formation of bonding interface in explosive welding—a
TRANSCRIPT
Journal of Physics: Condensed Matter
PAPER
Formation of bonding interface in explosive welding—a moleculardynamics approachTo cite this article: Jianrui Feng et al 2019 J. Phys.: Condens. Matter 31 415403
View the article online for updates and enhancements.
This content was downloaded from IP address 128.61.137.153 on 18/07/2019 at 12:40
1 © 2019 IOP Publishing Ltd Printed in the UK
1. Introduction
Explosive welding is well known for its capability to join a wide variety of both similar and dissimilar metals, which sometimes can hardly be joined by any other welding tech-niques [1, 2]. The plates, tubes or bars of large dimensions possessing a unique combination of properties, such as low weight, high corrosion resistance, high temperature resistance, high frictional properties and excellent mechanical strengths, can be easily fabricated using this method. Although explo-sive welding has been known for more than 70 years, the exact joining mechanism of the bonding interface is still controver-sial [3–7]. This is because of the extremely fast welding pro-cess, which results in the difficulty in obtaining the in situ experimental data. As a result, the bonding mechanism can only be constructed by structural examination of the recovered
samples, when the welding process is finished. Compared with the experiment, a numerical simulation, however, can be used to directly acquire valuable information of explosive welding.
Mousavia et al [8–10] simulated the collision process of explosive welding using Euler method. In this simulation, the pressure and temperature at the contact point rapidly increase during the drastic collision. When the contact point goes ahead, the pressure behind it instantly decreases due to the unloading. However, because of the limitation of thermal dif-fusivity, the heat cannot be dissipated instantaneously. Thus, for a short period of time, a thin layer with high temperature appear near the bonding region. Similar results were also reported by some other authors [11–23]. By analyzing the collision process of the simulations, we can divide the joining procedure of the bonding interface into three distinct stages. The first stage is the loading stage. At this stage, high pressure and temperature are achieved near the collision region during the violent impact [8–10]. After the collision, the pressure near
Journal of Physics: Condensed Matter
Formation of bonding interface in explosive welding—a molecular dynamics approach
Jianrui Feng1, Kaida Dai2,4, Qiang Zhou2, Jing Xie2, Rongjie Yang1, I A Bataev3 and Pengwan Chen2,4
1 School of Materials Science and Engineering, Beijing Institute of Technology, Beijing 100081, People’s Republic of China2 State Key Laboratory of Explosion Science and Technology, Beijing Institute of Technology, Beijing 100081, People’s Republic of China3 Novosibirsk State Technical University, K. Marks 20, 630073, Novosibirsk, Russia
E-mail: [email protected] and [email protected]
Received 16 May 2019, revised 30 June 2019Accepted for publication 10 July 2019Published 18 July 2019
AbstractThe bonding between copper (Cu) and iron (Fe) to form a bi-layer composite using explosive welding is investigated through molecular dynamics simulation. Three stages in the joining process, including loading, unloading and cooling, are sequentially considered in modelling the formation of the bonding interface. The results demonstrate that three types of bonding interfaces can be obtained, based on whether melting happens. The morphologies and the atomic structures of the three types bonding interfaces in each stage are analyzed. The formation of nanograins near the bonding interface is mainly due to the melting and subsequent cooling process. Atomic simulations of tensile tests reveal that melting is not a necessary factor to form the bonding interface. What’s more, depending on whether melting occurs, the joining mechanism can be regarded as pressure welding or fusion-diffusion welding.
Keywords: explosive welding, bonding interface, molecular dynamics, copper, iron
(Some figures may appear in colour only in the online journal)
J Feng et al
Printed in the UK
415403
JCOMEL
© 2019 IOP Publishing Ltd
31
J. Phys.: Condens. Matter
CM
10.1088/1361-648X/ab30d7
Paper
41
Journal of Physics: Condensed Matter
IOP
4 Author to whom any correspondence should be addressed.
2019
1361-648X
1361-648X/19/415403+9$33.00
https://doi.org/10.1088/1361-648X/ab30d7J. Phys.: Condens. Matter 31 (2019) 415403 (9pp)
J Feng et al
2
the bonding interface disappears, but the high temperature still remains. This period can be regarded as the unloading stage [15–17]. With the thermal diffusion throughout the body of the composite plates, a cooling process, which can be consid-ered as the third stage, gradually occurs [18, 24]. However, until now the joining processes of the bonding interface in each stage have not been deeply investigated.
