E ﬃ ciency Separation Process of H 2 / CO 2 / CH 4 Mixtures by a Hollow Fiber Dual Membrane Separator

: Hydrogen puriﬁcation and CO 2 capture are of great signiﬁcance in reﬁneries and pre-combustion power plants. A dual membrane separator o ﬀ ers an alternative approach for improving H 2 / CO 2 separation e ﬃ ciency. In this work, H 2 / CO 2 / CH 4 ternary gas mixtures separation can be achieved by a dual membrane separator with an integrated polyimide (PI) membrane and polydimethylsiloxane / polyetherimide (PDMS / PEI) composite membrane. A hollow ﬁber dual membrane separation equipment is designed and manufactured. Through the self-designed device, the e ﬀ ects of stage cut, operating temperature, operating pressure, and membrane area ratio on separation performance of dual membrane separator have been studied. The results indicate that, at a high stage cut, a dual membrane separator has obvious advantages over a single membrane separator. Operating temperature has a signiﬁcant impact on gas permeation rates. At 25 ◦ C, a dual membrane separator can obtain the highest purity of H 2 and CO 2 . By increasing operating pressure, the purity and recovery of H 2 and CO 2 can be improved simultaneously. The e ﬀ ect of the membrane area ratio on the performance of the dual membrane separator was studied. When the permeate ﬂows of two membranes are approximately equal by changing the membrane area ratio, the overall performance of the dual membrane separator is the best. On the basis of its synergy in promoting separation, the dual membrane separator holds great industrial application potential.


Introduction
Hydrogen purification and CO 2 capture are vital processes in the chemical industry and integrated gasification combined cycle (IGCC) power plants [1,2]. Syngas is first produced by the steam methane reforming or coal gasification process, and then converted to a mixture containing H 2 , CO 2 , CH 4 , CO, H 2 O, and hydrocarbons by water-gas shift reaction [3,4]. Solvent absorption is the most well-developed process to capture CO 2 and obtain pure H 2 [5]. However, the concentration of CO 2 in the shifted syngas may be high (up to 45%), which would require large equipment and high energy consumption for desorption [6]. Therefore, more economical and flexible methods are desired.
Gas membrane separation has great prospects in the petrochemical industry owing to extensibility and modularity, cost efficiency, and lower energy requirement [7][8][9]. The first industrial gas separation membrane is a polysulfone hollow fiber membrane for H 2 separation [10]. Nowadays, H 2 -selective membranes are widely used in the field of H 2 /N 2 and H 2 /hydrocarbon separation, which could reach PDMS/PEI hollow fiber composite membranes were prepared by the wet membrane dip-coating method [25]. Coating solution was prepared by dissolving the PDMS polymer resin, cross-linking agent, and catalyst with a weight ratio of 50:5:1 (w/w/w) in i-octane at room temperature to obtain a homogeneous solution. The water of PEI substrates surface was removed by filter paper, and then the wet membrane was submerged in PDMS coating solution. After the membranes were withdrawn and additional solution was dripped off, the composite membranes were cured in an oven for 2 h at 85 • C to complete PDMS cross-linking. The cross-sectional morphologies of the membranes were observed by field-emission scanning electron microscope (FE-SEM, Hitachi S4800, Tokyo, Japan).

Design and Packaging of Dual Membrane Separator
As shown in Figure 1, the dual membrane separator used in this study consisted of a stainless steel casing and two aluminum head. One side of several hollow fiber membranes was bundled with absorbent cotton, put into aluminum heads, and then sealed with epoxy resin. The other side of the membranes was blocked with epoxy resin. To assemble a dual membrane separator, the two heads equipped with PDMS/PEI hollow fiber membranes and PI hollow fiber membranes needed to be placed at the two ends of the stainless steel casing, respectively, and sealed by O-rings and flanges. The separator was connected to test the system to study its gas separation performance. The head of the PI membranes can be taken out for testing as a PDMS/PEI single membrane separator, and vice versa.
Processes 2020, 8, x FOR PEER REVIEW 3 of 12 PDMS/PEI hollow fiber composite membranes were prepared by the wet membrane dip-coating method [25]. Coating solution was prepared by dissolving the PDMS polymer resin, cross-linking agent, and catalyst with a weight ratio of 50:5:1 (w/w/w) in i-octane at room temperature to obtain a homogeneous solution. The water of PEI substrates surface was removed by filter paper, and then the wet membrane was submerged in PDMS coating solution. After the membranes were withdrawn and additional solution was dripped off, the composite membranes were cured in an oven for 2 h at 85 °C to complete PDMS cross-linking. The cross-sectional morphologies of the membranes were observed by field-emission scanning electron microscope (FE-SEM, Hitachi S4800, Tokyo, Japan).

