Study on the Compatibility of Gas Adsorbents Used in a New Insulating Gas Mixture C 4 F 7 N / CO 2

: An environment-friendly insulating gas, perﬂuoroisobutyronitrile (C 4 F 7 N), has been developed recent years. Due to its relatively high liquefaction temperature (around − 4.7 ◦ C), bu ﬀ er gases, such as CO 2 and N 2 , are usually mixed with C 4 F 7 N to increase the pressure of the ﬁlled insulating medium. During these processes, the insulating gases may be contaminated with micro-water, and the mixture of H 2 O with C 4 F 7 N could produce HF under breakdown voltage condition, which is harmful to the gas insulated electricity transfer equipment. Therefore, removal of H 2 O and HF in situ from the gas insulated electricity transfer equipment is signiﬁcant to its operation security. The adsorbents with the ability to remove H 2 O but without obvious C 4 F 7 N / CO 2 adsorption capacity are essential to be used in this system. In this work, a series of industrial adsorbents and desiccants were tested for their compatibility with C 4 F 7 N / CO 2 . Pulse adsorption tests were conducted to evaluate the adsorption performance of these adsorbents and desiccants on C 4 F 7 N and CO 2 . The 5A molecular sieve showed high adsorption of C 4 F 7 N (22.82 mL / g) and CO 2 (43.86 mL / g); F-03 did not show adsorption capacity with C 4 F 7 N, however, it adsorbed CO 2 (26.2 mL / g) clearly. Some other HF adsorbents, including NaF, CaF 2 , MgF 2 , Al(OH) 3 , and some desiccants including CaCl 2 , Na 2 SO 4 , MgSO 4 were tested for their compatibility with C 4 F 7 N and CO 2 , and they showed negligible adsorption capacity on C 4 F 7 N and CO 2 . The results suggested that these adsorbents used in the gas insulated electricity transfer equipment ﬁlled with SF 6 (mainly 5A and F-03 molecular sieves) are not suitable anymore. The results of this work suggest that it is a good strategy to use a mixture of desiccants and HF adsorbents as new adsorbents in the equipment ﬁlled with C 4 F 7 N / CO 2 .


Chemical Reagents
The chemical reagents used in this study, including Al(OH) 3 , CaCl 2 , MgSO 4 , Na 2 SO 4 , NaF, MgF 2 , CaF 2 and the zeolite molecular sieve materials 3A, 4A were purchased from Sinopharm Co. Ltd. The adsorbents, 5A and F-03 zeolite molecular sieves, were offered by Shandong Taikai High Voltage Switchgear Co. Ltd. All of the chemicals with analytical grade or adsorbents were dried in an oven at 120°C for 10 h to remove the moisture. Pure CO 2 (99.999%) used as a calibration gas was purchased from Xi'an Teda Cryogenic Equipment Co. Ltd.; and C 4 F 7 N was purchased from a commercial market with a purity of 99%. The chemical composition of the zeolite molecular sieves are listed in Table 1.

Adsorption Characterization
To study the adsorption performance of the selected chemicals and adsorbents toward C 4 F 7 N and CO 2 , pulse adsorption tests were conducted in chemical adsorption equipment (Builder PCA-1200, Beijing Builder Co. Ltd., Beijing, China). The schematic and picture of the pulse adsorption test is shown in Figure 1. When the pulse gas (tested gas) passed through the thermal conductivity detector (TCD), a pulse signal would show up, and the area of the signal peak is proportional to the amount of tested gas. Before testing the samples, the pulse adsorption procedure was run with an empty tube, and the obtained data were used as a blank control. To determine the adsorption performance of the samples, 0.05-0.20 g of each sample was filled in the sample tube, and the pulse adsorption procedures were conducted by turning a six-way valve to feed the calibration gas on a certain time interval. As shown in Figure 1A, for each test, in the first step, the quantitative loop was connected with the pulse gas line to fill with a fixed volume of pulse gas (0.30 mL), and in the six-way valve, gas passage 1 was connected with 6, while gas passage 4 was connected with 5. In the second step, the connection of the quantitative loop was switched to the sample tube, and in the six-way valve, gas passage 1 was connected with 2, and gas passage 4 was connected with 3. Then the carrier gas was purged and the pulse gas filled in the quantitative loop to pass through the sample and the TCD sequentially to record the pulse signal. Therefore, for each sample, 5-15 pulses were conducted depending on the adsorption performance of the testing sample, and the data obtained from the equipment were used to calibrate the adsorption capacity of the samples. For each sample, at least three tests were conducted, and the average data with less than 5% deviation were accepted.
Processes 2019, 7, 698 3 of 12 detector (TCD), a pulse signal would show up, and the area of the signal peak is proportional to the amount of tested gas. Before testing the samples, the pulse adsorption procedure was run with an empty tube, and the obtained data were used as a blank control. To determine the adsorption performance of the samples, 0.05-0.20 g of each sample was filled in the sample tube, and the pulse adsorption procedures were conducted by turning a six-way valve to feed the calibration gas on a certain time interval. As shown in Figure 1A, for each test, in the first step, the quantitative loop was connected with the pulse gas line to fill with a fixed volume of pulse gas (0.30 mL), and in the sixway valve, gas passage 1 was connected with 6, while gas passage 4 was connected with 5. In the second step, the connection of the quantitative loop was switched to the sample tube, and in the sixway valve, gas passage 1 was connected with 2, and gas passage 4 was connected with 3. Then the carrier gas was purged and the pulse gas filled in the quantitative loop to pass through the sample and the TCD sequentially to record the pulse signal. Therefore, for each sample, 5-15 pulses were conducted depending on the adsorption performance of the testing sample, and the data obtained from the equipment were used to calibrate the adsorption capacity of the samples. For each sample, at least three tests were conducted, and the average data with less than 5% deviation were accepted.

