Journal of Physics D: Applied Physics

PAPER You may also like

- Investigations of dynamics of a single
Temporal characteristics of the pulsed electric spark-induced bubble in saline water
Zhen Liu, Xiantao Guan, Liancheng Zhang
discharges in small gaps filled with hydrocarbon oil et al.

- Analysis of discharge regimes obtained by

microsecond underwater electrical
To cite this article: U Maradia et al 2015 J. Phys. D: Appl. Phys. 48 055202 breakdown in regard to energy balance
B Dufour, N Fagnon, A Vega et al.

- Electric characteristic and cavitation

bubble dynamics using underwater pulsed
discharge
View the article online for updates and enhancements. Minglei SHAN, , Bingyan CHEN et al.

This content was downloaded from IP address 103.37.201.175 on 12/03/2022 at 18:25

 Journal of Physics D: Applied Physics

J. Phys. D: Appl. Phys. 48 (2015) 055202 (15pp) doi:10.1088/0022-3727/48/5/055202

Temporal characteristics of the pulsed

electric discharges in small gaps filled
with hydrocarbon oil
U Maradia1, Ch Hollenstein2 and K Wegener1
1
Institute of Machine Tools and Manufacturing (IWF), Leonhardstrasse 21, ETH Zürich, CH-8092
Zürich, Switzerland
2
Centre de Recherches en Physique des Plasmas, Ecole Polytechnique Fédérale de Lausanne, CH-1015,
Lausanne, Switzerland

E-mail: [email protected]

Received 23 September 2014, revised 20 November 2014

Accepted for publication 10 December 2014
Published 15 January 2015

Abstract
In order to understand the role of electrode materials in electrical discharges with micro
gaps (<200 µm) filled with a liquid hydrocarbon dielectric, the post-breakdown phase of low
ignition voltage (100 V) and low current (<20 A) pulsed electric discharges is experimentally
investigated. The electric discharge energies are selected in the range from 1 to 150 mJ. Due
to the non-repetitive and transient nature of the micro-discharges, time-resolved imaging,
spectroscopy and electrical analysis of single discharges are performed. The plasma–material
interaction is investigated by analysing the erosion craters on anode and cathode. It is found
that the electrode materials in these multiphase discharges affect the gas bubble dynamics, the
transport properties of the discharge plasmas and the transition from the gaseous to metallic
vapour plasma. The change in the energy fractions dissipated in the electrodes in function of
the discharge time is influenced by the thermo-physical properties of the electrode materials.
The simulation of craters in multiple discharge process requires consideration of the gas
bubble dynamics due to different energy fractions and plasma flushing efficiencies.

Keywords: plasma spectroscopy, discharge plasma, discharge crater, micro discharge

S Online supplementary data available from stacks.iop.org/JphysD/48/055202/mmedia

(Some figures may appear in colour only in the online journal)

1. Introduction of a few milli-Joules for rough machining. Capacitor based

generators are used for the discharges with micro- and nano-
Compared to the discharges in large gaps (>1 mm), little Joule energies, which are required to erode fine surfaces,
research has been concentrated on the electric discharges in achieve high precision and erode micro features.
small gaps, mainly due to the experimental diagnostic difficul- Physics of low current discharges in liquid dielectrics has
ties [1]. While the breakdown phenomenon in sub-millimetre been previously analysed to calculate plasma channel proper-
gaps remains an active research topic, application of micro- ties such as radius, pressure, electron density and tempera-
discharges is typical in electrical discharge machining (EDM), ture [2–5]. Using the visible light emission spectroscopy,
where a large number of consecutive pulsed discharges are Albinski et al [6] found the EDM plasma temperature to be
applied to machine a precision cavity. The discharges take between 8000–10 000 K. Descoeudres et al [7] used spatial
place in hydrocarbon oil as a dielectric with inter-electrode and time-resolved emission spectroscopy to thoroughly char-
gaps of less than 0.2 mm. Here, a transistor based electric cir- acterize such plasmas. With electron density in the range of
cuit is used to generate discharges with energies in the range 1016–1018 cm−3 and electron temperature around 0.7 eV, the

0022-3727/15/055202+15$33.00 1 © 2015 IOP Publishing Ltd Printed in the UK

J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Table 1. EDM material removal simulations performed by different authors and used energy distribution values for cathode Fc and anode Fa.

Year Authors Concept Fc (%) Fa (%)

1973 Van Dijck and Dutré [14] Thermo-physical model with disc heat source and superheating theory for 50 50
material removal
1986 Pandey and Jilani [27] Heat source expansion according to constant cathode boiling temperature 50 50
1989 DiBitonto et al [28] Point heat source including photoelectric effect 18.3 —
1990 Patel et al [29] Disc heat source including photoelectric effect — 8
1996 Xia et al [17] Temperature measurements for single, multiple discharges to determine 25–34 40–48
energy distribution
2005 Revaz et al [19] Fast response temperature measurements 10–15 —
2012 Zahiruddin and Kunieda [30] Small energy capacitance discharges with duration of 70 ns — 10.4
2012 Singh [31] Temperature measurements in cathode 6–27
2013 Shabgard et al [32] Calculation of plasma flushing efficiency 4–9 4–36
2014 Zhang et al [33] Metallographic analysis of craters to determine energy input 42 – 60 36–50

