Comparative Analysis of Different Graphite Concentrations in Micro-PMEDM Drilling

Abstract

Micro-electrical discharge machining is valuable in industry thanks to its ability to realise precise micro-holes with high aspect ratios. However, a limitation of the technology is represented by its low material removal rate compared to other material removal technologies. Therefore, different strategies are under investigation to make the process faster. One of these strategies consists of adding powders into the dielectric. This process is called powder-mixed electrical discharge machining (PMEDM). This paper focusses on the optimisation of different aspects of this process, particularly the effects of the powder concentration, the presence of the surfactant, the stirring of the dielectric during the machining and the stability in time of the dielectric in micro-drilling. Graphite was used as powder in pure water, and in some tests a dispersant was also added. The concentration of the powder was varied, maintaining the same ratio between the graphite and the surfactant. The optimal graphite concentration was also used without the dispersant but with a changed parameter for the stirring system. The powder-mixed dielectrics showed better removal performance than pure water, and the best graphite concentration was the highest. The material removal rate increased by 40–150% compared to pure water. The tests made without dispersant showed that its presence did not improve the machining rate, while the stirring system deeply affected the process. The electrode wear benefitted from the reduction in machining time, and when the dispersant was used, electrode wear was lowered up to 50% compared to pure water. The trend of the electrode law of motion was affected by the concentration of the contaminant (debris from the erosion and powder). The geometrical characteristics were also affected by the presence of the powder, which changed the spark length. With the highest graphite concentration, radial overcut increased up to 50% compared to pure water. The stability in time of the dielectric when the powder was added was also evaluated and it was found that an efficient stirring system without the use of dispersant is a good solution, able to limit the possible sedimentation and aggregation of the powder.

1. Introduction

EDM (electrical discharge machining) technology is classified in unconventional thermal processing as the removal of material that takes place using a series of successive electric discharges, which occur at a high frequency between the workpiece and the tool; the temperature within the discharge channel rises to a range of 8000–12,000 °C, leading to the melting or vaporisation of the material. A dielectric fluid between the two electrodes isolates the tool and workpiece until the applied potential difference exceeds the breaking voltage of the dielectric itself. Moreover, the dielectric keeps the spark channel’s width shorter and helps remove debris from the machining zone. One of the limitations of the technology is that it machines only electrically conductive materials, regardless of their mechanical properties. Indeed, it represents a valid solution to machine workpiece materials with a high hardness [1]. However, it has a low material removal rate (MRR), high power consumption and hazardous releases as a function of the dielectric used.

An important application consists of micro-EDM, particularly micro-EDM drilling, which is widely used in the automotive, aerospace and medical sectors to realise micro-holes with diameters from tens to hundreds of micrometres with a high aspect ratio [2].

To increase the performance of the EDM process, a strategy studied by the researchers consists of adding particles into the dielectric fluid. In this case, the process is called PMEDM (powder-mixed EDM). It was observed that the impurities from the erosion of the workpiece, the electrode wear and the addition of powder improved the erosion rate, reducing the breakdown voltage of the dielectric [3]. The powder added into the dielectric affects all the phases taking place in the development of a single electrical discharge. During the breakdown phase, the particles, under the action of electrical forces, tend to gather around surface irregularities because of the electric field concentration around such points. Therefore, higher particle concentrations in these regions are expected. It may be assumed that the liquid dielectric breakdown occurs at a critical particle concentration. The time required to reach such a breakdown decreases as the particle concentration increases. The erosion phase is influenced by the powders in the dielectric as well, considering that the machining rate increases.

When a potential difference between the two electrodes is applied, the displacement of electrical charge takes place in the powder, resulting in a powder bridge formation between the tool and the workpiece [4]. The first discharge occurs where the electric field density is greatest, along the particle bridge. This discharge causes the bridge to break, with consequent dispersion of the particles in the gap between the workpiece and the tool and a redistribution of charges that leads to fragmentation of the spark into a series of discharges, each having lower energy [5]. The discharges between the powder particles and the workpiece cause a lower crater depth on the workpiece than traditional EDM [6]. Of course, the properties of the powder play an important role, such as the shape, dimension, electrical conductivity and concentration of the dielectric. In micro-machining, the size of the powder is crucial, considering that the gap is only some microns, and, therefore, the most suitable powders should be of nanometric size [6].

