1 Tailoring the structure of a sub-nano silica network via fluorine doping to enhance CO2

2 separation and evaluating CO2 separation performance under dry or wet conditions

3 Ikram Rana, Hiroki Nagasawa, Toshinori Tsuru, and Masakoto Kanezashi*

4 Chemical Engineering Program, Graduate School of Advanced Science and Engineering,

5 Hiroshima University, Higashi-Hiroshima, 739-8527, Japan

6 *Corresponding Author: M. Kanezashi, E-mail: kanezashi@hiroshima-u.ac.jp

8 Abstract

9 The amorphous nature of a silica network structure offers an attractive opportunity

10 for excellent separation of small molecules such as He and H2. CO2 separation systems
11 are restricted, however, by the density of amorphous silica structures. Hence, the
12 objective of this work was to tailor the network pore size of conventional
13 tetraethoxysilane (TEOS)-derived silica membranes via fluorine (NH4F) doping.
14 Fluorine doping can precisely enlarge the network pore size in a subnano range by
15 reducing the Si-OH groups in the network structure, while CO2 adsorption properties
16 are decreased as doped fluorine concentration increases. A fluorine-doped silica
17 membrane (F/Si=0.1/9.9) showed a CO2/N2 permeance ratio that reached as high as 50
18 with CO2 permeance of 1.6 × 10-7 mol m-2 s-1 Pa-1 at 50 ⁰C. In a wet system (200-300 oC,
19 H2O partial pressure of 3 kPa), fluorine-doped silica membranes were quite stable,
20 irrespective of the doped fluorine concentration, but a blocking effect at around 100-150
21 ⁰C was observed and gas permeance (He, CO2, N2) was apparently lower than that in a
22 dry system. Under steam with a high partial pressure (300 ⁰C, H2O partial pressure of 30
23 kPa), the hydrothermal stability of silica membranes was increased with increases in the
24 fluorine concentration.

25

26 Keywords: Amorphous silica; Fluorine doping; Network pore size tuning; CO2
27 separation; Hydrothermal stability

Electronic copy available at: https://ssrn.com/abstract=4091364

28 1. Introduction

29 The surge in CO2 emissions is creating immense environmental issues, and this
30 impurity gas (CO2) must be removed from mixtures of light gases such as CO2/H2,
31 CO2/N2, and CO2/CH4. Thus far, processes for the separation/capture of CO2 from
32 mixtures have depended on absorption, pressure swing absorption, and membrane
33 separation. Among all these alternatives, the membrane-based CO2 separation/capture
34 process is more advantageous for reasons that include membrane mechanical simplicity,
35 energy efficiency, easy fabrication, and suitability to small- and medium-scale
36 separation procedures [1-8]. Hence, various types of membranes such as polymeric [9],
37 carbon molecular sieve [10], zeolite [11, 12], graphene oxide [13], metal-organic
38 frameworks (MOFs) [14], and silica [15-17] have been designed for CO2 separation.
39 In industrial processing, however, membranes selective for CO2 must possess high
40 levels of chemical and thermal stability under severe conditions of elevated
41 temperatures and pressures. For example, polymeric membranes have shown CO2/N2
42 selectivity of 40 and CO2/CH4 selectivity approximating 100, but the weak
43 physicochemical properties of polymer structures restrict their applications [18]. In this
44 context, since the 1980s silica structures have garnered a considerable amount of
45 interest due to the promise of robust physicochemical stability (up to 800 ⁰C).
46 More specifically, commonly used tetraethoxysilane (TEOS) is considered one of
47 the most promising Si precursors for the fabrication of silica membranes via chemical
48 vapor deposition (CVD) and sol-gel methods [19]. For example, silica membranes
49 prepared via sol-gel methods have shown H2 permeance of 5.5 × 10-7 mol m-2 s-1 Pa-1
50 with a H2/N2 permeation ratio of more than 100 at 750 ⁰C [20]. This excellent
51 membrane performance, however, is associated with the sol-gel derived silica network
52 pore structure (average pore size: 0.34 nm) that is suitable for permeation by small
53 molecules (He, H2), but is not appropriate for larger gases such as CO2 (0.33 nm) and
54 N2 (0.36 nm) [21, 22]. In addition, conventional silica membranes fail to maintain their
55 stability against an atmosphere (moist) whereby water molecules may interact with the
56 membrane surface and generate Si-OH groups and an excessive amount of silanol
57 density (Si-OH). The generation of Si-OH in the silica structure leads to structural
58 rearrangements (densification) and simultaneously to an enlargement of grain

Electronic copy available at: https://ssrn.com/abstract=4091364

59 boundaries. The hydrothermal instability of conventional silica degrades membrane
60 performance [23].
61 Hence, a reduction in the Si-OH density to improve the hydrophobic properties of
62 the silica structure is considered an effective approach with the dual benefits of
63 improved hydrothermal stability and the ability to tune the network pore size. For
64 instance, the introduction of hydrophobic functional organic groups coupled with two Si
65 atoms (Si-R-Si) effectively controlled the silica network pore size and allowed H2
66 permeance of a sol-gel-derived organosilica 1,2-bis(triethoxysilyl)methane (BTESM,
67 Si-CH2-Si) membrane on the order of 10-6 mol m-2 s-1 Pa-1, which is one order of
68 magnitude higher than those of pure silica (10-7 mol m-2 s-1 Pa-1) membranes [24]. These
69 organic units shield the Si atoms in dry systems and prevent the regeneration of silanol
70 groups that is followed by the decomposition of siloxane due to interactions with water.
71 Over the past few years, several studies have evaluated the hydrothermal stability of
72 organosilica 1,2-bis(triethoxysilyl)ethane (BTESE, Si-C2H4-Si) membranes in PV
73 applications/humid gas separation and confirmed that organosilica membranes show
74 superior stability [25,26]. In a similar manner, Igi et al. [27] developed metal-doped
75 silica membranes and observed that small crystals of cations may be dispersed into the
76 silica matrix, which stabilizes the network structure over a time period of 60 h and
77 promises H2 permeance on the order of 10-7 mol m-2 s-1 Pa-1 with H2/N2 selectivity of
78 (250-270). Kanezashi et al. [28] reported the achievement of hydrothermal stability in
79 Ni-doped silica membranes with stable He permeance (10-6 mol m-2 s-1 Pa-1) even under
80 very high levels of steam partial pressure and temperature (500 ⁰C, partial pressure of
81 steam: 90-400 kPa).
82 Another substitute for the above-mentioned strategies is to induce Si-F bonds into
83 the silica matrix via the addition of NH4F into silica sols. During the sol-gel
84 polymerization process, highly reactive fluorine resulted in the Si-F bonds formation,
85 which replaced the silanol groups [19]. The formation of Si-F bonds in the silica matrix
86 affected the Si-O-Si bond angle, and consequently loose network structure was formed.
87 For instance, a tetraethoxyfluorosilane (TEFS) membrane showed H2 permeance of 10-6
88 mol m-2 s-1 Pa-1, which is superior to conventional silica membranes, and was quite
89 stable under a steam atmosphere (500 oC, partial pressure of steam: 30-90 kPa), because

