Micro Eletro Discharge Milling for Microfabrication

Micromachining is used to fabricate three-dimensional microstructures and it is the foundation of a technology called Micro-Electro-Mechanical-Systems (MEMS). Bulk micromachining and surface micromachining are two major categories (among others) in this field. This book presents advances in micromachining technology. For this, we have gathered review articles related to various techniques and methods of micro/nano fabrications, like focused ion beams, laser ablation, and several other specialized techniques, from esteemed researchers and scientists around the world. Each chapter gives a complete description of a specific micromachining method, design, associate analytical works, experimental set-up, and the final fabricated devices, followed by many references related to this field of research available in other literature. Due to the multidisciplinary nature of this technology, the collection of articles presented here can be used by scientists and researchers in the disciplines of engineering, materials sciences, physics, and chemistry. following:


Introduction
Miniaturization of product is increasingly in demand for applications in numerous fields, such as aerospace, automotive, biomedical, healthcare, electronics, environmental, communications and consumer products. Researchers have been working on the microsystems that promise to enhance health care, quality of life and economic growth. Some examples are micro-channels for micro fuel cell, lab-on-chips, shape memory alloy 'stents', fluidic graphite channels for fuel cell applications, miniature actuators and sensors, medical devices, etc. (Madou, 2002;Hsu, 2002). Thus, miniaturization technologies are perceived as potential key technologies. One bottleneck of product miniaturization is the lack of simpler and cheaper fabrication techniques. Currently the common techniques are based on silicon processing techniques, where silicon-based materials are processed through wet and dry chemical etching. These techniques are suitable for microelectronics, limited to few silicon-based materials and restricted to simple two dimensional (2D) or pseudo three dimensional (2.5D) planar geometries. Other fabrication processes, such as LiGA (lithography, electroforming and molding), laser, ultrasonic, focused ion beam (FIB), micro electro discharge machining (EDM), mechanical micromilling, etc. are expensive and required high capital investment. Moreover these processes are limited to selected materials and low throughput (Ehmann et al. 2002). A less expensive and simpler microfabrication technique is sought to produce commercially viable microcomponents. Micro electro discharge (ED) milling, a new branch of EDM, has potential to fabricate functional microcomponents. The influences of micro ED milling process parameters on surface roughness, tool wear ratio and material removal rate are not fully identified yet. Therefore, modeling of ED milling process parameters for surface roughness, tool wear ratio and material removal rate are necessary.

Micro electro discharge machining
EDM has been successfully used for micromachining with high precision regardless of the hardness of work piece material. It uses the removal phenomenon of electrical discharges in a dielectric fluid. Two conductive electrodes, one being the tool and the other the workpiece, are immersed in a liquid dielectric. A series of voltage pulses are applied between the electrodes, which are separated by a small gap. A localized breakdown of the dielectric www.intechopen.com occurs and sparks are generated across the inter-electrode gap, usually at regions where the local electric field strength is highest. Plasma channels towards workpiece are formed during the discharge and high speed electrons come into collision with the workpiece. Each spark erodes a small amount of work material by melting and vaporizing from the surface of both the electrodes. The momentary local plasma column temperature can reach as high as 40,000 K (DiBitonto et al., 1989).  (Kim et al., 2005) This high temperature causes the melting and vaporization of the electrode materials; the molten metals are evacuated by a mechanical blast, resulting in small craters on both the tool electrode and work materials. It is understood that the shock waves, electromagnetic and electrostatic forces involved in the process are responsible for ejection of the molten part (debris) into the dielectric medium. The repetitive impulse together with the feed movement (by means of a servo mechanism) of the tool electrode towards the workpiece enables metal removal along the entire surface of the electrodes. Figure 1.1 shows the schematic of EDM principle. Production of micro-features using EDM pays significant attention to the research community. This is due to its low set-up cost, high accuracy and large design freedom. Compared to lithography based miniaturization, micro EDM has the clear advantages in fabricating complex 3D shapes with higher aspect ratio (Lim et al., 2003). Moreover, all conductive materials regardless of hardness can be machined by EDM. Conventional EDM is especially useful to produce molds and dies. Micro EDM is now basically focused on the fabrication of miniaturized product, like molds and dies, with greater surface quality. In this endeavor new CNC systems and advanced spark generators are found as great assistance (Pham et al., 2004). Current micro EDM technology used for manufacturing micro-features can be categorized into four different types: a) Die-sinking micro-EDM, where an electrode with micro-features is employed to produce its mirror image in the workpiece. b) Micro-ED drilling, where micro-electrodes (of diameters down to 5-10 µm) are used to 'drill' micro-holes in the workpiece. c) Micro-ED milling, where micro-electrodes (of diameters down to 5-10 µm) are employed to produce complex 3D cavities by adopting a movement strategy similar to that in conventional milling. d) Micro-wire EDM, where a wire of diameter down to 20 µm is used to cut through a conductive workpiece. Precision micro holes required for gas and liquid orifices in aerospace and medical applications, pinholes for x-ray and nuclear fusion measurements, ink-jet printer nozzles and electron beam gun apertures can be fabricated with high precision accuracy and with www.intechopen.com surface roughness less than 0.1 µm using the micro EDM process (Dario et al., 1995). Although micro EDM plays an important role in the field of micromachining, it has disadvantages such as high tool-electrode wear ratio and low MRR. The wear of electrode must be compensated either by changing the electrode or by preparing longer electrode from the beginning or fabricating the electrode in situ for further machining (Asad et al., 2007).

