JMEPEG (2018) 27:2513–2523 https://doi.org/10.1007/s11665-018-3330-x

ÓASM International 1059-9495/$19.00

Investigation of Hydrogen Embrittlement Susceptibility of X80 Weld Joints by Thermal Simulation Huangtao Peng, Teng An, Shuqi Zheng, Bingwei Luo, Siyu Wang, and Shuai Zhang (Submitted July 21, 2017; in revised form February 4, 2018; published online April 2, 2018) The objective of this study was to investigate the hydrogen embrittlement (HE) susceptibility and influence mechanism of X80 weld joints. Slow strain rate testing (SSRT) under in situ H-charging, combined with microstructure and fracture analysis, was performed on the base metal (BM), weld metal (WM), thermally simulated fine-grained heat-affected zone (FGHAZ) and coarse-grained heat-affected zone (CGHAZ). Results showed that the WM and simulated HAZ had a greater degree of high local strain distribution than the BM; compared to the CGHAZ, the FGHAZ had lower microhardness and more uniformly distributed stress. SSRT results showed that the weld joint was highly sensitive to HE; the HE index decreased in the following sequence: FGHAZ, WM, CGHAZ and BM. The effect of the microstructure on HE was mainly reflected in microstructure, local stress distribution and microhardness. Keywords

hydrogen embrittlement, thermal simulation, weld joints, X80

1. Introduction Hydrogen, as an energy carrier for both traditional energy and new energy, is closely related to our economic benefit and safety through its transportation and storage. Considering their transmission efficiency and operational costs (Ref 1), largediameter high-strength pipelines have become a widely accepted choice. However, there is a serious issue that must be considered: Hydrogen can easily interact with high-strength low-alloy steels and degrade their mechanical properties (Ref 24). This is known as hydrogen embrittlement (HE). HE is a very serious problem in the oil and natural gas industry, especially for the weld joints. Although they are indispensable for connecting long-distance transmission pipelines, weld joints are complicated and cause frequent pipeline failures (Ref 5-7) owing to their heterogeneous microstructures (Ref 8, 9) and complex stress states (Ref 10). Several problems related to the microstructural characteristics and HE of weld joints have been studied. Researchers (Ref 8, 9) have found that weld joints are highly susceptible to hydrogen because of their heterogeneous microstructures and chemical composition caused by the grain recrystallization and coarsening during welding. Cheng et al. (Ref 11) found the hydrogen diffusion rate to be closely related to the content and distribution of trapping sites, such as inclusions and grain Huangtao Peng, Teng An, Bingwei Luo, Siyu Wang, and Shuai Zhang, State Key Laboratory of Heavy Oil Processing and Department of Materials Science and Engineering, China University of Petroleum, Beijing 102249, China; Shuqi Zheng, State Key Laboratory of Heavy Oil Processing and Department of Materials Science and Engineering, China University of Petroleum, Beijing 102249, China; and Beijing Key Laboratory of Failure, Corrosion and Protection of Oil/Gas Facility Materials, China University of Petroleum, Beijing 102249, China. Contact e-mail: [email protected].

Journal of Materials Engineering and Performance

boundaries, when they evaluated the hydrogen permeation behavior of different microstructures (BM, WM, HAZ) in X80 weld joints. Owing to the limited size of an actual HAZ, Zhao et al. (Ref 12) used thermally simulated CGHAZ microstructures instead and demonstrated their consistency with physical HAZs; SSRT results showed that the BM had the lowest HE susceptibility, while the simulated CGHAZ, owing to its coarse structure and the quick accumulation of hydrogen, had the highest susceptibility. A few researchers have studied the effect of residual stress in weld joints. Pardal et al. (Ref 13) found that hydrogen-induced cracking (HIC) tended to occur in weld joints owing to their high residual stresses, which can promote the permeation of atomic hydrogen into the crystalline structure. Widely accepted HE mechanisms are hydrogen enhanced localized plasticity (HELP) (Ref 14) mechanism, hydrogeninduced decohesion (HEDE) (Ref 15) mechanism, internal pressure (HP) (Ref 16), etc. However, all these mechanisms have corresponding restrictions in terms of their use. There have been several studies on the complexity (Ref 810) and HE of X80 weld joints in a broad range of temperatures. However, relatively few studies have focused on the HE susceptibility of more detailed HAZ microstructures, such as the overheated zone (Tpeak > 1300 °C) and normalized zone (Ac3 < Tpeak < 1300 °C). It has been proven that the overheated and normalized zones have different microstructures and properties, such as phase size and content (Ref 17, 18), toughness (Ref 19), microhardness (Ref 17, 20) and corrosion resistance (Ref 17). Although some researchers have noticed these issues, they have either treated the entire HAZ as a monolith (Ref 11, 21) or used simple representations of the more complicated actual HAZ (Ref 12). Additionally, although it has been reported (Ref 22, 23) that the hardness can influence the HE behavior by affecting cracks nucleation and propagation in stress-oriented hydrogen-induced cracking (SOHIC) and HIC, there have only been a few such reports for X80 weld joints. However, this is an important phenomenon and deserves to be studied further. Therefore, it is necessary to study the weld joints and especially their more detailed HAZ microstructures to investigate their HE susceptibility and corresponding mechanisms.

Volume 27(5) May 2018—2513

by adding liquid nitrogen. The tensile-specimen fracture surfaces were subjected to detailed analyses via SEM (FEI Quanta 200F) and EDS. Dislocations and microcracks distributions were also studied by TEM (JEOL 2010 LaB6).

