Analysis of Gaseous By-Products of CF3I and CF3I-CO2 after High Voltage Arcing Using a GCMS

Increasing demand for an alternative insulation medium to sulphur hexafluoride (SF6) has led to the investigation of new environmentally friendly insulation gases which could be used in high voltage equipment on the electrical power network. One such alternative, which is currently being explored by researchers, is Trifluoroiodomethane (CF3I) which could potentially be used in a gas mixture with carbon dioxide (CO2) as an insulation medium. In this paper an analysis of gaseous by-products detected as a result of high voltage breakdown through pure CF3I and a CF3I-CO2 gas mixture across a sphere-sphere electrode arrangement is given. Gas chromatography and mass spectrometry (GCMS) is used to identify the gaseous by-products produced as a result of high voltage arcing which causes the gas between the electrodes to dissociate. Analysing these gas by-products helps to identify the long-term behaviour of the gas mixture in high voltage equipment.


Introduction
In the power industry, sulphur hexafluoride (SF 6 ) is increasingly being used as a high voltage insulation medium in gas insulated lines (GIL) for power transmission purposes and in gas insulated switchgear (GIS) as an insulating and arc interrupting medium [1]. However, the use of SF 6 is receiving a lot of scrutiny from the international community as a gas of extremely high global warming potential (GWP) which is estimated to be 23,500 times that of CO 2 [2] when released into the atmosphere. SF 6 is an extremely damaging global warming gas with a long atmospheric lifetime of 3200 years [3].
At present the international scientific community is searching for an alternative insulation medium which has all the necessary qualities required for use in the power industry, similar to SF 6 , without the damaging environmental effects. At present the candidates being proposed include: Fluoronitrile [4], Fluoroketones [5], Hydrofluoroolefins HFO's [6] and Trifluoroiodomethane (CF 3 I) [7][8][9] all of which exhibit a lower global warming potential than SF 6 and a reduced atmospheric lifetime. However, there are still concerns regarding their toxicity and much research is still needed to evaluate their characteristics in full [10][11][12]. In the power industry, at high voltages of 11 kV and above, SF 6 gas insulated equipment is used in GIS and GIL. Long term partial arcing or full HV breakdowns to ground, in this gas insulated equipment, release large amounts of energy through the insulation gas, which leads to the gas dissociating. This irrevocable gas dissociation can affect the overall insulation strength and operational withstand level of the HV equipment. As such, these events are of real concern to the long-term operation of future alternatively insulated gas equipment. It is important to understand the effect this may have on the gas being used to insulate equipment and whether the gaseous by-products will affect the ability of the equipment to operate safely or pose future health or environmental problems. In this paper the by-products produced from the use of a CF 3 I-CO 2 gas

Results of Pure CF3I Gas By-product Analysis
Next a pure gas sample of CF3I is analysed using the GCMS to compare gas samples both before and after 100 breakdowns across a sphere-sphere gas gap. It can be noted that even after 15 different flushing cycles of the test arrangement and gas sampling system with Helium that residual CO2 is still detected by the GCMS due to its level of sensitivity, which is necessary because of the amount of by-products produced. It can also be noted that CHF3, C2HF5 and H2O are also present in the pure gas before any breakdown takes place, indicating that these slight impurities exist in the original gas

