Hydrogen Cracking in Welds: Causes, Prevention and Best Practices
Hydrogen cracking — also known as hydrogen-assisted cold cracking (HACC), underbead cracking, or delayed cracking — is one of the most insidious and potentially catastrophic weld defects that can occur in carbon and low-alloy steels. Unlike hot cracking, which is visible immediately upon solidification, hydrogen cracking may not appear until hours or even days after welding is complete. This delayed nature makes it especially dangerous: welds can pass initial inspection, be released for service, and then develop cracks that lead to structural failure under in-service loading.
The mechanism behind hydrogen cracking is well understood, and the defect is entirely preventable — but only when the welding engineer and inspector both understand the three conditions that must coexist for cracking to occur. This article covers the complete technical picture: the three-condition model, hydrogen sources and diffusion behaviour, susceptible microstructures, quantitative preheat assessment, low-hydrogen consumable classifications, post-weld heat treatment, and the inspection practices required to detect hydrogen cracks reliably. The article is part of the WeldFabWorld Welding Metallurgy Series.
Anyone working with hardenable steels — from structural fabrication and pressure vessel manufacture to pipeline welding and offshore construction — needs to understand hydrogen cracking prevention as a non-negotiable element of welding procedure design. Combined with proper carbon equivalent assessment and understanding of HAZ microstructure, the guidance in this article provides a complete engineering framework for eliminating HACC from fabrication.
The Three Conditions Required for Hydrogen Cracking
Hydrogen cracking is not a single-cause phenomenon. It requires the simultaneous coexistence of three distinct conditions. This three-condition model — first systematically described by Granjon and subsequently codified in welding standards worldwide — is the most practically useful framework for understanding, predicting, and preventing HACC.
If any one of these three conditions is absent, hydrogen cracking will not occur regardless of the severity of the other two. This means that prevention strategies do not need to eliminate all three conditions simultaneously — reducing any single factor sufficiently will break the cracking triangle. In practice, the most robust approach targets at least two conditions: eliminating the hydrogen source and reducing microstructural susceptibility through preheat.
Hydrogen in Welding: Sources, Entry, and Diffusion
The hydrogen involved in cold cracking is diffusible atomic hydrogen — individual hydrogen ions (H+) or atoms, not molecular hydrogen (H2). The distinction matters because atomic hydrogen is extraordinarily small, approximately 0.53 Å in atomic radius, and can move through the interstitial spaces of both the FCC austenite and BCC/BCT ferrite lattice by a diffusion mechanism. Molecular hydrogen, once formed, is too large to diffuse rapidly through the lattice and tends to become permanently trapped.
Primary Hydrogen Sources
Under the intense heat of the welding arc (temperatures in excess of 6,000°C at the arc core), water molecules dissociate completely:
Nascent hydrogen atoms (H•) dissolve directly into the molten weld pool
Organic compounds: CₙHₘ + arc energy → nC + mH•
Oil, grease, cellulosic coatings, paint — all release hydrogen under arc heat
The primary sources of moisture and hydrogen-bearing compounds that enter the weld zone include:
- Electrode coatings — particularly cellulosic electrodes (E6010, E6011) designed to contain significant moisture for shielding gas generation, and basic low-hydrogen electrodes that have been improperly stored and absorbed atmospheric moisture
- Base metal surface contamination — oil, grease, drawing compounds, rust (which contains hydroxyl groups), mill scale, paint, and any organic coating within approximately 25 mm of the joint
- Flux in SAW and FCAW — submerged arc flux and flux-cored wire flux can absorb moisture during storage if not properly sealed and heated
- Shielding gas moisture — GMAW and GTAW processes can introduce hydrogen if the shielding gas contains significant water vapour (dew point too high)
- Condensation on cold base metal — in cold ambient conditions, moisture condenses onto the base metal surface and must be removed by preheating before welding
Hydrogen Solubility and Diffusion in Steel
Hydrogen solubility in liquid steel is significantly higher than in solid steel — meaning that as the weld pool solidifies, it becomes supersaturated with hydrogen. Some hydrogen escapes as bubbles during solidification, but a significant amount is retained in the solidified weld metal and HAZ as diffusible atomic hydrogen.
