Effect of heat input on microstructure and mechanical properties of GH159 and GH4169 dissimilar joints by laser beam welding

Co-based superalloy GH159 and Ni-based superalloy GH4169 have been successfully joined by laser beam welding and the effect of heat input on microstructure and properties of dissimilar joints were investigated systematically. The results showed that weld seams exhibited a nail shape and full penetration was attained at all dissimilar joints. Increasing grain size towards the fusion zone (FZ) were observed in the heat affected zone (HAZ) on the GH159 side while an increasing dissolution of γ′ and γ′ towards the FZ was attained at the HAZ on the GH4169 side. These resulted in decreasing microhardness towards FZ. Tensile failure was found in the FZ with the lowest microhardness. Meanwhile, the ultimate tensile strength (UTS) of the dissimilar joints increased with the decreasing of heat input. The high UTS of dissimilar joint with low heat input can be ascribed to the lower volume fraction of the Laves phase and the smaller dendrite arm spacing.


Introduction
One of the most vital and crucial components in the aero-engines is the high pressure rotor bolt assembly which exerts as a connector between the turbine disc and the shaft. It consists of spring plates and bolts, and the permanent connection between them is achieved by laser beam welding (LBW). The requirements for bolt assembly material are to provide sufficient mechanical strength, good creep and oxidation resistance, especially for long-term service at temperatures of 650°C. All these demands can be met by the GH4169 alloy and thus it has been used in the practice production.
GH4169 is a precipitation-hardened Ni-based superalloy that is widely used in the aerospace industry due to its outstanding mechanical strength, good corrosion resistance and weldability [1][2][3][4]. However, recent studies have shown that the main strengthening phase γ′ coarsens and gradually transforms into δ phase after long-term thermal exposure of GH4169 at 650°C, finally giving rise to its properties degradation [5,6]. In addition, with the development of aerospace industry, it is of importance to develop the high-strength and large-size bolt assembly. To meet these requirements, GH159 of deformation-hardened cobalt-based superalloy is used in the aviation field. After deformation, its tensile strength can reach 1916 MPa at room temperature, and the tensile strength still remains at 1564 MPa at 595°C [7][8][9]. Considering the cost, GH4169 is still used for the spring plate material, which involves the LBW of dissimilar materials.
The main challenges that occur during the LBW of GH4169 are the formation of brittle phases, cracks in the fusion zone (FZ) or heat affected zone (HAZ), and porosity [10][11][12][13][14][15][16][17][18][19]. According to the available literature, these metallurgical problems are unavoidable and can only be improved. There are scanty reports on the LBW of GH159. These challenges become manifold when attempting to weld dissimilar materials due to the big divergence of chemical composition and thermo-physical parameters, which may tend to aggravate the issues. Karadge et al [20] attempted to achieve the jointing between single crystal CMSX4 and Ni-based superalloy RR1000 and pointed out the difficulties of welding two different Ni-based alloys with very different structures.
The change of weld microstructure was also briefly mentioned, but the mechanical properties of the weld joints were not given. Ramkumar et al [21] obtained the optimal process parameters for LBW by conducting iterative tests at different welding speeds on Inconel 718 and AISI 416. The tensile failure was attained at the weakest base metal AISI416 despite the presence of a small number of microstructural defects (porosity or cracks) in the FZ or HAZ. Kuo et al [22] demonstrated that compared with continuous LBW, pulsed LBW of SUS 304L stainless steel and Inconel 690 superalloy could effectively reduce the porosity in weld joints.
It is obvious from the open references that most of the research works describes on the LBW of dissimilar joints involving weldability evaluation, defect formation mechanism and improvement and welding parameters optimization. But, little information is available on LBW of GH159 and GH4169. Moreover, LBW is the last step in the production process of bolt assembly, so no post-weld heat treatment can be carried out to improve its mechanical properties. The selection of welding parameters becomes especially important, which determines the final mechanical properties of the bolt assembly. Therefore, the purpose of this paper is to evaluate the weldability of LBW of GH159 and GH4169 by comprehensively characterizing the evolution of the macrostructure, microstructure and mechanical properties of dissimilar joints under different heat inputs. Moreover, the relationship between heat input, microstructure and mechanical properties is further discussed.

