What Is Hydrogen Embrittlement

Hydrogen embrittlement (HE) is a delayed fracture phenomenon. Atomic hydrogen diffuses into high-strength steel during acid cleaning, electroplating, or corrosion reactions. Once inside, the hydrogen atoms accumulate at internal defects, grain boundaries, and stress concentrations - particularly at the thread root where service stress is highest. Under sustained tensile load, the hydrogen causes the steel to crack and fracture with little or no plastic deformation. The fracture surface is typically flat and brittle-looking, unlike a ductile overload which shows necking.

The timing is what makes HE dangerous. A bolt may pass the incoming tensile test perfectly, get installed, and run fine for days or weeks. Then, without warning, it fractures under normal service load. By the time the failure is noticed, the bolt has already broken and the joint may have come apart. There is no gradual warning - it is a time-delayed brittle fracture triggered by the combination of tensile stress, hydrogen, and a susceptible material.

Which Processes and Parts Are Most At Risk

Not all fasteners are equally at risk. Three factors combine to make a part susceptible: high hardness (above 320 HV, roughly 10.9 class and above), a tensile stress at or above 60-80% of yield, and a hydrogen source. The hardest high-strength bolts (12.9 class, hardness 380-430 HV) are the most vulnerable. Bolts up to 8.8 class (HV below 320) carry only minimal hydrogen risk, and low-carbon 4.6 class bolts are essentially immune.

The main hydrogen sources are: acid pickling before plating, electrolytic plating (zinc, zinc-nick, cadmium), phosphating under harsh conditions, and even atmospheric corrosion of uncoated high-strength steel. Any process that exposes a high-strength steel to acid or electrolytic action introduces hydrogen. This is why the standard mitigation - bake-out immediately after plating - is time-critical: the sooner the hydrogen is baked out, the less it has time to diffuse into the steel lattice.

How to Test: GB/T 3098.17 Preload Method

GB/T 3098.17-2000 (identical to ISO 15330) specifies a parallel-bearing-surface preload test. The test bolt is loaded to a specified preload (typically 60-80% of the bolt's specified breaking load) in a test fixture that holds it in tension. The bolt must not fracture within the specified test duration. The standard requires the test to start within 24 hours of plating (to catch the hydrogen before it dissipates) and continue for 48 hours. If the bolt fractures during the test, the lot fails.

This test deliberately does not use a tensile test to failure, because a simple pull test may not trigger the delayed fracture. The preload method holds the bolt just below yield for an extended period, mimicking service conditions and giving the hydrogen time to cause a crack. A batch that passes 48 hours under preload is considered resistant to hydrogen embrittlement under the test conditions.

How to Prevent: From Material Selection to Bake-Out

Prevention starts at the material selection stage. If the bolt is 10.9 or 12.9 class, choose a non-electrolytic coating such as dacromet or mechanical zinc plating rather than electrolytic zinc. If electrolytic plating is unavoidable, specify a bake-out within 2 hours of plating at 190-220 C for 4-8 hours, and require a GB/T 3098.17 test report. The bake-out drives hydrogen out of the steel before it can accumulate at crack-initiation sites.

Other preventive measures include: avoiding over-torque (which raises the tensile stress that drives HE), using materials with lower hardness where possible, and ensuring that storage after plating does not re-introduce hydrogen through corrosion. For safety-critical applications such as automotive chassis and aircraft, many customers mandate non-electrolytic coatings on all high-strength bolts and do not accept electrolytic plating at all.

On-Site Assessment Checklist

If you are investigating a suspected HE failure, check these points: (1) Is the bolt 10.9 class or above (hardness above 320 HV)? (2) Was it electroplated, and was a documented bake-out performed? (3) Did the fracture occur hours to weeks after installation rather than immediately? (4) Does the fracture surface look flat and brittle with no necking? If all four answers are yes, hydrogen embrittlement is the likely cause. The fix is not to replace the broken bolt with a stronger one - it is to switch to a non-electrolytic coating or a lower-hardness material.

What to do if you suspect a hydrogen failure. If a plated high-strength bolt fractures unexpectedly, immediately quarantine the remaining lot. Do not simply replace the broken bolt with another from the same batch. Send a sample to a lab for fractography: a brittle intergranular fracture surface is the signature of hydrogen embrittlement. Review the plating records to confirm whether bake-out was performed and whether it was timed correctly. Switch to a non-electrolytic coating (dacromet, mechanical plating) for future lots, and require a GB/T 3098.17 test report. Treat every unexplained brittle fracture on a plated high-strength bolt as a hydrogen embrittlement investigation until proven otherwise.

Designing out hydrogen risk. The best way to avoid hydrogen embrittlement is to not introduce hydrogen in the first place. Specify non-electrolytic coatings on high-strength bolts from the drawing stage. If electrolytic plating is unavoidable, choose a lower hardness class where the material is less susceptible (below 320 HV is generally safe). Avoid acid cleaning of high-strength parts where possible. After plating, enforce the bake-out window as a hard process control, not a suggestion. And always test production lots per GB/T 3098.17: a 48-hour preload test catches delayed fractures that would otherwise appear in the field. The few hundred dollars per lot for hydrogen testing is cheap insurance against a field failure that costs thousands.

Material and design choices that reduce HE risk. Beyond process controls, material selection affects hydrogen susceptibility. Lower-strength steels (below 10.9 class) are much less susceptible. If the design allows, use a lower hardness bolt rather than pushing to 12.9. Add a radius at the thread root to reduce stress concentration. Avoid sharp notches and head-to-shank transitions that concentrate stress. And always specify a bake-out on the drawing as a hard requirement, not a note. A drawing callout that says "bake per GB/T 3098.17 within 2 hours of plating" is enforceable; a note that says "consider baking" is not.

What to do if you suspect a hydrogen failure. If a plated high-strength bolt fractures unexpectedly, immediately quarantine the remaining lot. Do not simply replace the broken bolt with another from the same batch. Send a sample to a lab for fractography: a brittle intergranular fracture surface is the signature of hydrogen embrittlement. Review the plating records to confirm whether bake-out was performed and whether it was timed correctly. Switch to a non-electrolytic coating (dacromet, mechanical plating) for future lots, and require a GB/T 3098.17 test report. Treat every unexplained brittle fracture on a plated high-strength bolt as a hydrogen embrittlement investigation until proven otherwise.

Designing out hydrogen risk. The best way to avoid hydrogen embrittlement is to not introduce hydrogen in the first place. Specify non-electrolytic coatings on high-strength bolts from the drawing stage. If electrolytic plating is unavoidable, choose a lower hardness class where the material is less susceptible (below 320 HV is generally safe). Avoid acid cleaning of high-strength parts where possible. After plating, enforce the bake-out window as a hard process control, not a suggestion. And always test production lots per GB/T 3098.17: a 48-hour preload test catches delayed fractures that would otherwise appear in the field. The few hundred dollars per lot for hydrogen testing is cheap insurance against a field failure that costs thousands.

Material and design choices that reduce HE risk. Beyond process controls, material selection affects hydrogen susceptibility. Lower-strength steels (below 10.9 class) are much less susceptible. If the design allows, use a lower hardness bolt rather than pushing to 12.9. Add a radius at the thread root to reduce stress concentration. Avoid sharp notches and head-to-shank transitions that concentrate stress. And always specify a bake-out on the drawing as a hard requirement, not a note. A drawing callout that says "bake per GB/T 3098.17 within 2 hours of plating" is enforceable; a note that says "consider baking" is not.