Abstract-Traditionally, the semiconductor fabrication conducts several measurements within a wafer and condenses the measurements into output quality characteristics to progress advanced process control (APC). Therefore, traditional APC employs performance indices along "time" axis and losses the space information among different measurements within a wafer. In this paper, we present a new perspective on process control with pheromones propagation mechanism, in which we treats the disturbances within a two-dimensional layout wafer as digital pheromones,. Our novel space-effect algorithm is called the two-dimensional pheromone propagation controller (2D-PPC). The simulation results show 2D-PPC, which involves space-effect, can improve the uniformity of the wafer over the conventional time-effect controllers.Index Terms-Two-dimensional pheromone propagation controller, process control, two-dimensional pheromone basket, two-dimensional digital pheromone infrastructure, swarm intelligence
I. INTRODUCTIONIn semiconductor manufacturing, run-to-run control adjusts the recipe slightly based on in-line measurements to even out disturbances. Disregarding the methodology of the run-to-run control, some researches condense measurements to output quality characteristics [1]- [10]. Thus, traditional run-to-run controllers like EWMA employ the "time-effect", which means disturbance of a wafer affects a measurement or the output quality characteristics at different runs, among observed data to calculate the recipe for the next run and does not consider the "space-effect", which means disturbance of a wafer affects several measurements at the same time, among measurements within a wafer at a run.In this paper, we describe an algorithm that includes the effect among different measurements within a wafer. The concept comes from the observation that a disturbance of a wafer will affect a piece area, which may contain several measurements. The new algorithm uses swarm intelligence by assuming that the measurements have their own behavior and affect others nearby within a wafer at a run. Then, the intercepts with disturbances included of a linear regression model at different measurements within a wafer are modeled as digital pheromones. The interaction among the digital pheromones is modeled by a propagation mechanism. Because measurements within a wafer are a two-dimensional Manuscript received August 10, 2012; revised December 11, 2012. The authors would like to thank the National Science Council of the Republic of China for financially supporting this paper under Contract NSC 100-2221-E-009-063-MY2. D. S. Lee was with the Dept. of Mechanical Engineering, National Chiao Tung University, Taiwan (e-mail: dslee605@ ms37.hinet.net).A. C. Lee is with the Dept. of Mechanical Engineering, National Chiao Tung University, Taiwan (e-mail: aclee@ mail.nctu.edu.tw). layout, our novel algorithm is called the two-dimensional pheromone propagation controller (2D-PPC). This study modifies the propagation-out ratio of digital pheromone infrastruc...