2021 Volume 62 Issue 9 Pages 1376-1385
Two types of intermediate layer (IML) were formed at the magnetic pulse welded Al/Fe joint interface. To clarify the formation mechanism of IML, the cooling process at the joint interface and formation process of IML were analyzed by the simulation and experiment. The simulation results showed that the crest zone of wavy interface underwent an extremely high temperature rise during the collision process, the high temperature leads to a local melting behavior at the joint interface. After the collision, a large temperature difference of 1000∼1500 K between joint interface and base metals caused a rapid cooling occurred at the interface with a cooling rate reached around 107∼109 K/s. Meanwhile, local melting zone (LMZ) changed into IML during a rapid solidification that proceeded from the outside and finished at the center area of LMZ. The morphology and composition percent of IML calculated by the numerical analysis was quantitatively in good agreement with experimental analysis results. A reasonable explanation has been made on the formation mechanism of IML at magnetic pulse welded Al/Fe joint interface.
Magnetic pulse welding (MPW) is a method of joining metal plates by oblique collision at several hundred meters per second using electromagnetic force. Using the MPW method, it is possible to join metal combinations that are difficult to join by fusion welding. It is known that the joint interface morphology of dissimilar metal varies greatly depending on the density difference of metals, impact velocity of flyer plate and collision angle between two metal plates. The magnetic pulse welded Al/Fe joint interface usually presents a trigger-like wavy morphology, and usually intermediate layer (IML) is formed intermittently at the joint interface.1,2) When a hard and brittle IML is excessively generated along the joint interface, the bonding strength tends to decrease. Therefore, to prevent the decrease in bonding strength due to the excessive formation of IML, it is necessary to clarify its formation process as to control or suppress its formation.
In our previous report,3) an A1050-O aluminum plate and a low carbon steel (SPCC) plates were seam welded by magnetic pulse welding (MPW). Collision angle between two plates and impact velocity of Al plate driven by electromagnetic force was numerically solved by coupled mechanical and electromagnetic fields analysis. The metal jet emission behavior, wavy interface formation process and the temperature change during MPW collision process were analyzed by SPH method. The numerical results showed that the temperature near the interface exceeds the melting point of both Al and Fe at extremely high pressure, which leads to the formation of a local melting zone (LMZ) where Al and Fe are mixed in molten state. To clarify the formation process of IML, it is necessary to clarify the cooling process of the joint interface right after the collision process. However, in MPW, joining is completed in an extremely short time of several microseconds. Therefore, it is difficult to observe the cooling process of the joint interface and IML formation process experimentally. Conventionally, the cooling process can only be inferred from the results of microstructure observation after joining.
Crossland4) reported that the cooling rate at explosive welded joint interface was 105∼107 K/s based on the observation results of the microstructure of joint interface. Bataev et al.5) developed an approach to calculate the cooling rate near the interface in explosive welding by solving the heat equation numerically using the finite difference method. They use microstructures observation results as the initial conditions for solving the differential equation. The highest cooling rate near the joint interface is around 106∼107 K/s. Sapanathan et al.1) using thermo-mechanical simulation method calculated the cooling rate near the magnetic pulse welded Al/Al joint interface, which was on the order of 109 K/s.
However, study on the cooling rate of the impact welded interface is not enough, especially for material combination with a IML with complicated composition formed at the joint interface such as Al/Fe, there are still insufficient reports concerning cooling rate at the interface and the formation process of IML formed at the interface.
In this study, the cooling process of the magnetic pulse welded Al/Fe joint interface and the accompanying IML formation process right after the collision process were investigated by numerical analysis with OpenFOAM. The morphology of the joint interface obtained by numerical analysis was compared with the microstructure observation results to discuss the formation process of IML in Al/Fe joint.