Electronic copy available at: https://ssrn.com/abstract=4091364

90 a high concentration of Si-F bonds decreases silanol density (Si-OH) density which
91 results in improved hydrothermal stability as well as permeation properties of silica
92 membranes [29]. However, F-doped SiO2 membranes have never been applied to a
93 CO2/N2 separation system.
94 Therefore, a strategy for anion (NH4F) doping into a silica matrix was applied to
95 fabricate highly permeable membranes for CO2/N2 separation. NH4F and TEOS were
96 used as doping agent and Si precursor, respectively. The physiochemical structure of
97 fluorine-doped silica gels/films was characterized via Fourier transform infrared
98 spectroscopy (FT-IR), N2 and CO2 adsorption/desorption isotherms, XPS spectra, and
99 XRD patterns. H2O adsorption isotherms were carried out for
100 hydrophobicity/hydrophilicity analysis. Furthermore, activation energies (Ep) of gas
101 permeation were used to evaluate the effect that fluorine exerts on the pore structure of
102 a silica network. In subsequent experiments, the CO2/N2 separation performance of all
103 prepared membranes was compared based on Ep (CO2) and Ep (N2) to further examine
104 the CO2 separation potential of fluorine-doped silica membranes. Finally, hydrothermal
105 stability of fluorine-silica membranes was carried out in a humid environment.
106
107 2. Experimental section

108 2.1 Sol/gel preparation

109 A sol-gel method was employed to prepare fluorine doped and undoped silica sols
110 via hydrolysis/polymerization process [30-37] and final composition ratio (mole) of
111 TEOS:H2O:HNO3 (1:200:0.1) was maintained. To the ethanol solution, we
112 simultaneously added water, a catalyst (HNO3), and a Si precursor under vigorous
113 stirring at 500 rpm for 40-60 minutes at room temperature. The final precursor
114 concentration was fixed with weight % of 0.5, and the fluorine concentrations (F/Si)
115 were 1-20 mol%. Under an air atmosphere, slow drying process was employed to
116 prepare the F-doped silica gel at the temperature of 40ºC and then gel was ground in a
117 mortar. For analysis (XPS, XRD, N2, H2O/CO2 adsorption, XPS), the powder was
118 calcined at 350 °C under an air atmosphere.
119
120 2.2 Characterization of F-doped silica sols/gels
4

Electronic copy available at: https://ssrn.com/abstract=4091364

121 A DLS analyzer (Malvan Zetasizer Nano ZS) was used to measure the F-doped
122 silica sol sizes. KBr plate was used to evaluate the Fourier transform infrared (FT-IR)
123 spectra of F-doped and undoped silica samples in the range of 400-4,000 cm-1 using a
124 FT-IR spectrometer (FT/IR-4100, Jasco, Japan). To examine the fluorine status in the
125 silica matrix, X-ray photoelectron spectroscopy (XPS) (XPS, Shimadzu, Japan) analyses
126 were performed and each sample was vacuumed prior to measurement. Adsorption
127 behavior of all gels was assessed by utilizing a BELMAX (BELJAPAN INC.) and prior
128 to start the measurements, all samples were pretreated for 12 h at 200 ºC to remove the
129 adsorbed water. The measurement temperature of N2 adsorption isotherms was fixed at
130 77 K for samples calcined at 350 ºC. Similarly, H2O and CO2 adsorption measurements
131 were carried out at ambient temperature (25 ºC) to evaluate the
132 hydrophobicity/hydrophilicity and CO2 adoption properties, respectively.
133
134 2.3 Membrane fabrication

135 Porous alumina tubes with 60% porosity and pore size of 1.2 µm were supplied
136 from Nikkato Ltd., Japan and were used to fabricate the fluorine-doped silica
137 membranes. The tubes were coated by alumina particles (0.2 and 2 µm) diluted with a
138 SiO2-ZrO2 sol (5-10 wt%) by following the calcination at 550 °C for 30 min to achieve
139 a smooth particle layer. Next, an intermediate layer (average pore size of 1-2 nm) was
140 formed by depositing the sol (SiO2-ZrO2) with reduced concentration (0.5-1 wt%
141 diluted with H2O). Finally, a top separation thin layer was coated on the intermediate
142 layer with F-doped and undoped silica sol by following the calcination at 350 ºC in an
143 air environment. A defect-free top separation layer was achieved with 5-6 coats and
144 ramping rate of calcination temperature was controlled at 10 ºC min-1.

145

146 2.4 Membrane performance evaluation

147 Using the experimental apparatus shown in Fig. 1, we conducted single gase (He, H2,
148 CO2, N2, CH4, CF4, and SF6) permeation measurements at 50-300 ºC. Each gas was fed
149 from the outer surface of the membrane, and the permeate side pressure was maintained

Electronic copy available at: https://ssrn.com/abstract=4091364

150 at an atmospheric level. To desorb water from the membrane surface, each membrane
151 was kept overnight at 200 ⁰C under an inert environment. A film flow meter (SF-U,
152 Horiba Ltd., Japan) was used to measure gas permeate flow.
153 The binary separation performance was examined using a vacuum pump attached to
154 the permeate side to control the upstream and downstream pressures at 100 kPa and 10-
155 12 kPa, respectively. The composition of the gas mixture (CO2/N2) was evaluated using
156 gas chromatography (GC-14B, Shimadzu, Japan) and a TCD detector (column:
157 Unibeads 1S80/100, GL Science, Japan; column temperature: 75 ⁰C).
158 In order to evaluate the hydrothermal stability under steam at low partial pressure
159 (H2O partial pressure = 3 kPa), each gas (He, N2, CO2) was fed to the membrane cell
160 passing through the water evaporator at ambient temperature. Gas permeation properties
161 under a wet system were evaluated at temperatures ranging from 50-300 ⁰C. Finally, the
162 F-doped membranes were exposed to a steam atmosphere under a higher partial
163 pressure of steam (300 oC, 30 kPa) for one hour using an apparatus reported elsewhere
164 [38].
165

166

167

Electronic copy available at: https://ssrn.com/abstract=4091364

168 Fig. 1 Schematic illustration of the experimental apparatus used for single/binary gas
169 permeation measurement.
170
171 3. Results and discussion

172 3.1 Physico-chemical properties of F-doped silica gel/powders

173 Fig. 2 (a) indicates the FT-IR spectra of fluorine-doped silica films calcined at 350
174 ⁰C under an air atmosphere. Irrespective of the fluorine concentration, the asymmetric
175 vibrations of Si-O-Si ranged from 980 to 1,100 cm-1 and were detected in all samples,
176 which includes complete hydrolysis and polymerization reactions. A small shoulder
177 peak of Si-F was detected at around 940 cm-1 in F-doped samples and increased as
178 fluorine concentration was increased from 1-20%. As a result, the existence of these Si-
179 F groups into the silica structure affected the Si-O-Si bond angle, which led to a shift in
180 the Si-O-Si peak to a higher wave number (blue shift) [39].

