Automation Cuts Solar Costs, Levels the Playing Field
The solar industry needs to cooperate on automation if it expects to bring down PV's cost-per-watt to parity with conventional electricity-generating methods. Currently, the solar industry labors under islands of automation, single automated tools or clusters, and almost as many ways of linking those islands as there are PV manufacturers, experts said.
Paula Doe, Contributing Editor, SEMI, San Jose -- PV Society, 6/12/2009
Participants in the solar industry may agree that integration and automation are key to bringing costs down to grid parity, but there are plenty of different views on how best to do it.
Currently, there are islands of automation, single automated tools or clusters, and almost as many ways of linking those islands as there are photovoltaics manufacturers. "In fact, there are probably more ways than there are manufacturers," said John Hennessy, regional sales manager at The Peer Group (Kitchener, Ontario), which supplies factory automation software across industries. "I know of PV plants that have multiple ways to move data, to track work in progress. In some cases there is an overall design, in other cases there is a patchwork. Even at the MES level, there is no real consensus — we have helped in a couple of MES selections recently and there was only one contender in common between the sets of candidates."
Problems getting the tools to talk to each other have sometimes caused months of delay in ramping production. One defacto solution for getting tools to talk to each other is turnkey lines from single suppliers, a point demonstrated by the preliminary list of top PV manufacturing suppliers from VLSI Research Inc. (Bedford, England). Eight of the 10 largest solar suppliers are marketing crystalline or thin-film turnkey lines with some degree of automation.
Suppliers report that most new entrants are buying turnkey systems, and most new plants now being built are at least 50 MW in size, and are being automated from the beginning. Existing PV producers are adding process equipment, and from ~25 MW scale or so are also looking at retrofitting automation on key processes where it has the most impact, including taping, busing and framing, to remain cost-competitive with the new automated plants.
"Solar module sales prices are coming down, but if you can run 24/7 with the same number of people and the same equipment, you can increase tool usage from about 70% in a non-automated line to almost 100%," noted Patrick Hofer-Noser, CEO of 3S Swiss Solar Systems (Lyss, Switzerland). "The capex is not so big compared to the added volume."
Hofer-Noser started 3S to make building-integrated solar systems, and decided his experienced engineers could build a higher-yield laminator in house than they could buy from equipment suppliers. He then integrated that laminator into a semi-automated module line. Customers who saw it asked if they could buy the whole line, so the startup sold it off the floor. "Two times we built lines for ourselves and then sold them," said Hofer-Noser, who got the message and got into the integrated automated line business instead. He acquired Somont's stringer/tabber business and Pasan's tester business to be better able to integrate the whole assembly process around in-house equipment.
Putting automotive and packaging robots to work
 |
|
ABB Group (Zurich) had adapted its 6-axis robots with different grippers for solar modules. |
First big strides in automation of crystalline solar production have come from putting robots from other industries to work. With investment in new automotive lines slowing, automotive robotics suppliers like The ABB Group (Zurich, Switzerland) and KUKA (Augsburg, Germany) have jumped aggressively into the growing solar market. ABB's six-axis robots need only different grippers or suction holders to handle the wafers, and some new programming, for the company to offer standard setups for trimming and taping, busing, and framing solar modules.
Though a big automation systems supplier, ABB doesn't plan to get into that solar market, but sells to crystalline solar line builders like Jonas & Redmann and Spire Solar, leaving the integration to them. "Our robots have been around so long in so many industries, they can talk to just about anything by just adding another communications module," said ABB's Keith Fox, newly in charge of business development for solar manufacturing in the United States.
Conergy AG automates its new 250 MW solar module line now ramping in Frankfurt, Germany, with clusters of ABB six-axis robots to do the initial clean, the taping and busing, and the trimming and framing, churning out a 220 Wp module every 20 seconds. Four robots with vision systems reportedly replace 30-40 machine operators to do the labor-intensive taping and busing of wafer strings into matrixes with more reliable accuracy. But even in this big automated plant running 24/7, some steps like back lamination and sealing after trimming are still done manually.
The main solution for moving crystalline solar cells between process tools is pick-and-place Delta parallel robots and their relatives from the packaging industry, the same kind of fast-moving handlers that put candy bars into package trays at up to 150 picks per minute. Q-Cells' 240 MWp line installed in 2007 in Thalheim, Germany, uses 36 of ABB's FlexPickers to load and unload the wafers from the wet process steps. Others mainly use the high-speed handlers for shuffling the wafers through testing and sorting.
