Factory level issues and needs from NTRS
02/01/1998
Factory level issues and needs from NTRS
Court Skinner, National Semiconductor Corp., Santa Clara, California
Gary Gettel, SEMATECH/Texas Instruments Inc., Austin, Texas
The Factory Integration Technical Working Group (FI TWG) of the NTRS has identified fundamental factory level challenge areas.
Factory cost. If the escalating cost of semiconductor factories is not slowed down, a fabrication facility will cost $10 billion in 2005. This will be a significant investment for all semiconductor manufacturers and an insurmountable barrier for many, resulting in widespread consolidation of semiconductor manufacturing in the industry. Equipment as a percentage of total factory cost continues to increase and will reach about 80% by 2000. Consequently, continued emphasis on equipment capital productivity goal setting and tracking is imperative. Also, an acceleration in the improvement in both bottleneck and average tool OEE is vital to maximize output from expensive equipment.
Three of the primary drivers of the factory cost escalation are smaller feature sizes, larger wafer diameters, and higher wafer volumes (wafer starts/month). Equipment capital productivity improvement is the key to address the factory cost impact from smaller feature sizes and larger wafer diameters. These vital elements of continuing down the manufacturing cost learning curve drive the industry`s productivity.
Improved tool granularity is the key to address the third area of cost impact, namely, higher wafer volumes. The trend to larger factories is driven by economy of scale. Lower cost/wafer can be achieved in larger factories, due in large part to the impact of mismatched tool granularity in smaller factories. Since equipment is generally designed for maximum throughput, and attention is rarely paid to matching this throughput throughout the factory, the minimum economic scale continues to grow. It has gone from about 8000 wafers/month 15 years ago to 20,000-30,000/month for current 200-mm production fabs. Matched tool granularity across the whole process and factory at a lower minimum economic scale (i.e. smaller factory increment) is needed. Finally, increased sensor integration and process control embedded in the process tools are necessary to reduce the cost for separate metrology equipment.
Investment risk. With escalating cost comes escalating risk for a factory investment. To control this increased risk, new factories will have to be built more quickly and start up faster. However, as factories have gotten larger and more complex, the historic trend is for longer set-up times to break even on the factory investment. Successful investments require modular construction and start-up approaches, fuller tool accommodation standards, improved process and tool qualification, and better equipment performance tracking and improvement methodologies. Faster yield ramp will be critical and will require shorter cycle times in the early operation of the factory to maximize cycles of learning, improved data management and analysis tools for full traceability and in situ metrology, and process control to maintain tighter tolerances.
Factory effectiveness. Bottleneck and average tool OEE levels need to increase significantly to keep the industry on the cost/function manufacturing learning curve. Current average tool OEE levels of 43% need to reach the 60% level to support 180-nm/300-mm manufacturing. Improved OEE data collection and analysis tools, more reliable equipment embedded software, improved scheduling and dispatching, and enhanced process control capabilities are critical to achieve the required OEE improvement. In addition to the higher required OEE levels, the OEE ramp rate for a new factory (or a new piece of equipment in an existing factory) will need to quadruple during the timeframe in the NTRS.
Process complexity. Increasing process complexity drives information and control systems complexity in the factories. Data collected/process step is increasing exponentially, while the number of steps that a wafer goes through in a factory is multiplying linearly. Availability of trained operators and decision makers to staff the factories is becoming more and more limited, and the cost of misprocessing a wafer is inflating exponentially. Decisions in a factory are getting more interdependent with wider range effects. These factors lead to a need for more decision support and automation. Higher yield of data to useable knowledge requires standard interfaces/data models and a standard architecture to tie the data sources to the decision makers.
Wafer carrier size and weight are increasing with wafer diameter. The wafer value is more than doubling with each wafer diameter change. Larger, more complex factories have more sources of operational variability, including work-in-process delivery, that need to be controlled. For maximum efficiency, improved cubic space utilization of the factory is required. These factors are driving increased levels of tool-to-tool automation with carriers designed for automation and environmental isolation, better wafer/die traceability, and improved material control systems.
COURT SKINNER received his PhD degree in metallurgy from MIT. He is director of system technology integration at National Semiconductor Corp. and is co-chair of the SIA`s Factory Integration Technology Working Group. National Semiconductor Corp., 2900 Semiconductor Drive, MS C2-450, Santa Clara, CA 95052-8090; ph 408/721-7420, email [email protected].
GARY GETTEL received his BS degree in systems and industrial engineering from the University of Arizona, and MS degrees in electrical engineering and engineering management from Southern Methodist University. He is director of factory integration at SEMATECH, and is co-chair of the SIA`s Factory Integration Technology Working Group. SEMATECH/Texas Instruments Inc., 2706 Montopolis Drive, Austin, TX 78741-6499.