Design and Packaging of Dual Membrane Separator
As shown in Figure 1, the dual membrane separator used in this study consisted of a stainless steel casing and two aluminum head. One side of several hollow fiber membranes was bundled with absorbent cotton, put into aluminum heads, and then sealed with epoxy resin. The other side of the membranes was blocked with epoxy resin. To assemble a dual membrane separator, the two heads equipped with PDMS/PEI hollow fiber membranes and PI hollow fiber membranes needed to be placed at the two ends of the stainless steel casing, respectively, and sealed by O-rings and flanges. The separator was connected to test the system to study its gas separation performance. The head of the PI membranes can be taken out for testing as a PDMS/PEI single membrane separator, and vice versa.

Experimental Set-Up and Analysis System
Permeation and separation experiments were carried out in the setup represented schematically in Figure 2. The feed was introduced and removed from the shell. PI membranes permeate and feed kept a co-current flow pattern, while PDMS/PEI membranes kept a countercurrent flow pattern. The feed pressure and flow were adjusted by a pressure regulator and a back pressure regulator. The separator was immersed in a thermostatic water bath. The gas permeate or retentate volumetric flow rate was measured by a soap bubble flow meter. The design of the inlet section offered the possibility of testing pure and mixed gases. The pure gas permeation experiments were conducted under the operating conditions of 0.2 MPa feed pressure and 25 °C. The retentate was plugged and only the flow rate of permeate streams was measured. The pure gas test data could be used to determine the quality of the package and membranes. In the mixed gas experiment, in order to simulate syngas of water-gas shift reaction, H2/CO2/CH4 three-component gas mixtures with a molar ratio of 45:40:15 were used as feed gas. This process required an opened retentate and the flow rates and composition of the two permeates and the retentate needed to be measured. The gas composition of each pipeline was analyzed by gas chromatography equipped with TCD (Thermal conductivity detector) and FID (Flame ionization detecto) dual detectors (GC 7900, Techcomp Scientific Instruments Co., Ltd., Shenyang, China), and three samples were taken at each sampling point. Before starting the experiment, the pipeline was purged with nitrogen for 10 min. After changing the gas, sufficient

Experimental Set-Up and Analysis System
Permeation and separation experiments were carried out in the setup represented schematically in Figure 2. The feed was introduced and removed from the shell. PI membranes permeate and feed kept a co-current flow pattern, while PDMS/PEI membranes kept a countercurrent flow pattern. The feed pressure and flow were adjusted by a pressure regulator and a back pressure regulator. The separator was immersed in a thermostatic water bath. The gas permeate or retentate volumetric flow rate was measured by a soap bubble flow meter. The design of the inlet section offered the possibility of testing pure and mixed gases. The pure gas permeation experiments were conducted under the operating conditions of 0.2 MPa feed pressure and 25 • C. The retentate was plugged and only the flow rate of permeate streams was measured. The pure gas test data could be used to determine the quality of the package and membranes. In the mixed gas experiment, in order to simulate syngas of water-gas shift reaction, H 2 /CO 2 /CH 4 three-component gas mixtures with a molar ratio of 45:40:15 were used as feed gas. This process required an opened retentate and the flow rates and composition of the two permeates and the retentate needed to be measured. The gas composition of each pipeline was analyzed by gas chromatography equipped with TCD (Thermal conductivity detector) and FID (Flame ionization detecto) dual detectors (GC 7900, Techcomp Scientific Instruments Co., Ltd., Shenyang, China), and three samples were taken at each sampling point. Before starting the experiment, the pipeline was purged with nitrogen for 10 min. After changing the gas, sufficient purge was also purge was also required to eliminate the effect of the previous gas on the membrane material, and the recording was started after the flow rate and composition were stable.

