WASHINGTON – Squeezing another penny out of the cost of conventional solar cell manufacturing is no longer a viable approach for cleantech companies that have seen the price of solar panels drop by more than 40 percent since 2011. Addressing that reality, a manufacturing equipment maker promises to shift the focus of solar and chip manufacturing with a wafer production machine that it claims can greatly reduce silicon and other substrate material costs.
Twin Creeks Technologies (San Jose, Calif.) also claims its ultra-thin manufacturing process is suited to advanced production of LEDs and CMOS image sensors.
Reducing the amount of silicon needed to make solar cells is critical, argues Siva Sivaram, a former Intel manager and CEO of Twin Creeks, since it represents most of the manufacturing cost. Moreover, current solar manufacturing processes use only the upper layers of a silicon wafer, and Sivaram estimates that 90 percent of substrate is not used in the manufacturing process.
Twin Creeks has leveraged a technique called proton-induced exfoliation (PIE) to build a machine that produces extremely thin, flexible wafers. The company is taking the wraps off of a third generation version of its Hyperion wafer production tool this week (March 13), a machine that is says it ready for production. The company’s goal is to use its manufacturing and wafer handling and processing tools to produce monocrystalline wafers less than 1/10th the thickness of conventional wafers currently used to make solar cells.
Such an advance could help reduce solar cell material costs to the point where devices based on ultra-thin wafers would offset much of the price decline seen in the solar panel market over the last 18 months. Twin Creeks claims it lower cell production costs by about 50 percent.
The company’s version of PIE technology uses a beam of hydrogen ions to lift a super-thin layer of silicon measuring as little as 20 microns off a wafer that serves as the basis of a solar cell. The company claims the Hyperion machine can produce wafers for solar cells at a cost of about 10 cents per watt. The current cost is about 85 cents per watt. “I don’t want to compete [in the solar market] in taking out the last penny of cost” using traditional manufacturing techniques, says Sivaram.
The third-generation Hyperion tool is a “platform-level technology,” Sivaram insists. “We are now in manufacturing,” and the company expects to begin shipping systems this year.
Other solar startups are taking different approaches to reducing materials cost by using thinner wafers. For example, 1366 Technologies makes thin wafers for solar cells from molten silicon. The company’s approach automates several manufacturing steps like crystallization furnaces, sawing stations and ingot handling. Another solar manufacturer, Astrowatt, claims it can produce 25-micron-thick wafers.
Along with solar photovoltaics, Twin Creeks also is touting its wafer tool for manufacturing concentrated photovoltaics, LEDs, power electronics, 3-D packaging and CMOS sensors. The Hyperion system also handles other semiconductor materials like gallium nitride and germanium.
For now, Siveram says Twin Creeks is targeting the growing Chinese solar manufacturing markets. Being able to add value in the manufacturing process “resonates with the Chinese,” he adds.
Well, after all of our billions spent through American Recovery Act dollars, most were a waste with very little permanent job positions. Solar STILL does not do much for job creation in the US. This technology, while creating a nominal number of US jobs at least allows us to export technology and receive some trade balance dollars. (Until the Chinese rip off their technology.) Good for Twin Creeks and their innovation. I wish them the best
Great, high-level discussion, folks. The consensus seems to be that Twin Creeks and others are on the right track in terms of cutting materials costs for solar PV. One commentator asked about government backing of Twin Creeks. Don't think so, but we'll find out.
Seems to be the same basic method as used in the SOITEC SOI wafer fabrication scheme but with a twist. That's smart.
The thickness has so far to some extent been set by the wafer handling in manufacturing. Silicon solar cells have typically a screen printed upper electrical contact. This required from a processing point of view a mininum wafer thickness to avoid excessive cracking.
A new non direct contact metallization method based on inkjet printing has been demonstrated by Fraunhofer and it seems to work as a charm. So the minimum thickness requirement isn't much of an issure anymore.
This is why I'm a strong believer in silicon PV. Good enough performance, multidecade processing experience and constant innovation from that incredible knowledge base will be hard for the thinfilm competitors, that essentially have to invent their own manufacturing scheme, to compete with.
Yes, it's somewhat boring compared with the fascinating CIGS or CdTe but in the end price matters and the price of silicon solar will easily match anything the thinfilm companies try to do without having to invent a completely new manufacturing system.
