Zirconia-Doped Methylated Silica Membranes via Sol-Gel Process: Microstructure and Hydrogen Permselectivity

In order to obtain a steam-stable hydrogen permselectivity membrane, with tetraethylorthosilicate (TEOS) as the silicon source, zirconium nitrate pentahydrate (Zr(NO3)4·5H2O) as the zirconium source, and methyltriethoxysilane (MTES) as the hydrophobic modifier, the methyl-modified ZrO2-SiO2 (ZrO2-MSiO2) membranes were prepared via the sol-gel method. The microstructure and gas permeance of the ZrO2-MSiO2 membranes were studied. The physical-chemical properties of the membranes were characterized by Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), transmission electron microscopy (TEM), scanning electron microscope (SEM), and N2 adsorption–desorption analysis. The hydrogen permselectivity of ZrO2-MSiO2 membranes was evaluated with Zr content, temperature, pressure difference, drying control chemical additive (glycerol) content, and hydrothermal stability as the inferred factors. XRD and pore structure analysis revealed that, as nZr increased, the MSiO2 peak gradually shifted to a higher 2θ value, and the intensity gradually decreased. The study found that the permeation mechanism of H2 and other gases is mainly based on the activation–diffusion mechanism. The separation of H2 is facilitated by an increase in temperature. The ZrO2-MSiO2 membrane with nZr = 0.15 has a better pore structure and a suitable ratio of micropores to mesopores, which improved the gas permselectivities. At 200 °C, the H2 permeance of MSiO2 and ZrO2-MSiO2 membranes was 3.66 × 10−6 and 6.46 × 10−6 mol·m−2·s−1·Pa−1, respectively. Compared with the MSiO2 membrane, the H2/CO2 and H2/N2 permselectivities of the ZrO2-MSiO2 membrane were improved by 79.18% and 26.75%, respectively. The added amount of glycerol as the drying control chemical additive increased from 20% to 30%, the permeance of H2 decreased by 11.55%, and the permselectivities of H2/CO2 and H2/N2 rose by 2.14% and 0.28%, respectively. The final results demonstrate that the ZrO2-MSiO2 membrane possesses excellent hydrothermal stability and regeneration capability.


Introduction
It is well-known that hydrogen is a clean energy source [1]. At present, there are many ways to obtain H 2 , but the biggest problem preventing its commercialization is the purification and separation of H 2 . The purification of H 2 can be achieved in three main ways: pressure swing adsorption, cryogenic distillation, and membrane separation [2,3]. Although pressure swing adsorption and cryogenic distillation can be operated commercially, the economic benefits are low. The main commercial application of membranes in gas separation is the separation of hydrogen from nitrogen, methane, and argon in an ammonia sweep gas stream. In the past few years, hundreds of new polymer materials have been reported, and only eight or nine polymer materials have been used to make gas separation membrane bases. Surprisingly few of them were used to make industrial membranes [4]. Membrane separation technology is also one of the most promising hydrogen purification membrane, and silica membrane before and after hydrothermal treatment under the same circumstances. At 100 • C, the hybrid silica membrane and Zr-doped BTESE membrane maintained good hydrothermal stability, while the silica membrane lost selectivity for all the studied gases. After hydrothermal treatment at 200 or 300 • C, the CO 2 permeance of the Zr-doped BTESE membrane decreased significantly, and the H 2 /CO 2 permselectivity increased significantly, by 65.71%. So far, many scholars have demonstrated the effect of different conditions during preparation on the properties of zirconia-doped silica materials/membranes. The influence of the Zr/Si molar ratio on the microstructure of the membrane and the permeability of the gas is crucial. Unfortunately, there are few reports in this regard. Furthermore, the effects of methyl modification on the microstructure and steam stability of ZrO 2 -SiO 2 membranes were rarely described in papers. Some scholars have found that adding a drying control chemical additive (DCCA) in the process of preparing the membrane via the sol-gel method can effectively reduce the uneven shrinkage of the membrane during the heating process and during the calcining process [35], and improve the gas permselectivity of the membrane.
In this paper, methyl-modified ZrO 2 -SiO 2 (ZrO 2 -MSiO 2 ) materials/membranes with various Zr/Si molar ratios (n Zr ) were fabricated. Glycerol was chosen to be the DCCA. The impact of n Zr on the microstructures and H 2 permselectivities of ZrO 2 -MSiO 2 membranes was thoroughly addressed. The water vapor stability of ZrO 2 -MSiO 2 membranes was investigated further by comparing the gas permeability characteristics of the ZrO 2 -MSiO 2 membranes before and after steam treatment. The heat regeneration performance of ZrO 2 -MSiO 2 membranes was also investigated.

Preparation of MSiO 2 Sols
The MSiO 2 sols were prepared by tetraethylorthosilicate (TEOS, purchased from Xi'an chemical reagent Co., Ltd., Xi'an, China) as a silica source, methyltriethoxysilane (MTES, purchased from Hangzhou Guibao Chemical Co., Ltd., Hangzhou, China) as a hydrophobic modified agent, anhydrous ethanol (EtOH, purchased from Tianjin Branch Micro-Europe Chemical Reagent Co., Ltd., Tianjin, China) as a solvent, and nitric acid (HNO 3 , purchased from Sichuan Xilong Reagent Co., Ltd., Chengdu, China) as a catalyst. To begin, TEOS, MTES, and EtOH were completely combined in a three-necked flask using a magnetic stirrer. The flask was correctly immersed in an ice-water combination. The solution was then agitated for 50 min using a magnetic stirrer to ensure thorough mixing. The H 2 O and HNO 3 combination was then dropped into the mixture while it was still being stirred. The reaction mixture was then agitated in a three-necked flask at a constant temperature of 60 • C for 3 h to yield the MSiO 2 sol.

Preparation of ZrO 2 -MSiO 2 Materials
The ZrO 2 -MSiO 2 sols were then placed individually in petri plates for gelation at 30 • C. The gel materials were ground and pulverized with a mortar, and then calcined at a heating rate of 0.5 • C·min −1 at 400 • C for 2 h under nitrogen atmosphere protection, and then cooled down naturally. The ZrO 2 -MSiO 2 materials with different n Zr were prepared. The ZrO 2 -MSiO 2 materials with n Zr = 0 are also referred to as "MSiO 2 " materials.

Preparation of ZrO 2 -MSiO 2 Membranes
The ZrO 2 -MSiO 2 membranes were coated on top of composite interlayers supported by porous α-alumina discs. The discs are 5 mm-thick and 30 mm in diameter, with a porosity of 40% and an average pore size of 100 nm. ZrO 2 -MSiO 2 membranes were effectively prepared by dip-coating the substrates in three-fold ethanol-diluted silica sol for 7 s, then drying and calcining them. Each sample was dried at 30 • C for 3 h before being calcined at 400 • C in a temperature-controlled furnace in a N 2 environment with a ramping rate of 0.5 • C·min −1 and a dwell period of 2 h. The dip-coating-drying-calcining process was repeated three times. Figure 1 demonstrates the preparation process of the ZrO 2 -MSiO 2 materials/membranes. The ZrO 2 -MSiO 2 membranes with n Zr = 0 are also referred to as "MSiO 2 " membranes. Nanomaterials 2022, 12, x FOR PEER REVIEW 4 of 18 ethanol after 12 h. GL was used as a drying control agent at 0%, 10%, 20%, and 30% (DCCA). After 60 min of stirring, ZrO2-MSiO2 sols with varied GL contents were obtained.

