Microstructure Evolution and Toughening Mechanisms in the Nugget Zone of Friction-Stir-Welded X80 Pipeline Steel

Abstract

Coarse grains and blocky M-A constituents are often generated in the heat-affected zone (HAZ) of fusion welded pipeline steel joints owing to high heat input, causing a significant deterioration of toughness. This study demonstrated the effect of heat input in friction stir welding (FSW) on the microstructure and toughness of the nugget zone (NZ), elucidating the microstructure evolution and toughening mechanism. The results revealed a marked reduction in effective grain size within the NZ at low heat input (LHI) and a significant increase in the ratio of the refined M-A constituent. Furthermore, the decreased heat input leads to weak texture components (D1, D2, and F) accompanied by a decrease in the kernel average misorientation (KAM) value. This microstructural optimization clearly enhances toughness, and an excellent toughness value of 200 J, representing 95.3% of the basal metal (BM), was achieved in the NZ at LHI. The primary reason for this improvement is the refinement of effective grains and M-A constituents resulting from reduced heat input. During crack propagation, the high proportion of effective grain boundaries and fine M-A constituents acts as a barrier, arresting and deflecting cracks and thereby enhancing toughness.

1. Introduction

As the global economy rapidly develops and urbanization advances, the need for energy sources, including oil and natural gas, continues to rise [1,2]. At present, the transportation and distribution of oil and natural gas mainly rely on pipelines because of their safety, economy and convenience [3,4,5]. Welding is the most important process in the field construction of oil and gas pipelines. However, coarse grains and blocky martensite–austenite (M-A) constituents are often generated in the coarse-grained heat-affected zone (CGHAZ) due to the high heat input (HHI), which reduces the toughness of joints [6,7,8]. In addition, coarse network-like M-A constituents are easily formed in the inter-critical CGHAZ (ICCGHAZ) during multi-pass welding, which further deteriorates joint toughness [9,10,11]. Li et al. [12] pointed out that the toughness in the ICCGHAZ of double-pass submerged arc welding (SAW) X100 steel joint decreased by 72.2% compared to the base metal (BM) on account of the generation of coarse grains and network-like M-A constituents. Therefore, a new welding technology with low heat input (LHI) is urgently needed to inhibit grain coarsening and the development of blocky and network-like M-A constituents, enhancing the toughness of welded joints.

Friction stir welding (FSW), as a significant advancement in solid-state joining, can effectively retard the coarsening of microstructure and M-A constituent in the HAZ owing to its LHI, improving the toughness [13,14]. In contrast to HAZ, the microstructure in the nugget zone (NZ) is markedly coarsened because the high deformation resistance of steel materials makes the peak temperature higher during FSW. Therefore, many studies have tried to refine the microstructure and enhance the mechanical properties of the NZ by changing the welding parameters or applying cooling medium [15,16,17]. Barnes et al. [15] conducted FSW on X65 pipeline steel at different welding speeds, and pointed out that the microstructure in the NZ changed from coarse granular bainite (GB) to fine martensite + lath bainite (LB) as the welding speed increased. Cui et al. [16] found that reducing HI in the NZ of FSW X100 steel produced fine LB, and the toughness of the NZ reached 92% of the BM. Furthermore, Duan et al. [18] reported that fine LB, GB, and ferrite were obtained in the NZ of FSW X80 steel by the peak temperature into the (α + γ) region. Clearly, the microstructural characteristics of the NZ are predominantly determined by the thermal cycle process and plastic deformation, which also has a considerable impact on toughness. However, existing research on FSW pipeline steels has concentrated on the microstructural characteristics and toughness of the NZ, while the relationship between welding parameters, thermal cycle process and microstructure evolution has been ignored.

In this work, the X80 pipeline steel was subjected to FSW at various HIs, and the relationship among welding parameters, thermal cycle history and microstructure characteristics was established. Furthermore, the microstructure characteristics of the NZs under various parameters and its influence on toughness were systematically studied.

2. Experimental Procedure

The API 5 L X80 pipeline steel (180 mm × 80 mm × 6 mm) was used as the BM.

