Clarifications on the Behavior of Alternative Gases to SF₆ in Divergent Electric Field Distributions under AC Voltage

Abstract

Negative and positive partial discharge inception voltages and breakdown measurements are reported in a needle-plane electrode system as a function of pressure under AC voltage for natural gases (N₂, CO₂, and O₂/CO₂), pure Novec^TM gases (C₄F₇N and C₅F₁₀O) and Novec^TM in different natural gas admixtures. For compressed 4% C₄F₇N–96% CO₂ and 6% C₅F₁₀O–12% O₂–82% CO₂ gas mixtures, the positive-streamer mode is identified as the breakdown mechanism. Breakdown and negative partial discharge inception voltages of 6% C₅F₁₀O–12% O₂–82% CO₂ are higher than those of 4% C₄F₇N–96% CO₂. At 8.8 bar abs, the breakdown voltage of 6% C₅F₁₀O–12% O₂–82% CO₂ is equal to that of 12.77% O₂–87.23% CO₂ (buffer gas). Synergism in negative partial discharge inception voltage/electric field fits with the mean value and the sum of each partial pressure individually component for a 20% C₄F₇N–80% CO₂ and 6% C₅F₁₀O–12% O₂–82% CO₂, respectively. In 9% C₄F₇N–91% CO₂, the comparison of partial discharge inception electric fields is E_max (CO₂) = E_max(C₄F₇N), and E_max (12.77% O₂–87.23% CO₂) = E_max(C₅F₁₀O) in 19% C₅F₁₀O–81%(12.77% O₂–87.23% CO₂). Polarity reversal occurs under AC voltage when the breakdown polarity changes from negative to positive cycle. Polarity reversal electric field E_PR was quantified. Fitting results show that E_PR (CO₂) = E_PR(9% C₄F₇N–91% CO₂) and E_PR(SF₆) = E_PR (22% C₄F₇N–78% CO₂). E_PR (4% C₄F₇N–96% CO₂) = E_PR (12.77% O₂–87.23% CO₂) and E_PR (6% C₅F₁₀O–12% O₂–82% CO₂) < E_PR (4% C₄F₇N–96% CO₂) < E_PR (CO₂).

1. Introduction

In compressed SF₆ [1,2,3,4,5], the propagation mechanism (leader criterion) is the breakdown mode in the presence of conducting particles fixed/floating on conductors or insulators. In the presence of invasive pollutants, the streamer inception level is lower than the leader inception level. The conventional partial discharge (PDs) measurement technique constitutes a means to detect the presence of defects without conducting destructive tests.

On the other hand, for C₄F₇N/CO₂ mixtures [6], the PDs activity under AC voltage varies as a function of pressure and the magnitude of the electric field, including breakdown polarity reversal (PR), when the AC breakdown polarity changes from negative half-cycle to positive half-cycle with the increase of pressure. At the reversal critical pressure (PR pressure), breakdowns occur on positive or negative half-cycles. Before PR, PDs are observed during the negative half-cycle, and current pulses occur on the positive cycle. For the same electric field distribution, increasing the pressure leads to an increase of the partial discharge inception voltage (PDIV) and an increase of the intensity of the maximum electric field on the tip of the needle. The latter causes inception of a denser PDs activity on the positive cycle and an attenuation of the PDs on the negative cycle. For pressures above PR, negative PDs activity decreases until being hidden by the background noise of the current sensing circuit. In this case, only higher positive polarity PDs activity is detectable, and the positive PDIV is close to breakdown. It was reported that the breakdown voltage is equal under positive and negative polarities under lightning impulse voltage waveform in quasi-uniform fields and at PR pressure [7,8]. The PR pressure increases linearly with the increase of field utilization factor η given by Equation (1). It should be noted that η = E_mean/E_max, E_max being the maximum electric field for a given electrode configuration; it is computed using finite element method simulation using COMSOL Multiphysics^TM software.; and E_mean is the mean electric field given by E_mean = V/d where V is the applied voltage and d is the electrodes gap. The linear increase is characterized by a constant polarity reversal electric field E_PR calculated according to Equation (1), where V_i is the positive PDIV for divergent electric fields and the breakdown voltage for quasi-uniform electric fields, d is the electrode gap.

Table 1. Reported polarity reversal electric field E_PR according to Equation (1) [6,7,8].

