Improve process control for high-mix manufacturing with ML-based virtual metrology

Semiconductor fabs face a growing demand for custom-designed products from diverse customers, which requires a great amount of manufacturing flexibility. Managing multiple products in a single facility involves coordination of various chambers and processes. The challenges of this high product mix manufacturing can reduce yield and raise costs.

Calibre Fab Insights offers an ML-based virtual metrology approach to effectively managing high product mix manufacturing. The virtual metrology model of Calibre Fab Insights incorporates design features and fault detection and classification data to create an advanced process control system to achieve target deposition thickness across multiple chambers, nodes and products. Simulation results demonstrate the effectiveness of integrating virtual metrology into real-time, high-volume manufacturing control systems, especially in run-to-run settings.

What you’ll learn:

• What challenges fabs face in high product mix manufacturing
• How chemical vapor deposition (CVD) thickness variations occur and how it impacts key transistor parameters
• What is virtual metrology and how is it used for process control
• How Calibre Fab Insights trained virtual metrology models
• How an advanced process control system uses the virtual metrology model for run-to-run control

Who should read this:

• Semiconductor process control engineers
• Semiconductor fab engineers interested in solutions to the challenges of high product mix manufacturing