Search for articles:
Jian Yang Zhang
, Dian Zhong Li
Corresponding authors:
Received: 2019-07-21
Revised: 2019-08-21
Accepted: 2019-08-23
Online: 2020-03-01
Copyright: 2020 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology
More
Abstract
Plastic deformation bonding (PDB) has emerged as a promising solid state bonding technique with limited risk of phase transformations and residual thermal stresses in the joint. In this study, the PDB behavior of IN718 superalloy was systematically investigated by performing a series of isothermal compression tests at various processing conditions. It was revealed that, with increasing PDB strain rate at 1000 °C, different extents of dynamic recrystallization (DRX) occur in the bonding area of IN718 joints. The extent of DRX, average size of DRXed grains, and a newly proposed “interfacial bonding ratio (ΨBonding)” parameter (to quantify the bond quality) were initially reduced with increase in the strain rate up to 0.1 s-1 and later increased at further higher strain rates. Electron backscattered diffraction (EBSD) and transmission electron microscopy (TEM) based interfacial microstructure analyses indicated that the quality of the bonded joints is closely related with the development of fine DRXed grains at the bonding interface with the increasing strain, which promotes adiabatic temperature rise. It was revealed that the initial bulging and subsequent migration of the original interfacial grain boundary (IGB) were the main mechanisms promoting DRX in the well bonded IN718 superalloy joints. Moreover, the mechanical properties of the bonded joints were not only controlled by the recrystallized microstructure but also depended upon the Bonding parameter of the joints.
Keywords：
Inconel 718 (IN718) superalloy is widely used in manufacturing key components of aircraft engines, marine gas turbines and land-based gas turbines [[1], [2], [3]], due to their high strength, high ductility, excellent corrosion resistance and high fatigue resistance [[4], [5], [6]]. Traditional solid state joining methods such as diffusion bonding (DB) [7], friction stir welding (FSW) [8], linear friction welding (LFW) [9], accumulative roll bonding (ARB) [10], etc., have been widely used in the forming process of above applications. However, formation of increasingly complex parts requires an effective joining process to meet the demands of industrial scale production. As an emerging solid-state bonding technique, plastic deformation bonding (PDB) is considered as an advanced and highly reliable metallurgical joining process wherein the welded joints are thermomechanically compressed to achieve significant plastic deformation of the bonding interface [[11], [12], [13]]. In recent years, PDB has been successfully applied to titanium and aluminum alloys to improve microstructure of the bonded joints viz. excellent joint performance, high bonding efficiency, and reduced manufacturing time [11,14,15]. In contrast to conventional fusion welding processes, solid state PDB does not involve the formation of a fusion zone (which usually changes the initial as-cast microstructure) in the joint and no post-weld heat treatment is required [12,16,17]. The PDB process involves high temperature severe plastic deformation which promotes metallurgical bonding of the joint due to the increased contact area of the joining interface [11,16]. As the deformation temperature and strain increase, the original bond line is removed due to continuous migration of interfacial grain boundaries (IGBs) [13,18,19]. Hence, PDB technique provides a viable approach to manufacture or repair high performance Ni-based superalloy components with minimal risk of undesired phase changes in the joint.