Data Analysis
The pulse signal data obtained were integrated to obtain the area of each peak. The area of each peak is proportional to the volume of calibration gas passed through the sample tube, and the difference of the area between the blank control was proportional to the amount of gas adsorbed by the samples. For each sample that adsorbed the target gas, several peaks with lower area than the control could be obtained, and the amount of gas adsorbed by the sample could be calculated according to the following equations where Ab is the average area of peaks obtained with empty tubes for each tested calibrating gas in blank test; and V represents volume of the quantitative loop in the six-way vale, which is 0.30 mL in this work. The item Ic stands for the area of peaks for one milliliter calibrating gas; Ai is the peaks with lower area compared with the control, when pulsing calibrating gas through sample in the tube. The item n represents that the number of peaks showed lower integrated area than that of the control peaks and m was the mass of the testing samples that filled the tube (in the unit of g). Vad represents the volume of calibrating gas adsorbed by the sample (mL/g).

Data Analysis
The pulse signal data obtained were integrated to obtain the area of each peak. The area of each peak is proportional to the volume of calibration gas passed through the sample tube, and the difference of the area between the blank control was proportional to the amount of gas adsorbed by the samples. For each sample that adsorbed the target gas, several peaks with lower area than the control could be obtained, and the amount of gas adsorbed by the sample could be calculated according to the following equations where A b is the average area of peaks obtained with empty tubes for each tested calibrating gas in blank test; and V represents volume of the quantitative loop in the six-way vale, which is 0.30 mL in this work. The item I c stands for the area of peaks for one milliliter calibrating gas; A i is the peaks with lower area compared with the control, when pulsing calibrating gas through sample in the tube. The item n represents that the number of peaks showed lower integrated area than that of the control peaks and m was the mass of the testing samples that filled the tube (in the unit of g). V ad represents the volume of calibrating gas adsorbed by the sample (mL/g).

