The Effect of Laser Remelting on the Microstructure and Mechanical Properties of the Bonding Interface in a Hybrid Metal Additive Manufacturing Process

In order to realize both high-efficient forming with the wire arc additive manufacturing (WAAM) and precise forming with the laser metal deposition (LMD) for some complex-structure and high-precision parts, a hybrid metal additive manufacturing method is proposed. The part is decomposed into sub volumes, then the sub volumes with relatively simple-structure features are formed through WAAM as a substrate, and the other sub volumes with more complex-structure or small-sized features are formed through LMD on the former substrate. However, the mechanical properties of the bonding interface would be reduced, if the later sub volumes are directly deposited by LMD on the rough WAAM substrate surface. In order to avoid unnecessary machining process between WAAM and LMD for high efficiency, and ensure the mechanical properties of WAAM-LMD bonding interface the laser remelting method is applied for improving the profile of WAAM substrate surface. The simulation model of heat transfer and fluid flow in the laser remelting process is established, the influence of the laser power and the scanning speed on the surface-profile improvement is researched by simulation and verified by experiments, Based on that the remelting process parameters are optimized. Furthermore, based on the WAAM formed substrate, the LMD formed volumes are deposited directly, after surface milling and after laser remelting, respectively. Then the microstructure and the mechanical properties of the bonding interface are compared among the three process methods, the feasibility of the laser remelting method for improving the bonding interface performance is verified.


Introduction
and the performance evolution. Meng Yunlong et al. [15][16][17] proposed that the crystal morphology mainly depends on the parameter T/G (T is the temperature gradient, G It is the ratio of the solidification rate) in the laser remelting process, and the increase of cooling rate makes the crystal grains smaller and the microstructure denser. YAO O et al. [18][19][20][21][22][23] investigated the effect of laser remelting on the surface performance strengthening besides the surface finish improving, the corrosion resistance and the wear resistance on the remelten surface are also improved.
Aiming at the 316L stainless steel, the simulation model of heat transfer and fluid flow in the laser remelting process is established in this paper, and the remelting process parameters are optimized. Based on the WAAM formed substrate, the LMD formed volumes are deposited directly, after surface milling and after laser remelting, respectively, the microstructure and the mechanical properties of the bonding interface are compared among the three process methods, the feasibility of the laser remelting method for improving the bonding interface performance is verified.

Assumption
The final surface morphology after laser remelting largely depends on the fluid flow on the free surface during the remelting process. Therefore, based on the basic theories of heat transfer and fluid mechanics, the laser remelting process model is established by multi-physics coupling finite element method to simulate the flow of the molten pool and the formation of free surfaces. The effects of surface tension and the Marangoni convection during the remelting process lead to the improvement of the WAAM substrate surface profile. In order to make the simulation results easy to converge and simulate the laser remelting process more accurately, this model makes the following assumptions on the premise of ensuring that the main characteristics of the actual remelting process are reflected: (1) The remelted substrate material is considered to be an isotropic and uniform medium; (2) The material's specific heat, thermal conductivity, density, heat transfer coefficient and other thermophysical parameters change with temperature; (3) The flow in the molten pool is considered to be viscous incompressible flow; (4) The laser incident energy distribution is assumed to be an ideal Gaussian distribution; (5) Consider the heat radiation and heat convection of the substrate;

Main control equations
The transient temperature field is controlled by Fourier's law, and the energy conservation equation is as follows [24]: * Where * is the heat accumulation term, ⋅ (− ) is the heat conduction term, ⋅ is the heat generated by the flow, is the laser heat source term, is the density, is Time, is the thermal conductivity, is the fluid velocity vector calculated according to the Navier-Stokes equation, as shown below [25]: Where is the pressure in the fluid, is the buoyancy, is the gravity, and is the dynamic viscosity. For incompressible flow, the fluid continuity equation can be simplified as [25]: When the solid phase absorbs heat and melts into a liquid, the absorbed heat becomes latent heat of phase change, and the specific heat capacity of the material changes with the change of the phase.
The equation to correct the change of thermal melting * with time is [26]: In the formula, is the temperature, is the heat capacity, is the latent heat of melting, is the liquid fraction, which increases in a linear manner, and its expression is shown as [27]: In the formula, is the solidus temperature of the material, and is the liquidus temperature of the material.

Remelted substrate model
Aiming at the surface and interface morphological characteristics of CMT forming, a sine curve is fitted as a CMT forming substrate model, with a peak value of 0.2mm and a period of 2mm. As shown in Figure 1, the laser beam is scanned from left to right along the X axis. The boundaries 4, 5, and 6 are non-slip walls during the remelting process.

