Electrical Discharge Grinding of Polycrystalline

M. Zulafif Rahim1 School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University East Campus, Melbourne, VIC 3083, Australia; Faculty of Mechanical and Manufacturing Engineering, Universiti Tun Hussein Onn Malaysia (UTHM), P.O. Box 101, Parit Raja, Batu Pahat 86400, Johor, Malaysia e-mail: [email protected]

Songlin Ding Senior Lecturer School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University East Campus, Melbourne, VIC 3083, Australia e-mail: [email protected]

John Mo Professor School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University East Campus, Melbourne, VIC 3083, Australia e-mail: [email protected]

Electrical Discharge Grinding of Polycrystalline Diamond— Effect of Machining Parameters and Finishing In-Feed Electrical discharge grinding (EDG) is becoming more prevalent in the manufacturing of polycrystalline diamond (PCD) tools. This paper concerns investigation of the effects of machining parameters, as well as finishing in-feed, to the surface quality obtained when using EDG to erode PCD. With the aid of the morphological findings, different PCD erosion mechanisms are discussed. Experimental results demonstrated that the eroded surface quality of PCD was significantly affected by the selected parameters. High temperature due to the erosion process resulted in the partial conversion of diamond to graphite phase under the surface. Higher finishing in-feed produced better surface quality and caused lower surface graphitization and lower tensile residual stress. A model for the thermal stress prediction was developed and found to have good agreement with the experimental findings. [DOI: 10.1115/1.4029433]

Introduction In fabricating PCD tools, a PCD chip is cut from larger PCD blanks (produced by the sintering process), and this chip is brazed on the tool/carrier substrate [1] before being machined to the desired shape. The residual stress in the PCD tools is the result of three different phases of manufacture. These are the sintering process, the brazing process that mounts the PCD chip on the tool holder, and the final machining processes. It is believed that the overall residual stress, which accumulates on the finished tool surface, is related to the combination of these three phases of manufacture as summarized in Fig. 1. The residual stress should be managed properly so that the toughness properties of the tools can be sustained [2,3]. In fact, Yahiaoui et al. [4] mentioned that the tensile residual stress in diamond promotes cracks, weakens the grain boundaries, and contributes to lowering abrasion resistance. In rock drilling application, it was found that residual stress contributed to the formation of gross fracture wear of diamond tools [5]. The coefficient of thermal expansion (CTE) mismatch between PCD and tungsten carbide (WC), which constitutes the substrate, results in stress and strain development during the cooling process [3,6]. Chen et al. [7] discovered that the thickness ratio of the diamond layer to the tungsten carbide backing significantly affects the value and the distribution of residual stress. Jia et al. [8] proved that the PCD of bigger grain size has higher compression residual stress [8]. The compression stress is also reported to be higher at regions closer to the diamond–WC interface due to the significant difference in CTE between PCD and WC [6,9]. Significant thermally induced stress on the intersection tends to cause delamination failure during the application of PCD tools. Special 1 Corresponding author. Contributed by the Manufacturing Engineering Division of ASME for publication in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received June 20, 2014; final manuscript received December 15, 2014; published online February 4, 2015. Assoc. Editor: Y.B. Guo.

design of irregular surface topography on the intersection between the PCD and carbide layers was proposed in order to reduce the probability of delamination-crack propagation [10]. The magnitude and direction of residual stress for the third phase is different with different production strategies [11]. In the machining of steel, for example, the thermal effect produces a different residual stress value with respect to the surface depth due to the nonuniform temperature distribution [12]. Electrical discharge machining (EDM) of ordinary materials usually induces the highest tensile stress just below the machined surface [13,14] and the value is reported to be directly proportional to the pulse energy used [15,16]. However, no discovery was reported on the effect of EDM plasma thermal stress on the surface strength alteration of PCD. It was reported that surface graphitization occurred due to the EDM of PCD but this has not yet been thoroughly researched. To date, the relationship of the thermal damage to the controllable machining parameters had not been well defined in the literature. Theoretical approaches should be developed in order to predict and estimate the degree of thermally damage to the surface. Understanding of the material removal mechanism is required in order to fully understand the thermal damage development process but the scarcity of relevant information on the removal mechanism creates a challenge for further exploration. Olsen et al. [1] reported that the PCD diamond grains are detached from the surface due to the preferential erosion of the highly conductive cobalt network. On the other hand, Kozak et al. [17] believed that the

Fig. 1 Three phases of the residual stress development in PCD tools production

Journal of Manufacturing Science and Engineering C 2015 by ASME Copyright V

Downloaded From: http://asmedigitalcollection.asme.org/ on 02/25/2015 Terms of Use: http://asme.org/terms

APRIL 2015, Vol. 137 / 021017-1

Table 1 EDG parameters for the roughing experiment Specimen no.

Polarity

Open voltage (V)

Current (A)

On-time (ls)

Off-time (ls)

1 2 3

Negative

120

12 10 8

40

20

cobalt binder is fractured due to the thermal stress generated by the process, which causes diamond grains to separate from the structure. Zhang et al. [18] thought the high temperature plasma not only melts the cobalt but also graphitizes the diamond grains. The graphite consequently dissolves into the molten cobalt before being flushed away by the dielectric. This demonstrates disagreement on the fundamental mechanism of material removal and a better explanation is demanded. This manuscript reports on the surface graphitization and residual stress developed under the surface of PCD machined by the EDG process. The mechanism of removal and its theoretical model are discussed in detail and found to be in good agreement with the experimental findings.