Compatibility of Samples with C 4 F 7 N
Due to its relatively high boiling point, the content of C 4 F 7 N used in the mixture gas is usually no more than 20% [22,23]. Therefore, the adsorbents or desiccants used to remove the moisture or acidic by-products from C 4 F 7 N should not be able to adsorb C 4 F 7 N. Some of the moisture adsorbents, including 3A, 4A, 5A and F-03 molecular sieves, the desiccants including Na 2 SO 4 , CaCl 2 , and the HF adsorbents, including NaF, MgF 2 , Al(OH) 3 and CaF 2 , were tested for their adsorption capacities on C 4 F 7 N gas.
As shown in Figure 2, comparing with the pulse adsorption spectra using an empty tube as a control (Figure 2A), the 3A and 4A molecular sieves show slight adsorption capacity of C 4 F 7 N ( Figure 2B,C) with 0.39 and 1.44 mL/g, respectively as shown in Table 2. The 3A and 4A molecular sieves are usually used to dewater as they possess high surface areas and pore volumes [15], since the average pore sizes of 0.3 nm (for 3A molecular sieve) and 0.4 nm (for 4A molecular sieve) pore size are suitable to adsorb H 2 O molecules, however, it is calculated that the dynamic diameter for C 4 F 7 N is around 0.7599 nm [12], which is significantly larger than the pore sizes of 3A and 4A molecular sieves. The surface area in micropores contributed most of the surface area, therefore, C 4 F 7 N molecules are only able to adsorb on the surface of the 3A and 4A molecular sieves, which led to low adsorption capacity. With 5A molecular sieve, although its average pore diameter is 0.5 nm, there are significant pores with sizes larger than 0.5 nm, besides, on the axial direction of this molecule, the diameter of the CF 3 group is smaller than 0.5 nm (0.4896 nm) [24], and therefore more surface area could be reachable for C 4 F 7 N adsorption on 5A molecular sieve. As shown in Figure 2D, C 4 F 7 N shows significant adsorption on 5A molecular sieve. The pulse adsorption peaks shown in Figure 2D are trailing, which suggests that the interaction between 5A molecular sieve and C 4 F 7 N are strong. The adsorption capacity for C 4 F 7 N is 22.82 mL/g calculated according to Equations (1) and (2). As for the commonly used adsorbent F-03, it also shows slight adsorption of C 4 F 7 N as shown in Figure 2E, in which the intensity of the signal peaks is slightly lower than that of the blank control. usually used to dewater as they possess high surface areas and pore volumes [15], since the average pore sizes of 0.3 nm (for 3A molecular sieve) and 0.4 nm (for 4A molecular sieve) pore size are suitable to adsorb H2O molecules, however, it is calculated that the dynamic diameter for C4F7N is around 0.7599 nm [12], which is significantly larger than the pore sizes of 3A and 4A molecular sieves. The surface area in micropores contributed most of the surface area, therefore, C4F7N molecules are only able to adsorb on the surface of the 3A and 4A molecular sieves, which led to low adsorption capacity. Since the 5A molecular sieves could adsorb C 4 F 7 N, it is not suitable to use these materials to eliminate the moisture from C 4 F 7 N gas. One alternative strategy could be using the common desiccants, such as CaCl 2 , MgSO 4 , Na 2 SO 4 . These chemicals implement dewatering efficiently by forming crystal water. Since these chemicals possess low surface area, they should show negligible adsorption capacity of C 4 F 7 N. As shown in Figure 3, three desiccants, including CaCl 2 , MgSO 4 and Na 2 SO 4 , show negligible adsorption with C 4 F 7 N. The adsorption capacity data listed in Table 2 also show that these chemicals do not intend to adsorb C 4 F 7 N. Therefore, these three desiccants could be used for dewatering of C 4 F 7 N gas.
forming crystal water. Since these chemicals possess low surface area, they should show negligible adsorption capacity of C4F7N. As shown in Figure 3, three desiccants, including CaCl2, MgSO4 and Na2SO4, show negligible adsorption with C4F7N. The adsorption capacity data listed in Table 2 also show that these chemicals do not intend to adsorb C4F7N. Therefore, these three desiccants could be used for dewatering of C4F7N gas.  Some fluorides are good HF adsorbents, including NaF [17,25], MgF 2 and CaF 2 [26]. Due to the reactivity with HF, Al(OH) 3 has also proved to be good HF remover [27]. These chemicals are potential HF removers that could be placed in the gas insulated electricity transfer equipment filled with C 4 F 7 N gas. Therefore, the adsorption performances of these chemicals on C 4 F 7 N are significant data. The ideal situation of negligible adsorption with this gas was expected to be observed. The pulse adsorption data are shown in Figure 4. As shown in these patterns, NaF, CaF 2 and Al(OH) 3 show negligible adsorption of C 4 F 7 N, while MgF 2 shows clear interaction with C 4 F 7 N. The adsorption capacity data listed in Table 2 also support the conclusion. These data suggest that NaF, CaF 2 and Al(OH) 3 are compatible with C 4 F 7 N when used as a HF remover. Al(OH)3 are compatible with C4F7N when used as a HF remover.
A mixture of desiccant (CaCl2) and HF remover (Al(OH)3) was also tested for its compatibility with C4F7N gas. As shown in Figure 5, regardless if the mass ratio of desiccant to HF remover was 1 or 2, the mixture did not show clear adsorption performance on C4F7N. These data suggest that using a mixture of desiccant and HF remover to eliminate the moisture and HF could be a promising way to substitute the 5A or F-03 adsorbents.  A mixture of desiccant (CaCl 2 ) and HF remover (Al(OH) 3 ) was also tested for its compatibility with C 4 F 7 N gas. As shown in Figure 5, regardless if the mass ratio of desiccant to HF remover was 1 or 2, the mixture did not show clear adsorption performance on C 4 F 7 N. These data suggest that using a mixture of desiccant and HF remover to eliminate the moisture and HF could be a promising way to substitute the 5A or F-03 adsorbents. potential HF removers that could be placed in the gas insulated electricity transfer equipment filled with C4F7N gas. Therefore, the adsorption performances of these chemicals on C4F7N are significant data. The ideal situation of negligible adsorption with this gas was expected to be observed. The pulse adsorption data are shown in Figure 4. As shown in these patterns, NaF, CaF2 and Al(OH)3 show negligible adsorption of C4F7N, while MgF2 shows clear interaction with C4F7N. The adsorption capacity data listed in Table 2 also support the conclusion. These data suggest that NaF, CaF2 and Al(OH)3 are compatible with C4F7N when used as a HF remover.
A mixture of desiccant (CaCl2) and HF remover (Al(OH)3) was also tested for its compatibility with C4F7N gas. As shown in Figure 5, regardless if the mass ratio of desiccant to HF remover was 1 or 2, the mixture did not show clear adsorption performance on C4F7N. These data suggest that using a mixture of desiccant and HF remover to eliminate the moisture and HF could be a promising way to substitute the 5A or F-03 adsorbents.

