Experimental analysis of microhardness changes of subsurface areas affected by WEDM

The influence of the surface area by the impact of high temperatures after wire electric discharge machining (WEDM) is a known fact. However, the affected parameters also include a change in microhardness. In order to further investigate this statement, 5 different ferrous and non-ferrous materials were selected, from which three samples were always made with different settings of machine parameters (gap voltage, pulse on and off time, wire feed and discharge current). This examined not only the effect of the machining itself on the material but also whether the change in the microhardness of the material is affected by the setting of the machine parameters. In order to measure the microhardness of the subsurface layer, a metallographic preparation was made from each sample, which enabled accurate measurements always in the same area. Subsequent evaluation revealed that the microhardness may not be affected at all and everything depends only on the type of material being machined. The changes in microhardness affected by setting machine parameters are negligible.


Introduction
Wire electric discharge machining (WEDM) is a widely used unconventional machining technology that uses the thermoelectric principle to cut material. Between the two electrodes, which represent the material being machined and the wire (tool), periodically repetitive electrical pulses take place in the presence of a dielectric fluid, which systematically erode and evaporate the material. The machining takes place without the impact of conventional cutting forces, so it is possible to machine even very soft materials or thin-walled components. [1][2][3] With WEDM, all at least minimally electrically conductive materials can be machined, regardless of their mechanical or physical properties. For this reason, the technology is suitable for machining classically difficult to machine materials or materials that have already been heat-treated. WEDM is an indispensable technological operation in many industries, such as the aerospace, automotive, food, military, medical and sports industries. 4,5 Straka 6 in their paper focused on the direct influence of microhardened surface layer on the final surface quality of the machined tool steel EN X210Cr12. The hardness measurements on metallographic samples in heat affected zone were carried out using the Vickers Hardness Tester, however, for the hardness measurements of the basic material the Vickers microhardness test could not be applied due to the high values of the hardness of the evaluated samples. During the experiments, it was proved that the machining method has a minimal influence on the microhardness variations in the surface layer. But contrarily, there was a significant influence observed in the white layer zone, where a substantial growth of the microhardness and its subsequent drop was detected. Newton et al. 7 focused on hardness testing in their experiments in order to measure the hardness of the recast layer during the machining process. Within the experimental testing, the bulk hardness was indicated, with an increasing linear trend on each sample. This might suggest that the WEDM process softens the machined surface below the hardness of the bulk material. While studying the hardness of the recast layer, it was determined that the hardness values are lower than those of the bulk material. This proves the fact, that the decrease in recast layer hardness due to the WEDM process could be caused by the metallurgical changes uncovered in the EPMA. Moreover, this decrease might be also caused by the electrolysis happening in the de-ionized water dielectric, or also by residual stress and the dramatic rates of heating and cooling of the recast layer. Zakaria et al., 8 while studying the effect of the WEDM cutting parameters, also focused on the hardness of the hybrid metal material. They tried to develop the optimization of machining parameters to obtain high surface quality and maximum hardness. It was found out that the significant factors for surface hardness are pulse on and off time and peak current. The highest level of hardness was obtained with the lowest level of pulse off time, and the lowest hardnessusing the highest level of pulse off time. It was also proved that the decreasing of the surface hardness takes place due to the electrolytic erosion during the WEDM process. Li et al. 9 studied the WEDM characteristics of Inconel 718 and focused on the white layers by trim cut and their microhardness. The measurements of the microhardness were taken at different discharge energy, corresponding to the main cut and three trim cuts. During the experiments, it was found that the white layer has a significantly reduced hardness compared to the bulk material, and this dramatic reduction in hardness suggests that significant thermal degradation occurs on EDMed surfaces regardless of discharge energy. However, the white layer of the EDMed carbon steel has higher hardness than the bulk material, which is caused by the rich C element, whereas the low C content of IN 718 would not make the white layer hard after quenching in the dielectric. Lin et al. 10 studied the hardness of the WEDM surface of Fe-30Mn-6Si and Fe-30MN-6Si-5Cr shape memory alloys in their paper, which was measured by a microvickers tester. The average hardness value was calculated from at least five test readings