See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267595975

Experimental Investigation on Electro-Discharge Deposition Process

Conference Paper · November 2013

DOI: 10.1115/IMECE2013-62866

CITATIONS READS

3 23

3 authors:

Murlidharan B H. Chelladurai
VIT University PDPM Indian Institute of Information Technology, Design and Manufacturing Jaba…
5 PUBLICATIONS   6 CITATIONS    16 PUBLICATIONS   26 CITATIONS   

SEE PROFILE SEE PROFILE

Janakarajan Ramkumar
Indian Institute of Technology Kanpur
168 PUBLICATIONS   1,054 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CONDITION MONITORING View project

Surface texturing on biocompatible Titanium alloy for inducing hydrophobicity using electrochemical micro-texturing View project

All content following this page was uploaded by Murlidharan B on 01 December 2016.

The user has requested enhancement of the downloaded file.

Experimental analysis of electro-discharge
deposition process

B. Muralidharan & H. Chelladurai

The International Journal of

Advanced Manufacturing Technology

ISSN 0268-3768
Volume 76
Combined 1-4

Int J Adv Manuf Technol (2015) 76:69-82

DOI 10.1007/s00170-014-5943-z

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag London. This e-offprint is for personal
use only and shall not be self-archived
in electronic repositories. If you wish to
self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

1 23
 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82
DOI 10.1007/s00170-014-5943-z

ORIGINAL ARTICLE

Experimental analysis of electro-discharge deposition process

B. Muralidharan & H. Chelladurai

Received: 10 October 2013 / Accepted: 8 May 2014 / Published online: 27 May 2014
# Springer-Verlag London 2014

Abstract The application of miniaturized products is slowly biotechnology etc., reported in Alting et al. [1]. In the present
increasing due to their demand in micro feature-based sys- trend, micro products are designed and produced with more
tems, which needs multitude of functions to be integrated. functionalities and compactness. To meet the above con-
Several lithographic-based technologies are used to manufac- straints and need for reduced size of products, various
ture high-aspect ratio micro features, which has limitations to micro-fabrication techniques are developed. Research find-
produce 3D features. Electro-discharge deposition (EDD) is a ings of micro-nanofabrication technologies like, lithography,
newly developing nonlithographic additive manufacturing FIB, etching technology and replication technology with their
process to produce micro products. The potential of this advantages, limitations (manufacturability, material, econom-
process is that any material which is conductive in nature ic feasibility and dimensions like 1D, 2D, 2.5D, and 3D) are
can be made to deposit over the substrate. In the present paper, discussed in Cui [2] and Brousseau et al. [3]. Micro products
the use of magnetic field in EDD process has been investigat- prone to aggressive environments and enhanced force actua-
ed. It is observed from the results that height and weight of tion are 3D in nature, which is reported in Brousseau et al. [3].
deposition are increased significantly by using magnetic field Dimov et al. [4] discussed that lithographic technologies can
and at the same time width is reduced due to plasma confine- extend their hands only upto 2D and 2.5D features with
ment by magnetic field effects. From the scanning electronic limited material usage. So, complex 3D micro features with
microscope (SEM) images it is observed that there is unifor- high aspect ratios, suitable to unstable environments can be
mity in deposition of the tool material over the substrate. It is produced by nonlithographic based technologies whether by
concluded from the experimental results that, magnetic field is adding or by removing the material to get the final product,
also a one of the significant contributing parameter assisting in reported by Rajurkar et al. [5]. The micro manufacturing can
electro-discharge deposition process. be done both by additive and subtractive process. Subtractive
process in micro scale especially in case of micro machines, it
is very difficult to handle micro tools, workpiece position,
Keywords Electro-discharge deposition process . Magnetic
normally all those positions and its conditions of fittings are
field . SEM . Design of experiments
monitored by using sensors, which consumes huge investment
cost. So, additive process is preferred to subtractive process in
micro scale. EDD process is one of the innovative deposition
1 Introduction methods in which reverse electrical discharge machining
(EDM) principle is used. In reverse EDM, electrodes are
The demand for micro products and its components are grow- connected in reverse polarity, i.e., tool is connected to positive
ing significantly, owing to wide applications in many fields polarity and substrate is connected to negative polarity. The
like, electronics, MEMS, medical, automobile and advantage of EDD is that any conductive material irrespective
of hardness can be used. Gangadhar et al. [6] initially inves-
tigated the transfer of material from tool to work piece during
B. Muralidharan (*) : H. Chelladurai
reverse EDM process. Simao et al. [7] and Goto et al. [8]
Discipline of Mechanical Engineering, PDPM IIITDM Jabalpur,
Jabalpur, MP, India reported that EDD process can be used for modifying the
e-mail: [email protected] surface of the material and it is extended to form anti wear
 Author's personal copy
70 Int J Adv Manuf Technol (2015) 76:69–82