Results of Pure CF 3 I Gas By-Product Analysis
Next a pure gas sample of CF 3 I is analysed using the GCMS to compare gas samples both before and after 100 breakdowns across a sphere-sphere gas gap. It can be noted that even after 15 different flushing cycles of the test arrangement and gas sampling system with Helium that residual CO 2 is still detected by the GCMS due to its level of sensitivity, which is necessary because of the amount of by-products produced. It can also be noted that CHF 3 , C 2 HF 5 and H 2 O are also present in the pure gas before any breakdown takes place, indicating that these slight impurities exist in the original gas sample or that the gas is reacting with one or more of the storage/test vessel materials. H 2 O likely exists in small quantities due to the fact that the level of vacuum used, which is the normal level for gas insulated power equipment, is insufficient to remove all of the H 2 O from the pressure vessel materials.
Following 100 breakdown events it can be shown in Figure 2 that the original gas impurities/ by-products remain along with newly created gaseous by-products, detected only after the 100 breakdown events. These detected by-products of pure CF 3 I include: CF 4 , C 2 F 6 , C 2 F 4 , C 3 F 8 , C 3 F 6 or C 4 F 8 , C 2 F 5 I, C 3 F 7 I and C 3 F 7 IO. It is likely that C 3 F 7 IO is a by-product caused by the interaction of CF 3 I and the very small quantities of CO 2 that remain or the H 2 O content which is also present before the breakdown sample is taken. It can be noted that because the molecular weight of iodine is much heavier and larger than the other elements that make up the molecule CF 3 I, that all large molecular by-products that contain iodine have a longer retention time from this column than CF 3 I and the by-products that do not contain iodine have a shorter retention time along this column and are detected near the front end of each run. It is also possible that there are by-products which have the same retention time in this column as CF 3 I and therefore are not able to be distinguished from CF 3 I. It can be shown in Figure 2 that the large amount of CF 3 I in the gas sample takes approx. 2 min to pass the detector.
It is also possible to show that most of these by-products detected after 100 breakdowns are similar to the by-products detected in pure CF 3 I when subjected to PD events as described by M. kamarol, Y. Nakayama, T. Hara, S. Ohtsuka and M. Hikita in reference [13]. This means that the breakdown event method used in this paper and the partial discharge method used in reference [13] are comparable methods of determining both the long term and short-term arcing by-products of CF 3 I. sample or that the gas is reacting with one or more of the storage/test vessel materials. H2O likely exists in small quantities due to the fact that the level of vacuum used, which is the normal level for gas insulated power equipment, is insufficient to remove all of the H2O from the pressure vessel materials. Following 100 breakdown events it can be shown in Figure 2 that the original gas impurities / by-products remain along with newly created gaseous by-products, detected only after the 100 breakdown events. These detected by-products of pure CF3I include: CF4, C2F6, C2F4, C3F8, C3F6 or C4F8, C2F5I, C3F7I and C3F7IO. It is likely that C3F7IO is a by-product caused by the interaction of CF3I and the very small quantities of CO2 that remain or the H2O content which is also present before the breakdown sample is taken. It can be noted that because the molecular weight of iodine is much heavier and larger than the other elements that make up the molecule CF3I, that all large molecular by-products that contain iodine have a longer retention time from this column than CF3I and the byproducts that do not contain iodine have a shorter retention time along this column and are detected near the front end of each run. It is also possible that there are by-products which have the same retention time in this column as CF3I and therefore are not able to be distinguished from CF3I. It can be shown in Figure 2 that the large amount of CF3I in the gas sample takes approx. 2 min to pass the detector.

Results of 30:70% CF 3 I-CO 2 Gas By-Product Analysis
Following the tests carried out with pure CF 3 I, a pressure-pressure gas mixture ratio of 30% CF 3 I and 70% CO 2 was examined to show the by-products produced by this gas mixture using the same sphere-sphere test electrode arrangement. It can be shown that, before any breakdown event takes place, the 70% CO 2 is detected in the gas sample as well as a CO component. Before breakdown, the 30% CF 3 I component of the gas mixture is detected along with a long 2 min retention time, as well as components such as CHF 3 and C 2 HF 5 . A small amount of H 2 O is also detected before any breakdown event is undertaken, presumable retained during vacuum from the containing vessel wall materials.
After 100 breakdown events have been carried out across the sphere-sphere electrode gas gap containing 30:70% CF 3 I-CO 2 , a gas sample was taken and the results were analysed using the GCMS as shown in Figure 3. The results show that C 2 F 6 O 3 -Trioxide, bis(trifluoromethyl), C 2 F 6 -Ethane, hexafluoro-, C 3 F 8 -Perfluoropropane and CF 4 -Tetrafluoromethane were detected as by-products from the 30:70% CF 3 I-CO 2 gas mixture. It is also likely that C 2 H 3 F 3 -Ethane, 1,1,1-trifluoro-and C 2 F 5 I-Pentafluoroethyliodide or C 2 F 9 I-Tetrafluoro(pentafluoroethyl)iodine could also be by-products but these are more difficult to confirm. It should also be noted that C 2 HF 5 -Ethane, pentafluoro-and CHF 3 -Fluoroform were also detected before and after breakdown of this CF 3 I-CO 2 gas mixture was carried out so they could be potential by-products, however, it is difficult to confirm they are produced as part of the arcing breakdown event and are not merely a by-product of gas interaction with containment materials. More analysis of this result, when compared to pure CF 3 I, is given in the next section of this paper.

