Optimization of plasma-assisted surface treatment for adhesive bonding via Artificial Intelligence

Abstract

In this research, Artificial Intelligence (AI) was used to support the optimization of six bonding process parameters for maximal joint strength and minimal production costs. Two industrial bonding processes were investigated, one from electronic potting and another from the manufacturing industry. The focus was on optimizing the plasma treatment of the substrate materials. Two approaches for optimization were compared, namely the traditional approach where the adhesive expert proposes experiments and interpret the results, and an AI approach with Bayesian optimization and Gaussian process models. Similar joint strengths could be achieved via the Bayesian optimization approach with 40% less budget to find the optimum compared to the traditional approach. Additionally, in the electronic potting process, the AI approach resulted in 18% reduction in production cost, while achieving a similar joint strength, compared to the traditional approach. Ageing of the samples did not result in a significant drop in joint strength nor changes in failure type or mechanism. This indicates that AI can support adhesive experts to find the optimal bonding process settings and manufacture robust and cost-efficient adhesive bonds.