Compatibility of Samples with CO 2
In C 4 F 7 N/CO 2 , the ratio of CO 2 could be more than 90% (v/v), therefore, to remove moisture and HF, the compatibility of the adsorbents with CO 2 is significant. Both of CO 2 and HF are acidic gases, and the reactivity of the adsorbents with CO 2 may compromise the efficiency for HF removal. In this work, the compatibility of the above tested molecular sieves, including 3A, 4A, 5A, F-03, the desiccants, such as CaCl 2 , MgSO 4 , Na 2 SO 4 , and HF remover, NaF, MgF 2 , CaF 2 and Al(OH) 3 were tested with pulse adsorption procedures to determine their adsorption performance or interaction with CO 2 .
As shown in Figure 6B,C, Figures 3A and 4A molecular sieves show slight adsorption of CO 2 compared with the blank control in Figure 6A, besides, the data listed in Table 3 show that the adsorption capacity is 0.8 mL/g and 3.13 mL/g, respectively. The 5A molecular sieve showed clear adsorption with CO 2 , as shown in Figure 6D, and this result is consistent with the previous study [28]. The peak intensity is lower than the blank control and they are trailing clearly, which suggests the CO 2 is strongly interacting with the 5A molecular sieve. The adsorption capacity listed in Table 3 is 43.66 mL/g. It is well known that 5A molecular sieve has high adsorption capacity of CO 2 [11,28]. The F-03 adsorbents also show a high CO 2 adsorption capacity, which is 26.2 mL/g as listed in Table 3. Therefore, 5A molecular sieve and F-03 are not compatible with C 4 F 7 N/CO 2 insulating gas.

