Common build failures: what actually goes wrong on the shop floor

A single recurring screwup originates in the EDM step and propagates through nitride into service.

EDM cavity that spalls under nitride. White-layer flakes show up in the cavity after a few thousand hits. The compound layer is visibly cracked under a loupe. The diffusion zone is intact, the bulk is intact, but the surface comes off in sheets. The root cause is EDM recast left in place before nitriding. Sinker and wire EDM both leave a 10-50 µm layer of re-solidified, micro-cracked steel on the cavity surface. Nitride that surface and the compound layer grows into a cracked substrate. The cracks propagate through the case on the first thermal cycle and the compound layer separates.

Surface integrity inspection before nitride catches this. Cross-section a witness coupon EDM'd in the same setup as the cavity and look for recast under 200x magnification. Specify Ra under 0.4 µm on the cavity going into nitride, and require recast removal by polish, vapor hone, abrasive flow, or electropolish depending on geometry. The rule for this lesson is shorter than the EDM handoff: do not nitride over recast.

Finish-EDM settings labeled "fine" or "skim" do not mean recast-free. They mean lower-energy recast. The difference between a 30 µm recast layer and a 10 µm recast layer is real, and the 10 µm layer still has microcracks the nitride cycle inherits. The verification is the cross-section, not the vendor's claim.