Step 4 of 5
63%
Failure modes — white-layer spall, network cracking, distortion, embrittlement
Five ways a nitrided tool fails in service, how to read each one from a microsection, and when 'the nitride failed' is honestly 'the wrong recipe was specified.'
Step 4 of 5Hydrogen embrittlement
The failure mode that vendors reach for when they do not want to own the failure. Real, but rarely the actual cause on tool steel at moderate hardness.
True hydrogen embrittlement requires three things together: hydrogen in the lattice, sustained tensile stress, and a hydrogen-susceptible microstructure. The susceptibility threshold for tool and die steels sits around 40 HRC bulk hardness, with risk climbing sharply above 45 HRC and severe risk above 50 HRC in thin sections or PM grades. Gas nitriding produces atomic hydrogen at the surface from NH₃ dissociation, so the hydrogen source is real.
What protects most nitrided tools is the cycle itself. A 40-80 hour soak at 480-560°C gives hydrogen extensive time to out-diffuse, and the diffusion zone's nitride precipitates trap whatever remains in non-mobile sites. The Surface and Coatings Technology study by Brass and Chêne on plasma nitriding and hydrogen-environment embrittlement of austenitic stainless found the nitrided layer actively reduced hydrogen ingress rather than promoted it. On a typical H13, P20, or Nitralloy die at 40-50 HRC bulk, hydrogen embrittlement is not a realistic in-service failure mode.
Where it can happen: very high-hardness PM tool steels (CPM-10V at 60+ HRC), thin high-strength sections, and certain stainless grades in sustained tension. For those cases a 200°C bake for 4-8 hours after nitriding mitigates real hydrogen retention.
When a vendor invokes hydrogen embrittlement on a 50 HRC die that delaminated, the actual cause is almost always white-layer spall (mode 1). Demand a microsection. Hydrogen-driven failures show transgranular or intergranular cracks initiating in the substrate under stress and propagating outward; spall shows cracks within the compound layer or at its interface, with the substrate untouched. The two patterns are distinguishable on a single cross-section, which is what the vendor should be asked to produce before any bake-out cycle is approved.
Quick check
A CPM-10V tool insert hardened to 62 HRC in a thin-walled (4 mm) section fails three weeks after nitriding with intergranular cracks originating in the substrate, away from any compound-layer feature. Is hydrogen embrittlement plausible here, and what is the mitigation?