Advancing EDM through Fundamental Insight into the Process

1 2 3 4
M. Kunieda (2), B. Lauwers (2), K. P. Rajurkar (1), B. M. Schumacher (1)
1
Tokyo University of Agriculture and Technology, Japan
2
Katholieke Universiteit Leuven, Belgium
3
University of Nebraska-Lincoln, USA
4
University of Applied Science St Gallen, Switzerland

Abstract
This paper aims to show the prospects of electrical discharge machining (EDM) technology by interrelating
recent achievements in fundamental studies on EDM with newly developed advanced application technologies.
Although gap phenomena in EDM are very complicated and hence not yet very well understood, recent
improvements in computers and electronic measuring instruments are contributing to new discoveries and
inventions in EDM technology. Such newly acquired insight sometimes raises questions on the validity of the
established theories of EDM phenomena, and EDM processes once believed to be impossible or unrealistic are
now becoming practical.

Keywords:
Electrical discharge machining (EDM), Wire electrical discharge machining (WEDM)

1 INTRODUCTION theories and commonly accepted EDM phenomena.

During the 1930s, attempts were made for the first time to
machine metals and diamonds with electrical discharge.
V.E. Matulaitis and H.V. Harding of Elox US developed
“Disintegrators” to remove broken taps from valuable
workpiece materials such as cemented carbide and high
speed steel. Erosion was caused by intermittent arc
discharges occurring in air between the tool electrode and
workpiece connected to a DC power supply. Arc discharges
were initiated through short mechanical contact like welding
arcs and were interrupted by retraction using vibration of
the tool electrode. AEG developed equipment capable of
eroding diamond using the heat generated by discharges
occurring at high frequencies in the gap between a pair of
tool electrodes.
These processes are not very precise due to overheating of
the machining area and are defined as “Arc-Machining”,
while “Spark Machining”, as discussed in the following,
works precisely because the workpiece is protected from
excessive heat attacks. This distinction is apparent from
the shop floor application of spark machining. However, it is
difficult to distinguish between arc machining and spark
machining from the physical point of view.
During World War II physicists B.R. and N.I. Lazarenko in
Moscow conducted studies on the minimisation of wear on
electric power contacts. They tested different materials with
discharges of defined energy, generated by a capacitor.
B.R. Lazarenko published the paper “To invert the effect of
wear on electric power contacts”, in 1943 [1]. This idea
started the development of Electrical Discharge Machining
(EDM), using controlled discharge conditions, for achieving
precision machining [2].
Since then, EDM technology has developed rapidly and
become indispensable in manufacturing applications such
as die and mold making, micro-machining, prototyping, etc.
However, EDM gap phenomena are very complicated and
hence not yet fully understood. Understanding arc
discharges even in the steady state continues to be a
difficult problem. Electrical discharge phenomena in EDM
occur over a very short time period, in a very narrow space
filled with liquid, and involve evaporation and melting of the
electrodes, thus making both observation and theoretical
analysis extremely difficult.
On the other hand, recent developments in computer
technology and measuring and analyzing instruments are
accelerating new findings and knowledge. This trend
sometimes leads to results conflicting with established
Dramatic improvements in machining characteristics are
being achieved, and EDM processes believed to be
impossible or unrealistic before are now becoming practical.
Furthermore, applications of EDM phenomena other than
material removal are also being developed. This paper
aims to demonstrate the future prospects of EDM
technology by linking recent achievements in fundamental
studies with newly developed advanced application
technologies.
2 OVERVIEW OF EDM
2.1 Principle
Figure 1 shows the concept of EDM. Pulsed arc discharges
occur in the “gap” filled with an insulating medium,
preferably a dielectric liquid like hydrocarbon oil or
de-ionized (de-mineralized) water between tool electrode
and workpiece. The insulating effect of the dielectric
medium has some importance in avoiding electrolysis
effects on the electrodes during an EDM process. As the
electrode shape is copied with an offset equal to the
gap-size, the liquid should be selected to minimize the gap
(10-100µm) to obtain precise machining. On the other hand
a certain gap width is needed to avoid short circuiting,
especially when electrodes that are sensitive to vibration
(like wire-electrodes) or deformation are used. The ignition
of the discharge is initiated by a high voltage, overcoming
the dielectric breakdown strength of the small gap. A
channel of plasma (ionized, electrically conductive gas with
high temperature) is formed between the electrodes and
develops further with discharge duration. As the metal
removal per discharge is very small, discharges should
occur at high frequencies (103 -106 Hz). For every pulse,
discharge occurs at a single location where the electrode
Feed
Dielectric liquid Bubbles
Tool electrode
Workpiece
Gap width: several
Discharge current tens of micrometer
Discharge column
Figure 1： Concept of EDM.
materials are evaporated and/or ejected in the molten
phase. As a result, a small crater is generated both on the
tool electrode and workpiece surfaces. Removed materials
are cooled and resolidified in the dielectric liquid forming
several hundreds of spherical debris particles, which are
then flushed away from the gap by the dielectric flow.
After the end of the discharge duration, the temperature of
the plasma and the electrode surfaces contacting the
plasma rapidly drops, resulting in a recombination of ions
and electrons and a recovery of the dielectric breakdown
strength. To obtain stable conditions in EDM, it is essential
for the next pulse discharge to occur at a spot distanced
sufficiently far from the previous discharge location. Such a
spot may be the place where the gap is small or
contaminated with debris particles which may weaken the
dielectric breakdown strength of the liquid. Accordingly, the
interval time between pulse discharges must be sufficiently
long so that the plasma generated by the previous
discharge can be deionized and the dielectric breakdown
strength around the previous discharge location can be
recovered by the time the next pulse voltage is applied.
Otherwise discharges occur at the same location for every
pulse, resulting in thermal overheating and a non uniform
erosion of the workpiece.
2.2 Sinking EDM and Wire EDM
Figure 2 shows the configuration of a sinking electrical
discharge machine. The workpiece can be formed, either
by replication of a shaped tool electrode, or by 3D
movement of a simple electrode like in milling or a
combination of the above. The electrode material is
normally copper or graphite. The numerical control
monitors the gap conditions (voltage and current) and
synchronously controls the different axes and the pulse
generator. The dielectric liquid is filtrated to remove debris
particles and decomposition products.
Figure 3 outlines the wire electrical discharge machining
(WEDM) method. Complicated shapes can be cut using a
wire electrode of 0.02 to 0.33mm in diameter like a wire
sawing machine. The wire electrode is usually a plain brass
wire or coated wires, such as zinc coated brass or coated
steel wires. Sometimes tungsten or molybdenum wires are
used in case of thin wires. Since wire orientation can be
changed by controlling the horizontal position of the upper
wire guide relative to the lower guide, all types of ruled
surfaces can be cut. Since discharge currents with a high
Gap condition
analyzer
Z-axis
NC Tank Filter
Pulse
generator
XY table
Pumping system
Figure 2: Sinking electrical discharge machine.
Wire electrode
Upper wire Upper feeding brush
guide
Workpiece (+)
Lower feeding brush
Lower wire guide
Figure 3: Wire electrical discharge machining (WEDM).
peak value over a short duration are used, current is
supplied through both the upper and lower feeding brushes
to obtain a quick rise in the discharge current by reducing
the inductance and to avoid wire breakage due to Joule
heating. Tension is applied to the wire to reduce vibration
and deflection, which deteriorates cutting accuracy.
In WEDM, water is most often used as the dielectric liquid,
but its specific electrical conductivity should be decreased
using de-ionizing resins to avoid electrolysis and to keep
high open voltage. Use of de-ionized water is much
preferable to hydrocarbon dielectrics considering fire
hazards. In sinking EDM, however, hydrocarbon dielectrics
are normally used because surface roughness is better and
tool electrode wear is lower compared to de-ionized water.
2.3 Characteristics and applications
Since EDM is a thermal process, hard materials such as
quenched steel, cemented carbide, and electrically
conductive ceramics can be machined. EDM also allows
machining of complicated shapes. Since the tool electrode
does not need to rotate for material removal like milling or
grinding, holes with sharp corners and irregular contours
can be machined without difficulty. Reaction forces
generated in the EDM gap are insignificant, which also
facilitates the machining of thin and flexible parts, and deep
grooves and holes which are difficult to machine by milling.
Generally, machining accuracy of EDM is very high in the
order of several micrometers, and achievable surface
roughness is Rz 0.4µm. On the other hand, the material
removal rate of EDM is low compared to other machining
processes. Hence EDM is mostly used in die and mold
making, where complicated shapes in hard materials and
with high precision have to be machined. In spite of
decreased opportunities for the EDM process to be
adopted in the manufacturing processes of certain dies and
molds as a result of the recent improvement of high-speed
milling technology, EDM of micro-parts has paved the way
for expanding application fields.
2.4 Pulse generator
Early EDM equipment used relaxation type pulse
generators with capacitor discharges as shown in Figure 4.
This type of equipment has been used especially in WEDM
until recent years because discharge current with high peak
values and short duration is needed in WEDM due to the
reasons described in Section 5.1. With improved capability
of power transistors which can handle large currents with
high response, the relaxation type was replaced by the
transistor type shown in Figure 5. However, the relaxation
type pulse generators are still being used in finishing and
micro-machining because it is difficult to obtain significantly
short pulse duration with constant pulse energy using the
transistor type pulse generator as described in Section 6.1.
In Figure 5, a series of resistances and transistors are
connected in parallel between the direct current power
supply and the discharge gap. Suppose one transistor is
switched on and a discharge occurs, since the discharge
voltage is around 20V, the discharge current is 4A in this
case. Here the discharge voltage is almost constant at 20V
regardless of the discharge current, because a larger
current brings about a higher degree of ionization and
greater diameter of the arc column resulting in increased
electrical conductivity of the plasma. Hence, the discharge
current increases proportionally to the number of
R
C
Tool electrode
Charge Discharge Workpiece
Figure 4: Relaxation type pulse generator.

Gate control circuit
20 Ω
100V Tool electrode
Power supply Workpiece
Figure 5: Transistor type pulse generator.
Ignition delay time td
Pulse interval to
Gap voltage Open voltage uo
≒100V
0V
Discharge voltage
ue = 20V Time
Current Discharge
0A current ie
Discharge duration te Time
(µs – ms)
Figure 6: Gap voltage and current waveforms.
transistors which are switched on at the same time.
Figure 6 shows voltage and current waveforms measured
at the gap, which are typical in sinking EDM operations.
When the transistors are switched on, the open voltage uo,
100V in the case of Figure 5, is applied between the tool
electrode and workpiece, but discharge does not occur
immediately but occurs after the ignition delay time, td. This
is because dielectric breakdown “physically” needs a
statistical time lag in which initial electrons should be
generated by the bombardments of cosmic, radioactive,
ultraviolet, and X rays with the neutral species in the gap,
and the formative time lag in which the initial electrons are
accelerated to generate the electron avalanche [3]. In EDM,
since the dielectric liquid is contaminated with electrically
conductive debris particles, the movement of debris
particles in the gap should be considered. Bommeli et al.
[4] statistically investigated the distribution of ignition delay
time to demonstrate that discharge occurs after an
activation time which is “practically” needed in addition to
the statistical and formative time lags for debris particles to
form bridges between electrodes. Since dielectric
breakdown strength decreases with the decrease of the
gap width and increase of the density of debris particles,
the ignition delay time reflects the gap conditions.
After the dielectric breakdown, a discharge current, ie, is
passed through the gap. The gate control circuit keeps the
transistors on for the discharge duration, te, after the
dielectric breakdown, resulting in a uniform discharge
crater size independent of the ignition delay time which can
vary statistically for each discharge. Then after the fixed
discharge interval, to, the transistors are again switched on
and open voltage is applied between the electrodes.
The discharge energy per single pulse q is expressed as,
q = ue × i e × t e (1)
where ue is the discharge voltage, which is around 20V.
Hence discharge current and discharge duration are set by
operators considering whether the process is for a roughing
or a finishing operation. To obtain stable machining, the
discharge interval should be sufficiently long so that the
plasma is extinguished and hence the dielectric breakdown
strength is recovered during the interval. However, too long
an interval results in a decrease of the material removal
Gap voltage
Long delay Forward
0V
Time
Large gap
Gap voltage Gap voltage
Proper delay
Time Proper gap
Short delay
Retract
Time
Small gap
Figure 7: Principle of servo feed control.
rate. Advanced electrical discharge machines can control
the discharge interval adaptively by monitoring the gap
voltage waveforms.
2.5 Servo feed control (Gap control)
Tool electrode feed is not constant like in conventional
machining methods. The servo feed control shown in
Figure 7 keeps the working gap at a proper width. Larger
gap widths cause longer ignition delays, resulting in a
higher average gap voltage. When the measured average
gap voltage is higher than the servo reference voltage
preset by the operator, the feed speed increases. On the
contrary, the feed speed decreases or the electrode is
retracted when the average gap voltage is lower than the
servo reference voltage, which is the case for smaller gap
widths resulting in a smaller ignition delay. Thus short
circuits caused by debris particles and humps of discharge
craters can be avoided. Also quick changes in the working
surface area, when tool electrode shapes are complicated,
does not result in hazardous machining. In some cases, the
average ignition delay time is used in place of the average
gap voltage to monitor the gap width [5].
3 GAP PHENOMENA
3.1 Single pulse discharge
EDM gap phenomena are usually depicted as shown in
Figure 8. The arc column is depicted with a diameter
smaller than the gap width. Debris particles are very fine
like dust and minute bubbles are suspended in the gap.
However, this sketch is not entirely correct. As described
later, the gap is mostly occupied with gas and the diameter
of debris particles is in the same order as the gap width and
surface roughness of the electrode surfaces as
schematically shown in Figure 9. Dielectric breakdown in
consecutive pulse discharges therefore occurs at the
boundary between bubbles, where debris particles are
condensed, or inside the bubble.
Figure 10 shows a schematic view of the discharge spot.
The arc column diameter is considered to increase with the
passage of time [6-8] and equal to the diameter of the
generated discharge crater [6]. If this is true, the diameter
of the arc column measures several times larger than the
gap width. Electrode materials and dielectric liquid are
evaporated, molecules are dissociated, and atoms are
ionized, resulting in a rapid expansion of the bubble. Since
the expansion is restricted by the inertia and viscosity of the
dielectric liquid, the pressure inside the bubble becomes
extremely high and the boundary between the bubble and
liquid expands with the velocity of several tens m/s [9, 10].
It is still believed that the dielectric liquid plays a significant
role in material removal because the high pressure and
velocity field in the bubble may serve as the dynamics of
the material removal in EDM [11, 12].
After the end of the discharge duration, ions and electrons
are recombined and the dielectric breakdown strength is
recovered. The evaporated atoms and molecules are
 Discharge column Bubble role in flushing debris particles and cooling of the gap.
3.2 EDM plasma
Initial electrons, which are generated by the ultraviolet ray,
x-ray, cosmic ray and radiation from the earth crust, are
Melted pool accelerated by the electric field and ionize the neutral
Dielectric breakdown Melting and Expansion of bubble
evaporating
species due to collision, resulting in an electron avalanche.
Thus, the electric field is distorted and streamers are
developed toward both the anode and cathode, resulting in
Debris
an established discharge [3]. In EDM, since discharge
Bubble duration is normally over several µs and current density is
Recovery of dielectric Discharge interval
108 - 109A/m2, the established discharge is an arc
strength
?
Figure 8: Model of EDM gap phenomena.
discharge. An arc discharge is sustained by electron
emission from the cathode spot which is due not only to the
secondary emission but also to the thermionic and field
emissions. Since both temperature and electrical field are
Debris particles strong, the emission process is strongly dependent on both
variables (T-F theory) [14]. Thus the plasma is highly
Tool electrode ionized resulting in high current densities with the
comparatively low discharge voltage of about 20 V. The
discharge voltage is composed of an anode, cathode and
Bubble Bubble channel voltage drop, and its value will slightly change
depending on electrode materials, dielectric fluids, gap
width and pulse conditions.
Because of the complicated physical phenomena occurring
in the discharge gap, precise analysis of the EDM plasma
Workpiece is difficult. Eubank et al. [15] analyzed the expansion of the
cylindrical plasma considering the evaporation of water
Dielectric fluid Tar dielectric and enthalpy increase in plasma due to
dissociation and ionization of water. The fraction of energy
Figure 9: Schematic view of discharge gap. distributed to the plasma was obtained by subtracting the
fractions of the energy transferring to the anode and
Tool electrode cathode from 100%. The fractions of the anode and
Melted zone
cathode were obtained from comparison between the
Debris Bubble
measured material removal per pulse discharge and
molten material volume calculated using the point
Liquid heat-source on cathode model [16] and expanding circular
heat-source on anode model [17]. However, the plasma
itself was not analyzed.
Discharge current Discharge plasma
Melted zone
Hayakawa et al. [18-20] first conducted
Workpiece
magnetohydrodynamics analysis in the steady state of a
DC arc between parallel plane copper electrodes. The arc
Figure 10: Discharge phenomena in EDM gap. was assumed to be in air under a constant discharge
Diameter reaches several millimeters
current and gap width as those used in the actual EDM
process. They assumed that the species in
Bubble
high-temperature air which includes copper electrode vapor
Debris are N2, O2, NO, N, O, Cu, NO+, N+, O+, Cu+, N2+, O2+, Cu2+,
Tool electrode
and electrons. Considering the temperature dependence of
the thermophysical properties of the plasma, the
Boundary Discharge spot electromagnetic field, temperature, pressure, and velocity
Workpiece Boundary distributions were calculated for the regions including both
the electrodes and discharge gap. The conservation
Figure 11: Bubble and debris particles generated by
equations of mass, momentum and energy, Ohm’s law and
single pulse discharge.
Maxwell’s equations were solved. The energy equation
solidified or condensed to form debris particles or dielectric contained the Joule heating, conduction, convection, and
liquid, but gases such as hydrogen and methane which are radiation terms. Figure 12 shows an example of the
generated by the dissociation of the working oil are left to calculated temperature fields. It was found that most of the
form a bubble. The diameter of the bubble reaches several discharge power is distributed in the electrodes, and heat
millimeters, several tens of times larger than the gap width transfer due to convection and radiation is negligible. It was
as shown in Figure 11. Since pulse discharge occurs also found that the plasma is extinguished within a few
several thousand times or more per second, obviously the microseconds after the end of the discharge duration as
gap becomes filled with gas in typical EDM processes [12]. shown in Figure 13 [18,20]. However, the arc which they
Hundreds of debris particles are generated per single pulse analyzed was not in dielectric liquid but in air. The arc was
discharge, blown off through the bubble generated by the not in transient but in steady state, and removal of the
discharge and stopped at the boundary of the bubble due electrodes, i.e. mass transfer from the electrode surfaces,
to the viscosity of the dielectric liquid [13]. Melted and was not taken into consideration. Furthermore, the arc was
evaporated materials are cooled by the dielectric liquid and assumed to be in thermoequilibrium, and the equations of
solidified to form spherical debris particles. Thus debris motion of the three species: electrons, ions, and neutral
particles are removed from the gap with the dielectric liquid particles, were not solved separately. The gap phenomena
and not reattached on the electrode surfaces. Furthermore, were simplified to be symmetrical between the anode and
heat convection in the boundary layers of the dielectric the cathode regions. Therefore, the fractions of energy
liquid cools the electrode surfaces, resulting in machining distributed to anode and cathode were equal.
stability. Thus, the dielectric liquid plays another important
 Removal amount, ×10-9 g/pulse
16000

