Journal Information
Vol. 8. Issue 5.
Pages 4364-4373 (September - October 2019)
Download PDF
More article options
Vol. 8. Issue 5.
Pages 4364-4373 (September - October 2019)
Original Article
DOI: 10.1016/j.jmrt.2019.07.047
Open Access
The contribution of impurities to unexpected cold cracks in a thick C-Mn steel plate
Aline Raquel Vieira Nunesa,
Corresponding author

Corresponding author.
, Annelise Zeemannb, Luiz Henrique de Almeidaa
a Metallurgy and Materials Department at COPPE/UFRJ, Rio de Janeiro, RJ 21.941-972, Brazil
b TECMETAL, Rio de Janeiro, RJ, Brazil
This item has received

Under a Creative Commons license
Article information
Full Text
Download PDF
Figures (12)
Show moreShow less
Tables (5)
Table 1. Chemical Composition of the EN 10025-2 ST52-3 grade S355J C-Mn steel plate (wt %).
Table 2. Welding conditions for the C–Mn steel 6″ thick plate that was originally cracked.
Table 3. Hydrogenation conditions for test pieces A, B and C.
Table 4. Conditions and welding parameters for the autogenous pass on the three test pieces.
Table 5. Summary the microstructural analysis of the weld metal and HAZ, hardness values measured in the different regions and crack location for the three test pieces.
Show moreShow less

The internal cracking observed in a 6″-thick carbon-manganese steel plate after welding is discussed with respect to the presence of material impurities. The morphology and location of the cracks detected during fabrication of a heavy structure are typical of hydrogen-induced cracking (HIC) usually associated to hard untempered martensite. However, typically martensitic microstructures would not be expected for low carbon low alloy steels joints when a suitable welding procedure specification is adopted, which includes preheating and high heat input. This paper illustrates the morphology of the original cracks, which are unrelated to hard microstructures, and presents the results of the welding experiments conducted in samples cut from segregated regions of the 6″-thick plate. Through hydrogenation of thin slices containing impure regions and autogenous gas tungsten arc welding bead, it was possible to reproduce the same cracking characteristics formed during submerged arc welding, indicating that HIC may also be associated to softer phases. Hydrogen entrapment at the weld fusion line can be related to high impurities level due to the expected lower melting points at grain boundaries, explaining the HIC phenomenon. The present paper proposes a model for hydrogen cracking of bainitic structures at the welding heat affected zone that typically is not susceptible to crack.

Cold cracks
Impurity content
Segregation bands
C–Mn steel
Full Text

The adoption of suitable parameters, such as preheating temperature and heat input magnitude, for welding thick plates of carbon steel, is necessary to avoid hydrogen-induced cracking (HIC), also referred to as “cold cracking” as it forms near room temperature after welding [1–8].

Several theoretical mechanisms have been proposed to explain this type of cracking but there is still no universally accepted theory [2,6,9]. According to Lippold [2], the high mobility of atomic hydrogen in ferritic microstructures allows it to diffuse into regions with stress concentration raisers or susceptible microstructures, accumulating in trapping sites and causing “delayed” cracking when the hydrogen threshold concentration is reached.

The Carbon Equivalent (CE) is the parameter used to evaluate susceptibility of the base metal steel and to define ideal welding conditions to prevent HIC. The most commonly used CE formula is presented in Equation I, adopted by IIW [10], although others consider small differences in the hardening effects of the elements for martensite formation [11]. A steel with a low carbon content and a low CE does not exhibit cold cracking since no martensite is expected at the HAZ.

Then, the base metal steel chemical composition is the key to prevent cold cracks. HIC usually appears at the heat affected zone (HAZ) after cooling because of the presence of hydrogen, residual stresses and a “susceptible microstructure,” and less frequently in the weld metal (WM) since the carbon content in this region is usually lower. For C-Mn steels this HAZ susceptible microstructure corresponds to non-tempered martensite with hardness values usually greater than 350HV (35HRc) that forms at high cooling rates [12].

The influence of impurities have been studied concerning HIC associated to local chemistry enrichment in segregation bands [13,14], mainly carbon that increases hardenability, leading to alerts for the adoption of higher preheating temperatures to prevent HIC [15].