Preparation of ZrO2-MSiO2 Materials
The ZrO2-MSiO2 sols were then placed individually in petri plates for gelation at 30 °C. The gel materials were ground and pulverized with a mortar, and then calcined at a heating rate of 0.5 °C·min −1 at 400 °C for 2 h under nitrogen atmosphere protection, and then cooled down naturally. The ZrO2-MSiO2 materials with different nZr were prepared. The ZrO2-MSiO2 materials with nZr = 0 are also referred to as "MSiO2" materials.

Preparation of ZrO2-MSiO2 Membranes
The ZrO2-MSiO2 membranes were coated on top of composite interlayers supported by porous α-alumina discs. The discs are 5 mm-thick and 30 mm in diameter, with a porosity of 40% and an average pore size of 100 nm. ZrO2-MSiO2 membranes were effectively prepared by dip-coating the substrates in three-fold ethanol-diluted silica sol for 7 s, then drying and calcining them. Each sample was dried at 30 °C for 3 h before being calcined at 400 °C in a temperature-controlled furnace in a N2 environment with a ramping rate of 0.5 °C·min −1 and a dwell period of 2 h. The dip-coating-drying-calcining process was repeated three times. Figure 1 demonstrates the preparation process of the ZrO2-MSiO2 materials/membranes. The ZrO2-MSiO2 membranes with nZr = 0 are also referred to as "MSiO2" membranes.

Steam Treatment and Regeneration of ZrO2-MSiO2 Membranes
The ZrO2-MSiO2 membranes were subjected to a 7-day steam stability test in which they were placed into saturated steam at 25 °C. After steam treatment, for thermal regeneration of ZrO2-MSiO2 membranes, they were processed at a calcination temperature of 350 °C, with the same calcination technique as before. The gas permeances of ZrO2-MSiO2 membranes were investigated after steam treatment and regeneration, respectively.

Characterizations
Using Fourier transform infrared spectroscopy, the functional groups of ZrO2-MSiO2 materials were characterized (FTIR, Spotlight 400 and Frontier, PerkinElmer Corporation, Waltham, MA, US), and the wavelength measuring range was 400 to 4000 cm −1 using the KBr compression technique. Using a Rigaku D/max-2550pc X-ray diffractometer (XRD, Rigaku D/max-2550pc, Hitachi, Tokyo, Japan) with CuKα radiation at 40 kV and 40 mA, the ZrO2-MSiO2 materials' phase structure was found. The X-ray photoelectron spectra

Steam Treatment and Regeneration of ZrO 2 -MSiO 2 Membranes
The ZrO 2 -MSiO 2 membranes were subjected to a 7-day steam stability test in which they were placed into saturated steam at 25 • C. After steam treatment, for thermal regeneration of ZrO 2 -MSiO 2 membranes, they were processed at a calcination temperature of 350 • C, with the same calcination technique as before. The gas permeances of ZrO 2 -MSiO 2 membranes were investigated after steam treatment and regeneration, respectively.

Characterizations
Using Fourier transform infrared spectroscopy, the functional groups of ZrO 2 -MSiO 2 materials were characterized (FTIR, Spotlight 400 and Frontier, PerkinElmer Corporation, Waltham, MA, USA), and the wavelength measuring range was 400 to 4000 cm −1 using the KBr compression technique. Using a Rigaku D/max-2550pc X-ray diffractometer (XRD, Rigaku D/max-2550pc, Hitachi, Tokyo, Japan) with CuKα radiation at 40 kV and 40 mA, the ZrO 2 -MSiO 2 materials' phase structure was found. The X-ray photoelectron spectra (XPS) were acquired on a K-Alpha X-ray photoelectron spectroscope from Thermo Fisher Scientific with AlKα excitation and were calibrated regarding the signal of adventitious carbon (XPS, ESCALAB250xi, Thermo Scientific, Waltham, MA, USA). The binding energy estimates were derived using the C (1s) line at 284.6 eV as the reference point. Transmission electron microscopy (TEM, JEM 2100F, JEOL, Tokyo, Japan) was utilized to investigate the ZrO 2 -MSiO 2 powders' crystallization. Operating at 5 kV, scanning electron microscopy (SEM, JEOL JSM-6300, Hitachi, Tokyo, Japan) was utilized to study the surface morphologies of the ZrO 2 -MSiO 2 membranes. N 2 adsorption-desorption measurements were conducted using an automated Micromeritics, ASAP2020 analyzer (ASAP 2020, Micromeritics, Norcross, GA, USA). The ZrO 2 -MSiO 2 materials' BET surface area, pore volume, and pore size distribution were determined. Figure 2 is a schematic of the experimental setup used to evaluate the performance of single gas permeation. Prior to the experiment, the pressure and temperature were set to the desired values for thirty minutes to allow the gas permeation to stabilize. The permeation properties of MSiO 2 and ZrO 2 -MSiO 2 membranes were evaluated using H 2 , CO 2 , and N 2 . The gas permeability was determined based on the outlet gas flow. The gas permselectivity values (ideal permselectivities) were calculated by the permeance ratio between two gases. (XPS) were acquired on a K-Alpha X-ray photoelectron spectroscope from Thermo Fisher Scientific with AlKα excitation and were calibrated regarding the signal of adventitious carbon (XPS, ESCALAB250xi, Thermo Scientific, Waltham, MA, USA). The binding energy estimates were derived using the C (1s) line at 284.6 eV as the reference point. Transmission electron microscopy (TEM, JEM 2100F, JEOL, Tokyo, Japan) was utilized to investigate the ZrO2-MSiO2 powders' crystallization. Operating at 5 kV, scanning electron microscopy (SEM, JEOL JSM-6300, Hitachi, Tokyo, Japan) was utilized to study the surface morphologies of the ZrO2-MSiO2 membranes. N2 adsorption-desorption measurements were conducted using an automated Micromeritics, ASAP2020 analyzer (ASAP 2020, Micromeritics, Norcross, GA, USA). The ZrO2-MSiO2 materials' BET surface area, pore volume, and pore size distribution were determined. Figure 2 is a schematic of the experimental setup used to evaluate the performance of single gas permeation. Prior to the experiment, the pressure and temperature were set to the desired values for thirty minutes to allow the gas permeation to stabilize. The permeation properties of MSiO2 and ZrO2-MSiO2 membranes were evaluated using H2, CO2, and N2. The gas permeability was determined based on the outlet gas flow. The gas permselectivity values (ideal permselectivities) were calculated by the permeance ratio between two gases.

