Journal of Materials Processing Technology 214 (2014) 892898

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

journal homepage: www.elsevier.com/locate/jmatprotec

Experimental investigation of spark generation in electrochemical

discharge machining of non-conducting materials
Baoyang Jiang a , Shuhuai Lan a, , Jun Ni a , Zhaoyang Zhang b
a
Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI 48109, USA
b
School of Mechanical Engineering, Jiangsu University, Zhenjiang, Jiangsu 212013, China

a r t i c l e i n f o a b s t r a c t

Article history: Electrochemical discharge machining (ECDM), also known as spark assisted chemical engraving (SACE), is
Received 6 July 2013 an effective micro-machining process for non-conducting materials. Process modeling of ECDM, includ-
Received in revised form ing spark generation and material removal, is not fully established however. Empirical estimation for
20 September 2013
discharge energy results in large prediction error of material removal and is hard to experimentally
Accepted 7 December 2013
validate. In this paper, an experiment-based stochastic model for spark energy estimation is presented.
Available online 15 December 2013
Tapered tool electrodes were fabricated by electrochemical machining (ECM) to improve the consistency
of spark generation. Energy of sparks was experimentally determined and t into a two-component mix-
Keywords:
Electrochemical discharge machining
ture log-normal distribution to reveal electrochemical characteristics of tool electrodes. A nite element
(ECDM) based model was established to correlate spark energy and the geometry of removed material. Material
Spark assisted chemical engraving (SACE) removal was treated as heat transfer problem because electrical energy released by spark generation
Tapered tool transfers into thermal energy on the workpiece, resulting in material removal due to thermal melting
Spark energy and chemical etching. Predictions of material removal by the model demonstrated good consistency with
Stochastic model experimental results.
Material removal 2013 Elsevier B.V. All rights reserved.

1. Introduction with respect to machining efciency and accuracy. Mochimaru et al.

The demand of non-conductive material has grown rapidly with
the broad application in optical, electrical, and mechanical sys-
tems. Glass is expansively employed due to the properties including
optical transparency, high stiffness, and good chemical resistance.
Electrochemical discharge machining (ECDM), taking advantage of
electrochemical discharge phenomenon, is a non-traditional micro
machining process. Non-conducting materials can be machined by
ECDM by employing auxiliary electrodes. This process is also known
as spark assisted chemical engraving (SACE). Features with high
aspect ratio and complicated geometry can be created by ECDM
(Zheng et al., 2007).
Improving the machining efciency and quality has long been
the concentration for studies of ECDM of non-conducting mate-
rials. To achieve a better machining process, innovative methods
were introduced into conventional ECDM, including developing
new procedures and machining tools. Wuthrich and Hof (2006)
applied tool vibration and experimentally demonstrated the incre-
ment in the material removal rate. Yang et al. (2011) fabricated
a spherical tool electrode and showed the merits of the new tool
Corresponding author. Tel.: +1 734 763 7119.
E-mail addresses: byjiang@umich.edu (B. Jiang), lans@umich.edu (S. Lan),
junni@umich.edu (J. Ni), zhangzhaoyang73@gmail.com (Z. Zhang).
(2012) introduced a two-step machining method to reduce the
over-cutting in drilling process.
Conventional ECDM, as the fundamental of all innovative
machining techniques, is not fully revealed and optimized. The
machining quality by ECDM is sensitive to many process param-
eters. However, it is unclear that which of the parameters affect
the machining quality and how these parameters can be optimized
(Wuthrich and Fascio, 2005). Process modeling is an important
approach of understanding the mechanism of ECDM. Studies of
process modeling involve discharge phenomenon, gas lm for-
mation, and material removal (Wuthrich, 2009). Basak and Ghosh
(1996) estimated the material removal rate by evaluating energy,
and frequency of spark generation (Basak and Ghosh, 1997). Jalali
et al. (2009) quantitatively characterized material removal due to
thermal heating and chemical etching and estimated the machin-
ing temperature. Finite element analysis was widely employed
in modeling. Jain et al. (1999) developed a nite element based
heat transfer model to estimate the thermal effects of single spark.
Bhondwe et al. (2006) established a nite element model to predict
material removal rate.
Experimental verication for process modeling of ECDM, how-
ever, is often indirect and inaccurate. There are various reasons that
results in such difculty. First, the discharging and material removal
process is very sensitive and is vulnerable to changes in ambient
condition, which caused low repeatability in experiments. Second,

