Your daily dose of Photovoltaic Technology Developments and Solar News

The semiconductor industry has long had its eye on improving yields in the fabs, developing an ever-more sophisticated array of software and hardware tools to detect defects and faults, monitor and analyze process and design variations, zero in on the root causes, and crunch the giga-reams of resultant data to try and make sense of the perturbations of the production flows. As the solar photovoltaics manufacturing industry ramps up and seeks ways to improve its own best practices on the factory floor while reducing costs, the need for a comprehensive, PV-specific yield enhancement strategy has become more acute.

One company familiar to the design and yield communities in the chipmaking realm, Magma Design Automation, announced earlier this week that it is developing a new solar-specific software system based on its proven YieldManager platform. The San Jose-based firm is working with systems integrator Pegasus Semiconductor-Solar to "refine product specifications and test the new product," which will aid PV cellmakers in their quest to, as the press release noted, "better monitor all metrology, inspection, and performance data throughout the manufacturing process. This would enable fab operators to identify and correct root causes of solar-efficiency and yield degradation caused by subtle fab processing fluctuations or instability."

"Semiconductor manufacturing tools such as rapid fault detection, advanced process control, and integrated yield management--which enabled rapid technology introductions and fast yield ramps--are now critically needed in the solar industry to reduce costs," explained Pegasus president Sudhindra Tatti in a statement. "To keep up with the overwhelming demand, innovation in the solar fabrication process must be accelerated, and no enterprisewide yield enhancement software exists for solar fabs."

Ankush Oberai, VP of Magma's failure analysis business unit, is working as the company's point man for what he calls the "solar-targeting YieldManager product." In an interview conducted via email, he told me that they "started working in this market segment in third-quarter 2007 and on the product planning in first-quarter 2008." Magma is "in the design stages of the product and plan to have version 1.0 of the product commercially available in first-quarter 2009."

He offered some examples of key parameters in solar manufacturing that differ from semiconductor manufacturing. "In semiconductor, yield-impacting parameters that are regularly monitored are mostly extrinsic, i.e., from outside, such as particles, overexposure, underexposure, and misprocessing. In solar cells, the yield-impacting parameters are mostly intrinsic, that is, something built into the solar-cell material which cannot be easily seen. Thus, a different 'eye' is needed to see the solar parameters.

"It would not be [done using] either inspection tools nor lithographic optical proximity correction (OPC) detectors. The solar-cell efficiency is directly influenced by electrophysics of solar materials. A 'physics' eye is required to see any changes in those efficiency-impacting parameters."

"The most important parameter is the lifetime of current-generating carriers," he continued. "As the solar energy generates the hole-electron pairs, they are collected separately as electricity. If the solar material is 'dirty,' with many crystalline boundaries as in thin-film solar cells, the solar-generated hole-electrons get pulled into those crystalline boundaries and do not contribute to the electricity generation. Our goal is to find an 'eye' to detect the very subtle change in carrier lifetime, and then we’ll have a home run in the solar-cell business."

As for what changes to YieldManager might be needed to adapt the software to solar-cell manufacturing, Oberai said that "the data management methodology would be similar to the semiconductor applications. Many systems and modules developed for semiconductor YieldManager would be applied with ease and field-proven assurance. There would be more solar-specific modules and systems to be developed with time.

"The analysis and reporting engine requires modifications to the requirements of the solar-manufacturing process and efficiency-effecting parameters. The data structure will be designed to accommodate the tool interfaces and drivers. We do not expect any surprises but could come across data format issues which Magma is very capable of addressing, given its semiconductor experience."

Since partnerships and cooperation among the various vendors, providers, and customers are a critical element to the success of yield enhancement strategies in the chipmaking community, I wondered how that model might be applied to the PV arena.

"Just like in the case of YieldManager for semiconductors," noted Oberai, "Magma plans to enter into technical and OEM partnerships for solar-targeting YieldManager with the metrology/inspection/test/process control equipment and other manufacturing software companies serving the solar-cell sector. We are already in discussions with a couple of tool companies that would like tool-based analysis. This is very similar to our CadNav OEM business of providing Camelot to companies like FEI, Hamamatsu, Zyvex, etc."

