Run-to-Run Advanced Process Control for Overlay Correction with Machine Learning

# Run-to-Run Advanced Process Control for Overlay Correction with Machine Learning

## Introduction & Motivation

Run-to-Run Advanced Process Control for Overlay Correction with Machine Learning addresses a central problem in Overlay error between successive lithography layers accumulates from stage repeatability, reticle placement, wafer stress, and prior-layer process history, and left uncorrected it erodes process window and yield as design rules shrink. Run-to-run advanced process control (APC) systems apply per-lot or per-wafer overlay corrections computed from prior metrology feedback, but naive linear feedback controllers struggle with nonlinear, tool-specific, and time-varying overlay signatures. Machine learning models learn overlay correction policies from historical metrology, scanner correctable parameters, and upstream process context to recommend per-wafer overlay corrections that reduce residual error while remaining stable under process drift and metrology noise.: how to Learn a run-to-run overlay correction policy from historical metrology and scanner correctable data that minimizes residual overlay error on subsequent wafers while remaining stable against metrology noise and process drift, compared against a linear APC baseline.. The difficult part is not producing a demonstration. It is maintaining a trustworthy system while equipment, data distributions, objectives, and organizations change.

The system consumes post-exposure overlay metrology maps (field and wafer-level x/y residuals), scanner correctable parameters (translation, rotation, magnification, higher-order terms), reticle and stage identity, prior-layer process history (etch, CMP, anneal), lot and wafer sequence context, and metrology tool identity with associated measurement uncertainty. It should produce recommended per-wafer scanner correctable adjustments, a predicted residual overlay distribution after correction, and a controller-confidence flag indicating when the recommendation should defer to the baseline linear APC controller. Those outputs become useful only when their uncertainty, provenance, and decision rights are explicit. A production implementation therefore couples modeling with data contracts, version control, monitoring, human review, and a safe fallback.

Learning objectives:

• Translate the topic into states, observations, decisions, constraints, and measurable outcomes.
• Establish a transparent baseline before introducing a complex learning architecture.
• Separate offline predictive performance from operational value and safety.
• Design validation that covers time drift, missing data, rare events, and subgroup behavior.
• Build a practical laboratory workflow that can be adapted to governed industrial data.

---

## Core Concepts & Theory

### Run-To-Run Feedback Control: Using Metrology From Completed Wafers Or Lots To Adjust Correctable Parameters For Wafers Not Yet Exposed, Closing The Loop With A Measurement Delay

Run-To-Run Feedback Control: Using Metrology From Completed Wafers Or Lots To Adjust Correctable Parameters For Wafers Not Yet Exposed, Closing The Loop With A Measurement Delay is treated as an engineering capability, not a slogan. Define its inputs, owners, update frequency, uncertainty, and failure response before selecting software. A useful design review asks what evidence would falsify the current model and how the system behaves when that evidence arrives.

### Overlay Error Decomposition: Separating Translation, Rotation, Magnification, And Higher-Order Field/Wafer Terms That Scanners Can Correct From Residual Components That Cannot Be Corrected By Available Knobs

Overlay Error Decomposition: Separating Translation, Rotation, Magnification, And Higher-Order Field/Wafer Terms That Scanners Can Correct From Residual Components That Cannot Be Corrected By Available Knobs is treated as an engineering capability, not a slogan. Define its inputs, owners, update frequency, uncertainty, and failure response before selecting software. A useful design review asks what evidence would falsify the current model and how the system behaves when that evidence arrives.

### Nonlinear Correction Policy Learning: Modeling Overlay Response As A Function Of Prior-Layer Process Context Rather Than Assuming A Fixed Linear Correctable-To-Residual Relationship

Nonlinear Correction Policy Learning: Modeling Overlay Response As A Function Of Prior-Layer Process Context Rather Than Assuming A Fixed Linear Correctable-To-Residual Relationship is treated as an engineering capability, not a slogan. Define its inputs, owners, update frequency, uncertainty, and failure response before selecting software. A useful design review asks what evidence would falsify the current model and how the system behaves when that evidence arrives.

### Controller Stability Under Feedback: Avoiding Oscillation Or Overcorrection When Metrology Noise Or Sparse Sampling Introduces Uncertainty Into The Correction Signal

Controller Stability Under Feedback: Avoiding Oscillation Or Overcorrection When Metrology Noise Or Sparse Sampling Introduces Uncertainty Into The Correction Signal is treated as an engineering capability, not a slogan. Define its inputs, owners, update frequency, uncertainty, and failure response before selecting software. A useful design review asks what evidence would falsify the current model and how the system behaves when that evidence arrives.

