Hydrogen Embrittlement in High-Strength Bolts: Which Grades Need Controls, Which Coatings and Baking Schedules Are Acceptable, and What to Require Before the RFQ

Does your grade sit above the hardness threshold?

Start by converting the specified property class into a hardness range, because hardness is the first-order driver of susceptibility. As strength rises the steel becomes harder, less ductile and less tough, and it absorbs hydrogen more readily ¹. Above a specified hardness of 39 HRC (380 HV) susceptibility increases significantly. Below it, fasteners normally show no significant susceptibility, provided they were made by well-controlled processes from steel of adequate quality ¹. That threshold is not a fixed line. Second-order effects of chemistry, tempering temperature and sub-microstructure move it by roughly ±1 HRC (±10 HV), and some standards set lower critical limits of 31-35 HRC ¹. Apply that to the grades you are likely to specify. ISO 898-1 class 12.9 is specified at 39-44 HRC (385-435 HV), and the standard carries a stress-corrosion caution for it ⁵. Class 14.9, at 1400 MPa minimum and 44-49 HRC, sits at the top of the risk range and falls outside the ISO 898-1 scope altogether ⁹. Class 10.9 and SAE J429 Grade 8 are the examples given of fasteners normally without significant susceptibility, and the 25-39 HRC (240-380 HV) band is described as not normally embrittled by galvanizing ^{1, 6}. So the first decision is binary: if your specified hardness range reaches or crosses roughly 39 HRC, treat the part as HE-susceptible and carry the coating, baking, testing and installation controls through the rest of this sequence. If it sits wholly below, the controls are usually unnecessary — the sources note that Grade 8 and class 10.9 parts are often baked when they do not need to be ³. Confirm the hardness range on the drawing rather than inferring it from the class number alone.

Which coating processes introduce hydrogen, and which avoid it?

The hydrogen that causes delayed fracture enters during surface preparation and coating, not during service. Electroplating produces hydrogen at the steel surface, and it can penetrate into hardened high-strength parts ¹⁸. That makes zinc, zinc-nickel and cadmium electroplating HE-capable processes that require post-plating baking ^{16, 21}. Acid pickling is the other route: for spring steels it must be avoided entirely, with mechanical cleaning (shot blasting, tumble finishing) used instead ¹⁰. The non-electrolytic alternatives do not introduce hydrogen. Zinc flake is applied by dipping, spinning and curing rather than electroplating, and is stated to carry no hydrogen embrittlement risk from the process ¹⁷. Mechanical galvanizing, organic coatings and phosphate-and-oil are listed alongside it as preferred routes where electroplating is not essential ¹⁰. Phosphate coatings are very porous, and that porosity is given as the reason the IHE risk is described as non-existent for them ⁷. Two nuances matter when you write the callout. First, not all electroplated coatings behave alike: certain zinc-nickel processes containing 12-16% nickel are stated to carry significantly lower IHE risk than zinc or zinc-iron because they are more permeable. That lets hydrogen escape more readily ⁷. Second, hot-dip galvanizing is a special case — the process itself does not introduce hydrogen, but it can release trapped residual hydrogen through thermal up-quenching. That is why galvanized high-strength fasteners have failed without any acid contact ². Zinc electroplating is also flagged as possibly inappropriate for case-hardened screws at high hardness ². For a susceptible grade, the practical rule is to specify a non-electrolytic coating unless the application forces electroplating, and to name the specific process rather than the finish family. A zinc flake coating is the common non-electrolytic choice for high-strength bolts in automotive, structural, rail and off-highway work ¹⁶.

What baking schedule should the PO specify?

