ABSTRACT
The goal of a manufacturing system is to produce multiple copies of the same product, each having attributes within specified tolerances. Variation reduction is one of the major techniques for achieving process stability and requires increasing amounts of process and equipment control at various levels of manufacturing systems. In particular, controlling complicated processes to produce smaller feature sizes is inherently difficult in semiconductor manufacturing. Moyne et al. (1993) attribute this difficulty to an insufficient number of sensors and actuators at each manufacturing process step for establishing a desired level of concurrent control over process parameters. Moreover, mathematical models incorporated into the control scheme rely on empirical data and are consequently imprecise. As pointed out by many researchers (Box and Kramer, 1992; Vander Wiel et al., 1992), it is highly desirable to investigate different control methods to detail the scope of their usage and limitations, and to address the complementary utilization of those methods in a control system.