Heat source model
The laser used is a fiber laser, the spot diameter is set to 2mm, the laser absorption rate is 0.36, and the laser beam energy is a standard Gaussian distribution. Figure 2 shows the laser energy distribution when the laser power is 900W. The expression of laser heat flux I is shown in the following formula [28]: In the formula, is laser power, is laser absorptance, is beam radius, is distance from the current calculation unit to the heating center, 0 is spot center X-axis coordinate, 0 is spot center Y-axis coordinate.

Heat transfer boundary conditions
The heat transfer state of all boundaries of the geometric structure is marked as shown in Figure   3. Due to the boundary conditions of heat conduction, heat convection and heat radiation on the upper boundaries 1, 2, and 3, the boundary condition equations are as follows [29]: Where is the absorption rate of the material, ℎ is the natural convection coefficient, is the emissivity, is the Boltzmann constant, and is the ambient temperature. The expression of the laser heat flux of Gaussian distribution is [30]: Where is the laser power, is the beam radius, and is the distance from the current calculation unit to the heating center.
Boundaries 4, 5, and 6 are only affected by thermal convection and thermal radiation. Therefore, the boundary condition formula is as follows:

Laminar flow boundary conditions
The boundaries 1, 2, and 3 are set as open boundaries, and the surface can be freely deformed.
On the surface, the surface tension acts along the normal direction, while the Marangoni effect acts along the surface tangential direction [31]. Normal: Tangential: Where is the curvature of the surface profile, is the temperature gradient along the surface tangential direction, is the surface tension, and and are the unit normal vector and tangent vector of the local surface respectively. Figure 3 Schematic diagram of the physical field application of each boundary

Moving grid
Boundaries 4 and 6 are fixed in the X direction and boundary 5 in the Y direction, so the displacement is zero. The ALE method is used to simulate the deformation of the free surface. In this method, the displacement of boundary nodes is determined by fluid dynamics. The domain nodes are described by Euler, and the boundary nodes are described by Lagrangian. In this paper, Laplace smoothing method is used for calculation [32], the coupling equation of ALE and fluid flow is: is the grid velocity, is the material velocity calculated according to the liquid fraction expression.

Mesh division and calculator configuration
The calculation area is meshed with free triangular elements. In order to reduce the amount of calculation without affecting the precise calculation of the top surface deformation, the curve boundary (boundary 2) is meshed with a custom cell, and the minimum cell size is set to 0.001mm.
In order to simulate the deformation of the free surface, automatic remeshing technology is used.
The complete grid consists of 48701 area elements and 2427 boundary elements. The fully meshed geometry is shown in Figure 4. The direct solver used for the simulation is PARDISO, and the time required for the laser scanning speed to pass 1 mm is set as the study time. Figure 4 Geometric meshing

Material properties
In the simulation model, the substrate is 316L stainless steel as an example. Some of the thermophysical parameters and related parameter settings of this material are shown in Table 1. At the same time, the thermal conductivity k, specific heat capacity Cp, and density rho of 316L stainless steel are greatly affected by temperature. The changes of each parameter with time are shown in Table 2. Thermal expansion coefficient Surface Tension N·m -1 1.8

Analysis of simulation results
In this study, a sinusoidal area with a length of 2mm and a peak value of 0.2mm is selected for simulation analysis. Generally speaking, when the laser interacts with the material, the energy is absorbed by the surface, thereby increasing the surface temperature. Taking laser power of 900W and scanning speed of 10mm/s as an example, after remelting with 10ms, the highest temperature in the area near the highest point of the model reaches 1733K( ). Over time, the area where the temperature rises above the liquidus temperature gradually expands.
There are two main driving forces for fluid flow: one is the surface tension related to the curvature of the surface profile, and the other is the Marangoni effect caused by the temperature gradient along the surface of the molten pool. If the tangential force caused by the temperature gradient cannot overcome the viscous force, the surface tension dominates the molten pool; otherwise, the Marangoni effect becomes the dominant. The direction of Marangoni convection is controlled by the surface tension temperature gradient. For 316L stainless steel, the liquid solute will flow from the center to the surrounding [33][34]. Laser energy density (ED) is related to laser power and scanning speed, and the specific calculation formula is as follows [37]: Where is the reference coefficient, is the laser power, is the scanning speed, and is the spot diameter.
Converting the simulated data into a line graph is shown in Figure 6 and Figure 7. The following conclusions can be drawn: (1)When the laser power is 300W and the laser scanning speed is 10mm/s, 15mm/s, 20mm/s, the reduction rate of surface profile before and after remelting is 0, which indicates that the laser energy density under this parameter group state is not enough to reduce The substrate melts, so the surface morphology of the substrate remains unchanged after remelting; (2)When the laser scanning speed is 10mm/s, 15mm/s, 20mm/s, as the laser power increases, the laser energy density gradually increases. As shown in Figure 6 and Figure 7, the area of the laser can melt the substrate also When the laser power is 300W, 600W, 900W, the laser energy density gradually decreases as the laser scanning speed increases, so the surface profile decreases. The rate of negative correlation is gradually decreasing; (3)When the laser power is 1200W and the scanning speed is 5mm/s, the reduction rate of the surface profile by laser remelting is lower than the 900W-5mm/s and 1200W-10mm/s parameter groups. This is because the laser energy density is too large, such as As shown in Figure 8, after the laser completely melts the convex peaks, due to the continuous action of the Marangoni effect, the ends of the molten pool gradually rise, which reduces the reduction rate of surface profile.