Micro electro discharge milling
In micro ED milling the material is eroded by non-contact thermo-electrical process, where a series of discrete sparks occur between the workpiece and the rotating tool electrode. The workpiece and tool are immersed in a dielectric fluid. The dielectric fluid is continuously fed to the machining zone to convey the spark and flush away the eroded particles. The work feeding system of micro ED milling is similar to mechanical end milling process. Like mechanical end milling, here the workpiece is fed to the tool electrode while the tool is in rotation. The tool movement is controlled numerically to achieve the desired threedimensional shape with high accuracy. Figure 1.2 shows the schematic of micro ED milling.

Process parameters
Micro ED milling has a number of process parameters, which are influential on the machining performance. Depending upon the circuit type used in the machine, the importance of parameters is also varied. Table 1.1 shows different micro ED milling process parameters. Some researches were conducted to optimize different EDM process parameters for R a , R y , TWR and MRR (Puertas and Luis, 2004;Lin et al., 2006). But almost all of them were on conventional macro EDM. Because of the stochastic thermal nature of the EDM process, it is difficult to explain all of those effects fully. The full understanding of micro ED milling process parameters are yet to be developed because the process itself is a new branch of manufacturing. For electric discharge the discharge energy (E) generated in the discharge circuit can be expressed roughly as follows: Where, C = capacitance of capacitors, C ' = total stray capacitance, V = discharge voltage According to the equation (1.1), there are two ways to reduce the discharge energy: by reducing the discharge voltage or by reducing the total capacitance (C + C ' ). However, low voltage results in an unstable discharge. Therefore, the total capacitance should be controlled primarily. The total stray capacitance (C') in the circuit plays a significant role here. The following subsections discuss the influence of different micro ED milling parameters on surface roughness, tool wear and material removal rate.

Surface roughness
Variation of discharge energy affects surface roughness. It is experimentally identified that with the increase in discharge energy either by increasing voltage or capacitance, R a increased in die sinking EDM, wire EDM (WEDM) and WEDG (Uhlmann et al., 2005;Han et al., 2006;Chiang, 2007). A large discharging energy usually causes violent sparks and results a deeper erosion crater on the surface. Accompanying the cooling process after the spilling of molten metal, residues remain at the periphery of the crater to form a rough surface (Liao et al., 2004). Theoretical model for R y or R max was developed for transistorized pulse generator type EDM (Chen and Mahdivian, 2000). It showed that with the increase in current and pulse duration, R y increased. Surface roughness of ceramic due to EDM was also investigated. EDM of boron carbide (B 4 C) showed that R a and R y increased with pulse duration but decreased with current (Puertas and Luis, 2004). On the other hand, EDM of high speed steel showed that with the increase of current, R a increased and with the increase of pulse duration it decreased (Lin et al., 2006). It was suggested to use low conductive or higher resistive dielectric to ensure low R a and R y (Liao et al., 2004). Use of rotating electrode was helpful for high shape accuracy and higher MRR but it caused higher surface roughness (Her and Weng, 2001). Formation of crack is a challenge in EDM. The presence of high crack density is not desirable because it leads early product failure. Investigation of EDM parameters showed that higher pulse-on duration increased both the average white layer thickness and the induced stress (Lee and Tai, 2003;Ekmekci et al., 2005). These two conditions tend to www.intechopen.com promote crack formation. When the pulse current was increased, the increase in material removal rate caused a high deviation of thickness of the white layer. Thus with the increase in pulse current, crack formation decreased. It was also verified that lower thermal conductivity of workmaterial caused higher crack formation. There was no analysis reported on the effect of feed rate on machined surface.

