This paper describes the development of a run-to-run control algorithm using a feedforward neural network, trained using the backpropagation training method. The algorithm is used to predict the critical dimension of the next lot using previous lot information. It is compared to a common prediction algorithm - the exponentially weighted moving average (EWMA) and is shown to give superior prediction performance in simulations. The manufacturing
implementation of the final neural network showed significantly improved process capability when compared to the case where no run-to-run control was utilised.
The paper examines the viability of various levelling options1 on Nikon i11, i12 and i14 steppers in compensating for across field differences in focus position. The analysis was performed on both production wafers at various processing stages and test wafers with oxide deposited then etched to different depths. The main analysis technique used was the stepper focus measurement system along with Hitachi 9220 CDSEM measurements and levelling beam analysis using a CCD camera2. The conclusion from the paper is that due to diffraction effects of the levelling beam, levelling-on can introduce large wafer stage tilts and so reduce CD control in the i11's and i12's. Since the EGL method also uses the levelling sensor in conjunction with the focus sensors this also introduces large tilts causing large across field CD variation.