There is a wide range of powders used for PMEDM, which can be classified into conducting, semi-conducting and non-conducting [7]. The choice depends on several factors, such as the electrical parameters, the workpiece material and the size of the application (macro or micro). Among these powders, the most widely used are graphite, aluminium, silicon and silicon carbide; others less commonly used are titanium, copper, chromium, tungsten, manganese and other compounds [8]. In general, choices must be made about the chemical (type) and physical (particle size, concentration) properties of the powder, together with the systems able to provide suspension stability (i.e., the use of surfactants and mechanical stirring).

Most papers available in the literature deal with PMEDM using a hydrocarbon dielectric, while only around 12% use water as a dielectric [8]. The advantages of hydrocarbon oils include easier filtering and a non-corrosive action on the workpiece, but the waste products are toxic [9]. The thermal decomposition causes the formation of substances such as benzene and benzopyrene in the air, which are considered carcinogenic. A healthier working environment can be achieved by using water as a dielectric [10]. A comparison between water and hydrocarbon oil in micro-drilling [11] and in slicing by a foil electrode [12] was made, and it was found that water guarantees higher MRR to the detriment of some quality aspects. This result was confirmed in [13], where it was also found that by using water, the relative tool wear is lower, while the roughness gets worse. This is only partially in agreement with [14], in which the authors suggest that when high pulse energy is used, machining in distilled water results in lower tool wear and a higher MRR than when using kerosene oil. Water penalises machining accuracy, but the surface finish is better. The nature of the dielectric also affects the formation of different chemical compounds and the presence of micro-cracks: by using kerosene, carbides and a low number of micro-cracks are on the surface, while by using water, oxides and a high number of micro-cracks are on the surface [15].

The first experimentation of PMEDM on water was made in 1965 [16]. Graphite was added to water, permitting an increase in the MRR. The graphite concentration affected the results, and the optimal concentration was 0.3 g/L. The same results were obtained in [17], in which titanium alloy was machined, showing an improvement in performance and finishing surface compared to the traditional EDM process. The optimal concentration of graphite was 2–2.5 g/L. It was also found that graphite forms a protective layer on the surface, avoiding the solidification of melted drops on the surface of the workpiece. A different result is shown in [18], in which Inconel X750 was machined by comparing two powders in the water. The authors found that the powders caused a decrease in the MRR and an increase in the tool wear rate but a better finishing surface. The improvement in the finishing surface is confirmed by [19], in which the crater dimensions were estimated using a FEM (finite element method) simulation approach. The results, validated by experimental tests, showed that the craters had high diameters and shallow depths, resulting in improved MRR and a better surface finish.

The powder concentration in the dielectric is a critical factor, as already seen in [15]. In general, there is an optimal value: for a lower concentration than the optimal one, the presence of the powder is so scarce that its contribution to the machining is negligible; for a higher concentration than the optimal one, the machining slows down due to the relevant number of short circuits caused by the bridge of conductive powders between the tool and the workpiece [20,21,22].