In this study, two typical HAZ microstructures (FGHAZ, CGHAZ) were thermally simulated, followed by SSRT with in situ hydrogen charging on four weld-joint microstructures (BM, WM, FGHAZ and CGHAZ). Microhardness testing, along with microstructural analyses, such as optical microscopy (OM)), scanning electron microscopy (SEM) and energy dispersive x-ray spectroscopy (EDS), transmission electron microscopy (TEM) and electron backscatter diffraction (EBSD) were conducted to precisely evaluate the influence of the microstructure on the HE susceptibility of X80 weld joints.

2.3 Mechanical Testing SSRT was conducted on the four microstructure specimens with and without hydrogen charging at a strain rate of 1.0 9 106 s1 at room temperature. Hydrogen-charged specimen was performed cathodically in a mixed aqueous solution of 0.3 mol L1 sulfuric acid (H2SO4) and 3 g L1 ammonium thiocyanate (NH4SCN) at a 20 mA cm2 current density. An uncharged specimen was tested in air for comparison. Before the SSRT, all specimens were mechanically cleaned and polished to keep them smooth. All hydrogen-charged specimens were pre-charged for 0.5 h. The shape and size of the SSRT tensile specimen is shown in Fig. 2. The nanoindentation hardness of the four original microstructure specimens was measured to investigate the influence of HE mechanism. Hardness tests were performed on a UNATSEM 2 from ASMEC equipped with a Berkovich pyramidal-shaped indenter tip under a 10 mN load. For accuracy, all hardness specimens were electrically polished before the measurements.

2. Experimental 2.1 Material and Thermal Simulation The material studied was a high-strength API X80 steel used for pipeline manufacturing (U1219 9 22 mm). The chemical composition is given in Table 1. BM and WM specimens for SSRT and hardness measurements were directly cut from the pipelines, and the two simulated HAZ microstructure specimens (FGHAZ, CGHAZ) were acquired from the treated BM by a Gleeble 3500 thermal simulator. The thermal simulation parameters, such as heating rate, peak temperature, holding time at high temperature and cooling rate, were acquired from the submerged arc welding during pipe manufacture. The following empirical equation to predict the austenite formation temperature AC3 for low-alloy steel with < 0.6 wt.% C was proposed by Andrews (Ref 24) and has also been cited by Zhao (Ref 25) and Wang (Ref 9): h AC3 ð CÞ ¼ 910  230 xðCÞ0:5 15:2xðNiÞ þ 44:7xðSiÞ i þ104xðVÞ þ 31:5xðMoÞ þ 13:1xðWÞ

3. Results 3.1 Microstructure Evolution and Characterization Optical micrographs of the BM, WM, FGHAZ and CGHAZ are shown in Fig. 3, and the corresponding TEM images are shown in Fig. 4. The BM microstructure was composed of granular bainite (GB), polygonal ferrite (PF) and a small amount of retained martensite/austenite (M/A) (Fig. 3a). It has

ðEq 1Þ here, x is the mass fraction of the element. The AC3 of X80 tested in this work was 868.3 °C. To distinguish the non-phase transformation microstructure specimen from the BM, two peak temperatures above the complete phase transformation temperature were chosen: 1350 °C to represent the overheated zone (CGHAZ) and 950 °C to represent the normalized zone (FGHAZ). Finally, the cooling rate was identified based on the measured parameters and the actual HAZ microstructure. The thermal cycles of the simulated X80 HAZ are shown in Fig. 1.

2.2 Microstructure Observation The microstructure was characterized by OM (Leica DM2500M) and EBSD (EDAX-TSL). For more accurate microstructure information, the EBSD specimens were mechanically polished and electrically polished to remove the surface strain layer. The electropolished test was conducted in a mixed solution (glacial acetic acid (CH3COOH): perchloric acid (HClO4): glycerol (C3H10O3) = 200: 10: 11(vol.%)) at 30 V for 4 min, and the experiment temperature was held at 8-16 °C

Fig. 1 Schematic diagram of welding thermal cycles for the two simulated HAZ specimens

Table 1 Chemical composition of X80 steel used in this study (wt.%) C 0.05

Si

Mn

P

S

Cr

Ni

Ti

Cu

Al

Nb

N

Fe

0.22

1.65

0.01

0.001

0.24

0.01

0.01

0.12

0.03

0.05

0.004

Balance

2514—Volume 27(5) May 2018

Journal of Materials Engineering and Performance

been reported that GB consists of fine-lath ferrite (acicular ferrite: AF) and some island constituents which are often called martensite/austenite (M/A) (Ref 26). The average grain size of the BM was approximately 5 lm, and a high dislocations density was observed between the AF grain boundaries (Fig. 4a). The WM mainly consisted of intra-granular nucleated acicular ferrite (IAF) and small-size GB (Fig. 3b). The AF was intertwined like a woven fabric. The TEM image in Fig. 4b shows several closely connected elongated ferritic lath bundles with a potentially lower dislocation density than the BM. Formation of this microstructure could be explained by the presence of many inclusions or precipitates in the WM, such as Al2O3, SiO2 and Ti2O3. These inclusions and precipitates could not only act as the grain nucleation sites, causing the newly born phase to grow in a radial shape with different orientations in prior austenite grains (Ref 27), but also inhibit the excessive growth through a pinning effect. The two HAZ specimens had relatively similar microstructures with large GB and bainitic