### Confidence-Gated Deference: Recognizing When A Learned Policy'S Prediction Uncertainty Is Too High And Falling Back To The Validated Linear Apc Baseline Rather Than Applying An Aggressive Correction

Confidence-Gated Deference: Recognizing When A Learned Policy'S Prediction Uncertainty Is Too High And Falling Back To The Validated Linear Apc Baseline Rather Than Applying An Aggressive Correction is treated as an engineering capability, not a slogan. Define its inputs, owners, update frequency, uncertainty, and failure response before selecting software. A useful design review asks what evidence would falsify the current model and how the system behaves when that evidence arrives.

The five concepts form a loop. Measurement creates evidence; modeling compresses evidence into a decision state; optimization proposes an action; execution changes the process; monitoring tests whether the original assumptions remain valid. Breaking that loop into disconnected dashboards and models prevents learning from operations.

---

## Mathematical Formulation

Choose notation that distinguishes measured values, latent states, model parameters, actions, and uncertainty. The following relations capture a compact starting point for run-to-run advanced process control for overlay correction with machine learning.

Overlay residual after correction:

$$ \mathbf{r}(w) = \mathbf{o}(w) - M \, \mathbf{c}(w) $$

Learned correction policy objective:

$$ \mathbf{c}^\star(w) = \arg\min_{\mathbf{c}} \; \mathbb{E}\big[\|\mathbf{o}(w) - M\mathbf{c}\|_2^2 \mid \mathbf{x}_{ ext{context}}(w)\big] + \lambda \|\mathbf{c} - \mathbf{c}_{ ext{prev}}\|_2^2 $$

Controller confidence gate:

$$ ext{defer to baseline if } \; \mathrm{Var}\big[\mathbf{c}^\star(w)\big] > au $$

These equations are abstractions. Every deployment must state units, sampling intervals, boundary conditions, missing-value behavior, and how constraints are enforced. Parameters estimated from historical data should not be interpreted causally unless the data-generating process and intervention assumptions support that claim.

Multi-objective decisions can be written as a constrained utility problem:

$$ x^*=\arg\min_{x\in\mathcal X}\sum_j w_j f_j(x)\quad\mathrm{subject\ to}\quad g_r(x)\leq0 $$

Weights express policy, not physical truth. Report the trade-off frontier when multiple settings are defensible.

---

## Advanced Theory & Extensions

### Probabilistic State and Uncertainty

A point estimate hides epistemic uncertainty, sensor noise, and future variability. Predict distributions or calibrated intervals when decisions depend on tail risk. Propagate uncertainty through downstream optimization instead of attaching an interval after a deterministic decision has already been made.

### Hybrid Mechanistic and Learned Models

Known conservation laws, topology, symmetries, and operating envelopes should constrain learned components. A hybrid model can use a mechanistic core plus a residual learner, or a learned surrogate with explicit feasibility projection. This often improves extrapolation and makes failure analysis more concrete.

### Causal and Counterfactual Analysis

Prediction answers what is likely under observed behavior. Intervention planning asks what will happen after an action changes that behavior. Use randomized experiments, natural experiments, or carefully defended causal assumptions before treating correlations as control levers.

### Hierarchical and Multi-Scale Reasoning

Industrial decisions occur at device, cell, line, plant, and enterprise scales. Local gains can create global queues or quality losses. Hierarchical models exchange summaries across time scales while preserving fast local safety loops.

---

## Computational Considerations

The raw computational cost is only one constraint. End-to-end latency includes acquisition, serialization, queueing, preprocessing, inference, optimization, communication, and actuation. Profile the whole path at median and tail latency.

• Data volume: streaming cost grows with sample rate, channel count, precision, and retention duration.
• Model cost: record training time, peak memory, inference latency, and energy on the target hardware.
• Numerical stability: scale features, monitor condition numbers, and test singular or missing inputs.
• Reproducibility: pin code, data snapshots, random seeds, environments, and model artifacts.
• Resilience: define behavior during network loss, stale inputs, service restart, and partial sensor failure.

A practical complexity budget separates fast-path decisions from slower analytical updates. Fast safety and control logic should not wait for a cloud retraining job. Expensive optimization can run asynchronously and publish bounded policies to a deterministic runtime.

---

## Practical Implementation Strategies

### 1. Frame the Decision

Name the decision, decision owner, action frequency, available alternatives, and cost of false positive and false negative outcomes. Do not begin with a model family.

### 2. Establish Data Contracts

For every field, specify source, unit, clock, valid range, missingness meaning, calibration state, and lineage. Enforce contracts at ingestion and quarantine invalid records rather than silently coercing them.