Material and methods
The chemical compositions of GH159 and GH4169 and the welding parameters are shown in tables 1 and 2, respectively. The component parts of the bolt assembly, bolts and leaf springs, have different heat treatments due to different requirements. Therefore, a solution heat treatment at 1050°C for 1h, 48% cold deformation and hot deformation were carried out on GH159 alloy before LBW. A solution heat treatment at 1050°C for 1h and double aging heat treatment (720°C for 8h, followed by furnace cooling at a cooling rate of 55°C h −1 to 620°C and maintained for 8h at 620°C followed by air cooling) were conducted on GH4169 alloy before LBW. The dimensions for the weld sample as shown in figure 1(a). In order to investigate the influence of heat input on the microstructure and properties of LBW dissimilar joints of GH159 and GH4169, the welding process was conducted using a continuous laser CO 2 welding machine.
The macrostructure and microstructure evolutions of weld joints were characterized by optical microscopy (OM) and a field emission gun scanning electron microscope (SEM). The Image-Pro Plus software was employed to quantitatively calculate the grain size and the volume fraction and size of precipitates. The samples for OM and SEM observation were prepared using mechanical grinding and polishing methods. Next, the samples were chemically etched in a mixed solution containing 5g CuSO 4 +100 ml HCl+5ml H 2 SO 4 [23]. The detailed microstructures were examined by transmission electron microscope (TEM) and TEM foils were prepared using standard techniques [24]. Bright field (BF) images, dark field (DF) images, and diffraction patterns were recorded using a TEM at an acceleration voltage of 200 kV to observe the fine microstructural details.
Non-standard micro-sized tensile specimens were prepared from the weld joints the dimensions of which are illustrated in figure 1(b). Similar micro-tensile specimens were also successfully used for the determination of tensile properties of diffusion bonded Ti-alloys and laser beam welded steel joints [25][26][27][28][29][30][31][32]. Tensile tests were performed using a single column desktop electronic universal testing machine equipped with a laser extensometer at a cross head speed of 1 mm min −1 at room temperature to evaluate the mechanical properties of various weld joints. To ensure the accuracy of the experimental results, each specimen was subjected to three instances of tensile testing. The hardness values of the specimens were obtained at a load of 1000 g for 10 s by using a Vickers hardness tester. The hardness test needs to start at 0.2 mm away from the weld surface and end at 0.2 mm away from the other side of the weld surface, in which test one point every 0.1 mm.

Results and discussion
3.1. Effect of heat input on macrostructure of weld joints Figure 2 shows the macrostructures of the dissimilar joints under different heat inputs. No welding defects such as cracks and metal splashes were observed in all the weld joint surfaces under the heat input of 35.2 J mm −1 ∼44.8 J mm −1 , as shown in figures 2(a)-(d). All weld seams exhibited a nail shape and the porosity was observed in the bottom of the weld seams as shown in figures 2(e)-(h)). The presence of porosity is mainly related to the high Ni content in the base metal [22]. In addition, the width and depth of the weld seam were measured and the results are shown in figure 3. As the heat input increased from 35.2 J mm −1 to 44.8 J mm −1 , the width and depth of weld seams increased from 2050 μm and 1769 μm to 2212 μm and 2038 μm, respectively. These results indicate the heat input has significant influences on the weld seam geometry.

