- Email: [email protected]
Study of the recast layer of a surface machined by sinking electrical discharge machining using water-in-oil emulsion as dielectric
Applied Surface Science 257 (2011) 5989–5997
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Study of the recast layer of a surface machined by sinking electrical discharge machining using water-in-oil emulsion as dielectric Yanzhen Zhang, Yonghong Liu ∗ , Renjie Ji, Baoping Cai College of Mechanical and Electronic Engineering, China University of Petroleum, Dongying, Shandong 250100, China
a r t i c l e
i n f o
Article history: Received 15 November 2010 Received in revised form 19 January 2011 Accepted 19 January 2011 Available online 26 January 2011 Keywords: Electrical discharge machining (EDM) Recast layer Water-in-oil Emulsion
a b s t r a c t Electrical discharge machining (EDM) caused a recast layer to form at the machined surface of the workpiece. The characteristics of the recast layer have a great relationship with the type of dielectric. The research work in this paper aims to acquire a profound knowledge of the recast layers of a surface machined by sinking EDM using water-in-oil (W/O) emulsion as dielectric. Scanning electron microscopy (SEM), X-ray diffraction (XRD), energy dispersive spectrograph (EDS) and micro hardness analysis were performed. The characteristics of the recast layer formed in W/O emulsion were investigated by comparing them with those of the recast layer formed in kerosene and de-ionized water dielectric. It was found that the recast layer formed in W/O emulsion exhibited larger surface roughness, thickness and micro hardness compared with that formed in kerosene and de-ionized water. Both carbide and oxide were detected in the recast layer formed in W/O emulsion whereas only carbide was detected in the recast layer formed in kerosene. Due to the higher supersaturation of gases in the melted material, the recast layer formed in W/O emulsion was found to possess more micro-voids than that formed in kerosene and de-ionized water. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Electrical discharge machining (EDM) is a thermoelectric process that erodes workpiece material by a series of discrete electrical sparks between the workpiece and electrode flushed by or immersed in a dielectric fluid. After EDM, a recast layer will be formed on the machined surface regardless of the type of dielectric. It is well known that the main mode of erosion is caused by the thermal action of an electrical discharge. The non-traditional manufacturing process of sinking-EDM possesses many advantages over traditional machining during the manufacture of the hard-tocut materials. However, certain detrimental effects are also present and are due in large part to the formation of the recast layer. The action of EDM has actually altered the metallurgical structure and characteristics of the recast layer. Micro-cracks can form in this very hard, brittle layer due to an increase in nonhomogeneities of metallurgical phases within the recast layer [1]. The recast layer is the result of the re-solidification of the melted material which did not sweep away from the component’s surface by
∗ Corresponding author at: College of Electromechanical Engineering, China University of Petroleum, Dongying, Shandong 257061, China. Tel.: +86 0546 8392303; fax: +86 0546 8393620. E-mail addresses: [email protected], [email protected], [email protected] (Y. Liu). 0169-4332/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2011.01.083
the dielectric during the EDM process and is known to exhibit high hardness, good adherence to the bulk and good resistance to corrosion. However, the recast layer formed by EDM process increases surface roughness, makes the surface become hard and brittle, and decreases the fatigue strength due to the presence of micro-cracks and micro-voids [2]. The composition of dielectric has an important influence on the characteristics of the recast layer since the discharge gap is partially filled by the dielectric during the EDM process which is also a chemical process [3]. The intrinsic nature of EDM process results in a material removal of both workpiece and tool electrode. Formation of the plasma channel consisting of material vapors from the eroding work material and tool electrode, and pyrolysis of the dielectric affect the surface composition after machining and consequently, its characteristics [4]. Deliberate material transfer may be carried out under specific machining conditions by either using composite electrodes [5–11] or dispersing metallic powders in the dielectric [12–14] or both. In previous work, we studied the sinking-EDM performance using water-in-oil (W/O) emulsion as dielectric. The material remove rate (MRR) obtained in W/O emulsion is much higher than that obtained in kerosene, especially with rough machining parameters. In the sinking-EDM application, the traditional dielectrics are hydrogen–carbon oils. Normally water is not used as a dielectric for sinking-EDM. Although the use of plain water
5990
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
in sinking-EDM results in a better performance in some situations [15,16], hydrocarbon oils are superior in a wider range of machining conditions. Guu et al. [17] investigated the effect of electrical discharge machining on surface characteristics of AISI D2 tool steel using kerosene as a dielectric. Experimental results indicated that the thickness of the recast layer, and surface roughness were proportional to the power input. Ekmekci [1] investigated the composition and crack formation of the recast layer formed in kerosene and water dielectrics. The results revealed that characteristics of the crack are mainly affected by the properties of base material and white layer composition which is partially determined by the type of dielectrics. Previous results presented by Chen et al. [18], Yan et al. [19] and Ekmekci et al. [20] also supported this conclusion. Additionally, operational parameters, such as average discharge current and pulse-on duration were also found to play an important role in the formation of cracks [1,21]. Cusanelli et al. [22] studied the microstructure of the recast layer produced by EDM technique using kerosene as a dielectric at submicron scale. They found that the recast layer was composed of sublayers with particular structures and phases which depending on the machining energy. Cabanillas et al. [23] studied the formation of carbide by electro-discharge machining of alpha iron using kerosene as dielectric. They found that the type of carbide formed on the machined surface was determined by discharge energy. Kruth et al. [24] studied the property of recast layer formed in standard oil dielectric (BP180) by comparing it with that formed in deionized water. He found that the use of an oil dielectric increased the carbon content in the recast layer, whereas a water dielectric caused a decarbonisation. In our case, W/O emulsion was used as the dielectric, the simultaneous presence of water and oil in the discharge gap may contribute to a recast layer that is different from that formed only in water or oil-based dielectric. In this paper, the composition and crystallographical and metallurgical properties of the recast layer formed in W/O emulsion were studied by comparing them with those formed in kerosene and de-ionized water dielectrics.
2. Experimental procedures The W/O emulsion used in this experiment was prepared using, for the oil phase, 66 vol% of machine oil, and for the water phase, 34 vol% of de-ionized water. In order to stabilize the emulsion, 2.5 wt% of Span80 was added to the oil phase. Emulsification was carried out with a homogenizer (FJ200, with 18 and 12.7 mm of stator and rotor diameters, respectively) at 3000 rpm for 20 min. The microstructure of W/O emulsion after homogenization is shown in Fig. 1. The samples were machined by a spark-erosion sinking machine (NH250). The experimental set-up is illustrated in Fig. 2. Since the viscosity of W/O emulsion is much larger than that of kerosene or water, the emulsion was compelled into the discharge gap through a hole (Ø4.2 mm) in the centre of the electrode by a micro diaphragm pump. Subject of the tests was the W/O emulsion. For comparison, the tests were also done with kerosene and de-ionized water as dielectrics. During the EDM process, the pulse interval (48 s) was constant. Technologies going from roughing to half finishing have been used in order to produce a large range of the recast layer thicknesses. Samples were machined in a parametrical order using three different pulse-on durations and two different peak current settings (Table 1). A cylindrical copper rod of 30 mm diameter was used as the tool electrode. The material of the workpiece was ordinary carbon steel (C35) and the size of the workpiece was 75 mm × 35 mm × 4 mm. The machining depth was 1 mm. The copper electrode served as the positive polarity when machining was performed in W/O emulsion and kerosene, and negative polar-
Fig. 1. Optical micrograph of the W/O emulsion used in this experiment.
Fig. 2. Schematic representation of the EDM process (1, pulsed generator; 2, servocontrol; 3, electrode; 4, specimen; 5, dielectric fluid; 6, micro pump).
ity when machining was performed in de-ionized water during the EDM process. After machining, the samples were cut using a wire-EDM machine to obtain a transversal cross-section. In order to observe the recast layer using metallographic microscope and scanning electron microscopy (SEM), the samples were mirror-polished down to 1 m after having been embedded in epoxy, and then etched with nital. An energy dispersive spectrograph (EDS) associated with the SEM was also used to investigate the elements present in the recast layer after EDM. In order to identify the phases present in the recast layer, X-ray diffraction (XRD) tests have been performed with an OMNI diffractometer. The micro hardness was measured using a micro indentation device with a low load (100 mN) and a micrometric pyramidal imprint. The surface roughness (SR) was determined with a surface profilometer. The recast layer thickness (RLT) was measured with a metallurgical microscope (Nikon EclipseME600P) under a magnification of 200×. The RLT was measured at 40 locations across each cross-sectioned specimen, and an average value was calculated. The distance between the measured locations is uniform and is 50 m.
