Product Traceability in Microelectronics Manufacturing

Abstract

Alongside other Industry 4.0 practices, modern product traceability provides a valuable digital mechanism for process and quality control, product refinement, recall management and production efficiency enhancement. Although well established in some industries, traceability is rarely found in microelectronics due to the challenges that emerge from the small-sized, low cost and high volume nature of the products. This project was split into two parts; Creating a general internal traceability framework for use in any manufacturing industry, then using it to develop the foundations of a traceability system for an existing microelectronics production factory. Microelectronics manufacturer Rakon Limited produces products as small as 1.6mm x 2mm at a high volume and were used as a case study for this research. During the development of the general framework, it was found that the ideal traceability system is built up from many low-cost modular scanning stations. A scanning station can read a product's unique ID and link it to production data. The more scanning stations there are in the production line, the more comprehensive and valuable the traceability system can be. Because of this, scanning stations should be as low-cost as possible. By following the framework, technologies were explored for traceability in microelectronics and found that laser marked data matrices were the most suitable. This allowed a 16 digit serial number to be stored in an area of 2.1mm x 2.1mm with the potential of further reduction. As an industrial scanning station did not meet the requirement set, a custom one was developed for under $400 (NZD), based around low-cost open-source machine vision. The prototype contained a single-board computer, diffuse dome lighting system, camera, and a software application utilising remote server processing. This prototype was able to read the ID of 40 products simultaneously in 3.2 seconds. Occasional accuracy issues were found and could be mitigated through more premium decoding software but would have a higher cost. As a result, foundations of a system was created that could link physical microelectronic products to their corresponding production data, providing a significant opportunity for future data-driven applications.