Journal of Materials Science & Technology  2019 , 35 (8): 1671-1680 https://doi.org/10.1016/j.jmst.2019.04.005

Orginal Article

Interfacial microstructure evolution and bonding mechanisms of 14YWT alloys produced by hot compression bonding

Liying Zhouab, Shaobo Fenga, Mingyue Suna*, Bin Xua, Dianzhong Lia

a Shenyang National Lab. for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China
b School of Materials Science and Engineering, University of Science and Technology of China, Shenyang 110016, China

Corresponding authors:   *Corresponding author.E-mail address: mysun@imr.ac.cn (M. Sun).

Received: 2019-02-15

Revised:  2019-03-15

Accepted:  2019-03-21

Online:  2019-08-05

Copyright:  2019 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology

More

Abstract

Hot compression bonding was first used to join oxide-dispersion-strengthened ferrite steels (14YWT) under temperatures of 750-1100°C with a true strain range of 0.11-0.51. Subsequently, the microstructure evolution and mechanical properties of the joints were characterized, revealing that the 14YWT steels could be successfully bonded at a temperature of at least 950 °C with a true strain of 0.22, without degrading the fine grain and nanoparticle distribution, and the presence of inclusions or micro-voids along the bonding interface. Moreover, the joints had nearly the same tensile properties at room temperature and exhibited a similar fracture morphology with sufficient dimples compared to that of the base material. An electron backscattered diffraction technique and transmission electron microscopy were systematically employed to study the evolution of hot deformed microstructures. The results showed that continuous dynamic recrystallization characterized by progressive subgrain rotation occurred in this alloy, but discontinuous dynamic recrystallization characterized by strain-induced grain boundary bulging and subsequent bridging sub-boundary rotation was the dominant nucleation mechanism. The nuclei will grow with ongoing deformation, which will contribute to the healing of the original bonding interface.

Keywords: Oxide dispersion strengthened steels ; Hot compression bonding ; Dynamic recrystallization ; Tensile property

0

PDF (5076KB) Metadata Metrics Related articles

Cite this article Export EndNote Ris Bibtex

Liying Zhou, Shaobo Feng, Mingyue Sun, Bin Xu, Dianzhong Li. Interfacial microstructure evolution and bonding mechanisms of 14YWT alloys produced by hot compression bonding[J]. Journal of Materials Science & Technology, 2019, 35(8): 1671-1680 https://doi.org/10.1016/j.jmst.2019.04.005

1. Introduction

Oxide-dispersion-strengthened (ODS) steels have been considered as candidate materials for nuclear reactors because of their advanced creep properties [1], excellent irradiation tolerance [2], and superior corrosion resistance [3] based on nanoscale oxides with unparalleled stability at elevated temperatures. Such properties are necessary because the materials used in the core of a nuclear reactor are exposed to an extremely aggressive environment which includes high temperature, high stresses, a chemically aggressive coolant, and intense radiation fluxes [4].

In order to use ODS steels for structural components with large, complex structures, robust joining techniques are urgently needed. It widely reported that conventional fusion welding leads to the agglomeration of nanoparticles and disruption of the fine grain distribution [5], so solid-state routes are preferred such as pressure resistance welding (PRW) [6], friction stir welding (FSW) [7,8], and solid-state diffusion bonding (SSDB) [[9], [10], [11], [12]]. However, these techniques have some inherent shortcomings. The defects induced by PRW may include fine hairline cracks that are difficult to eliminate and the inevitably formed burr need extra machining process to remove [6]. The main limitations of FSW are the curvature and smaller thickness of the cladding tube for the FSW of the end plug with clad tubes because the frictional volume is too small to generate sufficient frictional heat [13]. Although SSDB is considered to be a more reliable method, where the atomistic interaction produced by diffusion can heal the joint interfaces [9], it is relatively time-consuming [13]. Therefore, access to a reliable, affordable, and universally applicable joining technology is increasingly needed with the rapid advancement of nuclear reactors. Hot compression bonding (HCB) can take advantage of the coupled effects of temperature and deformation to facilitate interfacial grain boundary migration and atomic diffusion. What's more, it does not create a local overheating zone at the bonding interface, which can prevent the fine-scaled microstructures from disrupting. This method has been successfully applied in joining not only identical materials, like 316 L N stainless steels [14] and IN718 superalloys [15], but also dissimilar materials, like high Cr cast iron and low carbon steel [16]. Hence, HCB was firstly proposed to join ODS steels in our research.

