Ken Heinz

Graduate Research Assistant

Shiv G. Kapoor
Grayce Wicall Gauthier Chair in Mechanical Science and Engineering

An Investigation of MagneticField-Assisted Material Removal in Micro-EDM for Nonmagnetic Materials

Previous magnetic-eld-assisted microelectrical discharge machining -EDM techniques have been limited to use with magnetic materials. Therefore, a novel process has been developed and tested to improve material removal rate in magnetic-eld-assisted -EDM for nonmagnetic materials. The workpiece electrodes were oriented to promote directionality in the current owing through the workpiece, while an external magnetic eld was applied in such a way as to produce a Lorentz force in the melt pool. Singledischarge events were carried out on nonmagnetic Grade 5 titanium workpieces to investigate the mechanical effects of the Lorentz force on material removal. Erosion efciency, melt pool volume analysis, plasma temperature, electron density, and debris eld characterization were used as the response metrics to quantify and explain the change in material removal with the applied Lorentz force. By orienting the Lorentz force to act in a direction pointing into the workpiece surface, volume of material removed was shown to increase by up to nearly 50%. Furthermore, erosion efciency is observed to increase by over 54%. Plasma temperature is unaffected and electron density shows a slight decrease with the addition of the Lorentz force. The distribution of debris around the crater is shifted to greater distances from the discharge center with the Lorentz force. Taken together, these facts strongly suggest that the Lorentz force process developed produces a mechanical effect on the melt pool to aid in increasing material removal. The application of the Lorentz force is not found to negatively impact tool wear. DOI: 10.1115/1.4003488

Richard E. DeVor
College of Engineering Distinguished Professor of Manufacturing Department of Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801

Vijay Surla
Postdoctoral Research Associate Department of Nuclear, Plasma, and Radiological Engineering, Center for Plasma-Material Interactions, University of Illinois at Urbana-Champaign, Urbana, IL 61801

Introduction

Microelectrical discharge machining -EDM is a unique machining method capable of creating micron-sized features irrespective of material hardness 1. This process is valuable in the manufacturing of miniaturized products where industry demand for increasingly hard materials has reached the limitations of conventional micromachining techniques. Some examples include tool steel, tungsten carbide, and titanium used in the manufacture of microscale turning and milling equipment, micromold and diemaking, and diesel fuel injector fabrication. However, the current material removal rates MRRs for -EDM range from 0.6 mm3 / h to 6.0 mm3 / h 1, which is far below the desired minimum level of 10 15 mm3 / h required for industrial viability. Magnetic elds have been successfully implemented at both the macro- and microscales for improving the MRR of magnetic workpiece materials. The application of magnetic elds for gap cleaning in EDM was rst suggested by De Bruijn et al. 2. They concluded that for magnetic workpiece materials, the application of magnetic elds increased debris transport out of the gap. Recently, Lin and Lee 3,4 linked enhanced debris removal from the application of magnetic elds to an increase in MRR. They observed that for a magnetic material, the MRR increased nearly three times that of a hole cut without the magnetic eld. The problem of debris buildup in the interelectrode gap is compounded in -EDM as gap voltages are reduced to decrease discharge energy, resulting in smaller interelectrode gaps that further hinder debris ushing. Thus the benets of improved debris reContributed by the Manufacturing Engineering Division of ASME for publication in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received May 3, 2010; nal manuscript received November 13, 2010; published online March 8, 2011. Assoc. Editor: Bin Wei.

moval are of even greater interest at the microscale. Yeo et al. 5 reported successfully extending enhanced MRR via improved debris ushing through the method used by De Bruijn et al. 2 to -EDM by observing that the application of a magnetic eld near the discharge location during a 360 min plunge cut of a 275 m hole in a magnetic material yielded a 26% deeper hole compared with the one machined without the magnetic eld. The above work examined the effects of magnetic elds on magnetic workpiece materials only, and the proposed mechanics behind the increases in MRR reported are dependent on the workpiece material being magnetic. Further, the above work examined the effects of magnetic elds over multiple discharges, which makes it difcult to explain the mechanics of the material removal process. The objective of this study is to investigate the mechanical effects of magnetic-eld-assisted -EDM for nonmagnetic workpiece materials with an eye toward enhancing material removal rates. A novel process is developed whereby the workpiece electrodes are oriented to promote directionality in the current owing through the workpiece, while an external magnetic eld is applied in such a way as to produce a Lorentz force that would affect melt pool behavior in nonmagnetic materials. The -EDM process typically involves a continuous pulse train of discharges; however, it is difcult to study the fundamental behavior of the discharge plasma and the melt pool behavior during a discharge under these conditions. Therefore single-discharge experiments have been chosen to enable a study on the fundamentals of melt pool behavior. Volume analysis and erosion efciency are utilized to quantify the material removal at the singledischarge level. Plasma temperature, electron density, and debris distribution are used to investigate and explain the mechanisms inuencing material removal. APRIL 2011, Vol. 133 / 021002-1

Journal of Manufacturing Science and Engineering Copyright 2011 by ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Fig. 1 Schematic of parallel magnetic eld with directional current

The rest of the paper is organized as follows. Section 2 covers the experimental methodology. Section 3 discusses the effects Lorentz forces have on the material removal process. Section 4 discusses tool wear for magnetic-eld-assisted -EDM. Finally, Sec. 5 draws specic conclusions about material removal in magnetic-eld-assisted -EDM for nonmagnetic materials.