Evolution of microstructure of weld joints
It can be clearly observed from figure 4 that the weld joint could be obviously divided into 5 zones, namely base metal (BM) GH4169, HAZ of GH4169 side (HAZ B), FZ, HAZ of GH159 side (HAZ A)and BM GH159. The phase composition in each area one by one was confirmed through the SEM and TEM analyses, as described in the following paragraphs.   Figure 5(a) shows the microstructure of BM GH4169 after solution treatment at 1050°C and double aging. SEM results indicate that BM GH4169 consisted of equiaxed grains and a few blocky particles in the grain boundaries. The blocky particles were determined to be MC carbides through the EDS analysis, as shown in figure 5(b). Similar results were also observed in Refs [17,33,34]. In addition to MC carbides, TEM images also show that nanometer precipitates were appeared in the BM GH4169 (figure 5(c)). Further confirmed by the selected area electron diffraction (SAED), the BM GH4169 contained strengthening phase γ′ and γ′ ( figure 5(d)). It can be seen from figure 5(e) that the γ′ precipitated phase was dispersefully distributed in the matrix. The situation is absolutely different in the microstructure of BM GH159 (figure 6(a)). Fine recrystallization grains (arrow 2), MC carbides and a much coarser grain with intersecting network of fine platelets (arrow 1) were observed in BM GH159. Further confirmed by SAED obtained along the most favorable orientation [011] [9], BF and DF images, the fine platelets were twins (figures 6(b)-(d)). GH159 is mainly strengthened by the cold deformation which forms intersecting network of fine twins. The intersecting network of thin twins, which function as 'cells' or 'subgrains' and divide the coarse grain into many small parts, which is equivalent to refining the coarse grain. Furthermore, it is difficult for dislocations to move long distances in this intersecting network of fine thin twins. Figure 7(a) depicts the microstructure of the HAZ A. Compared with the BM GH159, static recrystallization and the incremental grain size towards the FZ was clearly observed in HAZ A (The direction of the arrow in figure 7(a) is away from the FZ). Disappearance of cold deformation microstructure obviously indicates that the temperature during welding process near the weld line was more than the GH159 static recrystallization temperature of 920°C [35]. However, MC carbides were still present close to the weld line after welding (figures 7(a)-(b)), namely, MC carbides did not dissolve after welding. This reveals the temperature near the    weld line during welding was lower than the solvus temperature of MC carbides (∼1260°C) [36]. Consequently, the temperature near the weld line during welding process is estimated to be in the range from 920°C to 1260°C. But, this temperature range is higher than the solvus temperature of strengthening phases γ′ (910°C [37]) and γ′ (900°C [38]) in GH4169. Consequently, it is expected that the strengthening phases near the weld in HAZ B have been completely dissolved during the welding process. As expected, no strengthening phases γ′ or γ′ were found in the area close to the weld line (as shown in figures 7(c)-(d)). Consequently, it is inferred that the increasing dissolution of strengthening phases γ′ and γ′ towards the FZ was attained at in the HAZ B. This conclusion is also supported by the finding of Ye et al [39]. Figure 8(a) displays a typical SEM micrograph of dissimilar joints of GH159 and GH4169. The microstructure from the weld line to the weld center changed from dendritic structure to equiaxed grains as shown in figures 8(b) and (c). The irregular phases that appeared in the FZ were characterized by the SEM/EDS spectra analysis. Figure 8(d) verifies that irregular phases mainly contained Nb and Mo elements. The precipitates were further confirmed to be Laves phases by SEAD as shown in figure 8(f).

Microstructure of the FZ
The elemental distributions of the dissimilar joints are shown in figure 9. It is distinct that under the minimum and the maximum heat inputs the amounts of elements Ni, Fe and Nb decreased from the GH4169 side to the GH159 side. However, the change of the amounts of Co and Ti was the opposite. In particular, under the minimum heat input, the element on the interface increased abruptly, while under the maximum heat input, the element on the interface increased slowly. Therefore, it can be concluded that with the increasing of heat input, element diffusion increases. Figure 10 illustrates the dendritic microstructure of the FZ of different heat inputs. Obviously, the increasing of heat inputs caused a distinct increase in the dendrite arm spacing. To further determine the relationship between the heat inputs and the microstructure of the FZ, the dendrite arm spacing of the FZ was quantitative calculated. As shown in figure 11, the measured dendrite arm spacing increased from 2.72±0.62 μm to 3.94±0.63 μm as the heat input increased from 35.2 J mm −1 to 44.8 J mm −1 . The dendritic arm spacing is mainly affected by the cooling rate. The relationship between the cooling rate and the dendritic arm spacing can be described by equation (1)  where K is the dendrite arm spacing and R is the cooling rate. Equation (1) indicates that the cooling rate decreases with an increase in the dendritic arm spacing. Figure 12 represents the distribution of the Laves phase in the FZ under different heat inputs. It indicates that the number of Laves phases increased with the increasing of heat input. Particularly, the morphology of the Laves phase was semi-continuous in figure 12(c) and highly interconnected in figure 12(d). Figure 13 gives the quantitative result on size and volume fraction of the Laves phase in the FZ under different heat inputs. With the increasing of heat inputs, the size of Laves phase increased slightly. However, the volume fraction of Laves phase varies greatly with the heat input. As shown in figure13(b), the volume fraction of Laves phase was 1.86±0.15% (35.2 J mm −1 ), 2.88±0.09% (38.4 J mm −1 ), 4.93±0.31% (41.6 J mm −1 ), and 6.38±0.32% (44.8 J mm −1 ). The relationship between the heat input and the volume fraction of Laves phase can be explained as following: the increasing of the heat input due to the decreasing cooling rate corresponding to a high microsegregation. The extent of microsegregation with high heat input is higher that results in a large number of Laves phases.