Table 1 Experimental conditions of the samples. Working parameters
Description
Peak current (A) Pulse duration (s) Polarity Dielectric
9 76 Positive (+) W/O emulsion
15 150 Negative (−) Kerosene
308 De-ionized water
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
5991
Fig. 3. Comparing the surface roughness obtained in different dielectrics and peak currents. Pulse duration = 308 s.
3. Results and discussion 3.1. SR and RLT Fig. 3 compares SR and Fig. 4 compares the surface textures of the samples obtained in different dielectrics. Surfaces with different morphologies could be obtained using different dielectrics. The difference can be easily detected with lower magnification. The samples obtained in W/O emulsion, exhibit much coarser surface with respect to those obtained in kerosene and de-ionized water. The craters formed in W/O emulsion are much larger and deeper than those formed in kerosene and de-ionized water. A comparison of the surface textures reveals that the diameter and the depth of the craters are significantly affected by the type of dielectrics. The craters formed in W/O emulsion exhibit the largest depth whereas the craters formed in de-ionized water exhibit the largest diameter and the smallest depth. These results are consistent with the reports of Konig and Jorres [16] who point out that de-ionized water usually results in lower values of surface roughness. These results may be attributed to the high viscosity of W/O emulsion. Because dielectric with high viscosity can restrict the expansion of the discharge channel, the impulsive force is concentrated within a very small area; therefore results in larger and deeper craters. Another possible reason for the big craters is the oxidation reaction of the melted material which can intensify the discharge and contribute to larger craters. Fig. 4 also shows that more micro-voids existed on the surface of the sample machined in W/O emulsion than that machined in the other dielectrics. The reason will be discussed later. Fig. 5 shows the thickness of the recast layer obtained in different dielectrics and pulse duration. As has been stated above, the recast layer is formed by molten metal which is not flushed away by the dielectric, but re-solidifies on the sample’s machined surface during cooling. It can be seen that RLT is significantly influenced by the type of dielectrics. The average thickness of the recast layer formed in W/O emulsion is larger than that formed in kerosene and de-ionized water with the same pulse duration. This is explained by the fact that the amount of molten metal which can be flushed away by different dielectrics is different. Therefore, as W/O emulsion has a much higher viscosity than kerosene, it is unable to clear away the molten material, and so it builds up upon the surface of the sample. During subsequent cooling, this molten material resolidifies to form the recast layer, the depth of which depends upon the volume of molten material which was left on the sample surface during machining. Fig. 5 also shows that the RLT and its deviation increase with increasing pulse duration. This is due to the greater material removal rate when the pulse duration is larger. It will be
Fig. 4. SEM photographs of the machined surface of (a) sample W9; (b) sample K9. Peak current = 9 A. Pulse duration = 308 s.
seen that when the EDM surface undulates dramatically, there will be significant variation in the recast layer thickness. Fig. 6 shows the RLT obtained with various peak currents. The results have revealed that the thickness of the recast layer increases with increasing peak current regardless of the dielectric type. It should be noted that, although RLT increases as the peak current increases, the increase is not very significant. This phenomenon is consistent with Lee and Tai’s research [21]. Standard deviations for measured white layer thicknesses indicate that increasing peak current increases RLT deviation considerably. These results are consistent with the observations presented by Ekmekci [1].
5992
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
Fig. 5. Comparing the RLT of samples obtained in different dielectrics and pulse duration. Peak current = 9 A.
3.2. Oxide existed in the recast layer With higher magnification, large amount of oxide can be found on the surface of samples when machining is performed in W/O emulsion and de-ionized water, as shown in Fig. 7a and b. This can be attributed to the decomposition of the water droplets dispersed in the discharge gap due to the discharge energy. As illustrated in Fig. 1, the average diameter of the droplets is about 10 m which is comparable with the distance of the discharge gap. When using W/O emulsion as dielectric, the working area was partially occupied by the water droplets which can be easily vaporized by the heat generated by the discharge and subsequently reacted with the melted material. The melted metal is oxidized by oxygen decomposed from water vapors by the discharge energy. More oxides exist on the surface of samples when machining is performed with larger peak current. This indicates that the discharge energy is an important factor that affects the formation of oxides. Since larger discharge energy can decompose more water into H and O, larger amount of oxide will be generated when larger peak current was adopted. Furthermore, the higher discharge energy can result in a higher temperature of the discharge gap, which can accelerate the oxidation reaction rate. The oxides shown in Fig. 7a and b were rarely observed on the surface of samples machined in kerosene. Although they were observed on the surface of samples machined in kerosene with larger peak current, the area occupied by the oxides is much smaller compared with samples machined in W/O emulsion and de-ionized
Fig. 7. SEM photographs show the typical surface textures of samples machined in: (a) W/O emulsion and de-ionized water; (b) W/O emulsion and de-ionized water; (c) Kerosene.
Fig. 6. Comparing the RLT of samples obtained in different dielectrics and peak current. Pulse duration = 308 s.
water dielectric. With high magnification, the typical surface texture of samples machined in kerosene is shown in Fig. 7c. These results strengthen the opinion that the formation of the oxide is due to the presence of water in the discharge gap. The oxidation can intensify the discharge and contribute to a larger crater; therefore the samples machined in W/O emulsion have a larger material removal rate (MRR) and surface roughness (SR) compared with that machined in kerosene. This experimental finding is consistent with Kunieda et al., who concluded that oxygen could be fed into the gap between workpiece and electrode to increase MRR
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
5993
Table 2 The element contents in the white layer of sample K9 and W9. Peak current = 9 A. Pulse duration = 308 s. Element
W/O emulsion mass%
Kerosene mass%
De-ionized water mass%
C O Fe Cu Total
4.1 2.1 93.8 0 100
5.1 2.7 92.2 0 100
0.7 1.7 90.0 7.6 100
obtained in W/O emulsion and de-ionized water. However, when machining is performed in W/O emulsion, the presence of iron oxide was not detected. This may be attributed to its low contents in the recast layer. All the phases present in the recast layer formed in kerosene have been reported by Chen et al. [18] and Cabanillas et al. [23]. Cabanillas et al. [23] studied the formation of carbide by EDM of alpha iron using kerosene as dielectric. In Cabanillas’ reports, the trace of some other carbide, such as Fe7C3 and Fe5C2 was detected. However, in our case, the presence of Fe7C3 or Fe5C2 was not detected. This may be attributed to their low contents in the recast layer. 3.4. EDS measurement
Fig. 8. X-ray diffraction pattern of samples machined in various dielectric and pulse duration. Peak current = 9A.