Experimental Methodology

2.1 Purpose and Goals. One of the mechanical components in the material removal mechanism for the EDM process is the plasma pressure, which acts on the melt pool to aid in ejecting melt material from the discharge crater 6. Increasing this force on the melt pool through the use of magnetic force provides a possibility for enhanced material ejection. Nonmagnetic materials inherently do not experience any force in the presence of a magnetic eld. However, if additionally a directional current is owing through the nonmagnetic material Fig. 1a, a Lorentz force is produced if the current and magnetic-eld vectors are nonparallel. The Lorentz force is given by the cross product of the current with the magnetic eld F=JB
2

where J is the current per unit area A / mm and B is the magnetic eld T. When the two components are perpendicular, the Lorentz force is maximized in a direction that is mutually perpendicular to both J and B components. During a typical -EDM discharge, current ows normal to the workpiece surface in the plasma channel. By applying a parallel magnetic eld, dened as a magnetic eld parallel to the workpiece surface, a Lorentz force F2 resulting from the current in the 021002-2 / Vol. 133, APRIL 2011

plasma channel J2 can be produced in the plasma channel Fig. 1b. Once the current enters the workpiece, it disperses isotropically through the workpiece Fig. 1c, eliminating the possibility of producing a Lorentz force in the melt pool. However, if the current is given a preferential direction in the workpiece by providing a low resistance path Fig. 1c, an additional Lorentz force is developed, force F1 from current direction J1 in Fig. 1b, that acts on the melt pool in the same way a Lorentz force acts on a current-carrying wire in a magnetic eld. The actual direction of the current J1 in the melt pool will be a combination of the overall current direction in the workpiece J1 and the current direction in the plasma channel J2; thus the resulting force vectors F1 and F2 will combine to produce a force in the melt pool FR, as seen in Fig. 1b. For clarication, J1 and J2 are two designations for the same discharge current, used to differentiate between current in the plasma channel and current in the workpiece. By controlling the directionality of the magnetic eld, it is possible to control the direction of the Lorentz force. Figure 1b shows the conguration where the additional Lorentz force F1 is pointing into the workpiece surface; however, the magnetic-eld direction can be reversed 180 deg to direct the Lorentz force out of the workpiece. The time required for the Lorentz force to develop is limited by the timing of the current pulse during a discharge. Figure 2 shows a time resolved plot of a current pulse characteristic of those used in this paper. The pulse is initiated just before the 1.2 s mark on the x-axis, followed by approximately 200 ns of delay where current spikes can be seen that are indicative of streamer formation prior to dielectric breakdown 7. Once breakdown occurs, current begins to ow steadily between 6 A and 8 A for the remainder of Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Fig. 2 Single-shot spark current pulse 1 m spark gap, 100 V open gap voltage, and 100 W discharge power

Fig. 4

FEM simulation results at maximum current

the discharge. The timing and magnitude of the Lorentz force developed would follow the form of the current pulse based on Eq. 1. 2.2 Testbed Design. Figure 3 shows a solid model of the -EDM testbed. The testbed consists of two main modules; the electrode holder, with manual X-Y-Z control, and the workpiece holder, with a piezoelectric Z-axis capable of 10 nm positioning. The testbed is located on a precision granite surface plate with vibration isolation to prevent oscillation of the electrode during testing. 2.2.1 Electromagnet Design. For initial theory validation, the magnetic eld was produced by a permanent magnet adjacent to the discharge location and oriented as seen in Figs. 1a and 1b. In this conguration, the magnetic-eld lines in the discharge gap can be assumed to be uniform and perpendicular to the magnet face due to the close proximity of the discharge location to the center of the magnet face. To increase control over magnetic-eld line strength, direction, and uniformity for experimental investigations, the magnetic-eld source was switched to an electromagnet after successful initial testing. The electromagnet used in this research was designed with the aid of nite element model FEM simulations. The magnet consists of four 1006 annealed steel cores held by a 1006 annealed steel frame. Each core is wrapped in 900 turns of 24 American Wire Gauge AWG copper magnet wire and is independently controlled to allow the polarity on the poles to be switched for altering the eld direction. Figure 4 shows the FEM simulation results for magnetic-eld direction and density. The simulation indicates that the magneticeld lines are nearly perpendicular to the workpiece surface in this conguration and that the maximum eld strength is approximately 1 T at the discharge location under maximum magnetic coil current conditions. The orthogonality of the eld lines at the