0924-0136/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jmatprotec.2013.12.005
 B. Jiang et al. / Journal of Materials Processing Technology 214 (2014) 892898 893

- tool
power + chuck
supply
electrolyte

auxilary
electrode

workpiece fixture

Fig. 1. Experimental setup of ECDM.

real-time observation is not feasible for micro-scale machining,

especially in high temperature alkali solution. Sensors installed in
the machining area could interfere with the original setup and thus
change the electrochemical and uid dynamic characteristics of the
process.
In this paper, process modeling for spark generation and
material removal is presented. Spark energy was experimentally
determined and t into a stochastic model. A tapered tool electrode
was fabricated and employed in the experiments to increase the
consistency of spark generation. A nite element model was devel-
oped to correlate spark energy to material removal. Experiments
were delivered to validate the models.

2. Experiment preparation

An experimental setup was designed and fabricated to imple-

ment ECDM (Fig. 1). A machining chamber consists of chemicals
was xed to a CNC machine, with the tool electrode clamped on the
machine. The power supply could provide DC or pulse output from
0 to 64 V in the range of 010 A current. Electrode voltage directly
affects the machining efciency. If the electrode voltage is too high,
thermal cracks tend to happen (Wuthrich et al., 2006), while a min-
imum voltage must be maintained to ignite the electrochemical
reaction (Maillard et al., 2007). In most of the works presented in
this paper, the output was set to be 34 V if not specically noticed.
The workpiece was a sheet of soda-lime glass with 1 mm thickness,
which consisted of 74 wt.% SiO2 , 13 wt.% Na2 O, 10.5 wt.% CaO.
The process characteristics of ECDM are closely related to the
properties of tool electrode (Mousa et al., 2009). Cylindrical tools,
as are conventionally used in ECDM process, do not have desired
electrochemical characteristics in the study of discharging activity.
Fig. 2 is the picture of a crater created by ECDM process using a
0.5 mm diameter cylindrical tool with the discharging duration of
Fig. 2. Crater created by a cylindrical tool electrode. Machined with 35 V electrode
voltage and 2 s machining time.
Fig. 3. Finite element simulation of current density in electrochemical reaction. (a)
Cylindrical electrode. (b) Tapered electrode.
2 s. Fig. 2 illustrates that material near the rim of the cylindrical
tool was removed, indicating release of sparks were distributed
around the rim, or fringing effect. By replacing the cylindrical tool
with a tapered end tool, however, could make discharges focus on
a concentrated region.
A nite element simulation was deployed to investigate elec-
trochemical reaction in the machining process. Tool electrode was
chosen to be tungsten with 250 m diameter and the electrolyte
was 30 wt.% NaOH. Electrode voltage was set to 30 V. The simula-
tion proved that fringing effect existed for cylindrical tools, while
tapered tool has only one spot for spark generation. Fig. 3 shows
the current density around the tool electrodes (brighter color rep-
resents higher current density). It can be seen that current density
is more intense near the rim for cylindrical tools. Thus, sparks tend
to generate from any point along the rim of the tool electrode, if not
considering and geometrical defects. However, sparks can only be
generated at the tip of tapered tools, and thus increased consistency
in spark generation.
Electrochemical machining (ECM) was facilitated to fabricate
tools with tapered end (Fig. 4). Tapered tools were made from
tungsten rods with 250 m diameter. The rods were cut to approx-
imately one inch long and clamped on the chuck. Tool and chuck
were connected to the power supply as the anode in ECM pro-
cess. The auxiliary electrode was made of stainless steel. A hole
was drilled on the center of auxiliary electrode, surrounding the
tool electrode to create an axisymmetric electrical eld in the elec-
trochemical reaction. The auxiliary electrode was immersed 1 mm
under the upper surface of the electrolyte (8 wt.% NaOH).
894 B. Jiang et al. / Journal of Materials Processing Technology 214 (2014) 892898

Fig. 4. Tool electrode fabrication by ECM.