Table 1. Chemical composition (wt. %) of X80 pipeline steel [17].

Chemical composition	Fe	C	Si	Mn	P	S	Nb	V
	Balance	0.038	0.2329	1.611	0.0098	0.017	0.0535	0.0029

shows the chemical composition of X80 pipeline steel [17]. FSW was conducted on X80 pipeline steel using two parameter sets: 600 rpm with 100 mm/min (designated HHI) and 400 rpm with 100 mm/min (designated LHI). The FSW process employed a W-25Re alloy tool featuring a 15 mm shoulder diameter, a 9 mm pin root diameter, and a 5 mm pin length. During FSW, the thermal cycle history of the NZs was measured, and t_8/5 represents the average time of cooling from 800 to 500 °C, as shown in Figure 1a.

Figure 1b shows the sample collection position of the microstructure and impact specimens. The optical microscope (OM, Axio Imager M2m, Carl Zeiss Microscopy GmbH, Jena, Germany) and scanning electron microscope (SEM, Zeiss Gemini 300, Oberkochen, Germany) samples were corroded with 4% Nital. Electrolytic polishing was applied to prepare the back-scattered diffractometer (EBSD, Oxford Instruments Symmetry, Oxford, UK) and transmission electron microscope (TEM, FEI Tecnai G2-F20, Thermo Fisher Scientific, Waltham, MA, USA) specimens, using12.5% HClO₄ + 87.5% C₂H₆O solutions respectively. The morphological features of M-A constituents were quantified using three key parameters: W_max (maximum width), L_max (maximum length), and their aspect ratio (L_max/W_max). Image-pro Plus software （Version 9.x） was used to measure the size and volume ratio of the M-A constituents. Figure 1b reveals the sampling position of the impact samples, and the impact toughness was tested at 20 and −40 °C.

3. Results and Discussion

3.1. Microstructural Evolution in the NZs

Figure 2 presents the OM images of the BM and NZs under various HIs. The microstructure of the BM mainly consisted of fine acicular and polygonal ferrite. At HHI, coarse GB was obtained in the NZ, while fine GB and LB were observed at LHI. It was obvious that the varying HI has a significant impact on the phase constituents and grain size of the NZ. Furthermore, the thermal history information under various HIs was measured, as shown in Figure 3. The peak temperature and t_8/5 of the NZs at HHI and LHI was 1005 and 935 °C, and 1.6 and 9.5 s, respectively. This indicates that FSW with LHI can effectively reduce the peak temperature and increase the post-weld cooling rate. Analogously, Xie et al. [19] noted that the post-weld cooling rate in the NZ of FSW DP780 steel increased as the rotation rate decreased. In our previous study [18], the A_c3 and A_c1 temperatures of the BM were measured and found to be 874 and 672 °C, respectively. Clearly, the peak temperature of the NZs under various HIs was higher than that of the A_c3, resulting in complete austenitization of the BM. At HHI, the high peak temperature promoted the growth of austenite grains, and the slow post-weld cooling rate led to the transformation of coarse austenite into coarse GB during subsequent cooling. By comparison, the low peak temperature at LHI can inhibit the growth of austenite grains, and a high post-weld cooling rate was beneficial for increasing the undercooling degree, ultimately leading to the generation of fine LB and GB.

