Taguchi Optimization of Electrolytic Plasma Hardening Process Parameters on Ti‐6Al‐4V Alloy: Microstructure and Mechanical Properties

Herein, the effects of electrolytic plasma treatment (EPT) on the hardness distribution, wear resistance, and surface morphologies of titanium alloy (Ti‐6Al‐4V) at three different thermal cycles (4‐5‐6 times) are investigated. The microstructure is controlled by heat temperature and modification times of thermal cycles depending on EPT processing parameters. A novel technique to modify Ti‐6Al‐4V alloy by the process of EPT successfully occurs. The results show that the phase transformation is shown as follows: fine α and β for the low thermal cycle → irregular martensite α′ for the high thermal cycle → complete zigzag martensite α′ for the max thermal cycle. The hardness is increased from 350 ± 10HV0.05 to 530 ± 10HV0.05 after the modification. Wear tests are conducted according to the Taguchi L9 (3^3) orthogonal array. Three parameters (load, sliding speed, and sample type) with three levels examine wear rate and coefficient of friction. Optimum levels are obtained by Taguchi analysis from the experimental results. In addition, an analysis of variance is performed to find the effectiveness of the parameters. The wear surfaces are analyzed by scanning electron microscope. After the analysis, the best result is obtained in five thermal cycles.

surface properties of Ti-6Al-4V just in a few seconds.Another important point; the Taguchi method was used to plan experiments and to detect optimum parameter levels.Analysis of variance was conducted to find the effects of parameters over the results.The flowchart of the study is given in Figure 1.

Materials Selection
Ti-6Al-4 V plates with an in 100 Â 50 Â 5 mm were hardened with EPT, and the chemical composition of this alloy is given in Table 1.All the specimens were polished with 120 grit to 1000 grit, cleaned with alcohol, and dried before surface treatment.
Sodium carbonate (Na 2 CO 3 ) solution was used as a heating and cooling source.The composition of the electrolyte is 88% distilled water and 12% sodium carbonate (wt%).The samples were hardened at the same voltages of 310-250 V.The electrolytic plasma impulse (heating), pause values (cooling), treatment time, and thermal cycle are listed in Table 2.After the surface hardening, the microstructure was analyzed using NIKON ECLIPSE L150 optical microscope (OM) and Joel (JSM 6060-LU) scanning electron microscope (SEM).

Hardness Tests
The cross-sectioned hardened layer was measured using a Future Tech tester with 50 g load and 10 s.The presented hardness indentation results are an average of three profiles.

Wear Test
The linear wear tests were carried out on a CSM-Tribometer ball-on-disc tester using a 6 mm diameter Al 2 O 3 ball.The wear tests were suitable to DIN 50 324 and ASTM G 99. [24] Unlubricated wear tests with a total sliding distance of 300 m were carried out at room temperature (%25 °C), relative humidity of approximately 45%, a sliding speed of 10-20-30 cm s À1 , a normal load of 1-3-5 N, and the full amplitude 20 mm.The profiles were recorded before and after the wear tests using a profilometer to calculate the wear volume.

Experimental Design
A part of linear sliding wear tests was performed based on the study order generated by the Taguchi technique.The Taguchi method was used in planning experiments and finding optimum levels in this study.Taguchi is an experimental design method that saves cost and time by reducing the number of experiments and finding the optimum parameter levels with the signal-to-noise ratio (S/N) approach. [25]The parameters and levels used in tribological tests are given in Table 3.The wear tests were conducted according to Taguchi L9 (3 ^3) orthogonal array in Table 4.
According to this method, the data obtained from the experiments are converted into the signal-to-noise ratio and analyzed.S/N ratio calculation is done in three ways to decide the optimum value according to the desired result.These are "the lower is better", "nominal is best", and "the higher is better".In this study, wear rate and coefficient of friction were wanted to be low.So "the lower is better criteria" was chosen.According to this criteria, signal/noise (S/N) is calculated according to Equation (1) below.In the equation, y i represents the measured data and n is the number of experiments. [26] In addition, the effect percentages of the parameters on the results were calculated with analysis of variance (ANOVA).All statistical analyses of the measured values were performed with the Minitab 19 program at a 95% confidence level.