Results of 30:70% CF3I-CO2 Gas By-product Analysis
Following the tests carried out with pure CF3I, a pressure-pressure gas mixture ratio of 30% CF3I and 70% CO2 was examined to show the by-products produced by this gas mixture using the same sphere-sphere test electrode arrangement. It can be shown that, before any breakdown event takes place, the 70% CO2 is detected in the gas sample as well as a CO component. Before breakdown, the 30% CF3I component of the gas mixture is detected along with a long 2 min retention time, as well as components such as CHF3 and C2HF5. A small amount of H2O is also detected before any breakdown event is undertaken, presumable retained during vacuum from the containing vessel wall materials.

Discussion of Pure CF 3 I and 30:70% CF 3 I-CO 2 Gas By-Product Comparison
In order to determine which by-products are produced after 100 breakdown events, it is important to examine the pure gases separately and compare these results to the gas mixture result. To identify which by-products are a result of the interaction between the two molecules in the gas mixture, it is useful to compare the analysis of all of these gas samples together i.e. both pure and mixtures. Figure 4 shows the GCMS analysis from both the pure CF 3 I and the 30%:70% CF 3 I-CO 2 gas mixture results. As pure CO 2 does not show any pertinent distinguishable results it was not necessary to include it in this discussion. The molecules and by-products detected during the GCMS analysis carried out in Figure 4 are shown in Table 1.

Discussion of Pure CF3I and 30:70% CF3I-CO2 Gas By-product Comparison
In order to determine which by-products are produced after 100 breakdown events, it is important to examine the pure gases separately and compare these results to the gas mixture result. To identify which by-products are a result of the interaction between the two molecules in the gas mixture, it is useful to compare the analysis of all of these gas samples together i.e. both pure and mixtures. Figure 4 shows the GCMS analysis from both the pure CF3I and the 30%:70% CF3I-CO2 gas mixture results. As pure CO2 does not show any pertinent distinguishable results it was not necessary to include it in this discussion. The molecules and by-products detected during the GCMS analysis carried out in Figure 4 are shown in Table 1. . GCMS analysis of the gaseous by-products of a pure CF3I and a 30%:70% CF3I-CO2 gas mixture, red line-pure CF3I following 100 high voltage AC breakdowns, black line-30%:70% CF3I-CO2 gas sample following 100 high voltage AC breakdowns. (Blue numbers-detected molecule in both CF3I and CF3I-CO2 breakdown tests, Red numbers-detected molecules in pure CF3I breakdown test only, Black numbered molecules-detected in CF3I-CO2 breakdown test only) (Analysis of molecule numbers shown in Table 1). Table 1. GCMS analysis of the gaseous by-products of pure CF3I and a 30%:70% CF3I-CO2 gas mixture.   . GCMS analysis of the gaseous by-products of a pure CF 3 I and a 30%:70% CF 3 I-CO 2 gas mixture, red line-pure CF 3 I following 100 high voltage AC breakdowns, black line-30%:70% CF 3 I-CO 2 gas sample following 100 high voltage AC breakdowns. (Blue numbers-detected molecule in both CF 3 I and CF 3 I-CO 2 breakdown tests, Red numbers-detected molecules in pure CF 3 I breakdown test only, Black numbered molecules-detected in CF 3 I-CO 2 breakdown test only) (Analysis of molecule numbers shown in Table 1).

