Good engineering practices and intelligent process optimization are keys to cost-reduction, product differentiation and brand recognition …
Good engineering practices and intelligent process optimization are keys to cost-reduction, product differentiation and brand recognition, ultimately resulting in product bankability and competitiveness against established c-Si products.
Thin-film photovoltaic (TF PV) manufacturers have experienced major commercialization advances in the past two years, as many companies have ramped up their first mass production lines and are gearing up to expand their manufacturing volume, capture market share and establish the economic viability of multiple thin-film technologies. To compete with crystalline silicon (c-Si), new players must be extremely innovative in their approach to lower manufacturing costs, create brand recognition and assure the bankability of their products. Rapid and continuing improvements are pursued in panel efficiencies and factory productivity, ultimately leading to reduced production costs/Watt. Concurrently, new quality and reliability enhancements are implemented in order to validate thin-film products as reliable and rugged, and worthy candidates for project financing.
Many thin-film manufacturers recognize that a major enabler for many improvement efforts is a strong and well-engineered process control methodology, backed by appropriate factory automation and highly trained engineering and technical staff. Process control is central to many manufacturing domains, from the automobile industry to microelectronics, yet the thin-film industry cannot afford the long learning curve associated with those disciplines, and it must develop its own customized methods while at the same time racing to achieve viable cost-per-Watt figures. Concepts and knowledge derived from fundamental process control theory are giving rise to new methodologies and solutions that are increasingly implemented by thin-film manufacturers in order to reduce production costs, increase average panel efficiency and provide a solid technical foundation for emerging panel durability and long-term performance metrics [1, 2].
The 3-way scaling challenge
The challenges faced by thin-film manufacturers are best considered in the framework of the following 3-way scaling issues:
Scaling of cells from laboratory size to full panel size – The lab-to-fab efficiency gap has proven extremely frustrating, yet represents a major opportunity for efficiency improvement of thin-film panels. Efficiencies that can be reached in the lab cannot be reproduced by merely scaling-up the process equipment, as the adverse effects of process variations, non-uniformities and multiple local process defects accumulate across panels, and result in final performance, which is substantially lower than that achieved on small area cells utilizing the same basic production technology. Much of the effort of thin-film manufacturers is related to bridging this gap.
Scaling of production from R&D and pilot lines to mass volume production – Many of the subtler aspects of efficient manufacturing become apparent when production lines are throttled to their maximum designed output. Production tools and material supplies are operated near their designed maximum rates, maintenance time must be kept minimal and troubleshooting must take place faster in order to minimize scrap material or line down time. In another manifestation of the production scaling challenge, multiple production lines, either across the street or across the globe, must be operated such that products are closely matched, and in a manner which will allow propagation of new improvements and best methods to multiple sites.
Scaling of quality and reliability engineering – Thin-film manufacturers are expected to provide the physical evidence that will support the claim for product durability under harsh environmental conditions, with typical 25-year guarantees. Since long-term statistical data is not as abundant as for crystalline silicon modules, this type of scaling represents an uncharted territory for most thin-film manufacturers. However, the known link between production quality and long-term reliability can be exploited: by tightening production specifications, narrowing process windows, and increasing yields, manufacturers can improve the implied durability of their modules, thereby supporting end users’ risk mitigation when selecting thin-film panels for their projects. Process control thereby becomes a stamp of quality and a factor that determines product bankability.
Unique requirements for thin-film PV
Thin-film PV manufacturing shares many process technologies with two well-established industries: the glass industry for glass coatings and flat-panel-display industry for large area deposition. To some extent, aspects of the production process are also shared with the wafer-based semiconductor industry. While most of these technologies are not new, their adaptation to thin-film PV has introduced a set of requirements and constraints that require a fresh approach to process optimization and production line management. Consider, for example, two aspects of process control that are highly specific for thin-film PV manufacturing:
Extreme requirements on process uniformity across large panel areas: The total efficiency of a thin-film panel can be dramatically affected by relatively small non-functioning areas. For example, an area of <1% of the panel area with a missing Si p-layer can cause a 35% drop in total panel efficiency, effectively scrapping the panel. Such local deposition problems are usually related to process tool faults and process window drifts. Unlike flat-panel-displays, where several displays are cut from the same glass, PV manufacturers most often manufacture large panels, currently up to 5.7m2, and the impact of each scrapped panel is substantial.
Conflicting process integration requirements: The single-step optimization that is initially applied to each of the discrete process tools in the production flow often does not yield an optimally functioning cell. A typical example is the surface morphology of the front TCO layer in tandem Si cells, where optimal light scattering is often in contradiction to optimal structure seeding for the micro-crystalline silicon that will eventually be deposited on the stack. TCO deposition that has been optimized for a single junction process will therefore require much re-engineering for a tandem cell to achieve the expected performance gains. Such inter-dependencies are continuously being identified by thin-film manufacturers and academic research centers, and this knowledge is incorporated into intelligent process control systems.
