Early Detection of Corrosion-Induced Failures in Electronic Components
Blog post by Dr. Eyal Weiss CTO & Founder
Corrosion is one of the most prevalent failure modes in electronic components, leading to various issues ranging from component longevity to functional reliability. It can be caused by various environmental conditions, such as humidity, temperature, and acidity, and can manifest in multiple ways, such as surface pitting, component degradation, and irreversible damage. If left untreated, corrosion can lead to catastrophic failures in the field, resulting in costly recalls and damaged reputations. However, in many cases, it affects the longevity of the products by increasing their failure rate and shortening its mean time between failures.
In this article, we propose an AI-driven method for the early detection of corrosion contamination on soldering leads before they are assembled onto printed circuit boards (PCBs). This method relies on the correlation between the visual appearance of the leads and the extent of their corrosion. Corroded leads have noticeably rougher surfaces, pitting spots, and different light reflectance patterns than pristine leads. An AI model can be trained to detect these differneces in the component’s pictures and flag them for inspection.
We further present a real-world study in which corroded components were successfully detected during the pick-and-place process and subsequently failed during ICT testing. SEM/EDS and cross-section analysis were conducted to confirm the AI failure predictions on multiple components with corrosion in a full-scale production environment.
The method has been implemented on multiple production lines and is capable of inspecting all components without affecting throughput. It is exceptionally accurate, with a prediction accuracy of over 99.5%, and has been tested on over 2.5 billion components. As such, it is a powerful tool for preventing corrosion-induced failures in the field and improving the reliability of electronic components. It also allows explorability of the time-zero conditions of all the components in the assembled products after they were shipped out.
Red Next: Contaminated reel was successfully blocked by Cybord AI