**3. Results**

## *3.1. Sample Breakdown*

No arc discharge was observed during the experiments. Samples C, D, and E were broken-down under some conditions as shown in Table 2. Higher relative permittivity of these samples compared with samples A and B as shown in Table 1 generates higher localized electric field in the air gap, resulting in active partial discharge and shorter lifetime to breakdown. Also, lower resistivity of these samples may have some relation to breakdown.


**Table 2.** Number of cycles of sample breakdown.

## *3.2. Contact Angle*

Figure 3a,b shows the change in contact angle with number of cycle just before voltage application and just after voltage interruption in the 50-cycle test with 2 mm air gap spacing, respectively. Characteristics obtained in the 100-cycle test with 2 mm gap spacing are shown in Figure 4a,b. The initial value of contact angle is about 100 degrees for any sample. After 8-h voltage application, it decreased at most 20 degrees due to reduction of hydrophobicity of sample surface by partial discharge. Reduction in contact angle of Sample A is much lower than those of other samples. This corresponds to lower surface erosion of Sample A as described below. Recovery of contact angle to almost the initial value is confirmed after 16-h voltage interruption. It is considered that 16 h in the present experimental procedures are appropriate for low-molecular weight silicone oil to migrate from the bulk to the surface of a sample and to recover hydrophobicity. No remarkable reduction is observed with the number of cycle. Samples A and B are superior to the other samples when repeating cycles.

Large fluctuation of contact angle may be attributed to microscopic uneven surface caused by partial discharge, not only erosion like pitting but also accumulation of discharge by-product on the surface as suggested by the results of surface roughness measurement shown later in Figure 7. Contact angle does not necessarily reflect the surface profile in a small area, because the size of a water droplet for evaluation of contact angle is large enough to mask microscopic uneven surface and consequently an average information on surface in a larger area will be given.

*Energies* **2019**, *12*, 2756

**Figure 3.** Change in contact angle with number of cycles in 50-cycle test with gap spacing 2 mm. (**a**) Just before commencement of voltage application. (**b**) Just after voltage interruption.

**Figure 4.** Change in contact angle with number of cycles in 100-cycle test with gap spacing 2 mm. (**a**) Just before commencement of voltage application. (**b**) Just after voltage interruption.

#### *3.3. Observation of Sample Surface*

Optical and confocal microscope images of sample surface are shown in Figure 5, which indicate change in surface erosion in the area of 1.5 mm × 1.5 mm of samples A and E in the 100-cycle test with 1 mm air gap. It is clearly shown especially by confocal microscope images that surface erosion progresses gradually with the number of cycle and degree of erosion is much different between samples A and E; surface erosion of sample E is serious and sample A is eroded slightly even after completing 80th cycle.

Figure 6 shows confocal microscopic images of all samples taken after completing 80 cycles in the 100-cycle test with 1 mm air gap. Appearance of surface erosion of Sample B is similar to that of Sample C. Sample D has eroded surface similar to Sample E. Samples D and E are severely damaged, which relates to the shorter lifetime to breakdown as shown in Table 2.

A quantitative discussion of surface erosion will be given in the next section.


**Figure 5.** Optical and confocal microscope images of surface of samples A and E obtained in the 100-cycle test with 1 mm air gap.

 **Figure 6.** Confocal microscope images of sample surface obtained after completing 80 cycles in the 100-cycle test with 1 mm air gap. (**a**) Sample A; (**b**) Sample B; (**c**) Sample C; (**d**) Sample D; (**e**) Sample E.

(**d**) (**e**)

## *3.4. Surface Erosion*

#### 3.4.1. Analysis Based on Surface Roughness Meter

Figure 7 shows surface roughness profiles of samples A and E, which are obtained after completing 6 and 60 cycles in the 100-cycle test with 1 mm air gap. Negative surface roughness means pitting on a sample surface generated by partial discharge. Discharge by-product accumulated on the sample surface may be a possible reason for the positive surface roughness. No remarkable change in surface roughness of sample A is observed between after 6 and 60 cycles. On the contrary, the surface of sample E is more eroded after completing 60 cycles; surface roughness is ~10 times larger compared with that after completing six cycles. These results correspond well to lower contact angle and shorter lifetime to breakdown of sample E compared with those of sample A, which is described above.

To evaluate quantitatively surface roughness shown in Figure 7, the arithmetic mean roughness Ra is used. Change in Ra in the 50-cycle tests with 1 and 2 mm air gap are shown in Figure 8a,b, respectively. When applied voltage is low, it is clear from Figure 8a that erosion of samples B to E progresses gradually with the number of cycle. Surface erosion is progressed in the order of samples D and E, samples B and C, and A, but di fference among samples is not significant. In the case of 2 mm air gap, higher applied voltage intensifies partial discharge activity in the air gap and samples are clearly divided into three groups from standpoint of surface erosion. Samples D and E are eroded seriously by partial discharge and finally break down before completing 100 cycles. Erosion of samples B and C progresses steadily with the number of cycle. Sample A is eroded little. It is understood from Figure 8a,b that acceleration of partial discharge erosion is achieved by increasing applied voltage and gap spacing, especially for samples D and E.