Theoretical Analysis
The performances of the membranes were evaluated in terms of the permeation rate and the separation factor.
The permeability of pure gas i is determined by the following equation: The permeability of gas i in the mixed gas is determined by the following equation: where Ji is the permeation rate GPU (Gas permeation unit), GPU = 10 −6 × cm 3 •cm −2 •s −1 •cmHg −1 ); QP is the gas permeate volumetric flow rate (cm 3 •s −1 ); A is membrane effective surface area (cm 2 ); pF and pP are the pressure on the feed and permeate, respectively (cmHg); and fi , xi, and yi are the mole fractions of gas component i on the feed, retentate, and permeate, respectively. αA/B, as the selectivity of gas pair A/B, is defined as the ratio of the gas permeation rate of two gases, which is represented as below.
The separation efficiency of the membrane separator is expressed by the purity and recovery of the product. Dual membrane separator can obtain two enriched products at the same time: H2 product is obtained on the permeation side of PI membranes, and CO2 product is obtained on the permeation side of PDMS/PEI membranes.
The recovery of gas on the permeate sides: where QF and QR are the feed and retentate volumetric flow rate, respectively (cm 3 •s −1 ). The stage cut is defined as the relative magnitude between permeate and feed flow rate, which can be calculated by the following equation:

Theoretical Analysis
The performances of the membranes were evaluated in terms of the permeation rate and the separation factor.
The permeability of pure gas i is determined by the following equation: The permeability of gas i in the mixed gas is determined by the following equation: where J i is the permeation rate GPU (Gas permeation unit), GPU = 10 −6 × cm 3 ·cm −2 ·s −1 ·cmHg −1 ); Q P is the gas permeate volumetric flow rate (cm 3 ·s −1 ); A is membrane effective surface area (cm 2 ); p F and p P are the pressure on the feed and permeate, respectively (cmHg); and f i , x i , and y i are the mole fractions of gas component i on the feed, retentate, and permeate, respectively. α A/B , as the selectivity of gas pair A/B, is defined as the ratio of the gas permeation rate of two gases, which is represented as below.
The separation efficiency of the membrane separator is expressed by the purity and recovery of the product. Dual membrane separator can obtain two enriched products at the same time: H 2 product is obtained on the permeation side of PI membranes, and CO 2 product is obtained on the permeation side of PDMS/PEI membranes.
The recovery of gas ϕ R on the permeate sides: where Q F and Q R are the feed and retentate volumetric flow rate, respectively (cm 3 ·s −1 ). The stage cut θ is defined as the relative magnitude between permeate and feed flow rate, which can be calculated by the following equation: The membrane area ratio R is an important parameter of the dual membrane separator: where I stands for PI membranes and II stands for PDMS/PEI membranes.

Basic Properties of the Two Hollow Fiber Membranes
In this experiment, industrial PI hollow fiber membranes and self-made PDMS/PEI hollow fiber composite membranes were used. On the basis of the pure gas tests, the permeances and ideal selectivities were calculated. The results are summarized in Table 1. Compared with other gases, H 2 through PI membranes and CO 2 through PDMS/PEI membranes have the highest permeances. The cross-section micrographs of the PI membranes and PDMS/PEI composite membranes are shown in Figure 3. The PI displays sponge-like voids from Figure 3a. The PEI substrate displays finger-like voids originating from the inner and outer surfaces from Figure 3b. PDMS can be found well attached to the PEI hollow substrate forming uniform PDMS layers. These results reflect physicochemical properties of the particular polymer: PI is glassy, while PDMS is rubbery [25,26]. For glassy polymers, the mass transfer mainly depends on diffusion selectivity and smaller molecules preferentially pass the membrane. By contrast, solubility predominates the mass transfer through rubbery CO 2 -philic polymers such as PDMS.
The membrane area ratio R is an important parameter of the dual membrane separator: where Ι stands for PI membranes and ΙΙ stands for PDMS/PEI membranes.

