Micro-extrusion Process Parameter Modeling.

摘要

Direct write processes are a family of technologies with the ability to deposit functional structures directly onto planar and non-planar surfaces. Direct writing includes a variety of processes that use different mechanisms to transfer materials on to substrates and can be generally distinguished from conventional rapid prototyping processes by a feature resolution in the sub-micron to micron range. The dispensing system studied in this thesis is a pneumatically actuated micro-extruder which is capable of processing a wide variety of materials. This material dispensing tool is capable of depositing small amounts of material to build three dimensional structures in an accurate and repeatable manner.;The material dispensing system in this study has a variety of manufacturing applications ranging from printed electronics to biomedical applications. The material dispensing system employs a needle valve mechanism that allows ink or slurry to be deposited onto a substrate using air pressure. The dispensing tool used for this research is an nScrypt SmartPump. This research is focused on analyzing the extrusion process and developing and validating a parametric model for the input parameters using a design of experiments (DOE) approach. The aim is to improve the repeatability and accuracy of the process.;A two phase approach was used to identify significant input parameters impacting the dimensional properties of a printed track. The first set of experiments employed a 2-level fractional factorial screening design where all user controllable parameters were tested against the response variables—height and width of a printed track. Significant parameters from this analysis were then used to build a regression equation for both height and width. It was observed that while the regression equation for height was accurate in predicting the output at intermediate levels, the regression equation for width was unable to do so and displayed signs of curvature. A higher order three-level regression model was then fit to the significant parameters for width and was found to be satisfactory in predicting process output. The errors observed between predicted outputs from the regression equations and actual output dimensions from the validation experiments were less than 2% and 3% for height and width respectively.