Figure 4 reveals the SEM images and M-A constituent characteristics of the NZs under various HIs. It can be observed that the variation in HI has a significant impact on the features of the M-A constituents. At HHI, coarse M-A constituents preferentially precipitated along prior austenite grain (PAG) boundaries, in contrast to the fine M-A constituents that uniformly dispersed within the GB at LHI. To carefully analyze the effect of HI on the feature of the M-A constituent, its characteristics were classified into four types, namely, island-like (I, L_max/W_max < 3, L_max < 3 μm), fine slender (II, L_max/W_max > 3, L_max < 3 μm), coarse slender (III, L_max/W_max > 3, L_max > 3 μm), and blocky (IV, L_max/W_max > 3, L_max < 3 μm) M-A constituents, as shown in Figure 4c and d. At HHI, the volume ratios of blocky (19.8 ± 1.9%) and coarse slender (21.3 ± 1.7%) M-A constituents were obtained, while the volume ratios of island-like (57.5 ± 2.6%) and fine slender (21.1 ± 2.3%) M-A constituents were achieved at LHI. This may be attributed to the fact that the partitioning and diffusion of carbon were retarded owing to the combined effects of reduced peak temperatures and fast post-weld cooling rates. It is well known that dislocations and substructures can serve as diffusion channels for carbon atoms [18]. However, the introduction of dislocations and substructures under LHI was inhibited due to weak plastic deformation, further inhibiting the partitioning and diffusion of carbon. As a result, a large number of fine slender and island-like M-A constituents were observed in the NZ at LHI. Clearly, the peak temperature, cooling rate and plastic deformation have a pronounced impact on the feature of M-A constituent during FSW high strength pipeline steel.

To further clarify the effect of HI on the M-A constituent, EBSD analysis and TEM observation were utilized, as shown in Figure 5. At HHI, the diffraction spots are symmetrically distributed, and this indicates that the microstructure is a body-centered cubic structure. Combined with the measured interplanar spacing, it was identified as twinned martensite. In contrast, at LHI, two sets of diffraction spots (body-centered cubic and face-centered cubic) were generated in the NZ, indicating the presence of austenite within the M-A constituent. It can be found that twin martensite was formed in the M-A constituent in addition to the retained austenite. Furthermore, it can be seen that retained austenite is mainly distributed at the edges of the M-A constituent, which was attributed to the austenitic volume constraint and the slow diffusion of carbon at the austenite/ferrite interface, exhibiting higher stability [20,21]. From the SEM and Band contrast + phase images, the volume fractions of retained austenite and M-A constituent at HHI and LHI were 0.08 ± 0.009% and 0.18 ± 0.021%, and 17.8 ± 2.3% and 13.6 ± 1.4%, respectively. Thus, the austenite/martensite ratios at HHI and LHI were 0.5% and 1.3%, respectively. Obviously, at HHI, numerous prior-austenite grains transformed into twinned martensite, resulting in a decrease in the average carbon concentration within the austenite. Thus, the austenite stability of the NZ at HHI was poor, leading to a low volume fraction of austenite/martensite.

Figure 6 illustrates the EBSD analysis results of the NZs under various HIs, in which the yellow and black lines representing low- and high-angle boundaries (LABs, 2°  ≤  θ  <  15°; HABs, 15°  ≤  θ; θ-misorientation angle), respectively. HABs are generally defined as effective grain boundaries that impede crack propagation. [22,23]. The average effective grain size of the NZs was 12.7 ± 1.2 and 9.3 ± 0.7 μm at HHI and LHI, respectively. Meanwhile, the ratios of HABs in the NZs at HHI and LHI were 13.4 and 21.2%, respectively. Clearly, the ratio of HABs increased as the HI decreased, which is closely related to the refinement of the microstructure. Interestingly, the ratio of LABs was higher and reached 86.6% and 78.8% at HHI and LHI, respectively. By contrast, it was reported by Xue et al. that the NZ of FSW low-alloy steels contained a low proportion of LABs, approximately 5% [24]. In this work, the microstructure coarsening resulted in a reduction in the number of HABs, causing an increase in the ratio of LABs. In addition, a vast quantity of dislocations and substructures were generated during the post-weld cooling, further enhancing the ratio of LABs.

