Abstract
Throughout the past decade, the field of printed electronics has gained increased research impetus and has started to make entryways into the consumer electronics product market. The main advantage of additive printed electronics is that it allows for the creation of lightweight and flexible electronic components that can be produced at a lower cost compared to traditional manufacturing methods. There are several printing techniques, conductible inks, dielectrics, solder pastes, electrically conductive adhesives, etc. that have been investigated in the literature to find the best solutions for several use cases. However, one issue that is prevalent across printing techniques and materials is the difference between the designed and actualized geometrical and electrical characteristics of printed traces. Owing to these differences, the overall achieved circuit performance differs slightly than the idealized one, and in some cases, significant differences might also be observed. To this end, the current study aims at developing a closed-loop control process for process parameter adjustment to achieve the desired electrical and physical characteristics of printed traces. In closed loop control, the output is compared to a desired setpoint, and any differences between the two are used to adjust the input. The first step in developing a closed-loop control algorithm to establish a relation between print process parameters and realized print characteristics. An inkjet printer and a particle free silver ink are used for this study. Particle-free inks are a liquid solution containing dissolved silver ions, which are beneficial for inkjet printing as they can help prevent printhead clogging. The aforementioned ink will be used for printing lines with different process parameters to establish the process parameter relation with the line geometry. This has been achieved by different machine learning methods including nonlinear regression, k-nearest neighbor regression, and Gaussian process regression. Each of these methods have been analyzed and further optimized to improve their accuracy and provide lower line width estimation errors. Ultimately a GUI for closed-loop control algorithm, which returns corrective print parameters given the desired and realized print characteristics and current print parameters, has been developed and validated.