In the MPW experiment, the flyer plate used the pure aluminum 1050-O plate (200 × 70 × 0.4 mm) and parent plate used the low carbon steel SPCC plate (200 × 70 × 0.3 mm). The chemical composition of 1050-O and SPCC plate can be found in the previous report.3) They are referred to as the Al and Fe hereafter. MPW experiment is conducted by Bmax MP12.5/25 system (capacitance: 40 µF). The MPW setup is shown in Fig. 1. The initial distance between the flyer plate and parent plate is called ‘Gap’, which affects the collision angle between two plates (β) and the impact velocity of the flyer plate (v). The discharge energy (DE) mainly controls the kinetic energy, which determines v. This study used the same conditions as the previous report with that Gap is set as 2 mm and DE is set as 6 kJ.
Setup for magnetic pulse welding.
Specimens for microstructure observation and composition analysis were collected from the tensile shear tested joints and cut in cross-section along the welding direction. The microstructural observation of the joint interface was carried out using a field emission scanning electron microscope (FE-SEM: JEOL JSM-7500F). The chemical composition of the interface microstructure was measured by an energy dispersive X-ray spectroscopy (EDS) attached to SEM KEYENCE VE-9800.
In the previous report,3) the collision process of the plate in the electromagnetic field was reproduced, β and v were obtained by the Emag-Mechanical method. The initial collision angle β0 and impact velocity v0 were 15° and 619 m/s, respectively (as shown in Fig. 2). Here, the subscript 0 is used to represent the status when flyer plate starting to collide with parent plate. β0 and v0 were used as the initial conditions of an analytical model using the Smoothed Particle Hydrodynamics (SPH) method as a solver to reproduce the formation process of wavy joint interface (as shown in Fig. 3). By analyzing the results of SPH model, it is possible to obtain information on the wavy interface morphology (material distribution), the temperature distribution, the location and composition of LMZ immediately after the formation of the wavy interface (as shown in Fig. 4). However, heat conduction is not considered in the SPH solver of AUTODYN, which is the analysis code of the SPH model. Therefore, it is not possible to analyze the temperature change after the formation of the wavy interface, that is, the cooling process of the material at the joint interface.
Initial impact velocity v0 (vx0, vy0) of flyer plate and initial collision angle β0 at first impact point calculated by Emag-Mechanical method.3)
Collision process between flyer plate and parent plate reproduced by SPH model.3)
(a) Temperature distribution near the interface; (b) Location of local melting zone.3)
After the collision process in impact welding, the joint interface, which has become high temperature due to the formation of the wavy interface, should then be cooled by heat conduction to the base material, which is almost at room temperature. As the result, cooling of the joint interface and associated solidification of the LMZ should occur. Therefore, a new analytical model considering heat conduction was created using laplacianFoam in OpenFOAM.6) OpenFOAM (Open-source Field Operation and Manipulation) is a C++ toolbox for the development of customized numerical solvers for the solution of continuum mechanics problems.7) This model was conducted to reproduce the cooling process at the joint interface and the solidification process in the LMZ using the material and temperature distributions obtained from the SPH model as the initial conditions. The laplacianFoam in OpenFOAM is a solver that uses finite volume method (FVM) to discretizes the Laplace equation, which is represented by the heat diffusion equation for solids:
| \begin{equation} \partial T/\partial t + \mathit{\nabla}^{2} (\alpha,T) = 0 \end{equation} | (1) |
Figure 5(a) shows the flowchart of the analysis using the LF model. The first column of the flowchart in Fig. 5(a) shows the initial state of the LF model (material distribution (ρ) and temperature distribution (T) obtained from the SPH model). The material identification number (C), thermal diffusivity (α), and T in LF model are converted from ρ and T obtained from the SPH model by the method shown in shown in Fig. 5(b). The figure on the left side of Fig. 5(b) shows the SPH particles of the SPH model. Focusing on SPH particles A∼D, each SPH particle is either material 1 or 2 used for joining and has temperatures TA∼D and densities ρ1∼2, respectively. Here, C = 1 means material 1, C = 2 means material 2, and C = 3 means LMZ.
(a) Flowchart of the calculation cycle for the thermal conductivity analysis, (b) Conversion from SPH model to LF model, (c) Update method of the material number and thermal diffusivity in the calculation cycle.