It is a well-known fact that during PDB, severe plastic deformation results in dynamic recrystallization (DRX) of the microstructures in highly deformed regions. Generally, the recrystallization behavior of the interfacial grains has a significant influence on the migration of IGBs in the bonding area, which in turn affects the mechanical properties of the bonded joint [19]. Tang et al. [20] studied the effect of recrystallization on the interfacial grains and resulting shear strength of TiAl alloy joints. They suggested that the recrystallization of the initial microstructure in the bonding area improves the strength of the joint by changing the mode of failure. It is a well-known fact for metallic materials that the occurrence of recrystallization is often related to deformation parameters such as temperature, degree of deformation, and strain rate [21]. Many studies have shown that in order to obtain the most favorable microstructures and mechanical properties, the range of deformation temperature for IN718 superalloy should be selected from 950 °C to 1020 °C [22]. Liu et al. [23] indicated that, in order to avoid excessive growth of the grains and a large amount of δ precipitates, the deformation temperature is generally slightly higher than the δ phase dissolution temperature (~ 980 °C). On the other hand, Zhang et al. [24] studied the compressive deformation behavior of IN718 superalloy below 900 °C. The results indicated that a mixture of complex (γ′′ and δ) phases is precipitated in the IN718 superalloy matrix, which undoubtedly increase the complexity of the PDB process. Obviously, for solid state PDB, the effect of temperature is limited due to the small window of processing temperature. Moreover, IN718 is very sensitive to hot deformation/thermomechanical processing conditions, in particular, the strain rate strongly affects the microstructure of the alloy [25,26]. Satheesh Kumar et al. [27] investigated the effects of strain rate on the microstructural evolution of the IN718 superalloy during hot deformation. In their work, the DRX behavior was found to be sluggish at intermediate strain rates, however, it was accelerated during hot deformation performed at lower as well as higher strain rates. They believed that the main reason behind the accelerated DRX was the increased heating effect caused by the adiabatic temperature rise. Zhang et al. [28] have performed a systematic research on the macro and micro behaviors of IN718 superalloy under electrically-assisted tension condition. Their results prove that due to the local heating effect, the extent of DRX is enhanced even though the deformation temperature remains much lower than the critical temperature of recrystallization. Under practical industrial environment, the most of the forgings are produced by the hydraulic forge pressing and/or forge hammering, which may result in initial strain rates of about 0.001 and 1.0 s-1, respectively [29]. Therefore, it is of paramount importance to systematically study the effect of strain rates (from 0.001 to 1.0 s-1) on recrystallization and bonding behavior of IN718 during PDB.
This paper aims to understand the recrystallization behavior of interfacial grains and migration of the original IGBs in IN718 superalloy under different PDB conditions. i.e. variable strain rate at a certain fixed deformation temperature (1000 °C). Isothermal compression bonding tests were implemented to physically simulate PDB process and resulting interfacial microstructure in the joint. The effect of different strain rates on microstructure and mechanical properties of IN718 joints was investigated in detail using EBSD, TEM and tension tests and mechanism of PDB was explained thereof.
In this study, commercial grade IN718 superalloy with chemical composition (wt %): Ni 53.35, Cr 19.10, Nb 5.18, Mo 3.02, Ti 0.98, Al 0.56, Co 0.04, C 0.04, and Fe bal., was selected as the starting material for PDB. The as-received IN718 superalloy was initially subjected to solid solution treatment by heating at 1100 °C for 1 h followed by air cooling to room temperature, which resulted in a uniform microstructure. Afterwards, PDB of the homogenized IN718 superalloy was accomplished by performing a series of isothermal compression bonding tests on Gleeble-3500 thermal simulator using two round bar-shaped specimens (as shown in Fig. 1(a)). The dimensions of round bar-shaped specimens and bond formation stages are shown schematically in Fig. 1(b). Before PDB, the contacting surfaces were polished to remove any oxides or impurities such as oil stains etc. Subsequently, the two contacting cylindrical specimens were joined together at different strain rates ranging from 0.001 - 1 s-1 at 1000 °C. In all PDB experiments, specimens were heated at a rate of 5 °C/s to the process temperature, and held for 5 min to eliminate any temperature gradient. The vacuum level inside the equipment was maintained at 2.2 × 10-3 Torr. After isothermal compression bonding test, the specimens were immediately quenched in cold water.
Fig. 1. (a) Schematic of isothermal compression bonding tests. (b) Shape and dimensions of the round bar-shaped specimen for isothermal compression bonding tests. (c) Area selected for microstructural characterization of interfacial grain boundaries (IGBs). (d) Location and dimensions of the tensile test specimen.