Chemical Structure Analysis
FTIR spectra were used to investigate the functional groups of ZrO2-MSiO2 materials. The FTIR spectra of ZrO2-MSiO2 materials containing various nZr contents are displayed in Figure 3. The absorption peak at around 3448 cm −1 was assigned to the stretching and bending vibration of the -OH group from the absorbed water. The absorption peak at 1630 cm −1 corresponds to Si-OH and Zr-OH on the surface of ZrO2-MSiO2 materials [36]. The antisymmetric stretching vibration absorption peak -CH3 at 2985 cm −1 was mainly from unhydrolyzed TEOS and MTES. The absorption peak at 1278 cm −1 was attributed to the Si-CH3 group. It is also the main hydrophobic functional group of the membrane. The absorption peak observed at 1050 cm −1 was attributed to the Si-O-Si bond [37]. Compared with the materials with nZr = 0, the materials with nZr = 0.08-0.5 all showed a new absorption peak at the wavenumber of 448 cm −1 . This was related to the formation of Zr-O bonds [38]. Meanwhile, with the increase of nZr, the peak at 1050 cm −1 shifted to around 1100 cm −1 . This may be ascribed to the fact that partial substitution of Zr atoms for Si atoms in the Si-O-Si network to form Zr-O-Si bonds occurred [39], breaking the symmetry of SiO2 and leading to the shift of peak positions. However, there was no obvious Zr-O-Si bond in the FTIR spectrum of ZrO2-MSiO2 materials due to the overlap of the Zr-O-Si bond with Si-O-

Chemical Structure Analysis
FTIR spectra were used to investigate the functional groups of ZrO 2 -MSiO 2 materials. The FTIR spectra of ZrO 2 -MSiO 2 materials containing various n Zr contents are displayed in Figure 3. The absorption peak at around 3448 cm −1 was assigned to the stretching and bending vibration of the -OH group from the absorbed water. The absorption peak at 1630 cm −1 corresponds to Si-OH and Zr-OH on the surface of ZrO 2 -MSiO 2 materials [36]. The antisymmetric stretching vibration absorption peak -CH 3 at 2985 cm −1 was mainly from unhydrolyzed TEOS and MTES. The absorption peak at 1278 cm −1 was attributed to the Si-CH 3 group. It is also the main hydrophobic functional group of the membrane. The absorption peak observed at 1050 cm −1 was attributed to the Si-O-Si bond [37]. Compared with the materials with n Zr = 0, the materials with n Zr = 0.08-0.5 all showed a new absorption peak at the wavenumber of 448 cm −1 . This was related to the formation of Zr-O bonds [38]. Meanwhile, with the increase of n Zr , the peak at 1050 cm −1 shifted to around 1100 cm −1 . This may be ascribed to the fact that partial substitution of Zr atoms for Si atoms in the Si-O-Si network to form Zr-O-Si bonds occurred [39], breaking the symmetry of SiO 2 and leading to the shift of peak positions. However, there was no obvious Zr-O-Si bond in the FTIR spectrum of ZrO 2 -MSiO 2 materials due to the overlap of the Zr-O-Si bond with Si-O-Si [40]. Furthermore, the decrease in the intensity of the silanol band at 779 and 835 cm −1 with increasing n Zr could be attributed to the substitution of Si-OH bonds by Zr-O-Si bonds [41]. It demonstrates the formation of Zr-O-Si bonds in the produced materials. Si [40]. Furthermore, the decrease in the intensity of the silanol band at 779 and 835 cm −1 with increasing nZr could be attributed to the substitution of Si-OH bonds by Zr-O-Si bonds [41]. It demonstrates the formation of Zr-O-Si bonds in the produced materials.

Phase Structure Analysis
The XRD patterns of the ZrO2-MSiO2 materials with varied nZr are presented in Figure  4. The peaks of amorphous SiO2 were concentrated at 2θ = 23.1° [42]. The SiO2 peak moved progressively towards higher 2θ values as nZr rose, and it slowly dropped in intensity. This is attributable to the replacement of the portion of silicon atoms by the inserted Zr atoms, producing Zr-O-Si bonds, resulting in a drop in the SiO2 concentration. The peaks corresponding to a crystalline tetragonal structure of zirconia are clearly apparent in the ZrO2-MSiO2 materials with nZr = 0.15-0.5. The (101), (112), and (202) reflection planes of the body-centered ZrO2 (t-ZrO2) tetragonal phase were ascribed to the large diffraction peaks occurring at 60.2°, 50.7°, and 30.2°, respectively (JCPDS No. 79-1771). XRD analysis demonstrated that with the growth of the Zr concentration, the peak intensity corresponding to t-ZrO2 progressively increased. In other words, the content of t-ZrO2 increased with the growth in Zr content. Combined with the FTIR analysis, the Zr element in ZrO2-MSiO2 materials may exist in the form of Zr-O-Si bonds and t-ZrO2.

Phase Structure Analysis
The XRD patterns of the ZrO 2 -MSiO 2 materials with varied n Zr are presented in Si [40]. Furthermore, the decrease in the intensity of the silanol band at 779 and 835 cm −1 with increasing nZr could be attributed to the substitution of Si-OH bonds by Zr-O-Si bonds [41]. It demonstrates the formation of Zr-O-Si bonds in the produced materials.

Phase Structure Analysis
The XRD patterns of the ZrO2-MSiO2 materials with varied nZr are presented in Figure  4. The peaks of amorphous SiO2 were concentrated at 2θ = 23.1° [42]. The SiO2 peak moved progressively towards higher 2θ values as nZr rose, and it slowly dropped in intensity. This is attributable to the replacement of the portion of silicon atoms by the inserted Zr atoms, producing Zr-O-Si bonds, resulting in a drop in the SiO2 concentration. The peaks corresponding to a crystalline tetragonal structure of zirconia are clearly apparent in the ZrO2-MSiO2 materials with nZr = 0.15-0.5. The (101), (112), and (202) reflection planes of the body-centered ZrO2 (t-ZrO2) tetragonal phase were ascribed to the large diffraction peaks occurring at 60.2°, 50.7°, and 30.2°, respectively (JCPDS No. 79-1771). XRD analysis demonstrated that with the growth of the Zr concentration, the peak intensity corresponding to t-ZrO2 progressively increased. In other words, the content of t-ZrO2 increased with the growth in Zr content. Combined with the FTIR analysis, the Zr element in ZrO2-MSiO2 materials may exist in the form of Zr-O-Si bonds and t-ZrO2.  To further investigate the presence of Zr and Si species in the ZrO 2 -MSiO 2 materials, the XPS measurement was conducted. The survey XPS spectrum of ZrO 2 -MSiO 2 material with n Zr = 0.15 is shown in Figure 5. Figure 5 demonstrates that C, O, Si, and Zr elements are present in the ZrO 2 -MSiO 2 material, which indicates the successful incorporation of Zr into the silica frameworks. Figure 6 presents the Si 2p and Zr 3d XPS spectra of the ZrO 2 -MSiO 2 sample with n Zr = 0.15. In Figure 6a, the peaks at the binding energies of 102.8 and 104.7 eV correspond to Si-C and Si-O bonds, respectively. In Figure 6b, the peaks at 186.6 and 183.3 eV correspond to the Zr-O 3d 3/2 and Zr-O 3d 5/2 peaks, respectively.
To further investigate the presence of Zr and Si species in the ZrO2-MSiO2 materials, the XPS measurement was conducted. The survey XPS spectrum of ZrO2-MSiO2 material with nZr = 0.15 is shown in Figure 5. Figure 5 demonstrates that C, O, Si, and Zr elements are present in the ZrO2-MSiO2 material, which indicates the successful incorporation of Zr into the silica frameworks. Figure 6 presents the Si 2p and Zr 3d XPS spectra of the ZrO2-MSiO2 sample with nZr = 0.15. In Figure 6a, the peaks at the binding energies of 102.8 and 104.7 eV correspond to Si-C and Si-O bonds, respectively. In Figure 6b, the peaks at 186.6 and 183.3 eV correspond to the Zr-O 3d3/2 and Zr-O 3d5/2 peaks, respectively.