Byproduct
Using the analysis conducted with both pure CF 3 I and CF 3 I-CO 2 it is possible to identify the by-products that are produced by the CF 3 I component and the alternative/extra by-products that are produced when it is used in a gas mixture with CO 2 .
It can be shown that prior to the breakdown tests being carried out, in both the pure CF 3 I and CF 3 I-CO 2 gas samples, that both CHF 3 and C 2 HF 3 are detected. This means that these are either impurities in the original gas sample or are a result of interaction with storage/test vessel materials. It is also possible that CHF 3 and C 2 HF 3 could be produced as a result of the breakdown arcing tests carried out, however, it is not possible to distinguish these since they are present in the pre-breakdown gas analysis sample.
From both the pure CF 3 I and CF 3 I-CO 2 breakdown tests it can be shown that CF 4 , C 2 F 6 , C 3 F 8 and C 2 F 5 I are produced as gaseous by-products. As these by-products are produced in both tests, this means it is most likely that these by-products are a result of the CF 3 I molecules in the gas mixture only and are not due to any interaction with the CO 2 in the gas mixture during the arcing event. It is also possible to surmise from this that because the CF 3 I is only 30% of the partial pressure mixture with CO 2 that the amount of these gaseous by-products will reduce as the amount of CF 3 I gas molecules involved in the arcing event is reduced, however, more research is needed to conclusively prove this. Table 1. GCMS analysis of the gaseous by-products of pure CF 3 I and a 30%:70% CF 3 I-CO 2 gas mixture.

By-product
Labelled Number in Figure 4 CF 3 I and CF 3 I-CO 2 Gas Sample-Detected before Breakdown

CF 3 I and CF 3 I-CO 2 Gas
Sample-By-products Detected after Breakdown CF 3 I Gas Sample only-By-products Detected after Breakdown CF 3 I-CO 2 Gas Sample only-By-products Detected after Breakdown In the pure CF 3 I breakdown tests, the gas analysis showed additional by-products of C 2 F 4 , C 3 F 6 , C 3 F 7 I, C 3 F 7 IO and potentially C 4 F 8 which were not identified in the CF 3 I-CO 2 gas mixture breakdown tests. This means that these by-products are either not produced with a mixture of CF 3 I and CO 2 or their production is reduced in quantity so that they are not detectable by the GCMS when CF 3 I is mixed with CO 2 in a partial pressure mixture of 30% or less after 100 breakdown events.
In the GCMS analysis of the CF 3 I-CO 2 gas mixture after breakdown it is identified that additional by-products of C 2 F 6 O 3 and potentially C 2 H 3 F 3 and C 2 F 9 I are produced which are not detected in the analysis of pure CF 3 I after breakdown. These by-products are produced as a direct result of the interaction between CF 3 I and CO 2 when used as a mixture which is subjected to arcing events of this energy level and are not produced when both CF 3 I or CO 2 are used as an insulation/interruption medium which undergoes separate arcing events.

Discussion of 30:70% CF 3 I-CO 2 GCMS Column Comparison
A brief comparison for this paper was also conducted using two different GCMS columns.
In Figure 5a the plot shows the gas chromatogram analysed by using a SiliPlot Column (Agilent, Amstelveen, Netherlands). In Figure 5b the plot shows the gas chromatogram analysed using a GS-GasPro Column (Agilent, Santa Clara, CA, USA). Although these two separate gas samples were analysed using different columns under relatively similar breakdown conditions to a 30%-70% CF 3 I-CO 2 gas mixture some conclusions can be drawn about the use of these columns under this testing regime.
It can be shown from Figure 5 that at the front end of the run light gases such as CO and CO 2 are easier to distinguish using a GS-GasPro column as the separation of the column allows for individual identification which is not possible in the SilicaPlot column. It can also be shown that a gas with elution time similar to that of CF 3 I such as CF 4 (No. 9. In Figure 5b) might also be harder to distinguish because these gases elute together and due to the high concentration of CF 3 I it is harder to distinguish CF 4 at this retention time. It was therefore chosen to carry out all of the tests in the above sections using the GCMS with the GS-GasPro column.
their production is reduced in quantity so that they are not detectable by the GCMS when CF3I is mixed with CO2 in a partial pressure mixture of 30% or less after 100 breakdown events.
In the GCMS analysis of the CF3I-CO2 gas mixture after breakdown it is identified that additional by-products of C2F6O3 and potentially C2H3F3 and C2F9I are produced which are not detected in the analysis of pure CF3I after breakdown. These by-products are produced as a direct result of the interaction between CF3I and CO2 when used as a mixture which is subjected to arcing events of this energy level and are not produced when both CF3I or CO2 are used as an insulation/interruption medium which undergoes separate arcing events.

