The semiconductor industry, as the cornerstone of modern technology and electronics, has seen rapid expansion in response to increasing global demand. Amid ongoing chip shortages, companies are striving to enhance production capacity and efficiency. This study, conducted in collaboration with a leading semiconductor packaging and testing firm, focuses on developing an automated wafer laser marking recognition system tailored to the company’s operational needs. Wafer IDs, which are serial numbers laser-marked near the wafer notch, play a crucial role in tracking production data within the Manufacturing Execution System (MES). Frequent and accurate verification of these IDs is essential, particularly in mixed-production-line environments. However, manual identification is hindered by issues such as surface reflection, etching-induced blurriness, and inconsistent font styles, leading to inefficiencies and potential wafer damage. The current manual and OCR-based recognition methods are limited by low efficiency, risk of contamination, and poor generalizability. To overcome these limitations, this study proposes an end-to-end recognition framework based on object detection models. Alongside software development, dedicated hardware components were also designed and implemented. The resulting system significantly enhances automation and reliability in wafer identification, streamlining the production process and minimizing the need for human intervention.
Lin et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: