Semiconductor Behavior Recognition for Wafer Transfer Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of multi-chamber type cluster semiconductor manufacturing apparatuses makes it difficult to capture the root cause of productivity loss, as manufacturing operations are intricate and involve multiple heterogeneous units.
Innovation Solution
A behavior recognition device is used to process log data from semiconductor manufacturing apparatuses, building a machine learning model to recognize behaviors of manufacturing units, identify good/bad sequences and activities, and capture the root cause of bad behaviors, thereby optimizing manufacturing operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multi-chamber type cluster semiconductor manufacturing apparatus is used to increase throughput, then productivity is improved, but device complexity increases making it difficult to capture root cause of productivity loss
Solution Approach 1:
The patent segments the complex manufacturing operations into discrete activities and sequences, representing them as a state transition model with distinct states and transitions. This segmentation allows the system to break down the intricate multi-chamber operations into manageable units that can be individually analyzed and monitored, while maintaining the overall throughput benefits of the multi-chamber architecture.
Solution Approach 2:
The patent introduces a behavior recognition device as an intermediary between the manufacturing apparatus and the control system. This device processes log data and generates state transition models, serving as a mediator that translates complex operational data into actionable insights without requiring direct modification of the manufacturing apparatus itself.
2Measurement precision
If detailed monitoring of manufacturing operations is implemented to identify root causes, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The behavior recognition device performs self-service by automatically processing log data and generating state transition models without requiring external intervention. The system uses existing log data from the manufacturing apparatus and transforms it into behavioral insights through automated machine learning processes, reducing the need for additional monitoring hardware or manual analysis tools.
Solution Approach 2:
The patent creates a virtual copy of the manufacturing system's behavior through state transition models. Instead of adding physical monitoring devices to track every operation, the system generates a computational model that replicates the system's behavior patterns, allowing for detailed analysis without physical complexity.
Data Source
AI summary
A behavior recognition device for recognizing behaviors of a semiconductor manufacturing apparatus includes a storage device and a control unit. The storage device is configured to store log data of the semiconductor manufacturing apparatus. The control unit is cooperatively connected to the storage device, and configured to build a transition state model based on the log data to analyze behaviors related to wafer transfer sequences and manufacturing operations of the semiconductor manufacturing apparatus.