A critical issue in PMEDM is represented by the stirring of the powder into the dielectric liquid and the use of a surfactant. Only a few papers provide information about maintaining the powder suspension to avoid the deposition of the powder and its agglomeration. In [17], the authors underline that the suspension was mixed before the start of the machining to avoid the formation of agglomerates having a large size. During the machining, the dielectric was instead not mixed. A magnetic stirrer was used in [23] to mix the powder, and a surfactant was also added to the dielectric to minimise the risk of graphite agglomeration. Two surfactants were considered, sodium dodecyl sulfate and hexadecyltrimethylammonium bromide, at different concentrations. The addition of a surfactant always caused a lower MRR than using only a dielectric and graphite, while in terms of the roughness of the surface, the lowest value was obtained with 0.1 g of hexadecyltrimethylammonium bromide. In [24], Span 20 was added as a surfactant to graphite powder mixed with an oil dielectric for machining stainless steel, obtaining a higher MRR and surface finish. Surfactants make the distribution of the powder uniform and increase the dielectric conductivity. Consequently, the discharge time also increases. Moreover, Span 20 mixed in oil EDM using a stirring system reduced the agglomeration of both debris and powder particles [25,26]. From a sustainability perspective, a surfactant is more environmentally friendly than pure kerosene [27]. Tests were performed by adding only surfactant without powder into kerosene [28]. The performance improved, and, in the author’s opinion, it was due to the change in the physical properties of the dielectric. In particular, the dielectric density increased, causing a reduction of the plasma channel and, consequently, an increase in the discharge energy. Not only did the density change but the electrical conductivity also increased, permitting the formation of more discharges. The geometrical and surface characteristics also improved after using additives and kerosene compared to pure kerosene.

The effects of the powders in the dielectric on process performance are still poorly understood. The results available in the literature are not in agreement with each other. Many studies concerning specific cases and general results have not yet been conducted. This is particularly true in the case of micro-machining by EDM, for which research is limited.

This paper is focussed on PMEDM technology in micro-drilling. The role of the powder, its concentration, the surfactant and the stirring action were investigated. Titanium alloy was used for the experimental tests, adding graphite powder into pure water. The concentration of the powder was varied, maintaining the same ratio between graphite and dispersant. The optimal graphite concentration was also used without dispersant but while changing a parameter of the stirring system. The performance was evaluated by considering typical indicators such as MRR, TWR (tool wear ratio) and ROC (radial overcut). In addition, during the execution of the holes, the process was monitored through a multi-step strategy to obtain the law of motion of the electrode during erosion.

2. Materials and Methods

Micro-holes on titanium sheets (Ti6Al4V) with a thickness of 1 mm were created. A cylinder tungsten carbide electrode with a diameter of 0.1 mm was used. As a dielectric, the reference case was demineralised water. Two different conditions were tested: the first consisted of adding both graphite powder and a dispersant; in the second, only graphite with the highest graphite concentration was added without the dispersant (C100_ND). The graphite powder had a nanometric dimension (10–30 nm). As dispersant, Poloxamer 188 (commercial name, Sigma Aldrich, Milan, Italy) was used. The concentrations of both the graphite and the dispersant varied, but the ratio between them was kept constant. Figure 1 shows the experimental plan. Each experimental condition was tested fifty times. The high number of repetitions was selected to evaluate the time stability of the dielectric. The concentrations of graphite and dispersant used as reference (C100) were 2.6 g/L and 1.37 g/L, respectively. In condition C50, the concentrations were half that of condition C100, and so on.

The micro-holes were created using a Sarix SX-200 (Sarix, Sant’Antonino, Switzerland), a micro-EDM system. The available device was equipped with hydrocarbon oil as a dielectric in a closed circuit. Considering the different nature of the dielectric tested, it was necessary to design a small tank containing the water where the workpiece was fixed to avoid the contamination of the traditional dielectric used by the machine with water. The stirring effect was realised using a pipette connected to a small pipe to the tank. A piston, controlled by a solenoid valve, crushed the pipette with a programmable frequency. This system (Figure 2) helped to move the dielectric into the small tank, minimising the deposition of the powder on the bottom of the tank. The stirring cycle time was set to 22 s. In the case of C100_ND, a stirring cycle time of 7 s was also tested and labelled C100_ND++.

The process parameters used during the experimental campaign were as follows: peak current 100, voltage 140 V, polarity negative, pulse on time 5 µs, frequency 140 kHz, energy 103, gap 75, gain 40, regulation 02-00. Some of these parameters are typical of the used system. Through-holes were machined by setting an electrode Z displacement of 1.8 mm and an electrode length outside the clamping unit of 3 mm.

