Separation of CO2 from Small Gas Molecules Using Deca-Dodecasil 3 Rhombohedral (DDR3) Membrane Synthesized via Ultrasonically Assisted Hydrothermal Growth Method

Department of Chemical Engineering and Chemistry, Eindhoven University of Technology, Netherlands Department of Chemical Engineering, Universiti Teknologi PETRONAS (UTP), Seri Iskandar, Perak 32610, Malaysia Department of Chemical Engineering, COMSATS University Islamabad (CUI), Lahore Campus, Punjab 54000, Pakistan Department of Chemical, Polymer and Composite Materials Engineering, University of Engineering and Technology Lahore, New Campus, Pakistan School of Chemical and Materials Engineering, National University of Sciences and Technology, Islamabad 44000, Pakistan Department of Chemistry, College of Science, King Khalid University, Abha 61413, P.O. Box 9004, Saudi Arabia


Introduction
Over the last few decades, much research has been done on the CO 2 separation especially in syngas production (mainly H 2 ) and natural gas purification. The presence of CO 2 in the syngas reduces the yield of hydrogen and methane recovery during the Fischer-Tropsch (FT) process [1,2]. Similarly, the presence of CO 2 decreases the heating value of natural gas and caused equipment corrosion in the existence of water [3,4]. Meanwhile, membrane separation technology has received much attention in CO 2 separation mainly due to its advantages compared to the conventional separation technologies [5][6][7][8][9][10][11]. Zeolite membrane is favored in the CO 2 separation among the membrane materials over polymeric and mixed matrix membranes due to its characteristics, including well-specified pores, the molecular filtering property, and strong CO 2 permeation [12,13]. Table 1 shows the comparison of different zeolite membranes reported in the literature for CO 2 separation from CH 4 .
Recently, the DDR3 zeolite membrane has attracted various researchers due to its aperture size of 0:36 nm × 0:44 nm which resulted in a remarkable performance in the separation of CO 2 from CH 4 [17][18][19][20][21]. In 27 to 31 days, many scientists used a secondary hydrothermal growth process and evaluated membrane efficiency in CO 2 separation from CH 4 . DDR3 membrane is synthesized by different researchers. Tomita et al. documented CO 2 permeance over a membrane of DDR3 of 3:0 × 10 −7 mol/m 2 sPa and selectivity CO 2 /CH 4 of 200. CO 2 permeance was then obtained at 7:80 × 10 −9 mol/m 2 sPa and selectivity at CO 2 /CH 4 at 6.1 obtained by Nakayama et al. [19]. Recently, Bose et al. [17] investigated the fabrication of DDR3 membrane in 7 days and they found that CO 2 permeance of 1:1 × 10 −7 mol/m 2 sPa and H 2 /CO 2 selectivity of 3.70 was achieved at ambient conditions.
Although several studies have reported the gas permeation of DDR3 membrane in CO 2 separation from CH 4 , researchers rarely reported on the membrane performance in CO 2 separation from other small gas molecules including H 2 and N 2 . Therefore, in the present work, we focus on the performance of the DDR3 membrane synthesized in our previous work [22] for CO 2 separation from small gas molecules including H 2 , CH 4 , and N 2 . This research also explores the impact of the impurities of hydrocarbons on the efficiency of CO 2 distinguishing the DDR3 membrane from CH 4 .

Materials and Methods
Chemicals and the procedure used for the synthesis of DDR3 membrane have been reported in detail in our previous work [10,22]. Firstly, the mixture solution was prepared using 1adamantaneamine, tetramethoxy silane, ethylenediamine, and water. Subsequently, the mixture solution was undergoing ultrasonic-irradiation pretreatment for 1 h prior to hydrothermal growth for 1 day. Then, the seeds were recovered, washed with deionized water, and attached on α-alumina support via a vacuum-assisted seeding method. In the typical synthesis of DDR3 membrane, the seeded support was placed in an autoclaved filled with solution mixture prepared using a similar recipe and procedure as described in our previous work [10]. After that, the membrane was grown in 2 days under hydrothermal heating. Consequently, the resultant membrane was preceded for calcination and characterization. The crystallinity and morphology of the membrane were characterized using an X-ray diffractometer (Bruker D8 Advance) and field emission scanning electron microscopy (Zeiss Supra 55 VP).