A 1 × 1 µm grid equal to the smoothing length (h = 1 µm) of SPH particles is fitted to the SPH particle distribution, and the density and temperature at the intersection of the meshes in each particles A∼D were read using the line trace function of AUTODYN. The temperature and density of each particle were read into the mesh of the LF model. Considering the following case as an example of this operation: the temperatures of SPH particles A∼D are 1000, 500, 2000, 1500 K, respectively, the melting point of material 1 (Tm1) is 1800 K, and the melting point of material 2 (Tm2) is 900 K. Focusing on SPH particle A, the density is equal to material 1, but the temperature T1 does not exceed Tm1. Therefore, SPH particle A is regarded as solid phase material 1 and initial conditions for the mesh in this region are defined as T = TA, α = α1, and C = 1. Similarly, T = TB, α = α2, and C = 2 are defined as initial conditions for the mesh corresponding to SPH particle B. Next, focusing on SPH particles C and D, mesh C is determined to be material 2 and mesh D is determined to be material 1 based on the density. Furthermore, since the temperatures TC and TD exceed Tm1 and Tm2, respectively, the SPH particles of C and D can be regarded as melted. Therefore, for meshes C and D, α = α3, C = 3 are defined and these regions are considered as LMZ. By performing such conversion work on all SPH particles of the SPH model, the material identification number, temperature distribution, and thermal diffusivity, which are the initial conditions of the LF model, can be obtained from the temperature distribution and density distribution (material distribution) obtained by the SPH model.
As shown in the blue dotted line in Fig. 5(a), the LF model can reproduce the change of the temperature at the joint interface by solving the nonstationary thermal diffusion equation. As the temperature drops, it is necessary to update the material identification number and thermal diffusivity as shown in Fig. 5(c). The mesh C in Fig. 5(c) is shown as an example. Consider the case where the temperatures TA∼D of each mesh drops to the temperatures T′A∼D during to cooling. The temperature of the mesh C (T′C) is 750 K, and the solidification temperature of LMZ (Tm3) is 800 K. Tm3 was set to a certain temperature based on the composition of LMZ obtained by the SPH model and the equilibrium phase diagram of the metals used for joining. Since mesh C is defined as LMZ and T′C < Tm3, mesh C is considered to be solidified, and the material identification number and the thermal diffusivity of mesh C are updated as C = 4 and α = α4 of IML. In this way, the formation process of IML can be simulated.
3.2 Conversion of analysis results from AUTODYN to OpenFOAMFigure 6 shows the converted results from SPH model to LF model. The left side of Fig. 6(a), (c) shows the temperature distribution and material distribution at the joint interface after welding at 1.4 µs reproduced by the SPH model. SPH particles with h = 1 µm in the SPH model in one wave surrounded by a red frame in Fig. 6(a), (c) were converted into a 1 × 1 µm mesh in the LF model as shown in Fig. 6(b), (d). During the conversion from the SPH model to the LF model, particles exceeding the melting points of Al and Fe are defined as the mesh of LMZ (red in the figure). Therefore, the initial state of the LF model consists of green meshes for Al, blue meshes for Fe, and red meshes for LMZ.
Definition of the conversion area and initial time of the thermal conductivity analysis.
In this way, the temperature distribution and material distribution at 1.4 µs of the SPH model were defined as the initial state of the LF model (0 µs of the LF model), and analysis using the LF model was conducted based on the temperature distribution, material distribution, and thermal diffusivity. As a result of the preliminary analysis, it was found that the temperature change after 7 µs is extremely small under the welding conditions used in this study. Therefore, this analysis focused on the cooling process of the joint interface and the formation process of IML between 0∼7 µs.