Temperature [K] [K]

ie = 25 A
Temperature 12000

8000 anode (Cu)

cathode (Cu)
4000

0
0 AnoAnode
de
100
Gap
RadRiu
adius 200 Gap
s [μ[μm] 300 Catho
m] Cathode
de
Pulse duration, te µs
Figure 12: Temperature distribution in electrodes and
Figure 14: Relationship between removal amount and
gap obtained by magnetohydrodynamics analysis.
pulse duration.
10000

0.1mm
Temperature [K] [K]

Calculation model
Temperature

Heat flux q”
5000 Heat source area A
Measuring point
Fe
Fe
Foil electrode
Cu
Cu
Rod electrode
Foil electrode
0 Experimental Constantan wire
0 1 2 3 4 5
setup (φ0.1mm)
Time [µs] Thermocouple
Time [µs]
Figure 13: Calculated plasma temperature decrease
during discharge interval when material used for both Amplifier
anode and cathode is steel or copper. Figure 15: Method to obtain energy distributed
into electrode in single pulse discharge.
3.3 Energy distribution
The anode and cathode removal amounts are greatly
Energy distribution ratio

dependent on the discharge duration, and they are quite

different even when the materials of the two electrodes are
the same. Figure 14, for example, shows the relationship
between removal amount per pulse and discharge duration
obtained when copper was used both for anode and
cathode [21]. The anode removal amount is greater than
the cathode one for discharge duration shorter than 20µs, anode ie = 25 A
while it is smaller with the discharge duration longer than cathode
20µs. Motoki and Hashiguchi [22], and Van Dijck [23] Gap = 40 μm
explained that such a phenomenon is caused by the
variation of the energies distributed into the anode and
Pulse duration, te µs
cathode with the discharge duration, on the basis of the T-F
electron emission theory [14]. Figure 16: Relationship between energy distribution
Hence it is important to measure the energy distribution ratio and pulse duration.
between the anode, cathode and dielectric fluid. Koenig et It was concluded that the difference in removal amount
al. [24] measured the energy distribution by measuring the between anode and cathode is mainly caused by the
temperatures of the electrodes and dielectric fluid. Xia et al. phenomenon of carbon adhesion onto the anode surface,
[21,25,26] measured the energy distribution by comparing not by energy distribution, when hydrocarbon dielectrics
the measured temperatures of the electrodes with the are used and both anode and cathode are made of copper.
calculated results obtained under the assumed ratio of the Motoki et al. [28] and Ikai et al. [29] also explained that the
energy distributed in electrodes, using the calculation electrode wear amount is reduced due to the protective
model and experimental setup shown in Figure 15. When effects of carbon layer with high boiling temperature and
the calculated temperature agreed with the measured one, high thermal resistance. Mohri et al. [30] examined the
they found that the estimated energy distribution was catalysis and structure of the carbon layer generated in the
correct. Figure 16 shows the energy distribution obtained EDM process. Kunieda et al. [31] measured copper vapor
by Xia et al. when copper was used for both anode and density in the arc plasma using spectroscopic analysis and
cathode under the same pulse conditions as in Figure 14. proved that the carbon layer protects the copper anode
The energy distribution to anode and cathode is about 40% from wear based on the fact that the deposition of thicker
and 25%, respectively. They reported that the energy carbon layer on the copper anode results in lower copper
distributed to anode is always greater than that to cathode, vapor density.
and is rarely affected by the discharge duration, in both
single discharge [21] and continuous pulse discharges [26]. Figure 17 shows the energy distribution in the gap
measured by Xia et al. [26] in the continuous pulse
It is obvious that the difference in amount of material discharge. Ratio of energy used for removal of electrodes
removal cannot be fully explained by the energy distribution. is significantly low (1%), and most of the heat is lost by
Xia et al. [21,25] and Natsu et al. [27] investigated the conduction into the electrodes. Van Dijck [23] also indicated
influence of carbon adhesion to the anode on the removal that more than 90% of the heat is conducted into the
amount under different machining conditions, such as electrodes and that the ratio of the volume of metal ejected
discharge duration, polarity, and dielectric flushing flow rate. to the volume of metal melted is only 1 to 10%.
 Energy into tool electrode
Energy into workpiece
Energy 34%
carried away
48% by debris
Loss into 1%
Total discharge
gap
energy
18%
30%
34%
Loss due to
conduction
Energy into workpiece
(a) Distribution of discharge (b) Energy distribution in
energy workpiece
Figure 17: Energy distribution in EDM process
(Ie:16A, te:100μs , Cu(+) – Cu(-)).
It is also found from Figure 17 that 18% of the energy goes
to the discharge gap. In the magnetohydrodynamics
analysis of the steady state arc, however, Hayakawa et al.
[20] found that almost all the discharge power is conducted
into the electrodes and the heat dissipated by convection
and radiation is negligible. This is because the arc column
is established in steady state and both convection and
radiation are insignificant in the narrow gap between
parallel plane electrodes. This result agrees with the
distribution measured when discharge duration was several
thousand times longer than the duration actually used in
EDM [20]. The power distributed into the gap with the
actual discharge duration is considerably higher than that in
the steady arc. Thus it was concluded that the gap
condition is not in equilibrium during the actual discharge
duration in EDM, and a large fraction (18%) of the
discharge power is consumed in the formation of plasma
through ionization, excitation, dissociation, and
polymerization. So it means the plasma formation is still
going on.
3.4 Measurement of plasma temperature
The arc plasma temperature in EDM was measured using
spectroscopic analysis, and it was found that the
temperature reaches 6000 to 7000K [32-35]. Figure 18
shows the change in measured temperatures during and
after a discharge when a copper electrode of 2mm in
diameter was used both for anode and cathode under the
following pulse conditions: discharge current: 40A;
discharge duration: 300µs. It is found that the deionization
time for a 0.5mm gap width is considerably shorter than for
1.5mm. This is because a wide gap results in a small
temperature gradient and a great heat capacity of the
plasma. On the other hand, Hashimoto et al. [34] found that
the ignition delay time of the following pulse discharge
becomes zero when the discharge interval after the
preceding pulse discharge was shorter than 5µs. They also
found that the plasma temperature measured at 5µs after
the cease of discharge current was below 5000K. The
calculated temperature decrease during the discharge
interval shown in Figure 13 also supports their results. This
means that a plasma temperature of 5000K is considered
as the threshold of deionization. This temperature drop
during the pulse interval shows the minimum interval
10000
Gap distance 1.5mm
Temperature K
8000 Gap distance 0.5mm
6000
4000
Discharge duration: 300µs
2000
0 when the bubble pressure drops sharply, promoting the
-50 0 50 100 150 200 250 300 350
Time µs
Figure 18: Plasma temperature at gap center during
and after discharge measured by spectroscopy.
required to obtain stable machining.
3.5 Thermal aspects
Loss due to
convection To obtain the relationship between pulse conditions and
and radiation material removal rate, many attempts have been made to
3% calculate temperature distribution in the electrodes caused
by a single pulse discharge by solving time-dependent heat
transfer equations assuming various heat source models.
By integrating the solution for a point source which is
liberated instantaneously at a given point and time with
regard to appropriate space and time variables, one can
obtain solutions for instantaneous and continuous sources
of any spatial configuration (Carslaw and Jaeger [36]). The
analytical solutions of the temperature distributions at time t
produced by stationary Gaussian energy distribution
sources in a semi-infinite solid were given by Pittaway [37].
With the advent of powerful computers and numerical
methods, it is nowadays not difficult to take into account the
temperature dependence of thermophysical properties of
electrode materials and latent heat of melting and
vaporization. It is also possible to consider the time
dependent radius of the circular heat source and time
dependent heat flux, using the principle of superposition
based on the linearity of the heat conduction problem.
It should be noted, however, that boundary conditions, such
as the time dependent function of plasma radius, spatial
configuration of the heat flux, distribution of heat flux in the
heat source, and ratios of energy distributed to anode and
cathode, also exert significant influence on the calculation
results. Snoeys and Van Dijck [7,8,23] systematically
analyzed the temperature distribution assuming a circular
heat source with time dependent radius and time
dependent heat flux on a semi-infinite cylinder. The heat
source growth function was obtained from an iterative
calculation of the heat conduction equation by assuming
that the temperature at the center of the heat source
corresponds to the metal boiling temperature at a pressure
equal to an average pressure in the gas bubble calculated
from the thermodynamical model. Good agreement was
found between melting point isothermals calculated and
those measured from the pictures of the cross sections of
craters cut perpendicular to the surface. Since boiling of
superheated metal is suppressed by the bubble pressure
during the discharge duration, the most significant removal
was assumed to be associated with ejection of liquid metal
by bulk boiling inside the normal boiling point isothermal
surface, at the end of the discharge. The calculated
superheated volume agreed well with the measured
material removed per pulse. It was also found that the
metal removal efficiency, which was defined as the ratio of
the ejected to melted volume, was 0.01 to 0.1.
However, with regard to the heat source configuration,
some papers assume a point heat source [16], and many
other papers consider that the diameter of the heat source
is equal to that of the discharge crater generated by the
discharge [6,21,22,28,29]. Although, in most cases, it is
assumed that the heat source is stationary and uniform in
its heat flux, Kunieda et al. [38] measured the arc column
movement in a single pulse discharge and found that both
anode and cathode spots move in the same way and the
area of movement becomes larger with the progress of the
discharge. Their detection method, however, does not yet
allow the measurement of the diameter of the arc column.
About the timing of material removal, there are still different
findings. Zolotykh [11] reported that high-speed
5 µs photographs showed that most of the metal is ejected due
to boiling of the superheated crater at the end of discharge
discharge of the gas dissolved in the molten metal. On the
other hand, Yoshida et al. [13] found that metal removed
per pulse in air is almost equal to that in liquid, indicating
that metal removal can occur without a sharp drop of
bubble pressure. Observation of the gap of single pulse
discharge in dielectric liquid using X-ray also shows that
85% of material removal occurs during the discharge
duration [39].
Xia [25] examined the influence of the assumptions used in
the boundary conditions: heat source diameter, heat flux
distribution in the heat source, and timing of metal removal,
on the calculation results of temperature distributions in the
electrodes. It was found that these assumptions exert a
significant influence on the calculation results of boiling
point and melting point isothermal surfaces. Unfortunately,
however, determining the boundary conditions is still a
controversial issue. Hence fundamental studies on the
issue are highly looked forward to.
To show the influence of thermophysical properties of
electrode materials on temperature distribution inside the
electrode, the heat transfer equation was solved using a
simplified model assuming an expanding circular heat
source with a uniform heat flux and ignoring the metal
removal and temperature dependence of thermophysical
properties [40].
Figure 19 and Figure 20 show the calculated results when
steel and copper are used as the electrode material,
respectively, assuming the same energy distribution ratio of
25%. Based on the formula obtained by Saito et al. [41]
assuming that the heat source diameter is equal to the
discharge crater diameter, the time dependent function of
the heat source diameter d(t) was determined as:
0.4
d (t ) = 2.4 × 10 −3 × t 0.4 × i e [m]
Here, t is the time after ignition. Although the melting point
of copper (1357K) is lower than that of steel (1808K), the
melted zone of copper is considerably smaller than that of
steel due to its higher thermal diffusivity (1.1×10-4m2/s) than
that of steel (2.0×10-5m2/s). With the expansion of the
plasma radius, the melted zone is even resolidified during
the discharge duration in the case of copper. During the
discharge interval to, since the surface temperature at the
discharge spot drops rapidly in copper, the time needed for
Temperature, K
In-d
e
po pth l n
μmsition adia sitio
z, R po m
μ
t =10μs,
d(t) = 50.8μm
Figure 19: Temperature distribution in steel workpiece (cathode) (ie: 30A, te: 100μs, energy distribution to cathode: 25%)
Temperature, K
In-d
e r,
po pth l n
μmsition dia sitio
z,
Ra po m
μ
t =10μs,
d(t) = 50.8μm
Figure 20: Temperature distribution in copper workpiece (cathode) (ie: 30A, te: 100μs, energy distribution to cathode: 25%)
the plasma to be extinguished is short, leading to stable
machining. Difference in the plasma temperature decrease
calculated, shown in Figure 13, is a result of the difference
in the thermal diffusivity of the electrode material. Uno et al.
[42] and Suzuki et al. [43] used CVD-carbon and
electrically conductive CVD diamond as tool electrode
materials, respectively. Due to the high thermal diffusivity of
these materials, EDM could be performed with almost no
wear of the tool electrode.
Zingerman [44] states that Joule heating does not play a
substantial role in the machining of metallic materials,
because electrical potential drops in metallic workpieces
are negligibly small. According to the calculated results of
Rich [45], however, the Joule heating is extremely
concentrated in a hemispherical volume directly under the
discharge spot, and for high-resistivity metals, e.g., Hg, Sb,
Fe, and Bi, the Joule heating is comparable to the energy
input from the arc plasma. Saeki et al. [46] calculated the
workpiece temperature distribution considering the Joule
heating as well as the heat flux from the arc column and
found that the removal of high-electric-resistivity material
(Si3N4-30wt%SiC) in a single discharge is greater than that
of steel due to Joule heating under the same discharge
current. However, it was found that a longer discharge
interval is necessary to deionize the arc plasma and
recover the insulation of the gap after each pulse, which
results in a lower material removal rate of high-resistivity
materials in a practical EDM.
As a thermal machining process, EDM provides a means of
(2) machining ceramic materials, irrespective of their hardness
and strength, provided that their electrical conductivity
values are of the order of 0.01 S/cm (100Ωcm) [47]. Owing
to the sometimes very different physical properties of the
materials, electrical discharge processing produces a
porous surface on which the binder phase has been
preferentially removed. Processing of thermo-shock
sensitive materials causes material to be removed by
spalling.
r,
t =100μs, 20μs after
d(t)=121μm discharge
t =100μs, 20μs after
d(t)=121μm discharge
3.6 Chemical aspects
EDM is undeniably a thermal process. However, by
acknowledging that EDM is also a chemical process, new
findings and creative ideas can unexpectedly be obtained.
One of the applications of arc discharge was originally
production of fine particles of metals [48]. Even now the
Reactive-Electrode Submerged-Arc (RESA) method [49] is
under investigation for producing ultrafine powders based
on reaction between electrode materials and dielectric fluid.
It is well known that diamond-like carbon [50] and carbon
nanotubes [51] can be produced using arc discharge
evaporation methods.
The EDM arc plasma temperature is high enough to not
only evaporate the electrode materials and dielectric liquid
but dissociate and ionize their molecules and atoms. As
Hayakawa et al. [20] concluded, the EDM gap condition is
not in equilibrium during the discharge duration, and a large
fraction of the discharge power is consumed in chemical
reactions. Eubank et al. [15] analyzed the enthalpy
increase in plasma involving dissociation and ionization of
water. Lauwers et al. [52, 53] investigated the cutting rate of
WEDM of ceramics using de-ionized water, and pointed out
that besides typical material removal mechanisms like
melting and evaporation, other mechanisms such as
decomposition and oxidation were identified.
Toenshoff et al. [54] states that using a dielectric such as
mineral oil or an organic fluid for ED sinking, hazardous
fumes like polycyclic aromatic hydrocarbons, benzene,
vapour of mineral oil, mineral aerosols and various
by-products are generated by the dissociation of oil and its
additives. Kusano et al. [55] analyzed the components of
the bubble generated using a hydrocarbon dielectric by gas
chromatography and showed that the volume fractions of
hydrogen gas and hydrocarbon gases, which involve
molecules with carbon atoms less than seven, are 56% and
36%, respectively, under a medium finishing condition.
Another example is the deposition of carbon on the anode
surface by the thermal decomposition of the working oil
(Figure 21). The white layer generated on workpieces
machined in hydrocarbon oils contains more carbon than
the base material, while that in a water dielectric contains
less carbon than the base material [56]. Material removal
rates can be improved by supplying oxygen gas into the
discharge gaps filled with water-based dielectric liquid [57]
or air [58]. Material removal rate using water based
liquid, and electrode materials with low melting point, at
least at the surface. Thus, deterioration of surface quality
due to electrolysis in de-ionized water occurs [64]. When
ED-machining on high alloy steels in the commonly used
de-ionised water, toxic ions of heavy metals become
dissolved in the dielectric. These are examples which
demonstrate the importance of chemical aspects.
3.7 Role of dielectric liquid
As described in Section 3.5, Zolotykh [11] and Snoeys and
Van Dijck [7, 8, 23] reported that most of the metal removal
occurs due to boiling of the superheated molten mass in
the crater at the end of discharge because boiling of that
superheated metal is prevented by the bubble pressure
during the discharge duration. Hockenberry et al. [12]
stated that after the discharge termination when the
expanded gas bubble starts to collapse due to its pressure
drop below atmospheric pressure, the gas bubble becomes
fragmented and the liquid jets which penetrate the bubble
and impinge upon the electrode crater from all directions
will strike the discharge crater and eject the molten metal.
On the other hand, Yoshida et al. [13] observed debris
formation and their distribution on the working surface due
to a single discharge. Figure 22(a) and 22(b) show the
distribution of debris obtained from single pulse discharge
in air and liquid, respectively. In the case of discharge in
liquid, a solution of gelatin (25wt%) of 303K was used as
dielectric liquid, and after the discharge the temperature of
the experimental setup was lowered to freeze the gelatin to
8mm
Bubble boundary
1.5mm
- 8mm 8 -3mm
8 3mm
8mm
-1.5mm
- 8mm (b) In liquid
(a) In air
Ie 32A, te 120μs
Under 5μm 10 15 20 Over 30
Figure 22: Size and location of debris particles scattered
on electrode surface due to single pulse discharge.
h
h
Distribution of debris diameter

