With 22,000 visitors and 800 exhibitors, July's U.S. Intersolar show in San Francisco was a busy affair. Conference delegates heard how the U.S. solar market grew 67% in 2010 while demand for PV in the first quarter of 2011 was up 66% on the same period last year too.
The success of the California Solar Initiative (CSI) was noted, with 19,877 PV systems totaling 194 MW installed in 2010, while other discussions focused on issues like energy storage and, of course, on the policy and infrastructure required to create a 10 GW annual domestic U.S. market for the future.
But perhaps above all, the key theme of the week was innovation. And in this context, the key questions were can the Department of Energy's (DOE) SunShot Initiative target for module manufacturing costs of 50 cents per watt be met – and how?
Well, crystalline silicon was often hailed as the most promising answer.
Cell optimisation
Dr. Peter Wawer, senior vice president at Q-Cells is more than confident the goal will be met, with cell optimisation proving the critical factor. “We will see 50 cents for silicon manufacturing costs. Perhaps not this decade, but certainly in the next,” he said.
He outlined several areas ripe for cell optimisation on the front and rear side, as well as in bulk material quality, often referring to the company's Triple Q quality programme in the process. For hot spot detection, the company checks all cells under the programme because, as Wawer explained, localised overheating can inhibit performance and even destroy modules.
To prevent potential induced degradation (PID), it tests all of its finished modules to ensure they have been properly processed. Q-Cells maintain that these tests verify that modules do not degrade due to the presence of water, and that they can meet lifetime guarantees. “Those cells not resistant to PID lose up to 80% performance,” Wawer explained. “That's why it is important that such testing is standardised across the industry.”
Finally, the company makes a watermark ID on every component to attain full accountability on all long-term quality issues. By gathering this kind of data, the company has been able to note how efficiency relates to such factors as the height of the ingot – certain heights showed consistently better efficiency than others.
Q-Cells is also one of a growing legion of vendors adding lasers to its process. It begins with a thin dielectric layer, adds silicon and uses laser-fired contacts. Wawer said it is a faster process, and he reports average efficiency levels of 18%. And by transposing this approach from multi-crystalline to mono, the company is exceeding 20% efficiency.
“There remains huge room for improvement in the cell and the module,” he added. “Additionally, alternatives to silver is a big area to reduce cost and increase efficiency.”
Enhanced wafers
Meanwhile, Suntech unveiled two new products designed to improve the levelised cost of electricity (LCOE). The 72-cell Suntech 290W Vd Series module harnesses SuperPoly processing technology, which is characterised as a silicon casting and wafering process that produces higher-quality multi-crystalline wafers. This enables higher power output and resistance to light-induced degradation (LID), and represents the commercialisation of a laboratory-developed silicon casting using modified casting equipment.
“SuperPoly is a hybrid of mono-crystalline and multi-crystalline technologies that offers low-cost casting at high quality,” said Dr. Stuart Wenham, CTO of Suntech. “It operates at temperatures of around 1400°degrees C and takes two days to produce a crystal.”
Multi-crystal, he said, requires much less energy and is good quality, but not quite as high quality as mono due to random orientations. “We are working to improve the crystallisation of multi-crystalline by encouraging it to have one orientation,” he said. “This adds up to a 5% to 10% increase in efficiency and is no more expensive.”
SuperPoly technology grew out of a group of expired patents from 20 years ago (not exclusive to Suntech). While many companies are now taking advantage of this approach, Wenham claimed his company is ahead of everyone else in quality and scaling. “This wafer type will dominate industry in the near future,” he added.
Meantime, Suntech's 60-cell HiPerforma 245 watt module was also launched at the show. This uses Pluto cell processing technology involving a proprietary metallisation process – Pluto is inspired by PERL (Passivated Emitter Rear Locally Diffused) technology developed by the University of New South Wales in Australia, which Suntech has now commercialised as Pluto.
The result is grid contacts thinner than 30 microns wide, about a quarter the size of traditional screen-printed cells. These contacts are made primarily of copper instead of silver to reduce shading on the cell surface, thereby allowing the cells to absorb more sunlight and generate more electricity.
Wenham said Pluto heightens performance by 10% compared to traditional solar cells, even under low light conditions. Further, it should generate 2%-5% more power-per-Watt peak over time, due to better spectral response across all wavelengths of light and high shunt resistance. “The spectral response is stronger than with other solar cells on cloudy days or with a redder sun,” said Wenham.
He covered the limitations of the metallisation approach of conventional screen-printed solar cells. When a silver paste is used, he explained, the wide lines shade too much of the top surface. Suntech's process produces 25 micron metal lines (100 microns is the norm) spaced 1 mm apart. By using mostly copper and only a little silver, costs stay down. Heavy doping is only done under the lines, not across the rest of the cell in order to minimise damage to the silicon.
SuperPoly is compatible with Pluto but has not yet been used with it. As Suntech scales up SuperPoly, it plans to make it available for Pluto modules.
“Suntech's HiPerforma 245W module with Pluto cell processing technology fills the wattage gap that exists today between 235W and 270W modules,” said Ed Merrick, vice president of marketing and business development for Trinity Solar, a designer and integrator of solar electric systems in New Jersey: “Our preliminary view is that this will translate into a reduction in balance-of-systems costs, a reduction in inventory levels, and allow us to compete more effectively as we drive to grid parity.”
Reduced melting temperature
But Suntech and Q-Cells aren't the only game in town when it comes to better silicon. Calisolar has added aluminum to the silicon to reduce the melting temperature from 1500°C to 900°C. Treatment with hydrochloric acid comes next along with the removal of aluminum. The firms says all this results in far lower startup costs too.