Tool wear
Tool wear is responsible for shape inaccuracy and dimensional instability. In micro ED milling at higher feed rate tool wear was found decreasing (Lim et al., 2003). It was also observed that tool wear was increasing with feed rate but after reaching the peak it started decreasing (Kim et al., 2005). High discharge energy usually caused high TWR (Uhlmann et al., 2005). In EDM of high speed steel and boron carbide (B 4 C), tool wear was found to be proportional to current and inversely proportional to pulse duration (Puertas and Luis, 2004;Lin et al., 2006). It was also found that tool wear showed an optimum peak value while pulse duration had increased (Ozgedik and Cogun, 2006). However, use of shorter pulse ontime was suggested to achieve lower TWR (Son et al., 2007). The negative polarity of the tool gives a lower tool wear than that of positive polarity in the range of low to medium discharge current values. At high current settings, the polarity has no significant effect on the tool wear. A slight decrease in TWR was observed with increasing current in negative polarity. In the case of positive polarity, TWR decreased significantly with increasing current (Lee and Li, 2001). The type of dielectric fluid is also responsible for tool wear rate. Kerosene is frequently use as the dielectric medium. In EDM, carbide layer formed on the workpiece surface with the use of kerosene, which has a higher melting temperature than the oxide layer formed with the use of distilled water. The carbide layer formed by kerosene needed higher pulse energy for melting and evaporation which caused high tool wear (Chen et al., 1999). The use of low resistive deionized water as a dielectric fluid reduced the tool wear as compared to kerosene (Chung et al., 2007).

Material removal rate
Experimental investigation in conventional EDM found that MRR increased with the discharge current. With the increase in pulse duration MRR showed a peak and then downfall (Chen and Mahdivian, 2000;Lin et al., 2006;). Increase in discharge energy by discharge voltage or capacitance, resulted in higher MRR (Kim et al., 2005). It was also observed that material removal depth was inversely proportional to feed rate (Lim et al., 2003). MRR can be increased in both positive and negative polarity by rotating the toolelectrode (Her and Weng, 2001). Higher MRR was obtained for low ultrasonic vibration frequency combined with tool rotation (Ghoreishi and Atkinson, 2002). Dielectric flushing system is also responsible for the variation in MRR. The low material removal rate in static condition is mainly due to improper flushing of the molten and evaporated workpiece material from the machining gap (Ozgedik and Cogun, 2006). With the increase in flushing flow rate, MRR increased. Use of kerosene and tap water as the dielectric fluid showed higher MRR than distill water (Chen et al., 1999).

Tool electrode
EDM tool electrodes should be conductive. A list of different tool electrode materials is shown in Table 1.2. In micro range, the replacement of tool during machining is www.intechopen.com unacceptable because it leads to positional inaccuracy. For this, tool with low wear ratio are needed in micro EDM. Because of its low wear ratio, tungsten or tungsten carbide rods or tubes of diameters within the range 100-400 µm are preferred as tool electrodes for micro ED drilling and milling (Pham et al., 2004). The handling of tungsten and tungsten-carbide rods is difficult as they can be easily damaged. Therefore, sub-systems are incorporated into micro EDM machines for on-the-machine manufacture and holding of the required microelectrodes. The most common sub-systems are ceramic guides and dressing units such as WEDG.

