- PE
- Process Engineer responsible for unit module recipe performance
- PIE
- Process Integration Engineer responsible for end-to-end device integration and yield
- EE
- Equipment Engineer responsible for tool uptime, hardware maintenance, and calibration
- Evidence Gap
- Critical information currently unverified or missing that directly impacts engineering judgment
Distinct Engineering Perspectives Complement Each Other
PE focuses on a process module, PIE on cross-step structures and electrical behavior, and EE on hardware and maintenance; yield, inspection, and metrology teams may also participate. Responsibilities and collaboration vary by organization and problem type.
- 01ObservationTeam Aligns on the Same Physical Anomaly
Which wafers, spatial die coordinates, and process steps exhibit the excursion?
- 02BaselinePE and PIE Agree on Valid Reference Conditions
Are product codes, recipes, inspection thresholds, and material lots truly comparable?
- 03HypothesesDeconstructing Disagreements into Testable Claims
For example: upstream incoming litho CD shifts versus downstream etch chamber drift require distinct validation splits.
- 04ActionDelivering Clear Recommendations with Assigned Responsibilities
Who prepares supplementary evidence, performs checks, and reviews results is assigned according to the customer workflow.
Framing Disagreements as Two Inspectable Engineering Questions
Illustrative scenario:Residue is found after etching. One explanation is an incomplete upstream pattern opening; another is changed etch performance. Check post-development inspection coverage and representativeness, then align etch metrology and process records. Normal center sites do not rule out edge issues, and sensors within control limits do not guarantee every relevant state is normal.
Transparent Evidence Standards Outweigh Excessive Numeric Precision
- Record data source, system version, spatial coordinates, timestamps, and inspection sampling coverage.
- State units, denominators, rounding conventions, and measurement uncertainty; retain reproducible inputs without mandating arbitrary float precision.
- Explicitly state which data points support or weaken each candidate hypothesis, avoiding confusing non-significant tests with affirmative proof of absence.
- Propose discriminating verification and state how each result would change the judgment; retain an unresolved status when alternatives cannot be distinguished.
Subjective terms like 'severe drift' must be accompanied by explicit quantitative definitions and sensor data. Where client compliance demands immutable audit records, secure storage and access controls must be maintained.
Bring Reports into the Existing Responsibility Framework
| Workflow Stage | Key Questions to Clarify |
|---|---|
| Review Baseline Data | Who verifies wafer identities, inspection recipe consistency, and metrology measurement uncertainty? |
| Authorize Verification Splits | Who approves experimental wafers, split process windows, recipe parameters, and abort thresholds? |
| Execute Physical Actions | Who conducts equipment checks, recipe parameter updates, or lot holds in the MES? |
| Confirm Findings | Who evaluates post-split evidence and signs off on lot release or continued investigation? |
WAFERPILOT / APPLICATION TO THIS TOPIC
Delivering Shared, Auditable Investigation Dossiers
WaferPilot consolidates visual inspection observations, computational workflows, supporting and contradictory evidence, and proposed actions into a single shared workspace. The resulting report serves as verifiable input for human decision-making, without superseding fab governance.
Deliverables for Engineers:A unified team report: verified observations, competing candidate mechanisms, supporting and contradictory data points, and prioritized next steps with assigned stakeholder roles.
View the Current Demo and Capability Scope →Related Reading & References
Below are verifiable industry references; specific processes, permissions, and acceptance criteria are governed by fab site specifications.