for each sample. It was concluded, that the effect of hardness arises from the formation of the Cu 0.83 Si 0.17 compound, η'-(Cu, Si) and (Fe, Cr) solid solutions in the recast layer. It was also proved, that the hardness of the matrix is not affected by the WEDM process. Klocke et al. 11 in their studies about the structure and composition of the white layer in the WEDM process focused also on the microhardness testing according to the ISO-standard 4516 requirements. In order to be able to quantify the deviation in hardness and modulus of elasticity a sample with the same geometry but without the white layer, using grinding and polishing during the machining process. The experiments revealed that the alloyed molybdenum and copper increase the hardness of the steel. Moreover, all the presented chemical alterations of the white layer increase the hardness of the material, which could be further enhanced through the possible deposition of carbon from the dielectric. As the last step of the experiments, the local hardness in the white layer and in the area below this layer was quantified. It was found, that the hardness of the recast layer was significantly increased, and it was nearly doubled in the white layer compared to the parent material. Aspinwall et al. 12 researched the workpiece surface roughness and integrity after WEDM and focused on the microhardness variation and depth profile data as wellin the experiments. The experiments showed that there was no significant change in workpiecemicrohardness variation with cracking confined to the recast layer. Sharma et al. 13 while evaluating the WEDM performance characteristics of Inconel 706, studied the microhardness of the crosssection of WEDMed surface employing the micro Vickers hardness tester, and it was measured at 10 kgf load with the dwell time of 10 s the low and high settings of servo voltage and pulse on time were also employed for the measurements. According to the experiments, the subsurface microhardness was decreased due to quenching of dielectric fluid and due to multiple thermal loading during the WEDM process. The surface was also softened with the help of the low carbon content of Inconel 706. According to the EDAX analysis of the recast surface, the presence of Cu and Zn caused the lowered hardness of the recast region.
Abhilash and Chakradhar 14 investigated the effect of machining stability on the surface integrity of the Inconel 718 superalloy which was machined with the help of the WEDM process. Hard zinc coated brass was used as the material for the machining. The experiments were held on different levels of machining stabilities, the surface defects were detected and analysed with the help of scanning electron microscope images, and the recast layer thickness was analysed by observing the polished cross-sectional view of the machined samples. Abhilash and Chakradhar 15 focused on the WEDM failure prediction and the process control which was based on the sensor fusion and pulse train analysis. The artificial neural network classifier was employed for the analysis, and the extracted features are discharge energy, spark frequency, open spark ratio and short circuit ratio. The output classes were presented by normal machining, wire breakage, and spark absence. Abhilash and Chakradhar 16 in the studies correlated the mean gap voltage variation and wire breakage occurrences while the WEDM process of the Inconel 718 superalloy. The wire breakage was tried to be predicted by considering the mean gap voltage variation. It was found out that if the value of voltage variation crosses a threshold limit, and the process interruptions through wire breakages were observed. Abhilash and Chakradhar 17 tried to improve the sustainability of WEDM by predicting the wire breakages, which are process interruptions which increase the machining time, energy wastage and material consumptions. They tried to predict process continuity by binomial classification of machining outcomes. Using kernel-based naïve Bayes algorithm. The trained model was subsequently 96.7% accurate in wire breakage predictions. WEDM is a process with many input factors influencing the machining efficiency and the resulting quality of the machined surface, including its subsurface layer. Given that even in the current state of knowledge, the process of influencing a workpiece made of different materials is not fully explored, it is necessary to further expand the areas of research. The aim of this study was a comprehensive assessment of the degree of influence of the subsurface layer in the form of changes in its microhardness for individual materials. This is linked to very extensive research of the behaviour of individual materials after WEDM, where subsurface defects of several different materials affecting the life of the part were studied, 18 the machinability of Hadfield steel, 19 Inconel, 20 Creusabro steel 21 or aluminium alloy 7475-T7351 22 were studied. The knowledge and possible prediction of the behaviour of individual materials after WEDM is key knowledge for machining parts with the highest possible surface quality without the occurrence of defects. This study is completely new in its scope, as no similar study has examined several materials simultaneously under the same conditions. If we have knowledge about the degree of change of microhardness of the subsurface layer of the material, it will be possible to better predict the life of the part.

