Microstructure and high temperature fracture toughness of NG-TIG welded Inconel 617B superalloy

1. Introduction

The ultra supercritical (USC) power plants, using the conventional ferritic or austenitic heat-resistant steels, are operated at the service temperature of approximately 600 ℃ and steam pressure of 25-30 MPa. In order to meet the rapidly growing needs for energy, the advanced ultra-supercritical (AUSC) technology with the operating temperature of 700-750 ℃ and steam pressures of 35 MPa would be widely used to improve the heat efficiency and thereby reduce the carbon emission [1]. Nevertheless, the conventional heat-resistant steels used in USC power plants cannot meet the AUSC requirements [1]. Inconel 617B becomes a primary candidate material for AUSC due to its superior resistance to oxidation and hot corrosion, good creep-rupture strength, as well as outstanding performance in machining and welding [[1], [2], [3]].

The turbine rotor is a key component for the steam power plant, and its reliability and stability have the decisive effect on the safety and service life of the steam turbine set. Because of the complex structure and large dimension of the turbine rotor, welding is the most proper technique to fabricate the large components rather than integrated casting or hot forging. Compared with the conventional welding methods, i.e. oxyacetylene welding (OAW), submerged arc welding (SAW), tungsten inert gas welding (TIG), metal inert-gas wielding (MIG), and plasma transferred arc welding (PTA) [4], narrow gap tungsten inert gas welding (NG-TIG) is the most promising method to manufacture the Inconel 617B welded joint by multi-pass welding for its high quality and formability. In the welded joints of Inconel 617B, the microstructures of weld metal (WM) and heat affected zone (HAZ) are usually distinct from that of base metal (BM) [5], which may affect the mechanical properties and reliability of welded joint. Therefore, the study of the mechanical properties of Inconel 617B and its weld joint is essential. Among the mechanical properties, fracture toughness, the critical indicator for damage tolerance design, is a prerequisite [6,7]. The service safety of Inconel 617B NG-TIG welded rotor requires a thorough understanding of the fracture behavior of its welded joint. However, there is few literature about the effect of microstructure on facture toughness of 617B NG-TIG welded joint at the designed service temperature. For Inconel 617B employed in AUSC conditions, fracture toughness and fracture behavior of NG-TIG welded joints should be critically investigated at the service temperature.

In the study, the 617B welded trial rotor fabricated by NG-TIG welding was prepared, and high temperature fracture toughness and fracture behavior of its welded joint were experimentally investigated. In order to probe the reliability and durability of welded joint, three parts, WM, HAZ, and BM, were selected to conduct the fracture test at 700 ℃. The J-R curves of WM, HAZ, and BM were established, and then J_0.2 was determined to the investigate resistance to crack initiation and propagation. The correlation between the microstructural and fractographic features was further discussed with the aid of electron backscatter diffraction (ESBD) and scanning electron microscope (SEM).

2. Material and methods

2.1. Materials and welding parameters

The material used in this study was a pair of rings of Inconel 617B, with an external diameter of 910 mm, a thickness of 160 mm, and a width of 140 mm. The chemical compositions of BM and filler metal (ERNICrCoMo-1) were listed in Table 1. TIG root welding and multi-layer NG-TIG were used to fabricate the 617B welded trial rotor. The welding voltage, welding current, and welding speed were 11 V, 245 A, and 80 mm/min, respectively. In order to release the welding residual stress, a post-weld heat treatment (PWHT) was conducted at 980 ℃ holding for 10 h. The welded joint was chemically etched with Kalling’s reagent (100 ml HCl+100 ml CH₃CH₂OH +5 g CuCl₂) for 150 s, and then observed with an optical microscope (OM: CX14) and a field emission scanning electron microscope (FE-SEM: TESCAN LYRA3) to analyze the microstructures.