TEM Analysis
The TEM micrographs of the ZrO2-MSiO2 material with nZr = 0 and 0.15 at 400 °C under nitrogen atmosphere are illustrated in Figure 7. Figure 7a depicts that the silica particles in MSiO2 materials are amorphous, while in Figure 7b, a small amount of particles with darker color appear and are mixed in the silica skeleton, which may be due to the presence of t-ZrO2. Overall, the ZrO2-MSiO2 materials with nZr = 0.15 still maintained the amorphous state. To further investigate the presence of Zr and Si species in the ZrO2-MSiO2 materials, the XPS measurement was conducted. The survey XPS spectrum of ZrO2-MSiO2 material with nZr = 0.15 is shown in Figure 5. Figure 5 demonstrates that C, O, Si, and Zr elements are present in the ZrO2-MSiO2 material, which indicates the successful incorporation of Zr into the silica frameworks. Figure 6 presents the Si 2p and Zr 3d XPS spectra of the ZrO2-MSiO2 sample with nZr = 0.15. In Figure 6a, the peaks at the binding energies of 102.8 and 104.7 eV correspond to Si-C and Si-O bonds, respectively. In Figure 6b, the peaks at 186.6 and 183.3 eV correspond to the Zr-O 3d3/2 and Zr-O 3d5/2 peaks, respectively.

TEM Analysis
The TEM micrographs of the ZrO2-MSiO2 material with nZr = 0 and 0.15 at 400 °C under nitrogen atmosphere are illustrated in Figure 7. Figure 7a depicts that the silica particles in MSiO2 materials are amorphous, while in Figure 7b, a small amount of particles with darker color appear and are mixed in the silica skeleton, which may be due to the presence of t-ZrO2. Overall, the ZrO2-MSiO2 materials with nZr = 0.15 still maintained the amorphous state.

TEM Analysis
The TEM micrographs of the ZrO 2 -MSiO 2 material with n Zr = 0 and 0.15 at 400 • C under nitrogen atmosphere are illustrated in Figure 7. Figure 7a depicts that the silica particles in MSiO 2 materials are amorphous, while in Figure 7b, a small amount of particles with darker color appear and are mixed in the silica skeleton, which may be due to the presence of t-ZrO 2 . Overall, the ZrO 2 -MSiO 2 materials with n Zr = 0.15 still maintained the amorphous state.

Pore Structure Analysis
The N 2 adsorption-desorption isotherm of the ZrO 2 -MSiO 2 materials with various n Zr are shown in Figure 8a. According to the Brunauer-Deming-Deming-Teller (BDDT) classification, the ZrO 2 -MSiO 2 materials showed a type I adsorption isotherm, while n Zr = 0 indicated the formation of microporous structures. The isotherms for the four samples (n Zr = 0.08-0.5) all showed a similar trend, which could be categorized as type IV isotherms. However, the shapes of the hysteresis loops for the four samples were different, implying the variation of pore structures. A significant proportion of adsorption occurred in the range of low relative pressure, P/P 0 < 0.1, indicating that the materials contain a large quantity of micropores. The shape of the hysteresis loop of the ZrO 2 -MSiO 2 materials Nanomaterials 2022, 12, 2159 8 of 18 with n Zr = 0.5 was altered, indicating the presence of larger mesopores or macropores. In addition, the distributions of pore size for all samples are depicted in Figure 8b. It is found that the samples with n Zr = 0.08-0.5 showed a broader pore size distribution and a larger mean pore size than the samples with n Zr = 0. The conclusion was also confirmed by the pore structure parameters of ZrO 2 -MSiO 2 materials with various n Zr in Table 1. The average pore size, BET specific surface area, and total pore volume of the ZrO 2 -MSiO 2 materials gradually increased with n Zr = 0.08 and 0.15, and the pore size distribution became wider. However, the total pore volume of the ZrO 2 -MSiO 2 materials with n Zr = 0.3 and 0.5 decreased instead. The fact is that the bond length of Zr-O (1.78 Å) was slightly longer than that of Si-O (1.64 Å) [43]. Figure 9 shows the molecular structure models of MSiO 2 , ZrO 2 -MSiO 2 , and t-ZrO 2 crystallites, respectively. Hence, the formation of Zr-O-Si bonds contributes to the formation of pores. With the increase of Zr content, more and more t-ZrO 2 crystallites were formed and distributed in the framework of MSiO 2 materials, and the internal pore structure of the ZrO 2 -MSiO 2 materials was hindered from shrinking and pore collapse, resulting in the decrease of the BET surface area and pore volume. From Table 1, it can be seen that the ZrO 2 -MSiO 2 materials with n Zr = 0.15 had the maximal total pore volume (0.43 cm 3 ·g −1 ), BET surface area (616.77 m 2 ·g −1 ), and the minimum mean pore size (2.19 nm).