Each set of experiments was duplicated to obtain two bonded specimens for each condition. In order to observe the microstructure of the joints, one bonded sample was cut axially, as shown in Fig. 1(c). The deformed part (including the bonding interface) was selected for microstructural examination. In order to evaluate interface bonding ratio (ΨBonding) parameter in the deformed microstructure, seven metallographic images were taken sequentially at the central position of each specimen along the interface. For revealing fine microstructural details of the bonding interface, EBSD and TEM were performed. EBSD was conducted using an HKL Nordlysnano probe mounted on a ZEISS MERLIN Compact device. The kernel average misorientation (KAM) [30], describing the average misorientation spread between a reference pixel and its nearest neighbor pixels for a defined kernel size, was determined using an OIM Analysis software package [31]. The geometrically necessary dislocation (GND) density (ρGND) was calculated from the EBSD data using a strain gradient method [32]. For TEM observations, a disk specimen with a diameter of 3 mm and a thickness of 500 μm was carefully cut at the bonding interface. The interface bonding line was located at the center of the disk in the radial direction. The disk was then ground down to a thickness of 60 μm, and finally subjected to ion beam milling until an observable thin area was formed.
The second (duplicate) bonded specimen was used to perform tension tests. Before the tension tests, the as-deformed specimens were not heat treated. The tension test specimens were machined according to GB/T228-2010 standard by wire cutting [33]. The dimensions and exact location of collecting the tension test specimen are schematically shown in Fig. 1(d). Tension tests were performed at room temperature using an AG-100KNG machine at a strain rate of 3 × 10-3 s-1.
The evolution of the microstructure in the bonding area, under a deformation strain rate of 0.001 - 1 s-1 at 1000 °C, are shown in Fig. 2. It is quite clear that DRX takes place in the studied area as evidenced by the development of a large number of small grains in the vicinity of the deformed grains. It is worth noting that the bonding lines/interfaces, which were initially located at the center (highlighted with yellow lines in Fig. 2) of the specimens, are fully occupied with the small DRXed grains. Hence, it is believed that the disappearance of the bonding lines is closely related to the occurrence of DRX. Previous studies have shown similar behavior wherein the interface bonding of joint is associated with the migration of DRXed grain boundaries during hot deformation of copper [18] and nickel-based alloy [13].
Fig. 2. Optical microscopy images of IN718 joints at 1000 °C with a strain rate of (a) 0.001 s-1, (b) 0.01 s-1, (c) 0.1 s-1, and (d) 1 s-1. (The bonding zone is highlighted with yellow lines.).
Inverse pole figure (IPF) maps of the bonding interface at 1000 ℃ with different strain rates are shown in Figs. 3(a-d). Some un-recrystallized coarse grains, which are severely elongated along the deformation direction, are noticed in all samples. Furthermore, the deformed microstructure is characterized by a chain of fine pancake shaped grains evolved around the parent grains. The new grains emerge along the original grain boundaries, thus leading to the evolution of this so-called necklace-type microstructure [27]. The microstructure indicates that the preferred nucleation sites for recrystallized grains are the strained grain boundaries, which is a significant feature of the recrystallized IN718 surperalloy [27,34].
Fig. 3. Representative inverse pole figure maps of the bonding area of the joints obtained at 1000 ℃ with a strain rate of (a) 0.001 s-1, (b) 0.01 s-1, (c) 0.1 s-1, and (d) 1 s-1. Effects of deformation strain rates on (e) the volume fraction of different types of grains (recrystallized, substructured, and deformed) and (f) the average grain size at various strain rate levels.
