Journal of Materials Processing Technology 84 (1998) 90 – 96

Influence of EDM pulse energy on the surface integrity of

martensitic steels
J.C. Rebelo a, A. Morao Dias a,*, D. Kremer b, J.L. Lebrun b
a
Faculdade de Ciências e Tecnologia da Uni6ersidade de Coimbra, Largo D. Dinis, P-3000 Coimbra, Portugal
b
LPTPM and LM3, 151 Boul6ard de l’Hôpital, URA CNRS 1219, ENSAM, Paris, France

Received 4 August 1997

Abstract

Electro-discharge machining (EDM) is essentially a thermal process with a complex metal-removal mechanism, involving the
formation of a plasma channel between the tool and workpiece electrodes, resulting in metallurgical transformations, residual
tensile stresses and cracking. These properties determine the operational behaviour of the material and can be included in one
term: surface integrity. Results of different experimental analysis to characterise the surface integrity of steels used in the
production of moulds after EDM are presented. The roughness of the surface, the metallurgical structure, the residual stress state
and the surface crack network of the near-surface layers, in the electro-discharge steels studied, as well as their dependence on the
main processing parameters, are examined and discussed quantitatively and qualitatively. © 1998 Elsevier Science S.A. All rights
reserved.

Keywords: Electro-discharge machining; Martensitic steels; Surface integrity

1. Introduction 1. Ionisation and arc formation, at a localised area

between the electrodes, following the application of
Electrical-discharge machining (EDM) removes elec- a voltage exceeding the breakdown voltage;
trically conductive material by means of rapid, repeti- 2. The occurrence of the main discharge as an electron
tive spark discharges from a pulsating direct-current avalanche striking the anode, with low electrical
power supply with dielectric flow between the work- resistance in the discharge channel. The cathode is
piece and the tool. The shaped tool (electrode) is fed struck by ions and is heated less rapidly than the
into the workpiece under servo control. A spark dis- anode;
charge then breaks down the dielectric fluid. The servo 3. Local melting and evaporation follow, and some
control maintains a constant gap between the tool and material is removed from the site of the discharge by
the workpiece whilst advancing the electrode. The explosion occurring after the cessation of electrical
dielectric oil cools and flushes out the vaporised and
discharge. The current density decreases with in-
condensed material whilst re-establishing insulation in
creasing discharge duration, the discharge tending
the gap [1].
to become an arc.
Several theories were proposed by early investigators
to account for the erosion mechanism of the EDM The structural changes of electro-discharge machined
process [2]. It has been accepted generally that the (EDMed) surfaces have been studied extensively [2].
metal removal phenomenon is predominantly thermal For tool steels, it has been shown that the top-most
in nature [2]. The best explanation of EDM phenomena surface layer is an uneven, non-etchable layer, namely
has been established by extensive experimental studies the ‘white layer’. This is a re-cast layer formed by the
[3,4]. Three stages can be distinguished: molten metal solidifying at an extremely high rate after
the discharge process. Immediately beneath the ‘white
* Corresponding author. Tel.: + 351 39 7900759-22; fax: + 351 39 layer’ there is an intermediate layer, a heat-affected
7900701; e-mail: GTR@[Link] zone, where the heat is not high enough to cause

0924-0136/98/$ - see front matter © 1998 Elsevier Science S.A. All rights reserved.
PII S0924-0136(98)00082-X
 J.C. Rebelo et al. / Journal of Materials Processing Technology 84 (1998) 90–96 91

melting but is sufficiently high to induce micro-struc- and the austenite phases varies [8]. The increase in
tural transformation in the material [2,4 – 6]. The re-cast carbon content in the surface and sub-surface layers as
layer is found to be heavily alloyed with the pyrolysis a result of EDM has been attributed by most workers
products of the cracked dielectric. When the tool elec- to the pyrolysis of the dielectric, but others have sug-
trode, copper, for example, is compatible, alloying at gested that carbon is assimilated more rapidly from
the surface can be found also. With suitable reagents, it graphite electrodes than from carbonaceous dielectric
has been shown that, depending on the steel and on the [8].
machining conditions, a variety of micro-structures can EDM generates residual stresses, these being due
result [7]. Some authors [8] have described the spark-af- mainly to the non-homogeneity of heat flow and metal-
fected layer as being different if they are machined by lurgical transformations. Investigation of the residual
graphite or copper electrodes; dendritic austenite and a stresses of EDMed components revealed their tensile
cementite–austenite eutectic (ledeburite), or a fully nature, the narrowness of the superficial zone where
austenitic surface followed by an austenite – cementite they appear, their high magnitude at the surface layers
matrix, respectively. Others, however, have reported a and the increase of their magnitude with increasing
similar structure of carbides in an austenite matrix, but pulse energy [5,9,10].
stated that different electrodes do not change the mor- In this study the role of the EDM parameters on the
phology of the white layer: only the ratio of the carbide surface roughness, the metallurgical structure, the resid-