Electrode material
Volume wear ratio (tool/workpiece) Machinability Cost Best application  (Fleischer et al., 2004) www.intechopen.com The tool electrode can be dressed or prepared either by etching, grinding or lithography process. However, when the fabricated tool is chucked on the machine, an alignment error between tool center and spindle center and run-out error of tool which has an influence on the variation of the discharge gap occurs and cause serious problems in micro machining and assembly of micro parts. For this, on-machine tool dressing is always recommended. WEDG is an on-machine tool dressing method. The principle of WEDG was developed in the mid 1980s. The WEDG process is using a traveling wire between one or two guides. The wire is the tool electrode of 50 and 200 µm diameter. In Figure 1.3 the principle of the WEDG-unit with one wire guide is shown, where rotating workpiece is fed downwards and the discharge area is limited to the front edge of the wire. The straightness of the ground part depends only on the variation of the gap distance at the machining point. Minimum 5.5 m tool was fabricated by WEDG from experiment (Kim et al., 2005). But in WEDG it is difficult to control the shape, dimension and roughness of the ground electrode (Lim et al., 2003).

Experimental design
The experiment was designed based on 3-level full factorial statistical model, where the three micro ED milling process parameters were feed rate, capacitance and voltage. The experiment was designed to analyze the individual influences, interaction effects and quadratic effects of these parameters on R a , R y , TWR and MRR. Table 2.1 shows the experimental conditions. Total number of 32 experiments had been conducted to complete the analysis. Six experiments at the same combination were conducted to reduce the curvature effect. The response parameters were statistically analyzed by applying factorial ANOVA at probability value of 0.05. This experimental study has been briefly published before (Mehfuz and Ali, 2009

Tool and workpiece materials
The workpiece material was beryllium-copper (Be-Cu) alloy (Be = 0.4 %, Ni = 1.8 % Cu = 97.8 %), which is industrially known as Protherm. Some features of Be-Cu alloy are as follows: a) High thermal conductivity (245.00 W/m o C at 20 O C). b) Resistance to high temperature. c) Excellent corrosion resistance, which enables it to become a suitable mold material. d) Good machinability and polishability. The tool electrode material was tungsten (W). It has the following features: a) Low tool wear ratio, b) Suitable for small holes or slot, c) High cost, d) Machinable by WEDG.

Experimental procedures and measurements
In this subsection the equipments used to conduct the experiments and the experimental procedures are described. A commercial multi-purpose miniature machine tool (DT-110 Mikrotools, Singapore), shown in Figure 3.1, was used to conduct the micro ED milling experiments. The machine contained RC circuit to generate small but high frequency discharge energy. EDM-3 synthetic oil was used as the dielectric fluid. The following subsections discuss the different steps of the experiments.

Sample preparation
Before the experiments, the workpieces were prepared by grinding and polishing. Workpieces were ground by using (NAGASEI SGW 52, Japan) surface grinding machine. Next, the workpiece was grounded manually by using 240, 320, 400 and 1000 grade of sand papers respectively. Then the workpieces were polished by 3 µm diamond suspensions and lubricant. The polishing steps were needed to obtain mirror surface finish. After the final polishing, the workpiece were immersed in acetone and cleaned for 15 minutes by ultrasonic cleaning machine, (BRANSON 2510). Tungsten tool electrodes of 500 µm diameter were also cleaned by the same ultrasonic cleaning machine to clean. Ultrasonic cleaning is useful for the removal of micro sized dirt particle. Then both the workpieces and tool-electrodes were weighed on an electric balance (B204-S Mettler Toledo, Switzerland) with a resolution of 0.1 mg. The measured weights before and after machining were compared. The following shows the specifications of a sample workpiece and tool-electrode after the final polishing. The prepared samples were then used for micro ED milling as discussed in following subsection.