Equipment and Methodology Although PCD is a strong and hard material it is sensitive to high temperature [19,20]. It has been reported that the graphitization of diamond occurs at temperatures above of 973 K when in the presence of cobalt. In addition, up to 36% of diamond’s hardness reduction was reported when the temperature increases from 300 K to 550 K [20]. Since EDG of PCD tools generally involves a high energy roughing process the authors theorized that graphitization would be present on the PCD surface. The research was begun with the roughing process investigation of CTB010 (PCD with 10 lm grain size). The samples were machined with three different roughing parameters as shown in Table 1. Secondary electron images from a scanning electron microscope (SEM) indicated the formation of an affected layer up to 20 lm depth. In this layer, microstructure changes due to heat from roughing process were observed. Further investigation was done using three different PCD types of 4.6 mm width. Table 2 shows the properties of these three PCD types [21,22]. Due to depth of the observed modified zone, it was hypothesized that the finishing in-feed could make a significantly difference to the final surface quality. Two-stages of machining, roughing and finishing, were performed for every sample. For each test, the same roughing process was used to remove material to a depth of 0.5 mm. This was followed by different finishing processes. Table 3 shows the erosion parameters used in this series of experiments. A tungsten-copper wheel acted as the anode for the roughing operation and the cathode during the finishing stage. Before every surface analysis, the samples were ultrasonically cleaned using ethanol and acetone to remove any contamination on the surface. For structural quality assessment, Raman Spectra with a 785 nm (near infrared) laser wavelength (Perkin Elmer Spectrum Raman Station 400) was used to analyse several types of carbon formed on the surface. With 100 lm laser spot size, good averaging of the Raman values of the inspected surface was obtained.

Result and Discussion Effect of Roughing Process. Raman spectroscopy has been successfully used in the research relating to diamond material. Several studies show that the formation of peak bands on a particular range of wave number is an indicator, which differentiates between several types of carbon lattice structures formed [23,24]. The formation of the band that is centralized on a specific number is the indication of the presence of carbon lattice structure and the

deviation of peak can be used for the surface stress measurement [25,26]. SEM study of roughly eroded PCD found that the modified region had a thickness of approximately 10–20 lm from the eroded surface as shown in Figs. 2 and 3. The formation of this modified zone was believed to be due to the microstructure changes resulting from heat produced in the roughing operation. In the case of a single crystal diamond, simultaneous or explosive disintegration and graphitization of diamond with the activation energy (Eac) of 42 6 8 kJ/mol was reported when the temperature reached 2273 K [27]. Below this explosive disintegration level, graphitization occurred at an activation energy (Eac) of 336 6 21 kJ/mol, and this graphitization is known as the diffusive mechanism [27]. At this stage, a significant reduction of the graphitization rate is achieved. The preceding theories provide the best explanation for the appearance of the modified layer up to a certain depth until the level where the graphitization transition of explosive to diffusive mechanism is reached. In the case of machining PCD, the availability of residual cobalt catalyzes graphitization and this reduces the onset temperature for explosive graphitization. Therefore, explosive graphitization at temperature lower than 2273 K should be expected. The Raman spectrum obtained from the eroded surface after a roughing operation was compared with the spectrum of the unmodified PCD surface (as received from the material supplier). It was also compared with the spectrum from a single crystal diamond and the results are presented in Fig. 4. The spectrums obtained from the unmodified PCD demonstrate the three different Raman peaks at 1332 cm1, 1250 cm1, and 1600 cm1. These peaks indicate the presence of cubic diamond, nanocrystalline diamond, and graphite peaks, respectively [23,28–31]. Sumiya [32] stated that the graphite produces nanocrystalline diamond when subjected to high pressure and high temperature in the manufacturing process. The same conditions were applied to synthetic diamond during sintering process to make our PCD sample. Graphite, formed by the graphitization of diamond particles at high temperature, would have been converted to nanocrystalline diamond under the high pressure, high temperature conditions, thus explaining the appearance of the nanocrystalline diamond 1250 cm1 peak, similar to that found by Ali et al. [29]. In contrast to the spectra obtained for the unmodified PCD, the spectra obtained from the eroded PCD section demonstrated only the presence of graphite and wide cubic diamond, defined by the Raman peaks at position of 1306 cm1 and 1599 cm1, respectively. This means that during the erosion process, nanocrystalline diamond is fully graphitized. This phenomenon is due to the higher surface-area-to-volume ratio of nanocrystalline diamond that is in contact with cobalt. Since as an object gets smaller the surface-area-to-volume ratio goes up, a nanosized diamond in PCD will have a very large surface area in contact with cobalt relative to its volume when compared to the larger crystals which constitute the main form of diamond in PCD. This means that the graphitisation encouraged by the catalytic action of cobalt would be expected to quickly eliminate all diamond nanoparticles. As the intensity ratio between the graphite and diamond peaks relates to the graphitization degree [33,34], the lower diamond to graphite intensity ratio of the eroded PCD when compared to the unmodified PCD is an indication of partial conversion of diamond into graphite under the surface. The conversion not only weakens the structure but also causes significant irreversible expansion, due to the difference between diamond and graphite density [35].