Discussion of 30:70% CF3I-CO2 GCMS Column Comparison
A brief comparison for this paper was also conducted using two different GCMS columns. In Figure 5a the plot shows the gas chromatogram analysed by using a SiliPlot Column (Agilent, Amstelveen, Netherlands). In Figure 5b the plot shows the gas chromatogram analysed using a GS-GasPro Column (Agilent, Santa Clara, CA, USA). Although these two separate gas samples were analysed using different columns under relatively similar breakdown conditions to a 30%-70% CF3I-CO2 gas mixture some conclusions can be drawn about the use of these columns under this testing regime.
It can be shown from Figure 5 that at the front end of the run light gases such as CO and CO2 are easier to distinguish using a GS-GasPro column as the separation of the column allows for individual identification which is not possible in the SilicaPlot column. It can also be shown that a gas with elution time similar to that of CF3I such as CF4 (No. 9. In Figure 5b) might also be harder to distinguish because these gases elute together and due to the high concentration of CF3I it is harder to distinguish CF4 at this retention time. It was therefore chosen to carry out all of the tests in the above sections using the GCMS with the GS-GasPro column.  C 2 F 6 O 3 -Trioxide, bis(trifluoromethyl) or CF 4 -Tetrafluoromethane, 4. C 2 F 6 -Ethane, hexafluoro-, 6. C 2 F 6 O 3 -Trioxide, bis(trifluoromethyl), 7. C 3 F 8 -Perfluoropropane, 9. CF 4 -Tetrafluoromethane or C 2 F 6 O 3 -Trioxide, bis(trifluoromethyl) or C 2 H 3 F 3 -Ethane, 1,1,1-trifluoro-, 11. C 2 F 5 I-Pentafluoroethyliodide or C 2 F 9 I-Tetrafluoro(pentafluoroethyl)iodine.

Materials and Methods
In this paper the test electrode setup was fixed as shown in Figure 6. In this setup the electrodes were of sphere-sphere geometry with a diameter of 25 mm and manufactured from stainless steel 304 which is composed of the primarily alloys: iron (Fe), Chromium (Cr) and Nickel (Ni) and various other trace elements [14]. Stainless steel 304 has a melting point of 1450 • C [14] so can only play a role in the decomposition products if the temperature in the arc reaches above this. These sphere electrodes are insulated using pure CO 2 , CF 3 I gas or a 30%:70% CF 3 I-CO 2 gas mixture at a total pressure of 1.2 bar (g) which is contained within a pressurised tube and filled using stainless steel gas connectors. This insulated tube is connected to a high voltage bushing which is air insulated on one side and insulated by CO 2 at 2 bar (g) on the inside. This CO 2 , which insulates the bushing, also surrounds the pressurised tube vessel containing the gas/gas mixture under test therefore keeping the test gas contained under a high pressure differential. This differential pressure eliminates any potential leakage as any leakage is more likely to occur from the CO 2 vessel inwards rather than the CF 3 I gas tube outwards. The gases used have a purity level of >99.9% for CF 3 I [15] and 99.8% for CO 2 [16]. The vessel containing CF 3 I is electrically isolated from the main vessel. The ground electrode for the CF 3 I vessel is connected via a dedicated grounding earth connection which is isolated from the outside pressure vessel walls, therefore ensuring the voltage and current measurements are separate from any external effects on the main vessel.
The gas mixture 30%-70% CF 3 I-CO 2 was chosen to be analysed based on extensive research of the various characteristics of this gas mixture including electrical. In reference [9] the electrical characteristics of 10%-90%, 20%-80% and 30%-70% are discussed and it was found that 30%-70% had the best electrical performance. Another factor that greatly impacts the decision to use this gas mixture is the boiling point of this mixture which are limited by the restraints and specifications for indoor/outdoor high voltage electrical switchgear which states the operating temperatures of the equipment in the UK/Europe. 30%-70% CF 3 I-CO 2 has a boiling point of −12.5 • C at a pressure of 0.5 MPa [17], while a higher percentage concentration of CF 3 I in the gas mixture would fall outside the guided operating temperatures for gas insulated switchgear specifications at this common operating pressure.