workpiece surface is important for maximizing the cross-product seen in Eq. 1 to maximize the Lorentz force produced. 2.2.2 Single-Shot Spark Generator Design. To facilitate the study of the -EDM process at the single-spark discharge level, a hybrid RC-transistor single-shot spark generator was designed and fabricated. This design was chosen to take advantage of the small pulse energies offered by the RC circuit 8 and the pulse control offered by transistor-based circuits 9. The design consists of high-power metal oxide semiconductor eld effect transistor MOSFET 10 controlled gates timed by an adjustable monostable multivibrator to control discharge time and energy. 2.3 Experimental Design. For each experiment, trials were conducted by varying magnetic-eld strength or magnetic-eld direction. For a given trial type, several repetitions were done to obtain an average for the measured parameters. Due to the stochastic nature of the -EDM process, some repetitions in each trial produced unusable data due to short-circuiting or timing errors on the data collection equipment. Test conditions are shown in Table 1. The workpieces were prepared by polishing and cleaning. The polarity of the tool electrode was positive anode and the workpiece was negative cathode. The gap used in these experiments was set to 1 m based on the breakdown strength of de-ionized water, typically 65 70 V / m. Voltage and current data for the discharge were collected on a 2 gigasample/s Tektronix TDS2024B oscilloscope. After a discharge, a stepover of approximately 500 m was used to separate subsequent discharge locations. Dielectric decomposition during a discharge can alter the surface chemistry of the workpiece 11, thus the stepover was required to ensure consistent workpiece surface properties with each discharge. The electrode was rezeroed at each individual discharge location to ensure accurate gap setting. 2.4 Response Metrics

2.4.1 Erosion Efciency. Erosion efciency can be used to quantify the percentage of discharge energy going into material removal. The greater the erosion efciency, the higher the volume of material removed per unit energy expended during a discharge. Erosion efciency, originally dened by Bor-Jenq et al. 12, is the ratio of erosion energy to spark energy
Table 1 Experimental conditions Electrode anode Workpiece cathode Gap distance Open gap voltage Discharge power Dielectric Permanent magnet strength Electromagnet strength 100 m diameter W wire Grade 5 titanium 6% Al, 90% Ti, 4% V 1 m 100 V 100150 W De-ionized water 0.7 T surface eld strength 0.33 T, 0.66 T, or 1.0 T

Fig. 3

-EDM testbed

Journal of Manufacturing Science and Engineering

APRIL 2011, Vol. 133 / 021002-3

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Table 2 Titanium workpiece properties Density Melting point Boiling point Enthalpy of melting 4.43 g / cm3 1933 K 3560 K 1.156 103 kJ / kg Fig. 7 Side view of 3D surface map

E v H = e = e m Es Es where Es is the spark energy and Ee is the erosion energy. Spark energy was calculated from Es = VIT

2 terized by examining the area surrounding the discharge crater to allow for differentiation in the 3D laser data between depth data from the discharge crater and depth data from the surface roughness. To calculate volume, the following algorithm was used: 1. digitally correct any slope to the workpiece surface and move the average surface height to z = 0 m 2. determine the maximum depth and height of the surface roughness on the workpiece surface around the discharge crater to minimize inclusion of surface roughness in volume measurements 3. any points in the discharge crater found to be higher than the highest point of the workpiece surface were added to the total negative volume Fig. 7 4. any points in the discharge crater found to be lower than the lowest point on the workpiece surface were added to the total positive volume Fig. 7 2.4.2 Plasma Temperature and Electron Density. In order to determine if the melt pool effects were mechanical or plasmarelated, spectroscopic measurement techniques were utilized to nd plasma temperature and electron density. Plasma temperature was determined by the line-pair method Eq. 4, a widely used plasma temperature estimation technique that compares the relative intensities of two spectral peaks and has been shown to determine plasma temperature with an estimated 20% error 13. T= E 1 E 2 / k I1 1 g2 A2 ln I2 2 g1 A1

where V is voltage across the electrode gap, I is the current through the electrode gap, and T is the duration of the discharge. This calculation was performed by numerical integration over the duration of the discharge since both voltage and current vary with time during the discharge. Erosion energy as dened by Bor-Jenq et al. 12 is a relative energy quantity representative of the amount of material actually removed from the melt pool, assuming melting/splashing as the primary material removal mechanism. This is not the same as the energy required to remove that material, which would take into account ejection velocity, surface tension forces, etc. By the denition provided by Bor-Jenq et al., Ee can be found by the product seen in Eq. 2 where ve is the erosion volume, is the density of the material, and Hm is the enthalpy of melting for the workpiece material. The titanium workpiece properties used to calculate Ee are listed in Table 2. To obtain volume information for the discharge craters, 3D laser scanning of the workpiece surface was employed. A Keyence LT-9010M surface scanning confocal laser with 0.01 m distance resolution and a 2 m spot size was used to raster scan over the surface of the discharge craters Fig. 5. The data points collected from these scans were used to digitally recreate the surface of the crater. Figure 6 shows a sample of the digitized 3D surface topography. After initial examination of the discharge craters, two unique volumes were identied, a negative volume, dened as the material pushed above the workpiece surface around the rim of the discharge crater, and positive volume, dened as the material removed below the surface of the workpiece Fig. 7. The erosion volume actual amount of material removed was found by the difference in these two volumes. Surface roughness was not insignicant at the microscale, so the workpiece surface was charac-