The diffusion coefficient of hydrogen in ferritic (BCC) steel is relatively high at elevated temperatures — approximately 10-4 mm2/s at 200°C — which means hydrogen can migrate over significant distances in the hours following welding. In austenitic (FCC) steel, diffusion is much slower (approximately 10-7 mm2/s at 200°C), which is one reason austenitic stainless steels and nickel alloys are not susceptible to HACC.
As the weld cools, hydrogen preferentially migrates to regions of triaxial tensile stress — such as the HAZ beneath a weld bead — and to microstructural traps such as grain boundaries, dislocation tangles, and inclusion interfaces. When hydrogen accumulates faster than it can diffuse out, the local concentration builds until it is sufficient to initiate cracking.
| Hydrogen Source | Typical Diffusible H Content (ml/100g) | Risk Level | Mitigation |
|---|---|---|---|
| Cellulosic electrodes (E6010, E6011) — as supplied | 30–60+ | Very High | Use only where specifically required; accept risk with compensating preheat |
| Rutile electrodes (E6012, E6013) — as supplied | 15–30 | High | Avoid on hardenable steels CE >0.40; use low-H alternatives |
| Basic (low-H) electrodes — improperly stored | 8–20 | Moderate-High | Re-dry per manufacturer; maintain electrode oven |
| Basic (low-H) E7018 — properly stored/dried (H8) | <8 | Low | Standard for structural and pressure vessel work |
| Basic (low-H) electrode — H4 designation | <4 | Very Low | Specify for high-CE steels, high-restraint joints, critical applications |
| GMAW/FCAW with dry shielding gas, clean wire | 2–8 | Low | Verify dew point of gas supply; keep wire clean and dry |
| GTAW — clean filler wire, dry argon | <2 | Very Low | Inherently low; verify base metal cleanliness |
| Oil/grease contamination on base metal | Equivalent to 10–30+ | Very High | Degrease and clean all surfaces within 25 mm of joint |
Table 1 — Approximate diffusible hydrogen contributions from common welding sources. Values are indicative; actual levels depend on specific product formulation, storage conditions, and welding parameters.
The Susceptible Microstructure: Martensite and HAZ Hardness
Of the three conditions required for hydrogen cracking, the susceptible microstructure is the one most directly controlled by the welding procedure. As described in the companion article on martensite, bainite, and pearlite formation, the HAZ microstructure is determined by the combination of steel composition and HAZ cooling rate.
Martensite is by far the most susceptible microstructure because:
- Its BCT lattice contains a high density of dislocations and internal stress from trapped carbon — providing abundant trapping sites for hydrogen
- Its very high strength and low ductility mean that it cannot deform plastically to relieve the internal pressure generated by molecular hydrogen recombination at trapping sites
- The combination of high hardness (often 500–900 HV in as-quenched medium-carbon martensite) and high residual stress makes crack propagation easy once a microcrack initiates
The relationship between steel hardenability, HAZ hardness, and HACC risk is captured quantitatively by the carbon equivalent (CE). The IIW formula is the most widely used:
All elements in wt%. Source: IIW Doc. IX-535-67
CE < 0.35: Generally no preheat required — low HACC risk
0.35 ≤ CE < 0.45: Low-moderate risk — preheat 50-100°C typically required
CE ≥ 0.45: Moderate-high risk — preheat 100-250°C required; use H4/H8 consumables
CE ≥ 0.65: High risk — high preheat, H4 consumables, hydrogen bake mandatory
For steels where the Mn content is significant relative to carbon — as in many HSLA and microalloyed steels — the Pcm (parameter for cracking susceptibility of medium carbon steel) formula, developed by Ito and Bessyo, provides a more accurate assessment:
Better suited to steels with C < 0.18 wt% and multiple alloying elements
Pcm < 0.20: Generally safe without preheat for normal restraint
Pcm ≥ 0.25: Preheat required; magnitude depends on restraint and hydrogen level
Use the WeldFabWorld Carbon Equivalent Calculator to compute both IIW CE and Pcm values from your material test certificate composition.