Molecular dynamics (MD) simulation, which can reveal atomic-scale structure evolution and interpret the relevant experiments at the microscopic level, is an effective method to clarify the joining process in each stage. Chen et al [15] inves-tigated atomic diffusion behavior of Al–Cu in the loading stage and unloading stage. In their simulations, the collisions with and without transverse velocity are respectively simulated. The results demonstrate that the atomic diffusion mostly takes place in the unloading stage. If there is no transverse velocity, the diffusion coefficient is directly proportional to the longi-tudinal velocity. When the longitudinal velocity is fixed, the diffusion coefficient is proportional to the square of the trans-verse velocity. Chen et al [16] also carried out the Ni50Ti50–Cu joining process to study the concentration distribution of the elements across the diffusion layer. Zhang et al [17] analyzed atomic diffusion behavior of Al–Mg in the loading stage and unloading stage, and proposed a mathematical formula to cal-culate the thickness of diffusion layer. Saresoja et al [25] mod-eled the inclined collision process of Cu–Fe, and observed the nanograins with crystal sizes of 10–20 nm near the bonding interface after the cooling process. Comparing with the exper-imental results, they all concluded that their simulations are in good agreement with the experiments [15–17, 25]. In the joining process, however, the influence of melting on the for-mation of the bonding interface has not yet been deeply ana-lyzed. What’s more, the evolution of atomic structures in the joining process has not been discussed.
In the past decades, copper-steel composite has received widespread attention, owing to the excellent properties of high corrosion resistance, high frictional properties and excellent mechanical strengths [26, 27]. Compared with other welding methods, a composite with large bonding surface and high joining strength can be directly acquired using explosive welding method. In this paper, the explosive welding process
of Fe–Cu bonding interface was modeled using MD simula-tion to clarify the joining mechanism. Section 2 of this presen-tation addresses the methodology related to MD simulations and the analysis methods. The results, about the morphologies of the bonding interface, the structure transformation and the tensile loading tests, are presented in section 3. Based on the simulation results, the discussion is proposed in section 4, fol-lowed by conclusions in section 5.
2. Methodology
The large-scale atomic/molecular massively parallel simu-lator (LAMMPS) was used to simulate the joining process of the bonding interface [28]. In MD simulation, the force field sometimes significantly affects the atomistic simulation results [29, 30]. To accurately model the joining process of the bonding interface, the basic phenomenon including disloca-tions, diffusion layer, melting and nanograins must be simu-lated. In our simulation, single crystalline Fe (body-centered cubic (BCC) structure) and Cu (face-centered cubic (FCC) structure) were selected using an embedded atom method alloy potential developed by Pasianot and Malerba [31]. This potential was mainly built to study the microstructure evo-lution and subsequent mechanical property change in reactor pressure vessel steels. Saresoja et al [25] also used this poten-tial to simulate the collision process of explosive welding, which is in accordance with the experiment. Therefore, it can be considered that this potential can be used to simulate the joining process of explosive welding. However, it should be noted that phase diagram of Fe from the magnetic-induced transitions to the FCC phase is not considered in this potential.
The configuration of the initial model is presented in figure 1. The lattice constants of the Cu (lCu) and Fe (lFe) are respectively 3.62 Å and 2.86 Å at 300 K. Because the incommensurate unit cell sizes of the Fe and Cu crystals, the system size in the x- and y -direction had to be chosen care-fully to avoid any severe deformation. In this simulation, 51 Fe unit cells and 40 Cu unit cells in the x- and y -direction were fitted resulting in a mismatch of about 0.34%. Totally the initial model contained 1042 604 atoms, and the dimen-sions were 145.8 Å × 145.8 Å × 291.6 Å for Fe and 145.3
Figure 1. Fe and Cu samples prepared for MD simulation.