By adjusting electrode voltage and the vertical position of the

tool in ECM process, tool electrodes with different geometry could
be fabricated (Fig. 5). Chiou et al. (2012) empirically modeled ECM
process for high-aspect micro-rod to correlate machined geome-
try, including width and shape of the rod, with machining time
Fig. 6. Current output of DC power supply in discharging regime. Electrode voltage
ranged from 0 to 80 min. Mochimaru et al. (2012) studied ECM of was 35 V.
micro-tool electrode by speculating micro-ow from variation of
tool geometry in experiments. Tool electrode, made of tungsten,
sparks and evaluate the effect of geometry of tool electrode based
acted as the anode in electrochemical reaction. Tool electrode dis-
on the experimental measurements and observations.
solved faster at the rim of the auxiliary electrode (cathode) because
of higher intensity of electrical eld. Tools with different shape
3.1. Discharge phenomenon
could be fabricated by changing the relative position to the cathode.
As a result, tool electrode fabricated could either have long tapered
The output current from the power supply could track the dis-
tips with high aspect ratio, or have short tapered ends with high
charge phenomenon during the machining process. Once electrode
stiffness. Some other geometry (such as two-section tools) could
voltage was applied, spark generation could be clearly recognized
also be achieved by choosing process parameters appropriately.
from the sudden increments in current output (Fig. 6). Every single
impulse represents a spark generation process. It can be seen from
3. Spark generation with tapered tool Fig. 6 that there existed a small portion of constant current that
maintained the chemical reaction (about 0.1 A). However, the cur-
The discharge activity plays a key role in ECDM. Spark genera- rent output instantly reached approximately 0.30.8 A when sparks
tion with cylindrical tools, however, is complicated due to fringing were released.
effect. Tapered tools are more consistent in spark generation and It can be noticed that current output peaked at 1 A and main-
can be modeled more accurately. In this section, quantitative study tained at a high level for approximately 15 ms after power was
of spark generation with both cylindrical and tapered tools was turned on, indicating that gas lm had not been fully developed
presented. A stochastic model was developed to estimate energy of and the electrochemical reaction was not in discharging regime. In

Fig. 5. Tapered tool electrodes with different shape.

 B. Jiang et al. / Journal of Materials Processing Technology 214 (2014) 892898 895

Fig. 7. Current waveform of a typical spark generation process.

this case, even if the electrode voltage was set to be higher than the
critical voltage for discharging, sparks were not generated instantly
due to the instability of gas lm in the transient phase. In the exper-
iments, the minimum pulse width of power supply (if using pulse
output) was set higher than 30 ms to ensure a safety factor of 2.

3.2. Estimation of single spark energy

In ECDM process, electrical power of sparks generated is a direct

measure of machining power transferred to workpiece. The elec-
trical power of sparks can be derived from voltage and current
measured in the experiment (1).

q= u(t)i(t)dt (1)

Energy of sparks can be explicitly calculated using (1). The cur-

rent waveform (Fig. 7) indicates the duration of single sparks was
around 0.10.2 ms. The current sensor employed had a measur-
ing bandwidth of 200 kHz, resulting in a resolution of 5 s, and
therefore preventing aliasing in current sensing.
However, uncertainties were involved in discharging activity
during ECDM due to high instability of gas lm. Duration and
peak current varied among discharges, and therefore changed the
energy releasing rate. A histogram revealed the distribution of
energy released by sparks with a tapered tool electrode (Fig. 8(a))
(energy was labeled in logarithm scale). As a benchmark, energy
histogram of sparks generated by a cylindrical tool was presented
(Fig. 8(b)). Fig. 8 illustrated that energy of sparks generated by
tapered tool electrode distributed around one value, while the
distribution peaked twice in the case of using cylindrical tool elec-
trode.
Fig. 8 suggested that energy of sparks followed 2-component
mixture log-normal distribution with form shown in (2).
 
1 (log q 1 )2
P(log q) = p1 exp
1 2 21 2
 
1 (log q 2 )2
+ p2 exp
2 2 22 2
Fig. 8. Energy distribution of sparks. (a) Tapered tool electrode. (b) Cylindrical tool
electrode.
Fig. 8 illustrates that tapered tools created more consistent
sparks comparing with at-end tools. To be specic, the proportion
of sparks with low energy was larger when using at-end tools
(reected by tting parameters in Table 1), indicating two possible
discharging releasing mechanisms. For a cylindrical tool electrode,
sparks tend to release from the edge because of the higher electrical
eld intensity. Meanwhile, minor discharges could take place at
the at section of the tool, corresponding to the minor peak in the
distribution. By using the tapered tool electrodes, however, could
eliminate the minor discharging effect, resulting in uniform sparks.
Considering the single spark energy releasing model, the spark
energy follows a log-normal distribution if not accounting for
sparks with very low energy. Considering the fact that sparks with
low energy could not remove material from the workpiece, it is
Table 1
Fitting parameters of energy distribution of single spark.
(2)
Tapered Flat

