Application of a Neural Manufacturing Concept to Process Modeling, Monitoring and Control
- PDF / 1,320,934 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 107 Downloads / 149 Views
CHI YUNG FU, LOREN PETRICH, and BENJAMIN LAW Lawrence Livermore National Laboratory University of California P. 0. Box 808, L-271 Livermore, CA 94550 Email: [email protected]
ABSTRACT The cost of a fabrication line, such as one in a semiconductor house, has increased dramatically over the years, and it is possibly already past the point that some new start-up company can have sufficient capital to build a new fabrication line. Such capital-intensive manufacturing needs better utilization of resources and management of equipment to maximize its productivity. In order to maximize the return from such a capital-intensive manufacturing line, we need to work on the following: 1) increasing the yield, 2) enhancing the flexibility of the fabrication line, 3) improving quality, and finally 4) minimizing the down time of the processing equipment. Because of the significant advances now made in the fields of artificial neural networks, fuzzy logic, machine learning and genetic algorithms, we advocate the use of these new tools in manufacturing. We term the applications to manufacturing of these and other such tools that mimic human intelligence neural manufacturing. This paper describes the effort at the Lawrence Livermore National Laboratory (LLNL) [1] to use artificial neural networks to address certain semiconductor process modeling, monitoring and control questions. *
INTRODUCTION There are a number of problems associated with traditional manufacturing concepts and viewpoints towards equipment. They are as follows: 1) Process and equipment modeling are still confined to the researchers, and are generally not available for the front-line process engineers to put to work. To improve yield, and to add flexibility in manufacturing, these tools are needed at the front line. 2) Although existing process control is a closed-loop system, it is not a direct closedloop control of the fabrication process itself. Instead, it is a collection of individual closed-loop controls of peripheral parameters such as pressure and gas flow. This type of control is indirect, since it is not the individual parameters that control the outcomes of a process, but the synergistic effect of all the parameters that governs the results. 3) If a process deviates from specifications during processing, the process engineer has very few options to "rescue" a bad run. 4) At present, some processing errors are corrected by "reworking" or "processing feed-forward." Scheduling such reprocessing runs among "fresh" runs is a serious logistical problem. 5) For semiconductor processing, the cluster tool concept, increasingly becoming important, must be extended to include metrology. This complicates the cluster-tools' design and integration. 6) Diagnosing equipment problems only after the equipment is down is highly inefficient. It is much more preferable to try to predict which component will fail, and thus schedule timely repair with minimal impact on Work supported under the auspices of the US Department of Energy by the Lawrence Livermore National Laboratory unde
Data Loading...