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
3
Å × 145.3 Å × 290.5 Å for Cu. The contact surfaces of Cu and Fe are both (0 0 1) planes. A standoff distance of 2 nm was kept between the contact surfaces. In this Fe–Cu potential, the cutoff distance is 0.56 nm [31]. In the initial model, the standoff distance should be longer than the cutoff. To reduce the influence of the shock wave, on each end side along the z-axis, two transition regions consisting of 3 atomic layers were selected [15, 16]. Periodic boundary conditions were applied in the x and y directions, and free boundary conditions were applied to the z direction. For all simulations, a constant inte-gration time step of 2 fs was used.
Three stages, including the loading stage, the unloading stage and the cooling stage, were modeled respectively. The simulation procedures are summarized as the following. At the beginning of the following collision step, each Cu atom was given the same velocity to make the Cu bulk translate towards the Fe bulk, and at the same time, the transition region of Fe is fixed. Then the two bulks collide, while the length of the Cu bulk becomes shorter and shorter. When the Cu bulk reaches its minimum length, the transition region of Cu is fixed [16]. Then the system was relaxed for 1000 ps under the micro-canonical (NVE) ensemble. In the end, with the final equilibrium temperature of the NVE simula-tion kept, the system was relaxed for another 1000 ps at zero external pressure (NPT ensemble). Finally, the cooling pro-cess from the equilibrium temperature to 300 K was simu-lated, and the system was further equilibrated for 1000 ps at 300 K for relaxation. In explosive welding, the cooling rate mainly depends on the rate of the temperature redistribu-tion within the composite. However, due to the limitations of the sample size and simulation time in MD simulation, the cooling process cannot be accurately modeled. In order to avoid the influence of the high cooling rate, the system was thermalized from the equilibrium temper ature to 300 K for 50 ps every 50 K with a cooling rate of 1012 K s−1.
In this Fe–Cu potential, the melting points of Cu (TCu) and Fe (TFe) were respectively 1330 K and 1772 K, which are in very reasonable agreement with experiment (1358 K and 1812 K). Based on the equilibrium temperatures (Te) and the melting points of Fe and Cu, three statuses of the Fe–Cu bonding
interfaces, including solid–solid (Te < TCu < TFe), solid–liquid (TCu < Te < TFe) and liquid–liquid (TCu < TFe < Te), might be acquired in the joining process. In our simulation, impact velocities (Vz) from 1000 to 2000 m s−1 with an increment of 250 m s−1 were first attempted to acquire the equilibrium temperatures and pressures. Figures 2(a) and (b) respectively present the time history of the temperature and pressure of the system during the loading stage. As can be seen in figure 2, with the system’s kinetic energy changing into internal energy, the temperature and pressure of the system near the interface increase dramatically. After about 200 ps, the temperature and pressure achieve to a dynamic equilibrium state. The equilibrium temperature obtained by simulations are sequentially 672 K, 872 K, 1172 K, 1582 K and 1932 K. And the equilibrium pressure are respectively 17.82 GPa, 23.44 GPa, 29.68 GPa, 38.85 GPa and 43.75 GPa. At the impact velocities from 1000 to 1500 m s−1, melting does not occur, and solid–solid status of bonding interface might be acquired. At the impact velocity of 1750 m s−1, melting only occurs at the Cu side, and solid–liquid status of bonding interface might be acquired. However, at the impact velocity of 2000 m s−1, melting occurs at both the Cu and Fe sides, and liquid–liquid status of bonding interface might be acquired. To investigate the influence of melting on the forma-tion of the bonding interface, three types of the bonding inter-faces with the impact velocities of 1500 m s−1, 1750 m s−1 and 2000 m s−1 were respectively conducted and analyzed.