Basic Properties of the Two Hollow Fiber Membranes
In this experiment, industrial PI hollow fiber membranes and self-made PDMS/PEI hollow fiber composite membranes were used. On the basis of the pure gas tests, the permeances and ideal selectivities were calculated. The results are summarized in Table 1. Compared with other gases, H2 through PI membranes and CO2 through PDMS/PEI membranes have the highest permeances. The cross-section micrographs of the PI membranes and PDMS/PEI composite membranes are shown in Figure 3. The PI displays sponge-like voids from Figure 3a. The PEI substrate displays finger-like voids originating from the inner and outer surfaces from Figure 3b. PDMS can be found well attached to the PEI hollow substrate forming uniform PDMS layers. These results reflect physicochemical properties of the particular polymer: PI is glassy, while PDMS is rubbery [25,26]. For glassy polymers, the mass transfer mainly depends on diffusion selectivity and smaller molecules preferentially pass the membrane. By contrast, solubility predominates the mass transfer through rubbery CO2-philic polymers such as PDMS.    Figure 4 shows the influence of stage cut on the traditional PI membrane separator and PI/PDMS dual membrane separator at feed pressure of 0.2 MPa and operating temperature of 25 • C. It could be observed that H 2 purity (y H2 ) decreased with the increase of stage cut in the PI membrane separator. At a high stage cut, most of the feed permeated through the membranes, which means a high H 2 recovery. However, it also resulted in the similar concentration of feed and permeate stream. Therefore, the driving force for H 2 permeation was reduced and H 2 purity decreased. However, this phenomenon was not so obvious in the dual membrane separator. Because PDMS/PEI membranes in the dual membrane separator removed CO 2 from feed gas, CO 2 concentration (x CO2 ) in the retentate side was significantly reduced, thereby increasing the driving force for H 2 permeation through PI membranes. Thanks to its enhanced separation, dual membrane separator could simultaneously obtain products with higher purity and higher recovery.  Figure 4 shows the influence of stage cut on the traditional PI membrane separator and PI/PDMS dual membrane separator at feed pressure of 0.2 MPa and operating temperature of 25 °C. It could be observed that H2 purity (yH2) decreased with the increase of stage cut in the PI membrane separator. At a high stage cut, most of the feed permeated through the membranes, which means a high H2 recovery. However, it also resulted in the similar concentration of feed and permeate stream. Therefore, the driving force for H2 permeation was reduced and H2 purity decreased. However, this phenomenon was not so obvious in the dual membrane separator. Because PDMS/PEI membranes in the dual membrane separator removed CO2 from feed gas, CO2 concentration (xCO2) in the retentate side was significantly reduced, thereby increasing the driving force for H2 permeation through PI membranes. Thanks to its enhanced separation, dual membrane separator could simultaneously obtain products with higher purity and higher recovery.

Effect of Operating Temperature
The mixed gas experiments were carried out to investigate the effects of operating temperature on PI and PDMS/PEI membrane performances. The results were obtained at 0.2 MPa and a temperature range of 25-65 °C. As shown in Figure 5a, with the increase of temperature, the permeation rate of all gases through PI membranes increased. From 25 °C to 65 °C, H2 and CO2 permeation rate increased by 81% and 104%, respectively. As a result, the selectivity of H2/CO2 decreased by 11%. In the glassy PI membrane, the diffusion coefficient plays a major role and its temperature dependence follows Arrhenius's law [27]. Large gas molecules require high diffusion activation energy and are more sensitive to temperature. So, the increase of CO2 permeation rate was greater than that of H2. However, CO2 was less affected by temperature in PDMS/PEI membranes (Figure 5b). The effect of temperature on gas permeation rate can also be described by an Arrhenius relationship. CO2 is a condensable gas, and the apparent activation energy in rubbery polymers is often lower than that of H2 and CH4 [25]. As the temperature increases, the permeation rate of H2 increased significantly, and CO2/H2 selectivity in PDMS/PEI membranes decreased.