In Eq. 4, En is the excitation energy of the spectral line n, k is the Boltzmann constant, In is the intensity of the spectral line n, n is the wavelength of line n, gn is the statistical weight of line n, and An is the transition probability of line n. Values for all these parameters are shown in Table 3 14. Electron density was found by measuring the H 656.28 nm spectral line broadening, another common spectroscopic technique for plasma analysis 7,15. The full-width half-maximum FWHM of the H line can be directly correlated with electron density using 7 ne = 8.8308 1016 w1.6005
3

where ne is the electron density in cm and w is the width of the H line measured at FWHM in nm. The spectroscopic equipment used to collect these data consists of two f/2 lenses to focus the discharge light onto a 0.1
Fig. 5 Digitizing the discharge crater surface Parameter n En k In gn An

Table 3 Equation 4 values Line 1 498.173 26,910.712 0.695035613 Measured 13 0.66 Line 2 521.04 19,573.973 0.695035613 Measured 9 0.0411 Units nm cm1 cm1 / K 108 s1

Fig. 6 3D surface topography from laser scan

021002-4 / Vol. 133, APRIL 2011

Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Table 4 Summary of initial experiments Average positive volume m 3 No eld Lorentz in Lorentz out 3240 4230 2510 Average negative volume m 3 910 1010 520 Average erosion volume m 3 2330 3220 1990 Average erosion efciency % 4.5 5.6 4.5

Fig. 8

PEAKFIT

plots

5 mm2 slit in the spectroscope, where the spectrum is then captured with a 752 580 pixel camera yielding a resolution of approximately 0.5 nm/pixel. Three sets of spectra were collected during each experiment. First, a calibration spectrum was recorded by using a mercury lamp with distinct spectral lines at known wavelengths. Second, a dark image no spark was taken to allow for background subtraction from the signal images. Last, the trial discharge spectrum was collected. After background subtraction, the trial spectrum exhibits both a signicant amount of continuum radiation as well as convolution of many of the spectral peaks due to peak broadening brought on by instrument broadening as well as natural, Stark, and Doppler broadening. To remove the continuum radiation, several local minima were used to nd a baseline, which was then removed from all the data around those minima. To correct the convolution issue, automated peak separation analysis software 16 was used to both deconvolute and t Voigt peaks to a range of the spectral data around the spectral lines of interest, as seen in Fig. 8. The upper plot shows the original data points with a line overlaid that is produced by the convolution product of the individual deconvoluted peaks shown on the lower plot. Voigt line shapes were utilized because they combine characteristics of both a Gaussian line shape as well as a Lorentzian line shape. These two line shapes characterize different sources of peak broadening in optical spectroscopy; thus the Voigt line shape provides the most accurate model for experimental spectral data. Once the spectrum was deconvoluted into the appropriate Voigt peaks, the areas under the 498.173 nm and 521.04 nm lines were determined to nd the total intensity of the spectral lines for Eq. 4. The FWHM of the H line was also determined at this time. Before the FWHM of the H can be used in Eq. 5 for determining electron density, line broadening due to instrument effects needs to be removed as it is not accounted for in Eq. 5. This was done by measuring the spectral line broadening of several mercury lamp emission lines that were known to be singular wavelengths and subtracting the average FWHM of these peaks from the H FWHM. 2.4.3 Additional Plasma Characteristics. After determining the plasma temperature and electron density, it is possible to calculate the coupling factor , the mean interparticle distance a, and the Debye length D 7. The coupling factor describes the ratio of the potential energy of Coulomb interactions between particles to the thermal energy of the particles in the plasma. If 1, the plasma is labeled as ideal and indicates that all particles in the plasma are free to move and electrostatic interactions between particles are minimal. Ideal plasmas are typically very hot and have a low density. For a nonideal plasma 1, as well as a strongly coupled plasma 1, the interparticle distances are very small, which leads to high electrostatic interactions between plasma particles. Journal of Manufacturing Science and Engineering

The Debye length D describes the distance over which charge shielding occurs and gives insight into the cause of plasma ideality. The Debye length is important relative to the interparticle distance a in the sense that when D a an ideal plasma, many particles are present within the distance required for charge shielding. Only a few particles are actually needed for charge shielding, and as a result, the remaining particles are free to move around. However, if D a a nonideal plasma, few particles are present within the Debye length, resulting in a majority of particles in the plasma tied up in electrostatic interactions with one another. Because electrostatic forces are present between nearly all particles in nonideal plasmas, magnetic elds may have less of an effect on nonideal plasmas as the electrostatic forces likely dominate the magnetic forces. 2.4.4 Debris Field Characterization. The nal technique utilized to investigate the effects of the Lorentz force process was a characterization of the debris eld around each discharge crater. By quantifying the characteristics of the debris eld, insight can be had into any effects the Lorentz force may be having on material ejection from the melt pool and discharge gap. Scanning electron microscopy SEM images were taken of each discharge area, and the distance of every debris particle 2 m or larger contained inside a 640 m diameter zone, centered on the discharge crater, was measured from the discharge crater center. The 2 m particle size cutoff was used because debris smaller than that was difcult to distinguish from other surface features on the workpiece. The debris eld zone diameter was chosen as the maximum diameter allowed between discharges without creating overlapping discharge zones.