181 Fig. 2 (b) depicts the Si-OH/Si-O-Si peak area ratios for fluorine-doped and
182 undoped silica films. The Si-OH/Si-O-Si peak area ratio for the pure silica sample was
183 higher than those of the F-doped silica samples. The Si-OH/Si-O-Si peak area ratio was
184 decreased largely by lower concentrations of fluorine (F/Si = 0.1/9.9, 0.3/9.7, 0.5/9.5)
185 and was somewhat independent of doped-fluorine concentrations above 10 mol%
186 (F/Si=1/9). These results suggest that when Si-F groups replaced silanol groups (Si-OH),
187 small concentration of fluorine significantly decreased the Si-OH density.

188

189
190

Electronic copy available at: https://ssrn.com/abstract=4091364

191 Fig. 2 FTIR spectra (a) and peak area ratios (b) for F-SiO2 films (F/Si = 0/10-2/8)
192 calcined at 350 ⁰C in an air atmosphere.

193

194 Fig. 3 shows the XRD patterns for fluorine-doped (F = 0-20%) and undoped silica
195 gels calcined at 350 ⁰C under an air atmosphere. All samples demonstrated an
196 amorphous structure with a single peak irrespective of the fluorine concentration. This
197 peak was shifted to a lower degree of 2θ and this trend was more prominent in samples
198 with higher fluorine concentrations, which indicates an enlargement of the d-spacing.
199 The change in the bond angle (Si-O-Si) indicated via FT-IR increased with increases in
200 the fluorine concentration. Thus, the FT-IR and XRD results depict the effectiveness of
201 fluorine by showing the formation of a loose network structure with decreased silanol
202 (Si-OH) density [40, 41].

203

F/Si = 2/8
Intensity [-]

F/Si = 1/9

F/Si = 0.5/9.5

F/Si = o.3/9.7

F = 0.1/9.9

F=0

10 20 30 40
2 
204
205

206 Fig. 3 XRD patterns for fluorine-doped (F/Si = 0/10-2/8) and undoped silica gels
207 calcined at 350 ⁰C in an air atmosphere.

208

209 Fig. 4 (a) shows the water adsorption isotherms at 25 ⁰C for F-doped (F/Si = 0.1/9.9,
210 F/Si = 0.3/9.7, and F/Si = 0.5/9.5) and undoped (F = 0) silica gels. Conventional silica
211 reaches the upper limits of absorbed water due to hydrophilic characteristics that make

Electronic copy available at: https://ssrn.com/abstract=4091364

212 it attractive to water vapors. After fluorine doping, however, H2O adsorption properties
213 were dependent on the concentration of fluorine doping since doped fluorine can
214 dramatically reduce Si-OH density. For example, high levels of silanol (Si-OH) in a gel
215 (F/Si = 0.1/9.9) leads to higher amounts of adsorbed water: F/Si=0.3/9.7 and 0.5/9.5.
216 Kanezashi et al. [42] reported similar results and concluded that despite being
217 hydrophobic due to the organic liking units, a sample of fluorine containing 1,2-
218 bis(triethoxysilyl)methane (BTESM, Si-CH2-Si) showed a reasonably lower amount of
219 absorbed water by comparison with pure BTESM.

220 Fig. 4 (b) illustrates the CO2 adsorption isotherms at 25 ⁰C for F-doped and undoped
221 silica gels. Higher CO2 uptake was observed for pure silica (TEOS) gel by comparison
222 with F-doped silica gels, since silica interacts with CO2 due to polar characteristics. F-
223 doped silica gels, however, showed that fluorine effectively controls the
224 hydrophobic/hydrophilic characteristics of conventional silica with decreases in CO2
225 adsorption as the content of fluorine increases. For instance, the gel with an F/Si ratio of
226 0.1/9.9 contained a higher amount of silanol groups compared with other samples (F/Si
227 = 0.3/9.7 & 0.5/9.5), which resulted in higher levels of CO2 adsorption. This trend for
228 CO2 adsorption was consistent with that of Si-OH density; i.e., F-doped silica with less
229 Si-OH showed a lower amount of CO2 adsorption.

230

231
232

233 Fig. 4 H2O adsorption (a) CO2 adsorption (b) isotherms for fluorine-doped and undoped
234 silica gels calcined at 350 ⁰C in an air atmosphere.

Electronic copy available at: https://ssrn.com/abstract=4091364

235

236 3.2 Network pore size evaluation of fluorine-doped and undoped silica membranes

237 Fig. 5 shows gas permeance results and permeance ratios at 50 ⁰C as a function of
238 fluorine concentration for fluorine-doped and undoped silica membranes calcined at 350
239 ⁰C under an air atmosphere. Since conventional silica exhibits a dense network structure,
240 H2 permeance of 1.1 × 10-7 mol m-2 s-1 Pa-1 with permselectivity of H2/N2 = 64 was
241 observed. On the contrary, F-induced membranes displayed improved gas permeation
242 with decreased permselectivity followed by the loose network formation, and this trend
243 was independent of the measurement temperature (Fig. S1). For instance, a membrane
244 (F/Si = 0.1/9.9) showed H2 permeance of 4.0 × 10-7 mol m-2 s-1 Pa-1 with H2/N2
245 selectivity of 58 by comparison with another membrane with an F/Si ratio of 2/8, which
246 recorded H2 permeance of 9.6 × 10-6 mol m-2 s-1 with H2/N2 selectivity of 31.

