Neural network characterization of scanning electron microscopy

A scanning electron microscope (SEM) is a sophisticated equipment employed for fine imaging of processed film surfaces. In this study, a prediction model of scanning electron microscopy was constructed by using a generalized regression neural network (GRNN). The SEM components examined include condenser lens 1 and 2 and Objective lens (coarse and fine) referred to as CL1, CL2, OL-Coarse, and OL-Fine. For a systematic modeling of SEM characteristic, a Box-Wilson experiment was conducted. The prediction performance of GRNN was optimized by using a genetic algorithm (GA). The prediction error of GA-GRNN model is 1.96 ×10^-12 at a spread range of 0.2. From an optimized model, 3D plots were generated to interpret parameter effects on SEM resolution. For the variation in CL2 and OL-Coarse, the highest resolution (R) could be achieved in all conditions except for the two large sections, larger CL2 at smaller or larger OL-Coarse. For the variations in CL1 and CL2, the highest R was obtained in all conditions but those at larger CL2 and smaller CL1.