The second of a four-part series. Click here for part one.

LEXINGTON, Mass. -- Away from curious visitors' eyes, laboratory staff at a company called 1366 Technologies here are melting purified silicon in a furnace until it reaches the consistency of red-hot lava. Then they skim off wafers of silicon, the platforms for photovoltaic solar modules.

This may sound simple, but it isn't being done in China, in Germany or anywhere else, says 1366 CEO Frank van Mierlo, one of the Massachusetts Institute of Technology alumni and technology entrepreneurs behind the startup firm.

If it works at factory scale, 1366's process would go far toward making solar energy as cheap as coal power in this decade, he says.

Or it may not work, as van Mierlo acknowledges, although all results so far are on track. "It is really hard. It is not a zero-risk venture. We still have a steep hill to climb," he said.

The firm's project is a case study in the Energy Department's search for game-changing technologies to reduce carbon emissions from power plants and motor vehicles.

DOE's Advanced Research Projects Agency-Energy (ARPA-E) gave 1366 a $4 million grant in November 2009, the only grant to a solar firm in ARPA-E's initial awards. The grant raises a seminal question: How far and how fast can a government agency push commercial innovation?

Energy Secretary Steven Chu dropped by 1366 one afternoon in November, donning a face shield for protection from the glowing silicon furnace. He praised the firm's leaders as examples of "the innovative brilliance of Americans" who can not only compete for leadership in advanced energy technologies "but prevail convincingly," he said.

DOE has set an ambitious, if not audacious, goal of driving the price of installed residential and commercial photovoltaic solar power modules down to $1 per watt, compared to about $3.50 for utility-scale projects, and more than $5 for commercial and residential installations.

"With business-as-usual, by 2016, it will get to $2.20 a watt, and so we need a significant change to get to $1 a watt," said Arun Majumdar, whom Chu recruited from the University of California, Berkeley, and Lawrence Berkeley National Laboratory to run ARPA-E. Hitting $1 a watt would make the United States the global leader, he said.

DOE research grants seek scientific breakthroughs, while loan guarantees underwrite the expansion of solar unit manufacturing with today's technologies. Adding to these are federal subsidies and tax credits and state solar power targets. DOE research and development spending on solar research in fiscal 2010 exceeds $400 million.

Congress has approved the loan guarantees and incentives, and passed the American Recovery and Reinvestment Act, which made $117 million available for additional solar research. The Obama administration's initiatives aim higher, seeking to not only develop new energy technologies, but bring about their commercial success, generating jobs and U.S. leadership in clean energy technologies for the future, according to the president.

But the administration's goals have become another issue that divides congressional Democrats and Republicans. The DOE science and energy programs were funded by the lame-duck House on Dec. 21, with 212 Democrats and 16 Republicans voting in favor, and 130 Republicans voting no.

The price at which 'everything becomes possible'
If solar units could be installed for $1 per watt, the cost of solar energy would fall to 5 to 6 cents per kilowatt-hour, less than half the current average retail price in much of the country, according to DOE.

The calculation leaves out major costs, such as expanding and strengthening the power grid to handle far more complex flows of electricity from millions of solar installations, and creating breakthroughs in energy storage technology to lengthen solar power's impact beyond the daylight hours and to deal with sudden drops in output when clouds arrive.

But the $1 target would enable solar to compete without subsidies with new generation from any source, including coal boilers.

"At $1 per watt ... this will pay for itself," Chu told his science advisers last year. "Everything becomes possible."

Hitting that target will require breakthroughs on three separate fronts, DOE and industry experts agree. The cost of solar modules must be slashed from about 75 cents per watt to a third of that amount, or less. Similar reductions are required for the costs of the electronics that manage solar modules' output, and for installing solar units at homes, businesses and utilities.

Van Mierlo says 1366 aims to take the cost of a silicon photovoltaic solar module down to 25 cents per watt, exceeding this part of the $1 per watt goal.