Distribution of debris diameter

dielectrics compared with that of hydrocarbons is 100 100h

controversial. Although the material removal rate was 80 80h
based on volume, %

Over 30μm
based on volume, %

Over 30μm
improved by mixing organic compounds into water, it did 60 60h
h
not compete with that of hydrocarbon oil [59, 60]. Some 40 40
5 h
publications, however, showed higher material removal rate 20
15
20μm 20
h
20μm
2 10 15μm
of water-based dielectrics [52, 53, 61-63], because water is 0 0h 10
5
dissociated by discharge to hydrogen and oxygen, resulting 20 60 120 200 360
h20 60 120 200 360

in oxidation of workpiece materials [52, 53, 62]. Discharge duration, μs Discharge duration, μs
In order to reach stable machining conditions in WEDM In air In liquid
(large gap width for the vibration of the wire) and to obtain a
high rising speed of the discharge current, it is necessary to Figure 23: Debris diameter distribution based on volume.
use a relatively high open voltage, less insulating dielectric
Total volume of generated

In liquid
10-4
In air
10-5
Ａｎｏｄｅ（＋）
debris, mm3

10-6

Carbon 10-7
deposition
10-8
0 100 200 300
Discharge duration, μs
Cathode （－）
Figure 21: Deposition of pyrolytic carbon on anode in Figure 24: Comparison of total volume of
hydrocarbon dielectric (SEIBU ELECTRIC & MACHINERY). generated debris between air and liquid.
keep the debris from scattering. Then distribution of the
size of debris generated by the single discharge was
observed by an optical microscope. Figure 23 shows the
debris diameter distribution in volume. Figure 24 shows the
measured total volume of debris generated in air and in
liquid. It is found that there are differences in the total
volume of the generated debris and the distribution of the
debris size between single pulse discharges in liquid and in
air when the discharge duration is short. No difference
exists however for discharge durations longer than 90µs.
The debris volume difference is very small when the
discharge duration te is long. This is because the discharge
column is located at the center of the bubble generated by
the discharge itself and the diameter of the bubble grows
up to several millimeters as shown later in Figure 29.
Hence, as time elapses, the arc column environment
becomes equivalent to that of the discharge which occurs
in gas with the increase in the diameter of the bubble.
The above results indicate that metal removal can occur
without a liquid dielectric. In air, however, most of debris
particles were reattached to the workpiece surface, as
shown in Figure 25(a) [13]. This is because the melted
debris particles move with constant velocity in air and do
not solidify until they hit the electrode surface. In the case
of liquid, in contrast, the debris particles proceed straight
through the gas bubble and penetrate the bubble wall, and
as a result decelerate. They then solidify into a spherical
shape under the influence of surface tension [12] as shown
in Figure 25(b), confirming that the dielectric liquid is
important for the cooling and flushing of debris particles but
not for material removal.
Miyajima et al. [65] found that the working gap is mostly
occupied by bubbles although the working gap is
submerged in dielectric liquid. Figure 26 shows the bubble
conglomerated in the discharge gap when machining was
interrupted and the tool electrode was lifted to widen the
gap. From the volume of the conglomerated bubble it was
(a) In air (b) In iquid
Figure 25: Shapes of debris generated in air and liquid.
Liquid surface
Tool electrode (+)
(Copperφ20mm）
Workpiece (-)
(carbon steel)
Figure 26: Bubble unified in discharge gap when
machining is interrupted （ie 9A, te 120μs, to 40μs）.
Bubbles
Discharge location
Dielectric
contaminated
with debris
Figure 27: Schematic view of working surface.
found that 80 to 90% of the working area was occupied by
gas bubbles during the process as illustrated in Figure 27.
Imai et al. [66] measured the bubble quantity in the working
gap utilizing ultrasonic waves passed through the working
gap. The ultrasonic wave transmitted through the electrode
and incident normal to the working surface was mostly
reflected at the interface when the gap was not filled with
dielectric liquid. They found that more than half of the
working area is occupied by bubbles at higher discharge
currents, shorter discharge intervals and longer machining
periods between jump flushing motion of the tool electrode.
Thus Tanimura et al. [67] proposed a new EDM method in
which air mixed with water mist is supplied to the discharge
gap and found that the EDM in mist shows almost the same
machining abilities as conventional EDM. Karasawa et al.
[68] showed that the material removal rate with a dielectric
liquid poured into the gap in air is higher than that with the
gap submerged in a tub of the dielectric liquid because the
ability of flushing debris in the pouring method is higher
than that in the submerging method. All the above results
suggest that removal itself does not require the gap to be
entirely filled with liquid. It should be noted that findings
obtained from single pulse discharges in a gap filled with a
dielectric liquid cannot always be extrapolated in the actual
EDM process. Natsu et al. [69] found that removal in a
single pulse discharge is greater than the material removal
per pulse in consecutive pulse discharges. They measured
the change in the removal amount at anode and cathode
with increasing number of repetition of pulse discharges in
an originally fresh inter-electrode atmosphere which is
achieved by the jump action of the tool electrode, and
found that the removal amount per pulse decreases
strongly with repetition.
3.8 Forces applied to electrodes
The force accompanying the discharge is caused mainly by
the expansion and contraction of the bubble generated by
the evaporation, dissociation and ionization of the dielectric
liquid and electrode materials. Figure 28 shows a model to
calculate the bubble oscillation in the gap between
parallel plane electrodes [9]. The model illustrates a case
that a bubble with a high pressure is generated at the
center of the gap between cylindrical electrodes. The
bubble expands and contracts periodically. The reaction
force applied to the tool electrode can be calculated by
integrating both the pressure in the bubble and that in
dielectric liquid over the working surface [70].
Photos in Figure 29 taken by Ikeda [71] show the oscillation
of the bubble generated by a single pulse discharge in a
gap between parallel plane electrodes. The reaction force
in Figure 29 was measured by Kunieda et al. [72] using the
Split Hopkinson Bar method [73], with which the influence
of the natural frequency of the measurement system was
successfully eliminated. At the initial state in which the
bubble is compressed, the force indicates the highest peak.
With the expansion of the bubble, the force decreases
according to the decrease of the bubble pressure. The
P(r)
Pressure distribution
r
Tool electrode
Atmospheric
Gap width Bubble Dielectric
pressure
Workpiece
Initial diameter of bubble
Figure 28: Pressure distribution in bubble generated by
single pulse discharge in gap filled with dielectric liquid.

Dielectric reakdown
Reaction force, N 50
0
Dielectric breakdown
- 50
0 Dielectric
Time, ms
Figure 29: Oscilation of bubble observed by Ikeda [71] and reaction force measured by Split Hopkinson
Bar method (ie 30A, te 150μs, gap width 0.15mm, anode: copper φ20mm, cathode: steel φ20mm).
force even becomes negative because the bubble
continues to expand even after the bubble pressure falls
below the atmospheric pressure due to the inertia of the
dielectric liquid which is moving radially away from the
discharge spot. When the diameter of the bubble reaches
the maximum, the absolute value of the negative force
becomes maximum. Then, the bubble starts contracting,
and the force increases showing a dumping oscillation due
to the viscosity of the dielectric liquid.
Since the natural frequencies of EDM machine structures
are much lower than the frequency components included in
the reaction force waveform, the gap width cannot respond
to the change in the force. Furthermore, the reaction force
in a series of pulse discharges decreases with time while
the working gap fills with bubbles generated by each
discharge [70, 72]. Hence influence of the force caused by
each pulse discharge is negligibly small in sinking EDM.
However, it causes vibration and deflection of electrodes in
the cases of wire EDM and micro EDM.
It is also known that an electrostatic force occurs between
the tool electrode and workpiece [74, 75] mainly during the
discharge delay time in which an open voltage is applied
between them. In WEDM, when the discharge energy is
small, the influence of the electrostatic force cannot be
ignored in comparison with the above mentioned reaction
force. Although the electrostatic force per unit length along
the wire electrode is small compared with the reaction force,
the total electrostatic force may not be negligible, because
the force can be applied at any point on the wire which is
facing with the workpiece surface. Electromagnetic force
can also be applied to the wire electrode at any location
where the workpiece is facing the wire electrode, because
high frequency current pulses generate an electric current
flow on the workpiece surface in parallel with the wire
electrode due to skin-effects [5, 75, 76].
4 PROCESS STABILITY
4.1 Factors determining discharge location
Basically discharge occurs where the gap width is shortest.
This principle guarantees the replicating accuracy of EDM.
To obtain an electric discharge in clean oil the gap width
must be less than several µm. However, since the dielectric
liquid is contaminated with electrically conductive debris
particles, whose average diameter is one third to half of the
gap width, discharge occurs through the debris particles,
resulting in gap width much larger than that for an
uncontaminated dielectric [4, 77-79]. This fact sometimes
causes reduced replicating accuracy due to the uneven
Photographed
by Ikeda [71]
Discharge duration
2breakdown 4
distribution of the debris. On the other hand, due to the
extended gap width, servo gap control becomes easy.
Otherwise it is difficult to keep gap widths of several µm
constant.
A simplest explanation for the decrease in the dielectric
breakdown strength is that the electrically conductive
debris particles reduce the apparent gap width by a
distance equal to the diameter of the particles. However,
the measured gap width is occasionally larger than 100µm,
whereas the measured average diameter of debris particles
is only 25µm under a certain pulse condition. This result
can be explained in terms of the movement of debris
particles in the gap, which was observed by Suda et al. [78],
Bommeli et al [4], Schumacher [79] and Kunieda et al. [80].
They found that the debris particles move between the
anode and cathode in a direction perpendicular to the
electrode surfaces due to electrophoresis. They also
observed that some particles are linked in series to form
chains parallel to the electric fields.
From the observed results Bommeli et al. [4] and Kunieda
et al. [80] hypothesized that in the gap there are numerous
chains of particles which almost bridge the gap and that
discharge occurs at the site at which the end of a chain is
closest to the opposite electrode surface. Bommeli et al. [4]
statistically investigated the distribution of ignition delay
time to demonstrate that discharge occurs after an
activation time which is necessary for debris particles to
form bridges between the electrodes. Kunieda et al.
calculated the electrophoresis-induced motion of a particle
and found that the time interval for 30µm diameter steel
particle to reciprocate between the parallel plane
electrodes and the average velocity were 1.32ms and
0.136m/s, respectively showing a good agreement with the
results of observation using a high-speed video camera.
The velocity is high enough for the last particle to enter
small gaps between the end of the bridge already present
and the opposite electrode surface and ignite the discharge
during the ignition delay time.
In order to further study the influence of debris particles, a
debris particle with a diameter of 5µm is placed in a gap of
20µm in width as shown in Figure 30. If it is true that
discharge occurs at a point where the gap is shortest, the
discharge will probably occur at the point where the debris
particle is placed. In this experiment [81], the gap was filled
with a clean EDM working oil and the surface roughness of
the electrodes was 0.8µm Rz. However, the experimental
results showed that in most cases, the discharge crater
could not be found at the point where the debris particle
 50mm ie= i1+ i2

i1 i2

ro ro

20μm
Discharge Tool electrode
rc rc
5μm R1 R2

Debris particle
Workpiece
Figure 30: Discharge location is determined in
not deterministic but probabilistic way. Figure 31: Principle of measuring discharge locations.
was placed. This is because occurrence of discharge is a Feed of tool electrode
probabilistic phenomenon and the probability of occurrence
of discharge per unit area is not zero even if the gap width Elede