Table 1 Factors and their levels

S.No Factors Levels

−2 −1 0 1 2

1. Current (I), A 3 9 14 19 25
2. Pulse on time (Ton), μs 7.5 20 30 40 50
3. Duty factor (t) 2 5 7 9 12
4. voltage (v),V 2 30 50 70 98
5. Magnetic flux (F), Tesla 0.04 0.07 0.16 0.20 0.40

pulse duration having less spatial current density. EDD pro-

cess is advantageous to any additive process such that, com-
plex 3D microstructures with free form surface can be
manufactured at optimum cost which is observed by
Fig. 1 Schematic diagram of deposition process Rajurkar et al. [5]. Chi et al. [11] fabricated 3D spiral structure
with brass electrode as a tool, steel as workpiece and air as a
dielectric medium. From spectrum analysis it is observed that
coating of hard materials by using reverse EDM principle. the composition of deposited material is almost same as tool
This idea is developed further and used for metal layer depo- material. Wang et al. [12] fabricated brass micro cylinders
sition, which leads to fabrication of 3D micro parts. In EDD, with approximately 200 μm in diameter and 1.2 mm in height
thin wire is selected as a tool so that tool is melted and which is deposited on high-speed steel. Jain et al. [13] have
deposited over the substrate. Mohri et al. [9] noted that the attempted for micro-fabrication of parts by using conventional
wear of a thin electrode is greater than that of a thick electrode EDM machine and factors (current, duty cycle, and pulse on
because of the difference in thermal resistivity. Mohri et al. time) are varied, in order to find the significant factors con-
[10] observed that higher tool wear occurs in shorter pulse tributing to the EDD process. It is realized that the presence of
having more spatial current density comparatively with longer magnetic field in EDD may enhance the deposition process by
forming uniform layer, increasing height, weight, and reduc-
ing the width of deposition. Heinz [14] reported that magnetic
field when used around plasma, discharge plasma channel will
be altered and the plasma particles will behave according to
electromagnetic field interaction and Lorentz force. Kazufumi
et al. [15, 16] observed plasma arc shape can be controlled by
the permanent magnet to get sound weld beads in a magne-
tized tungsten inert gas (TIG) arc welding process. Many
researches have been carried out in the area of welding using
magnetic field to confine the discharge of plasmas by Tayler
[17], Keidar et al. [18], Beilis et al. [19] and Rondanini et al.
[20] in order to improve the quality of welding. Still
now, limited literatures are available in the area of EDD
process, particularly using magnetic field to improve the
quality of deposition.

Fig. 2 Schematic diagram of EDD setup Fig. 3 EDD deposited layer

 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82 71

Table 2 Experimental design and responses Table 2 (continued)

Ex. no Input parameters Responses Ex. no Input parameters Responses

I Ton t v F H W WD I Ton t v F H W WD

1 14 30 7 50 0.16 7.46 480 1,067.77 48 9 20 9 70 0.2 6.292 610 694.79

2 19 20 5 70 0.07 7.523 360 759.8 49 19 40 5 30 0.2 5.997 880 1,172.52
3 19 40 5 70 0.2 10.86 390 1,305.23 50 14 30 2 50 0.16 4.613 640 975.35
4 19 40 9 70 0.07 11.59 520 1,173.51
5 9 40 5 30 0.07 5.34 440 966.44
6 14 30 7 50 0.16 7.67 700 1,203.45
In this paper, an attempt has been made to find the influ-
7 9 40 9 30 0.07 9.03 550 975.16 ence of magnetic field in EDD process as well as analyses has
8 19 20 5 30 0.2 7.47 540 1,019.43 been carried out to improve the quality of deposition of
9 19 40 2 70 0.07 7.435 750 1,151.1 conductive materials over the substrate by using reverse
10 14 30 7 98 0.16 6.36 950 1,020.38 EDM principle. The mechanism of magnetic field on deposi-
11 9 40 5 30 0.2 4.7 420 1,142.67 tion process is shown in Fig. 1. When a high potential is given
12 14 30 12 50 0.16 9.344 770 1,118.48 between cathode and anode, field emission occurs and by that
13 9 40 9 70 0.07 10.1 540 962.61 instant, electrons start moving from the cathode to anode. The
14 19 40 9 30 0.07 7.533 620 1,101.8 direction of movement of electron will be perpendicular to
15 14 30 7 50 0.16 9.173 490 1,157.805
16 25 30 7 50 0.16 7.594 480 1,255.96
Table 3 ANOVA for developed response surface from Eq. (2)
17 9 20 9 30 0.2 5.763 300 807.58
18 14 30 7 2 0.16 3.16 350 1,052.11 Source Sum of df Mean F value p value
19 9 40 5 70 0.2 6.438 890 1,077.24 squares square Prob > F