EDM plasmas in liquid hydrocarbon have been found to be removal during the discharge itself or at the end of the discharge
weakly non-ideal with a coupling parameter of 0.45. In micro- is considered [14, 15, 21–24]. Molecular dynamic simulations
EDM, the electron density in the order of 3.5 × 1018 cm−3 and have also been performed in nanometre and picosecond scale
plasma temperatures between 5000–8000 K are found [8]. to understand the material removal mechanism [25, 26].
Also, broadening of the Hβ spectral line is observed for such Due to the lack of systematic investigations of the micro-
plasmas with nanosecond pulse durations. For similar dis- discharges through a coherent analysis, current research
charge plasmas in air, as in the case of dry-EDM, much lower attempts to gain insights in to the temporal characteristics
electron density of 1013–1014 cm−3 and plasma temperature of single discharges by understanding how the discharge
of 4000 ± 500 K have been found [9], which is typical for plasma and its interaction with the electrode surfaces affect
ideal plasmas. Recently, Braganca et al [10] used visible light each other. In particular, the role of electrode materials on the
emission spectroscopy and a semi-empirical resistive plasma gas bubble dynamics, discharge plasma and discharge craters
model to understand the effect of gap size, pulse duration and is analysed. While on one hand such analysis is essential for
current on electron density of single discharges, which was better description of the discharge plasma and plasma–surface
determined to be in the order of 1016–1017 cm−3. Additionally, interactions through modelling and simulation, understanding
using results from emission spectroscopy, simulations have of the role of electrode materials on the discharge phenomena
been performed with the method of net emission co-efficient is especially useful in EDM. For example, spark to arc tran-
[11], where the cooling due to iron vapour content in the sition boundaries dependent on time and electrode materials
plasma has been evaluated [12]. In most of these works, an may help to avoid the undesired arcing phenomenon in EDM.
average of thousands of discharges is considered due to the Also, state of the plasma and its effect on the generated cra-
weak plasma light emission. While on one hand this method ters is useful for optimal process parameter selection to obtain
provides better approximation of the average plasma proper- high material removal on workpiece and lower removal on
ties, it limits the precise characterization of discharge plasmas tool. Upon introduction of the used experimental methods and
due to the non-repetitive nature of these discharges. heat conduction simulation method, results are presented for
High speed imaging with up to one million frames per the discharge plasma and the erosion craters.
second and emission spectroscopy is used to determine EDM
arc plasma diameter in gas and further correlated to the dis-
2. Experimental set-up and methods
charges in oil [13], where the plasma region is found to be
considerably larger than the erosion craters on the electrodes.
2.1. Discharge conditions and imaging
In terms of plasma–material interactions, modelling of mate-
rial removal of cathode and anode has been developed, where A point-to-plane electrode configuration is used, where the
superheating of material is considered and other effects such point anode has a diameter of 1 mm. Copper and graphite
as joule heating have been found to be negligible [14–16]. are chosen as the anode materials. Cathode materials are
Although the electrode heating is treated as a simple heat con- Steel 1.2343, TiAl6V4 (titanium basis alloy) and Inconel 718
duction problem with the discharge plasma acting as a heat (nickel basis alloy). The electrodes and discharge region are
source, the boundary conditions for such simulations continue completely submerged in the liquid dielectric. The hydro-
to remain a focus of research [17–21]. The main boundary carbon oil IME110 from Oelheld GmbH with a kinematic vis-
conditions are heat source expansion, heat flux profile, energy cosity of 3.4 mm2 s−1 (at 20 °C) is used as dielectric. Electrical
entering into cathode and anode. The used boundary condi- discharges typical for roughing operation in EDM have been
tions differ considerably among various authors as shown in analysed in this work. The pulse duration of the discharges is
table 1 and are mainly based on assumptions. Apart from the set at 300 µs. The applied peak current is 20 and 5 A for rect-
heat conduction simulations, material removal mechanism angular shaped current pulses generated by a constant current
also remains an active topic of discussion, where the material source with 100 V ignition voltage as shown in figure 1.

2
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Figure 1. Electrode arrangement for gas bubble and plasma expansion characterization (see supplementary material stacks.iop.org/
JphysD/48/055202/mmedia). A discharge with 20 A is generated between a 1 mm diameter anode and cathode plane completely submerged
in the liquid dielectric oil. The images with a 4.5 µs exposure time show discharges after about 50 µs from the dielectric breakdown. The
voltage and current waveforms of a typical 5 A discharge are also shown.

Single discharges are generated using a commercial GFAC Table 2. Chosen spectral lines (λ1, λ2) of different elements for the
Form 1000 EDM machine, discharge images are recorded at electron temperature calculations using the two-line method.
200 000 frames per second (fps) using a high-speed imaging Elements Wavelength (nm) E (eV) g A (10−8 s−1)
camera Phantom V12 from Vision Research and the electrical
signals are recorded using a Lecroy 44Mxi oscilloscope. The Cu 510.554 3.82 4 0.02
521.820 6.19 6 0.75
use of telecentric lens attached to the high-speed camera
Fe 516.748 3.88 7 0.027
allows direct distance measurement from the images without
526.953 3.21 9 0.018
distortions, resulting in a spatial resolution of 18 µm. Craters Ti 444.914 4.67 11 0.97
generated on cathode and anode are analysed using a confocal 453.324 3.58 11 0.88
microscope Leica DCM 3D and metallographic analysis is Cr 529.827 0.982 5 0.033
performed to analyse the sub-surface structure of the polished 540.977 1.030 7 0.062
and etched crater cross-sections.
Since Descoeudres et al [7] found that the spatial proper- 2.2. Heat conduction simulations
ties of EDM plasma vary slightly due to the small size of
plasma, plasma is assumed to be spatially homogeneous and Finite element method using COMSOL Multiphysics 4.3b
only time-resolved emission spectroscopy is performed in software has been used for heat conduction simulation in the
this work. The light emitted from the discharge plasma is electrodes. A 2D axis-symmetric model of anode and cathode
collected by a 1 mm diameter optical fibre bundle. The light is used and assumptions for simulations are similar to that
is dispersed by a W-configuration SP2556 spectrometer from elaborated by DiBitonto et al [28]. The plasma heat source
Princeton instruments with a focal length of 500 mm. To on the electrode is assumed to be a time-dependent disc with
reduce the light intensity loss, the spectrometer is fitted with a uniform heat flux profile. For the heat source radius estima-
two silver coated mirrors with a reflectance above 95% in the tion, constant cathode boiling temperature condition as pro-
visible range. The spectrum is recorded by the high-speed posed by [27] and shown in equation (1) is considered.
camera at the frame rate of 50 000 fps. Using 150 l mm−1
⎡ ⋅ a ⋅ t ⋅ 106 ⎤ 2
1
gratings and complimentary metal-oxide-semiconductor E ⋅ Fc ⋅ 106 −1⎢ 4
 Tb = ⋅ tan ⎥ , (1)
(CMOS) sensor pixel width of 20 µm, spectral resolution of
k ⋅ rp ⋅ π
3
2 ⎣ rp2 ⎦
0.263 nm in the spectral range of 360–715 nm is obtained.
The plasma properties such as electron density, electron
 E = Ue ⋅ I ⋅ t . (2)
temperature and coupling parameter have been calculated
using the method as described in detail by [34], assuming the Here, Tb is boiling point temperature in K, rp is heat source
local thermal equilibrium (LTE) of the plasma. The electron radius in m, k is thermal conductivity in W m−1·K−1, a is
density is calculated by considering the stark broadening of thermal diffusivity in m2 s−1, t is pulse duration in µs, E is total
the Hα spectral line. The electron temperature is calculated discharge power in W and Fc is cathode energy fraction. Total
using the two-line method. The chosen emission lines for dif- discharge power is a function of the discharge voltage Ue and
ferent electrode elements are presented in table 2. current I in time. Temperature dependent thermal conductivity,
Plasma spectrum simulations have been performed using the specific heat and density of the electrode materials are used
commercial PrismSPECT 4.7 software from Prism Computation for the simulations. The latent heats of vaporization and fusion
science. For simplification, only the hydrogen plasma emission have been ignored. The melting temperature isotherm of the
spectrum has yet been simulated in this work. The hydrogen respective electrode material is considered as the melt front.
emission spectrum was obtained assuming zero-width plasma The ratio between the eroded crater volume and the molten
with a Maxwell electron distribution. The low temperature metal volume is termed as the plasma flushing efficiency (PFE)
spectroscopy model used ion density and electron temperature in EDM. Assuming the PFE of 100%, the energy fraction
as input. The simulations are performed for hydrogen including values Fc and Fa have been calculated by comparison with the
the bound-bound and bound-free transitions. measured crater volume and simulated molten metal volume.