[25–27]. However, when machining in de-ionized water, the MRR is much lower than machining in kerosene or W/O emulsion. This has been attributed to the lower viscosity of water, which produces less restriction of the discharge channel, thus reducing the energy density and, as consequence, decreasing the material removal rate [16]. Moreover, the large amount of energy required to heat and vaporize water compared with oil results in lower gas pressure in the gap. Consequently the molten metal is not removed properly in every discharge, because of the insufficient pressure produced by the burst of water [28]. 3.3. X-ray diffraction (XRD) measurement The examination of the XRD patterns allowed identifying the major phases present in the samples. The diffraction patterns were measured for 10.3366 s per step from 10◦ to 100◦ with a step of 0.017◦ and analyzed using the X’Pert HighScore Plus program. The phases have been identified from searches in the PDF-2 databases. X-ray diffraction patterns for samples have shown two different trends (Fig. 8). When samples are machined in kerosene and W/O emulsion, iron carbides and ␥-Fe peaks have been observed, and the compositions of the recast layer do not show much difference. On the other hand, when samples are machined in de-ionized water, iron oxides and ␣-Fe have been observed. When machining is performed in de-ionized water, the intensity of iron oxides increases with increasing pulse duration whereas the intensity of ␣-Fe decreases with increasing pulse duration. The intensity of ␣-Fe exhibited the same trend when machining is performed in both kerosene and W/O emulsion. This indicates that the discharge energy has a significant influence on the ␣ → ␥ transformation. Iron oxide was created through chemical reaction between the workpiece material and the oxygen element decomposed from water caused by the discharge energy. The result is consistent with the SEM detection of the iron oxide on the machined surface
EDS analysis was used to measure the element composition of the recast layer after EDM. Fig. 9 compares the EDS patterns of samples machined in different dielectrics. The elements present in the recast layer are clearly indicated by the peaks corresponding to their energy levels. The relative percentages of different materials found in the surface are given in Table 2. It was observed from the analysis that the percentage of carbon element in the recast layer formed in kerosene is larger than that in the recast layer formed in W/O emulsion and de-ionized water dielectrics. It should be noted that the accuracy of carbon content detected by EDS analysis is limited since it is not sensitive to the light elements. However, comparing the EDS results of the samples, we can make a qualitative conclusion that the carbon content of the recast layer formed in kerosene is higher than that formed in W/O emulsion and deionized water. Since carbon is generated due to the decomposition of the hydrocarbon oil, the lower carbon content in the recast layer formed in W/O emulsion and de-ionized water can be attributed to the presence of water in the discharge gap with respect to kerosene case. The results were consistent with Kruth et al. [24], who found that there was a decarbonisation of the machined surface by sinking EDM when using water as dielectric. It should be noted that, when machining was performed in de-ionized water, the presence of Cu was detected in the recast layer. This is due to the migration of material between electrodes and this is confirmed to be affected by the machining parameters, such as polarity, peak currents and pulse duration [29,30]. 3.5. Micro-cracks and micro-voids in the recast layer Crack formation can be attributed to the presence of thermal stress and tensile stress within the machined surface. The primary causes of the thermal stress in the machined surface were the drastic heating and cooling rates and the non-uniform temperature distribution [31]. Tensile stress within the sample is generated because not all of the material which melts during the machining process is swept away from the component’s surface by the dielectric. Due to the ingress of carbon, the melted material contracts more than the unaffected parent part during the cooling process, and when the stress in the surface exceeds the material’s ultimate tensile strength, cracks are formed [19,31–33].
5994
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
Fig. 9. EDS analysis of the samples machined in various dielectric. Pulse duration = 308 s; Peak current = 9A.
Ekmekci [1] classified the cracks formed in EDM process into three types. The first type of cracks, called surface cracks, exists in the white layer, initiating at its surface and traveling down perpendicularly toward the interferential zone, and usually terminates at this interference. Such cracks have high potentials to partial flaking off, which also proved the weakness of the intermediate layer between the recast and heat affected layers. The second type of cracks, called penetrating cracks, penetrates the entire white layer thickness to an extent into the parent material. In our experiments, the second type cracks that penetrated into the parent material were not found within all the machining conditions, although they have been reported by other researchers [21,34]. The third type of cracks is visualized usually around globular or irregularly shaped attachments on crater rims. Such cracks have very low penetration depth and are randomly distributed on the surface. Thus, have little influence on the surface properties. In our case, the cracks that existed on the recast layer formed in water dielectric are mainly the third type cracks which have very low penetration depth and are randomly distributed on the surface especially where the curvature is larger, such as the rims of the craters (Fig. 10). So, we infer that the larger density of cracks may be explained by the higher cooling rate when machining is performed in water, since a faster cooling rate has the ability to introduce a larger residual stresses, thus increasing the tendency for cracks to form. Because of their low penetration depth and width, it is hard to detect them in the cross section (Fig. 11). Observation of the external and sectional micrograph of the samples shows that the cracks formed in W/O emulsion and kerosene dielectric exhibit similar characteristic, and they can mainly be classified to the first type. The results presented in our experiments confirm the conclusion of Ekmekci that the first type of cracks are usually observed when machining is performed in a hydrocarbon-based dielectric liquid with high pulse-on duration and low discharge current [1]. In order to give a quantitative research of the first type of cracks, the cracks are quantified by the total length of cracks (mm) in a unit area (mm2 ). Fig. 12 presents an analysis of this measure for the samples, and shows how the sur-
face crack density varies with dielectric type. Observation of these results shows that the presence of surface cracking is much more evident when machining is performed in W/O emulsion, and the density of crack is greater than the kerosene and de-ionized water case. The results also show that the surface crack density decreases significantly with the increase of peak currents regardless of the dielectric type. Previous results presented by Lee et al. [21,35] also supported this conclusion. It should be noted that the cracks often originate from the border of micro-voids regardless of the dielectric type. The micro-cracks were associated with the development of thermal stresses exceeding the ultimate tensile strength of the material. The presence of micro-voids will reduce the local service strength of the recast layer and increase the cracking possibility. On the other hand, the presence of micro-voids leads to a drastic change of the surface curvature which is an important factor that affects the cooling rate of material. Fig. 11 compares the sectional micrograph of the samples obtained in W/O emulsion, kerosene and de-ionized water. There are many micro-voids scattered in the recast layer formed in W/O emulsion. The micro-voids can also be seen on the external surface of the sample (Figs. 4 and 10). In the micrograph of both sectional and external surface, the occurrence of such micro-voids was rarely observed in the recast layer formed in kerosene and de-ionized water. This is explained by the presence of water droplets in the discharge gap. When machining is performed in W/O emulsion, the gas volume generated in the discharge gap was larger than in the case of kerosene due to the decomposition and vaporization of water, since water has a much lower boiling point with respect to kerosene. When machining is performed in de-ionized water, just as the case of kerosene, the volume of generated gas is much smaller than the case of W/O emulsion. The results indicated that the small water droplets that scattered in the discharge gap can be vaporized more easily than the continuous water that filled the discharge gap due to its bigger specific surface area. Although some gases were generated when machining is performed in kerosene [36,37] and de-ionized water, their volume is much smaller than
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
5995
Fig. 11. Cross-sectional micrographs of samples machined in various dielectrics. Pulse duration = 308 s; Peak current = 9A.
Fig. 12. Comparing of the crack density of the recast layer formed in different dielectrics and peak currents. Pulse duration = 308 s.
Fig. 10. SEM photographs show micro-cracks on the surface of samples machined in various dielectrics. Pulse duration = 308 s; Peak current = 9A.
the W/O emulsion case. In order to provide quantitative results of the gases volume, the gases generated during the EDM process were collected by a gas gather devise. The volumes of gases generated in 1 min in different dielectrics are listed in Table 3. When
machining is performed in W/O emulsion, the volume of generated gas is much larger than that generated in the case of kerosene and de-ionized water. The micro-voids can be attributed to the gas bubbles expelled from the molten material during solidification [31]. The presence of large volume of gases in the gap, especially water vapor and hydrogen, will lead to a high supersaturation of gas in the molten pool, which is the prerequisite for the formation of micro-voids.
Table 3 The volume of gases generated in 1 min in various dielectrics and peak current. Dielectric Peak current (A) Volume of generated gas (ml)
W/O emulsion 9 114 ± 13
Kerosene 15 293 ± 20
9 16 ± 3
De-ionized water 15 34 ± 5
9 20 ± 5
15 42 ± 7
5996
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
Fig. 13. Comparing of the micro hardness of the recast layer formed in different dielectrics and peak currents. Pulse duration = 308 s.