Gas/Gas Mixture	EPR (kV/mm)	Pressure (bar abs)	η
SF6 [7,8]	86	7.5	0.17
CO2 [6]	64	2.5	0.0385
3.7% C4F7N–96.3% CO2 [8]	47	9.5	0.33
10% C4F7N–90% CO2 [7]	66	5.5	0.22

reports E_PR for different gas/gas mixtures [6,7,8].

$$\left({\mathrm{E}}_{\mathrm{PR}}\mathrm{or}{\mathrm{E}}_{\max }\right)=\frac{{\mathrm{V}}_{\mathrm{i}}}{\mathrm{d}\cdot \mathrm{n}}$$

(1)

The present work investigates both the negative/positive-inception and breakdown voltages of a classic defect in GIS (fixed particle on the live conductor) under AC voltage waveform for different gases and gas mixtures. Natural gases, such as N₂, CO_2, and O₂/CO₂, are also assessed since they constitute the buffer gas(es) when mixed with the fluorinated SF₆ replacement candidates (C₄F₇N or C₅F₁₀O). The main purpose of adding natural gases is to lower the boiling point of the final fluorinated/natural gas mixture to reach operating temperatures similar to SF₆ for indoor and outdoor high voltage applications. In addition, pure C₄F₇N and C₅F₁₀O were studied at the sub-atmospheric pressure level without liquefaction. Then, including minimal working temperature and its corresponding concentrations [7,8,9], the addition of highly electronegative gases to natural gases is investigated, and the synergism of the constituents is quantified. Inception electric field was calculated in positive/negative polarities for pure/gas mixtures, and synergism laws are deducted. The polarity reversal electric fields are calculated for the C₄F₇N/CO₂ and C₅F₁₀O/O₂/CO₂ gas mixtures.

2. Materials and Methods

To study pure gas/gas mixture pre-breakdown and breakdown characteristics under non-uniform fields, a needle/plane defect model was utilized as described in previous work [6]. The needle was placed between two plane electrodes of Bruce profile to provide homogeneous background electric field conditions. The plane electrodes were made of aluminum and the 100 µm tip radius needle from tungsten of 99.99% purity. The electrode set was placed inside a vessel made of stainless steel (volume = 12 L, max pressure = 12 bar abs). The high voltage needle electrode that was connected to the bushing was fixed and was connected to a high voltage AC source (3.75 kVA) with a maximum output voltage of 50 kV rms. The high voltage bushing installed on the test vessel was rated at 38 kV AC rms. The grounded plane electrode, however, was vertically moveable within the test chamber so that the desired gap length can be achieved with an accuracy of 0.1 mm. Current sensing was performed by coupling a high-frequency current transformer (HFCT), of 500 MHz upper cutoff frequency, to the lower (grounded) plane electrode, while the recording of the generated signals was performed through a 500 MHz, 2 GS/s oscilloscope (see Figure 1).

Figure 2 shows the field utilization factor η given by Equation (1) of the experimental arrangement as a function of gap distance. Such an electrode configuration gives a constant field utilization factor η = 0.0385 for gaps between 10 mm and 20 mm. For a 60 kV peak AC applied voltage, η = 0.0385 and d = 10 mm, the maximum electric field E_max is ≈156 kV/mm.

The different investigated gas/gas mixtures are listed in

Table 2. List of investigated gases and gas mixtures.

Gas/Gas Mixture	Maximum Pressure (bar abs)
CO2	8.8
N2	8.8
12.77% O2–87.23% CO2	10
C5F10O	0.6
6% C5F10O–12% O2–82% CO2	10
C4F7N	1
4% C4F7N–96% CO2	10
20% C4F7N–80% CO2	5

. A Dilo (mini-series ^C5 and mini-series ^C4) was used to circulate the Novec^TM 5110 (C₅F₁₀O) and Novec^TM 4710 (C₄F₇N) mixtures in the gaseous state. For a 6% C₅F₁₀O/12% O₂/82% CO₂ gas mixture [9], a premixed 12.77% O₂/87.23% CO₂ gas mixture was used for the preparation of the gas mixture and representing the buffer gas.

The inception voltage of pre-breakdown discharges, which originated from the needle tip, was investigated optically in the visible and ultraviolet light emission spectrum. The optical detection system consisted of a UV image intensifier, incorporating an S20 photocathode, mounted on the front of an 8-bit high-speed camera with a maximum resolution of 1024 × 1024 px. A chromatically corrected Nikon UV-105, 105 mm f/4.5 lens was installed on the optical input of the intensifier. Observations were performed through a UV-grade fused silica side viewport installed on the pressure vessel. The entire camera system was aligned to and focus adjusted towards the installed electrode configuration. A UV bandpass filter mounted on the front of the UV lens was used to absorb the visible spectrum light emitted by glow discharge. While current detection was difficult and sometimes impossible in the high-pressure range (at pressure levels between 5 and 10 bar abs), using a high-frequency current transformer (500 MHz) of high gain ratio (5 V/A), combined with the camera system described above, consisted an effective method to detect PDIV in the UV and VIS spectral range. Figure 3 shows an example of a negative-streamer light emission image of N₂ at 5 bar abs.

A new needle was used for each gas/gas mixture to prevent deformation in the radius of the needle and biased results. No erosion effect on needle tips was observed after partial discharge inception voltage (PDIV) measurements. Initially, the PDIV was investigated, followed by breakdown voltage (BDV) measurements. No difference in BDV results was observable between new and used needles up to a certain limit; however, the needle was replaced frequently (see Figure 4). In contrast, one single BDV caused a decrease in PDIV.