021017-2 / Vol. 137, APRIL 2015


Transactions of the ASME

Table 2 Properties of PCD PCD types

PCD grain size (lm)

Diamond fraction (Vol. %)

Cobalt fraction(Vol. %)

2 10 30 to 2

84.8 89.7 91.4

15.2 10.3 8.6

CTX002 CTB010 CTM302

Table 3 EDG parameters for the finishing experiment Operation

Annotation

Polarity

Open-voltage (V)

Wheel rotation speed (rpm)

Current (A)

On-time (ls)

Off-time (ls)

In-feed (mm)

Roughing Finishing

— A

Negative Positive

120

250

12 1

40 1

20 1

B

4

1

1

C

1

4

1

D

4

4

1

E

1

1

4

F

4

1

4

G

1

4

4

H

4

4

4

0.5 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02

The diamond peak value far from the unstressed cubic diamond value was identified as damaged diamond similar to that discussed in an earlier manuscript [36]. According to the previous research, the higher the diamond peak position shifted further away from the stress-free diamond peak position (pure diamond grains), the higher the stress value was [37]. With the position shifted to a lower value, it was concluded that the roughing operation stressed the surface in a tensile direction. The stress values in this study were calculated using the following formula [26]: r ¼ ðvs vr Þ=v

(1)

where r is the tensile residual stress (GPa), and s is the measured Raman shift value of the diamond (1306 cm1), r is the unstressed Raman value (1330 cm1) and v is the coefficient of

stress-induced frequency shift. Instead of using 1.98 cm1/GPa for the value of v as used in Ref. [38], 2.88 cm1/GPa was chosen as the more reasonable value, due to the usage of high volume of cobalt as referred to Ref. [26]. From the equation, 8.33 GPa tensile residual stress was obtained from the surface eroded by the 12 A roughing current. At high erosion powers such as this, the removal mechanism was believed to be due to mechanical failure, caused by thermal expansion of PCD and cobalt metal [17]. CTE mismatch between these two materials generates thermal stress during high temperature erosion, which causes dislocation and this promotes the breaking of diamond to diamond bonds. SEM investigation of the eroded surface found an average crater size of 10 lm, similar to that of the PCD grain size. This constitutes evidence for the hypothesized mechanical failure mechanism (Fig. 5). Grain detachment occurred, leaving cobalt as residue in

Fig. 2 SEM images of the PCD after roughing (modified zone was clearly observed on the 12 A PCD sample)

Journal of Manufacturing Science and Engineering


APRIL 2015, Vol. 137 / 021017-3

Fig. 3 Secondary electron (left) and Backscatter Image (right) of modified zone for 12 A PCD sample

Fig. 4

Raman analysis

the crater as shown by the backscatter images. Since the cobalt is a heavier element, it appears brighter in color. Theoretical Model. For the roughing process, a physical model of the thermal stress from cobalt–diamond interaction was

Fig. 5 Secondary electron images (left) and backscatter images (right) of the eroded PCD by roughing

obtained. It is similar to that used by the authors for the crater size prediction of PCD when undergoing EDM [39]. The model emphasizes diamond–to–cobalt interaction during the heating process. The stress components and annotations are indicated in Fig. 6. In this model, the thermal displacement, u, and the thermal stress, r, in radial and tangential directions was calculated using the following basic formula used by Kozak et al. [17]: d 1 d 2 að1 þ vÞ dh ðr (2) uÞ ¼ dr r 2 dr 1 v dr

Fig. 6 Thermal stress components considered in the model (r 5 position where the stress value is estimated, a 5 diamond grain radius, r 5 stress component for tangential, t and radial, r direction) [15]

021017-4 / Vol. 137, APRIL 2015



du 2u du þ þ 2l ð3k þ 2lÞah rr ¼ k dr r dr du 2u u rt ¼ k þ þ 2l ð3k þ 2lÞah dr r r

ð1Þ

rt

(3)

rð2Þ r ¼ (4) ð2Þ

where r is the position at which the stress value is determined, a is the diamond grain radius, u is the thermal displacement, a is the CTE, h is the temperature difference, and r is the stress component, which is annotated by t for tangential direction and r for radial direction. The elastic modulus, k, in terms of the Lame’s first parameter is a function of Poisson’s ratio, v, and modulus of elasticity, E, given by vE (5) k¼ ð1 þ vÞð1 2vÞ The lame’s second parameter (shear modulus), l, can be calculated as l¼

E 2ð1 þ vÞ

(6)

The equation was then simplified, and the general equations shown below were obtained: ð A C2 (7) u ¼ 2 hr 2 dr þ rC1 þ 2 r r ð 4lA C2 rr ¼ 3 hr 2 dr þ BC1 3 (8) r r ð 2lA 2lC2 rt ¼ 3 hr 2 dr þ BC1 3 2Ah (9) r r where the constants A and B are að1 þ vÞ ð1 vÞ

(10)

B ¼ 3k þ 2l

(11)

A¼

The following boundary condition was applied to the physical model derived: r¼0 u¼0 r¼a

ð1Þ

ð2Þ

uð1Þ ¼ uð2Þ rr ¼ rr ð2Þ

r ! 1 rr ¼ 0 Similar radial stress is gained when “r” is equal to “a”, where the subscript 1 and 2 represented the component for diamond and cobalt, respectively. By solving Eqs. (7), (8), and (9) using the above boundary condition, C1, and C2 were determined r¼0 r¼a

ð1Þ

C2 ¼ 0

(12)

1 ð1Þ ð2Þ ð2Þ A1 Ta þ aC1 ¼ aC1 þ a2 C2 3

(13)

4 ð1Þ ð2Þ ð2Þ l1 A1 T þ B1 C1 ¼ B2 C1 4l2 a3 C2 3 4 ð2Þ r!a l2 A2 T þ B2 C1 ¼ 0 3

¼ rð1Þ r

(14) (15)

Substituting the C1 and C2 into Eqs. (8) and (9), the thermal stress components satisfied the following equations: 4 1 ¼ ðA2 B1 ðB2 þ 4l2 Þ l2 rð1Þ r 3 B2 ðB1 þ 4l2 Þ þ A1 B1 B2 ðl1 l2 Þ A1 B2 ðB1 þ 4l2 Þl1 Þ h (16)

rt

(17)