3.6. Micro hardness In order to characterise the mechanical properties of the recast layer, a micro hardness test has been performed using a micro hardness tester with a low load (100 mN) and a micrometric pyramidal imprint. Before the measurement, the samples were mirror-polished down to 1 m after having been embedded in epoxy, then etched with nital for a very short time to make the recast layer visible. The micro hardness of the recast layer was measured at 30 locations in the middle zone of the recast layer, and an average value was calculated. Fig. 13 compares the micro hardness of the recast layers formed in different dielectrics. The micro hardness of the recast layer obtained in different dielectrics can be ranked as (in decreasing order) W/O emulsion–kerosene–water; so does the standard deviation of the micro hardness. This indicates that the recast layer formed in W/O emulsion has a more inhomogeneous structure than that formed in kerosene and de-ionized water. The micro hardness of recast layer machined in de-ionized water is lower than that of the recast layer machined in kerosene and W/O emulsion. This can be explained by its lower carbon contents. Previous results presented by Kruth et al. [24] also supported the same conclusion. Just as shown by the above analysis, the carbon content in the recast layer formed in W/O emulsion is lower than that formed in kerosene. But the micro hardness of the recast layer formed in W/O emulsion is higher than that formed in kerosene. It seems that besides the ingress of carbon, the high cooling rate is another possible reason for the hardness increase of the recast layer when machining is performed in W/O emulsion; since the high cooling rate can make the recast layer freeze into a martensitic structure which is also known to give a high hardness. Fig. 14 shows the martensitic structure in the recast layer when machining is performed in W/O emulsion and kerosene dielectric. Although it is observed in the recast layer when kerosene was used as dielectric, it is more common when machining is performed in W/O emulsion. The martensitic structure shown in Fig. 14 was not detected in the recast layer formed in water dielectric due to the lower carbon content. In order to study the influence of cooling rate, the micro hardness of the recast layer of the samples was also measured at different distances from the interface. As has been reported by Ekmekci [1] and Cusanelli et al. [22], there are two solidification fronts during the cooling of the melted material. Heat convection takes place at the outer surface which is in contact with the dielectric, whilst at the interface, convection and conduction both contribute to extract the heat. Within the recast layer, three points were chosen for measurement; A, point close to the external surface; B, point in the middle zone; C, point close to the inter-
Fig. 14. Comparing of the micro hardness depth profile of sample W9 and K9. Pulse duration = 308 s; Peak current = 9A.
face. Moreover, the micro hardness of HAZ and matrix was also measured. The outcome of the measurements reveals that the minimum value of the micro hardness usually appeared in the middle zone of the recast layer, whereas the maximum value appeared in the zone close to external surface or the interface. This indicates that the cooling rates of the melted material do have a remarkable influence on micro hardness of the recast layer.
4. Conclusion (1) The recast layer formed in W/O emulsion dielectric exhibits a greater surface roughness and thickness than that formed in kerosene and de-ionized water dielectrics. (2) Both carbide and oxide are formed in the recast layer when machining is performed in W/O emulsion dielectric, whereas only carbide is formed in the case of kerosene and only oxide is formed in the case of de-ionized water. The discharge energy has a positive effect on the formation of oxide. (3) The micro-cracks of the recast layer formed in water dielectric have very low penetration depth and are randomly distributed over the surface. However, cracks that formed in W/O emulsion and kerosene dielectrics exist in the recast layer and have much larger depth and usually terminate at the recast and heat affect layers interface. The density of crack decreases as the peak current increases regardless of the dielectric type. (4) Because of the high supersaturation of gas in the molten pool, there are many micro-voids within the recast layer when machining is performed in W/O emulsion. The micro-voids are rarely observed within the recast layer when machining is formed in kerosene and de-ionized water dielectrics. (5) Due to the ingress of carbon and high cooling rate, the recast layer formed in W/O exhibits a greater micro hardness than that formed in kerosene and de-ionized water dielectrics.
Acknowledgements The work is partially supported by a grant from Chinese National Natural Science Foundation (Grant No. 50675225) and a grant from Department of Science & Technology of Shandong Province (Grant No. 2006GG2204001).
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
References [1] B. Ekmekci, Metall. Mater. Trans. B: Process Metall. Mater. Process. Sci. 40 (2009) 70–81. [2] Y.S. Liao, J.T. Huang, Y.H. Chen, J. Mater. Process. Technol. 149 (2004) 165–171. [3] M. Kunieda, B. Lauwers, K.P. Rajurkar, B.M. Schumacher, CIRP Ann. Manuf. Technol. 54 (2005) 599–622. [4] S. Kumar, R. Singh, T.P. Singh, B.L. Sethi, J. Mater. Process. Technol. 209 (2009) 3675–3687. [5] D.I. Pantelis, N.M. Vaxevanidis, A.E. Houndri, P. Dumas, M. Jeandin, Surf. Eng. 14 (1998) 55–61. [6] T. Moro, N. Mohri, H. Otsobo, A. Goto, N. Saito, J. Mater. Process. Technol. 149 (2004) 65–70. [7] W.S. Zhao, Y. Fang, Z.L. Wang, L.H. Li, Mater. Sci. Forum 471–472 (2004) 750–754. [8] P.K. Patowari, U.K. Mishra, P. Saha, P.K. Mishra, Int. J. Manuf. Technol. Manage. 21 (2010) 83–89. [9] Z.L. Wang, Y. Fang, P.N. Wu, W.S. Zhao, K. Cheng, J. Mater. Process. Technol. 129 (2002) 139–142. [10] J. Simao, H.G. Lee, D.K. Aspinwall, R.C. Dewes, E.M. Aspinwall, Int. J. Mach. Tools Manuf. 43 (2003) 121–128. [11] P.K. Patowari, P. Saha, P.K. Mishra, Int. J. Adv. Manuf. Technol. (2010) 1–12. [12] Y.F. Chen, Y.C. Lin, J. Mater. Process. Technol. 209 (2009) 4343–4350. [13] B.H. Yan, Y.C. Lin, F.Y. Huang, C.H. Wang, Mater. Trans. 42 (2001) 2597– 2604. [14] K. Furutania, A. Saneto, H. Takezawa, N. Mohri, H. Miyake, Precis. Eng. 25 (2001) 138–144. [15] T. Masuzawa, K. Tanaka, Y. Nakamura, N. Kinoshita, CIRP Ann. Manuf. Technol. 32 (1983) 119–122. [16] W. Konig, L. Jorres, CIRP Ann. Manuf. Technol. 36 (1987) 105–109.
[17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28]
[29] [30] [31] [32] [33] [34] [35] [36] [37]
5997
Y.H. Guu, H. Hocheng, C.Y. Chou, C.S. Deng, Mater. Sci. Eng. A 358 (2003) 37–43. S.L. Chen, B.H. Yan, F.Y. Huang, J. Mater. Process. Technol. 87 (1999) 107–111. B.H. Yan, H.C. Tsai, F.Y. Huang, J. Mach. Tools Manuf. 45 (2005) 194–200. B. Ekmekci, O. Elkoca, A. Erden, Metall. Mater. Trans. B: Process Metall. Mater. Process. Sci. 36 (2005) 117–124. H.T. Lee, T.Y. Tai, J. Mater. Process. Technol. 142 (2003) 676–683. G. Cusanelli, W.A. Hessler, F. Bobard, R. Demellayer, R. Perez, R. Flukiger, J. Mater. Process. Technol. 149 (2004) 289–295. E.D. Cabanillas, J. Desimoni, G. Punte, R.C. Mercader, Mater. Sci. Eng. A 276 (2000) 133–140. J.P. Kruth, L. Stevens, L. Froyen, B. Lauwers, CIRP Ann. Manuf. Technol. 44 (1995) 169–172. M. Kunieda, S. Furuoya, N. Taniguchi, CIRP Ann. Manuf. Technol. 40 (1991) 215–218. M. Kunieda, M. Yoshida, N. Taniguchi, CIRP Ann. Manuf. Technol. 46 (1997) 143–146. M. Kunieda, Y. Miyoshi, T. Takaya, N. Nakajima, Y.Z. Bo, M. Yoshida, CIRP Ann. Manuf. Technol. 52 (2003) 147–150. T. Masuzawa, Machining characteristics of EDM using water as dielectric fluid, in: Proceedings of the 22nd Machine Tool Design and Research Conference, Manchester, 1981, pp. 441–447. J.S. Soni, G. Chakraverti, J. Mater. Process. Technol. 56 (1996) 439–451. S. Zinelis, Dent. Mater. 23 (2007) 607–1607. Y.H. Guu, M.T. Hou, Mater. Sci. Eng. A 466 (2007) 61–67. B. Ekmekci, A.E. Tekkaya, A. Erden, J. Mach. Tools Manuf. 46 (2006) 858–868. B. Ekmekci, Appl. Surf. Sci. 253 (2007) 9234–9240. C.P. Tabrett, J. Mater. Sci. Lett. 15 (1996) 1792–1794. H.T. Lee, F.C. Hsu, T.Y. Tai, Mater. Sci. Eng. A 364 (2004) 346–356. G.N. Levy, CIRP Ann. Manuf. Technol. 42 (1993) 227–230. F.N. Leão, I.R. Pashby, J. Mater. Process. Technol. 149 (2004) 341–346.