Mechanics of Material Removal

3.1 Preliminary Testing Using Permanent Magnets. Prior to running a full set of tests, a preliminary set of experiments was conducted to gauge the effectiveness of the Lorentz force technique. These experiments utilized a 0.7 T surface eld strength Grade 53 NdFeB sintered permanent magnet supplied from K&J Magnetics, Jamison, PA. The erosion efciency and crater volume were analyzed for differences between the Lorentz force process and the standard -EDM process without any magnetic elds or directional workpiece currents. As described in Sec. 2.1, the Lorentz force F1 Fig. 1b can be oriented into or out from the workpiece surface. Both cases were investigated for their effects of melt pool morphology. Table 4 shows the results from these experiments. With the Lorentz force pointing into the melt pool, there is an increase in positive volume of 30.5% and an increase in negative volume of 11% when compared with the no-eld case, resulting in an increase in erosion volume of 38.2%. Average erosion efciency for the no-eld case was 4.5%, while for the Lorentz force into the workpiece surface case it was 5.6%, a 25% increase in efciency over the no-eld case. The average negative volume decreases by 43% and the average positive volume decreases by 23% for the Lorentz force pointing out from the workpiece case compared with the no-eld case. The result is a 15% decrease in erosion volume and the erosion efciency appears unaffected. APRIL 2011, Vol. 133 / 021002-5

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Table 5 Summary of Lorentz force into the workpiece experiments Field strength T 0.33 Positive volume m 3 x: 2680 s: 1000; n: 4 x: 3430 s: 1130; n: 5 x: 3380 s: 1290; n: 4 Negative volume m 3 x: 510 s: 310; n: 4 x: 250 s: 125; n: 5 x: 250 s: 210; n: 4 Erosion volume m 3 x: 2140 s: 820; n: 4 x: 3180 s: 1040; n: 5 x: 3130 s: 1330; n: 4 Erosion efciency % x: 3.52 s: 1.36; n: 4 x: 5.42 s: 1.53; n: 5 x: 5.59 s: 2.12; n: 4

Table 7 Plasma characteristics for parallel eld experiments Field strength T 0.33 0.66 1.00 Temperature K x: 4960; s: 870; n: 5 x: 4840; s: 640; n: 5 x: 5070; s: 300; n: 5 Electron density 1 1017 cm3 x: 7.3; s: 9.9; n: 5 x: 2.6; s: 2.0; n: 5 x: 3.1; s: 3.1; n: 5

0.66

1.00

oriented to point out from the workpiece surface. Positive volume trends show decreases of 20% and 26% with each step increase in eld strength and the negative volume decreases by 26% and 9% as well. This indicates an overall decrease in the transport of melt pool material from the crater to the rim during a discharge. 3.2.2 Plasma Characteristics. The increases in positive volumes noted with the Lorentz force pointing into the workpiece surface could be the result of either mechanical effects such as increased debris ejection and/or thermal effects such as an increase in plasma temperature. The discharge plasma will be investigated rst for changes resulting from the Lorentz force pointing into the workpiece process. 3.2.2.1 Effects of Lorentz force pointing into workpiece on plasma characteristics. Table 7 shows the temperature and electron density data for the Lorentz force pointing into the workpiece experiments. No signicant difference is seen in the plasma temperature for the three eld strengths; however, plasma temperatures of 5000 K are signicantly colder than EDM plasmas produced at the macroscale, which are reported to be on average 8000 K 7. The electron density does appear to decrease at higher eld strengths, as the 0.33 T eld strength electron density 7.3 1017 cm3 is somewhat higher than the electron densities at the 0.66 T 2.6 1017 cm3 and 1.00 T 3.1 1017 cm3 eld strengths in Table 7. These electron densities are also lower than those reported by Descoudres 1018 cm3 7. At the lower eld strength, the interparticle distance, Debye length, and coupling factor are a = 7 nm, D = 6 nm, and = 0.434, indicating the plasma is nonideal and there are signicant electrostatic interactions between particles. These plasma characteristics are in agreement with those found by Descoudres for macroscale EDM plasmas within their rst microsecond a = 6 nm, D = 6 nm, and = 0.45 7. At the higher eld strengths, however, the interparticle distance increases to a = 9.4 nm, Debye length increases to D = 9 nm, and the coupling factor is = 0.356. These results show that for the increased eld strength, the plasma experiences greater expansion, as depicted by the larger interparticle distance. However, the Debye length increases as well so the resulting plasma does not become any more ideal. When the electric eld is perpendicular to the magnetic eld, as is the case in these experiments, the electrons develop a path of travel that is more restrictive than the straight path normally seen in an electric eld, as seen in Fig. 9. The particle motion shown is a 2D in-plane motion that occurs as the result of the electron both moving along the electric eld lines as well as orbiting about the magnetic-eld lines 17. The decrease in electron density at higher eld strengths seen in Table 7 is likely the result of increased resistance for electron travel caused by tighter orbits in the nonuniform path of travel shown in Fig. 9. Higher resistance to electron travel causes fewer electrons to be transported across the electrode gap, resulting in a lower electron density. These results suggest that the increased erosion efciency reported for the Lorentz force pointing into the workpiece case is likely not due to increased plasma temperature or electron density. Transactions of the ASME