Baking is a low-temperature heat treatment after coating that either extracts hydrogen by effusion or moves it to trap sites where it cannot migrate. As free hydrogen falls, both time-to-failure and threshold stress rise ³. Three variables govern how well it works: temperature, time, and the permeability of the coating ³. Because those vary, the sources give a family of schedules rather than one number, and the schedule you write should match the grade, the coating and the application. | Condition | Temperature | Duration | Plating-to-bake window | |---|---|---|---| | Zinc electroplated, above 39 HRC | 190-220 C | 8-10 h minimum, up to 24 h by size and strength | Keep short ³ | | Zinc electroplated | 200-220 C | Not stated | Keep short ⁴ | | Cadmium electroplated | 190-200 C | Not stated | Keep short ⁴ | | Electroplated Grade 10 nuts | 190-230 C | 8 h minimum | Within 4 h ¹¹ | | Electroplated Grade 12 nuts | 190-230 C | 8-24 h minimum | Within 4 h ¹¹ | | Springs, if electroplating is unavoidable | 190-230 C | 24 h minimum | Within 1 h ¹⁰ | | Aerospace, above 44 HRC | 190-210 C | 23-24 h minimum | Within 1 h ¹⁴ | The pattern is that higher strength and thinner sections push you toward the long end of the range and the short end of the window. Springs illustrate why: the fine wire gives hydrogen a short diffusion path, so it reaches critical concentration faster than in a large-diameter bolt ¹⁰. A 4-hour bake at roughly 190 C is the schedule to reject. For zinc-electroplated parts above 39 HRC it is described as insufficient, leading to occasional failures, and possibly even detrimental because zinc is an effective barrier to hydrogen diffusion ³. Note that the same source treats the plating-to-bake window differently from the bake duration: it calls a fixed 4-hour window subjective and a quality-assurance mechanism rather than a rigid acceptance criterion. It also warns against using it to assign root cause to a failure ⁴. The window is a process control; the duration is the de-embrittlement step. Ask your plater to validate the schedule they propose with sustained-load test data on production parts rather than quoting a generic table ⁴.

Which HE test and acceptance criterion do you require?

The acceptance test for a susceptible lot is sustained-load testing to ASTM F519 or BS ISO 16573-1: no specimen may fracture at 75% of notched fracture strength held for a minimum of 200 hours. Any fracture counts as failure ¹⁹. Under ASTM F519, if a single specimen fractures the load is increased and held a further hour; multiple fractures mean the plating process itself must be re-evaluated ¹⁹. That last clause is the one to put in the PO, because it forces the plater to fix the process rather than re-test until a lot passes. Process qualification is a separate requirement from lot acceptance. ASTM F1940 uses an embrittlement ratio — hydrogen-charged threshold stress divided by uncharged baseline strength — with acceptance above 0.60 for general applications and above 0.75 for critical ones ¹⁹. Other methods exist (preloading per BS EN ISO 15330, slow strain rate per BS ISO 16573-2/ASTM G129). But sustained load is the one that directly simulates service stress and is used for qualification and lot acceptance ¹⁹. Do not accept analytical methods as production evidence. ASTM G129, ISO 7539-7 and ASTM F1624 are described as unsuitable for production-scale embrittlement testing because of the time and cost involved, and are normally used for research and development ³. If a supplier offers an F1624 result as the acceptance evidence for a production lot, that is a mismatch with the intended use. For aerospace work above 44 HRC the bar is higher again: 100% sustained-load testing per ASTM F519 plus full documentation traceability ¹⁴. For general industrial lots, the sources describe the test methods and criteria but not sampling frequencies. So the number of specimens per lot is something you have to agree with the supplier rather than read off a standard.

Installation and post-installation checks

Tightening is what supplies the sustained tensile stress that drives hydrogen migration, and time to failure shortens as that stress rises ⁵. The checks below keep the installed stress and the monitoring period under control.

• Confirm the tightening method suits the product: impact wrenches may not be appropriate for some products and applications ².
• Verify the torque or preload procedure is qualified for the actual bolt, nut, washer, coating and lubricant combination, since changing the finish changes thread and bearing friction ¹².
• Treat the 24 hours to one week after installation as the critical monitoring window for internal hydrogen embrittlement failure ²⁰.
• Control the hydrogen-producing processes upstream: follow the applicable coating or product standard, apply the required relief treatment promptly, and qualify a non-electrolytic coating where appropriate ¹².
• Where the part is cold formed, cold bent, cold straightened, drilled or welded before plating, specify stress relief before electroplating ⁶.