at the place is longer than other places. Hence the Workpiece

probability of occurrence of discharge on a surface is
determined by the product between the probability of
occurrence of discharge per unit area and its surface area.
Since the projected area of the debris particle in the
direction normal to the electrode surface is negligible
compared with the area of the electrodes (50mm×50mm),
the probability of occurrence of discharge at a point where
the debris is not placed is greater than that at the point
where the debris is placed. It is still true that the probability
of discharge per unit area is higher where the gap width is
shorter and debris number density is higher, but the
discharge location is determined not in a deterministic way
but in a probabilistic way.
4.2 Method of discharge location measurement
To obtain stable conditions in EDM, it is essential for the
discharge locations to be dispersed over the working
surface. Figure 31 shows the equivalent circuit to detect
discharge locations developed by Kunieda et al. [82] and
Obara et al. [83]. The electric feeder is intentionally divided,
and each divided feeder is connected to the tool electrode
at each end. While the resistance of the feeder, ro, and
contact resistance at the feeding point, rc, are constant, the
resistance in the tool electrode from each end of the tool
electrode R1, R2, respectively, varies depending on the
discharge location. Consequently, by measuring the ratio
between currents i1 and i2 through each divided feeder, the
discharge location can be obtained in process.
Figure 32 shows the distribution of discharge locations
measured with the extension of the depth of erosion. The
working area of the tool electrode was a rectangle of 14mm
by 105mm. The abscissa shows the number of discharges
sampled every fifty pulse discharges, and the ordinate
shows the discharge location along the longitudinal
Electro Electrode Electrode Elde
(a) (b) (c) Electrode
(d) (e)
de
Figure 33: Mechanism of movement of localized discharge.
direction of the working surface. The dots indicate the
location of each discharge. It was observed that when the
depth of erosion was shallow, the process was stable. The
distribution of discharge location was uniform, and no
relationship between consecutive discharges was visible.
The tool electrode feed rate was constant and the
distribution of gap width was uniform. With the increase of
the eroded depth however, discharge became localized.
The area of localized discharges moved periodically with a
period of several minutes. When the machining depth
increased further, the localization became prominent, and
the area of localization narrowed. The movement of the
localized area was perfectly periodical. In this phase, the
tool electrode feed was not smooth and the distribution of
gap width was not uniform. Ultimately, discharge was
concentrated at a certain location. In this hazardous
situation, it was no longer possible to continue the process
and thermal damage was left on the workpiece surface.
The mechanism of the periodical movement of localized
discharge is shown in Figure 33. If the flushing of sludge is
difficult, local distribution cannot be eliminated once it is
created at some area, even if the gap width of the other
regions is narrower, since the dielectric strength is
considerably decreased due to contamination in the
localized area. Meanwhile, the localized area moves to
other regions, where the gap width is sufficiently short. At
this moment, the position of the tool electrode is retracted
resulting in the fluctuation of the feed speed.
Electrode
105mm

Electrode
location
Electrode

Electrode

105mm
location
105mm

Spark
location

Electrode
Spark
Spark

105

105

0 4000 8000 0 4000 8000 0 4000 8000

Sampling number Sampling number Sampling number
Servo feed
Servo feed

Servo feed
20μm

0 4000 8000 0 4000 8000 0 4000 8000

Electrode Sampling number Sampling number
Electrode Sampling number Electrode
20μm
20μm

20μm

10mm 10mm 10mm

Workpiece Workpiece Workpiece

(a) Stable machining （0.1mm in depth) (b) Periodical movement of localized area （2.2mm) (c) Instable machining （3.5mm)
Figure 32: Distribution of discharge locations measured with extension of depth of erosion.
4.3 Relation between distribution of discharge dimension (Figure 34), and thus the system is not random
location and process instability but chaotic. The higher the attractor dimensions, the more
Localization and concentration of discharges result from stable will machining be.
locally weakened dielectric breakdown strength in the gap. 4.5 Discharge location measurement for preventing
There seems to be two causes for the decrease of the wire breakage in WEDM
dielectric breakdown strength. Figure 35 shows the temperature distribution along the wire
One of the causes is thermal. It was found both analytically electrode in WEDM obtained from the heat transfer
and experimentally that it takes less than five microseconds analysis using the finite difference method [86, 87]. Zone
for the temperature at the center of the plasma generated 3 indicates the part of the wire electrode where discharge
by the preceding discharge to drop under 5000K, which is occurs. Zones 2 and 4 show the part from the upper and
the temperature where the dielectric strength of the gap is lower feeding points to the upper and lower surfaces of the
considered to recover sufficiently [18, 19, 34, 35]. When workpiece, respectively. Although the average temperature
deionization is insufficient during the discharge interval, is around the boiling point of water, which is used as the
subsequent discharges occur at the same location, working fluid, the temperature at the point where the
resulting in the further increase in the temperature at the preceding discharge occurred is significantly high so that
discharge spot, which accelerates the concentration. the tensile strength of the wire weakens at this point.
Formation of debris particles in the gap can be another Hence, in the case that subsequent discharges occur at the
reason. If flushing of the debris particles is insufficient, same location, the temperature on the wire at the discharge
localization of discharge causes increased concentration of spot increases further, causing wire breakage.
debris particles locally, leading to the acceleration of Consideration of this led to the development of adaptive
discharge localization. If discharge is localized, the control systems in which the pulse energy is reduced or
distribution of the gap width becomes uneven, resulting in stopped based on the distribution of discharge locations
deteriorated machining accuracy. measured in process [88-90]. Based on a thermal model,
the temperature distribution and the maximum temperature
4.4 Chaos found in distribution of discharge locations can be estimated. The control strategy is to reduce or stop
The distribution of discharge locations looks random when the pulse energy. The duration of pulse termination
the process is stable, as shown in Figure 32(a). Han et al. depends on several parameters such as the maximum
[84] however found that the distribution of discharge temperature value, the number of times the maximum
locations in EDM is not random but chaotic using the chaos temperature exceeded the limitation of temperature setting,
theory. Chaos is a deterministic development with chaotic and the moving average of the maximum temperature [89].
outcome. In other words, although the system is evolving 4.6 Flushing
with a simple rule in a deterministic way from moment to
moment, forecasting the future of the system is To maintain stable machining, it is critical to flush debris
substantially impossible because the initial condition of the particles and cool the working gap in order to prevent the
system affects the subsequent events dramatically [85]. localization and concentration of discharge locations.
Although the EDM discharge location is determined by a Normally, the tool electrode is lifted periodically to replace
simple rule that the probability of discharge is higher where the contaminated dielectric fluid in the gap with fresh one.
the gap width is shorter and density of debris particle is Masuzawa et al. [91] proposed a special rotary electrode
higher, the movement of discharge location appears movement to enhance the pumping action of dielectric fluid
complicated and irregular. The attractor dimension during the lifting motion. Cetin et al. [92] calculated the
calculated from the correlation integral of the vector defined three dimensional fluid flow in the gap considering the
using the time series of discharge locations indicates that suspended debris particles and obtained the relationship
the dynamics of this system can be described with a low between the tool lifting velocity and height and the flushing
capability. Linear motor equipped sinking EDM machines
[93] realized increased removal rate especially in the
Discharge location

Correlation dimension dm

machining of narrow and deep grooves due to the

100mm

Random high-speed jump flushing operation of tool electrode which

Discharge location

can prevent short circuiting. Machining accuracy was also

100mm

30
improved because the lateral gaps of the groove became
25
narrower and more uniform.
20
0 500 1000 15 The working gap can be flushed by a fresh dielectric fluid
0 500 1000
No. of dischages Discharge locations jetted from nozzles placed adjacent to the discharge gap.
10
No. of discharge Masuzawa et al. [94] however reported that jetting of
5
0
dielectric fluid merely from one direction causes increased
Attractor dimension of 0 5 10 15 20 25 30 density of debris particles in the downstream, resulting in
discharge locations Embedding dimensions m uneven distribution of gap width deteriorating the
Figure 34: Chaos found in distribution of discharge locations. machining accuracy. Hence they demonstrated that
flushing from both sides, alternate flushing, and sweeping
flushing is preferable. Pressure or suction flushing through
400
holes in the electrode or workpiece remains one of the
most efficient flushing methods at least if those holes have
Temperature, ℃

to be provided anyway or does not harm the workpiece.