Methods with single and multiple steps were used to execute the drilling operation. In the single-step method, the machining was made with a single descent of the electrode into the workpiece during the creation of each hole, so that only the total machining time and the final frontal electrode wear could be recorded. In the multiple-step method, the motion was divided into several steps having a descent stroke of 0.1 mm. In this case, both machining time and electrode wear were recorded at the end of each step. Using these data, the law of motion of the electrode during the drilling operation was able to be plotted, permitting us to achieve more information than in the single-step method. The multiple-step method required a longer execution time, but it affected the dielectric motion and, therefore, potentially altered the discharge phenomena. To evaluate such an effect, each experimental condition was tested by mixing holes machined using either method. For the experimental plan presented in Figure 1, single and multiple steps were performed differently only when low concentrations of graphite and dispersant were used, while pure water (WATER) and the highest graphite concentrations (C100 and C100_ND) did not show significant changes.

3. Results and Discussion

3.1. Effects of the Graphite Concentration on the Dielectric

Machining time, electrode wear, MRR, TWR and ROC were used as indexes to characterise process performance. Top and bottom diameters were measured by an optical microscope, and the volumes of the holes were estimated by the formula of a frustum of cone. ROC was calculated as the difference between the top radius of the hole and the electrode radius. Moreover, a comparison of the laws of motion of the electrode along the Z-axis in the different tested conditions was made.

Figure 3 and Figure 4 report machining time and electrode wear, respectively. By using graphite in the water, machining times were reduced by about 40–55% compared to the pure water case. The powder accelerated the machining thanks to the dielectric breakdown that occurred earlier. Moreover, the graphite concentration affected the machining time: as the graphite concentration increased, the machining took place more quickly. Even a low graphite concentration helped the process, but the benefits were increasingly evident with the highest powder concentrations.

Short machining times could also be observed without dispersant, especially when improved stirring was adopted. In this last case, not only were the shortest machining times achieved (probably because a more homogeneous dispersion allowed more graphite to be active in the process), but also the lowest data scatter was observed. Such results show that mechanical techniques may be an effective alternative to chemical solutions.

Another advantage of using powder regards the lower electrode wear compared to the water dielectric. The concentration of graphite also affected the wear: when the graphite increased, the wear decreased. This lower electrode wear was probably due to two phenomena. First, any reduction of machining time yields lower electrode wear because a lower number of sparks are involved. Second, the dispersant acts as a protective medium. Indeed, when dispersant-free mixtures were used (C100_ND and C100_ND++), the wear was higher than for C100.

During the drilling process, the electrode is subjected to two types of wear, frontal or axial wear and radial wear. In this paper, only frontal wear was considered, and it was measured by the difference between the starting length and the final length of the electrode. A typical shape of the electrode is shown in Figure 5.

Figure 6 shows the laws of motion of the electrode for the cases of water and C100. On the Y-axis, Za is the actual hole depth calculated and the difference between the programmed Z-axis position and the electrode wear. As can be seen, all repetitions of both experimental conditions were close to each other but there was a clear difference between the two cases. When using water, the initial steps were particularly slow. This behaviour can be explained by assuming that there is an optimal debris (or contaminant) concentration for machining. As reported in [3], the time required to reach the breakdown is lower when powders or debris are available in the dielectric and the frequency of discharges increases. As the electrode descends into the workpiece, debris evacuation from the working area becomes more and more difficult, causing an increase in the local contaminant concentration. When the optimal concentration is exceeded, the process slows again because of a lack of stability.

When graphite is added, it acts as a contaminant, and the process starts with good speed. The effect of the graphite concentration is shown in Figure 7, reporting only one law of motion for each graphite concentration, to improve clarity. With increasing graphite, the descent of the electrode was faster. However, the trends of the curves were similar to each other.

The trend of electrode wear during drilling for each experimental condition is reported in Figure 8. It can be noted that pure water has a different slope of the curve at the beginning of the drilling process, probably due to the higher machining time compared to the powder cases. For all the cases, the electrode wear for a step decreased as the actual hole depth increased. Considering that water was used as a dielectric, it is likely that some oxides covered the electrode surface and acted as a protective layer. The tests made without dispersant showed the same behaviour as the others. The stirring frequency did not alter either the values or the trends of the curves.

