It's hard to name another materials system that has been responsible for as much hype, loss and business failure as CIGS thin-film solar. VC investment in shuttered CIGS firms exceeds $2 billion.

Taiwan Semiconductor Manufacturing Company (TSMC) just announced that TSMC Solar, its fully owned subsidiary, will cease manufacturing operations next month, as "TSMC believes that its solar business is no longer economically sustainable." TSMC will honor all product warranties, however.

The aspirational claims and investment losses in the CIGS photovoltaic materials system have kept Greentech Media busy over the last few years. From Solyndra to MiaSolé to Nanosolar to AQT to SoloPower, the siren song of silicon efficiency at thin-film costs had long kept the CIGS hype -- and funding -- flowing freely.

The funding has stopped. As has the hype. And all that remains is Solar Frontier and a few other stalwarts.

TSMC blamed its "late entry into the market and lack of economies of scale" for "a substantial cost disadvantage," according to a release. The company added, "After careful consideration, TSMC has come to the conclusion that despite its world-class conversion efficiency for CIGS technology, TSMC Solar will not be viable even with the most aggressive cost reduction plan."

Last year, TSMC Solar posted a champion module efficiency of 15.7 percent with its Stion-licensed technology. Actual shipment volume from TSMC was unreported, but believed to be small. TSMC also notes that it had made gains in the module's temperature coefficient, improving performance in hot environments.

According to The Economic Times, "The closure of TSMC Solar follows the sale of its LED lighting unit, another sector which has struggled to reach profitability for the chip giant." 

Here's a partial list of CIGS solar players, past and present:

  • Solar Frontier (only true CIGS volume production)
  • Solibro (sold to Hanergy)
  • MiaSolé  (sold to Hanergy)
  • Solyndra  (bankrupt)
  • Avancis (sold to China's CNBM)
  • Bosch
  • Global Solar Energy (sold to Hanergy)
  • Soltecture (bankrupt)
  • Nanosolar (bankrupt)
  • AQT (bankrupt)
  • ISET (R&D)
  • HelioVolt (bankrupt)
  • Ascent Solar (shifted to consumer solar charging)
  • Stion (majority-owned by Khosla Ventures)
  • SoloPower (could go public in reverse merger on London stock exchange)
  • TSMC (closed)
  • Manz (recently hit 16 percent module efficiency, acquisition talks)
  • NuvoSun (acquired by Dow)
  • Siva Power (still plugging away, recently raised $10 million)

As Brad Mattson, CEO of Siva Power told GTM,  "The challenge with CIGS isn't the science. It's an equipment challenge. It's just process speed and control. And it can be done."  Mattson added, “TSMC is  a great company. The problem wasn’t with its effort, but with the technology licensed from Stion. The 2-step CIGS process is inherently slower and lower efficiency than the co-evaporation process pioneered by NREL (and others) and which holds all the world records for efficiency.  It’s worth noting that First Solar’s successful CdTe technology is also “evaporative” and not a 2-step process.”  Mattson's company looks to "apply semiconductor process control" and "focus on speed," to build a CIGS gigafactory with a line size of 300 megawatts to achieve 28 cents per watt cost.  

Markus Beck is CTO at Siva Power and has worked on the CIGS program at Solyndra, First Solar and GSE. He had something to say: "CIGS via indirect synthesis is penalized from the basic thermodynamic perspective as well as feedstock and processing equipment perspective. As such, it is no surprise that TSMC arrived at a cost that is not competitive to CdTe or x-Si made in China. The science and fundamentals of CIGS has been known since the 90s -- direct formation of CIGS is the only way to achieve fast deposition and high quality material using the lowest cost feedstock (the unprocessed elements)."

"The problem is not CIGS, but the lack of industry to follow the well established basics. A great number of researchers at NREL, IPE, ZSW and many other universities and institutes in the U.S., Europe, and Asia have shown time and time again how CIGS is best formed and reaches the highest efficiency -- direct formation, in particular coevaporation. Industry has failed CIGS by trying to use equipment applicable to a 2-step, indirect formation process, mostly sputtering metallic precursors from expensive sputter targets or by electroplating metallic precursors or coating nano particles. All of these processes then require to crystallize the CIGS in slow, time and energy intensive processes, typically emitting highly toxic and expensive H2Se and H2S (hydrogen selenide and hydrogen sulfide) gases.  This approach appears to be logical if you have a hammer and want everything to be a nail."

"However, what industry should have focused on is the equipment to enable high speed, large area coevaporation, the best known process. Industry turned CIGS into a science project where no science was required.  Instead industry shied away from the engineering challenge."

"Put differently, the Solar Frontier, Stion/TSMC, Avancis, Bosch, Hulk, Samsung approaches are all rooted in the late 1980s/early 1990s Arco/Shell/Siemens approach with no innovation on the equipment side. Solyndra employed coevaporation and yielded an amazingly good semiconductor -- the problem being the costly cylindrical substrate. Global Solar employed coevaporation, but on a steel substrate.  Solibro uses coevaporation, but overlooked the need to develop suitable sources -- the company basically scaled the point sources used in academia. Manz (Würth) uses linear evaporation sources, but they are too slow and don't last long enough."

"First Solar was quite successful with the approach, but due to unrelated issues had to terminate the CIGS program."

"So, the question is whether we decry CIGS as a failure based on obvious mistakes or apply what fundamental science has taught.  I sure hope that there are smart people out there who see the potential and have the means to fund the correct approach. From the perspective of a scientist it is extremely disappointing to see CIGS being misrepresented based on ignorance of the fundamentals of thermodynamics and lack of courage to tackle the engineering challenge of evaporation sources," concludes Beck.

First Solar has succeeded with thin film, albeit using a cadmium telluride absorber material and not CIGS. First Solar had a fleet average module conversion efficiency of 15.4 percent. Its best production line is now shipping solar modules with efficiencies of 16.2 percent. Modules built at the company's Malaysian facility have a cost of less than 40 cents per watt.

The only true volume producer of CIGS solar panels is Japan's Solar Frontier. We'll be interviewing the CEO of Solar Frontier, Mr. Atsuhiko Hirano, at next month's Solar Power International in Anaheim. Stay tuned for that interview -- and see you at SPI.