3.2

Full-Scale Testing Using Electromagnets

3.2.1 Volume Analysis and Erosion Efciency. Based on the results from the initial experiments, the Lorentz force process appears to have a signicant effect on the melt pool behavior, so a full-scale set of experiments utilizing multiple eld strengths generated by an electromagnet was conducted. 3.2.1.1 Effects of Lorentz force pointing into workpiece on material removal. Table 5 compares the volume data and erosion efciencies for varying eld strengths for the Lorentz force pointing into the workpiece surface conguration. When the eld strength is increased from 0.33 T to 0.66 T, there is an increase in erosion efciency of 54%, but there is only an increase of 3% from 0.66 T to 1 T. This indicates that signicantly more materials are removed per unit of spark energy going into each discharge when the magnetic eld is 0.66 T; however, there is little evidence to show that increasing the eld strength any higher than this will result in any signicant increase in MRR. To further understand the cause of this increase in erosion efciency, it can be seen that there is an increase in positive volume of approximately 28% and a decrease in negative volume of 50% between the 0.33 T and 0.66 T eld strengths. This indicates that there are more materials removed from the melt pool below the workpiece surface for the increased eld strength case, and this volume is not transported to the rim, but is instead completely removed from the melt pool, as evidenced by the decrease in negative volume. 3.2.1.2 Effects of Lorentz force pointing out from the workpiece surface on material removal. Table 6 compares the volume and erosion efciency data between eld strengths for the Lorentz force pointing out from the workpiece surface. Table 6 shows a decrease in erosion efciency of 15% when increasing the eld strength from 0.33 T to 0.66 T and a decrease of 21% from 0.66 T to 1 T. This indicates that material removal is hindered with increasing eld strength when the Lorentz force is

Table 6 Summary of Lorentz force pointing out from the workpiece experiments Field strength T 0.33 Positive volume m 3 x: 3100 s: 321; n: 5 x: 2460 s: 500; n: 3 x: 1830 s: 560; n: 4 Negative volume m 3 x: 720 s: 380; n: 5 x: 580 s: 400; n: 3 x: 530 s: 240; n: 4 Erosion volume m 3 x: 2330 s: 242; n: 5 x: 1810 s: 650; n: 3 x: 1330 s: 520; n: 4 Erosion efciency % x: 3.49 s: 0.39; n: 5 x: 2.87 s: 0.71; n: 3 x: 2.26 s: 0.75; n: 4

0.66

1.00

021002-6 / Vol. 133, APRIL 2011

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Fig. 9 Electron path of travel when E-eld and B-eld are perpendicular

3.2.3 Debris Field Characteristics. With the Lorentz force pointing into the workpiece surface, increases in positive volume and erosion efciency occurred. By examining the debris eld characteristics, it is possible to determine if material ejection is modied by the Lorentz force, causing increased erosion efciency. Figure 10 shows the results from 17 discharges done with and without the Lorentz force and with the debris data collected as specied in Sec. 2.4.4. It can be seen that when the Lorentz force is applied into the workpiece, the discharge debris distance shifts toward values further away from the crater. The average debris distance without the Lorentz force is 164 m, whereas with the Lorentz force applied, the average increases to 207 m. More importantly though is the 0 50 m category. The diameter of the electrode is 100 m, so any particles less than 50 m from the discharge center have not been cleared from the discharge gap assuming the discharge occurs at the center of the electrode. In the Lorentz force case, there are no debris particles in the 0 50 m range. This is important in -EDM as debris particles stuck in the gap can cause subsequent abnormal discharges; thus eliminating particles in this category is desirable. 3.3 Mechanics of Material Removal. Figure 11 shows a comparison of SEM images between the characteristic no-eld and Lorentz force into the workpiece discharge craters. The SEM images for the Lorentz force into the workpiece exhibit increased consistency from crater to crater in both shape and uniformity of melt material distribution versus the no-eld craters. Additionally, the depth of the crater is greater in the majority of the Lorentz force into the workpiece cases. The process of material removal is very complex, involving forces originating from electrodynamics, electromagnetics, thermodynamics, and hydrodynamics. As a result, comprehensive theories on the removal of melt material have not been forthcoming 18. One of the rst, but still widely accepted, theories was that of melting and splashing 6. Based on the SEM images in Fig. 11, volume data in Sec. 3.1, plasma data in Sec. 3.2, and debris data in Sec. 3.3, melting, and splashing are suspected as the material removal mechanism in -EDM, as is also suggested by

Fig. 11 SEM and laser scan images of Lorentz force into workpiece surface and no-eld experiments

Fig. 10 Debris eld distribution

Bor-Jenq et al. 12. In the melting/splashing model, the plasma channel causes workpiece material to melt, starting at the center of the plasma channel, and growing both axially into the workpiece surface and radially along the workpiece surface. The melted material ows radially to the edge of the discharge crater due to the Marangoni effect 19, creating a rim of material around the crater 12. Some of the material built up on the rim is removed via splashing due to recoil forces produced upon plasma channel collapse 6. The erosion efciencies seen in Tables 5 and 6 are based on the melting/ splashing assumption and are in agreement with those found by Bor-Jenq et al. 12, who reported an erosion efciency of 3.31% for titanium assuming melting/splashing. Calculations run for erosion efciency using vaporization as the primary material removal mechanism by substituting Hv, the enthalpy of vaporization 1.089 104 kJ / kg, for Hm in Eq. 2 yield efciencies well over 50%, far too large to be reasonable. Thus, the discharges examined in this study can be assumed to follow the model of melting/ splashing as the primary material removal mechanism. Based on the observation of both increased positive volume and increased debris distance for the Lorentz force pointing into the workpiece case, additional melt material is likely pushed away from the bottom of the melt pool as shown by the schematic in APRIL 2011, Vol. 133 / 021002-7

