1 Introduction

Micro electrical discharge machining (μEDM) is a noncontact machining process that relies on electrical discharges between a tool electrode and a workpiece to erode material from the substrate. It is widely used to manufacture highly accurate microscale components and structures, such as micromolds, microdie, microprobes, microtools, fuel nozzles, and photomasks. Additionally, μEDM can be effectively applied to machine-sloped surfaces, curved surfaces, thin sheet materials, and fragile and soft materials that are difficult to machine. Nonetheless, various challenges for μEDM, such as electrode wear, electrode preparation, and surface quality, remain [1]. One fundamental issue is that the micro electrodes used in μEDM gradually wear during the discharge process due to erosion. This wear or erosion always occurs and leads to loss of the original geometrical precision. Due to the miniaturization of the electrode, electrode wear will be more remarkable in μEDM, which adversely affects the geometrical accuracy of machined features. The electrode wear in some cases can even limit the application of μEDM [2].

Many investigations have sought to solve wear-related issues and improve machining accuracy. For example, the linear compensation method (LCM) has been applied to μEDM [3]. The tool (electrode) material is continuously fed along the route with a constant ratio to compensate for wear. However, this approach is only capable of generating microfeatures with straight sidewalls, and compensating for such nonuniform electrode wear and accurately generating microfeatures are very difficult. The uniform wear method (UWM) has been widely used for μEDM [4] and adopts layer-by-layer discharge milling and compensation of the worn electrode while maintaining the original electrode shape in every layer. The UWM is effective for precision 3D micromachining and has been realized through CAD/CAM, as discussed by Rajurkar et al. [5]. However, the compensation is executed at the beginning of every layer of discharge erosion, causing unstable discharging and low surface finish. In addition, the tool path requires specially design to restore original shape of the electrode after the erosion of each layer, and an empirical approach is needed to select tool paths and machining parameters. To overcome this issue, the combined linear uniform (CLU) method, which is a combination of the LCM and UWM processes, was proposed by Yu et al. [6]. Its tool path design is based on the UWM to ensure a constant electrode tip shape after one layer of discharge milling, while the total compensated length of tool wear (TW) in each layer is evenly divided into several pieces along the tool path. However, due to the scanning tool route overlap and the involvement of the sides of the tool in erosion, there is not always a linear relationship for the electrode wear ratio (EWR). Considering this nonuniform wear issue, a TW compensation technique to achieve a constant EWR was proposed by Li et al. [7] in which compensation points on the tool paths of every layer were set based on the scanned area (BSA). This approach was further improved through a real-time 3D servo scanning μEDM method [8]. Nevertheless, the machining performance and depth error need further improvement.

Based on the uniform TW assumption, the TW model was proposed by Yu et al. [9]. According to this model, Zhang et al. introduced the fix length compensation (FLC) method for improving the shape accuracy and dimensional precision, in which compensation was applied after a certain machining length [10]. Due to the discharge conditions and the potential for non-constant wear during the μEDM process, another type of compensation method, the multiple electrode strategy (MES), was proposed [11]. The MES can provide better precision and repeatability while preventing overcutting of the profiles. However, the fabrication of electrodes and multipass trajectories are a tedious task, considering the time wasted when an electrode follows an already eroded path.

Several researchers have applied the discharge counting method (DCM) for wear compensation, which relies on evaluating the volume of material wear per discharge and counting the number of pulse discharges [12, 13]. For example, Aligiri et al. proposed a real-time TW compensation method by estimating the material removal volume of single discharge based on the electrothermal model and pulse discrimination [14]. However, the measurement of small volumes was imprecise, and individual discharges do not illustrate the natural process. To overcome this issue, Puthumana et al. applied a statistical distribution of discharges rather than individual ones for TW compensation [15]. However, the errors associated with the estimation of TW per discharge will lead to longitudinal errors of the tool electrode. Modica et al. proposed a simple TW compensation method, fixed reference compensation (FRC) [16], in which a reference control point is fixed, and use an electrical control touch procedure to extract the coordinates. However, this approach can only estimate electrode linear wear, and its precision depends on the repeatability of machine tooling. Yan et al. proposed a straightforward measurement TW method based on an expensive machine vision system [17] in which the process must be temporarily interrupted to move the electrode to a specified position for detection. Although this wear compensation approach can directly measure and evaluate the bottom wear and side wear of the tool electrode, it is time-consuming and costly.