A new process is born
ARPA-E's funding enabled 1366 to build a prototype factory machine to produce ultra-thin silicon wafers directly from molten silicon. Van Mierlo won't detail how it's done. The process is protected by several patents, "and some things we just keep secret," he said. He compares the task to pulling newly formed sheets of thin ice off a pond in winter.

Today, molten silicon is cast into ingots and wafers are cut from the ingots, but half of the ingot is wasted in the sawing operation.

The 1366 process, invented by MIT professor and company co-founder Emanuel Sachs, would compress four costly production steps into one and eliminate the need to saw wafers from the ingots. "We have demonstrated we can make wafers directly from the melt, which should allow us to get down to less than 25 cents per watt. With the traditional sawing technology, you just can't get there," van Mierlo said.

"Now you have a yield [from the purified silicon] that is well above 90 percent instead of losing more than half the silicon in the sawing operation," van Mierlo said. "There are big, big savings there."

The "Direct Wafer" prototype at 1366 looks a lot like a factory machine, van Mierlo said. "What's missing is the automation. We don't have the throughput yet. But the size of the machine, the way it's built, the interfaces, and all of that is factory-ready."

"There are substantial technology challenges ahead," said 1366 board member Bob Metcalfe. He's with Polaris Venture Partners, one of the firm's initial backers. "That's not a Nobel Prize, but it's a technological challenge. It will take time and money."

Hopes for a strong U.S.-based solar industry suffered a hard blow last month when Evergreen Solar Inc., a leading solar panel manufacturer, announced it would shut down its manufacturing facility outside Boston by the end of March, at a cost of 800 jobs. It will relocate to Wuhan in central China to take advantage of low labor costs and strong government backing for solar expansion.

Evergreen's unique wafer manufacturing technology was invented by Sachs. It opened in 2008 with some financial support from Massachusetts state programs. But with U.S. subsidies uncertain and prices falling, solar panel manufacturing in this country no longer appeared viable.

The $1-per-watt goal has doubters, as well.

"We're still far away from a $1-per-watt cost" for a solar module, said Shyam Mehta, a senior analyst at GTM Research. "I'll be very surprised to see that happening in the next five years, even 2017. Ultimately, that has to happen. We have to hit somewhere around those costs."

"Is that a slogan or a real target?" asks Ric O'Connell, renewable energy consultant at Black & Veatch Corp. "I think the answer is, it's somewhere in between. It's sort of like 20 percent renewable by 2020. It sounds great, '$1 a watt.' In some sense, it is good to have a target, so it's more than just a slogan. But it's early days. They just rolled at that program. There's a lot of work to be done."

Cutting module costs is one step. The second, also a major challenge, requires lowering the costs and raising the performance of the electronics that convert direct current from photovoltaic solar power to alternating current used on the grid.

Getting to 'plug and play'
ARPA-E has created a program called ADEPT, for Agile Delivery of Electric Power Technology, that aims to shrink the size and cost of inverters that do the DC-to-AC conversion, miniaturizing them just as was done with the integrated circuits in computers. "There are some fascinating projects of how to reduce costs, improve performance and really get the technological lead over where the rest of the world is," Majumdar said in an interview last year.

One promising idea is to carry out the conversion at high frequency, enabling miniaturization and integration of components, and hopefully, an exponential drop in costs, just as has happened with computer chips, Majumdar said.

Finally, there are the back end, or "balance of system," costs surrounding module installation.

A report by the Rocky Mountain Institute last year concluded that back-end costs, now averaging $1.60 to $1.85 per watt, could be reduced to 60 cents for ground-mounted units and 90 cents for rooftops.

The lower costs would offer "a pathway to bring photovoltaic electricity into the conventional electricity price range," the report concludes.

But that step requires its own revolution, beginning with an automation of installation work and "plug and play" designs that could be hooked up without the need for electricians. Thousands of different local building codes create another drag on costs that needs an answer. These back-of-system costs are about equal to the costs of the modules, and they are not declining as fast as module costs, Majumdar said.