p1 0.57 0.41
In (2), q is the energy of single spark, while the function P (log q) p2 0.42 0.58
gives an estimation of the probability of spark generation given 1 5.56 4.67
certain energy. p1 , p2 , 1 , 2 ,  1 ,  2 are parameters of tting, as 2 6.74 5.86
1 0.28 0.34
shown in Table 1. p1 , p2 are weights of log-normal distributions,
2 1.16 1.48
1 , 2 ,  1 ,  2 are average and standard deviation, respectively.
896 B. Jiang et al. / Journal of Materials Processing Technology 214 (2014) 892898

reasonable to neglect sparks with energy lower than 30% of the

mean energy of all sparks. Therefore it is reasonable to neglect the
component in mixture log-normal function with smaller energy.
Specically, the mean spark energy is exp (1 ), i.e. 3.8 mJ.

4. Modeling material removal

The energy transferred to the material from discharge activity

was converted into heat. Since discharges took place at the tip of
the tool electrode, energy transfer can be modeled as a heat source
acting raising the temperature on surface of the workpiece. If the
temperature is high enough, material will be removed.
In ECDM, material was not only removed from the workpiece
by melting, but also from chemical reactions with the electrolyte.
As NaOH was chosen as the electrolyte, the following reaction dis-
solves glass workpiece and generates loose precipitate:

2NaOH + SiO2 Na2 SiO3 + H2 O (3) Fig. 9. Finite element simulation of material removal.

The reaction rate is considerable under high temperature (Jalali

et al., 2009). Although other mechanisms might also be involved
in material removal, including releasing thermal stress and shock-
waves created by uid motion, thermal melting and chemical
etching are considered dominating in ECDM process (Wuthrich and
Fascio, 2005).
In ECDM for conducting workpiece, material is removed by
both discharge and electro-chemical reaction. However, electro-
chemical reaction does not take place if the workpiece is
non-conducting. Instead, chemical etching of glass is consider-
able as the temperature increase. Given the fact that material
was removed by melting and etching, material can be considered
removed under a critical temperature. The critical temperature
should be lower than melting temperature of glass due to the efforts
of chemical etching. It was estimated that the critical temperature
of glass is around 600 C (Jalali et al., 2009). In this case, material
removal can be simulated by solving heat transfer problems.
Distribution of energy in spark generation affects the geometry
of craters on surface of the workpiece. The shapes of craters are
usually parabolic, judging from microscopic images, if tool elec-
trode is set stationary (Yang et al., 2010). A Gaussian distribution of
heat source is a good implementation of the transmission of spark
energy (Bhondwe et al., 2006), as is shown in (4).
 
1 r2
q(r) = cqs exp
 2 2 2
can reasonably predict the depth and diameter of machined holes
(Kulkarni et al., 2003).
4.1. Finite element analysis
A nite element model was deployed in software COMSOL Mul-
tiphysics 4.3. The problem was treated as heat transfer in solids.
A disk of 400 m diameter and 50 m thickness was constructed
with the axisymmetric assumption. A Gaussian heat source acting
on the top surface a disk. The other surfaces were set to have free
convective heat transfer, with the coefcient of 50 W/m2 K. Energy
of sparks was determined to follow log-normal distribution. Mate-
rial removal can be simulated by taking machining energy as an
input parameter. Since energy was measured in experiments, the
recorded values were taken in FEA models.
Material properties of glass involved were density (2520 kg/m3),
heat capacity (640 J/kg K), and thermal conductivity (1.05 W/m K).
However, once the temperature of material reached the criti-
cal temperature, 600 C, the element was replaced with air, with
density (1.2 kg/m3 ), heat capacity (1000 J/kg K), and thermal con-
ductivity (0.024 W/m K).
Simulation illustrated the material removed to have parabolic
geometry (Fig. 9), which was consistent with experimental obser-
vations.
(4)