3

a ð1Þ r r3 r

1 1 a3 ð1Þ ¼ rð2Þ r r ¼ 2 2 r3 r

(18)

(19)

The final equation for the thermal stress components was obtained by substituting diamond and cobalt properties found in Refs. [17], [40], and [41] (Table 4) into the equation. Unlike Kozak et al. [17], temperature dependent properties of diamond were taken into consideration in developing the final equation. CTE for diamond relative to temperature was obtained by Thewlis and Davey [40]. The value follows three different line trends, segregated by three different temperature stages, which are 175 K–275 K for the first stage, 275 K–400 K for the second stage, and 400 K–1175 K as the final stage. The stages mentioned above are satisfied by the piecewise linear equations as shown in Fig. 7. Diamond–to–diamond is the primary bonding mechanism in PCD as reported in Refs. [42–44]. Thus, grain dislodgement only occurs when the stress is more than the tensile limit of diamond. This is referred to as the removal mechanism through diamond fracture. The authors’ assumption seemed to contradict with Kozak et al., who believed the dislodgement happens when the tensile limits for cobalt is reached [17]. For this reason, only the stress for diamond is investigated further. By incorporating the piecewise linear equation with the thermal stress components, Eqs. (16) and (17), the stress value in MPa can be estimated using the following modified equations: 7:43h þ 2:4ð105 ÞhT 273:15 T < 400 (20) rt ¼ rr T 400 7:44h þ 4ð106 ÞhT where T is the temperature of the material. A thermal analysis model was created via ANSYS software to predict the temperature profile so that the stress value could be determined. The energy supply for every spark was calculated by the following equation [45,46]: Ed ¼

ð t0 iðtÞUðtÞdt

(21)

0

where Ed, t, i, and U is the energy, pulse duration, current peak value, and voltage, respectively. Hence, the heat flux q, which is the energy supplied per unit area can be calculated using the following equation [45] and by assuming that the plasma generated is symmetrical in shape: qðtÞ ¼

kEd prp2

(22)

where rp is the heat flux radius and k is the energy fraction obtained by the cathode. DiBibonto et al. [47,48] used the constant energy partition of 8% and 18% for the anode and cathode erosion model of EDM of steel using a copper electrode. Shankar et al. [49], in their investigation, found that the current, pulse offtime, and the electrode gap does not significantly affect the energy partition, but lower thermal diffusivity which would cause the lower share of input power. Their finding demonstrates that around 40% of the total energy is transferred to the electrode while using water as the dielectric and with steel and copper as the electrodes. There is no comprehensive method that has been proposed thus far to determine the exact energy partition of the EDM process. In this analysis, the k value is assumed to be 15% of the total energy supplied. The electrode materials may well influence the spark radius. However, it is extremely difficult to prove this due to extremely short pulse durations [50]. Compared with the steel, the lower



APRIL 2015, Vol. 137 / 021017-5

Table 4 Properties of diamond and cobalt

Modulus of elasticity, E (GPa) CTE, a (106 K1) Poisson’s ratio, v Shear Modulus, l (GPa)

Diamond

Cobalt

900 At 175 K T < 275 K; a ¼ 0:0075 T þ 2:1125 At 275 K T < 400 K; a ¼ 0:01856 T 5:0595 T 400 K; a ¼ 0:0032 T þ 1:1442 0.18 420

211 14.2

rp ¼

0.32 80

2:4 103 t0:4 i0:4 2

(23)

By applying this equation, it is assumed that the plasma radius grows with time. The value of voltage and current are obtained from the feedback system with which the machine is equipped. Calculated results obtained from Eqs. (19) and (20) are shown in Fig. 8. The following assumptions are applied to the thermal model: (a) The initial temperature of the process was set as 20 C, which is close to room temperature. (b) Heat flux is considered as originating in the heat source from the plasma created by the process. (c) Thermal convection is applied to the contact surface between PCD and dielectric. (d) The material is assumed to be an isotropic and homogeneous material. (e) The plasma radius was assumed to be symmetrical. Fig. 7 Temperature dependent properties of diamond

electrical conductivity of PCD directly results in a lower value of current in the gap than the input current set by the operator which consequently lowers the plasma radius and heat flux value on the gap. Following equation from Ref. [45], the heat flux radius, rp can be calculated by

Figure 9 shows the scheme for the boundary conditions applied. For boundary A 8 h ðT T0 Þ when rp > rp max @T < c ¼ q when rp rp max (24) k @y : 0 off time where hc is the heat transfer coefficient between the material surface and dielectric. In the case of single sparks, a semi-infinite body was considered. Thus, the effect of the surroundings could be neglected. In order to obtain this condition, the radius of the body is assumed to be at least six times that of the maximum spark radius calculated [49]. The following boundary condition was applied for boundary B, C, and D: k

@T ¼0 @g

(25)

where g is the direction normal to the surface.