In this work, HCB was employed to join 14YWT ODS steels with a series of bonding parameter, and then tensile tests were conducted at room temperature on the joints to evaluate the feasibility of this method. In order to elucidate the mechanisms involved in interfacial healing, electron backscattered diffraction (EBSD) in conjunction with transmission electron microscopy (TEM) were employed to study the microstructural evolution along the bonding interface.

2. Experiment

The chemical composition of 14YWT ODS steel is listed in Table 1. The ODS steel was produced by mechanical alloying and hot isostatic pressing (HIP). Metallic alloy powder and Y2O3 powder were mechanically alloyed in a planetary attritor under a highly pure Ar gas atmosphere at a rotation speed of 240 rpm for 50 h using a ball-to-powder weight ratio of 10:1. Subsequently, the powders were sieved and charged in mild steel capsules, which were degassed to 10-3 pa at 400 °C for 3 h, and then HIP was carried out at 1150 °C for 4 h.

Table 1   Element content (wt. %) of experimental materials. Fe is in balance.

ElementCrWTiYCON
14YWT13.791.850.290.200.0050.0070.008

New window

Samples (φ8 mm × 6 mm) were machined, and the joint surfaces were sequentially ground using SiC papers. Fig. 1(a) shows a schematic of the HCB process performed by a Gleeble-3500 thermo-mechanical simulator under a high vacuum of 10-3 Torr. Before bonding, the specimens were heated to the selected bonding temperature of 750-1100 °C at a rate of 5 °C s-1 and held there for 300 s to ensure heating uniformity. Then, the specimens were compressed with different true strains ranging from 0.11 to 0.51 at a strain rate of 0.01 s-1.

Fig. 1.   Schematic of hot compression bonding process (a) and specimen geometries used in tensile tests (b).

To evaluate the joint reliability, tensile tests were conducted at ambient temperature with strain rate of 10-3 s-1. Tensile specimens were taken from the core of the compressed specimen with the joint interface located at the center of the sample, and the specimens were carefully polished to eliminate any traces of wire-electrode cutting with a gauge length of 5 mm, width of 1.5 mm, and thickness of 1 mm, of which the geometries is shown in Fig. 1(b). At least 3 samples were measured for each bonding parameters to ensure the reliability of the results.

After polishing with ion milling to remove the surface deformation layer, the crystallographic microstructure of the grains was determined by the EBSD with the step size of 60 nm using an FEI Nova NanoSEM 430 field-emission scanning electron microscope equipped with a fully automatic HKL Technology EBSD attachment and an energy dispersive spectrometer (EDS). In order to further investigate the microstructure of the bonding interface, the specimens used for TEM observations were precisely sampled from the bonding interface, and then were twin-jet electro-polished with a solution of 10% perchloric acid in ethanol at a temperature of -30 °C and a voltage of 25 V. The samples in which a hole appeared at the interface were selected to investigate the microstructure characterization along the bonding interface. The TEM observations were performed using an FEI Tecnai G2 F20 field-emission transmission electron microscope at an acceleration voltage of 200 kV.

3. Results

3.1. Starting material

The grain size of 14YWT was characterized as a typical bimodal distribution [17] as shown in Fig. 2(a). The grain size distribution is dominated by small submicrometer, and even nanoscale grains produced by the high-energy ball milling and the presence of nanostructured oxide, with far fewer grains extending up to a few micrometers. In the TEM observation shown in Fig. 2(b), oxide particles are found within the grains and along the grain boundaries. The following atom probe tomography (APT) results demonstrates that the nano-particle is mainly composed of Y, Ti and O with a nominal diameter about 5 nm. From chemical concentration analysis of the particles, it can be found that the chromium-to-yttrium atom ratio was approximately 1:1 (the composition estimate of the core was 14.9 ± 0.017 at % Ti, 12.4 ± 0.06 at % Y). it was well established that the two main phases in 14YWT are usually considered to be Y2Ti2O7 and Y2TiO5 [18,19]. Therefore, it can be inferred that most of the nanoscale phases is more likely to be Y2Ti2O7 particles.