Under Chu, DOE's biggest effort has been the review and selection of solar projects to be funded with the 2009 stimulus money, including the ARPA-E grants. More than 3,600 applicants applied for the initial ARPA-E funding. Only 37, including the 1366 firm, were successful.

The various DOE solar programs had not been linked into a tightly coordinated attack on the $1-per-watt goal as of the end of the 2010, Majumdar said in an interview. "These are disparate programs," he said. ARPA-E was not funded until the stimulus bill passed in 2009. The ADEPT program is also new. "There are other parts of DOE, so we have been putting our heads together to come up with a well-coordinated initiative."

Throwing '$86 billion at the wall'?
DOE's approach has prompted strong opposition from some Republicans. During House debate last spring, Rep. James Sensenbrenner (R-Wis.) called DOE's R&D programs at attempt "to throw another $86 billion at the wall to see what sticks."

He asserted that the administration has shifted focus from basic research to increased spending "on later-stage technology development and commercialization efforts. ... I do not believe that the government ought to be in the business of picking winners and losers; however, that is exactly what the provisions of this legislation attempt to do."

"We need a broader strategy that prioritizes spending, reduces debt, eliminates deficits and provides clarity, stability and the appropriate regulatory environment," Rep. Dana Rohrabacher (R-Calif.) said during the debate. "This legislation makes no choices. It simply authorizes more and more spending."

Van Mierlo counters that it is not just taxpayers who have anted up for this test.

Two investment funds, North Bridge Venture Partners and Polaris Venture Partners, seeded Sachs and van Mierlo with $12 million in startup capital. The $4 million ARPA-E grant was next. That, in turn, helped attract $ 33 million more last year from outside backers, including a South Korean chemical manufacturer and a European clean energy investor. DOE's backing helped validate the project to these funders, van Mierlo said.

Sachs and van Mierlo spent a year, in 2007, exploring potential technology ventures that led to 1366 -- named for the average number of watts delivered by the sun on every square meter of the Earth's surface. Foremost in their minds was the potential of solar power to generate electricity without the greenhouse gas emissions that come from fossil fuels. "Right from the start, the goal of the company was 'solar at the cost of coal.' That has always been our tag line."

Building value-added exports
But when van Mierlo lists the potential payoffs from the venture, he doesn't start with the climate case. It's the economy.

The United States is still the most important producer of purified polysilicon used in solar cells and semiconductors, led by manufacturers in Michigan and Washington state, van Mierlo says. In 2009, $1.1 billion of that silicon product was exported to foreign solar module manufacturers, according to a study by GTM Research.

If the 1366 process works, U.S.-produced purified silicon could be turned into wafers, and those could be exported. That would boost the export value to $7 billion annually and generate perhaps 50,000 jobs, van Mierlo said.

"That is essentially what justified spending public money for a project like this. If you add up the number of researchers in this facility and you look at the money that's being spent, it doesn't jibe. If you think that you could create an American industry that is competitive and triple the value of the exports, that becomes interesting.

"The moment that you move to electric cars, then the electricity demand is such that you are creating a huge opportunity for new electricity sources to come online, and this will be the day that solar will prevail."

The fits and starts in energy policy that have marked the solar industry's history are big competitive burdens for U.S. companies facing China and Germany. "I think it will play out like in the next two decades or so. It won't be today or tomorrow," explained van Mierlo. "There will steps forward and steps backward. That's the nature of our democratic system. But I would not dismiss the wisdom of our system, either. Ultimately, the U.S. will take on leadership again in the industry it started, and we will make this happen."

"We need a bit of luck," he said. "But there's certainly been a tremendous amount of progress here, so I'm feeling optimistic. You kind of have to be in my position."

Reprinted from Climatewire with permission from Environment & Energy Publishing, LLC., 202-628-6500