로고

로고
  • Semiconductor
  • Semiconductor
  • Semiconductor

    임픽스의 발자취를 소개합니다.

    AI Autonomous Manufacturing in the Semiconductor Industry

    The semiconductor industry is a high-precision manufacturing sector where nanoscale control and yield optimization are critical. To improve nanoscale process precision and yield, AI-based process prediction and optimization technologies are needed to analyze equipment conditions in real time and detect anomalies early.
    In order to respond to rapidly changing process conditions, expertise in clean room environmental control, inter-equipment control, and yield optimization algorithms is required.

    Required expertise: QbD/PAT regulatory response, GMP-based process knowledge, AI-based quality prediction

    Best Practice

    SK Hynix

    We delivered the D² (data lake) solution to Hynix's high-bandwidth memory manufacturing process to implement an AI-based data collection, analysis, and utilization environment. In addition, we installed Non-SECS Boxes on equipment that does not support standard communications to enable integrated data collection from all facilities.

    • BIG DATA

      D2

      Supplying data processing software for high-
      bandwidth memory manufacturing processes

    • PHYSICAL AI

      Non-SECS Box

      Delivery of physical box equipment equipped
      with software to connect non-SECS
      standard equipment and MES/ERP systems.

    Adopted Technologies

    Equipment monitoring

    Visualize equipment and sensor status in real time to intuitively understand the overall process situation. Alarms, sensor values, equipment logs are integrated into a web-based dashboard.

    • Manage equipment network status and RFID information
    • Log monitoring

    Data analsys

    Early identification of process risks, such as anomaly detection and quality trend analysis, based on multidimensional data. Key influencing factors can be identified through AI interpretation-based analysis results, such as SHAP and Permutation.

    • Monitor PC resource usage and resource usage by each program

    AI inference query

    View AI-predicted quality/facility anomaly results in real time and visually verify them. The most recent inference values and itemized results are automatically linked to the dashboard.

    • Monitoring sensor information from connected equipment
    • Proactively predict risk information

    Data integration/storage

    Unifiedly collect, cleanse, and store heterogeneous data from equipment logs, sensors, RFID, CCTV. Automatically configure high-quality datasets for AI learning and data linkage between equipment.

    USE CASE

    Starting with the construction of a smart factory in 2019,
    IMPIX has created best practices optimized for SMEs
    through various AX(AI Transformation) projects.