Leland Teschler, Editor
Solar automation in a nutshell
Cell soldering, module conveyance is manual in plants up to about 25 MW.
Full automation becomes practical in plants above 50 MW.
Solar wafer thickness is unlikely to go much below 160 microns.
Next Big Challenge for PV Makers:
Wafer Handling, Machine Design,
National Renewable Energy Lab
solar research home page,
Spire Corp., Solar Operations,
The German solar maker Q-Cells AG has an interesting product. It is a multicrystalline solar cell roughly 7 in. square purported to convert about 14% of incident sunlight into electricity. This makes the device among the most efficient in its class.
You might think Q-Cells is happy with this state of affairs. You’d be wrong. The firm figures its costs are too high and says it aims to slash them by 50% within two years. In that regard it is pursuing avenues that include increasing the efficiency of its production and figuring out ways to make its processes more standard.
Q-Cells’ worries about manufacturing costs mirror those of other photovoltaics producers. Though the PV market is booming, the technology is not cost competitive with electricity generated by fossil fuels. Government incentives level the economics, but PV makers view these as unreliable. So to both get their costs down and meet mushrooming demand, they are automating production and trying to get their manufacturing operations into the 21st century.
“In some ways, their processes are not completely optimized,” says U.S. National Renewable Energy Lab Principal Project Leader Richard Mitchell. Mitchell coordinated a manufacturing R&D project credited with significant reductions in the production cost of PV modules. “Solar production processes are refined enough to make high-quality salable products but they will continue to be optimized over the next decade to get higher yields and lower costs. It is an evolution similar to what large manufacturers like automakers go through.”
Turmoil in PV manufacturing has brought a need for integrators experienced in automation. “PV makers have been in a rush to scale up production because they have presold the inventory of every plant they can build,” says Rockwell Automation Industry Solutions Manager for Solar and Semiconductor Bates Marshall. Rockwell has developed automation equipment for several main-stream thin-film PV suppliers. “PV thin-film people are focused on speed of integration, ramping up production, and expanding to support factories globally,” he says. “This contrasts with the semiconductor industry where there is still a great deal of invention going on in control systems because of a perception that more benefits can be had by inventing automation technologies.”
One factor complicating automation efforts in PV is that “In thin-film, manufacturing processes of different suppliers are completely unique. That’s why there are so many of them,” explains Marshall.
There is more commonality in crystalline production processes and today, they are more mature than thin-film lines. “The layout of different PV module assembly lines would look a lot like the ones we sell,” says Spire Corp. Sales and Marketing Vice President Mark Willingham. Spire manufactures equipment for making PV modules as well as turnkey PV-module production lines. It says about 90% of all PV manufacturers have pieces of its equipment in place.
In the making of a crystalline-silicon solar module, the first difficulty manufacturers must overcome in automating their operations is handling the in-coming silicon wafers. Modern silicon wafers for solar cells are only on the order of 100 to 200 microns thick. Separating the thin, fragile silicon wafers from the top of stacks is tricky. (see Next Big Challenge for PV Makers: Wafer Handling, Machine Design, 7/8/08) The typical approach is to employ an air knife to separate wafers from the stack and either vacuum or venturi-type grippers for moving them around.
The thinner the wafer, the more the likelihood of damage during handling. There are efforts to cut costs by making wafers even thinner than 100 microns, but industry veterans doubt the practicality of such devices because of the challenges in handling they would entail. ”I’ve never seen one below 100 microns,” says Spire’s Willingham. “Some big manufacturers have announced 160-micron devices but I doubt they will get even this thin. And the cost of raw silicon is expected to drop. This will alleviate the rationale for thinner wafers.”
“We have seen sample wafers at 90 microns,” says NREL’s Mitchell. “Though it becomes a cost problem if the yields aren’t high enough.”
Also, the thinner the wafer, the more problematic are such operations as soldering and laminating. “Contacting very-thin crystalline cells is an engineering concern. In any soldering process steps the thermal expansion of the silicon can be a problem,” says Mitchell.
Crystalline cells get soldered front-to-back style to make connections analogous to those of batteries in series. The resulting strings then are placed on a glass substrate and aligned for straightness.