247 These results indicate that the presence of Si-F groups into a silica matrix form a
248 loose network, which results in improved permeation properties. Several studies have
249 reported the controllability of network pore size via fluorine incorporation, and it was
250 thus emphasized that fluorine effectiveness is independent irrespective of the silica
251 precursor. For example, study results reported by our group explain how fluorine
252 effectively controls the network pore sizes for both pure silica (SiO2) and bridged-type
253 organosilica structures 1,2-bis(triethoxysilyl)methane (BTESM, Si-CH2-Si) and 1,2-
254 bis(triethoxysilyl)propane (BTESP, Si-C3H6-Si). F-doped bridged-type organosilica
255 membranes show a loose network structure by comparison with F-SiO2 membranes
256 using the same doped-fluorine concentration due to the existence of a non-hydrolyzed
257 organic chain [38, 43, 44].

258

10

Electronic copy available at: https://ssrn.com/abstract=4091364

259
260
261 Fig. 5 Gas permeance and permeance ratio at 50 ⁰C as a function of fluorine
262 concentration for fluorine-doped and undoped silica membranes calcined at 350 ⁰C in an
263 air atmosphere.

264

265 Fig. 6 shows the temperature dependence (50-200 ⁰C) of He permeance for fluorine-
266 doped silica membranes. The temperature dependence of He permeance was evaluated
267 to discuss how fluorine impacts the silica network pore structure since most He
268 permeates the network pores and is unaffected by the presence of pinholes. The
269 permeance of He for fluorine-doped silica membranes slightly increases as temperature
270 increases, which suggests an activated diffusion mechanism, but the slope of
271 temperature dependence was decreased as the doped-fluorine concentration increased.
272 The permeance of He for fluorine-doped silica membranes (F/Si=2/8) was somewhat
273 independent of temperature. Permeance was increased with increases in the fluorine
274 concentration, irrespective of temperature (50-200 oC), which further demonstrated how
275 significantly fluorine controls the network pore size of conventional silica (TEOS).

276 The activation energy of He permeation was calculated using equation (1) [45].

𝑘0 𝐸𝑝
277 𝑃= 𝑀𝑅𝑇 [ ]
𝑒𝑥𝑝 ‒ 𝑅𝑇 (1)

11

Electronic copy available at: https://ssrn.com/abstract=4091364

278 In equation (1), 𝑃 is the gas permeance 𝑘0 is the permeation constant, 𝑀 is the
279 molecular weight of gas [kg mol-1], 𝑅 is the gas constant [J mol-1 K-1], 𝑇 is the absolute
280 temperature [K], and Ep is the activation energy of permeation [kJ mol-1]. Table 1
281 summarizes the activation energy of He permeation (Ep (He)) for F-SiO2 membranes
282 with different amount of incorporated fluorine. A decreased Ep (He) was observed as the
283 amount of incorporated fluorine was increased, indicating that molecules can permeate
284 micropores with less resistance due to the enlargement of pores in a network.

285

Temperature [C]
200 150 100 50
10-5

F/Si = 2/8
He permeance [mol m-2 s-1 Pa-1]

F/Si = 0.5/9.5
F/Si = 1/9
10-6
F/Si = 0.3/9.7

F/Si = 0.1/9.9

10-7
2.0 2.2 2.4 2.6 2.8 3.0 3.2
286 1000/T [K-1]

287
288 Fig. 6 Temperature dependence of He permeance at 50 -200 ⁰C for fluorine-doped silica
289 membranes calcined at 350 ⁰C in an air atmosphere.

290

291 Table 1 Ep (He) for F-SiO2 membranes with different fluorine concentrations
292
Membranes EP (He) kJ mol-1

F/Si = 0.1/9.9 4.8

F/Si = 0.3/9.7 3.6

12

Electronic copy available at: https://ssrn.com/abstract=4091364

 F/Si = 0.5/9.5 3.1

F/Si = 1/9 2.79

F/Si = 2/8 1.6

293

294 3.3 Evaluation of binary (CO2/N2) separation for fluorine-doped silica membranes

295 Fig. 7 shows the temperature dependence of CO2 and N2 permeance in single and
296 binary separation for fluorine-doped silica membranes. In a single-permeation system,
297 the permeance of CO2 was increased with decreased temperature, which showed a
298 surface-diffusion property. This trend was independent of the concentration of doped
299 fluorine in SiO2. It should be noted that CO2 is an adsorptive gas, and the obtained trend
300 suggests that adsorption played a crucial role in the CO2 permeation properties. The
301 permeance of N2 for a fluorine-doped silica membrane (F/Si=0.1/9.9) was slightly
302 increased as the temperature increased, showing an activated permeation mechanism.
303 The permeance of N2 for fluorine-doped silica membranes (F/Si=0.3/9.7, 0.5/9.5) was
304 somewhat independent of permeation temperature. Accordingly, the CO2/N2
305 permselectivity was slightly increased with decreased temperature, irrespective of the
306 fluorine concentration.

307 In binary separation, all membranes showed a similar dependence on temperature

308 with respect to CO2 and N2 permeance in a single permeation system. On the other hand,
309 the deviation of CO2 permeance between single permeation and binary separation was
310 increased with decreased temperature, and this trend became more obvious in fluorine-
311 induced silica membranes (F/Si=0.1/9.9) due to the increased CO2 adsorptive effect, as
312 shown in Fig 4(b). For example, a fluorine-doped silica membrane (F/Si=0.1/9.9)
313 showed CO2 permeance of 1.3 x 10-7 and 2.8 x 10-7 mol m-2 s-1 Pa-1 in single and binary
314 separations at 50 ⁰C, respectively, and CO2/N2 selectivity in binary (5:5) separation was
315 higher than that of single gas separation systems. It should be noted that there was no
316 blocking of N2 permeation by adsorbed CO2 even at low temperature.

317

13

Electronic copy available at: https://ssrn.com/abstract=4091364

318
319
320 Fig. 7 Temperature-dependence of single and binary (CO2/N2=5/5) gas permeance at
321 50-200 ⁰C for fluorine-doped (F/Si = 0.1/9.9, 0.3/9.7, and 0.5/9.5) silica membranes
322 calcined at 350 ⁰C in an air atmosphere.

323

324 Fig. 8 shows the feed partial pressure dependence of CO2 permeance at 50 ⁰C for
325 fluorine-doped silica membranes (F/Si=0.1/9.9, 0.3/9.7, 0.5/9.5). The permeance of CO2
326 increased as the feed partial pressure of CO2 decreased in fluorine-doped silica
327 membranes (F/Si=0.1/9.9, 0.3/9.7). On the contrary, the permeance of CO2 in fluorine-
328 induced silica membranes with F/Si = 0.5/9.5 was approximately independent of the
329 partial pressure of the CO2 feed. These results suggested that the surface diffusion
330 mechanism was more obvious with lower fluorine concentrations in SiO2. As shown in
331 Fig. 4 (b), gels with a lower fluorine concentration (F/Si=0.1/9.9) showed reasonably
332 higher CO2 adsorption properties, which decreased as fluorine content increased from
333 0.3/9.7 to 0.5/9.5. That result demonstrated that hydrophobic fluorine apparently
334 decreases the CO2 adsorption properties of the silica surface as the fluorine
335 concentration increases.

