New Hampshire, USA — Armed with backgrounds and technologies from two related industries, the resurrected thin-film solar PV company vows to show the industry what scaling really is.
Siva Power, née Solexant, has re-emerged from stealth (again), having circled around its thin-film solar PV roots to reorganize behind copper-indium-gallium-selenide (CIGS). The company now says it’s “de-risked” its technology and is ready to scale up to a 300-MW “pilot” line, and ultimately build multi-gigawatt fabs on every major continent.
Three years ago the company planned to build a 100-MW CdTe fab in Oregon, but as the plunging prices and soaring demand of silicon began dominating the market and swamping other thin-film aspirants, Solexant dove back into stealth in 2011, and reemerged this spring as a CIGS developer. With Mattson, a former semiconductor equipment exec, at the helm, Siva Power is championing a new direction: three-stage co-evaporation on glass substrates with monolithic integration.
Two Keys to Cost-Effective CIGS
Mattson calls CIGS development “a blessing and a curse.” The technology has the allure of tweaking several materials any number of ways to enhance their integration and combined performance, but those arrangements are tricky to control. Other CIGS developers have sought simpler and cheaper manufacturing methods for CIGS, but they end up with subsequent recrystallization steps that “are just a killer in the manufacturing process,” Mattson said. This means that they typically have to go back and restart the process, which adds costs and time — that’s why the CIGS efficiency record has been stuck at around 20 percent in labs.
With its three-stage co-evaporation process, Siva says it can manipulate the material composition to optimize each layer throughout the device to maximize efficiency — a certain blend of material in the bulk layer to maximize the sunlight being trapped, and different blends at the back- and front-contacts to reduce recombination.
How to handle hot glass, and the design, materials and construction of the source, are two keys at the heart of Siva’s CIGS process development work, Mattson explained. Three-stage co-evaporation is linear with temperature, meaning the hotter it is the better the crystal quality and faster it can be processed. But high temperatures are also a limiting factor; typical solar glass loses strength and deforms far below the melting point of the copper, meaning much delicate balancing of temperatures — and related variables that can bring them more closely into line, such as vacuum pressure or dopants — are required during the deposition process. That also extends to the evaporative sources themselves; the extreme temperature and vapor pressures from processing the copper can overwhelm even the source components, causing reliability and repeatability issues, he explained.
Asked about Solar Frontier, currently the only CIGS thin-film company doing anything close to scale, Mattson questions both its process complexity (more glass layers and encapsulant material, more processing and deposition steps) and its cost structure within the giant parent company. “I don’t know how you can make them in a commodity market,” he says — adding that replicating complexity is not proper scaling (more on scaling below).
But CIGS three-stage co-evaporation with high-vacuum processing and handling needn’t be as burdensome or as cost-impactful as some believe, he said. They are “really no mystery” to the semiconductor manufacturing sector, which long ago figured out how to optimize material layers for multiple parameters and implement the very tight process control required to do it with repeatable quality.
The Real Way to Scale
“People have missed the scaling thing,” Mattson said. “They’ve scaled uneconomically, doing small lines and making a lot of them — but it turns out they’re all unprofitable.” Semiconductor and flat-panel display makers, on the other hand, have shown us the real way to scale up technology manufacturing, starting with exponentially bigger substrate generations and huge machine throughput efficiency gains. (That’s why Siva is pursuing deposition on large 2 m2 glass sheets.) Building larger factories to accommodate those types of scale-ups doesn’t require a comparable scale-up in spending, though, so the economics end up being favorable.
And some of they tools they’re using, such as depositing molybdenum and transparent conductive oxides (TCOs), can be adopted into CIGS production with a little customization, Mattson pointed out. Siva plans to outsource about 70 percent of tool development to third-parties under strict specifications and exclusivity, a la the First Solar model, while keeping the heart of the production equipment and processing: the chamber to co-evaporate the four materials. They’re on the fourth version of that tool, Mattson pointed out. “From our point of view, we’re not concerned about patents, but ownership of the technology,” Mattson said.
Mock-up of Siva Power’s 300-MW pilot line. Credit: Siva Power
Siva already has “de-risked” its technology, from process to performance to costs, on what Mattson calls “a megawatt-category” R&D line — and he argues that’s how it should be done, improving the tech and defining processes and costs at a small scale. Once the architecture is locked in, throughputs are hit and cost structures are determined, then it’s time to talk about scaling up.
The next step for Siva is securing financing to build out a 300-MW “pilot line,” ten times the scale of a typical silicon PV line. The company already has VCs as co-investors — $60 million over three rounds from Trident, Firelake, Medley, DBL, and others — but now they’re going after “bigger pots of money in private equity,” including strategic company partners and potentially even governments. “We have done deposition rate tests, we know we have the throughput,” Mattson said, so now “we have to build that toolset.” He’s agnostic where to put this pilot line, saying it’ll be “wherever we have our first major investor.”
Mattson says his 300-MW factory will cost roughly $100 million to build or $0.32/Watt in capital costs, less than a third of a 250-MW four-line First Solar plant (or, triple its space efficiency at 30 kW/m2). That would make it “the most efficient factory in the world from a capex point of view,” Mattson quipped. Off the planned success of that first 300-MW line, Siva would seek an IPO to launch a gigawatt-sized factory, which on those same metrics would cost about $300 million — but again that’s a gigawatt-sized line. The company is shooting for 2-GW factories in several regional locations, which Mattson suggests isn’t an outrageous goal given demand projections of 60 GW right around the corner.
There’s another metric Siva’s shooting for: $0.40/W manufacturing costs, significantly lower than even silicon PV today. Most other solar PV platforms “don’t pencil out” under a rigorous cost-of-ownership analysis, he asserts, spanning everything from processes to materials to electricity usage to labor to tool depreciation. But co-evaporated CIGS on glass with monolithic integration, with the fewest process steps and the least material and with tighter process control, will hit that target on a 15-percent-efficient module, Mattson said, adding that the company has “a game-plan to get to 16 percent and further” with a goal of setting the new CIGS efficiency record next year.