After simulation, Ovito software was used to analyze the joining process in each stage [32]. Radial distribution func-tion (RDF) was adopted to analyze the atomic structures of both the Cu and Fe in the joining process. And polyhedral template matching was used to investigate the micro-struc-tures of the bonding interface in each stage [33]. After equili-brated at 300 K for 1000 ps, tensile loading tests were applied to evaluate the mechanical properties of the bonding inter-faces. To generate the tensile deformation, the velocities of each boundary atoms are controlled as 30 m s−1. The strain rate for a loading step alone is approximately 1.5 × 109 s−1, which is several orders of magnitude higher than the rate in an experimental tensile test.
Figure 2. The temperature and pressure curves of the whole system during the loading stage.
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
4
3. Results
Using the MD simulation technique, the joining process of the Fe–Cu bonding interface, including the loading stage, the unloading stage and the cooling stage, were successfully mod-eled. At the same time, three types of bonding interfaces were obtained, depending on whether melting occurs on each side.
The morphologies of the three bonding interfaces in each stage are respectively shown in figure 3. As can be seen, in
the loading stage, the diffusion behavior is not obvious (fig-ures 3(a)–(c)), even though the equilibrium temperature is higher than the melting point. In the unloading stage, if the equilibrium temperature is lower than the melting point of both the Cu and Fe (Te < TCu < TFe), the diffusion layer is still not evident (figure 3(d)). However, when the equilibrium temperature is higher than the melting point, a diffusion layer can be observed in the unloading stage (figures 3(e) and (f)). If the equilibrium temperature is higher than the melting point
Figure 3. The morphologies of the bonding interfaces in each stage, when Vz = 1500 m s−1 (a), (d) and (g); Vz = 1750 m s−1 (b), (e) and (h); Vz = 2000 m s−1 (c), (f) and (i).
Figure 4. The RDF results of the Fe and Cu, when Vz = 1500 m s−1 at the Cu side (a) and the Fe side (b), Vz = 1750 m s−1 at the Cu side (c) and the Fe side (d), Vz = 2000 m s−1 at the Cu side (e) and the Fe side (f).
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
5
of the Cu but lower than the Fe (TCu < Te < TFe), the atoms mainly diffuse from the Fe side to the Cu side (figure 3(e)), and a thin diffusion layer with a thickness of less than 7 nm is formed. When the equilibrium temperature is higher than the melting points of both the Cu and Fe (TCu < TFe < Te), an inter-diffusion layer with a thickness of approximately 15 nm is generated (figure 3(f)). After the cooling process, if the equilibrium temperature is lower than the melting point (Te < TCu < TFe), a bonding interface without diffusion layer is generated (figure 3(g)). However, comparing the morpholo-gies in the unloading stage, a relatively thicker diffusion layer is formed in the cooling stage (figures 3(h) and (i)), when equilibrium temperature is higher than the melting point.
To investigate the atomic structures of both the Fe and Cu in each stage, the RDFs are analyzed, as shown in figure 4. In the initial models, the positions of the first two peaks sequen-tially represent their first neighbor distances and their lattice constants. Before welding, the lattice constants of the Cu and Fe are respectively 3.62 Å and 2.86 Å at 300 K, and the first neighbor distances are 2.56 Å and 2.48 Å at 300 K. In the loading stage, the neighbor distances move closer to each other, which is due to the high pressure. In the three simula-tions, the first neighbor distances of the Cu are 2.42 Å, 2.38 Å and 2.36 Å, and the first neighbor distances of the Fe are 2.34 Å, 2.32 Å and 2.31 Å. What’s more, both the Fe and Cu still
remain the solid states in the loading stage, even the temper-ature is higher than the melting points (figures 4(c), (e) and (f)). The atomic diffusion is restricted in the loading stage, which is due to the solid states of the Fe and Cu. However, in the unloading stage, with the disappearance of external pressure, the neighbor distances move almost to their original values. And liquid phases are observed in the unloading stage (figures 4(c), (e) and (f)), when the equilibrium temperature is higher than the melting points. After cooling to the room temperature, solid states are formed and the RDF results are almost similar with their initial models.