3
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

4000 3000
Measured diameter (µm) Steel: Bubble

Measured diameter (µm)

3500
2500
3000 Steel: Bubble Inconel: Bubble
2000
2500 Inconel: Bubble
TiAl6V4: Bubble
2000 1500
TiAl6V4: Bubble
1500 Steel: Plasma
1000
1000 Steel: Plasma
Inconel: Plasma
500
500
TiAl6V4: Plasma
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (µs) Time (µs)

Figure 2. Measured gas bubble and light emitting plasma region for 300 µs long 20 A discharges between the Ø1 mm anode and cathode
plane. Discharges are generated between steel, Inconel 718 and TiAl6V4 cathodes and graphite (left) and copper (right) anodes.

3. Discharge plasma 1600

Gr-Steel
C 5A: Bubble Cu-Steel 5A: Bubble
1400
3.1. Gas bubble and plasma expansion Gr-Steel
C 5A: Plasma Cu-Steel 5A: Plasma

Measured diameter (µm)

1200
The electrode arrangement inside dielectric oil, the gas bubble
1000
and plasma expansion characterization method and the typ-
ical voltage and current signals of a 5 A discharge are shown 800
in figure 1. The discharge voltage is measured to be within 600
23–20 V for 20 A discharges and between 20–18 V for 5 A
discharges. 400

The plasma region is considered identical with the visible 200

light emission region [34]. Although the gas bubble formation 0
from the liquid dielectric during the discharge breakdown is 0 50 100 150 200 250 300
instantaneous, the first measured point in the presented results Time (µs)
is after 5 µs from the breakdown, which is the interval time
for high-speed imaging. The gas bubble is mainly formed of Figure 3. Measured gas bubble and light emitting plasma region
for 300 µs and 5 A discharges between Ø1 mm anode and cathode
the hydrogen and carbon by the hydrocarbon oil vaporization. plane. Cathode material is steel and anode materials are graphite
Comparison of gas bubble and plasma expansion for 300 µs and copper.
long discharges with different anode and cathode materials is
shown in figure 2. In case of the graphite anode (left), the gas into the electrode bulk. Thus, melting and boiling point tem-
bubble expands during the discharge for steel and TiAl6V4 peratures, enthalpies for vaporization and fusion and thermal
cathodes, whereas bubble oscillation takes place during the conductivity may influence the energy balance, hence the gas
discharge when using Inconel 718 cathode. In comparison, for bubble dynamics.
the copper anode, bubble oscillations take place during the Due to its higher thermal conductivity and lower boiling
discharge for the analysed cathode materials. However, in this point, higher energy may be dissipated into copper compared
case, the bubble oscillation frequency changes depending on to the graphite. Also, with lower heat conductivity and higher
the cathode material. boiling point, more energy is dissipated into the dielectric in
Of the total discharge energy, the fractions of energy are case of the TiAl6V4 compared to the other cathode materials.
dissipated into anode, cathode, dielectric oil and the plasma On the other hand, the bubble expansions are identical for
itself. Since the bubble expansion is caused by the plasma copper and graphite anodes for 5 A discharges and steel cath-
energy released into the liquid dielectric and its resulting odes as shown in figure 3. Since the energy released by the
pressure, it can be inferred that the energy released through plasma channel and the resultant pressures are functions of
radiation and convection due to joule heating of the plasma the discharge energy, lower bubble diameters and the bubble
channel is lower for copper anode compared to the graphite oscillations are seen in the case of 5 A discharges.
anode. Similar observation can be made for the cathode
materials, where the energy dissipated into the dielectric is
3.2. Plasma characterization
highest for the TiAl6V4, followed by steel and Inconel in case
of the graphite anode. In order to explain this behaviour, one Figure 4 shows typical visible light emission spectrum of the
may consider the plasma properties and the electrode mate- discharge plasma with 20 A generated between a graphite
rial properties. Assuming similar plasma properties, one may anode (left), a copper anode (right) and a steel cathode. Since
infer that the energy dissipated into the electrodes is different. the discharge is generated in hydrocarbon oil, a high intensity
A part of the energy dissipated into the electrode is used for hydrogen spectral line Hα is clearly visible in the spectra. For
melting and evaporation of a crater and the rest is conducted the graphite anode, Fe and Cr spectral lines from steel cathode

4
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

30 20
Hα 18
25 Hα
16

Intensity [a.u.]
Intensity [a.u. ]

Fe 14
20 Cu
12
15 Fe Fe 10 Fe Cr
Fe Cr Fe Fe
Fe Fe Fe 8 Cu
Fe Fe
10 6 Fe Cu Fe Cr
Fe 4
Fe Cu
5
2
0 0
350 400 450 500 550 600 650 700 350 400 450 500 550 600 650 700
Wavelength (nm) Wavelength (nm)

Figure 4. Time integrated light emission spectrum of a 300 µs duration 20 A single discharge between the Ø1 mm graphite anode (left),
copper anode (right) and steel cathode.

Figure 5. Time resolved spectra of the 20 A single discharges, acquired at 20 µs interval with 19 µs exposure time and normalized to the Hα
peak. Discharges are generated between the steel cathode and graphite (left) and copper (right) anode.

1.9E+17 10000
Steel
Electron density ne (cm-3)

Electron temperature Te (K)

1.6E+17 Inconel 718

8000 TiAl6V4
1.3E+17
6000
1E+17
4000
7E+16
Steel
4E+16 Inconel 718 2000
TiAl6V4
1E+16 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (µs) Time (µs)

Figure 6. Calculated electron density ne and electron temperature Te for 20 A discharges between the graphite anode and different
cathode materials.

are visible, whereas Cu lines are also visible when using the for a vacuum or arc discharge plasmas. This may indicate the
copper anode. The metallic lines originate from the metal transition phase from spark to a more stable arc with higher
vapour which is excited and not from the debris, since the metallic content and electron density. Understanding of this
heated debris generate the continuous black-body radiation. transition is crucial for energy balance in the gap region, since
Time-resolved spectra of the same discharge reveal addi- the energy dissipation into the electrodes is higher in the case
tional information as shown in figure 5. Using a graphite of arc discharges.
anode (left), the metallic emission lines, especially the Fe The calculated electron density ne and electron tempera-
spectral lines have much higher intensity compared to Hα after ture Te for the plasma generated between graphite anode
200 µs, thus indicating a metal vapour content of the plasma. and different cathode materials are presented in figure 6. It
In the case of the Cu anode, Cu spectral lines have high inten- is seen that the electron density is in the order of 1017 cm−3
sity in the first 30 µs from discharge breakdown and Fe lines after about 30 µs from the discharge breakdown, which is
have higher intensity with the increasing duration. Thus, it can in agreement with the values found by [34]. However, for
be inferred that the cathode metal vapour content increases in longer pulse durations, when the metallic vapour content in
the discharge plasma with increasing pulse duration, typical the plasma increases, an increase in the electron density ne

5
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

1.9E+17 10000

Electron temperature Te (K)

Electron density n e (cm-3)
1.6E+17 8000
1.3E+17
6000
1.0E+17
4000 Steel
7.0E+16
Steel Inconel 718
4.0E+16 Inconel 718 2000 TiAl6V4
TiAl6V4
1.0E+16 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (µs) Time (µs)

Figure 7. Calculated electron density ne and electron temperature Te for 20 A discharges between the copper anode and different cathode
materials.