For breakdown measurements, an AC voltage ramp was increased at a rate of 2 kV/s up to 90% of the voltage level that initiated electrical discharges, and then 0.1 kV/s was used. For each breakdown event, measurement series of 10 individual discharges were performed. Three minutes were kept in between the individual electrical discharges. The test rig bushing flashover limit was 60 kV peak. The ramp method was employed to identify the inception voltage of partial discharge. The voltage was increased with the rate of 1 kV/s until the first pulse, or a light emission image (UV-vis spectrum), was detected.

3. Results

3.1. Natural Gases

N₂, CO₂, O₂/CO₂ mixtures are interesting because they are perfectly ecologically compatible, and they constitute a major part of the atmosphere. The pure nitrogen used contains around 0.01% oxygen. Here, negative and positive-inception voltages and AC breakdown measurements are reported for N₂ at various pressures up to 8.8 bar abs (see Figure 5). The effect of needle condition (non-arced vs. arced) was investigated. After breakdown measurements (arced needle), a change in negative-inception voltage was observed at high pressure, and a saturation trend occurs. To prevent this deviation, inception voltage was measured first, followed by breakdowns for all the following measurements. For each gas/gas mixture, a new needle was used. The negative-inception voltage increases linearly as the gas pressure increases. An example of negative-inception voltage–current characteristics and its visual proof (UV vs. UV + VIS) at 8.8 bar abs is shown in Figure 6. The captures show that it contained combined ultraviolet and visible emitted light. The glow discharge is detectable across the pressure range (see Figure 7).

Figure 5 shows a deviation between breakdown and negative-inception voltages at 5 bar abs, corresponding to an electric field of 67 mkV/mm according to Equation (1). By increasing the pressure, deviation increased. Positive-current pulses were detectable for pressures above 6 bar abs (see Figure 5 and Figure 8).

The breakdown voltage of N₂ showed a linear dependence on increasing gas pressure in contrast to electronegative gas/gas mixtures that present nonlinear tendency for the same needle-plane configuration, such as CO₂ and 4% C₄F₇N–96% CO₂ [6]. The breakdown occurred exclusively under a negative AC voltage peak for the pressure range. According to the pre-breakdown current measurements (see Figure 8) and the optical observations (see Figure 3 and Figure 8), the negative glow streamer mode was identified as the breakdown mechanism for compressed N₂. These findings agree with previous investigations [10,11,12,13,14,15].

Figure 9 summarizes the negative and the positive-inception voltage and AC breakdown measurement for 12.77% O₂–87.23% CO₂ at various pressures up to 10 bar abs. The negative-inception voltage increased linearly with the increase of gas pressure. An example of negative and positive partial discharge voltage–current characteristics and its emission image (UV + VIS) at 8.8 bar abs are shown in Figure 10. The intensity of negative partial discharge current decreased with an increase of pressure (see Figure 11a). The breakdown characteristic of 12.77% O₂–87.23% CO₂ showed a nonlinear behavior when the gas pressure increased. The breakdown occurred under the negative AC voltage peak for pressures below 1.6 bar abs and on the positive AC voltage peak above. The polarity reversal (PR) cross point at 1.6 bar abs corresponds to a positive-inception electric field of 50 kV/mm, calculated according to Equation (1). At pressures below PR, only negative PDs were detectable up to breakdown. At PR pressure, similar partial discharge activities occurred on positive and negative cycles (see Figure 10a). The luminosity of the positive cycle was lower than that of the negative cycle (see Figure 10b,c). At pressures above PR [3,6], the needle tip was covered with a cloud of charge carriers, which were stabilized by negative ions formed by the attachment of electrons. Consequently, the negative space charges generate a reduction of the electric field on the side of the tip. This causes an increase in the breakdown voltage. For high pressure (8.8 bar abs), one positive-current pulse was detectable before the breakdown; the positive-streamer channels transited to leader channels and arrived at the opposite electrode instantly. The space charge near the needle tip resulted in a reduced electric field. The electrode was virtually elongated, the gap width was reduced, and the electric field was enhanced in the remaining gap, resulting in a reduction in breakdown voltage. It established an electrical conduction path between the two electrodes. Heating and expansion of the conductive channel occurred, and multiple branches were created (see Figure 11c). The negative partial discharge inception voltage (PDIV) was identified with current and emission light images. The negative partial discharge inception electric field (PDIEF) E_max was calculated according to Equation (1) for different natural gases as a function of pressure. The results are shown in Figure 12. Along the pressure range investigated, PDIEF of N₂ and 12.77% O₂–87.23% CO₂ exhibits a linear trend. On the other hand, PDIEF of CO₂ evolved linearly and presented a saturation trend at higher pressure levels (>5 abs). As a result, at the highest and lowest pressures (8.8 bar abs and 1 bar abs), PDIEF (12.77% O₂–87.23% CO₂) > PDIEF (CO₂) > PDIEF (N₂). However, at the intermediate pressures, PDIEF (CO₂) was higher than PDIEF (12.77% O₂–87.23% CO₂ and N₂) and PDIEF (N₂) = PDIEF (12.77% O₂–87.23% CO₂).