20 19 20 9 30 0.07 4.082 430 680.65

Model 231.4474 20 11.57237 2.276215 0.0213 Significant
21 19 20 5 30 0.07 5.22 550 736.54
Current (I) 0.09896 1 0.09896 0.019465 0.8900
22 14 30 7 50 0.16 5.913 530 984.97
Pulse on 8.416931 1 8.416931 1.65556 0.2084
23 9 40 5 70 0.07 5.576 510 1,025.12 (Ton)
24 14 30 7 50 0.16 13.989 670 1,065.4 Duty factor 1.683141 1 1.683141 0.331064 0.5695
25 9 20 5 30 0.2 3.548 380 637.5 (t)
Voltage (V) 16.25367 1 16.25367 3.197 0.0842
26 14 7.5 7 50 0.16 1.02 490 482.93
Magnetic 0.408711 1 0.408711 0.080391 0.7788
27 9 20 9 70 0.07 4.461 180 727.2
flux (F)
28 19 40 9 30 0.2 9.989 550 1,326.84 I Ton 0.013237 1 0.013237 0.002604 0.9597
29 19 20 9 70 0.07 4.587 510 825.13 It 6.017069 1 6.017069 1.183521 0.2856
30 19 40 5 30 0.07 7.066 660 1,199 IV 8.284707 1 8.284707 1.629552 0.2119
31 19 20 9 70 0.2 5.116 550 942.642 IF 1.771144 1 1.771144 0.348373 0.5596
32 19 20 5 70 0.2 5.861 670 827.49 Ton t 29.08534 1 29.08534 5.720911 0.0235
33 19 20 9 30 0.2 19.02 510 896.69 Ton V 2.906545 1 2.906545 0.5717 0.4557
34 14 30 7 50 0.4 10.467 1400 972.99 Ton F 0.054942 1 0.054942 0.010807 0.9179
35 14 30 7 50 0.16 9.054 470 924.15 tV 6.154302 1 6.154302 1.210514 0.2803
36 9 20 9 30 0.07 4.047 460 744.3 tF 10.32487 1 10.32487 2.03084 0.1648
37 9 40 9 30 0.2 7.9 690 1,088.6 VF 5.663496 1 5.663496 1.113975 0.2999
38 14 30 7 50 0.16 6.244 600 952.122 I2 5.486305 1 5.486305 1.079123 0.3075
39 9 40 9 70 0.2 14.193 570 1,021.79 Ton2 22.79328 1 22.79328 4.4833 0.0429
40 19 40 9 70 0.2 7.65 800 1,281.51 t2 2.112668 1 2.112668 0.415549 0.5242
41 9 20 5 70 0.07 12.437 590 738.86 2
V 17.35475 1 17.35475 3.413575 0.0749
42 14 30 7 50 0.035 4.118 340 1,014.98 F2 0.196916 1 0.196916 0.038732 0.8454
43 3 30 7 50 0.16 8.591 560 873.49 Residual 147.4372 29 5.08404
44 9 20 5 70 0.2 4.8 480 758.55 Lack of fit 105.734 22 4.806 0.80674 0.6759 Not
45 9 20 5 30 0.07 3.236 390 637.52 significant
46 14 50 7 50 0.16 6.22 570 1,112.78 Pure error 41.70224 7 5.957464
47 14 30 7 50 0.16 6.6 540 916.1 Cor total 378.8846 49
 Author's personal copy
72 Int J Adv Manuf Technol (2015) 76:69–82

magnetic field (B) and parallel to electric field (E). So the Table 5 ANOVA for developed response surface from Eq. (4)
electron experiences a force (FB) perpendicular to both mag- Source Sum of df Mean F value p value
netic field (B) and velocity (V) and a force (FE) parallel to squares square Prob > F
electric field. The resultant force (FR) confines the electron
which confines the plasma column. The mean free paths of Model 1,652,959 20 82,647.97 12.64362 <0.0001 Significant

electrons are also reduced in the presence of transverse mag- Current (I) 5,371.797 1 5,371.797 0.821786 0.3721

netic field. This increases the more likelihood of collision Pulse on (Ton) 157,385.1 1 157,385.1 24.07703 < 0.0001

among the particles and neutral atoms which increases density Duty factor (t) 48.47143 1 48.47143 0.007415 0.9320
of electrons. When this avalanche of electrons strikes the Voltage (V) 8,097.503 1 8,097.503 1.23877 0.2748
anode surface it absorbs the electrons and its surface temper- Magnetic 67,835.23 1 67,835.23 10.37755 0.0031
flux F)
ature increases. This causes the tool surface to melt and
I Ton 8,082.501 1 8,082.501 1.236474 0.2753
vaporize. Also, at melted temperature, there will be more inter
It 66.53439 1 66.53439 0.010179 0.9203
molecular collision existing in the melted material. Since the
IV 509.0196 1 509.0196 0.077871 0.7822
tool wire used is conductive in nature, its work function at
IF 15,700.93 1 15,700.93 2.401955 0.1320
melted temperature will be less which may make the melted
Ton t 2,734.294 1 2,734.294 0.418297 0.5229
material atoms to ionize. Due to the presence of transverse
Ton V 235.4089 1 235.4089 0.036013 0.8508
magnetic field the heavier melted and ionized metal ion will
Ton F 14.64546 1 14.64546 0.00224 0.9626
experience Lorentz force where it gets confined and deposited
tV 329.303 1 329.303 0.050377 0.8240
over the substrate. Using the concept of Design of
tF 1,364.665 1 1,364.665 0.208769 0.6511
Experiments (DOE), experiments have been carried out to
VF 8,718.727 1 8,718.727 1.333805 0.2576
I2 97.79283 1 97.79283 0.014961 0.9035
Ton2 100,947.6 1 100,947.6 15.44313 0.0005
Table 4 ANOVA for developed response surface from Eq. (3)
t2 392.8641 1 392.8641 0.060101 0.8081
Source Sum of df Mean F value p value V2 591.4563 1 591.4563 0.090482 0.7657
squares square Prob > F F2 37,984.32 1 37,984.32 5.810905 0.0225
Residual 189,565.2 29 6,536.731
Model 1,188,803 20 59,440.13 2.667748 0.0079 Significant
Lack of fit 108,322.5 22 4,923.749 0.424238 0.9413 Not
Current (I) 7,204.046 1 7,204.046 0.323327 0.5740 significant
Pulse on (Ton) 15,607.02 1 15,607.02 0.700463 0.4095 Pure error 81,242.73 7 11,606.1
Duty factor (t) 7,056.53 1 7,056.53 0.316706 0.5779 Cor total 1,842,525 49
Voltage (V) 12,953.98 1 12,953.98 0.581391 0.4519
Magnetic flux 50,420.32 1 50,420.32 2.262928 0.1433
(F)
I Ton 3,350.714 1 3,350.714 0.150384 0.7010
Table 6 ANOVA for developed response surface from Eq. (5)
It 81.23145 1 81.23145 0.003646 0.9523
IV 36,643.03 1 36,643.03 1.644585 0.2099 Source Sum of df Mean F value p value
IF 1,003.472 1 1,003.472 0.045037 0.8334 squares square Prob > F
Ton t 4,602.898 1 4,602.898 0.206584 0.6528
Model 186.251 8 23.28138 4.955194 < 0.0 Significant
Ton V 4,682.371 1 4,682.371 0.210151 0.6501
Pulse on (Ton) 49.9689 1 49.96893 10.63536 0.0
Ton F 83.83852 1 83.83852 0.003763 0.9515
Duty factor (t) 3.52673 1 3.526733 0.750628 0.3
tV 745.9594 1 745.9594 0.03348 0.8561
Voltage (V) 24.7668 1 24.76687 5.271366 0.0
tF 6,526.998 1 6,526.998 0.29294 0.5925
Magnetic 8.36388 1 8.36388 1.780163 0.1
VF 54,592.14 1 54,592.14 2.450164 0.1284 flux (F)
2
I 5,895.71 1 5,895.71 0.264607 0.6109 Ton t 26.4066 1 26.40665 5.620375 0.0
Ton2 2,183.813 1 2,183.813 0.098012 0.7565 tF 13.8108 1 13.81087 2.939497 0.0
t2 55,363.9 1 55,363.9 2.484802 0.1258 Ton2 23.7728 1 23.7728 5.059788 0.0
V2 11,638.75 1 11,638.75 0.522362 0.4756 V2 18.8185 1 18.81854 4.005326 0.0
F2 255,705.6 1 255,705.6 11.47639 0.0020 Residual 192.633 41 4.698379
Residual 646,149.4 29 22,281.01 Lack of fit 150.931 34 4.439155 0.745142 0.7386 Not
Lack of fit 592,149.4 22 26,915.88 3.489096 0.0477 Significant significant
Pure error 54,000 7 7,714.286 Pure error 41.7022 7 5.957464
Cor total 1,834,952 49 Cor total 378.884 49
 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82 73