Journal of Manufacturing Science and Engineering

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

Table 8 EDS data for workpiece and melt pools %Al Workpiece Parallel eld only Parallel eld with directional current X: 4.9 X: 4.1 X: 4.4 %Ti X: 91.0 X: 90.7 X: 90.1 %V X: 3.9 X: 4.7 X: 4.9 %W X: 0.3 X: 0.5 X: 0.5

12, who found that there is a high correlation between melting temperature differences and erosion volume in single-spark discharge events.
Fig. 12 Melting/splashing material removal mechanism

Conclusions
The following conclusions can be drawn from this work. 1. A novel process has been developed to enhance the effects of magnetic-eld-assisted -EDM with nonmagnetic materials. In contrast to current magnetic-eld-assisted -EDM techniques that are only effective when machining magnetic workpiece materials, the process developed in this research utilizes directional workpiece current to produce an additional Lorentz force normal to the workpiece surface to affect melt pool behavior for enhanced material removal regardless of the magnetic properties of the workpiece material. 2. For the Lorentz forces pointing into the workpiece surface, positive crater volume increased by 28%, while negative rim volume decreased by 50% between the 0.33 T and 0.66 T eld strengths; thus much of the additional volume removed below the workpiece surface is removed as debris. This is also seen in a 54% increase in erosion efciency from 3.5% with a 0.33 T eld applied to 5.4% with a 0.66 T eld. 3. For the Lorentz force pointing outward from the workpiece surface, the ow of material is impeded, causing both the positive and negative volumes to decrease with increasing eld strength. 4. Magnetic elds at or below 1 T oriented parallel to the workpiece surface as in the Lorentz process do not affect plasma temperature T = 5000 K, but they tend to decrease electron density between eld strengths of 0.33 T ne = 7.3 1017 cm3 and 0.66 T ne = 2.6 1017 cm3 due to an increase in transport resistance for the electrons in the plasma channel. 5. The debris distance distribution shifts to values further from the discharge crater with the application of the Lorentz force into the workpiece surface. 6. Taken together, the volume, erosion efciency, plasma temperature, electron density, and debris eld data strongly suggest that the Lorentz force pointing into the workpiece surface causes a mechanical effect in the melt pool to enhance material removal. 7. The results from the Lorentz force process have the potential to be extended to other EDM platforms beyond -EDM without signicant alteration. 8. Equal trace amounts 1% of the electrode material were found in the parallel eld and parallel eld with directional current melt pools, indicating that magnetic elds and directional currents do not have a negative impact on electrode wear.

Fig. 12a as a result of the added Lorentz force. This is conceptually similar to the process of laser hole drilling as proposed by von Allmen and Blatter 20, who described the plasma pressure above the melt pool as a piston that exerts force downward on the melt pool. The added Lorentz force serves to augment the plasma pressure in -EDM. The plasma pressure above the melt pool is reported to be on the order of several bars several hundred kN / m2 by Descoeudres 7. An estimate of the additional force added from the Lorentz force can be calculated from Eq. 1, the current seen in Fig. 2 8 A, and the magnetic eld reported in Table 1 0.7 T. Based on a melt pool diameter of 50 m and a cathode spot diameter of 15 m 18, the Lorentz force in these experiments would exert approximately 100 N 50 kN / m2 of force on the melt pool, which is in the same order of magnitude as the existing plasma pressure. By adding the Lorentz force, the melt pool experiences an increased pressure above it. This clears away additional melted material from the bottom of the discharge crater, creating a deeper crater. Melted material is transported up the sides of the crater to the crater rim, where it is removed by splashing after the plasma channel collapses, as suggested in Fig. 12b. Furthermore, the effects of the Lorentz force are potentially scalable to other EDM platforms.

Tool Wear

Yeo et al. 5 reported signicantly increased electrode wear with the application of magnetic elds during a 360 min -EDM drilling operation of a 275 m diameter hole in a magnetic material. In the experiments conducted by Yeo et al., a hardened tool steel XW 42 with a melting point of 1700 K was used as the workpiece and a tungsten rod with a melting point of 3640 K was used as the electrode. To measure the effects of tool wear in the single-discharge Lorentz force-assisted experiments investigated in this paper, the resolidied melt pool was analyzed using energy-dispersive X-ray spectroscopy EDS to determine its composition and look for signs of electrode cross-contamination in the melt pool. The titanium workpiece has a composition of approximately 6% Al, 90% Ti, and 4% V. The electrode is composed of commercially pure W. A baseline was established using an area of the workpiece well away from the discharge crater and debris eld and is labeled as Workpiece in Table 8. Table 8 lists the averages for the data collected. The melt pool composition is nearly identical to the composition of the workpiece material in both the parallel eld only and the parallel eld with directional current cases. Only trace amounts of tungsten are seen, indicating that negligible electrode material wear occurs during single-spark discharges, irrespective of applied magnetic elds and directional currents. The lack of electrode wear found here is in agreement with Bor-Jenq et al. 021002-8 / Vol. 133, APRIL 2011

