|
|
AI to Predict Compatibility Between Solder Paste Residues and Coatings
Materials Tech |
|
Authored By:Mélanie MATHON, Mehdi GUELMOUSSI Inventec Performance Chemicals Bry-Sur-Marne, France SummaryThe electronic industry has always had to adapt to many challenges, such as miniaturization, higher performance or lifespan, and product reliability. To ensure a greater reliability, Printed Circuit Board Assemblies (PCBAs) need to be protected from harsh environmental conditions. That’s why, conformal coatings have been used for years in the automotive, military, aerospace, and energy industries. The aim of this study is to demonstrate the benefits of developing an AI solution to optimize the performance and robustness of conformal coating solutions. For optimum performance of these coatings, it is advisable to clean the PCBAs before application. Market constraints in terms of cost and time are forcing manufacturers to reduce the number of steps in the process. Sometimes, the cleaning stage is not considered as an option and the use of no-clean residues is preferred. An incompatibility between the solder paste residue and the coating applied could have more adverse effects than without coating. Then, physical and chemical compatibility between the two products must be ensured. These requirements triggered a new challenge: on one hand, there are many different types of coatings on the market; on the other hand, there are numerous possibilities for solder paste formulations. The ingredients of these two products are most of the time confidential and different from one manufacturer to the other. Today, there is no simple way to determine the reliability between a solder paste and a conformal coating. It requires physical and reliability measurements that often involve lengthy and costly qualification processes involving numerous tests. For these reasons, the development of a simple test to predict the compatibility between a solder paste and coating is helpful. Artificial Intelligence (AI) can help to determine a model to assess the compatibility. This paper shows how AI is used to predict compatibility results between solder paste residues and coatings. A physical compatibility study is carried out to benchmark some combinations of solder pastes and coatings of the market. Residues surface energy, coating surface tension and physical application tests are used as input data for the AI calculation. Results between AI prediction and the physical tests are compared to choose the right AI model for a good prediction of the compatibilities. ConclusionsThis study has allowed the development of a repeatable and reproducible test method based on the physico-chemical properties of solder paste residues and conformal coatings. This method is in agreement with the experiments carried out (application methods to define the adhesion) and the experimental observations of the compatibility of coatings on soldering paste residues. The method using surface energy and surface tension measurements has the advantage of being a simple and quick method to implement. The other advantage is that there is no need to re-test all residue/coating combinations for each new incoming material. Once a conformal coating or a solder paste residue has been tested, it is entered into the database, and testing the new coating or residue is the only information needed to see how compatible it is with its counterparts. With this method, the coating compatibility classification can be predicted with a simple law using the surface tension/energy measurements. These results were obtained using A.I. to correlate data from the application method. As part of the development of a new test method for the determination of coating/residue compatibilities, the use of AI has made it possible to:
Our latest research highlights the potential of AI to surpass human capabilities in identifying complex relationships. The use of automated learning models makes it possible to conceptualize new approaches to formulation analysis. Initially Published in the SMTA Proceedings |
|
Comments
|
|
|
|
|