The micro-structures of the bonding interfaces in each stage are respectively revealed, as shown in figure 5. Because of the propagation of shock wave, in the loading stage, a few slip dislocations with an inclined angle 45°are formed at the Cu side (figure 5(a)). Increasing the impact velocities, more dislocations are produced at the Cu side (figures 5(a)–(c)). However, probably due to the BCC structures, dislocations are not observed at the Fe side. At the joining surface, a thin transition layer with disordered atomic structures is formed, owing to the mismatch of the atomic structures between the Fe and Cu. What’s more, at the Cu and Fe sides, the spreads of the shock wave change some atoms into the disordered structures. In the unloading stage, with the disappearance of the high pressure, some dislocations are eliminated at the Cu
Figure 5. The micro-structures of the bonding interface in each stage, when Vz = 1500 m s−1 (a), (d) and (g); Vz = 1750 m s−1 (b), (e) and (h); Vz = 2000 m s−1 (c), (f) and (i).
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
6
side, and more disordered atoms are generated (figure 5(d). When the equilibrium temperature is higher than the melting point, all the atoms transform into the disordered states due to the melting (figures 5(e) and (f)). After cooling to the room temperature, three types of bonding interfaces are formed. If the equilibrium temperature is lower than the melting points of both the Cu and Fe (Te < TCu < TFe), a bonding interface with some dislocations is formed at the Cu side, and a thin transition layer with disordered atoms is formed at the surface (figure 5(g)). If the equilibrium temperature is only higher than the melting point of the Cu (TCu < Te < TFe), a defective FCC system with stacking faults (two layers of HCP atoms) is generated at the Cu side, and dislocations with an inclined angle 45°are formed away from the stacking faults (figure 5(h)). However, if the equilibrium temperature is higher than the melting points of both the Cu and Fe (TCu < TFe < Te), nanograins and some disordered atoms are created near the bonding interface (figure 5(i)). At the Fe and Cu sides, the disordered atoms are mainly grain boundaries.
In the simulations, nanograins are generated in the cooling process (figure 5(i)), when the equilibrium temperature is higher than the melting points of both the Cu and Fe. To investigate the forming process of nanograins in figure 5(i), during the cooling process, the ratio of atomic structures of the Cu and Fe are analyzed, as shown in figures 6(a) and (b). At the Cu side, during the cooling process from 850 K to 300 K, the numbers of the FCC, BCC and HCP structures begin to increase with the decrease of disordered atoms. Similarly, at the Fe side, during the cooling procedure from 950 K to 300 K, the quantities of BCC structure start to increase with the vanishing of disordered atoms. After cooling to 300 K, at the Cu side, the ratios of FCC, BCC, HCP and disordered atoms are respectively 44.41%, 5.65%, 18.83% and 31.11%. At the Fe side, the fractions of BCC and disordered atoms are respectively 86.32% and 12.17%, and the percentages of FCC and HCP structures are almost zero.
After cooling to the room temperature for 1000 ps, ten-sile loading tests were performed, and the nominal tensile stress–strain curves of the three bonding interfaces are ana-lyzed in figure 7. As can be seen, in the tensile loading tests,
the maximum tensile strengths of the bonding interfaces are 10.42 GPa, 9.69 GPa and 10.53 GPa, and the failure strains in this moment are 0.069, 0.071 and 0.086 respectively. According to the tensile stress–strain curves of the three bonding inter-faces, it is very interesting to note that, without the occur-rence of melting in the joining process (Vz = 1500 m s−1), the bonding interface can still be joined together. What’s more, in the three types of bonding interfaces, the interface with the formation of nanograins and the diffusion layer (Vz = 2000 m s−1) possesses the best mechanical property with a highest tensile strength and failure strain.
To gain a better insight into the observed mechanical behaviors, we now turn our attention to the deformed structures inside the Fe–Cu samples. Figure 8 shows the deformed configurations of the bonding interfaces at dif-ferent strains. Before the stress reaches the maximum, struc-tural change is not obvious, except for the elongation of the samples (ε = 4.8% in figure 8(a), ε = 4.8% in figure 8(b) and ε = 4.8% in figure 8(c)). After reaching the maximum stress, the deformation behaviors of the bonding interfaces
Figure 6. The fractions of atomic structures of Cu (a) and Fe (b) during the cooling process in figure 3(i).
