Hydrogen Embrittlement in High-Strength Bolts: 390 HV Is the Line
12 min·Yaxiio Engineering Alliance

Hydrogen Embrittlement in High-Strength Bolts: 390 HV Is the Line

Zinc-plated 10.9 and 12.9 bolts can crack after installation. The 390 HV trigger, 190-220 °C baking window, 24-200 h test and the PO clauses that prove it.

The Risk Matrix Comes First

A 10.9 bolt and a 12.9 bolt can leave the same zinc plating line with the same coating thickness and land in different risk classes. The number that decides it is not the property class on the head. It is hardness, and the line that changes the requirement sits at 390 HV.

GB/T 5267.1-2023 Table 2 sorts parts by hardness:

Surface hardness Class What the standard requires
360 HV or below A No supplemental process verification, no product test, no baking
Above 360 HV to 390 HV B Supplemental verification and/or product test, or baking — the manufacturer chooses
Above 390 HV C Verification and/or product test, and baking; temperature and time shall be specified

Read through the property classes and it becomes a grade rule. Below 10.9 is class A; 10.9 is B, where the manufacturer chooses; 12.9 is C, where baking is mandatory and temperature and time must be written down. Nuts up to class 12 are exempt because nut hardness stays below 390 HV, except class 10 and 12 all-metal prevailing torque lock nuts. Plain washers at 300 HV or below are class A; the 380 HV row is B.

For zinc-electroplated 10.9 bolts, the standard puts the batch in class B and leaves the decision with the manufacturer. Annex B.4 adds that a 10.9 fastener made with the correct process and mechanical properties is not prone to internal hydrogen embrittlement and does not need baking. Those statements agree only if the choice is deliberate: class B is a decision point, not a default pass.

GB/T 3098.1-2010 caps surface hardness at 390 HV for 10.9 and 435 HV for 12.9, so a conforming 10.9 part sits at the class C trigger and a batch measured near the top of its allowance is one reading from a different requirement. Argue about measured surface hardness, not the head marking. Our bolt grade strength chart lists class values; it does not replace a measurement on your parts.

What Hydrogen Embrittlement Actually Is

Internal hydrogen embrittlement is a failure mode of the steel, not a coating defect. Atomic hydrogen that entered during processing stays in the metal, and where the steel is hard and under tensile stress it can drive a crack that grows with little or no plastic deformation. The break looks brittle and arrives without the stretching a ductile overload would give.

That is why the standard gives it §4.4 and a full annex, Annex B. The consequence is a fastener that passes every dimensional and coating check and then fails in service. The annex note records that internal hydrogen embrittlement usually appears within hours of installation and generally in under 72 h. A part that sat in a warehouse for three months is not evidence of anything; the failure window opens when the joint is tightened.

Where the Hydrogen Comes From

In the coating route covered by GB/T 5267.1-2023, hydrogen enters during electroplating and the preparation around it. Acid cleaning and activation ahead of the bath, and the cathodic reaction in the bath itself, are where hydrogen is available to the steel. The remedy points the same way: baking is specified after electroplating and before the conversion coating, sealer or top coat.

The same requirements cover cadmium, zinc-nickel and zinc-iron systems, with a lower ceiling for cadmium. What changes is how much hydrogen is picked up and how easily it is removed, not whether the mechanism exists.

Two boundaries matter. These requirements are written around electroplated fasteners; hydrogen picked up in service from corrosion or cathodic protection is a different route and not what Table 2 answers. And coating thickness or salt-spray hours say nothing about hydrogen, because a thicker coating can mean longer in the bath. Do not let a corrosion report stand in for a dehydrogenation record.

The Three Conditions That Must Line Up

GB/T 5267.1-2023 §4.4 defines the risk as three conditions present at the same time:

  1. High tensile strength or high hardness, including a surface-hardened case on a softer core.
  2. Tensile stress, including residual tensile stress from cold working, straightening or welding.
  3. Atomic hydrogen in the steel.

The word that matters is “and”. Remove any one condition and the mechanism loses its footing. That gives three control levers, and a purchasing argument that separates them.

You cannot usually remove hardness, because the joint needs the property class. You can reduce residual stress by avoiding cold straightening or bending after heat treatment. You can remove hydrogen, within limits, by baking soon after plating. A supplier who answers a hydrogen question only with “we bake” has addressed one of three conditions and has not said what the batch hardness is or what stress the part carries.