ferrite (BF) (Fig. 3c, d), and a few small differences in shape and size, which were mid-temperature transition products at a high cooling rate. Optical micrographs showed that the HAZ contained fewer and smaller M/A islands as compared to the BM. Additionally, it was found that the grain sizes of the two simulated HAZs were larger than the others. The CGHAZ was somewhat coarser than the FGHAZ owing to the high heating peak temperature, causing greater grain coarsening. The TEM images show that the two simulated HAZs presented lathshaped structures and their dislocation density was lower than that of BM due to the recrystallization during thermal cycling (Fig. 4a, c, d). Figure 5 shows the corresponding microhardness values of the four specimens. The two simulated HAZs had lower nanoindentation hardness values (FGHAZ (2.61 GPa), CGHAZ (3.34 GPa)) than the WM (3.56 GPa) and BM (3.65 GPa). This may be attributed to the softening effect due to recrystallization in the HAZ during thermal cycling, which caused a large loss in the work hardening produced during the thermomechanical control process (TMCP). In addition, the hardness of the FGHAZ was significantly lower than that of the CGHAZ, potentially because the CGHAZ was formed at a higher temperature near the fusion zone, which contributed to a higher concentration of dissolved non-metallic and metallic impurity atoms. As the cooling rates during thermal cycling in this study did not differ substantially, more hard phases could remain in the CGHAZ grains. Although the WM also experienced a certain degree of softening, the magnitude of hard phases precipitated and the grain size of the WM were

Fig. 2

Shape and size of the tensile specimen (thickness: 1 mm)

Fig. 3

Optical microstructure of X80 weld joint: (a) BM; (b) WM; (c) FGHAZ; (d) CGHAZ

Journal of Materials Engineering and Performance

Volume 27(5) May 2018—2515

Fig. 4

TEM morphology of X80 welded joint: (a) BM; (b) WM; (c) FGHAZ; (d) CGHAZ

relatively small, which was beneficial to improving the hardness.

specimens. Instead, after the yield point, the tensile curves differed significantly with and without H-charging as shown in Fig. 6(a), (b), (c), and (d). The main reason may be explained as follows (Ref 28): As compared to the elastic-deformation stage, after the yield point, the plastic-deformation-induced dislocation multiplications could have acted as high rate transportation channels for hydrogen atoms to diffuse into crack tips under the applied stress (Ref 29). Moreover, the rapid increase in local stress triaxiality after necking could have increased the local hydrogen solubility and promoted hydrogen-atom absorption in the steel subsurface (Ref 30). The absorbed hydrogen atoms could lead to crack initiation at the specimen surface by decreasing the material surface energy (Ref 31). Thus, an increasing number of hydrogen atoms could diffuse to the cracks during SSRT, leading to further initiation and propagation of cracks. This could finally lead to accelerated brittle-fracture characteristics. For HE susceptibility, considering the specimen size effect on the relative measurement error, the HE index was calculated as follows:

3.2 SSRT

HE ¼

Fig. 5 Nanoindentation hardness of four specimens from the X80 weld joint

SSRT is a widely used test methods to characterize HE. Hence SSRT was conducted on the four specimens with and without H-charging to obtain the relationship between the microstructure and HE. The stress–strain curves are shown in Fig. 6, and the corresponding tensile data are presented in Table 2 and Fig. 7. The mechanical parameters of the four specimens, such as ultimate tensile strength (UTS), elongation (EL), and reduction of area (RA), decreased, especially for the two simulated HAZs. For example, in Fig. 7, the UTS of the FGHAZ decreased by 26.52%, CGHAZ by 11.43%, BM by 6.1% and WM by 7.61%, indicating the degenerative effect of hydrogen to the mechanical properties. However, it appears that hydrogen had no obvious effect on the elastic-deformation stage according to the coincidence curve section of all

2516—Volume 27(5) May 2018

ELO  ELH  100% ELO

ðEq 2Þ

where ELH and ELo are the strain elongations obtained with and without H-charging, respectively. Figure 7 shows that all four specimens all presented a relatively high HE index, which may be attributed to the significant poisoning effect of NH4SCN and H2SO4. It can also be seen that the microstructure of the weld joint was more susceptible to hydrogen than that of the BM. For example, the FGHAZ had the highest HE index (90.67%), followed by WM (86.2%), CGHAZ (84.91%) and BM (63.58%).

3.3 Fracture Surface SEM fracture morphologies of the four specimens with and without H-charging are shown in Fig. 8. The fracture mor-

Journal of Materials Engineering and Performance

Fig. 6

Stress-strain curves of the four X80 welded-pipe specimens with and without H-charging: (a) BM; (b) WM; (c) FGHAZ; (d) CGHAZ

Table 2 Tensile data of the four tension specimens in X80 welded joint UTS, MPa

BM WM FGHAZ CGHAZ

EL, %

RA, %

Air

H2

Air

H2

Air

H2

HE, %

672 684 592 594

631 632 435 527

21.8 18.05 24.87 15.04

7.94 2.49 2.32 2.27

42.9 35.51 58.83 31.69

15.9 3.59 5.18 3.3

63.58 86.2 90.67 84.91

Fig. 7 HE index and loss ratio of UTS curves after hydrogen charging for the four specimens

phology of the BM tested in air presented obvious ductile characteristics with numerous small dimples (Fig. 8a), and a subtle secondary crack was seen in the central region surrounding the inclusions. Besides, an obvious necking zone can be found from the macroscopic fracture in the upper left corner of Fig. 8(a). Conversely, the H-charged BM specimen con-

