Laser Welding Modes Optimization of the Selective Laser Melted Ti-6Al-4V Thin-Thickness Parts with Complex Shape

Abstract

The paper studies laser welding of thin-thickness Ti-6Al-4V parts, manufactured by selective laser melting (SLM). A full factorial experiment was carried out in order to construct a regression model of the technological parameters (laser power, welding speed, and defocusing amount) which influence the weld shape. Metallographic analysis was carried out and it was found that thermal cycles of product printing and laser welding are equivalent. The microhardness analysis also showed no differences between the weld metal and the base metal. The contour plots of the parameters influence on the response function was constructed, and the area of welding modes was determined.

1. Introduction

Titanium alloys are widely used in various industries due to their high specific strength, corrosion resistance, and ability to operate at relatively high temperatures [1,2]. Selective laser melting technology (SLM) makes it possible to produce functional titanium alloys parts with complex configuration for aerospace industry, shipbuilding, engine construction, automotive industry, etc. [3,4,5]. There are many articles devoted to laser treatment of titanium alloys [6,7,8,9]. The size of printed parts is limited by the dimensions of the printing hopper. Parts with dimensions larger than the hopper can be made by joining several separate parts [10]. One of the possible ways of joining is laser welding. Welding, an optimal joining technology, has been used for parts fabricated by SLM. Xu and Zhang et al. revealed that the Inconel 625 alloys fabricated by SLM were successfully welded by a laser, while the tensile tests demonstrated greatly enhanced high temperature strength of the laser-welded SLM specimens [11]. J. Yang and Y. Wang et al. describe the study of the welds of SLMed 304 stainless steel. The results show that the microstructure of laser-welded joints consists of cellular dendrites in austenite matrix within columnar grains, exhibiting a coarser dendrite structure and superior corrosion properties of the samples [12]. The rapid joining of SLM parts with laser welding provides new opportunities for production development and extends the scope of additive manufacturing (AM) technology.

Welding of thin-thickness parts presents technological difficulties associated with the assembly of the joint and optimization of modes. The following technological parameters influence the stability of weld formation: laser power, welding speed, depth of focus point, and flow of shielding gas. For example, Yu, Hanchen and Li, Fangzhi et al. have found that printed Ti-6Al-4V parts can be joined by laser welding without special arrangements. The stress-relieved SLMed Ti-6Al-4V has a good laser weldability; however, the formation of gas type pores affects the fatigued lives of the welded product [10], and the shielding gas must be used to prevent absorption of the hydrogen in the molten pool. To date, laser welding of SLMed Ti-6Al-4V parts has not been widely investigated. The use of SLM technology for manufacturing parts implies their complex configuration [13,14,15]. One such configuration is the ribbon pattern with longitudinal ribs, used in gas turbine engines. The purpose of this work is to optimize the laser welding modes of SLMed Ti-6Al-4V parts with a complex shape.

Design of experiments (DOE) is one of the best known optimization techniques, which helps in carrying out the analysis of experiments with the least experimental effort. DOE is a set of mathematical and statistical techniques that are useful for modeling and predicting the response of interest, which are influenced by a number of input parameters. Such a method can also be used to optimize laser welding modes [16]. In this paper, we consider the influence of laser welding parameters (laser power, welding speed, and focus depth) on the weld shape. In the course of this work, a regression model will be developed, which can be used to determine the area of optimal laser welding modes.

2. Materials and Methods

Titanium alloy (Ti-6Al-4V) samples were used for the experiments, and the chemical composition is presented in

Table 1. The chemical composition of the Ti-6Al-4V [17].

Ti	Al	V	Fe	O	C	N	H
86–90	5.5–6.7	3.5–4.5	up to 0.3	up to 0.2	up to 0.08	up to 0.05	up to 0.015

.

The samples are 5 × 150 mm ribbons of 125 µm thickness with longitudinal ribs of 150 µm height and 165 µm spacing. A cross-sectional SEM image of part of the sample is shown in Figure 1. The samples were printed vertically and then cut off from the substrate by electrical erosion.

These samples have real technological applications. They are used in the gas turbine engines. Samples should be enlarged on the long side for the purpose of their further winding into a coil. The grooves formed by the ribs are intended for laying the threads of special fiber, so it is necessary to weld them without disturbing the slot geometry.