Susceptible Grades and Finishes

Part and finish Class Practical reading
8.8 or 9.8, electroplated A No baking requirement from Table 2
10.9, electroplated B Manufacturer chooses verification or baking; a correctly made 10.9 is not prone
12.9, electroplated C Baking mandatory; temperature and time must be specified
Nuts up to class 12 Exempt Nut hardness stays below 390 HV; class 10 and 12 all-metal prevailing torque lock nuts excepted
Plain washers, 380 HV B Baking necessary for that special application

The finish side has one exemption buyers ask about more than any other. Alkaline zinc-nickel with 12 to 16 percent nickel carries a lower risk, and a decision not to bake should rest on testing and an agreement between purchaser and supplier. In aerospace this process is known as low hydrogen embrittlement, LHE. A different zinc-nickel route must support the exemption with its own test data.

The clause also exempts parts not designed to carry tensile stress, such as set screws, and induction-hardened tips such as those on thread-rolling screws. Both rest on design intent and, where relevant, testing and purchaser agreement.

Baking: What It Fixes and What It Does Not

Baking is the dehydrogenation step: the part is held at temperature long enough for hydrogen to leave the steel before it finds a stress concentration. Annex B.4 gives the working window, and the numbers are worth quoting exactly because so much shop practice rests on a shorter cycle.

Parameter Value in GB/T 5267.1-2023 Annex B.4
Typical bake, zinc-electroplated parts above 390 HV 190 °C to 220 °C for at least 8 h to 10 h
Adjustment May be shortened by type, size, hardness and coating system, or may need up to 24 h
The common short cycle 185 °C to 195 °C for 4 h is not sufficient and research indicates it may even be harmful
What to do instead Extend the time and/or raise the temperature, for example to at least 200 °C
Temperature ceiling Zinc-electroplated parts usually not above 220 °C; cadmium-plated parts not above 200 °C
Sequence After electroplating, before the conversion coating and/or sealer or top coat
Plating-to-bake interval As short as possible; not a batch acceptance criterion and never the basis for determining a failure cause

The 4 h row is the one to take to a supplier meeting. A 185–195 °C, 4 h cycle is common enough to appear on process cards as “dehydrogenation baking”, yet the standard says it is not enough and may do harm. Four hours at 190 °C does not meet the annex’s recommendation for a part above 390 HV.

The last row is often misused. A long gap between plating and baking is bad practice, so keep it short, but the standard refuses to turn the interval into a pass/fail number: it is not a batch acceptance criterion and cannot be used after the fact to explain a failure. Control it as a process requirement and audit the records.

Baking cannot repair a part that has already cracked. It is preventive, taken after plating and before the part is loaded, not a rework operation for failed hardware. It cannot compensate for a design that overloads a hard part, or for hydrogen picked up later in service. Nor does it replace the class C paperwork: temperature and time must be specified, so a certificate saying only “baked” does not close the requirement.

One administrative trap sits in the marking rules. Clause 9.2 says stress relief and/or baking is not included in the coating designation but specified separately. A callout such as Zn//Cn/T2 describes the coating system and says nothing about dehydrogenation. If the baking requirement lives only in a drawing note nobody copied onto the purchase order, the order does not carry it.

Design and Assembly Controls

Stress relief before plating is narrower than many specifications assume. Annex B.5 states that it is neither required nor suitable for quenched and tempered parts. It is needed only where an operation produced significant plastic deformation and residual tensile stress: cold forming, cold bending, cold straightening, drilling or welding. The annex also says the stress relief recommendations in GB/T 19349 are too broad for fasteners, a useful answer when a customer specification imports them wholesale.

Most high-strength bolts are quenched and tempered, so that question usually ends there. What remains is the stress the joint applies. Preload is the largest controlled tensile stress a bolt sees, and it is set by torque and by friction in the thread and under the head. Our bolt torque chart gives values by grade and condition; a joint designed at the top of the bolt’s capacity leaves less margin for any hydrogen that remains.

Two assembly habits add stress the design did not intend. Forcing a misaligned joint closed with the fastener bends the bolt and leaves residual tensile stress in the shank; reusing a bolt already tightened past yield changes the preload the next joint receives. Both feed the tensile-stress condition.

Testing and Acceptance

The standard’s test for internal hydrogen embrittlement is a sustained-load test: a static load is applied for a defined time and the part is observed for failure. Annex B.6 gives the duration as 24 h to 200 h depending on the specification, and names GB/T 3098.17, DIN 50969-2, NASM 1312-2 and ASTM F606/F606M as the methods in use.

Those names are not interchangeable, and duration is the first thing to read in a report. Because failures usually appear within hours of installation and generally under 72 h, a 24 h run covers the common window but is not the same test as a 200 h run. “Passed sustained load” without load level and duration is not a comparable result.

The second thing to read is what was tested. A report should identify the property class and measured hardness of the samples, the coating system, and whether the parts were tested after baking. A test on an unbaked sample says nothing about the shipped part.

