feedback control

**Feedback Control** is a **process control strategy that uses downstream measurements (after processing) to adjust the process recipe for subsequent lots** — correcting for systematic drift by comparing output measurements to targets and applying corrections to future runs. **How Does Feedback Control Work?** - **Measure Output**: Measure the critical parameter after processing (e.g., post-etch CD). - **Calculate Error**: $e = ext{measured} - ext{target}$. - **Adjust Recipe**: Modify the recipe for the next lot to reduce the error (e.g., change etch time). - **Controller**: EWMA (Exponentially Weighted Moving Average), PID, or model-based controller determines the correction. **Why It Matters** - **Drift Compensation**: Automatically corrects for slow process drifts (chamber aging, gas line degradation). - **Standard Practice**: Feedback R2R (run-to-run) control is implemented on nearly every critical process step. - **Combines with Feed-Forward**: Most production uses combined feed-forward (inter-step) + feedback (intra-step) control. **Feedback Control** is **learning from the last lot** — using post-process measurements to continuously improve the recipe for subsequent production.