Table 7 ANOVA for developed response surface from Eq. (6)

Source Sum of squares df Mean square F value p value Prob > F

Model 1,087,612 7 155,373.2 8.731871 < 0.0 Significant

Pulse on (Ton) 127,196.7 1 127,196.7 7.148368 0.0
Duty factor (t) 410.7153 1 410.7153 0.023082 0.8
Voltage (V) 12,155.24 1 12,155.24 0.683117 0.4
Magnetic flux (F) 51,485.69 1 51,485.69 2.893462 0.0
VF 53,286.92 1 53,286.92 2.99469 0.0
t2 63,648.41 1 63,648.41 3.576999 0.0
2
F 264,225 1 264,225 14.84927 0.0
Residual 747,339.7 42 17,793.8
Lack of fit 693,339.7 35 19,809.71 2.567925 0.0978 Not significant
Pure error 54,000 7 7,714.286
Cor total 1,834,952 49

study the influence of each parameter like current, pulse on steel and so it is selected as a substrate. Setup was made to
time, duty cycle and so on. By using the experimental output, hold the permanent magnets of varying capacity, near to the
the statistical models have been developed to evaluate the plasma channel such that magnetic flux cuts the plasma chan-
height, weight and width of deposited material on the sub- nel in such a way that magnet are always kept 15 mm apart.
strate in terms of input parameters. The developed statistical Table feed rate is also found to be one of the significant
model has been validated with experimental results. contributing factor in EDD process reported by
Muralidharan et al. [21]. So, table feed of 200 μm/min (X
and Y directions) is kept constant which is controlled by using
2 Experimental setup stepper motor and micro controller circuit using Lab VIEW
program. Bi layer deposition for a length of 2 mm is carried
The used experimental setup is shown in Fig. 2. A conven- out in every experiment. Material deposition depends on
tional die-sinking EDM machine (Electronica) is used to carry machining parameters like current, voltage, duty factor, mag-
out the experiments in reverse polarity. Copper wire of diam- netic flux, and pulse on time, etc. According to DOE concepts,
eter 0.5 mm is taken as tool and steel plate of dimension five independent parameters are chosen such as, current (I) in
(15 mm×15 mm×1.5 mm) is taken as a workpiece. Copper amperes, duty factor (t), pulse on time (Ton) in microseconds,
material is chosen as workpiece because of its good physical, voltage (V) in volt, and magnetic flux (F) in tesla to perform
chemical, thermal, and electrical properties made it suitable experiments. The above-mentioned parameters with their five
for wide applications, particularly in semiconductor levels are shown in Table 1. The output responses are mea-
manufacturing industry. Mostly, machine parts are made of sured in terms of height (H) in micrometers, weight (W) in

Table 8 ANOVA for developed response surface from Eq. (7)

