Cause And Prediction Of The Defects In The Process Of Injection Molding For Micro-fluidic Chip

With the maturity of micro molding technology, micro plastic parts and the parts with micro structure are widely used in many fields. As a new branch of injection molding technology, micro injection molding is on the stage of rapid development. Scholars at home and abroad have already studied a lot on it and had some achievements. However, the molding mechanism and defect theories still need to be further investigated in order to build process specification for micro injection molding.The micro-fluidic chip which is the typical plastic with micro-structure in the fields of micro-injection molding, was chosen as the subject. The injection molding process of micro-fluidic chip contains the coupling of macroscopical and micro injection molding. As a result, there are both conventional defects and micro defects in micro-fluidic chip. The possible conventional defects were firstly studied and the mechanism of defects were analysed, and then correction suggestions were advanced from the aspects of process design. Then, the differences between traditional and micro injection molding were discussed to explain the origins of the defects and to introduce the standard of quality-estimation for micro-fluidic chip, which was the opening width of the channel.A mold was designed and manufactured according to the structural characters of micro-fluidic chip. After the experiments, the chips was cut at given position in order to observe and photograph by microscope. The contour of channel was picked up through software Matlab. After that, the width of channel was measured by using image processing tools of software Matlab.The relationship between processing parameters and chip quality was analyzed by both simulation and BP Neural Network. Through software Moldflow, the injection process was simulated with the given parameters, and the pressure at the cross-section of the channels was got in the simulation results. By analyzing, the conclusion was that the bigger the pressure was, the better the quality of the channel was. But this method was only the predict one. The BP Neural Network method was on the basis of Matlab which builded the BP Neural Network and trained it with taking process parameters (mould temperature, melts temperature, injection pressure, injection rate and holding pressure) as the input and the width of the channel as the output until the fitting precision reached the requirement. The prediction of this model was quite close to the values of experiments. The error was very small which means the method of building BP Neural Network was reasonable and feasible. Injection molding process specifications of micro-fluidic chip for PMMA material was established on the base of experiment, and it's used to guide the actual production.