To address the erosion-related problems associated with μEDM, the present study investigates the no-wear feature of Galinstan μEDM. This work specifically focuses on experimental characterizations based on the preliminary study reported by Huang et al. [18]. Square cavity samples were well prepared using the liquid alloy microelectrode without loss of the electrode length, which was suggested to potentially overcome the issues related to electrode wear. The machining performance will be discussed in terms of the material removal rate (MRR) and liquid alloy consumption rate (LACR).

2 The principle of liquid alloy electrode μEDM

Figure 1a illustrates the machining principle of the proposed method. In this process, a liquid alloy is held and slowly flows through a capillary needle so that the liquid protrudes on the needle tip, forming a semi-ellipsoid (Fig. 1b). The semi-ellipsoid liquid alloy will deform into a cone when a voltage is applied between the workpiece and the liquid alloy, yielding an electric field (Fig. 1c). Moving the needle tip-spouted liquid cone toward the workpiece surface promptly generates discharges once reaching the breakdown condition and erodes the workpiece material (Fig. 1d). The liquid alloy will retract to a semi-ellipsoid (Fig. 1b) due to the electrostatic force disappearing after a pulse discharge, which enlarges the discharge gap and is conducive to the emission of erosion debris.

Fig. 1
figure 1

The machining principle of the liquid alloy electrode: a pulse generator and discharge general view; b formation of a semi-ellipsoid; c deformation into a cone under an electric field; d breakdown and discharge

Full size image

The consumption of the liquid alloy electrode in this method is similar to that of traditional μEDM; however, the liquid alloy can be continuously fed to the needle tip to avoid effects on the erosion process due to the application of pressure to compensate for consumption. The sparks are only generated between the workpiece and the liquid electrode by controlling the feed of the liquid alloy; consequently, the needle itself does not produce sparks that can thermally damage it and remains intact. Arbitrary erosion of the workpiece surface can be obtained through controlling the relative positions of the workpiece and the needle.

3 Liquid alloy feeding control

The liquid alloy electrode μEDM is different from conventional μEDM. Although no electrode wear occurs, liquid alloy needs to be continuously fed to compensate for consumption. If the feed volume is less than the consumption volume, then the end of the capillary cannot form a discharge electrode in time, which will lead to a low discharge efficiency. If the feed volume is greater than the consumption volume, then excessive liquid alloy at the end of the capillary will form large droplets that will drip down, thus easily causing short circuits, affecting the process stability, and causing waste. Therefore, the liquid feed volume should correspond to the volume of liquid consumed. The flow velocity of liquid alloy in the capillary determines the feed rate. Therefore, the relationship between the flow velocity and applied pressure should be clarified, which will provide a theoretical basis for stable continuous processing. The flow velocities of liquid alloys in capillaries are analyzed as follows.

Because the inside walls of the needle are not absolutely smooth, the liquid alloy flow in the needle will be subjected to resistance due to the liquid alloy viscosity. The velocity of the liquid is the fastest at the center of the tube and gradually decreases along the radius toward the tube wall. The fluid type is determined by the Reynolds number (Re) of the boundary layer. The Reynolds number of the boundary layer is given by

$$ \mathit{\operatorname{Re}}=\frac{\rho uL}{\mu } $$
(1)

where

ρ:

the density of the fluid (kg/m3).

u:

the velocity of the fluid (m/s).

L:

the characteristic length (m).

μ:

the dynamic viscosity of the fluid (Pa s).

In this work, Galinstan, a type of nontoxic eutectic liquid alloy, is used as the liquid electrode. The properties of Galinstan are shown in Table 1. It can easily flow through microscale needles [19]. Because the inner diameter of the needle is the smallest part of the liquid supply system, it has the greatest influence on the flow velocity. The other parts are larger, and their influence on the flow velocity can be neglected. The inner diameter of the needle is less than 0.2 mm, and the velocity of the liquid alloy is less than 1 mm/min. Therefore, the Reynolds number Re is less than 2000, and the fluid tends to be dominated by laminar flow, where viscous forces are dominant, and is characterized by smooth, constant fluid motion. Therefore, the flow type of the liquid alloy is laminar flow.

Table 1 Properties of Galinstan (at 20 °C)
Full size table

In the flow direction, the liquid alloy will be subjected to two types of forces: the driving force, P1, which is provided by the external pressure and is in the same direction as the flow direction, and the resistance, P2, which is caused by the tube wall and internal friction and prevents fluid flow. According to Newton’s second law, under the action of the two forces, the liquid alloy can maintain a constant-speed flow in the tube if it can reach the force balance.