336

14

Electronic copy available at: https://ssrn.com/abstract=4091364

337
338

339 Fig. 8 Feed partial pressure dependence of CO2 permeance at 50 ⁰C for fluorine-doped
340 silica membranes (F/Si=0.1/9.9, 0.3/9.7, 0.5/9.5) calcined at 350 ⁰C in an air atmosphere.

341

342 To further probe the CO2 separation potential of fluorine-doped silica membranes,
343 the relationship between the Ep of CO2 and that of N2 was calculated using equation (1).
344 Fig. 9 shows the relationship between Ep (CO2), Ep (N2), and Ep (CO2)-Ep (N2) as a
345 function of fluorine concentration. It should be noted that the apparent activation energy
346 (activation energy of permeation), Ep, is the sum of the activation energy for diffusion
347 and that of adsorption enthalpy (𝐸𝑝 = 𝐸𝑑 + ∆𝐻𝑎𝑑). The Ep of N2 could be considered
348 approximately equal to Ed, depending on the pore size of the permeating path and the
349 molecular size, since N2 has fewer adsorption properties [1]. The activation energy of
350 N2 permeation was clearly decreased with an increase in the fluorine concentration,
351 which indicates a loose network formation for F-doped silica membranes.

352 The activation energy of CO2 permeation was decreased from that of pure SiO2 (0
353 mol%) with a 5 mol% concentration of fluorine but was slightly increased at higher
354 fluorine concentrations. Here, if we assume the molecular size of CO2 (0.33 nm)
355 approximates that of N2 (0.364 nm), the activation energy for the diffusion for both CO2
356 and N2 could be assumed to be similar (Ed (CO2)  Ed (N2)). Thus, the Ep (CO2) – Ep
357 (N2) could be considered an expression of the adsorption enthalpy of CO2 [45], which
358 indicates the CO2 adsorption strength of membranes with different fluorine
359 concentrations. Ep (CO2) – Ep (N2) was clearly increased with increases in the fluorine

15

Electronic copy available at: https://ssrn.com/abstract=4091364

360 concentration, which indicated a small affinity for an increase in F-doping. Thus, we
361 concluded that fluorine doping could be used to precisely enlarge the network pore size
362 in a subnano range by reducing the concentration of Si-OH in the network structure.
363 CO2 adsorption was decreased with increased amount of incorporated fluorine.

364

365
366
367 Fig. 9 Relationship between Ep (CO2), Ep (N2), and Ep (CO2) - Ep (N2) as a function of
368 fluorine concentration.

369

370 Fig. S2 depicts the CO2/N2 selectivity at 50 ⁰C against CO2 permeance for various
371 fluorine concentrations of silica membranes. In the single-gas system shown in Fig. S2
372 (closed symbols), a conventional silica membrane exhibits a dense network structure
373 with a lower CO2 permeance of 6.2 × 10-8 mol m-2 s-1 Pa-1 and a moderate selectivity of
374 26. Conversely, fluorine doping induced silica membranes to display a loose network
375 structure with improved CO2 permeance. In a binary system, on the other hand,
376 membranes with an F/Si ratio of 0.1/9.9 showed a higher permeance ratio of 50 with
377 CO2 permeance of 1.6 × 10-7 mol m-2 s-1 Pa-1. This could be associated with the
378 fluorine-doped control of CO2 permeation/adsorption properties. Since CO2 is an
379 adsorptive gas, this adsorptive effect was increased with decreases in the fluorine

16

Electronic copy available at: https://ssrn.com/abstract=4091364

380 concentration, which resulted in improved selectivity. Thus, the simultaneous control of
381 molecular sieving and adsorption properties could be a strategy to achieve high CO2
382 selectivity with high CO2 permeance.

383

384 3.4 Evaluating the hydrothermal stability of fluorine-doped SiO2 membranes under
385 steam conditions

386 Hydrothermal stability for fluorine-doped silica membranes was evaluated by

387 measuring the time courses of He and N2 permeance in both dry and wet systems (H2O
388 partial pressure: 3 kPa). Fig. 10 shows the time courses of He and N2 permanence under
389 both dry and wet atmospheres at 200 ⁰C for F-SiO2 membranes. There were no changes
390 in He and N2 permeance, irrespective of fluorine concentration, neither before nor after
391 exposure to a wet system. In addition, the values of permeance for He and N2 in a wet
392 system were similar to those in a dry system, and permeation blocking by adsorbed
393 water molecules was not confirmed in these membranes.

394

395
396
397 Fig. 10 Time course of He and N2 permanence under both dry and wet (H2O partial
398 pressure: 3 kPa) systems at 200 ⁰C for F-SiO2 membranes calcined at 350 ⁰C in an air
399 atmosphere.

17

Electronic copy available at: https://ssrn.com/abstract=4091364

400 Fig. 11 shows the temperature dependence (50-300 ⁰C) of fluorine-doped silica
401 membranes in both dry and wet (H2O partial pressure: 3 kPa) systems. It worth noticing
402 that each permeance was plotted after a steady state was confirmed in each
403 measurement condition. In a wet system, a blocking effect at around 100-150 ⁰C was
404 observed since gas permeance was apparently lower than that in a dry system,
405 irrespective of the permeating molecules and doped fluorine concentration. This
406 indicates that gas permeation was inhibited by water vapor absorbed onto the membrane
407 surface and/or by capillary condensation in the pores.

408 In a wet system the gas permeance ratio (He/N2, CO2/N2) was lower than that in a
409 dry system below 150 ⁰C. Even though gas permeance in a wet system was lower than
410 that in a dry system, irrespective of doped fluorine concentration, the effect of blocking
411 by adsorbed H2O at 150 ⁰C was more obvious in a membrane with a lower amount of
412 fluorine (F/Si = 0.1/9.9) because the network was more interactive to water vapor. For
413 example, decreases in the ratio of He permeance at 50 ⁰C were approximately 90 and
414 50% for membranes with fluorine/silica ratios of 0.1/9.9 and 0.5/9.5, respectively.

415

416
417
418 Fig. 11 Temperature dependence of He, CO2 and N2 permeance under dry and wet (H2O
419 partial pressure: 3 kPa) atmospheres for fluorine-doped (F/Si=0.1/9.9, 0.3/9.7, and
420 0.5/9.5) silica membranes calcined at 350 ⁰C in an air atmosphere.