Effect of Operating Temperature
The mixed gas experiments were carried out to investigate the effects of operating temperature on PI and PDMS/PEI membrane performances. The results were obtained at 0.2 MPa and a temperature range of 25-65 • C. As shown in Figure 5a, with the increase of temperature, the permeation rate of all gases through PI membranes increased. From 25 • C to 65 • C, H 2 and CO 2 permeation rate increased by 81% and 104%, respectively. As a result, the selectivity of H 2 /CO 2 decreased by 11%. In the glassy PI membrane, the diffusion coefficient plays a major role and its temperature dependence follows Arrhenius's law [27]. Large gas molecules require high diffusion activation energy and are more sensitive to temperature. So, the increase of CO 2 permeation rate was greater than that of H 2 . However, CO 2 was less affected by temperature in PDMS/PEI membranes (Figure 5b). The effect of temperature on gas permeation rate can also be described by an Arrhenius relationship. CO 2 is a condensable gas, and the apparent activation energy in rubbery polymers is often lower than that of H 2 and CH 4 [25]. As the temperature increases, the permeation rate of H 2 increased significantly, and CO 2 /H 2 selectivity in PDMS/PEI membranes decreased.  Figure 6 depicts the effect of operating temperature on the purity and recovery of dual membrane separator. It could be observed from Figure 6a that as the temperature rose from 25 °C to 65 °C, H2 recovery increased by about 46%, while the CO2 recovery remained basically unchanged. This was consistent with the effects of temperature on the permeation rates of the two membranes. The increase of gas permeation rate in the membrane could bring a certain increase in gas recovery. Thus, increasing operating temperature is beneficial to improve H2 recovery of the dual membrane separator. As shown in Figure 6b, the purity of H2 and CO2 both showed a downward trend. It was mainly caused by the decrease in the selectivities of H2/CO2 of PI membranes and CO2/H2 of PDMS/PEI membranes at higher temperatures. The increase of the CH4 permeation rate would also reduce the purity of H2 and CO2.
In industry, product recovery can be improved by increasing the stage cut, but it is relatively difficult to improve product purity. Therefore, selecting a lower temperature (such as 25 °C in this experiment) to ensure higher purity is more beneficial to the application of the dual membrane separator.

Effect of Operating Pressure
The presence of CO2 in the feed gas made the polymer matrix easily plasticized, which resulted in an increase in gas permeation rate. As the pressure increased, the polymer matrix tended to be plasticized strongly [28,29]. Figure 7 illustrated the effects of operating pressure on PI and PDMS/PEI membrane properties at 25 °C and a pressure range of 0.2-0.8 MPa. The permeation rates of H2, CO2,  Figure 6 depicts the effect of operating temperature on the purity and recovery of dual membrane separator. It could be observed from Figure 6a that as the temperature rose from 25 • C to 65 • C, H 2 recovery increased by about 46%, while the CO 2 recovery remained basically unchanged. This was consistent with the effects of temperature on the permeation rates of the two membranes. The increase of gas permeation rate in the membrane could bring a certain increase in gas recovery. Thus, increasing operating temperature is beneficial to improve H 2 recovery of the dual membrane separator. As shown in Figure 6b, the purity of H 2 and CO 2 both showed a downward trend. It was mainly caused by the decrease in the selectivities of H 2 /CO 2 of PI membranes and CO 2 /H 2 of PDMS/PEI membranes at higher temperatures. The increase of the CH 4 permeation rate would also reduce the purity of H 2 and CO 2 .  Figure 6 depicts the effect of operating temperature on the purity and recovery of dual membrane separator. It could be observed from Figure 6a that as the temperature rose from 25 °C to 65 °C, H2 recovery increased by about 46%, while the CO2 recovery remained basically unchanged. This was consistent with the effects of temperature on the permeation rates of the two membranes. The increase of gas permeation rate in the membrane could bring a certain increase in gas recovery. Thus, increasing operating temperature is beneficial to improve H2 recovery of the dual membrane separator. As shown in Figure 6b, the purity of H2 and CO2 both showed a downward trend. It was mainly caused by the decrease in the selectivities of H2/CO2 of PI membranes and CO2/H2 of PDMS/PEI membranes at higher temperatures. The increase of the CH4 permeation rate would also reduce the purity of H2 and CO2.
In industry, product recovery can be improved by increasing the stage cut, but it is relatively difficult to improve product purity. Therefore, selecting a lower temperature (such as 25 °C in this experiment) to ensure higher purity is more beneficial to the application of the dual membrane separator.