Compatibility of Samples with CO2
In C4F7N/CO2, the ratio of CO2 could be more than 90% (v/v), therefore, to remove moisture and HF, the compatibility of the adsorbents with CO2 is significant. Both of CO2 and HF are acidic gases, and the reactivity of the adsorbents with CO2 may compromise the efficiency for HF removal. In this work, the compatibility of the above tested molecular sieves, including 3A, 4A, 5A, F-03, the desiccants, such as CaCl2, MgSO4, Na2SO4, and HF remover, NaF, MgF2, CaF2 and Al(OH)3 were tested with pulse adsorption procedures to determine their adsorption performance or interaction with CO2.
As shown in Figures 6B,C, 3A and 4A molecular sieves show slight adsorption of CO2 compared with the blank control in Figure 6A, besides, the data listed in Table 3 show that the adsorption capacity is 0.8 mL/g and 3.13 mL/g, respectively. The 5A molecular sieve showed clear adsorption with CO2, as shown in Figure 6D, and this result is consistent with the previous study [28]. The peak intensity is lower than the blank control and they are trailing clearly, which suggests the CO2 is strongly interacting with the 5A molecular sieve. The adsorption capacity listed in Table 3 is 43.66 mL/g. It is well known that 5A molecular sieve has high adsorption capacity of CO2 [11,28]. The F-03 adsorbents also show a high CO2 adsorption capacity, which is 26.2 mL/g as listed in Table 3. Therefore, 5A molecular sieve and F-03 are not compatible with C4F7N/CO2 insulating gas. Similar with the results tested in C4F7N, the three desiccants did not show clear adsorption with CO2, as shown in Figure 7, and the data listed in Table 3. The data also suggest the three chemicals would not react with CO2. Since no clear adsorption with C4F7N was observed, they could be used for removing the moisture in the insulating gas C4F7N/CO2.   Similar with the results tested in C 4 F 7 N, the three desiccants did not show clear adsorption with CO 2 , as shown in Figure 7, and the data listed in Table 3. The data also suggest the three chemicals would not react with CO 2 . Since no clear adsorption with C 4 F 7 N was observed, they could be used for removing the moisture in the insulating gas C 4 F 7 N/CO 2 . Similar with the results tested in C4F7N, the three desiccants did not show clear adsorption with as shown in Figure 7, and the data listed in Table 3. The data also suggest the three chemicals ld not react with CO2. Since no clear adsorption with C4F7N was observed, they could be used moving the moisture in the insulating gas C4F7N/CO2. All of the four HF removers are alkaline chemicals, one would suspect that these chemicals react with CO2. The pulse adsorption data presented in Figure 8 suggest that the four chemicals s negligible adsorption of CO2, and the adsorption capacity data listed in Table 3 are all below 0.5 m These data suggest that CO2 would not react with the four HF removers. The pKa of HF is 3.18 the pKa1 of H2CO3 is 6.38, therefore, the fluoride salts are stable in CO2 gas. Al(OH)3 is a weak a and it is also stable in CO2 gas. All of the four HF removers are alkaline chemicals, one would suspect that these chemicals may react with CO 2 . The pulse adsorption data presented in Figure 8 suggest that the four chemicals show negligible adsorption of CO 2 , and the adsorption capacity data listed in Table 3 are all below 0.5 mL/g. These data suggest that CO 2 would not react with the four HF removers. The pK a of HF is 3.18, and the pK a1 of H 2 CO 3 is 6.38, therefore, the fluoride salts are stable in CO 2 gas. Al(OH) 3 is a weak alkali, and it is also stable in CO 2 gas. All of the four HF removers are alkaline chemicals, one would suspect that these chemicals may react with CO2. The pulse adsorption data presented in Figure 8 suggest that the four chemicals show negligible adsorption of CO2, and the adsorption capacity data listed in Table 3 are all below 0.5 mL/g. These data suggest that CO2 would not react with the four HF removers. The pKa of HF is 3.18, and the pKa1 of H2CO3 is 6.38, therefore, the fluoride salts are stable in CO2 gas. Al(OH)3 is a weak alkali, and it is also stable in CO2 gas. Since both the desiccants and HF remover studied in this work did not show clear reaction or adsorption with CO2, logically, the mixture of a desiccant and HF remover should also not adsorb or react with CO2. The data shown in Figure 9 and Table 3 prove that the mixture of CaCl2 and Al(OH)3 are compatible in CO2, which is the same result as tested in C4F7N. Therefore, the mixture of desiccants with HF remover could be used in C4F7N/CO2. Since both the desiccants and HF remover studied in this work did not show clear reaction or adsorption with CO 2 , logically, the mixture of a desiccant and HF remover should also not adsorb or react with CO 2 . The data shown in Figure 9 and Table 3 prove that the mixture of CaCl 2 and Al(OH) 3 are compatible in CO 2 , which is the same result as tested in C 4 F 7 N. Therefore, the mixture of desiccants with HF remover could be used in C 4 F 7 N/CO 2 .

Conclusions
The pulse adsorption tests suggested that the commonly used adsorbents 5A and F-03 molecular sieves could not be used in C 4 F 7 N/CO 2 , due to the severe adsorption of the mixed gas on these molecular sieves. The 3A and 4A molecular sieves adsorb C 4 F 7 N and CO 2 slightly, and might be used as adsorbents for C 4 F 7 N/CO 2 . Desiccants, including Na 2 SO 4 , CaCl 2 and MgSO 4 show negligible adsorption with C 4 F 7 N and CO 2 . Some HF removers, such as NaF, CaF 2 , Al(OH) 3 also show negligible adsorption with the two gases, and could be compatible with them sealed in related gas insulated electricity transfer equipment. Using a mixture of desiccant and HF remover could be a good strategy to remove the moisture and HF produced in the C 4 F 7 N/CO 2 insulated equipment.