Table 3. Overview of the calculated electron density ne and electron temperature Te from the measured spectra for different anode and
cathode materials.
Temperature Te
Anode Cathode Current (A) 17
Density ne × 10 (cm ) −3
K eV Coupling factor Γ
Graphite Steel 1.2343 20 1.72 ± 0.14 5310 ± 600 0.46 ± 0.10 0.29 ± 0.06
5 1.64 ± 0.20 4712 ± 710 0.41 ± 0.10 0.34 ± 0.13
TiAl6V4 20 1.75 ± 0.10 5400 ± 120 0.47 ± 0.01 0.28 ± 0.01
Inconel 718 20 1.68 ± 0.20 4100 ± 700 0.38 ± 0.10 0.30 ± 0.10
Copper Steel 1.2343 20 1.58 ± 0.13 6250 ± 550 0.54 ± 0.10 0.24 ± 0.03
TiAl6V4 20 1.69 ± 0.19 5960 ± 750 0.51 ± 0.10 0.28 ± 0.05
Inconel 718 20 1.52 ± 0.21 5930 ± 600 0.51 ± 0.05 0.25 ± 0.03

is seen, typically after 200 µs for steel and Inconel cathodes. the bubble contraction and thus further analysis is needed
No increase in the electron density ne is observed when using using the concurrent imaging and spectroscopy of a single
TiAl6V4 cathode within the used pulse duration. The calcu- discharge. Also, high metal vapour density in the plasma may
lated electron temperatures Te in the range of 3000–6000 K are cause increased electron density due to the low excitation
found to be constant with time and are almost half compared energies of metallic elements, thus metal vapour density also
to the values found by [34]. The lower temperature Te and the need to be determined using spectroscopy.
increase in electron density ne may be attributed to the higher Time-resolved spectroscopy of single discharges offers
metallic vapour within the plasma. The metal vapour in plasma an additional benefit of analysing the subsequent discharges.
may provide additional cooling effect due to increased radia- Figure 8 shows time-resolved spectra of two distinct dis-
tive emission losses as observed by [12, 35, 36] and increased charges with 350 µs pause duration (left) and 200 µs pause
energy input into the electrodes. The observed discharge duration (right). It is seen that for the shorter pause dura-
plasmas have plasma coupling factor between 0.25–0.35, sug- tion, an emission continuum is observed in the spectra.
gesting cold dense or weakly non-ideal plasmas similar to A similar continuum is observed in the time-integrated
[34]. The calculated electron density ne and electron tempera- spectra of multiple discharges in the micro scale erosion
ture Te for the plasma generated between copper anode and over longer spectrum acquisition durations as shown in
different cathode materials are presented in figure 7. It is seen figure 9. Here, two scenarios have been analysed, where the
that the electron density ne is similar to the case of graphite discharges take place on the surface of the cathode or inside
anode, where as a slightly higher electron temperature Te a 5 mm deep cavity in the cathode. In order to collect light
values are observed for the copper anodes. Mean electron from the discharges inside a cavity, near real erosion con-
densities ne and temperatures Te calculated from at least five ditions as developed by [37] have been used. Here, two of
repetitions of the discharges with different anode and cathode the four metallic walls of the deep cavity are replaced by
materials are presented in table 3, where it can be inferred that the glass plates to allow light collection and visual observa-
the average plasma properties do not change significantly. The tion of the gap region. For the multiple discharge process,
mean values obtained by repetitive discharges have a standard the cavity is filled with the gas bubbles from the previous
deviation of less than 20% as shown in table 3. This signi- discharges. This can be correlated with the two subsequent
fies the importance of single discharge analysis compared to discharges with short pause duration between them, where
the averages derived from the multiple discharges, where the the second discharge takes place inside the gas bubble from
temporal changes in the plasma properties may not be clearly the previous discharge. Also, the metallic spectral lines have
identified due to the stochastic nature of the discharges. The higher intensity compared to Hα spectral line intensity in the
temporal change in the electron density ne may be caused by spectra of the subsequent discharge.

6
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Figure 8. Two consecutive 20 A discharges with 150 µs pulse duration between the Ø1 mm graphite anode and steel cathode with a pause
duration of 350 µs (left) and 200 µs (right).
Micro
(in cavity)
Fe

Anode
Hα

Gas bubble
Intensity [a.u.]

Fe
Fe
Micro

Anode
Cathode (on
surface)

Gas bubble

Cathode

350 450 550 650 750 850 950

Wavelength (nm)

Figure 9. Time integrated spectra of the multiple discharges with 20 A for different electrode configurations. The spectrum acquisition time
is 1 s and the discharge frequency is set at 1 kHz. The spectra are normalized to the Hα peak.
1
Elapsed time from discharge
Hβ Hα ne
breakdown (µs)
0.8 (cm-3)
14
Intensity [a.u.]
Intensity [a.u.]

1e20
34
0.6 1e19
114
1e18
154
0.4 1e17
314
1e16
0.2 1e15
1e14
0
645 650 655 660 665 670
Wavelength (nm) Wavelength (nm)

Figure 10. Broadening and shifting of the Hα peak seen in the measured spectra (left) and simulated spectra of pure hydrogen plasma
(right) with different electron densities showing a decrease in FWHM of the Hα peak with decreasing electron density ne.

In order to interpret the spectroscopic measurements 4. Discharge craters

and to clearly identify the spectral lines, emission spectrum
simulation can be useful. Thus, spectrum simulation of pure Heat dissipation into the electrodes from the plasma channel
hydrogen plasma is performed to validate the calculated elec- results in high temperatures causing vaporization and
tron densities as shown in figure 10. It is seen from the simu- melting, and leads to the erosion craters. The heat dissipa-
lated spectra that the Hα is broadened with increasing electron tion may be through radiation, conduction and electron/
density and the full-width half-maximum (FWHM) of Hα ion bombardment. In general, the discharge crater sizes are
peak for the measured electron densities correlates well with dependent on the discharge energy and energy entering into
the simulations. Also, due to the abundance of many Fe spec- the electrodes. Figure 11 shows a typical single discharge
tral lines around the Hβ spectral line, the spectra simulation crater generated on a steel cathode by a 20 A discharge with
provides a clear indication of the presence of the Hβ spectral 300 µs pulse duration using Ø1 mm graphite anode. It can be
line in the measured spectrum for the given electron densities. seen that a part of the volume is re-deposited in form of a
Simulation of the visual emission spectra may open a new rim around the crater. Metallographic analysis of the crater
way to gain insight into the interesting features of the EDM shows the re-solidified region along the crater depth with
discharge plasma. thickness less than 10 µm.