Effect of Operating Pressure
The presence of CO2 in the feed gas made the polymer matrix easily plasticized, which resulted in an increase in gas permeation rate. As the pressure increased, the polymer matrix tended to be plasticized strongly [28,29]. Figure 7 illustrated the effects of operating pressure on PI and PDMS/PEI membrane properties at 25 °C and a pressure range of 0.2-0.8 MPa. The permeation rates of H2, CO2, In industry, product recovery can be improved by increasing the stage cut, but it is relatively difficult to improve product purity. Therefore, selecting a lower temperature (such as 25 • C in this experiment) to ensure higher purity is more beneficial to the application of the dual membrane separator.

Effect of Operating Pressure
The presence of CO 2 in the feed gas made the polymer matrix easily plasticized, which resulted in an increase in gas permeation rate. As the pressure increased, the polymer matrix tended to be plasticized strongly [28,29]. Figure 7 illustrated the effects of operating pressure on PI and PDMS/PEI membrane properties at 25 • C and a pressure range of 0.2-0.8 MPa. The permeation rates of H 2 , CO 2 , and CH 4 by PI membranes were almost independent of the operating pressure, as shown in Figure 7a. and CH4 by PI membranes were almost independent of the operating pressure, as shown in Figure  7a. The results indicated that the industrialized PI membranes used in this study had a good ability to resist plasticizing. However, CO2 had higher solubility in the PDMS membrane. As shown in Figure 7b, in PDMS/PEI membranes, the permeation rates of all three gases increased with the operating pressure. Figure 8 shows the effect of operating pressure on the purity and recovery of the dual membrane separator. As is known to us, there is a clear trade-off relationship between recovery and purity in conventional single membrane separators. It was worth noting that, in the dual membrane separator, with the increase of operating pressure, the purity and recovery of H2 and CO2 increased simultaneously. This results fully reflected the enhanced separation effect of the dual membrane separator. The increase of CO2 permeation rate in PDMS/PEI membranes helps reduce the CO2 concentration in feed, and which leads to a larger driving force for H2 to permeate through PI membranes. At the same time, more H2 passed through the PI membranes, temporarily eliminating the effect of the PDMS/PEI membranes selectivity decline.
Therefore, a higher operating pressure is beneficial to the separation of the dual membrane separator. Certainly, the optimization of operating pressure should also consider the impact of compressor energy consumption and the pressure resistance of the hollow fiber in the separator.  The results indicated that the industrialized PI membranes used in this study had a good ability to resist plasticizing. However, CO 2 had higher solubility in the PDMS membrane. As shown in Figure 7b, in PDMS/PEI membranes, the permeation rates of all three gases increased with the operating pressure. Figure 8 shows the effect of operating pressure on the purity and recovery of the dual membrane separator. As is known to us, there is a clear trade-off relationship between recovery and purity in conventional single membrane separators. It was worth noting that, in the dual membrane separator, with the increase of operating pressure, the purity and recovery of H 2 and CO 2 increased simultaneously. This results fully reflected the enhanced separation effect of the dual membrane separator. The increase of CO 2 permeation rate in PDMS/PEI membranes helps reduce the CO 2 concentration in feed, and which leads to a larger driving force for H 2 to permeate through PI membranes. At the same time, more H 2 passed through the PI membranes, temporarily eliminating the effect of the PDMS/PEI membranes selectivity decline. and CH4 by PI membranes were almost independent of the operating pressure, as shown in Figure  7a. The results indicated that the industrialized PI membranes used in this study had a good ability to resist plasticizing. However, CO2 had higher solubility in the PDMS membrane. As shown in Figure 7b, in PDMS/PEI membranes, the permeation rates of all three gases increased with the operating pressure. Figure 8 shows the effect of operating pressure on the purity and recovery of the dual membrane separator. As is known to us, there is a clear trade-off relationship between recovery and purity in conventional single membrane separators. It was worth noting that, in the dual membrane separator, with the increase of operating pressure, the purity and recovery of H2 and CO2 increased simultaneously. This results fully reflected the enhanced separation effect of the dual membrane separator. The increase of CO2 permeation rate in PDMS/PEI membranes helps reduce the CO2 concentration in feed, and which leads to a larger driving force for H2 to permeate through PI membranes. At the same time, more H2 passed through the PI membranes, temporarily eliminating the effect of the PDMS/PEI membranes selectivity decline.
Therefore, a higher operating pressure is beneficial to the separation of the dual membrane separator. Certainly, the optimization of operating pressure should also consider the impact of compressor energy consumption and the pressure resistance of the hollow fiber in the separator.  Therefore, a higher operating pressure is beneficial to the separation of the dual membrane separator. Certainly, the optimization of operating pressure should also consider the impact of compressor energy consumption and the pressure resistance of the hollow fiber in the separator.