Journal of Materials Engineering and Performance

tained somewhat smaller dimples (Fig. 8b), and the necking zone was also relatively smaller, indicating that its ductile characteristics were not as not obvious, and that the BM was somewhat susceptible to hydrogen. Additionally, a large hole of approximately 80 lm of size appeared to surround an inclusion. The fracture-edge area of the WM tested in air also presented typical dimples (Fig. 8c). However, both the dimples and necking zone were smaller, and it contained more inclusions of larger size, causing a larger crack as compared with the BM. The H-charged specimen also presented a few holes and cracks around the inclusions, and some small flat facets can be found (Fig. 8d). These were obvious brittle features and may indicate that the WM was more susceptible to hydrogen than the BM, corresponding well to the HE index in Table 2. The FGHAZ tested in air had the most evident ductile characteristics with a large dimple-patterned area and an obvious necking zone (Fig. 8e). The dimple size was large, and the absence of large secondary cracks implied that the airtested FGHAZ had good plasticity and toughness properties. Conversely, the air-tested CGHAZ fracture surface showed smaller and shallower dimples, and a smaller necking zone. A few big cracks and flat facets had occurred on both simulated specimens (Fig. 8g). The fracture-edge surface was also flat overall, showing poor plasticity and toughness properties as

Volume 27(5) May 2018—2517

Fig. 8 SEM images of the fracture surfaces without H-charging: (a) BM; (c) WM; (e) FGHAZ; (g) CGHAZ. Images with H-charging: (b) BM; (d) WM; (f) FGHAZ; (h) CGHAZ

2518—Volume 27(5) May 2018

Journal of Materials Engineering and Performance

Fig. 9 EDS spectra of inclusions marked by arrows. (a), (c) are the morphologies of BM, FGHAZ; (b), (d) are the corresponding EDS spectra of the inclusions

compared with the FGHAZ. However, fracture surfaces of the two simulated HAZ microstructure specimens with H-charging exhibited similar quasi-cleavage morphologies and almost flat cross-fractures. Additionally, compared with the uncharged specimen, a few big cracks and flat facets were present on both the simulated HAZ specimens (Fig. 8f, h), indicating HE tendency in accordance with the HE index in Table 2. Inclusions in the four specimens had a similar composition. Figure 9 shows the composition of typical BM and FGAHZ inclusions as examples. Both these inclusions primarily contained Fe, Ti, Al, C, O, Si, etc., which can act as an important hydrogen traps (Ref 32). A large hole can be seen around the inclusion which is shown in Fig. 9(a), implying that a high hydrogen concentration had existed in its vicinity. The kernel average misorientation (KAM) map, which represents the numerical misorientation average of a given pixel with its neighbors, was mainly used to characterize the local strain distribution. The KAM is proportional to the microstrain caused by crystal defects, such as dislocations (Ref 33). It is well known that local strain can produce a local stress field, and it has also been reported that stress concentration can create a driving force for hydrogen diffusion and redistribution around stress fields (Ref 34), which is further facilitated by the moving dislocations during SSRT. Therefore, local strain can affect the HE susceptibility to some degree. As shown in Fig. 10(a), (b), (c), and (d), the WM and simulated HAZ specimens had a larger fraction of high KAM than the BM, indicating that they had a greater degree of local high strain distribution than the BM. Furthermore, the high strain in the FGHAZ was more uniformly distributed, while the strain distribution in the

Journal of Materials Engineering and Performance

CGHAZ was more concentrated due to the abnormal grain growth, resulting in an uneven stress distribution. This could indicate that the CGHAZ had a greater degree of local stress concentration than the FGHAZ.

4. Discussion The HE susceptibility of steels is closely related to the interaction between absorbed hydrogen and microstructural features, such as phase type, grain boundaries and inclusions (Ref 32). Therefore, microstructural analyses, combined with EBSD maps and fracture morphologies, were used to investigate the mechanism influencing the HE susceptibility of the four microstructures. In addition, the effect of hardness on crack initiation and propagation was also analyzed. The microstructure type and size are very important to the mechanical properties and HE susceptibility of metallic materials. The BM was mainly composed of PF and GB with fine grains and contained high density of dislocations in the AF. High dislocation density can impede the crack propagation (Ref 35), which can make the BM have high toughness inherently. In addition, dislocations can act as hydrogen traps: Hydrogen atoms can be absorbed by dislocations to increase the hydrogen concentration, which may imply that it can potentially increase its HE index to some degree; however, high-density dislocations in the AF can also decrease the hydrogen diffusion rate to decrease the HE tendency (Ref 36). Considering the uniform distribution of AF or so-called hydrogen in BM, it can avoid the