### 3. Build a Time-Aware Baseline

Use a chronological split and a simple model. Compare against current operating rules, last-value prediction, or a domain heuristic. A complicated method must beat these baselines on both accuracy and operational cost.

### 4. Validate in Shadow Mode

Run the system without action authority. Capture recommendations, operator responses, downstream outcomes, latency, and model confidence. Review disagreement cases and revise the decision policy.

### 5. Deploy with Bounded Authority

Use approval gates, rate limits, feasibility checks, and fallbacks. Increase autonomy only after stable shadow and canary evidence. Maintain a manual path that is tested rather than merely documented.

### 6. Operate a Learning Loop

Monitor inputs, outputs, outcomes, interventions, and data quality. Schedule reviews based on risk and drift, not an arbitrary retraining calendar. Every model update should have a change record and rollback artifact.

---

## Benchmark Datasets & Evaluation

A benchmark should approximate the deployment distribution and decision horizon. Random row splits overstate performance when adjacent records share time, equipment, batch, or specimen identity. Prefer forward-chaining evaluation, leave-one-site-out tests, and stress suites.

Primary evaluation dimensions:

• Mean Residual Overlay Error (Nm) After Correction Versus The Linear Apc Baseline: report a central estimate and uncertainty interval.
• Process Capability Index (Cpk) On Overlay Residuals Under The Learned Versus Baseline Controller: stratify by operating regime and data quality.
• Correction Stability, Measured As Run-To-Run Variance In Recommended Correctable Adjustments: measure the system effect, not only model output.
• Deferral Rate To Baseline Control And Its Correlation With Subsequent Residual Overlay Error: verify the result on production-like infrastructure.

Always include a naive baseline, a transparent statistical baseline, and the proposed method. Report performance by time period, asset, product family, and relevant risk group. Use ablations to identify which data sources or components create value.

---

## Key Challenges & Limitations

### Feedback-Induced Distribution Shift, Since Applied Corrections Change The Very Data Distribution The Model Was Trained On, Risking Compounding Drift If Unchecked

Feedback-Induced Distribution Shift, Since Applied Corrections Change The Very Data Distribution The Model Was Trained On, Risking Compounding Drift If Unchecked can invalidate an apparently strong offline result. Record the assumption explicitly, design a stress test, assign an owner, and define a bounded fallback. A dashboard without a response protocol only makes the failure more visible.

### Measurement Delay Between Exposure And Overlay Metrology Availability, Which Limits How Quickly The Controller Can React To Genuine Process Shifts

Measurement Delay Between Exposure And Overlay Metrology Availability, Which Limits How Quickly The Controller Can React To Genuine Process Shifts can invalidate an apparently strong offline result. Record the assumption explicitly, design a stress test, assign an owner, and define a bounded fallback. A dashboard without a response protocol only makes the failure more visible.

### Sparse Metrology Sampling Relative To Total Wafer Volume, Requiring The Model To Generalize Corrections To Unmeasured Wafers With Calibrated Uncertainty

Sparse Metrology Sampling Relative To Total Wafer Volume, Requiring The Model To Generalize Corrections To Unmeasured Wafers With Calibrated Uncertainty can invalidate an apparently strong offline result. Record the assumption explicitly, design a stress test, assign an owner, and define a bounded fallback. A dashboard without a response protocol only makes the failure more visible.

### Cross-Layer And Cross-Product Generalization, Since Overlay Signatures Differ By Layer Pair, Reticle Set, And Product Design Rule, Limiting Direct Policy Transfer

Cross-Layer And Cross-Product Generalization, Since Overlay Signatures Differ By Layer Pair, Reticle Set, And Product Design Rule, Limiting Direct Policy Transfer can invalidate an apparently strong offline result. Record the assumption explicitly, design a stress test, assign an owner, and define a bounded fallback. A dashboard without a response protocol only makes the failure more visible.

Limitations should travel with the model artifact. State where the system was validated, where it was not, and what conditions trigger abstention. Accuracy alone cannot justify action when consequences are asymmetric.

---

## Hyperparameter Tuning

Tune against a validation period that precedes the final test period. Optimize a deployment-aligned score that includes reliability and cost, then confirm robustness across seeds and operating regimes.

Avoid selecting a setting from a single best trial. Prefer a stable region where nearby settings behave similarly. Log the full search space, unsuccessful trials, random seeds, and resource consumption.