Table 1
Chemical composition and initial state of the steels tested

Steel Chemical composition (wt.%) Heat treatment Microhardness (HV) Notation

C Si Mn Cr Mo V

A 0.6 0.35 0.8 4.5 0.5 0.2 Annealed 234 914 A1

Quenched and tempered 700 915 A2
B 0.38 0.9 0.5 13.6 0.3 Annealed 240 911 B1
Quenched and tempered 610 910 B2

Fig. 1. (a) Showing the top surface of A2 EDMed at: (a) 32A and 800 ms; and (b) 2A and 6.4 ms (SEM pictures).

Fig. 2. (a) Showing cross-sections of A2 EDMed at: (a) 16A and 200 ms; and (b) 32A and 800 ms (optical pictures).
92 J.C. Rebelo et al. / Journal of Materials Processing Technology 84 (1998) 90–96

Fig. 5. Variation of ‘white layer’ thickness (wlt) and ‘heat-affected

layer’ thickness (halt) with Int [A]*time (s) for A2 EDMed steel
Fig. 3. Variation of surface roughness with Int (A)time (s) for plates.
EDMed steel plates.

graphic microscope. Etching was performed by immers-

ual stress state and the surface crack network, in
ing the specimen in 5% Nital reagent.
EDMed martensitic steels, are examined and discussed
(ii) Scanning electron microscopy: sample 5×5 mm2
qualitatively and quantitatively.
strips were cut from electro-discharge machined speci-
mens and their surfaces examined under an electron
microscope.
2. Experimental procedure (iii) Surface roughness measurements: surface rough-
ness was obtained from a ‘Perthometer’ recorder, the
A ROBOFORM 200-‘Charmilles Technology’ indus- cut-off being selected as 17.5 mm, the roughness values
trial machine was used for the electro-discharge ma- averaging at least five measurements per specimen.
chining. The experiments were conducted in (iv) Micro-probe analysis: this technique consists of
hydrocarbon dielectric with lateral flushing, with elec- spectrometer analysis of the X-ray characteristics of a
trolytic copper being used as the tool electrode (the sample after it has been subjected to an electron beam.
anode) of a rectangular working area of 30× 30 mm2. (v) The X-ray diffraction method [11] using a diffrac-
The generator produced rectangular pulses at currents tometer with V geometry was used for the following: (a)
of I = 2, 4, 8, 16, 32 A and at durations t =6.4, 50, 100, X-ray phase analysis: possible phase changes and for-
200, 800 ms. Rectangular steel plates of 15 mm thick- mation of new phases after processing. (b) A diffrac-
ness were used as the test specimens. Quenched and togram between 20° and 160° in 2u was made for each
tempered samples A and B (see Table 1) were cut by specimen, at the EDMed surface, using a Cu X-ray
wire EDM (WEDM) whilst the annealed samples were tube. The quantification of the in-depth austenite–
cut mechanically. The WEDM machine was a martensite phases was made by software developed at
ROBOFIL310-‘Charmilles Technology’. the University of Coimbra, Portugal. (c) Residual stress
After machining, the following experimental tech- profiles: measurements of in-depth residual stresses at
niques were employed for assessing the surface integrity EDMed surfaces were made. A Cr X-ray tube was
of the worked plates: used. The residual stresses were determined by the
(i) Optical microscopy: specimens were sectioned software STRESS AT. The in-depth profiles were
transversely and prepared under standard procedure for achieved by electrolytic polishing.
metallographic observation on a NIKON metallo-

3. Results and discussion

3.1. Surface topography

During each electrical discharge, intense heat is gen-

erated, causing local melting or even evaporation of the
workpiece material. With each discharge, a crater is
formed on the workpiece and a smaller crater is formed
on the tool electrode. Of the molten material produced
by the discharge, only 15% or less is carried away by
the dielectric [7]. The remaining melt re-solidifies to
Fig. 4. Semi-quantitative in-depth profiles of carbon content. form an undulating terrain. After magnification, the
 J.C. Rebelo et al. / Journal of Materials Processing Technology 84 (1998) 90–96 93

Fig. 6. X-ray determination of the surface phase changes in the EDM of A2 steel.

Fig. 7. In-depth diffraction peak width evolution of A2 and B steel plates EDMed for different regimes.

