An adaptive model for the control of critical dimension in photolithography process

In this paper, a method of adaptive modeling is studied for the photolithography process used in the semiconductor industry. Real data of the critical dimension (CD) in a shallow trench isolation area from an Intel flash manufacturing fab is used for model analysis. We focus on the problem of CD prediction in the presence of unpredictable drift. Multiple inputs are included in the model as interactive integral parts of the system. In addition, a feedback controller based on the adaptive model is designed to stabilize CID in the presence of unknown disturbances. Simulations are carried out based on the data from an Intel 8-inch fab of flash products. The results are generalized to systems with multiple outputs, in which both CD and focus indicator are controlled based on adaptive modeling. The simulation results prove that the designed controller is able to correct process drift automatically.