Experimental materials
The chemical composition of the 316L stainless steel welding wire material and powder material used in the experiment is shown in Table 3. The substrate used for welding is 304 stainless steel plate, the size of the substrate is 200×200×30mm, and alcohol scrubbing is used before the experiment.

Experimental equipment
The CMT forming volume of 316L stainless steel parts is prepared by welding robot forming system, and the welding machine uses the Fronius CMT Advanced 4000 welding power source, as shown in Figure 9.

Figure 9 Cold metal transfer (CMT) equipment
The wire grade of the wire feeder is ER316L, the diameter of the wire is 1.2mm, and the shielding gas is 98%Ar + 2%O2 gas. Other main process parameters are obtained through preliminary process experiments, as shown in Table 4. The LMD forming volume of 316L stainless steel parts is prepared by a 3-axis CNC machine, using argon gas (99.999%) for partial atmosphere protection of the cladding, equipped with a fiber laser, with a maximum output power of 2kW, equipped with a coaxial powder feeding cladding head, as shown in Figure 10.

Figure 10 Laser metal deposition (LMD) equipment
The defocus of the cladding head used in laser additive manufacturing is 14.5mm, and the spot diameter is about 2.5mm. The distance between the bottom of the nozzle of the cladding head and the substrate is 10mm. Other main process parameters are obtained through preliminary process experiments, as shown in Table 5.

Experimental program
Refer to the basic process parameters in the remelting simulation process, design the corresponding laser power and scanning speed to perform laser remelting verification experiments on the CMT forming substrate surface. Remelt a 25×20mm square area with each set of parameter, and use a dial indicator to measure the remelting surface profile. The surface height difference in each square area before and after remelting is calculated, the surface profile reduction rate c respect to each parameter is calculated, and the calculation formula of the profile reduction rate c is defined as follows: Among them, ℎ 1 is the maximum surface height difference before remelting, and ∆ℎ 2 is the maximum surface height difference after remelting.
(1) Metallographic preparation The metallographic samples of the LMD forming volume, the CMT forming volume and the bonding area of the CMT-LMD hybrid forming are prepared, and then they are embedded in the inlay machine after ultrasonic cleaning. Use 240#～2000# metallographic sand for polishing and 3.5μm diamond spray polishing agent for polishing. The 316L stainless steel corrosive solution uses 10gFeCL3+30mlHCL+120mlH2O, and the metallographic structure is observed under an optical microscope.
(2) Shear test Three 40×20×20mm experimental sample substrates are formed by CMT, which surfaces are respectively threated by three different methods, includings non-treatment, laser remelting, and milling, then based on the surface of each sample substrate, blocks with size of 30×15×18mm are formed by LMD. Taking the joint interface as the center line, three sets of shear test rod are prepared, and the size of the test rod is shown in Figure 11(a) below, and the sampling position of the test rods is shown in Figure 11

