What Data for This Category Looks Like
Data for this category originates from MES systems, equipment FDC monitoring devices, defect detection and metrology equipment, process recipe libraries, and experimental datasets. Update frequencies include per wafer batch, real-time, and daily. The data includes process flow cards, batch run records, equipment recipes, SPC charts, exception handling documents, defect maps, and yield reports. Core fields include wafer batch numbers, nanoscale process parameters, FDC thresholds, wafer map coordinates, and defect type codes. The data structure and parameter precision align with nanoscale control requirements for semiconductor manufacturing.
What Constraints Do These Characteristics Impose on Workflow Orchestration
The extremely long wafer fabrication process chain includes hundreds of processes and rework paths. It requires workflows to split nodes by wafer batch to avoid cross-batch data confusion. The coupling between nanoscale parameters and cleanroom environments requires variables passed by workflows to be precisely bound to the corresponding batch’s equipment data, preventing misjudgments from parameter drift. Real-time updated FDC data and daily generated yield reports require workflow node trigger timing to match data refresh rhythms, ensuring the timeliness of equipment inspection and maintenance Q&A. Limited rework windows require workflows to configure fast-failure handling branches to avoid abnormal spread to multiple batches.
How to Set Configurations
This configuration is based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
variables | Bind wafer batch numbers, real-time FDC parameters, equipment recipe ID | Match the data sources and update rhythms of this category, cover core parameters for equipment inspection |
nodeId | Configure separately for equipment inspection nodes, maintenance approval nodes, and exception handling nodes | Adapt to the node splitting requirements of extremely long process chains, facilitate cross-department collaboration |
| HTTP request timeout | 600 seconds | Handle long-duration scenarios of equipment FDC data reading, avoid process interruptions |
| Global variable trigger condition | Update synchronously when the request carries the custId parameter | Adapt to dynamic variable update requirements for multi-customer scenarios on production lines |
| Response variable extraction | Enable response content extraction function | Support extracting inspection results returned by HTTP requests into variables available for downstream nodes |
| Workflow export permissions | Configure according to production line department roles | Match the permission control requirements of semiconductor manufacturing, prevent leakage of process parameters |
Three Common Mistakes
- The response extraction for workflow HTTP request nodes has no available configuration entry, making it impossible to generate downstream variables. The symptom is that no variable extraction options appear in the interface, and downstream nodes cannot obtain inspection results. The cause is that the node’s variable extraction switch is not enabled, and the extraction rules are not configured correctly.
- Global variables are not updated based on query parameters, leading to incorrect custId binding. The symptom is that variable values for different production lines do not change with request parameters, resulting in data crossover. The cause is that no variable trigger conditions are configured, and fixed variable values are hardcoded directly.
- Exported workflows cannot adapt to different production line configurations, leading to mismatched node parameters. The symptom is that imported or exported workflows have missing or incorrect node parameters. The cause is that reusable variable templates are not used, and production line-specific equipment parameters are hardcoded.
How to Confirm Correct Configuration
- Trigger a simulated wafer batch inspection process, check whether
variablesare correctly bound to the corresponding batch’s MES data and FDC parameters. - Send test requests carrying different custId values, verify whether global variables update synchronously with the parameters.
- Call the HTTP request node, confirm that response content can be extracted into variables available for downstream nodes.
- Export the workflow configuration and import it into a test environment, verify that all node parameters and permission configurations are not lost.
The values provided are common starting points and should be measured against the reader's own samples.