Despite surface similarities to processes found in semiconductor fabrication, solar panel manufacturing facilities require a less rigid approach to contamination control, where contaminants have a smaller impact on yield.
By Sarah Fister Gale
As oil prices continue to climb and environmental concerns grow, the demand for and interest in solar power as an alternative to fossil fuel has increased dramatically. Solar energy demand for residential and industrial applications has grown 20 to 25 percent per year over the past 20 years. As its popularity increases, the price for solar has decreased due in large part to increasing efficiencies of solar cells, improvements in manufacturing technology, and growth in economies of scale.
Supporting this growth, companies have invested millions of dollars in new solar cell manufacturing facilities and silicon development over the past several years to meet the demand of this burgeoning market. But supply and demand challenges, particularly for highly pure silicon for wafers, continue to plague this industry as it struggles to find its footing.
Solar strives for greater efficiencies
Today, renewed interest in alternative fuels, coupled with regulatory requirements that cities and corporations invest in environmentally sustainable energy supplies, has drawn attention once again to solar power.
Solar cells are essentially semiconductors, conveying electrons from one place to another. Solar technology requires no toxic fuel and relatively little maintenance, is virtually inexhaustible, and, with adequate financial support, is capable of becoming directly competitive with conventional technologies in many locations. These attributes make solar energy one of the most promising sources for many current and future energy needs.
Early photovoltaic applications used to capture the sun’s energy were extraordinarily inefficient; however, the more recent advent of the transistor and accompanying semiconductor technology boosted the efficiency of photovoltaic power dramatically, making it a realistic solution, with steady increases in efficiencies on an almost monthly basis.
For example, last December, researchers at Boeing-Spectrolab in St. Louis, MO, produced a multi-junction photovoltaic cell that achieves 40.7 percent efficiency, which is twice the efficiency of Silicon Valley-based SunPower’s 22 percent efficient cell, which in itself was a breakthrough just a few months earlier.
Similarly, scientists at Lawrence Berkeley National Laboratory (LBNL) recently created a new type of semiconductor material designed to improve the efficiency of solar cells by capturing low-energy photons. Traditional solar cells respond only to a narrow spectrum of sunlight, making them highly inefficient, while photons with lower energy pass right through the material.
The new semiconductor material can capture these low-energy photons for electricity, which could result in solar cells with efficiencies around 45 percent, compared with 25 percent for conventional cells that use a single semiconductor and 39 percent for cells with layers of mixed semiconductors.
Cost vs. clean is always a concern
In general, the solar manufacturing process is similar to that of semiconductors, with the fabrication of highly sensitive wafers taking main stage in the production environment. Although the processes are somewhat similar, the economics of the two fields are different. Chipmakers require cutting-edge equipment and extremely clean environments to protect delicate machinery and chips from even the smallest contaminants. Solar cell manufacturers, on the other hand, don’t face nearly the same stringent contamination control requirements because contamination doesn’t have as significant an impact on yield.
A single wafer of finished semiconductors may be worth several thousand dollars; a wafer of solar cells isn’t valued nearly as high. Solar cell manufacturers are more concerned with producing large volumes of wafers, so they don’t have to stay on the same cutting edge as the chipmakers.
In the solar industry, manufacturers produce photovoltaic power using either discrete cell technology or integrated thin-film technology. With discrete cell technology, manufacturing facilities use single-crystal silicon wafers that are sliced from single-crystal boules of polycrystalline grown silicon, as thin as 200 microns. The starting material to manufacture silicon wafers is chunks or granules of chemically ultra-pure silicon. Research cells have reached nearly 24 percent efficiency, with commercial modules of single-crystal cells exceeding 15 percent.
Other facilities use multi-crystalline silicon sliced from blocks of cast silicon. These wafers/cells are less expensive to manufacture, but they are also less efficient than single-crystal silicon cells. Research cells approach 18 percent efficiency; commercial modules approach 14 percent.
The grade of silicon used directly affects efficiencies because the lower grade, or “solar grade,” silicon tends to have higher levels of metals and impurities in the material, says Bala Bathey, senior process engineer at Schott North America (Billerica, MA), a leading solar industry company. “The typical solar grade silicon isn’t very good because of the purity issues.”
Bathey points out that solar grade silicon has levels of metal contamination that can reach 1,000 ppb. Transition metals are one of the main culprits in degrading the efficiency of multi-crystalline solar cells, and studies have shown that the size, spatial distribution, and chemical binding of metals within clusters is just as important as the total metal concentration in limiting the performance of multi-crystalline silicon solar cells.
Figure 1: Solar-grade silicon may contain metal contaminants on the level of 1,000 ppb. Photo courtesy of M+W Zander.
The solar industry has historically taken this off-specification material that is rejected by the semiconductor industry; however, the drive to increase efficiencies of solar cells is pushing many manufacturers to invest in higher-quality silicon. “We use an electronic grade silicon, which has 100 ppb of impurities, to achieve higher efficiencies,” Bathey notes. Semiconductor grade silicon has less than 30 ppb of impurities.
