1 Introduction

Metal additive manufacturing (MAM) has been developing rapidly recently, showing great potential applications in medical, automotive, aerospace and other industry fields [1]. Metallic powders are essential to MAM process development, and the demand for high-quality and inexpensive metallic powders is growing substantially [2]. Metallic powders are mainly produced by gas atomization, water atomization, rotating electrode centrifugal atomization, et al. [3–6]. These powder production methods have their advantages, however, they all share some common disadvantages, namely, complex equipment, large energy consumption, and high cost.

Wire electrical discharge machining (WEDM) is a nontraditional, thermoelectric process which erodes material from the work piece by a series of discrete electric discharge sparks (EDS) between a work and tool electrode immersed in a liquid dielectric medium [7]. The major advantage of EDS is that the selection of machining parameters in a machining process significantly affects erosion products of both workpiece surfaces [8] and debris, making it a simple, effective technique to generate multi-scale metallic particles [9–11]. Most previous researches in EDS powder preparation focused on the minimization of the occurrence of the larger particle mode to produce nanometer-scale powders [12–15]. Limited works have been published in studying the effect of WEDM process parameters on micron-scale metallic powders which are required by MAM process.

In the present study, WEDM parameters for producing micron-scale bronze powders have been optimized based on the Taguchi method. The current investigation aims at highlighting the application potential of WEDM in preparation micron-scale metallic powders to meet the requirement of MAM process. The purpose of this study is to evaluate the feasibility of the WEDM process to prepare metal powders suitable for the MAM process.

2 Experimental

WEDM process was carried out using a commercial WEDM machine (DK350, China). The cathode was a molybdenum wire of 180 µm in diameter, and the anode workpiece was a CuSn8 bronze plate of 5 mm thickness. The eroded bronze particles were collected in a series precipitation tanks, as shown in Fig. 1. The collected bronze powder was ultrasonic cleaned in water and ethanol successively, and then was dried with hot air at 50 °C.

Fig. 1
figure 1

Electric spark erosion powder making device (a) and processing track (b)

Full size image

Factors influencing WEDM powder particles can be divided into two categories, namely discharge medium and processing parameters [16]. Four kinds of dielectric medium were selected in the present study. The selection of processing parameters in our WEDM machine is primarily dependent on the machining parameter tables provided by the machine-tool manufacturers. The parameter table of our WEDM machine contains three parameter patterns, namely cutting-class, cutting-division and cutting section. Each parameter pattern has several options: four in cutting-class (c1–c4), eight in cutting-division (d1–d8) and nine in cutting-sections (s1–s9). Another WEDM process parameter is cutting speed. Details of five input controlling factors and their four levels were listed in Table 1. A L16(45) orthogonal array was employed according to the Taguchi method, as shown in Table 2.

Table 1 Factors and their levels
Full size table
Table 2 Orthogonal array and observations
Full size table

Particle sizes were tested using a laser particle analyzer (NKT2020-L, China). Particle shapes were determined using an optical microscope (XJP-6A, China) and a scanning electron microscopy (SS-60, China), respectively.

Particle size data were analyzed according to the Taguchi method to estimate relative significant effect of each control factor, and to optimize WEDM process parameters for purpose of improving powder properties.

3 Results and discussions

3.1 Particle size data

The particle size distribution curves measured by laser particle size analyzer are shown in Fig. 2. All curves are presented as approximate Gaussian distributions. However, these curves are significant different in peak width. Characteristics of a powder particle size are generally described with d50, d10 and d90. The d50 value reflects the mean particle size of the powder, while the value of d90-d10 reflects the particle size distribution of the powder. Metallic powders for additive manufacturing are usually required uniform particle sizes of 20–60 μm with a narrow size distribution. Therefore, the particle size d50 and size distribution index di (= (d90-d10)/d50) were used for data analysis.

Fig. 2
figure 2

Cumulative percentage vs particle size for each sample powder

Full size image

Figure 2 shows that the d50 for most sample powders are below 10 μm, except sample 12 and sample 14. According to the Taguchi based methodology, the d50 is regarded as a larger-the-better type performance, and di is regarded as a smaller-the-better type performance [17]. In other words, the goal of process parameters optimization, based on the data of this Taguchi orthogonal test, is to maximize d50 and to minimize the particle size distribution di. The d50 and di data are summarized in Table 2.

3.2 Factor effects and optimal parameters

Based on the data in Table 2, the average value of each level for each factor has been calculated, and the calculated results are listed in Table 3. The range for each factor is also listed in Table 3. The large the range value of a factor, the greater the influence of a factor is. The factor effects are displayed graphically in Fig. 3 which make it easy to visualize the relative influence of the various factors on d50 and di.

Table 3 Average mean value and range analysis
Full size table
Fig. 3
figure 3

The average mean d50 (a) and di (b) for each level of each parameter

Full size image

The factor that has the greatest influence on d50 is D (Cutting-sections); the other factors are B (Cutting-classes) E (Wire speed), C (Cutting-divisions) and A (Dielectric medium) in order of importance. For the larger-the-better d50, the optimum factor-level setting is D1–B4–E3–C2–A3.

The factor that has the greatest influence on di is D (Cutting-sections) also; the other factors are E (Wire speed), C (Cutting-divisions), B (Cutting-classes) and A (dielectric medium) in order of importance. For the smaller-the-better di, the optimum levels setting is D1–E1–C3–B4–A3.