Pore Structure Analysis
The N2 adsorption-desorption isotherm of the ZrO2-MSiO2 materials with various are shown in Figure 8a. According to the Brunauer-Deming-Deming-Teller (BDDT) cl sification, the ZrO2-MSiO2 materials showed a type I adsorption isotherm, while nZr indicated the formation of microporous structures. The isotherms for the four samples ( = 0.08-0.5) all showed a similar trend, which could be categorized as type IV isotherm However, the shapes of the hysteresis loops for the four samples were different, imply the variation of pore structures. A significant proportion of adsorption occurred in range of low relative pressure, P/P0 < 0.1, indicating that the materials contain a large qu tity of micropores. The shape of the hysteresis loop of the ZrO2-MSiO2 materials with = 0.5 was altered, indicating the presence of larger mesopores or macropores. In additi the distributions of pore size for all samples are depicted in Figure 8b. It is found that samples with nZr = 0.08-0.5 showed a broader pore size distribution and a larger me pore size than the samples with nZr = 0. The conclusion was also confirmed by the p structure parameters of ZrO2-MSiO2 materials with various nZr in Table 1. The avera pore size, BET specific surface area, and total pore volume of the ZrO2-MSiO2 materi gradually increased with nZr = 0.08 and 0.15, and the pore size distribution became wid However, the total pore volume of the ZrO2-MSiO2 materials with nZr = 0.3 and 0.5 creased instead. The fact is that the bond length of Zr-O (1.78 Å) was slightly longer th that of Si-O (1.64 Å) [43]. Figure 9 shows the molecular structure models of MSiO2, Zr MSiO2, and t-ZrO2 crystallites, respectively. Hence, the formation of Zr-O-Si bonds c tributes to the formation of pores. With the increase of Zr content, more and more t-Z    (a) (b) Figure 8. (a) The nitrogen adsorption-desorption isotherms and (b) the corresponding pore size distribution curves for the ZrO2-MSiO2 materials with various nZr.

Gas Permselectivity Analysis
3.5.1. The Influence of nZr Figure 10 depicts the influence of nZr on the gas permeabilities and H2 permselectivities of ZrO2-MSiO2 membranes with varying nZr and 0% DCCA addition at 25 °C and 0.1 MPa. In Figure 10a, with the increase of nZr, the H2, CO2, and N2 permeances of the samples increased until nZr = 0.15, and then decreased. Compared with the MSiO2 membranes (nZr = 0), the H2, CO2, and N2 permeance of the ZrO2-MSiO2 membranes with nZr = 0.15 increased by 50.95%, 26.74%, and 36.36%, respectively. From the pore structure analysis (Table 1), the overall pore volumes of the ZrO2-MSiO2 membranes grew somewhat with increasing nZr until nZr = 0.15, and then decreased, which can explain why the ZrO2-MSiO2 membranes with nZr = 0.15 had the highest permeance to each gas. For the same In Figure 10a, with the increase of n Zr , the H 2 , CO 2 , and N 2 permeances of the samples increased until n Zr = 0.15, and then decreased. Compared with the MSiO 2 membranes (n Zr = 0), the H 2 , CO 2 , and N 2 permeance of the ZrO 2 -MSiO 2 membranes with n Zr = 0.15 increased by 50.95%, 26.74%, and 36.36%, respectively. From the pore structure analysis (Table 1), the overall pore volumes of the ZrO 2 -MSiO 2 membranes grew somewhat with increasing n Zr until n Zr = 0.15, and then decreased, which can explain why the ZrO 2 -MSiO 2 membranes with n Zr = 0.15 had the highest permeance to each gas. For the same membrane, the order of gas molecular permeance is H 2 > CO 2 > N 2 . Gas permeance decreased with increasing d k (0.289, 0.33, and 0.364 nm, respectively), indicating that all membranes exhibited molecular sieve properties. However, when n Zr ≥ 0.15, the permeance of CO 2 decreased more closely to that of N 2 . This behavior was related to the fact that following heat treatment at 400 • C, the Zr-O-Si bonds and t-ZrO 2 crystallites generated in the ZrO 2 -SiO 2 membranes will generate a significant number of Brønsted acid sites. High acidity leads to a reduction in the affinity of the membranes for CO 2 , hence lowering the CO 2 permeance [44].
It can be observed from Figure 10b that compared with MSiO 2 membranes, the H 2 /CO 2 and H 2 /N 2 permselectivities of ZrO 2 -MSiO 2 membranes with n Zr = 0.15 increased by 22.93% and 33.04%, respectively. Combined with the previous characterization test, it was found that the ZrO 2 -MSiO 2 membranes with n Zr = 0.15 had a good pore structure, which is beneficial to improve the permselectivity of gas. In addition, the acidic sites formed by the ZrO 2 -MSiO 2 membranes reduced the affinity of the membranes for CO 2 and helped to separate it from H 2 . However, the permselectivities of ZrO 2 -MSiO 2 membranes after n Zr = 0.15 showed a decreasing trend. Compared with the ZrO 2 -MSiO 2 membranes with n Zr = 0.15, the H 2 /CO 2 and H 2 /N 2 permselectivities of the membranes with n Zr = 0.5 decreased by 9.35% and 20.15%, respectively. As a result, just because the n Zr concentration is larger, it does not indicate that the separation effect is better. Since the Zr-O and Si-O bonds in zirconium-substituted siloxane rings are longer than in pure siloxane rings, for the ZrO 2 -MSiO 2 membranes with n Zr = 0.5, the number of siloxane rings containing Zr increased, and the pore size of the membranes became larger. Meanwhile, a large number of t-ZrO 2 crystals were produced, which led to the shrinkage of the pore structure inside the membranes and the collapse of the pores, resulting in the decrease of the permselectivities of the membranes.
which is beneficial to improve the permselectivity of gas. In addition, the acidic sites formed by the ZrO2-MSiO2 membranes reduced the affinity of the membranes for CO2 and helped to separate it from H2. However, the permselectivities of ZrO2-MSiO2 membranes after nZr = 0.15 showed a decreasing trend. Compared with the ZrO2-MSiO2 membranes with nZr = 0.15, the H2/CO2 and H2/N2 permselectivities of the membranes with nZr = 0.5 decreased by 9.35% and 20.15%, respectively. As a result, just because the nZr concentration is larger, it does not indicate that the separation effect is better. Since the Zr-O and Si-O bonds in zirconium-substituted siloxane rings are longer than in pure siloxane rings, for the ZrO2-MSiO2 membranes with nZr = 0.5, the number of siloxane rings containing Zr increased, and the pore size of the membranes became larger. Meanwhile, a large number of t-ZrO2 crystals were produced, which led to the shrinkage of the pore structure inside the membranes and the collapse of the pores, resulting in the decrease of the permselectivities of the membranes. SEM images of surface topography for MSiO2 and ZrO2-MSiO2 (nZr = 0.15) membranes calcined at 400 °C are shown in Figure 11. Compared to the MSiO2 membranes, the particle size and distribution of the ZrO2-MSiO2 membranes were more uniform, the membranes' surfaces had no obvious defects, and the surface was uniform and smooth. The particle size of MSiO2 membranes was between 1.1 and 5.8 nm, while the particle size of ZrO2-MSiO2 membranes was between 1.3 and 8.9 nm. The formed ZrO2-MSiO2 membranes with a smooth surface and uniform membrane pores were more conducive to the gas permselectivitiy. SEM images of surface topography for MSiO 2 and ZrO 2 -MSiO 2 (n Zr = 0.15) membranes calcined at 400 • C are shown in Figure 11. Compared to the MSiO 2 membranes, the particle size and distribution of the ZrO 2 -MSiO 2 membranes were more uniform, the membranes' surfaces had no obvious defects, and the surface was uniform and smooth. The particle size of MSiO 2 membranes was between 1.1 and 5.8 nm, while the particle size of ZrO 2 -MSiO 2 membranes was between 1.3 and 8.9 nm. The formed ZrO 2 -MSiO 2 membranes with a smooth surface and uniform membrane pores were more conducive to the gas permselectivitiy.