According to the characteristic grains in Fig. 3(a-d), the distribution of different types of grains (i.e. recrystallized, substructured, and deformed) and the average size of overall grains and DRXed grains are presented in Fig. 3(e) and (f), respectively. It is obvious that as the strain rate gradually increases from 0.001 s-1 to 0.1 s-1, the volume fraction of the recrystallized grains gradually decreases from ~75% to a minimum value of ~ 12%, and then increases sharply to ~ 35% at strain rate of 1 s-1. Likewise, the average size of the recrystallized grains shows the same trend, i.e. the recrystallized grain size reduces from ~ 6 μm to a minimum of ~ 3 μm at a strain rate of 0.1 s-1, and then rapidly increases to ~ 4 μm for strain rate of 1 s-1. Generally, the extent of recrystallization decreases with an increase in the strain rate [35]. However, in the present study, the extent of recrystallization shows abnormal increase at a strain rate of 1 s-1. This discrepancy is caused by the friction (between mating surfaces) during PDB [29]. During the bonding process, if the strain rate is too high, local temperature rise (ΔT) may take place due to adiabatic heating, caused by the stamping speed of the forging equipment, or local preferential material flow thereof [29]. Hence, the strain rate of 1 s-1 is considered as relatively high, while adiabatic heating cannot be neglected. The adiabatic temperature rise is estimated using the following expression [27]:
$ΔT=\frac{ησε}{ρC_p} $ (1)
where ΔT is the rise in temperature, σ and ε are measured true stress and true strain, respectively, η is the fraction of deformation energy converted into heat which is assumed to be 0.95, ρ is the density (kg/m3) and Cp (kJ/mol) is the specific heat capacity at deformation temperatures. The temperature rise induced by adiabatic heating is found to be as high as 22 °C at 1 s-1 whereas the temperature rise is less than 18 °C at lower strain rates. The availability of higher thermal energy, due to adiabatic temperature rise at higher strain rates, facilitates accelerated nucleation for DRX in conjunction with enhanced IGB migration. Previous studies have shown similar behavior wherein the kinetics of DRX are accelerated at higher strain rates during hot deformation of austenitic stainless steels [36] and nickel superalloys [37].
In order to quantitatively evaluate the extent of bonding of the joints, the ΨBonding parameter was proposed in this experiment. The ΨBonding is defined as the total original length of the segments that have bulged into opposite grains to the total length of the observable interfacial region, as demonstrated in Fig. 4(a). The effect of strain rate on the ΨBonding of the joints is shown in Fig. 4(b). For a strain rate of 0.001 s-1, the ΨBonding is estimated to be ~ 65%, which is progressively decreased to ~ 38% for 0.1 s-1, however, the ΨBonding further increases to a higher value of ~ 50% when the strain rate is increased to 1 s-1. The above results indicate that the strain rate has a significant influence on the bonding behavior of the joint during PDB. It is speculated that the sharp rise in ΨBonding, at a strain rate of 1 s-1, is related to the increase in the extent of DRX caused by localized adiabatic temperature rise.
Fig. 4. (a) Demonstration for calculating interfacial bonding ratio. (b) Effect of strain rate on interfacial bonding ratio (ΨBonding) in the bonding area.
3.2.1. DRX at the bonding interface
Fig. 3(a-d) shows dissimilar microstructures (with respect to the extent of DRX) in the bonding zones of all samples which seems to be influenced by the strain rate, without altering recrystallization mechanism. Therefore, in order to further clarify the relationship between DRX and interfacial bonding, it is logical to study the evolution of interfacial microstructure under different plastic deformations performed at a certain strain rate. Fig. 5(a-d) shows typical IPF maps of the deformed microstructure that evolved at the bonding interface under different deformation strains at a fixed strain rate of 0.01 s-1. Fig. 5(e) and (f) show evolution of volume fraction of different types of grains and average grain size at various strain levels, respectively. As the strain increases, DRX initially take place at the bonding interface and then gradually spreads towards the boundaries of the individual (parent) grains. The volume fraction of newly formed recrystallized grains was estimated to increase from ~ 3% to ~ 25% (Fig. 5(e)). The DRXed grain size gradually increases from ~ 7.5 μm to ~ 11 μm, while the overall grain size decreases from ~ 20 μm to ~13 μm (Fig. 5(f)). This indicates that DRX leads to the grain refinement, accompanied by the growth of new grains.
Fig. 5. Representative inverse pole figure maps of the bonding area in the joints obtained at strain of (a) 0.20, (b) 0.30, (c) 0.40, and (d) 0.50, under constant strain rate and PDB temperature of 0.01 s-1 and 1000 °C, respectively. Evolution of (e) volume fraction of different types of grains (i.e. recrystallized, substructured, and deformed) and (f) average grain size at various strain levels.