To simplify the calculation, the following conditions are assumed: (1) no bubbles occur in the needle, i.e., the needle is full of liquid alloy, and (2) the flow velocity varies linearly from the center to the tube wall. A fluid column with a length of l and a radius of r to the central axis is taken as the research object, as shown in Fig. 2. The driving force acting on the fluid column is

$$ \left({p}_1-{p}_2\right)\pi\ {r}^2=\varDelta {p}_f\pi\ {r}^2 $$
(2)
Fig. 2
figure 2

Sketch of the selected fluid column in a tube

Full size image

The velocity of the fluid layer at radius r is ur, and the velocity of the adjacent fluid layer at r + dr is (ur + dur). The velocity gradient along the radius is dur/dr. According to Newton’s law of viscosity, the internal friction due to the relative motion between adjacent fluid layers is

$$ {F}_r={\tau}_wS=-\mu \left(2\pi rl\right)\frac{d{u}_r}{dr} $$
(3)

where

τw:

shear stress (N/m2).

S:

the contact area of the fluid layer (m2).

Due to the constant fluid motion of Galinstan, the force is balanced, and the pressure difference counteracts the frictional resistance.

$$ \varDelta {p}_f\pi {r}^2=-\mu \left(2\pi rl\right)\frac{d{u}_r}{dr} $$
(4)

Thus, \( d{u}_r=-\frac{\varDelta {p}_f}{2\mu\ l} rdr \), and integrating both sides of Eq. (4),

$$ {\int}_0^{u_r}d{u}_r=-\frac{\varDelta {p}_f}{2\mu\ l}{\int}_R^r rdr $$
(5)

Thus,

$$ {u}_r=\frac{\varDelta {p}_f}{4\mu\ l}\left({R}^2-{r}^2\right) $$
(6)

The average flow velocity \( \overline{u} \) is usually used as the standard for calculating resistance loss. Therefore, the relationship between \( \overline{u} \) and ∆pf should be calculated. The average flow velocity can be expressed as the flow volume per unit area per unit time:

$$ \overline{u}=\frac{V}{A}=\frac{V}{\pi\ {R}^2} $$
(7)

To obtain the flow volume V passing through a unit cross-sectional area, a ring with radius r and width dr is considered (shown in Fig. 3); thus, dA = 2πrdr. Because dr is very small, the liquid velocity in the entire ring can be assumed to be ur, and the flow volume formula is then

Fig. 3
figure 3

Sketch of the cross-section of the tube

Full size image
$$ dV={u}_r dA={u}_r\left(2\pi rdr\right) $$
(8)

Substituting ur, which is calculated in Eq. (6), into Eq. (8) leads to

$$ dV=\frac{\varDelta {p}_f}{2\mu\ l}\left({R}^2-{r}^2\right)\pi rdr $$
(9)

Integrating both sides of Eq. (9),

$$ {\int}_0^V dV={\int}_0^R\frac{\varDelta {p}_f}{2\mu\ l}\left({R}^2-{r}^2\right)\pi rdr $$
(10)

and

$$ V=\frac{\pi\ \varDelta {p}_f}{2\mu\ l}\left(\frac{R^4}{2}-\frac{R^4}{4}\right)=\frac{\pi\ \varDelta {p}_f}{8\mu\ l}{R}^4 $$
(11)

Thus, the average velocity can be expressed as

$$ \overline{u}=\frac{V}{\pi\ {R}^2}=\frac{\frac{\pi\ \varDelta {p}_f}{8\mu\ l}{R}^4}{\pi\ {R}^2}=\frac{\varDelta {p}_f}{8\mu\ l}{R}^2 $$
(12)

When applying a voltage, an electrostatic force will be generated. The electrostatic force together with the liquid surface tension and end shape will mildly affect the flow velocity; hence, adjustment parameter k can be introduced into Eq. (12), resulting in

$$ \overline{u}=k\cdot \frac{\varDelta {p}_f}{8\mu\ l}{R}^2 $$
(13)

Parameter k is determined by the voltage, the gap distance between the workpiece and liquid electrode, the liquid surface tension and end shape and so on, which can be experimentally obtained. Hence, different flow volumes can be obtained by setting the pressure to different values.