Experiments by micro electro discharge milling
At the beginning of the micro ED milling set-up, the tool electrode was clamped in the spindle, while the workpiece was clamped on the worktable. After clamping, the spindle rotation was fixed at a speed of 2000 rpm, the worktable was fed and dielectric fluids were flown to the sparking zone. When discharge occurred between the tool-electrode and the workpiece, a small portion of workpiece melted and evaporated due to high local temperature, creating a crater on the work surface. The workpiece cooled rapidly by the influence of dielectric fluid. Figure 2.1a schematically depicts the experimental setup. The experiment was done to fabricate a microchannel for the dimensions of 13 mm length, 0.5 mm width and 0.2 mm depth. Figure 2.1b shows the schematic of the microchannels after micro ED milling. Computer numerical coding was used to conduct the experiment. The machining time was obtained from the auto stopwatch of the integrated computer. After the experiment, both the tool-electrode and workpiece were cleaned again in ultrasonic cleaning bath and dried. After this, the tool-electrode and workpiece were weighed to find out the weight loss due to machining. The differences of weight before and after machining were used to calculate the TWR and MRR. After weighing, the samples were put into a dry box to avoid the moisture www.intechopen.com contact. These samples were used for SEM investigation and measurement of R a and R y as discussed in the following subsections.

SEM Inspection
The work sample was cleaned by ultrasonic cleaning and sputter coated with carbon by Coater (SC 7640). Though the work material was conductive, it was coated to get high quality images. The machined surfaces were then inspected by scanning electron microscope (SEM) (JEOL JSM-5600). Figure 2.2 shows SEM micrographs of a microchannel produced by micro ED milling.

Measurement of surface roughness
Both the R a and R y were measured by using a precision surface profiler (Mitutoyo, Surftest SV-500, Japan). This profiler provided surface roughness along a line with a resolution of 10 nm. As the length of the microchannel was 13 mm long, the roughness was measured in three segments. Each segment was 3.5 mm long. Average of these three measurements was taken as the surface roughness value.

Results and discussions
The experimental results of R a , R y , TWR and MRR , are analyzed using variance approach (ANOVA) to check the adequacy of the developed statistical model. The analysis ultimately showed the main and interaction effects of the process variables on the responses. Main effect was the direct effect of an independent variable while interaction effect was the joint effect of two independent variables on the response.

Calculation of analysis of variance
ANOVA is a calculation procedure to allocate the amount of variation in a process and determine if it is significant or is caused by random noise. ANOVA tables are constructed for R a , R y , TWR and MRR respectively. The statistical significance of each effect by means of the comparison of the mean squares (MS) with an estimation of the experimental error was performed. In the   Table 3.2 shows the analysis of variance for R a . Using this analysis a second order quadratic model was developed, which is shown below in Equation (3.1). The model was developed for 95% level of confidence. The model F-value of 213.06 implies the model is significant. There is almost no influence, 0.01%, of noise on the model developed. By checking F value and P-value it is clearly seen that Factor B (capacitance), Factor C (voltage) and Factor B 2 are most influential on R a . The P-value of each of these factors indicates the confidence level is more than 99.00%, which shows their very strong influence. The P -value of interaction effects of BC shows the confidence level is above 95.0% and thus shows very good influence on R a . The P-value of Factor A (feed rate) and interaction Factor AC has insignificant influence over R a as it provides very high P values. The lack of fit F-value of 0.53 implies that the lack of fit is not significant compare to the pure error. The high P-value of lack of fit, 85.35%, indicates the model is fit, while the very low P-value of the model, 0.01%, indicates that the model is significant. The specific power transformation was chosen within the confidence level, which was suggested by the Design Expert software toolbox using Box-Cox plotting. In this case, natural log power transformation was suggested. Thus the developed statistical quadratic equation for R a is:  Capacitance and voltage are strongly influential on R a . R a increases along with the increase in voltage, which is same as the conventional EDM. With the increase of capacitance from 0.1 to 5 nF the R a value increases but for further increase of capacitance the R a values reduce. As the capacitance increased, large energy dissipated which erodes more materials with stronger spark. This strong spark erodes materials with high amount of debris from both the tool electrode and workpiece creating uneven crater. As these debris are trapped in between the plasma channel, it causes unwanted spark. Thus, a high amount of discharge energy is employed to spark with debris, while work material is effectively removed by a small portion of discharge energy. Thus lower R a is obtained.   Figure 4.2b shows the surface texture (R a = 0.22 µm) with the capacitance of 1 nF. Here, the surface was found rougher with increased amount of unwashed debris. Similarly the developed statistical quadratic equation for maximum peak-to-valley height roughness, R y is:

Tool Wear Ratio
A second order quadratic model was developed which is the mathematical expression of TWR: interaction effect of feed rate-voltage. TWR decreases with the increase in feed rate. For further increase in feed rate TWR starts rising. In the first phase, for a particular spark energy discharge if the feed rate is so small it will have high electron emission rate from the tool electrode, resulting high TWR. As the feed rate increase the spark energy is more involved in material erosion, which reduces TWR and reaches to minimum. In the second phase, for further increase of feed rate from the optimum, the unflashed eroded materials cause unwanted spark with the tool, which results more tool wear. Thus high TWR is obtained. With the increase in capacitance large energy dissipated which produces stronger spark resulting high work material erosion. Higher spark energy produces higher amount of debris. These debris sticking on the workpiece, trapped in and causes unwanted spark. The unwanted sparks erodes materials from the tool electrode, which results high tool wear. Thus, higher capacitance results higher TWR. As significant amounts of spark energy are employed in sparking with debris, effectively a lower amount of work material is eroded. TWR is a ratio of tool wear and workpiece wear, so decreases in material removal mean increase in TWR.
Therefore, the decrease of MRR 1 with the increase of any of the factors means the increase in MRR. MRR increases with the increase in feed rate. For further increase in feed rate MRR starts falling slightly. As the feed rate increase, the spark energy is more involved in material erosion, which increases MRR till reaches to the optimum. For further increase of feed rate from the optimum, the unflashed eroded materials cause unwanted spark with the tool, which changes the tool shape making it uneven and rough. The spark discharge from this uneven tool results less MRR. With the increase in capacitance large energy dissipated which erodes more work materials with stronger spark. With the material erosion, the unflashed debris sticking on the workpiece is trapped in between the machining zone, which causes unwanted spark with the tool-electrode. Thus a great portion of discharge energy occupies with unwanted sparking, while the remaining erodes the work material. Hence, effectively a lower amount of work material is eroded.

Multiple response optimizations
This subsection discusses the optimization of four output responses, R a , R y , TWR and MRR.
The developed non-linear models Equations (3.1) to (3.4) are used for multiple response optimization. Desirability function approach was used for optimization. The method finds operating conditions, e.g. feed rate, capacitance and voltage that provide the "most desirable" values of the responses, e.g. R a , R y , TWR and MRR. The following subsection describes the desirability function approach.

Desirability function approach
In the analysis the objective function, (3.5) where n is the number of responses in the measure. From the equation (4.5) it can be noticed that if any response Y i is completely undesirable (d i (Y i ) = 0), then the overall desirability is zero. In this case, the geometric mean of overall desirability is as follows: Depending on whether a particular response Y i is to be maximized, minimized, or assigned a target value, different desirability functions d i (Y i ) can be used. In this case, R a , R y and TWR are needed to be minimized while MRR are needed to maximized. Following are the two desirability functions: where,  Table 3.3 shows the process parameters obtained after multiple response optimization. For the shown values of process parameters, it is 88.06% likely to get the R a 0.04 µm, R y 0.34 µm, TWR 0.044 and MRR 0.08 mg/min. Any other combination of the process parameters will either statistically less reliable or give poor results of at least one of the responses. The analysis was done by using computer software, Design Expert.

Verification of optimized values
Experiments were conducted to verify the result obtained from the multiple response optimization. The actual values obtained from the experiments are compared with the predicted values in

Application of micro ed milling: Micro swiss-roll combustor mold
Micro swiss-roll combustor is a heat re-circulating combustor. It uses hydrocarbon fuels to generate high density energy. The advantage of a micro swiss-roll combustor is that it provides high density energy by reducing heat loss [Ahn and Ronney, 2005;Kim et al., 2007]. The generated heat inside the micro swiss-roll combustor is entrapped and recirculated. Thus, high density energy is obtained. One of the challenges in micro combustor design is to reduce the heat loss. To reduce the heat loss by reducing surface-to-volume ratio, wall thickness should be as small as possible [Ahn et al. 2004]. The application of micro swiss-roll combustor includes portable electronics, such as cell phone, laptop, space vehicles, military uses, telecommunication, etc. The proposed design of the micro swiss-roll combustor mold cavity is shown in Figure 5.1. Beryllium-copper alloy (Protherm) was selected as the mold material, because of its high thermal conductivity, high heat and corrosion resistance. The microchannel of the mold cavity was fabricated by using a tungsten tool-electrode of 100 µm diameter. The minimum gap between two microchannels was 380 µm. The preliminary drawing and the numerical code (NC) of the design was generated by using CATIA V.5 R14 computer aided drafting software.