Figure 7. Tensile stress–strain curves of the three bonding interfaces.
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
7
become very different. If melting does not occur, as strain is further increased, several small voids form in the joining sur-face (ε = 7.6% in figure 8(a)). Then the voids grow rapidly along the bonding surface, forming cracks as deformation proceeds (ε = 8.4% in figure 8(a)). In this situation, brittle fracture takes place and a flat fracture surface (ε = 12.0% in figure 8(a)) can be observed, which is due to the forma-tion of the amorphous layer at the bonding surface. However, if melting occurs, larger voids occur at the Cu side near the diffusion layer (ε = 8.4% in figure 8(b) and ε = 10.0% in figure 8(c)), and necking deformation takes place before frac-ture (ε = 16.0% in figure 8(b) and ε = 40.0% in figure 8(c)). In this condition, ductile fracture takes place and a rough frac-ture surface can be observed (ε = 24.0% in figure 8(b) and ε = 50.0% in figure 8(c)), which is because of the generation of the diffusion layer at the bonding surface. The lower tensile strength of the bonding interface (Vz = 1750 m s−1) is because of some atoms at the Fe side transformed into FCC structures (ε = 8.4% in figure 8(b)) under the influence of dislocations near the interface.
4. Discussion
4.1. Joining mechanism
Explosive welding is generally considered as pressure welding [3, 7], fusion welding [6] and diffusion welding [15, 16]. However, our MD simulations demonstrate that melting behavior is not a necessary factor in explosive
welding. Based on whether melting occurs, the joining mech-anism can be summarized as pressure welding or fusion-diffu-sion welding, as shown in figure 9. During the drastic collision of explosive welding, jet will be ejected in front of the col-lision point, which will strip away the surface contaminant, oxides and impurities [8, 14]. After the drastic collision, the generated high pressure can provide tight contact between the joining surfaces. In order to form the bonding interface, the atoms from each bonding surface must be brought together at atomic scale. If melting does not occur, the generated high pressure at the collision point helps to bring and keep the atoms together. In this situation, the joining mechanism can be regarded as pressure welding. Durgutlu et al [34] acquired the copper/steel composite using explosive welding method. Melting and diffusion are not observed in their bonding inter-face under the impact velocities between 1120–1453 m s−1, which is in good agreement with our simulation (Vz = 1500 m s−1). However, higher impact velocities (>1500 m s−1), which probably results in melting, are not performed in their experiments. If melting occurs, the formation of the diffusion layer joins the interface together. In this condition, the joining mechanism can be considered as fusion-diffusion welding.
4.2. Joining procedure
Although explosive welding is an extremely fast process, it is not instantaneous. The joining process of the bonding interface can be divided into three stages, including the loading stage, the unloading stage and the cooling stage. In the loading stage,
Figure 8. The deformed configurations of the bonding interfaces at different strains, when Vz = 1500 m s−1 (a), Vz = 1750 m s−1 (b) and Vz = 2000 m s−1 (c). Atoms are colored according to their coordination number.
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
8
the bonding surfaces come to a close contact due to high-speed collision which leads to increase of both temperature and pres-sure. The high pressure, which seriously restricts the atomic migration and increases the melting points, prevents the for-mation of the liquid phase. The lower diffusion rate at this stage do not allow formation of a diffusion layer. However, high pressure provides a tight contact between the welding surfaces. It should be noted that, if particularly higher impact velocities (Vz > 2000 m s−1) are selected, melting probably will happen in the loading stage. In the unloading stage, the pressure near the bonding interface disappears, but the high temperature still remains. If the temperature at the interface exceeds the melting points, liquid phase accelerates the diffu-sion coefficient and forms the diffusion layer. In the simulation performed by Chen et al [15, 16], they also proved that melting occurs in the unloading stage, and the higher diffusion coef-ficient results in the formation of the diffusion layer. In the unloading stage, the atoms mainly diffuse into the liquid side. After the cooling stage, the plates are tightly joined together. Based on whether melting occurs at each side, three types of Cu–Fe bonding interfaces can be identified. If melting does not take place, a bonding interface with some dislocations is gen-erated. If melting only occurs at the Cu side, stacking fault and dislocations are formed. When melting happens at both sides of the interface, nanograins are generated near the bonding interface during the cooling process.