“The CAPEX is US$20 per kg compared to US$100-US$200 for the Siemens process,” said Dr. Kamel Ounadjela, co-founder and chief development officer of Calisolar. “You can also build a plant within one year compared to two to three years for traditional plants.”
In terms of energy usage, the reduction in temperature means on average 20 kWh/kg is consumed as opposed to 120 kWh per kg for pure silicon. And no hazardous gases are involved. Performance wise, Ounadjela said this drops costs to two thirds of other types of cells.
Average efficiency is currently 16.5%, up from below 16% last year due to better quality silicon and advances in cell processing, he added. The highest cell efficiency recorded to date using this process is 17.82%.
Calisolar has also been working on the creation of modules of 60 to 72 cells, to increase the voltage generation. Moving forward, silicon doping, wafer texturisation and the emitter profile are all areas the company sees as providing further gains.
“LID is increased by oxygen content in the cells, so we have been working to reduce the amount of oxygen,” said Ounadjela. This has boosted cell efficiency by 0.4% so far. The thickness of the wafers is in the range of 190 micrometres. Low light performance is another zone where progress has been made with this methodology.
Automation for reliability
Other companies are working on efficiency on a variety of fronts. Instead of the commonplace 2-busbar cell architecture, for example, Conergy uses a 3-busbar design as a means of boosting reliability and efficiency. The company has also instituted robotic-based automation to engineer repeatability and greater speed into its assembly line: “We have a fully-automated and integrated process for producing silicon, cells and modules,” said Dr. Karl Heinz Kuesters, head of technology at Conergy.
In the cell department, the company is developing metallisation pastes to minimise breakages in crystals caused by stress along the interface between silicon and the metal paste/ribbon. Extensive electroluminescence testing reveals any flaws at the QC stage.
Kuesters is a fan of large amounts of testing to detect module failures before they ever make it into the field. Hail tests, for instance, simulate 55 mm hailstone storms, bombarding the modules to ensure they can stand up to extreme weather: “We need to develop standardised tests across industry to improve overall quality,” said Kuesters.
Like others covered above, Conergy is also investing in the creation of mono-like multi crystalline silicon using a multi-crystalline process. Kuesters estimates that further improvements in this sector can bring about an efficiency hike of 0.5%-0.9%. He's talking about cell efficiencies attaining more than 18.5% for mono-crystalline and more than 17% for multi-crystalline.
After covering the importance of weak light efficiency and achieving a better contact for selective emitter technology using lasers (a gain of about 10 milliVolts), he circled back to overall costs, which he said continue to be driven mainly by the price of materials – especially silicon. “When you begin using less pure silicon like Calisolar, it is even more essential to watch out for supplier quality,” Kuesters added.
However, by 2020, he estimates the industry will be at 40% of current costs.
Rose coloured spectacles
Not everyone is so sure. Thomas Surek of Surek PV Consulting thinks the SunShot target is unrealistic. As a 30-year PV programme veteran of the DOE's National Renewable Energy Lab (NREL), he does however expect COE can be brought down considerably via R&D. He drilled into the numbers to see how low the industry could go in the area of crystalline silicon.
“Last year, costs were estimated to be a little under US$2 per Watt to manufacture silicon,” he said. “Every 1% gain in module efficiency is worth about 10 cents per Watt at the system level.”
In the past year, he continued, incremental improvements in process and performance; more vertical integration; and oversupply across the value chain have driven costs down a little. He believes that in two to five years, with silicon feedstock prices continuing to drop, as well as developments in ingot growth and wafering, the cost of silicon manufacturing could hit the US$1/W range.
But the DOE, he explains, wants a US$1/W total for all manufacturing costs i.e. 50 UScents for the module, 10 UScents for power electronics and 40 UScents for balance of system and installation. Many manufacturers are successfully eking out steady but small gains in efficiency, he said. And each 1% in efficiency is also worth about US$20 per kg in lowered silicon feedstock costs, or about 10 UScents per Watt at the system level.
Surek outlined other areas of possible gain: Kerfless (no wastage due to cutting the silicon) approaches; improved light management; selective emitters; module packaging and longer-lifetime modules. In addition, wafering advances (or elimination) could perhaps boost efficiency another 1.5%.
“Higher quality multi-crystalline silicon ingots, which are mono-crystalline like, is one of the most promising areas,” he said: “And maybe it would be possible to use fixed diamond abrasives to produce thinner wafers.”
Even wearing rose coloured spectacles, Surek doesn't think all this will achieve the goal though. By looking at all the possible areas of shaving costs, manufacturing costs might get down to 60 UScents per Watt for silicon, he suggested. But that is still a very optimistic outlook, he said, and will probably take five to 10 years to occur.
That said, he believes silicon will continue to be the dominant PV technology for the foreseeable future. And he doesn't think the DOE target needs to be made to bring the technology into a cost effective range.
“The DOE goal may look possible on paper but it is not realistic due to the trade-offs involved,” Surek said. “In my opinion, the DOE programme has given too much emphasis to blue sky initiatives that won't be ready for 30 years.”
He wants the focus on proven areas such as silicon and thin film where R&D can bring immediate gain.
NB: If you are interested in the DoE's SunShot initiative, we covered it here:
SunShot: Solar PV's falling costs, part one, two and three;
About the author:
Drew Robb is a graduate of the University of Strathclyde in Glasgow. Currently living in Los Angeles, he is a freelance writer focusing on engineering and technology.