A vision for a next-generation 300mm factory
10/01/2008
EXECUTIVE OVERVIEW
To help meet next-generation factory (NGF) challenges, ISMI has been tasked by its member companies to explore ways in which improvement in cycle time and cost reductions can achieve targets of 50% and 30%, respectively, by 2012. This paper describes ISMI’s NGF Program activities that are focused to achieve these goals.
To remain profitable, the member companies of International SEMATECH Manufacturing Initiative (ISMI) must simultaneously increase product performance while reducing the cost of manufacturing operations. Alliances to share the increasing financial burden have been effective for addressing technology development, but have been less effective in reducing manufacturing cost escalation.
The largest single cost for a semiconductor factory is the depreciation of capital equipment. Equipment utilization remains almost unchanged over the last several years and ranges from ~40-70%. For this reason, ISMI’s efforts will initially focus on improving equipment utilization, availability, and predictability. Member collaboration throughout the industry supply chain will be needed to create a cost-effective NGF implementation.
ISMI’s efforts will be guided by the Detractors List and 19-Point Guidelines developed to identify and propose solutions for the major sources of inefficiency and waste in the factory. The current NGF projects have been prioritized in terms of both immediate needs and long-term returns.
NGF program overview
To enable the vision for an NGF, ISMI created a program that strategically foresees opportunities in equipment productivity and within the factory environment. It is expected that factory systems such as automated material handling systems (AMHS), decision-making capabilities, and manufacturing rules for dispatching and cascading will become the limiting factor for further productivity improvements. This program is not only intended to improve the productivity of future 300mm factories; it will build the bridge to 450mm factories, and provide retrofitable solutions, independent of wafer size.
The projects developed by the NGF program comprise five main categories. The Next Generation Realization Project serves as steering and coordinating body and is defining the strategy of the program and collaboration with all other industry resources. The project entitled Factory for Small Lot Size (SLS) defines the elements necessary to deploy smaller lot sizes in a next-generation factory. This project also uses factory simulation modeling to estimate the effect of small lot sizes on cycle time and cost to determine the most effective contributors. The Equipment Data Acquisition (EDA) and Equipment Data Quality (EDQ) projects focus on improving data collection from production tools, data quality, and lead the effort for more rigorous standardization. All projects depend on equipment data and these two projects are building the basis for success of the NGF program.
Equipment Chamber Matching (ECM), Predictive and Preventative Maintenance (PPM), and Enhanced Equipment Quality Assurance (EEQA) projects strive to improve not only equipment availability but also equipment predictability. And lastly, The Virtual Metrology (VM) project develops methods and algorithms to predict metrology measurement results based on current known data and equipment behavior. This will reduce the need for measurements and lead to less disruption of the manufacturing flow and decrease cycle time, as well as the costs for equipment purchases.
Project definition
The SLS project, considers all factors for equipment, factory, transport, and systems challenges as well as operating business model impacts and variations in factories utilizing small lot sizes. The project must also determine where equipment, AMHS, and other systems cannot operate effectively, and when radical new concepts and architectures must be introduced.
New generation data collection capabilities are the foundation for all other ISMI NGF projects and build the cornerstone for any next-generation factory. A more agile factory with lower cycle time, smaller lot size and single wafer manufacturing requires more accurate and timely data to make conscious and intelligent decisions about wafer and equipment management. The PPM project may serve as an example to underline this need.
To prevent disruptions in the manufacturing flow, it is essential to decide on the right timing for maintenance procedures or to actively manage the product flow of this particular equipment. ECM, PPM, and EEQA are three projects working on different aspects and levels of equipment productivity improvement. A next-generation factory requires not only increased equipment availability, but also improved predictability and control to meet productivity goals.
The VM project is developing methods and procedures for predicting metrology measurement results based on prior data and data collected from the current manufacturing step. Within a certain confidence interval, VM methods will reduce measurement frequencies. Removing a portion of the metrology steps will reduce cycle time and capital expenses
Single wafer manufacturing, while not currently included within the NGF program, is nevertheless considered to be important to achieve the stated productivity targets. Small lot sizes and single wafer manufacturing, in conjunction with lower cycle times and improved equipment productivity, will require better cascading scenarios and higher accuracy and speed of product delivery to the equipment. This will result in strains on AMHS and manufacturing execution systems (MES) which must also be addressed.
Waste reduction
To date, the Toyota Production Model (also called lean manufacturing) is one of the most efficient replicated manufacturing processes. Not only do integrated device manufacturers (IDMs) see value in bringing this method into semiconductor manufacturing [1, 2], equipment suppliers have identified a huge productivity improvement potential by applying Toyota’s waste reduction approach. As different as the actual manufacturing process may seem, most of the elements can be easily detected in any semiconductor factory:
- Overproduction (ahead of demand)
- Transportation, not actually necessary to perform processing
- Waiting (by far the biggest part from a wafer perspective in any factory)
- Inventory (WIP and finished products not being processed)
- Motion (people or equipment moving more than is required to perform the minimum required processing)
- Over production (to deliver wafers to less predictable equipment so they can be processed the moment the equipment becomes available) Defects (effort to inspect for defects versus prevention)
If implementing these principles into a semiconductor factory seems unlikely, some values from a wafer point-of-view suggest a much improved and more efficient production process. Implementation of a single-piece manufacturing flow (or extreme small lot manufacturing) and single-wafer processing addresses many of these principles. For example: the pull vs. push principle in dispatching material to equipment will ensure that no material is transported to any equipment before it is required for processing. Simply implementing these three elements will lead to dramatically increased data traffic and messaging, requiring more accurate delivery schedules from the AMHS and increase the pressure for timely and accurate decision making.