Experimental material
Samples for the experiment were made of nickel superalloys B1914 and Nimonic 263, Hadfield steel, Creusabro 4800 and Hardox 400. The properties of all these materials, including microstructures, were summarized in Tab. 1. Material B1914 was cast and delivered by PBS Velká Bíteš a.s. A 15 mm thick semi-product was used for all samples.

WEDM machine setup
A wire electric discharge machine of type EU64 from the company MAKINO was used for machining. The machine was equipped with CNC control in 5 axes and deionized water was used as a dielectric liquid. The wire electrode marked PENTA CUT E had a diameter of 0.25 mm and was made of 60% Cu and 40% Zn. To monitor the influence of individual input factors on the resulting microhardness of the subsurface layer, in our case the parameters of machine settings (gap voltage, pulse on and off time, wire feed and discharge current)

Experimental methods
Metallographic cuts of cross-sections were prepared for the analysis of subsurface microstructural changes. These preparations were prepared by common techniques -wet grinding and diamond paste polishing using the automatic preparation system TEGRAMIN 30 from Struers. The final mechanical-chemical polishing was performed by vibro-polishing. After etching (Hadfield steel, Creusabro and Hardox were etched with Nital, Nimonic and B1914 were etched with kalling 2), the structure of the material was observed and documented by electron and light microscopy on an inverted light microscope (LM) Axio Observer Z1m from ZEISS and on a scanning electron microscope (SEM) of Lyra3 type from TESCAN. A TI 950 triboindenter from HYSITRON with a Berkovich type tip with a diameter of 50 nm was used to measure the microhardness of the subsurface layer. The load curve followed a standard trapezoidal function with a total force of 8 mN during the analysis. The electron backscatter diffraction (EBSD) analysis was performed on a Mira 3 microscope from TESCAN equipped with an EBSD analyser from OXFORD. During the measurement, the change of microhardness according to the ASTM E384 methodology 24 depending on the distance from the machined surface was investigated using the WEDM technology. The maximum load of the indenter during the measurement reached 8 000 µN, while its basic microhardness was measured in the middle of each sample. To determine the change both in the vicinity of the recast layer and in the base material, a total of 45 indentation measurements were performed on each sample, arranged in a 3 × 15 matrix with a distance of 3 µm between the indents. In order to ensure the protection of the indentation tip, the beginning of the nut was always situated 3 µm from the resin in which each sample was cast An example of samples made of Creusabro 4800, a metallographic specimen, and a cross-section with indents is shown in Figure 1. The individual vertical rows of indents are hereinafter referred to as Line 1, Line 2 and Line 3 and the horizontal ones are numbered 1 to 15.

EBSD analysis
Due to the thermal stress of the upper layer of the machined material due to the effects of the WEDM and rapid cooling, some materials recrystallize and change the overall orientation of the grains. This phenomenon has been investigated previously by EBSD in studies such as Klocke et al., 25 Salvati and Korsunsky 26 or Liu et al.. 27 The investigation of this process was performed on a Mira 3 microscope equipped with an EBSD analyser. A total of 20 measurements were made, always measuring in the middle of the sample in the base and completely unaffected material and then on each of the individual samples in the recast layer with a part of the base and unaffected material. Due to the orientation, the resin was always shown to a small extent, so that it is clear that this is the correct area, i.e. the recast layer. The size and resolution of the measurements are always given in the description of the figure, because the individual materials examined had different grain sizes and therefore it was not desirable to use the same resolution for all of them. The evaluation of the data was performed by means of the final comparison within one material at a time.
The recast layer size is shown for each sample in Figure 2, wherein the thicknesses of the individual recast layers are recorded in Table 3.