Source Sum of squares df Mean square F value p value Prob > F

Model 1624860 6 270809.947 53.4988613 < 0.0 Significant

Current (I) 5423.243 1 5423.24275 1.07136874 0.3
Pulse on (Ton) 1106556 1 1106556.16 218.601626 0.0
Magnetic flux (F) 67506.11 1 67506.1056 13.33592 0.0
IF 16009.2 1 16009.2035 3.16263922 0.0
Ton2 101317.8 1 101317.799 20.0154645 0.1
F2 37696.17 1 37696.1724 7.44692845 0.0
Residual 217665 43 5061.97591
Lack of fit 136422.2 36 3789.50659 0.3265098 0.9876 Not significant
Pure error 81242.73 7 11606.1038
Cor total 1842525 49
 Author's personal copy
74 Int J Adv Manuf Technol (2015) 76:69–82

micrograms and width (WD) in micrometers. After each ex-

periment’s height, weight and width of deposition are mea-
sured using, TESA RUGOSURF roughness gauge 10G,
Digital weighing machine (least count=10 μg) and Zeiss
Axio Scope microscope, respectively. To increase the confi-
dence limit, three readings are recorded for height, weight, and
width. The average value is noted as every experimental
output response and these responses are used to develop a
statistical model. The microscopic image of deposited line
with width measurement is shown in the Fig. 3. Since the
process is not fully established, all experiments are de-
signed by the concept of design of experiments. In the
present research, Central Composite Rotatable Design
(CCRD) has been used to design the experiments through
Design-Expert statistical software. Box and Hunter [22]
reported that CCRD can be a useful alternate for factorial
design. CCRD requires very few experiments unlike full
factorial design to explore much information about a Fig. 4 Comparison between measured value and predicted value for
process which is reported by Obeng et al. [23]. In height (μm)
CCRD, experiments are designed on the basis of 2 k
Factorial, where k is the number of variables. In the
The height, weight, and width models are obtained by
present design, the number of experiments for full fraction
analyzing the data presented in Table 2 and it is given below
with five factors is 50. The designed experiments are
as Eqs. (2), (3), and (4)
shown in the Table 2.

Height ¼ −7:08 þ 0:10  I þ 0:17  T on þ 0:06  t þ 0:32  V −16:98  F

3 Empirical model and data analysis
þ0:0004  I  T on −0:04  I  t−0:005  I  V þ 0:71  I  F
þ 0:04  T on  t þ 0:001  T on  V −0:06  T on  F−0:01  t  V
Response surface methodology (RSM) is a collection of
statistical and mathematical techniques useful for devel- þ 4:10  t  F−0:31  V  F þ 0:01  I 2 −0:007  T on 2 −0:04  t2

oping, improving, and optimizing processes. The applica- −0:001  V 2 þ 7:58  F 2 ð2Þ
tion of RSM is developing an approximating model for
the response surface, which can be extracted from some
unknown physical mechanism reported by Mayers et al.
[24]. RSM is used to determine the relationship between
different input factors affecting the process and the output
of that process. When the results of these experiments are
analyzed, they help to identify the factors which most
influence the results. The relationship between single de-
pendent variable (response) and number of independent
variables (regressors) is characterized by a mathematical
model called regression model. Montgomery [25] ex-
plained that regression model is used to express the re-
sults of an experiment quantitatively, in terms of empirical
model. In general, response variable Y may be related to k
regressor variables is given by Eq. (1).

k k
Y ¼ β 0 þ ∑ β i X i þ ∑ βii X 2i þ∑∑ βij X i X j
i ¼1 i ¼1 i< j
Y ¼ corresponding response; e:g:weight; height etc:
X i Xj ¼ levels of k quantitative process variables:
Fig. 5 Comparison between measured value and predicted value for
β 0 βi βii β ij ¼ regression coefficients : ð1Þ
weight (μg)
 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82 75

Width ¼ −251:64−8:24  I þ 46:47  T on þ 21:90  t þ 3:03  V

þ923:89  F þ 0:31  I  T on þ 0:13  I  t þ 0:04  I  V
þ67:48  I  F−0:44  T on  t−0:013  T on  V þ 1:03
 T on  F−0:07  t  V þ 47:24  t  F−12:54  V
 F þ 0:06  I 2 −0:51  T on 2 −0:56  t2 −0:007
 V 2 −3; 333:01  F 2 ð4Þ

Analysis of variance (ANOVA) was used to check the

adequacy of the developed model. The p values of the predic-
tive model were calculated. Values of p value Prob>F which
is less than 0.0500 indicates that model terms are significant
and p value greater than 0.1000 indicates the model terms are
not significant reported in Mayers [24]. The ANOVA for the
second-order model as shown in Eq. (2) has been presented in
Table 3. It is observed that the p value Prob > F is 0.0213
Fig. 6 Comparison between measured value and predicted value for indicates that the developed model is significant. The ANOVA
width (μm) for the second-order model shown in Eq. (3) is presented in
Table 4. It is observed from Table 4 that p value Prob > F is
Weight ¼ 407:04 þ 44:45  I þ 11:54  T on −119:46  t
0.0079, which indicates that the model is significant. Similarly,
ANOVA for the second-order model shown in Eq. (4) is pre-
þ1:52  V −3509:89  F−0:20  I  T on −0:15  I  t−0:34
sented in Table 5, in which p value Prob > F is<0.0001, which
 I  V −17:05  I  F þ 0:57  T on  t−0:06  T on  V
indicates that the model significant. There are many insignificant
þ 2:46  T on  F−0:11  t  V þ 103:32  t  F þ 31:38 model terms found in Tables 3, 4, and 5.
 V  F−0:47  I 2 −0:075  T on 2 þ 6:72  t 2 þ 0:035  V 2 Since there are many insignificant model terms present in
þ 8; 647:80  F 2 ð3Þ the model, the model has to be improved using backward

Fig. 7 a Variation of pulse on

time vs. height (I=14, t=7, v=50,
F=0.16). b Variation of duty
factor vs. height (I=14, Ton =30,
v=50, F=0.16). c Variation of
voltage vs. height (I=14, Ton =30,
t=7, F=0.16). d Variation of
magnetic flux vs. height (I=14,
Ton =30, t=7, v=50)
 Author's personal copy
76 Int J Adv Manuf Technol (2015) 76:69–82