7
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Figure 11. Characterization of a crater generated on steel cathode using 20 A, 300 µs duration discharge with the Ø1 mm graphite anode.
The crater volume is shown by V−, whereas the re-solidified volume around the crater is shown as V+. Apart from an optical image of the
crater from top, the height profile from the 3D measurements of the crater is shown using pseudo colours.

1.6E+10 1.5E+11

Energy density (W/m 2)

1.4E+10
Current density (A/m2)

1.2E+11
1.2E+10 20A
20A 5.4A
1.0E+10 9.0E+10
5.4A
8.0E+09
6.0E+09 6.0E+10
4.0E+09 3.0E+10
2.0E+09
0.0E+00 0.0E+00
0 100 200 300 0 100 200 300
Time (µs) Time (µs)

Figure 12. Calculated average current density (left) and energy density (right) at the cathode spot for 20 and 5.4 A discharges between a
Ø1 mm graphite anode and steel cathode.

Figure 13. Variation of crater geometries for five repetitions of discharges with pulse duration of 20, 75, 150 and 300 µs. The 20 A
discharges are generated between the steel 1.2343 (cathode) and Ø1 mm graphite anode. The height profiles of the craters from 3D
measurements are shown in pseudo colours to clearly show rim region and eroded volume. Here, the green region is the zero reference
(polished cathode plane), yellow to red indicates increasing rim height from the reference plane and blue to violet shades indicate
increasing depth from the reference plane.

Using the measured cathode crater radius r, current den- considered to be the primary crater formation mechanism,
sities and energy densities are calculated on the cathode as opposed to joule heating used for the current densities
spots according to equations (3) and (4) respectively. As above 1 × 1010 A m−2 [38].
shown in figure 12, the current densities at the cathode
spot fall below 1 × 1010 A m−2 within a few microseconds I
 Current density = , (3)
after the breakdown, thus electron/ion-bombardment is π ⋅ r (t ) 2

8
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

1.40E-03 140
V+
1.20E-03 120
Crater volume (mm3) V-

Dimension (µm)
1.00E-03 100
Radius
8.00E-04 80
Depth
6.00E-04 60
4.00E-04 40
2.00E-04 20
0.00E+00 0
0 5 10 15 20 25 0 5 10 15 20 25
Gap width (µm) Gap width (µm)

Figure 14. Effect of the gap width on the crater geometry is found to be negligible. The craters are generated between the steel cathode
plane and Ø1 mm graphite anode with 250 V ignition voltage and 20 A with 300 µs pulse duration.

Figure 15. Tilted images of the cathode crater cross sections (top) and magnified images of the re-solidified layer on the crater surface
(bottom). The craters on steel cathode are eroded by 20 A discharges using a graphite anode. Re-solidified layer thickness is seen to be
increasing with increasing pulse duration.

Ue ⋅ I the latent heat of electrode material may also result in higher

 Energy density = . (4) re-solidified volume. In terms of crater geometry, the eroded
π ⋅ r (t )2
crater volume, crater radius and crater depth increase with
increasing discharge duration as seen in figure 16. Here, larger
variation in the crater depth is seen compared to the crater
4.1. Cathode craters
radius, further stressing the role of plasma flushing through
Craters generated by single discharges on steel cathode plane bubble implosion on the crater formation.
using a graphite anode of 1 mm diameter and 20 A with dif- In order to simulate the crater geometry, heat source radius
ferent pulse durations have been analysed as presented in rp and energy fraction into cathode Fc are varied. Three sepa-
figure 13. Despite the similar process conditions, there is a rate heat source radius rp expansion rates have been con-
large variation in the discharge crater geometries. To under- sidered. Apart from the heat source radius rp similar to the
stand this variation, the effect of inter-electrode gap width measured crater radius r, constant cathode boiling tempera-
was analysed, where it can be seen in figure 14 that almost ture condition as proposed by [27] and shown in equation (1)
no influence of gap width on the crater geometry has been has been used. Also, the melting temperature apart from the
observed within the analysed gap width range. In fact, dis- boiling temperature has been used in equation (1) to derive
charge location on the electrode and the resulting gas bubble the heat source expansion in time. It can be seen that simu-
shape [39] with the subsequent bubble implosion may affect lated crater geometries using heat source expansion based on
the molten metal ejection from the electrode bulk and thus both the measured crater diameter and constant boiling tem-
crater geometry. perature condition are in good agreement with the measured
The molten material not removed from the discharge crater geometry. On the other hand, constant melting tempera-
region is re-solidified by rapid cooling. The re-solidified layer ture condition results in over-estimation of the crater size and
thickness increases with increasing pulse duration as seen results in a different crater profile as shown in figure 17.
in figure 15. This may indicate lower pressure gradients on The erosion craters generated on steel cathode using copper
the gas bubble boundary for longer pulse durations, resulting anode with 20 A discharges are shown in figure 18. These cra-
in lower ejection of the molten metal from electrode. Also, ters are much smaller, especially in depth compared to the
cooling of the molten metal pool during the discharge due to craters generated using the graphite anode. Also, the plasma

9
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Figure 16. Simulated temperature distribution in the steel cathode induced by a 20 A discharge. Different heat source radius rp expansion
rates have been considered resulting in the different melt front profiles. The observed crater profile is consistent with the melt front derived
from the simulation using the constant cathode boiling temperature condition shown in equation (1).

0.0040 Measured
200 90 V-
0.0035
80

Crater volume V- (mm3)

175
Crater radius r ( µm)

Crater depth d ( µm)

70 0.0030 Fc=0.45
150
60 0.0025
125
50 0.0020 Fc=0.35
100
40
75 0.0015
30
50 0.0010 Fc=0.44
20
25 0.0005 (crater
10
measure)
0 0 0.0000 Fc=0.45
0 100 200 300 0 100 200 300 0 100 200 300 (Melting)
Time (µs) Time (µs) Time (µs)

Figure 17. Measured and simulated geometries of the craters on a steel cathode generated by 20 A discharges using a Ø1 mm graphite
anode. Three different heat source radius rp expansions have been considered, namely constant cathode boiling temperature using
equation (1) (Fc 0.45, 0.35), constant melting temperature (Fc 0.45) and measured cathode crater radius (rp = r, Fc 0.44).