4.3. Nanograin formation
In our simulations, nanograins are generated near the interface. In previous experimental studies, for some other composite plates fabricated by explosive welding, such as steel-Zr [35], Al–Mg [36], Ti–Ti [37] and Cu-metallic glass [38], nanograins are also observed near the bonding interface. Saresoja et al [25] considered the formation of nanograins is due to the severe plastic deformation. Yang et al [37] considered it is the high cooling rate, high pressure and high shear stress that create the nanograins. However, according to our simulations,
we can propose that when melting occurs at both sides of the bonding interface, it is the melting and the subsequent cooling process that produce the nanograins. Crystallization occurs at the temperature of approximately 850 K–950 K, which is due to the high cooling rate (1012 K s−1) in our MD simulations. High cooling rate also produces some amorphous structures near the bonding interface [39–41]. The existence of amor-phous state near the bonding zone has also been observed in various composite plates [36–38]. In explosive welding, the cooling rate mainly depends on the rate of temperature redis-tribution within the composite. According to the numerical model shown in Bataev et al [18], the cooling rate during the solidification of the liquid phase in explosively welded Cu/Cu bimetal was in the range 104–107 K s−1. In the numerical simulation performed by Liu et al [24], the cooling rate of the temperature is about 108 K s−1 in the explosive welding prog-ress. In the MD simulation, if the cooling rate in the range of 104–108 K s−1 is selected, maybe crystallization occurs at the higher temperature (>900 K).
5. Conclusions
In this paper, the joining mechanism of the Fe–Cu bonding interface in explosive welding was investigated through MD simulation. The results are summarized as follows.
(1) Melting is not a necessary factor in forming the bonding interface. Depending on whether melting takes place, the joining mechanism of the bonding interface is regarded as pressure welding or fusion-diffusion welding. If melting does not occur, the joining mechanism is regarded as pressure welding. When melting takes place, the joining mechanism is considered as fusion-diffusion welding.
(2) The joining process of the bonding interface can be divided into three stages. In the loading stage, high pres-sure prevents the atomic diffusion and keeps the Fe and Cu in a solid state. The atomic diffusion mainly takes place in the unloading stage, when melting occurs near
Figure 9. The joining mechanism of the bonding interface.
J. Phys.: Condens. Matter 31 (2019) 415403
J Feng et al
9
the bonding interface. After the cooling stage, depending on whether melting happens at each side, three types of bonding interfaces are obtained.
(3) The formation of nanograins is due to the melting and the subsequent cooling. The interface with the formation of nanograins possesses the best mechanical property with a highest tensile strength and failure strain.
Acknowledgments
This work was supported by the National Natural Science Foundation of China under grants No. 11521062 and 11472054, the Project of SKLEST under grants ZDKT18-01 and KFJJ18-05M, China Postdoctoral Science Foundation funded project under grant No. 2019M650504.
References
[1] Blazynski T Z 1983 Explosive Welding, Forming and Compaction (London: Applied Science Publishers)
[2] Crossland B 1982 Explosive Welding of Metals and its Applications (New York: Oxford University Press)
[3] Crossland B and Williams J D 1970 Metall. Rev. 15 79 [4] Reid S R 1974 Int. J. Mech. Sci. 16 399 [5] Bahrani A S and Crossland B 1964 Proc. Inst. Mech. Eng. C
179 264 [6] Wronka B 2010 J. Mater. Sci. 45 4078 [7] Zheng Y M 2008 The Principle and Application of Explosive
Welding and Metallic Composite (Changsha: Central South University Press)
[8] Mousavia A A A and Al-Hassani S T S 2005 J. Mech. Phys. Solids 53 2501
[9] Mousavi A A A, Burley S J and Al-Hassani S T S 2005 Int. J. Impact Eng. 31 719
[10] Mousavia A A A and Al-Hassani S T S 2008 Mater. Des. 29 1[11] Li X J, Mo F, Wang X H, Wang B and Liu K X 2012 Sci.