Fig. 8 Current, voltage spark radius, and heat flux value used in the model

Fig. 9 Scheme for the boundary conditions

021017-6 / Vol. 137, APRIL 2015



For defining the material characteristic, the temperature dependent properties of PCD of different grades are calculated based on a number of theoretical formulations developed for ceramic materials with granular grain shapes, as assumed to have similar physical morphology as the structure of PCD used. The thermal conductivity, K of PCD was calculated using Maxwell equation as follows (similar to as used in Ref. [51]) K ¼ kd

2kd þ kc 2fc ðkd kc Þ 2kd þ kc þ fc ðkd kc Þ

(26)

where kd, kc, and fc are the thermal conductivity of diamond, thermal conductivity of cobalt, and cobalt fraction, respectively. The heat capacitance C, which also known as the specific heat of material, was calculated using the Neumann–Kopp’s law C ¼ cc fc þ cd fd

the thermal stress is higher than the thermal stress limit for diamond. Therefore, the grains fracture and dislodgement occurs. The grains are separated by the pressure wave generated by the process and flushed away by the dielectric. Effect of Finishing Operation. As stated in the previous section “Effect of Roughing Process,” after roughing, a 20 lm deep modified zone is observed together with high surface graphitization and tensile surface residual stress. As partial graphitization was predicted to occur in only a few microns depth toward the diamond bulk, deeper finishing removal is expected to produce a better surface quality. As shown in Fig. 13, there is no evidence for the diamond breakage occurring on the surface eroded by the

(27)

where cc is the heat capacitance for cobalt, cd is the heat capacitance for diamond, and fd is the diamond fraction Figure 10 shows the calculated results for the temperature dependent properties of different PCD grades. The temperature profile obtained from the ANSYS analysis is shown in Fig. 11. Integrating the stress equation with the temperature profile obtained enables the residual stress to be predicted. The tensile residual stress measured earlier indicates a value of 8.33 GPa, which is in agreement with the highest residual stress reported by Ferreira et al. in their research regarding the residual stress of boron-doped diamond films [52]. As the roughing process removes diamond by the fracture mechanism, it is reasonably accurate to assume that the maximum thermal stress for diamond is close to that value (in this case, the thermal stress limit for diamond was assumed to be 8.5 GPa by taking into account the stress relaxation on the exposed surface). This value is reasonable, since the strength of diamond was reported as being close to that value at the temperature of 1573 K [53]. In Fig. 11, a straight horizontal line is used to indicate the thermal stress limit of synthetic diamond. The intersection of this line with the diamond grade profile will give an estimate of the fractured level of the PCD after roughing. The next vertical lines are the offset lines for 10 lm and 20 lm surface depth as representing the finishing in-feed. It was found that the higher the finishing in-feed, the lower the tensile stress on the surface. The finding thus supports the hypothesis that the finishing in-feed will result in a significant surface stress reduction. Figure 12 illustrates the diamond breakage mechanism. At a temperature of about 1400 K,

Fig. 10 Temperature dependent properties calculated

Fig. 11 Temperature–stress relationship for different PCD



APRIL 2015, Vol. 137 / 021017-7

Fig. 12 Diamond breakage mechanism

bulk material before the next spark is generated. In the case of smaller interpulse delays (off-time), higher sparking frequency happened, which resulted in the lower heat removal by flushing. Thus, the ratio of those parameters (on-time to off-time) considerably affects the level of surface stress (Fig. 14). The unclear trend observed on the CTM302 PCD grain was caused by the inconsistency of the machining process. Combination of grain size between 2 lm to 30 lm for this grade makes the sparking inconsistent. At certain stages, where the distribution of 30 lm grains is high, the erosion process is slower due to the difficulty of big grain erosion. At the same time, the heat was continually supplied towards the deeper PCD bulk and the tensile stress was continually generated due to the rapid heating up and cooling down. Due to this effect, the stochastic distribution of PCD grains size is regarded as a major factor influencing the residual stress value, especially during low energy erosion. The stress value obtained by the developed model was compared with the actual stress value measured by the Raman analysis (Table 5). In this comparison, the actual values were taken only from the surface that was eroded by the smallest current with the lowest pulse on to pulse off ratio. With these parameters, surface

finishing stage. As only small erosion energy is used for finishing, graphitization becomes the main removal mechanism. As mentioned by Zhang et al. [18], by this mechanism, the diamond grains on the top surface would graphitize. The graphite would then dissolve into the molten cobalt before being flushed away by the dielectric. Due to the small energy used in finishing, multiple sequential sparks are required to achieve the graphitisation temperature. Figure 14 shows the D-value obtained by the Raman spectrums. It is clear that, for every sample, a higher diamond peak value was obtained when using a 20 lm finishing in-feed. As expected, with higher finishing in-feed, a higher diamond peak value was obtained which indicates better surface stress (reduction of tensile residual stress). Additionally, the diamond peak location was found to be significantly affected by the machining parameters. The significant difference in diamond peak locations is primarily due to the difference in sparking energy used in the finishing operation. Higher energy caused a higher heat flux to be produced by the plasma, thus increasing the heat penetration depth. As discussed before, the roughing operation produced high residual stress on the surface and the stress relaxation occurred deeper into the material. Similarly to the roughing process, thermal stress was also generated by the finishing process but with a lower value. The smaller sparking energy with smaller pulse duration, and longer sparking interval was found to minimize the cumulative stress generated by the finishing process. Longer interpulse delays allow the heat generated by a single spark to be dissipated into the

Fig. 13 Eroded surface after the finishing operation

Fig. 14 Raman D-value (a) CTX002, (b) CTB010, and (c) CTM302

021017-8 / Vol. 137, APRIL 2015



Table 5 Comparative study between the theoretical values obtained from the physical model and the real value 10 lm finishing in-feed PCD types CTX002 CTB010 CTM302

20 lm finishing in-feed

Theoretical value (GPa)

Actual value (GPa)

Percentage error (%)

Theoretical value (GPa)

Actual value (GPa)

Percentage error (%)

5.0 5.0 5.0

5.4 5.2 5.2

7 4 4

3.8 3.3 2.8

4.7 3.8 2.3

19 13 21

Fig. 15 D/G ratio (a) CTX002, (b) CTB010, and (c) CTM302

stress modification was minimized, leaving only the stress originating from the roughing process. The results show that the model is reasonably accurate for the thermal stress prediction since it is closely matched by the actual value measured except in the case of CTM 302 in which the sparking inconsistency (as discussed) possibly caused the high percentage error obtained (more than 20%). The diamond peak to graphite peak intensity ratio was again examined after the finishing erosion. As shown in Fig. 15, the graphitization degree reduces with higher finishing in-feed (refer to the result variance between low and high finishing in-feed). This is an evidence of a partial graphitization phenomenon caused by the roughing process. The comparison of the graphitization degree of three different PCD types used shows that the lowest graphitization is for the

CTM302, followed by CTB010, and CTX002 (Fig. 16). It could be explained as being due to the thermal properties of the material. The PCD with the higher diamond fraction has better thermal conductivity thus reducing the probability of heat accumulation within the material. In addition, the lower the cobalt catalyst composition, the lesser the diamond–to–cobalt contact area and this consequently reduces graphitization at lower activation energies.