3.2. C₄F₇N Gas Based

C₄F₇N or Novec^TM 4710 was combined with a simple dilution natural gas/gas mixture (CO₂, O₂ and/or N₂). For quasi-homogeneous and divergent electric field distributions, the dielectric strength of the 20% C₄F₇N/80% CO₂ mixture was equivalent to that of pure SF₆ at equivalent pressure [7,8]. In practical applications, a reduction in the concentration of C₄F₇N was required to enable an increase in the filling pressure of the apparatus without liquefaction. Figure 13 summarizes the measured negative and the positive-inception and AC breakdown voltages for 20% C₄F₇N–80% CO₂ at various pressures up to 5 bar abs. The negative-inception voltage increased linearly with the increase of gas pressure. The breakdown characteristic of 20% C₄F₇N–80% CO₂ showed a nonlinear behavior over the gas pressure levels. The breakdown occurred under a negative AC voltage peak for pressure below 1.6 bar abs and on a positive AC voltage peak. The polarity reversal (PR) cross point at 1.6 bar abs corresponded to a positive-inception electric field of 84 kV/mm, calculated according to Equation (1). Below PR, negative and positive PDs were detectable until breakdown occurs. An example of negative and positive partial discharge voltage–current characteristics and its emission image (UV vs. UV + VIS) at 1.3 bar abs are shown in Figure 14.

Negative PDIV detection was possible via current measurement (current transformer 5-1) and via emission image. For high pressure (5 bar abs), multiple pulses around the positive cycle peak were detectable before the breakdown (see Figure 15). While negative partial discharge amplitude decreases with pressure increases, the intensity of positive partial discharge current increases. Increasing the applied voltage level induced denser current activity and consequently increased light emission (see Figure 14a and Figure 15a); in turn, this generated a sufficient critical space charge avalanche to create a discharge. The positive glow streamer mode was identified as the breakdown mechanism for compressed 20% C₄F₇N–80% CO₂. The findings agreed with previous investigations on a 4% C₄F₇N–96% CO₂ gas mixture [6]. The streamer criterion was proven to fit the breakdown values for different electric field distributions [16]. The critical space charge size was given by the natural logarithm of the number of electrons N_critical in the critical avalanche;

Table 3. Streamer criterion mechanism: critical breakdown space charge for different C₄F₇N based gas mixtures [17].

Gas/Gas Mixture	Ln (Ncritical)
3.7% C4F7N–96.3% CO2	6.228768
20% C4F7N–80% CO2	8.448
C4F7N	14.56

summarizes Ln (N_critical) for different C₄F₇N based gas mixtures. Increasing C₄F₇N content led to an increase in the breakdown critical space charge and a larger gap between positive PDIV and breakdown, leading to a higher rate of gas molecules decomposition.

To clarify the synergism between components of 20% C₄F₇N–80% CO₂ gas mixture, negative PDIV was identified for each pure component separately, i.e., C₄F₇N up to 1 bar abs (20% of the mixture), CO₂ up to 4 bar abs (80% of the mixture) and the final mixture of them up to 5 bar abs. The negative partial discharge inception electric field (PDIEF) E_max was calculated for each component according to Equation (1), reported to the final mixture pressure and plotted in Figure 16. For all the pressure ranges, PDIEF (C₄F₇N) was higher than PDIEF (20% C₄F₇N–80% CO₂), which in its turn was higher than PDIEF (CO₂). Synergism on PDIEF of 20% C₄F₇N–80% CO₂ gas mixture fits with the mean value of each individual component according to Equation (2). Based on PDIEF data of pure CO₂ (up to 8.8 bar abs, Figure 12) and pure C₄F₇N (up to 1 bar abs, Figure 15), three different variations with the increase of C₄F₇N content in a C₄F₇N–CO₂ gas mixture were identified.

Table 4. Negative partial discharge inception electric field (PDIEF) (E_max) comparison between CO₂ and C₄F₇N as a function of component concentration.

Case	C4F7N/CO2 Gas Mixture
	% C4F7N	% CO2
Emax (CO2) > Emax (C4F7N)	<9 ${\pm }_0^1$	>91 ${\pm }_1^0$
Emax (CO2) = Emax (C4F7N)	9 ${\pm }_0^1$	91 ${\pm }_1^0$
E max (CO2) < Emax (C4F7N)	>9 ${\pm }_0^1$	<91 ${\pm }_1^0$

summarizes the comparison between PDIEF (E_max) of CO₂ and C₄F₇N separately in C₄F₇N–CO₂ gas mixtures. At 9${\pm }_0^1\%$ C₄F₇N concentration, E_max (CO₂) = E_max (C₄F₇N). When C₄F₇N concentration < 9${\pm }_0^1$, Equation (2) was no longer applicable. Concerning polarity reversal, when the breakdown polarity changed from negative to positive half-cycle, the polarity reversal electric field E_PR was calculated according to Equation (1) for CO₂, SF_6, and different C₄F₇N–CO₂ gas mixtures and plotted in Figure 17. Fitting results show that E_PR (CO₂) = E_PR (9% C₄F₇N–91% CO₂) and E_PR(SF₆) = E_PR (22% C₄F₇N–78% CO₂). At the same time, the pressure of polarity reversal increased linearly with the increase of field utilization factor η. Increasing C₄F₇N content improved E_PR and polarity reversal pressure at the same time at a constant field utilization factor η (see

Table 5. Polarity reversal electric field E_PR according to Equation (1) (current work).

