Sub-fab environments, containing gas distribution, vacuum pumps, chemical supply and exhausts supporting wafer processing equipment, are critical facilities for semiconductor fabs. Despite extensive digitalization in the industry, Monitoring and Maintenance of sub-fabs still involve a significant amount of manual intervention. Applications such as valve operation, monitoring of analog flow sensors, pressure sensors for gases etc. require human presence, despite the hazards of toxic gases and pressurized fluids. The large footprint and complexity of this infrastructure adds to the challenge of detecting and resolving system issues.
We propose a method to develop 3D Digital Twins for Sub-Fab facilities. The key enabler for this is a Robot platform that contains sensors such as RGBD and LIDAR, and audio sensors moving autonomously through the sub-fab, and developing context-aware 3D models of the entire environment using Semantic SLAM techniques. Through this process, critical infrastructure identifiers (such as Hazard symbols and Pipe Flow directions) and physical devices (analog pressure/flow gauges, valve actuators etc.) are localized and their output values tracked, creating Position-linked Telemetry. By combining this spatial information with Floor Plans and aligning it with point clouds collected from Building Information Models (BIM), we construct a digital twin of the sub-fab environment. The robot subsequently uses this 3D Digital Twin to traverse through the facility, using the mapped device positions to track sensor data for remote data reporting.
Considering the large-scale, evolving nature of sub-fabs, enabling near real-time, remote inspection linked to physical positioning has potential to improve both human safety and efficiency in sub-fab operations. Linking this data through SEMI E167, which involves sub-fab to fab communication, would provide a concrete link between sub-fab inspection and maintenance and the rest of the enterprise. The Digital Twin and the Robot platform opens further avenues for investigation into localization of visual and audio anomalies of equipment, to identify causes based on existing Piping & Instrumentation Diagrams (P&ID).