Short cycle times and rapid fault detection and root cause identification play a critical role in maintaining line productivity. Keeping production costs under control requires that line utilization be kept high for long stretches of time, from weeks to even years. Unscheduled tool downs, long investigations and qualifications cycles reduce line utilization. To avoid such mishaps, the status of current process tools and manufactured panels must be continuously collected, processed against well-developed process capability standards and converted into actionable control outputs. Such outputs can be used to modify process conditions, trigger tool maintenance procedures or initiate further investigations. Feedback loops need to be gradually automated to enable around-the-clock monitoring and reduce the dependency on knowledgeable operator intervention. Remote monitoring, in particular, is useful in the case of multiple production sites.
Wide-area-metrology in process control
When considering the overall annual power output of a production line, as measured by MW per year, the effects of line productivity (good panels out) and average panel efficiency are equally important (Fig. 1). Figure 1 shows two bar graphs; the upper bar illustrates the improvement of productivity, while the lower has the narrowing of the histogram of panel efficiencies. A well-designed process control scheme will include goals for improvement on both fronts.
Wide-area-metrology (WAM) is a major source of information for process control. Measurements are taken in hundreds of locations across a panel, generating multiple spatial maps that capture the distribution and spatial patterns that are associated with various process steps. One such example, shown in Fig. 2, reveals the hidden pattern of thickness of an amorphous silicon layer, as the panel leaves the PECVD deposition chamber (i.e., a typical a-Si thickness map showing chamber signature). Most deposition patterns are not truly uniform, but rather exhibit a typical process and chamber signature, as in this example. Deviations from best process conditions, whether resulting from process recipe changes, disruption to gas flow or other fault sources, often manifest themselves as disturbances to the typical spatial signature (e.g., Fig. 3). Such spatial changes are extremely hard to detect with other local metrology methods, yet easily classified by dedicated spatial analysis software tools. WAM has been shown to be highly valuable in predicting chamber failures, rapidly detecting process excursions and dramatically shortening qualification time after chamber maintenance procedures.
In a second example of the role of full panel mapping (e.g., the slope of TCO coverage shown in Fig. 4), the TCO coating process is monitored inline to detect non-uniform coating. Such non-uniformities cannot be detected by conventional imaging techniques that are commonly employed for TCO inspection, yet their effect on panel performance is dramatic. Similarly, doping levels and light scattering can be monitored with spatial signature maps continuously compared against “golden maps” that represent best process conditions.
To fully utilize the power of WAM, inline monitoring solutions have been developed that enable the measurements to be taken in a non-intrusive manner during production, without requiring special dummy panels to be produced and scrapped, and without adding any delays or additional processing steps. This is achieved by placing inline WAM systems that are capable of generating spatial signature maps at production throughput. Depending on exact line design, such metrology tools can inspect up to 100% of the panels, providing near instant feedback for any process deviation. With highly advanced signal processing capabilities, WAM tools will simultaneously generate between four and eight spatial signature maps, capturing multiple layers within a single scan, and providing critical mapping of layer thickness, surface and light trapping properties, energy gap and more. Customized software reporting and alerting tools have been developed that detect spatial signature variations and panel-to-panel trends, and provide notification, either by modifying process parameters to compensate for the drift, or by alerting operators by means of SMS, email or a number of available software interfaces. By employing such software tools, a manufacturer can tailor-fit a set of process optimization routines that are employed during line ramp-up as well as during volume production to assure optimal productivity, increase cell efficiency and drive process enhancement efforts.
Whether relying on integrated production lines by turn-key providers or implementing a best-of-breed method to integrate multiple process tools, thin-film manufacturers have come to realize that good engineering practices and intelligent process optimization are keys to cost-reduction, product differentiation and brand recognition, ultimately resulting in product bankability and competitiveness against established c-Si products.
Figure 4. A full panel map of TCO layer thickness, capturing a sloped layer, a result of a bad coating process. Such variations are much too subtle to capture using imaging and must be detected using advanced optical technologies.
- C. Beitel, “Driving Down the Cost of Solar,” 2nd Thin-Film Solar Summit US, December 2009.
- R. Lahri, “Amorphous Silicon: Keeping the Price Low and Staying Competitive, 2nd Thin-Film Solar Summit US, December 2009.
Ariel Ben-Porath received his BSc in electrical engineering from the Technion, Haifa, Israel and an MSc in applied mathematics & computational biology from the Weizmann Institute of Science. He is the VP, marketing & products at BrightView Systems, 25 Basel St., Petach-Tikva 49510 Israel; ph.: +972-3-929-1400; email: email@example.com.
Benny Shoham received his BSc in electrical engineering from the Technion, Haifa, Israel, and an executive MBA from Bradford, U., UK He is co-founder and CEO at BrightView Systems.