Wide Carbon Nanopores as Efficient Sites for the Separation of SF6 from N2

SF6 and SF6-N2 mixed gases are used widely as insulators, but such gases have high greenhouse gas potential. The separation of SF6 from SF6-N2 mixed gases is an inevitable result of their use. Single-walled carbon nanohorns (CNHs) were used here for a fundamental study of the separation of SF6 and N2. The diameters of the interstitial and internal nanopores of the CNHs were 0.7 and 2.9 nm, respectively. The high selectivity of SF6 over N2 was observed only in the low-pressure regime in the interstitial 0.7 nm nanopores; the selectively was significantly decreased at higher pressures. In contrast, the high selectivity was maintained over the entire pressure range in the internal 2.9-nm nanopores. These results showed that the wide carbon nanopores were efficient for the separation of SF6 from the mixed gas.

SF 6 is used as an insulator for high voltage breakers and circuit breakers in electrical equipment, because of its low-dielectric properties, low toxicity, and high stability [1][2][3] . Despite these advantages, SF 6 has a global warming potential that is 24000 times larger than that of CO 2 . The restriction of SF 6 emissions is therefore necessary to allow the use of SF 6 in applications in a variety of fields [4][5][6] . The mixing of SF 6 with N 2 can reduce SF 6 emissions, and can save costs in industrial applications. SF 6 -N 2 mixed gases retain insulating properties, even with low SF 6 contents (SF 6 :N 2 = 1:9) 7,8 . Mixed SF 6 -N 2 gases have therefore been widely applied, and the purification and recovery of SF 6 have become more important. The separation of gases is typically achieved using liquefaction, adsorption, or membrane separation techniques. In liquefaction techniques using refrigeration, it is possible to separate a target molecule by choosing an appropriate temperature near the boiling temperature. However, the separation of SF 6 via liquefaction is costly, because the system must be maintained at 209 K (the boiling temperature of SF 6 ). Liquefaction techniques that use pressurization also have high energy costs; the liquefaction of pure and mixed SF 6 gases (SF 6 :N 2 = 1:9) at 293 K was observed above 2 and 20 MPa, respectively 8,9 . The hydration of SF 6 and N 2 was also conducted under high pressures 10 . Motivated by these high costs, adsorption 8,11-15 and membrane separation 9,16-18 have been proposed as low-energy-cost techniques, and the adsorption technique shows promise for applications in a variety of fields. SF 6 was preferentially adsorbed over N 2 in porous media, because of its strong adsorption potential; in contrast, SF 6 was rarely adsorbed in extremely narrow nanopores 8,[11][12][13][14][15] . SF 6 adsorption was rarely observed in zeolites with pores with a diameter of 0.5 nm, but N 2 was adsorbed easily in such narrow nanopores 8 ; in contrast, zeolites with pores with a diameter of 1.0 nm selectively adsorbed SF 6 molecules over N 2 molecules 11 . Metal organic frameworks, which have uniform and narrow nanopores, also show highly selective adsorption [12][13] . The pore size dependence of the SF 6 -and-N 2 separation abilities of mesoporous silica and zeolite-templated carbons was evaluated by Builes et al., using grand canonical Monte Carlo simulations 14 ; 1.1 nm-nanopores separated SF 6 and N 2 well. Those preceding studies demonstrated the SF 6 separation efficiency of various porous media having narrow pores, whereas in pores larger than 2.0 nm, the SF 6 and N 2 separation has been neglected. The mechanisms responsible for the adsorption and separation of SF 6 also have not yet been sufficiently clarified. Thus, narrow and wide pores have been simply considered as having the high separation ability in small pore volumes and low separation ability in large pore volumes, respectively. Porous media having the high separation ability and large pore volumes are ideal materials for application of the separation. We here propose to use wide nanopores for the purpose, because of having relatively high adsorption potentials. In this paper, nanopores are defined as narrower pores than 5 nm.