Texture evolution under various HIs is illustrated using orientation distribution function (ODF) maps. Figure 7a–c presents the body-centered cubic (bcc) texture component at the φ₂ = 0° and φ₂ = 45° section of Euler space during the transformation from austenite to ferrite. At HHI, D1, D2, and F shear texture were obtained in the NZs, while only D and D2 shear texture were observed at LHI. Compared with LHI, the NZ experienced high peak temperature and strong plastic deformation at HHI. Therefore, it is speculated that strong thermal–mechanical coupling is beneficial to the generation of F shear texture. Furthermore, the texture intensity was highest at HHI. Figure 7d–f reveals the kernel average misorientation (KAM) maps of the NZs under various HIs, reflecting dislocation density and local crystal orientation changes in the material. It can be found that the KAM value of the NZ was the highest at HHI, which was consistent with the texture evolution. It is well known that deformed textures and dynamic recrystallization (DRX) textures were easily generated in intense thermomechanical processing, which appear alternately [18,25]. However, no DRX texture was observed in the NZs despite the strong thermoplastic deformation occurring during FSW. Abbasi et al. [26] noted that in addition to undergoing DRX at elevated temperature, the NZ also underwent continuous deformation during continuous cooling transformation. Thus, it was deduced that the formation of D1, D2, and F texture components is attributed to the bainite transformation. Furthermore, Yasavol et al. [27] conducted a study of FSW AISI D2 tool steel and pointed out that the texture intensity decreased as the ratio of LABs decreased. In this work, at LHI, the NZ, including the lowest proportion of LABs, revealed a weak texture component. Clearly, the texture evolution is complex, and it depends heavily on plastic deformation and phase transformation.

3.2. Toughness Variation in the NZs

Figure 8 reveals the impact energy of the BM and NZs under various HIs. The room-temperature and low-temperature toughness at HHI and LHI were 188 and 172, 200 and 192 J, respectively, reaching 89.5% and 86.3%, 95.3% and 98.5% of the BM, respectively. It can be found that the toughness of the NZ at LHI was markedly enhanced. In contrast, Xie et al. [28] observed a significant toughness reduction in gas metal arc welding (GMAW) of X80 steel, with the joint toughness reaching only 81% that of the BM. Clearly, compared to the toughness of fusion-welded joints, the FSW joint exhibited higher toughness due to its low heat input. Figure 9 shows the impact fracture and main crack propagation paths of the NZs under various HIs. Numerous shallow dimples were obtained in the NZ at HHI, suggesting a predominantly brittle fracture. At LHI, deep dimples were observed in the NZ, indicating ductile fracture and improved toughness.

Previous studies [29,30,31] have indicated that the grain boundary characteristics and the shape and size of the M-A constituent are the main factors influencing toughness. The effective grain boundary with a high misorientation angle can effectively inhibit the straight propagation of cracks because the crack needs to consume more energy to cross the grain boundary. The tortuous crack path observed at LHI (Figure 9d) resulted from the crack inhibition and deflection by HABs, thereby effectively contributing to toughening. By comparison, the crack propagation path of the NZ at HHI was straighter (Figure 9b), resulting in toughness deterioration. In addition to effective grain boundaries, the M-A constituent also plays a critical role in determining the toughness [30]. It is well known that high stress concentration is easy to occur at coarse M-A constituent/matrix interface, and the critical stress required for microcracking is low. On the contrary, the fine M-A constituent is less susceptible to stress concentration, and the critical stress required for microcracking is high. Figure 9d illustrated that the crack propagation path was deflected at the island-like M-A constituent, indicating that more energy was required for the crack to propagate. Similarly, Duan et al. [18] attributed the tortuous crack path in the NZ of FSW high-strength pipeline steel to crack deflection by the fine M-A constituent. Obviously, the toughness of the NZ can be significantly improved by tailoring the size and shape of the M-A constituent.

4. Conclusions

(1)

Coarse granular bainite (GB) appeared in the NZ at high heat input (HHI), while fine GB and lath bainite (LB) were observed in the NZ at low HI (HHI). Meanwhile, there was a marked reduction in effective grain size within the NZ at LHI, and a significant increase in the ratio of refined M-A constituent was observed.

(2)

Different types of shear textures such as D1, D2, and F were generated in the NZ under various HIs. Furthermore, the decreased HI leads to weak texture component accompanied by a decrease in the kernel average misorientation (KAM) value.

(3)

An excellent toughness value of 200 J, representing 95.3% of the basal metal (BM), was achieved in the NZ at LHI. This is attributed to the fact that a high proportion of effective grain boundaries and fine M-A constituents can effectively arrest and deflect the crack propagation, enhancing the toughness.