3.3 Estimation of thermal diffusivity 3.3.1 Thermal diffusivity of Al and FeThe initial conditions for solving the thermal diffusion equation are the temperature distribution at the joint interface, the thermal diffusivity, and the solidification temperature of the IML. The thermal diffusivity of the solid phase Al and Fe, LMZ, IML and the solidification temperature of IML are determined as follows. First, the thermal diffusivities of the solid and liquid phases of Al and Fe and their temperature dependence are determined by referring to volume 10 of Thermophysical Properties of Matter-The TPRC Data Series,8) which contains the thermal diffusivities of various materials. Figure 7 shows the temperature dependence of the thermal diffusivity of the solid and liquid phases of Al and Fe. The relationship between the thermal diffusivity and temperature of the solid and liquid phases of Al and Fe in Fig. 7 can be approximated by the linear relationship shown in eq. (2)–(6).
| \begin{equation} \alpha_{\text{Al,solid}} = -4.81 \times 10^{-8}T + 1.13 \times 10^{-4} \end{equation} | (2) |
| \begin{equation} \alpha_{\text{Al,liquid}} = 8.14 \times 10^{-9}T + 3.07 \times 10^{-5} \end{equation} | (3) |
| \begin{equation} \alpha_{\text{${\alpha}$-Fe,solid}} = -2.47 \times 10^{-8}T + 2.82 \times 10^{-5} \end{equation} | (4) |
| \begin{equation} \alpha_{\text{${\gamma}$-Fe,solid}} = 1.15 \times 10^{-9}T + 4.53 \times 10^{-6} \end{equation} | (5) |
| \begin{equation} \alpha_{\text{Fe,liquid}} = 1.41 \times 10^{-9}T + 5.01 \times 10^{-6} \end{equation} | (6) |
Thermal diffusivity of the Al (solid and liquid) and Fe (solid and liquid).8)
The composition of LMZ is necessary for determining its thermal diffusivity. The composition of LMZ can be predicted from the ratio of the number of SPH particles between Al and Fe in the LMZ obtained from the analysis result in Fig. 4(b). The number of Al and Fe particles in LMZ (NAl and NFe) were calculated as 661 and 125, respectively. The number densities of Al and Fe (nAl and nFe) were calculated using eq. (7), and they were 0.84 and 0.16, respectively.
| \begin{equation} n_{\text{Al or Fe}} = N_{\text{Al or Fe}}/(N_{\text{Al}} + N_{\text{Fe}}) \end{equation} | (7) |
The thermal diffusivity of Al and Fe in the solid phase can be expressed by eqs. (2), (4), (5). However, in the LMZ shown in Fig. 4(b), Al and Fe are mixed, and the thermal diffusivity in the liquid phase of Al and Fe cannot be directly applied to this zone. The thermal diffusivity of the melted alloy of Al and Fe is not recorded in the data book.8) In this study, the thermal diffusivity of Al and Fe in the liquid phase shown in eqs. (3) and (6) and the number density of Al and Fe in the LMZ was calculated and added linearly to estimate the thermal diffusivity of the LMZ (as shown in eq. (8)):
| \begin{equation} \alpha_{\text{LMZ}} = \alpha_{\text{Al,liquid}} \times n_{\text{Al}} + \alpha_{\text{Fe,liquid}} \times n_{\text{Fe}} \end{equation} | (8) |
The thermal diffusivity of IML will be examined. Since it is considered that the LMZ at the Al/Fe joint interface solidifies to form an IML, the average composition of LMZ and the average composition of IML should be the same. However, the thermal diffusivity of IML that matches the composition of LMZ obtained by the SPH model has not been reported so far. Therefore, we considered that the thermal diffusivity of IML (αIML) could be obtained from the thermal conductivity (kIML), density (ρIML) and specific heat of IML (Cp,IML) as shown in eq. (9):
| \begin{equation} \alpha_{\text{IML}} = k_{\text{IML}}/\rho_{\text{IML}}C_{\text{p,IML}} \end{equation} | (9) |
The thermal conductivity and specific heat of the intermediate phase of the Al–Fe system (FeAl3) at 300 K, which is close to the composition of the IML (79%Al–21%Fe, atomic ratio) estimated from the particle number ratio of SPH, are given as k = 4 W/mK and Cp = 600 J/kgK, respectively.9) Therefore, these values are used here as kIML and Cp,IML. ρIML was calculated to be 3528 kg/m3 by eq. (10) using the density ρ and number density n of Al and Fe.
| \begin{equation} \rho_{\text{IML}} = \rho_{\text{Al}} \times n_{\text{Al}} + \rho_{\text{Fe}} \times n_{\text{Fe}} \end{equation} | (10) |
Based on kIML, Cp,IML, ρIML, the value of αIML was estimated to be 1.89 × 10−6 m2/s using eq. (9). The temperature dependence of thermal conductivity and specific heat of FeAl3 has not yet been reported. Therefore, in this analysis, the thermal diffusivity of the IML is treated as a constant.