Fabrication of tool electrode by WEDG
Commercially available 300 µm diameter cylindrical tungsten rod was first dressed to 100 µm diameter by WEDG. Later this rod was used as a tool-electrode in micro ED milling to fabricate microchannels. Figure 4.2a illustrates the mechanism of WEDG. Figure 4.2b is the picture taken during the experiment and Figure 4.2c illustrates the SEM image of fabricated tool electrode. Computer numerical coding was used to control the size and shape of the required tool-electrode. The process parameters used are shown in

Fabrication of micro mold cavity
The micro swiss-roll combustor mold cavity was fabricated by micro ED milling. Be-Cu alloy plate of 6 mm thickness was used as the workmaterial. The tool-electrode of 100 µm diameter was used, which produced microchannels of 120 µm width and 1 mm depth. Channel width comprises of the tool diameter and spark gap. Layer by layer approach was chosen to get better dimensional accuracy. The thickness of each layer was 200 µm. Figure  4.4 explains the layer by layer approach. The gap between two microchannels was 380 µm. After machining each 500 µm, the tool-electrode was dressed by WEDG to reduce the shape inaccuracy due to tool wear. The whole machining was done using computer numerical control. Figure 4.5a is the picture during experiments, Figure 4.5b shows the final product and Figure 4.5c shows the SEM micrographs of the window A in Figure 4.4b. The process parameters obtained from the multiple responses optimization were used in the microfabrication. The experimental condition is shown in  www.intechopen.com

Conclusion
Micro ED milling is shown as a potential fabrication technique for functional microcomponents. Influences of three micro ED milling parameters, feed rate, capacitance and voltage, were analyzed. Mathematical models were developed for output responses R a , R y , TWR and MRR. Analysis of multiple response optimization was done to get the best achievable response values. The micro ED milling process parameters obtained by the multiple response optimization were used in the fabrication of micro mold cavity. WEDG was used to dress the tool-electrode to a diameter of 100 µm. The final product was a micro swiss-roll combustor mold cavity. In brief, this research showed the followings: 1. Capacitance and voltage have strong individual influence on both the R a and R y , while the interaction effect of capacitance and voltage also affects the roughness greatly. Ususally higher discharge energy results higher surface roughness. The unflushed debris sticking on the workpiece causes higher R a and R y . At very high discharge energy the entrapped debris inside the plasma channel creates unwanted spark with the tool-electrode. Thus only a small portion of discharge energy involves in material erosion process, which results low R a and R y . 2. Capacitance and voltage plays significant role on TWR along with the interaction effect of feed rate and voltage. At high discharge energy large amount of debris are produced, which causes high TWR by generating unwanted sparks with the tool-electrode. 3. Feed rate, capacitance and voltage have strong individual and interaction effects on MRR.
Usually, MRR is higher at high discharge energy. But the presence of high amount debris in the plasma channel often creates unwanted spark with the tool electrode. Thus only a portion of energy involves in workmaterial removal, which reduces MRR. 4. Multiple response optimization shows 88.06% desirability for minimum achievable values of R a , R y , TWR and maximum achievable MRR, which are 0.04 µm, 0.34 µm, 0.044, 0.08 mg/min respectively when the feed rate, capacitance and voltage are 4.79 µm/s, 0.10 nF and 80.00 volts respectively. The achieved R a and R y values are in the acceptable range for many MEMS applications. 5. The result of multiple response optimization was verified by experiment. The percentages of errors for R a (0.0%), R y (5.56%) at 88.06% desirability were found within the acceptable range. For TWR (16.98%) and MRR (11.11%), it was found relatively unsteady. Low resolution (0.1 mg) of electric balance could be a reason behind this. 6. A micro swiss-roll combustor mold cavity was fabricated by using the WEDG dressed tool. Optimized and verified micro ED milling process parameters were used for fabrication. The final product has the channel dimension of 0.1 mm. 7. Combination of micro ED milling and molding can be a suitable route for the mass replication of miniaturized functional components at a lower cost.