Solar manufacturing models in dispute

Soitec Fraunhofer

Market research firms square off over question of outsourced ‘fabless’ solar manufacturing approach, Soitec shatters CPV cell efficiency record

While fabricating a crystalline-silicon solar cell is indeed another form of semiconductor manufacturing–and the respective supply chains, materials and equipment sets, and process flows of the PV and IC worlds have some similarities–the types of end-devices produced and the business models of the companies that make them are quite distinct. Still, the tendency to draw parallels between the two sectors continues among market analysts and observers familiar with both industries (including this observer), as can be seen in the latest contretemps between NPD Solarbuzz and IHS.

Solarbuzz has pointed to growing outsourcing of manufacturing by many major solar firms and made comparisons to the fabless or “fab-lite” (asset-lite) approaches used by most semiconductor concerns. IHS sees the likelihood of major PV manufacturers outsourcing most of their production and creating a fabless model as “highly unlikely,” according to its most recent press release. Both research outfits do concur on the probability of a return of capital spending by the PV fab companies by next year—welcome news for the equipment vendors that have endured a prolonged overcapacity-induced downturn.

A news analysis by Mark Osborne at PV-Tech examines the opposing viewpoints of the two market forecasters. His conclusion? Advantage, Team Solarbuzz. After offering several of IHS’s talking points and related information, he shares IHS’s central thesis. “The proposition that Tier 1 Chinese manufacturers would come to rely on capacity from Tier 2 and Tier 3 suppliers is highly unlikely,” noted Mike Sheppard, senior PV analyst at the firm. “The top players wouldn’t give away the fruits of their efforts during the past two years to improve their technology, quality and bankability with financiers or owners. This is especially true as these companies attempt to improve their positioning relative to Western players.”

Mark then serves up NPD Solarbuzz’s observations that regarding “mainstream bankable PV modules, excluding the likes of SunPower and Panasonic, the majority of manufacturers have used c-Si equipment from a select number of leading suppliers, enabling an almost de facto commonality for critical processing steps such as metallization and thermal diffusion. This has enabled the likes of Yingli Green and others to outsource without the need to specify specific in-house cell characteristics or provide third-parties with unique intellectual property. Indeed, even SunPower has adopted a joint venture approach to manufacturing its proprietary solar cell technology to lower capital costs, while Hanwha Q-Cells and Hanwha SolarOne are also sharing manufacturing IP.”

Although concentrating PV sector survivor/market leader Soitec may not be manufacturing anywhere close to the volume of a Yingli or even a SunPower, the French firm does have a claim to fame far beyond the capabilities of the c-Si set. The company, along with its research partners at Fraunhofer ISE, Helmholtz Center Berlin and CEA-LETI, has broken the overall PV cell conversion efficiency record with a verified 44.7% four-junction/subcell device measured under a concentration level of 297 suns. Just this past spring, the group announced a then-record 43.6% cell had been produced—meaning the team has pushed the numbers more than a full efficiency point higher in a matter of months, which says something for the efficacy of the company’s direct semiconductor wafer bonding technology.

A closer look at the charts provided in the Fraunhofer ISE version of the press release reveals a bit more granularity and cause for restraint. Although the bonding of two semiconductor crystals, “which otherwise cannot be grown on top of each other with high crystal quality” is touted, the precise formulation of the III-V compound material set is not divulged. However, the IV-characteristic graph for “the current best four-junction solar cell” shows the efficiency number with a standard deviation of plus-or-minus 3.1%, so there’s a lot of wiggle room in these experimental albeit confirmed results. Another stat of note: the fill factor for the 5.2 mm2 cell is a sturdy 86.5%; by comparison, Spectrolab’s C4MJ 40%-efficient production cell has a FF of 85.1%.

If the Soitec-led research team continues the torrid efficiency roadmap progress it has achieved of late, we could see 50% champion cells within a few short years. Although the theoretical path to that milestone metric seems clear, the ongoing velocity of efficiency improvements and the challenge of transitioning the breakthrough technology to volume production still remain works in progress.