Fig. 8 a Variation of pulse on

time vs. weight (I=14, t=7, v=50,
F=0.16). b Variation of duty
factor vs. weight (I=14, Ton =30,
v=50, F=0.16). c Variation of
voltage vs. weight (I=14, Ton =30,
t=7, F=0.16). d Variation of
magnetic flux vs. weight
(I=14, Ton =30, t=7, v=50)

Fig. 9 a Variation of pulse on

time vs. width (I=14, t=7, v=50,
F=0.16). b Variation of magnetic
flux vs. width (I=14, Ton =30,
t=7, v=50). c Variation of current
vs. width (Ton =30, t=7, v=50,
F=0.16)
 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82 77

Fig. 10 Effect of pulse on time on height (I=14, t=7, v=50) Fig. 12 Effect of voltage on height (I=14, Ton =30, t=7)

regression elimination method, which removes the terms, factors such as, Pulse on time, voltage and magnetic flux are
having an insignificant effect on output responses. The back- found to be significant factors for deposition height. The
ward elimination is a stepwise method begins by developing a ANOVA for the improved model described by Eqs. (6)
full regression model using all independent variables. Then a t and (7) has been presented in Tables 7 and 8. From
test or an F test for significance is performed on each regres- Table 7, it is observed that, p value Prob > F is<0.0001
sion coefficient at a specified alpha level. The backward and from Table 8 p value Prob > F is <0.0001 which
elimination continues until all independent variables remain- indicates that both the modified model for weight and
ing in the model have t values above a level specified by the width are significant. Pulse on time, and magnetic flux
decision maker. The advantage of backward elimination is that are found to be significant factors which contribute for
the decision maker has the opportunity to look at all the weight, which is observed from Table 7. Pulse on time
independent variables in the model before removing the var- and magnetic flux are found to be significant factors for
iables that are not significant. The new model for height, width of deposition, which is observed from Table 8.
weight and width of deposition, are obtained after applying Lack of fit is not significant for all the developed models
backward elimination model reduction method. The ANOVA which are desired. So, the developed model can represent
for improved model from Eq. (5) has been presented in appreciable responses, when the corresponding input pa-
Table 6. It is observed that the p value Prob > F is <0.0001 rameters are used in the model.
showing that the model is significant. Also, from Table 6, the

Fig. 11 Effect of pulse on time on height (I=14, t=7, F=0.16) Fig. 13 Effect of voltage on height (I=14, t=7, F=0.16)
 Author's personal copy
78 Int J Adv Manuf Technol (2015) 76:69–82

Fig. 14 Effect of magnetic flux on height (I=14, t=7, v=50) Fig. 16 Effect of pulse on time on weight (I=14, t=7, F=0.16)

Height ¼ 3:12 þ 0:27  T on −1:63  t þ 0:17  V −26:15 range of error percentage for height is from 20.43 to
−22.36 %, for weight is from 23.69 to −24.84 % and for width
 F þ 0:04  T on  t þ 4:62  t  F−0:007
is from 14.82 to −18.31 %, respectively. These variations in
 T on 2 −0:001  V 2 ð5Þ error percentage are due to occurrence of error. From Figs. 7a,
8a, and 9a it is observed that there is an increase in percentage
of error with increase in pulse on time. At higher pulse on time
Weight ¼ 798:51 þ 5:58  T on −93:71  t−2:26 the vaporization of metal may be more which leads to devia-
 V −2; 935:92  F þ 30:96  V  F þ 6:58 tion. There is an increase in variation of error is observed with
increase in duty factor from the Figs. 7b and 8b. This increase
 t 2 þ 8; 667:87  F 2 ð6Þ
in error may due to arc instability at higher duty factor. From
the Figs. 7c and 8c there is an increase in error deviation with
Width ¼ −169:89 þ 5:99  I þ 46:97  T on þ 642:39 increase in voltage. At higher voltage inter electrode gap
increases which are related to arc sensitivity of the quill.
 F þ 67:67  I  F−0:50 Therefore, error due to quill sensitivity to maintain spark with
 T on 2 −3; 280:44  F 2 ð7Þ variation in inter electrode gap increases the error deviation.
The increase in magnetic flux increases error deviation which
is observed from Figs. 7d, 8d, and 9b. This increase in error
The above-developed mathematical model can be used to due to variation in crater size formation on tool wire is due to
predict the values of height, weight, and width of deposition. high magnetic force, which affects the deposition layer. From
The differences between the measured and predicted re- Fig. 9c with an increase in current there is an increase in
sponses are shown in the Figs. 4, 5, and 6. The calculated percentage of error. At higher current there may be more

Fig. 15 Effect of magnetic flux on height (I=14, Ton =30, t=7) Fig. 17 Effect of pulse on time on weight (I=14, v=50, F=0.16)
 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82 79

