Application of neural network on LTCC fine line screen printing process

Design of experiment (DOE) is a useful tool for optimization of manufacturing process control. Two steps of experiment are designed to minimize experimental variables and find out the optimal control factor and its range. Combine neural network technique and first step experiment data of DOE, one can obtain the optimal conditions for a manufacturing process. This is study investigating the effect of screen-printing parameters on the width of conductor lines based on the implication of neural network design of experiments. The process parameters including paste viscosity, print squeegee speed, squeegee pressure, snap-off distance, emulsion thickness, screen mesh, and screen open area are examined to find out the optimal conditions. Line width finer than 50 μm can be successfully achieved through this kind of combination from DOE and neural network simulation.