Technol. Weld. Join. 17 36[12] Wang Y X, Beom H G, Sun M and Lin S 2011 Int. J. Impact
Eng. 38 51[13] Wang X, Zheng Y Y, Liu H X, Shen Z B, Hu Y, Li W,
Gao Y Y and Guo C 2012 Mater. Des. 35 210
[14] Feng J, Chen P, Zhou Q, Dai K, An E and Yuan Y 2017 Int. J. Multiphys. 11 315
[15] Chen S Y, Wu Z W, Liu K X, Li X J, Luo N and Lu G X 2013 J. Appl. Phys. 113 044901
[16] Chen S Y, Wu Z W and Liu K X 2014 Chin. Phys. B 23 066802
[17] Zhang T T, Wang W X, Zhou J, Cao X Q, Xie R S and Wei Y 2017 Acta Metall. Sin. 30 983
[18] Bataev I A, Lazurenko D V, Tanaka S, Hokamoto K, Bataev A A, Guo Y and Jorge A M Jr 2017 Acta Mater. 135 277
[19] Nassiri A and Kinsey B 2016 J. Manuf. Process. 24 376[20] Nassiri A, Chini G, Vivek A, Daehn G and Kinsey B 2015
Mater. Des. 88 345[21] Aizawa Y, Nishiwaki J, Harada Y, Muraishi S and Kumai S
2016 J. Manuf. Process. 24 100[22] Li Y, Liu C, Yu H, Zhao F and Wu Z 2017 Metals 7 407 [23] Grignon F, Benson D, Vecchio K S and Meyers M A 2004 Int.
J. Impact Eng. 30 1333[24] Liu K X, Liu W D, Wang J T, Yan H H, Li X J, Huang Y J,
Wei X S and Shen J 2008 Appl. Phys. Lett. 93 081918[25] Saresoja O, Kuronen A and Nordlund K 2012 Adv. Eng. Mater.
14 265[26] Zhang H, Jiao K X, Zhang J L and Li J 2018 Mat. Sci. Eng. A
731 278[27] Liu Y B, Li C, Hu X Y, Yin C F and Liu T S 2019 Metals 9 1[28] Plimpton S 1995 J. Comput. Phys. 117 1[29] Chandler A B, Francesca T, Zachary T T and Robert A B M
2013 Curr. Opin. Solid State Mater. Sci. 17 277[30] Etesami S A, Laradji M and Asadi E 2019 Model Simul.
Mater. Sc. 27 025005[31] Pasianot R C and Malerba L 2007 J. Nucl. Mater. 360 118[32] Stukowski A 2010 Model. Simul. Mater. Sci. 18 015012[33] Larsen P M, Schmidt S and Schiotz J 2016 Model. Simul.
Mater. Sci. 24 055007[34] Durgutlu A, Gulenc B and Findik F 2005 Mater. Des. 26 497[35] Paul H, Morgiel J, Baudin T, Brisset F, Prazmowski M and
Miszczyk M 2014 Arch. Metall. Mater. 59 1129 [36] Chen P W, Feng J R, Zhou Q, An E F, Li J B, Yuan Y and
Ou S L 2016 J. Mater. Eng. Perform. 25 2635[37] Yang Y, Wang B and Xiong J 2006 J. Mater. Sci. 41 3501[38] Feng J R, Chen P W and Zhou Q 2018 J. Mater. Eng. Perform.
27 2932[39] Feng J R, Chen P W and Li M 2018 Phys. Rev. B 98 024201[40] Feng J R, Chen P W and Li M 2018 J. Phys.: Condens. Matter
30 255402[41] Feng J R, Chen P W and Li M 2018 J. Appl. Phys. 124 165901
J. Phys.: Condens. Matter 31 (2019) 415403