Acknowledgment
The authors are grateful to the Grayce Wicall Gauthier Chair for funding this research. The authors would like to thank Professor David Ruzic and Professor Nick Glumac of the University of Illinois for their contributions to this project. This research was Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms

carried out in part in the Frederick Seitz Materials Research Laboratory Central Facilities, University of Illinois, which are partially supported by the U.S. Department of Energy under Grant Nos. DE-FG02-07ER46453 and DE-FG02-07ER46471.

References
1 Uruarte, L., Herrero, A., Ivanov, A., Oosterling, H., Staemmler, L., Tang, P. T., and Allen, D., 2006, Comparison Between Micro Fabrication Technologies for Metal Tooling, Proc. Inst. Mech. Eng., Part C: J. Mech. Eng. Sci., 220, pp. 16651676. 2 De Bruijn, H. E., Delft, T. H., and Pekelharing, A. J., 1978, Effect of a Magnetic Field on the Gap Cleaning in EDM, CIRP Ann., 27, pp. 9395. 3 Lin, Y. C., and Lee, H. S., 2008, Machining Characteristics of Magnetic Force-Assisted EDM, Int. J. Mach. Tools Manuf., 48, pp. 11791186. 4 Lin, Y. C., and Lee, H. S., 2009, Optimization of Machining Parameters Using Magnetic-Force-Assisted EDM Based on Gray Relational Analysis, Int. J. Adv. Manuf. Technol., 42, pp. 10521064. 5 Yeo, S. H., Murali, M., and Cheah, H. T., 2004, Magnetic Field Assisted Micro Electro-Discharge Machining, J. Micromech. Microeng., 14, pp. 15261529. 6 Gray, E. W., and Pharney, J. R., 1974, Electrode Erosion by Particle Ejection in Low-Current Arcs, J. Appl. Phys., 452, pp. 667671. 7 Descoeudres, A., 2006, Characterization of EDM Plasmas, Ph.D. thesis, Ecole Polytechnique Federale De Lausanne, Lausanne, Switzerland. 8 Jahan, M. P., Wong, Y. S., and Rahman, M., 2008, A Comparative Study of Transistor and RC Pulse Generators for Micro-EDM of Tungsten Carbide, Int. J. Precision Eng. and Manuf., 94, pp. 310.

9 Han, F., Chen, L., Yu, D., and Zhou, X., 2007, Basic Study on Pulse Generator for Micro-EDM, Int. J. Adv. Manuf. Technol., 33, pp. 474479. 10 Bose, B. K., 1992, Evaluation of Modern Power Semiconductor Devices and Future Trends of Converters, IEEE Trans. Ind. Appl., 282, pp. 403413. 11 Kibria, G., Sarkar, B. R., Pradhan, B. B., and Bhattacharyya, B., 2010, Comparative Study of Different Dielectrics for Micro-EDM Performance During Microhole Machining of Ti-6Al-4V Alloy, Int. J. Adv. Manuf. Technol., 485-8, pp. 557570. 12 Wang, B. J., Saka, N., and Rabinowicz, E., 1992, Static-Gap, Single-Spark Erosion of Ag-CdO and Pure Metal Electrodes, Wear, 157, pp. 3149. 13 Albinski, K., Musiol, K., Miernikiewicz, A., Labuz, S., and Malota, M., 1996, The Temperature of a Plasma Used in Electrical Discharge Machining, Plasma Sources Sci. Technol., 5, pp. 736742. 14 NIST, atomic spectra database, Version 3, URL: physics.nist.gov/ PhysRefData/ASD/index.html. 15 Gigosos, M. A., and Cardenoso, V., 1996, New Plasma Diagnosis Tables of Hydrogen Stark Broadening Including Ion Dynamics, J. Phys. B, 2920, pp. 47954838. 16 2009, PEAKFIT software, URL: http://www.sigmaplot.com. 17 Chen, F. F., 1984, Introduction to Plasma Physics and Controlled Fusion, Vol. 1: Plasma Physics, 2nd Ed., Plenum, New York, Chap. 2, pp. 1921. 18 Yeo, S. H., Kurnia, W., and Tan, P. C., 2007, Electro-Thermal Modeling of Anode and Cathode in Micro-EDM, J. Phys. D: Appl. Phys., 40, pp. 2513 2521. 19 Witz, G., Revaz, B., and Flukiger, R., 2005, Heat Transfer and Marangoni Effect in the Electron Discharge Machining EDM Process, Proceedings of the COMSOL Multiphysics Users Conference, Paris. 20 von Allmen, M., and Blatter, A., 1995, Laser-Beam Interactions With Materials, Springer, New York, Chap. 5, pp. 131133.

Journal of Manufacturing Science and Engineering

APRIL 2011, Vol. 133 / 021002-9

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 06/24/2013 Terms of Use: http://asme.org/terms