Ra obtained in the 100-cycle test is shown in Figure 9 for 1 mm air gap. Ra increases gradually with the number of cycle except sample A. The characteristics of samples B and C are similar to each other. The same can be said of samples D and E. The result for 2 mm air gap are not shown because only limited data were available due to failure of the measuring instrument. Comparing Figure 9 with Figure 8a, obtained under the same applied voltage and the same gap spacing, surface roughness after completing 40 cycles, for example, is almost the same for any sample. This shows a good repeatability of results, suggesting the proposed experimental procedures are suitable to investigate partial discharge degradation of silicone rubber.

**Figure 7.** Examples of surface roughness profiles obtained with surface roughness meter in the 100-cycle test with 1 mm air gap.

**Figure 8.** Change in Ra with number of cycle in the 50-cycle tests. (**a**) 1 mm air ga; (**b**) 2 mm air gap.

**Figure 9.** Change in Ra with number of cycle in the 100-cycle test with 1 mm air gap.

#### 3.4.2. Analysis Based on Confocal Microscopic Images

In the previous section, surface roughness along the 3 mm straight line is discussed. Analysis of surface roughness in the area (1.5 mm × 1.5 mm) is available with confocal microscopic images shown in Section 3.2. Three-dimensional coordinate can be obtained at the time of surface scanning, where Xand Y-coordinates indicate the location on the sample surface and Z-coordinate shows height with reference to the sample surface without erosion.

Figure 10a,b shows changes in the arithmetic average roughness Ra over the 1.5 mm × 1.5 mm area with the number of cycle in the 100-cycle tests with 1 mm and 2 mm air gap, respectively. Samples D and E give larger Ra when compared with the other samples, which is the same tendency with results obtained with the surface roughness meter. Ra in Figure 10 is much larger than that in Figures 8 and 9. A possible reason is the difference in resolution of measuring instruments used. As the surface roughness meter is a stylus type apparatus, a fine probe needle is moved on a sample surface. Surface roughness is obtained based on up-and-down movement of the probe. Thus, it is considered difficult to measure precisely the depth of a narrow pitting smaller than the probe diameter, resulting in lower value of Ra. Meanwhile, LASER beam is used to evaluate surface roughness in the confocal microscope. Since the resolution is higher and a long and narrow pitting can be evaluated precisely, larger Ra is obtained.

**Figure 10.** Change in Ra with number of cycle in the 100-cycle tests. (**a**) 1 mm air gap; (**b**) 2 mm air gap.

## *3.5. FTIR Spectrum*

Change in FTIR spectrum with the number of cycle is shown in Figure 11 for all samples, which are obtained in the 100-cycle test with 2 mm air gap. Characteristics of Samples B and C are similar to each other. The same can be said of Samples D and E. These results correspond to progress of surface roughness described above.

In the case of sample A without ATH, absorbance around 786 and 1257 cm<sup>−</sup><sup>1</sup> attributed to hydrophobic Si-CH3 [13] decreases with the number of cycle. Almost no change is observed in absorbance around 1008 cm<sup>−</sup><sup>1</sup> attributed to Si-O-Si. Increase of silica-related absorbance [13] is observed around 1060 cm<sup>−</sup><sup>1</sup> with increase in the number of cycle. A decrease in absorbance of hydrophobic Si-CH3 and almost no change in absorbance of Si-O-Si, shown in Figure 12a, sugges<sup>t</sup> that a side chain of silicone rubber is cleaved by partial discharge but main chain is damaged little, leading to limited erosion.

In sample E, eroded more by partial discharge compared with sample A, absorbance around 3620–3370 cm<sup>−</sup><sup>1</sup> vanishes when the number of cycle is large, which is related to ATH [13]. At the same time, absorbance around 940 cm<sup>−</sup><sup>1</sup> increases, which is attributed to Si-H [13]. It is suggested that ATH changes into aluminum hydroxide by losing hydroxyl groups. Not only absorbance of hydrophobic group Si-CH3 but also that of Si-O-Si of main chain of silicone rubber decrease rather rapidly with the number of cycle as shown in Figure 12b, suggesting progress of silicone rubber decomposition and short lifetime to breakdown.

 (**e**) 

**Figure 11.** Change in FTIR spectra with the number of cycle obtained in the 100-cycle test with 2 mm air gap. (**a**) Sample A; (**b**) Sample B; (**c**) Sample C; (**d**) Sample D; (**e**) Sample E.

**Figure 12.** Change in normalized absorption of Si-CH3 and O-Si-O with the number of cycle obtained in the 100-cycle test with 2 mm air gap. (**a**) Sample A; (**b**) Sample E.