4 Experimental setup and procedure

Figure 4 shows the experimental setup used for process evaluation. In this study, a liquid alloy, Galinstan, is selected as the electrode. Its high boiling point and high electrical conductivity are suitable for use as a μEDM electrode. A stainless-steel needle (here, two types of needles are applied, needle A: inner diameter DI = 160 μm, outer diameter DO = 310 μm, length L = 13 mm; needle B: inner diameter DI = 100 μm, outer diameter DO = 240 μm, length L = 13 mm) is used as the needle in the setup. A syringe that stores Galinstan is joined to the needle. A pressurizing unit is used to apply pressure to the syringe to feed Galinstan to the needle. A highly doped p-type silicon wafer is used as a workpiece and fixed on the work tank. A high-precision three-axis stage is used to position the needle microelectrode and the workpiece. To prevent oxidation of the Galinstan, a 2% H2SO4 solution is admixed in the syringe and floats on top of the Galinstan. Furthermore, the H2SO4 solution keeps the liquid alloy clean to avoid clogging of the needle. As shown in Fig. 1a, a resistor-capacitor (RC) pulse generator is selected to perform the Galinstan μEDM processing [20].

Fig. 4
figure 4

Experimental setup

Full size image

The distance, D, between the workpiece surface and the needle tip is measured by using electrical contact perception and retracting the needle to determine the gap. μEDM erosion is implemented by programmed XY scanning motions with a constant gap D while using a detection circuit to monitor the gap state. The experimental conditions are outlined in Table 2.

Table 2 The experimental conditions
Full size table

5 Experimental results and discussions

5.1 Electrically passivated needles

Figure 5 shows scanning electron microscopy images of the needle B tips. Figure 5a shows the original state of the needle tip. Figure 5b shows the needle tip used in Galinstan μEDM with Vs = 100 V, R = 1 kΩ, C = 27 nF, P = 1.2 atm, T = 40 min, and D = 100 μm. The comparison shows that the needle tip caused some discharge pulses in the process, leading to slight erosion of the needle surface. The gap distance D was too large to directly induce breakdown between the workpiece and the needle tip under this voltage condition. Thus, the sparks on the needle surface were presumably related to the secondary discharge caused by the erosion debris produced in the Galinstan μEDM.

Fig. 5
figure 5

Needle tip. a Original state, b used in Galinstan μEDM

Full size image

To eliminate the effects of discharge on the needle surface, a Parylene C film (thickness, 20 μm) was coated on the stainless steel needle for use in the process [21]. The coated needle B is presented in Fig. 6a. The coating consequently modified the DI and DO of the needle to 60 μm and 280 μm, respectively. Figure 6b displays the state of this needle tip after use for Galinstan μEDM (in which a 2-mm-long groove was eroded in 40 min). The insulation effect is clearly seen from this image, which indicates no sign of discharge on the needle surface.

Fig. 6
figure 6

The needle coated with 20-μm-thick Parylene C. a Coated state, b used in Galinstan μEDM

Full size image

5.2 Three-dimensional erosion

Three-dimensional erosion was tested using the programmed motions of the 3-axis stage. Figure 7a shows the design model of a square microcavity (top: 1 × 1 mm, depth: 0.3 mm). The tool scanning path in the X–Y plane is shown in Fig. 7b with an overlap of 30 μm. To prevent uneven side edges caused by single direction scanning, the scanning path adopts a snake pattern, and different scanning directions are set according to the cycle parity. When the scan number is odd, the scanning path is horizontal, and when the scan number is even, the scanning path is vertical; this process is repeated until the depth of the cavity is reached to ensure that the bottom surface is smoothly machined. Sample erosion was performed using coating needle B with C = 27 nF, R = 1 kΩ, Vs = 100 V, P = 1.2 atm, and D = 100 μm. A square microcavity was well realized on the silicon sample by controlling the XY stage scanning path, as shown in Fig. 7c. The square cavity has a draft angle of several degrees because the liquid electrode is a cone-shaped tip instead of a square tip. In addition, in Fig. 7c, visible pits exist around the edge of the processed cavity. The reason for the random distribution of micropits near the edges of the produced cavity is that some pulses were likely generated between the suspended liquid microelectrode and the liquid alloy that had already dripped on the workpiece surface.