Distribution of hardness across the profile of the weld joint
The above results indicate that the heat input has a significant effect on the microstructure of the dissimilar joint, which may lead to the change of mechanical properties. Figure 14 shows the microhardness changes of the dissimilar joints under different heat inputs and the average hardness values of each zone are listed in table 3. It is observed that the microhardness value of the whole weld joint decreased when the heat input increased at a same measuring position 0.2 mm from the weld surface in the weld joint, indicating that the heat input exerts a remarkable effect on the microhardness of the dissimilar joint of GH159 and GH4169. As can be observed from the curve in figure 14, the microhardness values decreased continuously when the measuring positions were gradually close to the center of the FZ. The lowest microhardness value in the whole dissimilar joint was acquired at the FZ. Thus, the following discussion mainly focuses on the relationship between heat input and FZ hardness. The microhardness of a material is generally defined as its resistance to plastic deformation. It can be seen from  precipitation depletes major strengthening elements Nb, Mo and Ti, thus weakening the γ matrix and making it soft. The volume fraction of Laves phase increased with increasing heat input and, consequently, a decrease in the content of elements Nb, Mo and Ti in the γ matrix. Moreover, the relationship between the microhardness of FZ and the secondary dendrite arm spacing can be explained by the Hall-Petch equation [39]: where C and K are material constants and λ is secondary dendrite arm spacing. From equation (2), it can be seen that the hardness value is inversely proportional to the secondary dendrite spacing. Namely, the hardness value decreases with increasing dendrite arm spacing. The results of dendrite arm spacing in section 3.2.3 show that when the heat input increased from 35.2 J mm −1 to 44.8 J mm −1 , and the dendrite arm spacing increased from 2.72 μm to 3.94 μm. Therefore, with increasing the heat input, the secondary dendrite arm spacing increases and the hardness value decreases. Obviously, the changes of microstructure in the FZ can serve as the compelling evidence of the changes of the microhardness in the FZ (see figures11 and 13).  The width of HAZ with different heat input could be estimated by hardness diagram distribution. The widths of HAZ B and HAZ A were 0.8 mm and 0.6 mm (35.2 J mm −1 ), 1.0 mm and 0.8mm (38.4 J mm −1 ), 1.5 mm and 0.9 mm (41.6 J mm −1 ), 1.5 mm and 0.9 mm (44.8 J mm −1 ), respectively. Obviously, the width of the HAZ B was larger than that of HAZ A and it could be attributed to the different thermal conductivity of two materials. The width of the HAZ can be roughly estimated using the following equation: where W HAZ is the width of the HAZ, a is the thermal conductivity of the material, t is heat conduction time. The thermal conductivity of GH159 and GH4169 is 11.0 W m −1 ·°C [40] and 14.7 W m −1 ·°C [41], respectively. Big divergence in the thermal conductivity can be conducted on two materials. Therefore, the width of the HAZ B of is larger than that of HAZ A. The continuously decreasing of microhardness values in the HAZs along FZ are mainly contributed to increasing dissolution of the strengthening phases γ′ and γ′ and the grain size along FZ.   Figure 15 depicts the engineering stress-strain curves and the corresponding mechanical properties are presented in , respectively. The UTS in dissimilar joints of GH159 and GH4169 under four heat inputs can meet the minimum requirements for using even without post-weld heat treatment. The elongations of all the dissimilar joints in time of tension tests which follow ASTM E8/8M standard were found to be above 15%. In addition, all the tensile fractures were observed in the FZ with the lowest microhardness. LBW is the last step in the production process of bolt assembly, so no post-weld heat treatment can be carried out to improve its mechanical properties. Heat input determines the final mechanical properties of the bolt assembly. Therefore, the following analysis mainly focuses on the relationship of heat input-UTS-FZ microstructure.