Volume 27(5) May 2018—2519

Fig. 10

Kernel average misorientation (KAM) maps obtained from EBSD: (a) BM; (b) WM; (C) FGHAZ; (d) CGHAZ

over-enrichment of hydrogen in local areas. So itÕs hard to reach the critical amount of hydrogen for cracks initiation (Ref 37). Besides, itÕs reported that the effect of hydrogen on HE is mainly attributed to the diffusible hydrogen rather than the indiffusible hydrogen (Ref 31), which implies the great importance of hydrogen diffusivity to HE. During the SSRT, hydrogen atoms are difficult to diffuse to the crack tips due to low hydrogen diffusion rate. Therefore, the fine structure with high dislocations density in the BM exhibited the low HE index. The WM mainly consisted of small intra-granular nucleated acicular ferrites (IAFs), which were intertwined with each other and presented a woven fabric-like structure. The IAF had high strength but was also highly brittle as seen in Table 2, resulting in the WM having an inherent tendency toward brittle fracture. Furthermore, the dislocation density in the WM was somewhat lower than in the BM (Fig. 4a, b, c), which was consistent with previous literature (Ref 12). This could also influence the HE susceptibility by affecting the hydrogen solubility and diffusivity. Both the FGHAZ and CGHAZ consisted of large GB and BF. Coarse grains have high HE susceptibility owing to their high diffusive hydrogen concentration and locally enriched hydrogen (Ref 38), which can reduce the grain boundary cohesion force through HEDE (Ref 15). Additionally, a threshold hydrogen concentration is important in inducing HIC in terms of both hydrogen and metallurgical factors, and the local concentration of hydrogen at some threshold level is necessary to initiate cracking (Ref 37, 39). For example, the CGHAZ in X80 has a lower critical hydrogen concentration for HIC than the BM (Ref 28). Thus, it is easier to make diffusible hydrogen atoms reach the critical

2520—Volume 27(5) May 2018

concentration for HIC initiation and propagation by HEDE (Ref 15) under similar test conditions. Therefore, structures with high local hydrogen concentration may have a greater probability for inducing crack initiation and propagation. Figure 3 shows that the average grain size increased in the following sequence: WM, BM, FGHAZ, CGHAZ. Therefore, from a size perspective, the two simulated HAZs exhibited somewhat higher HE susceptibility than BM and WM. In Table 2, it can be clearly seen that without H-charging, the elongation of the FGHAZ was higher than that of the CGHAZ, which was mainly attributed to its excellent mechanical properties due to a smaller, more uniform structure and softening effect. However, with H-charging, the reasons for the greater reduction in elongation of the FGHAZ as compared to that of the coarser CGHAZ were unclear. Namely, it was unclear why the FGHAZ exhibited a higher HE index than the CGHAZ. One reason may be that grain size was one of several factors affecting the HE of the two simulated HAZs and not the decisive one. This question will be analyzed in greater detail in the section addressing the microhardness effect. Inclusions are an important factor affecting the hydrogen diffusion and crack propagation rate. Inclusions are usually considered as hydrogen traps in steels (Ref 40), which can trap diffusible hydrogen atoms and exert hydrogen pressure by forming hydrogen molecules, and then develop a stress field (Ref 34). Additionally, defects like microvoids and microcracks, combined with their high stress concentration, tend to become the preferential sites for cracks initiation and propagation (Ref 41). Figure 9 shows numerous inclusions with Fe, Al, Ti, C, O, Si, etc. According to previous papers, hydrogen-

Journal of Materials Engineering and Performance

induced crack initiation and propagation are often linked to the interface between inclusions enriched with Si or Al-oxide and the steel matrix, as these inclusions are brittle and incoherent to the metal matrix (Ref 42). In addition,numerous microvoids existing at the interface may also enhance the HE susceptibility. By analyzing the quantity and distribution of the four specimens, it was determined that the WM and simulated HAZ had more inclusions owing to the decomposition of martensite during welding, which may be a factor in the relatively high HE index in weld joints as compared to the BM. The WM had the largest inclusions (Fig. 8c), and a large crack had emerged around these inclusions, potentially implying that larger size inclusions induce easier crack initiation and propagation owing to the high trapped hydrogen concentration and resulting stress concentration. It is well known that the local stress field can affect the hydrogen diffusion rate and redistribution state. According to the HEDE mechanism (Ref 15), hydrogen can embrittle such interfaces with high local stress by reducing the energy for interfacial decohesion and facilitating the initiation and propagation of hydrogen-induced cracks in high-stress areas. In Fig. 10(a), (b), (c), and (d), the WM and simulated HAZ specimens had higher stress distribution than BM, especially in the HAZ (CGHAZ). Local stress triaxiality can increase local hydrogen solubility (Ref 30), implying that the HAZ and WM had higher local hydrogen concertration. This was consistent with the 3D modeling result (Ref 43) of hydrogen distribution in X80 weld joints, which showed that HAZ had the highest hydrogen concentration, followed by the WM and BM. Therefore, this implied that the simulated HAZ or WM had greater tendency for a higher local hydrogen concentration. Furthermore, it was found that these local high strains were mainly distributed near partial grain boundaries or defects like voids, dislocations and cracks. For example, the local strain in BM was mainly distributed in the boundaries near the AF due to its high dislocation density. These local high-stress brittle regions like grain boundaries and defects can trap more hydrogen atoms, which had been proven through atomic simulations (Ref 44). This revealed that strong hydrogen-atom trapping regions map directly to regions of high stress. Thus, the brittle regions with high hydrogen concentration tend to become preferential sites for crack nucleation, and cause severe HE. Additionally, from Fig. 10, the high strain in the FGHAZ is distributed more uniformly, while the distribution of CGHAZ was more concentrated due to the abnormal grain growth, resulting in the local stress concentration. This may indicate CGHAZ had a greater degree of local stress concentration than the FGHAZ, resulting in a larger local stress field, which could induce more diffusible hydrogen atoms to accumulate near these areas. According to the above analysis, when the diffusible hydrogen atoms accumulated at the trapping sites and reached the critical value, HIC initiated and propagated through HELP (Ref 14). Namely, CGHAZ tended to induce the cracks initiation and propagation more easily with H-charging. With H-charging, more uniformly distributed stress led to more uniformly distributed hydrogen. This may help explain the somewhat reduced HE tendency of the FGHAZ as compared to the CGHAZ. Hardness is the measure of a solid to resist shape change when a force is applied and is dependent on strength, plasticity, ductility, etc. A few researchers have studied the effect of hardness on HE behavior. Azevedo et al. (Ref 23) suggested that SOHIC propagation is sensitive to the