Fig. 2.   Optical image (a), TEM micrograph (b) and APT maps (c) of studied 14YWT before hot deformation bonding in which Y-Ti-O particle is shown in the sample.

3.2. Overall interfacial evolution under different bonding parameters

The typical optical microscope (OM) photographs and corresponding SEM images in Fig. 3 illustrate the overall interfacial evolution under different bonding parameters. Fig. 3(a)-(d) shows the interfacial microstructure changes with temperatures from 750 °C to 1100 °C at a lower strain of 0.11. It can be clearly observed that the bonding boundary is distinct and relatively straight at 750 °C, and there is an evident inclusion layer along the bonding interface as marked by the yellow arrow in Fig. 3(a). The corresponding EDS profile reveals that the inclusions at the interface have a brighter contrast in the backscattered electron image and are rich in Cr and C elements (Fig. 4(a) and (b)), hence, it can be inferred that these inclusions may be Cr23C6 [20]. With the temperature increasing to 850 °C (Fig. 3(b)), the inclusion layer becomes partly indistinct and discontinuous, and at 950 °C we can see that the bonding interface is less distinguishable but still remain straight. Further SEM observations show that the carbides become fairly sparse, which implies that these carbides can gradually disintegrate with an increase in temperature. While further increasing to 1100 °C, it can be seen that there are hardly any inclusions appeared along the bonding interface, but the migration of interfacial grain boundaries occurs at triple grain boundary junction.

Fig. 3.   Microstructural changes shown in optical images and corresponding SEM images at increasing temperatures ((a) 750 °C, (b) 850 °C, (c) 950 °C, (d) 1100 °C) with strain at 0.11, and at different deformations ((e) 0.22 (f) 0.51) at temperature of 950 °C, where red arrow indicates original interface.

Fig. 4.   (a) The backscattered electron image, (b) EDS spectrum from the interfacial inclusion of the joint region bonded at 750 °C with the strain of 0.11, and (c) the EDS maps in regards of Fe, Cr, W, Ti, and Y elements of joint region bonded at 950 °C with the strain of 0.22.

The interfacial microstructure evolution when the true strain increases from 0.11 to 0.51 at 950 °C is displayed in Fig. 3(c),(e),(f). When the strain reaching 0.22 (Fig. 3(e)), the bonding interface is almost invisible in OM images. The SEM observation exhibits that the bonding interface gets fluctuated without any evident micro-voids and inclusions existed along the bonding interface. EDS map scanning (as shown in Fig. 4(c)) indicates that the element distributions of Fe, Cr, W, Ti, and Y across the bonding interface are virtually constant without element concentration phenomenon, and it is worth noting that the highlighted spot of Ti existed within the grain but is absent along the bonding interface. At the strain of 0.51 shown in Fig. 3(f), the bonding interface is almost undistinguishable and has almost been replaced by wavy grain boundaries.

3.3. Tensile property of joints compared to base material

The microstructure characterization suggested that HCB produced promising joints without microstructure deterioration. In order to further evaluate the reliability of the joints, tensile tests were performed at room temperature on the joints after bonding using optimized parameters compared with the base material. Fig. 5(a) shows the stress-strain curves of joints bonded at 950 °C with various strains. At a strain of 0.11, the joints ruptured much earlier in a brittle mode. When the strain was increased to 0.22 and 0.51, the strength and elongation rate of the joints were comparable to those of the base material. Similarly, Fig. 5(b) shows the stress-strain curves of joints bonded with a deformation of 0.22 at various temperatures. At a lower temperature of 850 °C, the tensile strength and elongation rate decreased to 86% and 79% of the base material, respectively. At the adequate temperature of 950 °C, the mechanical properties of the joint were almost identical to those of the base material. However, there was a slight reduction in strength but an increment in the elongation rate at 1100 °C, and this phenomenon might be ascribed to the effect of grain coarsening with temperature and strain increasing.

Fig. 5.   Stress-strain curves of joints at room temperature with various strains (a) and temperatures (b) compared with base material (as-received).