Nanoporous carbons are composed only of carbon and have simple geometries, and they are therefore useful for examining the above-mentioned adsorption and separation mechanisms. Single walled carbon nanohorns (CNHs) have a tubular structure similar to that of carbon nanotubes [19][20][21][22] . The internal and interstitial sites of CNH particles have cylindrical nanopores with diameters of 2.9 and 0.7 nm, respectively, and the adsorption in the two sets of nanopores can be evaluated separately 21,22 . Single-walled carbons also have the potential to adsorb large amounts of SF 6 . Thus, CNHs have the advantages of being a highly efficient separation medium, and allowing the evaluation of the mechanisms responsible for the adsorption and separation of SF 6 . Here, the selective adsorption of SF 6 over N 2 in the internal and interstitial nanopores of CNHs was evaluated by measuring the adsorption of SF 6 and N 2 at 273 K. The mechanisms responsible for the adsorption and separation of SF 6 and N 2 were also investigated in this study.

Results and Discussion
Both the internal and interstitial nanopores of the partially oxidized CNHs were available for the adsorption of molecules, whereas the internal nanopores of the as-grown CNHs were closed to adsorbed molecules; thus, molecules could be adsorbed in the interstitial nanopores, as reported elsewhere 23,24 . The internal nanopores were assessed by measuring the difference between the amounts of N 2 adsorbed for the partially oxidized CNHs and the as-grown CNHs. The micropore volumes of the interstitial and internal nanopores of the CNHs were determined from the N 2 adsorption isotherms measured at 77 K on the as-grown and partially oxidized CNHs, as shown in Fig. 1a. The nanopore volumes of the as-grown and partially oxidized CNHs were 0.13 and 0.56 mL g −1 , respectively, as obtained from the Dubinin-Radushkevich equation 25 . Thus, the nanopore volumes of the interstitial and internal nanopores of the CNHs were 0.13 and 0.43 mL g −1 , respectively. The adsorption isotherms and porosities of the CNHs agreed with results from previous studies [20][21][22][23][24]26 . The nanopore size distributions in Fig. 1b were obtained using the Barrett-Joyner-Halenda theory 27 . The interstitial and internal nanopore sizes of the CNHs were distributed mainly in the ranges of < 1.2 nm and 1-4 nm, respectively. Thus, the interstitial and internal nanopores were named as narrow and wide nanopores, respectively. The X-ray photoelectron spectroscopies in Fig. 1c and transmission electron microscopic images in Fig. 1d indicated that geometrical and chemical structures were rarely changed by the partial oxidation of as-grown CNHs.
The O/C ratios were approximately 5% and those CNHs had less surface oxygen groups. Thus, we discussed the change of selectivity by nanopore size. Figure 1e,f shows N 2 and SF 6 adsorption isotherms measured for the CNHs at 273 K. The N 2 adsorption isotherms measured at 273 K were Henry-type, and the adsorbed amounts were small. In contrast, SF 6 was adsorbed well in these nanopores even at 1 atm, because of the strong intermolecular interactions of the SF 6 , corresponding to P/P 0 = 0.07. Because the amounts of SF 6 adsorbed in the interstitial and internal nanopores were significantly different, the SF 6 was less densely packed in the interstitial, narrow nanopores with a diameter of 0.7 nm.