Figure 18. Discharge craters generated by 20 A discharges on a steel cathode using a Ø1 mm copper anode. A plasma affected region is
seen around the craters. Also, crater depth is much lower compared to the craters generated using a graphite anode.

affected region is seen around the craters. The smaller craters The erosion craters generated on Inconel 718 and TiAl6V4
sizes for similar discharge energy may result from the lower cathodes using graphite and copper anodes with 20 A dis-
cathode energy fraction or lower PFE. However, the re-solidi- charges are shown in figure 19. The measured crater geom-
fied layer thickness is found to be similar to the craters gener- etries of the craters are shown in figure 20 for graphite anodes
ated using the graphite anode. Here, a higher anode energy and figure 21 for copper anodes. In the case of graphite anode,
fraction for the copper anode compared to the graphite anode eroded craters on the steel cathode have highest crater volume,
may result in lower cathode energy fraction and lower dielec- followed by Inconel 718 and TiAl6V4. While a part of the
tric energy fraction. This may explain the lower erosion crater energy dissipated into the cathode is used for erosion, con-
volumes and lower gas bubble volumes for copper anodes, siderable amount of energy is dissipated into the electrode
assuming the bubble implosion as the primary mechanism for bulk through conduction. Steel has the highest thermal con-
crater generation. Thus, one may infer that the cathode energy ductivity of the investigated cathode materials and boiling
fraction Fc is lower in case of the copper anode. temperature higher than Inconel but lower than TiAl6V4.

10
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Figure 19. Erosion craters generated by 20 A discharges on Inconel 718 and TiAl6V4 cathode planes using Ø1 mm graphite (left) and
copper (right) anodes.

0.0030 200 70
Steel Steel
Crater volume V- (mm3)

0.0025 Inconel
Crater radius r (µm) 60 Inconel

Crater depth d (µm)

160 TiAl6V4
TiAl6V4 50
0.0020
120 40
0.0015
80 30
0.0010 Steel 20
40 Inconel
0.0005 10
TiAl6V4
0.0000 0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (µs) Time (µs) Time (µs)

Figure 20. Measured eroded volume V−, radius r and depth d of the craters on different cathode materials. The craters are generated by a
Ø1 mm graphite anode using 20 A discharges.

0.0016 200
30
Steel
Crater volume V- (mm3)

0.0014 Steel Steel

Crater depth d (µm)

Inconel
Crater radius r (µm)

160 Inconel 25 Inconel

0.0012 TiAl6V4 TiAl6V4 TiAl6V4
0.0010 120 20
0.0008
15
0.0006 80
0.0004 10
40
0.0002 5
0.0000 0
0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
0 50 100 150 200 250 300
Time (µs) Time (µs) Time (µs)

Figure 21. Measured eroded volume V−, radius r and depth d of the craters on different cathode materials. The craters are generated by a
Ø1 mm copper anode using 20 A discharges.

Table 4. Overview of the cathode crater geometries for 300 µs discharges and the used cathode energy fraction Fc values for simulations.

Energy
fraction Eroded crater volume Crater Max. crater
Anode Cathode Current (A) Fc V − (mm3) radius (µm) depth (µm)
Graphite Steel 1.2343 20 0.40 0.003 150 75
5 0.20 0.00005 60 15
Inconel 718 20 0.25 0.002 200 40
TiAl6V4 20 0.15 0.0008 130 35
Copper Steel 1.2343 20 0.20 0.0003 110 15

Thus, higher crater volumes and crater depth may be expected different materials to be near 100%, the energy fraction Fc
for the steel cathodes. It is seen that the crater generated on values used for the discharge crater simulations are shown in
TiAl6V4 using copper anode for longer duration is consider- table 4. It is noted that the values depend on the considered
ably large. Further investigation is needed to understand this PFE and thus detailed analysis of the re-solidified layer is
effect. Assuming the similar plasma flushing efficiencies for also required, similar to that by [33]. Nonetheless, it can be

11
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

Table 5. Overview of the cathode crater geometries for 300 µs long

discharges and the used energy fraction Fc values for simulations.
Current Energy
Anode Cathode (A) Crater volume fraction Fc
Graphite Steel 20 Eroded 0.40
1.2343 volume V −
Effective 0.20
volume Ve
Mean crater 0.09
volume Vm
Near real erosion 0.20, 0.10
Figure 22. Discharge craters generated by 20 A discharges on condition: H, L
copper anode using a Ø1 mm steel cathode. The crater volume Copper Steel 20 Eroded 0.20
decreases with increasing pulse duration and a black carbonaceous 1.2343 volume V −
layer is formed in the discharge region.
Effective 0.13
volume Ve
inferred that the cathode energy fraction Fc is also dependent Mean crater 0.11
on the anode material properties. volume Vm
Near real erosion 0.13, 0.10
condition: H, L
4.2. Anode craters

Analysis of the erosion craters on graphite is complex due to re-solidified volume V+ must be also eroded by the subse-
the graphite topography and the very small crater sizes due to quent discharges, thus effective crater volume Ve by a single
the high sublimation temperature of graphite. For the multiple discharge is (V−)–(V+). Table 5 presents the Fc values used
discharge process, it was seen that the crater volume decreases to simulate the crater volumes under different conditions. It
with the increasing pulse duration, reaching a zero value near is seen that Ve is much lower compared to V− when graphite
150 µs for 20 A discharges. A carbonaceous layer is built up anode is used. Also, Vm is considerably low when using the
on the anode for the longer pulse durations. Figure 22 shows micro scale (<Ø1 mm) anodes. In order to understand this
the craters generated on the copper anode, where the crater phenomenon, single discharges are generated inside a 5 mm
size decreases with increasing discharge duration and a carbo- deep cavity in steel cathode using Ø1 mm graphite anode. The
naceous layer is generated in the discharge region for longer use of a cavity creates a scenario where limited dielectric oil
pulse durations. The erosion crater volume on the copper is available in the discharge region. It can be seen in figure 24
anode is much higher than the graphite anode crater volume. that the craters generated inside the 5 mm deep cavity have
This may be due to the high sublimation temperature and the different geometries, some having higher volume H compared
lower thermal conductivity of graphite compared to copper. to the others with lower volumes L. In fact, the craters are
Using the heat source expansion calculated using equation (1), similar to the craters generated using the same electrode mate-
the crater geometry has been simulated and compared with the rials in air. Also, this effect is more pronounced when using
measured values for 20 A discharges generated between steel micro scale anode geometries (Ø1 mm, Fc: 0.08) compared to
cathode (point) and copper anode (plane) as seen in figure 23. the macro scale geometries (Ø5 mm, Fc: 0.15).
Further work is required for precise determination of Fa, as In case of the multiple discharges, one must consider the
the re-solidified layer thickness in copper anode has not been ratio of the discharges taking place in liquid, inside gas bubble
analysed. The decreasing anode crater volumes with discharge from the previous discharges and on liquid-gas interface.
time may indicate differing mechanisms for the anode and Also, the reduced instances of gas bubble implosion need to
cathode, such as plasma expansion rates and energy fractions. be considered for the micro scale anode geometry. It should be
noted that the Fc values differ less when using copper anode
4.3. Multiple discharges with micro (Ø1 mm, Fc: 0.11) and macro (Ø5 mm, Fc: 0.18)
dimensions. This effect may be attributed to the larger gas
The erosion craters generated by single discharges and by bubble volumes generated when using graphite anode com-
the multiple discharge process may differ due to the dynamic pared to the copper anode. The accumulation of the gas bub-
conditions in the gap region such as accumulation of gas bub- bles in the discharge region may lead to the discharges with
bles, debris and changes in the inter-electrode gap distance. a higher metal vapour content, which may cause different
This may explain the different energy faction Fc values used energy fractions into the electrodes. Thus, the PFE of a single
for simulations by various authors as presented in table 1. In discharge, a macroscopic flushing efficiency for the multi-
order to consolidate these values, the crater simulations for discharge process and differing cathode energy fractions must
single discharge and multi-discharge process are compared. be considered for the simulations of cathode crater volumes
For the multi-discharge process, the mean crater volume Vm in the multi-discharge process. An additional consideration of
is derived by dividing the eroded volume by the number of the carbonaceous layer formation is required for the simula-
discharges. In case of the multiple discharge process, the tion of anode craters.