Fig. 6 shows the fracture morphologies of the joints and base material in macro and micro view. At 950 °C with a strain of 0.11, the wrought joints shown in Fig. 6(a) exhibits major delamination, namely, macroscopically brittle fracture. The flat-bottomed fracture without the presence of dimples implies the crack starts from the unbonded area and propagates unimpeded along the bonding interface due to the inclusion layer. The joints bonded at 850 °C with a strain of 0.22 (Fig. 6(d)) reveals a typical 45° crack in macroscopic morphology with fine scale dimple, suggesting the joints has been bonded metallurgically. The other joint specimens occurs by double shear, as shown in the macroscopic profile, along with evident necking behavior in both the thickness and width directions. Furthermore, the micro observation manifests ductile failures with relatively deep and equiaxial dimples, and the occurrence of shear ledges indicates that the crack growth has undergone large plastic deformation, which bear much resemblance to the fracture morphology of the base material.

Fig. 6.   Fracture surfaces with corresponding macroscopic profile (inserted figures) after tensile tests at room temperature under different strains at 950 °C ((a) 0.11, (b) 0.22, and (c) 0.51, and at increasing temperature with strain of 0.22 ((d) 850 °C and (e) 1100 °C), compared with base material (f).

So far, it can be inferred that the structure of nanoparticles remain unchanged since the agglomeration of nanoparticles is bound to result in the degradation of the mechanical properties. And we found that both the deformation temperature and strain could affect the interfacial healing behavior, but its role remain undiscovered. Hence, the microstructure evolution with various bonding parameters will be further discussed in detail.

3.4. Microstructure evolution

3.4.1. Effects of strain and temperature on overall microstructure evolution

The EBSD technique is an effective method for identifying the characteristics of substructures and was applied here to investigate the interfacial microstructure evolution during the hot compression process. Fig. 7 shows inverse pole figures (IPF) with grain boundary distribution maps after bonding at 950 °C with increasing strains. In these maps, the black, green, and gray lines represent grain boundaries with different misorientation angles: greater than 15° (high-angle grain boundaries, HAGBs), in the range of 10-15° (middle-angle grain boundaries, MAGBs), and in the range of 2-10° (low-angle grain boundaries, LAGBs), respectively. It can be seen that the microstructures are obviously affected by the strains. During the early stage of deformation in Fig. 7(a), the bonding interface remain relatively straight with a small amount of newly formed grains appeared, and some sub-boundaries with low-angle misorientations develop inside of the deformed grains as a result of the generation and rapid rearrangement of dislocations in this alloy. When the strain increases to 0.22, the bonding interface began to bend since more new grain is formed at the vicinity of the interface. Under the condition of relatively higher strains (ε = 0.51), the bonding interface is rather undistinguishable since it has been replaced by the grain boundaries migration. It is worthwhile to point out that incomplete HAGBs partly connected by LAGBs can be observed within grains, indicating that the sub-boundaries progressively transform into general grain boundaries during the deformation. Generally, the formation of HAGBs inside the deformed grains as a result of gradually increasing the misorientation of the LAGBs is a typical feature of the continuous dynamic recrystallization (CDRX) mechanism [21].

Fig. 7.   Inverse pole figures with grain boundary distribution of joint region at 950 °C and various strains ((a) 0.11, (b) 0.22, and (c) 0.51). The LAGBs, MAGBs, and HAGBs are represented by silver, green, and black lines, respectively. Enlarged views of locations 1-5 in.(a) are shown.

Furthermore, in order to understand the effects of the deformation strain on the substructure behavior, a misorientation analysis was performed along the lines marked in Fig. 7. The local (point-to-point) and accumulative (point-to-origin) misorientation of the two characteristic directions are plotted in Fig. 8. At strain of 0.11 shown in Fig. 8(a) and (b), the point-to-origin misorientation along the grain boundary gradually increases but does not exceed 6°, and the accumulative misorientation within the grain is also lower than 7°, suggesting that sub-boundaries are under development at the early stage of deformation. With further deformation at a strain of 0.22, the misorientation gradient has a sight increment along the grain boundary reaching 11°, suggesting that subgrain rotation may occur within the grain. As seen in Fig. 8(e) and (f), the point-to-origin misorientation from the grain boundary to grain interior easily exceeds 20° when the strain increases to 0.51, while the cumulative misorientation along the grain boundary remains relatively low. It can be observed that the accumulative misorientation (C2) does not monotonically increase from the grain boundaries to grain interior but occurs by misorientation jumps with followed plateau, in consideration of the low misorientation gradient along the grain boundaries (C1), it is a strong evidence of the formation of orientation bands dividing the initial grain. It is widely acknowledged that the formation of the orientation bands [22] is correlated with the incompatibility of deformation with higher strain. Above all, the level of misorientation gradient increases with ongoing deformation, suggesting the CDRX is gradually developed since the CDRX nuclei can transform into HAGBs by accumulating adequate misorientation [23].