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
Study of the recast layer of a surface machined by sinking electrical discharge machining using water-in-oil emulsion as dielectric Yanzhen Zhang, Yonghong Liu ∗ , Renjie Ji, Baoping Cai College of Mechanical and Electronic Engineering, China University of Petroleum, Dongying, Shandong 250100, China
a r t i c l e
i n f o
Article history: Received 15 November 2010 Received in revised form 19 January 2011 Accepted 19 January 2011 Available online 26 January 2011 Keywords: Electrical discharge machining (EDM) Recast layer Water-in-oil Emulsion
a b s t r a c t Electrical discharge machining (EDM) caused a recast layer to form at the machined surface of the workpiece. The characteristics of the recast layer have a great relationship with the type of dielectric. The research work in this paper aims to acquire a profound knowledge of the recast layers of a surface machined by sinking EDM using water-in-oil (W/O) emulsion as dielectric. Scanning electron microscopy (SEM), X-ray diffraction (XRD), energy dispersive spectrograph (EDS) and micro hardness analysis were performed. The characteristics of the recast layer formed in W/O emulsion were investigated by comparing them with those of the recast layer formed in kerosene and de-ionized water dielectric. It was found that the recast layer formed in W/O emulsion exhibited larger surface roughness, thickness and micro hardness compared with that formed in kerosene and de-ionized water. Both carbide and oxide were detected in the recast layer formed in W/O emulsion whereas only carbide was detected in the recast layer formed in kerosene. Due to the higher supersaturation of gases in the melted material, the recast layer formed in W/O emulsion was found to possess more micro-voids than that formed in kerosene and de-ionized water. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Electrical discharge machining (EDM) is a thermoelectric process that erodes workpiece material by a series of discrete electrical sparks between the workpiece and electrode flushed by or immersed in a dielectric fluid. After EDM, a recast layer will be formed on the machined surface regardless of the type of dielectric. It is well known that the main mode of erosion is caused by the thermal action of an electrical discharge. The non-traditional manufacturing process of sinking-EDM possesses many advantages over traditional machining during the manufacture of the hard-tocut materials. However, certain detrimental effects are also present and are due in large part to the formation of the recast layer. The action of EDM has actually altered the metallurgical structure and characteristics of the recast layer. Micro-cracks can form in this very hard, brittle layer due to an increase in nonhomogeneities of metallurgical phases within the recast layer [1]. The recast layer is the result of the re-solidification of the melted material which did not sweep away from the component’s surface by
∗ Corresponding author at: College of Electromechanical Engineering, China University of Petroleum, Dongying, Shandong 257061, China. Tel.: +86 0546 8392303; fax: +86 0546 8393620. E-mail addresses: [email protected], [email protected], [email protected] (Y. Liu). 0169-4332/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2011.01.083
the dielectric during the EDM process and is known to exhibit high hardness, good adherence to the bulk and good resistance to corrosion. However, the recast layer formed by EDM process increases surface roughness, makes the surface become hard and brittle, and decreases the fatigue strength due to the presence of micro-cracks and micro-voids [2]. The composition of dielectric has an important influence on the characteristics of the recast layer since the discharge gap is partially filled by the dielectric during the EDM process which is also a chemical process [3]. The intrinsic nature of EDM process results in a material removal of both workpiece and tool electrode. Formation of the plasma channel consisting of material vapors from the eroding work material and tool electrode, and pyrolysis of the dielectric affect the surface composition after machining and consequently, its characteristics [4]. Deliberate material transfer may be carried out under specific machining conditions by either using composite electrodes [5–11] or dispersing metallic powders in the dielectric [12–14] or both. In previous work, we studied the sinking-EDM performance using water-in-oil (W/O) emulsion as dielectric. The material remove rate (MRR) obtained in W/O emulsion is much higher than that obtained in kerosene, especially with rough machining parameters. In the sinking-EDM application, the traditional dielectrics are hydrogen–carbon oils. Normally water is not used as a dielectric for sinking-EDM. Although the use of plain water
5990
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
in sinking-EDM results in a better performance in some situations [15,16], hydrocarbon oils are superior in a wider range of machining conditions. Guu et al. [17] investigated the effect of electrical discharge machining on surface characteristics of AISI D2 tool steel using kerosene as a dielectric. Experimental results indicated that the thickness of the recast layer, and surface roughness were proportional to the power input. Ekmekci [1] investigated the composition and crack formation of the recast layer formed in kerosene and water dielectrics. The results revealed that characteristics of the crack are mainly affected by the properties of base material and white layer composition which is partially determined by the type of dielectrics. Previous results presented by Chen et al. [18], Yan et al. [19] and Ekmekci et al. [20] also supported this conclusion. Additionally, operational parameters, such as average discharge current and pulse-on duration were also found to play an important role in the formation of cracks [1,21]. Cusanelli et al. [22] studied the microstructure of the recast layer produced by EDM technique using kerosene as a dielectric at submicron scale. They found that the recast layer was composed of sublayers with particular structures and phases which depending on the machining energy. Cabanillas et al. [23] studied the formation of carbide by electro-discharge machining of alpha iron using kerosene as dielectric. They found that the type of carbide formed on the machined surface was determined by discharge energy. Kruth et al. [24] studied the property of recast layer formed in standard oil dielectric (BP180) by comparing it with that formed in deionized water. He found that the use of an oil dielectric increased the carbon content in the recast layer, whereas a water dielectric caused a decarbonisation. In our case, W/O emulsion was used as the dielectric, the simultaneous presence of water and oil in the discharge gap may contribute to a recast layer that is different from that formed only in water or oil-based dielectric. In this paper, the composition and crystallographical and metallurgical properties of the recast layer formed in W/O emulsion were studied by comparing them with those formed in kerosene and de-ionized water dielectrics.
2. Experimental procedures The W/O emulsion used in this experiment was prepared using, for the oil phase, 66 vol% of machine oil, and for the water phase, 34 vol% of de-ionized water. In order to stabilize the emulsion, 2.5 wt% of Span80 was added to the oil phase. Emulsification was carried out with a homogenizer (FJ200, with 18 and 12.7 mm of stator and rotor diameters, respectively) at 3000 rpm for 20 min. The microstructure of W/O emulsion after homogenization is shown in Fig. 1. The samples were machined by a spark-erosion sinking machine (NH250). The experimental set-up is illustrated in Fig. 2. Since the viscosity of W/O emulsion is much larger than that of kerosene or water, the emulsion was compelled into the discharge gap through a hole (Ø4.2 mm) in the centre of the electrode by a micro diaphragm pump. Subject of the tests was the W/O emulsion. For comparison, the tests were also done with kerosene and de-ionized water as dielectrics. During the EDM process, the pulse interval (48 s) was constant. Technologies going from roughing to half finishing have been used in order to produce a large range of the recast layer thicknesses. Samples were machined in a parametrical order using three different pulse-on durations and two different peak current settings (Table 1). A cylindrical copper rod of 30 mm diameter was used as the tool electrode. The material of the workpiece was ordinary carbon steel (C35) and the size of the workpiece was 75 mm × 35 mm × 4 mm. The machining depth was 1 mm. The copper electrode served as the positive polarity when machining was performed in W/O emulsion and kerosene, and negative polar-
Fig. 1. Optical micrograph of the W/O emulsion used in this experiment.
Fig. 2. Schematic representation of the EDM process (1, pulsed generator; 2, servocontrol; 3, electrode; 4, specimen; 5, dielectric fluid; 6, micro pump).
ity when machining was performed in de-ionized water during the EDM process. After machining, the samples were cut using a wire-EDM machine to obtain a transversal cross-section. In order to observe the recast layer using metallographic microscope and scanning electron microscopy (SEM), the samples were mirror-polished down to 1 m after having been embedded in epoxy, and then etched with nital. An energy dispersive spectrograph (EDS) associated with the SEM was also used to investigate the elements present in the recast layer after EDM. In order to identify the phases present in the recast layer, X-ray diffraction (XRD) tests have been performed with an OMNI diffractometer. The micro hardness was measured using a micro indentation device with a low load (100 mN) and a micrometric pyramidal imprint. The surface roughness (SR) was determined with a surface profilometer. The recast layer thickness (RLT) was measured with a metallurgical microscope (Nikon EclipseME600P) under a magnification of 200×. The RLT was measured at 40 locations across each cross-sectioned specimen, and an average value was calculated. The distance between the measured locations is uniform and is 50 m.