Contamination control requirements
Regardless of the efficiency or purity level of the silicon for wafers, conventional solar cell manufacturing facilities rarely implement formal cleanroom conditions, relying instead on standard operating procedures for cleaning and maintenance to keep contamination at bay, says Robert Gattereder, managing director at M+W Zander FE (Stüttgart, Germany).
“Yields are already generally high in the solar industry, so that the more critical question than contamination control is effective production,” he says. “Standard contamination control issues, as known from semiconductor manufacturing, are normally not an issue. There is no general need for cleanroom installations because you can use standardized processes and solutions from other industries in order to avoid any damages caused by contaminants.”
However, that’s beginning to change. “Some photovoltaic manufacturers generally do need lower-class cleanrooms, to create separation of the operator and the product for standardized handling procedures and precautions,” Gattereder says.
He adds that the biggest contamination control challenges in the solar cell manufacturing process result from the combustible dusts that occur when slicing silicon blocks or ingots; gases, such as nitrous used in processing steps; and possibly heavy metals released in the environment. “But there is a big variety of manufacturing processes, and each manufacturer is mastering this,” he says.
Figure 2: Inside these vacuum chambers, solar cells are deposited simultaneously on six rolls of stainless steel, each 1.5 miles long, to make 9 miles of solar cells in 62 hours. Photo courtesy of United Solar Ovonic.
Dave Genova, project manager for Spire Corporation (Bedford, MA), a solar equipment and processing company, has had direct experience with the impact of metal dust on wafer processing steps. He’s had several clients experience contamination problems as a result of construction dust in the manufacturing environment. “We’ve run into situations where metals from construction dust, which can include titanium, silica, magnesium, and chromium in the parts per million or even parts per billion, hurt the wafer in a solar manufacturing line,” he says. “The metals interfere with the conductivity of the wafer. That can kill your productivity.”
For Genova’s clients, although the construction dust was being generated in another part of the facility well away from the manufacturing area, it was drawn into the duct system of the building and permeated the facility, pulling particulates into the air around the wafer processing steps. “When that happens you just have to wait for the metal to come out of the building through the ventilation system. That can take days and, during that time, work production goes way down,” Genova says, adding that metal contaminants from construction dust can drop wafer efficiencies from 14 percent to 13 percent. “Over the course of a year that can add up to a billion dollars,” he says. “A Class 100,000 [ISO Class 8] cleanroom would eliminate that problem.”
If a solar cell facility is doing construction work, even if it’s as simple as knocking out or replacing dry wall, Genova urges them to secure the area, sealing off the space with plastic barriers, taping doors and gaps in the floors and walls, and using blowers to pull dust out of the air and siphon it out of the facility, creating negative pressure in the affected room. “We’ve learned our lesson over the years. Dry wall dust kills wafers.”
Spire also now recommends to any solar cell manufacturing clients that they invest in a Class 100,000 (ISO Class 8) environment to ensure that no titanium or other metals present in dry wall can interfere with manufacturing. “It doesn’t cost a lot to install a Class 100,000 [ISO Class 8] cleanroom, and it ensures a smooth running process,” Genova says.
Along with maintaining dust-free air, these facilities also need to monitor and control ionic contamination, humidity, and temperature to protect the wafers during processing steps. “You want to avoid extreme heat or moisture because it can cause oxidization on the wafer that will cause it to be less robust,” Genova says.
Simple sensors can track temperature and humidity; ionic contamination is monitored by standard offline techniques such as impinger sampling and ion chromatography, adsorption tubes combined with thermodesorption gas chromatography coupled with mass spectrometry, notes Gattereder.
Other volatile organic compounds, such as oils used in compressors, can also cause problems in the wafer line if they are left unchecked. “Eventually these oils always get through the traps and into the line,” Genova says. “It makes the wafer slippery on contact, and when you put the metal grid lines on the surface of the wafer it will move or peel.”
Instead, Genova suggest clients invest in oil-free air compressors to avoid such problems. Similarly, he suggests that clients remove all equipment that uses Freon® or chlorofluorocarbons because they can also coat the wafer and make them less robust if they are released into the manufacturing environment.
These issues are still relatively minor for the solar industry; however, as it continues to expand and costs continue to drop, Genova predicts that, just as the semiconductor industry has pushed for smaller and thinner geometries, so too will the solar industry. “Silicon is so expensive, you want to cut the wafers thinner to generate more revenue and less waste,” he says.
With the thinner wafers, however, he expects processes, contamination control strategies, and environmental controls to get stricter. “When processes are tightened and more electricity is being produced by each wafer, the environment is less forgiving about contamination. Any mistake can affect yield.”
Genova also notes that, as manufacturers follow the trend to move their facilities to harsher environments, managing particulates will become much more challenging. “If you build a facility in the desert, you can’t walk into it without a gust of air coming in with you,” he says. “In those environments, facilities will need sealed entryways, locker and gowning rooms, and a cleanroom on the production floor.”
Still, his predictions for solar manufacturing aren’t entirely dire. “When it comes to contamination, the solar industry will never be like the semiconductor industry. It will never get that bad.”