Conclusion Grain fracture was found to be the PCD removal mechanism for roughing with high energy erosion and graphitization was found to be the removal mechanism for the low energy finishing stage. The appearance of the modified layer is due to surface graphitization. Simultaneous graphitization on the surface occurred, resulting in the formation of a “black layer” until some limit at which the graphitization rate reduced significantly. The study found that the high residual stress and surface graphitization could be defined as thermal damage of PCD. The thermal stress generated by the roughing process is the main contributor to the final residual stress on the surface. Higher finishing in-feed resulted in better surface quality due to higher graphitization and stressed structure removal. A model for the residual stress prediction was developed by integrating the physical model with the thermal profile obtained by use of ANSYS software. The results demonstrated that the developed model is reasonably accurate for the thermal residual stress prediction of EDM of PCD.

Nomenclature

Fig. 16 A comparison of graphitization degree of three different PCD

a¼ C¼ cc ¼ cd ¼ E¼ Ed ¼ fc ¼ fd ¼ i¼ K¼ kd ¼

diamond grain radius (lm) specific heat of material (J/kgK) heat capacitance for cobalt (J/kgK) heat capacitance for diamond (J/kgK) modulus of elasticity (N/m2) discharge energy (J) cobalt fraction diamond fraction current (A) thermal conductivity (W/mK) thermal conductivity of diamond (W/mK)



APRIL 2015, Vol. 137 / 021017-9

kc ¼ r¼ rp ¼ t¼ T¼ u¼ U¼ v¼ r ¼ s ¼ a¼ h¼ k¼ l¼ r¼ rt ¼ rt ¼ v¼

thermal conductivity of cobalt (W/mK) position where the stress value is determined (lm) plasma radius (lm) pulse duration (ls) temperature (K) thermal displacement (lm) voltage (V) Poisson’s ratio unstressed Raman value (cm1) Raman shift value (cm1) coefficient of thermal expansion (K1) temperature difference (K) elastic modulus shear modulus tensile residual stress (N/m2) tangential stress (N/m2) radial stress (N/m2) coefficient of stress-induced frequency shift (N/m2)

References [1] Olsen, R. H., Dewes, R. C., and Aspinwall, D. K., 2004, “Machining of Electrically Conductive CVD Diamond Tool Blanks Using EDM,” J. Mater. Process. Technol., 149(1–3), pp. 627–632. [2] Katayama, S., and Hashimura, M., 1997, “Effects of Microcracks in CVD Coating Layers on Cemented Carbide and Cermet Substrates on Residual Stress and Transverse Rupture Strength,” ASME J. Manuf. Sci. Eng., 119(1), pp. 50–54. [3] Bertagnolli, K., and Vale, R., 2000, “Understanding and Controlling Residual Stresses in Thick Polycrystalline Diamond Cutters for Enhanced Durability,” Finer Points, Technical Report. [4] Yahiaoui, M., Gerbaud, L., Paris, J. Y., Denape, J., and Dourfaye, A., 2013, “A Study on PDC Drill Bits Quality,” Wear, 298–299, pp. 32–41. [5] Che, D., Han, P., Guo, P., and Ehmann, K., 2012, “Issues in Polycrystalline Diamond Compact Cutter–Rock Interaction From a Metal Machining Point of View—Part II: Bit Performance and Rock Cutting Mechanics,” ASME J. Manuf. Sci. Eng., 134(6), p. 064002. [6] Paggett, J., Drake, E., Krawitz, A., Winholtz, R., and Griffin, N., 2002, “Residual Stress and Stress Gradients in Polycrystalline Diamond Compacts,” Int. J. Refract. Met. Hard Mater., 20(3), pp. 187–194. [7] Chen, F., Xu, G., Ma, C.-D., and Xu, G.-P., 2010, “Thermal Residual Stress of Polycrystalline Diamond Compacts,” Trans. Nonferrous Met. Soc. China, 20(2), pp. 227–232. [8] Jia, H., Jia, X., Xu, Y., Wan, L., Jie, K., and Ma, H., 2011, “Effects of Initial Crystal Size of Diamond Powder on Surface Residual Stress and Morphology in Polycrystalline Diamond (PCD) Layer,” Sci. China Phys., Mech. Astron., 54(1), pp. 98–101. [9] Krawitz, A., Andrew Winholtz, R., Drake, E., and Griffin, N., 1999, “Residual Stresses in Polycrystalline Diamond Compacts,” Int. J. Refract. Met. Hard Mater., 17(1), pp. 117–122. [10] Frushour, R. H., 1996, “Composite Polycrystalline Compact With Improved Fracture and Delamination Resistance,” Google U.S. Patent No. 5564511A. [11] Garcıa Navas, V., Ferreres, I., Mara~ n on, J. A., Garcia-Rosales, C., and Gil Sevillano, J., 2008, “Electro-Discharge Machining (EDM) Versus Hard Turning and Grinding—Comparison of Residual Stresses and Surface Integrity Generated in AISI O1 Tool Steel,” J. Mater. Process. Technol., 195(1–3), pp. 186–194. [12] Guu, Y. H., and Hocheng, H., 2001, “Improvement of Fatigue Life of Electrical Discharge Machined AISI D2 Tool Steel by TiN Coating,” Mater. Sci. Eng.: A, 318(1–2), pp. 155–162. [13] Yadav, V., Jain, V. K., and Dixit, P. M., 2002, “Thermal Stresses Due to Electrical Discharge Machining,” Int. J. Mach. Tools Manuf., 42(8), pp. 877–888. [14] B€ulent Ekmekci, O. E., Erman Tekkaya, A., and Erden, A., 2005, “Residual Stress State and Hardness Depth in Electric Discharge Machining: De-Ionized Water as Dielectric Liquid,” Mach. Sci. Technol., 9(1), pp. 39–61. [15] Guu, Y. H., Hocheng, H., Chou, C. Y., and Deng, C. S., 2003, “Effect of Electrical Discharge Machining on Surface Characteristics and Machining Damage of AISI D2 Tool Steel,” Mater. Sci. Eng.: A, 358(1–2), pp. 37–43. [16] Lee, H. T., Tai, T. Y., Liu, C., Hsu, F. C., and Hsu, J. M., 2011, “Effect of Material Physical Properties on Residual Stress Measurement by EDM HoleDrilling Method,” ASME J. Eng. Mater. Technol., 133(2), p. 021014. [17] Kozak, J., Rajurkar, K. P., and Wang, S. Z., 1994, “Material Removal in WEDM of PCD Blanks,” ASME J. Manuf. Sci. Eng., 116(3), pp. 363–369. [18] Zhang, Z., Peng, H., and Yan, J., 2013, “Micro-Cutting Characteristics of EDM Fabricated High-Precision Polycrystalline Diamond Tools,” Int. J. Mach. Tools Manuf., 65, pp. 99–106. [19] Che, D., Han, P., Guo, P., and Ehmann, K., 2012, “Issues in Polycrystalline Diamond Compact Cutter–Rock Interaction From a Metal Machining Point of View—Part I: Temperature, Stresses, and Forces,” ASME J. Manuf. Sci. Eng., 134(6), p. 064001.