18

Electronic copy available at: https://ssrn.com/abstract=4091364

421

422 Fig. 12 shows the molecular size dependence of single-gas permeance at 300 ⁰C for
423 fluorine-doped silica membranes calcined at 350 ⁰C before/after steam treatment (300
424 ⁰C, H2O partial pressure: 30 kPa) for 1 h. In steam with a low level of partial pressure
425 (H2O partial pressure: 3 kPa), a F-SiO2 (F/Si=0.1/9.9) membrane showed stable
426 performance, as shown in Fig. 10 (a). Under steam with a high level of partial pressure,
427 however, a F-SiO2 membrane with the same makeup (F/Si=0.1/9.9) showed drastic
428 decreases in permeance, particularly for He and H2. For example, the permeance of He
429 was decreased by 20% of the initial value after an exposure of only 1h to steam with a
430 high level of partial pressure.

431 That significant decrease in permeance was probably caused by doping with a very
432 low concentration of fluorine (F/Si=0.1/9.9), which increased the frequency of
433 interactions of the Si-OH-rich silica structure with water vapor by increasing the silanol
434 density and network rearrangements. On the contrary, when the fluorine content was
435 increased, the hydrothermal stability was increased accordingly and both membranes
436 (F/Si = 0.3/9.7 & 0.5/9.5) showed steady values of gas permeance under both dry and
437 wet systems. Thus, the membrane performance evaluation in a wet system apparently
438 depends on the hydrophobicity/hydrophilicity properties of the silica network structure.

439

440
441
442 Fig. 12 Molecular size dependence for F-SiO2 membranes (F/Si = 0.1/9.9, 0.3/9.7,
443 0.5/9.5) before/after steam treatment (300 ⁰C, partial pressure of steam: 30 kPa, 1h).

444

19

Electronic copy available at: https://ssrn.com/abstract=4091364

445 Fig. 13 shows the N2 adsorption isotherms for fluorine-doped silica gels calcined at
446 350 ⁰C before/after steam treatment (300 ⁰C, steam partial pressure: 30 kPa, 1h). Both
447 samples demonstrated type I isotherms under a low level of relative pressure with large
448 levels of N2 uptake, which confirms the microporous network structure of gels. Table
449 S1 summarizes the microporous properties (SBET & Vp) for these samples. The
450 adsorption properties increased with increases in the fluorine concentration, which
451 demonstrates a strong dependency on the amount of fluorine incorporated into the silica
452 matrix.

453 Following exposure to a steam atmosphere the gels showed similar results for the
454 sample (F/Si = 0.3/9.7) shown in Table S1. On the contrary, the sample with a small
455 amount of fluorine (0.1/9.9) underwent densification and exhibited decreases in the
456 surface area and in the micropore volume, which demonstrated that the network
457 structure was less resistive to water vapor and interacted with wet molecules, which
458 resulted in the generation of Si-OH density and a subsequent promotion of
459 condensation/densification.

460

461
462
463 Fig. 13 N2 adsorption isotherms at 77 K before/after steam treatment (300 oC, partial
464 pressure of steam: 30 kPa, 1 h) for fluorine-doped (F/Si = 0.1/9.9, 0.3/9.7) silica gels
465 calcined at 350 ⁰C in an air atmosphere.

466

20

Electronic copy available at: https://ssrn.com/abstract=4091364

467 Fig. S3 shows the magnified F 1s spectra ranging from 680 to 694 eV for fluorine-
468 containing silica gels. The results were obtained before/after steam treatment (300 ⁰C,
469 steam partial pressure: 30 kPa, 1 h) for all fluorine-containing silica gels. An F 1s peak
470 at around 688 eV was detected in all samples irrespective of the fluorine concentration;
471 however, in gels with F/Si ratios of 0.1/9.9, the physical appearance of the peak
472 intensity was decreased due to the low amount of fluorine incorporated into the silica
473 structure. Table S2 summarizes the F/Si ratios for F-doped silica gels before/after steam
474 treatment. It should be noted that the calculated F/Si ratios were approximately the same
475 as that of the concentrations of doped fluorine. After steam treatment, the F/Si ratio was
476 decreased, irrespective of fluorine concentration. The decreased fluorine concentration
477 could be a result of the reaction between Si-F and water vapor during steam treatment.

478 Fig. 14 features the schematic images of the network pore structures of fluorine
479 doped SiO2 membranes and the effect that steam treatment exerts on the network pore
480 structure. It is evident that hydrothermal stability involves several factors such as
481 membrane pore size, steam partial pressure, temperature, and exposure time [29, 44].
482 Although fluorine tremendously improved the hydrothermal stability as reported in
483 various studies [29, 38, 44], results in the present study indicate that network stability is
484 significantly affected by the amount of incorporated fluorine. For example, the network
485 structure created by an F/Si ratio of 0.1/9.9 possesses a high level of silanol density and
486 a smaller network pore size than networks doped with a higher concentration of fluorine.
487 After steam exposure, Si-F groups interacted with water vapor and the original
488 concentration of Si-OH along with regenerated silanol groups resulted in network
489 densification. On the contrary, network pore structures with a higher fluorine
490 concentration showed less condensation/densification due to reduced silanol density,
491 which resulted in improved hydrothermal stability.

492

21

Electronic copy available at: https://ssrn.com/abstract=4091364

493
494

495 Fig. 14 Schematic illustration for F-SiO2 network structures before/after steam
496 treatment (30 kPa).

497

498 4. Conclusions

499 A strategy was proposed to tune the network pore structure of silica via F-doping to
500 fabricate highly permeable membranes for CO2 separation. Various fluorine
501 concentrations (F/Si=0.1/9.9, 0.3/9.7, 0.5/9.5, 1/9, 2/8) were doped via a sol-gel method
502 to allow examination of the effect that fluorine exerts on a silica network structure.
503 Physicochemical analysis revealed that fluorine effectively improved the
504 hydrophobic/hydrophilic properties by eliminating the factors (Si-OH) involved in the
505 formation of a dense silica structure.

506 Fluorine-doped silica membranes fabricated with different concentrations of fluorine

507 showed consistent results according to structural characterization. For example, an F/Si
22

Electronic copy available at: https://ssrn.com/abstract=4091364

508 ratio of 0.5/9.5 showed a loose network structure by comparison with a smaller fluorine
509 content (F/Si=0.1/9.9). Fluorine doping precisely enlarges the network pore size in a
510 sub-nano range. Due to a reduction in Si-OH in a network structure, however, CO2
511 adsorption is decreased as the fluorine induced concentration increases. A fluorine-
512 doped silica membrane with an F/Si ratio of 0.1/9.9 showed the highest permeance ratio
513 of 50 with CO2 permeance of 1.6 × 10-7 mol m-2 s-1 Pa-1 at 50 ⁰C.