Materials and Methods
In this paper the test electrode setup was fixed as shown in Figure 6. In this setup the electrodes were of sphere-sphere geometry with a diameter of 25 mm and manufactured from stainless steel 304 which is composed of the primarily alloys: iron (Fe), Chromium (Cr) and Nickel (Ni) and various other trace elements [14]. Stainless steel 304 has a melting point of 1450 °C [14] so can only play a role in the decomposition products if the temperature in the arc reaches above this. These sphere electrodes are insulated using pure CO2, CF3I gas or a 30%:70% CF3I-CO2 gas mixture at a total pressure of 1.2 bar (g) which is contained within a pressurised tube and filled using stainless steel gas connectors. This insulated tube is connected to a high voltage bushing which is air insulated on one side and insulated by CO2 at 2 bar (g) on the inside. This CO2, which insulates the bushing, also surrounds the pressurised tube vessel containing the gas/gas mixture under test therefore keeping the test gas contained under a high pressure differential. This differential pressure eliminates any potential leakage as any leakage is more likely to occur from the CO2 vessel inwards rather than the CF3I gas tube outwards. The gases used have a purity level of >99.9% for CF3I [15] and 99.8% for CO2 [16]. The vessel containing CF3I is electrically isolated from the main vessel. The ground electrode for the CF3I vessel is connected via a dedicated grounding earth connection which is isolated from the outside pressure vessel walls, therefore ensuring the voltage and current measurements are separate from any external effects on the main vessel.
The gas mixture 30%-70% CF3I-CO2 was chosen to be analysed based on extensive research of the various characteristics of this gas mixture including electrical. In reference [9] the electrical characteristics of 10%-90%, 20%-80% and 30%-70% are discussed and it was found that 30%-70% had the best electrical performance. Another factor that greatly impacts the decision to use this gas mixture is the boiling point of this mixture which are limited by the restraints and specifications for indoor/outdoor high voltage electrical switchgear which states the operating temperatures of the equipment in the UK/Europe. 30%-70% CF3I-CO2 has a boiling point of −12.5°C at a pressure of 0.5 MPa [17], while a higher percentage concentration of CF3I in the gas mixture would fall outside the guided operating temperatures for gas insulated switchgear specifications at this common operating pressure. The high voltage bushing of the gas pressure vessel is connected to a 150 kV rms AC transformer (Ferranti, Hollinwood, England) and a capacitive divider (Ferranti, Hollinwood, England) with a ratio of 6736:1 which allows for voltage measurements to be taken as shown in Figure 7. The current was measured upon breakdown of the gas gap using a current transformer (CT) (Stangenes Industries Inc, Palo Alto, CA, USA) which was connected around the grounding strap connected to the ground electrode. Each breakdown across the tested gas gap produces a separate voltage and current waveform which is recorded using an oscilloscope (Teledyne LeCroy, Glasgow, UK) The high voltage bushing of the gas pressure vessel is connected to a 150 kV rms AC transformer (Ferranti, Hollinwood, England) and a capacitive divider (Ferranti, Hollinwood, England) with a ratio of 6736:1 which allows for voltage measurements to be taken as shown in Figure 7. The current was measured upon breakdown of the gas gap using a current transformer (CT) (Stangenes Industries Inc, Palo Alto, CA, USA) which was connected around the grounding strap connected to the ground electrode. Each breakdown across the tested gas gap produces a separate voltage and current waveform which is recorded using an oscilloscope (Teledyne LeCroy, Glasgow, UK) triggered by the rise in current on the grounded electrode, an example of this is shown in Figure 8. For each event recorded the voltage on the output of the transformer is slowly increased until a breakdown occurs at which point it is lowered until the arc is self-extinguished by the gas. A constant measurement of the voltage applied to the test gas gap is also recorded using a secondary oscilloscope (Teledyne LeCroy), which helps determine a breakdown or no-breakdown scenario. When a breakdown occurs, the output voltage of the transformer drops because it has an earth fault allowing the flow of current to ground, under no-breakdown conditions the current is zero and the voltage is a steady oscillating 50 Hz AC waveform.