In (4), qs is the total energy of a spark, c is the coefcient of

energy transmission. The diameters of craters created were around
150 m, as was experimentally determined (Kulkarni et al., 2003).
As the cumulative probability of Gaussian distribution reaches
99.7% in the range of 3,  was taken as 50 m. Energy of a spark
was taken as the mean value the experimental measurements.
The energy released from spark, however, did not fully transfer
to the workpiece. The reason is that sparks were released between
tool electrode and electrolyte, and did not directly act on the work-
piece. The transferring coefcient, c, was estimated to be 0.291 (Wei
et al., 2011).
Treating material removal as heat transfer problem, however, is
a simplication of the complicated material removal mechanism
of ECDM process. Discharge phenomenon generates arc plasma
and has different characteristics as heat source. In addition, mate-
rial removal is related to micro-ow of electrolyte and heat was
transferred through both liquid and gas phase. Therefore model-
ing material removal as steady state heat transfer process cannot
reveal the mechanism and geometry of removed material (Hinduja Fig. 10. Microscopic pictures of a tapered tool electrode. (a) Initial status. (b) After
and Kunieda, 2013). However, such empirical heat transfer models 1.0 s discharging under 34 V electrode voltage.
 B. Jiang et al. / Journal of Materials Processing Technology 214 (2014) 892898
Fig. 11. Experimental results and simulation of material removal. Total energy consumption: 0.321 J in (a) and 0.659 J in (b). (For interpretation of the references to color in
text, the reader is referred to the web version of the article.)
4.2. Comparison with experiments
Drilling experiments with tapered tool electrode were carried
out. Tapered tool electrodes were immersed 1 mm in the elec-
trolyte and the tips maintained a 50 m gap from the workpiece.
34 V DC voltage was applied to the electrodes, with machining time
ranging from 0.1 s to 1.0 s in order to vary total energy released to
workpiece.
Tool wear affect the quality of ECDM because discharge and gas
lm formation would become unpredictable. Unfortunately, tool
wear is generally not measurable during the machining process
(Abou Ziki and Wuthrich, 2012).Tool wear of tapered tool electrode
was a serious issue if machining time was long, because sparks were
always released from the tip. The temperature of the tip rose very
quickly and the tool would be worn out once melting temperature
of tungsten was reached. However, a tapered tool could survive
under short machining time. Pictures of a tapered tool before and
after 1.0 s discharging were taken (Fig. 10). Although there were few
amount of precipitates attached to the tool, the tapered remained
sharp and undamaged, indicating negligible tool wear within the
1.0 s time period. From this point, it can be concluded that tool wear
had minor impact in experiments since the maximum machining
time in experiments were 1.0 s.
Energy released through discharging was recorded by voltage
and current sensors. Finite element simulation using heat source
with same energy (multiplied by coefcient of energy transfer) was
carried out. Fig. 11(a) and (b) show SEM pictures of two machined
897
holes with total energy of 0.321 J and 0.659 J. Workpieces were
tiled by 60 when installed into SEM. Diameter and depth of the
holes were measured and recorded. Fig. 11(c) and (d) shows the
simulation results, with the energy input set up according to the
actual values respectively. Only half of the workpiece was shown
in the cross-sectional view due to the axisymmetric assumption. In
Fig. 11(c) and (d), red area indicates the region of removed material.
The gures demonstrated reasonable consistency between exper-
imental and simulation outcomes in terms or crater radius and
depth. It can be also noticed that the surface of machined hole
was smooth, indicating chemical etching took a signicant role in
material removal.
5. Conclusion
In this paper, works on process modeling of ECDM with respect
to spark generation and material removal were presented. Tapered
tool electrodes were employed as tool electrode in the study of
spark generation. Fabricated by ECM, tapered tool electrodes could
enhance consistency in spark generation comparing with conven-
tional tools. Energy of each spark generated was measured and
was t in a stochastic model with a two-component mixture log-
normal distribution. The energy distribution proved that tapered
tool improved the consistency of spark generation and suppressed
the generation of minor discharges. The average energy for sparks
was 3.8 mJ with 34 V electrode voltage. A nite element based
model was developed to predict the geometry of removed material