In the case of material B1914 from a comparison of the base material and the individual samples shown in Figure 3 (a), it is evident that due to WEDM there was recrystallization in the investigated area and individual grain size was reduced. From the comparison of inverse pole figures (IPF) it is clear that in the recast layer there was an overall decrease of individual grain orientations up to orientation 001. Since each pole figure describes the angular rotation of individual grains with respect to the scanning plane, it is possible to determine the change in the orientation of the grains. From this finding, it can be concluded that in the case of B1914, there was a preferential change to the 001 orientation due to machining in the recast layer area.
In the case of EBSD measurement with Nimonic material, which is shown in Figure 3 (b), no significant change in grain size was observed in the recast layer compared to the base material. This finding indicates only a minimal effect of WEDM on the material. A comparison of individual IPFs shows that the setting of the machine parameters significantly affected the overall orientation of the grains not only among themselves but also in comparison with the basic material.
From a comparison of the EBSD data for the Hadfield steel material shown in Figure 3 (c), it is clear that again the grain size in the recast layer area decreased due to the WEDM process. In addition, crystal orientations (green colour of grain in the middle of the study area) were observed, which were not identified by the Oxford device. Unfortunately, these grains have an unknown crystal structure. From the comparison of the IPF image of the base material of Hadfield steel with other samples, it is evident that in the base material a uniform distribution of individual crystal orientations prevailed compared to individual samples affected by the WEDM process. This finding shows that there were significant changes in   crystal orientation due to WEDM, both between individual samples and with respect to the base material.
In the case of EBSD measurements for Creusabro material (Figure 4 (a)), no significant change in the size and distribution of the individual grains was observed. From the IPF data for Creusabro, it is clear that there was no significant change in the orientation of individual grains due to WEDM in comparison of the basic material with individual samples measured near the recast layer. The only visible difference is the distribution of grains in the y-axis, where the uniform distribution of individual crystal orientations dominates for the base material. There are most differences in the y-axis between the individual samples.
From a comparison of the EBSD figures for Hardox material shown in Figure 4 (b) for individual samples and base material, that there was no significant change in grain size in the HAZ area. Mutual differences in orientation between individual samples are minimal in the xand y-axis. In the z-axis, however, the change is more significant. If we compare the grain orientation of the base material, the predominant proportion of grain orientation 1 0 1 in the y-axis for the BCC grid is clearly visible. The main reason for the change in grain orientation caused by WEDM is due to the fast hardening (thermal stress of the material).
The overall change in grain orientation depends on the material composition, the total thermal energy of the WEDM process, the thermal water content and the stability of the material. In the case of certain unknown conditions, there may be minimal or no change in crystal orientation.

Microhardnessanalysis
Due to the significant temperature rise and rapid cooling during WEDM machining, some materials have a change in microhardness H in the heat-affected zone. The microindentation method was used to determine this effect. The measurement was performed on a Hysitron TI 950 device, while the measurement methodology was described in detail in Chapter 2.3, including the representation of individual indents on the sample. In the following maps for microhardness and reduced elastic modulus Er, shown in Figures 5 and 6, for clarity, the distance from the sample surface is given instead of the number of individual indents (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15).
From the graphs shown in Figure 5, it is evident that for 2 of the studied materials there was no significant change in mechanical properties due to WEDM. These were the B1914 and Hardox 400 materials. A more significant change occurred with the Nimonic material, where the microhardness increased with increasing distance from the surface, which is supported by the Kumar et al. study. 28 There was also a more significant change in Hadfield steel and Creusabro 4800 steel, while in Creusabro steel the initial value of microhardness was approximately 2 times smaller than the value for the basic heat-unaffected material, in Hadfield steel, it was 3 times smaller. In his study, Straka also noted a change in the microhardness of the subsurface layer during the machining of X210Cr12 steel. 6 In addition to microhardness, the effect of machining on the reduced modulus of elasticity was evaluated, which includes the effect of the elastic response of the tip with the modulus of elasticity of the measured material. The obtained values were plotted in the graphs shown in Figure 6. The trend of individual changes was similar to the case of the microhardness measurement itself, with the only difference being that the growth of values from the surface was more significant. In the case of Nimonic, the overall growth was over 50%. For Creusabro steel the value increased 3 times and for Hadfield steel 4 times.