Factory simulation modeling
High-mix factory models contain a factory capacity of 34,500 wafer starts per month including 4,500 non-product wafers. Two percent of these wafer starts are higher priority. Simulations included two lot sizes???25-wafer and 12-wafer???and both of these models contain the same equipment. Hence, no additional capital was required for the 12-wafer factory model. High-mix simulation models contain 100 products distributed over five process flows. Baseline models assume batch equipment for furnace and wet bench process tools.
 |
ISMI’s approach to achieve fair comparisons for equipment loading was measured by calculating equipment utilization/availability (U/A)%. To make fair comparisons of each model at equal factory capacity, a U/A benchmark was established as the theoretical U/A target based on projected relative equipment capital costs. The utilization target of bottleneck equipment was set at 90%. The goal in each modeling activity was to optimize represented equipment U/A without exceeding the theoretical U/A target for each equipment family.
Single factor experiments testing the sensitivity of factory and equipment improvements include a 5% improvement in equipment availability and a 25% improvement in first wafer delay and setups. More extreme improvements simulated include the complete conversion of batch tools to single wafer processing (SWP) equipment. Equal wafer throughput was assumed and simulated for batch and SWP tools. Factory improvements include observing lot movement throughout the factory, including queue sizes at each tool family to reduce or remove the occurrence of idle equipment and keeping equipment utilized.
Simulation results for a factory modeled with 25-wafer lot sizes and a separate model representing a factory with 12-wafer lots, can show sizable improvements in cycle time (see Table). The most notable impact upon cycle-time improvement was achieved by simulating the effect of SWP in the 12-wafer model; a 29% reduction in cycle time compared to the 25-wafer model, which resulted in an 8% improvement in cycle time.
Although these cycle-time reductions are theoretical, the combined effects of the factors that have been simulated and the improvements expected from the NGF projects will likely achieve an overall cycle-time reduction of greater than 50%.
Cost modeling
The goal of the economic analysis work was to provide a quantitative cost benefit assessment of 300mm NGF productivity initiatives through in-depth research and comprehensive modeling. In addition, the team was to develop a methodology to assess the economics of both individual and overall solutions for eliminating productivity detractors.
The approach was to establish surrogate baseline 300mm wafer processing costs by technology node and product groups developed from external sources. A surrogate 300mm classic foundry factory was selected to represent the leading-edge industry production for CMOS Logic (Fig. 1). Starting material value assumptions were developed using a value curve by time derived from historical cost/area. Technology nodes were selected using the International Technology Roadmap of Semiconductors (ITRS) for guidance.
 Figure 1. Cost development projections. |
The cost modeling assumptions were then correlated with the factory modeling base assumption where applicable, and other nominal industry metrics were defined where appropriate. Future processed wafer costs by technology node for both base (ITRS) and accelerated technology roadmaps were then developed by extrapolating historical trends.
 Figure 2 Cost distribution in a 300mm classic logic factory. |
These baseline costs were then modified to reflect the enabling and implementation cost defined by the factory modeling scenarios for eliminating productivity detractors. The initial step in this process was to define a methodology for establishing the enabling cost with each of the productivity initiatives. The scenarios were first linked to key needs and possible approaches to solutions, and then evaluated considering all aspects of the supply chain based on development effort, enabling time, and implementation probability (Fig. 2). Relational enabling costs were then assigned to each initiative, based on its initial benefit as determined by simulation. Using input on increases to capacity from the factory model, cost variances for processed wafers were then calculated from each initiative. In addition, cost benefits were assigned to all cycle-time improvements to determine the overall cost benefit results.
Conclusion
To achieve the necessary 50% cycle time and 30% cost reduction, ISMI member companies have created the NGF strategic program. The current focus is on improving equipment productivity. The NGF focus will later expand into the factory side of productivity improvement. AMHS and MES systems and their suppliers, as well as factory operations, will be the focus of improvements in future NGF projects.
Interface A maturity, its availability on all production equipment, and a standardized Interface A test method are all critical to the success of the NGF Program.
The waste reduction concepts and approaches, successfully demonstrated by Toyota, and supported by many ISMI member companies, are getting more traction in the semiconductor industry and are considered a valuable approach for finding and implementing solutions. ??
References
- P. Singer, “Redefining Factory Productivity from a Waste Perspective,”Solid State Technology, May 2008, http://www.solid-state.com/display_article/329290/5/wnart/none/none/1/Redefining-fab-productivity-from-a-waste-perspective/&dcmp=confab2008.
- E. Englhardt, G. Rao, S. Kobayashi, et al., “Waste Reduction ??? A New ITRS Initiative,”May 2008, unpublished ITRS ??? Factory Integration Team draft, p. 1-5.
- M. Splinter, “The Future Depends on an Agile Fab,”SEMI Quarterly Report to Members, Winter 2007, p. 1-2.
Olaf Rothe received his diploma in electrical engineering from the Leningrad Electrotechnical Institute in 1986 and is the program manager for the Next Generation Factory Program at International SEMATECH Manufacturing Initiative, 2706 Montopolis Drive, Austin, TX 78741 USA; ph.: +1 512.356.3717; email [email protected].
Brad Van Eck received his undergraduate degree in chemistry from Calvin College and his PhD in inorganic chemistry from Michigan State U. in 1983 and is a project manager NGFR at International SEMATECH Manufacturing Initiative.
Denis Fandel received a BS in mathematics from the U. of Wisconsin and is a project manager at International SEMATECH Manufacturing Initiative.
Robert Wright received his BBA in management and his MS in industrial technology from Texas State U. and is a project manager at International SEMATECH Manufacturing Initiative.