Laser defect diagnose

Laser Defect Diagnosis

Operates under the rules in ../../INSTRUCTIONS.md (reference-first, reference-only, interactive mode).

When to use this skill

When the operator has already identified a specific defect on a marked or cut sample and wants to know:

• What's causing it?
• Which parameter to change?
• Whether it's a process problem or an alignment / contamination problem.

This skill is more specific than laser-troubleshoot — assume general triage has been done, or skip it if the defect points clearly at a process cause.

References to load at start

• references/laser_ablation.md — HAZ, plume, threshold, multi-pulse, recipe (§4, §10–§12)
• references/laser_dicing.md — cutting-specific defects, polarization, dissimilar materials
• references/laser_welding.md — mode collapse, weld defects
• references/scanner_optimization.md — scan-delay-related defects (cross-reference)

Inputs to gather first

Ask the operator:

1. Process type — ablation / dicing / cutting / welding?
2. Material + thickness if relevant.
3. Visible defect — describe in plain words. Photo helpful.
4. Where on the feature — start of line, end, corner, edge, throughout?
5. Current parameters — power, PRF, pulse width, scan speed, hatch, repeats, polarization.
6. What changed — same recipe used to work, or new recipe?

→ Pause: confirm inputs.

---

Workflow

Step 1 — Localize the defect

The same word ("burn", "rough edge", "incomplete") can mean very different things. Resolve ambiguity first.

Match against this triage:

Defect location	Likely category	Branch
Position-specific — at line start / line end / specific corner	Scan delay	Hand to laser-delay-tuner
Distributed across the cut / mark	Process parameter	Continue to Step 2
Edge or wall only	Polarization, wall-angle, or material-class issue (dicing)	Continue to Step 2 (dicing branch)
Surface marks away from cut, or beam-path issue	Contamination or stray reflection	Hand to laser-troubleshoot
Random / intermittent	Drift or alignment	Hand to laser-troubleshoot

→ Pause: confirm classification before continuing.

Step 2 — Process-specific defect mapping

Branch by process. Cite the reference for every fix proposed.

Ablation defects

Per laser_ablation.md §4, §10, §11, §12.

Defect	Likely root cause	Fix
Excessive HAZ around the mark	Fluence too far above threshold (§4)	Lower power or increase scan speed
Mark width grows but depth doesn't	Saturation — past optimum fluence (§4)	Lower power
Inconsistent depth along a single line	Plume shielding (§10)	Cross-flow gas; or vacuum/inert atmosphere
No mark at all	Below ablation threshold (§4)	Increase fluence — more power, slower speed, or smaller spot
Mark fades after first pass	Heat-tolerant process needing more passes (§9)	Add repeats
Diminishing returns from added passes	Strong thermal accumulation (§12)	Add a Pause between passes, or lower per-pass power
Recast / re-deposited debris in kerf	Plume not clearing (§10)	Cross-flow gas; lower repeat rate; vacuum
Wider HAZ than expected for cold ablation	Pushed beyond cold-ablation regime into "warm" ablation (§7)	Drop fluence closer to threshold — lower power, longer process time

Dicing / cutting defects

Per laser_dicing.md §3–§6.

Defect	Likely root cause	Fix
Cut won't go all the way through	Wall-angle / depth limit reached (§3)	Add parallel passes (depth² scaling); or change focal-spot width
Different quality in X vs Y	Linear polarization with asymmetric coupling (§4)	Switch to circular polarization
Charred epoxy on PCB cut	Heat too high for the matrix (§6)	Increase scan speed; add cooling gas
Glass fiber not fully cut	Pulse energy insufficient for fiber (§6)	Increase pulse energy; or change wavelength
Through-cut into substrate on blind via	Pulse parameters wrong for shock-delamination regime (§6)	Adjust pulse width and PRF for delamination; reduce per-pulse energy
Recast on Si dicing edge	Ablative path on transparent material (§2)	Switch to stealth dicing if material allows
Tapered kerf walls	Wall-angle approaching the geometric limit (§3)	Widen entry with parallel passes; or use larger focal spot

Welding defects

Per laser_welding.md.

Defect	Likely root cause	Fix
Weld much shallower than expected	Peak intensity dropped below keyhole threshold (§4)	Increase power; decrease spot; or shorten pulse to raise peak
Spatter, porosity, or humping	Operating in keyhole near the collapse boundary (§4)	Stabilize keyhole — adjust intensity, add shield gas, slow scan slightly
Wide shallow nugget when keyhole was intended	In conduction or transition mode unintentionally (§2–§3)	Increase peak intensity to clear ~1.5 MW/cm² (§4)
Weld appearance OK but no bond	Insufficient energy coupling	Check material absorption at wavelength; check surface state

→ Pause: ask which row matches. Confirm before proposing the fix.

Step 3 — Validate the proposed fix is reachable

Before changing the recipe:

• Compute the new peak intensity / fluence (per laser_process_calculations.md).
• Check it stays within optic damage thresholds (per laser_optics_selection.md §5 — apply the 2×–10× safety factor).
• Confirm the laser, scanner, and stage envelope can deliver the new parameters (PRF range, max power, scan-speed limit).

If any of these fails, the proposed fix is not feasible. Either:

• Find an alternative parameter knob.
• Recommend a hardware change (different optic, different focus lens, beam expander adjustment).
• Stop and report the constraint.

→ Pause: present the proposed change + validation. Wait for go.

Step 4 — Test and report

Operator runs a test sample with the proposed change. Compare to baseline.

• Defect resolved → done.
• Defect partially resolved → loop back to Step 2 with updated symptom.
• Defect unchanged → re-classify; the original root-cause hypothesis was wrong.

→ Pause: loop until clean.

---

Output

A defect-diagnosis record (markdown):

• Defect description (with photo if provided)
• Process type, material, parameters
• Localization verdict (Step 1)
• Root cause identified (Step 2) with reference citation
• Proposed fix + feasibility check (Step 3)
• Test result (Step 4)
• Final recipe delta

Save to outputs/laser-defect-diagnose/<YYYY-MM-DD>-<defect-id>.md.

---

Cross-skill links

• Position-specific corner / line-end defects → laser-delay-tuner
• General performance drop without a clear defect → laser-troubleshoot
• Optic damage suspected → laser-optics-cleaning (for inspection)
• Recipe rebuild needed → laser-doe-ablation or laser-doe-dicing