Molecules 2019, 24, x 9 of 16 triggered by the rise in current on the grounded electrode, an example of this is shown in Figure 8. For each event recorded the voltage on the output of the transformer is slowly increased until a breakdown occurs at which point it is lowered until the arc is self-extinguished by the gas. A constant measurement of the voltage applied to the test gas gap is also recorded using a secondary oscilloscope (Teledyne LeCroy), which helps determine a breakdown or no-breakdown scenario. When a breakdown occurs, the output voltage of the transformer drops because it has an earth fault allowing the flow of current to ground, under no-breakdown conditions the current is zero and the voltage is a steady oscillating 50 Hz AC waveform.  After filling the pressurised gas vessel under test to 1.2 bar (g) a sample of the gas or gas mixture is taken using a stainless steel sampling tube. This is then connected to the GCMS which allows for different molecules to separate through the gas chromatography column and its mass spectrum to be analysed using the MS. The gas in the pressurised vessel is not replaced and therefore every time a sample is taken the gas pressure and density between the sphere gas gap is lowered, therefore, reducing the overall high voltage insulation strength. An initial sample of each gas/gas mixture is analysed using the GCMS to confirm its identity and identify whether any contamination exists triggered by the rise in current on the grounded electrode, an example of this is shown in Figure 8. For each event recorded the voltage on the output of the transformer is slowly increased until a breakdown occurs at which point it is lowered until the arc is self-extinguished by the gas. A constant measurement of the voltage applied to the test gas gap is also recorded using a secondary oscilloscope (Teledyne LeCroy), which helps determine a breakdown or no-breakdown scenario. When a breakdown occurs, the output voltage of the transformer drops because it has an earth fault allowing the flow of current to ground, under no-breakdown conditions the current is zero and the voltage is a steady oscillating 50 Hz AC waveform.  After filling the pressurised gas vessel under test to 1.2 bar (g) a sample of the gas or gas mixture is taken using a stainless steel sampling tube. This is then connected to the GCMS which allows for different molecules to separate through the gas chromatography column and its mass spectrum to be analysed using the MS. The gas in the pressurised vessel is not replaced and therefore every time a sample is taken the gas pressure and density between the sphere gas gap is lowered, therefore, reducing the overall high voltage insulation strength. An initial sample of each gas/gas mixture is analysed using the GCMS to confirm its identity and identify whether any contamination exists After filling the pressurised gas vessel under test to 1.2 bar (g) a sample of the gas or gas mixture is taken using a stainless steel sampling tube. This is then connected to the GCMS which allows for different molecules to separate through the gas chromatography column and its mass spectrum to be analysed using the MS. The gas in the pressurised vessel is not replaced and therefore every time a sample is taken the gas pressure and density between the sphere gas gap is lowered, therefore, reducing the overall high voltage insulation strength. An initial sample of each gas/gas mixture is analysed using the GCMS to confirm its identity and identify whether any contamination exists before any high voltage arcing takes place, an example of this is shown in Figures 1-3. After the initial sample is taken the gas sampling section is connected to the pressure vessel for the duration of the high voltage tests and only removed to allow the gas sample to be introduced to the GCMS. For each gas mixture the gas sampler and GCMS are flushed using a carrier gas of pure Helium which is 99.999% pure [18], therefore ensuring no-cross contamination takes place. During each run the GC oven is kept at a constant temperature of 50 • C for 5 min, the temperature is then ramped at 15 • C/min for 8 min and then kept at a constant temperature of 170 • C for a final 5 min leading to a total run time of 18 min. The GC is running the sample through for analysis in split mode with a ratio of 10:1. The MS is scanning for a mass between 5-350 with a step size of 0.1 m/z for the entire run.
For each test the vessel was flushed and vacuumed (<1 mbar [19]) and re-filled with CO 2 five times before the final sample was introduced to remove cross contamination from the connecting hoses/vessel etc. For the pure CF 3 I test the vessel was flushed using helium and vacuumed fifteen times before the final sample of CF 3 I was introduced into the system. After each group of tests containing CF 3 I the gas/gas mixture was first recovered to a separate cylinder using a gas recovery system, thereby ensuring that the amount of CF 3 I released is minimal before the system was flushed. The experiments were conducted in the following order: pure CO 2 , 30%:70% CF 3 I-CO 2 and then pure CF 3 I to ensure as few solid by-products were left inside the vessel which could affect the next set of results. The initial sample was taken following filling but preceding any high voltage breakdown. The sampling tube was then re-attached to the vessel and 50 breakdowns across the gas sample were conducted. The sample was then connected to the GCMS where the gas was analysed, this procedure was then repeated once 100 breakdown events had been undertaken when a final gas sample was taken before the gas/gas mixture was recovered into a separate cylinder.