Little attention has been paid, however, to the contribution of impurities, such as sulfur, in hydrogen-assisted cracking [3]. Although low melting point elements do not increase the material hardenability from one side they may increase the hydrogen entrapment from other side, since segregated regions of the material will remain liquid longer, maintaining hydrogen in solution.

Thick plates require careful welding procedures since fast cooling rates are expected and consequently the presence of martensite. Experienced fabricators of heavy structures are familiar with the precautions required for welding thick plates and adopt only qualified welding procedure specifications (WPS), which include preheating the material, a high heat input submerged arc welding (SAW) process, and a mandatory post-welding heat treatment (PWHT). However, ultrasonic tests (UT) showed that this 6″ thick plate of EN 10025-2 ST52-3 grade S355J steel, welded by a qualified fabricator, cracked before the PWHT was applied. This unexpected result motivated the present study.

Metallurgical analysis of the cracked region of the 6″ plate, which was removed during repair, indicated the presence of a segregated central region, associated with intergranular cracks, at previous austenite grain boundaries of the coarse-grained HAZ. The cracking location and morphology suggested that cold cracking occurred, but the HAZ microstructure was fully bainitic and the maximum Vickers hardness value was only 220 HV, making the cracking mechanism unclear.

Thin slices were cut at the segregated regions of the plate to create test pieces. A high temperature dehydrogenation heat treatment was performed to ensure that no hydrogen remained entrapped. The test pieces were machined, polished and exposed to various hydrogenation conditions. After hydrogenation, the test pieces underwent an autogenous GTAW cycle, creating a liquid pool. The adopted welding parameters for the thin slices were suitable to create a low hardness HAZ with the same characteristic of the original microstructure of the SAW 6″ plate HAZ. Intergranular cracking was observed in the test piece welded immediately after hydrogen charging, indicating that even softer steels may crack if hydrogen is present.

A model including the contribution of hydrogen entrapment to cold cracking is proposed to explain the formation of cracks at segregated HAZ regions. In this model, the HIC is not associated to hydrogen migration into HAZ. Instead, it is associated to the hydrogen entrapment at the coarse grain boundaries.


The material used in this study was removed from the cracked region (near the weld metal) of a 6″ C–Mn carbon steel plate, that was cut in three sections for practical reasons, as shown in Fig. 1. The microstructure of the center of the plate, as shown in Fig. 2, is ferritic-pearlitic with marked segregation bands.

Fig. 1.

Macrography of a transverse section of the 6″ C–Mn steel plate after removal from the equipment. Nital etch.

Fig. 2.

Optical micrography of the center of the steel plate showing ferritic-pearlitic base metal with segregation bands. Nital etch.


The chemical composition of the EN 10025-2 ST52-3 grade S355J C-Mn steel plate is presented in Table 1 and indicates that the steel is of a low carbon content (0.164%), with a typical CE of 0.42%.

Table 1.

Chemical Composition of the EN 10025-2 ST52-3 grade S355J C-Mn steel plate (wt %).

Mn  Si  Cr  Ni  Mo  Al  Cu  Ti  Nb  N (ppm)  CE 
0.16  1.43  0.39  0.023  0.011  0.04  0.01  0.03  0.05  0.08  0.004  0.004  0.001  0.02  155  0.42 

Carbon equivalent.

The 6″ C–Mn carbon steel plate was welded from both sides, in a full penetration double V groove, using a high heat input SAW process and a preheating temperature of 80°C, all qualified parameters, suitable for the specific material. The weld conditions are summarized in Table 2.

Table 2.

Welding conditions for the C–Mn steel 6″ thick plate that was originally cracked.

Weld conditions  Welding parameters 
Groove and position  Double V – bevel, welded in 1G position, both sides 
Welding process  Multiple-pass submerged arc welding 
Wire/flux  EM 12K/neutral flux 
Preheating temperature (min)  80°C 
Interpass temperature (max)  250°C 
Post-welding heat treatment  None because cracking was detected after welding 

Fig. 3 shows the microstructure of the cracked region of the 6″ C–Mn steel plate. The cracking is intergranular at the coarse-grained HAZ, which is typical for HIC, but associated with a low hardness bainite structure. The cracks appeared only in the center of the plate thickness.

Fig. 3.