The Influence of Temperature
The permeances and permselectivities of the MSiO2 and ZrO2-MSiO2 (nZr = 0.15) membranes with 0% DCCA addition at a pressure difference of 0.1 MPa and temperature changing from 25 to 200 °C are shown in Figure 12. Obviously, with the increasing temperature, the permeance of H2 of MSiO2 and ZrO2-MSiO2 membranes increased gradually, as seen in Figure 12a. From 25 to 200 °C, the H2 permeance of MSiO2 and ZrO2-MSiO2 membranes rose by 19.66% and 7.12%, respectively, demonstrating that the H2 permeation behavior in the two membranes followed the activated diffusion transport mechanism. In contrast, the permeabilities of CO2 and N2 were similar to the Knudsen diffusion trend, whereby both slightly decreased. The CO2 permeance of MSiO2 and ZrO2-MSiO2 membranes decreased by 31.46% and 30.20% from 25 to 200 °C, respectively, and the N2 per-

The Influence of Temperature
The permeances and permselectivities of the MSiO 2 and ZrO 2 -MSiO 2 (n Zr = 0.15) membranes with 0% DCCA addition at a pressure difference of 0.1 MPa and temperature changing from 25 to 200 • C are shown in Figure 12. Obviously, with the increasing temperature, the permeance of H 2 of MSiO 2 and ZrO 2 -MSiO 2 membranes increased gradually, as seen in Figure 12a. From 25 to 200 • C, the H 2 permeance of MSiO 2 and ZrO 2 -MSiO 2 membranes rose by 19.66% and 7.12%, respectively, demonstrating that the H 2 permeation behavior in the two membranes followed the activated diffusion transport mechanism. In contrast, the permeabilities of CO 2 and N 2 were similar to the Knudsen diffusion trend, whereby both slightly decreased. The CO 2 permeance of MSiO 2 and ZrO 2 -MSiO 2 membranes decreased by 31.46% and 30.20% from 25 to 200 • C, respectively, and the N 2 permeance decreased by 18.60% and 29.98%, respectively. The major explanations for the decrease in CO 2 and N 2 permeance were the violent movement of molecules and the rise in the mean free path as temperature increased. membranes rose by 19.66% and 7.12%, respectively, demonstrating that the H2 permeation behavior in the two membranes followed the activated diffusion transport mechanism. In contrast, the permeabilities of CO2 and N2 were similar to the Knudsen diffusion trend, whereby both slightly decreased. The CO2 permeance of MSiO2 and ZrO2-MSiO2 membranes decreased by 31.46% and 30.20% from 25 to 200 °C, respectively, and the N2 permeance decreased by 18.60% and 29.98%, respectively. The major explanations for the decrease in CO2 and N2 permeance were the violent movement of molecules and the rise in the mean free path as temperature increased.
In Figure 12b, it can be seen that the permselectivities in the membranes gradually increased with the increase of temperature. At 200 °C, compared with the MSiO2 membranes, the H2/CO2 and H2/N2 permselectivities of the ZrO2-MSiO2 membranes increased by 21.11% and 23.34%, respectively. The above results show that the ZrO2-MSiO2 membranes had better permselectivity and permeance of H2 than the MSiO2 membranes under the same conditions.  In Figure 12b, it can be seen that the permselectivities in the membranes gradually increased with the increase of temperature. At 200 • C, compared with the MSiO 2 membranes, the H 2 /CO 2 and H 2 /N 2 permselectivities of the ZrO 2 -MSiO 2 membranes increased by 21.11% and 23.34%, respectively. The above results show that the ZrO 2 -MSiO 2 membranes had better permselectivity and permeance of H 2 than the MSiO 2 membranes under the same conditions.
Combined with the preceding studies, it was determined that a rise in temperature facilitated the separation of H 2 and that the separation process of H 2 from CO 2 and N 2 is dominated by activation diffusion, which is described by the Arrhenius equation [45]: In Formula (1), P is the permeation rate, E a is the apparent activation energy, and P 0 is a constant, which depends on the pore wall-gas molecule interaction, gas selective layer thickness, and pore shape and tortuosity [46]. For linear fitting, 1000/RT was used as the abscissa and lnP as the ordinate, and the slope of the fitting equation might be used to obtain the apparent activation energy. Figure 13 depicts the Arrhenius fitting diagrams for the three gases. Figure 13 demonstrates that the apparent activation energy of H 2 is positive, while that of various other gases is negative. This is related to the gas-activated transport behavior, whereby there are two parallel transmission channels for gas through the membrane: one is through selective micropores, with gas transport processed by a thermally activated surface diffusion mechanism in the micropore state [47], and the other is through larger pores [48]. The activation energies for CO 2 and N 2 are negative, indicating that there are permeation pathways large enough in these types of membranes to allow the diffusion of these larger gas. It is generally believed that E a is composed of two parts [46], the adsorption heat, Q st , of gas on the surface of the membrane and the activation energy, E m , of gas flowing through the solid surface. The larger the E m is, the harder it is for the gas to diffuse, E a = E m − Q st . Arrhenius equation parameter values are shown in Table 2. Table 2 shows that the E m values of the gases in the ZrO 2 -MSiO 2 membrane are all less than those in the MSiO 2 membrane. It shows that the structure of ZrO 2 -MSiO 2 membranes is not as dense as that of MSiO 2 membranes. This is in good accordance with the N 2 adsorption-desorption results. This finding also shows that ZrO 2 doping successfully diminishes the densification of the SiO 2 network. The higher porosity of the ZrO 2 -MSiO 2 membranes allows the gas to cross the membrane pore barrier using their kinetic energy. Therefore, the gas (H 2 , CO 2 , and N 2 ) permeance of ZrO 2 -MSiO 2 membranes is higher than that of MSiO 2 membranes.
Combined with the preceding studies, it was determined that a rise in temperature facilitated the separation of H2 and that the separation process of H2 from CO2 and N2 is dominated by activation diffusion, which is described by the Arrhenius equation [45]: In Formula (1), P is the permeation rate, Ea is the apparent activation energy, and P0 is a constant, which depends on the pore wall-gas molecule interaction, gas selective layer thickness, and pore shape and tortuosity [46]. For linear fitting, 1000/RT was used as the abscissa and lnP as the ordinate, and the slope of the fitting equation might be used to obtain the apparent activation energy. Figure 13 depicts the Arrhenius fitting diagrams for the three gases.  Figure 13 demonstrates that the apparent activation energy of H2 is positive, while that of various other gases is negative. This is related to the gas-activated transport behavior, whereby there are two parallel transmission channels for gas through the membrane: one is through selective micropores, with gas transport processed by a thermally activated surface diffusion mechanism in the micropore state [47], and the other is through larger pores [48]. The activation energies for CO2 and N2 are negative, indicating that there are permeation pathways large enough in these types of membranes to allow the diffusion of these larger gas. It is generally believed that Ea is composed of two parts [46], the adsorption heat, Qst, of gas on the surface of the membrane and the activation energy, Em, of gas flowing through the solid surface. The larger the Em is, the harder it is for the gas to diffuse, Ea = Em − Qst. Arrhenius equation parameter values are shown in Table 2. Table 2 shows that the Em values of the gases in the ZrO2-MSiO2 membrane are all less than those in the MSiO2 membrane. It shows that the structure of ZrO2-MSiO2 membranes is not as dense as that of MSiO2 membranes. This is in good accordance with the N2 adsorption-desorption results. This finding also shows that ZrO2 doping successfully diminishes the densification of the SiO2 network. The higher porosity of the ZrO2-MSiO2 membranes allows the gas to cross the membrane pore barrier using their kinetic energy. Therefore, the gas (H2, CO2, and N2) permeance of ZrO2-MSiO2 membranes is higher than that of MSiO2 membranes.   