Microstructural Analyses
The microstructure images of the sample cross-sections taken to examine the effect of the electrolytic plasma hardening process on the Ti6Al4V alloy, which is an alpha-beta alloy, are shown in Figure 2a.When we look at the microstructure of the untreated Ti6Al4V alloy, it is seen that the microstructure consists of the equiaxial alpha (α) structure as well as the intergranular beta (β) phase. [10,27]29] As shown in Figure 3, looking at the microstructure images, the effect of the EPT process on the microstructure can be examined into three different groups; heat-affected zone (HAZ), unaffected zone of based material (non-HAZ), and hardened zone (Figure 3). [10]As a result of the diffusionless transition following electrolytic plasma hardening, the initial microstructure of equiaxed phase α and β phase transferred into β phase, first, from α!β, and subsequently transformed into α martensite phase under a relatively high cooling rate from β!α. [29,30]n this study, the electrolytic plasma hardening process is carried out in three different cycles.When the optical microscope images were examined, although there was no dramatic change in the surface morphology of the T4 sample, the martensitic structure formed by the rapid heating and subsequently fast cooling in the T5 sample, which was applied for five cycles, was effective in a wide area.The surface image obtained after six cycles shows the molten solidified zone effect on the surface (Figure 4).

Wear Results and Discussion
Table 5 shows the wear rate and coefficient of friction results and S/N analysis from the Taguchi method.
Table 6 and 7 show the S/N table.The last lines of the tables show the order of effectiveness of the parameters on the results.The sample load is the most effective in the wear rate results, the speed is the second, and the sample type has a minor effect.On the coefficient of friction, while the speed is the most influential parameter, the load is second.Sample type has the lowest effect.
Figure 5 shows the main effects plot for S/N ratios for wear rate.Optimum levels are those with the highest S/N ratio.Levels to be used to have the lowest wear rate are A2B1C3 (3 N, 10 m s À1 , T5).
According to Figure 6, the minimum coefficient of friction can be achieved with A2B3C1 (3 N, 30 m s À1 , T6) levels that are optimum for the coefficient of friction.
ANOVA aims to determine how the parameters selected affect the results selected to measure the quality and how the different levels cause variability. [31]Table 8 and 9 show the analysis of variance for the coefficient of friction and wear rate, respectively.The effects of parameters in ANOVA were calculated by comparing the F-value of each parameter.Minitab uses the F-value to calculate the p-value, which is used to decide about the statistical significance of the terms and model.The p-value is a probability that measures the evidence against the null hypothesis.Lower probabilities provide stronger evidence against the null hypothesis. [32] sufficiently large F-value indicates that the term or model is significant.The effects of the parameters in the ANOVA were calculated automatically via Minitab by comparing the F-value of each parameter according to Equation (2).In this equation, SS is sum of squares of related factor, DF is degree of freedom of related factor, SS e is sum of squares of error term, and DF e is degree of freedom of error term.
The contribution of each factor to the total variation is shown in percent (%) in the last column of the tables.According to Table 8, the most influential parameter for the wear rate is the load with 52.60%, the speed is in the second order with 25.48%, and the sample type is in the last order with 13.21%.
As shown in Table 9, the dominant parameter in the coefficient of friction results is the speed at 78.59%, followed by the load at 6.24%, and finally, the sample type at 1.20%.The probability plot evaluates how well the data fits a particular theoretical distribution.When the probabilities of the predicted values tend to a point, and the normal line can be approached, a suitable model between the normal distribution and the actual values is obtained. [33]Figure 7a,b shows that the data points      are close to the standard distribution line.Therefore, the data can be used in experimental and optimization studies.
The surface plots of the wear rate are given in Figure 8. Figure 8a shows the three-dimensional (3D) surface plot of wear rate against speed and load.It is seen that the wear rate increases as the speed and load increase.It is seen that the wear is max when the load is 5 N and the speed is 30 cm s À1 .Figure 8b shows surface plot of wear rate against speed and sample type.It is seen that the highest wear rate occurs in the T4-type sample.The wear rate also increases with increasing speed.Figure 8c shows surface plot of wear rate against load and sample type.The highest wear is observed in the 5 N and T4 type sample combination.
