Feeding the Beast: Why Scaling Up Production Is So Hard

Solar startup companies like to talk publicly about their core technologies of creating efficient solar cells with various etching, texturing or deposition processes. But there is one manufacturing aspect that is, though less sexy, equally important to reach mass production: protocols to ensure workers run the machines efficiently.

One startup has been more public about its need to create the right protocols, which makes a significant impact on production costs: MiaSole. The Silicon Valley startup talked about enlisting Intel’s help one year ago, and the company seems to be on its way to produce highly efficient copper-indium-gallium-selenide thin films in large quantities. The company announced a $55 million equity round last week, and it’s looking for partners who can provide the money and market experience to help MiaSole build more factories and larger sales staff.

MiaSole’s CEO John Carrington declined to disclose the company’s current production cost, but did say that production costs should hit $0.80 per watt later this year. The Intel team remains at MiaSole, he said.

“Figuring out the right configuration for the recipe around CIGS is very difficult. Laying it down, the manufacturing process, is equally hard or harder. Being able to mass produce is the next difficult, and we’ve executed on all three,” Carrington said.

Creating best practices are crucial for controlling the quality and consistency of solar cells and products. These protocols address issues such as scheduling tool down time for maintenance and repair or employee shifts to minimize interruptions in production. Ideally, you want to spot and fix a problem before it cripples the operation.

When I visited the pilot fab of Alta Devices, which is developing gallium-arsenide thin films, I asked Alta’s CEO Christopher Norris about developing manufacturing best practices. Norris has worked at Intel, Cypress Semiconductor and MicroDisplay.

High-volume production is about “feeding the beast,” Norris said, and some of the key issues to work out, besides the tool down time and maintenance, include lining up materials, supplies and training workers — from people who designed the tools to those who complete the final assembly of components. Doing things right early will save a lot of money and headaches later on when a company scales up production.

Over the past five years, a good number of thin film startups have been late to start mass production. It’s not always easy to decipher what caused the delays. Given the financial market crisis, it was apparent that some companies couldn’t raise the large amount of money needed to build a first commercial-scale factory. When market demand is low, competition is fierce and prices fall fast, smaller manufacturers sometimes don’t have the production scale to compete in price or they need to suspend production in order to re-tool their factories for next-generation products. These scenarios played themselves out over the past year.

But creating manufacturing protocols has likely played a big part as well. The solar industry, though not new, is seeing a renaissance since it started fabricating solar cells in the 1950s. Understandably, the industry has been poaching people with manufacturing experience from the semiconductor industry, given some similarities between the solar and chip businesses. Joseph Laia had worked for semiconductor factory equipment supplier KLA-Tencor and Novellus before he took over MiaSole in 2007 (Carrington succeeded him). First Solar hired Bruce Sohn from Intel in 2007. Bruce left to become the chief executive of battery startup Fluidic Energy in mid-2011 but has since left and is now a principal at MEGE Associates.

Taking a solar company from the R&D phase to mass production has shown to be more difficult than what many CEOs and investors initially expected. As the industry grows, it should be able to attract more manufacturing veterans and train newcomers to run factories efficiently, like what companies such as First Solar have had time to develop and demonstrate.

No posts to display