Optical micrographies of the cracked region in the center of the 6″ C–Mn steel plate. A – Intergranular cracking at the HAZ; B – detail of the intergranular cracking (at previous austenite grain boundaries) in the bainitic structure. Nital etch.


Fig. 4 schematically summarizes the experimental procedure adopted to study the influence of the segregation bands on hydrogen cracking after welding. Fig. 4a shows the cracking in the original 6″ plate, Fig. 4b shows the location from which the segregated material was removed and Fig. 4c shows how the slices were cut to create test pieces.

Fig. 4.

Scheme of the experimental procedure adopted in the present study.


Three 100mm×28mm test pieces pf 4mm thickness were cut from the segregated plate sample (Fig. 4c) and identified as A, B and C. The test pieces were heat treated in a furnace at 900°C for 2h and slowly cooled, as shown in Fig. 4d. This treatment was applied to remove any hydrogen entrapped in the base metal and anneal the steel. After the heat treatment, the test pieces were polished (Fig. 4e) to ensure adequate conditions for hydrogen charging.

Fig. 4f presents the electrolytic hydrogen charging conditions for each test piece. The samples were hydrogen charged for 72h using the parameters listed in Table 3 to create uniform charging throughout the 4mm thickness of the test pieces.

Table 3.

Hydrogenation conditions for test pieces A, B and C.

Test piece 
Conditions  Hydrogenated  Hydrogenated  Non-hydrogenated 
Electrolyte  0.1M NaOH  0.1M NaOH  – 
J (mA/cm220  20  – 
Charging time (h)  72  72  – 
Waiting time between hydrogenation and welding (h)  48  None  – 

The amount of hydrogen in the material was not measured, but the effects that were later identified in the welded condition showed that the hydrogenation was effective in introducing the hydrogen.

Test piece A was welded 48h after hydrogen charging; test piece B was welded immediately after hydrogen charging and test piece C was welded without hydrogen charging.

Each test piece (A, B and C) was subjected to a welding thermal cycle via a single autogenous GTAW pass in the flat position (Fig. 4g), using the welding parameters listed in Table 4. Fig. 5 shows the test piece during and after welding (Fig. 5A) as well as the surface finish after welding (Fig. 5B).

Table 4.

Conditions and welding parameters for the autogenous pass on the three test pieces.

Conditions and welding parameters
Preheating temperature  Ambient 
Current (A)  160 
Voltage (V)  12 
Speed (mm/min)  160 
Heat input (kJ/mm)  0.72 
Post-welding heat treatment  None 
Fig. 5.

Autogenous TIG arc welding by automatic GTAW of the slices. The thermal cycle was applied in one pass. A – Thin slice fixed in the welding device. B – Surface finish of the weld bead.


The lack of filler metal and the welding parameters created a weld metal with the same composition as the base metal (100% dilution) and a good bead penetration without any perforation of the thickness of the test pieces. The start and stop arc regions were cut and discarded from the test pieces.

The welding cycle was applied to the material using an automatic autogenous GTAW bead on a plate (thin slice), in one pass. Bainite is the microstructure at the HAZ, in the same way as the original welded joint HAZ.

During welding no strong evidence of the presence of hydrogen, such as boiling of the melt pool, was detected. All test pieces were welded in the same day, one after the other.

Two days after welding, all the welded test pieces were evaluated, visually and through a metallurgical characterization. Macrographs of the center of the test pieces were taken. Vickers hardness tests using a 1kg (HV1) load were conducted at different regions of the weld metal and the HAZ. Transverse cross-sections of the samples were polished and etched with Nital. The microstructures were evaluated using optical microscopy (OM) and scanning electron microscopy (SEM).


The appearance of the three test pieces was similar after welding. No macropores or macrocracks were identified in the test pieces. All test pieces, welded at the same position along the thickness of the samples and with the same parameters, had similar welding profiles, microstructures and hardness values.

Fig. 6 shows a transverse cross-section of test piece B, welded immediately after hydrogen charging. It is important to note that only test piece B displayed intergranular cracks. Fig. 6A shows the diagonal stains that reveal the segregations in the steel (arrow) and the cracks appeared in a region aligned with a segregation. The cracks are under the bead (Fig. 6B) in a bainitic region of the coarse-grained HAZ (Fig. 6C).

Fig. 6.