3.4 Estimation of the solidification temperature of local melting zoneWhen the cooling process of the joint interface after the formation of the wavy interface is reproduced using the LF model, the solidification process of LMZ can be reproduced by defining the solidification temperature at which LMZ solidified and became an IML. Figure 8 is the equilibrium phase diagram of the Al–Fe binary system,10) as indicated by the red dashed line, the temperature at which all the alloy melt in the LMZ (79%Al–21%Fe, atomic ratio) become solid phase is 928 K. Although the actual composition of IML is probably differs from location to location and is considered inconstant, due to the limitation of numerical analysis method, 928 K was set as the solidification temperature of IML for convenience.
Al–Fe binary equilibrium phase diagram and estimation of solidification temperature.10)
Figure 9 shows the FE-SEM backscattered electron images and EDS composition analysis result of IML formed at the magnetic pulse welded Al/Fe joint interface. As reported by the previous report,3) this interface region contained two different type of IMLs with different contrast: the IML with lighter contrast was surrounded by Fe, while the IML with the darker contrast was sandwiched between the Fe and Al plate. This kind of IML with two different contrast was also confirmed at the Al/Fe joint interface by explosive welding.11) Based on the formation location, the two kinds of IML were defined as Tail-side layer (TSL), which formed at the rear of the wave interface, and Front-side layer (FSL), which formed at the front of the wave interface. A crack was found vertically impenetrated the TSL while some voids were found formed within the FSL. The similar phenomena of crack and void formed in the IML was also reported in the explosive welded interface.6) The mapping results suggested that TSL has a higher percent of Fe than FSL. The spot analysis results in Fig. 9(b)–(c) indicated that the average composition of TSL is about 54%Al–46%Fe, while that of FSL is about 85%Al–15%Fe (both in atomic ratio).
FE-SEM backscattered electron images and EDS mapping and point analysis of IML (atomic ratio).
Figure 10 shows the cooling process of the joint interface during 7 µs reproduced using the LF model. During 0∼0.5 µs, a high temperature region above 1500 K was observed only at the crest zone and near the interface, while the temperature of the base metal of Al and Fe far from the interface did not increase much. Thermal diffusion of Al and Fe occurred rapidly from the vicinity of the joint interface to the base metal. Furthermore, there is an area of 20∼30 µm width on both sides from the joint interface where the temperature exceeds 400 K. At 1 µs, the temperature near the center of the crest zone decreased to about 1200∼1400 K. Compared to 0.5 µs, the region where the temperature of the joint interface is above 400 K expands to about 20 µm in width on the Fe side and 40 µm in width on the Al side from the joint interface. During 2∼3 µs, the temperature near the center of the crest zone decreased to about 1000∼1200 K. The region where the temperature of the joint interface is above 400 K is wider than that at 1 µs, with a width of about 30 µm on the Fe side and 50 µm on the Al side from the joint interface. During 5∼7 µs, the temperature of the crest zone and the vicinity of the interface dropped to 700∼900 K. The region where the temperature of the joint interface is above 400 K is in the range of 30∼40 µm on the Fe side and in the range of 40∼50 µm on the Al side from the joint interface.
Cooling process of the joint interface during 7 µs.
To further investigate the change of temperature during cooling process, six reference points 1–6 (as shown in Fig. 11(a)) were selected to investigate the change of temperature and cooling rate at the specific location. 1, 3, 5 and 2, 4, 6 are located within Fe and Al, respectively. These six reference points are corresponding to the six gauge points preset in SPH model as described by the previous report. Reference points 1, 3 are located at the location where TSL is formed while 2, 4 are located at the location where FSL is formed. 1∼4 is 0∼4 µm away from joint interface while 5 and 6 are located 20 µm away from the joint interface.