Fig. 8.   Changes in misorientation angle measured along lines marked in Fig. 6: (a) A1, (b) A2, (c) B1, (d) B2, (e) C1, and (f) C2.

The microstructure evolution of the joint region under 850 °C and 1100 °C is displayed in Fig. 9. Here, the strain of the deformed specimen is 0.22. Compared with the characterization of joints bonded under 950 °C in Fig. 7b, it can be clearly seen that newly formed grain along the bonding interface has grown up with temperature increasing. At 850 °C, the sub-boundaries is relatively abundant due to the stronger dynamic recovery (DRV) process under lower temperature. While, the bonding interface at 1100 °C has been replaced by the coarsened grains since the grain boundaries migration is accelerated with temperature increased. In addition, it can be found that there are more small grain nucleated along the pre-existing grain boundaries, forming the so-called necklace microstructures locally [24], which indicates a different DRX process.

Fig. 9.   IPF maps with grain boundary distribution of joint region at 850 °C (a) and 1100 °C (b) with the true strain of 0.22. The LAGBs, MAGBs, and HAGBs are represented by silver, green, and black lines, respectively.

Fig. 10 shows the transformation of grain boundaries with different misorientation angle scopes with increasing strain and temperature. It can be concluded from Fig. 10(a) that the fraction of LAGBs increases when deformation strain is gradually increasing. This occurs because the dislocations introduced by the ongoing deformation first rearrange themselves to form new sub-boundaries (LAGBs), which is a strong dynamic recovery process and often considered to be the incubation period of CDRX [25]. Moreover, the fraction of MAGBs is below 5% and does not obviously change, which can contribute to the nucleation mechanism of CDRX because MAGBs are considered to be the transition from LAGBs to HAGBs. Likewise, Fig. 10(b) depicts the change in the misorientation angle with increasing temperature. The fraction of LAGBs decreases when the deformation temperature increases, which mainly results in the enhanced transformation from LAGBs to HAGBs since the dislocation motion is enhanced at higher deformation temperatures. In contrast, the MAGBs do not change much but remain at low level, showing that the CDRX is an important but not the dominant DRX mechanism in 14YWT ferrite steels.

Fig. 10.   Fractions of different misorientation angle scopes with varying strain at 950 °C (a) and varying temperature at strain of 0.22 (b).

3.4.2. Interfacial healing behavior

Fig. 3 shows that the bonding interface becomes less visible with increasing strain. The details of this evolution are elaborated in Fig. 7, and the interfacial healing behavior can be explicitly observed in the enlarged views of locations 1-5 indicated by the white circles in Fig. 7(a). The original grain boundaries along the bonding interface bulge out. It is followed by the formation of sub-boundaries behind the bulging area them and the transformation to LAGBs (location 2), MAGBs (location 3), and finally HAGBs (location 5). At location 4, it can be seen that the HAGBs formed behind the bulging area are discontinuous and mixed with MAGBs, indicating that the HAGBs evolved via progressive subgrain rotation. With ongoing deformation, the newly formed nuclei will grow in size, contributing to the replacement of the original bonding interface.

To further confirm the healing behavior, we conducted a TEM analysis at a strain of 0.11 under 950 °C. This condition were selected because the bonding interface was distinguishable so that we can ensure the TEM characterization can be performed on the joint region. As shown in Fig. 11(a), the grain boundaries along the original bonding interface bulge into the neighboring grains with evident contrast differences, and there seems to be a sub-boundary formed behind them. Meanwhile, the dislocation-free grains with convex grain boundaries along the bonding interface are supposed to be the newly formed DRX grains by discontinues dynamic recrystallization (DDRX) mechanism. This result is consistent with Fig. 7. After the DRX nuclei form, they can grow when the driving force associated with the stored energy of the deformed matrix is sufficient to surmount the surface-energy-driven shrinkage of the DRX grains [26,27]. Fig. 11(b) shows a higher magnification micrograph of the bulging area. The migration direction is from the grains with finer and more abundant particles to the grains with fewer nano-oxides, because there is a large strain gradient between these two neighboring of grains. In addition, it is worthwhile to point out that the nanoparticles were not found to agglomerate after HCB, which suggests that this joining technique will not disrupt the distribution of the original microstructure, including the fine grains and nanoscale strengthened phase.