Features of the weld bead produced by autogenous GTAW. A – Macrography of a transverse section of the welded region showing a segregation band (diagonal – arrows); B – Optical micrography showing the crack location at the HAZ in a segregation band; C – Detail of the bainitic structure cracked at which previous austenitic grain boundaries. Nital etch.


The microstructure of the weld metal, which has a similar composition to the base metal due to the 100% dilution, is bainitic with interdendritic martensitic islands and some micropores, as shown in Fig. 7. The martensitic-bainitic weld metal was harder than the bainitic HAZ, as summarized in Table 5, but no cracks were detected in the WM, suggesting that the amount of hydrogen entrapped during welding was not sufficient to form cracks in the melted region even though the WM microstructures are hardner than the HAZ. WM presenting up to 418 HV and HAZ with maximum 283 HV.

Fig. 7.

Metallographies of the weld metal, which has the same chemical composition as the base metal (autogenous weld). A – OM of the bainitic structure in the weld metal with interdendritic martensite islands; B – high magnification SEM image of the interdendritic regions of the weld metal. Nital etch.

Table 5.

Summary the microstructural analysis of the weld metal and HAZ, hardness values measured in the different regions and crack location for the three test pieces.

Test piece 
Weld metalMartensite and bainite  Martensite and bainite  Martensite and bainite 
Maximum 419 HV1  Maximum 418 HV1  Maximum 416 HV1 
HAZBainite  Bainite  Bainite 
Maximum 303 HV1  Maximum 283 HV1  Maximum 304 HV1 
Cracking  No  Yes at the GC HAZ  NO 

The intergranular cracks observed in test piece B are located at the grain boundaries of the HAZ, as illustrated in Fig. 8, in a bainitic region with hardness less than 350 HV similar to the original cracking at the HAZ in the center of the 6″ C-Mn steel plate. These observations indicate that the experiment successfully produced hydrogen cracks in the impure material.

Fig. 8.

Microstructure of the HAZ, cracked in the previous austenite grain boundary A – OM; B – SEM. Nital etch.


Fig. 9 shows two features of the effect of hydrogen on the low hardness HAZ of test piece B: blisters and intergranular cracks at the grain boundaries of the bainitic structure. Fig. 10 shows, at higher magnification, small voids (blisters), at the coarse grain boundaries. These voids were created due to the local reduction of the melting point by sulfur segregation. The EDS spectrum presented in Fig. 10 confirms the presence of sulfides at grain boundaries, opening blisters.

Fig. 9.

Intergranular crack of welded test piece B (hydrogen charged and immediately welded). A – Blisters and intergranular crack at HAZ previous austenite coarse grains; B – intergranular crack opening in a bainitic structure with a hardness of only 280 HV1. A carbide film is present at the cracks. Nital etch.

Fig. 10.

Test piece B. A – SEM micrography of the voids (blisters) created by hydrogen entrapment near the fusion line; B – EDS spectrum (arrow), showing evidence of sulfides. Nital etch..


The welding experiments carried out in the present study on a segregated region of a hydrogen charged plate led to two important observations.

First, the hydrogen embrittlement after welding can be associated with a low hardness bainitic microstructure, i.e., depends on the hydrogen availability in the material. In this experiment, the hydrogen was electrolytically introduced into the base metal before welding, but in industrial settings hydrogen can result from the ambient humidity, flux, and traces of oil, grease and rust during welding or be entrapped in the base metal if the plate is supplied in a non-degassed condition. Segregated thick plates facilitate hydrogen entrapment in regions containing impurities and increase the cold cracking susceptibility.

Second, and most important, the accepted theory of hydrogen diffusion from the solid WM into the HAZ, during and after welding, to crack regions of stress concentrations raisers plus susceptible microstructures [2], may not be accurate for all situations. In this study, the hard susceptible martensitic regions of the autogenous weld metal (418 HV1) did not crack, which indicates that the hydrogen was diffused out of the WM. On the other hand, cracks were formed in the soft bainitic regions (283HV1), associated with strong traps at the coarse grain boundaries of the HAZ due to a relatively low hydrogen mobility after cooling.

