As wafers are high-tech and high-value products, every stage of the production process must be meticulously controlled. Production anomalies that affect wafer yield or result in scrap can lead to significant financial losses. In the semiconductor manufacturing process, wafer saw is part of the back-end processes, aimed at dicing an intact wafer into individual chips. When the wafer saw machine detects a positional deviation in the dicing process, it issues an anomaly notification. At this point, the alignment process is carried out manually via remote control. Once the alignment process is completed, dicing operations resume. Since wafer saw deviations rely on manual alignment rather than automated processes, time and manpower are required to align equipment parameters until alignment is complete.
Edge AI systems provide solutions for wafer saw inspection by addressing the challenges of manual alignment through automation technology. AI enhances the inspection accuracy of wafer saw, leading to improvements in operational efficiency and manpower savings. In application scenarios, industrial cameras capture image data of wafer saw with the assistance of lighting. These images are transmitted from the wafer saw machine to the edge AI system via a switch. The edge AI system analyzes and compares the image data to determine whether the wafer saw is deviated. If an anomaly is detected, the edge AI system initiates an alignment process, enabling the wafer saw machine to automatically execute the alignment workflow.
The PJAI-100-ON is an edge AI system designed for semiconductor wafer saw inspection and automated alignment. Powered by the NVIDIA® Jetson Orin Nano™ SOM, it provides rapid and precise detection of wafer saw misalignments. Equipped with 4GB/8GB LPDDR5 memory, the system ensures fast detection capabilities and multitasking for automated alignment control. In terms of connectivity, the system features an M.2 slot supporting NVMe storage, enhancing data processing speeds. Dual RJ45 GbE ports facilitate connections with switches and servers, enabling seamless handling of image data transmission, alignment control and data backup. Additionally, USB and COM ports are available for expansion and connectivity with other devices. This application architecture includes three edge AI systems, each connected to multiple wafer saw machines via a switch. These systems not only enable AI-based detection but also execute automated alignment, making the detection process more accurate and efficient.
Portwell offers comprehensive DMS (Design, Manufacturing and Service) solutions, assisting customers in addressing diverse application scenarios and market demands. Through these services, Portwell adds value by supporting design, manufacturing and post-sale service processes. Portwell RPET (Remote Portwell Engineering Toolkits) provides cross-system and cross-platform API and applications, ranging from system monitoring and device control to remote management. Its streamlined system design enhances IoT application capabilities, enabling customers to build IoT solutions with ease. With full-spectrum remote management, RPET delivers comprehensive system status monitoring and supports firmware updates to ensure stable and efficient system operations. This capability addresses growing demands for improved management efficiency and remote system access in IoT environments.