2.2. Fracture toughness tests

In this work, fracture toughness of WM, HAZ, and BM was tested at 700 ℃ in terms of J-resistance by using a mechanical testing machine (MTS Landmark 1). During the whole tests, the temperature deviation between the actual value and setting value was controlled within 1 ℃. The compact tension samples were manufactured according to the test standard ISO 12,135 [8], and the sampling location and dimensions are shown in Fig. 1. The samples were fatigue pre-cracked at room temperature with the stress intensity factor decreasing method [8]. At least six samples were loaded to the specific displacement levels and then the secondary fatigue test was conducted to get the corresponding crack extension (Δa) [8]. The crack mouth opening displacement (CMOD) was measured by a clip gauge. After fracture toughness tests, the samples were broken with tension method to obtain the initial crack length (a₀) and Δa. The fracture appearances were observed by SEM and the crack profiles of the broken samples were observed by OM. The deformation microstructures near the crack in the ruptured samples were further analyzed by using an EBSD system (FE-SEM: JSM-7100F). The samples for EBSD observations were ground to 3000 grit and electropolished using a solution of ethyl alcohol (95 vol.%) and perchloric acid (5 vol.%) at 30 V for 30 s. The step size for the EBSD operation was 0.1 μm, and the raw data were processed using Orientation Imaging Microscopy software (TSL-OIM).

3. Results and discussion

3.1. Microstructure analysis

The microstructures of the NG-TIG welded joint were observed by OM, as shown in Fig. 2. BM and HAZ are composed of austenite with equiaxed grains. Ti (C,N) could be found in both BM and HAZ. WM mainly consisted of columnar crystals and grain size in WM is much bigger than that in BM and HAZ. The carbide morphologies in WM, HAZ, and BM were observed by SEM (Fig. 3, Fig. 4, Fig. 5). Two different morphologies of M₂₃C₆ phase, which were Cr-rich carbides, were observed in BM: the micro-scale M₂₃C₆ and small-scale M₂₃C₆ decorating the grain boundary, as shown in Fig. 5(b). The complex eutectic microstructure and M₂₃C₆ carbides were found in HAZ, as shown in Fig. 4(b) and (c). Furthermore, there were some nano-scale precipitations around the eutectic microstructure. In particular, a large amount of Mo-rich M₆C-carbides were found in WM, as shown in Fig. 3(b), which might be related to the complex welding thermal cycle and post-weld heat treatment. Moreover, lots of fine M₂₃C₆ carbides were also observed around the M₆C in WM. These results of microstructure were consistent with previous studies [5,9].

3.2. Fracture toughness

The J-R curves of WM, HAZ, and BM at 700 ℃ were shown in Fig. 6. J-Integral was calculated according to the following equations:

$J=\left[\frac{F}{(BBNW)^{0.5}}g_{2}(\frac{a_{0}}{W})\right]^{2}\cdot\frac{1-v^{2}}{E}+\frac{\eta_{p}U_{p}}{B_{N}(W-a_{0})}\cdot\left[1-\frac{(0.75\eta_{p}-1)\Delta a}{W-a_{0}}\right]$ (1)

$g_{2}(\frac{a_{0}}{W})=\frac{(2+\frac{a_{0}}{W})\left[0.884+4.64\frac{a_{0}}{W}-13.32(\frac{a_{0}}{W})^{2}+14.72(\frac{a_{0}}{W})^{3}-5.6(\frac{a_{0}}{W})^{4}\right]}{1-\frac{a_{0}}{W}}$ (2)

$\eta_{p}=2+0.522(1-a_{p}/W)$ (3)

where F (N) represents the applied force, B (mm) the thickness of sample, B_N (mm) the net thickness of sample between side grooves, W the width of sample, v the passion's ratio, E (GPa) elastic modulus at the testing temperature, U_p the area under the curve of force F versus load line displacement, and Δa (mm) the stable crack extension including blunting. The relationship between J-Integral and Δa can be fitted using the following power-law equation:

J=α+βΔa^γ (4)

where α, β, and γ are the material constant [8], and α and β≥0, 0≤γ≤n. The results are listed in Table 2 and Table 3. The intersection of the line of 0.2 mm offset from the blunting line J = 3.75R_mΔa and the J-R curve was determined as fracture toughness value (J_0.2). The results in Table 4 report the J_0.2 values of WM, HAZ, and BM. The typical force-CMOD curves obtained from the fracture toughness tests are shown in Fig. 7. As reaching the corresponding maximum force, BM showed larger CMOD value than the other materials did, indicating the significant ductile tearing of BM. It is another indicator of high toughness of BM associated with the low strength and high ductility.

3.3. Fractographic observations and fracture path analysis

Fig. 8 shows the typical fractograph of the broken sample, which reveals the eroded notch region, pre-fatigued crack region, crack extension region, secondary fatigue crack region, and quick fracture region. Specific attention was paid to the crack extension region, which composed of the stretch zone and followed propagation zone (Fig. 9, Fig. 10, Fig. 11). The stretch zone of WM exhibit cleavage fracture and the propagation zone was dominated by intergranular fracture along columnar crystals, as shown in Fig. 9, which is consistent with the lowest fracture toughness of WM. The quasi-cleavage fracture becomes the dominant failure mechanism in the stretch zone and propagation zone of HAZ, as shown in Fig. 10. Under the same testing conditions, the fracture surface in the stretch zone of BM began to show dimple characteristics, as shown in Fig. 11, indicating the better toughness of BM than WM and HAZ, as reported in Table 4. Moreover, almost the entire fracture surface in the propagation zone of BM was dominant by ductile dimples; this type of fracture apperance is another indicator of high ductility of BM.

In essence, the formation of the stretch zone is used to accommodate the desired plastic strains ahead of the crack propagation, and the stretch zone width (SZW) could be also used for obtaining fracture toughness [[10], [11], [12], [13]]. The intersection of the line of Δa=SZW and J-R curve was determined as fracture toughness value (J_str) [14] and the SZW was measured according to ISO 12,135 [5]. In this study, J_str was also obtained, as reported in Table 4. J_str of WM and HAZ was higher than their J_0.2, which indicates that J_0.2 might be stricter and safer than J_str to evaluate fracture toughness of Inconel 617B-WM and Inconel 617B-HAZ under the current test conditions. In contrast, J_str of BM is very close to its J_0.2, indicating that SZW method was applicable for predicting fracture toughness of Inconel 617B-BM.

Blunting angle can be regarded as an indicator of fracture toughness and smaller blunting angle indicates better fracture toughness [15]. Fig. 12 shows the crack profiles of the broken samples. BM suffered the severe plastic blunting around the crack tip (Fig. 12(c)), indicating its excellent fracture toughness. Blunting angle of HAZ (Fig. 12(b)) is slightly bigger than that of BM. For WM, the crack initiated with a small blunting deformation, and it is considered that the high strength and low ductility of WM (Fig. 7) restrained the blunting and opening deformation of crack tip. These observations show a good correspondence with the previous results of fracture toughness test.

3.4. Deformed microstructures and fracture behavior

After fracture, the microstructures of an area on the mid thickness plane of ruptured samples close to fracture surface were detected using EBSD. Fig. 13, Fig. 14, Fig. 15(a) display the inverse pole figure (IPF) maps; Fig. 13(b), 14 (b), and 15 (b) show the corresponding kernel average misorientation (KAM) maps. It is evident from these results that under the testing conditions, the strain distribution of WM is relatively homogeneous, compared with that of HAZ and BM. Strain localization for HAZ and BM mainly occurred at grain boundary. Fig. 13, Fig. 14, Fig. 15(c) display the misorientation distribution of the broken samples and the corresponding statistic results of misorientation angle (0-64°) are shown in Fig. 16. The fraction of large misorientation angle (>50°) of BM and HAZ is 41.6% and 24.4%, respectively, and no large misorientation angle (>50°) was found in WM. Further analysis shows that the average misorientation angles of WM, HAZ, and BM are 7.69592°, 20.3241° and 32.7517°, which indicates the distribution trend of misorientation angle on the whole. Figs. 13(d), 14 (d), and 15 (d) illustrate coincidence site lattice (CSL) ∑3 boundaries in WM, HAZ, and BM, respectively, where the black line boundaries represent random grain boundaries (RGBs) and red ones represent CSL ∑3 boundaries. Fig. 17 shows the statistic results of the CSL boundaries in the broken samples. The influence of the deformed microstructures on the fracture toughness will be discussed in more detail.