Screen the material and heat-treatment condition

Coating and baking controls cannot compensate for a bad lot of steel. Improper quenching and tempering is very often linked to hydrogen embrittlement failures, and its consequences include higher-than-expected hardness, unintended carburization and incomplete martensite transformation ⁵. Non-homogeneity of the metallurgical structure from poorly controlled heat treatment, impurities and non-metallic inclusions can dramatically increase susceptibility in ways that a correct coating schedule will not fix ¹. The San Francisco-Oakland Bay Bridge anchor rod failures are cited as an example where the root cause was the material condition of one lot rather than the galvanizing process ⁶. Residual tensile stress is the second material-side factor. Work hardening before electroplating can leave residual tensile stress that exceeds the HE threshold and initiates hydrogen-induced microcracks after plating ⁶. Where that applies — cold forming, cold bending, cold straightening, some drilling and welding — stress relief before electroplating is beneficial ⁶. It is not necessary or desirable for fasteners that are thread rolled after heat treatment, or where compressive residual stresses are intentionally introduced. Standard secondary machining such as grinding, turning, tapping and milling is not normally problematic ⁶. On the RFQ, this translates into two asks: the heat-treatment condition and hardness range of the delivered lot, and whether any cold-work operation precedes plating. If the answer to the second is yes, the stress-relief step belongs in the process sequence, not in a note.

Cost and order-size implications of HE-safe routes

HE-safe coatings are not cost-neutral, and the premium is large enough to affect which route you can justify. One manufacturer reports zinc flake at about twelve times the cost of phosphate-and-oil at their volumes, while noting that their zinc flake volume is much lower, which itself raises the price ¹⁷. That figure is a single manufacturer's observation at their own volumes, not a market ratio, so treat it as an order-of-magnitude signal rather than a quotation benchmark. Hot-dip galvanizing carries its own premium when the enhanced specification is required. ASTM F2329 is reported at a 10-25% cost premium over ASTM A153. It also demands facility capability that not every galvanizer has: a water quench tank, post-treatment processing, temperature monitoring equipment, trained quality control personnel and documentation systems ⁸. The same source advises specifying F2329 only where critical structural applications, high-strength fasteners or rigorous quality documentation genuinely justify it ⁸. On order size, one manufacturer states a typical minimum order quantity of 1000 pieces, with exceptions for extra-large sizes, special materials and special purposes ²². That is one supplier's stated policy, not an industry floor, but it is the kind of figure worth confirming early if your HE-safe coating choice narrows the supplier pool. The sources do not establish how baking or HE-safe coatings affect lead time, so ask for the plating and baking steps in the quoted lead time rather than assuming they are absorbed.

What certification and traceability to demand

Certification requirements scale with the sector, and the sources describe two very different levels. Aerospace work above 44 HRC is stated to require full documentation traceability, with Nadcap accreditation and the AMS 2759/9 (hydrogen embrittlement relief baking) and AMS 2750 (pyrometry) standards cited alongside it ¹⁴. At the other end, one manufacturer states ISO 9001 certification and ROHS/REACH compliance as its credentials ²². These are different sectors and different schemes, so they cannot be read as a ladder — an ISO 9001 certificate does not demonstrate the furnace control that AMS 2750 covers. For your own PO, the useful move is to name the specific evidence you need rather than a scheme name. If the part is susceptible and electroplated, the baking furnace records — loading method, time in furnace and temperature uniformity — are what the plater is expected to control closely ⁴. Pair those with the sustained-load test report for the lot. Whether IATF 16949 or Nadcap is required for general industrial high-strength fasteners, and whether lot traceability tying plating and baking records to delivered bolts is a general requirement, is not settled by these sources. If your sector demands it, that is a customer or regulatory requirement you impose, not one the fastener standards establish.

The two disagreements to resolve before you accept a schedule or finish

Two items in this sequence are genuinely disputed across the sources. Do not treat either position as settled without the supplier's own test data behind it.