The experiments were carried out using a unit based on the YLR300 ytterbium fiber laser (Figure 2). Laser beam sweeping over the surface of the workpiece is carried out by a galvanometer scanner with a travel speed up to 10 m/s and a three-coordinate CNC manipulator. The YLR300 laser has a wavelength of 1.06 µm and a power of 300 W. The welding surface is opposite to flat surface of the sample. Welding is performed without filler material.

The main purpose of the experiment is to develop a mathematical model that adequately describes the process and allows to control the process. When using statistical methods of experiment planning, the mathematical description is presented in the form of a polynomial:

y = f(x₁, x₂, x₃),

(1)

where y is the response function, and x₁, x₂, x₃...—are the factors (arguments) of the process under study. The plan of the experiment in this case determines the location of the experimental points in the k-dimensional factor space. The plan is set in the form of a planning matrix (each row of which defines conditions of the experiment while each column defines values of controlled and controllable parameters in the process under study, i.e., values of factors corresponding to the condition of the experiment). In this paper, a full factor experiment 2³ was used.

When laser welding thin thicknesses, it is advisable to minimize the vapor recoil pulse to stabilize joint formation. The vapor recoil pulse results from material evaporation and leads to surface deflection. The region of power densities, with minimum vaporization, can be estimated from the following relation:

$${\mathrm{E}}_1\le {\mathrm{E}}_{\mathrm{w}}<{\mathrm{E}}_2,$$

(2)

where E_w is the power density in the welding zone, and E1 and E2 are the power densities for reaching temperatures T_m and T_b, respectively obtained from solving the heat conduction equation for the one-dimensional semi-infinite body heating problem [18] (5.1.36).

$${\mathrm{E}}_1=\frac{\sqrt{\mathsf{\pi }}{\mathrm{T}}_{\mathrm{m}}\mathsf{\lambda }}{2\sqrt{\mathrm{at}}},{\mathrm{E}}_2=\frac{\sqrt{\mathsf{\pi }}{\mathrm{T}}_{\mathrm{b}}\mathsf{\lambda }}{2\sqrt{\mathrm{at}}}$$

(3)

where a—the coefficient of thermal diffusivity; τ ~ d²/4a—the energy storage time (d—laser spot size diameter); λ—the coefficient of thermal conductivity; T_m—the melting temperature of the material; and T_b—the boiling point of the material.

Taking into account the value of the radiation reflection coefficient R, the radiation power density, E_w, falling on the surface of the parts to be welded, must be in the range:

$$2.9\times {10}^4\le {\mathrm{E}}_{\mathrm{w}}\ast \left(1-\mathrm{R}\right)<5.2\times {10}^4\left(\mathrm{W}/{\mathrm{cm}}^2\right)$$

(4)

The factors and values of the levels are presented in

Table 2. Factors and experimental design levels.

Notation	Factor	Level
		−1	0	1
x1	Laser power (W)	200	250	300
x2	Welding speed (mm/s)	50	75	100
x3	Defocus amount (mm)	0	+5	+10

.

Laser welding experiments were performed in bead on plate (BOP) mode. The gaseous protection was carried out with argon shielding gas with rate 15 L/min and the forming gas with rate 15 L/min. The scheme of the experiment is shown in Figure 3. The focal position was above the surface of the welded part.

To study the microstructure of the welded seam, the samples surfaces were greeted using SiC grit papers up to 2500-grit, further polished with an aluminum oxide suspension of 1 μm. The sample surfaces were etched using the Kroll’s reagent (1 mL HF +  2 mL HNO₃ + 97 mL H₂O) within 60 s. Metallographic studies were carried out on an optical micro-scope DMi 8 (Leica, Germany) with Axalit software (Axalit, Russia). Scanning electron microscopy (SEM) Tescan Mira3 (TESCAN, Czech Republic) was used to obtain images of the seam weld surface. The SEM images were obtained using the secondary electron (SE) detector.

The microhardness tests were carried out on the sample with an etched surface with a load of 50 g for 15 s at tester FM-310 (Future Tech, Japan), measured by Vickers’s method. The hardness measurement scheme is shown in Figure 4.

The metallographic studies and microhardness tests were performed on the cross-section.

3. Results and Discussion

3.1. Metallographic Studies

The planning matrix and the results of the experiment (fusion depth and weld width measurements) are shown in

Table 3. DOE matrix.