Journal of Materials Engineering and Performance

microstructure and regions with minimum microhardness can become preferential crack-nucleation sites. Takahashi et al. (Ref 45) also held that the cracking susceptibility depends on the minimum microhardness rather than the maximum hardness of X65 or less; the local region with minimum microhardness, having lower yield strength, is deemed to become the preferential site for hydrogen-induced blister cracking; regions with minimum microhardness can also provide plastic zones through which the existing blisters are prone to linking with each other, leading to the cracking. Li et al. (Ref 22) reported that HIC resistance is related to the microstructure, and that regions with minimum microhardness, such as a softened HAZ, can become preferential sites for crack nucleation. Kim et al. (Ref 46) also suggested that a soft microstructure is more sensitive to HIC than a hard one, because more cracks form in soft microstructures in steels with the same oxide inclusion level. This may indicate that when other conditions, such as the defect state, are similar, the nucleation and propagation of microcracks is relatively easy in the soft zone under the combined action of hydrogen and stress. Thus, cracks in soft ferrite are easier to propagate than in hard bainite or martensite. In Fig. 5, the hardness of the FGHAZ is significantly lower than that of CGHAZ. Considering their similar inclusion levels, this may imply that the occurrence of microcracks was somewhat easier in the FGHAZ during SSRT with H-charging. This can be seen directly from Fig. 6(a), (b), (c), and (d). Under the same test conditions, yielding occurred first in the FGHAZ; the yield point of the FGHAZ was approximately 390 MPa, which was lower than that of the CGHAZ (410 MPa), BM (450 MPa), and WM (530 MPa). Generally, softer microstructures tend to produce plastic deformation more easily as compared to harder ones. This was reflected in the elongation comparison between the FGHAZ and CGHAZ without H-charging in Fig. 6(c) and (d). Plastic deformation is accompanied by a large number of dislocations and microcracks initiation and propagation, which can act as high-speed channels for hydrogen transportation. Hydrogen atoms can be transported quickly along the dislocations to the crack tips under concentrated stress (Ref 29), followed by more hydrogen atoms accumulating at the tips and reaching the critical hydrogen concentration for cracks. This implies that a soft microstructure may enhance the HE susceptibility by improving the hydrogen diffusion rate and inducing further initiation and propagation of dislocations and microcracks during SSRT. Therefore, a soft microstructure can cause the FGHAZ to exhibit significantly higher HE susceptibility than a hard microstructure. According to the analysis above, the influence of microhardness discussed here may help explain the question reflected in Table 2, i.e., what was the reason for the large difference in elongation between the two HAZ microstructures (FGHAZ and CGHAZ) after H-charging? Namely, why did the FGHAZ exhibit a higher HE index than the CGHAZ? It can be seen that the two HAZs had similar microstructures with GB and BF, but relatively large differences in grain size, stress distribution state, and microhardness. In addition, from the analysis of the grain-size influence and stress state, a smaller size and more uniformly distributed stress can cause the FGHAZ to exhibit a relatively low HE index. The lower hardness resulted in the FGHAZ exhibiting higher HE susceptibility than the CGHAZ due to the influence of the hydrogen diffusion rate and induced the further nucleation and propagation of microcracks under the combined action of hydrogen and stress. Therefore, the higher

Volume 27(5) May 2018—2521

HE index of FGHAZ may primarily be attributed to its soft microstructure as compared to the CGHAZ.

5. Conclusions 1. Two HAZ microstructure specimens were thermally simulated. The HE index of X80 weld joints decreased in the following sequence: FGHAZ, WM, CGHAZ and BM. The HE susceptibility was due to the combined effect of microstructure, local stress distribution and microhardness. 2. The BM had the highest HE resistance owing to its fine structure with high-density dislocations and low local stress distribution. The WM and the two simulated HAZs had relatively high HE index owing to their brittle microstructure (WM: IAF; HAZ: coarse BF, PF), relatively more distributed inclusions and high local stress distribution. 3. Compared to the CGHAZ, the FGHAZ had a lower microhardness and more uniform stress distribution. The higher HE index of the FGHAZ was mainly attributed to its soft microstructure. Thus, it can be concluded that hardness can affect the nucleation and propagation of microcracks and influence the hydrogen diffusion rate under the action of hydrogen and stress.

Acknowledgments This work was financially supported by the Natural Science Foundation of China (No. 51671215).