Effect of Membrane Area Ratio
The area ratio of the two membranes was a critical parameter in a dual membrane separator. Figure 9 presented the influence of membrane area radio (R) on membrane separation performance at different feed rates. The external separation efficiency of the dual membrane separator is closely related to the performance of the internal membrane material. Therefore, a fixed flow rate (100 mL/min) in Figure 9a,b was selected. At this point, the permeation rate and selectivity of the PI membrane in the dual membrane separator are shown in Table 2.

Conclusions
A PI/PDMS hollow fiber dual membrane separator was designed and applied to separate ternary gas mixtures. Higher gas separation performance could be achieved by adjusting various process parameters to match the permeability and selectivity of the two membranes. The membrane area ratio was a system parameter of the dual membrane separator, which could be used to adjust the purity and recovery of the single-sided product. With R increased, a higher H2 recovery could be achieved, but also a lower H2 purity. When R was 1.85, which means permeation of the two membranes was approximately equal, the dual membrane separator could have the best overall performance of the separation. At a high stage cut, the dual membrane separator effectively reduced the concentration polarization by reducing the CO2 concentration in feed gas, and the product purity was significantly improved compared with the PI single membrane separator. A high temperature was detrimental to the products' purity of the dual membrane separator. In order to obtain high purity products, the optimal temperature should be 25 °C. With the increase of operating pressure, the purity and recovery of H2 and CO2 increased at the same time. According to the requirements of  Mixed gas experiments were carried out at 25 • C and 0.2 MPa. The area of PDMS/PEI membranes was fixed, and R could be adjusted by changing the area of PI membranes. As illustrated in Figure 9a,b, a trade-off could also be found between the H 2 recovery and purity. When R increased from 0.87 to 2.94, H 2 recovery increased from 0.15 to 0.45 and H 2 purity dropped by about 12% at a feed flow rate of 100 ml/min. Figure 9c,d showed the trade-off of CO 2 purity and recovery on the PDMS/PEI membranes. "R = 0" implied that there was only a PDMS/PEI membrane in the separator.
The effect of the membrane area ratio on the dual membrane separator can be explained by the matching of gas permeability of PI membranes and PDMS/PEI membranes. When R was 2.94, most of the feed permeated through PI membranes, which resulted in the highest recovery of H 2 (Figure 9a). The H 2 purity and H 2 /CO 2 selectivity of PI membranes in the dual membrane separator were similar to