Both in pressure and suction flushing, one can observe
200 lower electrode wear and larger gap width at the outlet
point in comparison with the inlet point [95].
Tsukahara et al. [96] found that when an oxidized dielectric
oil is used, the metal elements in the tool electrode and
0 workpiece react with the oxidized oil, and organometallic
1 2 Zone 3 4 5 compounds are consequently formed. Thus they proposed
a dielectric liquid which is added with a fatty acid containing
Figure 35: Calculated temperature distribution along wire carboxylic acid. Use of this dielectric liquid results in a more
electrode (discharge current: 90A, discharge frequency: stable process since debris particles are partly dissolved as
18kHz, wire diameter: 0.25mm, workpiece thickness: 100mm).
organometallic compounds and not suspended in the gap
anymore.
4.7 Control of discharge locations
As described in Section 4.3, it is essential to disperse
discharge locations uniformly over the working surface. To
disperse the discharge locations, Narumiya et al. [97]
added electrically conductive fine powder into the working
fluid. Now this method is widely used in the finishing stage
to obtain a mirror-like surface. Nishikawa et al. [98] found
from experiment that the discharge locations can be
dispersed with this method because the gap width is
increased when powder is added to the working fluid. They
concluded that the dispersion of the discharge locations is
one of the reasons why mirror-like surface can be obtained
using the powder suspended working fluid. They reported,
incidentally, that another reason is the decrease of
discharge energy due to the decrease of the inherent
capacity of the discharge gap, because the surface electric
charge on the electrode surfaces may increase the
discharge current over the current supplied from the pulse
generator if the gap is narrow especially when the
discharging surface area is wide.
Certain attempts to control discharge location have been
made. Mohri et al. [99] controlled the discharge location by
sequentially applying one voltage pulse to each needle of a
bundled electrode. The bundled electrode is made of
copper needles which are tied together and electrically
insulated from each other. Luo et al. [100] and Kunieda et
al. [101] controlled the discharge location by applying a
high voltage pulse with a very short rise time to an arbitrary
point on the tool electrode around which discharge is
intended. Due to the internally existing inductance of the
tool electrode, which is small but not negligible, it takes a
finite time for the electric potential to transmit in the tool
electrode. Discharge is therefore forced to occur around
the place where the high voltage pulse is applied, because
the gap voltage is locally risen around there even though in
an extremely short period of time. Han et al. [102] utilized
this principle to WEDM as shown in Figure 36. The high
voltage pulse is applied either to the upper or to the lower
feeding brush. With the use of the discharge location
measurement, they demonstrated that the discharge
location can be controlled in the actual WEDM process.
It is common knowledge that, in conventional EDM, there
exists only one discharge point for each pulse. On the other
Superimposed to lower brush
Superimposed to upper brush
Probability
Without superimposition
0 40 80 100 160
Discharge location, mm
Wire electrode
Workpiece
Lower brush Upper brush
Figure 36: Control of discharge locations by
superimposing high voltage to feeding points in WEDM.
Discharges
L R
Conventional Multi spark
Figure 37: Principle of multi-spark EDM method.
hand, Figure 37 shows the basic circuit of the newly
developed Multi-spark EDM method [103], in which two
discharge points can be obtained for each pulse. The
circuit is comprised of a conventional pulse generator, a
workpiece and two tool electrodes, both of which are fixed
onto the quill of the same EDM machine. The pulse
generator is connected with one tool electrode, the
workpiece, and the other tool electrode serially in this order.
Accordingly, for each pulse, one discharge occurs in the
gap between one tool electrode and the workpiece, and
another discharge occurs at the same time in the gap
between the other tool electrode and workpiece. Hence the
removal rate and energy efficiency of the Multi-spark EDM
method are considerably higher than those of conventional
EDM. To balance the removal rates in both gaps, the
polarity of the pulse generator is changed adaptively to
equalize the gap voltages measured at both gaps.
4.8 Gap monitoring and control systems
In order to improve the machining rate, stability and quality,
extensive efforts have been made to develop advanced gap
monitoring and control systems. Discharge states can be
classified into five types, open circuit, normal discharge,
abnormal discharges (arcing 1 and arcing 2) and short
circuit as shown in Figure 38 [104, 105]. The ignition delay
time, td, is an important indicator of the isolation condition of
the discharge gap. A gap monitoring circuit was developed
to identify the five discharge states and output the ratios of
gap open, normal discharge, transient arcing, harmful
arcing and short circuit [106]. These ratios are used as input
parameters for EDM on-line control based on various
control strategies.
Adaptive control arose with the success of fundamental and
mathematical adaptive control theory and the development
of computer technology. A typical example is the steepest
ascent path algorithm as shown in Figure 39 [107]. In this
system, the pulse efficiency, i.e. the ratio of normal
discharge time to total pulse duration, is used as the
controlled parameters. The system adjusts on-line the
discharge interval, t0, and the control reference voltage
which is compared to the feedback gap voltage, while the
difference value of the comparison drives the servo
mechanism to maintain the average gap voltage at the level
80
Gap voltage [V]
td te to
40 ui
ti
ue
0
800 1600 2400 Time [μs]
Gap current [A]
4
2 ie is
0
Open Normal Arcing Arcing Short Time [μs]
circuit spark 1 2 circuit
Figure 38: Voltage and current waveforms of
the five basic EDM gap states [104, 105].
Discharge interval to [μs]
50 Start point
Arcing and short-circuit zone
30 Ridge Pulse efficiency: 20%
20 30%
40%
50%
60%
10 Steepest ascent path
70%
5 80%
3 Arcing zone
-2.0 -1.0 0 1.0 2.0
Servo reference voltage us [V]
Figure 39: Algorithm of steepest ascent path [107].
determined by the servo reference voltage, to seek the
highest value of the pulse efficiency in order to achieve the
maximum machining rate.
To reflect the dynamic behavior of the EDM process, a
dynamic modeling approach so called self-tuning control
was proposed [108]. In this model, the outputs which are
used to adjust the servo reference voltage and discharge
interval are based on the detection of the type of pulses that
occur in the gap (see Figure 38). During machining, the
outputs are maintained at the level of reference values
which are manually set according to given machining
conditions. To improve machining productivity, this model
has been simplified by linearizing the output value with the
machining gap size. During machining, the difference
between the output of the process and the output of the
reference model is controlled at a minimum value [109]. To
address the internal dynamic characteristics of the closed
loop servo system for stabilizing the machining process, the
adaptive servo system was modified into a self-tuning
regulator system [110].
When a deep and narrow cavity is machined, electrode
jumping operation is always used to retract the tool from the
machining area to improve the flushing condition. In the
adaptive jumping control system, the system output is
calculated based on the ratio of the sum of abnormal
discharge time to the total pulse on-time. When the
machining depth increases, the ratio of the jump-down time
in one jumping cycle is reduced to avoid harmful arcing [111,
112].
EDM lacks a direct link between the theoretical models of
material removal mechanism and industrial applications. In
many cases, empirical knowledge is helpful for deciding
machining parameters without explicit mathematical models.
Such expert knowledge is simply expressed as IF-THEN
rules to describe the relationship between the input and
output of system. Therefore, they are applied in on-line
control, off-line process planning and operator assistance.
There are many research results published on fuzzy logic
control either on sinking EDM or on wire EDM [113-118].
An artificial neural network consists of many neurons
arranged in different layers. Each neuron is connected to
every neuron in the previous and following layers with a
variable weight. The self-learning algorithm automatically
adjusts these weights based on training input-output sets.
Neural networks have been used to optimize EDM sinking
process to improve gap width control, harmful arc
prevention and high machining accuracy [119, 120]. They
are also used in WEDM to estimate the workpiece height
and maintain optimal and stable machining [121].
To integrate EDM within the next generation agile
manufacturing environment, a system with communication
function through internet has been developed. The Web
Home Page Browser window is used to browse information
on the Internet. The complete data during monitoring can be
recorded and displayed. The recorded data can then be
used for off-line analysis and comparison of the effects of
different machining conditions [122, 123].
5 MACHINING CHARACTERISTICS
5.1 Influence of pulse conditions
Figure 40 shows three machining characteristics which are
most important in EDM practice: material removal rate,
surface roughness and tool electrode wear ratio. Here the
tool electrode wear ratio is defined as the ratio of volume of
tool electrode wear to the volume of workpiece removal.
Requirements for any two of the characteristics can be
satisfied if the remaining one is sacrificed using the
discharge current waveforms indicated in Figure 40.
However, there is no current waveform which can satisfy all
the requirements.
Current waveforms with larger peak current and longer
Discharge
current
Removal rate
Time
Surface
Tool wear ratio
roughness
Figure 40: Relation between discharge current
waveforms and machining characteristics.
discharge duration result in higher material removal rates.
At the same time, low tool electrode wear can also be
satisfied because the carbon layer deposited on the anode
tool electrode is thicker when longer discharge durations
are used [21, 25-31]. However, surface roughness is not
good because the crater generated by each pulse
discharge is large. Thus this pulse condition is normally
used in rough machining.
On the other hand, the pulse condition with longer
discharge duration and lower peak current brings about
both lower tool electrode wear ratio and better surface
roughness. However, the lower heat flux due to the lower
current density at the discharge spot results in lower energy
efficiency of material removal compared with the pulse
condition with shorter discharge duration and higher peak
current even though the same pulse energy is applied. In
this situation, the melted and evaporated regions in the
workpiece are small and most of the energy distributed to
the workpiece is dissipated inside the workpiece due to
heat conduction. Obviously, the material removal per pulse
is small. In addition, longer discharge duration results in
lower discharge frequency per unit time. Thus this pulse
condition is suitable for finishing.
Finally, pulse conditions with shorter discharge duration
and higher peak current provide better surface roughness
due to smaller discharge crater. Moreover, although
material removal per pulse is not large, it is large enough
compared with the pulse condition with longer discharge
duration and lower peak current under the same discharge
energy per pulse because energy efficiency is higher due to
larger heat flux. Higher repetition rate of discharge per unit
time due to shorter discharge duration further increases the
material removal rate. However, the tool electrode wear
ratio is high because the carbon layer deposited on the tool
electrode is thin. This is why extremely short discharge
duration and high peak current compared with the case of
sinking EDM, for example 1µs and 100A, are used in
WEDM, where the problem of tool electrode wear is
insignificant.
5.2 Factors influencing tool electrode wear ratio
Influence of polarity
As shown in Figure 16, the energy dissipation into the
anode is greater than into the cathode. Nevertheless, in
sinking EDM, polarity of the tool electrode is normally
positive except when a very short discharge duration is
used. This is because the carbon layer which is deposited
on the anode surface due to thermal dissociation of the
hydrocarbon oil protects the anode surface from wear.
Since the carbon layer is thick when the discharge duration
is long, the tool electrode wear ratio is low with the polarity
of positive tool electrode under the pulse condition of
longer discharge durations. On the contrary, a negative tool
electrode is used considering the energy distribution in the
cases of finish machining and micromachining where
deposition of carbon layer is scarce and of WEDM in which
deionized water is normally used as dielectric liquid.
Influence of thermal properties of electrode materials
Since EDM is a thermal process, the influence of thermal
properties of electrode materials on the removal volume is
significant. When the heat flux from the arc column is equal,
higher heat conductivity results in lower temperature on the
electrode surface. Hence materials with higher heat
conductivity are suitable as tool electrodes. On the other
hand, under the same temperature distribution inside the
electrode, when the surface temperature is not over the
boiling point or melting point, removal does not occur.
Hence materials with higher melting point and boiling point
are also suitable as tool electrodes. The results of heat
conduction analysis shown in Figure 19 and 20 indicate
that although both melting point and boiling point of copper
are lower than those of steel, copper is more suitable as a
tool electrode because copper has a much higher heat
conductivity than steel. Thus, considering the higher
energy distribution to the anode, carbon layer deposition on
the anode, and the thermal properties of electrode
materials, it should be possible to select appropriate
discharge current waveforms, polarity, and tool electrode
materials to obtain minimum tool electrode wear ratio even
lower than several %.
6 CHALLENGES IN MINIATURIZATION
6.1 Strategy found in WEDG
EDM processes are widely used for the micro-machining of
fuel injection nozzles, spinneret holes for synthetic fibers,
electronics and optical devices, micro-mechanical parts,
and micro-tools for producing these devices. Figure 41
shows the Wire Electro-Discharge Grinding (WEDG)
method which was developed by Masuzawa et al. [124].
Using a wire electrode traveling along the groove of a wire
guide, a micro rod can be machined like in a turning
process. Then, on the same machine, by reversing the
polarity of the machined micro rod, micro holes of less than
10 µm in diameter can be drilled by EDM. The following are
the essential problems of micro EDM which were
elucidated and solved by Masuzawa [125] through the
development of the WEDG method.
On-machine forming of tool electrodes
When a micro tool electrode is purchased or made using
another machine in the machine shop and held by the
chuck of the micro EDM machine, the micro tool electrode
is always off-centered more or less from the rotation axis of
the spindle. In the WEDG machine, however, run-out of the
Workpiece
Dielectric liquid
Wire electrode
Wire guide
Figure 41: Wire electro-discharge grinding
(WEDG) method [124].
Pulse generator
Micro rod workpiece
Block tool electrode
Figure 42: On-the-machine forming of micro rod
using block electrode.
tool electrode is almost zero because a micro rod is first
made on the same machine by WEDG, followed by
reversing its polarity to use the micro rod as tool electrode
to machine micro holes without loosening the chuck.
Tool electrode wear
On-machine forming of micro rods is also possible using a
block tool electrode as shown in Figure 42. However, wear
of the block deteriorates the machining accuracy. In WEDG,
there is no problem of wear, because a renewable wire is
used as tool electrode. Since the wire is guided by the
groove of the wire guide, machining accuracy is not
deteriorated by the wire vibration unlike the WEDM
process.
Reaction force
As shown in Figure 29, the peak of the reaction force
generated by a single pulse discharge is remarkably large.
Use of a block electrode causes discharge even at the end
of the micro rod which is being formed, thus exciting
vibration of the micro rod. In WEDG, however, the bending
moment applied to the almost finished micro rod is
insignificant because discharge only occurs at the foot of
the rod being formed.
Relaxation type pulse generator
In micro EDM, the relaxation type pulse generator shown in
Figure 4 is still being used. If the transistor type is used, it
takes at least several tens of ns for the discharge current to
diminish to zero after detecting the occurrence of discharge
because the electric circuit for detecting the occurrence of
discharge, the circuit for generating an output signal to
switch off the power transistor and the power transistor