References 1. A.J. Slifka, E.S. Drexler, N.E. Nanninga, Y.S. Levy, J.D. McColskey, R.L. Amaro, and A.E. Stevenson, Fatigue Crack Growth of Two Pipeline Steels in a Pressurized Hydrogen Environment, Corros. Sci., 2014, 78, p 313–321 2. K. Shen, L. Xu, Y. Guo, J. Shi, and M. Wang, Effect of Microstructure on Hydrogen Diffusion and Notch Tensile Strength of Large Steel Forging, Mater. Sci. Eng. A, 2015, 628, p 149–153 3. X.F. Li, Y.F. Wang, P. Zhang, B. Li, X.L. Song, and J. Chen, Effect of Pre-Strain on Hydrogen Embrittlement of High Strength Steels, Mater. Sci. Eng. A, 2014, 616, p 116–122 4. X. Zhu, W. Li, T.Y. Hsu, S. Zhou, L. Wang, and X. Jin, Improved Resistance to Hydrogen Embrittlement in a High-Strength Steel by Quenching–Partitioning–Tempering Treatment, Scripta Mater., 2015, 97, p 21–24 5. V. Olden, A. Alvaro, and O.M. Akselsen, Hydrogen Diffusion and Hydrogen Influenced Critical Stress Intensity in an API, X70 Pipeline Steel Welded Joint- Experiments and FE Simulations, Int. J. Hydrogen Energy, 2012, 37, p 1147–1148 6. M. Zrilic, V. Grabulov, Z. Burzic, M. Arsic, and S. Sedmak, Static and Impact Crack Properties of a High-Strength Steel Welded Joint, Int. J. Press. Vessels Pip., 2007, 84, p 139–150 7. I.A. Bataev, A.A. Bataev, V.G. Burov, Y.S. Lizunkova, and E.E. Zakharevich, Structure of Widmanstatten Crystals of Ferrite and Cementite, ISSN 0967-0912, Steel Trans., 2008, 38, p 684–687 8. Y.D. Han, H.Y. Jing, and L.Y. Xu, Welding Heat Input Effect on the Hydrogen Permeation in the X80 Steel Welded Joints, Mater. Chem. Phys., 2007, 84, p 139–150 9. L.W. Wang, Z.Y. Liu, Z.Y. Cui, C.W. Du, X.H. Wang, and X.G. Li, In Situ Corrosion Characterization of Simulated Weld Heat Affected Zone on API, X80 Pipeline Steel, Corros. Sci., 2014, 85, p 401–410

2522—Volume 27(5) May 2018

10. J. Lin, N. Ma, Y. Lei, and H. Murakawa, Measurement of Residual Stress in Arc Welded Lap Joints by cosa X-ray Diffraction Method, J. Mater. Process. Technol., 2017, 243, p 387–394 11. H.B. Xue and Y.F. Cheng, Hydrogen Permeation and Electrochemical Corrosion Behavior of the X80 Pipeline Steel Weld, J. Mater. Eng. Perform., 2013, 22, p 170–175 12. W.M. Zhao, T.M. Zhang, Y.J. Zhao, J.B. Sun, and Y. Wang, Hydrogen Permeation and Embrittlement Susceptibility of X80 Welded Joint Under High-Pressure Coal Gas Environment, Corros. Sci., 2016, 111, p 84–97 13. J.M. Pardal, S.S.M. Tavares, B.A.R.S. Barbosa, F.B. Mainier, J.S. Corte, and J.P. Pardal, Investigation of Hydrogen Embrittlement Failure in a Steam Separator by Field Metallography, Eng. Fail. Anal., 2013, 35, p 46–53 14. C.D. Beachem, A new Model for Hydrogen-Assisted Cracking, Metall. Trans., 1972, 3, p 437–451 15. A.R. Troiano, The Role of Hydrogen and Other Interstitials in the Mechanical Behavior of Metals, Trans. ASM, 1960, 52, p 54–80 16. C. Zapffe and C. Sims, Hydrogen Embrittlement, Internal Stress and Defects in Steel, Trans. ASM, 1941, 145, p 225–271 17. F. Mohammadi, F.F. Eliyan, and A. Alfantazi, Corrosion of Simulated Weld HAZ of API, X80 Pipeline Steel, Corros. Sci., 2012, 63, p 323– 333 18. H. Aydin and T.W. Nelson, Microstructure and Mechanical Properties of Hard Zone in Friction Stir Welded X80 Pipeline Steel Relative To Different Heat Input, Mater. Sci. Eng. A, 2013, 586, p 313–322 19. S.P. Trasatti, E. Sivieri, and F. Mazza, Susceptibility of a X80 Steel to Hydrogen Embrittlement, Mater. Corros., 2005, 56, p 111–117 20. W.G. Zhao, W. Wang, S.H. Chen, and J.B. Qu, Effect of Simulated Welding Thermal Cycle on Microstructure and Mechanical Properties of X90 Pipeline Steel, Mater. Sci. Eng. A., 2011, 528, p 7417–7422 21. E. Fallahmohammadi, F. Bolzoni, G. Fumagalli, G. Re, G. Benassi, and L. Lazzari, Hydrogen Diffusion into Three Metallurgical Microstructures of a CeMn X65 and Low Alloy F22 Sour Service Steel Pipelines, Int. J. Hydrogen Energy, 2014, 39, p 13300–13313 22. X.F. Li, J. Zhang, Y.F. Wang, M.M. Ma, S. Shen, and X.L. Song, The Dual Role of Shot Peening in Hydrogen-Assisted Cracking of PSB1080 High Strength Steel, Mater. Des., 2016, 110, p 602–615 23. C.R.F. Azevedo, Failure Analysis of a Crude Oil Pipeline, Eng. Fail. Anal., 2007, 14, p 978–994 24. K.W. Andrews, Empirical Formulae for the Calculation of Some Transformation Temperatures, J Iron Steel Inst., 1965, 203, p 721–727 25. W. Zhao, Y. Zou, K.J. Matsuda, and Z.D. Zou, Corrosion Behavior of Reheated CGHAZ of X80 Pipeline Steel in H2S-Containing Environments, Mater. Des., 2016, 99, p 44–56 26. T. An, H.T. Peng, P.P. Bai, S.Q. Zheng, X.L. Wen, and L. Zhang, Influence of Hydrogen Pressure on Fatigue Properties of X80 Pipeline Steel, J. Hydrogen Energy, 2017, 42, p 15669–15678 27. X.Y. Zhang, H.L. Gao, and Z.Y. Bi, Effect of Welding Thermal Cycle to Microstructure and Properties of X80 Welded Steels, J. Mater. Sci. Technol., 2009, 17, p 159–163 28. T.M. Zhang, W.M. Zhao, Q.S. Deng, W. Jiang, Y.L. Wang, Y. Wang, and W.C. Jiang, Effect of Microstructure Inhomogeneity on Hydrogen Embrittlement Susceptibility of X80 Welding HAZ Under Pressurized Gaseous Hydrogen, Int. J. Hydrogen Energy, 2017, 42, p 25102–25113 29. N.E. Nanninga, Y.S. Levy, E.S. Drexler, R.T. Condon, A.E. Stevenson, and A.J. Slifka, Comparison of Hydrogen Embrittlement in Three Pipeline Steels in High Pressure Gaseous Hydrogen Environments, Corros. Sci., 2012, 59, p 1–9 30. W.Y. Chu, L.J. Qiao, J.X. Li, Y.J. Su, Y. Yan, Y. Bai et al., Hydrogen Embrittlement and Stress Corrosion Cracking, Science Press, Beijing, 2013 31. I. Moro, L. Briottet, P. Lemoine, E. Andrieu, C. Blanc, and G. Odemer, Hydrogen Embrittlement Susceptibility of a High Strength Steel X80, Mater. Sci. Eng. A, 2010, 527, p 7252–7260 32. I. Maroef, D.L. Olson, M. Eberhart, and G.R. Edwards, Hydrogen Trapping in Ferritic Steel Weld Metal, Int. Mater. Rev., 2013, 47, p 191–223 33. None, Hydrogen embrittlement and stress corrosion cracking, Brit. Corros. J., 2013, 20, p 158–158 34. R. Kirchheim, Revisiting Hydrogen Embrittlement Models and Hydrogen-Induced Homogeneous Nucleation of Dislocations, Scripta Mater., 2010, 62, p 67–70

