Journal Article

AI-Driven Process Optimization Framework for Enhancing Print Quality in Aerosol Jet Printing

Choi, Joon Phil, Zhang, Haining, Kim, Yongrae, Cui, Lin, Moon, Seung Ki

2025

Overview

Aerosol jet printing (AJP) is a promising printing technology in the printed electronics industry, offering significant advantages by enabling the creation of highly customized patterns with a wide range of materials. However, broader application of AJP technology still faces challenges due to issues with printed line quality. According to previous research, inconsistencies in line width could lead to functional failures in the printed circuits. While various process parameters can be manipulated to customize line widths, such approaches often overlook the impact of line thickness on overall quality. This can lead to sub-optimal printed line quality within the design space. Furthermore, external factors such as temperature fluctuations and solvent evaporation can affect the printing process, in turn impacting the reliability and performance of the final electronic components. Therefore, optimizing key parameters and enhancing print quality in the AJP process is essential for improving the electrical functionality of the produced devices. This paper presents an AI-driven framework for optimizing printing quality in AJP. The proposed approach begins with Latin Hypercube Sampling for the initial experimental design. Bayesian Neural Networks (BNNs) are then utilized to model the printed line morphology, taking advantage of their ability to provide uncertainty estimates in predictions. The BNN models subsequently are integrated with a multi-objective genetic algorithm, which systematically identifies optimal process parameters that balance line width customization and line thickness maximization in a cost-effective manner. Furthermore, a convolutional neural networks model is developed for real-time monitoring of the printing process, enabling early detection of anomalies and continuous evaluation of process performance. Finally, experimental results demonstrate the validity of the developed approach for enhancing printing performance and anomaly detection in AJP.

• Publisher: Korean Society for Precision Engineering,Springer Nature B.V,한국정밀공학회
• Subject: Aerodynamics / Aerosols / Anomalies / Artificial intelligence / Artificial neural networks / Bayesian analysis / Customization / Design of experiments / Electronic components / Electronics industry / Energy Efficiency / Engineering / Gas flow / Genetic algorithms / Hypercubes / Industrial and Production Engineering / Jet printing / Latin hypercube sampling / Machine learning / Manufacturing / Mathematical models / Morphology / Neural networks / Optimization / Parameter identification / Pareto optimum / Performance evaluation / Printing / Process parameters / Real time / Regular Paper / Screen printing / Sustainable Development / Thickness / Variables / 기계공학