Fig. 18 Effect of pulse on time on weight (I=14, t=7, v=50) Fig. 20 Effect of magnetic flux on weight (I=14, t=7, v=50)

vaporization as said above leads to this error. Also, the average effect of voltage on height of deposition for various pulse on
predicted error for height, weight, and width are 2.21, −0.06, time. It is observed that as voltage increases, the height of
and −12 %. Since the average predicted error percentages deposition increases. Also, at higher voltage, because of high
were less, the developed empirical model has better fit. potential difference, the inter electrode gap (IEG) increases
and ultimately plasma column starts diverge in between tool
and substrate. By using external magnetic field, plasma diver-
4 Results and discussions gence can be controlled. So the presence of magnetic field in
the experimental setup checks the number of charged particles
The effect of pulse on time on height for various flux values not to escape from plasma column due to vapourization. This
and effect of pulse on time on height for various voltage gives higher deposition rate with increase in voltage. The
values are shown in Figs. 10 and 11. It is observed that there effect of magnetic flux on deposited height is shown in
is increase in height until 50 μs and after that it starts decreas- Fig. 14 for various pulse on time and Fig. 15 for different
ing. Pulse on time determines the duration of current to be voltage values. The deposited height increases as magnetic
discharged and current is a function of energy generated. flux is increased. Plasma channel will be confined in the
Therefore, beyond 50 μs because of simultaneous increase presence of magnetic field as said above. So at higher field,
in pulse off time, the available heat energy to melt the tool plasma will be confined more, which leads to higher deposi-
material to a significant amount may be dissipated to the tion rate and height will be increased significantly.
surrounding which reduces tool wear rate which leads to lesser It is observed from Figs. 16, 17 and 18 that deposited
deposition over the substrate. material weight increases as the pulse on time increases for
Figure 12 shows the effect of voltage on height of deposi- different voltage, duty factor, and magnetic flux values. Pulse
tion for various ranges of magnetic flux and Fig. 13 shows the on time determines the discharge current duration which is

Fig. 19 Effect of magnetic flux on weight (I=14, Ton =30, v=50) Fig. 21 Effect of pulse on time on width (t=7, v=50, F=0.16)
 Author's personal copy
80 Int J Adv Manuf Technol (2015) 76:69–82

Fig. 22 Effect of pulse on time

on width (I=14, t=7, v=50)

function of current. Therefore, there is a possibility that more discharge channel will be confined in the presence of magnet-
material is melted from the tool at higher pulse on time and ic flux. So, it is observed from Fig. 23 which shows the effect
thereby increasing the deposited weight. Figures 19 and 20 of magnetic flux on deposited width, where the deposited
show that with an increase in magnetic flux, weight of depo- width decreases with an increase in magnetic field.
sition increases for different values of duty factor and pulse on EDAX analysis has been carried out to examine the chem-
time. This is because magnetic flux confines the discharge ical composition of deposited layer. It is observed from Fig. 24
channel which leads to more deposition of particles by hold- that the chemical composition of deposited layer by weight
ing them not to escape from the channel. The effect of pulse on percentage as, copper (Cu)—53.42 %, carbon (C)—41.00 %,
time on deposited width is shown in Figs. 21 and 22 for oxygen (O)—3.54 %, and ferrous (Fe)—2.04 %. The presence
various values of current and magnetic flux. The deposited of copper is directly from the tool wire by melting and depo-
width increases as pulse on time is increased. For a fixed time sition. The carbon and oxygen percentage deposited is from
higher pulse on retains more heat energy in IEG which in- the atmosphere, since the experiment is carried in dry mode
creases tool wear with higher vaporization. This process may
tend to deposit the ionized particles with increased width. The

Fig. 23 Effect of magnetic flux on width (Ton =30, t=7, v=50) Fig. 24 EDAX image showing compositional analysis
 Author's personal copy
Int J Adv Manuf Technol (2015) 76:69–82 81

Fig. 27 AFM image of deposited surface without magnetic field (I=19,

Ton =20, t=5, V=30)

deposited material in the presence of magnetic field. From

Fig. 26, it is found that there is no random deposition of small
particles as shown in Fig. 25 and has better uniformity in
deposited layer when compared to effect of deposited layer
Fig. 25 SEM image for the condition (I=9, Ton =20, t=9, V=70) without
magnetic field without a magnetic field. 3D images are taken using atomic
force microscope (AFM) to understand the uniformity of the
with air as dielectric medium. The percentage of carbon and deposited layer. Figure 27 shows the AFM image of deposited
oxygen may be significantly reduced by employing inert layer without magnetic field, where the maximum peak value
atmosphere at the sparking zone. This may increase the depo- is about 6 μm. Figure 28 represents the AFM image of
sition of copper by weight percentage and helps to produce deposited layer with magnetic field and the maximum peak
microparts. The traces of ferrous are from the base material value is observed to be about 4 μm. So, with the help of
where stainless steel is used as workpiece. From the EDAX magnetic field roughness may be reduced and uniformity
analysis, it is understood that EDD process can be employed may be achieved. The uniformity may be improved by reduc-
to build the product since the deposited material main chem- ing the Larmor radius of the metal ion which can be done by
ical composition is same as tool material. increasing the strength of magnetic field in the plasma
The SEM micrograph of a deposited material without column.
magnetic field is shown in Fig. 25. In the absence of magnetic
field, it is observed that the copper material is deposited in the
form of small particles over the substrate randomly without
uniformity. Figure 26 represents an SEM micrograph of 5 Conclusions

A systematic investigation of magnetic field assisted electro-

discharge deposition process is reported in this paper for
various conditions. Based on the current study, the following
conclusions can be drawn.

1. A multiple regression model has been developed and

validated with experimental results.