Experimental results
In order to verify the accuracy of the above-mentioned laser remelting simulation results, a study on the influence of laser power and scanning speed on the surface profile of the wire was carried out. The height difference between the highest point and the lowest point on the surface of each area was used as the evaluation of the surface profile. Indicators, the height difference of each area before and after remelting was measured by a dial indicator, and the intuitive cladding effect is shown in Figure 12.
Little improvement on the surface 300W-20mm·s -1 600W-5mm·s -1 600W-10mm·s -1 600W-15mm·s -1 600W-20mm·s -1 900W-5mm·s -1 900W-10mm·s -1 900W-15mm·s -1 900W-20mm·s -1 Figure 12 Actual laser remelting effect diagram corresponding to different laser power and scanning speed Surface warping 1200W-5mm·s -1 1200W-10mm·s -1 1200W-15mm·s -1 1200W-20mm·s -1 Figure 12 (continued) Actual laser remelting effect diagram corresponding to different laser power and scanning speed According to the measured data, it is converted into a line chart as shown in Figure 13 and Figure   14 below. It can be seen that the actual improvement trend of the laser remelting of each parameter group on the surface profile of each area is close to the simulation result, and the error of the flatness reduction rate of the corresponding parameter group is within the range of about 20%. The main reasons are: (1)The actual laser heat source is similar to Gaussian distribution, not regular Gaussian distribution; (2)There is a big gap between actual heat convection and heat conduction and theoretical heat convection and heat conduction.
According to process experience, the reduction rate of surface profile needs to be at least 55% or more. It can be seen from Figure 13 and Figure 14 that there are parameters that meet the optimization requirements in the actual laser remelting process parameter group. Therefore, it is proved that the flatness of the CMT forming surface can be achieved by laser remelting. The method required for LMD forming is feasible.

Microstructure analysis
Three kinds of substrates were obtained by remelting, milling and unremelting the forming surface of CMT, and LMD forming was performed on the surface after treatment, and metallographic samples were prepared at the bonding interface. The study of the mechanical behavior of materials must consider the influence of the interface and grain boundaries [38][39]. The metallographic structure is shown in Figure 15. It is found that the formed structures of the two processes are composed of δ ferrite + γ austenite, but in the CMT formed structure area, the dendrite spacing and grain size are larger than the LMD forming structure area, mainly due to the difference of the raw material composition and thermal history in the forming process of the two processes [3].
Because the dendrite spacing and grains of the LMD region are smaller, there are more grains and grain boundaries in the unit area. When external load is applied, the load and deformation can be distributed to more grains, which can It can better suppress the "slip" of atoms inside the grain, so the mechanical properties of the LMD forming area are better than the CMT forming area. And through observation, it was found that after cladding the surface of the remelted substrate, no remelted layer was found at the bonding interface. This is because the subsequent LMD cladding process completely covered it, and the bonding after remelting The morphology of the interface at the interface is similar to the bonding interface formed after milling, while Due to the uneven surface of the substrate, the growth direction of the microstructure formed by LMD is inconsistent with the growth direction of the structure formed by CMT, and there is a certain angle deviation. (c) Untreated bonding interface on the surface

Shear performance analysis
The corresponding shear test results of the shear samples prepared for different substrate bonding interfaces are shown in Figure 16 below. It can be seen that the non-remelted shear strength is slightly lower than the shear performance at the bonding interface after laser remelting or milling.
This is because the shearing direction is along the X-axis, and when the shearing direction is When the direction is perpendicular, the number of grain boundaries on the shear interface is the largest, so the shear resistance is the best, while the non-remelted bonding interface has a disorderly growth direction and is at a certain angle with the direction of the shear interface, so the grain in the shear section The number of boundaries is small, and when loaded, the ability to inhibit the "slip" of the atoms inside the grain is weakened, and the shear resistance is also reduced. Figure 16 Different interface processing methods corresponding to the bonding interface shear performance

Conclusion
(1) When the scanning speed is constant, as the laser power increases, the improvement degree of laser remelting on the surface profile of the CMT forming substrate first increases and then decreases; and when the laser power is constant, As the scanning speed increases, the improvement of the surface profile of the CMT forming substrate by laser remelting gradually decreases, but when the laser power is 1200W and the scanning speed is 5mm/s, the improvement of the surface profile is less than 10mm/s. This is because the laser energy density is too large, leading to the continuous action of the Marangoni effect.
(2) The dendrite spacing and grain size of the CMT forming structure are larger than the LMD forming structure area, and when the surface of the CMT forming substrate is not treated, the dendrite growth direction of the structure at the bonding interface is disordered, but when the CMT forming substrate surface After milling or laser remelting, the growth direction of LMD dendrite at the bonding interface is close to the same as that of CMT.
(3) When the surface of the CMT forming substrate is not processed, the LMD forming is directly performed, and the shear strength at the bonding interface obtained is slightly lower than that of the surface of the CMT forming substrate after milling or laser remelting, and then LMD forming , The resulting bonding interface is mainly due to the disordered growth direction of the dendrite, which is at a certain angle with the shear interface, so that the number of grain boundaries in the shear interface is relatively reduced, and the ability to inhibit the "slip" of atoms inside the grain is relatively Weakened, so the shear resistance is also reduced.