Wafer-Metrology Correlation Director — Advanced Logic Foundry

The mandate

Process-control teams are responding to inconsistent critical-dimension, overlay and defect signals from optical, CD-SEM and high-resolution inspection systems while downstream electrical structures show a different ranking of wafer risk. Recipe owners correlate on hand-selected sites, focus and charging effects vary by layer, and no one can explain whether a reported shift is physical, sampling-driven or measurement-induced. The defined problem is to restore a decision-grade measurement chain for priority process windows.

The deliverables are a metrology-use-case taxonomy, reference-wafer set, measurement-system analysis, sampling design, cross-tool matching model, physics-to-electrical correlation, excursion decision rules and ownership blueprint. Artefacts must bind layer, pattern context, site selection, recipe revision, tool state and reference standard. The work will state where correlation is conditional instead of forcing one global conversion among fundamentally different measurement modalities.

Four milestones govern the engagement: prioritised measurement conflicts and reference plan by 23 October 2026; completed gauge, sampling and tool-matching evidence by 4 December; verified correlations against electrical monitors and split wafers by 15 January 2027; and accepted control rules, training and client-led reproduction by 26 February. Each milestone uses production-relevant wafers whose genealogy and handling are preserved.

Acceptance requires integration owners to make the same hold or release decision from the approved evidence, metrology teams to reproduce matching after maintenance, and yield engineering to confirm predicted electrical risk on blinded samples. The sponsor returns one consolidated defect list within five working days. A high coefficient on pooled data cannot close acceptance when layer context, range restriction or outlier handling makes the relationship unstable.

The client provides metrology recipes, raw signals, reference artefacts, tool maintenance histories, wafer maps, electrical-test results, process splits, cleanroom access and empowered integration owners. It funds agreed reference sampling and protects production wafers from unrecorded remeasurement. The director cannot release product, change process specifications, select an equipment supplier or use customer designs outside the controlled correlation environment.

Why this is external work

Each measurement group has invested in its own recipes and naturally explains conflicts through another tool’s limitations. The foundry needs an independent specialist who can test the physics, sampling and decision consequence without defending an installed platform. The work ends when client engineers reproduce bounded correlations and know precisely when a result is not transferable.

What you will own

• Classify measurement decisions by layer, feature, range, spatial scale, destructiveness, latency and downstream electrical consequence.
• Establish reference wafers and structures whose stability, handling, contamination and ageing limits are independently understood.
• Direct gauge studies covering repeatability, reproducibility, matching, operator effect, charging, focus and recipe sensitivity.
• Redesign sampling around pattern context and spatial mechanism rather than convenient common sites or pooled wafer averages.
• Correlate dimensional, overlay and defect signals with split conditions, physical analysis and electrical monitors using guarded inference.
• Define tool-health, maintenance recovery, excursion, hold and escalation rules with owners and false-decision costs.
• Transfer reference custody, correlation limits, recipe governance, open discrepancies and witnessed client reproduction.

Candidate qualifications

• Directed advanced-semiconductor metrology or process-control programmes spanning multiple measurement modalities and production layers.
• Can evidence a strong apparent correlation overturned by sampling, range, tool-state or measurement-physics investigation.
• Understands optical scatterometry, CD-SEM, overlay, e-beam inspection, reference standards and electrical-monitor relationships.
• Has designed production measurement-system studies without consuming impractical wafer volume or hiding rare systematic failures.
• Challenged equipment suppliers and internal recipe owners using reproducible raw evidence rather than platform preference.
• Translated complex correlation limits into operational hold, release and maintenance decisions for integration leaders.

Non-negotiables

• Will work in Portland and Hillsboro and attend the scheduled Hsinchu equipment review.
• Has personally correlated semiconductor metrology to electrical or yield consequence on production-relevant wafers.
• Accepts milestone rejection when reference stability, sampling context or tool genealogy is not traceable.
• Will disclose metrology-equipment, foundry, analytics and process-control software relationships before mobilisation.

1. 49 words maximum. Which impressive metrology correlation failed when you changed sampling context or tool state?
2. 49 words maximum. How would you prove whether a reported shift is physical rather than measurement-induced?
3. 49 words maximum. What evidence makes a cross-tool conversion operationally safe but deliberately bounded?