---

## Real-World Applications & Case Studies

### In-Line Run-To-Run Overlay Correction Integrated Into The Lithography Cell'S Apc System For Critical Layers With Tight Overlay Budgets

For in-line run-to-run overlay correction integrated into the lithography cell's APC system for critical layers with tight overlay budgets, begin with one decision, one accountable owner, and one measurable baseline. Run the proposed system in shadow mode, compare its recommendation with actual outcomes, and expand authority only after reliability and recovery behavior are demonstrated.

### Reticle And Scanner Qualification Support, Identifying Systematic Correctable-Parameter Drift Attributable To Specific Reticle Or Stage Hardware

For reticle and scanner qualification support, identifying systematic correctable-parameter drift attributable to specific reticle or stage hardware, begin with one decision, one accountable owner, and one measurable baseline. Run the proposed system in shadow mode, compare its recommendation with actual outcomes, and expand authority only after reliability and recovery behavior are demonstrated.

### Process Window Recovery Analysis Quantifying How Much Overlay Budget Is Freed For Other Process Variation When Learned Corrections Outperform The Linear Baseline

For process window recovery analysis quantifying how much overlay budget is freed for other process variation when learned corrections outperform the linear baseline, begin with one decision, one accountable owner, and one measurable baseline. Run the proposed system in shadow mode, compare its recommendation with actual outcomes, and expand authority only after reliability and recovery behavior are demonstrated.

A credible case study reports the previous process, deployment boundary, data period, intervention policy, operational metric, uncertainty, and failure handling. Percentage improvement without a baseline definition is not sufficient evidence.

---

## Integration with Other Methods

Run-to-Run Advanced Process Control for Overlay Correction with Machine Learning is usually one component of a larger decision system:

• Overlay Metrology Systems That Supply Field And Wafer-Level Residual Maps As Both Training Data And Real-Time Feedback Signals: supplies a complementary capability and should exchange versioned data through a documented contract.

• Lithography Scanner Control Interfaces That Receive Recommended Correctable Parameters For The Next Exposure: supplies a complementary capability and should exchange versioned data through a documented contract.

• Fab-Wide Apc Orchestration Platforms That Arbitrate Between Multiple Run-To-Run Controllers Across Process Steps And Layers: supplies a complementary capability and should exchange versioned data through a documented contract.

Integration contracts should specify schemas, units, timestamps, confidence semantics, version compatibility, retry behavior, and ownership. Keep safety interlocks independent from probabilistic services unless the complete path is engineered and certified accordingly.

---

## Summary & Key Takeaways

Run-to-Run Advanced Process Control for Overlay Correction with Machine Learning can improve Overlay error between successive lithography layers accumulates from stage repeatability, reticle placement, wafer stress, and prior-layer process history, and left uncorrected it erodes process window and yield as design rules shrink. Run-to-run advanced process control (APC) systems apply per-lot or per-wafer overlay corrections computed from prior metrology feedback, but naive linear feedback controllers struggle with nonlinear, tool-specific, and time-varying overlay signatures. Machine learning models learn overlay correction policies from historical metrology, scanner correctable parameters, and upstream process context to recommend per-wafer overlay corrections that reduce residual error while remaining stable under process drift and metrology noise. when technical modeling and operational governance are designed together. Begin with a bounded decision and measurable baseline; encode data and safety contracts; validate chronologically; deploy with constrained authority; and monitor outcomes rather than model scores alone.

Core principles:

1. Run-To-Run Feedback Control: Using Metrology From Completed Wafers Or Lots To Adjust Correctable Parameters For Wafers Not Yet Exposed, Closing The Loop With A Measurement Delay: define it operationally and test it under representative stress.
2. Overlay Error Decomposition: Separating Translation, Rotation, Magnification, And Higher-Order Field/Wafer Terms That Scanners Can Correct From Residual Components That Cannot Be Corrected By Available Knobs: define it operationally and test it under representative stress.
3. Nonlinear Correction Policy Learning: Modeling Overlay Response As A Function Of Prior-Layer Process Context Rather Than Assuming A Fixed Linear Correctable-To-Residual Relationship: define it operationally and test it under representative stress.
4. Controller Stability Under Feedback: Avoiding Oscillation Or Overcorrection When Metrology Noise Or Sparse Sampling Introduces Uncertainty Into The Correction Signal: define it operationally and test it under representative stress.
5. Confidence-Gated Deference: Recognizing When A Learned Policy'S Prediction Uncertainty Is Too High And Falling Back To The Validated Linear Apc Baseline Rather Than Applying An Aggressive Correction: define it operationally and test it under representative stress.

The durable deliverable is not a notebook. It is a maintained learning system with evidence, ownership, recovery behavior, and an explicit path from observation to decision.