Above results reveal that the flat bonding interface experiences significant changes in the microstructure through DRX. With the increasing strain, both sides of the flat bonding interface are gradually covered with small recrystallized grains, which are intertwined to promote bonding of the interface. As the recrystallized grains (on both sides of the bonding interface) grow toward the inner part of the matrix grains, they also grow across the interface, thus resulting in the bonding of the microstructures on both sides of the interface. The interface gradually disappears due to the continuous growth of the recrystallized grains. Consequently, the microstructures on both sides of the interface amalgamate with each other. Previous researchers have reported that migration of the IGBs leads to perfect bond between the microstructures on both sides of the interface [38,39]. The evolution of the interface microstructure, such as the migration of IGBs, seems to be closely linked with DRX at a given processing temperature.
3.2.2. Strain-induced IGB migration
Representative EBSD maps, shown in Fig. 6, illustrate PDB phenomenon under different strain levels. This process involves nucleation and growth of the DRXed grains evolved at IGBs. In IPF map, different colors within one deformed grain show variation in the orientation within the grain. As the strain level increases from 0.20 to 0.40, IGB bulging takes place by migration of the original boundary (N0) and growth of the newly formed grain (N2) on the interfacial bonding line (Figs. 6(a, b)). This phenomenon indicates a successful bonding of the local interface in the joint. The presence of net-shaped low-angle grain boundaries (LAGBs) in Fig. 6(c) indicates formation of subgrains at a strain level of ε = 0.20, which seems to be the main reason for the variation of orientation in the vicinity of the IGBs. As deformation strain increases to ε = 0.40, the area in front of the bulged IGBs contains a large number of subgrains which are characterized by low (white and red lines) or high-angle grain boundaries (HAGBs) (black lines) in Fig. 6(d). Meanwhile, the subgrain (N1) is composed of the sections containing LAGBs as well as HAGBs. This indicates the operation of the continuous DRX mechanism in the vicinity of IGBs. Fig. 6(e-f) show KAM maps of the specimens deformed at different strain levels. The highlighted areas represent stress concentration regions which exactly overlaps with the position of sub-boundaries in Fig. 6(a-b). This indicates that the formation of subgrains in the vicinity of IGBs is strongly related to the distribution of strain. Hence, the bulging phenomenon can be considered as strain-induced boundary migration (SIBM) [40,41].
Fig. 6. (a, b) Inverse pole figure maps, (c, d) Grain boundary maps and (e, f) Kernel average misorientation maps of the bonding interface bonded at 1000 °C under deformation strains of: (a, c, e) 0.20 and (b, d, f) 0.40. (g) Geometrically necessary dislocation density (ρGND) profile along the lines ‘AB’ and ‘CD’ marked in panels (e) and (f), respectively. (h) Geometrically necessary dislocation density (ρGND) profile from point G1 to G2, (marked in inset of panel (h)).
Fig. 6(g) shows ρGND profiles of both samples along the lines ‘AB’ and ‘CD’ which are marked in Fig. 6(e) and (f), respectively. At ε = 0.20, multiple ρGND peak values appear at different positions along the line ‘AB’, covering a spatial distance of 14 μm. This indicates that a large number of subgrains evolved on both sides of the IGBs to efficiently coordinate with local inhomogeneous deformation. In contrast, with an increase in deformation strain to ε = 0.40, the peak values of ρGND tend to decrease and become uniform along the line ‘CD’. This suggests that the evolution of new grains is facilitated by the transformation of subgrains, thereby consuming GNDs. Fig. 6(h) shows ρGND profiles in the vicinity of the bulged IGB (marked by white dashed box in Fig. 6(f)). The bugling of original IGB (N0) is caused by the migration of the IGB from grain G2 into grain G1. A relatively higher density difference of GNDs (i.e., Δ ρ ≈ 4.0 × 1015 m-2) exists in the vicinity of the bulged IGB. This observation can be interpreted as the difference in stored deformation energy (induced by the GNDs which accumulate in the vicinity of the IGB) which promotes migration of the interface. Jiang et al. [42] reported that GNDs are usually concentrated in the vicinity of triple points and grain boundaries while studying the deformation behavior of polycrystalline copper. This process can also be proved as SIBM. SIBM involves a small bulging of the pre-existing grain boundaries, leaving an area with lower dislocation content behind the migrating interface. Hence, the nucleation sites are formed by relatively dislocation-free cells in the vicinity of the IGB. This mechanism is characterized by newly grown grains that have similar orientations to their parent grains [41]. Similar observation is noted in the present study, Fig. 6(b).