[20] Cai, M., Li, X., and Rahman, M., 2007, “Study of the Mechanism of Groove Wear of the Diamond Tool in Nanoscale Ductile Mode Cutting of Monocrystalline Silicon,” ASME J. Manuf. Sci. Eng., 129(2), pp. 281–286. [21] Iwai, M., Sano, S., Pan, W., Itagaki, K., Murakami, Y., Wang, M., and Uematsu, T., 2006, “Manufacturing of Micro V-Groove With an Electrically Conductive Diamond Electrode in EDM,” American Society for Precision Engineering 21st Annual Meeting, Monterey, CA. [22] Suzuki, K., Saito, T., Sano, S., Iwai, M., Ninomiya, S., and Uematsu, T., 2007, “Manufacturing of a Porous PCD With Skeleton Structure by EDM,” Key Eng. Mater., 329, pp. 471–476. [23] Chen, Y., Zhang, L. C., and Arsecularatne, J. A., 2007, “Polishing of Polycrystalline Diamond by the Technique of Dynamic Friction. Part 2: Material Removal Mechanism,” Int. J. Mach. Tools Manuf., 47(10), pp. 1615–1624. [24] Bhowmick, S., and Alpas, A., 2013, “The Performance of Diamond-Like Carbon Coated Drills in Thermally Assisted Drilling of Ti-6Al-4V,” ASME J. Manuf. Sci. Eng., 135(6), p. 061019. [25] Kang, Y., Qiu, Y., Lei, Z., and Hu, M., 2005, “An Application of Raman Spectroscopy on the Measurement of Residual Stress in Porous Silicon,” Opt. Lasers Eng., 43(8), pp. 847–855. [26] Catledge, S. A., Vohra, Y. K., Ladi, R., and Rai, G., 1996, “Micro-Raman Stress Investigations and X-Ray Diffraction Analysis of Polycrystalline Diamond (PCD) Tools,” Diamond Relat. Mater., 5(10), pp. 1159–1165. [27] Andreev, V., 1999, “Spontaneous Graphitization and Thermal Disintegration of Diamond at T > 2000 K,” Phys. Solid State, 41(4), pp. 627–632. [28] Prawer, S., Nugent, K., Jamieson, D., Orwa, J., Bursill, L. A., and Peng, J., 2000, “The Raman spectrum of Nanocrystalline Diamond,” Chem. Phys. Lett., 332(1), pp. 93–97. € [29] Ali, M., Urgen, M., and Atta, M., 2012, “Effect of Surface Treatment on HotFilament Chemical Vapour Deposition Grown Diamond Films,” J. Phys. D: Appl. Phys., 45(4), p. 045301. [30] Lewis, J. S., Gittard, S. D., Narayan, R. J., Berry, C. J., Brigmon, R. L., Ramamurti, R., and Singh, R. N., 2010, “Assessment of Microbial Biofilm Growth on Nanocrystalline Diamond in a Continuous Perfusion Environment,” ASME J. Manuf. Sci. Eng., 132(3), p. 030919. [31] Ramanathan, D., and Molian, P. A., 2002, “Micro-and Sub-Micromachining of Type IIa Single Crystal Diamond Using a Ti: Sapphire Femtosecond Laser,” ASME J. Manuf. Sci. Eng., 124(2), pp. 389–396. [32] Hitoshi Sumiya, T. I., 2008, “Microstructure and Mechanical Properties of HighHardness Nano-Polycrystalline Diamonds,” SEI Tech. Rev., 66, pp. 85–91. [33] Mlungwane, K., Herrmann, M., and Sigalas, I., 2008, “The Low-Pressure Infiltration of Diamond by Silicon to Form Diamond–Silicon Carbide Composites,” J. Eur. Ceram. Soc., 28(1), pp. 321–326. [34] Li, L., Zhu, Z., Yan, Z., Lu, G., and Rintoul, L., 2007, “Catalytic Ammonia Decomposition Over Ru/Carbon Catalysts: The Importance of the Structure of Carbon Support,” Appl. Catal. A, 320, pp. 166–172. [35] Kalyanasundaram, D., Schmidt, A., Molian, P., and Shrotriya, P., 2014, “Hybrid CO2 Laser/Waterjet Machining of Polycrystalline Diamond Substrate: Material Separation Through Transformation Induced Controlled Fracture,” ASME J. Manuf. Sci. Eng., 136(4), p. 041001. [36] Orwa, J. O., Nugent, K. W., Jamieson, D. N., and Prawer, S., 2000, “Raman