Industry News

Why do high-strength bolts snap without warning? Hydrogen embrittlement and baking essentials

2026-09-20

Compiled by JIALI Industrial (164.hk) — a Hong Kong fastener supplier specialising in screws, bolts, nuts and selection know-how.

Delayed fracture days after installation: how hydrogen embrittlement works

A bolt that torques up perfectly, then snaps days or weeks later with a flat, almost plastic-free fracture face — that is classic hydrogen-embrittlement delayed failure. Atomic hydrogen enters the steel during acid pickling and electroplating, migrates to highly stressed zones such as the first engaged thread and the head-shank fillet, and initiates intergranular cracking under sustained tensile stress. The harder the fastener, the greater the sensitivity: parts above roughly 320 HV (10.9 grade and up) are high-risk, especially 12.9 bolts, spring washers and hardened self-tapping screws. Fracture needs three factors together — hydrogen, tensile stress and a susceptible microstructure — which is why only some bolts in a batch actually fail. The timing is what makes this failure mode so dangerous in practice. It rarely appears during assembly or even during the first load test; the joint passes inspection, the equipment goes into service, and the failure arrives anywhere from a few hours to several weeks later, often with no warning and no visible corrosion. A Hong Kong case is typical: a mould-clamping application using black-oxide 12.9 socket head cap screws saw three failures in a single week, all at the head-shank fillet, on a set of screws installed ten to fourteen days earlier. The batch passed dimensional inspection and hardness checks, and the only common factor was that the fasteners had been acid-pickled and electroplated before blackening with no baking step in between. Investigators also find that failure concentrates in the first engaged thread and under the head, because those are the locations where the stress concentration and the triaxial stress state are highest and where hydrogen collects. That is why a hydrogen-related crack almost always starts at a thread root or a fillet, travels along prior austenite grain boundaries, and leaves a brittle, faceted appearance under the microscope rather than the dimpled ductile surface of an overload failure.

Where hydrogen gets in, and how baking removes it

Hydrogen uptake concentrates in three steps: acid pickling, pre-plate activation and the plating bath itself, where hydrogen evolves on the cathode. The countermeasure is prompt post-plate baking: per ISO 4042 / GB/T 5267.1, hold at about 190-220 degC, from 4 hours up to 22 hours as hardness increases; for 12.9 electroplated bolts, no less than 20 hours, into the oven within 4 hours of plating. Where stronger corrosion protection is needed, zinc-flake coatings such as Dacromet and mechanical galvanising involve no electrolysis and introduce virtually no hydrogen — the reason Dacromet is the preferred finish for fasteners above grade 10.9. Never machine, straighten or re-heat-treat fasteners after plating, as this multiplies both hydrogen and residual-stress risk. As a working comparison across grades: a grade 8.8 bolt below 320 HV is low risk and ordinary clear zinc plating is acceptable without a special baking regime; a grade 10.9 bolt in the 320-380 HV band needs baking at 190-220 degC for about 4 hours as a minimum; a grade 12.9 bolt at 390 HV and above needs at least 20 hours, and some specifications call for 22-24 hours. The reason longer times are needed is not arbitrary: hydrogen diffusion rate falls sharply as hardness rises, so a hard, fine-grained, heavily cold-worked part holds hydrogen longer and releases it more slowly. Baking also has to be done at the right temperature — too low and the hydrogen does not move, too high and the bolt loses the temper that gives it its strength. Practical process control therefore matters as much as the specification: keep the interval between plating and baking short, load the oven so parts are not stacked densely enough to shield each other, and record the actual hold time and temperature rather than the set point. Suppliers who cite the bake specification without a record of the actual cycle cannot demonstrate compliance, which is exactly what a failure investigation will ask for. There is also a design-level remedy: where a joint permits it, substituting a zinc-flake coating for electroplating removes the hydrogen source entirely instead of relying on a downstream bake to remove it, and this is nearly always cheaper than the risk it eliminates.

A procurement and acceptance checklist

Put four things in writing at order stage: (1) grade and hardness ceiling, with mandatory baking for 10.9 and above; (2) the coating process — electroplated zinc must come with baking records, and zinc-flake Dacromet preferred for high-strength parts; (3) the test basis — batch hydrogen-embrittlement sensitivity per the ISO 15330 / GB/T 3098.17 parallel-bearing test, or ASTM F1940 process-validation data from the supplier; (4) storage away from acidic environments on site, and installation to a torque value without over-yielding the bolt. If a failure has already occurred, keep the fracture face for metallographic analysis to separate hydrogen embrittlement from fatigue before taking corrective action. The parallel-bearing test is worth understanding, because it is the check most often quoted and least often understood. A batch of fasteners is assembled into a hardened parallel-bearing plate at a specified torque and held for a set period, typically 24 to 48 hours; any bolt that cracks or fractures has retained enough hydrogen to fail under sustained load. Because the test is destructive and time-consuming, it is usually run on a sample basis as a process validation rather than on every batch. The alternative, ASTM F1940, approaches the same question from the process side: a control specimen with a known notch geometry is processed through the same plating and baking line, then loaded and inspected, so the result validates the process rather than just the parts. For a buyer, the practical consequence is to decide which evidence is acceptable before the order is placed, and for critical applications to require the supplier to maintain periodic process validation records with dates. On site, storage and installation discipline adds its own layer of protection: keep high-strength plated fasteners in dry, ventilated storage away from acid fumes and pickling areas, avoid leaving them exposed to humid air for months before use, and install to a controlled torque rather than over-tightening, because an over-yielded bolt starts out with less margin against any residual hydrogen. Where a joint must be disassembled and reassembled, use a fresh bolt rather than reusing one that has already been stretched.

Common pitfalls and acceptance points in hydrogen-embrittlement control

Pitfall one is assuming that a high grade is automatically safe because it passed the hardness check. Hardness correlates with susceptibility; it does not certify that baking was done or done properly. Always ask for the baking record with the batch number. Pitfall two is accepting a verbal assurance that Dacromet is used when the parts in the box are clearly electroplated — the two finishes look different under a low-power lens, and a quick abrasion or a coating-thickness check distinguishes a metallic zinc layer from an inorganic zinc-flake layer. Pitfall three is letting a subcontractor re-plate or touch up a batch without re-baking; a small cosmetic touch-up after a scratch is a common root cause, and the rework often runs through an acid activation step that reintroduces hydrogen. Pitfall four is ignoring the washers: hardened spring washers in the same joint are just as susceptible as the bolt, so specify the same coating and baking requirements for them. Pitfall five is a documentation trail with no dates. A baking record without a plating completion timestamp cannot show that baking started within the required interval. At acceptance, a practical checklist is: confirm the grade and hardness range against the certificate; confirm the coating type physically or by report; obtain the baking record with times and temperatures for 10.9 and above; obtain batch or periodic evidence of ISO 15330 or ASTM F1940 testing; check that the packaging is dry and undamaged; and record torque values used at installation so that any later failure can be assessed against the actual load. Where the application is safety-critical, such as lifting equipment, pressure retaining parts or rotating machinery, treat hydrogen-embrittlement control as a design requirement from the outset rather than a purchasing detail. Need baking-compliant 10.9 / 12.9 bolts or zinc-flake coated supply? Contact JIALI Industrial (164.hk) with drawings and grades for matching specifications and certificates.

Related Reading

• How should you manage fastener inventory? ABC classification, MOQ and safety stock essentials
• Why do bolts break below their rated strength? Fatigue life and stress concentration
• JIALI launches Stainless Steel zone with full 304 / 316 range

← Back to News