Fears about contamination are certainly not slowing growth in this booming industry. With industry speculation forecasting continued steady growth, significant investments are being made to build and expand solar cell manufacturing facilities.
But the continued growth, combined with the drive to improve efficiencies and the desire to use higher-quality silicon, has had the effect of leaching much of the available silicon out of the market, leaving some facilities scrambling for silicon resources to meet demand. By 2010, many solar cell manufacturers are expected to be running solar plants with 10 or more production lines, capable of producing 100 megawatts worth of solar cells annually, but each watt of power requires 7 g of silicon, which means that each 1,000 megawatt plant will need 7,000 tons of processed silicon a year. Since 2004, demand for solar cells has outstripped supply, causing the price of silicon to skyrocket and access to shrink.
In response, many companies are picking up the slack, building silicon manufacturing operations to meet their own demands. Wacker Chemie AG (Munich, Germany) for example, recently announced that it would expand its polysilicon production capacity by an additional 4,500 metric tons to a total of 14,500 metric tons by the end of 2009.
Similarly, Prime Solar Power (Perth, Australia) has begun plans for a polysilicon manufacturing facility in Bitterfeld, Germany, to produce 7,000 metric tons of polysilicon per year, after a plan to launch a wafer manufacturing facility in Thalheim, Germany, was shelved in 2006 because the company was unable to secure polysilicon feedstock.
The thin-film alternative
Other photovoltaic manufacturers, such as United Solar Ovonic (Auburn Hills, MI), have opted for a different solution: using integrated thin-film technologies instead of conventional silicon wafers to produce solar cells.
“Conventional silicon wafers are 200 to 300 micrometers thick, but we use a thin film of silicon that is half a micrometer thick, so the silicon shortage doesn’t affect us,” says Subhendu Guha, president of United Solar Ovonic.
In the company’s processing environment, rolls of stainless steel sheets a mile and a half long are drawn into a 250-foot vacuum chamber. Using a plasma chemical vapor deposition process, silane gas is pumped into the chamber and voltage is applied to break down the silane into silicon and hydrogen. The hydrogen is pumped out of the chamber and burned while the remaining silicon is deposited as a film onto the steel sheets.
The thin-film cells are less efficient than conventional silicon wafer cells but the manufacturing costs are much lower, giving consumers a better dollar-per-kilowatt-hour ratio, Guha says. “We recognized from the beginning of operations in 1986 that material costs would be an issue in this industry. When you process, grow, and cut silicon into wafers, there is a lot of loss. We wanted a simple, robust, low cost operation, which is why we chose thin film.”
Because the thin-film process is completely automated and the most critical steps-when silicon is deposited onto the steel sheets-occur inside the vacuum chamber, there is little concern about contaminants in the environment.
Members of the operations team overseeing the 62-hour automated process monitor the chamber from a control room, where they can track humidity, temperature, film thickness, and gas pressure on computer screens to ensure that the environment meets quality processing standards. “If there is a problem in the chamber that doesn’t self correct, a warning will go off in the control room and the team can respond to the problem,” Guha says.
Once the film is deposited on the stainless steel sheet, the coated steel is removed from the end of the chamber in a roll so the delicate surface has minimal contact with the atmosphere. The roll is then sent through another automated vacuum chamber where 9×14-inch slabs are cut from the sheet and a layer of electrodes is applied, followed by a polymer encapsulate with an adhesive. “Once the encapsulate is applied, the cells can be touched without risk of damage,” Guha says.
The thin-film environment has the added advantage of being a predominantly closed system requiring no additional cleanroom spaces, and a smaller manufacturing environment, says Kees Jan Leliveld, head of development and engineering products for Bosch Rexroth Electronic Drives and Controls, Product Area Semiconductor and Medical, located in Eindhoven, the Netherlands. Like the system United Solar Ovonic uses, Bosch Rexroth’s Linear Motion System is a modular transportation system for solar wafer manufacturing that uses intelligent coils outside the vacuum environment, where all of the electronics are located. Only the moving parts necessary for conveying the products are located inside the vacuum along with sensors to monitor temperature, pressure, vapor deposition, and silicon thickness.
“The clean environment is miniaturized, which offers a big economical advantage,” Leliveld says. “Because it’s done with vacuum control, and it’s completely automated, you don’t need a cleanroom.”
The future of solar
Whether facilities are using thin-film technology or silicon wafers, the trend toward rapid growth means everyone needs to find ways to improve efficiencies and cut costs. To do this, Gattereder suggests looking to industries that have already paved the way. “There are a lot of parallels between the photovoltaic industry and semiconductor and flat-panel display industries,” he says. “They also began with smaller fabs, and today many companies have huge fab complexes, some with their own energy supply centers. The photovoltaic industry can learn a lot from these industries.”
But he concedes that there are also a lot of differences. “We have different cleanroom needs and we are pushing to simplify processes, not only in avoiding contamination but also in recycling materials to save costs,” he notes. “And even as we move to a tenfold increase in production in the solar industry, which will, of course, lead to new dimensions in managing chemical and metallic materials, we are in discussions with manufacturers and equipment suppliers on these issues and are sure that contamination problems will continue to stay manageable.”