3.2.3. Substructure formation at the bonding interface
Fig. 7 shows EBSD and TEM images of the typical substructure that evolves in the vicinity of the IGBs at 1000 °C and ε = 0.10. Fig. 7(a) shows an IPF map of the substructure, where the color shade variation within a grain (e.g., along the line ‘AB’) indicates a gradual change in the orientation of the local lattice. The variation in three-dimensional microstructure of the crystal from point 1 to 6 on the line ‘AB’ indicates that the local lattice of the observed grain rotates along <001> direction during initial stage of PDB process. Fig. 7(b) shows a corresponding KAM map of the same area. The dark green lines in the KAM map suggest that the lattice rotation of the local substructure results in stress concentration [43]. Fig. 8(c) shows a TEM image of the region marked by the white box in Fig. 8(b). It was further revealed that the local substructure generated by lattice rotation is composed of dislocation walls (as demarcated by yellow dashed lines in Fig. 7(c)). This is in excellent agreement with the distorted features of the local lattice highlighted in the KAM map (Fig. 7(b)). Hence, the dislocation walls observed in Fig. 7(c) could be referred to as GNDs associated with local inhomogeneous lattice strains [44]. Another type of dislocations, named as statistically stored dislocations (SSDs) [45], was also revealed in the deformed substructure. The SSDs are highlighted by white arrows in Fig. 7(c).The generation of SSDs may also be attributed to the local strain within the substructure. SSDs generally evolve through the random trapping of dislocations, whereas GNDs are related to rotation of the lattice [46]. Figs. 7(d, e) respectively show the misorientation angle and corresponding ρGND profiles along the line ‘AB’ marked in Fig. 7(a). The point-to-point misorientation angle at a distance of ~12 μm (starting from point 1 to point 6 in Fig. 7(a)) does not exceed 2°, while the point-to-origin misorientation angle approaches over 10°. The values of ρGND show a close correlation with the local misorientation along line ‘AB’. These results further elucidate the notion that the evolution of fine substructures in the vicinity of the IGBs is strongly dependent upon the local strains associated with the deformed grains. Therefore, the evolution of new grains in the vicinity of the IGBs should only be considered as a strain-induced phenomenon.
Fig. 7. (a) Inverse pole figure map, (b) kernel average misorientation map, and (c) TEM image of the subgrains evolved in the vicinity of IGBs at 1000 °C and a deformation strain of 0.10. (d) Misorientation angle (Δθ) and corresponding (e) geometrically necessary dislocation density (ρGND) profiles along the line ‘AB’ marked in panel (a).
Fig. 8. (a) Room temperature tensile stress-strain curves of the wrought and PDBed specimens deformed at different deformation strain rates. (b) Comparison between yield strengths (YSs), ultimate tensile strengths (UTSs), and elongations (ELs) extracted from the tensile curves in panel (a).