itself have a certain amount of delay time. Hence it is
difficult to keep the constant discharge duration shorter
than several tens of ns using the transistor type pulse
generator.
Stray capacitance
When the relaxation type pulse generator in Figure 4 is
used, capacitance of the capacitor should be decreased to
obtain smaller discharge energy per pulse. In the actual
EDM machine, however, stray capacitance exists between
the electric feeders, between the tool electrode holder and
work table, and between the tool electrode and workpiece.
Hence all the charge stored in the stray capacitance is
discharged to the working gap together with the charge
stored in the capacitor wired to the circuit as shown in
Figure 43. This means the minimum discharge energy per
pulse is determined by the stray capacitance. Hence
shortening of electric feeders and use of electrical insulator
for the tool electrode holder and work table materials are
effective for reducing discharge energy. In the final finishing,
when minimum discharge energy is necessary, the
capacitor is not wired and machining is conducted with the
stray capacitance only. Since the working surface area
between the tool electrode and workpiece in WEDG is
narrower than that in the block electrode method, smaller
stray capacitance of the working gap is advantageous to
WEDG for the limit of miniaturization.
Some attempts [126, 127] were made to replace the
relaxation type pulse generator with the transistor type
pulse generator in micro EDM, and a minimum discharge
R
C Stray capacitance
Tool electrode
Workpiece
Figure 43: Charge stored in both stray capacitance
and condenser is discharged.
duration of 30ns was achieved [126]. However, even if
future developments of electronics devices can further
reduce the delay time existent in the transistor type pulse
generator, the discharge energy can never be smaller than
the energy stored in the stray capacitance.
Masuzawa et al. conducted assembly [128], milling [129],
blanking [130], electrochemical discharge machining [131],
and ultrasonic machining [132], based on the
on-the-machine manufacturing of micro jigs, cutting tools,
blanking tools, tool electrodes, and horns using the WEDG
method.
6.2 Factors influencing limits of miniaturization in
micro EDM
Kawakami et al. [133] examined the limits of minimum
machinable size using the WEDG method. Cemented
tungsten carbide rods of 0.3mm in diameter with a WC
particle size of 0.4µm were used as the workpiece. The
relationship between the target rod diameter and machined
rod diameter was obtained as shown in Figure 44. Red
marks in Figure 44 indicate machining failure due to
breakage of the micro rod. It should be noted that micro
rods smaller than 2.3µm in diameter were not obtained.
Deviation in the machined rod diameter from the target rod
diameter was obviously caused by the positioning error and
thermal deformation of the equipment, and by the
inaccuracy of wire electrode diameter used in WEDG.
However, these are not the reasons why rod diameters
under 2.3µm were never obtained. If they were, one would
have obtained successful results in the vacant zone since
machining was repeated in great numbers. Therefore, in
this case, plausible reasons would be the size of the
discharge crater, sub-surface layer damaged by EDM,
residual stress, and material structure of the workpiece.
Machined rod diameter, μm
15
15
20μm
10
10
5
5 Failure
Vacant
zone
0
0
0 5
0 5
Target rod diameter, μm
Figure 44: Relationship between target rod diameter
and machined rod diameter (WC grain size : 0.4μm).
11.5μm
(a) C=3300 pF, open
voltage 110V
Figure 45: Discharge craters generated on tungsten anode.
10µm
Normal polarity:
workpiece positive
Figure 46: Surface of micro rods obtained by
WEDG (Cemented tungsten carbide, open
voltage 40V, stray capacitance only).
Discharge crater size
Figure 45(a) shows a discharge crater generated on the
tungsten anode surface when using a capacitance of
3300pF. Figure 45(b) shows a discharge crater generated
with stray capacitance only. Even with the minimum
discharge energy, diameter of the discharge crater
measures 2µm, indicating that the minimum discharge
crater size may influence on the limit of miniaturization. As
described in Section 5.2, under the pulse conditions used
in micro EDM, since the carbon layer is scarcely deposited
on the anode surface due to the extremely short discharge
duration, polarity of workpiece is normally positive because
energy distribution into anode is larger than that into
cathode. Thus Han et al. [134] compared the diameter of
craters generated on an anode and a cathode workpiece
and found that the diameter was 2µm and 0.8µm,
respectively (Figure 46). However, the minimum diameter
of the micro rod machined by WEDG using the reversed
polarity was almost the same as that using the normal
polarity. This was probably because the machining time
was about eight times longer than the normal polarity.
Since discharge energy is 1/2(Cu02), decreasing the open
voltage u0 is effective for reducing the crater size. Thus
Egashira et al. [135] decreased the voltage of the power
source to 20V and obtained the minimum rod diameter of
1µm as shown in Figure 47.
Residual stress
There are two kinds of residual stress which may influence
the limit of miniaturization. One is residual stress already
present prior to machining, and the other is residual stress
caused by EDM.
Residual stress already present prior to machining, for
example, can be found in tungsten rods which are
produced by plastic deformation using the wire drawing
process. The residual stress in the axial direction is not
distributed uniformly in the radial direction. When this
material rod is held by the chuck of the spindle of an EDM
Success
Target value
10 15
10 15
Figure 47: Tungsten rod of 1μm diameter fabricated by
WEDG (Open voltage: 20V, stray capacitance) [135].
300µm 300µm
2μm
Before annealing
Without After annealing at 1000K
(b) Stray capacitance
only, open voltage 40V Figure 48: Influence of residual stress already present prior to
machining (tungsten rod, C=3300pF, open voltage 110V).
10µm
Reversed polarity:
workpiece negative
Figure 49: Tooth edge deviation due to
tensile residual stress [136].
machine, the center of rotation does not always coincide
with the center of the material rod. Therefore the micro rod
is generally formed off-centered from the center of the
material rod, and this residual stress causes deformation of
the micro rod machined as shown in Figure 48 [133].
Since EDM is a thermal process, tensile residual stress is
generated in the re-solidified layer. Spur et al.[136], Kruth et
al.[137], and Klocke et al. [138] reported that thin fins cut by
wire EDM may bend due to the residual stress caused by
the EDM process when fin thickness is less than 0.1mm in
rough cutting (Figure 49).
Influence of material structure
Kawakami et al. [133] examined the influence of the grain
size of tungsten carbide (WC) in cemented tungsten
carbide material on the limit of miniaturization. Since the
minimum diameter obtained using cemented tungsten
carbide with WC grain size of 0.4µm was 2.3µm as shown
in Figure 44, change in the limit was examined using grain
size larger than 0.4µm. The results showed that vacant
zone with WC grain size of 2µm was wider than that in
Figure 44 and the minimum machinable rod diameter
increased to 3µm.
Almond et al. [139] examined the influence of voids existing
in the workpiece material prior to machining on micro EDM.
Han et al. [134] examined the sub-surface layer damaged
by micro EDM. Kawakami et al. [133] compared the
machining characteristics of micro EDM between
poly-crystal and mono-crystal tungsten. The minimum
machinable thickness of a micro wall was thinner when
mono-crystal tungsten was used compared with
poly-crystal tungsten. However, since cracks were
generated parallel to the {100} planes, it is not always true
that mono-crystal is more suitable for miniaturization than
poly-crystal. They also found that the machining speed
depends on the crystal orientation of mono-crystal tungsten,
indicating that micro EDM may not be a pure thermal
process.
6.3 Advancements in micro EDM
The use of a single tool electrode has limits in throughput
and precision because of positioning error and tool wear.
Thus high aspect ratio electrode arrays were fabricated by
the LIGA process [140, 141]. A LIGA-fabricated array of
400 Cu electrodes with 20µm diameter was used to
machine through-holes in 50µm thick stainless steel
Figure 50: LIGA fabricated electrodes [141].
Wire electrode Cylindrical electrode
n=1： WEDG Plate electrode
n=2： Pattern machining
n=3： Machining multi-electrode
n=4： Multi-hole machining
Figure 51: μEDMn method [142]
(Figure 50). The machining time was about 5 minutes,
which is 600 times less than that required for serial
machining by a single electrode. Arrayed electrodes of
even complicated cross section shapes, like hexagonal and
gear, can be fabricated using LIGA with a high aspect ratio.
Figure 51 shows another novel approach to improve the
throughput in micro EDM [142]. In the first step (n=1), a
single micro cylindrical electrode is made by WEDG. In the
second step (n=2), a plate electrode is perforated to have a
pattern of holes using the cylindrical electrode made in the
first step. In the third step (n=3), using the plate electrode
as tool electrode, the pattern is replicated to a block
workpiece. In the next step (n=4), the workpiece is used as
tool electrode to make many patterns of holes precisely and
efficiently. After this, steps of 3 and 4 may be repeated to
obtain numerous numbers of holes.
Figure 52 shows the simultaneous processing method for
micro rods and holes developed by Yamazaki et al. [143].
With the polarity of the rod electrode positive, which is
reverse to the normal micro EDM, the rod electrode is
rotated and fed into a plate electrode at a feed rate of ten
times higher than the normal one. Since the circumferential
region of the rod electrode wears at a much higher rate
than the center, the end of the rod electrode is sharpened
in just a few minutes. At the same time, a tapered micro
hole is generated in the plate electrode corresponding to
the conical shape of the rod electrode. Thus a pair of micro
rod
（＋）
plate 50μm
（－） rod rod plate
０ ２ ４ ６ ８
min ＜before＞ ＜after＞
Time
Figure 52: Simultaneous processing of micro rod and hole.
Figure 53: EDMing micro rods by self-drilled holes.
(a) φ4μm cemented (b) φ5μm hole machined
carbide micro rod by formed rod in Cu plate
100μm in thickness
Figure 54: Micro rod formed by self-drilled hole
method and micro hole machined by formed rod
(Open voltage:70V，C: stray capacitance).
Inlet Outlet
(a) Self-formed slots
150 μm
150 μm
(b) Formed rod
150 μm
Inlet Outlet
(c) Hole machined by formed rod
Figure 55: Irregular hole made by self-drilled hole
method (open voltage: 110V，capacitance: 100pF).
rod and micro hole can be machined in a single process
without positioning control in the radial direction.
Figure 53 shows a simple but novel idea proposed by
Yamazaki et al. [144] to machine micro rods and holes
without using any special equipment like WEDG. With the
polarity of a rod electrode negative, the rod electrode is
rotated and fed into a plate electrode to make a hole. After
the rod electrode returns to its initial position, the rod
electrode is moved from the center of the hole at a certain
distance. Next the polarity of the rod electrode is reversed,
and then the rod electrode is fed into the plate electrode
either with or without rotation. Positioning of the rod
electrode in reference to the tool electrode is not necessary
because self-drilled holes are utilized as the tool electrodes.
Figure 54 shows a micro rod of 4µm in diameter formed by
this method. Using this rod electrode as tool electrode,
micro holes with a diameter of 5µm was perforated on a
copper plate. In Figure 55, a cross-shaped right-angled slot
was formed using a rod electrode. Positioning the rod
electrode on each corner of the cross-shaped slot without a
rotation, EDM was performed to form the cross-shaped rod
with reversed polarity. Using this cross-shaped electrode,
cross-shaped holes were formed on a plate.
7 INNOVATIVE EDM TECHNOLOGIES
7.1 Electrical discharge coating and alloying
EDM is undeniably classified as a removal process.
However, the EDM process can also be used as a surface
treatment method and/or an additive process. White layers
generated on a steel workpiece machined by EDM using
hydrocarbon dielectric have a higher carbon content than
the base material and hence shows increased resistance to
abrasion and corrosion [56, 64, 97, 145]. Under the high
temperature of the discharge column, the white layer can
absorb carbon from the gases formed in the discharge
column from the hydrocarbon dielectric. Since WEDM
normally takes place in deionized water, a discharge
column with a high oxygen and hydrogen content is built up.
In particular the oxygen causes decarbonization of the
material metal surface, thus increasing the density of alloys,
which are generally chromium and nickel. This also
increases the corrosion-resistance of the surface [64].
Narumiya et al. [97] reported that the workpiece machined
using a powder (Si, SiC or Al) suspended oil shows not only
improved surface roughness but also increased corrosion
resistance compared to the surface machined without
powder. Masui et al. [146] added fine tungsten powder in
the dielectric fluid to form an alloyed layer and found that
great improvements in surface roughness, hardness,
Servo
Hydrocarbon
Ti electrode
dielectric Power
supply
Hard layer
Workpiece
Figure 56: Principle of EDM surface modification [148].
TiC layer (6µm)
S45C
Figure 57: Surface modification of carbon steel
by TiC semi-sintered electrode [149].
corrosion resistance, and tribological properties can be
obtained when the polarity of the workpiece is positive.
Mohri et al. [147] and Goto et al. [148] found that alloying
speed can be improved significantly by using partially
sintered or green compacted materials such as Cu, Al,
WC-Co, and Ti as tool electrode. This is because the wear
ratio of these materials is high due to the low thermal
conductivity of these materials. As shown in Figure 56 [148],
a hard material (TiC) layer is coated on the workpiece
surface since Ti is carbonized in the hydrocarbon oil. Using
the same setup as the usual EDM process, harder and
thicker layer coating with a higher bonding strength can be
conducted in a shorter time compared with the
conventional methods like CVD and PVD. Figure 57 shows
a cross-section of a TiC layer formed on carbon steel in
eight minutes [149]. Applications of this method to cutting
tools and blanking tools show that tool life becomes
significantly longer [148, 150]. Various types of powder and
compact were proposed depending on workpiece materials
and alloyed layers for the applications to steel rolls, molds,
dies, cutting tools, turbines and aeroengine components
[151-153]. To make this process possible, a negative
polarity of the tool electrode and a long discharge duration
is used. Under these conditions, as described in Section
3.3, the material removal rate of the cathode (tool
electrode) is higher than that of the anode and carbon
coming from the hydrocarbon dielectric is deposited on the
anode (workpiece). The worn tool electrode material (Ti)
and the deposited carbon yields the TiC coating.
Hayakawa et al. [154] fabricated a micro structure shown in
Figure 58 using EDM deposition in air. The suitable
discharge conditions for this process were predicted from
the transient temperature analysis of the tool electrode and
workpiece. To enhance the wear of the tool electrode, its
polarity was set positive, which is opposite to the above
coating and alloying methods, because removal rate of the
anode is higher than that of the cathode in air.
7.2 Coloring of Titanium alloy using EDM process
Minami et al. [155] developed a new method of coloring
titanium alloys using the WEDM process. Since deionized
water is normally used in WEDM, an oxide layer is formed
over the surface of the anode workpiece due to electrolysis.
It is known that the surface of titanium alloy and stainless
steel can be colored by anodic oxidation due to the
interference of light in the oxide film formed by electrolysis.
Tool electrode (+)
deposit
Work (-) Work (-) Work (-)
Figure 58: Fabrication of micro-structure using EDM deposition
(Tool and workpiece: carbon steel, medium: air) [154].
－
W ire e le c tro d e
Incident beam Interfering beams
W o rk p ie c e
(T ita n iu m a llo y)
＋
air P ro c e s s in g
d ire c tio n
h TiO2 film
Ti
absorption 1 s t-c u t s u rfa c e C o lo re d s u rfa c e
Interference of light
Figure 59: Coloring of titanium alloy with EDM [155].
Hence an arbitrary color is given to the WEDMed surface
during the finish cutting process as shown in Figure 59. The
color can be changed by controlling the feed rate of the
wire electrode or varying the open voltage because the
color is determined by the oxide layer thickness. Using a
simple electrode controlled by an NC servo system in
sink-EDM, fine and multicolored pattern can be drawn on
titanium products [156].
7.3 EDM of non-conducting materials
It has been believed that only electrically conductive
materials can be machined by EDM. Diamond however can
be machined by means of EDM by providing diamond with
a conducting coating of graphite by heating the diamond in
a non-oxidizing flame or in a pyrolitic atmosphere