The dependence of the observed characteristics (H and Er) on the distance from the sample surface cannot be described by the classical regression model, because due to local inhomogeneities the scattering of the model residues is not homogeneous. It is, therefore, necessary to test the dependencies using the correlation coefficient. The assumption of the independence test using the Pearson correlation coefficient is that the data come from a twodimensional normal distribution, which our data do not satisfy. Therefore, the Spearman correlation coefficient Rho was used, which does not have this assumption in the independence test, and moreover, it is not only a measure of linear dependence but is a measure of monotonic dependence in general. Supplementary Table 4 shows the Spearman correlation coefficient Rho of the observed characteristics (H and Er) and the distance from the sample surface for individual materials. Next, the p-value of the independence test is given in parentheses, where the quantities are independent if the p-value is greater than the selected significance level α, which was standardly selected as 5%. Because the Rho coefficient is always positive, there is always an increase in these characteristics with increasing distance from the The linear squares calculated by the least squares method and shown in graphs 7(a) and 7(b) are for illustrative purposes only, as residues (deviations from the model) do not correspond to the normal distribution, and performing a classical regression analysis would be statistically incorrect. In addition, the Spearman correlation coefficient Rho is an ameasure of monotonic dependence and linear dependence is one of many monotonic dependencies, so the dependence equations themselves are not given and the graphs illustrate only the increase of observed quantities for individual materials depending on the distance from the surface. Of a particular note, there is Nimonic 263, where the dependencies may seem graphically insignificant, yet they are significant.

Conclusions
In order to detect changes in the microhardness of the subsurface areas affected by WEDM, an extensive analysis was performed involving 5 different metallic materials (B1914, Nimonic 263, Hadfield steel, Creusabro 4800 steel and Hardox 400), 3 samples were produced from each material using different machine settings, while the following conclusions have been reached: • on the basis of the EBSD analysis, it was found that the samples of B1914, Hadfield steel and Hardox 400 steel were recrystallized and the size of the individual grains was reduced, while the samples of Nimonic 263 and Creusabro 4800 showed no visible change, • from the IPF comparison it is clear that in the case of samples from B1914 there was a preferential change to orientation 001 in the area of recast layer due to WEDM, in samples of theNimonic material the setting of machine parameters significantly affected the overall grain orientation not only between them but also in comparison with the base material, • for Hadfield steel samples there was a significant change in crystal orientation on the basis of IPF, both between individual samples and with respect to the basic material, for Creusabro samples on IPF there was no significant change in grain orientation due to WEDM and in Hardox steel the samples are minimal on the x-and y-axis, but the change on the z-axis is more significant, • the microhardness analysis revealed that the samples of B1914 and Hardox 400 did not show a significant change in mechanical properties due to WEDM, • the samples made of Nimonic increased microhardness with the increasing distance from the surface, • a more significant change in microhardness occurred in samples made of Hadfield steel and Creusabro 4800 steel, with the initial microhardness value for Creusabro steel being approximately 2 times smaller than the value for the base heat-unaffected material and 3 times smaller for Hadfield steel, • the reduced modulus of elasticity had a trend of individual changes similar to that in the case of the microhardness measurement itself, with the only difference that the growth of values from the surface was more significant, • based on the statistical evaluation, it can be said that the increases in microhardness and reduced modulus of elasticity are statistically significant for Creusabro 4800, Hadfield steel and Nimonic 263 (p-value <0.05), for the other two materials (B1914 and Hardox 400) these increases are statistically insignificant.
In the case of Nimonic and Hadfield steel the decrease of microhardness is connectedwith smaller crystals after WEDM process. But in case of Creusabro the change of microhardness could not by understand by microstructure change but by the thermal shockinfluence during WEDM quenching. Rest of the materials was not influencedby WEDM from point of microhardness. Based on the above-mentioned conclusions, it can be clearly stated that changes in the microhardness of the subsurface layer after WEDM are clearly influenced by the type of material being machined and may not occur at all. However, the changes in microhardness affected by the setting of the machine parameters are negligible.