Location of Hydrogen Cracks in the HAZ
Hydrogen cracks are not uniformly distributed through the HAZ. They occur preferentially at locations where the combination of maximum hydrogen concentration, maximum hardness (martensite content), and maximum tensile stress coincides. Three characteristic locations are recognised:
- Underbead cracking — occurs in the coarse-grained HAZ (CGHAZ) immediately adjacent to the fusion line, typically running parallel to it. This is the most common form, as the CGHAZ has the highest martensite content and the highest hydrogen trapping efficiency due to large prior austenite grain size.
- Toe cracking — initiates at the weld toe, where the geometric stress concentration of the weld profile combines with HAZ martensite and hydrogen concentration. Common in fillet welds and in joints with high restraint.
- Root cracking — occurs in the root pass HAZ, particularly in single-sided groove welds where the root region experiences high restraint and the root pass has limited heat input. Particularly problematic in pipeline root pass welding with cellulosic electrodes.
Prevention Strategy 1: Eliminate or Reduce Hydrogen Sources
Hydrogen source control is the first and most direct prevention measure. The goal is to minimise the quantity of diffusible hydrogen that enters the weld pool, because downstream measures (preheat, PWHT) become progressively less effective as hydrogen level increases.
Low-Hydrogen Consumable Selection
AWS A5.1 (SMAW electrodes for carbon steel) classifies electrodes by their maximum diffusible hydrogen content using an optional suffix designation:
| AWS H-Designation | Max. Diffusible H (ml/100g weld metal) | ISO Equivalent | Typical Application |
|---|---|---|---|
| H4 | ≤4 | ISO H5 | Critical — High CE steels, high restraint, offshore, pressure vessels |
| H8 | ≤8 | ISO H10 | Standard — General structural and pressure vessel fabrication on hardenable steels |
| H16 | ≤16 | ISO H15 | Limited use — Low-CE plain carbon steels with low restraint only |
| No H designation | Not controlled | — | Avoid — Not suitable for hardenable steels |
Table 2 — AWS A5.1 hydrogen designation suffixes for SMAW electrodes. The H-designator follows the electrode classification, e.g., E7018-H8 or E7018-H4.
Electrode Storage and Handling
Specifying a low-hydrogen electrode is only effective if the electrode arrives at the arc with low moisture content. Basic low-hydrogen electrodes are extremely hygroscopic — they absorb moisture from the atmosphere within hours of exposure. Proper storage and handling are therefore mandatory:
- Sealed containers: Electrodes should remain in their factory-sealed hermetic containers until immediately before use. Opened containers should be returned to an electrode oven.
- Electrode holding ovens: Opened electrodes should be kept in heated ovens at temperatures specified by the manufacturer — typically 66°C to 149°C (150°F to 300°F) for basic electrodes.
- Re-drying (reconditioning): Electrodes exposed to ambient conditions for more than the manufacturer’s specified time (typically 4–8 hours) must be re-dried at higher temperatures, typically 260°C to 430°C (500°F to 800°F) for 1–2 hours, per manufacturer instructions. Only one re-drying cycle is typically permitted.
- Portable heated quivers: At the welding station, electrodes should be kept in a portable heated quiver to prevent re-absorption during the welding shift.
Base Metal Surface Preparation
All surfaces within approximately 25 mm of the joint on both faces must be free of organic contamination. The minimum preparation standard is:
- Remove all oil, grease, and cutting fluids with a suitable solvent degreaser
- Remove rust, mill scale, and paint by mechanical means (grinding, wire brushing, blast cleaning) or chemical means (pickling)
- In cold or humid conditions: preheat to at least 10°C above ambient (minimum 10°C) to drive off condensed moisture before welding commences
- For submerged arc welding: verify that flux is properly dried — SAW flux in particular can absorb significant moisture during storage and should be recirculated or baked if the bag has been open for more than the supplier’s specified period
Prevention Strategy 2: Reduce Microstructural Susceptibility
The second prevention axis targets the microstructure. Since martensite susceptibility is directly related to HAZ cooling rate, any measure that slows the HAZ cooling rate reduces martensite content and HAZ hardness — and therefore HACC risk.
