Abstract
Because consumer electronic products, such as notebook computers or digital cameras, are becoming increasingly popular, product design is moving toward the trend of lightweight and compact size. Therefore, how to produce nanosized patterns on silicon wafer becomes the key technology to reducing electronic modules. In terms of the scanning probe lithography proposed in this paper, wire thickness was defined directly on the surface of silicon wafer using anodic oxidation of atomic force microscope (AFM), improving on the optical diffraction of traditional optical lithography. The control factor and level value were planned using the Taguchi method, and the quality characteristics were specified. The Taguchi method designs one single quality, in terms of process, the overall quality characteristic seldom being optimized. Therefore, this paper combined grey relational generating with the fuzzy inference system to integrate multiple quality characteristics. Finally, the back-propagation neural network was combined with the quasi-Newton method to build the process prediction system to simulate the experimental results. After multi-quality optimization design, silicon oxide wire of width 68 nm was produced. Results showed marked uniformity on the silicon wafer and could become a useful technology in reducing electronic modules.