comprising a carbon compound until the temperature of
conversion to graphite is reached. When discharge occurs
between the tool electrode and the graphite coating, the
temperature at the discharge spot increases above the
conversion temperature of diamond-graphite, the bottom of
the crater is coated by newly formed graphite, hereafter the
process is repeated [157, 158].
Fukuzawa et al. [159-161] found a method for EDMing
completely nonconductive ceramics. As shown in Figure 60,
a metal plate or mesh is placed on the ceramics to be
machined. In the first stage, discharges generate between
the metal plate and tool electrode. While the metal plate is
being eroded, the working oil is thermally decomposed and
the pyrolitic carbon is deposited on the workpiece whose
polarity is positive. This is the same phenomenon of carbon
deposition on the tool electrode observed in the usual die
sinking, which was described in Section 3.6. Even after the
metal plate has eroded out, the surface of the ceramics is
covered with the carbon layer which is electrically
conductive. Hence, discharge continues to occur over the
ceramics as long as the deposition of carbon due to the
discharge is maintained. This method enables EDM
machining of even 3D shapes of Si3N4, ZrO2, SiC, Al2O3,
AlN, glass, old ceramics, and diamond. Figure 61 [161]
shows an example of machining a chair-shaped product of
Si3N4 using WEDM.
Taniguchi et al. [162] used a micro wave to machine
alumina ceramics and diamond. These workpieces were
placed in the gap between a pair of needle electrodes
whose axis was oriented in parallel to the electric field of
the micro wave standing in a micro wave tube. Alumina
ceramic was melted due to the dielectric heating inside the
Linear variable differential
transformer
Recorder
Tool electrode
EDM power
(-)
supply
Digital
oscilloscope
Ceramics
Assisting
electrode (+)
Figure 60: Schematic illustration of EDM for insulating
ceramics with an assisting electrode method [159].
Figure 61: Machining of insulating ceramics by
wire EDM (workpiece: Si3N4 ) [161].
material,whereas diamond was evaporated due to the heat
flux from the discharge column generated between the tip
of the needle electrode and the diamond surface because
the dielectric hysteresis loss of diamond was insignificant.
Uno et al. [163], Kawada et al. [164] and Luo et al. [165]
reported that one of the difficulties encountered in the EDM
slicing of silicon wafers is the large voltage drop caused by
the high electric resistivity of the silicon single crystal, while,
Kawada et al. reported that this large voltage drop also
results from the high contact resistance at the interface
between the silicon wafer and metal electric feeder.
Generally, even at the interface between contacting smooth
plane plates, the area of real contact is very narrow and the
voltage drop is significant due to the highly concentrated
electric current density in the area of real contact. Hence,
the voltage drop at the interface between metal and
high-electric-resistivity material is enormous, whereas that
in a metal-to-metal contact can be ignored compared with
the voltage drop in the discharge channel. They have
therefore demonstrated that the machining rate can be
improved by forming a low electric resistivity layer over the
surface of the silicon wafer. On the other hand, the contact
resistance is caused not only by the small area of real
contact but by the difference in the work function between
materials in contact which creates the Schottky barrier. Luo
et al. [165] and Kunieda et al. [166] eliminated the effects of
the Schottky barrier in metal-silicon contact by
electroplating the silicon surface with metal like Ni, Al and
Sb-Au to convert the rectifying contact to an ohmic one and
reduce the contact resistance in the EDM of the silicon
single crystal.
7.4 Dry EDM
Based on the consideration described in Section 3.7,
Kunieda et al. [58, 167-169] developed the dry EDM
method shown in Figure 62. A high velocity gas flow is
supplied through a thin-walled pipe electrode, causing
molten workpiece material to be removed and flushed out
of the working gap without being reattached to the
electrode surfaces. During the pulse interval, the gas jet
blows off the plasma formed by the previous pulse
discharge and decreases the temperatures of the
discharge spots. Environmentally, this process is clean
because there is no generation of hazardous gas nor waste
from the dielectric liquid. There is no risk of fire hazards
because no flammable dielectric is used. NTIS Tech Note
[170] reports briefly that in NASA holes in complicated
hardware were drilled successfully by a dry EDM process
that uses a tubular copper electrode and argon or helium
gas as the dielectric medium. However, NASA does not
disclose any further information. Advantages of the dry
EDM are as follows:
- Using oxygen gas for the machining of steel, even
higher material removal rate than the conventional
EDM can be obtained due to the heat generation by
the oxidation of steel.
- Tool electrode wear is almost negligible for any
discharge duration.
Pipe tool electrode Oxygen gas
Guide
Debris
Workpiece
Figure 62: Dry EDM milling.
- Heat affected zones such as the white layer are thin
and micro cracks are few.
- Process reaction force is negligible.
- Gap is narrower than conventional EDM.
- There is no corrosion which is caused by electrolysis
when deionized water is used as dielectric fluid.
Hence dry EDM is also suitable for WEDM and micro EDM
owing to the negligible process reaction force and the
minimized surface damage. Furudate et al. [171,172] found
that, when it is applied to finish-cutting in WEDM (Figure
63), straightness and geometrical accuracy of the finished
surface are excellent owing to the significantly reduced
vibration of the wire electrode due to the considerably
smaller process force compared with that in conventional
WEDM.
Wire electrode
Workpiece
Figure 63: Dry WEDM.
7.5 EDM of non-linear holes
In plastic injection molding and aluminum die-casting, it is
necessary to cool the mold surface in order to reduce the
forming cycle time. For this purpose, channels for coolant
should be provided uniformly close to the curved surfaces
of the mold. However, it is difficult to machine curved holes
using mechanical processes such as cutting, grinding, etc.
Thus several novel electrical machining methods were
proposed for digging curved holes [173-176]. A coil spring
which was initially deformed to the curvature of the hole to
be machined was inserted into the hollow space of a
straight pipe to guide the coil. A spherical tool electrode is
attached to the end of the coil spring. First, the workpiece is
machined to a certain depth by feeding both the coil spring
and the guide pipe together. Then the position of the guide
pipe is fixed and the coil spring only is fed. Figure 64 shows
the curved hole thus obtained by Goto et al. [174]. An
electrochemical machining method was also developed by
Uchiyama et al. [176].
Figure 64: Curved Tunnel machined by EDM [174].
[170].
8 CONCLUSIONS
EDM processes involve transient phenomena occurring in
a narrow space (order of micrometers) and in a short period
of time (order of microseconds). These EDM phenomena
are not in thermal equilibrium, but include transitions
between solid, liquid, gas, and plasma, chemical reactions,
mass transfer, and displacement of boundaries. Hence,
compared to other discharge phenomena such as glow
discharge in dry etching processes and arc discharge in
welding processes, physics involved in EDM processes are
obviously most complicated, rendering observation and
theoretical analysis extremely difficult. Although it is a long
way to fully understand the EDM phenomena, thanks to
advancements of various surface analyzing equipments,
microscopes, high-speed imaging devices, and software
tools for numerical analysis, fundamental studies are
progressing step by step, and some of the common EDM
knowledge which have been accepted for a long time are
being modified. On the other hand, given that numerous
discharge gap phenomena have yet to be fully understood,
the potentials of EDM technology may not be fully realized.
Undoubtedly, continuous efforts are expected to lead to
further developments of the EDM technology in the future.
9 ACKNOWLEDGEMENTS
The authors would like to acknowledge the following
colleagues who contributed to this paper: Prof. D. Allen, Dr.
H. Altena, Dr. P. Bleys, Dr. A. Goto, Dr. S. Hayakawa, Dr. T.
Ishida, Prof. F. Klocke, Prof. J. Kozak, Prof. J.P. Kruth, Prof.
G. N. Levy, Dr. D. Lung, Mr. T. Masaki, Dr. K. Masui, Prof. T.
Masuzawa, Prof. T. Matsuo, Dr. H. Minami, Prof. N. Mohri,
Prof. W. Natsu, Prof. Y. Takeuchi, Prof. Y. Uno, Prof. Z.Y.
Yu.
10 REFERENCES
[1] Lazarenko, B.R., 1943, To invert the effect of wear on
electric power contacts”, Dissertation of The All-Union
Institute for Electro Technique in Moscow/CCCP (in
Russian).
[2] Lazarenko, B.R., 1943, SU-Pat 70010 / IPC B23p /
Priority: 3.4.1943.
[3] Meek J. M., Craggs J. D., 1978, Electrical Breakdown
of Gases, John Wiley & Sons.
[4] Bommeli B., Frei C., Ratajski A., 1979, On the
Influence of Mechanical Perturbation on the
Breakdown of a Liquid Dielectric, J. of Electrostatics, 7,
123-144.
[5] Altpeter, F., Perez, R., 2004, Relevant Topics in Wire
Electrical Discharge Machining Control, J. of Materials
Processing Technology, Vol. 149, Issues 1-3, 147-151.
[6] Zingerman, A. S., 1956, Propagation of a Discharge
Column, Soviet Physics-Technical Physics, 1 (1), 5,
992-996.
[7] Snoeys, R., Van Dijck, F., 1971, Investigations of EDM
Operations by Means of Thermomathematical Models,
Annals of the CIRP, 20, 35-36.
[8] Snoeys, R., Van Dijck, F., 1972, Plasma Channel
Diameter Growth Affects Stock Removal in EDM,
Annals of the CIRP, 21, 39-40.
[9] Eckman P.K., Williams E.M., 1960, Plasma Dynamics
in Arc Formed by Low-Voltage Sparkover of a Liquid
Dielectric, Appl. Sci. Res., Section B, 8, 299-320.
[10] Ikeda, M., 1972, The Movement of a Bubble in the
Gap Depending on the Single Electrical Discharge
(1st report): J. JSEME, 6, 11, 12-25 (in Japanese).
[11] Zolotykh, b. n., 1959, The Mechanism of Electrical
Erosion of Metals in Liquid Dielectric Media, Soviet
Physics-Technical Physics, 4, 12, 1370-1373.
[12] Hockenberry T.O., Williams E.M., 1967, Dynamic
Evolution of Events Accompanying the Low-Voltage
Discharge Employed in EDM, IEEE Trans. on Industry
and General Applications, IGA-3, 4, 302-309.
[13] Yoshida M., Kunieda M., 1998, Study on the
Distribution of Scattered Debris Generated by a Single
Pulse Discharge in EDM Process, IJEM, No.3, 39-47.
[14] Lee, T.H., 1959, T-F theory of Electron Emission in
High-Current Arcs, J. of Appl. Phys., 30, 2, 166-171.
[15] Eubank P.T., Patel M.R., Barrufet M.A., Bozkurt B.,
1993, Theoretical Models of the Electrical Discharge
Machining Process III. The Variable Mass, Cylindrical
Plasma Model, J. Appl. Phys., 73 (11), 1, 7900-7909.
[16] DiBitonto D.D., Eubank P.T., Patel M.R., Barrufet M.A.,
 1989, Theoretical Models of the Electrical Discharge
Machining Process I. A Simple Cathode Erosion
Model, J. Appl. Phys., 66 (9), 1, 4095-4103.
[17] Patel M.R., Barrufet M.A., Eubank P.T., DiBitonto D.D.,
1989, Theoretical Models of the Electrical Discharge
Machining Process II. The Anode Erosion Model, J.
Appl. Phys., 66 (9), 1, 4104-4111.
[18] Hayakawa S., Kunieda M., 1996, Numerical Analysis
of Arc Plasma Temperature in EDM Process Based on
Magnetohydrodynamics, Trans. JSME (B), 62, 600,
263-269 (in Japanese).
[19] Hayakawa S., Xia H., Kunieda M., Nishiwaki N., 1996,
Analysis of Time Required to Deionize an EDM Gap
during Pulse Interval, Proc. of Symposium on
Molecular and Microscale Heat Transfer in Materials
Processing and Other Applications, 368-377.
[20] Hayakawa, S., Yuzawa, M., Kunieda, M., Nishiwaki,
N., 2001, Time Variation and Mechanism of
Determining Power Distribution in Electrodes during
EDM Process, IJEM, 6, 19-26.
[21] Xia H., Kunieda M., Nishiwaki N., 1996, Removal
Amount Difference between Anode and Cathode in
EDM Process, IJEM, 1, 45-52.
[22] Motoki M., Hashiguchi K., 1967, Energy Distribution at
the Gap in Electric Discharge Machining, Annals of
the CIRP, 14, 485-489.
[23] Van Dijck, F., 1973, Physico-Mathematical Analysis of
the Electro Discharge Machining Process,
Dissertation of Katholieke Universiteit Leuven.
[24] Koenig, W., Wertheim, R., Zvirin, Y., Toren, M., 1975,
Material Removal and Energy Distribution in Electrical
Discharge Machining, Annals of the CIRP, 24, 1,
95-100.
[25] Xia H., 1995, Study on Factors Affecting Electrode
Wear Ratio and Improvement of Machining
Characteristics in EDM Process, Dissertation of Tokyo
University of Agriculture and Technology (in
Japanese).
[26] Xia H., Hashimoto, H., Kunieda M., Nishiwaki N.,
1996, Measurement of Energy Distribution in
Continuous EDM Process, J. of JSPE, 62, 8,
1141-1145 (in Japanese).
[27] Natsu W., Kunieda M., Nishiwaki N., 2004, Study on
Influence of Inter-electrode Atmosphere on Carbon
Adhesion and Removal Amount, IJEM, No. 9, 43-50.
[28] Motoki, M., Lee, C., Tanimura, T., 1967, Research on
electrode Erosion Caused by Transient Arc Discharge
in Dielectric Liquid, J. IEEJ, 87-943, 793-801 (in
Japanese).
[29] Ikai, T., Hashiguchi, K., 1988, On the Tool Electrode
Material with Low Erosion in the Electric Discharge
Machining, T. IEEJ. D, 108-3, 338-343 (in Japanese).
[30] Mohri N., Suzuki M., Furuya M., Saito N., 1995,
Electrode Wear Process in Electrical Discharge
Machining, Annals of the CIRP, 44/1, 165-168.
[31] Kunieda M., Kobayashi T., 2004, Clarifying
Mechanism of Determining Tool Electrode Wear Ratio
in EDM Using Spectroscopic Measurement of Vapor
Density, J. Materials Processing Technology, 149,
284-288.
[32] Palatnik, L.S., Liulichev, A.N., 1958, Investigation of
the Temperature in the Vapor Phase Occurring during
the Electrical Spark Treatment, Soviet
Physics-Technical Physics, 1, 818-824.
[33] Albinski K., Musiol K., Miernikiewicz A., Labuz S.,
Malota M., 1995, Plasma Temperature in Electro
Discharge Machining, Proc. ISEM XI, 143-152.
[34] Hashimoto H., Kunieda M., 1997, Spectroscopic
Analysis of Temperature Variation of EDM Arc Plasma,
J. JSEME, 31, 68, 32-40 (in Japanese).
[35] Natsu W., Ojima S., Kobayashi T., Kunieda M., 2004,
Temperature Distribution Measurement in EDM Arc
Plasma Using Spectroscopy, JSME International
Journal C, 47, 1, 384-390.
[36] Carslaw, H.S., Jaeger, J.C., 1959, Conduction of Heat
in Solids, Oxford Science Publications.
[37] Pittaway, L.G., 1964, The Temperature Distributions in
Thin Foil and Semi-Infinite Targets Bombarded by an
Electron Beam, Brit. J. Appl. Phys., 15, 967-982.
[38] Kunieda M., Xia H., Nishiwaki N., 1992, Observation
of Arc Column Movement during Monopulse
Discharge in EDM, Annals of the CIRP, 41, 1,
227-230.
[39] Fujimoto, R., Tojima, S., 1973, Observation of
Electrode Erosion Phenomena, J. of JSEME, 6, 11,
48-42 (in Japanese).
[40] Kunieda, M., 2005, Fundamentals and Future in
Electrical Discharge Machining – II Future Prospect -,
JSPE, 71, 2, 189-194 (in Japanese).
[41] Saito, N., Kobayashi, K., 1967, Machining Principle
and Characteristics of Electric Discharge Machining,
Mitsubishi Denki Giho, 41, 10, 1222-1230 (in
Japanese).
[42] Uno, Y., Okada, A., Nakanishi, H., Guo, C., Okamoto,
Y., Takagi, T., 1998, EDM Characteristics of
CVD-Carbon Electrode, IJEM, 3, 19-24.
[43] Suzuki K., Sharma A., Iwai M., Uematsu T., Shoda K.,
Kunieda M., 2004, Electrical Discharge Machining
Using Electrically Conductive CVD Diamond as an
Electrode, New Diamond and Frontier Carbon
Technology, 14, 1, 35-44.
[44] Zingerman, A. S., 1956, The Effect of Thermal
Conductivity upon the Electrical Erosion of Metals,
Soviet Physics-Technical Physics, 1, 2, 9, 1945-1958.
[45] Rich J.A., 1961, Resistance Heating in the Arc
Cathode Spot Zone, J. Appl. Phys., 32, 6, 1023.
[46] Saeki T., Kunieda M., Ueki M., Satoh Y., 1996,
Influence of Joule Heating on EDM Processes of
High-Electric-Resistivity Materials, ASME,
HTD-Vol.336, 95-103.
[47] Koenig, W., Dauw, D. F., Levy, G., Panten, U., 1988,
EDM-Fiture Steps towards the Machining of Ceramics,
Annals of the CIRP, 37, 2, 623-631.
[48] Zsigmondy R., 1905, Zur Erkenntis der Kolloide,
Fisher Verlag, Jena.
[49] Kumar A., Roy R., 1988, RESA-A Wholly new process
for fine oxide powder preparation, J. Mater. Res. 3(6),
1373-1377.
[50] Martin, P.J., Filipczuk, S.W., Netterfield, R.P., Field,
J.S., Whitnall, D.F., Mckenzie, D.R., 1988, Structure
and Hardness of Diamond-like Carbon Films Prepared
by Arc Evaporation, J. Mater. Sci. Lett., 7, 410-412.
[51] Iijima, S., 1991, Helical Microtubules of Graphitic
Carbon, Nature, 354, 7, 56-58.
[52] Lauwers B., Kruth J.P., Liu W., Eeraerts W., Schacht
B., Bleys P., 2004, Investigation of the material
removal mechanisms in EDM of composite ceramic
materials, Journal of Materials Processing Technology,
Vol. 149, Issues 1-3, 347-352.
[53] Lauwers, B., Liu, W., Kruth, J.P., Vleugels, J., Jiang,
D., Van der Biest, O., 2005, Wire EDM machining of
Si3N4, ZrO2 and Al2O3-based ceramics, IJEM, 10,
33-38.
[54] Toenshoff, H.K., Egger R., Klocke, F., 1996,
Environmental and Safety Aspects of Electrophysical
 and Electrochemical Processes, Annals of the CIRP,
45, 2, 553-568.
[55] Kusano T., Yoshida M., Kunieda M., 1993, Process
Monitoring of EDM by Measuring Volume of Gas
Generated in Discharge Gap, Proc. ’93 Annual
Meeting of JSEME, 87-90 (in Japanese).
[56] Kruth, J.-P., L. Stevens, Froyen, L., Lauwers, B., 1995,
Study on the White Layer of a Surface Machined by
