiation, but thin enough so that the excited electron could go out sideways. Lewis says that “it opens up the possibility of using cheap materials that could never be purified enough to get the electrons to the top.” A variant might use what Lewis calls “solar sand”—nanoscale balls of titanium dioxide, a cheap material used in toothpaste and white paint; they would be suspended in a liquid conductor that could then be painted onto a house.
THESE PHOTOVOLTAIC INVENTORS and entrepreneurs all know what happened in the 1980s when oil prices dropped after their 1970s surge and both public and private commitment to developing solar energy pretty much dried up. This time, they think the chances for technological and environmental breakthroughs are far greater. In the words of Conrad Burke of Innovalight, “Twenty years ago no one would have put money into these things. They would have viewed it as some niche-y environmental thing. Now it’s for real.”
At the same time, all these entrepreneurs agree that serious changes in government policy are needed. Until now, the main policy tools used to advance alternative energy have been subsidies and mandates, which have helped promote renewable energy. Although Japan and Germany account for just 10 percent of the global energy market, thanks to long-term subsidies they command 70 percent of the solar market worldwide. Germany has used a “feed-in tariff,” which requires utilities to buy electricity from renewable-energy producers, including owners of rooftop systems, at above-market rates. In 2007 German utilities paid up to 57 euro cents (about 72 cents American) per kilowatt-hour for solar energy, about triple the price at which they sold energy back to consumers. Not surprisingly, that has spurred huge investments in solar products. “All the high-value solar companies in the world have made it by selling their products in Germany,” Burke says. “That country will be so wired up to the sun many nations will be envious. It’s like Korea, which decided to wire the country with high-speed fiber-optics and now has the fastest communications network in the world—it makes ours look like bicycles.”
But those successes are a testament to the long-term commitment Japan and Germany made to alternative energy—rather than to the particular way they chose to promote it. Subsidies and mandates have several critical weaknesses. For one, they depend on a degree of detailed knowledge and a prescience about technology beyond the reach of government regulators.
They also reward lobbying prowess more than technologies that actually perform, and can result in perverse outcomes. “The European model of feed-in tariffs richly rewards certain players,” says John O’Donnell, a solar power entrepreneur you will meet in the next chapter. “And it creates bizarre situations which have nothing to do with slowing climate change.” He points to the German subsidy for photovoltaic power as a prime example. Because Germany is a generally unsunny place, it takes as much as six years for a photovoltaic cell to generate as much electricity as it took to manufacture it. Demand for fossil electricity, therefore, has yet to drop, says O’Donnell: “Not a single coal plant has yet been shut down by this initiative,” even while the net cost of electric power (including the big government subsidies) has risen to 50 cents per kilowatt-hour.
Subsidies are also notoriously fickle; they can get the ball rolling but also drop the ball when they are abruptly eliminated, as they so often have been for renewable energy. O’Donnell’s colleague Peter Le Lièvre points to Spain, which pays a premium of about 25 cents American per kilowatt-hour for solar power. With its ample sun, “Spain is a delightful hotpot right now,” says Le Lièvre. “But it is an aberration of a single policy setting that could be turned off at a moment’s notice.” In the U.S., “the federal investment tax credit is very nice, but is ultimately a destructive weapon in that it’s not a permanent market reform, but a bit-by-bit handout, as if we’re begging for assistance. Nothing could be further from the truth. Market reform is a much more durable and sustainable platform on which to build our long-term investments. We strongly believe that mobilizing capital markets is the best method for deploying these technologies rapidly.”
Policy, in other words, must harness the most basic and visceral impulse of capitalism: the pursuit of profit. Says Burke: “We’ll start taking out coal and oil, and the end result will be good. That makes us more motivated than if we were just designing some new process for a laptop to make the Web go faster. But my primary goal is to return a lot of money to my shareholders, including me and the employees. The scaling of solar is not going to happen with a bunch of environmentalists. You need big, serious money.” As Economist writer Vijay Vaitheeswaran puts it, “You link the technology to the capital, and that’s where the rubber hits the road.”*Events in California since its passage of a law capping carbon emissions offer proof of that, suggests venture capitalist John Doerr: “The biggest impact has been the availability of capital. Some would call it a bubble—I would call it a boom.”
Bill Gross of Energy Innovations predicts that since “there will be ninety-nine failures for every one success,” it will not be large, established companies that achieve the “disruptive innovations” that are necessary: “It’s a question of small and nimble versus big and powerful.”
What is the best way to encourage the small, nimble risk-takers? Gross argues for using tax policy to spur change. Fossil energy, he says, should be brought to its true price with carbon taxes and similar levies—“then let the market figure it out.” But nowhere in the world has a tax successfully solved an air pollution problem. And given the enthusiasm with which politicians generally greet taxes, it is unlikely to be the right solution to this one. More fundamentally, a tax fails to provide the certain pollution reductions offered by a cap: it tells polluters how much they must pay to pollute, but does not impose a legal limit.
That brings us to the cap-and-trade system—the best way to harness market forces to fix a market failure. Instead of forcing polluters to pay certain prices or to back particular technologies, the cap-and-trade system mandates only the pollution limit, then lets the competitive machinery of the market figure out the cheapest, most efficient way to get there. Mobilizing the market ensures that the hunt for the cheapest technologies will be as broad as possible, ranging as far as the human imagination; only with such a far-reaching search will the United States be able to reach the 80 percent reduction in global warming emissions that scientists tell us is necessary to stabilize the climate. That broad hunt, in turn, sets in motion a valuable cascading effect: as the market finds the most efficient technologies, and quickly brings down the cost of reducing pollution, the political will builds for even steeper carbon cuts—without the backlash that inevitably follows when the government tries to pick technologies and too often makes the wrong choice.
Dave Pearce, founder of Miasolé, is one of the growing number of clean-energy innovators arguing for this view. As one of fifteen members of the TechNet Green Technologies Task Force, he has been advocating in Washington for a federal cap on carbon emissions to “create an environment in which new energy technologies can emerge and thrive.”
A task force document spells out its top priorities: “Climate change policies must internalize the environmental cost of emitting greenhouse gases, thereby reducing or eliminating the price differential between high carbon-emitting activities and more environmentally sound activities. A cap-and-trade system built around market mechanisms will be the most powerful driver of demand for new technologies.” That dollar-a-watt threshold for solar energy to compete with fossil-derived energy, in other words, assumes that coal plants are not required to clean up their global warming pollution. Put a cap on carbon, and it has a price for everyone, including coal burners—and solar power becomes competitive in much less time. John Doerr says that if a carbon cap were enacted in 2008, Miasolé’s thin-film technology would achieve “grid parity”—make electricity for the same average price charged by utilities—by 2010.
The goal, the urgent necessity, is to reduce global warming pollution in the atmosphere enough to pull us back from the precipice before the changes in earth’s ecosystems and weather patterns become so rapid and so vast that we will no longer be able to reverse the catastrophe.