Vendor selection and QC sign-off — separating real nitriding houses from job shops

Before approving production at any new vendor, run a first article and sign it off against the print.

Measure surface microhardness at the spec load (HV 0.05 for the compound layer, HV 0.5 for the diffusion zone, per ASTM E384). Run a microhardness traverse from the surface to core on a sectioned and etched specimen, at the indentation spacing required by the depth standard on the print, typically 25-50 µm spacing for nitriding hardness depth per DIN 50190-3 or ISO 18203. Read case depth off the traverse against the definition specified on the print. Measure compound layer thickness optically from the etched microsection at the location indicated on the drawing, and confirm phase by etch behavior or EDX if the spec calls out ε or γ'. Dimensionally check every toleranced feature on the part; nitriding grows the surface by 3-7 µm per side from compound-layer formation, and a tight bore can lose its fit.

Keep on file: the certificate with the cycle log, the microhardness traverse data, the photo of the etched microsection, the witness coupon if one was run, and the dimensional report from the first article. Keep the file for the life of the tool. When the part fails in service eighteen months later, that file is what tells you whether the failure traces to the case or to something downstream.

Argue about any number outside spec by more than the measurement uncertainty. Microhardness uncertainty at HV 0.05 is roughly ±25 HV; case depth measurement uncertainty per DIN 50190-3 is roughly ±10% on a traverse; compound layer thickness from an optical microsection is good to ±1 µm. A number inside that band is a measurement noise question, not a process question. A number outside it is a process question, and the vendor owes you a root-cause analysis on the load, not a re-run on the next part.