The third is the sampling basis, and here the coating standard sets a trap. Chapter 8 makes coating thickness, thread gauging and appearance mandatory for every batch, but puts hydrogen embrittlement verification under process control, not per batch. For ordinary bolts, a batch report without a hydrogen test is normal, not evidence the requirement was met. If you want lot-level evidence, it has to be written into the purchase order. Acceptance is simple: a fracture within the specified duration is a failure, and a report that stops short of it is incomplete.

What to Write Into the PO

Chapter 10 of GB/T 5267.1-2023 already lists what the purchaser is expected to provide. Five items carry most of the hydrogen risk.

PO item What to write Why it matters
1. Part definition Property class, base material, surface condition, measured hardness or tensile data Hardness selects the Table 2 class; the grade can hide a surface-hardened case
2. Coating designation Full GB/T 5267.1-2023 designation, with thickness or minimum neutral salt spray hours Separates the corrosion requirement from the hydrogen requirement
3. Baking A separate clause: temperature, time, sequence after plating and before sealing, record per batch Clause 9.2 keeps baking out of the designation, so it must be stated separately
4. Hydrogen test Method, static load, duration within 24 h to 200 h, report content Makes results comparable and stops a short test being presented as a pass
5. Pre-plate stress relief Only where cold forming, bending, straightening, drilling or welding left residual tensile stress Annex B.5 rules out blanket stress relief on quenched and tempered parts

Add the test report requirement per GB/T 39310 if the project needs a document trail, and keep it with the batch records rather than the certificate file. A material certificate answers none of the five items.

When a 10.9 or 12.9 order reaches us, we treat the hydrogen question as a document check before it becomes a production question. We ask for measured surface hardness, not only the head marking; we look for baking as a separate line with a temperature, a time and a sequence; and we ask which test method and duration will be used. “All high-strength parts are baked” does not pass without temperature and time, and “class B, therefore exempt” does not pass either, because class B is a choice that has to be justified. If you are reviewing a batch now, send the grade, coating, measured hardness and existing bake and test records through our high-strength RFQ route, and we will tell you which of the five items is missing. The grade side is covered in 4.8 to 12.9 explained.

References

FAQ

Do all 10.9 bolts need baking?

No. Grade 10.9 is class B, where the manufacturer chooses between supplemental verification or product testing and baking. Annex B.4 adds that a correctly made 10.9 fastener is not prone to internal hydrogen embrittlement and does not need baking. Require the evidence behind whichever choice the supplier made.

Is a four-hour bake enough?

Not for a part above 390 HV. The standard says the common 185–195 °C, 4 h cycle is insufficient for dehydrogenation and that research indicates it may even be harmful. It points to extending the time or raising the temperature, for example to at least 200 °C, within the 220 °C ceiling for zinc-electroplated parts.

If you want the grade and coating combination checked against these clauses, start with the strength chart download or the torque chart download. How we source and state requirements like these is set out on our editorial policy page.

Key Standards

Standards and clauses referenced by this article.

  1. Fasteners - Electroplated coatings§4.4; Annex B (B.4, B.5, B.6); Table 2; Table 8; Chapter 8; 9.2; Chapter 10

    Primary source for the three IHE conditions, the hardness classes and the 390 HV trigger, the 190-220 °C / 8-10 h bake window and its limits, the 24-200 h sustained-load test, per-batch versus process-control testing, the marking rule that keeps baking out of the coating designation, and the ordering requirements.

  2. Mechanical properties of bolts, screws and studsTable 3, notes g-j

    Surface hardness limits: 10.9 not above 390 HV, 12.9 not above 435 HV (HV 0.3). Used to show that a conforming 10.9 part sits at the class C trigger.

  3. Fasteners - Preloading test for the detection of hydrogen embrittlement

    Sustained-load test method named in Annex B.6; no table values quoted.

  4. Prevention of hydrogen-induced brittle fracture in high-strength steel fasteners

    Process-control and test method named in Annex B.6; no table values quoted.

  5. Fastener test methods, method 2: tensile strength

    Test method named in Annex B.6; no table values quoted.

  6. Standard test methods for determining the mechanical properties of threaded fasteners

    Test method named in Annex B.6; method name only, no values quoted and no equivalence asserted.

  7. Fasteners - Test report requirements

    Report requirement referenced in Chapter 10 for orders that need a document trail.

References

Sources used for fact checking and background context.

  1. GB/T 5267.1-2023 - Fasteners: electroplated coatingsSAMR / SAC
  2. GB/T 3098.1-2010 - Mechanical properties of bolts, screws and studsSAMR / SAC
  3. ISO 4042:2022 - Fasteners: electroplated coating systemsISO

Deep Reading

More systematic selection, procurement, or inspection guides.

This page is one entry in the fastener engineering reference. The library covers standards, dimensions, materials, and failure analysis; every figure is traced back to its source.

Continue reading in the library →