Die Sinking Electro Discharge Machining, Annals of
the CIRP, 44, 1, 169-172.
[57] Kunieda M., Furuoya S., 1991, Improvement of EDM
Efficiency by Supplying Oxygen Gas into Gap, Annals
of the CIRP, 40/1, 215-218.
[58] Kunieda M., Yoshida M., 1997, Electrical Discharge
Machining in Gas, Annals of the CIRP, 46/1, 143-146.
[59] Masuzawa, T., Tanaka, K., 1983, Water-Based
Dielectric Solution for EDM, Annals of the CIRP, 32/1,
119-122.
[60] Koenig, W., Joerres, L., 1987, Aqueous Solutions of
Organic Compounds as Dielectrics for EDM Sinking,
Annals of the CIRP, 36/1, 105-109.
[61] Masuzawa T., Tsukamoto J., Fujino M., 1989, Drilling
of Deep Microholes by EDM, Annals of the CIRP, 38/1,
195-198.
[62] Yu Z., Kunieda M., 1999, Study on Material Removal
Rate of EDM in Deionized Water, J. JSEME, 33, 72,
28-36 (in Japanese).
[63] Leao, F.N., Pashby, I.R, 2004, A Review on the Use of
Evironmentally-Friendly Dielectric Fluids in Electrical
Discharge Machining, Journal of Materials Processing
Technology, Vol. 149, Issues 1-3, 341-346.
[64] Schumacher, B.M., 1983, EDM Technology for
Precision Workpieces with Excellent Surface Quality,
Proc. ISEM 7, 124-135.
[65] Miyajima J., Kunieda M., Masuzawa T., 1987,
Observation of Bubbles generated in EDM Gap,
Proc. ’87 Spring Meeting of JSPE, 723-724 (in
Japanese).
[66] Imai Y., Hiroi M., Nakano M., 2001, Investigation of
EDM Machining States Using Ultrasonic Waves, ISEM
13, 109-116.
[67] Tanimura T., Isuzugawa K., Fujita I., Iwamoto A.,
Kamitani T., 1989, Development of EDM in the Mist,
Proc. ISEM 9, 313-316.
[68] Karasawa T., Kunieda M., 1990, EDM Capability with
Poured Dielectric Fluids without a Tab, Bull. JSPE, 24,
3, 217-218.
[69] Natsu, W., Kunieda, M. Nishiwaki, N., 2004, Study on
Influence of Inter-electrode Atmosphere on Carbon
Adhesion and Removal Amount, IJEM, No. 9, 43-50.
[70] Kunieda M., Adachi Y., Yoshida M., 2000, Study on
Process Reaction Force Generated by Discharge in
EDM, Proc. of 2nd Int'l Conf. MMSS, 313-324.
[71] Ikeda, M.,1972, The Movement of a Bubble in the
Gap Depending on the Single Electrical Discharge
(1st report), J. JSEME, 6, 11, 12-26 (in Japanese).
[72] Kunieda M., Tohi M., Ohsako Y., 2003, Reaction
Forces Observed in Pulse Discharges of EDM, IJEM,
No.8, 51-56.
[73] Yanagihara, N., 1980, New Measuring System of
Impact Force - Development and Application to the
High Speed Blanking Test - , Bulletin of JSME, 23, 175,
140-144.
[74] Dekeyser, W. L., Snoeys, R., 1989, Geometrical
Accuracy of Wire-EDM, Proc. of ISEM 9: 226-232.
[75] Obara, H., Ishizu, N., Kawai, T., Ohsumi, T., Hayashi,
T., 2000, Simulation of Wire EDM (3rd Report),
JSEME, 34,75: 30-37(in Japanese).
[76] Panschow, 1974, Ueber die Krafte und ihre
Wirkungen beim Electroerosiven Schneiden mit
Ablaufender Drahtelektrode, Dissertation T.U.
Hannover, BRD.
[77] Masuzawa T., Sata T., Kinoshita N., 1971, The Role of
the Chips in Micro-EDM, J. JSPE, 37, 9, 52-58.
[78] Suda, T., Sata, T., 1974, Movement of Conductive
Particles in EDM Gap, J of JSEME, 7, 14, 19-28.
[79] Schumacher B.M., 1990, About the Role of Debris in
the Gap During Electrical Discharge Machining,
Annals of the CIRP, 39/1, 197-199.
[80] Kunieda M., Yanatori K., 1997, Study on Debris
Movement in EDM Gap, IJEM, No.2, 43-49 .
[81] Kunieda M., Nakashima T., 1998, Factors
Determining Discharge Location in EDM, IJEM, No.3,
53-58.
[82] Kunieda M., Kojima H., 1990, On-line Detection of
EDM Spark Locations by Multiple Connection of
Branched Electric Wires, Annals of the CIRP, 39/1,
171-174.
[83] Obara H., 1992, Detection of Discharging Position on
Wire EDM, Proc. ISEM 10, 404-409.
[84] Han F., Kunieda M., 2001, Chaos Found in
Distribution of EDM Spark, Proc. of ISEM XIII,
185-192.
[85] Rasband, S.N., 1990, Chaotic Dynamics of Nonlinear
Systems, John Wiley & Sons.
[86] Jennes, M., Snoeys, R., Dekeyser, W., 1984,
Comparison of Various Approaches to Model the
Thermal Load on the EDM-Wire Electrode, Annals of
the CIRP, 33, 1, 93-96.
[87] Kunieda M., Takeshita S., Okumiya K., 1998, Study
on Wire Electrode Temperature in WEDM, Proc. ISEM
12, 119-128.
[88] Shoda K., Kaneko Y., Nishimura H., Kunieda M. and
Fan M.X., 1992, Adaptive Control of WEDM with
On-line Detection of Spark Locations, Proc. of ISEM X,
410-416.
[89] Lauwers B., Kruth J.P., Bleys P., Van Coppenolle B.,
Stenvens L., Derighetti R., 1998, Wire rupture
prevention using on-line pulse localisation in WEDM,
Proc. of ISEM 12, 203-213.
[90] Obara, H., Abe, M., Ohsumi, T., 1999, Control of Wire
Breakage During Wire EDM, 1st Report: Monitoring of
Gap Signals According to Discharge Location,
International Journal of Electrical Machining, No. 4,
53-58.
[91] Masuzawa, T., Heuvelman, C.J., 1983, A
Self-Flushing Method with Spark-Erosion Machining,
Annals of the CIRP, 32, 1, 109-112.
[92] Cetin S., Okada A., Uno Y., 2004, Effect of Debris
Distribution on Wall Concavity in Deep-Hole EDM,
JSME Int’l J., Series C, 47, 2, 553-559.
[93] Kaneko Y., Yamada H., Toyonaga T., Shoda K., 2000,
Performance of Linear Motor Equipped Die-Sinking
EDM, IJEM, 5, 59-64.
[94] Masuzawa, T., Cui X., Taniguchi, N., 1992, Improved
Jet Flushing for EDM, Annals of the CIRP, 41, 1,
239-242.
[95] Koenig, W., Weill, R., Wertheim, R., Jutzler, W. I.,
1977, The Flow Fields in the Working Gap with
Electro-Discharge-Machining, Annals of the CIRP, 25,
1, 71-76.
[96] Tsukahara H., Minami H., Hagino H., Lee S., Masui K.,
Sone T., 2004, EDM Process Using Chemical
Reaction of Organometallic Compound, J. of JSEME,
 38, 87, 37-42 (in Japanese).
[97] Narumiya H., Mohri N., Saito N., Ootake H.,
Tsunekawa Y., Takawashi T., Kobayashi K., 1989,
EDM by Powder Suspended Working Fluid, Proc.
ISEM 9, 5-8.
[98] Nishikawa K., Mohri N., Saito N., Furutani K, Kojima
H., Kunieda M., 1993, Phenomena in EDM using
Powder Suspended Working Fluid, ’93 Spring Meeting
of JSPE, 247-248 (in Japanese).
[99] Mohri N., Furutani K., Shirai K., Enami T., 1997,
Forced Discharge Dispersion by Dot-Matrix Method,
Annals of the CIRP, 46/1, 139-142.
[100] Luo Y.F., Zhang Z.Y., Yu Y., 1988, Mirror Surface EDM
by Electric Field Partially Induced, Annals of the CIRP,
39/1, 177-180.
[101] Kunieda M., Saga S., Yoshino H., Ohta T., Kobayashi
M., 2001, Control of Discharge Locations in EDM with
Locally Imposed High Electric Field, Proc. of ISEM XIII,
485-495.
[102] Han F., Kunieda M., Asano H., 2004, Improvement of
Controllability of Discharge Locations in WEDM, IJEM
No. 9, 37-42.
[103] Kunieda M., Muto H., 2000, Development of
Multi-spark EDM, Annals of the CIRP, 49/1, 119-122.
[104] Kruth, J. P., 1979, Adaptive Control Optimization of
Electro-Discharge Machining, Ph.D. thesis 79D3, ME
dept., K. U. Leuven.
[105] Snoeys, R., Dauw D., Kruth, J. P., 1980, Improved
Adaptive Control System for EDM Processes, Annals
of the CIRP, 29, 1, 97-101.
[106] Wang, W. M., 1985, Investigations of Monitoring
Discharge Parameters and Adative Control for EDM,
Ph. D Dissertation, Harbin Institute of Technology,
Harbin, China.
[107] Kruth, J. P., Snoeys, R., Van Brussel, H., 1978,
In-Process Optimization of Electro-Discharge
Machining, Proceedings of 19th MTDR, 567-574.
[108] Wang, W. M., 1988, A New EDM Adaptive Control
Plan Using Self-Tuning Control Algorithm,
Proceedings of ASME Manufacturing International’88,
Vol. 1, 227-233.
[109] Wang, W. M., Rajurkar, K. P., 1989, New Servo
Control Interface and Model Reference Adaptive
Control for EDM, Transaction of NAMRI/SME,
281-286.
[110] Rajurkar, K. P., Wang, W. M., 1990, Real Time
Stochastic Model and Control of EDM, Annals of the
CIRP, Vol. 39/1, 197-190.
[111] Zhao, W. S., Masuzawa, T., 1990, Adaptive Control of
EDM-Jump with Self-Tuning Approach, Bulletin of the
Japan Society of Precision Engineering, Vol. 24/1,
45-50.
[112] Wang, W. M., Rajurkar, K. P., Akamatsu, K., 1993,
Digital Gap Monitor and Adaptive Integral Control for
Auto-Jumping in EDM, ASME Journal of Engineering
for Industry, Vol. 117, 253-258.
[113] Morita, A., Imai, Y., Maruyama, H., Kobayashi, K.,
1989, Fuzzy Controller for EDM, Proceedings of ISEM
9, 236-239.
[114] Boccadoro, M., Dauw, D., 1995, About the Application
of Fuzzy Controllers in High-Performance Die-Sinking
EDM machines, Proceedings of ISEM XI, 333-341.
[115] Yan, M. T., Liao, Y. S., 1995, Adaptive Control of
WEDM Process Using the Fuzzy Control Strategy,
Proceedings of ISEM XI, 343-352.
[116] Zheng, H., Liu, Q. B., Guo, Y. F., Zhao, W. S., Kong, Z.
Y., Liu, J. C., 1998, A Fuzzy Controller for EDM
Process, Proceedings of ISEM XII, 185-191.
[117] Liao, Y. S., Woo, J. C., 1998, A New Fuzzy Control
System for the Adaptive Control of WEDM Process,
Proceedings of ISEM XII, 192-202.
[118] Liu, Z., Wang, W., Hu, J., 2001, Study on Application
of Gap Monitoring and Fuzzy Control for EDM,
Proceedings of ISEM XIII, 85-92.
[119] Klocke, F., Raabe, R., Wiesner, G., 1998, Optimization
of the EDM Sinking Process Using Neuro-Fuzzy
Control, Proceedings of ISEM XII, 163-172.
[120] Cao, F., Yang, D., 2004, The Study of High Efficiency
and Intelligent Optimization System in EDM Sinking
Process, Proceedings of ISEM XIV, 83-87.
[121] Yan, M. T., Liao, Y. S., Chang, C. C., 2001, On-Line
Estimation of Workpiece Height by Using Neural
Networks and Hierarchical Adaptive Control of WEDM,
Proceedings of ISEM XIII, 513-522.
[122] Kapoor A., Rajurkar, K.P., Williams, R.E., Bishu, R.,
1999, Internet-based Real Time Remote Monitoring
System for Machining Processes, Transactions of the
NAMRI/SME XXVII.
[123] Kobayashi, K., Nagao, T., Li, Gunki, Kakiuchi, K.,
Noda, T., Nakamura, K., Inoue, T., 2001, The Study on
the Methods for Applying Internet-use Technology to
EDM Fields, Proc. Of ISEM XIII, 523-533.
[124] Masuzawa T., Fujino M., Kobayashi K., 1985, Wire
Electro-Discharge Grinding for Micro-Machining,
Annals of the CIRP, 34, 1, 431-434.
[125] Masuzawa T., 2000, State-of-the-art on
Micromachining, Annals of the CIRP, 49, 2, 473-488.
[126] Han, F., Wachi, S., Kunieda, M., 2004, Improvement
of Machining Characteristics of Micro-EDM using
Transistor Type Isopulse Generator and Servo Feed
Control, Precision Engineering, 28, 4, 378-385.
[127] Hara, S., 2001, Ultra-high Speed Discharge Control
for Micro Electric Discharge Machining, Initiatives of
Precision Engineering of a Millennium, Kluwer
Academic Publisher, 194-198.
[128] Langen, H., Masuzawa, T., Fujino, M., 1995, Modular
Method for Microparts Machining and Assembly with
Self-Alignment, Annals of the CIRP, 44, 1, 173-176.
[129] Fujino, M., Okamoto, N., Masuzawa, T., 1995,
Development of Multi-Purpose Microprocessing
Machine, Proc. of ISEM XI, 613-620.
[130] Masuzawa, T., Yamamoto, M., Fujino, M.,1989, A
Micro-punching System using Wire-EDG and EDM,
Proc. of ISEM IX, 86-90.
[131] Langen, H., Breguet, J-M., Bleuler, H., Renaud, Ph.,
Masuzawa, T., 1998, Micro Electrochemical Discharge
Machining of Glass, IJEM, 3, 65-69.
[132] Egashira, K., Masuzawa, M., Fujino, M., Sun, X.-Q.,
1997, Application of USM to Micromachining by
On-the-machine Tool Fabrication, IJEM, 2, 31-36.
[133] Kawakami, T., Kunieda, M., 2005, Study on Factors
Determining Limits of Minimum Machinable Size in
Micro EDM, Annals of the CIRP, 54, 1, 167-170.
[134] Han, F., Yamada, Y., Kawakami, T., Kunieda, M., 2003,
Investigations on Feasibility of Sub-micrometer Order
Manufacturing Using Micro-EDM, ASPE 2003 Annual
Meeting, Vol. 30, 551-554.
[135] Egashira, K., Mizutani, K., 2005, EDM at Low
Open-Circuit Voltage, IJEM, 10, 21-26.
[136] Spur G., Uhlmann E., Doll U., Daus N., 1999, WEDM
of Microstructured Component Parts – Heat
Conduction Model, IJEM, 4, 41-46.
[137] Kruth J.P., Bleys P., 2000, Measuring Residual Stress
Caused by Wire EDM of Tool Steel, IJEM, 5, 23-28.
[138] Kocke, F., Lung, D., Noethe, T., 2001, Micro
Contouring by EDM with Fine Wires, Proc. ISEM 13,
767-779.
[139] Almond H., Allen D., Logan P., 2000, Examination of
the Internal Profiles of Micro-Electrodischarge
Machined Holes and the Detection of Material Defects,
IJEM, 5, 29-34.
[140] Masaki, T., Kawata, K., Masuzawa, T., 1990, Micro
Electro-Discharge Machining and Its Applications,
Proc. IEEE MEMS, 21-26.
[141] Takahata, K., Gianchandani, Y.B., 2001, Batch Mode
Micro-EDM for High-Density and High-Throughput
Micromachining, Proc. 14th IEEE Int’l Conf. Micro
Electro Mechanical Systems (MEMS 2001), 72-75.
[142] Masaki, T., Wada, T., 2002, Micro Electro Discharge
Machining, J. of JSAT, 46, 12, 610-613 (in Japanese).
[143] Yamazaki M., Mori N., Suzuki T., Kunieda M., 2003,
Simultaneous Processing Method for Micro-Rods and
Holes using EDM, Int’l Conf. on Leading Edge
Manufacturing in 21st Century (LEM 21), 983-988.
[144] Yamazaki M., Suzuki T., Mori N., Kunieda M., 2004,
EDMing of Micro-Rods by Self-Drilled Holes, J.
Materials Processing Technology, 149, 134-138.
[145] Kruth, J.P., Toenshoff, H.K., Klocke, F., 1998, Surface
and Sub-surface quality in material removal process
for tool making, Proc ISEM 12, 33-64.
[146] Masui, K., Demizu, K., Sato, Y., Sone, T., 1995,
Surface Modification of Tool Steels by Alloying Method
Using EDM Process, Proc. ISEM 11, 419-426.
[147] Mohri N., Saito N., Tsunekawa Y., 1993, Metal
Surface Modification by Electrical Discharge
Machining with Composite Electrode, Annals of the
CIRP, 42, 1, 219-222.
[148] Goto A., Magara T., Moro T., Miyake H., Saito N.,
Mohri N, 1998, Formation of Hard Layer on Metallic
Material by EDM, Proc. ISEM 12, 271-278.
[149] Mohri N., Fukusima Y., Fukuzawa Y., Tani T., Saito N.,
2003, Layer Generation Process on Work-Piece in
Electrical Discharge Machining, Annals of the CIRP,
52/1, 157-160.
[150] Moro, T., Mohri, N., Otsubo, H., Goto, A., Saito, N.,
2004, Study on the Surface Modification System with
Electrical Discharge Machine in the Practical Usage, J.
Materials Processing Technology, 149, 65-70.
[151] Kruth J.P., Froyen L., Stevens L. Dejonghe P., 1998,
In-process alloying of the white layer of a workpiece
machined by die-sinking EDM, IJEM, 3, 33-38.
[152] Uno, Y., Okada, A., Hayashi, Y., Tabuchi, Y., 1999,
Surface Modification by EDM with Nickel Powder
Mixed Fluid, IJEM, 4, 47-52.
[153] Aspinwall D.K., Dewes R.C., Lee H.G., Simao J.,
2003, Electrical Discharge Surface Alloying of Ti and
Fe Workpiece Materials Using Refractory Powder
Compact Electrodes and Cu Wire, Annals of the CIRP,
52/1, 151-154.
[154] Hayakawa S., Ori R.I., Itoigawa F., Nakamura T.,
Matsubara T., 2001, Fabrication of Microstructure
using EDM Deposition, ISEM 13, 783-793.
[155] Minami H., Masui K., Tsukahara H., Hagino H., 1998,
Coloring Method of Titanium Alloy using EDM Process,
Proc. ISEM 12, 503-512.
[156] Minami H., Masui K., Tsukahara H., Hagino H., 2001,
Coloring of Titanium Alloy using EDM Process –
Drawing with Simple Electrode -, Proc. ISEM 13,
589-599.
[157] Heerschap, M., Levitt, C.M., 1960, Eroding of Hard
Crystalline Carbon, United States Patent, 2,939,941.
[158] Van Osenbruggen C., van Ruler J., Schoenmakers
T.M., 1977, Method of Finishing a Workpiece of a
Non-conducting Material, Particularly Diamond, by
Means of Spark Erosion, United States Patent,
4,013,863.
[159] Fukuzawa Y., Katougi H., Mohri N., Furutani K., Tani
T., 1998, Machining Properties of Insulating Ceramics
with an Electric Discharge Machine, Proc. ISEM 12,
445-453.
[160] Mohri N., Fukuzawa Y., Tani T., Sata T., 2002, Some
Considerations to Machining Characteristics of
Insulating Ceramics, Annals of the CIRP, 51/1,
161-164.
[161] Tani T., Fukuzawa Y., Mohri N., Saito N., Okada M.,
2004, Machining Phenomena in WEDM of Insulating
Ceramics, Journal of Materials Processing Technology,
Vol. 149, Issues 1-3, 124-128.
[162] Taniguchi N., Nagata T., 1968, Micro-wave Machining,
50th Denki-Kakoh-Kenkyukai, 1, 1-22 (in Japanese).
[163] Uno, Y., Kubota, S., Yokomizo, S., Okada, A., Tanaka,
H., 1996, Study on Fine Boring of Single Crystalline
Silicon by EDM, J. of JSEME, 30, 65, 9-16 (in
Japanese).
[164] Kawada, K., Masaki, K., Sato, T., Masuzawa, T., 1994,
Study on Micro-EDM (3rd Report), J. of JSEME, 28,
59, 1-10 (in Japanese).
[165] Luo Y.F., Chen C.G., Tong Z.F., 1992, Slicing Thin
Silicon Wafers by Wire EDM Cutting, ISEM 10,
287-294.
[166] Kunieda M., Ojima S., 2000, Improvement of EDM
Efficiency of Silicon Single Crystal Through Ohmic
Contact, Precision Engineering, 24, 185-190.
[167] Kunieda M., Miyoshi Y., Takaya T., Nakajima N., Yu
Z.B., Yoshida M, 2003, High Speed 3D Milling by Dry
EDM, Annals of the CIRP, 52/1, 147-150
[168] Yu Z.B., Takahashi J., Kunieda M., 2004, Dry
Electrical Discharge Machining of Cemented Carbide,
J. Materials Processing Technology, 149, 353-357.
[169] Kunieda M., Takaya T., Nakano S., 2004,
Improvement of Dry EDM Characteristics Using
Piezoelectric Actuator, Annals of the CIRP, 53/1,
183-186.
[170] NASA, 1985, Inert-Gas Electrical-Discharge
Machining, NTIS Tech Note.
[171] Furudate, C., Kunieda, M., 2000, Study on
Dry-WEDM, Proc. 2nd Int’l Conf. MMSS, Krakow,
325-330.
[172] Kunieda, M., Furudate, C., 2001, High Precision
Finish Cutting by Dry WEDM, Annals of the CIRP, 50/1,
121-124.
[173] Fukui M., Kinoshita N., 1989, Developing a Mole
Electric Discharge Digging Machining, Annals of the
CIRP, 38/1, 203-206.
[174] Goto A., Watanabe K., Takeuchi A., 2002, A Method to
Machine a Curved Tunnel with EDM, IJEM, 7, 43-46.
[175] Ishida T., Kogure S., Miyake Y., Takeuchi Y., 2004,
Creation of Long Curved Hole by Means of Electrical
Discharge Machining Using an In-pipe Movable
Mechanism, J. Materials Processing Technology, 149,
157-164.
[176] Uchiyama M., Shibazaki T., 2004, Development of an
Electromachining Method for Machining Curved Holes,
J. Materials Processing Technology, 149, 453-459.