Fig. 11.   TEM morphology of bonding interface at 950 °C with strain of 0.11 (a) and higher magnification image of bulging area (b).

4. Discussion

4.1. Dynamic recrystallization mechanism

DRX can occur in one step in a relatively slow and continuous way called CDRX or in two steps in a fast and discontinuous manner called DDRX [25,[28], [29], [30]]. CDRX can be regarded as an extended dynamic recovery process featuring progressive subgrain rotation, and thereby the transformation of LAGBs into HAGBs [25]. In contrast, DDRX includes the nucleation and subsequent growth of new strain-free grains at the expense of deformed grains. In this work, 14YWT, which has a high stacking fault energy, was expected to undergo strong dynamic recovery, which enhances subgrain rotation and thereby CDRX [24,31].

In essence, CDRX can turn the deformed and recovered state into DRX grains through the process shown in Fig. 12. At the early stage of hot compression, dislocations are generated and randomly distributed along the grain boundary and in the grains (Fig. 12(a)), which are then rearranged into highly tangled dislocation cells by the cross-slip and climb mechanisms. With ongoing deformation, the misorientation gradient will increase as demonstrated in Fig. 8, leading to the dislocation walls being converted into subgrains with LAGBs adjacent to the pre-existing grain boundary, which corresponds to the ascending fraction of LAGBs with increasing strain illustrated in Fig. 10. Afterward, these LAGBs will further evolve into HAGBs by continuously absorbing dislocations while the misorientation angles reach a critical θc value of 15° [25,32]. The transformation of LAGBs into HAGBs is in accord with the decreasing fraction of LAGBs with increasing temperature, because the dislocation motion will become faster, contributing to the transition of LAGBs into HAGBs. The subgrain coalescence can also be enhanced by the increase in the deformation temperature. The larger fraction of CDRX with temperature increasing also explains the slight decrease in strength but a small increase in the elongation rate as shown in Fig. 5. Accordingly, it can be concluded that a temperature increase will facilitate the CDRX process, whereas the largest strain (0.51) in this study is not sufficient to initiate overall CDRX.

Fig. 12.   Schematic representation of bonding mechanism (the full lines indicate HAGBs while dotted lines indicate LAGBs).

4.2. Interfacial bonding mechanism

Generally, CDRX can introduce microstructural changes from deformed and recovered structures to recrystallized grains by means of a gradual increase in the misorientation angles, subsequent formation of LAGBs, and final transformation to HAGBs [21], without the large-scale migration of pre-existing boundaries. Therefore, it can be inferred that the CDRX mechanism should contribute little to the healing of the bonding interface.

It is interesting that the current study found that there was an abundance of fine dislocation-free grains generated at the bonding interface (Fig. 11(a)). These new fine grains could be considered recrystallized nuclei. Moreover, almost all of the nuclei appeared to be associated with the bulging of the original bonding interface, followed by the formation and evolution of sub-boundaries, as shown in the circles of Fig. 7(a). From this, it can be inferred that the nucleation of DRX at the bounding interface should be strongly correlated to the appearance of bulging in the pre-existing bonding boundaries.