Fig. 8. Scheme of the interpretation of the EDM-affected zones, for a rough (above) and fine (below) regime, through diffraction peak broadening:
(a) quenched and tempered steels; (b) annealed steels.
94 J.C. Rebelo et al. / Journal of Materials Processing Technology 84 (1998) 90–96

Fig. 9. Profiles of percentage residual austenite in the heat-affected Fig. 11. Average maximum crack lengths (amcl), depth of maximum
zone of A steel samples for different regimes. value of stress (dmvs), of A2 samples for different regimes of machin-
ing.
surface is observed to be covered with overlapping
craters, globules of debris, and pockmarks or ‘chim- As can be seen from Fig. 3, the roughness of an
neys’, formed by entrapped gases escaping from the EDMed surface increases with the increase in pulse
re-deposited material. energy and can be represented by an empirical equation
Another prominent feature on EDM surfaces is the of the form Ra = k[It]a where the constants k and a
abundance of microcracks. These cracks are seen to depend only on the chemical composition of the steel.
radiate from, and also to circumvent, the craters. The These data corroborate the results of other authors [12],
crack formation is associated with the development of who have shown that constants k and a depend on the
high thermal stresses exceeding the ultimate tensile tool–work material combination.
strength of the material, as well as with plastic defor-
mation. The crater sizes increase with pulse energy, as 3.3. Carbon content and phase changes
does the density of surface cracks. Furthermore, the
cracks penetrate into the re-cast layer and extend to A semi-quantitative analysis of the carbon content in
depths that depend on the pulse energy. the surface layers was made by micro-probe analysis.
As an example, the top surfaces and cross-sections of The variation is plotted in Fig. 4 for different machin-
EDMed A2 steel are shown in Figs. 1 and 2. ing regimes for both steels, A and B. It can be seen that
the carbon intensity can be nine times greater at the
3.2. Effect of pulse energy on surface roughness and surface than in the bulk material. The carbon content
white layer thickness gradient increases with the pulse energy of the machin-
ing regime.
The values of the surface roughness were measured The ‘white layer’ thicknesses observed were an aver-
and plotted against intensity (A)time of discharge (s) age of about 30 readings, measured at various places on
for the materials studied. The logarithmic trend of the different cross-sections of the specimen concerned.
curves suggests an empirical equation of the form: Etching of the specimens with Nital reagent left this
Ra = k[It]a. The values of the constants k and a were area characteristically white. The data obtained for
determined by fitting the curves by the least-squares different materials but for the same regime are similar.
method. The high values of the x 2 obtained (0.996– The average of the ‘white layer’ thickness is plotted
0.998) indicate a good fit. against Int (A)time (s) in Fig. 5 for A2 specimens.

Fig. 10. Residual stress distribution below the surface of A2 and B steel samples.
 J.C. Rebelo et al. / Journal of Materials Processing Technology 84 (1998) 90–96 95