Change of temperature and cooling rate at each reference points during 0∼7 µs.
Figure 11(b) shows the change of temperature at points 1∼6. Figure 11(c) shows the change of the cooling rate between 0∼7 µs in 1∼4 obtained from (b). At reference point 1, during 0∼7 µs, the temperature dropped from approximately 2100 K to 800 K, with a particularly sharp drop between 0∼0.5 µs. At point 2, the temperature decreased from approximately 1600 K to 600 K during 0∼7 µs, and similar with point 1, the temperature decreased rapidly between 0∼0.5 µs. The cooling rate at point 1 is on the order of 107∼1010 K/s, which is similar to point 2. This indicates that Fe shares the same cooling rate with Al at TSL location. The temperature at point 3 dropped from 1500 K to 900 K on the order of 108∼109 K/s while temperature at point 4 dropped from 1100 K to 600 K on the order of 107∼108 K/s, which shows that Fe is cooling more faster than Al at the FSL location. At reference point 5 and 6, the temperature increased a little from 330 K to 370 K and 400 K to 430 K, respectively. Also, Fig. 11(c) shows that at all the reference points, the cooling rate between 0∼0.5 µs is higher than that between 0.5∼7 µs.
In summary, based on the numerical analysis results of SPH model and LF model, it was found that a wavy interface is formed within a few microseconds at the magnetic pulse welded joint interface, and the temperature rises rapidly especially at the crest zone of the wavy interface. This rapid temperature increase occurs in a very limited area near the joint interface, and the temperature of the Al and Fe base materials hardly increases at several hundred micrometers away from the interface. A temperature difference of 1000∼1500 K occurs between the joint interface and the surrounding base material with a width of several tens of micrometers, and under this extremely large temperature gradient, rapid heat removal from the joint interface to the base material occurs, and the area near the joint interface rapidly cools at a cooling rate on the order of 107∼109 K/s.
4.3 Formation process of the intermediate layer at the joint interfaceFigure 12 shows the solidification process in LMZ between 0∼7 µs. In this figure, green indicates solid phase Al and blue indicates solid phase Fe. The region where the temperature is higher than the melting point of Al and Fe is regarded as LMZ and is shown in red. When the temperature of LMZ is less than the solidification temperature, the region is considered to be solidified and become IML. The IML is shown in gray.
Solidification process of the local melting zone.
Between 0∼0.5 µs, the range of LMZ at the joint interface slightly expanded. This is because the temperature of the Al and Fe base materials near the interface increased due to thermal diffusion from the interface. Also, during 0.5∼1 µs, LMZ began to solidify from the side closer to the Al base metal. At 2 µs, the outside of the LMZ located at the crest zone of the Al wave began to solidify. During 2∼5 µs, solidification proceeded from the outside to the center of the crest zone. At this stage, a small part of LMZ still remained in the center of the crest zone, which indicated the final solidification zone of LMZ. At 7 µs, the solidification of LMZ was completed, and an IML was formed at the position of LMZ. The thickness of IML was not constant and varied in the range of approximately 2∼12 µm.
If the composition of LMZ is almost constant, the solidification temperature is equal, so solidification should start from the narrow part where temperature decreases first. In the wide part, solidification should proceed from the outside to the center of the melting zone. Therefore, the final solidification zone of IML is near the center of the wide part of LMZ. Solidification shrinkage occurred in the final solidification zone, and voids formed in this area. As shown in Fig. 9, cracks were observed in IML and was perpendicular to the joint interface in IML. This is thought to be caused by thermal contraction during cooling. Because IML has a smaller coefficient of thermal expansion and is more brittle than the surrounding Al and Fe, it is considered that cracks originated from the voids and formed mainly in the wide part of IML.