898 B. Jiang et al. / Journal of Materials Processing Technology 214 (2014) 892898
given specic spark energy input. Electrical energy transferred
converted into heat source acting on the workpiece in machining
process. Therefore material removal can be simulated by solving
heat transfer problems. It can be concluded from the comparison
with experimental results that the prediction of material removal is
reasonable in terms of diameter and maximum depth of machined
holes.
Acknowledgements
This work is sponsored by National Science Foundation (grant
no. 1068935, high-performance micromachining of glass using
electrochemical discharge machining (ECDM) for MEMS applica-
tions).
References
Abou Ziki, J.D., Wuthrich, R., 2012. Tool wear and tool thermal expansion during
micro-machining by spark assisted chemical engraving. International Journal of
Advanced Manufacturing Technology 61, 481486.
Basak, I., Ghosh, A., 1996. Mechanism of spark generation during electrochemical
discharge machining: a theoretical model and experimental verication. Journal
of Materials Processing Technology 62, 4653.
Basak, I., Ghosh, A., 1997. Mechanism of material removal in electrochemical dis-
charge machining: a theoretical model and experimental verication. Journal of
Materials Processing Technology 71, 350359.
Bhondwe, K.L., Yadava, V., Kathiresan, G., 2006. Finite element prediction of material
removal rate due to electro-chemical spark machining. International Journal of
Machine Tools and Manufacture 46, 16991706.
Chiou, Y.C., Lee, R.T., Chen, T.J., Chiou, J.M., 2012. Fabrication of high aspect ratio
micro-rod using a novel electrochemical micro-machining method. Precision
Engineering 36, 193202.
Hinduja, S., Kunieda, M., 2013. Modelling of ECM and EDM processes. CIRP Annals
Manufacturing Technology 62, 775797.
Jain, V.K., Dixit, P.M., Pandey, P.M., 1999. On the analysis of the electrochemical spark
machining process. International Journal of Machine Tools and Manufacture 39,
165186.
Jalali, M., Maillard, P., Wuthrich, R., 2009. Toward a better understanding of glass
gravity-feed micro-hole drilling with electrochemical discharges. Journal of
Micromechanics and Microengineering 19.
Kulkarni, A., Sharan, R., Lal, G.K., 2003. Measurement of temperature transients in
the electrochemical discharge machining process. AIP Conference Proceedings
684, 10691074.
Maillard, P., Despont, B., Bleuler, H., Wuthrich, R., 2007. Geometrical character-
ization of micro-holes drilled in glass by gravity-feed with spark assisted
chemical engraving (SACE). Journal of Micromechanics and Microengineering
17, 13431349.
Mochimaru, Y., Ota, M., Yamaguchi, K., 2012. Micro hole processing using electro-
chemical discharge machining. Journal of Advanced Mechanical Design Systems
6, 949957.
Mousa, M., Allagui, A., Ng, H.D., Wuthrich, R., 2009. The effect of thermal conductivity
of the tool electrode in spark-assisted chemical engraving gravity-feed micro-
drilling. Journal of Micromechanics and Microengineering 19.
Wei, C.J., Xu, K.Z., Ni, J., Brzezinski, A.J., Hu, D.J., 2011. A nite element based model for
electrochemical discharge machining in discharge regime. International Journal
of Advanced Manufacturing Technology 54, 987995.
Wuthrich, R., 2009. Micromachining Using Electrochemical Discharge Phenomenon:
Fundamentals and Application of Spark Assisted Chemical Engraving. Elsevier,
Oxford, UK.
Wuthrich, R., Fascio, V., 2005. Machining of non-conducting materials using elec-
trochemical discharge phenomenon an overview. International Journal of
Machine Tools and Manufacture 45, 10951108.
Wuthrich, R., Hof, L.A., 2006. The gas lm in spark assisted chemical engraving
(SACE) a key element for micro-machining applications. International Journal
of Machine Tools and Manufacture 46, 828835.
Wuthrich, R., Spaelter, U., Wu, Y., Bleuler, H., 2006. A systematic characterization
method for gravity-feed micro-hole drilling in glass with spark assisted chem-
ical engraving (SACE). Journal of Micromechanics and Microengineering 16,
18911896.
Yang, C.K., Cheng, C.P., Mai, C.C., Wang, A.C., Hung, J.C., Yan, B.H., 2010. Effect of sur-
face roughness of tool electrode materials in ECDM performance. International
Journal of Machine Tools and Manufacture 50, 10881096.
Yang, C.K., Wu, K.L., Hung, J.C., Lee, S.M., Lin, J.C., Yan, B.H., 2011. Enhancement of
ECDM efciency and accuracy by spherical tool electrode. International Journal
of Machine Tools and Manufacture 51, 528535.
Zheng, Z.P., Su, H.C., Huang, F.Y., Yan, B.H., 2007. The tool geometrical shape and
pulse-off time of pulse voltage effects in a Pyrex glass electrochemical dis-
charge microdrilling process. Journal of Micromechanics and Microengineering
17, 265272.