Table 1 Experimental conditions of the samples. Working parameters
Description
Peak current (A) Pulse duration (s) Polarity Dielectric
9 76 Positive (+) W/O emulsion
15 150 Negative (−) Kerosene
308 De-ionized water
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
5991
Fig. 3. Comparing the surface roughness obtained in different dielectrics and peak currents. Pulse duration = 308 s.
3. Results and discussion 3.1. SR and RLT Fig. 3 compares SR and Fig. 4 compares the surface textures of the samples obtained in different dielectrics. Surfaces with different morphologies could be obtained using different dielectrics. The difference can be easily detected with lower magnification. The samples obtained in W/O emulsion, exhibit much coarser surface with respect to those obtained in kerosene and de-ionized water. The craters formed in W/O emulsion are much larger and deeper than those formed in kerosene and de-ionized water. A comparison of the surface textures reveals that the diameter and the depth of the craters are significantly affected by the type of dielectrics. The craters formed in W/O emulsion exhibit the largest depth whereas the craters formed in de-ionized water exhibit the largest diameter and the smallest depth. These results are consistent with the reports of Konig and Jorres [16] who point out that de-ionized water usually results in lower values of surface roughness. These results may be attributed to the high viscosity of W/O emulsion. Because dielectric with high viscosity can restrict the expansion of the discharge channel, the impulsive force is concentrated within a very small area; therefore results in larger and deeper craters. Another possible reason for the big craters is the oxidation reaction of the melted material which can intensify the discharge and contribute to larger craters. Fig. 4 also shows that more micro-voids existed on the surface of the sample machined in W/O emulsion than that machined in the other dielectrics. The reason will be discussed later. Fig. 5 shows the thickness of the recast layer obtained in different dielectrics and pulse duration. As has been stated above, the recast layer is formed by molten metal which is not flushed away by the dielectric, but re-solidifies on the sample’s machined surface during cooling. It can be seen that RLT is significantly influenced by the type of dielectrics. The average thickness of the recast layer formed in W/O emulsion is larger than that formed in kerosene and de-ionized water with the same pulse duration. This is explained by the fact that the amount of molten metal which can be flushed away by different dielectrics is different. Therefore, as W/O emulsion has a much higher viscosity than kerosene, it is unable to clear away the molten material, and so it builds up upon the surface of the sample. During subsequent cooling, this molten material resolidifies to form the recast layer, the depth of which depends upon the volume of molten material which was left on the sample surface during machining. Fig. 5 also shows that the RLT and its deviation increase with increasing pulse duration. This is due to the greater material removal rate when the pulse duration is larger. It will be
Fig. 4. SEM photographs of the machined surface of (a) sample W9; (b) sample K9. Peak current = 9 A. Pulse duration = 308 s.
seen that when the EDM surface undulates dramatically, there will be significant variation in the recast layer thickness. Fig. 6 shows the RLT obtained with various peak currents. The results have revealed that the thickness of the recast layer increases with increasing peak current regardless of the dielectric type. It should be noted that, although RLT increases as the peak current increases, the increase is not very significant. This phenomenon is consistent with Lee and Tai’s research [21]. Standard deviations for measured white layer thicknesses indicate that increasing peak current increases RLT deviation considerably. These results are consistent with the observations presented by Ekmekci [1].
5992
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
Fig. 5. Comparing the RLT of samples obtained in different dielectrics and pulse duration. Peak current = 9 A.
3.2. Oxide existed in the recast layer With higher magnification, large amount of oxide can be found on the surface of samples when machining is performed in W/O emulsion and de-ionized water, as shown in Fig. 7a and b. This can be attributed to the decomposition of the water droplets dispersed in the discharge gap due to the discharge energy. As illustrated in Fig. 1, the average diameter of the droplets is about 10 m which is comparable with the distance of the discharge gap. When using W/O emulsion as dielectric, the working area was partially occupied by the water droplets which can be easily vaporized by the heat generated by the discharge and subsequently reacted with the melted material. The melted metal is oxidized by oxygen decomposed from water vapors by the discharge energy. More oxides exist on the surface of samples when machining is performed with larger peak current. This indicates that the discharge energy is an important factor that affects the formation of oxides. Since larger discharge energy can decompose more water into H and O, larger amount of oxide will be generated when larger peak current was adopted. Furthermore, the higher discharge energy can result in a higher temperature of the discharge gap, which can accelerate the oxidation reaction rate. The oxides shown in Fig. 7a and b were rarely observed on the surface of samples machined in kerosene. Although they were observed on the surface of samples machined in kerosene with larger peak current, the area occupied by the oxides is much smaller compared with samples machined in W/O emulsion and de-ionized
Fig. 7. SEM photographs show the typical surface textures of samples machined in: (a) W/O emulsion and de-ionized water; (b) W/O emulsion and de-ionized water; (c) Kerosene.
Fig. 6. Comparing the RLT of samples obtained in different dielectrics and peak current. Pulse duration = 308 s.
water dielectric. With high magnification, the typical surface texture of samples machined in kerosene is shown in Fig. 7c. These results strengthen the opinion that the formation of the oxide is due to the presence of water in the discharge gap. The oxidation can intensify the discharge and contribute to a larger crater; therefore the samples machined in W/O emulsion have a larger material removal rate (MRR) and surface roughness (SR) compared with that machined in kerosene. This experimental finding is consistent with Kunieda et al., who concluded that oxygen could be fed into the gap between workpiece and electrode to increase MRR
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
5993
Table 2 The element contents in the white layer of sample K9 and W9. Peak current = 9 A. Pulse duration = 308 s. Element
W/O emulsion mass%
Kerosene mass%
De-ionized water mass%
C O Fe Cu Total
4.1 2.1 93.8 0 100
5.1 2.7 92.2 0 100
0.7 1.7 90.0 7.6 100
obtained in W/O emulsion and de-ionized water. However, when machining is performed in W/O emulsion, the presence of iron oxide was not detected. This may be attributed to its low contents in the recast layer. All the phases present in the recast layer formed in kerosene have been reported by Chen et al. [18] and Cabanillas et al. [23]. Cabanillas et al. [23] studied the formation of carbide by EDM of alpha iron using kerosene as dielectric. In Cabanillas’ reports, the trace of some other carbide, such as Fe7C3 and Fe5C2 was detected. However, in our case, the presence of Fe7C3 or Fe5C2 was not detected. This may be attributed to their low contents in the recast layer. 3.4. EDS measurement
Fig. 8. X-ray diffraction pattern of samples machined in various dielectric and pulse duration. Peak current = 9A.