Gas/Gas Mixture	EPR (kV/mm)	Pressure (bar abs)	η
20% C4F7N–80% CO2	84	1.6	0.0385
4% C4F7N–96% CO2	50	0.7	0.0385

). This variation was governed by a constant E_PR intrinsic characteristic for each gas/gas mixture.

$${\mathrm{E}}_{\max }\left(20\%{\mathrm{C}}_4{\mathrm{F}}_7\mathrm{N}-80\%\mathrm{C}{\mathrm{O}}_2\right)=\frac{{\mathrm{E}}_{\max }\left({\mathrm{C}}_4{\mathrm{F}}_7\mathrm{N}-\mathrm{Partial}\mathrm{pressure}\right)+E_{\max }\left(\mathrm{C}{\mathrm{O}}_2-\mathrm{Partial}\mathrm{pressure}\right)}{2}$$

(2)

3.3. C₅F₁₀O Gas Based

The vaporization of C₅F₁₀O or Novec ^TM 5110 was obtained above 26.9 °C. C₅F₁₀O could only be used as an additive at a pressure below the saturated vapor pressure and could not be used in a pure state for insulation applications where the minimum operating temperature of an HV apparatus may drop to −25 °C or below. A mixture with a low concentration of a C₅F₁₀O (4–6%) and a buffer gas (O₂/N₂/CO₂) resulted in an improvement of the dielectric strength [9].

Figure 18 reports AC breakdown measurements for 12.77% O₂–87.23% CO₂, 6% C₅F₁₀O–12% O₂–82% CO₂, and 20% C₄F₇N–80% CO₂ at various pressures and for a gap distance of 5 mm. All the gas mixtures show nonlinear behavior as a function of pressure and with different peak values. At 8.8 bar abs, the breakdown voltage of 6% C₅F₁₀O–12% O₂–82% CO₂ was equal to that of 12.77% O₂–87.23% CO₂. A saturation trend of the breakdown voltage was observed for 12.77% O₂–87.23% CO₂ and 6% C₅F₁₀O–12% O₂–82% CO₂ above 8.8 bar abs, while the increase for 20% C₄F₇N–80% CO₂ was linear. At 5 bar abs, the breakdown voltage of 20% C₄F₇N–80% CO₂ was 2 times higher compared to that of 6% C₅F₁₀O–12% O₂–82% CO₂ at 8.8 bar abs.

The breakdown occurred under a negative AC voltage peak for pressures below 0.5 bar abs and on a positive AC voltage peak. The polarity reversal (PR) cross point at 0.5 bar abs corresponds to a positive-inception electric field of 36 kV/mm, calculated according to Equation (1). In terms of partial discharge activity, the 6% C₅F₁₀O–12% O₂–82% CO₂ exhibited a similar behavior compared with that of 12.77% O₂–87.23% CO₂ or 20% C₄F₇N–80% CO₂. Unlike 12.77% O₂–87.23% CO₂ and below PR, negative and positive PDs were detectable up to breakdown for the 6% C₅F₁₀O–12% O₂–82% CO₂. To explain the latter behavior, pure C₅F₁₀O at 0.54 bar abs, which corresponds to a 6% molar fraction of 9 bar abs total pressure, was evaluated separately, and the results are shown in Figure 19. High amplitude and positive polarity current pulses were detected during the positive cycle (see Figure 19a) before breakdown on the negative cycle (see Figure 19b), and a positive-inception electric field of 60 kV/mm was calculated according to Equation (1). On the other hand, C₄F₇N at 0.352 bar abs, which corresponds to a 4% molar fraction of 8.8 bar abs total pressure, exhibits similar behavior to C₅F₁₀O with less dense positive polarity discharge activity (see Figure 19c). The corresponding positive-inception electric field was 70 kV/mm calculated according to Equation (1). The breakdown voltage of the C₅F₁₀O (0.54 bar abs) was 1.09 times of C₄F₇N at 0.352 bar abs.

Above PR, the intensity of negative partial discharge current decreased with the increase of pressure. For high pressure (7 bar abs, see Figure 20), negative PDIV detection was not possible via current measurement (current transformer 5-1); however, the detection was performed via emission image. Before breakdown at high pressure (p = 6 bar abs), positive partial discharge appears in multiple pulsed within the positive AC half-cycle (see Figure 21a and Figure 22). Increasing the applied voltage levels induced more current activity and a brighter emission image. The breakdown on the positive cycle was led by a positive-streamer sufficient space charge. 6% The 6% C₅F₁₀O–12% O₂–82% CO₂ presents denser positive partial discharge activities before the breakdown compared to the 4% C₄F₇N–96% CO₂ [6].