Permeation Study
2.1.1. Pure Gas Permeability. The performance of the resultant DDR3 membrane was investigated using the permeation test rig as shown in Figure 1. The membrane was sealed in the permeation cell, and the flow rate of the targeted gases including, H 2 , CO 2 , N 2 , and CH 4 was fixed at 200 ml/min while the temperature (MKS instrument) and feed pressure were maintained at room temperature and 2 bar, respectively. Temperature and feed pressure were controlled with an error of +/-0.1°C and 0.1 bar using MKS instrument temperature and feed pressure controllers, respectively. The gas permeation was calculated using equation (1), while ideal selectivity was calculated using equation (2) as follows [22,23].
where J i is the flux (mol/m 2 s) and ΔP is the pressure drop (Pa) while α i,j is the gas pair selectivity.  4 was also conducted using the same permeation test rig. The experiments were conducted in three consecutive steps in which the first and third steps were used to investigate the stability of DDR3 membrane. Meanwhile, the second step of the experiment was carried out to study the effect of hydrocarbon impurities on the performance of the resultant membrane in CO 2 separation from CH 4 . During the first step of the experiment, temperature, inlet pressure, and CO 2 inlet composition were maintained at 30.1°C, 1 bar, and 10 vol% (90 vol% of CH 4 ), respectively, because this experimental condition was obtained as the optimum CO 2 /CH 4 separation condition as reported in our previous work [10]. After 24 h, the second step of the testing was conducted by feeding the CO 2 and CH 4 gas mixture containing hydrocarbon impurities with the compositions of 35.84% CO 2 , 44.50% CH 4 , 6.27% C 3 H 8 , 3.44% C 4 H 10 , and 9.94% C 2 H 6 into the system. This composition resembled the natural gas composition presence in North Sumatra Basin, Indonesia. After 48 h of testing, the feed gas was cut off and the experiment was continued using the feed gas composition applied in the first step of the testing. After each step, prior to the measurement of gas permeances, the system was allowed to stabilize for 30 min.
Gas compositions were calculated using gas chromatographs in feed, permeate, and retention streams (PerkinElmer Clarus 580). The CO 2 and CH 4 permeations were determined with equation (1) with the amount of pressure decrease obtained by equation (3) [24]. The selectivity of the composition of CO 2 /CH 4 was determined subsequently from the mole fractions as defined in equation (4).
where P f and P p are partial pressures in feed and permeate, respectively.
where X i and X j are mole fractions of the i and j elements of the binary mixture. P and F subscriptions reflect both permeate and feed.

Results and Discussion
3.1. Characterization. The details on XRD, FESEM, FTIR, and EDX analysis on DDR3 membrane have been described in our previous work [10]. The significant XRD peaks and FESEM images obtained from the resultant membrane showed comparable crystallinity and hexagonal morphology which are analogous with those findings documented in the researches for DDR3 membrane [9,10,25,26]. The surface and cross-sectional view of the membrane showed that a uniform, flat, and fully intergrown membrane layer was formed.  Figure 1: Schematic of permeation test rig used in the present work. Figure 2 shows the performance of the membrane at ambient temperature and feed pressure of 2 bar. Referring to Figure 2, H 2 , CO 2 , N 2 , and CH 4 permeances of 4:10 × 10 −7 mol/m 2 sPa, 2:46 × 10 −7 mol/m 2 sPa, 9:11 × 10 −8 mol/m 2 sPa, and 4:73 × 10 −8 mol/m 2 sPa are obtained, respectively. It can be observed from Figure 2 that H 2 permeance is higher than that of those values obtained for CO 2 , N 2 , and CH 4 gases. This could be due to the significant difference in the Knudsen diffusion value of the gases through the membrane. H 2 with a molecular diameter of 0.29 nm is smaller than of the molecular diameter of CO 2 , N 2 , and CH 4 , as well as the DDR3 pore channel. Meanwhile, the molecular diameter of CO 2 , N 2 , and CH 4 of 0.33 nm, 0.36 nm, and 0.38 nm, respectively, is closer to the DDR3 pore channel. Thus, H 2 can diffuse faster compared to CO 2 , N 2 , and CH 4 gases.

