Modular AI Defect Inspection Solution for Efficient Semiconductor Equipment Upgrades

• Drones - Drone Payloads & Accessories
• Functional Applications - Manufacturing Execution Systems (MES)

• Equipment & Machinery
• Semiconductors

• Maintenance
• Quality Assurance

• Infrastructure Inspection
• Security Claims Evaluation

• Testing & Certification
• Training

Smasoft Technology Co., Ltd., a System Integrator that develops industrial automation software and offers AI application solutions, was commissioned by a semiconductor equipment manufacturer to implement AI inspection features into their Extreme Ultraviolet Light (EUV) pod inspection machines. The existing AOI software in the EUV pod inspection machines could only identify defective products but could not trace the cause of the defects. The manufacturer wanted to upgrade their machines with AI features to make the products more useful. The AI solution needed to complete the analysis of 380 images for a single pod within two minutes and inspect different materials simultaneously. This required multiple sets of AI models for interpretation. The solution also needed to be installed in a cabinet in the lower half of the machines, which posed a challenge due to the limited space. Smasoft needed to purchase a hardware solution with strong computing performance, stable operation, compact size, and flexible configuration to overcome these challenges.

Smasoft Technology Co., Ltd.

The customer in this case study is a semiconductor equipment manufacturer that manufactures Extreme Ultraviolet Light (EUV) pod inspection machines. The company was looking to upgrade their EUV pod inspection machines with AI features to not only identify defective products but also trace the cause of the defects. The manufacturer required an AI solution that could complete the analysis of 380 images for a single pod within two minutes and inspect different materials simultaneously. The solution also needed to be compact in size and flexible in configuration to fit into a cabinet in the lower half of the machines. The manufacturer chose to work with Smasoft Technology Co., Ltd., a System Integrator that develops industrial automation software and offers AI application solutions, to implement the AI inspection features into their machines.

Smasoft chose Advantech's solution, which included compact fanless system MIC-770, GPU expansion module MIC-75G20, and AI inference system MIC-730AI, due to its performance, modularity, and stability. The AINavi-AOI-Seq and AINavi-AOI-Semicon software were installed on MIC-770 and the AINavi-AOI-Semicon on the MIC-730AI. The EUV pod inspection machine could take images and transmit them to the MIC-770 for preprocessing. The pre-processed data was then analyzed by MIC-770 and MIC-730AI for different AI models. After the analysis, AINavi-AOI-Seq classified and graded the selected defective products according to the types and severity of defects, while generating quality inspection reports for customers to review. The MIC-770, an industrial-grade computer with a compact design, was suitable for applications where space was limited. The built-in Intel Core i series processors gave it high-efficiency computing capabilities and low power consumption characteristics. The MIC-730AI, a high-performance computer with embedded NVIDIA Jetson AGX Xavier processor, was used as an inference machine for deep learning and was responsible for analysis tasks of defect inspection.

• The implementation of the AI defect inspection solution was highly successful for the semiconductor equipment manufacturer. The solution not only identified defective products but also traced the cause of the defects, thereby assisting the manufacturer in further optimizing the manufacturing process and increasing yields. The manufacturer was so satisfied with the results that they decided to implement AI defect inspection for all their other automated machines. This demonstrated that the combined solution of Advantech and Smasoft was recognized for its valuable contribution by the industry. The solution was not only suitable for semiconductor equipment but also for other manufacturing industries, making the implementation of AI applications much easier.

• The AI solution can complete the analysis of 380 images for a single pod within two minutes.
• The solution can inspect different materials simultaneously.
• The solution is compact in size and flexible in configuration, fitting into a cabinet in the lower half of the machines.