As can be seen, a regression analysis of the data The residual stress profiles show the high tensile
yields wlt=148[It]0.34, with a x 2 of 0.996. The regres- nature of the stresses. The residual stress increases from
sions obtained for the remaining materials are similar. the bulk material to a maximum and then decreases
This empirical relationship agrees with the empirical again near to the surface. This decrease is related to
relationship determined by Lee et al. [2], and holds for crack formation since the residual stresses exceed the
a wide range of tool steels. Furthermore Lee et al. [2] fracture strength of the material. The depth of the
concluded that neither the pulse current nor the pulse- maximum stress value corresponds to the average depth
on duration has a significant effect on the thickness of of the thermal cracks network induced by EDM. The
the white layer. The magnitude of the pulse current development of the cracks network near to the surface
depends on the area or size of the pulse form alone, and of the specimens is related to a stress-relieving process.
not on its shape. In this way, discharge energy mainly affects the depth
X-ray diffraction was used to identify the new phases at which the maximum value of residual stress occurs:
appearing during the process. After machining, a new the greater the energy, the greater the depth. Thus, the
phase Fe3C, cementite, is formed, predominating in the surface residual stress level decreases with increasing
surface layers. The same behaviour was observed for all pulse energy. Empirical relationships between average
of the studied steels. The increase of carbon content at maximum crack lengths (amcl) and Int (A)time (s),
the surface as a result of EDM is due to the pyrolysis of and depth of maximum stress value (dmsv) and Int
the dielectric. (A)time (s), have been determined for A2 samples
Two diffractograms, taken at the surface, for A2 steel (Fig. 11). The remaining results have similar empirical
before and after machining are shown as an example in relationships.
Fig. 6. Under the ‘white layer’, a heat-affected layer is
observed for both steels A and B, through the breadth
of the {211} martensite diffraction peak. The results are
plotted in Fig. 7, for different regimes of machining. An
example of the influence of machining pulse energy on 4. Conclusions
the EDM heat-affected layer thickness (halt), for A2
steel samples, with Int (A)time (s) is plotted in Fig. 5. Summarising the main features of the present experi-
As far as the diffraction peak breadth is concerned, mental work, the following conclusions may be drawn:
different heat-affected zones can be distinguished for 1. A stochastic process is revealed using optical and
quenched and annealed steels. Fig. 8 shows a scheme for scanning electron microscopy. The dimensions of ran-
interpreting the EDM heat-affected layers through the dom overlapping surface craters increases with machin-
broadening of the diffraction peak. ing pulse energy. The cracks radiate from, and
Three zones were identified in rough regimes of ma- circumvent, the craters. The density and penetration
chining for the two initial states of steels: a quenched depth of the cracks in the re-cast layer increases with
layer, a tempered layer and bulk material were observed the machining pulse energy.
for quenched and tempered steels; and a quenched layer, 2. A new phase, cementite, formed at the ‘white
a treated layer and bulk material for annealed steels. In layer’, was determined by X-ray diffraction and confi-
polished regimes of machining, two zones were distin- rmed by microprobe analysis.
guished: a tempered layer and bulk material for 3. Different heat-affected zones were observed, de-
quenched and tempered steels; and a treated layer and pending on the machining energy and the initial state of
bulk material for annealed steels. the steels, and modelled using X-ray diffraction peak
Inside the EDM heat-affected zones, the in-depth broadening.
evaluations of residual austenite were quantified, cor- 4. X-ray diffraction enables the accurate determina-
roborating the EDM quenched layer. Fig. 9 shows some tion of high residual stresses of tensile nature.
of these experimental data for EDMed A steel samples. 5. Network crack formation is associated with the
development of high tensile stresses, exceeding the frac-
3.4. Residual stresses ture strength of the material.
6. The possibility of quantifying the extended surface
The cementite on the surface layer of the samples was damage and relating it to a parameter that can be more
removed by electrolytic polishing before the residual readily measured or controlled, such as Int (A)time (s),
stress determination on the martensite or ferritic phase has been thoroughly examined. Thus, empirical rela-
of the steels. As soon as the {211} diffraction peak of tionships between surface roughness, ‘white layer’
Fe was free from overlapping by other peaks, the thickness, heat-affected layer thickness, average maxi-
residual stresses were evaluated. Examples of residual mum crack length, depth of maximum value of residual
stresses in in-depth profiles are presented in Fig. 10 for stress, and Int (A)time (s) were determined for both
steels A and B. materials.
96 J.C. Rebelo et al. / Journal of Materials Processing Technology 84 (1998) 90–96

Acknowledgements Working Technol. 15 (1987) 335 – 356.

J.C. Rebelo gratefully acknowledges the support [4] J.P. Devimeux, et al., Guide des Usinages par Électro-Érosion et
par Électrochimie, CETIM et l’ENIMS, France, 1980.
given to this work by the PRAXIS XXI Program [5] Report of AGIE: Control of surface quality, including thermal
(Portugal). and stress-induced cracking, in sinking EDM, American Machin-
ist and Automated Manufacturing, Nov. 1987, vol. 9, no. 6, pp.
80 – 83.
[6] R. Reda Mostafa, EDM Digest 5 (5) (Sept./Oct. 1983) 8–9.
References [7] L.C. Lim, L.C. Lee, Y.S. Wong, H.H. Lu, Mater. Sci. Technol.
7 (1991) 239 – 248.
[1] R. Williams, Machining Hard Materials, Machining Data Hand- [8] P.F. Thomson, Mater. Sci. Technol. 5 (1989) 1153 – 1156.
book, 3rd ed., vol. 2, Society of Manufacturing Engineers, [9] A.G. Mamalis, G.C. Vosniakos, N.M. Vaxevanidis, Ann. CIRP
Dearborn, MI, 1980. 37 (1) (1988) 531 – 535.
[2] L.C. Lee, L.C. Lim, V. Narayanan, V.C. Venkatesh, Int. J. [10] B. Hosari, Ph.D. Thesis, ENSAM, Paris, July 1988.
Mach. Tools Manufact. 28 (4) (1988) 359–372. [11] G. Maeder, Chem. Scripta, 26A (1986) 23 – 31.
[3] A.G. Mamalis, G.C. Vosniakos, N.M. Vaxevanidis, J. Mech. [12] M.L. Jewsvani, Wear 51 (1978) 227 – 236.