The task of determining the best setting for each control factor is complicated when multiple characteristics are to be optimized [18]. However, it is fortunate in the present study. Either for d50 or di, factor D (Cutting-sections) is the biggest influence factor, and factor A is the smallest influence factor. And furthermore, D1 and A3 are the optimization parameters to both of them. This is definitely a good result, as it makes the WEDM process a great potential process for the preparation of metallic powders with proper suitable particle size and a narrow particle size distribution. In addition, dielectric medium (Factor A) has little effect on the particle size of the powder, which is beneficial to select suitable electrolytes for the preparation of different metallic powders, so as to reduce the pollution of the electrolyte chemical to the metallic powder and the pollution to the environment. The B factor is B4 in both d50 and di optimization parameters. The influence of C factor on di is slightly greater than its influence on d50, so C3 can be selected according to di. The influence of E factor on di is also slightly greater than its influence on d50. And the d50 value of E1 is close to that of E3, so E1 can also be selected according to di. Therefore, considering the optimization results of d50 and di, the optimized process parameter combination in this study is determined as: A3–B4–C3–D1–E1.

3.3 Verification experiment

Conducting a verification experiment is a crucial, final and indispensable part of the Taguchi method project. Its aim is to verify the optimum condition suggested by the matrix experiment estimating how close are the respective predictions with the real ones. Experimental bronze powder has been prepared using the optimum parametric setting of A3–B4–C3–D1–E1. The laser powder sizes measured by laser particle analyzer is shown in Fig. 4. Several feature sizes are: d10 = 11.05 μm, d50 = 31.13 μm, d90 = 67.48 μm, and di = 1.81. Compared with the powder prepared under the basic parameters of the Taguchi orthogonal test, the particle size distribution of the powder prepared under the optimized parameters is smaller.

Fig. 4
figure 4

Cumulative percentage vs particle size for the optimized powder

Full size image

The SEM observations of the powders prepared under the basic parameters of the Taguchi orthogonal experiment and the powder prepared under the optimized parameter are shown in Figs. 5 and 6, respectively. As can be seen from Fig. 5, the appearance of the various powders prepared by the basic parameters of Taguchi test is very different. The S1, S12 and S14 samples powders contain some larger spherical particles. And other thirteen groups of powder samples are mainly composed of fine irregular-shape particles.

Fig. 5
figure 5

SEM micrographs of the base Taguchi experiment powders

Full size image
Fig. 6
figure 6

SEM images of a appearance and b cross-section of the optimized powder

Full size image

Figure 6 is SEM micrographs of the optimized powder. Apparently, the powder is composed of spherical particles of uniform size. Most powder particles are between 10 and 50 microns in diameter. The powder particles are spherical or nearly spherical with smooth surface. There is almost no particle-sticking and satellite-ball phenomenon. Cross-sectional images of powder particles show that a certain number of powder particles have internal voids. The internal pores of individual powder particles are large, forming an eggshell structure. It is worth mentioning that small-sized particles are more prone to the hollow structure.

The phenomenon of voids in metallic powder prepared by WEDM has been well known for a long time, and has also been studied extensively [19, 20]. Mechanisms for voids forming inside of a powder particle are possibly bagging effect and/or density gradient effect. In bagging effect, a molten platelet closes on itself forming a bubble, then freezes rapidly into a hollow particle. In density gradient effect, a molten droplet solidifies initially on its surface, leaving a small central pore after complete solidification [21]. Large size particles can form spherical or near-spherical shape results from the spheroidization time is shorter than solidification time [22].

During electrical discharge machining, it has been reported that materials can be transferred among the electrodes [23]. The same phenomenon has been reported in WEDM [24]. Hence, it is reasonable that the electrode molybdenum wire will also be consumed during WEDM process, resulting in electrode debris mixed in powder products. EDS analysis was used to identify chemical compositions on surfaces of both powder particles and the workpiece. Mo element could not be found either on surfaces of powder particles and workpiece, as shown in Fig. 7.

Fig. 7
figure 7

SEM images of a powder particles and b machined workpiece surface as well as (c) and d their corresponding composition maps

Full size image

It was reported that the adherence of wire electrode elements on machined workpiece surface reduced by longer pulse-on time and smaller discharge current [25]. Considering the electrode molybdenum wire is the cathode and the workpiece is the anode, the heat of the spark is distributed more on the anode workpiece, and the melting point of molybdenum is much higher than that of the bronze workpiece. the loss of the electrode molybdenum wire is extremely low compared with the bronze workpiece. In fact, we observed that during WEDM process, the color of the electrode molybdenum wire gradually changed from dark black to bright yellow (see Fig. 8a, b), due to copper depositing on surface of the cathode molybdenum wire. The diameters of molybdenum wire before and after machining are 0.178 mm and 0.181 mm, respectively. It shows that the copper coating phenomenon occurs on the surface of the molybdenum wire. Therefore, during WEDM process, the electrode is almost not consumed, and the WEDM products contains almost no debris from the wire electrode.

Fig. 8
figure 8

Images of molybdenum wires a before and b after machining process

Full size image

4 Conclusions

  1. 1.

    Spherical bronze powders with uniform particle size can be prepared using a commercial WEDM machine and a suitable process parameters optimized by Taguchi method.

  2. 2.

    Within the range of equipment and parameters in this study, Cutting-sections are the most important factor affecting the particle size and particle size distribution of the bronze powders. And the electrolyte has the least effect on the particle size and particle size distribution of the bronze powders.

  3. 3.

    Spherical bronze powder with particle size of d10 11.05 μm, d50 31.13 μm and d90 67.48 μm are prepared using the optimized WEDM process parameter, which is A3-B4-C3-D1-E1 including dielectric medium water, cutting class 4, cutting division 3, cutting Sect. 1 and wire speed 200 μm/s.

  4. 4.

    It confirms that a convenient WEDM process is practical for fabricating micron-sized spherical metallic powders used for metal additive manufacturing.