4.4 Consistency between numerical analysis results and experimental resultsNumerical analysis combining the SPH model and the LF model can reproduce not only the formation process of the wavy interface, but also the formation position, shape, composition of LMZ, and the subsequent formation process of IML. Here, the consistency between the numerical analysis results and the actual magnetic pulse welded joint interface is discussed.
First, about the wavy interface morphology. The wave height and wavelength of the wavy interface reproduced by the SPH model were 17 µm and 60 µm, respectively. The wave height and wavelength of the wavy interface observed in the Al/Fe joint obtained in the magnetic pulse welding experiment, which was conducted at the same impact velocity and collision angle as the numerical analysis conditions, were 21 µm and 78 µm, respectively. Although the numerical analysis results are slightly smaller than the experimental results, the trigger-like wavy interface with IML formed at the crest zone of wave is in good agreement with experiment results. In addition, the location of the final solidification zone predicted from the formation process of IML reproduced by the LF model corresponds well with the position of the void observed in the IML at the actual joint interface.
In terms of the composition, using the obtained number densities of Al and Fe n, the densities of Al and Fe ρ, and the volume per SPH particle V (the volume per SPH particle of equal smoothing length is same regardless of the material), the composition in LMZ (C′) was calculated using eq. (11).
| \begin{align} C'{}_{\text{Al or Fe}} &= n_{\text{Al or Fe}} \times \rho_{\text{Al or Fe}} \times V \\ &\quad \times 100/(n_{\text{Al}} \times \rho_{\text{Al}} \times V + n_{\text{Fe}} \times \rho_{\text{Fe}} \times V) \end{align} | (11) |
As the result, 79%Al–21%Fe (atomic ratio) was obtained as the numerical analysis results of the average composition of LMZ. In this study, LMZ was considered completely transformed to IML in LF model, the composition of IML is therefore equal to LMZ. The average composition of IML based on EDS results was 72%Al–28%Fe. Specifically, the average composition of TSL was 36 54%Al–46%Fe, while that of FSL was 85%Al–15%Fe. Due to the technique limitation, the simulation model cannot separately reproduce the TSL and FSL, so only the average composition of IML can be compared. The numerical analysis results of the average composition of IML showed 7% more Al than experimental results. In the actual situation, LMZ consists of both locally melted Al and Fe. At locations where the temperature is above the melting point of Al but below the melting point of Fe, LMZ should not actually be generated because only Al is considered as melted. However, the present simulation model cannot distinguish whether only Al is melted. Therefore, all Al particles above the melting point were considered as LMZ and then became IML in simulation. This is why the IML reproduced by simulation contains more Al than the IML obtained by experiment. Considering the above-mentioned difference, the average composition of IML estimated by the SPH model was quantitatively in good agreement with the average composition obtained by EDS analysis. Thus, the consistency between the numerical analysis results and the experimental results was quite well.
4.5 Formation mechanism of the intermediate layer at the Al/Fe joint interface in MPWWith the formation of the wavy interface, a rapid temperature rise above the melting points of Al and Fe occurred in the immediate vicinity of the interface and the crest zone of wave interface. This caused the local melting and resulted in the formation of IML.
