Die Casting Porosity After Refurbishment: Vacuum Seal, Thermal Balance, or Both?

Classification: gas porosity from entrapped air/vapor, with a secondary cold-shut contribution.

Evidence pointing to gas:

• Dark, irregular walls. Entrapped air and vaporized die lubricant oxidize the void surface. Shrinkage voids have bright, vacuolated, interdendritic walls because they form in a vacuum as the last liquid is drawn away.
• Elongation along the fill direction and clustering near the overflow. This is where the flow front arrives last and where gas that was not evacuated collects - the overflow exists to receive exactly that gas.
• Round isolated pores are rare. Classic hydrogen or lubricant gas porosity would appear scattered through the section; here it is location-specific, which indicates flow and evacuation rather than gas in solution.
• Vacuum at 320 mbar against a ≤150 mbar target. More than double the target means the cavity was not evacuated, so air had to be compressed into the last-filled region.

Dominant cause: the vacuum seal that was installed without a leak check. The timing is decisive - the rejection rate changed immediately after the refurbishment, and the refurbishment introduced the one change that directly governs cavity evacuation. Vacuum degradation is the only change that explains both the location (near the overflow, last-filled) and the morphology (oxidized walls).

Secondary contributors: die surface at 168 °C is well below the 200–240 °C window, and metal at 655 °C is 25 °C low; together these shorten the time the section stays liquid and can create cold shuts that act as porosity initiation sites. The deeper cooling channels explain the low die temperature. Note the reduced spray time did not offset it - the deeper channels dominate.

Rank the fixes accordingly: vacuum first, thermal second, shot profile third.

Containment.

• Quarantine everything produced since the refurbishment, not just since the rejection rate was noticed. The rate moved from 1.8% to 6.4%, which means a substantial population of defective parts passed X-ray sampling before the trend was detected.
• 100% leak test is already required and is the functional gate; keep it and add X-ray or sectioning on a statistically valid sample per shift to confirm that leak test is actually sensitive to this defect. A leak test only detects porosity that breaches a sealed boundary - an internal void in the boss may pass leak test and still fail in service.
• MRB every quarantined lot. Sort to conforming, rework-eligible, and scrap, with records.

Impregnation: no, not as a production disposition.

• It seals porosity with a polymer; it does not remove the void or restore the local wall section. Structural margin and fatigue life in a 2.1 kg housing boss remain reduced.
• AlSi9Cu3 housings see elevated service temperatures; anaerobic and some methacrylate sealants degrade above roughly 150–200 °C, so a seal that passes the leak test at room temperature can reopen in service.
• The customer permits it only with written approval, and approval for a 6.4% systemic defect would be approving an uncontrolled process. It is appropriate only for a defined, small number of parts already in process, with engineering justification, temperature validation, and a documented concession.

Position to take. Do not spend money on impregnation capacity to protect an unfixed process. Fix vacuum and thermal conditions first; if porosity returns to 1.8%, the residual can be handled with normal sort and scrap economics.

Corrective actions, in order.

1. Vacuum system. Leak-check the entire cavity and shot system - seal, vacuum valve, hose, and tank - with a documented pressure-decay test. Repair and verify ≤150 mbar at the die before running production. Add a per-shot vacuum trace with an alarm and automatic part marking when vacuum exceeds the limit, so this never again depends on someone noticing an X-ray trend.
2. Thermal balance. Restore spray time to 6 s and re-establish the die thermal map by scanning the cover half at steady state. The 8 mm deeper cooling channels may need reduced flow or a different circuit routing to bring the boss to 200–240 °C. Verify with a thermal scan, not a single point reading.
3. Metal and shot. Return metal temperature to 680 °C and re-establish the 3.4 m/s final shot velocity using a shot-profile study; do not simply dial velocity up, since excessive velocity increases turbulence and air entrainment. Tune the velocity-to-vacuum relationship together.
4. Restore the original cycle-time target by attacking the real cycle losses, not the spray - the spray was cut to hide a cycle-time problem and created a quality problem.

Verification.

• Three-level DOE on vacuum level, die temperature, and shot velocity; 50-shot stabilization runs, then 30 consecutive shots sampled per condition.
• Acceptance: X-ray rejection ≤1.5% (the pre-refurbishment baseline), zero leak-test failures in 3 consecutive shifts, and process capability on the boss wall thickness.
• Run 500 shots and re-scan the die thermally at the end, since thermal balance drifts with production time.
• Only then release the die to normal production, with vacuum and die temperature on the control plan and monitored per shot.