Negative partial discharge inception electric field (PDIEF) E_max was calculated according to Equation (1) for pure C₅F₁₀O and C₄F₇N at sub-atmospheric pressures corresponding to the operational molar fraction without liquefaction in real indoor/outdoor high voltage applications and the results are shown in Figure 23. For this pressure range, PDIEF(C₄F₇N) was higher than PDIEF(C₅F₁₀O) and, at 0.6 bar abs, PDIEF(C₄F₇N) was 2.18 times higher than PDIEF(C₅F₁₀O). Figure 24 shows (PDIEF) E_max for 4% C₄F₇N–96% CO₂, 6% C₅F₁₀O–12% O₂–82% CO₂, and 20% C₄F₇N–80% CO₂ as a function of pressure. For all the gas mixtures, PDIEF was linear as a function of pressure, and PDIEF (6% C₅F₁₀O–12% O₂–82% CO₂) was higher than PDIEF (4% C₄F₇N–96% CO₂). Above 3 bar abs, PDIEF (20% C₄F₇N–80% CO₂)> PDIEF (4% C₄F₇N–96% CO₂ and 6% C₅F₁₀O–12% O₂–82% CO₂).

To study the synergism between components of 6% C₅F₁₀O–12% O₂–82% CO₂ gas mixture, negative PDIV was identified for each pure component separately, i.e., C₅F₁₀O up to 0.6 bar abs (6% of the mixture), 12.77% O₂–87.23% CO₂ up to 9.4 bar abs (94% of the mixture), and the final mixed gas up to 10 bar abs. The negative partial discharge inception electric field (PDIEF) E_max was calculated for each component according to Equation (1), reported to the final mixture pressure and plotted in Figure 25. For all the pressure ranges, PDIEF (6% C₅F₁₀O–12% O₂–82% CO₂) was higher than PDIEF (12.77% O₂–87.23% CO₂), which, in its turn, was higher than PDIEF (C₅F₁₀O). Synergism on PDIEF of 6% C₅F₁₀O–12% O₂–82% CO₂ gas mixture fits with the sum of each negative partial discharge inception electric field E_max individually component according to Equation (3).

Based on PDIEF data of pure 12.77% O₂–87.23% CO₂ (up to 9.4 bar abs) and pure C₅F₁₀O (up to 0.6 bar abs), three different variations with the increase of C₅F₁₀O content in a C₅F₁₀O–(12.77% O₂–87.23% CO₂) gas mixture were identified.

Table 6. PDIEF (E_max) comparison between 12.77% O₂–87.23% CO₂ and C₅F₁₀O as a function of component concentration.

Case	C5F10O/12.77% O2–87.23% CO2 Gas Mixture
	% C5F10O	%(12.77% O2–87.23% CO2)
Emax (12.77% O2–87.23% CO2) > Emax (C5F10O)	<19 ${\pm }_0^1$	>81 ${\pm }_1^0$
Emax (12.77% O2–87.23% CO2) = Emax (C5F10O)	19 ${\pm }_0^1$	81 ${\pm }_1^0$
Emax (12.77% O2–87.23% CO2) < Emax (C5F10O)	>19 ${\pm }_0^1$	<81 ${\pm }_1^0$

summarizes the comparison between PDIEF (E_max) of 12.77% O₂–87.23% CO₂ and C₅F₁₀O separately in C₅F₁₀O–(12.77% O₂–87.23% CO₂) gas mixtures. At 19${\pm }_0^1$ C₅F₁₀O concentration, E_max (12.77% O₂–87.23% CO₂) = E_max(C₅F₁₀O).

Regarding polarity reversal, when the breakdown polarity changes from negative to positive cycle, the polarity reversal electric field E_PR was calculated according to Equation (1) for CO₂, 12.77% O₂–87.23% CO_2, and 6% C₅F₁₀O–12% O₂–82% CO₂ gas mixtures (see

Table 7. Polarity reversal electric field E_PR according to Equation (1) for different gas mixtures-based CO_2.