Gas Permeation.
From Figure 2, it is found that the gas permeances are decreased in the order of H 2 > CO 2 > N 2 > CH 4 . This decreasing trend is mainly attributed to the molecular sieving behaviour of DDR3 membrane and Knudsen diffusion mechanism of the gases [25][26][27][28]. The diffusion rate of these gases through the DDR3 membrane decreases when the kinetic diameter of the gases increases, and thus, decreasing order of gas permeance has been observed [17]. Moreover, H 2 , CO 2 , N 2 , and CH 4 permeances obtained from the membrane were comparable with those results reported in the literature [17,20,25].
As shown in Figure 2 also, the ideal selectivities of CO 2 /CH 4 , CO 2 /N 2 , and H 2 /CO 2 of 5.22, 2.27, and 1.66, respectively, are obtained. It is interesting to note that the ideal selectivities of the gas pairs decreased in the sequence of CO 2 /CH 4 > CO 2 /N 2 > H 2 /CO 2 . This result is attributed to the CH 4 permeance, which is much smaller than the permeance of H 2 and N 2 as shown in Figure 3. Overall, although the gas permeance and ideal selectivity results obtained in the present work are comparable with those results reported in the literature [25], the relatively low performance of DDR3 membrane could be due to the presence of minor defects or intercrystalline pores which have reduced the performance of the membrane. Thus, it is suggested that the membrane should be modified with a TEOS solution to minimized intercrystalline defects to enhance its performance in gas separation [24]. TEOS could penetrate into intercrystalline defects and seal it through the formation of hydrogen bonding between membrane zeolite materials and TEOS. In addition, it also slows down the CH 4 permeation by providing extra resistance due to its higher kinetic diameter [25]. Figure 4 shows the stability test and effect of hydrocarbon impurities on the performance of the membrane in CO 2 separation from CH 4 for the testing duration of 96 h. Sections І and III shown in Figure 4 display the stability results of DDR3 membrane in CO 2 separation from a binary gas mixture of CO 2 and CH 4 at a temperature of 30.1°C, inlet pressure of 1 bar, and CO 2 inlet composition of 10 vol% because this experimental condition was obtained as the optimum CO 2 /CH 4 separation condition which provided higher permeation performance of DDR3 membrane as reported in our previous work [10].

Effect of Hydrocarbon Impurities on the Performance of DDR3 Membrane.
Meanwhile, section II in Figure 4 exhibits the effect of hydrocarbon impurities on the performance of DDR3 membrane in CO 2 separation from CH 4 at a temperature of 30.1°C and feed pressure of 1 bar. Referring to Figure 4, at points C and E, CO 2 and CH 4 permeances of 3:13 × 10 −8 mol/m 2 sPa and 2:81 × 10 −8 mol/m 2 sPa are obtained, respectively. Meanwhile, CO 2 /CH 4 selectivity of 2.59 was obtained at point G.
Results at points C * , E * , and G * were obtained when the feed gas was changed from a pure binary mixture of CO 2 /CH 4 to CO 2 /CH 4 gas mixture containing hydrocarbon impurities. At points C * and E * , CO 2 and CH 4 permeances of 3:11 × 10 −8 mol/m 2 sPa and 2:83 × 10 −8 mol/m 2 sPa are obtained, respectively. On the other hand, CO 2 /CH 4 selectivity of 2.59 was obtained at point G * .
It can be seen from Figure 4 that the performance of the membrane slightly decreases when the gas mixture containing hydrocarbon impurities is fed into the system. Besides, from Figure 4, it is found that CO 2 and CH 4 permeances decrease from point E * to F and C * to D, respectively, after    [3]. Adsorption of hydrocarbon impurities into the pores of DDR3 pore reduced its capacity for the diffusion of CO 2 and CH 4 molecules. Subsequently, adsorption of hydrocarbon impurities may also cause saturation for adsorption of CO 2 and CH 4 molecules which may also resulted to drop the performance of the membrane. In addition, a competitive sorption effect has been reported previously that mixed gas performance is always lower than pure gas due to a competitive sorption mechanism [4,16]. However, CO 2 /CH 4 selectivity of 2.54 was found at point H, which is slightly lower than the CO 2 /CH 4 selectivity obtained at point G * . It is found that    Figure 4: Effect of hydrocarbon impurities on the performance of DDR3 membrane and its durability test (separation results points C, E, and G were determined at a temperature of 30.1°C, feed pressure of 1 bar, with the CO 2 composition of 35.84% (CH 4 composition of 64.16%)).

Advances in Polymer Technology
the loss of CO 2 /CH 4 selectivity at point H was only about 4.2% compared to the CO 2 /CH 4 selectivity which was initially obtained at point G * .
Furthermore, it can be seen from Figure 4 (section I) that, when the hydrocarbon impurities are not presented in the feed gas, the performance of DDR3 membrane remains stable. After 24 h, nearly constant CO 2 and CH 4 permeances of 7:23 × 10 −8 mol/m 2 sPa and 4:51 × 10 −8 mol/m 2 sPa were obtained, respectively, while CO 2 /CH 4 selectivity of 3.66 was achieved (line A). These results are similar to those values obtained for CO 2 permeance, CH 4 permeance, and CO 2 /CH 4 selectivity when the feed gas was changed to the binary gas mixture of CO 2 and CH 4 at the testing duration from 80 h to 96 h. In section III, CO 2 and CH 4 permeances of 7:08 × 10 −8 mol/m 2 sPa and 4:41 × 10 −8 mol/m 2 sPa were obtained (line B), while CO 2 /CH 4 selectivity of 3.69 was achieved.
Therefore, it can be concluded that the presence of hydrocarbon impurities in CO 2 and CH 4 gas mixture has affected the performance of DDR3 membrane and caused a slight reduction of CO 2 and CH 4 gas over a testing duration of 40 h (from 32 h to 72 h). Consequently, CO 2 /CH 4 selectivity of DDR3 membrane remained stable at testing condition of temperature of 30.1°C and feed pressure of 1 bar and CO 2 feed composition of 10 vol%.