Preheat: Temperature and Zone
Preheat is the most widely used and most effective measure for controlling HAZ microstructure in production welding. Its effects are:
- Reduced cooling rate: Higher preheat reduces the temperature differential between the HAZ and the surrounding metal, slowing the rate at which heat is conducted away. This shifts the HAZ transformation from martensite toward bainite or pearlite — a softer, more hydrogen-tolerant microstructure.
- Increased hydrogen effusion: Maintaining the joint at temperatures in the range 93–232°C (200–450°F) significantly accelerates hydrogen diffusion out of the weld metal and HAZ before the structure cools to the critical cold cracking range below ~150°C.
- Reduced thermal shock: Preheat reduces the temperature gradient across the joint, lowering the magnitude of thermally-induced residual stresses — the third condition required for cracking.
Preheat requirements are specified in ASME B31.3 (process piping), ASME Section VIII Division 1 and Division 2, AWS D1.1 (structural steel), and EN 1011. The minimum preheat temperature is a function of the steel’s carbon equivalent, the material thickness (which controls heat sink effect), and the expected hydrogen level from the welding process.
Preheat Zone Width
Preheat must be applied over a sufficient area to actually influence the HAZ cooling rate — not just the visible surface adjacent to the joint. The preheat zone must extend at least 75 mm (3 inches) on each side of the weld centreline, or three times the wall thickness (whichever is greater), measured from the edge of the weld preparation. Preheat applied to a narrow strip immediately adjacent to the groove is ineffective and may give a false sense of security.
Heat Input Control
Welding heat input (kJ/mm or kJ/inch) directly controls the Δt8/5 cooling time — the time taken for the HAZ to cool from 800°C to 500°C — which in turn determines the transformation product. Higher heat input means slower cooling and a less martensitic HAZ. The heat input formula is:
V = arc voltage (V), I = welding current (A), TS = travel speed (mm/min)
Example: V = 24V, I = 180A, TS = 200 mm/min
HI = [24 × 180 × 60] / [200 × 1000]
HI = 1.30 kJ/mm
Thermal efficiency factor k applies: 1.0 for SMAW, 0.8 for GMAW, 0.6 for GTAW
Minimum heat input limits for hardenable steels are commonly specified in the Welding Procedure Specification (WPS) alongside preheat requirements. Straying below the minimum heat input limit — e.g., by running at lower amperage or higher travel speed than specified — can produce an excessively fast HAZ cooling rate and unacceptable martensite content, even when preheat is correctly applied.
Prevention Strategy 3: Reduce Residual Stress
Residual stress is the third condition in the cracking triangle. While it is not always practical to eliminate residual stress — it is inherent in the welding process — it can be managed and reduced to levels below those required to propagate hydrogen-assisted cracks.
Post-Weld Hydrogen Release Bake
For high-restraint joints or high-CE materials where maximum risk mitigation is required, a hydrogen release bake (also called a post-heat or post-weld bake) is applied immediately after welding completion, before the joint is allowed to cool to ambient temperature. The procedure is:
- Heat the completed joint (without allowing it to cool below the preheat temperature) to 200–300°C (400–570°F)
- Hold at temperature for at least 2 hours per 25 mm of thickness, minimum 2 hours
- Allow to cool slowly, wrapped in insulating blankets
At these temperatures, the diffusion coefficient of hydrogen in ferritic steel is high enough that most of the remaining diffusible hydrogen migrates to the surface and escapes within the hold period. This is distinct from full PWHT — the temperature is below the range where significant metallurgical changes occur, so it does not affect mechanical properties.
Full Post-Weld Heat Treatment (PWHT)
Full PWHT as required by ASME Section VIII UCS-56 or equivalent code provisions achieves both stress relief and HAZ martensite tempering simultaneously. At PWHT temperatures of 580–760°C, residual stresses are reduced by stress relaxation (creep at elevated temperature), and any martensite present in the HAZ is tempered to a softer, tougher condition. PWHT is mandatory for carbon steel above 38 mm thickness and at lower thicknesses for alloy steels under the ASME pressure vessel code. See the heat treatment guide for full PWHT requirements by material group.