Figure 9 shows the ROC for all the tested conditions. Increasing the graphite concentration increased the ROC. Pure water displayed a much lower ROC than the other cases. Since ROC is mainly affected by spark length, such a trend agrees with assumptions about the spark bridging effect. In this case, bridging was more evident at higher graphite concentrations. Data scatter should be discussed, using powder, and the standard deviation of ROC became a few times larger. This could be a problem in industrial applications, and it is worth further investigation.

The dispersant did not show significant effects, while the stirring frequency influenced the ROC. Stirring more efficiently moved the powder into the dielectric, so more powder was available in the machining zone and the bridging effect was enlarged.

Figure 10 and Figure 11 report MRR and TWR. These indexes are related to machining time and electrode wear. The same considerations made when discussing Figure 3 and Figure 4 can be made here. However, MRR and TWR also consider the actual sizes of the holes. As already seen for ROC, the diameters showed some degree of scatter. This is why the standard deviations of both MRR and TWR were higher than the machining time and electrode wear.

By using graphite, MRR increased by 40–150% with respect to pure water. In the last tested configuration, double frequency stirring without dispersant, the MRR was more than three times that in the condition using only water. The presence of graphite made the process less stable (the standard deviations were higher) but generally more efficient. The increased efficiency of the process was due to both the larger geometric characteristics of the holes and the reduction of machining time. In fact, from Figure 8, the holes obtained using the graphite showed a larger top diameter due to an increase in spark length compared to the reference cases.

The TWR values reported in Figure 11 follow the electrode wear trend. Additionally, in this case, taking into account the actual hole size, the TWR for pure water was significantly higher than that for the powder cases. Improved stirring is the best solution.

3.2. Dielectric Stability

A common problem when PMEDM is used is related to changes in the dielectric properties over time. This is especially true when no stirring system is adopted due to the sedimentation of the powder. Figure 12 reports the law of motion of the electrode during the execution of five micro-holes, both with and without a stirring system when the dielectric C100 was used. The execution of the holes without stirring degraded as early as the second hole and became even worse by the third hole (blue curves in Figure 12). The machining time ranged between 18 s (the first hole) and 80 s (more than four times greater for the fifth hole). The stirring system instead was able to limit the sedimentation of the powder into the tank. In fact, in this case, the machining time range was very limited (see the orange curves in Figure 12).

In addition to sedimentation, the stirring system could also help prevent the possible aggregation of the powder and the formation of large conglomerates that cause problems during machining. Considering the dimension of the gap, these conglomerates are difficult to evacuate and increase the risk of frequent short circuits that penalise the continuation of the machining.

Another aspect that should be considered is the removed material from both the workpiece and the electrode that remains in the dielectric tank and that could modify some properties of the dielectric. Dielectric contamination due to debris formation is proportional to the removed volume, that is, the number of holes. This is the main reason why the number of repetitions for each experimental condition was as large as fifty. In this way, it was also possible to evaluate the trend of machining times as a function of the number of machined holes.

In all cases, a linear model can satisfactorily fit the machining times (T, in sec) as a function of the hole number (J) according to the following equation:

T = m∙J + q

(1)

Table 1. Regression parameters of machining times for different dielectrics.

Dielectric	m	q
Water	0.0173	40.422
C12.5	0.1998	19.716
C25	0.1675	21.422
C50	0.1547	19.179
C100	0.0701	18.338
C100_ND	0.0886	17.341
C100_ND++	0.0122	17.398

reports the regression coefficients evaluated for all experimental conditions mentioned in Figure 1. Not only was the slope m always positive (confirming that the machining performance gets worse with time), but the slope monotonically depended on graphite concentration as well. In particular, lower graphite concentrations displayed a faster increase in machining time. Conversely, pure water showed almost negligible changes in machining times, signifying that the effect of dispersed debris was not significant. This result agrees with the literature [8], which states that dielectric stability is especially critical when dealing with PMEDM. Powder-mixed dielectrics may be less stable than pure water because of either mechanical sedimentation over time (which can only be hindered by stirring) or the effect of metal debris, acting as nuclei for graphite aggregation.