Redesigning tools for solar automation
|
| Adept Technology developed pick and place robots for the the solar industry. |
Instead of just adding robots to automate the traditional manual process, others are redesigning the equipment to integrate the process steps more efficiently. Diversified robotics supplier Adept Technology (Pleasanton, Calif.) also aims to sell its overhead parallel Quattro pick-and-place robots to the solar industry, but it has added specific solar wafer inspection systems to check for defects on the fly.
The company improved the resolution of the onboard vision system so the handler can now not only locate the wafers and notice major defects like missing corners, but also inspect for micro-cracks. "Automation and inspection have to go together, with the belts kept running at full throttle, and all inspection between tools," said Peter Lundquist, marketing consultant for Adept. "The goal of automation of a solar line should be 100% inspection with no slowdown in the production line."
He noted that although improving the resolution is not trivial, the key is really the intelligence of the control system, and how tool talks to tool. The company aims to sell the tool through line builders, and there the slowdown has helped get the new tool into evaluation. CEO John Dulchinos told analysts on Adept's most recent conference call that line builders were too busy to talk in the fall, but now are evaluating the Quattro inspection systems. The company aims at getting design-ins for the next investment cycle.
The most important place to first focus inspection for best return is the front end, so the plant knows the quality of product that it's starting with, Lundquist noted. And the key metric to track is not just the micro-cracks, as all substrates will likely have some level of these defects, but rather how they propagate and change during processing and how that influences yield.
3S tried to rethink the whole process for the integrated automated module line it introduced in August, according to Hofer-Noser. It eliminated a step from layout and busing, and reduced breakage by not making holes in the cover glass and EVA to bring the contacts out of the module sandwich as usual. Instead, it uses sensors to detect the position of the contacts after processing, and opens the backsheet to reach them. Similarly for trimming, instead of trying to detect an edge with few distinguishable reflective properties with a camera, it marks the edge before processing so it only has to find the mark, which reduces cycle time. Modules can be integrated into the stringer for optical inspection, electrical testing of finished strings, or full current-voltage (I-V) curves.
Product is moved through the process equipment with some gantry-like robotics and some belts, with inline inspection and tracking, using software based on that from the automotive industry. Hofer-Noser said that the company tries to avoid using six-axis robots if possible, since more axes mean more variability. The company announced the first sale of a segment of the new integrated automated system — the lamination through packaging flow — to a customer in Germany in February.
Integration cuts costs
Taking the idea of integrated turnkey line efficiency to its logical extreme, Centrotherm Photovoltaics (Blaubeuren, Germany) argues that vertical integration of the entire production process — from polysilicon to finished module — within one company will bring costs down to grid parity for many regions. In a conference call detailing the plan to analysts, CTO Peter Fath cited not only savings on things like sales, shipping and incoming/outgoing inspection, but also argued that better integrated processes with fast feedback loops across the company could optimize product yield, and expanding a single factory management system across the whole value chain would improve logistics.
Centrotherm aims to supply this mega integrated turnkey system, though it doesn't actually yet have all the necessary technologies in house. The company reported that its first offerings in module and polysilicon equipment are about to start production at customers, and it's eyeing acquisitions in automation, as well as chemical etching, wet cleaning and module assembly.
Since introducing the idea early this year, the company says its has five customers seriously considering the idea and in various stages of negotiations, some looking at investing ~$280M for full vertical integrated production of 60 MW of modules, others interested in combining only two or three steps in the value chain.
Flexibility allows enhancements
There is also plenty of push for more flexible lines that can incorporate changes to improve cell efficiency and module yields, to give the producer some competitive advantage besides geographic location over all the other companies using the same turnkey line.
"We get many very different requests from customers," said 3S's Hofer-Noser. Some users want SEMI-10, but Hofer-Noser argues that although SEMI standards are good for cells, they're too much for the module business, which has more in common with the auto and glass industries. "We have ongoing discussions with users on how standardizing on one size would help simplify packaging technology," he said. "But if they have a reactor that can do a larger substrate, they win some efficiency, and efficiency is a more important driving factor than capex costs. So we get requests to make machines that can handle substrates that are only 5 cm bigger."
The PV industry has ramped so quickly in the last couple of years that there really has not been time to think about factory-scale automation or standards in detail, said Peer Group's Hennessy. The level of automation will determine in part who will be the low-cost producer, and a well-designed factory automation system will be even more important for easily replicating the lines to get the second and third plants up and making money quickly. But there's no IP in getting the furnace to talk to the saw, he said, so engineers should use one of the solutions someone has already figured out, and spend their time on the things more likely to give competitive advantage. "Everybody will end up with an automated factory," he said. "But only one company will have the most efficient cells or the best price/performance cells in each of the market niches."