Generally speaking, during hot deformation the microstructure inhomogeneity is bound to introduce a high stored energy gradient between neighboring grains, especially along the bonding interface. According to the investigation of E. Aydogan et al. [33], nano-oxide play a critical role for static recrystallization mechanisms in 14YWT alloys. The nanoparticles in the unrecrystallized grains are finer with higher density, whereas in recrystallized grains they are coarser and less. Hence, it can be inferred that the different grains will undergo different extent of plastic deformation, especially at the vicinity of the bonding interface. During hot compression, This can directly translate into stored energy differences acting as a driving pressure for grain boundary migration, called strain-induced boundary migration (SIBM) [17,34,35], namely the bulging of the bonding boundaries. As shown in Fig. 11(b), there are nanoparticles dispersed unevenly across the bonding interface. Given the nonuniformity of the size, shape, density, and distribution of the nanoparticles, and thereby the grain size, these nanoparticles will contribute to the deformation incompatibility, therefore strain gradient and stored energy differences around the bonding boundaries during the HCB process. This would promote the occurrence of SIBM along the bonding boundaries. Generally, the bulging areas of the boundaries are potential sites for the formation of DRX nuclei through the formation and development of sub-boundaries by the rearrangement of the dislocations with ongoing deformation [24,27,36,37]. With a further increase in strain, the LAGBs will rotate into MAGBs and finally HAGBs by the absorption of dislocations to achieve full isolation [38,39]. Therefore, the DRX nucleation around the bonding interface during hot deformation is dominant because of the bulging through SIBM and the subsequent evolution of sub-boundaries to separate the bulged areas into new nuclei [40,41]. Afterwards, the fine dislocation-free grains marked in Fig. 11(a), which is considered as DRX nuclei, are delimited by the HAGBs and can grow up by the consumption of work-hardened surroundings. This growth process involves a large-scale boundary migration, which can effectively remove the original bonding interface and integrate the two parts into one. With increasing strain, the stored energy gradient to overcome the energy required for grain boundary curvature increases, while the migration of grain boundaries is accelerated with increasing temperature. Thus, there is a higher interfacial DDRX fraction, which contributes to the enhanced tensile strength of the bonding interface.

To the best knowledge of the author, there are limited studies concerned about recovery, recrystallization and grain growth of ODS steels [33,[42], [43], [44], [45]], and these literature pay attention to the annealing behavior, namely the static recrystallization phenomenon and texture evolution where the nano-oxides is generally considered as barrier of grain boundaries migration. The aforementioned analysis suggests that the nanoscale oxides play a different role. Meanwhile, the introduction of the bonding interface leads to a distinct difference in the substructure evolution of the grains along the bonding interface during the hot deformation. Thus, a different DRX mechanism for the deformed matrix along the bonding interface is proposed on the basis of the current experimental observations, as illustrated schematically in Fig. 12.

During the deformation, an abundance of dislocations would be produced as a result of the plastic deformation of the alloy. For the ferrite matrix, some dislocations can easily rearrange themselves into sub-boundaries. Meanwhile, high strain gradients are gradually built up close to the bonding interface to maintain the strain compatibility between neighboring grains. These high strain gradients at the interface initiate the SIBM (i.e., bulging) of the bonding boundaries. Moreover, the bulging boundaries usually serve as preferential DRX nucleation sites because of the accumulation of high-density dislocations behind the bulging areas, and these dislocations can transform into LAGBs, MAGBs, and finally HAGBs, which separate the bulging areas into new DRX nuclei [40,41]. During a further strain, the growth of the DRX nuclei can efficiently eliminate the bonding boundaries. Therefore, the evolution of the bonding boundaries involves a clear nucleation and growth stage, which is a typical characteristic of the DDRX mechanism. In other words, the effective healing of the bonding interface was found to be the result of the beneficial effect of the DDRX mechanism during the current study.

5. Conclusions

14YWT ODS steels were bonded via hot compression bonding at different strains in the range of 0.11-0.51 between 750 °C and 1100 °C with a strain rate of 0.01 s-1. Tensile tests were performed to assess the mechanical properties of the joints compared to the base material, and interfacial microstructural characterizations were carried out to investigate the microstructure evolution and corresponding bonding mechanism. The main conclusions are as follows:

(1) The 14YWT could be successfully bonded by hot compression bonding at only 950 °C with a minimum strain of 0.22. The bonding interface was free of defects and inclusions, with no degradation of the fine grains and nanoparticle distribution.

(2) The successfully bonded joints manifested the same tensile properties at room temperature as the base material. These joints did not fracture at the bonding interface but exhibited sufficient dimples.

(3) CDRX occurred in the matrix of the 14YWT ferrite steel. However, the strain-induced grain boundary migration driven by stored energy differences and subsequent bridging sub-boundary rotation boosted the DDRX nucleation at the original bonding interface, then the growth of the nuclei along the bonding interface could contribute to the healing of the bonding interface.

Acknowledgements

The authors acknowledge the financial support from National Key Research and Development program (Grant No. 2016YFB0300401), National Natural Science Foundation of China (Grant Nos. U1508215, 51774265) and key Program of the Chinese Academy of Sciences (Grant No. ZDRW-CN-2017-1).

The authors have declared that no competing interests exist.


/