The surface plots of the coefficient of friction are given in Figure 9. Figure 9a shows the 3D surface plot of the coefficient of friction (CoF) against speed and load.The lowest coefficient of friction is observed under a load of 30 cm s À1 and 3 N.  shows surface plot of CoF against speed and sample type.The lowest coefficient of friction is seen in the T6 sample.Figure 9c shows surface plot of CoF against load and sample type.The lowest coefficient of friction was found in the combination of T6 and 3 N.
In this study, a linear sliding wear test set was performed based on the Taguchi technique.Various results were obtained according to the determined variations.Wear volume ratios and friction coefficients are given in Table 10.According to the parameters applied according to the Taguchi method, the wear rate increased with increasing the applied sliding speed.It has been observed that the speed is more effective in the wear rate than the applied load effect.The highest wear rate at maximum speed was obtained in T4-coded samples.The surface morphology of the T4 sample was not changed, and the lowest hardness data after EPT treatment were seen in sample T4 that is why this was an expected result for the wear rate.
The hardness of the substrate material was measured as 350 AE 10HV0.05.After processing, the hardness value was measured as 530 AE 10HV0.05.High hardness generally indicates good wear resistance.The average hardness of the EPT-modified Ti-6Al-4V was 530 AE 10HV0.05,which is relatively high compared to compere literature. [8,14,34]In general, high hardness was obtained in samples showing the martensitic transformation.Microhardness during EPT is that the β phase with lower hardness is transformed into acicular type α martensite as well as plate-like martensite with high hardness.Therefore, there is almost no difference in hardness between the thermal cycle 5 and 6.The hardness of the segregation layer formed in the molten zone layer was also measured similarly.
The modification of the structure of T5 showed an improvement in wear rate with an improvement in hardness.The best results were obtained for T5 at middle wear load and low sliding speed (Figure 10).Although the wear load decreased, the wear rate increased with the sliding speed.This showed that speed  was more effective than the load-on-wear mechanism.With increasing speed, an unstable titanium oxide layer may have formed on the surface.Furthermore, it strengthens the possibility that this layer was broken and acted as an extra abrasive on the surface.
Although the wear rate at low speed and low load in the T6 sample was better than the other samples, this situation depends on the changing load and speed (Figure 10).The most important observation was that the wear rate of the remelted sample (T6) was higher than that of the EPT modified samples under the high sliding speed conditions.According to T4 and T5, the T6 coded sample, it was observed that the weight loss did not increase much with the increase of the sliding speed (30 cm s À1 ).The microhardness values of EPT remelted coating was 530 AE 10HV0.05.Compared with the wear rate of EPT treated, the wear rate of remelted was much lower, especially at a higher speed.While there are many ways to increase the wear resistance of titanium, nowadays the development of structural modification technologies has been very popular. [35,36]EPT is a virgin in academic studies and offers the opportunity to improve surface properties in a few seconds.Similar observations on Ti alloys have been reported by Mohazzab [37] with laser treatment.Jing et al. [38] and Chan et al. [39] reported titanium alloy surface prepared by laser cladding and laser nitriding.
Figure 11 shows the SEM morphologies of the worn surface after EPT treatment.As shown in Figure 11 first group, the sample surfaces were not severely worn, without plastic deformation.Ti alloys have low hardness and wear resistance.We aim to improve the strength by increasing the surface properties.The first group was tested on 1 N wear load.The damage to the surfaces intensified with increasing the load.Spalling is a kind of typical damage caused by hard abrasive through the surface.Meanwhile, a large number of wear debris and grooves can be seen on sample T4-5N-30 cm s À1 .It is seen that the wear is intensified with increasing load and sliding speed.However, this damage was more affected by the speed.The expected result is that  the sliding speed increase will lead to changes in the strain rate and friction heating. [6,40,41]he last step in the Taguchi method is verification tests.These tests confirm the accuracy of optimization.Therefore, confirmation tests were performed at optimum levels of parameters and results are given in Table 11.According to experimental and predicted results with the optimized levels of factors, smaller values for wear rate and coefficient of friction were obtained.The error between the experimental and estimated result is 8.718% for wear rate and 9.972% for the coefficient of friction.From these results, we can say that optimization is successful.