For the sake of comparison, mechanical properties of the PDBed specimens together with the original wrought IN718 are shown in Figs. 8. Fig. 8(a) shows representative room temperature tensile curves for specimens bonded at 1000 °C under different strain rates. Comparison of mechanical properties (like yield strengths (YSs), ultimate tensile strengths (UTSs), and elongations (ELs)) extracted from the tensile curves is presented in Fig. 8(b). It can be observed that the yield strength (YS) and elongation (EL) values of the PDBed specimens are similar to that of the wrought IN718. However, the UTS values of PDBed specimens are ~ 100 MPa lower than that of the wrought IN718. It is clearly that PDBed joints are not as strong as the wrought IN718. For the PDBed specimens, an increase in the bonding strain rate from 0.001 s-1 to 0.1 s-1 results in a slowly decreasing in values of UTS, YS and EL, while further increase in strain rate to 1 s-1, the UTS, YS and EL values of the tested specimens are slightly increased by ~ 50 MPa, ~20 MPa, and ~ 0.1, respectively. These improvements in the mechanical properties of PDBed specimens can be attributed to the increased ΨBonding, as discussed in section 3.1. It is because that the bulged boundaries formed at the IGBs might hinder the propagation of the crack. The cracks could not propagate along the bonding interface.
According to the evolution of microstructures in the bonding area of IN718 joints (as investigated in 3.2), it can be inferred that the DRX evolved at IGBs is closely related to the bonding process. This type of bonding process could be considered as a nucleation and growth process of DRXed grains evolved under different deformation environments. Therefore, the bonding mechanism can be explained on the basis of current experimental observations, as schematically illustrated in Fig. 9.
Fig. 9. Schematic diagram explaining bonding mechanism through nucleation and growth of DRXed grains in the bonding area: (a) Subgrains, acting as nuclei for DRX, are initially generated within the parent grain on both sides of the interface under continuous deformation. (b) Boundaries of DRXed grains move into the opposite side of the IGBs in order to bond the interface. (c) Nucleation and growth of DRXed grains in the vicinity of the IGBs leading to a fully bonded joint.
During the initial stage of the bonding, slight plastic deformation, occurring at the interface, causes heterogeneous strain inside the interface grains (Fig. 7(a, b)). The evolution of sub-boundaries is attributed to the local strain gradients originated from strain incompatibility between adjacent deformed grains. The local strain gradients show a small misorientation, indicating formation of the GNDs in the deformed areas (Fig. 7(c)). These GNDs form the sub-boundaries [47], which ultimately transform into HAGBs at higher strain levels (Fig. 6(a, b)). Hence, a large number of subgrains are generated within the deformed grains on both sides of the interface (Fig. 9(a)). The formation of subgrain, which acts as nuclei for the evolution of DRXed grain, is mainly associated with the transformation of strain-induced subboundaries from LAGBs to HAGBs through the operation of CDRX mechanism [48]. This evolution behavior of recrystallized grains in the bonding area seems to be similar to that formed in an adiabatic shear band (created under severe plastic deformation) in the high entropy alloys. For instance, Li et al. [49] demonstrated for CrMnFeCoNi high entropy alloy that the grains generated in the adiabatic shear band (with LAGBs) can easily rotate which result in increase of subgrain misorientation under severe plastic deformation.
Further increase in the strain level results in the gradual transformation of subgrains into DRXed grains (Fig. 6(b)). With the evolution of DRX, the new grains on the interface gradually grow up into the adjacent deformation regions, causing the IGBs to migrate into the opposite sides (Fig. 9(b)). The migration of recrystallized grain boundary at the pre-existing HAGBs is also considered as SIBM (Fig. 6(b)), especially occurring at lower strain levels [50]. The bulging and further migration of the original IGBs is characterized by HAGBs which are generally located in regions with high strain gradients. Due to the difference in deformed storage energy on both sides of the IGBs, a section of the IGB migrates into the adjacent grain at the expense of total stored energy, thus creating a room for the evolution of recrystallized grains [51]. In short, the gradient of deformed storage energy across the IGB is the main driving force for SIBM. Humphreys [52] calculated the critical size required for a large subgrain to undergo SIBM. The results showed that the critical subgrain size for single-subgrain SIBM (as shown in Fig. 6(b)) decreases if there exists a stored energy difference across the boundary. The viability of single subgrain SIBM is therefore depend upon the size, distribution and boundary energy (or misorientation) of the subgrains located within two adjacent grains. Based on this conclusion, it is evident from the present study that the SIBM is strongly related to the size and misorientation of the subgrains, generated on the both sides of the bonding interface (Fig. 6(b)). This generally occurs after substructure formation in the vicinity of the IGBs.