Magma has already lined up its first major partner in Pegasus. The FA VP offered a few more details about how that arrangement works. "Magma (which owns the IP of the product) is developing the solar-targeting YieldManager product, and Pegasus will be providing solar-manufacturing domain expertise and helping with the beta testing of the product."

Since the whole raison d'etre of Magma's foray into solar manufacturing is to help companies improve their yields and increase their cells' conversion efficiencies while reducing costs (and expand its own potential for revenue generation, of course), I asked Oberai to explain in more depth what the product will do and what he believes it can achieve.

"This software will monitor the parameters that directly impact the solar efficiency, which, in turn, would cause the efficiency distribution to be wider than normal. That is, if 400,000 wafers are processed every day, many wafers may show <15% or even <14% in efficiency instead of the target 16%. Those would be considered 'bad' wafers and may not be sold at a premium price and some would be even scrapped. The 'effective' yield of the solar fab is appreciably lowered by those low-efficiency wafers.

"The efficiency distribution will be tightly controlled by monitoring the parameters impacting solar efficiencies and by implementing automatic countermeasures as any sign of efficiency degradation or fluctuation is detected with the unstable parameter values. The major cost of the solar cell is the starting silicon wafer cost, which would be 60-70% of the solar cell price. If a fab had to run 440,000 wafers every day for a year to produce 400,000 'good' wafers to meet the target output of 500 MW, the net profit would be detrimentally impacted. Thus, the cost reduction comes from tightly controlling the solar-efficiency distribution and the wafer costs."

What about the differences between mono- and multicrystalline-type solar cell production and YieldManager's ability to deal with those variations? "The parameters which influence the solar conversion efficiency would be the same for any types of solar cells," he said. "However, there would sensitivity differences in detecting those parameters from one material to another."

"Since the polycrystalline silicon is 'dirtier' in electronic terms than the single-crystalline silicon, the parameters may be detected with more signal noise in the background. However, this would be possible, as long as the quality of polysilicon stays within a certain boundary. Thus, the solar-targeting YieldManager would work more easily with 'cleaner' electronic materials than polycrystalline or amorphous materials in general."

Since crystalline silicon isn't the whole ballgame any more in solar manufacturing, I wondered about the yield software package's thin-film playmaking abilities.

"It would be a sensitivity issue, not a capability-limit issue," Oberai countered. "Dirtier materials with many crystalline boundaries would directly impact the solar conversion and lower the conversion efficiencies. A powerful signal-to-noise-improving filter or algorithm needs to be developed to monitor a subtle change in those efficiency-impacting parameters. This is a challenge no one has attemped to solve yet.

"Yes, we do foresee applying this to the thin-film sector. Developing a mechanism to monitor and control the solar-efficiency-impacting parameters for thin-film solar cells is more challenging, simply because it is very dirty in an electronic sense, i.e., there is a high level of noise in monitored data. Since the thin-film solar conversion efficiency is relatively low, 6-10%, it may not be very fruitful to monitor efficiency-impacting parameters but rather [to look at] the factors directly impacting the solar power output in thin-film solar cells.

"There are many factors that would cause power output degradation of thin-film solar cells in processing aspects, as in the semiconductor manufacturing. That is, the thin film can be affected by various particles and processing instabilities more critically than crystalline silicon solar cells. Inspection data, metrology data and performance data can be managed by solar-targeting YieldManager similarly as they are with semiconductor fabrication, to implement automatic real-time countermeasures as yield-impacting data or instabilities start appearing."

Oberai concluded with some thoughts about the the state of the PV manufacturing industry and the timing of bringing Magma's software package to market. "Solar-cell technology is old, but it's a very new industry, simply because there was very little demand nor investment in it. Now, there is considerable interest, and money is pouring into the solar industry, and products like solar-targeting YieldManager will start to appear. We don't know of any other commercially developed product that has comparable capabilities to Magma's and can be applied across the enterprise."

Now that Magma has kicked open the door to this un(der)exploited market segment, expect other companies with deep semiconductor yield management and EDA knowledge to bring their own similar products and services to the solar fab floor.