Fig. 7
figure 7

Square microcavity images produced on a silicon substrate using the developed process: a design model; b planning path; c square microcavity

Full size image

5.3 MRR and LACR

The machining performance was evaluated in terms of the material removal rate (MRR) and liquid alloy consumption rate (LACR). The MRR is computed as the ratio of the material removed from the workpiece (approximated as the volume of the frustum of the cone) to the EDM system’s recorded machining time:

$$ MRR=\frac{V_F}{t}\kern0.5em \left({\mathrm{mm}}^3/\min \right) $$
(14)

where VF is the volume of removed material and t is the machining time recorded by the EDM system. In this paper, a square microcavity was adopted to verify the MRR and LACR and its erosion characteristics, as mentioned above. Thus, VF can be expressed as

$$ {V}_F=\frac{1}{3}h\left({a}^2+ ab+{b}^2\right)\kern0.5em \left({\mathrm{mm}}^3\right) $$
(15)

where a and b are the top and base side lengths of the truncated pyramid and h is the erosion depth. The taper is very small, so to simplify the calculation, a can be set equal b; then,

$$ {V}_F={a}^2h\kern0.5em \left({\mathrm{mm}}^3\right) $$
(16)

The LACR was calculated as the ratio of the liquid alloy consumed to the machining time, as given in the following:

$$ LACR=\frac{V_L}{t}\kern0.5em \left({\mathrm{mm}}^3/\min \right) $$
(17)

where VL is the volume lost due to liquid alloy consumption.

Experiments were carried out to evaluate the effects of the open voltage VS, capacitance C, resistance R, pressure P and needle inner diameter on the MRR and the LACR under both positive and negative polarities (the workpiece connected to the positive terminal of the pulse generator and the tool electrode connected to the negative terminal of the pulse power supply is called positive processing, and the opposite setup is called negative processing). The workpiece was silicon with a thickness of 3 mm and was immersed in dielectric EDM fluid in the experiment. a and h were measured using the 3-axis stage through electrical contact with nonelectrically passivated needles. VL can be directly recorded from the syringe. The detailed experimental variables are summarized in Table 3.

Table 3 Summary of experimental conditions
Full size table

Figure 8 shows the relationship between the MRR and VS (needle B, with C = 100 nF, R = 0.5 kΩ and P = 1.2 atm, as shown in Table 3), which clearly shows that the MRR consistently increases with the voltage increases. This trend is observed for both polarity processing conditions. This result is reasonable because the discharge energy of a single pulse, which determines the amount of material eroded, increases with increasing voltage.

Fig. 8
figure 8

The effect of open voltage VS on the MRR

Full size image

Figure 8 also shows that the MRR under positive processing is slightly greater than that under negative processing. The reason is that the RC pulse generator has a high discharge frequency and a short discharge pulse time. With a short discharge pulse time, due to the small mass, small inertia, and flexible movement of electrons, a large number of electrons easily travel toward and impact the positive surface, which causes the positive surface to rapidly melt and gasify. In contrast, the positive ion inertia is large, and the movement is slow. Only a small part of these ions can reach the negative surface, so the damage effect of electrons is greater than the damage effect of positive ions; therefore, more of the positive material is eroded compared to the negative material.

Figure 9 shows a similar trend for the MRR obtained by varying capacitance C. When the capacitance is less than 3.3 nF, a short circuit can easily occur in the discharge gap because the discharge energy of single pulse is too small to remove the debris away from the discharging location.

Fig. 9
figure 9

The effect of capacitance C on the MRR

Full size image

Figure 10 plots the MRR as a function of resistance R. MRR decreases as resistance R increases in accordance with the RC pulse generator principle. In other words, the higher the resistance R, the lower the discharge frequency, leading to a smaller removal rate or smaller MRR of the process. Here, the frequency of the RC pulse generator with R = 0.5 kΩ and C = 100 nF was less than 3.2 kHz.

Fig. 10
figure 10

The effect of resistance R on the MRR

Full size image

Real-time feeding of consumed Galinstan is very important for high efficiency and stable processing, which largely depends on the applied pressure, as previously discussed. Therefore, the effect of the feeding speed on the MRR cannot be ignored. Figure 11 shows the relationship between the MRR and pressure P. As shown in the graph, an optimal pressure value for MRR is found under the same discharge conditions. The reason is that the consumed Galinstan cannot be compensated in time during the discharge process when the pressure is small, leading to an open circuit state most of the time, thus resulting in the small value of the MRR. With increasing pressure, the frequency of discharge will increase with increasing Galinstan compensation, which increases the MRR. However, above a certain value, as the pressure increases, excess Galinstan will be suspended on the needle tip, and this droplet will easily fall due to the discharge force and electrostatic force and cause unstable discharge, thus resulting in the MRR declining. Thus, the pressure is very significant for the stable process. Based on the graph, 1.3 atm pressure is the optimum value.