Joint Design and Restraint Reduction
Joint design can significantly affect restraint levels. Where possible, design choices that reduce restraint include:
- Allowing free movement of one joint member during welding (avoid simultaneous multi-pass fully restrained situations)
- Using balanced welding sequences to distribute and minimise net residual stress
- Avoiding thick-to-thin intersections that create high triaxial stress states at the junction
- Specifying a generous root opening and included angle to reduce weld metal notch effects at the root
Inspector Responsibilities and NDE Timing
The welding inspector’s role in hydrogen cracking prevention is both proactive (before and during welding) and reactive (after welding, through NDE). The key responsibilities at each stage are:
Before Welding
- Verify that the consumables specified in the WPS are in use, with the correct H-designation suffix
- Check electrode oven temperature and confirm electrodes have not been exposed beyond the allowed out-of-oven time
- Confirm joint surfaces are clean, dry, and within the specified surface preparation standard
- Measure and record preheat temperature using a calibrated contact pyrometer or temperature-indicating crayons (thermocouples for critical applications)
- Verify that the preheat zone extends the required minimum distance from the joint
During Welding
- Monitor interpass temperature to confirm it remains within the specified range (minimum preheat to maximum interpass)
- Verify that the welding process parameters (current, voltage, travel speed) are within the WPS limits, ensuring adequate heat input
- If the weld is interrupted, confirm that preheat is maintained or that the joint is properly post-heated before the restart
NDE After Welding — The Critical Hold Time
Perhaps the most important inspection responsibility related to hydrogen cracking is the timing of NDE. Performing surface NDE immediately after welding is complete may miss the majority of hydrogen cracks, because they can continue to develop and propagate for hours or days after the weld cools. Applicable hold times before NDE include:
| Condition | Minimum Hold Time Before NDE | Reference |
|---|---|---|
| Carbon steel, CE <0.40, low restraint, low-H consumables | No specific hold required (24h recommended) | General practice |
| Carbon steel, CE 0.40–0.55, moderate restraint | 24 hours minimum after completion | AWS D1.1, EN 1090 |
| Low-alloy steel, high restraint, or CE >0.55 | 48–72 hours minimum | Project-specific QP |
| P91, Cr-Mo alloy steels — pre-PWHT inspection | 48 hours minimum at ambient; re-inspect post-PWHT | ASME B31.3, EN 13480 |
| High-strength structural steel (YS ≥550 MPa) | 48–72 hours; apply to full HAZ volume | Fabricator QP per EN 1011-2 |
Table 3 — Recommended minimum hold times before post-weld NDE for hydrogen crack detection. More conservative hold times are appropriate for thick sections, high CE, or high-restraint joints.
The NDE methods best suited to detecting hydrogen cracks are:
- Magnetic particle testing (MT) — most sensitive method for surface and near-surface (up to ~3 mm depth) cracks. Requires the steel to be ferromagnetic (not applicable to austenitic stainless).
- Dye penetrant testing (PT) — effective for surface-breaking cracks where MT is not applicable or as a complement to MT
- Phased array ultrasonic testing (PAUT) — required for detection of subsurface underbead cracks not accessible from the surface
- Time-of-flight diffraction (TOFD) — high sensitivity for planar defects in the HAZ volume, increasingly used for thick-section pressure vessel welds
Frequently Asked Questions
What are the three conditions required for hydrogen cracking in welds?
Hydrogen cracking requires the simultaneous presence of three conditions: (1) a source of diffusible atomic hydrogen in the weld zone, (2) a susceptible microstructure — typically hard martensite or high-hardness bainite in the HAZ or weld metal, and (3) sufficient residual or applied tensile stress to propagate a crack. Remove any one of these three conditions and hydrogen cracking cannot occur. All prevention strategies are directed at eliminating or reducing one or more of these three factors. See the companion article on martensite formation for more detail on the susceptible microstructure condition.
Why is hydrogen cracking called “delayed” or “cold” cracking?
The term “cold cracking” refers to the fact that hydrogen cracking initiates below approximately 150°C — it does not occur at elevated temperatures because the material is more ductile and hydrogen diffuses too rapidly to accumulate in dangerous concentrations. It is called “delayed” because cracking may not appear until hours or even days after welding is complete, as hydrogen continues to redistribute and accumulate in critical locations over time. This is why NDE performed immediately after welding completion may miss developing cracks — hold times of 24–72 hours are necessary for reliable detection.