This is good work. Semiconductors can shrink in die area for cost reductions, but solar cells can't. It good to see that they can shrink in thickness for cost savings. I don't think they will be able to follow Moore's law that way, but at least they have some incremental improvement.
There are hints that TwinCreeks has inhouse ( via an inventor roster ) a large ion implanter design team with history back to Varian /Extrion and Eaton /Nova, and are likely using an RF Linac possibly designed inhouse or at a Northern California Proton Linac manufacturer near Pleasanton.
If SiGen is doing thin wafer exfoliation from ingots, this might be less desirable due to beam power limits and wafer transfer removal overhead than a batch ion implanter wheel for exfoliation form thick wafer blocks that Twin Creeks seems to be pursuing, based on references in their patents of ion implanter batch wheels of up to 50 wafer loads ( of comparable endstation to the Eaton Nova ion implanter spinning disks / other spinning wheel batch ion implanters used in conventional silicon production high dose ion implants )
I used to be a production process engineer in silicon ion implants, and recently came off 4yrs of cleantech technical due diligence, with ~80 reviews done, some 30 photovoltaics firms looked over.
TwinCreeks since they are intent on volume production seems more interesting than Sigen, but there are hints that SiGen might be working with SunPower. Progress is unknown.
TwinCreeks is unusual in that they built a large endstation to implant (for later peeling / transfer) many wafers in a single huge endstation. They use large thicker wafers ( thin blocks ) to peel ? 20-50? wafers via iterative ion implants / peel cycles.
SiGen might be peeling wafers directly from a single crystal ingot - end on in the ingot, or they might have built a large endstation for thick silicon wafer blocks much as Twin Creeks. Their powerpoints intimated they peel directly from ingot, but this might be less practical, since it is easier to mechanically scan ( rotating translating disk ) than electrostatically / magnetically scan a multiMev beam and maintain desireable constant angle of incidence desired for tighter implant depth control? Yet energy spread might be larger in an RF Linac - so the thickness uniformity results of linac (mech scanned ) vs electrostatic/ electromagnetic scanned DC accleration ( hypothetically ) are a bit unclear as yet.
Photos of the SiGen multimev ion implanter that were apparently published in the solar magazine Photon International, hints that the SiGen implanter is likely a more conventional ?tandem VanDeGraff DC acelerator ( there was a HUGE SF6 gas insulating vessel visible in the Photon International pictures ) which means it is possible parts of the SiGen implanter might have been contracted to National Electrostatics Corp in Wisconsin, a well known physics DC high voltage ion accelerator equipment desinger and manufacturer.
TwinCreeks PIE proton exfoliation is comparable to SIGEN Silicon Genesis' multi MEV Hydrogen Ion exfoliation, which is comparable to methods used by SIGEN and SOITEC in forming SOI wafers via high energy proton wafer bond and transfer.
Notably there was experimental multi Mev silicon exfoliation experiments done in Japan for examining this application for wafer peeling, in the mid 90s, with a static non scanned beam, that produced the requisite blistering in a multi 10's micron blister layer - that is indicative of what is needed for proton exfoliation for useful modest thickness silicon solar cells.
FWIW, making the peeled layer too thin does tradeoff thin-ness ( silicon cost ) versus upper PV % efficiency due to the indirect bandgap that Silicon is. The energies needed for useful silicon PV cells, are proton beams of order of 10 or more MeV particle energy, and this is done more cost effectively with Linacs ( RF linear accelerators ) since there is no need for a large SF6 dielectric gas insulation chamber otherwise needed for a DC accelerating voltage MultiMeV Ion Implanter.
But they do compete on scale when U.S. develops and opens factories in China. Immediately, China takes over and begins to lead. They do not need to invest in R&D, just for U.S to do that and then ship the idea to them
First of all, congratulations to Twin Creaks!
As you have mentioned, I agree that it will be difficult to keep this in US. I depends on how much money did Twin Creaks invest behind the development of this new process technology and who funded that money. Any stake of US Govt in it? If not, then, won't it be difficult recovering the investment if selling this to one of the fast growing manufacturer & market such as China is avoided?
We are seeing a growing trend in the solar industry toward new materials and reducing material requirements for silicon wafers. The other trend is adding value in the manufacturing process. To remain competitive in the solar business, companies like 1366 Technologies and Twin Creeks are developing tools that will allow solar companies to add value in the manufacturing process. One challenge will be keeping the intellectual property behind these new processes and tools in the U.S. For now, China can't compete with the U.S. on manufacturing innovation.