The trend of electrode wear as a function of the number of machined holes was always linear with a small positive slope. This behaviour is consistent with the results shown in Figure 4, in which the data indicate a low standard deviation. Moreover, electrode wear was negatively correlated with graphite concentration. Assuming that during the tests aggregation and sedimentation phenomena reduced the graphite concentration, the small increase in wear with the number of the hole in the sequence is justified.

A similar conclusion can be reached by analysing the ROC as a function of the hole number. In this case, ROC decreases with the hole number, and this behaviour can be justified with same considerations made for electrode wear.

The dispersant effect can also be seen in Figure 13, which reports the machining times of certain tests. The experiments with (C100) and without (C100_ND) dispersant displayed very similar behaviour (as well as having close values of slope m); the effect of the dispersant, in this case, was small. When increasing the stirring frequency (C100_ND++), the dielectric properties were as stable as when using water, and consequently, so were the machining times.

The stability of the powder dielectric was also verified after several uses. The same dielectric C100 was used more times: new (first day), first reuse (after 1 day) and second reuse (after 7 days). In all cases, before putting the dielectric into the small tank, it was stirred in an ultrasonic system for 10 min. For the first reuse, the average machining time increased by around 10%, and the process became more unstable. The performance became significantly worse for the second reuse, with an increase in machining time of 50% and a further deterioration of the stability of the dielectric (Figure 14). This was probably due to the level of contamination: the high amount of debris made graphite aggregation more probable. In this way, less graphite was available for the process.

4. Conclusions

The addition of graphite to a water dielectric improved process performance. Machining time and electrode wear were reduced, whereas overcut significantly increased since the bridging action of particles led to a higher spark length. Because of the random nature of particle bridges, the process was affected by a larger scatter of all the collected data.

Graphite concentration is a parameter affecting process performance. By using graphite in water, machining times were reduced by about 40–55%. By increasing the graphite concentration, all the effects became increasingly stronger. The best results occurred with the highest powder concentration, a value of 3 g/L. However, even a small amount of powder triggered significant changes in the drilling evolution over time and led to different patterns of the electrode law of motion.

The graphite had positive effects on electrode wear. The amount of wear was reduced by up to 50% compared to using pure water. The comparison between the dielectric with and without dispersant at the same graphite concentration showed that the dispersant acted as a protective medium on the electrode, and electrode wear was slightly lower. ROC was mainly affected by spark length, and it increased as the graphite concentration increased. With the highest graphite concentration, ROC increased up to 50% compared to pure water. MRR increased by 40–150% compared to pure water. In the configuration with double frequency stirring without dispersant, the MRR was more than three times greater compared to using only water. The increased effectiveness of the process by using graphite was also confirmed by the TWR, which was reduced by about 50% at the highest graphite concentration.

The ability to preserve powder-mixed dielectric properties is a key factor in PMEDM because the conventional technique, based on filtering, is no longer available in this case. In fact, there is an increasing trend of machining times becoming longer as more and more material is removed. Lower graphite concentrations display a faster reduction in machining performance. This effect can be related to the interaction between graphite and debris, activating aggregation and precipitation phenomena among the solid components of the dielectric mix. The net effect of debris is to reduce the amount of effective (‘active’) graphite available for the process, especially notable for low graphite concentrations. When high graphite concentration is considered, the rich amount of powder compensates for the action of the debris, making the process increasingly stable, as pure water does. An excessive increase in the amount of debris in the dielectric reduces the performance of all powder-mixed dielectrics.

The stirring system deeply affects the process and is essential to make the process more stable in time, improve performance and reduce data scatter. Within the limits of this experiment, the configuration obtained by increasing the frequency of stirring without a dispersant permitted the achievement of the lowest machining time, confirming the key role of the process of the stirring action.

From a sustainability perspective, graphite added to water as a dielectric with an efficient stirring system represents a good solution to not only minimise environmental impact but also increase productivity.

A limitation of the proposed solution in industrial applications is represented by the management of the dielectric with added graphite. The graphite would be retained by the filter and should be added continuously. Further work will be focussed on finding possible solutions to facilitate the industrial implementation of this technique.