Table 3 displays the E a of H 2 , pore diameter, H 2 permeance, and H 2 permselectivities for several membranes/films prepared by other researchers. It is challenging to concurrently enhance the membranes' permselectivity and gas permeability, as seen in Table 3. Generally, the larger the average pore size of the membrane, the higher the permeance to H 2 , which is accompanied by a smaller E a . Meanwhile, the E a of H 2 is related to the interaction of H 2 with the membrane pore wall. It can be seen from Table 3 that the as-prepared ZrO 2 -MSiO 2 membrane has a large H 2 permselectivity compared to other membranes.  Figure 14 illustrates the effect of pressure difference on the gas permeances and H 2 permselectivities of MSiO 2 and ZrO 2 -MSiO 2 (n Zr = 0.15) membranes at 200 • C with 0% DCCA addition. In Figure 14a, it can be observed that the H 2 permeance of the ZrO 2 -MSiO 2 membranes with n Zr = 0.15 improved with the increase of the pressure difference, and the pressure dependence increased. However, the H 2 permeance of MSiO 2 membranes remained basically unchanged. The MSiO 2 and ZrO 2 -MSiO 2 membranes increased their H 2 permeance by 2.16% and 19.96%, respectively, when the pressure was increased from 0.10 to 0.40 MPa. In Figure 14b, the H 2 permselectivities of MSiO 2 membranes did not change significantly, and the H 2 /CO 2 and H 2 /N 2 permselectivities decreased by 5.27% and 1.12%, respectively. However, the H 2 permselectivities of the ZrO 2 -MSiO 2 membranes with n Zr = 0.15 changed greatly, where the H 2 /CO 2 and H 2 /N 2 permselectivities decreased by 22.04% and 21.12%, respectively. Clearly, it can be seen that the permselectivity of the ZrO 2 -MSiO 2 membranes was reduced more than that of the MSiO 2 membranes, that is, the pressure had a relatively small effect on the gas permeation of the MSiO 2 membranes. This phenomenon is attributed to the relatively dense MSiO 2 membranes with micropores as the main component. With the increase of the pressure difference between the two sides of the membranes, the power of the gas passing through the ZrO 2 -MSiO 2 membranes increased, making it easier for the gas to pass through the mesopores or even the macropores, which has a greater impact on the permselectivity of the membranes. At the same time, it is shown that the H 2 diffusion mechanism of ZrO 2 -MSiO 2 membranes was different from that of MSiO 2 membranes due to the influence of doping ZrO 2 . The gas transport in the ZrO 2 -MSiO 2 membranes follows the surface diffusion mechanism. In addition, when the pressure difference increased to 0.4 MPa, the permselectivities of H 2 /CO 2 and H 2 /N 2 were still higher than their respective Knudsen diffusion (4.69 and 3.74, respectively), indicating that they still have good gas permeation performance under high pressure.  Comparing and analyzing the previous results, it was found that the addition of DCCA (glycerol) by the sol-gel method can effectively reduce the uneven shrinkage of the membrane when it is heated during firing. The gas permeances and H2 permselectivities of ZrO2-MSiO2 membranes (nZr = 0.15) with various DCCA additions at 200 °C and 0.1 MPa are shown in Figure 15.  Figure 15 demonstrates that the H2 permeance of the ZrO2-MSiO2 membranes reduced by 21.30% as the DCCA addition increased from 0 to 30%. The permselectivities of H2/CO2 and H2/N2 increased by 21.77% and 14.07%, respectively. However, compared with the 20% membranes, the H2/CO2 and H2/N2 permselectivities of the membranes with the addition of 30% only increased by 2.14% and 0.28%, respectively. Figure 16 shows the relationship between DCCA (GL) contents and FH2 × αH2 (FH2 is the H2 permeance, αH2 is the permselective of H2). It was clearly observed that the addition of DCCA from 0 to 20% enhanced the FH2 × αH2 value, and after more than 20%, the FH2 × αH2 value showed a lower  Figure 15 demonstrates that the H 2 permeance of the ZrO 2 -MSiO 2 membranes reduced by 21.30% as the DCCA addition increased from 0 to 30%. The permselectivities of H 2 /CO 2 and H 2 /N 2 increased by 21.77% and 14.07%, respectively. However, compared with the 20% membranes, the H 2 /CO 2 and H 2 /N 2 permselectivities of the membranes with the addition of 30% only increased by 2.14% and 0.28%, respectively. Figure 16 shows the relationship between DCCA (GL) contents and F H2 × α H2 (F H2 is the H 2 permeance, α H2 is the permselective of H 2 ). It was clearly observed that the addition of DCCA from 0 to 20% enhanced the F H2 × α H2 value, and after more than 20%, the F H2 × α H2 value showed a lower level. This is attributed to the fact that the addition of glycerol will gradually surround the sol particles, reduce the agglomeration caused by the collision of the colloid particles, accelerate the creation of the sol-gel network, and improve the stability of the sol structure [52]. In addition, the membrane layer collapsed and cracked easily during the drying and calcination processes, and the addition of glycerol can effectively reduce the liquid-gas surface tension to a certain extent, thereby protecting the gel skeleton from deformation [55]. However, the particles of the sol were surrounded by steric effects when too much GL was added, making the sol sticky and difficult to dry to form a membrane. The extension of the drying time makes the cross-linking of the sol more thorough and eventually leads to the densification of the membrane, which is not conducive to gas separation. From the above point of view, the membranes with 20% DCCA addition were the more worthy choice. level. This is attributed to the fact that the addition of glycerol will gradually surround the sol particles, reduce the agglomeration caused by the collision of the colloid particles, accelerate the creation of the sol-gel network, and improve the stability of the sol structure [52]. In addition, the membrane layer collapsed and cracked easily during the drying and calcination processes, and the addition of glycerol can effectively reduce the liquid-gas surface tension to a certain extent, thereby protecting the gel skeleton from deformation [55]. However, the particles of the sol were surrounded by steric effects when too much GL was added, making the sol sticky and difficult to dry to form a membrane. The extension of the drying time makes the cross-linking of the sol more thorough and eventually leads to the densification of the membrane, which is not conducive to gas separation. From the above point of view, the membranes with 20% DCCA addition were the more worthy choice.   Figure 17 shows the effects of steam treatment and thermal regeneration on the gas permeances (H 2 , CO 2 , and N 2 ) and H 2 permselectivities of MSiO 2 and ZrO 2 -MSiO 2 (n Zr = 0.15) membranes with 0% DCCA addition at a pressure difference of 0.1 MPa and 25 • C. The permeances of H 2 , CO 2 , and N 2 for MSiO 2 and ZrO 2 -MSiO 2 membranes appear to have reduced after steam treatment. After steam aging for 7 days, the permeance of H 2 for MSiO 2 and ZrO 2 -MSiO 2 membranes dropped by 20.63% and 3.70%, respectively, as compared to untreated fresh samples, and the permselectivities of H 2 /CO 2 and H 2 /N 2 for MSiO 2 membranes decreased by 1.59% and 1.04%, respectively, whereas those of ZrO 2 -MSiO 2 membranes increased by 0.09% and 0.43%, respectively.