Spire says it has appreciably automated its crystalline cell-soldering operations. “Silicon’s thermal coefficient of expansion is essentially zero but the cooling solder’s coefficient of expansion is significant. To compensate we have built in a proprietary mechanism,” says Mark Willingham. “But it is still an area of concern. Inside the stringer/tabber, there is a real opportunity for yield loss because you are picking up every cell. After that, the cells forever more sit on glass so downstream operations are less threatening for breakage.”
The next operation in the sequence, called busing, makes an electrical connection between the strings. In plants up to about a 50-MW capacity, this is a manual operation, says Willingham. And below about 25 MW, substrates tend to be manually pushed around on carts rather than move on automated conveyors, he says.
Indeed, 50-MW capacity seems to be the point at which crystalline PV plants contemplate full automation. Willingham says Spire is installing a 50-MW line for a customer in Europe that intends to eventually expand it to 100 MW. The plant is fully automated with software handling the queuing of modules at each assembly step. Only 10 workers run the facility, and they mostly feed material to robots.
Trends toward larger-capacity PV plants will probably foster more work in automation and improvements in assembly processes. “You will probably see innovative soldering methods that handle more than just one or two cells at a time,” says Willingham. “Installations will also be less about large-scale automotive-style robots and more about simpler mechanisms that are doing simple tasks. Right now the larger Scara-style devices are widely used because they are available in volume and are reasonably priced. As the cost of silicon comes down, we expect to see more distributed manufacturing located in a demand center. This will keep the factory size at or below 100 MW.”
Laminators are likely to see advances in the form of presses able to handle substrates having larger areas and to work with multiple levels of material.
Lamination is followed by framing and trimming steps. Here excess lamination material gets removed. Then a double-sided tape goes onto the substrate area that will lie underneath the aluminum frame bordering the solar module. The taping operation tends to be automated only in plants over 50MW capacity, says Willingham.
However, you won’t find any of these operations in a thin-film PV plant. The assembly steps involved in thin-film differ dramatically from those in crystalline PV. For one thing, the making of crystalline cells and their assembly into modules are completely separate operations. That means they can reside far away from each other.
In contrast, thin-film cell-deposition and module packaging operations tend to be intimately related. So both deposition and packaging normally take place on the same line. One result: “The PV factory will operate 24/7 in a lights-out mode,” says Rockwell’s Bates Marshall. “Thin-film PV makers are interested in dictating a standard way of handling controls so when they extend their operations, they will not have a one-of-a-kind automation strategy for each plant.”
Nevertheless, thin-film lines have their own knotty issues when it comes to improving process efficiency. “Uniformity is one of the parameters in thin films that has been an engineering concern,” says NREL’s Richard Mitchell. “If you don’t have uniformity, pin holes can form. There is also a potential for these devices to operate differently in the middle of the deposition than they do near the edges. But if you are manufacturing these devices you have solved these problems.”
Thin-film PV also entails multiple laser-scribing steps to make electrical connections with buried layers in the cell structure. “We are seeing the PV makers doing a lot of research into laser scribing to get a high throughput,” says Rockwell’s Marshall.
It turns out that information about PV cells themselves is another problem area. “Modules are going out under warranties of up to 25 years. If one comes back, it’s important to quickly figure out not only the problem but also what the scope of it could be and how many other modules could be affected,” says Marshall. The need for such information is one reason Rockwell developed a manufacturing execution system package optimized specifically for the thin-film solar industry, he says.
And plants outside the U.S. have the same information needs. “In China’s PV industry, it seems they are anticipating automation needs and are more forward-thinking than in the semiconductor fabs there,” says Marshall.
It is also likely semiconductor equipment suppliers will recycle knowledge about making flat-panel displays in their PV operations. “One of the biggest proponents of big substrates is Applied Materials,” says Marshall. “That’s because large substrates play into what they have learned about making flat-panel displays.”
It isn’t just thin-film lines that can work with large-area modules. “There are some difficulties associated with working with long strings and laying them up in large panels, but we have demonstrated that this is not a problem for crystalline cells,” says Spire’s Willingham. “There is no issue with the automation of large PV modules. The only reason you haven’t seen them is the expense of the silicon, which is coming down. The larger the module the lower the cost for items such as cables and clamps. Those items can add up to half the system cost.”