Gas/Gas Mixture	EPR (kV/mm)	Pressure (bar abs)	η
CO2	64	2.5	0.0385
12.77% O2–87.23% CO2	50	1.6	0.0385
6% C5F10O–12% O2–82% CO2	36	0.5	0.0385

). At a constant field utilization factor η, decreasing the concentration of CO₂ reduces final E_PR and polarity reversal pressure. A linear trend was observed between polarity reversal pressure and final polarity reversal electric field E_PR. Comparing data in Table 5 and Table 7, it could be observed that E_PR (4% C₄F₇N–96% CO₂) = E_PR (12.77% O₂–87.23% CO₂) and E_PR (6% C₅F₁₀O–12% O₂–82% CO₂) was less than E_PR (4% C₄F₇N–96% CO₂).

$$\begin{array}{l}{\mathrm{E}}_{\max }\left(6\%{\mathrm{C}}_5{\mathrm{F}}_{10}\mathrm{O}-12\%{\mathrm{O}}_2-82\%\mathrm{C}{\mathrm{O}}_2\right)={\mathrm{E}}_{\max }\left({\mathrm{C}}_5{\mathrm{F}}_{10}\mathrm{O}-\mathrm{Partial}\mathrm{pressure}\right)+\\ {\mathrm{E}}_{\max }\left(12.77\%{\mathrm{O}}_2-87.23\%\mathrm{C}{\mathrm{O}}_2-\mathrm{Partial}\mathrm{pressure}\right)\end{array}$$

(3)

4. Discussion

Compared to CO₂, the addition of O₂ (12.77% O₂–87.23% CO₂) increased the critical reduced electric field strength from 82 Td to 90 Td (1 Td = 10⁻²¹ V.m²) with ≈10% improvement [18,19]. N₂-O₂ gas mixtures exhibited better critical reduced electric field strength than CO₂–O₂ for oxygen content higher than 8% (see Figure 26). At 15 Td, the mixture 16% O₂–84% CO₂ exhibited the same value of reduced attachment Townsend coefficient as pure O₂ [18]. Compared to N₂ or N₂-O₂ gas mixtures, CO₂ or CO₂–O₂ mixtures had better arc-breaking capabilities. Oxygen was selected to be part of the gas mixture (C₄F₇N/CO₂/O₂) to reduce the generation rate of both carbon monoxide (main decomposition product) and fluorinated byproducts during arc quenching [20].

The analysis of the effective ionization coefficients in C₅F₁₀O/O₂/CO₂ [21,22] and C₄F₇N/CO₂ [23] gas mixtures showed a strong dependence of the effective ionization coefficient on the gas density. C₅F₁₀O is characterized by an intrinsic dielectric strength highly dependent on the pressure, i.e., the increase of pressure increases the limiting dielectric strength. For a stationary value E/N of density reduced electric field (see Equation (4)), where:

• V: DC applied voltage;
• d: Insulation distance; and
• N: gas density, increasing the pressure decreases the applied voltage V, resulting in a decrease of the velocity and the energy of the emitted electrons.

The decrease of the energy of the swarm inhibits the generations in fractional negative ions result of dissociative attachment. The pure C₅F₁₀O and C₄F₇N exhibit PDs activities on the AC positive half-cycle when the breakdown occurs on the negative half-cycle (see Figure 19b,c). The pure C₅F₁₀O shows denser PDs activity than that of pure C₄F₇N. Moreover, for similar breakdown voltages, negative partial discharge inception electric field of C₄F₇N at 0.352 bar abs is 1.75 times higher than that of C₅F₁₀O at 0.54 bar abs, as shown in Figure 23. The positive PDs incept by dissociative attachment of electrons in C₅F₁₀O or C₄F₇N. Thynne and Harland [24] reported negative ions formed by low energy electron impact and attachment to C₄F₇N, using a Bendix time of flight spectrometer. Multiple dissociative attachment ions are formed (F⁻, CN⁻, CF₃⁻, C₂F₅⁻ and C₃F₇⁻) in the range of appearance potential 0.3–12 eV. The appearance potential is defined as the energy required to produce the ion and its accompanying neutral fragment from a given molecule. The attachment cross-section of C₄F₇N is continuous and higher than SF₆ in the energy range between 0.1 eV and 1 eV [25]. Hösl et al. [23] proved the existence of three types of anions (F⁻ or CN⁻ and/or C₄F₇N⁻) at low energies and developed a model, including the anions electron detachment to fit their experimental breakdown data. Ranković et al. [26] detected and identified the C₄F₇N fragments (F⁻, CN⁻, C₃F₄N⁻, C₄F₆N⁻) by the electron energy loss spectroscopy and the dissociative electron attachment spectroscopy created at energies around ~1 eV.

The attachment cross-section C₅F₁₀O [25] is similar to that of SF₆ for electron energies below 0.7 eV, and that it is substantially larger than that of SF₆ in the region between 0.7 eV and 10 eV. The formation of anions is governed by dissociative attachment to C₅F₁₀O. Dissociative electronic attachment leads to fragmentation of the C₅F₁₀O molecules; different chemical species appear in the gas phase. The species generated in the gas phase cause a destabilization of the molecular texture and a weakening of the dielectric behavior of gas or gas mixture.

$$\frac{E}{N}=\frac{V}{Nd}$$

(4)

The SF₆ presents different behaviors of the electron attachment cross-section [27] in the incident energy of the electron E_i between 0 and 10 eVas shown in Figure 27. Electrons with low-energy < 0.6 eV and low-speed are generally absorbed by SF₆ molecules to form SF₆⁻ anion in a stable state. For unstable SF₆⁻ anions, the electrons auto-detaches after 68 ± 0.2 μs [28]. Dissociative attachment of electrons in SF₆ is dependent on the incident energy of the electrons E_i. For E_i between 0.6 eV and 2 eV, SF₅⁻ anions are produced by fragmentation of the parent SF₆ molecule. For E_i between 2 eV and 3.5 eV, no dissociative attachment occurs. For E_i between 3.5 eV and 10 eV, SF₄⁻, SF₃⁻ and SF₂⁻ anions are produced. The previous dissociative attachment patterns in SF₆ result in multiple polarity reversal (at least 2) (+)/(−) and (−)/(+) as a function of the maximum electric field given by Equation (1) [7,8,14,29,30].