Conclusion
In terms of microstructural and mechanical properties of EPT the effect on Ti-6Al-4V alloy were studied in this study.The findings obtained in the study are briefly summarized below: The initial microstructure of equiaxed phase α and β phase transferred into β phase, first, from α!β, and subsequently transformed into α martensite phase under a relatively high cooling rate from β!α.The microhardness of EPT modified zone is significantly higher than that of base material (BM).The average microhardness value of the modified Ti alloy increased from 350 AE 10HV0.05 to 530 AE 10HV0.05,which was improved by about 51%.With the increasing sliding speed, the coefficient of friction gradually decreases; at the same time, the wear resistance of the modified samples decreases, too.The thermal cycle is very effective in the modification.The wear rate decreased with increasing thermal cycling.As a result of the probability analysis using the experimental results, a confidence level of 95% was obtained.This value confirms that the experimental results can be used for optimization work.To Taguchi optimization, the levels to be used to have the lowest wear rate are A2B1C3 (3 N, 10 m s À1 , T5).The minimum coefficient of friction can be achieved with A2B3C1 (3 N, 30 m s À1 , T6) levels that are optimum for the coefficient of friction.The effects of factors on the tribological results were investigated with ANOVA.The most influential parameter for the wear rate is the load with 52.60%, the speed is in the second order with 25.48%, and the sample type is in the last order with 13.21%, according to ANOVA results.The dominant parameter in the coefficient of friction results is the speed at 78.59%, followed by the load at 6.24%, and finally, the sample type at 1.20%.The analysis was verified by performing confirmation tests at optimum levels.According to experimental and predicted results with the optimized levels of factors, optimization has been successful.

Figure 1 .
Figure 1.The flowchart of the study.

Figure 5 .
Figure 5. Main effect plot of S/N for wear rate.

Figure 6 .
Figure 6.Main effect plot of S/N for the coefficient of friction.

Figure 8 .
Figure 8. Surface plots of wear rate a) wear rate versus load; speed b) wear rate versus sample type; speed c) wear rate versus sample type; load.

Figure 7 .
Figure 7. a) Probability plot of coefficient of friction and b) probability plot of wear rate.

Figure 9 .
Figure 9. Surface plots of coefficient of friction a) CoF versus load; speed b) CoF versus sample type; speed c) CoF versus sample type; load.

Figure 11 .
Figure 11.The SEM images of the wear track morphology of EPT-modified Ti-6Al-4V alloys.

Figure 10 .
Figure 10.Wear rate and friction coefficient of Ti-6Al-4V alloy under different times of thermocycles hardening.

Table 3 .
Parameters and levels.

Table 5 .
Experimental and S/N results.

Table 6 .
Response table for S/N of wear rate.

Table 7 .
Response table for S/N of coefficient of friction.

Table 8 .
ANOVA table of wear rate.
DF: Degree of freedom, SS: Sum of squares, MS: Mean of squares, Cont %: Percentage contribution.

Table 9 .
ANOVA table of coefficient of friction.
DF: Degree of freedom, SS: Sum of squares, MS: Mean of squares, Cont %: Percentage contribution.

Table 10 .
Wear results of EPH-modified samples.