---

## Appendix: Practical Labs

### Lab 1: Build a reproducible synthetic operating dataset

This lab creates correlated features, a noisy target, and a chronological split. Replace the synthetic generator with governed source data while retaining the assertions and metadata checks.

import numpy as np

rng = np.random.default_rng(101548)
n_samples, n_features = 720, 6
time = np.arange(n_samples)
features = rng.normal(size=(n_samples, n_features))
features[:, 1] = 0.65 * features[:, 0] + 0.35 * features[:, 1]
features[:, 2] += 0.4 * np.sin(time / 35.0)
weights = np.array([1.4, -0.9, 0.6, 0.25, -0.35, 0.8])
target = features @ weights + 0.3 * np.sin(time / 20.0)
target += rng.normal(0.0, 0.25, n_samples)

cut = int(0.75 * n_samples)
x_train, x_test = features[:cut], features[cut:]
y_train, y_test = target[:cut], target[cut:]

assert x_train.shape == (540, 6)
assert x_test.shape == (180, 6)
assert np.isfinite(features).all() and np.isfinite(target).all()
print("Run-to-Run Advanced Process Control for Overlay Correction with Machine Learning")
print("train/test:", x_train.shape, x_test.shape)
print("target mean/std:", round(target.mean(), 3), round(target.std(), 3))

### Lab 2: Train and evaluate a transparent baseline

A ridge baseline is deliberately simple. It establishes whether a more complex method adds value and supplies a stable reference for the primary metric, mean residual overlay error (nm) after correction versus the linear APC baseline.

import numpy as np

def standardize_fit(x):
 mean = x.mean(axis=0)
 scale = x.std(axis=0)
 scale[scale < 1e-9] = 1.0
 return mean, scale

def ridge_fit(x, y, alpha=1.0):
 design = np.column_stack([np.ones(len(x)), x])
 penalty = np.eye(design.shape[1])
 penalty[0, 0] = 0.0
 return np.linalg.solve(design.T @ design + alpha * penalty, design.T @ y)

mean, scale = standardize_fit(x_train)
xtr = (x_train - mean) / scale
xte = (x_test - mean) / scale
coef = ridge_fit(xtr, y_train, alpha=1.0)
prediction = np.column_stack([np.ones(len(xte)), xte]) @ coef
rmse = float(np.sqrt(np.mean((prediction - y_test) ** 2)))
r2 = 1.0 - float(np.sum((prediction - y_test) ** 2) / np.sum((y_test - y_test.mean()) ** 2))

assert np.isfinite(coef).all()
assert rmse >= 0.0 and r2 <= 1.0
print("RMSE:", round(rmse, 4))
print("R2:", round(r2, 4))

### Lab 3: Tune regularization without test-set leakage

The final chronological segment remains untouched. Candidate settings are compared on a validation tail drawn only from the training period.

import numpy as np

split = int(0.8 * len(xtr))
x_fit, x_val = xtr[:split], xtr[split:]
y_fit, y_val = y_train[:split], y_train[split:]
grid = [0.0, 0.01, 0.1, 1.0, 10.0, 100.0]
scores = []

for alpha in grid:
 candidate = ridge_fit(x_fit, y_fit, alpha=alpha)
 val_prediction = np.column_stack([np.ones(len(x_val)), x_val]) @ candidate
 val_rmse = float(np.sqrt(np.mean((val_prediction - y_val) ** 2)))
 scores.append((val_rmse, alpha))

best_rmse, best_alpha = min(scores)
assert best_alpha in grid
assert all(np.isfinite(score) for score, _ in scores)
print("best alpha:", best_alpha)
print("validation RMSE:", round(best_rmse, 4))

### Lab 4: Add an online drift and intervention monitor

This monitor separates model error from input drift. In production, route alerts through approval and fallback policies appropriate to the consequence of a wrong action.

import numpy as np

reference = xtr[-160:]
recent = xte[-60:].copy()
recent[:, 0] += 0.8 # controlled drift injection

mean_shift = np.abs(recent.mean(axis=0) - reference.mean(axis=0))
pooled_scale = np.maximum(reference.std(axis=0), 1e-9)
standardized_shift = mean_shift / pooled_scale
drift_score = float(np.max(standardized_shift))
warning_threshold = 0.5
critical_threshold = 1.0

if drift_score >= critical_threshold:
 action = "fallback_and_investigate"
elif drift_score >= warning_threshold:
 action = "review_and_collect_labels"
else:
 action = "continue_monitoring"

assert drift_score >= 0.0
assert action in {"fallback_and_investigate", "review_and_collect_labels", "continue_monitoring"}
print("drift score:", round(drift_score, 3))
print("recommended action:", action)

---