Conclusions
This paper experimentally evidences the use of CF 3 I and CF 3 I-CO 2 gas mixtures in high voltage gas insulated equipment as an alternative insulation medium and examines the by-products produced when AC breakdown events occur through these gases.
The experimental gas samples analysed using a GCMS show that in pure CF 3 I the by-products from arcing events are: CF 4 , C 2 F 6 , C 2 F 4 , C 3 F 8 , C 3 F 6 or C 4 F 8 , C 2 F 5 I, C 3 F 7 I and C 3 F 7 IO with potential by-products of CHF 3 and C 2 HF 5 . In a 30:70% CF 3 I-CO2 gas mixture the gas samples analysed showed by-products of: C 2 F 6 O 3 , C 2 F 6 , C 3 F 8 and CF 4 with potential by-products of C 2 H 3 F 3 , C 2 F 5 I or C 2 F 9 I, C 2 HF 5 and CHF 3 .
It can be shown that prior to the breakdown tests being carried out, in both the pure CF 3 I and CF 3 I-CO 2 gas samples, that both CHF 3 and C 2 HF 3 are detected so these are potential by-products, but it is difficult to conclusively prove this from these results. From both the pure CF 3 I and CF 3 I-CO 2 breakdown tests it can be shown that CF 4 , C 2 F 6 , C 3 F 8 and C 2 F 5 I are produced as gaseous by-products. In the pure CF 3 I breakdown tests, the gas analysis showed additional by-products of C 2 F 4 , C 3 F 6 , C 3 F 7 I, C 3 F 7 IO and potentially C 4 F 8 which were not identified in the CF 3 I-CO 2 gas mixture breakdown tests. In the GCMS analysis of the CF 3 I-CO 2 gas mixture after breakdown it is identified that additional by-products of C 2 F 6 O 3 and potentially C 2 H 3 F 3 and C 2 F 9 I are produced which are not detected in the analysis of pure CF 3 I after breakdown, these by-products are produced as a direct result of the interaction between CF 3 I and CO 2 . All potential by-products of CF 3 I and CF 3 I-CO 2 and their associated hazards and exposure controls are detailed more extensively in Table A1.
It is necessary to point out that if CF 3 I reacted with any of the materials used in the filling valve, hose, pressure vessel wall or gas connections/sampling port during its 10 hour filling time that these by-products may be included in these results. It is also important to note that iodine is produced in a semi-solid state from arcing events in CF 3 I gas mixtures as shown in reference [20].
Future work will involve further analysis into the gaseous by-products of CF 3 I gas mixture with other potential buffer gases.

Conflicts of Interest:
The authors declare no conflict of interest. Table A1. Potential By-Products of CF 3 I and CF 3 I-CO 2 gas mixtures and their associated hazards and exposure controls. Note: data retrieved was correct at time of writing 19/03/2019, however, it is subject to frequent change and re-evaluation.   [46]