[25–27]. However, when machining in de-ionized water, the MRR is much lower than machining in kerosene or W/O emulsion. This has been attributed to the lower viscosity of water, which produces less restriction of the discharge channel, thus reducing the energy density and, as consequence, decreasing the material removal rate [16]. Moreover, the large amount of energy required to heat and vaporize water compared with oil results in lower gas pressure in the gap. Consequently the molten metal is not removed properly in every discharge, because of the insufficient pressure produced by the burst of water [28]. 3.3. X-ray diffraction (XRD) measurement The examination of the XRD patterns allowed identifying the major phases present in the samples. The diffraction patterns were measured for 10.3366 s per step from 10◦ to 100◦ with a step of 0.017◦ and analyzed using the X’Pert HighScore Plus program. The phases have been identified from searches in the PDF-2 databases. X-ray diffraction patterns for samples have shown two different trends (Fig. 8). When samples are machined in kerosene and W/O emulsion, iron carbides and ␥-Fe peaks have been observed, and the compositions of the recast layer do not show much difference. On the other hand, when samples are machined in de-ionized water, iron oxides and ␣-Fe have been observed. When machining is performed in de-ionized water, the intensity of iron oxides increases with increasing pulse duration whereas the intensity of ␣-Fe decreases with increasing pulse duration. The intensity of ␣-Fe exhibited the same trend when machining is performed in both kerosene and W/O emulsion. This indicates that the discharge energy has a significant influence on the ␣ → ␥ transformation. Iron oxide was created through chemical reaction between the workpiece material and the oxygen element decomposed from water caused by the discharge energy. The result is consistent with the SEM detection of the iron oxide on the machined surface
EDS analysis was used to measure the element composition of the recast layer after EDM. Fig. 9 compares the EDS patterns of samples machined in different dielectrics. The elements present in the recast layer are clearly indicated by the peaks corresponding to their energy levels. The relative percentages of different materials found in the surface are given in Table 2. It was observed from the analysis that the percentage of carbon element in the recast layer formed in kerosene is larger than that in the recast layer formed in W/O emulsion and de-ionized water dielectrics. It should be noted that the accuracy of carbon content detected by EDS analysis is limited since it is not sensitive to the light elements. However, comparing the EDS results of the samples, we can make a qualitative conclusion that the carbon content of the recast layer formed in kerosene is higher than that formed in W/O emulsion and deionized water. Since carbon is generated due to the decomposition of the hydrocarbon oil, the lower carbon content in the recast layer formed in W/O emulsion and de-ionized water can be attributed to the presence of water in the discharge gap with respect to kerosene case. The results were consistent with Kruth et al. [24], who found that there was a decarbonisation of the machined surface by sinking EDM when using water as dielectric. It should be noted that, when machining was performed in de-ionized water, the presence of Cu was detected in the recast layer. This is due to the migration of material between electrodes and this is confirmed to be affected by the machining parameters, such as polarity, peak currents and pulse duration [29,30]. 3.5. Micro-cracks and micro-voids in the recast layer Crack formation can be attributed to the presence of thermal stress and tensile stress within the machined surface. The primary causes of the thermal stress in the machined surface were the drastic heating and cooling rates and the non-uniform temperature distribution [31]. Tensile stress within the sample is generated because not all of the material which melts during the machining process is swept away from the component’s surface by the dielectric. Due to the ingress of carbon, the melted material contracts more than the unaffected parent part during the cooling process, and when the stress in the surface exceeds the material’s ultimate tensile strength, cracks are formed [19,31–33].
5994
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
Fig. 9. EDS analysis of the samples machined in various dielectric. Pulse duration = 308 s; Peak current = 9A.
Ekmekci [1] classified the cracks formed in EDM process into three types. The first type of cracks, called surface cracks, exists in the white layer, initiating at its surface and traveling down perpendicularly toward the interferential zone, and usually terminates at this interference. Such cracks have high potentials to partial flaking off, which also proved the weakness of the intermediate layer between the recast and heat affected layers. The second type of cracks, called penetrating cracks, penetrates the entire white layer thickness to an extent into the parent material. In our experiments, the second type cracks that penetrated into the parent material were not found within all the machining conditions, although they have been reported by other researchers [21,34]. The third type of cracks is visualized usually around globular or irregularly shaped attachments on crater rims. Such cracks have very low penetration depth and are randomly distributed on the surface. Thus, have little influence on the surface properties. In our case, the cracks that existed on the recast layer formed in water dielectric are mainly the third type cracks which have very low penetration depth and are randomly distributed on the surface especially where the curvature is larger, such as the rims of the craters (Fig. 10). So, we infer that the larger density of cracks may be explained by the higher cooling rate when machining is performed in water, since a faster cooling rate has the ability to introduce a larger residual stresses, thus increasing the tendency for cracks to form. Because of their low penetration depth and width, it is hard to detect them in the cross section (Fig. 11). Observation of the external and sectional micrograph of the samples shows that the cracks formed in W/O emulsion and kerosene dielectric exhibit similar characteristic, and they can mainly be classified to the first type. The results presented in our experiments confirm the conclusion of Ekmekci that the first type of cracks are usually observed when machining is performed in a hydrocarbon-based dielectric liquid with high pulse-on duration and low discharge current [1]. In order to give a quantitative research of the first type of cracks, the cracks are quantified by the total length of cracks (mm) in a unit area (mm2 ). Fig. 12 presents an analysis of this measure for the samples, and shows how the sur-
face crack density varies with dielectric type. Observation of these results shows that the presence of surface cracking is much more evident when machining is performed in W/O emulsion, and the density of crack is greater than the kerosene and de-ionized water case. The results also show that the surface crack density decreases significantly with the increase of peak currents regardless of the dielectric type. Previous results presented by Lee et al. [21,35] also supported this conclusion. It should be noted that the cracks often originate from the border of micro-voids regardless of the dielectric type. The micro-cracks were associated with the development of thermal stresses exceeding the ultimate tensile strength of the material. The presence of micro-voids will reduce the local service strength of the recast layer and increase the cracking possibility. On the other hand, the presence of micro-voids leads to a drastic change of the surface curvature which is an important factor that affects the cooling rate of material. Fig. 11 compares the sectional micrograph of the samples obtained in W/O emulsion, kerosene and de-ionized water. There are many micro-voids scattered in the recast layer formed in W/O emulsion. The micro-voids can also be seen on the external surface of the sample (Figs. 4 and 10). In the micrograph of both sectional and external surface, the occurrence of such micro-voids was rarely observed in the recast layer formed in kerosene and de-ionized water. This is explained by the presence of water droplets in the discharge gap. When machining is performed in W/O emulsion, the gas volume generated in the discharge gap was larger than in the case of kerosene due to the decomposition and vaporization of water, since water has a much lower boiling point with respect to kerosene. When machining is performed in de-ionized water, just as the case of kerosene, the volume of generated gas is much smaller than the case of W/O emulsion. The results indicated that the small water droplets that scattered in the discharge gap can be vaporized more easily than the continuous water that filled the discharge gap due to its bigger specific surface area. Although some gases were generated when machining is performed in kerosene [36,37] and de-ionized water, their volume is much smaller than
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
5995
Fig. 11. Cross-sectional micrographs of samples machined in various dielectrics. Pulse duration = 308 s; Peak current = 9A.
Fig. 12. Comparing of the crack density of the recast layer formed in different dielectrics and peak currents. Pulse duration = 308 s.
Fig. 10. SEM photographs show micro-cracks on the surface of samples machined in various dielectrics. Pulse duration = 308 s; Peak current = 9A.
the W/O emulsion case. In order to provide quantitative results of the gases volume, the gases generated during the EDM process were collected by a gas gather devise. The volumes of gases generated in 1 min in different dielectrics are listed in Table 3. When
machining is performed in W/O emulsion, the volume of generated gas is much larger than that generated in the case of kerosene and de-ionized water. The micro-voids can be attributed to the gas bubbles expelled from the molten material during solidification [31]. The presence of large volume of gases in the gap, especially water vapor and hydrogen, will lead to a high supersaturation of gas in the molten pool, which is the prerequisite for the formation of micro-voids.
Table 3 The volume of gases generated in 1 min in various dielectrics and peak current. Dielectric Peak current (A) Volume of generated gas (ml)
W/O emulsion 9 114 ± 13
Kerosene 15 293 ± 20
9 16 ± 3
De-ionized water 15 34 ± 5
9 20 ± 5
15 42 ± 7
5996
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
Fig. 13. Comparing of the micro hardness of the recast layer formed in different dielectrics and peak currents. Pulse duration = 308 s.
