The photovoltaics/solar power manufacturing industry has come a long way in the last 12 years, increasing productivity by a factor of ten, thanks to increases in throughput and wafer sizes. Speaking at Intersolar in Munich, Germany, Dr. Axel Metz, director of solar cell R&D for SCHOTT Solar AG (Mainz, Germany) made a case for inline wafer processing, saying it is best suited to new throughput and yield challenges. He said that, by 2020, wafer throughput is expected to increase to 7200 wafers per hour, and that yield loss below 1% is expected, even on much thinner (100 micron thick) wafers.
Metz noted that cost reduction in the main technology driver in the PV industry, and that increases in efficiency have a relatively minor impact, reducing module manufacturing costs by only about 4% (Fig. 1). The learning rate shown is based on historical progress. Metz said the annual growth rate of 35% has always been exceeded in the last ten years. “This year and next we might not quite reach this number but no one knows,” he said.
Cell efficiency gain is assumed to incrase 15% to 20%. “If you analyze the cost per piece, you see a dramatic reduction of 50%,” Metz said (indicated by the red line). The cost benefit from the increase of efficiency over time is relatively small. “This is because the efficiency gain is only on the order of 4% absolute, or 20% relative,” Metz said. “Cost reduction is clearly dominated by the manufacturing reduction of cost per piece. Of course, one has to mention that, on the system level, efficiency plays an important role as well. A high efficiency module has the potential to further reduce system cost. But at the module the efficiency contribution is relatively small compared to other cost impacts,” he said.
Fig. 1: At the module level, cost reduction is dominated by manufacturing cost per piece. Efficiency gains play a minor role. Source: SCHOTT Solar AG.
Metz made a case for a manufacturing strategy based on inline processing, citing key productivity parameters of equipment throughput; tool uptime (defined by SEMI Standard E10); mechanical, optical and electrical yield; and overall equipment efficiency (OEE); He said that the International Technology Roadmap for PV (ITRPV) — the second version of which was released in March 2011 — calls for a throughput of 7200 wafers per hour by 2020, for both front- and back-end processes. In the 2011-2012 timeframe, throughput for front-end (chemical, thermal and SiN deposition) processes is expected to be 3600 wafers/hr, increasing to 5000 in 2013, 6400 in 2015 and then 7200 in 2020. There is a mismatch with the throughput of back-end processes, which are expected to be 3000 wafers/hr in 2011-2012, then increasing to 3600 in 2013 and 5400 in 2015. By 2020, the mismatch should be eliminated, with back-end processes also capable of 7200 wafers/hr, due to expected advances in metallization technology. “We expect throughput numbers to double by 2020. We also expect to change metallization technology around 2015 so that the numbers will match,” Metz said. “This is an important message to equipment manufacturers who need to prepare for these high throughputs,” he said.
Wafer thickness is also expected to drop dramatically over this same timeframe, going from 180 micron now to about 100 micron. “Still, we expect the yield, despite the thinner wafers, to increase. We should aim for yield loss numbers for below 1%. That’s for the whole production line, from wafer-in to tested cell out,” Metz said.
Metz said inline processing has many advantages over other types of strategies. Equipment throughput is higher, handling steps are minimized which helps yield, it’s suitable for thin wafers and it offers higher process stability. “If you have an inline process up and running, it can run very smoothly without much operator interference, unless there’s an unscheduled event,” Metz said. “Since you have less equipment, you have less equipment to control and this adds to the process stability.” The main disadvantage: less process flexibility. “At the beginning, when you build the equipment, you have to think of the process and define the process parameters,” Metz said, but noted that tost modern inline equipment – especially wet chemistry – is built on a modular basis. “You can add modules later on if you want. That helps lessen the fear of the constraints of the process flexibility for inline processing,” he said.
Figure 2: Throughput of state-of-the-art inline process tools, power production capacity of a MWp/year basis and number of tools required for a GWp/year production line. Source: SCHOTT Solar AG.
The throughput of existing inline tools is shown in Fig. 2. Noting the higher number of diffusion furnaces compared to other tools on a GWp/year basis, Metz said “there are still some inline diffusion furnaces out there and their throughput is a little more than 2000 wafers/hr.” He said the number of tools needed is an advantage for big fabs. For small fabs (i.e., 200 MW) if a piece of texturing equipment or chemical equipment goes down, then the entire production line goes down. “For the future, as we all think, photovoltaics will grow, bigger fabs will come up, and then this a clear advantage of inline processing,” Metz said.