Investigation of Damage Caused by Deep Ion Implantation in Diamond,” Phys. Rev. B, 62(9), pp. 5461–5472. [37] Johnston, C., Crossley, A., Chalker, P., Buckley-Golder, I., and Kobashi, K., 1992, “High Temperature Raman Studies of Diamond Thin Films,” Diamond Relat. Mater., 1(5), pp. 450–456. [38] Rahim, M. Z., Pourmoslemi, A., Ding, S. L., and Mo, J., 2013, “Residual Stress Analysis of Polycrystalline Diamond After Electrical Discharge Machining,” Adv. Mater. Res., 820, pp. 106–109. [39] Rahim, M. Z., Ding, S., Hu, B., and Mo, J., 2014, “Crater Size Prediction in Electrical Discharge Grinding (EDG) of Polycrystalline Diamond (PCD),” Nonconv. Technol. Rev., 18, pp. 92–97. [40] Thewlis, J., and Davey, A., 1956, “XL. Thermal Expansion of Diamond,” Philos. Mag., 1(5), pp. 409–414. [41] Li, Z., Jia, H., Ma, H., Guo, W., Liu, X., Huang, G., Li, R., and Jia, X., 2012, “FEM Analysis on the Effect of Cobalt Content on Thermal Residual Stress in Polycrystalline Diamond Compact (PDC),” Sci. China Phys., Mech. Astron., 55(4), pp. 639–643. [42] Shin, T. J., Oh, J. O., Hwan Oh, K., and Lee, D. N., 2004, “The Mechanism of Abnormal Grain Growth in Polycrystalline Diamond During High PressureHigh Temperature Sintering,” Diamond Relat. Mater., 13(3), pp. 488–494. [43] Uehara, K., and Yamaya, S., 1990, “High Pressure Sintering of Diamond by Cobalt Infiltration,” Science and Technology of New Diamond, S. Satio, O. Fukunaga, and M. Yoshikawa, eds., KTK Scientific Publishers/Terra Scientific Publishing Company (TERRAPUB), Tokyo, Japan, pp. 203–209. [44] Evans, T., Davey, S., and Robertson, S., 1984, “Photoluminescence Studies of Sintered Diamond Compacts,” J. Mater. Sci., 19(7), pp. 2405–2414. [45] Han, F., Jiang, J., and Yu, D., 2007, “Influence of Discharge Current on Machined Surfaces by Thermo-Analysis in Finish Cut of WEDM,” Int. J. Mach. Tools Manuf., 47(7–8), pp. 1187–1196. [46] Tao, J., Ni, J., and Shih, A. J., 2012, “Modeling of the Anode Crater Formation in Electrical Discharge Machining,” ASME J. Manuf. Sci. Eng., 134(1), p. 011002. [47] DiBitonto, D. D., Eubank, P. T., Patel, M. R., and Barrufet, M. A., 1989, “Theoretical Models of the Electrical Discharge Machining Process. I. A Simple Cathode Erosion Model,” J. Appl. Phys., 66(9), pp. 4095–4103. [48] Patel, M. R., Barrufet, M. A., Eubank, P. T., and DiBitonto, D. D., 1989, “Theoretical Models of the Electrical Discharge Machining Process. II. The Anode Erosion Model,” J. Appl. Phys., 66(9), pp. 4104–4111.

021017-10 / Vol. 137, APRIL 2015



[49] Shankar, P., Jain, V., and Sundararajan, T., 1997, “Analysis of Spark Profiles During EDM Process,” Mach. Sci. Technol., 1(2), pp. 195–217. [50] Marafona, J., and Chousal, J., 2006, “A Finite Element Model of EDM Based on the Joule Effect,” Int. J. Mach. Tools Manuf., 46(6), pp. 595–602. [51] Kidalov, S. V., and Shakhov, F. M., 2009, “Thermal Conductivity of Diamond Composites,” Materials, 2(4), pp. 2467–2495.

[52] Ferreira, N. G., Abramof, E., Corat, E. J., and Trava-Airoldi, V. J., 2003, “Residual Stresses and Crystalline Quality of Heavily Boron-Doped Diamond Films Analysed by Micro-Raman Spectroscopy and X-Ray Diffraction,” Carbon, 41(6), pp. 1301–1308. [53] Weidner, D. J., Wang, Y., and Vaughan, M. T., 1994, “Strength of Diamond,” Science, 266(5184), pp. 419–422.



APRIL 2015, Vol. 137 / 021017-11