514 In a wet system (200-300 oC, H2O partial pressure: 3 kPa), fluorine-doped silica
515 membranes were quite stable, irrespective of the doped amount of the fluorine. A
516 blocking effect was observed at around 100-150 ⁰C, however, and it was apparent that
517 gas permeance (He, CO2, N2) was lower than that in a dry system. Under steam with a
518 high level of partial pressure (300 oC, H2O partial pressure: 30 kPa), the incorporated
519 amount of fluorine strongly affected the hydrothermal stability. The F-SiO2 membrane
520 with an F/Si ratio of 0.1/9.9 showed drastic decreases for all examples of permeance.
521 On the contrary, when fluorine content was increased, it was apparent that the change in
522 pore size distribution accordingly suppressed the hydrothermal stability and the
523 membranes with F/Si ratios of 0.3/9.7 & 0.5/9.5 showed steady values for gas
524 permeance in both dry and wet systems.

525

526 References

527 1. L. Yu, M. Kanezashi, H. Nagasawa, J. Oshita, A. Naka, T. Tsuru, Pyrimidine-

528 bridged organoalkoxysilane membrane for high-efficiency CO2 transport via mild
529 affinity. Sep. Purif. Technol. 178 (2017) 232-241.
530 2. A.M.W. Hillock, W.J. Koros, Cross-linkable polyimide membrane for natural gas
531 purification and carbon dioxide plasticization reduction. Macro. 40 (2007) 583-587.
532 3. A. Car, C. Stropnik, W. Yave, K.V. Peinemann, Tailor‐made polymeric membranes
533 based on segmented block copolymers for CO2 separation. Adv. Func. Mater. 18
534 (2008) 2815-2823.
535 4. T.S. Chung, L.Y. Jiang, Y. Li, S. Kulprathipanja, Mixed matrix membranes
536 (MMMs) comprising organic polymers with dispersed inorganic fillers for gas
537 separation. Prog. Poly. Sci. 32 (2007) 483-507.
23

Electronic copy available at: https://ssrn.com/abstract=4091364

538 5. D.L. Gin, R.D. Noble, Designing the next generation of chemical separation
539 membranes. Sci. 332 (2011) 674-676.
540 6. W. He, Z. Wang, W. Li, S. Li, Z. Bai, J. Wang, S. Wang, Cyclic tertiary amino
541 group containing fixed carrier membranes for CO2 separation. J. Membr. Sci. 476
542 (2015) 171-181.
543 7. H. Lin, B.D. Freeman, Materials selection guidelines for membranes that remove
544 CO2 from gas mixtures. J. Mol. Struc. 739 (2005) 57-74.
545 8. M. Ostwal, R.P. Singh, S.F. Dec, M.T. Lusk, J.D. Way, 3-
546 Aminopropyltriethoxysilane functionalized inorganic membranes for high
547 temperature CO2/N2 separation. J. Membr. Sci. 369 (2011) 139-147.
548 9. D.E. Gottschlich, D.L. Roberts, G.J. Wijmans, C.M. Bell, R.W. Baker, Economic
549 comparison of several membrane configurations for H2/N2 separation. Gas Sep.
550 Puri. 3 (1989) 170-179.
551 10. W. Jiao, Y. Ban, Z. Shi, X. Jiang, Y. Li, W. Yang, Gas separation performance of
552 supported carbon molecular sieve membranes based on soluble
553 polybenzimidazole. J. Membr. Sci. 533 (2017) 1-10.
554 11. L. Yu, M. Grahn, P. Ye, J. Hedlund, Ultra-thin MFI membranes for olefin/nitrogen
555 separation. J. Membr. Sci. 524 (2017) 428-435.
556 12. Y. Kuwahara, Y. Kamegawa, K. Mori, H. Yamashita, Fabrication of hydrophobic
557 zeolites using triethoxyfluorosilane and their application as supports for TiO2
558 photocatalysts. Chem. Commun. 39 (2008) 4783-4785.
559 13. H.P. Li, X.P. Shen, K. Han, G. Tang, Z.H. Zhang, Quantum chemistry study on the
560 third-order nonlinear optical properties of spirobifluorene derivatives. Comput.
561 Theor. Chem. 1023 (2013) 95-98.
562 14. Y. Peng, Y. Li, Y. Ban, H. Jin, W. Jiao, X. Liu, W. Yang, Metal-organic framework
563 nanosheets as building blocks for molecular sieving membranes. Sci. 346 (2014)
564 1356-1359.
565 15. S. Chai, H. Du, Y. Zhao, Y. Lin, C. Kong, L. Chen, Fabrication of highly selective
566 organosilica membrane for gas separation by mixing bis (triethoxysilyl) ethane with
567 methyltriethoxysilane. Sep. Purif. Technol. 222 (2019) 162-167.

24

Electronic copy available at: https://ssrn.com/abstract=4091364

568 16. R.M. De Vos, W.F. Maier, H. Verweij, Hydrophobic silica membranes for gas
569 separation. J. Membr. Sci. 158 (1999) 277-288.
570 17. J. Campaniello, C.W. Engelen, W.G. Haije, P.P.A.C. Pex, J.F. Vente, Long-term
571 pervaporation performance of microporous methylated silica membranes. Chem.
572 Commun. (2004) 834-835.
573 18. X. Ren, K. Nishimoto, M. Kanezashi, H. Nagasawa, T. Yoshioka, T. Tsuru, CO2
574 permeation through hybrid organosilica membranes in the presence of water
575 vapor. Ind. Eng. Chem. Res. 53 (2014) 6113-6120.

576 19. M. Kanezashi, T. Matsutani, T. Wakihara, H. Tawarayama, H. Nagasawa, T.