Only one polarity reversal happens in C₄F₇N/CO₂ mixtures and pure CO₂. The breakdown voltage in positive polarity becomes constraining and goes below the voltage in negative polarity when the pressure increases and/or the field tends to be divergent.

5. Conclusions

The breakdown characteristic as a function of pressure under non-uniform electric field distribution is linear for N₂ and strongly nonlinear for weak electronegative gas/gas mixture (CO₂, 12.77% O₂–87.23% CO₂) or strong electronegative gas/gas mixture (C₄F₇N, C₅F₁₀O, 4% C₄F₇N–96% CO2, 20% C₄F₇N–80% CO₂, 6% C₅F₁₀O–12% O₂–82% CO₂). The negative glow streamer mode is identified as the breakdown mechanism for compressed N₂.

For compressed 4% C₄F₇N–96% CO₂ and 6% C₅F₁₀O–12% O₂–82% CO₂ gas mixtures under non-uniform electric field distribution, the positive-streamer mode is identified as the breakdown mechanism. Breakdown and negative partial discharge inception voltages of 6% C₅F₁₀O–12% O₂–82% CO₂ are higher than those of 4% C₄F₇N–96% CO₂. For high pressure, negative PDIV detection is not possible via current measurement; the detection is feasible via emission image. At 8.8 bar abs, the breakdown voltage of 6% C₅F₁₀O–12% O₂–82% CO₂ is equal to that of 12.77% O₂–87.23% CO₂ (buffer gas). At 5 bar abs, the breakdown voltage of 20% C₄F₇N–80% CO₂ is 2 times higher compared with that of 6% C₅F₁₀O–12% O₂–82% CO₂ at 8.8 bar abs.

For 20% C₄F₇N–80% CO₂ gas mixture, synergism in negative partial discharge inception voltage/electric field fits with the mean value of each individual partial pressure component (C₄F₇N and CO₂). Based on partial discharge inception voltage/electric field data of pure CO₂ (up to 8.8 bar abs) and pure C₄F₇N (up to 1 bar abs), three different variations with the increase of C₄F₇N content in a C₄F₇N–CO₂ gas mixture was identified. At 9 ${\pm }_0^1$ C₄F₇N concentration, partial discharge inception electric field E_max (CO₂) = E_max(C₄F₇N). When C₄F₇N concentration < 9 ${\pm }_0^1$, the mean value of each partial pressure individual component (C₄F₇N and CO₂) is no longer applicable.

For 6% C₅F₁₀O–12% O₂–82% CO₂ gas mixture, synergism in negative partial discharge inception voltage/electric field (E_max) fits with the sum of each individual partial pressure component (C₅F₁₀O and 12.77% O₂–87.23% CO₂). Based on partial discharge inception voltage/electric field data of 12.77% O₂–87.23% CO₂ (up to 9.4 bar abs) and pure C₅F₁₀O (up to 0.6 bar abs), three different variations with the increase of C₅F₁₀O content in a C₅F₁₀O–(12.77% O₂–87.23% CO₂) gas mixture was identified. At 19 ${\pm }_0^1$ C₅F₁₀O concentration, E_max (12.77% O₂–87.23% CO₂) = E_max(C₅F₁₀O).

Polarity reversal occurs under AC voltage when the breakdown polarity changes from negative to positive cycle. Polarity reversal electric field E_PR was quantified. Fitting results show that E_PR (CO₂) = E_PR(9% C₄F₇N–91% CO₂) and E_PR(SF₆) = E_PR (22% C₄F₇N–78% CO₂). The pressure of polarity reversal increases linearly with the increasing field utilization factor η. At a constant field utilization factor η, increasing C₄F₇N content improves E_PR and polarity reversal pressure. This variation is ruled by a constant E_PR intrinsic characteristic for each gas/gas mixture. The addition of O₂ in 12.77% O₂–87.23% CO₂ and C₅F₁₀O in C₅F₁₀O–(12.77% O₂–87.23% CO₂) reduces final E_PR and polarity reversal pressure, E_PR (4% C₄F₇N–96% CO₂) = E_PR (12.77% O₂–87.23% CO₂) and E_PR (6% C₅F₁₀O–12% O₂–82% CO₂) < E_PR (4% C₄F₇N–96% CO₂) < E_PR (CO₂).