3.6. Micro hardness In order to characterise the mechanical properties of the recast layer, a micro hardness test has been performed using a micro hardness tester with a low load (100 mN) and a micrometric pyramidal imprint. Before the measurement, the samples were mirror-polished down to 1 m after having been embedded in epoxy, then etched with nital for a very short time to make the recast layer visible. The micro hardness of the recast layer was measured at 30 locations in the middle zone of the recast layer, and an average value was calculated. Fig. 13 compares the micro hardness of the recast layers formed in different dielectrics. The micro hardness of the recast layer obtained in different dielectrics can be ranked as (in decreasing order) W/O emulsion–kerosene–water; so does the standard deviation of the micro hardness. This indicates that the recast layer formed in W/O emulsion has a more inhomogeneous structure than that formed in kerosene and de-ionized water. The micro hardness of recast layer machined in de-ionized water is lower than that of the recast layer machined in kerosene and W/O emulsion. This can be explained by its lower carbon contents. Previous results presented by Kruth et al. [24] also supported the same conclusion. Just as shown by the above analysis, the carbon content in the recast layer formed in W/O emulsion is lower than that formed in kerosene. But the micro hardness of the recast layer formed in W/O emulsion is higher than that formed in kerosene. It seems that besides the ingress of carbon, the high cooling rate is another possible reason for the hardness increase of the recast layer when machining is performed in W/O emulsion; since the high cooling rate can make the recast layer freeze into a martensitic structure which is also known to give a high hardness. Fig. 14 shows the martensitic structure in the recast layer when machining is performed in W/O emulsion and kerosene dielectric. Although it is observed in the recast layer when kerosene was used as dielectric, it is more common when machining is performed in W/O emulsion. The martensitic structure shown in Fig. 14 was not detected in the recast layer formed in water dielectric due to the lower carbon content. In order to study the influence of cooling rate, the micro hardness of the recast layer of the samples was also measured at different distances from the interface. As has been reported by Ekmekci [1] and Cusanelli et al. [22], there are two solidification fronts during the cooling of the melted material. Heat convection takes place at the outer surface which is in contact with the dielectric, whilst at the interface, convection and conduction both contribute to extract the heat. Within the recast layer, three points were chosen for measurement; A, point close to the external surface; B, point in the middle zone; C, point close to the inter-
Fig. 14. Comparing of the micro hardness depth profile of sample W9 and K9. Pulse duration = 308 s; Peak current = 9A.
face. Moreover, the micro hardness of HAZ and matrix was also measured. The outcome of the measurements reveals that the minimum value of the micro hardness usually appeared in the middle zone of the recast layer, whereas the maximum value appeared in the zone close to external surface or the interface. This indicates that the cooling rates of the melted material do have a remarkable influence on micro hardness of the recast layer.
4. Conclusion (1) The recast layer formed in W/O emulsion dielectric exhibits a greater surface roughness and thickness than that formed in kerosene and de-ionized water dielectrics. (2) Both carbide and oxide are formed in the recast layer when machining is performed in W/O emulsion dielectric, whereas only carbide is formed in the case of kerosene and only oxide is formed in the case of de-ionized water. The discharge energy has a positive effect on the formation of oxide. (3) The micro-cracks of the recast layer formed in water dielectric have very low penetration depth and are randomly distributed over the surface. However, cracks that formed in W/O emulsion and kerosene dielectrics exist in the recast layer and have much larger depth and usually terminate at the recast and heat affect layers interface. The density of crack decreases as the peak current increases regardless of the dielectric type. (4) Because of the high supersaturation of gas in the molten pool, there are many micro-voids within the recast layer when machining is performed in W/O emulsion. The micro-voids are rarely observed within the recast layer when machining is formed in kerosene and de-ionized water dielectrics. (5) Due to the ingress of carbon and high cooling rate, the recast layer formed in W/O exhibits a greater micro hardness than that formed in kerosene and de-ionized water dielectrics.
Acknowledgements The work is partially supported by a grant from Chinese National Natural Science Foundation (Grant No. 50675225) and a grant from Department of Science & Technology of Shandong Province (Grant No. 2006GG2204001).
Y. Zhang et al. / Applied Surface Science 257 (2011) 5989–5997
References [1] B. Ekmekci, Metall. Mater. Trans. B: Process Metall. Mater. Process. Sci. 40 (2009) 70–81. [2] Y.S. Liao, J.T. Huang, Y.H. Chen, J. Mater. Process. Technol. 149 (2004) 165–171. [3] M. Kunieda, B. Lauwers, K.P. Rajurkar, B.M. Schumacher, CIRP Ann. Manuf. Technol. 54 (2005) 599–622. [4] S. Kumar, R. Singh, T.P. Singh, B.L. Sethi, J. Mater. Process. Technol. 209 (2009) 3675–3687. [5] D.I. Pantelis, N.M. Vaxevanidis, A.E. Houndri, P. Dumas, M. Jeandin, Surf. Eng. 14 (1998) 55–61. [6] T. Moro, N. Mohri, H. Otsobo, A. Goto, N. Saito, J. Mater. Process. Technol. 149 (2004) 65–70. [7] W.S. Zhao, Y. Fang, Z.L. Wang, L.H. Li, Mater. Sci. Forum 471–472 (2004) 750–754. [8] P.K. Patowari, U.K. Mishra, P. Saha, P.K. Mishra, Int. J. Manuf. Technol. Manage. 21 (2010) 83–89. [9] Z.L. Wang, Y. Fang, P.N. Wu, W.S. Zhao, K. Cheng, J. Mater. Process. Technol. 129 (2002) 139–142. [10] J. Simao, H.G. Lee, D.K. Aspinwall, R.C. Dewes, E.M. Aspinwall, Int. J. Mach. Tools Manuf. 43 (2003) 121–128. [11] P.K. Patowari, P. Saha, P.K. Mishra, Int. J. Adv. Manuf. Technol. (2010) 1–12. [12] Y.F. Chen, Y.C. Lin, J. Mater. Process. Technol. 209 (2009) 4343–4350. [13] B.H. Yan, Y.C. Lin, F.Y. Huang, C.H. Wang, Mater. Trans. 42 (2001) 2597– 2604. [14] K. Furutania, A. Saneto, H. Takezawa, N. Mohri, H. Miyake, Precis. Eng. 25 (2001) 138–144. [15] T. Masuzawa, K. Tanaka, Y. Nakamura, N. Kinoshita, CIRP Ann. Manuf. Technol. 32 (1983) 119–122. [16] W. Konig, L. Jorres, CIRP Ann. Manuf. Technol. 36 (1987) 105–109.
[17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28]
[29] [30] [31] [32] [33] [34] [35] [36] [37]
5997
Y.H. Guu, H. Hocheng, C.Y. Chou, C.S. Deng, Mater. Sci. Eng. A 358 (2003) 37–43. S.L. Chen, B.H. Yan, F.Y. Huang, J. Mater. Process. Technol. 87 (1999) 107–111. B.H. Yan, H.C. Tsai, F.Y. Huang, J. Mach. Tools Manuf. 45 (2005) 194–200. B. Ekmekci, O. Elkoca, A. Erden, Metall. Mater. Trans. B: Process Metall. Mater. Process. Sci. 36 (2005) 117–124. H.T. Lee, T.Y. Tai, J. Mater. Process. Technol. 142 (2003) 676–683. G. Cusanelli, W.A. Hessler, F. Bobard, R. Demellayer, R. Perez, R. Flukiger, J. Mater. Process. Technol. 149 (2004) 289–295. E.D. Cabanillas, J. Desimoni, G. Punte, R.C. Mercader, Mater. Sci. Eng. A 276 (2000) 133–140. J.P. Kruth, L. Stevens, L. Froyen, B. Lauwers, CIRP Ann. Manuf. Technol. 44 (1995) 169–172. M. Kunieda, S. Furuoya, N. Taniguchi, CIRP Ann. Manuf. Technol. 40 (1991) 215–218. M. Kunieda, M. Yoshida, N. Taniguchi, CIRP Ann. Manuf. Technol. 46 (1997) 143–146. M. Kunieda, Y. Miyoshi, T. Takaya, N. Nakajima, Y.Z. Bo, M. Yoshida, CIRP Ann. Manuf. Technol. 52 (2003) 147–150. T. Masuzawa, Machining characteristics of EDM using water as dielectric fluid, in: Proceedings of the 22nd Machine Tool Design and Research Conference, Manchester, 1981, pp. 441–447. J.S. Soni, G. Chakraverti, J. Mater. Process. Technol. 56 (1996) 439–451. S. Zinelis, Dent. Mater. 23 (2007) 607–1607. Y.H. Guu, M.T. Hou, Mater. Sci. Eng. A 466 (2007) 61–67. B. Ekmekci, A.E. Tekkaya, A. Erden, J. Mach. Tools Manuf. 46 (2006) 858–868. B. Ekmekci, Appl. Surf. Sci. 253 (2007) 9234–9240. C.P. Tabrett, J. Mater. Sci. Lett. 15 (1996) 1792–1794. H.T. Lee, F.C. Hsu, T.Y. Tai, Mater. Sci. Eng. A 364 (2004) 346–356. G.N. Levy, CIRP Ann. Manuf. Technol. 42 (1993) 227–230. F.N. Leão, I.R. Pashby, J. Mater. Process. Technol. 149 (2004) 341–346.