12
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

3.5E-04 70 70
Measured V- Measured Measured
Crater volume V- (mm3)

Crater depth d (µm)

3.0E-04 60 60
Fc=0.29 Fc=0.29
a Fc=0.29

Crater radius r (µm)

a a
2.5E-04 50 50
Fc=0.285
a Fc=0.285
a Fc=0.285
a
2.0E-04 40 40
1.5E-04 30 30
1.0E-04 20 20
5.0E-05 10 10
0.0E+00 0 0
0 100 200 300
0 100 200 300 0 100 200 300
Time (µs) Time (µs) Time (µs)

Figure 23. Comparison of the measured and simulated crater geometries on copper anode for 20 A discharges.

Figure 24. Left: craters generated on a steel cathode plane using 20 A discharges with a Ø1 mm graphite anode inside a 5 mm deep cavity
filled with dielectric oil. Right: craters generated on a steel cathode by 20 A discharges using a Ø1 mm graphite anode with air as the dielectric.

5. Conclusions and outlook anode. This may be attributed to higher energy dissipation
into copper anode due to its lower boiling temperature and
A coherent overview of the post-breakdown phase of low higher thermal conductivity. Also, the erosion craters on steel
current electric discharges in micro-gaps filled with a liquid cathode are larger compared to Inconel 718 and TiAl6V4 as
hydrocarbon dielectric is provided through a systematic inves- the cathode materials. Since the thermal conductivity of steel
tigation of discharge plasma and erosion craters. A few mil- is considerably higher than Inconel 718 and TiAl6V4, higher
lijoule energy pulse discharges are generated in point-to-plane energy dissipation into steel cathode may be expected. The re-
electrode configuration, where graphite and copper are used solidified layer thickness has been found to be increasing with
as point anode materials and steel, Inconel 718 and TiAl6V4 the discharge duration and found to be similar for the tested
are selected as plane cathode materials. The following insights cathode materials.
may be derived from the presented results: Simulations of the erosion craters using the heat source
The thermal properties of the electrode materials influence expansion using the constant cathode boiling temperature con-
the energy balance and hence the temporal characteristics of dition show good agreement with the observed craters. Also,
the single discharges. the cathode energy fraction values differ based on the cathode
Higher gas bubble volumes are observed in the case of and anode material properties. The cathode energy fraction
graphite anodes compared to the copper anodes. This may is highest for steel, followed by Inconel 718 and TiAl6V4.
be attributed to the higher anode energy fraction for copper A precise analysis of the re-solidified layer is required to sepa-
due to its lower boiling temperature and higher thermal con- rately determine energy fraction values and plasma flushing
ductivity. Also, lower thermal conductivity and higher boiling efficiencies for different discharge conditions and electrode
temperature of TiAl6V4 may result in higher dielectric energy materials.
fraction compared to steel and Inconel 718, leading to larger Crater simulations for the single discharges and the multi-
gas bubble volumes. discharge process require different energy fraction values.
The average plasma properties such as electron density and The cathode energy fraction Fc ranging from 0.40–0.20 for
electron temperature do not change significantly for different the single discharge conditions is reduced to 0.15–0.10 for
investigated anode and cathode materials. However, using the the multiple discharge conditions, mainly due to a large pro-
single discharge analysis, temporal changes in the plasma portion of discharges taking place inside the gas bubble and
properties and the plasma species have been observed based reduction of the gas bubble implosion instances. A combined
on the electrode materials. Effect of the gas bubbles oscilla- effect of the electrode energy fractions dependent on the
tions and metal vapour density on the electron density of the plasma state and the plasma flushing efficiencies due to the
plasma needs to be evaluated using concurrent imaging and gas bubble dynamics needs further analysis.
plasma spectroscopy. Further in-depth investigations are required using the same
The erosion craters on cathode are affected by the anode and different material pairs for anode and cathode in order
material properties. Higher crater volume on steel cathode is to construct a coupled quantitative model of the energy bal-
observed in case of the graphite anode compared to copper ance in the discharge region. Similar analysis of the electric