Journal of Materials Engineering and Performance

35. G.T. Park, S.U. Koh, H.G. Jung, and K.Y. Kim, Effect of Microstructure on the Hydrogen Trapping Efficiency and Hydrogen Induced Cracking of Linepipe Steel, Corros. Sci., 2008, 50, p 1865–1871 36. M.C. Zhao, Y.Y. Shan, F.R. Xiao, K. Yang, and Y.H. Li, Investigation on the H2S- Resistant Behaviors of Acicular Ferrite and Ultrafine Ferrite, Mater. Lett., 2002, 57, p 141–145 37. T.Y. Jin, Z.Y. Liu, and Y.F. Cheng, Effect of Non-Metallic Inclusions on Hydrogen-Induced Cracking of API5L X100 Steel, Int. J. Hydrogen Energy, 2010, 35, p 8014–8021 38. K. Takasawa, R. Ikeda, N. Ishikawa, and R. Ishigaki, Effects of Grain Size and Dislocation Density on the Susceptibility to High-Pressure Hydrogen Environment Embrittlement of High-Strength Low-Alloy Steels, Int. J. Hydrogen Energy, 2012, 37, p 2669–2675 39. H.Y. Liou, R.I. Shieh, F.I. Wei, and S.C. Wang, Roles of Microalloying Elements in Hydrogen Induced Cracking Resistant Property of HSLA Steels, Corrosion, 1993, 49, p 389–398 40. D. Hardie, E.A. Charles, and A.H. Lopez, Hydrogen Embrittlement of High Strength Pipeline Steels, Corros. Sci., 2006, 48, p 4378–4385

Journal of Materials Engineering and Performance

41. F. Huang, J. Liu, Z.J. Deng, J.H. Cheng, Z.H. Lu, and X.G. Li, Effect of Microstructure and Inclusions on Hydrogen Induced Cracking Susceptibility and Hydrogen Trapping Efficiency of X120 Pipeline Steel, Mater. Sci. Eng. A, 2010, 527, p 6997–7001 42. Z.Y. Liu, X.G. Li, C.W. Du, G.L. Zhai, and Y.F. Cheng, Stress Corrosion Cracking Behavior of X70 Pipe Steel in an Acidic Soil Environment, Corros. Sci., 2008, 50, p 2251–2257 43. C.Y. Yan, C.Y. Liu, and Y. Bo, 3D Modeling of the Hydrogen Distribution in X80 Pipeline Steel Welded Joints, Comp. Mater. Sci., 2014, 83, p 158–163 44. X. Zhou and J. Song, Effect of Local Stress on Hydrogen Segregation at Grain Boundaries in Metals, Mater. Lett., 2017, 196, p 123–127 45. A. Takahashi and H. Ogawa, Influence of Microhardness and Inclusion on Stress Oriented Hydrogen Induced Cracking of Line Pipe Steels, Trans. Iron Steel Inst., 1996, 36, p 334–340 46. W.K. Kim, H.G. Jung, G.T. Park, S.U. Koh, and K.Y. Kim, Relationship between Hydrogen-Induced Cracking and type I, Sulfide Stress Cracking of High-Strength Linepipe Steel, Scripta Mater., 2010, 62, p 195–198

Volume 27(5) May 2018—2523