Previous study has shown that large misorientation (mostly>50°) could effectively hinder fracture propagation [16]. The crack is inclined to initiate and propagate at region of strain localization. In the study, strain of HAZ and BM mainly localised at grain boundaries with large misorientation (mostly>50°), which might lead to an inhibitory effect of cracking and the better fracture toughness of BM and HAZ than WM. Dislocation is the lattice defect which has adverse effects on fracture toughness, and the piling-up of dislocations is usually the initiation of crack [16]. The results of KAM could be linked with dislocations and the region with large KAM revealed a local high density of dislocations that might result in crack initiation [[16], [17], [18], [19]]. Compared with BM, HAZ contained more large KAM regions (Figs. 14(b) and 15 (b)), which had local high density of dislocations and high tendency of crack initiation in the process of deformation, thus resulting in the lower fracture toughness of HAZ than BM. Moreover, more twins, as the barrier of crack growth [20], were observed in BM than that in HAZ, leading to the higher fracture toughness of BM. The special boundaries may also play a role in the different fracture toughness of WM, HAZ, and BM. The grain boundary energy could be estimated according to geometrical matching of the boundaries, as described by the CSL theory. The grain boundary energy of the special boundaries with remarkable atomic fit is lower than that of RGBs [21]. Grain boundaries with the low ∑ values in the CSL framework are known to have the better resistance to cracking than RGBs with higher energy [21]. It is indicated that most of the CSL boundaries in HAZ and BM were ∑3 boundaries, with the fraction being 14.9% and 28.8% for HAZ and BM, respectively. In contrast, very few grain boundaries with low ∑ values were observed in WM and no ∑3 boundary was even found. Therefore, BM and HAZ exhibited the better fracture toughness compared with WM. Moreover, fraction of ∑3 boundaries in BM was higher than that in HAZ, resulting in the more outstanding fracture toughness of BM. CSL Σ3ⁿ (n = 1, 2, 3) boundaries are related to twins [22]. No twin was observed in WM, while there were many twins in BM and HAZ (Figs. 13(a), 14 (a), and 15 (a)). As expected, the number of twin in BM was higher than that in HAZ.

4. Conclusions

The fracture behavior of Inconel 617B NG-TIG welded joint was investigated at 700 ℃ in this study. The correlation between microstructure and fracture toughness was discussed. The following conclusions were drawn:

(1) The J_0.2 values of WM, HAZ, and BM were 138.09 kJ m², 225.21 kJ m², and 293.36 kJ m², respectively, which implied that BM had the best fracture resistance, HAZ took the second place, and WM was the weakest part of the whole welded joint.

(2) The high fracture toughness of BM and HAZ was related to strain localization occurring at grain boundaries with misorientation greater than 50° and large amounts of CSL ∑3 boundaries inside grains.

(3) More twins, higher fraction of CSL ∑3 boundaries (28.8%), and less serious strain localization were responsible for the better resistance of crack initiation and growth of BM than HAZ.

(4) These findings may provide guidance for future research aimed at improving the high temperature fracture toughness of Ni-based alloy welded joint.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Project No. 51775300); Shanghai Turbine Company, Shanghai, China and the State Key Laboratory of Tribology, Beijing, China.

---