The formation mechanism of IML containing TSL and FSL is discussed in detail here. As described in 4.2, four reference points 1∼4 were selected to investigate the change of temperature after the collision process. 1 and 2 correspond to Fe and Al in TSL while 3 and 4 correspond to Fe and Al in FSL. The temperature of 1 and 2 were appreciably higher than the melting point of Fe and Al, respectively. The simulation result showed that local melting occurred in TSL. In the previous report,11) Aizawa et al. observed the microstructure of TSL formed at explosive welded Al/Fe joint interface. They found fine dendrite branches, voids, and cracks in TSL. Their results indicated that the formation of TSL resulted from local melting and successive solidification. The present observation showed no dendrite structure at the MPWed joint interface. This is probably because the sample size and impact energy of MPW are several tens of times smaller than EXW. For example, in the case of Al/Fe EXW,11) the thickness of flyer plate and parent plate were both 2 mm and the estimated β and v were 15° and 750 m/s. In the case of the present Al/Fe MPW, the plate thicknesses were 0.3 mm and 0.4 mm, and the estimated β and v were 15° and 619 m/s. Considering that impact energy was absorbed at the interface, when converted it to kinetic energy per unit area, the size of IML can vary depending on the thickness of the flyer plate at the same β and v. Since β was the same between EXW and MPW, comparing the difference of v and thickness in EXW and MPW, the kinetic energy per unit area of MPW was estimated to be about 10 times smaller than that of EXW. However, since MPW shares a similar welding mechanism with EXW, and the characteristic of IML observed in the present study was also similar with EXW, there is still a high possibility that TSL was formed by local melting and rapid solidification at the joint interface. About FSL, the simulation results showed that the temperature of 3 did not exceed the melting point of Fe, but the temperature of 4 was slightly higher than that of Al. Therefore, local melting only occurred at Al and no locally melted Fe was contained in FSL. A lot of fragments embedded in FSL with light contrast were observed, as shown in Fig. 9(c). The composition analysis results indicated that the area of fragments with light contrast rich in Fe composition. The average composition of FSL was about 85%Al–15%Fe (atomic ratio). These Fe-rich fragments may be firstly ejected out of the Fe surface by the intense impact force during MPW collision process, and then wrapped into the local melted Al by an intense shear force.12,13) Lee et al.14) observed the magnetic pulse welded Al/Fe joint interface, and reported that an IML containing Fe-rich fragments, intermittently formed at the interface. This IML is similar to the FSL in the present study. They reported that IML consisted of fine grains of Al with diameters of around 100 nm, much finer intermetallic particles and fine fragments of Fe. Combined with the simulation results in this work, the formation of FSL took place as the follwing three steps. Firstly, fine Fe fragments were ejected out of Fe surface during collision process. Secondly, the ejected fine Fe fragments were drawn into the local melted Al at the foot of wavy and underwent a mechanical alloying process caused by intense plastic deformation during wavy interface formation process. Finally, the mixture of fine Fe fragments and local melted Al rapidly solidified to become FSL.
In this study, Al/Fe magnetic pulse welded material was prepared using Al for the flyer plate and Fe for the parent plate at the conditions of Gap = 2 mm and DE = 6 kJ. At the joint interface, the formation of two types of IML: TSL and FSL, was observed along with a trigger-like wavy interface. The average composition of TSL and FSL measured by EDS analysis is 54%Al–46%Fe and 85%Al–15%Fe (atomic ratio), respectively. Cracks perpendicular to the joint interface were observed in TSL, and voids were observed in FSL. The formation process of IML and cooling process at the magnetic pulse welded Al/Fe joint interface after the collision process were reproduced by numerical analysis method. The numerical results were compared with the experimental results to investigate the formation mechanism of IML.
The initial conditions of the LF model (thermal diffusivity and temperature distribution) were obtained from the material and temperature distributions at the Al/Fe wavy interface and the composition of LMZ at the interface reproduced by the SPH method. By setting the solidification temperature of LMZ, the cooling process at the joint interface and the solidification process of LMZ after the formation of the wavy interface were reproduced by the LF model.
Right after the formation of the wavy interface, there was a large temperature difference between the joint interface and the base metal. This caused a rapid heat removal from the interface to the base metal. The cooling rate near the interface is on the order of 107∼109 K/s. The base metal remained almost unchanged at room temperature. The solidification of LMZ proceeded from the outside to the center of LMZ and lasted for several microseconds. The final solidification zone of IML was the center of LMZ. Therefore, it is considered that the voids observed in the center of IML were formed due to the solidification shrinkage that occurred in the final solidification zone.
The estimated chemical composition from the particle number ratio by SPH model and the reproduced morphology of wavy interface were quantitatively in good agreement with the experimental results. By analyzing the change of temperature at the joint interface during and after the MPW collision process, a plausible explanation for the formation of IML at the Al/Fe interface was provided.
A part of this work was financially supported by the Japan Society for the Promotion of Science (JSPS) 19K05028 (Shinji Kumai).