13
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

discharges with nanosecond durations and micro-Joule ener- nitrogen dielectric medium IEEE Trans. Plasma Sci.
gies which are used for surface finishing and precision in 40 853–62
[12] Adineh V R, Coufal O and Zivny O 2012 Thermodynamic
EDM is also sought. Simulation of the visual emission spectra
and radiative properties of plasma excited in EDM process
can be a powerful tool to investigate interesting features of the through N2 taking into account Fe IEEE Trans. Plasma Sci.
EDM discharge plasma. It can be used to precisely identify 40 2723–35
the spectral lines and validate the plasma properties obtained [13] Kojima A, Natsu W and Kunieda M 2008 Spectroscopic
from spectroscopy measurements. In addition, plasma opacity measurement of arc plasma diameter in EDM CIRP Ann.
57 203–7
and the degree of ionization of elements can be analysed.
[14] Dijck FSv and Dutré W L 1974 Heat conduction model for
the calculation of the volume of molten metal in electric
Acknowledgments discharges J. Phys. D: Appl. Phys. 7 899
[15] Zolotykh B 1960 The mechanism of electrical erosion of
metals in liquid dielectric media Sov. Phys.—Tech. Phys.
The authors wish to acknowledge the financial support by the
4 1370–3
Swiss National Science Foundation (SNF) Project nr. 140823. [16] Zingerman A 1959 Regarding the problem of the volume of
The authors would also like to thank Mr N Gulde from Roper molten metal during electrical erosion Sov. Phys.—Solid
Scientific and GF Machining Solutions (AgieCharmilles SA) State 1 255–60
for their support regarding the equipment and know-how. [17] Xia H, Kunieda M and Nishiwaki N 1996 Removal amount
difference between anode and cathode in EDM process
The authors also acknowledge the support by the Electron
Int. J. Electr. Mach. 1 45–52
Microscopy Center of the ETH Zurich (EMEZ), especially Dr [18] Perez R, Rojas H, Walder G and Flükiger R 2004 Theoretical
K Kunze. The authors would like to thank Ms B Wegmann modeling of energy balance in electroerosion. J. Mater.
from LMPT of ETH Zurich for help with the metallographic Process. Technol. 149 198–203
analysis. [19] Revaz B et al 2005 Local temperature response to pulsed
discharges in electronic discharge machining (EDM)
environment IEEE Trans. Plasma Sci. 33 1066–71
References [20] Yeo S H, Kurnia W and Tan PC 2008 Critical assessment and
numerical comparison of electro-thermal models in EDM
J. Mater. Process. Technol. 203 241–51
[1] Bruggeman P and Brandenburg R 2013 Atmospheric pressure [21] Hinduja S and Kunieda M 2013 Modelling of ECM and EDM
discharge filaments and microplasmas: physics, chemistry processes CIRP Ann. 62 775–97
and diagnostics J. Phys. D: Appl. Phys. 46 464001 [22] Hockenberry T O and Williams E M 1967 Dynamic
[2] Eckman P K and Williams E M 1960 Plasma dynamics in an evolution of events accompanying the low-voltage
arc formed by low-voltage sparkover of a liquid dielectric discharges employed in EDM IEEE Trans. Ind. Gen. Appl.
Appl. Sci. Res. 8 299–320 IGA-3 302–9
[3] Hayakawa S, Yuzawa M, Kunieda M and Nishiwaki N 2001 [23] Hayakawa S, Sasaki Y, Itoigawa F and Nakamura T 2013
Time variation and mechanism of determining power Relationship between occurrence of material removal and
distribution in electrodes during EDM process Int. J. Electr. bubble expansion in electrical discharge machining Proc.
Mach. 6 19–25 CIRP 6 174–9
[4] Eubank P T, Patel M R, Barrufet M A and Bozkurt B 1993 [24] Zhang Y, Liu Y, Shen Y, Ji R, Li Z and Zheng C 2014
Theoretical models of the electrical discharge machining Investigation on the influence of the dielectrics on the
process: III. The variable mass, cylindrical plasma model material removal characteristics of EDM J. Mater. Process.
J. Appl. Phys. 73 7900–9 Technol. 214 1052–61
[5] Mujumdar S S, Curreli D, Kapoor S G and Ruzic D [25] Yang X, Guo J, Chen X and Kunieda M 2011 Molecular
2014 A model of micro electro-discharge machining dynamics simulation of the material removal mechanism in
plasma discharge in deionized water J. Manuf. Sci. Eng. micro-EDM Precis. Eng. 35 51–7
136 031011 [26] Timko H et al 2010 Mechanism of surface modification in the
[6] Albinski K, Musiol K, Miernikiewicz A, Labuz S and plasma–surface interaction in electrical arcs Phys. Rev. B
Malota M 1996 The temperature of a plasma used in 81 184109
electrical discharge machining Plasma Sources Sci. [27] Pandey P C and Jilani S T 1986 Plasma channel growth and
Technol. 5 736 the resolidified layer in EDM Precis. Eng. 8 104–10
[7] Descoeudres A, Hollenstein C, Wälder G and Perez R 2005 [28] DiBitonto D D, Eubank P T, Patel M R and Barrufet M A 1989
Time-resolved imaging and spatially-resolved spectroscopy Theoretical models of the electrical discharge machining
of electrical discharge machining plasma J. Phys. D: Appl. process: I. A simple cathode erosion model J. Appl. Phys.
Phys. 38 4066 66 4095–103
[8] Nagahanumaiah N A, Ramkumar J, Glumac N, Kapoor S G [29] Patel M R, Barrufet M A, Eubank P T and DiBitonto D D
and DeVor R E 2009 Characterization of plasma in micro- 1989 Theoretical models of the electrical discharge
EDM discharge using optical spectroscopy J. Manuf. machining process: II. The anode erosion model J. Appl.
Process. 11 82–7 Phys. 66 4104–11
[9] Kanmani Subbu S, Karthikeyan G, Ramkumar J and [30] Zahiruddin M and Kunieda M 2012 Comparison of energy and
Dhamodaran S 2011 Plasma characterization of dry μ-EDM removal efficiencies between micro and macro EDM CIRP
Int. J. Adv. Manuf. Technol. 56 187–95 Ann. 61 187–90
[10] Braganca I M F, Rosa P A R, Dias F M, Martins P A F and [31] Singh H 2012 Experimental study of distribution of energy
Alves L L 2013 Experimental study of micro electrical during EDM process for utilization in thermal models Int. J.
discharge machining discharges J. Appl. Phys. 113 Heat Mass Transfer 55 5053–64
233301–14 [32] Shabgard M, Ahmadi R, Seyedzavvar M and Oliaei S N B
[11] Adineh V R 2012 Net emission coefficient of plasma excited 2013 Mathematical and numerical modeling of the effect
in the electrical discharge machining through liquid of input-parameters on the flushing efficiency of plasma

14
J. Phys. D: Appl. Phys. 48 (2015) 055202 U Maradia et al

channel in EDM process Int. J. Mach. Tools Manuf. [36] Anthony B M 2013 Influence of metal vapour on arc
65 79–87 temperatures in gas–metal arc welding: convection versus
[33] Zhang Y, Liu Y, Shen Y, Li Z, Ji R and Cai B 2014 A novel radiation J. Phys. D: Appl. Phys. 46 224004
method of determining energy distribution and plasma [37] Maradia U, Knaak R, Boos J, Boccadoro M, Stirnimann J and
diameter of EDM Int. J. Heat Mass Transfer 75 425–32 Wegener K (ed) 2013 EDM process analysis using high-
[34] Descoeudres A, Ch H, Demellayer R and Wälder G 2004 speed imaging Proc. 13th Int. Conf. of the European Society
Optical emission spectroscopy of electrical discharge for Precision Engineering & Nanotechnology (Berlin, May
machining plasma J. Phys. D: Appl. Phys. 37 875 2013) pp 39–42
[35] Schnick M, Füssel U, Hertel M, Spille-Kohoff A and [38] Meek J M and Craggs J D 1978 Electrical Breakdown of
Murphy A B 2010 Metal vapour causes a central minimum Gases (New York: Wiley)
in arc temperature in gas–metal arc welding through [39] Schumacher B M 2004 After 60 years of EDM the discharge
increased radiative emission J. Phys. D: Appl. Phys. process remains still disputed J. Mater. Process. Technol.
43 022001 149 376–81

15