577 Yoshioka, T. Tsuru, Tailoring the subnano silica structure via fluorine doping for
578 development of highly permeable CO2 separation membranes. ChemNanoMat. 2
579 (2016) 264-267.
580 20. X. Yu, H. Nagasawa, M. Kanezashi, T. Tsuru, Improved thermal and oxidation
581 stability of bis (triethoxysilyl) ethane (BTESE)-derived membranes, and their gas-
582 permeation properties. J. Mater. Chem. A 6 (2018) 23378-23387.
583 21. T. Yoshioka, M. Kanezashi, T. Tsuru, Micropore size estimation on gas separation
584 membranes: A study in experimental and molecular dynamics. AIChE J. 59 (2013)
585 2179-2194.
586 22. H.R. Lee, T. Shibata, M. Kanezashi, T. Mizumo, J. Oshita, T. Tsuru, Pore-size-
587 controlled silica membranes with disiloxane alkoxides for gas separation. J. Membr.
588 Sci. 383 (2011) 152-158.
589 23. M. Kanezashi, D. Fuchigami, T. Yoshioka, T. Tsuru, Control of Pd dispersion in
590 sol–gel-derived amorphous silica membranes for hydrogen separation at high
591 temperatures. J. Membr. Sci. 439 (2013) 78-86.
592 24. M. Kanezashi, Y. Yoneda, H. Nagasawa, T. Tsuru, Gas permeation properties for
593 organosilica membranes with different Si/C ratios and evaluation of microporous
594 structures. AIChE J. 63 (2017) 4491-4498.
595 25. N. Moriyama, H. Nagasawa, M. Kanezashi, T. Tsuru, Selective water vapor
596 permeation from steam/non-condensable gas mixtures via organosilica membranes
597 at moderate-to-high temperatures. J. Membr. Sci. 589 (2019) 117254.

25

Electronic copy available at: https://ssrn.com/abstract=4091364

598 26. H.L. Castricum, A. sah, R. Kreiter, D.H.A. Blank, J.F. Vente, J.E. ten Elshof,
599 Hydrothermally stable molecular separation membranes from organically linked
600 silica. J. Mater. Chem.18 (2008) 2150-2158.
601 27. R. Igi, T. Yoshioka, Y.H. Ikuhara, Y. Iwamoto, T. Tsuru, Characterization of
602 Co‐doped silica for improved hydrothermal stability and application to hydrogen
603 separation membranes at high temperatures. J. Amer. Ceram. Soci. 91 (2008) 2975-
604 2981.
605 28. M. Kanezashi, M. Asaeda, Stability of H2-permselective Ni-doped silica
606 membranes in steam at high temperature. J. Chem. Eng. J. 38 (2005) 908-912.
607 29. M. Kanezashi, N. Hataoka, I. Rana, H. Nagasawa, T. Tsuru, Hydrothermal stability
608 of fluorine‐induced microporous silica membranes: effect of steam treatment
609 conditions. AIChE J. 67 (2021)e 17292.
610 30. Q. Wei, F. Wang, Z.R. Nie, C.L. Song, Y.L. Wang, Q.Y. Li, Highly hydrothermally
611 stable microporous silica membranes for hydrogen separation. J. Phys. Chem.
612 B. 112 (2008), 9354-9359.
613 31. A. Darmawan, J. Motuzas, S. Smart, A. Julbe, J.C.D. da Costa, Binary iron cobalt
614 oxide silica membrane for gas separation. J. Membr. Sci. 474 (2015) 32-38.
615 32. J. Yang, W. Fan, C.M. Bell, Effect of calcination atmosphere on microstructure and
616 H2/CO2 separation of palladium-doped silica membranes. Sep. Purif. Technol. 210
617 (2010) 659-669.
618 33. H. Song, S. Zhao, J. Lei, C. Wang, H. Qi, Pd-doped organosilica membrane with
619 enhanced gas permeability and hydrothermal stability for gas separation. J. Mater
620 Sci. 51 (2016) 6275-6286.
621 34. D.L. Martens, J. Motuzas, S. Smart, J.C. Diniz da Costa, Structural investigation of
622 cobalt oxide seeded silica xerogels under harsh hydrothermal condition. J. Sol-Gel
623 Sci. Technol. 98 (2021) 470-477.
624 35. S. Smart, J.F. Vente, J.D. Da Costa, High temperature H2/CO2 separation using
625 cobalt oxide silica membranes. Inter. J. Hydr. Ener. 37 (2012) 12700-12707.
626 36. E.S. Mirza, B. Topuz, Nanoscale tailoring on thin bimetallic organo-oxide
627 membranes for H2/CO2 separation. Sep. Purif. Technol. 280 (2022) 119801.

26

Electronic copy available at: https://ssrn.com/abstract=4091364

628 37. H. Qi, J. Han, N. Xu, Effect of calcination temperature on carbon dioxide
629 separation properties of a novel microporous hybrid silica membrane. J. Membr.
630 Sci. 382 (2011) 231-237.
631 38. I. Rana, H. Nagasawa, K. Yamamoto, T. Gunji, T. Tsuru, M. Kanezashi, Effect of
632 fluorine doping on the network pore structure of non-porous organosilica bis
633 (triethoxysilyl) propane (BTESP) membranes for use in molecular separation. J.
634 Membr. Sci. (2021) 120083.
635 39. Y.H. Kim, M.S. Hwang, H.J. Kim, Infrared spectroscopy study of low-dielectric-
636 constant fluorine-incorporated and carbon-incorporated silicon oxide films. J. of
637 App. Phy. 90 (2001) 3367-3370.
638 40. U. Anggarini, L. Yu, H. Nagasawa, M. Kanezashi, T. Tsuru, Metal-induced
639 microporous aminosilica creates a highly permeable gas-separation membrane.
640 Mater. Chem. Front. 5 (2021) 3029-3042.
641 41. Q. Wang, Y. Kawano, L. Yu, H. Nagasawa, M. Kanezashi, T. Tsuru, Development
642 of high-performance sub-nanoporous SiC-based membranes derived from
643 polytitanocarbosilane. J. Membr. Sci. 598 (2020) 117688.
644 42. M. Kanezashi, M. Murata, H. Nagasawa, T. Tsuru, Fluorine doping of microporous
645 organosilica membranes for pore size control and enhanced hydrophobic
646 properties. ACS omega. 3 (2018) 8612-8620.
647 43. M. Takenaka, H. Nagasawa, T. Tsuru, M. Kanezashi, Hydrocarbon permeation
648 properties through microporous fluorine-doped organosilica membranes with
649 controlled pore sizes. J. Membr. Sci. 619 (2021) 118787.
650 44. M. Kanezashi, T. Matsutani, H. Nagasawa, T. Tsuru, Fluorine-induced microporous
651 silica membranes: Dramatic improvement in hydrothermal stability and pore size
652 controllability for highly permeable propylene/propane separation. J. Membr.
653 Sci. 549 (2018) 111-119.
654 45. L. Yu, M. Kanezashi, H. Nagasawa, J. Ohshita, A. Naka, T. Tsuru. Fabrication and
655 microstructure tuning of a pyrimidine-bridged organoalkoxysilane membrane for
656 CO2 separation. Ind. Eng. Chem. Res. 56 (2017) 1316-1326.

27

Electronic copy available at: https://ssrn.com/abstract=4091364