Fig. 11
figure 11

The effect of pressure P on the MRR

Full size image

Figure 12 shows the effect of two different needles on the MRR. The graph shows that under both polarity processing conditions, the larger the inner diameter of the needle, the larger the MRR. The reason is likely that the larger the inner diameter of the needle, the higher the discharge probability due to the larger discharge area facing the workpiece. The MRR under positive processing is slightly higher than that under negative processing for both types of needles, which arises from the characteristics of RC pulse discharge mentioned above. However, as shown in Fig. 12, regardless of positive or negative processing, the needle inner diameter only has a slight influence on the MRR, and the most important factor affecting the MRR is still the discharge energy of a single pulse.

Fig. 12
figure 12

The effect of different needles on the MRR

Full size image

As shown in Figs. 13, 14, and 15, the LACR clearly has the same trends as the MRR with voltage, capacitance, and resistance. Increasing the discharge energy or discharge frequency causes a greater spark intensity, which generates more melting material in the spark region. Thus, the workpiece and the electrode material are both subjected to increased sparks, which increases both the MRR and the LACR. Because the electrons flow from the negative terminal toward the positive terminal, more of the positive material is eroded compared to the negative material during the short pulse discharge of the RC pulse generator. As a result, the MRR and LACR substantially increase with increasing discharge energy, and the MRR under positive processing is larger than that under negative processing; however, the LACR under negative processing is larger than that under positive processing.

Fig. 13
figure 13

The effect of voltage VS on the LACR

Full size image
Fig. 14
figure 14

The effect of capacitance C on the LACR

Full size image
Fig. 15
figure 15

The effect of resistance R on the LACR

Full size image

Figure 16 shows the relationship between the LACR and pressure P. As shown in the graph, the LACR increases with the pressure increases in a consistent manner. This result can be explained by Eq. (13). The higher the pressure is, the higher the flow velocity, which increases the consumption of Galinstan. However, in the lower pressure region, LACR differs under the two polarity processing conditions. The LACR under negative processing is larger than that under positive processing. This phenomenon is due to the combination of the narrow pulse discharge characteristics of the RC pulse generator and the electrostatic force. In negative processing, the impact force of high concentration electrons causes the Galinstan droplet suspended at the end of the needle to fall more easily. At the same time, gasification is easier under the high temperature of the discharge, which will cause substantial liquid alloy loss. Meanwhile, due to the siphoning effect, the liquid in the needle will be more quickly fed to the end. The comprehensive result is that the LACR is larger. When the pressure increases, the difference in the LACR for the two polarities decreases. This outcome is reasonable because the flow rate, which determines the amount of Galinstan consumed based on the velocity of the flow, increases with the pressure increases. The amount of Galinstan fed is much greater than the amount of Galinstan consumed by discharge at an excessive flow velocity. Therefore, regardless of which polarity processing is used, the LACR will be the same.

Fig. 16
figure 16

The effect of pressure P on the LACR

Full size image

As shown in Fig. 17, the smaller the needle, the smaller the LACR under the same energy discharge conditions. The reason is that the liquid alloy is viscous, and the flow rate is proportional to the square of the inner diameter under the same pressure, as depicted in Eq. (13). The smaller the inner diameter of the needle, the smaller the flow rate, which results in a decrease in the LACR. At the same time, the LACR does not obviously vary with polarity for needle B. The LACR under negative processing is slightly larger than that under positive processing. This phenomenon may be caused by inadequate Galinstan feeding. If the amount of newly supplied Galinstan is less than the amount of Galinstan consumed by discharge, then the LACR largely depends on the flow rate.

Fig. 17
figure 17

The effect of different needles on the LACR

Full size image

6 Conclusion

The experimental results verified the feasibility of using liquid alloy as an electrode during μEDM processing for the purpose of resolving the problems related to electrode wear in traditional μEDM. The coating of an electrically passivated material on the needle suitably protects the needle tip from the thermal action of the pulse discharge. Square microcavities eroded on silicon samples using a liquid alloy electrode were well demonstrated.

The influence of the open voltage VS, capacitance C, resistance R, pressure and needle inner diameter on the MRR and LACR under both polarities were also investigated. Both MRR and LACR consistently exhibited the same trend with the single pulse discharge energy; that is, they increased with the discharge energy. In the case of the MRR, the discharge energy was the most influential factor. In the case of the LACR, the most influential factor was pressure. Moreover, the smaller the needle, the higher the utilization rate of liquid alloy. For the RC pulse generator, both the MRR and LACR were slightly different under the different polarities; however, the MRR and LACR trends under the different polarities were the opposite.