Which coatings trap hydrogen, which let it escape, and how should the bake be sequenced?

The coating you specify determines both how much hydrogen enters the steel and whether a post-plate bake can actually remove it, which is why two electroplated finishes with the same corrosion rating can carry very different embrittlement risk.

During zinc, nickel, chromium or cadmium electroplating the fastener acts as the cathode, and the hydrogen-evolution side reaction generates atomic hydrogen directly at the surface, where it is absorbed into the steel before it can recombine into gas (trojansf.co.uk). Once the deposit is on, its permeability governs relief baking: matte zinc and zinc-nickel are more permeable and let hydrogen diffuse out, while chromium and nickel deposits are highly impermeable, making subsequent baking less effective (trojansf.co.uk). Cadmium charges hydrogen by the same cathodic mechanism, but its deposit is more hydrogen-permeable than zinc, so baking removes hydrogen more efficiently, and it holds corrosion protection to roughly 230 degrees Celsius (trojansf.co.uk). Hard chrome is the worst case: one plating supplier reports its extremely high current densities generate substantially more hydrogen than typical plating processes (plateco.net). Tin adds a constraint of its own — standard bake temperatures of 190-230 degrees Celsius degrade the deposit, forcing lower temperatures and longer durations (trojansf.co.uk).

Several process-sequence decisions matter as much as the bake schedule itself. Mechanical cleaning — shot blasting, wire brushing or tumble finishing — removes mill scale with no chemical reaction and therefore generates zero hydrogen, and shot blasting additionally leaves compressive residual stress that resists hydrogen-assisted crack initiation (trojansf.co.uk, citing BS ISO 9587:2007). Sequencing also has a trap: per ASTM F1941 the relief bake must be completed before passivation, because baking after passivating degrades the conversion coating and cuts its salt-spray performance (plateco.net). Stripping and re-plating a susceptible fastener stacks a second full hydrogen cycle on the first, so best practice is to scrap the parts; if re-plating is unavoidable, use alkaline strippers, bake after stripping, and bake again at the maximum permitted 24 hours after re-plating (trojansf.co.uk). For threads rolled after heat treatment, one source advises starting the bake within 1 hour rather than 4, because the cold-worked region absorbs hydrogen faster (trojansf.co.uk). Suppliers should document the actual elapsed plating-to-bake time, since a technically correct bake performed too late may not fully mitigate hydrogen that has already concentrated at stress points (plateco.net).

The historical record shows how much these requirements have tightened. Most plating specifications once required only 3 hours at 375 ± 25 °F within 4 hours after plating; that was found inadequate and revised in 1981-82 to not less than 23 hours at 375 ± 25 °F within 2 hours, per MIL-N-25027D (engineeringlibrary.org, referencing NASA fastener design criteria). Buyers should also read 'low-hydrogen' claims critically: one supplier notes that low-embrittlement zinc-nickel formulations still require qualification testing and baking to confirm the risk is controlled, not eliminated by process design (plateco.net). For fasteners at or above 1000 MPa tensile strength or 320 HV (~32 HRC), one coating source states zinc flake is often the only acceptable coating option, needs no post-coating bake, and is validated with ISO 15330 embrittlement testing (trojansf.co.uk). When a failure does occur, SEM fractography showing intergranular facets supports a hydrogen diagnosis, but EDS cannot detect hydrogen directly because it is too light for standard detectors (trojansf.co.uk).

The corpus segments do not quantify hydrogen generation rates per bath chemistry, and they do not address hot-dip galvanizing trade-offs for high-strength fasteners in this candidate set, so that comparison cannot be sourced here.

FAQ

Related pages

• Surface finishes product page
• Bolts product page
• Zinc Plating for Fasteners: When It Fits, When Hardness or Exposure Rules It Out
• Fastener Failure Modes: Which Fatigue, Creep and SCC Controls the Evidence Supports
• What Bolt Property Classes 8.8, 10.9 and 12.9 Encode — and When High Strength Stops Being the Right Choice