N	Factors			Weld Depth, mm	Weld Width, mm	Depth/Width
	x1	x2	x3
8	−1	−1	−1	0.20	0.31	0.64
6	1	−1	−1	0.23	0.34	0.68
5	−1	1	−1	0.22	0.27	0.81
7	1	1	−1	0.24	0.27	0.89
1	−1	−1	1	0.17	0.56	0.30
4	1	−1	1	0.27	0.77	0.35
2	−1	1	1	0.11	0.46	0.24
3	1	1	1	0.16	0.66	0.24

. The weld width was measured on the surface of the plate, and the weld depth was measured by cross sections.

All obtained samples are free of external defects. For example, Figure 5 shows SEM images of sample number 1. The SEM micrograph in Figure 5b shows a rough surface, which is detrimental for mechanical properties. This characteristic is common in powder deposition methods, such as SLM and EBM [19].

Figure 6 shows images of cross sections.

Based on the analysis of cross sections microstructure (Figure 7), and according to previous results [20,21], the phase structure of the weld metal, heat-affected zone (HAZ) and the base metal are identical and represent a fine martensite α’-phase. Based on this, it can be assumed that the thermal cycles in the laser welding and product printing are equivalent. The results of microhardness measurements show no appreciable changes between SLMed part and laser weld of microhardness measurement, which indicates that there is no difference in the phase composition of the weld and base metal, as well as between the samples. The microhardness values in the HAZ are reduced in comparison with the SLMed part and the laser weld, which is associated with the measurement of the hardness in the superheated area. Based on research results [22], additional heating of the SLMed part leads to partial decomposition of the martensite α’-phase and a decrease in microhardness. Results of measurements of microhardness are presented in Figure 8.

3.2. Development of Mathematical Model

To develop a regression model of the process, the first-order polynomial (5) was chosen. Here, a dimensionless number in an aspect ratio (obtained by dividing weld depth by weld width) was defined as the response function. This ratio characterizes the geometry of the weld.

$$\mathrm{y}={\mathrm{b}}_0+{\displaystyle {{\displaystyle \sum }}_{\mathrm{i}=1}^3}{\mathrm{b}}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{i}}+{{\displaystyle \sum }}_{\mathrm{i}<\mathrm{j}}{\displaystyle {{\displaystyle \sum }}_{\mathrm{i}=1}^3}{\mathrm{b}}_{\mathrm{ij}}{\mathrm{x}}_{\mathrm{i}}{\mathrm{x}}_{\mathrm{j}},$$

(5)

where b₀ is the response of f(0,0,0), namely the response of the central point; and b_i, b_ii, and b_ij are regression coefficients that depend on respective linear, squared, and interaction terms of factors, respectively. After calculating the coefficients, the final mathematical model is presented below:

$$\mathrm{y}=0.52+0.038{\mathrm{x}}_1+0.052{\mathrm{x}}_2-0.47{\mathrm{x}}_3-0.01{\mathrm{x}}_1{\mathrm{x}}_2-0.01{\mathrm{x}}_1{\mathrm{x}}_3-0.13{\mathrm{x}}_2{\mathrm{x}}_3,$$

(6)

where ${\mathrm{x}}_1$—the laser power, ${\mathrm{x}}_2$— the welding speed, and ${\mathrm{x}}_3$— the defocus amount.

The adequacy of the model was evaluated by the R² criterion, obtaining a value of 0.97. The signs of the factor coefficients correspond to the logical sense. Defocus has the greatest influence on the response function. The second most important factor is the interaction of defocus and welding speed. Figure 9a shows the contour plot of the defocus and welding speed influence on the response function (at laser power 250 W). Figure 9b shows the contour plot of the defocus and laser power influence on the response function (at welding speed 75 mm/s). The maximum value of the response function is determined at the following values of the main factors: power of 300 W, welding speed of 100 mm/s, and defocus of 0 mm.

Based on these graphs, it is possible to determine the mode for laser welding of thin thicknesses with a given aspect ratio, obtained by dividing weld depth by weld width.

4. Conclusions

In this work, laser welding of thin-thickness and complex configuration parts manufactured by the SLM method was carried out. Metallographic studies were carried out, which showed that the weld metal structure coincides with the structure of the base metal and is an α’-phase martensite solution. Thus, we can conclude that thermal cycles occurring during the printing of products by the SLM and during laser welding are identical. In particular, in this work, the influence of technological parameters of laser welding (laser power, welding speed, the value of defocusing) on the weld shape was studied. A regression model between the process parameters and weld shape was developed. Defocus has the greatest influence on the response function. Finally, the range of modes for laser welding thin-thickness SLMed parts was determined.