Voids associated with the small sulfide inclusions at previous austenitic grain boundaries, Fig. 10, were observed in the bainitic regions, suggesting that the base material impurities create regions at the coarse grain boundaries near the fusion line, with low melting points. These regions act as traps for hydrogen, which create blisters as the material cools. When a carbide film is present at the grain boundary, the blisters have a crack-like appearance, as shown in Figs. 8B and 9B, suggesting that the carbides at the grain boundaries are stronger entrapment sites. Depending on the microstructure near the grain boundary, the hydrogen pressure in the traps creates voids or cracks.

The HIC in this case can be explained by the entrapment of hydrogen in regions with low melting points that remain liquid longer after the pool solidifies and by the relatively low hydrogen mobility in these regions after cooling. The impurities from the base metal increase the susceptibility for HIC since they increase hydrogen entrapment at the coarse grain boundaries of the partially melted zone at the fusion line.

4.1Proposed model for hydrogen entrapment during welding

A new model based on hydrogen entrapment at the fusion line of a welded joint is presented to explain HIC. Fig. 11 presents the proposed sequence of events that caused the hydrogen assisted cracking associated with a highly segregated region of a steel plate in a non-susceptible microstructure.

Fig. 11.

Schematic representation of the proposed sequence of events that lead to HIC in a transverse section of a test piece during the welding cycle. A – Base metal before the welding cycle; B – maximum temperature during arc welding, showing partially melted regions of the base metal; C – solidification of the pool, and hydrogen evolution at the surface; D – almost fully solid weld metal that is austenitic and contains dissolved hydrogen; E – regions of the partially melted zone in the base metal in which hydrogen was entrapped during welding; F – intergranular cracking in the coarse grains of the HAZ due to hydrogen embrittlement after cooling.


Fig. 11A shows the test piece that was previously hydrogen charged. During the autogenous arc welding, Fig. 11B, the hydrogen concentration in the melted pool (WM) increases due to the migration of atoms from low to high temperature regions.

Fig. 11B illustrates the partial melting, at the fusion line, of coarse grain boundaries. This liquid grain boundary is enriched with alloying elements and impurities, and the hydrogen concentration is increased since these are regions of low melting point. Additionally, the concentration of impurities and hydrogen is higher for a coarse austenitic grain (at the fusion line).

It is proposed that upon melting more hydrogen is dissolved in the partially melted regions of the grain boundaries because, although these regions are at the same temperature as the adjacent solid, they will remain liquid for a longer time allowing the grain boundaries to trap more hydrogen. The reduction in the melting temperature depends on the concentration of solute or impurities, as shown in Fig. 12, so the grain boundaries may remain liquid longer for more impure steels.

Fig. 12.

Schematic detail of the partially melted region at the fusion line (Fig. 11 B). The extension is higher for a more segregated steel.


The continuous solidification of the weld metal forces the hydrogen out of the weld metal, as shown in Fig. 11C. After the liquid transforms into austenite (Fig. 11D), ferrite, bainite or martensite (Fig. 11E), the amount of hydrogen decreases because hydrogen diffusion is faster through the ferritic structures.

By the end of the welding cycle only the grain boundaries at the fusion line are enriched in entrapped hydrogen. The level of entrapment depends, as previously stated, on the amount of impurities, and for impure steels a higher cold cracking susceptibility may occur. It is interesting to note that this proposal for cold cracking in fact combines the liquation susceptibility and the hydrogen susceptibility, that is a combination previously proposed by Savage et al. for HY80 steel [16], when they analyze the effect of impurities on hot and cold cracking, associated with partially melted boundaries.

After cooling the weld, the hydrogen-enriched grain boundaries are not able to accommodate the internal pressure of hydrogen, which leads to the formation of voids or cracking, depending on the diffusivity and ability of the metal to accommodate hydrogen (Fig. 11F). In the present study the cracking occurred only at large grain boundaries, which had a higher concentration of impurities, where sulfides melted and solidified.

Thus, the observation of the effects of impurities in entrapping hydrogen at the fusion line allows a new model for explaining HIC, even for steels that form martensite. During welding the partially melted region at the fusion line entraps hydrogen at grain boundaries. This region is the coarse grain HAZ that later forms martensite. Since the hydrogen mobility is reduced with temperature reduction, and the diffusion rate through the martensitic microstructure is low, associated to an increase in residual stresses, the cracks appear when the grain boundaries are not able to accommodate the internal pressure.