Fig. 26 SEM image for the condition (I=9, Ton =20, t=9, V=70, F= Fig. 28 AFM image of deposited surface with magnetic field (I=19,
0.07) in the presence of magnetic field Ton =20, t=5, V=30, F=0.20)
 Author's personal copy
82 Int J Adv Manuf Technol (2015) 76:69–82

2. The major finding in the present research is that, in the 9. Mohri N, Takezawa H, Furutani K, Ito Y, Sata T (2000) New process
of additive and removal machining by EDM with a thin electrode.
presence of magnetic field, height, and weight of the
Ann CIRP 49:123–126
deposition can be increased and the width can be signif- 10. Mohri N, Suzuki M, Furuya M, Saito N (1995) Electrode wear
icantly reduced. process in electric discharge machining. Ann CIRP 44:165–168
3. In the presence of magnetic field random deposition of 11. Chi G, Wang Z, Xiao K, Cui J, Jin B (2008) The fabrication of a
micro-spiral structure using EDM deposition in the air. J Micromech
particles, which leads to nonuniformity in deposited layer
Microeng 18:1–9
is reduced and the deposited layer is formed with better 12. Wang YK, Xie BC, Wang ZL, Peng ZL (2011) Micro EDM deposi-
uniformity. tion in air by single discharge thermo simulation. Trans Nonferrous
Metals Soc China 21:450–455
13. Jain VK, Shashank, Ajay Sidpara, Himanshu Jain (2012) Some
So, it is concluded that magnetic field is one of the impor-
aspects of micro- fabrication using electro-discharge deposition pro-
tant contributing factors in the EDD process, and in the cess. Proceedings of International Symposium on Flexible
presence of magnetic field the process performance is signif- Automation ISFA2012, ISFA2012-7108:419–424
icantly improved. 14. Kenneth G Heinz JR (2010) Fundamental study of magnetic field
assisted micro –EDM for non magnetic materials. Dissertation,
University of Illinois at Urbana-Champaign, Urbana, Illinois, US
Acknowledgments The authors acknowledge the contribution of Dr. 15. Nomuraa K, Oginob Y, Hirataa Y (2012) Influence of magnet con-
Mukesh Kumar Roy, PDPM-IIITDMJ for valuable suggestion and con- figurations on magnetic controlled TIG arc welding. Trans JWRI
stant support to carry out EDAX analysis and AFM experiments. 39(2):209–210
16. Nomuraa K, Oginob Y, Hirataa Y (2012) Shape control of TIGArc
plasma by cusp-type magnetic field with permanent magnets. Weld
Int 26(10):759–764
References 17. Tayler RJ (1957) The influence of an axial magnetic field on the
stability of a constricted gas discharge. Proc Phys Soc 70(11):1049–
1063
1. Alting L, Kimura F, Hansen HN, Bissacco G (2003) Micro engineer- 18. Keidar M, Beilis I, Boxman RL, Goldsmith S (1996) 2D expansion
ing. Ann CIRP 52(2):635–658 of the low-density interelectrode vacuum arc plasma jet in an axial
2. Cui Z (2005) Micro-nano fabrication. Higher Education Press, magnetic field. J Phys D Appl Phys 29(7):1973–1983
Springer, China 19. Beilis II, Keidar M, Boxman RL, Goldsmith S (1998) Theoretical
3. Brousseau EB, Dimov SS, Pham DT (2010) Some recent advances in study of plasma expansion in a magnetic field in a disk anode vacuum
multi-material micro and nanomanufacturing. Int J Adv Manuf arc. J Appl Phys 83(2):709–717
Technol 47(1–4):161–180 20. Rondanini M, Cavallotti C, Ricci D, Chrastina D, Isella G, Moiseev
4. Dimov SS, Matthews CW, Glanfield A, Dorrington PA (2006) T, VonKanel H (2008) An experimental and theoretical investigation
Roadmapping study in multi-material micro manufacture. of a magnetically confined dc plasma discharge. J Appl Phys 104:1–
Proceedings of the Second International Conference on Multi- 13
Material Micro Manufacture 4M2006, Grenoble,France, 20–22 21. Muralidharan B, Chelladurai H, Ramkumar J (2013) Experimental
September, xi–xxv investigation on electro discharge deposition (EDD) process.
5. Rajurkar KP, Levy G, Malshe A, Sundaram MM, McGeough International Conference-ASME 2013 International Mechanical
J, Hu X, Resnick R, DeSilva A (2006) Micro and nano Engineering Congress & Exposition, San Diego
machining by electro-physical and chemical processes. Ann 22. Box GEP, Hunter JS (1957) Multi-factor experimental design for
CIRP 55(2):643–666 exploring response surfaces. Ann Math Stat 28(1):195–241
6. Gangadhar A, Shanmugam MS, Philip PK (1991) Surface modifica- 23. Obeng DP, Morrell S, Napier TJN (2005) Application of central
tion in electro-discharge processing with powder compact tool elec- composite rotatable design to modeling the effect of some operating
trode. Wear 143:45–55 variables on the performance of the three-product cyclone. Int J
7. Simao J, Lee HG, Aspinwall DK, Dewes RC, Aspinwall EM (2003) Miner Process 769:181–192
Workpiece surface modification using electric discharge machining. 24. Mayers RH, Montgomery DC, Anderson-Cook CM (2009)
Int J Mach Tools Manuf 43:121–128 Response surface methodology process and product optimization
8. Goto A, Magara T, Imai Y, Miyake H, Saito N, Mohri N (1997) using designed experiments. Wiley, New York
Formation of hard layer on metallic material by EDM. J Jpn Soc 25. Montgomery DC (2009) Design and analysis of experiments. Wiley,
Electr Mach Eng 31(68):26–31 New York

View publication stats