The adsorption density results for SF 6 and N 2 in the interstitial and internal nanopores (shown in Fig. 2a) clearly showed that the SF 6 adsorption density was significantly higher in the internal nanopores than in the interstitial nanopores, whereas the N 2 adsorption density in the interstitial nanopores was higher than in the internal nanopores, as reported elsewhere 28 . Here, the adsorption densities were obtained from the number of adsorbed molecules and the micropore volume. The slightly higher density of N 2 in the interstitial nanopores was a result of the stronger adsorption potential in the interstitial nanopores, compared with that in the internal nanopores. The significantly lower density of SF 6 in the interstitial nanopores was a result of the size restriction imposed by the narrow nanopores, which had a diameter of 0.7 nm. Thus, the adsorption density was controlled by two factors: the adsorption potential, and the steric restriction. These factors led to the highly selective adsorption of SF 6 and N 2 . The selectivity for SF 6 over N 2 was defined by a theoretical ideal adsorption expression, as follows 29 : Here, x i and y i are the molar fractions of molecule i in the adsorbed and bulk phases, respectively. This theoretical ideal adsorption expression has been adopted previously to evaluate the selectivity of binary mixture gas adsorption at relatively low pressures in various porous media 12,[30][31][32][33][34] . The selectivity for the adsorption of SF 6 over N 2 in the interstitial and internal nanopores of the CNHs (shown in Fig. 2b,c) was determined from the adsorption densities shown in Fig. 2a. The selectivity is a measure of the molecular sieving ability 8,[35][36][37] . The selectivity in the interstitial nanopores decreased exponentially from 60 to 10 at y SF6 :y N2 = 0.1:0.9, which is a preferable molar fraction for industrial applications 7,8 . For the interstitial nanopores, the selectivity was consistently the greatest at y SF6 :y N2 = 0.1:0.9, for all of the molar fractions analyzed in this study. In the internal nanopores, the selectivity was also consistently the greatest at y SF6 :y N2 = 0.1:0.9; however, the decreases in the selectivity with increasing pressure were more moderate than those observed in the interstitial nanopores. The exponential decrease of the selectivity observed in the interstitial nanopores was a result of the restriction of the high-density SF 6 adsorption by the narrow nanopores. At very low pressures, the strong intermolecular potential of SF 6 promoted the adsorption of SF 6 in the nanopores, but N 2 was rarely adsorbed. In the internal nanopores with a diameter of 2.9 nm (i.e., in wide nanopores), SF 6 was adsorbed without any imposition of a size restriction by the nanopores. Thus, the strong adsorption potential of SF 6 resulted in high selectivity over the entire range of pressures. High selectivity was therefore achieved via the negative contribution of the steric restriction produced by the narrow interstitial nanopores alone, and the positive contribution of the strong adsorption potential of SF 6 in both the narrow interstitial and wide internal nanopores. At y SF6 :y N2 = 0.1:0.9, the selectivity was 44 in the internal nanopores at 1 atm, which was equivalent to the greatest selectivity value determined in previous studies, as shown in Table 1 [11][12][13] . Thus, the results of this study indicated that the wide nanopores with large pore volumes provided the greatest selectivity-yielding values similar to the greatest values achieved previously in extremely narrow nanopores, under ambient conditions-although grand canonical Monte Carlo simulations suggested that wide nanopores rarely show separation abilities at high pressures 14 .
In this study, the SF 6 and N 2 separation abilities of nanoporous carbons were evaluated using adsorption isotherms measured for the narrow interstitial and wide internal nanopores of CNHs (considered as narrow and wide nanopores, respectively). The narrow nanopores of the CNHs had an average diameter of 0.7 nm, and showed high SF 6 separation abilities in the low-pressure regime; however, the selectivity decreased rapidly with increasing pressure. In contrast, the selectivity in the wide nanopores, which had an average diameter of 2.9 nm, was maintained over the entire range of pressures. These results showed that the wide nanopores functioned well as sites for the separation of SF 6 .

Methods
CNHs were prepared by the Iijima and Yudasaka groups 19 . The internal nanopores were made accessible using partial oxidation, which was performed at 673 K for 1 h, in an atmosphere of flowing O 2 gas (using a flow rate of 100 mL min −1 ). N 2 adsorption isotherms were measured at 77 K and 273 K, and SF 6 adsorption isotherms were measured at 273 K; in both cases the adsorption isotherms were measured for CNHs and open-CNHs using a volumetric apparatus (Autosorb-1, Quantachrome Co., Florida, USA), after heating at 423 K for more than 2 h, at pressures below 10 mPa. Dubinin-Radushkevich analysis was conducted for the above adsorption isotherms, to evaluate the micropore volumes. The micropore volumes were evaluated using a Dubinin-Radushkevich analysis range of [ln (P/P 0 )] 2 = 20-60, which is typically applied for the analysis of micropores. X-ray photoelectron microscopy (Mg Kα radiation at 10 kV and 10 mA; JPS-9010MX, JEOL Co., Tokyo, Japan) was used to assess the surface oxygen groups in CNHs. Transmission electron microscopy at 120 keV (JEM-2100F, JEOL Co., Tokyo, Japan) was used for direct observations of CNHs.