What is the most effective way to prevent hydrogen cracking?
The most effective combination is low-hydrogen consumables (such as E7018-H8 or E7018-H4) together with adequate preheat. Low-hydrogen consumables eliminate the primary hydrogen source, while preheat slows the HAZ cooling rate to prevent hard martensite and accelerates hydrogen effusion from the weld zone. For highly restrained joints or high-CE steels, a post-weld hydrogen release bake at 200–300°C immediately after welding provides additional assurance. Use the carbon equivalent calculator to determine the CE of your material and the appropriate preheat level.
How does preheat prevent hydrogen cracking?
Preheat prevents hydrogen cracking through two mechanisms simultaneously. First, it reduces the HAZ cooling rate by reducing the temperature gradient between the weld and surrounding base metal — this promotes softer ferritic-pearlitic or bainitic microstructures rather than brittle martensite. Second, higher preheat temperatures accelerate hydrogen diffusion out of the weld zone — hydrogen diffuses rapidly out of ferritic steel between approximately 93°C and 232°C, so maintaining the metal in this range for longer allows more hydrogen to escape before the weld cools to the cold cracking initiation range below ~150°C. Preheat also reduces thermally-induced residual stresses, addressing the third condition in the cracking triangle.
What is the diffusible hydrogen content limit for low-hydrogen consumables?
AWS A5.1 classifies SMAW electrodes by diffusible hydrogen content in ml per 100g of deposited weld metal. H4 designates a maximum of 4 ml/100g, H8 a maximum of 8 ml/100g, and H16 a maximum of 16 ml/100g. For critical applications with higher CE steels, H4 or H8 designations should be specified. Basic low-hydrogen electrodes such as E7018 achieve H8 or better when properly stored and baked; cellulosic electrodes such as E6010 can produce 30–60 ml/100g — making them entirely unsuitable for hardenable steels without significant compensating measures. The H-designator must appear on the electrode box and in the manufacturer’s certification.
When should NDE be performed to detect hydrogen cracking?
NDE for hydrogen cracking should never be performed immediately after welding. A minimum hold time must elapse to allow delayed cracks to develop before inspection. Typical hold times range from 24 hours for plain carbon steel with moderate CE and low restraint, to 48–72 hours for alloy steels, high-CE steels, or high-restraint joints. AWS D1.1 and EN 1090 specify applicable hold times. Magnetic particle testing (MT) is the preferred method for surface and near-surface cracks in ferritic steel; phased array UT (PAUT) or TOFD is required to detect subsurface underbead cracks.
Which steels are most susceptible to hydrogen cracking?
Susceptibility increases with carbon equivalent. Steels with IIW CE above 0.45 are considered susceptible and require preheat and low-hydrogen consumables. Medium and high-carbon steels, low-alloy hardenable grades such as Cr-Mo (P11, P22, P91), quenched-and-tempered structural steels, and high-strength steels with yield strength above 550 MPa are all highly susceptible. Austenitic stainless steels and nickel alloys are not susceptible to HACC because their FCC crystal structure allows rapid hydrogen diffusion at ambient temperature and their microstructures are far more ductile.
What is the difference between underbead cracking and toe cracking?
Both are forms of hydrogen-assisted cold cracking but occur at different locations. Underbead cracking forms in the coarse-grained HAZ beneath the weld bead, parallel to the fusion line, in the region of maximum martensite content and maximum hydrogen trapping due to large prior austenite grain size. Toe cracking initiates at the weld toe — the geometric notch at the junction between the weld face and the base metal surface — where stress concentration combines with a hard HAZ and accumulated hydrogen. Root cracking occurs in the root pass HAZ of single-sided welds under high restraint conditions, particularly common in pipeline root pass welding. All three are detected by surface NDE (MT or PT) after the appropriate hold time, with subsurface variants requiring volumetric UT methods.
Recommended Technical References
Disclosure: WeldFabWorld participates in the Amazon Associates programme (StoreID: neha0fe8-21). If you purchase through these links, we may earn a small commission at no extra cost to you. This helps support free technical content on this site.