During the evolution of DRX, the boundary movement (caused by the growth of newly formed daughter grains) can successfully bond the two parts of the joint (Fig. 9(c)). Therefore, the dynamic grain boundary migration plays a prominent role during PDB. It is well established that the grain boundary movement involved in the nucleation and growth of the DRXed grains is associated with an extremely strong diffusion-controlled process that induces long-distance migration of grain boundaries [53]. Hence, it is deduced that the grain boundary movement rate should be higher owing to adiabatic temperature rise. This could also be the main reason for the sharp increases in the ΨBonding caused by the localized adiabatic temperature rise (Fig. 4(b)), as well as increase in the average size of recrystallized grains (Fig. 3(f)). Accordingly, the deformation strain rate should be improved to obtain better bonded joint.
Above results suggest that higher strain rate (e.g., ≥ 1 s-1) should result in the optimum bonded joint, however, achieving such a high strain rate is relatively difficult from a practical point of view. If the effect of localized adiabatic temperature rise on DRXed grain growth is not considered, it is much clear that a better condition for obtaining a good joint should be a lower strain rate. This is due to the fact that a lower strain rate means the isothermal deformation takes longer time and the new grains have enough time for their growth. For PDBed joint, it is quite clear that the more complete new grain growth is, the higher is the ΨBonding of the joint. Therefore, it is a good choice to reduce the strain rate without increasing the deformation temperature.
The microstructural evolution and bonding mechanisms of IN718 superalloy joints during isothermal compression bonding were studied in detail at different strain rates. The main results can be summarized as follows:
(1) Different extents of DRX occur in the bonding area of IN718 joints prepared by a PDB process at 1000 °C under different strain rates. The fraction of DRX, as well as the size of new grains, initially decreases with increasing strain rate up to 0.1 s-1 and increases later at a higher strain rate owing to the localized adiabatic temperature rise.
(2) Interfacial microstructure studies indicate that the bonding process of joints is closely related to the development of fine DRXed grains at the bonding interface under high strain. Small subgrains (acting as nuclei for DRX) are developed by lattice rotation on both sides of the IGBs due to continuous increase in strain hardening effect. Subsequently, the subgrains gradually grow up into the adjacent deformation regions, causing the IGBs to migrate into the opposite sides. The bulging and further migration of the original IGBs (which is considered as the growth process of DRX) are imperative for successful bonding of the joints.
(3) The strain rate has a significant influence on the bonding behavior of the joint during PDB. With increasing strain rate up to 0.1 s-1, the ΨBonding decreases with the decrease in the extent of DRX, while, at the higher strain rate, ΨBonding increases simultaneously. The sharp rise in ΨBonding, at a strain rate of 1 s-1, is related to the increase in the extent of DRX caused by a localized adiabatic temperature rise.
(4) The UTS values of PDBed specimens are ~ 100 MPa lower than that of the wrought ones, while the YS and EL of PDBed specimens are almost similar to the wrought alloys. For the PDBed specimens, an increase in bonding strain rate from 0.001 s-1 to 0.1 s-1 results in a slowly decreasing in values of UTS, YS and EL, while further increase strain rate to 1 s-1, these values are slightly increased by ~ 50 MPa, ~20 MPa, and ~ 0.1, respectively. These improvements in the mechanical properties of PDBed specimens can be attributed to the increased ΨBonding.
This work was supported by the National Key Research and Development Program [grant number 2018YFA0702900], the National Natural Science Foundation of China [grant numbers U1508215, 51774265], the National Science and Technology Major Project of China [Grant No. 2019ZX06004010], the Key Program of the Chinese Academy of Sciences [grant number ZDRW-CN-2017-1], and Program of CAS Interdisciplinary Innovation Team.
WeChat
/
| 〈 |
|
〉 |