Tensile testing results
As we all know, chemical composition, grain size and precipitation of nanoscale γ′ and γ′ (solid solution strengthening, grain boundary strengthening and precipitation strengthening) have a remarkable effect on the strength of superalloys [39]. It can be seen from section 3.2.3 that there was no precipitation of strengthening phases γ′ and γ′ in the FZ, and the change of heat input only affects the dendrite arm spacing and the volume fraction of the Laves phase. Therefore, in this study, only the influences of solid solution strengthening and grain boundary strengthening on strength of FZ are considered [42,43]:  The resulting strength s s is the sum of the effects of the intrinsic strength s , 0 grain boundary strengthening s gb and solid solution strengthening s . ss Each of these contributions affects the strength in the FZ of the GH159 and GH4169 dissimilar joints. Thus, each microstructural feature will be discussed below to investigate the heat input-structure-property.
According to figure 11, the measured dendrite arm spacing increased from 2.72±0.62 μm to 3.94± 0.63 μm as the heat input increased from 35.2 J mm −1 to 44.8 J mm −1 . The dendrite arm spacing increases with the increasing of heat input. According to Hall-Petch equation, the contribution of grain size is: where D is the grain size, and κ is the Hall-Petch slope. According to equation (5), grain boundary strengthening is inversely proportional to grain size, namely, the effect of fine grain strengthening decreases with the increasing of grain size. Therefore, it can be concluded that with the increasing of heat input, the dendrite arm spacing increases and the fine grain strengthening effect become weaker. Laves phase is the main precipitation phase in the FZ, but it has no direct effect on the strength. Its precipitation depletes major strengthening elements Nb, Mo and Ti in the matrix. These elements are known to be as the most significant solid solution strengthening elements in superalloy [44]. Therefore, only the solid solution strengthening induced by these elements is considered in this work. In this study, the model established by Gypen is adopted because of its simplicity and widespread application in nickel-based superalloys [44][45][46]. The solid solution strengthening can be defined as where a i is the strengthening constant of atomic species i and c i is the atomic concentration of atomic species i.
is the modifying factor, which is used to account for the solid solution strengthening confined only to the matrix. The following equation can be obtained: 1 , Laves f Laves is the volume fraction of the Laves precipitates. According to equation (6), as the volume fraction of Laves phase increases, the strengthening effect of solid solution weakens. According to the quantitative statistical results in figure 14(b), the volume fraction of Laves phase was 1.86±0.15% (35.2 J mm −1 ), 2.88±0.09% (38.4 J mm −1 ), 4.93±0.31 (41.6 J mm −1 ), 6.38±0.32 (44.8 J mm −1 ), respectively. Laves phase increased with the increasing of heat input. Therefore, it can be concluded that the solid solution strengthening effect weakens with the increasing of heat input.
On the basis of this detailed microstructure study, each individual strengthening contribution in equation (1) has been discussed. It can be concluded that the decreasing of UTS in FZ could be attributed to the higher volume fraction of Laves phase and the larger dendrite size.

Conclusions
In summary, different heat inputs were applied to the LBW of dissimilar GH159 and GH4169 superalloys. The macrostructure and microstructure evolution were examined, the microhardness distribution and the tensile mechanical properties of dissimilar joints GH159 and GH4169 were tested. The main conclusions are as follows: (1) Weld seams exhibited a nail shape and full penetration was attained at the heat input from 35.2 J mm −1 to 44.8 J mm −1 . With the increasing of heat input, the width and depth of weld seams increased. Heat input has significant influences on the weld seam geometry.
(2) The microstructure result showed that FZ consisted of dendritic microstructure and Laves phase. The microstructure of GH4169 consisted of NbC, and the strengthening phases γ′ and γ′. But increasing dissolution of these two phases towards the FZ was attained at in the HAZ B. Static recrystallization and the incremental grain size towards the FZ was observed in the HAZ A.
(3) The width of the HAZ B was larger than that of HAZ A and it could be attributed to the different thermal conductivity of two materials. The significant reduction in the microhardness at the HAZs could be attributed to the dissolution of the main strengthening phases and the disappearance of intersecting network of fine twins.
(4) Tensile failures were occurred in the FZ with the lowest microhardness. With the decreasing of heat input, the UTS of the dissimilar joints increased. The high UTS of dissimilar joints with low heat input can be ascribed to the lower volume fraction of the Laves phase and the smaller dendrite arm spacing.