Stability of DDR3
Membrane. Figure 4 also shows the stability of DDR3 membrane in CO 2 separation from CH 4 . It can be seen from Figure 4 that CO 2 permeability, CH 4   Figure 4. From these results, it has been found that membrane performance was dropped in small quantity and thus, it was decided that the membrane was stable in the presence of hydrocarbons.
In conclusion, the permeation performance of DDR3 membrane in the separation of CO 2 /CH 4 remained steady up to 96 h with small losses of CO 2 permeability, CH 4 permeability, and CO 2 /CH 4 selectivity. Thus, the separation performance of DDR3 membrane is durable and stable in the presence of hydrocarbon impurities.
The findings seen in the literature under the same temperature and pressure condition assess the efficiency of DDR3 membrane in CO 2 /CH 4 , H 2 /CO 2 , and CO 2 /N 2 separations. With respect to Figure 5(a), the ideal CO 2 /CH 4 selectivity obtained in this analysis is comparable to the published findings by Yajima and Nakayama [21]. However, this value is below the published results by Himeno et al. [18] and van den Bergh et al. [25]. Relatively low CO 2 /CH 4 ideal selectivity of the DDR3 membrane obtained during the present research may be triggered by the existence of slight membrane defects which may influence CO 2 or CH 4 transport pathways via the membrane.
On the other hand, the performance of DDR3 membrane in CO 2 /N 2 and H 2 /CO 2 separation is shown in Figures 5(b) and 5(c). From Figure 5(b), it is found that the CO 2 /N 2 selectivity obtained in the present work is lower than those results reported by Nakayama et al. [19]. Referring to Figure 5(c), the ideal selectivity of H 2 /CO 2 obtained in the present study is slightly higher than those results reported by Tomita et al. [20] and Zheng et al. [26]. However, these values were lower than those results reported by Bose et al. [17]. Low ideal selectivity values of CO 2 /N 2 and H 2 /CO 2 obtained in the present work could be because of the similar reason as described previously for the performance of the membrane in CO 2 /CH 4 separation. The presence of defects on the membrane affected the transport mechanisms of N 2 , H 2 , and CO 2 through the membrane and lowered the selectivity values of CO 2 /N 2 and H 2 /CO 2 .

Conclusions
In the present work, the performance of DDR3 membrane for the CO 2 separation from gas molecules exhibiting comparatively smaller kinetic diameters, i.e., N 2 and CH 4 , has been successfully conducted. The permeances of the gases obtained over the membrane decreased in the order of H 2 > CO 2 > N 2 > CH 4 . Subsequently, it was also found that the DDR3 membrane demonstrated an ideal selectivity of 5.22 for CO 2 /CH 4 which was higher than the ideal selectivity of H 2 /CO 2 and CO 2 /N 2 gas pairs. Besides, ideal selectivities obtained from DDR3 membrane decreased in the sequence of CO 2 /CH 4 > CO 2 /N 2 > H 2 /CO 2 . Furthermore, it was observed that the presence of hydrocarbon impurities in the CO 2 and CH 4 gas mixture directly affects the performance of DDR3 membrane and contributed to the losses of CO 2 permeability, CH 4 permeability, and CO 2 /CH 4 selectivity of 39.1%, 14.8%, and 4.2%, respectively. Consequently, from the stability test, it can be suggested that the permeation performance of DDR3 membrane remained stable for 96 h, even after the separation testing of a gas mixture containing hydrocarbon impurities, with losses of CO 2 permeability, CH 4 permeability, and CO 2 /CH 4 selectivity of 2.35%, 4.43%, and 0.55%, respectively. From the comparison study, it was concluded that the permeation performance of DDR3 membrane in CO 2 separation from H 2 , CH 4 , and N 2 still needs to be improved mainly due to the presence of defects in the membrane.

Data Availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.