Conclusions
The ZrO2-MSiO2 membranes were manufactured to enhance the steam stability and H2 permselectivity of SiO2 membranes. It was found that with the increase of ZrO2 content, the pore size distribution of the materials became wider and the average pore size increased, indicating that the doping of ZrO2 had the effect of expanding the pores. The ZrO2-MSiO2 membranes with nZr = 0.15 had a good pore structure and suitable micropore/mesoporous ratio, which is beneficial to improve the permeance of natural gas. At 200 °C, the H2/CO2 and H2/N2 permselectivities of ZrO2-MSiO2 membranes were 79.18% and 26.75% greater than those of MSiO2 membranes, respectively. Furthermore, when the pressure was increased to 0.4 MPa, the permselectivities were still higher than their respective Knudsen diffusion, indicating that they still had good gas permeance at high pressure. With the addition of DCCA from 20% to 30%, the H2/CO2 and H2/N2 permselectivities of ZrO2-MSiO2 membranes only increased by 2.14% and 0.28%, respectively, and the FH2 × αH2 value with 20% addition was the highest. In conclusion, it is worthwhile to choose 20% GL as the DCCA addition for ZrO2-MSiO2 membranes. Compared with the untreated fresh sample, after 7 days of water vapor aging, the permeance of ZrO2-MSiO2 membranes to H2 decreased by only 3.70%, and the permselectivities of H2/CO2 and H2/N2 increased by only 0.09% and 0.43%, respectively. After regeneration at 350 °C, the H2 permeance of the ZrO2-MSiO2 membranes decreased by 1.65%, and the permselectivities of H2/CO2 and H2/N2 increased by 0.08% and 1.21%. It is enough to show that the prepared ZrO2-MSiO2 membranes had a good hydrothermal stability and certain regeneration performance. In the future, the influence of high temperature (for example, ≥300 °C) and mixed gases on the gas permeances and permselectivies of ZrO2-MSiO2 membranes should be explored, which is important to the practical engineering applications.  The gas permeances (H 2 , CO 2 , and N 2 ), as well as the permselectivities of H 2 /CO 2 and H 2 /N 2 for two membranes, all exhibit an increased trend after regeneration by calcination at 350 • C. However, as compared to untreated fresh samples, the H 2 permeances of MSiO 2 and ZrO 2 -MSiO 2 membranes after regeneration dropped by 9.96% and 1.65%, respectively, whereas the permselectivities of H 2 /CO 2 and H 2 /N 2 for MSiO 2 membranes improved by 1.12% and 4.71%, respectively, and those for ZrO 2 -MSiO 2 membranes increased by 0.08% and 1.21%, respectively. The decrease in gas permeances in both membranes suggests that membrane pore shrinking occurs after calcination at 350 • C. Lower permeance and greater permselectivities are produced as a result of the smaller pores. This is attributed to the partial Zr atoms replacing the Si atoms in Si-O-Si to form more stable Zr-O-Si bonds, which further improves the hydrothermal stability of the membrane material. Therefore, the above results indicated that the ZrO 2 -MSiO 2 membranes had a better hydrothermal stability and reproducibility than MSiO 2 membranes.

Conclusions
The ZrO 2 -MSiO 2 membranes were manufactured to enhance the steam stability and H 2 permselectivity of SiO 2 membranes. It was found that with the increase of ZrO 2 content, the pore size distribution of the materials became wider and the average pore size increased, indicating that the doping of ZrO 2 had the effect of expanding the pores. The ZrO 2 -MSiO 2 membranes with n Zr = 0.15 had a good pore structure and suitable micropore/mesoporous ratio, which is beneficial to improve the permeance of natural gas. At 200 • C, the H 2 /CO 2 and H 2 /N 2 permselectivities of ZrO 2 -MSiO 2 membranes were 79.18% and 26.75% greater than those of MSiO 2 membranes, respectively. Furthermore, when the pressure was increased to 0.4 MPa, the permselectivities were still higher than their respective Knudsen diffusion, indicating that they still had good gas permeance at high pressure. With the addition of DCCA from 20% to 30%, the H 2 /CO 2 and H 2 /N 2 permselectivities of ZrO 2 -MSiO 2 membranes only increased by 2.14% and 0.28%, respectively, and the F H2 × α H2 value with 20% addition was the highest. In conclusion, it is worthwhile to choose 20% GL as the DCCA addition for ZrO 2 -MSiO 2 membranes. Compared with the untreated fresh sample, after 7 days of water vapor aging, the permeance of ZrO 2 -MSiO 2 membranes to H 2 decreased by only 3.70%, and the permselectivities of H 2 /CO 2 and H 2 /N 2 increased by only 0.09% and 0.43%, respectively. After regeneration at 350 • C, the H 2 permeance of the ZrO 2 -MSiO 2 membranes decreased by 1.65%, and the permselectivities of H 2 /CO 2 and H 2 /N 2 increased by 0.08% and 1.21%. It is enough to show that the prepared ZrO 2 -MSiO 2 membranes had a good hydrothermal stability and certain regeneration performance. In the future, the influence of high temperature (for example, ≥300 • C) and mixed gases on the gas permeances and permselectivies of ZrO 2 -MSiO 2 membranes should be explored, which is important to the practical engineering applications.