In this model, hydrogen embrittlement is not only associated with the hardness of the HAZ (although harder microstructures reduce diffusion rate and reduce the capacity to accommodate the hydrogen), but also with the impurity content, which increases the extension of the liquidus to solidus region at the fusion line, favoring hydrogen entrapment and reducing the release of hydrogen after cooling.

Applying the above concepts to the original impure 6″ C-Mn steel plate indicates that the segregation bands in the center of the plate created a condition for hydrogen entrapment at the welding fusion line that could be responsible for later cracking. The hydrogen source for the original cracking may be related to the SAW welding (wet flux) or the non-degassed base metal. The thickness of the plate could be an additional barrier for hydrogen diffusion out of the base metal.

The adoption of higher preheating temperatures and hydrogen evolution treatment after welding shows to be essential procedures to avoid HIC in thick plates.


Thick plates with high impurity contents may suffer HIC independently of the steel cold cracking susceptibility predicted by their carbon contents and CE, due to their hydrogen entrapment potential. In this work, hydrogen was intentionally introduced prior to welding, which resulted in cracking of coarse grain boundaries at partially melted regions rather than higher hardness regions.

The proposed model contributes to the understanding of the mechanism of cold cracking formation.

Conflicts of interest

The authors declare no conflicts of interest.


The authors gratefully acknowledge the financial support of the CNPq (132120/2013-5) and FAPERJ. The authors would like also to thank the TECMETAL Company for their support.

H.W. Lee, S.W. Kang, D.S. Um.
A study on transverse weld cracks in thick steel plate with the FCAW process I.
Weld Res Suppl MVC, (1998), pp. 503-510
J.C. Lippold.
Welding metallurgy and weldability.
J Chem Inform Model, 53 (2013),
S. Kou.
Welding metallurgy.
John Wiley & Sons, Inc, (2003), pp. 410-417
Y.D. Park, I.S. Maroef, A. Landau, D.L. Olson.
Retained austenite as a hydrogen trap in steel welds.
Weld J, 81 (2002), pp. 27-s-35-s
N. Yurioka, H. Suzuki.
Hydrogen assisted cracking in c-mn and low alloy steel weldments.
Int Mater Rev, 35 (1990), pp. 217-249
Gaseous hydrogen embrittlement of materials in energy technologies: mechanisms, modelling and future developments,
R.P. Gangloff.
Critical issues in hydrogen assisted cracking of structural alloys.
Environment Induced Cracking of Metals (EICM-2), Elsevier Science, (2006), pp. 1-26
K.O. Findley, M.K. O’Brien, H. Nako.
Mechanisms of hydrogen induced cracking in pipeline steels.
Mater Sci Technol, 31 (2015), pp. 1673-1680
M.B. Djukic, G.M. Bakic, V.S. Zeravcic, A. Sedmak, B. Rajicic.
Hydrogen embrittlement of industrial components: prediction, prevention, and models.
Corrosion, 72 (2016), pp. 943-961
J.F. Lancaster.
Metallurgy of welding.
6th ed, Abington Publishing, Elsevier, (1999), pp. 464
V.B. Ginzburg, R. Ballas.
Flat rolling fundamentals.
CRC Press, (2000), pp. 141-142
W.A. Bruce, P.E. IWE.
Control of hydrogen cracking for in-service repair of pipelines.
IIW international conference on welding for repair and life extension of plants and infrastructure,
W.B. Morrison, K. Abernethy.
Influence of controlled levels of centreline segregation on the toughness of structural steels: final report (technical steel research). EUR 12771 EN.
(1990), pp. 52
J. Ćwiek.
Prevention methods against hydrogen degradation of steel.
J Achiev Mater Manufact Eng, 43 (2010), pp. 214-221
N. Bailey.
Welding steels without hydrogen cracking.
Woodhead Publishing, (1993),
W.F. Savage, E.F. Nippes, E.S. Szekeres.
Hydrogen induced cold cracking in a low alloy steel.
Weld Res Suppl, 55 (1976), pp. 276-284
Copyright © 2019. The Authors
Journal of Materials Research and Technology

Subscribe to our newsletter

Article options
Cookies policy
To improve our services and products, we use cookies (own or third parties authorized) to show advertising related to client preferences through the analyses of navigation customer behavior. Continuing navigation will be considered as acceptance of this use. You can change the settings or obtain more information by clicking here.